No title

HAZ IFC 195(C).qxd:INNER COVER 10/11/11 2:31 PM Page 1 SUBMISSION OF ARTICLES Articles must be written in good Engli...

5 downloads 7240 Views 41MB Size Report

This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!

Report copyright / DMCA form

DOWNLOAD PDF

HAZ IFC 195(C).qxd:INNER COVER

10/11/11

2:31 PM

Page 1

SUBMISSION OF ARTICLES Articles must be written in good English. We recommend that authors whose native language is not English have their manuscripts checked before submitting. A high standard of English is a criterion of acceptance for the review process. Articles not written in acceptable English will be returned to the author without review. Language services: Authors who require information about language editing and copyediting services pre- and post-submission please visit http://webshop. elsevier.com/languageediting or our customer support site at http://support.elsevier.com.

Aims and Scope The Journal of Hazardous Materials publishes full length research papers, reviews, case studies and short communications which improve our understanding of the hazards and risks certain materials pose to people and the environment or deal with ways of controlling these hazards and associated risks. The Editors particularly welcome proposals for review articles. To limit the scope the following areas are excluded: standard municipal and waste and wastewater treatment processes; manufacturing of explosives. The Journal publishes high-impact contributions on: • Characterization of the harmful effects of chemicals and materials • Measurement and monitoring of hazardous materials • Transport and fate of HM in environment • Risk assessment and management • Physico-chemical and separation processes • Biological processes • Advanced oxidation processes • Safer and cleaner technologies The Journal publishes papers with significant novelty and scientific impact. The Editors reserve the right to decline, without external review, papers that do not meet these criteria, including papers that: • Are very similar to previous publications, with changed target substrates, employed materials, analyzed sites or experimental methods, • Deal with parameter optimization of known processes without new concepts and/or interpretations, • Report the environmental analysis and monitoring of specific geographic areas without presenting new insights and/or hypothesis testing, or • Do not focus on the environmental relevance and significance of the studied systems or materials. Editors Professor Diana S. Aga, Department of Chemistry, State University of New York (SUNY) at Buffalo, 359 Natural Sciences Complex, Buffalo, NY 142603000, USA, Tel: (716) 645-4220, E-mail: [email protected] Professor Wonyong Choi, School of Environmental Science & Engineering Pohang University of Science and Technology (POSTECH) Pohang, Korea 790-784 Phone: +82-54-279-2283 Fax: 82-54-279-8299 E-mail:[email protected] Professor Andrew Daugulis, Department of Chemical Engineering, Queen’s University, Dupuis Hau, Kingston, K7L 3N6, Canada, Tel.: (613) 533-2784; Fax: (613) 533-6637; E-mail: [email protected] Professor Gianluca Li Puma, Department of Chemical Engineering, Loughborough University, Loughborough LE11 3TU, United Kingdom. Tel: +44 (0) 1509 222 510; Fax: +44 (0) 1509 223 923; E-mail: [email protected] Professor Gerasimos Lyberatos, School of Chemical Engineering, National Technical University of Athens, Zografou Campus, Athens 15780, Greece. Tel.: +302107723256; Fax: +302107723155; E-mail: [email protected] Professor Tay Joo Hwa, Division of Environmental & Water Resources Engineering, School of Civil & Environmental Engineering, Nanyang Technology University, Nanyang Avenue, Singapore 639798, Tel: +65 6790 6228, Fax: +65 6791 0676, E-mail: [email protected] Associate Editors Jaesang Lee, Water Environment Center, National Agenda Research Division, Korea Institute of Science and Technology (KIST), 39-1 Hawolgok-Dong, Seongbuk-Gu, 136-791 Seoul, South Korea, Tel: +82-2-958-6947, E-mail: [email protected] Constantina Marazioti, Laboratory of Biochemical & Environmental Technology, Dept. of Chemical Engineering, University of Patras, Patras, Greece, Tel: +30 6976 128157, E-mail: [email protected]

Online Submission of Articles Submission to this journal proceeds entirely on-line via http://ees.elsevier.com/hazmat. You will be guided stepwise through the creation and uploading of the various files. The system automatically converts source files to a single Adobe Acrobat PDF version of the article, which is used in the peer-review process. Please note that even though manuscript source files are converted to PDF at submission for the review process, these source files are needed for further processing after acceptance. All correspondence, including notification of the Editor’s decision and requests for revision, takes place by e-mail and via the Author’s homepage, removing the need for a hard-copy paper trail.

MANUSCRIPT PREPARATION Authors are requested to refer to the detailed guide for authors and artwork instructions available via the Journal’s homepage at http://www.elsevier. com/locate/hazmat

AUTHOR ENQUIRIES For enquiries relating to the submission of articles (including electronic submission) please visit this journal’s homepage at http://www.elsevier.com/locate/jhazmat. Contact details for questions arising after acceptance of an article, especially those relating to proofs, will be provided by the publisher. You can track accepted articles at http://www.elsevier.com/trackarticle. You can also check our Author FAQs at http://www.elsevier.com/authorFAQ and/or contact Customer Support via http://support.elsevier.com.

For a full and complete Guide for Authors, please go to: http://www.elsevier.com/locate/ jhazmat

OFFPRINTS The corresponding author, at no cost, will be provided with a PDF file of the article via e-mail. The PDF file is a watermarked version of the published article and includes a cover sheet with the journal cover image and a disclaimer outlining the terms and conditions of use. Additional paper offprints can be ordered by the authors. An order form with prices will be sent to the corresponding author.

FUNDING BODY AGREEMENTS AND POLICIES Elsevier has established agreements and developed policies to allow authors whose articles appear in journals published by Elsevier, to comply with potential manuscript archiving requiremens as specified as conditions of their grant awards. To learn more about existing agreements and policies please visit http://www.elsevier.com/fundingbodies

INFORMATION FOR ADVERTISERS Advertising information: If you are interested in advertising or other commercial opportunities please e-mail [email protected] and your enquiry will be passed to the correct person who will respond to you within 48 hours. USA mailing notice: Journal of Hazardous Materials (ISSN 0304-3894) is published semi-monthly by Elsevier B.V. (Radarweg 29, 1043 NX Amsterdam, The Netherlands). Periodical postage paid at Rahway NJ and additional mailing offices. USA POSTMASTERS: Send change of address to Journal of Hazardous Materials, Elsevier Customer Service Department, 3251 Riverport Lane, Maryland Heights, MO 63043, USA. AIRFREIGHT AND MAILING in the USA by Mercury International Limited, 365, Blair Road, Avenel, NJ 07001. The paper used in this publication meets the requirements of ANSI/NISO Z39.48-1992 (Permanence of Paper).

Printed by Polestar Wheatons Ltd, Exeter, UK

Editorial Board R. Andreozzi (Naples, Italy) J. Baeyens (Tremelo, Belgium) A. Ballester (Madrid, Spain) E. Brillas (Barcelona, Spain) V. Cozzani (Bologna, Italy) G. de la Rosa (Guanajuato, Mexico) D. Dong (Changchun, China) M. Erdem (Elazig˘, Turkey) S. Esplugas (Barcelona, Spain) M. Fingas (Alberta, Canada) V. K. Gupta (Roorkee, India) O. Hamdaoui (Annaba, Algeria)

W.-J. Lee (Tainan, Taiwan, ROC) P. Lianos (Patras, Greece) E. Lima (Porto Alegre, Brazil) T. Mangialardi (Rome, Italy) A. Mittal (Bhopal, India) D. Mohan (New Delhi, India) C. Mulligan (Montreal, Canada) A. Nzihou (Albi, France) J.M. Park (Pohang-si, Republic of Korea) H.J. Pasman (Delft, The Netherlands) S. Pavlostathis (Atlanta, GA, USA) H. Potgieter (WITS, South Africa)

P. Randall (Cincinnatti, OH, USA) K. Reddy (Chicago, IL, USA) J. Sneddon (Lake Charles, Louisiana, USA) M. Soylak (Kayseri, Turkey) M. Talawar (The Hague, Netherlands) T. Townsend (Gainesville, FL, USA) S. Tunali Akar (Eskis¸ehir, Turkey) S. Wang (Perth, WA, Australia) H.-Q. Yu (Hefei, China) J. Yu (Wuhan, China) S. Zeman (Pardubice, Czech Republic) A.I. Zouboulis (Thessaloniki, Greece)

Publication schedule and subscription information

Journal of Hazardous Materials (ISSN 0304-3894). For 2011 volumes 182–192 are scheduled for publication. Subscription prices are available upon request from the Publisher or from the Elsevier Customer Service Department nearest you or from this journal’s website (http://www.elsevier.com/locate/jhazmat). Further information is available on this journal and other Elsevier Products through Elsevier’s website (http://www.elsevier.com). Subscriptions are accepted on a prepaid basis only and are entered on a calendar year basis. Issues are sent by standard mail (surface within Europe, air delivery outside Europe). Priority rates are available upon request. Claims for missing issues should be made within six months of the date of dispatch. Orders, claims, and journal enquiries: please contact the Elsevier Customer Service Department nearest you: St. Louis: Elsevier Customer Service Department, 3251 Riverport Lane, Maryland Heights, MO 63043, USA; phone: (877) 8397126 [toll free within the USA]; (+1) (314) 4478878 [outside the USA]; fax: (+1) (314) 4478077; e-mail: [email protected] Oxford: Elsevier Customer Service Department, The Boulevard, Langford Lane, Kidlington, Oxford OX5 1GB, UK; phone: (+44) (1865) 843434; fax: (+44) (1865) 843970; e-mail: [email protected] Tokyo: Elsevier Customer Service Department, 4F Higashi-Azabu, 1-Chome Bldg, 1-9-15 Higashi-Azabu, Minato-ku, Tokyo 106-0044, Japan; phone: (+81) (3) 5561 5037; fax: (+81) (3) 5561 5047; e-mail: [email protected] Singapore: Elsevier Customer Service Department, 3 Killiney Road, #08-01 Winsland House I, Singapore 239519; phone: (+65) 63490222; fax: (+65) 67331510; e-mail: [email protected] doi:10.1016/S0304-3894(11)01250-7

Journal of Hazardous Materials 195 (2011) 1–10

Contents lists available at ScienceDirect

Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat

Review

The modern paradox of unregulated cooking activities and indoor air quality Ki-Hyun Kim a,∗ , Sudhir Kumar Pandey a,1 , Ehsanul Kabir a , Janice Susaya a , Richard J.C. Brown b a b

Department of Environment & Energy, Sejong University, Seoul, Republic of Korea Analytical Science Division, National Physical Laboratory, Teddington TW11 0LW, UK

a r t i c l e

i n f o

Article history: Received 17 June 2011 Received in revised form 10 August 2011 Accepted 11 August 2011 Available online 17 August 2011 Keywords: Food Cooking Hazardous pollutants Cancer Human health Emission factor

a b s t r a c t Pollutant emission from domestic and commercial cooking activities is a previously neglected area of concern with respect to human health worldwide. Its health effects are relevant to people across the globe, not only those using low quality food materials in lesser-developed countries but also to more afﬂuent people enjoying higher quality food in developed countries. Based on the available database of pollutant emissions derived from ﬁre-based cooking, its environmental signiﬁcance is explored in a number of ways, especially with respect to the exposure to hazardous vapors and particulate pollutants. Discussion is extended to describe the risk in relation to cooking methods, cooking materials, fuels, etc. The observed pollutant levels are also evaluated against the current regulations and guidelines established in national and international legislation. The limitations and future prospects for the control of cooking hazards are discussed. © 2011 Elsevier B.V. All rights reserved.

Contents 1. 2. 3. 4. 5. 6. 7.

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Pollutants released from cooking fuels and their effect on indoor air pollution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The effect of different cooking methods and ingredients on IAQ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Emission inventories of pollutants released from cooking activities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The level of cooking-related emission in relation to regulations and guidelines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Health and environmental impacts of cooking activities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Future directions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1. Introduction With the progress of healthcare science, human life expectancy has increased gradually over the years. Likewise, with the increasing pervasiveness of advanced civilization and urbanization many primitive risks that threatened human life previously have been reduced or eliminated. As such, the pattern of risks to livelihood and their relative magnitudes have also been altered dramatically. Among many risks in our normal everyday life, not many people are aware of the risks associated with cooking activities. As the use of ﬁre became part of human culture, all populations have become

∗ Corresponding author. Tel.: +82 2 3408 3233; fax: +82 2 3408 4320. E-mail addresses: [email protected], [email protected] (K.-H. Kim). 1 Present address: Department of Botany, Guru Ghasidas Central University Bilaspur (C.G.), 495 009, India. 0304-3894/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.jhazmat.2011.08.037

1 2 2 2 5 7 9 9 9

prone to this cooking-related risk, regardless of their race, age, wealth, cultural food preferences, etc. The selection of ﬁre-based cooking approaches such as frying, roasting, and grilling can exert a signiﬁcant impact not only on the quality of the food but also on pollutant emissions [1,2]. The extent of the latter can be controlled by the combined effects of different recipes, cooking procedures, food materials (and ingredients), fuel types, extraction/ventilation equipment, etc. [3]. The style of these cooking activities and their impact can also be affected by macroscale variables like population, culture, climate, and geographical location. Thus assessment of these cooking related risks becomes a delicate issue with sociological sensitivities, if certain cooking types with deeply ingrained traditional methods are labeled as increased risk. Humans can be subject to cooking-related risks via various intake routes either directly (overcooked foodstuffs) or indirectly (fumes). Our emphasis in this review is directed mainly

2

K.-H. Kim et al. / Journal of Hazardous Materials 195 (2011) 1–10

towards indoor air pollutants (IAP) liberated from ﬁre-based cooking activities, the ubiquitous risks of their exposure, and guidelines/regulations for maintaining indoor air quality (IAQ) from such activities. 2. Pollutants released from cooking fuels and their effect on indoor air pollution As most ﬁre-based cooking cannot be carried out without fuels, the effect of fuel combustion can add to the risks of cooking activities. In fact, cooking fuels are one of the most important causes of IAP, particularly in developing countries [4]. Nearly half of the world’s population use solid fuels and biomass (e.g., coal, wood, animal dung, and crop residues) as their primary energy source [5,6]. Solid fuel is used commonly among the poor, particularly in rural areas in developing countries [7]. In urban areas comparatively clean fuels (like liquid petroleum gas (LPG), biogas, natural gas, and kerosene) are available for cooking purposes, along with electricity [7]. Biomass combustion produces a large number of harmful pollutants including respirable particulate matter (PM10 ), carbon monoxide (CO), nitrogen oxides (NO2 ), formaldehyde, benzene, poly-cyclic aromatic hydrocarbons (PAH), and many other toxic compounds [8,9]. Furthermore, because of the relatively conﬁned nature of indoor spaces with low air turnover rates, pollutants liberated inside will not disperse quickly to sustain low IAQ. Pollutants emitted from fuel consumption (for cooking purposes) have been studied in relation to cooking fuel type as part of IAP research (Tables 1 and 2). Traditional open-ﬁre cooking stoves, used extensively in rural households in many developing countries, generally release high quantity of particles and harmful pollutants in smoke [10–12]. Indoor PM10 concentrations from cooking via biomass combustion were measured as 1545 ␮g m−3 in Kenya [13] and 1200 ␮g m−3 in Mozambique [14]. In rural Bolivia, the 6h mean levels of PM10 , when cooking from indoor and outdoor kitchens by cow dung, were 1830 and 250 ␮g m−3 , respectively [15]. Additionally, common volatile organic compounds (VOCs) such as benzene, toluene, and xylene (commonly called BTX) were signiﬁcantly higher with the use of biomass fuel relative to natural gas [16]. Likewise, considerable emissions of VOCs and metals have also been detected from combustion of BBQ charcoals produced from several countries [17,18]. As cooking activities occur in the close proximity of people, they are usually exposed to high levels of these pollutants [19]. 3. The effect of different cooking methods and ingredients on IAQ Cooking ingredients vary widely, reﬂecting local environmental, geographical, economic, and cultural factors. Moreover, cooking

methods are diverse enough to encompass baking, roasting, frying, grilling, barbecuing, smoking, boiling, steaming, microwaving, braising, etc. As such, the cooking of each food reﬂects its own combinations for the above factors. The data on air pollutant emissions as a function of different food ingredients are summarized in Table 3. Moreover, as the pollutant type and levels are also inﬂuenced by the cooking methods, the data for air pollutant emissions is also compiled in relation to cooking style in Table 4. For instance, stir-frying in a wok is the most common cooking practice in China through which many HAPs are released [20]. Schauer et al. [21] estimated emission rates of gas-phase, semi-volatile, and particle-phase organic compounds (C1 to C27) from commercial-scale food cooking operations using seed oils. In Korean-style barbecue restaurants using hot steel pan and broiling steel bars (above a charcoal burner), a list of 99 pollutants (including respirable suspended particulates (RSP), CO, and VOCs) were detected [22]. Lee et al. [20] investigated the IAP at four restaurants in metropolitan Hong Kong and found high concentrations of formaldehyde (177 ␮g m−3 ) and benzene (18.4 ␮g m−3 ) in the dining areas of the Korean-style barbecue restaurant. Furthermore, meat charbroiling was thus identiﬁed as one of the previously unconsidered sources of heavy aldehydes in urban air [23]. Acrolein is also released from heated oils during domestic cooking. Heated canola, extra-virgin olive, and olive oils, when heated at 180 ◦ C, were reported to emit acrolein at 52.6, 9.3, and 9.6 mg h−1 L−1 [24], respectively. Moreover, indoor acrolein levels are found to persist for a considerable time (a half-life of 14.4 ± 2.6 h) after cooking under poor ventilation [25].

4. Emission inventories of pollutants released from cooking activities Cooking related emissions can be important sources of major airborne pollutants (e.g., PM, SO2 , and CO) as well as trace-level pollutants (e.g., secondary organic aerosols (SOA), organic carbon (OC), and elemental carbon (EC)). The latter also represents important constituents of the global carbon balance. Emissions inventories for cooking activities are generally lacking, and considered a “missing or unaccounted fraction of the area source category”, regardless of pollutant type. Several researchers estimated emissions inventories for many cooking activities based on various statistical approaches (Table 5), and some limited national inventories exist. In the UK, the National Atmospheric Emissions Inventory (NAEI – www.naei.org.uk) estimated VOC emissions for commercial food production activities such as animal feed manufacture, biscuit, cake and cereal production, coffee roasting, sugar production, and margarine and vegetable oil production.

Table 1 Comparison of indoor air pollutant concentrations measured from different fuels used for cooking (concentrations in ␮g m−3 ). Fuels used for cooking

CO

SO2

Natural gas

4170

185

Coal Charcoal LPG

Biomass

6550

PM10

– 56.2 –

247

710 50 147 744 1545

Benzene 13.7

4.24

30.3

27.3 6.03

70.9 36.3

959

61.5

3322

60.1 135

Toluene 2.70

Xylene 3.81

Formaldehyde 17.2

708 315

– 115 284 –

Wood Dung

436

NO2

54.2 353 1200 300 1830

625

34.2

337

18.5

9.92

Country

References

China Bangladesh China Korea India Malaysia India India Kenya Bangladesh India Mozambique Malaysia Bolivia

[34] [16] [34] [17] [11] [12] [35] [11] [13] [16] [35] [14] [12] [15]

Table 2 Comparison of emission concentrations between different cooking activities in various indoor (and outdoor) environments. CO2 CO (ppm) (ppb)

Environment/ emission source

Hongkong (Lee et al. [20])

1648 15.7 Korean BBQ-style restaurant (indoor) 2344 Chinese hot-pot 8.11 restaurant (indoor) 1031 2.23 Chinese dim sum restaurant (indoor) 636 0.01 Western canteen (indoor) 512 Korean BBQ-style 1.92 restaurant (outdoor) 780 1.21 Chinese hot-pot restaurant (outdoor) 4.2 Chinese dim sum 1.26 restaurant (outdoor) 435 1.13 Western canteen (outdoor) Chinese restaurant (Hunan cooking style emissions) Chinese restaurant (Cantonese cooking style emissions) Chinese restaurant (ambient concentrations) Average of different kitchen environments Average of different Living room environments 13,100 Indoor with traditional stoves Outdoor 1800 Indoor with improved stoves Outdoor Personal (using traditional stoves)a Personal (using improved stoves)a

China (He et al. [49])

Bangladesh (Beghum et al. [50])

Honduras (Clark et al. [51])

PM10 (ppb)

PM2.5 (ppb)

HCHO THC Benzene (ppb) (ppm) (ppb)

Toluene Methyl (ppb) Chloride (ppb)

Chloroform (ppb)

1.44

1.17

0.18

11.40

0.02

0.16

0.001

0.015

0.11

0.08

0.04

8.50

0.01

0.09

0.020

0.010

0.03

0.03

0.02

5.60

0.01

0.08

0.007

0.003

0.04

0.02

0.02

4.00

0.00

0.02

0.001

0.001

0.08

0.06

0.13

6.78

0.01

0.21

0.011

0.004

0.08

0.07

0.06

6.50

0.00

0.03

0.002

0.003

0.08

0.06

0.02

5.10

0.01

0.04

0.014

0.002

0.10

0.07

0.02

4.10

0.01

0.04

0.001

0.001

1.41

0.67

n-Fatty acids (ppm)

Dicarboxylic acids (ppm)

PAH (ppb)

287

8.15

25.0

250

5.14

41.7

357

0.04

65.9

129

97.6

0.31

0.60 0.19

n-Alkanes Total (ppb) carbonyls (ppb)

K.-H. Kim et al. / Journal of Hazardous Materials 195 (2011) 1–10

Country

1.00 0.36 0.27 0.22 0.14 0.07

3

4

Table 2 (Continued) Environment/ emission source

Hongkong (Ho et al. [52])

General Chinese Restaurant Large Medium 1 Medium 2 Small Sinchuan Spicy Food Restaurant Hongkong style Fast food Demonstration Kitchen Single dish 1 Single dish 2 Chinese barbeque kitchen A B C Korean BBQ Western fast-food chain stops A B Western small fast-food chain stops A B Western restaurant

a

CO2 CO (ppm) (ppb)

PM10 (ppb)

PM2.5 (ppb)

HCHO THC Benzene (ppb) (ppm) (ppb)

Toluene Methyl (ppb) Chloride (ppb)

Chloroform (ppb)

n-Fatty acids (ppm)

Honduras. Personal PM2.5 was assessed by attaching the sampler to the participant’s clothing nearest breathing zone and placing the pump in a pack worn around waist.

Dicarboxylic acids (ppm)

PAH (ppb)

n-Alkanes Total (ppb) carbonyls (ppb)

831 96.6 277 289 715 152

226 81.8

179 188 414 473

762 350

113 149 160

K.-H. Kim et al. / Journal of Hazardous Materials 195 (2011) 1–10

Country

K.-H. Kim et al. / Journal of Hazardous Materials 195 (2011) 1–10

5

Table 3 Differences in pollutant emissions between different oil types under varying cooking conditions (mean: mg h−1 L−1 ). Ordera

Oil type/treatments

1

24.7 Canola oil (heated at 180 ◦ C 33 50.5 for 15 h) 25.1 Extra virgin olive oil 46.7 8.6 (heated at 180 ◦ C for 15 h) 30.2 51 7.41 Olive oil (heated at 180 ◦ C for 15 h) ◦ 38.5 Canola oil (heated at 240 C 84.7 51.9 for 7 h) 64.5 124.3 7.09 Extra virgin olive oil (heated at 240 ◦ C for 7 h) 69.2 138.1 7.61 olive oil (heated at 240 ◦ C for 7 h) Safﬂower oil heated at 210 ◦ C for different intervals (h) 0 317 653 28.7 1 408 947 48.1 349 949 43.7 2 365 489 45.4 3 400 447 45 4 5 375 635 43.3 6 369 611 42.4 Safﬂower oil heated for 6 h at different temperatures (◦ C) 388 799 47.5 180 369.36 611.1 42.4 210 1658.5 969.2 25.2 240 1371.4 1942.4 80.7 270 ◦ Coconut oil heated for 6 h at different temperatures ( C) 180 370.1 198.7 15.61 210 951.3 255.4 13.88 240 1853.5 493 34.6 3429.6 420.7 18.38 270 Canola oil heated for 6 h at different temperatures (◦ C) 303.07 304 110.7 180 607.6 158.2 210 335.4 240 339.07 806.1 195.3 270 1691.1 1549.1 419 Extra virgin olive oil heated for 6 h at different temperatures (◦ C) 481.1 310.8 27.3 180 210 855.6 939.1 162.5 240 1288.8 1562.5 637.2 270 1392.1 1699.7 736

2

3

4

5

a

Total akanals

Total alkenals (mg h−1 L−1 )

Total alkadienals (mg h−1 L−1 )

Total aldehydes (mg h−1 L−1 )

Total oleic acid derivatives (mg h−1 L−1 )

Total linoleic acid derivatives (mg h−1 L−1 )

Total linolic acid derivatives (mg h−1 L−1 )

108.1

46.3

26.5

34.3

80.4

61.4

13.5

4.1

88.5

70.7

13.8

2.48

175

112

30.7

31.7

195.9

171.1

18.8

2.73

214.9

156.2

44

2.81

999 1403 886 900 892 1053 1022 1235 1023 2653 3395 584 1221 2381 3869 718 1101 1340 3659 819 1957 3489 3828

Source of the data: For order 1 (Fullana et al. [53]) and orders 2 through 5 (Katragadda et al. [54]).

In 2009, these estimates ranged from 70 (for coffee roasting) to 10,400 tonnes (animal feed production). The US EPA [26] estimated emissions from cooking beef and chicken by street vending cooking devices (charcoal grilling). It revealed that marinated meat resulted in higher pollutant emissions than non-marinated meat, while no signiﬁcant differences exist in emission strengths between meat types. It was also pointed that charcoal did not contribute signiﬁcantly to the pollutant emissions relative to the meat stuffs. In order to acquire a broader understanding of the emission proﬁle of different cooking activities, Roe et al. [27] developed a national emissions inventory for commercial cooking in the USA (Table 5). Apart from these examples, there are relatively few efforts to develop emission inventories of the range of cooking activities in domestic and commercial settings. As the extent and nature of cooking activities can vary considerably, there is a pressing need to establish emission inventories across a much wider range of geographies, cultures and cooking styles. Such efforts will help us assess both short and long-term health impacts of human exposure to cooking related pollution and accelerate the implementation of regulation to govern safe levels of emission (especially in commercial settings).

5. The level of cooking-related emission in relation to regulations and guidelines To protect the public from the possible health effects of cooking emissions, various regulations and guidelines have been issued by various authorities (Table 6). The pollutants measured from cooking fuels and food smoke were compared based on the literature survey (Table 1). As health criteria for IAP are generally limited, this focuses on CO, BTX, and formaldehyde. Note that CO and xylene however did not exceed any of their regulations and guidelines (Table 6). Kabir et al. [17] reported levels of toluene (625 ␮g m−3 ) that exceeded the chronic-duration inhalation MRL (300 ␮g m−3 ) and the EPA reference air concentration (400 ␮g m−3 ). Their benzene data (315 ␮g m−3 ) likewise exceeded the chronic (10 ␮g m−3 ), intermediate (20 ␮g m−3 ), and acute (30 ␮g m−3 )-duration inhalation MRLs set by ATSDR, and the 30 ␮g m−3 reference air concentration set by EPA. Similarly, formaldehyde levels greatly exceeded the chronic, intermediate, and acute-duration inhalation MRL set by ATSDR, the REL (8-h TWA), and the 15-min ceiling limit set by NIOSH. Aside from charcoal, other cooking fuels can yield considerable emissions. Khalequzzaman et al. [16] reported 17.2 ␮g m−3 of

6

K.-H. Kim et al. / Journal of Hazardous Materials 195 (2011) 1–10

Table 4 Comparison of pollutant emissions between different food/cooking methods. [A] Microwaving popcorn (Rosati et al. [55]) Concentration range in chamber (ng mL−1 )

Compound name Butyric acid Diacetyl Acetoin Propylene glycol 2-Nonanone Triacetin Acetic acid 2-Butoxy-1-methyl-2-oxoethyl ester butanoic acid p-Xylene Pentanal Toluene Hexanal 2-Methyl propanoic acid 2-Octanone Heptanal Benzaldehyde 2-(2-Hydroxypropoxy) 1-propanol Acetophenone Siloxanes 2-Tridecanone 3-Methyl butanal 2-Methyl butanal Furfural 4-Methyl-3-penten-2-one 2-Pentyl furan 2-(2-Ethoxyethoxy) ethanol 2-Ethyl 1-hexanol 3-Hexanone Ethyl ester butanoic acid Butyl ester 2-propenoic acid 2,3-Butanedioldiacetate Cyclotetrasiloxane Decamethyl cyclopentasiloxane Octanoic acid Dodecamethyl cyclohexasiloxane Dodecamethyl pentasiloxane Dihydro-5-pentyl-2(3H)-furanone Octanal Styrene 1-Ethoxy-2-methyl propane Methyl ester octanoic acid Ethyl ester octanoic acid Tridecane 2-(Perﬂuorooctyl)ethanol 8:2-telomer

0.1–8.6 0.02–5.8 0.01–4.2 0.005–1.3 0.005–1.4 0.01–1.2 0.005–0.5 0.005–0.7 0.01–0.4 0.01–0.02 0.01–0.04 0.01–0.05 0.01–0.27 0.01–1.28 0.01–0.02 0.01–0.02 0.01–0.5 0.015–0.01 0.01–0.03 0.01–0.16 0.01–0.01 0.01–0.03 0.01–0.37 0.01–1.20 0.01–0.01 0.01–0.3 0.01–0.06 0.01–0.17 0.01–0.05 0.01–0.04 0.01–0.33 0.01–0.09 0.01–0.02 0.01–0.16 0.01–0.05 0.01–0.03 0.01–0.08 0.015–0.01 0.01–0.02 0.01–0.02 0.01–0.01 0.01–0.05 0.01–0.05 0.0005–0.009

[B] Emission from combination of food and cooking style (all concentration in ppb: Kabir et al. [2]) Compound

Steamed cabbage

Boiled clam

Hydrogen sulﬁde Methane thiol Dimethyl sulﬁde Dimethyl sulﬁde Acetaldehyde Propionaldehyde Butyraldehyde Isovaleraldehyde Styrene Toluene para-xylene Methylethyl ketone Methylisobutylketone Butylacetate Isobutylalcohol Propionic acid Butyric acid Isovaleric acid Valeraldehyde

0.86 0.15 9.44 1.2 12 0.39 0.39 0.44 0.37 26.3 1.62 3.21 0.04 0.44 0.09 2.27 0.06 3.46 0.06

0.2 0.15 0.26 0.06 18.7 2.81 0.39 0.44 0.31 19.8 1.51 5.45 0.48 0.04 0.09 2.5 0.2 5.75 0.06

Brewed coffee 0.2 13.5 16.9 4.32 153 31.8 77.6 0.44 0.36 24 1.95 52.6 0.04 0.04 3.08 5.84 0.06 15.9 0.06

Fried cabbage

Grilled clam

0.2 63.8 25.6 9.34 12.5 5.4 15.3 0.44 0.07 51.2 1.57 3.21 0.04 0.04 0.09 4.39 0.06 0.05 0.14

39.6 0.15 31.3 35.5 253 8.65 12.9 0.44 0.2 51.1 1.99 28.2 0.04 0.04 3.91 36.1 5.11 1.97 0.12

Roasted coffee 2398 2070 98.7 24.5 5233 366 458 600 8.36 123 0.03 964 0.04 0.04 0.09 695 67 132 8.39

K.-H. Kim et al. / Journal of Hazardous Materials 195 (2011) 1–10

7

Table 5 Emission inventory of air pollutants released from different cooking activities. [A] Emission factors considering a mass balance approach Emission rate (␮g kg−1 )

Cooking activity

Meat charbroiling

Reference

Air pollutant category

Gas phase

Alkanes Oleﬁns Carbonyls Aromatics and napthenes Unidentiﬁed organic compounds Aliphatic aldehydes Ketones Alkanoic acids Alkenoic acids Unresolved mixture

1,470,000 2,450,000 5,480,000 200,000 4,590,000

Alkanes

15.8

Saturated n-aldehydes Ketones n-Alkanoic acids Others Oleﬁnic n-aldehydes n-Alkenoic acids Non-extractable organics

138.6 12.9 25.2 25.6

Particle phase [23]

260,000 220,000 480,000 320,000 1,300,000

Stir frying vegetables 0.96

[21]

4.9 0.89 0.72 7.98 0.34

[B] Total emission rate estimated for USA (based on Roe et al. [27]). Air pollutant category

Emission rate (tonnes year−1 ) Conveyorized charbroiling

VOCs PAHs CO PM10 PM2.5

2113 43 7401 8460 8201

Under-ﬁred charbroiling 7234 122 23,662 60,304 58,295

formaldehyde from Bangladeshi cooking fuels which slightly exceeded its chronic-duration inhalation MRL (10 ␮g m−3 ). The same research team measured 13.7 ␮g m−3 of benzene from natural gas which exceeded the 10 ␮g m−3 chronic-duration inhalation MRL set by ATSDR. They also measured 54.2 ␮g m−3 of benzene from biomass which exceeded the chronic, intermediate, and acute-duration inhalation MRLs and the reference air concentration set by the EPA. According to Qing [34], very high levels of SO2 from natural gas (185 ␮g m−3 ) and coal (436 ␮g m−3 ) exceeded not only the acute-duration inhalation MRL (30 ␮g m−3 ) but the air quality guidelines of the WHO (40–60 ␮g m−3 ) and NAAQS (80 ␮g m−3 ). The SO2 emission from coal (436 ␮g m−3 ) further exceeded the NAAQS (24-h exposure limit: 365 ␮g m−3 ) of the EPA and 1-h exposure limit (350 ␮g m−3 ) set by the WHO. Its emission from biomass fuel (61.5 ␮g m−3 ) [35] also exceeded the acute-duration inhalation MRL (40–60 ␮g m−3 ) of the WHO. In addition, acrolein levels reported in Section 3 commonly exceeded many exposure guidelines. Its indoor concentrations (26.4–64.5 ␮g m−3 ) during cooking exceeded not only the inhalation reference concentration (RfC: 0.02 ␮g m−3 ) of the EPA but also the intermediate (0.09 ␮g m−3 ) and acute-duration inhalation MRLs (6.88 ␮g m−3 ) [25]. Considering the frequent exceedance of the IAP due to cooking, one could easily extrapolate its effect on human health. In this regard, it is worth assessing the carcinogenic potentials of the pollutants discussed above. It should be noted that benzene is a known human carcinogen for all routes of exposure based on convincing evidence from both human and animal studies by IARC, EPA, and NTP. Furthermore, formaldehyde has been classiﬁed as a probable human carcinogen based on limited (human) and sufﬁcient (animal) evidence [36]. As such, formaldehyde (and toluene) are regulated as hazardous air pollutants (HAPs) by the U.S. Congress [37] and are subject to the regulations for various manufacturing

Deep fat frying 1173

Flat griddle frying 39 41 1941 15,679 11,916

Clamshell griddle frying 940

1073 909

processes and operations [38]. However, as to the carcinogenicity, various regulatory agencies have not yet ﬁrmly assigned cancer classiﬁcations for xylene, toluene, CO, SO2 , and acrolein or assessed their carcinogenic potential due to inadequacy of data or evidence. 6. Health and environmental impacts of cooking activities There is a line of evidence that cooking related emissions can cause severe health problems. For instance, Yang et al. [39] demonstrated that cooking fume is a major cause of lung cancer in Chinese women. Based on an epidemiological study, Yu et al. [40] also concluded that cumulative exposure to cooking emissions by means of any form of frying could increase the risk of lung cancer for nonsmoking women in Hong Kong. Despite a low smoking rate, these subjects recorded one of the highest non-smoking lung cancer rates worldwide which was ascribed to cumulative exposure to cooking fume rather than to the peak concentrations experienced during cooking [40]. Furthermore the risk of active tuberculosis increased in Indians (aged 20 years and older) cooking with biomass fuel relative to cleaner fuels [41]. This estimate is comparable to the report made by WHO [42] based on non-clinical measures [43]. In addition, chronic exposure to biomass fuel combustion products was also suspected to cause chromosomal and DNA damage and upregulation of DNA repair mechanisms in premenopausal women in rural areas [44]. Evidence also indicates an etiological link between indoor coal burning and lung cancer. For instance, high lung cancer rates in Chinese women were closely associated with the combustion of smoky coal emitting submicron particles with mutagenic organics, especially aromatic and polar compounds [45]. There have been many attempts to estimate the global burden of disease due to the use of solid fuels by applying disease speciﬁc

8

K.-H. Kim et al. / Journal of Hazardous Materials 195 (2011) 1–10

Table 6 International and national regulations, advisories, and guidelines issued by various agencies and published by the Agency for Toxic Substances and Diseases Registry (ATSDR).a Agency

Description

International guidelines (Air) ATSDR Acute-duration MRL Intermediate-duration MRL Chronic-duration MRL Cancer classiﬁcationb IARC National regulations and guidelines (Air) TLV – 8 h TWA ACGIH STEL Ceiling Limit for Occupation Exposure (TLV-STEL) Cancer classiﬁcationc Hazardous Air Pollutants EPA Cancer classiﬁcationd Inhalation reference concentration OSHA PEL (8-h TWA) for general, construction, and shipyard industries 15-min STEL Acceptable ceiling concentration Acceptable max. peak above the acceptable ceiling conc. for an 8-h shift for a max. duration of 10 min NIOSH REL (8-h TWA) REL (10-h TWA) REL (15-min ceiling) IDLH STEL NTP Cancer classiﬁcatione Agency

Formaldehyde (Ref. [28]) (mg m−3 )

Benzene (Ref. [29]) (mg m−3 )

Toluene (Ref. [30]) (mg m−3 )

0.05 0.04 0.01 Group 2A

0.03 0.02 0.01 Group 1

3.8

1.6f 7.99f

188

0.3 Group 3

0.37 Yes B1

Yes A 0.03 3.19

0.92

A4 Yes D 0.4 754

2.46 1130 1884 0.02

377 0.32g

0.12 24.6

1597g 3.19g A

B

Description

International guidelines Inhalation MRL ATSDR Acute-duration Intermediate-duration STEL (occupational exposure) Cancer classiﬁcationb IARC Air quality guidelines WHO 10-min exposure limit 1-h exposure limit 24-h exposure limit Annual arithmetic mean TWA based on effects other than cancer or odor/annoyance: 15 min-TWA 30 min-TWA 1 h-TWA 8 h-TWA National regulations and guidelines (Air) TLV (TWA) ACGIH TLV (ceiling limit) Carcinogenicity classiﬁcationc Hazardous Air Pollutants EPA Cancer classiﬁcationd Inhalation reference concentration National Ambient Air Quality Standards (NAAQS) 24-h exposure limit Annual arithmetic mean 3-h exposure limit 8-h averaging time 1-h averaging time OSHA PEL (8-h TWA) for general industry REL TWA NIOSH REL (10-h TWA) IDLH STEL Ceiling Cancer classiﬁcatione NTP

SO2 (Ref. [31]) (mg m−3 )

CO (Ref. [32]) (mg m−3 )

Acrolein (Ref. [33]) (mg m−3 )

0.003 4.0E−05

0.03 10 Group 3

565

No data

Group 3

0.5 0.35 0.10–0.15 0.04–0.06 100 60 30 10 5.2

29

No No No

0.23f A4 Yes ID 2.00E−05

10h 40h 55

0.23

0.365h 0.08 1.3h

13 5

40 1375 13

0.23 4.59 0.69

229 None

a Deﬁnitions: ACGIH = American Conference of Governmental Industrial Hygienists; EPA = Environmental Protection Agency; IARC = International Agency for Research on Cancer; IDLH = immediately dangerous to life or health; MRL = inhalation Minimum Risk Level; NIOSH = National Institute for Occupational Safety and Health; NTP = National Toxicology Program; OSHA = Occupational Safety and Health Administration; PEL = permissible exposure limit; REL = recommended exposure limit; STEL = short-term exposure limit; TLV = threshold limit values; TWA = time-weighted average; WHO = World Health Organization. b IARC cancer classiﬁcation: Group 1 (carcinogenic to humans); Group 2A (probable human carcinogen), Group 3 (Not classiﬁable as to carcinogenicity to humans). c ACGIH cancer classiﬁcations: A4 (not classiﬁable as a human carcinogen). d EPA cancer classiﬁcation: A (known human carcinogen); B1 (probable human carcinogen); D (substances are unclassiﬁable as to their carcinogenicity); ID (data are inadequate for an assessment of the carcinogenic potential). e NTP cancer classiﬁcations: A (substance known to be carcinogenic); B (reasonably anticipated to be a human carcinogen). f Refers to the potential signiﬁcant contribution to the overall exposure by the cutaneous route, including mucous membranes and the eyes, either by contact with vapors or, of probable greater signiﬁcance, by direct skin contact with the substance. g NIOSH potential occupational carcinogen. h Not to be exceeded more than once per year.

K.-H. Kim et al. / Journal of Hazardous Materials 195 (2011) 1–10

9

Fig. 1. Global distribution estimates of deaths caused by indoor smoke from solid fuels by WHO sub-region for 2000. Source: Ref. [47].

relative risks or odds ratios to global estimates of the household number relying on such fuels. IAP from solid fuel use and urban outdoor air pollution are in fact estimated to cover 3.1 million premature deaths worldwide every year and 3.2% of the global burden of disease expressed in disability-adjusted life years (DALYs) [46]. If this burden is evaluated on a regional basis, it varies signiﬁcantly due to many contributing factors (Fig. 1). For instance, almost 80% of ill health effects occur in Africa and Southeast Asia. Globally, indoor smoke from solid fuels ranks as the eighth risk factor. This rises to fourth (after (1) childhood and maternal underweight, (2) unsafe sex, and (3) unsafe water, sanitation, and hygiene) in developing countries (∼40% of the world population) [47]. Indoor smoke risks present with a high mortality thus ranks higher than micronutrient deﬁciencies and tobacco risks. Hence, to minimize the exposure to cooking related emissions, efforts should be directed to improve cooking devices, development of alternate energy sources (such as sun light), living environments, and cooking behavior. It is important to note that cooking emissions can also exert inﬂuences on climate change. For instance, solid fuel dependency exacerbates deforestation which indirectly contributes to the build-up of greenhouse gases (e.g., CO2 ). Deforestation can also cause soil erosion, pollution of streams with sediment and debris, loss of biodiversity, and alteration of vector-borne disease transmission patterns [48]. Moreover, these emissions themselves contain a wide array of pollutants that can contribute to the global climate change.

understanding of the toxic effects of IAP is critical in establishing the safety of indoor air. It is an interesting modern paradox that most legislation in developed countries governs the allowed concentration of HAPs in ambient outdoor air, and yet the citizens which this legislation is designed to protect spend the majority of their time in indoor locations, especially at home, where the concentrations of these regulated pollutants are often much higher. While it is unlikely that regulatory air quality legislation will ever penetrate the threshold of the domestic private home, it seems increasingly important to educate the public of the potential dangers of IAP so that they may take informed choices about their behaviors, and if desired install abatement and extraction technologies to improve their IAQ. Despite the recognition of the potent role of cooking emission, it has scarcely been investigated in relation to possible human health impact. Most studies were directed to emission estimates for selected cooking procedures under certain conditions, while epidemiological studies are lacking for particular cooking procedures. Future research should thus be directed towards a comprehensive survey of the most common cooking procedures with respect to HAPs emission and their direct or indirect impacts on human health and the surrounding atmosphere. Acknowledgements This work was supported by a Grant from the National Research Foundation (NRF) of Korea funded by the Ministry of Education, Science and Technology (MEST) (No. 2010-0007876).

7. Future directions The results of our review conﬁrm that a wide array of HAPs is released during food preparation using common solid fuels and the main materials (and ingredients) for cooking. The level of pollutants released via such activities can pose serious threats to human health, especially to those performing the cooking and their household members. The health effects of such exposure are not restricted to the respiratory tract but can readily cross the alveolar–capillary barrier to reach vital body organs through the circulatory system. It is thus important to monitor IAP in residential settings, restaurant kitchens, and dining areas. In addition, a better

References [1] K. Skog, Cooking procedures and food mutagens: a literature review, Food Chem. Toxicol. 31 (1993) 655–675. [2] E. Kabir, K.-H. Kim, An investigation on hazardous and odorous pollutants emission during cooking activities, J. Hazard. Mater. 188 (2011) 443–454. [3] Q. Zhang, R.H. Gangupomu, D. Ramirez, Y. Zhu, Measurement of ultraﬁne particles and other air pollutants emitted by cooking activities, Int. J. Environ. Res. Public Health 7 (2010) 1744–1759. [4] D. Behera, G. Aggarwal, Domestic cooking fuel exposure and tuberculosis in Indian women, Indian J. Chest Dis. Allied Sci. 52 (2010) 137–138. [5] A.K.N. Reddy, R.H. Williams, T.B. Johansson, Energy after Rio: Prospects and Challenges, United Nations Publications, New York, 1996.

10

K.-H. Kim et al. / Journal of Hazardous Materials 195 (2011) 1–10

[6] N. Bruce, R. Perez-Padilla, R. Albalak, Indoor air pollution in developing countries: a major environmental and public health challenge, Bull. WHO 98 (2008) 1078–1092. [7] K.-H. Kim, S.A. Jahan, E. Kabir, A review of diseases associated with household air pollution due to the use of biomass fuels, J. Hazard. Mater. 19 (2011) 425–431. [8] K.R. Smith, J.M. Samet, I. Romieu, N. Bruce, Indoor air pollution in developing countries and acute respiratory infections in children, Thorax 55 (2000) 518–532. [9] A.J.S. Jan, Risk assessment of acute health and comfort effects of indoor air pollution, Ann. N. Y. Acad. Sci. 641 (1992) 56–61. [10] D. Schwela, Cooking smoke: a silent killer, People Planet 6 (1997) 24–25. [11] R. Kumar, J.K. Nagar, N. Raj, P. Kumar, A.S. Kushwah, M. Meena, S.N. Gaur, Impact of domestic air pollution from cooking fuel on respiratory allergies in children in India, Asian Pac. J. Allergy Immunol. 26 (2008) 213–222. [12] M.A.K. Lodhi, A. Zain-al-Abdin, Indoor air pollutants produced from fossil fuel and biomass, Energy Convers. Manage. 40 (1999) 243–248. [13] M. Ezzati, D.M. Kammen, Indoor air pollution from biomass combustion and acute respiratory infections in Kenya: an exposure-response study, Lancet 358 (2001) 619–624. [14] A. Ellegard, Cooking fuel smoke and respiratory symptoms among women of low income areas of Mapato, Environ. Health Prospect. 104 (1996) 980–985. [15] R. Albalak, G.J. Keeler, A.R. Frisancho, M. Haber, Assessment of PM10 concentrations from domestic biomass fuel combustion in rural Bolivian highland villages, Environ. Sci. Technol. 33 (1999) 2505–2509. [16] M. Khalequzzaman, M. Kamijima, S. Kiyoshi, B.A. Hoque, Indoor air pollution and the health of children in biomass- and fossil-fuel users of Bangladesh: situation in two different seasons, Environ. Health Prev. Med. 15 (2010) 236–243. [17] E. Kabir, K.-H. Kim, J.W. Ahn, O.F. Hong, J.R. Sohn, Barbecue charcoal combustion as a potential source of aromatic volatile organic compounds and carbonyls, J. Hazard. Mater. 174 (2010) 492–499. [18] J. Susaya, K.-H. Kim, J.-W. Ahn, M.-C. Jung, C.-H. Kang, BBQ charcoal combustion as an important source of trace metal exposure to humans, J. Hazard. Mater. 176 (2010) 932–937. [19] K.R. Smith, Indoor air pollution in developing countries: recommendations for research, Indoor Air 12 (2002) 1–7. [20] S.C. Lee, W.M. Li, L.Y. Chan, Indoor air quality at restaurants with different styles of cooking in metropolitan Hong Kong, Sci. Total. Environ. 279 (2001) 181–193. [21] J.J. Schauer, M.J. Kleeman, G.R. Cass, B.R.T. Simoneit, Measurement of emissions from air pollution sources: 4. C1–C27 organic compounds from cooking with seed oils, Environ. Sci. Technol. 36 (2002) 567–575. [22] S.O. Baek, Y.S. Kim, R. Perry, Indoor air quality in homes, ofﬁces and restaurants in Korean urban areas – indoor/outdoor relationships, Atmos. Environ. 31 (1997) 529–544. [23] J.J. Schauer, M.J. Kleeman, G.R. Cass, B.R.T. Simoneit, Measurement of emissions from air pollution sources: 1. C1 through C29 Organic compounds from meat charbroiling, Environ. Sci. Technol. 33 (1999) 1566–1577. [24] A. Fullana, A.A. Carbonell-Barrachina, S. Sidhu, Comparison of volatile aldehydes present in the cooking fumes of extra virgin olive, olive, and canola oils, J. Agric. Food Chem. 52 (2004) 5207–5214. [25] V.Y. Seaman, D.H. Bennett, T.M. Cahill, Indoor acrolein emission and decay rates resulting from domestic cooking events, Atmos. Environ. 43 (2009) 6199–6204. [26] United States Environmental Protection Agency (US EPA), Emissions from street vendor cooking devices (charcoal grilling) EPA/600/Sr-99/048, June 1999. [27] S.M. Roe, M.D. Spivey, H.C. Lindquist, P. Hemmer, R. Huntley, National Emissions inventory for commercial cooking, 2004. Available at: www.p2pays.org/ref/43/42612.pdf. [28] Agency of Toxic Substance and Disease Registry (ATSDR), Toxicological proﬁle for formaldehyde, US DHHS, Public Health Service, 1999. [29] Agency of Toxic Substance and Disease Registry (ATSDR), Toxicological proﬁle for benzene, US DHHS, Public Health Service, 2007. [30] Agency of Toxic Substance and Disease Registry (ATSDR), Toxicological proﬁle for toluene, US DHHS, Public Health Service, 2000. [31] Agency of Toxic Substance and Disease Registry (ATSDR), Toxicological proﬁle for sulfur dioxide, U.S. Department of Health and Human Services (US DHHS), Public Health Service, 1998.

[32] Agency of Toxic Substance and Disease Registry (ATSDR), Draft toxicological proﬁle for carbon monoxide, US DHHS, Public Health Service, 2009. [33] Agency of Toxic Substance and Disease Registry (ATSDR), Toxicological proﬁle for Acrolein, US DHHS, Public Health Service, 2007. [34] Y.H. Qing, Effect of indoor air pollution on respiratory diseases in children, in: Proceeding (in Chinese) of China–Japan environment health conference, Zhengzhou, China, 1992, pp. 53–55. [35] B.K. Padhi, P.K. Padhy, Domestic fuels, indoor air pollution, and children’s health, Ann. N. Y. Acad. Sci. 1140 (2008) 209–217. [36] IRIS, Online. U.S. Environmental Protection Agency, Ofﬁce of Health and Environmental Assessment, Environmental Criteria and Assessment Ofﬁce, Cincinnati, OH, April 1999. Available at: http://www.epa.gov/ngispgm3/iris/subst. [37] U.S. Congress, Clean air act amendments, Hazardous air pollutants, Section 112, national emissions standards for hazardous air pollutants, Public Law 101-549, U.S. Congress. November 15, 1990. [38] US EPA, Standards of performance for volatile organic compounds (VOC) emissions from synthetic organic chemical manufacturing industry (SOCMI) distillation operations, U.S. EPA, Code of Federal Regulations, 40 CFR 60, Subpart NNN, 1995. [39] S.C. Yang, S.N. Jenq, Z.C. Kang, H. Lee, Identiﬁcation of Benzo[a]pyrene 7,8Diol 9,10-Epoxide N2-Deoxyguanosine in human lung adenocarcinoma cells exposed to cooking oil fumes from fFrying ﬁsh under domestic conditions, Chem. Res. Toxicol. 13 (2000) 1046–1050. [40] I.T.S. Yu, Y.L. Chiu, J.S.K. Au, T.W. Wong, J.L. Tang, Dose–response relationship between cooking fumes exposures and lung cancer among Chinese nonsmoking women, Cancer Res. 66 (2006) 4961–4967. [41] V.K. Mishra, R.D. Retherford, K.R. Smith, Biomass cooking fuels and prevalence of tuberculosis in India, Int. J. Infect. Dis. 3 (1999) 119–129. [42] WHO, Tuberculosis Programme Review, World Health Organization Geneva, India, 1992. [43] B.N. Gupta, N. Mathur, P.N. Mahendra, A.K. Srivastava, A study of the household environmental risk factors pertaining to respiratory diseases, Energy Environ. Rev. 13 (1997) 61–67. [44] N.K. Mondal, B. Mukherjee, D. Das, M.R. Ray, Micronucleus formation, DNA damage and repair in premenopausal women chronically exposed to high level of indoor air pollution from biomass fuel use in rural India, Mutat. Res. 697 (2010) 47–54. [45] IARC, Environmental carcinogens methods of analysis and exposure measurement, Indoor air, Vol. 12, IARC Scientiﬁc Publications No. 109, Lyon, France, 1993. [46] World Health Organization (WHO), Global Health Risks: Mortality and Burden of Diseases Attributable to Selected Major Risks, Geneva, 2009. Available at: http://www.who.int/healthinfo/global burden disease/GlobalHealthRisks report full.pdf. [47] World Health Organization (WHO), World Health Report, 2002. Available at: http://www.who.int/whr/en/. [48] Emerging challenges – new ﬁndings: emerging and re-emerging infectious diseases: links to environmental change, in: P. Harrison (Ed.), GEO Yearbook 2004/5: An Overview of our Changing Environment, United Nations Environment Programme, Nairobi, 2005. [49] L.-Y. He, M. Hu, X.-F. Huang, B.-D. Yu, Y.-H. Zhang, D.-Q. Liu, Measurement of emissions of ﬁne particulate organic matter from Chinese cooking, Atmos. Environ. 38 (2004) 6557–6564. [50] B.A. Begum, S.K. Paul, M.D. Hossain, S.K. Biswas, P.K. Hopke, Indoor air pollution from particulate matter emissions in different households in rural areas of Bangladesh, Build. Environ. 44 (2009) 898–903. [51] M.L. Clark, S.J. Reynolds, J.B. Burch, S. Conway, A.M. Bachand, J.L. Peel, Indoor air pollution, cookstove quality, and housing characteristics in two Honduran communities, Environ. Res. 110 (2010) 12–18. [52] S.S.H. Ho, J.Z. Yu, K.W. Chu, L.L. Yeung, Carbonyl emissions from commercial cooking sources in Hong Kong, J. Air Waste Manage. Assoc. 56 (2006) 1091–1098. [53] A. Fullana, A.A. Carbonell-Barrachina, S. Sidhu, Volatile aldehyde emissions from heated cooking oils, J. Sci. Food Agric. 84 (2004) 2015–2021. [54] H.R. Katragadda, A. Fullana, S. Sidhu, Á. Carbonell-Barrachina, Emissions of volatile aldehydes from heated cooking oils, Food Chem. 120 (2010) 59–65. [55] J.A. Rosati, K.A. Krebs, X. Liu, Emissions from cooking microwave popcorn, Crit. Rev. Food Sci. Nutr. 47 (8) (2007) 701–709.

Journal of Hazardous Materials 195 (2011) 11–29

Contents lists available at SciVerse ScienceDirect

Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat

Review

A review on techniques to enhance electrochemical remediation of contaminated soils Albert T. Yeung a,∗ , Ying-Ying Gu b,1 a

Department of Civil Engineering, The University of Hong Kong, Pokfulam Road, Hong Kong Department of Environmental Science & Engineering, College of Chemical Engineering, China University of Petroleum (East China), 66 West Changjiang Road, Qingdao 266555, People’s Republic of China b

a r t i c l e

i n f o

Article history: Received 20 June 2011 Received in revised form 15 August 2011 Accepted 15 August 2011 Available online 22 August 2011 Keywords: Electrochemical remediation Soil remediation Enhancement techniques Contaminant solubilization Soil pH control Coupling of remediation technologies

a b s t r a c t Electrochemical remediation is a promising remediation technology for soils contaminated with inorganic, organic, and mixed contaminants. A direct-current electric ﬁeld is imposed on the contaminated soil to extract the contaminants by the combined mechanisms of electroosmosis, electromigration, and/or electrophoresis. The technology is particularly effective in ﬁne-grained soils of low hydraulic conductivity and large speciﬁc surface area. However, the effectiveness of the technology may be diminished by sorption of contaminants on soil particle surfaces and various effects induced by the hydrogen ions and hydroxide ions generated at the electrodes. Various enhancement techniques have been developed to tackle these diminishing effects. A comprehensive review of these techniques is given in this paper with a view to providing useful information to researchers and practitioners in this ﬁeld. © 2011 Elsevier B.V. All rights reserved.

Contents 1. 2. 3.

4.

5.

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Classiﬁcation of enhancement techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Techniques to solubilize contaminants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1. Lowering of soil pH . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2. Introduction of enhancement agents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2.1. Chelants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2.2. Complexing agents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2.3. Surfactants and cosolvents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2.4. Oxidizing/reducing agents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2.5. Cation solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Soil pH control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1. Electrode conditioning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2. Use of ion exchange membrane . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Coupling with other remediation technologies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.1. Oxidation/reduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2. Bioremediation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.3. Permeable reactive barriers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.3.1. Lasagna process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.3.2. Zero-valent iron (ZVI) PRB . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.3.3. PRBs of different reactive media . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

∗ Corresponding author. Tel.: +852 28598018; fax: +852 25595337. E-mail addresses: [email protected] (A.T. Yeung), [email protected] (Y.-Y. Gu). 1 Tel.: +86 532 86984668; fax: +86 532 86984668. 0304-3894/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.jhazmat.2011.08.047

12 12 12 12 13 13 14 14 15 15 16 16 18 18 18 20 21 21 22 23

12

6.

A.T. Yeung, Y.-Y. Gu / Journal of Hazardous Materials 195 (2011) 11–29

5.3.4. PRB – summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.4. Phytoremediation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.5. Ultrasonication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.6. Other remediation technologies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1. Introduction Electrochemical remediation is a promising technology to remediate ﬁne-grained soils contaminated by inorganic, organic, and mixed contaminants. A direct-current (dc) electric ﬁeld is imposed on the contaminated soil. The contaminants are migrated by the combined mechanisms of electroosmosis, electromigration, and/or electrophoresis. Detailed description of these fundamental electrokinetic phenomena in soil is given by Yeung [1] and Yeung and Gu [2]. As a dc electric ﬁeld is a much more effective force in driving ﬂuid through ﬁne-grained soils than a hydraulic gradient [3], electrochemical remediation is particularly applicable to ﬁne-grained soils of low hydraulic conductivity and large speciﬁc area. Milestone developments and future research directions of the technology are given in Yeung [4]. However, as the soilchemical ﬂuid system is an electrochemical system [5], many electrochemical reactions are occurring simultaneously during electrochemical remediation of contaminated soil [6]. Moreover, the large speciﬁc area of the ﬁne-grained soil provides numerous sites for soil-contaminant interactions. These interactions are soil speciﬁc, contaminant speciﬁc, dynamic, reversible, and pHdependent. The coupling of electrochemical reactions with the soil-contaminant interactions makes the electrochemical remediation process extremely complex. Similar to most remediation technologies, electrochemical remediation can only extract mobile contaminants from soil [7,8]. Contaminants can exist as sorbed species on soil particle surfaces, sorbed species on colloidal particulates suspended in soil pore ﬂuid, dissolved species in soil pore ﬂuid, or solid species as precipitates. Only contaminants exist as dissolved species in the soil pore ﬂuid or sorbed species on colloidal particulates suspended in soil pore ﬂuid can be extracted by most remediation technologies, and electrochemical remediation is no exception [7]. Therefore, enhancement techniques are developed to solubilize contaminants in soil and to keep them in a mobile chemical state. Electrolytic decomposition of electrolytes occurs at the electrodes, generating H+ ions at the anode (the positive electrode) and OH− ions at the cathode (the negative electrode). These ions are migrated into the contaminated soil, resulting in changes in soil pH as a function of time and space. The change in soil pH can change the chemical states of contaminants, rendering them immobile. It can also change the magnitude and direction of electroosmotic ﬂow, affecting the advective transport of contaminants in soil pore ﬂuid by electroosmosis. Moreover, these ions can polarize the electrodes and reduce the effectiveness of the dc electric ﬁeld imposed. Therefore, controlling soil pH is very important for the success of electrochemical remediation. In many cases, application of electrochemical remediation alone is not adequate to remediate the contaminated soil to the required acceptance level. Therefore, the technology is enhanced by coupling with other remediation technologies as part of a remediation train of processes. The synergy can achieve results that are better than the sum of technologies applied individually. Many techniques to enhance the extraction efﬁciency of electrochemical remediation of contaminated soil have been developed throughout the years. A comprehensive review of these techniques

24 24 25 25 26 26 26

is given in this paper to facilitate effective applications of these enhancement techniques by researchers and practitioners in the ﬁeld of electrochemical remediation of contaminated ﬁne-grained materials such as clay, sediment, and sludge. 2. Classiﬁcation of enhancement techniques The primary objectives of these enhancement techniques are: (1) to solubilize contaminants in soil and to keep them in mobile states; (2) to control the soil pH within a range of values favoring the application of electrochemical remediation; and (3) to destruct, breakdown, or transform the contaminants simultaneously or sequentially. Therefore, the enhancement techniques are broadly classiﬁed into three groups: (1) techniques that solubilize contaminants and keep them in mobile states; (2) techniques that control soil pH; and (3) remediation techniques that can be coupled with electrochemical remediation synergistically to destruct, breakdown, or transform the contaminants simultaneously or sequentially. However, these three groups of techniques are inter-related. Detailed classiﬁcation of these techniques is presented in Fig. 1. 3. Techniques to solubilize contaminants Contaminants in soil can be sorbed on soil particle surfaces or exist as precipitates in soil pores under certain environmental conditions, rendering them immobile. These contaminants may go into dissolved phases again when the environmental conditions change. Therefore, the temporary immobility of contaminants cannot be considered as permanent containment. However, it does create a difﬁcult hurdle for the remediation process. Enhancement techniques have been developed to solubilize contaminants during electrochemical remediation including: (1) lowering of soil pH; and (2) introduction of enhancement agents. 3.1. Lowering of soil pH Most metals can be solubilized in a low pH environment. During the electrochemical remediation process, H+ ions are generated at the anode and migrated towards the cathode, an acid front is thus developed. A low pH environment can be generated in soil of low acid/base buffer capacity and extraction of metals can be achieved with a reasonable degree of success. For natural soils of high acid/base buffer capacity, strong acids and weak acids have been used as enhancement agents to neutralize the OH− ions generated at the cathode and to lower of the soil pH. Weak acids, such as acetic acid CH3 COOH and citric acid, can also serve as a complexing agent and a chelant, respectively. Strong acids are observed to be more effective than weak acids in many studies. However, it should be noted that when the soil pH is lower than the point of zero charge (PZC) [9], the direction of electroosmotic ﬂow is reversed, i.e., from the cathode towards the anode. The advective transport of contaminant by electroosmosis would diminish the electromigration of cations towards the cathode. Moreover, a very low pH environment developed during the

A.T. Yeung, Y.-Y. Gu / Journal of Hazardous Materials 195 (2011) 11–29

13

Fig. 1. Classiﬁcation of enhancement techniques for electrochemical remediation.

remediation process may impact the environment adversely and render the remediated soil not readily arable afterwards. More details on lowering of soil pH are presented in Section 4 later in this review paper. 3.2. Introduction of enhancement agents When the acid/base buffer capacity of soil is high, i.e., the resistance of soil to pH change is high, it is very difﬁcult to lower the soil pH by the H+ ions generated by electrolysis or introduction of acid to the soil. Therefore, other enhancement agents have to be utilized to desorb contaminants sorbed on soil particle surfaces and to keep them in the dissolved phase. These enhancement agents include: (1) chelants or chelating agents; (2) complexing agents; (3) surfactants and cosolvents; (4) oxidizing/reducing agents; and (5) cation solutions. 3.2.1. Chelants Chelation is the formation or presence of two or more separate bonds between a bi-dentate or multi-dentate ligand, i.e., the chelant, and a single metal central atom or ion. Chelants can thus desorb toxic metals from soil particle surfaces by forming strong water-soluble complexes which can be removed by the chelant-enhanced electrochemical remediation. An example on how ethylenediamine-N,N -disuccinic acid (EDDS), a biodegradable chelant, solubilizes sorbed Pb from soil particle surfaces is illustrated in Fig. 2 [10]. The chelant-enhanced electrochemical remediation is thus a four-step process: (1) injection of the chelant into the contaminated soil by electroosmosis and/or electromigration; (2) formation of soluble Pb–EDDS complex on soil particle surfaces; (2) dislodgement of Pb–EDDS complex from soil particle surfaces to soil pore ﬂuid; and (3) extraction of Pb as Pb–EDDS complex by electroosmosis and/or electromigration. Chelants, such as carboxylates, organophosphonates, polyamines, and industrial wastewaters [11], have been used or investigated as enhancement agents in electrochemical remediation. Among all the chelating agents, aminopolycarboxylates, such as ethylenediaminetetraacetic acid (EDTA) and (diethylenetriamine)pentaacetic acid (DTPA), and hydroxycarboxylates, such as citric acid, have been most frequently used in electrochemical

remediation. A detailed review of use of chelants in electrochemical remediation is given by Yeung and Gu [2], and will not be repeated in this review paper. In addition to solubilizing sorbed contaminants from soil particle surfaces, chelants also change the zeta potential of soil particle surfaces. In general, chelants lower (becomes more negative) the zeta potential of soil particle surfaces [12]. The lowering of the zeta potential of soil particle surfaces increases the positive electroosmotic volume ﬂow rate of soil pore ﬂuid, i.e., from the anode towards the cathode, facilitating the advective transport of contaminants by electroosmosis towards the cathode. Post-remediation of treatment and disposal of the used extraction ﬂuid is a problem as it is rich in metal-chelant complexes.

Fig. 2. Solubilization of sorbed Pb from soil particle surfaces by EDDS.

14

A.T. Yeung, Y.-Y. Gu / Journal of Hazardous Materials 195 (2011) 11–29

Some chelants, such as EDTA, are toxic, especially in their free forms [13,14], and are poorly photo-, chemo-, or biodegradable in the environment [15]. Different methods to handle used extraction ﬂuids are presented by Lestan et al. [16]. The methods currently available to recover chelants from used extraction ﬂuids still encounter operational difﬁculties and they work well only for a few contaminants and soil types [2,16]. The development of more robust recycling methods for used chelants would greatly increase the economic value of chelant-enhanced electrochemical remediation. 3.2.2. Complexing agents Complexing agents are chemicals which form coordination complexes with metal ions. Coordination complexes differ from chelate complexes by the formation of only a single bond between the metal central atom or ion and the complexing agent. Some complexing agents, such as I− , Cl− , NH3 − , and OH− , are introduced into soil as conditioning acids or bases during electrochemical remediation process. These ligands can form soluble complexes with metals such as [HgI4 ]2− , [CuCl2 ]− , [CuCl4 ]2− , [Cu(NH3 )4 ]2+ , [Zn(OH)4 ]2− , [Cr(OH)4 ]− , and [Cr(OH)3 ]2− . It has been demonstrated by many researchers [17–20] that mercury could be efﬁciently extracted by iodide-enhanced electrochemical remediation as soluble complex HgI4 2− . Sulfate reducing bacteria were shown to be a viable tool for treatment of the acidic and oxidative Hg-contaminated iodide waste solution resulting from the enhanced electrochemical remediation [21]. Acetic acid, CH3 COOH, is a complexing agent frequently utilized to enhance electrochemical remediation [22–24]. Although it is not as effective as strong acid such as HNO3 , it is preferred in soil remediation. It can neutralize the electrolysis product at the cathode to reduce energy consumption, and keep the electrolyte pH within a certain range by its acid/base buffer capacity. Moreover, it is relatively cheap, biodegradable, and environmentally safe. Similarly, lactic acid was used to enhance electrochemical remediation of Cu-contaminated soil [25]. Cyclodextrins are nontoxic, biodegradable, and have low afﬁnity of sorption onto the soil particle surfaces in a wide pH range [26]. Moreover, they have the ability to form inclusion complexes with many substrates in aqueous solutions. Hydroxypropyl-␤cyclodextrin, carboxymethyl-␤-cyclodextrin, ␤-cyclodextrin, and methyl-␤-cyclodextrin have been utilized to enhance electrochemical remediation of soils and sediments contaminated with organic compounds and heavy metals [26–32], with varying degrees of success. Ammonium acetate, CH3 COONH4 , was used as anolyte by Chen et al. [33] in their bench-scale experiments on electrokinetic removal of Cu from soil using a constant electrical current density of 1.33 A/m2 . Their results reveal that a concentration of CH3 COONH4 of higher than 0.1 M was needed to sustain the electroosmotic ﬂow. The apparent electrical conductivity of the specimen was controlled by the 10-mm thick layer of soil close to the cathode. The high pH condition in the vicinity of the cathode favors copper–ammonia complex reactions, thus increasing the solubility and removal rate of Cu during electrochemical remediation. The extraction efﬁciency of Cu increased with the concentration of CH3 COONH4 used. When 0.5 M CH3 COONH4 was used, the proportion of soil containing Cu was less than 10% after treatment. 3.2.3. Surfactants and cosolvents Cationic, anionic, or non-ionic surfactants are amphiphilic compounds containing both hydrophilic groups (heads) and hydrophobic groups (tails). There are both synthetic and natural surfactants. Natural surfactants are also known as biosurfactants, as they are biologically produced from yeast or bacteria from various substrates including sugars, oils, alkanes, and wastes [34].

Fig. 3. Variation of surface tension, interfacial tension, and contaminant solubility with surfactant concentration (after Mulligan et al. [35]).

Surfactants can lower the surface tension of a liquid to allow easier spreading, and the interfacial tension between two liquids, or between a liquid and a solid. Therefore, they may act as adhesives, ﬂocculating agents, wetting agents, foaming agents, detergents, de-emulsiﬁers, penetrants, and dispersants. Typical desirable functions of surfactants include solubility enhancement, surface tension reduction, critical micelle concentration, wetting ability, and foaming capacity [35]. Surfactant monomers form spheroid or lamellar structures with organic pseudo-phase interiors, which lowers surface or interfacial tensions The minimum concentration at which any added surfactant molecules appear with high probability as micellar aggregates is called the critical micelle concentration (CMC) [36]. The variation of surface tension, interfacial tension, and contaminant solubility with surfactant concentration is schematically shown in Fig. 3. Both synthetic surfactant and natural surfactants can be used as additives in the phase separation processes for remediation of organic compound-contaminated soils by enhancing the aqueous solubility and mobility of organic contaminants [35,37–40]. Moreover, surfactants have been observed by many researchers to be feasible in enhancing heavy metal extraction from soil and sludge [41]. Several factors can adversely affect the efﬁciency of soil ﬂushing using surfactants including: (1) hardness of groundwater; (2) sorption of surfactants onto clay particle surfaces; (3) inactivation of surfactants due to rapid biodegradation; and (4) difﬁculties in recovering the surfactant from used ﬂushing solution [42]. Therefore, factors that need to be considered in the selection of surfactants in electrochemical remediation include: (1) efﬁciency and effectiveness of the surfactant in remediating the contamination; (2) biodegradability of the surfactant and degradation products; (3) toxicity of the surfactant and its degradation products to humans, animals, plants, and the ecology; (4) ability to be recovered, recycled, and reused; (5) public perception and regulatory restrictions; (6) functionality of the surfactant at different pHs; (7) electrical charges, if any, carried by the surfactant; and (8) cost. Overall, desirable surfactant characteristics for soil remediation include biodegradability, low toxicity, solubility at groundwater temperatures, low sorption onto soil particles, effective at concentrations lower than 3%, low soil dispersion, low surface tensions, and low CMC. Anionic and non-ionic surfactants are less likely to be sorbed onto soil particle surfaces but anionic surfactants may precipitate. However, co-injection of an anionic surfactant with a non-ionic surfactant can reduce precipitation and also CMC values [43].

A.T. Yeung, Y.-Y. Gu / Journal of Hazardous Materials 195 (2011) 11–29

Biosurfactants or natural surfactants are known for their biodegradability, reduced toxicity, and environmental-friendliness [44], and they may be less expensive in some cases [45]. Moreover, their efﬁciency is often higher than those of synthetic surfactants, i.e., a similar surface tension reduction can be achieved by introduction of a smaller quantity of biosurfactant [46]. They are proven to be more tolerant to extreme variations in temperature, ionic strength, and pH [47–49]. Moreover, they may potentially be produced in situ using the organic contaminants as substrates for their production. Biosurfactants are essentially classiﬁed either as low- or high-molecular-mass. High-molecular-mass biosurfactants consist of particulate and polymeric amphiphiles. Low molecular-mass biosurfactants can broadly be classiﬁed into three groups: (1) glycolipids or lipopolysaccharides, such as rhamnolipids, trehalolopids [50], and sophorolipids [51,52]; (2) lipoproteins–lipopeptides, such as acyclic [53], and cyclic ones (cyclolipopeptides) [54,55]; and (3) hydroxylated cross-linked fatty acids (mycolic acids) or phospholipids [46]. Both synthetic surfactant and natural surfactants have been reported to be efﬁcient in mobilizing organic contaminants during electrochemical remediation of soil contaminated by organic compounds. The feasibility of using synthetic surfactants such as alkyl polyglucoside (APG), sodium dodecyl sulfate (SDS), sodium dodecylbenzenesulfonate (SDBS), Pannox 110, Brij 30, Triton X-100, Griton ALM 100, Calfax 16L-35, Igepal CA-720, Tergitol 15-S-7, Tergitol NP-10, and Tween 80 have been studied by many researchers to enhance electrochemical remediation of soils contaminated by petroleum hydrocarbons [56], polycyclic aromatic hydrocarbons (PAHs) [27,57–64], 1,2-dichlorobenzene (1,2-DCB) [65], hexachlorobenzene (HCB) [29,63], dichlorodiphenyltrichloroethane (DDT) [66], ethylbenzene [67], chlorobenzene [68], trichloroethylene (TCE) [68], and diesel oil [69,70]. The viability of these synthetic surfactants in enhancing electrochemical remediation of soils contaminated by various organic compounds has been established. However, few studies have been carried out on the use of natural surfactants to enhance electrochemical remediation. Nonetheless, rhamnolipid is the most frequently used biosurfactant as an enhancement agent for electrochemical remediation of soils contaminated by organic contaminants. Chang et al. [61] compared the performance of rhamnolipid with Triton X-100, a synthetic surfactant, in enhancing the extraction of phenanthrene from unsaturated soils by electrochemical remediation. Their results indicate that rhamnolipid was more efﬁcient in removing phenanthrene from soil than Triton X-100. Moreover, the electroosmotic ﬂow rate in the rhamnolipid system was higher than that in Triton X-100. In addition to the higher electroosmotic ﬂow rate, the higher remediation efﬁciency may also be attributed to the promotion of microbial growth in the soil-water system in the presence of rhamnolipid. Gonzini et al. [71] also studied the effects of rhamnolipid on enhancing electrochemical remediation of a gasoil-contaminated soil. Their results indicate that the remediation efﬁciency of gasoil could be increased up to 86.7% by increasing the dose of rhamnolipid. Moreover, the lower concentration of the gasoil in the liquid phase at the higher concentration of the biosurfactant demonstrated evidently that rhamnolipid could enhance gasoil biodegradation, possibly through two mechanisms: (1) increasing the aqueous solubility of hydrocarbons and thus their bioavailability to microorganisms; and (2) interacting with microorganisms to make their cell surfaces more hydrophobic and thus easier to associate with hydrophobic substrates. They also identiﬁed the need for future development on surfactant production by autochthonous microorganisms, so as to reduce the surfactant cost for ﬁeld application of the technology. Groboillot et al. [72] studied the feasibility of using amphisin, a biosurfactant, to enhance electrochemical remediation of dredged

15

harbor sediments contaminated by PAHs. Their results indicate pure amphisin from Pseudomonas ﬂuorescens DSS73 was more effective in solubilizing and mobilizing PAHs strongly sorbed to sediments than a synthetic anionic surfactant. Amphisin production by bacteria in natural environment was also considered. Although the growth of P. ﬂuorescens DSS73 was weakened by the three model PAHs above saturation, amphisin was still produced. Kaya and Yukselen [73] studied the effects of anionic, cationic, and non-ionic surfactants on the zeta potential of soil particle surfaces of kaolinite, montmorillonite, and quartz powder in the presence of Li+ , Ca2+ , Cu2+ , Pb2+ , and Al3+ . Understanding the variations of zeta potential of soil particle surfaces with the introduction of surfactants is important because the zeta potential controls the direction and rate of electroosmotic ﬂow which impact the contaminant extraction efﬁciency of electrochemical remediation [1,2]. Their results indicate that the presence of cationic surfactant significantly increases (becomes less negative) the zeta potential of soil particle surfaces in an acidic environment (pH ∼4). The presence of the anionic surfactant makes the zeta potential of soil particle surfaces more negative. However, the non-ionic surfactant has little effect on the zeta potential of soil particle surfaces. They recommended the determination of zeta potential of soil particle surfaces prior to electrochemical remediation to maximize the remediation efﬁciency of the technique. However, the results of using surfactants to enhance the extraction efﬁciency of metal contaminants from soil by electrochemical remediation are mixed. Some researchers reported positive results [74,75], while other researchers reported insigniﬁcant enhancement [64,76,77]. Cosolvent is a second solvent added in small quantity to the primary solvent to form a mixture that may greatly enhance the solvent power of the primary solvent due to synergism. They can enhance the aqueous solubility of many organic contaminants through cosolvent effect. Several cosolvents, such as ethanol [78,79], n-butylamine [80–82], n-propanol [70], acetone [80], and tetrahydrofuran [80], have been examined for their ability to enhance the solubilization of organic compounds such as PAHs and diesel oil in soil during the electrochemical remediation process. 3.2.4. Oxidizing/reducing agents Oxidizing or reducing agents can be injected into contaminated soil to manipulate the in situ chemistry and microbiology, so as to enhance extraction of contaminants or to reduce their toxicity through oxidation or reduction reactions. Oxidizing agents may include air or oxygen, or chemical oxidants, such as hydrogen peroxide H2 O2 , potassium permanganate KMnO4 or sodium permanganate NaMnO4 , ozone, chlorine, or oxygen releasing compounds. Contaminants are chemically or microbially oxidized. Similarly, reducing agents such as Fe2+ , Fe0 , calcium polysulﬁde, or sodium dithionite can be used to reduce contaminants in soil. The injection of oxidizing/reducing agents during electrochemical remediation of contaminated soil is equivalent to coupling electrochemical remediation with oxidation/reduction to remediate contaminated soil. Therefore, the subject will be treated in Section 5. 3.2.5. Cation solutions Coletta et al. [83] used natural solutions containing clay extracts and synthetic solutions with varying concentrations of Al3+ , Ca2+ , and Na+ as anodic ﬂushing solutions to investigate the feasibility of enhancing electrochemical remediation of Pb-contaminated clay of initial Pb concentration of 340–410 mg/kg dry clay (dry) and moisture content of 80–83%. Natural ﬂushing solutions were prepared by mixing water and clay in ratios varying from 2:1 to 40:1 by weight, and the supernatant was used as an anodic ﬂushing solution. The 7:1 natural solution was observed to be most effective

16

A.T. Yeung, Y.-Y. Gu / Journal of Hazardous Materials 195 (2011) 11–29

for Pb removal. The solution was composed of 2.289, 0.314, 1.464, 0.803, and 0.908 ppm of Ca2+ , A13+ , Na+ , Mg2+ , and K+ , respectively. Synthetic solutions were prepared using AlCl3 , Ca(NO3 )2 , and NaCl solutions. The Pb extraction efﬁciency was highest when the solution ionic strength was approximately 0.001 M for each element group, and with trivalent Al3+ and divalent Ca2+ ions at concentrations of 0.064 and 0.31 mM, respectively. Moreover, the 0.31 mM Ca synthetic solution exhibited the highest overall Pb extraction efﬁciency due to its high ionic mobility, large hydrated ionic radius, and near optimum ionic strength. Energy requirement was determined to be 8–31 kWh/m3 of soil. Reddy and Chinthamreddy [84] investigated the feasibility of enhancing electrochemical remediation of glacial till spiked with Cr6+ , Ni2+ , and Cd2+ of concentrations of 1000, 500, and 250 mg/kg, respectively by simultaneous injection of 0.1 M NaCl from the anode and 0.1 M EDTA from the cathode. Their experimental results indicate that the presence of NaCl sustained the electric current and electroosmotic ﬂow. The remediation efﬁciency of Cr was increased considerably to 79%. Ni and Cd were migrated signiﬁcantly towards the anode but eventually accumulated in the soil near the anode. The accumulation of these metals was attributed to the preferential complexation of EDTA with H+ ions in an acidic environment. The thickness of the diffuse double layer around soil particles 1/ (m) is given by 1 =

εRT 2000 × cF 2 z 2

(1)

where ε is the permittivity of soil pore ﬂuid (F/m); R is the universal gas constant (8.314 J/mol K); T is the absolute temperature (K); c is the concentration of cations in the diffuse double layer (mol/L); F is the Faraday constant (96,485 C/mol); and z is the valence of cations in the diffuse double layer. The electroosmotic volume ﬂow rate is given by Q = ke ie A

(2) (m3 /s);

ke is where Q is the electroosmotic volume ﬂow rate the coefﬁcient of electroosmotic conductivity (m2 /V s); ie is the electrical gradient (V/m); and A is the total cross-sectional area perpendicular to the direction of ﬂow (m2 ). It should be noted that the polyvalent cations injected into contaminated soil may replace contaminant ions or H+ ions in the diffuse double layer of the soil. Cation exchange in clay follows a replaceability series that favors the adsorption of cations of higher valence. If two atoms have the same valence, the larger cation is favored. The order of adsorption is shown with corresponding ionic radii in A˚ as follows [3], Al3+ 0.57

>

Pb2+ 1.19

>

Ca2+ 1.06

>

Mg2+ 0.78

>

K+ 1.33

>

Na+ 0.98

>

Li+ 0.7

In addition to valence and ionic radius, ion concentration is another important factor affecting cation exchange. A cation of high concentration and low replacing power may be preferred to a cation of low concentration and high replacing power [3]. Adsorption of cations of high valence in the diffuse double layer of clay particle surfaces causes a decrease in the thickness of the diffuse double layer around clay platelets as predicted by Eq. (1) [85,86]. Moreover, a high concentration of cations in the soil pore ﬂuid causes further decrease in the thickness of the diffuse double layer and an increase in ionic strength of the system. The decrease in thickness of the diffuse double layer decreases the repulsive forces among clay particles and allows the van der Waals attractive forces among clay platelets to dominate, resulting in ﬂocculation of clay particles. The ﬂocculated structure in the clay fabric causes an increase in porosity and decrease in tortuosity of ﬂow paths, leading to an increase in tortuosity factor, and resulting in increases in the hydraulic conductivity and coefﬁcient of electroosmotic conductivity of the clay, and the effective

ionic mobilities of the ionic species in the soil pore ﬂuid. However, an increase in ionic strength (electrolyte concentration) of the soil pore ﬂuid increases the electrical conductivity of soil and energy consumption of the process. Conversely, a decrease in ionic strength (electrolyte concentration) increases the thickness of the diffuse double layer, leading to a decrease in the coefﬁcient of electroosmotic conductivity and a reduction of electroosmotic ﬂow rate and electromigration of ions. However, the electric current ﬂowing through the soil is reduced, leading to lower energy consumption. As the electroosmotic ﬂow rate and electromigration are the dominant transport mechanisms in electrochemical remediation, there is an optimized ionic strength to maximize the overall remediation efﬁciency of electrochemical remediation. 4. Soil pH control As a result of electrolytic decomposition of electrolytes at the electrodes, H+ and OH− ions are generated at the anode and the cathode, respectively during the electrochemical remediation process as follows: Oxidation at the anode : Reduction at the cathode :

2H2 O − 4e− → 4H+ + O2 ↑ −

−

4H2 O + 4e → 4OH + 2H2 ↑

(3) (4)

The generated H+ and OH− ions are migrated into the soil by the dc electric ﬁeld imposed on the soil. As a result, the soil pH near the anode is lowered and that near the cathode is raised. Different techniques have been developed to condition the electrode reservoir solutions, i.e., the anolyte and catholyte, so as to eliminate the adverse impacts of electrode reactions. The primary purpose of electrode reservoir conditioning is to maintain the pHs of anolyte and/or catholyte within appropriate ranges speciﬁc to the contaminants being remediated. In most cases, the pH of anolyte is raised and that of catholyte is lowered. The conditioning is particularly important for electrochemical remediation of soils of low acid/base buffer capacity, as the resistances to pH change of these soils are low. Speciﬁc objectives of reservoir conditioning include [87]: (1) precipitation of metal contaminants should be avoided and/or precipitates should be solubilized and mobilized; (2) electrical conductivity of the specimen should not be increased excessively in a short duration so as to avoid diminishing of the advective transport of contaminant by electroosmosis prematurely; (3) the electrolysis reaction at the cathode should possibly be depolarized to avoid the generation of OH− ions and their transport into the specimen; (4) the depolarization would also assist in decreasing the electrical potential difference across the specimen and reduce energy consumption of the process; (5) if any chemical is used, the metal precipitate with this new chemical should be soluble within the pH ranges maintained by reservoir conditioning; (6) any special chemicals introduced should not result in any increase in toxic residue in the soil; and (7) the additional cost of chemicals and/or equipment for reservoir conditioning should not increase the overall cost of the electrochemical remediation process signiﬁcantly. The most frequently used reservoir conditioning techniques in electrochemical remediation are: (1) electrode conditioning by conditioning agents; and (2) use of ion exchange membranes. 4.1. Electrode conditioning Weak acids may be introduced to neutralize the OH− ions generated at the cathode during the electrochemical remediation process. However, improper use of some acids in the process may pose a health hazard. For example, the use of HCl may pose a health hazard as: (1) it may increase Cl− concentration in groundwater; (2) it may promote the formation of some insoluble chloride salts,

A.T. Yeung, Y.-Y. Gu / Journal of Hazardous Materials 195 (2011) 11–29

for example, PbCl2 ; and (3) Cl2 gas will be generated by electrolysis if it reaches the anode. Organic acids, such as CH3 COOH or citric acid, are weak acids that undergo partial dissociation in water. There are several advantages in using these weak acids to depolarize the OH− ions generated at the cathode: (1) they are environmentally safe and biodegradable; (2) they possess certain acid/base buffer capacities so that they can maintain the electrolyte pH to some extent; (3) they are complexing agents that can form soluble complexes with metals to enhance solubilization of heavy metals sorbed on soil particle surfaces and to maintain mobility of heavy metals in soil; (4) the concentration of ions generated by acid dissociation is very low as their pKa values are relatively high, the resulting increase in the electrical conductivity of soil and thus the power consumption are small; and (5) these weak acid ions prevent the formation of other insoluble salts in the vicinity of the cathode, preventing the development of a low electrical conductivity zone and dissipation of excessive electrical energy in the soil near the cathode [2]. Experimental results on removal of Pb from kaolinite by Lee and Yang [88] indicate that external circulation of the electrolyte solution from the cathode reservoir to the anode reservoir could control pore ﬂuid pH, and prevent excessive H+ from decreasing electroosmotic ﬂow rate and excessive OH− from increasing heavy metal precipitation. Saichek and Reddy [89] demonstrated that the use of NaOH to control pH at the anode could improve the extraction efﬁciency of phenanthrene from kaolin by electrochemical remediation. Experimental results of Hicks and Tondorf [90] on removal of Zn from Georgia kaolinite, a soil of low acid/base buffer capacity, reveal that problems related to isoelectric focusing could be prevented by rinsing away the OH− ions generated at the cathode, achieving an extraction efﬁciency of 95%. The experimental results of Puppala et al. [87], Rødsand et al. [91], and Reed et al. [92] indicate that the addition of CH3 COOH to the cathode reservoir prevented the development of alkaline conditions in the soil. The technique could improve the extraction efﬁciency of Pb, as the soil pH nearest to the cathode was lowered to prevent precipitation of Pb(OH)2 . Zhou et al. [93] studied the performance of electrochemical remediation of the low pH Chinese red soil contaminated by Cu and Zn enhanced by catholyte conditioning. Without catholyte conditioning, the soil pH near the cathode was increased from 4.2 to above 6, resulting in accumulation of large quantities of Cu and Zn precipitates in the vicinity of the cathode. Application of lactic acid as catholyte pH conditioning agent improved the extraction efﬁciency of Cu and Zn from the soil. Increasing the ionic strength of the conditioning agent by adding 10 mM CaCl2 further enhanced Cu removal, but did not cause a signiﬁcant improvement for Zn extraction. The feasibility of using reservoir conditioning to enhance electrochemical remediation of heavy Cd-contaminated soil was investigated by Gidarakos and Giannis [94]. 0.01 M CH3 COOH or 0.01 M citric acid was used as catholyte to prevent Cd from precipitating as hydroxide. Their results reveal that when the catholyte pH was controlled to be lower than 4, signiﬁcant amounts of H+ ions produced at the anode could be migrated throughout the specimen, resulting in desorption of Cd from soil particle surfaces and a very high extraction efﬁciency. Ryu et al. [95] studied the performance of laboratory-scale electrochemical remediation on Cu-, As-, and Pb-contaminated soil enhanced by electrolyte conditioning. Their results reveal that catholyte conditioning using HNO3 increased the removal of Cu and Pb from the soil, and the maximum removal was 60.1% for Cu and 75.1% for Pb. Anolyte conditioning using NaOH enhanced the migration of As which exists in an anionic form and 43.1% of As was removed.

17

Fig. 4. The NEOCHIM electrode (after Leinz et al. [100]).

Genc et al. [96] used CH3 COOH to keep both the anolyte and catholyte at pH ≤4 in their laboratory study on electrochemical remediation of contaminated sediment from Cuyahoga River, OH, USA. The river sediment was contaminated by Mn, Cu, Zn, and Pb. However, the low pH of catholyte generated reverse electroosmotic ﬂow, i.e., from the cathode towards the anode. As a result, they observed the accumulation of Mn near the cathode. However, other metals, such as Cu, Zn, and Pb were mostly in the middle section of the specimen. Moreover, as a result of reverse electroosmotic ﬂow, the extraction efﬁciencies of metals were low. The highest extraction efﬁciencies of Mn, Cu, and Pb observed were 18%, 20% and 12%, respectively, and no removal of Zn was observed in all their experiments. Buffer solutions, such as CH3 COOH and NaHCO3 , have also been successfully used to control the pH of electrode reservoir electrolytes so as to control the electroosmotic ﬂow direction and to maintain the electroosmotic volume ﬂow rate during the electrochemical remediation of Pb- or Cd-contaminated Milwhite kaolinite, a natural clay of high acid/base buffer capacity [97,98]. The NEOCHIM technology was developed by the U.S. Geological Survey on the foundation of Russian scientists’ research results on CHIM, a method of electrogeochemical sampling for use in the exploration of buried mineral deposits. A schematic of the NEOCHIM electrode is shown in Fig. 4. The technology solves the problems associated with the presence of H+ and OH− ions in the vicinity of electrodes by using an electrode made of two compartments linked by a salt bridge [99]. The power electrode is immersed in a conducting ﬂuid in the inner compartment where H+ and OH− ions produced by electrolysis are retained and prevented from reaching the outer compartment by the salt bridge. The salt bridge is retained by a semipermeable parchment membrane at the base of the inner compartment. A further conducting ﬂuid is retained by the outer compartment. Electrical contact of the electrode with soil is made through a semipermeable parchment membrane at the base of the outer compartment. The membrane allows the passage of ions from the conducting ﬂuid into the soil and from the soil into the ﬂuid, while retaining the ﬂuid in the compartment. The experimental results of Leinz et al. [100] on electrochemical remediation also indicate the high potential of the NEOCHIM process for the monitoring and remediation of hazardous waste sites.

18

A.T. Yeung, Y.-Y. Gu / Journal of Hazardous Materials 195 (2011) 11–29

4.2. Use of ion exchange membrane Another technique of reservoir conditioning is the use of ion exchange membranes or ion-selective membranes to isolate speciﬁc ions generated by electrode reactions from the contaminated soil. Cation exchange membranes essentially allow only cations to pass through, and anion exchange membranes allow only anions to pass. Therefore, a cation exchange membrane installed between the cathode and contaminated soil can prevent the OH− ions generated at the cathode from migrating into the contaminated soil and precipitating with metal contaminants as hydroxides. The metal contaminant cations can be migrated from the soil through the cation exchange membrane into the catholyte to precipitate with the OH− ions on the membrane surface or in the catholyte. However, precipitation with the OH− ions on the membrane surface causes fouling of the membrane. The deterioration of the membrane performance is essentially caused by deposition of foulants on the membrane surface, resulting in an increase in ﬂow resistance of the membrane and decreases in ﬂuxes ﬂowing through the membrane. Nonetheless, the technique is promising as it does not introduce any additional chemicals into the system [101]. However, membrane fouling remains one of the most crucial factors limiting the use of ion exchange membranes in electrochemical remediation [102]. Moreover, the experimental results of Rødsand et al. [91] indicate that the membrane extraction technique did not enhance the extraction of Pb from spiked Norwegian marine clay by electrochemical remediation as expected. Puppala et al. [87] studied the use of NaﬁonTM membrane to limit the transport of OH− ions into soil during electrochemical remediation of an illitic deposit contaminated by Pb. The advantage of the membrane technology is that it would not be necessary to neutralize the cathode by continuous introduction of acid, resulting in considerable saving of acid cost. However, the energy consumption of the process was increased by the electrical resistance of the membrane. Lower expenditures are anticipated if: (a) the membrane is changed periodically and cleaned to prevent fouling, and/or (b) the post-membrane catholyte is removed frequently for precipitation. Moreover, the relative high cost of NaﬁonTM membrane may increase the cost of in situ electrochemical remediation unless the system can be engineered and optimized to decrease the cost during real-life ﬁeld implementation. Therefore, cost-efﬁcient ﬁeld techniques should be devised. The results of Li et al. [103] indicate the use of a cation selective membrane installed at the front of the cathode to prevent OH− ion migration towards the anode could greatly enhance the extraction efﬁciency of Cu by electrochemical remediation. However, they observed that very little Cu ions could penetrate the cation-selective membrane to precipitate in the cathode compartment. Although a cation selective membrane should ideally not permit anions, such as OH− , to enter, most of the Cu precipitated as hydroxides in the compartment between the soil and the membrane, indicating the membrane was not 100% effective and some OH− ions still entered the compartment and precipitated the Cu ions there. Kim et al. [104] installed an anion exchange membrane between the anode and contaminated soil specimen and a cation exchange membrane between the cathode and the soil specimen to enhance electrochemical remediation of a Cd- and Pb-contaminated kaolinite. Moreover, an auxiliary solution cell was installed between the cation exchange membrane and the contaminated soil. Small holes were punched in the membrane to allow OH− ions to move into the auxiliary solution cell from the catholyte so that metal contaminants were precipitated in the auxiliary solution cell instead of at the catholyte. Their results indicate the overall extraction efﬁciencies of membrane-enhanced electrochemical remediation were improved tremendously due to the prevention of hydroxide

precipitation in the soil and increase in electric current efﬁciency. Moreover, the installation of the auxiliary solution cell could nullify the fouling problem within the cation exchange membrane and thus improve the overall effectiveness of the electrochemical remediation process. 5. Coupling with other remediation technologies There are many technologies available for remediation of contaminated soil and groundwater [8,105]. They all have their advantages and disadvantages. Some of these technologies can be coupled with electrochemical remediation synergistically so that the coupled remediation efﬁciency is higher the sum of the individual technologies applied individually. Some of the remediation technologies with feasibility of coupling with electrochemical remediation are presented here. However, the feasibility of many other remediation technologies coupling with electrochemical remediation has yet to be investigated and it should be noted that there are numerous opportunities of coupling these remediation technologies with electrochemical remediation to improve the remediation efﬁciency of contaminated soil and groundwater drastically for the beneﬁt of mankind and the environment. 5.1. Oxidation/reduction The oxidation/reduction remediation technologies focus on modifying the chemistry and microbiology of the environment by injecting selected reagents into the subsurface to enhance degradation and extraction of contaminants by in situ chemical oxidation/reduction reactions [8]. The technologies are applicable for a wide range of inorganic, organic, and mixed contaminants. The most widely studied and utilized oxidation technology in environmental engineering is probably the Fenton process. All processes that involve catalytic reaction between hydrogen peroxide H2 O2 and Fe2+ ions can be denoted as Fenton processes [106]. The Fenton process involves two major steps: (1) oxidation of Fe2+ ions to Fe3+ ions with decomposition of H2 O2 and generation of hydroxyl radicals, as illustrated in Eq. (5); and (2) degradation of organic contaminants by hydroxyl radicals through oxidation as illustrated in Eqs. (6) and (7), Fe2+ + H2 O2 → Fe3+ + OH• + OH−

(5)

RH + HO• → H2 O + R •

(6)

R•

(7)

+ Fe

3+

→ Fe

2+

+ products

By-products of the chemical reactions presented in Eq. (7) can be further degraded by radical mechanism to complete mineralization. Although Eq. (5) is often referred as the Fenton reaction, other important reactions, such as the occurrence of the Fenton catalytic cycle, also occur: Fe2+ + HO• → Fe3+ + HO− 3+

+ H2 O2 → Fe

2+

2+

•

+ HO2 → Fe

3+

3+

+ HO2 • → Fe

2+

Fe Fe Fe

(8) •

+ HO2 + HO + OH2

−

+ O2 + H

−

(9) (10)

+

(11)

The presence of Fe is catalytic. The hydroxyl radicals so generated are strong and relatively unspeciﬁc oxidants that react with most organic contaminants. Therefore, the Fenton process is widely used for the destruction of biorefractory organic contaminants such as benzene, phenols and chlorophenols in wastewater or drinking water. The radicals oxidize the organic molecule by abstracting hydrogen atoms as illustrated in Eq. (6) or by adding themselves to double bonds and aromatic rings. The hydroxyl radicals are only active in aqueous form and thus cannot attack contaminants sorbed

A.T. Yeung, Y.-Y. Gu / Journal of Hazardous Materials 195 (2011) 11–29

on soil particle surfaces [107]. However, it has been demonstrated that it is technically feasible to use high concentration H2 O2 to oxidize contaminants sorbed on soil particle surfaces [108,109], as high concentration H2 O2 favors the generation of highly reactive species, such as HO2 • (hydroperoxyl radicals), O2 •− (superoxide anions), and HO2 •− (hydroperoxide anions), other than hydroxyl radicals. The generation of these non-hydroxyl and highly reactive radicals in the presence of high concentration H2 O2 leads to aggressive reactions that ultimately oxidize the contaminants sorbed on soil particle surfaces [110,111]. However, the Fenton process is effective only at low pHs of 3–5. Therefore, pH adjustment may be required during the remediation process. Recently, there are many investigations into the Fenton-like processes for the degradation of organic contaminants. These processes can be broadly classiﬁed into three groups: (1) processes that use ferric salts as catalyst to incite the Fenton reaction, i.e., Eq. (5); (2) processes that use heterogeneous Fenton type catalysts such as iron powder, iron-oxides, iron-ligands, or iron ions doped in zeolites, pillared clays or resins; and (3) processes that use other metal ions, e.g., copper, manganese or cobalt, as catalyst. The major advantages of the Fenton type processes are: (1) they are able to degrade many organic contaminants to harmless or biodegradable products; (2) they use relatively cheap reagents; and (3) the reagents are safe to handle and environmentally benign [106]. The bench-scale laboratory experimental results of Yang and Long [112] and Yang and Liu [113] indicate that it is technically feasible to couple the Fenton-like process with electrochemical remediation using a permeable reactive barrier of granular scrap iron powder to extract and degrade phenol and trichloroethylene (TCE) in situ, respectively. The overall contaminant remediation efﬁciency is contributed by two mechanisms: (1) destruction of organic contaminants by the Fenton-like process; and (2) extraction of contaminants by electrochemical remediation. Their experimental results also reveal that the percentage of organic contaminant destruction increased with the quantity of scrap iron powder used in the process. However, a larger quantity of scrap iron powder embedded in soil would decrease the coefﬁcient of electroosmotic conductivity, resulting in lower efﬁciency of advective transport of the contaminant by electroosmosis and thus lower contaminant extraction efﬁciency. Moreover, the smaller was the size of the scrap granular iron powder, the higher was the destruction efﬁciency, but the lower was the overall contaminant remediation efﬁciency. Kim et al. [114] explored the feasibility of coupling the Fenton process with electrochemical remediation to remediate phenanthrene-contaminated EPK kaolinite, using the iron minerals on soil particle surfaces as catalyst. Their results reveal that the intermediate anions, i.e., HO2 − and O2 •− , generated by the Fentonlike reactions changed the electrical current intensity signiﬁcantly. The addition of 0.01 N H2 SO4 to the anode reservoir improved the stability of H2 O2 and treatment efﬁciency of phenanthrene in the soil specimen. More than a half of the spiked phenanthrene was destructed or extracted after 21 days of treatment. Therefore, the use of H2 O2 and dilute acid, as an anode purging solution, is a feasible technology for the remediation of halogenated organic compound-contaminated soil of low hydraulic conductivity, low acid/base buffer capacity, and high iron content. Kim et al. [115] attempted to remediate phenanthrene-contaminated Hadong clay similarly, however, the acid/base buffer capacity of Hadong clay is high due to its high carbonate content. Their results reveal that the presence of carbonates of high acid/base buffer capacity reduced the stability of H2 O2 and treatment efﬁciency of phenanthrene, and conﬁrmed that the Fenton reaction is effective only at low pHs of 3–5.

19

Different methods have been attempted to overcome the problem of high acid/base buffer capacity of soil. Kim et al. [116] studied the stabilizing effects of phosphate and sodium dodecyl sulfate (SDS) on H2 O2 during electrochemical remediation of phenanthrene-contaminated Hadong clay coupled with the Fenton-like process. Both stabilizers decreased (becomes more negative) the zeta potential of soil particle surfaces due to complexation of phosphate and SDS with oxides, resulting in increase of electroosmotic volume ﬂow rate. Complexation with phosphate hindered the migration of dissolved Fe ions towards the cathode signiﬁcantly. However, SDS could dissolve the Fe ion from the Fe oxide of soil and transport the dissolved Fe ions towards the cathode. Nonetheless, transition metal complexation with phosphate and SDS improved the stability of H2 O2 , in particular, in the high pH region near the cathode by SDS. The increase of H2 O2 stability allowed more reaction time for the Fenton-like process, resulting in better treatment efﬁciency of phenanthrene. Kim et al. [117] studied the performance of H2 SO4 and HCl injected from the anode for pH control in the remediation of phenanthrene-contaminated Hadong clay by electrochemical remediation. When H2 SO4 was utilized, the reduced species of sulfate may increase the decomposition rate of H2 O2 near the anode signiﬁcantly as follows: SO4 2− + 2e− + 2H+ → SO3 2− + H2 O −

(12)

−

HSO3 + H2 O2 → SO2 OOH + H2 O

(13)

SO2 OOH− + H+ → SO4 2− + 2H+

(14) HS− ,

Moreover, reduced sulfur species, such as H2 S and accumulated in the region near the cathode due to the reducing environment of the region. The generation of these sulfur species is accompanied by a signiﬁcant stoichiometric decrease of H+ ions in the soil pore ﬂuid, SO4 2− + 8e− + 10H+ → H2 S + 4H2 O

(15)

resulting in a sharp increase in soil pH, rapid decomposition of H2 O2 , and generation of O2 gas. Such decomposition of H2 O2 was not observed in experiments using HCl as the pH control agent. Moreover, H2 O2 may be re-generated near the cathode by the reaction, O2 + 2H+ + 2e− → H2 O2

(16)

The remediation efﬁciency of phenanthrene-contaminated soil by the Fenton-like process is dependent on both the extent of degradation and migration by electroosmosis. Alcantara et al. [118] studied the electrochemical remediation of phenanthrene-contaminated kaolinite of initial concentration of 500 mg/kg of soil. Electrochemical remediation alone resulted in negligible remediation of phenanthrene. Fenton-like reaction was thus generated in kaolinite which was also contaminated by Fe. When both the anode and cathode reservoirs were ﬁlled with 10% H2 O2 , an overall extraction and destruction efﬁciency of phenanthrene of 99% was obtained in 14 days by applying an electrical gradient of 300 V/m across the soil specimen. It should be noted that the soil pH was maintained at approximately 3.5 without pH control, favoring the Fenton-like processes. Reddy and Karri [119] applied electrochemical remediation enhanced by the Fenton-like process to kaolin contaminated with a mixture of Ni and phenanthrene each at a concentration of 500 mg/kg of dry soil. The objective of the coupled remediation processes was simultaneous oxidation of phenanthrene and extraction of Ni. Experiments were conducted using H2 O2 solution in concentrations of 5%, 10%, 20%, and 30% and deionized water as control. Native Fe was used as catalyst for the Fenton-like process. A dc electrical gradient of 1 V/cm was applied and H2 O2 solution was

20

A.T. Yeung, Y.-Y. Gu / Journal of Hazardous Materials 195 (2011) 11–29

introduced at the anode for 4 weeks. The volume of electroosmotic ﬂow was substantial in all the experiments, approximately one pore volume in the control experiment and 1.2–1.6 pore volumes in the H2 O2 experiments. Oxidation of phenanthrene increased with concentration of H2 O2 and a maximum of 56% oxidation was observed with 30% H2 O2 . Nickel was migrated from the anode towards the cathode but it was precipitated near the cathode as a result of the high pH environment. They concluded that optimization of H2 O2 /catalyst concentration and electrical gradient applied, and control of soil pH are required to improve the efﬁciency of oxidation of phenanthrene and extraction of Ni simultaneously. Oonnittan et al. [120] studied the feasibility of electrochemical remediation of hexachlorobenzene (HCB)-contaminated kaolin enhanced by the Fenton process with and without using ␤cyclodextrin to enhance the solubility of HCB in the soil pore ﬂuid. The initial concentration of HCB in kaolin was 100 mg/kg of soil. The native iron in kaolin was utilized to catalyze the Fenton-like reaction and no soluble iron was added during the process. After 15 days of treatment, a maximum remediation efﬁciency of 76% was observed when 30% H2 O2 was used in the absence of ␤cyclodextrin. However, the introduction of ␤-cyclodextrin as an enhancing agent led to a slower rate of oxidation. Tsai et al. [121] studied the feasibility of electrochemical remediation of diesel-contaminated soils enhanced by the use of 0.1 M NaCl as purging solution and corroded iron electrodes. Their experimental results indicate the concentration of total petroleum hydrocarbon diesel in the contaminated soil was reduced from 10,000 to 300 mg/kg by electrokinetically enhanced oxidation in the presence of both 8% H2 O2 and Fe3 O4 (corroded iron electrodes), i.e., remediation efﬁciency of 97%. However, individually applied electrochemical remediation and Fenton oxidation can only yield remediation efﬁciencies of 55% and 27%, respectively. The synergistic effect of the two remediation technologies is thus evident. Oonnittan et al. [122] identiﬁed the importance of efﬁcient oxidant delivery methodologies for effective contaminant oxidation to occur. The success of electrochemical remediation coupled with the Fenton process depends heavily on the good contact between the contaminant and the oxidant facilitated under optimized reaction conditions. Isosaari et al. [123] coupled persulfate oxidation with electrochemical remediation to cleanup creosote-contaminated soil for 8 weeks. Their results reveal that electrokinetically enhanced oxidation with sodium persulfate Na2 S2 O8 resulted in remediation efﬁciency of creosote removal of 35% which is better than that of electrochemical remediation of 24% or persulfate oxidation of 12% individually. The oxidant generated more positive redox potential than electrochemical remediation alone. Moreover, the persulfate treatment decreased the electroosmotic volume ﬂow rate. The results of elemental analyses indicate decrease in the natural Al and Ca concentrations, increase in Zn, Cu, P, and S concentrations, and migration of several metal cations towards the cathode. The effectiveness of electrokinetically enhanced persulfate oxidation for destruction of TCE spiked in a sandy clay soil was evaluated by Yang and Yeh [124]. Their experimental results indicate that electroosmosis could greatly enhance the transport of the injected Na2 S2 O8 from the anode reservoir to the cathode reservoir via the contaminated soil, enhancing the in situ chemical oxidation of TCE. Moreover, the injection of nano-scale Fe3 O4 was observed to have a profound impact in the activation of persulfate oxidation. Reddy and Chinthamreddy [125] studied the electromigration of Cr6+ , NI2+ , and Cd2+ in clayey soils containing different in situ reducing agents in bench-scale experiments. Two different clays, kaolin and glacial till, were used with or without a reducing agent. Kaolin is a soil of low acid/base buffer capacity and glacial till is a soil of high acid/base buffer capacity. The reducing agent used was humic acid, ferrous sulfate, or sodium sulﬁde of concentration of

humic acid, Fe2+ , and S− of 1000 mg/kg soil. The soils were then spiked with Cr6+ , Ni2+ , and Cd2+ in concentrations of 1000, 500, and 250 mg/kg, respectively, and treated by an electrical gradient of 1 V/cm for more than 200 h. The reduction of chromium from Cr6+ to Cr3+ was completed prior to electrochemical remediation. Their results indicate that the extent of Cr6+ reduction was dependent on the type and quantity of reducing agent in the soil in the order of sulﬁde > ferrous iron > humic acid. Moreover, electromigration of Cr6+ was signiﬁcantly retarded in the presence of sulﬁde because of: (1) the opposite directions of migration of Cr6+ and Cr3+ ; (2) sorption and precipitation of Cr3+ in high pH regions near the cathode in kaolin and throughout the glacial till; and (3) sorption of Cr6+ in low pH regions near the anode in both soils. Both Ni2+ and Cd2+ were migrated towards the cathode in kaolin. However, the migration was signiﬁcantly retarded in the presence of sulﬁde due to the pH increase throughout the soil. The initial high pH conditions within the glacial till caused Ni2+ and Cd2+ to precipitate, so the effects of reducing agents were inconsequential. The study demonstrated evidently that the reducing agents, particularly sulﬁde, in soils may affect the redox chemistry and pH of the soil, ultimately affecting the remediation efﬁciency of electrochemical remediation. Weeks and Pamukcu [126] conducted a study to demonstrate the feasibility of in situ reduction of Cr6+ to Cr3+ by introducing ferrous iron Fe2+ , a reducing agent, to the contaminated soil electrokinetically. Their results indicate that the Cr6+ in soils could be effectively reduced to Cr3+ by electrochemical remediation. Moreover, they demonstrated that the Nernst equation may be applicable to model the soil-water system to estimate the concentrations of different Cr species after electrochemical remediation.

5.2. Bioremediation Bioremediation is the use of microorganisms (mainly bacteria) to decompose hazardous contaminants, transform them to less harmful forms, and/or immobilize them under suitable environmental conditions [8]. The success of bioremediation requires the simultaneous existence of microorganisms, contaminants (food for the microorganism), electron acceptors, and essential nutrients for the microorganisms to grow. In ﬁne-grained soils of low hydraulic conductivity, it is difﬁcult to supply microorganism and the required electron acceptors or nutrients to the contaminants, or to supply the contaminants to natural occurring microorganisms. Electrokinetics-enhanced bioremediation or bioelectrokinetics is the technology that couples bioremediation with electrochemical remediation by supplying the microorganisms, electron acceptors, or nutrients to the contaminants, or migrating the contaminants to the microorganisms by electrokinetic ﬂow processes. The ability to directionally transport bacteria from injection points into zones of contamination is a distinct advantage of electrokinetics-enhanced bioremediation for in situ remediation [127]. Electroosmosis and/or electrophoresis have been utilized successfully to inject a Pseudomonas strain (bacterial cell capable of degrading diesel) into diesel-contaminated soil [128]; Sphingomonas sp. L138 and Mycobacterium frederiksbergense LB501TG (polycyclic aromatic hydrocarbon-degrading bacteria) into model aquifers made of glass beads, alluvial sand from Lake Geneva, and historically polluted clayey soil in the laboratory [129]; Pseudomonas putida, Bacillus subtilis, and Klebsiella pneumoniae to stimulate bacterial cell migration and biodegradation of crude oil in soil [130]; Sphingomonas sp. LB126 (ﬂuorene-degrading bacteria) into a laboratory model aquifer [131]; Bacillus spp. (nitrate reducing bacteria) to remove nitrate from soil [132]; B. subtilis LBBMA 155 and nitrogen-starved cells of Pseudomonas sp. LBBMA 81 into a residual

A.T. Yeung, Y.-Y. Gu / Journal of Hazardous Materials 195 (2011) 11–29

soil [133]; and Acidithiobacillus thiooxidans (sulfur-oxidizing bacteria) into tailing soil contaminated by Cd, Cu, Pb, Zn, Co, and As [134]. However, it was observed that electrokinetic transport of strongly charged and highly adhesive cells of M. frederiksbergense LB501TG in different model aquifers was poor [129]. Lee and Kim [135] injected A. thiooxidans (sulfur-oxidizing bacteria) into shooting range soil contaminated by Cu, Zn, and Pb. The bioleaching process improved the extraction efﬁciencies of Cu and Zn by electrochemical remediation. However, PbSO4 , a byproduct of sulfur oxidation, existed as precipitates and was immobile. Nonetheless, the problem was overcome by subsequent injection of EDTA. Electrokinetics was used successfully to inject ammonium nitrogen into ﬁne-grained soil [136], benzoic acid cometabolite into TCE-contaminated soil [137], acetate and phosphate amendment into Cr6+ -contaminated soil [138], KH2 PO4 and triethyl phosphate into kaolin soils [139], oxygenated and nutrient-rich liquid into creosote-contaminated soil [140], and nitrate to toluenecontaminated soil under denitrifying conditions [141]. The results of Schmidt et al. [142] indicate the feasibility of injecting nitrate and ammonium into a very humid clayey silt of high plasticity, high electrical conductivity, low hydraulic conductivity, low density, high acid/base buffer capacity, and high cation exchange capacity. However, injection of phosphorous into this type of soil did not prove to be successful. Lohner et al. [143,144] studied the distributions of microbial electron acceptors nitrate and sulfate and of the nutrients ammonium and phosphate by electrokinetics in a model sandy soil. Their results reveal that the ion distribution in the soil was signiﬁcantly inﬂuenced by the pH proﬁle and the imposed electrical gradient. The results of Xu et al. [145] reveal that ammonium and nitrate ions could be distributed more uniformly in phenanthrene contaminated-soil by reversal of electrode polarity. The results of Jackson et al. [146] indicate electrokinetics could enhance the bioremediation of 2,4-dichlorophenoxyacetic acid-contaminated soil by increasing the bioavailability of the contaminant to microorganism. Similar observation was made by Fan et al. [147] during their study on in situ bioremediation of 2,4dichlorophenol-contaminated soil. Wu et al. [148] demonstrated experimentally that electrokinetic injection of lactate, a negatively charged biodegradable organic, in sand was dependent on electric current density. However, the increase in electric current intensity did not result in a proportional increase in lactate transport due to development of an appreciable electroosmotic ﬂow from the anode to the cathode. Tiehm et al. [149] observed that the microbial activities of vinyl chloride degrading microorganisms were inhibited by electrochemical reaction products when stainless steel electrodes and titanium electrodes with mixed oxide coating type DN201 were used. However, when the electrodes were separated from the microorganisms by bipolar membranes, no inhibition by the electric ﬁeld was observed. Li et al. [150] demonstrated that a dc electric current could stimulate microbial activities and accelerate the biodegradation of petroleum, and there is a strong positive correlation between the electric intensity and the bioremediation efﬁciency of petroleum. The results of Wick et al. [151] suggest that the presence of an electric ﬁeld, if suitably applied, would not inﬂuence the composition and physiology of soil microbial communities and hence would not affect their potential to biodegrade subsurface contaminants. The results of Kim et al. [152] also suggest that the application of electrokinetics could be a promising soil remediation technology if soil parameters, electric current, and electrolyte were suitably controlled based on the understanding of interaction between electrokinetics, contaminants, and indigenous microbial community. Moreover, the increase in soil temperature during electrochemical remediation promotes microbial activities in general.

21

However, the microbial activities can also be inhibited if the soil temperature is higher than 45 ◦ C. 5.3. Permeable reactive barriers A permeable reactive barrier (PRB) is an engineered barrier made of reactive treatment media placed across the ﬂow path of a contaminant plume in aquifer that removes or degrades contaminants in the groundwater ﬂowing through it [153]. It relies on the ﬂow of contaminants through the barrier. The use of granular zero-valent iron (ZVI) mixed with soil to construct PRBs for entrapping or decomposing contaminants in the subsurface has gained widespread acceptance by the environmental remediation and regulatory communities in recent years. Considerable investigation has been conducted to understand the interfacial chemistry of granular iron, and the sorption and degradation mechanisms of contaminants. When a PRB is coupled with electrochemical remediation, the ﬂow of contaminants through the barrier is not provided by the advective transport of contaminants driven by the natural hydraulic gradient of groundwater. It is driven by the electroosmotic ﬂow of soil pore ﬂuid, electromigration of charged species, and/or electrophoresis of charged particulates. In most cases, particularly in ﬁne-grained soils, these transport mechanisms are far more signiﬁcant than that driven by the natural hydraulic gradient of groundwater. The sorption characteristics of most solid particle surfaces are pH-dependent. The degradation reactions of many contaminants are also pH-dependent. As a result, the pH gradient generated by the electrochemical remediation process in the PRB may affect the sorption and degradation mechanisms of the reactive medium in the PRB. The use of enhancement agents in electrochemical reaction would further complicate the situation. Moreover, it is possible to construct a PRB in the subsurface by electrokinetic ﬂow processes. Therefore, there are many additional aspects that need to be considered when a PRB is coupled with the electrochemical remediation process to improve the remediation efﬁciencies of organic, inorganic, and mixed contaminants. 5.3.1. Lasagna process Electrochemical remediation is coupled with sorption/degradation of contaminants in treatment zones installed directly in contaminated soils in the Lasagna process. The Lasagna process is an in situ remediation technique that applies the concept of Integrated In situ Remediation [154]. A dc electric ﬁeld is applied to migrate the contaminants from soil into treatment zones where the contaminants are removed by sorption, immobilization, or degradation as shown in Fig. 5. The technique is called “Lasagna” because of the layered appearance of electrodes and treatment zones. Theoretically, it can remediate organic, inorganic, and mixed contaminants. Electrodes and treatment zones can be of any orientation depending upon the emplacement technology used and the characteristics of the site and contaminant. The treatment process is composed of these key steps [154]: (1) Highly permeable zones in close proximity of the contaminated soil are created by hydrofracturing or similar technologies. Appropriate materials such as sorbents, catalytic agents, microbes, oxidants and buffers are introduced to these highly permeable zones to transform them into treatment zones. (2) Electrokinetic ﬂow processes are utilized to migrate contaminants from soil into treatment zones. Since these zones are located close to each other, the time taken for the contaminants to move from zone to zone can be very short. (3) For highly non-polar contaminants, surfactants can be introduced into the ﬂuid or incorporated into the treatment zones

22

A.T. Yeung, Y.-Y. Gu / Journal of Hazardous Materials 195 (2011) 11–29

5.3.2. Zero-valent iron (ZVI) PRB Chew and Zhang [161] investigated the feasibility of electrochemical remediation coupled with a ZVI PRB installed at the anode to treat a nitrate-contaminated soil according to the chemical reaction, 5Fe0 + 2NO3 − + 12H+ → 5Fe2+ + N2 ↑ + 6H2 O

Fig. 5. Principle of Lasagna Process (after Ho et al. [154]).

to solubilize the organics. For a mixture of organics and metals, the treatment zones can contain sorbents for binding the metals and/or microbes or catalysts for degrading the organics. (4) If needed, the ﬂuid ﬂow direction can be reversed periodically by switching the electrical polarity. The operation would enable multiple passes of the contaminants through the treatment zones for complete sorption/destruction. The polarity reversal also serves to minimize complications associated with longterm operation of uni-directional electrokinetic ﬂow processes. The high pH cathode efﬂuent can be re-circulated through the contaminated soil when the polarity of the electric ﬁeld applied is reversed, i.e., the cathode has been reversed to become the anode and vice versa. The recycling of efﬂuent provides a convenient means for pH neutralization of the contaminated soil and minimization of wastewater generation.

The technique has been proved to be technically feasible in bench-scale laboratory experiments on the degradation of paranitrophenol in kaolinite [154] and ﬁeld-scale experiments on remediation of TCE-contaminated soils at various sites [155–158]. Jackman et al. [159] demonstrated the feasibility of migrating 2,4-dichlorephenoxyacetic acid in contaminated silt soil by electrokinetic ﬂow processes into microorganism active treatment zone for biodegradation of organic contaminants. A bench-scale experiment was conducted by Ma et al. [160] to investigate the simultaneous removal of 2,4-dichlorophenol (2,4DCP) and Cd from a sandy loam by the Lasagna process using a newtype of bamboo charcoal as sorbent and periodic polarity reversals at different intervals. Their results indicate that the Lasagna process was effective in the simultaneous extraction of 2,4-DCP and Cd from sandy soil. Moreover, the extraction efﬁciencies were higher when the electrical polarity was reversed at 24-h intervals.

(17)

The amount of nitrate–nitrogen transformed by electrochemical remediation was increased signiﬁcantly by coupling with a ZVI PRB. The major transformation products were ammonia–nitrogen and nitrogen gas. Moon et al. [162] investigated the mechanisms of TCE degradation during electrochemical remediation coupled with a ZVI PRB. Their results indicate the rate of reductive dechlorination of TCE was improved 1.3–5.8 times of that of a ZVI PRB alone. The most effective conﬁguration of electrode and ZVI PRB for TCE removal was with the cathode installed at the hydraulic down-gradient. The enhancement was attributed to the availability of more electron sources including: (1) the dc power supply; (2) electrolysis of water; (3) oxidation of ZVI; (4) oxidation of dissolved Fe2+ ; (5) oxidation of molecular hydrogen at the cathode; and (6) oxidation of Fe2+ in mineral precipitates. Each of these electron sources was evaluated for their potential inﬂuences on the TCE removal capacity through the electron competition model and energy consumption. A strong correlation between the quantity of electrons generated, removal capacity, and energy-effectiveness was identiﬁed. Yuan [163] investigated the effect of ZVI PRB position and ZVI quantity on the efﬁciency of electrochemical remediation of tetrachloroethylene (PCE)-contaminated clay coupled with a ZVI PRB. The PRB was composed of 2–16 g of ZVI mixed with Ottawa sand in a ratio of 1:2 by weight. Her results indicate that the best position of the PRB was at the cathode and the remediation efﬁciency of PCE was 2.4 times that of electrochemical remediation alone. The remediation efﬁciency also increased with the quantity of ZVI in the barrier. The highest remediation efﬁciency of 90.7% was observed when the quantity of ZVI in the barrier was increased to 16 g. Moreover, it was observed that the more was ZVI in the barrier, the higher was the electroosmotic ﬂow rate, and the lower was ﬁnal soil pH after treatment. The effectiveness of a ZVI PRB barrier installed at the middle of the soil specimen during electrochemical remediation of hyperCr6+ -contaminated clay (2497 mg/kg) was investigated by Weng et al. [164]. The barrier was composed of 1:1 ratio of granular ZVI and sand by weight. Their results indicate that the migration of H+ ions was greatly retarded by the strong opposite migration of anionic CrO4 2− ions, resulting in a reverse electroosmotic ﬂow and development of alkaline zone across the specimen. The alkaline environment promoted the release of Cr6+ from the clay. Chromium removal was indicated by the high Cr6+ concentration in the anolyte and the presence of Cr3+ precipitates in the catholyte. The reduction efﬁciency of Cr6+ to Cr3+ was increased by the ZVI PRB. The electrochemical remediation coupled with a ZVI PRB has transformed the contaminant in the hyper-Cr6+ -contaminated soil to the less toxic form of Cr3+ . Yuan and Chiang [165] investigated the removal mechanisms of As from soil by electrochemical remediation coupled with a PRB made of ZVI and FeOOH. The extraction efﬁciency for As was increased by 60–120% by the PRB. The best performance was achieved when a FeOOH layer was installed at the middle of the soil specimen. The improvement was attributed to higher surface area of FeOOH and the migration of HAsO4 2− towards the anode by electromigration. The presence of As on the surface of the reactive media of the PRB was conﬁrmed by results obtained by SEM coupled with energy dispersive spectroscopy. Moreover, the extraction of As contributed by surface sorption/precipitation on the PRB reactive media was much more than that by the electrokinetic ﬂow

A.T. Yeung, Y.-Y. Gu / Journal of Hazardous Materials 195 (2011) 11–29

23

voltage. The hydraulic conductivity and unconﬁned compressive strength of the iron-rich band so produced were 1 × 10−9 m/s or less and 10.8 N/mm2 , respectively. The barrier may function as a PRB to degrade contaminants or an impervious barrier to contaminant transport. By monitoring the dc electric current intensity passing through the barrier, the integrity of the Fe-rich band may be assessed. Moreover, the barrier may ‘self-heal’ by continuing application of a dc electric current.

Fig. 6. Principle of iron-rich barrier generation by electrokinetics (after Faulkner et al. [169]).

processes. However, electromigration was a more dominant contaminant migration mechanism than the advective transport by electroosmosis. Cang et al. [166] investigated the feasibility of treating a Crcontaminated soil by electrochemical remediation coupled with a ZVI PRB. The reactions between Cr6+ in groundwater and the ZVI in the PRB are as follows: 2Fe0 + Cr2 O7 2− + 7H2 O → 2Cr(OH)3 ↓ + 2Fe(OH)3 ↓ + 2OH− (18) Fe0 + CrO4 2− + 4H2 O → 2Cr(OH)3 ↓ + Fe(OH)3 ↓ + 2OH− ZVI is ultimately oxidized to Fe3+

(19)

which precipitates as Fe(OH)3 , while Cr6+ is reduced to Cr3+ and also precipitates in the PRB. During the reactions, the OH− ions released increases the soil pH and decreases the sorption capacity of Cr6+ on soil particle surfaces. Their results indicate that the technique was feasible for the remediation of Cr-contaminated soil. The maximum remediation efﬁciency of Cr achieved was 72%. The quantities of Cr in the anolyte and catholyte with a PRB were smaller than those without. The position of the PRB affected both the direction and rate of electroosmotic ﬂow. The optimum positions of the PRBs are between the contaminated soil specimen and the electrodes. Wan et al. [167] investigated the feasibility of surfactantenhanced electrochemical remediation coupled with a PRB composed of microscale Pd/Fe for the treatment of a HCBcontaminated soil. The reduction kinetics of HCB by nanoscale Pd/Fe bimetallic particles was faster than that by nanoscale Fe particles. The degradation products of HCB using nanoscale Pd/Fe bimetallic particles have less chloro substituents than those using nanoscale Fe particles. The effects can be attributed to the catalytic effect of Pd on the Fe surface [168]. The nonionic surfactant Triton X-100 was selected as the solubility-enhancing agent. Their results indicate that HCB removal was generally increased by a factor of 4 as HCB was removed from soil through several sequential processes: (a) advective transport of HCB from the anode towards the cathode by electroosmosis; (b) complete sorption/degradation by the reactive Pd/Fe particles in the PRB; and (3) probable electrochemical reactions near the cathode. ZVI PRBs may be constructed in situ by electrokinetics. Faulkner et al. [169] have successfully generated subsurface barriers of continuous Fe-rich precipitates in situ by electrokinetics in their laboratory-scale experiments. Continuous vertical and horizontal Fe-rich bands up to 2 cm thick have been generated by applying a voltage of less than 5 V over a period of 300–500 h, using sacriﬁcial iron electrodes 15–30 cm apart as shown in Fig. 6. The Fe-rich barrier is composed of amorphous iron, goethite, lepidocrocite, maghemite, and native iron. The applied dc electric ﬁeld dissolved the sacriﬁcial anode and injected the Fe ions into the soil. The Fe ions then re-precipitated in an alkaline environment to form the barrier. The thickness of the Fe-rich band increased with the applied

5.3.3. PRBs of different reactive media Chung and Lee [170] investigated the potential use of atomizing slag as an inexpensive PRB reactive medium coupled with electrochemical remediation for simultaneous treatment of soil contaminated by TCE and Cd by laboratory-scale experiments. Their results indicate that the TCE concentration of the efﬂuent through the PRB during electrochemical remediation were much lower than that of electrochemical remediation alone. Some of the TCE passing through the PRB would have been dechlorinated by the atomizing slag as indicated by the higher chloride concentration of the efﬂuent. In general, both the remediation efﬁciencies of TCE and Cd achieved approximately 90%. The removal rate of Cd from the soil specimen was higher than that of TCE as a result of the additional transport by electromigration due to its positive charge. Kimura et al. [171] investigated the possibility of coupling electrochemical remediation with a ferrite treatment zone (FTZ) to treat Cu-contaminated kaolinite. The FTZ was constructed between the cathode and contaminated kaolinite of soil containing polyferric sulfate solution so that the concentration of ferrite in the FTZ was 1000 ppm (mg/kg). Their results indicate 92% of Cu ions in contaminated kaolinite were migrated into the FTZ by electrochemical remediation and ferritized by the alkaline environment generated by the process after 48 h of treatment. The Cu ions were insolubilized by the ferrite reagent in the FTZ and accumulated as copper-ferrite through these chemical reactions [172], nCu2+ + (3 − n)Fe2+ + 6OH− → Cun Fe(3−n) (OH)6

(20)

Cun Fe(3n−1) (OH)6 + (1/2)O2 → Cun Fe(3−n) O4 + 3H2 O

(21)

Barrado et al. [173] suggested the co-precipitation mechanism for Fe2+ and divalent or polyvalent metal ions as follows, xCu2+ +FeSO4 +6NaOH + (1/2)O2 → Cux Fe(3−x) O4 + 3Na2 SO4 + 3H2 O + x[Fe2+ ]

(22)

The copper-ferrite precipitates are magnetic and can be separated from solution easily. Therefore, the advantages of coupling a FTZ with electrochemical remediation include: (1) it is possible collect the extracted heavy metals in a speciﬁc FTZ; (2) the treatment of a large quantity of Cu-rich wastewater produced by electrochemical remediation can be avoided; and (3) there is a possibility that copper-ferrite can be recovered by magnetic separation. It is envisaged in ﬁeld implementation that the FTZ can be constructed near the cathode by injecting ferrite reagent into the soil, and the contaminated water is migrated to the FTZ by the electrochemical remediation process. Afterwards, the FTZ is excavated and the Cu is recovered by appropriate processes, such as soil washing using acid and magnetic separation. The feasibility of electrochemical remediation of Crcontaminated clay enhanced by a PRB made of transformed Red Mud (TRM) was investigated by De Gioannis et al. [174] in bench-scale experiments. The TRM is primarily composed of micron-sized NaOH etched aggregates of (hydrated) Fe oxides (hematite and ferrihydrite 35% by weight) and hydrated alumina (boehmite and gibbsite 20% by weight). These are impregnated by newly formed and more or less soluble alkaline minerals, including sodalite (15% by weight), Ca(OH)2 , hydroxycarbonates

24

A.T. Yeung, Y.-Y. Gu / Journal of Hazardous Materials 195 (2011) 11–29

and calcium alumino silicates (portlandite, calcite, cancrinite, hydrocalumite and aluminohydrocalcite 15 wt%), Mg(OH)2 and magnesium alumino silicates (brucite and hydrotalcite 4 wt%). Their results reveal that the remediation efﬁciency of Cr6+ was proportional to treatment duration. The acidic environment near the anode generated by the electrochemical remediation process improved the sorption capacity of TRM for metal-oxyanions. Therefore, the PRB made of TRM installed near the anode improved the remediation efﬁciency of metal-oxyanions by electrochemical remediation. Yuan et al. [175] investigated the feasibility of surfactantenhanced electrochemical remediation of 1,2-dichlorobenzene (1,2-DCB)-contaminated soil coupled with a carbon nanotube (CNT) PRB installed at the middle of the specimen. CNT is becoming a prominent material being applied in the removal of aqueous and gaseous pollutants due to its high speciﬁc area, high reaction ability, and high electron transfer capacity [176–179]. It is highly expected that CNT will become an effective reactive medium in the PRB for removal of organic contaminants from the subsurface. Their results indicate the remediation efﬁciency of electrochemical remediation could be signiﬁcantly improved by the introduction of SDS and coupling with a CNT PRB. Removal of 1,2 DCB was primarily contributed by surface sorption of the contaminant on CNT rather than by electrokinetic ﬂow processes. However, electrophoresis of anionic SDS micelles towards the anode became a more critical contributor when the surfactant was used as processing ﬂuid. An enhanced electrochemical remediation process coupled with a PRB made of carbon nanotube coated with cobalt (CNT-Co) was investigated for As5+ removal by Yuan et al. [175]. Their experimental results indicate the PRB made of CNT did not contribute much to the remediation efﬁciency of As5+ . However, the PRB made of CNTCo increased the remediation efﬁciency from 35% to 62%. The better remediation efﬁciency of electrochemical remediation enhanced by the PRB made of CNT-Co was attributed to the higher sorption of As5+ onto CNT-Co surfaces than CNT surfaces. Removal of As5+ was thus primarily contributed by the surface sorption of As5+ onto CNT-Co instead of the electrokinetic ﬂow processes. The surface characteristics of CNT-Co, as revealed by SEM coupled with energy dispersive spectroscopy, evidently conﬁrmed that As was adsorbed on the passive layer surface. The results of an investigation using sequential extraction revealed that the binding between As5+ and soil particles was shifted considerably from strong binding forms, i.e., Fe–Mn oxide, organic, and residual, to weak binding forms, i.e., exchange and carbonate, after electrochemical remediation. Han et al. [180] investigated the feasibility of enhancing the electrochemical remediation of Cu-contaminated kaolinite by coupling with a PRB made of carbonized foods waste (CFW). The CFW is composed of more than 85% oxygen, calcium and carbon. The size range of the CFW is 75–150 ␮m within porous structures. The speciﬁc area, total pore volume, and average pore diameter of ˚ CFW were determined to be 14.16 m2 /g, 46.9 mm3 /g, and 132.4 A, respectively. The sorption efﬁciency of CFW used as a PRB reactive medium was found to be 4–8 times more efﬁcient than that of zeolite. Throughout the experiment, an electrical gradient of 1 V/cm was implemented and acetic acid was injected from the anode to improve the remediation efﬁciency. Their results indicate the installation of a CFW PRB did not inﬂuence the electroosmotic ﬂow. However, the electroosmotic ﬂow was increased by the injection of CH3 COOH with time. The majority of Cu2+ extracted from kaolinite was sorbed by CFW. 5.3.4. PRB – summary The remediation efﬁciency of electrochemical remediation can be enhanced by coupling with a PRB. Depending on the type of contaminant to be treated, different reactive media of the PRB can be utilized. However, experimental results from different researchers

to date indicate the remediation efﬁciency is primarily contributed by the sorption capacity of the reactive medium of the PRB. The role of electrochemical remediation lies in the migration of contaminants towards the PRB, generation of an acidic environment near the anode, and generation of an alkaline environment near the cathode. However, the sorption capacity of the reactive medium of the PRB can be promoted by the acidity or alkalinity of the environment to improve the remediation efﬁciency. 5.4. Phytoremediation Phytoremediation is the use of plants to remove, degrade, or sequester inorganic and organic contaminants from soil and/or groundwater [8]. It is an emerging cost-effective alternative to conventional remediation technologies. However, contaminants may have limited bioavailability in the soil, methods to facilitate its transport to the shoots and roots of plants are thus required for successful application of phytoremediation. O’Connor et al. [181] investigated the use of coupling phytoremediation with electrochemical remediation to decontaminate soils contaminated by Cu, Cd, and As. It can be observed in their results that the dc electric ﬁeld could transport metal contaminants from the anode towards the cathode, and generate signiﬁcant changes in soil pH. Moreover, perennial ryegrass could be grown in the treated soils to take up a proportion of the mobilized metals into its shoot system. In their bench-scale studies, Lim et al. [182] demonstrated the effectiveness of Indian mustard (Brassica juncea) grown in contaminated soil in accumulating high tissue concentration of Pb, with the addition of EDTA in the soil and the application of a dc electric ﬁeld around the plants. The accumulation of Pb in the shoots using EDTA and a dc electric ﬁeld was increased by two- to fourfold that of using EDTA only. Similarly, the shoot Cu concentrations of ryegrass in the phytoremediation of contaminated soil enhanced by EDTA and EDDS was increased by 46% and 61%, respectively when coupled with electrochemical remediation [183]. Aboughalma et al. [184] studied the use of potato tubers to decontaminate soils polluted with Zn, Pb, Cu, and Cd in their laboratory-scale experiments using: (1) a dc electric ﬁeld; (2) an alternating-current (ac) electric ﬁeld; and (3) no electric ﬁeld, i.e., the control. Their results reveal that metal accumulation in plant roots treated with electrical ﬁelds was generally higher than the control. The overall metal uptake in plant shoots treated with a dc electric ﬁeld was lower than those treated with an ac electric ﬁeld and the control, although there was a higher accumulation of Zn and Cu in the plant roots treated with electrical ﬁelds. The Zn uptake in plant shoots treated with an ac electric ﬁeld was higher than that treated with a dc electric ﬁeld and the control. Zn and Cu accumulation in plant roots treated with a dc electric ﬁeld and an ac electric ﬁeld were similar and higher than that of the control. Bi et al. [185] studied the growth of rapeseed (Brassica napus) plants and tobacco (Nicotiana tabacum) plants under a dc electric ﬁeld and an ac electric ﬁeld and their abilities to decontaminate a soil contaminated by Cd, and a soil contaminated by Cd, Zn, and Pb. Their results reveal that the biomass production of rapeseed plants was enhanced by the ac electric ﬁeld. However, the ac electric ﬁeld has no effect on the biomass production of tobacco plants and the dc electric ﬁeld even has a negative effect. Moreover, metal uptake by the rapeseed plant shoot was enhanced by the application of the ac electric ﬁeld. Cang et al. [186] studied the effects of dc electric current on the growth of Indian mustard (B. juncea) and speciation of soil heavy metals in pot experiments for 35 days. The soil was contaminated by Cd, Cu, Pb, and Zn. Their results indicate that plant uptake of metals was increased by the electrokinetics-assisted phytoremediation. Moreover, electrical gradient was identiﬁed to be the most

A.T. Yeung, Y.-Y. Gu / Journal of Hazardous Materials 195 (2011) 11–29

important factor in affecting the plant growth, soil properties, and metal concentrations in the soil and plant. 5.5. Ultrasonication Acoustic waves or acoustic energy can enhance migration of contaminants in soil and facilitate their subsequent remediation and/or removal by these effects: (1) increase in kinetic energy of soil pore ﬂuid, causing rise in temperature, and increase in volume and pressure of the soil pore ﬂuid; (2) decrease in viscosity of soil pore ﬂuid, increasing the volume ﬂow rate of the soil pore ﬂuid; (3) increase in molecular motion of contaminants, inducing disintegration and mobilization of contaminants sorbed on soil particle surfaces; and (4) cavitation (forming bubbles) in soil pore ﬂuid, causing increase in porosity and hydraulic conductivity of soil [187]. Chung and Kamon [187] studied the performance of electrochemical remediation coupled with ultrasonication on simultaneous remediation of Pb and phenanthrene from contaminated natural clay. Their bench-scale experiments were conducted using specially designed and fabricated devices. Their experimental results reveal that both the ﬂuid outﬂow rate and remediation efﬁciencies for both heavy metal and PAH were increased by the coupled remediation technologies in comparison to electrochemical remediation alone. The average outﬂow rate was increased from 120 mL/h to 143 mL/h, an increase of 19% by the coupling effects of electrokinetic and ultrasonic phenomena. The average remediation efﬁciency for Pb was increased from 88% to 91%, an increase of 3.4%; and the average remediation efﬁciency for phenanthrene was increased from 85% to 90%, an increase of 5.9%. Chung [188] evaluated the performance of four remediation technologies, i.e., soil ﬂushing, electrochemical remediation, ultrasonication, and electrochemical remediation coupled with ultrasonication, in the remediation of river sand from Korea contaminated by diesel fuel and Cd. His results indicate the coupled remediation technologies increased both the volume ﬂow rate and contaminant extraction efﬁciencies. After 100 min, the ﬁnal accumulated ﬂow volume was 2200 mL, 2400 mL, 3800 mL, and 4000 mL by soil ﬂushing, electrochemical remediation, ultrasonication, and electrochemical remediation coupled with ultrasonication, respectively. The ﬁnal accumulated ﬂow volume was thus increased by 9%, 73%, and 82% by electrochemical remediation, ultrasonication, and electrochemical remediation coupled with, ultrasonication, respectively. The remediation efﬁciencies for diesel fuel were 65%, 67%, 85%, and 87% by soil ﬂushing, electrochemical remediation, ultrasonication, and electrochemical remediation coupled with ultrasonication, respectively, Similarly, the remediation efﬁciencies for Cd were 62%, 76%, 65%, and 83%, respectively. It is evident that electrochemical remediation coupled with ultrasonication is the most effective technique to extract heavy metal and hydrocarbon simultaneously from the contaminated sandy soil. Moreover, electrochemical remediation was observed to be the most effective method for the treatment of heavy metal, e.g., Cd, while ultrasonic remediation was the most effective for hydrocarbon, e.g., diesel fuel. As a result, the coupled techniques can be used effectively to extract both the heavy metal and hydrocarbon from contaminated soils simultaneously. Pham et al. [189] studied the performance of electrochemical remediation enhanced by ultrasonication in the cleanup of kaolin contaminated by a mixture of three persistent organic pollutants: HCB, phenanthrene, and ﬂuoranthene. Their bench-scale experimental results conclude that the remediation efﬁciencies for these three persistent organic pollutants by electrochemical remediation coupled with ultrasonication was higher than those of electrochemical remediation alone. Although the ultrasonic enhancement could increase both the electric current intensity and electroos-

25

motic volume ﬂow rate, it could only increase the remediation efﬁciency for less than 10%. HCB is the most difﬁcult contaminant to extract because of its high stability, while ﬂuoranthene is the easiest contaminant to extract. Enhancement of electrochemical remediation by ultrasonication can be considered as one of the feasible technology to extract PAHs from contaminated soil. Pham et al. [162] studied the feasibility of using electrochemical remediation enhanced by ultrasonication or the surfactant 2-hydroxylpropyl-␤-cyclodextrin to remediate soil contaminated by the hydrophobic compounds of HCB and phenanthrene. Their results indicate that both contaminants could be mobilized by electrochemical remediation enhanced by either ultrasonication or surfactant. However, it is more difﬁcult to extract HCB because of its stability and low water-solubility. Moreover, remediation of phenanthrene enhanced by ultrasonication was more efﬁcient than that by surfactant, as ultrasound can degrade the contaminant through oxidation by free radicals. Shrestha et al. [190] utilized electrochemical remediation coupled with ultrasonication to treat kaolin contaminated by chrysene of concentrations of 25, 50, 75, and 100 mg/kg. Their results indicate the coupled technologies could improve the remediation efﬁciency of electrochemical remediation. Moreover, the remediation efﬁciency decreased with increase in the initial concentration of chrysene.

5.6. Other remediation technologies There are many mature remediation technologies for contaminated soil and groundwater [8,105], and they can potentially be coupled with electrochemical remediation to enhance their individual remediation efﬁciencies synergistically. However, these possibilities have yet to be investigated. For example, production of electrochemical oxidation equivalents in situ by inserting anodes in contaminated soil appears to be a promising idea, but the approach is proven to have poor remediation efﬁciency and the effects are much localized in soil. However, in situ production of oxidants has many advantages: (1) oxidants of short lifetimes can be used in the remediation process; (2) no stabilization of peroxides is necessary; (3) the hazard of storing large quantities of chemicals is avoided; and (4) logistics of handling chemicals is much simpler. A new approach is being investigated by Wesner et al. [191] to separate the in situ production of tailored oxidants and the transport of the oxidants by electrokinetics. Thermal desorption is a technology that heats contaminated soil or sludge in situ or ex situ to volatize the contaminants and remove them from soil [8]. Volatile and semi-volatile organics are removed from contaminated soil in thermal desorbers at 100–300 ◦ C for low-temperature thermal desorption, or at 300–550 ◦ C for hightemperature thermal desorption [192]. When a dc or ac electrical current is ﬂowing through a contaminated soil, resistive or ohmic heating occurs. The heating can be used to accelerate many chemical and biological reactions occurring in the contaminated soil, and to modify many physical properties of contaminants. For example, heating can be used to increase desorption of many organic contaminants from soil particle surfaces and to remove dense non-aqueous phase liquids. Increased temperature may increase the aqueous solubility, decrease the density, decrease the viscosity, and increase the volatilization of organic contaminants, facilitating their transport in soil. Elevated temperatures not exceeding the temperature tolerance of microbial consortia can increase their metabolic activity and bioavailability, resulting in enhancement of biodegradation of organic contaminants. However, these thermal effects during electrochemical remediation have not been well studied to date [193].

26

A.T. Yeung, Y.-Y. Gu / Journal of Hazardous Materials 195 (2011) 11–29

6. Conclusions Electrochemical remediation is a promising technology for the remediation of ﬁne-grained soil contaminated by inorganic, organic, and mixed contaminants. However, enhancement techniques are often required to improve the remediation efﬁciency of the technology. A comprehensive review on techniques to enhance electrochemical remediation of contaminated ﬁne-grained materials is given in this paper. A comprehensive and updated list of references is also provided for the reader who is interested in a particular enhancement technique to perform further study. Acknowledgements The sabbatical leave of the ﬁrst author in California supported by the Faculty of Engineering of The University of Hong Kong to complete these review papers is appreciated. The ﬁnancial support for the doctoral study of the second author provided by The University of Hong Kong is gratefully acknowledged. References [1] A.T. Yeung, Electrokinetic ﬂow processes in porous media and their applications, in: M.Y. Corapcioglu (Ed.), Advances in Porous Media, vol. 2, Elsevier, Amsterdam, 1994, pp. 309–395. [2] A.T. Yeung, Y.-Y. Gu, Use of chelating agents in electrochemical remediation of contaminated soil, in: D.C.W. Tsang, I.M.C. Lo (Eds.), Applications of Chelating Agents for Land Decontamination Technologies, ASCE Press, Reston, 2011. [3] J.K. Mitchell, K. Soga, Fundamentals of Soil Behavior, 3rd ed., John Wiley & Sons, Hoboken, 2005. [4] A.T. Yeung, Milestone developments, myths, and future directions of electrokinetic remediation, Sep. Purif. Technol. 79 (2011) 124–132. [5] A.T. Yeung, Fundamental aspects of prolonged electrokinetic ﬂows in kaolinites, Geomech. Geoeng.: Int. J. 1 (2006) 13–25. [6] A.T. Yeung, Geochemical processes affecting electrochemical remediation, in: K.R. Reddy, C. Cameselle (Eds.), Electrochemical Remediation Technologies for Polluted Soils, Sediments and Groundwater, John Wiley & Sons, Hoboken, 2009, pp. 65–94. [7] A.T. Yeung, Contaminant extractability by electrokinetics, Environ. Eng. Sci. 23 (2006) 202–224. [8] A.T. Yeung, Remediation technologies for contaminated sites, in: Y. Chen, X. Tang, L. Zhan (Eds.), Advances in Environmental Geotechnics, Zhejiang University Press, Hangzhou, 2009, pp. 328–369. [9] G. Sposito, On points of zero charge, Environ. Sci. Technol. 32 (1998) 2815–2819. [10] D. Lestan, B. Kos, Soil washing using a biodegradable chelator, in: B. Nowack, J.M. VanBriesen (Eds.), Biogeochemistry of Chelating Agents, ACS Symposium Series 910, American Chemical Society, Washington, DC, 2005, pp. 383–397. [11] Y.-Y. Gu, A.T. Yeung, Desorption of cadmium from a natural Shanghai clay using citric acid industrial wastewater, J. Hazard. Mater. 191 (2011) 144–149. [12] Y.-Y. Gu, A.T. Yeung, A. Koenig, H.-J. Li, Effects of chelating agents on zeta potential of cadmium-contaminated natural clay, Sep. Sci. Technol. 44 (2009) 2203–2222. [13] M. Sillanpaa, A. Oikari, Assessing the impact of complexation by EDTA and DTPA on heavy metal toxicity using Microtox bioassay, Chemosphere 32 (1996) 1485–1497. [14] N. Dirilgen, Effects of pH and chelator EDTA on Cr toxicity and accumulation in Lemma minor, Chemosphere 37 (1998) 771–783. [15] B. Nortemann, Biodegradation of EDTA, Appl. Microbiol. Biotechnol. 51 (1999) 751–759. [16] D. Lestan, C.L. Luo, X.D. Li, The use of chelating agents in the remediation of metal-contaminated soils: a review, Environ. Pollut. 153 (2008) 3–13. [17] C.D. Cox, M.A. Shoesmith, M.M. Ghosh, Electrokinetic remediation of mercurycontaminated soils using iodine/iodide lixiviant, Environ. Sci. Technol. 30 (1996) 1933–1938. [18] K.R. Reddy, C. Chaparro, R.E. Saichek, Removal of mercury from clayey soils using electrokinetics, J. Environ. Sci. Health A – Tox. Hazard. Subst. Environ. Eng. 38 (2003) 307–338. [19] P. Suèr, T. Lifvergren, Mercury-contaminated soil remediation by iodide and electroreclamation, J. Environ. Eng., ASCE 129 (2003) 441–446. [20] Z.M. Shen, J.D. Zhang, L.Y. Qu, Z.Q. Dong, S.S. Zheng, W.H. Wang, A modiﬁed EK method with an I− /I2 lixiviant assisted and approaching cathodes to remedy mercury contaminated ﬁeld soils, Environ. Geol. 57 (2009) 1399–1407. [21] T. Hakansson, P. Suer, B. Mattiasson, B. Allard, Sulphate reducing bacteria to precipitate mercury after electrokinetic soil remediation, Int. J. Environ. Sci. Technol. 5 (2008) 267–274. [22] O. Hanay, H. Hasar, N.N. Kocer, O. Ozdemir, Removal of Pb from sewage sludge by electrokinetics: effect of pH and washing solution type, Environ. Technol. 30 (2009) 1177–1185.

[23] G.N. Kim, Y.H. Jung, J.J. Lee, J.K. Moon, C.H. Jung, An analysis of a ﬂushing effect on the electrokinetic-ﬂushing removal of cobalt and cesium from a soil around decommissioning site, Sep. Purif. Technol. 63 (2008) 116–121. [24] L.M. Rajic, B.D. Dalmacija, J.S. Trickovic, M.B. Dalmacija, D.M. Krcmar, Behavior of zinc, nickel, copper and cadmium during the electrokinetic remediation of sediment from the Great Backa Canal (Serbia), J. Environ. Sci. Health A – Tox. Hazard. Subst. Environ. Eng. 45 (2010) 1134–1143. [25] D.M. Zhou, L. Cang, A.N. Alshawabkeh, Y.J. Wang, X.Z. Hao, Pilot-scale electrokinetic treatment of a Cu contaminated red soil, Chemosphere 63 (2006) 964–971. [26] K. Maturi, K.R. Reddy, Simultaneous removal of organic compounds and heavy metals from soils by electrokinetic remediation with a modiﬁed cyclodextrin, Chemosphere 63 (2006) 1022–1031. [27] S.O. Ko, M.A. Schlautman, E.R. Carraway, Cyclodextrin-enhanced electrokinetic removal of phenanthrene from a model clay soil, Environ. Sci. Technol. 34 (2000) 1535–1541. [28] A.P. Khodadoust, K.R. Reddy, O. Narla, Cyclodextrin-enhanced electrokinetic remediation of soils contaminated with 2,4-dinitrotoluene, J. Environ. Eng., ASCE 132 (2006) 1043–1050. [29] S.H. Yuan, M. Tian, X.H. Lu, Electrokinetic movement of hexachlorobenzene in clayed soils enhanced by Tween 80 and beta-cyclodextrin, J. Hazard. Mater. 137 (2006) 1218–1225. [30] T.P. Li, S.H. Yuan, J.Z. Wan, L. Lin, H.Y. Long, X.F. Wu, X.H. Lu, Pilot-scale electrokinetic movement of HCB and Zn in real contaminated sediments enhanced with hydroxypropyl-␤-cyclodextrin, Chemosphere 76 (2009) 1226–1232. [31] T.P. Li, S.H. Yuan, J.Z. Wan, X.H. Lu, Hydroxypropyl-␤-cyclodextrin enhanced electrokinetic remediation of sediment contaminated with HCB and heavy metals, J. Hazard. Mater. 176 (2010) 306–312. [32] C.L. Wan, M.A. Du, D.J. Lee, X. Yang, W.C. Ma, L.N. Zheng, Electrokinetic remediation and microbial community shift of ␤-cyclodextrin-dissolved petroleum hydrocarbon-contaminated soil, Appl. Microbiol. Biotechnol. 89 (2011) 2019–2025. [33] J.L. Chen, S.F. Yang, C.C. Wu, S. Ton, Effect of ammonia as a complexing agent on electrokinetic remediation of copper-contaminated soil, Sep. Purif. Technol. 79 (2011) 157–163. [34] S.C. Lin, Biosurfactants: recent advances, J. Chem. Technol. Biotechnol. 66 (1996) 109–120. [35] C.N. Mulligan, R.N. Yong, B.F. Gibbs, Surfactant-enhanced remediation of contaminated soil: a review, Eng. Geol. 60 (2001) 371–380. [36] E. Ruckenstein, R. Nagarajan, Critical micelle concentration – transition point for micellar size distribution, J. Phys. Chem. 79 (1975) 2622–2626. [37] W. Chu, C.Y. Kwan, Remediation of contaminated soil by solvent/surfactant system, Chemosphere 53 (2003) 9–15. [38] C.N. Mulligan, F. Eftekhari, Remediation with surfactant foam of PCPcontaminated soil, Eng. Geol. 70 (2003) 269–279. [39] D. Fabbri, A. Crime, M. Davezza, C. Medana, C. Baiocchi, A.B. Prevot, E. Pramauro, Surfactant-assisted removal of swep residues from soil and photocatalytic treatment of the washing wastes, Appl. Catal. B – Environ. 92 (2009) 318–325. [40] C.N. Mulligan, Recent advances in the environmental applications of biosurfactants, Curr. Opin. Colloid Interface Sci. 14 (2009) 372–378. [41] R.A. Doong, Y.W. Wu, W.G. Lei, Surfactant enhanced remediation of cadmium contaminated soils, Water Sci. Technol. 37 (1998) 65–71. [42] J.H. Harwell, Factors affecting surfactant performance in groundwater remediation applications, in: D.A. Sabitini, R.C. Knox (Eds.), Transport and Remediation of Subsurface Contaminants: Colloidal, Interfacial, and Surfactant Phenomena, ACS Symposium Series 491, American Chemical Society, Washington, DC, 1992, pp. 124–132. [43] D.A. Sabatini, R.C. Knox, J.H. Harwell, Emerging technologies in surfactantenhanced subsurface remediation, in: D.A. Sabitini, R.C. Knox, J.H. Harwell (Eds.), Surfactant-enhanced Subsurface Remediation: Emerging Technologies, ACS Symposium Series 594, American Chemical Society, Washington, DC, 1995, pp. 1–9. [44] D.G. Cooper, Biosurfactants, Microbiol. Sci. 3 (1986) 145–149. [45] C.C. West, J.H. Harwell, Surfactant and subsurface remediation, Environ. Sci. Technol. 26 (1992) 2324–2330. [46] J.D. Desai, I.M. Banat, Microbial production of surfactants and their commercial potential, Microbiol. Mol. Biol. Rev. 61 (1997) 47–64. [47] E.Z. Ron, E. Rosenberg, Biosurfactants and oil bioremediation, Curr. Opin. Biotechnol. 13 (2002) 249–252. [48] S. Mukherjee, P. Das, R. Sen, Towards commercial production of microbial surfactants, Trends Biotechnol. 24 (2006) 509–515. [49] K. Muthusamy, S. Gopalakrishnan, T.K. Ravi, P. Sivachidambaram, Biosurfactants: properties, commercial production and application, Curr. Sci. 94 (2008) 736–747. [50] G. Soberón-Chávez, F. Lépine, E. Déziel, Production of rhamnolipids by Pseudomonas aeruginosa, Appl. Microbiol. Biotechnol. 68 (2005) 718–725. [51] Y. Uchida, R. Tsuchiya, M. Chino, J. Hirano, T. Tabuchi, Extracellular accumulation of mono- and di-succinoyl trehalose lipids by a strain of Rhodococcus erythropolis grown on n-alkanes, Agric. Biol. Chem. 53 (1989) 757–763. [52] Y. Hirata, M. Ryua, Y. Odaa, K. Igarashia, A. Nagatsukaa, T. Furutaa, M. Sugiuraa, Novel characteristics of sophorolipids, yeast glycolipid biosurfactants, next term as biodegradable low-foaming surfactants, J. Biosci. Bioeng. 108 (2009) 142–146. [53] G.S. Kiran, T.A. Thomas, J. Selvin, B. Sabarathnam, A.P. Lipton, Optimization and characterization of a new lipopeptide biosurfactant produced by marine

A.T. Yeung, Y.-Y. Gu / Journal of Hazardous Materials 195 (2011) 11–29

[54]

[55]

[56]

[57]

[58]

[59]

[60]

[61]

[62] [63]

[64]

[65]

[66] [67]

[68]

[69]

[70]

[71]

[72]

[73] [74] [75]

[76]

[77]

[78]

[79]

[80]

[81]

Brevibacterium aureum MSA13 in solid state culture, Bioresour. Technol. 101 (2010) 2389–2396. T.H. Nielsen, J. Sorensen, Production of cyclic lipopeptides by Pseudomonas ﬂuorescens strains in bulk soil and in the sugar beet rhizosphere, Appl. Environ. Microbiol. 69 (2003) 861–868. J.M. Raaijmakers, I. De Bruijn, M.J.D. De Kock, Cyclic lipopeptide production by plant-associated Pseudomonas spp.: diversity, activity, biosynthesis, and regulation, Mol. Plant-Microbe Interact. 19 (2006) 699–710. C.S. Jeon, J.S. Yang, K.J. Kim, K. Baek, Electrokinetic removal of petroleum hydrocarbon from residual clayey soil following a washing process, Clean-Soil Air Water 38 (2010) 189–193. K.R. Reddy, R.E. Saichek, Effect of soil type on electrokinetic removal of phenanthrene using surfactants and cosolvents, J. Environ. Eng., ASCE 129 (2003) 336–346. R.E. Saichek, K.R. Reddy, Surfactant-enhanced electrokinetic remediation of polycyclic aromatic hydrocarbons in heterogeneous subsurface environments, J. Environ. Eng. Sci. 4 (2005) 327–339. J.W. Yang, Y.J. Lee, J.Y. Park, S.J. Kim, J.Y. Lee, Application of APG and Calfax 16L-35 on surfactant-enhanced electrokinetic removal of phenanthrene from kaolinite, Eng. Geol. 77 (2005) 243–251. J.Y. Park, H.H. Lee, S.J. Kim, Y.J. Lee, J.W. Yang, Surfactant-enhanced electrokinetic removal of phenanthrene from kaolinite, J. Hazard. Mater. 140 (2007) 230–236. J.H. Chang, Z.M. Qiang, C.P. Huang, A.V. Ellis, Phenanthrene removal in unsaturated soils treated by electrokinetics with different surfactants-Triton X-100 and rhamnolipid, Colloids Surf. A 348 (2009) 157–163. M.T. Alcántara, J. Gómez, M. Pazos, M.A. Sanromán, Electrokinetic remediation of PAH mixtures from kaolin, J. Hazard. Mater. 179 (2010) 1156–1160. T.D. Pham, R.A. Shrestha, M. Sillanpaa, Removal of hexachlorobenzene and phenanthrene from clayey soil by surfactant- and ultrasound-assisted electrokinetics, J. Environ. Eng., ASCE 136 (2010) 739–742. K.R. Reddy, C. Cameselle, P. Ala, Integrated electrokinetic-soil ﬂushing to remove mixed organic and metal contaminants, J. Appl. Electrochem. 40 (2010) 1269–1279. C. Yuan, C.H. Hung, W.L. Huang, Enhancement with carbon nanotube barrier on 1,2-dichlorobenzene removal from soil by surfactant-assisted electrokinetic (SAEK) process – the effect of processing ﬂuid, Sep. Sci. Technol. 44 (2009) 2284–2303. A. Karagunduz, A. Gezer, G. Karasuloglu, Surfactant enhanced electrokinetic remediation of DDT from soils, Sci. Total Environ. 385 (2007) 1–11. C. Yuan, C.H. Weng, Remediating ethylbenzene-contaminated clayey soil by a surfactant-aided electrokinetic (SAEK) process, Chemosphere 57 (2004) 225–232. A.Y. Kolosov, K.I. Popov, N.A. Shabanova, A.A. Artemeva, B.M. Kogut, A.S. Frid, V.Y. Zelvenskii, E.M. Urinovich, Electrokinetic removal of hydrophobic organic compounds from soil, Russian J. Appl. Chem. 74 (2001) 631–635. J. Kim, K. Lee, Effects of electric ﬁeld directions on surfactant enhanced electrokinetic remediation of diesel-contaminated sand column, J. Environ. Sci. Health A – Tox. Hazard. Subst. Environ. Eng. 34 (1999) 863–877. H. Han, Y.J. Lee, S.H. Kim, J.W. Yang, Electrokinetic remediation of soil contaminated with diesel oil using EDTA-cosolvent solutions, Sep. Sci. Technol. 44 (2009) 2437–2454. O. Gonzini, A. Plaza, L. Di Palma, M.C. Lobo, Electrokinetic remediation of gasoil contaminated soil enhanced by rhamnolipid, J. Appl. Electrochem. 40 (2010) 1239–1248. A. Groboillot, F. Portet-Koltalo, F. Le Derf, M.J.G. Feuilloley, N. Orange, C.D. Poc, Novel application of cyclolipopeptide amphis feasibility study as additive to remediate polycyclic aromatic hydrocarbon (PAH) contaminated sediments, Int. J. Mol. Sci. 12 (2011) 1787–1806. A. Kaya, Y. Yukselen, Zeta potential of soils with surfactants and its relevance to electrokinetic remediation, J. Hazard. Mater. 120 (2005) 119–126. C.H. Weng, C. Yuan, Removal of Cr(III) from clay soils by electrokinetics, Environ. Geochem. Health 23 (2001) 281–285. P.W. Chang, Y.C. Tan, I.S. Wang, On-farm remediation and fertility enhancement of cadmium-contaminated silt loam by electrokinetic treatment process in Taiwan, Irrigation Drain. 57 (2008) 187–201. A. Giannis, E. Gidarakos, A. Skouta, Application of sodium dodecyl sulfate and humic acid as surfactants on electrokinetic remediation of cadmiumcontaminated soil, Desalination 211 (2007) 249–260. R.K. Srivastava, R.P. Tiwari, P. Bala Ramudu, Removal of cadmium using electrokinetics – effect of enhancement agents, in: Proc. 1st Sri Lankan Geotech. Soc. Int. Conf. Soil Rock Eng., Colombo, 2007. R.E. Saichek, K.R. Reddy, Effects of system variables on surfactant enhanced electrokinetic removal of polycyclic aromatic hydrocarbons from clayey soils, Environ. Technol. 24 (2003) 503–515. J.Z. Wan, S.H. Yuan, J. Chen, T.P. Li, L. Lin, X.H. Lu, Solubility-enhanced electrokinetic movement of hexachlorobenzene in sediments: a comparison of cosolvent and cyclodextrin, J. Hazard. Mater. 166 (2009) 221–226. A. Li, K.A. Cheung, K.R. Reddy, Cosolvent-enhanced electrokinetic remediation of soils contaminated with phenanthrene, J. Environ. Eng., ASCE 126 (2000) 527–533. K.R. Reddy, P.R. Ala, S. Sharma, S.N. Kumar, Enhanced electrokinetic remediation of contaminated manufactured gas plant soil, Eng. Geol. 85 (2006) 132–146.

27

[82] K. Maturi, K.R. Reddy, Cosolvent-enhanced desorption and transport of heavy metals and organic contaminants in soils during electrokinetic remediation, Water Air Soil Pollut. 189 (2008) 199–211. [83] T.F. Coletta, C.J. Bruell, D.K. Ryan, H.I. Inyang, Cation-enhanced removal of lead from kaolinite by electrokinetics, J. Environ. Eng., ASCE 123 (1997) 1227–1233. [84] K.R. Reddy, S. Chinthamreddy, Enhanced electrokinetic remediation of heavy metals in glacial till soils using different electrolyte solutions, J. Environ. Eng., ASCE 130 (2004) 442–455. [85] R.J. Hunter, Zeta Potential in Colloid Science: Principles and Applications, Academic Press, London, 1981. [86] A.T. Yeung, Diffuse double layer equations in SI units, J. Geotech. Eng., ASCE 118 (1992) 2000–2005. [87] S.K. Puppala, A.N. Alshawabkeh, Y.B. Acar, R.J. Gale, M. Bricka, Enhanced electrokinetic remediation of high sorption capacity soil, J. Hazard. Mater. 55 (1997) 203–220. [88] H.H. Lee, J.W. Yang, A new method to control electrolytes pH by circulation system in electrokinetic soil remediation, J. Hazard. Mater. 77 (2000) 227–240. [89] R.E. Saichek, K.R. Reddy, Effect of pH control at the anode for the electrokinetic removal of phenanthrene from kaolin soil, Chemosphere 51 (2003) 273–287. [90] R.E. Hicks, S. Tondorf, Electrorestoration of metal contaminated soils, Environ. Sci. Technol. 28 (1994) 2203–2210. [91] T. Rødsand, Y.B. Acar, G. Breedveld, Electrokinetic extraction of lead from spiked Norwegian marine clay, in: Y.B. Acar, D.E. Daniel (Eds.), Characterization, Containment, Remediation, and Performance in Environmental Geotechnics, Geotech. Spec. Publ. No. 46, vol. 2, ASCE, New York, 1995, pp. 1518–1534. [92] B.E. Reed, M.T. Berg, J.C. Thompson, J.H. Hatﬁeld, Chemical conditioning of electrode reservoirs during electrokinetic soil ﬂushing of Pb-contaminated silt loam, J. Environ. Eng., ASCE 121 (1995) 805–815. [93] D.M. Zhou, C.F. Deng, L. Cang, A.N. Alshawabkeh, Electrokinetic remediation of a Cu–Zn contaminated red soil by controlling the voltage and conditioning catholyte pH, Chemosphere 61 (2005) 519–527. [94] E. Gidarakos, A. Giannis, Chelate agents enhanced electrokinetic remediation for removal cadmium and zinc by conditioning catholyte pH, Water Air Soil Pollut. 172 (2006) 295–312. [95] B.G. Ryu, G.Y. Park, J.W. Yang, K. Baek, Electrolyte conditioning for electrokinetic remediation of As, Cu, and Pb-contaminated soil, Sep. Purif. Technol. 79 (2011) 170–176. [96] A. Genc, G. Chase, A. Foos, Electrokinetic removal of manganese from river sediment, Water Air Soil Pollut. 197 (2009) 131–141. [97] A.T. Yeung, C. Hsu, R.M. Menon, EDTA enhanced electrokinetic extraction of lead, J. Geotech. Eng., ASCE 122 (1996) 666–673. [98] A.T. Yeung, C. Hsu, Electrokinetic remediation of cadmium-contaminated clay, J. Environ. Eng., ASCE 131 (2005) 298–304. [99] R.W. Leinz, D.B. Hoover, D.L. Fey, D.B. Smith, T. Patterson, Electrogeochemical sampling with NEOCHIM – results of tests over buried gold deposits, J. Geochem. Explor. 61 (1998) 57–86. [100] R.W. Leinz, D.B. Hoover, A.L. Meier, NEOCHIM: an electrochemical method for environmental application, J. Geochem. Explor. 64 (1998) 421–434. [101] Z.M. Shen, X.J. Chen, J.P. Jia, L. Qu, W.H. Wang, Comparison of electrokinetic soil remediation methods using one ﬁxed anode and approaching anodes, Environ. Pollut. 150 (2007) 193–199. [102] H.J. Lee, J.M. Choi, J.W. Cho, S.H. Moon, Characterization of anion exchange membranes fouled with humate during electrodialysis, J. Membr. Sci. 203 (2002) 115–126. [103] Z. Li, J.-W. Yu, I. Neretnieks, Electroremediation: removal of heavy metals from soils by using cation selective membrane, Environ. Sci. Technol. 32 (1998) 394–397. [104] W.S. Kim, S.O. Kim, K.W. Kim, Enhanced electrokinetic extraction of heavy metals from soils assisted by ion exchange membranes, J. Hazard. Mater. 118 (2005) 93–102. [105] F.A. Caliman, B.M. Robu, C. Smaranda, V.L. Pavel, M. Gavrilescu, Soil and groundwater cleanup: beneﬁts and limits of emerging technologies, Clean Technol. Environ. Policy 13 (2011) 241–268. [106] N. Koprivanac, H. Kusic, AOP as an effective tool for the minimization of hazardous organic pollutants in colored wastewater; chemical and photochemical processes, in: A.A. Lewinsky (Ed.), Hazardous Materials and Wastewater: Treatment, Removal and Analysis, Nova Science Publishers, New York, 2007, pp. 149–200. [107] E. Neyens, J. Baeyens, A review of classic Fenton’s peroxidation as an advanced oxidation technique, J. Hazard. Mater. 98 (2003) 33–50. [108] R.J. Watts, S. Kong, M. Dippre, W.T. Barnes, Oxidation of sorbed hexachlorobenzene in soils using catalysed hydrogen peroxide, J. Hazard. Mater. 39 (1994) 33–47. [109] R.J. Watts, P.C. Stanton, Mineralization of sorbed and NAPL-phase hexadecane by catalyzed hydrogen peroxide, Water Res. 33 (1999) 1405–1414. [110] F.J. Rivas, Polycyclic aromatic hydrocarbons sorbed on soils: a short review of chemical oxidation based treatments, J. Hazard. Mater. 138 (2006) 234–251. [111] E. Ferrarese, G. Andreottola, I.A. Oprea, Remediation of PAH-contaminated sediments by chemical oxidation, J. Hazard. Mater. 152 (2008) 128–139. [112] G.C.C. Yang, Y.W. Long, Removal and degradation of phenol in a saturated ﬂow by in-situ electrokinetic remediation and Fenton-like process, J. Hazard. Mater. 69 (1999) 259–271.

28

A.T. Yeung, Y.-Y. Gu / Journal of Hazardous Materials 195 (2011) 11–29

[113] G.C.C. Yang, C.Y. Liu, Remediation of TCE contaminated soils by in situ EKFenton process, J. Hazard. Mater. 85 (2001) 317–331. [114] S.S. Kim, J.H. Kim, S.J. Han, Application of the electrokinetic-Fenton process for the remediation of kaolinite contaminated with phenanthrene, J. Hazard. Mater. 118 (2005) 121–131. [115] J.H. Kim, S.J. Han, S.S. Kim, J.W. Yang, Effect of soil chemical properties on the remediation of phenanthrene-contaminated soil by electrokinetic-Fenton process, Chemosphere 63 (2006) 1667–1676. [116] J.H. Kim, S.S. Kim, J.W. Yang, Role of stabilizers for treatment of clayey soil contaminated with phenanthrene through electrokinetic-Fenton process – some experimental evidences, Electrochim. Acta 53 (2007) 1663–1670. [117] J.H. Kim, J.Y. Kim, S.S. Kim, Effect of H2 SO4 and HCl in the anode purging solution for the electrokinetic-Fenton remediation of soil contaminated with phenanthrene, J. Environ. Sci. Health A – Tox. Hazard. Subst. Environ. Eng. 44 (2009) 1111–1119. [118] T. Alcantara, M. Pazos, S. Gouveia, C. Cameselle, M.A. Sanroman, Remediation of phenanthrene from contaminated kaolinite by electroremediation-Fenton technology, J. Environ. Sci. Health A – Tox. Hazard. Subst. Environ. Eng. 43 (2008) 901–906. [119] K.R. Reddy, M.R. Karri, Effect of oxidant dosage on integrated electrochemical remediation of contaminant mixtures in soils, J. Environ. Sci. Health A – Tox. Hazard. Subst. Environ. Eng. 43 (2008) 881–893. [120] A. Oonnittan, R.A. Shrestha, M. Sillanpaa, Effect of cyclodextrin on the remediation of hexachlorobenzene in soil by electrokinetic Fenton process, Sep. Purif. Technol. 64 (2009) 314–320. [121] T.T. Tsai, J. Sah, C.M. Kao, Application of iron electrode corrosion enhanced electrokinetic-Fenton oxidation to remediate diesel contaminated soils: a laboratory feasibility study, J. Hydrol. 380 (2010) 4–13. [122] A. Oonnittan, P. Isosaari, M. Sillanpaa, Oxidant availability in soil and its effect on HCB removal during electrokinetic Fenton process, Sep. Purif. Technol. 76 (2010) 146–150. [123] P. Isosaari, R. Piskonen, P. Ojala, S. Voipio, K. Eilola, E. Lehmus, M. Itavaara, Integration of electrokinetics and chemical oxidation for the remediation of creosote-contaminated clay, J. Hazard. Mater. 144 (2007) 538–548. [124] G.C.C. Yang, C.F. Yeh, Enhanced nano-Fe3 O4 /S2 O8 2− oxidation of trichloroethylene in a clayey soil by electrokinetics, Sep. Purif. Technol. 79 (2011) 264–271. [125] K.R. Reddy, S. Chinthamreddy, Electrokinetic remediation of heavy metalcontaminated soils under reducing environments, Waste Manage. 19 (1999) 269–282. [126] A. Weeks, S. Pamukcu, Electrokinetically enhanced reduction of [Cr(Vl)] in contaminated soils, in: Proc. 15th Int. Conf. Soil Mech. Geotech. Eng., vols. 1–3, Istanbul, 2001, pp. 1935–1938. [127] M.F. DeFlaun, C.W. Condee, Electrokinetic transport of bacteria, J. Hazard. Mater. 55 (1997) 263–277. [128] H.S. Lee, K. Lee, Bioremediation of diesel-contaminated soil by bacterial cells transported by electrokinetics, J. Microbiol. Biotechnol. 11 (2001) 1038– 1045. [129] L.Y. Wick, P.A. Mattle, P. Wattiau, H. Harms, Electrokinetic transport of PAHdegrading bacteria in model aquifers and soil, Environ. Sci. Technol. 38 (2004) 4596–4602. [130] A. Olszanowski, K. Piechowiak, The use of an electric ﬁeld to enhance bacterial movement and hydrocarbon biodegradation in soils, Polish J. Environ. Stud. 15 (2006) 303–309. [131] L. Shi, S. Muller, H. Harms, L.Y. Wick, Factors inﬂuencing the electrokinetic dispersion of PAH-degrading bacteria in a laboratory model aquifer, Appl. Microbiol. Biotechnol. 80 (2008) 507–515. [132] J.H. Choi, S. Maruthamuthu, H.G. Lee, T.H. Ha, J.H. Bae, Nitrate removal by electro-bioremediation technology in Korean soil, J. Hazard. Mater. 168 (2009) 1208–1216. [133] U.N. Da Rocha, M.R. Totola, D.M.M. Pessoa, J.T. Araruna, J.C.L. Neves, A.C. Borges, Mobilisation of bacteria in a ﬁne-grained residual soil by electrophoresis, J. Hazard. Mater. 161 (2009) 485–491. [134] K.Y. Lee, H.A. Kim, B.T. Lee, S.O. Kim, Y.H. Kwon, K.W. Kim, A feasibility study on bioelectrokinetics for the removal of heavy metals from tailing soil, Environ. Geochem. Health 33 (2011) 3–11. [135] K.Y. Lee, K.W. Kim, Heavy metal removal from shooting range soil by hybrid electrokinetics with bacteria and enhancing agents, Environ. Sci. Technol. 44 (2010) 9482–9487. [136] M. Elektorowicz, V. Boeva, Electrokinetic supply of nutrients in soil bioremediation, Environ. Technol. 17 (1996) 1339–1349. [137] M.F. Rabbi, B. Clark, R.J. Gale, E. Ozsu-Acar, J. Pardue, A. Jackson, In situ TCE bioremediation study using electrokinetic cometabolite injection, Waste Manage. 20 (2000) 279–286. [138] K.R. Reddy, S. Chinthamreddy, R.E. Saichek, T.J. Cutright, Nutrient amendment for the bioremediation of a chromium-contaminated soil by electrokinetics, Energy Sources 25 (2003) 931–943. [139] G. Lee, H. Ro, S. Lee, S. Lee, Electrokinetically enhanced transport of organic and inorganic phosphorus in a low permeability soil, Geosci. J. 10 (2006) 85–89. [140] S. Suni, E. Malinen, J. Kosonen, H. Silvennoinen, M. Romantschuk, Electrokinetically enhanced bioremediation of creosote-contaminated soil: laboratory and ﬁeld studies, J. Environ. Sci. Health A – Tox. Hazard. Subst. Environ. Eng. 42 (2007) 277–287. [141] A. Tiehm, T. Augenstein, D. Ilieva, H. Schell, C. Weidlich, K.M. Mangold, Bioelectro-remediation: electrokinetic transport of nitrate in a ﬂow-through

[142]

[143]

[144]

[145]

[146]

[147]

[148]

[149]

[150]

[151]

[152]

[153] [154]

[155]

[156]

[157]

[158]

[159]

[160]

[161]

[162]

[163] [164]

[165]

[166]

[167]

[168]

system for enhanced toluene biodegradation, J. Appl. Electrochem. 40 (2010) 1263–1268. C.A.B. Schmidt, M.C. Barbosa, M.D.S. de Almeida, A laboratory feasibility study on electrokinetic injection of nutrients on an organic, tropical, clayey soil, J. Hazard. Mater. 143 (2007) 655–661. S.T. Lohner, D. Katzoreck, A. Tiehm, Electromigration of microbial electron acceptors and nutrients: (I) Transport in synthetic media, J. Environ. Sci. Health A – Tox. Hazard. Subst. Environ. Eng. 43 (2008) 913–921. S.T. Lohner, D. Katzoreck, A. Tiehm, Electromigration of microbial electron acceptors and nutrients: (II) Transport in groundwater, J. Environ. Sci. Health A – Tox. Hazard. Subst. Environ. Eng. 43 (2008) 922–925. W. Xu, C.P. Wang, H.B. Liu, Z.Y. Zhang, H.W. Sun, A laboratory feasibility study on a new electrokinetic nutrient injection pattern and bioremediation of phenanthrene in a clayey soil, J. Hazard. Mater. 184 (2010) 798–804. W.A. Jackson, M.A. Jeon, J.H. Pardue, T. Anderson, Enhanced natural attenuation of perchlorate in soils using electrokinetic injection, in: Bioremediation of Inorganic Compounds, Proc. 6th International In Situ On-Site Bioremediation Symposium, vol. 9, San Diego, 2001, pp. 273–280. X.Y. Fan, H. Wang, Q.S. Luo, J.W. Ma, X.H. Zhang, The use of 2D non-uniform electric ﬁeld to enhance in situ bioremediation of 2,4-dichlorophenolcontaminated soil, J. Hazard. Mater. 148 (2007) 29–37. X. Wu, A.N. Alshawabkeh, D.B. Gent, S.L. Larson, J.L. Davis, Lactate transport in soil by DC ﬁelds, J. Geotech. Geoenviron. Eng., ASCE 133 (2007) 1587– 1596. A. Tiehm, S.T. Lohner, T. Augenstein, Effects of direct electric current and electrode reactions on vinyl chloride degrading microorganisms, Electrochim. Acta 54 (2009) 3453–3459. T.T. Li, S.H. Guo, B. Wu, F.M. Li, Z.X. Niu, Effect of electric intensity on the microbial degradation of petroleum pollutants in soil, J. Environ. Sci. – China 22 (2010) 1381–1386. L.Y. Wick, F. Buchholz, I. Fetzer, S. Kleinsteuber, C. Hartig, L. Shi, A. Miltner, H. Harms, G.N. Pucci, Responses of soil microbial communities to weak electric ﬁelds, Sci. Total Environ. 408 (2010) 4886–4893. S.H. Kim, H.Y. Han, Y.J. Lee, C.W. Kim, J.W. Yang, Effect of electrokinetic remediation on indigenous microbial activity and community within diesel contaminated soil, Sci. Total Environ. 408 (2010) 3162–3168. P.L. Palmer, Permeable treatment barriers, in: In situ Treatment Technology, 2nd ed., Lewis Publishers, Boca Raton, 2001, pp. 459–482. S.V. Ho, C.J. Athmer, P.W. Sheridan, A.P. Shapiro, Scale-up aspects of the LasagnaTM process for in situ soil decontamination, J. Hazard. Mater. 55 (1997) 39–60. S.V. Ho, P.W. Sheridan, C.J. Athmer, M.A. Heitkamp, J.M. Brackin, D. Weber, P.H. Brodsky, Integrated in situ soil remediation technology: the Lasagna process, Environ. Sci. Technol. 29 (1995) 2528–2534. S.V. Ho, C. Athmer, P.W. Sheridan, B.M. Hughes, R. Orth, D. Mckenzie, P.H. Brodsky, A. Shapiro, R. Thornton, J. Salvo, D. Schultz, R. Landis, R. Grifﬁth, S. Shoemaker, The Lasagna technology for in situ soil remediation. 1. Small ﬁeld test, Environ. Sci. Technol. 33 (1999) 1086–1091. S.V. Ho, C. Athmer, P.W. Sheridan, B.M. Hughes, R. Orth, D. Mckenzie, P.H. Brodsky, A.M. Shapiro, T.M. Sivavec, J. Salvo, D. Schultz, R. Landis, R. Grifﬁth, S. Shoemaker, The Lasagna technology for in situ soil remediation. 2. Large ﬁeld test, Environ. Sci. Technol. 33 (1999) 1092–1099. M. Roulier, M. Kemper, S. Al-Abed, L. Murdoch, P. Cluxton, J.-L. Chen, W. Davishoover, Feasibility of electrokinetic soil remediation in horizontal LasagnaTM cells, J. Hazard. Mater. 77 (2000) 161–176. S.A. Jackman, G. Maini, A.K. Sharman, G. Sunderland, C.J. Knowles, Electrokinetic movement and biodegradation of 2,4-dichlorephenoxyacetic acid in silt soil, Biotechnol. Bioeng. 74 (2001) 40–48. J.W. Ma, F.Y. Wang, Z.H. Huang, H. Wang, Simultaneous removal of 2,4-dichlorophenol and Cd from soils by electrokinetic remediation combined with activated bamboo charcoal, J. Hazard. Mater. 176 (2010) 715–720. C.F. Chew, T.C. Zhang, In-situ remediation of nitrate contaminated groundwater by electrokinetics iron wall processes, Water Sci. Technol. 38 (1998) 135–142. J.W. Moon, H.S. Moon, H. Kim, Y. Roh, Remediation of TCE-contaminated groundwater using zero valent iron and direct current: experimental results and electron competition model, Environ. Geol. 48 (2005) 805–817. C. Yuan, The effect of Fe(0) on electrokinetic remediation of clay contaminated with perchloroethylene, Water Sci. Technol. 53 (2006) 91–98. C.H. Weng, Y.T. Lin, T.Y. Lin, C.M. Kao, Enhancement of electrokinetic remediation of hyper-Cr(VI) contaminated clay by zero-valent iron, J. Hazard. Mater. 149 (2007) 292–302. C. Yuan, T.S. Chiang, The mechanisms of arsenic removal from soil by electrokinetic process coupled with iron permeable reaction barrier, Chemosphere 67 (2007) 1533–1542. L. Cang, D.M. Zhou, D.Y. Wu, A.N. Alshawabkeh, Coupling electrokinetics with permeable reactive barriers of zero-valent iron for treating a chromium contaminated soil, Sep. Sci. Technol. 44 (2009) 2188–2202. J.Z. Wan, Z.R. Li, X.H. Lu, S.H. Yuan, Remediation of a hexachlorobenzenecontaminated soil by surfactant-enhanced electrokinetics coupled with microscale Pd/Fe PRB, J. Hazard. Mater. 184 (2010) 184–190. Y.H. Shih, Y.C. Chen, M.Y. Chen, Y.T. Tai, C.P. Tso, Dechlorination of hexachlorobenzene by using nanoscale Fe and nanoscale Pd/Fe bimetallic particles, Colloids Surf. A 332 (2009) 84–89.

A.T. Yeung, Y.-Y. Gu / Journal of Hazardous Materials 195 (2011) 11–29 [169] D.W.S. Faulkner, L. Hopkinson, A.B. Cundy, Electrokinetic generation of reactive iron-rich barriers in wet sediments: implications for contaminated land management, Mineral. Mag. 69 (2005) 749–757. [170] H.I. Chung, M. Lee, A new method for remedial treatment of contaminated clayey soils by electrokinetics coupled with permeable reactive barriers, Electrochim. Acta 52 (2007) 3427–3431. [171] T. Kimura, K.I. Takase, N. Terui, S. Tanaka, Ferritization treatment of copper in soil by electrokinetic remediation, J. Hazard. Mater. 143 (2007) 662–667. [172] S.S. Mandaokar, D.M. Dharmadhikari, S.S. Dara, Retrieval of heavy-metal ions from solution via ferritisation, Environ. Pollut. 83 (1994) 277–282. [173] E. Barrado, F. Prieto, M. Vega, F. Fernandez-Polanco, Optimization of the operational variables of a medium-scale reactor for metal-containing wastewater puriﬁcation by ferrite formation, Water Res. 32 (1998) 3055–3061. [174] G. De Gioannis, A. Muntoni, R. Ruggeri, J.J.P. Zijlstra, Chromate adsorption in a transformed red mud permeable reactive barrier using electrokinesis, J. Environ. Sci. Health A – Tox. Hazard. Subst. Environ. Eng. 43 (2008) 969–974. [175] C. Yuan, C.H. Hung, K.C. Chen, Electrokinetic remediation of arsenate spiked soil assisted by CNT-Co barrier – the effect of barrier position and processing ﬂuid, J. Hazard. Mater. 171 (2009) 563–570. [176] X.J. Peng, Y.H. Li, Z.K. Luan, Z.C. Di, H.Y. Wang, B.H. Tian, Z.P. Jia, Adsorption of 1,2-dichlorobenzene from water to carbon nanotubes, Chem. Phys. Lett. 376 (2003) 154–158. [177] Y. Yu, J.C. Yu, C.Y. Chan, Y.K. Che, J.C. Zhao, L. Ding, W.K. Ge, P.K. Wong, Enhancement of adsorption and photocatalytic activity of TiO2 by using carbon nanotubes for the treatment of azo dye, Appl. Catal. B – Environ. 61 (2005) 1–11. [178] K. Yang, X.L. Wang, L.Z. Zhu, B.S. Xing, Competitive sorption of pyrene, phenanthrene, and naphthalene on multiwalled carbon nanotubes, Environ. Sci. Technol. 40 (2006) 5804–5810. [179] K. Yang, L.Z. Zhu, B.S. Xing, Adsorption of polycyclic aromatic hydrocarbons by carbon nanomaterials, Environ. Sci. Technol. 40 (2006) 1855–1861. [180] J.G. Han, K.K. Hong, Y.W. Kim, J.Y. Lee, Enhanced electrokinetic (E/K) remediation on copper contaminated soil by CFW (carbonized foods waste), J. Hazard. Mater. 177 (2010) 530–538. [181] C.S. O’Connor, N.W. Lepp, R. Edwards, G. Sunderland, The combined use of electrokinetic remediation and phytoremediation to decontaminate metalpolluted soils: a laboratory-scale feasibility study, Environ. Monit. Assess. 84 (2003) 141–158.

29

[182] J.M. Lim, A.L. Salido, D.J. Butcher, Phytoremediation of lead using Indian mustard (Brassica juncea) with EDTA and electrodics, Microchem. J. 76 (2004) 3–9. [183] D.M. Zhou, H.F. Chen, L. Cang, Y. Wang, Ryegrass uptake of soil Cu/Zn induced by EDTA/EDDS together with a vertical direct-current electrical ﬁeld, Chemosphere 67 (2007) 1671–1676. [184] H. Aboughalma, R. Bi, M. Schlaak, Electrokinetic enhancement on phytoremediation in Zn, Pb, Cu and Cd contaminated soil using potato plants, J. Environ. Sci. Health A – Tox. Hazard. Subst. Environ. Eng. 43 (2008) 926–933. [185] R. Bi, M. Schlaak, E. Siefert, R. Lord, H. Connolly, Inﬂuence of electrical ﬁelds (AC and DC) on phytoremediation of metal polluted soils with rapeseed (Brassica napus) and tobacco (Nicotiana tabacum), Chemosphere 83 (2011) 318–326. [186] L. Cang, Q.Y. Wang, D.M. Zhou, H. Xu, Effects of electrokinetic-assisted phytoremediation of a multiple-metal contaminated soil on soil metal bioavailability and uptake by Indian mustard, Sep. Purif. Technol. 79 (2011) 246–253. [187] H.I. Chung, M. Kamon, Ultrasonically enhanced electrokinetic remediation for removal of Pb and phenanthrene in contaminated soils, Eng. Geol. 77 (2005) 233–242. [188] H.I. Chung, Treatment of contaminated groundwater in sandy layer under river bank by electrokinetic and ultrasonic technology, Water Sci. Technol. 55 (2007) 329–338. [189] T.D. Pham, R.A. Shrestha, J. Virkutyte, M. Sillanpaa, Combined ultrasonication and electrokinetic remediation for persistent organic removal from contaminated kaolin, Electrochim. Acta 54 (2009) 1403–1407. [190] R.A. Shrestha, T.D. Pham, M. Sillanpaa, Electro ultrasonic remediation of polycyclic aromatic hydrocarbons from contaminated soil, J. Appl. Electrochem. 40 (2010) 1407–1413. [191] W. Wesner, A. Diamant, B. Schrammel, M. Unterberger, Electrosynthesis of oxidants and their electrokinetic distribution, in: K.R. Reddy, C. Cameselle (Eds.), Electrochemical Remediation Technologies for Polluted Soils, Sediments and Groundwater, John Wiley & Sons, Hoboken, 2009, pp. 473–482. [192] J.A. Soesilo, S.R. Wilson, Site Remediation Planning and Management, Lewis Publishers, Boca Raton, 1997. [193] G.J. Smith, Coupled electrokinetic-thermal desorption, in: K.R. Reddy, C. Cameselle (Eds.), Electrochemical Remediation Technologies for Polluted Soils, Sediments and Groundwater, John Wiley & Sons, Hoboken, 2009, pp. 505–535.

Journal of Hazardous Materials 195 (2011) 30–54

Contents lists available at SciVerse ScienceDirect

Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat

Review

Non-thermal plasmas for non-catalytic and catalytic VOC abatement Arne M. Vandenbroucke, Rino Morent ∗ , Nathalie De Geyter, Christophe Leys Research Unit Plasma Technology, Department of Applied Physics, Faculty of Engineering, Ghent University, Jozef Plateaustraat 22, 9000 Ghent, Belgium

a r t i c l e

i n f o

Article history: Received 27 February 2011 Received in revised form 19 August 2011 Accepted 22 August 2011 Available online 27 August 2011 Keywords: Non-thermal plasma Plasma–catalysis Volatile organic compounds Waste gas treatment

a b s t r a c t This paper reviews recent achievements and the current status of non-thermal plasma (NTP) technology for the abatement of volatile organic compounds (VOCs). Many reactor conﬁgurations have been developed to generate a NTP at atmospheric pressure. Therefore in this review article, the principles of generating NTPs are outlined. Further on, this paper is divided in two equally important parts: plasmaalone and plasma–catalytic systems. Combination of NTP with heterogeneous catalysis has attracted increased attention in order to overcome the weaknesses of plasma-alone systems. An overview is given of the present understanding of the mechanisms involved in plasma–catalytic processes. In both parts (plasma-alone systems and plasma–catalysis), literature on the abatement of VOCs is reviewed in close detail. Special attention is given to the inﬂuence of critical process parameters on the removal process. © 2011 Elsevier B.V. All rights reserved.

Contents 1. 2.

3.

4.

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Non-thermal plasmas for VOC abatement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1. Without catalyst . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1.1. What is a non-thermal plasma? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1.2. Reactor concepts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1.3. VOC abatement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2. Combined with catalyst . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2.1. What is plasma–catalysis? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2.2. Different types of catalysts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2.3. VOC abatement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Critical process parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1. Temperature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2. Initial VOC concentration. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3. Humidity level . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.4. Oxygen content . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.5. Gas ﬂow rate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Future trends . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1. Introduction Environmental protection is becoming an issue of growing concern in our globalized world. The industrialization of many economies has led to the emission of various kinds of substances that danger both human and ecological life [1]. Since World War II,

∗ Corresponding author. E-mail address: [email protected] (R. Morent). 0304-3894/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.jhazmat.2011.08.060

30 31 31 31 31 34 38 38 40 40 47 47 47 47 48 48 48 49

governments have become aware that emission legislation needs to become increasingly severe in order to ensure the protection of our environment for future generations. International treaties like the Kyoto protocol (1997) and the protocol of Gothenburg (1999) are important examples. Exhausts from mobile (e.g. cars) and stationary sources (e.g. plants) pollute the air with a variety of harmful substances that threat human and ecological life [2]. Next to NOx , SOx , H2 S,. . ., volatile organic compounds (VOCs) are a large and important group of pollutants. Their high volatility causes them to rapidly

A.M. Vandenbroucke et al. / Journal of Hazardous Materials 195 (2011) 30–54 Table 1 Typical VOCs and their health effects. VOC

Formula

Effects

O Acetone

H3C

C

CH3

Carcinogen

O Formaldehyde

H Dichloroethane

C

Sore throat, dizziness, headache

H

H

H

H

H

Cl

Cl

H

Cl

Cl

Cl

Cl

Cl

Cl

Cl

Trichloroethylene

Tetrachloroethylene

Paralysis of nerve center

Liver and kidney disease, paralysis of nerve center

Probable heart and liver disease, skin irritation

Benzene

Carcinogen

Toluene

Headache, dizziness

Xylene

Headache, dizziness

Styrene

Probable carcinogen

31

non-selective, creating a chemical reactive environment in which harmful substances are readily decomposed. Although NTP for end-of-pipe application has been frequently proposed in literature for the removal of VOCs, NOx and SO2 ,. . . [14–16] formation of unwanted byproducts, poor energy efﬁciency and mineralization are serious roadblocks towards industrial implementation. To overcome these problems, researchers are combining the advantages of NTP and catalysis in a technique called plasma–catalysis. This innovative method has become a hot topic over approximately the last ten years. The primary idea is that, by placing catalysts inside or in close vicinity of the discharge zone, retention time can be increased through adsorption of target molecules, favoring complete oxidation to CO2 and H2 O [17]. Interestingly, combining both techniques creates a synergism that is caused by various mechanisms [18–20]. This review gives an overview of the literature on plasmaassisted decomposition of VOCs with a special focus on plasma–catalysis. In the ﬁrst part, an overview of different NTP reactors is given. In the second part, based on a large number of papers, an extended review is presented dealing with the treatment of VOCs with plasma-alone as well as with plasma–catalytic systems. Particular attention is paid to the most studied target compounds, i.e. trichloroethylene, benzene and toluene. Also general mechanisms that govern plasma–catalysis are summarized. In the third section, special attention is given to the inﬂuence of critical process parameters on the removal process. In the ﬁnal section, future trends for this promising hybrid technique are discussed.

2. Non-thermal plasmas for VOC abatement 2.1. Without catalyst

evaporate and enter the earth’s atmosphere. Depending on their chemical structure and concentration, they can cause various effects such as the creation of photochemical smog, secondary aerosols and tropospheric ozone [3]. They also have an effect on the intensiﬁcation of global warming and on the deterioration of the stratospheric ozone layer. Some of them are toxic and cause odour nuisance while others have carcinogenic effects, proving their adverse effects on human health [4]. Table 1 provides an overview of typical VOCs that have been studied for removal with non-thermal plasma (NTP) and plasma–catalysis along with their related health effects. Conventional methods to control VOC emissions are wellestablished technologies such as adsorption [5], thermal and catalytic oxidation [6], membrane separation [7], bioreaction [8] and photocatalysis [9]. The disadvantage of these methods is that they become cost-inefﬁcient and difﬁcult to operate when low concentrations of VOC need to be treated [10]. With the increased severity of emission limits in mind, this creates the need for an alternative technology that overcomes these weaknesses. For the abatement of VOCs, NTP technology has attracted growing interest of scientists over the last 2 decades [11,12]. The energy that is delivered to the system, is almost completely consumed for accelerating electrons. They gain a typical temperature of 10,000–250,000 K (1–20 eV) [13], while the background gas remains at room temperature. This non-equilibrium state makes it unnecessary to heat the entire treated gas ﬂow. Accelerated primary electrons collide with background molecules (N2 , O2 , H2 O,. . .) producing secondary electrons, photons, ions and radicals. These latter species are responsible for the oxidation of VOC molecules, although ionic reactions are also possible. This process is highly

2.1.1. What is a non-thermal plasma? Non-thermal plasmas are generated by applying a sufﬁciently strong electric ﬁeld to ensure the discharge of a neutral gas. This creates a quasi neutral environment containing neutrals, ions, radicals, electrons and UV photons. Due to their light mass, electrons are selectively accelerated by the ﬁeld and gain high temperatures while the heavier ions remain relatively cold through energy exchange by collisions with the background gas. The bulk gas molecules (e.g. N2 , O2 ) are bombarded by the electrons, typically having temperatures ranging from 10,000 K to 250,000 K (1–20 eV). This produces excited gas molecules (N2 *, O2 *) which lose their excess energy by emitting photons or heat. Next to excitation, other processes like ionization, dissociation and electron attachment occur in the discharge zone. Through these reaction channels, unstable reactive species like ions and free radicals are formed. Free radicals, such as OH• and O• , are highly reactive species which are ideal for the conversion of environmental pollutants to CO2 , H2 O and other degradation products at uncharacteristic low temperatures. The generation of NTP at atmospheric pressure and ambient temperature has been the subject of many research papers during the last two decades. This has lead to great advances, mainly on laboratory scale. However, large-scale demonstrations of NTP technology for waste gas cleaning are also currently operative [13,21].

2.1.2. Reactor concepts Researchers have investigated a variety of NTP reactors for environmental purposes. The classiﬁcation of these different reactors is rather complex and depends on multiple characteristics, such as:

32

A.M. Vandenbroucke et al. / Journal of Hazardous Materials 195 (2011) 30–54

Fig. 1. Illustrations of various NTP reactor conﬁgurations.

- type of discharge: (DC or pulsed) corona discharge, surface discharge, dielectric barrier discharge, ferro-electric packed bed discharge,. . . - type of power supply: AC, DC, pulse, microwave, RF,. . . - other characteristics: electrode conﬁguration, voltage level, polarity, gas composition,. . . For the conventional NTP reactors that are employed in laboratory experiments, only the main characteristics will be brieﬂy discussed here. A more detailed discussion can be found in literatures [22–27]. A dielectric barrier discharge (DBD) or silent discharge, typically has at least one dielectric (e.g. glass, quartz or ceramic) between the electrodes. DBDs are generally operated in one of the planar or cylindrical conﬁgurations shown in Fig. 1. When the local electron density at certain locations in the discharge gap reaches a critical value, a large number of separate and short-lived current ﬁlaments are formed, also referred to as microdischarges. These bright, thin ﬁlaments are statistically distributed in space and time and are formed by channel streamers with nanosecond duration [28]. When a microdischarge reaches the dielectric, it spreads into a surface discharge and the accumulation of the transferred charge on the surface of the dielectric barrier reduces the electric ﬁeld. As the electric ﬁeld further reduces, electron attachment prevails over ionization and the microdischarges are extinguished. When the polarity of the AC voltage changes, the formation of a microdischarge is repeated at the same location if the

electron density again reaches a critical value necessary for electrical breakdown. Therefore, the use of the dielectric in the discharge zone has two functions: (1) limiting the charge transferred by an individual microdischarge, thereby preventing the transition to an arc discharge, and (2) spreading the microdischarge over the electrode surface which increases the probability of electron–molecule collisions with bulk gas molecules [28]. This type of arrangement is often referred to as a volume discharge [29]. Another type of arrangement to generate NTP in a DBD is the surface discharge [29] (Fig. 1(c)). Here for example, a series of strip electrodes are attached to the surface of a high-purity alumina ceramic base. A ﬁlm like counter electrode is embedded in the inside of the alumina ceramic base and functions as an induction electrode. The ceramic can be either planar or cylindrical [30,31]. When an AC voltage is applied between the strip electrodes and the embedded counter electrode, a surface discharge starts from the peripheral edges of each discharge electrode and stretches out along the ceramic surface. The surface discharges actually consist of many nanosecond surface streamers. In another conﬁguration to generate a surface discharge, strip electrodes can be placed on the inner surface of a cylindrical surface discharge reactor [32]. In this set-up, a DBD discharge is also formed between the central rod electrode and the surface electrodes. A pulsed corona discharge (Fig. 1(d)) applies a pulsed power supply with a fast voltage rise time (tens of nanoseconds) to enable an increase in corona voltage and power without formation of sparks, which can damage the reactor and decrease the process efﬁciency.

A.M. Vandenbroucke et al. / Journal of Hazardous Materials 195 (2011) 30–54

The required voltage level to energize the discharge depends on the distance between the electrodes, the pulse duration and the gas composition [33]. The duration of a pulse voltage is typically in the order of 100–200 ns to ensure that spark formation is prevented and that the energy dissipation by ions is minimal. The latter is important to enhance the energy efﬁciency of the system. The electrode conﬁguration of a pulsed corona discharge reactor can be either wire-to-cylinder [34–36] or wire-to-plate [37,38], although the former allows a better spatial distribution of the streamers and a higher energy density deposition in the gas [34]. The pulsed corona discharge usually consists of streamers, for which the ionization zone ﬁlls the entire electrode gap (e.g. 10 cm). This is favorable in terms of up-scaling and reducing the pressure drop. However, upscaling is hampered by the high demands on the electronics of large pulsed power voltage sources. A ferroelectric pellet packed-bed reactor (Fig. 1(e)) is a packed-bed reactor ﬁlled with perovskite oxide pellets. These reactors can have a parallel-plate or a coaxial conﬁguration. Barium titanate (BaTiO3 ) is the most widely used ferroelectric material for environmental purposes, owing to its high dielectric constant (2000 < ε < 10,000). Other used ferroelectric materials are NaNO2 [39], MgTiO4 , CaTiO3 , SrTiO3 , PbTiO3 [40] and PbZrO3 –PbTiO3 [41]. Application of an external electric ﬁeld leads to polarization of the ferroelectric material and induces strong local electric ﬁelds at the contact points between the pellets and between the pellets and electrodes. This enables the production of partial discharges in the vicinity of each contact point between pellets. The presence of ferroelectric pellets in the discharge zone is beneﬁcial for a uniform gas distribution and electrical discharge but causes an increase in pressure drop over the reactor length. Ferroelectric packed-bed reactors could serve as an alternative approach to enhance the energy efﬁciency, because the increase of the electric ﬁeld will lead to a higher mean electron energy. Hence, the energetic electrons tend to form active species through dissociation and ionization, rather than forming less useful species through rotational and vibrational excitation. This leads to a more favorable consumption of the energy delivered, because electron-impact reactions are mainly

33

responsible for the plasma chemistry that destroys environmental pollutants. A DC corona discharge is generated at atmospheric pressure when sharp points, edges or thin wires are subjected to a sufﬁciently large electric ﬁeld. This causes a local increase of the electric ﬁeld in the vicinity of the sharp curvature of the electrode. This is e.g. the case for a point-to-plate or for a wire-to-cylinder conﬁguration. The corona discharge is initiated by acceleration of free electrons and subsequent electron collision processes. Due to formation of electron/positive-ion pairs and their separating process, an electron avalanche is created which sustains the corona discharge. Visually, this discharge is characterized by a weak glow region around the sharp electrode. Depending on the polarity of this electrode, the formation mechanism of the electron avalanche physically differs [22]. When the electrode with the strongest curvature is connected to the positive output of the power supply, a positive DC corona discharge is generated. Propagation of the discharge mainly depends on secondary photo-ionization processes around the sharp tip. The positive corona is characterized by the presence of streamers, i.e. numerous thin current ﬁlaments which are chaotically distributed in the gap. At a certain threshold voltage the discharge transitions from the stable corona mode to an unstable spark discharge regime. In the case that the sharp electrode is connected to the negative output, a negative DC corona discharge is formed. Here, impact ionization of gas molecules is generally responsible for the propagation of the discharge. As the applied voltage increases, the negative corona will initially form Trichel pulse corona, followed by pulseless corona and spark discharge [22]. However, certain research groups [42–48] have succeeded in generating a glow discharge at atmospheric pressure before the negative corona shifts to a spark discharge. Akishev et al. [47] applied a special electrode geometry and a fast gas ﬂow to stabilize the discharge, hence delaying it from creating sparks. Vertriest et al. [49] successfully tested the multipin-to-plate reactor concept for VOC abatement. Antao et al. [50] recently reviewed the operating regimes of atmospheric pressure DC corona discharges and their potential applications.

Table 2 Overview of published papers on TCE removal with NTP. Plasma type

Carrier gas

Flow rate (mL/min)

Concentration range (ppm)

Maximum removal efﬁciency (%)

DBD

Ar/O2 Ar/O2 /H2 O

104

500

>99 90

DBDa DBDb

Air Humid air

700 500

250 150–200

DBD

Dry air

400

DBD DBD DBD DBD

Dry air Dry air Dry air Humid air

DBD Surface discharge

Energy density (J/L)

Energy yield (g/kWh)

References

50 150

193.5d 58d

[51]

>99 >99

140 480

34.6d 8.1d

[115] [133]

1000 100

95 >99

150 135

122.5d 14.3d

[138]

400 2000 510 200–510

100 250 430 750

99 98 >99 98–99

200 120 350 2400

9.6d 39.5d 23.8d 6d

[161] [171] [172] [229]

Dry air

400

1000

99 95–99

1400 1150

13.7d 16.3d

[143]

DBD Pulsed corona

Dry air

2 × 104

160

85 90

100 50

26.3d 55.7d

[53]

Pulsed coronac Pulsed corona Positive corona DC negative glow discharge Capillary tube discharge

Humid air Dry air Dry air Humid air Dry air

– – 1500 106 1000

1000 100 100 120 452

90 80 67 47 80

100 50 580 37 –

174.1d 30.9 2.2d 29.5d –

a b c d

Copper rod inner electrode. Inner electrode made of sintered metal ﬁbres. Pulsed corona discharge with reticulated vitreous carbon electrodes. Calculated from data retrieved from reference.

[52] [72] [118] [49] [78]

34

A.M. Vandenbroucke et al. / Journal of Hazardous Materials 195 (2011) 30–54

Table 3 Overview of published papers on benzene removal with NTP. Plasma type

Carrier gas

Flow rate (mL/min)

Concentration range (ppm)

Maximum removal efﬁciency (%)

Energy density (J/L)

Energy yield (g/kWh)

References

DBD DBD DBD DBD DBD DBD DBD DBD DBD DBD DBD Pulsed corona Positive DC coronaa DC glowb BaTiO3 packed-bed BaTiO3 packed-bed BaTiO3 packed-bed BaTiO3 packed-bed BaTiO3 packed-bed BaTiO3 packed-bedc

Air (0–90% RH) Dry air Humid air Dry air (5% O2 ) Ar/2–40% O2 Dry air Dry air Dry air Dry air Dry air Dry air Dry air Dry air Dry air Dry air Humid air (0.5% H2 O) Dry air Dry air Dry air Humid air (0.5% H2 O)

500–2000 200 104 4000–5000 275 250 500 400 4000 1667 35 × 103 100 – 100 200 200 203–210 – 200 200

500–2700 100 276 200 500–104 300–380 105 200 203–210 407 250 300 300 296 200 200 200 110 200 200

>99.9 90 >99 75 30 11 35 70 40 50 50 75 >99 90 >99 75 65 98 60 95

2000–3000 680 810 305 – 170 360 3150 370 – 230 30 – 4000 3000 1800 400 130 600 1600

– 1.5d 3.9d 5.7d – 2.5 1.2d 0.5d 2.6d – 6.3d 86.1d – 0.9d 0.8d 1d 3.7d 9.4d 2.3d 1.4d

[69] [70] [71] [75] [77] [108] [129] [163] [175] [179] [230] [90] [73] [72] [40] [163] [174] [176] [212] [231]

a b c d

Corona reactor is sealed after addition of benzene/air mixture. Microhollow cathode. Glass layer between two concentric electrodes. Calculated from data retrieved from reference.

2.1.3. VOC abatement Tables 2–5 give an overview of published papers on VOC removal with NTP. For each reference, experimental conditions are given, along with the maximum removal efﬁciency and the corresponding energy yield in g/kWh calculated as followed: C × × M × 0.15 Energy yield = in ε where Cin is the initial concentration (ppm) of the VOC with molecular weight M (g/mol), the maximum removal efﬁciency and ε the corresponding energy density (J/L), i.e. the energy deposited per unit volume of process gas. Each calculation is based on the fact that one mole of a gas occupies 24.04 L volume at standard ambient temperature and pressure (293 K and 101325 Pa). In what follows, particular attention is paid to the most studied target compounds, i.e. trichloroethylene, benzene and toluene. In

Table 5, a selection has been made of other relevant, but less frequently studied VOCs. For more details about operating conditions and results, the reader can consult the corresponding references. 2.1.3.1. Trichloroethylene. As can be seen from Table 2, TCE is a chlorinated oleﬁn which has attracted a lot of attention because it can be relatively easy removed by NTP without the addition of considerable energy. This results from the fact that reactive radicals, produced in the plasma discharge, easily add to the carbon-carbon double bond thereby initiating the oxidation process. Evans et al. [51] carried out an experimental and computational study of the plasma remediation of TCE in dry and wet Ar/O2 mixtures using a silent discharge plasma. They found that the ClO radical is an important intermediate which oxidizes TCE. In wet mixtures, ClO is partially consumed by OH radicals, resulting in a lower decomposition rate of TCE. They suggest a diagram

Table 4 Overview of published papers on toluene removal with NTP. Plasma type

Carrier gas

Flow rate (mL/min)

Concentration range (ppm)

Maximum removal efﬁciency (%)

Energy density (J/L)

Energy yield (g/kWh)

References

DBD

Dry air (5% O2 )

4000–5000

200

75

310

6.6b

[75]

DBD

N2 dry air

2000

400

21 23

240

4.7b 5.2b

[81]

DBD DBD DBD packed with glass pellets DBD packed with glass pellets DBD packed with glass beads Multicell DBD packed with glass beadsa DC back corona Pulsed corona Positive corona BaTiO3 packed-bed Dielectric capillary plasma electrode discharge Capillary tube discharge

N2 /5% O2 (0.2% RH) Humid air (55% RH) Dry air Humid air (95% RH) Dry air Dry air

100 1000 600 500 315 1000

50 100 1100 500 240 110

73 46 75–80 91 36 72

600 2100 1000 18.5 172 2502

0.8b 0.3b 11.5b 11.5 6.8b 0.4b

[82] [203] [167] [232] [202] [224]

Dry air Dry air Humid air (26% RH) Dry air Air

100–750 450 104 – –

5–200 500 0.5 101 266.5

93 >99 80 95 >99

2400 1000 65 125 3500

0.4b 6.7b 0.1b – 1b

[79] [84] [83] [176] [233]

Dry air

350

1246

86

–

–

[78]

a b

Three cells. Calculated from data retrieved from reference.

A.M. Vandenbroucke et al. / Journal of Hazardous Materials 195 (2011) 30–54

35

Table 5 Published papers on removal of other VOCs with NTP. Target compound

Reference

Acetaldehyde Acetone Acetylene Dichloromethane Formaldehyde Methane Methanol Propane Propene Styrene Tetrachloromethane Xylene

[215,234] [31,235–240] [241,242] [210,235,243–246] [191,194] [72,218,219,247] [53,61,210,248–250] [34,190,193,251–253] [34,241,252,254,255] [256,257] [61,216,235,245,258–260] [35,192,233,238,261]

of the dominant reaction pathways in plasma remediation of TCE giving CO, CO2 , COCl2 and HCl as main byproducts. According to the authors, the toxic byproduct phosgene (COCl2 ) can easily be removed from the exhaust stream by placing a water scrubber downstream of the plasma discharge. This is a cost effective posttreatment because removal of phosgene with their DBD reactor requires high energy. The authors propose the following reactions as the dominant degradation pathway for TCE in dry Ar/O2 mixtures: C2 HCl3 + O• → CHOCl + CCl2

(1)

ClO•

(2)

CCl2 + O2 →

+ COCl

CHOCl +

O•

→ COCl +

OH•

(3)

CHOCl +

Cl•

→ COCl + HCl

(4)

COCl +

O•

→ CO +

COCl +

Cl•

→ CO + Cl2

COCl + O2 → CO2

ClO•

+ ClO•

(5) (6) (7)

The ClO radical rapidly back-reacts with TCE leading to the formation of phosgene and methylchloride by the following reaction: C2 HCl3 + ClO• → COCl2 + CHCl2

(8)

Methylchloride then quickly reacts with oxygen in the subsequent reaction: CHCl2 + O• → CHOCl + Cl

(9)

In wet mixtures, two additional species can be produced by reaction of OH with TCE, CHCl2 –COCl (dichloroacetylchloride; DCAC) and CHCl2 . DCAC is detected as main byproduct of TCE decomposition with a pulsed corona discharge by Kirkpatrick et al. They suggest the reaction of TCE with ClO radicals leading to the formation of DCAC under dry conditions, as follows [52]: C2 HCl3 + ClO• → CHCl2 COCl + Cl

(10)

C2 HCl3 + OH• → CHCl2 COCl + H

(11)

Under humid conditions the formation of DCAC is suppressed, suggesting that ClO radicals are quenched by OH radicals by the reaction: ClO• + OH• → HCl + O2

(12)

Cl radicals can further attack DCAC, leading to the formation of CO, HCl, CCl4 , CHCl3 and COCl2 as ﬁnal products. The effect of temperature on the removal chemistry and byproduct formation of TCE is studied by Hsiao et al. [53]. Experiments, carried out with a pulsed corona and a DBD reactor, have shown that the removal of TCE and the formation of COx depend on temperature but not on reactor type. Moreover, higher temperatures

Fig. 2. TCE decomposition mechanism. Reprinted from Ref. [54], with permission from Elsevier.

cause a decrease in energy yield for TCE. The formation of byproducts (CO, CO2 , COCl2 , HCl and DCAC) is almost the same as found by Evans et al. [51]. Prager et al. [54] report the degradation of TCE with electron beam treatment. They found CO, HCl, COCl2 , DCAC and CHCl3 as main byproducts next to traces of CCl4 and CCl3 –COCl (trichloroacetylchloride; TCAA). In the proposed degradation mechanism (Fig. 2), OH radicals add to the double bond of TCE forming OH adducts. These adducts decompose and produce chlorine radicals or to a minor extend dichloromethyl radicals. Next, chlorine radicals add to the double bond and in a subsequent reaction with oxygen, the corresponding peroxyl radical is formed. In a bimolecular reaction step, molecular oxygen and alkoxy radicals are formed, which fragmentate to DCAC and chlorine which in turn re-enters the ﬁrst chain reaction. In a second chain reaction, DCAC is further decomposed to HCl, COCl2 and CO. To minimize the formation of chloroacetic acids and phosgene, a wet scrubbing system is installed downstream of the electron beam system. Hakoda et al.

36

A.M. Vandenbroucke et al. / Journal of Hazardous Materials 195 (2011) 30–54

[55,56] also conclude that TCE decomposition with electron beam proceeds via a Cl radical addition chain reaction induced by OH radicals via the following reactions: C2 HCl3 + OH• → C2 HCl3 OH•

(13)

C2 HCl3 OH• + O2 → OOC2 HCl3 OH•

(14)

2OOC2 HCl3 OH∗ → 2OC2 HCl3 OH∗ + O2

(15)

OC2 HCl3 OH• → C2 HCl2 OH(O) + Cl•

(16)

Vitale et al. [57] examined the effect of a carbon-carbon double bond on electron beam treatment of TCE. The primary decomposition products found in their study are CO, CO2 , COCl2 , DCAC and HCl. Chloroform and TCAA were found as minor decomposition products. These researchers propose a reaction pathway in which dissociative electron attachment is believed to be the primary initiation step. This reaction produces chlorine radicals and a doubly chlorinated ethylene anion: C2 HCl3 + e− → CHClCCl− + Cl

(17)

In a study by Yamamoto and Futamura [27], the pronounced decomposition of TCE in dry nitrogen also strongly argues for dissociative electron attachment as the ﬁrst stage in the decomposition of TCE. Vitale et al. propose that the chlorinated ethylene anion will likely decompose by direct oxidation: CHClCCl− + O2 → CHClCClOO− −

CHClCClOO → CHOCl + COCl

(18) −

CHOCl → CO + HCl −

COCl → CO + Cl

−

(19)

electrons seems to be the likely primary decomposition mechanism. These reactions initiate the detachment of Cl radicals, which in turn decompose more TCE molecules by Cl radical addition to the carbon-carbon double bound causing a chain reaction as proposed by Vitale et al. [57]. Futamura and Yamamoto [62] have applied a pulsed corona and a ferroelectric (BaTiO3 ) packed-bed reactor for TCE removal. In wet nitrogen, dichloromethane, chloroform, pentachloroethane, carbon tetrachloride, 1,1,2,2,- and 1,1,1,2-tetrachloroethanes and tetrachloroethylene are detected as major byproducts for the packed-bed reactor by using GC–MS (gas chromatography–mass spectrometry). Chloro- and dichloroacetylenes, (Z)- and (E)-1,2dichloroethylenes, and 1,1,2-trichloroethane are obtained as minor byproducts. With a pulsed corona reactor, 1,1,2-trichloroethane is the main byproduct along with tetrachloroethylene, (Z)-1,2dichloroethylene and negligible amounts of polychloromethanes. When air is used as carrier gas for the decomposition with the packed-bed reactor, only phosgene could be detected. For both reactors and for both carrier gases CO, CO2 , NOx and N2 O are also formed as byproducts. Formation of DCAC is, however, not observed in aerated conditions, which is in contrast with previous mentioned studies. The authors propose a plausible reaction mechanism under deaerated conditions. In the presence of O2 , they suggest that triplet oxygen molecules scavenge intermediate carbon radicals derived from TCE decomposition in an autoxidation process. Unstable alkylperoxy radicals are generated and further oxidatively decompose to render CO and CO2 , as shown in the following general reaction:

(20)

R • + O2 → ROO• → intermediates → CO + CO2

(21)

Urashima and Chang suggest that electron impact processes produce C, H, N radicals and negative ions. According to the authors the oxidation processes will take place directly by radicals or via oxidation of negative ions [63]. They propose a mechanism of TCE destruction based on 162 reactions [64]. In a study performed by Han and Oda [65], the effect of oxygen concentration on byproduct distribution is examined. TCE decomposition efﬁciency improves with decreasing oxygen content except for 0% oxygen. The formation of DCAC is maximal for 2% oxygen, while TCAA formation decreases with decreasing oxygen concentration. They suggest that oxygen species, like O(1 D) or other states in the discharge, react more strongly with the precursor of DCAC (CHCl2 –CCl2 • ) than that of TCAA (CCl3 –CH• ). When nitrogen is used as carrier gas, the GC–MS could detect HCl, Cl2 , C2 H2 Cl2 , CHCl3 , CCl4 and C2 HCl5 as byproducts. The authors suggest that collisions between TCE and electrons and (or) N2 excited species (N2 *) generate chlorine radicals. The main decomposition mechanism is considered to be the chlorine radical chain reaction as mentioned before by other authors.

Then, in a secondary autocatalytic radical reaction, chlorine radicals add to the least substituted carbon atom of the double bond of TCE resulting in the start of a chlorine radical chain reaction [58,59]. Bertrand et al. [60] suggest that addition to the least chlorinated site is favored over addition at the more chlorinated site by at least a factor 8. A possible chlorine addition reaction mechanism for the favored reaction is as follows: CHCl2 CCl2 + O2 → CHCl2 CCl2 OO

(22)

2 CHCl2 CCl2 OO → 2 CHCl2 CCl2 O + O2

(23)

CHCl2 CCl2 O → CHCl2 COCl + Cl

(24)

CHCl2 CCl2 O → CHCl2 + COCl2

(25)

CHCl2 + O2 → CHClO + Cl + O

(26)

CHClO → CO + HCl

(27)

DCAC decomposes to form HCl, COCl2 and chlorinated radicals through the following reaction: CHCl2 COCl + Cl (or O2 ) → CCl2 COCl + HCl (or H2 O)

(28)

CCl2 COCl + O2 → CCl2 OOCOCl

(29)

2 CCl2 OOCOCl → 2 CCl2 OCOCl + O2

(30)

CCl2 OCOCl → COCl + COCl2

(31)

COCl → CO + Cl

(32)

Phosgene may further decompose through Cl abstraction by chlorine, oxygen or other radicals forming CO and Cl2 or Cl radicals. The TCE removal rate is reduced by the presence of reaction products such as phosgene, HCl and DCAC through scavenging of electrons in the plasma which could otherwise initiate more dissociative electron attachment reactions of TCE. The study of Penetrante et al. [61] shows that for small initial concentrations of TCE in dry air, the reaction with O radicals and

(33)

2.1.3.2. Benzene. Benzene has attracted attention for NTP removal because it is a carcinogenic compound that has detrimental effects on human health. Table 3 summarizes published papers on benzene removal with NTP. In order to minimize operation costs for NTP removal it is important to optimize the operation conditions. Ogata et al. [40] investigated the effects of properties of ferroelectric materials, AC frequency, initial concentration of benzene and the concentration of O2 in the background gas for the removal of benzene in air using a ferroelectric packed-bed reactor. Under dry conditions benzene removal results in a low CO2 -selectivity and in the formation of various byproducts, such as CO, C2 H2 , N2 O, NO and NO2 . To improve this technique for practical applications, Ogata et al. [66] have studied the effect of water vapor on the removal of benzene with a ferroelectric packed-bed reactor. They suggest that a portion of the lattice oxygen species in BaTiO3 pellets is deactivated

A.M. Vandenbroucke et al. / Journal of Hazardous Materials 195 (2011) 30–54

37

efﬁciency of benzene to a large extent and proves that DBD treatment is competitive with other technologies although formation of solid residues and aerosol particles are issues that must be solved to secure an effective operation. In a study by Jiang et al. [72], a DC microhollow cathode glow discharge is applied to remove 300 ppm benzene from dry air. The authors use a zero-dimensional plasma chemistry code (KINEMA) [73] to model the benzene dissociation mechanism in a benzene/dry air mixture plasma. Important dissociation reactions predicted by the model are: C6 H6 + O• → C6 H5 + OH

(34)

C6 H6 + O2 → C6 H5 + HO2 −

C6 H6 + e → C6 H5 + H + e

Fig. 3. Effect of input power on benzene removal efﬁciency and energy yield. Reprinted from Ref. [71], with permission from Elsevier.

by adsorption of H2 O on the surface of the pellets. This results in a suppressed formation of CO and N2 O, a higher CO2 -selectivity and a lower decomposition of benzene. These observations are conﬁrmed by Kim et al. [67]. The negative effect of humidity on benzene removal is ascribed to the interaction of water vapor with the surface of BaTiO3 pellets which may alter the surface state. As a consequence, plasma properties may be negatively affected [68], resulting in slower chemical destruction pathways. Cal and Schluep [69] investigated the decomposition of benzene in a DBD reactor as a function of the relative humidity (RH) without the presence of ferroelectric pellets. In both dry and wet gas streams, near complete destruction (>99.9%) of benzene is achieved and no intermediate hydrocarbons are observed with GC–MS. However, in wet gas streams the mineralization degree is greatly improved compared to dry air. Unfortunately, at high RH, a polymeric ﬁlm is produced on the dielectric plates which slowly decreases the removal efﬁciency of benzene through time. Lee et al. [70] also used a DBD discharge to decompose 100 ppm of benzene in air. The authors suggest a plausible reaction mechanism that includes the formation of all byproducts detected by GC–MS and FT-IR (Fourier transform-infrared) spectroscopy. According to the authors, the plasma can produce O radicals from O2 , which can react with benzene to form CO2 , H2 O and benzene cation by a series of reactions. Benzene could directly be decomposed by the plasma to form phenol and benzenediol. The plasma is also capable of decomposing stable CO2 to form CO radicals that would add to phenol. This leads to the formation of secondary products such as benzaldehyde and benzoic acid. Finally, decomposition of H2 O by the discharge forms H and OH radicals which lead to the formation of benzene, phenol and benzenediol. Also, Ye et al. [71] have investigated the feasibility of benzene destruction with a DBD discharge. Experiments are carried out with a laboratory scale and a scale-up DBD reactor. With the former reactor, high removal efﬁciencies are obtained with lower ﬂow rates, lower initial concentrations and higher input power (Fig. 3). In contrast, higher initial concentration and input power provide a high-energy efﬁciency for benzene removal. For the scale-up reactor, adding DBD systems in series can enhance the decomposition efﬁciency to a large extent. However, after a certain treatment time brown polymeric deposits are formed on the inside wall of the reactor which can ﬁnally lead to mechanical failure of the dielectric due to thermal energy built up. The deposit can be removed by passing air through the reactor at 6 kV for several minutes. GC–MS analysis revealed that phenol, hydroquinone and nitrophenol are the main products contained in the deposition. The feasibility study shows that multiple DBD systems in series can enhance the removal

(35) −

(36)

Modeling results reveal that the dominant dissociation reactions for benzene destruction in the DC glow discharge are atomic oxygen impact reactions. They suggest that the benzene destruction rate and efﬁciency are limited due to atomic oxygen losses in the boundary layer of the dielectric walls, which conﬁne the discharge in the direction perpendicular to the gas ﬂow direction. Satoh et al. [73] have applied a positive DC corona discharge between a multi-needle and a plane electrode for the removal of 300 ppm benzene in different N2 /O2 mixtures. Analysis of the exhaust stream is performed with FT-IR and shows C2 H2 , HCN, NO and HCOOH as intermediate products and CO2 as an end product. At low oxygen concentrations (0.2%) benzene is primarily converted into CO2 via CO, whereas at high oxygen concentrations (20%) benzene is converted into CO2 via CO and HCOOH. After treatment, benzene fragments are deposited on the plane electrode and discharge chamber at low oxygen concentrations. It is found that an increase in the oxygen concentration inhibits the decomposition of benzene, which is also the case with a DBD discharge [74]. However, with a packed-bed reactor, a higher N2 /O2 ratio improves the decomposition of benzene [40] which indicates that the effect of the amount of oxygen in the background gas depends on the type of discharge. Kim et al. [75] also investigated the inﬂuence of oxygen and found an optimum O2 concentration of 3–5% for benzene removal with a DBD discharge. Further increase of the oxygen concentration drastically decreases the decomposition efﬁciency. They suggest that higher benzene destruction at lower O2 partial pressure is due to the contribution of N radicals and excited N2 molecules. Comparison of the reaction rate constants indicates that the reaction with N2 (A3 +u ) is more plausible and is even faster than the reaction with O radicals (k = 1.6 × 10−14 cm3 molecule−1 s−1 ). However, as O2 partial pressure increases, quenching of N2 (A3 +u ) becomes signiﬁcant [76] and the rate of reaction slows down. In addition, more O atoms are produced due to direct electron-impact dissociation and collision dissociation by N2 (A3 +u ), but at the same time O atoms are also consumed in the formation of O3 . Because the gas-phase reaction between ozone and benzene is very slow (k = 1.72 × 10−22 cm3 molecule−1 s−1 ), it does not contribute to the decomposition of benzene. N2 (A3 +u ) + C6 H6 → products k = 1.6 × 10−10 cm3 molecule−1 s−1 N + C6 H6 → products k < 10−15 cm3 molecule−1 s−1 N2 (A3 +u ) + O2 → N2 + O2

(37) (38)

k = 2.4 × 10−12 cm3 molecule−1 s−1 (39)

N2 (A3 +u ) + O2 → N2 + 2 O k = 2.5 × 10−12 cm3 molecule−1 s−1 (40)

38

A.M. Vandenbroucke et al. / Journal of Hazardous Materials 195 (2011) 30–54

C6 H5 CH3 + O3 → C6 H5 CHO2 + H2 O k = 1.5 × 10−22 cm3 molecule−1 s−1

Fig. 4. Gas chromatogram of surface washed ethanolic solution from 10 min DBD discharge. Inset: Gas chromatogram of ethanolic solution of various standards employed. Reprinted from Ref. [77], with permission from Elsevier.

In a recent study by Dey et al. [77] the formation of byproducts of benzene oxidation in a Ar/O2 ﬂow with a DBD reactor is carefully analyzed with GC–FID (ﬂame ionization detector) and GC–MS. A plausible sequential reaction mechanism is given to rationalize the observation of the various byproducts. In the gas phase, only phenol and biphenyl are detected at a maximum conversion of 3%. GC analysis of an ethanolic solution of the polymeric deposit on the dielectric surface reveals the presence of substituted phenols besides phenol and biphenyl (Fig. 4). It is suggested that the intermediate phenyl radical plays the role of the primary precursor. 2.1.3.3. Toluene. Table 4 tends to give an overview of published work on toluene removal with the aid of NTP. Toluene can be regarded as the most studied VOC for abatement on laboratory scale. Therefore only a selection of papers will be discussed here. Other references can be found in Table 4. Kohno et al. [78] have applied a DC capillary tube discharge reactor and investigated the effect of gas ﬂow rate, initial toluene concentration and reactor operating conditions. According to the authors, the following destruction process can be expected in a NTP environment: C6 H5 CH3 + e− → products

k = 10−6 cm3 molecule−1 s−1

(41)

C6 H5 CH3 + O+ , O2 + , N, N2 + → C6 H5 CH3 + + O, O3 , N, N2 k = 10−10 cm3 molecule−1 s−1 C6 H5 CH3 + + e− → C6 H5 + CH3

(42) k = 10−7 cm3 molecule−1 s−1 (43)

C6 H5 CH3 + OH → C6 H5 CH3 OH k = 5.2 × 10−12 cm3 molecule−1 s−1

(47)

FT-IR spectroscopy detects CO2 , CO, NO2 and H2 O as gaseous byproducts and a signiﬁcant amount of brown particles are deposited at the exit of the reactor. It is suggested that CO2 and CO mainly form carbon and nitrogen hydride bonded aerosol particles and tars. CO2 and H2 O are observed as main reaction products by Mista and Kacprzyk [79]. They also detect a thin polymeric ﬁlm (brown residues) covering the discharge electrode and dielectric layer. Operation at higher energy densities can successively be applied to oxidize the condensed polymeric species to CO2 . Machala et al. [80] suggest that formation of aerosols including peroxy-acetylnitrates species (PANs) may be possible during toluene removal through a mechanism that is similar to formation of photochemical smog in the atmosphere. In pure nitrogen [81], GC–MS analysis showed that N2 plays a major role in the polymerisation process through the formation of C–N C and C–(NH)–bonds. A proposal of the polymerisation process is given to explain the formation of micrometric sized particles in the plasma reactor. In Ref. [82] a wire plate DBD has been used to examine the humidity effect on toluene decomposition. A maximum removal efﬁciency of 73% was achieved in a gas stream containing 0.2% H2 O in N2 with 5% O2 . This controlled humidity is governed by two opposite effects: as humidity increases, more H2 O molecules collide with high-energy electrons and form OH radicals, resulting in a higher removal efﬁciency. On the other hand, the electronegative characteristic of H2 O limits the electron density in the plasma and quenches activated chemical species, as concluded by Van Durme et al. [83]. Kim et al. [75] have conﬁrmed that 5% O2 is the optimum oxygen partial pressure in a dry nitrogen stream, as is the case for benzene. Recently, Schiorlin et al. [84] have tested three different corona discharges (positive DC, negative DC, positive pulsed) for toluene removal and have observed that process efﬁciency increases in the order positive DC < negative DC < positive pulsed. By investigating the effect of humidity on the removal efﬁciency, it is concluded that for both negative DC and positive pulsed corona, OH radicals are involved in the initial stage of toluene oxidation. When the RH was greater than 60%, removal efﬁciency slightly drops due to saturation and inhibition of the OH radical forming reactions, i.e. dissociation of H2 O molecules induced by interaction with electrons or by reaction with O(1 D). A positive DC corona discharge has been applied by Van Durme et al. [83] in order to abate toluene from indoor air and to unravel the degradation pathway. The removal of toluene is achieved with a characteristic energy density of 50 J/L. Fig. 5 shows that partially oxidized intermediates are formed under the applied conditions. By determining the effect of humidity, the authors ﬁnd out that OH radicals play a major role in the oxidation kinetics due to initiation by H-abstraction or OH-addition. The byproducts detected by GC–MS consist of benzaldehyde, benzylalcohol, formic acid, nitrophenols and furans.

(44) 2.2. Combined with catalyst

C6 H5 CH3 + OH → C6 H5 CH2 + H2 O k = 7 × 10−13 cm3 molecule−1 s−1

(45)

C6 H5 CH3 + O → C6 H5 CH2 O + H k = 8.4 × 10−14 cm3 molecule−1 s−1

(46)

2.2.1. What is plasma–catalysis? Many studies have shown that NTP is attractive for the removal of NOx , SOx , odours and VOCs. There is, however, a consensus among researchers that application of NTP for VOC abatement suffers from 3 main weaknesses, i.e. incomplete oxidation with emission of harmful compounds (CO, NOx , other VOCs), a poor energy efﬁciency and a low mineralization degree.

A.M. Vandenbroucke et al. / Journal of Hazardous Materials 195 (2011) 30–54

39

Fig. 5. Chromatogram of GC–MS analysis for the identiﬁcation of toluene degradation products. Reprinted from Ref. [83], with permission from Elsevier.

The combination of NTP with heterogeneous catalysts can be divided in two categories depending on the location of the catalyst: in-plasma catalysis (IPC) and post-plasma catalysis (PPC). The latter is a two-stage process where the catalyst is located downstream of the plasma reactor while the former is a single stage process with the catalyst being exposed to the active plasma. In literature, several different terms and corresponding abbreviations have already been proposed to represent IPC and PPC. For in-plasma catalysis, one can ﬁnd among others: plasma-driven catalysis (PDC) [13], inplasma catalysis reactor (IPCR) [85], single-stage plasma–catalysis (SPC) [86], plasma and catalyst integrated technologies (PACT) [87] or combined plasma catalysis (CPC) [88,89]. For PPC, the following terms have been proposed: plasma-enhanced catalysis (PEC) [13], post-plasma catalysis reactor (PPCR) [85], two-stage plasma catalysis (TPC) [86]. In plasma–catalysis, synergetic effects are related to the activation of the catalyst by the plasma. Activation mechanisms include ozone, UV, local heating, changes in work function, activation of lattice oxygen, adsorption/desorption, creation of electron–hole pairs and direct interaction of gas-phase radicals with adsorbed pollutants [75]. The plasma–catalyst interactions described in the following paragraphs contribute to one or more of these catalyst activation mechanisms. The presented experimental ﬁndings, applying to speciﬁc working conditions, may appear as scattered pieces of information. Indeed further research is needed to connect the loose ends and unravel the detailed mechanisms. However, it is meaningful to try and extract some general pathways at this stage. 2.2.1.1. Inﬂuence of the catalyst on the plasma processes. Discharge mode: The physical properties of a discharge will be affected if a catalyst is introduced into the discharge zone. When for example a dielectric surface is introduced in the gap of a streamer-type discharge, the discharge mode at least partially changes from bulk streamers to more intense streamers running along the surface (surface ﬂashover) [90]. Similar ﬁeld effects can lead to higher average electron energies when the discharge zone is ﬁlled with ferroelectric pellets, leading to a more oxidative discharge [91]. Parameters that inﬂuence the effect of the packed bed on the discharge are the dielectric constant of the pellet material and the size and shape of the pellets. The dielectric constant affects the electric ﬁeld in the void between the pellets and thereby the mean electron energy. With increasing pellet size the number of microdischarges decreases, but the amount of charge that is transferred per microdischarge increases [92]. Reactive species production: Obviously, introducing a heterogeneous catalyst changes the physical characteristics of the discharge, so the chemical activity will be affected as well. Roland et al. [18]

studied the oxidation of various organic substances immobilized on porous and non-porous alumina and silica catalysts and concluded that short-living active species are formed in the pore volume of porous materials when exposed to NTP. On the other hand, introducing a catalyst can reduce the concentration of ionic species [93]. However, this effect did not impair the catalyst’s role in reducing the emissions of ozone and carbon monoxide for this particular application (indoor air control).

2.2.1.2. Inﬂuence of the plasma on the catalytic processes. Catalyst properties: Non-thermal plasmas are used for catalyst preparation [94–99]. Plasma treatment of the catalyst enhances the dispersion of active catalytic components [100,101] and inﬂuences the stability and catalytic activity of the exposed catalyst material [102]. The oxidation state of the catalyst can also be altered by NTP. For instance, when a Mn2 O3 catalyst is exposed for a long time to a DBD plasma, X-ray diffraction spectra reveal the presence of Mn3 O4 , a lower-valent manganese oxide with a larger oxidation capability. Due to plasma–catalyst interactions, less parent Ti–O bonds are found on TiO2 surfaces after several hours of discharge operation [103]. Even new types of active sites with unusual properties may be formed [104], such as stable Al–O–O* with a lifetime exceeding more than two weeks, as observed in the pores of Al2 O3 in IPC experiments [104]. Plasma exposure can result in an increase or decrease of the speciﬁc surface area or in a change of catalyst structure [100,102,105]. Adsorption: Adsorption processes play an important role in plasma–catalytic reaction mechanisms. If the catalyst has a significant adsorption capacity for pollutant molecules, it prolongs the pollutant retention time in the reactor. In the case of IPC, the pollutant concentration in the discharge zone is increased. The resulting higher collision probability between pollutant molecules and active species enhances the removal efﬁciency. Adsorption of VOC and active species increases with the porosity of the catalyst [106]. Under conditions where plasma-generated ozone is not effective in itself to destroy pollutants, high decomposition rates are obtained due to the adsorption of ozone on the catalyst surface and the subsequent dissociation into atomic oxygen species [107]. Humidity is a critical parameter in plasma–catalytic processes. The adsorption of water on the catalyst surface results in a decrease of the reaction probability of the VOC with the surface and therefore reduces the catalyst activity [82]. Thermal activation: Although gas heating will result in higher catalyst surface temperatures [108], the heating effect is in general too small to account for thermal activation of the catalyst. However, hot spots can be formed in packed-bed reactors as a result of localized heating by intense microdischarges that run between

40

A.M. Vandenbroucke et al. / Journal of Hazardous Materials 195 (2011) 30–54

sharp edges and corners of adjacent pellets. Increased catalyst temperatures can promote catalytic VOC removal [109]. Plasma-mediated activation of photocatalysts: In photocatalysis, VOCs are adsorbed on the surface of a porous semiconductor material that is exposed to UV radiation. The UV photons generate electron–hole pairs, inducing the subsequent oxidation of the adsorbed VOC by valence band holes. In a ﬁnal step the oxidation products are desorbed. Among other photocatalysts (e.g. ZnO, ZnS, CdS, Fe2 O3 , WO3 ), TiO2 is one of the most efﬁcient for the decomposition of a wide range of VOCs. Moreover, the combination of TiO2 with NTP results in higher oxidation efﬁciencies and better selectivity to CO2 . For the anatase phase of TiO2 , having a bandgap of 3.2 eV, it takes a photon with a wavelength shorter than 388 nm to create an electron–hole pair. Although there are excited nitrogen states that emit light in this wavelength range, there is experimental evidence that photocatalysis induced by UV light from the plasma cannot explain the observed synergy in several hybrid plasma/TiO2 systems reported in literature. For instance, Sano et al. [110] has detected no enhancement in acetylene conversion when the reactor walls are coated with TiO2 . Emission spectra of the surface discharge plasma with and without catalyst coating reveal that UV light from the plasma is absorbed by TiO2 , but the intensity is too weak for photoactivation. This observation has been conﬁrmed by Huang et al. [111], who employed a wire-cylinder DBD reactor with a photocatalyst sheet stuck along the inner wall of the tube. Kim et al. have tested a DBD reactor packed with Ag/TiO2 for benzene removal [112]. When O2 -benzene mixtures are diluted with argon, signiﬁcantly higher decomposition efﬁciencies are observed compared to N2 dilution. This result suggests that the role of UV light for photoactivation is negligible because light emission from excited argon ranges in the visible range (400–850 nm). However, other groups report that UV light emitted from the plasma can act as a source for activation of TiO2 [70,113,114]. Subrahmanyam et al. [115] suggest that the increased activity with sintered metal ﬁbres modiﬁed with TiO2 might be related to activation as well as to photocatalytic action in the presence of UV light emitted by the plasma discharge. In some cases, TiO2 shows plasma-induced catalytic activity under conditions where there is no or very little UV emitted by the plasma [112,116]. Direct plasma activation has been observed when TiO2 is exposed to an atmospheric pressure argon discharge at room temperature [117]. The question then arises how the plasma-exposed TiO2 is activated, if not by UV photons. Different mechanisms to bridge the TiO2 band-gap by plasma-driven processes can be envisaged, but to date there is insufﬁcient information to elaborate on the relative importance of electrons, ions, metastables, charging effects, surface recombination, etc. 2.2.2. Different types of catalysts As in classical heterogeneous catalysis, the catalyst material can be introduced in the hybrid system in different ways for both IPC (Fig. 6) and PPC: in the form of pellets (a so-called packed-bed conﬁguration) [91,118–121], foam [82,100,122,123,102] or honeycomb monolith [93,124–127], as a layer of catalyst material [128] or as a coating on the reactor wall [110,129] or electrodes [115,130–135]. Many catalysts have been tested for VOC abatement with IPC and PPC. Historically, the ﬁrst materials tested were porous adsorbents placed inside the discharge region as in references [41,136]. The idea is that, by introducing these materials, the retention time of VOC molecules would increase along with the probability of surface reactions with active chemical plasma species (electrons, radicals, ions, photons). Adsorbents that were used to achieve a more complete oxidation are ␥-Al2 O3 [17,18,136,137] and zeolites or molecular sieves [17,138–142]. Furthermore, these materials are coated or impregnated with (noble) metals such as silver, palladium, platinum, rhodium, nickel, molybde-

Fig. 6. Most common catalyst insertion methods for IPC conﬁguration. Reprinted from Ref. [228], with permission from Elsevier.

num, copper, cobalt or manganese to provide catalytic activity [75,105,107,109,142–151]. Adsorbents also function as support for metal oxides [20,149,152–158]. Extensive attention has been given during recent years to the use of photocatalysts, in particular to TiO2 . In most studies, TiO2 is inserted in the discharge region in order to achieve activation through different mechanisms. This catalyst has also been coated with (noble) metals [67,75,107,112,116,159,160] and metal oxides [138,161–163]. Additionally, it has been used as a coating on activated carbon ﬁlter [111] or ﬁbre [164], on glass ﬁbres [165,166] or beads [70,113,167], nickel foam [168], silica gel pellets [129] and on UV lamp [169]. 2.2.3. VOC abatement Tables 6–9 give a summary of literature on VOC removal with plasma–catalysis. For each paper catalyst information and operating conditions are presented along with the maximum removal efﬁciency and energy density. In this section, particular attention is again paid to the most studied target compounds, i.e. trichloroethylene, benzene and toluene. Table 9 presents a list of other relevant, but less frequently studied VOCs that have been examined in plasma–catalytic studies. For more details about operating conditions and results, the reader can consult the corresponding references. 2.2.3.1. Trichloroethylene. Table 6 presents published papers regarding TCE abatement. Oda et al. [162] have investigated the effect of TCE initial concentration, pellet size and sintering temperature for TiO2 catalysts on the TCE decomposition performance. When the barrier type reactor was ﬁlled with TiO2 sintered at 673 K, the breakdown voltage to generate NTP greatly reduces in comparison to the empty reactor and the reactor ﬁlled with TiO2 sintered at 1373 K. They suggest that the nonuniform geometrical distribution of the disk-like dielectric pellets sintered at 673 K disturbed the electric ﬁeld and generated an electric ﬁeld concentration at the contacting area of the pellets. This results in the formation of contacting point discharges or surface discharges on the pellet surfaces, lowering the breakdown voltage and improving the decomposition energy efﬁciency. Moreover, they indicate that too ﬁne TiO2 particles disturb the gas ﬂow and cause insufﬁcient ﬁlling of the discharge area with plasma. In another study by Oda et al. [170], MnO2 is used as a postplasma catalyst in a direct (contaminated air is directly processed by the plasma) and an indirect process (plasma-processed clean air is mixed with the contaminated air). Manganese oxide is very efﬁ-

A.M. Vandenbroucke et al. / Journal of Hazardous Materials 195 (2011) 30–54

41

Table 6 Overview of published papers on TCE removal with plasma–catalysis. Plasma type

Catalyst

Position

Tcat (K)

Carrier gas

Flow rate (mL/min)

Concentration Maximum range removal (ppm) efﬁciency (%)

Energy density (J/L)

References

DBD DBD

MnO2 TiO2 /SMF CoOx (3 wt%)/SMF MnOx (3 wt%)/SMF TiO2 /MnOx /SMF MnOx (3 wt%)/SMF MnO

PPC IPC

293 –

Air Air

500 700

250 250

95–99 >99

240 –

[65] [131]

IPC IPC

293 293

Air Dry air

500 10002000

150–200 250

95–99 >99>99

550 120

[133] [144]

DBD

TiO2 V2 O5 (0.7 wt%)/TiO2 V2 O5 (4.6 wt%)/TiO2 WO3 (4.2 wt%)/TiO2

IPC

293

Dry air

400

100

>99 95–99 90–95 >99

180 140 140 180

[161]

DBD

TiO2 sintered at 1373 K 0.5–1 mm 1–2 mm 2–3 mm TiO2 sintered at 673 K MnO2 Au/SBA-15 TiO2 Pd(0.05 wt%)/Al2 O3

IPC

293

Dry air

400

1000

DBD DBD

DBD DBD DC positive corona DC negative glow Surface discharge

V2 O5 /TiO2 Cu–ZSM-5

[162] >99 >99 >99 >99

200 120 120 120

IPC PPC IPC PPC

293 – – 373

Dry air Dry air Dry air Humid air

400 510 1500 2000

1000 430 100 600–700

>99 >99 85 80

120 670 600 300

[170] [172] [118] [147]

IPC

293

Dry air

400

1000

>95 >95

50 50

[143]

cient in enhancing the decomposition efﬁciency for both processes. The catalysts effectiveness to dissociate ozone generates oxygen radicals which are excellent oxidizers for TCE removal. Han et al. [171] have further examined the effect of the manganese dioxide post-plasma catalyst for the direct and indirect process. For the direct process oxygen species, generated from collisions between excited species (or electrons) with O2 , mainly oxidize TCE into DCAC. The increased decomposition efﬁciency for the direct process is ascribed to the oxidation of the remaining TCE into trichloroacetaldehyde (CCl3 –CHO, TCAA) by oxygen species produced during ozone decomposition at the surface of MnO2 . The COx yield increases from 15% to 35% at an energy density of 120 J/L when MnO2 is present. When the energy density is raised to 400 J/L, a COx yield of 98% is established. For the indirect process, similar

Fig. 7. Selectivity to CO and CO2 as a function of input energy for inner electrodes made of SMF and MnOx /SMF. Reprinted from Ref. [133], with permission from Elsevier.

conclusions are made although the COx yield is not as good as for the direct process. Magureanu et al. [133] have tested a plasma–catalytic DBD reactor with an inner electrode made of sintered metal ﬁbres (SMF) coated by transition metal oxides. Fig. 7 shows the CO and CO2 selectivity over the range of energy densities used. The selectivity to CO2 reaches 25% with the SMF and showed a signiﬁcant improvement with MnOx /SMF, up to 60%. The use of MnOx /SMF does, however, not substantially lower the selectivity to CO. Thus, as compared to the reactor with SMF electrode, TCE conversion and CO2 selectivity were signiﬁcantly enhanced using MnOx /SMF. The ability of MnO2 to decompose ozone in situ, produces strong oxidizing atomic oxygen species on the catalyst surface. These species may lead to an enhanced oxidation of TCE resulting in a high CO2 selectivity [115,131,133]. After reaction, XPS (X-ray photoelectron spectroscopy) analysis of the catalyst has revealed that both manganese and iron have preserved their initial oxidation state. The used catalyst, however, shows an enrichment of iron on the catalyst surface suggesting a redispersion of manganese on the surface during reaction. Finally, XPS also reveals some chlorine deposition on the catalyst surface after reaction. In another study conducted by Magureanu et al. [172], gold nano-particles embedded in SBA-15 have been tested for PPC. The catalyst with the least amount of Au (0.5 wt%) seems to enhance the COx selectivity the most and has the best catalytic performance. As for MnO2 , the Au/SBA-15 can dissociate ozone, produced in the plasma, to oxygen radicals that decompose TCE. They suggest that in the presence of ozone generated in the plasma, isolated gold cations are the active sites that elucidate the catalytic behaviour. To achieve a more complete oxidation of TCE at a reduced energy cost, Morent et al. [118] have used a hybrid plasma–catalyst system with cylindrical TiO2 pellets for IPC. They suggest that the increased removal fraction for the plasma–catalytic system can be explained through adsorption on and/or photoactivation of TiO2 . Adsorption of TCE molecules on the surface of TiO2 increases the residence time of TCE in the discharge.

42

A.M. Vandenbroucke et al. / Journal of Hazardous Materials 195 (2011) 30–54

Table 7 Overview of published papers on benzene removal with plasma–catalysis. Flow rate (mL/min) Concentration range Maximum removal (ppm) efﬁciency (%)

Energy density (J/L)

References

Dry air

100 500 500

100 105 105

60 54 50

900 360 320

[93]

293

Dry air

250

300–380

12 16

170 170

[108]

IPC

–

Dry air

400

200

90 >99 >99

3150

[163]

Ag(2 wt%)/TiO2 Kr/I2 (KrI* excimer UV radiation) TiO2 TiO2 /Al2 O3 silica gel

IPC IPC

373 293

Dry air Dry air

4000 13 × 103 –130 × 103

110 30–940

>99 66.5

125 –

[176] [230]

IPC IPC IPC

293 – –

Dry air Dry air Dry air

– 200 100

188 100 300

98 50 85

– 140 –

[262] [70] [90]

Multistage corona a

TiO2 Sol–gel TiO2 Pt/Sol–gel TiO2

IPC

293

Dry air

60

1500

92.7 91.7 >99

–

[222]

Surface discharge

Ag(1 wt%)/TiO2

IPC

373

Dry air Humid air

200–3000

200–210

89 86

383 391

[67]

Surface discharge

Ag(4 wt%)/TiO2 IPC Ni(2 wt%)/TiO2 Ag(0.5/5 wt%)/Al2 O3 Pt(0.5 wt%)/Al2 O3 Pd(0.5 wt%)/Al2 O3 Ferrierite Ag(2 wt%)/H–Y

373

Dry air

4000–104

200

>99 >99 >99 >99 >99 >99 >99

89–194

[75]

Packed-bed DBD

TiO2 Pt(1 wt%)/TiO2 V2 O5 (1 wt%)/TiO2

IPC

373

Dry air

2000

203–210

82 80 90

388 391 383

[174]

BaTiO3 packed-bed

TiO2 Ag(0.5 wt%)/TiO2 Al2 O3 Ag(0.5 wt%)/Al2 O3 TiO2 Ag(0.5 wt%)/TiO2 Al2 O3 Ag(0.5 wt%)/Al2 O3

IPC

292

Dry air

1000

500

60

[180]

PPC

292

66 60 52 49 34 46 28 39

Plasma type

Catalyst

Position Tcat (K) Carrier gas

DBD

TiO2 MnO2 TiO2 –silica

IPC

–

DBD

TiO2 MnO2

IPC

DBD

TiO2 Pt(1 wt%)/TiO2 V2 O5 (1 wt%)/TiO2

DBD DBD DBD DBD glow discharge Pulsed corona

a

Four stages in serie.

To conﬁrm the presence of excited species of nitrogen, Subrahmanyam et al. give an UV–vis emission spectrum of the DBD plasma discharge in the wavelength range 250–500 nm. It is proven that emission of excited nitrogen molecules (N2 *) is in the range of the band gap of the TiO2 /SMF catalyst [115]. They suggest that the increased activity of TiO2 /SMF might be due to photocatalytic action in the presence of UV light as well as activation of TiO2 by the plasma discharge. Vandenbroucke et al. [147] have investigated the use of a DC glow discharge combined with Pd/␥-Al2 O3 located in an oven downstream. When the catalyst temperature was set at 373 K, the combined system showed synergistic effects on the removal of TCE. By comparing the experimental removal efﬁciency of the hybrid system with the removal calculated by multiplying the individual effects (plasma and catalyst alone), 12–22% additional TCE was decomposed. A more elaborated review on plasma–catalytic abatement of TCE can be found in [173]. 2.2.3.2. Benzene. Ogata et al. [136] have performed much research on the removal of benzene with plasma–catalysis. In a ﬁrst study, they test an adsorbent hybrid reactor packed with a mixture of BaTiO3 and Al2 O3 pellets and compare the results with a BaTiO3

packed reactor and a two-stage reactor (BaTiO3 packed reactor with Al2 O3 downstream). The hybrid reactor shows the best performance, owing to its better energy efﬁciency, CO2 -selectivity and suppressed N2 O formation. The combined effect of benzene concentration on Al2 O3 followed by surface decomposition and gas-phase reaction is thought to be responsible for the enhanced decomposition. Cyclic operation of adsorption and plasma discharge is suggested to further improve the energy efﬁciency. In Ref. [137] they continue examining a catalyst hybrid reactor with metal supported Al2 O3 and have found that Ag-, Co-, Cu and Ni-supported Al2 O3 shows a slightly better CO/CO2 ratio and a lower N2 O formation than the adsorbent hybrid reactor. Next, a zeolite hybrid plasma reactor (mixture of zeolite and BaTiO3 ) has been applied for dilute benzene decomposition [139]. The higher adsorption capacity of zeolite structures compared to alumina allows a higher decomposition efﬁciency and CO/CO2 ratio if the micropore surface area is large enough for accommodation of benzene molecules. The authors have also found that benzene adsorbed outside of a zeolite crystalline pore decomposed more easily than that inside a zeolite pore. In Ref. [145] they expand the study and examine the effect of BaTiO3 pellet size and mixing ratio of BaTiO3 and adsorbent, catalyst or zeolite. Plasma energy is found to be almost independent of the pellet size. However, with pellets larger than 2 mm in diam-

A.M. Vandenbroucke et al. / Journal of Hazardous Materials 195 (2011) 30–54

43

Table 8 Overview of published papers on toluene removal with plasma–catalysis. Plasma type

Catalyst

Position

Tcat (K)

Carrier gas

Flow rate (mL/min)

Concentration range (ppm)

Maximum removal efﬁciency (%)

Energy density (J/L)

References

DBD

Al2 O3

IPC

Air (18% RH)

104

220

[17]

Fe2 O3 /MnO honeycomb SMF CoOx (3 wt%)/SMF MnOx (3 wt%)/SMF

PPC

Dry air

2500

85

60–65 80 65

110

DBD

293 373 293

72

[127]

IPC

–

Air

500

250

60 70 65

160

[131]

[134]

DBD

DBD

SMF CoOx (3 wt%)/SMF MnOx (3 wt%)/SMF Cu

IPC

–

Air

500

500

90 92 95 90

298

DBD

MnPO4 Mn–APO-5 Mn–SAPO-11

PPC

673

Air

50–150

560

70 65 70

900–2700 [142]

DBD

Ag/TiO2

IPC

373

Air

4000

101

95

125

[176]

DBD

Ti–MPS Mn(5 wt%)–Ti–MPS Mn(10 wt%)–Ti–MPS

PPC

–

Air

200

1000

45 58 75

300

[263]

Wire-cylinder DBD

TiO2 /activated carbon ﬁlter Al2 O3 TiO2 /Al2 O3 MnO2 (5 wt%)/Al2 O3 MnO2 (10 wt%)/Al2 O3 MnO2 (15 wt%)/Al2 O3

IPC

–

Air (0.5% H2 O)

200

100

55

–

[111]

IPC

–

Dry air

2000

186

75 86 84 96 96

700

[20]

TiO2 /glass pellets TiO2 /Al2 O3 /Ni foam MnO2 /Al/Ni foam Mn-1 Mn-2 Mn-3 N150 (MnO2 –Fe2 O3 ) Al2 O3 MnO2 (9 wt%)/Al2 O3 Activated carbon (AC) MnO2 (3 wt%)/AC

IPC PPC IPC PPC

293 – – 300

Dry air Dry air 5% O2 /N2 Air

600 200 100 300

1100 50 50 200

80 95 >95 90–95

1000 900 750 1400

[167] [122] [100] [264]

PPC

–

Air

588

240

76 74 88 98.5 99.7

172

[202]

Wire-cylinder DBD

Wire-cylinder DBD Wire-plate DBD Wire-plateDBD DBD (pulsed)

DBD packed with glass beads

Multistage packed-bed DBD

MnO2 MnO2 –CuO

PPC

–

Air

104

70

>99 >99

340

[265]

BaTiO3 packed-bed

Al2 O3 Ag2 O(7 wt%)/Al2 O3 MnO2 (7 wt%)/Al2 O3 Al2 O3 Ag2 O(7 wt%)/Al2 O3 MnO2 (7 wt%)/Al2 O3

IPC

673 573 603 698 573 603

Dry air

1000

500

95 >99 >99 78 >99 >99

60

[153]

TiO2 Al2 O3 Ag(0.5 wt%)/Al2 O3 TiO2 Al2 O3 Ag(0.5 wt%)/TiO2 Ag(0.5 wt%)/Al2 O3

IPC

753

Dry air

1000

500

91

[180]

PPC

886

60 >99 >99 95 >99 95–99 99

PPC IPC

513 433

Air Dry air

2 × 104 1000

330 200

90–95 85

142 140–150

[126] [130]

IPC

–

Air

100

300

>95

5.43.5

[90]

Pulsed corona

Pt-honeycomb Reticulated vitreous carbon Pt/Rh coated electrodes AlO2 Silica gel Al2 O3

IPC

–

Air

400

1100

>99

1100

[181]

DC positive corona

TiO2

IPC PPC

–

Air

1000

80–100

75 70

160 330

[182]

DC positive corona

TiO2 CuO–MnO2 /TiO2

IPC PPC

293

Dry air

104

0.5

82 78

17 2.5

[186]

DC positive corona

Cu–Mn/TiO2 (a) N140 N150 Pd(0.5 wt%)/Al2 O3 Cu–Mn/TiO2 (b)

PPC

293

Air (50% RH)

104

0.5

40 47 34 47 62

14 16 16 10 20

[107]

BaTiO3 packed-bed

Pulsed corona Pulsed corona

Pulsed corona

PPC

44

A.M. Vandenbroucke et al. / Journal of Hazardous Materials 195 (2011) 30–54

Table 8 (Continued) Plasma type

Catalyst

Position

Tcat (K)

Carrier gas

Flow rate (mL/min)

Concentration range (ppm)

Maximum removal efﬁciency (%)

Energy density (J/L)

References

Wire-cylinder corona

TiO2 (3 wt%)/glass beads TiO2 (3 wt%)/Al2 O3

IPC

–

10% O2 /N2

4000

1000

70–75

–

[113]

Positive DC streamer Surface discharge

Surface discharge Pulsed wire-cylinder DBD

Cu–Mn/Al2 O3 Ni/cordierit honeycomb Mn–Cu/cordierit honeycomb V/cordierit honeycomb zeolites Pt/Al2 O3

80–85 PPC PPC

573 –

Air Air

133 × 10 666 × 103

45 30

96 40–45

20 –

[146] [93]

IPC IPC

– 468

Air (0.5% H2 O) Air

500 2 × 104

200 300

– 92

– –

[140] [266]

eter sparking occurs earlier. For the catalyst hybrid reactor (with metal supported Al2 O3 ), larger BaTiO3 pellets in comparison to catalyst pellets, are beneﬁcial because high-energy plasma is formed around the contact points of the BaTiO3 pellets (Fig. 8). This shows the importance of the combination method to effectively induce catalytic properties. Kim et al. have also tested various catalyst formulations and reactor types to enhance the decomposition of benzene with NTP. A BaTiO3 packed-bed reactor has been modiﬁed by replacing the ferroelectric material with TiO2 , Pt/TiO2 or Ag/TiO2 pellets [174]. The reactor is placed in an oven that controls the temperature at 373 K. Experiments reveal that the catalytic activity for benzene decomposition is in the order Ag/TiO2 > TiO2 > Pt/TiO2 . The silver catalyst also improves the CO2 -selectivity with 15% compared to the BaTiO3 packed-bed reactor. Beside CO2 and CO, no other byproducts are formed, which is conﬁrmed by good carbon balances. Results indicate that the energy density is the governing factor for benzene decomposition rather than the amount of Ag/TiO2 (grams of catalyst) in the reactor [175] or the gas residence time [176]. However, larger amounts of Ag/TiO2 slightly reduce the formation of N2 O. In this study, formic acid is found as minor byproduct at lower energy density. While a pulsed corona and surface discharge reactor form aerosols during benzene removal, negligible amounts are detected in the reactor packed with Ag/TiO2 [175]. In a subsequent study [177], Ag-loading amount on TiO2 (percentage of Ag on catalyst) conﬁrms to have no effect on the benzene removal. This parameter,

3

however, plays an important role for the oxidative decomposition of intermediates on the TiO2 surface, indicated by the carbon balance. Larger Ag-loading seems to beneﬁt the carbon balance and CO2 -selectivity. Further work has examined the activation mechanism of the Ag/TiO2 catalyst in the hybrid reactor [112]. Thermal catalytic experiments and comparison of the effects of dilution gases (Ar, N2 ) on benzene removal respectively reveal that temperature is not an important parameter and contribution of plasma generated UV light to the photoactivation of the catalyst is negligible. The authors therefore suggest that in situ decomposition of ozone over Ag/TiO2 and plasma-induced catalysis at higher energy density play a dominant role. The observed zero-order kinetics to benzene concentration supports the latter assumption. The catalyst shows good durability against catalyst deactivation for over 150 h of continuous operation tests [176]. Finally, Kim et al. [75] have tested a cycled system of adsorption and oxygen plasma as earlier proposed by Ogata et al. [136] and Song et al. [17]. Benzene oxidation is examined as function of oxygen partial pressure (0–80% O2 ) and different catalyst types (TiO2 , ␥-Al2 O3 , zeolites) inside the reactor. An increase of O2 partial pressure improves both the decomposition and CO2 -selectivity of benzene regardless of the catalyst used. Tests with the cycled system demonstrate that the regeneration mode must be done with pure oxygen to fully suppress harmful Nx Oy formation. The authors suggest a plausible reaction mechanism where removal of benzene mainly proceeds on the surface of the main catalysts (Fig. 9).

Fig. 8. Image of plasma discharge in a (a) BaTiO3 packed-bed DBD and in hybrid reactors with mixtures of (b) BaTiO3 > catalyst and (c) BaTiO3 < catalyst. Reprinted from Ref. [145], with permission from Elsevier.

A.M. Vandenbroucke et al. / Journal of Hazardous Materials 195 (2011) 30–54

45

Fig. 9. Plausible mechanism for IPC for VOCs on various catalysts. Reprinted from Ref. [75], with permission from Elsevier.

Fig. 10. Mechanism for MnO2 -catalyzed oxidation of benzene. Reprinted from Ref. [129], with permission from Elsevier.

Recently, Fan et al. [178] have also investigated a cycled system with a storage and a discharge stage packed with a metal supported zeolite (Ag/HZSM-5). High oxidation rate of adsorbed benzene as well as low energy cost (3.7 × 10−3 kWh/m3 ) are achieved at a moderate discharge power. Additionally, Ag/HZSM-5 exhibited good stability during cycled operation. In a study by Futamura et al. [129], a DBD discharge is applied to investigate the synergistic effect of ﬁlling the plasma reactor with different catalysts (TiO2 , MnO2 and TiO2 -silica gel). They suggest a mechanism for MnO2 -catalyzed oxidation of benzene (Fig. 10). Apparently, adsorption of ozone forms oxygen atoms on the MnO2

surface which partially desorb as O(3 P) in the gas phase, acting as possible oxidants for benzene decomposition. Park et al. [163] have attached sheet type catalysts (TiO2 , Pt/TiO2 and V2 O5 /TiO2 ) on the dielectric barrier of a DBD discharge. Benzene decomposition efﬁciency decreases in the order V2 O5 /TiO2 > Pt/TiO2 > TiO2 . Suppression of N2 O formation and improved mineralization degrees are obtained with all catalysts. Results indicate that high-energy electrons along with UV light generated from DBD plasma excite the TiO2 catalysts. A hybrid plasma-photocatalyst system has also been tested by Lee et al. [70]. Comparison of OES (optical emission spectroscopy) spectra of the DBD glow discharge and an UV lamp conﬁrms that the discharge emits UV light with an energy corresponding to 3–4 eV. The authors assume that photocatalysis could be possible using plasma as a photoactivation source, as proposed by Park et al. [163]. Titanium dioxide is coated on glass beads and on three types of ␥Al2 O3 with different surface area, pore volume and pore diameter. High porous alumina dramatically enhances the benzene conversion and mineralization degree.

Table 9 Published papers on removal of other VOCs with plasma–catalysis. Target VOC

References

Acetaldehyde Acetone Acetylene Dichloromethane Formaldehyde Methane Methanol Propane Propene Styrene Tetrachloromethane Xylene

[110,267,268] [91,160,167] [106,165,166,241,269] [105,144,270,271] [220,272] [90,148,273] [249] [17,149,157] [149] [151,176,274,275] [150,198] [159,176,276–282]

Fig. 11. FT-IR spectra showing the plasma–catalytic destruction of benzene with Ag/␥-Al2 O3 , as a function of temperature in a two-stage conﬁguration. Reprinted from Ref. [180], with permission from Elsevier.

46

A.M. Vandenbroucke et al. / Journal of Hazardous Materials 195 (2011) 30–54

Fig. 12. Effect of energy density on energy yield of different catalysts (RH: 20%; initial toluene concentration: 105 ppm; gas ﬂow rate: 450 mL/min). Reprinted from Ref. [183], with permission from Elsevier. Fig. 13. Inﬂuence of SMF modiﬁcation and energy density on the conversion of 100 ppm toluene.

In Ref. [179] the inﬂuence of humidity on benzene removal is investigated with a DBD packed with Raschig rings coated with nano TiO2 ﬁlms. Humidity negatively affects decomposition of benzene for three reasons: deactivation of high-energy electrons, inhibition of ozone formation and suppression of the catalyst activity of TiO2 for benzene oxidation with ozone. Harling et al. [180] have examined the effect of temperature (293–886 K) and catalyst position (IPC/PPC). Fig. 11 shows IR spectra of the plasma–catalytic destruction of benzene as a function of temperature with Ag/␥-Al2 O3 in a two-stage conﬁguration. When compared to thermal catalysis, NOx formation is detected and increasing amounts are produced at elevated temperatures. Additionally, higher levels of destruction are observed at lower temperatures for plasma–catalysis.

2.2.3.3. Toluene. Table 8 gives an extensive overview of the papers that have been published on the plasma–catalytic abatement of toluene. A concise discussion is given on selected papers. Other references can be found in Table 8. Song et al. [17] have applied a DBD packed with macro-porous ␥-Al2 O3 and investigated the effect of adsorption and elevated temperature. Higher operating temperatures (373 K) cause a reduction in adsorption capability. However, toluene removal is more favorable under these conditions in comparison with the use of non-adsorbing glass beads. The use of ␥-Al2 O3 beads proves to reduce some of the gas-phase byproducts, such as O3 and HNO3 , generated by the NTP process. Malik and Jiang [181] have also indicated that selecting the alumina packing with higher overall surface area can lower ozone generation without affecting the destruction efﬁciency of toluene. In a study by Li et al. [182], a DC streamer corona discharge is employed in combination with TiO2 pellets. Positioning the photocatalyst between the needle and mesh electrodes beneﬁts the plasma discharge due to a higher streamer repetition rate. This conﬁguration shows the best performance for decomposition (76%) and energy efﬁciency (7.2 g/kWh). This is attributed to the simultaneous decomposition of gas phase and adsorbed toluene and to possible TiO2 activation by plasma inducing catalytic reactions. In absence of the TiO2 layer, both the decomposition (44%) and efﬁciency (3.2 g/kWh) signiﬁcantly drop. The authors claim that intermittent operation can improve the efﬁciency due to the regeneration of the catalyst surface through desorption during the discharge.

Reprinted from Ref. [134], with permission from Elsevier.

Guo et al. [183] have applied a DBD to study the effect of MnOx /Al2 O3 /nickel foam for IPC. Earlier results have conﬁrmed that MnOx /Al2 O3 /nickel foam is the most effective for toluene removal among different catalysts tested [102]. Fig. 12 shows that the MnOx catalyst greatly improves the energy yield as compared to the plasma-alone system. A sampling method has been developed to detect OH radicals in the gas phase and on the catalyst surface [184]. The catalyst can enhance the toluene removal efﬁciency due to efﬁcient reactions of OH radicals with toluene on the surface or the active sites and other active species on the catalyst. With the plasma–catalytic system toluene removal decreases with increased humidity. It is suggested that water molecules cover the catalyst surface, resulting in a lower reaction probability [185]. Indeed, Van Durme et al. [186] have concluded that water molecules adsorb on the catalyst surface to form mono- or multilayers that block active sites and create an extra diffusion layer for toluene to reach the catalyst surface. This hypothesis has also been conﬁrmed by Huang et al. [122,123]. In a recent paper [187], Huang et al. have investigated the effect of water vapor on toluene removal efﬁciency, carbon balance, CO2 selectivity and outlet ozone concentration. A wire-plate DBD ﬁlled with MnOx /Al2 O3 /nickel foam or TiO2 /Al2 O3 /nickel foam is used to perform experiments. The results show that increased humidity lowers the formation of ozone through quenching of energetic electrons. Also, catalytic decomposition of ozone is depressed by the presence of water vapor due to competitive adsorption causing deactivation of the catalyst and suppression of catalytic ozonation. The carbon balance and CO2 selectivity reach maximum values when RH is in the range 25–75%. Van Durme et al. [107] have also studied the effect of humidity on PPC removal of toluene. As for IPC, PPC is less efﬁcient when RH increases. With Pd/Al2 O3 as PPC removal efﬁciencies are >90% and 37% at dry air and air with 74% RH (298 K), respectively. The negative humidity effect is mostly attributed to changing Van der Waals interactions. In a recent study by Huang et al. [168], NTP has been combined with a photocatalyst located downstream. Experimental results indicate that catalytic ozonation plays a vital role in toluene decomposition. The dominant active species in the NTP-driven photocatalyst system are active oxygen species formed from ozone catalytic decomposition. The decomposition pathway of toluene

A.M. Vandenbroucke et al. / Journal of Hazardous Materials 195 (2011) 30–54

has been elucidated in subsequent work [188]. Detected byproducts for IPC removal of toluene with TiO2 /Al2 O3 /nickel foam include benzene, benzaldehyde, formic acid and small amounts of acetic acid and 2-methylamylene. Subrahmanyam et al. [134] modiﬁed a sintered metal ﬁbre ﬁlter, which acts as inner electrode, with MnOx and CoOx . Fig. 13 shows the inﬂuence of this modiﬁcation and energy density on the conversion of 100 ppm toluene. At an energy density of 235 J/L, nearly 100% conversion has been achieved with both MnOx and CoOx /SMF. Whereas SMF only shows 50% CO2 selectivity, MnOx /SMF reaches 80% even at 235 J/L. Interestingly, no polymeric carbon deposits are detected. All the catalytic electrodes maintain the same activity during almost 3 h of toluene decomposition. This proves that the electrodes maintain their stability during VOC destruction. Magureanu et al. [142] have tested MnPO4 , Mn-APO-5 and Mn-SAPO-11 as PPC catalysts in an oven for temperatures up to 673 K. Even at low temperature, a remarkable synergetic effect has been observed while the catalysts alone are not active at that temperature level. The authors expect a further increase in the plasma–catalytic synergy by placing the catalyst in the discharge region, where short-lived species produced in the plasma will most likely contribute to oxidation on the catalyst surface. 3. Critical process parameters Various process parameters determine the initial condition of the feeded gas stream. In the following section these parameters are discussed which are critical for an effective operation of both catalytic and non-catalytic NTP systems. For each parameter, the different inﬂuences on the removal performance of the conﬁguration will be discussed and compared if possible. 3.1. Temperature In most cases, the NTP process removes VOCs more effectively as the process temperature increases. This is ascribed to an increased reaction rate of O and OH radicals with VOCs due to the endothermic behaviour of these reactions [53,127,189–195]. This is, however, only the case for VOCs that are primarily decomposed through radical reactions. When electron impact is thought to be the primary decomposition step (e.g. CCl4 ), no temperature dependence on the removal is observed because the electron density is not really inﬂuenced hereby [196,197]. However, in Ref. [198] CCl4 destruction is greatly improved at high temperature. This can be explained by the fact that the maximum energy density is also signiﬁcantly higher than in [196], which might lead to higher decomposition. The improved removal rate and energy efﬁciency can also be explained by an increase in the reduced electric ﬁeld (E/n) with increasing temperatures. The reduced electric ﬁeld, being the ratio of the electric ﬁeld (E) and the gas density (n), is an important factor that determines the electron energy in the plasma. Since the gas density decreases as the gas temperature increases at constant pressure, NTP systems tend to operate at a higher reduced electric ﬁeld [192,199]. When the catalyst is located downstream, NTP produced ozone can be decomposed by reaction with molecular oxygen in the gas phase: O3 + O2 → O + O2 + O2

(48)

The rate constant of this reaction is accelerated at elevated temperatures (5 times higher at 573 K compared to 373 K). However, the lifetime of the produced oxygen atoms in the gas phase is too short to react with VOCs adsorbed on the catalyst surface. At the same time, reactions at the catalyst surface between adsorbed oxygen atoms and VOCs are also accelerated. The net-result of these two competing effects is the most likely explanation for

47

the different temperature dependencies found in literature: with increasing temperature, VOC decomposition efﬁciency can remain almost constant [142], can increase [149,153,180] or can decrease [127]. 3.2. Initial VOC concentration Generally, the VOC concentration of actual industrial exhaust streams strongly varies. Therefore the effect of VOC concentration on the removal process has been abundantly studied. When the initial concentration rises, each VOC molecule shares fewer electrons and reactive plasma species. Consequently, numerous research papers have pointed out that higher initial VOC concentrations are detrimental for the removal efﬁciency in catalytic and non-catalytic NTP systems. Some papers also indicate that the characteristic energy [200–204] (i.e. the energy density needed to decompose 63% of the initial VOC concentration) and the energy yield [49,199,205] are an increasing function of the initial VOC concentration. For some halogenated carbons, the initial concentration barely seems to affect the decomposition efﬁciency. This is the case for HFC-134a [206], CFC-12 [207], HCFC-22 [208] and TCE [62]. This may be partly attributed to secondary decomposition induced by fragment ions and radicals produced by primary destruction steps [62]. Another plausible explanation may be that for these compounds the primary destruction by reactive plasma species is the rate-determining step, leading to similar decomposition efﬁciencies regardless of the initial concentration [208]. 3.3. Humidity level The effect of humidity is of great interest for practical applications in industry since process gas consists of ambient air that usually contains water vapor at ﬂuctuating concentrations. It appears that the effect of water vapor strongly depends on its concentration as well as on the type of the target VOC and the type of discharge. Water plays an important role in the plasma chemistry since it decomposes into OH and H radicals in a NTP environment as follows: H2 O + e− → OH• + H• + e− 3

+u )

→ N2

H2 O + O( D) →

2 OH•

H2 O + N2 (A 1

+ OH•

(49) + H•

(50) (51)

The oxidation power of OH is generally much stronger than those of other oxidants such as oxygen atoms and peroxyl radicals. The introduction of water vapor can induce changes in the electrical and physical properties of the discharge. The effect of water vapor has been mostly studied with (packed) DBD reactors. For this type of discharge, the presence of water vapor is known to reduce the total charge transferred in a microdischarge which ultimately decreases the volume of the reactive plasma zone [68]. The plasma characteristics of corona discharges are also affected by the presence of water vapor. At higher RH, lower currents are observed for a given voltage [83]. This is attributed to a higher probability of the plasma attachment processes resulting in a reduced OH production [209]. Water also has an adverse effect on VOC removal due to its electronegative characteristic which limits the electron density and quenches activated chemical species [82]. The effect of humidity has been tested for several VOCs. It seems that the addition of water negatively inﬂuences the properties of the discharge irrespective of the VOC chemical structure. However, the enhanced production of OH caused by higher water vapor content competes with the latter effect, depending on the VOC chemical structure [210]. The inﬂuence on the removal process is designated as an enhancement, a suppression or a neutral effect depending on the chemical structure of the target VOC. Table 10

48

A.M. Vandenbroucke et al. / Journal of Hazardous Materials 195 (2011) 30–54

Table 10 Inﬂuence of humidity on VOC removal with NTP.

Table 11 Optimal oxygen content for VOC removal with NTP.

Target VOC

Plasma type

Inﬂuence

References

Target VOC

Optimal O2 content (%)

References

Acetylene Benzene Benzene Benzene Butane Chloroform Dichloromethane Formaldehyde Methane Methanol Propane TCE TCE TCE Tetrachloromethane Tetrachloromethane Toluene Toluene Toluene Toluene Toluene 1,1,1-Trichloroethane p-Xylene

DBD Packed DBD DBD Packed DBD DBD Gliding arc Packed DBD DBD DBD Packed DBD DBD DBD Surface discharge SPCPa Packed DBD Gliding arc Corona Packed DBD Pulsed corona Gliding arc Surface discharge DBD DBD

Suppression Suppression Suppression Suppression Suppression Suppression Suppression Neutral Enhancement Neutral Neutral Suppression Suppression Neutral Suppression Suppression Enhancement Neutral Suppression Neutral Neutral Suppression Enhancement

[166] [66] [69] [283] [213] [284] [205] [194] [247] [205] [193] [213] [285] [286] [216] [284] [169] [205] [244] [199] [140] [287] [192]

Acetaldehyde Benzene Benzene Carbon tetraﬂuoride Dichloromethane 2-Heptanone HCFC-22 HFC-134a Methylbromide TCE Toluene Toluene Trichloromethane p-Xylene

3–5 0.2 3–5 1 1–3 2–3 0.5 0.5 2 2 3–5 2 0.5 5

[215] [73] [75] [288] [243] [200] [208] [206] [289] [290] [75] [102] [217] [192]

a

Surface discharge induced plasma chemical processing.

gives an overview of research results concerning the effect of humidity on the decomposition efﬁciency in various plasma reactors. Some studies have shown that an optimal water vapor content exists for achieving a maximum VOC removal efﬁciency. Interestingly, this optimum is around 20% RH for both TCE [211] and toluene [82,83]. Furthermore, addition of water counteracts the formation of ozone due to consumption of O(1 D) (reaction (51)) which is the most important origin of ozone formation [179]. It has also been shown that water vapor decreases the formation of CO and enhances the selectivity towards CO2 [166,212,213]. In case of a PPC system, catalytic ozonation will play a minor role due to the inhibition of ozone formation by humidity. Secondly, the catalyst surface can be covered with layers of H2 O preventing the adsorption of ozone and VOCs and consequently minimalizing direct catalyst/VOC intermolecular interactions [107,122,186,214]. In this context, the morphology and chemical composition of the catalyst are important factors that inﬂuence the interactions with H2 O. Therefore, it is desirable to choose a catalyst that is less susceptible to H2 O adsorption. Finally, increased humidity can poison catalytic active sites and lower the catalysts activity [122,214].

level because process gas depends on its industrial environment and in a lot of situations it consists of ambient air. Similar effects are observed for IPC systems. Additionally, direct reactions between oxygen radicals and VOC molecules adsorbed on the catalyst surface add to the positive effect of a moderate O2 addition to N2 [220]. However, in [75], several catalysts (TiO2 , ␥-Al2 O3 , zeolites) have been tested at varying oxygen content for the removal of toluene and benzene with a cycled system of removal and adsorption. For all catalysts tested, the removal efﬁciency increased with oxygen content ranging from 0% to 100%. Operation at higher oxygen content is also able to reduce the formation of N2 O and NO2 . As for the inﬂuence of the oxygen content on the performance of PPC conﬁgurations, no studies were found in literature. 3.5. Gas ﬂow rate The gas ﬂow rate applied in laboratory experiments generally ranges from 0.1 L/min to 10 L/min. The effect of decreasing the gas ﬂow rate logically implies an increase in residence time of the VOC in the system. Hence, the collision probability for electron-impact reactions and for reactions between VOCs and plasma generated radicals and metastables is enhanced which increases the decomposition efﬁciency. When the NTP system is combined with a catalyst, the same argumentation can be made. In that case, the increased probability of surface reactions is beneﬁcial for the removal process. In the interest of practical operation, some groups have studied multistage NTP reactors with the aim to increase the residence time without decreasing the gas ﬂow rate [116,221–224].

3.4. Oxygen content 4. Future trends Similar to the presence of water vapor, the oxygen content in the gas stream affects the discharge performance and plays a very important role in the occurring chemical reactions. A small increase in oxygen concentration generally leads to an enhanced generation of reactive oxygen radicals, resulting in a higher removal efﬁciency. However, due to its electronegative character, higher oxygen concentrations tend to trigger electron attachment reactions. Consequently, this limits the electron density and changes the electron energy distribution functions [215,216]. Also, oxygen and oxygen radicals are able to consume reactive species such as excited nitrogen molecules and nitrogen atoms, which are otherwise used for destroying VOCs [217–219]. Collectively, the phenomena described above ensure the existence of an optimal oxygen content for VOC removal with NTP (Table 11). It appears that the optimal oxygen content ranges between 1% and 5%. For practical application in industrial waste gas treatment, it is, however, in most cases difﬁcult to control the oxygen content to this

The extensive literature review regarding NTP and plasma–catalytic decomposition of VOCs demonstrates that there are still challenges that have to be adressed in future work. For plasma-alone systems, incomplete oxidation of VOCs leads to the formation of various intermediate and unwanted byproducts. Although for certain compounds decomposition mechanisms are proposed, there is still need to expand the knowledge on plasma-chemical kinetics. The derived information about e.g. the distribution of byproducts can be very useful to choose an appropriate catalyst or to enhance existing catalytic formulations in order to increase the efﬁciency of the hybrid system. In the case of plasma–catalytic systems, synergistic effects on the overall removal efﬁciency are often observed. The mechanisms that contribute to this synergy are thoroughly investigated and different elucidations are proposed in literature often showing discrepancies between them. Therefore, better understanding of

A.M. Vandenbroucke et al. / Journal of Hazardous Materials 195 (2011) 30–54

which mechanisms have the most important contributions and which chemical species play a dominant role in the decomposition of VOCs is still of great interest. From this point of view, development of well-designed instruments specialized in in situ measurements is crucial. The research of plasma material interactions for VOC treatment is recently looking at the opportunity to regenerate VOC saturated surfaces with the aid of NTP systems [225–227]. This process of alternate adsorption and desorption can convert ﬂue gases with a large ﬂow rate and low VOC concentration into that with a low ﬂow rate and high concentration. Further developments could yield an economical VOC removal process for small- and medium-sized facilities that emit diluted VOC waste gases. Therefore, further progress on this subject is suspected in the nearby future.

References [1] P.S. Monks, C. Granier, S. Fuzzi, A. Stohl, M.L. Williams, H. Akimoto, M. Amann, A. Baklanov, U. Baltensperger, I. Bey, N. Blake, R.S. Blake, K. Carslawn, O.R. Cooper, F. Dentener, D. Fowler, E. Fragkou, G.J. Frost, S. Generoso, P. Ginoux, V. Grewet, A. Guenther, H.C. Hansson, S. Hennew, J. Hjorth, A. Hofzumahaus, H. Huntrieser, I.S.A. Isaksen, M.E. Jenkin, J. Kaiser, M. Kanakidou, Z. Klimont, M. Kulmala, P. Laj, M.G. Lawrence, J.D. Lee, C. Liousse, M. Maione, G. McFiggans, A. Metzger, A. Mieville, N. Moussiopoulos, J.J. Orlando, C.D. O’Dowd, P.I. Palmer, D.D. Parrish, A. Petzold, U. Platt, U. Pöschl, A.S.H. Prevot, C.E. Reeves, S. Reimann, Y. Rudich, K. Sellegri, R. Steinbrecher, D. Simpson, H. ten Brink, J. Theloke, G.R. van der Werf, R. Vautard, V. Vestreng, C. Vlachokostas, R. von Glasow, Atmospheric composition change—global and regional air quality, Atmospheric Environment 43 (2009) 5268–5350. [2] R. Atkinson, Atmospheric chemistry of VOCs and NOx , Atmospheric Environment 34 (2000) 2063–2101. [3] J.R. Odum, T.P.W. Jungkamp, R.J. Grifﬁn, H.J.L. Forstner, R.C. Flagan, J.H. Seinfeld, Aromatics, reformulated gasoline, and atmospheric organic aerosol formation, Environmental Science & Technology 31 (1997) 1890–1897. [4] R.A Larson, E.J. Weber, Reaction Mechanisms in Environmental Organic Chemistry, Lewis Publishers, 2010. [5] J. Chorria, M. Jaroniec, M. Kloske, Ordered siliceous and carbonaceous nanomaterials in environmental protection, Ochrona Srodowiska 29 (2007) 3–12. [6] W.B. Li, H. Gong, Recent progress in the removal of volatile organic compounds by catalytic combustion, Acta Physico-Chimica Sinica 26 (2010) 885–894. [7] A. Kumar, J. Dewulf, H. Van Langenhove, Membrane-based biological waste gas treatment, Chemical Engineering Journal 136 (2008) 82–91. [8] S. Mudliar, B. Giri, K. Padoley, D. Satpute, R. Dixit, P. Bhatt, R. Pandey, A. Juwarkar, A. Vaidya, Bioreactors for treatment of VOCs and odours—a review, Journal of Environmental Management 91 (2010) 1039–1054. [9] K. Demeestere, J. Dewulf, H. Van Langenhove, Heterogeneous photocatalysis as an advanced oxidation process for the abatement of chlorinated, monocyclic aromatic and sulfurous volatile organic compounds in air: state of the art, Critical Reviews in Environmental Science and Technology 37 (2007) 489–538. [10] F.I Khan, A.K. Ghoshal, Removal of volatile organic compounds from polluted air, Journal of Loss Prevention in the Process Industries 13 (2000) 527–545. [11] T. Yamamoto, K. Ramaanathan, P.A. Lawless, D.S. Ensor, J.R. Newsome, N. Plaks, G.H. Ramsey, Control of volatile organic compounds by an AC energized ferroelectric pellet reactor and a pulsed corona reactor, IEEE Transactions on Industry Applications 28 (1992) 528–534. [12] C.M. Nunez, G.H. Ramsey, W.H. Ponder, J.H. Abbott, L.E. Hamel, P.H. Kariher, Corona destruction: an innovative control technology for VOCs and air toxics, Air and Waste 43 (1993) 242–247. [13] H.H. Kim, Nonthermal plasma processing for air-pollution control: a historical review, current issues, and future prospects, Plasma Processes and Polymers 1 (2004) 91–110. [14] J.Y Park, J.G. Jung, J.S. Kim, G.H. Rim, K.S. Kim, Effect of nonthermal plasma reactor for CF4 decomposition, IEEE Transactions on Plasma Science 31 (2003) 1349–1354. [15] T. Yamamoto, B.S. Rajanikanth, M. Okubo, T. Kuroki, M. Nishino, Performance evaluation of nonthermal plasma reactors for NO oxidation in diesel engine exhaust gas treatment, IEEE Transactions on Industry Applications 39 (2003) 1608–1613. [16] H.B. Ma, P. Chen, M.L. Zhang, X.Y. Lin, R. Ruan, Study of SO2 removal using non-thermal plasma induced by dielectric barrier discharge (DBD), Plasma Chemistry and Plasma Processing 22 (2002) 239–254. [17] Y.H. Song, S.J. Kim, K.I. Choi, T. Yamamoto, Effects of adsorption and temperature on a nonthermal plasma process for removing VOCs, Journal of Electrostatics 55 (2002) 189–201. [18] U. Roland, F. Holzer, E.D. Kopinke, Combination of non-thermal plasma and heterogeneous catalysis for oxidation of volatile organic compounds Part 2. Ozone decomposition and deactivation of gamma-Al2 O3 , Applied Catalysis B: Environmental 58 (2005) 217–226.

49

[19] C. Subrahmanyam, Catalytic non-thermal plasma reactor for total oxidation of volatile organic compounds, Indian Journal of Chemistry Section A: Inorganic Bio-Inorganic Physical Theoretical & Analytical Chemistry 48 (2009) 1062–1068. [20] T. Zhu, J. Li, W.J. Liang, Y.Q. Jin, Synergistic effect of catalyst for oxidation removal of toluene, Journal of Hazardous Materials 165 (2009) 1258–1260. [21] A. Mizuno, Industrial applications of atmospheric non-thermal plasma in environmental remediation, Plasma Physics and Controlled Fusion 49 (2007) A1–A15. [22] J.S. Chang, P.A. Lawless, T. Yamamoto, Corona discharge processes, IEEE Transactions on Plasma Science 19 (1991) 1152–1166. [23] A. Fridman, A. Chirokov, A. Gutsol, Non-thermal atmospheric pressure discharges, Journal of Physics D: Applied Physics 38 (2005) R1–R24. [24] U. Kogelschatz, Dielectric-barrier discharges: their history, discharge physics, and industrial applications, Plasma Chemistry and Plasma Processing 23 (2003) 1–46. [25] V. Nehra, A. Kumar, H.K. Dwivedi, Atmospheric non-thermal plasma sources, International Journal of Engineering 2 (2008) 53–68. [26] R. Hackam, H. Akiyama, Air pollution control by electrical discharges, IEEE Transactions on Dielectrics and Electrical Insulation 7 (2000) 654–683. [27] T. Yamamoto, S. Futamura, Nonthermal plasma processing for controlling volatile organic compounds, Combustion Science and Technology 133 (1998) 117–133. [28] B. Eliasson, M. Hirth, U. Kogelschatz, Ozone synthesis from oxygen in dielectric barrier discharges, Journal of Physics D: Applied Physics 20 (1987) 1421–1437. [29] G.J. Pietsch, Peculiarities of dielectric barrier discharges, Contributions to Plasma Physics 41 (2001) 620–628. [30] S. Masuda, S. Hosokawa, X.L. Tu, K. Sakakibara, S. Kitoh, S. Sakai, Destruction of gaseous-pollutants by surface-induced plasma chemical process (Spcs), IEEE Transactions on Industry Applications 29 (1993) 781–786. [31] S. Masuda, S. Hosokawa, X. Tu, Z. Wang, Novel plasma chemical technologies—Ppcp and Spcp for control of gaseous-pollutants and air toxics, Journal of Electrostatics 34 (1995) 415–438. [32] K. Urashima, J.S. Chang, T. Ito, Reduction of NOx from combustion ﬂue gases by superimposed barrier discharge plasma reactors, IEEE Transactions on Industry Applications 33 (1997) 879–886. [33] R. McAdams, Prospects for non-thermal atmospheric plasmas for pollution abatement, Journal of Physics D: Applied Physics 34 (2001) 2810–2821. [34] J. Jarrige, P. Vervisch, Decomposition of three volatile organic compounds by nanosecond pulsed corona discharge: study of by-product formation and inﬂuence of high voltage pulse parameters, Journal of Applied Physics 99 (2006). [35] M. Nifuku, M. Horvath, J. Bodnar, G.Y. Zhang, T. Tanaka, E. Kiss, G. Woynarovich, H. Katoh, A study on the decomposition of volatile organic compounds by pulse corona, Journal of Electrostatics 40–1 (1997) 687–692. [36] E.H.W.M. Smulders, B.E.J.M. van Heesch, S.S.V.B. van Paasen, Pulsed power corona discharges for air pollution control, IEEE Transactions on Plasma Science 26 (1998) 1476–1484. [37] K.F. Shang, Y. Wu, J. Li, G.F. Li, D. Li, N.H. Wang, Reduction of NOx /SO2 by wire-plate type pulsed discharge reactor with pulsed corona radical shower, Plasma Chemistry and Plasma Processing 26 (2006) 443–454. [38] Y. Shi, J. Ruan, X. Wang, W. Ll, T. Tan, Decomposition of mixed malodorants in a wire-plate pulse corona reactor, Environmental Science & Technology 39 (2005) 6786–6791. [39] W.J. Liang, J. Li, Y.Q. Jin, Abatement of toluene from gas streams via ferroelectric packed bed dielectric barrier discharge plasma, Journal of Hazardous Materials 170 (2009) 633–638. [40] A. Ogata, N. Shintani, K. Mizuno, S. Kushiyama, T. Yamamoto, Decomposition of benzene using a nonthermal plasma reactor packed with ferroelectric pellets, IEEE Transactions on Industry Applications 35 (1999) 753–759. [41] U. Roland, F. Holzer, F.D. Kopinke, Improved oxidation of air pollutants in a non-thermal plasma, Catalysis Today 73 (2002) 315–323. [42] S. Okazaki, M. Kogoma, M. Uehara, Y. Kimura, Appearance of stable glowdischarge in air, argon, oxygen and nitrogen at atmospheric-pressure using a 50-Hz source, Journal of Physics D: Applied Physics 26 (1993) 889–892. [43] Y. Akishev, A. Deryugin, V. Karal’nik, I. Kochetov, A. Napartovich, N. Trushkin, Numerical simulation and experimental study of an atmospheric pressure direct-current glow discharge, Plasma Physics Reports 20 (1994) 511–524. [44] D. Trunec, A. Brablec, F. Stastny, Experimental study of atmospheric pressure glow discharge, Contributions to Plasma Physics 38 (1998) 435–445. [45] E.E. Kunhardt, Generation of large-volume, atmospheric-pressure, nonequilibrium plasmas, IEEE Transactions on Plasma Science 28 (2000) 189–200. [46] Y.S Akishev, M.E. Grushin, I.V. Kochetov, A.P. Napartovich, M.V. Pan’kin, N.I. Trushkin, Transition of a multipin negative corona in atmospheric air to a glow discharge, Plasma Physics Reports 26 (2000) 157–163. [47] Y. Akishev, O. Goossens, T. Callebaut, C. Leys, A. Napartovich, N. Trushkin, The inﬂuence of electrode geometry and gas ﬂow on corona-to-glow and glowto-spark threshold currents in air, Journal of Physics D: Applied Physics 34 (2001) 2875–2882. [48] T. Callebaut, I. Kochetov, Y. Akishev, A. Napartovich, C. Leys, Numerical simulation and experimental study of the corona and glow regime of a negative pin-to-plate discharge in ﬂowing ambient air, Plasma Sources Science & Technology 13 (2004) 245–250. [49] R. Vertriest, R. Morent, J. Dewulf, C. Leys, H. Van Langenhove, Multi-pinto-plate atmospheric glow discharge for the removal of volatile organic

50

[50]

[51]

[52]

[53]

[54]

[55]

[56]

[57]

[58] [59] [60]

[61]

[62]

[63]

[64]

[65]

[66]

[67]

[68]

[69]

[70]

[71]

[72]

[73] [74]

[75]

A.M. Vandenbroucke et al. / Journal of Hazardous Materials 195 (2011) 30–54 compounds in waste air, Plasma Sources Science & Technology 12 (2003) 412–416. D.S. Antao, D.A. Staack, A. Fridman, B. Farouk, Atmospheric pressure DC corona discharges: operating regimes and potential applications, Plasma Sources Science & Technology 18 (2009) 035016 (035011 pp.). D. Evans, L.A. Rosocha, G.K. Anderson, J.J. Coogan, M.J. Kushner, Plasma remediation of trichloroethylene in silent discharge plasmas, Journal of Applied Physics 74 (1993) 5378–5386. M.J. Kirkpatrick, W.C. Finney, B.R. Locke, Chlorinated organic compound removal by gas phase pulsed streamer corona electrical discharge with reticulated vitreous carbon electrodes, Plasmas and Polymers 8 (2003) 165–177. M.C. Hsiao, B.T. Merritt, B.M. Penetrante, G.E. Vogtlin, P.H. Wallman, Plasmaassisted decomposition of methanol and trichloroethylene in atmosphericpressure air streams by electrical-discharge processing, Journal of Applied Physics 78 (1995) 3451–3456. L. Prager, H. Langguth, S. Rummel, R. Mehnert, Electron-beam degradation of chlorinated hydrocarbons in air, Radiation Physics and Chemistry 46 (1995) 1137–1142. T. Hakoda, S. Hashimoto, Y. Fujiyama, A. Mizuno, Decomposition mechanism for electron beam irradiation of vaporized trichloroethylene-air mixtures, Journal of Physical Chemistry A 104 (2000) 59–66. T. Hakoda, G. Zhang, S. Hashimoto, Decomposition of chloroethenes in electron beam irradiation, Radiation Physics and Chemistry 54 (1999) 541–546. S.A. Vitale, K. Hadidi, D.R. Cohn, P. Falkos, The effect of a carbon-carbon double bond on electron beam-generated plasma decomposition of trichloroethylene and 1,1,1-trichloroethane, Plasma Chemistry and Plasma Processing 17 (1997) 59–78. E. Sanhueza, I.C. Hisatsune, J. Heicklen, Oxidation of haloethylenes, Chemical Reviews 76 (1976) 801–826. G. Huybrech, L. Meyers, Gas-phase chlorine-photosensitized oxidation of trichloroethylene, Transactions of the Faraday Society 62 (1966) 2191. L. Bertrand, J.A. Franklin, P. Goldﬁng, G. Huybrech, Point of attack of a chlorine atom on trichloroethylene, Journal of Physical Chemistry 72 (1968) 3926. B.M. Penetrante, M.C. Hsiao, J.N. Bardsley, B.T. Merritt, G.E. Vogtlin, A. Kuthi, C.P. Burkhart, J.R. Bayless, Identiﬁcation of mechanisms for decomposition of air pollutants by non-thermal plasma processing, Plasma Sources Science & Technology 6 (1997) 251–259. S. Futamura, T. Yamamoto, Byproduct identiﬁcation and mechanism determination in plasma chemical decomposition of trichloroethylene, IEEE Transactions on Industry Applications 33 (1997) 447–453. K. Urashima, J.S. Chang, Removal of volatile organic compounds from air streams and industrial ﬂue gases by non-thermal plasma technology, IEEE Transactions on Dielectrics and Electrical Insulation 7 (2000) 602–614. J.S. Chang, K. Urashima, T. Ito, Mechanism of non-thermal plasma treatment of volatile organic compounds in dry air, in: D.W. Tedder, F.G. Pohland (Eds.), Emerging Technologies in Hazardous Waste Management, ACS Press, Atlanta, 1994, pp. 203–206. S.B. Han, T. Oda, Decomposition mechanism of trichloroethylene based on byproduct distribution in the hybrid barrier discharge plasma process, Plasma Sources Science & Technology 16 (2007) 413–421. A. Ogata, N. Shintani, K. Yamanouchi, K. Mizuno, S. Kushiyama, T. Yamamoto, Effect of water vapor on benzene decomposition using a nonthermaldischarge plasma reactor, Plasma Chemistry and Plasma Processing 20 (2000) 453–467. H.H. Kim, H. Kobara, A. Ogata, S. Futamura, Comparative assessment of different nonthermal plasma reactors on energy efﬁciency and aerosol formation from the decomposition of gas-phase benzene, IEEE Transactions on Industry Applications 41 (2005) 206–214. Z. Falkenstein, J.J. Coogan, Microdischarge behaviour in the silent discharge of nitrogen-oxygen and water-air mixtures, Journal of Physics D: Applied Physics 30 (1997) 817–825. M.P. Cal, M. Schluep, Destruction of benzene with non-thermal plasma in dielectric barrier discharge reactors, Environmental Progress 20 (2001) 151–156. B.Y. Lee, S.H. Park, S.C. Lee, M. Kang, S.J. Choung, Decomposition of benzene by using a discharge plasma—photocatalyst hybrid system, Catalysis Today 93–5 (2004) 769–776. Z.L. Ye, Y.N. Zhang, P. Li, L.Y. Yang, R.X. Zhang, H.Q. Hou, Feasibility of destruction of gaseous benzene with dielectric barrier discharge, Journal of Hazardous Materials 156 (2008) 356–364. C.Q. Jiang, A.A.H. Mohamed, R.H. Stark, J.H. Yuan, K.H. Schoenbach, Removal of volatile organic compounds in atmospheric pressure air by means of direct current glow discharges, IEEE Transactions on Plasma Science 33 (2005) 1416–1425. K. Satoh, T. Matsuzawa, H. Itoh, Decomposition of benzene in a corona discharge at atmospheric pressure, Thin Solid Films 516 (2008) 4423–4429. N. Goto, H. Kurimoto, S. Kudo, Y. Wanatabe, Decomposition of benzene by barrier discharge in N2 at low concentrations of O2 , IEEJ Transactions on Fundamentals and Materials 123 (2003) 900–906. H.H. Kim, A. Ogata, S. Futamura, Oxygen partial pressure-dependent behavior of various catalysts for the total oxidation of VOCs using cycled system of adsorption and oxygen plasma, Applied Catalysis B: Environmental 79 (2008) 356–367.

[76] M.F. Golde, Reactions of N2(A3-Sigma-U+), International Journal of Chemical Kinetics 20 (1988) 75–92. [77] G.R. Dey, A. Sharma, K.K. Pushpa, T.N. Das, Variable products in dielectricbarrier discharge assisted benzene oxidation, Journal of Hazardous Materials 178 (2010) 693–698. [78] H. Kohno, A.A. Berezin, J.S. Chang, M. Tamura, T. Yamamoto, A. Shibuya, S. Hondo, Destruction of volatile organic compounds used in a semiconductor industry by a capillary tube discharge reactor, IEEE Transactions on Industry Applications 34 (1998) 953–966. [79] W. Mista, R. Kacprzyk, Decomposition of toluene using non-thermal plasma reactor at room temperature, Catalysis Today 137 (2008) 345–349. [80] Z. Machala, E. Marode, M. Morvová, P. Lukáˇc, DC glow discharge in atmospheric air as a source for volatile organic compounds abatement, Plasma Processes and Polymers 2 (2005) 152–161. [81] S. Ognier, S. Cavadias, J. Amouroux, Aromatic VOC removal by formation of microparticles in pure nitrogen discharge, Plasma Processes and Polymers 4 (2007) 528–536. [82] Y.F. Guo, D.Q. Ye, K.F. Chen, Y.F. Tian, Humidity effect on toluene decomposition in a wire-plate dielectric barrier discharge reactor, Plasma Chemistry and Plasma Processing 26 (2006) 237–249. [83] J. Van Durme, J. Dewulf, W. Sysmans, C. Leys, H. Van Langenhove, Abatement and degradation pathways of toluene in indoor air by positive corona discharge, Chemosphere 68 (2007) 1821–1829. [84] M. Schiorlin, E. Marotta, M. Rea, C. Paradisi, Comparison of toluene removal in air at atmospheric conditions by different corona discharges, Environmental Science & Technology 43 (2009) 9386–9392. [85] F. Holzer, U. Roland, F.D. Kopinke, Combination of non-thermal plasma and heterogeneous catalysis for oxidation of volatile organic compounds Part 1. Accessibility of the intra-particle volume, Applied Catalysis B: Environmental 38 (2002) 163–181. [86] H.L. Chen, H.M. Lee, S.H. Chen, M.B. Chang, S.J. Yu, S.N. Li, Removal of volatile organic compounds by single-stage and two-stage plasma catalysis systems: a review of the performance enhancement mechanisms, current status, and suitable applications, Environmental Science & Technology 43 (2009) 2216–2227. [87] X. Chen, J. Rozak, J.C. Lin, S.L. Suib, J. Hayashi, H. Matsumoto, Oxidative decomposition of chlorinated hydrocarbons by glow discharge in PACT (plasma and catalyst integrated technologies) reactors, Applied Catalysis A: General 219 (2001) 25–31. [88] M.B. Chang, H.M. Lee, Abatement of perﬂuorocarbons with combined plasma catalysis in atmospheric-pressure environment, Catalysis Today 89 (2004) 109–115. [89] S.J. Yu, M.B. Chang, Oxidative conversion of PFC via plasma processing with dielectric barrier discharges, Plasma Chemistry and Plasma Processing 21 (2001) 311–327. [90] M.A. Malik, Y. Minamitani, K.H. Schoenbach, Comparison of catalytic activity of aluminum oxide and silica gel for decomposition of volatile organic compounds (VOCs) in a plasmacatalytic reactor, IEEE Transactions on Plasma Science 33 (2005) 50–56. [91] F. Holzer, F.D. Kopinke, U. Roland, Inﬂuence of ferroelectric materials and catalysts on the performance of non-thermal plasma (NTP) for the removal of air pollutants, Plasma Chemistry and Plasma Processing 25 (2005) 595–611. [92] H.L. Chen, H.M. Lee, S.H. Chen, M.B. Chang, Review of packed-bed plasma reactor for ozone generation and air pollution control, Industrial & Engineering Chemistry Research 47 (2008) 2122–2130. [93] J.O. Chae, V. Demidiouk, M. Yeulash, I.C. Choi, T.G. Jung, Experimental study for indoor air control by plasma–catalyst hybrid system, IEEE Transactions on Plasma Science 32 (2004) 493–497. [94] J.G. Wang, C.J. Liu, Y.P. Zhang, K.L. Yu, X.L. Zhu, F. He, Partial oxidation of methane to syngas over glow discharge plasma treated Ni-Fe/Al2 O3 catalyst, Catalysis Today 89 (2004) 183–191. [95] Z.H. Li, S.X. Tian, H.T. Wang, H.B. Tian, Plasma treatment of Ni catalyst via a corona discharge, Journal of Molecular Catalysis A: Chemical 211 (2004) 149–153. [96] X.L. Zhu, P.P. Huo, Y.P. Zhang, C.J. Liu, Characterization of argon glow discharge plasma reduced Pt/Al2 O3 catalyst, Industrial & Engineering Chemistry Research 45 (2006) 8604–8609. [97] C.J. Liu, J.J. Zou, K.L. Yu, D.G. Cheng, Y. Han, J. Zhan, C. Ratanatawanate, B.W.L. Jang, Plasma application for more environmentally friendly catalyst preparation, Pure and Applied Chemistry 78 (2006) 1227–1238. [98] C. Ratanatawanate, M. Macias, B.W.L. Jang, Promotion effect of the nonthermal RF plasma treatment on Ni/Al2 O3 for benzene hydrogenation, Industrial & Engineering Chemistry Research 44 (2005) 9868–9874. [99] Y.R. Zhu, Z.H. Li, Y.B. Zhou, J. Lv, H.T. Wang, Plasma treatment of Ni and Pt catalysts for partial oxidation of methane, Reaction Kinetics and Catalysis Letters 87 (2005) 33–41. [100] Y.F. Guo, D.Q. Ye, K.F. Chen, J.C. He, W.L. Chen, Toluene decomposition using a wire-plate dielectric barrier discharge reactor with manganese oxide catalyst in situ, Journal of Molecular Catalysis A: Chemical 245 (2006) 93–100. [101] Y.P. Zhang, P.S. Ma, X.L. Zhu, C.J. Liu, Y.T. Shen, A novel plasma-treated Pt/NaZSM-5 catalyst for NO reduction by methane, Catalysis Communications 5 (2004) 35–39. [102] Y.F. Guo, D.Q. Ye, K.F. Chen, J.C. He, Toluene removal by a DBD-type plasma combined with metal oxides catalysts supported by nickel foam, Catalysis Today 126 (2007) 328–337.

A.M. Vandenbroucke et al. / Journal of Hazardous Materials 195 (2011) 30–54 [103] A.E. Wallis, J.C. Whitehead, K. Zhang, Plasma-assisted catalysis for the destruction of CFC-12 in atmospheric pressure gas streams using TiO2 , Catalysis Letters 113 (2007) 29–33. [104] A.S. Pribytkov, G.N. Baeva, N.S. Telegina, A.L. Tarasov, A.Y. Stakheev, A.V. Tel’nov, V.N. Golubeva, Effect of electron irradiation on the catalytic properties of supported Pd catalysts, Kinetics and Catalysis 47 (2006) 765–769. [105] A.E. Wallis, J.C. Whitehead, K. Zhang, The removal of dichloromethane from atmospheric pressure nitrogen gas streams using plasma-assisted catalysis, Applied Catalysis B: Environmental 74 (2007) 111–116. [106] A. Rousseau, O. Guaitella, J. Ropcke, L.V. Gatilova, Y.A. Tolmachev, Combination of a pulsed microwave plasma with a catalyst for acetylene oxidation, Applied Physics Letters 85 (2004) 2199–2201. [107] J. Van Durme, J. Dewulf, K. Demeestere, C. Leys, H. Van Langenhove, Postplasma catalytic technology for the removal of toluene from indoor air: effect of humidity, Applied Catalysis B: Environmental 87 (2009) 78–83. [108] B. Lu, X. Zhang, X. Yu, T. Feng, S. Yao, Catalytic oxidation of benzene using DBD corona discharges, Journal of Hazardous Materials 137 (2006) 633–637. [109] H.H. Kim, A. Ogata, S. Futamura, Effect of different catalysts on the decomposition of VOCS using ﬂow-type plasma-driven catalysis, IEEE Transactions on Plasma Science 34 (2006) 984–995. [110] T. Sano, N. Negishi, E. Sakai, S. Matsuzawa, Contributions of photocatalytic/catalytic activities of TiO2 and gamma-Al2 O3 in nonthermal plasma on oxidation of acetaldehyde and CO, Journal of Molecular Catalysis A: Chemical 245 (2006) 235–241. [111] H.B. Huang, D.Q. Ye, M.L. Fu, F. Da Feng, Contribution of UV light to the decomposition of toluene in dielectric barrier discharge plasma/photocatalysis system, Plasma Chemistry and Plasma Processing 27 (2007) 577–588. [112] H.H. Kim, S.M. Oh, A. Ogata, S. Futamura, Decomposition of gas-phase benzene using plasma-driven catalyst (PDC) reactor packed with Ag/TiO2 catalyst, Applied Catalysis B: Environmental 56 (2005) 213–220. [113] M. Kang, B.J. Kim, S.M. Cho, C.H. Chung, B.W. Kim, G.Y. Han, K.J. Yoon, Decomposition of toluene using an atmospheric pressure plasma/TiO2 catalytic system, Journal of Molecular Catalysis A: Chemical 180 (2002) 125–132. [114] M. Kang, Y.R. Ko, M.K. Jeon, S.C. Lee, S.J. Choung, J.Y. Park, S. Klim, S.H. Choi, Characterization of Bi/TiO2 nanometer sized particle synthesized by solvothermal method and CH3 CHO decomposition in a plasma-photocatalytic system, Journal of Photochemistry and Photobiology A: Chemistry 173 (2005) 128–136. [115] C. Subrahmanyam, M. Magureanu, D. Laub, A. Renken, L. Kiwi-Minsker, Nonthermal plasma abatement of trichloroethylene enhanced by photocatalysis, Journal of Physical Chemistry C 111 (2007) 4315–4318. [116] S. Chavadej, K. Saktrakool, P. Rangsunvigit, L.L. Lobban, T. Sreethawong, Oxidation of ethylene by a multistage corona discharge system in the absence and presence of Pt/TiO2 , Chemical Engineering Journal 132 (2007) 345–353. [117] A. Ogata, H.H. Kim, S.M. Oh, S. Futamura, Evidence for direct activation of solid surface by plasma discharge on CFC decomposition, Thin Solid Films 506 (2006) 373–377. [118] R. Morent, J. Dewulf, N. Steenhaut, C. Leys, H. Van Langenhove, Hybrid plasma–catalyst system for the removal of trichloroethylene in air, Journal of Advanced Oxidation Technologies 9 (2006) 53–58. [119] K. Takaki, K. Urashima, J.S. Chang, Ferro-electric pellet shape effect on C2F6 removal by a packed-bed-type nonthermal plasma reactor, IEEE Transactions on Plasma Science 32 (2004) 2175–2183. [120] K. Urashima, K.G. Kostov, J.S. Chang, Y. Okayasu, T. Iwaizumi, K. Yoshimura, T. Kato, Removal of C2F6 from a semiconductor process ﬂue gas by a ferroelectric packed-bed barrier discharge reactor with an adsorber, IEEE Transactions on Industry Applications 37 (2001) 1456–1463. [121] C. Leys, D. Neirynck, R. Morent, E. Temmerman, DC-excited cold atmospheric pressure plasmas, Czechoslovak Journal of Physics 56 (2006) B896–B902. [122] H.B. Huang, D.Q. Ye, Combination of photocatalysis downstream the nonthermal plasma reactor for oxidation of gas-phase toluene, Journal of Hazardous Materials 171 (2009) 535–541. [123] H.B. Huang, D.Q. Ye, X.J. Guan, The simultaneous catalytic removal of VOCs and O-3 in a post-plasma, Catalysis Today 139 (2008) 43–48. [124] N. Blin-Simiand, P. Tardiveau, A. Risacher, F. Jorand, S. Pasquiers, Removal of 2-heptanone by dielectric barrier discharges—the effect of a catalyst support, Plasma Processes and Polymers 2 (2005) 256–262. [125] K. Krawczyk, B. Ulejczyk, H.K. Song, A. Lamenta, B. Paluch, K. SchmidtSzalowski, Plasma–catalytic reactor for decomposition of chlorinated hydrocarbons, Plasma Chemistry and Plasma Processing 29 (2009) 27–41. [126] V. Demidiouk, J.O. Chae, Decomposition of volatile organic compounds in plasma–catalytic system, IEEE Transactions on Plasma Science 33 (2005) 157–161. [127] K. Hayashi, H. Yasui, M. Tanaka, S. Futamura, S. Kurita, K. Aoyagi, Temperature dependence of toluene decomposition behavior in the discharge-catalyst hybrid reactor, IEEE Transactions on Industry Applications 45 (2009) 1553–1558. [128] K. Hensel, S. Katsura, A. Mizuno, DC microdischarges inside porous ceramics, IEEE Transactions on Plasma Science 33 (2005) 574–575. [129] S. Futamura, H. Einaga, H. Kabashima, L.Y. Hwan, Synergistic effect of silent discharge plasma and catalysts on benzene decomposition, Catalysis Today 89 (2004) 89–95. [130] M.J. Kirkpatrick, W.C. Finney, B.R. Locke, Plasma–catalyst interactions in the treatment of volatile organic compounds and NOx with pulsed corona

[131]

[132]

[133]

[134]

[135]

[136]

[137]

[138]

[139]

[140]

[141]

[142]

[143]

[144]

[145]

[146]

[147]

[148]

[149]

[150]

[151]

[152]

[153]

[154]

[155]

[156]

51

discharge and reticulated vitreous carbon Pt/Rh-coated electrodes, Catalysis Today 89 (2004) 117–126. C. Subrahmanyam, A. Renken, L. Kiwi-Minsker, Novel catalytic non-thermal plasma reactor for the abatement of VOCs, Chemical Engineering Journal 134 (2007) 78–83. C. Subrahmanyam, A. Renken, L. Kiwi-Minsker, Novel catalytic dielectric barrier discharge reactor for gas-phase abatement of isopropanol, Plasma Chemistry and Plasma Processing 27 (2007) 13–22. M. Magureanu, N.B. Mandache, V.I. Parvulescu, C. Subrahmanyam, A. Renken, L. Kiwi-Minsker, Improved performance of non-thermal plasma reactor during decomposition of trichloroethylene: optimization of the reactor geometry and introduction of catalytic electrode, Applied Catalysis B: Environmental 74 (2007) 270–277. C. Subrahmanyam, A. Renken, L. Kiwi-Minsker, Catalytic non-thermal plasma reactor for abatement of toluene, Chemical Engineering Journal 160 (2010) 677–682. X. Chen, J. Rozak, J.C. Lin, S.L. Suib, Y. Hayashi, H. Matsumoto, Oxidative decomposition of chlorinated hydrocarbons by glow discharge in PACT (plasma and catalyst integrated technologies) reactors, Applied Catalysis A: General 219 (2001) 25–31. A. Ogata, K. Yamanouchi, K. Mizuno, S. Kushiyama, T. Yamamoto, Oxidation of dilute benzene in an alumina hybrid plasma reactor at atmospheric pressure, Plasma Chemistry and Plasma Processing 19 (1999) 383–394. A. Ogata, K. Yamanouchi, K. Mizuno, S. Kushiyama, T. Yamamoto, Decomposition of benzene using alumina-hybrid and catalyst-hybrid plasma reactors, IEEE Transactions on Industry Applications 35 (1999) 1289–1295. T. Oda, K. Yamaji, Dilute trichloroethylene decomposition in air by using nonthermal plasma—catalyst effect, Journal of Advanced Oxidation Technologies 6 (2003) 93–99. A. Ogata, D. Ito, K. Mizuno, S. Kushiyama, T. Yamamoto, Removal of dilute benzene using a zeolte-hybrid plasma reactor, IEEE Transactions on Industry Applications 37 (2001) 959–964. S.M. Oh, H.H. Kim, A. Ogata, H. Einaga, S. Futamura, D.W. Park, Effect of zeolite in surface discharge plasma on the decomposition of toluene, Catalysis Letters 99 (2005) 101–104. S.M. Oh, H.H. Kim, H. Einaga, A. Ogata, S. Futamura, D.W. Park, Zeolitecombined plasma reactor for decomposition of toluene, Thin Solid Films 506 (2006) 418–422. M. Magureanu, N.B. Mandache, P. Eloy, E.M. Gaigneaux, V.I. Parvulescu, Plasma-assisted catalysis for volatile organic compounds abatement, Applied Catalysis B: Environmental 61 (2005) 12–20. T. Oda, T. Takahashi, S. Kohzuma, Decomposition of dilute trichloroethylene by using nonthermal plasma processing-frequency and catalyst effects, IEEE Transactions on Industry Applications 37 (2001) 965–970. A.E. Wallis, J.C. Whitehead, K. Zhang, The removal of dichloromethane from atmospheric pressure air streams using plasma-assisted catalysis, Applied Catalysis B: Environmental 72 (2007) 282–288. A. Ogata, H. Einaga, H. Kabashima, S. Futamura, S. Kushiyama, H.H. Kim, Effective combination of nonthermal plasma and catalysts for decomposition of benzene in air, Applied Catalysis B: Environmental 46 (2003) 87–95. H. Grossmannova, D. Neirynck, C. Leys, Atmospheric discharge combined with Cu-Mn/Al2 O3 catalyst unit for the removal of toluene, Czechoslovak Journal of Physics 56 (2006) B1156–B1161. A. Vandenbroucke, R. Morent, N. De Geyter, M.T. Nguyen Dinh, J.M. Giraudon, J.F. Lamonier, C. Leys, Plasma–catalytic decomposition of TCE, International Journal of Plasma Environmental Science and Technology 4 (2010) 135–138. P. Da Costa, R. Marques, S. Da Costa, Plasma catalytic oxidation of methane on alumina-supported noble metal catalysts, Applied Catalysis B: Environmental 84 (2008) 214–222. T. Blackbeard, V. Demidyuk, S.L. Hill, J.C. Whitehead, The effect of temperature on the plasma–catalytic destruction of propane and propene: a comparison with thermal catalysis, Plasma Chemistry and Plasma Processing 29 (2009) 411–419. T. Yamamoto, K. Mizuno, I. Tamori, A. Ogata, M. Nifuku, M. Michalska, G. Prieto, Catalysis-assisted plasma technology for carbon tetrachloride destruction, IEEE Transactions on Industry Applications 32 (1996) 100–105. C.L. Chang, H.L. Bai, S.J. Lu, Destruction of styrene in an air stream by packed dielectric barrier discharge reactors, Plasma Chemistry and Plasma Processing 25 (2005) 641–657. M.B. Chang, H.M. Lee, Abatement of perﬂuorocarbons with combined plasma–catalysis in atmospheric pressure environment, Catalysis Today 89 (2004) 109–115. V. Demidyuk, J.C. Whitehead, Inﬂuence of temperature on gas-phase toluene decomposition in plasma–catalytic system, Plasma Chemistry and Plasma Processing 27 (2007) 85–94. V. Demidiouk, S.I. Moon, J.O. Chae, D.Y. Lee, Application of a plasma–catalytic system for decomposition of volatile organic compounds, Journal of the Korean Physical Society 42 (2003) S966–S970. C.L. Chang, T.S. Lin, Elimination of carbon monoxide in the gas streams by dielectric barrier discharge systems with Mn catalyst, Plasma Chemistry and Plasma Processing 25 (2005) 387–401. H.L. Chen, H.M. Lee, L.C. Cheng, M.B. Chang, S.J. Yu, S.N. Li, Inﬂuence of nonthermal plasma reactor type on CF4 and SF6 abatements, IEEE Transactions on Plasma Science 36 (2008) 509–515.

52

A.M. Vandenbroucke et al. / Journal of Hazardous Materials 195 (2011) 30–54

[157] J. Jarrige, P. Vervisch, Plasma-enhanced catalysis of propane and isopropyl alcohol at ambient temperature on a MnO2 -based catalyst, Applied Catalysis B: Environmental 90 (2009) 74–82. [158] X. Fan, T.L. Zhu, M.Y. Wang, X.M. Li, Removal of low concentration BTX in air using a combined plasma catalysis system, Chemosphere 75 (2009) 1301–1306. [159] T. Hakoda, K. Matsumoto, A. Mizuno, K. Hirota, Role of metals loaded on a TiO2 surface in the oxidation of xylene in air using an electron beam irradiation/catalytic process, Applied Catalysis A: General 357 (2009) 244–249. [160] A.S. Besov, A.V. Vorontsov, Acceleration of acetone destruction process under synergistic action of photocatalytic oxidation and barrier discharge, Plasma Chemistry and Plasma Processing 27 (2007) 624–634. [161] T. Oda, K. Yamaji, T. Takahashi, Decomposition of dilute trichloroethylene by nonthermal plasma processing—gas ﬂow rate, catalyst, and ozone effect, IEEE Transactions on Industry Applications 40 (2004) 430–436. [162] T. Oda, T. Takahahshi, K. Yamaji, Nonthermal plasma processing for dilute VOCs decomposition, IEEE Transactions on Industry Applications 38 (2002) 873–878. [163] D.W. Park, S.H. Yoon, G.J. Kim, H. Sekiguchi, The effect of catalyst on the decomposition of dilute benzene using dielectric barrier discharge, Journal of Industrial and Engineering Chemistry 8 (2002) 393–398. [164] R.B. Sun, Z.G. Xi, F.H. Chao, W. Zhang, H.S. Zhang, D.F. Yang, Decomposition of low-concentration gas-phase toluene using plasma-driven photocatalyst reactor, Atmospheric Environment 41 (2007) 6853–6859. [165] O. Guaitella, F. Thevenet, E. Puzenat, C. Guillard, A. Rousseau, C2 H2 oxidation by plasma/TiO2 combination: inﬂuence of the porosity, and photocatalytic mechanisms under plasma exposure, Applied Catalysis B: Environmental 80 (2008) 296–305. [166] F. Thevenet, O. Guaitella, E. Puzenat, C. Guillard, A. Rousseau, Inﬂuence of water vapour on plasma/photocatalytic oxidation efﬁciency of acetylene, Applied Catalysis B: Environmental 84 (2008) 813–820. [167] C.L. Chang, T.S. Lin, Decomposition of toluene and acetone in packed dielectric barrier discharge reactors, Plasma Chemistry and Plasma Processing 25 (2005) 227–243. [168] H.B. Huang, D.Q. Ye, D.Y.C. Leung, Removal of toluene using UV-irradiated and nonthermal plasma-driven photocatalyst system, Journal of Environmental Engineering: ASCE 136 (2010) 1231–1236. [169] M. Kogoma, Y. Miki, K. Tanaka, K. Takahashi, Highly efﬁcient VOC decomposition using a complex system (OH Radical, Ozone-UV, and TiO2 ), Plasma Processes and Polymers 3 (2006) 727–733. [170] T. Oda, T. Takahashi, K. Yamaji, TCE decomposition by the non-thermal plasma process concerning ozone effect, IEEE Transaction on Industrial Applications 40 (2004) 1249–1256. [171] S.B. Han, T. Oda, R. Ono, Improvement of the energy efﬁciency in the decomposition of dilute trichloroethylene by the barrier discharge plasma process, IEEE Transactions on Industry Applications 41 (2005) 1343–1349. [172] M. Magureanu, N.B. Mandache, J.C. Hu, R. Richards, M. Florea, V.I. Parvulescu, Plasma-assisted catalysis total oxidation of trichloroethylene over gold nanoparticles embedded in SBA-15 catalysts, Applied Catalysis B: Environmental 76 (2007) 275–281. [173] A.M. Vandenbroucke, R. Morent, N. De Geyter, C. Leys, Decomposition of trichloroethylene with plasma–catalysis: a review, Journal of Advanced Oxidation Technologies 14 (2011) 165–173. [174] H.H. Kim, Y.H. Lee, A. Ogata, S. Futamura, Plasma-driven catalyst processing packed with photocatalyst for gas-phase benzene decomposition, Catalysis Communications 4 (2003) 347–351. [175] H.H. Kim, A. Ogata, S. Futamura, Y.H. Lee, Decomposition of gas-phase benzene using hybrid systems of a non-thermal plasma and catalysts, Journal of the Korean Physical Society 44 (2004) 1163–1167. [176] H.H. Kim, A. Ogata, S. Futamura, Atmospheric plasma-driven catalysis for the low temperature decomposition of dilute aromatic compounds, Journal of Physics D: Applied Physics 38 (2005) 1292–1300. [177] H.H. Kim, S.M. Oh, A. Ogata, S. Futamura, Decomposition of benzene using Ag/TiO2 packed plasma-driven catalyst reactor: inﬂuence of electrode conﬁguration and Ag-loading amount, Catalysis Letters 96 (2004) 189–194. [178] H.Y. Fan, C. Shi, X.S. Li, D.Z. Zhao, Y. Xu, High-efﬁciency plasma catalytic removal of dilute benzene from air, Journal of Physics D: Applied Physics 42 (2009) 22510, 225105. [179] T. Zhu, J. Li, Y. Jin, Y. Liang, G. Ma, Decomposition of benzene by non-thermal plasma processing: photocatalyst and ozone effect, International Journal of Environmental Science and Technology 5 (2008) 375–384. [180] A.M. Harling, V. Demidyuk, S.J. Fischer, J.C. Whitehead, Plasma–catalysis destruction of aromatics for environmental clean-up: effect of temperature and conﬁguration, Applied Catalysis B: Environmental 82 (2008) 180–189. [181] M.A. Malik, X.Z. Jiang, Catalyst assisted destruction of trichloroethylene and toluene in corona discharges, Journal of Environmental Sciences 12 (2000) 7–11. [182] D.A. Li, D. Yakushiji, S. Kanazawa, T. Ohkubo, Y. Nomoto, Decomposition of toluene by streamer corona discharge with catalyst, Journal of Electrostatics 55 (2002) 311–319. [183] Y.F. Guo, X. Liao, J.H. He, W. Ou, D.Q. Ye, Effect of manganese oxide catalyst on the dielectric barrier discharge decomposition of toluene, Catalysis Today 153 (2010) 176–183. [184] Y.F. Guo, X.B. Liao, D.Q. Ye, Journal of Environmental Science: China 20 (2008) 1429.

[185] X.B. Liao, Y.F. Guo, J.H. He, W.J. Ou, D.Q. Ye, Hydroxyl radicals formation in dielectric barrier discharge during decomposition of toluene, Plasma Chemistry and Plasma Processing 30 (2010) 841–853. [186] J. Van Durme, J. Dewulf, W. Sysmans, C. Leys, H. Van Langenhove, Efﬁcient toluene abatement in indoor air by a plasma catalytic hybrid system, Applied Catalysis B: Environmental 74 (2007) 161–169. [187] H.B. Huang, D.Q. Ye, D.Y.C. Leung, Plasma-driven catalysis process for toluene abatement: effect of water vapor, IEEE Transactions on Plasma Science 39 (2011) 576–580. [188] H. Huang, D.Q. Ye, D.Y.C. Leung, F. Feng, X.J. Guan, Byproducts and pathways of toluene destruction via plasma–catalysis, Journal of Molecular Catalysis A: Chemical 336 (2011) 87–93. [189] B.M. Penetrante, M.C. Hsiao, J.N. Bardsley, B.T. Merritt, G.E. Vogtlin, A. Kuthi, C.P. Burkhart, J.R. Bayless, Decomposition of methylene chloride by electron beam and pulsed corona processing, Physics Letters A 235 (1997) 76–82. [190] V. Demidyuk, S.L. Hill, J.C. Whitehead, Enhancement of the destruction of propane in a low-temperature plasma by the addition of unsaturated hydrocarbons: experiment and modeling, Journal of Physical Chemistry A 112 (2008) 7862–7867. [191] N. Blin-Simiand, S. Pasquiers, F. Jorand, C. Postel, J.R. Vacher, Removal of formaldehyde in nitrogen and in dry air by a DBD: importance of temperature and role of nitrogen metastable states, Journal of Physics D: Applied Physics 42 (2009). [192] H.M. Lee, M.B. Chang, Abatement of gas-phase p-xylene via dielectric barrier discharges, Plasma Chemistry and Plasma Processing 23 (2003) 541–558. [193] O. Aubry, J.M. Cormier, Improvement of the diluted propane efﬁciency treatment using a non-thermal plasma, Plasma Chemistry and Plasma Processing 29 (2009) 13–25. [194] D.G. Storch, M.J. Kushner, Destruction mechanisms for formaldehyde in atmospheric-pressure low-temperature plasmas, Journal of Applied Physics 73 (1993) 51–55. [195] J. Li, S.P. Bai, X.C. Shi, S.L. Han, X.M. Zhu, W.C. Chen, Y.K. Pu, Effects of temperature on benzene oxidation in dielectric barrier discharges, Plasma Chemistry and Plasma Processing 28 (2008) 39–48. [196] B.M. Penetrante, M.C. Hsiao, J.N. Bardsley, B.T. Merritt, G.E. Vogtlin, P.H. Wallman, Electron beam and pulsed corona processing of volatile organic compounds in gas streams, Pure and Applied Chemistry 68 (1996) 1083–1087. [197] A.M. Harling, H.H. Kim, S. Futamura, J.C. Whitehead, Temperature dependence of plasma–catalysis using a nonthermal, atmospheric pressure packed bed; the destruction of benzene and toluene, Journal of Physical Chemistry C 111 (2007) 5090–5095. [198] L.W Huang, K. Nakajyo, T. Hari, S. Ozawa, H. Matsuda, Decomposition of carbon tetrachloride by a pulsed corona reactor incorporated with in situ absorption, Industrial & Engineering Chemistry Research 40 (2001) 5481–5486. [199] C.M. Du, J.H. Yan, B. Cheron, Decomposition of toluene in a gliding arc discharge plasma reactor, Plasma Sources Science & Technology 16 (2007) 791–797. [200] A.S. Chiper, N. Blin-Simiand, M. Heninger, H. Mestdagh, P. Boissel, F. Jorand, J. Lemaire, J. Leprovost, S. Pasquiers, G. Popa, C. Postel, Detailed characterization of 2-heptanone conversion by dielectric barrier discharge in N2 and N2 /O2 mixtures, Journal of Physical Chemistry A 114 (2010) 397–407. [201] N. Blin-Simiand, F. Jorand, L. Magne, S. Pasquiers, C. Postel, J.R. Vacher, Plasma reactivity and plasma-surface interactions during treatment of toluene by a dielectric barrier discharge, Plasma Chemistry and Plasma Processing 28 (2008) 429–466. [202] S. Delagrange, L. Pinard, J.M. Tatibouet, Combination of a non-thermal plasma and a catalyst for toluene removal from air: manganese based oxide catalysts, Applied Catalysis B: Environmental 68 (2006) 92–98. [203] J.H. Byeon, J.H. Park, Y.S. Jo, K.Y. Yoon, J. Hwang, Removal of gaseous toluene and submicron aerosol particles using a dielectric barrier discharge reactor, Journal of Hazardous Materials 175 (2010) 417–422. [204] L.A. Rosocha, R.A. Korzekwa, Advanced oxidation and reduction processes in the gas phase using non-thermal plasmas, Journal of Advanced Oxidation Technologies 4 (1999) 247–264. [205] S. Futamura, M. Sugasawa, Additive effect on energy efﬁciency and byproduct distribution in VOC decomposition with nonthermal plasma, IEEE Transactions on Industry Applications 44 (2008) 40–45. [206] Y.S. Mok, V. Demidyuk, J.C. Whitehead, Decomposition of hydroﬂuorocarbons in a dielectric-packed plasma reactor, Journal of Physical Chemistry A 112 (2008) 6586–6591. [207] H.C. Kang, Decomposition of chloroﬂuorocarbon by non-thermal plasma, Journal of Industrial and Engineering Chemistry 8 (2002) 488–492. [208] J.H. Oh, Y.S. Mok, S.B. Lee, Destruction of HCFC-22 and distribution of byproducts in a nonthermal plasma reactor packed with dielectric pellets, Journal of Korean Physical Society 54 (2009) 1539–1546. [209] L. Fouad, S. Elhazek, Effect of humidity on positive corona discharge in a 3electrode system, Journal of Electrostatics 35 (1995) 21–30. [210] M. Sugasawa, T. Terasawa, S. Futamura, Additive effect of water on the decomposition of VOCs in nonthermal plasma, IEEE Transactions on Industry Applications 46 (2010) 1692–1698. [211] Y. Nakagawa, H. Fujisawa, R. Ono, T. Oda, Dilute trichloroethylene decomposition by using high pressure non-thermal plasma-humidity effects, in: 2010 IEEE Industry Applications Society Annual Meeting, 2010.

A.M. Vandenbroucke et al. / Journal of Hazardous Materials 195 (2011) 30–54 [212] A. Ogata, D. Ito, K. Mizuno, S. Kushiyama, A. Gal, T. Yamamoto, Effect of coexisting components on aromatic decomposition in a packed-bed plasma reactor, Applied Catalysis A: General 236 (2002) 9–15. [213] S. Futamura, A.H. Zhang, T. Yamamoto, The dependence of nonthermal plasma behavior of VOCs on their chemical structures, Journal of Electrostatics 42 (1997) 51–62. [214] H. Einaga, T. Ibusuki, S. Futamura, Performance evaluation of a hybrid system comprising silent discharge plasma and manganese oxide catalysts for benzene decomposition, IEEE Transactions on Industry Applications 37 (2001) 1476–1482. [215] H.M. Lee, M.B. Chang, Gas-phase removal of acetaldehyde via packed-bed dielectric barrier discharge reactor, Plasma Chemistry and Plasma Processing 21 (2001) 329–343. [216] R.G. Tonkyn, S.E. Barlow, T.M. Orlando, Destruction of carbon tetrachloride in a dielectric barrier/packed-bed corona reactor, Journal of Applied Physics 80 (1996) 4877–4886. [217] Y.S. Mok, S.B. Lee, J.H. Oh, K.S. Ra, B.H. Sung, Abatement of trichloromethane by using nonthermal plasma reactors, Plasma Chemistry and Plasma Processing 28 (2008) 663–676. [218] A. Ogata, K. Mizuno, S. Kushiyama, T. Yamamoto, Methane decomposition in a barium titanate packed-bed nonthermal plasma reactor, Plasma Chemistry and Plasma Processing 18 (1998) 363–373. [219] K.J. Pringle, J.C. Whitehead, J.J. Wilman, J.H. Wu, The chemistry of methane remediation by a non-thermal atmospheric pressure plasma, Plasma Chemistry and Plasma Processing 24 (2004) 421–434. [220] H.X. Ding, A.M. Zhu, F.G. Lu, Y. Xu, J. Zhang, X.F. Yang, Low-temperature plasma–catalytic oxidation of formaldehyde in atmospheric pressure gas streams, Journal of Physics D: Applied Physics 39 (2006) 3603–3608. [221] A.M. Harling, D.J. Glover, J.C. Whitehead, K. Zhang, Industrial scale destruction of environmental pollutants using a novel plasma reactor, Industrial & Engineering Chemistry Research 47 (2008) 5856–5860. [222] S. Chavadej, W. Kiatubolpaiboon, P. Rangsunvigit, T. Sreethawong, A combined multistage corona discharge and catalytic system for gaseous benzene removal, Journal of Molecular Catalysis A: Chemical 263 (2007) 128–136. [223] Y. Shi, J.J. Ruan, X. Wang, W. Li, T.E. Tan, Evaluation of multiple corona reactor modes and the application in odor removal, Plasma Chemistry and Plasma Processing 26 (2006) 187–196. [224] A.M. Harling, D.J. Glover, J.C. Whitehead, K. Zhang, Novel method for enhancing the destruction of environmental pollutants by the combination of multiple plasma discharges, Environmental Science & Technology 42 (2008) 4546–4550. [225] T. Kuroki, T. Fujioka, R. Kawabata, M. Okubo, T. Yamamoto, Regeneration of honeycomb zeolite by nonthermal plasma desorption of toluene, IEEE Transactions on Industry Applications 45 (2009) 10–15. [226] H.H. Kim, S. Tsubota, M. Date, A. Ogata, S. Futamura, Catalyst regeneration and activity enhancement of Au/TiO2 by atmospheric pressure nonthermal plasma, Applied Catalysis A: General 329 (2007) 93–98. [227] T. Kuroki, T. Fujioka, M. Okubo, T. Yamamoto, Toluene concentration using honeycomb nonthermal plasma desorption, Thin Solid Films 515 (2007) 4272–4277. [228] J. Van Durme, J. Dewulf, C. Leys, H. Van Langenhove, Combining non-thermal plasma with heterogeneous catalysis in waste gas treatment: a review, Applied Catalysis B: Environmental 78 (2008) 324–333. [229] M. Magureanu, N.B. Mandache, V.I. Parvulescu, Chlorinated organic compounds decomposition in a dielectric barrier discharge, Plasma Chemistry and Plasma Processing 27 (2007) 679–690. [230] Z.L. Ye, Y. Shen, R.X. Zhang, H.Q. Hou, Destruction of benzene in an air stream by the outer combined plasma photolysis method, Journal of Physics D: Applied Physics 41 (2008). [231] A. Ogata, K. Miyamae, K. Mizuno, S. Kushiyama, M. Tezuka, Decomposition of benzene in air in a plasma reactor: effect of reactor type and operating conditions, Plasma Chemistry and Plasma Processing 22 (2002) 537–552. [232] C.H. Lin, H. Bai, Energy effectiveness of nonthermal plasma reactors for toluene vapor destruction, Journal of Environmental Engineering 127 (2001) 648–654. [233] A.D. Koutsospyros, S.M. Yin, C. Christodoulatos, K. Becker, Plasmochemical degradation of volatile organic compounds (VOC) in a capillary discharge plasma reactor, IEEE Transactions on Plasma Science 33 (2005) 42–49. [234] K. Faungnawakij, N. Sano, D. Yamamoto, T. Kanki, T. Charinpanitkul, W. Tanthapanichakoon, Removal of acetaldehyde in air using a wetted-wall corona discharge reactor, Chemical Engineering Journal 103 (2004) 115–122. [235] T. Oda, A. Kumada, K. Tanaka, T. Takahashi, S. Masuda, Low-temperature atmospheric-pressure discharge plasma processing for volatile organiccompounds, Journal of Electrostatics 35 (1995) 93–101. [236] T. Oda, R. Yamashita, I. Haga, T. Takahashi, S. Masuda, Decomposition of gaseous organic contaminants by surface discharge induced plasma chemical processing SPCP, IEEE Transactions on Industry Applications 32 (1996) 118–124. [237] R. Rudolph, K.P. Francke, H. Miessner, Concentration dependence of VOC decomposition by dielectric barrier discharges, Plasma Chemistry and Plasma Processing 22 (2002) 401–412. [238] H.C. Wang, D. Li, Y. Wu, J. Li, G.F. Li, Removal of four kinds of volatile organic compounds mixture in air using silent discharge reactor driven by bipolar pulsed power, Journal of Electrostatics 67 (2009) 547–553.

53

[239] M.N. Lyulyukin, A.S. Besov, A.V. Vorontsov, The inﬂuence of corona electrodes thickness on the efﬁciency of plasmachemical oxidation of acetone, Plasma Chemistry and Plasma Processing 31 (2011) 23–39. [240] M.G. Sobacchi, A.V. Saveliev, A.A. Fridman, A.F. Gutsol, L.A. Kennedy, Experimental assessment of pulsed corona discharge for treatment of VOC emissions, Plasma Chemistry and Plasma Processing 23 (2003) 347–370. [241] J. Hoard, T.J. Wallington, R.L. Bretz, A. Malkin, R. Dorai, M.J. Kushner, Importance of O(P-3) atoms and OH radicals in hydrocarbon oxidation during the nonthermal plasma treatment of diesel exhaust inferred using relative-rate methods, International Journal of Chemical Kinetics 35 (2003) 231–238. [242] A. Rousseau, O. Guaitella, L. Gatilova, M. Hannemann, J. Ropcke, Measurement of the C2 H2 destruction kinetics by infrared laser absorption spectroscopy in a pulsed low pressure dc discharge, Journal of Physics D: Applied Physics 40 (2007) 2018–2025. [243] C. Fitzsimons, F. Ismail, J.C. Whitehead, J.J. Wilman, The chemistry of dichloromethane destruction in atmospheric-pressure gas streams by a dielectric packed-bed plasma reactor, Journal of Physical Chemistry A 104 (2000) 6032–6038. [244] R.A. Korzekwa, M.G. Grothaus, R.K. Hutcherson, R.A. Roush, R. Brown, Destruction of hazardous air pollutants using a fast rise time pulsed corona reactor, Review of Scientiﬁc Instruments 69 (1998) 1886–1892. [245] E. Marotta, G. Scorrano, C. Paradisi, Ionic reactions of chlorinated volatile organic compounds in air plasma at atmospheric pressure, Plasma Processes and Polymers 2 (2005) 209–217. [246] M. Sugasawa, G. Annadurai, S. Futamura, Reaction behavior of toluenedichloromethane mixture in nonthermal plasma, IEEE Transactions on Industry Applications 45 (2009) 1499–1505. [247] A. Mfopara, M.J. Kirkpatrick, E. Odic, Dilute methane treatment by atmospheric pressure dielectric barrier discharge: effects of water vapor, Plasma Chemistry and Plasma Processing 29 (2009) 91–102. [248] M. Derakhshesh, J. Abedi, H. Hassanzadeh, Mechanism of methanol decomposition by non-thermal plasma, Journal of Electrostatics 68 (2010) 424–428. [249] R. Taranto, D. Frochot, P. Pichat, Combining cold plasma and TiO2 photocatalysis to purify gaseous efﬂuents: a preliminary study using methanol-contaminated air, Industrial & Engineering Chemistry Research 46 (2007) 7611–7614. [250] T. Sato, M. Kambe, H. Nishiyama, Analysis of a methanol decomposition process by a nonthermal plasma ﬂow, JSME International Journal Series B: Fluids and Thermal Engineering 48 (2005) 432–439. [251] S.L. Hill, J.C. Whitehead, K. Zhang, Plasma processing of propane at hyperatmospheric pressure: experiment and modelling, Plasma Processes and Polymers 4 (2007) 710–718. [252] S.L. Hill, H.H. Kim, S. Futamura, J.C. Whitehead, The destruction of atmospheric pressure propane and propene using a surface discharge plasma reactor, Journal of Physical Chemistry A 112 (2008) 3953–3958. [253] N. Moreau, S. Pasquiers, N. Blin-Simiand, L. Magne, F. Jorand, C. Postel, J.R. Vacher, Propane dissociation in a non-thermal high-pressure nitrogen plasma, Journal of Physics D: Applied Physics 43 (2010), -. [254] Y.S. Mok, C.M. Nam, M.H. Cho, I.S. Nam, Decomposition of volatile organic compounds and nitric oxide by nonthermal plasma discharge processes, IEEE Transactions on Plasma Science 30 (2002) 408–416. [255] I. Orlandini, U. Riedel, Oxidation of propene and the formation of methyl nitrate in non-thermal plasma discharges, Catalysis Today 89 (2004) 83–88. [256] G.K. Anderson, H. Snyder, J. Coogan, Oxidation of styrene in a silent discharge plasma, Plasma Chemistry and Plasma Processing 19 (1999) 131–151. [257] A. Ostapczuk, A.G. Chmielewski, V. Honkonen, J. Ruuskanen, J. Tarhanen, B. Svarfvar, Preliminary test in decomposition of styrene by electron beam treatment, Radiation Physics and Chemistry 56 (1999) 369–371. [258] B.M. Penetrante, M.C. Hsiao, J.N. Bardsley, B.T. Merritt, G.E. Vogtlin, P.H. Wallman, A. Kuthi, C.P. Burkhart, J.R. Bayless, Electron-beam and pulsed corona processing of carbon-tetrachloride in atmospheric-pressure gas streams, Physics Letters A 209 (1995) 69–77. [259] A. Lamenta, S. Jodzis, K. Krawczyk, K. Schmidt-Szalowski, Carbon tetrachloride decomposition in spark discharge plasma, Polish Journal of Chemistry 83 (2009) 169–174. [260] S.J. Rubio, M.C. Quintero, A. Rodero, Application of microwave air plasma in the destruction of trichloroethylene and carbon tetrachloride at atmospheric pressure, Journal of Hazardous Materials 186 (2011) 820–826. [261] H. Kohno, M. Tamura, S. Honda, A. Shibuya, T. Yamamoto, A.A. Berezin, J.S. Chang, Generation of aerosol particles in the process of xylene and TCE decomposition from air streams by a ferroelectric packed-bed barrier discharge reactor, Journal of Aerosol Science 28 (1997) S413–S414. [262] T. Zhu, J. Li, Y.Q. Jin, Y.H. Liang, G.D. Ma, Gaseous phase benzene decomposition by non-thermal plasma coupled with nano titania catalyst, International Journal of Environmental Science and Technology 6 (2009) 141–148. [263] J.Y. Ban, Y.H. Son, M. Kang, S.J. Choung, Highly concentrated toluene decomposition on the dielectric barrier discharge (DBD) plasma-photocatalytic hybrid system with Mn-Ti-incorporated mesoporous silicate photocatalyst (Mn-TiMPS), Applied Surface Science 253 (2006) 535–542. [264] M. Magureanu, N.B. Mandache, E. Gaigneaux, C. Paun, V.I. Parvulescu, Toluene oxidation in a plasma–catalytic system, Journal of Applied Physics 99 (2006). [265] A.M. Harling, D.J. Glover, J.C. Whitehead, K. Zhang, The role of ozone in the plasma–catalytic destruction of environmental pollutants, Applied Catalysis B: Environmental 90 (2009) 157–161.

54

A.M. Vandenbroucke et al. / Journal of Hazardous Materials 195 (2011) 30–54

[266] V. Demidiouk, S.I. Moon, J.O. Chae, Toluene and butyl acetate removal from air by plasma–catalytic system, Catalysis Communications 4 (2003) 51–56. [267] A. Mizuno, Y. Kisanuki, M. Noguchi, S. Katsura, S.H. Lee, U.K. Hong, S.Y. Shin, J.H. Kang, Indoor air cleaning using a pulsed discharge plasma, IEEE Transactions on Industry Applications 35 (1999) 1284–1288. [268] T. Ohshima, T. Kondo, N. Kitajima, M. Sato, Adsorption and plasma decomposition of gaseous acetaldehyde on ﬁbrous activated carbon, IEEE Transactions on Industry Applications 46 (2010) 23–28. [269] A. Rousseau, O. Guaitella, L. Gatilova, F. Thevenet, C. Guillard, J. Ropcke, G.D. Stancu, Photocatalyst activation in a pulsed low pressure discharge, Applied Physics Letters 87 (2005). [270] A.M. Harling, A.E. Wallis, J.C. Whitehead, The effect of temperature on the removal of DCM using non-thermal, atmospheric-pressure plasma-assisted catalysis, Plasma Processes and Polymers 4 (2007) 463–470. [271] A. Ogata, K. Saito, H.H. Kim, M. Sugasawa, H. Aritani, H. Einaga, Performance of an ozone decomposition catalyst in hybrid plasma reactors for volatile organic compound removal, Plasma Chemistry and Plasma Processing 30 (2010) 33–42. [272] H.X. Ding, A.M. Zhu, X.F. Yang, C.H. Li, Y. Xu, Removal of formaldehyde from gas streams via packed-bed dielectric barrier discharge plasmas, Journal of Physics D: Applied Physics 38 (2005) 4160–4167. [273] R. Marques, S. Da Costa, P. Da Costa, Plasma-assisted catalytic oxidation of methane—on the inﬂuence of plasma energy deposition and feed composition, Applied Catalysis B: Environmental 82 (2008) 50–57. [274] J. Kim, B. Han, Y. Kim, J.H. Lee, C.R. Park, J.C. Kim, J.C. Kim, K.J. Kim, Removal of VOCs by hybrid electron beam reactor with catalyst bed, Radiation Physics and Chemistry 71 (2004) 429–432. [275] K.J. Kim, J.C. Kim, J.Y. Kim, Y. Sunwoo, Development of hybrid technology using E-beam and catalyst for aromatic VOCs control, Radiation Physics and Chemistry 73 (2005) 85–90. [276] T. Hakoda, A. Shimada, A. Kimura, M. Taguchi, Y. Sugo, K. Arak, E.B. Dally, K. Hirota, An electron-beam irradiation/catalytic oxidation system for puriﬁcation of aromatic hydrocarbons/air mixture under practical gasﬂow condition, Industrial & Engineering Chemistry Research 49 (2010) 5517–5522. [277] T. Hakoda, K. Matsumoto, A. Shimada, T. Narita, T. Kojima, K. Hirota, Application of ozone decomposition catalysts to electron-beam irradiated xylene/air mixtures for enhancing carbon dioxide production, Radiation Physics and Chemistry 77 (2008) 585–590.

[278] T. Hakoda, K. Matsumoto, A. Mizuno, T. Kojima, K. Hirota, Catalytic oxidation of xylene in air using TiO2 under electron beam irradiation, Plasma Chemistry and Plasma Processing 28 (2008) 25–37. [279] T. Hakoda, K. Matsumoto, A. Mizuno, T. Narita, T. Kojima, K. Hirota, Oxidation process of xylene in air using Ag/TiO2 under electron beam irradiation, IEEE Transactions on Industry Applications 44 (2008) 1950–1956. [280] T. Hakoda, K. Matsumoto, A. Mizuno, K. Hirota, Oxidation of xylene and its irradiation byproducts using an electron-beam irradiating gamma-Al2 O3 bed, Journal of Physics D: Applied Physics 41 (2008). [281] K.P. Francke, H. Miessner, R. Rudolph, Cleaning of air streams from organic pollutants by plasma–catalytic oxidation, Plasma Chemistry and Plasma Processing 20 (2000) 393–403. [282] T. Kuroki, K. Hirai, R. Kawabata, M. Okubo, T. Yamamoto, Decomposition of adsorbed xylene on adsorbents using nonthermal plasma with gas circulation, IEEE Transactions on Industry Applications 46 (2010) 672–679. [283] S. Futamura, A.H. Zhang, H. Einaga, H. Kabashima, Involvement of catalyst materials in nonthermal plasma chemical processing of hazardous air pollutants, Catalysis Today 72 (2002) 259–265. [284] K. Krawczyk, B. Ulejczyk, Inﬂuence of water vapor on CCl4 and CHCl3 conversion in gliding discharge, Plasma Chemistry and Plasma Processing 24 (2004) 155–167. [285] S. Futamura, H. Einaga, A.H. Zhang, Comparison of reactor performance in the nonthermal plasma chemical processing of hazardous air pollutants, IEEE Transactions on Industry Applications 37 (2001) 978–985. [286] R. Yamashita, T. Takahashi, T. Oda, Humidity effect on non-thermal plasma processing for VOCs decomposition, in: IAS’96—Conference Record of the 1996 IEEE Industry Applications Conference, Thirty-First IAS Annual Meeting, vol. 1–4, 1996, pp. 1826–1829. [287] S. Agnihotri, M.P. Cal, J. Prien, Destruction of 1,1,1-trichloroethane using dielectric barrier discharge nonthermal plasma, Jorunal of Environmental Engineering 130 (2004) 349–355. [288] Y. Kim, K.T. Kim, M.S. Cha, Y.H. Song, S.J. Kim, CF4 decompositions using streamer- and glow-mode in dielectric barrier discharges, IEEE Transactions on Plasma Science 33 (2005) 1041–1046. [289] S. Futamura, A.H. Zhang, T. Yamamoto, Mechanisms for formation of inorganic byproducts in plasma chemical processing of hazardous air pollutants, IEEE Transactions on Industry Applications 35 (1999) 760–766. [290] Z. Falkenstein, Effects of the O2 concentration on the removal efﬁciency of volatile organic compounds with dielectric barrier discharges in Ar and N2 , Journal of Applied Physics 85 (1999) 525–529.

Journal of Hazardous Materials 195 (2011) 55–61

Contents lists available at ScienceDirect

Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat

Emulsion stabilization using ionic liquid [BMIM]+ [NTf2 ]− and performance evaluation on the extraction of chromium Rahul Kumar Goyal, N.S. Jayakumar, M.A. Hashim ∗ Department of Chemical Engineering, University of Malaya, Malaysia

a r t i c l e

i n f o

Article history: Received 7 December 2010 Received in revised form 23 February 2011 Accepted 9 March 2011 Available online 15 March 2011 Keywords: Emulsion liquid membrane [BMIM]+ [NTf2 ]− TOMAC Chromium Removal

a b s t r a c t This study focuses on the role of a hydrophobic ionic liquid 1-butyl-3-methylimidazolium bis(triﬂuoromethylsulfonyl)imide, [BMIM]+ [NTf2 ]− in the preparation of emulsion liquid membrane (ELM) phase containing kerosene as solvent, Span 80 as surfactant, NaOH as internal phase and TOMAC (tri-n-octylmethylammonium chloride) a second ionic liquid as carrier. The ﬁrst time used [BMIM]+ [NTf2 ]− in ELM was found to play the role of a stabilizer. The emulsion prepared using [BMIM]+ [NTf2 ]− has a long period of stability of about 7 h (at 3% (w/w) of [BMIM]+ [NTf2 ]− ) which otherwise has a brief stability up to only 7 min. The stability of the emulsion increases with the increase in concentration of [BMIM]+ [NTf2 ]− up to 3% (w/w). Nevertheless, with further increase in concentration of [BMIM]+ [NTf2 ]− , a reduction in the stability occurs. The extraction experiments were carried out after holding the ELM for 2 h after the preparation and a removal efﬁciency of approximately 80% was obtained for Cr. The destabilization of the emulsion was studied by observing the change in the interface height. An empirical correlation for the stability of the emulsion has been proposed. © 2011 Elsevier B.V. All rights reserved.

1. Introduction Liquid membrane technology is well known for its wide application in extraction processes to separate hydrocarbons [1–3], heavy metals [4–6], amin o acids [7,8] and biological compounds [9,10]. The potential advantages of liquid membrane techniques are low capital and operating costs, low energy and extractant consumption, high concentration factors and high ﬂuxes. This technology has an edge over solvent extraction because it requires less energy and operates in a single stage for extraction and stripping. The main types of liquid membrane systems include emulsion liquid membrane, supported liquid membrane and bulk liquid membrane. However, the liquid membrane techniques have not been adopted for large scale industrial processes primarily due to problems in maintaining its stability. Emulsion liquid membrane (ELM) was invented by Li [11] to separate hydrocarbons and this technique has been utilized for many other applications such as metal extraction [5,6], wastewater treatment [6] and bio-medical separation [9]. Stability of emulsion is a major concern in the effective use of ELM either in laboratory scale or industrial scale. The resistance to rupture of liquid membrane at high shear stress deﬁnes the stability of the emulsion liquid membrane. Repeated coalescence of

∗ Corresponding author. Fax: +60 3 79675319. E-mail address: [email protected] (M.A. Hashim). 0304-3894/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.jhazmat.2011.03.024

the internal droplets on the interface, creaming due to density difference, Ostwald Ripening and ﬂocculation cause the instability of the emulsion. Several techniques to overcome the stability problem have been proposed, and these include the use of aliphatic solvent instead of aromatic solvent [12], the increment of the carbon chain length of the aliphatic solvent [13], the increment of the surfactant concentration [14], the increment of membrane viscosity [13,15], the use of co-surfactants [16], non-Newtonian conversion of the membrane phase [17], the use of Janus particles as stabilizers in emulsion polymerization [18] and the use of functionalized silica particles for high internal phase emulsion [19]. All of the remedies have their own tradeoffs and compromises with the overall extraction efﬁciency. Room temperature ionic liquids (RTILs) possess unique and exceptional properties such as negligible vapor pressure, inﬂammability, thermal stability even at high temperatures, highly polar yet non-coordinating solvent and application based adjustable miscibility/immiscibility in chemical processes [20–25]. These properties have made them potentially useful in a wide range of applications in industries as well as in research. Ionic liquids possess a very negligible vapor pressure that has enabled them to be used as a “green solvent” in synthesis [23,24,26–28], separation and puriﬁcation [29–34], and electrochemical applications [35]. RTILs being stable and in the liquid form at room temperature, are made of organic cation and organic/inorganic anion. The physical and chemical properties of RTILs can be altered by changing the cation or anion or both to facilitate a particular task, hence they are

56

R.K. Goyal et al. / Journal of Hazardous Materials 195 (2011) 55–61

(a)

(b) C8H17 C8H17

+ N C8H17

[Cl-]

CH3

Table 1 Physical and thermodynamic properties of [BMIM]+ [NTf2 ]− [36].

N N C4H13 O O NS S CF3 CF3 O O +

[NTf2]-

Fig. 1. Structures of TOMAC and [BMIM]+ [NTf2 ]− .

sometimes referred as “task speciﬁc” ionic liquids [24]. However, very few studies have been reported on the application of ionic liquids in emulsion liquid membrane. Hence, an effort to incorporate ionic liquids with emulsion liquid membrane has been made by investigating the stability and % removal efﬁciency of emulsion liquid membrane in the presence of ionic liquids. Chromium metal was selected to investigate the removal efﬁciency of emulsion liquid membrane. The present study focuses on enhancing the stability of ELM, identiﬁcation of the role of a hydrophobic ionic liquid [BMIM]+ [NTf2 ]− in it and extraction efﬁciency of the ELM. Ionic liquid [BMIM]+ [NTf2 ]− was chosen over other ionic liquids due to its hydrophobicity, minimum toxicity, relatively less viscosity and density. 2. Materials and methods 2.1. Chemicals Ionic liquids [BMIM]+ [NTf2 ]− and TOMAC, with structural formulae illustrated in Fig. 1, were directly obtained from Merck (Germany) and used without any further puriﬁcation while kerosene of boiling point ranged from 180◦ C to 280 ◦ C was received from ACROS (USA). Span 80 (Sorbitan oleate or Sorbitan (Z)mono-9-octadecenoate) a non-ionic surfactant with a ratio of 4:3 of hydrophilic to lipophilic (HLB), was purchased from Merck (Malaysia). Sodium hydroxide pellets, potassium dichromate and hydrochloric acid were procured from R&M Chemicals (UK). The solution of sodium hydroxide of desired normality was prepared by dissolving appropriate weight of pellets in de-ionized water. Similarly, Cr solution of 500 mg/L was prepared by mixing suitable amount of potassium dichromate in de-ionized water. The prepared Cr solution was diluted with de-ionized water according to the required concentration.

Property

Temperature (◦ C)

Value

Density (g/ml)

25

1 .43

Viscosity (cP)

25

52

Surface tension (dyne/cm) (Water equilibrateda )

25

36 .8

Thermal decomposition temperature (◦ C) (Water equilibrateda ) Water content (mg/l) (Water equilibrateda ) Melting point (◦ C) (Driedb )

394 25

3280 4

Water equilibrated denotes that [BMIM]+ [NTf2 ]− was kept in contact with water. Dried stands for water equilibrated [BMIM]+ [NTf2 ]− that was dried at 70 ◦ C for 4 h on a vacuum line. a

b

bilizer) were added to the mixture. The mixture was homogenized for up to 5 min by the homogenizer at 8400 rpm. NaOH was added drop-wise into the mixture, keeping the whole mixture homogenized for the next 5 min. The ratio of internal to organic phase (I/O) was kept at 1:3 for all the experiments. The surfactant concentration (Span 80) (wherever applicable) was kept 3% (w/w) which is an optimized concentration in order to avoid swelling and to provide sufﬁcient stability. Photographs of the beaker containing the emulsion were taken at regular intervals to analyze its stability. The photographs were analyzed by AUTOCAD to determine the phase separation rate of the emulsion. 2.3.2. Extraction of chromium The prepared emulsion was poured into another 250 mL beaker containing the Cr solution of 100 mg/L. The ratio of emulsion to feed phase (E/F) was kept at 1:2 for all the extraction experiments. The pH of the feed phase was maintained below 1.5 to establish a pH difference between the internal and external phases, hence maintaining a driving force for Cr to diffuse through the membrane. The whole mixture was gently stirred by a mechanical stirrer, and an agitation speed of 300 rpm was found to be the best to generate ﬁne globules of emulsion with lowest possible breakage. Samples were taken at a regular interval using disposable syringes and the syringes were kept left undisturbed for some time until the emulsion and the feed phase were separated. The feed phase was then taken out, ﬁltered and analyzed using ICPspectrophotometer.

2.2. Analytical instruments 3. Results and discussion An ICP-spectrophotometer (Perkin Elmer, model: Optima 7000 DV) was used for the measurement of the Cr concentration. The emulsion was prepared using a high speed homogenizer (IKA, model: T25 digtal Ultra Turrax) and the dispersion of the emulsion in the feed phase was carried out by a stirrer (IKA, model: RW11 Lab Egg). pH values were measured using a CyberScan 510 pH meter while photographs were taken using a digital camera (NIKON, model: DSLR D3000). Surface tension was measured by a tensiometer (Fisher Scientiﬁc, model: Tensiomat 21® ) using a Pt/Ir Du Noüy ring.

Physico-chemical properties of this ionic liquid are as shown in Table 1. 3.1. Identiﬁcation of the role of [BMIM]+ [NTf2 ]− in emulsion without TOMAC

2.3. Procedure

As the ﬁrst stage of this study, the role of ionic liquid [BMIM]+ [NTf2 ]− is conjectured to behave as one of the following: either as a carrier, surfactant, solvent or stabilizer. In order to substantiate its role, the following experiments were conducted and discussed below.

2.3.1. Preparation of emulsion and stability analysis The emulsion was prepared in a 100 mL unbafﬂed beaker by mixing organic solvent and an appropriate amount of non ionic surfactant Span 80. Subsequently, the carrier and ionic liquid (sta-

3.1.1. Consideration of [BMIM]+ [NTf2 ]− as a carrier In order to identify the role of [BMIM]+ [NTf2 ]− as a carrier, emulsion was prepared by taking kerosene as solvent, Span 80 as surfactant, NaOH (0.1 N) as internal phase and varying amount

R.K. Goyal et al. / Journal of Hazardous Materials 195 (2011) 55–61

57

Fig. 2. % removal of Cr with time as a function of [BMIM]+ [NTf2 ]− concentration (% w/w) of Cr without any TOMAC. Span 80 = 3% (w/w); I/O = 1:3. Emulsion to feed phase ratio was taken at 1:2.

of [BMIM]+ [NTf2 ]− . I/O ratio was maintained at 1/3 and Span 80 concentration was kept at the value of 3% (w/w). The initial Cr concentration in the feed phase was 100 mg/L and the pH of the feed was maintained below 1.5. The effect of ionic liquid on the percentage removal of Cr in an emulsion liquid membrane is, as shown in Fig. 2. Generally, the role of a carrier in emulsion liquid membrane extraction is to enhance the ﬁnal removal and to increase the rate of extraction. In the absence of TOMAC (carrier), Cr extraction is facilitated by type-I mechanism if [BMIM]+ [NTf2 ]− cannot play the role of a carrier. If the ionic liquid was presumed to act as a carrier in the emulsion, then the ﬁnal % removal of Cr and the rate of % removal of Cr should have increased with the increase in the ionic liquid concentration. On the contrary, an insigniﬁcant decrease in the ﬁnal removal and a signiﬁcant decrease in the rate of removal can be seen upon increasing the concentration of [BMIM]+ [NTf2 ]− from 0 to 2% (w/w), as shown in Fig. 2. Moreover, both the parameters keep on decreasing with the increase in the concentration of ionic liquid up to the value of 4% (w/w). This indicates that the [BMIM]+ [NTf2 ]− is not involved in making complex and transporting the metal from the feed phase to the internal phase. Hence, it’s a type-I facilitation where no carrier is present. The decrease in the % removal of chromium with an increase in concentration of [BMIM]+ [NTf2 ]− can be explained as the increased mass transfer resistance caused by [BMIM]+ [NTf2 ]− during the time of extraction and stripping. It was believed that the diffusion of Cr was hindered by the big size of [BMIM]+ [NTf2 ]− . Electrostatic and Van der waal’s attraction also slowed down the transport of Cr. A sudden decrease in the % extraction of chromium was observed at 6% (w/w) of [BMIM]+ [NTf2 ]− after 50 min. This discrepancy can be explained by the aggregated sedimenting tendency of [BMIM]+ [NTf2 ]− due to its high density after a long time and at higher concentration of the ionic liquid. From these observations and facts, it can be concluded that [BMIM]+ [NTf2 ]− cannot act as a carrier for this operation.

3.1.2. Consideration of [BMIM]+ [NTf2 ]− as a surfactant The role of a surfactant in ELM is to minimize the interfacial energy (interfacial tension) between the organic and the aqueous phase. There is no literature available regarding the use of [BMIM]+ [NTf2 ]− as a surfactant, so is for their HLB number. [BMIM]+ has the properties that can make this ionic liquid to act as a surfactant. Interfacial tension of kerosene and NaOH interface and CMC (Critical Micelle Concentration) of [BMIM]+ [NTf2 ]− were experimentally determined. Emulsion preparation with kerosene as solvent, NaOH as internal reagent and [BMIM]+ [NTf2 ]− as surfactant (assumed) was also tried out to check the feasibility of [BMIM]+ [NTf2 ]− acting as a surfactant. [BMIM]+ [NTf2 ]− is a hydrophobic ionic liquid. It has a density more than kerosene and NaOH. Several combinations of the concentration of organic phase (kerosene); aqueous phase (NaOH) and [BMIM]+ [NTf2 ]− were tested. No emulsion was yielded even if the mixture was homogenized at 15,000 rpm. The concentration of [BMIM]+ [NTf2 ]− was also varied from in the range of 0.40–7% (w/w) for all of the above combinations of phases but no emulsion was observed. The failure of micelle formation of [BMIM]+ [NTf2 ]− in kerosene can be explained by two possible reasons [37]. The ﬁrst is the small hydrocarbon tail (butyl) attached to the cationic group of [BMIM]+ [NTf2 ]− that does not interact well enough with the kerosene hydrocarbon chain to yield the micelles of the ionic liquids. The other reason that may be attributed is the big size of [NTf2 ]− anion which is hard to ﬁt the micelle surface region (Stern Layer). Therefore, it can be concluded that [BMIM]+ , the cationic part of [BMIM]+ [NTf2 ]− does not behave as a surfactant for the above mentioned solvent and the internal phase. 3.1.3. Consideration of [BMIM]+ [NTf2 ]− as a solvent Ionic liquids have been proved to be the solvents of future, primarily based on their unique properties over organic solvents. In order to verify the feasibility of ionic liquid [BMIM]+ [NTf2 ]− as a

58

R.K. Goyal et al. / Journal of Hazardous Materials 195 (2011) 55–61

Fig. 5. The effect of the concentration of [BMIM]+ [NTf2 ]− on the stability time of ELM. Span 80 = 3% (w/w); TOMAC = 0.29% (w/w), I/O = 1/3. Fig. 3. Emulsion prepared with 3% (w/w) Span 80, TOMAC = 0.29% (w/w); [BMIM]+ [NTf2 ]− = 0% (w/w) (after 7 min); scale bar = 1 cm.

solvent for emulsion liquid membrane the emulsion was prepared using [BMIM]+ [NTf2 ]− as a solvent, Span 80 as a surfactant and NaOH as the receiving phase while keeping the proportions of each component the same as it was prepared with kerosene as a solvent in the previous sections. The concentrations and volumes of [BMIM]+ [NTf2 ]− , Span 80 and NaOH and homogenizing speed and time were varied in order to get a stabilized membrane but the stability lasted only for 5 min with the best composition. The density difference between [BMIM]+ [NTf2 ]− and NaOH solution is higher than the difference between kerosene and NaOH therefore, proclivity of sedimenting of the former emulsion is much higher than the latter. Hence, emulsion formed with [BMIM]+ [NTf2 ]− as solvent lasted only for a short duration. The adsorbed amount of Span 80 on the surface of [BMIM]+ [NTf2 ]− was found to be very small and hence interfacial tension was not too reduced to make ﬁne internal droplets of NaOH. The formation of small droplets of Span 80 took place on the surface of ionic liquid upon increasing the Span 80 concentration above 0.3% (w/w). Hence, another reason for the reduced stability may be explained as the insufﬁcient reduction in the interfacial tension of the solvent by Span 80. From the above discussion, it could be concluded that [BMIM]+ [NTf2 ]− cannot be used as a solvent for the above mentioned surfactant and the internal phase to make a stable emulsion liquid membrane.

3.1.4. [BMIM]+ [NTf2 ]− as a stabilizer when TOMAC is used as a carrier TOMAC is a very good phase transfer catalyst which is relatively less expensive, easily available and less toxic. Hence, TOMAC was selected as a carrier to study the effect of ionic liquid ([BMIM]+ [NTf2 ]− ) on the stability of emulsion and subsequent removal efﬁciency of the same. The problem with TOMAC when it is used as a carrier in the emulsion liquid membrane with no concentration of [BMIM]+ [NTf2 ]− , stability lasted only for 7 min, as shown in Fig. 3 which is not sufﬁcient for the extraction to take place and for subsequent demulsiﬁcation. From Fig. 4(a), it can be observed clearly that the emulsion was stable for up to 7 h when 3% (w/w) of the ionic liquid [BMIM]+ [NTf2 ]− is added. On the other hand, Fig. 4(b) depicts the separated organic and aqueous phases after 5 h when there is no concentration of [BMIM]+ [NTf2 ]− present in the emulsion. In fact, the phase separation started only after 7 min without [BMIM]+ [NTf2 ]− in the emulsion. The stability time with the varying concentration of [BMIM]+ [NTf2 ]− is, as shown in Fig. 5. The stability time increases with an increase in the concentration of [BMIM]+ [NTf2 ]− for up to 3% (w/w) of [BMIM]+ [NTf2 ]− . After 3% (w/w) of [BMIM]+ [NTf2 ]− onwards, the stability time decreased which consolidated the fact that the [BMIM]+ [NTf2 ]− helped to stabilize the emulsion liquid membrane when it was present up to a certain maximum concentration. If the emulsion liquid membrane contained more than this concentration then emulsion sedimentation took place due to the higher density of ionic liquid. Each experiment was

Fig. 4. (a) Emulsion prepared with [BMIM]+ [NTf2 ]− = 3% (w/w); TOMAC = 0.29% (w/w) (after 7 h), (b) Emulsion prepared with [BMIM]+ [NTf2 ]− = 0% (w/w); TOMAC = 0.29% (w/w) (after 5 h); scale bar = 1 cm.

R.K. Goyal et al. / Journal of Hazardous Materials 195 (2011) 55–61

59

Fig. 6. % removal of Cr with time as a function of [BMIM]+ [NTf2 ]− concentration (% w/w) with TOMAC. TOMAC = 0.29% (w/w); Span 80 = 3% (w/w); I/O = 1:3. E/F =1:2.

conducted twice and the results for the stability time were reproducible with minor difference between two corresponding values. The increased stability of the emulsion liquid membrane by addition of [BMIM]+ [NTf2 ]− may be explained by Coulombic interactions of the charges on the ions of ionic liquids [BMIM]+ [NTf2 ]− and TOMAC. The other interactions present in the emulsion are between other chemical complex groups such as Span 80 and TOMAC; TOMAC and NaOH; [BMIM]+ [NTf2 ]− and NaOH. These strong interactions help to avoid the coalescence of the internal droplets but they also cause the hindrance to Cr–TOMAC complex diffusion through the membrane. Apart from strong interactions between ions, there is a possibility of hydrogen bonding present between [BMIM]+ [NTf2 ]− and [OH]− group of NaOH. The hydrogen bonding may cause a strong protection surrounding the internal droplets to avoid coalescence. [BMIM]+ [NTf2 ]− is capable of developing a polymeric structure with large cavities [38] when it is used for different kinds of reactions. These polymeric structures of ionic liquid may also help to understand the cause for the enhanced stability. The polymeric structure of ionic liquid [BMIM]+ [NTf2 ]− may behave like polymeric surfactant of the A–B, A–B–A or (BA)n graft type to generate a repulsive barrier to prevent the collapse of the emulsion liquid membrane.

3.1.5. The removal efﬁciency of the emulsion liquid membrane stabilized by ionic liquid [BMIM]+ [NTf2 ]− with TOMAC The prepared emulsion was kept for 2 h to verify its stability then it was poured into an unbafﬂed beaker containing Cr feed phase. The emulsion was prepared with varied concentration of [BMIM]+ [NTf2 ]− . The samples were taken at regular intervals. The effect of ionic liquid [BMIM]+ [NTf2 ]− on the removal efﬁcient of the emulsion liquid membrane having TOMAC as a carrier is, as shown in Fig. 6. TOMAC concentration was kept at a constant value of 0.29% (w/w) for all the experiments. On the contrary, the percentage removal of the emulsion liquid membrane prepared with TOMAC and [BMIM]+ [NTf2 ]− decreases due to the hindrance caused by both of the compounds. From Fig. 6, the time taken for 70% of the removal of chromium is only 5 min.

3.1.6. Effect of [BMIM]+ [NTf2 ]− concentration on phase separation rate of the stabilized emulsion The emulsion was prepared by taking kerosene as solvent, Span 80 as surfactant, NaOH (0.1 N) as internal phase, TOMAC as carrier and varying amount of [BMIM]+ [NTf2 ]− . I/O ratio was maintained at 1/3 and Span 80 concentration was kept 3% (w/w). TOMAC concentration was kept at a constant value of 0.29% (w/w). The stabilized membranes started to separate into organic and aqueous phases after their maximum time of stability which is dependent on the concentration of [BMIM]+ [NTf2 ]− . Creaming and coalescence are the main causes for emulsion sedimentation for the current composition of emulsion. The sedimentation of the emulsion due to Ostwald ripening is insigniﬁcant, since aqueous NaOH and kerosene are almost insoluble in each other. The stabilized membrane was held for the next 3 h after it started to destabilize to analyze the stability with respect to the concentration of [BMIM]+ [NTf2 ]− . The calculation of the phase separation was done by noting the height of the interface from the bottom of the beaker at a regular interval. The normalized height of the emulsion is a ratio of the height of the interface from the ground level and the total height of the emulsion. Therefore, it’s a dimensionless quantity. The stabilized membrane stability time and their phase separation with respect to time are shown in Fig. 7. It can be observed from Fig. 7 that an increment in the concentration of [BMIM]+ [NTf2 ]− to 3% (w/w) increases the ﬁnal (after 3 h) interface height of the destabilized emulsion. It implies that the sedimentation rate decreases upon increasing the concentration of [BMIM]+ [NTf2 ]− up to 3% (w/w). The decreased sedimentation rate may be explained by the effective electrostatic interactions between the two ionic liquids, [BMIM]+ [NTf2 ]− and TOMAC, over the density of ionic liquids and NaOH. Fig. 7 illustrates that an increment in the concentration of [BMIM]+ [NTf2 ]− above 3% (w/w) decreases the interface height of the destabilized emulsion. It means that the sedimentation rate of the destabilized emulsion increases upon increasing the concentration of [BMIM]+ [NTf2 ]− above 3% (w/w). The increased sedimentation could be understood by the dominance of the density of [BMIM]+ [NTf2 ]− and NaOH over electrostatic interactions between ionic liquids [BMIM]+ [NTf2 ]− and TOMAC. However, the complete phase separation of the stabilized membrane into its original phases took place only after 2–3 days.

60

R.K. Goyal et al. / Journal of Hazardous Materials 195 (2011) 55–61

Fig. 7. The effect of ionic liquid concentration (% w/w)) [BMIM]+ [NTf2 ]− on the phase separation of the stabilized membrane. TOMAC = 0.29% (w/w); Span 80 = 3% (w/w); I/O = 1:3.

The experimental data of stability time and initial rate of sedimentation of the emulsion prepared with TOMAC = 0.29% (w/w); Span 80 = 3% (w/w); I/O = 1:3 has been given in Table 2. The stability time of the emulsion as a function of [BMIM]+ [NTf2 ]− concentration (% w/w) (x1 ) and initial rate of sedimentation (x2 ) is correlated to be as ts, pred = ax1b + x2c + 7

(1)

where a, b and c are parameters. The term 7 min in the above correlation takes care of stabilized time when concentration of [BMIM]+ [NTf2 ]− is equal to zero in emulsion liquid membrane containing TOMAC as carrier, NaOH as stripping phase and kerosene as solvent. The parameters are estimated using data in Table 1, by nonlinear parameter estimation scheme with the help of MATLAB 7.0.4 software and the predicted stabilized time relationship is given as ts, pred = 1446.6x10.54549 x20.38057 + 7

Table 3 Comparison between experimental stability time and predicted and stability time. Concentration of [BMIM]+ [NTf2 ]−

ts,exp (min)

ts,pred (min)

% deviation

0.6 1.5 2.0 3.0 4.0

220 330 400 425 465

221.4 335.3 365.9 430.5 457.7

0.6 1.6 8.5 1.3 1.5

(2)

The stability time of the emulsion is dependent more on the concentration of [BMIM]+ [NTf2 ]− than the initial rate of sedimentation of the emulsion, as observed from Eq. (2). The comparison between the predicted values and experimental values has been reported in Table 3. The low values of deviations as summarized in Table 3 imply the accuracy of the correlation. The predicted and experimental stabilized times are in good agreement within ±9% deviation as shown in Fig. 8. However, the

Fig. 8. Experimental stability time versus predicted stability time.

Table 2 Experimental data on stabilized time with ionic liquid concentration [BMIM]+ [NTf2 ]− and initial rate of sedimentation height with TOMAC = 0.29% (w/w); Span 80 = 3% (w/w); I/O = 1:3.

correlation is only applicable for lower range of the ionic liquid concentrations. It does not hold the accuracy for the higher concentrations of ionic liquid [BMIM]+ [NTf2 ]− .

Concentration of [BMIM]+ [NTf2 ]− (% w/w)

Initial rate of sedimentation time (min−1 )

Experimental stability (min)

4. Conclusion

0.6 1.5 2.0 3.0 4.0

0.0150 0.0120 0.0100 0.0086 0.0067

220 330 400 425 465

The present work focuses on the stability aspects of emulsion liquid membrane and in this context, the experimental investigation identiﬁes the use of hydrophobic ionic liquid 1-butyl-3-methylimidazolium bis(triﬂuoromethylsulfonyl)imide, [BMIM]+ [NTf2 ]− as a stabilizer with the preparation of the emul-

R.K. Goyal et al. / Journal of Hazardous Materials 195 (2011) 55–61

sion liquid membrane containing TOMAC as a carrier. The enhanced stability of the emulsion liquid membrane caused by the addition of [BMIM]+ [NTf2 ]− could be explained by the strong interactions such as coulombic, dipolar and ionic interactions among the ionic liquids and NaOH. It was observed that the stability of the emulsion liquid membrane could be enhanced for a duration of up to 7 h. Experiments had proved that the stability of the emulsion liquid membrane can be enhanced for a duration of up to 7 h. 80% removal of chromium could be achieved even after keeping the emulsion for 2 h before the extraction experiments were carried out. The sedimentation rate of the stabilized membrane for the next 3 h after its maximum stability time was found to be decreasing with the increase in concentration of [BMIM]+ [NTf2 ]− up to 3% (w/w). It starts to increase with further increase in the concentration of [BMIM]+ [NTf2 ]− . An empirical correlation relating stability time of emulsion as a function of [BMIM]+ [NTf2 ]− concentration (% w/w) and initial rate of sedimentation of the emulsion is proposed and the predicted stability times are in good agreement with the experimental stability times. Ultimately, this paper reﬂects the potential use of “task speciﬁc” ionic liquids as a stabilizer in the ﬁeld of emulsion liquid membrane. List of symbols a, b, c constant E/F emulsion to feed phase ratio I/O internal to organic phase ratio ts, exp experimental stability time of the emulsion (min) ts, pred predicted stability time of the emulsion (min) x1 concentration of [BMIM]+ [NTf2 ]− (% w/w) initial rate of sedimentation of the emulsion (min−1 ) x2 Acknowledgement The authors acknowledge funding from University of Malaya, Malaysia. References [1] C.M. Das, G. Rungta, S.D. Arya, S. De, Removal of dyes and their mixtures from aqueous solution using liquid emulsion membrane, J. Hazard. Mater. 159 (2008) 365–371. [2] C.C. Lin, R.L. Long, Removal of nitric acid by emulsion liquid membrane: experimental results and model prediction, J. Membr. Sci. 134 (1997) 33–45. [3] P. Venkateswaran, K. Palanivelu, Recovery of phenol from aqueous solution by supported liquid membrane using vegetable oils as liquid membrane, J. Hazard. Mater. 131 (2006) 146–152. [4] S. Chowta, P.K. Mohapatra, B.S. Tomar, K.M. Michael, A. Dakshinamoorthy, V.K. Manchanda, Recovery of americium(III) from low acid solutions using an emulsion liquid membrane containing PC-88A as the carrier extractant, Desalination Water Treat. 12 (2009) 62–67. [5] R.A. Kumbasar, I. Sahin, Separation and concentration of cobalt from ammoniacal solutions containing cobalt and nickel by emulsion liquid membranes using 5,7-dibromo-8-hydroxyquinoline (DBHQ), J. Membr. Sci. 325 (2008) 712– 718. [6] H.R. Mortaheb, H. Kosuge, B. Mokhtarani, M.H. Amini, H.R. Banihashemi, Study on removal of cadmium from wastewater by emulsion liquid membrane, J. Hazard. Mater. 165 (2009) 630–636. [7] H. Itoh, M.P. Thien, T.A. Hatton, D.I.C. Wang, Water transport mechanism in liquid emulsion membrane process for the separation of amino acids, J. Membr. Sci. 51 (1990) 309–322. [8] M. Matsumoto, T. Ohtake, M. Hirata, T. Hano, Extraction rates of amino acids by an emulsion liquid membrane with tri-n-octylmethylammonium chloride, J. Chem. Technol. Biotechnol. 73 (1998) 237–242.

61

[9] T. Kaghazchi, A. Kargari, R. Yegani, A. Zare, Emulsion liquid membrane pertraction of L-lysine from dilute aqueous solutions by D2EHPA mobile carrier, Desalination 190 (2006) 161–171. [10] J.-Q. Shen, W.-P. Yin, Y.-X. Zhao, L.-J. Yu, Extraction of alanine using emulsion liquid membranes featuring a cationic carrier, J. Membr. Sci. 120 (1996) 45–53. [11] N.N. Li, Separating Hydrocarbons with Liquid Membranes, U.S. Patent 3,410,794 (1968). [12] F. Nakashio, M. Goto, M. Matsumoto, J. Irie, K. Kondo, Role of surfactants in the behavior of emulsion liquid membranes—development of new surfactants, J. Membr. Sci. 38 (1988) 249–260. [13] T. Kinugasa, K. Watanabe, H. Takeuchi, Effect of organic solvents on stability of liquid surfactant membranes, J. Chem. Eng. Japan 22 (1989) 593–597. [14] E.C. Hsu, N.N. Li, Membrane recovery in liquid membrane separation processes, Sep. Sci. Technol. 20 (2–3) (1985) 115–130. [15] R.E. Terry, N.N. Li, W.S. Ho, Extraction of phenolic compounds and organic acids by liquid membranes, J. Membr. Sci. 10 (1982) 305–323. [16] Y.S. Mok, K.H. Lee, W.K. Lee, Control of swelling in liquid emulsion membranes employed for lactic acid separation, J. Chem. Tech. Biotechnol. 65 (1996) 309–316. [17] A.H.P. Skelland, X. Meng, Non-Newtonian conversion solves problems of stability, permeability, and swelling in emulsion liquid membranes, J. Membr. Sci. 158 (1999) 1–15. [18] H.E. Muller, A. Walther, M. Hoffmann, Emulsion polymerization using janus particles as stabilizers, Angew. Chem. Int. Ed. 47 (2008) 711–714. [19] A. Bismarck, V.O. Ikem, A. Menner, High internal phase emulsions stabilized solely by functionalized silica particles, Angew. Chem. Int. Ed. 47 (2008) 8277–8279. [20] J.D. Holbrey, K.R. Seddon, Ionic liquids, Clean Prod. Processes 1 (1999) 223–237. [21] M.J. Earle, K.R. Seddon, Ionic liquids. Green solvents for the future, Pure Appl. Chem. 72 (7) (2000) 1391–1398. [22] K. Mikami, Green Reaction Media in Organic Synthesis, ﬁrst ed., Blackwell, UK, 2005. [23] T. Welton, Room-temperature ionic liquids. Solvents for synthesis and catalysis, Chem. Rev. 99 (1999) 2071–2083. [24] P. Wasserscheid, T. Welton, Ionic Liquids in Synthesis, Second ed., WILEY-VCH, Weinheim, 2008. [25] M. Freemantle, New horizons for ionic liquids, Chem. Eng. News 79 (2001) 21–25. [26] C.J. Adam, M.J. Earle, G. Robert, K.R. Seddon, Friedel–crafts reactions in room temperature ionic liquids, Chem. Commun. 19 (1998) 2097–2098. [27] P.J. Dyson, D.J. Ellis, D.C. Parker, T. Welton, Arene hydrogenation in a roomtemperature ionic liquid using a ruthenium cluster catalyst, Chem. Commun. 1 (1999) 25–26. [28] C.E. Song, E.J. Roh, Practical method to recycle a chiral (salen) Mn epoxidation catalyst by using an ionic liquid, Chem. Commun. 10 (2000) 837–838. [29] A.E. Visser, R.P. Swatloski, W.M. Reichert, R. Mayton, S. Sheff, A. Wierzbicki, J.H. Davis Jr., R.D. Rogers, Task-speciﬁc ionic liquids for the extraction of metal ions from aqueous solutions, Chem. Commun. 1 (2001) 135–136. [30] A.P. de los Rios, F.J. Hernandez-Fernandez, L.J. Lozano, S.S. Sanchez, J.I. Moreno, C. Godlinez, Removal of metal ions from aqueous solutions by extraction with ionic liquids, J. Chem. Eng. Data 55 (2010) 605–608. [31] A.E. Visser, R.P. Swatloski, W.M. Reichert, S.T. Grifﬁn, R.D. Rogers, Traditional extractants in nontraditional solvents: group 1 and 2 extraction by crown ethers in room-temperature ionic liquids, Ind. Eng. Chem. Res. 39 (10) (2000) 3596–3604. [32] S. Dai, Y.H. Ju, C.E. Barnes, Solvent extraction of strontium nitrate by a crown ether using room temperature ionic liquids, J. Chem. Soc. Dalton Trans. (1999) 1201–1202. [33] L.C. Branco, J.G. Crespo, C.A.M. Afonso, High selective transport of organic compounds by using supported liquid membranes based on ionic liquids, Angew. Chem. Int. Ed. 41 (2002) 2771–2773. [34] A.G. Fadeev, M.M. Meagher, Opportunities for ionic liquids in recovery of biofuels, Chem. Commun. 3 (2001) 295–296. [35] I.W. Sun, C.L. Hussey, Electrochemistry of niobium chloride and oxide chloride complexes in the basic aluminum chloride-1-methyl-3-ethylimidazolium chloride room-temperature ionic liquid, Inorg. Chem. 28 (1989) 2731–2737. [36] J.G. Huddleston, A.E. Visser, W.M. Reichert, H.D. Willauer, G.A. Broker, R.D. Rogers, Characterization and comparison of hydrophilic and hydrophobic room temperature ionic liquids incorporating the imidazolium cation, Green Chem. 3 (2001) 156–164. [37] M. Blesic, M.H. Marques, N.V. Plechkova, K.R. Seddon, L.P.N. Rebelo, A. Lopes, Self-aggregation of ionic liquids: micelle formation in aqueous solution, Green Chem. 9 (2007) 481–490. [38] H. Weingrtner, Understanding ionic liquids at the molecular level: facts, problems, and controversies, Angew. Chem. Int. Ed. 47 (2008) 654–670.

Journal of Hazardous Materials 195 (2011) 62–67

Contents lists available at ScienceDirect

Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat

The removal of some rare earth elements from their aqueous solutions on by-pass cement dust (BCD) O.I.M. Ali a , H.H. Osman a , S.A. Sayed a,b , M.E.H. Shalabi c,∗ a b c

Faculty of Science, Helwan University, Ain Helwan, Cairo, Egypt Ha’el University, Saudi Arabia1 CMRDI, Tabbin, Cairo, Egypt

a r t i c l e

i n f o

Article history: Received 20 October 2010 Received in revised form 30 July 2011 Accepted 4 August 2011 Available online 25 August 2011 Keywords: BCD Rare earth elements Uptake Sorption

a b s t r a c t The sorption behavior of yttrium (Y3+ ), neodymium (Nd3+ ), gadolinium (Gd3+ ), samarium (Sm3+ ) and lutetium (Lu3+ ) from their aqueous solutions by by-pass cement dust (BCD) has been investigated using a batch technique. The sorption on BCD was studied as a function of pH, shaking time, initial concentration, mass of BCD and temperature. It was found that the sorption capacity of BCD had the order of Lu3+ > Sm3+ > Y3+ > Gd3+ ≈ Nd3+ following Freundlich isotherm at the determined optimum conditions. The results also demonstrated that the sorption data ﬁt well the pseudo-second-order kinetic model. Thermodynamic parameters such as H◦ , S◦ and G◦ indicated that the sorption of the investigated REEs on BCD was endothermic, favored at high temperature and spontaneous in nature. © 2011 Elsevier B.V. All rights reserved.

1. Introduction Recently, more and more rare earth elements (REEs) enter the environment through various pathways as a result of rapid increase of the utilization of REE resources and their applications in many ﬁelds of modern industry and daily life [1–3]. Rare earth elements are used as petrogenetic tracers in internal geodynamic studies of the earth [4]. Moreover, millions of tons of fertilizers containing REEs are also used worldwide for increasing agricultural productivity [5]. The increased demand for REEs means increased public exposure to the REEs both from various commercial products and from production wastes/efﬂuents. In regions with high levels of REEs, elevated levels of REEs are found in human [6] and other organisms [7]. REEs are entering the human body due to exposure to various industrial processes can affect metabolic processes. Trivalent ions, especially La(III) and Gd(III) can interfere with calcium channels in human and animal cells. They can also alter or even inhibit the action of various enzymes and when they found in neurons can regulate synaptic transmission, as well as block some receptors (for example, glutamate receptors) [8]. Numerous techniques are available for the separation and recovery of REEs such as chemical precipitation, ion exchange, solvent extraction and adsorption [9–12]. Adsorption represents the most

∗ Corresponding author. Tel.: +20 223591108. E-mail address: [email protected] (M.E.H. Shalabi). 1 On leave. 0304-3894/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.jhazmat.2011.08.014

efﬁcient and widely applied technique. During the last decade, research efforts have been directed towards using low cost alternative techniques or development of low cost sorbents applicable for the removal and separation of undesirable metal ions from an aqueous phase [13–15]. A variety of low cost sorbents such as ﬂy ash [16], rice husk [17], peat [18], peat moss [19], red mud [20], teawaste [21], olive stones [22], date bits [23] have been tried. Recently, research efforts have been directed towards the use of wastes as sorbent materials in an attempt to minimize the processing costs and to protect the environment and public health [24]. By-pass cement dust (BCD) is the by-product of the manufacture of Portland cement. It is a ﬁne grained material that is generated during the calcination process in the kiln. CaO is the major constitutes of BCD composition. Other constituents include SiO2 , Al2 O3 , Fe2 O3 , K2 O, Na2 O, Cl, etc. Most of cement companies generate high quantities of BCD every year [25]. Potential effectiveness of BCD as a low cost sorbent for some pollutants has been demonstrated. For instance, BCD was used for removal; some heavy metals from textile industrial efﬂuents [26], Cu(II), Ni(II), Zn(II), Fe(III), Co(III), U(VI) and Th(IV) from aqueous solutions [27–29], Cr(III) from tanning wastewater efﬂuents [30] and in wastewater treatment [31]. Therefore, the objective of this research is to investigate the inﬂuence of various experimental parameters on sorption of some rare earth elements (including yttrium, neodymium, gadolinium, samarium, and lutetium) on BCD as an effective and low cost sorbent material in aqueous solutions. Moreover, the kinetics, isotherms and thermodynamics characteristics of the sorption

O.I.M. Ali et al. / Journal of Hazardous Materials 195 (2011) 62–67

63

complexing agent at 655 nm against reagent blank [34]. Uptake percentage (%E) and distribution constant KD (mL/g) were calculated from the equations: %E =

Co − Ce × 100 Co

KD (mL/g) =

(1)

C − C V o e Ce

m

(2)

where Co and Ce are the initial and equilibrium REE concentration in the solution (mg/L), respectively. V is the volume of the aqueous solution (mL) contacted with BCD and m is the mass of BCD in grams. 2.3. Sorption kinetics

Fig. 1. X-ray diffraction of by-pass cement dust.

process of the investigated elements from their aqueous solutions on BCD will be discussed.

To examine the controlling mechanism of the sorption process, two kinetic models were used to test experimental data. The Lagergren pseudo ﬁrst-order equation is a simple kinetic analysis of sorption in the following form [35]: log(qe − qt ) = log qe −

2. Experimental 2.1. Materials 2.1.1. By-pass cement dust (BCD) By-pass cement dust (BCD) was brought from National Cement Co., Egypt. The received BCD was placed in a glass container which in turn was kept in a desiccator all the time of the experiments. BCD chemical composition (Table 1) was identiﬁed using X-ray ﬂuorescence (PANalytical Axios advanced, The Netherlands). The constituents’ phases of BCD were identiﬁed by X-Ray diffraction analysis (XRD brucker axs D8 advance, Germany) with CuK␣ radia˚ with a typical scanning begin at 2 equal to 20–80◦ tion (1.5406 A) and scan rate of 20 min−1 . X-Ray diffraction pattern of BCD is shown in Fig. 1. It indicates that BCD mainly consists of calcite, calcium sulfate, mono calcium silicates, calcium carbonate, quartz and sodium chloride (where CaCO3 = 38.34%, NaCl = 3.77%, KCl = 5.54%, CaSO4 = 4.93%, SiO2 free = 7.08%, CaSiO3 = 5%, CaO free = 12.20%). The cation exchange capacity (CEC) of BCD was determined using ammonium acetate saturation method [32] and it was 6.53 meq/g. 2.1.2. Reagents Standard individual REE solutions (1000 mg/L) for Y(III), Nd(III), Gd(III), Sm(III) and Lu(III) were prepared from the corresponding oxides (Aldrich Co., Germany) by dissolving 0.254, 0.162, 0.231, 0.116 and 0.26 g, respectively in 10 mL of perchloric acid. The solutions were heated till complete evaporation and another 5 mL of perchloric acid was added, then the solutions were diluted to volume with 0.1 N perchloric acid. Arsenazo (III) was obtained from Aldrich Co, Germany. All other reagents used were analytical reagent grade. In all experiments, doubled distilled water was utilised for preparation and dilution of solutions.

k t 1 2.303

(3)

where qe and qt are the sorbed concentration of REE at equilibrium and at any time t, respectively and k1 is the overall rate constant of ﬁrst-order sorption. The slope of the plot of log(qe − qt ) as a function of t can be used to determine the ﬁrst-order rate constant k1 . The activation energy of the sorption process can be determined using Arrhenius equation. It can be calculated from the slope of a plot of log k and 1/T, since the slope equal to (−Ea /2.303R). In addition, the pseudo second-order equation based on sorption equilibrium capacity may be expressed in the following form [35]: 1 1 = + k2 t qe − qt qe

(4)

where k2 is the rate constant of the second-order sorption. Eq. (4) can be rearranged to: t 1 t = + 2 qt q k2 qe e

(5)

Similarly, the slope of the plot of t/qt as a function of t was used to determine the second order rate constant k2 that is used to determine the activation energy of the sorption process using Arrhenius equation. 2.4. Thermodynamic parameters Determination of thermodynamic parameters were based on experiments that carried out by shaking 0.02 g BCD with solutions of each REE of concentration (100 mg/L) adjusted at pH 7 ± 0.1 for 5 min at different temperatures. The thermodynamic parameters (H◦ , G◦ and S◦ ) were calculated from the sorption results. 3. Results and discussion

2.2. Sorption experiments 3.1. Sorption experiments The sorption experiments were studied by a batch technique. In the experiments, BCD was separately shaken with each REE solution at various experimental conditions. Separating of solid phase from liquid was done by centrifuging at 4000 rpm for 15 min. The pH of the solutions was maintained by thiel buffer [33] in the range (2–7). After equilibration, the REEs’ concentrations were determined spectrophotometrically employing Shimadzu UV–Vis160A Spectrophotometer using Arsenazo III (0.05%, w/v) as a

The parameters which may affect the uptake of REEs by pass cement dust, such as shaking time, temperature, pH, initial concentration of REE and sorbent mass were investigated. The results showed that the equilibrium reached to its maximum within 9 min of shaking, while uptake percentage only slightly increased on raising temperature up to 60 ◦ C. Therefore, the sorption experiments were carried out at room temperature for 9 min.

64

O.I.M. Ali et al. / Journal of Hazardous Materials 195 (2011) 62–67

Table 1 Major chemical constituents of BCD. Constituents

Mass (%)

CaO

SiO2

P2 O5

Na2 O

Fe2 O3

MgO

K2 O

Al2 O3

Cl−

SO3

LOI

44.24

10.54

6.35

2.00

1.92

1.45

3.5

4.7

4.93

2.9

17.47

Table 2 Variation of the KD with the REEs’ initial concentrations for their sorption on BCD. [REE]

50 100 200 300 400 500

KD × 104 Y3+

Nd3+

Gd3+

Sm3+

Lu3+

6.12 0.61 0.08 0.03 0.02 0.02

6.12 0.68 0.06 0.04 0.02 0.02

6.12 6.12 6.12 6.12 0.07 0.03

6.12 6.12 6.12 6.12 0.92 0.28

6.12 0.74 0.11 0.05 0.03 0.03

Fig. 3. Variation of the meq (REE)/g BCD with the REEs’ initial concentrations for their sorption on BCD. Operating conditions: 50 mL solution, 0.04 g BCD, pH = 7, and shaking time = 9 min.

Fig. 2. Variation of the uptake percentage with pH for REEs sorption on BCD. Operating conditions: 50 mL solution, [REE] = 100 mg/L, 0.02 g BCD, and shaking time = 9 min.

3.1.1. Effect of pH Fig. 2 shows the inﬂuence of pH on the sorption of the investigated REEs. The data reveals that the percentage of sorption steeply increases with increasing pH up to 7 ± 0.1. Consequently, in the subsequent work, the sorption experiments were carried out at pH 7. Generally, below pH 2 dissolution of BCD was occurred. Above pH 7 the formation of a precipitate was observed. Low REEs uptake at low pH values is most probably due to the protonation of the active sites in BCD, which inhibits their binding ability towards the REEs [36]. In addition, as pH increases, surface positive charge decreases, this would result in lower columbic repulsion of the adsorbed REE ions [37]. In aqueous solution, the hydrolysis of trivalent lanthanides begins at pH as low as 6 and various species can be formed, such as Ln(OH)2+ , Ln(OH)2 + , Ln(OH)3 , Ln(OH)4 − [38]. Thus, as pH increases, hydrolysis precipitation most probably would start due to the formation of various hydrocomplexes in aqueous solution. The observed reduction in the percentage of REE uptake on BCD at low pH by the sorbent indicates that the interaction between them is most probably due to an ion exchange process [39]. 3.1.2. Effect of initial concentration of REE The sorption of each REE as a function of their initial concentrations was studied at room temperature by varying the REE initial concentration from 50 to 500 mg/L. Table 2 shows the calculated

distribution coefﬁcients (KD ) at different REE initial concentrations. Over the studied range, the distribution coefﬁcients vary by more than 6 orders of magnitude. The inverse correlation between REE initial concentration and distribution coefﬁcient reﬂects the greater partitioning of REEs into the solid phase (up to 100 mg/L for Y3+ , Nd3+ , Gd3+ and 300 mg/L for Sm3+ and Lu3+ ). With increasing REEs concentration, the distribution coefﬁcients stabilize, owing to saturation of the BCD surface. These results indicate that energetically less favorable sites become involved with increasing REEs concentration in the aqueous solution. Additionally, the results were expressed in terms of meq of REE sorbed per unit mass of BCD (Fig. 3) to investigate the difference in afﬁnity of BCD towards each REE. It is obvious from this ﬁgure that the sorption capacity of BCD has the order of Lu3+ > Sm3+ > Y3+ > Gd3+ ≈ Nd3+ . The amount of REE sorbed per unit mass of BCD increased with the initial REE concentration as expected. The plateau values for Lu3+ , Y3+ , Gd3+ , and Nd3+ represent saturation of the active sites on BCD available for interaction. The decrease in Sm3+ sorption at high concentration may be due to the Sm3+ afﬁnity to sorbate–sorbate interaction than sorbate–sorbent one. This is more or less in agreement with the sorption of Sm3+ by natural clinoptilolite-containing Tuff [40]. The data from Fig. 3 reveals that the sorption capacity of Sm3+ and Lu3+ was higher than CEC of BCD which reﬂects again that sorption mechanism of these elements is a mixed mechanism of ion exchange and hydrolysis precipitation as well. 3.1.3. Effect of sorbent mass Effect of sorbent mass on the sorption process of the investigated REEs is represented in Fig. 4. The experimental results reveal that the sorption efﬁciency of REEs increases up to the optimum mass of 0.04 g BCD for Y3+ , Gd3+ , Nd3+ and 0.01 g for Sm3+ and Lu3+ beyond which the sorption efﬁciency does not change with the sorbent mass.

O.I.M. Ali et al. / Journal of Hazardous Materials 195 (2011) 62–67

65

Fig. 5. Langmuir sorption isotherm of REEs sorption on BCD.

3.4. Sorption isotherms models Fig. 4. Variation of the uptake percentage with the sorbent mass for REEs sorption on BCD. Operating conditions: 50 mL solution, [REE] = 100 mg/L, pH = 7, and shaking time = 9 min.

The increase in uptake percentage with increasing the BCD mass may be due to increasing number of sorbent particles in the solution that allows more REEs ions to interact with more binding sites. 3.2. Sorption kinetics For simplicity, the kinetics experiments were carried out for Gd3+ and Sm3+ . The studies shows that the pseudo-ﬁrst-order kinetic model did not ﬁt the data for the sorption process since the values of correlation factor R2 were small. Therefore, the pseudosecond-order kinetic model was applied. The calculated values of k2 and Ea with the values of the linear correlation coefﬁcients (R2 ) are represented in Table 3. The values of correlation factor R2 of the pseudo-second-order were better than those of pseudo-ﬁrst-order model indicating second-order kinetics of the sorption process on BCD. The natural logarithms of the rate constants (k) were used according to the Arrhenius equation to calculate the activation energy of sorption process. It was found that the activation energy was within the range of 0–40 kJ/mol, which means that the sorption process is a physical one [41]. 3.3. Thermodynamic parameters For better understanding the mechanism of the sorption process of REEs on BCD, thermodynamic parameters were determined. They were calculated using the equation: ln KD =

H ◦ S ◦ − R RT

(6)

where KD is the distribution coefﬁcient (mL/g), S◦ is standard entropy (J/mol K), H◦ is standard enthalpy (kJ/mol), T is the absolute temperature (K) and R is the gas constant (8.314 J/mol K). The standard Gibbs free energy G◦ values (kJ/mol) were calculated from the equation: G◦ = H ◦ − TS ◦ H◦ ,

(7) S◦

G◦

The values of and were calculated from the slopes and intercepts of plots of ln KD versus 1/T. The values are presented in Table 4. The positive value of enthalpy change H◦ for the process conﬁrms the endothermic nature of the process, while the positive entropy of sorption S◦ reﬂects the afﬁnity of BCD towards these elements. The obtained values of G◦ point to that the feasibility of the sorption process of the investigated elements on BCD and its spontaneous nature without an induction period.

The sorption data of Sm3+ , Gd3+ and Nd3+ have been subjected to different sorption isotherms models namely Langmuir and Freundlich over a range of 100–300 mg/L for Nd3+ and 100–400 mg/L for Gd3+ and Sm3+ . The goodness-of-ﬁt between experimental data and the model predicted values was expressed by the correlation coefﬁcient (R2 , values close or equal 1). When the value of R2 is close to 1, it does not mean that the ﬁt is necessarily good [42]. Therefore, the conformity between experimental data and the model predicted values was expressed by the total mean error (ε%), which is the discrepancy between the experimental data and the predicted values [43]:

n

ε% =

i=1

|qe(exper.) − qe(calc.) |

n

q i=1 e(exper.)

(8)

A relatively low (ε%) value indicates which model can be successfully used to describe the sorption equilibrium on BCD. 3.4.1. Langmuir isotherm The Langmuir isotherm was applied for the sorption equilibrium on BCD according to the equation: Ce 1 Ce = + qe Qe bQe

(9)

where Ce is the equilibrium concentration of REE in solution (mg/L), qe is the amount of solute sorbed per unit mass of BCD at equilibrium (mg/g) and Qe (mg/g) and b (L/mg) are the Langmuir constants related to monolayer sorption capacity and free energy of sorption, respectively. From Table 5, it can be concluded that Qe parameter for Sm3+ was 8.32 meq/g, which was higher than the CEC of BCD. These data reﬂects again the mixed mechanism of sorption on BCD of ion exchange process and hydrolysis precipitation as well. According to the Langmuir model, sorption occurs uniformly on the active sites of the sorbent and once a sorbate occupies a site, no further sorption can take place at this site. A plot of Ce /qe versus Ce would result in a straight line with a slope of (1/Qe ) and intercept of 1/bQe as shown in Fig. 5. The values of the slopes, intercepts of the plots and total mean errors are presented in Table 5. The values of total mean error (ε%) between this model and the experimental data for Sm3+ , Gd3+ and Nd3+ were 3.76, 1.38 and 6.31, respectively. 3.4.2. Freundlich isotherm The simple Freundlich isotherm is often used for heterogeneous surface energy systems according to the equation: log qe = log k +

1 log Ce n

(12)

where qe is the amount of solute sorbed per unit mass of BCD at equilibrium (mg/g) and Ce is the equilibrium concentration of REE

66

O.I.M. Ali et al. / Journal of Hazardous Materials 195 (2011) 62–67

Table 3 The calculated parameters of the pseudo-second order kinetics model for sorption of Gd3+ and Sm3+ on BCD at different temperatures. Gd3+

Temperature (K)

Sm3+ −3

k2 (×10 303 313 323 333

2

g/mg min)

R

3.22 5.51 7.04 8.99

Ea (kJ/mol)

0.9927 0.9843 0.9813 0.9908

27.96

k2 (×10−3 g/mg min)

R2

10.23 11.31 12.04 13.98

0.9987 0.9928 0.999 0.9941

Ea (kJ/mol)

9.97

Table 4 Thermodynamic parameters for sorption of Y3+ , Nd3+ , Gd3+ , Sm3+ and Lu3+ on BCD. H◦ (kJ/mol)

REE

3+

3+

3+

Y , Gd , Nd Sm3+ , Lu3+

S◦ (kJ/mol K)

50.72 19.99

G◦ (kJ/mol)

0.250 0.124

303 K

313 K

323 K

−24.90 −17.60

−27.03 −18.82

−30.03 −20.06

333 K −32.53 −21.30

Table 5 Isotherms constants and values of correlation factors (R2 ) for sorption of Sm3+ , Nd3+ and Gd3+ on BCD. REE

Freundlich isotherm

3+

Sm Nd3+ Gd3+

Langmuir isotherm

1/n

log k

R2

ε%

Qe (meq/g)

b

R2

RL

ε%

0.599 0.220 0.260

1.828 2.130 1.860

0.9388 0.9722 0.9541

2.83 0.50 0.68

8.32 4.44 5.87

0.123 0.089 0.096

0.9835 0.9892 0.9945

0.010 0.0270 0.075

3.76 2.31 1.38

Fig. 6. Freundlich sorption isotherm of REEs sorption on BCD.

in solution (mg/L). K and n are constants characteristics of the system. Log k and 1/n are the Freundlich constants related to sorption capacity and sorption intensity of the sorbent, respectively. A plot of log qe as a function of log Ce would result in a straight line with a slope of (1/n) and intercept of log k as shown in Fig. 6. The values of 1/n, log k and ε% are presented in Table 5. The values of 1/n < 1 correspond to a heterogeneous surface with an exponential distribution of energy of the sorption sites [44]. The total mean error (ε%) between this model and the experimental data represents the best ﬁt of experimental data than Langmuir one. The ﬁt of the data to Freundlich isotherm indicates that the sorption process is not restricted to one speciﬁc class of sites and assumed surface heterogeneity [45]. This trend is due to the high surface area of the sorbent and multilayer of sorption on the BCD. This trend was also investigated by M. Al-Meshragi et al. [46] for sorption of Cr(III) on BCD. 4. Conclusions • The efﬁciency of BCD for the sorption of Y3+ , Nd3+ , Gd3+ , Sm3+ and Lu3+ from their aqueous solutions was investigated. It was

found that maximum sorption capacity was achieved at pH 7 using thiel buffer and the sorption capacity of BCD has the order of Lu3+ > Sm3+ > Y3+ > Gd3+ ≈ Nd3+ . The sorption of these elements on BCD was found to follow pseudo-second-order kinetics. • The thermodynamic parameters H◦ , S◦ and G◦ values of the REEs sorption onto BCD show endothermic heat of sorption, favored at high temperatures. The positive entropy value is an indication of the probability of favorable nature of sorption and the process is spontaneous. • The equilibrium data have been analyzed using Langmuir and Freundlich isotherms. The Freundlich isotherm was demonstrated to provide the best correlation and the lowest total error for sorption of the studied elements on BCD. • BCD may be successfully used as effective, low cost and abundant source for the removal of Y3+ , Nd3+ , Gd3+ , Sm3+ and Lu3+ from their aqueous solutions and may be used as alternative to more costly materials. References [1] J. Kalinowski, J. Mezyk, F. Meinardi, R. Tubino, M. Cocchi, D. Virgili, Electric ﬁeld and charge induced quenching of luminescence in electroluminescent emitters based on lanthanide complexes, Chem. Phys. Lett. 453 (2008) 82–86. [2] G.A. Molander, J.A.C. Romero, Lanthanocene catalysts in selective organic synthesis, Chem. Rev. 102 (2002) 2161–2185. [3] K. Kuriki, Y. Koike, Y. Okamoto, Plastic optical ﬁber lasers and ampliﬁers containing lanthanide complexes, Chem. Rev. 102 (2002) 2347–2356. [4] K.H. Johannesson, K.J. Stetzenbach, V.F. Hodge, Rare earth elements as geochemical tracers of regional groundwater mixing, Geochim. Cosmochim. Acta 61 (1997) 3605–3618. [5] W. Bremmer, Rare earth applications in Chinese agriculture elements, Rare Earths Spec. Metals Appl. Technol. 3 (1994) 20–24. [6] S.-L. Tong, W.-Z. Zhu, Z.-H. Gao, Y.-X. Meng, R.-L. Peng, G.-C. Lu, Distribution characteristics of rare earth elements in children’s scalp hair from a rare earths mining area in southern China, J. Environ. Sci. Health A: Toxic/Hazard. Subst. Environ. Eng. 39 (2004) 2517–2532. [7] L. Weltje, H. Heidenreich, W.Z. Zhu, H.T. Wolterbeek, S. Korhammer, J.J.M. de Goeij, B. Markert, Lanthanide concentrations in freshwater plants and molluscs, related to those in surface water, pore water and sediment. A case study in The Netherlands, Sci. Total Environ. 286 (2002) 191–214. [8] A. Pałasz, P. Czekaj, Toxicological cytophysiological aspects of lanthanides action, Acta Biochim. Pol. 47 (2000) 1107–1114. [9] F. Wiberg, N. Wiberg, A.F. Hollemann, Inorganic Chemistry, Academic Press, San Diego, CA.

O.I.M. Ali et al. / Journal of Hazardous Materials 195 (2011) 62–67 [10] M. Atanassovaa, Synergistic solvent extraction and separation of lanthanide (III) ions with 4-benzoyl-3-phenyl-5-isoxazolone and the quaternary ammonium salt, Solvent Extr. Ion Exch. 27 (2009) 159–171. [11] S.D. Faust, O.M. Aly, Chemistry of Water Treatment, Butter Worth Publishers, NY, USA, 1976. [12] H.A. Omar, H. Moloukhia, Use of activated carbon in removal of some radioisotopes from their waste solutions, J. Hazard. Mater. 157 (2008) 242–246. [13] M. Grover, M.S. Narayanswamy, Removal of hexavalent chromium by adsorption on ﬂy ash, Institute of Engneers Indian, J. Environ. Eng. 63 (1982) 36–39. [14] N. Kannan, A. Vanangamudi, A study on removal of chromium Cr(VI) by adsorption lignite coal, Ind. J. Environ. Pollut. 11 (1991) 241–245. [15] S. Chand, V.K. Agarwal, C. Pavankumar, Removal of hexavalent chromium from wastewater by adsorption, Ind. J. Environ. Health 36 (1994) 151–158. [16] A. Sen, A. De, Adsorption of mercury (II) by coal ﬂy ash, Water Res. 21 (1987) 885–888. [17] K. Srinivasan, N. Balasubramaniam, T.V. Ramakrishna, Studies on chromium removal by rice husk carbon, Ind. J. Environ. Health 30 (1988) 376–387. [18] P.A. Brown, S.J. Allen, Metal removal from wastewater using peat, Water Res. 34 (16) (2000) 3907–3916. [19] D. Humelnicu, L. Bulgariu, M. Macoveanu, On the retention of uranyl and thorium ions from radioactive solution on peat moss, J. Hazard. Mater. 174 (2010) 782–787. [20] V.K. Gupta, M. Gupta, S. Sharma, Process development for the removal of lead and chromium from aqueous solutions using red mud—an aluminum industry waste, Water Res. 35 (2001) 1125–1134. [21] A.H. Mahvi, D. Naghipour, F. Vaezi, S. Nazmara, Teawaste as an adsorbent for heavy metal removal from industrial wastewaters, Am. J. Appl. Sci. 2 (1) (2005) 372–375. [22] C. Kütahyali, M. Eral, Sorption studies of uranium and thorium on activated carbon prepared from olive stones: kinetics and thermodynamics aspects, J. Nucl. Mater. 396 (2010) 251–256. [23] E.M. Saad, R.A. Mansour, A. El-Asmy, M.S. El-Shahawi, Sorption proﬁle and chromatographic separation of uranium (VI) ions from aqueous solutions onto date pits solid sorbent, Talanta 76 (2008) 1041–1046. [24] A. Mackie, S. Boilard, M.E. Walsh, C.B. Lake, Physicochemical characterization of cement kiln dust for potential reuse in acidic wastewater treatment, J. Hazard. Mater. 173 (2010) 283–291. [25] H.Y. Ahmed, A.M. Othman, A.A. Mahmoud, Effect of using waste cement dust as a mineral ﬁlter on the mechanical properties of hot asphalt, Ass. Univ. Bull. Environ. Res. 9 (2006) 51–59. [26] E.K. Mahmoued, Cement kiln dust and coal ﬁlters treatment of textile industrial efﬂuents, Desalination 255 (2009) 175–178. [27] N.G. Zaki, I.A. Khattab, N.M. Abd El-Monem, Removal of some heavy metals by BCD leachate, J. Hazard. Mater. 147 (2007) 21–27. [28] M.H. El-Awady, T.M. Sami, Removal of heavy metals by cement kiln dust, Bull. Environ. Contam. Toxicol. 59 (1997) 603–610. [29] O.M. Ali, H.H. El-Kashif, S.A. Sayed, M.E.H. Shalabi, The removal of uranium and thorium from aqueous solutions onto by-pass cement dust (BCD), Eurasian Chem. Tech. J. 11 (2009) 105–113.

67

[30] O.A. Fadali, E.E. Ebrahiem, Y.H. Magdy, A.A.M. Daifullah, M.M. Nassar, Removal of chromium from tannery efﬂuents by adsorption, J. Environ. Sci. Health A: Environ. Sci. Eng. 9 (2005) 465–472. [31] B. Hegazy, M.A. El-Khateeb, A.A. El-adly, M.M. Kamel, Low-cost wastewater treatment technology, J. Appl. Sci. 7 (2007) 815–819. [32] M. Horshall, A.I. Spiff, Studies on the effect of pH on the sorption of Pb2+ and Cd2+ ions from aqueous solutions by Caladium bicolor (Wild Cocoyam) biomass, J. Biotechnol. 7 (3) (2004) 313–323. [33] H.T.S. Britton, Hydrogen Ions, fourth ed., Chapman and Hall, London, 1952, pp. 354–358. [34] Z. Marczenko, Spectrophotometric Determination of Elements, third ed., Ellis Horwood Chichester, UK, 1986. [35] Y.S. Ho, G. Mckay, Pseudo-second order model for sorption processes, Water Res. 34 (5) (1999) 578–584. [36] M.S. El-Shahawi, M.A. Othman, M.A. Abdel-Fadeel, Kinetics, thermodynamic and chromotographic behavior of the uranyl ions sorption from aqueous thiocyanate media onto polyurethane foams, Anal. Chim. Acta 546 (2005) 221–228. [37] K.E. Laintz, J. Yu, C.M. Wai, Separation of lanthanides with sodium bis-(triﬂuoroethyl) dithiocarbamate chelation and supercritical ﬂuid chromatography, Anal. Chem. 64 (3) (1992) 311–315. [38] E. Bentouhami, G.M. Bouet, J. Meullemeestre, F. Vierling, M.A. Khan, Physicochemical study of the hydrolysis of rare-earth elements (III) and thorium (IV), Comptes Rendus Chimie 7 (5) (2004) 537–545. [39] E.A. El-Sofany, Sorption of Cd(II) and Se(IV) from aqueous solution using modiﬁed rice husk, J. Hazard. Mater. 147 (2007) 546–555. [40] N.M. Kozhevnikova, E.P. Ermakova, A study of sorption of samarium(III) ions by natural clinoptilolite-containing tuff, Russ. J. Appl. Chem. 81 (2008) 2095–2098. [41] A. Mellah, S. Chegrouche, M. Barkat, The removal of uranium (VI) from aqueous solution onto activated carbon: kinetic and thermodynamic investigation, Colloid Interface Sci. 169 (2006) 146–152. [42] S.C. Chapra, R.P. Canale, Numerical Methods for Engineers, third ed., McGrawHill Co., Singapore, 1998. [43] W.H. Press, B.P. Flannery, S.A. Teukolsky, W.T. Vetterling, Numerical Recipes in Pascal: The Art of Scientiﬁc Computing, Cambridge University Press, Cambridge, 1989. [44] M.M. Saeed, Adsorption proﬁle and thermodynamic parameters of the preconcentration of Eu(III) on 2-thenoyltriﬂuoroacetone loaded polyurethane (PUR) foam, J. Radioanal. Nucl. Chem. 256 (2003) 73–80. [45] E.A. El-Sofany, Removal of lanthanum and gadolinium from nitrate medium using Aliquat-336 Impregnated Onto Amberlite XAD-4, J. Hazard. Mater. 153 (2008) 948–949. [46] M. Al-Meshragi, H.G. Ibrahim, M.M. Aboabboud, Equilibrium and kinetics of chromium adsorption on cement kiln dust, in: Proceedings of the World Congress on Engineering and Computer Science, San Francisco, USA, October 22–24, 2008.

Journal of Hazardous Materials 195 (2011) 68–72

Contents lists available at ScienceDirect

Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat

Silver nanotoxicity using a light-emitting biosensor Pseudomonas putida isolated from a wastewater treatment plant R.I. Dams a,∗ , A. Biswas a , A. Olesiejuk a , T. Fernandes b , N. Christoﬁ a a b

Centre for Nano Safety, Edinburgh Napier University, Scotland, UK Nano Safety Research Group, Heriot-Watt University, Edinburgh EH14 4AS, Scotland, UK

a r t i c l e

i n f o

Article history: Received 8 November 2010 Received in revised form 29 July 2011 Accepted 4 August 2011 Available online 10 August 2011 Keywords: Pseudomonas putida Silver nanoparticles Nanotoxicity

a b s t r a c t The effect of silver ions, nano- and micro-particles on a luminescent biosensor bacterium Pseudomonas putida originally isolated from activated sludge was assessed. The bacterium carrying a stable chromosomal copy of the lux operon (luxCDABE) was able to detect toxicity of ionic and particulate silver over short term incubations ranging from 30 to 240 min. The IC50 values obtained at different time intervals showed that highest toxicity (lowest IC50 ) was obtained after 90 min incubation for all toxicants and this is considered the optimum incubation for testing. The data show that ionic silver is the most toxic followed by nanosilver particles with microsilver particles being least toxic. Release of nanomaterials is likely to have an effect on the activated sludge process as indicated by the study using a common sludge bacterium involved in biodegradation of organic wastes. © 2011 Elsevier B.V. All rights reserved.

1. Introduction Engineered nanomaterials, that have seen signiﬁcant increases in use recently, are mainly composed of metallic nanoparticles. The knowledge of the ecotoxicology of nanoparticles (NP) to bacteria and other microbes is still limited, even though some manufactured nanoparticles (NP; materials with three dimensions between 1 and 100 nm) [1] such as silver and titanium, which are constantly released into the environment, are known as antibacterial agents. As such it is essential that technology that fully assesses their effects on natural microbial organisms, and on biogeochemical cycling in the environment, is available. Clearly there is a concern that these novel materials could be released into the environment. Whole microbial cell biosensors are now widely used as research tools in the testing of substances likely to elicit cytotoxic and genotoxic events, and, in the determination of bioavailability of chemicals [2]. They embrace genetically engineered bacteria that have a toxicant detecting gene that is coupled with a reporter gene (e.g. luminescence gene such as lux or luc) capable of producing a detectable response on activation. Wiles et al. [3] argue that autochthonous microorganisms would be appropriate in toxicity testing with the potential for in situ relevance. Silver, a metal used extensively in various consumer products because of its effective antimicrobial properties, is subject to release to sewer the sewerage system. Therefore it is important

∗ Corresponding author. Tel.: +44 0131 455 2291; fax: +44 0131 455 2291. E-mail address: [email protected] (R.I. Dams). 0304-3894/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.jhazmat.2011.08.013

to determine whether has an effect on activated sludge microorganisms. Silver nanoparticles are well known for their antibacterial activity [4–7] and have become one of the nanomaterials mostly used in consumer product [8]. Being considered as the most prevalent of engineered materials [9] is likely to enter the wastewater treatment plants as nanowaste. Blaser et al. [10] pointed out that silver released to wastewater is incorporated into sewage sludge and may spread further on agricultural ﬁelds where will mainly stay in the top layer of soils [11]. This work aims to assess the effect of silver nanoparticles on the activated sludge process by examining the response of a common bacterial member of the sludge consortium involved in sewage organic degradation. In this study ecotoxicological was performed using a genetically-modiﬁed Pseudomonas isolated from a polluted, phenolic-rich, wastewater treatment system by transposon mutagenesis [3]. The Pseudomonas carries a stable chromosomal copy of the lux operon (luxCDABE) derived from Photorhabdus luminescens with continuous output of light. The bioluminescence bioassay performed in this study has the advantage of allowing monitoring the presence of these nanoparticles, specially silver nanoparticles, the object of this study. 2. Materials and methods 2.1. Silver chemicals All of the chemicals used were analytical grade. Silver powder nanoparticles (average size: 35 nm, 99.5% metal basis, spherical morphology and cubic crystallographic structure) and

R.I. Dams et al. / Journal of Hazardous Materials 195 (2011) 68–72

69

microparticles (average size: 0.6–1.6 ␮m) were obtained from Nanoamor (Nanostructured and Amorphous Materials Inc., TX, USA). Stock suspensions of silver ion (AgNO3 ), silver nanoparticles (Ag-NP) and silver microparticles (Ag-MP) were freshly made in 20 ml universal ﬂasks and placed into an ultrasonic bath (XB6 Grant Instruments Cambridge Ltd., UK) at 25 kHz, 25 ◦ C for 30 min. Two fold dilutions were externally prepared of these suspensions and added to the 96 well black microtitre plates (Sterling, Caerphilly, UK) to give ﬁnal toxicant concentration in the range to be tested. Two different stabilisers were added to the nanoparticles and microparticles working solutions: 0.1% citric acid, 0.1% BSA (bovine serum albumin) and comparisons were also made with preparations without stabilisers. Previous studies have used BSA for stabilisation of ZnO nanoparticles [12] and of carbon black nanoparticles [13]. Thus improving the stability of particle suspension reducing particle agglomeration and settling over time [13]. Citric acid is known to act as a chelating agent and may thus be able to also act to quench toxicity of dissolved metal toxicants. All working solutions were light protected and used within 30 min of preparation. In order to assess any shading effects of the particles on the bacterial cells, the light output was measured before and after addition of particles. No shading effects were determined.

For error analysis, all of the experiments were conducted 3 times on different plates. Data from eight wells were used for one concentration and coefﬁcients of variation (CV) between independent assays were calculated using Microsoft Excel 97. Differences among treatments were tested using a two-way analysis of variance (ANOVA) to determine which treatments were statistically different (P < 0.05).

2.2. Media and growth conditions

3. Results and discussion

The bacterial strain used in this study, Pseudomonas putida BS566::luxCDABE was constructed based upon chromosomal expression of the luxCDABE operon derived from an entomopathogenic nematode symbiont, P. luminescens [14]. Originally isolated from the treatment system [15], this reporter organism encompassing a dynamic xenobiotic sensing range, is suitable for placement around an industrial processing system to monitor remediation in multiple compartments. Cultures were grown in Luria–Bertani (LB) broth (10.0 g L−1 of Tryptone; 5.0 g L−1 of yeast extract; 5.0 g L−1 of sodium chloride), containing 100 mg L−1 Kanamycin, and overnight at 26 ◦ C in shaking conditions (200 rpm) to late exponential stage. Prior to the assay, cultures were diluted to approximately 107 cells/ml and regrown under the same conditions for two to three generations without Kanamycin. When OD (optical density) reached 0.2 (approximately 108 cells/ml) toxicity tests were carried out.

IC50 values following challenge with AgNO3 , Ag-NP and Ag-MP (with and without dispersant) are shown in Table 1. The experimental uncertainty of these bioluminescence bioassays is within the coefﬁcients of variation. Calculated coefﬁcients of variation (CV) between independent assays were found to be between 1 and 15%. Fig. 1 shows the light output reduction by P. putida BS566::luxCDABE when challenged with different concentrations up to 2500 ␮g L−1 silver ion. Among the silver species tested, AgNO3 is by far the most toxic to P. putida highlighting the action of ionic Ag+ (P < 0.05). It is well known that silver ion and silver based compounds are highly toxic to micro organisms and have strong biocide effects to many bacteria species [17–20]. Figs. 2 and 3 show the light output reduction by P. putida BS566::luxCDABE when challenged with Ag nanoparticles and Ag microparticles. In the presence of BSA as stabilizer, Ag nanoparticles showed to be statistically more toxic than Ag microparticles (P < 0.05). The same was observed when using citric acid as stabilizer. The highest toxicity (Table 1) was observed after 90 min incubation and indeed this was the case for all silver species tested with or without stabilisation. The order in toxicity was Ag+ > AgNP (35 nm) > Ag-MP (0.6–1.6 ␮m). In this current study we tested three different silver species, and ionic Ag+ being the most toxic to the biosensor tested. Similar results were found by Choi et al. [4] when testing the toxicity of silver species. In their studies, the

2.3. Biosensor assay Luminescence measurements were undertaken using a 96well plate luminometer (FLUOstar Optima, BMG Labtech, UK) in 96 well black microtitre plates (Sterling, Caerphilly, UK) whereby each well contained bacterial inoculum and toxicant at the required concentration in 100 ␮L volumes, using an integration time of 1 s at a temperature of 28 ◦ C. Readings were taken every 30 min for 240 min. Control wells containing LB broth with P. putida BS566::luxCDABE were run and changes in toxicity for the test systems expressed as percentages of the control. Luminescence values were expressed in the instrument’s arbitrary relative light units (RLU). The maximal response ratios were the highest ratios of luminescence in the samplecontaining wells to luminescence in wells containing untreated cells determined during a speciﬁed period: 30, 90, 180 and 240 min [16]. Inhibitory concentration which represents 50% inhibition of light output (IC50 ) in relation to the control was assessed for all toxicants tested at each time point. All experiments were run at least 3 times (most were ran 4 times) at different dates with different batches.

2.4. Calculation of IC50 (IC, inhibitory concentration) values The IC values were calculated using a statistical program developed in-house. The program ﬁts a three parameter logistic model to the logarithm of the concentration by weighted least squares. The parameters are the initial response, the slope and the intercept. It is assumed that the response would decline to zero at sufﬁciently high concentrations. The initial response effectively uses the information from both the controls (if present) and low concentrations. The weights used are taken to be proportional to the ﬁtted response but with adjustments for high and low responses; this is to protect against bias due to “hormesis” effects (stimulatory effects causing increased light output when challenged with low toxicant concentrations) and the effective omission of data respectively. 2.5. Data analysis

Table 1 IC50 (mg L−1 ) values for light emission reduction by P. putida BS566::luxCDABE after 30, 90, 180 and 240 min. Values presented are an average of at least 3 independent experiments carried out with different batches, standard deviation between 1 and 15%. IC50 values (mg L−1 ) Time (min)

30

90

180

240

AgNO3 Ag-NP Ag-NP BSA Ag-NP CA Ag MP Ag MP BSA Ag MP CA

0.44 88 102 147 715 375 700

0.18 81 35 126 530 256 240

0.25 91.5 43 136 765 308 300

0.30 184 50 149 1075 330 337

70

R.I. Dams et al. / Journal of Hazardous Materials 195 (2011) 68–72

Fig. 1. Light output reduction by P. putida BS566::luxCDABE when challenged with 2500 ␮g L−1 silver ion after 30, 90, 80 and 240 min. Mean of three replicates with different batches.

ionic form Ag+ was more toxic to heterotrophic Escherichia coli than 16 nm silver nanoparticles, while to autotrophic nitrifying bacteria Ag nanoparticles were more toxic than the ionic form. It has been pointed out that the bactericidal effect of nanoparticles is dependent on the concentration of nanoparticles and the initial bacterial concentration [19]. In our study we used 35 nm silver nanoparticles which were dispersed in liquid cultures with an initial bacterial concentration of 108 UFC cells/ml for the toxi-

Fig. 3. Light output reduction by P. putida BS566::luxCDABE when challenged with 1500 mg L−1 Ag micro particles: (a) without stabilizer, (b) with 0.1% citric acid, and (c) with BSA after 30, 90, 80 and 240 min. Mean of three replicates with different batches.

Fig. 2. Light output reduction by P. putida BS566::luxCDABE when challenged with 200 mg L−1 Ag nanoparticles: (a) without stabilizer, (b) with 0.1% citric acid, and (c) with BSA after 30, 90, 80 and 240 min. Mean of three replicates with different batches.

city tests. Using E. coli as a model, Sondi and Sondi-Salopeki [18] showed that the bacterial growth in LB liquid medium (with an initial bacterial concentration of 107 UFC cells/ml) was delayed by nanoparticles size 12 nm. However, other studies using the Microtox system [21] for bioluminescence testing have shown no toxicity on exposures to silver nanoparticles. In their study, however, these authors used a marine microrganism Photobacterium phosphoreum which requires a sodium chloride concentration of 22% and is not indicated for ecotoxicity of terrestrial systems [21]. Whole cell biosensors are of greater environmental relevance for luminescence-based testing in terrestrial systems compared to the Microtox system testing. Furthermore, whole cell biosensor have shown a greater sensitivity than Microtox system [16]. In this current study we used a whole cell biosensor P. putida BS566::luxCDABE, a terrestrial bacterium for the bioluminescence testing. This methodology has been used in many studies formerly [2,15,16,22–25]. Beaton et al. [22] showed that the biosensor E. coli HB101 pUCD607 is a sensitive indicator of changes in toxicity in a soil system spiked with 2,4-dichlorophenol; Shaw et al. [23] used the biosensor lux marked Burkholderia RASC c2 in bioluminesce inhibition studies and Boyd et al. [24] used Burkholderia RASC c2 and Pseudomonas ﬂuorescens 10586 to assess the toxicity of chlorophenols. Sinclair et al. [25] showed the toxic response of the lux-marked biosensors as Pseudomonas ﬂuorencens and E. coli to 2,4-dichlorophenol.

R.I. Dams et al. / Journal of Hazardous Materials 195 (2011) 68–72

Our biosensor responded within 90 min (as see in Table 1) to the presence of toxicants. Thus, this strain can be used for the rapid and sensitive detection of potentially toxic silver compounds. Overall, the toxicity of silver was found not to be dependent. Overtime, the IC50 values obtained over 90-min for ionic silver, Ag-NP and Ag-MP was about 1.5–2.0 times lower than the 240-min test. Therefore, according to our ﬁndings, a 90-min test should be taken when monitoring and evaluating wastewater treatment plants for silver toxicity. If we consider 90 min assay results, which shows the highest toxicity, to compare the data, it can be noticed that Ag-NP are nearly 200 times less toxic that Ag+ , with Ag-MP ∼3000 times less toxic in non-stabilised systems (P < 0.05). The effect with Ag-NP was more pronounced in BSA systems with a calculated IC50 value ∼4 times lower than Ag-NP with citric acid (P < 0.05). Addition of BSA seems to result in higher toxicity perhaps through better dispersion of the NP providing more surface area for Ag to have an effect. This however may not be the case as citric acid may be just as effective in preventing agglomeration. Citric acid is an effective chelating/complexing agent for metals in solution. The effect of citric acid on the toxicity value might be due to immobilization of Ag+ released in solution or those on surfaces. On the one hand one may need to show a worst case toxic effect of well dispersed NP systems but on the other hand real systems will have a range of cheating/complexing agents that will affect the toxicity of substances released into its environment. Dissolution is likely to be a critical step for some metallic nanoparticles in determining fate in the environment and within the organism. In this study when particles were well dispersed using BSA as stabilizer, a higher toxicity was observed. Solubility strongly inﬂuences the toxicity and when no stabilizer was used the IC50 values were higher, probably indicating that less soluble compounds were available for the bacterial cells. Brunner et al. [26] observed that nanoparticles with a low solubility such as TiO2 showed no toxicity to mammalian cells while more soluble nanoparticles like ZnO showed a higher toxicity. The use of different stabilisers and their effect on toxicity values must be assessed particularly if laboratory bioassay results are to be used to derive wastewater discharge consents. There were no signiﬁcant differences between the IC50 values obtained for Ag-MP with BSA and citric acid stabilisation (P > 0.05) and both values were ∼2 times statistically lower than Ag-MP tested without stabilisation (P < 0.05). Particle size does cause a toxicity difference. In the present study, if we compare the toxicity between nanoparticles and their micro size counterparts, we noticed that when Ag-NP were stabilized showed a higher toxicity when compared with the micro size counterparts (P < 0.05). Other studies have also showed no toxicity of micro scaled particles when compared with nano size counterparts. For instance, Jiang et al. [20] observed a higher toxicity of nanoparticles of Al2 O3 , TiO2 and ZnO than their micro size scale counterparts which showed no or lower toxicity. Sinha et al. [27] have noticed that ZnO nanoparticles disintegrate Gram negative bacteria cell membrane and accumulate in cytoplasm, while when these cells were grown in micro particle counterparts the cell membrane and cytoplasm were intact. In this study we used 35 nm spherical nanosilver particles. Smaller nanosilver particles are more active than larger ones because of their higher surface area. However, in this study we used relatively large size of silver nanoparticles (35 nm) which proved to have a higher level of toxicity against the biosensor tested. Silver nanoparticles surface area plays quite an important role for antibacterial activity which depends on its exposed surface area concentration. This dependency is originated from the released Ag+ from the nanosilver surface. Recent studies [28,29] indicated that when nanosilver particles are small and release many Ag+ ions, the antibacterial activity is dominated by these

71

ions rather than the nanosilver particles. However, when larger sizes are used (as the ones used in this study) which have a low release of Ag+ ions, the nanosilver particles themselves also inﬂuence the antibacterial activity as indicated by these results here presented. Metal nanoparticles as silver have a tendency to attach on the cell wall [20]. So, toxicity is not dependent only on release of Ag+ , but also it depends on other factors such as the attachment of particles on the cell surface, disruption of cell membrane and consequently accumulation of nanoparticles in the cytoplasm. Surprisingly, the counterpart micro sized particles showed a lower toxicity if we consider that smaller particles release more Ag+ ions that the larger ones. However, one should consider other factors that affect toxicity such as the initial concentration of particles and bacteria, and particle bioavailability, among many other factors. So, although silver NPs have a higher surface area and probably a higher release of Ag+ , there was probably not enough contact in order to cause damage to cell membrane and cytoplasm since toxicity is dependent on contact and/or bacterial attachment to the particle as pointed by Jiang et al. [20] and Sinha et al. [27]. Furthermore, it has been demonstrated that small particles as the counterpart microsized are in suspension only in small numbers and are not able to attach to the bacterial surface as the nanoparticles do [20]. Then, in this case of the micro sized particles as the ones used in this study, they caused less harm to the bacterial cells than the nanoparticles ones due to their inability to attach on cell surface. The antimicrobial effect is related to the amount and the rate of silver released by NP. Severe structural changes occur in the bacterial cell wall when ionised silver binds to cell membrane proteins which leads to protein distortion and cell death [30,31]. Silver is classiﬁed as the “soft” metal group [10] and it complexes with many organic or inorganic materials such as chloride, sulphide, thiosulphate [32]. In order to evaluate the impact of silver discharge in the environment it is important to understand the fate and transport of silver in wastewater treatment plants. The applicability of our sensor in wastewater treatment plants has been previously demonstrated [3,33]. The biosensor here tested, P. putida BS566::luxCDABE had accurately predicted toxicity shifts in wastewater treatment plants, with a high tolerance to a phenolic cocktail, thus demonstrating an effective biosensing in all treatment compartments [3]. Philp et al. [33] have tested immobilised P. putida BS566::luxCDABE in phenolic wastewater treatment plant. The biosensor tested proved to be able to discriminate toxicity of various zones within the waste treatment plant [33]. Further studies in our laboratories using wastewater samples affected by silver toxicity have being carried out using this biosensor and results will be published somewhere else. Nanowaste is likely to increase and therefore enter the wastewater treatment plants which are the ﬁnal step to control silver discharge. Estimation of silver load in sewage sludge and its microorganisms growth inhibition has been predicted. Blaser et al. [10] predicted that an expected silver concentration in sewage treatment plant range from 2 ␮g L−1 to 18 ␮g L−1 . Shafer et al. [34] reported a range of ∼2–4 ␮g L−1 of silver in sewage treatment plants treating common wastewater and a much higher load from industrial discharges (from 24 to 105 ␮g L−1 ). The removal of silver ion by chloride free sludge is dependent on the silver-sludge loading, the solution pH and the concentration of dissolved organic matter. Studying the interactions of silver with wastewater constituents, Wang et al. [35] showed that silver ion can be removed through chloride precipitation and sludge adsorption. However, the authors [35] pointed out that the formation of silver-ion-dissolved organic matter complexes, which is increased in alkaline conditions, reduces the silver ion adsorption by sludge.

72

R.I. Dams et al. / Journal of Hazardous Materials 195 (2011) 68–72

4. Conclusions In conclusion, our results demonstrate that the use of a bacterial biosensor as P. putida BS566::luxCDABE provides a robust, early warning system of acute toxicity which could lead to process failure. This strain is suitable for toxicity monitoring in a highly polluted industrial wastewater treatment streams. The information regarding the inhibition of microbial growth by different Ag compounds, especially in wastewater treatment systems, is valuable for operating planning and control. The presence and activity of microorganisms in biological wastewater treatment are vital to the process. One of the projected applications of such strains is its combined use as analytical panel for toxicant detection. An important advantage of using these organisms is that a positive response will not only indicate the presence of a toxicant but will also provide some idea as to its character. Acknowledgements This work was part of a project funded by the Brazilian National Research and Development Council (CNPq). Also we would like to acknowledge the Environmental Microbiology Research Laboratory at Edinburgh Napier University, Edinburgh, Scotland. References [1] J. Bridges, W. Jong, T. Jung, D. Williams, T.F. Fernandes, J.-P. Marty, T. Butz, Opinion on the scientiﬁc aspects of the existing and proposed deﬁnitions relating to products of nanoscience and nanotechnologies, SCENIHR (Scientiﬁc Committee on Emerging and Newly-Identiﬁed Health Risks), European Commission, Brussels, Belgium, November 2007, . [2] S. Belkin, Genetically engineered microorganisms for pollution monitoring, in: Soil and Water Pollution Monitoring, Protection and Remediation, NATO Science Series: IV: Earth and Environmental Sciences, vol. 69, 2007, pp. 147–160. [3] S. Wiles, A.S. Whiteley, J.C. Philp, M.J. Bailey, Development of bespoke bioluminescent reporters with the potential for in situ deployment within a phenolic-remediating wastewater treatment system, J. Microbiol. Methods 5 (2003) 667–677. [4] O. Choi, K.K. Deng, N-J. Kim, L. Ross, Z.R.Y. Surampalli, The inhibitory effects of silver nanoparticles, silver ions, and silver chloride colloids on microbial growth, Water Res. 42 (2008) 3066–3074. [5] J.P. Ruparelia, A.K. Chatterjee, S.P. Duttagupta, S. Mukherji, Strain speciﬁcity in antimicrobial activity of silver and copper nanoparticles strain speciﬁcity in antimicrobial activity of silver and copper nanoparticles, Acta Biomater. 4 (2008) 707–716. [6] Q. Chang, L. Yan, M. Chen, H. He, J. Qu, Bactericidal mechanism of Ag/Al2 O3 against Escherichia coli, Langmuir 23 (2007) 11197–11199. [7] S. Ghosh, R. Kaushik, K. Nagalakshmi, S.L. Hoti, G.A. Menezes, B.N. Harish, H.N. Vasan, Antimicrobial activity of highly stable silver nanoparticles embedded in agar–agar matrix as a thin ﬁlm, Carbohydr. Res. 345 (2010) 2220–2227. [8] A.D. Maynard, E. Michelson, The Nanotechnology Consumer Product Inventory (2006), . [9] D. Rejeski, D. Lekas, Nanotechnology ﬁeld observations: scouting the new industrial waste, J. Cleaner Prod. 16 (2008) 1014–1017. [10] S.A. Blaser, M. Scheringer, M. MacLeod, K. Hungerbuhler, Estimation of cumulative aquatic exposure and risk due to silver: contribution of nanofunctionalised plastics and textiles, Sci. Total Environ. 390 (2008) 396–409. [11] H. Hou, T. Takamatsu, M.K. Koshikawa, M. Hosomi, Migration of silver, indium, tin, antimony and bismuth and variations in their chemical fractions on addition to uncontaminated soils, Soil Sci. 170 (2005) 624–639.

[12] R. Brayner, R. Ferrari-Iliou, N. Brivois, S. Djediat, F. Marc, M.F. Benedetti, F. Fieˇıvet, Toxicological impact studies based on Escherichia coli bacteria in ultraﬁne ZnO nanoparticles colloidal medium, Nano Lett. 86 (2006) 866–870. [13] L. Foucaud, M.R. Wilson, D.M. Brown, V. Stone, Measurements of reactive species production by nanoparticles prepared in biologically relevant media, Toxicol. Lett. 174 (2007) 1–9. [14] M. Winson, S. Swift, P.J. Hill, C.M. Sims, G. Griesmayr, B.W. Bycroft, P. Williams, S.A.B. Gordon, G.S.A.B. Stewart, Engineering the luxCDABE genes from Photorhabdus luminescens to provide a bioluminescent reporter for constitutive and promoter probe plasmids and mini-Tn5 constructs, FEMS Microbiol. Lett. 163 (1998) 193–202. [15] A.S. Whiteley, S. Wiles, A.K. Lilley, J. Philp, M.J. Bailey, Ecological and physiological analyses of pseudomonad species within a phenol remediation system, J. Microbiol. Methods 44 (2001) 79–88. [16] S. Belkin, D.R. Smulski, A.C. Vollmer, T.K. Van Dyk, R.A. Larossa, Oxidative stress detection with Escherichia coli, Appl. Environ. Microbiol. 62 (1996) 2252–2256. [17] G. Zhao, E. Stevens, Multiple parameters for the comprehensive evaluation of the susceptibility of Escherichia coli to the silver ion, Biometals 11 (1998) 27–32. [18] I. Sondi, B. Salopek-Sondi, Silver nanoparticles as antimicrobial agent: a case study on E. coli as a model for Gram-negative bacteria, J. Colloid Interface Sci. 275 (2004) 177–182. [19] S. Pal, Y.K. Tak, J.M. Song, Does the antibacterial activity of silver nanoparticles depend upon the shape of the nanoparticle? A study of the Gram-negative bacterium Escherichia coli, Appl. Environ. Microbiol. 73 (2007) 1712–1720. [20] W. Jiang, H. Mashayekhi, B. Xing, Bacterial toxicity comparison between nanoand micro-scaled oxide particles, Environ. Pollut. 157 (2009) 1619–1625. [21] R. Barrena, E. Casals, J. Colón, X. Font, A. Sánchez, V. Puntes, Evaluation of the ecotoxicity of model nanoparticles, Chemosphere 75 (2009) 850–857. [22] Y. Beaton, L.J. Shaw, L.A. Glover, A. Mehard, K. Killkam, Biosensing 2,4dichlorophenol toxicity during biodegradation by Burkholderia sp. RASC c2 in soil, Environ. Sci. Technol. 33 (1999) 4086–4091. [23] L.J. Shaw, Y. Beaton, L.A. Glover, K. Killkam, Development and characterization of a lux-modiﬁed 2,4-dichlorophenol-degrading Burkholderia sp. RASC, Appl. Environ. Microbiol. 1 (1999) 393–399. [24] E.M. Boyd, K. Killkam, A.A. Mehard, Toxicity of mono-, di- and tri-chlorophenols to lux marked terrestrial bacteria Burkholderia species RASC c2 and Pseudomonas ﬂuorescens, Chemosphere 43 (2001) 157–166. [25] G.M. Sinclair, G.I. Paton, A.A. Meharg, K. Killham, Lux-biosensor assessment of pH effects on microbial sorption and toxicity of chlorophenols. FEMS (Federation of European Microbiological Societies), Microbiol. Lett. 174 (1999) 273–278. [26] T.J. Brunner, P. Wick, P. Manser, P. Spohn, R.N. Grass, L.K. Limbach, A. Bruinink, W.J. Stark, In vitro cytotoxicity of oxide nanoparticles: comparison to asbestos, silica, and the effect of particle solubility, Environ. Sci. Technol. 40 (2006) 4374–4381. [27] R. Sinha, R. Karan, A. Sinha, S.K. Khare, Interaction and nanotoxic effect of ZnO and Ag nanoparticles on mesophilic and halophilic bacterial cells, Bioresour. Technol. (2010), doi:10.1016/j.biortech.2010.07.117. [28] G.A. Sotiriou, S.E. Pratsinis, Antibacterial activity of nanosilver ions and particles, Environ. Sci. Technol. 44 (2010) 5649–5654. [29] G.A. Sotiriou, A. Telekia, A. Camenzinda, F. Krumeicha, A. Meyerb, S. Pankeb, S.E. Pratsinis, Nanosilver on nanostructured silica: antibacterial activity and Ag surface area, Chem. Eng. J. (2011), doi:10.1016/j.cej.2011.01.099. [30] A.B.C. Landsdown, Silver I: its antibacterial properties and mechanism of action, J. Wound Care 11 (2002) 125–138. [31] J.J. Castellano, S.M. Shaﬁi, F. Ko, G. Donate, T.E. Wright, R. Mannari, Comparative evaluation of silver-containing antimicrobial dressings and drugs, Int. Wound J. 4 (2007) 114–122. [32] H.T. Ratte, Bioaccumulation and toxicity of silver compounds: a review, Environ. Toxicol. Chem. 18 (1999) 89–108. [33] J.C. Philp, S. Balmand, E. Hajto, M. Bailey, S. Wiles, A. Whiteley, A.K. Lilley, J. Hajto, S.A. Dunbar,.Whole cell immobilised biosensors for toxicity assessment of a wastewater treatment plant treating phenolics-containing waste, Anal. Chim. Acta 487 (2003) 61–74. [34] M.M. Shafer, J.T. Overdier, D.H. Armstong, Removal portioning and fate of silver and other metals in wastewater treatment plants and efﬂuent-receiving streams, Environ. Toxicol. Chem. 17 (4) (1998) 630–641. [35] J. Wang, C.P. Huan, D. Pirestan, Interactions of silver with wastewater constituents, Water Res. 37 (2003) 4444–4452.

Journal of Hazardous Materials 195 (2011) 73–81

Contents lists available at ScienceDirect

Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat

Utilization of different crown ethers impregnated polymeric resin for treatment of low level liquid radioactive waste by column chromatography M.F. Attallah a,∗ , E.H. Borai a , M.A. Hilal a , F.A. Shehata a , M.M. Abo-Aly b a b

Analytical Chemistry and Control Department, Hot Laboratories and Waste Management Center, Atomic Energy Authority, Post Code 13759, Abu Zaabal, Cairo, Egypt Chemistry Department, Faculty of Science, Ain Shams University, Cairo, Egypt

a r t i c l e

i n f o

Article history: Received 2 February 2011 Received in revised form 2 August 2011 Accepted 4 August 2011 Available online 25 August 2011 Keywords: Liquid radioactive waste treatment Impregnated polymeric resin Separation Radionuclides Crown ether derivatives

a b s t r a c t The main goal of this study was to ﬁnd a novel impregnated resin as an alternative for the conventional resin (KY-2 and AN-31) used for low and intermediate level liquid radioactive waste treatment. Novel impregnated ion exchangers namely, poly (acrylamide-acrylic acid-acrylonitril)N,N’-methylenedi-acrylamide-4,4’(5 )di-t-butylbenzo 18 crown 6 [P(AM-AA-AN)-DAM/DtBB18C6], poly (acrylamide-acrylic acid-acrylonitril)-N,N’-methylenediacrylamide-dibenzo 18 crown 6 [P(AM-AA-AN)DAM/DB18C6], and poly (acrylamide-acrylic acid-acrylonitril)-N,N’-methylenediacrylamide-18 crown 6 [P(AM-AA-AN)-DAM/18C6] were prepared and their removal efﬁciency of some radionuclides was investigated. Preliminary batch experiments were performed in order to study the inﬂuence of the different derivatives of 18 crown 6 on the characteristic removal performance. Separation of 134 Cs, 60 Co, 65 Zn and (152+154) Eu radionuclides from low level liquid radioactive waste was investigated by using column chromatography with P(AM-AA-AN)-DAM/DtBB18C6 and metal salt solutions traced with the corresponding radionuclides. Breakthrough data was obtained in a ﬁxed bed column at room temperature (298 K) using different bed heights and ﬂow rates. The breakthrough capacities were found to be 94.7, 83.3, 58.7, 43.1 (mg/g) for 60 Co, 65 Zn, 134 Cs, and (152+154) Eu, respectively. Pre-concentration and separation of all radionuclides under study have been carried out using different concentration of nitric and/or oxalic acids. © 2011 Elsevier B.V. All rights reserved.

1. Introduction Frequently used methods for treatment of liquid radioactive waste include chemical precipitation, evaporation, solvent extraction and ion exchange processes. Among the ion exchanger materials, impregnated polymeric resins have been used for the selective removal and separation of some radionuclides from radioactive liquid waste [1–6] as well as for the pre-concentration of metal species [7–10]. Crown ethers are effective extractants due to their ability to form stable complexes with metal ions. This property of crown ethers has led to the elaboration of new processes to extract radioactive elements from radioactive waste solutions [11–14]. Among the crown ethers, which are selective for alkali metal ions, derivatives of 21 crown 7 (21C7) have been extensively used for cesium extraction [15]. The key for the extraction is the good match between the cavity of the crown ether and the ionic radius of the metal ion. However, there are also several reports

∗ Corresponding author. Tel.: +20 2 446 20 806; fax: +20 2 446 20 784. E-mail addresses: [email protected], mohamedfathy [email protected] (M.F. Attallah). 0304-3894/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.jhazmat.2011.08.007

involving the extraction of cesium by 18-membered crown ethers (18C6) [16–21]. The new cesium-selective macrocycle calix[4]arene-bis[4-(2ethylhexyl)benzo-crown-6] (“BEHBCalixC6”) has been studied by Engel et al. The other calixcrown extractant, calix[4]arene-bis[4tert-octyl-benzo-crown-6] (“BOBCalixC6”) was used to synthesis of this new extractant “BEHBCalixC6”. It was found that replacement of the tert-ocytl alkyl chains on the benzo-crown portion of the calixcrown by 2-ethylhexyl chains improves the equilibrium solubility of the free calixcrown in aliphatic diluents, while not affecting the cesium extraction strength [3]. Development of the chromatographic partitioning of cesium and strontium utilizing two impregnated polymeric composites was also studied by Zhang et al. A novel, speciﬁc macro porous silica-based 4,4 (5 )di-t-butylcylohexano 18 crown 6 (DtBuCH18C6) chelating polymeric material was synthesized by impregnating DtBuCH18C6 molecule into Si-polymer particles that was prepared by a series of polymerization reactions. They found that, DtBuCH18C6/Si-polymer is highly selective for Sr2+ , where as DtBuCH18C6 acts as chelating agent for Sr2+ [4]. In acidic HLW, Cs(I) has been separated by a novel silica-based polymeric adsorption material, Calix[4]arene-R14/SiO2 -P, which is an excellent

74

M.F. Attallah et al. / Journal of Hazardous Materials 195 (2011) 73–81

molecular recognition reagent for Cs(I) from Sr(II) and other ﬁssion products [5]. The present work was oriented to study the effect of different derivatives of 18 crown 6 based on poly (acrylamide-acrylic acid-acrylonitrile) resin as a novel impregnated polymeric material which could be used to remove some key radionuclides from low level liquid radioactive waste. 2. Experimental 2.1. Chemicals and reagents All chemicals and reagents used in this study were of analytical grade purity and were used without further puriﬁcation. Cesium chloride and europium nitrate, were obtained from Prolabo (England). Acrylic acid, acrylonitrile purity (99%) and cobalt chloride were obtained from Merck (Germany). Acrylamide was supplied from BDH (England), 4,4 (5 )di-t-butylbenzo 18 crown 6 (DtBB18C6) were purchased from Fluka (Switzerland) while 18 Crown 6 (18C6) and dibenzo 18 Crown 6 (DB18C6) were supplied from Aldrich (USA). N,N’-methylenediacrylamide (DAM) was obtained from Aldrich (USA). Oxalic acid and sodium hydroxide were purchased from Adwic (Egypt). Nitric acid and zinc chloride were obtained from Winlab (England). A radioactive waste sample containing mixed radionuclides 134 Cs, 65 Zn, 60 Co and 152+154 Eu was collected from various laboratory research activities in Hot Laboratories Center, Egypt. 2.2. Synthesis of impregnated polymeric material (acrylamide-acrylic acid-acrylonitrile) N,N’Poly methylenediacrylamide P(AM-AA-AN)-DAM was prepared to use ␥-radiation induced template copolymerization and reported in our previous work [19,20]. In order to convert the polymer from the H+ -form to the Na+ form, the polymer (H+ ) was soaked in 0.1 M NaOH for 24 h. The solid material was separated from the solution by decantation and dried in electric oven (at ∼100 ◦ C). P(AM-AA-AN)-DAM (particle size 1.0–0.5 mm) was mixed individually with different concentration of each of 18 crown 6 (18C6), dibenzo 18 crown 6 (DB18C6) and 4,4 (5 )di-t-butylbenzo 18 crown 6 (D-t-BB18C6) that were dissolved in nitrobenzene and soaked overnight, decanted, then dried at ∼50 ◦ C for 24 h in an electric oven. The obtained three types of impregnated polymeric ion exchangers were subsequently used for batch and/or column experiments.

and 152+154 Eu radionuclides from radioactive liquid waste. Different forms of P(AM-AA-AN)-DAM such as H+ and Na+ were prepared to test the effect of counter ion on the removal efﬁciency. Furthermore, the effect of crown ether loading (10, 20 and 40%, w/v) on the polymers was studied in order to ﬁnd the optimal impregnation of the polymeric resin used for the removal of radionuclides. For this, 5 mL of radioactive liquid waste (at pH 8) was mixed with 50 mg of desired impregnated polymeric materials. The mixture was contacted on the thermostatic shaker at room temperature for 24 h to attain equilibrium. The activity concentration of the radionuclides in solution was determined radiometrically using the HPGe detector. The sorption percent (S %) of the impregnated ion exchange resin is calculated according to the following equation: S (%) =

Ci − Cf Ci

× 100

(1)

where Ci and Cf are the initial and ﬁnal counting rates per unit volume for the radionuclide, respectively; C0 is the initial concentration (mg/L) of metal ions used. 2.5. Column chromatography studies Fixed bed sorption studies were conducted to evaluate the column performance for Cs, Co, Zn and Eu ions removal on P(AMAA-AN)-DAM (Na+ )/DtBB18C6. Experiments were carried out in column of 0.8 cm inner diameter and 12.0 cm length packed with prepared P(AM-AA-AN)-DAM (Na+ )/DtBB18C6 at pH of 5.0 that was selected from our previous batch experiments [19]. Sampling of efﬂuent was done at predetermined time intervals in order to investigate the breakthrough point. The effects of inlet eluent ﬂow rate (1.0, 3.0 and 5.0 mL/min) and the resin bed height (2.0, 4.0, 6.0 and 8.0 cm) on the performance of the breakthrough curves for each ion were studied. The initial concentration of all inactive ions was kept at 100 mg/L. The break-through capacity (Q0.5 ) of the impregnated ion exchange resin is calculated according to the following equation: Breakthrough capacity (Q0.5 ) =

V(50%) × C0 m

(2)

where, V(50%) is the volume to break through at 50% uptake in L and m is the weight of the impregnated polymeric material (g). Set of experimental trials has been performed in order to elute and/or separate radionuclides that retained on the impregnated polymeric materials. In this respect different eluent reagents such as oxalic and nitric acid were used.

2.3. Instruments

2.6. Characterization of real radioactive liquid waste

The impregnated polymers were investigated using a FT-IR spectrometer (Bomen, Hartman & Braun, and model MB-157, Canada). The sample was ground into ﬁne powder and dried to eliminate the moisture content. Representative amount of the impregnated polymer (2.0 mg) was then mixed with (98.0 mg) of potassium bromide (KBr). The mixture was compressed into the disc of 5 mm diameter and 1 mm thickness. The IR spectra of the prepared disc was then measured and recorded. Measurements of the gamma radioactivity of the different radionuclides in the samples were carried out using a non-destructive ␥-ray spectroscopic technique with high purity germanium (HPGe) detector model 2201-Oxford (USA).

Two types of liquid wastes were collected. The ﬁrst waste sample, including mono-radionuclide (137 Cs only) was collected from the storage tank in Egyptian plant for treatment of radioactive liquid waste. While the second waste sample, including mixed radionuclides 134 Cs, 65 Zn, 60 Co and 152+154 Eu was collected from various laboratory research activities in Hot Laboratories Center, located at Abu Zaable site, Cairo, Egypt. Characterization of liquid radioactive waste used in this work has been done in our previous work [19,22], and are reported in Table 1. 3. Results and discussion

2.4. Batch experiments

3.1. Sorption percentage of radionuclides using different derivatives of 18 crown 6 based on polymeric resin

Preliminary batch experiments were performed to investigate the efﬁciency of different crown ethers impregnated into P(AMAA-AN)-DAM particles towards the removal of 134 Cs, 65 Zn, 60 Co

In order to investigate the sorption of some hazardous radionuclides from radioactive liquid waste, various types of impregnated polymeric resin were prepared based on some derivatives of 18

M.F. Attallah et al. / Journal of Hazardous Materials 195 (2011) 73–81

75

Fig. 1. Structure formula of 18 crown 6 (18C6), dibenzo 18 crown 6 (DB18C6) and 4,4 (5 )di-t-butylbenzo 18 crown 6 (DtBB18C6).

Table 1 Some chemical and radiochemical properties of liquid radioactive wastes [19,22]. Name

Individual radionuclide waste

pH 8 1844 TDS (mg/L) 6460 Conductivity (␮S/cm) Anions and cations (mg/L) − 199.0 Cl 443.0 NO3 − SO4 2− 1063 40.0 PO4 3− 33.5 Li+ 8.50 Na+ 2+ 9.0 Ca 2+ 25.0 Pb 25.0 Zn2+ ND Cu2+ ND Ni2+ Activity concentration (Bq/L) 137 Cs 25,397 ± 715 134 Cs ND 60 Co ND 152 Eu ND 154 Eu ND 65 Zn ND

Mixed radionuclide waste 7.5 809 1716 400.0 90.0 200.0 75.0 20.0 ND ND ND 12.5 2.50 15.0 ND 128,549 ± 1399 8320 ± 971 71,921 ± 2667 9707 ± 489 5175 ± 200

ND: not detected.

crown 6, including 18 crown 6 (18C6), dibenzo 18 crown 6 (DB18C6) and 4,4 (5 )di-t-butylbenzo 18 crown 6 (DtBB18C6) that were impregnated individually into P(AM-AA-AN)-DAM polymeric resin. Structural formula of 18C6, DB18C6 and DtBB18C6 are presented in Fig. 1. The obtained sorption percentages have been calculated and reported in Table 2. It appeared that the P(AM-AA-AN)-DAM poly-

meric resin in the Na+ form was more efﬁcient for the sorption of the radionuclides than in the H+ form. Signiﬁcant enhancement from 9, 13.3, 33 and 56.3% to 55.1, 54, 87.6 and 88.4% was obtained for the sorption of 137 Cs, 134 Cs, 60 Co and 65 Zn, respectively. This is attributed to ion exchange process that takes place for the radionuclides with Na+ ions more favorably than with H+ ions. This characteristic behavior was reported in other investigations. In extraction chromatography, many authors activate the ion exchange resin by NaCl solution to improve the uptake percentage especially for monovalent cations [19,20,22–26]. For example, Borai et al. [24] showed this idea for the impregnated zeolite materials that used for Cs removal. Based on their results, they demonstrated that the distribution coefﬁcients and the corresponding uptake percentages of Cs-134 are highly affected (decreases) by potassium rather than sodium ions in the waste solution. This may be due to the close similarity in ionic radii between Cs+ and K+ rather than that between Cs+ and Na+ . Therefore, K+ ion could compete more with Cs+ ion during the sorption process. This ﬁnding has a typical explanation to our results that showed signiﬁcant improvement in the uptake percentage of Cs ion due to the activation of the resin with sodium rather than H+ form. This phenomena is clear in uni-univalent cations exchange rather than divalent cases. Therefore, interesting high uptake values for Co(II) and Zn(II) with P(AM-AA-AN)-DAM were obtained even without impregnation. Better sorption of radionuclides on Na+ form was attributed to the consistence ionic radius of Na+ ion with the radionuclides rather than H+ ion [19,20,23]. Total dissolved salts and electric conductivity were found to be 1844 and 809 mg/L and 6460 and 1716 ␮S/cm (as reported in Table 1) for individual 137 Cs and mixed radionuclides radioactive

Table 2 Uptake percentage of some radionuclides from LLLRW using different crown ether derivatives based on P(AM-AA-AN)-DAM ion exchanger. Crown ether

a

+

R(H ) R(Na+ ) R(H+ ) + 10% 18C6 R(H+ ) + 20% 18C6 R(H+ ) + 40% 18C6 R(H+ ) + 10% DB18C6 R(H+ ) + 20% DB18C6 R(H+ ) + 40% DB18C6 R(H+ ) + 10% DtBB18C6 R(H+ ) + 20%DtBB18C6 R(H+ ) + 40% DtBB18C6 R(Na+ ) + 10% DtBB18C6 R(Na+ ) + 20% DtBB18C6 R(Na+ ) + 40% DtBB18C6

Mono

Mixed radionuclide

137

134

Cs

9.0 55.1 10.0 18.8 17.0 11.8 18.7 29.2 56 53.6 54.7 82.7 82.5 83.2

Cs

13.3 54 8.9 15.1 9.5 8.9 25.8 17.4 32.8 32.0 41.4 81.6 82.0 82.7

Experimental condition: real waste samples, contact time 24 h at room temperature. a R: means resin (P(AM-AA-AN)-DAM).

60

Co

33 87.6 11.7 23.2 10.5 10.0 16.2 24.6 33.1 33.3 38.4 84.7 85.3 86.7

65

Zn

56.3 88.4 45.4 62.5 55.7 54.1 44.3 45.1 65.2 68.2 66.7 81.1 82.2 82.5

152+154

63.3 42.3 72.2 85.8 77.3 83.9 91.6 84.7 89 93.4 92.5 83.8 80.7 84.3

Eu

76

M.F. Attallah et al. / Journal of Hazardous Materials 195 (2011) 73–81

Table 3 Distribution behavior of hazardous radionuclide using different crown ether. Impregnated resin or extractant Impregnated resin SiO2 -P/DtBuCH18C6 PVA/DCH18C6 P(AM-AA-AN)-DAM/DtBB18C6

Extractant DB18C6 DAB18C6 DHB18C6 DNB18C6 DtBB18C6 DtBB18C6 BEHBCalixC6 BOBCalixC6 BOBCalixC6 Calix[4]arene-BC6 a

Radionuclide

Distribution coefﬁcient (mL/g)

References

Sr Cs Sr Cs Co Zn Eu

946 <3 30–45 457 549 407 354

Zhang et al. [30] Zhang et al. [30] Zakurdaeva et al. [31] The present work The present work The present work The present work

Cs Cs Cs Cs Cs Cs Cs Cs Cs Cs

1.8a 1.17a 0.78a 0.8a 3.34a <0.01–2.6a 1.78–33.4a 1.22–35.3a 10.32–16.35a 10–40a

Kumar et al. [17] Kumar et al. [17] Kumar et al. [17] Kumar et al. [17] Kumar et al. [17] Mohapatra et al. [21] Engle et al. [3] Engle et al. [3] Delmau et al. [32] Gorbunova et al. [33]

This value is distribution ratio.

Table 4 Breakthrough capacity of 134 Cs, 60 Co, 65 Zn and (152+154) Eu sorbed onto P(AM-AA-AN)-DAM (Na+ )/DtBB18C6 at different process parameters.a Radionuclide

Bed height, cm

134

2 4 6 8 2 4 6 8 2 4 6 8 2 4 6 8

Cs

60

Co

65

Zn

(152+154)

a

Eu

Breakthrough capacity, mg/g 1 mL/min

3 mL/min

5 mL/min

– 58.70 – – – 94.67 – – – 83.34 – – – 43.09 – –

– 39.11 – – – 59.73 – – – 53.10 – – – 29.37 – –

11.92 20.22 28.01 36.88 14.75 25.14 36.13 45.02 14.06 23.00 31.90 40.20 8.69 15.91 22.10 31.76

Experimental condition: real waste samples pH 5, initial concentration 100 mg/L.

Fig. 2. Comparison of IR spectrum of (a) P(AM-AA-AN)-DAM and (b) P(AM-AA-AN)-DAM-(Dt-BB18C6) as impregnated polymeric material.

1.0

1.0

0.8

0.8

+

0.6

Cs

0.4

2.0 cm 4.0 cm 6.0 cm 8.0 cm

0.2

0.0

0

C/Co

C/Co

M.F. Attallah et al. / Journal of Hazardous Materials 195 (2011) 73–81

100 200 300 400 500 600 700 800 900 1000

77

2+

Co

0.6

0.4

2.0 cm 4.0 cm 6.0 cm 8.0 cm

0.2

0.0

0

200

400

600

800

1000

Effluent Volume. mL

Effluent Volume, mL

1.0

1.0

0.8

0.8 3+

Eu

2+

0.6

C / Co

C / Co

Zn

0.6

0.4

0.4

2.0 cm 4.0 cm 6.0 cm 8.0 cm

0.2

0.0 0

1200

0.2

0.0 100 200 300 400 500 600 700 800 900 1000 1100

Effluent volume, mL

2.0 cm 4.0 cm 6.0 cm 8.0 cm

0

100

200

300

400

500

600

700

800

900 1000

Effluent volume, mL

Fig. 3. Breakthrough curve of Cs+ , Co2+ , Zn2+ and Eu3+ sorbed onto P(AM-AA-AN)-DAM (Na+ )/DtBB18C6 at different bed height and at 5.0 mL/min ﬂow rate, pH 5, initial concentration (100 mg/L).

liquid wastes, respectively. Table 2, shows signiﬁcant variation in the removal % of 137 Cs and 134 Cs at the same impregnated resin. This variation of results was attributed to the high TDS in the individual radioactive waste solution, leads to various exchange potentials due to the competition between non-radioactive ions and the radioactive species during sorption of radionuclides by the adsorption or ion exchange process. Clearly, the impregnation process of both derivatives 18C6 and DB18C6 demonstrated insigniﬁcant improvement of the sorption efﬁciency of different radionuclides under study, except in case of (152+154) Eu. It was found that maximum sorption of (152+154) Eu reached to 85.8 and 91.6% at 20% loading of each of 18C6 and DB18C6, respectively. This may be due to their low afﬁnity to the other radionuclides. Moreover, the impregnation of these two derivatives may be blocked the active sites in the polymeric resins, and therefore, decrease the sorption percent of radionuclides.

On the other hand, it was found that P(AM-AA-AN)-DAM (Na+ )/DtBB18C6 gave a higher sorption percent for all radionuclides than the other derivatives of crown ether. This is likely attributed to the substituted di-tertiary butyl groups that attached to the crown ether rings. This could be probably due to the greater electron withdrawing ability of these groups when connected via the benzo group. The electron donating ability of the tertiary butyl group in DtBB18C6 has a positive inductive effect (+I) which helps in increasing the electron density on the oxygen ‘O’ atoms [17,19]. Moreover, ring cavity size of 18C6 is 2.6–3.2 A˚ [27], is large enough to allow the entry of any cation. However, the ligand is more selective for the ˚ to almost Cs+ . Cesium has the proper dimension (diameter ∼3.40 A) ﬁt in the ring cavity of 18C6, which favors the participation of all oxygen atoms of the macro ring cavity on the coordination, and leads to a more favorable stabilization [28]. The ionic radii of the other metals are too small comparing to the cavity of 18C6, therefore less number of oxygen atoms of 18C6 cavity coordinate with

78

M.F. Attallah et al. / Journal of Hazardous Materials 195 (2011) 73–81

the metal ion [28,29]. More efﬁcient interaction between metal ions and the dipole donor (O) atoms of 18C6 may takes place via ion–dipole mechanism. The mechanism of interaction of Co(II), Zn(II) and Eu(III) with the impregnated polymeric resin may be taken place cationic exchange between Na+ of the carboxylic group of AA (–COONa) and metal ion. This is agree with the previous ﬁnding of interaction of polymeric materials such as P(AANa), P(AM-AA), P(AM-AANaDAEA-HCl), P(AM-AA) with metal. Furthermore, insigniﬁcant improvement of the sorption percentage was obtained by an increase the concentration of 18C6, DB18C6 and DtBB18C6 impregnated from 10 to 40%. The sorption percent of radionuclides using impregnated polymeric resin follows the order: P(AM-AA-AN)-DAM (Na+ )/ DtBB18C6 > P(AM-AA-AN)-DAM (H+ ) / DtBB18C6 > P(AM-AA-AN)-DAM (H+ )/ DB18C6 > P(AM-AA-AN)-DAM (H+ )/ 18C6 Based on these results, P(AM-AA-AN)-DAM (Na+ )/DtBB18C6 (10%) as impregnated ion exchanger was selected for the subsequent investigations. The obtained distribution coefﬁcients of radionuclides under study have been compared with different impregnated crown ethers, derivative of crown ether and calix crown ether as extractant. It was observed the impregnated materials with different crown ether that used to remove and separate radionuclides are limited in literature. As shown in Table 3, the prepared resin (P(AM-AA-AN)-DAM/DtBB18C6 provided signiﬁcantly high distribution coefﬁcient for Cs radionuclides compared to the other derivative crown ether and calixarene.

sites for sorption of metal ions beside various factional groups in the polymer support materials. 3.3. Chromatographic column studies The operation and performance of a column are known to be inﬂuenced by it a number of parameters such as type, concentration and ﬂow rate of the feed solution as well as column bed height. In this respect optimization of some variables is essential to evaluate the column performance. Fixed bed column experiments were carried out to study the sorption dynamics. The shape of the breakthrough curve and the time for the breakthrough appearance are the predominant factors for determining the operation and the dynamic response of the sorption column. The general position of the breakthrough curve along the volume/time axis depends on the capacity of column with respect to bed height, the feed concentration and the ﬂow rate [40–43]. In this concern different ﬂow rates as well as various bed heights were tested at ﬁxed initial ion concentration of 100 mg/L for all ions under investigation. 3.3.1. Effect of bed height As shown in Fig. 3 (see also Table 4), the breakthrough capacity (Q0.5 ), breakthrough time was increased with increasing bed height. The increase in the ion sorption with bed height was due to the increase of the sorbent mass in larger beds, which provide greater sorption sites for the metal ions. The obtained results are agree with the same trend by other authors [19,43–45]. Based on the obtained result it could be found that breakthrough capacity (Q0.5 ) obeyed the following sequence at the same corresponding bed height: Co > Zn > Cs > Eu

3.2. Impregnated polymer structure The infrared spectra of the polymeric material and impregnated polymeric material (P(AM-AA-AN)-DAM (Na+ )/DtBB18C6) show (Fig. 2) that there are many vibrationally absorption bands, characterized mainly to carboxylate, carboxylic, ester, ether and nitrile groups. The broad absorption band at ∼3443, 3427 cm−1 is characterized to stretching vibrations of CONH2 related to amide group content in polymeric resin and impregnated polymeric material. This band was conﬁrmed by the appearance of another band at 1520 cm−1 . Moreover, there is a strong absorption band at ∼2922 and 2875 cm−1 , attributed to stretching vibrations of CH2 group, which is conﬁrmed by another band at 1076, 1165, and 1146 cm−1 . Two characteristic absorption bands at ∼1428, 1418 cm−1 are related to carboxylate group as well as absorption bands at ∼1710, 1428, 1418, 962 cm−1 are attributed to the carboxylic group. The absorption bands at 2242, 2240 cm−1 , are due to the nitrile group, as well as a band at 1655, 1597 cm−1 are attributed to C O bond. The new other two bands at 2957, 1366 cm−1 are attributed to t-butyl in impregnated polymeric material as well as a band at 1459 cm−1 is characterized to nitro-aromatic group [34–37]. The presence of the carboxylate and ester groups in P(AM-AAAN)-DAM indicated the interaction of DAM with carboxylic groups of acrylic acid of the polymeric chain. It was found that DAM acts as a crosslinker in the polymerization of acrylamide, acrylamide–acrylic acid and acrylic acid–acrylonitrile [38,39]. This implies the presence of acrylamide, acrylic acid, acrylonitrile and ether units in the impregnated polymeric chains, as shown in Fig. 2b. The spectroscopy revealed that the resin, including DAM is linked between the polymeric chains according to the mechanism for the template copolymerizeation of AA–AN on P(AM) in the presence of DAM while DtBB18C6 may be linked to the polymeric chains according to hydrogen bond [19,39]. Crown ethers were providing the active

3.3.2. Effect of ﬂow rate The effect of ﬂow rate on 134 Cs, 60 Co, 65 Zn and (152+154) Eu sorption by P(AM-AA-AN)-DAM (Na+ )/DtBB18C6 was studied by varying the ﬂow rate for 1.0, 3.0 and 5.0 mL/min at the ﬁxed bed height (4.0 cm) and initial concentrations (100 mg/L) for all ions under study. The plots of the breakthrough curves of 134 Cs, 60 Co, 65 Zn and (152+154) Eu at various ﬂow rates are shown in Fig. 4. As shown from Fig. 4, an increase in ﬂow rate reduces the efﬂuent breakthrough volume and thereby decreases the retention time of the elements. This is due to the decrease in the residence time of the 134 Cs, 60 Co, 65 Zn and (152+154) Eu within the bed at higher ﬂow rates. Much sharper breakthrough curves for 134 Cs, 60 Co, 65 Zn and (152+154) Eu sorption onto P(AM-AA-AN)-DAM (Na+ )/DtBB18C6 were obtained at higher ﬂow rates. The breakthrough time and the amount of total 134 Cs, 60 Co, 65 Zn and (152+154) Eu sorbed also decreased with increasing ﬂow rate, as presented in Table 4. This is attributed to the reduced contact time causing a weak distribution of the liquid inside the column, which leads to a lower diffusively of the solute among the particle of the P(AM-AA-AN)DAM (Na+ )/DtBB18C6 [41]. 3.4. Separation of 134 Cs, 60 Co, 65 Zn and (152+154) Eu from radioactive liquid waste Based on the previous results, removal and separation of 134 Cs, and (152+154) Eu radionuclides from low level liquid radioactive waste was investigated using column containing P(AMAA-AN)-DAM (Na+ )/DtBB18C6 at ﬂow rate 3.0 mL/min and 4.0 cm bed height. The loading process was carried out by passing an appropriate volume of the radioactive waste solution. Some set of an experiment were preferred for removal and separation process towards 134 Cs, 60 Co, 65 Zn and (152+154) Eu radionuclides using

60 Co, 65 Zn

M.F. Attallah et al. / Journal of Hazardous Materials 195 (2011) 73–81

1.0

1.0

0.8

0.8

79

0.6

2+

0.4

C / Co

C / Co

+

Cs

Co

0.6

5.0 mL 3.0 mL 1.0 mL

0.4

5.0 mL 3.0 mL 1.0 mL

0.2

0.2

0.0 0

200

400

600

800

1000

1200

1400

0.0

0

200 400 600 800 1000 1200 1400 1600 1800

Effluent volume, mL

Effluent volume, mL

1.0

1.0

0.8

0.8 2+

0.6 5.0 mL 3.0 mL 1.0 mL

0.4

3+

Eu

0.6

5.0 mL 3.0 mL 1.0 mL

0.4

0.2

0.2

0.0

C / Co

C / Co

Zn

0

200 400 600 800 1000 1200 1400 1600 1800

0.0

0

100 200 300 400 500 600 700 800 900

Effluent volume, mL

Effluent volume, mL

Fig. 4. Breakthrough curve of Cs+ , Co2+ , Zn2+ and Eu3+ sorbed onto P(AM-AA-AN)-DAM (Na+ )/DtBB18C6 at different ﬂow rate and at ﬁxed bed height (4.0 cm) as well as initial concentration (100 mg/L) at pH 5.

80

Concentration, (mg/L)

nitric and oxalic acids as eluent reagents. Separation and removal of radionuclides under study are presented in Figs. 5–7. Different concentrations of nitric acid such as 0.1, 0.5 mol/L, and 0.12 mol/L of oxalic acid at pH 4.5 were investigated for removal and/or separation process. As shown in Fig. 5, 0.5 M of nitric acid was used as eluent for removal and separation process. It was found that preconcentration and removal of 134 Cs, 60 Co, 65 Zn and (152+154) Eu radionuclides was done by 60 mL with recovery percent >98%, as presented in Table 5. It can be inferred that 0.5 mol/L nitric acid is a good eluent for preconcentration and removal of all radionuclides under study, but it is not capable of the separation of radionuclides from each other. The second trial was carried out using 0.1 mol/L nitric acid as eluent as shown in Fig. 6. Separation of 134 Cs from 60 Co, 65 Zn and (152+154) Eu radionuclides was obtained by 50 mL, with recovery percent 98%. No release of any other radionuclides on elution by 0.1 mol/L nitric acid took place. Therefore, higher concentration of nitric acid (0.5 mol/L) was applied to elute 60 Co, 65 Zn and (152+154) Eu radionuclides.

0.5M HNO3 Cs Co Zn Eu

60

40

20

0

0

20

40

60

80

100

Effluent volume, mL Fig. 5. Elution curves of 134 Cs, 3.0 mL/min, 4.0 cm bed height.

60

Co,

65

Zn and

(152+154)

Eu by 0.5 mol/L nitric acid at

80

M.F. Attallah et al. / Journal of Hazardous Materials 195 (2011) 73–81

//

0.1M HNO 3

0.5M HNO 3

60 Cs Co Zn Eu

50

Concentration, (mg/L)

40

30

20

10

0

0

20

40

60

80

100

120

140

160

Effluent volume, mL Fig. 6. Gradient separation and removal of 4.0 cm bed height.

134

Cs,

60

Co,

65

Zn and

(152+154)

Eu from radioactive liquid waste using 0.1 mol/L followed by 0.5 mol/L nitric acid at 3.0 mL/min,

Table 5 Recovery percent of 134 Cs, 60 Co, 65 Zn and (152+154) Eu using different eluent. Eluent

Recovery, % Efﬂuent volume, mL 134

0.12 M Oxalic acid (Fig. 7) 0.1 M nitric acid (Fig. 6) 0.5 M nitric acid (Fig. 6)

60

Cs

99 (60 mL) 98.3 (50 mL) 98.7 (50 mL)

65

Co

96.8 (70 mL) – 97.6 (50 mL)

The third trial was carried out using 0.12 mol/L of oxalic acid at pH 4.5, and it appeared that 0.12 mol/L of oxalic acid is highly efﬁcient eluent for the preconcentration of both 134 Cs and 60 Co from 65 Zn and (152+154) Eu within about the ﬁrst 50 mL. On the 0.12 M of oxalic acid

(152+154)

Zn

90 (80 mL) – 99 (60 mL)

Eu

– – 99.5 (60 mL)

other hand, 65 Zn was separated from (152+154) Eu within the second 50 mL, while (152+154) Eu is not eluted by oxalic acid. Then the last stage was performed for the separation of (152+154) Eu successfully by gradient elution of 120 mL of 0.5 mol/L nitric acid as 0.5 M of nitric acid

//

60 Cs Co Zn Eu

Concentration, (mg/L)

50

40

30

20

10

0

0

20

40

60

80

10 0

120

140

16 0

18 0

200

220

240

Effluent Volume, mL Fig. 7. Gradient separation and removal of 134 Cs, 60 Co, 65 Zn and (152+154) Eu from radioactive liquid waste using 0.12 mol/L of oxalic acid at pH 4.5 followed by 0.5 mol/L nitric acid at 3.0 mL/min, 4.0 cm bed height.

M.F. Attallah et al. / Journal of Hazardous Materials 195 (2011) 73–81

depicted in Fig. 7. The recovery percent of 134 Cs, 60 Co, and 65 Zn were 99, 96.8, and 90% using 0.12 mol/L of oxalic acid as well as 99.5% for (152 + 154) Eu using 0.5 mol/L of nitric acid, as shown in Table 5. 4. Conclusions Impregnated polymeric materials, namely P(AM-AA-AN)DAM/D-t-BB18C6, P(AM-AA-AN)-DAM/DB18C6 and P(AM-AAAN)-DAM/18C6 were prepared and the removal properties for some hazardous radionuclides from radioactive liquid waste were investigated. P(AM-AA-AN)-DAM/D-t-BB18C6 exhibits promising high sorption characteristics for the removal of 137 Cs and other mixed radionuclides such as 60 Co, 65 Zn and (152+154) Eu. Therefore, it can be used as an alternative sorbent material to the Egyptian plant for treatment of radioactive liquid waste. Separation and removal of some hazardous ﬁssion products were successfully applied from low level liquid radioactive waste (LLLRW) containing mixed radionuclides (134 Cs, 65 Zn, 60 Co and 152+154 Eu) using extraction column chromatography packed with P(AM-AAAN)-DAM/D-t-BB18C6. The separation process was done with high recovery by gradient elution of nitric and oxalic acids. References [1] S.A. Ansari, P.N. Pathak, M. Husain, A.K. Prasad, V.S. Parmar, V.K. Manchanada, Talanta 68 (2006) 1273. [2] B.S. Shaibu, M.L.P. Reddy, A. Bhattacharyya, V.K. Manchanda, J. Magn. Magn. Mater. 301 (2006) 312. [3] N.L. Engel, P.V. Bonnesen, B.A. Tomkins, T.J. Haverlock, B.A. Moyer, Solv. Extr. Ion Exch. 22 (4) (2004) 611. [4] A. Zhang, E. Kuraoka, H. Hoshi, M. Kumagai, J. Chromatogr. A 1061 (2004) 175. [5] A. Zhang, E. Kuraoka, M. Kumagai, Sep. Purif. Technol. 50 (2006) 35. [6] A. Zhang, E. Kuraoka, M. Kumagai, J. Chromatogr. A 1157 (2007) 85. [7] T. Honjo, H. Kitayama, K. Terada, T. Kiba, Fresenius Z. Anal. Chem. 330 (1988) 159. [8] A.K. Kostad, P.Y.T. Chow, F.F. Cantwell, Anal. Chem. 60 (1988) 1569. [9] J.P. Bernal, E. Rodriguez De San Miguel, J.C. Aguilar, G. Salazar, J. De Gyves, Sep. Sci. Technol. 35 (10) (2000) 1661. [10] K.A.K. Ebraheem, M.S. Mubarak, Z.J. Yassien, F. Khalil, Sep. Sci. Technol. 35 (13) (2000) 2115. [11] I.H. Gerow, J.E. Smith Jr., M.W. Davis Jr., Sep. Sci. Technol. 16 (1981) 519. [12] K.L. Nas, Solv. Extr. Ion Exch. 11 (1993) 729. [13] V.S. Talanov, G.G. Talanova, M.G. Gorbunova, R.A. Bartsc, J. Chem. Soc., Prekin Trans. 2 (2002) 209. [14] V.V. Yakshin, V.I. Zhilov, S.V. Demin, G.A. Pribylova, I.G. Tananaev, A.Y. Tsivadze, B.F. Myasoedov, C. R. Chim. 10 (2007) 1020.

81

[15] Y. Takeda, Topics Curr. Chem. 121 (1984) 1. [16] M.L. Dietz, E.P. Horwitz, M.P. Jensen, S. Rhoads, R.A. Bartsch, A. Palka, J. Krzykawski, J. Nam, Solv. Extr. Ion Exch. 14 (1996) 357. [17] A. Kumar, P.K. Mohapatra, V.K. Manchanda, J. Radioanal. Nucl. Chem. 229 (1998) 169. [18] Y. Kikuchi, Y. Sakamoto, Anal. Chim. Acta 403 (2000) 325. [19] M.F. Attallah, Radiochemical studies on separation and removal of some radionuclides from radioactive wastes, Ph.D. Thesis, Chemistry Department, Faculty of Science, Ain Shams University, Cairo, Egypt, 2009. [20] M.F. Attallah, E.H. Borai, K.F. Allan, Radiochemistry 51 (6) (2009) 622. [21] P.K. Mohapatra, D.S. Lakshmi, D. Mohan, V.K. Manchanda, J. Membr. Sci. 232 (2004) 133. [22] E.H. Borai, M.A. Hilal, M.F. Attallah, F.A. Shehata, Radiochim. Acta 96 (2008) 441. [23] F.A. Shehata, M.F. Attallah, E.H. Borai, M.A. Hilal, M.M. Abo-Aly, Appl. Radiat. Isot. 68 (2010) 239. [24] E.H. Borai, R. Harjula, L. Malinen, A. Paajanen, J. Hazard. Mater. 172 (2009) 416. [25] T. Siyam, M.M. Abdel-Hamid, I.M.M. El-Naggar, Macromol. Rep. A 32 (Suppls. 5 & 6) (1995) 871. [26] T. Siyam, I.M. El-Naggar, H.F. Aly, Intern. Topical Meeting on Nuclear and Hazards Waste Management Spectrum 96, 18–23 August, Seatle, Washington, 1996, p. 66. [27] T.M. Letcher, J.D. Mercer-Chalmers, R.L. Kay, Pure Appl. Chem. 66 (3) (1994) 419. [28] J.M. Caridade Costaa, P.M.S. Rodrigues, Port. Electrochim. Acta 20 (2002) 167. [29] H.J. Buschmann, H. Dong, E. Schollmeyer, J. Coord. Chem. 30 (1993) 311. [30] A. Zhang, Y.-Z. Wei, M. Kumagai, T. Koyama, J. Radioanal. Nucl. Chem. 262 (3) (2004) 739. [31] O.A. Zakurdaeva, S.V. Nesterov, N.A. Shmakova, G.K. Semenova, E.O. Sozontova, V.I. Feldman, Nucl. Instr. Meth. Phys. Res. B 265 (2007) 356. [32] L.H. Delmau, P.V. Bonnesen, B.A. Moyer, Hydrometallurgy 72 (2004) 9. [33] M.G. Gorbunova, P.V. Bonnesen, N.L. Engle, E. Bazelaire, L.H. Delmau, B.A. Moyer, Tetrahedron Lett. 44 (2003) 5397. [34] J.A. Dean, Lange’s Handbook of Chemistry, thirteenth ed., McGraw-Hill Inc., USA, 1985. [35] B. Staurt, Modern Infrared Spectroscopy, John Wiley and Sons, Ltd, West Sussex, PO 19 IUD, England, 1996. [36] R.A. Nyquist, R.O. Kagel, Infrared and Raman Spectra of Inorganic Compounds and Organic Salts, Academic Press Inc, 1997. [37] K. Nakamoto, Infrared and Raman Spectra of Inorganic and Coordination Compounds, John Wiley and Sons Publications, New York, 1978. [38] K.F. Allan, T. Siyam, W.A. Sanad, 6th Arab Intern. Conf. on Polym. Sci. and Technology, 1–5 September, Sharm El Sheikh-Sinaa, Egypt, 2001, p. 121. [39] K.F., Allan, Gamma radiation-induced preparation of some polymeric resins and their use for the treatment of waste water, Ph.D. Thesis, Chemistry Department, Faculty of Science, Suez Canal University, Egypt, 2004. [40] S. Netpradit, P. Thiravetyan, S. Towprayoon, Water Res. 38 (2004) 71. [41] T.S. Singh, K.K. Pant, Sep. Purif. Technol. 48 (2006) 288. [42] E. Malkoc, Y. Nuhoglu, J. Hazard. Mater. 135 (2006) 328. [43] M. Sarkar, A.R. Sarkar, J.L. Goswami, J. Hazard. Mater. 149 (2007) 666. [44] S.S. Metwally, Ion exchange characteristics of poly-acrylamide and polyacrylonitrile based Ce (IV) phosphate for removal and separation of some radioactive nuclides from waste solutions, Ph.D. Thesis, Chemistry Department, Faculty of Science, Ain Shams University, 2008. [45] A.M. El-Kamash, J. Hazard. Mater. 151 (2008) 432.

Journal of Hazardous Materials 195 (2011) 82–91

Contents lists available at ScienceDirect

Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat

Characteristics and source apportionment of PM1 emissions at a roadside station Y. Cheng a,b,c,∗ , S.C. Zou d , S.C. Lee c , J.C. Chow a,b,e , K.F. Ho c , J.G. Watson a,b,e , Y.M. Han b , R.J. Zhang f , F. Zhang a , P.S. Yau c , Y. Huang c , Y. Bai a , W.J. Wu a a Department of Environmental Science and Technology, School of Human Settlements and Civil Engineering, Xi’an Jiaotong University, No.28 Xianning West Road, Xi’an, Shaanxi, 710049, China b SKLLQG, Institute of Earth and Environment, CAS, Xi’an, Shaanxi, 710075, China c Department of Civil and Structural Engineering, Research Center for Environmental Technology and Management, The Hong Kong Polytechnic University, Hung Hom, Kowloon, Hong Kong d School of Marine Sciences, Sun Yat-sen University, Guangzhou 510275, China e Division of Atmospheric Sciences, Desert Research Institute, Reno, NV, USA f Key Laboratory of Regional Climate-Environment Research for Temperate East Asia, Institute of Atmospheric Physics, Chinese Academy of Sciences, Beijing, 100029, China

a r t i c l e

i n f o

Article history: Received 22 March 2011 Received in revised form 4 August 2011 Accepted 4 August 2011 Available online 25 August 2011 Keywords: PM1 Chemical composition PMF

a b s t r a c t The mass concentrations of PM1 (particles less than 1.0 ␮m in aerodynamic diameter), organic carbon (OC), elemental carbon (EC), water-soluble ions, and up to 25 elements were reported for 24 h aerosol samples collected every sixth day at a roadside sampling station in Hong Kong from October 2004 to September 2005. Annual average PM1 mass concentration was 44.5 ± 19.5 ␮g m−3 . EC, OM (organic matter, OC × 1.2), and SO4 = were the dominant components, accounting for ∼36%, ∼26%, and ∼24% of PM1 , respectively. Other components, i.e., NO3 − , NH4 + , geological material, trace elements and unidentiﬁed material, comprised the remaining ∼14%. Annual average OC/EC ratio (0.6 ± 0.3) was low, indicating that primary vehicle exhaust was the major source of carbonaceous aerosols. The seasonal variations of pollutants were due to gas-particle partitioning processes or a change in air mass rather than secondary aerosol produced locally. Vehicle exhaust, secondary aerosols, and waste incinerator/biomass burning were dominant air pollution sources, accounting for ∼38%, ∼22% and ∼16% of PM1 , respectively. Pollution episodes during summer (May–August) which were frequently accompanied by tropical storms or typhoons were dominated by vehicle emissions. During winter (November–February) pollution episodes coincided with northeasterly monsoons were characterized by secondary aerosols and incinerator/biomass burning emissions. © 2011 Elsevier B.V. All rights reserved.

1. Introduction Adverse health effects of exposure to particles have been proven in numerous toxicological and epidemiological studies [1–5]. Fine particles, like PM1 (particles with an aerodynamic diameter of less than 1.0 ␮m), are those most harmful to human beings, as they are able to penetrate into the human respiratory and circulation system, resulting in adverse health effects [4,6]. The mechanism of these adverse health effects is unclear; however, previous research indicates that toxic elements and compounds carried in ﬁne particles may play an important role. About 70–80% of toxic trace

∗ Corresponding author at: Department of Environmental Science and Technology, School of Human Settlements and Civil Engineering, Xi’an Jiaotong University, No.28 Xianning West Road, Xi’an, Shaanxi, 710049, China. Tel.: +86 29 83395078; fax: +86 29 83395078. E-mail address: [email protected] (Y. Cheng). 0304-3894/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.jhazmat.2011.08.005

elements, like lead (Pb), cadmium (Cd), arsenic (As) and nickel (Ni), as well as Polycyclic Aromatic Hydrocarbon compounds (PAHs), were found in PM1 , with the majority of sulfur (S), vanadium (V), selenium (Se) and zinc (Zn) found in the submicron particle size range [7,8]. Fine particles, mainly arising from vehicle exhaust, comprised the majority of airborne total suspended particles (TSP) in the atmosphere of Hong Kong [9–12]. Lee et al. [11] found that PM1 constituted ∼70% of the PM2.5 mass at the Roadside Air Quality Monitoring Station on the campus of Hong Kong Polytechnic University (i.e., the PU Roadside Station). They suggested using PM1 as an indicator for vehicular emissions at the PU Roadside Station due to less inﬂuence from non-vehicle sources [11]. The concentrations of the elements Cr, Fe, Co, Cu, As, and Ba at the roadside location were a factor of two higher than those measured at the background Hok Tsui (HT) station in Hong Kong [7]. In a roadside microenvironment, increasing particles counts were found when vehicles accelerate (e.g., after stopping at a

39,214 41,268 41,747 38,066 39,955 631 543 401 538 536 ± ± ± ± ±

(m)

1040 1116 1053 1044 1069 6 7 6 3 7 ± ± ± ± ± 11 14 15 10 12 3 4 6 5 5 ± ± ± ± ± 4 5 7 5 5 10 9 16 9 10 ± ± ± ± ± 21 22 23 24 23 43 74 83 27 82 ± ± ± ± ± 67 172 92 55 102 4 55 33 2 46 81.6 83.5 72.5 76.0 78.9 20.2 28.0 26.6 18.5 23.3

± ± ± ± ±

4.2 1.6 2.0 3.6 5.2

(%) (◦ C)

± ± ± ± ±

9.0 6.3 11.4 11.4 10.6

(mm)

± ± ± ± ±

(km hr−1 )

Total bright sunshine (h) Wind speed

Prevailinig wind (◦ ) Rainfall RH

Seven days with mixing height higher than 3000 were not counter here.

The Teﬂon-membrane ﬁlters were analysed for the presence of 51 elements (from Na to U) by X-ray ﬂuorescence (XRF, Watson et al. [21]) at the Environmental Analysis Facility of the

a

2.3. Chemical analyses

Temperature

The 24 h PM1 sampling was performed once every sixth day from 8 October 2004 to 23 September 2005 and forty valid sample sets were obtained. Approximately 5% of additional ﬁeld blanks were collected for blank subtraction and error propagation. A URG3000ABC sampler (URG corporation, Chapel Hill, USA) with one PM1 inlet (Teﬂon® coated aluminum, URG corporation, Chapel Hill, USA), operated at 16.7 L/min, was used to collect samples. The sampling inlet was about 1.5 m above street level. The PM1 sampler was equipped with two parallel channels containing 47 mm Teﬂonmembrane and quartz-ﬁber ﬁlters at the ﬂow rate of 8.3 L/min for each channel. Both the Teﬂon-membrane and quartz-ﬁber ﬁlters were weighed twice before and after sampling, respectively, using a Sartorius Model MC5 Microbalance (Göttingen, Germany) with a sensitivity of ±1 ␮g in the 0–250 mg range. Before weighing, ﬁlters were equilibrated for 24 h in a desiccator at 20–30 ◦ C and a relative humidity of 30–40%. Prior to sampling quartz-ﬁber ﬁlters were preheated in an electric furnace at 900 ◦ C for 3 h to remove carbonaceous contaminants. Collected quartz ﬁber samples were stored in a refrigerator at about 4 ◦ C to prevent the evaporation of volatile components prior to chemical analysis.

Season

2.2. Sampling method

Table 1 Summary of daily meteorological parameters and vehicle numbers in four seasons from October 2004 to September 2005.

The climate in Hong Kong is sub-tropical, inﬂuenced by the Asian monsoons. The cooling of the great Asian land mass during winter and its heating during summer give rise to monsoonal winds on a very large scale, which leads to four seasons of unequal duration in Hong Kong [18]. The four seasons in this study are deﬁned as prolonged summer (May–August) and winter (November–February), and transitional, short spring (March–April) and autumn (September–October), as listed in Table 1. An examination of the historical climatology records shows that meteorological characteristics during the study period did not deviate from the norm. Daily meteorological data was obtained from the Hong Kong Observatory. PM1 samples were collected at the PU Roadside Station (22.30◦ N, 114.17◦ E), located in a residential and commercial area near Victoria Harbour. The sampling site [11] is about 1–2 m away from the curb of Hong Chong Road, which is approximately 30 m wide with four lanes for each direction leading to the busiest crossharbour tunnel in Hong Kong. During the sampling period, daily trafﬁc ﬂow remained at roughly 120,000 vehicles per day [19,20]. Trafﬁc data were obtained from the toll data maintained for Victoria Harbour.

Daily global solar radiation (MJ m−2 )

2.1. Climate and sampling location

Spring Summer Autumn Winter Average

2. Experimental method

3 35 13 1 15

Mixing heighta

Daily diesel vehicle (#)

Daily total vehicle (#)

signal light or a bus stop), especially for diesel-fueled vehicles [13]. Previous studies [e.g., 11,14,15,16,17] have provided limited information on PM1 in Hong Kong. For this study, 24 h sampling of PM1 was conducted every sixth day at the PU Roadside Station from 8 October 2004 to 23 September 2005. Study objectives were to: (1) characterize the chemical composition and seasonal variation in speciated PM1 ; (2) quantify the source contributions to PM1 by the Positive Matrix Factorization (PMF) receptor model; and (3) investigate meteorological characteristics that may affect the occurrence and strength of each air pollution source.

83

123,130 121,997 123,517 122,690 122,580

Y. Cheng et al. / Journal of Hazardous Materials 195 (2011) 82–91

84

Y. Cheng et al. / Journal of Hazardous Materials 195 (2011) 82–91

PM1 massQuartz (μg m-3 )

100

was present as soluble K+ in the atmosphere. These physical consistencies validate the effectiveness of the sampling and chemical analysis methods used in this study.

y=1.01(±0.07)x-4.76(±2.17) R=0.92 n=40

80 60

3.2. PM1 mass concentration and mass closure

40 20 0

50

0

100

PM1 massTeflon (μg m-3 ) Fig. 1. PM1 gravimetric mass concentrations from Teﬂon-membrane and quartzﬁber ﬁlters at the PU Roadside Station from October 2004 to September 2005.

Desert Research Institute (Desert Research Institute, DRI, Reno, NV, USA). Half of the quartz-ﬁber ﬁlters were extracted with distilled-deionized water and the extracts were analysed for chloride (Cl− ), nitrate (NO3 − ), sulfate (SO4 = ), water-soluble sodium (Na+ ), potassium (K+ ), calcium (Ca++ ) and ammonium (NH4 + ) ions by ion chromatography (DIONEX, USA), following the methodology described by Chow and Watson [22], in the Air Laboratory of the Hong Kong Polytechnic University. OC and EC were measured on a 0.5 cm2 quartz-ﬁber ﬁlter punch from the remaining half of the ﬁlters using a DRI Model 2001 carbon analyzer with a thermal-optical reﬂectance (TOR) method following the Interagency Monitoring of Protected Visual Environments (IMPROVE) protocol [23,24]. Field blanks were analysed for blank subtraction and error propagation. Major mass constituents, including OC, EC, SO4 = , NO3 − , NH4 + , and elements (e.g., Al, Si, Ca, Fe, and Zn), were detected in almost every sample. Concentrations of only 25 elements were reported here because elements, such as Sc, Co, Ga, Se, Y, Nb, Mo, Pd, Ag, Cd, In, Sb, Cs, La, Ce, Sm, Eu, Tb, Hf, Ta, Wo, Ir, Au, Hg, Tl, and U, seldom showed concentrations higher than three times their respective minimum detectable limits (MDLs, [25]). 3. Results and discussion 3.1. Measurement validation Because PM1 samples were acquired on two different substrates and chemical analyses were performed in different laboratories, consistency tests are needed as part of the quality assurance process. Fig. 1 shows the comparison of PM1 gravimetric mass measurements from the collocated Teﬂon-membrane and quartz-ﬁber ﬁlters. Good agreements (slope close to unity) and high correlation (R = 0.92) demonstrated the consistency of sample and gravimetric analyses. Similar comparisons were also carried out by Engelbrecht et al. [26] in the USA and Louie et al. [14] in Hong Kong for PM2.5 samples. Because quartz-ﬁber ﬁlters are known to have positive sampling artifacts due to absorption of gaseous organic compounds and water [27–31] and known to have a tendency to shred and fragment during sample handling, the following discussion refers to Teﬂon-membrane mass unless otherwise speciﬁed. Regarding different chemical analysis methods, Fig. 2 shows reasonable agreement for SO4 = versus S (R = 0.97), and K+ versus K (R = 0.98), sampled on different sampling substrates. The ratio of SO4 = to S was 2.57 with a small intercept (−0.05 ␮g m−3 ), indicating that more than ∼85% of S was present as soluble SO4 = in the atmosphere and that both XRF and IC measurements were valid. The scatter plot of K+ and K also showed a slope of 0.92 and close to zero intercept (−0.02 ␮g m−3 ) suggesting that over ∼90% of total K

The annual average PM1 mass concentration from October 2004 to September 2005 was 44.0 ± 19.4 ␮g m−3 (Table 2). This level was higher than that measured at the urban Chung Shan site in Taiwan (17.1, 13.1, 9.7 ␮g m−3 in spring, autumn and winter, respectively, [32]) and at the Virolahti background station in Finland (4.3 ± 3.8 and 3.8 ± 3.6 ␮g m−3 in summer and winter, respectively, [33]), but much lower than that (127.3 ± 62.1 ␮g m−3 ) at an urban site in Xi’an, China [34]. Mass balance measurements of PM1 showed that EC, OM (OM = OC × 1.2, [35]), and SO4 = were the major components of PM1 , accounting for ∼36%, ∼26%, and ∼24% of the PM1 mass, respectively. Low abundances were found for NO3 − (∼5%), NH4 + (∼3%), geological material and trace elements (∼3%), and unidentiﬁed material (∼3%). The statistical summary of PM1 mass concentration in Table 2 lists maximum PM1 mass concentration in winter (52.9 ± 20.1 ␮g m−3 ), followed by autumn (48.7 ± 24.8 ␮g m−3 ), spring (41.3 ± 7.5 ␮g m−3 ) and summer (34.8 ± 17.9 ␮g m−3 ). Hong Kong is located at south edge of East Asia and China. Monsoon winds exert a profound inﬂuence on the air quality of Hong Kong, as previously reported [11,14–16,36,37]. In the summer, prevailing southerly winds, with the resultant vector of 172◦ (Table 1), brought clean marine air masses to Hong Kong. In the autumn, winter, and spring, with the resultant vector of 71◦ (Table 1), prevailing northeasterly winds transported continental emissions from interior Asia to Hong Kong and the South China Sea. This explains the higher PM1 mass concentrations in winter, autumn, and spring compared to summer. Seasonal variation in mixing height, daily diesel vehicle numbers, and total vehicle numbers on the Hong Chong road (Table 1) are not signiﬁcant factors that explain the seasonality of air pollution. 3.3. Carbonaceous aerosols During the sampling period, PM1 OC ranged from 3.2 to 29.8 ␮g m−3 and EC ranged from 8.3 to 26.8 ␮g m−3 . Annual average OC and EC were 9.6 ± 4.9 and 15.8 ± 5.1 ␮g m−3 , respectively. Compared to one of the most polluted inland cities (Xi’an) in China, average OC from this study was a factor of two lower than OC (21.0 ␮g m−3 ) reported by Shen et al. [34], and average EC was three times higher than in Xi’an (5.1 ␮g m−3 ), suggesting different source categories for carbonaceous aerosols between the two cities. The PU Roadside Station was dominated by fresh vehicle emissions, with a low average OC/EC ratio of 0.6 ± 0.3, and Xi’an was dominated by coal combustion emissions, with a high OC/EC ratio of 4.4 [34]. It has been reported that fresh vehicle emissions accounted for more than 60% of OC at a typical roadside Mong Kok (MK) station in Hong Kong after examining detailed organic species in PM2.5 using gas chromatography–mass spectrometry (GC/MS) method [38]. Average wintertime OC was 12.2 ± 6.0 ␮g m−3 (Table 2), approximately 60% higher than summer. High OC concentrations at the PU Roadside Station in winter were not due to the nearby on-road primary vehicle exhaust because the daily percentage of diesel-fueled vehicles (∼32%) and total trafﬁc numbers on the Hong Chong road is consistent throughout the entire sampling period, as shown in Table 1. In addition, measurements of OC are sensitive to ambient and sampling conditions because gas-particle partitioning of OC are impacted by surrounding meteorological parameters. The seasonal data showed an inverse relationship between OC concentrations (Table 2) and temperature (Table 1) in line with the dynamic equi-

Y. Cheng et al. / Journal of Hazardous Materials 195 (2011) 82–91

b

45

Sulfate (μg m-3 )

36

Soluble potassium (μg m-3 )

a

y=2.57(±0.10)x-0.05(±0.52) R=0.97 n=40

27 18 9 0

0

5

10

Sulfur (μg

15

85

3

y=0.92(±0.03)x-0.02(±0.02) R=0.98 n=40

2

1

0

0

1

2

Total potassium (μg

m-3 )

3

m -3 )

Fig. 2. Physical consistency tests of PM1 measurements for (a) sulfate (SO4 = ) by ion chromatography on quartz-ﬁber ﬁlters versus total sulfur (S) by X-ray ﬂuorescence (XRF) on Teﬂon-membrane ﬁlters; (b) soluble potassium (K+ ) versus total potassium (K).

librium between particle and gas phase OC, which supports high OC concentrations observed in winter. Except for primary sources, OC can be formed in the atmosphere as secondary aerosol that is typically enhanced when solar intensity is higher and daylight hours are longer. Table 1 shows that seasonal sunshine hours are highest in autumn when the OC/EC ratio is low indicating that it is not secondary aerosol produced locally but perhaps gas to particle partitioning processes or a change in air mass that is responsible for the seasonal changes. Pollutants transported from mainland China have been reported to increase the PM2.5 OC concentrations at the PU Roadside Station [11] and other ambient monitoring stations [14–16,36,37]. A previous study observed high OC concentrations (17.8 ± 10.2 ␮g m−3 ) and OC/EC ratios (2.9) in an upwind

area of Guangdong province [39], compared to those in Hong Kong. The seasonal EC concentrations showed highest value in summer (17.5 ± 5.8 ␮g m−3 ), followed by autumn (16.0 ± 8.0 ␮g m−3 ), winter (15.3 ± 3.8 ␮g m−3 ), and spring (12.2 ± 3.5 ␮g m−3 ). Good relationships between daily EC and wind speeds were found from November 2004 to April 2005 (winter and spring), with a correlation coefﬁcient (R) of 0.72. EC decreased from 20.1 to 8.6 ␮g m−3 as northeasterly wind speeds increased from 8 to 39 km h−1 . In addition, the upwind area has less impact on EC levels at the PU Roadside Station when winter monsoons prevail because signiﬁcantly lower EC (6.0 ± 3.2 ␮g m−3 ) has been reported over there [39]. Above evidence suggests that winter monsoons had a dispersive effect on EC concentrations.

Table 2 Statistical summary of 24 h PM1 measurements at the PU Roadside Station from October 2004 to September 2005. Totala

PM1

Spring

␮g m−3

Average

SDb

Average

SDb

Average

SDb

Average

SDb

Average

SDb

Mass (Teﬂon) Organic carbon (OC) Elemental carbon (EC) Chloride (Cl− ) Nitrate (NO3 − ) Sulfate (SO4 2− ) Soluble sodium (Na+ ) Ammonium (NH4+ ) Soluble potassium (K+ ) Sodium (Na) Magnesium (Mg) Aluminum (Al) Silicon (Si) Phosphorus (P) Sulfur (S) Chlorine (Cl) Potassium (K) Calcium (Ca) Titanium (Ti) Vanadium (V) Manganese (Mn) Iron (Fe) Nickel (Ni) Copper (Cu) Zinc (Zn) Arsenic (As) Bromine (Br) Rubidium (Rb) Strontium (Sr) Zirconium (Zr) Tin (Sn) Antimony (Sb) Barium (Ba) Lead (Pb)

41.3 8.9 12.2 0.4 2.8 10.0 1.2 3.0 0.42 0.43 0.12 0.14 0.18 0.18 4.4 0.41 0.50 0.06 0.0048 0.015 0.018 0.14 0.0048 0.011 0.11 0.0078 0.011 0.0031 0.0051 0.0081 0.025 0.027 0.029 0.040

7.5 0.4 3.5 0.3 0.9 5.3 0.1 1.1 0.20 0.17 0.04 0.05 0.09 0.08 2.0 0.60 0.26 0.02 0.0032 0.011 0.006 0.04 0.0036 0.002 0.06 0.0024 0.005 0.0036 0.0042 0.0011 0.016 0.010 0.023 0.022

34.8 7.3 17.5 0.2 0.8 6.7 1.3 1.4 0.18 0.39 0.11 0.12 0.14 0.12 2.7 0.10 0.20 0.08 0.0095 0.017 0.015 0.25 0.0054 0.012 0.18 0.0050 0.007 0.0017 0.0041 0.0082 0.015 0.038 0.025 0.017

17.9 3.3 5.8 0.2 0.7 5.0 0.2 1.3 0.17 0.36 0.07 0.11 0.17 0.10 2.4 0.03 0.21 0.11 0.013 0.019 0.013 0.38 0.0051 0.013 0.21 0.0042 0.005 0.0009 0.0022 0.0049 0.008 0.007 0.023 0.022

48.7 7.9 16.0 0.1 1.2 15.6 1.1 3.4 0.39 0.69 0.20 0.12 0.17 0.29 6.2 0.12 0.44 0.06 0.0058 0.016 0.016 0.18 0.0035 0.014 0.24 0.0147 0.005 0.0039 0.0043 0.0087 0.019 0.042 0.044 0.043

24.8 2.6 8.0 0.0 0.6 14.5 0.1 2.7 0.31 0.43 0.09 0.06 0.09 0.23 5.1 0.10 0.34 0.02 0.0025 0.006 0.005 0.08 0.0025 0.004 0.11 0.0000 0.001 0.0024 0.0011 0.0036 0.011 0.007 0.026 0.029

52.9 12.2 15.3 0.4 2.8 13.9 1.3 3.3 0.80 0.53 0.15 0.15 0.30 0.21 5.1 0.19 0.89 0.08 0.0085 0.016 0.029 0.24 0.0055 0.016 0.26 0.0150 0.017 0.0070 0.0046 0.0062 0.040 0.039 0.025 0.077

20.1 6.0 3.8 0.2 2.0 6.0 0.2 1.4 0.62 0.21 0.05 0.08 0.17 0.08 2.0 0.15 0.58 0.05 0.0067 0.018 0.021 0.17 0.0061 0.008 0.18 0.0103 0.013 0.0049 0.0031 0.0043 0.021 0.007 0.011 0.049

44.0 9.6 15.8 0.3 1.9 10.7 1.3 2.5 0.47 0.48 0.13 0.13 0.21 0.18 4.2 0.21 0.54 0.07 0.0082 0.016 0.021 0.23 0.0052 0.013 0.21 0.0127 0.012 0.0045 0.0045 0.0074 0.027 0.037 0.027 0.047

19.4 4.9 5.1 0.2 1.6 7.1 0.2 1.6 0.49 0.29 0.06 0.09 0.17 0.11 2.6 0.26 0.51 0.07 0.0091 0.017 0.017 0.26 0.0051 0.009 0.18 0.0093 0.010 0.0041 0.0027 0.0043 0.019 0.008 0.019 0.044

a

Summer

Total number of samples are equal to 40; b standard deviation.

Autumn

Winter

Y. Cheng et al. / Journal of Hazardous Materials 195 (2011) 82–91

3.4. Water-soluble ions and elements Table 2 shows that SO4 = (10.7 ± 7.1 ␮g m−3 ) was by far the major PM1 ion, followed by NH4 + (2.5 ± 1.6 ␮g m−3 ), and NO3 − (1.9 ± 1.6 ␮g m−3 ). Abundances of other ions were low: Na+ (1.3 ± 0.2 ␮g m−3 ), K+ (0.47 ± 0.49 ␮g m−3 ), and Cl− (0.3 ± 0.2 ␮g m−3 ). Crustal elements (i.e., Fe, Si, Mg, Al, and Ca in decreasing concentrations) were low, in the range of 0.07–0.23 ␮g m−3 . Trace elements (i.e., Sb, Sn, Ba, Mn, V, Cu, As, Br, Ti, Zr, Ni, Rb, and Sr) were in the range of 0.0045 ± 0.0027–0.037 ± 0.008 ␮g m−3 , with the exception of Zn (0.21 ± 0.18 ␮g m−3 ) and P (0.18 ± 0.11 ␮g m−3 ). On average, watersoluble ions, crustal elements, and the remaining elements accounted for ∼39%, 1.8%, and 0.6% of the PM1 mass, respectively. Out of all water-soluble ions and elements, eight species, such as SO4 = , NH4 + , NO3 − , Cl− , K+ , As, Br and Pb, exhibited signiﬁcant seasonal differences at the 0.05 level using one-way analysis of variance (ANOVA) [41]. The concentrations of these pollutants were generally lowest during summer and highest during autumn and winter periods, as shown in Table 2. SO4 = and NH4 + had the highest concentrations in autumn, with average values of 15.6 ± 14.5 and 3.4 ± 2.7 ␮g m−3 , respectively. Fine-mode SO4 = , NO3 − , and NH4 + are secondary aerosols, arising from oxidization of gaseous precursors in air. Strong correlation (R = 0.96) between SO4 = and NH4 + suggests their co-existence in the atmosphere. The relationships between NO3 − and NH4 + were moderate (R = 0.61). Comparisons between the calculated and observed NH4 + concentrations were conducted to evaluate the formation of ions. NH4 + concentration can be calculated based on the stoichiometric ratios of the major compounds (i.e., ammonium sulfate [(NH4 )2 SO4 ], ammonium bisulfate [NH4 HSO4 ] and ammonium nitrate [NH4 NO3 ]); assuming that NO3 − is in the form of NH4 NO3 and that SO4 = is in the form of either (NH4 )2 SO4 or NH4 HSO4 . Fig. 3 shows the good correlation (R = 0.98) between calculated and measured NH4 + concentrations. The slope was 1.7 when (NH4 )2 SO4 was

-3

Calculated ammonium (µg m )

14

y=1.74(±0.06)x+0.21(±0.17) R=0.98 n=40

12 10

PM1

(NH4)2SO4+NH4NO3 NH4HSO4 +NH4NO3

8 6 4

y=0.97(±0.03)x+0.14(±0.08) R=0.98 n=40

2 0

0

2

4

6

8

10

12

14

-3

Measured ammonium (µg m ) Fig. 3. Comparison between calculated and measured ammonium in PM1 (calculated NH4 + = 0.38 × [SO4 = ] + 0.29 × [NO3 − ]) or NH4 HSO4 (i.e., NH4 + = 0.192 × [SO4 = ] + 0.29 × [NO3 − ]).

assumed and 1.0 when NH4 HSO4 was assumed. This suggests that aerosol is acidic (i.e., not fully neutralized with available NH4 + ) and in the form of NH4 HSO4 . The anion and cation balance in Fig. 4 also shows high correlation (R = 0.98). A deﬁciency of 11% in cations was found, especially at high loading concentrations, conﬁrming the existence of acid aerosol. The seasonal anion-to-cation equivalent ratios (A/C) were 1.2 ± 0.04, 1.0 ± 0.2, 1.2 ± 0.3 and 1.2 ± 0.1 in spring, summer, autumn, and winter, respectively. Most samples had an A/C ratio higher than unity, especially during cold seasons. Increasing industrial activities in mainland China and prevailing northeasterly winds during winter may have contributed to the elevated SO4 = concentrations. Only ∼27% of the PM1 samples, mostly from summer, gave A/C ratios less than unity. Similar to those reported by Lin et al. [8], concentrations of crustal elements (i.e., Si, Al, Ca, Ti) were correlated with each other (R > 0.75), while Fig. 2 shows that over 90% of K is in the form of K+ , showing poor correlation with crustal elements. Abundant K in the form of K+ suggests the inﬂuence of biomass burning and waste incinerator in the Macao Special Administrative Region (SAR) [14]. This is conﬁrmed by good correlations (R = 0.8) of K+ with Rb and Pb. The annual average Pb level (47 ± 44 ng m−3 ) at the PU Roadside Station is three times lower than the annual USA standard of 150 ng m−3 . Annual average PM1 V (16 ± 17 ng m−3 ), Mn (21 ± 17 ng m−3 ), and Pb (47 ± 44 ng m−3 ) are much lower than the World Health Organization [42] guideline values of 1, 0.15, and 0.5 ␮g m−3 , respectively. Two carcinogenic substances, As 0.8 -3

In summer, a higher fraction of OC exists in the vapor phase as temperatures increase, which results in low particle OC concentration (7.3 ± 3.3 ␮g m−3 ) and low OC/EC ratio of 0.4 ± 0.1. Among all potential sources (e.g., vehicle exhaust, cooking, and vegetative burning), vehicle exhaust is the most likely to produce such low OC/EC ratios [40]. Secondary organic aerosols were insignificant at the PU Roadside Station also in the summer because low OC/EC ratios and a moderate correlation between OC and total sunshine hours (R = 0.42) and solar radiation (R = 0.06) were observed. The summer daily OC and EC concentrations correlated well with prevailing wind directions, with correlation coefﬁcients (R) of 0.76 and 0.84, respectively. Concentrations increased from 5.8 to 20.7 ␮g m−3 for OC, and from 9.4 to 27.1 ␮g m−3 for EC as the direction of the vector of prevailing winds changed from 50◦ to 300◦ . However, OC/EC ratios did not follow the changes in wind directions, with the minimum and maximum values of 0.31–0.62, respectively. The evidence above suggests the existence of primary sources southwest of the Hong Chong Road. Vehicle exhaust from the Victoria Harbour tunnel may contribute to observed carbonaceous aerosols, because the exit of the tunnel is about 800 m southwest of the PU Roadside Station in summer. In addition to vehicle exhaust from the tunnel, ship/container terminal emissions may contribute as well. Several container ports are distributed to the southwest of the PU sampling station, stretching for miles along the south coast of Kowloon Peninsula, Hong Kong. Elevated pollution levels around the Victoria Harbour area (near the sampling site) due to the inﬂuences of local vehicle exhaust and ship emissions have been reported by the Institute for the Environment of the Hong Kong University of Science and Technology (http://envf.ust.hk) and Civic Exchange (http://www.civic-exchange.org/).

Anion equivalence (ueq m )

86

0.6

y=1.45(±0.05)x-0.06(±0.01) R=0.98 n=40

0.4

0.2

0.0 0.0

0.2

0.4

0.6

0.8

-3

Cation equivalence (ueq m ) Fig. 4. Scatter plots of PM1 anion versus cation measurements from PM1 data. The anion equivalence was calculated from Cl− , NO3 − , SO4 = and the cation equivalence was calculated from Na+ , NH4 + , K+ , Mg++ , and Ca++ .

70

Factor1_Vehicle

60

50

50

40

40

30

30

10

0

0

Factor3_ Secondary aerosol

50 40

60 50 40 30

20

20

10

10

0

0

30

80 60

Factor 6_ Residul oil combustion

40

20

20

10

0 NO3SO4= NH4+ K+ Ca++ OC EC Na Mg Al Si P S K Ca Ti V Mn Fe Ni Cu Zn Br Pb

0 NO3SO4= NH4+ K+ Ca++ OC EC Na Mg Al Si P S K Ca Ti V Mn Fe Ni Cu Zn Br Pb

Factors' contribution (%)

40

Factor5_Waste incinerator/ biomass burning

Factor4_Resuspended road dust

NO3SO4= NH4+ K+ Ca++ OC EC Na Mg Al Si P S K Ca Ti V Mn Fe Ni Cu Zn Br Pb

30

NO3SO4= NH4+ K+ Ca++ OC EC Na Mg Al Si P S K Ca Ti V Mn Fe Ni Cu Zn Br Pb

Factors' contribution (%)

60

50

Factor2_Industrial exhaust

NO3SO4= NH4+ K+ Ca++ OC EC Na Mg Al Si P S K Ca Ti V Mn Fe Ni Cu Zn Br Pb

10

60

87

20

20

NO3SO4= NH4+ K+ Ca++ OC EC Na Mg Al Si P S K Ca Ti V Mn Fe Ni Cu Zn Br Pb

Factors' contribution (%)

Y. Cheng et al. / Journal of Hazardous Materials 195 (2011) 82–91

Fig. 5. Factor loadings obtained from positive matrix factorization (PMF) analysis of chemical constituents of PM1 .

(13 ± 9 ng m−3 ) and Ni (5 ± 5 ng m−3 ), and one toxic substance, Cu (13 ± 9 ng m−3 ), have concentrations lower than their California chronic exposure limits. Evidence indicates that all elements in PM1 related to human health were lower than corresponding guideline values over the duration of four seasons at the PU Roadside Station. 3.5. Source apportionment by the PMF model

Percentage of PM (%)

PMF has been shown to be a powerful tool for source identiﬁcation [43,44] and has been used to assess PM2.5 and PM10 source

contributions in the Arctic [45], Hong Kong [12,46], Thailand [47], Vermont [48], and cities in the USA [44,49–51]. For this study, measured concentration values and uncertainties (sampling and chemical analytical errors) were used as input data for the PMF3.0 model [52–54]. Species with a signal-to-noise ratio less than 0.2 were excluded from the analysis [54]. Finally, only 25 species were included in the PMF3.0, including mass, NO3 − , SO4 = , NH4 + , K+ , Ca++ , OC, EC, Na, Mg, Al, Si, P, S, K, Ca, Ti, V, Mn, Fe, Ni, Cu, Zn, Br, and Pb. Forty samples were involved in the calculation.

Chemical Profile of PM1 from the PMF

10

Chemical Profile of PM2.5from the tunnel

measurement

1

NO3SO4= NH4+ K+ OC EC Na Mg Al Si P S K Ca Ti V Mn Fe Ni Cu Zn Br

0.1

Chemical profile for mixed-vehicles Fig. 6. Comparison of PMF-calculated and tunnel-measured vehicle chemical proﬁles in Hong Kong.

Fig. 7. Average source contributions of each factor to PM1 mass.

88

Y. Cheng et al. / Journal of Hazardous Materials 195 (2011) 82–91

25 20

EC_Vehicle

10

5

5

0

0

0.07 0.06

=

SO4 _Secondary aerosol

15

10

V_Residul oil combustion

0.05

2.0

K_Waste incinerator / biomass burning

1.5

0.04 1.0

0.03 0.02

0.5

0.01 0.00

0.0

0.7

0.8

0.6

Si_Suspended road dust

0.5

0.7 0.6

Zn_Industrial exhaust

0.5

0.4

0.4

0.3

0.3 0.2

0.1

0.1

0.0

0.0 2004-10-8 2004-10-14 2004-10-20 2004-10-26 2004-10-29 2004-11-1 2004-11-4 2004-11-13 2004-11-19 2004-11-24 2004-12-22 2004-12-28 2005-1-4 2005-1-10 2005-1-22 2005-1-30 2005-2-3 2005-2-11 2005-4-4 2005-4-12 2005-4-19 2005-4-27 2005-5-4 2005-5-21 2005-6-7 2005-6-14 2005-6-21 2005-6-28 2005-7-5 2005-7-13 2005-7-20 2005-7-22 2005-7-28 2005-8-4 2005-8-9 2005-8-12 2005-8-31 2005-9-10 2005-9-16 2005-9-23

0.2

2004-10-8 2004-10-14 2004-10-20 2004-10-26 2004-10-29 2004-11-1 2004-11-4 2004-11-13 2004-11-19 2004-11-24 2004-12-22 2004-12-28 2005-1-4 2005-1-10 2005-1-22 2005-1-30 2005-2-3 2005-2-11 2005-4-4 2005-4-12 2005-4-19 2005-4-27 2005-5-4 2005-5-21 2005-6-7 2005-6-14 2005-6-21 2005-6-28 2005-7-5 2005-7-13 2005-7-20 2005-7-22 2005-7-28 2005-8-4 2005-8-9 2005-8-12 2005-8-31 2005-9-10 2005-9-16 2005-9-23

-3

25 20

15

Species concentration (µg m )

30

Fig. 8. Temporal patterns of marker species from the six source categories during the time period from October 2004 to September 2005.

Six factors were generated by the PMF model and the contributions of each factor are shown in Fig. 5. The ﬁrst factor with high loading of OC, EC, and Ca++ , are characteristic of vehicle exhaust [40,55–57] and lube oil additives [58], respectively. As shown in Fig. 6, the species abundances in this PMF-derived vehicle source proﬁle are comparable to those measured in an urban tunnel in Hong Kong [55], especially for the major species (i.e., OC, EC, NO3 − , SO4 = , NH4 = , Na, Mg, and K). PMF-derived vehicle source proﬁles underestimate crustal elements (i.e., Al, Si, Ca, Fe), which is reasonable because these elements mainly occur in PM2.5 but not in PM1 . The second factor, loaded with Mn and Zn, represents exhaust from industry [59,60]. The third factor is identiﬁed as secondary aerosols, based on the high abundances of NO3 − , SO4 = , and NH4 + . The fourth factor, enriched with Al, Si, Ca, Ti, and Fe, is best explained as geological material or resuspended road dust [61,62]. The ﬁfth factor with abundant K+ , Br, and Pb, is indicative of waste incinerator/biomass burning emissions, as previously found by Louie et al. [14]. The sixth factor, loaded with V and Ni, is consistent with residual oil combustion, likely from ship emissions or utilities at container terminals [12,17]. Fig. 7 shows that vehicle exhaust is the largest contributor, accounting for ∼38% of the PM1 mass. This is consistent with sampling in a roadside vehicle exhaust-dominated environment. Secondary aerosols are the second largest contributor, accounting for ∼22% of PM1 . It has been shown through previous analyses that long-range transported secondary aerosols impact the air quality at the sampling site when the air mass has traveled over China before reaching Hong Kong. Waste incinerator/biomass burning and residual oil combustion account for ∼16% and ∼12% of PM1 , respectively. Industrial exhaust and resuspended road dust account for ∼7% and ∼5% of PM1 , respectively.

Marker species (i.e., EC, Zn, SO4 = , Si, K+ , and V, respectively) were selected as representative components of their six respective source categories. Temporal patterns of marker species (Fig. 8) and meteorological characteristics on episode days (Table 3) were examined in order to identify essential meteorological parameters that may affect the occurrence and intensity of certain types of air pollution. The highest ﬁve concentrations for each marker species were considered to air pollution episodes, as shown in Table 3. Meteorological characteristics were obtained from the Hong Kong Observatory. The temporal patterns of SO4 = and K+ suggest that secondary aerosols and waste incinerator/biomass burning frequently occurred in autumn, winter, and spring. The prevailing wind directions were 70◦ and 54◦ on the SO4 = and K+ episode days, respectively. Moreover, elevated SO4 = , K+ , and PM1 concentrations often existed simultaneously. Vehicle exhaust episodes mainly occurred in summer, accompanied by typhoon or tropical storm events. Elevated PM1 concentrations were often observed at the same time. The temporal variation in V showed weak seasonality for residual oil combustion sources. V and Ni correlated well with each other (R = 0.97) and did not change with the changes in wind directions throughout the entire sampling periods. Therefore, ship emissions may also inﬂuence the PU Roadside Station as a regional pollution source. Wind direction was the most important meteorological parameter for industrial exhaust sources as they are stationary emitters. Resuspended road dust episodes predominantly occurred in winter, with less rainfall and dry northeasterly winds prevail. Throughout the entire sampling period, air pollution on 4 January 2005 was the most serious, with all marker species showing elevated concentrations in the atmosphere, which could be explained by the extremely low mixing height of 477 m on that

Table 3 Summary of air pollution episodes (highest ﬁve concentrations of each marker species) from October 2004 to September 2005. Episode day

Vehicle Secondary exhaust aerosols

a

EC ∼38%

b

8October2004 14 October 2004 20 October 2004 26 October 2004 1 November 2004 4 November 2004 19 November 2004 4 January 2005 10 January 2005 19 April 2005 21 June 2005

67.0 55.9 84.4 67.1 62.4 40.4 41.4 93.6 76.2 48.9 32.6

13 July 2005 20 July 2005

40.9 64.2

4 August 2005

25.1

12 August 2005

74.5

26.8

31 August 2005 10 September 2005 23 September 2005 Annual average 44.0

60.4 72.4 50.6 15.7

22.5

a

SO4 = ∼22%

Waste incinerator/bio moss burning K+ ∼16%

Resuspended road dust

Residual oil combustion

Si ∼5%

V ∼12%

Air temperature (◦ C)

Relative humidity (%)

Rainfall (mm)

Prevailing wind direction (◦ )

Wind speed (km h−1 )

Weather

904 1268 1022 967 663 620 733 477 732 660 641

25.9 25.1 24.8 25.5 23.1 23.1 20.4 16.5 15.2 23.3 28.5

45% 70% 74% 58% 70% 73% 51% 70% 73% 84% 83%

0.0 0.1 5.8 2.3 2.2 0.0 0.0 1.2 0.0 4.7 9.7

10 90 80 20 80 80 90 70 90 70 230

17 36 31 37 28 29 21 26 24 19 22

842 573

29.8 28.7

75% 81%

0.2 16.2

230 230

20 23

836

28.5

82%

14.9

240

20

0.5

738

28.3

83%

16.4

230

21

29.4 28.1 28.6

1067

728 898 838 23.0

76% 78% 67% 75%

0.3 6.6 Trace 10.8

270 90 10 87.8

18 18 42 23

Fine and dry Fine Haze Fine Some haze Sunshine Fine Haze Haze Sunshine Heavy rains and Thunderstorms Sunny and hot Sunny and hot; Thunderstorms Sunny and hot; Thunderstorms Haze; Thunderstorms Fine Haze Hot and hazy Annual average 44.0

Industrial Mixing exhaust height (m) Zn ∼7%

1.3 19.2 24.1 18.4

26.5

1.6 1.6

2.2 1.4

0.7 0.5 0.4 0.5

0.054

0.7 0.4

0.067

0.5

0.031 0.7

25.2 22.8 0.7 0.070 0.050 32.2 25.2 10.7

0.5

0.2

0.016

0.2

Y. Cheng et al. / Journal of Hazardous Materials 195 (2011) 82–91

PM1 mass

Marker species; b percentage of PM1 .

89

90

Y. Cheng et al. / Journal of Hazardous Materials 195 (2011) 82–91

day. Overall, average concentrations of each marker species, except for EC, on episode days were around two to three times higher than the annual average values. The average concentration of EC was 24.5 ± 1.8 ng m−3 on episode days, approximately 1.6 times the annual average value. Meteorological data shows that low mixing heights (averaged at 786 m) occurred on all air pollution episode days and relative humidity ranged from 70 to 85%. Except for vehicle exhaust episodes, all air pollution episodes were associated with hazy, reduced visibility conditions, similar to those found in Beijing [63] and Guangzhou, China [64]. 4. Conclusions Twenty-four-hour PM1 samples were collected every sixth day at an urban roadside monitoring station in Hong Kong from October 2004 to September 2005. Concentrations of OC, EC, water-soluble ions, and up to 25 elements were reported. The seasonal average PM1 concentrations were 41.3 ± 7.5, 34.8 ± 17.9, 48.7 ± 24.8, and 52.9 ± 20.1 ␮g m−3 for spring, summer, autumn, and winter, respectively. PM1 and major component species (e.g., OC, SO4 = , NO3 − , NH4 + , etc.) showed distinct seasonal patterns with elevated concentrations typically in autumn, winter and spring, which were associated with northeasterly winds that transport from the continental Asian interior. Except for long-range transported regional pollutants, OC were impacted by gas-particle partitioning as well. EC was an exception with elevated concentrations in summer associated with primary emissions from local nearby tunnel and container ports. The low OC/EC ratio (<1) reﬂects that primary carbon emissions, rather than secondary organic aerosol, dominate this nearby roadside location. Source apportionment of PM1 was carried out using a PMF model. Vehicle exhaust was the largest contributor to PM1 (∼38%), followed by secondary aerosols (∼22%), waste incinerator/biomass burning emissions (∼16%), residual oil combustion (∼12%), industrial exhaust (∼7%) and resuspended road dust (∼5%). Elevated PM1 mass concentrations were found when the sampling station was experiencing vehicle exhaust episodes in summer or secondary aerosols and waste incinerator/biomass burning emissions in winter. However, this was not the case when the station was impacted by residual oil combustion, industrial emissions, or resuspended road dust. Most air pollution episodes are associated with hazy, reduced visibility conditions, and low mixing heights (averaged at 786 m). Acknowledgements This project is supported by Hong Kong Polytechnic University (G-YX3L, G-YF23), Xi’an Jiaotong University (no. 0814100; no. 51100033), State Key Laboratory of Loess & Quaternary Geology (SKLLQG0804), and the Research Grants Council of Hong Kong (RGC 5197/05E, PolyU 5175/09E and BQ01T). References [1] J. Ferin, G. Oberdörster, D.P. Penney, Pulmonary retention of ultraﬁne and ﬁne particles in rats, Am. J. Respir. Cell Mol. Biol. 6 (1992) 535–542. [2] K. Donaldson, X.Y. Li, W. MacNee, Ultraﬁne (nanometer) particle mediated lung injury, J. Aerosol Sci. 29 (1998) 553–560. [3] G. Oberdörster, Pulmonary effects of inhaled ultraﬁne particles, Int. Arch. Occup. Environ. Health 74 (2001) 1–8. [4] C.A. Pope III, D.W. Dockery, Critical review: health effects of ﬁne particulate air pollution: lines that connect, J. Air Waste Manage. Assoc. 56 (2006) 709–742. [5] J.D. McDonald, M.D. Reed, M.J. Campen, E.G. Barrett, J. Seagrave, J.L. Mauderly, Health effects of inhaled gasoline engine emissions, Inhal. Toxicol. 19 (2007) 107–116. [6] J.C. Chow, J.G. Watson, J.L. Mauderly, D.L. Costa, R.E. Wyzga, S. Vedal, G.M. Hidy, S.L. Altshuler, D. Marrack, J.M. Heuss, G.T. Wolff, C.A. Pope III, D.W. Dockery, Critical review discussion – health effects of ﬁne particulate air pollution: lines that connect, J. Air Waste Manage. Assoc. 56 (2006) 1368–1380.

[7] Y. Cheng, S.C. Lee, J.J. Cao, K.F. Ho, J.C. Chow, J.G. Watson, C.H. Ao, Elemental composition of airborne aerosols at a trafﬁc site and a suburban site in Hong Kong, Int. J. Environ. Pollut. 36 (2009) 166–179. [8] C.C. Lin, S.J. Chen, K.L. Huang, W.I. Hwang, G.P. Chang, W.Y. Lin, Characteristics of metals in nano/ultraﬁne/ﬁne/coarse particles collected beside a heavily trafﬁcked road, Environ. Sci. Technol. 39 (2005) 8113–8122. [9] H. Guo, A.J. Ding, K.L. So, G. Ayoko, Y.S. Li, W.T. Hung, Receptor modeling of source apportionment of Hong Kong aerosols and the implication of urban and regional contribution, Atmos. Environ. 43 (2009) 1159–1169. [10] K.F. Ho, S.C. Lee, C.K. Chan, J.C. Yu, J.C. Chow, X.H. Yao, Characterization of chemical species in PM2.5 and PM10 aerosols in Hong Kong, Atmos. Environ. 37 (2003) 31–39. [11] S.C. Lee, Y. Cheng, K.F. Ho, J.J. Cao, P.K.K. Louie, J.C. Chow, J.G. Watson, PM1.0 and PM2.5 characteristics in the roadside environment of Hong Kong, Aerosol Sci. Technol. 40 (2006) 157–165. [12] Z.B. Yuan, A.K.H. Lau, H.Y. Zhang, J.Z. Yu, P.K.K. Louie, J.C.H. Fung, Identiﬁcation and spatiotemporal variations of dominant PM10 sources over Hong Kong, Atmos. Environ. 40 (2006) 1803–1815. [13] H. Tsang, R. Kwok, A.H. Miguel, Pedestrian exposure to ultraﬁne particles in Hong Kong under heavy trafﬁc conditions, Aerosol Air Qual. Res. 8 (2008) 19–27. [14] P.K.K. Louie, J.C. Chow, L.W.A. Chen, J.G. Watson, G. Leung, D.W.M. Sin, PM2.5 chemical composition in Hong Kong: urban and regional variations, Sci. Total Environ. 338 (2005) 267–281. [15] P.K.K. Louie, J.G. Watson, J.C. Chow, L.W.A. Chen, D.W.M. Sin, A.K.H. Lau, Seasonal characteristics and regional transport of PM2.5 in Hong Kong, Atmos. Environ. 39 (2005) 1695–1710. [16] K.L. So, H. Guo, Y.S. Li, Long-term variation of PM2.5 levels and composition at rural, urban, and roadside sites in Hong Kong: increasing impact of regional air pollution, Atmos. Environ. 41 (2007) 9427–9434. [17] J.Z. Yu, J.W.T. Tung, A.W.M. Wu, A.K.H. Lau, P.K.K. Louie, J.C.H. Fung, Abundance and seasonal characteristics of elemental and organic carbon in Hong Kong PM10 , Atmos. Environ. 151 (2004) 1511–1521. [18] P.C. Chin, in: T.N. Chiu, C.L. So (Eds.), Climate and Weather: A Geography of Hong Kong, Oxford University Press, New York, 1986, pp. 69–85. [19] The annual trafﬁc census, (2004), Transport Department, The Government of the Hong Kong Special Administrative Region, http://www.td.gov.hk/. [20] The annual trafﬁc census, (2005), Transport Department, The Government of the Hong Kong Special Administrative Region, http://www.td.gov.hk/. [21] J.G. Watson, J.C. Chow, C.A. Frazier, X-ray ﬂuorescence analysis of ambient air samples, in: S. Landsberger, M. Creatchman (Eds.), Elemental Analysis of Airborne Particles, vol. 1, Gordon and Breach Science, Amsterdam, 1999, pp. 67–96. [22] J.C. Chow, J.G. Watson, Ion chromatography in elemental analysis of airborne particles, in: S. Landsberger, M. Creatchman (Eds.), Elemental Analysis of Airborne Particles, vol. 1, Gordon and Breach Science, Amsterdam, 1999, pp. 97–137. [23] J.C. Chow, J.G. Watson, L.C. Pritchett, W.R. Pierson, C.A. Frazier, R.G. Purcell, The Dri thermal optical reﬂectance carbon analysis system – description, evaluation and applications in United-States air-quality studies, Atmos. Environ. 27 (1993) 1185–1201. [24] J.C. Chow, J.G. Watson, L.W.A. Chen, W.P. Arnott, H. Moosmuller, Equivalence of elemental carbon by thermal/optical reﬂectance and transmittance with different temperature protocols, Environ. Sci. Technol. 38 (2004) 4414–4422. [25] J.C. Chow, J.G. Watson, S.D. Kohl, M.P. Gonzi, L.W.A. Chen, Measurements and validation for the twelve month particulate matter study in Hong Kong, Report to Hong Kong Environmental Protection Department. (2002). [26] J.P. Engelbrecht, L. Swanepoel, J.C. Chow, J.G. Watson, R.T. Egami, PM2.5 and PM10 concentrations from the Qalabotjha low-smoke fuels macro-scale experiment in South Africa, Environ. Monit. Assess. 69 (2001) 1–15. [27] Y. Cheng, S.C. Lee, K.F. Ho, K. Fung, Positive sampling artifacts in particulate organic carbon measurements in roadside environment, Environ. Monit. Assess. 168 (2010) 645–656. [28] J.C. Chow, J.G. Watson, L.W.A. Chen, J. Rice, N.H. Frank, Quantiﬁcation of PM2.5 organic carbon sampling artifacts in US networks, Atmos. Chem. Phys. 10 (2010) 5223–5239. [29] T.W. Kirchstetter, C.E. Corrigan, T. Novakov, Laboratory and ﬁeld investigation of the adsorption of gaseous organic compounds onto quartz ﬁlters, Atmos. Environ. 35 (2001) 1663–1671. [30] B.J. Turpin, J.J. Huntzicker, S.V. Hering, Investigation of organic aerosol sampling artifacts in the Los Angeles Basin, Atmos. Environ. 28 (1994) 3061–3071. [31] J.G. Watson, J.C. Chow, L.W.A. Chen, N.H. Frank, Methods to assess carbonaceous aerosol sampling artifacts for IMPROVE and other long-term networks, J. Air Waste Manage. Assoc. 59 (2009) 898–911. [32] C.S. Li, C.H. Lin, PM1.0 /PM2.5 /PM10 characteristics in the urban atmosphere of Taipei, Aerosol Sci. Technol. 36 (2002) 469–473. [33] U. Makkonen, H. Hellén, P. Anttila, M. Ferm, Size distribution and chemical composition of airborne particles in south-eastern Finland during different seasons and wildﬁre episodes in 2006, Sci. Total Environ. 408 (2010) 644–651. [34] Z.X. Shen, J.J. Cao, R. Arimoto, Y.M. Han, C.S. Chu, J. Tian, S.X. Liu, Chemical characteristics of ﬁne particles (PM1 ) from Xi’an, China, Aerosol Sci. Technol. (2010) 461–472. [35] L.M. Russell, Aerosol organic-mass-to-organic-carbon ratio measurements, Environ. Sci. Technol. 37 (2003) 2982–2987. [36] Y. Cheng, K.F. Ho, S.C. Lee, S.W. Law, Seasonal and diurnal variations of PM1.0, PM2.5 and PM10 in the roadside environment of Hong Kong, Particuology 4 (2006) 312–315.

Y. Cheng et al. / Journal of Hazardous Materials 195 (2011) 82–91 [37] R.K. Pathak, X.H. Yao, A.K.H. Lau, C.K. Chan, Acidity and concentrations of ionic species of PM2.5 in Hong Kong, Atmos. Environ. 37 (2003) 1113–1124. [38] M. Zheng, G.S.W. Haqler, L. Ke, M.H. Berqin, F. Wang, P.K.K. Louie, L. Salmon, D.W.M. Sin, J.Z. Yu, J.J. Schauer, Composition and sources of carbonaceous aerosols at three contrasting sites in Hong Kong, J. Geophys. Res. 111 (2006) D20313, doi:10.1029/2006JD007074. [39] J.J. Cao, S.C. Lee, K.F. Ho, X.Y. Zhang, S.C. Zou, K. Fung, J.C. Chow, J.G. Watson, Characteristics of carbonaceous aerosol in the Pearl River Delta Region, China during 2001 winter period, Atmos. Environ. 37 (2003) 1451–1460. [40] J.G. Watson, J.C. Chow, D.H. Lowenthal, L.C. Pritchett, C.A. Frazier, G.R. Neuroth, R. Robbins, Differences in the carbon composition of source proﬁles for dieseland gasoline-powered vehicles, Atmos. Environ. 28 (1994) 2493–2505. [41] W.J. Conover, R.L. Iman, Rank transformations as a bridge between parametric and nonparametric statistics, Am. Stat. 35 (1981) 124–129. [42] WHO, World Health Organization, Guidelines for Air Quality; Geneva, 2000. [43] R.D. Willis, Workshop on Unmix and PMF as applied to PM2.5, EPA 600-A-00048, 2000. [44] X.H. Song, A.V. Polissar, P.K. Hopke, Sources of ﬁne particle composition in the northeastern US, Atmos. Environ. 35 (2001) 5277–5286. [45] Y.L. Xie, P.K. Hopke, P. Paatero, L.A. Barrie, S.M.J. Li, Identiﬁcation of source nature and seasonal variations of arctic aerosol by positive matrix factorization, J. Atmos. Sci. 56 (1999) 249–260. [46] E. Lee, C.K. Chan, P. Paatero, Application of positive matrix factorization in source apportionment of particulate pollutants in Hong Kong, Atmos. Environ. 33 (1999) 3201–3212. [47] W. Chueinta, P.K. Hopke, P. Paatero, Investigation of source of atmospheric aerosol at urban and suburban residential areas in Thailand by positive matrix factorization, Atmos. Environ. 34 (2000) 3319–3329. [48] A.V. Polissar, P.K. Hopke, R.L. Poirot, Atmospheric aerosol over vermont: chemical composition and sources, Environ. Sci. Technol. 35 (2001) 4604–4621. [49] E. Kim, P.K. Hopke, E.S. Edgerton, Source identiﬁcation of Atlanta aerosol by positive matrix factorization, J. Air Waste Manage. Assoc. 53 (2003) 731–739. [50] E. Kim, P.K. Hopke, T.V. Larson, D.S. Covert, Analysis of ambient particle size distributions using Unmix and positive matrix factorization, Environ. Sci. Technol. 38 (2004) 202–209. [51] L.W.A. Chen, J.G. Watson, J.C. Chow, D.W. DuBois, L. Herschberger, Chemical mass balance source apportionment for combined PM2.5 measurements from U.S. non-urban and urban long-term networks, Atmos. Environ. 44 (2010) 4908–4918.

91

[52] P. Paatero, U. Tapper, Positive matrix factorization: a non-negative factor model with optimal utilization of error estimates of data values, Environmetrics 5 (1994) 111–126. [53] P. Paatero, Least squares formulation of robust non-negative factor analysis, Chemom. Intell. Lab. Syst. 37 (1997) 23–35. [54] P. Paatero, P.K. Hopke, Discarding or downweighting high-noise variables in factor analytic models, Anal. Chim. Acta 490 (2003) 277–289. [55] Y. Cheng, S.C. Lee, K.F. Ho, J.C. Chow, J.G. Watson, P.K.K. Louie, J.J. Cao, H. Xu, Chemically speciated on-road PM2.5 motor vehicle emission factors in Hong Kong, Sci. Total Environ. 408 (2010) 1621–1627. [56] J.M. Norbeck, T.D. Durbin, T.J. Truex, Measurement of primary particulate matter emissions from light-duty motor vehicles. Report to the coordinating research council under CRC Project No. E-24-2, by Center for Environmental Research and College of Engineering Technology (CE-CERT) U.C., Riverside, CA, 1998. [57] A.W. Gertler, J.A. Gillies, W.R. Pierson, C.F. Rogers, J.C. Sagebiel, M. Abu-Allaban, W. Coulombe, L. Tarnay, T.A. Cahill, D. Grosjean, E. Grosjean, Emissions from diesel and gasoline engines measured in highway tunnels, in: Health Effects Institute, Capital City Press, Montpelier VT, Boston, MA, USA, 2002. [58] H.J. Wei, H.L. Yu, D.L. Guan, P.T. Sun, H.Z. Wang, MOA spectral analysis of additive element contents in lube oil, Spectrosc. Spect. Anal. 26 (2006) 738–740. [59] J.M. Pacyna, in: J.O. Nriagu, C.I. Davidson (Eds.), Toxic Metals in the Atmosphere, Wiley, New York, 1986. [60] G.M. Marcazzan, S. Vaccaro, G. Valli, R. Vecchi, Characterisation of PM10 and PM2.5 particulate matter in the ambient air of Milan (Italy), Atmos. Environ. 35 (2001) 4639–4650. [61] I.F. Al-Momani, Trace elements in atmospheric precipitation at Northern Jordan measured by ICP-MS: acidity and possible sources, Atmos. Environ. 37 (2003) 4507–4515. [62] M. Singh, P.A. Jaques, C. Sioutas, Size distribution and diurnal characteristics of particle-bound metals in source and receptor sites of the Los Angeles Basin, Atmos. Environ. 36 (2002) 1675–1689. [63] Y.L. Sun, G.S. Zhuang, A.H. Tang, Y. Wang, Z.S. An, Chemical characteristics of PM2.5 and PM10 in haze-fog episodes in Beijing, Environ. Sci. Technol. 40 (2006) 3148–3155. [64] J.H. Tan, J.C. Duan, K.B. He, Y.L. Ma, F.K. Duan, Y. Chen, J.M. Fu, Chemical characteristics of PM2.5 during a typical haze episode in Guangzhou, J. Environ. Sci. 21 (2009) 774–781.

Journal of Hazardous Materials 195 (2011) 92–99

Contents lists available at ScienceDirect

Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat

Cancer risk assessment of polybrominated diphenyl ethers (PBDEs) and polychlorinated biphenyls (PCBs) in former agricultural soils of Hong Kong Yu Bon Man a,b , Brenda Natalia Lopez b , Hong Sheng Wang a,b , Anna Oi Wah Leung b , Ka Lai Chow a,b , Ming H. Wong a,b,∗ a b

School of Environmental and Resource Sciences, Zhejiang Agriculture and Forestry University, Lin’an, Zhejiang 311300, PR China Croucher Institute for Environmental Sciences and Department of Biology, Hong Kong Baptist University, Hong Kong, PR China

a r t i c l e

i n f o

Article history: Received 7 April 2011 Received in revised form 2 August 2011 Accepted 4 August 2011 Available online 10 August 2011 Keywords: Land use change Farm soils Lifetime cancer risk PBDEs PCBs

a b s t r a c t The major objective of this study was to evaluate the carcinogenic risk posed to humans through PBDEs and PCBs of changing agricultural land use for recycling of e-waste and open burning of municipal waste. Nine locations were selected to represent 6 different types of land use such as e-waste dismantling workshop (EW (DW)) and e-waste open burning site (EW (OBS)). The total concentrations for PBDEs and PCBs, and the bioaccessibility of PCBs were determined using Soxhlet extraction and in vitro simulated gastric solution, respectively. Both total and bioaccessible concentrations were subsequently used to establish the cancer risk probabilities in humans via ingestion, dermal contact and inhalation of soil particles. It was found that very low cancer risk in all 6 types of different land use was caused by BDE209. Nevertheless, at the 95th centile, the concentration of PCBs in EW (DW) and EW (OBS) indicate a low cancer risk to humans of 40 and 2.1 in a million, respectively, while the same was also observed for the bioaccessible PCBs in EW (DW) of 1.71 ± 2.96 in a million. © 2011 Elsevier B.V. All rights reserved.

1. Introduction Polybrominated diphenyl ethers (PBDEs) and polychlorinated biphenyls (PCBs) are both persistent organic pollutants (POPs), which are commonly found at e-waste recycling sites [1,2]. Polybrominated diphenyl ethers and PCBs tend to readily accumulate in the fats of organisms and get passed along the food chain due to their high lipophilicity [3,4]. For example, the former were detected in Indo-Paciﬁc humpback dolphins (Sousa chinensis) and ﬁnless porpoises (Neophocaena phocaenoides) in Hong Kong [5], while the latter were identiﬁed in human milk [6]. Polybrominated diphenyl ethers are believed to act as endocrine disruptors that affect hormone regulation [7]. It has been shown that BDE-209, the major ingredient of commercial brominated ﬂame retardants can cause neurobehavioral derangements in adult mice [8]. Animal studies also revealed that PBDEs can cause other health problems such as thyroid hormone disruption, and possibly cancer [9–11]. Polybrominated diphenyl ethers are commonly used as ﬂame retardants, while PCBs are components of transformers and capacitors, as well as, hydraulic and heat exchange ﬂuids [12], which

∗ Corresponding author at: Croucher Institute for Environmental Sciences, Hong Kong Baptist University, Hong Kong, PR China. Tel.: +852 3411 7746; fax: +852 3411 7743. E-mail addresses: [email protected], [email protected] (Ming H. Wong). 0304-3894/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.jhazmat.2011.08.010

may explain why they can always be found at electronic waste (e-waste) recycling sites [1,2]. In Hong Kong, there are sites for e-waste recycling and open burning that represent a signiﬁcant number of potential releasing sources of PBDEs and PCBs. These sites are former agricultural lands in which their initial purpose has been changed to other forms of land use. The reasons behind this may be clariﬁed by the burgeoning urbanization leading to the disappearance and fragmentation of large areas of farmland, which in turn depreciated the value of the land and thus further exacerbating the problem. Large-scale cultivation was no longer feasible after the “critical mass” of farmland had been destroyed [13,14]. In the 1980s, both abandoned and existing farmlands were dramatically converted for other purposes, directly as a result of rapid economic development and fragmentation [13,15,16]. Consequently, substantial areas of agricultural land were changed to greater proﬁt generating applications such as storage sites (for container), car dismantling workshops, and more recently, for the storing, dismantling, recycling, and open burning of e-waste. Most of the previous studies focused on the concentrations of PBDEs and PCBs in biota samples [17–19], where no studies relating to their concentrations in former agricultural soils of Hong Kong could be found. Therefore, it was crucial to perform an investigation on estimating the risks and potential health effects of these POPs in soils after land use changes in Hong Kong. The health risks exerted on humans was generally overestimated in other studies when risk assessments were conducted by means of total pollutant

Y.B. Man et al. / Journal of Hazardous Materials 195 (2011) 92–99

93

100 m × 100 m, which were divided into 5 equal sub-areas, where each had 5 random soil samples (0–5 cm, 0.5 kg) assembled using a stainless steel spade. Consequently, leading to the formation of individual composite samples, hence, each site contained 5 of these. The soil samples from EW (S) were taken from the surrounding agricultural area as the ﬂoor of the EW (S) was concerted. Half of the soil samples were air-dried for two weeks, while the other half was freeze-dried for at least two weeks and sieved through a 2-mm mesh. Soil texture and soil organic matter (SOM) were determined by the Bouyoucos Soil Hydrometer Method [22] and the method for analyzing the total organic carbon (TOC) in soils and sediments [23], respectively. Every 5th soil sample was tested in duplicate to check the consistency of the data. 2.2. The extraction and analysis of total PBDEs and PCBs

Fig. 1. The location of 9 soil sampling sites in Hong Kong. OF = organic farm, A = agricultural, EW (S) = e-waste storage, OBS = open burning site, EW (DW) = ewaste dismantling workshop, EW (OBS) = e-waste open burning site and CDW = car dismantling workshop.

concentrations [20]. Therefore, this study included the analysis of bioaccessible PBDE and PCB concentrations to combat this issue, thus ensuring that the resultant cancer risk assessment presented was a realistic portrayal of events. The objectives of this study were to determine the concentrations of PBDEs and PCBs, especially their bioaccessible concentrations in current and former agricultural soils of Hong Kong, as well as, to conduct a human health risk assessment on cancer, in order to evaluate the potential risks based on their concentrations within soils. 2. Materials and methods 2.1. Sampling, preparation and analysis The PBDE and PCB concentrations were previously reported in a study by Lopez et al. [21], where its major difference with this study is that it involved the addition of 2 open burning sites, in which bioaccessible PCB concentrations were analyzed and the pollutant concentrations subsequently used for human cancer risk evaluation. Fig. 1 brieﬂy shows the location of the 9 sampling sites, whereby a total of 45 composite soil samples were collected from these sites within the existing and former agricultural lands, located to the north west of the New Territories in Hong Kong. The sampling sites were grouped into 6 soil types according to their current land use: agricultural (A); organic farm (OF); e-waste storage (EW (S)); e-waste dismantling workshop (EW (DW)); ewaste open burning site (EW (OBS)); and open burning site (OBS). Descriptions of each type of land use and their number of sites are given in Table 1. The size of each location totaled approximately

The extraction of PBDEs and PCBs from the soil samples was conducted by using the Standard Method 3540C [24]. Five grams (g) of each soil sample was spiked with PCB standard solutions (28 PCB congeners mixture, Standard Reference Materials Group of NIST, USA), native (12 C12 ) PBDE solution/mixture (BDE-MXE) for precision and recovery (PAR) and mass-labelled (13 C12 ) solution/mixture (MBDE-MXE) (Wellington Laboratories Inc., Canada) for PBDE recoveries. The samples were then Soxhlet extracted with 150 mL acetone (pesticide grade, Tedia), dichloromethane (DCM) (pesticide grade, Tedia) and n-hexane mixture (1:1:1, v:v) in a 65 ◦ C water bath for 18 h with 3 g of anhydrous sodium sulphate for moisture removal during extraction. One g of activated copper granules (Riedel-de Haën) was added to remove the sulphur during the process (activated copper granules was prepared by mixing with hydrochloric acid (1 N) then washed with distilled water and DCM). The extracts were then concentrated to roughly 1 mL by a rotary evaporator, which were subsequently cleaned up by the standard clean up method 3620B [25]. Extracts were brieﬂy eluted with a 20 mL (7:3, v:v) mixture of n-hexane (95%, pesticide grade, Tedia) and DCM through a column packed with 5 g of anhydrous sodium sulphate and 8 g of ﬂorisil. The extracts were reduced to around 1 mL by a rotary evaporator post-clean up stage and allowed to further evaporate in the fume hood until the volume reached 200 ␮L. GC–MS analysis was performed on a Hewlett Packard 6890 GC system equipped with a mass selective detector and a 30 m × 0.25 mm × 0.25 ␮m DB-5 capillary column (J & W Scientiﬁc Co. Ltd., USA). The Standard Method 8270 C [26] was adopted for the veriﬁcation of the following 37 PCB congeners including 7 PCB indicators (PCB-28, -52, -101, -118, -138, -153 and -180). In addition, the concentrations of 22 PBDEs were determined using Shimadzu QP2010 GC/MS, following the methods described in [27], with a slight modiﬁcation by Zheng [28]. The low molecular weight PBDE congeners (BDE-3 to -191) were measured with a DB-1 (30 m × 0.25 mm i.d. ×0.25 ␮m) column, whereas a short column (12 m) was used for high molecular weight BDE congeners (BDE-197, -196, -207, -206 and -209).

Table 1 Descriptions of different agricultural land uses and their respective number of sites under investigation. Different types of agricultural land use

Number of sites

Site description

Organic farm

1

Agricultural E-waste storage Open burning site E-waste dismantling workshop E-waste open burning site

1 1 2 3 1

Vegetables are grown without the use of chemical fertilizers and pesticides in which nutrients are recycled within the cultivated area Traditional farming system and the application of legal chemical fertilizers and pesticides are allowed Electronic waste storage sites with concrete ﬂooring surrounded by concrete walls Existing agricultural land using small areas to burn bulky woody furniture, household waste and wild grass Breakdown of electronic components such as refrigerators, computers and printers on existing agricultural land Burning of electronic components such as refrigerators, computers, cables and printers on existing agricultural land

94

Y.B. Man et al. / Journal of Hazardous Materials 195 (2011) 92–99

Table 2 Slope factors of PCBs and PBDEs via the exposure pathways of ingestion, dermal contact and inhalation [35]. PCBs and PBDEs

Slope factors for evaluating cancer risks

Polychlorinated biphenyls (high risk) Decabromodiphenyl ether, 2,2 ,3,3 ,4,4 ,5,5 ,6,6 -(BDE-209)

Ingestion SFO (mg/kg/day)−1

Dermal contact SFO × GIBAS (mg/kg/day)−1

Inhalation IUR (mg/m3 )−1

2.00E+00 7.00E−04

2.00E+00 7.00E−04

5.70E−01 –

SFO = oral slope factors, GIABS = gastrointestinal absorption factor, IUR = inhalation unit risk.

2.3. Quality control A Standard Reference Material (SRM) 2585 (Organic Contaminants in House Dust) for PCBs was obtained from the National Institute of Standards and Technology (NIST, USA). The SRM and an analytical blank were included in every batch of extraction, where the mean SRM recoveries of PCBs ranged between 94 and 117%. The mean recoveries for the matrix spike of low molecular PBDEs were between 70 and 108% and ranged from BDE-3 to BDE-191, whereby the high molecular weight BDE-209 made up 51%. The detection limit, deﬁned as a signal of three times the noise level, was 0.5 ng/g (dw) for PCBs, and 0.1 ng/g (dw) for PBDEs. 2.4. Risk characterization and estimation Cancer risks via ingestion, dermal contact and inhalation of soil particles were estimated, based on the following Eqs. (1) and (2) [29] and Eq. (3) [30]. Cancer riskingest =

Csoil × IngR × EF × ED × CF × SFO BW × AT

(1)

where Cancer riskingest is cancer risk via ingestion of soil, C is soil concentration of the contaminant in soil (mg/kg), IngR is ingestion rate of soil (mg/day), EF is exposure frequency (days/year), ED is exposure duration (years), BW is average body weight (kg), AT is averaging time (days), CF is conversion factor (1 × 10−6 kg/mg), SFO is oral slope factor (mg/kg/day)−1 . Cancer riskdermal =

Csoil × SA × AFsoil × ABS × EF × ED BW × AT ×CF × SFO × GIABS

(2)

where Cancer riskdermal is cancer risk via dermal contact of soil, SA is surface area of the skin that contacts the soil (cm2 ), AFsoil is skin adherence factor for soil (mg/cm2 ), ABS is dermal absorption factor (chemical speciﬁc), GIABS is gastrointestinal absorption factor. Cancer riskinhale =

Csoil × EF × ED × IUR PEF × AT

updated human health evaluation manual of 2009 for the ventilation rate and body weight to be applied in Eq. (3) [30], as “the amount of the chemical that reaches the target site of the chemical through the inhalation pathway is not the simple function of the ventilation rate and body weight”. Therefore, the ventilation rate and body weight were excluded from Eq. (3). The estimation of cancer risk via ingestion, dermal contact and soil inhalation was based on a human lifespan of 70 years, and a slope factor (SF) that can be deﬁned as the human cancer risk per unit (mg/kg/day) dose, obtained from either animal bioassays or human data [29]. According to the Human Health Evaluation Manual [34], cancer risks can increase across different exposure pathways, but only when assessing the risks for the same individuals. Hence, the estimation of combined cancer risks through ingestion, dermal contact and inhalation may be applied in this study by adding the results of Eqs. (1)–(3) together. Furthermore, the total and bioaccessible concentrations of BDE-209 were used only to conduct the cancer risk assessment through ingestion and dermal contact of soils, as no IUR of BDE-209 was available from the US EPA [35]. In contrast, the total and bioaccessible PCB concentrations were utilized to conduct the cancer risk assessment of 3 different exposure pathways, namely ingestion, dermal contact and inhalation [35]. The SF for both BDE-209 and PCBs are listed in Table 2. Lifetime cancer risks can be qualitatively described with the following of: very low when the estimated value is ≤10−6 ; low in the range of 10−6 < to <10−4 ; moderate in the range of 10−4 ≤ to <10−3 ; high in the range of 10−3 ≤ to <10−1 ; and very high when the value is ≥10−1 [36].

(3)

where Cancer riskinhale is cancer risk via inhalation of soil particles, InhR is inhalation rate (m3 /day), IUR is inhalation unit risk (mg/m3 )−1 = slope factor via inhalation, PEF is particle emission factor = 1.36 × 109 m3 /kg. The PEF concerns the inhalation of pollutants adsorbed to respirable particles (PM10 ) [31]. A conservative soil ingestion rate (IngR) of 100 mg/day was recommended for adults [32], which was based on an exposure duration (ED) of 70 years (life time exposure period) and an assumed exposure frequency (EF) of 350 days/year. The EF took into account the working pattern of workers and farmers toiling the farmland all year round, in which the average time (AT) was calculated (excluding the 15 days of holiday) as (AT) = ED × 365 = 25,550 days. In this study, a body weight of 60 kg per adult worker was selected to accommodate the current local situation [33], in which the contact surface area of skin with soil (SA) was deemed as 3300 cm2 , while the skin adherence factor of soil (AFsoil ) was 0.2 mg/cm2 [29]. It was not recommended in the

2.5. In vitro digestion model to extract bioaccessible PBDEs and PCBs The cancer risks from total BDE-209 were classiﬁed as very low amongst the 6 types of land use and so no analysis of bioaccessible PBDEs was completed. The bioaccessible PCBs were established by the physiologically based extraction test described in Ruby et al. [37], with a slight modiﬁcation, which involved simulating the conditions of both the human stomach and intestine [38]. The gastric solution used in this study consisted of a mixture of 17.55 g of NaCl, 1.0 g of citrate, 1.0 g of malate, 0.85 mL of lactic acid, 1.0 mL of acetic acid, and 2.5 g of pepsin (P7000, Sigma Chemical Co.) into 2 L of deionised water, which was adjusted to pH 1.5 with 12 M HCl. Then, 1 g of soil was added into a 50 mL plastic centrifuge tube followed by 30 mL of gastric solution. This mixture was then shaken in a shaking incubator (SHEL LAB 1575 R) above 55 rpm for 1 h at 37 ◦ C and subjected to simulated intestinal conditions by adjusting the pH to 7.0 with 1 M NaOH. Meanwhile, 0.06 g of porcine bile extract (B8631, Sigma Chemical Co.) and 0.018 g of porcine pancreatin (P1500, Sigma Chemical Co.) were added to each tube. These samples were also shaken with the same shaking incubator above 55 rpm for 4 h at 37 ◦ C, during the intestinal condition simulation. Next, the samples were centrifuged at 3300 rpm for 10 min at 37 ◦ C and ﬁltered with an Advantec 5C ﬁlter paper. Finally, the ﬁltrate was diluted and topped up to 35 mL with deionised water.

US EPA [35] US EPA [35] – – 430,000 250,0000 16,000 200,000 7800 100,000 7800 100,000 180,000 180,0000

HK, China HK, China HK, China HK, China HK, China HK, China HK, China HK, China HK, China HK, China Guiyu, China Guiyu, China Guiyu, China Guiyu, China

US EPA US EPA

Organic farm (OF) Agricultural (A) E-waste storage EW (S) Open burning site (OBS) E-waste dismantling workshop EW (DW) E-waste open burning site EW (OBS) Within e-waste storage site Outside e-waste storage site Pak Heung e-waste storage site Fanling e-waste storage site Ash 1 (charred plastic parts) Ash 2 (mixture of ash, mud, sand) Ash 3 (mixture of ash, mud, sand) Combusted residue

Regional screening level Resident soil Industrial soil

1 2 1 1 1 3

120,000 120,0000

23.5 27.5 50.5 28,111 6875 32,337 – – 32,746 274–306 9224 8291 12,094 63,300 0 2.24 0 55.2 1542 5806 – – 4610 148–162 – – – 48,633 0 0.902 2.06 14.8 117 799 – – – – – – – – 0 0.714 4.99 9.3 236 1410 – – 478 15.2–22.2 758 1270 521 24.0 1.62 1.77 5.67 12.9 384 2287 – – 274 9.57–16 899 1350 372 15.4 2.18 7.07 0 4453 787 5731 – – – – – – – –

BDE-99 BDE-47 Octa Penta Sample no. Country/region Types of soil

The two PBDE congeners namely, BDE-47 and BDE-99, attract the most public concern due to their toxicity and persistence, with reports implying that they can disturb the activities of thyroid hormones and neurobehavioral development [40,41]. Amongst the 6 sampling sites, EW (OBS) contained the highest concentration of BDE-47 (2287 ␮g/kg) and BDE-99 (1410 ␮g/kg) (Table 3). However, when the Guiyu study was compared with this study [39], it was deduced that the former site contained higher total PBDEs of 63,300 ␮g/kg, but lower BDE-47 of 15.4 ␮g/kg and BDE99 of 24 ␮g/kg, than the latter site of 32,337 ␮g/kg, 2287 ␮g/kg and 1410 ␮g/kg, respectively (Table 3). The commercial products of PBDEs include BDE-99 (also known as penta-BDE), octa-BDE and deca–BDE. The soil of EW (OBS) and OBS contained relatively high concentrations of BDE-47 and BDE-99 than Guiyu, which may be attributed to the e-waste soil samples containing mostly BDE-99 in Hong Kong and also the degradation of BDE99 to BDE-47 (also known as tetra-BDE). However, since only 5 samples were taken for each studied land use type, a greater number of soil samples from the EW (OBS) and OBS have to be taken for analysis in order to prove this conjecture. In addition, the US EPA imposed guidelines on PBDEs for residential (BDE-47 = 7800 ␮g/kg; BDE-99 = 7800 ␮g/kg) and industrial soils (BDE-47 = 100,000 ␮g/kg; BDE-99 = 100,000 ␮g/kg) [35], where the concentrations in the samples of the current study were below these standards. However, it is important to point out that these standards are not speciﬁc for agricultural soils but it is envisaged that more stringent guidelines should be adopted for these types of soils. Electronic waste dismantling workshop (1061 ␮g/kg) contained the highest concentration of total PCBs (Table 4), which far exceeded the background surface soil from other places including Hong Kong (2.45 ␮g/kg) [42], China (0.515 ␮g/kg) [43], North America (196 ␮g/kg) [44] and other parts of the world (5.41 ␮g/kg) [45], as well as, the Canadian soil quality guideline for environmental health (500 ␮g/kg) [46]. Electronic waste dismantling workshop also demonstrated the most elevated total of the 6 PCB indicators and 7 PCB indicators. The total 6 PCB indicators (57.7 ␮g/kg) found in EW (DW)

Table 3 Summary of PBDEs concentrations in this and other studies, and the regional screening level (RSL) tables (␮g/kg) taken from the US EPA [35].

3. Results and discussion

BDE-153

The minimum, median and maximum cancer risks of PCBs via the ingestion pathway for each land use type were chosen for determining their bioaccessible PCB concentrations: 3 (minimum, median and maximum) × 6 types of land use = 18 soil samples, in which each sample was performed in duplicates. The SRM 2585 (Organic Contaminants in House Dust) and an analytical blank were also included in every batch of extraction, and the mean bioaccessible PCB recoveries of SRM 2585 found to range from 0.71 to 29.7%. The cancer risk assessment of bioaccessible PCBs was estimated by the same method in Section 2.4.

2.28 3.02 14.6 28.6 375 1993 7.5–190 16–89 607 18.9–28 2080 2100 2090 64.4

BDE-209

2.6. Selection of soil samples for the In vitro digestion model, bioaccessible PCB recoveries and risk assessment of bioaccessible PAHs

5 5 5 5 5 5

Total BDEs

Reference

Bioaccessible PCBs were subsequently extracted by using the liquid–liquid extraction method. The 35 mL ﬁltrate was mixed with 35 mL of DCM and acetone (v:v 1:1) for 2 min in a separatory funnel and left to stand for half an hour for extraction to occur. The organic sample extracts were then collected and excess anhydrous sodium sulphate was used to absorb any water left in the extracts. The sample extracts were concentrated to 1 mL by a rotary evaporator and the same cleanup and analytical procedures were applied as for the total PCBs (Section 2.2).

95

This study This study This study This study This study This study EPD [62] EPD [62] Greenpeace China [63] Greenpeace China [63] Luksemburg et al. [64] Luksemburg et al. [64] Luksemburg et al. [64] Leung et al. [39]

Y.B. Man et al. / Journal of Hazardous Materials 195 (2011) 92–99

500 1000 20 Netherlands Netherlands Canada Soil quality standards Dutch Target Value Dutch Intervention Value SQEE

Note: In this study: Total of 6 PCB indicators (PCB 28, 52, 101, 138, 153 (+157) and 180) and 7 PCB indicators (PCB 28, 52, 101, 118, 138, 153 (+157) and 180). Other studies: Total of 6 PCB indicators (PCB 28, 52, 101, 138, 153 and 180) and 7 PCB indicators (PCB 28, 52, 101, 118, 138, 153 and 180). SQEE = soil quality guideline for environmental health and N.D. = not detected.

This study This study This study This study This study This study Zhang et al. [42] Ren et al. [43] Meijer et al. [45] Wilcke and Amelung [44]

VROM [47] VROM [47] CCME [46]

Reference

3.25 (2.92–3.38) 5.96 (2.67–16.5) 9.53 (3.5–22.8) 37.7 (19.5–64.2) 1061 (7.13–14,542) 144 (2.51–110) 2.45 (0.04–9.87) 0.515 (0.138–1.84) 5.41 (0.0260–96.6) 196 (7.9–3136) 1.27 (1.08–1.38) 1.22 (0.887–1.48) 3.69 (1.54–8.41) 22.3 (11.3–36.1) 64.3 (448–3.05) 49.5 (4.55–127) 2.45 (0.04–9.87) 0.110 (N.D.–1.06) 2.91 (0.00950–57.91) 185 (6.9–3042) 1.11 (0.945–1.22) 1.42 (0.926–2.76) 3.09 (1.25–6.68) 2.16 (N.D.-8.80) 57.7 (2.56–404) 41.5 (2.51–110) 2.42 (0.04–9.87) 0.0945 (N.D.–0.799) 2.47 (0.00900–51.2) 173 (5.2–2839)

Total 6 indicator PCBs

HK, China HK, China HK, China HK, China HK, China HK, China HK, China China Global North America Organic farm Agricultural E-waste storage Open burning site E-waste dismantling workshop E-waste open burning stie Hong Kong surface soil Chinese surface soil Global background surface soil

Sample no. Country/region Types of soil

5 5 5 5 5 5 7 52 191 18

Total 7 indicator PCBs

Total PCBs

Y.B. Man et al. / Journal of Hazardous Materials 195 (2011) 92–99

Table 4 Summary of PCB concentrations in this and other studies, and the soil quality standards (␮g/kg) of VROM (2000) and CCME (1991) [46,47].

96

Table 5 Total cancer risks from exposure via ingestion and dermal contact of soils in humans from different types of agricultural land use, based on the total BDE-209 concentrations at the 5th, median and 95th centiles. Sampling site

Organic farm Agricultural E-waste storage Open burning site E-waste dismantling workshop E-waste open burning site

Total cancer risks from exposure via ingestion and dermal contact 5th centile

Median

95th centile

N.D. N.D. N.D. N.D. N.D. N.D.

N.D. 2.08E−05 6.62E−05 N.D. 2.98E−03 9.25E−03

N.D. 2.08E−05 3.68E−04 N.D. 5.60E−03 1.70E−02

Note: N.D. means not detected and the values of cancer risk are in the unit of 10−6 .

were more than twice the Dutch Target Value (20 ␮g/kg) [47], indicating its potential health risk. Nonetheless, EW (DW) contained a lower total 7 PCB indicator concentration (64.3 ␮g/kg) than the Dutch Intervention Value (1000 ␮g/kg) [47]. The demolition of electronic components during e-waste dismantling activities causes the release of PCBs into the soils, which may have resulted in the detection of high PCB concentrations in EW (DW). In another e-waste dissembling site in Zhejiang, China, cable coating was revealed to contain the highest concentration of total PCBs (680 ␮g/kg dw) [48]. This was attributed to the addition of PCBs into polyvinyl chloride in order to improve the insulating ability of highvoltage cables [49]. Consequently, the dismantling of cables most probably will lead to high PCB concentrations in the surrounding environment. On the other hand, signiﬁcantly higher soil organic matter (SOM) was noticed in the soils of OBS, EW (DW) and EW (OBS) (6.11, 8.63 and 6.30%, respectively) (Table S1 – supplementary data). In addition, there were positive correlations between SOM and total PCBs (r = 0.357; p < 0.05) and PBDEs (r = 0.318; p < 0.05) (Table S2 – supplementary data). SOM may be able to bind the PCBs and PBDEs once they had been deposited into the soils [50,51]. The strong sorption may render the PBDEs and PCBs more resistant to degradation and leaching from soils [50,51]. Table S3 shows the supplementary data for the cancer risks via ingestion and dermal contact of soils on humans, based on the total BDE-209 concentration. The carcinogenic risks for PBDEs by combining the above exposure pathways are displayed in Table 5. Rodent experiments have illustrated that the intake of BDE-209 in contaminated food may cause liver tumors [52], and in the present study only BDE-209 was considered for cancer risk assessment, as there were no SF for other PBDEs. Furthermore, there was no SF for BDE-209 via the exposure pathway of inhalation (Table 2) due to the scarcity of inhalation toxicity factors needed to obtain IUR (mg/m3 )−1 [53]. Therefore, the cancer risks of BDE-209 via inhalation were not taken into account for this study. The results show that the cancer risk values of the soils from all the 6 sites were below 1 in a million people at the 5th, 50th and 95th centiles (Table 5), which means that the cancer risks imposed by BDE-209 are very low. At the 95th centile, soils from OF and EW (S) did not show any cancer risks as BDE-209 was not detected. While in the case of the other four land use types the cancer risk trend was as follows: EW (OBS) > EW (DW) > OBS > OF in both ingestion and dermal contact pathways. For both pathways, even the highest cancer risk for PBDEs of EW (OBS) (0.0102 × 10−6 for ingestion; 0.00676 × 10−6 for dermal contact), calculated to be 3 times greater than the other types of land use, and still grouped under the category of ‘very low’. The higher PBDE carcinogenic risk of the soil from EW (OBS) may be due to the burning activities of e-waste, as PBDEs are commonly applied to circuit boards and as coatings of ﬂame retardants [3]. Open burning of the PBDE containing

Y.B. Man et al. / Journal of Hazardous Materials 195 (2011) 92–99 Table 6 Total cancer risks from exposure via ingestion, dermal contact and inhalation of soils in humans from different types of agricultural land use based on the total PCBs concentrations at the 5th, median and 95th centiles. Sampling sites

Organic farm Agricultural E-waste storage Open burning site E-waste dismantling workshop E-waste open burning site

Total cancer risks from exposure via ingestion, dermal contact and inhalation 5th centile

Median

95th centile

0.018 0.0164 0.0215 0.120 0.0438 0.0445

0.0198 0.0242 0.0418 0.229 0.107 0.183

0.0222 0.102 0.145 0.389 40 2.1

Note: Cancer risks are in bold and the values of cancer risk are in the unit of 10−6 .

e-waste may thus be the major cause for the release of PBDEs and its associated elevated carcinogenic risk. Positive correlations were observed between the concentrations of PBDEs in ﬁsh (tilapia and bug head) and sediment from Guiyu [54]. Since PBDEs can be bioaccumulated and biomagniﬁed in the ecosystem [55], they may be accumulated in human bodies and cause cancer after sustained ingestion of contaminated ﬁsh. It has been clearly shown in Guiyu that PBDE contamination can affect humans and cause adverse health effects via various pathways, such as ingestion of contaminated ﬁsh, hence, PBDE pollution must not be overlooked in Hong Kong. The carcinogenic risks for PCBs were estimated via a tiered approach, which depended on different levels of human SF for environmental PCBs, through various exposure pathways. In this study, the PCB cancer risks were estimated using tiers of high risk and persistence (a relatively high SF), based on the criteria of ingestion, dermal contact and inhalation of soil being within this tier [56]. The estimated cancer risks for PCBs are given in Table 6 with the supplementary data included in Table S4. All 6 types of land use exhibited very low carcinogenic risks at the 5th and 50th centiles, while EW (DW) and EW (OBS) showed low cancer risks through ingestion and dermal contact at the 95th centile. Higher cancer risks were found in EW (DW) (ingestion: 20.8 × 10−6 ; dermal contact: 19.2 × 10−6 ; inhalation: 0.000873 × 10−6 ) and EW (OBS) (ingestion: 1.09 × 10−6 ; dermal contact: 1.01 × 10−6 ; inhalation: 4.59 × 10−11 ) when compared to the other types of land uses, which may be a consequence from the release of PCBs from ewaste. PCBs are commonly detected at e-waste recycling sites since they are used in the manufacturing of capacitors and transformers [1] and the dismantling and burning activities promote its leakage into surrounding soils. For the exposure pathways, the trend of decreasing cancer risks was as follows: ingestion > dermal contact > inhalation. Inhalation presented the lowest cancer risks (EW (DW): 0.000873 × 10−6 ; EW (OBS): 4.59 × 10−11 ) because PCBs possess a low volatility [56].

97

Ingestion is a signiﬁcant exposure pathway for PCBs usually due to involuntary consumption of soil. As mentioned above, cancer risks for PCBs through ingestion (EW (DW): 20.8 × 10−6 ; EW (OBS): 1.09 × 10−6 ) were relatively high when compared to dermal contact and inhalation. Consequently, an in vitro system that simulated the human digestive system, was adopted in the current study in order to investigate the amount of bioaccessible carcinogenic PCBs in the soil samples. According to Table 7, it can be observed that the cancer risk for bioaccessible PCBs from EW (DW) was signiﬁcantly higher than the other types of soil (mean = 1.71 × 10−6 ± 2.96 × 10−6 ), implying that the bioaccessible PCBs concentrations in EW (DW) may pose potential cancer risk in humans. The major exposure pathway of PBDEs and PCBs to workers or farmers in OF, A, EW (S) and EW (DW) was discovered to be dermal contact. But the pathway in OBS and EW (OBS) in contrast was via inhalation, as the combustion activities in these land use types tend to generate ultra ﬁne particles less than PM0.1 , which can penetrate deeply into the lungs and cause adverse health effects [57]. The estimation of health risks via inhalation should be based on pollutants adsorbed onto respirable particles of soils (less than PM10 ) [31]. Only the inhaled soil particles with a size of less than PM10 can be deposited in the upper part of the respiratory tract or penetrate deeply into the lungs [58,59]. Fine soil particles (less than PM10 ) with organic pollutants (such as PAHs) and inorganic pollutants (such as Cu, Cd and Zn), may be able to cause oxidative stress and inﬂammation after penetrating into the lungs [60,61]. This study used soil particles with a diameter of less than 2 mm to estimate cancer health risks in humans via the exposure route of inhalation, implying that not all soil particles were able to penetrate into the lungs. In addition, the concentrations of pollutants in soil particles with a diameter of less than 2 mm should be lower than the particles smaller than PM10 . Consequently, the human health risks based on pollutant concentrations would more than likely be underestimated. Furthermore, the absent IUR of BDE209 causes this evaluation of cancer risks to be underestimated. There is a need to derive pollutant toxicity values based on the inhalation pathway from experimental data in order to ﬁll the gap of risk assessment via this means especially in the OBS and EW (OBS). Evaluating the health risks by using bioavailable pollutant concentrations is commonly regarded as the most accurate way, because only the bioavailable portion of the contaminants will ultimately reach our bloodstream and exert adverse effects on our body [20]. However, this method usually brings along ethical concerns due to the involvement of animal experiments, therefore, assessing bioaccessible fractions of pollutants may be a suitable alternative in portraying the reality [20]. In this study, bioaccessible fractions of PCBs were used to estimate the health risks via ingestion using an in vitro digestion model. No bioaccessible fractions of pollutants were used in the cancer risk estimations of the other two studied pathways of the present

Table 7 Cancer risks via ingestion of soils in humans from different types of agricultural land uses based on bioaccessible PCB toxic equivalent concentrations at minimum, medium, maximum and mean. Sampling sites

Organic farm Agricultural E-waste storage Open burning site E-waste dismantling workshop E-waste open burning site

Cancer risks via ingestion Min

Median

Max

Mean

N.D. N.D. N.D. N.D. N.D. N.D.

N.D. N.D. N.D. 0.0203 0.00136 0.00611

N.D. 0.0152 0.00359 0.0365 5.13 0.139

N.D. 0.00507 0.00120 0.0189 1.71 0.0483

Note: cancer risk are in bold, N.D. means not detected and the values of cancer risk are in the unit of 10−6 .

± ± ± ± ±

0.00879 0.00207 0.0183 2.96 0.0784

98

Y.B. Man et al. / Journal of Hazardous Materials 195 (2011) 92–99

study, consequently, the cancer risks from PCBs may be underestimated. 4. Conclusion Inhalation of soil particles is the major exposure pathway of PBDEs and PCBs to humans from OBS and EW (OBS). Whereas, the major exposure pathway of other land use types including (OF, A. EW (S) and EW (DW)) is via dermal contact of soils. Soils from EW (DW) and EW (OBS) were of the greatest concern in terms of threatening human health as they contained the highest concentrations of PCBs and PBDEs, resulting in relatively high cancer risks amongst the 6 types of land use. The burning and dismantling activities in e-waste sites may still potentially pose cancer risks to humans. Although the cancer risks of PBDEs via the exposure pathways of ingestion and dermal contact of soils in EW (DW) and EW (OBS) were still very low, these two pathways were not the major exposure pathways in EW (OBS). Hence, regular monitoring is required as these pollutants may be continuously deposited on soils and eventually accumulated to hazardous levels. Acknowledgements The authors would like to thank the Public Policy Research Grants (2002-PPR-3), Special Equipment Grant (SEG HKBUO09) of the Research Grants Council of Hong Kong and the Mini-AoE (Areas of Excellence) Fund from Hong Kong Baptist University (RC/AoE/08–09/01) for ﬁnancial support. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.jhazmat.2011.08.010. References [1] H. Liu, Q. Zhou, Y. Wang, Q. Zhang, Z. Cai, G. Jiang, E-waste recycling induced polybrominated diphenyl ethers, polychlorinated biphenyls, polychlorinated dibenzo-p-dioxins and dibenzo-furans pollution in the ambient environment, Environ. Int. 34 (2008) 67–72. [2] M.H. Wong, S.C. Wu, W.J. Deng, X.Z. Yu, Q. Luo, A.O.W. Leung, C.S.C. Wong, W.J. Luksemburg, A.S. Wong, Export of toxic chemicals – a review of the case of uncontrolled electronic-waste recycling, Environ. Pollut. 149 (2007) 131–140. [3] F. Rahman, K.H. Langford, M.D. Scrimshaw, J.N. Lester, Polybrominated diphenyl ether (PBDE) ﬂame retardants, Sci. Total Environ. 275 (2001) 1–17. [4] US EPA (United States Environmental Protection Agency), Polychlorinated Biphenyls (PCBs) Update: Impact on Fish Advisories, EPA-823-F-99-019, Environmental Protection Agency, Washington, 1999. [5] J.C.W. Lam, R.K.F. Lau, M.B. Murphy, P.K.S. Lam, Temporal trends of hexabromocyclododecanes (HBCDs) and polybrominated diphenyl ethers (PBDEs) and detection of two novel ﬂame retardants in marine mammals from Hong Kong, South China, Environ. Sci. Technol. 43 (2009) 6944–6949. [6] C.K.C. Wong, K.M. Leung, B.H.T. Poon, C.Y. Lan, M.H. Wong, Organochlorine hydrocarbons in human breast milk collected in Hong Kong and Guangzhou, Arch. Environ. Contam. Toxicol. 43 (2002) 364–372. [7] J. De Boer, P.G. Wester, H.J.C. Klamer, W.E. Lewis, J.P. Boon, Do ﬂame retardants threaten ocean life? Nature 394 (1998) 28–29. [8] J.E. Goodman, Neurodevelopmental effects of decabromodiphenyl ether (BDE209) and implications for the reference dose, Regulat. Toxicol. Pharmacol. 54 (2009) 91–104. [9] ATSDR (Agency for toxic substances and disease registry), Toxicological Proﬁle for Polybrominated Diphenyls and Polybrominated Diphenyl Ethers-draft for Public Comment, Agency for Toxic Substances and Disease Registry, US Department of Health and Human Services, Public Health Service, Atlanta, 2002. [10] T.A. McDonald, A perspective on the potential health risks of PBDEs, Chemosphere 46 (2002) 745–755. [11] L.S. Birnbaum, D.F. Staskal, Brominated ﬂame retardents: a cause for concern? Environ. Health Perspect. 12 (2004) 9–17. [12] US EPA (United States Environmental Protection Agency), Polychlorinated Biphenyls (PCBs), Basic Information, 2009. [13] C.R. Bryant, L.H. Russwurm, A.G. McLellan, The City’s Countryside: Land and its Management in the Rural-Urban Fringe, Longman, New York, 1982. [14] T. Lindstrom, E. Hansen, H. Juslin, Forest certiﬁcation: the view from Europe’s NIPFs, J. Forest. 97 (1999) 25–30.

[15] C.Y. Jim, Rural blight and land use planning in Hong Kong, Environment 17 (1997) 269–281. [16] Y.W. Li, Agricultural Land in Hong Kong: A Solution Space for Urban Development, MSc. Dissertation. The University of Hong Kong, Hong Kong, 1998. [17] J.P. Wu, X.J. Luo, Y. Zhang, Y. Luo, S.J. Chen, B.X. Mai, Z.Y. Yang, Bioaccumulation of polybrominated diphenyl ethers (PBDEs) and polychlorinated biphenyls (PCBs) in wild aquatic species from an electronic waste (e-waste) recycling site in South China, Environ. Int. 34 (2008) 1109–1113. [18] J. She, A. Holden, M. Sharp, M. Tanner, C. Williams-Derry, K. Hooper, Polybrominated diphenyl ethers (PBDEs) and polychlorinated biphenyls (PCBs) in breast milk from the Paciﬁc Northwest, Chemosphere 67 (2007) S307–S317. [19] A. Mazdai, N.G. Dodder, M.P. Abernathy, R.A. Hites, R.M. Bigsby, Polybrominated diphenyl ethers in maternal and fetal blood samples, Environ. Health Perspect. 111 (2003) 1249–1252. [20] Environment Agency, UK In-vitro Methods for the Measurement of the Oral Bioaccessibility of Selected Metals and Metalloids in Soils: A Critical Review, Environment Agency, Bristol, Britain, 2002. [21] B.N. Lopez, Y.B. Man, Y.G. Zhao, J.S. Zheng, A.O.W. Leung, J. Yao, M.H. Wong, Major pollutants in soils of abandoned agricultural land contaminated by e-waste activities in Hong Kong, Arch. Environ. Contam. Toxicol. 61 (2011) 101–114. [22] S.E. Allen, Chemical Analysis of Ecological Materials, second ed., Blackwell Scientiﬁc Publications, Oxford England, Boston, 1989. [23] US EPA (United States Environmental Protection Agency), Method for Determination of Total Organic Carbon (TOC) in Soils and Sediments Method, Environmental Protection Agency, Washington, 2002. [24] US EPA (United States Environmental Protection Agency), Method 3540C: Soxhlet Extraction, Environmental Protection Agency, Washington, 1996. [25] US EPA (United States Environmental Protection Agency), Method 3620B: Florisil Cleanup, Environmental Protection Agency, Washington, 1996. [26] US EPA (United States Environmental Protection Agency), Method 8270C: Semivolatile Organic Compounds by Gas Chromatography/Mass Spectrometry (GC/MS), Environmental Protection Agency, Washington, 1996. [27] US EPA (United States Environmental Protection Agency), Ofﬁce of Water, Ofﬁce of Science and Technology, Engineering and Analysis Division (4303T), EPA Method 1614, Brominated Diphenyl Ethers in Water, Soil, Sediment and Tissue by HRGC-HRMS, Environmental Protection Agency, Washington, 2007. [28] M.H. Zheng, Progress and prospect for dioxins analysis in China, in: The China International Symposium on Persistent Toxic Substances, Beijing, China, 2004, pp. 7–11. [29] US EPA (United States Environmental Protection Agency), Exposure Factors Handbook. EPA/600/P-95/002F, Environmental Protection Agency, Washington, 1997. [30] US EPA (United States Environmental Protection Agency), Risk Assessment Guidance for Superfund, Vol. I: Human Health Evaluation Manual (F, Supplemental Guidance for Inhalation Risk Assessment) EPA/540/R/070/002, Environmental Protection Agency, Washington, 2009. [31] US EPA (United States Environmental Protection Agency), Supplemental Guidance for Developing Soil Screening Levels for Superfund Sites, OSWER 9355. 4-24, Environmental Protection Agency, Ofﬁce of Solid Waste and Emergency Response, Washington, 2001. [32] E.J. Calabrese, P.T. Gilbert, C.E. Kostecki, How much dirt do children eat? An emerging environmental health question, Comment. Toxicol. 1 (1987) 229–241. [33] M.M. Lee, A. Wuwilliams, A.S. Whittemore, S. Zheng, R. Gallagher, C.Z.et al. Teh, Comparison of dietary habits physical-activity and body-size among Chinese in North-America and China, Int. J. Epidemiol. 23 (1994) 984–990. [34] US EPA (United States Environmental Protection Agency), Risk Assessment Guidance for Superfund, Vol. I: Human Health Evaluation Manual, EPA/540/189/002, Environmental Protection Agency, Ofﬁce of Solid Waste and Emergency Response, Washington, 1989. [35] US EPA (United States Environmental Protection Agency), Mid Atlantic Risk Assessment, Regional Screening Level (RSL) Summary Table, Environmental Protection Agency, Washington, 2010. [36] ATSDR (New York State Department of Health under cooperative agreement with the Agency for Toxic Substances and Disease Registry), Public Health Assessment. Johnstown City Landﬁll, Johnstown, Fulton Country, (CERCLIS NO. NYD980506927), Department of Health and Human Services, Public Health Service, Atlanta, 1995. [37] M.V. Ruby, A. Davis, R. Schoof, S. Eberle, C.M. Sellstone, Estimation of lead and arsenic bioavailability using a physiologically based extraction test, Environ. Sci. Technol. 30 (1996) 422–430. [38] X. Tang, L. Tang, Y. Zhu, B. Xing, J. Duan, M. Zheng, Assessment of the bioaccessibility of polycyclic aromatic hydrocarbons in soils from Beijing using an in vitro test, Environ. Pollut. 140 (2006) 279–285. [39] A.O.W. Leung, W.J. Luksemburg, A. Wong, A. Wong, Spatial distribution of polybrominated diphenyl ethers and polychlorinated dibenzo-p-dioxins and dibenzofurans in soil and combusted residue at guiyu, an electronic waste recycling site in Southeast China, Environ. Sci. Technol. 41 (2007) 2730–2737. [40] S.C. Lema, J.T. Dickey, I.R. Schultz, P. Swanson, Dietary exposure to 2,2,4,4tetrabromodiphenyl ether (PBDE-47) alters thyroid status and thyroid hormone–regulated gene transcription in the pituitary and brain, Environ. Health Perspect. 116 (2008) 12. [41] J. Cheng, J. Gu, J. Ma, X. Chen, M. Zhang, W. Wang, Neurobehavioural effects, redox responses and tissue distribution in rat offspring developmental exposure to BDE-99, Chemosphere 75 (2009) 963–968.

Y.B. Man et al. / Journal of Hazardous Materials 195 (2011) 92–99 [42] H.B. Zhang, Y.M. Luo, M.H. Wong, Q.G. Zhao, G.L. Zhang, Concentration and possible sources of polychlorinated biphenyls in the soils of Hong Kong, Geoderma 138 (2007) 244–251. [43] N. Ren, M. Que, Y.F. Li, Y. Liu, X. Wan, D. Xu, E. Sverko, J. Ma, Polychlorinated biphenyls in Chinese surface soils, Environ. Sci. Technol. 41 (2007) 3871–3876. [44] W. Wilcke, W. Amelung, Persistent organic pollutants in native grassland soils along a climosequence in North America, Soil Sci. Soc. Am. J. 64 (2000) 2140–2148. [45] S.N. Meijer, W.A. Ockenden, A. Sweetman, K. Breivik, J.O. Grimalt, K.C. Jones, Global distribution and budget of PCBs and HCB in background surface soils: Implications for sources and environmental processes, Environ. Sci. Technol. 37 (2003) 667–672. [46] CCME (Canadian Council of Ministers of the Environment), Canadian Environmental Quality Guideline, CCME, Canada, 1991. [47] VROM (Volkshuisvesting, Ruimtelijke Ordening en Milieubeheer) (Ministry of Housing, Spatial Planning and Environment), Circular on Target Values and Intervention Values for Soil Remediation, Spatial Planning and Environment, Ministry of Housing, Netherlands, 2000. [48] G. Zhao, Z. Wang, M.H. Dong, K. Rao, J. Luo, D. Wang, J. Zha, S. Huang, Y. Xu, M. Ma, PBBs, PBDEs, and PCBs levels in hair of residents around e-waste disassembly sites in Zhejiang Province, China, and their potential sources, Sci. Total Environ. 397 (2008) 46–57. [49] United Nations Environment Programme (UNEP), Draft Guidelines on BAT for Smouldering Copper Cable, UNEP/POPS/EGB.2/INF/12, UNEP, Kenya, 2003. [50] S.A. Boyd, S. Sun, Residual petroleum and polychlorobiphenyl oils as sorptive phases for organic contaminants in soils, Environ. Sci. Technol. 24 (1990) 142–144. [51] B.N. Zegers, W.E. Lewis, K. Booij, R.H. Smittenberg, W. Boer, J.P.B.J. deBoon, Levels of polybrominated diphenyl ether ﬂame retardants in sediment cores from Western Europe, Environ. Sci. Technol. 37 (2003) 3803–3807. [52] ATSDR (Agency for toxic substances and disease registry), Polybrominated Diphenyl Ethers, Agency for Toxic Substances and Disease Registry, US Department of Health and Human Services, Public Health Service, Atlanta, 2004. [53] US EPA (United States Environmental Protection Agency), Mid Atlantic Risk Assessment, Frequently Asked Questions, Environmental Protection Agency, Washington, 2011.

99

[54] Q. Luo, Z.W. Cai, M.H. Wong, Polybrominated diphenyl ethers in ﬁsh and sediment from river polluted by electronic waste, Sci. Total Environ. 383 (2007) 115–127. [55] J. Boon, W. Lewis, A. Tjoen, C. Allchin, R. Law, J. de Boer, Levels of polybrominated diphenyl ether (PBDE) ﬂame retardants in animals representing different trophic levels of the North Sea food web, Environ. Sci. Technol. 36 (2002) 4025–4032. [56] US EPA (United States Environmental Protection Agency), Polychlorinated Biphenyls (PCBs), CASRN 1336-36-3, Environmental Protection Agency, Washington, 1997. [57] G.R. Cass, L.A. Hughes, P. Bhave, M.J. Kleeman, J.O. Allen, L.G. Salmon, The chemical composition of atmospheric ultraﬁne particles, Philos. Trans. R. Soc. London A: Math. Phys. Eng. Sci. 358 (2000) 2581–2592. [58] J. Ferin, Pulmonary retention and clearance of particles, Toxicol. Lett. 72 (1994) 121–125. [59] C.S. Kim, S.C. Hu, Regional deposition of inhaled particles in human lungs: comparison between men and women, J. Appl. Physiol. 84 (1998) 1834–1844. [60] P.T. Scheepers, R.P. Bos, Combustion of diesel fuel from a toxicological perspective II, toxicity, Int. Arch. Occup. Environ. Health 64 (1992) 163–177. [61] K. Donaldson, W. MacNee, Mini-review potential mechanisms of adverse pulmonary and cardiovascular effects of particulate air pollution (PM10), Int. J. Hyg. Environ. Health 203 (2001) 411–415. [62] EPD (Hong Kong Environmental Protection Department), Press Release: Test Results of Soil Samples Collected from e-Waste Workshops, Environmental Protection Department, Hong Kong, 2005. [63] Greenpeace China, Test report for analysis for polychlorinated biphenyls (PCBs), polybrominated biphenyls (PBBs), polybrominated diphenyl ether (PBDEs) and selected metal compounds, 2005. [64] W.J. Luksemburg, R.S. Mitzel, R.G. Peterson, M.M. Maier, M. Schuld, J.M. Hedin, H.D. Zhou, A.S. Wong, Polychlorinated dibenzodioxins and dibenzofurans (PCDDs/PCDFs) and polybrominated diphenylether (PBDE) levels in environmental and human hair samples around an electronic waste processing site in Guiyu, Guangdong Province, China, in: Poster presented at 22nd International Symposium on Halogenated Environmental Organic Pollutants and Persistent Organic Pollutants (POPs) (Dioxins 2002), Spanish Council for Scientiﬁc Research Laboratory for Dioxins, Barcelona, Spain, 2002.

Journal of Hazardous Materials 195 (2011) 100–106

Contents lists available at ScienceDirect

Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat

Critical assessment of suitable methods used for determination of antibacterial properties at photocatalytic surfaces ´ b,∗ , Eva Musilová b , Jiˇrí Zita a Josef Krysa a b

Institute of Chemical Technology Prague, Department of Inorganic Technology, Technická 5, CZ-166 28 Prague, Czech Republic Institute of Chemical Technology Prague, Department of Water Technology and Environmental Engineering, Technická 5, CZ-166 28 Prague, Czech Republic

a r t i c l e

i n f o

Article history: Received 21 April 2011 Received in revised form 29 July 2011 Accepted 4 August 2011 Available online 10 August 2011 Keywords: TiO2 Antibacterial activity ISO 27447:2009(E) E. coli E. faecalis

a b s t r a c t This work describes the development of methods necessary for antibacterial effect evaluation on irradiated TiO2 layers. Two methods using bacteria suspensions and the glass adhesion method (based on ISO 27447:2009(E)) were critically assessed and compared. As test bacteria gram negative Escherichia coli and gram positive Enterococcus faecalis were employed. The method using 50 cm3 of bacteria suspension is convenient for testing layers with strong antibacterial effect (prepared from powder photocatalysts). For the evaluation of the antibacterial effect of sol gel layers, the glass adhesion method based on the ISO is more appropriate than the method with 3 cm3 of bacteria suspension. The reason is that the later does not allow a distinction between the inhibition effect of TiO2 and UV light itself. Some improvements of the ISO method were suggested, namely the use of gelatinous pills (CCM) of bacteria, using saline solution instead of nutrient broth for bacteria suspension preparation and the application of selective media for bacteria cultivation. Decreasing the light intensity from 0.6 mW cm−2 to 0.2 mW cm−2 (fulﬁlling the requirements of the ISO) results in almost negligible effect of UV light itself, thus enabling proper testing of the antibacterial properties of TiO2 thin ﬁlms. © 2011 Elsevier B.V. All rights reserved.

1. Introduction Titanium dioxide (TiO2 ) is one of the most popular photocatalysts [1]. In the presence of ultraviolet light (UV-A), TiO2 in anatase form is capable of decomposing organic compounds and microorganisms on its surface. Due to this ability TiO2 has high potential in many ﬁelds of application, such as medicine [2], architecture and water and air puriﬁcation [3,4]. So far chlorine is the most common agent for water disinfection. Inhibition of bacteria by chlorine is very fast and efﬁcient. However, it is well-known that chlorine reacts with organic materials (humic substances) producing chloroorganic compounds (e.g. trihalomethanes (THMS)) which are considered to be carcinogenic [5,6]. This has led to the development of alternative methods for water treatment based on the interaction of a photocatalyst with UV light [6–8]. Among the photocatalysts investigated TiO2 is the most suitable because it is stable, non-toxic and relatively cheap [9–13]. Many different microorganisms are used for antibacterial tests on photocatalytic surfaces namely, Pseudomonas aeruginosa [14,18], Enterococcus faecim [14], Candida albicans [14], Staphylococcus aureus [14,19–21], Bacillus pumilus [22] and Bacillus

∗ Corresponding author. Phone +420 220 444 112; Fax: +420 220 444 410. ´ E-mail address: [email protected] (J. Krysa). 0304-3894/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.jhazmat.2011.08.009

megaterium [23] but the most commonly used is Escherichia coli [14,16,17,23–30]. E. coli belongs to the group of Gram negative thermo tolerant coliform bacteria. Usually it appears in the digestive tract of humans and warm-blooded animals, where it is useful for the host (synthesising vitamins and supporting the overall balance of microorganisms in the intestines by suppressing the growth of harmful bacteria) [31]. E. coli usually remains harmlessly conﬁned to the intestinal lumen; however, in a debilitated or immunosuppressed host, or when gastrointestinal barriers are violated, even normal “nonpathogenic” strains of E. coli can cause infection. Infections due to pathogenic E. coli may be limited to the mucosal surfaces or may disseminate throughout the body. Three general clinical syndromes result from infection with inherently pathogenic E. coli strains: (i) urinary tract infection, (ii) sepsis/meningitis, and (iii) enteric/diarrheal disease [32]. E. coli is considered as an indicator of faecal contamination and is widely used, not only as a model microorganism for physiological, biochemical and genetic experiments, but also for antibacterial tests of different chemical substances and materials. There are many papers describing the antibacterial testing of photocatalytic surfaces, but methods and conditions are often different. The most common arrangement is an experimental setup where a drop of bacterial suspension is laid on a glass support covered by TiO2 layer [14–17]. Another experimental set-up consists

J. Kr´ ysa et al. / Journal of Hazardous Materials 195 (2011) 100–106

in the bacterial suspension placed in Teﬂon ring placed on titania thin ﬁlm [18]. The variety of the test conditions and parameters described in the literature requires the creation of a uniﬁed system for antibacterial testing. The main reason for such system is so that results obtained on different photocatalytic surfaces and in different laboratories can be easily and clearly compared. The ISO 27447:2009(E) [33] standard (Fine ceramics (advanced ceramics, advanced technical ceramics) – test method for antibacterial activity of semiconducting photocatalytic materials) was introduced in 2009, because all currently described methods for antibacterial tests of TiO2 photocatalytic layers have different procedures and conditions. The standard includes selection of suitable microorganisms and determines the conditions of the testing methods such as light intensity, amount of microorganisms and design of apparatus. The aim of the present work was the critical assessment of several methods used in laboratories worldwide for the determination of antibacterial properties of TiO2 thin ﬁlms and their comparison with the ISO standard method. The special attention was given to the question of whether it is really necessary to follow all the conditions mentioned in the ISO standard or if it is possible to adjust some conditions according to the experience and facilities of each laboratory. 2. Experimental 2.1. Chemicals For bacterial suspension preparation and for bacteria cultivation NaCl (Penta, p.a.), m-FC Agar Base (Himedia), Rosolic acid (Himedia), Slanetz and Bartley Medium (Himedia), NaOH (Penta, p.a.) were used. Titanium(IV) isopropoxide (97%; Sigma-Aldrich) and tetraethyl orthosilicate TEOS (purity 98%; Fluka) were used to prepare the titania and silica in the TiO2 /SiO2 /glass ﬁlms. Absolute ethanol (p.a. Penta) and ethyl acetoacetate (purity p.a. 99%; Fluka) were utlised as solvents and hydrochloric acid (p.a. 36%; Penta) and nitric acid (p.a. 65%; Penta) were employed as sol–gel catalysts. Evonik-Degussa P25 TiO2 powder was used for particulate layer preparation. 2.2. Preparation of TiO2 thin ﬁlms The microscope (75 × 25 × 1 mm2 ) soda-lime glass substrates were ﬁrst dip-coated (withdrawal speed: 60 mm min−1 ) into the SiO2 sol [34] to form the necessary SiO2 barrier against metal ion (mainly Na+ ) diffusion from the glass substrate into the titania ﬁlm [35]. The SiO2 interlayer was calcined at 530 ◦ C for 3 h. The titania layer was then produced by subsequent dip-coating in the titania sol [34]. After the dip-coating process, the titania ﬁlms were calcined at 530 ◦ C for 3 h. The resulting layers were around 250 nm thick and the amount of titania in each layer was around 0.04 mg cm−2 . Particulate layers were prepared by sedimentation of ultrasonically pretreated suspensions of P25 TiO2 (75 % of anatase, 25 % of rutile, crystalline size around 30 nm, BET surface area around 50 m2 g−1 ) on the same glass substrate as for sol–gel layers followed by calcination for 2 h at 300 ◦ C. The amount of photocatalyst deposited on the glass supports was 0.1, 0.2, 0.5 and 1.0 mg TiO2 cm−2 . 2.3. Microorganism used The tested microorganisms were Gram negative (G−) bacterium E. coli (CCM 3954) and Gram positive (G+) bacterium Enterococcus faecalis (CCM 4224). The pure cultures of bacteria were

101

obtained as gelatinous pills from the Czech Collection of Microorganisms (CCM), Masaryk University, Brno. The pills consist of the lyophilisated form of preserved bacteria (cca 108 CFU/ml) and the main composition of the protecting medium is gelatine. The pills must be stored at low temperature (+2 to +8 ◦ C) and used within 3 years. Before each test, it was necessary to dissolve the pill of bacteria for each culture in 9 cm3 of sterile saline solution (8.5 g dm−3 NaCl) and cultivate it for 24 h at 37 ◦ C. The bacterial suspension was than diluted with saline solution (10-times dilution method) to obtain the required concentration (CFU/ml) for each test. For the purpose of analysis, the bacterial suspension was diluted several times (10-times dilution method) to obtain the count of 30 colonies to 300 colonies in each Petri dish. To avoid contamination, selective medium m-FC agar for E. coli [36] and Slanetz-Bartley for E. faecalis [37] were used (selective media were chosen according to the water quality standards). Petri dishes with E. coli were than incubated for 24 h at 43 ◦ C and with E. faecalis for 48 h at 37 ◦ C. The number of colonies was counted and the results were expressed as the number of colony-forming units per millilitre (CFU/ml). 2.4. 50 cm3 test Particulate layers of P25 and sol–gel layers were placed in 50 cm3 of E. coli suspension. The scheme of the reactor is shown in Fig. 1A. The incident light intensity was 1.0 mW cm−2 (SYLVANIA Lynx CFS BLB, maximum at 365 nm) and the initial bacteria concentration was around 1 × 104 CFU/ml. During irradiation a 1 cm3 sample from the reaction solution was taken every 30 min. The sample was then diluted, cultivated and the results of the experiment were recorded as dependence of log(CFU/ml) versus time. To observe the effect of UV light itself, the clear glass substrate (blank) was also tested in 50 cm3 of E. coli suspension. 2.5. 3 cm3 test In this case the sol–gel TiO2 sample (25 mm × 30 mm) was placed in the small Petri dish (diameter 45 mm). Then the 3 cm3 of bacterial suspension (3.3 × 106 and 2.5 × 104 CFU/ml) was added and the dish was covered by a glass lid to minimize the vaporization (Fig. 1B). The whole system was placed on a platform shaker to insure mixing of the bacterial suspension in contact with TiO2 surface. In this test, the light intensity was 0.6 mW cm−2 (BLB Philips TL-D 15W, 300–400 nm, broad maximum at 365 nm). At regular time intervals, 0.1 cm3 of the irradiated cell suspension were taken, diluted and analysed. To see the effect of UV light itself, the clear glass substrate (blank) was also tested in another Petri dish. 2.6. Glass adhesion test This method is based on the ISO 27447:2009(E) standard, which works with the bacteria (E. coli) spread on the test surface (25 × 30 mm2 ) and covered by adhesive glass (24 × 24 mm2 ). This so called “sandwich” was then put in the Petri dish (diameter 45 mm) with wet paper ﬁlter and the dish was covered with the cap (Fig. 1C). The volume of the cell suspension was 0.05 cm3 and the concentration of bacteria was within the interval 2.0 × 106 to 8.0 × 106 CFU/ml. In this experimental set-up, the irradiation conditions were the same as in the 3 cm3 test (0.6 mW cm−2 ). After a given interval of time, the cap was removed and the cover glass together with the TiO2 /glass sample were shaken out in 10 cm3 of saline solution, diluted and analysed. The effect of light intensity on E. coli and E. faecalis degradation was studied for light intensities in the range 0.2 to 0.6 mW cm−2 . Different intensities of the incident light were achieved by changing the distance of the sample from the light source and also by placing the stainless steel grid in front of the light source. For the

102

J. Kr´ ysa et al. / Journal of Hazardous Materials 195 (2011) 100–106

1 2 4 5 8 B

A 25 ml method

3 ml method

1 2 3 4 5 6 7

C

Fig. 1. Schematic diagram of the antibacterial tests arrangement: (A) 50 cm3 method, (B) 3 cm3 method, (C) thin ﬁlm method – according to the ISO standard [33]; 1-light source, 2-moisture preservation glass, 3-cover slide glass, 4-bacterial suspension, 5-TiO2 sample under test, 6-glass rod, 7-wet ﬁlter paper, 8-magnetic stirrer.

distances 24 and 36 cm the light intensities were 0.6 and 0.4 mW cm−2 , respectively. When the grid was used the light intensity for the distance of 36 cm was 0.2 mW cm−2 . 3. Results and discussion 3.1. Test method For the determination of the photocatalytic activity of titania layers, photochemical reactors where the TiO2 layer is placed in the solution of model dye or organic compound are often used. Such types of reactor have been commonly employed in our laboratory for the determination of photoactivity using dye Acid Orange 7 [38]. Our ﬁrst approach to the antibacterial test was simply to replace the dye solution with bacterial suspension. In Fig. 2 the results of antibacterial test on sol–gel and particulate TiO2 layers of different titania loading in 50 cm3 of bacterial suspension can be seen. It is clear that the inhibition effect of the TiO2 layer on E. coli increased as the amount of titania in the layer increased. The advantages of this method and experimental setup are: (i) the effect of UV light irradiation on the inhibition of bacteria in the suspension without TiO2 layer is almost negligible, (ii) the bacterial suspension is stable in dark even when the TiO2 layer is present, (iii) samples of bacteria suspension can be taken during the experiment, (iv) only one sample of TiO2 layer is necessary for the whole experiment (in the glass adhesion test – the ISO standard – one TiO2 sample is necessary for each point of the CFU dependence on the time of irradiation). It seems, that the 50 cm3 method is an ideal test for particulate TiO2 layers with high activity prepared from powder suspensions.

−1

Survival E. coli / log (CFU ml )

3.5 3.0 2.5 2.0

−2

P25 1.0 mg cm −2 P25 0.5 mg cm −2 P25 0.2 mg cm −2 P25 0.1 mg cm just UV Light dark −2 Sol-gel 0.043 mg cm

1.5 1.0 0.5 0.0 0

25

50

75

100

125

150

175

200

Time / min Fig. 2. 50 cm3 method for E. coli antibacterial test. Log scale of surviving bacteria under UV irradiation (1 mW cm−2 ) for particulate titania layers with different amount of P25 as a function of illumination time. The sol–gel layer is also included (250 nm, 0.043 mg cm−2 ).

But use of particulate layers in the practical application (antibacterial glasses and tiles) is not favoured due to the low mechanical stability. TiO2 layers prepared by the sol–gel method have much higher application potential in this ﬁeld. Sol–gel layers can be applied on various surfaces, such as tiles, glass and metal surfaces and their mechanical stability, compared to powder layers, is much better. On the other hand due to the non-porous structure [39] and much smaller layer thickness (146 nm) than 0.1 mg cm−2 particulate ﬁlm (thickness 800 nm) the resulting photoactivity of sol gel ﬁlm measured using Acid Orange 7 as model compound is about 8 times smaller than that for particulate ﬁlm [38]. Thus we can expect the similar behaviour when comparing the antibacterial properties of particulate and sol gel ﬁlms. In fact, from Fig. 2 it is clear that the antibacterial activity of the sol–gel layer in the 50 cm3 test is comparable with the antibacterial effect of UV light itself. As a consequence, we had to ﬁnd and verify different methods for the antibacterial testing of sol–gel layers. An experimental setup with a drop of bacterial suspension (0.2 cm3 ) pipetted onto the coated substrates has been described by Kühn et al. [14]. The volume of the drop in the “drop test” can be smaller, e.g. 0.1 cm3 [16,17], 0.07 cm3 [30] or even 0.01 cm3 [29]. This approach is simple but it has two serious drawbacks. At ﬁrst, samples with the drop of bacterial suspension were not covered and drying of the drop during irradiation may take place. Secondly, the surface area of TiO2 ﬁlm in contact with the bacteria drop is not properly deﬁned. Kikuchi et al. [19] solved the drying problem by placing the TiO2 sample with the bacterial drop into a Petri dish with a small amount of water and covering it with a glass lid. However the problem with the deﬁnition of surface area remained. The above mentioned drawbacks result in a number of discrepancies as is visible, for example, from comparison of the results on TiO2 layers prepared from Degussa P25. Kühn et al. [14] observed 4 log decrease of E. coli CFU after 1 h of irradiation using a 0.2 cm3 drop (UV light had no effect on the drop of bacterial suspension). Hajková et al. [29] also described a 4 log decrease of E. coli CFU after 1 h of irradiation using a 0.01 cm3 drop, but a 2 log decrease using UV light itself was observed. As a next step, we eliminated the problem of the ill-deﬁned area and drop drying by creating a new antibacterial test method. Our method deﬁnes the size of the tested TiO2 sample (25 × 30 mm2 ) which ﬁts well into a small Petri dish (diameter 45 mm). Then we put 3 cm3 of bacterial suspension into the Petri dish to create a thin liquid ﬁlm above the TiO2 layer and cover the whole system with a glass lid to eliminate evaporation (Fig. 1B). Using this method we decreased the volume of bacterial suspension from 50 to 3 cm3 and also the ratio of the irradiated area to the volume was changed from 1:5 (50 cm3 method) to 1:0.4 (3 cm3 method). Initially, we tried similar initial bacteria concentration as in the 50 cm3 test (2.5 × 104 CFU/ml). After irradiation of the system, we expected a faster decrease of bacteria concentration. However the results showed almost no killing of bacteria (Fig. 3). Secondly, we increased the initial concentration to the range recommended in the ISO

J. Kr´ ysa et al. / Journal of Hazardous Materials 195 (2011) 100–106

3.2. Critical assessment of the ISO standard – adhesion glass method

7 6

−1

Survival E. coli / log (CFU ml )

103

5 4 3 Survival bacteria / %

100

2 1

6

90

3.3×10 CFU ml TiO2 + UV UV

80 70

4

2.5×10 CFU ml TiO2 + UV UV

−1

60 50

0

0 0

−1

20 40 60 80 100 120 140 160 180 Time / min

15

30

45

60

75

90

105 120 135 150 165 180

Time / min Fig. 3. 3 cm3 method for the antibacterial test of E. coli (two initial concentrations). Log scale of surviving bacteria under UV irradiation (0.6 mW cm−2 ) for sol–gel titania layer and pure glass substrate as a function of illumination time. Insert diagram shows the percentage of surviving bacteria.

standard [33] (3.3 × 106 CFU/ml). Again the differences between the effects of UV light itself and irradiated TiO2 layer was not signiﬁcant. This is possibly due the insufﬁcient contact of bacteria with the TiO2 layer and the existence of “dead volumes of bacteria suspension” with small or no exchange with the volume of bacteria suspension in contact with TiO2 layer. The problem of the dead volume was solved by Sunada et al. [24] who placed a cylindrical frame directly on the TiO2 sol–gel layer and then 1 cm3 of the E. coli suspension was pipetted into it. After 1 h around 50% of the bacteria were killed (only 5% due to UV itself) [24]. If we compare this test (1 cm3 ) with our 3 cm3 test, where we have 35% of killing after 1 h and the same effect of UV (5%), it is clear that, if all the volume of bacterial suspension is in direct contact with the TiO2 layer, the photocatalytic de-activation of microorganisms is faster. Similarly to Sunada et al. [24], Dunlop et al. [40] used a silicone cylinder placed on a TiO2 layer and ﬁlled this with a 1 cm3 bacterial suspension of lower concentration (1 × 103 CFU/ml). Even though the experimental setups [24,40] were similar the difference between inhibition efﬁciency of the TiO2 + UV light and UV light itself is much smaller in the work of Dunlop et al. [40]. In addition to this observed discrepancy the scale of bacteria concentration used may make comparison difﬁcult. Fig. 3 shows that a percentage scale shows a decrease of viable bacteria, but a log scale suggests negligible antibacterial effect. It seems that a log scale is more suitable for conﬁrmation of photocatalytic inhibition effect of TiO2 layers, but for evaluation of the effect of UV itself the percentage scale is more useful.

In the next step, we adapted our experimental setup according to the ISO [33]. However in our laboratory we are not able to fulﬁl all the recommendations and requirements of the ISO. In Table 1 we show the differences between the ISO and our own glass adhesion test. The differences are in detail discussed in the following three paragraphs. At ﬁrst, according to the ISO standard, E. coli (G−) is the species of bacterium recommended for the tests (glass adhesion method), but other types of bacteria can be tested, if necessary. In our work we used E. faecalis (G+) as the second test microorganism. The preparation of microorganism suspension according to the ISO is complicated and time consuming (repeated subcultures with one month expiration, many cultivations and dilutions before each experiment). Using the gelatinous pills (CCM) has many advantages: after 24 h the bacterial suspension is ready for the experiment, the concentration of bacteria in the pill is guaranteed, the pill can be stored for 3 years, the purity of bacterium strain is also guaranteed and, ﬁnally, it is easy to use. Secondly, according to the ISO standard, nutrient broth must be used for the preparation of the bacterial suspension. However we think that saline solution is better than nutrient broth because it does not contain organic compounds (meat extract and peptone in nutrient broth) which could also be photocatalytically degraded by the TiO2 layers during the test and thus slow down the rate of bacteria inactivation. According to the ISO standard, nutrient agar must be used for bacteria cultivation. From a microbiological point of view, this is not the best choice because of possible contamination from the surrounding environment. For this reason we are using selective media in our laboratory. Finally, according to the ISO standard, the specimen size should be 50 × 50 ± 2 mm2 and the size of adhesive glass should be 40 × 40 ± 2 mm2 . It is also possible to use a different specimen size but the specimen surface must be covered by adhesive glass of dimension in the range from 400 mm2 to 1600 mm2 . Our specimen size was 24 × 30 mm2 and the area covered by adhesive glass was 576 mm2 . The volume of the cell suspension spread on the specimen was different from that mentioned in ISO standard and was adjusted according to the size of the adhesive glass. The amount of bacteria was the same as the concentration recommended in the ISO standard (∼2 × 106 CFU/ml). Fig. 4 shows the results of the adhesive glass test. It is apparent that the difference between the antibacterial effect of the TiO2 layer and UV light itself is much higher than in the case of the 3 cm3 test. The explanation is that in the adhesive glass test, the irradiated surface of TiO2 is in direct contact with bacteria. Comparing the drop test and glass adhesion test the later seems to be more appropriate for TiO2 thin ﬁlms. In the present adhesive glass test

Table 1 List of signiﬁcant parameters of ISO 27447:2009(E) method and their comparison with the glass adhesion method used in our laboratory.

Bacterium Bacteria suspension preparation Bacteria cultivation Specimen size Sample size covered by adhesive glass Volume of test bacterial suspension Initial bacteria concentration Exposure time Light source UV light intensity

ISO 27447:2009 (E)

Our laboratory adhesion glass test

Staphylococcus aureus (G+) Escherichia coli (G−) Cultivation in nutrient broth Nutrient agar

Enterococcus faecalis (G+) Escherichia coli (G−) Gelatinous pill m-FC agar (E. coli) Medium Slanetz-Bartley (E. faecalis) 24 × 24 mm2 576 mm2 0.05 cm3 2 × 106 CFU/ml 3h Fluorescent BLB lamp 300–400 nm 0.2–0.6 mW cm−2

50 × 50 ± 2 mm2 400–1600 mm2 0.15 cm3 6.7 × 105 –2.6 × 106 CFU/ml 4–8 h Fluorescent BLB lamp 300–400 nm 0.001–0.25 mW cm−2

104

J. Kr´ ysa et al. / Journal of Hazardous Materials 195 (2011) 100–106

7.0

5.5

−1

6.0 5.5

Survival bacteria / %

5.0 4.5 4.0 3.5

100 90 80 70 60 50 40 30 20 10 0

6

−1

8×10 CFU ml TiO2 + UV UV

5.0 4.5 4.0 100

Survival bacteria / %

−1

Survival E. coli / log (CFU ml )

6.5

Survival E. coli / log (CFU ml )

a

3.5 1

60 40

20 40 60 80 100 120 140 160 180

3.0 30

45

60

75

0

90

30

45

60

105

120

135

150

165

180

5.5 −1

5.0

4.5 100

Survival bacteria / %

Light intensity is an important parameter in the antibacterial tests. Firstly, light intensity is one of the rate determining steps in semiconductor photocatalysis. Secondly, UV light itself (especially of low wavelengths) may inactivate bacteria. In the ISO standard ﬂuorescent black light blue (BLB) lamps are recommended (wavelength 300–400 nm, light intensity 0.001–0.25 mW cm−2 ). However the light sources and their intensities and wavelengths employed in reported antibacterial tests [6,8,10,11,15,21] often vary and, as a consequence, comparison of results is very difﬁcult. For example, Soken et al. [6] describe E. coli disinfection using a AgTiO2 suspension and a UV light intensity of 5.8 mW cm−2 (300 W, 294 nm). Sunada et al. [15] studied the photocatalytic inhibition of E. coli on TiO2 thin ﬁlms by BLB lamp (15 W, 365 nm, 1.0 mW cm−2 ). Wu et al. [21] used a metal halogen desk lamp to investigate disinfection induced by visible light. The light intensity below 400 nm was less than 0.01 mW cm2 and the visible light intensity was in the range 1.6 mW cm−2 to 0.4 mW cm−2 [21]. Ibanez et al. [10] used an UV-A lamp (maximum at 365 nm) for studying the antibacterial effect of TiO2 (P25) suspension on different Gram(−) microorganisms. Because of the high sensitivity of P. aeruginosa to UV-A, suspensions of these bacteria were exposed to a lower UV-A intensity, i.e. 1.4 mW cm2 . For other microorganisms (E. coli, Salmonella typhimurium, Enterobacter cloacae) a light intensity of 5.5 mW cm−2 was chosen. Benabbou et al. [11] used HPK 125 W light to investigate the disinfection of E. coli in TiO2 suspension. Appling an optical ﬁlter they were able to work in UVC, UVB and UVA wavelength regions. In the case of UVA light, the intensity varied from 0.48 mW cm−2 to 3.85 mW cm−2 by virtue of the distance from the light source and the presence of the appropriate grid [11]. It must be emphasized that in all the above mentioned cases, the intensity of the light sources did not ﬁt the interval set by the ISO standard [33]. Fig. 5 shows the results of adhesion glass tests (A – E. coli, B – E. faecalis) using three different light intensities (0.6, 0.4 and 0.2 mW cm−2 ). The lowest value, (0.2 mW cm−2 ) fulﬁls the ISO standard. It can be seen that the effect of light intensity on E. coli and

90

b Survival E. faecalis / log (CFU ml )

3.3. Inﬂuence of light intensity

75

Time / min

Time / min

60% of the bacteria were killed after 20 min and after 2 h almost 99% of the surface was disinfected. On the other hand the drop test (100 ␮l, 106 CFU/ml) on sol–gel layers shows inactivation of only 30% of viable bacteria after 3 h [16]

0.2 mW cm TiO + UV UV

Time / min

15

105 120 135 150 165 180

Fig. 4. Adhesion glass method for the antibacterial test of E. coli. Log scale of surviving bacteria under UV irradiation (0.6 mW cm−2 ) for titania sol–gel layer and pure glass substrate as a function of illumination time. Insert diagram shows the percentage of surviving bacteria.

0.4 mW cm TiO + UV UV

20 40 60 80 100 120 140 160 180

0 0

15

0.6 mW cm TiO + UV UV

20 0

0

Time / min

0

80

4.0

3.5 1

80 60 40

0.6 mW cm TiO + UV UV

20 0

0

0.4 mW cm TiO + UV UV

0.2 mW cm TiO + UV UV

20 40 60 80 100 120 140 160 180

Time / min

0 0

15

30

45

60

75

90

105 120 135 150 165 180

Time / min Fig. 5. Log scale of surviving bacteria under UV irradiation for titania sol–gel layer and pure glass substrate as a function of illumination time – adhesive glass method. (a) E. coli (initial bacteria concentration – 3.8 × 105 CFU/50 ␮l). (b) E. faecalis (initial bacteria concentration – 3.2 × 105 CFU/50 ␮l). Open symbols – UV light itself, full symbols – UV light + TiO2 , light intensity 0.6 mW cm−2 (), 0.4 mW cm−2 (♦) and 0.2 mW cm−2 ().

E. faecalis inactivation is different. In the case of Gram(−) bacterium E. coli, the effect of UV light itself on the bacteria inhibition decreased as the light intensity decreased (Fig. 5A). The percentage of surviving bacteria after 180 min irradiation increased from 38% for the highest light intensity (0.6 mW cm−2 ) to 77% for the lowest light intensity (0.2 mW cm−2 ). After 60 min irradiation the difference was even higher: 40% for 0.6 mW cm−2 and almost 90% for 0.2 mW cm−2 . In the case of Gram(+) bacterium E. faecalis, a decrease in UV light intensity did not have such a deﬁnite effect on the bacteria inhibition (Fig. 5B). The percentage of surviving bacteria after 60 min irradiation was around 80% for all UV light intensities. Even after 180 min irradiation the effect of UV light was not as strong as observed in the case of E. coli. The effect of UV light on the amount of surviving E. faecalis has moved from 35% (highest intensity – 0.6 mW cm−2 ) to 46% (lowest intensity – 0.2 mW cm−2 ). According to our experiments 60 min is the minimum irradiation time necessary to distinguish the antibacterial effect of TiO2 from the effect of UV light itself. In the case of E. coli (light intensity 0.2 mW cm−2 ), we observed 10% inhibition by UV light and 60% by TiO2 layer (after 60 min irradiation). When a higher intensity was used (0.6 mW cm−2 ), 60% of the bacteria were killed only by UV, but with a TiO2 layer more than 95% bacteria were inactivated (Fig. 5A). This trend (increasing light intensity) is consistent

J. Kr´ ysa et al. / Journal of Hazardous Materials 195 (2011) 100–106

with recent results of Dunlop et al. [40] who observed, after 40 min irradiation (UVA, 3 mW cm−2 ), 70% inhibition by UV light and 90% inhibition when TiO2 layer was applied. A strong effect of UV light itself was also observed by Foster et al. [41] but in the log scale (2 log decrease after 6 h for 2 mW cm−2 ) while for an order lower light intensity (the ISO) the inhibition due to UV light itself was only 60%. In the case of E. faecalis, for all studied light intensities after 60 min of irradiation 20% of bacteria was inactivated by UV light itself and around 55% was inactivated by using an irradiated TiO2 layer (Fig. 5B). Thus E. faecalis are not as sensitive to UVA light as E. coli. On the other hand their inactivation proceeds with similar rate as that of E. coli. The results are important because there are few other few other data concerning the photocatalytic degradation of E. faecalis on TiO2 . Only Malato et al. [42], in his review, reports that bacterium E. faecalis is generally more difﬁcult to disinfect than E. coli and Mitoraj et al. [43] conﬁrm this experimentally, but for the case of VIS light irradiation. 4. Conclusions Gramnegative E. coli and gram positive E. faecalis were found to be suitable for antibacterial effect evaluation on irradiated TiO2 layers. It was found that the method using 50 cm3 of bacteria suspension is convenient for testing layers with strong antibacterial effect (prepared from powder photocatalysts). A decrease in the bacteria suspension volume to 3 cm3 did not bring the expected result (improvement of the difference between antibacterial effect of irradiated TiO2 and UV light itself). The possible reason is insufﬁcient contact of bacteria with the TiO2 layer and the existence of “dead volumes of bacteria suspension” with small or no exchange with the suspension adjacent to the TiO2 layer. Thus for evaluation of the antibacterial effect of transparent sol gel layers the adhesion glass method based on the ISO standard is the most appropriate. Some parameters stated in ISO 27447:2009(E) can be adapted according to the working conditions used in particular laboratories (sample size, type of microorganism, irradiation time). Furthermore we suggest some improvements: (i) the use of gelatinous pills (CCM) of bacteria leading to simplicity and reproducibility, (ii) the use of saline solution instead of nutrient broth for bacteria suspension preparation, (iii) the application of selective media instead of nutrient agar for bacteria cultivation. Experiments at three UV light intensities (0.2–0.6 mW cm−2 ) conﬁrm the inhibition effect of UV light (even at 365 nm) itself. The lowest value of 0.2 mW cm−2 , fulﬁlling the requirements of the ISO standard, and irradiation time 60 min was found to be optimal for testing. Acknowledgements The authors acknowledge ﬁnancial support (project 1M0577) of the Ministry of Education, Youth and Sport of the Czech Republic, the Grant Agency of the Czech Republic (project number 104/08/0435) and the FP7 EU project PILGRIM (No.: 223050). The authors gratefully acknowledge the English correction done by Prof. A.A. Wragg from Exeter University, UK. References [1] Q. Li, S. Mahendra, D.Y. Lyon, L. Brunet, M.V. Liga, D. Li, P.J.J. Alvarez, Antimicrobial nanomaterials for water disinfection and microbial control: potential applications and implications, Water Res. 42 (2008) 4591–4602. [2] M. Yoshinari, Y. Oda, T. Kato, K. Okuda, Inﬂuence of surface modiﬁcations to titanium on antibacterial activity in vitro, Biomaterials 22 (2001) 2043–2048. [3] A. Paleologou, H. Marakas, Ni.P. Xekoukoulotakis, A. Moya, Y. Vergara, N. Kalogerakis, P. Gikas, D. Mantzavinos, Disinfection of water and waste water by TiO2 photocatalysis, sonolysis and UV-C irradiation, Catal. Today 129 (2007) 136–142.

105

[4] A. Vohra, D.Y. Goswami, D.A. Deshpande, S.S. Block, Enhanced photocatalytic inactivation of bacterial spores on surfaces in air, J. Ind. Microbiol. Biotechnol. 32 (2005) 364–370. [5] J.A.H. Melián, J.M.D. Rodríguez, A.V. Suárez, E.T. Rendón, C.V.d. Campo, J. Arana, ˇ The photocatalytic disinfection of urban waste waters, Chemosphere J.P. Pena, 41 (3) (2000) 323–327. [6] M. Sökmen, F. Candan, Z. Sümerb, Disinfection of E. coli by the Ag-TiO2 /UV system: lipid peroxidation, J. Photochem. Photobiol. A: Chem. 143 (2–3) (2001) 241–244. [7] R. van Grieken, J. Marugán, C. Sordo, P. Martínez, C. Pablos, Photocatalytic inactivation of bacteria in water using suspended and immobilized silver-TiO2 , Appl. Catal. B: Environ. 93 (2009) 112–118. [8] N. Lydakis-Simantiris, D. Riga, E. Katsivela, D. Mantzavinos, N.P. Xekoukoulotakis, Disinfection of spring water and secondary treated municipal waste water by TiO2 photocatalysis, Desalination 250 (2010) 351–355. [9] J.M.C. Robertson, P.K.J. Robertson, L.A. Lawton, A comparison of the effectiveness of TiO2 photocatalysis and UVA photolysis for the destruction of three pathogenic micro-organisms, J. Photochem. Photobiol. A: Chem. 175 (2005) 51–56. ˜ [10] J.A. Ibánez, M.I. Litter, R.A. Pizarro, Photocatalytic bactericidal effect of TiO2 on Enterobacter cloacae: comparative study with other Gram (−) bacteria, J. Photochem. Photobiol. A: Chem. 157 (2003) 81–85. [11] A.K. Benabbou, Z. Derriche, C. Felix, P. Lejeune, C. Guillard, Photocatalytic inactivation of Escherischia coli: effect of concentration of TiO2 and microorganism, nature, and intensity of UV irradiation, Appl. Catal. B: Environ. 76 (2007) 257–263. [12] H.-L. Liu, T.C.-K. Yang, Photocatalytic inactivation of Escherichia coli and Lactobacillus helveticus by ZnO and TiO2 activated with ultraviolet light, Process Biochem. 39 (2003) 475–481. [13] D.M.A. Alrousan, P.S.M. Dunlop, T.A. McMurray, J.A. Byrne, Photocatalytic inactivation of E. coli in surface water using immobilised nanoparticle TiO2 ﬁlms, Water Res. 43 (2009) 47–54. [14] K.P Kühn, Ir.F. Chaberny, K. Massholder, M. Stickler, V.W. Benz, H.-G. Sonntag, L. Erdinger, Disinfection of surfaces by photocatalytic oxidation with titanium dioxide and UVA light, Chemosphere 53 (2003) 71–77. [15] K. Sunada, T. Watanabe, K. Hashimoto, Studies on photokilling of bacteria on TiO2 thin ﬁlm, J. Photochem. Photobiol. A: Chem. 156 (2003) 227–233. [16] C.C. Trapalis, P. Keivanidis, G. Kordas, M. Zaharescu, M. Crisan, A. Szatvanyi, M. Gartner, TiO2 (Fe3+ ) nanostructured thin ﬁlms with antibacterial properties, Thin Solid Films 433 (2003) 186–190. [17] O. Akhavan, R. Azimirad, Photocatalytic property of Fe2 O3 nanograin chains coated by TiO2 nanolayer in visible light irradiation, Appl. Catal. A: Gen. 369 (2009) 77–82. [18] P. Amézaga-Madrid, G.V. Nevárez-Moorillón, E. Orrantia-Borunda, M. Miki-Yoshida, Photoinduced bactericidal activity against Pseudomonas aeruginosa by TiO2 based thin ﬁlms, FEMS Microbiol. Lett. 211 (2002) 183–188. [19] Y. Kikuchi, K. Sunada, T. Iyoda, K. Hashimoto, A. Fujishima, Photocatalytic bactericidal effect of TiO2 thin ﬁlms: dynamic view of the active oxygen species responsible for the effect, J. Photochem. Photobiol. A: Chem. 106 (1997) 51–56. [20] X. Zhao, Q. Zhao, J. Yu, B. Liu, Development of multifunctional photoactive selfcleaning glasses, J. Non-Cryst. Solids 354 (2008) 1424–1430. [21] P. Wu, R. Xie, J.A. Imlay, J.K. Shang, Visible-light-induced photocatalytic inactivation of bacteria by composite photocatalysts of palladium oxide and nitrogen-doped titanium oxide, Appl. Catal. B: Environ. 88 (2009) 576–581. [22] J.C. Yu, W. Ho, J. Lin, H. Yip, P.K. Wong, Photocatalytic activity, antibacterial effect and photoinduced hydrophilicity of TiO2 ﬁlms coated on a stainless steel substrate, Environ. Sci. Technol. 37 (2003) 2296–2301. [23] G. Fu, P.S. Vary, C.-T. Lin, Anatase TiO2 nanocomposites for antimicrobial coatings, J. Phys. Chem. B 109 (2005) 8889–8898. [24] K. Sunada, Y. Kikuchi, K. Hashimoto, A. Fujishima, Bactericidal and detoxiﬁcation effects of TiO2 thin ﬁlm photocatalysts, Environ. Sci. Technol. 32 (1998) 726–728. [25] W. Zhang, Y. Chen, S. Yu, S. Chen, Y. Yin, Preparation and antibacterial behavior of Fe3+ -doped nanostructured TiO2 thin ﬁlms, Thin Solid Films 516 (2008) 4690–4694. [26] L. Caballero, K.A. Whitehead, N.S. Allen, J. Verran, Inactivation of Escherichia coli on immobilized TiO2 using ﬂuorescent light, J. Photochem. Photobiol. A: Chem. 202 (2009) 92–98. [27] N. Baram, D. Starosvetsky, J. Starosvetsky, M. Epshtein, R. Armon, Y. Ein-Eli, Enhanced inactivation of E. coli bacteria using immobilized porous TiO2 photoelectrocatalysis, Electrochim. Acta 54 (2009) 3381–3386. [28] R.v. Grieken, J. Marugán, C. Sordo, C. Pablos, Comparison of the photocatalytic disinfection of E. coli suspensions in slurry, wall and ﬁxed-bed reactors, Catal. Today 144 (2009) 48–54. ˇ J. Krumeich, P. Exnar, A. Kolouch, J. Matouˇsek, P. Koˇcí, [29] P. Hájková, P. Spatenka, Antibacterial effect of metal modiﬁed TiO2/PECVD ﬁlms, Eur. Phys. J. D 54 (2009) 189–193. [30] F.R. Marciano, D.A. Lima-Oliveira, N.S. Da-Silva, A.V. Diniz, E.J. Corat, V.J. TravaAiroldi, Antibacterial activity of DLC ﬁlms containing TiO2 nanoparticles, J. Colloid Interface Sci. 340 (2009) 87–92. [31] M.T. Madigan, J.M. Martinko, P.V. Dunlap, D.P. Clark, Brock Biology of Microorganisms, 12th edn., Pearson Education Inc., Glenview, IL, USA, 2009, ISBN 978-0321-53615-0. [32] J.P. Nataro, J.B. Kaper, Diarrheagenic Escherichia coli, Clin. Microbiol. Rev. (1998) 142–201.

106

J. Kr´ ysa et al. / Journal of Hazardous Materials 195 (2011) 100–106

[33] ISO 27447:2009(E), Fine ceramics (advanced ceramics, advanced technical ceramics) — test method for antibacterial activity of semiconducting photocatalytic materials, ﬁrst edition 2009-06-01. ´ A. Mills, Correlation of oxidative and reductive dye bleach[34] J. Zita, J. Krysa, ing on TiO2 photocatalyst ﬁlms, J. Photochem. Photobiol. A: Chem. 203 (2009) 119–124. ´ Multilayer TiO2 /SiO2 thin sol–gel ﬁlms: effect of [35] J. Zita, J. Maixner, J. Krysa, calcinations temperature and Na+ diffusion, J. Photochem. Photobiol. A: Chem. 216 (2010) 194–200. [36] ISO 9308-1:2000, Water quality – enumeration of Escherichia coli and coliform bacteria. [37] ISO 7899-2:2000, Water quality – detection and enumeration of intestinal enterococci. ´ ´ Photocatalytic degradation of Acid Orange 7 [38] M. Zlámal, J. Krysa, J. Jirkovsky, on TiO2 ﬁlms prepared from various powder catalysts, Catal. Lett. 133 (2009) 160–166.

[39] J. Zita, J. Krysa, U. Cernigoj, U. Lavrencic Stangar, J. Jirkovsky, J. Rathousky, Photocatalytic properties of different TiO2 thin ﬁlms of various porosity and titania loading, Catal. Today 161 (2011) 29–34. [40] P.S.M. Dunlop, C.P. Sheeran, J.A. Byrne, M.A.S. McMahon, M.A. Boyle, K.G. McGuigan, Inactivation of clinically relevant pathogens by photocatalytic coatings, J. Photochem. Photobiol. A: Chem. 216 (2010) 303– 310. [41] H.A. Foster, D.W. Sheel, P. Sheel, P. Evans, S. Varghese, N. Rutschke, H.M. Yates, Antimicrobial activity of titania/silver and titanoa/copper ﬁlms prepared by CVD, J. Photochem. Photobiol. A: Chem. 216 (2010) 283–289. ˜ M.I. Maldonado, J. Blanco, W. Gernjak, Decon[42] S. Malato, P. Fernández-Ibánnez, tamination and disinfection of water by solar photocatalysis: recent overview and trends, Catal. Today 147 (2009) 1–59. [43] D. Mitoraj, A. Jánczyk, M. Strus, H. Kisch, G. Stochel, P.B. Heczko, W. Macyk, Visible light inactivation of bacteria and fungi by modiﬁed titanium dioxide, Photochem. Photobiol. Sci. 6 (2007) 642–648.

Journal of Hazardous Materials 195 (2011) 107–114

Contents lists available at ScienceDirect

Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat

CO2 sequestration by carbonation of steelmaking slags in an autoclave reactor E.-E. Chang a , Shu-Yuan Pan b , Yi-Hung Chen c , Hsiao-Wen Chu b , Chu-Fang Wang d , Pen-Chi Chiang b,∗ a

Department of Biochemistry, Taipei Medical University, Taipei, Taiwan Graduate Institute of Environmental Engineering, National Taiwan University, No. 71 Chou-shan Rd., Taiwan c Department of Chemical Engineering and Biotechnology, National Taipei University of Technology, Taiwan d Biomedical Engineering and Environmental Sciences, National Tsing Hua University, Taiwan b

a r t i c l e

i n f o

Article history: Received 13 May 2011 Received in revised form 3 August 2011 Accepted 4 August 2011 Available online 10 August 2011 Keywords: Accelerated carbonation Alkaline solid waste Calcite Surface coverage model Life cycle assessment

a b s t r a c t Carbon dioxide (CO2 ) sequestration experiments using the accelerated carbonation of three types of steelmaking slags, i.e., ultra-ﬁne (UF) slag, ﬂy-ash (FA) slag, and blended hydraulic slag cement (BHC), were performed in an autoclave reactor. The effects of reaction time, liquid-to-solid ratio (L/S), temperature, CO2 pressure, and initial pH on CO2 sequestration were evaluated. Two different CO2 pressures were chosen: the normal condition (700 psig) and the supercritical condition (1300 psig). The carbonation conversion was determined quantitatively by using thermo-gravimetric analysis (TGA). The major factors that affected the conversion were reaction time (5 min to 12 h) and temperature (40–160 ◦ C). The BHC was found to have the highest carbonation conversion of approximately 68%, corresponding to a capacity of 0.283 kg CO2 /kg BHC, in 12 h at 700 psig and 160 ◦ C. In addition, the carbonation products were conﬁrmed to be mainly in CaCO3, which was determined by using scanning electron microscopy (SEM) and X-ray powder diffraction (XRD) to analyze samples before and after carbonation. Furthermore, reaction kinetics were expressed with a surface coverage model, and the carbon footprint of the developed technology in this investigation was calculated by a life cycle assessment (LCA). © 2011 Elsevier B.V. All rights reserved.

1. Introduction Carbon sequestration is a promising option for reducing carbon dioxide (CO2 ) emissions and alleviating global warming. Both CO2 captured from emission sources and subsequent transport of the captured CO2 to isolated reservoirs are essential for carbon sequestration. Carbon capture is affected by environmental factors, capacity, and cost. Mineral sequestration is a method of carbon capture that accelerates the natural weathering of silicate minerals, allowing them to react with CO2 to form stable products, carbonate minerals, and silica for further usage or disposal [1]. In addition, carbonation is an exothermal reaction; thus, energy consumption and costs may be limited by its inherent properties [1,2]. In all cases, the sequestration chemicals must provide base ions such as monovalent sodium and potassium, or divalent calcium and magnesium ions to neutralize the carbonic acid. Other carbonate-forming elements such as iron carbonates are not practical due to their unique and precious features [3]. In addition to controlling the reaction conditions, choosing suitable mineral feedstocks and properly designing the reactor are crucial to achieving high CO2 sequestration efﬁciencies.

∗ Corresponding author. Tel.: +886 2 23622510, fax: +886 2 23661642. E-mail address: [email protected] (P.-C. Chiang). 0304-3894/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.jhazmat.2011.08.006

One possible feedstock for CO2 sequestration by accelerated carbonation is industrial solid waste, including steelmaking slags, combustion residues, and ﬂy ash, which generally are alkaline and rich in calcium. The use of industrial waste is advantageous because of its low cost and widespread availability in industrial areas [4]. Interest in using industrial alkaline solid wastes as sources of calcium or magnesium oxide for CO2 sequestration has arisen because these materials are readily available, cheap, and usually produced near large-emission sources of CO2 [5]. In this study, carbonation reactions were performed primarily via the reaction of CO2 with raw CaO-based materials, and calcium carbonate (CaCO3 ) was observed to be the predominant carbonation product [6]. The use of this material simultaneously can reduce the amount of waste and neutralize a hazardous material. The objectives of this study were to investigate the carbonation of several steelmaking slags, including ultra-ﬁne (UF) slag, ﬂy-ash (FA) slag, and blended hydraulic slag cement (BHC), in an autoclave reactor. The effects of the operational conditions, including the type of steelmaking slag, reaction time, liquid-to-solid ratio (L/S), temperature, CO2 pressure, and initial pH, on the performance of the carbonation process were evaluated. In addition, reaction kinetics of the carbonation process were tested using a surface coverage model.

108

E.-E. Chang et al. / Journal of Hazardous Materials 195 (2011) 107–114

Fig. 1. Schematic diagram of the experimental set-up for the carbonation of steelmaking slag in an autoclave reactor. 1. CO2 gas cylinder; 2. Circulating bath; 3. Syringe pump; 4. Magnetic stirrer and heater; 5. Reactor (autoclave); 6. Thermo couple; 7. Needle valve; 8. Vent to hood.

2. Materials and methods 2.1. Experiments The aqueous carbonation of UF slag, FA slag, and BHC were conducted in an autoclave reactor that contained distilled water at a designated temperature of 40–160 ◦ C. The UF slag, FA slag, and BHC with a diameter of approximately 1 cm were provided by the CHC

100

Fresh-FA Fresh-UF Fresh-BHC

Weight Percent (%)

95

FA UF

90

Weight loss between 500 - 850 ºC

85

Weight loss due to decomposition of CaCO3

Fresh-UF Fresh-FA Fresh-BHC UF Slag FA Slag BHC

80

2.2. Composition analysis BHC

75 0

200

Resources Corporation (Kaohsiung, Taiwan). All slags were ground and sieved to less than 44 ␮m for all experiments. The BHC contains an intimate and uniform blend of Portland cement and ﬁne granulated blast furnace (BF) slag. The BHC used in this investigation is classiﬁed as CEM III/C (∼90% BF slag content) according to EN standards [7]. A schematic diagram demonstrating the carbonation of the steelmaking slag in an autoclave reactor is shown in Fig. 1. CO2 was injected continuously into the reactor at a designated pressure and a constant ﬂow rate. The operational factors, including the reaction time (t), liquidto-solid ratio (L/S), reaction temperature (T), CO2 pressure (P), and initial pH, systematically were varied with the various feedstocks to minimize energy and chemical consumption. After the reaction, the samples of reacted slurry immediately were ﬁltered through a PTFE membrane ﬁlter (Millipore, 45-␮m pore size and 47 mm diameter), and then heated in an oven (105 ◦ C) for use. The conversion of the carbonation products was determined quantitatively by thermogravimetric analysis (TGA) and qualitatively by X-ray diffraction (XRD) and scanning-electron microscopy (SEM).

400

600

800

1000

Temperature (ºC) Fig. 2. TGA curves of fresh and carbonate of UF slag, FA slag, and BHC (Carbonation conditions: PCO2 = 650 psig; T = 60 ◦ C; t = 1 h; particle size < 44 ␮m; L/S = 10 mL g−1 ).

Prior to examining the capacity for CO2 capture, the chemical compositions of steelmaking slags were characterized by inductively coupled plasma atomic emission spectroscopy (ICP-AES), after total digestion using aqua regia to dissolve the solid materials in the sample, by the chemistry analysis laboratory in the China Himent Corporation. However, SiO2 was dissolved further by using hydroﬂuoric acid with increasing temperatures and pressures in a

E.-E. Chang et al. / Journal of Hazardous Materials 195 (2011) 107–114

microwave digestion. The contents of each metal in the extracted solution were measured by the ICP-AES method. Then, the content of metal oxide could be computed using the ICP-AES results. 2.3. TGA The thermal characteristics of the slag before and after carbonation were examined using a thermo-gravimetric analyzer (TGA-51, Shimadzu); this analysis was performed to determine the weight loss using different temperatures for the selected samples. Three weight fractions corresponding to (1) moisture, (2) organic elemental carbon, calcium hydroxide, and MgCO3 , and (3) CaCO3 content were determined mainly at the following temperature ranges: (1) 25–105 ◦ C, (2) 105–500 ◦ C, and (3) 500–850 ◦ C. The weight loss between a temperature range of 500–850 ◦ C is contributed mainly by the decomposition of CaCO3 due to its release of CO2 [6,8,9]. However, it has to be remarked that a continuous weight loss between temperatures of 105 and 1000 ◦ C is due to the dehydration of calcium hydroxide, calcium silicate hydrates, calcium aluminate hydrates, and other minor hydrates [10]. In order to prevent overestimating the CaCO3 content, the weight losses due to the dehydration of hydrates have been modiﬁed by a graphical technique and are illustrated in Fig. 2 [10]. Samples were heated linearly in the temperature range of 25–850 ◦ C at a heating rate of 10 ◦ C/min. The TGA weight fraction determined by means of a graphical technique, based on the dry weight, was assumed to be the CaCO3 content, expressed in terms of CO2 (wt%): CO2 (wt%) =

mCaCO3 m105◦ C

(1)

The carbonation conversion (ıCa ) was determined from the total calcium content of the carbonation product, assuming the initial carbonate content was negligible [8,11]: ıCa (%) = [CO2 (wt%)/100 − CO2 (wt%)] × [MWCa /MWCO2 ]/Catotal (2) where ıCa is the carbonation conversion, MWCa and MWCO2 are the molar weights of Ca and CO2 , respectively, in kg/mol, and Catotal is the total Ca content of the fresh sample in kg/kg. 2.4. SEM and XRD SEM (JSM-6500F, JEOL) was used in this study to produce highresolution three-dimensional images of the sample and to study the surface structure of the slag. SEM was useful particularly in identifying CaCO3 formed on the surface of the slag in the carbonation reaction. XRD (X’ Pert Pro, PANalytical) was used to identify and characterize the CaCO3 crystals in the carbonation products. Monochromatic X-rays were used to determine the interplanar spacing of the sample atoms using Cu as the anode material (K␣˚ at 1 wavelength = 1.540598 A˚ and K␣-2 wavelength = 1.544426 A) an angular step of 1◦ held for 1 s with 2 spanning from 20◦ to 70◦ . When the Bragg conditions for constructive interference were obtained, a “reﬂection” was produced in which the relative peak height was proportional to the number of grains in a preferred orientation. 2.5. Aqueous carbonation Theoretically, the extent of carbonation increases with reaction time. The aqueous carbonation experiments were conducted with reaction times of up to 720 min. The experimental procedures included the following three steps: aqueous CO2 dissolution, Ca leaching, and CaCO3 precipitation. Previous studies by Huijgen et al. [9] had indicated that the inﬂuence of the L/S ratio on carbonation was insigniﬁcant. Therefore, the L/S ratio was ﬁxed at 10 mL g−1

109

Table 1 Physico-chemical properties of UF slag, FA slag, and BHC (CHC Resources Corporation). Parameters

UF slag

Physical properties True density (g/cm3 ) Mean diameter (␮m) BET surface area (m2 /kg) Total pore area (m2 /g) Porosity

FA slag

BHC

2.89 11.67 148 1.76 0.63

2.78 17.35 237 1.32 0.59

2.94 20.63 115 1.13 0.60

Chemical properties SiO2 (%) Al2 O3 (%) Fe2 O3 (%) CaO (%) MgO (%) S2− (%) SO3 (%)

33.93 14.35 0.35 42.43 6.42 0.24 0.52

34.89 15.75 1.97 38.80 5.59 – 0.52

27.34 8.42 2.71 52.82 4.66 – 1.49

Total (%)

98.24

97.52

97.44

in this study. The conditions of the carbonation experiment were as follows: L/S ratio of 10 mL g−1 , a PCO2 of 700 psig, particle size of less than 44 ␮m, and reaction time of 60 min, unless otherwise speciﬁed. To assess the effective initial pH of the water samples for the carbonation conversion, the pH values of the water were prepared from 2 to 12 at a ﬁxed reaction temperature of 100 ◦ C, and a pressure of 700 psig. The initial pH value of solution was adjusted to the designed value using KOH and HNO3 solutions. Then, different steelmaking slags were dispersed intensively in the prepared water at an L/S of 10 mL g−1 . 3. Results and discussion 3.1. Physico-chemical properties of steelmaking slags The physico-chemical properties of the UF slag, FA slag, and BHC feedstocks are presented in Table 1, which shows that the major components of these three steelmaking slags were CaO: 42.43 wt%, 38.80 wt%, and 52.82 wt% for UF slag, FA slag, and BHC, respectively. Minor amounts of SiO2 , Al2 O3 , MgO, Fe2 O3 , S2− , and SO3, which do not contribute to CO2 sequestration because the CO2 -capturing capacity of the slag material is attributed mainly to the CaO components, are also listed in Table 1. A higher adsorption capacity of CO2 on the slags was expected in the carbonation reaction, which was validated by the following experiments. Assuming that all CaO was converted to CaCO3 , the theoretical capacities of the UF slag, FA slag, and BHC were 0.333, 0.305 and 0.415 kg CO2 /kg dry solid, respectively. Fig. 2 shows the weight variation of the fresh and carbonated specimen obtained by using TGA. According to the TGA results, the weight losses of the fresh UF slag, FA slag, and BHC between 500 and 850 ◦ C was insigniﬁcant which indicated that the initial hydrate and carbonate contents of the three feedstocks were negligible. In addition, the TGA curves of carbonated specimen indicate that the CO2 release at high temperatures (above 800 ◦ C) can be neglected due to the lack of a peak at 800 ◦ C in the TGA curves. The weight loss of the carbonated BHC was higher signiﬁcantly than the carbonated UF and FA samples, which can be attributed to the greater CaO content and lower silicon dioxide content of the BHC. Therefore, the BHC conclusively captured a higher amount of CO2 than the other two slags during the carbonation reaction. Fig. 3 shows the SEM images of the fresh (Fig. 3a, c, e) and carbonated (Fig. 3b, d, f) UF slag, FA slag, and BHC, respectively. Comparisons of the SEM images of the feedstocks before and after carbonation showed that cubic particles adhered to the feedstocks

110

E.-E. Chang et al. / Journal of Hazardous Materials 195 (2011) 107–114

Fig. 3. Scanning electron micrographs (SEMs) of (a) fresh and (b) carbonated UF slags; (c) fresh and (d) carbonated FA slags; and (e) fresh and (f) carbonated BHC.

after carbonation. Based on the SEM images and qualitative analysis from XRD, it was determined that the cubic particles were composed of CaCO3 and had diameters ranging from 1 to 2 ␮m, which is similar to the literature [4,11]. The dark cubic particles in the SEM images were found to be CaCO3 which was conﬁrmed by the EDX and XRD analyses. The mineralogical characterizations of fresh and carbonated UF slag, FA slag, and BHC were performed based on the XRD patterns shown in Fig. 4a–c, respectively, which indicated that the main crystal phase of the fresh slags was CaO. In contrast to the XRD results from fresh BHC, CaCO3 was identiﬁed as the primary phase in the reaction products. The peaks in the XRD analysis of the carbonated material appeared at 2 values of 23.02◦ , 29.41◦ , 35.97◦ , 43.15◦ , 47.49◦ , 48.50◦ , 57.40◦ , 60.68◦ , and 64.68◦ (in red line), which are indicative of calcium carbonate. These results suggest that the steelmaking slags should be carbonated with CO2 to form CaCO3 in an autoclave reaction.

3.2. Effects of feedstocks and operating factors The effect of reaction time on the conversion ratio of the three feedstocks at 160 ◦ C and 700 psig is shown in Fig. 5, which indicates that the carbonation rate decreased as the reaction time increased. The reaction leveled off after 60 min, indicating that the carbonation reaction had a stationary phase due to the formation of a SiO2 barrier, which strongly blocks the reactive surface sites and inhibits the release of calcium ions from the slag. These effects exhibit a limited conversion of CO2 during the carbonation reaction, which is consistent with the ﬁndings suggested by Huijgen et al. [4]. Therefore, the maximum efﬁciencies in carbonation conversion (ıCa ) after a reaction time of 720 min for the UF slag, FA slag, and BHC were found to be 38.1%, 34.7%, and 68.3%, respectively. The relatively higher conversion of the BHC also could be explained by the chemical compositions of the slags shown in Table 1, indicating that the CaO content of the BHC was (52.82 wt%) and thus higher than that of the UF (42.43 wt%) and FA (38.80 wt%)

E.-E. Chang et al. / Journal of Hazardous Materials 195 (2011) 107–114

a

111

100

700

Intensity

600 500

80

400 300

δCa (%)

200 100 0 20

25

30

35

40

45

50

55

60

65

70

2θ[ °]

Intensity

b

UF slag FA slag BHC

500 450 400 350 300 250 200 150 100 50 0 20

60

40

20

0 0

2

4

6

8

10

12

Reaction Time (hr)

25

30

35

40

45

50

55

60

65

70

Fig. 5. The inﬂuence of reaction time on carbonation conversion of steelmaking slags (carbonation conditions: PCO2 = 700 psig; T = 160 ◦ C; particle size <44 ␮m; L/S = 10 mL g−1 ).

2θ [°]

c

1800

Intensity

1600 1400 1200 1000 800 600 400 200 0 20

25

30

35

40

45

50

55

60

65

70

2θ [°] Fig. 4. XRD spectra of fresh (in blue line) and carbonated (in red line) slag with peak identiﬁcations ( = CaCO3 ); (a) UF slag; (b) FA slag; and (c) BHC (Carbonation conditions: 700 psig, 160 ◦ C, and 2 h). (For interpretation of the references to color in this ﬁgure legend, the reader is referred to the web version of the article.)

that the use of a large amount of water in the sequestration process could inhibit the carbonation reaction. From a thermodynamic equilibrium perspective, the actual concentration of CO3 2− signiﬁcantly was smaller than the concentration of Ca2+ , even at high partial pressures of CO2 . The solution became strongly alkaline (pH 11.8) before carbonation, whereas the pH of the suspension slurry dropped rapidly after the introduction of pure CO2 gas into the system. The solution ultimately stabilized at a pH of approximately 6.3. This reaction suggests that the carbonation process should eliminate the alkalinity, which was in agreement with the literature [11,12]. The effect of temperature on the feedstocks was assessed by varying the temperature from 40 ◦ C to 160 ◦ C, with a ﬁxed reaction time of 60 min and a CO2 pressure of 700 psig. The conversion increased with increasing temperature (Fig. 7) due to the higher leaching rate of calcium at higher temperatures. However, in the 100

UF Slag FA Slag BHC

80

δCa (%)

slags. The actual CO2 capture capacities per gram of dry solid were 0.127, 0.107, and 0.283 kg CO2 for the UF slag, FA slag, and BHC, respectively. The L/S ratio represents the weight ratio of water-to-waste in the slurry used in the aqueous carbonation. The L/S ratios of the three solid wastes in this study ranged from 0 to 100 mL g−1 . The particle size was controlled to be less than 44 ␮m, the reaction temperature was maintained at 100 ◦ C, and the reaction pressure was set at 700 psig. As shown in Fig. 6, the conversions of the UF slag, FA slag, and BHC were quite low in the absence of water (L/S = 0) because of the slow reaction kinetics of dry carbonation, which is reﬂective of the natural carbonation of minerals in the environment. When the L/S ratio was increased to 10 mL g−1 , the conversions were improved to be 18.3%, 13.9%, and 57.5% for the UF slag, FA slag, and BHC, respectively, because the aqueous carbonation process was dominant. As the L/S ratio further increased, the conversions were not increased signiﬁcantly because the presence of excessive water formed a mass transfer barrier, lowered the ionic strength, and resulted in decreasing the dissolution rate of Ca2+ . The L/S ratio was held from 10 to 20 for these three feedstocks, which implies

60

40

20

0

0

20

40

60

80

100

L/S Ratio (mL/g) Fig. 6. The inﬂuence of liquid-to-solid ratio (L/S) on carbonation conversion of steelmaking slags (carbonation conditions: PCO2 = 700 psig; T = 100 ◦ C; t = 1 h; particle size <44 ␮m).

112

E.-E. Chang et al. / Journal of Hazardous Materials 195 (2011) 107–114

80 70

UF-1300 psi

FA-700 psi

FA-1300 psi

BHC-700 psi

BHC-1300 psi

Simulated Conversion (%)

100 UF-700 psi

60

δCa (%)

50 40 30

80

60

40 UF Slag FA Slag BHC

20 20 0 10

0

20

40

60

80

100

Experimental Conversion (%) 0 20

40

60

80

100

120

140

160

180

Temperature (ºC) Fig. 7. The inﬂuence of reaction temperature and pressure on carbonation conversion of (a) UF slag, (b) FA slag; and (c) BHC (carbonation conditions: t = 1 h; particle size <44 ␮m; L/S = 10 mL g−1 ).

BHC case, the conversion started to decrease at temperatures over 120 ◦ C if the pressure was set at 700 psig. The above evidence may be caused by the reduction of the CO2 dissolution as temperatures increased. These observations were similar to a previous study [11] which found that the CO2 solubility could be the key factor affecting the carbonation conversion at higher temperatures in BHC case. In contrast to the supercritical condition (1300 psig), under which the CO2 ﬂuid has a relatively higher solubility of liquid and lower dynamic viscosity of gas, the conversion of BHC increases due to its superior CO2 solubility at higher temperatures. It was concluded that the temperature signiﬁcantly inﬂuenced the conversion of the carbonation reaction, with increasing temperatures resulting in higher conversions. The CO2 pressure was varied between two types of conditions: a normal condition (700 psig) and a supercritical condition (1300 psig). The conversion of supercritical CO2 was slightly less than that of normal CO2 (Fig. 7), due to the inhibition of CaCO3 crystal growth under this supercritical pressure. However, in general, a sufﬁciently high pressure increases the rate of the carbonation reac100

Fig. 9. Comparison of simulated and experimental conversion values for the carbonation of UF slag, FA slag, and BHC.

tion, which precludes the formation of CaCO3 crystals and inhibits the reaction. To assess the effective initial pH for the carbonation of the three feedstocks, the initial pH values of the water samples were varied from 2 to 12 at a ﬁxed reaction temperature of 100 ◦ C, a pressure of 700 psig, and an L/S of 10 mL g−1 . As shown in Fig. 8, the conversion of the BHC was higher than the other two at all pH values due to its higher CaO content. It also was observed that the three feedstocks had the highest conversions at pH 12 because there were large amounts of CO3 2− in the slurry, which enhanced the carbonation process. The conversions of all three feedstocks were lower when the pH ranged from 6 to 10. However, when the pH was lower than 4, the conversions of the three feedstocks increased because the Ca2+ dissolved thoroughly. Furthermore, in UF slag case, the trace S2− content (∼0.24 wt%) may cause a potential problem of H2 S emissions if the operational pH value was lower than 4 [13]. Therefore, it is required to control experiments at a higher pH (e.g., 10) to increase the degree of aqueous carbonation and prevent an environmental issue. 3.3. Kinetic modeling of the carbonation reaction A surface coverage model, which was originally developed by considering the carbonation of hydrated lime [14], was chosen to determine the kinetics of the carbonation reactions in this study.

0.12

UF slag FA slag BHC

80

0.04 60

CO2 Emission (kg/kg)

δCa (%)

Energy Consumption

0.08

40

20

Sequestration

0.00

Net Emission

-0.04 -0.08 -0.12 -0.16 -0.20

0

Transportation

UF slag FA slag BHC

-0.24 0

2

4

6

8

10

12

14

Initial pH Fig. 8. The inﬂuence of initial pH on carbonation conversion of steelmaking slags (carbonation conditions: t = 1 h; particle size <44 ␮m; L/S = 100 mL g−1 ).

-0.28 Fig. 10. Comparison of CO2 emissions including transportation, energy consumption, and sequestration for various steelmaking slags.

E.-E. Chang et al. / Journal of Hazardous Materials 195 (2011) 107–114

113

Table 2 Surface coverage model and LCA results for the UF slag, FA slag, and BHC. Item

Unit

UF slag

FA slag

BHC

Surface coverage model results Sg M ks kp k1 k2

(m2 /g) (g/mol) (mol/h/m2 ) (m2 /mol) (1/h) (–)

0.15 61.24 12.11 0.27 111.2 0.0290

0.24 61.83 10.69 0.46 158.7 0.0312

0.12 60.21 27.28 0.12 197.01 0.0167

Life cycle assessment results CO2 (Emission)a CO2 (Sequestration) CO2 (Net)b

(kg CO2 /kg solid) (kg CO2 /kg solid) (kg CO2 /kg solid)

a b

0.054 0.035 0.019

0.062 0.025 0.037

0.13 0.098 0.023

Sum of CO2 emission during transportation and energy consumption. Net = CO2 emission − CO2 sequestration.

The measurements of the fresh and carbonated materials by SEM in conjunction with XRD provided evidence that indicated the suitability of using the surface coverage model in this investigation. The surface composition and molecular structure changed during the course of carbonation. The small CaCO3 particles formed on the surface of slags, which indicated that the carbonation reaction occurred on the surface of slags and formed a protective layer around the reacting particles, thus inhibiting further reaction. Assume that the rate-determining step of carbonation is the surface reaction, which occurs only at unreacted surface sites not covered by product, then the carbonation reaction rate per initial surface area of the solid can be expressed by Eq. (3): rs = ks ˚

(3)

resulted in the largest kp , which suggested that the product was deposited on its surface faster than it was on the other materials because of its higher SiO2 content. It was concluded that the BHC was the most reactive material for carbonation among the three feedstocks tested. Fig. 9 presents the relationship between the experimental and model values and indicates that the experimental values were consistent with the model values, which suggests that the surface coverage model is suitable to describe the carbonation of the steelmaking slag. The standard errors (%) between the experimental and model values of the UF slag, FA slag, and BHC were 5%, 6%, and 7%, respectively, indicating that the surface coverage model could simulate successfully the carbonation reaction of these slag materials.

(mol/h/m2 )

where ks is the rate constant which is a function of temperature, the concentrations of reacting species, and the relative humidity; and ˚ is the fraction of surface sites (or area) still active and not covered by the reaction product. The rate of conversion, ıCa (%), can then be expressed by Eq. (4): dı = Sg M rs = Sg M ks ˚ dt

(4)

where Sg (m2 /g) is the initial speciﬁc surface area of the solid waste, and M (g/mol) is the weight of solid waste per mole. ˚ is a function of time and is indicative of the manner in which the product is deposited on the surface. Hence, ˚ changes with reaction time and is dependent on the reaction rate, and one may assume that ˚ can be expressed by the following equation: −d˚ = kp rs = kp ks ˚ dt

(5)

where kp (m2 /mol), a function of temperature, the concentrations of reacting species, and the relative humidity, is a proportional constant that reﬂects the fraction of the surface that is reactive and not covered by the reaction product. Integration of Eq. (5) enables ˚ to be expressed as a function of time, as shown in Eq. (6), and assuming that k1 = ks Sg M and k2 = kp /(Sg M) ˚ = exp(−k1 k2 t)

(6)

By substituting Eq. (6) into Eq. (4), the integration of Eq. (4) can be used to describe the relationship between the conversion and reaction times: ı=

[1 − exp(−k1 k2 t)] k2

(7)

Table 2 shows the values of Sg, M, ks , kp , k1 , and k2 determined by non-linear regressions for each of the three materials based on the data obtained from experiments conducted at 160 ◦ C, which indicated that the BHC had the fastest reaction rate (ks ) due to its higher CaO content and lower SiO2 content. However, the FA slag

3.4. Comparison of various carbonation processes CCS (carbon capture and storage) is an energy-intensive process; thus, it may consume additional energy, leading to further CO2 emissions compared to the non-CCS treatment process. However, the consumption of energy for accelerated sequestration has decreased with recent technology developments. The factors that affected the system operation were reaction temperature, initial pH with various feedstocks, and CO2 pressure. Many researchers have attempted to capture CO2 with lower power consumption and less chemical usage. In this study, the highest conversion (ıCa ) for the BHC was 68%, when the aqueous carbonation was conducted at an L/S of 10 mL g−1 , a partial pressure of CO2 of 700 psig, and a temperature of 160 ◦ C, with a reaction time of 12 h in an autoclave reactor. Table 3 shows a comparison of the experimental results in this study with others reported in literature. In a previous investigation [11], the highest conversion for the BHC was 48%, in a reaction conducted at 101.3 kPa and 60 ◦ C in a slurry reactor. A similar study by Huijgen et al. [9] on accelerated carbonation experiments using wollastonite in an autoclave was conducted with a reaction time of 15 min at 200 ◦ C, CO2 partial pressure of 20 bar and a particle size <38 ␮m, which resulted in a maximum conversion of 70%. In addition, the BF slag conversion by the pH-swing method from Kodama et al. [15] was 80% at a relatively low temperature of 40 ◦ C and pressure of 1.9 psig. However, previous studies using the pHswing method consumed great amounts of base and acid solvents, which could result in adverse effects on the environment. Furthermore, O’Connor et al. [16] and Lackner et al. [17] demonstrated that an extremely high carbonation conversion (91–100%) could be achieved at higher temperatures (185–500 ◦ C) and pressures (1682–4930 psig). It was concluded that this investigation exhibited a higher BHC carbonation conversion (68.3%) at relatively a lower temperature (160 ◦ C) and pressure (600 psig).

114

E.-E. Chang et al. / Journal of Hazardous Materials 195 (2011) 107–114

Table 3 Comparison of experimental results among the literature and this study. Material type

Method

CO2 conc.

Pressure

Temperature

Particle size (␮m)

Time

Conversion

Reference

Wollastonite Olivine Mg(OH)2 Converter slag Blast furnace slag APC Residues CKD BHC BHC

Aqueous carbonation Aqueous carbonation Direct carbonation pH-swing pH-swing Aqueous carbonation Direct carbonation Aqueous carbonation Aqueous carbonation

100 vol% 100 vol% 100 vol% 13 vol% 100 vol% 100 vol% 35,600 ppm 100 vol% 100 vol%

20 bar 1682 psig 4930 psig 1.9 psig 14.7 psig 3 bar Atmospheric 14.7 psig 700 psig 1300 psig

200 ◦ C 185 ◦ C 500 ◦ C 40 ◦ C 30 ◦ C 30 ◦ C Ambient 60 ◦ C 160 ◦ C 160 ◦ C

<38 <37 <20 <63 <10 Not speciﬁed Not speciﬁed <44 <44 <44

15 min 1 day 2h 1h 15 min 5h 12 days 1h 12 h 1h

70% 91% 100% 80% 72.5% 67% 70.6% 47.5% 68.3% 59.2%

Huijgen et al. [9] O’Connor et al., [16] Lackner et al., [17] Kodama et al., [15] Eloneva et al., [2] Baciocchi et al., [18] Huntzinger et al., [6] Chang et al., [11] This study

Life cycle assessment (LCA) is a method in which the energy and raw material consumptions, types of emissions, and other important issues related to a speciﬁc product are measured, analyzed, and evaluated from an environmental point of view. In this investigation, LCA was utilized to determine the CO2 emission including transportation, energy consumption, and sequestration (Fig. 10). The transportation and energy consumptions including grinding, sieving, heating, and pressuring, could lead to additional CO2 emissions. In addition, although BHC contains cement as small as 10% (CEM III/C), the process of limestone decomposition alone produces CO2 for about 0.05 kg/kg BHC during the cement manufacturing process. Therefore, in case of BHC, the amounts of CO2 emission from cement manufacturing also were taken into account. The calculated net CO2 emission and sequestration are shown in Table 2, which indicates that the net CO2 emissions of the UF slag, FA slag, and BHC were 0.019, 0.037, and 0.023 kg CO2 /kg solid, respectively. The heating process was the most energy intensive throughout all systems, which accounts for 92.6% of total energy consumption. In commercial operation, however, this problem could be improved by utilizing the waste heat from the industrial process to minimize the additional energy consumption. Therefore, it suggests that the BHC should be a feasible feedstock to sequester CO2 through the carbonation according to the preliminary results of LCA. 4. Conclusions UF slag, FA slag, and BHC were selected as feedstocks for the aqueous carbonation process in this study. The initial carbonation levels of the fresh slags were found to be negligible based on the TGA curves. The three feedstocks were alkaline, calcium-rich particles that reacted with CO2 dissolved in the aqueous slurry to form CaCO3 coating on the surface of the slag. In this study, the highest conversion (ıCa ), 68.3%, for BHC was achieved when the aqueous carbonation was conducted at an L/S of 10 mL g−1 , a CO2 partial pressure of 700 psig, and a temperature of 160 ◦ C, with a reaction time of 12 h in an autoclave reactor. In comparison with a previous report, the BHC exhibited a better performance at a lower temperature (160 ◦ C) and pressure (700 psig). Most of the carbonation reaction occurred within the ﬁrst hour of reaction time, and the conversion efﬁciency was lower signiﬁcantly after 12 hr of reaction time. In addition, it is required to control experiments at a higher pH (e.g., 10) to increase the degree of aqueous carbonation and prevent an environmental issue. Kinetics of this reaction were described using a surface coverage model, in which the experimental data were shown to be consistent with the predicted values. The CaCO3 product formed was identiﬁed as crystallized calcite, and the surface composition and molecular structure were found to be varied in the course of carbonation based on SEM and XRD measurements. Conclusively,

the aqueous carbonation of these steelmaking slags, including UF slag, FA slag, and BHC, in an autoclave reactor is feasible and results in a high conversion. Acknowledgements This study was support by National Science Council of Taiwan, R.O.C., under Grant No. NSC-100-3113-E-007-005. The China Steel Corporation and CHC Resources Corporation were highly appreciated by authors for providing the steelmaking slag in this investigation. References [1] K.S. Lackner, A guide to CO2 sequestration, Science 300 (2003) 1677–1678. [2] S. Eloneva, S. Teir, et al., Fixation of CO2 by carbonating calcium derived from blast furnace slag, Energy 33 (2008) 1461–1467. [3] K.S. Lackner, Carbonate chemistry for sequestration fossil carbon, Annu. Rev. Energy Environ. 27 (2002) 193–232. [4] W.J.J. Huijgen, G.J. Witkamp, R.N.J. Comans, Mineral CO2 sequestration by steel slag carbonation, Environ. Sci. Technol. 39 (2005) 9676–9682. [5] W. Huijgen, G.J. Witkamp, R. Comans, Mineral CO2 sequestration in alkaline solid residues, Greenhouse Gas Control Technol. 7 (2005) 2415–2418. [6] D.N. Huntzinger, J.S. Gierke, S.K. Kawatra, T.C. Eisele, L.L. Sutter, Carbon dioxide sequestration in cement kiln dust through mineral carbonation, Environ. Sci. Technol. 43 (2009) 1986–1992. [7] European Standards EN 197-1: 2000 (Amendment A1: 2004): Cement – Part 1: Composition, speciﬁcations and conformity criteria for common cements, European Committee for Standardization, in, 2004. [8] E. Bouquet, G. Leyssens, C. Schönnenbeck, P. Gilot, The decrease of carbonation efﬁciency of CaO along calcination–carbonation cycles: experiments and modelling, Chem. Eng. Sci. 64 (2009) 2136–2146. [9] W.J.J. Huijgen, G.J. Witkamp, R.N.J. Comans, Mechanisms of aqueous wollastonite carbonation as a possible CO2 sequestration process, Chem. Eng. Sci. 61 (2006) 4242–4251. [10] B.K. Marsh, R.L. Day, Pozzolanic and cementitious reactions of ﬂy ash in blended cement paste, Cem. Concr. Res. 18 (1988) 301–310. [11] E.E. Chang, C.H. Chen, Y.H. Chen, S.Y. Pan, P.C. Chiang, Performance evaluation for carbonation of steel-making slags in a slurry reactor, J. Hazard. Mater. 186 (2011) 558–564. [12] M. Uibu, R. Kuusik, Mineral trapping of CO2 via oil shale ash aqueous carbonation: controlling mechanism of process rate and development of continuous-ﬂow reactor system, Oil Shale 26 (2009) 40–58. [13] F.M.M. Morel, J.G. Hering, Principles and Applications of Aquatic Chemistry, John Wiley & Sons Inc., 1993. [14] C.F. Liu, S.M. Shih, R.B. Lin, Kinetics of the reaction of Ca(OH)2 /ash sorbent with SO2 at low temperatures, Chem. Eng. Sci. 57 (2002) 93–104. [15] S. Kodama, T. Nishimoto, et al., Development of a new pH-swing CO2 mineralization process with a recyclable reaction solution, Energy 33 (2008) 776–784. [16] W.K. O’Connor, D.C. Dahlin, D.N. Nilsen, S.J. Gerdemann, G.E. Rush, R.P. Walters, P.C. Turner, Research status on the sequestration of carbon dioxide by direct aqueous mineral carbonation, in: 18th Annual International Pittsburgh Coal Conference, Newcastle, Australia, 2001. [17] K.S. Lackner, D.P. Butt, C.H. Wendt, Progress on binding CO2 in mineral substrates, Energy Convers. Manage. 38 (1997) S259–S264. [18] R. Baciocchi, G. Costa, A. Polettini, R. Pomi, V. Prigiobbe, Comparison of different reaction routes for carbonation of APC residues, Energy Procedia 1 (2009) 4851–4858.

Journal of Hazardous Materials 195 (2011) 115–123

Contents lists available at ScienceDirect

Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat

The oxidative corrosion of carbide inclusions at the surface of uranium metal during exposure to water vapour T.B. Scott a,∗ , J.R. Petherbridge b , N.J. Harker a , R.J. Ball a , P.J. Heard a , J. Glascott b , G.C. Allen a a b

Interface Analysis Centre, University of Bristol, 121 St Michaels Hill, Bristol, BS2 8BS, United Kingdom AWE, Aldermaston, Reading, Berkshire, RG7 4PR, United Kingdom

a r t i c l e

i n f o

Article history: Received 13 May 2011 Received in revised form 3 August 2011 Accepted 4 August 2011 Available online 10 August 2011 Keywords: Uranium Water vapour Oxidation Carbide Inclusions

a b s t r a c t The reaction between uranium and water vapour has been well investigated, however discrepancies exist between the described kinetic laws, pressure dependence of the reaction rate constant and activation energies. Here this problem is looked at by examining the inﬂuence of impurities in the form of carbide inclusions on the reaction. Samples of uranium containing 600 ppm carbon were analysed during and after exposure to water vapour at 19 mbar pressure, in an environmental scanning electron microscope (ESEM) system. After water exposure, samples were analysed using secondary ion mass spectrometry (SIMS), focused ion beam (FIB) imaging and sectioning and transmission electron microscopy (TEM) with X-ray diffraction (micro-XRD). The results of the current study indicate that carbide particles on the surface of uranium readily react with water vapour to form voluminous UO3 ·xH2 O growths at rates signiﬁcantly faster than that of the metal. The observation may also have implications for previous experimental studies of uranium–water interactions, where the presence of differing levels of undetected carbide may partly account for the discrepancies observed between datasets. Crown Copyright © 2011 Published by Elsevier B.V. All rights reserved.

1. Introduction

2. Experimental

The interaction between metallic uranium surfaces and water vapour is considered to be most important in regard to the environmental corrosion of the metal. Numerous studies have examined the initial stages of these interactions [1–7]. However, there have been discrepancies in the published data describing kinetic laws, pressure dependence of the reaction rate constant and activation energies. The precise mechanism for uranium corrosion is not entirely clear with different mechanisms proposed arising from the results of the undertaking studies [1–7]. Existing discrepancies in the published data may, in part, be related to differences in the provenance and purity of the metal used by different groups. The reactivity of impurity species such as carbide, may have affected recorded data. This work aims to provide data for an improved understanding of the role of impurity phases in the uranium–water reaction, samples of uranium containing 600 ppm carbon were analysed during and after exposure to water vapour at 19 mbar pressure, in an environmental scanning electron microscope (ESEM). Samples were analysed using secondary ion mass spectrometry (SIMS), focused ion beam (FIB) imaging and sectioning and transmission electron microscopy (TEM) with X-ray diffraction (micro-XRD).

2.1. Starting materials

∗ Corresponding author. Tel.: +44 117 3311176. E-mail address: [email protected] (T.B. Scott).

The uranium used in the experimental work was cast, depleted uranium containing 600 ppm carbon. Prior analysis of the microstructure before experiment revealed a coarsely grained metal, with grains frequently >100 ␮m in width with long, relatively straight, low angle (<25◦ ) grain boundaries and occasional sets of well deﬁned crystal twins. Inclusion particles were frequently observed across the metal surface, present as individual particles and conjoined clusters. A surface inclusion number density of 575 per mm2 (±20) was measured for the metal, with an average inclusion diameter of 5 ␮m. TEM analysis of a small number of inclusions (removed from a high-carbon uranium sample using FIB) indicated that the inclusions were of mixed UC-UN (carbonitride) composition with FCC crystal structure [8,9]. Hereafter, and for simplicity, these inclusions are referred to as carbides. Examples of carbide particles exposed at the uranium surface are shown in Figs. 1–3 and highlight the occurrence of both individual and clustered ‘compound’ particles. 2.2. Experimental methods Uranium sample coupons (15 mm diameter and 1 mm thick), containing 600 ppm carbon were mechanically polished using wetted Buehler SiC grit papers of increasingly ﬁne grade down to

0304-3894/$ – see front matter. Crown Copyright © 2011 Published by Elsevier B.V. All rights reserved. doi:10.1016/j.jhazmat.2011.08.011

116

T.B. Scott et al. / Journal of Hazardous Materials 195 (2011) 115–123

Fig. 1. Secondary electron images of typical electropolished surface regions on the sample metal. Carbide inclusions are highlighted as dark spots 3–8 ␮m in size.

Fig. 2. Secondary electron image of a typical carbide inclusion at the surface of the uranium. The inclusion is obviously zoned, with an outer shell assumed to be uranium carbide, and an inner core of nitride.

4000 grit (equivalent to a 2–3 ␮m surface ﬁnish). Once a satisfactory ﬁnish was achieved, each coupon was rinsed and washed using ethanol and electro-polished in a 10:6:6 mixture of ethanol, orthophosphoric acid and ethylene glycol, for a 10 min period. The sample was then dipped in the unbiased electrolyte and rinsed successively with deionised water, ethanol and acetone. Directly after electro-polishing, the samples were immediately loaded into an Electroscan environmental secondary electron microscope system (ESEM). Water vapour was admitted to the chamber at a constant controlled pressure of 19 mbar, at approximately 20 ◦ C, and the sample surface was imaged at regular intervals for a period of up to one week. Secondary electron images

Fig. 3. Secondary electron image of a carbide inclusion cluster at the surface of the uranium, showing numerous compound particles.

taken from speciﬁc areas on the sample surface were compiled, to produce a ‘time-lapse’ sequence of surface reaction. The experiment was repeated under identical conditions three times using deionised water as the vapour source and a further three times using a 1:1 mixture of high performance liquid chromatography (HPLC) grade and distilled water as the vapour source. After each experiment the ‘reacted’ sample coupon was immediately transferred for analysis by FIB imaging and milling with supplementary analysis by SIMS. TEM specimens were also prepared using FIB milling and ex-situ lift-out, as described in Ishitani et al. [10]. Further samples were prepared then exposed to water vapour in the ESEM after using only mechanical polishing (down to a 4000 grit paper) to level the sample surface and remove any oxide present. This experiment was performed in order to determine if sample reactivity differed between preparation methods. In addition, water exposure was carried out using a separate gas treatment rig to repeat the ESEM experiments. A sample coupon was prepared following the aforementioned procedure (polishing and electropolishing) and loaded into a gasket sealed stainless steel reaction cell of ∼30 cm3 volume. The cell was evacuated to better than 10−5 mbar prior to the experiment, and then ﬁlled with water vapour to 20 mbar pressure at approximately 22 ◦ C. High purity (<10 ppb O2, <20 ppb H2 O) argon gas was then immediately added to generate a total system pressure of 1000 mbar and the cell was left for a 24 h period to allow reaction of the metal surface with the water vapour.

2.3. Analytical methods Both an Electroscan environmental secondary electron microscope system and a FEI FIB Strata 201 focused ion beam system were used to examine the sample morphology and micro-texture. The resolution of each system is dependent on the operating conditions employed, for example, at 30 keV beam energy, the FIB resolution is 500 nm for an operating current above 11 nA and 5–7 nm for a current of 1 pA. Surface images were recorded from the sample at tilt angles of both 0◦ and 45◦ to provide complimentary topographical and structural information. Whilst the ESEM system was primarily used for the water vapour exposure of the metal (rather than imaging), the FIB instrument was used for sample sectioning and preparation of TEM lamella for analysis. TEM images and electron diffraction patterns were obtained using a Philips EM 430 TEM operating at 250 keV beam energy. TEM lamella were mounted on 200 mesh carbon-coated copper grids using an ex situ lift-out method, after FIB preparation. Subsequently a selected area aperture was used to record images

T.B. Scott et al. / Journal of Hazardous Materials 195 (2011) 115–123

117

Fig. 4. Secondary electron images of carbide inclusions on the surface of the cast uranium following reaction with water vapour at 20 mbar for 12 h at 20 ◦ C. Horn-like growths were observed to grow at inclusion edges and corners.

and diffraction patterns from speciﬁc areas of the TEM lamella to identify the various phases present. Secondary ion mass spectrometry was performed in negative ion mode using a spectrometer previously constructed at the University of Bristol. The SIMS system employed a focused gallium ion source (FEI electronically variable aperture type) ﬁtted to a Vacuum Generators model 7035 double-focusing magnetic sector mass analyser. During analysis numerous ion maps were recorded at different magniﬁcations with beam currents of 0.5–1 nA. Negative ions and ion clusters corresponding to H− , C− , O− , OH− and CN− were mapped from discreet areas of the sample surface, with those for C− and CN− being considered to be most characteristic of the carbide inclusions, and H− , O− and OH− ion clusters considered to best represent the observed surface reaction products. Ion maps for UO+ and UO2 + clusters were also recorded and found to be representative of surface oxides formed on the metal. 3. Results 3.1. Water-sample interactions under ESEM conditions Upon exposure to water vapour at 20 ◦ C (equivalent to ∼85% relative humidity) oxidation of the uranium surfaces was observed. FIB sections cut through the sample material indicated that the oxide on the metal surface thickened considerably over the duration of experiments. After preparatory electropolishing the oxide thickness on the starting material was determined to be 5–20 nm thick (depending on time taken to transfer to the instrument). For comparison, exposure to water vapour for a 24 h period resulted in recorded oxide thicknesses of 90–120 nm, with close to 900 nm thickness after one week. Whilst the ESEM instrument could not easily detect thickening of oxide on the metal, the oxidative decomposition of the surface carbides was readily apparent, forming voluminous horn-like growths as imaged in the FIB system (Figs. 4 and 5). Corrosion growth was often observed to initiate at the corners and edges

of carbide particles and also at boundaries between conjoined particles. Figs. 4 and 5 show a number of representative growth sites, where corrosion product can clearly be seen extending from the carbide to overlie the surrounding surface of the metal. The period of time between the introduction of water vapour and the onset of growth formation was observed to vary from carbide to carbide. Whilst some carbide particles exhibited corrosion initiation after periods of only minutes, other particles showed no signs of reaction until days into the experiment. Analysis of the samples exposed to water vapour for one week revealed that all visible surface carbide inclusions had undergone some degree of oxidative corrosion and hydrous growth formation. Figs. 6 and 7 display images recorded from deﬁned areas of the sample material before, during and after water vapour exposure in the ESEM system. Fig. 7 clearly illustrates the gradual nature of corrosion growth on a carbide particle. After a period of 6 h water exposure the growth had apparently halted, whilst new ones were formed (and grew) elsewhere on the same carbide particle. High resolution SEM imaging of reacted carbide particles clearly indicated that the growths initiated beneath a protective surface layer and burst outwards to form a voluminous reaction product (Fig. 5). The protective surface layer on the carbide particles was ascribed to a mixed layer of uranium oxide, uranium hydroxide and free carbon [11] formed during sample preparation and subsequent transfer to the ESEM instrument. Sections through the growths and parent carbide particles were made with a FIB instrument. Numerous growths were sectioned at different periods of water exposure to chart their relative growth into the carbide particles. An example of a carbide inclusion after 24 h water vapour exposure is shown in Fig. 8. The image shows that the carbide had been completely consumed by a polycrystalline growth of much larger volume, with individual grains ≤100 nm in size. The section also highlights that the oxide covering the metal surface (∼87 nm thick) thickens down the margins of the inclusion site. This phenomenon was not observed for other sections where the carbide had not fully decomposed (Fig. 9).

118

T.B. Scott et al. / Journal of Hazardous Materials 195 (2011) 115–123

Fig. 5. Secondary electron images of carbide inclusions on the uranium surface following reaction with for 24 h at 20 mbar water vapour pressure and 20 ◦ C. The horn-like precipitates are seen to fragment the surface layer of the carbide during growth.

For each section it was clear that the growth has formed solely from the carbide inclusion and showed no reaction with the surrounding metal (Fig. 9). The cross-sectional shape of the reaction front in the carbides indicated that, once initiated, growth occurred more slowly at the carbide margins than in the bulk of the crystal (Fig. 9). Within the ESEM, growths were observed to stop developing after a period of reaction and it is presumed that either the growth’s reaction-front reached an internal chemical boundary within the carbide (possibly zoning from nitride to carbide), which prevented continued reaction. Concurrent work using combined FIB and electron backscatter diffraction (EBSD) technique has demonstrated that the carbide inclusions present in the metal are single crystals. Additionally SIMS and TEM analysis has shown that these carbide inclusions may also be chemically zoned from nitride to carbide, without a signiﬁcant change in lattice structure.

In the current experiments, the apparent preference for growth nucleation at the edges and corners of inclusion particles indicates that these parts of the carbide crystals represent low-energy zones for reaction initiation, possibly due to the presence of incomplete (dangling) bonds and/or structural defects at the termination of the crystal lattice. TEM analysis was performed on a number of lamella produced by FIB ion milling and the structure of the metal, carbide and growth phases was identiﬁed using micro-XRD. An example of FIB preparation is shown in Fig. 10. The metal was conﬁrmed as ␣-uranium and the carbide inclusions were determined to have the expected cubic monocarbide (UC) structure. The growth forming from the carbide inclusions was determined as a nanocrystalline UO3 ·xH2 O with 0.5 < x < 1.0. The conﬁrmation of hydrated UO3 (metaschoepite) as the corrosion product phase provides direct chemical evidence for the reaction of the carbide with water vapour.

T.B. Scott et al. / Journal of Hazardous Materials 195 (2011) 115–123

119

Fig. 6. In situ secondary electron images of carbide inclusions on the uranium surface before and after water vapour exposure for a 24 h period at 19 mbar and 20 ◦ C.

Fig. 7. Secondary electron images of a carbide inclusion cluster on the surface of the cast uranium with increasing lengths of exposure to water vapour at 19 mbar and 20 ◦ C.

Fig. 8. Secondary electron images of a carbide inclusion cluster on the surface of the cast uranium with increasing lengths of exposure to water vapour at 19 mbar and 20 ◦ C.

120

T.B. Scott et al. / Journal of Hazardous Materials 195 (2011) 115–123

The TEM-determined composition of the hydrated product was found to correlate well with supplementary SIMS data, which recorded a signiﬁcant concentration of positive UOH+ ion clusters from the growth phases that was not recorded from the metal or un-reacted carbide inclusions (Fig. 11). 3.2. Water-sample interactions – water purity and gas effects

Fig. 9. Secondary electron images of a FIB cut section through a water-reacted carbide inclusion showing consumption of the carbide through hydrous growth formation. The reaction front is clearly visible.

Repeated experiments carried out in the ESEM, using water of different purity (deionised versus distilled/HPLC), showed an apparent difference in the rate of carbide attack and metaschoepite growth, with the water of higher purity displaying more rapid reaction with the surface carbides. Additionally, in some ESEM experiments the vapour exposure was deliberately halted (often overnight) by removal of the water vapour using vacuum pumping. During these dormant periods atmospheric gases were allowed to enter the chamber at very low pressures (∼5 torr) to assist with sample imaging. Compared with samples that experienced uninterrupted H2 O exposure for the duration of their deﬁned reaction period, samples which were ‘stop-started’ showed apparently slower subsequent reaction of

Fig. 10. Secondary electron images of a water-reacted carbide inclusion, before and after preparation for TEM analysis using the FIB system. The section face clearly shows the hydrous growth, carbide inclusion, surrounding metal, surface oxide layer and FIB deposited protective Pt strap.

Fig. 11. Negative ion maps recorded for an area of the water-reacted uranium surface, showing the association of different ion clusters with the metal, carbide and metaschoepite growths.

T.B. Scott et al. / Journal of Hazardous Materials 195 (2011) 115–123

121

Fig. 12. Secondary electron images of a uranium sample, prepared only by mechanical polishing, before and after exposure to water vapour a 20 mbar and 20 ◦ C in an ESEM system. The image after 24 h exposure clearly shows the development of numerous hydrous growths at carbide inclusions in the metal.

the surface carbides. It is inferred that during dormant periods, atmospheric gas species entering the ESEM chamber would have adsorbed on the sample surface (including carbides) and occupied available sorption sites, subsequently limiting water sorption when the experiment was resumed. A similar behaviour was observed for water exposure carried out in the gas rig, which showed relatively little reaction of the carbide inclusions on the metal surface when exposed to an H2 O–Ar gas mixture. Although metaschoepite growths were observed, the number recorded was signiﬁcantly less than would have been expected from a sample grown in the ESEM for the same period. It is not yet clear whether carbide–H2 O decomposition reactions were limited by the presence of the argon (chosen as an inert bystander gas), the overall gas pressure of the system, the free volume of water vapour available or the presence of impurities in the water or argon used for experiment.

3.3. ESEM water-sample interactions – surface preparation effects From the initial tranche of experimental work using electropolished coupons it was not clear if the observed reactivity of the surface carbides was inadvertently related to the electropolishing process, which may have ‘activated’ the carbide surface in some

way. Consequently, to test this theory, further samples were prepared and exposed to water vapour in the ESEM after using only mechanical polishing down to a 4000 grit ﬁnish. The results from analysis by electron microscopy clearly showed similar reactivity of the carbides in the presence of water vapour. Examples of the voluminous metaschoepite growths observed are shown in Figs. 12 and 13 and demonstrate that carbide reactivity is unaffected by preparatory electropolishing of the samples.

4. Discussion Although the metallic uranium surface was determined to undergo oxidation in the presence of water vapour in the ESEM, the oxidative reaction of the carbide particles, present as impurity phases, was also observed. However, attempts to grow UO3 ·H2 O on carbide inclusions outside of the ESEM have met with more limited success, attributed to a number of possible factors including water purity, contaminant gases, and overall gas pressure of the reaction system. Further work will be aimed at determining which factors are most signiﬁcant in limiting carbide decomposition, with the use of isotopic labelling to clarify further the mechanism of UO3 ·H2 O growth. Analysis also indicated that the surfaces of the carbide inclusions were covered by a thin ﬁlm, assumed to be uranium oxide

Fig. 13. High resolution secondary electron images of hydrous metaschoepite growths shown in Fig. 12.

122

T.B. Scott et al. / Journal of Hazardous Materials 195 (2011) 115–123

(UO2 ). The oxide layer (also present on the metal) is thought to have acted as a barrier preventing immediate contact between the water vapour and carbide particles and evidenced by an apparent ‘induction’ period of a few minutes. The carbide particles clearly provide zones of chemical impurity in the metal for initiation of corrosion reactions. The observation of UO3 ·H2 O growths at individual points on the carbide surfaces suggests that oxidative decomposition is energetically favourable in these zones compared to the rest of the exposed particle. Particle corners, edges and boundaries were found to be the primary sites for corrosion initiation. This could be due to disruption of either the inclusion crystal structure or alternatively the inclusion surface oxide layer, resulting in sites of increased susceptibility to oxidative corrosion. Previously reported reactions of UC and UN with water are given in the equations below [11–15]. UC(s) + 2H2 O(g) → UO2(s) + C(s) + 2H2(g)

(1)

UC(s) + 2H2 O(g) → UO2(s) + CH4(g)

(2)

3UN(s) + 2H2 O(g) → UO2(s) + U2 N3(s) + 2H2(g)

(3)

UN(s) + 2H2 O(g) → UO2(s) + NH3(g) + 1/2H2(g)

(4)

These support the detection of various small organic molecules (e.g. acetylene, methane, ethane, ethene, higher order alkenes and alkanes, as well as H2 ) highlighted in various reports [2–9,11–21]. However, it should be noted that in the current work the observed growth species is UO3 based, rather than the UO2 which is traditionally expected to be formed [11–13,21]. The current result is partly supported by the work of Bradley and Ferris [22], who reported a “gelatinous, hydrous, tetravalent uranium oxide” as the product of bulk UC hydrolysis. According to Dell et al. [14] UN is much more stable with respect to water than UC, which might provide some explanation for the observed cessation of some metaschoepite growths on carbide particles, related to chemical zoning of the inclusion from UN to UC. Dell’s proposed reason for greater stability of UN is the presence of a thin coherent epitaxial ﬁlm of ␣-U2 N3 and super imposed UO2 at the surface of the nitride which acts to protect the surface from ready reaction. By comparison, a similar ﬁlm on UC is considered unlikely due to the lack of an intermediate phase for UO2 to bond to (as UO2 appears incapable of bonding directly to UC due to the signiﬁcant difference in lattice parameters). Based on the data presented in the current work the reactions considered occurring between the carbide particles and water vapour are given in Eqs. (5) and (6). UC(s) + 4H2 O(g) → UO3 ·H2 O(s) + CH4(s) + H2(g)

(5)

2UN(s) + 8H2 O(g) → 2UO3 ·H2 O(s) + 2NH3(g) + 3H2(g)

(6)

The change in Gibbs free energy for the reactions given in Eqs. (5) and (6) are −431.8 and −462.6 kJ mol−1 , respectively. Thus, both of the proposed reactions are concluded to be thermodynamically viable (at 298 K and 1 atm). However, these reactions are not yet considered deﬁnitive, and further work using residual gas analysis will be used to determine the gases generated during oxidative decomposition of the carbide inclusions. Although the rate of reaction of H2 O with the carbo-nitride inclusions has not been empirically assessed separate to the metal, it is possible to make an estimation of the relative rates of reaction. This is undertaken using the experimental observations that some inclusions were completely consumed in 7 days and the surface UO2 layer present on the surrounding uranium metal thickened by ∼880 nm over the same period. In addition, it is observed that the inclusions present in the uranium samples have partitioned zones of UC and UN composition, thus calculations were undertaken assuming inclusion composition of either UC or UN.

A rate of H2 O consumption can be calculated for a hypothetical UC inclusion assuming: (i) an average inclusion diameter of 5 ␮m, giving an inclusion volume of 125 ␮m, (ii) the density of UC is 13.63 g cm−3 , (iii) reaction proceeds as set out in Eq. (5), (iv) the entire inclusion is consumed in 7 days. This yields a rate of H2 O consumption by UC of 1.80 × 10−10 mol cm−2 s−1 . A similar calculation can be carried out for a hypothetical UN inclusion (density = 14.31 g cm−3 , volume = 125 ␮m) being consumed after 7 days by the reaction as set out in Eq. (6). This gives a rate of H2 O consumption by UN of 1.87 × 10−10 mol cm−2 s−1 . Using the same approach, the rate of H2 O consumption can be calculated for the uranium metal surface. This is done by assuming the following: that a surface area of 1 ␮m is considered; the thickness of UO2 produced after 7 days is 880 nm; the density of UO2 is 10.97 g cm−3 and UO2 is produced as described in Eq. (7). Thus, a rate of H2 O consumption by the uranium surface of 1.18 × 10−11 mol cm−2 s−1 is obtained U(s) + 2H2 O(g) → UO2 + 2H2

(7)

It is evident that the estimated rate of H2 O consumption is approximately 15 times faster for decomposition of either a hypothetical UC or UN inclusion, as compared to oxidation of the uranium surface. Thus, it is reasonable to conclude that the experimental rate of H2 O consumption due to reaction with the carbo-nitride inclusions to form UO3 ·H2 O will be of the order of 15 times faster than the consumption of H2 O due to the formation of UO2 at the surface of the surrounding uranium. However, it is vital to determine the relative contributions to the total H2 O consumption of reaction with the inclusions and the surrounding metal. This can be achieved by considering the relative surface areas of inclusion and (UO2 covered) metal available for reaction with H2 O. If an area of 1 mm2 is considered, the number of inclusions present at the surface has been determined to be 575. Given an average inclusion diameter of 5 ␮m, each inclusion has an exposed surface area of 25 ␮m and the total surface area of inclusions within an area of 1 mm2 is 0.0144 mm2 , i.e. 1.44% of the surface area. Therefore, if it is assumed that 100% of the volume of surface inclusions is consumed then the relative rates of H2 O consumption can be calculated for an area of 1 cm2 , using the area speciﬁc rates given above and the inclusion/metal surface area ratio of 1.44%. This gives H2 O consumption rates of 2.60 × 10−12 mol s−1 , 2.70 × 10−12 mol s−1 and 1.17 × 10−11 mol s−1 for UC, UN and U, respectively. Assuming that the rates are additive, the contribution from inclusions is ∼18% of the total rate. However, it was observed that not all inclusions underwent total reaction during the course of the experiment, therefore the relative reaction rates must be scaled depending on the proportion of total surface inclusion volume consumed. If it is assumed that the contribution to the H2 O consumption rate for inclusions versus surrounding uranium is directly proportional to the proportion of the available inclusion volume that has reacted, then the calculated contribution from inclusions reduces to 10% for a 50% consumption of surface inclusion volume and 5% for a 25% volume consumption. In conclusion, the reactions between uranium metal and water have only previously been considered as ‘uranium–water’, with omnipresent impurity phases either ignored due to their limited abundance or considered as un-reactive bystander species. In the current work the recorded ‘inclusion–water’ reactions appeared to occur more vigorously than concurrent ‘uranium–water’ reactions, in actively consuming water vapour. It is therefore likely that reaction results recorded for uranium metals of different purity are very likely to differ, providing a potentially signiﬁcant source of error. This may partly explain the conﬂicting data reported by different groups in the literature [1–7].

T.B. Scott et al. / Journal of Hazardous Materials 195 (2011) 115–123

5. Conclusions Examination of cast ␣-uranium surfaces after exposure to water vapour in an ESEM instrument at 19 mbar and 20 ◦ C indicated surface corrosion had occurred, with secondary growths determined to be UO3 ·H2 O (metaschoepite) forming at carbide inclusions across the surface. Over the period of a week the growths were observed to increase in size, consuming only the carbide particles and not the surrounding metal. In some cases complete decomposition of the surface carbides was observed. From the results of the current study it is apparent that that the carbide particles reacted more readily with the water vapour than the metal. Resultantly it is suggested that disparities between previous studies of the uranium–water reaction may be attributable to differential purities of uranium metal used by different research groups, with resultantly different populations of carbide particles. References [1] F. Weigel, Uranium, in: J.J. Katz, G.T. Seaborg, L.R. Morss (Eds.), The Chemistry of the Actinide Elements, Chapman & Hall, London, 1986, p. 245. [2] J.M. Haschke, Corrosion of uranium in air and water vapor: consequences for environmental dispersal, Journal of Alloys and Compounds 278 (1998) 149–160. [3] V.S. Yemel’yanov, A.L. Yevstyukhin, The Metallurgy of Nuclear Fuel, Pergamon Press, 1969. [4] M.M. Baker, L.N. Less, S. Orman, Uranium + water reaction. 1 Kinetics products and mechanism, Transactions of the Faraday Society 62 (1966) 2513–2524. [5] G.C. Allen, P.M. Tucker, R.A. Lewis, X-ray photoelectron-spectroscopy study of the initial oxidation of uranium metal in oxygen + water-vapor mixtures, Journal of the Chemical Society, Faraday Transactions 2 80 (1984) 991– 1000. [6] A.G. Ritchie, The kinetics and mechanism of the uranium–water vapour reaction – an evaluation of some published work, Journal of Nuclear Materials 120 (1984) 143–153.

123

[7] G.W. McGillivray, D.A. Geeson, R.C. Greenwood, Studies of the kinetics and mechanism of the oxidation of uranium by dry and moist air: a model for determining the oxidation rate over a wide range of temperatures and water vapour pressures, Journal of Nuclear Materials 208 (1994) 81–97. [8] B.R.T. Frost, The carbides of uranium, Journal of Nuclear Materials 10 (1963) 265–300. [9] P.E. Evans, T.J. Davies, Uranium nitrides, Journal of Nuclear Materials 10 (1963) 43–55. [10] T. Ishitani, H. Tsuboi, T. Yaguchi, H. Koike, Transmission electron microscope sample preparation using a focused ion beam, Journal of Electron Microscopy 43 (1994) 322–326. [11] R. Asuvathraman, S. Rajagopalan, K. Ananthasivan, C.K. Mathews, R.M. Mallya, Surface studies on uranium monocarbide using XPS and SIMS, Journal of Nuclear Materials 224 (1995) 25–30. [12] Y. Hori, T. Mukaibo, Kinetic studies of the reaction between uranium monocarbide and water vapour, Bulletin of the Chemical Society of Japan 40 (1967) 1878–1883. [13] A. Schürenkämper, Kinetic studies of the hydrolysis of uranium monocarbide in the temperature range 30–90 ◦ C, Journal of Inorganic and Nuclear Chemistry 32 (1970) 417–429. [14] R.M. Dell, V.J. Wheeler, N.J. Bridger, Hydrolysis of uranium mononitride, Transactions of the Faraday Society 63 (1967) 1286–1294. [15] G.A.R. Rao, S.K. Mukerjee, V.N. Vaidya, V. Venugopal, D.D. Sood, Oxidation and Hydrolysis kinetic-studies on UN, Journal of Nuclear Materials 185 (1991) 231–241. [16] L.J. Colby Jr., Kinetics of the reaction of uranium monocarbide with water, Journal of the Less Common Metals 10 (1966) 425–431. [17] C.P. Kempter, Hydrolysis properties of uranium monocarbide and dicarbide, Journal of the Less Common Metals 4 (1962) 419–425. [18] M.J. Bradley, L.M. Ferris, Hydrolysis of uranium carbides between 25 and 100◦ . III. Uranium sesquicarbide and mixtures of the sesquicarbide with monocarbide or dicarbide, Inorganic Chemistry 3 (1964) 730–734. [19] M.J. Bradley, L.M. Ferris, Hydrolysis of uranium carbides between 25 and 100◦ . II. Uranium dicarbide, uranium metal-monocarbide mixtures, and uranium monocarbide-dicarbide mixtures, Inorganic Chemistry 3 (1964) 189–195. [20] M.I. Ermolaev, G.V. Tishchenko, Reaction of uranium monocarbide with acid and alkaline solutions, Izvestija vysˇsich uˇcebnych zavedenij, Khim. Khim. Tekhnol. 16 (11) (1973) 1631–1633. [21] B. Hájek, P. Karen, V. Broˇzek, Hydrolysis of uranium monocarbide, Collection of Czechoslovak Chemical Communications 49 (1984) 793–804. [22] M.J. Bradley, L.M. Ferris, Hydrolysis of uranium carbides between 25 and 100◦ . I. Uranium monocarbide, Inorganic Chemistry 1 (1962) 683–687.

Journal of Hazardous Materials 195 (2011) 124–131

Contents lists available at ScienceDirect

Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat

Adsorption of volatile organic compounds by metal–organic frameworks MIL-101: Inﬂuence of molecular size and shape Kun Yang a,b,∗ , Qian Sun a,b , Feng Xue a,b , Daohui Lin a,b a b

Department of Environmental Science, Zhejiang University, Hangzhou 310058, China Zhejiang Provincial Key Laboratory of Organic Pollution Process and Control, Hangzhou, Zhejiang 310058, China

a r t i c l e

i n f o

Article history: Received 4 May 2011 Received in revised form 5 August 2011 Accepted 5 August 2011 Available online 11 August 2011 Keywords: Metal–organic frameworks MIL-101 VOCs Adsorption

a b s t r a c t Adsorption of gaseous volatile organic compounds (VOCs) on metal–organic frameworks MIL-101, a novel porous adsorbent with extremely large Langmuir surface area of 5870 m2 /g and pore volume of 1.85 cm3 /g, and the inﬂuence of VOC molecular size and shape on adsorption were investigated in this study. We observed that MIL-101 is a potential superior adsorbent for the sorptive removal of VOCs including polar acetone and nonpolar benzene, toluene, ethylbeznene, and xylenes. MIL-101 is of higher adsorption capacities for all selected VOCs than zeolite, activated carbon and other reported adsorbents. Adsorption of VOCs on MIL-101 is captured by a pore ﬁlling mechanism, showing the size and shape selectivity of VOC molecules. These prove to be a negative linear relationship between the volume adsorption capacities of VOCs and their molecular cross-sectional area values. Most VOC molecules, such as acetone, benzene, toluene, ethylbenzene and p-xylene, enter into MIL-101 pores with the planes having the minimum diameters. However, m-xylene and o-xylene may ﬁll into the pores with the planes having the maximum diameters because of the preferred interaction of MIL-101 with the two methyl groups of adsorbate molecules. © 2011 Elsevier B.V. All rights reserved.

1. Introduction Volatile organic compounds (VOCs), including BTEXs (i.e., benzene, toluene, ethylbenzene, and xylenes), aldehydes, ketones, and chlorinated hydrocarbons, are among the most common contaminants in indoor/outdoor air with worldwide concerns [1]. They are largely emitted by the chemical process industries (e.g., thinner, degreasers, cleaners, lubricants, and liquid fuels) and can pose great threat to human being, including sensory irritation symptoms [2], severe disorder of respiratory system or mucous membrane of vision organ [3] and even the cancerization of natural cells [2,4,5]. VOCs in ambient environment may also bring on other serious environmental and health risks such as photochemistry smog [6]. Therefore, a number of add-on-control techniques, including the destruction techniques (e.g., bioﬁltration, thermal oxidation, and catalytic oxidation) and the recovery techniques (e.g., absorption, adsorption, condensation, and membrane separation), have been developed to control VOCs emissions and to remove VOCs from contaminated air [7]. Among these techniques, adsorption using porous materials as adsorbents is a well-established and effective technique for the removal and recovery of VOCs from air. Porous

∗ Corresponding author at: Department of Environmental Science, Zhejiang University, Hangzhou 310058, China. Tel.: +86 571 88982589; fax: +86 571 88982590. E-mail address: [email protected] (K. Yang). 0304-3894/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.jhazmat.2011.08.020

materials, having large surface area and pore volume, including zeolites [8–10], resins [11], activated carbon together with their derivatives [12–14], are the most frequently utilized adsorbents with high adsorption capacity for VOCs. Efforts are still needed to develop new porous materials with larger surface area and pore volume and higher adsorption capacity for VOC adsorption. Metal–organic frameworks (MOFs), a new class of hybrid porous solids, are potentially a type of prominent porous adsorbents for VOCs because of the extremely large pore volume and surface area (usually >3000 m2 /g) [15]. MOFs are crystalline hybrid porous solids with ordered three-dimensional network frameworks via strong metal–ligand bonds between metal cations and organic linkers [15,16]. Since their discovery, MOFs received lots of signiﬁcant attentions in their potential applications in gas storage, separation, heterogeneous catalysis, and sensing [15–17]. In the great family of MOFs, MIL-101 with the formula of Cr3 F(H2 O)2 OE(O2 C)–C6 H4 –(CO2 )3 ·nH2 O (n is ∼25), which is an automated assembly product of chromium nitrate nonahydrates and terephthalic acid, and ﬁrstly reported by Ferey and his coworkers in 2005 [18], is a porous material possessing the largest surface area (5900 ± 300 m2 /g) and pore volume (≈2.0 cm3 /g) among the reported porous materials, which make MIL-101 a potential candidate as adsorbent on gas adsorption [18]. Adsorption researches on MIL-101 initially focused on characteristics of H2 adsorption [19,20]. Later, the adsorption characteristics of other gases, such as CO2 [21,22], CH4 [21,22],

K. Yang et al. / Journal of Hazardous Materials 195 (2011) 124–131

H2 S [23], SF6 [22], C3 H8 [22], benzene [24,25], toluene [26], nalkanes [27], and n-butane [28] on MIL-101 were examined. All of these studies indicated that MIL-101 was a potential superior adsorbent for gas adsorption, especially for organic vapors. For example, Chowdhury and co-workers [22] observed that MIL-101 had higher adsorption capacity of C3 H8 than that of other gases (i.e., CO2 , CH4 , and SF6 ). MIL-101 also showed higher adsorption capacity of benzene (i.e., 16.7 mmol/g) than other frequently used adsorbents including SBA-15, HZSM-5 and activated carbon [24]. In addition, MIL-101 was reported to be a very promising candidate for the application of high-resolution capillary gas chromatograph (GC) as the stationary phase in separation of xylene isomers and ethylbenzene due to its superior adsorption characteristics [29]. In addition to the surface characteristics of porous materials, molecular properties, such as the size and shape of VOCs, play a commonly important role in their adsorption on porous materials [8–14,30]. However, as far as we know, very limited studies were conducted to examine the inﬂuence of molecular size and shape of VOCs on their adsorption by MIL-101. The relationship between molecular properties of VOCs and their adsorption on MIL-101 has not been established yet [27]. So it became the main objective of this study. Moreover, only the nonpolar VOCs (i.e., C3 H8 , benzene, toluene, n-alkanes, xylene isomers, ethylbenzene and n-butane) were examined for their adsorption by MIL-101 in previous studies [22,24–30]. Polar VOCs (such as acetone, methanol, ethanol, tetrachloroethane, methyl chloride, various chlorohydrocarbons and perﬂuorocarbons) were also environmentally concerned pollutants in air due to their high toxicity and volatility [7]. However, adsorption characteristics of these polar VOCs by MIL-101 were still not examined. Therefore, in this study, polar acetone and nonpolar BTEXs with various sizes and shapes were selected as adsorbates to examine the adsorption performance of MIL-101 towards polar and nonpolar VOCs as well as the inﬂuences of VOC molecular properties. We also hope to establish the relationships between molecule properties of VOCs and their adsorption on MIL-101, and to explore the underlying adsorption mechanisms. 2. Materials and methods 2.1. Chemicals Chromium nitrate nonahydrates (99+%), terephthalic acid (TPA) (99+%) and ﬂuorhydric acid (HF) (40+%) were purchased from Sinopharm Chemical Reagent Co., Ltd. (China), Acros organic (USA) and Juhua Reagent Co., Ltd. (China), respectively; acetone (98+%), benzene (98+%), toluene (98+%) and ethylbenzene (98+%) were purchased from Hangzhou Chemical Reagent Co., Ltd (China); o-xylene (98+%), p-xylene (98+%) and m-xylene (98+%) were purchased from Sinopharm Chemical Reagent Co., Ltd. (China). These chemicals were used without any further puriﬁcation. Selected properties of the VOCs [30,31] are listed in Table 1.

125

2.2. Synthesis of MIL-101 MIL-101, the highly crystallized green powder of the chromium terephthalate, was synthesized according to the method described in the literature [18]. Brieﬂy, 4.0 g Cr (NO3 )3 (0.01 mol), 1.64 g TPA (0.01 mol), 125 ␮L HF and 70 mL ultrapure water were transformed into a 100 mL Teﬂon-lined stainless steel autoclave, sealed, heated up to 220 ◦ C for 8 h, and then were slowly cooled down to atmospheric temperature. After that, the green suspension of MIL-101 was ﬁltered by using a stainless steel meshwork (with a diameter of 0.061 mm) to remove the re-crystallized needle-shaped colorless TPA which retained on the meshwork and the MIL-101 suspension passed through the meshwork. The ﬁltrated MIL-101 suspension was sequently centrifuged at 3500 × g (for 15 min to collect the ﬁrst precipitates of MIL-101) and 8000 × g (for 15 min to collect the second precipitates of MIL-101). And then, the second precipitates of MIL-101 were washed several times with ultrapure water, and dried at 70 ◦ C for 24 h in a hot air oven for the usage of adsorption experiments. 2.3. Characterization of synthesized MIL-101 Adsorption isotherm of nitrogen on MIL-101 was obtained at 77 K by an Autosorb-1MP-VP apparatus (Quantachrome Corp., USA). The MIL-101 was evacuated overnight at 105 ◦ C to remove the water molecules in MIL-101 prior to the nitrogen adsorption experiment. Speciﬁc surface area values of MIL-101were calculated by the Brunauer–Emmett–Teller (BET) method and the Langmuir method using adsorption data of nitrogen adsorption isotherm. Pore size of MIL-101 was calculated by the Quenched Solid State Density Functional Theory (QSDFT) method. Powder Xray diffraction (XRD) pattern of MIL-101 was recorded on a XPert diffractometer (Panalytical Corp., Netherlands) which was operated at 40 kV for Cu K␣ ( = 0.1543 nm) radiation from 3◦ to 70◦ (2 angle range) with a scan step size of 0.02◦ . Transmission electron microscope (TEM) images were taken at 80 kV by a JEM-1230 (JEOL Corp., Japan). The thermal stability of MIL-101 was investigated by thermal gravimetric analysis (TGA) with SDT Q600 (AT Corp., USA) by using approximate 7.3 mg sample from 35 ◦ C to 605 ◦ C with a ramping rate of 2.0 ◦ C/min in air atmosphere. Fourier transform infrared (FT-IR) spectrum was determined using Bruker-vector-22 (German) with a range of 500–4000 cm−1 . MIL-101 sample for FTIR analysis was pretreated by grinding power of MIL-101 together with KBr in an agate mortar and then by pressing them into ﬂakes by a tablet machine. 2.4. Adsorption measurements The vapor-phase adsorptions of BETXs and acetone were carried out at 25 ◦ C on a volumetric adsorption apparatus, i.e., Autosorb1MP-VP (Quantachrome Corp., USA). This apparatus was modiﬁed

Table 1 Selected properties of VOC molecules. VOCs

a (g/mL, 25 ◦ C)

MWb (g/mol)

c (nm2 )

SPd (kPa, 25 ◦ C)

Xe

Ye

Ze

Acetone Benzene Toluene Ethylbenzene m-Xylene o-Xylene p-Xylene

0.786 0.876 0.865 0.901 0.877 0.858 0.861

58 78 92 106 106 106 106

0.270 0.305 0.344 0.368 0.379 0.375 0.380

30.414 12.573 3.776 1.320 1.117 0.876 0.725

6.600 6.628 6.625 6.625 8.994 7.269 6.618

4.129 3.277 4.012 5.285 3.949 3.834 3.810

5.233 7.337 8.252 9.361 7.315 7.826 9.146

a b c d e

Density. Molecule weight. Molecule cross-sectional area, data cited from Ref. [30]. Saturation pressure. X, Y and Z are the molecular width, thickness and length, respectively, data cited from Ref. [31].

126

K. Yang et al. / Journal of Hazardous Materials 195 (2011) 124–131

with an additional organic vapor generation system and a heating system. The VOC vapors were generated by the vapor generation system at 50 ◦ C. The constant adsorption temperature was achieved by putting sample cell into water bath in Dewar ﬂask. The MIL101 in sample cell was evacuated at 105 ◦ C overnight to remove the water molecules in MIL-101 prior to the adsorption experiments. Adsorption equilibrium was assumed to be reached when the vapor pressure drop in sample cell was less than 0.0001 atm (=8 Pa) within 3 min. The vapor pressure in sample cell was measured by using three transducers (i.e., 1 Torr transducer, 10 Torr transducer and 1000 Torr transducer). The accuracy of 1, 10 and 1000 Torr transducers are ±0.15% of reading, ±0.15% of reading and ±0.1% of full scale, respectively. The linearity of 1, 10 and 1000 Torr transducers are ±0.1% of reading, ±0.1% of reading and ±0.05% of full scale, respectively. The resolution of each transducer

is 0.000025%. The adsorbed amount of VOCs on MIL-101 for each pressure at adsorption equilibrium was calculated with the ideal gas law (the fundamental principle of the volumetric adsorption method as well as the volumetric adsorption apparatus) by the difference between the pressure at adsorption equilibrium and the initial vapor pressure in sample cell. 3. Results and discussion 3.1. Characteristics of MIL-101 Nitrogen adsorption isotherm of the dehydrated MIL-101 is shown in Fig. 1A. The speciﬁc surface area of MIL-101, calculated by the BET method and the Langmuir method, is about 3980 and 5870 m2 /g, respectively. These values are close to the reported val-

Fig. 1. Nitrogen adsorption isotherm (A), pore size distribution (B), XRD pattern (C), TEM image (D), FT-IR spectrum (E) and DSC/TGA curve (F) of MIL-101.

K. Yang et al. / Journal of Hazardous Materials 195 (2011) 124–131

127

1600 1200 1000

Q (mg/g)

Q (mg/g)

1200

800 benzene acetone toluene

400

800 600

p-xylene o-xylene m-xylene ethylbenzene

400 200

0

0 0.0

0.2

0.4

0.6

0.8

1.0

Relative pressure (P/P0)

0.0

0.2

0.4

0.6

0.8

Relative pressure (P/P0)

Fig. 2. Isotherms of acetone, benzene, toluene, ethylbenzene and xylenes on MIL-101 at 25 ◦ C.

ues for MIL-101 [18]. The total pore volume of MIL-101 is estimated to be 1.85 cm3 /g at a relative pressure of P/P0 = 0.55. Pore sizes of ˚ and conMIL-101, shown in Fig. 1B, mainly range from 8.5 to 26 A, ˚ which is similar to ﬁrm two domain pore sizes (i.e., 12 and 18 A), the pore sizes estimated from the crystal structure [18]. The XRD pattern of the sample obtained (Fig. 1C) is similar to the simulated pattern of MIL-101 reported previously [32]. The TEM image shown in Fig. 1D conﬁrms that the synthesized MIL-101 is a highly crystallized regular octahedron with a perfect cubic symmetry. The FT-IR spectrum of MIL-101 (Fig. 1E) is similar to the previous results [24,29,33]. The strong bands, at 1650 and 1430 cm−1 , can be assigned to the vibrational stretching frequencies of the framework (O–C–O), conﬁrming the presence of dicarboxylate linker in the MIL-101 framework [24,33]. The bands at 1055 and 750 cm−1 can be assigned to the vibrations of benzene rings [33]. The bands near 600 cm−1 are most likely to ascribe to in-plane and out-ofplane bending modes of COO-groups [33]. The TGA proﬁle shows that MIL-101 is stable up to 220 ◦ C (Fig. 1F). A total of 71% weight loss of MIL-101, occurred between 220 and 350 ◦ C, may result from the framework decomposition of the organic moieties [22,24]. 3.2. Adsorption characteristics of selected VOCs on MIL-101 Isotherms of acetone, benzene and toluene exhibit a sharp increase of adsorption amount at the relative pressure (P/P0 ) less than 0.1 (Fig. 2A). However, there are almost no adsorption of ethylbenzene and xylenes by MIL-101 at P/P0 < 0.1 (Fig. 2B). A sharp increase of adsorption amount of ethylbenzene and xylenes on MIL101 occurred approximately at P/P0 from 0.1 to 0.2. The negligible adsorption of ethylbenzene and xylenes on MIL-101 at P/P0 < 0.1 (Fig. 2B) should not be a fact. It may be an artiﬁcial result derived from the possibly slow diffusion of ethylbenzene and xylenes into pores of MIL-101 since adsorption equilibrium was assumed to be reached when the pressure drop in sample cell was less than 0.0001 atm (=8 Pa) within 3 min. The low absolute pressure of ethylbenzene and xylenes at P/P0 < 0.1, which is lower than 0.001 atm (=100 Pa) calculated from their P0 (Table 1), and the big molecular size, which is identiﬁed by the large molecule cross-sectional area (Table 1), may be responsible for the slow diffusion of ethylbenzene and xylenes into the pores of MIL-101 [34,35]. Slow diffusion could make the pressure drop of ethylbenzene and xylenes in the sample cell within 3 min undetectable. In addition, the rapid diffusion of acetone, benzene and toluene into the pores of MIL-101, identiﬁed by their signiﬁcant adsorption at P/P0 < 0.1 (Fig. 2A), could be attributed to their smaller molecular sizes as compared with the molecular sizes of ethylbenzene and xylenes [24,34,35]. Another evidence of the slow diffusion and the un-equilibrium of adsorption of ethylbenzene and xylenes is that their desorption isotherms are a

litter higher than their adsorption isotherms at P/P0 > 0.1, while the desorption isotherm of acetone with smaller size is almost the same with its adsorption isotherm (Fig. S1 in Supporting Information). Isotherms of acetone, benzene, toluene, ethylbenzene and xylenes show that adsorption of these VOCs on MIL-101 reached a plateau approximately at P/P0 > 0.2 (Fig. 2). The plateau values (i.e., saturated adsorption capacity) are listed in Table 2. MIL-101 shows the highest adsorption capacity for these VOCs as compared with other commonly utilized adsorbents such as zeolites, resins, activated carbon together with their derivatives [11,24,30,36–44] (Table 2), indicating that MIL-101 is a potential superior adsorbent for the sorptive removal of unwanted VOCs such as polar acetone and apolar BTEXs from contaminated air. For example, the saturated adsorption capacity of benzene on MIL-101 in this study was estimated to be 1291 ± 77 mg/g (i.e., 16.4 mmol/g) at P/P0 = 0.55 ± 0.05, which is close to the reported value (16.7 mmol/g) of benzene on MIL-101 at P/P0 = 0.5 in a previous study [24] but higher than the saturated adsorption capacity of benzene on HY zeolite (3.3 mmol/g), whose speciﬁc surface area is one of the largest among the aluminosilicate zeolites [24], and a pitch-based activated carbon (12.4 mmol/g, the maximum value was reported in the literatures for benzene on adsorbents except for MIL-101) [24]. 3.3. Inﬂuence of VOC structure on their adsorption into MIL-101 and the underlying adsorption mechanisms A negative linear relationship between the volume adsorption capacity (Va ) of selected VOCs at P/P0 = 0.55 ± 0.05 (Table 3) and their molecular cross-sectional area (i.e., the total area of the orthographic projection of a molecule where the molecular geometry is accepted as a sphere) in Table 1 was established (Fig. 3). This phenomenon implies that adsorption of selected VOCs into MIL-101 could be attributed to the continuous micropore ﬁlling, an adsorption mechanism suggested by previous studies widely for adsorption of various vapors into porous materials including zeolites, resins, activated carbons and MOFs [9–14,24,25,45]. The dependence of the volume adsorption capacity of VOCs (Table 3) on their molecular cross-sectional area (Table 1 and Fig. 3) but not their minimum dimension (Table 1) could be attributed to the pore type (i.e., cylindrical pore) of MIL-101 [18,31]. The decrease of the volume adsorption capacity of VOCs with the increase of VOC molecular cross-sectional area is a result of that VOC molecules cannot enter into a fraction of MIL-101 cylindrical micropores whose sizes are smaller than the VOC molecular size [31]. Therefore, VOC molecules with larger size presented lower volume adsorption capacity. Another evidence of the micropore ﬁlling mechanism for adsorption of VOCs into MIL-101 is that the ﬁrst derivatives of

128

K. Yang et al. / Journal of Hazardous Materials 195 (2011) 124–131

Table 2 Adsorption capacities of MIL-101 and other adsorbents for selected VOCs. Adsorbates

Acetone

Benzene

Toluene

Ethylbenzene

m-Xylene

o-Xylene

p-Xylene

Asurf (m2 /g)

Adsorbents

PCH ACFC CDAC Y-Zeolite MIL-101 MOF-5 IRMOF-3 MOF-74 MOF-177 MOF-199 IRMOF-62 SBA-15 HZSM-5 zeolite Activated carbon NDA-201 resin ACC-963 ACFC MIL-101 MIL-101 MIL-101 PCH ACC-963 Y-Zeolite MIL-101 Zn(BDC)-(Dabco)0.5 MOF-5 MOF-monoclinic MIL-47 PCH MIL-101 Zn(BDC)-(Dabco)0.5 MOF-5 MOF-monoclinic MIL-47 MIL-96 PCH MIL-101 Zn(BDC)-(Dabco)0.5 MOF-5 MOF-monoclinic MIL-47 PCH MIL-101 Zn(BDC)-(Dabco)0.5 MOF-5 MOF-monoclinic MIL-47 MIL-96 PCH MIL-101

740 1604 965 704 3980 2205 1568 632 3875 1264 1814 805 550 1600 855.6 1705 1604 3900 3054 3980 740 1705 704 3980 1450 773 225 930 740 3980 1450 773 225 930 532 740 3980 1450 773 225 930 740 3980 1450 773 225 930 532 740 3980

Qa

Va (mL/g)

4.8 mmol/g 595 mg/g 381 mg/g 7.2 mmol/g 1291 ± 71 mg/g 2 mg/g 56 mg/g 96 mg/g 1 mg/g 176 mg/g 109 mg/g 3.0 mmol/g 1.9 mmol/g 8.0 mmol/g 5.2 mmol/g 7.8 mmol/g 634 mg/g 16.7 mmol/g 15.5 mmol/g 1291 ± 77 mg/g 2.9 mmol/g 6.1 mmol/g 1.625 mmol/g 1096 ± 142 mg/g 347 mg/g 99 mg/g 5 mg/g 35 wt% 4.25 mmol/g 1105 ± 116 mg/g 345 mg/g 151 mg/g 4 mg/g 37 wt% 0.81 mL/g 3.5 mmol/g 727 ± 88 mg/g 338 mg/g 125 mg/g 4 mg/g 36 wt% 2.6 mmol/g 758 ± 176 mg/g 342 mg/g 138 mg/g 13 mg/g 40 wt% 0.105 mL/g 3.4 mmol/g 1067 ± 83 mg/g

Conditions

0.354 0.752 0.481 0.528 1.645 ± 0.090 0.002 0.064 0.109 0.001 0.200 0.124 0.269 0.170 0.717 0.466 0.668 0.719 1.495 1.388 1.477 ± 0.081 0.308 0.645 0.172 1.270 ± 0.164 0.527 0.146 0.007 0.526 0.434 1.228 ± 0.128 0.511 0.217 0.006 0.542 0.814 0.431 0.846 ± 0.100 0.505 0.181 0.006 0.532 0.314 0.866 ± 0.205 0.506 0.198 0.019 0.586 0.105 0.420 1.246 ± 0.096

Refs.

Ce

T

0.98P/P0 0.8P/P0 0.8P/P0 0.17 atm 0.55 ± 0.05P/P0 440 ppm 440 ppm 440 ppm 440 ppm 440 ppm 440 ppm 0.5P/P0 0.5P/P0 0.5P/P0 10 kPa 0.80P/P0 0.8P/P0 0.5P/P0 55 mbar 0.55 ± 0.05P/P0 0.95P/P0 0.83P/P0 0.022 atm 0.55 ± 0.05P/P0 0.1 bar 2.8 kPa 1.3 kPa 0.035 bar 0.98P/P0 0.55 ± 0.05P/P0 0.1 bar 3.0 kPa 1.3 kPa 0.03 bar 0.87P/P0 0.97P/P0 0.55 ± 0.05P/P0 0.1 bar 3.4 kPa 1.3 kPa 0.028 bar 0.85P/P0 0.55 ± 0.05P/P0 0.1 bar 2.5 kPa 1.2 kPa 0.035 bar 0.78P/P0 0.98P/P0 0.55 ± 0.05P/P0

298 K 20 ◦ C 20 ◦ C 20 ◦ C 25 ◦ C 25 ◦ C 25 ◦ C 25 ◦ C 25 ◦ C 25 ◦ C 25 ◦ C 30 ◦ C 30 ◦ C 30 ◦ C 303 K 273 K 20 ◦ C 30 ◦ C 288 K 25 ◦ C 298 K 273 K 20 ◦ C 25 ◦ C 120 ◦ C 150 ◦ C 150 ◦ C 130 ◦ C 298 K 25 ◦ C 120 ◦ C 150 ◦ C 150 ◦ C 130 ◦ C 30 ◦ C 298 K 25 ◦ C 120 ◦ C 150 ◦ C 150 ◦ C 130 ◦ C 298 K 25 ◦ C 120 ◦ C 150 ◦ C 150 ◦ C 130 ◦ C 30 ◦ C 298 K 25 ◦ C

[30] [36] [36] [37] This work [38] [38] [38] [38] [38] [38] [24] [24] [24] [11] [39] [36] [24] [25] This work [30] [39] [37] This work [40,41] [42] [42] [42,43] [30] This work [40,41] [42] [42] [42,43] [44] [30] This work [40,41] [42] [42] [42,43] [30] This work [40,41] [42] [42] [42,43] [44] [30] This work

Asurf is the speciﬁc surface area calculated using the BET method; Qa is the reported data of saturation adsorbed amount of VOCs; Ce is the equilibrium concentration (ppm), relative pressure (P/P0 ) or absolute pressure (kPa, atm, mbar or bar) at which the reported Qa values obtained; T is the temperature at which the reported Qa values obtained; Va is adsorbed capacity in volume, calculated from Qa and chemical density (Table 1), by the following equation: Va = Qa /.

Table 3 Molecule dimension and the possible and determined volume adsorption capacities of selected VOCs into MIL-101. Compounds

DXY (nm)

V0.55 − VXY (mL/g)

DYZ (nm)

V0.55 − VYZ (mL/g)

DXZ (nm)

V0.55 − VXZ (mL/g)

Va (mL/g)

Acetone Benzene Toluene Ethylbenzene m-Xylene o-Xylene p-Xylene

0.778 0.739 0.775 0.847 0.982 0.822 0.746

1.392 1.393 1.291 1.288 1.084 1.260 1.357

0.666 0.804 0.918 1.075 0.831 0.871 0.991

1.510 1.291 1.138 0.997 1.228 1.228 1.075

0.842 0.989 1.058 1.147 1.159 1.068 1.129

1.228 1.078 0.918 0.928 0.918 1.003 0.945

1.645 1.477 1.270 1.228 0.846 0.866 1.246

± ± ± ± ± ± ±

0.090 0.081 0.164 0.128 0.100 0.205 0.096

V0.55 is the adsorbed volume of N2 into MIL-101 at P/P0 = 0.55; DXY , DYZ , DXZ are diameters of circumcircles of the XY, YZ and XZ plane of molecules, calculated by the following equations: DXY = (X2 + Y2 )0.5 , DYZ = (Y2 + Z2 )0.5 , DXZ = (X2 + Z2 )0.5 , where X, Y and Z are the molecular width, thickness and length (Table 1), respectively; VXY , VYX and VXZ are the adsorbed volumes of N2 into MIL-101 pores whose sizes are smaller than DXY , DYZ and DXZ , respectively; VXY , VYX and VXZ are obtained directly from the relationship between pore diameter and accumulative pore volume of N2 on MIL-101(please see the detailed illustration in Supporting Information); Va is the adsorbed capacity in volume (Table 2).

K. Yang et al. / Journal of Hazardous Materials 195 (2011) 124–131

2.0 acetone

in xylene molecules and the consequent difference in molecule shape [47]. For VOC molecules, their shape is commonly accepted as rectangular solid, having three important dimensions, i.e., the width (X), thickness (Y) and length (Z) (Table 1). The rectangular VOC molecule is made of six rectangle planes. The six rectangle planes come in three congruent pairs, i.e., XY, YZ and XZ (Table 1), respectively. Therefore, they could possibly enter into pores of MIL101 with one of the three congruent rectangle pairs. Adsorption capacity of VOCs by ﬁlling into pores of MIL-101 is dependent on the length of diagonal line of the molecular rectangle plane (i.e., the diameter of circumcircle of the plane including DXY , DYZ and DXZ listed in Table 3) by which VOC molecules entered into MIL-101 pores since VOC molecules cannot enter into pores whose diameters smaller than the diameter of VOC molecule plane. Mostly, VOC molecules, such as acetone, benzene, toluene, ethylbenzene and p-xylene, enter into MIL-101 pores with the plane having the minimum diameter (i.e., YZ plane of acetone and XY plane of benzene, toluene, ethylbenzene and p-xylene) [48], showing that V0.55 − VYZ value of acetone and V0.55 − VXY values of toluene, benzene, ethylbenzene and p-xylene are respectively equal to their Va values (Table 3). However, Va values of o-xylene and m-xylene are respectively equal to their V0.55 − VXZ values, implying that they ﬁll into MIL-101 pores with the XZ plane (i.e., the plane having maximum diameter) (Table 3). A possible reason is that methyl groups of xylene molecules, prior to their benzene ring, interact with active sites of MIL-101 [47], and thus, o-xylene and m-xylene could not ﬁll into the pores with diameters smaller than their DXZ (i.e., the diameter of the maximum plane of o-xylene and m-xylene which depends on the distance between the two methyl groups) as shown in Fig. 5. However, the two methyl groups in p-xylene, linking benzene ring with an angle of about 180◦ , form a line-type structure, and which allow p-xylene molecules, as similar to ethylbenzene

p-xylene

1.6

ethylbenzene

Va (ml/g)

benzene 1.2

toluene

0.8

o-xylene m-xylene

0.4 0.0 0.25

0.30

0.35

129

0.40

Molecular cross-sectional area (nm 2) Fig. 3. Relationship between molecular cross-sectional area and volume adsorption capacity (Va ) of seven selected VOCs.

adsorption amount of benzene and toluene, as an example, present two peaks (Fig. 4). The two peaks ascribe to the presence of the two types of microporous windows existed in the skeleton of MIL-101 [46], which can be detected in the bimodal pore size distribution shown in Fig. 1B. The high and low peaks in Fig. 4 could be resulted from the ﬁlling of VOC molecules into 12 and 18 A˚ micropores of MIL-101 (Fig. 1B), respectively. In addition to molecule size, shape of VOC molecules may also affect their adsorption into MIL-101. For example, m-xylene and o-xylene have almost equal molecule cross-sectional area to pxylene, but lower adsorption capacity than p-xylene (Fig. 4). This could be attributed to the position difference of methyl side chains

60,000

40,000

40,000

dQ

60,000

toluene

benzene 20,000

20,000

0

0.0

0

0.2

0.4

0.6

0.8

1.0

0.0

0.2

0.4

0.6

0.8

1.0

Relative pressure (P/P0 ) Fig. 4. First derivative (dQ) of adsorption amount of toluene and benzene into MIL-101 versus the relative pressure (P/P0 ).

Fig. 5. Scheme of ethylbenzene, p-xylene, o-xylene and m-xylene entering into MIL-101 pores.

130

K. Yang et al. / Journal of Hazardous Materials 195 (2011) 124–131

molecules, to enter into MIL-101 pores with diameters larger than DXY as shown in Fig. 5. Therefore, volume adsorption capacity of pxylene in MIL-101 was higher than that of m-xylene and o-xylene (Fig. 3). 4. Conclusions The results presented that the porous MIL-101 material is a potential superior adsorbent for sorptive removal of unwanted VOCs from contaminated air due to its large surface area and pore volume. Adsorption of VOCs by MIL-101 is captured by a pore ﬁlling mechanism, showing the size and shape selectivity of VOC molecules. These prove to be a negative linear relationship between the volume adsorption capacities of selected VOCs and their molecular cross-sectional area values. Mostly, VOC molecules, such as acetone, benzene, toluene, ethylbenzene and p-xylene, entered into MIL-101 pores with the planes having the minimum diameters. However, m-xylene and o-xylene entered into the MIL-101 pores with the molecular planes having the maximum diameters (i.e., DXZ , which depends on the distance between the two methyl groups) because of the preferred interaction of MIL-101 with the two methyl groups of VOC molecules. In addition, the size and shape selectivity of VOC molecules into MIL-101 pores should be mentioned in the application of MIL-101 as an adsorbent for the removal and recovery of VOCs from air. Acknowledgements This work was in part supported by the National High Technology Research and Development Program of China (Nos. 2010AA064902 and 2007AA061402), the Science and Technology Department of Zhejiang Province (2006R10022) and the Key Innovation Team for Science and Technology of Zhejiang Province (2009R50047). Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.jhazmat.2011.08.020. References [1] WHO, Indoor air quality: organic pollutants, Euro Reports and Studies No. 111, 1989. [2] P. Wolkoff, C.K. Wilkins, P.A. Clausen, G.D. Nielsen, Organic compounds in ofﬁce environments—sensory irritation, odor, measurements and the role of reactive chemistry, Indoor Air 16 (2006) 7–19. [3] J.A. Bernstein, N. Alexis, H. Bacchus, I.L. Bernstein, P. Fritz, E. Horner, N. Li, S. Mason, A. Nel, J. Oullette, K. Reijula, T. Reponen, J. Seltzer, A. Smith, S.M. Tarlo, The health effects of nonindustrial indoor air pollution, J. Allergy Clin. Immunol. 121 (2008) 585–591. [4] H. Geiger, K.H. Becker, P. Wiesen, Effect of gasoline formulation on the formation of photosmog: a box model study, J. Air Waste Manage. Assoc. 53 (2003) 425–433. [5] J.E. Cometto-Muniz, W.S. Cain, M.H. Abraham, Detection of single and mixed VOCs by smell and by sensory irritation, Indoor Air 14 (2004) 108–117. [6] A.E. Pouli, D.G. Hatzinikolaou, C. Piperi, A. Stavridou, M.C. Psallidopoulos, J.C. Stavrides, The cytotoxic effect of volatile organic compounds of the gas phase of cigarette smoke on lung epithelial cells, Free Radic. Biol. Med. 34 (2003) 345–355. [7] F.I. Khan, A.K. Ghoshal, Removal of volatile organic compounds from polluted air, J. Loss Prevent. Process Ind. 13 (2000) 527–545. [8] V.R. Choudhary, K. Mantri, Adsorption of aromatic hydrocarbons on highly siliceous MCM-41, Langmuir 16 (2000) 7031–7037. [9] J. Pires, A. Carvalho, M.B. de Carvalho, Adsorption of volatile organic compounds in Y zeolites and pillared clays, Micropor. Mesopor. Mater. 43 (2001) 277–287. [10] H.F. Cheng, M. Reinhard, Sorption of trichloroethylene in hydrophobic micropores of dealuminated Y zeolites and natural minerals, Environ. Sci. Technol. 40 (2006) 7694–7701. [11] P. Liu, C. Long, Q.F. Li, H.M. Qian, A.M. Li, Q.X. Zhang, Adsorption of trichloroethylene and benzene vapors onto hypercrosslinked polymeric resin, J. Hazard. Mater. 166 (2009) 46–51.

[12] D. Das, V. Gaur, N. Verma, Removal of volatile organic compound by activated carbon ﬁber, Carbon 42 (2004) 2949–2962. [13] M. Lordgooei, M.J. Rood, M. Rostam-Abadi, Modeling effective diffusivity of volatile organic compounds in activated carbon ﬁber, Environ. Sci. Technol. 35 (2001) 613–619. [14] A. Aizpuru, L. Malhautier, J.C. Roux, J.L. Fanlo, Bioﬁltration of a mixture of volatile organic compounds on granular activated carbon, Biotechnol. Bioeng. 83 (2003) 479–488. [15] S.T. Meek, J.A. Greathouse, M.D. Allendorf, Metal–organic frameworks: a rapidly growing class of versatile nanoporous materials, Adv. Mater. 23 (2011) 249–267. [16] J.L.C. Rowsell, O.M. Yaghi, Metal–organic frameworks: a new class of porous materials, Micropor. Mesopor. Mater. 73 (2004) 3–14. [17] M.B. Sander, L. Pan, X.Y. Huang, J. Li, M. Smith, E. Bittner, B. Bockrath, Microporous metal organic materials: promising candidates as sorbents for hydrogen storage, J. Am. Chem. Soc. 130 (2004) 8–1309. [18] G. Ferey, C. Mellot-Draznieks, C. Serre, F. Millange, J. Dutour, S. Surble, I. Margiolaki, A chromium terephthalate-based solid with unusually large pore volumes and surface area, Science 309 (2005) 2040–2042. [19] M. Latroche, S. Surble, C. Serre, C. Mellot-Draznieks, P.L. Llewellyn, J.H. Lee, J.S. Chang, S.H. Jhung, G. Ferey, Hydrogen storage in the giant-pore metal–organic frameworks MIL-100 and MIL-101, Angew. Chem. Int. Edit. 45 (2006) 8227–8231. [20] Y.W. Li, R.T. Yang, Hydrogen storage in metal–organic and covalent-organic frameworks by spillover, AICHE J. 54 (2008) 269–279. [21] P.L. Llewellyn, S. Bourrelly, C. Serre, A. Vimont, M. Daturi, L. Hamon, G. De Weireld, J.S. Chang, D.Y. Hong, Y.K. Hwang, S.H. Jhung, G. Ferey, High uptakes of CO2 and CH4 in mesoporous metal–organic frameworks MIL-100 and MIL-101, Langmuir 24 (2008) 7245–7250. [22] P. Chowdhury, C. Bikkina, S. Gumma, Gas adsorption properties of the chromium-based metal organic framework MIL-101, J. Phys. Chem. C 113 (2009) 6616–6621. [23] L. Hamon, C. Serre, T. Devic, T. Loiseau, F. Millange, G. Ferey, G. De Weireld, Comparative study of hydrogen sulﬁde adsorption in the MIL-53(Al, Cr, Fe), MIL-47(V), MIL-100(Cr), and MIL-101(Cr) metal–organic frameworks at room temperature, J. Am. Chem. Soc. 131 (2009) 8775–8777. [24] S.H. Jhung, J.H. Lee, J.W. Yoon, C. Serre, G. Ferey, J.S. Chang, Microwave synthesis of chromium terephthalate MIL-101 and its benzene sorption ability, Adv. Mater. 19 (2007) 121–124. [25] Z.X. Zhao, X.M. Li, S.S. Huang, Q.B. Xia, Z. Li, Adsorption and diffusion of benzene on chromium-based metal organic framework MIL-101 synthesized by microwave irradiation, Ind. Eng. Chem. Res. 50 (2011) 2254–2261. [26] C.Y. Huang, M. Song, Z.Y. Gu, H.F. Wang, X.P. Yan, Probing the adsorption characteristic of metal–organic framework MIL-101 for volatile organic compounds by quartz crystal microbalance, Environ. Sci. Technol. 45 (2011) 4490–4496. [27] T.K. Trung, N.A. Ramsahye, P. Trens, N. Tanchoux, C. Serre, F. Fajula, G. Ferey, Adsorption of C5–C9 hydrocarbons in microporous MOFs MIL-100(Cr) and MIL101(Cr): a manometric study, Micropor. Mesopor. Mater. 134 (2010) 134–140. [28] N. Klein, A. Henschel, S. Kaskel, n-Butane adsorption on Cu-3(btc)(2) and MIL101, Micropor. Mesopor. Mater. 129 (2010) 238–242. [29] Z.Y. Gu, X.P. Yan, Metal–organic framework MIL-101 for high-resolution gaschromatographic separation of xylene isomers and ethylbenzene, Angew. Chem. Int. Edit. 49 (2010) 1477–1480. [30] F. Qu, L.Z. Zhu, K. Yang, Adsorption behaviors of volatile organic compounds (VOCs) on porous clay heterostructures (PCH), J. Hazard. Mater. 170 (2009) 7–12. [31] C.E. Webster, R.S. Drago, M.C. Zerner, Molecular dimensions for adsorptives, J. Am. Chem. Soc. 120 (1998) 5509–5516. [32] A. Henschel, K. Gedrich, R. Kraehnert, S. Kaskel, Catalytic properties of MIL-101, Chem. Commun. 419 (2008) 2–4194. [33] N.V. Maksimchuk, M.N. Timofeeva, M.S. Melgunov, A.N. Shmakov, Y.A. Chesalov, D.N. Dybtsev, V.P. Fedin, O.A. Kholdeeva, Heterogeneous selective oxidation catalysts based on coordination polymer MIL-101 and transition metalsubstituted polyoxometalates, J. Catal. 257 (2008) 315–323. [34] J.G. Bell, X.B. Zhao, Y. Uygur, K.M. Thomas, Adsorption of chloroaromatic models for dioxins on porous carbons: the inﬂuence of adsorbate structure and surface functional groups on surface interactions and adsorption kinetics, J. Phys. Chem. C 115 (2011) 2776–2789. [35] M.B. Sweatman, N. Quirke, W. Zhu, R. Kapteijn, Analysis of gas adsorption in Kureha active carbon based on the slit-pore model and Monte–Carlo simulations, Mol. Simul. 32 (2006) 513–522. [36] D. Ramirez, S.Y. Qi, M.J. Rood, Equilibrium and heat of adsorption for organic vapors and activated carbons, Environ. Sci. Technol. 39 (2005) 5864–5871. [37] J.H. Kim, S.J. Lee, M.B. Kim, J.J. Lee, C.H. Lee, Sorption equilibrium and thermal regeneration of acetone and toluene vapors on an activated carbon, Ind. Eng. Chem. Res. 46 (2007) 4584–4594. [38] D. Britt, D. Tranchemontagne, O.M. Yaghi, Metal–organic frameworks with high capacity and selectivity for harmful gases, Proc. Natl. Acad. Sci. U.S.A. 105 (2008) 11623–11627. [39] M.E. Ramos, P.R. Bonelli, A.L. Cukierman, M.M.L.R. Carrott, P.J.M. Carrott, Adsorption of volatile organic compounds onto activated carbon cloths derived from a novel regenerated cellulosic precursor, J. Hazard. Mater. 177 (2010) 175–182. [40] M.P.M. Nicolau, P.S. Barcia, J.M. Gallegos, J.A.C. Silva, A.E. Rodrigues, B.L. Chen, Single- and multicomponent vapor-phase adsorption of xylene isomers and

K. Yang et al. / Journal of Hazardous Materials 195 (2011) 124–131

[41]

[42]

[43]

[44]

ethylbenzene in a microporous metal–organic framework, J. Phys. Chem. C 113 (2009) 13173–13179. D.N. Dybtsev, H. Chun, K. Kim, ﬂexible. Rigid, A highly porous metal–organic framework with unusual guest-dependent dynamic behavior, Angew. Chem. Int. Edit. 43 (2004) 5033–5036. Z.Y. Gu, D.Q. Jiang, H.F. Wang, X.Y. Cui, X.P. Yan, Adsorption and separation of xylene isomers and ethylbenzene on two Zn-terephthalate metal–organic frameworks, J. Phys. Chem. C 114 (2010) 311–316. V. Finsy, H. Verelst, L. Alaerts, D. De Vos, P.A. Jacobs, G.V. Baron, J.F.M. Denayer, Pore-ﬁlling-dependent selectivity effects in the vapor-phase separation of xylene isomers on the metal–organic framework MIL-47, J. Am. Chem. Soc. 130 (2008) 7110–7118. J.S. Lee, S.H. Jhung, Vapor-phase adsorption of alkylaromatics on aluminum–trimesate MIL-96: an unusual increase of adsorption capacity with temperature, Micropor. Mesopor. Mater. 129 (2010) 274–277.

131

[45] I. Senkovska, S. Kaskel, High pressure methane adsorption in the metal–organic frameworks Cu-3(btc)(2), Zn-2(bdc)(2)dabco, and Cr3F(H2O)(2)O(bdc)(3), Micropor. Mesopor. Mater. 112 (2008) 108–115. [46] D.Y. Hong, Y.K. Hwang, C. Serre, G. Ferey, J.S. Chang, Porous chromium terephthalate MIL-101 with coordinatively unsaturated sites: surface functionalization, encapsulation, sorption and catalysis, Adv. Funct. Mater. 19 (2009) 1537–1552. [47] G.Q. Guo, H. Chen, Y.C. Long, Separation of p-xylene from C-8 aromatics on binder-free hydrophobic adsorbent of MFI zeolite. I. Studies on static equilibrium, Micropor. Mesopor. Mater. 39 (2000) 149–161. [48] A.M. Tolmachev, D.A. Firsov, T.A. Kuznetsova, K.M. Anuchin, DFT modeling of the adsorption of benzene, methanol, and ethanol molecules in activated carbon nanopores, Prot. Met. Phys. Chem. Surf. 45 (2009) 163–168.

Journal of Hazardous Materials 195 (2011) 132–138

Contents lists available at SciVerse ScienceDirect

Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat

Exceptional catalytic efﬁciency in mineralization of the reactive textile azo dye (RB5) by a combination of ultrasound and core–shell nanoparticles (CdS/TiO2 ) Narjes Ghows, Mohammad H. Entezari ∗ Department of Chemistry, Ferdowsi University of Mashhad, 91775, Mashhad, Iran

a r t i c l e

i n f o

Article history: Received 14 May 2011 Received in revised form 1 August 2011 Accepted 5 August 2011 Available online 22 August 2011 Keywords: Titanium dioxide Cadmium sulﬁde Core–shell Composite Reactive black 5 Mineralization Sonocatalytic degradation

a b s t r a c t A novel composite with a core–shell structure (CdS/TiO2 ) was prepared through the combination of microemulsion and ultrasound (20 kHz). The degradation of reactive black 5 (RB5) was carried out in aqueous solution in a series of experiments by CdS/TiO2 nanoparticles. This composite with mole ratio of 1/6 has shown an exceptional sonocatalytic activity in comparison to the pure nanoparticles of TiO2 and CdS. A signiﬁcant decrease in the concentration of RB5 (≈94%) was observed in 3 min sonication of the solution containing the core–shell nanocomposite. While at the same time, the concentration was reduced to 4% under sonication without nanocomposite and 50% under UV light with nanocomposite. The increased catalytic activity of nanocomposite in the presence of ultrasound is due to the enhancement of mass transfer, cleaning and sweeping the surface of catalyst, and preventing the aggregation of particles. In addition, the presence of CdS nanoparticles in the composite acts as photosensitizer which not only extends the spectral response to the visible region but also reduces the charge recombination. The selected combined method (sonocatalysis) was able to decolorize and oxidize simultaneously the organic dye with a complete mineralization into SO4 2− and NO3 − ions. © 2011 Elsevier B.V. All rights reserved.

1. Introduction Titanium dioxide as a heterogeneous photocatalyst has received extensive attention for the degradation and mineralization of organic pollutants from water in recent 10 years [1–4]. One of the major disadvantage of TiO2 as a photocatalyst is due to its relatively large band gap (Eg = 3.2 eV for anatase phase) that can only be excited by UV radiation with wavelength less than 380 nm. This strongly causes a limited use of the solar spectra as a source for the photoreaction. Another major limiting factor is the high rate of recombination of the photogenerated electron/hole pairs. Due to these limitations, a coupling of a large band gap semiconductor with a smaller one which contains suitable potential energies for activating in the visible range [5] is of great interest for the degradation of organic pollutants by the solar radiation. In particular, CdS with ideal band gap energy (2.4 eV) is very unstable against photocorrosion in aqueous solutions. Hence, it has been combined with other materials such as ZnO, and TiO2 [6–8]. These composite materials improve the photo-efﬁciency, photostability, and maximize the interfacial areas between the two compounds [5,9,10] to enhance their catalytic properties. In addition to the ﬂat band potential of the components,

∗ Corresponding author. Tel.:+98 511 8797022x306; fax: +98 511 8795457. E-mail address: moh [email protected] (M.H. Entezari). 0304-3894/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.jhazmat.2011.08.049

the photocatalytic performance of the coupled semiconductors is also related to the crystallinity of the particles, the particle size, and the geometry [9,11,12]. These parameters strongly depend on the manner that the couples are prepared. A few examples concerning the combination of CdS and TiO2 have been reported for the photoelectrochemistry [13], water splitting [14], and degradation of pollutants [9,11]. But, it requires a post-thermal treatment for obtaining a coupled nanocomposite and a long time for their photocatalytic performance. Furthermore, the UV light is screened by catalyst particles itself and therefore the region exhibiting the catalytic power is spatially limited in the reactor [15]. Hence, it seems that these limitations can be eliminated by ultrasonic irradiation as an energy source. This is attributed to the synergetic effect of ultrasonic/nanocomposite. Since, the presence of solid particles in a liquid increases the nucleation sites for cavity formation, resulting in the generation of more free radicals [15,16]. In fact, the particle size and its amount could affect the production of OH radical [3,17–22]. However, the large particles which are close to the ultrasonic transducer can block the energy input into the solution [3]. In addition, acoustic cavitation produced by ultrasonic waves profoundly increases the surface area and mass transfer between the two phases. Both of them enhance the diffusion at the interface mixing better than the conventional agitation [2,23]. Moreover, it seems that ultrasound under the mild conditions can improve the contact of the two components in the nanocomposite, crystallinity, and the uniform deposition of the nanosized inorganic particles

N. Ghows, M.H. Entezari / Journal of Hazardous Materials 195 (2011) 132–138

onto another surface or template with the removal of surface contamination [24–28]. Our recent works conﬁrmed that the synthesis of crystalline nanomaterials without any post-thermal treatment is easier with ultrasound than the other methods [29–31]. Ultrasound is able to cause a thin and homogeneous coating [32]. This is due to the high temperatures produced during the cavitation which facilitates the crystallization of the semiconductors [30,33–35]. In this study, CdS/TiO2 couples have been prepared by a combination of ultrasound and microemulsion in order to: (i) exploit the maximum optical absorption in the visible range, (ii) increase the surface contact between CdS and TiO2 nanoparticles, (iii) facilitate the crystallization of the semiconductors. As a result, the prepared composite has high sonocatalytic efﬁciency for the degradation of the selected dye. This is attributed to the synergetic effect of ultrasound and CdS/TiO2 , where the former annihilates any mass-transfer limitations with unblocking catalyst active sites and the latter provides increased nucleation sites for the cavity formation. The sono-catalytic efﬁciency of the CdS/TiO2 coupling is highly dependent on the proportion ratio of the components. They exhibited a faster degradation rate of reactive black 5 than the individual components. In addition, easier workup, higher degradation efﬁciency, shorter reaction time, better control in preparation of core–shell with uniform shape can be considered as advantageous of this study.

133

Fig. 1. Effect of irradiation source on the dye removal (50 mL solution 200 ppm, temperature 38 ◦ C, power of ultrasound 66%, stirring 600 rpm).

The formation of inorganic ions was detected with IC 761 ionchromatograph (model Metrohm).

2. Experimental

3. Results and discussion

2.1. Materials

3.1. Irradiation source

Ethylenediamine, sulfur, p-xylen, titanium tetra-isopropoxide (TTIP), CTAB (cetyl trimethyl ammonium bromide) and 1-butanol from Merck, CdCl2 ·2H2 O from Fluka, and RB5 from Germany (Dystar Company) have been used without further puriﬁcation. De-ionized water was used for the sample preparation.

Fig. 1 compares the rate of dye removal from aqueous solution with different combinations. The combination of ultrasound and nanocomposite (sonocatalyst) shows the highest activity with respect to the combination of nanocomposite with UV light and nanocomposite with sunlight. This is attributed to the synergetic effect of ultrasound and semiconductor. The solid particles increase the nucleation sites for cavity formation, resulting in the generation of more free radicals [16]. Moreover, ultrasonic shockwave prevents blocking of catalyst active site (in contrast to UV light, that the light is screened by catalyst particles itself), increases the availability of the effective sites, and reduces any mass-transfer limitations.

2.2. Procedure 2.2.1. Synthesis of core–shell nanocomposite The synthesis of core–shell nanocomposite and the structure and morphology of the ﬁnal products has been reported in our recent work [32]. 2.2.2. Sonication of dye solution with core–shell nanocomposite A total of 50 mL of RB5 solution (100 mg/L) containing nanocomposite (0.05 g) as a catalyst was sonicated (20 kHz Soniﬁer W-450, output acoustic power 41 W, horn with 1.9 cm diameter) for 10 min in a Rosset cell at initial pH of 6.5. The temperature was controlled by the circulating bath at 38 ◦ C. Then the sample was centrifuged at 8000 rpm for 3 min to separate the suspended catalyst particles from aqueous solution. Some other experiments were carried out with the same conditions under UV lamp (30 W, = 360 nm, distance from liquid surface = 20 cm) and some under sunlight with clear sky in August 1990 (GPS coordinates: N = 36◦ 18 41.6 , E = 59◦ 31 54.2 ).

3.2. Adsorption isotherm Adsorption isotherm of RB5 on nanocomposite surface was determined by mixing 50 mL aqueous solution of dye at various initial concentrations at pH 6.5 for 30 min in a dark place. Data obtained from the adsorption experiments was ﬁtted to the Langmuir equation (Supporting Fig. S1): 1 Ce Ce = + qe Qm bQm where Ce is the equilibrium concentration of RB5 after 30 min, qe is the adsorbed dye concentration on the catalyst surface, and b and Qm are Langmuir adsorption constants which were calculated as 64 mg g−1 and 0.50 L mg−1 , respectively.

2.3. Sample analysis 3.3. Effect of pH The absorbance of the sample was measured by spectrophotometer (model Unico 2800) at 599 nm which corresponds to the maximum absorbance of RB5. The absorption was converted to concentration through the standard curve of reactive black 5. The total organic carbon (TOC) of the samples was determined by a TOC-V CPH (model Shimadzu) analyzer after separation of solid phase by centrifuge.

The pH has a great effect on the sonocatalytic degradation of dye (Fig. 2). The surface charge of TiO2 particles and the adsorption of dye on the surface vary by changing the pH. The interpretation of pH effects on the efﬁciency of dye removal is a very difﬁcult task. This is due to the multiple roles of pH on the ionization state of the surface, hydroxyl radical formation, agglomeration of

134

N. Ghows, M.H. Entezari / Journal of Hazardous Materials 195 (2011) 132–138

Fig. 2. The effect of pH on the removal of pollutant (50 mL solution 100 ppm, 0.05 g catalyst, temperature 38 ◦ C, time 10 min, stirring 600 rpm).

particles, and the speciﬁcation of dye and products [36]. The pH can affect the adsorption and agglomeration which are important factors on sonocatalytic degradation of RB5. The literature indicated that pHzpc of TiO2 for anatase is 6.8 [37,38]. Therefore, when the solution pH is less or more than pHzpc , the surface hydroxyl groups of TiO2 would undergo a proton association or dissociation reaction and lead to a positive or negative charge on TiO2 surface. Repulsion between charged surfaces prevents the agglomeration of the catalyst particles in the medium. The acidic solution favors adsorption of the dye molecules onto TiO2 surface due to the change in the speciﬁcation of the dye. The pKa values for the dye are almost close to 3.8 and 6.9 [39]. Then, the dye molecules are in the molecular form in acidic medium. But, the sulfonate group cannot be protonated under the applied pH and it can be existed in the anionic form. Therefore, there is an attraction between the surface of the nanocomposite and the dye molecule through the sulfonate group at acidic pH. This results to the adsorption of dye molecules onto the surface and consequently its degradation with hydroxyl radicals produced by catalyst [2]. But, a small decrease of degradation was observed in acidic medium in comparison to the neutral conditions in sonocatalytic degradation. This should be due to the adding of inorganic acid (HCl) to the solution which led to an increase amount of Cl− in the solution. The Cl− ions can react with hydroxyl radicals and produce inorganic radical anions with lower activity than hydroxyl radical in the degradation of dye [40]. In alkaline medium, the dye molecules are mostly in their anionic forms. In this situation, the columbic repulsion is appeared between the negative charged surface of the catalyst at high pH and the negative charge of anionic groups. Therefore, the removal was lower in basic pH medium (pH 10.9). The results clearly demonstrate that the combination of catalyst and sonication in near neutral pH causes an additive effect on the degradation rates of the pollutant which is due to the more efﬁcient generation of hydroxyl radicals by nanocomposite. The rate of degradation was very slow by sonication alone. Since, RB5 is a non-volatile compound and the place of degradation would be mostly at the exterior of the cavitation bubbles [41–43]. A little increase of degradation rate in acidic conditions might be due to the more hydrophobic character of the resulting dye molecule with respect to the basic conditions which the dye molecule has more hydrophilic character. As the hydrophobic character of the molecule increases, the chance of presence of the molecule near the cavitation is higher and this leads to more degradation [44].

Fig. 3. The effect of concentration on the removal of pollutant in the presence of ultrasound (50–400 ppm solution, 0.05 gr catalyst, temperature 38 ◦ C, power of ultrasound 66%, time 10 min).

3.4. Effect of initial concentration Fig. 3 shows the inﬂuence of initial concentration of RB5 in the range of 50–400 mg/L on its sonocatalytic degradation using the synthesized nanocomposite (TiO2 /CdS) as a catalyst. The amount of dye removal increased by increasing the dye concentration and then decreased. But, the percent removal of RB5 decreased with increasing the initial concentration. For instance at the concentration of 100 mg/L, a complete removal (100%) was achieved after about 10 min while at the concentration of 300 mg/L, it decreased to 76.8% after 10 min. As the amount of composite is ﬁxed, the number of adsorption sites is constant. By increasing the dye concentration, the chance of trapping the dye molecule at higher concentrations is less than lower concentrations. At higher concentrations, most of the composite sites are occupied and the chance of ﬁnding sites for the adsorption is low. This is the reason for the lower removal percentage at higher concentrations. In addition, for the highest removal under sonication there is an optimum concentration for the pollutant in the mixture. The concentration below and above the optimum concentration can cause a lower effect of cavitation. The dye degradation happens in different ways. Some dye molecules degrade by hydroxyl radical through oxidation (Eq. (1)) [4]. In addition, several authors have proposed the dye removal occurs through direct electron (e− ) transfer from the semiconductor surface to the dye molecule [36,45,46], as shown by Eq. (2). Direct oxidation by reaction with holes (h+ ) has also been reported [36,47], as shown by Eq. (3). Thus, with increasing the dye concentration, it is not possible to degrade 100% of the pollutant as the available species responsible for degradation is approximately constant under the applied conditions. Dye + OH• → Dye• − + H2 O → degradation products

(1)

−

Dye• −

→ degradation products

(2)

+

Dye• −

→ degradation products

(3)

Dye + e → Dye + h →

3.5. Effect of sonication time The removal of RB5 from aqueous solution was shown by three different methods: sonication of solution with nanocomposite (Fig. 4a), sonication alone (Fig. 4b), and comparison of the mentioned methods with mixing of solution by nanocomposite under UV (Fig. 4c). It was found that, all absorption peaks decayed more or less with respect to the original dye solution spectrum. The largest reduction (more than 95%) was achieved in the

N. Ghows, M.H. Entezari / Journal of Hazardous Materials 195 (2011) 132–138

135

Fig. 4. The effect of contact time on the removal of dye. (a) Catalyst with ultrasound (sonocatalyst), (b) sonication alone, (c) the change of concentration with time (50 mL solution 100 ppm, 0.05 g catalyst, temperature 38 ◦ C, intensity of ultrasound 66%).

presence of nanocomposite powder with ultrasound (sonocatalyst) in less than 3 min. These results indicate that the removal of RB5 by nanocomposite CdS/TiO2 combined with ultrasonic irradiation was more effective than those by UV method and sonication alone. Ultrasonic irradiation of semiconductor nanocomposite as well as light in the environment may promote • OH formation or superoxide radicals (• O2 ) on the surfaces of the catalyst. Formation of these reactive oxygen species initiates a series of chemical reactions that promotes the power of oxidation. The reactive species might be formed by the absorption of light from the environment by the nanocomposite and hot energies from ultrasonic cavitation which leads to the excitement of semiconductor and production of electron and hole. Finally, the reactive species can be formed by the redox reactions of electron and hole with adsorbed species on the surface of nanocomposite. In addition, the pyrolysis of H2 O molecules and direct formation of hydroxyl radical through the cavitation is another source of production. They are capable of undergoing further reaction with solutes adsorbed at the bubble–solution interface [41,42,48] or with solutes in bulk solution [49]. Moreover, ultrasonic shockwave prevents blocking of catalyst active site. But, the efﬁciency of ultraviolet light was lower which is due to the UV-screening of the catalysts. Sonication alone can degrade the dye molecule by the radicals produced through the cavitation. It is assumed that a local high concentration of hydroxyl radicals exist at the interface region of the collapsing bubbles [42,50] and some of them escape to the bulk of solution [44]. RB5 is a non-volatile compound and the region of degradation would be at the exterior of the cavitation bubbles. 3.6. IR spectrum The IR study was done for understanding the adsorbed species on the catalyst and also their degradation through appearance and

disappearance of different peaks. Supporting Fig. S2 shows the IR spectrum of the samples. For sonocatalytic reactions, the adsorbed species are signiﬁcant. They will react with sonoexcited products on the catalyst surface and produce reactive radicals which are powerful oxidants in degrading organics in water. The band at 1625 cm−1 corresponds to the bending mode of adsorbed water. The intense band at 1504–1514 cm−1 has been attributed to the –N N– bond vibrations or to the aromatic ring vibrations sensitive to the interaction with the azo bond [46]. The bands at 1599, 1572, and 1452 cm−1 are linked to C C aromatic skeletal vibrations [51]. Finally, the peaks at 1122, 1226, 1130, and 1047 cm−1 are respectively assigned to the stretching of the SO4 group and sulfoxide [52]. A slight shift of the bands toward lower wave numbers and a major decrease of the band intensities are observed. The most important shift is observed for the band associated with the chromophore part of the dye, which is shifted from 1514 to 1504 cm−1 . While the bands of associated to the sulfonate group are almost disappeared. The changes observed show that degradation of the adsorbed species is taking place. This is particularly true for the band at 1504 cm−1 , which reﬂects the destruction of the chromophore part of the azo dye. 3.7. Effect of catalyst composition The rate of dye removal was examined by different catalysts from pure components to different composites. Fig. 5 shows the rate of dye removal by TiO2 , CdS and CdS/TiO2 with two different ratios (1/2.5, 1/6.0). The degradation rate of RB5 in the presence of CdS/TiO2 with ratio equal to 1/6 shows a signiﬁcant difference with respect to the other samples. The high activity of the composite photocatalyst is due to the fast charge separation and transportation throughout the particles [5,9,53] as well as the contact between TiO2 and solution which is sufﬁcient for the creation of active

136

N. Ghows, M.H. Entezari / Journal of Hazardous Materials 195 (2011) 132–138

Fig. 6. Concentration and TOC of the RB5 solution versus time.

Fig. 5. Comparison of different catalysts for the dye removal in the presence of ultrasound (50 mL solution 100 ppm, 0.05 gr catalyst, temperature 38 ◦ C, power of ultrasound 66%).

radicals [11]. The possible role of TiO2 nanoparticles is to provide sites for collecting the electrons generated from CdS, enabling thereby an efﬁcient electron–hole separation as depicted. The charge separated state can be followed by the emission decay of the sample. In nanocomposite a decrease in the intensity of the luminescence emission was observed for CdS. This means the signiﬁcant quenching in emission measurements can be attributed to the electron transfer between excited CdS and TiO2 . This quenching behavior represents the deactivation of the excited CdS via electron transfer to TiO2 particles [53]. The sonocatalytic performance of coupled semiconductors is related to the charge separation, the surface contact between nanoparticles, the crystallinity, and sizes in the coupled particles.

acids which could be the reason for the decrease of pH from neutral to slightly acidic (from 6.5 to 5) [57]. Evolution of sulfate ions in the solution conﬁrmed the results of TOC. The concentration of this ion was 16 mg/L and 19 mg/L in the solution after 10 min and 60 min sonication, respectively. Sulfate ions are primary products from the initial attack on the sulfonyl group of dye by a direct reaction with positive holes or hydroxyl radicals. Between both oxidative agents (OH• and h+ ), it is suggested that the reaction with holes could be preferential since the adsorbed molecule is in contact with the surface of titania via one of its sulfonate groups [58]. Surprisingly, Only 40% of the sulfate was present in the solution (less than the expected stoichiometric quantity). This behavior could be described as a partially irreversible adsorption of SO4 2− ions. The strong adsorption of SO4 2− could partially inhibit the reaction rate which, however, remains acceptable [56,59]. The concentration of nitrate ions was 17 mg/L and 19 mg/L in the solution after 10 min and 60 min sonication, respectively. On the other hand, substantial increase in the conductivity from101 ␮S/cm to 121 ␮S/cm suggests the formation of ions such as nitrate or sulfate.

3.8. Mineralization of the dye 3.9. Proposed mechanism of sonocatalytic degradation A complete photocatalytic mineralization reaction of reactive black 5 has been suggested as follows [54,55]: UV/TiO2

C26 H21 N5 O19 S6 Na4 38O2 −→ 26CO2 + 4H2 O + 4NaNo3 + HNO3 + 6H2 SO4 The formation of CO2 , NO3 − , and SO4 2− has been extensively studied by Poulios and Tsachpinis [55]. They reported that after the decolorization, the intermediates were present in the medium. The sonocatalytic mineralization of RB5 was followed by measuring the TOC during the process. Fig. 6 shows the TOC and its variation with time during the sonocatalytic degradation. The TOC was abated to about 50% of the initial value in less than 10 min of irradiation, and then decreased slowly. This is might be due to a detrimental effect of the adsorbed SO4 2− ions on the catalyst surface [56]. It is another fact that the intermediates such as carboxylic acids are difﬁcult to oxidize than their parent compound (azo dye in this case). Therefore, a complete oxidation may proceed with slower rate [36]. But, the degree of mineralization in UV method was very low. The results obtained from the absorbance and TOC measurements indicate that the decay of the chromophore in the dye molecule is a relatively fast process (disappearance of chromophore peak in the visible spectra) but overall degradation requires more time. Some intermediates are in the form of organic

A few mechanism and satisfying explanation can be found about the sonocatalytic degradation of organic pollutants in the presence of various semiconductor materials [2,4,9,26]. It is obvious that further studies are necessary to determine the mechanism in greater details. Kamat and Patrick [60] have demonstrated the simultaneous migration of both electrons and holes in coupled semiconductor photocatalysts. An increase in the lifetime of the photo-generated pairs due to the hole and electron transfers between the two semiconductors in nanocomposite, was invoked in many cases as the key factor for the improvement of the photoactivity. In this work a novel composite of core–shell CdS/TiO2 with a high sonocatalytic activity and appropriate adsorbability were obtained through the combination of nano-sized CdS and TiO2 particles. The possible process of sonocatalytic degradation and reactions are proposed in Fig. 7 based on the following statements: (i) Based on the cavitation in our experiments, the ultrasonic irradiation can result in the formation of high temperatures as a spot (hot spot theory). The high temperatures produced by ultrasound brings many holes and electrons which are changed to reactive radicals on the surface of semiconductor catalyst [3,4,9]. Under these conditions, a series of reactions can take place on the surface of nanocomposite. In addition,

N. Ghows, M.H. Entezari / Journal of Hazardous Materials 195 (2011) 132–138

137

Fig. 7. Proposed mechanism for RB5 removal.

the cavitation promotes the electron transfer from CdS to TiO2 through the crystal interphase between TiO2 and CdS, which effectively results in the separation of electron–hole pairs. This separation enhances the sonocatalytic activity. (ii) Enhancing the mass transfer of organic materials such as dye molecule between the liquid phase and the catalyst surface which is due to the shockwave propagation of the cavitation process. (iii) Increasing the catalytic activity by de-aggregation of the catalyst particles which tend to aggregate due to its high potential surface energy. (iv) Cleaning and sweeping the catalyst surface due to acoustic micro-streaming which allows more active catalyst sites to be available. The contribution of the mentioned effects is difﬁcult to be differentiated and the overall net effect is an enhancement in the rate of dye degradation. 4. Conclusion In this study, the sonocatalytic activities of nano-sized TiO2 , nano-sized CdS and its composite CdS/TiO2 with two different ratios were compared by the degradation of RB5. The experimental results showed that the prepared nanocomposite of TiO2 /CdS exhibited a high sonocatalytic activity in comparison with pure nano-sized TiO2 and CdS. The presence of catalyst enhanced the rate of RB5 removal. The extent of removal depends on the operating conditions employed such as the type and concentration of catalyst, initial dye concentration and solution pH. The removal of dye increases with the increase of ultrasonic irradiation time, while the removal percentage was decreased with the increase of initial concentration. A signiﬁcant decrease in the concentration of RB5 was observed at initial times of sonication in sonocatalytic method for the higher mole ratio of the nanocomposite. In addition to the decolorization, sonocatalytic method was able to mineralize the organic compound into SO4 2− and NO3 − ions completely. The increased catalytic activity was due to the improvement of charge separation by the nanocomposite and de-aggregation of the nanocatalyst by ultrasound which allows more active sites to be available for the reaction. Acknowledgments The authors acknowledge the help given by Mrs. M. Hassanzadeh from Solid State Physics Research Center, Damghan

University of Basic Sciences and Mrs. R. Pesyan from Central Research Laboratory of Ferdowsi University of Mashhad for analysis. This work has been supported by the “Iranian National Science Foundation: INSF” (No. 85103/31). Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.jhazmat.2011.08.049. References [1] M. Ni, M.K.H. Leung, D.Y.C. Leung, K. Sumathy, A review and recent developments in photocatalytic water-splitting using TiO2 for hydrogen production, Renew. Sust. Energy Rev. 11 (2007) 401–425. [2] D.E. Kritikos, N.P. Xekoukoulotakis, E. Psillakis, D. Mantzavinos, Photocatalytic degradation of reactive black 5 in aqueous solutions: effect of operating conditions and coupling with ultrasound irradiation, Water Res 41 (2007) 2236–2246. [3] N. Shimizu, C. Ogino, M. Farshbaf Dadjour, T. Murata, Sonocatalytic degradation of methylene blue with TiO2 pellets in water, Ultrason. Sonochem. 14 (2007) 184–190. [4] J. Wang, W. Sun, Z. Zhang, Z. Xing, R. Xu, R. Li, Y. Li, X. Zhang, Treatment of nano-sized rutile phase TiO2 powder under ultrasonic irradiation in hydrogen peroxide solution and investigation of its sonocatalytic activity, Ultrason. Sonochem. 15 (2008) 301–307. [5] P.V. Kamat, Quantum dot solar cells. Semiconductor nanocrystals as light harvesters, J. Phys. Chem. C 112 (2008) 18737. [6] L. Spanhel, H. Weller, A. Henglein, Photochemistry of semiconductor colloids. 22. Electron injection from illuminated CdS into attached TiO2 and ZnO particles, J. Am. Chem. Soc. 109 (1987) 6632–6635. [7] S.J. Jum, G.K. Hyun, A.J. Upendra, W.J. Ji, S.L. Jae, Fabrication of CdS nanowires decorated with TiO2 nanoparticles for photocatalytic hydrogen production under visible light irradiation, Int. J. Hydrogen Energy 33 (2008) 5975–5980. [8] J.C. Kim, J. Choi, Y.B. Lee, J.H. Hong, J.I. Lee, J.W. Yang, et al., Enhanced photocatalytic activity in composites of TiO2 nanotubes and CdS nanoparticles, Chem. Commun. 502 (2006) 4–6. [9] J. Wang, Z. Jiang, L. Zhang, P. Kang, Y. Xie, Y. Lv, R. Xu, X. Zhang, Sonocatalytic degradation of some dyestuffs and comparison of catalytic activities of nanosized TiO2 , nano-sized ZnO and composite TiO2 /ZnO powders under ultrasonic irradiation, Ultrason. Sonochem. 16 (2009) 225–231. [10] P.A. Sant, P.V. Kamat, Inter-particle electron transfer between size-quantized CdS and TiO2 semiconductor nanoclusters, Phys. Chem. Chem. Phys. 4 (2002) 198–203. [11] Y. Bessekhouad, N. Chaoui, M. Trzpit, N. Ghazzal, D. Robert, J.V. Weber, UV–vis versus visible degradation of Acid Orange II in a coupled CdS/TiO2 semiconductors suspension, J. Photochem. Photobiol. A: Chem. 183 (2006) 218–224. [12] P.R. Yu, J.M. Nedeljkovic, P.A. Ahrenkiel, R.J. Ellingson, A.J. Nozik, Size dependent femtosecond electron cooling dynamics in CdSe quantum rods, Nano Lett. 4 (2004) 1089–1092. [13] S. Chen, M. Paulose, C. Ruan, G.K. Mor, O.K. Varghese, D. Kouzoudis, et al., Electrochemically synthesized CdS nanoparticle-modiﬁed TiO2 nanotube-array photoelectrodes: preparation, characterization, and application to photoelectrochemical cells, J. Photochem. Photobiol. A: Chem. 177 (2006) 77–184. [14] J.S. Jang, W. Li, S.H. Oh, J.S. Lee, Fabrication of CdS/TiO2 nano-bulk composite photocatalysts for hydrogen production from aqueous H2 S solution under visible light, Chem. Phys. Lett. 425 (2006) 278–282.

138

N. Ghows, M.H. Entezari / Journal of Hazardous Materials 195 (2011) 132–138

[15] M. Kubo, K. Matsuoka, A. Takahashi, N. Shibasaki-Kitakawa, T. Yonemoto, Kinetics of ultrasonic degradation of phenol in the presence of TiO2 particles, Ultrason. Sonochem. 12 (2005) 263. [16] C. Sauter, M.A. Emin, H.P. Schuchmann, S. Tavman, Inﬂuence of hydrostatic pressure and sound amplitude on the ultrasound induced dispersion and de-agglomeration of nanoparticles, Ultrason. Sonochem. 15 (2008) 517–523. [17] A. Keck, E. Gilbert, R. Köster, Inﬂuence of particles on sonochemical reactions in aqueous solutions, Ultrasonics 40 (2002) 661–665. [18] H.B. Marschall, K.A. Mørch, A.P. Keller, M. Kjeldsen, Cavitation inception by almost spherical solid particles in water, Phys. Fluids 15 (2003) 545–553. [19] T. Tuziuti, K. Yasui, M. Sivakumar, Y. Iida, Correlation between acoustic cavitation noise and yield enhancement of sonochemical reaction by particle addition, J. Phys. Chem. A 109 (2005) 4869–4872. [20] E. Cho, B. Park, M. Lim, J. Khim, Sonochemical degradation in presence of particle, Proc. Symp. Ultrason. Electron. 31 (2010) 89–90. [21] T. Tuziuti, K. Yasui, T. Kozuka, A. Towata, (AIST), inﬂuence of particle addition on sonochemiluminescence intensity under pulsed ultrasound, Proc. Symp. Ultrason. Electron. 31 (2010) 355–356. [22] T. Tuziuti, K. Yasui, T. Kozuka, A. Towata, Inﬂuence of liquid-surface vibration on sonochemiluminescence intensity, J. Phys. Chem. A 114 (2010) 7321–7325. [23] M.H. Entezari, N. Ghows, M. Chamsaz, Combination of ultrasound and discarded tire rubber: removal of Cr(III) from aqueous solution, J. Phys. Chem. A 109 (2005) 4638. [24] N.A. Dhas, K.S. Suslick, Sonochemical preparation of hollow nanospheres and hollow nanocrystals, J. Am. Chem. Soc. 127 (2005) 2368–2369. [25] S. Anandan, F. Grieser, M. Ashokkumar, Sonochemical synthesis of Au–Ag core–shell bimetallic nanoparticles, J. Phys. Chem. C 112 (2008) 15102– 15105. [26] W. Ho, J.C. Yu, Sonochemical synthesis and visible light photocatalytic behavior of CdSe and CdSe/TiO2 nanoparticles, J. Mol. Catal. A: Chem. 247 (2006) 268–274. [27] A. Tiehm, S. Krabnitzer, Y. Koltypin, A. Gedanken, Chloroethene dehalogenation with ultrasonically produced air-stable nano iron, Ultrason. Sonochem. 16 (2009) 617. [28] A.L. Morel, S.I. Nikitenko, K. Gionnet, A. Wattiaux, J. Lai-Kee-Him, C. Labrugere, B. Chevalier, G. Deleris, C. Petibois, A. Brisson, M. Simonoff, Sonochemical approach to the synthesis of Fe3 O4 @SiO2 core–shell nanoparticles with tunable properties, ACS Nano 2 (2008) 847. [29] M.H. Entezari, N. Ghows, Micro-emulsion under ultrasound facilitates the fast synthesis of quantum dots of CdS at low temperature, Ultrason. Sonochem. 18 (2011) 127–134. [30] N. Ghows, M.H. Entezari, Ultrasound with low intensity assisted the synthesis of nanocrystalline TiO2 without calcinations, Ultrason. Sonochem. 17 (2010) 878–883. [31] N. Ghows, M.H. Entezari, A novel method for the synthesis of CdS nano-particles without surfactant, Ultrason. Sonochem. 18 (2011) 269–275. [32] N. Ghows, M.H. Entezari, Fast and easy synthesis of core–shell nanocrystal (CdS/TiO2 ) at low temperature by micro-emulsion under ultrasound, Ultrason. Sonochem. 18 (2011) 629–634. [33] S. Avivi (Levi), O. Palchik, V. Palchik, M.A. Slifkin, A.M. Weiss, A. Gedanken, Sonochemical synthesis of nanophase indium sulﬁde, Chem. Mater. 13 (2001) 2195–2200. [34] W.P. Huang, X.H. Tang, Y.Q. Wang, Y. Koltypin, A. Gedanken, Selective synthesis of anatase and rutile via ultrasound irradiation, Chem. Commun. 141 (2000) 5–1416. [35] S.S. Lee, K.W. Seo, S.H. Yoon, I.-W. Shim, K.-T. Byun, H.-Y. Kwak, CdS coating on TiO2 nanoparticles under multibubble sonoluminescence condition, Bull. Korean Chem. Soc. 26 (2005) 1579. [36] I.K. Konstantinou, T.A. Albanis, TiO2 -assisted photocatalytic degradation of azo dyes in aqueous solution: kinetic and mechanistic investigations. A review, Appl. Catal. B: Environ. 49 (2004) 1–14. [37] J. Wang, Z. Pan, Z. Zhang, X. Zhang, F. Wen, T. Ma, et al., Sonocatalytic degradation of methyl parathion in the presence of nanometer and ordinary anatase titanium dioxide catalysts and comparison of their sonocatalytic abilities, Ultrason. Sonochem. 13 (2006) 493–500.

[38] J. Zhao, K. Wu, T. Wu, H. Hidaka, N. Serpone, Photodegradation of dyes with poor solubility in an aqueous surfactant/TiO2 dispersion under visible light irradiation, J. Chem. Soc., Faraday Trans. 94 (1998) 673–676. [39] D. Lucio, D. Laurent, G. Roger, Adsorption of remazol black B dye on activated carbon felt, Carbon-Sci. Technol. 1 (2) (2008) 66–71. [40] M.A. Rauf, S.B. Bukallah, A. Hammadi, A. Soliman, F. Hammadi, The effect of operational parameters on the photoinduced decoloration of dyes using a hybrid catalyst V2 O5 /TiO2 , Chem. Eng. J. 129 (2007) 167–172. [41] A. Tauber, G. Mark, H.-P. Schuchmann, C. von Sonntag, Sonolysis of tert-butyl alcohol in aqueous solution, J. Chem. Soc., Perkin Trans. 2 (1999) 1129–1135. [42] K. Okitsu, K. Iwasaki, Y. Yobiko, H. Bandow, R. Nishimura, Y. Maeda, Sonochemical degradation of azo dyes in aqueous solution: a new heterogeneous kinetics model taking into account the local concentration of OH radicals and azo dyes, Ultrason. Sonochem. 12 (2005) 255–262. [43] R. Vijayakumar, Yu. Koltypin, I. Felner, A. Gedanken, Sonochemical synthesis and characterization of pure nanometer-sized Fe3 O4 part, Mater. Sci. Eng. A 286 (2000) 101–105. [44] S. Vajnhandl, A.M.L. Marechal, Case study of the sonochemical decolouration of textile azo dye reactive black 5, J. Hazard. Mater. 141 (2007) 329–335. [45] W.Z. Tang, H. An, UV/TiO2 photocatalytic oxidation of commercial dyes in aqueous solutions, Chemosphere 31 (1995) 4157–4170. [46] K. Vinodgopal, D.E. Wynkoop, P.V. Kamat, Environmental photochemistry on semiconductor surfaces: photosensitized degradation of a textile azo dye, acid orange 7, on TiO2 particles using visible light, Environ. Sci. Technol. 30 (1996) 1660–1666. [47] A. Houas, H. Lachheb, M. Ksibi, E. Elaloui, C. Guillard, J.M. Herrmann, Photocatalytic degradation pathway of methylene blue in water, Appl. Catal. B Environ. 31 (2001) 145–157. [48] A. Henglein, Adv. Sonochem. 3 (1993) 17. [49] R.A. Caruso, M. Ashokkumar, F. Grieser, Sonochemical formation of gold sols, Langmuir 18 (2002) 7831. [50] B. Yim, H. Okuno, Y. Nagata, R. Nishimura, Y. Maeda, Sonolysis of surfactants in aqueous solutions: an accumulation of solute in the interfacial region of the cavitation bubbles, Ultrason. Sonochem. 9 (2002) 209. [51] R. Silverstein, T. Clayton, T. Morril, Spectrophotometric Identiﬁcation of Organic Compounds, Wiley, New York, 1991. [52] C. Bradua, L. Frunzab, N. Mihalchea, S.-M. Avramescua, M. Neata, I. Udreaa, Removal of reactive black 5 azo dye from aqueous solutions by catalytic oxidation using CuO/Al2 O3 and NiO/Al2 O3 , Appl. Catal. B: Environ. 96 (2010) 548–556. [53] A. Kongkanand, K. Tvrdy, K. Takechi, M. Kuno, P.V. Kamat, Quantum dot solar cells. Tuning photoresponse through size and shape control of CdSe–TiO2 architecture, J. Am. Chem. Soc. 130 (2008) 4007–4015. [54] A. Mills, R. Davies, The photomineralization of reactive black 5 sensitized by titanium oxide: a study of the initial kinetics of dye photobleaching, in: D.F. Ollis, H. Al-Ekabi (Eds.), Photocatalytic Puriﬁcation and Treatment of Water and Air, Amsterdam, The Netherlands, 1993, pp. 595–600. [55] I. Poulios, I. Tsachpinis, Photodegradation of the textile dye reactive black 5 in the presence of semiconducting oxides, J. Chem. Technol. Biotechnol. 74 (1999) 349–357. [56] C. Gomes da Silva, J.L. Faria, Photochemical and photocatalytic degradation of an azo dye in aqueous solution by UV irradiation, J. Photochem. Photobiol. A: Chem. 155 (2003) 133. [57] K. Sahel, N. Perol, H. Chermette, C. Bordes, Z. Derriche, C. Guillard, Photocatalytic decolorization of Remazol Black 5 (RB5) and Procion Red MX-5B—isotherm of adsorption, kinetic of decolorization and mineralization, Appl. Catal. B: Environ. 77 (2007) 100–109. [58] M. Karkmaz, E. Puzenat, C. Guillard, J.M. Herrmann, Photocatalytic degradation of the alimentary azo dye amaranth. Mineralization of the azo group to nitrogen, Appl. Catal. B: Environ. 51 (2004) 183–194. [59] M. Kerzhentsev, C. Guillard, J.M. Herrmann, P. Pichat, Photocatalytic pollutant removal in water at room temperature: case study of the total degradation of the insecticide fenitrothion (phosphorothioic acid O,O-dimethyl-O-(3-methyl4-nitro-phenyl) ester), Catal. Today 27 (1996) 215. [60] P.V. Kamat, B. Patrick, Photophysics and photochemistry of quantized zinc oxide colloids, J. Phys. Chem. 96 (1992) 6829–6834.

Journal of Hazardous Materials 195 (2011) 139–146

Contents lists available at ScienceDirect

Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat

Waste oil shale ash as a novel source of calcium for precipitated calcium carbonate: Carbonation mechanism, modeling, and product characterization O. Velts a,b,∗ , M. Uibu a , J. Kallas a , R. Kuusik a a b

Laboratory of Inorganic Materials, Tallinn University of Technology, Ehitajate tee 5, Tallinn 19086, Estonia Laboratory of Separation Technology, Lappeenranta University of Technology, P.O. Box 20, Lappeenranta FI-53851, Finland

a r t i c l e

i n f o

Article history: Received 14 May 2011 Received in revised form 11 July 2011 Accepted 6 August 2011 Available online 12 August 2011 Keywords: Oil shale ash Precipitated calcium carbonate Modeling Carbonation mechanism CaCO3 polymorphs

a b s t r a c t In this paper, a method for converting lime-containing oil shale waste ash into precipitated calcium carbonate (PCC), a valuable commodity is elucidated. The mechanism of ash leachates carbonation was experimentally investigated in a stirred semi-batch barboter-type reactor by varying the CO2 partial pressure, gas ﬂow rate, and agitation intensity. A consistent set of model equations and physical–chemical parameters is proposed to describe the CaCO3 precipitation process from oil shale ash leachates of complex composition. The model enables the simulation of reactive species (Ca2+ , CaCO3 , SO4 2− , CaSO4 , OH− , CO2 , HCO3 − , H+ , CO3 2− ) concentration proﬁles in the liquid, gas, and solid phases as well as prediction of the PCC formation rate. The presence of CaSO4 in the product may also be evaluated and used to assess the purity of the PCC product. A detailed characterization of the PCC precipitates crystallized from oil shale ash leachates is also provided. High brightness PCC (containing up to ∼96% CaCO3 ) with mean particle sizes ranging from 4 to 10 ␮m and controllable morphology (such as rhombohedral calcite or coexisting calcite and spherical vaterite phases) was obtained under the conditions studied. © 2011 Elsevier B.V. All rights reserved.

1. Introduction In order to sustainably meet the ever-rising demand for energy, it is becoming necessary to exploit lower-quality fossil fuels such as oil shale. Well-explored oil shale reserves include the Green River deposits in the western United States, the Tertiary deposits in Queensland, Australia, the El-Lajjun deposit in Jordan, and deposits in Sweden, Estonia, France, Germany, Brazil, China, and Russia. In Estonia, large-scale combustion of calcareous kerogenous oil shale (8–12 MJ kg−1 ) provides over 90% of the basic electric power supply. The technology used in oil shale processing for heat and power production exerts strong environmental effects. Due to the extensive use of oil shale, per capita CO2 emissions in Estonia (15.2 metric tonnes in 2007) are about twice the European average and rank 13th worldwide [1]. In addition the process produces approximately 5–7 Mt of hazardous ash annually. A small portion of the waste ash is used for construction materials, road construction, and agricultural purposes [2], while most of the ash is transported as a slurry to be deposited on waste piles near the power plants. These ash dumps occupy an area of approximately 20 km2 . The

∗ Corresponding author at: Laboratory of Inorganic Materials, Tallinn University of Technology, Ehitajate tee 5, Tallinn 19086, Estonia. Tel.: +372 5283756; fax: +372 620 2801. E-mail address: [email protected] (O. Velts). 0304-3894/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.jhazmat.2011.08.019

combustion waste ash is rich in free lime and anhydrite that under aqueous conditions produces highly alkaline leachates (pH 12–13). These pose a potential long-term environmental risk as neutralization of ash ﬁelds under natural conditions may take hundreds of years [3,4]. In the context of reducing the environmental burden and enhancing economic beneﬁt, strategies for upgrading waste ashes into products of commercial value have arisen into focus, for instance [5–8]. Related to the aforementioned issues in Estonia, the authors recently introduced a novel approach for synthesizing precipitated calcium carbonate (PCC) crystals utilizing alkaline waste ash as an alternative low-cost source of water-soluble calcium [9]. PCC is currently produced from lime in a multi-stage process that requires large amounts of energy and uses expensive high-quality raw material. PCC production using oil shale ash could have considerable commercial importance in the paint, plastics, rubber, and paper industries. Other potential advantages of this process such as safer disposal of wastes, CO2 emissions reduction, and wastewater neutralization were elaborated in our earlier studies [10,11]. Also, a new method for intensive heterogeneous gas–liquid processing was proposed [12]. One of the main challenges in this work was establishing a quantitative understanding of heterogeneous gas–liquid–solid system kinetics and dynamics. In this paper, the mechanism of calcium carbonate precipitation during gas–liquid reaction of oil shale ash leachates is discussed as well as a mathematical model describing the precipitation process reported. The

140

O. Velts et al. / Journal of Hazardous Materials 195 (2011) 139–146

current study also examines the impact of the complex composition of ash leachates on the main characteristics (composition, morphology, surface area, and particle size) of the solid product over a wide range of operating conditions.

2. Experimental 2.1. Preparation of alkaline mother solutions (leachates) The leaching of Ca2+ and other ions from oil shale ash was previously studied by the authors [3,13,14]. In this paper, oil shale (pulverized ﬁring) combustion ash (containing ∼8.0% free CaO) was dispersed in distilled water (10:1 w/w liquid to solid ratio) under atmospheric pressure and room temperature for 15 min in a 15 L reactor equipped with a turbine-type impeller (Fig. 1(a)). The alkaline suspension was ﬁltered from the solid ash residue. The solutions were analyzed for Ca2+ (titrimetric method ISO 6058:1984), SO4 2− , Cl− , K+ , PO4 3− (using a Lovibond SpectroDirect spectrophotometer), CO3 2− , HCO3 − , and OH− (titrimetric method ISO 9963–1:1994(E)). The oil shale ash leachates (pH ∼12.65) had the following average ion concentration: (in g L−1 ) Ca2+ : ∼1.23, SO4 2− : ∼0.75, K+ : ∼0.076, Cl− : ∼0.038, PO4 3− : ∼0.011 and (in mol L−1 ) OH− : ∼0.047.

2.2. Synthesis of PCC particles Carbonation of oil shale ash leachates was performed in a semibatch barboter-type reactor. A turbine-type impeller was used to provide effective mechanical mixing of the gas and liquid phases to increase the interfacial contact area (Fig. 1(b)). Recirculating alkaline mother solution (à 10 L) was treated with a model gas mixture containing pre-determined concentrations of CO2 in air (cCO2 ). The CO2 content was based on typical industrial ﬂue gas compositions. The ﬂow rate (QG ) and composition of the inlet gas were controlled using calibrated rotameters and an infrared CO2 analyzer (Duotec). The reactor was operated batch-wise with respect to the liquid phase and continuously with respect to the gas phase. A 23 full-factorial experimental plan was designed in which the process variables were maintained near the center of the operating range (Table 1). Operating variables potentially inﬂuencing the precipitation conditions were varied (base value and step in parenthesis): (a) Air–CO2 gas mixture ﬂow rate, QG (b.v. = 1000 L h−1 , step 500) (b) CO2 concentration in the inlet gas, cCO2 (b.v. at 25 ◦ C and 1 atm = 5 vol%, step 5) (c) Stirring rate, N (b.v. = 400 rpm, step 300)

2.3. Characterization of solid products The solid product was analyzed to determine total carbon (TC; ELTRA CS-580 Carbon/Sulfur Determinator). Phase/composition identiﬁcation was carried out using X-ray diffraction (XRD), FT-IR spectroscopy and thermal analysis techniques. XRD was performed using a Bruker D8 Advanced instrument. Fourier transform infrared (FT-IR) spectra (Interspec 2020) were acquired using samples prepared as KBr pellets and using a thermoanalyzer (Setaram Setsys 1750) coupled to a FT-IR spectrometer (Nicolet 380). Determination of total sulfur and its bonding forms was carried out according to EVS 664:1995. The crystal morphology of the precipitate particles was monitored during the course of the experiment using a scanning electron microscope (Jeol JSM-8404A). The particle size distribution (PSD) of the ﬁnal product was determined using a laser diffraction analyzer (Beckman Coulter LS 13320). BET-surface area and total and micropore volume were measured using a nitrogen dynamic desorption analysis method (Sorptometer KELVIN 1042). The brightness of the PCC samples was measured according to ISO 2470:1999. 3. Results and discussion 3.1. Reaction mechanism of oil shale ash leachates carbonation process Formation of PCC from lime-containing oil shale ash is an innovative yet complex multi-stage process. Recently, the mechanisms and modeling algorithms for intermediate stages of the process including calcium leaching [3,13], dissolution of gaseous CO2 into the alkaline liquid phase [15], and calcium carbonate precipitation via CO2 absorption into pure lime based model solutions [16] have been reported by the authors. In the present study, a mechanism for the reaction of CO2 with Ca2+ and SO4 2− rich alkaline oil shale ash leachates is proposed. The carbonation process is described by Eqs. (1)–(9) beginning with the physical dissolution of gaseous CO2 into solution: CO2 (q) ↔ CO2 (l)

(1)

The solubility equilibrium follows Henry’s law (at pressures below approximately 5 atm): [CO2 (l)]eq = kH × PCO2

(2)

where kH is the Henry’s law constant and PCO2 is the CO2 partial pressure. Formation of bicarbonate: k11

CO2 (l) + OH− HCO3 −

(3)

k12

Dissociation of bicarbonate: Samples of the suspension were collected through a valve on the reactor body. During the carbonation experiments, the concentrations of Ca2+ , SO4 2− , Cl− , K+ , PO4 3− , CO3 2− , HCO3 − , OH− in the (ﬁltered) liquid phase samples, pH (Mettler Toledo GWB SG2) and conductivity (HI9032) in the reactor, and the CO2 content of the outlet gas ﬂow were continuously monitored. When the pH of the solution had stabilized and the CO2 concentration in the outlet gas became equal to the inlet values, CO2 addition was stopped. Immediately after carbonation, the suspension was ﬁltered (Whatman “blue ribbon” ﬁlter paper) and the resulting solid was dehydrated at 105 ◦ C. The solid material was analyzed as received with no subsequent washing. The synthesis of PCC particles (including the preparation of alkaline mother solution from waste ash) is schematically represented in Fig. 1.

k21

HCO3 − + OH− CO3 2− + H2 O

(4)

k22

Ionization of water: k31

OH− + H+ H2 O

(5)

k32

CO2 hydration [17]: k41

CO2 (l) + H2 O HCO3 − + H+

(6)

k42

Nucleation and growth of CaCO3 crystals: k51

Ca2+ + CO3 2− CaCO3 k52

(7)

O. Velts et al. / Journal of Hazardous Materials 195 (2011) 139–146

a

b Ash

141

CO2 + air QG; cCO2

CO2 analyzer Conductivity/pH meter

VG2 Carbonator VL + VG

Separator

N

Leachate Ash residue

L/S=10 w/w mixing time: 15 min

samples

PCC

Separator

Treated water solution recirculation 50 L/h Fig. 1. Principal experimental scheme: (a) leaching step; (b) carbonation step.

Formation of anhydrate phase: Ca2+ + SO4 2− ↔ CaSO4

(8)

Back-dissolution of CaCO3 crystals at lower pH: k61

CaCO3 + H+ Ca2+ + HCO− 3

• For Ca2+ , OH− , SO4 2− , HCO3 − , CO3 2− , and H+ ions: ∗ d[Ca2+ ] = k52 − k51 [Ca2+ ][CO3 2− ] + kL aCaSO4 × ([SO4 2− ] dt

−[SO4 2− ]) + k61 [H+ ] − k62 [Ca2+ ][HCO3 − ]

(9)

k62

The reactions of other ions present such as K+ , Cl− , and PO4 3− were neglected as their concentrations in the solution remained unchanged during carbonation. It is therefore assumed that they do not take part in the precipitation process in signiﬁcant amounts. 3.2. Modeling of calcium carbonate precipitation from oil shale ash leachates

• For CO2 dissolved in the liquid phase:

d[CO2 (l)] = dt

2

n kH ×MCO2 ×P×[CO2 i (g)] i=1

CO

−

−

2

− [CO2 (l)]

×

VL +VG n

VL −

d[OH− ] = −k11 [CO2 (l)][OH− ] + k12 [HCO3 − ] − k21 [HCO3 − ][OH− ] dt +k22 [CO3 2− ] + k32 − k31 [OH− ][H+ ]

(12)

∗ d[SO4 2− ] = kL aCaSO4 × ([SO4 2− ] − [SO4 2− ]) dt

The model proposed in this paper accounts for absorption and reaction kinetics taking place in the liquid phase (Eqs. (1)–(9)), including formation of the solid product, as well as the hydrodynamic conditions within the system. The concentration proﬁles of all species participating in the precipitation process may be modeled as a function of time using the following differential equations (assuming that the system is operated isothermally at 25 ◦ C):

kL a0CO × E ×

(11)

+

−k11 [CO2 (l)][OH ] + k12 [HCO3 ] − k41 [CO2 (l)] + k42 [HCO3 ][H ]

(13)

d[HCO3 − ] = k11 [CO2 (l)][OH− ] − k12 [HCO3 − ] − k21 [HCO3 − ][OH− ] dt + k22 [CO3 2− ] + k41 [CO2 (l)] − k42 [HCO3 − ][H+ ] + k61 [H+ ] − k62 [Ca2+ ][HCO3 − ]

(14)

d[CO3 2− ] = k21 [HCO3 − ][OH− ] − k22 [CO3 2− ] + k52 dt −k51 [Ca2+ ][CO3 2− ]

(15)

d[H+ ] = k32 − k31 [OH− ][H+ ] + k41 [CO2 (l)] dt −k42 [HCO3 − ][H+ ] − k61 [H+ ] + k62 [Ca2+ ][HCO3 − ] (16)

(10) Table 1 Parameters of the oil shale ash leachates carbonation experiments. Nr

QG (L h−1 )

Air ﬂow rate (L h−1 )

CO2 ﬂow rate (L h−1 )

cCO2 (vol%)

Na (rpm)

1 2 3 4 5 6 7 8 9

1000 1000 1000 1000 2000 2000 1500 2000 2000

950 950 850 850 1900 1900 1350 1700 1700

50 50 150 150 100 100 150 300 300

5 5 15 15 5 5 10 15 15

400 1000 400 1000 400 1000 700 400 1000

a

The stirring rate, as measured experimentally, corresponds to a power consumption 1.1, 2.0 and 3.7 W L−1 for N = 400, 700 and 1000 rpm, respectively.

142

O. Velts et al. / Journal of Hazardous Materials 195 (2011) 139–146

• For CO2 mixture:

exiting

the

ith

section

of

the

reaction

The neutralization rate constant, k31 , was determined by Eigen [21] to be 1.4 × 108 m3 (mol s)−1 . The rate constant k41 for the

QG ([CO2 (g)]IN − [CO2 i (g)]) − kL a0CO × E × ((kH × MCO2 × P × [CO2 i (g)]/CO2 ) − [CO2 (l)]) × (VL + VG )/n d[CO2 i (g)] 2 = dt VG /n • For CO2 exiting the reactor e.g. headspace VG2 above the reaction mixture (see Fig. 1(b)): QG ([CO2 (g)] − [CO2 (g)]OUT ) d[CO2 (g)]OUT = VG2 dt

(18)

• For CaCO3 forming during the carbonation process: d[CaCO3 ] = k51 [Ca2+ ][CO3 2− ] − k52 − k61 [H+ ] dt + k62 [Ca2+ ][HCO3 − ]

reaction between CO2 and water is 0.024 s−1 [23]. The values of the backward reaction rate constants k32 and k42 may be calculated from the equilibrium constants and are equal to k31 /Kw and k41 /K1 . The value of the Henry’s law constant kH (mol (L bar)−1 ) may be expressed as a function of temperature using the equation of Pohorecki and Moniuk [24]: log kH = 9.1229 − 5.9044 × 10−2 T + 7.8857 × 10−5 T 2

(19)

• For CaSO4 forming during the carbonation process: ∗ d[CaSO4 ] = kL aCaSO4 × ([SO4 2− ] − [SO4 2− ] ) dt

(17)

(20)

(25)

The average values of the reaction rate constants k51 and k52 in Eq. (7) were estimated by Velts et al. [16] to be 1.88 × 106 L (mol s)−1 and 0.009 mol (L s)−1 . Based on our study of CO2 uptake kinetics in hydroxide solutions under various process conditions [15], the volumetric CO2 mass transfer coefﬁcients for the system in the absence of chemical reaction kL a0CO (s−1 ) were calculated using an empirical 2

In Eqs. (10)–(20) concentrations are expressed in molar units, QG is the gas volumetric ﬂow rate in L s−1 , kL a0CO is the volumetric mass 2

transfer coefﬁcient of CO2 in the absence of chemical reaction in s−1 , E is the CO2 mass transfer enhancement factor, VL is the solution volume in L, VG is the volume of gas in the gas–liquid mixture in L, VG2 is the gas volume in the reactor headspace in L, kH is the Henry’s Law constant in mol·(L atm)−1 , P is the atmospheric pressure in atm, MCO2 is the CO2 molar mass in g mol−1 , and CO2 is the CO2 gas density in g L−1 . A program feature accounting for changes in VL , VG , and VG2 due to sample collection was implemented in the modeling algorithm. The gas phase in the reaction mixture was divided into a number of theoretical sections n with a volume VG /n (gas phase in approximately plug ﬂow, liquid phase in perfectly mixed ﬂow due to solution recirculation). Each of these sections (high correlation coefﬁcient observed at n = 10) was treated as a non-equilibrium stage governed by Eq. (17). Considering the near inﬁnite-dilution ionic strength of the leachates (I = 0.1), the value of the second-order rate constant k11 (in L (mol s)−1 ) of reaction (3) was calculated as a function of temperature T (K) using a relationship proposed by Pohorecki and Moniuk [18]: log k11 = 11.916 −

2382 T

(21)

The backward reaction rate k12 in Eq. (3) is deﬁned by the value of the equilibrium constant for this reaction (k12 = k11 Kw /K1 ). The value of the solubility product Kw (mol2 m−6 ) is given by Tsonopoulos [19]:

log

Kw

=−

2 w

5839.5 − 22.4773 log(T ) + 61.2062 T

(22)

The value of the equilibrium constant K1 (mol m−3 ) is given as a function of temperature by Edwards et al. [20]:

12092.1

K1 = exp −

T

− 36.786 ln(T ) + 235.482 w

(23)

where w is the density of water (kg m−3 ). The reaction rate constant k21 of reaction (4) was reported as 6 × 106 m3 (mol s)−1 by Eigen [21]. The equilibrium constant K2 (m3 mol−1 ) at inﬁnite dilution that determines the value of the backward reaction rate, k22 = k21 /K2 , is given by Hikita et al. [22]: log(K2 ) =

1568.9 − 2.5866 − 6.737 × 10−3 T T

(24)

equation (R2 = 0.91) applicable to barboter-type reactors:

Q 0.386 P 0.330

kL a0CO = 2.953 × 10−3 × 2

G

N

VL

VL

0.114 cCO 2

(26)

in which PN is the power consumed by the stirrer in watts. The effect of chemical reaction on the process performance was accounted for by introducing the CO2 mass transfer enhancement factor, E. This value was determined using an empirical equation (R2 = 0.97) proposed by Velts et al. [16], where E is a function of the initial Ca2+ concentration (mmol L−1 ): 2

E = 0.0027 × [Ca2+ ]0 + 0.0224 × [Ca2+ ]0 + 1.0

(27)

Based on the experimental data obtained in this study, the SO4 2− dynamic equilibrium concentration [SO4 2− ]* (mmol L−1 ) was calculated using an empirical equation (R2 = 0.98) dependent on the operating parameters and the initial concentrations of Ca2+ and SO4 2− ions (mmol L−1 ) in the leachate: ∗

[SO4 2− ] = 0.761 × [SO4 2− ]0 ×

Q 0.066 G

VL

0.976

cCO2 0.038

[Ca2+ ]0

−0.073

P −0.012 N

(28)

VL

The volumetric mass transfer coefﬁcient of anhydrite, kL aCaSO4 and the reaction rate constants k61 and k62 in Eq. (9) were evaluated from the differential equations (10)–(20). The set of model equations was solved by means of linear multi-step methods implemented in ODESSA, which is based on the LSODE software [25]. The calculations were performed using the MODEST 6.1 software package [26] designed for various model-building tasks such as simulation, parameter estimation, sensitivity analysis, and optimization. The software consists of a FORTRAN 95/90 library of objective functions, solvers, and optimizers linked to model problem-dependent routines and the objective function. Based on the estimated values of kL aCaSO4 (s−1 ), an empirical equation (R2 = 0.8) applicable to barboter-type reactors was proposed as a function of the main process parameters: kL aCaSO4 = 1.95 × 10−7 ×

Q 1.702 G

VL

cCO2 1.134

P 0.126 N

VL

(29)

The average values of the reaction rate constants k61 and k62 were estimated to be 0.1 (±0.021)×107 s−1 and 0.4 (±0.013)×103 L (mol s)−1 . The correlation coefﬁcients for all data sets were greater than 0.93. The reaction rate constants used in Eqs. (3)–(9) and other parameters used in this paper (T = 298.1 K) are summarized in Table 2.

O. Velts et al. / Journal of Hazardous Materials 195 (2011) 139–146

143

Fig. 2. Modeling of ash leachate carbonation process accompanied by the formation of PCC at (a): QG = 2000 L h−1 , cCO2 = 5 vol%; N = 1000 rpm; (b): QG = 1000 L h−1 , cCO2 = 15 vol%; N = 1000 rpm; (c): QG = 1500 L h−1 , cCO2 = 10 vol%; N = 700 rpm; (d): QG = 2000 L h−1 , cCO2 = 15 vol%; N = 400 rpm: experimental vs. simulated Ca2+ (), SO4 2− (), OH− (), CaCO3 (䊉), CaSO4 (♦), HCO3 − (), CO2 (l) concentration (mmol L−1 ) and pH () proﬁles.

The model was veriﬁed by comparing the predictions of concentration changes for the reactive species (Ca2+ , OH− , SO4 2− , CaCO3 , CaSO4 , HCO3 − , CO2 , H+ , and CO3 2− ) with the experimental data. Plots of experimental and simulated concentration proﬁles corresponding to experiments 4, 6, 7, and 8 (Table 1) are provided in Fig. 2. The relatively small deviations between the measured and estimated data conﬁrm the ability of the proposed model to quite accurately describe the process course including re-dissolution of PCC due to increased solubility of CaCO3 at lower pH. It is also worth emphasizing that the model enables the prediction of pH (Fig. 2(a)), which suggests potential applications in wastewater neutralization process design.

Table 2 Parameters used in the modeling of oil shale ash leachates carbonation at 298 K. Parameter k11 k12 k21 k22 k31 k32 k41

(L (mol s)−1 ) (s−1 ) (L (mol s)−1 ) (s−1 ) (L (mol s)−1 ) (mol·(L s)−1 ) (s−1 )

Value

Parameter

Value

8.4 × 103 2.0 × 10−4 6.0 × 109 1.2 × 106 1.4 × 1011 1.3 × 10−3 2.4 × 10−2

k42 (L (mol s)−1 ) k51 av (L (mol s)−1 ) k52 av (mol (L s)−1 ) k61 av (s−1 ) k62 av (L (mol s)−1 ) kH (mol (L atm)−1 ) CO2 (25 ◦ C) (kg m−3 )

5.7 × 104 1.9 × 106 9.0 ×10−3 0.1 × 107 0.4 × 103 3.5 × 10−2 1.8 × 100

3.3. Characterization of PCC crystallized from oil shale ash leachates Among other parameters, the shape, size, and texture of crystals play a crucial role in determining the properties and application suitability of a material. For this reason, a detailed characterization of the ﬁnal precipitates formed during ash leachate carbonation under different conditions was performed. The characteristics of samples PCC1–PCC9 were determined using numerous characterization techniques (see Section 2.3) and are presented in Table 3. The unwashed precipitates were a bright white color with a ﬁne and powdery texture. The brightness value (∼93%) exceeded that of PCC (∼89%) obtained from pure lime under the same conditions [9]. Total carbon (TC) analysis indicated that the solid samples predominantly contained CaCO3 (∼94.3–96.2%), with minor amounts of CaSO4 (∼4–6%), evidently adsorbed on the surface of the CaCO3 crystals (Table 3). Washing of the precipitate cake would be expected to improve the purity of the solid product by a few percentage points. The phase composition was also conﬁrmed using FT-IR spectroscopy. The morphology of the precipitated particles was examined using scanning electron microscopy (SEM). Fig. 3 contains SEM images of the ﬁnal precipitates PCC1–PCC9 crystallized under various carbonation conditions (Table 3). Under the conditions in experiments 1–5 (Table 1), well-deﬁned rhombohedral

144

O. Velts et al. / Journal of Hazardous Materials 195 (2011) 139–146

Fig. 3. SEM micrographs of PCC samples (a) PCC1, (b) PCC2, (c) PCC3, (d) PCC4, (e) PCC5, (f) PCC6, (g) PCC7, (h) PCC8, (i) PCC9 formed via oil shale ash leachate carbonation under experimental conditions presented in Table 1.

O. Velts et al. / Journal of Hazardous Materials 195 (2011) 139–146

145

Table 3 Synthesis conditions and main characteristics of oil shale ash leachates carbonation products. Sample

Operating variables QG (L h−1 )

cCO2 (vol%)

Solid product characteristics N (rpm)

CaCO3 av (%) TC

1000 1000 1000 1000 2000 2000 1500 2000 2000

5 5 15 15 5 5 10 15 15

400 1000 400 1000 400 1000 700 400 1000

94.5 94.5 94.6 94.4 94.4 95.1 95.6 96.2 95.4

SSA (m2 g−1 )

Vtot (mm3 g−1 )

Vmicro (mm3 g−1 )

Dmean (␮m)

Brightness ISO (%)

XRD calcite

PCC1 PCC2 PCC3 PCC4 PCC5 PCC6 PCC7 PCC8 PCC9

CaSO4 av (%)

vaterite

100.0

97.3 77.2 63.2 95.7

2.6 21.9 36.3 3.6

5.7 5.8 5.5 5.9 5.9 5.0 4.4 4.0 4.5

crystals with a mean size ranging from ∼4 to 8 ␮m were produced (Fig. 3(a–e)). X-ray powder diffraction analysis (XRD) of these carbonated samples (PCC1–PCC5) identiﬁed calcite as the only crystal form of calcium carbonate detected. Carbonation under the intensiﬁed hydrodynamic conditions (experiments 6–8, Table 1) resulted in formation of distinctly spherical particles in the precipitates along with the rhombohedral crystals of calcium carbonate (Fig. 3(f–h)). Initial analysis of these images indicated that the rhombohedral crystal was calcite, while the calcium carbonate microspheres were forms of vaterite. The coexistence of calcite and vaterite in the product (PCC6–PCC9) was conﬁrmed using XRD measurements (Table 3), which enabled us to distinguish different morphologies of PCC. The XRD results indicated that sample PCC6 contained only small amount of the vaterite phase (2.6%), while the relative mass percentages of vaterite in samples from experiments 7 and 8 were ∼22 and 36% (Table 3). The presence of a signiﬁcant amount of spherical vaterite could explain the greater surface area and pore volume in samples PCC7 and PCC8. The shape and surface observations conﬁrmed the results of the particle size distribution analysis. Interestingly, precipitation under the most rapid conditions (experiment 9, Table 1) decreased the amount of vaterite in the product, leading to formation of pseudo-cubic or randomly aggregated rhombohedral (Fig. 3(i)) and spherical structures with a mean diameter of ∼10 ␮m and a calcite content of ∼96% (PCC9, Table 3). The results suggest that the carbonation conditions may direct the morphology of the CaCO3 crystals and indicate that the coexisting vaterite polymorph can be stabilized under speciﬁc experimental conditions. Whether the vaterite phase is formed prior to or during the PCC re-dissolution stage is a matter of great interest and requires further investigation. A closer examination of the morphological development of the CaCO3 crystals at different crystallization times will be undertaken. 4. Conclusion In this study, modeling, simulation, and experimental results describing the carbonation of leachates from oil shale ash are presented. This work is part of our effort to develop a promising calcium carbonate production process employing an abundant waste material. A mathematical model of the multi-step PCC formation process incorporating mechanisms of CO2 dissolution and CaCO3 and CaSO4 precipitation was introduced. The model provided results that were in good agreement with experimental data, conﬁrming its accuracy. The modeling algorithm presented in this paper may be applied to design, energetic, and economic assessment of PCC pilot plants using oil shale ash or other lime-containing wastes or calcium-rich wastewaters as feedstock.

2.28 2.56 3.15 1.33 1.35 2.38 4.70 7.29 1.95

3.38 3.51 4.03 1.93 1.74 3.77 11.05 17.44 3.44

– 0.11 – – – – 0.01 0.43 –

4.1 5.1 4.8 7.8 6.5 8.1 7.7 8.0 9.9

93.2

92.7 92.3

Carbonation of oil shale ash leachates resulted in precipitation of high brightness PCC containing up to ∼96% CaCO3 with mean particle sizes ranging from 4 to 10 ␮m. Depending on the carbonation conditions, formation of rhombohedral calcite crystals or co-precipitation of calcite and spherical vaterite structures occurred, suggesting control over CaCO3 crystallization and the ability to construct crystals with a desired morphology. The PCC morphogenesis will be further investigated to determine the relationship between formation conditions and morphology. A description of the carbonation reaction mechanism and the properties of the precipitated product are important for understanding and estimating the potential reusability of alkaline wastes associated with oil shale-based power production. According to simpliﬁed calculations, 1 tonne of ash (containing ∼20% of free lime on the average) would allow producing near to 360 kg of CaCO3 , while via carbonation of 1 m3 of leachates at least 1.3 kg of CO2 can be captured and up to 3 kg of PCC formed. Due to availability of enormously large amounts (10–15 million m3 ) of highly alkaline ash leachates in the proximity of CO2 emission source, the direct capture and storage of CO2 -containing ﬂue gas by leachates could further improve the technology. Hence, oil shale energetics could beneﬁt from this innovative process by utilizing and valorizing its own waste-products into a valuable commodity, lowering the environmental impact of deposited waste material, alkaline leachates and CO2 emissions at the same time. Acknowledgements The ﬁnancial support of the Estonian Ministry of Education and Research (SF0140082s08) and the Estonian Science Foundation (Grant No. 7379) are gratefully acknowledged. The assistance of Prof. Kalle Kirsimäe and Dr. Valdek Mikli in performance of XRD and SEM measurements is highly appreciated. Authors are also grateful for the contribution of Esko Kukkamäki (UPM-Kymmene Corporation). This work has been partially supported by graduate school “Functional materials and processes” receiving funding from the European Social Fund under project 1.2.0401.09-0079 in Estonia. References [1] Carbon Dioxide Information Analysis Center, Environmental Sciences Division (US), CO2 emissions—Estonia. Available at: http://data.worldbank. org/indicator/EN.ATM.CO2E.PC/countries/EE-7E-XR (accessed May 2011). [2] A. Ots, Oil Shale Fuel Combustion, Tallinn University of Technology Press, Tallinn, 2006. [3] O. Velts, M. Hautaniemi, J. Kallas, M. Kuosa, R. Kuusik, Modeling calcium dissolution from oil shale ash: part 2. Continuous washing of the ash layer, Fuel Process. Technol. 91 (5) (2010) 491–495. [4] R. Mõtlep, T. Sild, E. Puura, K. Kirsimäe, Composition, diagenetic transformation and alkalinity potential of oil shale ash sediments, J. Hazard. Mater. 184 (1–3) (2010) 567–573.

146

O. Velts et al. / Journal of Hazardous Materials 195 (2011) 139–146

[5] A.S.M. Ribeiro, R.C.C. Monteiro, E.J.R. Davim, M.H.V. Fernandes, Ash from a pulp mill boiler—characterisation and vitriﬁcation, J. Hazard. Mater. 179 (1–3) (2010) 303–308. [6] C. Ferreira, A. Ribeiro, L. Ottosen, Possible applications for municipal solid waste ﬂy ash, J. Hazard. Mater. 96 (2–3) (2003) 201–216. [7] R. Ciofﬁ, M. Marroccoli, L. Sansone, L. Santoro, Potential application of coal–fuel oil ash for the manufacture of building materials, J. Hazard. Mater. 124 (1–3) (2005) 101–106. [8] Yu-Fen Yang, Guo-Sheng Gai, Zhen-Fang Cai, Qing-Ru Chen, Surface modiﬁcation of puriﬁed ﬂy ash and application in polymer, J. Hazard. Mater. 133 (1–3) (2006) 276–282. [9] O. Velts, M. Uibu, J. Kallas, R. Kuusik, Prospects in waste oil shale ash sustainable valorization, World Acad. Sci. Eng. Technol. 76 (2011) 451–455. [10] M. Uibu, M. Uus, R. Kuusik, CO2 mineral sequestration in oil-shale wastes from Estonian power production, J. Environ. Manage. 90 (2) (2009) 1253–1260. [11] M. Uibu, O. Velts, R. Kuusik, Developments in CO2 mineral carbonation of oil shale ash, J. Hazard. Mater. 174 (1–3) (2010) 209–214. [12] R. Kuusik, M. Uus, M. Uibu, et al., Method for neutralization of alkaline waste water with carbon dioxide included in ﬂue gas, Patent nr EE05349B1. [13] O. Velts, M. Hautaniemi, J. Kallas, R. Kuusik, Modeling calcium dissolution from oil shale ash: part 1. Ca dissolution during ash washing in a batch reactor, Fuel Process. Technol. 91 (5) (2010) 486–490. [14] O. Velts, M. Uibu, I. Rudjak, J. Kallas, R. Kuusik, Utilization of oil shale ash to prepare PCC: leachibility dynamics and equilibrium in the ash–water system, Energy Procedia 1 (1) (2009) 4843–4850. [15] O. Velts, M. Hautaniemi, M. Uibu, J. Kallas, R. Kuusik, Modelling of CO2 mass transfer and hydrodynamics in a semi-batch reactor, J. Int. Sci. Publ. Mater. Methods Technol. 4 (2) (2010) 68–79. Available at: http://www.sciencejournals.eu/mmt/index.html (accessed May 2011).

[16] O. Velts, M. Uibu, J. Kallas, R. Kuusik, CO2 mineral trapping: modeling of calcium carbonate precipitation in a semi-batch reactor, Energy Procedia 4 (2011) 771–778. [17] A.H.G. Cents, D.W.F. Brilman, G.F. Versteeg, CO2 absorption in carbonate/bicarbonate solutions: the Danckwerts-criterion revisited, Chem. Eng. Sci. 60 (2005) 5830–5835. [18] R. Pohorecki, W. Moniuk, Calculation of the rate constant for the reaction of carbon dioxide with hydroxyl ions in mixed electrolyte solutions, Rep. Inst. Chem. Eng. Warsaw. Tech. Univ. 5 (1976) 179–192. [19] C. Tsonopoulos, Ionization constants of water pollutants, J. Chem. Eng. Data 21 (1976) 190–193. [20] T.J. Edwards, G. Maurer, J. Newman, J.M. Prausnitz, Vapor–liquid equilibria in multicomponent aqueous solution of volatile weak electrolytes, AIChE J. 24 (1978) 966–976. [21] M. Eigen, Protonenübertagung, säure-base-katalyse und enzymatische hydrolyse. teil I: Elementarvorgänge, Angew. Chem. 75 (1963) 489–508. [22] H. Hikita, S. Asai, T. Takatsuka, Absorption of carbon dioxide into aqueous sodium hydroxide and sodium bicarbonate solutions, Chem. Eng. J. 11 (1976) 131–141. [23] P.V. Danckwerts, M.M. Sharma, Absorption of carbon dioxide into solutions of alkalis and amines, Chem. Eng. (London) 10 (1966) 244–280. [24] R. Pohorecki, W. Moniuk, Kinetics of reaction between carbon dioxide and hydroxyl ions in aqueous electrolyte solutions, Chem. Eng. Sci. 43 (1988) 1677–1684. [25] A.C. Hindmarsh, ODEPACK, a systematized collection of ODE solvers, Sci. Comput. IMACS Trans. Sci. Comput. 1 (1983) 55–64. [26] H. Haario, Modest user manual, Profmath Oy, Finland, 1994.

Journal of Hazardous Materials 195 (2011) 147–154

Contents lists available at ScienceDirect

Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat

Post-treatment of anaerobically degraded azo dye Acid Red 18 using aerobic moving bed bioﬁlm process: Enhanced removal of aromatic amines E. Hosseini Koupaie a , M.R. Alavi Moghaddam a,∗ , S.H. Hashemi b a b

Civil and Environmental Engineering Department, Amirkabir University of Technology (AUT), Hafez Ave., Tehran 15875-4413, Iran Environmental Science Research Institute, Shahid Beheshti University, Tehran, Iran

a r t i c l e

i n f o

Article history: Received 28 December 2010 Received in revised form 28 July 2011 Accepted 7 August 2011 Available online 12 August 2011 Keywords: Azo dye Aromatic amine Biodegradation Mineralization Moving bed sequencing batch bioﬁlm reactor

a b s t r a c t The application of aerobic moving bed bioﬁlm process as post-treatment of anaerobically degraded azo dye Acid Red 18 was investigated in this study. The main objective of this work was to enhance removal of anaerobically formed the dye aromatic metabolites. Three separate sequential treatment systems were operated with different initial dye concentrations of 100, 500 and 1000 mg/L. Each treatment system consisted of an anaerobic sequencing batch reactor (An-SBR) followed by an aerobic moving bed sequencing batch bioﬁlm reactor (MB-SBBR). Up to 98% of the dye decolorization and more than 80% of the COD removal occurred anaerobically. The obtained results suggested no signiﬁcant difference in COD removal as well as the dye decolorization efﬁciency using three An-SBRs receiving different initial dye concentrations. Monitoring the dye metabolites through HPLC suggested that more than 80% of anaerobically formed 1-naphthylamine-4-sulfonate was completely removed in the aerobic bioﬁlm reactors. Based on COD analysis results, at least 65–72% of the dye total metabolites were mineralized during the applied treatment systems. According to the measured bioﬁlm mass and also based on respiration–inhibition test results, increasing the initial dye concentration inhibited the growth and ﬁnal mass of the attachedgrowth bioﬁlm in MB-SBBRs. © 2011 Elsevier B.V. All rights reserved.

1. Introduction Azo dyes are the largest class (60–70%) of synthetic dyes in the textile, food, rubber, plastic, paper, and cosmetic industries [1]. The release of colored wastewaters in the environment even at low concentrations of water soluble azo dyes (10–50 mg/L) not only is a matter for aesthetic point of view, but also leads to the reduction in sunlight penetration diminishing the photosynthesis and oxygen solubility [2]. Moreover, both mutagenic and carcinogenic effects of several azo dyes and their intermediates have been reported so far [3,4]. Various physicochemical processes including electrochemical [5], adsorption [6], chemical coagulation/ﬂocculation [7], advanced oxidation [8] and photocatalysis [9] have been effectively used for treatment of azo dye-containing wastewaters. However, most physicochemical dye removal methods are quite expensive and energy consuming [10] and usually generate large amounts of sludge which require safe disposal [10,11]. These methods also interfere with other wastewater constituents [3] and in some cases

∗ Corresponding author. Tel.: +98 912 2334600; fax: +98 21 66414213. E-mail addresses: [email protected] (E. Hosseini Koupaie), [email protected], [email protected] (M.R. Alavi Moghaddam), h [email protected] (S.H. Hashemi). 0304-3894/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.jhazmat.2011.08.017

generate toxic by-products which are difﬁcult to dispose [2]. Therefore, biological azo dye removal methods as the environmentally friendly and cost-competitive alternatives to the physicochemical degradation processes have been considered in several researches [12–14]. Due to the large degree of complex aromatics present in azo dyes molecules and also strong electron-withdrawing property of the azo group (–N N–), most azo dyes are recalcitrant to the conventional aerobic treatment [15]. Combined anaerobic–aerobic biological processes have been studied for treatment of wastewaters containing different azo dyes such as monoazo acid orange 7, 8 and 10; hydrolyzed and non-hydrolyzed reactive black 5 and diazo reactive red 141 [16–19]. In the case of azo dye Acid Red 18 (AR18), the previous researches are limited to the study accomplished by FitzGerald and Bishop. They utilized a two stage reactor system which consisted of an anaerobic ﬁxed-ﬁlm ﬂuidized bed reactor followed by a conventional aerobic reactor for treatment of a wastewater containing low concentration (10 mg/L) of AR18 [20]. In two-stage anaerobic–aerobic processes, the reductive cleavage of the azo bond occurs in the anaerobic stage resulting in formation of the dye aromatic intermediates [4], while further mineralization of these intermediates is expected in the aerobic stage, [3,21]. Although the intermediate metabolites residue from anaerobic degradation of azo dyes can theoretically be mineralized aerobically, previous studies have shown that several aromatic amines

148

E. Hosseini Koupaie et al. / Journal of Hazardous Materials 195 (2011) 147–154

such as aminobenzene sulfonates and naphthylamine sulfonates are resistant to degradation even under the aerobic conditions [10]. Aside from the mineralization difﬁculties, the autoxidation of azo dyes breakdown metabolites which leads to the formation of more aerobically recalcitrant products, has been mentioned by some researchers [17,22]. Therefore there is still a need to research on modiﬁed biological treatment techniques to achieve complete biodegradation and mineralization of azo dyes and their aromatic metabolites. The attached-growth bioﬁlm systems have shown to be more drastic than suspended-growth processes for the removal of compounds which are difﬁcult to degrade [23]. It has also been reported that the bioﬁlm cells are more resistant to toxicity than freely suspended ones [24]. Moving bed bioﬁlm reactor (MBBR) as one of the attached-growth bioﬁlm systems was introduced about 15 years ago in order to offer the advantages of former bioﬁlm processes without their limitation including head loss, clogging and hydraulic instability [25]. MBBRs have been efﬁciently used for treatment of different municipal and industrial wastewaters during the last decade [25–27]. Recently, some reports have been published on successful application of the MBBR process (individually or in combination with other treatment methods) in biodegradation of some aromatic compounds such as aniline [28], phenol [29] and polycyclic aromatic hydrocarbons [30] and also for treatment of wastewaters containing aerobically recalcitrant compounds such as pesticides [31]. This indicates that MBBR can be a good option to be applied as post-treatment of anaerobically degraded azo dyes, which to our knowledge has not been considered so far. The main objective of the present study was to investigate the performance of moving bed sequencing batch bioﬁlm reactor (MBSBBR) as post-treatment of anaerobically degraded azo dye Acid Red 18 (AR18) in order to enhance the removal breakdown aromatic amines. For this purpose, the anaerobic–aerobic degradation of the dye was studied and the concentration of 1-naphthylamine4-sulfonate (1N-4S) as one of the main aromatic constituents of azo dye AR18 was monitored through the applied treatment systems. In addition, the change of attached-growth bioﬁlm mass during the operation period as well as the bioﬁlm morphology was investigated.

Table 1 General characteristics of C.I. Acid Red 18 (AR18). Parameter

Value

Molecular formula Molecular weight COD of 1 g-AR18/L max

C20 H11 N2 Na3 O10 S3 604.5 (g/mol) 597 ± 17 (mg/L) 507 (nm)

Chemical structure

to 100/2/0.3 in the feed solution of An-SBRs. In order to keep the COD/N/P ratio favorable for the aerobic bioﬁlm process, additional nutrients (N and P) were added to the MB-SBBRs at the beginning of each aerobic reaction cycle. The azo dye C.I Acid Red 18 (AR18) was obtained from Alvan Sabet Company (Tehran, Iran) and used without further puriﬁcation. The general characteristics of the dye AR18 are listed in Table 1. 2.3. Operation of lab-scale treatment systems

2. Materials and methods

Granulated anaerobic sludge was obtained from a full scale UASB reactor treating a dairy factory wastewater (Pegah Dairy Company, Tehran, Iran) and used as seed in An-SBRs. The MB-SBBRs were inoculated with the activated sludge taken from a municipal wastewater treatment plant (Zargandeh, Tehran, Iran). A 24 h operation cycle of both anaerobic and aerobic reactors consisted of ﬁve phases including ﬁlling, reaction, settling, draw and idle which was controlled by a digital timer. In the ﬁlling phase, 2 liters of new SWW were supplied to each An-SBR and the efﬂuent of An-SBRs was passed to the MB-SBBRs by gravity. The complete mixing condition in the anaerobic reactors was provided by a low speed (100 rpm) gear motor driving two paddle-shaped impellers. In MB-SBBRs, an electromagnetic air pump (RESUN; ACO-006, China) was used for supplying air and keeping dissolved oxygen concentration above 3 mg/L during the reaction phase. The main operating parameters of the laboratory treatment systems are listed in Table 2.

2.1. Reactors conﬁguration

2.4. Analytical methods and procedures

The study was carried out using three separate lab-scale sequential anaerobic–aerobic treatment systems (treatment system 1, 2 and 3). As shown in Fig. 1, each treatment system consisted of an anaerobic sequencing batch reactor (An-SBR) followed by an aerobic moving bed sequencing batch bioﬁlm reactor (MB-SBBR). The reactors were made of plexiglas having an inner diameter of 14 cm and height of 50 cm. A type of plastic bioﬁlm carrier (2H-BCN017KL, Germany) was used for biomass immobilization in MB-SBBRs. About 50 percent of the effective volume of MB-SBBRs was ﬁlled with these carriers and a coarse bubble aeration system made them thoroughly immersed. 2.2. Composition of synthetic wastewater

2.4.1. Dye and COD analysis The concentration of the dye was determined by measuring the absorbance of the test samples at the maximum absorbent wavelength of AR18 (max : 507 nm) using UV–Vis spectrophotometer (DR 4000, HACH, USA). The limit of detection (LOD) for UV–Vis analysis was 0.3 mg AR18/L. Before the analysis, the samples withdrawn from the treatment systems were centrifuged at 6000 rpm for 10 min. The qualitative information related to the decolorization of AR18 as well as formation of the dye metabolic intermediates was determined by scanning of complete spectrum from 200 to 800 nm. Chemical oxygen demand (COD) analysis was carried out using colorimetric method according to the Standard Methods [32]. All experiments were conducted at room temperature (22 ± 2 ◦ C).

The composition of synthetic wastewater (SWW) was as follows: glucose (1.5 g/L), lactose (1.5 g/L), urea (116.5 mg/L), KH2 PO4 (23.3 mg/L), K2 HPO4 (30 mg/L) and NaHCO3 (1.5 g/L) supplied with different concentration of AR18 including 100, 500 and 1000 mg/L as the feed solution of An-SBR1, An-SBR2 and An-SBR3, respectively. The SWW was prepared in tap water and the chemicals were analytical grade (Merck, Germany). The COD/N/P ratio was adjusted

2.4.2. Abiotic adsorption test To investigate the contribution of abiotic conditions to decolorization, the anaerobic sludge was sterilized in autoclave (20 min at 121 ◦ C and 0.103 MPa). The autoclaved sludge was added to the 500 mL of SWW solution containing 100 mg-AR18/L plus other auxiliary substrates similar to the composition of feed solution in An-SBRs. The mixture was anaerobically incubated for 24 h with

E. Hosseini Koupaie et al. / Journal of Hazardous Materials 195 (2011) 147–154

149

Table 2 Operating parameters of the applied treatment systems. Parameter

Inﬂuent dye (mg AR18/L) Inﬂuent COD (mg/L) OLR (mg COD/L.d) Inﬂuent BAa (mg CaCO3 /L) Effective volume (L) Fill (min) Reaction (h) Settling (h) Draw (h) Idle (min) Hydraulic retention time (d) Temperature (◦ C) a

Treatment system 1

Treatment system 2

Treatment system 3

An-SBR1

MB-SBBR1

An-SBR2

MB-SBBR2

An-SBR3

MB-SBBR3

100 3040 ± 90 1105 ± 33 1163 ± 96 5.5 20 21 2 0.5 10 2.75 35 ± 0.2

– – – – 5.5 20 22.5 0.5 0.5 10 2.75 22 ± 2

500 3337 ± 75 1213 ± 27 1280 ± 107 5.5 20 21 2 0.5 10 2.75 35 ± 0.2

– – – – 5.5 20 22.5 0.5 0.5 10 2.75 22 ± 2

1000 3620 ± 105 1316 ± 38 1148 ± 83 5.5 20 21 2 0.5 10 2.75 35 ± 0.2

– – – – 5.5 20 22.5 0.5 0.5 10 2.75 22 ± 2

BA, bicarbonate alkalinity.

Fig. 1. Schematic of the experimental anaerobic–aerobic treatment systems used in this study.

a constant mixing speed of 100 rpm at 35 ◦ C. After the incubation period, the UV–Vis absorption was used as the measure of abiotic decolorization.

2.4.3. Respiration–inhibition study The respiration–inhibition test was performed to assess the inhibitory effect of An-SBRs efﬂuent containing the dye intermediates on the activity of aerobic microorganisms in MB-SBBRs. For this purpose, the speciﬁc oxygen uptake rate (SOUR) was measured for 400 mL of the mixed liquor samples withdrawn immediately after the start of the aeration phase in MB-SBBRs. The SOUR tests were accomplished according to the procedure outlined in the Standard Methods [32].

2.4.4. Bioﬁlm mass measurement The mass of attached-growth bioﬁlm in MB-SBBRs was determined after detachment of the bioﬁlm from the surfaces of sample moving carriers (15 pieces) and weighing the dried biomass. The total attached-growth bioﬁlm mass in MB-SBBRs was calculated as: Attached − growth bioﬁlm mass (mg) =

n (m1 − m2 ) 15

(1)

where m1 and m2 are the mass of the sample carriers before and after the bioﬁlm washout (mg), respectively and n is the total number of the bioﬁlm carriers in each MB-SBBR. All the moving carriers were freely immersed in the whole effective volume of the reactors under complete mixing conditions during the aeration phase.

150

E. Hosseini Koupaie et al. / Journal of Hazardous Materials 195 (2011) 147–154

Fig. 2. UV–Vis spectral analysis: An-SBR3 inﬂuent (I); An-SBR3 efﬂuent (II) and MBSBBR3 efﬂuent (III). Fig. 3. HPLC chromatogram of the standard solution.

Therefore, the distribution of the attached-growth bioﬁlm and the suspended biomass were uniform in the MB-SBBRs. 2.4.5. High performance liquid chromatography (HPLC) HPLC analysis was carried out using an Agilent 1200 chromatograph equipped with a variable wavelength detector (190–600 nm). A 5 ␮m C18 column (1.6 mm × 250 mm) was connected for analytical, reverse-phase separation. The mobile phase was a gradient started with 91% water, 5% acetonitrile and 4% methanol. The gradient changed linearly to 69% water, 27% acetonitrile and 4% methanol over 25 min. The detection was performed at 254 nm. The limit of detection for monitoring the concentration of 1naphthylamine-4-sulfonate (1N-4S) was 0.06 mg 1N-4S/L. 2.4.6. Scanning electron microscopy (SEM) In order to observe the attached-growth bioﬁlm morphology, the micrograph images were taken by a digital scanning electron microscope (Philips-XL30; Holland) applying 25 kV accelerating voltage.

tration, the peaks are more visible than those of treatment system 1 and 2. The disappearance of the absorbance peak at 507 nm indicates the complete decolorization of AR18. Considering a selective absorbance decrease only in the visible region and the change of UV–Vis absorption pattern after the treatment process (Fig. 2), it can be concluded that the anaerobic decolorization was caused by the azo bond degradation. The biodegradation of AR18 was also conﬁrmed by the abiotic adsorption test which demonstrated that the dye removal efﬁciency has been only 5–7% due to the adsorption of the dye into the inactivated anaerobic cells. It is noteworthy that the anaerobic efﬂuents exhibited two strong absorbance peaks at around 263 and 325 nm (Fig. 2), while, the intensity of these peaks was signiﬁcantly diminished after the aerobic treatment in MB-SBBRs. This indicates the possible ability of the applied bioﬁlm process to decompose the anaerobically formed the dye aromatic metabolites.

3. Results and discussion

3.2. Monitoring of the dye metabolites through HPLC

3.1. Anaerobic–aerobic degradation of AR18

The HPLC chromatogram of the standard solution containing AR18 (100 mg/L) and 1N-4S (100 mg/L) is presented in Fig. 3. Two peaks are respectively observed at retention times (RT) of 1.99 and 10.8 min related to 1N-4S and AR18. The results of HPLC analysis of the samples extracted from treatment system 1, 2 and 3 are presented in Figs. 4–6, respectively. It should be noted that in addition to the reactors efﬂuents, the HPLC analysis was also carried out on the samples extracted immediately after the complete anaerobic decolorization of AR18. According to the kinetic study tests, complete anaerobic decolorization was obtained after 8, 13 and 16 h from the start of anaerobic phase in An-SBR1, 2 and 3, respectively. All the HPLC chromatograms of the samples taken from the An-SBRs (Figs. 4a, 5a and 6a) showed a peak with the retention time almost similar to 1.99 min (2.02 min). Accordingly, 1N-4S was identiﬁed as one of the dye intermediates formed during the anaerobic AR18 degradation. The appearance of different peaks in the HPLC chromatograph of the An-SBRs efﬂuent (Figs. 4b, 5b and 6b) as well as the high AR18 decolorization efﬁciency achieved anaerobically (Table 3) shows that the inﬂuent dye was signiﬁcantly degraded to its metabolic intermediates. In other words, the applied anaerobic process in An-SBRs improved the biodegradability of AR18 for further aerobic treatment by transforming the original dye AR18 to its metabolites that could be mineralized aerobically. The results reported by Sponza and Is¸ik [33] and An et al. [34] also prove the positive effect

The general experimental data including the efﬂuent and the bio-sludge properties of the reactors are summarized in Table 3. According to Table 3, up to 98% of the AR18 decolorization and more than 80% of the COD removal occurred in the An-SBRs. To realize whether the differences observed among three An-SBRs in decolorization and COD removal efﬁciency are meaningful or not, the experimental data were compared using one-way ANOVA (95% conﬁdence interval). The results (not presented) showed that there was no statistically signiﬁcant difference among three An-SBRs in the case of decolorization (P = 0.885) as well as the COD removal efﬁciency (P = 0.136). Since the AR18 decolorization and also the COD removal efﬁciency were not affected by the initial dye concentration, it can be inferred that the dye or its breakdown metabolites had no inhibition effects on performance of the An-SBRs. The complete biological removal of azo dyes occurs in a twostage anaerobic–aerobic process. As shown in Eq. (2), the ﬁrst stage (anaerobic) involves the reductive cleavage of the azo bond resulting in production of aromatic compounds which are expected to be mineralized in the second stage (aerobic) [16]. 4H+ + 4e− + R1 − N = N − R2 → R1 − NH2 + R2 − NH2

(2)

Fig. 2 shows the UV–Vis spectrum of the samples taken from treatment system 3 in which due to the higher initial dye concen-

E. Hosseini Koupaie et al. / Journal of Hazardous Materials 195 (2011) 147–154

151

Table 3 The general experimental data of the applied treatment systems. Experimental dataa

Treatment system 1 An-SBR1

Efﬂuent COD (mg/L) (n = 8) COD removal (%) AR18 decolorization (%) (n = 8) b MLSS (mg/L) (n = 8) MLVSS/MLSS (%) (n = 8) Efﬂuent pH (n = 8) Efﬂuent BA c (mg CaCO3 /L) (n = 4) a b c

532 82.5 98.1 7646 74.9 7.5 1113

± ± ± ± ± ± ±

51 1.7 1.4 318 3.1 0.4 41

Treatment system 2 MB-SBBR1

An-SBR2

17 ± 3 16.9 ± 0.1
625 81.3 98.0 8289 72.0 7.3 1246

± ± ± ± ± ± ±

31 0.9 0.6 301 3.3 0.4 55

Treatment system 3 MB-SBBR2

An-SBR3

84 ± 17 16.1 ± 0.5
672 81.4 97.9 8763 72.5 7.3 1088

± ± ± ± ± ± ±

40 1.1 0.4 418 4.0 0.3 62

The average values were determined from the steady-state data obtained during the last 30 days of the operation period. LOD, limit of detection. BA, bicarbonate alkalinity.

Fig. 4. HPLC chromatograms: (a) 8 h after the start of anaerobic phase; (b) An-SBR1 efﬂuent and (c) MB-SBBR1 efﬂuent.

Fig. 5. HPLC chromatograms: (a) 13 h after the start of anaerobic phase; (b) An-SBR2 efﬂuent and (c) MB-SBBR2 efﬂuent.

MB-SBBR3 210 ± 30 12.8 ± 0.8
152

E. Hosseini Koupaie et al. / Journal of Hazardous Materials 195 (2011) 147–154

Fig. 6. HPLC chromatograms: (a) 16 h after the start of anaerobic phase; (b) An-SBR3 efﬂuent and (c) MB-SBBR3 efﬂuent.

Fig. 7. Change in 1N-4S concentration during the applied anaerobic–aerobic treatment systems.

Fig. 8. Attached-growth bioﬁlm mass in MB-SBBRs.

of the anaerobic process on the aerobic biodegradability of some azo dyes. The similarity of the chromatographic peak area corresponding to 1N-4S (RT: 2.02 min) in Fig. 5a and b and also in Fig. 6a and b, reveals that the anaerobic process had no effect on further biodegradation of 1N-4S in An-SBR2 and An-SBR3. However, in the case of An-SBR1 with the lower inﬂuent dye concentration (100 mg/L), 1N-4S was no longer present after the complete anaerobic cycle (Fig. 4b). The anaerobic removal of 1N-4S in AnSBR1 could be due to the almost long anaerobic hydraulic retention time (HRT = 2.75 day) and low concentration of 1N-4S in An-SBR1 compared to those of An-SBR2 and An-SBR3. The HPLC analysis also reveals that after the aerobic treatment, the chromatographic peak area decreased and shifted to the lower retention times (Figs. 4c, 5c and 6c). This ﬁnding indicates the formation of less aromatic and more polar compounds during the treatment process

in MB-SBBRs. The similar results has also been reported by other study groups [11,19]. Stoichiometrically, the complete reduction of the azo bond with the assumption of no anaerobic degradation of aromatic amines, leads to the accumulation of 40.5, 202.6 and 405.3 mg/L of 1N-4S in An-SBR1, 2 and 3, respectively. The change in 1N-4S concentration during the anaerobic–aerobic treatment systems which was determined from the developed area–concentration curve is shown in Fig. 7. Fig. 7 proves that the MB-SBBRs were dramatically able to mineralize 1N-4S. The concentration of 1N-4S in the efﬂuent of MB-SBBR1, 2 and 3 were below than detection limit (0.06 mg/L), 11.5 mg/L and 14.9 mg/L resulting in almost 100, 83.9 and 88.7% of 1N-4S removal efﬁciency, respectively. Contrary to the several studies which reported difﬁculties with the conventional aerobic decomposition of sulfonated aromatic amines [3,10,35], in this

E. Hosseini Koupaie et al. / Journal of Hazardous Materials 195 (2011) 147–154

153

Fig. 9. SEM micrographs of 80-day old bioﬁlm grown on surfaces of the bioﬁlm carriers.

study, a considerable removal efﬁciency of 1N-4S was obtained even at high initial dye concentration (1000 mg/L). Since the authentic standard was available just for 1N-4S, the other anaerobically formed intermediates could not be quantiﬁed through the HPLC analysis. However, as described by Libra et al. [17], the overall mineralization of the azo dyes metabolites can be followed by using lumped parameters such as dissolved organic carbon (DOC), total organic nitrogen (TON), etc. In this regard, considering the average COD concentration of the MB-SBBRs efﬂuent and the COD concentration of AR18 (597 mg-COD/L for 1000 mg/L of the dye solution), it is inferred that at least 71.5, 71.9 and 64.8% of the dye metabolites were completely mineralized using treatment system 1, 2 and 3, respectively. Comparison of the overall peak area among the HPLC chromatogram of the An-SBRs efﬂuent to that of the MB-SBBRs efﬂuent, reveals that almost 59.9, 51.7 and 50.1% of the total dye metabolites remaining in the efﬂuent of An-SBRs were removed in MB-SBBR1, 2 and 3, respectively. 3.3. Bioaccumulation of attached-growth bioﬁlm The attached-growth bioﬁlm mass in MB-SBBRs which was evaluated in the 40th, 60th and 90th day of the operation are compared in Fig. 8. It is evident that as the reactors operation progressed, the bioﬁlm mass increased in all three MB-SBBRs. It is also obvious that the bioﬁlm mass decreased with increasing the initial dye concentration. The ratio of the attached-growth bioﬁlm concentration to the suspended biomass concentration in the MB-SBBRs was also increased as the reactors operation progressed and reduced as the initial dye concentration increased. This ratio was 0.76, 0.7 and 0.51 at 90th day of the operation in MB-SBBR1, 2 and 3, respectively. Since all the parameters except the initial dye concentration were the same for all three treatment systems, the reduction in bioﬁlm growth is most likely due to the toxicity effects of the dye metabolites which was increased with increasing the initial AR18

concentration. This ﬁnding is in agreement with those reported by Asad et al. [2] in which the reduction in cell growth was observed during the decolorization of azo dye Remazol Black B. In another study carried out by Renganathan et al. [12], the inhibitory effect of AR18 at higher dye concentrations on growth and ﬁnal concentration of White rot fungus S. has been observed. In our study, the inhibitory effect of the dye metabolites on the microorganisms’ activity was also conﬁrmed by the respiration–inhibition test. The SOUR values were determined 31.9 ± 3.6, 18.0 ± 4.9 and 16.9 ± 2.7 mg O2 /g VSS/h for the mixed liquor withdrawn from MBSBBR1, 2 and 3, respectively. 3.4. Bioﬁlm morphology Bioﬁlm usually contains three-dimensional structures composed of interstitial voids, channels, cell clusters and extracellular polymeric substances (EPS) [36,37]. The SEM photographs of the 80-day old attached-growth bioﬁlm are presented in Fig. 9 where the micro pores and channels (15–50 ␮m in diameter) are clearly visible. Areal porosity of a bioﬁlm section is deﬁned as the ratio of the porous area to the total area of the section [38]. In our study, the areal porosity of the bioﬁlm was determined 15–25% with an image analysis of Fig. 9a. As noted by Villena et al. [37], the advantage of a channeled structure in comparison with the non-porous one is that it allows ﬂuids to pass through, enhancing mass transfer [37]. Therefore, one reason for signiﬁcant removal efﬁciency of the dye intermediates achieved in MB-SBBRs could be the high percentage of porosity present in the bioﬁlm structure (Fig. 9). 4. Conclusions In this study, up to 98% of AR18 decolorization and more than 80% of COD removal efﬁciency occurred anaerobically. Based on statistical analysis, there was no signiﬁcant difference among three

154

E. Hosseini Koupaie et al. / Journal of Hazardous Materials 195 (2011) 147–154

An-SBRs in the case of AR18 decolorization as well as the COD removal efﬁciency. The aerobic moving bed bioﬁlm process applied in this study has shown to be efﬁcient in elimination of aromatic amines formed through the anaerobic degradation of AR18. More than 80% of the anaerobically formed 1N-4S was removed in all three MB-SBBRs. The COD analysis results showed that at least 71.5, 71.9 and 64.8% of the dye total metabolites were completely mineralized through the treatment system 1, 2 and 3, respectively. The attached-growth bioﬁlm mass measurement as well as the respiration–inhibition test proved the inhibitory effect of the dye metabolites on the microorganisms’ activity in MB-SBBRs. The outcome of present study showed that the applied moving bed sequencing batch bioﬁlm reactor was highly efﬁcient in the removal of aromatic amines formed during the anaerobic degradation of azo dye AR18. Acknowledgements We are grateful to the Amirkabir University of Technology for providing Research materials, equipments and fund. In addition, the authors wish to thank Dr. Babak Bonakdarpour and Dr. Jalal Hassan for their consultation and also Ms. Lida Ezzedinloo and Ms. Shirin Mehrali for their assistance during the experiments. References [1] E. Acuner, F.B. Dilek, Treatment of tectilon yellow 2G by Chlorella vulgaris, Process Biochem. 39 (2004) 623–631. [2] S. Asad, M.A. Amoozegar, A.A. Pourbabaee, M.N. Sarbolouki, S.M.M. Dastgheib, Decolorization of textile azo dyes by newly isolated halophilic and halotolerant bacteria, Bioresour. Technol. 98 (2007) 2082–2088. [3] F.P. Van der Zee, S. Villaverde, Combined anaerobic–aerobic treatment of azo dyes—a short review of bioreactor studies, Water Res. 39 (2005) 1425–1440. ˜ X. Domènech, J.A. García-Hortal, F. Torrades, J. Peral, The [4] J. García-Montano, testing of several biological and chemical coupled treatments for Cibacron Red FN-R azo dye removal, J. Hazard. Mater. 154 (2008) 484–490. [5] M.Y.A. Mollah, S.R. Pathak, P.K. Patil, M. Vayuvegula, Treatment of orange II azo-dye by electrocoagulation (EC) technique in a continuous ﬂow cell using sacriﬁcial iron electrodes, J. Hazard. Mater. 109 (2004) 165–171. [6] S. Netpradit, P. Thiravetyan, S. Towprayoon, Application of ‘waste’ metal hydroxide sludge for adsorption of azo reactive dyes, Water Res. 37 (2003) 763–772. [7] S. Sadri Moghaddam, M.R. Alavi Moghaddam, M. Arami, A comparison study on Acid Red 119 dye removal using two different types of waterworks sludge, Water Sci. Technol. 61 (2010) 1673–1681. [8] A. Kunz, V. Reginatto, N. Durán, Combined treatment of textile efﬂuent using the sequence Phanerochaete chrysosporium–ozone, Chemosphere 44 (2001) 281–287. [9] Y. Dong, J. Chen, C. Li, H. Zhu, Decoloration of three azo dyes in water by photocatalysis of Fe (III)eoxalate complexes/H2 O2 in the presence of inorganic salts, Dyes Pigments 73 (2007) 261–268. [10] A. Pandey, P. Singh, L. Iyengar, Bacterial decolorization and degradation of azo dyes, Int. Biodeterior. Biodegrad. 59 (2007) 73–84. [11] N. Supaka, K. Juntongjin, S. Damronglerd, M.-L. Delia, P. Strehaiano, Microbial decolorization of reactive azo dyes in a sequential anaerobic–aerobic system, Chem. Eng. J. 99 (2004) 169–176. [12] S. Renganathan, W.R. Thilagaraj, L.R. Miranda, P. Gautam, M. Velan, Accumulation of acid orange 7, acid red 18 and reactive black 5 by growing Schizophyllum commune, Bioresour. Technol. 97 (2006) 2189–2193. [13] N. Dafale, S. Wate, S. Meshram, T. Nandy, Kinetic study approach of remazol black-B use for the development of two-stage anoxic–oxic reactor for decolorization/biodegradation of azo dyes by activated bacterial consortium, J. Hazard. Mater. 159 (2008) 319–328.

[14] P.I.M. Firmino, M.E.R. da Silva, F.J. Cervantes, A.B. dos Santos, Colour removal of dyes from synthetic and real textile wastewaters in one- and two-stage anaerobic systems, Bioresour. Technol. 101 (2010) 7773–7779. [15] K. Kumar, S. Saravana Devi, K. Krishnamurthi, S. Gampawar, N. Mishra, G.H. Pandya, T. Chakrabarti, Decolorisation, biodegradation and detoxiﬁcation of benzidine based azo dye, Bioresour. Technol. 97 (2006) 407–413. [16] S. Seshadri, P.L. Bishop, Anaerobic/aerobic treatment of selected azo dyes in wastewater, Waste Manage. 14 (1994) 127–137. [17] J.A. Libra, M. Borchert, L. Vigelahn, T. Storm, Two stage biological treatment of a diazo reactive textile dye and the fate of the dye metabolites, Chemosphere 56 (2004) 167–180. [18] M. Is¸ik, D.T. Sponza, Decolorization of azo dyes under batch anaerobic and sequential anaerobic/aerobic conditions, J. Environ. Sci. Health 39 (2004) 1107–1127. [19] C. O’Neill, A. Lopez, S. Esteves, F.R. Hawkes, D.L. Hawkes, S. Wilcox, Azo-dye degradation in an anaerobic–aerobic treatment system operating on simulated textile efﬂuent, Appl. Microbiol. Biotechnol. 53 (2000) 249–254. [20] S.W. FitzGerald, P.L. Bishop, Two stage anaerobic/aerobic treatment of sulfonated azo dyes, J. Environ. Sci. Health 30 (1995) 1251–1276. [21] M. Is¸ık, D.T. Sponza, Monitoring of toxicity and intermediates of C.I. Direct Black 38 azo dye through decolorization in an anaerobic/aerobic sequential reactor system, J. Hazard. Mater. 114 (2004) 29–39. [22] P. Barsing, A. Tiwari, T. Joshi, S. Garg, Application of a novel bacterial consortium for mineralization of sulphonated aromatic amines, Bioresour. Technol. 102 (2010) 765–771. [23] H. Jiang, P.L. Bishop, Aerobic biodegradation of azo dyes in bioﬁlms, Water Sci. Technol. 29 (1994) 525–530. [24] M. Farhadian, D. Duchez, C.d. Vachelard, C. Larroche, Monoaromatics removal from polluted water through bioreactors—a review, Water Res. 42 (2008) 1325–1341. [25] B. Rusten, B. Eikebrokk, Y. Ulgenes, E. Lygren, Design and operations of the Kaldnes moving bed bioﬁlm reactors, Aquacultur Eng. 34 (2006) 322–331. [26] J. Jing, J. Feng, W. Li, Y. Xu, Removal of COD from coking-plant wastewater in the moving-bed bioﬁlm sequencing batch reactor, Korean J. Chem. Eng. 26 (2008) 564–568. [27] G. Andreottola, P. Foladori, M. Ragazzi, R. Villa, Dairy wastewater treatment in a moving bed bioﬁlm reactor, Water Sci. Technol. 45 (2002) 321–328. [28] M. Delnavaz, B. Ayati, H. Ganjidoust, Prediction of moving bed bioﬁlm reactor (MBBR) performance for the treatment of aniline using artiﬁcial neural networks (ANN), J. Hazard. Mater. 179 (2010) 769–775. [29] B. Ayati, H. Ganjidoust, M. Mir Fattah, Degradation of aromatic compounds using moving bed bioﬁlm reactors, Iran. J. Environ. Health Sci. Eng. 4 (2007) 107–112. [30] B.G. Plósz, C. Vogelsang, K. Macrae, H.H. Heiaas, A. Lopez, H. Liltved, K.H. Langford, The BIOZO process – a bioﬁlm system combined with ozonation: occurrence of xenobiotic organic micro-pollutants in and removal of polycyclic aromatic hydrocarbons and nitrogen from landﬁll leachate, Water Sci. Technol. 61 (2010) 3188–3197. [31] S. Chen, D. Sun, C. Jong-Shik, Treatment of pesticide wastewater by movingbed bioﬁlm reactor combined with Fenton-coagulation pretreatment, J. Hazard. Mater. 144 (2007) 577–584. [32] APHA, AWWA, WPCF, Standard Methods for the Examination of Water and Wastewater, 19th ed., American Public Health Association, Washington, DC, USA, 1998. [33] D.T. Sponza, M. Is¸ik, Ultimate azo dye degradation in anaerobic/aerobic sequential processes, Water Sci. Technol. 45 (2002) 271–278. [34] H. An, T. Qian, X. Gu, W.Z. Tang, Biological treatment of dye wastewaters using an anaerobic–oxic system, Chemosphere 33 (1996) 2533–2542. [35] H.M. Pinheiro, E. Touraud, O. Thomas, Aromatic amines from azo dye reduction: status review with emphasis on direct UV spectrophotometric detection in textile industry wastewaters, Dyes Pigments 61 (2004) 121–139. [36] X.-M. Zhan, M. Rodgers, E. O’Reilly, Bioﬁlm growth and characteristics in an alternating pumped sequencing batch bioﬁlm reactor (APSBBR), Water Res. 40 (2006) 817–825. [37] G.K. Villena, T. Fujikawa, S. Tsuyumu, M. Gutiérrez-Correa, Structural analysis of bioﬁlms and pellets of Aspergillus niger by confocal laser scanning microscopy and cryo scanning electron microscopy, Bioresour. Technol. 101 (2010) 1920–1926. [38] Z. Lewandowski, H. Beyenal, J. Myers, D. Stookey, The effect of detachment on bioﬁlm structure and activity: the oscillating pattern of bioﬁlm accumulation, Water Sci. Technol. 55 (2007) 429–436.

Journal of Hazardous Materials 195 (2011) 155–161

Contents lists available at ScienceDirect

Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat

Structural characterisation of Arquad® 2HT-75 organobentonites: Surface charge characteristics and environmental application Binoy Sarkar a,b , Mallavarapu Megharaj a,b , Yunfei Xi a,b , Ravi Naidu a,b,∗ a b

CERAR – Centre for Environmental Risk Assessment and Remediation, Building X, University of South Australia, Mawson Lakes, SA 5095, Australia CRC CARE – Cooperative Research Centre for Contamination Assessment and Remediation of the Environment, P.O. Box 486, Salisbury, SA 5106, Australia

a r t i c l e

i n f o

Article history: Received 29 December 2010 Received in revised form 3 August 2011 Accepted 7 August 2011 Available online 12 August 2011 Keywords: Bentonite Arquad® 2HT-75 Organoclay Surface charge Adsorption

a b s t r a c t Organoclays are increasingly being used to remediate both contaminated soils and waste water. The present study was attempted to elucidate the structural evolution of bentonite based organoclays prepared from a commercially available, low-cost alkyl ammonium surfactant Arquad® 2HT-75. XRD, FTIR, SEM and zeta potential measurement were used to characterise the organoclays. In particular, the relationship between surface charge characteristics of the organoclays and their ability to remediate organic contaminants such as phenol and p-nitrophenol was investigated. The investigation revealed that the arrangement and conformation of surfactant molecules in the bentonite became more regular, ordered and solid-like as of Arquad® 2HT-75 loading increased. This also led to the formation of a positive zeta potential on the surface of organobentonites prepared with 3.57:1 and 4.75:1 surfactant–clay (w/w) ratio. The zeta potential values decreased with increasing pH of the suspension. The adsorption data of phenol and p-nitrophenol were best ﬁtted to Freundlich isotherm model. The adsorption was controlled by multiple mechanisms of partitioning, physico-sorption and chemisorption. The outcomes of this study are useful for the synthesis of low cost organobentonite adsorbents for the remediation of ionisable organic contaminants such as phenol and p-nitrophenol from waste water. © 2011 Elsevier B.V. All rights reserved.

1. Introduction Organoclay is often prepared by modifying natural clay mineral with quaternary ammonium surfactant. During this preparation, the hydrated cations (Na+ , K+ , Ca2+ , Mg2+ ), which neutralise the excess negative charge evolving due to isomorphous substitution in the silicon tetrahedra and/or aluminium octahedra, are replaced by the quaternary ammonium cations in the clay interlayer. The quantity of surfactant molecules in excess to the cation exchange capacity (CEC) of clay mineral can incorporate into the clay structure by aggregation of organic molecules through mechanisms such as van der Wall interaction and hydrophobic bonding [1,2]. As a result, the natural clay minerals that are intrinsically hydrophilic in nature become hydrophobic. These modiﬁed clay products are extensively used in nanocomposites synthesis and adsorbent development. Organoclays are prepared from a range of clay minerals and alkylammonium surfactants. Arquad® 2HT-75 is one of the surfactants which is commercially available and inexpensive, but

∗ Corresponding author at: CERAR – Centre for Environmental Risk Assessment and Remediation, Building X, University of South Australia, Mawson Lakes, SA 5095, Australia. Tel.: +61 8 8302 5041; fax: +61 8 8302 3124. E-mail addresses: [email protected], [email protected] (R. Naidu). 0304-3894/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.jhazmat.2011.08.016

not been much studied for organoclay preparation. Chemically Arquad® 2HT-75 is di(hydrogenated tallow) dimethylammonium chloride (∼75%) with 2-propanol (∼14%) and water (∼11%) impurity. Tallow is a hard fat consists chieﬂy of glyceryl esters of oleic, palmitic, and stearic acids (16–18 carbon chains) and extracted from fatty deposits of animals, especially from suet (fatty tissues around the kidneys of cattle and sheep). This surfactant does not have an accurate molecular weight because of the impurities and homologs present in it. Recently we demonstrated that Arquad® 2HT-75 is much less toxic to soil inhabiting microorganisms than other most widely used surfactants namely, hexadecyl trimethylammonium bromide (HDTMA) and octadecyl trimethylammonium bromide (ODTMA) [3]. The sorption behaviour of Arquad® 2HT-75 in soils also differs from that of HDTMA and ODTMA [3]. Being comparatively less toxic to soil microorganisms, Arquad® 2HT-75 derived organoclays might be more suitable for contaminant degrading microbes to thrive on them and thus successful development of bio-reactive organoclays [2]. Bio-reactive organoclay can serve dual functions of contaminant immobilisation and subsequent biodegradation or biotransformation by the microorganisms growing on the adsorbents [2]. Organoclays are proven technology for remediation of both organic and inorganic contaminants by adsorption [2,3]. Bentonite based Arquad® 2HT-75 organoclay was successfully used

156

B. Sarkar et al. / Journal of Hazardous Materials 195 (2011) 155–161

for adsorptive removal of hexavalent chromium from aqueous system under environmental conditions [4]. However, for the elucidation of mechanisms of interaction between organoclay adsorbent and the target contaminants in the environment, it is very much necessary to study the structural evolution of the modiﬁed clay adsorbents. The complete knowledge about the structural behaviour of the organoclays will also help to decide subsequent other use of the materials such as nanocomposites or catalysts synthesis. In this paper we report the preparation of organobentonites using various loading rates of Arquad® 2HT-75 and structural characteristics evolution of the clay products. We also present the surface charge behaviour of these modiﬁed clay minerals and their application for the adsorption of phenol and p-nitrophenol in the environment. These contaminants are frequently released into the environment from pesticides, dyes and pigments, polymer engineering, olive processing, petrochemicals and pharmaceutical industries and hence need remediation consideration. 2. Experimental 2.1. Materials and preparation A locally available bentonite (QB) was used in this study. It was passed through a 300 ␮m sieve before use. The cation exchange capacity of the bentonite (66.7 cmol (p+ ) kg−1 ) was determined by the ammonia electrode method [5]. The dominant inorganic cation in the bentonite was Ca2+ . Arquad® 2HT-75 was purchased from Sigma–Aldrich, Australia. The loadings of surfactant for modiﬁcation were adjusted as 0.60:1, 1.2:1, 1.79:1, 2.38:1, 3.57:1 and 4.75:1 surfactant–clay (w/w) ratio for QB-Aq1, QB-Aq2, QB-Aq3, QB-Aq4, QB-Aq5 and QB-Aq6, respectively. The organoclays were prepared in a cation exchange reaction at mildly elevated temperature (80 ◦ C) as reported earlier [4,6,7]. The required quantity of surfactant was dissolved in 500 mL deionised water on a magnetic stirrer heated to 80 ◦ C. Then 30 g clay (≤300 ␮m size fraction) was added to it and the mixture was stirred continuously for 5 h. The reaction temperature was maintained at 80 ◦ C. Care was taken while stirring so that excess spume did not form during the reaction. Later, the organoclay was separated from the mixture by centrifugation and was washed several times with deionised water until a negative result for the presence of chloride was obtained with AgNO3 . The organoclays were then dried at 60 ◦ C for 24 h in a hot air oven and crushed into powders in agate mortars. They were stored under moisture free conditions for further use. 2.2. Material characterisation X-ray diffractograms of powdered QB and the organoclays were ˚ on a Lab X, XRD-6000, obtained using CuK˛ radiation ( = 1.5418 A) Shimadzu diffractometer (Shimadzu Corporation, Japan) operating at 40 kV and 30 mA between 1.5 and 65◦ (2) at a step size of 0.016◦ . The basal spacing (d) was calculated from the 2 value using Bragg’s equation. Infrared (IR) spectra were obtained using a Magna-IRTM spectrometer 750 (Nicolet Instrument Corp., USA) equipped with liquid nitrogen cooled mercury–cadmium–telluride (MCT) detector and DRIFT (Diffuse Reﬂectance Infra-red Fourier Transform) accessories. Spectra over the 4000–400 cm−1 range were obtained by the co-addition of 64 scans with a resolution of 4 cm−1 and a mirror velocity of 0.6329 cm s−1 . The band component was analysed using PeakFit v4.12 software package (Hearne Scientiﬁc Software). Also, morphology of the organoclays was examined by a Philips XL30 FEG scanning electron microscope (SEM). The ﬁnely ground clay products were dried at room temperature and then coated with platinum under vacuum in argon atmosphere for the SEM studies. The surface charge behaviour of the organoclays was determined

from their zeta potential values in aqueous suspension (0.01% w/v) at pH values ranging from 2 to 12. Zeta potential was measured on a Malvern Zetasizer Nano instrument (Malvern Instruments, USA). Total organic carbon (TOC) in the organoclays was determined by high temperature combustion in an atmosphere of oxygen using a Leco CNS-2000. Carbon was converted to CO2 and determined by infrared detection [8]. 2.3. Adsorption study To test the adsorption of phenol and p-nitrophenol onto organoclays, batch experiments were conducted by equilibrating 0.1 g of adsorbent with 10 mL of aqueous phenol and p-nitrophenol solution separately in 40 mL centrifuge tube. The initial concentration of phenol and p-nitrophenol in the reaction mixture ranged from 50 to 700 mg L−1 and 50–1500 mg L−1 , respectively. The mixture was continuously agitated on an end-over-end shaker for 4 h. Preliminary experiment showed that the adsorption of phenol and p-nitrophenol onto these organoclays reach equilibrium within 2 h of agitation. The batch adsorption was started at an initial pH value very close to neutral (pH 6.7–7). However, after 2 h of agitation the pH of the adsorption mixtures varied from 6.9 to 7.5. The increase in pH was not affected by the concentration of p-nitrophenol in the adsorption mixtures, but was more prominent for the organoclays prepared with surfactant concentration greater than the CEC of the bentonite. All adsorption experiments were carried out in triplicate in deionised water without any background electrolyte. The solute-adsorbent mixture was centrifuged at 11,000 rpm for 10 min to separate the sorbent from the liquid. Then the concentration of phenol and p-nitrophenol in the supernatant was analysed on an Agilent 8453 UV–vis Spectrophotometer (Agilent Technologies, Japan) at 270 and 317 nm wavelength, respectively. The pKa of p-nitrophenol at 25 ◦ C is 7.15. Above pH 7.15, molecular p-nitrophenol dissociates into p-nitrophenol anion and gives yellow colour which causes a shift of the absorption peak (Kmax ) from 317 nm to 400 nm [6]. For this reason, all the aliquots (following separation of the adsorbents by centrifugation) including the standards were mildly acidiﬁed with a drop of 0.5 M HCl immediately before spectrophotometric measurement of p-nitrophenol at 317 nm [9]. The quantity of phenol and p-nitrophenol adsorbed was calculated using the following equation: qe = V

(Ci − Ce ) (M × 1000)

(1)

where, qe is the amount of solute adsorbed on the adsorbent (mg g−1 ), Ci the initial liquid phase concentration of the solute (mg L−1 ), Ce the equilibrium liquid phase concentration of the solute (mg L−1 ), V the volume of liquid phase (mL) and M the mass of the adsorbent (g). 2.4. Statistical analysis The ﬁtness of the adsorption data into isothermal model was tested by SPSS version 18 (SPSS Inc., Chicago, USA) packages. The distribution of residues after model ﬁtting was checked as described by Draper and Smith [10]. 3. Results and discussion 3.1. Characterisation of the organobentonites 3.1.1. X-ray diffraction Fig. 1 shows the X-ray diffraction patterns of the unmodiﬁed bentonite and its organoclays. The basal spacing of unmodiﬁed ´˚ As observed in Fig. 1, the X-ray diffracbentonite was 15.03 A. tion reﬂections of the organoclays shifted towards lower 2 values

B. Sarkar et al. / Journal of Hazardous Materials 195 (2011) 155–161

157

Fig. 2. Changes in the basal spacings (d) of bentonite and the Arquad® 2HT-75modiﬁed organobentonites with increasing TOC contents (%).

Fig. 1. X-ray diffraction patterns of bentonite and the Arquad® 2HT-75-modiﬁed organobentonites.

as the loadings of surfactant in the organoclays increased. The corresponding d values of all the organobentonites are shown in Fig. 1. It is apparent from Fig. 2 that the gradual increase in surfactant density in the clay interlayer resulted in a corresponding increase in the basal spacing from 15.15 A´˚ to 35.04 A´˚ for QB-Aq1, QB-Aq2, QB-Aq3, QB-Aq4, QB-Aq5 and QB-Aq6. We modiﬁed QB with Arquad® 2HT-75 which does not have an accurate molecular weight due to the presence of impurities and homologs. For this reason, the calculation of exact loadings of the surfactant in the QB interlayer was not possible to explain in terms of clay CEC. Rather we used the surfactant–clay (w/w) ratio of 0.60:1, 1.2:1, 1.79:1, 2.38:1, 3.57:1 and 4.75:1 to obtain organoclays QB-Aq1, QB-Aq2, QB-Aq3, QB-Aq4, QB-Aq5 and QB-Aq6, respectively. The corresponding TOC contents of the organoclays were 6.9%, 12.5%, 17.4%, 21.7%, 28.1% and 30.6%, respectively. Fig. 2 also shows that the basal spacing for QB-Aq2 (12.5% TOC) was slightly higher than QB-Aq1 (6.9% TOC). Thereafter, for QB-Aq3 (17.4% TOC) and QB-Aq4 (21.7% TOC) the basal spacing increased sharply and then arrived to almost a plateau for QB-Aq5 (28.1% TOC) and QB-Aq6 (30.6% TOC). The basal spacing (d 0 0 1) give details of arrangement of surfactant molecules in the organoclays [11–14]. Xi et al. [11] described that (a) at surfactant concentrations equivalent to 0.2–0.4 CEC of the bentonite clay mineral, lateral monolayer and lateral bilayer conformation is formed, (b) at 0.6–0.8 CEC modiﬁcations, lateral bilayer

arrangement is obtained and (c) above 1 CEC, pseudo-trilayer or parafﬁn layer orientation is formed. However, this model is applicable for the structure of organoclays prepared with surfactants having only single alkyl tail. Arquad® 2HT-75 used in the present study contains di(hydrogenated tallow) dimethylammonium chloride which has two long alkyl chains. Xu and Zhu [14] observed the typical d spacing of 15.1 A´˚ (0.1–0.5 CEC) and 37.1 A´˚ (0.5–0.7 CEC) for organobentonites prepared with two alkyl tail containing surfactant dioctadecyl dimethylammonium (DODMA). They [14] suggested formation of a ﬂat monolayer arrangement for DODMA+ chains in the organobentonite at low surfactant loadings where the two alkyl chains linked to the same N-atom of DODMA+ both laid parallel to the silicate plane. They also proposed that a parafﬁntype bilayer with a tilt angle of 33◦ to the silicate planes was the predominant arrangement of adsorbed DODMA+ at high surfactant contents [14]. In the present study, QB-Aq1 and QB-Aq2 might produce lateral monolayer conformation, whereas QB-Aq3 and QB-Aq4 might reﬂect formation of lateral bilayer conformation [14–16]. QB-Aq5 might produce parafﬁn type bilayer in the clay interlayer [14–16]. At further higher surfactant concentration, QB-Aq6 could give rise to more ﬂattened plate formation resulting from interactions between surfactant alkyl chain-silicate surface and the alkyl chain-alkyl chain of the alkyl ammonium cation [11–13,15]. Such regular stacking of the surfactant molecules in the clay interlayer is further supported by the appearance of higher order of peaks (0 0 2 and 0 0 3 planes) in the XRD patterns [4,12]. Fig. 1 clearly shows that QB-Aq4, QB-Aq5 and QB-Aq6 produce 0 0 2 peaks at 15.03, 17.31 ´˚ respectively. Given a double chain cationic surfactant and 17.53 A, was used to prepare organoclay, evolution of two expansions at each of the surfactant loading levels was proposed due to the existence of two structural arrangements of the surfactant molecules in the clay interlayer [15,17]. Although Xu and Zhu [14] used a surfactant (DODMA) with well deﬁned chain length (double chain) and accurately known molecular weight (630.95) to propose the above mentioned organoclay structure, the organobentonites prepared with Arquad® 2HT-75 would also produce a similar arrangement pattern of the surfactant molecules in the clay minerals due to the fact that Arquad® 2HT-75 is also a double chain surfactant having comparable equivalent mass (576) with DODMA.

158

B. Sarkar et al. / Journal of Hazardous Materials 195 (2011) 155–161

Fig. 3. FTIR spectra of bentonite and the Arquad® 2HT-75-modiﬁed organobentonites.

3.1.2. FTIR characterisation of the organoclays Fig. 3 shows FTIR spectra of QB and the organoclays. The IR spectra were presented in two distinguished regions. The peaks appearing as the signature of CH2 stretching vibration (both symmetric and asymmetric) and CH2 bending vibrations are shown in Fig. 3. The bands appearing at 3000–2800 cm−1 region were indicative of the ordering (gauche/trans conformer ratio), packing density of the surfactants in organoclays and the interactions between the alkyl chains [4,6,18,19]. As shown in Fig. 3, the CH2 asymmetric stretching vibration (vasy ) appeared at region ranging 2929–2920 cm−1 wave number for the six organoclays studied, whereas CH2 symmetric stretching vibration (vsy ) for them appeared at 2855–2850 cm−1 wave number. Both of these peak regions did not appear in unmodiﬁed QB. As the distinctive feature of the spectra it appeared that there was a shift for vasy and vsy bands towards a lower frequency as the surfactant loading rate (TOC contents in organoclays) increased, which was indicative of highly ordered all-trans conformations of Arquad® 2HT-75 in the QB structure [19–21]. In addition, the CH2 bending vibration bands for the organoclays appeared at wave numbers 1477–1471 cm−1 and here also the shift of bands towards lower frequency with increasing surfactant loadings (TOC contents in organoclays) was prominently observed. The spectral position of these peaks shown by QB-Aq6, the organoclay prepared with maximum surfactant loading density, had been exactly similar to that of pure Arquad® 2HT-75. It implies that the conformation of the adsorbed surfactant molecules in the organoclays progressively develops from low packing density and ordering (liquid like) to high packing density and ordering (solid like) [4,6]. Fig. 3 also shows an IR peak appearing

Fig. 4. SEM micrographs of bentonite and the Arquad® 2HT-75-modiﬁed organobentonites.

for QB and the organoclays within 1639–1645 cm−1 region which could be attributed to OH deformation vibration [22]. However, no additional structural information of the organoclays could be obtained from this peak as they did not follow any particular trend. 3.1.3. SEM characterisation of the organoclays SEM study indicated that the morphological differences among the prepared organoclays were not signiﬁcantly observable. Fig. 4 shows the SEM micrographs of QB, QB-Aq4 and QB-Aq6. It was found that unmodiﬁed QB showed massive and aggregated morphology with little number of ﬂakes, whereas the organoclays had less aggregated morphology with large amount of ﬂakes with curled and crumpled structure. As the surfactant density in organoclay increased, the number of ﬂakes also increased, but became more ﬂat in appearance (Fig. 4). The surfactant modiﬁcation of

B. Sarkar et al. / Journal of Hazardous Materials 195 (2011) 155–161

159

chains of the tallow compound led to stronger lateral interaction binding more cations within the shear plane [23–25]. In all the organoclays and unmodiﬁed QB, values decreased with increasing pH of the suspension (Fig. 5). Only QB-Aq6 (30.6% TOC) could maintain positive values over the whole range of pH studied (pH 2–12). QB-Aq5 (28.1% TOC) also maintained positive values at pH range 2–6 after which it developed negative surface charge. The values of all other organobentonites were negative over the entire pH range studied and the values became more negative as the pH increased. As indicated by the slope of the vs. pH graphs, the values were less sensitive to the change of pH as the contents of surfactant (TOC) decreased in the organoclays, which is supported by Bate and Burns [25]. The effect of pH on the values of the organoclays might depend on the interactions between surfactant cation and the clay mineral surface. By imparting hydrophobic and van der Waals interactions among themselves, the alkyl tails of the surfactant cation shielded the negative charge on the bentonite surface and developed positive charge when the surfactant addition was in excess to the CEC. With the change in pH from the point of zero charge of the bentonite, the dissolution and subsequent release of metals cations increased that in turn changed the interaction between surfactant cations and bentonite surface [25–28]. 3.2. Adsorption of phenol and p-nitrophenol

Fig. 5. Changes in the zeta potentials () of bentonite and the Arquad® 2HT-75modiﬁed organobentonites as a function of pH.

the bentonite might cause fragmentation of the ﬂakes and thus enhance adsorption of contaminants. 3.1.4. Surface charge characteristics of the organoclays There is limited information in the literature on the relationship between surface charge characteristics of organoclays and contaminant adsorption. In our previous investigations we reported that organopalygorskites prepared with 200% CEC equivalent surfactant concentration could produce positive surface charge and enhance p-nitrophenol adsorption [6,23]. We also reported that zeta potential () values of bentonite modiﬁed with Arquad® 2HT-75 at high concentration could reach as high as 51.1 mV at pH ∼ 4 as compared to −21.3 mV measured for unmodiﬁed QB [4]. In this study we investigated in detail the surface charge behaviour of Arquad® 2HT-75 modiﬁed organobentonites as a function of pH. Positive was formed on the surface of organobentonites when the TOC contents were 28.1% and 30.6% (QB-Aq5 and QB-Aq6) (Fig. 5). When the concentration of surfactant in organoclays exceeded the CEC of the clay (as observed from the increasing TOC contents), the excess surfactant molecules adhered to the surface-adsorbed surfactant cations by van der Waal forces [11]. Because of the presence of cationic surfactant molecules in excess to the CEC of the clay mineral covering the total internal and external surface of the bentonite, the resultant organoclays exhibited positive surface charge [6,23,24]. Bate and Burns [25] also reported that the values of organomontmorillonites became less negative as the total organic carbon (surfactant concentration) and the length of attached carbon chain increased. It was attributable to the hydrophobicity of the organoclay, which increased the hydrophobic lateral interaction between surfactant cations giving positive charge within the shear plane [25]. As Arquad® 2HT-75 contains glyceryl esters of oleic, palmitic, and stearic acids (16–18 carbon chains) [1], the longer

We tested the adsorption efﬁciency of Arquad® 2HT-75 organobentonites for ionisable organic pollutants such as phenol and p-nitrophenol. Among various isothermal models tested, the Freundlich model [29] best ﬁtted the adsorption data. Freundlich model: lnqe = lnKF +

lnCe n

(2)

where, Ce is equilibrium adsorbate concentration, qe is amount of adsorbate adsorbed at equilibrium, KF and 1/n are the Freundlich constants related to adsorption capacity and intensity of adsorption (n is the heterogeneity factor). The Freundlich model implies the bonding energy of the adsorbate on a given surface decreases with fractional coverage of the surface area and in case of heterogeneous surfaces it is closer to reality than the assumption of constant bonding energy [6,28]. Freundlich plot of phenol adsorption is shown in Fig. 6a. The adsorption data of phenol by Arquad® 2HT-75 modiﬁed bentonites ﬁtted very well to Freundlich isothermal model. The r2 values ranged from 0.985 to 0.997 (Table 1). The corresponding F statistics (the degree of reduction of heterogeneity in a population) and p (probability) values also indicated strong ﬁt of the adsorption data to this model (Table 1). The distributions of the model ﬁtting residues were systematic bell-shaped except for the adsorbent QB-Aq1 (ﬁgures not shown). The values of 1/n and KF increased gradually with increasing surfactant loadings in the organobentonites (Table 1). The 1/n values were less than unity in case of all the adsorbents which indicated that at higher adsorbate concentrations, the ﬁlling of total adsorption capacity occurred without increasing the adsorption intensity [6]. The gradually increasing KF values indicated the increase in afﬁnity of phenol to the adsorbents [6]. Freundlich plot of p-nitrophenol adsorption is shown in Fig. 6b. Similar to phenol, p-nitrophenol adsorption data also ﬁtted very well to the Freundlich model (Table 1). The model ﬁtting parameters for p-nitrophenol indicated stronger afﬁnity of the adsorbate to the organobentonites surface as compared to phenol (Table 1). The distributions of the model ﬁtting residues were bell-shaped for all the adsorbents (ﬁgures not shown). However, compared to phenol, p-nitrophenol adsorption data showed slight non-linearity

160

B. Sarkar et al. / Journal of Hazardous Materials 195 (2011) 155–161 Table 2 Normalized amount of adsorption of phenol and p-nitrophenol per percent of carbon incorporated into the organoclays. Organoclays

TOC content (%)

QB-Aq1 QB-Aq2 QB-Aq3 QB-Aq4 QB-Aq5 QB-Aq6

6.9 12.5 17.4 21.7 28.1 30.6

Normalized adsorption capacity (mg g−1 ) Phenol

p-Nitrophenol

2.1 2.5 2.7 2.8 2.9 3.2

6.0 4.2 3.4 3.5 3.1 3.3

was indicative of adsorption predominantly through partitioning process [29]. Phenol and p-nitrophenol adsorption capacity of the organobentonites was normalized against the TOC content of the modiﬁed adsorbents (Table 2). The normalized adsorption capacity of phenol per percent of incorporated carbon gradually increased as the surfactant loadings increased in the organoclays. However, the normalized adsorption capacity of p-nitrophenol decreased with increasing surfactant loadings (TOC contents). The relationship between the KF value and the TOC content in organoclays for phenol adsorption was linear with high correlation coefﬁcient value (r2 = 0.99 at p < 0.01). The 1/n values also gradually increased (Table 1). However, KF value for p-nitrophenol adsorption maintained a less linear relationship with TOC contents in organoclays (r2 = 0.86 at p < 0.01). The results indicated that the mechanism of phenol removal gradually shifted from surface adsorption to partitioning as the TOC contents in the adsorbents increased [30,31], whereas multiple mechanisms such as partitioning, physio-sorption and chemisorption controlled p-nitrophenol adsorption [6,22]. Overloading (relative to CEC of the bentonite) of organoclays with surfactants was associated with the appearance of the adsorbed surfactant-anion (halide) pairs such that anion exchange with p-nitrophenol anion occurred [6,23]. Evidences exist that the halide counter ions released from the organoclays due to the adsorption of p-nitrophenol anion at pH > 7.15 [6]. This also caused greater non-linearity of the p-nitrophenol adsorption isotherms (the reduction of 1/n values) as compared to phenol (Table 1). 3.3. Mechanisms of adsorption QB modiﬁed with Arquad® 2HT-75 having long alkyl chain attracted phenol both by adsorption and partition mechanisms [30–32]. Juang et al. [30] proposed adsorption of phenol also by ion exchange on the surfaces of organomontmorillonite. As apparent from the KF values (Table 1), the afﬁnity of p-nitrophenol for the adsorption sites of organobentonites was higher than that of phenol because of the difference in the polarity level of these two contaminants. The adsorption of p-nitrophenol to organobentonites

Fig. 6. Freundlich ﬁttings for the adsorption of (a) phenol and (b) p-nitrophenol by Arquad® 2HT-75-modiﬁed organobentonites. Bars represent standard errors of mean at 95% conﬁdence level (n = 3).

during model ﬁttings. The KF values increased remarkably with the increase of surfactant contents (TOC contents) in the adsorbents and reached as high as 19 for QB-Aq6. Such higher KF value

Table 1 Fitting of the phenol and p-nitrophenol adsorption data by the Freundlich model at 95% conﬁdence level. Sample

Phenol

p-nitrophenol −1

1/n QB-Aq1 QB-Aq2 QB-Aq3 QB-Aq4 QB-Aq5 QB-Aq6 * a

0.71 0.78 0.76 0.77 0.79 0.82

± ± ± ± ± ±

0.01a 0.006 0.012 0.014 0.017 0.013

KF (L g

)

0.0913 0.1453 0.2949 0.3953 0.5052 0.5805

± ± ± ± ± ±

2

r 0.013 0.019 0.017 0.027 0.025 0.033

0.985 0.992 0.993 0.993 0.997 0.989

*

F

368 715 825 818 2214 525

*

p

1/n

<0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001

0.48 0.50 0.45 0.38 0.24 0.24

F statistics – the degree of reduction of heterogeneity in a population; p – probability. Standard error at 95% conﬁdence level, n = 3.

KF (L g−1 ) ± ± ± ± ± ±

0.008 0.01 0.013 0.016 0.027 0.012

1.6 1.9 3.4 6.5 17.0 18.8

± ± ± ± ± ±

0.09 0.12 0.23 0.29 0.37 0.33

r2

F*

p*

0.978 0.986 0.978 0.993 0.986 0.990

265 412 271 838 431 612

<0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001

B. Sarkar et al. / Journal of Hazardous Materials 195 (2011) 155–161

with surfactant contents up to the CEC of the bentonite involved both surface adsorption and partition, whereas its adsorption to organobentonites with surfactant contents in excess to the CEC involved mainly partitioning [11,33]. Adsorption of p-nitrophenol by organopalygorskites prepared with surfactant loadings in excess to the CEC of palygorskite occurred through anion exchange also following formation of p-nitrophenol anion at higher pH (pH > 7.15) of the adsorption system [6,23]. In one of our previous publication, we reported that p-nitrophenol adsorption by organoclays was controlled by multiple mechanisms including hydrophobic bonding, electrostatic attraction and anion exchange (at pH > 7.15) [6,23]. The electrostatic attraction mechanism can come into effect when organoclay possess positive surface charge () and p-nitrophenol anion (after ionisation at pH > 7.15) gets adsorbed thereon [23]. 4. Conclusions Bentonite could be effectively modiﬁed with Arquad® 2HT-75. Due to insertion of the surfactant molecule into the clay interlayer the basal spacing of the bentonite gradually increased with increasing surfactant loadings. The arrangement and conformation of surfactant molecules in the bentonite became more regular, ordered and solid-like as the loadings of Arquad® 2HT-75 increased. Positive zeta potential was formed on the surface of organobentonites prepared with 3.57:1 and 4.75:1 surfactant–clay (w/w) ratio where surfactant molecules in excess to the CEC of the clay mineral aggregated on the surface. In all the organoclays and unmodiﬁed bentonite the zeta potential values decreased with increasing pH of the suspension. The adsorption capacities of phenol and p-nitrophenol improved signiﬁcantly due to the modiﬁcation of bentonite with Arquad® 2HT-75. The adsorption data best ﬁtted to Freundlich isothermal model. Predominantly partitioning and adsorption, and to some extent ion exchange and physiosorption were the mechanisms of these contaminants removal by these adsorbents from aqueous solutions. This study illustrates the potential of low cost organobentonite adsorbents for the remediation of ionisable organic contaminants such as phenol and p-nitrophenol from contaminated waste water. Acknowledgement Binoy Sarkar is thankful to the University of South Australia for the award of University President Scholarship and to the Cooperative Research Centre for Contamination Assessment and Remediation of the Environment (CRC CARE) for PhD fellowship. The authors would like to acknowledge the ﬁnancial and infrastructural support from the CRC CARE and the Centre for Environmental Risk Assessment and Remediation (CERAR), University of South Australia. Suggestions from anonymous reviewers to improve the quality of the manuscript are gratefully acknowledged. References [1] L.B. de Paiva, A.R. Morales, F.R. Valenzuela Díaz, Organoclays: properties preparation and applications, Appl. Clay Sci. 42 (2008) 8–24. [2] B. Sarkar, Y. Xi, M. Megharaj, G. Krishnamurti, M. Bowman, H. Rose, R. Naidu, Bio-reactive organoclay: a new technology for environmental remediation, Crit. Rev. Environ. Sci. Technol. (in press), doi:10.1080/10643389.2010.518524. [3] B. Sarkar, M. Megharaj, Y. Xi, G.S.R. Krishnamurti, R. Naidu, Sorption of quaternary ammonium compounds in soils: implications to the soil microbial activities, J. Hazard. Mater. 184 (2010) 448–456. [4] B. Sarkar, Y. Xi, M. Megharaj, G.S.R. Krishnamurti, D. Rajarathnam, R. Naidu, Remediation of hexavalent chromium through adsorption by bentonite based Arquad® 2HT-75 organoclays, J. Hazard. Mater. 183 (2010) 87–97.

161

[5] D. Borden, R.F. Giese, Baseline studies of the clay minerals society source clays: cation exchange capacity measurements by the ammonia-electrode method, Clays Clay Miner. 49 (2001) 444–445. [6] B. Sarkar, Y. Xi, M. Megharaj, G.S.R. Krishnamurti, R. Naidu, Synthesis and characterisation of novel organopalygorskites for removal of p-nitrophenol from aqueous solution: Isothermal studies, J. Colloid Interface Sci. 350 (2010) 295–304. [7] B. Sarkar, Y. Xi, M. Megharaj, R. Naidu, Orange II adsorption on palygorskites modiﬁed with alkyl trimethylammonium and dialkyl dimethylammonium bromide – an isothermal and kinetic study, Appl. Clay Sci. 51 (2011) 370–374. [8] I. Matejovic, Determination of carbon and nitrogen in samples of various soils by the dry combustion, Commun. Soil Sci. Plant Anal. 28 (1997) 1499–1511. [9] H. He, Q. Zhou, W.N. Martens, T.J. Kloprogge, P. Yuan, Y. Xi, J. Zhu, R.L. Frost, Microstructure of HDTMA+ -modiﬁed montmorillonite and its inﬂuence on sorption characteristics, Clays Clay Miner. 54 (2006) 689–696. [10] N.R. Draper, H. Smith, Applied Regression Analysis, Wiley, New York, 1998. [11] Y. Xi, Z. Ding, H. He, R.L. Frost, Structure of organoclays – an X-ray diffraction and thermogravimetric analysis study, J. Colloid Interface Sci. 277 (2004) 116–120. [12] J. Zhu, H. He, J. Guo, D. Yang, X. Xie, Arrangement models of alkylammonium cations in the interlayer of HDTMA+ pillared montmorillonites, Chin. Sci. Bull. 48 (2003) 368–372. [13] H. He, R.L. Frost, T. Bostrom, P. Yuan, L. Duong, D. Yang, Y. Xi, J.T. Kloprogge, Changes in the morphology of organoclays with HDTMA+ surfactant loading, Appl. Clay Sci. 31 (2006) 262–271. [14] L. Xu, L. Zhu, Structures of OTMA- and DODMA-bentonite and their sorption characteristics towards organic compounds, J. Colloid Interface Sci. 331 (2009) 8–14. [15] Y. Xi, R.L. Frost, H. He, Modiﬁcation of the surfaces of Wyoming montmorillonite by the cationic surfactants alkyl trimethyl, dialkyl dimethyl, and trialkyl methyl ammonium bromides, J. Colloid Interface Sci. 305 (2007) 150–158. [16] L. Zampori, P. Gallo Stampino, G. Dotelli, D. Botta, I. Natali Sora, M. Setti, Interlayer expansion of dimethyl ditallowylammonium montmorillonite as a function of 2-chloroaniline adsorption, Appl. Clay Sci. 41 (2008) 149–157. [17] R. Liu, R.L. Frost, W.N. Martens, Y. Yuan, Synthesis, characterization of mono, di and tri alkyl surfactant intercalated Wyoming montmorillonite for the removal of phenol from aqueous systems, J. Colloid Interface Sci. 327 (2008) 287–294. [18] R. Zhu, L. Zhu, J. Zhu, L. Xu, Structure of cetyltrimethylammonium intercalated hydrobiotite, Appl. Clay Sci. 42 (2008) 224–231. [19] Y. Xi, Z. Ding, H. He, R.L. Frost, Infrared spectroscopy of organoclays synthesized with the surfactant octadecyltrimethylammonium bromide, Spectrochim. Acta A 61 (2005) 515–525. [20] J. Zhu, H. He, L. Zhu, X. Wen, F. Deng, Characterization of organic phases in the interlayer of montmorillonite using FTIR and 13 C NMR, J. Colloid Interface Sci. 286 (2005) 239–244. [21] Y. Xi, R.L. Frost, H. He, T. Kloprogge, T. Bostrom, Modiﬁcation of Wyoming montmorillonite surfaces using a cationic surfactant, Langmuir 21 (2005) 8675–8680. [22] G. Socrates, Infrared and Raman Characteristic Group Frequencies, 3rd ed., John Wiley and Sons Ltd, New York, 2000. [23] B. Sarkar, M. Megharaj, Y. Xi, R. Naidu, Surface charge characteristics of organopalygorskites and adsorption of p-nitrophenol in ﬂow-through reactor system, Chem. Eng. J. (in press), doi:10.1016/j.cej.2011.05.062. [24] D. Zadaka, A. Radian, Y.G. Mishael, Applying zeta potential measurements to characterize the adsorption on montmorillonite of organic cations as monomers, micelles, or polymers, J. Colloid Interface Sci. 352 (2010) 171–177. [25] B. Bate, S.E. Burns, Effect of total organic carbon content and structure on the electrokinetic behavior of organoclay suspensions, J. Colloid Interface Sci. 343 (2010) 58–64. [26] I. Barshad, Signiﬁcance of the presence of exchangeable magnesium ions in acidiﬁed clays, Science 131 (1960) 988–990. [27] E. Tombácz, M. Szekeres, Colloidal behavior of aqueous montmorillonite suspensions: the speciﬁc role of pH in the presence of indifferent electrolytes, Appl. Clay Sci. 27 (2004) 75–94. [28] M. Rozalen, F.J. Huertas, P.V. Brady, Experimental study of the effect of pH and temperature on the kinetics of montmorillonite dissolution, Geochim. Cosmochim. Acta 73 (2009) 3752–3766. [29] H. Freundlich, Über die adsorption in lösungen, Z. Phys. Chem. (Leipzig) 57 (1906) 385–470. [30] R.-S. Juang, S.-H. Lin, K.-H. Tsao, Mechanism of sorption of phenols from aqueous solutions onto surfactant-modiﬁed montmorillonite, J. Colloid Interface Sci. 254 (2002) 234–241. [31] J.A. Smith, A. Galan, Sorption of nonionic organic contaminants to single and dual organic cation bentonites from water, Environ. Sci. Technol. 29 (1995) 685–692. [32] C.H. Ko, C. Fan, P.N. Chiang, M.K. Wang, K.C. Lin, p-Nitrophenol, phenol and aniline sorption by organo-clays, J. Hazard. Mater. 149 (2007) 275–282. [33] L. Zhu, B. Chen, X. Shen, Sorption of phenol, p-nitrophenol, and aniline to dual-cation organobentonites from water, Environ. Sci. Technol. 34 (2000) 468–475.

Journal of Hazardous Materials 195 (2011) 162–169

Contents lists available at ScienceDirect

Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat

Retention–oxidation–adsorption process for emergent treatment of organic liquid spills Xianjun Liu, Yu Li, Xingwang Zhang, Lecheng Lei ∗ Institute of Industrial Ecology and Environment, Zhejiang University, Hangzhou 310027, PR China

a r t i c l e

i n f o

Article history: Received 5 April 2011 Received in revised form 5 August 2011 Accepted 7 August 2011 Available online 11 August 2011 Keywords: Retention–oxidation–adsorption process Organic liquid spill Organobentonite Ferrate Ozone

a b s t r a c t The feasibility and effectiveness of retention–oxidation–adsorption process (ROA) for the elimination of organic contaminants induced by chemical accidents were investigated in this study. Organobentonites (DTMA-, TTA-, CTMA- and OTMA-bentonite), potassium ferrate (Fe(VI)), ozone and granular activated carbon (GAC) were used as rapid and efﬁcient materials in the treatment and recovery of organic liquid spills. Results indicated that the retention capacities of organobentonites (especially CTMA-bentonite) were much higher than that of natural bentonite towards the chosen organic compounds. Additionally, pH, oxidant dosage, initial concentration of contaminant and chemical structure had signiﬁcant inﬂuences on the effectiveness of the oxidation process. In a pilot-scale experiment, the ferrate/GAC (F/G) and ozone/GAC (O/G) processes made a comparatively good performance in the treatment of wastewater containing aniline or nitrobenzene, with the removal efﬁciencies of the contaminants greater than 80%. Overall, the ROA process showed a high efﬁciency and steady operation in the removal of hazardous organic liquids and subsequent clean up of the contaminated site. Crown Copyright © 2011 Published by Elsevier B.V. All rights reserved.

1. Introduction In recent years, there has been growing concern in China about the widespread occurrence of chemical incidents in industry and road transport [1]. Every year, sudden and unexpected contamination events (accidental leakage, spills, explosions, etc.) put large numbers of the public in the danger of exposure to hazardous organic pollutants (benzene, nitrobenzene, petroleum, etc.), which account for about 34% of the total involved pollutants [2]. Since the accident scenes and contaminants varied, the systematic study of remediation technologies and processes seems to be very difﬁcult and necessary, especially in China. When chemical incidents take place accidently, response actions must be quick and accurate to minimize the impact of the contaminants [3]. The method of retention is considered to be one of the most common processes for the removal of hazardous organic liquids and subsequent cleanup of the contaminated site, and many types of sorbents are developed accordingly. Among them, organobentonites, normally synthesized by cationic surfactants and bentonite with cationic exchanging, are widely used as effective sorbents for organic contaminants and numerous stud-

∗ Corresponding author. Tel.: +86 571 88273090; fax: +86 571 88273090. E-mail address: [email protected] (L. Lei).

ies have investigated their mechanisms in the past decades [4–6]. Ozonation [7–10] and ferrate oxidation [11–16] have also shown their own advantages in removal of a wide range of organic pollutants both in wastewater and drinking water. Meanwhile, activated carbon has proven to be one of the most effective adsorbent materials, which is widely used in the elimination and recovery of hazardous organic pollutants [17–19]. However, more information remains to be gathered to evaluate their application in organic contamination treatment. Considering the ideas addressed above, this research chose eight organic compounds as representative pollutants on the basis of their predominant occurrence in chemical accidents, including aromatics (benzene, toluene, o-xylene, nitrobenzene, chlorobenzene, aniline) and chlorophenols (2,4dichlorophenol, 2,4,6-trichlorophenol), and introduced an effective retention–oxidation–adsorption process for the emergent treatment of organic liquid spills induced by chemical accidents. The present study was conducted to investigate the effectiveness and the inﬂuence factors (sorbent material, solution pH, oxidant dosage, initial concentration of contaminant, chemical structure, etc.) of retention–oxidation–adsorption process, and to try to develop a mobile treatment system based on the ROA process for environmental emergency management departments, corporations that deal with organic chemicals, etc., which could be applied for the treatment of chemical incidents that occur in industry and road transport.

0304-3894/$ – see front matter Crown Copyright © 2011 Published by Elsevier B.V. All rights reserved. doi:10.1016/j.jhazmat.2011.08.018

X. Liu et al. / Journal of Hazardous Materials 195 (2011) 162–169

2. Experimental 2.1. Chemicals Benzene, toluene, o-xylene, nitrobenzene, chlorobenzene, aniline, 2,4-dichlorophenol, 2,4,6-trichlorophenol were purchased from Sinopharm Chemical Reagent Co. (China) and were of 99% purity or higher. Dodecyltrimethylammonium bromide (DTMAB), tetradecyltrimethylammonium bromide (TTAB), cetyltrimethylammonium bromide (CTMAB) and octadecytrimethylammonium bromide (OTMAB) of analytical grade were obtained from Shanghai Chemical Co., China. The natural bentonite composed primarily of Ca2+ montmorillonite was obtained from Inner Mongolia, China. The bentonite BET-N2 surface area, organic carbon content, and cation exchanged capacity (CEC) was 73.3 m2 g−1 , 0.17%, and 108 cmol kg−1 , respectively. Potassium ferrate (97%) was purchased from Sigma–Aldrich Co., USA. Ozone was generated from oxygen by an ozone generator (NPL5W, Green Continent Co., China) with a maximum ozone production rate of 5 g h−1 . All other chemicals were of reagent grade, and used as received. Doubly distilled water was used throughout the study except in the pilot-scale experiment. 2.2. Batch retention experiment In this investigation, DTMA-bentonite, TTA-bentonite, CTMAbentonite, OTMA-bentonite and natural bentonite were chosen as sorbents. The organobentonites were synthesized with the following procedure [20]: 30.0 g natural bentonite was mixed with 500 mL of an aqueous solution containing 9.99 g DTMAB, 10.9 g TTAB, 11.8 g CTMAB or 12.7 g OTMAB, respectively, equal to 100% of bentonite’s CEC. The mixture was stirred at 60 ◦ C for 4 h. The product was separated from water by vacuum ﬁltration and washed with distilled water six times, then dried at 70 ◦ C for 12 h. Finally, the organobentonite was grounded to pass a 100 mesh sieve. The retention capacity measurements were undertaken as follows [21]: 200 mL of pure pollutant liquid was poured into a 250 mL pyrex glass beaker. 5.0 g of sorbent was weighed and the value recorded. The sorbent material was spread evenly on a ﬁlter paper, placed into a circular brass wire mesh (2 mm2 ) basket ( 5.0 × 8.0 cm), and then lowered into the beaker so that the sorbent was completely covered by the liquid. After 15 min of immersion,

163

the basket with the sorbent was hung up for about 5 min to allow the excess liquid to be drained off. Then, the saturated sorbent was transferred to a suitable pre-weighed dish and weighed. Additionally, to evaluate the effect of retention time between the chosen pollutants and sorbents, the experiments were conducted with the same procedure other than the volume of pure organic liquid was reduced to 25 mL and the contact time varied from 1 to 30 min. All tests were carried out in triplicate. The retention capacity of the sorbents and the retention rate of organic liquids on a weight basis were calculated as follows: retention capacity (g g−1 ) = (S1 − S0 )/S0 , retention rate (%) = (S1 − S0 )/S2 × 100, where S0 is the initial dry weight of sorbent, S1 is the weight of the saturated sorbent, S2 is the weight of the pure organic liquid. 2.3. Degradation by ozonation and ferrate oxidation The experimental set-up used for wastewater treatment by ozonation consisted of a pressurized oxygen gas cylinder, an ozone generator, a gas ﬂow meter, and a reaction vessel ( 5 × 50 cm) of about 1 L in volume. The reactor was equipped with a gas diffuser at the bottom to sparge the ozone/oxygen stream into the water. The exhaust gas vented from the top of the reactor was captured in a pair of adsorption bottles containing 2% (w/v) KI solution. 500 mL of a solution that contained different contaminants at various concentrations was adjusted to the desired pH value by using either sulfuric acid or sodium hydroxide, and added to the reactor before the ozone bubbled. In this work, the ozone dosage added into the system was controlled between 5.59 mg min−1 and 35.1 mg min−1 by a gas ﬂow meter. Samples were collected at ﬁxed intervals and 1.0 mol L−1 Na2 SO3 was added to remove the residual ozone and organic radicals prior to analysis. Batch Fe(VI) oxidation experiments were conducted in 100 mL conical ﬂasks in which 50 mL of a solution was stirred vigorously by magnetic stirrer. 1.0 mol L−1 Na2 SO3 was added into the samples immediately to stop any further reaction. After the quenching, samples were centrifuged at 8000 rpm for 10 min, and then ﬁltered with 0.45 ␮m ﬁlters before analysis. All the experiments were conducted in triplicate at room temperature (23 ± 2 ◦ C). The elimination of the target pollutant through volatilization was evaluated and deducted. Removal is deﬁned as (%) = (C0 − C)/C0 × 100, where C0 is the initial contaminant concentration or TOC.

Fig. 1. The schematic diagram of the retention–oxidation–adsorption system.

X. Liu et al. / Journal of Hazardous Materials 195 (2011) 162–169

10

2.4. Pilot-scale experiment

2.5. Analytical methods The concentration of Fe(VI) in the aqueous solution was determined by monitoring the absorbance at 510 nm using a GBC Cintra 303 spectrophotometer (Australia) [22]. The ozone concentration in the inlet and outlet gas was determined by a standard iodometry method [23]. The dissolved ozone in solution was measured by the indigo method [24]. TOC was measured by a TOC analyzer (Shimadzu TOC-VCPH , Japan) and pH was measured by a PHSJ-3 F instrument (Shanghai Precision & Scientiﬁc Instrument Co., China). The determination of benzene, toluene and o-xylene was carried out by a gas chromatograph (FULI 9790, Fuli Analytical Instrument Co., China) equipped with a SE30 capillary column (30 m × 0.32 mm, 0.33 ␮m ﬁlm), a headspace sampler DK3001A (Zhongxinghuili Science and Technology Development Co., China), and an ECD detector according to the method

Natural bentonite

9

DTMA-bentonite

8 -1

Retention capacity (g g )

TTA-bentonite CTMA-bentonite

7

OTMA-bentonite

6 5 4 3 2

N itr ob en ze ne

en e

Ch lo ro be nz

oX yl en e

0

To lu en e

1

A ni lin e

According to the results of the lab experiment, a pilot-scale experiment was performed to simulate the treatment and recovery of an accidental organic liquid spill. The indoor temperature was about 20 ± 2 ◦ C during the period of the experiment. Fig. 1 shows the schematic diagram of the retention–oxidation–adsorption system used in this study. The pilot-scale retention experiments were conducted in a tank (2 × 2 × 0.5 m) made of polyurethane. With the consideration of environment protection, a certain concentration (about 50 g L−1 with most of the pollutant remained undissolved) of efﬂuent containing the chosen pollutant (aniline or nitrobenzene) was used as the treatment target instead of pure organic liquid. When all the efﬂuent stored in a polyurethane cylinder ( 0.5 × 0.5 m, working volume 20 L) was gradually released through a needle valve into the tank, and spread over most of the closed area, the CTMAbentonite was quickly transferred through a plastic ﬂexible pipe from its container, a stainless steel cylinder ( 1.0 × 2 m), and spread evenly across the polluted area by an air-operated powder pump (650780-X, Ingersoll Rand Company, USA). The consumed weight of CTMA-bentonite was calculated by the product of jet velocity (kg min−1 ) and handling time (min). After that, the waste solid was removed for ﬁnal disposal and tap water was used to ﬂush the contaminated site. All the wastewater was pumped directly into an adjusting tank ( 2.0 × 1.0 m), mixed with tap water for a working volume of 2.0 m3 and conditioned by agitation (300 rpm) for 4 h. Before further treatment, a small sample was withdrawn and the concentration of the pollutant was measured subsequently. Then, the wastewater was treated by Fe(VI) oxidation or ozonation. In Fe(VI) oxidation, an optimal amount of potassium ferrate was added directly into the adjusting tank, reacting with the contaminant in the condition of agitation (300 rpm) for 20 min. The ozonation system consisted of an air compressor, a refrigeration ﬁlter drier, and an ozone generator (maximum ozone output: 100 g h−1 ) supplied by Shandong Zhiwei Science and Technology Co., China. The wastewater with a ﬂow rate of 1 m3 h−1 was pumped into a stainless steel reactor ( 0.5 × 2 m) which was a magniﬁcation of the lab-scale reactor. The effective volume of the reactor and the actual ozone production were approximately 300 L and 76 g h−1 , respectively. Both of the oxidation processes were followed by an activated carbon adsorption process since most of the pollutants could not be completely degraded. The GAC adsorption column ( 0.5 × 1.5 m) consisted of a 30 cm high gravel layer in the bottom and a 100 cm carbon layer above the gravel layer. The ﬂow rate of the efﬂuent in the packed bed was ﬁxed at 1.0 m3 h−1 . The pollutant concentration was measured at the inlet and outlet of each reactor by the methods mentioned below.

Be nz en e

164

Fig. 2. The retention capacity (g g−1 ) of natural bentonite and organobentonites towards the chosen pollutants.

reported by Menendez et al. [25]. The other compounds were measured by high performance liquid chromatography (HPLC) using an Agilent instrument (1200 series, Agilent, USA) equipped with an Eclipse XDB-C18 column (25 ␮m particle size, 250 × 4.6 mm i.d.). The proportions of the speciﬁc mobile phase and UV detection wavelength used for each compound by HPLC were as follows: nitrobenzene, methanol/water-70/30 (v/v), UV-262 nm; chlorobenzene, methanol/water-80/20 (v/v), UV-212 nm; aniline, methanol/water-40/60 (v/v), UV-280 nm; 2,4-dichlorophenol, methanol/water-80/20 (v/v), UV-280 nm; 2,4,6-trichlorophenol, methanol/water-80/20 (v/v), UV-295 nm. The ﬂow rate was 1 mL min−1 , and injection volume was 20 ␮L. 3. Results and discussion 3.1. Retention by organobentonites 3.1.1. Retention capacity of organobentonites As shown in Fig. 2, the retention capacities of organobentonites were much higher than that of natural bentonite towards the chosen organic compounds, and compared with other three organobentonites (DTMA-, TTA- and OTMA-bentonite), CTMAbentonite stood out as the best performing sorbent. As for CTMA-bentonite, benzene, toluene, o-xylene and chlorobenzene showed similar retention capacities (5.14–5.73 g g−1 ), while aniline (2.71 g g−1 ) and nitrobenzene (9.01 g g−1 ) showed the lowest and highest retention capacities, respectively. This trend coincides with the hydrophobicity of the chemicals and may be greatly inﬂuenced by the viscosity and other factors [5]. 3.1.2. Effect of retention time Based on the retention capacity test, the effect of retention time between toxic chemicals and CTMA-bentonite was investigated. The simulated leakage volume of organic liquid was 25 mL and the dosage of CTMA-bentonite was ﬁxed at 200 g L−1 , which was more than the theoretical optimal value. As shown in Fig. 3, the CTMA-bentonite adsorbed the pollutants (aniline, benzene, chlorobenzene and nitrobenzene) very quickly, and the retention rates were all over 80% within 4 min and changed little after 6 min. It was noticeable that about 50% of the hazardous chemicals were retained in 1 min, which indicates the CTMA-bentonite could be used as a rapid and efﬁcient material for preventing the dispersal of hazardous organic liquids.

X. Liu et al. / Journal of Hazardous Materials 195 (2011) 162–169

95 90

Retention rate (%)

85 80 75 Aniline Benzene Chlorobenzene Nitrobenzene

70 65 60 55 50 45

1

2

3

4

5

6

7

8

9

10

Retention time (min) Fig. 3. Effect of retention time on the retention rate of organic liquid by CTMAbentonite.

3.2. Ozonation and ferrate oxidation 3.2.1. Effect of initial pH value Effect of initial pH values of the aqueous solution on the degradation of aniline was investigated at different pH in the range of 3.0–11.0 with aniline initial concentration of 10.7 mM. The ozone dosage was kept constant at 26.5 mg min−1 . As shown in Fig. 4(a), there was an increase in the degradation efﬁciency with an increas-

a

100 90

Aniline removal (%)

80 70 60 50 pH=3.0 pH=5.0 pH=7.3 pH=9.0 pH=11.0

40 30 20 10 0

0

5

10

15

20

25

30

35

40

Reaction time (min)

b

70

Aniline removal (%)

60 50 40 30

pH=3.0 pH=5.0 pH=7.3 pH=9.0

20 10 0

0

5

10

15

20

25

30

35

40

Reaction time(min) Fig. 4. Effect of initial pH on the degradation efﬁciency of aniline by ozone (a) and ferrate (b). [aniline] = 10.7 mM, [ozone] = 26.5 mg min−1 , [ferrate] = 15.1 mM.

165

ing pH value in the range of 3.0–7.3 (after which a decrease was observed). When the pH value was 7.3 (the initial pH value without adjustment), the degradation of aniline reached 90% within 20 min. The intermediate products can also inﬂuence the reactions by consuming ozone and • OH [26]. Sarasa et al. [27] have conﬁrmed that nitrobenzene is the major aromatic by-product formed after ozonation of aniline, which is one of the target compounds for remove in this study. Under the conditions of this experiment, nitrobenzene was gradually accumulated in solution, and was detected in higher concentrations at acidic conditions. The ﬁnal concentrations of nitrobenzene during the reactions at pH 3.0, 5.0, 7.3, 9.0 and 11.0 were 3.2 × 10−2 , 2.1 × 10−2 , 1.1 × 10−2 , 5.3 × 10−3 and 1.2 × 10−2 mM, respectively. Meanwhile, the TOC removal in 40 min at pH 3.0, 5.0, 7.3, 9.0 and 11.0 was 16%, 25%, 34%, 41% and 28%, respectively. It can be seen that a considerable high pH (pH 7.3) of the solution can beneﬁt the mineralization of aniline. The inﬂuence of pH on the Fe(VI) oxidation process for aniline degradation was investigated with potassium ferrate concentration of 15.1 mM. As presented in Fig. 4(b), the degradation efﬁciencies of aniline at different pH values showed no signiﬁcant differences. When the pH value was 3.0, 5.0, 7.3 and 9.0, the degradation efﬁciency of aniline after 20 min of treatment was 61%, 68%, 62% and 60%, respectively. At the same time, the TOC removal at pH 3.0, 5.0, 7.3 and 9.0 was found to be 27%, 31%, 22% and 20%, respectively. Nitrobenzene was also detected as a by-product during the ferrate oxidation of aniline, and its concentration increased quickly in the ﬁrst 10–20 min, then decreased slowly in the residual time. For example, when the initial pH value was 5.0, the nitrobenzene concentration reached its maximum value (4.2 × 10−2 mM) in the ﬁrst 15 min, and its ﬁnal concentration after 40 min of treatment was 3.0 × 10−2 mM. The solution pH also played an important part in the formation and degradation of nitrobenzene. The ﬁnal concentrations of nitrobenzene at pH 3.0, 7.3 and 9.0 were 4.4 × 10−3 , 1.5 × 10−2 and 3.7 × 10−2 mM, respectively. As for Fe(VI), its reactivity with reactants varies signiﬁcantly with pH [28]. It is highly unstable but more powerful at pH < 6. In contrast, it becomes more chemically stable but has a weaker oxidizing ability at pH > 9 [29]. In these experiments, the ﬁnal decomposition rates of ferrate at pH 3.0, 5.0, 7.3 and 9.0 were 87%, 72%, 48% and 44%, respectively. Since the removal of aniline showed no signiﬁcant difference between pH 5.0 and 7.3, a pH of 7.3 was chosen as the optimal pH for Fe(VI) oxidation of aniline in this study. By avoiding the pH regulation there is a considerable saving of time and money, which is very important for wastewater treatment in cases of emergency. 3.2.2. Effect of oxidant dose The dosage of oxidant was found to be the most important factor in achieving better degradation of aniline. To evaluate the effect of ozone dose on the degradation of aniline, experiments were conducted over a range of ozone doses of 5.59–35.1 mg min−1 under the condition of optimal pH 7.3 and the results were shown in Fig. 5(a). It was clear that the degradation efﬁciency of aniline increased with the increasing ozone dosage. When the ozone dosages were 5.59, 21.6, 26.5, 31.1 and 35.1 mg min−1 , the aniline removal in 20 min was found to be 33%, 75%, 90%, 96% and 98%, respectively. The TOC removal was 8.8%, 17%, 20%, 33% and 36%, respectively. During the reaction time, the average ozone ﬂuxes in the outlet gas were about 0.024, 0.24, 1.0, 2.3 and 4.7 mg min−1 , so the utilization rate of ozone decreased with the increasing ozone dosage. It was also found that the ﬁnal nitrobenzene concentration in solution decreased when ozone dosage was higher than 21.6 mg min−1 , though its maximum concentration during the reaction increased with the increasing ozone dosage. The ﬁnal nitrobenzene concentrations with the ozone dosages of 21.6, 26.5, 31.1 and 35.1 mg min−1 were 1.1 × 10−2 , 1.5 × 10−2 , 4.2 × 10−3 , 3.7 × 10−3 mM, respectively. Therefore it can be concluded that

166

X. Liu et al. / Journal of Hazardous Materials 195 (2011) 162–169

a

a

100

90

80 70

[O3 ]=5.59 mg min

60

[O3]=21.6 mg min

Aniline removal (%)

Aniline removal (%)

90

-1

-1

50 40 30

-1

[O3]=26.5 mg min

20

5

10

15

20

25

30

70 60 50 40

20 10

35

[Aniline]=2.68 mM [Aniline]=5.37 mM [Aniline]=8.05 mM [Aniline]=10.7 mM

30

-1

[O3]=35.1 mg min 0

80

-1

[O3]=31.1 mg min

10 0

100

0

40

0

5

10

Reaction time (min) 100

b

90

25

30

100 90

80 70 60 50 40 30

[Ferrate]=5.05 mM [Ferrate]=15.1 mM [Ferrate]=25.2 mM [Ferrate]=35.3 mM

20 10 0

20

Reaction time(min)

Aniline removal (%)

Aniline removal (%)

b

15

0

5

10

15

20

25

30

35

70 60 50

[Aniline]=2.68 mM [Aniline]=5.37 mM [Aniline]=8.05 mM [Aniline]=10.7 mM

40 30 20 10

40

0 0

Reaction time (min) Fig. 5. Effect of oxidant dosage on the degradation efﬁciency of aniline by ozone (a) and ferrate (b). [aniline] = 10.7 mM, pH = 7.3.

an appropriate ozone dosage can accelerate the reaction rate of ozonation and shorten the overall reaction time. The degradation efﬁciencies with different dosages of Fe(VI) were also tested under the condition of optimal pH 7.3. As shown in Fig. 5(b), the aniline degradation efﬁciency increased by increasing the Fe(VI) dosage. When the Fe(VI) dosage was 25.2 mM, 81% of aniline was degraded after 20 min. The degradation efﬁciency of aniline changed little after 20 min and about 69% of aniline was degraded during the ﬁrst 5 min. It was found that the decomposition rates of Fe(VI) in 5, 20 and 40 min were 45%, 56% and 65%, respectively. However, when the Fe(VI) dosage was 35.3 mM, the degradation efﬁciency of aniline after 20 min of treatment was 87%, and the decomposition rates of Fe(VI) in 5, 20 and 40 min were 47%, 66% and 79%, respectively. In this experiment, the ﬁnal nitrobenzene concentration and TOC removal both increased slightly with the increasing of Fe(VI) dosage. With a Fe(VI) dosage of 5.05, 15.1, 25.2 and 35.3 mM, the ﬁnal nitrobenzene concentrations were 4.1 × 10−3 , 3.0 × 10−2 , 3.5 × 10−2 , 4.9 × 10−2 mM, while the TOC removals after 40 min of treatment were 16%, 22%, 27% and 33%, respectively. This can be explained by the kinetics of Fe(VI) ion reactions in aqueous solutions. In general, the Fe(VI) ion decrease is caused by its reaction with organic compounds and by its selfdecay. The spontaneous decomposition of Fe(VI) in an aqueous solution is described by the following reaction (Eq. (1)) [30]. 2FeO4 2− + 5H2 O → 2Fe(OH)3 + 3/2O2 + 4OH−

80

(1)

The decomposition rate of Fe(VI) is dependent upon pH, initial Fe(VI) concentration, co-existing ions and temperature [31,32]. The iron(III) hydroxide formed upon decomposition (reaction (1))

5

10

15

20

25

30

35

40

Reaction time (min) Fig. 6. Effect of initial contaminant concentration on the degradation efﬁciency of aniline by ozone (a) and ferrate (b). pH = 6.9–7.3, [ozone] = 31.1 mg min−1 , [ferrate] = 25.2 mM.

also controls the decomposition of Fe(VI) [33]. When the concentration of Fe(VI) is higher than the optimal dose, its reaction with water cannot be neglected [14]. Considering economic and technical issues, the optimal dose of Fe(VI) to be added into the aniline solution with the concentration of 10.7 mM was 25.2 mM (the molar ratio of Fe(VI) to aniline = 2.4). 3.2.3. Effect of initial contaminant concentration The effects of the initial contaminant concentrations on the oxidation processes were investigated. In these experiments, the initial pH values varied in the range of 6.9–7.3 due to different aniline concentrations. Considering the mass transfer and economic efﬁciency, ozone dosage was ﬁxed at 31.1 mg min−1 . Fig. 6(a) shows the results of aniline degradation by ozonation for various initial aniline concentrations. Clearly, a higher initial concentration of aniline in the range of 2.68–10.7 mM resulted in a lower degradation efﬁciency and TOC removal in 30 min of oxidation. However, as the initial aniline concentration increased, the ozone utilization rate and ﬁnal nitrobenzene concentration in solution increased signiﬁcantly. When the initial aniline concentration increased from 2.68 to 10.7 mM, the average ozone ﬂux in the outlet gas decreased from 6.8 to 2.3 mg min−1 , and the ﬁnal nitrobenzene concentration in 30 min increased from 0 to 2.9 × 10−2 mM. As shown in Fig. 6(b), when the Fe(VI) dosage was 25.2 mM there was a small increase in degradation efﬁciency as the initial concentration of aniline decreased. The degradation efﬁciency of aniline with an initial concentration of 2.68 mM was 96%, while 81% of

X. Liu et al. / Journal of Hazardous Materials 195 (2011) 162–169 Table 1 Half life and pseudo-ﬁrst-order rate constant for selected target compounds by ozonation.

167

Table 2 The optimal molar ratio of ferrate to contaminant and its corresponding degradation efﬁciency by ferrate oxidation.

Compound

pH

t1/2 (min)

Observed 1st order reaction rate constant, k (min−1 )

R2

Compound

pH

The optimal molar ratio of ferrate to contaminant

Degradation efﬁciency (%)

Nitrobenzene Chlorobenzene Toluene Benzene o-Xylene 2,4-Dichlorophenol Aniline 2,4,6-Trichlorophenol

5.7 6.1 5.6 5.8 5.6 5.4 6.6 4.5

5.1 4.4 4.3 2.1 2.0 1.3 1.1 0.9

0.136 0.158 0.160 0.331 0.353 0.522 0.643 0.791

0.992 0.996 0.991 0.998 0.999 0.997 0.997 0.995

Benzene Chlorobenzene Toluene o-Xylene Aniline Nitrobenzene 2,4-Dichlorophenol 2,4,6-Trichlorophenol

5.8 6.1 5.6 5.6 6.6 5.7 5.4 4.5

0.10 0.23 0.47 0.54 3.0 3.1 5.6 12

82 84 88 83 83 47 57 31

aniline was degraded as the initial concentration increased to 10.7 mM in 20 min. The ﬁnal nitrobenzene concentration with an initial aniline concentration of 2.68, 5.37, 8.05 and 10.7 mM was 1.6 × 10−2 , 3.8 × 10−2 , 4.1 × 10−2 and 3.5 × 10−2 mM, respectively. 3.2.4. Effect of chemical structure The kinetics of ozone reactions with organic and inorganic compounds are typically second order [34], and can be described by −

d[M] = (kO3 [O3 ] + k• OH [• OH])[M] dt

(2)

Here [M] is the target compound, [O3 ] and [• OH] are concentrations (mol L−1 ) of dissolved ozone and hydroxyl radical, t is time (s), kO3 and k• OH represent reaction rate constants for molecular ozone and the hydroxyl radical. The ozone concentration in water is limited by its solubility and is inﬂuenced by intermediates and ﬁnal products that may be formed during the ozonation of a contaminant [35]. It was observed that the ozone concentration in the liquid phase increased slowly with reaction time. For example, with a pH of 7.3, aniline concentration of 10.7 mM, and ozone dosage of 31.1 mg min−1 , the dissolved ozone in 2.5, 5 and 10 min was 6.1, 7.7 and 10.5 mg L−1 , respectively. However, for speciﬁc processing of the system and gas phase ozone concentration, the dissolved ozone still could be estimated as ﬁxed values in a very short reaction time. Therefore the reaction kinetics can be treated as a pseudo-ﬁrstorder reaction (the inﬂuence of the hydroxyl radical was negligible at neutral and acidic conditions) and the integration of Eq. (2) is simpliﬁed as follows: −ln

[M] t

[M]0

= kt

(3)

As shown in Table 1, the time for 50% contaminant removal and observed ﬁrst-order reaction rate constants are used to evaluate the degradation efﬁciency for selected contaminants without pH adjustment. In this set of experiments, contaminants with 100 mg L−1 initial concentration were degraded with the ozone dosage of 31.1 mg min−1 . Among the selected compounds, nitrobenzene, chlorobenzene and toluene were shown to be more difﬁcult to degrade by ozone. The observed ﬁrst-order rate constants for degradation of these compounds were about only 17–20% of 2,4,6-trichlorophenol. In this study, we found aniline and chlorophenols were easy and fast to eliminate by ozonation, while benzene and o-xylene had a moderate reaction rate. Moreover, the degradation of 2,4,6-trichlorophenol was faster than that of 2,4-dichlorophenol, since the presence of chlorine enhanced the dechlorination step. This ﬁnding is generally in accordance with previous established literatures [36–38]. According to the above results, it can be seen that the chemical structures of the compounds have a dominant inﬂuence on the oxidation process.

The kinetic studies for the reaction of Fe(VI) with a large number of organic compounds were reported to be ﬁrst order with respect to both reactants [39]: −d[Fe(VI)] = k[Fe(VI)][M] dt

(4)

The values of second-order rate constant (k) for the reaction of Fe(VI) with aniline were determined as a function of pH in a few studies [40–42]. For example, with a Fe(VI) concentration of 90 ␮M at pH 7.0 (25 ◦ C), the second-order rate constant (k) and half-life (t1/2 ) of this reaction was reported to be 6.6 × 103 M−1 s−1 , 2.1 s, respectively [42]. Potassium ferrate has shown an appreciable reactivity with a variety of compounds. Most of the half-lives for reactions of Fe(VI) with possible contaminants under excess Fe(VI) conditions are reported to be in seconds to minutes [39], and a similar result was obtained in this study. As observed in Fig. 5(b), the degradation efﬁciency increased with the molar ratio of Fe(VI) to contaminant, and further increases in the extent of contaminant degradation were likely to be minor at molar ratios greater than optimal. This behavior was repeated for all of the compounds studied. Table 2 shows the optimal molar ratios of Fe(VI) to contaminants, and the corresponding degradation efﬁciencies of the selected compounds with 100 mg L−1 initial concentration after 20 min of treatment. Considering the dosage of Fe(VI) and the oxidative elimination, the results clearly show that the oxidation reactivity increases in the order of benzene > chlorobenzene > toluene > oxylene > aniline. Nitrobenzene and chlorophenols were found to be hard to degrade in acidic and neutral conditions. The electron attracting effect of nitro-substituting groups and chlorine atoms is believed to make the compound more electronically stable [29], and hence, less reactive. For chlorophenols, the electronegativity of the molecule increases with the degree of chlorine substitution, and thus the degradation of 2,4,6-trichlorophenol is much more difﬁcult than 2,4-dichlorophenol, even at a higher molar ratio of Fe(VI) to contaminant, as shown in Table 2. 3.3. Simulated treatment by ROA process In the pilot experiment, the amounts of CTMA-bentonite consumed were ﬂexible due to operational factors. Approximately 26 kg and 21 kg of CTMA-bentonite were used in the retention of wastewater containing aniline or nitrobenzene, respectively. Meanwhile, the concentrations of aniline and nitrobenzene in the adjusting tank were found to be 72 mg L−1 and 43 mg L−1 , respectively. Therefore more than 85% of the pollutants were removed by the retention process. The results of pollutant and TOC removal in each treatment unit of the experimental system are shown in Fig. 7(a) and (b). With the methods of Fe(VI) oxidation, ozonation and GAC adsorption, aniline removal efﬁciencies could reach 78%, 62% and 55%, respectively. However, the TOC removal efﬁciencies were

168

X. Liu et al. / Journal of Hazardous Materials 195 (2011) 162–169

a

Aniline TOC

80

Removal (%)

Additionally, compared with ozonation, the advantage of the Fe(VI) treatment is its easy adaptation to variations in wastewater quality and quantity. Ozonation also shows its advantage over Fe(VI) oxidation in the treatment of chlorophenols, and if given sufﬁcient time, complete degradation is possible. In the pilot-scale experiment, The F/G and O/G processes made a comparatively good performance in the treatment of wastewater containing aniline or nitrobenzene, with the removal efﬁciencies of the contaminants greater than 80%. Further research is needed to enhance the removal efﬁciency and cost effectiveness of these processes. Meanwhile, much work should be carried out to assess the feasibility and application of the combined processes in pilot or full-scale emergent treatment of organic liquid spills.

100

60 40 20 0

Ferrate

Ozone

GAC

F/G

O/G

Acknowledgements

b

Nitrobenzene TOC

80

Removal (%)

This work was ﬁnancially supported by the National High Technology Research and Development Program of China (No. 2007AA06A409). The authors are grateful to Scott Loughery at Duke University for reviewing the manuscript. This paper has beneﬁtted from comments of four anonymous reviewers.

100

60

References

40

20

0

Ferrate

Ozone

GAC

F/G

O/G

Fig. 7. The removal of pollutant and TOC in each treatment unit. [ozone] = 76 g h−1 , the molar ratio of ferrate to contaminant = 5.0, pH = 6.8–7.0. (a) [aniline] = 0.773 mM, [TOC] = 64.4 mg L−1 . (b) [nitrobenzene] = 0.406 mM, [TOC] = 31.5 mg L−1 .

only 23%, 12% and 48%, respectively. The degradation efﬁciency of ozonation was restricted to the mass transfer and the hydraulic retention time, which is beyond the scope of this paper. It is noticeable that there was a great improvement in aniline and TOC removal efﬁciencies with the Fe(VI) oxidation/GAC adsorption treatment processes, as well as the ozonation/GAC adsorption processes. In the lab-scale study, it was observed that nitrobenzene was hard to degrade by Fe(VI) oxidation and ozonation. However, activated carbon has proven to be an effective sorbent for nitrobenzene elimination in many publications [43,44]. In this study, using GAC adsorption, the nitrobenzene and TOC removal efﬁciencies were as high as 70%, 65%, respectively, and with the treatment of F/G and O/G processes, nitrobenzene removal efﬁciencies were both over 80%. 4. Conclusions the present study, we introduce a In retention–oxidation–adsorption process to remedy the contaminated environment caused by organic spills. The results showed that the retention capacities of organobentonites were much higher than that of natural bentonite with respect to the chosen organic compounds. CTMA-bentonite was successfully used as a rapid and efﬁcient material for preventing the dispersal of hazardous liquids in the pilot scale experiment. As a subsequent treatment, ozonation and Fe(VI) oxidation have proven to be efﬁcient, rapid processes for the elimination of selected pollutants in wastewater. The results indicated that pH, oxidant dosage, initial concentration of contaminant, and chemical structure had signiﬁcant inﬂuences on the effectiveness of degradation.

[1] J. Yang, F.Y. Li, J.B. Zhou, L. Zhang, L. Huang, J. Bi, A survey on hazardous materials accidents during road transport in China from 2000 to 2008, J. Hazard. Mater. 184 (2010) 647–653. [2] Y. Hou, T.Z. Zhang, Evaluation of major polluting accidents in China-results and perspectives, J. Hazard. Mater. 168 (2009) 670–673. [3] L. Alfonso, A. Jonoski, D. Solomatine, Multiobjective optimization of operational responses for contaminant ﬂushing in water distribution networks, J. Water Res. Pl.-ASCE 136 (2010) 48–58. [4] J.A. Smith, P.R. Jaffe, Benzene transport through landﬁll liners containing organophilic bentonite, J. Environ. Eng.-ASCE 120 (1994) 1559–1577. [5] L.Z. Zhu, B.L. Chen, X.Y. Shen, Sorption of phenol, p-nitrophenol, and aniline to dual-cation organobentonites from water, Environ. Sci. Technol. 34 (2000) 468–475. [6] R. Koswojo, R.P. Utomo, Y.H. Ju, A. Ayucitra, F.E. Soetaredjo, J. Sunarso, S. Ismadji, Acid green 25 removal from wastewater by organo-bentonite from Pacitan, Appl. Clay Sci. 48 (2010) 81–86. [7] W.R. Zhao, F.F. Liu, Y. Yang, M. Tan, D.Y. Zhao, Ozonation of cationic Red XGRL in aqueous solution: kinetics and modeling, J. Hazard. Mater. 187 (2011) 526–533. [8] A.H. Konsowa, M.E. Ossman, Y.S. Chen, J.C. Crittenden, Decolorization of industrial wastewater by ozonation followed by adsorption on activated carbon, J. Hazard. Mater. 176 (2010) 181–185. [9] R. Broseus, S. Vincent, K. Aboulfadl, A. Daneshvar, S. Sauve, B. Barbeau, M. Prevost, Ozone oxidation of pharmaceuticals, endocrine disruptors and pesticides during drinking water treatment, Water Res. 43 (2009) 4707–4717. [10] K. Ikehata, M. Gamal El-Din, S.A. Snyder, Ozonation and advanced oxidation treatment of emerging organic pollutants in water and wastewater, Ozone Sci. Eng. 30 (2008) 21–26. [11] B. Yang, G.G. Ying, L.J. Zhang, L.J. Zhou, S. Liu, Y.X. Fang, Kinetics modeling and reaction mechanism of ferrate(VI) oxidation of benzotriazoles, Water Res. 45 (2011) 2261–2269. [12] V.K. Sharma, G.W. Luther III, F.J. Millero, Mechanisms of oxidation of organosulfur compounds by ferrate(VI), Chemosphere 82 (2011) 1083–1089. [13] I. Ciabatti, F. Tognotti, L. Lombardi, Treatment and reuse of dyeing efﬂuents by potassium ferrate, Desalination 250 (2010) 222–228. [14] N. Noorhasan, B. Patel, V.K. Sharma, Ferrate(VI) oxidation of glycine and glycylglycine: kinetics and products, Water Res. 44 (2010) 927–935. [15] V.K. Sharma, Oxidation of nitrogen-containing pollutants by novel ferrate(VI) technology: a review, J. Environ. Sci. Health A 45 (2010) 645–667. [16] Y.Y. Eng, V.K. Sharma, A.K. Ray, Ferrate(VI): green chemistry oxidant for degradation of cationic surfactant, Chemosphere 63 (2006) 1785–1790. [17] L.Y. Huang, Z.D. Yang, B.J. Li, J.A. Hu, W. Zhang, W.C. Ying, Granular activated carbon adsorption process for removing trichloroethylene from groundwater, AlChE J. 57 (2011) 542–550. [18] S.J. Zhang, T. Shao, H.S. Kose, T. Karanﬁl, Adsorption of aromatic compounds by carbonaceous adsorbents: a comparative study on granular activated carbon, activated carbon ﬁber, and carbon nanotubes, Environ. Sci. Technol. 44 (2010) 6377–6383. [19] A. Dabrowski, P. Podkoscielny, Z. Hubicki, M. Barczak, Adsorption of phenolic compounds by activated carbon—a critical review, Chemosphere 58 (2005) 1049–1070. [20] B.L. Chen, W.H. Huang, J.F. Mao, S.F. Lv, Enhanced sorption of naphthalene and nitroaromatic compounds to bentonite by potassium and cetyltrimethylammonium cations, J. Hazard. Mater. 158 (2008) 116–123.

X. Liu et al. / Journal of Hazardous Materials 195 (2011) 162–169 [21] O. Carmody, R. Frost, Y.F. Xi, S. Kokot, Adsorption of hydrocarbons on organoclays-implications for oil spill remediation, J. Colloid Interface Sci. 305 (2007) 17–24. [22] Y. Lee, J. Yoon, U.V. Gunten, Spectrophotometric determination of ferrate (Fe(VI)) in water by ABTS, Water Res. 39 (2005) 1946–1953. [23] K. Rakness, G. Gordon, B. Langlais, W. Masschelein, N. Matsumoto, Y. Richard, C.M. Robson, I. Somiya, Guideline for measurement of ozone concentration in the process gas from an ozone generator, Ozone Sci. Eng. 18 (1996) 209–229. [24] H. Bader, J. Hoigne, Determination of ozone in water by the indigo method, Water Res. 15 (1981) 449–456. [25] J.C.F. Menendez, M.L.F. Sanchez, J.E.S. Uria, E.F. Martinez, A. Sanz-Medel, Static headspace, solid-phase microextraction and headspace solid-phase microextraction for BTEX determination in aqueous samples by gas chromatography, Anal. Chim. Acta 415 (2000) 9–20. [26] F.J. Beltran, J. Rivas, P.M. Alvarez, M.A. Alonso, A kinetic model for advanced oxidation processes of aromatic hydrocarbons in water: application to phenanthrene and nitrobenznene, Ind. Eng. Chem. Res. 38 (1999) 4189–4199. [27] J. Sarasa, S. Cortes, P. Ormad, R. Gracia, J.L. Ovelleiro, Study of the aromatic byproducts formed from ozonation of anilines in aqueous solution, Water Res. 36 (2002) 3035–3044. [28] K. Osathaphan, P. Tiyanont, R.A. Yngard, V.K. Sharma, Removal of cyanide in Ni(II)–cyanide, Ni(II)–cyanide–EDTA, and electroplating rinse wastewater by ferrate(VI), Water Air Soil Pollut. 219 (2011) 527–534. [29] N. Graham, C.C. Jiang, X.Z. Li, J.Q. Jiang, J. Ma, The inﬂuence of pH on the degradation of phenol and chlorophenols by potassium ferrate, Chemosphere 56 (2004) 949–956. [30] R.H. Wood, The heat, free energy and entropy of the ferrate(VI) ion, J. Am. Chem. Soc. 80 (1958) 2038–2041. [31] T. Schink, T.D. Waite, Inactivation of F2 virus with ferrate(VI), Water Res. 14 (1980) 1705–1717. [32] M.D. Johnson, K.D. Sharma, Kinetics and mechanism of the reduction of ferrate by one-electron reductants, Inorg. Chim. Acta 293 (1999) 229–233.

169

[33] V.K. Sharma, Oxidation of inorganic contaminants by ferrates (VI, V, and IV) kinetics and mechanisms: a review, J. Environ. Manage. 92 (2011) 1051–1073. [34] M.M. Huber, S. Canonica, G.Y. Park, U. Von Gunten, Oxidation of pharmaceuticals during ozonation and advanced oxidation processes, Environ. Sci. Technol. 37 (2003) 1016–1024. [35] E. Kusvurana, O. Gulnazb, A. Samilc, O. Yildirima, Decolorization of malachite green, decolorization kinetics and stoichiometry of ozone-malachite green and removal of antibacterial activity with ozonation processes, J. Hazard. Mater. 186 (2011) 133–143. [36] J. Hoigne, H. Bader, Rate constants of reactions of ozone with organic and inorganic-compounds in water. 1. Non-dissociating organic-compounds, Water Res. 17 (1983) 173–183. [37] J. Hoigne, H. Bader, Rate constants of reactions of ozone with organic and inorganic-compounds in water. 2. Dissociating organic-compounds, Water Res. 17 (1983) 185–194. [38] F.J. Benitez, J. Beltran-Heredia, J.L. Acero, F.J. Rubio, Rate constants for the reactions of ozone with chlorophenols in aqueous solutions, J. Hazard. Mater. 79 (2000) 271–285. [39] V.K. Sharma, Potassium ferrate(VI): an environmentally friendly oxidant, Adv. Environ. Res. 6 (2002) 143–156. [40] V.K. Sharma, S. Hollyﬁeld, Ferrate(VI) oxidation of aniline and substituted anilines, Abstr. Pap. Am. Chem. Soc. 210 (1995) 186. [41] H. Huang, D. Sommerfeld, B.C. Dunn, C.R. Lloyd, E.M. Eyring, Ferrate(VI) oxidation of aniline, J. Chem. Soc. Dalton Trans. (2001) 1301–1305. [42] Y. Lee, S.G. Zimmermann, A.T. Kieu, U. Von Gunten, Ferrate (Fe(VI)) application for municipal wastewater treatment: a novel process for simultaneous micropollutant oxidation and phosphate removal, Environ. Sci. Technol. 43 (2009) 3831–3838. [43] M. Chen, L. Cui, C.H. Li, G.W. Diao, Adsorption, desorption and condensation of nitrobenzene solution from active carbon: a comparison of two cyclodextrins and two surfactants, J. Hazard. Mater. 162 (2009) 23–28. [44] S.X. Liu, R. Wang, Modiﬁed activated carbon with an enhanced nitrobenzene adsorption capacity, J. Porous Mater. 18 (2011) 99–106.

Journal of Hazardous Materials 195 (2011) 170–174

Contents lists available at ScienceDirect

Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat

Inhibitory effects of Cu (II) on fermentative methane production using bamboo wastewater as substrate Donglei Wu ∗ , Zhizhong Yang, Guangming Tian College of Environmental and Resources, Zhejiang University, Hangzhou 310029, PR China

a r t i c l e

i n f o

Article history: Received 7 April 2011 Received in revised form 1 August 2011 Accepted 7 August 2011 Available online 12 August 2011 Keywords: Wastewater Inhibition Methanogenesis Copper IC50

a b s t r a c t The toxic effects of Cu (II) present in bamboo industry wastewater (BIWW) upon its anaerobic biodegradability of organic content were investigated. The analysis through the Modiﬁed Gompertz model indicated that the optimum chemical oxygen demand (COD) concentration for digestion was 22,780 mg L−1 with a maximum Rm (maximum CH4 production rate) value of 2.8 mL h−1 , corresponding to a speciﬁc methanogenic activity (SMA) of 2.38 mLCH4 gVSS−1 h−1 . The inhibitory effects of Cu (II) on cumulative methane production depended on its concentration and contact time. Low concentrations (5 mg L−1 ) of Cu (II) showed a stimulating effect on methanogenesis. Methane was not detected when the Cu (II) concentration was increased beyond 300 mg L−1 . The IC50 value of Cu (II), the Cu (II) concentration that causes a 50% reduction in the cumulative methane production, was 18.32 mg L−1 (15.9 mg Cu (II) gVSS−1 ). © 2011 Elsevier B.V. All rights reserved.

1. Introduction Bamboo industry is a traditional industry in China with a history of about one thousand years, generating many important bamboo products that are used today [1]. However, production of these products involves boiling and dyeing processes which generates tonnes of bamboo industry wastewater (BIWW). The wastewater produced from boiling organics out of the bamboo material contains a bulk of organics mainly consisting of plant residues like carbohydrates and organic acids, and wastewater produced from dyeing the bamboo material usually contains organic dyes and high concentrations of Cu (II). The wastewater is characterized by a high chemical oxygen demand (20,000–50,000 mg L−1 ), low pH values (2.5–5), high concentrations of Cu (II) (800–2000 mg L−1 ) and strong color content. Direct discharge of BIWW to the environment could be hazardous to aquatic life. Therefore, the treatment of BIWW is a challenging task which should be accomplished with urgency. Anaerobic treatment is an effective and widely used technology for decomposition of high strength wastewater containing organics [2–5]. However, the anaerobic bio-treatment is affected by substrate concentration, toxic materials, microbial biomass and contact time [6]. Moreover, a suitable substrate concentration plays an important role in the stable operation of anaerobic reactors. Many studies have suggested that the presence of heavy metals causes instability or operation failure of anaerobic reactors [7–9].

∗ Corresponding author. Tel.: +86 571 3805739113; fax: +86 571 8838 8393. E-mail addresses: [email protected], [email protected] (D. Wu). 0304-3894/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.jhazmat.2011.08.021

The main reason for this instability is attributed to disruption of enzymatic structure and function due to metal binding with thiol and other groups present in protein molecules or by replacing naturally occurring metals in their prosthetic groups [10]. Copper, being one of most prevalent heavy metals in BIWW, imparts potentially negative impacts on anaerobic microorganisms. Previous studies suggested variable concentrations of Cu (II) had inhibitory effects on methanogenesis. Mori et al. [11] indicated that 63.5 mg L−1 of Cu (II) could completely inhibit the activities of methanogens. Meanwhile, Lin [12,13] reported that the half-inhibition concentrations were 12.50 mg L−1 and 130 mg L−1 , respectively. Such huge variations in the inhibitory concentrations were partly due to differences in the precipitation and adsorption of soluble metals during the assays [14] and the different wastewaters used in their studies. Since none of the data has been reported in relation to the treatment of BIWW, the present study was aimed to conduct a series of batch experiments based on the Modiﬁed Gompertz model to explore the biodegradability of BIWW. The inhibitory effect of its Cu (II) content on the methanogenesis and some important kinetic parameters were investigated. The investigations would provide guidelines for treating high strength BIWW. 2. Materials and methods 2.1. Wastewater and sludge The raw wastewater, which contains no Cu (II), was collected from one of biggest and typical BIWW producing factory (Anji County, Zhejiang Province, China). The COD of the wastewater

D. Wu et al. / Journal of Hazardous Materials 195 (2011) 170–174

2.2. Batch tests The batch tests were performed in two experiments. Since high organic concentrations are typically encountered with raw wastewater, the ﬁrst experiment was aimed to study the effects of the COD concentration on the biodegradability of the wastewater. The dyeing wastewater commonly has high concentrations of Cu (II), which can inhibit the anaerobic process. Therefore the second experiment was conducted to investigate the inhibitory effects of Cu (II) on methanogenic activity. During the ﬁrst experiment, raw wastewater was diluted with deionized water into 8 concentration gradients. Batch studies were performed to analyze the methane production during the treatment of each wastewater concentration. In the second experiment combined raw wastewater containing COD = 6300 mg L−1 as substrate with different Cu (II) concentrations as CuCl2 ·2H2 O (5, 20, 70, 130, 200, 250 and 300 mg L−1 ) had been treated. Batch studies were performed to analyze methane production of each combination during treatment. Control units were operated without any Cu (II) dosage. All batch tests were conducted in a 150 mL glass serum vials, and each vial was ﬁlled with 20 mL anaerobic sludge and 100 mL wastewater along with nutrients. Although the pH values of raw wastewater was in the range of 2.5–5, the recycle efﬂuent of anaerobic reactor can increase it to 6–8. Therefore, the initial pH values of the study were adjusted to 7.0 ± 0.2 using NaOH (2 M) or HCl (2 M) solutions [15]. All vials were then capped with butyl rubber and incubated in a thermostatic water bath at 30 ± 0.5 ◦ C. All vials were shaken 8–10 times a day, and CH4 production was measured until it leveled off. Quadruplicate parallels were set for each batch test. 2.3. Analytical procedures Methane production was monitored during the assays using a displacement system. Biogas was allowed to move into the Smith fermentation tube containing 15% (w/v) NaOH solution where CO2 and H2 S were absorbed, and the remaining gas was CH4 , as indicated by the readings of fermentation tube. COD, TN, TP, VSS and TSS were measured according to the Standard Methods [16]. 2.4. Kinetic modeling The Modiﬁed Gompertz model (Eq. (1)) has been widely used in the studies of fermentative hydrogen production [17–20] and to predict rates of fermentative gas production processes [21,22]: H(t) = P exp

−exp

R e m P

( − t) + 1

(1)

where H(t) is cumulative CH4 production (mL), is lag time (h), P is CH4 production potential (mL) and Rm is the maximum CH4 production rate (mL h−1 ) and e = 2.718282828; values of H(t) , , P and Rm for each batch were estimated using Origin 9.0 for nonlinear regression analysis.

700

Cumulative Methane Production (mL)

was 36,650 mg L−1 , while its COD:TN:TP was 190:4:1. The seed sludge was obtained from a full-scale Up-ﬂow Anaerobic Sludge Bed (UASB) treating pig slurry (Hangzhou, Zhejiang Province, China). Volatile suspended solid (VSS) of the sludge was 58.7 g L−1 , accounting for 71% of total suspended solids (TSS). The essential nutrients fed to the reactor for the growth of microorganisms were (mg L−1 ): MgCl2 ·7H2 O (400), KCl (400), CaCl2 (38), FeCl2 ·4H2 O (40), COCl2 ·6H2 O (10), KI (10), Al2 (SO4 )3 (0.3), MnCl2 ·4H2 O (0.5), CuCl2 ·2H2 O (0.5), ZnCl2 (0.5), NaMoO4 ·2H2 O (0.5), H3 BO3 (0.5), and NiCl2 ·6H2 O (0.5).

171

-1

2480 mg L -1 4450 mg L -1 8300 mg L -1 12550 mg L -1 17550 mg L -1 22780 mg L -1 27840 mg L -1 36650 mg L

600 500 400 300 200 100 0 0

100

200

300

400 Time (h)

500

600

700

Fig. 1. Cumulative CH4 production at different COD levels.

To describe the biodegradability of organic content of wastewater and inactivation of anaerobic culture by Cu (II), cumulative CH4 production curves with respect to time were initially obtained from the methane production experiments. Then, the Modiﬁed Gompertz Equation was applied to quantify the methane production. The SMA (mLCH4 gVSS−1 h−1 ) was calculated by dividing the Rm by the initial VSS in the serum vial. 3. Results and discussion 3.1. Biodegradation of organics Cumulative CH4 production of raw wastewater, which contains no Cu (II), at different supplied COD levels is shown in Fig. 1. It was obvious that the lag time and CH4 production generally exhibited a linear relationship with increasing COD, whereby CH4 production increased from 60 to 655 mL when COD was increased from 2480 to 36,650 mg L−1 . At low COD concentrations (<8300 mg L−1 ), the reactor operation was stable without any obvious lag time, but further increase in COD concentration delayed CH4 production, which might be due to the fact that unacclimated microbial communities might have required longer duration to adapt concentrated substrates. In order to evaluate the effects of COD on methanogens, the experimental data obtained from cumulative CH4 production at different COD concentrations were ﬁtted to Modiﬁed Gompertz Equation with R2 > 0.98. The kinetic parameters obtained from the experiment are summarized in Table 1. It was evident that the lag time and methanogenesis potential P increased with the increasing COD concentrations. The respective increases in lag time and methanogenesis potential P were in the range of 2.1–130.3 h and 59–639 mL when the COD was increased from 2480 to 36,650 mg L−1 . The maximum methanogenesis rate Rm initially increased from 0.8 to the maximum 2.8 mL h−1 , corresponding to a SMA range of 0.68 to 2.38 mLCH4 gVSS−1 h−1 , for the COD increase from 2480 to 22,780 mg L−1 . The methanogenesis subsequently decreased to 2.6 mL h−1 with a corresponding SMA of 1.87 mLCH4 gVSS−1 h−1 , when the COD was increased to 36,650 mg L−1 . Theoretically, 1 gCOD can produce 350 mL of methane [23], while unit methanogenesis for COD was in range of 178–241 mLCH4 gCOD−1 during the present investigation which was 50.9–68.9% of theoretical methane production. The efﬁciency of methanogenesis is highly dependent on the optimal control of substrate to biomass (S X−1 ) ratio. This ratio signiﬁcantly affects the metabolic and kinetic characteristics of

172

D. Wu et al. / Journal of Hazardous Materials 195 (2011) 170–174

Table 1 Kinetic parameters of cumulative CH4 production at various COD concentrations.

2480 4450 8300 12,550 17,500 22,780 27,840 36,650

(h)

2.1 9.5 24.4 62.3 106.7 114.4 116.7 130.3

P (mL)

Rm (mL h−1 )

0.8 1.4 2.0 2.1 2.3 2.8 2.4 2.2

59 104 199 259 373 453 540 639

Methane production per unit COD (mLCH4 gCOD−1 )

SMA R2 (mLCH4 gVSS−1 h−1 )

241 232 239 208 213 207 202 178

0.68 1.19 1.70 1.78 1.95 2.38 2.04 1.87

Control -1 5 mg L -1 20 mg L -1 70 mg L -1 130 mg L -1 200 mg L -1 250 mg L -1 300 mg L

160 140 Cumulative Methane Production (mL)

COD (mg L−1 )

180

0.99535 0.9975 0.99884 0.99761 0.99653 0.98595 0.9874 0.9912

120 100 80 60 40 20

: lag time; P: CH4 production potential; Rm : maximum CH4 production rate. SMA: calculated by dividing the Rm by the initial VSS (1.175 g).

0 0

50

100

150

200

250

Time(h)

microbial communities involved [24]. Lobos et al. [25] demonstrated that the bacterial growth is optimum at higher S X−1 ; and in the case of low S X−1 , the MLVSS concentration begins to decrease. In the case of slow anaerobic bacterial growth, the microbial biomass concentration was assumed to remain virtually constant throughout the experiment (with values 1.175 mg VSS L−1 ). Fig. 2 shows the cumulative methanogenesis per unit COD at different S X−1 values. It was evident that the cumulative methanogenesis per unit COD slightly decreased with the increasing S X−1 . Linear ﬁtting showed a high correlation (R2 = 0.8671) between methanogenesis per unit COD and S X−1 .

240

genesis was completely inhibited, at the Cu (II) concentration of 300 mg L−1 , with no CH4 detected within 250 h. To evaluate the effects of Cu (II) on methanogenesis, the experimental data obtained from cumulative CH4 production at different Cu (II) concentrations were ﬁtted to the Modiﬁed Gompertz Equation and its kinetic parameters are summarized in Table 2. All estimated values of the kinetic parameters for Cu (II) concentrations of 5 mg L−1 were larger than that of control, except the lag time which was 51.8 h compared to 54.3 h of control. Cu (II) concentrations of 5 mg L−1 showed a stimulatory effect on methanogenic activity. The results presented in Table 2 showed a number of irregularities which may be due to the complexity of the real wastewater used in the batch test as a substrate. The lag time for 70 and 130 mg L−1 of Cu (II) were 77.8 and 74.5 h, respectively, which were obviously higher than that of the control keeping other parameters constant. It was indicated that the inhibitory effect of certain concentrations of Cu (II) (70–130 mg L−1 ) could be eliminated by a longer adaption time. It was previously reported that metals are toxic to microbial activity when they exist in free soluble state [26–28]. Moreover, microorganisms secrete extracellular polymeric substances (EPS) after enough acclimation time, which can adsorb or precipitate the soluble or free Cu (II) to eliminate its inhibitory effects [29,30]. Compared to the control, 200 mg L−1 of Cu (II) had a higher lag time (110 h) and lower methanogenesis potential P (139 mL). The maximum methanogenesis rate Rm and SMA values were close to each other. The lag time for Cu concentrations of 250 mg L−1 (99 h) was shorter than that for 200 mg L−1 , the reason behind that phenomenon was not clear. However, the , Rm , P and SMA values were far less than those of the control for 200 mg L−1 which further suggested that the methanogenic reaction was strongly blocked.

220

3.3. The recovery effects and IC50 of Cu (II)

3.2. Effects of Cu (II) on cumulative methanogenesis

The inhibitory effects of Cu (II) on methanogens could be judged by relative activity (RA), which could be calculated by the following formula:

-1

(mLCH4 gCOD )

Fig. 3 illustrates the cumulative methanogenesis at various Cu (II) concentrations. The data suggested that Cu (II) concentrations of 5 mg L−1 have a slightly stimulating effect on the activity of methanogenic bacteria. Such behavior could be attributed to fact that the presence of Cu (II) is required for the activation or functioning of many microbial enzymes and coenzymes [15]. Similar phenomenon was reported for other heavy metals; for example, Altas [15] indicated that Zn and Ni doses of 8–32 mg L−1 and 0.5–16 mg L−1 , respectively, could increase the cumulative CH4 production. Li and Fang [9] also found that the fermentative H2 production could be increased by Zn and Pb respective concentrations of 80–400 mg L−1 and mg L−1 . As Cu (II) concentrations increased above 70 mg L−1 , the microbial activity was temporarily inhibited with obvious lag time. A signiﬁcant inhibition of cumulative methanogenesis was evident at 250 mg L−1 with the gas volume of 35 mL which was far less than 152 mL for the control. Methano-

Cumulative CH 4 production per unit COD

Fig. 3. Cumulative CH4 production at different Cu (II) concentrations.

200

180

y = - 19.6869 + 242.98 2 R = 0.8671

RA=

160

0.0

0.5

1.0

1.5 -1

2.0

2.5

3.0

3.5

-1

S X (gCOD gVSS )

Fig. 2. Cumulative CH4 production per unit COD at various S X−1 value.

cumulative CH4 production of testing at particular time ×100% cumulative CH4 production of control at the same time

From Fig. 3 we can see that the RA of 5 and 20 mg L−1 increased within 112 h, and then deceased until 194 h; the maximum RA for 5 and 20 mg L−1 Cu (II) concentration were 110% and 125% (at 112 h), respectively. All RA values for 5 mg L−1 were above 100%, which implied that the Cu (II) concentration played a stimulatory

D. Wu et al. / Journal of Hazardous Materials 195 (2011) 170–174

173

Table 2 Kinetic parameters of cumulative CH4 production at various concentrations of Cu (II). Cu (II) (mg L−1 )

0 5 20 70 130 200 250 300

(h)

54.3 51.8 54.5 77.8 74.5 110 99 –

Rm (mL h−1 )

1.3 1.6 1.4 1.3 1.2 1.2 0.3 0

P (mL)

166 169 147 165 172 139 40 0

Methane production per unit COD (mLCH4 gCOD−1 )

SMA (mLCH4 gVSS−1 h−1 )

R2

242 252 221 238 239 189 55 0

1.1 1.4 1.2 1.1 1.0 1.0 0.25 0

0.98678 0.98664 0.99044 0.99238 0.99841 0.997 0.99597 –

: lag time; P: CH4 production potential; Rm : maximum CH4 production rate. SMA: calculated by dividing the Rm by the initial VSS (1.175 g).

role on methanogens throughout the batch tests. The initial and ﬁnal RA values for Cu (II) concentration of 20 mg L−1 were lower than 100%, indicating initial inhibitory effect then a stimulatory one. Such behavior may be explained in terms of adaptation that occurred after 112 h of adaption, microorganisms secreted a lot of EPS, absorbing and precipitating soluble or free Cu (II) in the wastewater that caused free Cu (II) concentrations which might have favored the growth of microbes. However, the reason for the inhibitory effect still needs further investigation. The RA values of 70, 130, 200 and 250 mg L−1 increased with time and were almost unchanged after 240 h. For 70 and 130 mg L−1 of Cu (II), the system could be restored to 80% of relative methanogenic activity after 194 h while the inhibitory effects could be completely eliminated after 240 h. But for 200 and 250 mg L−1 , the RA values for anaerobic microorganisms were far lower than 100%. Therefore, when the Cu (II) concentration was higher than 200 mg L−1 , a serious inhibitory effect would occur and the activity of microorganisms was hard to be recovered. The possible reason leading to this observation may be attributed to the heavy metal (Cu (II)) binding to the sulfydryl or other groups of the protein molecules causing alterations in delicate structure and subsequent function of microbial enzymes [10]. Jin et al. [31] reported that recovery was possible with a feed copper to volatile suspended solids ratio (Cu (II):VSS−1 ) in the range of 0.011–0.022 up to 15 mg L−1 of added Cu (II). However, if the Cu (II) concentration increased to more than 20 mg L−1 , the methanogenic activity could not be recovered even the Cu (II):VSS−1 was 0.015. The inhibitory effects of Cu (II) may be expressed by its halfinhibitory concentration (IC50 ) – the concentration of Cu (II) that causes a 50% reduction in the cumulative methanogenesis relative to the control sample over a ﬁxed period of exposure time (250 h), expressed as mg Cu (II) L−1 . Therefore, the IC50 value of Cu (II) was calculated to be 18.32 mg L−1 (R2 = 0.9061) by plotting the RA values at different Cu (II) concentrations. It has been suggested that inhibition caused by heavy metals would be more comparable if metal dosage was expressed as milligram of metal per gram of volatile solids [27]. The IC50 value of Cu (II) for the present study was 15.9 mg Cu (II) gVSS−1 . It is reported that the corresponding IC50 data of Cu (II) for methanogenic activities of granular sludge degrading cattail [32], benzoate [33] and volatile fatty acids (VFA) [34] were 6.4, 175 and 130 mg L−1 , respectively. Moreover, the SMA of granules for the degradation of starch was reduced by 50% when 180 mg Cu (II) gVSS−1 was added [35], while for ﬂocculent digester sludge for the degradation of mixed VFA the value was decreased to 23.3 mg Cu (II) gVSS−1 [36]. Results indicate that the inhibition effect was dependent on substrate and granulation degree of seed sludge. Overall, with the exception of degrading cattail, the IC50 values in this study were substantially lower than those reported in the literature, indicating that Cu (II) was quite inhibitive to the

CH4 -producing sludge when using raw wastewater’s organics as substrate. This may be attributed to the following two factors: (1) the capabilities for the precipitation and adsorption of soluble Cu (II) during individual assays [14]; (2) the concentration of EPS of the sludge, which are the microstructures of anaerobic granular sludge [37] and protect the CH4 -producing cells against the stressful environment conditions. 4. Conclusions The anaerobic treatment of the BIWW was quite feasible; the adaptation time required for the anaerobic microorganisms increased with the increasing COD in wastewater. The optimum COD concentration for methane production was 22,780 mg L−1 . A continuous experiment is needed to further prove the possibility of anaerobic treatment of the BIWW. Copper concentrations of 5 mg L−1 showed a stimulatory effect which veriﬁed the Modiﬁed Gompertz model. In general, the cumulative methane production of microorganisms decreased with increasing Cu (II) concentrations from 70 to 250 mg L−1 . Methaneproducing sludge had the ability to recover after a certain period. However, methanogenic activity was completely blocked at the Cu (II) concentration of 300 mg L−1 , and the IC50 value for Cu (II) was 18.32 mg L−1 (15.9 mg Cu (II) gVSS−1 ). Acknowledgments This study has been supported ﬁnancially by the Major Projects on Control and Management Technology of Water Pollution of China (Grant No. 2009ZX07317-008) and National Natural Science Foundation of China (Grant No. 20206019). References [1] D. Wenyuan, Historical opportunities for the sustainable development of China’s bamboo industry in the 21st century, World Forest. Res. 16 (2003) 42–45. [2] R.C. Speece, L. Yaxin, Anaerobic Biotechnology of Industrial Wastewater, China Building Industry Press, Beijing, 2001, pp. 123–311. [3] D. Rani, K. Nand, Ensilage of pineapple processing waste for methane generation, Waste Manage. 24 (2004) 523–528. [4] R. Leitao, A. Van Haandel, G. Zeeman, G. Lettinga, The effects of operational and environmental variations on anaerobic wastewater treatment systems: a review, Bioresour. Technol. 97 (2006) 1105–1118. [5] K. Yetilmezsoy, Z. Sapci-Zengin, Stochastic modeling applications for the prediction of COD removal efﬁciency of UASB reactors treating diluted real cotton textile wastewater, Stoch. Environ. Res. Risk Assess. 23 (2009) 13–26. [6] J. Yang, R. Speece, The effects of chloroform toxicity on methane fermentation, Water Res. 20 (1986) 1273–1279. [7] H. Yu, H. Fang, Inhibition on acidogenesis of dairy wastewater by zinc and copper, Environ. Technol. 22 (2001) 1459–1465. [8] S. Juliastuti, J. Baeyens, C. Creemers, D. Bixio, E. Lodewyckx, The inhibitory effects of heavy metals and organic compounds on the net maximum speciﬁc growth rate of the autotrophic biomass in activated sludge, J. Hazard. Mater. 100 (2003) 271–283.

174

D. Wu et al. / Journal of Hazardous Materials 195 (2011) 170–174

[9] C. Li, H. Fang, Inhibition of heavy metals on fermentative hydrogen production by granular sludge, Chemosphere 67 (2007) 668–673. [10] D.H. Nies, Microbial heavy-metal resistance, Appl. Microbiol. Biotechnol. 51 (1999) 730–750. [11] K. Mori, M. Hatsu, R. Kimura, K. Takamizawa, Effect of heavy metals on the growth of a methanogen in pure culture and coculture with a sulfate-reducing bacterium, J. Biosci. Bioeng. 90 (2000) 260–265. [12] C. Lin, Effect of heavy metals on volatile fatty acid degradation in anaerobic digestion, Water Res. 26 (1992) 177–183. [13] C. Lin, C. Chen, Effect of heavy metals on the methanogenic UASB granule, Water Res. 33 (1999) 409–416. [14] S. Karri, R. Sierra-Alvarez, J. Field, Toxicity of copper to acetoclastic and hydrogenotrophic activities of methanogens and sulfate reducers in anaerobic sludge, Chemosphere 62 (2006) 121–127. [15] L. Altas, Inhibitory effect of heavy metals on methane-producing anaerobic granular sludge, J. Hazard. Mater. 162 (2009) 1551–1556. [16] APHA, Standard Methods for the Examination of Water and Wastewater, 21st ed., American Public Health Association, Washington, DC, 2005. [17] D. Cheong, C. Hansen, Bacterial stress enrichment enhances anaerobic hydrogen production in cattle manure sludge, Appl. Microbiol. Biotechnol. 72 (4) (2006) 635–643. [18] C. Wang, P. Lin, J. Chang, Fermentative conversion of sucrose and pineapple waste into hydrogen gas in phosphate-buffered culture seeded with municipal sewage sludge, Process Biochem. 41 (2006) 1353–1358. [19] Y. Mu, G. Wang, H. Yu, Kinetic modeling of batch hydrogen production process by mixed anaerobic cultures, Bioresour. Technol. 97 (2006) 1302–1307. [20] J. Wang, W. Wan, Kinetic models for fermentative hydrogen production: a review, Int. J. Hydrogen Energy 34 (2009) 3313–3323. [21] B. Zhu, P. Gikas, R. Zhang, J. Lord, B. Jenkins, X. Li, Characteristics and biogas production potential of municipal solid wastes pretreated with a rotary drum reactor, Bioresour. Technol. 100 (2009) 1122–1129. [22] H. Benbelkacem, R. Bayard, A. Abdelhay, Y. Zhang, R. Gourdon, Effect of leachate injection modes on municipal solid waste degradation in anaerobic bioreactor, Bioresour. Technol. 101 (2010) 5206–5212. [23] K. Stamatelatou, A. Kopsahelis, P.S. Blika, C.A. Paraskeva, G. Lyberatos, Anaerobic digestion of olive mill wastewater in a periodic anaerobic bafﬂed reactor (PABR) followed by further efﬂuent puriﬁcation via membrane separation technologies, J. Chem. Technol. Biotechnol. 84 (2009) 909–917.

[24] Y. Liu, Bioenergetic interpretation on the S0 /X0 ratio in substrate-sufﬁcient batch culture, Water Res. 30 (1996) 2766–2770. [25] J. Lobos, C. Wisniewski, M. Heran, A. Grasmick, Effects of starvation conditions on biomass behaviour for minimization of sludge production in membrane bioreactors, Water Sci. Technol. 51 (2005) 35. [26] K.T. Kim, I.S. Kim, S.H. Hwang, S.D. Kim, Estimating the combined effects of copper and phenol to nitrifying bacteria in wastewater treatment plant, Water Res. 40 (2006) 561–568. [27] R. Hickey, J. Vanderwielen, M. Switzenbaum, The effect of heavy metals on methane production and hydrogen and carbon monoxide levels during batch anaerobic sludge digestion, Water Res. 23 (1989) 207–218. [28] H. Zhiqian, K. Chandran, D. Grasso, B.F. Smets, Impact of metal sorption and internalization on nitriﬁcation inhibition, Environ. Sci. Technol. 37 (2003) 728–734. [29] Y. Liu, H.H.P. Fang, Inﬂuence of extracellular polymeric substances (EPS) on ﬂocculation, settling, and dewatering of activated sludge, Crit. Rev. Environ. Sci. Technol. 33 (2003) 237–273. [30] P. Principi, F. Villa, M. Bernasconi, E. Zanardini, Metal toxicity in municipal wastewater activated sludge investigated by multivariate analysis and in situ hybridization, Water Res. 40 (2006) 99–106. [31] P. Jin, S. Bhattacharya, C. Williams, H. Zhang, Effects of sulﬁde addition on copper inhibition in methanogenic systems, Water Res. 32 (1998) 977–988. [32] Y. Zheng-Bo, Y. Han-Qing, W. Zhi-Liang, Anaerobic digestion of cattail with rumen culture in the presence of heavy metals, Bioresour. Technol. 98 (2007) 781–786. [33] H.H.P. Fang, O.C. Chan, Toxicity of electroplating metals on benzoate-degrading granules, Environ. Technol. 18 (1997) 93–99. [34] C.Y. Lin, C.C. Chen, Effect of heavy metals on the methanogenic UASB granule, Water Res. 33 (1999) 409–416. [35] H.H.P. Fang, H.H. Hui, Effect of heavy metals on the methanogenic activity of starch-degrading granules, Biol. Lett. 16 (1994) 1091–1096. [36] C. Lin, Effect of heavy metals on acidogenesis in anaerobic digestion, Water Res. 27 (1993) 147–152. [37] L. Yong-Qiang, L. Yu, T. Joo-Hwa, The effects of extracellular polymeric substances on the formation and stability of biogranules, Appl. Microbiol. Biotechnol. 65 (2004) 143–148.

Journal of Hazardous Materials 195 (2011) 175–181

Contents lists available at ScienceDirect

Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat

Recycle of electrolytically dissolved struvite as an alternative to enhance phosphate and nitrogen recovery from swine wastewater YingHao Liu a , Sanjay Kumar a , JungHoon Kwag b , JaeHwan Kim b , JeongDae Kim a , ChangSix Ra a,∗ a b

Department of Animal Life System, Kangwon National University, Hyoja 2, 192-1, Chunchon 200-701, South Korea National Institute of Animal Science, RDA, Suwon 441-350, South Korea

a r t i c l e

i n f o

Article history: Received 31 May 2011 Received in revised form 26 July 2011 Accepted 7 August 2011 Available online 12 August 2011 Keywords: Electrolysis Struvite Recycling Swine wastewater

a b s t r a c t Operational parameters such as electric voltage, NaCl, reaction time (RT) and initial struvite amount were optimized for struvite dissolution with a designed electrolysis reactor, and the effect of recycling the dissolved solution on the performance of struvite crystallization was also assessed. The electrolytic reactor was made of plexiglas having titanium plate coated with iridium oxide as anode (surface area: 400 cm2 ) and stainless steel plates as cathodes. For reutilization of dissolved struvite, four runs were conducted with different recycle ratio of the solution. Optimum conditions for the electric voltage, NaCl, RT and initial struvite amount were 7 V, 0.06%, 1.5 h and 1.25 g/L, respectively. At the above optimized conditions, 49.17 mg/L phosphate (PO4 3− –P) was dissolved and ammonium–nitrogen (NH4 –N) got completely removed from the solution. When 0.0, 0.5, 1.0 and 2.0 moles of the dissolved struvite with respect to PO4 3− –P in swine wastewater were recycled along with 0.5 M magnesium chloride (MgCl2 ), the PO4 3− –P removal was 63, 69, 71 and 79%, and NH4 –N was 9, 31, 40 and 53%, respectively. Hence, the performance of struvite formation process was proportionally increased. It is concluded that struvite can be re-dissolved by electrolysis and reused as a source of P and Mg. © 2011 Elsevier B.V. All rights reserved.

1. Introduction Phosphorus (P) is a valuable and limited resource. It is estimated that there are seven billion tons of phosphate rock as P2 O5 remaining in reserves that could be economically mined. The human population consumes as much as 40 million tons of P as P2 O5 each year and P demand is increasing by 1.5% each year. Hence, global reserves of high-quality mined phosphate deposits are being gradually depleted. It is predicted that the resource could be exhausted in 100–250 years [1]. Consequently researchers are concerned for alternative and renewable source of P. Swine wastewater is considered to be a good source for P recovery because of the presence of high concentrations of P. Struvite crystallization is an ideal and efﬁcient technique to remove and recover P from wastewater due to its high removal effectiveness and reaction rate. According to our previous study [2], if proper operation parameters were provided, about 90% P could be easily recovered as struvite from the swine wastewater. Struvite consists of equal molar concentrations of magnesium, ammonium and phosphate. The chemical reaction is expressed as follows [3]: Mg2+ + NH4 + + PO4 3− + 6H2 O ↔ MgNH4 PO4 ·6H2 O

∗ Corresponding author. Tel.: +82 33 250 8618; fax: +82 33 251 7719. E-mail address: [email protected] (C.s. Ra). 0304-3894/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.jhazmat.2011.08.022

Since swine wastewater contains high amounts of NH4 –N with respect to orthophosphate (PO4 3− –P) (more than 20 times), the NH4 –N removal during crystallization of struvite is very low [2]. To enhance the NH4 –N removal, high amounts of external P and Mg source should be added. However, the high cost of Mg salts and P consumption for the effective removal of ammonia would become the main obstacle to the wide application of the struvite precipitation process [4]. Therefore, it is preferable to reduce the usage of Mg and P. If the struvite could be re-dissolved and ammonia could be completely removed from the struvite, the remaining Mg and P would not only be a good source for ammonia removal, but also decrease the external addition of Mg and P sources. In recent years, many researchers found that the residues of struvite decomposed by heating under alkali conditions could be used as P and Mg sources, reducing operation cost [5–8]. Electrolysis, a modern treatment technology, has also attracted a great deal of attention in the application of the wastewater treatment. It is reported that the electrolysis is a simple, reliable, cost-effective and promising technique [9]. The key process in electrolysis is the interchange of atoms and ions by the removal or addition of electrons from an external circuit. The electrolysis efﬁciency depends on several factors such as electric voltage, ionic concentration, electrode material and reaction time, and optimizing such factors play an important role towards the success of the process. This technology has been successfully used for the removal of ammonia from various types of wastewaters such as landﬁll

176

Y. Liu et al. / Journal of Hazardous Materials 195 (2011) 175–181

leachate [10], tannery wastewater [11,12], municipal discharges [13], power plant efﬂuents [14], and sludge digester efﬂuents [15]. So, the electrolysis method would be a ﬁne approach to completely remove ammonia from the struvite and recycle the residue as an alternate source of Mg and P. The removal mechanism of ammonia in the electrolytic process is poorly understood in terms of the oxidation route. A direct oxidation of ammonia to nitrogen gas has been observed by Panizza et al. [16]. The removal of ammonia also took place through an indirect oxidation route by both hydroxyl radicals and hypochlorous acid (HOCl) [13]. Till date, according to our knowledge, there are no reports in the literature speciﬁcally on the applicability of electrolysis technology for struvite dissolution. Therefore, in the present study, operational parameters for electrolysis process to achieve the dissolution of Mg2+ and PO4 3− –P from struvite as well as NH4 + removal were determined. In addition, the effectiveness of the recycling the electrolytically dissolved struvite on the performance of the crystallization process was also examined.

2. Materials and methods 2.1. System conﬁguration A schematic illustration of the experimental apparatus is shown in Fig. 1. The system consisted of an electrolytic reactor for struvite dissolution, struvite crystallization and recovery reactor for swine wastewater treatment and a puriﬁcation system to recover pure struvite. The electrolytic reactor was made of Plexiglas and spigots were installed at the top and bottom of the reactor in order to facilitate the inﬂuent loading and efﬂuent decanting, respectively. The working volume, height and diameter were 4.0 L, 120 cm and 8 cm, respectively. Ti plate coated with IrO2 (thickness: 1 mm) was used as anode and placed at the center of the reactor. The anode surface area in the electrolytic reactor was 400 cm2 (100 cm2 /L, 100 cm × 4 cm). Two stainless steel plates (thickness: 1.2 mm) were used as cathodes, which were arranged parallel to each other on either side of the anode plate by an acrylic band. The distance between each electrode was ﬁxed as 1 cm in order to obtain an efﬁcient electric ﬁeld. The apex of the reactor was sealed along with the electrodes by acrylic and silicone gel. The electricity was regulated by a digital DC power supply (DC 12 V 30 A, Model: WER 312) by connecting it with the electrodes. The struvite crystallization and recovery reactor had a reaction zone (A) and a settling zone (B). The working volume of the reactor and the inner reaction zone was 12.3 and 2.72 L, respectively. A spigot was installed at the bottom of the settling zone to retrieve the formed struvite crystals from the reactor.

In the ﬁrst set of experiment, the effect of NaCl and voltage on the electrolytic struvite dissolution was examined by gradually increasing the NaCl concentration (0.01, 0.02, 0.04, 0.06, and 0.08% or 0.1, 0.2, 0.4, 0.6 and 0.8 g/L) at three different voltages (5, 7 and 9 V, respectively). The electrolysis process was operated in a batch mode and the RT was 6 h and initial struvite amount of 1.25 g/L. Pure struvite and NaCl were mixed well with tap water (pH 8.0) before loading into the electrolytic reactor. The ideal NaCl and voltage level selected based on the obtained results was used for further investigation. In the second set, the effect of RT and struvite amount on the electrolytic process for struvite dissolution was studied. For which different struvite amount (0.25, 0.50, 1.00, 1.25 and 2.50 g/L) was used and RT was taken as 2.5 h. The electrolytically dissolved struvite is collected in the container and stored till further use. To test the effect of recycling of the electrolytically dissolved struvite (as the source of P and Mg) for swine wastewater treatment, four sets of experiment (one control and three runs) were carried out. For control only 0.5 M MgCl2 with respect to the PO4 3− –P in the swine wastewater and distilled water (pH 8.0) in the same volume as the recycled supernatant was added into the struvite crystallization reactor at the same time of inﬂuent feeding. Different recycle ratio of the dissolved struvite (i.e. 0.5, 1 and 2 M respectively) along with 0.5 M of MgCl2 was loaded into the process in runs I, II and III. The detailed layout of the experimental design is presented in Table 1. The struvite crystallization and recovery reactor was operated in a continuous ﬂow mode and 4 h RT. The reaction zone was continuously aerated through an air stone to achieve CO2 air stripping, and the aeration rate of 0.73 L/L min was controlled by an air ﬂow meter. MgCl2 as a source of Mg2+ was loaded into the reaction zone along with the inﬂuent and the recycled supernatant. 2.3. Sampling and analytical method For struvite dissolution experiment, samples were taken at every 30 min from the electrolytic reactor and ﬁltered immediately through membrane ﬁlter (0.45 ␮m). The ﬁltrate was mixed with one drop of 1 N HCl before analysis. Samples were also collected from the inﬂuent and efﬂuent storage buckets of the crystallization and recovery reactor daily. The samples were stored at 4 ◦ C until analysis for a week. All the samples were centrifuged at 3000 rpm for 5 min to separate the solids. The parameters studied immediately were total solids (TS) and suspended solids (SS), while supernatant of all the samples were analyzed for PO4 3− –P and NH4 –N with the auto water analyzer (Quick Chem 8500, LACHAT) later on. All analyses were performed according to standard methods [17]. 2.4. Swine wastewater

2.2. Experimental procedure Based on information from our previous study [2] on struvite puriﬁcation, the sediment slurry was collected from the bottom of the reactor by centrifuging at 3000 rpm for 10 min. The obtained struvite-solid was dissolved into an acidic solution (pH < 4.0) after which the supernatant was collected. Then pH of the supernatant was increased over 10.0 by addition of 1 M NaOH solution to derive pure struvite crystal formation. Subsequently, the white crystals were recovered, washed for three times with distilled water (pH 8.5) and allowed to dry at room temperature. Pure struvite was collected and prepared for dissolution experiment. Four factors (NaCl concentration, electric voltage, RT and struvite amount) affecting the pattern of struvite dissolution (PO4 3− –P dissolution and NH4 –N removal) in the designed electrolytic reactor were studied to optimize the operational parameters.

The swine wastewater used during this study was collected from a local farm. The swine wastewater was stored in a 25 L container at 4 ◦ C until it was required. Prior to utilization, the swine wastewater was screened by a sieve with 0.5-mm mesh openings to remove large solids before being fed into the inﬂuent container. The characteristics of the swine wastewater used in this experiment are shown in Table 2. 3. Results and discussion 3.1. Effect of NaCl addition In electrolysis process, the ionic conductivity of the solution is a signiﬁcant parameter. Since the current passing through the circuit is a function of the conductivity under a certain applied

Y. Liu et al. / Journal of Hazardous Materials 195 (2011) 175–181

177

Table 1 Operational conditions for struvite crystallization reactor. Parameters

Control

Run I

Run II

Run III

Struvite recycle ratio to PO4 3− –P (M) MgCl2 to PO4 3− –P (M)

0.0 + (DW) 0.5

0.5 + (DW) 0.5

1.0 + (DW) 0.5

2.0 0.5

DW, distilled water, added to maintain the volume of recycled struvite same in all the operations; PO4 3− –P, ortho-phosphate. Table 2 Characteristics of the swine wastewater. Parameter

Mean

Maximum

Minimum

Std. Deviation

pH PO4 3− –P (mg/L) NH4 –N (mg/L) TS (g/L) SS (g/L)

8.31 31.62 2623.50 10.73 2.21

8.41 38.09 2846.12 12.13 3.30

8.19 26.95 2403.52 9.62 1.66

0.05 2.61 68.01 0.63 0.35

PO4 3− –P, orthophosphate; NH4 –N, ammonium nitrogen; TS, total solids; SS, suspended solids.

electric voltage hence, ionic conductivity of the solution affects the current efﬁciency, applied electric voltage and consumption of electric energy in the electric systems [18]. In the present study, NaCl was used as the supporting electrolyte because it increases the electric conductivity of the solution and thus reduces the energy consumption. Irdemez et al. [19] also reported that NaCl was the best electrolyte among four different types of chemicals, namely, NaCl, NaNO3 , Na2 SO4 , and CaCl2 . Furthermore, Abdelwahab et al. [20] also reported that the presence of chloride ions could remove the passive oxide layer formed on the electrode surface. On observing the pattern of PO4 3− –P dissolution at different NaCl concentration from Fig. 2a–c, it is evident that when the concentration of NaCl was increased from 0.01% to 0.08%, the PO4 3− –P concentration of the solution increased noticeably. At 0.01% and

0.02% NaCl, the PO4 3− –P dissolution is comparably lower. However, the PO4 3− –P amount (mg/L) in the solution increases considerably at 0.04, 0.06 and 0.08% NaCl concentration, but no much difference in PO4 3− –P dissolution was observed among those concentrations. From the above results it can be concluded that the addition of NaCl lead to an increase of PO4 3− –P concentration in the solution, which indicated that increased concentration of NaCl helped in better dissolution of the struvite. Bhuiyan et al. [21] stated that solubility of struvite changes with ionic strengths of solution. Wang et al. [22] also reported that the addition of NaCl to wastewater served as a useful technique to enhance the performance of electrolysis. Fig. 2d–f showed that the NH4 –N concentration decreases rapidly with increased level of NaCl in the solution. At 0.01% NaCl, no considerable change in NH4 –N concentration (mg/L) was observed in the solution, whereas at 0.02%, NH4 –N level started decreasing slowly. At 0.04% NaCl, NH4 –N level decreases considerably and sharp depletion in NH4 –N concentration was observed at 0.06 and 0.08%. Li and Liu [23] demonstrated that chloride present in electrolytic solution just played a catalytic role in NH4 –N degradation rather than reacting directly with NH4 –N and most of the NH4 –N in the presence of chloride gets converted into N2 gas. In addition, Vanlangendonck et al. [14] reported that NH4 –N can be efﬁciently removed with an appropriate chloride concentration. Vlyssides and Isralildes [11] and Vlyssides et al. [13] also reported increased NH4 –N depletion with increasing concentration of

V

A

+D.C. power supply

Anode (Ti / IrO2 )

pump

Cathode ( stainless steel)

pump

pump MgCl2

effluent

Struvite & water

A influent Purification system

B

struvite airstone

Sediment withdrawl Fig. 1. Schematic illustration of the experimental apparatus.

Harvesting pure struvite

178

Y. Liu et al. / Journal of Hazardous Materials 195 (2011) 175–181

a

0.02%

0.06%

0.08%

0.04%

b 55

0.01%

0.02%

0.06%

0.08%

0.04%

c 55

50

50

45

45

45

40 35 30 25

20

PO43--P level (mg/L)

50

40 35 30 25 20

15

15

10

10 Time (h)

Time (h)

0.01% 0.06%

16

0.02% 0.08%

0.04%

16

30 25 20

0.01% 0.06%

0.02% 0.08%

0.04%

0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6 Time (h)

f 16

12

4

10 8 6 4

2

2

0

0

0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6 Time (h)

NH4-N level (mg/L)

14

12

NH4-N level (mg/L)

14

6

0.04%

35

12

8

0.08%

40

14

10

0.02%

0.06%

10 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6

e

0.01%

15

0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6

d

NH4-N level (mg/L)

0.01%

PO43--P level (mg/L)

PO43--P level (mg/L)

55

0.01%

0.02%

0.06%

0.08%

0.04%

10 8 6 4 2 0

0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6 Time (h)

0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6 Time (h)

Fig. 2. Effect of NaCl and voltage on struvite dissolution (a and d) 5 V, (b and e) 7 V, (c and f) 9 V.

initial chloride. The removal of NH4 –N in the present study might be due to the conversion of the dissolved NH4 –N into N2 gas during electrolysis. It is well known that when electrochemical degradation proceeds by direct redox reactions on the electrode, the electrolysis of NaCl solution results in very strong oxidants of HOCl/OCl− that could ensure indirect oxidation in the solution [24,25]. With the addition of NaCl to the solution, chlorine forms at the anode and it is then converted to hypochlorous acid and hypochlorite [26,27]. The overall reaction occurring in an aqueous solution during electrolysis can be expressed as follows: MgNH4 PO4 ·6H2 O ↔ Mg2+ + NH4 + + PO4 3− + 6H2 O 2Cl− → Cl2 + 2e− Cl2 + H2 O → HOCl + H+ + Cl− HOCl → H+ + OCl−

9 V with 0.06% NaCl concentration, 6 h RT and struvite amount of 1.25 g/L. It can be seen from Fig. 3a that at 5 V, the PO4 3− –P dissolution was comparably lower than 7 and 9 V, however, slight decrease in PO4 3− –P concentration was observed when the applied electric voltage was increased from 7 V to 9 V. Hence, it can be concluded that the electric voltage affects the PO4 3− –P dissolution. Moreover, the results (Fig. 3b) show that the electric voltage also strongly inﬂuence the NH4 –N removal from the solution. NH4 –N depletion was positively correlated with the applied electric voltage. The required treatment times for complete removal of NH4 –N from the solution were 2.0, 1.5 and 1.0 h at 5, 7 and 9 V, respectively. Liu et al. [28] stated that higher electric voltage increased the rate of Cl− losing electron at anodes, which could improve higher NH4 –N oxidation rate. Mollah et al. [29] reported that the current density is the most important parameter for controlling the reaction rate within an electrochemical reactor. Since the current density went up with the electric voltage, resulting in the increased energy consumption, the optimum value of the applied electric voltage was selected as 7 V considering the efﬁciency and energy consumption.

NO3 − + 6H2 O + 8e− → NH3 + 9OH−

3.3. Effect of reaction time and struvite amount

2NH4 + + 3HOCl → N2 + 3H2 O + 5H+ + 3Cl−

Because RT is another vital factor affecting the treatment efﬁciencies of the electrolysis process [30], the effect of RT on the struvite dissolution was investigated with 7 V, 0.06% NaCl concentration and different struvite amount (0.25, 0.50, 1.00, 1.25 and 2.50 g/L). As shown in Fig. 4a, the PO4 3− –P concentration increased with the increase of RT. It could be easily observed that with initial struvite amount of 0.25 g/L, PO4 3− –P reached maximum level at 0.5 h and no effect of RT was noticed later on. However, with increased struvite amount i.e. 0.50–1.00 g/L, the PO4 3− –P level reached maximum at 1.0 h and not much difference in PO4 3− –P concentration was noticed by extending the RT up to 2.5 h. At 1.25 and 2.50 g/L of initial struvite amount, PO4 3− –P dissolution

Hence, from the present study it can be concluded that a concentration of 0.06% NaCl would be optimum to achieve the PO4 3− –P dissolution and complete removal of NH4 + from the struvite within short time of electrolysis. 3.2. Effect of applied electric voltage Since another important parameter that can strongly affect the performance of the electrolysis is the applied electric voltage, the effect of the voltage on struvite dissolution was studied at 5, 7 and

Y. Liu et al. / Journal of Hazardous Materials 195 (2011) 175–181

a

5v

55

7v

9v

b

5v

20

179

7v

9v

45 NH4-N level (mg/L)

PO43--P level (mg/L)

50

40 35 30 25 20

15

10

5

15 10

0 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6

0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6

Time (h)

Time (h)

Fig. 3. Effect of electric voltage on struvite dissolution at 0.06% NaCl and intial struvite amount of 1.25 g/L: (a) phosphorous concentration (mg/L) and (b) ammonia concentration (mg/L).

a 70

0.25 g/L

0.50 g/L

0.75 g/L

1.00 g/L

1.25 g/L

2.50 g/L

b

14

0.25 g/L

0.50 g/L

0.75 g/L

1.00 g/L

1.25 g/L

2.50 g/L

12

NH 4-N level (mg/L)

PO43--P level (mg/L)

60 50 40 30 20

10

8 6 4 2

10 0

0.5

1

1.5

2

2.5

0 0

0.5

1

Time (h)

1.5

2

2.5

Time (h)

Fig. 4. Struvite dissolution at 7 V and 0.06% NaCl: (a) phosphorous concentration (mg/L) and (b) ammonia concentration (mg/L) with reaction time.

40 35 30 CT -C0(mg/L)

increases with increase in time. In addition, from Fig. 4b it is evident that NH4 –N concentration in the solution decreased considerably with the increase in RT. With initial struvite amount of 1.25 and 2.50 g/L, NH4 –N completely depleted from the solution in 1.5 h. Hence, to summarize, longer electrolysis times (>1.5 h) unnecessarily linger the process of electrolysis for struvite dissolution. Therefore, an electrolysis time of 1.5 h was considered to be optimal RT for high PO4 3− –P dissolution and complete NH4 –N depletion, with due concern to the energy consumption. The results from Fig. 5 clearly show that with increase in struvite amount, the PO4 3− –P level increased in the solution. The dissolved amount (CT − C0 ; ﬁnal PO4 3− –P concentration–initial PO4 3− –P concentration) of PO4 3− –P during 1.5 h electrolysis was 3.26, 11.55, 19.46, 23.08 and 28.79 mg/L at 0.25, 0.50, 0.75, 1.00 and 1.25 g/L respectively, reaching plateau beyond 1.25 g/L of struvite amount. Therefore, 1.25 g/L was considered suitable struvite amount for the dissolution through the designed electrolytic reactor. Under the above optimized parameters (7 V, 0.06% NaCl, 1.5 h and 1.25 g/L), 49.17 mg/L PO4 3− –P was dissolved, whereas NH4 –N was completely depleted in the solution.

25 20 15 10 5 0 0.25

0.5

0.75

1 1.25 1.5 Struvite (g/L)

1.75

2

2.25

2.5

Fig. 5. Phosphate dissolution at 7 V and 0.06% NaCl with different sturivte amount (g/L); CT , ﬁnal PO4 3− –P concentration; C0 , initial PO4 3− –P concentration.

3.4. Recycle ratio of struvite To examine the effectiveness of reutilization of the dissolved struvite by electrolysis method on crystallization, different recycle ratio (0.0, 0.5, 1.0 and 2.0 M) of dissolved struvite with respect to PO4 3− –P in inﬂuent was pumped with 0.5 M of MgCl2 into the crystallization reactor. From Table 3, it can be seen that with increase

in the recycle ratio, the PO4 3− –P removal (%) increases. In run I, the PO4 3− –P removal was 69%, which is signiﬁcantly (p < 0.01) higher than that of control (63%), however no signiﬁcant difference was found between run I and II. In run III, the PO4 3− –P removal (79%) was signiﬁcantly higher than run I and II. Similarly, the percent NH4 –N removal was signiﬁcantly increased with the increase

180

Y. Liu et al. / Journal of Hazardous Materials 195 (2011) 175–181

Table 3 Effect of different molar ratio of recycled struvite on swine wastewater treatment. pH Inﬂuent C RI R II R III

8.37 8.37 8.39 8.36

± ± ± ±

0.59 0.06 0.05 0.04

Efﬂuent 8.78 8.68 8.61 8.51

± ± ± ±

0.08 0.09 0.06 0.06

PO4 3− –P (mg/L) Inﬂuent 33.68 31.02 30.98 30.79

± ± ± ±

2.51 2.90 2.28 1.69

Efﬂuent 12.36 9.60 8.71 6.27

± ± ± ±

1.01 1.13 0.81 1.14

Removal (%)

NH4 –N (mg/L) Inﬂuent

63a 69b 71b 79c

2612 2624 2647 2610

± ± ± ±

21 86 33 97

Efﬂuent 2358 1806 1587 1216

± ± ± ±

58 46 26 54

Removal (%) 9a 31b 40c 53d

C, control; RI, Run I; R II, Run II; R III, Run III; PO4 3− –P, Ortho phosphate; NH4 –N, ammonium nitrogen; ±standard deviation of mean (n = 16); unequal superscripts in same column shows signiﬁcant difference at p < 0.01.

Table 4 Comparison of cost between dissolved struvite recycling process and pure chemical addition. Chemical used

Price (USD/Kg)

MgCl2 ·6H2 O KH2 PO4 NaCl Energy consumption Total

10.20 21.72 3.77 0.049/kWh

* **

Cost for the dissolved struvite recycling (USD)*

Cost for pure chemicals (USD)** 13.35 19.05

9.07 2.07 11.34

32.40

Cost of 1 m3 solution of dissolved struvite by electrolysis. Cost of pure chemicals used to make 1 m3 solution.

of recycle ratio (p < 0.01), showing 9, 31, 40 and 53% in C, run I, II and III, respectively. Thus results clearly revealed that recycle ratio affect the percent PO4 3− –P and NH4 –N removal signiﬁcantly (p < 0.01). The correlation coefﬁcient (R2 ) value for NH4 –N and PO4 3− –P% removal was found to be 0.91and 0.98, respectively, indicating that the struvite crystallization can be enhanced gradually with increased recycle ratio of dissolved struvite as Mg and P source. Since swine wastewater contains high amounts of NH4 –N compared to PO4 3− –P and Mg this may affect the stoichiometric conditions for struvite precipitation and thus affect the efﬁciency [2]. Therefore, after efﬁciently removing the NH4 –N from struvite by electrolysis, addition of the remaining Mg and PO4 3− –P increase the Mg:PO4 3− –P molar ratio compared to NH4 –N which lead to more struvite formation and thus more NH4 –N removal from the swine wastewater. Furthermore, seeding the dissolved struvite solution might have added small struvite particles, which leads to increase absorption of PO4 3− –P and thus faster growth rate of struvite [32]. In addition increased concentration of Mg2+ , NH4 –N and PO4 3− –P also improves the PO4 3− –P removal efﬁciency [33]. The struvite crystallization, struvite puriﬁcation and dissolution are three separate processes. The main focus is on struvite crystallization (works in continuous mode) whereas puriﬁcation and dissolution are auxiliary processes which can be carried out separately (once in a month or twice in six months). Struvite dissolution can also be done without puriﬁcation depending on the utility and can be used for making alternative source of PO4 3− –P and Mg for further struvite formation in the crystallization reactor. The electrolytically dissolved struvite can be collected and stored in a container, and pumped into the crystallization reactor along with the inﬂuent feeding. An additional source of Mg can be added to match the stoichiometric conditions of struvite formation. Thus keeping the main purpose of our experiment (i.e. struvite crystallization) in mind, if we could easily and economically dissolve struvite by electrolysis, we could use the dissolved solution as an alternative source of Mg and PO4 3− –P for struvite formation. 3.5. Economic evaluation of the dissolution process Under the optimum conditions obtained (7 V, 0.06% and RT 1.5 h), about 50 mg of PO4 3− –P was dissolved in the 1 L of solution. Based on these results, an economic analysis for making 1 m3 dissolved struvite solution from the electrolysis reactor was carried out and compared to the cost of using pure chemicals of PO4 3− –P

and Mg in the same quantity. In this assessment chemicals used and energy consumed were considered. The prices of chemical used and energy consumed are shown in Table 4. It is calculated that the total chemical cost and energy is 11.34 USD/m3 when dissolved struvite was recycled for crystallization, whereas the cost by using pure chemicals is 32.40 USD/m3 . The results of economic analysis indicate that the average cost could be decreased to about 1/3rd and thus greatly lower the cost of the struvite precipitation. He et al. [5] reported that about 44% of the chemical costs were reduced by recycling struvite for three cycles. Huang et al. [7,31] also reported that 48.7% and 59% of struvite precipitation cost was decreased by using a struvite recycle technology, respectively. 4. Conclusions In the present study, operating parameters to achieve electrolytically struvite dissolution were determined and effectiveness of the reutilization of the dissolved struvite as Mg and PO4 source was also investigated. From the obtained results, following conclusions can be drawn. 1. The struvite dissolution was enhanced with increased NaCl concentration and electric voltage level, however from the study 0.06% NaCl and 7 V could be considered as optimum. 2. RT of 1.5 h was found to be suitable for PO4 3− –P dissolution and complete removal of NH4 + from struvite. PO4 3− –P dissolution increased with increase of initial struvite amount, but a plateau occurred after 1.25 g/L, suggesting 1.25 g/L of struvite to be suitable. 3. Recovery of P and N removal from swine wastewater was significantly enhanced with the recycle of electrolytically dissolved struvite (p < 0.01) and the efﬁciencies were proportional to the recycle ratio. 4. Therefore, electrolysis could be a practical approach for struvite dissolution and recycling of the dissolved struvite as Mg2+ and PO4 3− –P source would be a good strategy to enhance struvite crystallization. Acknowledgements This research was performed with the support of Rural Development Administration (RDA), Korea. Also, this work was supported

Y. Liu et al. / Journal of Hazardous Materials 195 (2011) 175–181

in part by a grant from the Institute of Animal Resources at Kangwon National University. References [1] L. Shu, P. Schneider, V. Jegatheesan, J. Johnson, An economic evaluation of phosphorus recovery as struvite from digester supernatant, Bioresour. Technol. 97 (2006) 2211–2216. [2] Y.H. Liu, J.H. Kwag, J.H. Kim, C.S. Ra, Recovery of nitrogen and phosphorus by struvite crystallization from swine wastewater, Desalination 277 (2011) 364–369. [3] J.D. Doyle, S.A. Parsons, Struvite formation, control and recovery, Water Res. 36 (16) (2002) 3925–3940. [4] S.I. Lee, S.Y. Weon, C.W. Lee, B. Koopman, Removal of nitrogen and phosphate from wastewater by addition of bittern, Chemosphere 51 (4) (2003) 265–271. [5] S.L. He, Y. Zhang, M. Yang, W. Du, H. Harada, Repeated use of MAP decomposition residues for the removal of high ammonium concentration from landﬁll leachate, Chemosphere 66 (2007) 2233–2238. [6] M. Türker, I. Celen, Removal of ammonia as struvite from anaerobic digester efﬂuents and recycling of magnesium and phosphate, Bioresour. Technol. 98 (2007) 1529–1534. [7] H.M. Huang, C.L. Xu, W. Zhang, Removal of nutrients from piggery wastewater using struvite precipitation and pyrogenation technology, Bioresour. Technol. 102 (2011) 2523–2528. [8] T. Zhang, L. Ding, H. Ren, Pretreatment of ammonium removal from landﬁll leachate by chemical precipitation, J. Hazard. Mater. 166 (2009) 911–915. [9] K. Yetilmezsoy, Z. Sapci-Zengin, Recovery of ammonium nitrogen from the efﬂuent of UASB treating poultry manure wastewater by MAP precipitation as a slow release fertilizer, J. Hazard. Mater. 166 (2009) 260–269. [10] P. Wang, I.W.C. Lau, H.H.P. Fang, Electrochemical oxidation of leachate pretreated in an up ﬂow anaerobic sludge blanket reactor, Environ. Technol. 22 (2001) 373–381. [11] A.G. Vlyssides, C.J. Israilides, Detoxiﬁcation of tannery waste liquors with an electrolysis system, Environ. Pollut. 97 (1997) 147–152. [12] L. Szpyrkowicz, S.N. Kaul, R.N. Neti, S. Satyanarayan, Inﬂuence of anode material on electrochemical oxidation for the treatment of tannery wastewater, Water Res. 39 (2005) 401–413. [13] A.G. Vlyssides, P.K. Karlis, N. Rori, A.A. Zorpas, Electrochemical treatment in relation to pH of domestic wastewater using Ti/Pt electrodes, J. Hazard. Mater. B95 (2002) 215–226. [14] Y. Vanlangendonck, D. Cornisier, A.V. Lierde, Inﬂuence of operating conditions on the ammonia electro-oxidation rate in wastewaters from power plants, Water Res. 39 (2005) 3028–3034. [15] Z.M. Xie, X.Y. Li, K.Y. Chan, Nitrogen removal from the saline sludge liquor by electrochemical denitriﬁcation, Water Sci. Technol. 54 (2006) 171–179.

181

[16] M. Panizza, G. Cerisola, Electrochemical oxidation as a ﬁnal treatment of synthetic tannery wastewater, Environ. Sci. Technol. 38 (2004) 5470–5475. [17] American Public Health Association (APHA), Standard Methods for the Examination of Water and Wastewater, 10th edition, APHA, Washington, DC, 1995. [18] W.L. Chou, C.T. Wang, K.Y. Huang, Effect of operating parameters on indium (III) ion removal by iron electrocoagulation and evaluation of speciﬁc energy consumption, J. Hazard. Mater. 167 (2009) 467–474. [19] S. Irdemez, Y.S. Yildiz, V. Tosunoglu, Optimization of phosphate removal from wastewater by electrocoagulation with aluminum plate electrodes, Sep. Purif. Technol. 52 (2006) 394–401. [20] O. Abdelwahab, N.K. Amin, E-S.Z. El-Ashtoukhy, Electrochemical removal of phenol from oil reﬁnery wastewater, J. Hazard. Mater. 163 (2009) 711–716. [21] M.I.H. Bhuiyan, D.S. Mavinic, R.D. Beckie, A solubility and thermodynamic study of struvite, Environ. Technol. 28 (2007) 1015–1026. [22] C.T. Wang, W.L. Chou, Y.M. Kuo, Removal of COD from laundry wastewater by electrocoagulation/electroﬂotation, J. Hazard. Mater. 164 (2009) 81–86. [23] L. Li, Y. Liu, Ammonia removal in electrochemical oxidation: mechanism and pseudo-kinetics, J. Hazard. Mater. 161 (2009) 1010–1016. [24] S.H. Lin, C.T. Shyu, M.C. Sun, Saline wastewater treatment by electrochemical method, Water Res. 32 (1998) 1059–1066. [25] J.S. Do, W.C. Yeh, Paired electrooxidative degradation of phenol with in situ electrogenerated hydrogen peroxide and hypochlorite, J. Appl. Electrochem. 26 (1996) 673–678. [26] G. Chen, Electrochemical technologies in wastewater treatment, Sep. Purif. Technol. 38 (2004) 11–41. [27] M. Kobya, H. Hiz, E. Senturk, C. Aydiner, E. Demirbas, Treatment of potato chips manufacturing wastewater by electrocoagulation, Desalination 190 (2006) 201–211. [28] Y. Liu, L. Li, R. Goel, Kinetic study of electrolytic ammonia removal using Ti/IrO2 as anode under different experimental conditions, J. Hazard. Mater. 167 (2009) 959–965. [29] M.Y.A. Mollah, S.R. Pathak, P.K. Patil, M. Vayuvegula, T.S. Agrawal, J.A.G. Gomes, M. Kesmez, D.L. Cocke, Treatment of orange II azo-dye by electrocoagulation (EC) technique in a continuous ﬂow cell using sacriﬁcial iron electrodes, J. Hazard. Mater. 109 (2004) 165–171. [30] A.R. Khataee, V. Vatanpour, A.R.A. Ghadim, Decolorization of C.I. Acid Blue 9 solution by UV/Nano-TiO2 , Fenton Fenton-like, electro-Fenton and electrocoagulation processes: a comparative study, J. Hazard. Mater. 161 (2009) 1225–1233. [31] H.M. Huang, X.M. Xiao, B. Yan, Recycle use of magnesium ammonium phosphate to remove ammonium nitrogen from rare-earth wastewater, Water Sci. Technol. 59 (6) (2009) 1093–1099. [32] Z.G. Liu, Q.L. Zhao, D.J. Lee, N. Yang, Enhancing phosphorous recovery by a new internal recycle seeding MAP reactor, Bioresour. Technol. 99 (2008) 6488–6493. [33] J.S. Wang, Y.H. Song, P. Yuan, J.F. Peng, M.H. Fan, Modeling the crystallization of Magnesium ammonium phosphate for phosphorous recovery, Chemosphere 65 (2006) 1182–1187.

Journal of Hazardous Materials 195 (2011) 182–187

Contents lists available at ScienceDirect

Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat

Genotoxicity and biodegradation of quaternary ammonium salts in aquatic environments ∗ ˙ ´ Elzbieta Grabinska-Sota Department of Environmental Biotechnology, Technical University of Silesia, 2 Akademicka Street, Gliwice 44-101, Poland

a r t i c l e

i n f o

Article history: Received 8 January 2011 Received in revised form 14 July 2011 Accepted 8 August 2011 Available online 16 August 2011 Keywords: Quaternary ammonium salts Biodegradation Genotoxicity Degradation kinetics Spectral analysis 1 H NMR and 13 C NMR

a b s t r a c t Biodegradation tests were conducted for three groups of quaternary ammonium salts (QAS) that differed in hydrophobic chain length or in hydrophilic properties. The degradation rate was inﬂuenced by the hydrocarbon chain length, the presence of aromatic or cyclic rings, and the occurrence of sulphur and oxygen atoms in the alkyl substituent. All tested QAS variants were biodegradable in an aquatic environment. The half life of the different QAS under these conditions ranged from 0.5 to 1.6 days and depended on the properties of the compound. Biodegradation intermediate products were identiﬁed by nuclear magnetic resonance spectrometry (1 H NMR and 13 C NMR). Both the initial preparations and their biodegradation products were not genotoxic. © 2011 Elsevier B.V. All rights reserved.

1. Introduction

2. Experimental

Quaternary ammonium salts (QAS) are molecules with at least one long, hydrophobic alkyl chain attached to a positively charged nitrogen atom. QAS are cationic surfactants that are widely used in industry, biotechnology, medicine, pharmacology and in biocide production. They are also used as active ingredient of many cosmetics. They have antimicrobial, fungicidal, algaecidal, antielectrostatic and anticorrosive properties [1–4]. Due to the wide spread use of QAS, leakage into aquatic environments occurs with rain and wastewater from municipal and industrial sources. QAS can harm organisms that live in water and affect both animals and plants. There is also a possibility that the salts can undergo biological and chemical degradation. These substances, considered harmful to human health, have been found in surface water at concentrations up to 3.8 ␮g L−1 [5]. There is evidence that QAS have genotoxic properties and interact with DNA molecules [6–8]. Consequently, studies on genotoxicity and biodegradation of newly synthesised QAS in aquatic environments are of great interest.

2.1. Compounds Three groups of quaternary ammonium salts were investigated. They differed from each other in hydrophobic chain length or hydrophilic properties. Properties of the substances used in the study are given in Table 1. Three groups of ten compounds that were tested are classiﬁed as A, B and I (Table 1): (A) alkylalkoxymethylammonium chlorides (B) alkylbenzyldimethylammonium chlorides, and (C) imidazolium chlorides. The compounds were prepared by Professor Juliusz Pernak’s ´ Poland). group (University of Technology, Poznan, All compounds were soluble in well water at the investigated concentrations and contained over 95% active molecules. Group A and I compounds act strongly against microorganisms. They possess antiseptic, fungicidal and antielectrostatic properties [4,9,10]. Group B compounds are used as a coating in galvanic baths during nickel plating and zinc plating. Group I chlorides have also recently been used as fungicides for wood protection. 2.2. Biodegradation tests

∗ Corresponding author. Tel.: +48 32 237 21 29. E-mail address: [email protected] 0304-3894/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.jhazmat.2011.08.026

Laboratory tests on QAS biodegradation were conducted using a modiﬁed version of the Simulation Test described in the Orga-

E.. Grabi´ nska-Sota / Journal of Hazardous Materials 195 (2011) 182–187

183

Table 1 Symbols and molecular weights of tested quaternary ammonium compounds. Code

Name of compound

Molecular weight (g mol−1 )

A-1 A-2 A-3 A-4 B-1 B-2 B-3 I-1 I-2 I-3

Octyldimethyldecyloxymethylammonium chloride Dodecyldimethyloctyloxymethylammonium chloride Dodecyldimethylnonyloxymethylammonium chloride Dodecyldimethyldecyloxymethylammonium chloride Benzyldimethyloctyloxymethylammonium chloride Benzyldimethyldecyloxymethylammonium chloride Benzyldimethyldodecyloxymethylammonium chloride 1–Decyl-3-hexyloxymethylimidazolium chloride 1–Decyl-3-cyklohexyloxymethylimidazolium chloride 1–Decyl-3-hexylthiomethylimidazolium chloride

363.5 391.5 405.0 419.0 313.5 341.5 369.5 358.5 356.5 374.5

nization for Economic Cooperation and Development (OECD) Guidelines [11]. Assays were conducted under conditions that simulate a natural aquatic environment. However, instead of 1.0 L conical ﬂasks, 40 L glass tanks were used. Such a large quantity of water was required for the analysis of degradation products by (1 H and 13 C NMR). Tanks were ﬁlled with settled tap water and inoculated with activated sludge microorganisms at a concentration of 1 ml L−1 . Each QAS variant was then added at a initial concentration of 1 mg L−1 . Cultures were aerated because shaking was technically impossible. No foaming was observed during aeration of the water in the glass tanks. The QAS biodegradation test was performed under conditions where the substrates were the only source of carbon and energy for a mixed population of microorganisms.The population of microorganisms used had been previously adapted to degrade the substrates. The cultures were grown at 20 ± 2 ◦ C in the, dark. The initial concentration was the result of the following issues: - investigated preparations appeared to be very toxic to aquatic microﬂora, - concentrations of QAS appearing in surface waters ranging from 0.1 to 3.8 ␮g L−1 [5]. 2.3. Genotoxicological tests Genotoxicological tests were performed using Bacillus subtilis H17 Rec+ and M45 Rec− strains described by Kada et al. [12]. The strains were cultivated in B-2 bouillon for 24 h and then sieved into a Petri dish containing the same media solidiﬁed with agar. A sterile, 1 cm paper disc soaked with each compound was placed at the centre of each dish. B. subtilis strains were incubated at 4 ◦ C for 24 h and then at 37 ◦ C for the next 24 h. After 48 h of incubation, the diameter of single colonies for both strains was measured and compared with that of growth controls. Based on the differences in size, each tested compound was assigned to one of three different classiﬁcations: not genotoxic (−), with a difference in diameter equal to = 2 mm; potentially genotoxic (±), with a difference in diameter between >2 and 4 mm; and genotoxic (+) with a difference larger than >4 and 6 mm. QAS concentrations varied between 0.1 mg and 10 mg of active ingredient per litre. B. subtilis strains H17 Rec+ and M45 Rec− used in the test were obtained from the Department of Microbiology, Faculty of Biology and Environmental Protection at, Silesian University in Katowice, Poland.

2.4. Analytical methods 2.4.1. Disulphine blue analysis Concentrations of QAS were measured by a conventional disulphine blue active substance test (DBAS) [13]. Brieﬂy, this method is a spectrophotometric test based on the formation of a chloroform

soluble blue complex of the cationic surfactant with the anionic dye, disulphine blue. The sensitivity of the DBAS test is 0.01–0.02 mg L−1 . 2.4.2. 1 H NMR and 13 C NMR spectrum analysis 1 H and 13 C NMR were used for the identiﬁcation of QAS and their degradation products. The volume of collected samples was 10.0 L. The samples were extracted with chloroform three times, and the extracted samples were concentrated under vacuum, resulting in a solid residue. The extracts were concentrated in two stages, using 1000 ml and 25 ml ﬂasks in succession. The solid residue obtained was dissolved in 1 ml of deuterated chloroform (CDCl3 ). 1 H and 13 C NMR spectra were recorded with a Varian Model XL 300 MHz Spectrometer in DMSO-d6 or CDCl3 , at 20 ◦ C with tetramethylsilane as a standard. 2.4.3. DOC analysis The concentration of dissolved organic carbon (DOC) in ﬁltered culture medium was determined with a Beckman Industrial Model 915B Tocamaster total organic carbon analyser. 2.5. Interpretation of biodegradation results The biodegradation of the compounds in “river water” was measured as a function of time and evaluated by statistical methods. For most QAS, the degradation process in water proceeded via ﬁrst-order kinetic; the concentration decreased exponentially with time: dC = −KC dt

(1)

where K is the degradation constant, C is the QAS concentration at time t, and t is the duration of the experiment. The half-life (t50 ) of the compounds could also be calculated. t50 =

ln 2 K

(2)

3. Results and discussion The dynamics of QAS degradation differed in individual groups of substrates (Figs. 1–3). It was found that, among the examined compounds, the most rapid, primary biodegradation occurred for benzyl(alkoxymethyl)dimethylammonium chlorides. After 6 days of incubation, these compounds were no longer present determined by disulﬁne blue analysis (Fig. 2). These results conﬁrm previous research that suggests that reports of the negative effects attributed to the biodegradation of QAS compounds containing benzene rings may not be correct [14]. Alkyl(alkoxymethyl)dimethylammonium chlorides were no longer present in water after 7 days (Fig. 1). Cultures containing alkyldimethylimidazolium chlorides had the slowest rate of biodegradation. Trace concentrations of these compounds were

184

E.. Grabi´ nska-Sota / Journal of Hazardous Materials 195 (2011) 182–187

Concentration [mg/L]

1,2

Table 2 Biodegradabilities of tested QAS.

1

Compound

0,8 0,6 0,4 0,2 0 0

5

10

15

20

Time [day] A-1

A-2

A-3

Concentration [mg/L]

1 0,8 0,6 0,4 0,2 0 10

15

20

Time [day] B-1

B-2

B-3

Fig. 2. Changes in concentration of benzyl(alkoxymethyl)dimethylammonium chlorides during biodegradation in the “river-water” test.

found after 10 days of incubation, but a after 19 days, there was no measurable amount based on disulphine blue analysis (Fig. 3). Analysis of organic compounds by DOC indicated 83.0–93.4% biodegradation depending on the starting QAS (Table 2). The results presented here are similar to those of French studies that found that alkylimidazolium compounds decay in river water after 7–14 days [15].

Concentration [mg/L]

1,2 1 0,8 0,6 0,4 0,2 0 0

5

10

15

20

Time [day] I-1

I-2

DOC

100 100 100 100 100 100 100 100 100 100

87.0 88.0 89.4 90.0 93.4 93.0 89.5 85.8 83.0 84.4

A-4

1,2

5

DBAS

DBAS – disulphine blue active substances, DOC–dissolves organic carbon.

Fig. 1. Changes in concentration of alkyl(alkoxymethyl)dimethylammonium chlorides during biodegradation in the “river-water” test.

0

A-1 A-2 A-3 A-4 B-1 B-2 B-3 I-1 I-2 I-3

Biodegradability (%) after 19 days

I-3

Fig. 3. Changes in concentration of alkyldimethylimidazolium chlorides during biodegradation in the “river-water” test.

Biodegradation of tested compounds in a model aquatic ecosystem was described by the function c = f(t), where c is the – substance C concentration at time t and f (t) = ln Cp (Table 3). Correlation coef0 ﬁcients, which ranged from −0.811 to −0.971 and are listed in Table 3, conﬁrm the assumption that degradation of QAS may be estimated based on the ﬁrst-order equation described above. Half-life times of eight-carbon chlorides ranged from 0.38 d to 0.52 d (for B-1 and A-4, respectively). Maximum values, up to 0.67 d and 0.81 d, were estimated for compounds that contained twelve carbon atoms (B-3 and A-2) (Table 3). Compound A-1 has the highest antiseptic properties, which probably accounts for the longest measured half-life (0.97 d) in this group of molecules (Table 3) [16]. Values obtained in this study are similar to those presented by Ruiz Cruz and Garcia who have conducted extensive studies about the dependence of the chemical structure of cationic, surface-active agents on biodegradation in river water [17]. They have shown that the half-life of QAS compounds, with 12 carbon atoms in the alkyl chain ranged from less than one day to up to a few days. Nishiyama has suggested that biological oxidation velocity of group A compounds was two, or even three, times slower than for group B and I preparations, which was probably caused by their high antiseptic properties[18]. Pernak et al. have shown that imidazolium salts have a higher antimicrobial activity than alkyldimethylammonium salts [19]. They also stated that the antiseptic effect was increased when a cycloalkoxymethyl substituent was attached to the molecule. These studies indicated that replacement of one alkyl chain (I-1) with cyclohexane caused an increase half-life time from 1.36 d to 1.69 d (Table 3). Replacement of the oxygen atom (I-1) with a sulphur atom (I-3) had less inﬂuence on decreasing the reaction velocity than with the increase of the half-life of the compound. Values obtained for all the groups of compounds tested were lower than the ones reported by Vives-Rego and Woltering, who showed that the half-life for quaternary ammonium salts ranged from 4 to 14 days [20,21]. Previous adaptation of microorganisms for QAS preparations could be a possible explanation for this discrepancy. It is known that previous adaptation of microorganisms can shorten the half-life by a factor of 14–50 [17]. The half-life of the I-2 compound was the longest biodegradation mechanism of the based on 1 H NMR (Fig. 4) and 13 C NMR (Fig. 5) analysis. Analysis of the NMR spectra showed, that degradation of 1decyl-3-cyclohexyloxymethylimidazolium chloride was correlated to the decomposition of the molecule at the position, where the hydrophobic chain was connected to the nitrogen. As a result of microbiological degradation, intermediate products, such as alcohols and dimethyloamines, were formed. Alcohols were not oxidised and were detected after the biodegradation process. It is assumed, that dimethyloamine was degraded to methyloamine, and this intermediate was further degraded to NH4 + , which can be used as a nutrient for the microorganisms. After the biodegrada-

E.. Grabi´ nska-Sota / Journal of Hazardous Materials 195 (2011) 182–187

Fig. 4.

1

Fig. 5.

13

H NMR spectrum for 1-decyl-3-cyclohexyloxymethylimidazolium chloride at different stages of the biodegradation process.

C NMR spectrum for 1-decyl-3-cyclohexyloxymethylimidazolium chloride at different stages of the biodegradation process.

CH3

N-dealkylation

N+ CH2O

ω-oxidation

H

N

H + CH3(CH2 )8CH2OH +

+ CO2 + H2 O + biomass

CH3

N

biomass CO2 + H2 O + biomasa

CH2(CH2 )8CH3 CH 3 H

N

+

CH 3

CH 3

N-demethylation

H

H HCHO H2O + CO2

N H

+

N-demethylation

H

NH4+

biomasa biomass

HCHO H2O + CO2

Fig. 6. Hypothetical mechanism of the biological decomposition of 1-decyl-3-cyclohexyloxymethyl imidazolium chloride.

185

186

E.. Grabi´ nska-Sota / Journal of Hazardous Materials 195 (2011) 182–187

Table 3 Kinetic parameters of reaction and correlation coefﬁcients for tested quaternary ammonium salts. Cp C0

= f (t)

Correlation coefﬁcient between C and tR2

Degradation constant (1/d) K ± s

Half-life time t50 (d)

−0.31x + 0.093 −0.37x + 0.083 −0.39x + 0.093 −0.59x + 0.063

−0.912 −0.879 −0.811 −0.956

0.71 ± 0.08 0.85 ± 0.07 0.89 ± 0.07 1.36 ± 0.1

0.97 0.81 0.77 0.50

B-1 B-2 B-3

−0.78x + 0.027 −0.67x + 0.06 −0.45x + 0.10

−0.966 −0.889 −0.830

1.8 ± 0.06 1.54 ± 0.14 1.03 ± 0.05

0.38 0.45 0.67

I-1 I-2 I-3

−0.22x + 0.012 −0.18x + 0.023 −0.2x + 0.19

−0.951 −0.971 −0.894

0.5 ± 0.04 0.41 ± 0.04 0.46 ± 0.06

1.38 1.69 1.50

Quternary ammonium salts group

Compound symbol

Equation ln

Alkylalkoxymethyldime-thylammonium chlorides

A-1 A-2 A-3 A-4

Benzylalkoxymethyl-dimethylammonium chlorides Alkylmethylimidiazolium chlorides

Cp – initial concentration of tested compound, C0 – concentration of tested compound after time t, t – time, s – standard deviation. Table 4 Genotoxicological effects of the investigated QAS on Bacillus subtilis. Compound concentration mg L−1

0.01 0.1 0.5 1 3 10

Difference between grow controls [mm]

(−) – non genotoxic properties, (±) – potentially genotoxic properties

A-1

A-2

A-3

A-4

B-1

B-2

B-3

I-1

I-2

I-3

0 (−) 1 (−) 1 (−) 1 (−) 1 (−) 2 (−)

0 (−) 0 (−) 0 (−) 0 (−) 1 (−) 2 (−)

0 (−) 0 (−) 0 (−) 1 (−) 2 (−) 2 (−)

0 (−) 1 (−) 0 (−) 2 (−) 1 (−) 2 (−)

1 (−) 1 (−) 1 (−) 1 (−) 2 (−) 2 (−)

0 (−) 0 (−) 0 (−) 0 (−) 2 (−) 3 (±)

0 (−) 0 (−) 0 (−) 0 (−) 1 (−) 1 (−)

0 (−) 1 (−) 1 (−) 1 (−) 1 (−) 1 (−)

0 (−) 1 (−) 1 (−) 2 (−) 2 (−) 2 (−)

0 (−) 1 (−) 1 (−) 1 (−) 2 (−) 2 (−)

tion process, neither the 1 H spectrum or 13 C spectrum showed the presence of any protons connected to nitrogen. The cyclohexane ring was presumed to be oxidised to CO2 and H2 O because, at the end of the process, the aromatic compounds were not present in the solution (Fig. 6). Intermediates were not genotoxic in any of the tests. Data relating to mutagenic properties are given in Table 4. The results indicated that tested compounds did not have mutagenic properties. Only the B-2 compound seems to be potentially mutagenic and, even then,only at concentrations 10 mg L−1 . Nevertheless, this concentration is much higher than normally occurs in drinking water or natural bodies of water. Cationic surfactant concentrations have been measured between 0.1 and 3.8 ␮g L−1 , in the Tamagawa, Arkawa, Edogawa and Yodogawa Rivers in Japan [5]. There is no monitoring system of QAS concentration in the surface water in Poland. Genotoxicity of the products formed from QAS biodegradation was also investigated in this study, and no toxic effect was found.

4. Conclusions Degradation testing with simulated river water was conducted with quaternary ammonium chlorides that contained alkoxymethyl or alkylmethyl substituents. The test showed that examined preparations were degraded at different periods of time. The calculated biodegradation rate constants indicate that the shortest degradation time was obtained for alkyl- and benzyl(alkoxymethyl)dimethylammonium chlorides. It was also found that biodegradation velocity decreased with increasing number of carbon atoms in the alkyl chain. The half-lives of the compounds were also estimated,and they ranged from 0.5 to 1.6 days depending on the type of QAS tested. Analysis of the data showed that all examined compounds were biodegradable in an aquatic environment. Based on the 1 H and 13 C NMR analysis of 1-decyl-3-cyklohexyloxymethyloimidazolium chloride, the hypothetical mechanism of QAS biodegradation was proposed. After biodegradation, only alcohols were present in the solution. The

examined substances, their intermediates, and ﬁnal biodegradation products did not cause genotoxic effects. These ﬁndings suggest that the presecnce of QAS compounds in natural bodies of water and thier consumption by humans will not create any carcinogenice health hazards. Acknowledgement The author is grateful to the team of Professor Juliusz Pernak from the Technical University of Poznan´ for the synthesis of the compounds used in this study. References [1] Ch. Zhang, U. Tezel, K. Li, D. Liu, R. Ren, J. Du, S.G. Pavlostathis, Evaluation and modeling of benzalkonium chloride inhibition and biodegradation in activated sludge, Water Res. 45 (2011) 1238–1246. [2] L. Huang, Y. Xiao, X. Xing, F. Li, S. Ma, L. Qi, J. Chen, Antibacterial activity and cytotoxicity of two novel cross-linking antibacterial monomers on oral pathogens, Arch. Oral Biol. 56 (2011) 367–373. [3] Y.S. Kim, H.W. Kim, S.H. Lee, K.S. Shin, H.W. Hur, Y.H. Rhee, Preparation of alginate-quaternary ammonium complex beads and evaluation of their antimicrobial activity, Int. J. Biol. Macromol. 41 (2007) 36–41. ´ J. Placzek, A. Skrzypczak, B. Predki, Analysis of relationships [4] J. Krysinski, between structure, surface properties, and antimicrobial activity of quaternary ammonium chlorides, QSAR Comb. Sci. 28 (2009) 995–1002. [5] K. Miura, N. Nishiyama, A. Yamamoto, Aquatic environmental monitoring of detergent surfactants, J. Oleo Sci. 57 (2008) 161–170. [6] P.S. Kuhn, M.C. Barbosa, Y. Levin, Complexation of DNA with cationic surfactant, Physica A 269 (1999) 278–284. [7] T. Deutschle, U. Porkert, R. Reiter, T. Keck, H. Riechelmann, In vitro genotoxicity and cytotoxicity of benzalkonium chloride, Toxicol. In Vitro 20 (2006) 1472–1477. [8] F. Ferk, M. Misik, C. Hoelzl, M. Uhl, M. Fuerhacker, Benzalkonium chloride (BAC) and dimethyldioctadecyl-ammonium bromide (DDAB), two common quaternary ammonium compounds, cause genotoxic effects in mammalian and plant cells at environmentally relevant concentrations, Mutagenesis 22 (2007) 363–370. ´ A. Skrzypczak, G. Demski, B. Predki, Application of the rough set [9] J. Krysinski, theory in structure activity relationships of antielectrostatic imidazolium compounds, Quant. Struct.-Act. Relat. 20 (2002) 395–401. [10] J. Zabielska-Matejuk, Antifungal properties of new quaternary ammonium compounds in relation to their surface activity, Wood Sci. Technol. 39 (2005) 235–243.

E.. Grabi´ nska-Sota / Journal of Hazardous Materials 195 (2011) 182–187 [11] Guideline for Testing of Chemicals, Aerobic Mineralisation in Surface WaterSimulation Biodegradation Test, Procedure 309, Organization for Economic Cooperation and Development OECD, Paris, 2004. [12] T. Kada, K. Hirano, Y. Shirasu, Screening of environmental chemical mutagens by the rec-assay system with Bacillus subtilis, in: F.J. de Serres, Hollaender (Eds.), Chemical Mutagenesis, Plenum Publishing Corp., New York, 1980, pp. 149–173. [13] J. Waters, Analysis of low concentrations of cationic surfactants in laboratory test liquars and environmental sample, in: J. Cross, E. Singer (Eds.), Cationic Surfactants. Analytical and Biological Evaluation, M. Dekker Inc., New York, Basel, 1994, pp. 235–256. [14] J. Sánchez, K. González, K.L.E. Kaiser, V.S. Palabrica, F. Comelles, M.T. Garcia, On toxicity and biodegradaton of cationic surfactants, Acta Hydrochim. Hydrobiol. 22 (1994) 13–18. [15] B. Baleux, P. Caumette, Biodegradation de quelques agents de surface cationiques, Water Res. 11 (1976) 833–841.

187

[16] A. Skrzypczak, B. Brycki, I. Mirska, J. Pernak, Synthesis antimicrobial activities of new quants, Eur. J. Med. Chem. 32 (1997) 661–668. [17] R.S. Boethling, Environmental aspects of cationic surfactants, in: J. Cross, E. Singer (Eds.), Cationic Surfactants. Analytical and Biological Evaluation, M. Dekker Inc., New York, Basel, 1994, pp. 96–135. [18] N. Nishiyama, Y. Toshima, Y. Ikeda, Biodegradation of alkyltrimethylammonium salts in activated sludge, Chemosphere 30 (1995) 593– 603. [19] J. Pernak, J. Zabielska-Matejuk, E. Urbanik, New quaternary ammonium chorides-wood preservatives, Holzfforschung 52 (1998) 249– 254. [20] J. Vives-Rego, M.D. Vaque, L.J. Sanchez, J. Parra, Surfactants biodegradation in sea water, Tenside Surf. Det. 24 (1987) 20–22. [21] D.M. Woltering, R.J. Larson, W.D. Hosing, R.A. Jamieson, The environmental fate and effects of detergent, Tenside Surf. Det. 24 (1987) 286–295.

Journal of Hazardous Materials 195 (2011) 188–194

Contents lists available at ScienceDirect

Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat

Ecotoxicological impacts of cloﬁbric acid and diclofenac in common carp (Cyprinus carpio) ﬁngerlings: Hematological, biochemical, ionoregulatory and enzymological responses Manoharan Saravanan, Subramanian Karthika, Annamalai Malarvizhi, Mathan Ramesh ∗ Unit of Toxicology, Department of Zoology, School of Life Sciences, Bharathiar University, Coimbatore 641 046, Tamil Nadu India

a r t i c l e

i n f o

Article history: Received 18 March 2011 Received in revised form 6 August 2011 Accepted 8 August 2011 Available online 16 August 2011 Keywords: Cloﬁbric acid Diclofenac Cyprinus carpio Acute toxicity Hematology Biochemical parameters

a b s t r a c t Investigation on the toxic effects of pharmaceutical drugs namely cloﬁbric acid (CA) and diclofenac (DCF) were studied in a common carp Cyprinus carpio at different concentrations such as 1, 10 and 100 ␮g L−1 for a short-term period of 96 h under static bioassay method. At all concentrations, red blood cell (RBC), plasma sodium (Na+ ), potassium (K+ ), and glutamate oxaloacetate transaminase (GOT) levels were decreased in ﬁsh treated with CA and DCF. Contrastingly, white blood cell (WBC), plasma glucose, protein, lactate dehydrogenase (LDH) and gill Na+ /K+ -ATPase level were increased. However, a mixed trend was observed in hemoglobin (Hb), hematocrit (Hct), plasma chloride (Cl− ), mean cellular volume (MCV), mean cellular hemoglobin (MCH), mean cellular hemoglobin concentration (MCHC) and glutamate pyruvate transaminase (GPT) levels. There was a signiﬁcant (P < 0.01 and P < 0.05) change in all parameters measured in ﬁsh exposed to different concentrations of CA and DCF. In summary, the alterations in hematological, biochemical, ionoregulatory and enzymological parameters can be used as biomarkers in monitoring the toxicity of CA and DCF in aquatic environment. However, more detailed studies on using of speciﬁc biomarkers to monitor the human pharmaceuticals are needed. © 2011 Elsevier B.V. All rights reserved.

1. Introduction Pharmaceuticals drugs are produced and prescribed to cure the diseases, and to improve human health [1,2]. These drugs enter into aquatic environment through domestic waste waters, disposal from medical centres, excretion via water and sewage treatment systems [3–5]. The occurrence and detection of various pharmaceutical drugs in the environment, particularly in surface water, ground water, drinking water and inﬂuents and efﬂuents from the wastewater treatment plant has been reported [6–9]. Rosal et al. [4] reported that the presence of pharmaceutical drugs even at low concentration level (ranging from ␮g L−1 to ng L−1 ) may lead to public health problems. Cloﬁbric acid (CA) is an active derivative substance of cloﬁbrate, designed to improve lipid metabolism in human [10]. Occurrence of CA in the environment has been reported in many countries. For example, 1 ng L−1 in the North Sea [11]; 0.55l ␮g L−1 in surface waters of Swiss lakes [12]; 1.6 ␮g L−1 in efﬂuents from sewage treatment plants, Germany [13]; 0.8–2 ␮g L−1 in USA [14] and 5 ng L−1 in Greece [15]. Diclofenac (DCF), one of the most important non-steroidal anti-inﬂammatory drugs (NSAID) is used in conditions like inﬂam-

∗ Corresponding author. Tel.: +91 422 2428394, fax: +91 422 2422387. E-mail address: [email protected] (M. Ramesh). 0304-3894/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.jhazmat.2011.08.029

mation, arthritis, menstrual, dysmenorrheal and rheumatic disease and also acts as a cyclooxygenase inhibitor [16]. DCF is detected extensively in different water bodies throughout the world because of its higher amount of usage and production. There are many reports available on the occurrence of DCF such as 1030 ng L−1 in surface waters [17]; 2 ng L−1 in drinking water well (Mediterranean region) [18]; 0.38 ␮g L−1 in groundwater (Berlin) [19] and 195 ng L−1 in Mersey Estuary (UK) [20] and 6.2 ng L−1 in Estuary of the River Elbe (North Sea) [11]. The continuous discharge and occurrence of pharmaceutical drugs in the aquatic ecosystem has become a major problem due to either high persistence or biological activity [2]. Many studies have been conducted based on acute toxicity test using laboratory organisms belonging to different tropic levels. Embry et al. [21] reported that aquatic toxicity data on acute responses to anthropogenic chemicals by ﬁsh plays a very important role. Powers [22] suggested that ﬁsh models are increasingly used in the early phases of pharmaceutical development and its toxicity evaluation. However, most investigations have been limited to lethal effects during acute exposures [23]. Assessment of polluted water bodies and aquatic animal health biomarkers are widely used as early diagnostic tools [24]. Hematological, biochemical, ionoregulatory and enzymological parameters have been routinely used as valuable biomarkers to assess the toxicity of environmental contaminants in aquatic ecosystem [25]. The hematological variables such as hemoglobin (Hb), hematocrit

M. Saravanan et al. / Journal of Hazardous Materials 195 (2011) 188–194

(Hct), red blood cell (RBC) count, white blood cell (WBC) count, and hematological indices such as mean cellular volume (MCV), mean cellular hemoglobin (MCH) and mean cellular hemoglobin concentration (MCHC), and biochemical parameters like plasma glucose and protein are widely used to assess the toxic stress induced by environmental contaminants. Ion levels in plasma as measured by osmolality or speciﬁc ion concentrations of sodium (Na+ ), potassium (K+ ), and chloride (Cl− ) have potential as sensitive biomarkers of environmental chemical exposure. However these ions are very sensitive to environment stressors [26,27]. Enzyme activities have also been used as sensitive indicator of stress in ﬁsh exposed to diverse group of water pollutants and also to predict the possible level of threat to life [28]. Among the enzymes, transaminases like glutamate oxaloacetate transaminase (GOT) and glutamate pyruvate transaminase (GPT) play a vital role in protein and carbohydrate metabolism and act as an indicator for tissue damage and cell rupture [29]. Lactate dehydrogenase (LDH) also used as indicative criteria of exposure due to chemical stress and anaerobic capacity of tissue [30]. In addition to these, gill Na+ /K+ -ATPase is involved in osmoregulation of ﬁsh and widely used as a sensitive indicator of environmental contaminants. Studies on the potential adverse ecological impacts of pharmaceutical drugs and its residues on the physiology of aquatic organisms are scarce. CA is persistent for a long time (e.g., approximately 21 years) in the environment [7,31] and is considered as a potential endocrine disruptor, because it interferes with the synthesis of cholesterol [32]. On the other hand, lethality and teratogenicity were observed in DCF exposed zebra ﬁsh embryos (Danio rerio) after 96 h exposure to 480 ± 50 ␮g L−1 (LC50 /96 h) and 90 ± 20 ␮g L−1 (EC50 /96 h), respectively [33]. Further, cytological alterations in liver, kidney and gills even at 1 ␮g L−1 in rainbow trout have also been observed [34]. More speciﬁcally, DCF is responsible for higher population decline among vultures in Indian subcontinent, Pakistan, and Nepal [35]. However, the knowledge on the toxicity and effects of pharmaceutical drugs on the aquatic organisms are meager, particularly on freshwater ﬁsh. Consequently, we attempted to study the impacts of CA and DCF at different concentrations such as 1, 10 and 100 ␮g L−1 on hematological, biochemical, ionoregulatory, and enzymological parameters in a common carp, C. carpio. The carp is widely cultivated in major water bodies of India.

189

(one third of the water) daily and feeding was withheld 24 h before the commencement of the experiment. The tap water free from chlorine was used and the water had the following physicochemical characteristics [36]; temperature (27.0 ± 1.2 ◦ C), pH (7.2), dissolved oxygen (6.2 mg L−1 ), total hardness (89 mg L−1 , as CaCO3 ) and salinity (0.4 ± 0.02‰). Before the experiment, ﬁsh were randomly divided into two groups which were housed in 200 L aquaria with tap water and continuously aerated. Photoperiod of the study was a 12:12 light-dark cycle. 2.3. Experimental design and acute toxicity test A 96 h acute test was conducted in order to determine the shortterm impacts of CA and DCF. The nominal concentrations of CA and DCF including 1, 10 and 100 ␮g L−1 were added in each glass aquaria (120 cm × 80 cm × 40 cm) containing 60 L of water. Three replicates were maintained for each concentration groups and 30 ﬁsh of equal size and weight were introduced. The test water was renewed at the end of 24 h and freshly prepared solution was added to maintain the concentration of CA and DCF at a constant level. A concurrent control of 30 ﬁsh in three different glass aquaria was maintained under identical conditions. The mortality/survival of ﬁsh was recorded in every 24 h. The dead ﬁsh were removed from the aquaria immediately. Feeding was withheld during the bioassay experiment. At the end of 96 h period ﬁsh from the control and drug treated groups were taken for further analysis. 2.4. Blood sample collection Blood samples were collected by heart puncture using plastic disposable syringes ﬁtted with 26 gauge needles. The syringe and needle were prechilled and coated with heparin (Beparine R heparin sodium, IP 1000 IU mL−1 derived from beef intestinal mucosa containing 0.15% w/v chlorocresol IP preservative), an anticoagulant manufactured by Biological E Limited, Hyderabad, India. The collected blood was transferred into small vials, which is previously rinsed with heparin. Whole blood was used for the estimation of hemoglobin, RBC and WBC counts. The remainder of the blood sample was centrifuged at 9392 g, at 4 ◦ C for 20 min to separate the plasma, which was used for the estimation of biochemical parameters (glucose and protein), electrolytes (Na+ , K+ and Cl− ) and enzymes (GOT, GPT and LDH).

2. Materials and methods 2.5. Hematological studies 2.1. Chemicals Cloﬁbric acid (␣-(p-Chlorophenoxy) isobutyric acid, CAS No. 882-09-7) and diclofenac (2-[(2, 6-Dichlorophenyl) amino] benzene acetic acid sodium salt, CAS No. 15307-79-6) were purchased from Sigma–Aldrich Chemie GmbH, Germany. Dimethyl sulphoxide (CAS No. 67-68-5) was purchased from Fischer Scientiﬁc India Pvt. Ltd, India and 0.2 mL L−1 used to prepare the stock solution at different concentrations (1, 10, and 100 ␮g L−1 ) due to their low water solubility.

RBC and WBC were counted by haemocytometer method [37]. Hb content of the blood was estimated by the method of cyanmethemoglobin [38]. Hct was estimated by the microhematocrit method [39]. Erythrocyte indices of ﬁsh viz., MCV, MCH and MCHC were also calculated according to standard formulas. MCV(cubic micra) = MCH(picograms) =

2.2. Experimental ﬁsh and water Fingerlings of C. carpio were obtained from Tamil Nadu Fisheries Development Corporation Limited, Aliyar Fish Farm, Tamil Nadu, India in the weight range of 8.0 ± 0.4 g and body length of 7.0 ± 0.5 cm (mean ± SD). They were safely brought to the laboratory and acclimatized for 20 days in a large cement tank (containing 1000 L of water) prior to the experiment. During the acclimatization period, ﬁsh were fed ad libitum with rice bran and groundnut oil cake in the form of dough one time a day. Water was renewed

MCHC(g/dl) =

HCT(%) RBC (millions × cu × 106 ) Hb(g/dl) RBC (millions × cu × 106 )

Hb(g/dl) × 100 HCT(%)

× 100

× 100

(1)

(2)

(3)

2.6. Biochemical studies 2.6.1. Estimation of plasma glucose and protein Plasma glucose was estimated following the method of Cooper and Mc Daniel [40]. To 5.0 mL of O-toluidine colour reagent, 0.1 mL of plasma was added and the content was mixed well and placed in

190

M. Saravanan et al. / Journal of Hazardous Materials 195 (2011) 188–194

Table 1a Hematological proﬁles (Hb, Hct, RBC, and WBC) in a freshwater ﬁsh C. carpio after 96 h exposure to different concentrations of CA and DCF. Hb (g/dL)

Hct (%)

Control 1 10 100

4.627b 2.468d 3.667c 5.485a

13.72b 7.20d 10.56c 16.02a

0.332a 0.250c 0.244c 0.280b

16.780d 24.663c 30.065b 47.101a

Control 1 10 100

4.716a 4.364a 3.316b 2.793b

13.840a 12.664a 9.982b 7.380c

0.377a 0.338b 0.284c 0.243d

18.460d 25.874c 47.515a 43.205b

15.09**

13.18**

43.76**

784.31**

3.09 ns 38.06**

3.56 ns 36.45**

27.93** 15.95**

84.43** 106.73**

Drug

Concentrations (␮g L−1 )

CA

DCF

RBC (million/cu mm)

WBC (1000/cu mm)

F-statistics# Concentration (C, F3 ,32 ) Drug (D, F1,32 ) C × D (F3,32 )

# ns, Not signiﬁcant, **, Signiﬁcant at P < 0.01. Means in a column followed by common superscript for the drug are not signiﬁcantly different (P < 0.05) according to DMRT.

boiling water for 10 min. The content was then cooled under running tap water for 5 min and the optical density (OD) of the sample was measured at 630 nm within 30 min in a UV spectrophotometer. Plasma protein was estimated following the method of Lowry et al. [41]. To 0.90 mL of distilled water, 0.10 mL of plasma was added and treated with 5.0 mL of solution C [50 mL of solution A (2.00 gm of sodium carbonate was dissolved in 100.00 mL of 0.1 N NaOH), was dissolved with 1 mL of solution B (500.00 mg of copper sulphate was dissolved in 100.00 mL of 1% sodium potassium tartarate solution)]. The prepared content was allowed to stand at room temperature for 10 min, and then 0.5 mL of Folin-phenol was added. After 15 min, the colour intensity was read at 720 nm in a UV spectrophotometer.

2.7. Analysis of plasma electrolytes Plasma sodium (Na+ ) and potassium (K+ ) were estimated by the method of Trinder [42] and Sunderman [43] and chloride (Cl− ) was estimated by the modiﬁed method of Young et al. [44].

2.8. Enzymological analysis 2.8.1. Determination of GOT, GPT and LDH activity Plasma GOT and GPT activities were estimated by 2,4-DNPH method described by Reitmen and Franckel [45], and LDH activity was measured following the methodology described by Tietz [46].

2.8.2. Estimation of gill Na+ /K+ -ATPase activity The gills were isolated from the control and drug treated ﬁsh and 100 mg of each tissue was weighed and homogenized with Teﬂon homogenizer along with 1 mL of 0.1 M Tris–HCl buffer adjusted to pH 7.4). The homogenates were centrifuged at 93.9 g for 15 min at 4 ◦ C and the clear supernatant was used for the estimation of Na+ /K+ -ATPase activity following the method of Shiosaka et al. [47]. 2.9. Statistical analysis All values were expressed as means and analyzed by analysis of variance (ANOVA), followed by a DMRT (Duncan Multiple Range Test) test to determine the signiﬁcant differences (P < 0.01 and P < 0.05) among the concentrations, between the drugs, and the difference between the concentrations and drugs on each parameters. 3. Results 3.1. Hematological indices The hematological parameters viz., Hb, Hct, RBC, WBC, MCV, MCH and MCHC in C. carpio exposed to CA and DCF for 96 h exposure showed alterations when compared to control groups (Tables 1a and 1b). Hb and Hct contents were decreased in both CA and DCF treatments (except in 100 ␮g L−1 of CA treatment). RBC count was decreased in both treatments. However, WBC count was increased at all concentrations of CA and DCF treatments. Among the hematological indices, MCV and MCH values were increased at

Table 1b Hematological indices (MCV, MCH, and MCHC) in a freshwater ﬁsh C. carpio after 96 h exposure to different concentrations of CA and DCF. Drug

Concentrations (␮g L−1 )

MCV (cu m)

MCH (pg)

CA

Control 1 10 100

413.770b 288.920c 435.112b 573.389a

139.369b 99.074c 151.073b 196.327a

33.713a 34.290a 34.742a 34.198a

DCF

Control 1 10 100

367.330a 374.924a 352.888a 304.846a

125.144a 129.166a 117.032a 115.396a

293.400a 290.090a 280.400a 292.094a

F-statistics#

Concentration (C, F3 ,32 ) Drug (D, F1,32 ) C × D (F3,32 )

6.11** 18.96** 16.74**

9.52** 20.04** 17.33**

<1 ns 34105.59** <1 ns

# ns, Not signiﬁcant, **, Signiﬁcant at P < 0.01. Means in a column followed by common superscript for the drug are not signiﬁcantly different (P < 0.05) according to DMRT.

MCHC (g/dl)

M. Saravanan et al. / Journal of Hazardous Materials 195 (2011) 188–194

191

Table 2 Biochemical changes (plasma glucose and plasma protein) in a freshwater ﬁsh C. carpio after 96 h exposure to different concentrations of CA and DCF. Drug

Concentrations (␮g L−1 )

Glucose (mg 100 mL−1 )

Protein (␮g mL−1 )

c

CA

Control 1 10 100

74.028 111.866b 134.411a 132.401a

1.784b 2.426a 2.502a 2.472a

DCF

Control 1 10 100

88.82c 106.08b 122.77a 132.30a

1.851c 2.619b 2.769a 2.606b

F-statistics#

Concentration (C, F3 ,32 ) Drug (D, F1,32 ) C × D (F3,32 )

44.35** <1 ns 2.63 ns

229.16** 44.06** 2.92*

# ns, Not signiﬁcant, *, Signiﬁcant at P < 0.05; **, Signiﬁcant at P < 0.01. Means in a column followed by common superscript for the drug are not signiﬁcantly different (P < 0.05) according to DMRT.

10 and 100 ␮g L−1 of CA treatment, whereas these parameters were found to be decreased at 1 ␮g L−1 . In contrast, MCV and MCH values were increased at 1 ␮g L−1 of DCF treatments whereas the values were found to be decreased at 10 and 100 ␮g L−1 treatments. But the MCHC values in both drugs treated groups were found to be similar to control groups. A signiﬁcant (P < 0.01) change was found in all hematological parameters among concentrations (C), between drugs (D) and also between the concentrations and drugs (C × D) (Tables 1a and 1b). 3.2. Biochemical parameters Plasma glucose level was elevated at all concentrations of CA and DCF exposed ﬁsh when compared with controls (Table 2). A significant (P < 0.01) difference was observed among the concentrations of CA and DCF. No signiﬁcant difference in plasma glucose level was observed between the drugs and between the concentrations and drugs. Furthermore, plasma protein level was also increased at all concentrations of CA and DCF exposed ﬁsh comparatively to control groups (Table 2). A signiﬁcant (P < 0.01) relationship was found among the concentrations of both drugs and also between the drugs. The interaction between the concentrations and drugs on plasma protein level showed a signiﬁcant value at P < 0.05. 3.3. Plasma electrolytes During 96 h exposure period, plasma Na+ level was decreased at all concentrations of CA and DCF (Table 3). However, a maximum decreased level was observed in 100 ␮g L−1 of CA and DCF concentrations when compared to other concentrations. A signiﬁcant (P < 0.01) difference in plasma Na+ level was observed among the concentrations of both CA and DCF. The difference between

the drugs and also between the concentrations and drugs were noted at signiﬁcant (P < 0.05) level on plasma Na+ levels. Similarly, plasma K+ level was also decreased in all CA and DCF concentrations (Table 3). Besides, a maximum decrease in K+ level was noted in 100 ␮g L−1 concentrations of both CA and DCF when compared to other concentrations. There was no signiﬁcant difference in plasma K+ level among the concentrations. However, a signiﬁcant (P < 0.01) difference was noted between the drugs and also between the concentrations and drugs (P < 0.05). Plasma Cl− level was found to be increased at all concentrations of CA whereas in DCF concentrations plasma Cl− level was decreased when compared to control groups (Table 3). There was no signiﬁcant difference among the concentrations of CA and DCF. A signiﬁcant (P < 0.01) difference was found between the drugs and also between the concentrations and drugs. 3.4. Enzyme assay The GOT activity was decreased at all concentrations (1, 10, and 100 ␮g L−l ) of CA and DCF treated ﬁsh. A maximum decrease was noted in CA concentrations when compared to DCF (except 1 ␮g L−l ) concentrations (Table 4). A signiﬁcant (P < 0.01) difference was observed among the concentrations and also between concentrations and drugs. Between the drugs a signiﬁcant difference in GOT activity was found at P < 0.05 level. GPT activity was increased at all concentrations of both CA and DCF (except 1 ␮g L−l concentration of CA) (Table 4). A maximum increase in GPT activity was noted in 100 and 10 ␮g L−l concentrations of CA and DCF, respectively. LDH activity was also increased at all concentrations of CA and DCF treatments (Table 4). However, a maximum increase was noted in CA treatment when compared to DCF treatment throughout the study period (96 h). A signiﬁcant (P < 0.01) difference was observed in GPT and LDH activities among the concentrations, between the

Table 3 Ionoregulatory (Na+ , K+ and Cl− ) responses in a freshwater ﬁsh C. carpio after 96 h exposure to different concentrations of CA and DCF. Drug

Concentrations (␮g L−1 )

Na+ (mmol L−1 ) a

K+ (mmol L−1 ) a

Cl− (mmol L−1 )

CA

Control 1 10 100

142.57 118.29b 110.05b 116.34b

9.345 8.832a 9.280a 7.935a

125.76c 175.51a 144.75bc 164.09ab

DCF

Control 1 10 100

136.52a 128.13ab 124.64b 119.05b

8.985a 7.797b 7.887ab 6.756ab

132.41a 109.08ab 112.16ab 89.89b

F-statistics#

Concentration (C, F3 ,32 ) Drug (D, F1,32 ) C × D (F3,32 )

20.67** 5.07* 4.01*

2.05 ns 9.35** 3.90*

# ns, Not signiﬁcant; *, Signiﬁcant at P < 0.05; **, Signiﬁcant at P < 0.01. Means in a column followed by common superscript for the drug are not signiﬁcantly different (P < 0.05) according to DMRT.

1.27 ns 43.51** 8.55**

192

M. Saravanan et al. / Journal of Hazardous Materials 195 (2011) 188–194

Table 4 Enzymological (GOT, GPT, LDH and Na+ /K+ -ATPase) responses in a freshwater ﬁsh C. carpio after 96 h exposure to different concentrations of CA and DCF. Drug

Concentrations (␮g L−1 )

GOT (IU L−1 )

GPT (IU L−1 )

LDH (IU L−1 )

CA

Control 1 10 100

40.468a 14.052b 14.608b 14.007b

52.593c 46.035d 69.219b 74.912a

1775.8c 1855.3c 2737.9b 2980.3a

18.88b 24.32a 25.92a 25.44a

DCF

Control 1 10 100

32.604a 32.463a 11.950b 29.230a

54.623c 67.553b 73.436a 71.466ab

1314.7c 1686.9b 1709.6b 1985.1a

19.04b 22.08a 22.96a 24.11a

F-statistics#

Concentration (C, F3 ,32 ) Drug (D, F1,32 ) C × D (F3,32 )

18.22**

79.56**

102.72**

6.57* 8.30**

29.63** 23.31**

233.79** 21.60**

Na+ /K+ -ATPase (␮g/h/g)

15.33** 5.43* <1 ns

# ns, Not signiﬁcant, *, Signiﬁcant at P < 0.05; **, Signiﬁcant at P < 0.01. Means in a column followed by common superscript for the drug are not signiﬁcantly different (P < 0.05) according to DMRT.

drugs, and also between the concentrations and drugs. The Na+ /K+ ATPase activity in gill was increased in 1, 10, and 100 ␮g L−1 of CA and DCF after 96 h exposure, compared with the respective controls (Table 4). Statistically, a signiﬁcant (P < 0.01) difference was observed among the concentrations of both CA and DCF. Further, the differences induced between CA and DCF shows a signiﬁcant value at P < 0.05. The interaction between concentrations and drugs on gill Na+ /K+ -ATPase activity was not signiﬁcant. 4. Discussion The occurrence of pharmaceutical chemicals and its residues in aquatic environments pose a major problem in most of the countries. Consequently, environmental risk assessment of these emerging pollutants is needed to evaluate their impacts on non target organisms. In ecotoxicological studies, static bioassay tests have been widely used for evaluating the impacts of toxic chemicals on aquatic organisms [48]. The scientiﬁc reports of APHA [49] accepted bioassay studies as standard methods for assessing the toxicity of any new chemical compounds that enter into the aquatic ecosystems. In bioassay method, acute toxicity test is commonly used to evaluate the potential threat of several aquatic pollutants. 4.1. Hematological indices The decrease in Hct, Hb and RBC count levels (except in CA at higher concentration of 100 ␮g L−1 ) may be indicators of anemia [50] and reduction in RBC count caused either by the inhibition of erythropoiesis or by the destruction of red cells by the drugs CA and DCF. Similar decrease in RBC count, Hb and Hct values were also reported in carps exposed to toxicants [27,51]. Further, the elevated level of Hb content in the CA exposure at 100 ␮g L−1 might have resulted from replacement of oxidized denatured Hb and to supply more oxygen to tissues. In this study, the increase in Hct level at 100 ␮g L−1 of CA treated ﬁsh indicates impaired respiratory capacity of the ﬁsh due to damage in the gill caused by the drug. Most of the pharmaceutical compounds in the aquatic environment enter into ﬁsh body through gill and food. Swelling of RBCs due to stress may also contribute an increase in Hct level [29]. The alterations in MCV, MCH and MCHC levels with the lower and higher concentrations of CA and DCF might be due to stress response to the drugs. The increase in MCV may also result from the increase of immature RBC [52]. Primarily, WBCs are involved in the regulation of immunological function in many organisms and the changes in WBC number to pollutants reﬂect the decrease in the non speciﬁc immunity of the ﬁsh [28]. In the present study also the increase in WBC count in both CA and DCF treatments indicate a

generalized immune response to drug toxicity or the immune system may be stimulated by the drugs to protective against toxicity. An increase in MCV and MCH levels indicates the swelling of RBCs due to drug toxicity, whereas the decrease in these hematological parameters might have resulted from impaired oxygen uptake due to gill damage caused by the drugs. No signiﬁcant change in MCHC value was observed in both the treatments at all concentrations. The changes in hematological parameters of rainbow trout (Oncorhynchus mykiss) exposed to a lipophilic drug, verapamil indicates a compensatory responses to maintain the gas transfer [53]. A similar mechanism may be operated in the present study on hematological parameters of CA and DCF treated ﬁsh.

4.2. Biochemical proﬁles In general environmental contaminants in aquatic media induce signiﬁcant changes in biochemical parameters of aquatic organisms. We found a signiﬁcant increase in plasma glucose and protein level in ﬁsh exposed to various concentrations of CA and DCF. Basically, environmental stress alters carbohydrate metabolism in ﬁsh. An elevation of blood glucose level in the present study may indicate gluconeogenesis to compensate the increased metabolic demands imposed by the drugs [25]. Increased plasma protein level in both the treatments indicates an adaptation of the ﬁsh to the drug toxicity.

4.3. Ionoregulatory response Gills of freshwater ﬁsh play an important role in the transport of ions like Na+ , K+ and Cl− to maintain acid base balance, osmotic pressure of the body and regulation of water inﬂux and ion efﬂux [54]. In ﬁsh, gills due to their intimate contact with water likely to be the important target organ for aquatic pollutants. In this study, the reduced level of plasma Na+ and Cl− ions in CA and DCF treated ﬁsh might have resulted from histological alterations of gills or disturbances in the membrane permeability due to drug toxicity. Osmoregulatory failure may be also a reason for decreased levels of major plasma ions [55]. The decrease in plasma K+ level indicates the inhibition of the Na+ /K+ -ATPase due to drug toxicity (in this study the inhibition of Na+ /K+ -ATPase was noted). The decrease in plasma Cl− level in DCF treated ﬁsh indicates an apparent decrease of blood chloride concentrations in ﬁsh due to reduced activity of carbonic anhydrase or interference of cortisol [56]. The elevation in plasma Cl− level in CA treatment may be due to transportation of chloride ions from other tissues into blood due to imbalances in the osmoregulation process [57]. Fletcher [58] suggested that loss

M. Saravanan et al. / Journal of Hazardous Materials 195 (2011) 188–194

of water from the circulation may also leads to signiﬁcant rise in plasma electrolytes in Pseudopleuronectes americanus during stress.

193

support in the form Senior Research Fellowship for this study (Fellowship award letter no: 09/472(0141)/2009-EMR-I, dated: 01.05.2009).

4.4. Enzymological activity The enzymes GOT, GPT and LDH can be used for detection of tissue damage and bioindicators in animals subjected to acute and chronic exposure of xenobiotics [59]. The decrease in plasma GOT activity in both CA and DCF treatments indicates the accumulation and toxicity of these drugs in liver which might have caused the necrosis and death of liver cells. A similar observation was also made in C. punctatus exposed to As2 O3 [60]. The enzyme GPT is mainly present in the liver and any damage to liver leads to release of these enzymes into blood stream. Moreover, the damage and severity of the organ (liver) is mainly depends on the exposure period and the type of toxicant [61]. In our study, the elevations of GPT activity indicate that the organism tries to mitigate the drug induced stress. The increase of GOT and GPT activity in O. mykiss exposed to verapamil indicates ampliﬁed transamination processes [53]. LDH involved in carbohydrate metabolism, can be used as a good indicator to chemical exposure and stress in ﬁsh. Jose et al. [62] suggested that the increase in LDH activity in cultures exposed to carbamazepine may be due to stabilization of cytoplasmic membrane. Elevation of LDH activity in mosquito ﬁsh, Gambusia holbrooki after acute exposure to CA may be due to stress response caused by CA [10]. Similarly, in carbamazepine treated rainbow trout O. mykiss, plasma LDH activity was increased during the chronic exposure [63]. The observed elevation of LDH activity in CA and DCF treated ﬁsh indicate structural damage of the cell membranes or hepatic or heart tissues. In addition, changes in protein and carbohydrate metabolism may cause a change in LDH activity [64]. Gill Na+ /K+ -ATPase is intimately involved in electrolyte balance and the determination of ATPase activity would prove to be an important index for tolerable levels of a large group of environmental contaminants [65,66]. However, the impacts of pharmaceutical drugs on gill ATPase activity in freshwater ﬁsh have not been given enough attention. The signiﬁcant increase in gill Na+ /K+ -ATPase activity at all the concentrations of CA and DCF indicating the direct action of these drugs on ATPase function. Further, the increase in gill Na+ /K+ -ATPase activity may be a compensation for a dysfunctional regulation of ionic levels or a process to restore electrolyte levels [67,68]. On the whole, in our study a signiﬁcant alteration in ionic levels were noted in both CA and DCF treatments and therefore we warrant a further in depth study. 5. Conclusion Collectively this study concluded that different concentrations (1, 10, and 100 ␮g L−1 ) of CA and DCF have a profound inﬂuence on the hematological, biochemical, ionoregulatory and enzymological proﬁles of freshwater ﬁsh C. carpio. These parameters could be effectively used as potential biomarkers of pharmaceutical toxicity to freshwater ﬁsh in the ﬁeld of environmental biomonitoring. Furthermore, chronic effects of CA, DCF and their metabolites on these parameters along with other parameters such as hormonal and histopathological studies need to be investigated in the future studies. Acknowledgments The author (Manoharan Saravanan) would like to thank and acknowledge the Council of Scientiﬁc and Industrial Research (CSIR), New Delhi, Government of India for providing ﬁnancial

References [1] Y. Zhang, S.U. Geißen, C. Cal, Carbamazepine and diclofenac: removal in wastewater treatment plants and occurrence in water bodies, Chemosphere 73 (2008) 1151–1161. [2] V.L. Cunningham, S.P. Binks, M.J. Olson, Human health risk assessment from the presence of human pharmaceuticals in the aquatic environment, Regul. Toxicol. Pharmacol. 53 (2009) 39–45. [3] H. Sanderson, D.J. Johnson, T. Reitsma, R.A. Brain, C.J. Wilson, K.R. Solomon, Ranking and prioritization of environmental risks of pharmaceuticals in surface waters, Regul. Toxicol. Pharmacol. 39 (2004) 158–183. [4] R. Rosal, A. Rodríguez, J.A. Perdigón-Melón, M. Mezcua, A. Agüera, M.D. Hernando, P. Letón, E. García-Calvo, A.R. Fernández-Alba, Removal of pharmaceuticals and kinetics of mineralization by O3 /H2 O2 in a biotreated municipal wastewater, Water Res. 42 (2008) 3719–3728. [5] S. Kar, K. Roy, First report on interspecies quantitative correlation of ecotoxicity of pharmaceuticals, Chemosphere 81 (2010) 738–747. [6] Daughton, T.A. Ternes, Pharmaceuticals and personal care products in the environment: agents of subtle change? Environ. Health. Perspect. 107 (1999) 907–938. [7] M. Gros, M. Petrovic, D. Barcelo, Development of a multiresidue analytical methodology based on liquid chromatography-tandem mass spectrometry (LC-MS/MS) for screening and trace level determination of pharmaceuticals in surface and wastewaters, Talanta 70 (2006) 678–690. [8] M.J. Benotti, R.A. Trenholm, J.V. Brett, J.C. Holady, B.D. Stanford, S.A. Snyder, Pharmaceuticals and endocrine disrupting compounds in U.S. drinking water, Environ. Sci. Technol. 43 (2009) 597–603. [9] C. Zwiener, F.H. Frimmel, LC-MS analysis in the aquatic environment and in water treatment – a critical review. Part II: applications for emerging contaminants and related pollutants, microorganisms and humic acids, Anal. Bioanal. Chem. 378 (2004) 862–874. [10] F. Nunes, F. Carvalho, Guilhermino, Acute and chronic effects of cloﬁbrate and cloﬁbric acid on the enzymes acetylcholinesterase, lactate dehydrogenase and catalase of the mosquitoﬁsh, Gambusia holbrooki, Chemosphere 57 (2004) 1581–1589. [11] S. Weigel, J. Kuhlmann, H. Hühnerfuss, Drugs and personal care products as ubiquitous pollutants: occurrence and distribution of cloﬁbric acid, caffeine and DEET in the North Sea, Sci. Total Environ. 295 (2002) 131–141. [12] H.R. Buser, M.D. Müller, N. Theobald, Occurrence of the pharmaceutical drug cloﬁbric acid and the herbicide Mecoprop in various Swiss Lakes and in the North Sea, Environ. Sci. Technol. 32 (1998) 188–192. [13] T.A. Ternes, Occurrence of drugs in German sewage treatment plants and rivers, Water. Res. 32 (1998) 3245–3260. [14] C. Hignite, D.L. Azarnoff, Drugs and drug metabolites as environmental contaminants-chlorophenoxyisobutyrate and salicylic-acid in sewage water efﬂuent, Life Sci. 20 (1977) 337–341. [15] V. Koutsouba, T. Heberer, B. Fuhrmann, K. Schmidt-Baumler, D. Tsipi, A. Hiskia, Determination of polar pharmaceuticals in sewage water of Greece by gas chromatography-mass spectrometry, Chemosphere 51 (2003) 69–75. [16] T. Kosjek, E. Heath, S. Perez, M. Petrovic, D. Barcelo, Metabolism studies of diclofenac and cloﬁbric acid in activated sludge bioreactors using liquid chromatography with quadrupole-time-of-ﬂight mass spectrometry, J. Hydrol. 372 (2009) 109–117. [17] T. Heberer, K. Reddersen, A. Mechlinski, From municipal sewage to drinking water: fate and removal of pharmaceutical residues in the aquatic environment in urban areas, Water Sci. Technol. 46 (2002) 81–88. [18] M. Rabiet, A. Togola, F. Brissaud, J.L. Seidel, H. Budzinski, E.F. Poulichet, Consequences of treated water recycling as regards pharmaceuticals and drugs in surface and ground waters of a medium-sized Mediterranean catchment, Environ. Sci. Technol. 40 (2006) 5282–5288. [19] U. Jux, R.M. Baginski, H.G. Arnold, M. Krönke, P.N. Seng, Detection of pharmaceutical contaminations of river, pond, and tap water from Cologne (Germany) and surroundings, Int. J. Hyg. Environ. Health. 205 (2002) 393–398. [20] K.V. Thomas, M.J. Hilton, The occurrence of selected human pharmaceutical compounds in UK estuaries, Mar. Pollut. Bull. 49 (2004) 436–444. [21] M.R. Embry, E.B. Scott, A.B. Thomas, G.B. Malyka, H. Marlies, E.H. David, A.L. Marc, L. Adam, N.K. Teresa, W. Graham, The ﬁsh embryo toxicity test as an animal alternative method in hazard and risk assessment and scientiﬁc research, Aquat. Toxicol. 97 (2010) 79–87. [22] D.A. Powers, Fish as model system, Science 246 (1989) 352–358. [23] S. Han, K. Choi, J. Kim, K. Ji, S. Kima, B. Ahn, J. Yun, K. Choi, J.S. Khim, X. Zhang, J.P. Giesy, Endocrine disruption and consequences of chronic exposure to ibuprofen in Japanese medaka (Oryzias latipes) and freshwater cladocerans Daphnia magna and Moina macrocopa, Aquat. Toxicol. 98 (2010) 256–264. [24] M.P. Cajaraville, M.J. Bebianno, J. Blasco, C. Porte, C. Sarasquete, A. Viarengo, The use of biomarkers to assess the impact of pollution in coastal environments of the Iberian peninsula: a practical approach, Sci. Total Environ. 247 (2000) 295–311. [25] Z.H. Li, J. Velisek, V. Zlabek, R. Grabic, J. Machova, J. Kolarova, T. Randak, Hepatic antioxidant status and hematological parameters in rainbow trout,

194

[26]

[27]

[28]

[29] [30]

[31] [32]

[33]

[34]

[35]

[36] [37]

[38]

[39]

[40] [41] [42] [43] [44] [45]

[46] [47]

[48]

M. Saravanan et al. / Journal of Hazardous Materials 195 (2011) 188–194 Oncorhynchus mykiss, after chronic exposure to carbamazepine, Chem. Biol. Interact. 183 (2010) 98–104. D.G. McDonald, T.R.K. Dalxical, The combined effects of pH and trace metals on ﬁsh ion regulation, in: R. Morris, E. Toylor, D.J.A. Brown, J.A. Brrown (Eds.), Acid Toxicity and Aquatic Animals, Cambridge University Press, Cambridge, 1989, pp. 221–242. S. Lavanya, M. Ramesh, C. Kavitha, A. Malarvizhi, Hematological, biochemical and ionoregulatory responses of Indian major carp Catla catla during chronic sublethal exposure to inorganic arsenic, Chemosphere 82 (2011) 977–985. C. Kavitha, A. Malarvizhi, S. Senthil Kumaran, M. Ramesh, Toxicological effects of arsenate exposure on hematological, biochemical and liver transaminases activity in an Indian major carp, Catla catla, Food Chem. Toxicol. 48 (2010) 2848–2854. J. Nemcsok, L. Boross, Comparative studies on the sensitivity of different ﬁsh species to metal pollution, Acta Biol. Hung. 33 (1982) 23–27. T.C. Diamantino, E. Almeida, A.M.V.M. Soares, L. Guilhermino, Lactate dehydrogenase activity as an effect criterion in toxicity tests with Daphnia magna straus, Chemosphere 45 (2001) 553–560. T.E. Doll, F.H. Frimmel, Fate of pharmaceuticals-photodegradation by simulated solar UV-light, Chemosphere 52 (2003) 1757–1769. P. Pﬂuger, D.R. Dietrich, Effects on pharmaceuticals in the environment-an overview and principle considerations, in: K. Kummerer (Ed.), Pharmaceuticals in the Environment-Sources, Fate, Effects and Risks, Springer, Berlin, 2001, pp. 11–17. D.R. Dietrich, A. Prietz, Fish embryotoxicity and teratogenicity of pharmaceuticals, detergents and pesticides regularly detected in sewage treatment plant efﬂuents and surface waters, Toxicologist 48 (1999) 151. R. Triebskorn, H. Casper, A. Heyd, R. Eikemper, H.R. Kohler, J. Schwaiger, Toxic effects of the non-steroidal anti-inﬂammatory drug diclofenac, Part II. Cytological effects in liver, kidney, gills and intestine of rainbow trout, Oncorhynchus mykiss, Aquat. Toxicol. 68 (2004) 151–166. J.L. Oaks, M. Gilbert, M.Z. Virani, R.T. Watson, C.U. Meteyer, B.A. Rideout, H.L. Shivaprasad, S. Ahmed, M.J.I. Chauddry, M. Arshad, S. Mahmood, A. Ali, A.A. Khan, Diclofenac residues as the cause of population decline of vultures in Pakistan, Nature 427 (2004) 630–633. APHA, Standard Methods for the Examination of Water and Wastewater, twentieth ed., American Public Health Association, Washington DC, 1998. V. Rusia, S.K. Sood, Routine hematological tests, in: L. Mukerjee Kanai (Ed.), Medical Laboratory Technology, vol. 1, Tata McGraw Hill Publishing Company Limited, New Delhi, 1992, pp. 252–258 (ﬁfth reprint). D.L. Drabkin, Spectrometric studies, XIV-the crystallographic and optimal properties of the hemoglobin of man in comparison with those of other species, Biol. Chem. 164 (1946) 703–723. D.A. Nelson, M.W. Morris, Basic methodology, Hematology and coagulation, part IV, in: D.A. Nelson, J.B. Henry (Eds.), Clinical Diagnosis and Management by Laboratory Methods, seventeenth ed., Saunder Company, Philadelphia, USA, 1989, pp. 578–625 (chapter 27). G.R. Cooper, V. McDaniel, The determination of the glucose by the orthotoluidine method, Standard Methods Clin. Chem. 6 (1970) 159–170. O.H. Lowry, N.J. Rosebrough, A.L. Farr, R.I. Randall, Protein measurement with folin phenol reagent, J. Biol. Chem. 193 (1951) 265–275. P. Trinder, A rapid method for the determination of sodium in serum, Analyst 76 (1951) 596–599. F.W. Sunderman, Am. J. Clin. Pathol. 29 (1958) 95. D.S. Young, L.C. Pestaner, V. Gibberman, Effects of drugs on clinical laboratory tests, Clin. Chem. 21 (1975) 1D–432D. S. Reitmen, S. Frankel, A colorimetric method for the determination of serum glutamic oxalo acetic and glutamic pyruvic transaminase, Am. J. Clin. Pathol. 2 (1957) 8–56. N.W. Tietz, Fundamentals of Clinical Chemistry, W.B. Saunders Co., Philadelphia, PA, 1976, p. 652. T. Shiosaka, H. Okuda, S. Fungi, Machanisms of phophorylation of thymidine by the culture ﬁltrate of Clostridium perfringens and rat liver extract, Biochem. Biophys. Acta. 246 (1971) 171–183. P. Doudoroff, B.G. Anderson, G.E. Burdick, P.S. Galtsoff, W.B. Hart, R. Patrick, E.R. Strong, E.W. Surber, W.M. van Horn, Bio-assay methods for the evaluation of acute toxicity of industrial wastes to ﬁshes, Sewage Ind. Wastes. 23 (1951) 1380–1397.

[49] APHA, in: Standard Methods for the Examination of Water and Waste Water, fourteenth ed., Am. Publ. Hlth. Ass, New York, 1974. [50] N. Thakur, S. Sahai, Carbaryl induced haematological alterations in a teleost Garra gotyla (Garay), in: R.C. Dalela (Ed.), Environment and Ecology, Proceedings of the Academy of Environmental Biology, Muzaffarnagar, 1987, pp. 339–344. [51] F. Jenkins, J. Smith, B. Rajanna, U. Shameem, K. Umadevi, V. Sandhya, R. Madhavi, Effect of sublethal concentrations of endosulfan on haematological and serum biochemical parameters in the carp Cyprinus carpio, Bull. Environ. Contam. Toxicol. 70 (2003) 993–997. [52] C.S. Caravalho, M.N. Fernandes, Effect of temperature on copper toxicity and hematological responses in the neotrophical ﬁsh, Prochilodus scrofa at low and high pH, Aquaculture 251 (2006) 109–117. [53] Z.H. Li, J. Velisek, V. Zlabek, R. Grabic, J. Machova, J. Kolarova, P. Li, T. Randak, Chronic toxicity of verapamil on juvenile rainbow trout (Oncorhynchus mykiss): effects on morphological indices, hematological parameters and antioxidant responses, J. Hazard. Mater. 185 (2011) 870–880. [54] F.L. Mayer, D.J. Versteeg, M.J. McKee, L.C. Folmar, R.L. Graney, D.C. McCume, B.A. Rattner, Physiological and nonspeciﬁc biomarkers, in: R.J. Huggett, R.A. Kimerle, P.M. Mehrle Jr., H.L. Bergman (Eds.), Biomarkers, Biochemical, Physiological and Histological Markers of Anthropogenic Stress, Proceedings of the Eighth Pellston Workshop, Keystone, Colorado, Lewis Publishers, Boca Raton, USA, 1989, 1992, pp. 5–85. [55] A. Larsson, B.E. Bengtsson, O. Svanberg, Some haematological and biochemical effects of cadmium on ﬁsh, in: A.P.M. Lockwood (Ed.), Effects of Pollutants on Aquatic Organisms, Cambridge University Press, Cambridge, UK, 1976, pp. 35–45. [56] P.C. Thomas, T.C. Murthy, Studies on the impact of a ferric organic pesticide on certain ﬁsh enzymes, Indian J. Anim. Sci. 46 (1976) 619–624. [57] L. Suvetha, M. Ramesh, M. Saravanan, Inﬂuence of cypermethrin toxicity on ionic regulation and gill Na+ /K+ -ATPase activity of a freshwater teleost ﬁsh. Cyprinus carpio, Environ. Toxicol. Pharmacol. 29 (2010) 44–49. [58] G.L. Fletcher, The effects of capture stress and storage of whole blood on the red blood cells, plasma proteins, glucose and electrolytes of the winter ﬂounder, Pseudopleuronectes americanus, Can. J. Zool. 53 (1975) 197–206. [59] J. Nemcsok, I. Benedeczky, Effect of sublethal concentrations of phenol on some enzyme activities and blood sugar level of carp Cyprinus carpio (L.), Environ. Monit. Assess. 14 (1990) 377–383. [60] S. Roy, S. Bhattacharya, Arsenic-induced histopathology and synthesis of stress proteins in liver and kidney of Channa punctatus, Ecotoxicol. Environ. Saf. 65 (2005) 218–229. [61] D. Jacobson-Kram, K.A. Keller, Toxicology Testing Handbook: Principles, Applications and Data interpretation, Marcel Decker, New York, 2001. [62] A. Jos, G. Repetto, J.C. Riosc, M.J. Hazend, M.L. Molerod, A. Del Pesoa, M. Salgueroa, P. Fernandez-Freired, J.M. Perez-Martınd, A. Cameanb, Ecotoxicological evaluation of carbamazepine using six different model systems with eighteen endpoints, Toxicol. In Vitro 17 (2003) 525–532. [63] Z. Li, V. Zlabek, J. Velisek, R. Grabic, J. Machova, T. Randak, Responses of antioxidant status and Na+ /K+ -ATPase activity in gill of rainbow trout, Oncorhynchus mykiss, chronically treated with carbamazepine, Chemosphere 77 (2009) 1476–1481. [64] T. Szegletes, C.S. Polyhos, T. Balint, A.A. Reddy, G. Lang, O. Kufesak, J. Nemcsok, In vivo effects of deltamethrin on some biochemical parameters of carp, Cyprinus carpio (L.), Environ. Monit. Assess. 35 (1995) 97–111. [65] R. Mathan, S.K. Kurunthachalam, M. Priya, Alterations in plasma electrolyte levels of a freshwater ﬁsh Cyprinus carpio exposed to acidic pH, Toxicol. Environ. Chem. 92 (2010) 149–157. [66] E.O. Oruc, N. Unea, L. Tamer, Comparison of Na+ /K+ ATPase activities and malondialdehyde contents in livers tissue for three ﬁsh species exposed to azinphosmethyl, Bull. Environ. Contam. Toxicol. 69 (2002) 271– 277. [67] C. Hogstrand, E.A. Ferguson, F. Galvez, J.R. Shaw, N.A. Webb, C.M. Wood, Physiology of acute silver toxicity in the starry ﬂounder Platichthys stellatus in seawater, J. Comp. Physiol. B 169 (1999) 461–473. [68] N.J. Bouskill, R.D. Handy, T.E. Ford, T.S. Galloway, Differentiating copper and arsenic toxicity using biochemical biomarkers in Asellus aquaticus and Dreissena polymorpha, Ecotoxicol. Environ. Saf. 65 (2006) 342–349.

Journal of Hazardous Materials 195 (2011) 195–200

Contents lists available at ScienceDirect

Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat

Toluene and chlorobenzene dinitration over solid H3 PO4 /MoO3 /SiO2 catalyst ´ Joanna Adamiak ∗ , Dorota Kalinowska-Alichnewicz, Michał Szadkowski, Wincenty Skupinski Warsaw University of Technology, Faculty of Chemistry, Division of High Energetic Materials, Noakowskiego 3, 00-664 Warsaw, Poland

a r t i c l e

i n f o

Article history: Received 20 May 2011 Received in revised form 12 July 2011 Accepted 8 August 2011 Available online 16 August 2011 Keywords: Nitration Continuous process Dinitrotoluene Dinitrochlorobenzene Solid catalyst

a b s t r a c t A new catalyst, H3 PO4 /MoO3 /SiO2 , was prepared by modiﬁcation of MoO3 /SiO2 using phosphoric acid. The characterization of the catalyst was performed using Infrared and Raman Spectroscopy, potentiometric titration and nitrogen adsorption–desorption methods. Molybdenum oxides were identiﬁed along with phosphomolybdic acid and polymolybdates on the modiﬁed surface. The suitability of the catalysts for toluene and chlorobenzene nitration in continuous process was examined. Toluene is effectively nitrated to dinitrotoluene (DNT) in one-stage process (96 wt.% of DNT in the product) and in mild conditions i.e. at room temperature and only with ten-fold excess of nitric acid. In chlorobenzene nitration only twelvefold excess of nitric acid is needed to obtain as high yield as 95 wt.%. Most importantly, the novel catalysts we have developed, provide the opportunity for sulfuric acid- free nitration of aromatic compounds. © 2011 Elsevier B.V. All rights reserved.

1. Introduction Nitro aromatic compounds are produced in great amounts for civilian and military uses [1–4]. One of the most common processes in the chemical industry are dinitrotoluene (DNT) and dinitrochlorobenzene (DNClB) synthesis. DNT is used in manufacture of polyurethanes (used as e.g. glues, lacquers), which are produced in amount of ca. 10 million tons per year [5]. Furthermore, DNT is an important compound used in the explosive trade as constituent of gun powders, propellants and explosive mixtures [4]. On the other hand, DNClB is one of the compound that is commonly used in synthesis of explosives [6] and is a starting material in production of 2,4,6trinitrophenylmethylnitramine (Tetryl), hexanitrodiphenylamin (HNDP), 2,4,6-trinitrophenol (Picric acid, TNP), trinitroanisole, etc. Furthermore, it is used in manufacture of dyes, e.g. Sulphur Black. Basically, nitro aromatic compounds synthesis is not an ecofriendly process. DNT and DNClB are obtained by nitration in the presence of nitric and sulfuric acid mixture [1–6]. Sulfuric acid is relatively effective catalyst in nitration because the reaction yields are high. However, the catalyst cannot be reused and its regeneration process is expensive and onerous. There can be also difﬁculties in the separation of the products from the reaction mixture, because they are dissolved in the mixture of acids. Furthermore, the nitration mixture is corrosive and dangerous in handling which additionally increases the plant and operating costs. After

∗ Corresponding author. Tel.: +48 22 2347991; fax: +48 22 2347991. E-mail address: [email protected] (J. Adamiak). 0304-3894/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.jhazmat.2011.08.031

nitration there is a great amount of waste in the form of diluted acids and by-products (effect of oxidation or sulfonation). All these disadvantages show that applying the mixture of nitric and sulfuric acid is neither ecological nor economical, despite of relatively low price of sulfuric acid. A new class of catalysts used in nitration comprises the solid acid catalysts, which have on their surface strong acidic centers. The use of these catalysts in nitration reactions can be advantageous due to the ease of product and catalyst separation from the reaction mixture. They may be characterized by high activity and selectivity in the reactions and their application very often reduces quantities of hazardous waste e.g. diluted sulfuric acid [7,8]. According to the recent literature data, acid strength of the catalysts may be increased by their modiﬁcation by phosphoric acid. An example is TN nitration over zeolite ZSM-5 modiﬁed by phosphoric acid [9,10] whose activity was shown to be higher than activity of unmodiﬁed zeolite. Moreover, in nitration with this system, high selectivity to isomer para-nitrotoluene was obtained. Solid acid catalysts prepared by mixing titania and phosphoric acid in the molar ratio of 1:1 and heating at 200–220 ◦ C on a sand bath were also used in nitration [11]. The catalysts were shown to be effective in nitration of various aromatic compounds (e.g. naphthalene, anthracene, phenanthrene) that may be further used as substrates in ﬁne chemicals synthesis. Furthermore, solid oxide catalysts modiﬁed by phosphoric acid or its salts are described in literature [12]. The phosphate-modiﬁed TiO2 –SiO2 catalyst was found to be an efﬁcient and selective catalyst for solvent-free mononitration of TN. Other method of catalyst modiﬁcation by phosphoric ions involves direct synthesis of active domains on the carrier surface.

196

J. Adamiak et al. / Journal of Hazardous Materials 195 (2011) 195–200

As an example, benzene nitration in the presence of a catalyst prepared by heating ammonium molybdate with ammonium orthophosphate can be presented [13]. The catalytic system consisted of ammonium salt of 12-phosphomolybdic acid (HPM, very strong Brönsted acid [14,15]) was effective in mild conditions and the nitration yield and selectivity to nitrobenzene were satisfactory. Our research is basically devoted to the solid catalyst systems consisted of molybdenum oxide (VI) supported on SiO2 carrier that are obtained by ammonium molybdate thermal decomposition [16,17]. From our experience, we know that the highest activity in toluene (TN) nitration (i.e. the yield over 90 wt.%) were observed for catalyst containing 15 wt.% of MoO3 (MoO3 /SiO2 ). In order to improve the catalyst activity (according to described data) we modiﬁed the MoO3 /SiO2 system by phosphoric acid and investigated its application to nitration of toluene and chlorobenzene (ClB). It was determined that H3 PO4 /MoO3 /SiO2 catalyst, in which 1 mole of H3 PO4 is used on 3 moles of MoO3 , was the most effective in nitration in batch process since substantial amount of DNT (i.e. 15 wt.%) was obtained in TN mononitration (molar substrate ratio TN:HNO3 is 1:1.5) in mild conditions. In this work, we continue our studies on nitration with the H3 PO4 /MoO3 /SiO2 catalyst. The results of TN and ClB nitration under continuous conditions are reported and their impact on making the process more ecological and economical in comparison to the batch process is highlighted.

2. Experimental 2.1. General Materials taken from POCH Gliwice: toluene (pure), chlorobenzene (pure), fuming nitric acid (pure), ammonium molybdate tetrahydrate [(NH4 )6 Mo7 O24 ·4H2 O] (pure), phosphoric acid (85%, pure), sodium hydrogen carbonate (NaHCO3 , pure), 1,2dichloroethane (pure), hydrochloric acid (pure). Silica gel (SiO2 ) ˛ was purchased from Matwy, Poland, 12-phosphomolybdic acid (pure) from Sigma Aldrich and n-butylamine (pure) from Fluka Analytical. Analysis of the reaction products was made using gas chromatography with a PerkinElmer Auto System XL and a Rxi-5Sil MS (30 m × 0.53 mm × 1.5 ␮m) column. The sample of post-reaction mixture was dissolved in 1,2-dichloroethane. The quantitative composition of prepared substances were calculated by internal standard method using peak areas and chlorobenzene as internal standard (IS). Raman spectra were recorded on a Nicolet Almega Dispersive Raman Spectrometer. Infrared spectrum was recorded as KBr pellets on Nicolet 6700 interferometer (4000–400 cm−1 , resolution 4 cm−1 ). Textural properties of the catalysts surface were determined using a speciﬁc surface area analyzer and the microporosity Micromeritics ASAP 2020M. Adsorption/desorption experiments were made for dinitrogen within the p/p0 range of 0.02–1.0 at −196 ◦ C. Speciﬁc surface area was determined on the basis of the BET method for p/p0 range of 0.08–0.30 [18]. The acidity of the catalyst was determined by measurements of initial electrode potential and potentiometric titration method [13,19,20]. Certain 1 g of catalyst was suspended in acetonitrile and the system was magnetically stirred for 1 h. The suspension was titrated with a solution of 0.05 N n-butylamine in acetonitrile. Then the variation in the electrode potential was measured with pH meter, using a standard glass electrode. The total number of acidic sites were determined by the amine titration method. The catalyst (500 mg) was suspended in solution of 0.1 N n-butylamine in dry toluene and the suspension was left for 24 h. Then 5 mL of solution

Fig. 1. Raman spectra of main domains present on the H3 PO4 /MoO3 /SiO2 surface: a. molybdenum oxide, b. polymolybdates c. phosphomolybdic acid (HPM).

was taken from above of the solid. To this solution 5 mL of distilled water was added and the mixture was titrated with 0.05 N HCl to phenolphthalein. 2.2. Catalysts preparation To obtain 5 g of the MoO3 /SiO2 catalyst, 0.9 g (0.75 mmol) of ammonium molybdate tetrahydrate was dissolved in 7 mL 3% H2 O2 . The solution of the salt was then applied by wet impregnation method on 4.25 g of SiO2 (grains 0.6–1 mm). First, the catalyst was dried at 110 ◦ C. Next it was heated at 300 ◦ C for about 16 h. As a result, 0.75 g (5.2 mmol) of MoO3 (the oxide constituted 15 wt.% of the catalyst mass) was obtained. Solution of phosphoric acid (1 mol of acid on 3 mol of MoO3 ) was applied by wet impregnation onto the catalyst made in the ﬁrst step. Re-heating was conducted at 300 ◦ C for about 16 h. To obtain 5 g of HPM/SiO2 catalyst 0.8 g of HPM was dissolved in 7 mL distilled water. Then the solution was applied on 4.2 g of SiO2 , dried at 110 ◦ C and heated at 300 ◦ C for 1 h. 2.3. Nitration process In a ﬂow reactor provided with a cooling jacket the catalysts was placed. Fuming nitric acid and substrate (toluene or chlorobenzene) was fed onto the catalyst bed and the substrate ﬂow was selected to obtain a HNO3 :toluene molar ratio required. The mixture ﬂow rate was set at ca. 10 mL/h. The post-reaction mixture was washed with aqueous NaHCO3 and then with water. In toluene nitration the post-reaction mixture contained mononitrotoluenes (MNT): ortho-nitrotoluene (o-NT), meta-nitrotoluene (m-NT), paranitrotoluene (p-NT) and dinitrotoluenes (DNT): 2,4-DNT, 2,6-DNT and 3,4-DNT. In chlorobenzene nitration the post-reaction mixture contained mononitrochlorobenzene (MNClB) and dinitrochlorobenzene (DNClB). 3. Results and discussion 3.1. Catalyst characterization In order to determine domains obtained on the surface Raman spectra were recorded. Representative spectra of main domains present on the H3 PO4 /MoO3 /SiO2 are presented in Fig. 1. In Fig. 1a the observed bands at 994, 818, 665, 470, 373, 332, 282 and 236 cm−1 are characteristic for the crystalline orthorombic molybdenum oxide (␣-MoO3 ) [19–22]. During the impregnation by phosphoric acid, a local excess of molybdate concentrations may occur and lead to formation of these crystallites.

J. Adamiak et al. / Journal of Hazardous Materials 195 (2011) 195–200

197

Fig. 3. Potentiometric titration curves of SiO2 H3 PO4 /MoO3 /SiO2 catalysts.

Fig. 2. IR spectrum of H3 PO4 /MoO3 /SiO2 .

In the spectrum in Fig. 1b broad bands at 800–900 and 900–1000 cm−1 are observed. These bands could be attributed to polymeric structures of molybdates with Mo–O groups [23–26]. The polymeric layer is known to be built from MoO6 octahedra. A broad, weak band at 200 cm−1 could be assigned to a deformation mode of Mo–O–Mo groups and indicates the oxygen bridges between molybdenum atoms in polymolybdates. In addition, the characteristic bands for Keggin structure of phosphomolybdic acid (HPM) are observed in Fig. 1c. The peaks at 990 and 970 cm−1 are assigned to the stretching mode ( Mo–Od ) of the terminal Mo O groups in the molecule of HPM [27,28]. The sharp band at 250 cm−1 can be assigned to the stretching mode ( Mo–Oa ) between the molybdenum atom and the oxygen atom that links central tetrahedron of each PO4 to Mo–O octahedron. It is clear that domains of HPM are formed in the reaction of phosphoric acid and molybdate ions, however, all molybdenum compounds do not react with added phosphoric acid (Fig. 1b). Fig. 2 shows the IR spectrum for the H3 PO4 /MoO3 /SiO2 catalyst. Bands at ca. 1100, 800 and 470 cm−1 can be assigned to SiO2 [28]. The typical pattern of HPM is therefore partially obscured by the SiO2 bands. However, a Keggin unit may still be characterized by the as Mo–Ot (t-terminal O atom in Keggin unit) band at about 961 cm−1 [28]. This band may conﬁrm that domains of HPM are indeed present on the surface. Potentiometric titration technique enables the determination of the total number of acidic sites and their strength. The maximum acidic strength of the surface sites may be described by initial electrode potential (Ei ) [15,19,20]. Ei amounts 230 mV for SiO2 , 561 mV for MoO3 /SiO2 and increases for catalysts modiﬁed by H3 PO4 to 646 mV. Therefore, the acidic strength of the surface sites increases after addition of phosphoric acid. From the plot shown in Fig. 3, it can be found that titration process of H3 PO4 /MoO3 /SiO2 requires higher amounts of n-butylamine than it is needed for MoO3 /SiO2 . The total number of acidic sites were determined by n-butylamine titration and equals 1.0 mmol g−1 for MoO3 /SiO2 and 1.2 mmol g−1 for H3 PO4 /MoO3 /SiO2 . Therefore, it may be concluded that the amount of H3 PO4 /MoO3 /SiO2 surface acidic centers is higher than amount present on MoO3 /SiO2 and SiO2 . The results of surface analysis of catalytic systems made by nitrogen desorption–adsorption methods are presented in Table 1. The formation of molybdenum oxide on SiO2 causes surface area to be reduced by about 30 m2 g−1 with respect to the carrier (BET isotherm). Phosphoric acid impregnation on the MoO3 /SiO2 brings about a slightly decrease in the surface size compared to

(carrier), MoO3 /SiO2

and

MoO3 /SiO2 . Probably domains of HPA and polymolybdates are responsible for this surface area reduction. 3.2. Toluene nitration Firstly, toluene nitration in continuous process was carried out with a little excess of nitric acid at 60 ◦ C. The results are presented in Table 2. The H3 PO4 /MoO3 /SiO2 catalyst is active in nitration from the beginning. TN conversion is high and amounts ca. 96 wt.%. The utilized amount of nitric acid is not enough to obtain a complete toluene conversion to DNT but is sufﬁcient to obtain 25 wt.% of DNT in the post-reaction mixture. When the reaction is continued, the yield decreases to 70 wt.% and basically stabilizes at this level. It is probably caused by surface active centers blocking by the products that are not totally washed off from the catalyst. This amount of nitric acid is not enough to wash off all the products from catalyst deposit and in the result the products are accumulated. Since reaction yield decreases as a result of centers blocking and the amount of DNT is stable (20 wt.%), it may be concluded that accumulated mononitrotoluenes are nitrated to DNT and then they are washed off. Toluene nitration was carried out with eight- and ten-fold excess of nitric acid (substrate molar ratios 8:1 and 10:1). A comparison of nitration results are presented in Fig. 4. In both cases almost complete toluene conversion is achieved in mild conditions (room temperature, 5 g of catalyst). Almost 100 wt.% of DNT is obtained with substrate ratio 10:1 and this amount is stable at least for 15 h. In the reaction carried out with eight-fold excess of HNO3 an increase of DNT amount is observed along with process time. Therefore, such an amount of nitric acid enables to obtain DNT but is not sufﬁcient to wash off the product from the catalyst deposit. It is the same situation as described above nitration with threefold excess of nitric acid where the main product was accumulated. Only when next portions of acid are provided, compounds adsorbed on the grains are washed off. However, as it is seen in Fig. 4, tenfold excess of nitric acid is sufﬁcient to prevent permanent product

Table 1 Textural parameters of SiO2 , MoO3 /SiO2 and H3 PO4 /MoO3 /SiO2 catalysts. Catalyst

SiO2 MoO3 /SiO2 H3 PO4 /MoO3 /SiO2

Surface area (m2 g−1 )

Micropore area (m2 g−1 )

BET

Langmuir

t-Plot

279 249 245

453 405 400

27 17 15

198

J. Adamiak et al. / Journal of Hazardous Materials 195 (2011) 195–200

Table 2 Toluene nitration in continuous process with substrate molar ratio HNO3 :TN = 3:1 at 60 ◦ C (5 g of catalyst). Time (h)

0 10 15

TN Conversion (wt.%)

95.9 70.4 76.9

Product composition (wt.%)

DNT composition (wt.%)

2.4/2.6 ratio

MNT

DNT

2,4-DNT

2,6-DNT

3,4-DNT

75 83.3 79.4

25 16.7 20.6

77.2 75.9 77.3

18 17 17.5

4.8 7.1 5.2

Fig. 4. Toluene nitration in continuous process with HNO3 :TN substrate ratio 10:1 and 8:1 (5 g of catalyst, room temperature).

accumulation and DNT is obtained in post-reaction mixture from the beginning. According to dinitration mechanism [1,2] o-NT is nitrated to 2,4-DNT easier than other isomers. Steric barierrs that occurs in o-NT molecule (spatial methyl group next to nitric group) cause lower amount of 2,6-DNT to be formed from this isomer. 2,4-DNT is formed also in p-NT nitration, however, it is also hindered on account of steric barriers. Since 2,6-DNT is not formed from p-NT there should be the highest amount of 2,4-DNT in the reaction product. Indeed, in nitration products (Table 3) there is ca. 80 wt.% of 2,4-DNT among DNT and o-NT is in the minority among MNT. This product composition is maintained for all reaction times. In order to compare the effectiveness of H3 PO4 /MoO3 /SiO2 and HPM, which is one of the strongest inorganic acid, the continuous process of toluene nitration over HPM supported on SiO2 (HPM/SiO2 ) was carried out (Table 3). In the latter toluene conversion equals to ca. 100 wt.%, however the amount of DNT is relatively low in comparison to the reaction with H3 PO4 /MoO3 /SiO2 . Moreover, this amount decreases along with the process time, which indicates that HPM/SiO2 loses its activity. To sum up, the H3 PO4 /MoO3 /SiO2 system has similar activity in

4.29 4.46 4.42

MNT composition (wt.%)

o-NT

m-NT

p-NT

52.1 53.3 53.1

3.6 3.7 3.5

44.3 43.3 43.3

Fig. 5. Toluene nitration in continuous process with substrate molar ratio HNO3 :TN = 8:1 at 60 ◦ C (5 g of catalyst).

toluene nitration to HPM/SiO2 , but is more efﬁcient and stable for longer time. Toluene nitration to DNT occurs only in the presence of nitronium cation (NO2 + ) [1,2]. In the nitration with HNO3 and H3 PO4 /MoO3 /SiO2 catalytic system in continuous process there are two factors that facilitate NO2 + formation i.e.: the excess of nitric acid and the presence of active acidic species on the catalyst surface. Nitronium cation generation in the presence of the excess of nitric acid occurs by acid autoprotonation: HNO3 + HNO3 ↔ H2 ONO2 + + NO3 − ↔ H2 O + NO2 + + NO3 −

(1)

However, in the presence of the catalyst NO2 + can be also generated in nitric acid protonation by strong protonic acid: HNO3 + H+ ·H2 NO3 + ·NO2 + + H2 O

(2)

As shown in Section 1, on the surface of the examined catalyst there are domains of HPM, which is a very strong Brönsted acid (pK1 = 2.0), stronger than nitric acid (pK1 = 9.4) and sulfuric acid (pK1 = 6.6) [29]. Therefore HPM is able to protonate nitric acid according to the reaction (2) and it can generate nitronium cation very effectively.

Table 3 The composition of nitration product in the reaction with substrate molar ratio HNO3 :TN = 10:1 (5 g of catalyst, room temperature). Time (h)

Amount of DNT (wt.%)

H3 PO4 /MoO3 /SiO2 5 97.5 91.6 10 96 15 HPM/SiO2 5 81.2 10 74.3 57.7 15

DNT Composition (wt.%)

2.4-/2.6 ratio

2,4-DNT

2,6-DNT

3,4-DNT

80 80.5 81.5

18 18 16.9

2 1.5 1.4

4.44 4.47 4.82

79.9 80 80.5

18.3 18.2 18.1

1.8 1.8 1.4

4.37 4.39 4.43

Amount of MNT (wt.%)

MNT composition (wt.%) o-NT

m-NT

p-NT

2.5 8.4 4

24.4 37 7.9

6 3.6 3

69.6 59.4 89.1

18.8 25.7 42.3

44.2 48 50.4

4.3 3.8 3.1

51.5 48.2 46.5

J. Adamiak et al. / Journal of Hazardous Materials 195 (2011) 195–200

199

Table 4 The comparison of reaction yield in batch and continuous process (5 g of catalyst, 2.5 h, room temperature, substrate molar ratio HNO3 :TN = 10:1). Type of reaction

Batch Continuous

Amount of TN (mmol)

4 47.3

TN conversion (wt.%)

100 93.7

Product composition (wt.%) DNT

MNT

34.6 97.5

65.4 2.5

Furthermore, molybdenum oxide in the presence of water transforms into polymeric structures of molybdates [16]. The polymolybdates are known to be built of linear chains linked by oxygen atoms Mo–O–Mo. In these structures active Brönsted acid centers may be located on terminal and bridge oxygen atoms where hydrogen cations can be bound. Since polymeric layer covers both the surface and the catalyst pores (as it is manifested in lower surface area of MoO3 /SiO2 than SiO2 , Table 1) it can be expected that there is a relatively great amount of active acidic sites on the surface. In summary, the presence of active acidic centers (domains of HPM and polymolybdates) and excess of nitric acid enable to obtain high nitronium cation concentration in a short time which leads to high-yield nitration of toluene to DNT. The comparison of the toluene nitration yield in continuous process to the yield in batch process has been also made (Table 4). In batch process with 5 g of catalyst complete toluene conversion is achieved. However, this amount of catalyst is not enough to obtain the same amount of DNT as it is obtained in continuous process. Thirty-fold increase of product amount is obtained in the continuous process in the same time in comparison to batch process. Therefore, nitration in continuous process is signiﬁcantly more efﬁcient and enables to use the catalyst more effectively (catalyst activity factor is about ten-fold higher than in batch process). Space time yield is order of magnitude higher in continuous process. Nitration with lower excess of nitric acid (substrate molar ratio HNO3 :TN = 8:1) was examined at higher temperature (60 ◦ C) (Fig. 5). At this temperature also complete TN conversion to DNT is possible. Higher temperature decreases DNT absorption in catalyst pores and then this amount of nitric acid enables to wash off the product from the beginning. After 16 h the amount of DNT decreases from 97 wt.% to 70 wt.%. Therefore, after this time the catalyst was regenerated by phosphoric acid impregnation (the same amount of acid as was used in the preparation) and re-heating at 300 ◦ C for 1 h. This operation causes catalyst reactivation probably by HPM domains re-formation, because in the next hours the amount of DNT is increased. The composition of DNT is stable in this conditions, ca. 80% of 2,4-DNT, ca. 17% of 2,6-DNT and ca. 3% of 3,4-DNT are obtained. Therefore, applying higher temperature to the process enables to decrease the amount of nitric acid from 10:1 to 8:1 (substrate molar ratio HNO3 :TN) retaining the similar yield

Amount of DNT (mmol)

Catalyst Activity (molTN molMo −1 h−1 )

Space time yield (gTN gkat −1 h−1

1.4 43.2

0.31 3.32

0.074 0.87

and selectivity. The simple regeneration process allows to use the catalyst for a long time. 3.3. Chlorobenzene nitration Chlorobenzene nitration in continuous process was carried out with various excess of nitric acid in the presence of H3 PO4 /MoO3 /SiO2 catalyst. Chlorine is a functional group attached to a benzene molecule that removes electron density from the benzene ring [30]. This causes deactivation of aromatic ring to electrophilic aromatic substitution reaction. ClB nitration occurs slower and requires more drastic conditions in comparison to TN nitration [1–4,30,31]. Therefore, in ClB nitration greater amounts of nitric acid, longer reaction time and higher temperature are necessary. In Table 5 results of ClB nitration in continuous process in the presence of H3 PO4 /MoO3 /SiO2 catalyst are presented. Complete ClB conversion is obtained in the reaction with tenfold excess of nitric acid. However, after 5 h in these conditions only 27.9 wt.% of DNClB is formed. In the next 10 h the amount of DNClB is increased. This suggests accumulation of the product in the catalyst deposit and that ten-fold excess of nitric acid is not enough to wash off the whole product from catalyst. When substrate molar ratio HNO3 :ClB increases to 12:1 and 16:1 the amounts of DNClB after 5 h are even greater and equal to 80.8 wt.% and 95.7 wt.%, respectively. However, after 10 h the amount of DNClB is decreased as the catalyst presumably undergoes slow deactivation. The inﬂuence of catalyst amount on nitration yield with substrate molar ratio HNO3 :ClB = 12:1 was further examined. The greater amount of catalyst with retaining the reactor geometry causes contact time between substrate and catalyst to be extended. The longer contact time results in increase in amount of DNClB from 80.8 wt.% to 85 wt.% after ﬁrst 5 h. After 10 h of the process ClB conversion to DNClB increases to 94.5 wt.%. This indicates that greater amount of catalyst provides greater amount of acid active centers which enables to obtain higher reaction yield. The research showed that examined catalyst may be used in the DNClB synthesis in continuous nitration process. In the reaction with 10 g of the catalyst and twelve-fold excess of nitric acid carried out for 10 h the summary yield to DNClB is as high as 90 wt.%.

Table 5 Chlorobenzene nitration in continuous process with various substrate molar ratios HNO3 :ClB (5 g of catalyst, 80 ◦ C). Substrate molar ratio HNO3 :ClB

Time (h)

Product composition (wt.%)

ClB Conversion (wt.%)

MNClB

DNClB

10:1

5 10

72.1 33.3

27.9 66.4

100 100

12:1

5 10

19.2 41.6

80.8 58.2

100 100

12:1, 10 g

5 10

15.0 5.0

85.0 94.5

100 100

16:1

5 10

4.3 13.1

95.7 86.8

100 100

200

J. Adamiak et al. / Journal of Hazardous Materials 195 (2011) 195–200

4. Conclusions A novel catalyst system – H3 PO4 /MoO3 /SiO2 – was examined in toluene and chlorobenzene nitration in continuous process. The catalytic system possesses very active surface Brönsted acid centers that enable effective nitration of toluene to DNT in one-stage process (96 wt.% of DNT in the product) and in mild conditions i.e. at room temperature and only with ten-fold excess of nitric acid. On the other hand, in chlorobenzene nitration greater amounts of nitric acid and higher temperature are required, however, only twelvefold excess of nitric acid is needed to obtain as high yield to DNClB as 95 wt.%. To summarize, H3 PO4 /MoO3 /SiO2 catalyst has been shown to be very effective in toluene and chlorobenzene nitration that can be carried out in mild conditions in the more eco-friendly way. The greatest advantage is undoubtedly the organic solvent and sulfuric acid elimination from the process. All these qualities can make nitration reaction more ecological and cheaper process. Acknowledgments Financial support from the State Committee for Scientiﬁc Research of Poland (KBN–Project Nr 1198/B/H03/2009/37) is gratefully acknowledged. References [1] G.A. Olah, R. Malhotra, S.C. Narang, Nitration Methods and Mechanism, VCH, New York, 1989. [2] J.G. Hoggett, R.B. Moodie, J.R. Penton, K. Schoﬁeld, Nitration and Aromatic Activity, Cambridge University Press, 1971. [3] B.E. Berkman, The Industrial Synthesis of Aromatic Nitro Compounds and Amines, Scientiﬁc and Technical Publishing, Warsaw, 1967 (in Polish). ´ [4] T. Urbanski, Chemistry, Technology of Explosives, 1, Polish Scientiﬁc Publishers, Warsaw, 1965. [5] SIDS Initial Assessment Report, Paris, France, April 20–23, UNEP Publications, 2004. [6] W.P. Cetner, Chemistry and Technology of Explosives, Preparation of Explosives and Intermediate Products, Military University of Technology Press, Warsaw, 1986 (in Polish). [7] K. Wilson, J.H. Clark, Solid acids and their use as environmentally friendly catalysts in organic synthesis, Pure Appl. Chem. 72 (2000) 1313– 1319. [8] G. Centi, P. Ciambelli, S. Perathoner, P. Russo, Environmental catalysis: trends and outlook, Catal. Today 75 (2002) 3–15. [9] M. Ghiaci, Internal versus external surface active sites in ZSM-5 zeolite. Part 1. Fries rearrangement catalyzed by modiﬁed and unmodiﬁed H3 PO4/ ZSM-5, Appl. Catal. A 298 (2006) 32–39. [10] R.J. Kalbasi, Highly selective vapor phase nitration of toluene to 4-nitro toluene using modiﬁed and unmodiﬁed H3 PO4 /ZSM-5, Appl. Catal. A 353 (2009) 1–8.

[11] S.K. Bharadwaj, S. Hussain, M. Kar, M.K. Chaudhuri, Al(H2PO4)3: an efﬁcient catalyst for nitration of organic compounds with nitric acid, Catal. Commun. 9 (2008) 919–923. [12] S.K. Samantaray, K. Parida, Effect of phosphate ion on the textural and catalytic. Activity of titania–silica mixed oxide, Appl. Catal. A 220 (2001) 9–20. [13] S. Gong, L. Liu, Q. Cui, J. Ding, Liquid phase nitration of benzene over supported ammonium salt of 12-molybdophosphoric acid catalysts prepared by sol–gel method, J. Hazard. Mater. 178 (2010) 404–408. [14] I.V. Kozhevnikov, Catalysis by heteropoly acids and multicomponent polyoxometalates in liquid-phase reactions, Chem. Rev. 98 (1998) 171–198. [15] I.V. Kozhevnikov, New catalysts for cleaner organic synthesis innovations, Pharm. Technol. (2003) 96–102. ´ [16] W. Skupinski, M. Malesa, An infrared study on the MoO3 /SiO2 catalytic system employed in toluene nitration, Appl. Catal. A 236 (2002) 223–234. ´ Nitration of aro[17] T. Miłczak, J. Jacniacki, J. Zawadzki, M. Malesa, W. Skupinski, matic compounds on solid catalysts, Synth. Commun. 31 (2001) 173–187. [18] E.P. Barrett, L.G. Joyner, P.P. Halenda, The determination of pore volume and area distributions in porous substances. I. Computations from nitrogen isotherms, J. Am. Chem. Soc. 73 (1951) 373–383. [19] H. Hu, I.E. Wachs, S.R. Bare, Surface structures of supported molybdenum oxide catalysts: characterization by raman and Mo L3-edge XANES, J. Phys. Chem. 99 (1995) 10897–10910. [20] P. Villabrille, P. Vázquez, M. Blanco, C. Cáceres, Equilibrium adsorption of molybdosilicic acid solutions on carbon and silica: basic studies for the preparation of ecofriendly acidic catalysts, J. Colloid Interface Sci. 251 (2002) 151–159. [21] T. Mizushima, K. Fukushima, T.M. Huong, H. Ohkita, N. Kakuta, Synthesis of ␤MoO3 by simple evaporation of molybdic acid solution containing nitric acid, Chem. Lett. 34 (2005) 986–987. [22] E.M. McCarron III, ␤-MoO3 : a metastable analogue of WO3 , J. Chem. Soc. Chem. Commun. (1986) 336–338. [23] O. Collart, P. Van Der Voort, E.F. Vansant, E. Gustin, A. Bouwen, D. Schoemaker, R. Ramachandra Rao, B.M. Weckhuysen, R.A. Schoonheydt, Spectroscopic characterization of an MoOx layer on the surface of silica. An evaluation of the molecular designed dispersion method, Phys. Chem. Chem. Phys. 1 (1999) 4099–4104. [24] N. Kakuta, K. Tohji, Y. Udagawa, Molybdenum oxide structure on silicasupported catalysts studied by Raman spectroscopy and EXAFS, J. Phys. Chem. 92 (1988) 2583–2587. [25] H. Jeziorowski, H. Knözinger, Raman and ultraviolet spectroscopic characterization of molybdena on alumina catalysts, J. Phys. Chem. 83 (1979) 1166–1173. [26] E. Payen, J. Grimblot, S. Kasztelan, Study of oxidic and reduced aluminasupported molybdate and heptamolybdate species by in situ laser Raman spectroscopy, J. Phys. Chem. 91 (1987) 6642–6648. [27] C. Srilakshmi, W. Chew, K. Ramesh, M. Garland, Observation of the ammonium salt of 12-molybdophosphoric acid by in situ Raman spectroscopy during solidstate synthesis: spectral analysis and reconstruction using the band-target entropy minimization (BTEM) algorithm, Inorg. Chem. 48 (2009) 1967–1973. [28] C. Rocchiccioli-Deltcheff, A. Aouissi, M.M. Bettahar, S. Launay, M. Fournier, Catalysis by 12-molybdophosphates – 1. catalytic reactivity of 12molybdophosphoric acid related to its thermal behavior investigated through IR, Raman, polarographic, and X-ray diffraction studies: a comparison with 12-molybdosilicic acid, J. Catal. 164 (1996) 16–27. [29] M.N. Timofeeva, Acid catalysis by heteropoly acids, Appl. Catal. A 256 (2003) 19–35. [30] J.D. Hepworth, D.R. Waring, M.J. Waring, Aromatic Chemistry, Polish Scientiﬁc Publishers, Warsaw, 2009 (in Polish). [31] M.G. Kuba, R. Prins, G.D. Pirngruber, Batch and continuous nitration of toluene and chlorobenzene with acetyl nitrate over zeolite beta, Appl. Catal. A 333 (2007) 24–29.

Journal of Hazardous Materials 195 (2011) 201–207

Contents lists available at ScienceDirect

Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat

Anaerobic degradation of benzene by enriched consortia with humic acids as terminal electron acceptors Francisco J. Cervantes a,∗ , Ana Rosa Mancilla a , E. Emilia Ríos-del Toro a , Ángel G. Alpuche-Solís b , Lilia Montoya-Lorenzana b a División de Ciencias Ambientales, Instituto Potosino de Investigación Cientíﬁca y Tecnológica (IPICyT), Camino a la Presa San José 2055, Col. Lomas 4a . Sección, San Luis Potosí, SLP, 78216 Mexico b División de Biología Molecular, Instituto Potosino de Investigación Cientíﬁca y Tecnológica (IPICyT), Camino a la Presa San José 2055, Col. Lomas 4a . Sección, San Luis Potosí, SLP, 78216 Mexico

a r t i c l e

i n f o

Article history: Received 26 May 2011 Received in revised form 19 July 2011 Accepted 8 August 2011 Available online 16 August 2011 Keywords: Anaerobic benzene oxidation Hydrocarbon biodegradation Humus respiration Quinones

a b s t r a c t The anaerobic degradation of benzene coupled to the reduction of humic acids (HA) was demonstrated in two enriched consortia. Both inocula were able to oxidize benzene under strict anaerobic conditions when the humic model compound, anthraquinone-2,6-disulfonate (AQDS), was supplied as terminal electron acceptor. An enrichment culture originated from a contaminated soil was also able to oxidize benzene linked to the reduction of highly puriﬁed soil humic acids (HPSHA). In HPSHA-amended cultures, 9.3 ␮M of benzene were degraded, which corresponds to 279 ± 27 micro-electron equivalents (␮Eq) L−1 , linked to the reduction of 619 ± 81 ␮Eq L−1 of HPSHA. Neither anaerobic benzene oxidation nor reduction of HPSHA occurred in sterilized controls. Anaerobic benzene oxidation did not occur in soil incubations lacking HPSHA. Furthermore, negligible reduction of HPSHA occurred in the absence of benzene. The enrichment culture derived from this soil was dominated by two ␥-Proteobacteria phylotypes. A benzene-degrading AQDS-reducing enrichment originated from a sediment sample showed the prevalence of different species from classes ␤-, ␦- and ␥-Proteobacteria. The present study provides clear quantitative demonstration of anaerobic degradation of benzene coupled to the reduction of HA. © 2011 Elsevier B.V. All rights reserved.

1. Introduction Benzene is a widespread contaminant commonly found in aquifers due to leaks in underground fuel storage tanks, improper disposal techniques and spills of petroleum products. Contamination of water reservoirs by benzene has deserved particular scrutiny, as it is the most water-soluble and carcinogenic of all gasoline hydrocarbons [1]. Microbial degradation of benzene readily occurs under aerobic conditions by a wide variety of aerobic bacteria [2]. However, benzene often persists as contaminant in sedimentary environments because anaerobic conditions prevail [3–5]. Even though benzene is considered one of the most recalcitrant compounds in anaerobic environments, evidence collected during the last two decades indicates that anaerobic benzene degradation can be achieved under anaerobic conditions. Anaerobic benzene degradation has been demonstrated under methanogenic conditions [6,7] and linked to

∗ Corresponding author. Present address: Department of Biotechnology, Norwegian University of Science and Technology, N7491 Trondheim, Norway. E-mail address: [email protected] (F.J. Cervantes). 0304-3894/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.jhazmat.2011.08.028

the reduction of different terminal electron acceptors (TEA), including sulfate [8–10], nitrate [7,11–13], Fe(III) [14–16], and Mn(IV) [17,18]. More recently, anaerobic benzene oxidation was demonstrated with graphite electrodes as a TEA in sediment incubations [19]. In the present study, humic substances (HS) are evaluated as TEA for benzene biodegradation. HS constitute the most abundant organic fraction of the biosphere accumulating in terrestrial and aquatic environments. Although these compounds were previously considered to be very inert, evidence has been accumulated indicating that they have active roles in the anaerobic oxidation of several distinct organic compounds by serving as TEA for humic-reducing microorganisms [20]. The redox mediating properties of HS have mainly been attributed to quinone moieties [21], which are very abundant in HS and the humic model compound, anthraquinone-2,6-disulfonate (AQDS), has been used in several studies documenting the role of HS as TEA during the anaerobic oxidation of priority pollutants, such as vinyl chloride, dichloroethene [22], phenol, p-cresol [23,24] and toluene [25]. Furthermore, HS have been shown to promote the anaerobic degradation of benzene by serving as an electron shuttle between Fe(III)-reducing bacteria and insoluble Fe(III) oxides [26]. Nevertheless, no direct

202

F.J. Cervantes et al. / Journal of Hazardous Materials 195 (2011) 201–207

evidence has been reported quantifying the role of HS as TEA linked to the anaerobic degradation of benzene. The aim of the present study was to evaluate the capacity of humic-reducing consortia to oxidize benzene with AQDS and humic acids (HA) as TEA. Benzene-degrading consortia were enriched under AQDS-reducing conditions and were phylogenetically characterized. 2. Materials and methods 2.1. Reagents, materials and inocula Benzene (ACS grade, 99.8% purity) and AQDS were purchased from Sigma–Aldrich (Milwaukee, Mis.). Highly puriﬁed soil humic acids (HPSHA) were obtained from the International Humic Substances Society (IHSS, catalogue number 1S104H-5). All other reagents were of ACS grade. Two different inocula were selected for the present study based on their capacity to oxidize organic substrates (glucose, acetate) with AQDS as TEA during a preliminary screening and due to their past exposure to hydrocarbons. A soil sample derived from a petroleum reﬁnery in Poza Rica (Veracruz, Mexico) was obtained from a place in which a petroleum spill occurred and it is referred to as PR soil in the present study. The sediment derived from Marland Lagoon in Ébano (San Luis Potosí, Mexico) has historically been exposed to hydrocarbons as the ﬁrst petroleum production well exploited in Mexico was located next to the lagoon. Nowadays, periodic petroleum emanations naturally occur in the lagoon contaminating this water body. This consortium is referred to as ML sediment in the present study.

association between anaerobic benzene oxidation and AQDS reduction. All experimental treatments were performed by triplicate and statically incubated in the dark at 28 ◦ C. Manual shaking during sampling was conducted to promote homogenous distribution of benzene. 2.3. Anaerobic benzene oxidation with humic acids as TEA Enriched consortia in which benzene biodegradation linked to AQDS reduction was observed were evaluated for their capacity to achieve anaerobic benzene oxidation with HPSHA as TEA. Microbial incubations were performed in 60 mL glass serum bottles with basal medium (25 mL) provided with 5 g L−1 of HPSHA. Benzene was supplied from an anaerobic stock solution at the initial concentration of 30 ␮M referred to the liquid volume. Previous studies documented the role of this HA source as TEA during the anaerobic mineralization of toluene by enriched sediments [25]. The electron accepting capacity of this humic sample was determined as described below. During benzene biodegradation, benzene concentration and reduction of HPSHA were monitored as described below. Sterile controls were prepared under the same experimental conditions and autoclaved for 20 min at 120 ◦ C two times before addition of benzene. Incubation controls lacking benzene were also included in order to correct for the endogenous reduction of HPSHA. Finally, biologically active controls incubated without HPSHA were also included to verify the association between anaerobic benzene oxidation and the reduction of HPSHA. All experimental treatments were performed by triplicate and statically incubated in the dark at 28 ◦ C. Manual shaking during sampling was conducted to promote homogenous distribution of benzene. 2.4. DNA extraction and PCR ampliﬁcation

2.2. Benzene-degrading enrichment cultures with AQDS as TEA Incubations were conducted in 117 mL glass serum bottles with basal medium, which composition is as follows (g L−1 ): NaHCO3 , (1.68); NH4 Cl, (0.3); KH2 PO4 , (0.2); MgCl2 ·6H2 O, (0.03); CaCl2 , (0.1); Na2 S, (0.1) and 1 mL L−1 of trace elements solution with the composition previously described [23]. AQDS (5 mM) was included in the medium as TEA and the pH was controlled at 6.7 ± 0.2 by the bicarbonate added and a headspace composed of N2 /CO2 (80%/20%). Portions (50 mL) of basal medium were dispensed in serum bottles under anaerobic conditions (anaerobic hood with an atmosphere composed of N2 /H2 (95%/5%)) and then inoculation took place by adding 10 g (dry weight) per liter of previously homogenized inocula. Bottles were sealed with Viton Stoppers (Maag Technic AG, Dübendorf, Switzerland) and aluminum crimps and were ﬂushed with N2 /CO2 (80%/20%) to establish anaerobic conditions. Benzene was supplied from an anaerobic stock solution to the initial concentration of 50 ␮M referred to the liquid volume. After benzene biodegradation was observed linked to AQDS reduction, consecutive cycles were performed by replacing the medium in an anaerobic hood (atmosphere composed of N2 /H2 (95%/5%)) by freshly prepared anaerobic medium. Strict anaerobic conditions were guaranteed by ﬂushing the basal medium with N2 /CO2 (80%/20%), by adding sulﬁde as a reducing agent (0.1 g L−1 ) and by allowing initial AQDS reduction (up to 0.1 mM with endogenous substrates remaining from previous cycles) before addition of benzene under strict anaerobic conditions. Benzene degradation and AQDS reduction was determined as described below. Sterile controls were prepared under the same experimental conditions and autoclaved for 20 min at 120 ◦ C two times before addition of benzene. Incubation controls lacking benzene were also included in order to correct for the endogenous reduction of AQDS. Finally, biologically active controls incubated without AQDS were also included to verify the

Enriched biomass samples derived from PR soil and ML sediment were washed with 1 mL ice-cold PBS 1: 2-propanone (10:1 vol/vol) and then centrifuged for 15 min at 14,000 rpm. A second wash was performed with PBS 1: ethanol (1:1 vol/vol) and a third one with PBS 1. Genomic DNA was extracted by enzymatic lysis and chloroform-isoamyl (24:1) alcohol extraction and 1 volume isopropanol precipitation based on a previously published protocol [27]. The yield and quality of DNA were analysed electrophoretically on 1% (wt/vol) agarose gels. Bacterial 16S rRNA gene ampliﬁcations were performed using 27f (5 -AGAGTTTGATCMTGGCTCAG-3 ) and 1492r (5 TACGGYTACCTTGTTACGACTT-3 ) primers and conditions were modiﬁed from Bond et al. [28] (annealing 57 ◦ C; 30 cycles). Reaction mixture consisted in primers (0.2 ␮M each), 200 ␮M dNTPs, 2 mM MgCl2 , 1× PCR buffer and 0.04 U/␮L GoTaq DNA Polymerase (Promega). Ampliﬁcations were performed using a DNA Engine Peltier Thermal Cycler (Bio-Rad). 2.5. Cloning of 16S rRNA Four clone libraries were constructed: Original PR soil and ML sediment samples, and enriched PR soil and ML sediment samples obtained under AQDS-reducing conditions with benzene as a sole energy source. Amplicons were ligated into pGEM-T Easy cloning vector (Promega) and transformed by thermal shock of competent Escherichia coli TOP-10F’ cells. Plasmidic DNA was extracted from all transformed colonies by the method of Birnboim and Doly [29]. A terminal restriction fragment length polymorphism analysis (T-RFLP) was performed in order to avoid redundancies; plasmidic DNA was digested in a single reaction with EcoRI and MspI [30]. Enzymatic digestion was performed by incubating 3 ␮L of the insert with 0.3 U of each enzyme and the corresponding enzyme buffer at 37 ◦ C for 4 h. The digestion products were analysed in 2%

F.J. Cervantes et al. / Journal of Hazardous Materials 195 (2011) 201–207

agarose gels in TAE 1 and stained on 0.5 ␮g mL−1 ethidium bromide. Colonies showing distinct ﬁngerprint patterns were sequenced using the vector-speciﬁc M13f and M13r primers in a DNA automated sequencer ABI Prism Model 377r (Applied Biosystems, USA). 2.6. Phylogenetic analysis Electropherograms were transformed into contiguous fasta sequences with DNA baser software (www.dnabaser.com). Sequences orientation and chimeras were detected by using Orientation Checker v.1.0 and Pintail v. 1.0 [31,32]. The redundant phylotypes (sharing ≥97%) were collapsed in operational taxonomic units (OTUs) by distance analyses using the Ribosomal Database Project (RDP) web page (http://rdp.cme.msu.edu) [33]. The 16S rRNA gene sequences were aligned with ClustalW algorithm and optimized manually with type sequences using eBioX v. 1.5.1. [34]. Alignments included the phylotypes and the closest sister genera reported in The All-Species Living Tree project [35] and the Bergey’s Manual [36]. Posterior probability and topology of the phylogenetic trees were inferred by Bayesian analysis with Mr. Bayes v. 3.1.2 [37] deﬁning the parameters GTR + I + G, after a consensus of 3 × 105 generations (“burnin” of 50% a standard deviation of 0.003). The 16S rRNA gene sequences were deposited in the GenBank database under accession numbers: HQ602825–HQ602872 and HQ694510–HQ694512. 2.7. Analytical methods Benzene concentrations were determined in 100 ␮L headspace samples by gas chromatography (Agilent Technology, Model 6890N) and a ﬂame ionization detector. The chromatograph was equipped with a capillary column (Agilent DB-624) and nitrogen (25 ml per min) was used as a carrier gas. The temperature of the injection port, oven and detector, were 230, 60 and 230 ◦ C, respectively. Standard bottles were previously autoclaved for 20 min at 120 ◦ C two times and incubated at 28 ◦ C overnight before adding benzene. Electron accepting capacity (EAC) of HPSHA was determined by the ferrozine technique in a N2 /H2 (95%/5%) anoxic chamber following the protocol described by Lovley et al. [38]. The biological method to determine the EAC of this humic sample was based on the reduction by Geobacter sulfurreducens according to the protocol of Lovely et al. [38]. Reduction of AQDS (as AH2 QDS) was determined in the same anaerobic chamber according to Cervantes et al. [39]. Reduction of HPSHA was determined by the ferrozine technique according to Lovley et al. [38]. Subsamples in which Fe(III) citrate was no added were also monitored by the ferrozine technique in order to subtract the reducing equivalents received by intrinsic Fe(III) present in HPSHA. 3. Results 3.1. Anaerobic degradation of benzene with AQDS as TEA ML sediment degraded benzene under anaerobic conditions when AQDS was provided as TEA (supplementary data, Fig. S1). After more than two months of lag phase, concomitant benzene degradation and AQDS reduction occurred in biologically active sediment incubations. No benzene degradation was observed in sediment incubations lacking AQDS or in sterilized controls including AQDS. About 5% of the benzene initially spiked disappeared in all experimental treatments during the ﬁrst 4 weeks of incubations presumably due to adsorption on the sediment (no benzene removal occurred in sterilized controls without sediment); however, no further benzene removal was detected in sterile controls

203

and in sediment incubations lacking AQDS after this period. After 112 days of incubation, 12.9 ± 2.3 ␮M of benzene (corrected for the amount adsorbed) were degraded linked to the reduction of 293 ± 21 ␮M of AQDS (corrected for the endogenous control incubated in the absence of benzene). Thus, the stoichiometric relationship established during the anaerobic biodegradation of benzene was 1 ␮mol benzene degraded per 22.7 ␮mol of AH2 QDS produced, which is higher than the expected from the following stoichiometry: C6 H6 + 15 AQDS + 12 H2 O → 6 CO2 + 15 AH2 QDS Nevertheless, the results suggest that the amount of benzene degraded was completely oxidized by ML sediment. The higher than expected ratio benzene degraded/AQDS reduced may partly be explained by the biodegradation of adsorbed benzene, which could not be quantiﬁed. Anaerobic benzene oxidation was also observed in PR soil incubations (supplementary data, Fig. S2). Adsorption of benzene (∼45% of the amount initially added) occurred in all experimental treatments during the ﬁrst 3 months of incubation. However, after this period of time no further benzene removal was observed in sterilized controls. Clear distinction between abiotic benzene removal (e.g. adsorption) and biodegradation could be established after 4 months of incubation since benzene degradation only occurred in biologically active PR soil incubations. In the absence of AQDS, 2.67 ± 0.17 ␮M of benzene were degraded after 198 days of incubation. No methanogenic activity was detected during the anaerobic degradation of benzene in PR soil incubations lacking AQDS, suggesting that unidentiﬁed TEA present in this consortium were responsible for the anaerobic benzene oxidation observed. Moreover, simultaneous benzene biodegradation and AQDS reduction occurred in AQDS-amended soil incubations. After 198 days of incubation, 3.4 ± 0.67 ␮M of benzene were degraded (corrected for the amount of benzene degraded in the absence of AQDS) and 344 ± 20 ␮M of AH2 QDS were produced (corrected for the endogenous control lacking benzene), which corresponds to a ratio of 1 ␮M of benzene degraded per 101 ␮M of AH2 QDS produced. The ratio benzene degraded: AQDS reduced is very distant from that expected from the stoichiometry (see above). Thus, it is conceivable to assume that an important amount of benzene adsorbed might have been oxidized in PR soil incubations amended with AQDS. In fact, considering the amount of AH2 QDS produced and the stoichiometry, the expected amount of benzene that might have been oxidized under these conditions is ∼23 ␮M, corresponding to 46% of benzene originally spiked, which agrees with the amount of benzene adsorbed. Both benzene-degrading quinone-reducing consortia were incubated in several consecutive cycles in which AQDS reduction was consistently observed with benzene as a sole energy source for 18 months (data not shown). Enriched consortium derived from PR soil was selected for documenting the anaerobic benzene oxidation with HPSHA as TEA based on its superior capacity to achieve anaerobic benzene oxidation with AQDS as TEA (see below).

3.2. Anaerobic oxidation of benzene with humic acids as TEA The EAC of HPSHA was measured in order to supply with enough EAC to achieve complete anaerobic benzene oxidation (30 ␮M). Chemical EAC determined in a H2 /Pd reaction system yielded 342 ± 23 micro-reducing equivalents (␮Eq) g−1 . Moreover, biological EAC determined with G. sulfurreducens yielded 216 ± 6 ␮Eq g−1 . Thus, 5 g HPSHA L−1 was supplied in order to have an excess of EAC for anaerobic benzene oxidation (e.g. 30 ␮M of benzene require 900 ␮Eq L−1 for complete oxidation).

204

F.J. Cervantes et al. / Journal of Hazardous Materials 195 (2011) 201–207

A

35 Sterile control without sediment

Benzene (µM)

30 Sterile control

25

Only Benzene

20 15

Benzene + HPSHA 10 0

Reduced humic acids (mEq/L)

B

20

40

60

80

100

0.8 Benzene + HPSHA 0.6

0.4

0.2

Sterile control Endogenous control

0 0

20

40

60

80

100

Time (days) Fig. 1. Anaerobic benzene degradation (A) coupled to reduction of HPSHA (B) by PR soil in incubations containing bicarbonate-buffered medium supplemented with 5 g of HPSHA per liter. Endogenous controls were incubated in the absence of benzene. Sterile controls including HPSHA. Data represent average from triplicate measurements and standard deviations in each treatment.

An enrichment culture derived from PR soil achieved anaerobic benzene oxidation linked to the microbial reduction of HPSHA (Fig. 1). Adsorption to PR soil accounted for ∼40% of benzene removal, which occurred during the ﬁrst 4 weeks of incubation in all experimental treatments including PR soil, but no benzene removal was observed in chemical controls lacking PR soil (Fig. 1). However, a clear distinction between adsorption and biodegradation could be established after this incubation period. Certainly, anaerobic benzene oxidation was only observed in soil incubations amended with HPSHA after this period of time. Furthermore, the role of HPSHA as TEA supporting anaerobic benzene oxidation could be corroborated by measuring a signiﬁcant amount of reduced humic acids as compared with the endogenous control lacking benzene (Fig. 1). In fact, 9.3 ± 0.4 ␮M (corrected for the sterilized control) of benzene were degraded in HPSHA-amended cultures, which corresponds to 279 ± 27 ␮Eq L−1 , linked to the reduction of 619 ± 81 ␮Eq L−1 of HPSHA (corrected for the endogenous control lacking benzene). Neither anaerobic benzene oxidation nor reduction of HPSHA occurred in sterilized controls. Furthermore, benzene oxidation did not occur in soil incubations lacking HPSHA. The superior reduction of HPSHA as compared with that expected from the anaerobic benzene oxidation quantiﬁed could be explained by biodegradation of adsorbed benzene, which could not be quantiﬁed. 3.3. Phylogenetic characterization of benzene-degrading humus-reducing enriched consortia The characterization of the four clone libraries: original PR soil and ML sediment and AQDS-benzene enriched PR soil and ML sediment; resulted in a total of 227 clones which were grouped into

32 different digestion patterns by T-RFLP analysis and 72 clones were sequenced redundantly to conﬁrm the patterns. From the 72 sequenced clones, 23 sequences (30%) were identiﬁed as chimeras and were excluded from the phylogenetic analysis. Tree branching was resolved unambiguously for most 16S rRNA gene OTUs at level of genera. However, ﬁve clones (C11, A6, A10, E1 and E2) remained afﬁliated until Class level. A posterior probability of 100% was calculated in most nodes. In some cases, the weakly supported nodes were examined with sequences from clones available at the GenBank database, and the branching within the family reached a clear deﬁnition. A phylogenetic tree based on the sequences derived from this study and GenBank accession numbers is shown in Fig. 2. AQDS-benzene enrichment libraries were predominantly composed of members of the phylum Proteobacteria and included the highest proportion of OTUs (48%). In PR consortia, two OTUs (C9 and C3) were not observed in the original sample and were assigned to ␥-Proteobacteria. OTU C9 was closely related to Pseudoxanthomonas clade with a posterior probability of 100%. OTU C3 (HQ602844) allowed a phylogenetic assignment to ␥Proteobacteria in Pseudomonas clade. As for OTU C3 the search against the GenBank and RDP databases (using BLAST and Classiﬁer tools, respectively), rendered the following results: BLAST retrieved sequences where the best hit showed a similarity of 93% to Pseudomonas clone (EU266810) and Classiﬁer assigned a similarity of 94% to Pseudomonas genera. However, caution for species identiﬁcation by local alignment (BLAST and Classiﬁer) was taken into account for all phylotypes and the phylotype identiﬁcation was preferably interpreted by phylogenetic analysis. This assignment revealed a close relation of phylotype C3 to the Pseudomonadaceae clade, because preliminary phylogenies discarded a clustering to the Moraxellaceae (the other family in Pseudomonadales), Halomonadales and Oceanospirillales clades. In ML consortia, 22 OTUs were detected and resolved at species level, the sequences were predominantly composed by ␥-Proteobacteria species. Moreover, several species belonging to the following phyla were detected in ML sediment enrichment culture: Bacteroidetes, Chloroﬂexi, uncultured phylum TM-7 and as well as the classes ␤-, ␦- and ␥-Proteobacteria.

4. Discussion The aim of the present study was to evaluate the capacity of humic-reducing consortia to achieve anaerobic benzene oxidation with HPSHA and the humic model compound, AQDS, as TEA. Multiple experimental proofs evidenced the role of humic substances as TEA during anaerobic benzene oxidation by enriched consortia. Both inocula studied (ML sediment and PR soil) could couple the anaerobic oxidation of benzene to AQDS reduction. Furthermore, a benzene-degrading enrichment culture derived from PR soil achieved anaerobic benzene oxidation when HPSHA were supplied as TEA. Concomitant anaerobic benzene oxidation and microbial reduction of HPSHA occurred in soil incubations supplemented with this TEA. Neither anaerobic benzene oxidation nor reduction of HPSHA occurred in sterilized controls. Furthermore, benzene oxidation did not occur in soil incubations lacking HPSHA. Furthermore, minor reduction of HPSHA occurred in endogenous controls lacking benzene. The present study shows a clear quantitative demonstration that anaerobic benzene oxidation can be accomplished by anaerobic consortia with humic acids as TEA. Previously, it was hypothesized that HA had served as a direct electron acceptor during benzene biodegradation when HA were added as chelators to increase Fe(III) oxide bioavailability for a benzene-degrading Fe(III)-reducing consortium in contaminated

F.J. Cervantes et al. / Journal of Hazardous Materials 195 (2011) 201–207

205

Fig. 2. Phylogenetic tree based on 16S rRNA gene sequences obtained from the clone libraries in this study. Two species of Aquifex were used as outgroup. The posterior probability support is indicated in nodes. The number of collapsed phylotypes assigned to OTU by sequencing and T-RFLP pattern analysis is indicated in parenthesis next to the GenBank accession number. Phlylotypes in black font correspond to members of original samples (ML sediment and PR soil origin site). The color font mark the phylotypes detected in the benzene-AQDS enrichments: red (ML sediment) and blue (PR soil). The symbol * denotes the genera or family reported to biodegrade aromatic hydrocarbons. The scale bar represents expected changes per site. (For interpretation of the references to color in this ﬁgure legend, the reader is referred to the web version of the article.)

sediment [38]. This hypothesis was based on the observation that HA promoted benzene biodegradation better than synthetic chelators (e.g. EDTA and NTA) even though humus had inferior chelating properties [26]. The mechanism proposed implies that benzene had been degraded with HA serving as TEA and the obtained reduced HA had been recycled back to the oxidized form by chemical reaction

with Fe(III) oxides present in the cultures. However, the reduction of HA during benzene biodegradation was not demonstrated and the EAC of HA could not be differentiated from its impact as chelating agent. The impact of AQDS on the anaerobic benzene oxidation was also studied in three different sites of Fe(III)-reducing sediments [14]. Stimulation of anaerobic benzene oxidation was

206

F.J. Cervantes et al. / Journal of Hazardous Materials 195 (2011) 201–207

observed at one site when 600 ␮M AQDS was applied, which may have been due to the use of AQDS as TEA, but the reduction of AQDS was not demonstrated. The same sediment sample did not oxidize benzene when 300 ␮M AQDS was applied, yet the amount of benzene added (12 ␮M) would have only required 180 ␮M AQDS for complete oxidation. The role of HA and the humic model compound, AQDS, as TEA has previously been demonstrated during the anaerobic oxidation of other priority pollutants, such as vinyl chloride, dichloroethene [22], phenolic compounds [23,24], and toluene [25]. Thus, these ﬁndings suggest that HS may play a signiﬁcant role during the intrinsic bioremediation of contaminates sites by serving as TEA. The ubiquity of HS and humic-reducing microorganisms in many different terrestrial and sedimentary environments [39,40] further emphasize their potential impact in the anaerobic oxidation of priority pollutants. Moreover, reduced HA can be recycled to its oxidized form by chemical reaction with metal oxides commonly found in anaerobic habitats [38,41], thus allowing its role as TEA at sub-stoichiometric concentrations [25]. However, a greater contribution of metal oxides on the recycling of HA is expected in environments that continuously shift from anoxic to aerobic conditions, such as shallow water bodies, so that metal oxides could be replenished. Reduced HA can also be recycled back to their oxidized form by transferring electrons to electron-accepting pollutants, such as azo dyes, nitroaromatics, polyhalogenated compounds and radionuclides [20]. Furthermore, reduced HA could also serve as electron donor for the microbial reduction of other electron acceptors, such as nitrate, nitrite, nitrous oxide, fumarate, perchlorate, arsenate and selenate [42–44]. The phylogenetic characterization of the microbial communities sampled from four PR soil and ML sediment clone libraries via 16S rRNA gene targeted T-RFLP analyses revealed that a signiﬁcant OTU fraction was detected exclusively after enrichment with AQDS-benzene (22 of 32 OTUs). In ML sediment the selective AQDSbenzene enrichment revealed an occurrence of ﬁve OTUs belonging to the anaerobic taxa Anaerolineaceae and Desulfobacca acetoxidans, which have previously been reported to degrade monoaromatic hydrocarbons under different redox conditions [45,46]. Additionally, in ML sediment AQDS-benzene enrichment the facultative genera Shewanella, Pseudomonas and Comamonas were detected, which agrees with previous reports documenting the anaerobic degradation of benzene by other members of ␤-Proteobacteria, such as species from Dechloromonas [47,48] and Pelomonas [49]. On the other hand, in AQDS-benzene enriched PR soil a Pseudomonadaceae phylotype was observed, which is a relevant ﬁnding since members of this genera have also been reported to degrade monoaromatic compounds [2,50,51]. The strict aerobes Achromobacter and Stenotrophomonas have been reported to metabolize monoaromatic hydrocarbons [52]. The phylum TM-7 has no cultured species; however, it has been described in anoxic conditions where the biodegradation of nitrogen containing aromatic compounds occurs [53]. Furthermore, a wide variety of microorganisms belonging to several phyla identiﬁed in both enrichments, have previously been reported to reduce either HS or AQDS [54]. Therefore, the results suggest that the enriched microorganisms detected are involved during the anaerobic degradation of benzene using HA or AQDS as TEA.

5. Conclusions The anaerobic degradation of benzene coupled to the reduction of HPSHA or AQDS was demonstrated in two enriched consortia. Both benzene-degrading enrichment cultures were phylogenetically characterized and the results indicate the prevalence of different species from classes ␤-, ␦- and ␥-Proteobacteria. The

present study provides a clear quantitative demonstration of anaerobic degradation of benzene coupled to the reduction of HA. Acknowledgements We thank the technical assistance of Dulce Partida, Guillermo Vidriales, Ma. Del Carmen Rocha-Medina and Jorge GonzalezEstrella. This study was ﬁnancially supported byCouncil of Science and Technology of Mexico (grant SEP-CONACYT 55045). Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.jhazmat.2011.08.028. References [1] M.E. Caldwell, J.M. Suﬂita, Detection of phenol and benzoate as intermediates of anaerobic benzene biodegradation under different terminal electron-accepting conditions, Environ. Sci. Technol. 34 (2000) 1216–1220. [2] S.A.B. Weelink, M.H.A. Van Eekert, A.J.M. Stams, Degradation of BTEX by anaerobic bacteria: physiology and application, Rev. Environ. Sci. Biotechnol. 9 (2010) 359–385. [3] C.M. Reddy, T.I. Eglinton, A. Hounshell, H.K. White, L. Xu, R.B. Gaines, G.S. Frysinger, The West Falmouth oil spill after thirty years: the persistence of petroleum hydrocarbons in marsh sediments, Environ. Sci. Technol. 36 (2002) 4754–4760. [4] S.W. Rogers, S.K. Ong, B.H. Kjartanson, J. Golchin, G.A. Stenback, Natural attenuation of polycyclic aromatic hydrocarbon-contaminated sites: review, Pract. Per. Hazard. Toxic. Rad. Waste Mgmt. 6 (2002) 141–155. [5] G.S. Frysinger, R.B. Gaines, L. Xu, C.M. Reddy, Resolving the unresolved complex mixture in petroleum-contaminated sediments, Environ. Sci. Technol. 37 (2003) 1653–1662. [6] T.M. Vogel, D. Grbic-Galic, Incorporation of oxygen from water into toluene and benzene during anaerobic fermentative transformation, Appl. Environ. Microbiol. 52 (1986) 200–202. [7] A.C. Ulrich, E.A. Edwards, Physiological and molecular characterization of anaerobic benzene-degrading mixed cultures, Environ. Microbiol. 5 (2003) 92–102. [8] E.A. Edwards, D. Grbic-Galic, Complete mineralization of benzene by aquifer microorganisms under strictly anaerobic conditions, Appl. Environ. Microbiol. 58 (1992) 2663–2666. [9] J.D. Coates, R.T. Anderson, J.C. Woodward, E.J.P. Phillips, D.R. Lovley, Anaerobic hydrocarbon degradation in petroleum-contaminated harbor sediments under sulfate-reducing and artiﬁcially imposed iron-reducing conditions, Environ. Sci. Technol. 30 (1996) 2784–2789. [10] R.T. Anderson, D.R. Lovley, Anaerobic bioremediation of benzene under sulfatereducing conditions in a petroleum contaminated aquifer, Environ. Sci. Technol. 34 (2000) 2261–2266. [11] D.W. Major, C.I. Mayﬁeld, J.F. Barker, Biotransformation of benzene by denitriﬁcation in aquifer sand, Ground Water 26 (1988) 8–14. [12] S.M. Burland, E.A. Edwards, Anaerobic benzene biodegradation linked to nitrate reduction, Appl. Environ. Microbiol. 65 (1999) 529–533. [13] Y. Kasai, Y. Takahata, M. Maneﬁeld, K. Watanabe, RNAbased stable isotope probing and isolation of anaerobic benzene-degrading bacteria from gasolinecontaminated groundwater, Appl. Environ. Microbiol. 72 (2006) 3586–3592. [14] R.T. Anderson, D.R. Lovley, Naphthalene and benzene degradation under Fe(III)reducing conditions in petroleum-contaminated aquifers, Bioremediat. J. 3 (1999) 121–135. [15] S. Botton, J.R. Parsons, Degradation of BTX by dissimilatory iron-reducing cultures, Biodegradation 18 (2007) 373–381. [16] U. Kunapuli, T. Lueders, R.U. Meckenstock, The use of stable isotope probing to identify key iron-reducing microorganisms involved in anaerobic benzene degradation, ISME J. 1 (2007) 643–653. [17] W.R. Villatoro-Monzon, A.M. Mesta-Howard, E. Razo-Flores, Anaerobic biodegradation of BTEX using Mn(IV) and Fe(III) as alternative electron acceptors, Water Sci. Technol. 48 (2003) 125–131. [18] W.R. Villatoro-Monzon, M.G. Morales-Ibarria, E.K. Velázquez, H. Ramírez-Saad, E. Razo-Flores, Benzene biodegradation under anaerobic conditions coupled with metal oxides reduction, Water Air Soil Pollut. 192 (2008) 165–172. [19] T. Zhang, S.M. Gannon, K.P. Nevin, A.E. Franks, D.R. Lovley, Stimulating the anaerobic degradation of aromatic hydrocarbons in contaminated sediments by providing an electrode as the electron acceptor, Environ. Microbiol. 12 (2010) 1011–1020. [20] F.P. Van der Zee, F.J. Cervantes, Impact and application of electron shuttles on the redox (Bio)transformation of contaminants: a review, Biotechnol. Adv. 27 (2009) 256–277. [21] D.T. Scott, D.M. McKnight, E.L. Blunt-Harris, S.E. Kolesar, D.R. Lovley, Quinone moieties act as electron acceptors in the reduction of humic substances by humics-reducing microorganisms, Environ. Sci. Technol. 32 (1998) 2984–2989.

F.J. Cervantes et al. / Journal of Hazardous Materials 195 (2011) 201–207 [22] P.M. Bradley, F.H. Chapelle, D.R. Lovley, Humic acids as electron acceptors for anaerobic microbial oxidation of vinyl chloride and dichloroethene, Appl. Environ. Microbiol. 64 (1998) 3102–3105. [23] F.J. Cervantes, S. Van der Velde, G. Lettinga, J.A. Field, Quinones as terminal electron acceptors for anaerobic microbial oxidation of phenolic compounds, Biodegradation 11 (2000) 313–321. [24] F.J. Cervantes, C.H. Gutiérrez, K.Y. López, M.I. Estrada-Alvarado, E.R. MezaEscalante, A.-C. Texier, F. Cuervo, J. Gómez, Contribution of quinone-reducing microorganisms on the anaerobic biodegradation of organic compounds under different redox conditions, Biodegradation 19 (2008) 235–246. [25] F.J. Cervantes, W. Dijksma, T. Duong-Dac, A. Ivanova, G. Lettinga, J.A. Field, Anaerobic mineralization of toluene by enriched sediments with quinones and humus as terminal electron acceptors, Appl. Environ. Microbiol. 67 (2001) 4471–4478. [26] D.R. Lovley, J.C. Woodward, F.H. Chapelle, Rapid anaerobic benzene oxidation with a variety of chelated Fe(III) forms, Appl. Environ. Microbiol. 62 (1996) 288–291. [27] J.D. Wisotzkey, P.J. Jurtshuck, G.E. Fox, PCR ampliﬁcation of 16S rDNA from lyophilized cell cultures facilitates studies in molecular systematics, Curr. Microbiol. 21 (1990) 325–327. [28] P.L. Bond, G.K. Druschel, J.F. Banﬁeld, Comparison of acid mine drainage microbial communities in physically and geochemically distinct ecosystems, Appl. Environ. Microbiol. 66 (2000) 4962–4971. [29] H. Birnboim, J. Doly, A rapid alkaline extraction procedure for screening recombinant plasmid DNA, Nucleic Acids Res. 7 (1979) 1513–1523. [30] J.A. Mauricio-Castillo, G.R. Arguello-Astorga, A.G. Alpuche-Solis, C.T. MonrealVargas, O. Diaz-Gomez, R. De la Torre-Almaraz, First report of tomato severe leaf curl virus in Mexico, Plant Dis. 90 (2006) 1116. [31] K.E. Ashelford, N.A. Chuzhanova, J.C. Fry, A.J. Jones, A.J. Weightman, At least 1 in 20 16S rRNA sequence records currently held in public repositories is estimated to contain substantial anomalies, Appl. Environ. Microbiol. 71 (2005) 7724–7736. [32] K.E. Ashelford, N.A. Chuzhanova, J.C. Fry, A.J. Jones, A.J. Weightman, New screening software shows that most recent large 16S rRNA gene clone libraries contain chimeras, Appl. Environ. Microbiol. 72 (2006) 5734–5741. [33] J.R. Cole, Q. Wang, E. Cardenas, J. Fish, B. Chai, R.J. Farris, A.S. Kulam-SyedMohideen, D.M. McGarrell, T. Marsh, G.M. Garrity, J.M. Tiedje, The Ribosomal Database Project: improved alignments and new tools for rRNA analysis, Nucleic Acids Res. 37 (2009) D141–D145. [34] A.M. Barrio, E. Lagercrantz, G.O. Sperber, J. Blomberg, E. Bongcam-Rudloff, Annotation and visualization of endogenous retroviral sequences using the Distributed Annotation System (DAS) and eBioX, BMC Bioinf. 10 (2009) S18. [35] P.M. Yarza, R.J. Peplies, J. Euzeby, R. Amann, K.H. Schleifer, W. Ludwig, F.O. Glöckner, R. Rossello-Mora, The all-species living tree project: A 16S rRNAbased phylogenetic tree of all sequenced type strains, Syst. Appl. Microbiol. 31 (2008) 241–250. [36] G.M. Garrity, J.A. Bell, T.G. Lilburn, Taxonomic Outline of the Prokaryotes Bergey’s Manual of Systematic Bacteriology, Springer, New York, 2004. [37] F. Ronquist, J.P. Huelsenbeck, MrBayes 3: Bayesian phylogenetic inference under mixed models, Bioinformatics 19 (2003) 1572–1574. [38] D.R. Lovley, J.D. Coates, E.L. Blunt-Harris, E.J.P. Phillips, J.C. Woodward, Humic substances as electron acceptors for microbial respiration, Nature 382 (1996) 445–448.

207

[39] F.J. Cervantes, S. Van der Velde, G. Lettinga, J.A. Field, Competition between methanogenesis and quinone respiration for ecologically important substrates in anaerobic consortia, FEMS Microbiol. Ecol. 34 (2000) 161– 171. [40] J.D. Coates, D.J. Ellis, E. Roden, K. Gaw, E.L. Blunt-Harris, D.R. Lovley, Recovery of humics-reducing bacteria from a diversity of sedimentary environments, Appl. Environ. Microbiol. 64 (1998) 1504–1509. [41] A.T. Stone, J.J. Morgan, Reduction of manganese(III) and manganese(IV) oxides by organics. 1. Reaction with hydroquinone, Environ. Sci. Technol. 18 (1984) 450–456. [42] R.A. Bruce, L.A. Achenbach, J.D. Coates, Reduction of (per)chlorate by a novel organism isolated from a paper mill waste, Environ. Microbiol. 1 (1999) 319–331. [43] D.R. Lovley, J.L. Fraga, J.D. Coates, E.L. Blunt-Harris, Humics as an electron donor for anaerobic respiration, Environ. Microbiol. 1 (1999) 89–98. [44] C. Aranda-Tamaura, M.I. Estrada-Alvarado, A.-C. Texier, F. Cuervo, J. Gómez, F.J. Cervantes, Effects of different quinoid redox mediators on the removal of sulphide and nitrate via denitriﬁcation, Chemosphere 69 (2007) 1722–1727. [45] A. Alfreider, C. Vogt, Bacterial Diversity, Aerobic biodegradation potential in a BTEX-contaminated aquifer, Water Air Soil Pollut. 183 (2007) 415–426. [46] S. Herrmann, S. Kleinsteuber, T.R. Neu, H.H. Richnow, C. Vogt, Enrichment of anaerobic benzene-degrading microorganisms by in situ microcosms, FEMS Microbiol. Ecol. 63 (2008) 94–106. [47] J.D. Coates, R. Chakraborty, J.G. Lack, S.M. O’Connor, K.A. Cole, K.S. Bender, L.A. Achenbach, Anaerobic benzene oxidation coupled to nitrate reduction in pure culture by two strains of Dechloromonas, Nature 411 (2001) 1039–1043. [48] R. Chakraborty, J.D. Coates, Hydroxylation carboxylation – two crucial steps of anaerobic benzene degradation by Dechloromonas strain RCB, Appl. Environ. Microbiol. 71 (2005) 5427–5432. [49] J.S.-C. Liou, C.M. DeRito, E.L. Madsen, Field-based and laboratory stable isotope probing surveys of the indetities of both aerobic and anaerobic benzenemetabolizing microorganisms in freshwater sediment, Environ. Microbiol. 10 (2008) 1964–1977. [50] N. Gülensoy, P.J. Alvarez, Diversity and correlation of speciﬁc aromatic hydrocarbon biodegradation capabilities, Biodegradation 10 (1999) 331–340. [51] M. Xu, J.J. Guo, S.G. Guoping, Biodegradation of textile azo dye by Shewanella decolorationis S12 under microaerophilic conditions, Appl. Microbiol. Biotechnol. 76 (2007) 719–726. [52] D.R. Nielsen, P.J. McLellan, J.A. Daugulis, Direct estimation of the oxygen requirements of Achromobacter xylosoxidans for aerobic degradation of monoaromatic hydrocarbons (BTEX) in a bioscrubber, Biotechnol. Lett. 28 (2006) 1293–1298. [53] X.Y. Liu, B.J. Wang, C.Y. Jiang, K.X. Zhao, H.L. Drake, S.J. Liu, Simultaneous biodegradation of nitrogen-containing aromatic compounds in a sequencing batch bioreactor, J. Environ. Sci. 19 (2007) 530–535. [54] J.A. Field, F.J. Cervantes, Microbial redox reactions mediated by humus and structurally related quinones, in: I.V. Perminova, K. Hatﬁeld, N. Hertkorn (Eds.), Use of Humic Substances to Remediate Polluted Environments: From Theory to Practice, Springer Publishers, The Netherlands, 2005, pp. 343–352.

Journal of Hazardous Materials 195 (2011) 208–213

Contents lists available at ScienceDirect

Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat

Protective effect of Panax ginseng against serum biochemical changes and apoptosis in liver of rats treated with carbon tetrachloride (CCl4 ) Emre Karakus a , Ali Karadeniz b,∗ , Nejdet Simsek c , Ismail Can c , Adem Kara c , Serap Yildirim d , Yildiray Kalkan e , Fikrullah Kisa a a

Department of Pharmacology and Toxicology, Faculty of Veterinary Medicine, University of Atatürk, 25240, Erzurum, Turkey Department of Physiology, Faculty of Veterinary Medicine, University of Atatürk, 25240, Erzurum, Turkey c Department of Histology and Embryology, Faculty of Veterinary Medicine, University of Atatürk, 25240, Erzurum, Turkey d Department of Physiology, Faculty of Medicine, University of Atatürk, 25240, Erzurum, Turkey e Department of Histology and Embryology, Faculty of Medicine, University of Rize, 53100, Rize, Turkey b

a r t i c l e

i n f o

Article history: Received 28 February 2011 Received in revised form 1 August 2011 Accepted 9 August 2011 Available online 16 August 2011 Keywords: Panax ginseng Carbon tetrachloride Liver Biochemistry CD68+ Caspase-3

a b s t r a c t The purpose of this study was to investigate possible beneﬁcial effects of Panax ginseng (PG) on carbon tetrachloride (CCl4 )-induced acute hepatotoxicity in rats. CCl4 challenge elevated serum enzyme activities of liver and some biochemical parameters, but these effects were prevented by the pretreatment of rats with PG. Histologically, a great amount of mononuclear cells inﬁltration, necrotic cells and few ﬁbroblasts were observed in liver of CCl4 group. Also, CD68+ and caspase-3 staining cells were diffused in both lobular and portal areas. However, PG pretreatment had a little inﬂuence on the number of caspase-3 immunpositive staining cells in the liver, but CD68+ staining areas were signiﬁcantly decreased in the PG + CCl4 when compared to CCl4 group. We conclude that PG treatment may play a protective role by enhancing liver enzyme activities and recovering biochemical parameters, and improving the changes in histological structure against CCl4 -induced liver damages in rats. Crown Copyright © 2011 Published by Elsevier B.V. All rights reserved.

1. Introduction The liver is a vital organ that plays a key role in many toxication cases. The hepatotoxicants including carbon tetrachloride (CCl4 ), nitrosamines, and polycyclic aromatic hydrocarbons are transformed into the intermediate reactive oxygen species (ROS), including oxygen free radicals then, they show their hepatotoxic effects in experimental animals and humans [1,2]. Carbon tetrachloride is a common industrial solvent used as hepatotoxin in the experimental studies for liver diseases. It is metabolized by cytochrome P450 in liver cells to yield the reactive metabolic hepatotoxic metabolites that trichloromethyl free radicals (CCl3 • ) and/or trichloromethyl peroxyl radicals (CCl3 O2 (). The toxicity of CCl4 probably depends on formation of the trichloromethyl radical (CCl3 (), which in the presence of oxygen interacts with it to form the more toxic trichloromethyl peroxyl radical (CCl3 O2 () [3]. CCl3 O2 ( is capable of abstracting hydrogen from polyunsaturated fatty acids to initiate lipid peroxidation. Therefore, both CCl3 ( and CCl3 O2 ( causes damage in cell membrane, change enzyme activity and ﬁnally induce hepatic injury or necro-

∗ Corresponding author. Tel.: +90 442 2315525; fax: +90 442 236 08 81. E-mail address: [email protected] (A. Karadeniz).

sis [4]. The number of inﬁltrated neutrophils, macrophages, Kupffer cells, lymphocytes and natural killer cells are signiﬁcantly increase after liver injury induced by hepatotoxins such as CCl4 , alcohols, d-galactosamine, etc. most of which causes activation of liver resident macrophages and/or chemoattraction of extrahepatic cells (e.g. neutrophils and lymphocytes) [5]. The activated macrophages are release to cell death ligand (CD95L and TNF alfa, etc.) that mediated apoptosis might be contributed to liver ﬁbrosis, inﬂammation and injury [6,7]. Following damages and inﬂammation in the liver tissue, repairing hepatocytes by anti-inﬂammatory agents take places where death necrotic and apoptotic cells [8]. Herbal compounds obtained from plant extracts that reduce chemical activating enzymes could be considered as good candidates for protection against chemically induced toxicities such as CCl4 and cisplatin. Panax ginseng, a traditional multipurpose herb in Asia, has become the World’s most popular herbal supplements in recent years. Ginseng has a variety of beneﬁcial biological processes that include anti-carcinogenic, anti-diabetic and anti-inﬂammatory effects, as well as cardiovascular- and neuro-protection [9–11]. Most of the pharmacological actions of ginseng are attributed to a variety of ginsenosides, which are phenolic acids, ﬂavonoids and triterpenoid saponins [12,13]. These properties of the ginseng are thought to provide many beneﬁcial effects against organ damages. Thus, we investigated effects on

0304-3894/$ – see front matter. Crown Copyright © 2011 Published by Elsevier B.V. All rights reserved. doi:10.1016/j.jhazmat.2011.08.027

E. Karakus et al. / Journal of Hazardous Materials 195 (2011) 208–213

serum lipid proﬁle and liver enzymes to determine the protective effect of P. ginseng against CCl4 -induced damage in male rats.

209

i50). For each specimen, CD68+ and caspase-3 immunoreactivity were determined in 10 randomly selected areas of approximately X20 objective.

2. Material and methods 2.6. Statistical analysis 2.1. Chemicals P. ginseng was purchased from trade cooperation SGM (Ankara, Turkey). CCl4 and all other chemicals of analytical grade were purchased from Sigma Chemical Co. (St. Louis, MO) and IBL Chemical Co. (Ankara, Turkey).

The results were expressed as mean ± SEM of ten animals in each group. The data were subjected to one-way ANOVA followed by Tukey’s multiple comparison tests. Student’s t-test was used in the two-group comparison. Statistical signiﬁcance was accepted for all tests at p < 0.05.

2.2. Animals and hausing

3. Results

Forty adult male Wistar albino rats (n = 10 × 4) weighing about 250 g were used. The animals were given standard rat pellets and tap water ad libitum. The rats were housed in individual cages (360 mm × 200 mm × 190 mm), each containing 2 or 3 animals from 15 days before the start of the experiment. All animals were housed in stainless steel cages under standard laboratory conditions (light period 07.00 a.m. to 8.00 p.m. h, 21 ± 2 ◦ C, relative humidity 55%), and received humane care according to the criteria outlined in the Guide for the Care and Use of Laboratory Animals prepared by the National Academy of Sciences and published by the National Institute of Health. 2.3. Experimental design Rats were divided into four groups, each containing 10 animals. Control group (C) was injected intraperitoneally (i.p.) with 1 ml physiological saline for 7 consecutive days. CCl4 toxication group (CCl4 ) was given a single i.p. dose of CCl4 10 ml/kg. P. ginseng + CCl4 toxication group (PG + CCl4 ) was injected i.p. with ginseng (300 mg/kg) for 7 consecutive days prior to CCl4 injection. P. ginseng group (PG) was injected i.p. with ginseng (300 mg/kg) for 7 consecutive days. 2.4. Blood sampling and analysis All treated animals were anesthetized by ether inhalation for blood sample collection 24 h after administration of CCl4 . Blood samples were collected from hearts of rats using a syringe with 24-gauge needle under ether anesthesia. The samples were centrifuged at 3200 g for 10 min within 1 h after collection. The sera were stored in the −20 ◦ C freezer before they were analysed. Serum enzyme activities [aspartate aminotransferase (AST), alanine aminotransferase (ALT), gama-glutamil transferase (GGT)], biochemical parameters [urea, blood urea nitrogen (BUN), glucose, creatinine, total protein, calcium and phosphorus] and serum total triglyceride and cholesterol levels were analysed using diagnostic kits (IBL Chemical Co., Ankara, Turkey). 2.5. Histochemical and immunhistochemical examination The liver tissue samples were ﬁxed in 10% buffered neutral formalin, and embedded in parafﬁn. The parafﬁn blocks were cut 5–7 ␮m thick and stained with Mallory’s triple stain modiﬁed by Crossman. CD68 (universal marker for monocyte/macrophage lineage cells) and capase-3 (apoptotic marker) positive cells were determined with streptavidin-biotin-peroxidase staining method. For immunohistochemistry examinations were used monoclonal mouse anti-CD68 (Clone KP1, Invitrogen, 08-0125) and monoclonal caspase-3 (Biovision-3015-100, dilution: 1/25) primary antibodies and biotinylated secondary antibody (DAKO-Universal LSAB KitK0690). The binding sites of antibody were visualized with DAB (Sigma), and evaluated by high-power light microscopic (Nikon

3.1. Biochemical results The activities of AST, ALT and GGT were estimated in serum samples as the liver function markers. These results are given in Table 1. The CCl4 treatment markedly affected the liver speciﬁc enzymes. It was found that a signiﬁcant increase in serum AST, ALT and GGT activities of rats given alone CCl4 (p < 0.05). This result suggests that liver function markers are elevated in the serum due to release of the enzymes from damaged liver. However a signiﬁcant decrease was observed in above serum activities of rats given PG + CCl4 compared with the alone CCl4 treated groups (p < 0.05). The levels of serum biochemical parameters of rats in all groups are presented in Table 2. Cholesterol, triglyceride and glucose levels were increased in the CCl4 groups compared with the control group (p < 0.05). There were decreases in cholesterol, triglyceride and glucose levels in the PG + CCl4 group compared with the CCl4 treated groups (p < 0.05). A signiﬁcant change was not determined in calcium and phosphorus levels of control and all treated groups (p > 0.05). However, a noticeable increase in protein levels of rats given alone CCl4 was observed compared with control groups (p < 0.05). But, this increase was reduced in pretreated with PG compared with alone CCl4 treated group (p < 0.05). The level of kidney markers of rats in control and all treated groups are showed in Table 3. A signiﬁcant increases change in creatinine, urea and BUN levels of rats given CCl4 were determined (p < 0.05). On the contrary, a marked reduction was observed in the amount of creatinine, urea and BUN of rats given PG + CCl4 group compared with the alone CCl4 treated groups (p < 0.05). 3.2. Histochemical and Immunohistochemical ﬁndings The livers of the control and the PG-only groups were seen normal histological structure. However, a great amount of mononuclear cells inﬁltration, necrotic cells, steatozis and few ﬁbroblasts were signiﬁcantly determined in both lobular and portal areas in the CCl4 group. On the other hand, these changes were slightly decreased in the PG + CCl4 treated animals. Table 1 Effects of CCl4 and Panax ginseng on serum enzyme activities of liver in rat. Groups

Parameters AST U/L

C CCl4 PG + CCl4 PG

126.19 395.43 133.69 247.09

ALT U/L ± ± ± ±

10.30 20.35a 14.59b 20.24

67.63 258.46 77.67 66.46

GGT U/L ± ± ± ±

6.57 8.52a 7.15b 9.25

Each value represents the mean ± SEM of 6 animals. a Signiﬁcantly different from control group (p < 0.05). b Signiﬁcantly different from CCl4 group (p < 0.05). C: Control, PG: Panax ginseng, CCl4 : Carbon tetrachloride.

130.84 542.24 118.72 246.18

± ± ± ±

12.35 20.25a 14.26b 15.20

210

E. Karakus et al. / Journal of Hazardous Materials 195 (2011) 208–213

Table 2 Effects of CCl4 and Panax ginseng on serum biochemical parameters in rat. Groups

Parameters Cholesterol mg/dl

C CCl4 PG + CCl4 PG

145.56 264.98 113.17 200.77

± ± ± ±

10.25 15.24a 15.55b 16.48

Triglyceride mg/dl 169.62 320.62 145.76 216.44

± ± ± ±

15.25 12.35a 14.65b 12.05

Glucose mg/dl 218.73 357.6 223.36 294.43

± ± ± ±

10.98 12.58a 16.5b 14.56

Total protein g/dl 62.10 44.26 61.54 65.97

± ± ± ±

5.65 6.24a 7.21 6.58

Calcium mg/dl 13.56 11.85 14.12 13.65

± ± ± ±

3.56 4.58 5.32 5.78

Phosphorus g/dl 8.14 7.94 8.4 7.91

± ± ± ±

1.25 2.03 1.64 2.58

Each value represents the mean ± SEM of 6 animals. a Signiﬁcantly different from control group (p < 0.05). b Signiﬁcantly different from CCl4 group (p < 0.05). C: Control, PG: Panax ginseng, CCl4 : Carbon tetrachloride.

Fig. 1. Caspase-3 positive reactions in rat livers. (A) Control, (B) PG, (C) CCl4 , (D) PG + CCl4 Arrow heads: Caspase-3 positive apoptotic cells (streptavidin-biotin peroxidase staining), Bar 35 ␮m.

Fig. 2. CD68+ positive reactions in rat livers. (A) Control, (B) PG, (C) CCl4 , (D) PG + CCl4 Arrow heads: CD68+ positive phagocytic cells (streptavidin-biotin peroxidase staining), Bar 35 ␮m.

E. Karakus et al. / Journal of Hazardous Materials 195 (2011) 208–213 Table 3 Effects of CCl4 and Panax ginseng on the kidney markers in rat. Groups

Parameters Creatinine mg/dl

C CCl4 PG + CCl4 PG

0.44 1.80 0.53 0.68

± ± ± ±

Urea mg/dl

0.12 0.15a 0.08b 0.10

45.78 89.20 44.63 59.54

± ± ± ±

5.62 7.25a 6.32b 5.58

BUN mg/dl 16.81 79.69 14.02 44.42

± ± ± ±

5.25 4.87a 6.59b 8.25

Each value represents the mean ± SEM of 6 animals. a Signiﬁcantly different from control group (p < 0.05). b Signiﬁcantly different from CCl4 group (p < 0.05). C: Control, PG: Panax ginseng, CCl4 : Carbon tetrachloride.

Table 4 Semiquantitative analysis of caspase-3 and CD68+ reactivity in liver. Groups

C PG CCl4 PG + CCl4

Apoptotic and CD68 immune staining density Caspase-3 Liver

CD68 Liver

−/+ + +++ ++

+ + ++++ +++

Caspase-3 and CD68+ reaction density was estimated as follows: none = −, weak = +, moderate = ++, strong = +++, very strong = ++++. C: Control, PG: Panax ginseng, CCl4 : Carbon tetrachloride.

In this study, apoptotic and monocyte/macrophage lineage cells in the liver tissue were investigated with caspase-3 and CD68+ antibodies, respectively. The caspase-3 and CD68+ activities are showed in Figs. 1 and 2, respectively. Caspase-3 immune-reactive cells were high levels observed around of the central vein and lobuler areas in the CCl4 group (Fig. 1). We determined that the numbers of apoptotic cell were increased in the CCl4 group when compared with other all groups, but it was slightly decreased in the PG + CCl4 group (Table 4). In control and PG treated animals, slightly immune-reactive CD68+ cells were seen diffusely throughout the hepatic tissue. In liver of rats treated with CCl4 observed a signiﬁcant increased numbers of CD68+ cells when compared to PG + CCl4 group (Table 4). CD68+ cells were mainly localized within the damaged lobuler and portal area (Fig. 2). 4. Discussion Previous studies have reported that herbal medicines have a signiﬁcant contribution to the treatment of liver ﬁbrosis which seems to be related to their antioxidant potentials [14,15]. P. ginseng, a traditional Asia and Chinese herb, has been used to important roles in maintaining oxidative status, by possessing either direct or indirect antioxidant functions, and has been a component of effective formulations for treatment of liver disease [16]. Carbon tetrachloride is one of the most commonly used hepatotoxins in the experimental liver studies. CCl4 induced liver injuries are the best characterized system of xenobiotic-induced hepatotoxicity and commonly used model for the screening of hepatoprotective activities of drugs [17–19]. Liver ﬁbrosis induced by CCl4 is associated with the severity of lipid peroxidation and the depletion of antioxidant status which causing by damage in the cell membrane and the organelles of the hepatocyte [20,21]. In the present study, showed that serum AST, ALT and GGT activities which hepatic markers were greatly increased in rats with the CCl4 treatment alone in comparison with control group. The increased serum levels of hepatic markers have been attributed to the liver injury, because these enzymes are place in cytoplasmic area of the cell and are released into circulation in case of cellular damage [18,22]. Zimmerman et al. [23] stated that the CCl4

211

induced the increase of serum ALT and AST levels which source from cell membrane and mitochondrial damages in liver cells. There are many authors’ reports that these enzymes activities were signiﬁcantly elevated after CCl4 treatment [24–27]. The ﬁrst reports about of hepatotoxic effects by CCl4 , are lipid peroxidation origin, and are largely due to its active metabolite CCl3 (. This metabolite can abstract hydrogen from fatty acids, initiating the lipid peroxidation, lead to cell injury, and ﬁnally liver damage [28,29]. On the other hand, pretreatment with ginseng was found to signiﬁcant suppressed an increase in serum AST, ALT and GGT activities induced by CCl4 in rats. This ﬁnding implies that ginseng challenge to protect liver tissue from CCl4 injury. Current studies have provided a considerable support for evidencing the protective effects of ginseng on liver damage [30–34]. Also, these studies declared that the antioxidant properties of ginsenosides those phenolic acids, ﬂavonoids and saponins contribute to protection against CCl4 induced hepatotoxicity in rats. These compounds may be responsible for its hepatoprotective action by scavenge and destroy lipid peroxyl radicals and reactive oxygen species such as like the superoxide anion (O2 − ), the hydrogen peroxide (H2 O2 ) and the hydroxyl radical ((OH) [35]. In addition to the increased level of hepatic marker, the results of the present study have also established that the CCl4 treatment could have affected the lipid metabolism of liver (triglyceride and cholesterol levels), renal markers (creatinine, urea, BUN, levels) and glucose level in rats. This is evidenced from our observations that, CCl4 was caused a signiﬁcant increase in the levels of lipid parameters and renal markers, but it decreased to the total protein levels. Muller et al. [34] stated that CCl4 intoxication is similar the hepatitis in case of the triglycerides catabolism. This situation could be also attributed to the reduction of lipase activity, which could lead to decrease in triglyceride hydrolysis [35]. On the other hand, it can be assumed that hypercholesterimea in CCl4 intoxicated rats has resulted in damage of hepatic parenchymal cells that lead to disturbance of lipid metabolism in liver [36]. However, in group of pretreatment with ginseng showed a signiﬁcant decline in above parameters compared with CCl4 -intoxicated group. The mechanisms of lipid lowering effects of ginseng are mainly unknown, but recently ginseng sapogenins which produced from ginseng saponins, were shown to exert a strong inhibitory activity on microsomal acyl coenzyme A: cholesterol acyltransferease in vitro [37]. This enzyme is responsible for acylation of cholesterol to cholesterol esters in liver [38]. However, there are some authors reported that ginseng may have a supportive effect as a antiatherosclerotic agent by reducing elevated serum total cholesterol level and enhancing antioxidant capacity [39,40]. Also, Inoue et al. [41] stated that oral administration of ginseng saponins are decreased the elevated serum triglycerides and cholesterol levels in cyclophosphamide treated rabbit. This result indicates that ginseng or sapogenins might be affecting the pathway of cholesterol biosynthesis. The histological changes in the liver injury induced by CCl4 are know as apoptosis, necrosis, steatosis and mononuclear cell inﬁltration in both lobuler area and portal septa [6,42,43]. As similar with above reports, our ﬁndings were revealed high level inﬂammation, steatosis, necrosis, CD68 and caspase-3 positive cells within the lobuler areas in the CCl4 group. Caspases enzymes have remarkably a role on apoptotic pathway of endoplasmic reticulum induced by CCl4, various injuries and stress that these protein expressions are accelerated to cells death [7,44,45]. In agreement with above report, we observed an expansion of the caspase-3 positive cell population in CCl4 treated rats when compared with PG + CCl4 group. These ﬁndings indicated that the PG pretreated may contribute to inhibition of apoptosis There are many studies about of monoclonal antibody CD68 that showed as immunhistochemically in the mononuclear phago-

212

E. Karakus et al. / Journal of Hazardous Materials 195 (2011) 208–213

cyte lineage cells (monocytes and most macrophages) [7,45,46]. Numerously researchers had shown that growth of the Kupffer cells and other mononuclear phagocytes population had higher in the administration of CCl4 compare with control animals [46–49]. Released mediators from cytotoxic cells cause an increase in the numbers of activated phagocytic cell in hepatotoxic CCl4 treatment [47]. Edwards et al. [45] have shown that gadolinium chloride were protected from CCl4 -induced liver damage in rats that it extensively decreased inﬁltration of neutrophils and destroys mononuclear phagocytes resulting from exposure to CCl4 . In our experiments, the numbers of CD68 and caspase-3 positive cells were slightly decreased within the lobuler areas and portal tracts in the PG + CCl4 group when compared to CCl4 group, which showing anti-inﬂammatory effects and inhibition of hepatocyte apoptosis through PG. As shown in our results, CCl4 treated rats have much higher blood glucose levels than control rats. By ginseng treatment to CCl4 rats, however, blood glucose levels were declined to a certain degree. The main mechanism of hypoglycemic activity for P. ginseng is not clearly, but three possibilities mechanisms can be suggested that modulation of glucose transport [50], glucose disposal [51] and insulin secretion [52]. On the other hand, Ragunathan and Sulochana [53] stated that some components such as phenolics and ﬂavonoids are known to be responsible for hypoglycemic activity. The main signs of kidney damage induced by CCl4 treatment are the high levels of creatinine, urea and BUN in serum. In this study we showed that CCl4 treatment caused a noticeable injury in kidney functions. Serum creatinine, urea and BUN concentrations were signiﬁcantly higher in the CCl4 treated rats compared with control group. Khan et al. [54] stated that the CCl3 ( radical initiates the of lipid peroxidation which is supposed to the most important mechanism in the pathogenesis of kidney injury induced by CCl4 . Also, there are many authors reported that the nephrotoxic effects of CCl4 are connected with oxidative damage of the lipids and proteins in rat kidney tissue as well as humans [55–58]. However P. ginseng signiﬁcantly decreased the elevated levels of creatinine, urea and BUN in our study. Ginsenosides are present in the P. ginseng, which may have improved the kidney functions through different activity properties such as scavenging of reactive oxygen species and inhibition of the free radicals generation. Similar results were also documented that different plant extracts signiﬁcantly improved the kidney injuries induced by CCl4 treatment [59–61]. In conclusion, we found that P. ginseng caused a protective effect against CCl4 -induced liver damage and improved the biochemical parameters. Also, we showed that P. ginseng has a hepatoprotective effect against apoptosis, and increased CD68+ cell activation in the liver of CCl4 -treated rats. We suggest that P. ginseng may be used to protect against toxic effects of CCl4 and other chemical agents in liver. Acknowledgement The authors would like to express their gratitude to the Center of Experimental Research and Practice at the Atatürk University for providing the animals. References [1] V. Robins, S.L. Kumar, Basic Pathology, 4th ed., W.B. Saunder Co., Philadelphia, PA., 1987. [2] R. Gonzales, I. Corcho, D. Remirez, Hepatoprotective effects of propolis extract on carbon tetrachloride-induce liver injury in rats, Phytother. Res. 9 (1995) 114–117. [3] F.F. Behar-Cohen, S. Heydolph, V. Faure, M.T. Droy-Lefaix, Y. Courtois, O. Goureau, Peroxynitrite cytotoxicity on bovine retinal pigmented epithelial cells in culture, Biochem. Biophys. Res. Commun. 226 (1996) 842–849.

[4] S. Wasser, C.E. Tan, Experimental models of hepatic ﬁbrosis in the rat, Ann. Acad. Med. Singapore 28 (1999) 109–111. [5] G. Ramadori, B. Saile, Portal tract ﬁbrogenesis in the liver, Lab. Invest. 84 (2004) 153–159. [6] A. Canbay, A. Feldstein, E. Baskin-Bey, S.F. Bronk, G.J. Gores, The caspase inhibitor IDN-6556 attenuates hepatic injury and ﬁbrosis in the bile duct ligated mouse, J. Pharm. Exp. Ther. 3 (2004) 1191–1196. [7] B. Saile, G. Ramadori, Inﬂammation, damage repair and liver ﬁbrosis-role of cytokines and different cell types, Z. Gastroenterol. 45 (1) (2007) 77–86. [8] D. Zhang, T. Yasuda, Y. Yu, P. Zheng, T. Kawabata, Y. Ma, S. Okada, Ginseng extract scavenges hydroxyl radical and protects unsaturated fatty acids from decomposition caused by iron-mediated lipid peroxidation, Free Rad. Biol. Med. 20 (1996) 145–150. [9] T.K. Yun, S.Y. Choi, H.Y. Yun, Epidemiological study on cancer prevention by ginseng: are all kinds of cancers preventable by ginseng? J. Korean Med. Sci. 16 (2001) S19–S27. [10] S.S. Joo, T.J. Won, I. Lee do, Reciprocal activity of ginsenosides in the production of proinﬂammatory repertoire, and their potential roles in neuroprotection in vivo, Plan. Med. 71 (2005) 476–481. [11] C.H. Jung, H.M. Seog, I.W. Choi, H.D. Choi, H.Y. Cho, Effects of wild ginseng (Panax ginseng C.A. Meyer) leaves on lipid peroxidation levels and antioxidant enzyme activities in streptozotocin diabetic rats, J. Ethnopharm. 98 (2005) 245–250. [12] A.S. Attele, J.A. Wu, C.S. Yuan, Ginseng pharmacology: multiple constituents and multiple actions, Biochem. Pharm. 58 (1999) 1685–1693. [13] Y.C. Huang, C.T. Chen, S.C. Chen, P.H. Lai, H.C. Liang, Y. Chang, L.C. Yu, H.W. Sung, A natural compound (ginsenoside Re) isolated from Panax ginseng as a novel angiogenic agent for tissue regeneration, Pharm. Res. 22 (2005) 636–646. [14] M. Kaplan, E.A. Mutlu, M. Benson, J.Z. Fields, A. Banan, A. Keshavarzian, Use of herbal preparations in the treatment of oxidant-mediated inﬂammatory disorders, Complement. Ther. Med. 15 (3) (2007) 207–216. [15] C. Balsano, A. Alisi, V. Nobili, Liver ﬁbrosis and therapeutic strategies: the goal for improving metabolism, Curr. Drug Targets 10 (6) (2009) 505–512. [16] H.F. Chang, Y.H. Lin, C.C. Chu, S.J. Wu, Y.H. Tsai, J.C. Chao, Protective effects of Ginkgo biloba, Panax ginseng, and Schizandra chinensis extract on liver injury in rats, Am. J. Chin. Med. 35 (6) (2007) 995–1009. [17] R.O. Recknagel Jr., E.A. Glender, J.A. Dolak, R.L. Waller, Mechanism of carbon tetrachloride toxicity, Pharmacol. Ther. 43 (1989) 139–154. [18] J.A. Brent, B.H. Rumack, Role of free radicals in toxic hepatic injury II, Clin. Tox. 31 (1993) 173–196. [19] N. Brautbar, J. Williams 2nd., Industrial solvents and liver toxicity: risk assessment, risk factors and mechanisms, Int. J. Hyg. Environ. Health. 205 (2002) 479–491. [20] J. Liu, Y. Liu, C.D. Klaassen, The effect of Chinese hepatoprotective medicines on experimental liver injury in mice, J. Ethnopharmacol. 42 (3) (1994) 183–191. [21] M.H. Shyu, T.C. Kao, G.C. Yen, Hsian-tsao, (Mesona procumbens Heml.) prevents against rat liver ﬁbrosis induced by CCl4 via inhibition of hepatic stellate cells activation, Food Chem. Toxicol. 46 (2008) 3707–3713. [22] R.O. Recknagel Jr., E.A. Glender, R.S. Britton, Free radical damage and lipid peroxidation, in: R.G. Meeks (Ed.), Hepatotoxicology, CRC Press, Florida, 1991, pp. 401–436. [23] H.J. Zimmerman, Y. Kodero, M. West, Rate of increase of plasma level of cytoplasmic and mitochondrial enzyme in experimental CCl4 hepatotoxicity, J. Lab. Clin. Med. 66 (2) (1965) 315–323. [24] P.Y. Wang, T. Kaneko, H. Tsukada, M. Nakano, T. Nakajima, A. Sato, Time courses of hepatic injuries induced by chloroform and by carbon tetrachloride: comparison of biochemical and histopathological changes, Arch. Toxicol. 71 (10) (1997) 638–645. [25] G. Mehmetc¸ik, G. Özdemirler, N. Koc¸ak-Toker, U. C¸evikbas¸, M.M. Uysal, Role of carnosine in preventing thioacetamide-induced liver injury in the rat, Peptides 29 (2008) 425–429. [26] O.F. Arıcı, N. Cetin, Protective role of ghrelin against carbon tetrachloride (CCl4 induced coagulation disturbances in rats, Regul. Pept. 166 (2011) 139–142. [27] D.L. Tribble, T.Y. Aw, D.P. Jone, The pathophysiological signiﬁcance of lipid peroxidation in oxidative cell injury, J. Hepatol. 7 (1987) 377–386. [28] L.G. Forni, J.E. Packer, T.F. Slater, R.L. Willson, Reaction of the trichloromethyl radical and halothane derived peroxy radical with unsaturated fatty acids: a pulse radiolysis study, Chem. Boil. Interact. 85 (1983) 171–177. [29] W.H. Park, S.K. Lee, C.H. Kim, A Korean herbal medicine, Panax notoginseng, prevents liver ﬁbrosis and hepatic microvascular dysfunction in rats, Life Sci. 76 (2005) 1675–1690. [30] A. Karadeniz, A. Yıldırım, A. Karakoc¸, Y. Kalkan, F. C¸elebi, Protective effect of Panax ginseng on carbon tetrachloride induced liver, heart and kidney injury in rats, Rev. Med. Vet. 160 (2009) 237–243. [31] A. Karadeniz, M. Cemek, N. Simsek, The effects of Panax ginseng and Spirulina platensis on hepatotoxicity induced by cadmium in rats, Ecotoxicol. Environ. Saf. 72 (2009) 231–235. [32] Y.G. Li, D.F. Ji, S. Zhong, L.G. Shi, G.Y. Hu, S. Chen, Saponins from Panax japonicus protect against alcohol-induced hepatic injury in mice by up-regulating the expression of GPX3, SOD1 and SOD3, Alcohol Alcohol. 45 (2010) 320–331. [33] D. Kitts, C. Hu, Efﬁcacy and safety of ginseng, Public Health Nutr. 3 (2000) 473–485. [34] P. Muller, R. Fellin, J. Lambreacht, B. Agostini, H. Wicland, W. Rost, D. Sadel, Hypertriglyceridemia, secondary to liver disease, Eur. J. Clin. Invest. 4 (1974) 419–428. [35] C.E. Jahn, E.J. Schaegfetr, L.A. Taam, J.H. Hoofnagle, F.T. Lindgren, J.J. Albers, E.A. Jones, H.B. Brever, Lipoprotein abnormalities in primary biliary cirrhosis

E. Karakus et al. / Journal of Hazardous Materials 195 (2011) 208–213

[36] [37]

[38] [39]

[40]

[41]

[42]

[43] [44] [45]

[46]

[47]

[48]

association with hepatic lipase inhibition as well as altered cholesterol esteriﬁcation, Gastroenterology 89 (1985) 1266–1278. R.J. Havel, Functional activities of hepatic lipoproteins receptors, Ann. Rev. Physiol. 48 (1986) 119–134. B.M. Kwon, M.K. Kim, N.I. Baek, D.S. Kim, J.D. Park, Y.K. Kim, H.K. Lee, S.I. Kim, Acyl-CoA: cholesterol acyltransferase inhibitory activity of ginseng sapogenins, produced from the ginseng saponins, Bioorg. Med. Chem. Lett. 9 (1999) 1375–1378. K. Matsuda, ACAT inhibitors as antiatherosclerotic agents: compounds and mechanisms, Med. Res. Rev. 14 (1994) 271–305. Y. Liu, H.G. Zhang, Y. Jia, X.H. Li, Panax notoginseng saponins attenuate atherogenesis accelerated by zymosan in rabbits, Biol. Pharm. Bull. 33 (8) (2010) 1324–1330. J. Li, Z.Z. Xie, Y.B. Tang, J.G. Zhou, Y.Y. Guan, Ginsenoside-Rd, a puriﬁed component from Panax notoginseng saponins, prevents atherosclerosis in apoE knockout mice, Eur. J. Pharmacol. 652 (1-3) (2011) 104–110. M. Inoue, C.Z. Wu, D.Q. Dou, Y.J. Chen, Y. Ogihara, Lipoprotein lipase activation by red ginseng saponins in hyperlipidemia model animals, Phytomedicine 6 (1999) 257–265. R.V. Rao, E. Hermel, S. Castro-Obregon, G. del Rio, L.M. Ellerby, H.M. Ellerby, D.E. Bredesen, Coupling endoplasmic reticulum stress to the cell death program. Mechanism of caspase activation, Jour. Biol. Chem. 36 (2001) 33869–33874. S. Friedman, Mac the knife? Macrophages – the double-edged sword of hepatic ﬁbrosis, J. Clin. Invest. 1 (2005) 29–32. M.H. Kang, C.P. Reynolds, Bcl-2 inhibitors: targeting mitochondrial apoptotic pathways in cancer therapy, Clin. Cancer Res. 15 (2009) 1126–1132. M.J. Edwards, B.J. Keller, F.C. Kauffman, R.G. Thurman, The involvement of Kupffer cells in carbon tetrachloride toxicity, Toxicol. Appl. Pharmacol. 119 (1993) 275–279. C. Orﬁla, J.C. Lepert, L. Alric, G. Carrera, M. Beraud, J.P. Vinel, B. Pipy, Expression of TNF-alpha and immunohistochemical distribution of hepatic macrophage surface markers in carbon tetrachloride-induced chronic liver injury in rats, Histochem. J. 31 (1999) 677–685. L.A. Morio, H. Chiu, K.A. Sprowles, P. Zhou, D.E. Heck, M.K. Gordon, D.L. Laskin, Distinct roles of tumor necrosis factor-alpha and nitric oxide in acute liver injury induced by carbon tetrachloride in mice, Toxicol. Appl. Pharmacol. 172 (2001) 44–51. S.B. Yee, P.E. Ganey, R.A. Roth, The role of Kupffer cells and TNF-alpha in monocrotaline and bacterial lipopolysaccharide-induced liver injury, Toxicol. Sci. 71 (2001) 124–132.

213

[49] M.D. Wheeler, Endotoxin and Kupffer cell activation in alcoholic liver disease, Alcohol. Res. Health 27 (2003) 300–306. [50] K. Yamasaki, C. Murakami, K. Ohtani, R. Kasai, T. Kurokawa, S. Ishibashi, Effects of standardized Panax ginseng extract G115 on the d-glucose transport by Ehrlich ascites tumour cells, Phytother. Res. 7 (1993) 200–202. [51] T. Yokozawa, T. Kobayashi, A. Kawai, H. Oura, Y. Kawashima, Stimulation of lipid and sugar metabolism in ginsenoside-Rb2 treated rats, Chem. Pharm. Bull. 32 (1984) 2766–2772. [52] I. Waki, H. Kyo, M. Yasuda, M. Kimura, Effects of a hypoglycemic component of ginseng radix on insulin biosynthesis in normal and diabetic animals, J. Pharmacobiodyn. 5 (1982) 547–554. [53] V. Ragunathan, N. Sulochana, A new ﬂavonol bioside from the ﬂowers of Hibiscus vitifolius Linn. and its hypoglycemic activity, J. Indian Chem. Soc. 71 (1994) 705–706. [54] R.A. Khan, M.R. Khan, S. Sahreen, J. Bokhari, Prevention of CCl4-induced nephrotoxicity with Sonchus asper in rat, Food Chem. Toxicol. 48 (8–9) (2010) 2469–2476. [55] A.J. Perez, M. Courel, J. Sobrado, L. Gonzalez, Acute renal failure after topical application of carbon tetrachloride, Lancet 1 (1987) 515–516. [56] P. Manna, M. Sinha, P.C. Sil, Aqueous extract of Terminalia arjuna prevents carbon tetrachloride induced hepatic and renal disorders, B.M.C. Complement. Altern. Med. 6 (2006) 33. [57] P. Abraham, G. Wilfred, S.P. Catherine, Oxidative damage to the lipids and protein in the lungs, testis and kidney of rats during carbon tetrachloride intoxication, Clin. Chim. Acta. 289 (1999) 177–179. [58] M. Ogeturk, I. Kus, N. Colakoglu, I. Zararsiz, N. Ilhan, M. Sarsilmaz, Cafeic acid phenethyl ester protects kidneys against carbon tetrachloride toxicity in rats, J. Ethnopharm. 97 (2005) 273–280. [59] T. Jayakumar, M. Sakthivel, P.A. Thomas, P. Geraldine, Pleurotus ostreatus, an oyster mushroom, decreases the oxidative stress induced by carbon tetrachloride in rat kidneys, heart and brain, Chem. Biol. Interact. 176 (2-3) (2008) 108–120. [60] A.P. Manjrekar, V. Jisha, P.P. Bag, B. Adhikary, M.M. Pai, A. Hegde, M. Nandini, Effect of Phyllanthus niruri Linn. treatment on liver, kidney and testes in CCl4 induced hepatotoxic rats, Indian J. Exp. Biol. 46 (7) (2008) 514–520. [61] T.H. Khan, S. Sultana, Antioxidant and hepatoprotective potential of Aegle marmelos Correa. against CCl4-induced oxidative stress and early tumor events, J. Enzyme Inhib. Med. Chem. 24 (2) (2009) 320–327.

Journal of Hazardous Materials 195 (2011) 214–222

Contents lists available at ScienceDirect

Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat

Photocatalytic degradation of dimethoate using LbL fabricated TiO2 /polymer hybrid ﬁlms D. Neela Priya a , Jayant M. Modak b , Polonca Trebˇse c , Romina Zˇ abar c , Ashok M. Raichur a,∗ a

Department of Materials Engineering, Indian Institute of Science, Bangalore 560012, India Department of Chemical Engineering, Indian Institute of Science, Bangalore 560012, India c Laboratory for Environmental Research, University of Nova Gorica, Vipavska 13, P.O. Box 301, 5000 Nova Gorica, Slovenia b

a r t i c l e

i n f o

Article history: Received 20 April 2011 Received in revised form 6 August 2011 Accepted 9 August 2011 Available online 16 August 2011 Keywords: Layer-by-layer method Photocatalysis TiO2 Dimethoate Immobilized catalyst

a b s t r a c t Degradation of dimethoate under UV irradiation using TiO2 /polymer ﬁlms prepared by the layer-by-layer (LbL) method was investigated. The thin ﬁlms were fabricated on glass slides and the surface morphology and roughness of the thin ﬁlms were characterized using X-ray diffraction (XRD), scanning electron microscopy (SEM) and atomic force microscopy (AFM). The effect of lamp intensity, catalyst loading in the layers, number of bilayers, pH and initial dimethoate concentration on the degradation of dimethoate was systematically studied. The degradation was monitored using high performance liquid chromatography (HPLC) analysis and total organic carbon (TOC) measurements as a function of irradiation time, to see the change in concentration of dimethoate and mineralization, respectively. Complete degradation of dimethoate was achieved under TiO2 optimum loading of 4 g/L at an UV irradiation time of 180 min. Increase in the lamp intensity, catalyst loading and number of bilayers increased the rate of degradation. At a pH of 4.62, complete degradation of dimethoate was observed. The degradation efﬁciency decreased with increase in initial dimethoate concentration. The degradation byproducts were analyzed and conﬁrmed by gas chromatography–mass spectra (GC–MS). Toxicity of the irradiated samples was measured using the luminescence of bacteria Vibrio ﬁscheri after 30 min of incubation and the results showed more toxicity than the parent compound. Catalyst reusability studies revealed that the fabricated thin ﬁlms could be repeatedly used for up to ten times without affecting the photocatalytic activity of the ﬁlms. The ﬁndings of the present study are very useful for the treatment of wastewaters contaminated with pesticides. © 2011 Elsevier B.V. All rights reserved.

1. Introduction In the last two decades, organophosphorous pesticides have been extensively used worldwide in agriculture as an alternative to organochlorides. Dimethoate is a systemic organophosphorous pesticide that is widely used to kill mites and insects on contact. In India, it is one of the commonly used pesticides besides monocrotophos, phorate, phosphamidon and methyl parathion [1]. It is considered as ‘moderately hazardous, class II’ compound by World Health Organization (WHO) and the maximum permissible limit of dimethoate in drinking waters is 0.006 mg/L [2]. Considering the toxicity of this compound, there is an urgent need to develop an effective treatment method to remove the compound from water. A review on organophosphorous pesticides and the breakdown products, which are stated to be very harmful and toxic when compared to the parent chemical, discusses their fate in

∗ Corresponding author. Tel.: +91 80 22933238; fax: +91 80 23600472. E-mail address: [email protected] (A.M. Raichur). 0304-3894/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.jhazmat.2011.08.030

the aquatic environment via several processes [3]. Conventional methods for the removal of pesticides from wastewaters include adsorption [4], ozonation [5], reverse osmosis [6], nanoﬁltration [7], etc. These methods, however, possess certain limitations, likeozone has relatively low solubility and stability in water. Also the cost of production of ozone is very high [8]. Whereas, a competitive adsorption of natural organic matter (NOM) reduces the efﬁciency of the carbon ﬁlters and therefore limits its usage [9]. Hence, in order to protect the environment and to meet the stringent enforcement regulations, many researchers are developing effective, reliable and economical way for pesticide-containing water treatment system. Owing to the non-biodegradable nature of these pollutants, the chemical oxidation or biological treatment methods are not effective as these are unable to mineralize all organic substances [10]. Advanced oxidation process (AOP’s), like photocatalysis using semiconductor catalyst such as TiO2 , has gained interest of researchers recently due to its ability to destruct organic compounds [11]. TiO2 is widely used as a photocatalyst because of its availability, stability, low cost, and favorable band gap energy [12].

D.N. Priya et al. / Journal of Hazardous Materials 195 (2011) 214–222

Several studies on the photocatalytic degradation of dimethoate in aqueous solutions using nanosized TiO2 have been reported in literature [13,14]. The degradation products as well as reaction pathways have also been evaluated [14–17]. However, it is known that any wastewater puriﬁcation process using catalyst in its suspended form posses several disadvantages related to ﬁltration and reuse of the catalyst. As a result, processes employing catalyst in suspension form directly incurs high treatment costs making such process economically unviable [18]. These problems can be overcome by immobilizing TiO2 particles as thin ﬁlms on some substrates. Catalyst can be immobilized with various techniques such as dip coating, sol–gel synthesis, chemical vapor deposition method, spin coating, sputtering technique, layer-by-layer (LbL) assembly multilayer technique, etc. [19–24]. The photocatalytic degradation of organophosphorus pesticides: dichlorvos, monocrotophos, phorate and parathion using TiO2 thin ﬁlms was studied [25]. Till date, very few studies reported the photodegradation of dimethoate using TiO2 thin ﬁlms [26–28]. These studies used the sol–gel method to immobilize TiO2 and the substrates used were silica gel particles and PVF ﬁlms in comparison to our study that is based on electrostatic interactions. Also, most of the fabricating methods need either high temperatures (as in CVD) or post treatment (as in sol–gel synthesis) of the coated samples. On the other hand, the LbL method makes use of biodegradable or biocompatible polymers and one can achieve controlled thickness without the use of expensive equipment. Further the deposition of particles is independent of size or shape of substrate and is very environment friendly. The LbL fabricated thin ﬁlms showed to possess a great potential for the degradation of chemical contaminants [29]. The signiﬁcance of LbL method for the photodegradation of Rhodamine B, in terms of using a strong polyelectrolyte system such as PSS/TiO2 , its stability and applicability, compared to other fabrication methods were reported in detail in our previous study [29]. In the present study, the photocatalytic degradation of dimethoate with LbL immobilized commercial Degussa Aeroxide TiO2 P25 catalyst on a glass substrate is investigated. To our knowledge, there are no reports on the photocatalytic degradation of dimethoate with the catalyst immobilized by LbL technique. The effect of various operating parameters like lamp intensity, catalyst loading, number of bilayers, pH and initial dimethoate concentration on degradation has been investigated. The degradation products were identiﬁed by GC–MS and the treated solutions were tested for their toxicity. In addition, the present study highlights the reusability of the catalyst to see its performance efﬁciency in degrading dimethoate.

other chemicals used in the present work were of AR grade. Stock solution of 1000 ppm dimethoate was prepared and refrigerated at 4 ◦ C. HPLC-graded organic solvents were used for HPLC analysis. 2.2. Apparatus The catalyst was characterized by X-ray diffraction (X’Pert Pro, PANalytical instruments). A T60U Spectrophotometer (PG Instruments Ltd., UK) was used for the analysis of the samples. The surface morphology of the thin ﬁlms was examined using a ﬁeld emission scanning electron microscope (FEI-SIRION, Eindhoven, The Netherlands). AFM images were obtained using a MFP-3D-SA atomic force microscope (Asylum Research, USA). HPLC analysis was performed using a Hewlett Packard/Agilent 1100 series HPLC analyzer. Total organic carbon (TOC) was measured using a Multi N/C® 3100 TOC analyzer (Germany). The GC–MS analysis was performed using a Shimadzu QP5050A analyzer. The toxicity tests were conducted using a LUMISTox Dr Lange Analyzer. 2.3. Photocatalytic reactor The photoreactor consists of a jacketed quartz tube with dimensions of 3.4 cm inner diameter, 4 cm outer diameter, and 21 cm length and the reaction vessel was a 100 mL beaker (Qualigens, Mumbai, India). The UV source used in this study was a 125 W and 400 W mercury vapor lamp. The outer glass shell was removed and placed inside the quartz tube for use. The ballast and capacitor were connected in series with the lamp to avoid ﬂuctuations in the input power supply. Submersible water pump was used to circulate water through the jacket of the quartz tube to avoid heating caused by dissipative loss of UV light. The reactant solution was taken in the reaction vessel and placed 10 cm away from the quartz tube. The whole reactor setup was enclosed in a wooden box. The schematic of photoreactor set up is shown in Fig. 1. 2.4. Experimental procedure The catalyst was immobilized on clean glass slides by LbL technique. The polyelectrolyte solutions, PSS and PAH each of concentration 1 g/L, and the TiO2 solution were prepared in deionized (DI) water. For immobilizing the substrates using LbL technique, glass slides and silicon wafers were ﬁrst cleaned by sonicating for

2. Experimental 2.1. Materials Dimethoate (O,O-dimethyl S-[2-(methylamino)-2-oxoethyl] phosphorodithioate) technical grade (90% pure), was supplied by Hyderabad Chemicals Limited, Hyderabad, India. Commercial Degussa Aeroxide P25 TiO2 (80% of anatase and 20% of rutile, average particle size: 21 nm, speciﬁc surface area: 50 ± 15 m2 /g) (Degussa AG, Germany) was used as the photocatalyst. Poly(styrene sulfonate) sodium salt (PSS, MW = 70,000 g/mol) (Sigma–Aldrich, USA) and poly (allylamine hydrochloride) (PAH, MW = 70,000 g/mol) (Sigma–Aldrich, USA) were used for thin ﬁlm preparation. Microscopic glass slides (25.4 mm × 63.5 mm) were used as substrates for the ﬁlm deposition. Millipore water (resistivity: 18 M) was used throughout the work. Sulfuric acid, ascorbic acid, ammonium molybdate and potassium antimonyl tartarate used for the preparation of combined reagent and all

215

Fig. 1. The schematic of photoreactor set up.

216

D.N. Priya et al. / Journal of Hazardous Materials 195 (2011) 214–222

10 min in a solution containing 2:1(v/v) ratio of iso-propanol and water mixture. PSS and PAH each of concentration 1 g/L in DI water were prepared and adjusted to the desired pH. Water pH for rinsing purpose was adjusted to that of the polyelectrolytes and TiO2 . To deposit TiO2 on glass substrates and silicon wafers, a colloidal solution of TiO2 was made in DI water and pH was adjusted such that it was stable during deposition and at the same time it had the desired charge to be suitable for LbL technique. Initially, one precursor monolayer of PAH was deposited by immersing the substrates in a solution of PAH for 15 min to reverse the charge on the substrate followed again by rinsing with water for 3 min. The alternate multilayers layers of PSS/TiO2 were then assembled by repeatedly immersing the substrates in PSS and TiO2 solutions. The ﬁnal layer was TiO2 in all the experiments. In all the catalyst immobilization experiments TiO2 , PAH and PSS were deposited at pH 2.5. In our previous work using the LbL-TiO2 thin ﬁlms for the photodegradation of Rhodamine B [29], we showed the effectiveness of LbL method in comparison to methods like drop casting and spin coating in terms of catalyst immobilization and photodegradation. The SEM images of the catalysts prepared by these three methods are shown in supplementary information (Fig. SI 1). Dimethoate solution (100 mL of 10 ppm, pH 4.62, unless otherwise stated) with appropriate amount of catalyst deposited on the slides was taken in the reaction vessel. The immobilized catalyst slides were placed vertically in the solution along the sides of the beaker. The solution was stirred continuously using a magnetic stirrer for the complete mixing of catalyst and the solution during the course of reaction. Before starting the degradation experiment, the catalyst-dimethoate solution was stirred for 60 min in dark to establish the adsorption–desorption equilibrium. The concentration of dimethoate solution after the adsorption–desorption equilibrium was analyzed by HPLC (SI Fig. SI 2). It was observed that the concentration of dimethoate after 60 min was same as the initial concentration and was not affected. Hence, further experiments to establish adsorption–desorption equilibrium were performed for 30 min in dark. At regular irradiation time intervals, 2 mL of the sample was collected for HPLC and UV–vis spectrophotometer analysis. The combined reagent was added to the samples analyzed using UV–vis spectrophotometer. Stoichiometrically, degradation of each mole of dimethoate releases one mole of phosphate. Therefore, the concentration of dimethoate degraded was indirectly calculated from the phosphate remaining, using the standard phosphate calibration curve. In order to check the reproducibility of results, all the degradation experiments were repeated at least 2 times and the experimental error was found to be within 5%. The percentage of dimethoate remaining in the solution was calculated from the following expression: %dimethoate remaining =

Ct × 100 C0

and Ct = C0 − Pt ; where Ct is the concentration of dimethoate remaining at time t, Pt is the amount of phosphate present after time t and C0 is the original concentration of dimethoate. 2.4.1. GC–MS analysis The GC–MS analysis of dimethoate samples was performed using a Shimadzu QP5050A attached with Shimadzu GC-17A with a split injector. The GC–MS was equipped with a Chrompack WCOT FS CP-Sil 5CB column of 25 m length and 0.25 i.d. Helium was used as the carrier gas (at a ﬂow rate of 38.9 mL/min) with a column ﬂow of 1.3 mL/min. The solution samples were ﬁltered through a 0.45 ␮m ﬁlter, prior to analysis. The chromatographic conditions used for the identiﬁcation of intermediates were: initial injector temperature of 250 ◦ C with an interface temperature of 230 ◦ C. The oven temperature was programmed from 50 ◦ C to 250 ◦ C at a rate of 5 ◦ C/min. The interface temperature of the MS was set at 230 ◦ C

and the spectrum was obtained at a scan range from m/z 40 to 500. The total scan time was 30 min. 2.4.2. HPLC analysis Dimethoate concentration was analyzed using a reverse-phase liquid chromatography equipped with diode array detector (DAD) with Zorbax C8 column (250 mm × 4.6 mm, 5 ␮m particle size) at 25 ◦ C, using an isocratic separation with mobile phase of water and acetonitrile (60:40, v/v) at a ﬂow rate of 1 mL/min. At speciﬁc time intervals, samples were withdrawn from the reactor and ﬁltered through a 0.45 ␮m ﬁlter prior to analysis. The injection volume was 75 ␮L and the detection was realized at 210 nm, which was identiﬁed on the basis of retention time comparison with authentic standard. 2.4.3. Toxicity tests The toxicity of dimethoate samples collected before and after different irradiation time intervals were examined by means of a LUMISTox 300 luminometer (Dr Lange GmbH, Dusseldorf, Germany). The reagents and Vibrio ﬁscheri NRRL-B-11177 luminescent bacteria supplied by the same manufacturer were used. The inhibition of the luminescence of bacteria V. ﬁscheri after 30 min of incubation was determined. NaCl, 2 vol.% was added to all samples and the pH was adjusted to a value between 6.8 and 7.2 by addition of HCl or NaOH, respectively. After the reactivation of the liquid-dried bacteria for 15 min at 15 ◦ C, 0.5 mL of the bacteria suspension was mixed with 0.5 mL of dimethoate samples of different irradiation times. The sample in the measuring cell was incubated for 30 min at 15 ◦ C. The inhibition of the luminescence was measured in comparison to a blank sample of 0.5 mL NaCl (2%) solution and 0.5 mL bacteria suspension. The percentage of luminescence inhibition was calculated for each concentration relative to the control. 3. Results and discussion 3.1. Catalyst characterization studies The mean crystallite size of the catalyst determined by XRD patterns using Scherrer formula is 20 ± 2 nm, which is close to the data supplied by the manufacturer. Fig. 2 shows the X-ray diffraction patterns of PSS/TiO2 thin ﬁlms with increase in TiO2 loading (Fig. 2a) and bilayers (Fig. 2b). It was observed that the peaks corresponding to the anatase phase have appeared at 2 = 25.33 in all cases. Further, the peaks were getting sharper with increase in catalyst loading and number of bilayers. But there was no remarkable difference in the peak intensity of TiO2 after a loading of 4 g/L and appears to be the same with further increase in catalyst loading from 6 g/L to 10 g/L. On the other hand, with increase in number of bilayers from 1 to 10, a difference in peak intensity was noticed. This observation implies that higher intensity of TiO2 peak is due to the presence of more amount of TiO2 available for the degradation of dimethoate making the process favorable for degradation. The LbL-TiO2 thin ﬁlms were characterized by SEM and AFM to evaluate the surface morphology and effectiveness of the multilayer assembly technique. The thickness of ﬁlms with increasing number of layers 1, 3, 5 and 10 was measured from the cross sections of these ﬁlms using SEM. For clear imaging, the multilayer thin ﬁlms were assembled on silicon wafers at pH 2.5 using 4 g/L TiO2 suspension. The thickness of the ﬁlms was found to be 125 nm, 730 nm, 1.563 ␮m and 3.268 ␮m for 1-, 3-, 5- and 10-layered ﬁlms respectively. Thickness of the layers is also dependent on the concentration of TiO2 used. It was observed that initial layers were not uniform but as the number of layers increased, the surface of the ﬁlm looks homogeneous and smooth. Interestingly, the increase in thickness for each layer was approximately linear. Fig. 3a shows

D.N. Priya et al. / Journal of Hazardous Materials 195 (2011) 214–222

217

tional view SEM images of the PAH/(PSS/TiO2 )5 thin ﬁlm showing well-deﬁned porous multilayers. The PSS/TiO2 multilayer thin ﬁlms were further examined by AFM to determine the surface roughness. PAH/(PSS/TiO2 )n ﬁlms were deposited with 1, 3 and 5 bilayers using 4 g/L TiO2 and PAH, PSS concentrations of 1 g/L. Fig. 4a–c shows the 3D images for the three samples with PAH/(PSS/TiO2 )1 , PAH/(PSS/TiO2 )3 and PAH/(PSS/TiO2 )5 , respectively. It was observed that the roughness of the ﬁlms for 1, 3 and 5 bilayers were 71, 86 and 92 nm, respectively. The roughness increased with increase in the number of bilayers, which is due to the increase in the amount of TiO2 deposited per each layer.

Fig. 2. XRD images of LbL-TiO2 ﬁlms with increase in (a) TiO2 loading and (b) number of bilayers.

the top view surface morphology of a PAH/(PSS/TiO2 )5 thin ﬁlm examined by ﬁeld emission scanning electron microscope. A close view surface of the ﬁlm shows a ﬂat and dense surface morphology of polyelectrolytes and uniformly distributed TiO2 nanoparticles embedded in the ﬁlm. The ﬁlm shows sufﬁcient porous structure to permit free diffusion of pollutant in and out of the ﬁlm and these porous structures can effectively trap light into the inner layers. Also, the optical absorbance as measured by the UV–vis spectra (data not shown) increases linearly with ﬁlm thickness showing that the ﬁlm is accessible to light. Fig. 3b shows the cross sec-

3.1.1. Comparison of LbL-TiO2 ﬁlms with other ﬁlms The ability of immobilized catalysts prepared by LbL-TiO2 ﬁlms in comparison with ﬁlms prepared by drop casting and spin coating methods is studied. Fig. SI 1 shows the SEM images of PAH/(PSS/TiO2 )5 deposited with TiO2 loading of 4 g/L and PELs of 1 g/L each at deposition pH of 2.5. It is very clear from these images that ﬁlms prepared by drop casting (Fig. SI 1a) and spin coating (Fig. SI 1b) methods do not show uniform surface morphology when compared to LbL-TiO2 ﬁlms (Fig. SI 1c) deposited under the same experimental conditions. It can also be seen that polyelectrolytes and TiO2 are aggregated as clusters in case of ﬁlms assembled using drop casting and spin coating methods. On the other hand, the SEM image of LbL-TiO2 ﬁlm illustrates a smooth surface morphology and a complex network of cross-linked polyelectrolytes with TiO2 nanoparticles uniformly distributed in the multilayer thin ﬁlms. The image also shows a high degree of porosity, which plays an important role in the photocatalytic activity, as larger surface area provides for more photocatalysis reactions to take place. The efﬁciency of these highly stable LbL-TiO2 ﬁlms in terms of photodegradation (studied in our previous paper but data not reported here) have also proven their ability to degrade the pollutant for number of cycles with the same efﬁciency as compared to the other two ﬁlms. 3.2. Effect of lamp intensity The effect of lamp intensity on degradation of dimethoate was studied with an UV irradiation source of 125 and 400 W lamp. Prior to the experiment, O-nitrobenzaldehyde actinometry experiment was performed to calculate the light intensity of the UV power source and the values were found to be 12 mW/cm2 and

Fig. 3. (a) Top view SEM image of PAH/(PSS/TiO2 )5 and (b) cross sectional SEM image of PAH/(PSS/TiO2 )5 deposited by LbL method.

218

D.N. Priya et al. / Journal of Hazardous Materials 195 (2011) 214–222

100

% dimethoate remaining

80

60

40

1 g/L, 125 watt 2 g/L, 125 watt 1 g/L, 400 watt 2 g/L, 400 watt

20

0 0

60

120

180

240

300

Time (min) Fig. 5. Effect of lamp intensity on degradation of dimethoate. In all the experiments, 5 slides were immobilized having 5 bilayers on each.

remaining was 65% and 45%, respectively. However, under the same set of conditions with 400 W lamp, the amount of dimethoate remaining after 300 min of irradiation was 31.7% and 5.1%, respectively. With increase in lamp intensity increase in the rate of degradation was observed. This is due to the increase in number of photons generated with 400 W lamp leading to more reactive species generation and more degradation of dimethoate. The observed results are in agreement with the earlier studies [13]. Since the degradation was more effective with a 400 W lamp as compared to a 125 W lamp, further experiments were carried out with a 400 W lamp that would yield better degradation of dimethoate. 3.3. Effect of catalyst loading The effect of catalyst loading on dimethoate degradation was studied by varying TiO2 concentration from 1 g/L to 10 g/L with ﬁve slides immobilized with PAH/(PSS/TiO2 )5 and subjecting for an irradiation time of 300 min with a 400 W lamp. The results are shown in Fig. 6. Photolysis alone (without catalyst) had no effect on degra-

100

Fig. 4. AFM images of LbL-TiO2 thin ﬁlms with (a) 1 bilayer of PAH/(PSS/TiO2 )1 , (b) 3 bilayers of PAH/(PSS/TiO2 )3 and (c) 5 bilayers of PAH/(PSS/TiO2 )5 .

31.6 mW/cm2 , respectively for 125 W and 400 W lamps. The experiment on effect of lamp intensity was then performed with ﬁve numbers of immobilized catalysts (PAH/(PSS/TiO2 )5 ) each with a TiO2 loading of 1 g/L and 2 g/L, for an irradiation time of 300 min. The results are shown in Fig. 5. It was observed that, at TiO2 loading of 1 g/L and 2 g/L with 125 W lamp, the amount of dimethoate

% dimethoate remaining

80

60

40

Photolysis 1 g/L 2 g/L 4 g/L 6 g/L 8 g/L 10 g/L

20

0 0

60

120

180

240

300

Time (min) Fig. 6. Effect of catalyst loading on degradation of dimethoate, with 400 W lamp. In all the experiments, 5 slides were immobilized having 5 bilayers on each.

D.N. Priya et al. / Journal of Hazardous Materials 195 (2011) 214–222

3.4. Effect of number of bilayers Fig. 7 shows the effect of number of bilayers on degradation of dimethoate. Experiments were carried out by varying the bilayers from 1 to 10, at an optimum catalyst loading of 4 g/L and at a lamp intensity of 400 W. It was observed that with PAH/(PSS/TiO2 )1 , the amount of dimethoate remaining in the solution was 22.5% for the irradiation time of 300 min. On the other hand, when the number of bilayers was increased from 1 to 5, the amount of dimethoate remaining decreased from 22.5% to 19.6% and 0% with 3 and 5 bilayers, respectively. Further, when the bilayers were increased to 10, it was observed that dimethoate was completely degraded with in 180 min. Increase in dimethoate degradation with increase in the number of bilayers can be attributed mainly to two reasons – the ﬁrst being more amount of catalyst deposited per each layer, i.e., as the number of layers increase, more TiO2 gets immobilized in the multilayers which in turn increases the degradation efﬁciency due to more catalyst available for reaction. The second is the porous structure of the ﬁlm (as evident from the SEM images Fig. 3a and b where

1 bilayer 3 bilayers 5 bilayers 10 bilayers

100

80

% dimethoate remaining

dation. It was observed that with increase in catalyst loading, the percentage of dimethoate remaining in the solution decreased from 100% to 31.7%, 5.1% and 0%, respectively for 1 g/L, 2 g/L and 4 g/L. Complete degradation of dimethoate was achieved with a catalyst loading of 4 g/L at an irradiation time of 300 min. However, any further increase in the catalyst loading than 4 g/L reduced the rate of degradation. This indicates that optimum loading for immobilized catalyst was achieved with 4 g/L TiO2 . The increase in percentage degradation of dimethoate with increase in TiO2 concentration up to 4 g/L is due to the availability of more amount of catalyst for the reaction. The sudden decrease in the percentage degradation with further loading from 6 g/L to 10 g/L is attributed mainly to the light scattering effect as reported in the earlier studies [13,30]. Though the ﬁlms are homogeneous, the coating becomes turbid with increase in concentration after certain number of layers. When all the dimethoate molecules are adsorbed on TiO2 , the addition of higher quantities would also have no effect on the degradation efﬁciency. And this is the reason why at high catalyst concentration, in spite of more amount of catalyst available for reaction, the degradation efﬁciency decreases due to the reduced transparency of the ﬁlm which proportionately lowers the photocatalytic activity. This was also well supported by the XRD patterns of the ﬁlms (Fig. 2a) where there was negligible difference in peak intensity with further increase in TiO2 loading from 6 g/L to 10 g/L. Catalyst loading beyond this point was not signiﬁcant and resulted in decreased degradation efﬁciency. Hence, we considered 4 g/L TiO2 as the optimum loading concentration that would efﬁciently degrade dimethoate and further experiments were performed at this loading concentration. A comparison was made on the optimum amount of TiO2 required in powder form and that for thin ﬁlm deposition for the complete degradation of dimethoate. The amount of catalyst deposited by the LbL method was roughly calculated from the difference between the weight of the empty slide and the weight after 10 bilayers of TiO2 deposition. The optimal amount of TiO2 required in powder form as reported by Evgenidou et al. [14] and Chen et al. [13] was 0.1 g/L and 0.6 g/L for the complete degradation of a 10 ppm and 50 ppm dimethoate solution respectively. On the other hand, using the LbL technique, the average deposition of TiO2 per each layer was ∼0.33 mg and on each slide was ∼3.69 mg. Hence the amount of TiO2 required for the preparation of catalyst with 10 bilayers and 5 numbers of slides is approximately 18.45 mg, which is very less as compared to that required by the earlier researchers to achieve complete degradation of 100 mL of 10 ppm dimethoate solution. Further, the catalyst can be reused several times with the same efﬁciency.

219

60

40

20

0 0

60

120

180

240

300

Time (min) Fig. 7. Effect of bilayers on the degradation of dimethoate at an optimum catalyst loading of 4 g/L and with 400 W lamp. In all the experiments, 5 slides were immobilized having 5 bilayers on each.

the growth of the ﬁlms is uniform), which allows the free diffusion of dimethoate molecules in and out of the layers promoting enhanced degradation efﬁciency. This observation is consistent to our previous study where we proved that the degradation is not only carried out by TiO2 on the outermost layer but also by the inner layers [29]. This result is also supported by the explanation given by Andrea Malagutti et al. [31] that, in ﬁlms with higher number of layers, the electron recombination process becomes slower because electrons have to diffuse in to the inner layers and interact before returning to the surface of the semiconductor owing to a greater density of holes at the surface leading to higher photocatalytic activity as compared to ﬁlms with less number of layers. Further, the investigations made by Chen et al. [13] and Evgenidou et al. [14] on the degradation of dimethoate stated that the addition of oxidants like H2 O2 and K2 S2 O8 to the pesticide solutions improved the photodegradation rate. On the contrary, in our study, the increase in number of bilayers had a signiﬁcant effect on enhancing the degradation efﬁciency of dimethoate, which proportionately reduced the UV irradiation time. In comparison to the earlier reported methods this is much safer in terms of environmental perspectives since new compounds are not introduced in to the system. 3.5. Effect of pH The effect of pH on degradation of dimethoate was studied by varying the pH from 2.0 to 10.0. The catalyst was immobilized at an optimum loading (4 g/L) on 5 slides with 10 bilayers and the experiment was carried out for 180 min. The natural pH of dimethoate was observed to be 4.62. To carry out the experiments, the initial pH of the dimethoate solution was adjusted (using 0.1 M of HCl or NaOH) along with the immobilized catalyst to cover both the acidic and basic regions. Fig. 8 shows the effect of pH on the degradation of dimethoate. It was observed that pH had a very signiﬁcant effect on dimethoate degradation. For an acidic pH of 2.0, the amount of dimethoate remaining in the solution was 27.1%, after 180 min of photocatalysis. Increasing the initial pH values of 4.62, a complete degradation was observed. However, further increase in the initial pH values to 6, 8 and 10 resulted in decreased degradation. At a pH of 10, the least degradation was observed and 60% of the dimethoate was remained in the solution. Thus from the results obtained, it can be concluded that a better degradation can be achieved at natural pH of the dimethoate solution. The dependency of dimethoate degradation on pH values is attributed to the surface charge interactions between dimethoate

220

D.N. Priya et al. / Journal of Hazardous Materials 195 (2011) 214–222

100

pH 2.0 pH 4.62 pH 6.0 pH 8.0 pH 10.0

% dimethoate remaining

80

60

40

20

0 0

60

Time (min)

120

180

Fig. 8. Effect of pH on the degradation of dimethoate at an optimum catalyst loading of 4 g/L and with 400 W lamp. In all the experiments, 5 slides were immobilized having 10 bilayers on each.

and TiO2 . The point of zero charge (pzc) of commercial TiO2 catalyst is around 6.0, below this pzc, the catalyst surface is positively charged and above it is negatively charged. The maximum degradation around pzc of TiO2 at higher pH values has been reported by Evgenidou et al. [14]. Other reason is that hydroxyl ions with elevated concentration would scavenge photogenerated holes, which yield highly oxidative • OH species through interaction with H+ ions. At lower pH, the degradation results from photogenerated holes whose oxidizing ability is less as compared to that of • OH species [13]. Hence the degradation efﬁciency of dimethoate is less at low pH values than at higher pH. 3.6. Effect of initial dimethoate concentration The effect of initial dimethoate concentration on the degradation of dimethoate was studied by varying the initial concentration between 10 and 100 ppm, at optimum catalyst loading of 4 g/L TiO2 (pH 4.62) with PAH/(PSS/TiO2 )10 for an irradiation time of 300 min. The results are shown in Fig. 9. Complete degradation of dimethoate was observed at 10 ppm within 180 min of exposure. With increase in the initial concentration of dimethoate from 10 ppm to 100 ppm, the degradation reduced and the percentage

100

% dimethoate remaining

80

60

40 10 ppm 25 ppm 50 ppm 75 ppm 100 ppm

20

0 0

60

120

180

240

300

Time (min) Fig. 9. Effect of initial dimethoate concentration (pH 4.62) on the photocatalytic degradation at an optimum catalyst loading of 4 g/L and with 400 W lamp. In all the experiments, 5 slides were immobilized having 10 bilayers on each.

of dimethoate remaining in the solution drastically increased from 0% to 87.21%, respectively for an irradiation time of 300 min. When the initial dimethoate concentration was increased from 10 ppm to 100 ppm, the catalyst loading, light intensity and illumination time were held constant. Moreover the available reaction sites are also constant. Therefore the free radicals formed on the catalyst surface might not be sufﬁcient enough to degrade the pollutant molecules on the surface of the catalyst resulting in decreased degradation efﬁciency. From the previous studies, it is evident that the photocatalytic degradation reaction of the organophosphorus pesticides occurs on the surface of TiO2 in trapped holes and generation of reactive species play an important role in efﬁcient degradation of pollutants. The contaminants cannot be degraded unless they are adsorbed on the surface of TiO2 . Since other parameters like light intensity and illumination time are constant, as the initial concentration increases, more and more organic substances are adsorbed on the surface of the TiO2 , but the free radicals formed on the surface of TiO2 will be constant. Hence the relative number of free radicals attacking the organophosphorus compounds decreases thereby decreasing the degradation efﬁciency [25]. Similarly, phosphate concentration increases with increase in initial concentration of dimethoate to a certain extent but remains constant when the surface concentration of dimethoate reaches a steady state. Therefore the availability of total adsorption sites is the dominating factor which results in different degradation efﬁciencies between low and high concentrations [13]. However, the present study shows that dimethoate concentrations of ≤10 ppm could be effectively degraded. 3.7. GC–MS and HPLC data The degradation byproducts of dimethoate were analyzed and conﬁrmed by GC–MS. Table 1 shows the GC–MS data of dimethoate with respect to retention time. The data of dimethoate degradation intermediates was obtained after 180 min of UV irradiation (spectra not shown). Eight degradation products were identiﬁed that matched with the data reported earlier [15–17]. The ﬁrst compound identiﬁed was N-methyl-2-sulfanylacetamide by a peak at m/z = 105 and the characteristic ions at m/z = 71 and m/z = 57 which matched with the data reported by Evgenidou et al. [16]. The second compound was identiﬁed as O,S,S-trimethyl phosphorothioate which exhibited a peak at m/z = 156 [16]. The third compound was identiﬁed as 2-S-methyl-(N-methyl)acetamide and it exhibits ions at m/z = 119 that correspond to the molecular ion [M]+ and the characteristic ions m/z = 73 and m/z = 58. In addition, this compound was also detected as by-product of dimethoate thermal decomposition [16,17]. Compound 4 was identiﬁed as O,O,S-trimethyl thiophosphorothioate [16] with a peak at m/z = 172. This product has also been identiﬁed as an intermediate during thermal decomposition [17]. Compound 5 exhibits a peak at m/z = 151 which corresponds to the molecular ion [M]+ and the characteristic ions m/z = 105, m/z = 73 and m/z = 58 [16]. The sixth compound which exhibited a peak at m/z = 182 was identiﬁed as O,O,S-trimethyl thiophosphorodithioate, a product obtained during the thermal decomposition of dimethoate [17]. Compound 7 is identiﬁed as omethoate which gives a characteristic rearrangement ion peak at m/z = 156 and compound 8 was identiﬁed as 1,2-bis(acetyl-N-methyl-)methane disulﬁde that exhibited a peak at m/z = 206, which corresponds to the molecular ion [M]+. A detailed explanation of the characteristic ions of these compounds has been given by Evgenidou et al. [16]. The concentration of dimethoate was also determined using HPLC (SI Fig. SI 2) and observations are made from the changes in dimethoate parent signal at retention time (Rt ) of 6.04. When experiments with dimethoate solution along with the catalyst were performed in dark for 60 min, there was no change in the signal at all. After irradiating with UV, there was a decrease in the parent

D.N. Priya et al. / Journal of Hazardous Materials 195 (2011) 214–222

221

Table 1 GC–MS data of dimethoate degradation products with retention times and characteristic ions. Degradation products

Rt (min)

Characteristic ions (m/z)

N-methyl-2-sulfanylacetamide O,S,S-trimethyl phosphorothioate 2-S-methyl-(N-methyl)acetamide O,O,S-trimethyl thiophosphorothioate O,O,S-trimethyl thiophosphoro-dithioate Omethoate Dimethoate 1,2-Bis(acetyl-N-methyl-) methane disulﬁde

8.7 9.3 12.5 14.6 20.7 23.7 28.2 30.41

105, 71, 57 156, 141, 126, 110, 95, 79 119, 73, 62, 58 172, 125, 95, 57 182 213, 156, 141, 126, 110, 79, 58 229, 143, 125, 93, 87 206, 105, 73, 58

signal with increase in time. At the end of 180 min of exposure to UV irradiation, the signal became totally ﬂay indicating that dimethoate is completely removed from the solution and at the same time new peaks were detected with increasing intensities indicating the formation of new intermediates during the course of reaction. 3.8. Mineralization studies Complete mineralization of organic compounds is of great signiﬁcance in wastewater treatment. Hence TOC measurements were carried out to determine the extent of mineralization and the results are shown in Fig. 10. It was expected that total mineralization of the compound would be achieved within 180 min due to the total disappearance of dimethoate peak (as determined by HPLC) at this irradiation time. On the contrary, the TOC removal efﬁciency at 180 min was 21%. The irradiation time was further extended to 300 min in order to achieve complete mineralization but TOC removal at this stage was only 50%. The rate of TOC removal was very slow which might be due to the presence of more stable intermediates produced during the course of degradation. Similar trend of results was reported by Evgenidou et al. [14] where comparison of the TOC removal of dimethoate with TiO2 alone and with the addition of H2 O2 was made. They were able to achieve 59% TOC removal with TiO2 alone and 100% with TiO2 –H2 O2 system. 3.9. Toxicity tests using V. ﬁscheri The treated dimethoate solutions were also tested for their toxicity. When complete mineralization is not achieved even after 300 min, it is obvious that the ﬁnal products could be more toxic than the parent compounds. Hence toxicity measurements were performed for the treated solutions. The toxicity of the samples collected was measured using the LUMIStox luminometer. The results

3.10. Catalyst reusability The amount of dimethoate left in the solution after each cycle of catalyst reuse. Experiments were performed with 4 g/L TiO2 (pH 4.62) with PAH/(PSS/TiO2 )10 for an irradiation time of 180 min for 10 cycles. To compare the dimethoate degradation efﬁciency, after the ﬁrst cycle of reaction the degraded dimethoate solution was removed and a fresh solution (10 ppm) of dimethoate was placed in the beaker without removing the catalysts. The same procedure was repeated for nine more cycles. Samples were collected at the end of each cycle and analyzed to see the percentage degradation of the dimethoate in 180 min. The results after each cycle showed complete degradation of dimethoate by reusing the immobilized slides. The results indicate that there was no alteration in the surface characteristics of LbL-TiO2 thin ﬁlms even after several repeated uses indicating its stability and thus retaining the same photocatalytic activity after several cycles of operation. These

100

100

Dimethoate TOC

90 80

80

70 Inhibition [%]

% dimethoate remaining & TOC removal

obtained are presented in Fig. 11. Initially, there was an increase in toxicity from 34% to 62% after 120 min of irradiation. When the samples were further irradiated till 300 min, only a slight decrease (<8%) in toxicity was observed (54%). These results can be related to the TOC measurements where, TOC removal decreased in the initial stages but appears to be more or less the same with increase in irradiation time. Similarly there was not much effect on the reduction of toxicity also. The intermediates obtained during the degradation of dimethoate were already discussed under GC–MS analysis stating that the intermediates produced during the course of reaction are more toxic than the parent compound especially omethoate, one of the main degradation products, is proved to be more toxic than dimethoate. Complete mineralization and total reduction in toxicity can be achieved in the presence of an oxidant as reported by Evgenidou et al. [14,16].

60

40

60 50 40 30 20

20

10 0

0 0

60

120

180

240

300

360

Time (min) Fig. 10. Mineralization studies of dimethoate. Immobilized catalysts with TiO2 loading of 4 g/L, 5 slides with 10 bilayers were used. The UV irradiation source was 400 W lamp.

0

60

120

180

240

300

360

Irradiation time (min) Fig. 11. Measurement of toxicity of dimethoate toxicity using Vibrio ﬁscheri. Immobilized catalysts with TiO2 loading of 4 g/L, 5 slides with 10 bilayers were used. The UV irradiation source was 400 W lamp.

222

D.N. Priya et al. / Journal of Hazardous Materials 195 (2011) 214–222

studies are very important for any wastewater treatment system in terms of economic perspectives. Although the degradation of dimethoate with TiO2 in powder form was extensively studied by previous researchers [13–16], none of these studies focused on the reusability of the catalyst. Complete degradation was achieved but there are some drawbacks associated with the use TiO2 in powder form. One major concern is the difﬁculty in separating the TiO2 powder from water after the completion of the reaction. In this case a centrifugation step is deﬁnitely required to recover the catalyst for further use which might lead to the loss of catalyst during the process and practically not feasible to scale up the process. On the other hand, with the solutions treated using LbL-TiO2 ﬁlms, the ﬁnal solution is absolutely clean and free of any suspended nanoparticles that can be safely discharged to the environment. For real time applications this method can be used for continuous systems thus making it an option for scale up. 4. Conclusions The present study deals with the application of LbL-TiO2 thin ﬁlms for the photocatalytic degradation of dimethoate in presence of UV irradiation. Complete degradation of 10 ppm dimethoate at an optimum loading of 4 g/L TiO2 was achieved. Increase in number of bilayers increased the degradation with a reduction in UV illumination time to 180 min. pH had a signiﬁcant effect on degradation and at pH 4.62, maximum degradation was observed. The degradation rates reduced with increase in the initial dimethoate concentration. The GC–MS analysis conﬁrmed complete degradation. Though complete removal of dimethoate was achieved within 180 min, mineralization of dimethoate was only 50% after an irradiation time of 300 min. The toxicity increased during the course of the reaction due to the presence of stable intermediates. Hence future studies will be focused on this aspect by bringing some modiﬁcations in the system like irradiating for a longer time or increasing the number of bilayers. Though the illumination time is longer as compared to that used in suspension form, it is apparent that the reaction time can be reduced with increase in number of bilayers. Further, the main advantage of LbL technique is its applicability to any kind of shape and substrate which makes an attractive option for scale up. The immobilized catalyst has advantages such as catalyst reusability, solution free of TiO2 after the completion of reaction, etc. This technique is a worthwhile endeavor and a new approach that alleviates all the drawbacks associated with TiO2 in suspension form and can be successfully applied to continuous systems with signiﬁcant savings in terms of TiO2 and polymer requirement. The results obtained can be applied for the treatment of industrial wastewaters and others contaminated with pesticides. Acknowledgements The authors express their sincere thanks to the Hyderabad Chemicals Limited for supplying the free sample of dimethoate. Neela Priya thanks the UGC for the award of Dr. D. S. Kothari’s Postdoctoral fellowship under the programme [(F.42/2006(BSR)/13-190/2008(BSR)]. Financial assistance under the GKP Indo-Korea Program is also acknowledged. The authors also express their sincere thanks to the Department of Science and Technology, New Delhi, India for ﬁnancial support under the IndoSlovenian Bilateral Exchange Programme. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.jhazmat.2011.08.030.

References [1] P.C. Abhilash, N. Singh, Pesticide use and application: an Indian scenario, J. Hazard. Mater. 165 (2009) 1–12. [2] WHO, The WHO Recommended Classiﬁcation of Pesticides by Hazard and Guidelines to Classiﬁcation, World Health Organization, 2004. [3] O.P. Simo, Q. Zhang, The degradation of organophosphorus pesticides in natural waters: a critical review, Crit. Rev. Environ. Sci. Technol. 32 (2002) 17–72. [4] G.S. Henry, Adsorption of selected pesticides on activated carbon and mineral surfaces, Environ. Sci. Technol. 1 (1967) 332–337. [5] K. Ikehata, M.G. El-Din, Aqueous pesticide degradation by ozonation and ozonebased advanced oxidation processes: a review (part II), Ozone Sci. Eng. 27 (2005) 173–202. [6] E.S.K. Chian, W.N. Bruce, H.H.P. Fang, Removal of pesticides by reverse osmosis, Environ. Sci. Technol. 9 (1975) 52–59. [7] B. Van der Bruggen, J. Schaep, W. Maes, D. Wilms, C. Vandecasteele, Nanoﬁltration as a treatment method for the removal of pesticides from ground waters, Desalination 117 (1998) 139–147. [8] B. Kasprzyk-Hordern, M. Ziolek, J. Nawrocki, Catalytic ozonation and methods of enhancing molecular ozone reactions in water treatment, Appl. Catal. B: Environ. 46 (2003) 639–669. [9] R. Boussahel, S. Bouland, K.M. Moussaoui, A. Montiel, Removal of pesticide residues in water using the nanoﬁltration process, Desalination 132 (2000) 205–209. [10] I.K. Konstantinou, V.A. Sakkas, T.A. Albanis, Photocatalytic degradation of the herbicides propanil and molinate over aqueous TiO2 suspensions: identiﬁcation of intermediates and the reaction pathway, Appl. Catal. B: Environ. 34 (2001) 227–239. [11] E.T. Soares, M.A. Lansarin, C.C. Moro, A study of process variables for the photocatalytic degradation of rhodamine B, Braz. J. Chem. Eng. 24 (2007) 29–36. [12] D.L. Liao, B.Q. Liao, Shape, size and photocatalytic activity control of TiO2 nanoparticles with surfactants, J. Photochem. Photobiol. A 187 (2007) 363–369. [13] J.-Q. Chen, D. Wang, M.-X. Zhu, C.-J. Gao, Photocatalytic degradation of dimethoate using nanosized TiO2 powder, Desalination 207 (2007) 87–94. [14] E. Evgenidou, K. Fytianos, I. Poulios, Photocatalytic oxidation of dimethoate in aqueous solutions, J. Photochem. Photobiol. A 175 (2005) 29–38. [15] J. Chen, Z. Hu, D. Wang, C. Gao, R. Ji, Photocatalytic mineralization of dimethoate in aqueous solutions using TiO2 : parameters and by-products analysis, Desalination 258 (2010) 28–33. [16] E. Evgenidou, I. Konstantinou, K. Fytianos, T. Albanis, Study of the removal of dichlorvos and dimethoate in a titanium dioxide mediated photocatalytic process through the examination of intermediates and the reaction mechanism, J. Hazard. Mater. 137 (2006) 1056–1064. [17] R. Andreozzi, G. Ialongo, R. Marotta, R. Sanchirico, The thermal decomposition of dimethoate, J. Hazard. Mater. B 64 (1999) 283–294. [18] J. Li, W. Ma, C. Chen, J. Zhao, H. Zhu, X. Gao, Photodegradation of dye pollutants on one-dimensional TiO2 nanoparticles under UV and visible irradiation, J. Mol. Catal. A: Chem. 261 (2007) 131–138. [19] N. Negishi, K. Takeuchi, T. Ibusuki, Preparation of the TiO2 thin ﬁlm photocatalyst by the dip-coating process, J. Sol–Gel Sci. Technol. 13 (1998) 691. [20] Y. Ma, J.-N. Yao, Photodegradation of rhodamine B catalyzed by TiO2 thin ﬁlms, Photochem. Photobiol. A 116 (1998) 167. [21] K.S. Yeung, Y.W. Lam, A simple chemical vapour deposition method for depositing thin TiO2 ﬁlms, Thin Solid Films 109 (1983) 169. [22] S. Song, H. Liu, X. Guo, N. Hu, Comparative electrochemical study of myoglobin loaded in different polyelectrolyte layer-by-layer ﬁlms assembled by spin-coating, Electrochim. Acta 54 (2009) 5851–5857. [23] M. Takeuchi, T. Itoh, H. Nagasaka, Dielectric properties of sputtered TiO2 ﬁlms, Thin Solid Films 51 (1978) 83–88. [24] G. Decher, in: G. Decher, J.B. Schlenoff (Eds.), Multilayer Thin Films, Polyelectrolyte Multilayers an Overview, Wiley-VCH, Weinheim, Germany, 2003, p. 3. [25] M. Zhao, S. Chen, Y. Tao, Photocatalytic degradation of organophosphorus pesticides using thin ﬁlms of TiO2 , J. Chem. Technol. Biotechnol. 64 (1995) 339–344. [26] G.R.M. Echavia, F. Matzusawa, N. Negishi, Photocatalytic degradation of organophosphate and phosphonoglycine pesticides using TiO2 immobilized on silica gel, Chemosphere 76 (2009) 595–600. [27] C. Shifu, C. Gengyu, Photocatalytic degradation of organophosphorus pesticides using ﬂoating photocatalyst TiO2 ·SiO2 /beads by sunlight, Solar Energy 79 (2005) 1–9. [28] F. Mazille, T. Schoettl, N. Klamerth, S. Malato, C. Pulgarin, Field solar degradation of pesticides and emerging water contaminants mediated by polymer ﬁlms containing titanium and iron oxide with synergistic heterogeneous photocatalytic activity at neutral pH, Water Res. 44 (2010) 3029–3038. [29] D.N. Priya, J.M. Modak, A.M. Raichur, LbL fabricated poly(styrene sulfonate)/TiO2 multilayer thin ﬁlms for environmental applications, Appl. Mater. Interfaces 1 (2009) 2684–2693. [30] K. Nagaveni, G. Sivalingam, M.S. Hegde, G. Madras, Solar photocatalytic degradation of dyes: high activity of combustion synthesized nano TiO2 , Appl. Catal. B: Environ. 48 (2004) 83–93. [31] R. Andrea Malagutti, A.J.L. Henrique Mourao, R. Jose Garbin, Caue Ribeiro, Deposition of TiO2 and Ag:TiO2 thin ﬁlms by the polymeric precursor method and their application in the photodegradation of textile dyes, Appl. Catal. B: Environ. 90 (2009) 205–212.

Journal of Hazardous Materials 195 (2011) 223–229

Contents lists available at ScienceDirect

Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat

Investigation of simultaneous adsorption of SO2 and NO on ␥-alumina at low temperature using DRIFTS Ying Xie a , Ying Chen b,∗ , Yugang Ma a , Zhenglai Jin a,c a b c

Department of Chemical and Environmental Engineering, Guangdong University of Petrochemical Technology, Maoming 525000, Guangdong, China Zhejiang Ocean University, Zhoushan, 316000, Zhejiang, China School of Environmental Science and Engineering, Nanjing University of Information Science & Technology, Nanjing 210044, Jiangsu, China

a r t i c l e

i n f o

Article history: Received 28 April 2011 Received in revised form 7 August 2011 Accepted 9 August 2011 Available online 16 August 2011 Keywords: SO2 NO ␥-Al2 O3 Adsorption mechanism DRIFTS

a b s t r a c t The interaction mechanism between SO2 and NO on ␥-Al2 O3 was explored by diffuse reﬂectance infrared Fourier transform spectroscopy (DRIFTS) and outlet response of the concentrations of NO, NO2 and SO2 under exposure of Al2 O3 to SO2 and/or NO in the absence or presence of oxygen at 150 ◦ C. The results showed that SO2 promoted NO oxidation and NO transformed weakly adsorbed SO2 into strongly adsorbed species on ␥-Al2 O3 , and the presence of O2 facilitated this transformation. An interaction mechanism between SO2 and NO on ␥-Al2 O3 was thus postulated. The exposure of Al2 O3 to SO2 and NO in the presence of O2 resulted in the formation of at least two types of intermediates. One type was [SO3 NO], which decomposed to form NO2 , and the other type was [SO3 NO2 ], which decomposed to form SO3 . The decomposition of both intermediates probably formed O vacancies replaceable by gaseous O2 . Crown Copyright © 2011 Published by Elsevier B.V. All rights reserved.

1. Introduction The emission of sulphur oxides (SOx ) and nitrogen oxides (NOx ) from ﬂue gases, causing acid rain and urban air pollution, is a major environmental issue. Normally, SOx and NOx in ﬂue gases consist of more than 98% sulphur dioxide (SO2 ) and over 90–95% nitric oxide (NO) [1,2]. To control SO2 and NOx emission, a great deal of simultaneous removal processes have been developed [3–9]. Flue gas treatment technologies are broadly classiﬁed as dry and wet techniques. The wet techniques use scrubber columns in which the ﬂue-gas mixture is subjected to liquid wash to remove gaseous SO2 and NOx with high efﬁciency, however, the wet process induces the difﬁculty of product disposal. Therefore, it is highly desirable to have a suitable single-step dry process for the removal of SO2 and NOx from ﬂue gas. As a promising dry process, the NOXSO process uses a regenerable sorbent (prepared by spraying sodium carbonate on ␥-Al2 O3 ) to remove SO2 and NOx simultaneously by catalytic oxidation. The process was tested at different scales, which was still in stage of demonstration industrial plant [10]. FLS-miljØ-Denmark has developed a new process derived from NOXSO process. In the process, the simultaneous adsorption of SO2 and NOx was performed on Na-␥-Al2 O3 in a circulating dilute phase riser reactor. De Wilde

∗ Corresponding author. Tel.: +86 580 2556212; fax: +86 580 2551439. E-mail address: [email protected] (Y. Chen).

et al. performed simultaneous SO2 and NOx removal on Na-␥-Al2 O3 at lower temperature (150 ◦ C) [11]. The interaction of SO2 and NOx on Na-␥-Al2 O3 is described. They explained the inﬂuence of the SO2 presence on the simultaneous adsorption of NO and O2 by the adsorbed SO2 as an intermediate in the NO and O2 adsorption. With respect to the role of supporter ␥-Al2 O3 and interaction of SO2 and NO on ␥-Al2 O3 without Na-impregnation, however, not much information is available in literature. Moreover, few studies related to the sequential adsorption of SO2 and NOx on ␥-Al2 O3 . In this paper, the interaction among NO, SO2 and O2 on ␥-Al2 O3 at low temperature(150 ◦ C) was systematically studied. Different with De Wilde’s research [11] sequential adsorption experiment was carried out for better understanding the reactions occurring on ␥-Al2 O3 surface. Finally, the interaction mechanism of SO2 and NO was proposed in this paper. 2. Experimental The sample of ␥-Al2 O3 was obtained from Merck (Merck Co., Germany) in the form of powder with a particle size of 0.10–0.15 mm. The speciﬁc surface area was 128 m2 /g, and the average pore diameter and pore volume were 7 nm and 0.2484 cm3 /g, determined by ourselves. The feed gas mixture contained 0.075% NO, 0.51% SO2 , 4.5% O2 , and balance Ar. The adsorption experiments were performed in a ﬁxed-bed reactor apparatus. The sample (1 g) was charged in a stainless reactor (Ø19 mm) and then purged under inert ﬂow at a total ﬂow rate

0304-3894/$ – see front matter. Crown Copyright © 2011 Published by Elsevier B.V. All rights reserved. doi:10.1016/j.jhazmat.2011.08.032

224

Y. Xie et al. / Journal of Hazardous Materials 195 (2011) 223–229

of 100 ml/min at 600 ◦ C for 1 h in order to remove containing oxygen compounds (H2 O and CO2 ); it was then cooled to 150 ◦ C and exposed to a mixture of NO and/or SO2 in Ar at a total ﬂow rate of 100 ml/min until the concentration of NO and SO2 in the outlet gas became steady. A series of NO and SO2 adsorption experiments on ␥-Al2 O3 were performed by exposing the samples to NO and/or SO2 in Ar with or without oxygen. SO2 and NO sequential experiments were also performed (termed PreSO2 and PreNO). PreSO2 indicates that SO2 /O2 was ﬁrst introduced to a fresh catalyst (‘clean’ Al2 O3 ). After saturation (sulphated Al2 O3 ), the SO2 gas ﬂow was changed to inert gas for 5 min, followed by exposure of the sulphated Al2 O3 to NO/O2 in Ar. PreNO indicates that NO/O2 was ﬁrst introduced to a fresh catalyst (‘clean’ Al2 O3 ). After saturation (nitrated Al2 O3 ), the nitrated Al2 O3 was exposed to SO2 /O2 in Ar. The measured outlet response curves were determined in a ﬂow-reactor equipped with a Total Sulphur/Nitrogen Analyzer for monitoring the concentrations of NO, NO2 and SO2 in the outlet gas (detection limit S or N with 0.2 mg/m3 ). After saturation, the sample was cooled to 50 ◦ C and then purged in Ar for 1 h. Finally, a temperature ramp of 10 ◦ C/min from 50 to 727 ◦ C was applied with an Ar gas rate of 20 ml/min. The diffuse reﬂectance infrared Fourier transform spectroscopy (DRIFTS) spectrum was carried out on a Bruker vector 33 spectrometer 3. Results and discussion 3.1. Adsorption of SO2 and NO The amounts of adsorbed SO2 and NO under different atmospheres over ␥-Al2 O3 at 150 ◦ C are summarised in Table 1. 3.1.1. Separate adsorption of SO2 and NO Comparing SO2 with SO2 /O2 (experiments a and b in Table 1), the amount of adsorbed SO2 in the absence of O2 was 0.265 mmol g−1 , whereas that in the presence of O2 was 0.321 mmol g−1 . This demonstrated that the adsorption of SO2 alone on ␥-Al2 O3 at 150 ◦ C occurred, and that the presence of O2 enhanced SO2 adsorption (the amount of adsorbed SO2 increased by 20%). In addition, 0.014 mmol g−1 of SO2 desorbed when sweeping with Ar at 150 ◦ C, which indicated that some adsorbed SO2 was unstable. Comparing NO with NO/O2 (experiments c and d), the amount of adsorbed NO in the absence of O2 was 0.0298 mmol g−1 , whereas in the presence of O2 , it was 0.038 mmol g−1 . These data indicated that NO could be independently adsorbed on ␥-Al2 O3 at 150 ◦ C. Additionally, the amount of adsorbed NO increased in the presence of O2 . In experiment c for NO in the absence of O2 , oxygen-containing compounds were removed from the ␥-Al2 O3 surface by purging with Ar at 600 ◦ C for 1 h before adsorption. However, a trace of NO2 still occurred in the outlet gas. This was likely that lattice oxygen of the ␥-Al2 O3 participated in the oxidation reaction [12,13]. 3.1.2. Simultaneous adsorption of SO2 and NO Comparing the amount of adsorbed SO2 over ␥-Al2 O3 under SO2 /NO and SO2 atmospheres (experiments e and a in Table 1), the amount of adsorbed SO2 for simultaneous adsorption of SO2 and NO was higher than that in the absence of NO, which indicated that NO enhanced the adsorption of SO2 . Comparing SO2 /NO with NO (experiments e and c in Table 1), although the amount of adsorbed NO was almost the same, NO2 was also detected in the outlet gas for simultaneous adsorption of SO2 and NO. Based on thermodynamics, it was unlikely that the oxygen of NO2 was from SO2 . Thus, the oxidation of NO may be attributed to the lattice oxygen of ␥-Al2 O3 with SO2 , which promoted this oxidation reaction.

When SO2 /NO/O2 was compared with SO2 /NO (experiments f and e in Table 1), both the amounts of adsorbed SO2 and NO were signiﬁcantly higher than in the absence of O2 . This observation revealed that O2 facilitated the simultaneous adsorption of SO2 and NO, whereas NO promoted the adsorption of SO2 and vice versa. 3.1.3. Sequential adsorption of SO2 and NO Comparing SO2 /O2 over “nitrated Al2 O3 ” with SO2 /O2 over “clean Al2 O3 ” (experiments h and b in Table 1), the amount of adsorbed SO2 over the “clean Al2 O3 ” was 0.321 mmol g−1 , whereas that over the “nitrated Al2 O3 ” increased to 0.377 mmol g−1 . Thus, pre-adsorbed NO species on Al2 O3 promoted SO2 adsorption. Moreover, in experiment h, SO2 /O2 was exposed to “nitrated Al2 O3 ” after NO/O2 was saturated on the Al2 O3 and NO and NO2 were detected in the outlet gas. It might have been that some adsorbed NO species on the Al2 O3 were replaced by SO2 due to its stronger acidity, leading to the discharge of NO and NO2 into the outlet gas. When NO/O2 over “sulphated Al2 O3 ” was compared with “clean Al2 O3 ” (experiments g and d in Table 1), the amount of adsorbed NO increased by 27%, which indicating that pre-adsorbed SO2 species on Al2 O3 promoted NO adsorption. In the PreSO2 experiment (experiment g in Table 1), the desorbed SO2 was 0.0533 mmol g−1 , whereas that for SO2 /O2 over “clean Al2 O3 ” was 0.014 mmol g−1 (experiment b in Table 1) by sweeping with Ar. The reason for this phenomenon was probably that some pre-adsorbed SO2 was replaced by NO. In conclusion, both the adsorption of SO2 on nitrated Al2 O3 (PreNO) and NO on sulphated (PreSO2 ) were promoted regardless of the sequence of exposure for SO2 or NO. 3.2. DRIFTS studies of adsorbed species 3.2.1. Separate adsorption of SO2 and NO The surface species formed from the reaction of SO2 or NO on Al2 O3 were studied by DRIFTS (Fig. 1). Fig. 1(a, b and i) shows the spectra of SO2 or SO2 /O2 adsorption on Al2 O3 and the desorption after SO2 /O2 saturation at 150 ◦ C for 1 h. Datta et al. [14] identiﬁed at least ﬁve different adsorption SO2 sites on Al2 O3 : a species physically adsorbed on hydroxyl groups (Al–OH–SO2 ) with bands at 1334 and 1148 cm−l , a weakly chemisorbed species (Al–O–SO2 ) with bands at 1322 and1140 cm−l , two species chemisorbed on acidic (positively charged aluminium ions, Al–SO2 ) Al3+ sites with bands at 1255 and 1189 cm−1 , and one strongly chemisorbed species (Al–SO3 ) with a broad band at approximately 1060 cm−l . As seen from spectra (a) and (b), the bands at 1325 cm−1 were observed and assigned to a weakly chemisorbed species (Al–O–SO2 ) [15]. In addition, the bands between 1200 and 1000 cm−1 might all be characteristic peaks of mixtures of the above-mentioned SO2 surface species. It was noted that the intensity for the bands from spectrum (b) was greater than that from spectrum (a). Additionally, the increase in intensity at 1325 cm−1 indicated that O2 had an effect on SO2 chemisorption at 150 ◦ C, which probably enhanced SO2 adsorption on O2− basic sites of Al2 O3 (Al–O–SO2 ). In the literature, Andersson et al. [16] reported that the amount of SO2 adsorbed in the presence of O2 was much higher than that absorbed in the absence of O2 . The results (seen from experiments b and a in Table 1) were in agreement with Andersson’s report. It was probable that O2 formed new basic sites at the lattice defect-sites on Al2 O3 , thus increasing the amount of Al–O–SO2 species present. Furthermore, compared with spectrum (b), the intensity for bands of spectrum (i) at 1325 and 1140 cm−1 decreased, indicating that adsorbed SO2 was bonded to O2− basic sites through the sulphur atom (Al–O–SO2 ), a weak surface species readily desorbed by sweeping in Ar at 150 ◦ C. Fig. 1(c and d) displays representative DRIFTS data for Al2 O3 exposed to NO at 150 ◦ C. The bands in the region from 1640 to 1000 cm−1 were assigned to surface nitrate and nitrite species [17].

Y. Xie et al. / Journal of Hazardous Materials 195 (2011) 223–229

225

Table 1 The amount of SO2 or NO adsorbed on ␥-Al2 O3 at 150 ◦ C. Experimental no.

Feed gas, Ar balance

Amount adsorbed SO2 (mmol g−1 )

Amount adsorbed NO (mmol g−1 )

SO2 desorption rate below 600 ◦ C (%)f

a b c d e f g h

SO2 SO2 /O2 a NOb NO/O2 SO2 /NOc SO2 /NO/O2 PreSO2 NO/O2 d PreNO SO2 /O2 e

0.265 0.321 – – 0.324 0.582 – 0.377

– – 0.0298 0.0380 0.0308 0.0866 0.0484 –

30 31 – – 16 0.5 0.6 0.5

f

0.04

1180

1325

Kubella Munck

0.06

i 0.02

b a

0.00 1500

1400

1300

1200

1100

1390

1078

0.12

1500

1640

0.16

1000

-1

1100

Wavenumber,cm

Kubellka Munck

1050

e

1088

c d

Sweeping with inert gas (Ar) at 150 ◦ C after absorption, SO2 (0.014 mmol g−1 ) in sweeping gas. NO2 was found in outlet gas during the adsorption process. NO2 was found in outlet gas during the adsorption process. SO2 (0.053 mmol g−1 ) in outlet gas during the adsorption process. NOx was found in outlet gas during the adsorption process. Mole ratio of SO2 desorption below 600 ◦ C and SO2 absorbed at 150 ◦ C.

1140 1110 1080

a b

0.08 0.04 0.00 2000

h g d c 1800

1600

1400

1200

1000

1100

Wavenumber cm-1

0.6

0.3

1390

0.4

1500

0.5 1640

Kubelka Munck

0.7

0.2 0.1 0.0 2000

f e g 1800

1600

1400

1200

1000

wavenumber,cm-1 Fig. 1. DRIFT spectra of Al2 O3 exposed to different atmospheres at 150 ◦ C. (a) SO2 ; (b) SO2 /O2 ; (c) NO; (d) NO/O2 ; (e) SO2 /NO; (f) SO2 /NO/O2 ; (g) PreSO2 : pre-adsorbed SO2 and NO/O2 ; (h) PreNO: pre-adsorbed NO and SO2 /O2 ; (i) desorption at 150 ◦ C for 1 h after exposure to SO2 /O2. .

The bands at 1630 and 1246 cm−1 were assigned to bridged nitrate, 1570 and 1249 cm−1 were assigned to bidentate nitrate, 1574 and 1290 cm−1 were assigned to monodentate nitrate, and the bands at 1470 and 1080 cm−1 were assigned to linear nitrite. In addition, the bands at 1240 and 1180 cm−1 were assigned to bidentate nitrite, whereas 1320 and 1230 cm−1 were assigned to bridged nitrite. In general, the bands ranging from 1640 to 1500 cm−1 were mainly assigned to surface nitrate species and bands between 1400 and 1000 cm−1 were mainly assigned to surface nitrite species. It has been reported that the exposure Al2 O3 to NO mainly formed nitrites on basic sites, and exposure of Al2 O3 to NO2 primarily formed nitrates at these sites [17–19]. In Fig. 1(c and d), it can be seen that bands at 1640–1500 cm−1 and 1400–1000 cm−1 were again observed, indicating that surface nitrite and nitrate species might have formed. Therefore, it was suggested that NO2 formed due to the oxidation reaction of adsorbed NO and lattice oxygen on Al2 O3 . Compared with (c) and (d) in Fig. 1, the band intensities increased in the presence of O2 , suggesting that gaseous O2 might migrate into O vacancies to enhance adsorption of NO and NO2 [20]. 3.2.2. Simultaneous adsorption of SO2 and NO Compared with the result of exposure of Al2 O3 to NO, the intensity for bands at 1640–1500 cm−1 increased in the case of simultaneous adsorption of SO2 and NO (Fig. 1(e and c)). This observation indicated that more surface nitrate species formed in the case of SO2 /NO. It thus seemed that SO2 promoted the oxidation of NO to form NO2 absorbed species. In Fig. 1(e and a), the bands at 1390 cm−l and 1100 cm−l grew in intensity under conditions of simultaneous adsorption of SO2 and NO, an effect assigned to surface sulphates (SO4 2− )[1,15]. This indicated that NO could enhance SO2 oxidation and lead to the formation of surface sulphates. These data were in good agreement with results reported by Tomohiro et al. [21]. Compared with the spectra of exposure to SO2 /NO (e), all of the band intensities for exposure to SO2 /NO/O2 (f) increased, especially the bands at 1390 cm−l and 1100 cm−l , assigned to surface sulphates (SO4 2− ). These results indicated that gaseous O2 promoted mutual oxidation of SO2 and NO. 3.2.3. Sequential adsorption of NO and SO2 Fig. 1(g and h) depicts the DRIFTS results of “nitrated Al2 O3 ” exposed to SO2 /O2 (PreNO) and “sulphated Al2 O3 exposed to NO/O2 (Pre SO2 ). It was found that both nitrate (1640–1500 cm−1 ) and SO4 2− (1390 and 1100 cm−l ) formed regardless of the sequence of exposure to SO2 and NO. Compared with sulphated Al2 O3 exposed to NO/O2 (g in Fig. 1), the bands at 1390 and 1100 cm−l increased in intensity when nitrated Al2 O3 was exposed to SO2 /O2 (h in Fig. 1). In this case, more SO4 2− formed. Thus, the amounts

226

Y. Xie et al. / Journal of Hazardous Materials 195 (2011) 223–229

of nitrate and SO4 2− formed were correlated with the sequence of exposure. 3.3. Response curve Figs. 2 and 3 display the measured outlet responses of NO, NO2 and SO2 . 3.3.1. Separate adsorption of SO2 and NO In Chang’s report [22], it was demonstrated that the desorption of weakly absorbed SO2 species on basic sites occurred below 600 ◦ C, whereas that of strongly absorbed SO2 species on acidic sites occurred above 600 ◦ C. Fig. 2(c) shows that the SO2 response curve exhibited three desorption peaks, indicating different degrees of basicity for the surface basic sites of ␥-Al2 O3 . By comparison to SO2 /O2 and SO2 , it appeared that the area of the SO2 desorption peak (below 600 ◦ C) was greater in the presence of O2 than in the absence of O2 and that the total amount of absorbed SO2 increased (Table 1). It is likely that adsorbed O2 created new basic sites on defect sites of the Al2 O3 surface, leading to an increased number of weakly absorbed SO2 species. Based on the decomposition equations for magnesium nitrate [23], aluminium nitrate decomposition equations were determined as follows: Al2 (NO2 )3 → Al2 O3 + 3NO

(1)

Al2 (NO3 )3 → Al2 O3 + 3NO + 1.5O2

(2)

Al2 (NO3 )3 → Al2 O3 + 3NO2

(3)

It was noted that NO was formed by the decomposition of nitrite and nitrate and that NO2 was formed by the decomposition of nitrate. Because nitrite decomposed more easily than nitrate, the low temperature peaks for NO represented the decomposition of nitrite, whereas the high-temperature peaks for NO and NO2 represented the decomposition of nitrate. Therefore, the low- and high-temperature peaks represented weakly and strongly absorbed species, respectively. After exposure of Al2 O3 to NO in the absence of O2 , the high-temperature peak for NO2 appeared, indicating that nitrate species formed on the surface of Al2 O3 (Fig. 2(b)). The reason for this phenomenon might have been the oxidation of adsorbed NO to NO2 by lattice oxygen to form nitrate species. After exposure to NO/O2 , the position of the low-temperature peak of NO shifted towards lower temperature, and the area of the low temperature peak increased within a wide range. This development indicated that new basic sites formed on the defect sites of Al2 O3 in the presence of O2 , increasing the amount of weakly adsorbed species. The area of the response peak at high temperature for NOx (NO and NO2 ) also greatly increased. This indicated that a considerable amount of NO was oxidized into NO2 (i.e., more surface nitrate on ␥-Al2 O3 ). This observation was likely attributed to the formation of O vacancies after the lattice oxygen participated in the reaction which were subsequently replaced by gaseous O2 for the reaction. Otherwise, the adsorbed oxygen on defect sites could also have participated in the oxidation of NO. 3.3.2. Simultaneous adsorption of SO2 and NO From Fig. 2(c) and Table 1 (experiments a and e), it can be seen that the area of three SO2 response peaks below 600 ◦ C for NO/SO2 decreased compared to SO2 , and the rate of SO2 desorption dropped from 30% to 16%. However, the amount of SO2 absorbed increased. The reason for this trend was probably that NO transformed weakly absorbed SO2 species into strongly absorbed SO2 species. It was presumed that weakly absorbed SO2 species (Al–O–SO2 ) was the active species. According to the DRIFT results, the strongly absorbed species were most likely Al–O–SO3 or SO4 2− .

By comparing the SO2 /NO and SO2 /NO/O2 cases, the SO2 response peaks nearly disappeared (the rate of SO2 desorption decreased from 16% to 0.5%, experiments a and e) below 600 ◦ C. However, the amount of SO2 absorbed greatly increased. Thus, the weakly adsorbed SO2 species were likely transformed into more strongly adsorbed species (surface sulphates) in the presence of O2 . It also seemed that gaseous O2 promoted this transformation process. These results were in accordance with DRIFT results where the bands at 1390 cm−l and 1100 cm−l , characteristic of surface sulphates, were found to be greatly increased. Compared with the results of response peaks for NO alone, the low-temperature response peak of NO disappeared in the case of NO/SO2 . Meanwhile, the area of the NOx response peaks increased and the position for the high-temperature peaks of NOx both shifted towards lower temperatures. It was thus presumed that lattice oxygen participated in the oxidation of weakly adsorbed NO species and that the presence of SO2 enhanced the transformation of NO to NO2 on Al2 O3 . As for the shift towards lower temperatures, it was concluded that SO2 promoted nitrate decomposition [21]. Compared with the response peaks for NO/SO2 , the area for the NOx response peaks at high temperature were signiﬁcantly larger than in the absence of O2 . This indicated that O2 could facilitate the oxidation NO by SO2 . 3.3.3. Sequential adsorption of NO and SO2 In Fig. 3(c), it can be seen that the response peak of SO2 disappeared over a wide range of temperatures (90–600 ◦ C). In addition, less SO2 was desorbed from the sulphated or nitrated Al2 O3 compared to the ‘clean’ Al2 O3 . This indicated that weakly adsorbed SO2 could be converted into strongly adsorbed species on Al2 O3 in both cases (preSO2 and preNO), regardless of the sequence of exposure to SO2 and NO. The NO and NO2 response peaks of sulphated Al2 O3 exposed to NO/O2 (preSO2 ) and nitrated Al2 O3 exposed to SO2 /O2 (preNO) were shown in Fig. 3(a and b). Compared with ‘clean’ Al2 O3 exposed to NO/O2 , the area of the low-temperature peak for NO decreased and the area of the high-temperature peak for NO2 increased. It therefore seemed that the surface species associated with NO lowtemperature desorption took part in oxidative reactions to form surface nitrate as long as SO2 existed, regardless of the sequence of NO exposure. From Fig. 3(a and b), it was also noted that the position of the high-temperature peaks shifted towards lower temperatures. It is known that when the amount of acidic species (surface sulphates) increases on Al2 O3 , the desorption of NOx species occurs more easily [19]. Therefore, compared with sulphated Al2 O3 exposed to NO/O2 (preSO2 ), a greater shifting towards lower temperatures indicated that more SO4 2− formed on the Al2 O3 in the case of nitrated Al2 O3 exposed to SO2 /O2 (preNO). According to the results of DRIFTS and response curve (Fig. 3), it was determined that at least two intermediates formed when SO2 and NO interacted on Al2 O3 and that the sequence of SO2 and NO exposure to Al2 O3 had a signiﬁcant effect on the amount of intermediates. Furthermore, the decomposition of some intermediates in this scenario proceeded more easily, leading to more surface sulphates in the case of nitrated Al2 O3 exposed to SO2 /O2 (preNO). Based on the discussion of separate, simultaneous and sequential adsorption of SO2 and NO on Al2 O3 , ﬁve main conclusions could be drawn: (1) The adsorption and oxidation of SO2 and NO on the Al2 O3 was promoted by each other. (2) The lattice oxygen of Al2 O3 may have lower activity but probably could participate in oxidation reactions. Adsorbed O2 on defect sites of the Al2 O3 could also participate in oxidation reactions. Gaseous O2 facilitated the replacement of O vacancies.

Y. Xie et al. / Journal of Hazardous Materials 195 (2011) 223–229 NO

a

1000

b

NO/O2

227

2000

NO NO/O2

NO/SO2 SO2/NO+O2

SO2/NO

ϕ(NO2) /10-6

ϕ(NO) /10-6

800 600 400

1500

SO2/NO/O2

1000

500

200

0

0 300

400

500

600

700

800

900 1000

300

400

500

600

700

800

900 1000

T/K

c

2500

ϕ (SO2) /10-6

T/K

2000

SO2 SO2/O2 NO/SO2 NO/SO2/O2

1500 1000 500 0

300

400

500

600

700

800

900 1000

T/K Fig. 2. Measured outlet response of NO, NO2 and SO2 in different atmospheres. (a) NO; (b) NO2 ; and (c) SO2 .

a

b

1000

2000 NO/O 2

800

ϕ(NO2) /10-6

ϕ(NO) /10-6

NO/O2 PreNO PreSO2

600

PreNO PreSO2

1500

1000

400 500

200 0

300

400

500

600

700

800

0

900 1000

300

400

500

T/K

600

T/K

ϕ(SO2) /10-6

c

2500

SO2/O2 PreNO PreSO2

2000 1500 1000 500 0

300

400

500

600

700

800

900

T/K Fig. 3. Measured outlet response of NO, NO2 and SO2 on sulphated (preSO2 ) and nitrated (preNO) Al2 O3 at 150 ◦ C. (a) NO; (b) NO2 ; and (c) SO2 .

1000

700

800

900 1000

228

Y. Xie et al. / Journal of Hazardous Materials 195 (2011) 223–229

Fig. 4. Intermediates formed for SO2 and NO interactions on an Al2 O3 surface.

Fig. 5. Possible modes of interacting for SO2 /NO/O2 on an Al2 O3 surface.

(3) The weakly absorbed NO and SO2 species (Al–O–SO2 and Al–O· · ·NO) associated with low-temperature desorption might have activity. (4) The weakly absorbed NO and SO2 species could be partly replaced with each other on sulphated and nitrated Al2 O3 (preSO2 and preNO). The adsorptions of SO2 and NO were also promoted by each other regardless of the sequence of exposure to SO2 and NO. (5) At least two intermediates formed when SO2 and NO interacted on Al2 O3 , and the amount of intermediates was affected by the amount of surface SO2 and NOx species. Some intermediates decomposed more easily to form surface sulphates (SO4 2− ) in the case of SO2 exposure to nitrated Al2 O3 (preNO) rather than NO exposure to sulphated Al2 O3 (preSO2 ).

The other intermediate [SO3 NO2 ] decomposed according to Eqs. (6)–(8). For the preNO process, the main intermediate formed with more SO4 2− on the Al2 O3 was [SO3 NO2 ]. It could therefore be concluded that (6) and (8) were possible decomposition pathways. However, in Eq. (6), as the lattice oxygen is inactive and two lattice oxygens are consumed in the formation of [SO3 NO2 ], there was little chance that decomposition occurred via Eq. (6). Thus, (8) was the main process for decomposition of [SO3 NO2 ]. In the following, two modes were proposed for the interaction of SO2 and NO on Al2 O3 . Mechanism I: Al–O· · ·SO2 reacting with gaseous NO in the presence of O2 formed intermediate [SO3 NO], which decomposed to form NO2 , SO2 and an O vacancy. The lattice oxygen and defect-site oxygen was assumed to be regenerated by gaseous O2 .

3.4. Mechanism Wilde et al. [24] proposed several intermediates, as shown in Fig. 4(a). Apparently, more NO adsorbed species were favoured in the formation of intermediate [SO3 NO2 ]. For the PreNO process, when the nitrated Al2 O3 was exposed to SO2 /O2 , the basic sites of the Al2 O3 were mostly occupied by NOx (NO or NO2 ). The amount of [SO3 NO2 ] formed was then much higher than [SO3 NO] due to a considerable amount of NO adsorbed species (Al–O· · ·NO). The decomposition process for intermediates [SO3 NO] and [SO3 NO2 ] are shown in Eqs. (4)–(8), where [] represents O vacancies. [SO3 NO] → SO3 + NO + []

(4)

[SO3 NO] → SO2 + NO2 + []

(5)

[SO3 NO2 ] → SO3 + NO2 + 2[]

(6)

[SO3 NO2 ] → SO2 + NO2 + []

(7)

[SO3 NO2 ] → SO3 + NO + []

(8)

The intermediate [SO3 NO] decomposed according to Eqs. (4) and (5). However, for the preSO2 process, the predominant intermediate formed was [SO3 NO]. It was therefore concluded that (5) was the main process for decomposition of [SO3 NO].

Mechanism II: Al–O· · ·SO2 or Al–OH· · ·SO2 interacting with adjacent Al–O· · ·NO formed intermediate [SO3 NO2 ], which decomposed to form NO, an O vacancy and SO4 2− (or NO2 , two O vacancies and SO4 2− ). The lattice oxygen and defect site oxygen on Al2 O3 was assumed to be regenerated by gaseous O2 . The possible interacting modes of SO2 and NO on Al2 O3 are illustrated in Fig. 5. 4. Conclusions The interaction of adsorbed SO2 and NO on ␥-Al2 O3 at low temperature (150 ◦ C) was studied by DRIFT and measured outlet response curves. The adsorptions of SO2 and NO on Al2 O3 were promoted by each other, especially in the presence of O2 . The weakly absorbed species of NO (Al–O· · ·NO) and SO2 (Al–O· · ·SO2 ) might have activity and could be transformed into strongly absorbed species. In addition, the lattice oxygen of Al2 O3 likely had a low activity, but could directly participate in oxidation reactions. At least two types of intermediates formed when SO2 and NO interacted on Al2 O3 . One type was [SO3 NO], which decomposed to form NO2 , and [SO3 NO2 ], which decomposed to form SO3 . The decomposition of both intermediates could form O vacancies, which could then be replaced by gaseous O2 .

Y. Xie et al. / Journal of Hazardous Materials 195 (2011) 223–229

References [1] I. Dahlan, K.T. Lee, A.H. Kamaruddin, A.R. Mohamed, Sorption of SO2 and NO from simulated ﬂue gas over rice husk ash (RHA)/CaO/CeO2 sorbent: evaluation of deactivation kinetic parameters, J. Hazard. Mater. 185 (2011) 1609–1613. [2] K.C. Pillai, S.J. Chung, T. Raju, Experimental aspects of combined NOx and SO2 removal from ﬂue-gas mixture in an integrated wet scrubber-electrochemical cell system, Chemosphere 76 (2009) 657–664. [3] D.S. Jin, B.R. Deshwal, Y.S. Park, H.K. Lee, Simultaneous removal of SO2 and NO by wet scrubbing using aqueous chlorine dioxide solution, J. Hazard Mater. B135 (2006) 412–417. [4] I. Dahlan, K.T. Lee, A.H. Kamaruddin, A.R. Mohamed, Selection of metal oxides in the preparation of rice husk ash (RHA)/CaO sorbent for simultaneous SO2 and NO removal, J. Hazard. Mater. 166 (2009) 1556–2155. [5] W.Y. Sun, S.L. Ding, S.S. Zeng, W.J. Jiang, Simultaneous absorption of NOx and SO2 from ﬂue gas with pyrolusite slurry combined with gas-phase oxidation of NO using ozone, J. Hazard. Mater. 192 (2011) 124–130. [6] G.Y. Xie, Z.Y. Liu, Z.P. Zhu, Q.Y. Liu, J. Ge, Z.G. Huang, Simultaneous removal of SO2 and NOx from ﬂue gas using a CuO/Al2 O3 catalyst sorbent I. Deactivation of SCR activity by SO2 at low temperatures, J. Catal. 224 (2004) 36–41. [7] G.Y. Xie, Z.Y. Liu, Z.P. Zhu, Q.Y. Liu, J. Ge, Z.G. Huang, Simultaneous removal of SO2 and NOx from ﬂue gas using a CuO/Al2 O3 catalyst sorbent II: promotion of SCR activity by SO2 at high temperatures, J. Catal. 224 (2004) 42–49. [8] A.K. Das, J. De. Wilde, G.J. Heynderickx, G.B. Marin, CFD simulation of dilute phase gas–solid riser reactors: part II-simultaneous adsorption of SO2 –NOx from ﬂue gases, Chem. Eng. Sci. 59 (2004) 187–200. [9] S. Sumathi, S. Bhatia, K.T. Lee, A.R. Mohamed, Cerium impregnated palm shell activated carbon (Ce/PSAC) sorbent for simultaneous removal of SO2 and NO—process study, Chem. Eng. J. 162 (2010) 51–57. [10] W.T. Ma, A.M. Chang, J.L. Haslbeck, L.G. Neal, NOXSO SO2 /NOx ﬂue gas treatment process adsorption chemistry and kKinetics: novel adsorbents and their environmental applications, AIChE Symp. Ser. 309 (1995) 18–31.

229

[11] J. De. Wilde, G.B. Marin, Investigation of simultaneous adsorption of SO2 and NOx on Na-(-alumina with transient techniques, Catal. Today 62 (2000) 319–328. [12] S. Blonski, S.H. Garofalini, Molecular dynamics simulations of ␥-alumina and ␥-alumina surfaces, Surf. Sci. 295 (1993) 263–274. [13] I. Levin, D. Brandon, Metastable alumina polymorphs: crystal structures and transition sequences, J. Am. Ceram. Soc. 81 (1998) 1995–2012. [14] A. Datta, R.G. Cavell, R.W. Tower, Z.M. George, Claus catalysis. 1. Adsorption of sulfur dioxide on the alumina catalyst studied by FTIR and EPR spectroscopy, J. Phys. Chem. 89 (3) (1985) 443–449. [15] S.W. Nam, G.R. Gavalas, Adsorption and oxidative adsorption of sulfur dioxide on (-alumina, Appl. Catal. 55 (1989) 193–213. [16] S. Andersson, R. Pompe, N. Vannerberg, SOx adsorption/desorption processes on (-alumina for SOx transfer catalyst, Appl. Catal. 16 (1985) 49–58. [17] B. Westerberg, E. Fridell, A transient FTIR study of species formed during NOx storage in the Pt/BaO/Al2 O3 system, J. Mol. Catal. A: Chem. 165 (2001) 249–263. [18] R. Burch, E. Halpin, J.A. Sullivan, A comparison of the selective catalytic reduction of NOx over Al2 O3 and sulphated Al2 O3 using, CH3 OH and C3 H8 as reductants, Appl. Catal. B: Environ. 17 (1998) 115–129. [19] P. Li, G.Z. Lu, Effect of SO2 on NO catalytic oxidation (V): the mechanism of SO2 over NiO/␥-Al2 O3 , Acta Chim. Sin. 61 (2003) 660–665. [20] N. Apostolescu, T. Schröder, S. Kureti, Study on the mechanism of the reaction of NO2 with aluminium oxide, Appl. Catal. B: Environ. 51 (2004) 43–50. [21] I. Tomohiro, K. Hajime, Y. Tsutomu, Initial step of ﬂue gas desulfurization-an IR study of the reaction of SO2 with NOx on CaO, Environ. Sci. Technol. 34 (2000) 2799–2803. [22] C.C. Chang, Infrared studies of SO2 on ␥-alumina, J. Catal. 53 (1978) 374–385. [23] W.S. Epling, L.E. Campbell, A. Yezerets, Overview of the fundamental reactions and degradation mechanisms of NOx storage/reduction catalysts, Catal. Rev. 46 (2) (2004) 163–245. [24] J. De. Wilde, A.K. Das, G.H. Heynderickx, G.B. Marin, Development of a transient kinetic model for the simultaneous adsorption of SO2 −NOx over Na/␥-Al2 O3 sorbent, Ind. Eng. Chem. Res. 40 (2001) 119–130.

Journal of Hazardous Materials 195 (2011) 230–237

Contents lists available at ScienceDirect

Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat

Inoculation of endophytic bacteria on host and non-host plants—Effects on plant growth and Ni uptake Ying Ma a,c,∗ , Mani Rajkumar b , YongMing Luo a , Helena Freitas c a b c

Key Laboratory of Soil Environment and Pollution Remediation, Institute of Soil Science, Chinese Academy of Sciences, Nanjing 210008, China National Environmental Engineering Research Institute (NEERI), CSIR Complex, Taramani, Chennai 600113, India Centre for Functional Ecology, Department of Life Sciences, University of Coimbra, Coimbra 3001-401, Portugal

a r t i c l e

i n f o

Article history: Received 10 May 2011 Received in revised form 9 August 2011 Accepted 10 August 2011 Available online 17 August 2011 Keywords: Phytoremediation Endophyte Plant growth promoting traits Hydrolyzing enzymes Ni contaminated soil

a b s t r a c t Among a collection of Ni resistant endophytes isolated from the tissues of Alyssum serpyllifolium, four plant growth promoting endophytic bacteria (PGPE) were selected based on their ability to promote seedling growth in roll towel assay. Further, the PGPE screened showed the potential to produce plant growth promoting (PGP) substances and plant polymer hydrolyzing enzymes. These isolates were further screened for their PGP activity on A. serpyllifolium and Brassica juncea under Ni stress using a phytagar assay. None of the four isolates produced any disease symptoms in either plant. Further, strain A3R3 induced a maximum increase in biomass and Ni content of plants. Based on the PGP potential in phytagar assay, strain A3R3 was chosen for studying its PGP effect on A. serpyllifolium and B. juncea in Ni contaminated soil. Inoculation with A3R3 signiﬁcantly increased the biomass (B. juncea) and Ni content (A. serpyllifolium) of plants grown in Ni contaminated soil. The strain also showed high level of colonization in tissue interior of both plants. By 16S rRNA gene sequencing analysis, A3R3 was identiﬁed as Pseudomonas sp. Successful colonization and subsequent PGP potentiality of Pseudomonas sp. A3R3 indicate that the inoculation with PGPE might have signiﬁcant potential to improve heavy metal phytoremediation. © 2011 Elsevier B.V. All rights reserved.

1. Introduction Industrial operations such as mining, smelting, metal forging, manufacturing of alkaline storage batteries, combustion of fossil fuel, and sewage sludge, cause accumulation of metals or metalloids in natural resources such as soil, water and air. Since the heavy metals seriously affect terrestrial and aquatic ecosystems and induce potential health risks [1], various physicochemical and biological methods have been developed to remove the metals from the environment. Phytoremediation refers to the use of plants that can uptake high levels of heavy metals from soil and accumulate them in a harvestable part [2]. Although the efﬁciency of heavy metal phytoremediation is dependent on an adequate yield of plants and their capacity for metal ion accumulation, the plant associated beneﬁcial microbes also play signiﬁcant roles as they can provide nutrients and reduce the deleterious effects of metals to the plants [3,4]. Considering such beneﬁcial features, it may be envisaged that inoculation of metal resistant plant growth promot-

∗ Corresponding author at: Key Laboratory of Soil Environment and Pollution Remediation, Institute of Soil Science, Chinese Academy of Sciences, Nanjing 210008, China. Tel.: +86 025 86881844; fax: +86 025 86881126. E-mail address: [email protected] (Y. Ma). 0304-3894/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.jhazmat.2011.08.034

ing bacteria would increase plant growth and phytoremediation potential in metal contaminated soils [5–7]. In recent years, the use of metal resistant endophytic bacteria in phytoremediation of heavy metal contaminated soils has attracted more attention [8–11]. Although the heavy metals such as Ni, Pb are toxic to plants and their associated microbes at high concentrations, the metal resistant plant growth promoting endophytic bacteria (PGPE) have been reported to occur widely in tissue interiors of various hyperaccumulator plants [12–14]. This indicates that endophytic bacteria have evolved to be resistant to high levels of heavy metals and that they might confer to the plant higher tolerance to heavy metal stress. Moreover, the endophytic bacteria are proved to be able to enhance the plant growth by various mechanisms including production of siderophores, 1-aminocyclopropane-1carboxylic acid (ACC) deaminase, indole-3-acetic acid (IAA) or solubilization of phosphate (P) [15]. In addition, certain endophytes have also been shown to alter heavy metal toxicity/availability to the plant by producing siderophores, biosurfactants and organic acids [10,16]. Despite these beneﬁcial actions on plant, however, the endophytic bacteria must be compatible with various hyperaccumulators and able to colonize the tissues of the host plants without producing any disease symptoms. Because of the ability to produce the plant growth beneﬁcial substance in metal stressed environment, the colonization potential of PGPE in the rhizosphere and/or tissue interior of plants has been considered as a major

Y. Ma et al. / Journal of Hazardous Materials 195 (2011) 230–237

factor that determines the inoculum efﬁciency for microbial assisted heavy-metal phytoremediation [4,10]. Although the endophytic bacteria are reported to be present in various plants growing on heavy metal contaminated soils, only a few attempts have been made to study their role on the growth and Ni accumulation in plants. Moreover, the role of PGPE on the growth and phytoremediation potential of a non-host plant in metal contaminated soils has not been adequately deﬁned. The objectives of our study were (1) to isolate and characterize Ni resistant PGPE from the tissues of the Ni hyperaccumulator plant, Alyssum serpyllifolium ssp. lusitanicum, (2) to test whether the PGPE with known plant growth promoting (PGP) traits promote the plant growth and Ni accumulation in their host (A. serpyllifolium) and non-host plants (Brassica juncea L. Czern.), and (3) to select a Ni resistant PGPE strain which might be useful to increase the plant Ni uptake and biomass production for improving the efﬁciency of phytoremediation of Nicontaminated soils.

231

IAA production by PGPE was determined as described previously [24]. Cultures of the isolates were raised in LB broth amended with 500 ␮g of tryptophan mL−1 at 27 ◦ C for 96 h at 200 rpm. Bacterial cells were removed by centrifugation at 7000 rpm and the supernatant was analyzed for IAA. The phosphate solubilizing activity of the isolates was analyzed in modiﬁed Pikovskayas medium [25] amended with tricalcium phosphate. The isolates were grown at 27 ◦ C for 192 h at 200 rpm. The solubilized phosphate in the culture supernatant was quantiﬁed as described by Fiske and Subbarow [26]. Siderophores production by PGPE was determined using in chrome azurol S (CAS) agar medium [27]. The presence of catechol and hydroxamate type siderophores in culture supernatants obtained from bacteria grown under iron-limiting conditions in casamino acids (CAA) medium was quantitatiﬁed according to the method of Ma et al. [18]. 2.3. Extracellular enzyme activities

2. Materials and methods 2.1. Isolation of Ni resistant PGPE Endophytic bacteria were isolated from leaves, stems and roots of Ni accumulators, A. serpyllifolium grown in serpentine soils in Braganc¸a, north-east of Portugal, previously described by Freitas et al. [17]. Brieﬂy, plant samples were washed with tap water followed by three rinses with deionized water and then separated into roots, stems and leaves. Healthy plant tissues were sterilized by sequential immersion in 70% (v/v) ethanol for 1 min, and 3% sodium hypochlorite for 3 min and washed three times with sterile water to remove surface sterilization agents. In order to conﬁrm the surface disinfection process was successful, sterility was checked by plating on Luria–Bartani (LB) agar. No contamination was found. After surface sterilization, the leaf, stem or root tissue was cut and titrated in distilled water; appropriate dilutions were plated onto sucrose-minimal salts low-phosphate (SLP) agar medium (sucrose 1%; (NH4 )2 SO4 0.1%; K2 HPO4 0.05%; MgSO4 0.05%; NaCl 0.01%; yeast extract 0.05%; CaCO3 0.05%; pH 7.2) amended with 50 mg of Ni L−1 (NiCl2 ). To isolate Ni resistant strains, the bacterial strains picked from the Ni resistant colonies were puriﬁed on the LB agar medium containing 50 mg L−1 of Ni and gradually taken to higher concentration of Ni (100–1000 mg L−1 ) according to the procedure of Ma et al. [18]. In order to isolate the Ni resistant PGPE, the Ni resistant strains were assessed for the PGP activity by roll towel method [19]. Seeds of A. serpyllifolium obtained from the Botanical Garden of the University of Coimbra, Coimbra, Portugal, were surface sterilized in 2% Ca(OCl)2 (2 h) and rinsed several times with sterile distilled water. The seeds were inoculated by soaking in a bacterial suspension containing 108 cell mL−1 for 2 h then placed in wet blotters and incubated in a growth chamber for 30 days. The vigour index was calculated as described by Abdul-Baki and Anderson [20]. 2.2. Characterization of PGP traits of endophytic bacteria To determine ACC deaminase activity, the PGPE were grown in test tubes containing 10 mL of Dworkin–Foster (DF) salts minimal medium [21]. The medium was supplemented with 3 mM ACC. After cultivation for 72 h at 27 ◦ C, the cells were harvested by centrifugation at 9000 rpm for 10 min at room temperature. The ACC deaminase activity in cells was determined by monitoring the amount of ␣-ketobutyrate generated by the enzymatic hydrolysis of ACC as described by Belimov et al. [22]. The protein concentration of cell suspensions was determined by the method of Bradford [23].

Cellulase and pectinase activities were assayed on the indicator plates. For the cellulase assay, nitrogen-freebase (NFB) [28] plates supplemented with 0.2% carboxymethyl cellulose and 0.5% tryptone were spotted with bacterial cells. After incubating for 48 h at 30 ◦ C the plates were ﬂooded with congo red (1 mg mL−1 ) solution for 30 min. Excess stain was discarded and the agar was destained with 1 M of NaCl solution [29]. Plates were kept overnight at 4 ◦ C and examined for clearing zone around the point of inoculations. For the pectinase assay, the bacterial isolates were spotted on nutrient agar supplemented with 0.5% pectin. After incubating the plates for 5 days at 30 ◦ C, the surface of the medium was overlayed with 2% hexadecyl trimethyl ammonium bromide (CTAB) solution for 30 min. CTAB solution was then discarded and the plate surface was washed with 1 M NaCl to visualize the zone around the bacterial growth [30]. 2.4. Phytagar assay This experiment was carried out to screen the Ni resistant PGPE for their ability to promote the growth and Ni accumulation of A. serpyllifolium and B. juncea growing in Ni treated agar media (5 mg Ni L−1 ). The growth media were prepared using 0.5% phytagar and one-quarter strength Hoagland’s nutrient solution with or without Ni. The surface sterilized seeds of A. serpyllifolium and B. juncea were inoculated with PGPE as detailed in earlier section and placed in 150 mL test tubes containing 25 mL of phytagar. The tubes were closed, placed in a growth chamber, and harvested after 60 days. At harvest, the plants were removed from the boxes and rinsed thoroughly in distilled water to remove adhering agar. Growth parameters such as fresh weight and dry weight of the plants were measured. The total Ni accumulation in plants was also determined [17]. 2.5. Pot experiment For pot experiments, the soil was collected from the Botanical Garden of the University of Coimbra, Coimbra, Portugal. The physicochemical properties of soil were: pH (1:1, w/v water) 7.3; organic matter 1.6%; nickel 18 mg kg−1 ; zinc 86 mg kg−1 ; chromium 31 mg kg−1 . The soil was sieved (2 mm) and sterilized by steaming (100 ◦ C for 1 h on three consecutive days). After sterilization the soil was amended with aqueous solution of NiCl2 to achieve the ﬁnal concentrations of 150, 300 or 450 mg Ni kg−l and left in a greenhouse for a 2-week period (for metal stabilization). Before inoculation, the mutant of A3R3 marked with antibiotic resistance was obtained after plating of the parental strain onto LB agar amended with ampicillin (75 mg L−1 ). The surface sterilized seeds

232

Y. Ma et al. / Journal of Hazardous Materials 195 (2011) 230–237

of A. serpyllifolium ssp. lusitanicum and B. juncea L. Czern. were inoculated with A3R3 as detailed in earlier section. Seeds soaked in sterile water were used as control. The inoculated and noninoculated seeds were planted in plastic pot containing 300 g of soil. The plants were grown in a greenhouse at 25 ◦ C and a 16/8 day/night regime. After 60 days, the plants were carefully removed from the pots and the root surface was cleaned several times with distilled water. Plant fresh weight and dry weight were measured. The accumulation of Ni in root and shoot system was quantiﬁed as described above. The population dynamics of introduced bacteria was also studied using the intrinsic antibiotic marker. The plant interior colonization was quantiﬁed according to the procedure of the above endophytic bacterial isolation. The resulting suspensions were evaluated for colony forming units (CFU) according to the dilution-plate method on LB agar with addition of 75 mg L−1 ampicillin. The plates were incubated for 4 days at 28 ◦ C. 2.6. Genetic characterization of Ni-resistant PGPE For genotypic characterization, the PGPE strain was grown in LB medium in presence of 1 mM Ni for 20 h and total DNA was isolated using standard procedure [31]. The 16S rDNA was ampliﬁed using the primers pA (5 -AGAGTTTGATCCTGGCTCAG; Escherichia coli bases 8–27) and pC5B (5 -TACCTTGTTACGACTT; E. coli bases 1507–1492) [32] under the reaction conditions described by Branco et al. [33]. Partial nucleotide sequence of the ampliﬁed 16S rDNA was determined using automated DNA sequencer. The sequences obtained were matched against nucleotide sequences present in GenBank using the BLASTn program [34]. 2.7. Statistical analysis Analysis of variance (ANOVA) followed by post hoc Fisher Least Signiﬁcant Difference (LSD) test (p < 0.05) were used to compare treatment means. All the statistical analyses were carried out using SPSS 10.0. 3. Results and discussion 3.1. Isolation of Ni resistant PGPE The plants growing in metal contaminated soils accumulate higher amounts of heavy metals and can therefore provide a metalstressed environment for endophytes where they can develop mechanisms to resist the toxic effects of metals [13]. Moreover, these metal resistant endophytes promote host plant growth by improving mineral nutrition or conferring tolerance to various biotic and abiotic stresses [15]. During the initial screening (50 mg L−1 ), 95 Ni resistant bacterial strains were isolated from root, stem and leaf tissues of A. serpyllifolium. After secondary screening, 27 bacterial strains showing a high degree of Niresistance were selected (data not shown). In order to isolate the PGPE, the metal resistant strains were assessed for PGP activity

Table 1 Inﬂuence of Ni resistant endophytes on root length, shoot length and vigour index of A. serpyllifolium. Treatment

Shoot length (cm)

Control A3R3 A3S4 A2R6 A3S6

2.0 2.9 2.5 2.6 2.6

± ± ± ± ±

0.2 b 0.2 a 0.2 a 0.2 a 0.3 a

Root length (cm) 1.9 2.5 2.3 2.2 2.3

± ± ± ± ±

0.3 b 0.2 a 0.2 ab 0.3 ab 0.2 ab

Vigour index 312.0 483.0 408.0 402.3 422.2

± ± ± ± ±

21.2 c 27.5 a 22.5 b 41.9 b 19.6 b

Average ± standard deviation from three samples. Data of columns indexed by the same letter are not signiﬁcantly different between Ni resistant endophyte treatments according to Fisher’s protected LSD test (p < 0.05). Vigour index = germination (%) × seedling length (shoot length + root length).

on A. serpyllifolium by roll towel method. Among the 27 strains tested, four isolates, namely A3R3, A3S4, A2R6 and A3S6 induced an increase in root length, shoot length and vigour index of A. serpyllifolium (Table 1). However, A3R3 induced a maximum increase in root length, shoot length and vigour index by 32%, 45% and 55%, respectively, compared with non-inoculated control. 3.2. Characterization of Ni resistant PGPE Since the PGPE could exert their beneﬁcial effects on host plant by several mechanisms including nitrogen ﬁxation, phosphate solubilization, IAA and siderophore production [35], the PGP traits of Ni resistant PGPE were further investigated in detail. Assessment of the parameters of PGP revealed the intrinsic ability of the Ni resistant PGPE for the utilization of ACC as the sole nitrogen source, production of IAA, siderophore and solubilization of phosphate (Table 2). The role of ACC deaminase in decreasing stress ethylene levels by the enzymatic hydrolysis of ACC into ␣-ketobutyric acid and ammonia has been presented as one of the major mechanisms of PGPE in promoting root and plant growth [36]. Among the four strains tested, strain A3R3 recorded the highest ACC deaminase activity followed by A3S4. Another important PGP mechanism is the solubilization of P, by which microbes enhance P availability to the host plant [37]. The strains A3R3 and A3S6 showed the phosphate solubilizing ability by utilizing the insoluble tricalcium phosphate in modiﬁed Pikovskayas medium. Further screening of the production of IAA by Ni resistant PGPE indicated that all the four strains utilized l-tryptophan as a precursor for growth and IAA production. However, strain A3R3 produced the highest amount, 69.4 mg L−1 of IAA, whereas A2R6 produced only 43.8 mg L−1 of IAA. The IAA released by bacteria enhances plant growth directly by stimulating elongation of the cell or affecting cell division [38]. Besides, A3R3, A3S4 and A3S6 showed the production of catechol and hydroxamate type siderophores in iron-restricted conditions in CAA medium. It is known that the siderophores produced by bacteria bind to the unavailable form of Fe3+ and make iron available to the plants, leading thereby to an increase in plant growth [39]. Production of plant cell wall degrading enzymes such as cellulase and pectinase was analyzed because this is an important mechanism for endophytic colonization [40]. Among the four

Table 2 Some key traits of Ni resistant endophytes. Parameters

ACC deaminase, ␣-ketobutyrate mg−1 protein h−1 (nmol) Phosphate solubilization (mg L−1 ) IAA synthesis (mg L−1 ) Catechol-type siderophore production (mg L−1 ) Hydroxamate-type siderophore production (mg L−1 ) Ni tolerance level (mg L−1 ) Average ± standard deviation from three samples. Nd: not detected.

Endophytic bacterial strain A3R3

A3S4

A2R6

A3S6

67.9 ± 6.2 138.2 ± 21.4 69.4 ± 3.2 83.3 ± 7.5 60.5 ± 6.3 750

40.0 ± 1.2 nd 53.0 ± 1.6 69.1 ± 3.9 79.8 ± 3.6 750

24.5 ± 1.4 nd 43.8 ± 1.6 nd nd 1000

nd 83.5 ± 11.1 60.8 ± 3.8 47.3 ± 2.1 31.9 ± 2.1 750

Y. Ma et al. / Journal of Hazardous Materials 195 (2011) 230–237

strains tested, A3R3 displayed a positive cellulase activity, as indicated by the development of yellow-colored zone on NFB plates. However, all the four strains exhibited pectinase activity, which was expected considering the endophytic nature of the strains. Jha and Kumar [41] recently reported the existence of cell wall-hydrolyzing enzyme mediated endophytic colonization and suggested that cellulose and pectinase produced by Klebsiella oxytoca GR-3 might play an important role in plant–microbe interactions and intercellular colonization of root. 3.3. Screening of Ni resistant PGPE In order to isolate an efﬁcient PGPE, the Ni resistant PGPE (A3R3, A3S4, A2R6 and A3S6) were screened for their ability to promote the growth and Ni accumulation of A. serpyllifolium and B. juncea in Ni treated phytagar medium. It was interesting to ﬁnd that none of the introduced Ni resistant PGPE showed any signs of pathogenicity towards A. serpyllifolium and B. juncea. In the absence of Ni, inoculation of PGPE induced an increase in fresh and dry weight of both A. serpyllifolium and B. juncea (Fig. 1). However, the maximum PGP effect was observed in A3R3. In the case of A. serpyllifolium, the strain A3R3 enhanced plant fresh weight and dry weight by 87% and 60%, respectively. Similarly, in B. juncea the strain A3R3 enhanced the fresh weight and dry weight by 162% and 177%, respectively. The application of Ni (5 mg Ni L−1 ) to the phytagar medium did not affect the growth of A. serpyllifolium. However, B. juncea exposed to Ni demonstrated a signiﬁcant (p < 0.05) reduction in plant growth. Reduction in the growth with application of Ni has been reported in various plant species [18,42]. Alteration of fundamental physiological/biochemical processes, e.g. leaf photosynthetic and transpiration activities [43], leaf chlorophyll content [44], have been attributed to excess Ni which could decrease fresh and dry matter yield. A. serpyllifolium inoculated with Ni resistant

233

PGPE exhibited an increase in plant fresh and dry weight in the presence of Ni. However, the highest PGP effect was found for strain A3R3, which enhanced plant fresh and dry weight by 185% and 175%, respectively, compared with non-inoculated plants. Similarly the maximum PGP effect on Ni-treated B. juncea was observed after inoculation with strain A3R3. Metal resistant PGPE belonging to different genera such as Pseudomonas, Microbacterium, Methylobacterium and Burkholderia were found to have PGP characteristics that can potentially promote the plant growth and reduce metal stress symptoms in plants [10,11,45]. In addition to plant growth promotion, PGPE strain A3R3 signiﬁcantly increased the Ni concentration in A. serpyllifolium and B. juncea by 36% and 20%, respectively (Fig. 2). Similar observations were also made by Zhang et al. [46]. The authors found that the inoculation of Brassica napus growing in 2.5 mg kg−1 of Cu-contaminated substrates with Cu resistant endophyte Pantoea agglomerans Jp3-3 signiﬁcantly increased the concentration of Cu in above-ground tissues and roots by 31% and 78%, respectively, compared with respective non-inoculated control. The results clearly indicate that, among the four Ni resistant PGPE tested, the strain A3R3 was highly efﬁcient at protecting both A. serpyllifolium and B. juncea from growth inhibition caused Ni and at enhancing the uptake of Ni by plants. 3.4. Inﬂuence of A3R3 on Ni toxicity in plants grown in Ni contaminated soil Based on the promotion of plant growth and Ni accumulation of plants in phytagar assay, the strain A3R3 was selected for studying the effects of the strain on the plant growth and the uptake of Ni by A. serpyllifolium and B. juncea in soil. In the absence of Ni, inoculation of A3R3 did not greatly inﬂuence the growth of A. serpyllifolium (Fig. 3a and b) Fig. 3. However, B. juncea inoculated with A3R3 demonstrated a signiﬁcant (p < 0.05) increase in plant

Fig. 1. Inﬂuence of Ni resistant PGPE on the fresh weight and dry weight of A. serpyllifolium (a, b) and B. juncea (c, d) grown in Ni-amended phytagar. Each value is the mean of triplicates. Error bars represent standard deviation. Data of columns indexed by the same letter are not signiﬁcantly different according to Fisher’s protected LSD test (p < 0.05).

234

Y. Ma et al. / Journal of Hazardous Materials 195 (2011) 230–237

Fig. 2. Ni concentration in A. serpyllifolium and B. juncea grown in Ni-amended phytagar (5 mg L−1 ). Each value is the mean of triplicates. Error bars represent standard deviation. Data of columns indexed by the same letter are not signiﬁcantly different according to Fisher’s protected LSD test (p < 0.05).

fresh and dry weight (Fig. 3c and d). As shown in phytagar experiments, the application of Ni to the soil did not affect the growth of A. serpyllifolium negatively, but caused a signiﬁcant increase in plant fresh and dry biomass. The beneﬁcial effects of Ni on plant growth have already been reported in Ni hyperaccumulators, Alyssum lesbiacum, A. bertolonii, and Thlaspi goesingenses [47]. In contrast to A. serpyllifolium, the non-inoculated B. juncea exposed to different concentrations of Ni demonstrated a signiﬁcant (p < 0.05) inhibition in plant growth. Panwar et al. [48] also reported similar results in B. juncea with increasing Ni content of soil (0–80 mg kg−1 ). These results are not surprising, since serpentinophytes including

A. serpyllifolium are generally understood as Ni hyperaccumulators that is well adapted to a high Ni concentration that B. juncea cannot tolerate [48,49]. A3R3 inoculations did not exhibit great inﬂuence on fresh and dry weight of A. serpyllifolium in the presence of Ni as compared with non-inoculated plants (Fig. 3a and b), whereas B. juncea inoculated with A3R3 exhibited a signiﬁcant increase in plant fresh and dry weight in the presence of different concentrations of Ni. For instance, the strain A3R3 increased the fresh weight and dry weight of B. juncea by 50% and 45%, respectively, even at 450 mg Ni kg−1 soil, compared to non-inoculated but amended with the same dose

Fig. 3. Inﬂuence of Ni resistant PGPE A3R3 on the fresh weight and dry weight of A. serpyllifolium (a, b) and B. juncea (c, d) grown in Ni-amended soil. Each value is the mean of triplicates. Error bars represent standard deviation. Data of columns indexed by the same letter are not signiﬁcantly different according to Fisher’s protected LSD test (p < 0.05).

Y. Ma et al. / Journal of Hazardous Materials 195 (2011) 230–237

235

Table 3 Colonization of strain A3R3 in the root and shoot interior of A. serpyllifolium and B. juncea (values in log CFU g−1 of shoot/root fresh weight). Ni concentration (mg kg−1 )

A. serpyllifolium Shoot interior

0 150 300 450

4.62 4.60 4.56 4.50

± ± ± ±

0.02a 0.02a 0.01ab 0.10b

B. juncea Root interior 4.71 4.64 4.55 4.38

± ± ± ±

0.02a 0.01b 0.04c 0.02d

Shoot interior 4.58 4.45 4.27 4.19

± ± ± ±

0.05a 0.04b 0.07c 0.08c

Root interior 4.65 4.62 4.44 4.30

± ± ± ±

0.08a 0.09a 0.09b 0.11b

Each value is the mean of triplicates. Error bars represent standard deviation. Data of columns indexed by the same letter are not signiﬁcantly different according to Fisher’s protected LSD test (p < 0.05).

of Ni. This result is in agreement with a previous report describing increased biomass production of Nicotiana tabacum inoculated with Cd resistant endophytes and grown in Cd-supplemented soil [9]. In general, high concentrations of heavy metals in the rhizosphere soil interfere with the uptake of essential nutrients such as P, Fe and lead to plant nutrient deﬁciency and growth retardation [50]. Under such condition PGPE offer a biological rescue system capable of solubilizing/scavenging/mobilizing mineral nutrients (e.g. Fe, P, Zn) of soil and make them available to the plant roots during the initial colonization [37,51]. Further, ACC deaminase producing bacteria have been reported to reduce the deleterious effects of heavy metals on root elongation by reducing the stress ethylene level through hydrolytic cleavage of its precursor ACC [45,52]. In our study the observed beneﬁts on the growth of A. serpyllifolium and B. juncea may be attributed to cumulative effects of A3R3, such as cleavage of ACC to ␣-ketobutyrate and ammonia by ACC deaminase, supply of Fe and P to the crop in addition to growth promoting substance IAA produced by this organism. Although the PGPE possess several traits to promote the plant growth, colonization and survival in metal stress environment are very important factors for microbial assisted phytoremediation, as the activity of inoculated PGPE is necessary to produce beneﬁcial substances. Hence, survival rate of the inoculated bacteria in plant tissue interiors was assessed. Though A3R3 originally isolated from

the root tissues of A. serpyllifolium, it showed high level of colonization in shoot and root interior of both plants (Table 3). These observations strongly suggest that strain A3R3 is non-host speciﬁc colonizer and can move within tissues of the plant. Further, the endophytic nature of A3R3 is also evident from the presence of plant cell wall hydrolyzing enzymes, pectinase and cellulase by which the bacteria enter and colonize the plant tissues [53]. Further, the population density of A3R3 in tissue interior of A. serpyllifolium and B. juncea was not signiﬁcantly affected in Ni (150 and 300 mg kg−1 ) treated soil. However, at the highest concentration of Ni (450 mg kg−1 ) a slight decrease in the population density was observed. Since most plant–microbe interactions are initiated at the level of colonization, the survival potential of the colonized microbes in metal stressed environment is likely to be closely linked to their metal resistance level. In addition to successful colonization in both host and non-host plants, the strain A3R3 was able to grow at increasing Ni concentrations (from 150 to 450 mg Ni L−1 ) in liquid medium (data not shown) indicating an adaptation of A3R3 in Ni contaminated environment. Overall, our results indicate that A3R3 can form sustaining endophytic populations in the tissue interior of A. serpyllifolium and B. juncea and exhibit beneﬁcial effects on plants irrespective of Ni stress. Fig. 4 shows the Ni distribution proﬁle in shoot and root systems of A. serpyllifolium and B. juncea grown at varying concentrations

Fig. 4. Ni concentration in shoot and root of A. serpyllifolium (a, b) and B. juncea (c, d) grown in Ni-amended soil. Each value is the mean of triplicates. Error bars represent standard deviation. Data of columns indexed by the same letter are not signiﬁcantly different according to Fisher’s protected LSD test (p < 0.05).

236

Y. Ma et al. / Journal of Hazardous Materials 195 (2011) 230–237

of Ni in the soil. Accumulation of Ni in the root and shoot systems increased with increase in the initial concentration of Ni in soil. However, A. serpyllifolium accumulated more Ni in both the shoot and root tissues compared with B. juncea. Further, the inoculation of the strain A3R3 signiﬁcantly increased the accumulation of Ni in the tissues of A. serpyllifolium and B. juncea compared with respective non-inoculated controls. For instance, strain A3R3 increased the Ni concentration in the shoot tissues of A. serpyllifolium and B. juncea by 10% and 15%, respectively, when plants were grown in soil amended with 450 mg Ni kg−1 compared to respective non-inoculated plants. Recently, Sheng et al. [10] reported that inoculation of B. napus with PGPE P. ﬂuorescens signiﬁcantly increased the plant uptake of Pb when compared with the dead bacterial-inoculation control. They attributed this effect to the ability of PGPE to produce siderophores and to reduce soil pH. Similarly, Zaidi et al. [54] have also reported that microbial solubilizing inorganic phosphates facilitate the uptake of the metals from soil. In our study, the signiﬁcant increase in Ni accumulation of plants caused by A3R3 could be attributed to the production of siderophore and the solubilization of P. Analysis of the 1224 bp 16S rDNA sequence of the strain A3R3 using the BLASTn program at NCBI showed 100% sequence homology to Pseudomonas sp. Based on the biochemical features (data not shown) and 16S rDNA sequence analysis, the isolate A3R3 was identiﬁed as a strain of Pseudomonas sp. The sequences obtained have been deposited in NCBI databases under the accession number GU550663. 4. Conclusion Modiﬁcation of plants to obtain organisms with improved phytoremediation capabilities is generally carried out by integrating foreign DNA into plant genomes to produce transgenic plants [55]. Although the genetic manipulation may be a promising approach, these methods are expensive, time consuming and dependent on speciﬁc plant being studied. As an alternative approach, PGPE have been used to improve the phytoremediation efﬁciency without requiring integration of foreign DNA into the plant genome. Our study demonstrated that the inoculation of metal resistant PGPE, Pseudomonas sp. A3R3 seemed to be effective in promoting the phytoremediation potential of both host (A. serpyllifolium) and nonhost (B. juncea) plants by improving either the Ni accumulation or biomass production. Although PGPE Pseudomonas sp. A3R3 might have signiﬁcant potential to colonize tissue interior of host and non-host plants, not every PGPE possesses the ability to colonize multiple plants, and not every plants that has the potential to harbour several PGPE. Hence, further studies, including the role of Pseudomonas sp. A3R3 on the growth and phytoremediation potential of various hyperaccumulator plants in metal contaminated soils are under progress in order to test the usefulness of this novel isolate for future phytoremediation application. Acknowledgements Y. Ma thankfully acknowledges the ﬁrst-class ﬁnancial support from the China Postdoctoral Science Foundation (Grant No. 20110490137). M. Rajkumar acknowledges the Department of Biotechnology (DBT), Government of India for the ﬁnancial support through Ramalingaswami Re-entry Fellowship. References [1] A.G. Khan, C. Kuek, T.M. Chaudhry, C.S. Khoo, W.J. Hayes, Plants, mycorrhizae and phytochelators in heavy metal contaminated land remediation, Chemosphere 41 (2000) 197–207. [2] P.B.A.N. Kumar, V. Dushenkov, H. Motto, I. Raskin, Phytoextraction: the use of plants to remove heavy metals, Environ. Sci. Technol. 29 (1995) 1232–1238.

[3] A.A. Belimov, V.I. Safronova, T. Mimura, Response of spring rape to inoculation with plant growth-promoting rhizobacteria containing 1-aminocyclopropane1-carboxylate deaminase depends on nutrient status of the plant, Can. J. Microbiol. 48 (2002) 189–199. [4] Y. Ma, M. Rajkumar, J. Vicente, H. Freitas, Inoculation of Ni-resistant plant growth promoting bacterium Psychrobacter sp. strain SRS8 for the improvement of nickel phytoextraction by energy crops, Int. J. Phytoremediation 13 (2011) 126–139. [5] Y. Gao, C. Miao, L. Mao, P. Zhou, Z. Jin, W. Shi, Improvement of phytoextraction and antioxidative defense in Solanum nigrum L. under cadmium stress by application of cadmium-resistant strain and citric acid, J. Hazard. Mater. 181 (2010) 771–777. [6] M. Kuffner, S. De Maria, M. Puschenreiter, K. Fallmann, G. Wieshammer, M. Gorfer, J. Strauss, A.R. Rivelli, A. Sessitsch, Culturable bacteria from Zn- and Cd accumulating Salix caprea with differential effects on plant growth and heavy metal availability, J. Appl. Microbiol. 108 (2010) 1471–1484. [7] Y. Ma, M.N.V. Prasad, M. Rajkumar, H. Freitas, Plant growth promoting rhizobacteria and endophytes accelerate phytoremediation of metalliferous soils, Biotechnol. Adv. 29 (2011) 248–258. [8] S.L. Doty, Enhancing phytoremediation through the use of transgenics and endophytes, New Phytol. 179 (2008) 318–333. [9] C. Mastretta, S. Taghavi, D. van der Lelie, A. Mengoni, F. Galardi, C. Gonnelli, T. Barac, J. Boulet, N. Weyens, J. Vangronsveld, Endophytic bacteria from seeds of Nicotiana tabacum can reduce cadmium phytotoxicity, Int. J. Phytoremediation 11 (2009) 251–267. [10] X.F. Sheng, J.J. Xia, C.Y. Jiang, L.Y. He, M. Qian, Characterization of heavy metal-resistant endophytic bacteria from rape (Brassica napus) roots and their potential in promoting the growth and lead accumulation of rape, Environ. Pollut. 156 (2008) 1164–1170. [11] L.N. Sun, Y.F. Zhang, L.Y. He, Z.J. Chen, Q.Y. Wang, M. Qian, X.F. Sheng, Genetic diversity and characterization of heavy metal-resistant-endophytic bacteria from two copper-tolerant plant species on copper mine wasteland, Bioresource Technol. 101 (2010) 501–509. [12] R. Barzanti, F. Ozino, M. Bazzicalupo, R. Gabbrielli, F. Galardi, C. Gonnelli, A. Mengoni, Isolation and characterization of endophytic bacteria from the nickel hyperaccumulator plant Alyssum bertolonii, Microb. Ecol. 53 (2007) 306–316. [13] R. Idris, R. Trifonova, M. Puschenreiter, W.W. Wenzel, A. Sessitsch, Bacterial communities associated with ﬂowering plants of the Ni hyperaccumulator Thlaspi goesingense, Appl. Environ. Microbiol. 70 (2004) 2667–2677. [14] C. Lodewyckx, S. Taghavi, M. Mergeay, J. Vangronsveld, H. Clijsters, D. van der Lelie, The effect of recombinant heavy metal resistant endophytic bacteria in heavy metal uptake by their host plant, Int. J. Phytoremediation 3 (2001) 173–187. [15] M. Rajkumar, N. Ae, H. Freitas, Endophytic bacteria and their potential to enhance heavy metal phytoextraction, Chemosphere 77 (2009) 153–160. [16] V.S. Saravanan, M. Madhaiyan, M. Thangaraju, Solubilization of zinc compounds by the diazotrophic, plant growth promoting bacterium Gluconacetobacter diazotrophicus, Chemosphere 66 (2007) 1794–1798. [17] H. Freitas, M.N.V. Prasad, J. Pratas, Analysis of serpentinophytes from north–east of Portugal for trace metal accumulation relevance to the management of mine environment, Chemosphere 54 (2004) 1625–1642. [18] Y. Ma, M. Rajkumar, H. Freitas, Isolation and characterization of Ni mobilizing PGPB from serpentine soils and their potential in promoting plant growth and Ni accumulation by Brassica spp., Chemosphere 75 (2009) 719–725. [19] ISTA, International rules for seed testing, Proc. Int. Seed Test Assoc. 31 (1966) 1–152. [20] A.A. Abdul-Baki, J.D. Anderson, Vigour determination in soya bean seed by multiple criteria, Crop Sci. 13 (1973) 630–633. [21] M. Dworkin, J. Foster, Experiments with some microorganisms which utilize ethane and hydrogen, J. Bacteriol. 75 (1958) 592–601. [22] A.A. Belimov, N. Hontzeas, V.I. Safronova, S.V. Demchinskaya, G. Piluzza, S. Bullitta, B.R. Glick, Cadmium-tolerant plant growth-promoting bacteria associated with the roots of Indian mustard (Brassica juncea L. Czern.), Soil Biol. Biochem. 37 (2005) 241–250. [23] M.M. Bradford, A rapid and sensitive method for the quantitation of microgram quantities of protein utilizing the principle of protein dye binding, Anal. Biochem. 72 (1976) 248–254. [24] J.M. Bric, R.M. Bostock, S.E. Silversone, Rapid in situ assay for indole acetic acid production by bacteria immobilization on a nitrocellulose membrane, Appl. Environ. Microbiol. 57 (1991) 535–538. [25] W.V.B. Sundara-Rao, M.K. Sinha, Phosphate dissolving microorganisms in the soil and rhizosphere, Indian J. Agric. Sci. 33 (1963) 272–278. [26] C.H. Fiske, Y. Subbarow, A colorimetric determination of phosphorus, J. Biol. Chem. 66 (1925) 375–400. [27] B. Schwyn, J.B. Neilands, Universal chemical assay for the detection and determination of siderophores, Anal. Biochem. 160 (1987) 47–56. [28] Y. Bashan, H. Levanony, An improved selection technique and medium for the isolation and enumeration of Azospirillum brasilense, Can. J. Microbiol. 31 (1985) 947–952. [29] B. Reinhold-Hurek, T. Hurek, M. Claeyssens, M. Montagu, Cloning, expression in Escherichia coli, and characterization of cellulolytic enzymes of Azoarcus sp., a root-invading diazatroph, J. Bacteriol. 175 (1993) 7056–7065. [30] P.F. Mateos, J.I. Jimenez-Zurdo, J. Chen, A.S. Squartini, S.K. Haack, E. MartinezMolina, D.H. Hubbell, F.B. Dazzo, Cell-associated pectinolytic and cellulolytic enzymes in Rhizobium leguminosarum bv. trifolii, Appl. Environ. Microbiol. 58 (1992) 1816–1822.

Y. Ma et al. / Journal of Hazardous Materials 195 (2011) 230–237 [31] J. Sambrook, E.F. Fritsch, T. Maniatis, Molecular Cloning: A Laboratory Manual, second ed., Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY, 1989. [32] J. Dunbar, S. Takala, S.M. Barns, J.A. Davis, C.R. Kuske, Levels of bacterial community diversity in four arid soils compared by cultivation and 16S rRNA gene cloning, Appl. Environ. Microb. 65 (1999) 1662–1669. [33] R. Branco, A.P. Chung, A. Verıssimo, P.V. Morais, Impact of chromium contaminated wastewaters on the microbial community of a river, FEMS Microbiol. Ecol. 54 (2005) 35–46. [34] S.F. Altschul, T.L. Madden, A.A. Schaffer, J. Zhang, Z. Zhang, W. Miller, D.J. Lipman, Gapped BLAST and PSIBLAST: a new generation of protein database search programs, Nucleic Acids Res. 25 (1997) 3389–3402. [35] R.P. Ryan, K. Germaine, A. Franks, D.J. Ryan, D.N. Dowling, Bacterial endophytes: recent developments and applications, FEMS Microbiol. Lett. 278 (2008) 1–9. [36] P.R. Hardoim, L.S. van Overbeek, J.D. van Elsas, Properties of bacterial endophytes and their proposed role in plant growth, Trends Microbiol. 16 (2008) 467–471. [37] M.E. Puente, C.Y. Li, Y. Bashan, Rock-degrading endophytic bacteria in cacti, Environ. Exp. Bot. 66 (2009) 389–401. [38] S. Spaepen, J. Vanderleyden, R. Remans, Indole-3-acetic acid in microbial and microorganism-plant signaling, FEMS Microbiol. Rev. 31 (2007) 425–448. [39] M. Rajkumar, N. Ae, M.N.V. Prasad, H. Freitas, Potential of siderophoreproducing bacteria for improving heavy metal phytoextraction, Trends Biotechnol. 28 (2010) 142–149. [40] S.C. Verma, J.K. Ladha, A.K. Tripathi, Evaluation of plant growth promoting and colonization ability of endophytic diazotrophs from deep water rice, J. Biotechnol. 91 (2001) 127–141. [41] P.N. Jha, A. Kumar, Endophytic colonization of Typha australis by a plant growthpromoting bacterium Klebsiella oxytoca strain GR-3, J. Appl. Microbiol. 103 (2007) 1311–1320. [42] M.A. Ali, M. Ashraf, H.R. Athar, Inﬂuence of nickel stress on growth and some important physiological/biochemical attributes in some diverse canola (Brassica napus L.) cultivars, J. Hazard. Mater. 172 (2009) 964–969.

237

[43] C. Chen, D. Huang, J. Liu, Functions and toxicity of nickel in plants: recent advances and future prospects, CLEAN 37 (2009) 304–313. [44] E. Gajewska, M. Skłodowska, Differential biochemical responses of wheat shoots and roots to nickel stress: antioxidative reactions and proline accumulation, Plant Growth Regul. 54 (2008) 179–188. [45] M. Madhaiyan, S. Poonguzhali, T. Sa, Metal tolerating methylotrophic bacteria reduces nickel and cadmium toxicity and promotes plant growth of tomato (Lycopersicon esculentum L.), Chemosphere 69 (2007) 220–228. [46] Y.F. Zhang, L.Y. He, Z.J. Chen, Q.Y. Wang, M. Qian, X.F. Sheng, Characterization of ACC deaminase-producing endophytic bacteria isolated from copper-tolerant plants and their potential in promoting the growth and copper accumulation of Brassica napus, Chemosphere 83 (2011) 57–62. [47] H. Kupper, E. Lombi, F.J. Zhao, G. Wieshammer, P. MacGrath, Cellular compartimentation of nickel in the hyperaccumulators Alyssum lesbiacum, Alyssum bertolonii and Thlaspi goesingense, J. Exp. Bot. 52 (2001) 2291–2300. [48] B.S. Panwar, K.S. Ahmad, S.B. Mittal, Phytoremediation of nickel-contaminated soils by Brasssica species, Environ. Dev. Sustain. 4 (2002) 1–6. [49] V. de la Fuente, N. Rodríguez, B. Díez-Garretas, L. Rufo, A. Asensi, R. Amils, Nickel distribution in the hyperaccumulator Alyssum serpyllifolium Desf. spp. from the Iberian Peninsula, Plant Biosyst. 141 (2007) 170–180. [50] A. Zayad, N. Terry, Chromium in the environment: factors affecting biological remediation, Plant Soil 249 (2003) 139–156. [51] H. Rodriguez, R. Fraga, Phosphate solubilizing bacteria and their role in plant growth promotion, Biotechnol. Adv. 17 (1999) 319–339. [52] B.R. Glick, D.M. Penrose, J. Li, A model for the lowering of plant ethylene concentrations by plant growth promoting bacteria, J. Theor. Biol. 190 (1998) 63–68. [53] P.Q. Hung, S.M. Kumar, V. Govindsamy, K. Annapurna, Isolation and characterization of endophytic bacteria, Biol. Fertil. Soils 44 (2007) 155–162. [54] S. Zaidi, S. Usmani, B.R. Singh, J. Musarrat, Signiﬁcance of Bacillus subtilis strain SJ-101 as a bioinoculant for concurrent plant growth promotion and nickel accumulation in Brassica juncea, Chemosphere 64 (2006) 991–997. [55] R.B. Malabadi, K. Nataraja, Genetic transformation of conifers: applications in and impact on commercial forestry, Transgenic Plant J. 1 (2007) 289–313.

Journal of Hazardous Materials 195 (2011) 238–244

Contents lists available at ScienceDirect

Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat

Liquid–liquid extraction and ﬂat sheet supported liquid membrane studies on Am(III) and Eu(III) separation using 2,6-bis(5,6-dipropyl-1,2,4-triazin-3-yl)pyridine as the extractant A. Bhattacharyya a , P.K. Mohapatra a,∗ , T. Gadly b , D.R. Raut a , S.K. Ghosh b , V.K. Manchanda a a b

Radiochemistry Division, Bhabha Atomic Research Centre, Trombay, Mumbai 400085, India Bio-organic Division, Bhabha Atomic Research Centre, Trombay, Mumbai 400085, India

a r t i c l e

i n f o

Article history: Received 16 March 2011 Received in revised form 13 July 2011 Accepted 10 August 2011 Available online 17 August 2011 Keywords: Supported liquid membrane Lanthanide actinide separation BTP Americium

a b s t r a c t Solvent extraction and supported liquid membrane transport studies for the preferential removal of Am3+ from feeds containing a mixture of Am3+ and Eu3+ was carried out using 2,6-bis(5,6-dipropyl-1,2,4triazin-3-yl)pyridine (n-Pr-BTP) as the extractant. Diluent plays an important role in these studies. It was observed that the distribution coefﬁcients deteriorate signiﬁcantly for both Am3+ and Eu3+ though the separation factors were affected only marginally. The transport studies were carried out at pH 2.0 in the presence of NaNO3 to result in the preferential Am3+ transport with high separation factors. Effect of different experimental parameters, viz. feed composition, stripping agents, diluents of the organic liquid membrane and membrane pore size was studied on the transport and separation behaviour of Am3+ and Eu3+ . The supported liquid membrane studies indicated about 85% Am3+ and 6% Eu3+ transport in 6 h using 0.03 M n-Pr-BTP in n-dodecane/1-octanol (7:3) diluent mixture for a feed containing 1 M NaNO3 at pH 2 and a receiver phase containing pH 2 solution as the strippant. Consequently, a permeability coefﬁcient of (1.75 ± 0.21) × 10−4 cm s−1 was determined for the Am3+ transport. Stability of the n-Pr-BTP and its SLM was also studied by carrying out the distribution and transport experiment after different time intervals. © 2011 Elsevier B.V. All rights reserved.

1. Introduction One of the major drawbacks of nuclear energy programme is the generation of high level liquid wastes (HLLW) containing long lived radionuclides (t1/2 = 102 –106 years) such as minor actinides (237 Np, 241,243 Am and 245 Cm) and ﬁssion products (93 Zr, 99 Tc, 129 I, 135 Cs, etc.). Therefore, safe management of HLLW is required for the public acceptability of the nuclear energy programme. The strategy of vitriﬁcation and storage in deep geological repositories is now accepted worldwide for the management of HLW. It, however, requires surveillance over millions of years to monitor deformation of glass and the migration of the long lived radionuclides to aquatic environment under natural calamities such as earth quake, volcanic eruption, etc. An alternative strategy is to separate the long-lived radionuclides followed by their transmutation in high ﬂux reactors or accelerator driven sub-critical systems (ADSS) and is known as the ‘Partitioning and Transmutation’ strategy [1]. The partitioning step proposed to selectively extract the minor actinides co-extracts

∗ Corresponding author. Fax: +91 22 25505151. E-mail addresses: [email protected], [email protected] (P.K. Mohapatra). 0304-3894/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.jhazmat.2011.08.033

the lanthanides. Many of the lanthanide isotopes have high neutron capture cross-sections resulting in adverse effect on the transmutation process. It is, therefore, required to carry out separation of lanthanides from the trivalent minor actinides immediately after the partitioning step [2]. Lanthanide–actinide separation is, therefore, a key step in the back end of the fuel cycle. Amongst the early processes employed for the Ln(III)/An(III) separation, the TRAMEX (tertiary amine extraction) process and the TALSPEAK (trivalent actinide lanthanide separation by phosphorus extractants aqueous komplexes) process using Alamine-336 and di-2-ethylhexylphosphoric acid (D2EHPA), respectively as the extractants were found to be promising [3–5]. Due to similar charge and ionic size (ionic potential), trivalent actinides and lanthanides show similar chemical behaviour towards the hard ‘O’ donor ligands. Actinides, however, can form stronger covalent bond with the soft donor ligands because of higher spatial distribution of the ‘5f’ valence orbitals of the actinides as compared to the valence ‘4f’ orbitals of the lanthanides and this property is being exploited for the separation of trivalent actinides and lanthanides. Dithiophosphinic acids with ‘S’ donor atoms such as bis(2,4,4-trimethyl pentyl)dithiophosphinic acid (Cyanex-301) and bis-(chlorophenyl)dithiophosphinic acid have also been used as selective extractants for the trivalent

A. Bhattacharyya et al. / Journal of Hazardous Materials 195 (2011) 238–244

actinides [6,7]. In spite of very high selectivity for the trivalent actinides over the lanthanides, the major disadvantage of Cyanex301 is its poor extractability at low pH region (pH < 3) [8,9]. Bis-(chlorophenyl)dithiophosphinic acid, however, can extract the trivalent actinides from acidic medium with poor selectivity over the lanthanides [7]. Recently, several ‘N’ donor reagents such as 2,6-bis(5,6-dipropyl-1,2,4-triazin-3-yl)pyridine (n-Pr-BTP) ligand have been proposed for the selective extraction of trivalent actinides from acidic feeds (0.1–1.0 M HNO3 ) with reasonably high S.F. (∼150) values with respect to the trivalent lanthanides [10,11]. Softer nature of this aromatic ‘N’ donor ligand is responsible for the selectivity of n-Pr-BTP towards trivalent actinides over the lanthanides. A number of reports appeared in the literature on the complexation studies of n-Pr-BTP with trivalent actinides and lanthanides and its selectivity towards the actinides over lanthanides was explained on the basis of formation of ML3 ·[NO3 ]3 type of species in higher extent in case of actinides [12] due to higher stability constant of this complex in case of actinides [13]. Most of the separation studies of trivalent actinides and lanthanides involving n-Pr-BTP as the extractant were carried out using solvent extraction [14]. The major drawback of solvent extraction is the requirement of large solvent inventory and formation of third phase. These problems can be alleviated by the use of alternative methods such as extraction chromatography and liquid membrane. There are very limited reports available using BTP derivatives for lanthanide–actinide separation by extraction chromatography [15–18] and liquid membrane [19,20]. Supported liquid membrane (SLM) based separation methods have received considerable attention by the researchers for various separations and puriﬁcation of metal ions [21–23]. SLM based techniques are particularly attractive as they involve low inventory of organic extractants and simultaneous extraction and stripping of the metal ion [21]. It is, therefore, pertinent to evaluate supported liquid membranes (SLMs) using n-Pr-BTP as the carrier extractant for the selective transport of trivalent actinides. The present study involves the synthesis of n-Pr-BTP derivative and its evaluation for the extraction and supported liquid membrane transport studies of Am(III) and Eu(III) using PTFE ﬂat sheets of different pore size and a variety of feed and strip conditions, viz. varying the NaNO3 concentration in the feed solution and using different stripping agents (pH 2.0 and 0.01 M EDTA). Effect of organic diluent was also studied on the extraction and transport behaviour of Am3+ and Eu3+ . 2. Experimental 2.1. Reagents 2.1.1. Synthesis of n-Pr-BTP The synthesis of the bis-triazinyl pyridine 4 is depicted in Scheme 1. Stetter reaction of butanal in the presence of 3-benzyl-5(2-hydroxyethyl)-4-methylthiazolium chloride and triethylamine in ethanol under reﬂux gave the hydroxy ketone 1 (68%) which under Swern oxidation conditions gave the diketone 2 in very good yield (87%). The required dihydrazide partner 3 was prepared from commercially available 2,6-dicyanopyridine by treating with excess of hydrazine hydrate at room temperature in a quantitative yield. The desired n-Pr-BTP 4 was prepared by reﬂuxing a 1:2 mixture of hydrazide 3 and the diketone 2 in ethanol. The crude product was crystallized from 2-propanol to give pure 4 in 88% yield. The following are the NMR characterization data. 1 H NMR (200 MHz, CDCl3 ): ı = 0.93–1.12 (m, 12H, 4 × CH3 ), 1.77–2.0 (m, 8H, 4 × CH3 –CH2 –CH2 ), 2.87–2.95 (m, 4H, 4 × CH3 –CH2 –CH2 –), 2.99–3.07 (m, 4H, 4 × CH3 –CH2 –CH2 –), 8.08 (t, 8.0 Hz, 1H, H4 of Py), 8.7 (d, 8.0 Hz, 2H, H3 , H5 of Py). 13 C NMR (200 MHz, CDCl3 ): ı = 13.9 (2C), 14.0 (2C), 21.1 (2C), 21.7 (2C), 34.3 (2C), 35.9 (2C), 125.1 (2C), 138.0, 153.9 (2C), 159.7 (2C), 161.1 (2C), 161.9 (2C).

239

2.2. Methods 2.2.1. Distribution studies Distribution studies were carried out with 241 Am and 152 Eu as the tracers in different aqueous phase conditions using different organic phases containing Et-BTP as the extractant. Equal volumes (0.5 mL) of the organic phase and the aqueous phase containing the required tracer were taken in stoppered glass tubes and agitated in a thermostated water bath at 25.0 ± 0.1 ◦ C for 1 h. The two phases were then centrifuged and assayed by taking suitable aliquots (usually 0.1 mL) from both the phases. The gamma activities were measured using a NaI(Tl) scintillation detector when only one radiotracer was used and using HPGe detector when a mixture of radiotracers was used. The distribution ratio (DM ) was calculated as the ratio of counts per minute per unit volume in the organic phase to that in the aqueous phase. The separation factor (S.F.) is deﬁned as the ratio of DAm to DEu . All the experiments were carried out in duplicate and the accepted data points have material balance with in ±5%.

2.2.2. Transport studies The transport studies were carried out in a two-compartment glass cell, each of 20 mL capacity. The feed and the receiver compartments were separated by microporous PTFE ﬁlters (0.45 ␮m) containing the liquid membrane. The supported liquid membranes were prepared by soaking the PTFE ﬁlters with the solutions of n-Pr-BTP in different organic diluent mixtures (vide infra). The compositions of the feed, strip solution and carrier concentration were varied and the transport of Am and Eu were measured. The feed, containing the required radiotracers, and the strip compartments were stirred using magnetic stirrer at a constant speed of 200 rpm, which was found to be adequate in an earlier communication [24]. Aliquots were withdrawn at regular intervals and the radioactivity was assayed by gamma counting as mentioned above.

2.3. Transport equations The transport of a metal ion using a carrier solution as the supported liquid membranes usually involves ﬁve distinct steps, viz. (1) diffusion of the metal ion from the bulk feed phase to the feed–membrane interface; (2) extraction of the metal ion by the carrier molecule at the feed–membrane interface; (3) diffusion of the metal–carrier complex inside the membrane phase ; (4) stripping of the metal ion at the membrane–receiver interface; and (5) diffusion of the stripped metal ion from the membrane–receiver interface. The transport experiments are carried out under the conditions that the distribution coefﬁcient (symbolized as D) is much larger at the feed–membrane interface as compared to that at the membrane–receiver interface. Under steady state condition, by ignoring the concentration of the metal ion in the receiver phase one can get the ﬂux (J) from the following equation [25]: J = PCf

(1)

where P is the permeability coefﬁcient at the feed–membrane interface and Cf is the concentration of the metal ion at the feed side. The ﬂux can alternatively be expressed as J=−

1 dV C f f Q

·

dt

(2)

where Vf is the feed volume and Q is the exposed area of the membrane. Assuming that the volume of the feed does not change

240

A. Bhattacharyya et al. / Journal of Hazardous Materials 195 (2011) 238–244

Me

CHO

N

Ph Cl

O

Et3N, EtOH, Reflux 68%

1

HO

S

1. (COCl)2, DMSO, CH2Cl2, -60oC to -50oC 2. Et3N

OH

O

87%

2

O

O NH2NH2.H2O NC

N

CN

100%

H2N

N

O C N NNH 2

C NH2 NNH 2

3

N

EtOH, reflux

N

N N

N

N

4

88%

Scheme 1. Synthesis scheme for n-Pr-BTP followed in the present study.

signiﬁcantly with time and after combining Eqs. (1) and (2), and integrating one obtains,

ln

Cf,0 Cf,t

=

QPt Vf

(3)

where Vf , Cf,0 and Cf,t represent the volume of the feed and concentrations of feed at starting time and after time ‘t’, respectively. Q is expressed as the product of the geometrical surface area (A) and the porosity (ε). The permeability coefﬁcient (P) values were calculated using Eq. (3). The cumulative percent transport (%T) of the metal ions at a given time is determined by the following equation, %T = 100 ×

Cr,t Cf,0

(4)

where Cr,t is the concentration of the metal ion in the receiver phase at a given time ‘t’. The data treatment that follows generally includes plots of %T vs time as well as calculations of permeability coefﬁcient using Eq. (3). All the experiments were repeated and the accepted data were within the error limits of ±5%. 3. Results and discussion

non-polar diluents were used in combination with a polar diluent, i.e. 2-ethyl-hexanol, which was used as the modiﬁer. In the present work, n-dodecane along with polar modiﬁer diluents viz. 1-decanol, 1-octanol, nitrobenzene and NPOE were used to make 0.04 M n-Pr-BTP solution. The polar to non-polar diluent volume ratio was maintained as 3:7 with the exception of nitrobenzene, which is miscible with dodecane with a volume ratio of 2:3 (dodecane:nitrobenzene) only. The distribution ratios of both Am3+ and Eu3+ was found to be increasing in the following order of the diluent mixtures, n-dodecane/NPOE < n-dodecane/1decanol < n-dodecane/1-octanol < n-dodecane/nitrobenzene. DAm and DEu values were found to be marginally higher in case of n-dodecane/1-octanol mixture as compared to that in case of ndodecane/1-decanol system. Higher DAm and DEu values were also observed by Kolarik et al. [14] in case of TPH/1-butanol mixture as compared to TPH/1-octanol mixture. Smaller the carbon chain length of the alcohol, higher the polarity of the organic phase and higher is the metal ion extraction. This suggests that the extractable species for Am3+ and Eu3+ by the n-Pr-BTP is polar in nature due to the presence of the nitrate ions in the outer sphere (ion-pair complex). In case of n-dodecane/nitrobenzene mixture, the polarity of nitrobenzene is much higher as compared to the alcoholic solvents,

3.1. Distribution studies

3.1.2. Effect of organic diluents Low volatile hydrocarbon solvents, viz. n-dodecane, kerosene or TPH are ideal for the solvent extraction or liquid membrane based separation processes. The BTP derivatives are generally insoluble in hydrocarbon solvents and can be solubilized by using polar solvent as modiﬁers along with the hydrocarbon solvents [11]. A detailed study on the effect of organic diluents was carried out by Kolarik et al. [14], where several aliphatic and aromatic

10 8 6

DAm

3.1.1. Effect of equilibrium time The attainment of equilibrium of Am3+ extraction by n-Pr-BTP was investigated as a function of equilibration time and the results are plotted in Fig. 1. A continuous increase in the DAm value was observed up to 60 min beyond which no change was seen suggesting that 1 h equilibration time is sufﬁcient. All subsequent experiments were carried out with 2 h equilibration time to ensure attainment of equilibrium. It has been reported earlier that Am3+ extraction rate is linearly dependent on the free n-Pr-BTP concentration. As in the presence of 1 M HNO3 in the aqueous phase, the free n-Pr-BTP concentration is very less, Am3+ extraction is, therefore relatively less in the present case [26]. Studies have been carried out at low acidity to ensure higher Am3+ extraction and efﬁcient separation from Eu3+ .

4 2 0 0

30

60

90

120

Time (min) Fig. 1. Extraction of Am3+ at different time of equilibration; org. phase: 0.04M n-PrBTP in n-dodecane/1-octanol (7:3) mixture; aq. phase: pH 2 containing 1 M NaNO3 .

A. Bhattacharyya et al. / Journal of Hazardous Materials 195 (2011) 238–244 Table 1 Effect of organic diluents on the extraction of Am(III) and Eu(III); org. phase: 0.04 M n-Pr-BTP in different diluent mixture; aq. phase: pH 2 containing 1 M NaNO3 . Org. phase

DAm

n-Dodecane:octanol (7:3) n-Dodecane:decanol (7:3) n-Dodecane:NPOE (7:3) n-Dodecane:nitrobenzene (2:3)

26.71 22.4 0.34 43.8

DEu ± ± ± ±

1.77 4.9 0.01 4.5

0.33 0.26 0.02 1.03

S.F. ± ± ± ±

0.02 0.01 0.002 0.01

82 85 18.5 42

moreover nitrobenzene is present in larger proportion, DAm and DEu values are therefore expected to be still higher in this case and it was observed experimentally also (Table 1). 3.1.3. Effect of n-Pr-BTP concentration Literature [26,27] shows that the complex species using BTP derivatives usually contain three molecules of the N-donor extractant. Accordingly, the extraction equilibrium involving Am3+ can be presented as: Am3+ + 3NO3 − + 3BTP(o) → Am(NO3 )3 ·3BTP(o)

(5)

where the species in the organic phase are indicated with subscript ‘(o)’ while those in the aqueous phase are without any subscript. Effect of the n-Pr-BTP concentration in n-dodecane/1-octanol phase was studied on the DAm value using 1 M HNO3 as the aqueous phase. A slope value of ∼2.6 was observed in the logarithmic plot of DAm vs n-Pr-BTP concentration (Fig. 2) contrary to the expected dependence of 3 [27]. Similar to our result, lower slope value was also reported in the literature, which was explained on the basis of the aggregation of the n-Pr-BTP molecules in the organic phase [26]. 3.1.4. Effect of NaNO3 concentration in the aqueous phase Though BTP derivatives extract trivalent actinides and lanthanides from nitric acid medium, it was of interest to understand the extraction behaviour in the absence of hydrogen ions, which may be forming cationic species such as BTP. H+ thereby decreasing the free extractant concentration. NaNO3 concentration variation study was therefore carried out using an aqueous phase of pH 2 containing varying amounts of NaNO3 . As indicated in Eq. (5) above, increasing the nitrate ion concentration should favour the metal ion

241

Table 2 Effect of NaNO3 concentration on Am(III) and Eu(III) extraction org. phase: 0.04 M n-Pr-BTP in n-dodecane/n-octanol (7:3); aq. phase: pH 2 with varying concentration of NaNO3 . [NaNO3 ]/(M)

DAm

0 0.5 1.0 1.5 2.0

0.020 8.74 26.71 58.8 96

DEu ± ± ± ± ±

0.002 0.38 1.77 0.2 4

0.0010 0.11 0.33 0.73 1.14

S.F. ± ± ± ± ±

0.0001 0.01 0.02 0.01 0.07

13.7 80 82 80 84

extraction. Both the DAm and DEu values were also found to increase with increasing NaNO3 concentration in the aqueous phase resulting in no change in the S.F. values (Table 2). Hoshi et al. [16], have also shown a increase in the distribution coefﬁcient of Am3+ , Nd3+ and Ce3+ with increasing nitrate concentration up to 2 M NaNO3 . 3.1.5. Stability of n-Pr-BTP Stability of n-Pr-BTP was studied by measuring the DAm value over a period of time (Fig. 3) and it was observed that the DAm value decreases with time. The experiments were carried out in two modes. In the ﬁrst case, the solid BPT was stored and intermittently taken to prepare a solution in the diluent mixture mentioned above and the distribution studies were carried out with both Am3+ and Eu3+ . In the second case, the solution itself was stored over a period of time and was used intermittently at varying time intervals. The decrease in DAm value is higher when n-Pr-BTP was stored in the solution form in n-dodecane/n-1-octanol mixture as compared to the case when n-Pr-BTP was stored as such in the solid form. The presence of labile ␣-benzylic hydrogen atom is mainly responsible for the poor chemical stability of the n-Pr-BTP molecules [28] and the degradation of n-Pr-BTP molecules is, therefore, much faster when it is stored in solution form. 3.2. Supported liquid membrane studies From the solvent extraction studies, signiﬁcant extraction of Am3+ was observed with a S.F. value of >100 at 1 M HNO3 . SLM studies at this acidity, however, show no transport of Am3+ even

100

Slope: 2.56 ± 0.08 10

DAm

DM / S.F.

10

DAm (Fresh Soln) DEu (Fresh Soln) S.F. (Fresh Soln) DAm DEu S.F.

1

0.1

1

0.01

0.02

0.03

0.04 0.05

[nPr-BTP] / (M) Fig. 2. Effect of n-Pr-BTP concentration on the distribution ratio of Am3+ ; Org. Phase: n-Pr-BTP in dodecane/octanol (7:3) mixture; aq. phase: pH 2 containing 1 M NaNO3 .

0

20

40

60

80

Time (days) Fig. 3. Variation of distribution coefﬁcient of Am3+ and Eu3+ and their separation factor with time; org. phase: 0.03 M n-Pr-BTP in n-dodecane/1-octanol (7:3) mixture; aq. phase: pH 2 containing 1 M NaNO3 .

242

A. Bhattacharyya et al. / Journal of Hazardous Materials 195 (2011) 238–244

100 100

10

Am (pH 2) Eu (pH 2) Am (0.01M EDTA) Eu (0.01M EDTA)

1

0

400

800

1200

Time (min) Fig. 4. Role of stripping agent on the transport of Am3+ and Eu3+ ; feed: pH 2 containing 1 M NaNO3 ; membrane: 0.03 M n-Pr-BTP in n-dodecane/1-octanol (7:3) mixture.

after 24 h, which might be due to the protonation of the n-Pr-BTP molecules. Unlike the solvent extraction studies, the SLM studies involve large volumes of aqueous phase as compared to the organic phase, which can result in the protonation of major fraction of the extractant. Apparently, the protonated form of BTP (BTPH+ ·NO3 − or analogous complex) comes out of the membrane phase leading to negligible Am3+ transport. The leaching of the carrier was conﬁrmed by the red coloration of the feed and receiver phases when small amount of Fe3+ was spiked in the respective aqueous phases. Subsequent studies on the transport behaviour of Am3+ and Eu3+ through the SLM containing n-Pr-BTP were, therefore, carried out after adjusting the feed solution to pH 2 containing varying amounts of NaNO3 . 3.2.1. Effect of stripping agents From the nitrate ion concentration variation study, it was observed that the presence of nitrate ion facilitates the metal ion extraction. Consequently, very low DAm value was observed at pH 2 in absence of NaNO3 and justifying its subsequent use in the receiver phase solution in the SLM studies. 0.01 M EDTA was also evaluated for the stripping and compared with the transport data with pH 2 solution as the strippant. No improvement in the transport proﬁle was observed using EDTA as the strippant (Fig. 4). In the subsequent SLM studies, pH 2 solution was chosen as the strippant while EDTA solution was not preferred. 3.2.2. Effect of organic diluents Transport studies were carried out in mixtures of two diluents where one component was n-dodecane and the other was a polar diluent like 1-octanol, 1-decanol, nitrobenzene or NPOE (2nitrophenyl octyl ether) in the same composition as used in the solvent extraction studies. Out of all the solvent system studied, most favourable transport was observed when n-dodecane/1octanol mixture was used (Fig. 5). n-Dodecane/1-decanol system, however, showed marginally slower transport of Am3+ and Eu3+ initially which was ascribed to the lower distribution ratio values in n-dodecane/1-decanol mixture (Table 1) and higher viscosity of 1-decanol (11.05 mPa s [29]) as compared to 1-octanol (7.24 mPa s [29]). However, overnight transport data for both the systems were comparable (Fig. 5). Though in the case of

% Transport

% Transport

80

Am (Dod/octanol) Eu (Dod/octanol) Am (Dod/decanol) Eu (Dod/decanol) Am (Dod/PhNO2) Eu (Dod/PhNO2) Am (Dod/NPOE) Eu (Dod/NPOE)

60

40

20

0 0

400

800

1200

1600

Time (min) Fig. 5. Effect of organic diluents on the transport of Am3+ and Eu3+ ; feed: pH 2 containing 1 M NaNO3 ; membrane: 0.03 M n-Pr-BTP in different diluent mixture; strip: pH 2 solution.

n-dodecane/nitrobenzene system the DAm value was higher as compared to other solvent systems and viscosity of nitrobenzene is also not so high, transport was found to be very slow, which could be due to leaching out of the carrier solution from the membrane pores. We have reported inefﬁcient transport behaviour with nitrobenzene as the diluent and the use of higher proportion (60% v/v) of nitrobenzene in the diluent mixture may result resulting in poor membrane stability due to its appreciable aqueous solubility [30]. This may be one of the reasons for inefﬁcient cation transport using nitrobenzene as the diluent [31]. In the n-dodecane/NPOE mixture, however, the slower transport is expected due to low DM values as well as low diffusivity as a consequence of higher viscosity of the diluent mixture [32]. For subsequent transport experiments, the n-dodecane/1-octanol mixture, which shows highest transport rate, was chosen as the organic solvent for the liquid membrane. 3.2.3. Effect of NaNO3 concentration in the aqueous phase From the solvent extraction studies, DAm and DEu values were found to be increasing with increasing the concentration of NaNO3 in the aqueous phase. A systematic study was, therefore, carried out in order to investigate the transport behaviour of Am3+ and Eu3+ with increasing the NaNO3 concentration and as expected the permeability coefﬁcient values for both Am3+ and Eu3+ increased (Fig. 6). The increase in PEu value was less signiﬁcant up to 1 M NaNO3 and increased signiﬁcantly beyond 1 M NaNO3 , whereas, signiﬁcant enhancement in PAm value was observed even up to 1 M NaNO3 . The separation factor (PAm /PEu ) value was, therefore, found to be highest at 1 M NaNO3 . Decrease in the S.F. values with increasing NaNO3 concentration from 1 to 2 M was also seen during our extraction chromatography studies [17]. 3.2.4. Effect of carrier concentration The PAm value was found to increase with increasing the carrier (n-Pr-BTP) concentration (Fig. 7), which can be explained on the basis of increase in the DAm value with increasing n-Pr-BTP concentration (Fig. 2). The permeability coefﬁcient is dependent on the distribution ratio of Am3+ in the feed–membrane interface and the diffusion coefﬁcient of the metal–carrier complex in the liquid

A. Bhattacharyya et al. / Journal of Hazardous Materials 195 (2011) 238–244

[ML3]3+[NO3-]3

M3+ + 3NO3+ Feed pH 2 + 1 M NaNO3

M3+ + 3NO3+ Strip pH 2

Membrane

3H+

243

3H+

3HL+

Scheme 2. Transport mechanism of Am3+ and Eu3+ through the SLM of n-Pr-BTP.

10

-1

PAm x 10 (cm.sec )

membrane phase. The diffusion coefﬁcient can be considered to be constant irrespective of the carrier concentration studied in the present work as the increase in viscosity of the organic phase with increasing the carrier concentration is insigniﬁcant at such a low concentration (0.01–0.04 M) and the metal–carrier complex is also remained unchanged throughout this concentration range. Hence higher DAm value is expected to positively inﬂuence the PAm value. The increase in PAm value, however, is more signiﬁcant at the lower carrier concentration and PAm value increases with lower slope at higher carrier concentration, which might be due to the aggregation of n-Pr-BTP with increasing concentration. Similar observation was also made in the solvent extraction studies (vide supra). Based on these studies a mechanism (Scheme 2) can be proposed for the selective transport of Am3+ over Eu3+ , where formation of the complex [ML3 ]3+ ·[NO3 − ]3 (L is n-Pr-BTP) is more favoured in case of Am3+ as compared to Eu3+ and this reﬂects in the selectivity of the membrane for Am3+ over Eu3+ .

4

1

0.1

3.2.5. Effect of membrane pore size Pore size of the PTFE ﬂat sheet has signiﬁcant role in controlling the transport behaviour of the metal ions. The PAm value was found to increase with increasing membrane pore size from 0.2 to 0.45 ␮m and then decrease up to 5.0 ␮m (Fig. 8). We have reported that increasing the membrane pore size usually results in a decrease 100

4

0.02

0.03

0.04 0.05

[nPr-BTP] / (M) Fig. 7. Effect of n-Pr-BTP concentration on the transport behaviour of Am3+ ; feed: pH 2 containing 1 M NaNO3 ; membrane: n-Pr-BTP in n-dodecane/1-octanol (7:3) mixture; strip: pH 2 solution. Table 3 Effect of operational time of the n-Pr-BTP SLM on the transport of Am3+ ; feed: pH 2 solution containing 1 M NaNO3 ; strip: pH 2 solution; liq. membrane: 0.03 M n-Pr-BTP in dodecane/octanol (7:3) mixture.

10

PM (x10 ) or S.F.

0.01

Time (day)

PAm × 104 (cm s−1 )

0 1

1.75 ± 0.21 0.17 ± 0.02

1

0.1 PAm PEu S.F. (PAm / PEu)

0.01 0.5

1.0

1.5

2.0

[NaNO3] / (M) Fig. 6. Effect of NaNO3 concentration in the feed solution on the transport and separation behaviour of Am3+ and Eu3+ ; feed: pH 2 with varying concentration of NaNO3 ; membrane: 0.04 M n-Pr-BTP in n-dodecane/1-octanol (7:3) mixture; strip: pH 2 solution.

in the P-values, which was attributed to lower capillary forces and higher leachability of the carrier solvent with membranes of higher pore size [33]. Similar observation is made in the present work. However, the increase in the P-value from 0.2 to 0.45 ␮m can only be explained on the basis of favourable transport of a bulky complex species at 0.45 ␮m which is hindered when membranes of 0.2 ␮m pore size are used. Similar observation was reported when UO2 2+ ion transport was investigated using Aliquat 336 as the carrier extractant [34]. 3.2.6. Stability of the membrane The stability of the supported liquid membrane of n-Pr-BTP in n-dodecane/1-octanol (7:3) mixture was studied by measuring the PAm value with time. PAm value was found to decrease from (1.75 ± 0.21) × 10−4 to (0.17 ± 0.02) × 10−4 cm s−1 even in 1 day (Table 3). This poor stability is attributed to two reasons viz., (i) leaching out of the organic phase from the pores of the PTFE support

244

A. Bhattacharyya et al. / Journal of Hazardous Materials 195 (2011) 238–244

4

PAm x 10

4

3

2

1

0

0.2

0.45

1.2

5.0

Pore Size (µm) Fig. 8. Effect of the pore size of the PTFE ﬂat sheet on Am3+ transport; feed: pH 2 solution containing 1 M NaNO3 ; membrane: 0.04 M n-Pr-BTP in n-dodecane/1octanol (7:3) mixture; strip: pH 2 solution.

as water has signiﬁcant miscibility with the 1-octanol and (ii) poor stability of the carrier molecules as described in earlier section the n-Pr-BTP molecules degrades with time resulting in decrease in DAm value (Fig. 3), which ultimately reduces the PAm value. 4. Conclusions From the solvent extraction studies highest distribution ratio of Am3+ was observed by n-Pr-BTP in dodecane/nitrobenzene mixture. The SLM study, however, shows promising result in dodecane/octanol mixture, where ∼85% transport of Am3+ was observed in 6 h, which was accompanied by ∼6% of Eu3+ , using 1 M NaNO3 at pH 2 as feed and pH 2 solution as the strippant. The distribution ratio and permeability coefﬁcient values were found to be increasing with the ligand as well as aqueous NaNO3 concentration. Poor stability of the carrier molecules and water miscibility of the organic phase has been reﬂected in the poor membrane stability suggesting the membranes need to be regenerated with fresh carrier solutions for more efﬁcient separation.

References [1] V.P. Bhatnagar, S. Casalta, M. Hugon (EC), Partitioning and Transmutation Research in the EURATOM Fifth and Sixth Framework Programmes, Proceedings of the Seventh Information Exchange Meeting on Actinide and Fission Product Partitioning and Transmutation, Jeju, South Korea, 14–16 October, 2002, pp. 77–88. [2] B. Christiansen, C. Apostolidis, R. Carlos, O. Courson, J.P. Glatz, R. Malmbeck, G. Pagliosa, K. Romer, D. Serrano-Purroy, Advanced aqueous reprocessing in P&T strategies: process demonstrations on genuine fuels and targets, Radiochim. Acta 92 (2004) 475–480. [3] R.D. Baybarz, B.S. Weaver, H.B. Kinser, Isolation of transplutonium elements by tertiary amine extractants, Nucl. Sci. Eng. 17 (1963) 457–463. [4] B. Weaver, F.A. Kappelmann, Preferential extraction of lanthanides over trivalent actinides by monoacidic organophosphates from carboxylic acids and from mixtures of carboxylic and aminopolyacetic acids, J. Inorg. Nucl. Chem. 30 (1968) 263–272. [5] K.L. Nash, A review of the basic chemistry and recent developments in trivalent f-element separations, Solv. Extr. Ion. Exch. 11 (1993) 729–768. [6] Y. Zhu, J. Chen, R. Jiao, Extraction of Am(III) and Eu(III) from nitrate solution with puriﬁed cyanex-301, Solv. Extr. Ion Exch. 14 (1996) 61–69.

[7] C. Madic, M.J. Hudson, J.O. Liljenzin, J.P. Glatz, R. Nannicini, A. Facchini, Z. Kolarik, R. Odoj, Report EUR 19149EN , 2000. [8] A. Bhattacharyya, P.K. Mohapatra, V.K. Manchanda, Separation of Am3+ and Eu3+ from nitrate medium using a mixture of Cyanex-301 and N-donor extractants, Solv. Extr. Ion Exch. 24 (2006) 1–17. [9] C. Hill, C. Madic, P. Baron, M. Ozawa, Y. Tanaka, Trivalent minor actinides/lanthanides separation, using organophosphinic acids, J. Alloys Compd. 159 (1998) 271–273. [10] C. Ekberg, A. Fermvik, T. Retegan, G. Skarnemark, M.R.S. Foreman, M.J. Hudson, S. Englund, M. Nilsson, An overview and historical look back at the solvent extraction using nitrogen donor ligands to extract and separate An(III) from Ln(III), Radiochim. Acta 96 (2008) 225–233. [11] Z. Kolarik, U. Mullich, F. Gassner, Selective extraction of Am(III) over Eu(III) by 2,6-ditriazolyl- and 2,6-ditriazinylpyridines, Solv. Extr. Ion Exch. 17 (1999) 23–32. [12] M. Steppert, C. Walther, A. Geist, T. Fanghänel, Direct nano ESI time-ofﬂight mass spectrometric investigations on lanthanide BTP complexes in the extraction-relevant diluent 1-octanol, New J. Chem. 33 (2009) 2437–2442. [13] S. Trumm, P.J. Panak, A. Geist, T. Fanghänel, A TRLFS study on the complexation of Cm(III) and Eu(III) with 2,6-bis(5,6-dipropyl-1,2,4-triazin-3-yl)pyridine in water/methanol mixture, Eur. J. Inorg. Chem. 19 (2010) 3022–3028. [14] Z. Kolarik, U. Mullich, F. Gassener, Extraction of Am(lII) and Eu(lII) nitrates by 2,6-di-(5,6-dipropyl-1,2,4-triazin-3-yl)pyridines, Solv. Extr. Ion Exch. 17 (1999) 1155–1170. [15] Y. Wei, K.N. Sabharwal, M. Kumagai, T. Asakura, G. Uchiyama, S. Fujine, Preparation of novel silica-based nitrogen donor extraction resins and their adsorption performance for trivalent americium and lanthanides, J. Nucl. Sci. Technol. 37 (2000) 1108–1110. [16] H. Hoshi, Y.Z. Wei, M. Kumagai, T. Asakura, Y. Morita, Separation of trivalent actinides from lanthanides by using R-BTP resins and stability of R-BTP resin, J. Alloys Compd. 408–412 (2006) 1274–1277. [17] A. Bhattacharyya, P.K. Mohapatra, T. Gadly, S.K. Ghosh, D.R. Raut, V.K. Manchanda, Extraction chromatographic study on the separation of Am3+ and Eu3+ using ethyl-BTP as the extractant, J. Radioanal. Nucl. Chem. 288 (2011) 571–577. [18] A. Zhang, E. Kuraoka, M. Kumagai, Preparation of a novel macroporous silica-based 2,6-bis(5,6-diisobutyl-1,2,4-triazine-3-yl)pyridine impregnated polymeric composite and its application in the adsorption for trivalent rare earths, J. Radioanal. Nucl. Chem. 274 (2007) 455–464. [19] A. Geist, M. Weigl, K. Gompper, Small-scale actinide(III) partitioning processes in miniature hollow ﬁbre modules, Radiochim. Acta 93 (2005) 197–202. [20] A. Bhattacharyya, P.K. Mohapatra, A. Roy, T. Gadly, S.K. Ghosh, V.K. Manchanda, Ethyl-bis-triazinylpyridine (Et-BTP) for the separation of americium(III) from trivalent lanthanides using solvent extraction and supported liquid membrane methods, Hydrometallurgy 99 (2009) 18–24. [21] P.K. Mohapatra, V.K. Manchanda, Liquid membrane based separations of actinides and ﬁssion products, Indian J. Chem. 42A (2003) 2925–2939. [22] L. Boyadzhiev, Z. Lazarova, in: R.D. Noble, S.A. Stern (Eds.), Membrane Separations Technology: Principles and Applications, Elsevier Science B.V., 1995, pp. 283–300. [23] N.M. Kocherginsky, Q. Yang, L. Seelam, Recent advances in supported liquid membrane technology, Sep. Purif. Technol. 53 (2007) 171–177. [24] S. Sriram, P.K. Mohapatra, A.K. Pandey, V.K. Manchanda, L.P. Badheka, Facilitated transport of americium(III) from nitric acid media using dimethyldibutyltetradecyl-1,3-malonamide, J. Membr. Sci. 177 (2000) 163–175. [25] D.L. Hofman, W.M. Craig, E.M. Buchalter, R.S. Birkill, J.J. Smit, Supported liquid membrane technology applied to the recovery of useful isotopes from reactor pool water. I, in: Chem. E. Symposium Series No. 103, Extraction’87, EFCE Publication Series No. 347, 2006, pp. 180–182. [26] M. Weigl, A. Geist, U. Muellich, K. Gompper, Kinetics of americium(III) extraction and back extraction with BTP, Solv. Extr. Ion Exch. 24 (2006) 845–860. [27] M.A. Denecke, P.J. Panak, F. Burdet, M. Weigl, A. Geist, R. Klenze, M. Mezzanine, K. Gompper, A comparative spectroscopic study of U(III)/Am(III) and Ln(III) complexed with N-donor ligands, C. R. Chimie 10 (2007) 872–882. [28] M.J. Hudson, C.E. Boucher, D. Braekers, J.F. Desreux, M.G.B. Drew, M.R. St, J. Foreman, L.M. Harwood, C. Hill, C. Madic, F. Markend, T.G.A. Youngsa, New bis(trizinyl) pyridines for selective extraction of americium(III), New J. Chem. 30 (2006) 1171–1183. [29] S. Matsuo, T. Makita, Viscosities of six 1-alkanols at temperatures in the range 298–348 K and pressures up to 200 MPa, Int. J. Thermophys. 10 (1989) 833–843. [30] H.E. Vermillion, B. Werbel, J.H. Saylor, P.M. Gross, Solubility studies. VI. The solubility of nitrobenzene in deuterium water and in ordinary water, J. Am. Chem. Soc. 63 (1941) 1346–1347. [31] D.R. Raut, P.K. Mohapatra, S.A. Ansari, A. Sarkar, V.K. Manchanda, Selective transport of radio-cesium by supported liquid membranes containing calix[4]crown-6 ligands as the mobile carrier, Desalination 232 (2008) 262–271. [32] D.R. Raut, P.K. Mohapatra, S.A. Ansari, V.K. Manchanda, Evaluation of a calix[4]bis-crown-6 ionophore-based supported liquid membrane system for selective 137 Cs transport from acidic solutions, J. Membr. Sci. 310 (2008) 229–236. [33] D.S. Lakshmi, P.K. Mohapatra, D. Mohan, V.K. Manchanda, Uranium transport using a PTFE ﬂat-sheet membrane containing alamine 336 in toluene as the carrier, Desalination 163 (2004) 13–18. [34] D.S. Lakshmi, P.K. Mohapatra, D. Mohan, V.K. Manchanda, Uranium pertraction across a PTFE ﬂat sheet membrane containing Aliquat 336 as the carrier, Sep. Purif. Technol. 51 (2006) 24–30.

Journal of Hazardous Materials 195 (2011) 245–253

Contents lists available at ScienceDirect

Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat

Performance and mechanisms of a microbial-earthworm ecoﬁlter for removing organic matter and nitrogen from synthetic domestic wastewater Longmian Wang a , Zheng Zheng b,∗ , Xingzhang Luo b , Jibiao Zhang b a b

State Key Laboratory of Pollution Control and Resource Reuse, School of the Environment, Nanjing University, Nanjing 210046, PR China Environmental Science & Engineering Department, Fudan University, No. 220, Handan Road, Shanghai 200433, PR China

a r t i c l e

i n f o

Article history: Received 16 January 2011 Received in revised form 22 July 2011 Accepted 11 August 2011 Available online 17 August 2011 Keywords: Ammonia nitrogen Chemical oxygen demand Denaturing gradient gel electrophoresis Rural domestic sewage Vermiﬁltration

a b s t r a c t The performance of a microbial-earthworm ecoﬁlter for the treatment of synthetic domestic wastewater is evaluated, and the mechanisms of organic matter and nitrogen transformation investigated. Vermiﬁltration efﬁciently reduced chemical oxygen demand (COD) and ammonia nitrogen (NH3 -N) from the inﬂuent. A combination of soil with sawdust possessed higher porosity and speciﬁc surface area than other media, and this microporous structure together with wormcast surface greatly facilitated COD reduction at depths from 5 to 35 cm. Nitrogen variations in wastewater were inﬂuenced by soil properties, earthworm activities, and wormcast characteristics. Their interaction with added nitrogen determined soil nitrogen distribution. In addition, denaturing gradient gel electrophoresis (DGGE) proﬁles revealed a highly diverse community of ammonia-oxidizing bacteria (AOB) and Nitrospira in soil layers. There was a positive correlation between the Shannon biodiversity index for AOB and decreasing NH3 -N concentration, indicating that dominant soil microbes played a major role in removing NH3 -N and nitrogen conversion. In contrast to previous reports, identiﬁcation of retrieved sequences of AOB species showed that most belonged to an uncertain AOB genus. This bioﬁltration system is a low cost, efﬁcient alternative for decontaminating local domestic wastewater. © 2011 Elsevier B.V. All rights reserved.

1. Introduction In China, watershed pollution is now a prominent form of water contamination. In contrast to the United States and other developed countries, rural people account for most of the Chinese population, and more than 96% of domestic wastewater in rural areas is discharged directly into aquatic environments without any treatment [1]. It is reported that rural wastewater has seriously affected surface water and groundwater quality, and is the main source of serious river and lake pollution [2,3]. Nitrogen (mainly in the form of ammonia) in wastewater causes eutrophication and potential toxicity to aquatic species [4]. Therefore, the key to watershed pollution is decontamination of wastewater (especially of nitrogen compounds) emitted in rural China. The characteristics of rural wastewater change greatly in quality, quantity, and spatial distribution; therefore, the conventional ﬁxed centralized wastewater treatment technologies used for municipal wastewater treatment are often not suitable [5]. In

Abbreviations: CWs, constructed wetlands; OM, organic matter; DO, dissolved oxygen; S1, topsoil; S2, midlevel soil; S3, subsoil; SD, sand; DE, detritus; VF, vermiﬁltration. ∗ Corresponding author. Tel.: +86 21 65643342; fax: +86 21 65643342. E-mail address: [email protected] (Z. Zheng). 0304-3894/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.jhazmat.2011.08.035

order for resource-scarce, economically developing rural areas to use wastewater treatment, the treatment system should be lowcost, easy to maintain, and highly efﬁcient. To date, many onsite independent and synthetic systems for decentralized sewage treatment have been applied in rural areas, including constructed wetlands (CWs), soil trenches, high-rate algal ponds, vegetationbased wastewater treatment, septic tanks, compound media ﬁlter bed combined systems, and three-stage step-feed wastewater treatment systems combined with drop-aeration bioﬁlms [6–11]. All of these methods have their own unique advantages, with different treatment efﬁcacies for different pollutants. Currently, vermiﬁltration (VF) is a promising method that combines biological technologies and ecological methods, and has been applied in small pilot-scale tests. VF is a process that adapts traditional vermicomposting to a passive wastewater treatment. A typical system will separate the wastewater solids by allowing wastewater to be gravity-fed over ﬁltration material such as ﬁne mesh [12]. Sinha [13] reported that earthworms work as bioﬁlters; they can reduce the 5-day BOD (Biochemical Oxygen Demand) level (BOD5 ) by over 90% and chemical oxygen demand (COD) by 80–90%. The worms ingest and biodegrade organic wastes by absorption through their body walls. Li [14,15] found that a VF system could be a stage in the reuse of swine wastewater by demonstrating that it could successfully treat the sewage produced each day by more than 100

246

L. Wang et al. / Journal of Hazardous Materials 195 (2011) 245–253

Table 1 Physical characteristics of initial ﬁllings.a Items

Unit

Soil + sawdust

Sand

Detritus

Bulk density Porosity Saturated hydraulic conductivity pH Speciﬁc surface area

g cm−3 % cm s−1

0.86 67 0.19

1.96 23 0.27

2.23 16 0.45

m2 g−1

7.14 40.59

8.02 24.35

8.33 1.01

a

All data represent average of triplicates.

swine. COD, ammonia nitrogen (NH3 -N), and BOD5 were efﬁciently reduced in these studies. Furthermore, some innovative technologies combined with traditional methods have been described at both laboratory and pilot scales. These include extensive systems to treat diluted manure consisting of a screen, vermiﬁlter, and settling tank, macrophytes ponds and constructed wetlands. To recycle the water, laboratory-scale ceramsite-vermiﬁlters and VF enhancement with a slag–coal cinder converter ﬁlter have been used for domestic wastewater treatment [16–18]. Although a considerable reduction in the level of COD and NH3 -N in end efﬂuents has been reported using VF, some of the key processes remain to be understood or optimized. If this technology is to be widely applied, then for example, the composition and properties of the different layers, characteristics of wormcast in soil, nitrifying bacterial community structure, and how these interact to effect COD, NH3 -N removal and nitrogen transformation in the process need to be better understood. To obtain a more detailed understanding of organic matter (OM) degradation, NH3 -N removal, and N distribution processes in VF, this study focused on the following four areas: (1) determining the concentration of COD, dissolved oxygen (DO), NH3 -N, nitrate nitrogen (NO3 -N), and nitrite nitrogen (NO2 -N) before and after VF of wastewater at different substratum layers in a laboratory experiment; (2) assessing the effects of the physical characteristics of each layer, the nutrient constitution of soil and wormcast, and the structural features of the wormcast surface on the removal of organic matter, NH3 -N, and nitrogen conversion in sewage; (3) identifying and evaluating the ammonia oxidizing bacterium (AOB) and Nitrospira in ﬁlters by polymerase chain reaction-denaturing gradient gel electrophoresis (PCR-DGGE); and (4) investigating microbial community diversity and composition and determining how the microbial population inﬂuences nitrogen distribution. 2. Experimental 2.1. Experimental design A cuboid shaped plexiglass VF system (40 cm in length, 40 cm in width, and 115 cm in depth) was equipped with a plexiglass wastewater tank. Wastewater was introduced to the apparatus via a peristaltic pump (Fig. 1). The reactor was packed with a 30 cm soil-earthworm bed (6–9 mm diameter), 30 cm sand (100–800 ␮m diameter), 30 cm detritus (3–10 mm diameter) and a 20 cm supporting layer of cobblestone (10–50 mm diameter) layered from top to bottom. Soil and sawdust were mixed at a volume ratio of 3:1. Sawdust was added as a bulking agent because it has been shown to improve soil permeability and enhance earthworm growth and survival [19]. Some physical characteristics of the ﬁllings are presented in Table 1. The artiﬁcial soil was inoculated with Eisenia foetida (Savigny) at an initial earthworm density of 12.5 g L−1 . This species has been widely used in VF [18]. During operation the surface loading of the wastewater was adjusted to 1 m3 m−2 d−1 , and the wet to dry time ratio was 1:3. These approaches prevented the blockage of the soil layers and

sustained the penetrability of the ecoﬁlter. The entire test volume of the synthetic sewage was applied in a single batch through a rotating glass pipe (2 cm in diameter and 36 cm in length) with 1.5 mm holes to ensure uniform distribution of the inﬂuent. The perspex pipe was placed on the upside of the VF surface, 15 cm from the top layer of VF. Synthetic wastewater based on that of Fang [20] was used, see Table 2. From June, 2010 to August, 2010 at Nanjing University, Nanjing (32◦ 03 N, 118◦ 47 E), the system was fed daily and one operation cycle was performed each day, controlled by a digital timer (Kerde TW-K11, Zhejiang, China). Each cycle included wastewater ﬂow for 1 h, retention for 3 h, and ﬁnally emptying of the water. The operation and monitoring of the VF were conducted between operating periods. All the samples including controls were performed in triplicate.

2.2. Water sampling and chemical analysis Inﬂuent and efﬂuent water from the laboratory-scale VF were sampled every week to evaluate treatment performance. Efﬂuent outlets 1–3 allowed separate analysis of direct soil-earthworm layer outﬂow, sand layer outﬂow, and detritus layer outﬂow (Fig. 1) at depths of 35, 65, and 95 cm below the surface, respectively. The total efﬂuent outlet was at the bottom left. Water samples were analyzed for COD, TN, total phosphorus (TP), NH3 -N, NO3 -N, and NO2 -N according to the methods described in Examination of Water and Wastewater [21]. The DO, pH, and temperature were measured in situ (using a YSI Model no. 550A DO meter, USA, and a pH meter from Shanghai Kangyi Instrument Co. Ltd., PHS-2C, China).

2.3. Sampling and analysis of soil layers and wormcast When emptying the sewage packing, samples of topsoil (S1, 5–15 cm), midlevel soil (S2, 15–25 cm), and subsoil (S3, 25–35 cm) were collected from sampling points at intervals of 15 days. In total, 35–65 cm sand (SD) and 65–95 cm detritus (DE) were collected at the end of each period. Samples from the same depth were mixed to give one composite sample. Subsequently, plant roots, earthworms, and other waste were removed from the substrates. Earthworm cast was gathered before and after the operation as described by Zhao [22], and the system was replenished with an equivalent weight of earthworms. Finally, both the ﬁlter materials and the casts were freeze-dried, sieved (<2 mm) except DE, and stored at −20 ◦ C for analysis. The speciﬁc surface area of various ﬁlters was measured at room temperature and atmospheric pressure by a BET method with nitrogen using a NOVA 3000e surface area and pore size analyzer (Quantachrome Instruments, USA). Porosity and bulk density were determined using standard soil science methods [23]. The saturated hydraulic conductivity was measured using the penetration tube method. The pH of the ﬁlter material was measured in a 10% (w/v) aqueous solution using a digital pH meter. The measurements of OM, total organic carbon (TOC), TN, NH3 -N, and NO3 -N were carried out in soil before the operation of VF (at day 0) and every 15 days thereafter, and in wormcast before and after the ﬁnal running period. Organic matter was determined by the loss on ignition at 550 ◦ C for 2 h [24]. Total organic carbon and TN were determined using an elemental analyzer (Elementar Vario EL III, Germany), while NH3 -N and NO3 -N were measured using the KCl extraction–distillation method and the Cu–Cd reductioncolorimetry method, respectively [23]. Organic nitrogen (Norg ) was calculated by subtracting NH3 -N and NO3 -N from TN. Scanning electron microscopy (SEM) of wormcast collected at the end of the operation period was conducted using an S-3400N II microscope (Hitachi, Japan).

L. Wang et al. / Journal of Hazardous Materials 195 (2011) 245–253

247

Table 2 Physical and chemical characteristics of the synthetic wastewater used as the inﬂuent.a Parameters a

Values a

pH

Temperature (◦ C)

DO (mg L−1 )

COD (mg L−1 )

TN (mg L−1 )

NH3 -N (mg L−1 )

TP (mg L−1 )

7.62 ± 0.25

27.82 ± 1.36

5.76 ± 0.51

243.76 ± 13.97

41.00 ± 1.17

39.30 ± 0.77

3.12 ± 0.11

Values (mean ± standard deviation) are averages of three replicates.

Table 3 Primers used in this study. Prime set

Target group

Primer sequence 5 –3

Reference

amoA-1f-GC amoA-2r NSR1113f-GC NSR1264r

ammonia oxidizing bacteria ammonia oxidizing bacteria Nitrospira Nitrospira

CGCCCGCCGCGCGCGGCGGGCGGGGCGGGG GCACGGGGGGa GGG GTT TCTA CTG GTG GT CCC CTC KGS AAA GCC TTC TTC CGC CCG CCG CGC GCG GCG GGC GGG GCG GGG GCA CGG GGG Ga CCT GCT TTC AGT TGC TAC CG GTT TGC AGC GCT TTG TAC CG

[25,26] [25] [26,27] [27]

a

The GC-clamp sequences used for PCR-DGGE are written in bold.

Fig. 1. Diagram of the laboratory-scale VF system used in this study (unit: cm).

2.4. DNA extraction and PCR-DGGE Samples of DNA were extracted from soil, sand, and detritus samples taken after 90 days of wastewater treatment using the Ultra Clean Soil DNA Isolation Kit (MO BIO Laboratories, Loker Ave West, Carlsbad, CA, USA), according to the manufacturer’s instructions. These DNA preparations were used as template DNAs for PCR. To amplify speciﬁc 16S rDNA from samples, we used a nested PCR approach. Each PCR reaction was conducted in a volume of 50 ␮L using EDC-810 Thermal Cyclers (Eastwin, Beijing, China). The PCR mixture contained 0.25 ␮L DNA polymerase (5 U ␮L−1 ) (TaKaRa, Ex Taq, Japan), 5 ␮L 10× Ex Taq Buffer, 4 ␮L MgCl2 (25 mM), 4 ␮L dNTP mixture (2.5 mM each), and 1 ␮L primer (20 ␮M each). One ␮L of template DNA (0.6–1.0 ng) was added to each reaction with the speciﬁc primer pair amoA-1F and amoA-2R for ammonia oxidizing bacteria or NSR1113f/NSR1264r for Nitrospira, see Table 3. The latter genus was selected because it is the most abundant nitrite oxidizer in wastewater treatment systems [28]. As PCR was performed to generate products for subsequent DGGE analysis, we

used primers containing GC-clamp at the amoA-1F and NSR1113f 5 -end. The cycling program for the PCR was 94 ◦ C for 3 min, followed by 35 cycles of 1 min at 94 ◦ C, 90 s at 55 ◦ C, and 90 s at 72 ◦ C, and a ﬁnal extension of 10 min at 72 ◦ C. The program used for ampliﬁcation of Nitrospira 16S rDNA sequences was 30 cycles of 1 min at 95 ◦ C, 1 min at 55 ◦ C, and 2 min at 72 ◦ C. For veriﬁcation, aliquots of the PCR products were separated on an agarose gel (1.2%, 100 V, 25 min) and DNA bands were visualized by ethidium bromide. Denaturing gradient gel electrophoresis (DGGE) was performed using the Bio-Rad D gene system (Bio-Rad, USA). Aliquots of the PCR fragments were loaded onto 8% (wt/vol) polyacrylamide gels in 1× TAE buffer. The gel was run on denaturing gradients of 35–60% (100% denaturant is 7 M urea plus 40% formamide) for AOB, or 40–60% for Nitrospira for about 6 h at 60 ◦ C and 150 V. After electrophoresis, the gels were soaked for 15 min in ethidium bromide (250 mL Milli-Q water, 25 ␮L ethidium bromide stock of 5 mg mL−1 ) and subsequently rinsed for 20 min with Milli-Q water. Gel images using UV translumination were stored by using the Gel Doc 2000 System from Bio-Rad (Bio-Rad, USA).

248

L. Wang et al. / Journal of Hazardous Materials 195 (2011) 245–253

Table 4 Organic matter, total organic carbon, and nitrogen distribution in the layers of soil and wormcast. Parameters

Initial soil + sawdusta

OM (%) TOC (g kg−1 ) TN (g kg−1 ) NH3 -N (g kg−1 ) NO3 -N (mg kg−1 ) Norg (g kg−1 )

8.59 29.4 4.6 1.7 24.8 2.9

a b

± ± ± ± ± ±

0.21 0.5 0.1 0.01 0.2 0.03

5–15 cm Soilb 12.64 39.7 5.2 3.1 37.6 2.1

± ± ± ± ± ±

0.85 1.7 0.3 0.1 2.7 0.03

15–25 cm Soilb 12.78 35.4 4.8 2.7 38.5 2.1

± ± ± ± ± ±

1.03 2.1 0.6 0.04 3.9 0.2

25–35 cm Soilb 12.90 46.9 6.1 3.4 40.2 2.7

± ± ± ± ± ±

0.46 2.4 1.04 0.5 3.2 0.06

Initial wormcasta 15.56 66.1 6.2 0.042 75.2 6.16

± ± ± ± ± ±

1.34 4.7 0.01 0.012 5.4 0.051

Wormcasta 17.31 64.3 6.3 0.036 86.3 6.18

± ± ± ± ± ±

0.13 3.6 0.01 0.003 4.8 0.062

Values (mean ± standard deviation) are averages of three replicates. Arithmetic averages of six samplings (mean ± standard deviation) during operation period.

2.5. Cloning, sequencing, and phylogenetic analysis Gel slices from the DGGE were put into 1.5 mL micro-centrifuge tubes containing 25 ␮L TE buffer and incubated for 24 h at 4 ◦ C. The DNA eluted from a small gel chip was used as a direct template for PCR to recover the DNA fragments. The PCR conditions were the same as for the original PCR, except that the primer pairs amoA-1F/amoA-2R and NSR1113f/NSR1264r were without GC-clamps. The PCR products ampliﬁed from DGGE gel were subjected to agarose gel electrophoresis. Subsequently, DNA fragments were repuriﬁed with an Agarose Gel DNA Puriﬁcation Kit Ver.2.0 (TaKaRa, Japan). Puriﬁed DNA from AOB was used as a template for direct sequencing at GenScript Inc. (Nanjing, China). Fragments less 200 bp were not suitable for direct sequencing [29], so puriﬁed PCR-ampliﬁed DNA from Nitrospira (151 bp) DGGE gel were further cloned using the pTG19-T PCR Product Cloning Kit (Generay, Shanghai, China) prior to sequencing, following the same method described above. The Shannon index was calculated to assess the structural diversity (richness and evenness) of the microbial community: s

H=−

i=1

s n i

pi log pi = −

i=1

N

log

n i

N

where H is the Shannon Index, ni is the height of the peak, and N is the sum of all peak heights in the curves [22]. The nucleotide sequences were compared with those deposited in the GenBank (NCBI) using Basic Local Alignment Search Tool (BLAST), then the sequences determined in this study and obtained from the DNA database were aligned using ClustalW. Distance matrix analyses using the p-distance, and neighbor-joining trees were constructed by pair-wise deletion using MEGA version 4.0 (Molecular Evolutionary Genetics Analysis). Tree topology was evaluated by bootstrap analysis using 1000 replicates. The sequences generated in this study have been deposited in the National Center for Biotechnology Information under accession numbers JF742542 to JF742558.

Fig. 2. Performance of laboratory-scale microbial-earthworm ecoﬁlters for wastewater treatment a: Removal of chemical oxygen demand (COD) and dissolved oxygen (DO) at different depths; b: ammonia nitrogen (NH3 -N), nitrate nitrogen (NO3 -N) and nitrite nitrogen (NO2 -N) variations in inﬂow and different efﬂuents by analysis of distribution in the ﬁlters.

3. Results and discussion 3.1. The performance and mechanisms involved in the OM removal The efﬂuent COD declined sharply from 5 to 35 cm, whereas it was more or less constant between 35 and 115 cm (Fig. 2a). The total efﬂuent COD was maintained below 50 mg L−1 at all times in the VF unit during synthetic wastewater treatment. The physical, chemical, and biological processes and synergistic effects of earthworms and microorganisms, including the adsorption of small particle organisms, colloid organisms, and organic molecules, as well as the oxidation-reduction of organic matter and activity of earthworms, made VF effective for COD removal [30]. Among the VF layers, the soil-earthworm layer with added sawdust played a major role because of its higher porosity and surface area (Table 1), which

proved beneﬁcial for removing most of the organic contaminants by precipitation and adsorption in the voids of the soil. Meanwhile, the micrographs of wormcast (Fig. 3) revealed micro-pores on the surface and abundant cylindrical organic matter within, suggesting that wormcast acted as a kind of wastewater ﬁlter medium for removing OM. Since the main decrease in COD content was found over the ﬁrst 5–35 cm depth, and the porosity and speciﬁc surface area were lower in the sand and detritus layers than those of the soil, the organic substrates had poor availability. Thus, COD content did not change dramatically over these lower layers. Because the COD was removed efﬁciently from wastewater mainly by means of precipitation and adsorption, the soil OM and TOC contents were observed to increase (Table 4). A large number of wormcasts in the VF produced by earthworms were rich in

L. Wang et al. / Journal of Hazardous Materials 195 (2011) 245–253

249

Fig. 3. A SEM image of a sample of wormcast (5000×).

Fig. 4. Denaturing gradient gel electrophoresis (DGGE) analysis of ammonia-oxidizing bacteria (AOB) (a) and Nitrospira (b) fragments retrieved from layers of the VF system. The excised bands 1–11 and bands 12–18 denote AOB and Nitrospira, respectively. S1–S3, SD, and DE represent topsoil, midlevel soil, subsoil, sand, and detritus, respectively. The same mobilities in the DGGE gel are represented by and .

organic matter and TOC (Table 4). It has been estimated that after several years all the surface soil in an earthworm habitat may be completely transformed into casts [31]. Therefore, the content of wormcast brings about sustained growth of the nutrient content of soil. 3.2. NH3 -N removal and nitrogen transformations 3.2.1. The performance of nitrogen variation in wastewater Fig. 2b shows the effect of depth on N species variation in sewage. The NH3 -N in inﬂow and outﬂow at 35 cm depth

comprised the largest component of the N-species. NH3 -N content underwent a dramatic decline between depths of 5 and 35 cm, followed by a further decline from 35 cm to 65 cm. No further signiﬁcant change occurred below 65 cm. There was both a low concentration and low variation of NO2 -N between the inﬂuent and different efﬂuents. It was also shown that efﬂuent NO3 -N concentration greatly increased from 35 to 65 cm, while it was dominant in the various forms of N between 65 cm and 115 cm. The majority of nitrogen present as NH3 -N from the synthetic wastewater was removed mainly in the soil and sand parts of the reactor through rapid adsorption by the biomass and ﬁlters,

250

L. Wang et al. / Journal of Hazardous Materials 195 (2011) 245–253

Conversely, the ratio of Norg /TN decreased slightly with soil depth, although the majority of N in the form of Norg was detected in wormcast. Consistent with previous work [34], earthworm activity was found to be a major determinant of Norg in soil, as these animals promote nitrogen mineralization. The increased NH3 -N in the top soil might be from the adsorption or absorption of NH4 + ions from inﬂuent onto the mineral or organic fraction of the soil, as ammonium ions are relatively inert to cation exchange [35]. The increased VF NH3 -N was concomitant with removal from the wastewater. Meanwhile, earthworm activity and the production of cast was found to be critical for NH3 -N reduction in the wastewater, as it oxygenates the inﬂuent and this oxygenation facilitates the nitriﬁcation of ammonia by microbes [36]. The ﬁber shown in Fig. 3 might be derived from sawdust. The inclusion of sawdust was also useful to decrease NH3 -N in sewage because it acted as a biosorbent through complexation with NH4 + ions [37]. The concentration of NO3 -N was low in all soils and wormcast, but increased with soil depth (Table 4). This is likely caused by surface earthworm activity, nitriﬁcation, and soil properties. It is well known that earthworms mediate the conversion of organic nitrogen to inorganic nitrogen and that nitriﬁcation promotes the formation of nitrate [33,34]. Furthermore, soil particles have the capacity to retain pollutants and capture NO3 -N from wastewater. Similar ﬁndings were obtained by Wong et al. [38]. Though nitrate nitrogen in soil increased with depth, the majority represented downward transport with sewage outﬂow.

Fig. 5. The relationship between decreasing NH3 -N and Shannon index for AOB (a), and increasing NO3 -N and Shannon index for Nitrospira (b) in different padding. The soil Shannon index is the average value of topsoil, midlevel soil, and subsoil.

and the adsorbed NH3 -N was subsequently converted to NO3 N via biological nitriﬁcation, which was carried out by aerobic, autotrophic bacteria using molecular oxygen as an electron acceptor [32]. Meanwhile the high surface DO concentration (Fig. 2a) was beneﬁcial for aerobic microbial survival, which in turn was advantageous for nitriﬁcation. Thus, the majority of the NH3 -N removal occurred at depths between 5 and 65 cm. In addition, NO2 -N concentrations remained low in all efﬂuent samples. This is thought to be due to the role of nitrites as intermediates in nitriﬁcation. Furthermore, the nitriﬁcation step for NH3 -N removal led to a substantial increase in NO3 -N and high DO consumption, the accumulated NO3 -N could not be removed effectively between 65 cm and 115 cm through denitriﬁcation because of the large quantities of organic carbon which served as a carbon source for denitriﬁcation in upper layers but not in deeper layers [33]. 3.2.2. Effect of the physicochemical characteristics of soil and wormcast The nitrogen distribution across the soil strata and wormcast, which play dominant roles in the N cycle, was studied to better understand nitrogen transformation mechanisms in VF. In the upper soil, TN increased, mainly due to the NH3 -N proﬁle that increased from 1.7 g kg−1 initially, to 3.1 g kg−1 at 5–15 cm, 2.7 g kg−1 at 15–25 cm and 3.4 g kg−1 at 25–35 cm (Table 4).

3.2.3. The effect of AOB and Nitrospira community structure and diversity The AOB and Nitrospira communities in soil, sand, and detritus from the lab-scale VF are shown in Fig. 4. The DGGE proﬁles revealed that the dominant communities were in S1–S3, with no signiﬁcant differences between them. However, soil nitrifying bacterial structure greatly differed between sand and detritus. Similarly, evaluation of the bacterial diversity (Table 5) based on the Shannon index showed that microbial diversity was higher in soil than in sand or detritus. These results indicate that soil is the most suitable substrate for nitrifying bacterial communities in this reactor, and that earthworm activities had no signiﬁcant effect on the community structure of nitrobacteria at different soil depths. Filter body effects in combination with the loss of organic substrates most likely contributed to this observation. Microorganisms easily attach to the surfaces and micro-pores of soil particles. Some groups like indigenous ammonium oxidizers attach more strongly to clay particles than do most heterotrophic bacteria [39,40]. In addition, lower substrate and nutrient availability in the sand and detritus, due to removal of organic pollutants in the upper layer of the VF, resulted in lower growth of microbes in these media [41]. It is probable that the earthworms maintain the balance of nitriﬁers with the help of aerobic and anaerobic microﬂora in their guts, which could explain why adding earthworms had no obvious inﬂuence on the diversity of nitrobacterium. The distinct features of these bands are also shown in Fig. 4. The majority (e.g., bands 2, 3, 4, and 14) exhibited higher intensities in S1–S3 (especially S2), indicating that particular bacterial species existed in soil at relatively high abundance within the nitriﬁcation groups. These ﬁndings may also be due to the higher density of earthworms in the 10–25 cm strata. Here, sufﬁcient oxygen was available, and the burrowing action of the earthworms resulted in improved aerobic conditions in the soil body, thereby creating a favored microenvironment for aerobic nitrobacteria. By examining the relationship between concentration changes of different N forms and the Shannon index for AOB and Nitrospira, the effect of bacterial diversity on NH3 -N removal and NO3 -N distribution for wastewater treatment was determined. It was found that there was a signiﬁcant positive correlation between

L. Wang et al. / Journal of Hazardous Materials 195 (2011) 245–253

251

Table 5 Contrast of Shannon-Wiener diversity index in different paddings of the VF.

Number of bands Shannon index (H)

Targets

S1

S2

S3

SD

DE

AOB Nitrospira AOB Nitrospira

4 7 0.58476 0.76296

6 6 0.65495 0.69021

6 6 0.64583 0.63185

3 1 0.45888 0

3 1 0.39854 0

Fig. 6. Phylogenetic tree based on neighbor-joining analysis of gene sequences from ammonia oxidizers with a length of 491 bp (a) and Nitrospira with a length of 151 bp (b). Bands 1–18 represent sample sequences. Numbers at the branches represent bootstrap values, and the scale bar indicates 5 changes per 100 amino acid positions.

decreasing NH3 -N and Shannon index for AOB (R2 = 0.9995), while there was no apparent relation between increasing NO3 -N and Shannon index for Nitrospira (Fig. 5). It is obvious that AOB diversity increased with decreasing ammonia nitrogen concentration (or as a function of improved ammonia nitrogen removal), which is consistent with the relatively large decline in the NH3 -N concentration in the soil-earthworm layer (Fig. 2). The correlation between nitrate nitrogen concentration and Shannon index suggests that there are other factors affecting NO3 -N transformation besides the diversity of Nitrospira. Because of nitrate transformation into nitrogen via denitriﬁcation [33], a minor increase in NO3 -N concentration with

increasing Shannon index for Nitrospira in TOC-enriched soil was observed. 3.2.4. Identiﬁcation of AOB and Nitrospira in the VF The numbered bands in the DGGE gel were excised for sequencing. Bands with the same mobility in the DGGE gel (e.g., bands 3, 3 , and 3 ) had the same nucleotide sequences. Thus, the same band name (e.g., band 3) was used to describe the sequences used for phylogenetic analysis. Their phylogenetic positions are illustrated by the neighbor-joining tree in Fig. 6. The closest matches of amoA for all sequences ampliﬁed from these bands

252

L. Wang et al. / Journal of Hazardous Materials 195 (2011) 245–253

(except band 6 which was not related to others) were either uncultured ammonia oxidizers or environmental clone sequences falling within the beta proteobacterial class. The obtained sequences showed between 97 and 100% similarity with previously identiﬁed gene sequences. Results indicated that the AOB community consisted mainly of unidentiﬁed ammonia oxidizers and a unique Nitrosomonas-like genus grouped in the Nitrosomonadaceae family. The unidentiﬁed AOB contributed four bands, six bands, ﬁve bands, two bands, and three bands to the ammonia oxidizer community in S1, S2, S3, SD, and DE, respectively (Fig. 4a and 6a). Sequences similar to these unidentiﬁed fragments have been observed in other soils (EU275281, GQ143590, FJ890547, AB514947, FJ517385, GU377312), and in a sediment (FJ951797) and constructed wetland (GQ255588), where they had a signiﬁcant inﬂuence on nitriﬁcation and the AOB community composition. Furthermore, the only sequence retrieved from band 9 in SD was placed within the Nitrosomonas-like genus. Obtained sequences from AOB in VF were chemolithoautotrophs, having in common the ability to utilize ammonia as a sole source of energy and mediate transformation of NH3 -N to NO2 -N [42]. In contrast, within the population of AOB, a higher percentage of Nitrosospira and Nitrosomonas sequences have been found in constructed wetlands and conventional wastewater treatment plants [43–45]. This apparently distinct microbial distribution in different bioﬁlters could arise for two types of reason; (1) if there was an insufﬁcient quantity of bacterial template DNA for PCR ampliﬁcation for the rarer groups in the AOB community, then the single more abundant group could appear to be the only one present, or, (2), it may also be that the system is relatively immature and thus the different populations have not had time to reach a long-term equilibrium. As shown in Fig. 5b, all the strains shared 95–97% of the sequences identical with those retrieved from the database. Sequences analysis revealed that the whole 16S DNA sequences retrieved from bands 12 to 18, except for 15, were related to Nitrospira, while band 15, which was present in all ﬁlters, was grouped with the unidentiﬁed bacterium in bioﬁlm or ﬁlters. Since Nitrospira thrive under conditions of relative nitrite scarcity and neutral pH [46,47], they were able to survive in this system (see Table 2 and Fig. 2) and were responsible for an important step in the N cycle, the oxidation of NO2 -N to NO3 -N [42]. This outcome is consistent with results showing that Nitrospira members of nitrite-oxidizing bacteria, usually dominant in bioﬁlms from wastewater treatment plants and soils [47,48], were present here.

4. Conclusion Because the microbial-earthworm ecoﬁlter was able to significantly reduce the COD and NH3 -N from artiﬁcial wastewater, it may be an effective technology for domestic wastewater treatment. A stable removal of COD was achieved and the majority was trapped by the upper 5–35 cm depths. This COD reduction was greatly facilitated by the addition of sawdust to the soil, which enhanced porosity and speciﬁc surface area, and by the structure of the wormcast surface. The soil layers were responsible for the major fraction of nitrogen variations, in addition to determining wormcast characteristics, earthworm activity, and nitrifying bacteria diversity. According to sequences analysis, the dominant AOB members were unidentiﬁed ammonia oxidizers and the majority of sequences retrieved from bands 12 to 18 were related to Nitrospira. In conclusion, these results could yield a deeper understanding of the mechanism of organic matter and nitrogen removal in VF systems. These laboratory-scale studies may allow extension of this methodology to applications like controlling water pollution in rural areas that cannot be served by conventional sewage systems.

Acknowledgments The authors would like to express their gratitude to the National Water Pollution Project for Taihu Lake Pollution Control of China (2008ZX07101-004, 2008ZX07101-005) and Shanghai Natural Science Foundation (11ZR1402900) for their ﬁnancial support. They would also like to acknowledge the ﬁnancial assistance from the National Natural Science Foundation of China (40803025, 11075039).

References [1] MHURDPRC, Present Situation and Problems of Rural Human Settlements, Ministry of Housing and Urban-Rural Development of the People’s Republic of China Beijing, China, 2005. [2] M. Chen, J. Chen, P. Du, An inventory analysis of rural pollution loads in China, Water Sci. Technol. 54 (2006) 65–74. [3] Q.M. Li, W. Zhang, X.X. Wang, Y.Y. Zhou, H. Yang, G.L. Ji, Phosphorus in interstitial water induced by redox potential in sediment of Dianchi Lake, China, Pedosphere 17 (2007) 739–746. [4] F.X. Ye, Y. Li, Enhancement of nitrogen removal in towery hybrid constructed wetland to treat domestic wastewater for small rural communities, Ecol. Eng. 35 (2009) 1043–1050. [5] G. Ho, Small water and wastewater systems: pathways to sustainable development? Water Sci. Technol. 48 (2003) 7–14. [6] J.H. Ham, C.G. Yoon, J.H. Jeon, H.C. Kim, Feasibility of a constructed wetland and wastewater stabilisation pond system as a sewage reclamation system for agricultural reuse in a decentralised rural area, Water. Sci. Technol. 55 (2007) 503–511. [7] D.Y. Xu, B.H. Zhou, X.X. Duan, Z.K. Li, J.L. Wang, Y.H. Zhang, M. Zhao, Application of soil trench system in rural sewage treatment in Erhai Lake Basin, in: Beijing International Environmental Technology Conference Beijing, Peoples R China, 16–19 October, 2009, pp. 110–115. [8] X.F. Huang, Y. Wen, S.L. He, G. Chen, X.D. Li, D.H. Yang, Q. Zhou, Migration and transformation of nitrogen in a HRAP treating domestic wastewater in rural area, Huanjing Kexue (Chinese) 29 (2008) 2219–2226. [9] A. Kaoru, K. Kunihiko, O. Yasuo, Vegetation-based wastewater treatment technologies for rural areas in Japan, JARQ 44 (2010) 231–242. [10] L. Zhang, S. Hong, C. Wen, X.H. Mao, A. Hui, F.X. Gan, A novel combined system for onsite domestic wastewater treatment in rural areas, Environ. Eng. Sci. 26 (2009) 775–782. [11] H.W. Liang, M. Gao, J.X. Liu, Y.S. Wei, X.S. Guo, A novel integrated step-feed bioﬁlm process for the treatment of decentralized domestic wastewater in rural areas of China, J. Environ. Sci. China 22 (2010) 321–327. [12] R.J. Hughes, J. Nair, G. Ho, The risk of sodium toxicity from bed accumulation to key species in the vermiﬁltration wastewater treatment process, Bioresource. Technol. 100 (2009) 3815–3819. [13] R.K. Sinha, B. Bharambe, U. Chaudhari, Sewage treatment by vermiﬁltration with synchronous treatment of sludge by earthworms: a low-cost sustainable technology over conventional systems with potential for decentralization, Environmentalist 28 (2008) 409–420. [14] Y.S. Li, P. Robin, D. Cluzeau, M. Bouché, J.P. Qiu, A. Laplanche, M. Hassouna, P. Morand, C. Dappelo, J. Callarec, Vermiﬁltration as a stage in reuse of swine wastewater: Monitoring methodology on an experimental farm, Ecol. Eng. 32 (2008) 301–309. [15] Y.S. Li, Y.Q. Xiao, J.P. Qiu, Y.Q. Dai, P. Robin, Continuous village sewage treatment by vermiﬁltration and activated sludge process, Water Sci. Technol. 60 (2009) 3001–3010. [16] P. Morand, P. Robin, A. Escande, B. Picot, J.P. Qiu, Y.S. Li, G. Hamon, C. Amblard, S. Fievet, D. Oudart, A.M. Pourcher, B. Picot, B. Landrain, Test of an extensive system combining screen, vermiﬁlter, macrophyte lagoons, and constructed wetlands to recover nutrients from fresh manure and recycle water for ﬂushing in pig housing, in: ISEST. International Symposium on Environmental Science and Technology, Shanghai, China, 02–05 Jun, 2009, pp. 1122–1132. [17] J. Liu, Z.B. Lu, J. Yang, M.Y. Xing, F. Yu, Ceramsite-vermiﬁlter for domestic wastewater treatment and reuse: an option for rural agriculture, in: ICEET. International Conference on Energy and Environment Technology, Guilin, China, 16–18 October, 2009, pp. 472–475. [18] S. Wang, J. Yang, S.J. Lou, J. Yang, Wastewater treatment performance of a vermiﬁlter enhancement by a converter slag–coal cinder ﬁlter, Ecol. Eng. 36 (2010) 489–494. [19] M. Khwairakpam, R. Bhargava, Vermitechnology for sewage sludge recycling, J. Hazard. Mater. 161 (2009) 948–954. [20] C.X. Fang, Z. Zheng, X.Z. Luo, F.H. Guo, Effect of hydraulic load on domestic wastewater treatment and removal mechanism of phosphorus in earthworm ecoﬁlter, Fresen. Environ. Bull. 19 (2010) 1099–1108. [21] China State Environmental Protection Administration, Water and Wastewater Testing Analysis Methods, fourth ed., China Environmental Science, Beijing, 2002. [22] L.M. Zhao, Y.Y. Wang, J. Yang, M.Y. Xing, X.W. Li, D.H. Yi, D.H. Deng, Earthworm–microorganism interactions: a strategy to stabilize domestic wastewater sludge, Water Res. 44 (2010) 2572–2582.

L. Wang et al. / Journal of Hazardous Materials 195 (2011) 245–253 [23] G.S. Liu, Soil Physical and Chemical Analysis & Description of Soil Proﬁles, ﬁrst ed., Standards Press of China, Beijing, 1996. [24] J.Y. Liu, H. Wang, H.J. Yang, Y.J. Ma, O.C. Cai, Detection of phosphorus species in sediments of artiﬁcial landscape lakes in China by fractionation and phosphorus-31 nuclear magnetic resonance spectroscopy, Environ. Pollut. 157 (2009) 49–56. [25] J.H. Rotthauwe, K.P. Witzel, W. Liesack, The ammonia monooxygenase structural gene amoA as a functional marker: molecular ﬁne-scale analysis of natural ammonia-oxidizing populations, Appl. Environ. Microbiol. 63 (1997) 4704–4712. [26] G. Muyzer, E.D. Waal, A.G. Uitterlinden, Proﬁling of complex microbial populations by denaturing gradient gel electrophoresis analysis of polymerase chain reaction-ampliﬁed genes coding for 16S rRNA, Appl. Environ. Microbiol. 59 (1993) 695–700. [27] H.M. Dionisi, A.C. Layton, G. Harms, I.R. Gregory, K.G. Robinson, G.S. Sayler, Quantiﬁcation of autotrophic ammonia-oxidizing bacteria and Nitrospira spp. from full-scale wastewater treatment plants using competitive PCR, Appl. Environ. Microbiol. 68 (2002) 245–253. [28] H. Daims, J.L. Nielsen, P.H. Nielsen, K.H. Schleifer, M. Wagner, In situ characterization of Nitrospira-like nitrite-oxidizing bacteria active in wastewater treatment plants, Appl. Environ. Microbiol. 67 (2001) 5273–5284. [29] N. Fromin, J. Hamelin, S. Tarnawski, D. Roesti, K. Jourdain-Miserez, N. Forestier, S. Teyssier-Cuvelle, F. Gillet, M. Aragno, P. Rossi, Statistical analysis of denaturing gel electrophoresis (DGGE) ﬁngerprinting patterns, Environ. Microbiol. 4 (2002) 634–643. [30] M.B. Bouché, P. Soto, An industrial use of soil animals for environment: the treatment of organically polluted water by lumbriﬁltration, in: Proceedings of the 14th International Colloquium on Soil Zoology and Ecology, University of Rouen, Mont Saint Aignan, France, August 30–September 3, 2004, pp. 1–13. [31] C.A. Edwards, Earthworm Ecology, second ed, CRC, Boca Raton, 2004. [32] J. Zhang, X. Huang, C.X. Liu, Nitrogen removal enhanced by intermittent operation in a subsurface wastewater inﬁltration system, Ecol. Eng. 25 (2005) 419–428. [33] G. Maltais-Landry, R. Maranger, J. Brisson, F. Chazarenc, Nitrogen transformations and retention in planted and artiﬁcially aerated constructed wetlands, Water Res. 43 (2009) 535–545. [34] Y. Ma, N.M. Dickinson, M.H. Wong, Beneﬁcial effects of earthworms and arbuscular mycorrhizal fungi on establishment of leguminous on Pb/Zn mine tailings, Soil Biol. Biochem. 38 (2006) 1403–1412. [35] R. Nieder, D.K. Benbi, H.W. Scherer, Fixation and deﬁxation of ammonium in soils: a review, Biol. Fertil. Soils (2010), doi:10.1007/s00374-010-0506-4.

253

[36] M. Taylor, W.P. Clarke, P.F. Greenﬁeld, The treatment of domestic wastewater using small-scale vermicompost ﬁlter beds, Ecol. Eng. 21 (2003) 197–203. [37] M.A. Wahab, S. Jellali, N. Jedidi, Ammonium biosorption onto sawdust: FTIR analysis, kinetics and adsorption isotherms modeling, Bioresour. Technol. 101 (2010) 5070–5075. [38] Y.S. Wong, C.Y. Lan, G.Z. Chen, S.H. Li, X.R. Chen, Z.P. Liu, Effect of wastewater discharge on nutrient contamination of mangrove soils and plants, Hydrobiologia 295 (1995) 243–254. [39] A. Aakra, M. Hesselsoe, L.R. Bakken, Surface attachment of ammonia-oxidizing bacteria in soil, Microbiol. Ecol. 39 (2000) 222–235. [40] V. Torsvik, L. Ovreas, Microbial diversity and function in soil: from genes to ecosystems, Curr. Opin. Microbiol. 5 (2002) 240–245. [41] A. Tietz, A. Kirschner, G. Langergraber, K. Sleytr, R. Haberl, Characterisation of microbial biocoenosis in vertical subsurface ﬂow constructed wetlands, Sci. Total Environ. 380 (2007) 163–172. [42] K.G. Robinson, H.M. Dionisi, G. Harms, A.C. Layton, I.R. Gregory, G.S. Sayler, Molecular assessment of ammonia- and nitrite-oxidising bacteria in full-scale activated sludge wastewater treatment plants, Water Sci. Technol. 48 (2003) 119–126. [43] A.K. Rowan, J.R. Snape, D. Fearnside, M.R. Barer, T.P. Curtis, I.M. Head, Composition and diversity of ammonia-oxidising bacterial communities in wastewater treatment reactors of different design treating identical wastewater, FEMS Microbiol. Ecol. 43 (2003) 195–206. ˜ [44] O. Ruiz-Rueda, S. Hallin, L. Baneras, Structure and function of denitrifying and nitrifying bacterial communities in relation to the plant species in a constructed wetland, FEMS Microbiol. Ecol. 67 (2009) 308–319. [45] R. Gorra, M. Coci, R. Ambrosoli, H.J. Laanbroek, Effects of substratum on the diversity and stability of ammonia-oxidizing communities in a constructed wetland used for wastewater treatment, J. Appl. Microbiol. 103 (2007) 1442–1452. [46] R. Nogueira, L.F. Melo, Competition between Nitrospira spp. and Nitrobacter spp. In nitrite-oxidizing bioreactors, Biotechnol. Bioeng. 95 (2006) 169–175. [47] E. Attard, F. Poly, C. Commeaux, F. Laurent, A. Terada, B.F. Smets, S. Recous, X.L. Roux, Shifts between Nitrospira- and Nitrobacter-like nitrite oxidizers underlie the response of soil potential nitrite oxidation to changes in tillage practices, Environ. Microbiol. 12 (2010) 315–326. [48] R. Blackburnea, V.M. Vadivelu, Z.G. Yuan, J. Keller, Kinetic characterisation of an enriched Nitrospira culture with comparison to Nitrobacter, Water Res. 41 (2007) 3033–3042.

Journal of Hazardous Materials 195 (2011) 254–260

Contents lists available at ScienceDirect

Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat

Evidence for shifts in the structure and abundance of the microbial community in a long-term PCB-contaminated soil under bioremediation ´ I. Petric´ a,c,∗ , D. Bru b , N. Udikovic-Koli c´ a , D. Hrˇsak a , L. Philippot b , F. Martin-Laurent b,c a b c

Rudjer Boskovic Institute, Division for Marine and Environmental Research, 10002 Zagreb, Croatia INRA, Université de Bourgogne, Soil and Environmental Microbiology, 21065 Dijon Cedex, France Welience Agro-Environnement, BP 66517, 21065 Dijon Cedex, France

a r t i c l e

i n f o

Article history: Received 26 April 2011 Received in revised form 22 July 2011 Accepted 11 August 2011 Available online 17 August 2011 Keywords: Polychlorinated biphenyls Bioremediation Microbial community structure Quantitative PCR Ribosomal intergenic spacer analysis

a b s t r a c t Although the impact of bioremediation of PCB-contaminated sites on the indigenous microbial community is a key question for soil restoration, it remains poorly understood. Therefore, a small-scale bioremediation assay made of (a) a biostimulation treatment with carvone, soya lecithin and xylose and (b) two bioaugmentation treatments, one with a TSZ7 mixed culture and another with a Rhodococcus sp. Z6 pure strain was set up. Changes in the structure of the global soil microbial community and in the abundances of different taxonomic phyla were monitored using ribosomal intergenic spacer analysis (RISA) and real-time PCR. After an 18-month treatment, the structure of the bacterial community in the bioremediated soils was signiﬁcantly different from that of the native soil. The shift observed in the bacterial community structure using RISA analysis was in accordance with the monitored changes in the abundances of 11 targeted phyla and classes. Actinobacteria, Bacteriodetes and ␣- and ␥-Proteobacteria were more abundant under all three bioremediation treatments, with Actinobacteria representing the dominant phylum. Altogether, our results indicate that bioremediation of PCB-contaminated soil induces signiﬁcant changes in the structure and abundance of the total microbial community, which must be addressed to implement bioremediation practices in order to restore soil functions. © 2011 Elsevier B.V. All rights reserved.

1. Introduction Polychlorinated biphenyls (PCBs) comprise a group of particularly persistent pollutants. They are considered as one of the most widely distributed class of chlorinated chemicals in the food chains, released into the environment by inappropriate use, improper disposal or accidental leakages [1,2]. Even though their production was banned more than 40 years ago they are still found in different compartments of the environment, causing serious threats to human health. In addition, by polluting the soil ecological system, they can seriously affect the capacity of the soil to perform its primary functions [3,4]. The physicochemical properties of PCBs make their biodegradation a wide-scale challenge. As a consequence, an important number of studies have focused on PCB degradation, and various microbial strains able to transform PCBs have been isolated [1,5–9]. Different studies have aimed at using microorganisms for removing PCBs from contaminated sites, which is considered as a potentially simple, economically and environmentally friendly

∗ Corresponding author at: Rudjer Boskovic Institute, Division for Marine and Environmental Research, Bijenic Street 54, P.O. Box 180, HR-10002 Zagreb, Croatia. Tel.: +3851 4680944; fax: +3851 4680242. ´ E-mail address: [email protected] (I. Petric). 0304-3894/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.jhazmat.2011.08.036

bioremediation strategy [6,10,11]. Up to now, their focus has been set on: (i) the efﬁciency of the bioremediation process by monitoring PCB-disappearance, (ii) the description of active PCB-degrading bacterial populations by using cultivation-based approaches and, more recently, (iii) the monitoring of the PCB-degrading ability of the soil microﬂora by determining the occurrence of bph catabolic genes in the soil [12–18]. Undoubtedly microorganisms play fundamental roles in different soil ecosystemic services such as nutrient cycling, ﬁltering, organic matter decomposition and climate regulation [19–21]. In this context, estimating the impact of bioremediation on the structure, composition and abundance of the soil microbial communities arises as a major issue [22–25]. However, although the effect of PCB-contamination on the soil microbial community has been investigated repeatedly [26–30], the impact of PCB-contaminated soil bioremediation on the microbial community has scarcely been monitored. The former studies, in which short-term, artiﬁcially contaminated soil was used, provide evidence that bioaugmentation of soils with PCB-degraders, biostimulation by the addition of inducers and rhizoremediation can lead to changes in the bacterial community structure [18,31–34]. However, to improve bioremediation practices, it is important to understand how the microbial community responds to bioremediation in less artiﬁcial systems such as long-term PCB-contaminated sites.

I. Petri´c et al. / Journal of Hazardous Materials 195 (2011) 254–260

In order to assess the impact of bioremediation on the soil microbial community, a small-scale bioremediation assay was designed to bioremediate the soil of a transformer station contaminated with PCBs since the 1991 Balkanian war events [35]. A previous study at the site had shown that biostimulation as well as bioaugmentation approaches resulted in the (i) degradation of 40% of the PCBs from the soil within a 1-year period and (ii) modiﬁcation in the structure and the abundance of the functional PCB-degrading community [35]. The objective of this work was to monitor the response of the total bacterial community to three different bioremediation treatments in this PCB-contaminated soil. For this purpose, the structure of the bacterial community was monitored by ribosomal intergenic spacer analysis (RISA) while the abundances of 10 bacterial phyla and classes as well as of the total bacteria and crenarchaea were quantiﬁed by real-time PCR (qPCR) [36–38] after an 18-month bioremediation period. 2. Materials and methods 2.1. Bioremediation assay design The soil used here was composed of 51.4% clay, 30.6% silt, 18.0% sand, 39.0 g kg−1 of organic matter, 22.6 g kg−1 of organic carbon, 2.1 g kg−1 of total nitrogen, with a C/N ratio of 11 and a pH value of 7.2. It was contaminated with up to 32 ␮g of PCB per gram of soil [35]. A large amount of soil (approx. 500 kg) was excavated with a shovel from a 10 m × 4 m plot (down to a 30-cm depth) alongside a transformer station damaged in 1991 during a war event (Zadar, Croatia). The soil was prepared for the assay by repeated sieving on a 4-mm mesh and manual mixing. The soil was placed in three plastic containers (0.84 × 0.41 × 0.16 m, approximately 90 kg per container). Each container was divided into two compartments, and submitted to three different bioremediation treatments: (a) bioaugmentation with a mixed culture (BAM, inoculation with TSZ7 culture along with mineral medium containing xylose (1 g l–1 ) as a supplemental carbon source, carvone (100 mg l–1 ) as an inducer of the PCB-degrading pathway and soya lecithin (5 g l–1 ) as a surfactant to enhance PCB bioavailability), (b) bioaugmentation with a pure strain culture (BAP, inoculation with Rhodococcus sp. Z6 strain along with mineral medium containing xylose, carvone and soya lecithin) and (c) biostimulation (BS, treatment with a mineral medium containing xylose, carvone and soya lecithin). Inoculums were prepared by growing cultures at 30 ◦ C under agitation in a phosphate-buffered mineral salt medium supplemented with biphenyl. The amendments were supplied every two weeks over an 18-month period. A detailed experimental design is described in [35], together with detailed instructions about how to prepare the inoculating cultures. GC–MS analysis revealed that by the end of the 18-month-long bioremediation period approximately 40% of total PCBs, representing mainly tri- and tetra-chlorinated congeners, had been degraded from the contaminated soil with all three bioremediation treatments. The monitoring of physicochemical parameters suggested continuous pH and no extreme changes in the soil moisture throughout the bioremediation treatment. 2.2. Extraction and puriﬁcation of total DNA from the soil At the end of the 18-month-long incubation period soil samples were collected from the three containers (n = 3 per treatment, ntot = 9). A composite soil sample (approximately 1 kg) made of 5 random samplings (0–10-cm depth layer) was collected on the contaminated site, subsampled. and used as control soil (n = 3). Samples were kept at −20 ◦ C until use (ntot = 12). Total DNA was extracted following ISO 11063 [39] and soil DNA extracts were then

255

puriﬁed using polyvinyl polypyrrolidone (PVPP) and Sepharose 4B spin columns (Sigma–Aldrich, USA) and the Geneclean Turbo Kit puriﬁcation kit (Qbiogen, France) according to [35]. The integrity of soil DNA was checked by electrophoresis (1% agarose gel) and its amount was estimated using a BioPhotometer at 260 nm (Eppendorf, Germany). 2.3. Inhibition test The absence of inhibitors in our soil extracts was veriﬁed for all samples by mixing a known amount of the plasmid pGEM-T Easy Vector (Promega, France) with the soil DNA extracts or water before running a qPCR with plasmid-speciﬁc T7 and SP6 primers as previously described [40]. The measured cycle threshold (Ct) values obtained for the different DNA extracts and for the water controls were not signiﬁcantly different, indicating that no inhibition occurred. 2.4. Ribosomal intergenic spacer analysis The global structure of the soil bacterial community was investigated using Ribosomal Intergenic Spacer Analysis (RISA). The 16S-23S intergenic spacer of the bacterial ribosomal operon was ampliﬁed from 25 ng of the DNA template using universal primers 38r and 72f [41]. PCRs were carried out in a PTC-200 gradient cycler (MJ Research, USA) under the following conditions: 5 min at 94 ◦ C; 35 cycles of 1 min at 94 ◦ C, 1 min at 55 ◦ C and 2 min at 72 ◦ C, followed by a 15 min cycle at 72 ◦ C. The resulting amplicons were quantiﬁed on native agarose gels. For each sample approximately 100 ng of amplicon were loaded on a 6% acrylamide gel (16 h, 8 mA). The gels were then stained with SYBR green II (Molecular Probes, Netherlands) and scanned with a Storm 960 Phosphor Imager (Molecular Dynamisc, Sunnyvale, CA, USA). 2.5. Real-time PCR quantiﬁcation (qPCR) The abundances of the total bacteria and crenarchaea were quantiﬁed by real time PCR (qPCR) as previously described. Taxaspeciﬁc 16S rRNA primers were used for quantiﬁcation of the Actinobacteria, Acidobacteria, ␣-Proteobacteria, ␤-Proteobacteria, ␥-Proteobacteria, Bacteriodetes, Firmicutes, Gemmatimonadetes, Verrucomicrobia, Planctomycetes and Archaea – Crenarchaeota by real-time PCR [36,37,42–44]. qPCR assays were conducted on an ABI 7900 HT Real-time PCR System (Applied Biosystems, USA) in 15 ␮l ﬁnal volume containing SYBR green PCR Master Mix (Absolute QPCR SYBR Green Rox Abgene, France), 250 ng of T4 gp32 (Qbiogene, France), 1 ␮M of each primer and 2 ng of template DNA. For each 16S rRNA target, a standard curve was established using serial dilutions of linearized plasmid pGEM-T (102 to 107 copies) containing cloned 16S rRNA. No-template controls (NTC, n = 2) were also included in all the assays. Melting curves were generated after ampliﬁcation in order to check the speciﬁcity of the assays. 2.6. Statistical analysis The signiﬁcance of differences between the data obtained with the three different bioremediation treatments was tested using XLStat 2009 (Addinsoft, Brooklyn, USA). Based on the normality test showing that our data were not following a normal distribution, we chose to analyse our results with the non-parametric Kruskal–Wallis test (p < 0.05). According to the test requirements, all data points were independent from each other and sample sizes were equal with three data points analysed for each treatment. RISA ﬁngerprints were analysed using PrepRISA [45] and ADE-4 package [46]. Using PrepRISA, the data from the 1D-Scan (ScienceTec, France) were converted into a matrix summarizing the

256

I. Petri´c et al. / Journal of Hazardous Materials 195 (2011) 254–260

Fig. 1. RISA ﬁngerprint (6%, acrylamid stained with SYBR green II) obtained from DNA extracted directly from soils under bioremediation treatments (BS, BAP and BAM) and control soil (n = 3 for each soil, ntot = 12). The size of molecular marker BVIII 19–1114 bp (Roche, USA) is shown on the right side of the gel (A). Principal component analysis performed from RISA ﬁngerprints obtained from DNA extracted from soils under bioremediation treatments (BS, BAP and BAM) and control soil (n = 3 for each soil, ntot = 12) (B). BAM: bioaugmentation with mixed culture TSZ7, xylose, carvone and soya lecithin; BAP: bioaugmentation with Rhodococcus sp. Z6, xylose, carvone and soya lecithin; BS: biostimulation with xylose, carvone and soya lecithin.

bands’ presence (i.e. peaks) and intensity (i.e. peak heights). Then, using ADE-4, principal component analysis (PCA) on the covariance matrix was performed. This method provided an ordination of bacterial communities and of the encoded bands, which were plotted in two dimensions based on the scores in the ﬁrst principal components. Three data points were analysed for each treatment (ntot = 12, per marker tested). 3. Results 3.1. Global structure of microbial communities in soils under different bioremediation treatments RISA, revealing the length polymorphism of the intergenic spacer of the bacterial 16S rRNA operon, was used to estimate the impact of bioremediation treatments (BAM, BAP, and BS) on the global structure of the soil microbial community. Visual examination revealed that RISA ﬁngerprints, which revealed up to 20 major bands, were well replicated for each treatment, revealing that DNA extraction and PCR ampliﬁcation from the different treatments were efﬁcient and reproducible. Interestingly, the structure of the communities from BAM, BAP, BS and control soils differed as to their numbers (19, 18, 16 and 13 in BAM, BAP, BS and con-

trol, respectively) and as to the relative intensities of the detected bands (Fig. 1A). For comparison analysis, RISA ﬁngerprints were digitized and further analysed by pairwise comparison using Principal Component Analysis (PCA) in order to ordinate the microbial communities on the plane deﬁned by the ﬁrst two principal components, in accordance to the bioremediation treatment applied (Fig. 1B). The ﬁrst principal component (PC1) represented 59.4% of the variances in the data while the second principal component (PC2) represented 21.6%. Analysis of the factorial map revealed that ordination along PC1 allowed us to distinguish control from BS/BAM /BAP treatments. We could not differentiate between bioremediation treatments on PC1 but ordination on PC2 showed a clear separation of these three bioremediation treatments. 3.2. Abundance of the total bacterial community and 11 targeted phyla in our soils The qPCR values were expressed as copy numbers per ng of soil DNA to minimize the possible bias related to the DNA extraction yield. All targeted 16S rRNA genes were successfully ampliﬁed from BAM, BAP, BS and from control soil with PCR efﬁciencies ranging between 81% and 100% except for the crenarchaea for which a lower efﬁciency of 64% was observed (Table 1).

Table 1 Primer pairs used for the qPCR assays to estimate the abundances of phyla and class-speciﬁc bacteria. Target group

Primers

Amplicon size (bp)

Annealing T (◦ C)

All groups Acidobacteria Actinobacteria ␣-Proteobacteria ␤-Proteobacteria ␥-Proteobacteria Bacteroidetes Firmicutes Gemmatimonadetes Verrucomicrobia Planctomycetes Crenarchaea

341F/534R Acid31/Eub518 Actino235/Eub518 Eub338/Alf685 Eub338/Bet680 Gamma395f/Gamma 871r Cbf319/Eub518 Lgc353/Eub518 Gem440/534R Ver349/Eub518 Plancto352f/Plancto920r 771F/957R

194 500 300 342 360 497 210 181 461 186 565 228

60 55 60 60 55 56 60 55 55 60 60 55

PCR efﬁciency (%) 93 94 90 100 96 81 95 95 96 87 81 64

Reference [43] [36] [36] [36] [36] [42] [36] [36] [37] [37] [42] [44]

I. Petri´c et al. / Journal of Hazardous Materials 195 (2011) 254–260

16S rRNA (copy number ng -1 DNA)

A 10 6

BAM

BAP

BS

257

control

b b b a

c b bc a

b b b a

ab b b a

ab b ab a

b b b a

b b ab a

b ab ab a

10 5 a a a b

c bc a ab

a a a a

ab b a ab

10 4

10 3

B

Actinobacteria 30 b

Firmicutes

Bacteroidetes 20

ab b a ab a

10

β-Proteobacteria

γ-Proteobacteria ab

b

a ab a a ab a ab a a a ab ab b

ab b

0

b

a

Acidobacteria

b

b

α-Proteobacteria

BAM BAP

Planctomycetes

Gemmatimonadetes

BS control

Verrucomicrobia

Fig. 2. 16S rRNA sequences copy number of total and targeted bacterial taxonomic groups (Actinobacteria, Acidobacteria, ˛-Proteobacteria, ˇ-Proteobacteria, -Proteobacteria, Bacteriodetes, Firmicutes, Gemmatimonadetes, Verrucomicrobia, and Planctomycetes) and Archaea (Crenarchaeota) determined by qPCR in soils under different bioremediation treatments (BAM, BAP, and BS) and in control soil (A). Relative abundances of targeted bacterial taxonomic groups within the total bacterial community determined by qPCR in soils under different bioremediation treatments (BAM, BAP, and BS) and in control soil, presented in percentages (%) (B). BAM = mixed culture TSZ7 + xylose + carvone + soya lecithin; BAP = Rhodococcus sp. Z6 culture + xylose + carvone + soya lecithin; BS = xylose + carvone + soya lecithin. Error bars represent standard deviation of a mean value (n = 3 for each soil, ntot = 12). Letters (a, b, c) assigned to each value represent groups appointed by the Kruskal–Wallis statistical analysis (p < 0.05). Values in the same group are not signiﬁcantly different from each other.

No-template negative controls yielded negligible values in all qPCR assays. The copy numbers of the total 16S rRNA genes ranged from 8.5 × 104 to 4.3 × 105 genes per ng of DNA extracted from the different soils, with a signiﬁcantly lower abundance of the total bacteria in the control soil as compared to the bioremediated soils (Fig. 2A). The abundances of the different bacterial taxa ranged from 3.4 × 103 to 8.5 × 104 16S rRNA gene copies per ng DNA. Lower abundances were observed in the control as compared to BAM, BAP and BS treatments for most taxa. However differences were signiﬁcant only for the Actinobacteria, Bacteriodetes and ˛-Proteobacteria. On the contrary, a signiﬁcantly higher abundance of Acidobacteria was found in the control soil compared to BAM, BAP and BS treatments while the abundance of the Firmicutes did not signiﬁcantly differ between treatments (Fig. 2A). For further insights into the composition of the bacterial community, we calculated the relative abundances of the different phyla and classes within the total bacterial community (Fig. 2B). We found that the ten targeted groups represented 79–90% of the abundance of the total microbial community, depending on the treatment. Actinobacteria and Bacteroides were the dominant phyla in bioremediated soils with relative abundances of 23% and 17% while they represented 17% and 13% of the community in the con-

trol soil, respectively. ␥- and ␣-Proteobacteria were less abundant with an average 12% (bioremediated soils) and 7% (control soil) of the bacterial community. Other groups were less abundant and represented less than 5% of the total community. Interestingly, the abundance of those groups was higher in the control than in the bioremediated soil. This trend was particularly clear for Acidobacteria which represented less than 1% in bioremediated soils while it averaged 12% in the control soil. 4. Discussion In the year 1991, a PCB-ﬁlled transformer station situated in Zadar (Croatia) was damaged during a military attack, and as a consequence the surrounding area was contaminated with substantial amounts of PCBs [47]. In order to restore this contaminated site a series of experiments were conducted. Among these, a small-scale assay was designed to test the efﬁciency of different bioremediation strategies on PCB degradation from the soil. A TSZ7 mixed bacterial culture and the Rhodococcus sp. Z6 strain, enriched on biphenyl from the contaminated soil, were used as seed cultures [48]. Our results suggest that all three bioremediation treatments were efﬁcient, as they led to the degradation of up to 40% of PCBs from our contaminated soil over an 18-month period. To get a better

258

I. Petri´c et al. / Journal of Hazardous Materials 195 (2011) 254–260

insight into the biological processes that accompany PCB degradation, we decided to study the structure, the composition and the abundance of the bacterial community in response to our bioremediation strategies of the site with molecular tools based on direct soil DNA extraction. An insight into the structure of the bacterial community was achieved by RISA. The analysis revealed that after an 18-monthlong treatment the bacterial communities in the BAM, BAP and BS bioremediation treatments were signiﬁcantly different from that of the control soil. This indicated that bioremediation resulted in a shift in the structure of the total bacterial community, suggesting the development of communities in the bioremediated soils that differed from the original one from the contaminated site. Interestingly, we noticed that the PCB-contaminated site was characterized by a bacterial community made of fewer phylotypes (13 phylotypes, compressed between 250 and 340 bp), which is a typical feature of stress environments [49] and could result from PCB toxicity to living cells [50]. Conversely, bioremediated soils were characterized by an increase in the number of phylotypes (16, 18 and 19 phylotypes in BS, BAM and BAP, respectively, with sizes up to 880 bp), suggesting modiﬁcations in the soil bacterial community composition, with several dominating phylotypes. All bioremediation treatments were biostimulated by carvone to induce the synthesis of the PCB-degrading enzyme system, and by soya lecithin as a surfactant to promote PCB bioavailability. Besides their expected effect, the amendments may also have been used as nutrient sources allowing the growth of opportunistic microbial populations [51]. Moreover, they may have had negative effects on some bacterial populations, such as suggested by some authors [52]. Consequently, they may lead to changes in the structure of the global microbial communities. As shown previously, the availability of carbon substrates in the highly oligotrophic soil environment can greatly modify the structure of microbial communities by favouring the growth of r-strategist microbial groups [33,53]. This could explain the shift in the structure of the bacterial community observed between control and bioremediated soils along the ﬁrst principal component (ﬁrst axis). However, the differentiation between biostimulated and bioaugmented soils on axis two of the PCA suggested that inoculation with PCB degraders also has an effect on the structure of the bacterial community. Even though the inoculated Rhodococcus strain can survive in the soil and be detected throughout the 18-month-long assay [35], it cannot fully explain the differences observed on the RISA proﬁles due to its low abundance (i.e. less than 1% of the total bacterial community). Nevertheless, it should be emphasized that the differences among all three bioremediated soils were small compared to those observed between control and bioremediated soils. Altogether, these results show that bioremediation treatments affect the structure of the soil bacterial community. Interestingly, we observed that even if the different treatments led to the establishment of bacterial communities differing in their structure, their overall performance was similar with approximately 40% depletion of PCBs [35]. This might be explained by the fact that RISA only gave an insight into the abundant populations within the total bacterial community but not into the functional community responsible for PCB degradation. Indeed we previously showed that the PCB-degrading potential quantiﬁed by targeting bphA genes was increased in a similar way by the three bioremediation treatments [35]. Therefore the efﬁciency of the bioremediation treatment most likely resulted from the increase in the abundance of this functional community in combination with the increase in PCB bioavailability. RISA was used to get a ﬁrst insight into the impact of bioremediation on the structure of the total soil bacterial community. Although this approach was shown to be of interest to observe shifts in the bacterial community structure in response to different

stresses, it does not allow us to monitor the changes in the taxonomic composition of the community. Therefore, the composition of the microbial communities was also studied by quantifying the relative abundances of the ten bacterial (Actinobacteria, Acidobacteria, ˛-Proteobacteria, ˇ-Proteobacteria, -Proteobacteria, Bacteriodetes, Firmicutes, Gemmatimonadetes, Verrucomicrobia, and Planctomycetes) and one Archaea (Crenarchaeota) taxa. According to the meta-analysis of 32 16S rRNA gene libraries from a variety of soils [54] these groups represent up to 90% of the soil bacterial community. Our qPCR results showed that the bacterial communities responded similarly to biostimulation and bioaugmentation approaches, which is in accordance with our RISA ﬁndings. We observed that Actinobacteria, Bacteriodetes and Acidobacteria dominated in the control transformer station soil in which elevated concentrations of PCBs up to 32 ␮g g−1 soil were found, while Firmicutes and Planctomycetes were lowly represented. Even though abundances of different taxa were shown to vary between different soils and are closely related to soil characteristics, these results are in an agreement with recent studies showing that these taxa were also dominant in other soils [38,54–56]. Interestingly, while Gemmatimonadetes are usually considered one of the less numerous taxa [54,56], they were quite abundant in our control soil in which they represented approximately 10% of the community. This suggests that this group has been favoured by exposure to PCB at the contaminated site but the effect of other environmental conditions at the PCB-contaminated site cannot be ruled out. The decline in the soil PCB content observed at the end of the 18-month small-scale bioremediation assay suggests that bioremediation could have induced the shift observed in the community structure. This signiﬁcant degradation of PCB can therefore be correlated with the selection of speciﬁc bacterial groups in soil under bioremediation. For example, the abundance of Actinobacteria and Bacteroidetes as well as ␣- and ␥-Proteobacteria increased up to threefold as a result of the bioremediation treatment compared to the control soil. Even though changes in the microbial community structure during PCB degradation have seldom been monitored, a few studies record higher abundances of Proteobacteria, Grampositive bacteria, and Actinobacteria along with PCB degradation [26,31,32]. This can be explained by the fact that these groups contain well-known PCB degraders (Rhodococcus, Arthrobacter, Corynebacterium, Sphingomonas, Pseudomonas, Acinetobacter, etc.). However, Bacteroidetes have not been known to respond to PCBs so far. Different studies correlate the predominance of Bacteroidetes with agricultural practices such as C amendments [54,55,57]. It can be hypothesized that the selection of these phyla might be due to the alteration of soil characteristics as a consequence of bioremediation treatments. Interestingly, our bioremediation treatments led to a decrease in the proportion of most of the targeted taxa, with the highest decrease observed for Acidobacteria (tenfold), which represented less than 1% of the overall community at the end of bioremediation. The microbial communities developed under our three different bioremediation treatments had similar structures when compared at higher taxonomical ranks. This could further be correlated with the similar PCB-degrading activity observed in all three bioremediated soils. Overall we found that the bioremediation treatments in our small-scale assay resulted in signiﬁcant shifts in the soil microbial community at high taxonomical ranks, which might reﬂect the ecological coherence of the targeted phyla and classes [58]. Even though it is difﬁcult to determine the functional role of the targeted taxa, the predominance of the Actinobacteria phylum in all three bioremediated soils (>20% of the total community) suggests that it might be of importance in the PCB degradation process. Indeed, the ability of bacterial populations belonging to the Rhodococcus genus to degrade xenobiotics is well documented, along with their ability to persist in soils even in starvation

I. Petri´c et al. / Journal of Hazardous Materials 195 (2011) 254–260

conditions [59,60]. This hypothesis is further supported by the fact that Rhodococcus-like bacterial populations, belonging to the Actinobacteria phylum and harbouring bphA and bphC catabolic genes for PCB-degradation, were promoted in response to bioremediation treatments at the same site [35]. The addition of the particular inducer of PCB-catabolic enzymes [61], along with the presence of hydrophobic cell walls and production of surfactants may have led to the development of this Rhodococcus bacterial population which appears to be more competitive for PCBs. As reported by [33] the choice of particular amendments combined with habitat characteristics has a strong enrichment effect on certain bacterial populations, thereby affecting bacterial community composition along with the pattern of PCB degradation. 5. Conclusions Our small-scale bioremediation assay revealed that bioremediation can signiﬁcantly modify the structure of the total bacterial community as well as the abundance of the targeted bacterial taxa. Indeed, the microbial community at the contaminated transformer station site was signiﬁcantly different from the ones present in the three bioremediation treatments. However, despite differences in the bioremediation strategy that was used (i.e. biostimulation or bioaugmentation), similar responses of the microbial community were observed. Bioremediation was also shown to stimulate the abundances of Actinobacteria, Bacteroidetes and ␣- and ␥Proteobacteria phyla but had a negative impact on the abundance of Acidobacteria. The predominance of the Actinobacteria phylum in bioremediated soils suggested their relevance in the PCB degradation process. Since microorganisms are key players in crucial soil functions, further research is required to assess whether the changes we observed in microbial community structures in response to bioremediation also altered the functioning of the bioremediated soil. Acknowledgements This work was ﬁnancially supported by the Croatian Ministry of Science, Education and Sports, by the contract ICA2-CT200210007 (APOPSBAL) within European Commission – The Fifth Framework Programme. I Petric was funded by ADEME and Conseil Régional de Bourgogne under the supervision of Welience Agro-Environnement. We are grateful to N. Rouard and J. Beguet (Laboratoire de Microbiologie du Sol et de l’Environnement, Dijon, France) for their help in data analysis. References [1] J. Borja, D.M. Taleon, J. Auresenia, S. Gallardo, Polychlorinated biphenyls and their biodegradation, Process Biochem. 40 (2005) 1999–2013. [2] K. Hornbuckle, L. Robertson, Polychlorinated biphenyls (PCBs): sources, exposures, toxicities, Environ. Sci. Technol. 44 (2010) 2749–2751. [3] P. Wilmes, P.L. Bond, Microbial community proteomics: elucidating the catalysts and metabolic mechanisms that drive the Earth’s biogeochemical cycles, Curr. Opin. Microbiol. 12 (2009) 310–317. [4] X.M. Xiao, D. Niyogi, D. Ojima, Changes in land use and water use and their consequences on climate, including biogeochemical cycles, Global Planet. Change 67Sp. Iss (2009). [5] W.R. Abraham, B. Nogales, P.N. Golyshin, D.H. Pieper, K.N. Timmis, Polychlorinated biphenyl-degrading microbial communities in soils and sediments, Curr. Opin. Microbiol. 5 (2002) 246–253. [6] J.A. Field, R. Sierra-Alvarez, Microbial transformation and degradation of polychlorinated biphenyls, Environ. Pollut. 155 (2008) 1–12. [7] K. Furukawa, H. Fujihara, Microbial degradation of polychlorinated biphenyls: biochemical and molecular features, J Biosci. Bioeng. 105 (2008) 433–449. [8] T. Ohmori, H. Morita, M. Tanaka, K. Miyauchi, D. Kasai, K. Furukawa, K. Miyashita, N. Ogawa, E. Masai, M. Fukuda, Development of a strain for efﬁcient degradation of polychlorinated biphenyls by patchwork assembly of degradation pathways, J. Biosci. Bioeng. 111 (2011) 437–442. [9] D.H. Pieper, Aerobic degradation of polychlorinated biphenyls, Appl. Microbiol. Biotechnol. 67 (2005) 170–191.

259

[10] T. Iwamoto, M. Nasu, Current bioremediation practice and perspective, J. Biosci. Bioeng. 92 (2001) 1–8. [11] Y. Ohtsubo, T. Kudo, M. Tsuda, Y. Nagata, Strategies for bioremediation of polychlorinated biphenyls, Appl. Microbiol. Biotechnol. 65 (2004) 250–258. [12] S. Capodicasa, S. Fedi, M. Carnevali, L. Caporali, C. Viti, F. Fava, D. Zannoni, Terminal-restriction fragment length polymorphism analysis of biphenyl dioxygenase genes from a polychlorinated biphenyl-polluted soil, Res. Microbiol. 160 (2009) 742–750. [13] R.W. Erb, I. Wagnerdobler, Detection of polychlorinated biphenyl degradation genes in polluted sediments by direct DNA extraction and polymerase chainreaction, Appl. Environ. Microbiol. 59 (1993) 4065–4073. [14] E.S. Gilbert, D.E. Crowley, Repeated application of carvone-induced bacteria to enhance biodegradation of polychlorinated biphenyls in soil, Appl. Microbiol. Biotechnol. 50 (1998) 489–494. [15] M.J. Hoostal, G.S. Bullerjahn, R.M.L. McKay, Molecular assessment of the potential for in situ bioremediation of PCBs from aquatic sediments, Hydrobiologia 469 (2002) 59–65. [16] A.C. Layton, C.A. Lajoie, J.P. Easter, R. Jernigan, J. Sanseverino, G.S. Sayler, Molecular diagnostics and chemical-analysis for assessing biodegradation of polychlorinated-biphenyls in contaminated soils, J. Ind. Microbiol. 13 (1994) 392–401. [17] A.C. Singer, E.S. Gilbert, E. Luepromchai, D.E. Crowley, Bioremediation of polychlorinated biphenyl-contaminated soil using carvone and surfactant-grown bacteria, Appl. Microbiol. Biotechnol. 54 (2000) 838–843. [18] A.C. Singer, D. Smith, W.A. Jury, K. Hathuc, D.E. Crowley, Impact of the plant rhizosphere and augmentation on remediation of polychlorinated biphenyl contaminated soil, Environ. Toxicol. Chem. 22 (2003) 1998–2004. [19] J.W. Doran, M. Saﬂey, Deﬁning and assessing soil health and sustainable productivity, in: C. Pankhurst, B.M. Doube, V.V.S.R. Gupta (Eds.), Biological Indicators of Soil Health, CAB International, Wallingford, Oxon, UK, 1997, pp. 1–28. [20] P. Nannipieri, J. Ascher, M.T. Ceccherini, L. Landi, G. Pietramellara, G. Renella, Microbial diversity and soil functions, EUR J. Soil Sci. 54 (2003) 655–670. [21] V. Torsvik, L. Ovreas, Microbial diversity and function in soil: from genes to ecosystems, Curr. Opin. Microbiol. 5 (2002) 240–245. [22] L. Ciric, R.I. Grifﬁths, J.C. Philp, A.S. Whiteley, Field scale molecular analysis for the monitoring of bacterial community structures during on-site diesel bioremediation, Bioresour. Technol. 101 (2010) 5235–5241. [23] D.K. Mills, K. Fitzgerald, C.D. Litchﬁeld, P.M. Gillevet, A comparison of DNA proﬁling techniques for monitoring nutrient impact on microbial community composition during bioremediation of petroleum-contaminated soils, J. Microbiol. Meth. 54 (2003) 57–74. [24] D. Paul, G. Pandey, C. Meier, J.R. van der Meer, R.K. Jain, Bacterial community structure of a pesticide-contaminated site and assessment of changes induced in community structure during bioremediation, FEMS Microbiol. Ecol. 57 (2006) 116–127. [25] A.S. Whiteley, M.J. Bailey, Bacterial community structure and physiological state within an industrial phenol bioremediation system, Appl. Environ. Microbiol. 66 (2000) 2400–2407. [26] P.A. Correa, L. Lin, C.L. Just, D. Hu, K.C. Hornbuckle, The effects of individual PCB congeners on the soil bacterial community structure and the abundance of biphenyl dioxygenase genes, Environ. Int. (2009), doi:10.1016/j.envint.2009.07.015. [27] M.B. Leigh, V.H. Pellizari, O. Uhlik, R. Sutka, J. Rodrigues, N.E. Ostrom, J.H. Zhou, J.M. Tiedje, Biphenyl-utilizing bacteria and their functional genes in a pine root zone contaminated with polychlorinated biphenyls (PCBs), ISME J. 1 (2007) 134–148. [28] G. Lloyd-Jones, P.C.K. Lau, A molecular view of microbial diversity in a dynamic landﬁll in Quebec, FEMS Microbiol. Lett. 162 (1998) 219–226. [29] B. Nogales, E.R.B. Moore, E. Llobet-Brossa, R. Rossello-Mora, R. Amann, K.N. Timmis, Combined use of 16S ribosomal DNA and 16S rRNA to study the bacterial community of polychlorinated biphenyl-polluted soil, Appl. Environ. Microbiol. 67 (2001) 1874–1884. [30] G. Zanaroli, A. Balloi, A. Negroni, D. Daffonchio, L.Y. Young, F. Fava, Characterization of the microbial community from the marine sediment of the Venice lagoon capable of reductive dechlorination of coplanar polychlorinated biphenyls (PCBs), J. Hazard. Mater. 178 (2010) 417–426. [31] D.A. de Carcer, M. Martin, U. Karlson, R. Rivilla, Changes in bacterial populations and in biphenyl dioxygenase gene diversity in a polychlorinated biphenylpolluted soil after introduction of willow trees for rhizoremediation, Appl. Environ. Microbiol. 73 (2007) 6224–6232. [32] N. Ding, H.C. Guo, T. Hayat, Y.P. Wu, J.M. Xu, Microbial community structure changes during Aroclor 1242 degradation in the rhizosphere of ryegrass (Lolium multiﬁorum L.), FEMS Microbiol. Ecol. 70 (2009) 305–314. [33] W. Luo, E.M. D’Angelo, M.S. Coyne, Organic carbon effects on aerobic polychlorinated biphenyl removal and bacterial community composition in soils and sediments, Chemosphere 70 (2008) 364–373. [34] C. Tu, Y. Teng, Y.M. Luo, X.H. Sun, S.P. Deng, Z.G. Li, W.X. Liu, Z.H. Xu, PCB removal, soil enzyme activities, and microbial community structures during the phytoremediation by alfalfa in ﬁeld soils, J. Soils Sediment 11 (2011) 649– 656. ´ D. Hrˇsak, S. Fingler, N. Udikovic-Koli ´ ´ D. Bru, F. Martin-Laurent, Insight c, [35] I. Petric, in the PCB-degrading functional community in long-term contaminated soil under bioremediation, J. Soils Sediment 11 (2011) 290–300. [36] N. Fierer, J.A. Jackson, R. Vilgalys, R.B. Jackson, Assessment of soil microbial community structure by use of taxon-speciﬁc quantitative PCR assays, Appl. Environ. Microbiol. 71 (2005) 4117–4120.

260

I. Petri´c et al. / Journal of Hazardous Materials 195 (2011) 254–260

[37] L. Philippot, D. Bru, N.P.A. Saby, J. Cuhel, D. Arrouays, M. Simek, S. Hallin, Spatial patterns of bacterial taxa in nature reﬂect ecological traits of deep branches of the 16S rRNA bacterial tree, Environ. Microbiol. 11 (2009) 3096–3104. [38] E. Wessén, S. Hallin, L. Philippot, Differential responses of bacterial and archaeal groups at high taxonomical ranks to soil management, Soil Biol. Biochem. 42 (2010) 1759–1765. [39] I. Petric, L. Philippot, C. Abbate, A. Bispo, T. Chesnot, S. Hallin, K. Laval, T. Lebeau, P. Lemanceau, C. Leyval, K. Lindström, P. Pandard, E. Romero, A. Sarr, M. Schloter, P. Simonet, K. Smalla, B.-M. Wilke, F. Martin-Laurent, Inter-laboratory evaluation of the ISO standard 11063 “Soil quality—method to directly extract DNA from soil samples”, J. Microbiol. Meth. 84 (2011) 454–460. [40] S. Henry, D. Bru, B. Stres, S. Hallet, L. Philippot, Quantitative detection of the nosZ gene, encoding nitrous oxide reductase, and comparison of the abundances of 16S rRNA, narG, nirK, and nosZ genes in soils, Appl. Environ. Microbiol. 72 (2006) 5181–5189. [41] V. Gurtler, V.A. Stanisich, New approaches to typing and identiﬁcation of bacteria using the 16S-23S rDNA spacer region, Microbiol. -Uk 142 (1996) 3–16. [42] M. Mühling, J. Woolven-Allen, J.C. Murrell, I. Joint, Improved group-speciﬁc PCR primers for denaturing gradient gel electrophoresis analysis of the genetic diversity of complex microbial communities, ISME J. 2 (2008) 379–392. [43] G. Muyzer, E.C. Dewaal, A.G. Uitterlinden, Proﬁling of complex microbialpopulations by denaturing gradient gel-electrophoresis analysis of polymerase chain reaction-ampliﬁed genes-coding for 16S ribosomal-RNA, Appl. Environ. Microbiol. 59 (1993) 695–700. [44] T. Ochsenreiter, D. Selezi, A. Quaiser, L. Bonch-Osmolovskaya, C. Schleper, Diversity and abundance of Crenarchaeota in terrestrial habitats studied by 16S RNA surveys and real time PCR, Environ. Microbiol. 5 (2003) 787– 797. [45] L. Ranjard, F. Poly, J.C. Lata, C. Mougel, J. Thioulouse, S. Nazaret, Characterization of bacterial and fungal soil communities by automated ribosomal intergenic spacer analysis ﬁngerprints: Biological and methodological variability, Appl. Environ. Microbiol. 67 (2001) 4479–4487. [46] S. Dray, A.B. Dufour, The ade4 package: implementing the duality diagram for ecologists, J. Stat. Softw. 22 (2007) 1–20. [47] M. Picer, N. Picer, I. Holoubek, J. Klanova, V.H. Kobasic, Chlorinated hydrocarbons in the atmosphere and surface soil in the areas of the city of Zadar and Mt. Velebit Croatia, Fresen. Environ. Bull. 13 (2004) 712–718.

[48] I. Petric, D. Hrsak, S. Fingler, E. Voncina, H. Cetkovic, A.B. Kolar, N. UdikovicKolic, Enrichment and characterization of PCB-degrading bacteria as potential seed cultures for bioremediation of contaminated soil, Food Technol. Biotech. 45 (2007) 11–20. [49] V. Torsvik, R. Sorheim, J. Goksoyr, Total bacterial diversity in soil and sediment communities—a review, J. Ind. Microbiol. 17 (1996) 170–178. [50] B. Camara, C. Herrera, M. Gonzalez, E. Couve, B. Hofer, M. Seeger, From PCBs to highly toxic metabolites by the biphenyl pathway, Environ. Microbiol. 6 (2004) 842–850. [51] G. Misra, S.G. Pavlostathis, E.M. Perdue, R. Araujo, Aerobic biodegradation of selected monoterpenes, Appl. Microbiol. Biotechnol. 45 (1996) 831–838. [52] L. Paavolainen, V. Kitunen, A. Smolander, Inhibition of nitriﬁcation in forest soil by monoterpenes, Plant Soil 205 (1998) 147–154. [53] P. Marschner, C.H. Yang, R. Lieberei, D.E. Crowley, Soil and plant speciﬁc effects on bacterial community composition in the rhizosphere, Soil Biol. Biochem. 33 (2001) 1437–1445. [54] P.H. Janssen, Identifying the dominant soil bacterial taxa in libraries of 16S rRNA and 16S rRNA genes, Appl. Environ. Microbiol. 72 (2006) 1719–1728. [55] V. Acosta-Martinez, S. Dowd, Y. Sun, V. Allen, Tag-encoded pyrosequencing analysis of bacterial diversity in a single soil type as affected by management and land use, Soil Biol. Biochem. 40 (2008) 2762–2770. [56] C.L. Lauber, M. Hamady, R. Knight, N. Fierer, Pyrosequencing-based assessment of soil pH as a predictor of soil bacterial community structure at the continental scale, Appl. Environ. Microbiol. 75 (2009) 5111–5120. [57] K. Jangid, M.A. Williams, A.J. Franzluebbers, J.S. Sanderlin, J.H. Reeves, M.B. Jenkins, D.M. Endale, D.C. Coleman, W.B. Whitman, Relative impacts of landuse, management intensity and fertilization upon soil microbial community structure in agricultural systems, Soil Biol. Biochem. 40 (2008) 2843–2853. [58] L. Philippot, S.G.E. Andersson, T.J. Battin, J.I. Prosser, J.P. Schimel, W.B. Whitman, S. Hallin, The ecological coherence of high bacterial taxonomic ranks, Nat. Rev. Microbiol. 8 (2010) 523–529. [59] K.S. Bell, J.C. Philp, D.W.J. Aw, N. Christoﬁ, A review—the genus Rhodococcus, J. Appl. Microbiol. 85 (1998) 195–210. [60] A.M. Warhurst, C.A. Fewson, Biotransformations catalyzed by the genus Rhodococcus, Crit. Rev. Biotechnol. 14 (1994) 29–73. [61] A.C. Singer, D.E. Crowley, I.P. Thompson, Secondary plant metabolites in phytoremediation and biotransformation, Trends Biotechnol. 21 (2003) 123–130.

Journal of Hazardous Materials 195 (2011) 261–275

Contents lists available at SciVerse ScienceDirect

Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat

Synthesis of a novel silica-supported dithiocarbamate adsorbent and its properties for the removal of heavy metal ions Lan Bai a , Huiping Hu a,∗ , Weng Fu a , Jia Wan a , Xiliang Cheng b , Lei Zhuge b , Lei Xiong b , Qiyuan Chen a a b

School of Chemistry and Chemical Engineering, Central South University, Changsha 410083, China School of Metallurgical Science and Engineering, Central South University, Changsha 410083, China

a r t i c l e

i n f o

Article history: Received 8 December 2010 Received in revised form 12 August 2011 Accepted 12 August 2011 Available online 22 August 2011 Keywords: Heavy metal Silica-supported dithiocarbamate Adsorption Thermodynamics

a b s t r a c t Silica-supported dithiocarbamate adsorbent (Si-DTC) was synthesized by anchoring the chelating agent of macromolecular dithiocarbamate (MDTC) to the chloro-functionalized silica matrix (SiCl), as a new adsorbent for adsorption of Pb(II), Cd(II), Cu(II) and Hg(II) from aqueous solution. The surface characterization was performed by FT-IR, XPS, SEM and elemental analysis indicating that the modiﬁcation of the silica surface was successfully performed. The effects of media pH, adsorption time, initial metal ion concentration and adsorption temperature on adsorption capacity of the adsorbent had been investigated. Experimental data were exploited for kinetic and thermodynamic evaluations related to the adsorption processes. The characteristics of the adsorption process were evaluated by using the Langmuir, Freundlich and Dubinin–Radushkevich (D–R) adsorption isotherms and adsorption capacities were found to be 0.34 mmol g−1 , 0.36 mmol g−1 , 0.32 mmol g−1 and 0.40 mmol g−1 for Pb(II), Cd(II), Cu(II) and Hg(II), respectively. The adsorption mechanism of Hg(II) onto Si-DTC is quite different from that of Pb(II), Cd(II) or Cu(II) onto Si-DTC, which is demonstrated by the XPS and FT-IR results. © 2011 Elsevier B.V. All rights reserved.

1. Introduction Heavy metals are persistent pollutants because of the toxic, non-biodegradable, accumulative and mobile characters. Due to the rapid increase of the industrial and mining activities which are responsible for the growing metal content in surface and ground water, heavy metal pollution of water bodies has become a serious problem in the world [1]. Many physicochemical methods have been proposed for toxic heavy metal removal from aqueous solution, including chemical precipitation, reverse osmosis, electrochemical reduction, ion exchange and adsorption [2–5]. Among these methods, adsorption is generally preferred for the removal of heavy metal ions due to its high efﬁciency, easy handling, availability of different adsorbents and cost effectiveness. As one of the adsorbent materials, chelating resin is reusable, easy to be separated, and often has high adsorption capacities [6]. Thus, much attention has been drawn to the synthesis of chelating resins and the investigation of their adsorption behaviors of metal ions from various matrices [7–9]. The dithiocarbamates, due to its strong binding ability to heavy metals while not complexing alkali and alkaline earth metals [10], have been widely used in the environ-

∗ Corresponding author. Tel.: +86 731 88877364; fax: +86 731 88879616. E-mail address: [email protected] (H. Hu). 0304-3894/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.jhazmat.2011.08.038

ment treatment and have got a good effect in the removal and preconcentration of heavy metals [11–13]. Silica-based organic/inorganic materials have been well studied with attractive characteristics of high chemical reactivity and good mechanical stability [14]. Due to the silanol groups (Si-OH) distributed on the silica gel surface, the silica-supported resins have favorable hydrophilicity, thus leading to higher adsorption efﬁciency and faster ion diffusion rate than the other resins supported by polystyrene, polyamide, polyvinyl chloride, etc. [15,16]. Most of the reported works about the synthesis of silicasupported dithiocarbamate resins have concerned to add carbon disulﬁde to modify the surface of silica gel phases containing amine moieties. However, the reaction between carbon disulﬁde and the amines modiﬁed on the silica gel surface is generally conducted under harsh conditions, using of strong alkaline conditions, which results in the degradation of silica matrix [17]. In addition, the great majority of the active amines are polyamine compounds with small molecular weight, such as ethanediamine, diethylenetriamine and triethylenetetramine, which may lead to the less amount of the dithiocarbamate groups on the silica [14,18,19]. In this paper, a new silica-supported dithiocarbamate adsorbent (Si-DTC) was synthesized following a novel synthesis route: a kind of macromolecular dithiocarbamate (MDTC) was synthesized through the reaction of polyethyleneimine with carbon disulﬁde under strong alkaline conditions and then immobilized onto the chloro-functionalized silica matrix (SiCl) which had

262

L. Bai et al. / Journal of Hazardous Materials 195 (2011) 261–275

Scheme 1. Synthesis of chloro-functionalized silica gel (SiCl).

been modiﬁed by ␥-chloropropyltriethoxysilane. By adopting this method, the improvement of the nucleophilic reaction rate of polyethyleneimine with carbon disulﬁde under the strong alkaline condition could be achieved and the degradation of silica matrix could be avoided. The synthesized adsorbents were used to remove Pb(II), Cd(II), Cu(II) and Hg(II), which are known to be common heavy metals from aqueous solution, and the corresponding removal performance has been studied. In addition, the adsorption characteristics and thermodynamic properties of the adsorbent were examined.

solution) was purchased from Saibo Chemical Factory, Wuhan, Hubei Province, China. Metal salts of CuSO4 ·5H2 O, Pb(NO2 )2 , 3CdSO4 7H2 O and HgSO4 which used to prepare metal ion stock solutions were obtained from Sinopharm Group Chemical Reagent Co. Ltd., Shanghai, China. The stock solutions of Pb(II), Cd(II), Cu(II) and Hg (II) (25 mmol L−1 for each ion) were prepared by dissolving appropriate amounts of metal salts in deionized water. The working solutions ranged from 0.25 mmol L−1 to 3.0 mmol L−1 of the metal ions were prepared by diluting the stock solutions to appropriate volumes. All the other reagents used were of analytical reagent grade except the silica gel.

2. Experimental 2.2. Apparatus 2.1. Materials The silica gel used was of chromatographic grade (100–120 mesh size), obtained from Qingdao Haiyang Chemical Co., Ltd., Qingdao, Shandong Province, China. ␥Chloropropyltriethoxysilane (CPTS) was purchased from Xiya Chemical Factory, Chengdu, Sichuan Province, China. Polyethyleneimine (PEI, molecular weight = 1000, 50 wt% aqueous

Infrared spectra were obtained on a Nicolet 6700 Fourier Transform Infrared Spectrophotometer (Thermo Scientiﬁc Co., USA), using KBr pellets in the 4000–400 cm−1 region with a resolution of 4 cm−1 , by accumulating 16 scans. X-ray photoelectron spectroscopy measurements were carried out on a K-Alpha 1063 XPS spectrometer (Thermo Scientiﬁc Co., USA). A monochromatic Al K␣ X-ray source (1253.6 eV of photons) was used, with a spot area of

Scheme 2. Synthesis of silica-supported dithiocarbamate adsorbent (Si-DTC).

L. Bai et al. / Journal of Hazardous Materials 195 (2011) 261–275

263

400 ␮m. The base pressure in the working chamber was less than 10−12 Pa. All binding energies were referenced to the neutral C1s peak at 284.6 eV to compensate for the surface-charging effects. The data from XPS were analyzed by the XPSpeak 4.1 software. A Shirley baseline was used for the subtraction of the background, and Gaussian/Lorentzian (80/20) peaks were used for spectral decomposition. Elemental analysis of C, H, N and S for Si-DTC were subjected to be analyzed by the SC600 Carbon/Sulfur Determinator (LECO Co., USA) and TCH600 Nitrogen/Oxygen/Hydrogen Determinator (LECO Co., USA). The morphology of the adsorbents was examined on JSM-6360LV scanning electron microscope (JEOL Co., Japan). The concentrations of metal ions were determined by an IRIS Intrepid II XSP inductively coupled plasma spectrometer (Thermo Scientiﬁc Co., USA) for Hg(II) and a ﬂame atomic absorption spectrophotometer TAS-990F (Beijing Purkinje General Instrument Co., Ltd., China) for Pb(II), Cd(II) and Cu(II). The pH values of metal ion solutions were measured with a pHS-3C digital pH meter (Shanghai Precision & Scientiﬁc Co. Ltd., China). 2.3. Synthesis of silica-supported dithiocarbamate adsorbent Fig. 1. FT-IR spectra of activated silica gel (a), SiCl (b) and Si-DTC (c).

2.3.1. Synthesis of SiCl The chloro-functionalized silica gel (SiCl) was prepared as described in literature [20]. The silica gel (30 g) was ﬁrst activated with 1 mol L−1 nitric acid (200 mL) by reﬂuxing at 50 ◦ C stirring for 6 h. Afterwards the silica gel was washed with deionized water and methanol then the activated silica was air dried overnight. Activated silica gel (5 g) was dispersed into dry toluene (80 mL) in a three-necked round-bottomed ﬂask of total volume 250 mL. CPTS (12.5 mL) and dry toluene (40 mL) were well mixed in a separating funnel and then slowly poured into the ﬂask with continuous stirring. The suspension was mechanically stirred under reﬂux of the solvent at 100 ◦ C in N2 atmosphere for 48 h. The ﬁnal product chloro-functionalized silica (SiCl), was ﬁltered off, washed with toluene and methanol, extracted 6 hours by Soxhlet extraction with methanol at 90 ◦ C, and dried under vacuum at 50 ◦ C for 12 h successively. The reaction scheme is represented in Scheme 1. 2.3.2. Synthesis of MDTC Under certain conditions, dithiocarbamate was synthesized by the reaction of PEI with carbon disulﬁde and sodium hydroxide [21]. NaOH (21 g) dissolved in deionized water (50 mL) was added dropwise into a mixture of PEI (15 mL) and methanol (50 mL) in a three-necked round-bottomed ﬂask of total volume 250 mL under ice-bath with continuous mechanical stirring. Excess solution of carbon disulﬁde (30 mL) and ethanol (50 mL) were well mixed in a beaker and added dropwise through a funnel into the ﬂask with continuous stirring. After all the reagents had been added into the ﬂask, the mixture was heated to 40 ◦ C stirring for 22 h. As the reaction going on, the color of reaction solution changed from faint yellow to orangered. The synthesized product, macromolecular dithiocarbamate (MDTC), was white sodium salt and precipitated at the bottom of the ﬂask. Then the MDTC was ﬁltered and washed with methanol and air dried overnight. 2.3.3. Synthesis of Si-DTC MDTC (10 g) was dissolved in deionized water (100 mL) in a three-necked round-bottomed ﬂask and the pH of the MDTC solution was checked and adjusted to between 9 and 10 pH units by manually dropwise adding 0.1 mol L−1 sulfuric acid in methanol. Then SiCl (3 g) was dispersed into the ﬂask and the suspension reﬂuxed under 60 ◦ C for 10 h. The ﬁnal product, silica-supported dithiocarbamate adsorbent (Si-DTC), was ﬁltered off, washed with deionized water and methanol, and dried under vacuum at 50 ◦ C for 12 h. The reaction scheme is represented in Scheme 2.

2.4. Adsorption experiments Static adsorption experiment was employed to determine the adsorption capability of Si-DTC. The experiments were carried out by shaking 0.05 g of adsorbent with 20 mL of working solution with a certain metal ion concentration. The pH of the working solutions were adjusted adding diluted solution of nitric acid (for Hg(II)) or sulfuric acid (for Pb(II), Cd(II), Cu(II)) and sodium hydroxide, the acetic-acid–sodium-acetate buffer solutions were used when required. The samples were shaken for predetermined time period at a certain temperature and the solid was separated by ﬁltration. Temperature experiments were carried out in a constanttemperature water bath oscillator at the temperature ranged from 303 K to 353 K. Initial and equilibrium metal ion concentrations in the aqueous solutions were determined by using ﬂame atomic absorption spectrometer (AAS) for Pb(II), Cd(II), Cu(II) and inductively coupled plasma spectrometer (ICP) for Hg(II). The amount of metal ions adsorbed by Si-DTC was calculated according to the following Eq. (1): Q =

(C0 − C) , W

(1)

where Q is the amount of metal ions adsorbed onto unit amount of the Si-DTC (mmol g−1 ), C0 and C are the initial and equilibrium concentrations of the metal ions in aqueous phase (mmol L−1 ), respectively. V is the volume of the aqueous phase (L), and W is the dry weight of the adsorbent (g). The adsorption kinetics on the uptake of metal ions by resins was studied by placing 0.20 g resins with 100 mL working solution in a ﬂask. 1 mL solution was taken at different time intervals and the concentration of metal ion was determined by AAS or ICP. 3. Results and discussion 3.1. Characterization of the adsorbent 3.1.1. FT-IR Fig. 1 shows the FT-IR spectra of activated silica gel (a), SiCl (b) and Si-DTC (c). A large broad band between 3600 cm−1 and 3200 cm−1 was attributed to the presence of the O–H stretching frequencies of silanol groups and also to the remaining adsorbed water. The sharp features around 1100 cm−1 and 462 cm−1 indicate Si–O–Si stretching vibrations and bending vibrations, respectively [22]. A characteristic feature of the SiCl when compared with the

264

L. Bai et al. / Journal of Hazardous Materials 195 (2011) 261–275

Fig. 2. FT-IR spectra of MDTC, DTC-Pb, DTC-Cd, DTC-Cu and DTC-Hg.

activated silica gel was aliphatic C–H stretching frequencies at 2920 cm−1 and 2846 cm−1 , indicating that organic silane had been grafted onto the surface of silica gel. There were no prominent features of the dithiocarbamate group in the FT-IR spectra of Si-DTC because they were obscured by the strong absorption peaks of silica gel matrix around 1100 cm−1 and 470 cm−1 . Thus, the structure of MDTC and the metal chelate complexes of MDTC (DTC-Pb, DTC-Cd, DTC-Cu and DTC-Hg) were conﬁrmed through FT-IR. Fig. 2 shows the FT-IR spectra of MDTC, DTC-Pb, DTC-Cd, DTCCu and DTC-Hg. The characteristic adsorption peaks appearing at 1165 cm−1 , 653 cm−1 and 1453 cm−1 were attributed to C S vibration, C–S vibration and N–CS2 vibration, respectively [12,21,23]. The vibration frequencies of C S and C–S for DTC-Pb, DTC-Cd, DTC-Cu and DTC-Hg shifted to lower frequencies while the N–CS2 vibration shifted to higher frequencies when compared with MDTC. All of these indicate that a stronger metal-ligand bond had formed between metal ions and chelating groups in MDTC [24]. The characteristic absorption frequencies were summarized in Table 1.

substituted by MDTC during the synthesis of Si-DTC. Because of the high reactivities between SiCl and the amino groups of PEI which had not been subjected to dithiocarbamation, the MDTC should be attached to the surface of SiCl by C–N bond to form Si-DTC (as shown in Scheme 2). However, if the Cl atom of SiCl had been only substituted by N atom from MDTC, the content of N in Si-DTC should not have been less than the content of Cl in SiCl. Actually, from Table 2, the content of N (1.24 at.%) in Si-DTC is lower than the content of Cl (1.54 at.%) in SiCl. Therefore, we can infer that the covalent attachment between MDTC and SiCl was carried out by C–N bond and C–S bond, because the dithiocarbamate groups in MDTC contain a negative ion form and can interact with C–Cl bond of SiCl to form Si-DTC-1 through a new kind of C–S bond (as shown in Scheme 3).

3.1.2. XPS X-ray photoelectron spectroscopy (XPS) was performed to determine the chemical composition of the adsorbent surface. A portion of the wide-scan XPS spectra for the activated silica gel (a), SiCl (b) and Si-DTC (c) were shown in Fig. 3. The contents (expressed as atomic percent) of Si, O, C, Cl, S and N on the surface of activated silica gel, SiCl and Si-DTC were listed in Table 2. From Table 2, the contents of Si and O in SiCl and Si-DTC decreased because the silica matrix was covered with organic functional groups. The Cl element was not detected on the surface of Si-DTC, indicating that the Cl atom of SiCl was almost completely

3.1.3. SEM The SEM images of the activated silica gel (A and B), SiCl (C and D), Si-DTC (E and F) were shown in Fig. 4. It could be seen that the morphology of the three samples were similar, demonstrating that the silica gel have good mechanical stability and they have not been destroyed during the whole reaction. There appeared more multihole structures on the surface of Si-DTC (F) comparing with that of activated silica gel (B) and Si-Cl (D), mainly because of the microcorrosion of the silica matrix surface by the solution with pH values of 9–10 during the synthesis of Si-DTC.

Fig. 3. The XPS spectra of activated silica gel (a), SiCl (b) and Si-DTC (c).

Table 1 The absorption frequencies of MDTC, DTC-Pb, DTC-Cd, DTC-Cu and DTC-Hg. Group

Strength

C S C–S N–CS2

Strong Weak Very strong

Absorption frequencies MDTC

DTC-Pb

DTC-Cd

DTC-Cu

DTC-Hg

1165.48 653.02 1453.74

1098.22 617.19 1462.40

1117.44 614.21 1466.55

1107.01 617.79 1472.95

1123.84 611.39 1456.94

Table 2 The surface compositional data from XPS. Samples

Activated silica gel SiCl Si-DTC

XPS analysis (at.%) Si2p

O1s

C1s

Cl2p

N1s

S2p

33.29 29.89 29.45

66.71 59.83 59.94

– 8.75 9.07

– 1.54 –

– – 1.24

– – 0.3

L. Bai et al. / Journal of Hazardous Materials 195 (2011) 261–275

Scheme 3. The structure of Si-DTC-1.

Fig. 4. SEM images of the activated silica gel (A and B), Si-Cl (C and D), Si-DTC (E and F).

265

266

L. Bai et al. / Journal of Hazardous Materials 195 (2011) 261–275

Fig. 5. The effect of pH on the adsorption of Pb(II), Cd(II), Cu(II) and Hg(II) metal ions. [() Pb(II)] C0 = 200 mg L−1 ; [(䊉) Cd(II)] C0 = 100 mg L−1 ; [() Cu(II)] C0 = 40 mg L−1 ; [() Hg(II)] C0 = 200 mg L−1 ; t = 60 min; m(Si-DTC) = 0.05 g; temperature = 25 ◦ C.

Scheme 4. The acid decomposition of dithiocarbamates.

3.1.4. Elemental analysis Elemental analysis indicated 1.375% carbon, 28.96% nitrogen, 2.35% hydrogen and 1.10% sulfur in Si-DTC. It could be calculated that 1 g Si-DTC contained 0.17 mmol dithiocarbamate groups. 3.2. Adsorption behavior of Si-DTC 3.2.1. The effect of pH on adsorption The medium pH value affected the surface charge of the adsorbent, the degree of ionization and speciation of adsorbate during reaction [25].Thus the adsorption of metal ions on Si-DTC was examined at the pH range of 0.5–8.0 and the results were presented in Fig. 5. As shown in Fig. 5, a general increase in adsorption capacities with increasing pH of solution was observed. There was an abrupt increase at pH 3, which was close to the isoelectric point of the dithiocarbamate [26]. Below pH of 3, the H3 O+ ions of higher concentration will compete with M(II) to seize the adsorption sites, and as a result, less adsorption capacities were observed at low pH. When the pH increased, the concentration of H3 O+ ions decreased and the active adsorption sites mainly turned into disso-

Fig. 6. Adsorption rates of heavy metal ions onto Si-DTC. [() Pb(II)] C0 =200 mg L−1 , pH 5.0; [(䊉) Cd(II)] C0 = 100 mg L−1 , pH 7.0; [() Cu(II)] C0 = 40 mg L−1 , pH 5.0; [() Hg(II)] C0 = 200 mg L−1 , pH 6.0; m(Si-DTC) = 0.20 g; temperature = 25 ◦ C.

ciated forms, which resulted in the high afﬁnity of adsorption sites towards the metal ions. According to the literature [27], the dithiocarbamate compounds are generally unstable in acidic media and can be decomposed into the amine and carbon disulﬁde (presented in Scheme 4). And the dithiocarbamates began to decompose when the pH of the media was less than 4, and the decomposition rate reached maximum below pH of 2, which had been demonstrated in previous literatures [28,29]. All these indicate that during the pH range of 0.5–2.0 in our study, the dithiocarbamate groups were partially hydrolyzed, which also led to the poor adsorption capacities of Si-DTC at lower pH. The optimum pH value at which the maximum metal uptake were obtained as 5.0, 7.0, 5.0 and 6.0 for Pb(II), Cd(II), Cu(II) and Hg (II), respectively. These optimum pH values were used for all subsequent experiments. However, Pb(II), Cd(II), Cu(II) and Hg(II) start to be precipitated as hydroxide at pH 6.0, 8.0, 6.0 and 7.0, respectively. 3.2.2. Adsorption dynamics To determine an optimum contact time between the Si-DTC and heavy metal ion solutions, adsorption capacities of metal ions were measured as a function of contact time and the results were presented in Fig. 6. As shown in Fig. 6, there was a rapid uptake kinetics and adsorption equilibria, which would be attained within 10 min. The adsorption equilibrium time of our study and some other dithiocarbamate based materials given in the literatures were summarized in Table 3. As listed in Table 3, the adsorption dynamics of Si-DTC

Table 3 Adsorption equilibrium time and adsorption capacities of some dithiocarbamate functionalized materials reported in the literatures. Adsorption capacities (mmol g−1 )

Supporting material

Adsorption equilibrium time (min)

Pb

Cd

Cu

Hg

Sporopollenin Monosize polystyrene microspheresa Macroreticular styrene-divinylbenzene copolymer MCM-41-[N-(2-aminoethyl) dithiocarbamate] Polymer/organosmectite Silica (present study)

30 20 20

0.45 0.031 0.11

0.06 0.017 –

0.27 0.059 –

– 0.11 0.55

[30] [16] [31]

90 60/15 10

0.018 0.34 0.34

0.013 0.28 0.36

0.032 – 0.32

– 0.038 0.40

[32] [33,34] –

a

Adsorption capacities in the presence of the heavy metals.

Reference

L. Bai et al. / Journal of Hazardous Materials 195 (2011) 261–275

Fig. 7. Lagergren’s pseudo-ﬁrst-order plots for heavy metal ions on Si-DTC. [() Pb(II)] C0 = 200 mg L−1 , pH 5.0; [(䊉) Cd(II)] C0 = 100 mg L−1 , pH 7.0; [() Cu(II)] C0 = 40 mg L−1 , pH 5.0; [() Hg(II)] C0 = 200 mg L−1 , pH 6.0; m(Si-DTC) = 0.20 g; temperature = 25 ◦ C.

were much faster than the other dithiocarbamate functionalized adsorbents reported in the literatures [30,16,31–34]. Probably the fast adsorption of heavy metal ions onto Si-DTC could be ascribed to the good hydrophilicity of the silica matrix and the prominent chelate capacity of dithiocarbamate towards heavy metal ions. To ensure the equilibrium condition was achieved, the contact time of 60 min was chosen in the subsequent experiments. In order to interpret the kinetic characteristics of metal adsorption processes, Lagergren ﬁrst-order equation and pseudosecond-order equation model were used to evaluate experimental data. The linearized form of the ﬁrst-order rate equation by Lagergren and Svenska [35] is given as: log(Qe − Qt ) = log Qe −

k t 1 2.303

,

(2)

where Qe and Qt are the amounts of the metal ions adsorbed (mg g−1 ) at equilibrium and at contact time t (min), respectively, k1 (min−1 ) is the rate constant. The plots of log(Qe − Qt ) versus t were depicted in Fig. 7 and the rate constants (k1 ) and theoretical equilibrium adsorption capacities, Qe (theor.) were presented in Table 4. The experimental data were also ﬁtted by the pseudo-secondorder kinetic model which was given with the equation below [36]: t 1 = + Qt k2 Qe2

1 Qe

t,

267

Fig. 8. Pseudo second-order plots for heavy metal ions on Si-DTC. [() Pb(II)] C0 = 200 mg L−1 , pH 5.0;[(䊉) Cd(II)] C0 = 100 mg L−1 , pH 7.0; [() Cu(II)] C0 = 40 mg L−1 , pH 5.0;[() Hg(II)] C0 = 200 mg L−1 , pH 6.0; m(Si-DTC) = 0.20 g; temperature = 25 ◦ C.

pseudo-second-order model is based on the assumption that the rate-determining step may be a chemical adsorption involving valence forces through sharing or exchanging of electrons between adsorbent and adsorbate [37]. 3.2.3. Adsorption isotherms The effect of initial concentrations on metal ion adsorption was investigated by varying the initial concentrations of the metal ions at optimum pH value and 60 min of equilibration time. The obtained results were presented in Fig. 9. From Fig. 9, the higher is the initial concentration of the metal ion, the larger is the amount of the metal ion taken up. This increase in loading capacity of the adsorbent with relation to the metal ions concentration can be explained with the high driving force for mass transfer [38]. The maximum adsorption capacity could reach to 0.34 mmol g−1 , 0.36 mmol g−1 , 0.32 mmol g−1 and 0.40 mmol g−1 for Pb(II), Cd(II), Cu(II) and Hg(II), respectively, in the range of the experimental concentrations. The adsorption capacities of our study and some other dithiocarbamate based materials given in

(3)

where k2 (g mg−1 min−1 ) is the rate constant of pseudo secondorder adsorption reaction. The plots of t/Qt versus t were shown in Fig. 8 and the rate constants (k2 ) and theoretical equilibrium adsorption capacities, Qe (theor.) were presented in Table 4. From Table 4, the calculated equilibrium adsorption capacities, Qe (theor.) of the pseudo-ﬁrst-order kinetic model were not in accordance with the experimental adsorption capacities Qe (exp.). Furthermore, linear correlation coefﬁcients of the plots for all the four metal ions were not good. Therefore, it suggested that the adsorption of Pb(II), Cd(II), Cu(II) and Hg(II) onto Si-DTC cannot be described by ﬁrst-order kinetic reaction. As to the pseudosecond-order model, the correlation coefﬁcients for the slopes were superior to 0.999 in all the systems and the theoretical Qe values were close to the experimental Qe values. It was possible to suggest that the adsorption of Pb(II), Cd(II), Cu(II) and Hg(II) onto Si-DTC followed a second-order type reaction kinetics. The

Fig. 9. Effect of initial concentration on heavy metal adsorption for Si-DTC. [() Pb(II)] pH 5.0; [(䊉) Cd(II)] pH 7.0; [() Cu(II)] pH 5.0; [() Hg(II)] pH 6.0 ; m(SiDTC) = 0.05 g; temperature = 25 ◦ C.

268

L. Bai et al. / Journal of Hazardous Materials 195 (2011) 261–275

Table 4 First-order, second-order rate constants. Metal

Pb Cd Cu Hg

First-order rate constants

Second-order rate constants

Qe (exp) (mg g−1 )

k1 (min−1 )

Qe (theor.) (mg g−1 )

R2

k2 (g mg−1 min−1 )

Qe (theor.) (mg g−1 )

R2

67.39 43.47 16.40 82.71

0.0450 0.1031 0.0365 0.0335

15.92 16.90 1.604 9.492

0.9587 0.9901 0.9314 0.9913

0.000027 0.000133 0.000960 0.000044

69.40 44.00 16.45 83.64

0.9996 0.9999 0.9997 0.9999

the literature were summarized in Table 3. As can be seen from Table 3, these adsorption capacities of our materials are comparable to some other dithiocarbamate functionalized adsorbents reported before. The adsorption data for heavy metal ions were analyzed by ﬁtting the Langmuir, Freundlich and D–R adsorption isotherm models. The Langmuir isotherm model assumes a monolayer adsorption which takes place at speciﬁc homogeneous sites within the adsorbent and all the adsorption sites are energetically identical. In this model, a M(II) monolayer is formed on a modiﬁed adsorbent with a metal complex formation [39]. Linearized form of the Langmuir equation was given by the following equation [40]: 1 1 1 = 0 + , Qe Q bQ 0 Ce

(4)

where Qe is the amount of metal ion adsorbed on the adsorbent (mmol g−1 ), Ce is the equilibrium metal ion concentration in the solution (mmol L−1 ), Q0 represents a practical limiting adsorption capacity when the surface of adsorbent is completely covered with adsorbate and b is the Langmuir adsorption constant (L mmol−1 ). The plots of 1/Qe versus 1/Ce were shown in Fig. 10 and the values of Q0 and b were presented in Table 5. The Freundlich model assumes a heterogeneous adsorption surface with sites having different adsorption energies. Linearized form of the Freundlich equation was given by the following equation [41]: log Qe = log KF +

1 log Ce , n

Fig. 10. Langmuir isotherm plots for heavy metal ions on Si-DTC. [() Pb(II)] pH 5.0; [(䊉) Cd(II)] pH 7.0; [() Cu(II)] pH 5.0; [() Hg(II)] pH 6.0 ; m(Si-DTC) = 0.05 g; temperature = 25 ◦ C.

and linearized form of the equation was given as Eq. (7): ln Q = ln Qm − kε2 ,

where Q is the amount of metal ion adsorbed per unit weight of adsorbent (mol g−1 ), k is a constant related to the adsorption energy (mol2 kJ−2 ) and Qm is the maximum adsorption capacity (mol g−1 ), ε is the Polanyi potential (J mol−1 ) that can be written as:

(5)

where Qe is the equilibrium metal ion concentration on adsorbent (mmol g−1 ), Ce is the equilibrium concentration of the metal ion (mmol L−1 ), KF is the Freundlich constant (mmol g−1 ) which indicates the adsorption capacity and represents the strength of the adsorptive bond and n is the heterogeneity factor which represents the bond distribution. The plots of log Qe versus log Ce were shown in Fig. 11 and the values of KF and n were presented in Table 5. The D–R isotherm is more general than the Langmuir isotherm, because it does not assume a homogeneous surface or constant adsorption potential. The equilibrium data were also examined by the Dubinin–Radushkevich (D–R) model to determine the nature of adsorption processes whether it is physical or chemical [42].The D–R isotherm is given with the following Eq. (6) [43]: Q = Qm exp(−kε2 ),

(7)

1 Ce

ε = RT ln 1 +

.

(8)

The plots of ln Q versus ␧2 were shown in Fig. 12 and the values of Qm and k were presented in Table 5. Know from Table 5, the adsorption processes could not be described by the Langmuir adsorption isotherm model as negative values of b and Q0 were obtained and the linear correlation coefﬁcients (R2 ) of the plots for all the four metal ions were not good. The R2 values were in the range 0.85–0.99 indicating that the adsorption processes did not ﬁt well with the Freundlich model. Consequently, the D–R isotherm model ﬁtted best to the experimental data when the R2 values were compared and the maximum adsorption Qm values calculated from the experimental data were in good accordance

(6)

Table 5 Langmuir, Freundlich and D–R isotherm constants. Metal

Langmuir isotherms parameters 0

−1

Q (mmol g Pb Cd Cu Hg

−0.0327 0.3753 2.0133 0.4455

)

b (L mmol −3.4313 8.4767 0.1400 14.9447

−1

Freundlich isotherms parameters )

2

−1

R

KF (mmol g

0.9489 0.9471 0.9924 0.9659

0.3725 0.3319 0.1901 0.2806

)

D–R isotherms parameters

n

R

Qm (mmol g−1

k (mol2 kJ−2 )

R2

0.9656 9.3703 1.2220 4.9576

0.9207 0.9818 0.9952 0.8553

0.3635 0.3609 0.3144 0.3893

0.00046 0.00012 0.00003 0.00278

0.9960 0.9995 0.9998 0.9974

2

L. Bai et al. / Journal of Hazardous Materials 195 (2011) 261–275

Fig. 11. Freundlich isotherm plots for heavy metal ions on Si-DTC. [() Pb(II)] pH 5.0; [(䊉) Cd(II)] pH 7.0; [() Cu(II)] pH 5.0; [() Hg(II)] pH 6.0 ; m(Si-DTC) = 0.05 g; temperature = 25 ◦ C.

269

Fig. 12. D–R isotherm plots for heavy metal ions on Si-DTC. [() Pb(II)] pH 5.0; [(䊉) Cd(II)] pH 7.0; [() Cu(II)] pH 5.0; [() Hg(II)] pH 6.0 ; m(Si-DTC) = 0.05 g; temperature = 25 ◦ C.

Fig. 13. The high-resolution core-level spectra of N1s and S2p for Si-DTC and Si-DTC-Hg.

270

L. Bai et al. / Journal of Hazardous Materials 195 (2011) 261–275

Fig. 14. The high-resolution core-level spectra of (a) Hg4d in Si-DTC-Hg, (b) Cu2p3/2 in Si-DTC-Cu, (c) Pb4f7/2 in Si-DTC-Pb and (d) Cd3d5/2 in Si-DTC-Cd.

with the experimental Q values. The Dubinin–Radushkevitch (D–R) equation was originally proposed as an empirical adaptation of the Polanyi adsorption potential theory. In our study, there appeared multihole structure on the surface of Si-DTC as shown in Fig. 4F, mainly because the microcorrosion of the silica matrix surface by the solution with a pH value of 9–10 during the synthesis of SiDTC. Therefore, the adsorption of heavy metal ions onto the Si-DTC surface was pore-ﬁlling rather than layer-by-layer surface cover-

age, which was in correspondence with the postulate of the D–R equation [44]. To evaluate the nature of interaction between heavy metal ions and the binding sites, the mean free energy of adsorption (E) was calculated from the k values using the following equation [45]:

E = (2k)

−1 2

.

Scheme 5. The structure of Si-DTC-Hg.

(9)

L. Bai et al. / Journal of Hazardous Materials 195 (2011) 261–275

271

Fig. 15. The high-resolution core-level spectra of N1s and S2p for Si-DTC-Cu, Si-DTC-Pb and Si-DTC-Cd.

The values of E were found to be 32.96 kJ mol−1 , 64.55 kJ mol−1 , 129.10 kJ mol−1 and 13.41 kJ mol−1 for Pb(II), Cd(II), Cu(II) and Hg(II), respectively. The mean free energy of adsorption (E) per mole of the adsorbate is the energy required to transfer one mole of an adsorbate to the surface from inﬁnity in solution [46]. The E

value of Hg(II) is much smaller than the E values of Pb(II), Cd(II) and Cu(II), i.e., it needs much more energy to transfer Pb(II), Cd(II) and Cu(II) onto the surface of Si-DTC than Hg(II), which results in stronger adsorbing tendency of Hg(II) onto the surface of Si-DTC than that of Pb(II), Cd(II) and Cu(II), and this also indicates that the

272

L. Bai et al. / Journal of Hazardous Materials 195 (2011) 261–275

Scheme 6. The structure of Si-DTC-A (A = Pb(II), Cd(II) or Cu(II))

adsorption mechanism of Hg(II) onto Si-DTC may be quite different from that of Pb(II), Cd(II) or Cu(II) onto Si-DTC. Moreover, the magnitude of E values indicates the adsorbing tendency of Si-DTC to heavy metal ions, which basically matches the adsorption capacities order in our study because the adsorption capacity of Hg(II) from Si-DTC is much higher than that of Pb(II), Cd(II) and Cu(II).

3.2.4. Adsorption mechanism of Pb(II), Cd(II), Cu(II) and Hg(II) onto Si-DTC In order to clarify the adsorption mechanism of Pb(II), Cd(II), Cu(II) and Hg(II) onto Si-DTC, the XPS analyses for SiDTC and M(II) loaded Si-DTC (abbreviated as Si-DTC-M) were done.

Table 6 Thermodynamic parameters for the adsorption of Pb(II), Cd(II), Cu(II) and Hg(II) on Si-DTC. Metal

Pb

Cd

Cu

Hg

C0 (mg L−1 )

H◦ (kJ mol−1 )

S◦ (J K−1 mol−1 )

1.69

66.46

23.46

135.96

1.72

64.47

23.83

135.07

1.34

66.10

24.03

138.96

4.40

95.45

11.31

116.60

200

100

60

200

T (K)

G◦ (kJ mol−1 )

303 308 313 318 323 328 333 338 343 348 353 303 308 313 318 323 328 333 338 343 348 353 303 308 313 318 323 328 333 338 343 348 353 303 308 313 318 323 328 333 338 343 348 353

−18.50 −18.84 −19.17 −19.50 −19.83 −20.16 −21.84 −22.52 −23.20 −23.88 −24.56 −17.82 −18.15 −18.47 −18.79 −19.11 −19.44 −21.17 −21.84 −22.52 −23.19 −23.87 −18.70 −19.03 −19.36 −19.69 −20.03 −20.36 −22.27 −22.96 −23.66 −24.35 −25.05 −24.54 −25.02 −25.49 −25.97 −26.45 −26.93 −27.54 −28.12 −28.71 −29.29 −29.87

R2

0.9913

0.9945

0.9928

0.9913

0.9933

0.9926

0.9933

0.9908

L. Bai et al. / Journal of Hazardous Materials 195 (2011) 261–275

3.2.4.1. The XPS results of Si-DTC and Si-DTC-Hg. The highresolution core-level spectra of S2p and N1s of Si-DTC and Si-DTC-Hg resolved into individual component peaks were shown in Fig. 13. The S2p binding energies of C–S bond of dithiocarbamate group in Si-DTC and Si-DTC-Hg were 162.88 eV (peak 2 in Fig. 13a) and 165.00 eV (peak 2 in Fig. 13c), respectively, i.e., the S2p binding energy of C–S bond of dithiocarbamate groups in Si-DTC-Hg was much higher than that in Si-DTC, which results from the donation of the electrons from S atoms of C–S bonds of dithiocarbamate groups to Hg(II) [47]. The peak 3 at 163.62 eV in Fig. 13a and peak 3 at 163.99 eV Fig. 13c were attributed to a new C–S bond [48] in Si-DTC-1(presented in Scheme 3) and this new C–S bond in SiDTC-Hg (presented in Scheme 5) without the complexing formation between Hg(II), respectively. The S2p binding energies of the C S bond of dithiocarbamate groups [49] in Si-DTC (peak 1 in Fig. 13a) and Si-DTC-Hg (peak 1 in Fig. 13c) were both 162.22 eV, which indicates no chelating bond is formed between Hg(II) and the C S bond. The N1s binding energies of N–CS2 bond in Si-DTC and in SiDTC-Hg were 398.30 eV (peak 1 in Fig. 13b) and 399.30 eV (peak 1 in Fig. 13d), respectively, i.e., the N1s binding energy of N–CS2 bond in Si-DTC-Hg was higher than that in Si-DTC. This is because of the chelating formation between Hg(II) and the N atom of dithiocarbamate group [47], which results in the transfer of electrons from N atom of dithiocarbamate group to Hg(II). The peak 2 at 399.75 eV in Fig. 13b corresponded to the –NH/–NH2 groups, which had not been subjected to dithiocarbamation in Si-DTC. The peak 2-2 at 400.51 eV in Fig. 13d corresponded to the –NH/–NH2 groups which forms complexes with Hg(II) in Si-DTC-Hg, i.e., the N1s binding energy of a portion of –NH/–NH2 groups in Si-DTC increased about 0.8 eV after the adsorption of Hg(II), which results from the donation of the lone pair of electrons of N atom to the shared bond between the N atom of –NH/–NH2 groups and Hg(II), and as a consequence, the electron cloud density of this N atom decreased [50,51]. The N1s binding energies of –NR2 in Si-DTC and in Si-DTC-Hg were 401.10 eV (peak 3 in Fig. 13b) and 401.30 eV (peak 3 in Fig. 13d) in Si-DTC-Hg [52], respectively, i.e., there was no obvious change of N1s binding energies of –NR2 after the adsorption for Hg(II), which indicates that there was no or little chelating interaction between tertiary amines and Hg(II). Since the spectra of Hg4f overlaps with that of Si2p , the signal of Hg4d was presented in Fig. 14a to investigate the adsorption of Hg(II) on Si-DTC. The binding energy of Hg4d5/2 and Hg4d3/2 both revealed single peak around 360.10 eV and 378.30 eV, respectively, which indicates that Hg(II) was adsorbed only through the chelating binding between Hg(II) and sulfur or nitrogen [53].

3.2.4.2. The XPS results of Si-DTC-Cu, Si-DTC-Pb and Si-DTC-Cd. The high-resolution core-level spectra of S2p and N1s of Si-DTC-Cu, SiDTC-Pb and Si-DTC-Cd resolved into individual component peaks were shown in Fig. 15. The S2p binding energies of C S bond of dithiocarbamate groups (peak 2 in Fig. 15a), C–S bond of dithiocarbamate group (peak 2 in Fig. 15a) and a new C–S bond (peak 3 in Fig. 15a) in Si-DTC-1, were 162.10 eV, 162.98 eV and 163.86 eV, respectively, i.e., the S2p binding energies of all these three kinds of bonds in Si-DTC-Cu were almost the same with those in SiDTC. From Fig. 15b, the N1s binding energies of N–CS2 bond (peak 1), –NH/–NH2 groups (peak 2-1/peak 2-2) and –NR2 (peak 3) in Si-DTC-Cu, were 398.78 eV, 399.55 eV/400.15 eV and 401.25 eV, respectively. According to the analysis results for N1s binding energies of Si-DTC and Si-DTC-Hg, these results indicate that the –NH/–NH2 groups were also formed as complexes with Cu(II) in Si-DTC-Cu. The S2p and N1s results of Si-DTC-Pb and Si-DTC-Cd (presented in Fig. 15c–f) were almost the same with those of Si-DTC-Cu, and

273

this shows that the adsorption mechanisms of Si-DTC for Pb(II), Cd(II) or Cu(II) were the same. It seemed no chemical combination between Cu(II), Pb(II) or Cd(II) and the C–S bond of dithiocarbamate groups. Nevertheless, the high-resolution core-level spectra of Cu2p3/2 , Pb4f7/2 and Cd3d5/2 reveal notable phenomena in Fig. 14b–d. Fig. 14b shows the main peak of Cu2p3/2 around 934.40 eV, along with its weaker satellite peak around 942.78 eV [54]. The peak of Cu2p3/2 was resolved into two component peaks. Peak 1 at 932.60 eV was characteristic of the complexation which was formed between –NH/–NH2 groups and Cu(II) [55]. Peak 2 at 934.30 eV corresponded to the Cu2p3/2 electron binding energies of the cupric forms [56], which suggests that Cu(II) was seized in cupric forms by the dissociated dithiocarbamate sites at the same time. The spectral decomposition results of Pb4f7/2 in Fig. 14c and Cd3d5/2 in Fig. 14d also show two valence states, respectively, one is of divalent ion form and another of metal-amine complexes [55,57–59], which were similar to Cu2p3/2 . It also suggests that the adsorption mechanisms of Si-DTC for Pb(II) or Cd(II) were in accordance with that for Cu(II). 3.2.4.3. The differences of adsorption mechanisms between Pb(II), Cd(II), Cu(II) or Hg(II) onto Si-DTC. The N1s binding energy results for Si-DTC and Si-DTC-M indicate that there were a portion of the –NH/–NH2 groups conformed coordination with all of the four heavy metal ions, that’s the reason why adsorption amounts of Si-DTC to metal ions were much higher than the content of dithiocarbamate groups attached to silica matrix. As discussed in Sections 3.2.4.1 and 3.2.4.2, the adsorption mechanism of Hg(II) (presented in Scheme 5) is quite different from that of Pb(II), Cd(II) or Cu(II) (presented in Scheme 6). Besides the –NH/–NH2 groups, the N atom of N–CS2 and the S atom of the C–S bond in dithiocarbamate groups were also coordinated to the Hg(II), while no chelating reactions for Pb(II), Cd(II) or Cu(II) took place with the N atom of N–CS2 and the S atom of the C–S bond. However, the XPS results of Cu2p3/2 , Pb4f7/2 and Cd3d5/2 suggest that a portion of Pb(II), Cd(II) and Cu(II) were adsorbed in a divalent ion form by the dithiocarbamate anions, i.e., there were also chemical bonds which were formed between Pb(II), Cd(II) or Cu(II) with the C–S bond of dithiocarbamate groups. Moreover, the FI-IR results of MDTC, DTC-Pb, DTC-Cd, DTCCu and DTC-Hg in Section 3.1.1 also evidenced the differences of the adsorption mechanisms. As shown in Table 1, the order of absorption redshift for the C–S bond of dithiocarbamate group was DTC-Hg > DTC-Cd > DTC-Pb > DTC-Cu. The combination between Hg(II) and the C–S bond was of strong covalent properties, which leads to the most prominent shift of the adsorption frequency. Meanwhile, the chemical bindings of Pb(II), Cd(II) or Cu(II) to the dithiocarbamate anions had weak tendencies of sharing electrons, which results in the relatively small variation of the C–S bond adsorption frequencies. The stretching vibration of N–CS2 bond of MDTC at 1453 cm−1 revealed considerable double bond characteristics, because the stretching mode of the single C–N bond is shortened under the inﬂuence of the polar ionic bond of C+ –S− [47,60]. From Table 1, the stretching vibration frequencies of N–CS2 bond of DTC-Pb, DTC-Cd and DTC-Cu shifted to higher frequencies comparing with MDCT, it is because the chemical bindings of Pb(II), Cd(II) or Cu(II) to the C–S bond of dithiocarbamate group produce the effect of electrostatic induction, which enhances the double bond characteristics of the N–CS2 bond. As for DTC-Hg, the N atom of N–CS2 bond and the S atom of the C–S bond in dithiocarbamate groups both have the tendencies to transfer electrons to Hg(II), which counteracts the tendency of the electron cloud dispersing to the two sides of N–CS2 bond, as a result, the stretching vibration of N–CS2 bond of DTC-Hg does not change, and is almost the same with MDTC.

274

L. Bai et al. / Journal of Hazardous Materials 195 (2011) 261–275

11.31 kJ mol−1 for Pb(II), Cd(II), Cu(II) and Hg(II), respectively, in the range of 333K–353K. This is probably because in the range of 333K–353K, the transportation of heavy metal ions was not appreciably improved by the increasing temperature, but the chelating reactivity of Si-DTC for heavy metal ions was promoted by the higher temperature. In this way, the chemical reaction became the leading factor during the adsorption process. 4. Conclusions

Fig. 16. The log KD −1 /T graphs for the adsorption of heavy metal ions on Si-DTC. [() Pb(II)] C0 = 200 mg L−1 , pH 5.0; [(䊉) Cd(II)] C0 = 100 mg L−1 , pH 7.0; [()Cu(II)] C0 = 60 mg L−1 , pH 5.0; [() Hg(II)] C0 = 200 mg L−1 , pH 6.0; m(Si-DTC) = 0.05 g; t = 60 min.

3.2.5. Adsorption thermodynamics The effect of temperature on the adsorption of heavy metal ions onto Si-DTC was given from the plots and curves of the distribution coefﬁcient Kd values versus adsorption temperatures in Fig. 16. It can be found in Fig. 16 that Kd increased with temperature increasing, a certiﬁcation of the endothermic adsorption nature. Thermodynamic parameters including free energy change (Go ), enthalpy change (Ho ) and entropy change (So ) were calculated according to Eqs. (10)–(12). The Gibbs free energy change of the process was related to the distribution coefﬁcient (Kd ) in the linearized form of Eq. (10): Go = −RT ln Kd

(10)

Go = H o − TS o

(11)

S o H o − , log Kd = 2.303R 2.303RT

(12)

where Kd is the distribution coefﬁcient (mL g−1 ), and R is the gas constant (8.314 J mol−1 K−1 ). The calculated values of thermodynamic parameters were listed in Table 6. As shown in Table 6, the values of Go were negative at all temperatures, conﬁrming that the adsorption of heavy metal ions onto Si-DTC was spontaneous and thermodynamically favorable. The increase in absolute value of −Go as temperature rising indicates that the adsorption process of metal ions on Si-DTC becomes more favorable at higher temperature. The positive values of So suggest the randomness at the solid-solution interface increases during the adsorption of heavy metal ions on Si-DTC. The enthalpy changes Ho were positive at all temperatures represented the endothermic natures of adsorption processes and the heat maybe consumed to transfer the heavy metal ions from aqueous solution onto Si-DTC. Although there are no certain criteria related to the Ho values that deﬁne the adsorption type, the heat of adsorption values between 20.9 kJ mol−1 and 418.4 kJ mol−1 are frequently assumed as the comparable values for chemical adsorption processes [61]. Known from Table 6, Ho values were 1.69 kJ mol−1 , 1.72 kJ mol−1 , 1.34 kJ mol−1 and 4.40 kJ mol−1 for Pb(II), Cd(II), Cu(II) and Hg(II), respectively, at the range of 303K–328K. We hypothesized that in the range of 303K–328K, the transportation of heavy metal ions from aqueous solution onto the surface of Si-DTC was improved by the increasing temperature, which reveals a physical nature of the adsorption process. And the Ho values were 23.46 kJ mol−1 , 23.83 kJ mol−1 , 24.03 kJ mol−1 and

In the presented study, a new procedure for the efﬁcient synthesis of silica-supported dithiocarbamate (Si-DTC) has been developed which observably improved the adsorption capacities of dithiocarbamate adsorbent and the properties of this adsorbent have been examined. The adsorption behavior of heavy metal ions by Si-DTC is pH-dependent and the decomposition of dithiocarbamate in acid media leads to the poor adsorption capacities of Si-DTC at lower pH. The adsorption process for the heavy metal ions for Si-DTC can be explained with pseudo second-order type kinetic model, which is based on the assumption that the rate-determining step is a chemical adsorption. The adsorption of all the metal ions on SiDTC could be expressed by D–R type adsorption isotherms, which shows the non-homogenous characteristics of the adsorption sites on the adsorbent. Chelating interactions are accompanied by an increase in entropy and exhibit endothermic enthalpy values. The dithiocarbamate groups and the amino groups in Si-DTC both take part in the adsorption process for M(II) from aqueous solutions but the adsorption mechanism of Hg(II) onto Si-DTC is quite different from that of Pb(II), Cd(II) or Cu(II) onto Si-DTC, which is testiﬁed by the XPS and FT-IR results. Acknowledgements The authors are grateful to the National Basic Research Program of China (973 Project) (No. 2007CB613601), the College Student Innovation Experiment Program of Central South University (No. LC09096) and the Program for New Century Excellent Talents in University (2008) for their ﬁnancial support and assistance in this study. References [1] H. Arslanoglu, H.S. Altundogan, F. Tumen, Heavy metals binding properties of esteriﬁed lemon, J. Hazard. Mater. 164 (2009) 1406–1413. [2] S. Mauchauffée, E. Meux, Use of sodium decanoate for selective precipitation of metals contained in industrial wastewater, Chemosphere 69 (2007) 763–768. [3] L. Melita, M. Popescu, Removal of Cr (VI) from industrial water efﬂuents and surface waters using activated composite membranes, J. Membr. Sci. 312 (2008) 157–162. [4] C.A. Basha, N.S. Bhadrinarayana, N. Anantharaman, K.M. Meera Sheriffa Begum, Heavy metal removal from copper smelting efﬂuent using electrochemical cylindrical ﬂow reactor, J. Hazard. Mater. 152 (2008) 71–78. [5] C.H. Xiong, C.P. Yao, Study on the adsorption of cadmium(II) from aqueous solution by D152 resin, J. Hazard. Mater. 166 (2009) 815–820. [6] C.Y. Chen, C.L. Chiang, P.C. Huang, Adsorptions of heavy metal ions by a magnetic chelating resin containing hydroxy and iminodiacetate groups, Sep. Purif. Technol. 50 (2006) 15–21. [7] N.M. Abd El-Moniem, M.R. El-Sourougy, D.A.F. Shaaban, Heavy metal ions removal by chelating resin, Pigment Resin Technol. 34 (2005) 332–339. [8] E. Pehlivan, T. Altun, The study of various parameters affecting the ion exchange of Cu2+ , Zn2+ , Ni2+ , Cd2+ and Pb2+ from aqueous solution on Dowex 50W synthetic resin, J. Hazard. Mater. B134 (2006) 149–156. [9] I. Karadjova, Determination of Cd, Co, Cr, Cu, Fe, Mn, Ni and Pb in natural waters, alkali and alkaline earth salts by electrothermal atomic adsorption spectrometry after preconcentration by column solid phase extraction, Mikrochim. Acta 130 (1999) 185–190. [10] J.F. Dingman, K.M. Gloss, E.A. Milano, S. Siggia, Concentration of heavy metals by complexation on dithiocarbamate resins, Anal. Chem. 46 (1974) 774–777. [11] M.E. Mahmoud, M.M. El-Essawi, E.M.I. Fathallah, Characterization of surface modiﬁcation, thermal stability, and metal selectivity properties of silica gel phases-immobilized dithiocarbamate derivatives, J. Liq. Chromatogr. Related Technol. 27 (2004) 1711–1727.

L. Bai et al. / Journal of Hazardous Materials 195 (2011) 261–275 [12] K.A. Venkatesan, T.G. Srinivasan, P.R. Vasudeva Rao, Removal of complexed mercury from aqueous solutions using dithiocarbamate grafted on silica gel, Sep. Sci. Technol. 37 (2002) 1417–1429. [13] K. Dimos, P. Stathi, M.A. Karakassides, Y. Deligiannakis, Synthesis and characterization of hybrid MCM-41 materials for heavy metal adsorption, Microporous Mesoporous. Mater. 126 (2009) 65–71. [14] S. G-Renaudin, F. Gaslain, C. Marichal, B. Lebeau, R. Schneider, A. Walcarius, Synthesis of dithiocarbamate-functionalized mesoporous silica-based materials: interest of one-step grafting, New J. Chem. 33 (2009) 528–537. [15] B. Mathew, V.N.R. Pillai, Divinylbenzene-crosslinked polyacrylamidesupported dithiocarbamates as metal complexing agents, Polymer Bull. 26 (1991) 603–610. [16] A. Denizli, K. Kesenci, Y. Arica, E. Pis¸kin, Dithiocarbamate-incorporated monosize polystyrene microspheres for selective removal of mercury ions, React. Funct. Polym. 44 (2000) 235–243. [17] H.A. Claessens, M.A. van Straten, J.J. Kirkland, Effect of buffers on silica-based column stability in reversed-phase high-performance liquid chromatography, J. Chromatogr. A 728 (1996) 259–270. [18] M.E. Mahmoud, Selective solid phase extraction of mercury(II) by silica gel-immobilized-dithiocarbamate derivatives, Anal. Chim. Acta 398 (1999) 297–304. [19] S.G. Renaudin, R. Schneider, A. Walcarius, Synthesis of new dithiocarbamatebased organosilanes for grafting on silica, Tetrahedron Lett. 48 (2007) 2113–2116. [20] S.T. Beatty, R.J. Fischer, E. Rosenberg, Comparison of novel and patented silicapolyamine composite materials as aqueous heavy metal ion recovery materials, Sep. Sci. Technol. 34 (1999) 2723–2739. [21] H.L. Zheng, X.P. Sun, Q. He, K. Liang, P. Zhang, Synthesis and trapping properties of dithiocarbamate macromolecule heavy-metal ﬂocculants, J. Appl. Polym. Sci. 110 (2008) 2461–2466. [22] A.G.S. Prado, C. Airoldi, The pesticide 3-(3,4-dichlorophenyl)-1, 1-dimethylurea (Diuron) immobilized on silica gel surface, J. Colloid. Interface Sci. 236 (2001) 161–165. [23] A. McClain, Y.-L. Hsieh, Synthesis of polystyrene-supported dithiocarbamates and their complexation with metal ions, J. Appl. Polym. Sci. 92 (2004) 218–225. [24] X.S. Jing, F.Q. Liu, X. Yang, P.P. Ling, L.J. Li, C. Long, A.M. Li, Adsorption performances and mechanisms of the newly synthesized N,N -di(carboxymethyl) dithiocarbamate chelating resin toward divalent heavy metal ions from aqueous media, J. Hazard. Mater. 167 (2009) 589–596. [25] E. Repo, T.A. Kurniawan, J.K. Warchol, M.E.T. Sillanpää, Removal of Co(II) and Ni(II) ions from contaminated water using silica gel functionalized with EDTA and/or DTPA as chelating agents, J. Hazard. Mater. 171 (2009) 1071–1080. [26] E.M. Tyapochkin, E.I. Kozliak, Kinetic and binding studies of the thiolate-cobalt tetrasulfophthalocyanine anaerobic reaction as a subset of the Merox process, J. Mol. Catal. A: Chem. 242 (2005) 1–17. [27] D.J. Halls, The properties of dithioearbamates, Mikrochim. Acta (1969) 62–77. [28] E. Humeres, N.A. Debacher, M.M. de, S. Sierra, J.D. Franco, A. Schutz, Mechanisms of acid decomposition of dithiocarbamates. 1. Alkyl dithiocarbamates, J. Org. Chem. 63 (1998) 1598–1603. [29] E. Humeres, N.A. Debacher, M. Marta de, S. Sierra, Mechanisms of acid decomposition of dithiocarbamates. 2. Efﬁciency of the intramolecular general acid catalysis, J. Org. Chem. 64 (1999) 1807–1813. [30] N. Ünlü, M. Ersoz, Removal of heavy metal ions by using dithiocarbamatedsporopollenin, Sep. Purif. Technol. 52 (2007) 461–469. [31] P.K. Roy, A.S. Rawat, P.K. Rai, Synthesis, characterisation and evaluation of polydithiocarbamate resin supported on macroreticular styrene-divinylbenzene copolymer for the removal of trace and heavy metal ions, Talanta 59 (2003) 239–246. [32] P. Stathi, K. Dimos, M.A. Karakassides, Y. Deligiannakis, Mechanism of heavy metal uptake by a hybrid MCM-41 material: surface complexation and EPR spectroscopic study, J. Colloid Interface Sci. 343 (2010) 374–380. [33] R. Say, E. Birlik, Z. Erdemgil, A. Denizli, A. Ersöz, Removal of mercury species with dithiocarbamate-anchored polymer/organosmectite composites, J. Hazard. Mater. 150 (2008) 560–564. [34] R. Say, E. Birlik, A. Denizli, A. Ersöz, Removal of heavy metal ions by dithiocarbamate-anchored polymer/organosmectite composites, Appl. Clay Sci. 31 (2006) 298–305. [35] S. Lagergren, Zur theorie der sogenannten adsorption geloster stoffe, K. Sven. Vetenskapsakad. Handl. 24 (1898) 1–39. [36] Y.S. Ho, G. McKay, Pseudo-second order model for sorption processes, Process. Biochem. 34 (1999) 451–465.

275

[37] V.C. Taty-Costodes, H. Fauduet, C. Porte, A. Delacroix, Removal of Cd(II) and Pb(II) ions, from aqueous solutions, by adsorption onto sawdust of Pinus sylvestris, J. Hazard. Mater. B105 (2003) 121–142. [38] N. Ünlü, M. Ersoz, Adsorption characteristics of heavy metal ions onto a low cost biopolymeric sorbent from aqueous solutions, J. Hazard. Mater. B136 (2006) 272–280. [39] I˙ .H. Gubbuk, R. Gup, H. Kara, M. Ersoz, Adsorption of Cu(II) onto silica gelimmobilized Schiff base derivative, Desalination 249 (2009) 1243–1248. [40] I. Langmuir, The adsorption of gases on plane surfaces of glass, mica and platinum, J. Am. Chem. Soc. 40 (1918) 1361–1403. [41] H. Freundlich, Ueber die adsorption in loesungen, Z. Physik. Chem. 57 (1907) 385–470. [42] H. Dikici, K. Saltali, Equilibrium and kinetics characteristics of copper (II) sorption onto gyttja, Bull. Environ. Contam. Toxicol. 84 (2010) 147–151. [43] M.M. Dubinin, L.V. Radushkevich, Equation of the characteristic curve of activated charcoal, Proc. Acad. Sci. Phys. Chem. Sect. U.S.S.R. 55 (1947) 331–333. [44] N.D. Hutson, R.T. Yang, Theoretical basis for the Dubinin–Radushkevitch (D–R) adsorption isotherm equation, Adsorption 3 (1997) 189–195. [45] J.P. Hobson, Physical adsorption isotherms extending from ultrahigh vacuum to vapor pressure, J. Phys. Chem. 73 (1969) 2720–2727. [46] M.S. Onyango, Y. Kojima, O. Aoyi, E.C. Bernardo, H. Matsuda, Adsorption equilibrium modeling and solution chemistry dependence of ﬂuoride removal from water by trivalent-cation-exchanged zeolite F-9, J. Colloid Interface Sci. 279 (2004) 341–350. [47] R. Payne, R.J. Magee, J. Liesegang, Infrared and X-ray photoelectron spectroscopy of some transition metal dithiocarbamates and xanthates, J. Electron. Spectrosc. Relat. Phenom. 35 (1985) 113–130. [48] J.A. Gardella, S.A. Ferguson, R.L. Chin, ␲* ← ␲ Shakeup satellites for the analysis of structure and bonding in aromatic polymers by X-Ray Photoelectron Spectroscopy, Appl. Spectrosc. 40 (1986) 224–232. [49] P. Morf, F. Raimondi, H.G. Nothofer, B. Schnyder, A. Yasuda, J.M. Wessels, T.A. Jung, Dithiocarbamates: functional and versatile linkers for the formation of self-assembled monolayers, Langmuir 22 (2006) 658–663. [50] C. Ji, S. Song, C. Wang, C.M. Sun, R.J. Qu, C.H. Wang, H. Chen, Preparation and adsorption properties of chelating resins containing 3-aminopyridine and hydrophilic spacer arm for Hg(II), Chem. Eng. J. 165 (2010) 573–580. [51] M. Saad, C. Gaiani, M. Mullet, J. Scher, B. Cuq, X-ray Photoelectron Spectroscopy for wheat powders: measurement of surface chemical composition, J. Agric. Food Chem. 59 (2011) 1527–1540. [52] E.O. Guareno, F.S. Gutiérrez, J.L.M. Quiroz, S.L.H. Olmos, V. Soto, W. Cruz, R. Manríquez, S.G. Salazar, Removal of Cu(II) ions from aqueous streams using poly(acrylic acid-co-acrylamide) hydrogels, J. Colloid Interface Sci. 349 (2010) 583–593. [53] J. Dong, Z.H. Xu, F. Wang, Engineering and characterization of mesoporous silica-coated magnetic particles for mercury removal from industrial efﬂuents, Appl. Surf. Sci. 254 (2008) 3522–3530. [54] B. Kannamba, K. Laxma Reddy, B.V. AppaRao, Removal of Cu(II) from aqueous solutions using chemically modiﬁed chitosan, J. Hazard. Mater. 175 (2010) 939–948. [55] C. Liu, R. Bai, Q.S. Ly, Selective removal of copper and lead ions by diethylenetriamine-functionalized adsorbent: behaviors and mechanisms, Water Res. 42 (2008) 1511–1522. [56] S. Biniak, M. Pakula, G.S. Szymanski, A. Swiatkowski, Effect of activated carbon surface oxygen and/or nitrogen containing groups on adsorption of copper(II) ions from aqueous solution, Langmuir 15 (1999) 6117–6122. [57] T.J. Zhu, X-ray diffraction and photoelectron spectroscopic studies of (0 0 1)oriented Pb(Zr0.52 Ti0.48 )O3 thin ﬁlms prepared by laser ablation, J. Appl. Phys. 95 (2004) 241–247. [58] D.W. Niles, G. Herdt, M. Al-Jassim, An X-ray photoelectron spectroscopy investigation of O impurity chemistry in CdS thin ﬁlms grown by chemical bath deposition, J. Appl. Phys. 81 (1997) 1978–1985. [59] S. Thirumaran, K. Ramalingam, G. Bocelli, L. Righi, XPS, single crystal Xray diffraction and cyclic voltammetric studies on 1,10-phenanthroline and 2,2 -bipyridine adducts of bis(piperidinecarbodithioato-S,S ) cadmium(II) with CdS4 N2 environment—a stereochemical and electronic distribution investigation, Polyhedron 28 (2009) 263–268. [60] R. Payne, R.J. Magee, Chemical shits in transition metal dithiocarbamates from infrared and X-ray photoelectron spectroscopies, Chem. Phys. Lett. 93 (1982) 103–106. [61] J.M. Smith, Chemical Engineering Kinetics, third ed., McGraw-Hill, New York, 1981, pp. 310–322.

Journal of Hazardous Materials 195 (2011) 276–280

Contents lists available at SciVerse ScienceDirect

Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat

Enhanced degradation of p-nitrophenol in soil in a pulsed discharge plasma-catalytic system Tiecheng Wang a , Na Lu a,b , Jie Li a,b,∗ , Yan Wu a,b , Yan Su c a b c

Institute of Electrostatics and Special Power, Dalian University of Technology, Dalian 116024, PR China Key Laboratory of Industrial Ecology and Environmental Engineering, Ministry of Education of the People’s Republic of China, Dalian 116024, PR China Faculty of Chemical, Environmental and Biological Science and Technology, Dalian University of Technology, Dalian 116024, PR China

a r t i c l e

i n f o

Article history: Received 19 April 2011 Received in revised form 20 July 2011 Accepted 12 August 2011 Available online 10 September 2011 Keywords: Pulsed discharge plasma TiO2 photocatalyst Soil remediation p-Nitrophenol Density functional theory

a b s t r a c t A pulsed discharge plasma-TiO2 catalytic (PDPTC) system was developed to investigate the degradation of p-nitrophenol (PNP) in soil. The effects of TiO2 amount, soil pH and air moisture on PNP degradation were evaluated, and PNP degradation processes were predicted with Gaussian 03W combined with density functional theory (DFT). Experimental results showed that 88.8% of PNP could be smoothly removed in 10 min in the PDPTC system with the speciﬁc energy density of 694 J gsoil −1 , compared with 78.1% in plasma alone system. The optimum TiO2 amount was 2% in the present study, and higher TiO2 amount exhibited an inhibitive effect. Alkaline soil was favorable for PNP removal. The increase of air moisture to a certain extent could enhance PNP removal. A DFT calculation presented that there was a high preference for the –ortho and –para positions with respect to the functional –OH group of PNP molecule for • OH radicals attack. The main intermediates were hydroquinone, benzoquinone, catechol, phenol, benzo[d][1,2,3]trioxole, acetic acid, formic acid, NO2 − , NO3 − and oxalic acid. The generation of hydroxylated intermediates, NO2 − and NO3 − suggested that the experimental results were consistent with those of the theoretical prediction. © 2011 Elsevier B.V. All rights reserved.

1. Introduction Phenols, widely used as chemical intermediates in the manufacture of dyes, pesticides and medicines, are of great environmental interest because of their toxicity and stability [1]. Most phenols in soil come from fugitive emissions during their production and use, causing serious health hazards. Therefore, remediation of these phenols contaminated soil has been called to task. Several technologies such as chemical methods [2,3], bioremediation [4,5], electrokinetics remediation [6], and photocatalysis [7] have been employed to remedy phenols contaminated soils. With the strengthening of industrial standard and the increasing of economic values of lands, high efﬁcient and rapid soil remediation method is becoming a necessity. In this case, the conventional remediation technologies will not meet the requirement of high efﬁcient and rapid remediation due to the drawbacks such as second pollution and time-consuming. Recently, non-thermal discharge plasma, one of the advanced oxidation processes, has been widely exploited for organic pollutants removal [8,9]. Chemical effects (such as • OH, • O, H2 O2 and

∗ Corresponding author at: Institute of Electrostatics and Special Power, Dalian University of Technology, No.2 Linggong Road, Ganjingzi District, Dalian 116024, PR China. Tel.: +86 411 84708576; fax: +86 411 84709869. E-mail address: [email protected] (J. Li). 0304-3894/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.jhazmat.2011.08.041

O3 ) and physical effects (such as strong electric ﬁeld and ultraviolet (UV) light) in the discharge processes have been both received great attention [9,10]. In our previous studies, the chemical effects in discharge plasma were conﬁrmed to play a decisive role in pentachlorophenol removal in soil [11,12]. The physical effects in discharge plasma possess parts of discharge energy, and they played important roles in organic pollutants removal in water [9]. If these physical effects can be sufﬁciently utilized for pollutants removal in soil, it is of great beneﬁt to enhance the soil remediation. Anatase TiO2 , an economic and photosensitive semiconductor material with a band gap of about 3.2 eV, can be excited by strong electric ﬁeld and UV light radiation to generate electron–hole pair [13]. The electrons and holes are capable of initiating oxidation and reduction reactions on the surface of TiO2 particles. Heterogeneous photocatalysis of organic pollutants using TiO2 under UV-irradiation for soil remediation has been received great attention, and the results suggested that the photogenerated electrons could reduce the organic compounds or react with electron acceptors such as O2 , reducing it to superoxide radical anion O2 •− , and the photogenerated holes could oxidize organic pollutants, OH− ions and H2 O molecule to • OH radicals [7]. Moreover, it has been proved that the physical effects in discharge plasma process could be employed to excite TiO2 , resulting in accelerated formation of active species, and thus the degradation efﬁciency and energy efﬁciency of organic pollutants were enhanced in water [9,14]. Therefore, the physical effects in discharge plasma are expected

T. Wang et al. / Journal of Hazardous Materials 195 (2011) 276–280

100

PNP degradation efficiency (%)

to be utilized to improve organic pollutants removal in soil in the presence of TiO2 catalyst. The aim of this study is to investigate organic pollutant degradation by pulsed discharge plasma-TiO2 catalytic (PDPTC) system. p-Nitrophenol (PNP) was used as the model pollutant, which has been widely used as an important raw material for production of insecticides, herbicides and various synthetic compounds [1], and has been listed as the 129 priority toxic pollutants by U.S. Environmental Protection Agency [15]. The effects of some factors, such as TiO2 amount, soil pH and air moisture on PNP removal were evaluated. Furthermore, PNP degradation process was predicted by the Gaussian 03W program [16], and its main degradation intermediates were analyzed.

277

90 80

TiO2 amount (%)

70

0 1

60

2 5

50 0

2. Experimental

500

1000

1500

2000

-1

SED (J gsoil )

2.1. Materials

Fig. 1. Effect of TiO2 amount on PNP degradation.

PNP was used in the study, and its detailed introduction was presented in S1 of Supplementary Data (SD). Soil samples were collected from a suburb of Dalian, China. The details were presented in S2 of Supplementary Data. The original PNP concentration in the soil was 800 mg kg−1 . Soil pH was adjusted with NaOH and H2 SO4 solutions as described by Hultgren et al. [17]. TiO2 (Degussa, P25) (BET area = 50 m2 g−1 ) was used as the catalyst. 2.2. Treatment of contaminated soil sample The schematic diagram of the experimental apparatus was illustrated in Fig. S1 of Supplementary Data, which was similar with our previous work [11]. The details of the reactor were showed in S3 of Supplementary Data. The pulse frequency, pulsed discharge voltage and pulse-forming capacitance Cp were 100 Hz, 20 kV and 200 pF, respectively, and the input energy per pulse was 0.023 J. In each experiment, a certain amount of TiO2 was added into PNP contaminated soil and then homogenized. The soil sample (approximately 2.0 g) was spread on the ground electrode with a thickness of about 1.3 mm. Prior to discharge treatment, the moisture content of the soil sample was adjusted to 20% with deionized water. Air was injected from one side of the reactor and out from the other side with ﬂow rate of 0.5 l min−1 . Air moisture was adjusted by making the air pass through a scrubbing bottle containing deionized water, and a heating unit was used to adjust water vapor from the scrubbing bottle. Herein the moisture content (g m−3 ) means the mass of water vapor in a stere of air.

15.1 g m−3 , respectively. The introduction of TiO2 enhanced PNP removal in soil. When the TiO2 amount increased from 0 to 2%, PNP degradation efﬁciency increased by 10.7% at the SED of 694 J gsoil −1 . However, further increase presented an inhibitive effect. Maximum PNP degradation efﬁciency was obtained at the TiO2 amount of 2% in the present study. Less TiO2 addition (<2%) does not sufﬁciently utilize the energy of physical effects, therefore as the TiO2 amount increases, more photons may be adsorbed on the catalyst surface, and then more active species are formed, accelerating PNP degradation processes. However, at higher TiO2 amount, particles aggregation may reduce the interfacial area between pollutants and catalyst surface sites, and thus the number of active sites on the catalyst surface is decreased, resulting in the decrease of PNP degradation. Sohrabi et al. [18] found that aggregation of TiO2 particles at high concentrations caused a decrease in the number of surface active sites. Wang et al. [7] reported that 0.5% of TiO2 was effective for PNP photodegradation in soil, and further increase of the TiO2 amount from 0.5 to 2% had no signiﬁcant enhancement effect. In addition, TOC removal efﬁciency was enhanced in the PDPTC system. At SED of 1387 J gsoil −1 , TOC removal efﬁciency was 45.8% at TiO2 amount of 2%, compared with 28.3% in plasma alone system. 3.2. Effect of soil pH The effect of soil pH on PNP degradation was presented in Fig. 2. TiO2 amount and air moisture were 2% and 15.1 g m−3 , respectively.

2.3. Extraction and analysis

3. Results and discussion

PNP degradation efficiency (%)

After discharge treatment, PNP in soil was extracted immediately, and the extraction procedure was described in S4 of Supplementary Data. The extractions produced average recoveries of 90.1–95.3%. PNP concentration, intermediates and total organic carbon (TOC) were analyzed and the details were shown in S5 of Supplementary Data. The input energy per discharge, speciﬁc energy density and energy efﬁciency were deﬁned as shown in S6 of Supplementary Data. All experiments were conducted in duplicates.

100 80 60 40

The effect of TiO2 amount (w/w) on PNP degradation was presented in Fig. 1. Herein, soil pH and air moisture were 7.51 and

pH = 7.51 pH = 3.03

0 3.1. Effect of TiO2 amount on PNP degradation

pH = 9.12

20

173.5

347

693.5

1387.8 -1

SED (J gsoil ) Fig. 2. Effect of soil pH on PNP degradation.

2081.7

278

T. Wang et al. / Journal of Hazardous Materials 195 (2011) 276–280

Greatest PNP degradation efﬁciency occurred in alkaline soil, and followed by neutral soil, and the lowest in acidic soil. At SED of 694 J gsoil −1 , PNP degradation efﬁciencies were 48.2%, 88.8% and 94.5% in acidic, neutral, and alkaline soils, respectively. The characteristics of electron–hole pairs are inﬂuenced by pH changes. High OH− ion content could enhance the separation of electron–hole pairs, and then more • OH radicals were generated [7]. On the other hand, the changes of pH could affect the ionization state of PNP [19]. PNP is primarily in the molecular state when the pH is lower than pKa of PNP (pKa = 7.15), and it exists in ionic form when the pH is higher than the pKa . The weaker adsorption occurs between the ionic PNP and soil particles, and thus PNP has more chances to react with active species at pH 9.12 and 7.51. Thorstensen et al. [20] reported that in acidic soil, some weak acidic compounds had higher adsorbility on soil particles. The lower degradation efﬁciency of PNP in acidic soil might be attributed to the stronger adsorption of PNP on soil particles. On the other hand, NOx could also be generated through pulsed discharge plasma established in humid air [10], which suppressed O3 generation, and therefore, resulting in the decrease of PNP degradation. Furthermore, the ionic form of organic pollutant was generally more reactive towards oxidants than the molecular form [21]. Therefore, higher PNP degradation efﬁciencies occurred at soil pH 9.12 and 7.51. 3.3. Effect of air moisture The effect of air moisture on PNP degradation was presented in Fig. 3. Herein, the soil pH value was 7.51. PNP degradation efﬁciency ﬁrstly increased up to a certain point, and then decreased slowly with the increase of the air moisture. PNP degradation efﬁciency enhanced from 72.0% to 88.8% when the air moisture increased from 0 to 15.1 g m−3 , while it decreased to 65.8% at the air moisture of 48.6 g m−3 . Greatest PNP degradation efﬁciency was obtained at the air moisture of 15.1 g m−3 in the present study. The presence of water molecules is considered to play an important role in the formation of active species on the surface of TiO2 . On the one hand, water molecules can promote the active species generation through the following reactions [10,22,23]: energy

TiO2 −→ TiO2 + e− + h+ +

• OH

−

−

h + H2 O →

e + H2 O → e 1

O( D) + H2 O → N2 (A3

+H

+ •H

(1)

+

+

(2) • OH

(3)

2• OH

(4)

) + H2 O → N2 + • OH + • H

(5)

PNP degradation efficiency (%)

100

The active species can react with organic pollutant, and then enhance its degradation efﬁciency; on the other hand, the presence of excessive amount of water molecules could compete for the activated surface of TiO2 and occupied it, resulting in the decrease of pollutant removal efﬁciency [24]. Therefore, a positive effect of air moisture occurred ﬁrstly in the present experiment, and followed by a negative effect. Similar results were also observed by Date et al. [25], where the increase of air moisture to a certain extent could enhance the activity of TiO2 , and further increase would exhibit an inhibitive effect. In addition, a certain amount of water molecules could enhance the generation of • OH radicals [26], and thus improved PNP removal. The important role of • OH radicals played in PNP removal could be conﬁrmed by the increase of PNP degradation efﬁciency with air moisture increased from 0 to 15.1 g m−3 in Fig. 3, because the presence of water and N2 molecules in humid air could inhibit O3 production during non-thermal discharge plasma process [10]. However, with the further increase in air moisture, more highenergy electrons were captured by water molecules, and the generation of O3 was further suppressed and the formation of • OH radicals also decreased, resulting in the decrease of PNP degradation efﬁciency [27]. 3.4. Theoretical prediction of PNP degradation In order to predict the primary intermediates, the attack of active species such as • OH radicals was discussed by theoretical calculation. There are several shortcut methods for determining the attack positions of the • OH radicals. One of the most successful theories is the “frontier orbital theory”, which states that in electrophilic reactions, the attack point is at the position of the greatest electron density in the highest occupied molecular orbital (HOMO) of the aromatic molecule [28]. The HOMO coefﬁcients for PNP were shown in Fig. 4. The italic data in parentheses in Fig. 4 represented the HOMO coefﬁcients. These results indicated that there was a high preference for the two –ortho positions with respect to the functional –OH group for • OH radical attack. The main bond lengths, bond angles and atomic charges of PNP were calculated with the help of Gaussian 03W, and the results were shown in Table 1. The corresponding PNP structure could be seen in Fig. 4. As could be found in Table 1, the length of C(4)–N(13) ˚ which was the longest in PNP molecule, and bond was 1.462 A, therefore it would be potential ﬁrstly to be attacked by active species such as • OH radicals [29]. After the • OH radicals attacked the bond of C(4)–N(13) with the –NO2 group removed from the aromatic ring, intermediates such as phenol, hydroquinone, benzoquinone and catechol would be generated. On the other hand, the atomic charge can indicate its electronegativity to some extent in quantum chemistry. Endou et al. [30] have tested the validity of

80 60 40 20 0

0

8

15.1

-3

28.3

48.6

Air moisture (g m ) Fig. 3. Effect of air moisture on PNP degradation.

Fig. 4. HOMO coefﬁcients for PNP.

T. Wang et al. / Journal of Hazardous Materials 195 (2011) 276–280

279

Table 1 Main bond lengths, bond angles and atomic charges of PNP. Bond name

˚ Bond length (A)

Name of bond angle

Bond angles (◦ )

Atom

Atomic charges

C(1)–C(2) C(1)–C(6) C(1)–O(11) C(2)–C(3) C(2)–H(7) C(3)–C(4) C(3)–H(8) C(4)–C(5) C(4)–N(13) C(5)–C(6) C(5)–H(9) C(6)–H(10) N(13)–O(14) N(13)–O(15) O(11)–H(12)

1.386 1.403 1.357 1.386 1.084 1.397 1.083 1.394 1.462 1.389 1.083 1.087 1.232 1.233 0.967

O(11)–C(1)–C(2) O(11)–C(1)–C(6) C(2)–C(1)–C(6) C(3)–C(2)–C(1) C(4)–C(3)–C(2) C(5)–C(4)–C(3) C(3)–C(4)–N(13) C(5)–C(4)–N(13) C(6)–C(5)–C(4) C(1)–C(6)–C(5) C(4)–N(13)–O(14) C(4)–N(13)–O(15) O(14)–N(13)–O(15) – –

117.102 122.625 120.273 119.835 119.299 121.505 119.321 119.174 119.131 119.957 117.777 117.797 124.426 – –

C(1) C(2) C(3) C(4) C(5) C(6) H(7) H(8) H(9) H(10) O(11) H(12) N(13) O(14) O(15)

0.349 −0.112 −0.091 0.241 −0.092 −0.142 0.116 0.141 0.141 0.095 −0.542 0.326 0.381 −0.403 −0.406

the qualitative electronegativity using quantum chemical calculations based on the density functional theory. In the present study, the atomic charges of C(2) and C(6) had larger negative values than those of C(3) and C(5), which indicated that the electronegativities

of C(2) and C(6) were stronger than those of C(3) and C(5), and thus C(2) and C(6) would be attacked more easily by electrophilic radicals. Therefore, the substituted intermediates of the two –ortho positions with respect to the functional –OH group were much easier to generate than those of the two –meta positions. 3.5. Intermediates of PNP degradation in PDPTC system PNP degradation intermediates in soil were analyzed using ion chromatography (IC), HPLC and HPLC/MS. The total ion chromatogram of the intermediates after 30 min of discharge treatment was shown in Fig. 5. As presented in Fig. 5(a), hydroquinone, benzoquinone, catechol and phenol were detected as the intermediates using HPLC system. Hydroquinone, catechol, benzo[d][1,2,3]trioxole, and the dimer of PNP with diphenol were monitored using HPLC/MS system in Fig. 5(b). In addition, acetic acid, formic acid, NO2 − , NO3 − and oxalic acid were identiﬁed by IC in Fig. 5(c). Herein the formation of NO2 − and NO3 − in the case of pulsed discharge in air atmosphere was eliminated through control experiments. Considering these intermediates, the hydroxylated products were the major intermediates, and which were based on the substitution of –ortho and –para positions with respect to the functional –OH group. In addition, the formation of NO2 − and NO3 − indicated that the broken of the bond of C(4)–N(13) occurred with the –NO2 group removed from the aromatic ring during PNP degradation process. These experimental results were consistent with those of the theoretical prediction. Based on these intermediates, it could be deduced that PNP was ﬁrst oxidized into kinds of hydroxylated intermediates, and then further oxidized into acetic acid, formic acid and oxalic acid after aromatic rings were broken. The organic acids could also be further decomposed into carbon dioxide. The detailed degradation mechanisms of PNP in soil in the PDPTC system would be still further studied. 4. Conclusions

Fig. 5. Total ion chromatograph (TIC) of intermediates of PNP degradation after 30 min of discharge treatment in PDPTC system: (a) HPLC, (b) HPLC/MS, and (c) IC.

We have investigated the feasibility of utilizing TiO2 catalyst to promote PNP degradation in soil in a PDPTC system. Greater PNP degradation performance was presented in this system, compared with in plasma alone system. The increase of TiO2 amount to a certain extent could enhance PNP degradation, while further increase would present a negative effect. By inﬂuencing the ionization state of PNP and the physicochemical properties of TiO2 , soil pH exhibited signiﬁcant effects on PNP degradation, and greatest PNP degradation efﬁciency occurred at alkaline soil. The change of air moisture could affect the formation of active species and the

280

T. Wang et al. / Journal of Hazardous Materials 195 (2011) 276–280

activity of TiO2 catalyst, and therefore there exists appropriate air moisture for PNP degradation. The degradation intermediates, analyzed from PNP degradation experiment, were consistent with those predicted using the Gaussian 03W combined with density functional theory. This study is a fundamental research effort, trying to offer an alternative solution to utilize catalyst to improve pollutant removal in soil. Further work needs to be carried out from exploring the enhancement mechanisms for PNP degradation in soil in this PDPTC system. Supplementary data Text S1–S6 include introduction of PNP and other reagents, details of the soil sample, reactor introduction, extraction procedure, analysis methods, calculation of input energy per discharge, speciﬁc energy density and energy efﬁciency. Fig. S1 presents the schematic diagram of the experimental setup. Acknowledgements The authors thank the National Natural Science Foundation, P.R. China (Project No. 40901150), the Ministry of Science and Technology, P.R. China (Project No. 2008AA06Z308), and Program for Liaoning Excellent Talents in University, China (Project No. 2009R09) for their ﬁnancial support to this research. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.jhazmat.2011.08.041. References [1] V. Uberoi, S.K. Bhattacharya, Toxicity and degradability of nitrophenols in anaerobic systems, Water Environ. Res. 69 (1997) 146–156. [2] P. Ye, A.T. Lemley, Adsorption effect on the degradation of 4,6-o-dinitrocresol and p-nitrophenol in a montmorillonite clay slurry by AFT, Water Res. 43 (2009) 1303–1312. [3] G. Qiu, Y.F. Chen, S.J. Chen, Effect of different pre-processing methods on reductive degradation of p-nitrophenol in soils by iron chips, Chin. J. Environ. Eng. 4 (2010) 654–658. [4] S. Laha, K.P. Petrova, Biodegradation of 4-nitrophenol by indigenous microbial populations in Everglades soils, Biodegradation 8 (1997) 349–356. [5] S. Labana, G. Pandey, D. Paul, N.K. Sharma, A. Basu, R.K. Jain, Pot and ﬁeld studies on bioremediation of p-nitrophenol contaminated soil using Arthrobacter protophormiae RKJ100, Environ. Sci. Technol. 39 (2005) 3330–3337. [6] S.V. Ho, C.J. Athmer, P.W. Sheridan, A.P. Shapiro, Scale-up aspects of the LasagnaTM process for in situ soil decontamination, J. Hazard. Mater. 55 (1997) 39–60. [7] J.X. Wang, S. Chen, X. Quan, H.M. Zhao, Y.Z. Zhao, Enhanced photodegradation of PNP on soil surface under UV irradiation with TiO2 , Soil Sediment Contam. 16 (2007) 413–421.

[8] B. Sun, M. Sato, J.S. Clements, Oxidative processes occurring when pulsed high voltage discharges degrade phenol in aqueous solution, Environ. Sci. Technol. 34 (2000) 509–513. [9] Y.S. Mok, J.O. Jo, Degradation of organic contaminant by using dielectric barrier discharge reactor immersed in wastewater, IEEE Trans. Plasma Sci. 34 (2006) 2624–2629. [10] P. Lukes, M. Clupek, V. Babicky, V. Janda, P. Sunka, Generation of ozone by pulsed corona discharge over water surface in hybrid gas–liquid electrical discharge reactor, J. Phys. D: Appl. Phys. 38 (2005) 409–416. [11] T.C. Wang, N. Lu, J. Li, Y. Wu, Evaluation of the potential of pentachlorophenol degradation in soil by pulsed corona discharge plasma from soil characteristics, Environ. Sci. Technol. 44 (2010) 3105–3110. [12] T.C. Wang, N. Lu, J. Li, Y. Wu, Degradation of pentachlorophenol in soil by pulsed corona discharge plasma, J. Hazard. Mater. 180 (2010) 436–441. [13] A. Mills, S. LeHunte, An overview of semiconductor photocatalysis, J. Photochem. Photobiol. A 108 (1997) 1–35. [14] X.L. Hao, M.H. Zhou, L.C. Lei, Non-thermal plasma-induced photocatalytic degradation of 4-chlorophenol in water, J. Hazard. Mater. 141 (2007) 475–482. [15] USEPA, July 2002, http://www.scorecard.org. [16] M.J. Frisch, G.W. Trucks, H.B. Schlegel, et al., Gaussian 03, Gaussian Inc., Pittsburgh, PA, 2003. [17] R.P. Hultgren, R.J.M. Hudson, G.K. Sims, Effects of soil pH and soil water content on prosulfuron dissipation, J. Agric. Food Chem. 50 (2002) 3236–3243. [18] M.R. Sohrabi, M. Ghavami, Photocatalytic degradation of Direct Red 23 dye using UV/TiO2 : effect of operational parameters, J. Hazard. Mater. 153 (2008) 1235–1239. [19] D.Y. Tang, Z. Zheng, K. Lin, J.F. Luan, J.B. Zhang, Adsorption of p-nitrophenol from aqueous solutions onto activated carbon ﬁber, J. Hazard. Mater. 143 (2007) 49–56. [20] C.W. Thorstensen, O. Lode, O.M. Eklo, A. Christiansen, Sorption of bentazone, dichlorprop, MCPA, and propiconazole in reference soils from Norway, J. Environ. Qual. 30 (2001) 2046–2052. [21] J. Hoigne, H. Bader, Role of hydroxyl radical reactions in ozonation processes in aqueous-solutions, Water Res. 10 (1976) 377–386. [22] M.R. Ghezzar, F. Abdelmalek, M. Belhadj, N. Benderdouche, A. Addou, Enhancement of the bleaching and degradation of textile wastewaters by gliding arc discharge plasma in the presence of TiO2 catalyst, J. Hazard. Mater. 164 (2009) 1266–1274. [23] P. Pichat, J. Disdier, C. Hoang-Van, D. Mas, G. Goutailler, C. Gaysse, Puriﬁcation/deoderization of indoor air and gaseous efﬂuents by TiO2 photocatalysis, Catal. Today 63 (2000) 363–369. [24] K.H. Wang, H.H. Tsai, Y.H. Hsieh, The kinetics of photocatalytic degradation of trichloroethylene in gas phase over TiO2 supported on glass bead, Appl. Catal. B 17 (1998) 313–320. [25] M. Date, Y. Ichihashi, T. Yamashita, A. Chiorino, F. Boccuzzi, A. Haruta, Performance of Au/TiO2 catalyst under ambient conditions, Catal. Today 72 (2002) 89–94. [26] S. Zheng, L. Zhang, Y. Liu, W.C. Wang, X.G. Wang, Modeling of the production of OH and O radicals in a positive pulsed corona discharge plasma, Vacuum 83 (2008) 238–243. [27] R. Ono, T. Oda, Dynamics of ozone and OH radicals generated by pulsed corona discharge in humid-air ﬂow reactor measured by laser spectroscopy, J. Appl. Phys. 93 (2003) 5876–5882. [28] N. San, A. Hatipoglu, G. Kocturk, Z. Cinar, Photocatalytic degradation of 4-nitrophenol in aqueous TiO2 suspensions: theoretical prediction of the intermediates, J. Photochem. Photobiol. A 146 (2002) 189–197. [29] Q.Z. Dai, L.C. Lei, X.W. Zhang, Enhanced degradation of organic wastewater containing p-nitrophenol by a novel wet electrocatalytic oxidation process: parameter optimization and degradation mechanism, Sep. Purif. Technol. 61 (2008) 123–129. [30] A. Endou, T.W. Little, A. Yamada, K. Teraishi, M. Kubo, S.S.C. Ammal, A. Miyamoto, M. Kitajima, F.S. Ohuchi, Chemical interaction of NF3 with Si (Part II): density functional calculation studies, Surf. Sci. 445 (2000) 243–248.

Journal of Hazardous Materials 195 (2011) 281–290

Contents lists available at SciVerse ScienceDirect

Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat

Improving the mechanical characteristics and restraining heavy metal evaporation from sintered municipal solid waste incinerator ﬂy ash by wet milling Chang-Jung Sun a,∗ , Ming-Guo Li b , Sue-Huai Gau b , Ya-Hui Wang b , Yi-Lin Jan c a

Department of Civil and Environment Engineering, Nanya Institute of Technology, No.414, Sec. 3, Zhongshan E. Rd., Zhongli City, Taoyuan County 320, Taiwan, ROC Department of Water Resources and Environmental Engineering, Tamkang University, 151 Ying-chuan Road, Tamsui, Taipei County 251, Taiwan, ROC c Department of Civil Engineering, Ching Yun University, No. 229, Jianxing Road, Zhongli City, Taoyuan County, 320, Taiwan, ROC b

a r t i c l e

i n f o

Article history: Received 1 December 2010 Received in revised form 28 June 2011 Accepted 13 August 2011 Available online 22 August 2011 Keywords: Municipal solid waste incinerator ﬂy ash Milling Sintering Heavy metal evaporation

a b s t r a c t The milling process has a veriﬁed stabilizing effect on the leaching of heavy metals into the environment from municipal solid waste incinerator (MSWI) ﬂy ash. The aim of this current study is to further improve and conﬁrm the effectiveness of the process by exploring its effects on the evaporation of heavy metals and on the mechanical characteristics of the sintered MSWI ﬂy ash. The chemical composition of the MSWI ﬂy ash is ﬁrst altered by the addition of water treatment plant sludge (WTS) and cullet, and then processed to produce sintered specimens suitable for reuse as an aggregate. In the experiments, ﬂy ash, WTS and cullet (40%: 30%: 30%, respectively) were mixed and milled for 1 h. Samples were sintered for 60 min at temperatures of 850, 900, 950 and 1000 ◦ C. Test results conﬁrm that milling increased the compressive strength of the sintered specimens. The compressive strength of unmilled specimens sintered at 900 ◦ C was only 90 kg/cm2 , but that of milled specimens was 298 kg/cm2 when sintered at only 850 ◦ C. There was also an improvement in the soundness ranging from 11.04% to 0.02% and a reduction in the evaporation rates of Pb, Cd, Cu, Cr and Zn from 54–64%, 43–49%, 38–43%, 30–40% and 14–35% (900–1000 ◦ C) to 19–21%, 19–21%, 14–19%, 12–19% and 14–17% (850–1000 ◦ C), respectively. The improvement in compressive strength was attained by the combination in the liquid sintering stage of powdered ash with the amorphous material. The amorphousness of the material also helped to seal the surface of the ﬂy ash, thereby reducing the evaporation of heavy metals during the heating process. © 2011 Elsevier B.V. All rights reserved.

1. Introduction In recent years, Taiwan has been actively promoting the recycling of municipal solid waste incinerator (MSWI) ﬂy ash. Sintering technology has been adapted to modify the MSWI ﬂy ash so that it can be recycled as a building material. Parameters of interest in this process generally include sintering temperature, sintering time, compressive force during pellet fabrication and the composition of the material itself. The characteristics of the municipal solid waste (MSW) affect the characteristics of the ﬂy ash so the sintering parameters must be modiﬁed accordingly. Another problem that must be considered is the evaporation of heavy metals in the ﬂy ash during the sintering process, especially under higher temperatures. In past studies of the thermal treatment of MSWI ﬂy ash, the powdered ﬂy ash has been used without any pretreatment. Results

∗ Corresponding author. Tel.: +886 3 4361070x3409; fax: +886 3 4372193. E-mail address: [email protected] (C.-J. Sun). 0304-3894/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.jhazmat.2011.08.040

show that when sintering is carried out at 1000 ◦ C, the evaporation rates of Pb, Cd, Cu, and Zn are around 83–95%, 48–95%, 70–80%, and 20–40%, respectively [1–4]. The effects of vitriﬁcation treatment (at 1400 ◦ C) on both boiler ash and electrostatic precipitator (ESP) ash have also been investigated [5,6] with an obvious reduction in heavy metal leaching from melted slag after vitriﬁcation treatment. Results reveal that the heavy metals stabilized and became ﬁxed in the melted slag. The problem is that the vitriﬁcation treatment at 1400 ◦ C tends to cause a large amount of weight loss from the ash. This material loss contributes secondary ﬂue dust particles that could contain volatile elements such as chloride, sulfate, Pb and Cd. Crystallization technology has been combined with vitriﬁcation treatment in a two-stage process for the generation of glass–ceramic or brick material from incinerator ﬂy ash [7–9]. In the ﬁrst stage, vitriﬁcation is carried out at 1400–1500 ◦ C to melt the ﬂy ash after which the molten sample is re-crystallized by annealing or sintering at 700–1050 ◦ C. The mechanical properties of glass–ceramic material made from ﬂy ash (such as bending, compressive strength and hardness) are good and the process does inhibit the leaching of heavy

282

C.-J. Sun et al. / Journal of Hazardous Materials 195 (2011) 281–290

Table 1 The composition and heavy metal content of the experimental materials. Rawa Composition (%) Ca Si Al Na K Mg Fe Ti Heavy metal content (mg/kg) Cd Cr Cu Pb Zn

Extractedb

WTSc

32.1 2.57 0.44 3.83 4.44 0.72 0.40 0.09

± ± ± ± ± ± ± ±

0.75 0.10 0.06 0.21 0.19 0.03 0.07 0.01

40.89 4.95 0.90 0.77 0.66 1.69 0.96 0.22

± ± ± ± ± ± ± ±

1.21 0.38 0.07 0.09 0.04 0.06 0.05 0.01

246 288 1270 5090 10,800

± ± ± ± ±

4 72 80 90 200

521 562 3250 5140 29,800

± ± ± ± ±

61 63 70 220 1500

Cullet

3.78 26.81 10.51 3.13 5.05 1.46 4.56 0.39

± ± ± ± ± ± ± ±

1.07 0.92 0.80 0.22 0.35 0.06 0.18 0.01

119 184 242 613 150

± ± ± ± ±

39 20 36 88 43

6.35 36.12 1.55 5.63 0.97 1.55 0.75 0.04

± ± ± ± ± ± ± ±

0.25 1.80 0.04 0.15 0.02 0.09 0.03 0.00

11 158 75 23 206

± ± ± ± ±

1 3 2 1 6

Mean ± standard deviation (n = 3). a Raw reaction ﬂy ash. b Extracted ﬂy ash. c Water treatment sludge.

metals, however, the energy costs are high. The evaporation rate of heavy metals during the two-stage thermal treatment process itself is problematical as is the formation of secondary ﬂue dust. Additives can be utilized to affect the ﬂux and reduce the operating temperature which helps to save on energy consumption during the melting and sintering processes. Polettini et al. [10] for example, used feldspar residue and cullet as additives which were mixed with ﬂy ash in order to adjust the composition to the optimal range for sintering. The blend was sintered at 1100 and 1150 ◦ C to obtain specimens with high compressive strength that also satisﬁed the requirements for the immobilization of some heavy metals although, the evaporation rates of Pb, Cd and Zn were still very high. Zhang et al. [11] used ﬂy ash as an additive for the production of ceramic tile. Although they did not analyze the evaporation of heavy metals during thermal treatment, the compressive strength of sintered specimens did meet the required standard. Furthermore, when 20% ﬂy ash was added before sintering at 960 ◦ C, the leaching of the heavy metals could meet the horizontal vibration extraction procedure (HVEP) standard. The main factor leading to a large amount of heavy metal evaporation is a high chloride content in the ﬂy ash [1–4]. The presence of chloride and other soluble salts such as sylvite and sodium is also detrimental to the durability of sintered specimens. In short, the chemical composition of the ﬂy ash is one of the critical parameters in sintering technology. Water extraction is a common method for the removal of easily soluble salts which is needed in order to avoid the development of an unstable structure in the sintered specimen [12–14]. However, it is necessary to increase the sintering temperature after water extraction because of the removal of soluble salts such as NaCl and KCl which produce ﬂux that is good for reducing sintering temperature.

According to sintering theory, the smaller the size of the particles the better the sintering action will be [15]. Milling can be used to decrease particle size and to facilitate the mixing process [9–11]. Particle size after milling is determined by sieving [13]. However, in past studies, milling has only been used as a pretreatment prior to sintering. The relationship between variations in particle characteristics and the effects produced by milling on sintering have not been investigated. De Casa et al. [14] did investigate the effects of milling on the physical–mechanical properties of sintered municipal incinerator ﬂy ash processed by washing, dry milling and sintering to manufacture ceramic materials. They found that milling could reduce the ﬁring temperature of milled-washed ﬂy ash from 1210 ◦ C to 1140 ◦ C, as well as facilitate the immobilization of heavy metals. In recent years, milling technology has been used to help stabilize heavy metals in the ﬂy ash. Both dry milling and wet milling can effectively decrease the release of heavy metals. For example, Nomura et al. [16] discovered that the dry milling (with a planetary ball mill) of a mixture of MSWI ash and calcium oxide can reduce heavy metal leaching. In another study they also found that milling could reduce the leaching effects of lead in simulated polluted soil [17]. These two studies are indicative of the potential of the milling technology for stabilizing heavy metals in solid phase materials. Li et al. [18] found that wet milling helped to stabilize Pb in MSWI ash. Sun et al. [19] found that milling increased the stabilization of Pb in a phosphoric acid MSWI ﬂy ash solution. In the present study, the chemical composition of the samples was adjusted by the addition of water treatment plant sludge (WTS) and cullet to improve the sintering characteristics. Wet milling was used to decrease the particle size and destroy the crystalline structure of the material so as to decrease the sintering temperature, for the manufacture construction materials. The crystalline

Table 2 The TCLP leaching concentrations of the experimental materials (Unit, mg/L). Reaction asha Cd Cr Cu Pb Zn Final pH

0.01 0.09 1.49 101.04 3.93 12.78

± ± ± ± ± ±

0.00 0.00 0.05 1.12 0.04 .03

Extracted ashb

WTSc

ND 0.16 ± 0.01 0.20 ± 0.19 6.85 ± 0.31 1.69 ± 0.55 12.76 ± 0.01

0.02 3.10 0.28 1.91 1.72 3.65

± ± ± ± ± ±

0.01 0.01 0.05 0.61 0.09 0.08

Cullet

Milled ash

ND ND 0.01 ± 0.01 0.01 ± 0.01 0.03 ± 0.02 7.88 ± 0.08

0.07 0.33 0.43 0.09 0.99 8.10

Mean ± standard deviation (n = 3). ND, not detectable, detection limit Pb = 5ppb,Zn = 0.3ppb,Cu = 0.6ppb,Cd = 0.35ppb,Cr = 0.5ppb. a Raw reaction ﬂy ash. b Water-extracted ﬂy ash. c Water treatment sludge.

± ± ± ± ± ±

0.01 0.01 0.01 0.01 0.01 0.01

Regulation limits 1 5 15 5 –

C.-J. Sun et al. / Journal of Hazardous Materials 195 (2011) 281–290

100

4 12

Cumulative Percent (%)

Milled 80 60

283

8. CaSO4 · 0.5 H2O 9. (Mg0.03Ca0.97)CO3 10. (Mg, Al)6(Si, Al)4O10(OH)8 11. K-Mg-Fe-Al-Si-O-H2O 12. KAl2(AlSi3O10)(OH)2 13. KAl2(Si3Al)O10(OH)2

1. CaClOH 2. KCl 3. NaCl 4. SiO2 5. CaSO4 6. CaOH2 7. CaCO3

WTS Cullet Extracted

40 20 0 0.1

1

10

100

1000

Particle Size (μm)

12

4 11 12 10

10

10

10 12

Water treatment sludge 11 12

12

4

4

4

4

12

4

4

11

4 13

Fig. 1. Cumulative curve showing results for experimental materials and milled powder (milled: milled mixed ﬂy ash; WTS: water treatment sludge; extracted: water-extracted ﬂy ash).

Milled powder

structure, microstructure, heavy metal leaching and evaporation, and mechanical characteristics of the sintered specimens were studied in order to understand the effects of the recycling process and to explore the mechanism of heavy metal stabilization.

10

13 10

2.1. Experimental materials

8

2.2. Experimental methodology The ﬂy ash was extracted with deionized water before the addition of the WTS and the cullet. For a description of the process refer to Chuang [20]. The process was carried out twice, with a liquid to solid ratio of 5 and an extraction time of 5 min. After the water extraction process the ﬂy ash becomes extracted ﬂy ash. The ratios of extracted ﬂy ash, WTS and cullet in the mixture were 40%, 30% and 30%, respectively [21]. The mixed ingredient is called the mixed ﬂy ash.

Elements

1 h of milling time

Pb Zn Cd Cu Cr

2.86 2.07 0.05 0.02 0.04

0.03 0.05 0.01 0.01 0.01

96 h of milling time 0.05 0.06 0.01 0.07 0.01

7 4 74

13

4

4

7 8 9

9

8 6 5 4 4 1 2

10

15

20

25

30

5 67 9 6 8

7 7 9 7 9 5 5 9

8 5

Water-extracted fly ash 6

1 3

Raw fly ash

1 1 4 5 37

1

5

6

2 4

1 4 7

35

40

31 1 2 4

45

50

13

55

11

60

1 3 2 4

2 3

65

70

75

3 4

80

85

2θ Fig. 2. XRD patterns for water-extracted ﬂy ash, water treatment sludge, cullet and milled blend.

The liquid to solid ratio during the wet milling of the mixed ﬂy ash was 9, the rotational speed was 93 rpm and the milling time was 1 h; for further details please refer to Li et al. [18]. After milling, the thick mixed ﬂy ash liquid was ﬁltered, then dried and pelleted at 34,474 kPa (5000 psi). The pellets were cylindrical in shape with a diameter of 20.5 mm and about 22.0–27.7 mm high. An electro-thermal rectangular furnace was used in the experiments. The heating rate was 20 ◦ C/min. The sintering times for specimens processed without and with milling were 900, 950 and 1000 ◦ C and 850, 900, 950 and 1000 ◦ C, respectively. The sintering time was 1 h. The evaporation rate of heavy metals during the sintering process can be calculated using the formula below E(%) =

(W1 × C1 ) − (W2 × C2 ) × 100% W1 × C1

(1)

where E (%) is the evaporation rate; W1 (kg) is the weight of the specimen before sintering; W2 (kg) is the weight of the specimen after sintering; C1 (mg/kg) is the concentration in the specimen Table 4 Heavy metal content in the experimental samples (Unit: mg/kg).

Table 3 Heavy metal concentration of the milling solution (Unit: mg/L).

Ave ± SD sample number: 3.

10 13

6

Fly ash was collected from a large MSW incineration plant in northern Taiwan. On average, 1350 metric tons are disposed of each day by this plant. Fly ash is generally comprised of boiler ash, cyclone ash or reaction ash depending on the air pollution control devices (APCD) position. Reaction ash sampled from the fabric baghouse ﬁlter behind the semi-dry scrubber system was used in these experiments. The composition of the ﬂy ash was adjusted by the addition of WTS and cullet. The WTS was collected from a water treatment plant, also in northern Taiwan. The sludge generation process included thickening and dewatering to form sludge cakes which were sun-dried in a ﬁeld. The sludge cakes were then crumbled, dried and stored for future use. The cullet powder was produced from achromatic, transparent glass vessels which had been soaked in a strong nitric acid solution, washed in de-ionized water, broken into pieces, and ﬁnally crushed in a jaw crusher. The pieces were gathered and sifted through a No. 150 screen sieve.

4 13 7 4 7

Cullet

2. Materials and methods

± ± ± ± ±

7

13 10 4

± ± ± ± ±

0.01 0.01 0.00 0.01 0.00

Elements

Before milling

Pb Zn Cd Cu Cr

1100 7400 74 590 260

Ave ± SD sample number:3.

± ± ± ± ±

41 260 0.95 15 9.3

After 1 h of milling 1100 8100 93 590 260

± ± ± ± ±

76 430 5.4 26 28

284

C.-J. Sun et al. / Journal of Hazardous Materials 195 (2011) 281–290

Table 5 The TCLP results of the sintered specimens for different sintering temperatures with and without milling (Unit, mg/L). Element

Cd Cr Cu Pb Zn pH

850 ◦ C

900 ◦ C

With milling

Without milling

0.14 0.80 0.16 0.39 1.36 5.07

± ± ± ± ± ±

0.01 0.00 0.00 0.00 0.06 0.01

± ± ± ± ± ±

0.27 0.76 0.10 0.11 1.82 5.40

950 ◦ C With milling

0.01 0.03 0.00 0.01 0.01 0.02

0.13 0.57 0.26 0.38 1.23 4.81

± ± ± ± ± ±

1000 ◦ C

Without milling

0.01 0.00 0.00 0.00 0.00 0.02

0.47 0.48 0.14 0.22 1.49 5.21

± ± ± ± ± ±

With milling

0.02 0.01 0.01 0.02 0.07 0.02

0.13 0.55 0.25 0.37 1.21 4.70

± ± ± ± ± ±

Regulation limits

Without milling

0.01 0.03 0.03 0.02 0.01 0.01

0.95 0.49 0.17 0.24 1.63 4.94

± ± ± ± ± ±

With milling

0.03 0.01 0.00 0.03 0.19 0.02

± ± ± ± ± ±

0.10 0.32 0.19 0.34 1.13 4.43

0.01 0.06 0.00 0.03 0.21 0.02

1 5 15 5 – –

Mean ± standard deviation (n = 3), ND, not detectable, detection limit Pb = 5ppb,Zn = 0.3ppb,Cu = 0.6ppb,Cd = 0.35ppb,Cr = 0.5ppb.

before sintering; C2 (mg/kg) is the concentration in the specimen after sintering.

3. Results and discussions 3.1. Characteristics of the material

2.3. Analysis The size of the particles was analyzed with a laser particle size analyzer (Honeywell Microtrac X-100). The leaching concentration of heavy metals was extracted using the toxicity characteristic leaching procedure (TCLP) USEPA method 1311. After digestion of the samples using the alkali fusion method, the heavy metal content and chemical composition were analyzed by inductively coupled plasma atomic emission spectroscopy (ICP-AES; JOBINYVON JORIBA, Ultima-2000). X-ray diffraction (XRD; Bruker D8A) was used to identify the crystallographic structure of the ﬂy ash during the different stages. The microstructure of the surface of the ﬂy ash particles was observed by scanning electron microscopy (SEM; Leo 1530). The water absorption rate, soundness test, and compressive strength of the sintered specimens were analyzed by the CNS 488, CNS 1167 and NIEA R206.20T methods, respectively.

(a)

1. Quartz 2. Labradorite 3. Albite 4. Anhydrite 5. Diopside

9 5

Details about the composition of the materials are shown in Table 1. The largest elementary component in the reaction ﬂy ash is calcium (Ca) followed by are potassium (K) and sodium (Na). A water extraction process was used to remove soluble salts such as KCl and NaCl from the MSWI ﬂy ash, which could be helpful in long-term stabilization of the structure of the sintered specimens but also caused the ratio of some elements in the ﬂy ash to increase. For example, the Ca content reached 40%. The change in the composition of the extracted ﬂy ash necessitates the raising of the sintering temperature to obtain specimens with the same mechanical characteristics. The sintering temperature can be reduced by adjusting the composition of the ﬂy ash, with the addition of cullet and WTS. Since the major elements in the cullet and the WTS are silicon and aluminum they can increase the Si and Al concentration and decrease the Ca concentration for suitable sintering of the material.

6. Gehlenite 7. Anorthite 8. Augite 9. Wollastonite 10. Hauyne

2 3 5 7 8 9 2 5 3 4 8 72 1 3 9 4 1 7

10 4 9 9 1 5 4 9 10

1

o

1000 C 1 5 9 9 9 10 4 4

1 9

9

9

2 3 7

9 5 6 10 1 4 9 1 9

4 6 6 99 5

2 3 4 72 3 1 7

o

950 C 1 5 10 9

10

1 9 6

9 4 4

9

1

1

6. Gehlenite 7. Anorthite 8. Augite 9. Wollastonite 10. Hauyne

1. Quartz 2. Labradorite 3. Albite 4. Anhydrite 5. Diopside

(b)

5 7 8 5 8

o

1000 C 5 7 4 4 8

1 9

1

5 8 9

5 8 9 4

5 7 8 5 8

o

900 C 5 5 8 7 4 4 8

1 9 5

5 8

1

1

5 8 1

1

2 3

4 4 9 6 10 1 9 10

5

15

25

96 o

1

900 C

9

1 9

6 9 10

35

1 1 6 9

9 4 4

45

2-Theta (o)

4

1 6

1

55

3 2 6 3

65

75

5

15

25

5 6

4 6

o

850 C

2 5 5 6 6 5 4 4

35

1 5 6

45

2-Theta (o)

Fig. 3. XRD patterns for sintered specimens: (a) without milling; (b) with milling.

1 6

1

55

65

75

C.-J. Sun et al. / Journal of Hazardous Materials 195 (2011) 281–290

285

Fig. 4. SEM observations of sintered specimens without milling treatment: (a) 900 ◦ C, 1 k×; (b) 900 ◦ C, 10 k×; (c) 950 ◦ C, 1 k×; (d) 950 ◦ C, 10 k×; (e) 1000 ◦ C, 1 k×; (f) 1000 ◦ C, 10 k×.

The TCLP concentrations of the three materials are shown in Table 2. Water extraction has removed some of the easily releasable Pb from the ﬂy ash, which causes a steep decrease in the leaching concentration from 101.04 to 6.85 mg/L. However, this leaching concentration still does not conform to Taiwan’s regulatory limits (5 mg/L). Further treatment of the ﬂy ash for stabilization of the Pb is thus necessary. The leaching concentrations of heavy metals from the cullet and the WTS all do fall within the regulatory limits. 3.2. Change in the characteristics after milling The cumulative curves of the particle size for the extracted ﬂy ash, cullet, WTS and milled mixed ﬂy ash are shown in Fig. 1. The d50 of the extracted ﬂy ash, WTS and cullet are 10.65, 7.87 and 45.84 ␮m, respectively. The extracted ﬂy ash and WTS can be classiﬁed as ﬁne powders. After 1 h of milling, the d50 of the mixed ﬂy ash decreases to 5.11 ␮m. Smaller sized particles have higher surface energies which affects the heat needed for sintering. The milling process has the beneﬁt of making the size of the mixed ﬂy ash particles uniform thereby helping to produce stable sintered specimens.

The XRD spectra of the reaction ﬂy ash, extracted ﬂy ash, cullet, WTS and milled mixed ﬂy ash are shown in Fig. 2. The main compounds in the reaction ﬂy ash are CaClOH, CaSO4 , KCl, NaCl and SiO2 . After water extraction, the CaClOH, NaCl and KCl became dissociated so that the chlorine (Cl), sodium (Na), potassium (P) and some of the calcium (Ca) can be extracted into the solution. The rest of the calcium combines with carbonates or hydroxyls to form calcium carbonate or calcium hydroxide which then precipitates. The lower solubility of CaSO4 makes water extraction difﬁcult which is why it can be identiﬁed in the extracted ﬂy ash. The main mineral found in the WTS is quartz. The extraordinarily obvious main peak in the XRD spectrum is indicative of the crystalline phase. The other main minerals, mica, chlorite and serpentine, are silicate minerals, commonly found in Taiwan. Examination of the cullet spectrum shows no obvious peak, but rather the gently rising curve typical of the XRD spectra of amorphous material. Comparison of the three materials shows no identiﬁable compounds in the XRD spectra of the milled mixed ﬂy ash. The identiﬁable compounds are calcium carbonate from the extracted ﬂy ash, mica from the water treatment plant sludge, and mineral quartz from both the extracted ﬂy ash and the WTS. However, after milling the mixed ﬂy ash for 1 h, the height of all of the peaks decreased. For example, the peak for

286

C.-J. Sun et al. / Journal of Hazardous Materials 195 (2011) 281–290

Fig. 5. SEM observations of sintered specimens with milling treatment: (a) 900 ◦ C, 1 k×; (b) 900 ◦ C, 10 k×; (c) 950 ◦ C, 1 k×; (d) 950 ◦ C, 10 k×; (e) 1000 ◦ C, 1 k×; (f) 1000 ◦ C, 10 k×.

quartz in 2 = 26.7 decreased 40% after 1 h of milling. Clearly, the mechanical energy from the milling was not sufﬁcient to allow the formation of new compounds, but did increase the surface energy of the particles which acted to increase defects in the crystalline structure, which in turn allowed for a decrease in the energy required for sintering and led to the easier formation of different compounds after sintering. Table 3 shows the concentration of the heavy metals in the milling solution. The results indicate that the concentration of heavy metals decreased to less than 2.86 mg/L and 0.07 mg/L after 1 h and 96 h of milling time, respectively. It can be seen that little heavy metal leached into the milling solution. Table 4 shows an increase in the concentration of the heavy metals in the milled ﬂy ash, because of the dissolution of some materials into the milling solution. The TCLP concentration of Pb in the ﬂy ash could be reduced by milling, as shown in Table 2. This result has been described in previous studies [18,19]. In this study, it is found that the TCLP concentrations of heavy metals, with the exception of Cu, were less than one tenth the regulatory limits, as shown in Table 2. This conﬁrms that the milling process is an effective method for stabilizing Pb, Cr, and Zn present in the ﬂy ash.

3.3. Crystalline phase and microstructure of the sintered specimens XRD patterns of sintered specimens produced with and without milling are shown in Fig. 3. The intensity of the XRD patterns of the quartz crystal decrease with increasing sintering temperature. After sintering at 900 ◦ C and 1000 ◦ C, the main peaks were from 153 to 88 without milling and from 124 to 78 with milling, respectively. At a high sintering temperature, the quartz will react with other elements (such as Ca and Si) to produce other minerals, as indicated by the lessening of the main quartz peak. Wollastonite, a calcium inosilicate mineral, and diopside, a monoclinic pyroxene mineral were the major minerals that formed in the specimens without milling. All of these materials have a simple chain-like structure, as shown in Fig. 3a. The next most common mineral to form from the Na, Al and Ca was hauyne, of the tectosilicates, and gehlenite, of the sorosilicates. In milled specimens sintered at temperatures above 900 ◦ C, we see the formation of albite, anorthite and labradorite of the feldspar group of tectosilicates (see Fig. 3b) followed by augite and diopside of the inosilicate group.

C.-J. Sun et al. / Journal of Hazardous Materials 195 (2011) 281–290

10

16

0

Volume change (%)

Weight loss (%)

(b)

(a)

15 14 13 12 11 10 850

900

950

-10 -20 -30 -40 -50 850

1000

900

Temperature (ºC)

1000

20

(c) Water absorption rate (%)

2.5

Density (g/cm3)

950

Temperature (ºC)

3.0

2.0 1.5 1.0 0.5 0.0 850

900

950

(d) 16 12 8 4 0 850

1000

900

Temperature (ºC)

950

1000

Temperature (ºC)

400

20

(e)

(f) 16

300

Soundness (%)

Compressive strength (kg/cm2)

287

200

100

8 4

0 850

12

900

950

1000

Temperature (ºC)

0 850

900

950

1000

Temperature (ºC)

Fig. 6. The physical and mechanical properties of sintered specimens produced with and without milling treatment, where hollow circles indicate non-milled, full circles indicate milled: (a) weight loss; (b) volume change; (c) density; (d) water absorption rate; (e) compressive strength; (f) soundness.

In other words, the milling process affected the type and hardness of the material that formed in the sintered specimens. All of the aforementioned minerals are important rock-forming silicate minerals. Albite, labradorite and anorthite belong to the tectosilicate group, having a hardness rating of between 6 and 6.5, comprising the most important rock-forming minerals on earth. The hardness of the augite and dispside ranges from 5 to 6 and that of the wollastonite from 4.5 to 5. Fig. 3 shows that, with suitable adjustment, milling and sintering could produce mineral substances with the potential to form specimens with high compressive strength. SEM examination of the microstructure of the sintered specimens shows an obvious difference between those produced with and without milling, for a variety of sintering temperatures (Figs. 4 and 5). Fig. 4 shows the microstructure of sintered specimens produced without milling. At magniﬁcations of 1 k and 10 k,

the temperature effect is obvious. At a sintering temperature of 900 ◦ C, the particles in the mixed ﬂy ash are basically pushed together, not displaying the characteristics of the sintered state. In the specimens sintered at 950 ◦ C (Fig. 4c), it can be seen that the particles have gradually become connected and bound together, but that pores still exist inside the specimens. Multiple zoom images of the specimens shown in Fig. 4d conﬁrm the appearance of the liquid sintering state in some places and some imprinted crystalline shapes on the fractured particle surfaces, indicative of the generation of some crystals at this temperature. When the sintering temperature is 1000 ◦ C, the liquid sintering state around the powdered particles grows to form larger particles, although there are still some pores between these large particles. Fig. 5 shows the microstructure of the milled and sintered specimens. It is very clear that liquid sintering has occurred. The fractured particle surfaces

288

C.-J. Sun et al. / Journal of Hazardous Materials 195 (2011) 281–290

80

80

(Cr)

Evaporation rate (%)

Evaporation rate (%)

(Cd) 60

40

20

0 850

900

950

60

40

20

0

1000

850

Temperature (ºC) 80

950

1000

80 (Pb)

Evaporation rate (%)

(Cu)

Evaporation rate (%)

900

Temperature (ºC)

60

40

20

0 850

900

950

1000

Temperature (ºC)

60

40

20

0

850

900

950

1000

Temperature (ºC)

80

Evaporation rate (%)

(Zn) 60

40

20

0 850

900

950

1000

Temperature (ºC) Fig. 7. Evaporation rate of heavy metals from sintered specimens produced with and without milling treatment where hollow circles indicate non-milled and full circles indicate milled.

are smoother than was the case without milling (Fig. 5a). It can be seen in Fig. 5b that the particles have moved together and are covered with amorphous material. In this liquid sintering stage, it is not easy to discriminate between single particles. This does not mean that all the particles have become completely amorphous; it simply shows that the space between the particles is now ﬁlled with amorphous material. Surface fracturing always occurs where particles in the specimen are cracked by force. When the sintering temperature reaches 950 ◦ C and 1000 ◦ C, the amount and dimensions of the amorphous material and closed pores in the specimens will increase. Comparison of specimens produced with and without milling shows that the milling process can increase the quantity of amorphous material during sintering. This occurs because of the

destruction of particles in the powder allowing for redistribution of the crystalline structure [18]. Milling reduces the elementary binding forces and increases the surface energy of the particles, which tends to increase the formation of amorphous material at lower temperatures, and enhance the mechanical characteristics of the sintered specimens. This contributes to the early formation of more complex structures (e.g., tectosilicates), as shown in the XRD patterns discussed above. In other words, a suitable amount of amorphous material is helpful for the formation of suitable sintered specimens, however, if the amount of amorphous material is too great, the strength of the specimen is insufﬁcient and the original shape of the sintered specimen will collapse. In this study, when the sintering temperature reaches 1050 ◦ C, the milled and sintered specimen are not no longer cylindrical in shape.

C.-J. Sun et al. / Journal of Hazardous Materials 195 (2011) 281–290

3.4. Effect of milling on the sintered specimens The characteristics of the sintered specimens are shown in Fig. 6. The weight loss of sintered specimens, with and without milling was 13% and 14–15%, respectively (see Fig. 6a). The data show the weight loss to be very similar possibly due to the decomposition of some of the calcium carbonate and calcium hydroxide in the extracted ﬂy ash at high sintering temperatures (see Fig. 2). The shrinkage of the milled specimens increased considerably when the sintering temperature reached 1000 ◦ C (see Fig. 6b). When the temperature rose to 1050 ◦ C, the shape of the milled sintered specimens collapsed. In Fig. 5, it can be seen that the amount of amorphous material increases with the temperature, indicating that a surfeit of liquid phase material would be harmful for recycling. The amorphous material eliminates open pores between the particles in the milled and sintered specimen. Thus, the density increases and the water absorption rate decreases with an increase in temperature as well (see Fig. 6c and d). The compressive strength is an important factor affecting the mechanical characteristics of sintered specimens. As shown in Fig. 6e, the compressive strength of specimens produced without milling and sintered at 900 ◦ C is less than 100 kg/cm2 . The compressive strength increases with the sintering temperature, reaching 389 kg/cm2 (38.2 MPa) when the temperature reaches 1000 ◦ C. However, if the powder is treated by milling, the compressive strength rises to 300 kg/cm2 (29.4 MPa) at 850 ◦ C, and 389 kg/cm2 at 1000 ◦ C. Comparison of the results shows that milling can help to produce suitable specimens at lower sintering temperatures. The formation of complex tectosilicates and amorphous material are the main factors leading to the increased compressive strength, ﬁrmly binding the particles together; see Figs. 3 and 5. Clearly, milling can improve the soundness of sintered specimens, as shown in Fig. 6f. Table 5 shows the TCLP leaching concentrations of sintered specimens produced with and without milling as well as the pH of the ﬁnal extraction solution. The pH decreases with both sintering temperature and milling. The TCLP leaching concentrations of Cd, Cu, and Pb increase slightly with increased temperature in the specimens produced without milling. However, in the milled specimens, the amount decreases slightly with increased temperature, although in all cases, amounts are lower than the regulatory limits. The pH ranges between 4.43 and 5.40, an interval where heavy metals are easily released, but the TCLP concentrations of all the sintered specimens (especially the milled specimens) fall within the regulatory limits. 3.5. Restraining heavy metal evaporated by milling As noted above, the evaporation of heavy metals is affected by the composition of the ﬂy ash. For example, chlorides combine easily with heavy metals during the sintering process, which could lead to an increase in their evaporation rate. Water extraction is an efﬁcient method to remove these chlorides from the ﬂy ash. However, the main chloride compounds, KCl and NaCl, also produce good ﬂux during sintering, which could improve the sintering characteristics. How to restrain heavy metal evaporation from the ﬂy ash during sintering or other types of thermo-treatment while keeping the temperature as low as possible is an important question. Fig. 7 shows the evaporation rates of the heavy metals, Cd, Cr, Cu, Pb and Zn, from sintered specimens, fabricated with and without milling. For specimens sintered at 900–1000 ◦ C without milling, the evaporation rates of Pb, Cd, Cu, Cr and Zn are 54–64%, 43–49%, 38–43%, 30–40% and 14–35%, respectively. The specimens in these experiments were produced through a pelleting process, so the evaporation rate was lower than that of powdered ﬂy ash without pelleting. The demands for the compressive strength and heavy

289

metal evaporation are often in conﬂict. For example, although increasing the sintering temperature will improve the compressive strength, the heavy metal evaporation rate will also increase, which raises the environmental risk of secondary pollution by heavy metals. On the other hand, although a lower sintering temperature can reduce the evaporation rate of the heavy metals, the mechanical characteristics of the sintered specimens might not meet the standards required for recycled building materials. In this respect, our results show the heavy metal evaporation rate of sintered specimens produced with milling to be better than those produced without. For example, after sintering at 850–1000 ◦ C, the evaporation rates of Pb, Cd, Cu, Cr and Zn are 19–21%, 19–21%, 14–19%, 12–19% and 14–17%, respectively. Obviously, the addition of milling to the recycling process could efﬁciently reduce the evaporation of heavy metals from the specimens. When the sintering temperature reached 1000 ◦ C, there was a decrease in the amount of Pb, Cd, Cu, Cr and Zn evaporation of 43%, 22%, 24%, 21% and 18%, respectively. In this study, it is found that the liquid phase in milled and sintered specimens was generated at 900 ◦ C (see Fig. 5). At this time, the surface of the ﬂy ash powder was covered by amorphous material, sealing the heavy metals within the samples. The evaporation rate of the milled sintered specimens increased little with the sintering temperature, as shown in Fig. 7. This reveals that the amorphous material helped to seal the heavy metals within the sintered specimens, and was not affected by the sintering temperature. Therefore, the milling process can effectively restrain heavy metal evaporation during sintering. 4. Conclusions The milling process acts to destroy crystalline structures in the ﬂy ash. Compounds recombine to form new ones during the sintering process, as evidenced by the formation of the tectosilicates, albite, labradorite and anorthite in the experiments. Such compounds offer a structural foundation to the sintered specimens. Our experimental results show that amorphous materials are more easily generated at lower temperatures during the liquid sintering stage in milled sintered specimens than in specimens produced without milling. The amorphous material helps to combine the particles in the ﬂy ash together which affects the mechanical characteristics of the sintered specimens, leading to increased compressive strength, soundness, density, and shrinkage, and decreased water adsorption rate and weight loss. If the material generated is sufﬁcient to cover the surfaces of the particles, it can also restrain the evaporation of heavy metals from the ﬂy ash during the heating process. Therefore, the milling process can help to stabilize heavy metals in the ﬂy ash, improve the compressive strength, and restrain heavy metal evaporation from sintered specimens. Acknowledgement The authors would like to thank the National Science Council of Taiwan, ROC, for their ﬁnancial support of this research under contract number (NSC- 92-2211-E-032-014). References [1] A. Jakob, S. Stucki, P. Kuhn, Evaporation of heavy metals during the heat treatment of municipal solid waste incinerator ﬂy ash, Environ. Sci. Technol. 29 (1995) 2429–2436. [2] C. Chan, C.Q. Jia, J.W. Graydon, D.W. Kirk, The behaviour of selected heavy metals in MSW incineration electrostatic precipitator ash during roasting with chlorination agents, J. Hazard. Mater. 50 (1996) 1–13. [3] L.L. Forestier, G. Libourel, High temperature behavior of electrostatic precipitator ash from municipal solid waste combustors, J. Hazard. Mater. 154 (2008) 373–380.

290

C.-J. Sun et al. / Journal of Hazardous Materials 195 (2011) 281–290

[4] H.C. Ho, J.D. Chow, S.H. Gau, Thermal mobility of heavy metals in municipal solid waste incinerator ﬂy ash (MSWIFA), Environ. Eng. Sci. 25 (2008) 649–656. [5] Y. Yang, Y. Xiao, J.H.L. Voncken, N. Wilson, Thermal treatment and vitriﬁcation of boiler ash from a municipal solid waste incinerator, J. Hazard. Mater. 154 (2008) 871–879. [6] Y. Yang, Y. Xiao, N. Wilson, J.H.L. Voncken, Thermal behaviour of ESP ash from municipal solid waste incinerators, J. Hazard. Mater. 166 (2009) 567–575. [7] Y.J. Park, J. Heo, Vitriﬁcation of ﬂy ash from municipal solid waste incinerator, J. Hazard. Mater. 91 (2002) 83–93. [8] T.W. Cheng, Y.S. Chen, Characterisation of glass ceramic made from incinerator ﬂy ash, Ceram. Int. 30 (2004) 343–349. [9] K.L. Lin, Feasibility study of using brick made from municipal solid waste incinerator ﬂy ash slag, J. Hazard. Mater. 137 (2006) 1810–1816. [10] A. Polettini, R. Pomi, L. Trinci, A. Muntoni, S. Lo Mastro, Engineering and environmental properties of thermally treated mixtures containing MSWI ﬂy ash and low-cost additives, Chemosphere 56 (2004) 901– 910. [11] H. Zhang, Y. Zhao, J. Qi, Study on use of MSWI ﬂy ash in ceramic tile, J. Hazard. Mater. 141 (2007) 106–114. [12] T. Mangialardi, Sintering of MSW ﬂy ash for reuse as a concrete aggregate, J. Hazard. Mater. 87 (2001) 225–239.

[13] K.S. Wang, K.Y. Chiang, K.L. Lin, C.J. Sun, Effects of a water-extraction process on heavy metal behavior in municipal solid waste incinerator ﬂy ash, Hydrometallurgy 62 (2001) 73–81. [14] G. De Casa, T. Mangialardi, A.E. Paolini, L. Piga, Physical–mechanical and environmental properties of sintered municipal incinerator ﬂy ash, Waste Manage. 27 (2007) 238–247. [15] R.M. German, Sintering Theory Practice, Wiley Interscience Publication, New York, 1996. [16] Y. Nomura, T. Okada, S. Nakai, M. Hosomi, Inhibition of heavy metal elution from ﬂy ashes by mechanochemical treatment and cementation, Kagaku Kogaku Ronbun 32 (2006) 196–199. [17] S. Montinaro, A. Concas, M. Pisu, G. Cao, Remediation of heavy metals contaminated soils by ball milling, Chemosphere 67 (2007) 631–639. [18] M.G. Li, C.J. Sun, S.H. Gau, C.J. Chuang, Effects of wet ball milling on lead stabilization and particle size variation in municipal solid waste incinerator ﬂy ash, J. Hazard. Mater. 174 (2010) 586–591. [19] C.J. Sun, M.G. Li, S.H. Gau, C.J. Chuang, Effect of the milling solution on lead stabilization in municipal solid waste incinerator ﬂy ash during the milling processes, Waste Manage. 31 (2011) 318–324. [20] C.J. Chuang, M.D. Thesis, University of Tamkang, Tamsui, Taiwan, R.O.C., 2006 (in Chinese). [21] Y.H. Wang, M.D. Thesis, University of Tamkang, Tamsui, Taiwan, R.O.C., 2009 (in Chinese).

Journal of Hazardous Materials 195 (2011) 291–297

Contents lists available at SciVerse ScienceDirect

Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat

Tantalum (oxy)nitrides: Preparation, characterisation and enhancement of photo-Fenton-like degradation of atrazine under visible light Yingxun Du ∗ , Lu Zhao, Yaling Su Nanjing Institute of Geography and Limnology, State Key Laboratory of Lake Science and Environment, Chinese Academy of Sciences, Nanjing 210008, China

a r t i c l e

i n f o

Article history: Received 22 May 2011 Received in revised form 16 July 2011 Accepted 13 August 2011 Available online 22 August 2011 Keywords: Tantalum (oxy)nitrides Photo-Fenton-like Fe3+ reduction Visible light Atrazine

a b s t r a c t Tantalum (oxy)nitrides were prepared by the nitridation of Ta2 O5 and were added to a photo-Fentonlike system to enhance Fe3+ reduction and atrazine degradation under visible light. The samples were characterized by XRD, XPS, DRS and BET analyses. XPS analysis showed that the nitrogen content of the tantalum (oxy)nitride samples increased noticeably with the nitridation temperature and nitridation time but slightly with the ﬂow rate of NH3 . XRD results showed Ta2 O5 was ﬁrst converted to TaON and then to Ta3 N5 when the nitridation temperature increased. DRS analysis showed that the sample obtained at 800 ◦ C displayed the strongest absorption of visible light. However, the ability of the tantalum (oxy)nitrides to reduce Fe3+ did not increase continuously with the nitrogen content. Sample 7 (700 ◦ C, QNH3 = 0.3 L/ min, 6 h) showed the highest level of photocatalytic activity for Fe3+ reduction. This is because the photocatalytic activity of TaON for Fe3+ reduction is higher than that of Ta3 N5 . And a slight synergetic effect was observed between TaON and Ta3 N5 . With the addition of sample 7, H2 O2 decomposition and atrazine degradation were signiﬁcantly accelerated in a photo-Fenton-like system under visible light. The regenerated tantalum (oxy)nitrides catalyst displayed considerably stable performance for atrazine degradation. © 2011 Elsevier B.V. All rights reserved.

1. Introduction photo-Fenton-like (Fe3+ /H2 O2 ) and photo-Fenton systems constitutes one kind of the most attractive advanced oxidation processes, as the materials required are relatively abundant, inexpensive, and environmentally benign. In photo-Fenton-like and photo-Fenton system, the degradation of the contaminant is based on the attack of the active hydroxyl radical (• OH), which is generated by the reaction of Fe2+ and H2 O2 (Eqs. (1) and (2)). A variety of refractory organics such as chlorinated phenols, herbicides and dyes can be decomposed effectively by the photo-Fenton(-like) reactions [1–6]. In photo-Fenton(-like) reaction, the cycling of the catalyst (Fe2+ /Fe3+ ) is key to the effective degradation of the contaminant. Eqs. (1) and (3) showed the cycling of iron ions driven by H2 O2 . Fe2+ is oxidized by H2 O2 with a high rate constant. But the reduction of Fe3+ by H2 O2 is slow and thus ineffective. In photo-Fenton(like)processes, the efﬁcient way to reduce Fe3+ is the irradiation of UV light ( < 360 nm) (Eq. (4)) [7]. However, the industrial appliThe

(Fe2+ /H2 O2 )

∗ Corresponding author. Tel.: +86 025 86882116; fax: +86 025 57714759. E-mail address: [email protected] (Y. Du). 0304-3894/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.jhazmat.2011.08.042

cability of this process is limited since natural UV comprises only 3–5% of the solar light energy that reaches the earth. Fe2+ + H2 O2 → Fe3+ + • OH + OH− k• OH • OH + P −→intermediates

Fe3+ + H2 O2 → Fe2+ + HO2 • + H+

k1 = 51 M−1 s−1 [8]

(1) (2)

k3 = 0.001–0.01 M−1 s−1 [9] (3)

h

Fe(OH)2+ −→Fe2+ + • OH

(4)

Many semiconductor compounds, especially TiO2 have been used as the photocatalysts in the reactions driven by solar energy such as water splitting and photocatalytic degradation [10–13]. The modiﬁed TiO2 such as metal-doped or/and nonmetal-doped TiO2 possessed the photocatalytic activity under visible light [12,14–18]. Tantalum (oxy)nitrides (Ta3 N5 and TaON) have narrow band gaps of 2.08 and 2.4 eV [19], respectively, making them suitable as visible-light driven photocatalysts. With respect to the modiﬁed TiO2 such as Ce–N codoped TiO2 , C–N codoped TiO2 and bimetal codoped (Bi-Co and Fe-Co) TiO2 [12,15,17], Ta3 N5 and TaON have relatively simple composition and structure. It has been found that TaON and Ta3 N5 were novel photocatalysts suitable for water decomposition and pollutant degradation reactions driven by visible light [20–22]. Compared with TiO2−x Nx in the same size, Ta3 N5

292

Y. Du et al. / Journal of Hazardous Materials 195 (2011) 291–297

Table 1 Composition, BET surface area and the ability to reduce Fe3+ of the tantalum (oxy)nitride samples prepared under various conditions. No. 1 2 3 4 5 6 7 8 9 10 11

Conditions ◦

600 C, 0.1 L/min, 6 h 600 ◦ C, 0.5 L/min, 6 h 600 ◦ C, 0.3 L/min, 6 h 700 ◦ C, 0.1 L/min, 6 h 700 ◦ C, 0.3 L/min, 2 h 700 ◦ C, 0.3 L/min, 4 h 700 ◦ C, 0.3 L/min, 6 h 800 ◦ C, 0.3 L/min, 2 h 800 ◦ C, 0.3 L/min, 6 h 800 ◦ C, 0.1 L/min, 6 h 800 ◦ C, 0.5 L/min, 6 h

Ta:O:N

Speciﬁc surface area (m2 /g)

Fe2+ (mg/L)

Reduction of Fe3+ (%)

1:3.0:0.10 1:2.67:0.26 1:2.62:0.34 1:2.18:0.66 1:3.32:0.24 1:2.0:0.86 1:1.86:0.80 1:1.88:0.94 1:1.50:1.07 1:1.44:1.03 1:1.16:1.17

0.896 1.000 1.548 9.014 1.894 9.796 14.25 9.433 8.549 9.656 9.654

0.129 0.317 0.534 1.489 0.425 1.737 2.003 1.586 1.171 1.320 0.700

0.45 1.12 1.90 5.31 1.51 6.19 7.15 5.65 4.17 4.70 2.50

showed much higher photocatalytic activity for the degradation of methylene blue under visible light irradiation [20]. To improve the efﬁciency of photo-Fenton system under visible light irradiation, Wang et al. [23] introduced Ta3 N5 into the photo-Fenton system and found that Ta3 N5 was effective to promote the reduction of Fe3+ under visible light and thus the degradation of N,N-dimethylaniline and 2,4-dichlorophenol. The mechanism of iron ions cycling in the presence of Ta3 N5 was also proposed. Under visible light, Ta3 N5 is excited to form electrons in the conduction band and holes in the valence band. The photoelectrons are captured by Fe3+ to generate Fe2+ , while the generated Fe2+ is oxidized by H2 O2 immediately to regenerate Fe3+ and thus a rapid iron ions cycling is established. Tantalum (oxy)nitrides are usually obtained by the nitridation of Ta2 O5 in a NH3 ﬂow. The nitridation reaction always leads to the formation of a mixture of TaON and Ta3 N5 [19]. The composition of the prepared sample was affected by the nitridation temperature, the nitridation time and the ﬂow rate of NH3 . Thus, the preparation condition should inﬂuence the photocatalytic activity of the samples. However, to the best of our knowledge, little information on the photocatalytic activity for Fe3+ reduction of the samples prepared under various conditions is available. In this study, the tantalum (oxy)nitride samples were prepared by the nitridation of Ta2 O5 at ﬁrst. XPS and DRS analysis showed that increasing the nitridation temperature, the nitridation time and the ﬂow rate of NH3 led to the higher content of nitrogen and the stronger absorption of visible light. But it was interesting to ﬁnd that the photocatalytic activity for Fe3+ reduction is not increased continuously with the nitrogen content. This is because the photocatalytic activity of TaON for Fe3+ reduction is higher than that of Ta3 N5 , conﬁrmed by the experimental data. And a slight synergetic effect was observed between TaON and Ta3 N5 . In the photo-Fenton-like system under visible light, the degradation of atrazine was signiﬁcantly accelerated by the tantalum (oxy)nitrides sample. And the regenerated catalyst was considerably stable for atrazine degradation. 2. Experimental 2.1. Materials Ta2 O5 , Fe2 (SO4 )3 and hydrogen peroxide (30%) were purchased from Sinopharm Chemical Reagent Co., Ltd., China. NH3 (99.99%) was supplied by Special Gas Co., Ltd., Nanjing, China. Atrazine was from Tokyo Chemical Industry Co., Ltd., Japan. Deionized water was used throughout this study. 2.2. Preparation and characterisation of tantalum (oxy)nitrides For the preparation of tantalum (oxy)nitrides, 2.0 g of Ta2 O5 was put into a quartz tube, which was put into a tube furnace and then subjected to nitridation under a ﬂow of NH3 at rates

ranging from 0.1 to 0.5 L/min. The nitridation temperature ranged from 600 to 800 ◦ C, and the nitridation time was 2–6 h. After the reaction, the sample was cooled to room temperature in the ﬂow of N2 . The detailed conditions for the preparation of each sample are summarized in Table 1. The phase composition of the sample was characterized by Xray diffraction (XRD) (model D/max-rA, Rigaku Co., Tokyo, Japan), using Cu K␣ radiation. X-ray photoelectron spectroscopy (XPS) analysis was carried out on a RBD-upgraded PHI-5000C ESCA system (Perkin Elmer) with Mg K␣ radiation (h = 1253.6 eV). Binding energies were calibrated with containment carbon (C1s = 284.6 eV). The speciﬁc surface area of the powders was determined by nitrogen absorption (BET, ASAP 2020M Analyzer, Micromeritics). UV–vis diffuse reﬂectance spectra (DRS) were recorded on a UV-2401 Shimadzu spectrometer.

2.3. Photocatalytic reactions and analysis Experiments were carried out in a photoreaction apparatus, which is shown in Fig. 1. The apparatus consists of two parts. The ﬁrst part is an annular quartz tube and cooling water passes through an inner thimble of the annular tube. In the axial center of the reactor, there is a 500W xenon lamp as the light source. The wavelength of the visible light is controlled through a 400 nm cut ﬁlter. The second part is a 50 mL quartz tube, which is put on a magnetic stirrer. The reaction solution in the quartz tube was stirred during the experiments. The distance between the light source and the surface of the reaction solution is 4 cm. In the experiment, the solution of atrazine was put into the quartz tube at ﬁrst, of which pH was adjusted to 2.6 with H2 SO4 solution (VH2 SO4 : Vdeionized water = 1 : 20). Then Fe3+ stock solution (pH 2.6) and 0.03 g tantalum (oxy)nitrides were put into the tube. To achieve an adsorption/desorption equilibrium of Fe3+ and atrazine on the surface of the tantalum (oxy)nitrides, the suspension was kept in the dark and stirred for 30 min. After that, the stock solution of H2 O2 was put into the tube and at the same time the lamp was turned on to initiate the photoreaction. The samples were taken out at the desired time intervals and ﬁltered with a 0.22 ␮m Millipore ﬁlter to remove the catalyst. In the experiment of Fe3+ photoreduction, the procedure was similar except that no atrazine and H2 O2 was added. Analysis of atrazine was conducted on an Agilent 1120 compact HPLC system with a reversed phase C 18 column and UV detector. The UV detector wavelength was set at 235 nm. The mobile phase was 70:30 (v/v) of methanol and deionized water with a ﬂow rate of 1.0 mL/min. The potential reaction of atrazine with hydroxyl radical was prevented by adding 1.0 mL of 1.0 M tert-butyl alcohol to the sample. The concentration of Fe2+ was measured by the o-phenanthroline colorimetric method ( = 510 nm, ε = 1.1 × 104 M−1 cm−1 ) [24]. The concentration of H2 O2 was

Y. Du et al. / Journal of Hazardous Materials 195 (2011) 291–297

293

16000

Ta2O5 600 °C 700 °C 800 °C

a O 1s 14000 12000

N(E)

10000 8000 6000 4000 2000 0 540

537

534

531

528

525

Binding Energy (eV) 16000

b N 1s

Ta2O5 600 °C 700 °C 800 °C

14000 12000

N(E)

10000 8000 6000 4000

Fig. 1. Schematic diagram of the experimental set-up: 1. Lamp, 2. Water-cooled jacket, 3. 400 nm cut ﬁlter, 4. Quartz tube, 5. Magneton, 6. Magnetic stirrer.

determined by a spectrophotometric analysis using the potassium titanium (IV) oxalate method [25].

2000 0 398

396

394

392

Binding Energy (eV)

3. Results and discussion 395.67 eV

c N 1s 3.1. Characterisation of tantalum (oxy)nitride samples 3.1.1. XPS analysis of tantalum (oxy)nitrides XPS analysis was used to determine the composition of the tantalum (oxy)nitride samples. Table 1 summarizes the ratios of Ta:O:N in the tantalum (oxy)nitride samples prepared under different conditions. The content of nitrogen in the tantalum (oxy)nitride samples increased with increasing nitridation temperature, nitridation time and ﬂow rate of NH3 . Within the range adopted in this study, the effects of nitridation temperature and nitridation time on the nitrogen content of the samples are signiﬁcant. At nitridation temperatures of 600, 700 and 800 ◦ C (in a 0.3 L/min NH3 ﬂow for 6 h), the ratios of N to O were 9.7%, 43.0% and 71.3%, respectively. The ratios of N to O were 7.2%, 43.0% for the nitridation times of 2 and 4 h, respectively (at 700 ◦ C in a 0.3 L/min NH3 ﬂow). While the effect of NH3 ﬂow rate was relatively slight. The ratios of N to O were 30.6% and 43.0% at NH3 ﬂow rates of 0.1 and 0.3 L/min, respectively (at 700 ◦ C for 6 h). Fig. 2(a) and (b) shows the O1s and N1s XPS spectra of Ta2 O5 and the samples produced at the nitridation temperatures of 600, 700 and 800 ◦ C, respectively. The peak positions of the O1s and N1s spectra were very similar in all samples. While the peak areas of O1s and N1s changed in opposite, as the areas assigned to O1s decreased with nitridation temperature, and the areas assigned to N1s increased. Fig. 2(c) was the typical simulation of N1s peaks (sample 9 as an example). As could be seen, the N1s peak consisted of the peak with the binding energy of 395.67 eV, which was generally considered to be the evidence for the presence of Ta–N bonds when N atoms replace O atoms in Ta2 O5 crystal lattices [19]. The

398

397

396

395

394

393

392

Binding Energy (eV) Fig. 2. O1s (a) and N1s (b) XPS spectra of Ta2 O5 and the tantalum (oxy)nitride samples prepared at 600, 700 and 800 ◦ C for 6 h at a ﬂow rate of 0.3 L/min for NH3 ; N1s (c) simulation of sample 9 (preparation conditions: temperature = 800 ◦ C, QNH3 = 0.3 L/ min, time = 6 h).

replacement of the N atom by an O atom results in the band gap narrowing and a red shift of absorption for the tantalum (oxy)nitride samples [26,27]. Higher nitrogen content leads to a narrower band gap and stronger optical absorption in visible light domain. 3.1.2. XRD and DRS analyses The nitridation reaction of Ta2 O5 always leads to the formation of the mixture of TaON and Ta3 N5 . Fig. 3 depicts the XRD

294

Y. Du et al. / Journal of Hazardous Materials 195 (2011) 291–297

1.2

Ta3N5 TaON

800 °C 700 °C 600 °C

1.0

Ta2O5

Ta2O5

0.8

d Abs

0.6

c

0.4

b

0.2

0.0

a

200 10

20

30

40

50

60

70

Two-Theta (deg) Fig. 3. XRD patterns of Ta2 O5 (a) and tantalum (oxy)nitride samples prepared at 600 ◦ C (b), 700 ◦ C (c) and 800 ◦ C (d).

patterns of Ta2 O5 and tantalum (oxy)nitride samples prepared at 600, 700 and 800 ◦ C for 6 h at a NH3 ﬂow rate of 0.3 L/min. The XRD patterns showed that reaction temperature had a signiﬁcant effect on the nitridation of Ta2 O5 . With a nitridation temperature as low as 600 ◦ C, the colour of Ta2 O5 changed to yellow, indicating that the nitridation reaction was starting. However, almost none of the crystalline phase of TaON and Ta3 N5 was detected by XRD. The samples which were prepared at 700 ◦ C and 800 ◦ C both consisted of Ta2 O5 , TaON and Ta3 N5 . Compared to the sample prepared at 700 ◦ C, stronger diffraction peaks corresponding to Ta3 N5 and weaker peaks assigned to TaON appeared in the sample prepared at 800 ◦ C. During the nitridation process, tantalum in the oxidation state V can form two ionic-covalent nitride types, Ta3 N5 and TaON. TaON is regarded as the intermediate between the oxide precursor of Ta2 O5 and the fully nitrided phase of Ta3 N5 [28]. As the reaction of Ta2 O5 with NH3 to form TaON and Ta3 N5 is endothermic [29], the content of nitrogen increased with the nitridation temperature. The colours of samples prepared at 600, 700 and 800 ◦ C are yellow, orange and red, respectively. Fig. 4 shows the UV–vis diffuse reﬂectance spectra of tantalum (oxy)nitride samples prepared at 600, 700 and 800 ◦ C for 6 h in the NH3 ﬂow rate of 0.3 L/min. The UV–vis diffuse reﬂectance spectrum of Ta2 O5 is also shown. Compared with the absorption spectrum of Ta2 O5 , the tantalum (oxy)nitride samples showed an obvious red shift with a stronger absorption in the wavelength range from 400 to 800 nm. When the nitridation temperature increased, a further red shift and a stronger absorption of the tantalum (oxy)nitride samples were observed. Samples obtained at 600, 700 and 800 ◦ C showed absorption edges of approximately 520 nm, 620 nm and 640 nm, respectively, while the absorption edge of Ta2 O5 was approximately 330 nm. According to the theoretical calculation, Ta3 N5 and TaON show band gaps at 2.08 eV and 2.4 eV, respectively [19,30]. The strongest optical absorption of the sample prepared at 800 ◦ C was mainly attributed to the highest content of Ta3 N5 . 3.1.3. BET analysis BET surface areas of the samples are also given in Table 1. From Table 1, the surface areas of the samples obtained at 600 ◦ C decreased compared to that of the precursor of Ta2 O5 (2.965 m2 /g). Since little tantalum (oxy)nitrides was formed at 600 ◦ C, the decrease of the surface area might be due to the sintering of Ta2 O5 particles at the high temperature. The speciﬁc surface area of the sample obtained at 700 ◦ C was largest compared to those at 600 ◦ C

400

600

800

1000

Wavelengh(nm) Fig. 4. UV–vis diffuse reﬂectance absorption spectra of Ta2 O5 and tantalum (oxy)nitride samples prepared at 600, 700 and 800 ◦ C for 6 h at the NH3 ﬂow rate of 0.3 L/min.

and 800 ◦ C (in a NH3 ﬂow rate of 0.3 L/min for 6 h). XRD results showed that the contents of TaON followed 700 ◦ C > 800 ◦ C > 600 ◦ C and the contents of Ta3 N5 were 800 ◦ C > 700 ◦ C > 600 ◦ C. Combined the results of XRD with BET, it may be included that the surface areas were TaON > Ta3 N5 > Ta2 O5 . 3.2. Photocatalytic activity of tantalum (oxy)nitrides 3.2.1. Photocatalytic activity of tantalum (oxy)nitrides for Fe3+ reduction Prior to the investigation into the photocatalytic activity of tantalum (oxy)nitrides, the adsorption of Fe3+ , H2 O2 and atrazine on the samples was investigated; little adsorption was observed. Table 1 also lists the generation of Fe2+ in Fe3+ solution after 180 min of irradiation with visible light with the addition of tantalum (oxy)nitride. It is interesting to ﬁnd that the ability of tantalum (oxy)nitrides to reduce Fe3+ was apparently not determined by the nitrogen content and the absorption under visible light domain. The highest concentration of Fe2+ was achieved in the presence of sample 7 (preparation conditions: temperature = 700 ◦ C, QNH3 = 0.3 L/ min, time = 6 h). The amount of Fe2+ generated in the presence of sample 7 was approximately 1.7 times the amount generated with sample 9. However, sample 9 has a higher nitrogen content and stronger visible light absorption. XRD results showed that more TaON was contained in sample 7 and when the nitridation temperature increased to 800 ◦ C (sample 9), some TaON changed to Ta3 N5 . Here we tried to compare the photocatalytic activity of TaON and Ta3 N5 for Fe3+ reduction under visible light irradiation as such information is limited. Firstly, the following two nitridation processes were conducted to obtain the pure TaON and pure Ta3 N5 . Based on the previous reports [19,31,32], Ta2 O5 was nitrided at 750 ◦ C in a 0.06 L/min NH3 ﬂow for 12 h and at 850 ◦ C in a 0.9 L/min NH3 ﬂow for 12 h to form TaON (sample 12) and Ta3 N5 (sample 13), respectively. The XRD patterns of sample 12 and sample 13 are shown in Fig. 5. As could be seen, sample 13 was almost the pure Ta3 N5 and the majority of sample 12 was TaON. The photoreduction of Fe3+ under visible light irradiation for 3 h by sample 12, sample 13 and a mixture of sample 12 and sample 13 (w:w = 1:1) were investigated. The reduction of Fe3+ over sample 12 (2.34 mg/L of Fe3+ formed after 3 h visible light irradiation) was signiﬁcantly faster than that over sample 13 (1.03 mg/L of Fe3+ formed after 3 h visible light irradiation). It is concluded that the ability of TaON to reduce Fe3+ is

Y. Du et al. / Journal of Hazardous Materials 195 (2011) 291–297

a

Ta3N5 TaON

295

3.0

without catalyst with Ta2O5 with tantalum (oxy)nitrides (sample 7)

2.5

[Fe ](mg/L)

2.0

1.5

2+

(b)

1.0

0.5

(a) 0.0 0 10

20

30

40

50

60

40

70

Two-Theta (deg)

b

Fig. 5. XRD patterns of sample 12 (a) (nitrided at 750 ◦ C in a 0.06 L/min NH3 ﬂow for 12 h) and sample 13 (b) (at 850 ◦ C in a 0.9 L/min NH3 ﬂow for 12 h).

3.2.2. Enhancement of Fenton reaction and atrazine degradation by tantalum (oxy)nitrides Fig. 6(a) shows the generation of Fe2+ by various systems under visible light. No H2 O2 was added in these systems. Little ferrous ion was generated in Fe3+ solution either without catalyst or with Ta2 O5 . With the addition of sample 7, Fe2+ was produced continuously under the irradiation of visible light and then slowed down. After 180 min of irradiation, more than 2 mg/L of Fe2+ had been produced. In the presence of H2 O2 , the generated Fe2+ could decompose H2 O2 to • OH. Fig. 6(b) shows the decomposition of H2 O2 in a photoFenton-like system, a photo-Fenton-like system with Ta2 O5 (called photo-Fenton-like-Ta2 O5 system) and a photo-Fenton-like system with tantalum (oxy)nitride (sample 7) (called photo-Fenton-liketantalum (oxy)nitride system) under visible light. In the systems of photo-Fenton-like and photo-Fenton-like-Ta2 O5 , the decomposition of H2 O2 was very slow, only 20–30% of H2 O2 was consumed after 180 min. The decomposition of H2 O2 in these two systems was due to Eqs. (1) and (3). Fe3+ ﬁrstly reacted with H2 O2 to generate Fe2+ at a very slow rate of 0.01–0.001 M−1 s−1 , and then Fe2+

120

160

2.5

2.0

photo-Fenton-like photo-Fenton-like-Ta2O5 photo-Fenton-like-tantalum (oxy)nitrides (sample 7)

H2O2(mM)

1.5

1.0

0.5

0.0 0

20

40

60

80

100

120

140

160

180

reaction time(min) Fig. 6. Reduction of Fe3+ (a) and decomposition of H2 O2 (b) in various systems under visible light: [Fe3+ ]0 = 28 mg/L, [H2 O2 ]0 = 2.5 mM, input of tantalum (oxy)nitride = 0.6 g/L, pH 2.6.

1.0

0.8

0.6

photo-Ta2O5

0.4

photo-Fe photo-H2O2 dark-Fenton-like-tantalum (oxy) nitrides(sample 7) photo-Fenton-like photo-Fenton-like-tantalum (oxy)nitrides(sample 7)

3+

C/C0

higher than that of Ta3 N5 . In the presence of the mixture of sample 12 and sample 13, there was 1.91 mg/L of Fe2+ formed, a little higher than 1.69 mg/L (the average of 2.34 mg/L and 1.03 mg/L). This may be due to the synergism between TaON and Ta3 N5 for the photoreduction of Fe3+ . TaON and Ta3 N5 have been reported as the visible-light photocatalyst to decompose water to H2 and O2 . It was also found that the reduction of H+ to H2 over TaON was higher than over Ta3 N5 [31]. The difference of the photocatalytic activity between TaON and Ta3 N5 is similar to that of the two phases of TiO2 , anatase and rutile. Anatase has a band gap of 3.2 eV, while rutile has a smaller band gap of 3.0 eV. Nevertheless, anatase is generally regarded as the more photochemically active phase of TiO2 , presumably due to the combined effect of lower rates of recombination and higher surface adsorptive capacity [33]. In addition, the mixing of an active phase (anatase) with a comparatively inactive phase (rutile) produces a class of photocatalysts with higher activity [34,35]. The BET results showed that sample 12 has higher speciﬁc surface area (10.052 m3 /g) than sample 13 (6.731 m3 /g), which could be one of the reasons for the higher photocatalytic activity of TaON. Other factors to inﬂuence the photocatalytic activity of TaON and Ta3 N5 need exhaustive investigation in the further work.

80

reaction time(min)

0.2

0.0 0

20

40

60

80

100

120

140

160

180

reaction time (min) Fig. 7. Degradation of atrazine in various systems under visible light: [Fe3+ ]0 = 28 mg/L, [H2 O2 ]0 = 2.5 mM, [atrazine]0 = 18 mg/L, input of tantalum (oxy)nitride = 0.6 g/L, pH 2.6.

296

Y. Du et al. / Journal of Hazardous Materials 195 (2011) 291–297

a

tion (Eq. (1)) produced • OH, resulting in a slight degradation of atrazine. With the addition of tantalum (oxy)nitrides (sample 7), the degradation of atrazine in a photo-Fenton-like system was obviously enhanced. After 120 min of irradiation with visible light, atrazine was degraded completely.

1.0

first run second run third run fourth run

0.8

C/C0

0.6

0.4

0.2

0.0 0

20

40

60

80

100

120

140

160

180

reaction time(min)

b

fresh catalyst regenerated catalyst TaON

3.2.3. Reusability of tantalum (oxy)nitrides photocatalyst Fig. 8(a) shows the reusability of the tantalum (oxy)nitrides (sample 7) in photo-Fenton-like system during four experimental runs. As could be seen, the degradation of atrazine in the second run was slower than that in the ﬁrst run. Compared with that over the fresh catalyst, the percentage of atrazine degradation decreased by about 25% at 60 min and 8% at 90 min in the second run. Nevertheless, almost all atrazine was also removed after 120 min in the second run. In addition, it was encouraging that the degradation rate of atrazine was almost the same in the second, third and fourth runs. The XRD patterns of a fresh catalyst sample and a regenerated catalyst sample are compared in Fig. 8(b). The two XRD patterns was similar except that the peaks corresponding to TaON at 2 of 29.0 and 32.6◦ were a little lower in the regenerated catalyst. When the fresh tantalum (oxy)nitrides catalyst was subjected to the photoFenton-like system in the ﬁrst run, the component of the catalyst changed a little and the photocatalytic activity decreased. But after that, the tantalum (oxy)nitrides catalyst became stable and showed good activity for atrazine degradation.

4. Conclusion

10

20

30

40

50

60

70

Two-Theta (deg) Fig. 8. (a) Degradation of atrazine over the fresh and regenerated tantalum (oxy)nitride samples in photo-Fenton-like system for four experimental runs (b) XRD patterns of a fresh and a regenerated samples: [Fe3+ ]0 = 28 mg/L, [atrazine]0 = 18 mg/L, input of tantalum (oxy)nitride = 0.6 g/L, pH 2.6.

catalyzed H2 O2 to • OH and was oxidized to Fe3+ . A cycling of Fe2+ /Fe3+ was established, with slow decomposition of H2 O2 . In the system of photo-Fenton-like-tantalum (oxy)nitrides (sample 7), more than 96% of H2 O2 was decomposed after 180 min. This is due to the enhancement of Fe3+ reduction by the tantalum (oxy)nitrides. The enhancement of the photo-Fenton-like degradation of contaminants by tantalum (oxy)nitrides under visible light was also investigated. Atrazine was chosen as a target pollutant because atrazine has no absorption under visible light. Because atrazine or the intermediates of atrazine cannot interact with Fe3+ /Fe2+ , which was different to aromatic compounds [36], the generation of Fe2+ during the degradation of atrazine could be regarded solely as the role of tantalum (oxy)nitrides. Fig. 7 shows the changes in atrazine concentration in different systems under visible light. Atrazine scarcely underwent direct photolysis (data no shown). In systems where only Fe3+ or Ta2 O5 was added, little degradation of atrazine was observed under visible light. In the presence of tantalum (oxy)nitrides and H2 O2 , atrazine could hardly be degraded. In the process of dark-Fentonlike-tantalum (oxy)nitrides, approximately 19% of the atrazine was degraded after 3 h reaction. In the process of photo-Fenton-like, the trend of atrazine degradation is similar to that of dark-Fentonlike-tantalum (oxy)nitrides. In these two systems, Fe3+ was slowly reduced to Fe2+ through H2 O2 (Eq. (3)). And then the Fenton reac-

Tantalum (oxy)nitrides were prepared by the nitridation of Ta2 O5 at 600–800 ◦ C in a NH3 rate of 0.1–0.5 L/min for 2–6 h. XPS analysis showed that the degree of nitridation was signiﬁcantly affected by the nitridation temperature and the nitridation time. The N1s XPS spectra conﬁrmed the replacement of O atoms by N atoms to form Ta–N bonds in the Ta2 O5 crystal lattice. With the increase of the nitridation temperature, Ta2 O5 was converted into TaON and then to Ta3 N5 . DRS analysis showed that the tantalum (oxy)nitride samples prepared at 600, 700 and 800 ◦ C possessed absorption edges around 520, 620 and 640 nm, respectively. But the ability of the tantalum (oxy)nitride samples to reduce Fe3+ did not increase continuously with the nitrogen content and the absorption range under visible light. The generation of Fe2+ in the presence of sample 7 (nitrided at 700 ◦ C) was approximately 1.7 times that generated with sample 9 (nitrided at 800 ◦ C). This was because the photoactivity for Fe3+ reduction of TaON is much higher than Ta3 N5 . With the addition of sample 7, H2 O2 was efﬁciently decomposed to generate hydroxyl radicals in a photo-Fenton-like system under visible light. The degradation of atrazine was signiﬁcantly accelerated and almost all atrazine was degraded completely after 120 min. The regenerated tantalum (oxy)nitrides catalyst became stable after the ﬁrst run and showed good activity for atrazine degradation.

Acknowledgements The authors would like to acknowledge ﬁnancial support for this work provided by the National Science Foundation of China (No. 20906098), the National Science Foundation of Jiangsu Province, China (No. BK2010602), the Foundation of Director of Nanjing Institute of Geography and Limnology, Chinese Academy of China (No. NIGLAS2010QD03) and the State Key Laboratory of Hydrology—Water Resources and Hydraulic Engineering, Hohai University (No. 2009490811).

Y. Du et al. / Journal of Hazardous Materials 195 (2011) 291–297

References [1] W. Chu, C.Y. Kwan, K.H. Chan, S.K. Kam, A study of kinetic modelling and reaction pathway of 2,4-dichlorophenol transformation by photo-Fenton-like oxidation, J. Hazard. Mater. 121 (1–3) (2005) 119–126. [2] I. Arslan-Alaton, G. Tureli, T. Olmez-Hanci, Treatment of azo dye production wastewaters using photo-Fenton-like advanced oxidation processes: optimization by response surface methodology, J. Photochem. Photobiol. A: Chem. 202 (2–3) (2009) 142–153. [3] T.M. Elmorsi, Y.M. Riyad, Z.H. Mohamed, H.M.H. Abd El Bary, Decolorization of Mordant red 73 azo dye in water using H2 O2 /UV and photo-Fenton treatment, J. Hazard. Mater. 174 (1–3) (2009) 352–358. [4] M. Jiménez, I. Oller, M.I. Maldonado, S. Malato, A. Hernández-Ramírez, A. Zapata, J.M. Peralta-Hernández, Solar photo-Fenton degradation of herbicides partially dissolved in water, Catal. Today 161 (1) (2011) 214–220. [5] S. Navarro, J. Fenoll, N. Vela, E. Ruiz, G. Navarro, Removal of ten pesticides from leaching water at pilot plant scale by photo-Fenton treatment, Chem. Eng. J. 167 (1) (2011) 42–49. [6] M.R.A. Silva, W. Vilegas, M.V.B. Zanoni, R.F. Pupo Nogueira, Photo-Fenton degradation of the herbicide tebuthiuron under solar irradiation: iron complexation and initial intermediates, Water Res. 44 (12) (2010) 3745–3753. [7] M. Pera-Titus, V. Garcia-Molina, M.A. Banos, J. Gimenez, S. Esplugas, Degradation of chlorophenols by means of advanced oxidation processes: a general review, Appl. Catal. B: Environ. 47 (4) (2004) 219–256. [8] C. Walling, Fenton’s reagent revisited, Accounts Chem. Res. 8 (4) (1975) 125–131. [9] A. Safarzadeh-Amiri, J.R. Bolton, S.R. Cater, The use of iron in advanced oxidation processes, J. Adv. Oxid. Technol. 1 (1996) 18–26. [10] L.G. Devi, B.N. Murthy, S.G. Kumar, Heterogeneous photo catalytic degradation of anionic and cationic dyes over TiO2 and TiO2 doped with Mo6+ ions under solar light: correlation of dye structure and its adsorptive tendency on the degradation rate, Chemosphere 76 (8) (2009) 1163–1166. [11] B. Tryba, A.W. Morawski, M. Inagaki, M. Toyoda, Effect of the carbon coating in Fe–C–TiO2 photocatalyst on phenol decomposition under UV irradiation via photo-Fenton process, Chemosphere 64 (7) (2006) 1225–1232. [12] Z. Wang, C. Chen, F. Wu, B. Zou, M. Zhao, J. Wang, C. Feng, Photodegradation of rhodamine B under visible light by bimetal codoped TiO2 nanocrystals, J. Hazard. Mater. 164 (2–3) (2009) 615–620. [13] J. Sato, N. Saito, Y. Yamada, K. Maeda, T. Takata, J.N. Kondo, M. Hara, H. Kobayashi, K. Domen, Y. Inoue, RuO2 -loaded ␤-Ge3 N4 as a non-oxide photocatalyst for overall water splitting, J. Am. Chem. Soc. 127 (12) (2005) 4150–4151. [14] J. Ananpattarachai, P. Kajitvichyanukul, S. Seraphin, Visible light absorption ability and photocatalytic oxidation activity of various interstitial N-doped TiO2 prepared from different nitrogen dopants, J. Hazard. Mater. 168 (1) (2009) 253–261. [15] X.-Z. Shen, Z.-C. Liu, S.-M. Xie, J. Guo, Degradation of nitrobenzene using titania photocatalyst co-doped with nitrogen and cerium under visible light illumination, J. Hazard. Mater. 162 (2–3) (2009) 1193–1198. [16] A. Ghicov, J.M. Macak, H. Tsuchiya, J. Kunze, V. Haeublein, L. Frey, P. Schmuki, Ion implantation and annealing for an efﬁcient N-doping of TiO2 nanotubes, Nano Lett. 6 (5) (2006) 1080–1082. [17] X. Wang, T.-T. Lim, Effect of hexamethylenetetramine on the visible-light photocatalytic activity of C–N codoped TiO2 for bisphenol A degradation: evaluation of photocatalytic mechanism and solution toxicity, Appl. Catal. A: Gen. 399 (1–2) (2011) 233–241.

297

[18] V.M. Menéndez-Flores, D.W. Bahnemann, T. Ohno, Visible light photocatalytic activities of S-doped TiO2 –Fe3+ in aqueous and gas phase, Appl. Catal. B: Environ. 103 (1–2) (2011) 99–108. [19] W.-J. Chun, A. Ishikawa, H. Fujisawa, T. Takata, J.N. Kondo, M. Hara, M. Kawai, Y. Matsumoto, K. Domen, Conduction and valence band positions of Ta2 O5 , TaON, and Ta3 N5 by UPS and electrochemical methods, J. Phys. Chem. B 107 (8) (2003) 1798–1803. [20] Q.H. Zhang, L. Gao, Ta3 N5 nanoparticles with enhanced photocatalytic efﬁciency under visible light irradiation, Langmuir 20 (22) (2004) 9821–9827. [21] R. Abe, M. Higashi, K. Domen, Facile fabrication of an efﬁcient oxynitride TaON photoanode for overall water splitting into H2 and O2 under visible light irradiation, J. Am. Chem. Soc. 132 (34) (2010) 11828–11829. [22] X. Feng, T.J. LaTempa, J.I. Basham, G.K. Mor, O.K. Varghese, C.A. Grimes, Ta3 N5 nanotube arrays for visible light water photoelectrolysis, Nano Lett. 10 (3) (2010) 948–952. [23] Y.F. Wang, W.H. Ma, C.C. Chen, X.F. Hu, J.C. Zhao, J.C. Yu, Fe3+ /Fe2+ cycling promoted by Ta3 N5 under visible irradiation in Fenton degradation of organic pollutants, Appl. Catal. B: Environ. 75 (3–4) (2007) 256–263. [24] H. Tamura, K. Goto, T. Yotsuyanagi, M. Nagayama, Spectrophotometric determination of iron(II) with 1,10-phenanthroline in the presence of large amounts of iron(III), Talanta 21 (4) (1974) 314–318. [25] R.M. Sellers, Spectrophotometric determination of hydrogen peroxide using potassium titanium(IV) oxalate, Analyst 150 (1980) 950–954. [26] G. Hitoki, T. Takata, J.N. Kondo, M. Hara, H. Kobayashi, K. Domen, An oxynitride, TaON, as an efﬁcient water oxidation photocatalyst under visible light irradiation (≤500 nm), Chem. Commun. 16 (2002) 1698–1699. [27] I. Tsuji, H. Kato, H. Kobayashi, A. Kudo, Photocatalytic H2 evolution reaction from aqueous solutions over band structure-controlled (AgIn)x Zn2(1−x) S2 solid solution photocatalysts with visible-light response and their surface nanostructures, J. Am. Chem. Soc. 126 (41) (2004) 13406–13413. [28] W.R. Matizamhuka, I. Sigalas, M. Herrmann, Synthesis, sintering and characterisation of TaON materials, Ceram. Int. 34 (6) (2008) 1481–1486. [29] E. Orhan, F. Tessier, R. Marchand, Synthesis and energetics of yellow TaON, Solid State Sci. 4 (8) (2002) 1071–1076. [30] C.M. Fang, E. Orhan, G.A. de Wijs, H.T. Hintzen, R.A. de Groot, R. Marchand, J.Y. Saillard, G. de With, The electronic structure of tantalum (oxy)nitrides TaON and Ta3 N5 , J. Mater. Chem. 11 (4) (2001) 1248–1252. [31] T. Takata, G. Hitoki, J. Kondo, M. Hara, H. Kobayashi, K. Domen, Visible-lightdriven photocatalytic behavior of tantalum-oxynitride and nitride, Res. Chem. Intermediat. 33 (1) (2007) 13–25. [32] T. Hisatomi, M. Otani, K. Nakajima, K. Teramura, Y. Kako, D. Lu, T. Takata, J.N. Kondo, K. Domen, Preparation of crystallized mesoporous Ta3 N5 assisted by chemical vapor deposition of tetramethyl orthosilicate, Chem. Mater. 22 (13) (2010) 3854–3861. [33] D.C. Hurum, A.G. Agrios, K.A. Gray, T. Rajh, M.C. Thurnauer, Explaining the enhanced photocatalytic activity of Degussa P25 mixed-phase TiO2 using EPR, J. Phys. Chem. B 107 (19) (2003) 4545–4549. [34] D.C. Hurum, K.A. Gray, T. Rajh, M.C. Thurnauer, Recombination pathways in the Degussa P25 formulation of TiO2 : surface versus lattice mechanisms, J. Phys. Chem. B 109 (2) (2004) 977–980. [35] T. Ohno, K. Tokieda, S. Higashida, M. Matsumura, Synergism between rutile and anatase TiO2 particles in photocatalytic oxidation of naphthalene, Appl. Catal. A: Gen. 244 (2) (2003) 383–391. [36] Y. Du, M. Zhou, L. Lei, Role of the intermediates in the degradation of phenolic compounds by Fenton-like process, J. Hazard. Mater. 136 (3) (2006) 859–865.

Journal of Hazardous Materials 195 (2011) 298–305

Contents lists available at SciVerse ScienceDirect

Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat

Photostabilization of the herbicide norﬂurazon microencapsulated with ethylcellulose in the soil–water system ˜ ∗ , Jaime Villaverde, Celia Maqueda, Esmeralda Morillo Fatima Sopena Institute of Natural Resources and Agrobiology (CSIC), Reina Mercedes 10, Apdo 1052, 41080 Seville, ES, Spain

a r t i c l e

i n f o

Article history: Received 6 May 2011 Received in revised form 15 July 2011 Accepted 14 August 2011 Available online 31 August 2011 Keywords: Ethylcellulose Norﬂurazon Photolysis Microencapsulation Controlled release Soil colloidal components

a b s t r a c t Ethylcellulose-microencapsulated formulations (ECFs) of norﬂurazon have been shown to reduce leaching, maintaining a threshold concentration in the topsoil than the commercial formulation (CF). Since photodegradation contributes to ﬁeld dissipation of norﬂurazon, the objective of the present work was to study if such formulations can also protect from its photodescomposition. For this purpose, aqueous solutions of CF and ECFs, containing the most important soil components (goethite, humic and fulvic acids and montmorillonite) were tested. To get a more realistic approach, studies in soil were also performed. The results were well explained by a simple ﬁrst order model. DT50 value was 3 h for CF under irradiation, which was considerably lower than those corresponding to the systems where ECF was used (35 h for ECF; 260 h for ECF–goethite; 53 h for ECF–humic acids; 33 h for ECF–montmorillonite; and 28 h for ECF–fulvic acids). ECF protected against photodegradation in both aqueous solution and soil due to the gradual release of the herbicide, which reduced the herbicide available to be photodegraded. These lab-scale ﬁndings proved that ECF could reduce the herbicide dosage, minimizing its photolysis, which would be especially advantageous during the ﬁrst hours after foliar and soil application. © 2011 Elsevier B.V. All rights reserved.

1. Introduction Transformation of pesticides in the environment is a highly complex process affected by biological and physicochemical factors. Photodegradation is one of the most destructive pathways for pesticides after their ﬁeld application. Water is the most studied environmental compartment in which the pesticides reside. This is because of the need to predict the photochemical transformations of xenobiotics in aquatic environments [1–3]. However, plants, especially, leaf surface, and soil surface are the ﬁrst reaction environment for a pesticide after application. Photolysis in such situations becomes an important mechanism of pesticide dissipation. The heterogeneity of both, soil and plant, together with the capricious transmission of sunlight onto these media makes the photolysis on them difﬁcult to understand [4]. Therefore, additional studies should be conducted to determine photolysis rates on such surfaces, since few studies have been reported so far. Recently, new methodologies with different degree of complexity have been developed to study the pesticide photochemistry on leaf and soil surfaces [5–8]. However, no standardized

∗ Corresponding author. Present address: School of Life Sciences, University of Warwick, Wellesbourne Campus, CV35 9EF, UK. Tel.: +044 024 7657 5170. ˜ E-mail address: [email protected] (F. Sopena). 0304-3894/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.jhazmat.2011.08.039

predictive method for assessing the importance of photolysis on soil surfaces has been reported yet. Likewise, soils are a complex system containing several variable parameters, each of which can affect the photochemistry of a chemical adsorbed on the surface. For instance, both photoproducts and photolysis rates of several pesticides have been shown to be affected by moisture content, mineral base, texture and organic matter content [4,9,10]. Pesticide photolysis on soil surface results in a loss of its efﬁcacy in controlling weeds and pests [11]. Thus, the application rate of the pesticide is increased to overcome its photodegradation. In this sense, several approaches have been used to reduce the environmental and economic costs of pesticides losses by photolysis, being one of them the development of photo-stable formulations of pesticides. Most of them rely on in the use of clay and modiﬁed clays with a different degree of photo-stabilization [12,13]. Information about obtaining pesticide encapsulated formulations by using other materials or a cheaper technology has been barely dealt. Effects of cyclodextrins on the photodegradation rates of organophosphorous pesticides were studied by Kamiya and Nakamura [14]. Demchak and Dybas [15] showed a decreased photodegradation of a natural pesticide by its encapsulation in protein zein. Huston and Pignatello [16] reported alachlor photodegradation from its microencapsulated commercial formulation when studied a waste water treatment. No further studies have been reported, which

F. Sope˜ na et al. / Journal of Hazardous Materials 195 (2011) 298–305

implies that the information about pesticide photodegradation from microencapsulated formulations is scarce so far. Norﬂurazon (NFZ) is a ﬂuorinated pyridazinone herbicide that is registered for soil-applied usage on cotton, soybean, tree fruit and nut crops, citrus, and cranberries [17]. Unfortunately, this herbicide suffers losses when it is applied to soil [18]. Losses by leaching in sandy soil columns have been also observed by Singh et al. [19] and Morillo et al. [20]. To prevent herbicides leaching, ethylcellulosemicroencapsulated formulations (ECFs) of norﬂurazon and alachlor have been prepared by using a simple and non-expensive technique [21,22]. These formulations not only provided a suitable release, reduced leaching, while maintaining the threshold concentration of the herbicide in the topsoil [23–25], but also remained the herbicidal activity longer than the commercial formulation [26,27]. Since photodegradation contributes signiﬁcantly to ﬁeld dissipation when norﬂurazon remains on the soil surface, with a half-life of 41 days [28], it would be interesting to study if such norﬂurazon formulations could also protect it from photodegradation. Therefore, the aim of the present work was to examine the effect of the microencapsulation in ethycellulose (EC) on the photodegradation of the herbicide NFZ. For this purpose, the photoprotective effect of CF vs. ECFs was tested by using different aqueous solutions containing the most important soil components (goethite, humic and fulvic acids and montmorillonite). Since up to 50% of the norﬂurazon initially applied as ECF remained in the topsoil layer in soil columns [23] it would be interesting to investigate how the microencapsulation in EC could protect NFZ at this soil level, which is also the most exposed to sunlight. Considering this, a realistic study, using an agricultural sandy soil has also been performed.

2. Materials and methods 2.1. Chemicals Norﬂurazon Technical grade (97.8% purity) and its commercial formulation (Zorial 80, content of norﬂurazon 80%) were kindly supplied by Syngenta Agro S.A. (Barcelona, Spain). Ethylcellulose (30–50 mPa, 48–49.5% w/w as ethoxyl) was purchased from Fluka (Buchs, Switzerland). Polyvinyl alcohol (PVA) with MW 30,000–70,000 was obtained from Sigma (St. Louis, USA). HPLCgrade acetronitrile, methanol, and chloroform were purchased from Merck (Darmstadt, Germany). All reagents were of analytical grade.

2.2. Soil Soil surface sample (0–20 cm depth) was collected, air-dried, sieved through a 2 mm sieve, and stored in plastic containers until used in the experiments. The soil used was a loamy sand (classiﬁed as Typic Xeropsament) with 84 g/kg clay, 40 g/kg silt, 876 g/kg sand, 9.2 g/kg organic matter, pH 8.0, 69 g/kg calcium carbonate, and a cationic exchange capacity of 4.8 cmol/kg.

299

2.4. Microsphere preparation and characterization Using a previously described procedure [22], the ethylcellulosemicroencapsulated formulation was obtained by the oil-in-water emulsion solvent evaporation technique, using EC as the polymer and PVA as the emulsiﬁer. Brieﬂy, EC (1 g) was dissolved in 15 mL of chloroform. Norﬂurazon (0.2 g) was dissolved in this polymer solution at room temperature. The herbicide–polymer solution was then emulsiﬁed into an aqueous phase by dropwise addition into 150 mL of aqueous solution containing 112.5 mg (0.075%) of PVA with stirring at 600 rpm. The ratio of organic to aqueous phase was 1/10. After 24 h of stirring to allow the total evaporation of the inner organic phase, the microspheres obtained were ﬁltered and washed with 250 mL of distilled water to remove any undesired residuals. The product was dried in vacuum desiccators until a constant weight was obtained. All experiments were performed in triplicate. The herbicide loading (amount of herbicide encapsulated by the microspheres) and encapsulation efﬁciency (amount of herbicide encapsulated with respect to the herbicide used) of microspheres were 15.9% and 78% (w/w), respectively, as previously shown in ˜ et al. [23]. Sopena 2.5. Photochemical procedures Photodegradation studies have been carried out with a Suntest CPS photoreactor (Heraeus, Hanau, Germany) equipped with a Xe lamp (500 W/m2 ) with a permanent ﬁlter, which selects wavelengths ≥290 nm. 2.5.1. Experiments in aqueous solution An aqueous NFZ suspension (250 mL, 20 mg L−1 ) applied as CF and ECF was magnetically stirred and irradiated in quartz ﬂasks during 32 h. Previous studies with the herbicide norﬂurazon had shown that the time of 32 h is fair enough to get a reliable evaluation of the herbicide photodegradation in aqueous solutions [31]. The same experiment was carried out in the presence of 20 mg of different soil components. Dark control experiments were conducted in a similar manner over the irradiation periods but ﬂasks were covered with aluminum foil. All experiments were carried out in triplicate. Samples collection and subsequent ﬁltration (0.22 ␮m Millipore glass ﬁber membrane) was carried out at different time intervals. NFZ quantiﬁcation in the samples was performed by HPLC (see below). At the end of the experiments, the ECF-suspensions were centrifuged (10,000 × g for 10 min) to eliminate the majority of water and the rest was lyophilized. The solids collected were extracted with methanol (recovery was 97.5%). Degradation curves were ﬁtted to three kinetic models: a simple ﬁrst-order equation, a ﬁrst-order multicompartment (Gustafson and Holden) model and a ﬁrst-order sequential (hockey-stick) model. Parameters were optimized according to recommendations by FOCUS [32] using the least-squares method with Microsoft Excel Solver. Simple ﬁrst-order kinetics (Eq. (1)) was found to be the best descriptor for the experimental data [33]. The linearized form of Eq. (1) (Eq. (2)) was used to calculate the degradation constants (k): Ct = C0 e−kt

(1)

LnCt = LnC0 − kt

(2)

2.3. Soil colloidal components SWy Montmorillonite (Mo, from Source Clay Minerals Repository, USA); synthetic humic acid (HA, from FLUKA); natural metal–fulvic acid complex (FA) extracted from a Typic Haphorthod soil, Scotland [29], with an iron content three times that of aluminum (22.16% Fe and 7.54% Al); synthetic acicular goethite (Go) with a speciﬁc surface area of 43 m2 g−1 and point zero of charge 8.2 [30].

and the time required for 50% disappearance (DT50 ) of NFZ (Eq. (3)), as follows: LnC50 = Ln t50

C 0

2

2 = Ln = DT50 k

= LnC0 − kt50

(3)

300

F. Sope˜ na et al. / Journal of Hazardous Materials 195 (2011) 298–305

where Ct is the concentration of pesticide remaining in soil (mg kg−1 ) at time t (h), C0 is the initial concentration of pesticide (mg kg−1 ) and k is the rate of degradation (h−1 ).

All statistical analyses were performed using the statistical analysis program SPSS® version 16 (SPSS Inc., Somers, NY 10589, USA). 3. Results and discussion

2.5.2. Experiments in soil Sterilized soil (10 g) was mixed in triplicate with 1 mg of norﬂurazon applied as CF and ECF and then shaken thoroughly for 24 h. After mixing, the samples were transferred to Petri-dishes (13 cm), which provided a surface of 132.7 cm2 and a soil layer thickness less than 1 mm. Previous studies have shown that when soil thickness was of 1.5 mm or greater, sunlight penetration is very slight [34]. The soil moisture content was maintained at ﬁeld capacity throughout experiment by difference of weight, adding sterilized water when needed. 1 g of treated soil (dry basis) was sampled at different times (0, 6, 14, 22, 28, 34, 60, 94 h). The NFZ remaining in each soil sample was extracted with 6 mL of methanol in centrifuge tubes, which were shaken by using a vortex for 3 min and later on allowing that soil was settled down. Quantiﬁcation of NFZ was made by HPLC (see below). Dark controls were also performed in the same manner by covering the Petri-dishes with aluminum foil. All experiments were performed in triplicate.

3.1. Photodegradation of norﬂurazon As shown in Fig. 1a, the commercial form of NFZ provided an exponential decay in the amounts of the herbicide remaining in solution under irradiated conditions (CF exposed), whereas the NFZ kept almost unaltered in dark experiments (CF dark), which provides a clear evidence of the great degradative effect of the light on NFZ stability in both water and soil (Fig. 1b), with almost 100% and 60% degraded, respectively. The irradiation of NFZ in both soil and aqueous solution produced desmethylnorﬂurazon (DM-NFZ) as main photodegradation product, observed by the mass fragment of 289.60 (m/z) obtained by HPLC/MS. A very good agreement was observed between the amount of DM-NFZ detected and NFZ degraded (data not shown). It has been previously reported that the major photodegradation pathway of NFZ involves the loss of methyl group (demethylation) and the chloro group (dechlorination), resulting in the formation of desmethylnorﬂurazon (DMNFZ) and deschloronorﬂurazon (DCMNFZ) [35,36].

2.6. Analysis of norﬂurazon and its main metabolite 3.2. Photostability of NFZ formulations in water and soil As shown in Fig. 1a, the herbicide release from the microencapsulated formulation to the aqueous solution (ECF dark) was greatly delayed in comparison with that from the commercial formulation (CF dark), in which almost 100% of the herbicide was kept in solution throughout the experiment, and consequently available to be photodegraded. The reason is that the herbicide is entrapped in the polymer and has to diffuse out from EC microspheres into the outside sink solution [22]. The herbicide releasing from ECF was

a NFZ remaining in soluon (%)

NFZ samples were analysed by HPLC under the following conditions: mobile phase, 50:50 acetonitrile/water; ﬂow 0.5 mL min−1 ; chromatographic column, Kromasil C18 (15 × 0.40 i.d.) (Teknokroma, Spain); ﬂuorescence detector (Shimadzu RF-10 A XL) at excitation wavelength of 310 and emission 405 nm. The retention time for norﬂurazon under these conditions was 6.5 min. Limit of detection was 0.01 mg L−1 . NFL metabolite was determined by employing a HPLC/MS instrument (Waters) under the following conditions: mobile phase, Milli-Q water with formic acid 0.1% (60%) and acetonitrile (40%); a ﬂow rate of 0.3 mL min−1 ; a C18 silica-based analytical column, Xterra TM MS 2.1 mm × 150 mm × 5 ␮m. Column efﬂuent was directly input into the electrospray ionization (ESI) interface operating in the positive ion mode. The characteristic ion used for analysis was m/z 304, corresponding to the molecular weight of the herbicide protonated form. A Waters ZQ 2000 Detector (single quadrupole) mass spectrometer (Waters, Milford, MA) with a Z-spray–electrospray interface was used. Desolvation and cone gas was nitrogen which ﬂow was set to approximately 200 and 50 L h−1 , respectively. Capillary voltage of 3.5 kV was used in positive ionization mode. The source temperature was set to 80 ◦ C and the desolvation temperature to 200 ◦ C. Dwell times of 0.20 s scan−1 were chosen. Data station operating software was MassLynx v4.0.

100

CF dark

80

CF exposed 60

ECF dark 40

ECF exposed 20 0

2.7. Statistical analysis

b

Photodegradation data were subjected to ANOVA analysis and signiﬁcant differences (p < 0.05) determined by least signiﬁcant differences (LSD) comparisons to identify if photoprotective effect of ECF was signiﬁcantly different from CF as well as to test how the presence of soil colloidal components can provide any signiﬁcant differences on the NFZ photostability observed from ECF. Stepwise multiple linear regression analysis was used to identify predictive equations for the herbicide photodegradation from ECF based on soil components variables, in order to see which variable made a more signiﬁcant contribution to norﬂurazon photodegradation in aqueous solution. Parameters were added to the model in order of signiﬁcance, non-signiﬁcant (p > 0.05) variables were eliminated, leaving only signiﬁcant variables at p ≤ 0.05 in the ﬁnal model.

NFZ remaining in soil (%)

0

5

10

15

20

25

30

35

Time (hours)

100 80

ECF exposed

60

CF exposed

40 20 0

0

20 ‡

40

60

80

Time (hours)

Fig. 1. Norﬂurazon in solution from microencapsulated (ECF) and commercial (CF) formulations in water (a) and soil (b) during the photodegradation study [‡ Values are mean of three replicated].

F. Sope˜ na et al. / Journal of Hazardous Materials 195 (2011) 298–305 Table 1 Norﬂurazon in solution and residuala (%) after photodegradation experiments in aqueous solution.b

CF ECF ECF + FA ECF + Mo ECF + HA ECF + Go a b

Irradiated experiments

NFZ In solution

NFZ Residual

NFZ In solution

NFZ Residual

94.9 (±4.9) 68.3 (±3.4) 57.6 (±5.9) 53.3 (±4.4) 63.1 (±3.1) 72.1 (±2.5)

94.9 (±4.9) 97.2 (±7.8) 93.8 (±7.4) 98.5 (±8.0) 94.5 (±8.2) 98.7 (±6.4)

4.02 (±1.8) 11.6 (±3.9) 3.2 (±3.9) 2.4 (±3.9) 22.4 (±3.9) 70.3 (±3.9)

4.02 (±1.8) 43.3 (±6.0) 45.6 (±9.6) 49.0 (±6.6) 59.2 (±8.6) 96.2 (±8.2)

NFZ in solution plus NFZ encapsulated. Values are mean of three replicated (±standard deviation).

100

NFZ remainingin soluon (%)

Dark experiments

301

80

60

40

b 20

a 0

nonlinear (ECF dark), characterized by an initial fast release in the ﬁrst few hours. This initial rapid discharge results from the NFZ located in surface of the particles [22,37]. Afterwards, lower constant concentrations of NFZ were released over a longer period of time, which is due to the lengthening of the diffusional pathway of the herbicide through the polymer [38]. When the formulation ECF is exposed to the light (ECF exposed), this sustained release compensates the losses by photodegradation of NFZ in aqueous solution, and even no NFZ photodegradation was observed during the ﬁrst 6 h of light exposition (Fig. 1a). On the contrary, only 12% of the NFZ applied as CF was detected in solution after 6 h (CF exposed), since the herbicide is immediately released to the media and thus completely available for its degradation. This is the reason why the residual NFZ applied as ECF was approximately a 40% higher than that applied as CF at the end of the experiment (Table 1). The residual NFZ results from the NFZ in solution under irradiation plus NFZ that remains encapsulated. When NFZ microspheres are applied to the soil (Fig. 1b), almost 100% of the initially applied NFZ was detected all over the experiment (ECF exposed). In contrast, 60% of the initially applied NFZ as CF was photodegraded at the end of the experiment (CF exposed). This fact shows the photoprotective effect provided by the microencapsulation. Dark controls for both ECF and CF were not shown, since they kept unaltered after 96 h. The slower NFZ photodegradation of CF formulation observed in soil compared to that in water can be explained by the screening effect of the soil matrix, which slows down the dissipation process [5,39]. The encapsulation in ethylcellulose itself acts as a physical barrier, in which the encapsulated NFZ is protected against the action of the light for some period of time. This could explain that the residual NFZ (NFZ in solution plus NFZ encapsulated) in the irradiated samples was much higher when it was applied as ECF (43.3%) than that obtained from CF (4.02%) after 32 h (Table 1). Considering the results shown in Fig. 1a, curves from dark control and irradiated samples are overlapped for ﬁrst 6 h, suggesting that the light had not degradative effect until this time. This would imply that EC could exhibit a protective effect on NFZ photodegradation, even when the herbicide is not encapsulated. After 6 h, the released NFZ from ECF in the exposed samples begins to be degraded. To conﬁrm this hypothesis a simple test was made by using the same photoreactor conditions, but a blank of microspheres without NFZ encapsulated were suspended in 250 mL of an aqueous solution of commercial NFZ (20 ppm). After 6 h, 97% of the initially dissolved NFZ was unaltered in presence of EC empty microspheres (Fig. 2b). This could be explained by a screening effect of the EC particles (physical action) coupled with a competition between NFZ and EC for the available photons, and an energy transfer processes between both herbicide and polymer (chemical action). Demchak and Dybas [15] studied the photostability of abamectin in protein zein microspheres, which was enhanced due to retarded reaction between oxygen and the diene chromophore of abamectin in presence of

0

10

20

30

Time(hours) Fig. 2. Irradiation of commercial NFZ aqueous solution (20 mg L−1 ) (a) and in the presence of empty EC microspheres (non-encapsulated NFZ) (b) [‡ Values are mean of three replicated].

light. These authors stated that zein could act as a quencher of singlet oxygen (chemical action) and/or oxygen barrier (physical action). 3.3. Inﬂuence of soil components and microencapsulation on NFZ photodegradation in aqueous solution In order to know which of soil colloidal components are involved in the NFZ photodegradation process, experiments were carried out in the presence of goethite, fulvic acid, montmonillonite (Mo) and humic acid (HA). The percentages of NFZ remaining in solution in the presence of different soil colloidal components are shown in Fig. 3 and greatly varied for each soil component considered. On the other hand, the percentages of residual NFZ in the system (not degraded NFZ, in solution + encapsulated) were plotted versus irradiation time, ﬁtting the data to a simple ﬁrst order model (Fig. 4). Degradation constants (k) and DT50 values obtained ranged from 0.002 to 0.22 h−1 and from 260 to 3 h, respectively. The mean scores and deviations for the NFZ in solution and residual for each soil component after photodegradation experiments were presented in Table 1. In the case of Go, although previous studies have reported that the photodegradation rate of some organic molecules is related to their sorption ability to the oxide surface, NFZ has not been found to be adsorbed on Go [31,40], which obstruct the action of photons on the herbicide molecule. In the present work, the 70.3% of the applied NFZ remained in solution in the presence of Go, which is quite similar to the percentage obtained in the corresponding dark experiment (72.1%) that shows the released NFZ (Table 1). This fact indicates that practically all the NFZ released into aqueous media has kept unaltered, but not adsorbed on the oxide surface, and thus greatly protected against its photo-descomposition (97.5% of the NFZ initially applied has not been degraded) (Table 1). In fact, exposed and dark proﬁles of the NFZ remaining in solution were almost coincident (Fig. 3a), which provides a visual evidence of this. This high photo-protection can be explained by the combination of both the screening effect of Go and the slower gradual release provided by EC microspheres. The high photoprotective effect of Go is observed also in Fig. 4c, since the DT50 value obtained was the highest (260 h). In the absence of soil components, the ﬁnal residual NFZ from ECF formulations was 43.2% under irradiation, whereas this value enhanced to 59.2% in the presence of HA. Similarly, DT50 values increased from 35 to 53 h when HA were present in the

302

F. Sope˜ na et al. / Journal of Hazardous Materials 195 (2011) 298–305

100

80

60

40

20

100

ECF+HA dark ECF+HA exposed

80

60

40

20

Time (h)

Time (h)

0 0

NFZ remaining in solution (%)

NFZ remaining in solution (%)

ECF+Go exposed

0

c

b

ECF+Go dark

5

10

15

20

100

25

30

35

d

ECF+Mo dark ECF+Mo exposed

80

0

60

40

20

Time (h) 0

NFZ remaining in solution (%)

NFZ remaining in solution (%)

a

5

10

15

20

100

25

30

35

ECF+FA dark ECF+FA exposed

80

60

40

20

Time (h) 0

0

5

10

15

20

25

30

35

0

5

10

15

20

25

30

35

Fig. 3. Norﬂurazon remaining in solution from microencapsulated (ECF) formulations during photodegradation studies in aqueous suspensions of different soil components [‡ Values are mean of three replicated].

NFZ-microspheres suspension. This indicates that this soil component also provided a high photo-protective effect, but lower than Go did (Table 1, Fig. 4d). In some previous studies, it has been shown that HA act as photo-sensibilizer by photo-induced radical generation [41], but this effect has been demonstrated to depend on the pesticide afﬁnity for the generated free radicals [42]. Conversely to Go, HA protection can be attributed to the NFZ adsorption on this component, which would cause a deactivation of the excited herbicide molecule [40], although only a small percentage of NFZ (about 5%) seems to be adsorbed on HA (Table 1) comparing ECF dark controls in water and in the presence of HA. Additionally, a screening effect of HA can also contribute to explain its photoprotective effect, as reported with certain pesticides in aqueous solution [43,44]. Graphically (Fig. 3b), HA clearly protected NFZ against decomposition for 24 h, afterwards the amount of herbicide remaining in solution under light conditions progressively declined, being lower than that detected in the absence of light. Other authors have also reported a decreasing of the pesticide photodegradation in presence of HA [8,45]. In presence of Mo, only 49% of the applied herbicide remained in the system at the end of the irradiated experiment versus 43.3% in absence of the clay mineral (Table 1). It indicates that Mo slightly protected the herbicide against photodegradation. However, this is contradictory with DT50 and k-values for ECF and ECF in the presence of Mo, which are quite similar (Table 1, Fig. 4b and 4e). No statistical differences (p > 0.05) were found in the rate of NFZ-photolysis in both cases. As shown in Fig. 3c, no photodegradation was observed during the ﬁrst 2 h. Afterwards the amounts of NFZ were progressively declining to 13% after 24 h and ﬁnally

dropped off at the end of the experiment, that is, almost all the released NFZ (53.3%) was degraded (Table 1). Some authors have reported the photostabilization of bensulfuron-methyl after being adsorbed in clay minerals [9], whereas other authors have reported that clay minerals enhance the photolysis of metolachlor in water [46]. Undabeytia et al. [12] observed increased photodegradation of norﬂurazon when the herbicide was irradiated in an aqueous suspension containing Mo. It was attributed to the attack of hydroxyl radicals produced by the clay under irradiation. However, it has been previously found to be dependent on the concentration of Mo in the aqueous media. In this sense, Villaverde et al. [31] observed an increase of NFZ photo-degradation rate as Mo concentration increased. It was because of higher concentration of the clay yielded higher amounts of NFZ absorbed and available to be attacked by reactive species formed at the clays surface. The results found here support this, although in the present work the NFZ concentration in the aqueous media was gradually increasing with time and Mo concentration was ﬁxed (80 mg L−1 ). The Mo concentration used here coincides with the lowest concentration used by Villaverde et al. [31], which had provided a light photostabilization of the herbicide. During the ﬁrst 2 h, NFZ released into aqueous suspension was very low to be susceptible to be absorbed and attacked by active radicals on the Mo surface. At this time, the screening effect provided by Mo protects the herbicide against its photodegradation. However, longer periods of irradiation rendered a higher number of reactive species attacking NFZ on the clay mineral. The adsorption of NFZ on Mo is supported by the decrease in the amount of NFZ remaining in solution (15%) observed when compared the ECF dark controls in water and in the presence of Mo after 32 h (Table 1).

F. Sope˜ na et al. / Journal of Hazardous Materials 195 (2011) 298–305

CF

ECF

b

120

Residual Norflurazon (%)

Residual Norflurazon (%)

a

100 80

K= 0,22 h-1 DT50= 3 h

60 40 20

120 100 80 60

K= 0,019 h-1 DT50= 35 h

40 20 0

0 0

10

20

30

0

40

10

ECF+Go

Residual Norflurazon (%)

Residual Norflurazon (%)

d

100 80 60

K= 0,002 h-1 DT50= 260 h

20

0

120 100 80 60

K= 0,013 h-1 DT50= 53 h

40 20

10

20

30

0

40

10

Time (hours)

80 60 -1

K= 0,021 h DT50= 33 h

f

120

Residual Norflurazon (%)

100

20

30

40

30

40

ECF+FA

120

40

20

Time (hours)

ECF+Mo Residual Norflurazon (%)

40

0

0

e

30

ECF+HA

120

40

20

Time (hours)

Time (hours)

c

303

100 80 60

K= 0,024 h-1 DT50= 28 h

40 20 0

0 0

10

20

30

40

Time (hours)

0

10

20

Time (hours)

Fig. 4. NFZ photodegradation proﬁles in aqueous solution of commercial (CF) (a) and microencapsulated (ECF) formulation (b), and ECF in the presence of gohetite (ECF + Go) (c), humic acids (ECF + HA) (d), montmorillonite (ECF + Mo) (e), and fulvic-acids (ECF + FA) (f), and the corresponding kinetic parameters after their adjustment to simple ﬁrst-order kinetics: photodegradation constants (k) and DT50 values. Model (dash line); measured (symbol, ).

This suggests that the concentration of the released herbicide was fair enough to be absorbed by the clay, and thus a great part is available to be degraded by hydroxyl radicals. This fact would become more important than the screening effect from Mo after the ﬁrst 2 h, making that all the released NFZ was photo-decomposed at the end of the experiment under irradiation. FA did not protect to NFZ from its photodecomposition, and the residual herbicide at the end of the experiment was comparable to that observed for ECF (Table 1). As shown in Fig. 4b and f, DT50 value was even lower than that for ECF (28 and 35 h−1 , respectively).

This result is in agreement with that from Villaverde et al. [31] using water solutions of NFZ and cyclodextrins. The iron content in natural-metal–fulvic acid complex is 22.2% and has been attributed to have an attenuation effect on its herbicide photostabilization capability. The luminescence of humic substances can be attenuated by complex formation with paramagnetic metals, decreasing their photochemically excited states. A slight adsorption of NFZ on FA was shown by the decrease in the NFZ remaining in solution from the dark controls (Table 1, Fig. 3d in comparison to Fig. 1). However, it should be noticed that this fact did not provide a NFZ

304

F. Sope˜ na et al. / Journal of Hazardous Materials 195 (2011) 298–305

Table 2 Standardized regression coefﬁcients of the soil component predictors in the multiple linear regression analysis for the photodegradation of microencapsulated norﬂurazon in aqueous solution. Standardized coefﬁcient Constant Go HA Mo FA

−1.101 −0.286 −0.162 −0.073

t

p

112.973 −72.618 −18.850 −10.684 −4.789

0.000 0.000 0.000 0.000 0.005

Acknowledgements The authors thank Mrs M.J. Calderon for the technical assistance using the HPLC/MS and Novartis Spain for providing NFZ. This work was supported by Junta the Andalucía withing the Research Project P06-FQM-01909, and the Spanish Ministry of Science and Innovation within the Project CTM2009-07425, (coﬁnanced by Fondo Europeo de Desarrollo Regional, FEDER).

References photostabilization by energy transfer from the herbicide to the FA. The presence of the ferrous ions counteracted the photoprotective effect of the herbicide adsorption to FA. ANOVA analysis was performed and LSD comparisons were used to identify if the presence of soil colloidal components can provide any signiﬁcant differences (p < 0.05) on the NFZ photostability observed from ECF. The NFZ photodegraded (NFZD) was calculated from the difference between the herbicide detected in aqueous solution under dark and irradiated conditions. The ANOVA showed all the soil components have a great effect on NFZD, as it was signiﬁcant beyond the 0.01 level [F(4,5) = 1939.2, p < 0.01, R2 = 0.999]. To see which soil variable made a more signiﬁcant contribution into NFZD aqueous solution, a stepwise multiple linear regression analysis was also used. The resultant predictive equation for herbicide photodegradation based on soil variables is as follows: NFZD = 57.71 (±0.51)–52.47 (±0.72) Go–13.62 (±0.72) HA–7.72 (±0.72) Mo–3.46 (±0.72) FA. Table 2 shows the standardized coefﬁcients of the soil predictors and their signiﬁcance. The p-values of all predictors indicate their signiﬁcant inﬂuence on NFZD in aqueous solution at a 95% conﬁdence level. The results agree what observed above. Go was the most signiﬁcant variable, followed by HA, Mo and FA. Although Mo and FA had a weak inﬂuence on NFZD, they were still signiﬁcant (p < 0.01). 4. Conclusions The present work extends the knowledge about ethylcellulosemicroencapsulated formulations. The ﬁndings found here clearly demonstrate that EC-microspheres provided a gradual and sustained release of NFZ, which considerably counteracts the herbicide photolysis in comparison to the commercial formulation. ECF protected against photodegradation in both aqueous solution and soil. The gradual release of the herbicide from ECF reduces the amount of herbicide available to be photodegraded. Moreover, an additional reduction in the rate of NFZphotodecomposition in aqueous solution was observed in the presence of colloidal soil components such as goethite and humic acids, but not in the cases of montmorillonite of fulvic acids, that was similar to that of ECF alone. UV screening effect was mainly responsible for the photoprotective effect of goethite and humic acids. Adsorption of NFZ to humic acids also contributed to the photoprotective effect. NFZ photodegradation in aqueous suspensions were well explained by a simple ﬁrst order model. DT50 value from CF (3 h) was really lower than those obtained from (ECF): 35 h for ECF; 260 h for ECF–goethite; 53 h for ECF–humic acids; 33 h for ECF–montmorillonite; and 28 h for ECF–fulvic acids. These lab-scale ﬁndings prove that the use of ethylcellulosemicroencapsulated formulations could reduce the herbicide dosage, since they can minimize the pesticide losses by photolysis. These facts would be especially advantageous during the ﬁrst hours after foliar and soil application, besides the reduced losses of the herbicide due to leaching and dissipation observed in previous papers.

[1] D.V. Sojic, V.N. Despotovic, N.D. Abazovic, M.I. Comor, B.F. Abramovic, Photocatalytic degradation of selected herbicides in aqueous suspensions of doped titania under visible light irradiation, J. Hazard. Mater. 179 (2010) 49–56. [2] A. Kiss, D. Virag, Photostability and photodegradation pathways of distinctive pesticides, J. Environ. Qual. 38 (2009) 157–163. [3] S. Chiron, L. Comoretto, E. Rinaldi, V. Maurino, C. Minero, D. Vione, Pesticide by products in the Rhone delta (Southern France). The case of 4-chloro-2methylphenol and of its nitroderivative, Chemosphere 74 (2009) 599–604. [4] T. Katagi, Photodegradation of pesticides on plant and soil surfaces, Rev. Environ. Contam. Toxicol. 182 (2004) 1–189. [5] S. Gupta, V.T. Gajbhiye, R.K. Gupta, Effect of light on the degradation of two neonicotinoides viz acetamiprid and thiacloprid, Bull. Environ. Contam. Toxicol. 81 (2008) 185–189. [6] B. Eyheraguibel, A. Ter Halle, C. Richard, Photodegradation of bentazon, clopyralid, and triclopyr on model leaves: importance of a systematic evaluation of pesticide photostability on crops, J. Agric. Food. Chem. 57 (2009) 1960–1966. [7] J. Thomas, A. Bejjani, B. Nsouli, A. Gardon, J.M. Chovelon, In situ studies of pesticides photodegradation on soils using PD-TOFMS technique: application to norﬂurazon and oxyﬂuorfen, Int. J. Mass Spectrom. 279 (2009) 59–68. [8] J.P. Da Silva, L.F.V. Ferreira, I. Osipov, I.F. Machado, Surface photochemistry of pesticides containing 4-chlorophenoxyl chromophore, J. Hazard. Mater. 179 (2010) 187–191. [9] P. Graebing, J.S. Chib, Soil photolysis in a moisture- and temperature-controlled environment. 2. Insecticides, J. Agric. Food Chem. 52 (2004) 2606–2614. [10] Y. Si, J. Zhou, H. Chen, D. Zhou, Y. Yue, Effects of humic substances on photodegradation of bensulfuron-methyl on dry soil surfaces, Chemosphere 56 (2004) 967–972. [11] V.R. Herbert, G.C. Miller, Depth dependence of direct and indirect photolysis on soil surfaces, J. Agric. Food Chem. 38 (1990) 913–918. [12] T. Undabeytia, S. Nir, E. Tel-Or, B. Rubin, Photostabilization of the herbicide norﬂurazon by using organoclays, J. Agric. Food Chem. 48 (2000) 4774–7479. [13] Y. El-Nahhal, T. Undabeytia, T. Polubesova, Y.G. Mishael, S. Nir, L. Marguiles, B. Rubin, Organo-clay formulations of pesticides: reduced leaching and photodegradation, J. Appl. Clay Sci. 18 (2001) 309–326. [14] M. Kamiya, K. Nakamura, Cyclodextrin inclusion effects on photodegradation rates of organophosphorous pesticides, Environ. Int. 21 (1995) 299–304. [15] R.J. Demchak, R.A. Dybas, Photostability of abamectin/zein microspheres, J. Agric. Food Chem. 45 (1997) 260–262. [16] P.L. Huston, J.J. Pignatello, Degradation of selected pesticide active ingredients and commercial formulations in water by the photo-assisted Fenton reaction, Water Res. 33 (1999) 1238–1246. [17] W. Ahrens, Norﬂurazon, in: Herbicide Handbook, 7th ed., Weed Sci. Soc. Am., Champaign, IL, 1994, pp. 9–11. [18] E. Morillo, C. Maqueda, R. Reinoso, T. Undabeytia, Effect of two organic amendments on norﬂurazon retention and release by soils of different characteristics, Environ. Sci. Technol. 36 (2002) 4319–4325. [19] M. Singh, S. Tan, D. Sharma, Leaching and sorption of norﬂurazon in soils as affected by cationic surfactants, Bull. Environ. Contam. Toxicol. 68 (2002) 901–907. [20] E. Morillo, T. Undabeytia, A. Cabrera, J. Villaverde, C. Maqueda, Effect of soil type on adsorption–desorption, mobility and activity of the herbicide norﬂurazon, J. Agric. Food Chem. 52 (2004) 884–890. [21] R. Fernández-Urrusuno, J.M. Ginés, E. Morillo, Development of controlled release formulations of alachlor in Ethylcellulose, J. Microencapsul. 17 (2000) 331–342. ˜ A. Cabrera, C. Maqueda, E. Morillo, Controlled release of the herbi[22] F. Sopena, cide norﬂurazon into water from Ethylcellulose, J. Agric. Food Chem. 53 (2005) 3540–3547. ˜ C. Maqueda, E. Morillo, Norﬂurazon mobility, dissipation, activity and [23] F. Sopena, persistence in a sandy soil as inﬂuenced by formulation, J. Agric. Food Chem. 55 (2007) 3561–3567. ˜ A. Cabrera, C. Maqueda, E. Morillo, Ethylcellulose formulations for [24] F. Sopena, controlled release of the herbicide alachlor in a sandy soil, J. Agric. Food Chem. 55 (2007) 8200–8205. ˜ C. Maqueda, E. Morillo, Inﬂuence of soil characteristics and formu[25] F. Sopena, lation on alachlor dissipation in soil, Soil Sci. Soc. Am. J. 72 (2008) 767–774. ˜ C. Maqueda, E. Morillo, Formulation affecting alachlor efﬁcacy and [26] F. Sopena, persistence in sandy soils, Pest Manag. Sci. 65 (2009) 761–768. ˜ T. Sánchez-Verdejo, J. Villaverde, S. Nir, E. Morillo, C. [27] T. Undabeytia, F. Sopena, Maqueda, Performance of slow release of formulations of alachlor, Soil Sci. Soc. Am. J. 74 (2010) 898–905.

F. Sope˜ na et al. / Journal of Hazardous Materials 195 (2011) 298–305 [28] C.W. Hubbs, T. Lavy, Dissipation of norﬂurazon and other persistent herbicides in soil, Weed Sci. 38 (1990) 81–88. [29] C. Maqueda, E. Morillo, T. Undabeytia, F. Martin, Sorption of glyphosate and Cu (II) on a natural fulvic acid complex: mutual inﬂuence, Chemosphere 37 (1998) 1063–1072. [30] L.A. Perez-Maqueda, J.M. Criado, J. Subrt, C. Real, Synthesis of acicular hematite catalyst with tailored porosity, Catal. Lett. 60 (1999) 151–156. [31] J. Villaverde, C. Maqueda, T. Undabeytia, E. Morillo, Effect of various cyclodextrins on the photodegradation of a hydrophobic herbicide in aqueous suspensions of different soil coloidal components, Chemosphere 69 (2007) 575–584. [32] Guidance document on estimating persistence and degradation kinetics from environmental fate studies on pesticides in EU registration. Report of the FOCUS Work Group on Degradation Kinetics, EC Document Reference Sanco/10058/2005 version 2.0, 2006. [33] J. Villaverde, W. van Beinum, S. Beulke, C.D. Brown, The kinetics of sorption by retarded diffusion into soil aggregate pores, Environ. Sci. Technol. 43 (2009) 8227–8232. [34] M.P. Frank, P.G. Graebing, J.S. Chib, Effect of soil moisture content and sample depth on photolysis, J. Agric. Food. Chem. 50 (2002) 2607–2614. [35] R. Roberts, H.D. Huston, in: Royal Soc. of Chem. (Ed.), Metabolic Pathways of Agrochemicals: Herbicides and Plant Growth Regulators, Royal Soc. of Chem., Cambridge, UK, 1998, p. p. 411. [36] G.-G. Ying, B. Williams, Photodegradation of norﬂurazon in water, Toxicol. Environ. Chem. 71 (1999) 261–269. [37] J. Zhao, R. Wilkins, Controlled release of a herbicide from matrix granules based on solvent-fractionated organosolv lignins, J. Agric. Food Chem. 48 (2000) 3651–3661.

305

[38] K. Papadokostaki, S. Amarantos, J. Petropoulos, Kinetics of release of particulate solutes incorporated in cellulosic polymer matrices as a function of solute solubility and polymer swellability. I. Sparingly soluble solutes, J. Appl. Pol. Sci. 67 (1998) 277–287. [39] S. Gupta, V.T. Gajbhiye, R.K. Gupta, Effect of ultraviolet and sunlight on persistence of thiamethoxam in soil, Pest. Res. J. 18 (2006) 211–214. [40] J. He, W. Ma, W. Song, J. Zhao, X. Qian, S. Zhang, J. Yu, Photoreaction of aromatic compounds at ␣-FeOOH/H2 O interface in the presence of H2 O2 : evidence for organic-goethite surface complex formation, Water Res. 39 (2005) 119–128. [41] X. Zhao, M. Quan, H. Zhao, S. Chen, Y. Zhao, J. Chen, Different effects of humic substances on photodegradation of p,p -DDT on soil surfaces in the presence of TiO2 under UV and visible light, J. Photochem. Photobiol. A 167 (2004) 177–183. [42] M. Kamiya, K. Kameyama, Photochemical effects of humic substances on the degradation of organophosphorous pesticides, Chemosphere 36 (1998) 2337–2344. [43] M. Elazzouzi, A. Bensaoud, A. Bouhaouss, A. Dahchour, P. Meallier, A. Piccolo, Photodegradation of imazapyr in the presence of humic substances, Fresen. Environ. Bull. 8 (1999) 478–485. [44] C. Zamy, P. Mazellier, B. Legube, Phototransformation of selected organophosphorous pesticides in dilute aqueous solutions, Water Res. 38 (2004) 2305–2314. [45] J.R. Gabin, D.M.P. Milori, M.L. Simoes, W.T. da Silva, L.M. Neto, Inﬂuence of humic substances on the photolysis of aqueous pesticide residues, Chemosphere 66 (2007) 1692–1698. [46] R. Mathew, S. Khan, Photodegradation of metolachlor in water in the presence of soil mineral and organic constituents, J. Agric. Food Chem. 44 (1996) 3996–4000.

Journal of Hazardous Materials 195 (2011) 306–317

Contents lists available at SciVerse ScienceDirect

Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat

Water quality assessment in the rivers along the water conveyance system of the Middle Route of the South to North Water Transfer Project (China) using multivariate statistical techniques and receptor modeling Siyue Li ∗ , Jia Li, Quanfa Zhang ∗ Key Laboratory of Aquatic Botany and Watershed Ecology, Wuhan Botanical Garden, the Chinese Academy of Sciences, Wuhan 430074, China

a r t i c l e

i n f o

Article history: Received 13 February 2011 Received in revised form 15 August 2011 Accepted 15 August 2011 Available online 22 August 2011 Keywords: Water quality assessment Multivariate statistic techniques Source apportionment Receptor modeling Anthropogenic activities

a b s t r a c t A total of 190 grab water samples were collected from 19 rivers along the water conveyance system of the Middle Route of China’s interbasin South to North Water Transfer Project (MRSNWTP). Multivariate statistics including principal component/factor analysis (PCA/FA), analysis of variance (ANOVA), and cluster analysis (CA) were employed to assess water quality, and the receptor model of factor analysismultiple linear regression (FA-MLR) was used for source identiﬁcation/apportionment of pollutants from natural processes and anthropogenic activities to river waters. Our results revealed that river waters were primarily polluted by CODMn , BOD, NH4 + -N, TN, TP, and Cd with remarkably spatio-temporal variability, and there were increasing industrial efﬂuents in rivers northward. FA/PCA identiﬁed four classes of water quality parameters, i.e., mineral composition, toxic metals, nutrients, and organic pollutants. CA classiﬁed the selective 19 rivers into three groups reﬂecting their varying water pollution levels of moderated pollution, high pollution, and very high pollution. The FA-MLR receptor modeling revealed predominantly anthropogenic inputs to river solutes in Beijing and Tianjin, i.e., 77% of nitrogen and 90% of phosphorus from industry, and 80% of CODMn from domestics. This study is critical for water allocation and division in the water-receiving areas using the existing rivers for MRSNWTP. © 2011 Elsevier B.V. All rights reserved.

1. Introduction The impressive growths of human populations and economic development have resulted in the current worldwide deterioration in water quality, particularly the elevation of certain nutrients leading to eutrophication and heavy metals in the aquatic environment [1–8]. Natural processes and anthropogenic activities such as precipitation and soil erosion, and domestic consumptions, industrial and agricultural activities largely contribute to chemical pollutants in the ﬂuvial systems [2,9,10], and diffuse sources by agricultural runoff have been increasingly of great concern for nutrients due to widely fertilizer overuse. Meanwhile, industrial developments, particularly electroplating, metallurgy, mining and mineral processing, have been changing the biogeochemical cycles of heavy metals [1,4]. Recent studies have also revealed that urban develop-

∗ Corresponding authors at: Key Laboratory of Aquatic Botany and Watershed Ecology, Wuhan Botanical Garden, The Chinese Academy of Sciences, Wuhan 430074, China. Tel.: +86 27 87510702; fax: +86 27 87510251. E-mail addresses: [email protected], [email protected] (S. Li), [email protected] (Q. Zhang). 0304-3894/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.jhazmat.2011.08.043

ment and hydrological impacts by climate change greatly degraded water quality [11,12]. Water pollution by nutrients and heavy metals contributes to biodiversity loss, environmental degradation, and human health hazard [13,14]. Intake of considerable amounts of metals could result in varying life threatening cancers and mental disease, e.g., Cd could cause kidney damage and cardiovascular disease [14]. Meanwhile, excessive nutrients can lead to water eutrophication, causing a hypoxia environment, the reductions of species diversity and microbial growth, mortality of benthic communities, and stress in ﬁshery resources [14]. Studies have indicated that many rivers/streams particularly in developing countries are heavily polluted due to industrial and municipal wastewater, as well as agricultural runoff [3,4,6,15], for example, tributaries in the Changjiang system [5–8], Gomti River in India [4], the Han River in South Korea [16,17], Dil Deresi stream in Turkey [15], Pisuerga River in Centre-North Spain [3], and a few rivers in Chile [18]. Despite that there are numerous studies on spatial and temporal variability in water quality in river systems [i.e., [3,6,7,17]], many studies investigated either spatial variations of water quality or temporal trends for a few selected sites [i.e., [3,7]], while others tended to be simply exercises in statistics [i.e., [19,20]]. Chang et al. [17] and Pizarro et al. [18] provided good cases on spatial and

S. Li et al. / Journal of Hazardous Materials 195 (2011) 306–317

307

Fig. 1. Locations of sampling sites along the water conveyance system of the Middle Route of the South to North Water Transfer Project, China (these selective rivers are within the four large basins of Changjiang, Huaihe, Yellow and Haihe Rivers).

temporal patterns of water quality; however, they did not quantify source proﬁles of pollutants. In fact, source contribution could help to verify the relations between landscape and water quality, as well as implement pollution management policies. A number of receptor-oriented source apportionment models have been developed to quantify the contributions of various sources to measured pollutants. The most widely used models included multiple linear regression with factor analysis (FA-MLR) [21], positive matrix factorization (PMF) [22], U.S. Environment Protection Agency UNMIX [23], and chemical mass balance (CMB) [24]. The CMB model requires detailed knowledge of source types. Three multivariate receptor models (FA-MLR, PMF and UNMIX) can simultaneously analyse a series of observations and quantitatively assess the contributions of various sources to each observation without prior knowledge about number and nature of sources [25]. These models have been well-documented on assessment of air pollutants [21,24–27]; however, studies on river quality concentrated on qualitatively potential pollutant sources [i.e., [3,7,19,28,29]]. Recently, FA-MLR has been successively applied in source apportionment of heavy metals in surface waters [4,15,30]. In our study, the eigenvector model, FA-MLR, relatively simple and easily performed using common

software package, was therefore used for large variety of water variables. China possesses the total water resources of 2800 billion m3 , ranking the sixth in the world, but only 2300 m3 of water occupation per capita, amounting to only 1/4 of the world’s average. The uneven spatial distribution of water resources and water pollution further complicate the water shortage particularly in the Northwest and Northern China. The Changjiang River (Yangtze) basin and southern China, with a cultivated land of less than 40%, yields a runoff accounting for more than 80% of the national total. Conversely, the Yellow River, Huaihe River, Haihe River basins and northwest inland have half of national total area and 45% of total cultivated land and 36% of the total population, but only possess less than 12% of the total water resources [31–33]. Since 2002, China has been implementing the three-route South–North Water Transfer Project (i.e., East, Middle and West) (SNWTP), transferring water from the water-rich Yangtze River to Northwest and North China. The Middle Route, with a total length of 1230 km, will divert 13 billion m3 per year of water from the upper Han River to the North China Plain including the municipalities of Beijing and Tianjin. The water conveyance system, crossing Henan, Hebei, Beijing and Tianjin, intersects four large rivers of

308

S. Li et al. / Journal of Hazardous Materials 195 (2011) 306–317

China including the Yangtze, Huaihe, Yellow and Haihe Rivers (Fig. 1). Research studies in relation to the interbasin water transfer project have been focused on water quality [5–9,34,35] and the relations between land use/land cover and water chemistry [10,36] in the water source area (the upper Han river), and demonstrated chemical pollutants including nitrogen, chemical oxygen demand (CODMn ) and several heavy metals in the Han river. However, water quality and associated source identiﬁcation and apportionment in the rivers along the water conveyance system is unavailable. This information is critical for water allocation and division in the waterreceiving areas using the existing rivers. The present study was carried out to ﬁrstly investigate water quality in the rivers along the water conveyance canal of the Middle Route of SNWTP (MR-SNWTP). The objectivities of the study were to (1) assess the spatial and temporal patterns of water quality in the rivers using multivariate statistical analyses, and (2) quantify natural and anthropogenic sources to water pollution in the rivers using the receptor model of FA-MLR. The results were expected to ﬁll the knowledge gap of water contamination status, distribution and source apportionment of chemical pollutants in rivers along the canal system and ultimately help develop water management and conservation strategies in the water-receiving areas of the interbasin water transfer project.

2. Materials and methods 2.1. Study area

Fig. 2. Waste water, COD discharges and relative importance of industrial and domestic contributions to COD from 2002 to 2009 in Henan, Hebei, Beijing and Tianjin, respectively. Source: [37].

2.2. Sampling and analysis Four sampling campaigns were conducted in 19 rivers along the canal system of the MR-SNWTP during 2006–2007 (September and December 2006, and April and June 2007), and there were 10 rivers in Hennan, four rivers in Hebei, two rivers in Beijing and three rivers in Tianjin, respectively (Table 1; Fig. 1). Three water samples in a distance of 1 km from upper to downstream in each river were 80

Land use composition (%)

The water conveyance system of MR-SNWTP (canal system), ca. 1230 km long, starting at the Danjiangkou Reservoir, goes across the Tangbaihe plain, and passes through the pediment plain and the south foot of the Funiu Mountain (Fig. 1). It then crosses the Yellow River in the west of Zhengzhou city, goes further northward along the pediment plain at the east foot of the Taihang Mountain and the west side of the Beijing–Guangzhou Railway and ﬁnally reaches its destiny of Beijing and Tianjin [31]. The water-receiving region of the MR-SNWTP is affected by the continent monsoon climate with subtropic and warm temperate zones, the annual rainfall decreases from south to north (800–400 mm). There is large inter- and intra-annual variability in precipitation, and the ﬂood season occurs in July, August and September (80% of the annual total in south). Meanwhile, evaporation capacity increases from 1073 mm (Tanbaihe Plain) to 1190 mm (Haihe basin) northward due to transformation of humid-subtropic to semi-arid warm temperate climate. The annual mean temperature is 11.5–14.9 ◦ C with an increase of 0.02–0.04 ◦ C/yr, and the highest and lowest daily values of 40 ◦ C and −27.4 ◦ C in a period of 1951–2000, respectively. Water-receiving region of the water transfer project covers a total area of approximately 150,000 km2 and includes four metropolises such as Zhengzhou, Shijiazhuang, Beijing and Tianjin, and other 19 medium-sized cities. Thus, the region is highly urbanized with rapid economic development in the past decade, resulting in large amount of waste water discharges and chemical oxygen demand (COD) (Fig. 2). Waste water (together industrial and domestic) and CODMn discharges in the region account for about 12% and 11% of the nation’s total, respectively [37]. There is intensive land utilization along the canal system, indicated by 78.7% of agriculture and 2.2% of urban, respectively in an area of 6110 km2 along the water conveyance system of the MR-SNWTP. Vegetated land including forest and shrub covers about 7.1% and others including grass lands and waters together account for about 12% of the total land area (Fig. 3) [38]. Agriculture is more intense in the Henan and Hebei provinces while industry is much more developed in Beijing and Tianjin.

70 60 50 40 30 20 10 0 Forest

Shrub

Wild grass Agriculture

Waters

Urbans

Fig. 3. Land use/land cover composition in an area of 6110 km2 along the water conveyance system of the MR-SNWTP. Source: [38].

S. Li et al. / Journal of Hazardous Materials 195 (2011) 306–317

309

Table 1 Rivers and geo-locations of sampling site along the water conveyance canal. No.

River

Province

a.s.l (m)

Longitude

Latitude

Yangtze 1 2 3

Tuanhe Zhaohe Yahe

Henan Henan Henan

123 173 153

112◦ 07.469 112◦ 10.631 112◦ 37.973

32◦ 42.078 33◦ 00.049 33◦ 17.059

Huaihe 4 5 6 7

Shahe Beiruhe Yinhe Jialuhe

Henan Henan Henan Henan

126 99 116 106

112◦ 55.978 112◦ 13.523 112◦ 32.591 113◦ 46.643

33◦ 42.785 33◦ 55.299 33◦ 07.007 34◦ 50.553

Yellow 8 9 10

Qinhe Qihe Weihe

Henan Henan Henan

99 75 65

113◦ 23.998 114◦ 16.899 114◦ 21.706

35◦ 03.317 35◦ 36.239 35◦ 31.195

Haihe 11 12 13 14 15 16 17 18 19

Zhihe Jiaohe Taipinghe Menglianghe Liulihe Yongdinghe Beiyunhe Duliujian Ziyahe

Hebei Hebei Hebei Hebei Beijing Beijing Tianjin Tianjin Tianjin

51 65 67 61 21 88 10 18 28

114◦ 33.746 114◦ 30.234 114◦ 31.188 114◦ 54.635 116◦ 01.378 116◦ 16.147 117◦ 03.596 117◦ 00.959 117◦ 00.951

37◦ 22.521 37◦ 55.222 38◦ 06.518 38◦ 29.887 39◦ 36.208 39◦ 52.879 39◦ 22.257 39◦ 01.358 39◦ 01.359

collected. Ultimately, a total of 190 grab water samples, taken at a depth of approximate 10 cm using previously acid-washed 5 l high density polyethylene (HDPE) containers, were pretreated for laboratory measurements. A 1000 ml sub-sample was ﬁltered through pre-washed 0.45 ␮m Millipore nitrocellulose ﬁlters. The initial portion of the ﬁltrate was discarded to clean the membrane and the following ones were stored using previously acid-washed HDPE bottles. The two aliquots, one acidiﬁed to pH 2 using ultra-pure concentrated nitric acid for metal determination and another nonacidiﬁed for the anions measurements were prepared. The rest samples were acidiﬁed by concentrated sulfuric acid for the determination of biochemical oxygen demand (BOD), chemical oxygen demand (CODMn ), total nitrogen (TN), and total phosphorus (TP). All samples were stored in a fridge at 4 ◦ C before analysis, and the analytes were ﬁnished in the two weeks after the 10-day ﬁeld work. Water temperature (T), dissolved oxygen (DO), pH, oxidationreduction potential (ORP), electrical conductivity (EC), total dissolved solid (TDS), turbidity, and ammonium-nitrogen (NH4 + -N) of water samples were detected in situ using multiparameter water quality monitoring instrument (YSI Incorporated, Yellow Springs, Ohio, USA). Calibration of sensors was performed before every survey. The membranes used for ﬁltration were dried at 63 ◦ C to constant mass, and total suspended solid (TSS) was calculated from the difference in the ﬁlter paper weights before and after ﬁltering. Bicarbonate (HCO3 − ) was titrated using hydrochloric acid on the sampling day, while chloride (Cl− ), sulfate (SO4 2− ) and hardness (Hard) were determined in laboratory by titration using silver nitrate, barium chromate and EDTA, respectively. BOD was the differences of dissolved oxygen of the samples before and after the 5-day incubation at 20 ◦ C in a biochemistry incubator, and CODMn was analysedanalysed by potassium permanganate method. TP was analysedanalysed by digestion and a colorimetric method (ammonium molybdenum blue method/ascorbic acid method) after the samples were digested with concentrated nitric and sulfuric acids. The concentration of the TN was determined by alkaline potassium persulfate oxidation-UV spectrophotometric method. Heavy metals including Cu, Zn, Cd, Mn, and Fe were determined using ﬂame atomic absorption spectrometry (FASS; WYX-9004). All the procedures were strictly following Chinese State Environment Protection Bureau (CSEPB) [39]. Quality control procedures, including internal

quality control using reference materials, were employed to prove the validity of the measured results. 2.3. Statistical analyses Multivariable statistic approach, including factor and principal component analyses (FA/PCA) has been widely used to characterize spatial and temporal variations [4,7,8], and they are capable of exploring the hidden complex and relations between features in a data set through reducing the dimensionality of the data set. This reduction generates new orthogonal (non- correlated) variables, and the principal components (PCs) are arranged in a descending order of importance for explaining variance of all original property. FA further reduces the contribution of variables with minor signiﬁcance obtained by PCA and the new group of variables called varifactors (VFs) are extracted through varimax rotated PCA. These reduced PCs/VFs can describe a large set of original variables without losing much information [3,4,28,29] and the PCs/VFs can be interpreted as origins or common sources of environmental pollutants. Cluster analysis (CA) is an unsupervised pattern recognition approach and classiﬁes objects into categories or clusters on the basis of similarities and dissimilarities of samples. In the present study, Q-model hierarchical agglomerative CA without any prior knowledge of number of clusters was performed on the normalized data set by means of the Ward’s method using squared Euclidean distance as a measure of similarity [3,4]. One way analysis of variance (ANOVA) was employed to compare the seasonal and spatial differences of water quality variables (p < 0.05; leastsigniﬁcance difference, LSD). All the statistical procedures were conducted using statistical product and service solution (SPSS) 15.0 for Windows. 2.4. Receptor modeling (FA-MLR) After determination of possible sources inﬂuencing the river water quality using PCA/FA, a receptor modeling based on multiple linear regression of factor score (FA-MLR) was used to quantify the contribution of each pollutant. Detailed description including operating principle and procedures about the FA-MLR model could be found in Thurston and Spengler [21] and Guo et al. [26]. This

310

S. Li et al. / Journal of Hazardous Materials 195 (2011) 306–317

Fig. 4. Spatial and temporal variability of water quality in the selective 19 rivers along the water conveyance system of the MR-SNWTP.

multivariate receptor model, conducted in SPSS software, used the absolute factor scores as independent variables, while the chemical concentrations as the dependent variables.

3. Results and discussion 3.1. Seasonal and spatial variations of water quality Water quality variables including physico-chemicals, nutrients, major inorganic matters, and heavy metals were illustrated in Fig. 4 and Table 2. pH varied between 4.7 and 9 with water acidiﬁcation

in sites 1 and 18. Orp varied from −301 to 191 mV and the 95% conﬁdence interval for the average was −27 to 4 mV. EC and TDS showed the similar trends and their maximum concentrations occurred in site 18. TSS and turbidity seemed to have higher concentrations in the intermediate sites, averaging 85.3 mg/l and 28.7 NTU, respectively. NH4 + -N and TN displayed highest concentrations in site 7. CODMn had its highest concentration (424 mg/l, site 5) in September, and BOD exhibited the highest concentration (33 mg/l, site 7) in April, while TP and hardness in summer. Heavy metals tended to increase northward. Furthermore, variables such as pH, EC, TDS, nitrogen, turbidity, CODMn , BOD, Cd and Fe of three transverse sections in a distance of 1000 m of the same river exhibited

Table 2 Seasonal concentrations (mean values with standard errors (S.E.)) for water quality of rivers along the canal system, as well as comparison with environmental quality standards for surface water (GB3838-2002) in China. September 2006

April 2007

Totala

June 2007

Grade

n

Mean

S.E

n

Mean

S.E

n

Mean

S.E

n

Mean

S.E

Mean ±S.E

19 19 19 19 19 19 13 19 19 19 19 11 19 19 19 19

23.41 1249.05 811.11 7.87 −30.74 27.62 27.86 68.09 12.07 5.9 44.48 0.63 3.79 145.85 213.25 266.81

0.35 269.96 174.85 0.12 23.59 7.74 7.73 21.91 1.33 2.59 1.52 0.17 0.49 50.21 67.79 28.82

56 56 56 56 56 56 56 56 56 56 56 56 56 56 53 56 46 56 56 53 23

19.08 1380.68 902.05 7.73 −68.46 30.39 136.73 11.8 6.21 10.76 88.36 0.64 4.17 152.61 117.04 289.38 0.14 0.08 0.02 0.13 0.1

0.54 191.28 124.42 0.05 16.45 7.15 36.53 1.17 1.11 2.74 4.71 0.14 0.29 31.86 10.4 21.3 0.01 0.01 0 0.02 0.02

0.29 333.54 216.72 0.07 12.71 7.47 7.75 0.83 1.03 1.26

0.01 0 0.08 0.07

0.32 153.09 100.57 0.07 8.93 4.09 32.22 10.74 0.39 0.39 2 0.09 0.27 29.89 4.78 16.64 0.01 0.01 0 0.01 0.02

27.34 1425.06 927.77 7.69 0.02 34.71 34.45 8.5 7.53 5.29

0.02 0.02 0.17 0.27

5.64 1282.26 822.58 7.84 41.31 21.68 90.17 69.03 4 1.7 61.45 0.51 4.04 157.96 65.54 292.42 0.05 0.04 0.02 0.12 0.17

53 53 53 53 53 53 44 53 53 53

9 19 9 5

57 57 57 57 57 57 20 57 57 57 57 17 57 57 57 57 30 57 42 57 45

53 53 53 49 53 11 5 53 53 50

0.73 5.32 162.41 141.76 319.7 0.01 0.05 0.02 0.15 0.06

0.13 0.63 36.34 15.36 25.41 0 0.02 0 0.05 0.02

17.8 1349.6 875.6 7.8 −11.1 28.7 85.3 34.3 6.5 5.9 70.4 0.66 4.4 156.4 117.6 296.7 0.09 0.06 0.02 0.13 0.12

± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ±

0.7 123.6 80.4 0.04 7.8 3.4 16.8 4.5 0.5 1 2.6 0.08 0.2 17.5 9.5 11.4 0.01 0.006 0.001 0.02 0.01

Min

Max

0.5 244 66 4.7 −300.9 0 0 1.7 0 0.01 29.72 0 1.25 2.6 1.49 43.9 ND ND ND 0.01 ND

32.8 12527 8141 8.97 191.3 242 1380 424 33 78.7 193.68 3.62 22.9 1159.4 1353.5 665.1 0.35 0.37 0.07 2.08 0.74

I

I

III

IV

V

15 3 0.5 0.5 0.1

20 4 1 1 0.2

30 6 1.5 1.5 0.3

40 10 2 2 0.4

750b 6–9

15 3 0.15 0.2 0.02 4.5b 250b 250b 300b 0.01 0.05 0.001 0.1b 0.3b

1 1 0.005

1 1 0.005

1 2 0.005

1 2 0.01

S. Li et al. / Journal of Hazardous Materials 195 (2011) 306–317

T (◦ C) EC (␮s/cm) TDS (mg/l) pH Orp (mV) Turb (NTU) TSS (mg/l) CODMn (mg/l) BOD (mg/l) NH4 + -N (mg/l) TN (mg/l) TP (mg/l) Hard (mmol/l) Cl− (mg/l) SO4 2− (mg/l) HCO3 − (mg/l) Cu (mg/l) Zn (mg/l) Cd (mg/l) Mn (mg/l) Fe (mg/l)

December 2006

Grade I: clean water from headwater and national conservation area that can be used for domestic purposes after simple disinfection, for recreational purposes and irrigation. Grade II: fairly clean water that can be used as domestic water after treatment, for recreational purposes, for ﬁsh farming etc., and the area is strictly protected. Grade III: water also can be used for domestic, recreational purposes after suitable treatment. Grade IV: polluted water which can only be used as industrial water after treatment. Grade V: heavily polluted water that should not be used at all. a Total average; ND, lower than detection limit. b Maximum desirable concentration.

311

312

S. Li et al. / Journal of Hazardous Materials 195 (2011) 306–317

Table 3 Comparison of water quality in the present study with other rivers. Indiad

China

T (◦ C) EC (␮s/cm) TDS (mg/l) pH Orp (mV) Turb (NTU) TSS (mg/l) CODMn (mg/l) BOD (mg/l) NH4 + -N (mg/l) TN (mg/l) TP (mg/l) Hard (mmol/l) Cl− (mg/l) SO4 2− (mg/l) HCO3 − (mg/l) Cu (mg/l) Zn (mg/l) Cd (mg/l) Mn (mg/l) Fe (mg/l) Reference a b c d e f g h

Our study

Han Rivera

17.8 1349.6 875.6 7.8 -11.1 28.7 85.3 34.3 6.5 5.9 70.4 0.66 4.4 156.4 117.6 296.7 0.09 0.06 0.02 0.13 0.12

21.3 285.6 185.3 8.2 60.3 31.1 26.8 2.5 1.1 0.3

1.3 6.3 32.1 148 0.013 0.002 0.031 0.031 [5,7]

Jinshui Riverb

c

Yangtze

Gomti River 26.7 421.71 264.22 8.35

137.3 7.8

Patancheru industrial town

2608.5 7.8

Han River, South Koreae

Turkeyf

Spaing

Chileh

Gurogongdan

Yangjae

Dil Deresi

Pisuerga River

Elqui

17.3 727

14.6 462

7.3

7.6

70.3 51 95.8

10.3 8.6 9.7

27.2 3.9

11.4 0.5

14.3 629 427 7.6

3.2 1.7 0.17 0.12 1 5.68 19.72 109.8 0.01

0.17 [41]

1.95 0.07

0.0084 0.019 0.0003 1.66 [42,43]

42.7 16.09 5.16 0.17 2.82 0.22 1.8 12.34 18.41 0.042 0.121 0.001 0.161 4.98 [4]

5 3.7 1.3

2.6 28.3 112.7 156.1 0.037 0.7 0.008

0.099 0.073 0.16 [20]

[16]

[16]

4.03 [15]

445.7 6.1 0.028

0.04 0.11 [3]

[18]

Han River is the water source area of the MR-SNWTP. The Jinshui River is located in the headwater of the Han River, China. Yangzte: N and P in Nanjing section from Muller et al. [42]; while metals from Wang et al. [43]. Polluted rivers in industrial areas. Sampling sites locating in Seoul, South Korea. Polluted rivers in industrial areas. Polluted rivers in industrial areas. Very high-polluted.

large differences, and these sampling sites (i.e., sites 18 and 19) were primarily distributed in the Haihe River near Beijing and Tianjin city. Compared to China’s environmental quality standards for surface water (GB3838-2002) [[40]; Table 2], rivers were heavily polluted by CODMn , BOD, NH4 + -N, TN, TP, and Cd. By comparing with results in the water source area of the MRSNWTP, abroad rivers (Table 3), and world averages (i.e., 1 ␮g/l for Cu, 10 ␮g/l for Zn, 0.02 ␮g/l for Cd and 6 ␮g/l for Mn) [44], chemical species along the canal system had considerably higher mean concentrations. The concentration of CODMn in our study was ca. 14

times higher than in the upper Han River (China), 20 times higher than in an unpolluted river (the Jinshui River in China). This could be explained by high vegetated coverage of 77% in the upper Han river, 96% for the Jinshui River [5]. CODMn was even two times higher than in the Gomti River in India, four times higher than in the Han River (Seoul, Korea), and seven times higher than the polluted river in Spain. TN concentration in the considered region was 35 times higher than in the Yangtze, 25 times higher than in the Gomti River in India, 2.6 times higher than in the high polluted section in the Han River in South Korea. Considering the priority toxic pollutants, i.e,

Table 4 ANOVA for water quality in each sampling time along the canal system, China.

T EC TDS pH Orp Turbidity TSS CODMn BOD NH4 + -N TN TP Hard Cl− SO4 2− HCO3 − Cu Zn Cd Mn Fe

Sum of squares

df

Mean square

F

p-value

13945.49 806392.72 422730.79 0.84 354647.83 4910.69 305207.65 154064.25 1006.40 2346.44 35397.21 0.65 62.23 4965.38 356919.84 49064.68 0.22 0.05 0.00 0.05 0.40

3 3 3 3 3 3 3 3 3 3 2 3 3 3 3 3 2 3 3 3 3

4648.50 268797.57 140910.26 0.28 118215.94 1636.90 101735.88 51354.75 335.47 782.15 17698.61 0.22 20.74 1655.13 118973.28 16354.89 0.11 0.02 0.00 0.02 0.13

551.03 0.09 0.12 1.14 12.42 0.77 2.84 17.27 7.77 4.68 27.84 0.26 2.22 0.03 8.29 0.68 25.42 4.72 0.82 0.27 7.56

0.000 0.963 0.951 0.334 0.000 0.513 0.041 0.000 0.000 0.004 0.000 0.852 0.087 0.993 0.000 0.565 0.000 0.004 0.485 0.845 0.000

df: degrees of freedom. Signiﬁcance at p < 0.05.

S. Li et al. / Journal of Hazardous Materials 195 (2011) 306–317

313

Table 5 Varimax rotated factor loadings for water quality parameters in the four sampling seasons (the signiﬁcance of Kaiser–Meyer–Olkin (KMO) and Bartlett’s sphericity test is <0.001). September 2006

Component

Communality

1

2

3

4

Orp EC TDS Turbidity CODMn BOD NH4 + -N TN Hard Cl− SO4 2− HCO3 − Cd

−0.59 0.98 0.98 0.03 −0.1 −0.11 0.21 −0.42 0.93 0.93 0.94 0.14 −0.34

−0.18 0.09 0.09 0.01 −0.16 −0.08 0.77 0.54 0.29 −0.05 −0.17 0.93 0.15

0.19 −0.02 −0.02 −0.04 −0.17 −0.91 0.35 −0.07 0.05 0.01 −0.13 0.06 0.74

−0.48 0.11 0.11 0.89 −0.53 0.06 0.24 0.25 0.02 0.09 −0.01 −0.02 0.21

Eigenvalue Cumulative %

5.41 40.54

2.51 55.66

1.34 67.9

1.06 79.37

April 2007

Orp EC TDS Turbidity TSS CODMn BOD NH4 + -N TN TP Hard Cl− SO4 2− HCO3 − Cu Zn Mn Cd Eigenvalue Cumulative %

December 2006

0.65 0.98 0.98 0.79 0.34 0.84 0.81 0.54 0.95 0.88 0.93 0.88 0.73

Component

Orp EC TDS Turbidity CODMn BOD NH4 + -N TN Hard Cl− SO4 2− HCO3 − Zn Mn Cd Fe Eigenvalue Cumulative % Communality

1

2

3

4

5

−0.41 0.97 0.97 −0.08 0.02 0.59 −0.09 0.11 0.10 −0.08 0.94 0.94 0.24 −0.17 0.16 0.03 0.74 0.55 6.32 28.74

−0.73 0.12 0.12 0.69 0.38 0.72 0.84 0.80 0.94 −0.14 0.07 0.00 −0.49 0.90 0.04 −0.06 −0.16 −0.09 5.13 56.72

−0.33 0.14 0.14 −0.21 −0.01 −0.01 −0.19 −0.20 0.06 −0.32 0.22 0.19 0.55 0.18 0.73 −0.84 −0.30 0.73 2.14 71.43

−0.18 0.03 0.03 0.60 0.87 0.22 0.00 0.31 0.07 −0.21 0.03 −0.07 0.28 0.15 −0.34 −0.05 −0.03 −0.03 1.14 80.19

0.18 0.01 0.01 0.27 0.12 0.19 0.39 0.30 −0.06 −0.83 −0.02 −0.03 −0.38 −0.11 0.32 −0.09 0.10 −0.01 1.09 87.80

0.87 0.98 0.98 0.96 0.91 0.95 0.90 0.88 0.91 0.86 0.94 0.92 0.82 0.90 0.78 0.71 0.67 0.85

Component

Communality

1

2

3

4

5

0.03 0.98 0.97 0.06 −0.22 −0.26 0.1 −0.02 0.93 0.94 −0.19 −0.07 0.31 0.83 0.15 0.07 5.24 28.98

−0.45 0.04 0.04 0.11 −0.19 −0.37 0.86 0.8 0.21 −0.04 −0.34 0.78 0.68 0.07 −0.08 0.64 3.66 50.56

−0.62 0.02 0 0.87 −0.03 −0.72 0.06 0.35 −0.1 0.08 0.48 0.32 0 0.29 −0.18 −0.36 2.1 65.44

0.45 −0.07 −0.07 −0.14 0.87 −0.08 −0.13 0.06 −0.12 −0.05 −0.61 −0.05 −0.04 0.22 −0.07 0.03 1.29 74.56

−0.18 −0.09 −0.09 0.24 0.19 0.22 0.24 −0.08 −0.01 −0.27 0.21 0.1 −0.24 0.24 −0.83 0.46 1.06 83.35

June 2007

Orp EC TDS Turbidity TSS CODMn BOD NH4 + -N TP Hard Cl− SO4 2− HCO3 − Mn Cd Fe Eigenvalue Cumulative %

Component

0.84 0.98 0.95 0.84 0.88 0.78 0.82 0.78 0.94 0.96 0.79 0.74 0.62 0.88 0.76 0.76

Communality

1

2

3

0.33 0.96 0.96 0.33 0.47 0.83 0.77 0.66 0.01 0.95 0.94 0.73 −0.26 0.94 0.32 0.78 8.34 49.97

−0.77 −0.05 −0.05 0.86 0.55 0.40 0.44 0.64 0.93 0.03 −0.11 −0.27 0.67 −0.01 −0.15 0.12 3.73 73.86

0.33 0.21 0.21 0.05 −0.12 −0.08 −0.02 0.18 0.18 0.08 0.16 −0.50 −0.33 0.28 0.84 0.20 1.23 83.16

0.81 0.97 0.97 0.84 0.54 0.86 0.79 0.87 0.89 0.91 0.92 0.85 0.62 0.97 0.83 0.67

Extraction method: Principal component analysis. Rotation method: Varimax with Kaiser normalization.

Cd, Cu, and Zn presented in US EPA, 2006 for aquatic life protection [45], their mean values showed that Cu concentration was about 90 times as high as world average, Zn was six times, and Cd was 100 times. Further, Cd concentration was evidently higher (3–67 times) than both home and abroad rivers, and comparable to a very highpolluted river by industrial efﬂuents in Chile. Thus, surface water in the water-receiving region (North China) was highly polluted and therefore lost their natural ecosystem service function. Analysis of variance indicated remarkable seasonal differences for water temperature, Orp, TSS, CODMn , BOD, NH4 + -N, TN, SO4 2− , Cu, Zn, and Fe, and the maximum concentrations of CODMn , BOD, SO4 2− and Fe occurred in autumn (Sept. 2006), while other variables showed their maximum concentrations in spring (Apr. 2007) (Tables 2 and 4). Due to highly urbanization along the canal, municipal and industrial wastes are primarily responsible for water pollution. In September and June (wet season), large amounts of rain runoff showed diluted effects though agricultural activities, partly contributed to riverine nutrients in Henan and Hebei. However, drought in the period from October to May together with large amounts of industrial and domestic waste water resulted in highest monthly nitrogen concentration in spring. This was conﬁrmed by industrial markers of metals, i.e., highest concentrations of Cu and Zn (Table 2). Result indicated that waters in varying sampling

time were always polluted by CODMn , BOD, NH4 + -N, TN, TP and Cd (Table 2), reﬂecting intense anthropogenic activities particularly industrial efﬂuents, e.g., waste water in Henan was 33 × 108 t/yr, 23 × 108 t/yr for Hebei, 11 × 108 t/yr for Beijing and 6 × 108 t/yr for Tianjin, respectively (Fig. 2). This was understandable considering more than 20 cities are located along the water conveyance system. FA/PCA extracted four, ﬁve, ﬁve and three principal components (PCs) for the measurement data in September and December 2006, and April and June 2007, respectively, and these PCs with eigenvalue >1 explained 79%, 83%, 88% and 83% of the total variance in the respective data sets (Table 5). For the data set pertaining to September 2006, the ﬁrst principal component (PC1) explained 40.5% of the total variance and had strong positive loadings on EC, TDS, Hard, Cl− and SO4 2− and moderate negative loadings on Orp, which could be interpreted as a mineral component of the river water [3,4]. These variables had a common origin in rock dissolution (limestone) and soils, and high Cl− and SO4 2− also indicated anthropogenic inputs to total solutes. PC2 explaining 15% of the total variance had strong positive loading on NH4 + -N and moderate positive loading on TN, representing the nutrient component in waters. The third and fourth components together explained 23.6% of the total variance and were characterized by BOD and Cd, and turbidity and CODMn , respectively. PC3 represented organic and metal

314

S. Li et al. / Journal of Hazardous Materials 195 (2011) 306–317

pollutants primarily caused by industrial efﬂuents, while domestic and industrial, and waste disposal for PC4. Similar to September 2006, EC, TDS, Hard, and Cl− in December 2006 and April 2007 were well correlated with PC1 and major pollutants were not included in the ﬁst component. As regards December 2006, PC2 represented nutrient pollution, and the third and fourth components had higher loadings on BOD and CODMn , representing biodegradable and organic pollutants. PC5 explaining relatively lower variance (<9%) was dominated by the toxic metalCd, and thus indicating the inﬂuence of industrial wastes. While for April 2007, the second component explaining 18% of variance included nitrogen and organic pollutants with positive loadings on CODMn and BOD and a negative loading on Orp, thus this component pointed to organic pollution from anthropogenic activities. Further, high concentrations of biodegradable and organic pollutants in urban wastewater consumed large amounts of oxygen, leading to anaerobic fermentation processes and higher levels of ammonium and organic acids. Hydrolysis of these acidic materials thus decreased water pH values (Fig. 4) [4]. In contrasts to other three seasons, PC1 in June 2007 explained the total variance of 50% and included EC, TDS, CODMn , BOD, NH4 + N, Hard, Cl− , SO4 2− , Mn and Fe (Table 4), reﬂecting the mixes of anthropogenic and natural processes. PC2 explaining 24% of total variance had strong positive loadings on turbidity, TP, and moderate positive loadings on TSS, NH4 + -N and HCO3 − , and a negative loading on Orp. This also indicated alga blooming and anaerobic environment and therefore lower Orp. PC3 (9.3% of total variance) represented heavy metal pollution (Cd) by industrial efﬂuents. Overall, FA/PCA demonstrated great seasonality for water quality parameters, i.e., CODMn , BOD, NH4 + -N, TN, TP and Cd, however, a few variables such as EC, TDS, Hard, Cl− and Mn were always in PC1. This is consistent with the results that seasonal differences in concentrations are statistically signiﬁcant shown in ANOVA test reported in Table 4. Moreover, the correlation coefﬁcients between PCs and water quality parameters in Table 4 revealed the relative importance of a variable in a PC [8]. In this study, an absolute correlation coefﬁcient value greater than 0.80 was considered to be an important variable inﬂuencing seasonal water quality. EC, TDS, Hard, Cl− and Mn in PC1 were the most important variables contributing to water quality variability, reﬂecting natural processes and great impacts of salts by domestics and Cl-contained industry such as dyeing, pharmaceutical and pesticide on water quality. Major pollutants such as CODMn , BOD, NH4 + -N, TN, TP and Cd had varying correlation coefﬁcients with PCs, indicating seasonal heterogeneity in anthropogenic activities and hydro-climatic variability, i.e., precipitation and agriculture concentrating in summer and autumn, as well as daily differences of industrial efﬂuents. Also, FA/PCA revealed four major groups of parameters: (1) EC, TDS, Hard, Cl− and Mn (mainly natural), (2) toxic metals (industrial), (3) nutrients (agricultural and industrial) and (4) organic pollutants (domestic, municipal and industrial efﬂuents). Three to ﬁve factors could explain 79%–88% of the total variance, thus, this statistical technique successfully deciphered possible pollution sources [3,4,7]. Hierarchical CA was employed to detect spatial similarity for sites and rendered a dendrogram grouping the sampling sites into three statistically signiﬁcant clusters at (Dlink/Dmax) × 100 <65 (Fig. 5). Cluster 1 (sites 1–6, 8, 9, 11, 13, 16), cluster 2 (sites 7, 10, 12, 14, 15, 17 and 19) and cluster 3 (site 18) corresponded to moderated pollution, high pollution, and very high pollution regions, respectively (Fig. 5). This clustering was conﬁrmed by the total factor scores (TFS) from FA versus sampling sites, reﬂecting their pollution levels [19], i.e., TFS for site 18 was the highest (1.6), and TFS with greater than 0 for sites 7, 10, 12, 14, 15, 17 and 19. Cluster 3 (site 18) with very high mineralization (4850 mg/l) was consistent

Fig. 5. Dendrogram showing clustering of sampling sites according to Ward’s method using squared Euclidean distance.

with the fact that this river went through the Tianjin city and thus was polluted by urban’s industrial and domestic wastes. ANOVA showed 75% of total variables with spatial signiﬁcant variations (Table 6), indicating remarkably spatial heterogeneity in anthropogenic activities and economic development, i.e., agriculture in Henan and Hebei, while Beijing and Tianjin with developed economy and high dense population. Pollutants thus sourced from a combination of agricultural, industrial and domestic processes in Henan and Hebei provinces, while from industrial and domestic activities in Beijing and Tianjin, i.e., 60 × 104 t/yr of COD for Henan and Hebei, and about 10 × 104 t/yr of COD for Beijing and Tianjin, respectively (Fig. 2). The Haihe River near Beijing and Tianjin city is one of the most contaminated rivers in China [Fig. 5; [37]]. 3.2. Source identiﬁcation Spatial heterogeneity of anthropogenic activities and economic development has resulted in diverse pollution source types in the region. In our work, precisely source identiﬁcation/apportionment was calculated based on geographic characterization. Two categories were therefore obtained: (1) sites 1–14 in Henan and Hebei, where agriculture accounted for more than 50% of its respective GDP (gross domestic product), and (2) sites 15–19 in Beijing and Tianjin with little agricultural activities. We hypothesized that pollutants in group 1 were contributable to agricultural, industrial and domestic sources, while those in group 2 were industrial and domestic in origin. Five PCs, with eigenvalue >1 and explaining 88% of the total variance, were extracted for sites in Henan and Hebei (group 1, Table 7). PC1 accounting for 29% of the total variance had strong loadings on Orp, EC, TDS, turbidity and TSS, and moderate positive loadings on NH4 + -N, TN, Hard and HCO3 − , indicating its association with dissolution of soil constituents mainly carbonates. Nitrogen variables and TSS were the primarily origins in run-off from ﬁelds with high load of soils and waste disposal activities. PC2 was heavily weighted

S. Li et al. / Journal of Hazardous Materials 195 (2011) 306–317

315

Table 6 Mean values with standard errors (S.E.) and ANOVA for water quality in different clusters of rivers along the canal system, China. Moderate polluted

T (◦ C) EC (␮s/cm) TDS (mg/l) pH Orp (mV) Turb (NTU) TSS (mg/l) CODMn (mg/l) BOD (mg/l) NH4 + -N (mg/l) TN (mg/l) TP (mg/l) Hard (mmol/l) Cl− (mg/l) SO4 2− (mg/l) HCO3 − (mg/l) Cu (mg/l) Zn (mg/l) Cd (mg/l) Mn (mg/l) Fe (mg/l)

High polluted

Very high polluted

ANOVA

n

Mean

S.E

n

Mean

S.E

n

Mean

S.E

Sum of squares

105 105 105 105 105 105 65 105 105 105 76 67 105 105 104 105 35 68 91 93 67

16.94 622.86 406.97 7.93 31.26 12.25 21.34 42.29 4.79 1.29 58.63 0.6 2.97 42.38 84.20 212.53 0.09 0.06 0.02 0.08 0.09

0.86 26.51 17.52 0.04 4.76 1.62 3.68 7.79 0.38 0.48 1.68 0.12 0.11 3.82 6.68 8.09 0.01 0.01 0 0.01 0.01

70 70 70 70 70 70 59 70 70 70 49 63 70 70 64 70 44 49 69 69 50

18.77 1566.81 1010.74 7.66 −76.08 51.51 155.86 22.81 8.48 11.54 89.78 0.73 5.15 204.58 141.38 448.52 0.09 0.06 0.02 0.1 0.13

1.14 63.62 43.3 0.05 15.03 7.56 35.47 1.55 1.06 2.2 5.39 0.1 0.12 16.55 11.00 14.23 0.01 0.01 0 0.01 0.02

10 10 10 10 10 10 9 10 10 10 7 7 10 10 10 10 8 10 10 10 6

19.13 7459 4850.1 6.83 −1.96 41.06 84 30.23 10.73 14.78 62.98 0.57 14.5 1015.74 313.19 117.46 0.1 0.06 0.04 0.84 0.36

3.15 820.61 533.11 0.21 53.15 14.53 12.32 3.58 3.44 4.05 9.26 0.09 1.79 31.43 122.63 16.03 0.03 0.01 0 0.21 0.1

160.28 432006720.32 182304291.41 12.29 484819.71 66355.92 559673.76 16114.46 758.71 5243.25 29328.58 0.65 1271.45 8912211.72 534791.36 2678623.41 0.00 0.00 0.00 5.24 0.40

in BOD, NH4 + -N, TN and SO4 2− . Nitrogen fertilizers were widely used in Henan and Hebei provinces (China’s major grain production base), thus agricultural runoff could be dominantly responsible this PC, whereas, SO4 2− was primarily contributable to fossil fuel combustion for thermal power and acid rain in this region. The third and fourth PCs together explaining 29% of total variance were most dependent upon CODMn , Cl− , Cu, Mn, Cd, TP and Zn, representing heavy metal contamination, organic and reductive pollutants. This cluster was primarily due to industrial processes such as metallurgy, petrochemical plants, chemical fertilizer and pesticides etc. However, domestics were also largely contributed to CODMn and Cl− (salts). The last PC was interpreted as natural processes because of high loading on Hard (Ca and Mg) and Fe, and these elements were found abundant in earth crust. Three PCs explaining 80% of the total variance were extracted from the rivers in Beijing and Tianjin (group 2, Table 7). The ﬁrst

df

Mean square

F

p-value

2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2

80.14 216003360.16 91152145.70 6.14 242409.86 33177.96 279836.88 8057.23 379.35 2621.63 14664.29 0.32 635.73 4456105.86 267395.68 1339311.70 0.00 0.00 0.00 2.62 0.20

0.95 447.60 436.75 33.72 27.70 18.66 8.32 2.17 8.56 17.46 21.48 0.40 241.77 516.08 20.17 142.06 0.02 0.00 11.93 92.45 11.41

0.388 0.000 0.000 0.000 0.000 0.000 0.000 0.117 0.000 0.000 0.000 0.674 0.000 0.000 0.000 0.000 0.978 0.997 0.000 0.000 0.000

PC included more than half variables and explained 42% of total variance. These variables represented solutes, ammonium, major elements (Hard, Cl− and HCO3 − ), and heavy metals, thus, PC1 indicated the combinations of natural and anthropogenic sources. PC2 was well correlated with turbidity, CODMn , BOD, SO4 2− and Cu and contributed 21.5% to total variance. 90% CODMn in Beijing and 80% CODMn in Tianjin sourced from domestics (Fig. 2), hence, we mainly ascribed this PC to domestics. Coal combustion (primarily coal-ﬁre power plants, thermal power industry, and heating of buildings in winter) and motor vehicle exhaust partially contributed to SO4 2− and Cu. This is consistent with the fact that high pollution levels of atmospheric sulfur oxide occur in winter and spring in North China, i.e., SO2 in Beijing increased from 10 ppb in summer to 46 ppb in winter with the highest value of 113 ppb [46,47]. PC3 (16% of total variance) had strong positive loadings on TN and TP, representing nutrient pollution. Excess nutrients led to eutrophication and

Table 7 Varimax rotated factor loading and corresponding possible source type (the signiﬁcance of KMO and Bartlett’s sphericity test is <0.001). Sites 1–14

Component 1

2

3

4

5

Orp EC TDS Turbidity TSS CODMn BOD NH4 + -N TN TP Hard Cl− SO4 2− HCO3 − Cu Zn Mn Cd Fe Eigenvalue Cumulative %

−0.79 0.83 0.83 0.79 0.91 0.35 0.35 0.59 0.58 −0.32 0.60 0.38 0.04 0.70 −0.01 0.03 −0.11 0.14 0.13 8.29 29.13

−0.44 0.21 0.20 0.50 −0.02 0.29 0.84 0.72 0.77 −0.35 0.00 −0.15 −0.84 0.59 0.16 0.13 −0.05 −0.25 0.09 4.12 48.80

0.06 0.31 0.17 −0.07 −0.07 0.65 −0.18 −0.24 −0.12 −0.10 0.30 0.81 −0.35 0.22 −0.66 0.34 0.86 −0.70 −0.01 2.01 66.57

−0.08 0.04 −0.04 0.11 0.17 0.21 0.13 0.06 −0.01 −0.80 0.14 0.01 0.05 0.11 0.58 −0.79 −0.20 0.53 −0.13 1.26 77.63

−0.20 0.35 0.26 0.09 −0.16 0.46 −0.15 0.11 0.11 0.05 0.65 0.22 −0.16 0.29 −0.27 −0.02 −0.10 0.12 0.93 1.12 88.45

Communality

Sites 15–19

Component

Communality

1

2

3

0.88 0.96 0.82 0.90 0.89 0.88 0.90 0.95 0.95 0.88 0.88 0.88 0.86 0.98 0.86 0.77 0.81 0.87 0.91

Orp EC TDS Turbidity

−0.21 0.97 0.97 −0.08

−0.31 0.00 0.00 −0.82

−0.82 0.08 0.08 −0.10

0.81 0.95 0.95 0.68

CODMn BOD NH4 + -N TN TP Hard Cl− SO4 2− HCO3 − Cu Zn Mn Cd

0.21 −0.14 0.95 −0.04 −0.04 0.97 0.93 0.35 −0.76 0.14 0.79 0.72 0.73

−0.69 0.83 0.00 0.23 0.06 0.05 0.06 0.77 −0.38 0.85 −0.23 0.02 0.42

−0.18 0.04 0.08 0.92 0.95 0.03 0.03 0.26 0.47 0.09 −0.08 −0.05 0.13

0.55 0.70 0.90 0.90 0.91 0.95 0.88 0.78 0.94 0.74 0.68 0.52 0.73

Eigenvalue Cumulative %

7.31 41.90

4.10 63.47

2.18 79.94

Possible sources type: Sites 1–14, PC1, Natural (primarily) + anthropogenic; PC2, Agricultural + fossil fuel; PCs 3 and 4, Industrial + domestic; PC5, Natural. Sites 15–19, PC1, Natural + anthropogenic (primarily); PC2, Domestic 1 + industrial 1 (coal combustion, vehicle); PC3, Domestic 2 (detergent, faeces) + industrial 2 (petroleum reﬁning).

316

S. Li et al. / Journal of Hazardous Materials 195 (2011) 306–317

Table 8 Source contribution (in %) to elements calculated using FA-MLR technique. Sites 1–14

F1

F2

F3

F4

F5

R2

Sites 15–19

F1

F2

F3

R2

Orp EC TDS Turbidity TSS CODMn BOD NH4 + -N TN TP Hard Cl− SO4 2− HCO3 − Cu Zn Mn Cd Fe

50.05 48.06 55.22 50.85 68.12 17.78 21.08 34.51 36.47 19.51 35.46 23.95 2.99 36.79 0.77 2.27 8.56 8.24 10.25

27.81 12.01 13.40 31.86 1.52 14.76 51.27 41.98 48.64 21.44 0.03 9.72 58.31 30.86 9.51 9.67 3.47 14.07 7.10

4.08 17.71 11.69 4.50 5.61 33.31 10.89 13.71 7.51 6.38 17.74 51.55 24.23 11.66 39.22 26.09 65.44 40.05 0.79

5.23 2.32 2.37 7.26 13.00 10.65 7.77 3.62 0.36 49.37 8.30 0.58 3.57 5.62 34.59 60.69 15.09 30.58 9.88

12.84 19.90 17.32 5.54 11.75 23.50 8.99 6.18 7.02 3.30 38.48 14.20 10.90 15.07 15.91 1.27 7.44 7.06 71.98

0.84 0.94 0.77 0.87 0.86 0.84 0.87 0.93 0.93 0.84 0.84 0.84 0.81 0.97 0.82 0.70 0.81 0.84 0.89

Orp EC TDS Turbidity

15.44 91.87 91.71 8.44

23.06 0.46 0.43 81.58

61.50 7.67 7.86 9.98

0.77 0.94 0.95 0.63

CODMn BOD NH4 + -N TN TP Hard Cl− SO4 2− HCO3 − Cu Zn Mn Cd

19.67 41.74 91.52 2.98 4.21 92.16 90.68 25.35 54.33 12.89 71.33 91.65 56.87

63.71 46.79 0.37 19.56 5.46 5.11 6.30 55.59 8.69 79.06 21.18 2.45 33.00

16.62 11.46 8.11 77.46 90.33 2.73 3.01 19.06 36.98 8.05 7.49 5.90 10.12

0.49 0.65 0.89 0.88 0.89 0.94 0.85 0.74 0.94 0.69 0.62 044 0.68

Possible sources type: Sites 1–14, F1, Natural (primarily) + anthropogenic; F2, Agricultural + fossil fuel; F3 and F4, Industrial + domestic; F5, Natural. Sites 15-19, F1, Natural + anthropogenic (primarily); F2, Domestic 1 + industrial 1 (coal combustion, vehicle); F3, Domestic 2 (detergent, faeces) + industrial 2 (petroleum reﬁning).

alga bloom, hence anaerobic environment, thus this PC had strong negative loading on Orp. 3.3. Source apportionment Using the PCA results above, FA-MLR receptor modeling was applied to quantify the source contributions to each measured water quality variables (Table 8). It is worth noting that for each individual factor containing mixed sources, the tracer for a source was assumed to be exclusively emitted from that speciﬁc source. High R2 listed in Table 8 and the ratio of mean calculated to measured values around one suggested goodness of the receptor model on source apportionment. For group 1 (rivers in Henan and Hebei), the model showed that the ﬁrst factor was dominantly controlled by natural processes (e.g., soil weathering) which contributed 50% to solute load, while the ﬁfth factor (purely earth crust source) contributed 17% to river solutes, resulting in 67% of solutes apportioned to natural processes. Over 50% of nitrogen was attributable to agriculture, which was consistent with the fact of intense agricultural activities in these two provinces. Industrial and domestic sources contributed 70% of heavy metals and more than a half of CODMn , phosphorus and Cl− . The ﬁrst factor for the group 2 (rivers in Beijing and Tianjin) represented multiple sources due to that the major pollutants such as ammonium and toxic metals were primarily contributed by this factor, resulting smaller proportion of natural sources to riverine solutes. Domestic sources contributed 64% of CODMn and 47% of BOD, while SO4 2− (56%) was predominantly attributable to coal combustion and vehicle exhaust. The last factor was largely dominated by industrial efﬂuents, contributing 77% of nitrogen and 90% of phosphorus, though some domestic sources such as detergent and faeces also contributed to them. Considering CODMn partially attributed to domestic source of factor 3, CODMn from domestics was high as 80%, which was close to the results obtained from emission inventory (80–90%) in Beijing and Tianjin city (Fig. 2) [37]. 4. Conclusion There are considerably spatio-temporal variabilities in water quality in the rivers along the water conveyance system of the Middle Route of the South to North Water Transfer Project. Rivers northward were more polluted by industrial efﬂuents. Hierarchical cluster analysis in combination with total

factor scores versus sampling sites grouped the sampling sites into three clusters of similar characteristics between sampling sites corresponding to moderate, high and very high polluted rivers. Seasonal FA/PCA allowed four categories of parameters such as mineral composition (primarily natural), toxic metals (industrial), nutrients (agricultural, domestic and industrial) and organic pollutants (domestic, municipal and industrial sources). Multi-linear regression with factor analysis (FA-MLR) receptor modeling provided apportionment of various sources contributing to river pollution. For rivers in Henan and Hebei provinces, industrial and domestic sources contributed 70% of heavy metals and more than half of CODMn , phosphorus, and Cl− , while 50% of nitrogen from agriculture. Industrial efﬂuents were the major sources for nitrogen and phosphorous, and domestics contributed 80% to CODMn for rivers in Beijing and Tianjin. Clean techniques in industry and reduction of agricultural runoffs should be adopted to minimize pollutants to rivers. This study demonstrated usefulness of multivariate statistic techniques in water quality assessment, source identiﬁcation/apportionment, which would help obtain better information on the water quality for water conservation. Acknowledgements The research was funded by Dr. Zhang Q.F.’s “Hundred-talent Project” of the Chinese Academy of Sciences (O629221C01) and the National Natural Science Foundation of China (No. 31130010), Dr. Li S.Y.’s National Natural Science Foundation of China (No. 31100347) and the Youth Innovation Foundation of the Chinese Academy of Sciences, China. Special thanks are given to Professor Gianluca Li Puma and two anonymous reviewers. References [1] J.O. Nriagu, A history of global metal pollution, Science 272 (1996) 223–224. [2] S.R. Carpenter, N.F. Caraco, D.L. Correll, R.W. Howarth, A.N. Sharpley, V.H. Smith, Non-point pollution of surface waters with phosphorus and nitrogen, Ecol. Appl. 8 (3) (1998) 559–568. [3] M. Vega, R. Pardo, E. Barrado, L. Deban, Assessment of seasonal and polluting effects on the quality of river water by exploratory data analysis, Water Res. 32 (1998) 3581–3592. [4] K.P. Singh, A. Malik, S. Sinha, Water quality assessment and apportionment of pollution sources of Gomti river (India) using multivariate statistical techniques-a case study, Anal. Chim. Acta 538 (2005) 355–374. [5] S. Li, S. Gu, W. Liu, H. Han, Q. Zhang, Water quality in relation to the land use and land cover in the Upper Han River basin, China, Catena 75 (2008) 216–222. [6] S. Li, W. Liu, S. Gu, X. Cheng, Z. Xu, Q. Zhang, Spatio-temporal dynamics of nutrients in the upper Han River basin, China, J. Hazard. Mater. 162 (2009) 1340–1346.

S. Li et al. / Journal of Hazardous Materials 195 (2011) 306–317 [7] S. Li, Q. Zhang, Spatial characterization of dissolved trace elements and heavy metals in the Upper River (China) using multivariate statistical techniques, J. Hazard. Mater. 176 (2010) 579–588. [8] S. Li, Q. Zhang, Risk assessment and seasonal variations of dissolved trace elements and heavy metals in the Upper Han River, China, J. Hazard. Mater. 181 (2010) 1051–1058. [9] S. Li, Q. Zhang, Geochemistry of the upper Han River basin, China. 2. Seasonal variations in major ion compositions and contribution of precipitation chemistry to the dissolved load, J. Hazard. Mater. 170 (2009) 605–611. [10] S. Li, S. Gu, X. Tan, Q. Zhang, Water quality in the upper Han River basin, China: the impacts of land use/land cover in riparian buffer zone, J. Hazard. Mater. 165 (2009) 317–324. [11] J. Duh, V. Shandas, H. Chang, L. George, Rates of urbanization and the resilience of air and water quality, Sci. Total Environ. 400 (2008) 238–256. [12] S. Praskievicz, H. Chang, A review of hydrologic modelling of basin-scale climate change and urban development impacts, Prog. Phys. Geog. 33 (2009) 650–671. [13] L. Jarup, Hazards of heavy metal contamination, Br. Med. Bull. 68 (2003) 167–182. [14] K.K. Treseder, Nitrogen additions and microbial biomass: a meta-analysis of ecosystem studies, Ecol. Lett. 11 (2008) 1111–1120. [15] H. Pekey, D. Karakas, M. Bakoglu, Source apportionment of trace metals in surface waters of a polluted stream using multivariate statistical analyses, Mar. Pollut. Bull. 49 (2004) 809–818. [16] H.J. Chang, Spatial and temporal variations of water quality in the Han River and its tributaries, Seoul, Korea, 1993–2002, Water Air Soil Pollut. 161 (2005) 267–284. [17] H.J. Chang, Spatial analysis of water quality trends in the Han River basin, South Korea, Water Res. 42 (2008) 3285–3304. [18] J. Pizarro, P.M. Vergara, J.A. Rodriguez, A.M. Valenzuela, Heavy metals in northern Chilean rivers: spatial variation and temporal trends, J. Hazard. Mater. 181 (2010) 747–754. [19] T. Kowalkowski, R. Zbytniewski, J. Szpejna, B. Buszewski, Application of chemometrics in river water classiﬁcation, Water Res. 40 (2006) 744–752. [20] A.K. Krishna, M. Satyanarayanan, P.K. Govil, Assessment of heavy metal pollution in water using multivariate statistical techniques in an industrial area: a case study from Patancheru, Medak District, Andhra Pradesh, India, J. Hazard. Mater. 167 (2009) 366–373. [21] G.D. Thurston, N.D. Spengler, A quantitative assessment of source contributions to inhalable particulate matter pollution in metropolitan Boston, Atmos. Environ. 19 (1985) 9–25. [22] P. Paatero, Least squares formulation of robust, nonnegative factor analysis, Chemometr. Intell. Lab. Syst. 37 (1997) 23–35. [23] R.C. Henry, History and fundamentals of multivariate air quality receptor models, Chemometr. Intell. Lab. Syst. 37 (1997) 37–42. [24] USEPA, CMB 8.2 User’s Manual, Ofﬁce of Air Quality Planning and Standards, Research Triangle Park, NC, USA, 2000. [25] R.K. Larsen, J.E. Baker, Source apportionment of polycyclic aromatic hydrocarbons in the urban atmosphere: a comparison of three methods, Environ. Sci. Technol. 37 (2003) 1873–1881. [26] H. Guo, T. Wang, P.K.K. Louie, Source apportionment of ambient non-methane hydrocarbons in Hong Kong: application of a principal component analysis/absolute principal component scores (PCA/APCS) receptor model, Environ. Pollut. 129 (2004) 489–498. [27] H. Guo, A.J. Ding, K.L. So, G. Ayoko, Y.S. Li, W.T. Receptor modeling of source apportionment of Hong Kong aerosols and the implication of urban and regional contribution, Atmos. Environ. 43 (2009) 1159–1169.

317

[28] B. Wu, D. Zhao, Y. Zhang, X. Zhang, S. Cheng, Multivariate statistical study of organic pollutants in Nanjing reach of Yangtze River, J. Hazard. Mater. 169 (2009) 1093–1098. [29] B. Helena, R. Pardo, M. Vega, E. Barrado, J.M. Fernandez, L. Fernandez, Temporal evolution of groundwater composition in an alluvial aquifer (Pisuerga River, Spain) by principal component analysis, Water Res. 34 (2000) 807–816. [30] S. Li, Q. Zhang, Response of dissolved trace metals to land use/land cover and their source apportionment using a receptor model in a subtropic river, China, J. Hazard. Mater. 190 (2011) 205–213. [31] L. Wang, C. Ma, A study on the environmental geology of the Middle Route Project of the South–North water transfer, Eng. Geol. 51 (1999) 153–165. [32] J. Berkoff, China The South–North Water Transfer Project—is it justiﬁed? Water Policy 5 (2003) 1–28. [33] S. Feng, L. Li, Z. Duan, J. Zhang, Assessing the impacts of South-to-North Water Transfer Project with decision support systems, Decis. Support Syst. 42 (2007) 1989–2003. [34] S. Li, Z. Xu, H. Wang, J. Wang, Q. Zhang, Geochemistry of the Upper Han River basin, China. 3. Anthropogenic inputs and chemical weathering to the dissolved load, Chem. Geol. 264 (2009) 89–95. [35] S. Li, Q. Zhang, Geochemistry of the Upper Han River basin, China. 1. Spatial distribution of major ion compositions and their controlling factors, Appl. Geochem. 23 (2008) 3535–3544. [36] S. Li, Z. Xu, X. Cheng, Q. Zhang, Dissolved trace elements and heavy metals in the Danjiangkou Reservoir, China, Environ. Geol. 55 (2008) 977–983. [37] CMEP (Chinese Ministry of Environmental Protection), Bulletin for China’s environment status in 2009 (2009). [38] WRPBC (Water Resource Protection Bureau of Changjiang), Environmental Impacts of the Middle Route of the South to North Water Transfer Project (1995) (in Chinese). [39] CSEPB (Chinese State Environment Protection Bureau), Water and Wastewater Monitoring Analysis Methods, fourth ed., Chinese Environment Science Press, Beijing, China, 2002. [40] CSEPB (Chinese State Environment Protection Bureau), Environmental quality standards for surface water (GB3838-2002) (2002). [41] H. Bu, X. Tan, S. Li, Q. Zhang, Water quality assessment of the Jinshui River (China) using multivariate statistical techniques, Environ. Earth Sci. 60 (2010) 1631–1639. [42] B. Müller, M. Berg, Z. Yao, X. Zhang, D. Wang, A. Pﬂuger, How polluted is the Yangtze River? Water quality downstream from the Three Gorges Dam, Sci. Total Environ. 402 (2008) 232–247. [43] L. Wang, Y. Wang, C. Xu, Z. An, S. Wang, Analysis and evaluation of the source of heavy metals in water of the River Changjiang, Environ. Monit. Assess. 173 (2011) 301–313. [44] M. Klavinˇs, A. Briede, V. Rodinov, I. Kokorite, E. Parele, I. Klavina, Heavy metals in rivers of Latvia, Sci. Total Environ. 262 (2000) 175–183. [45] U.S. Environmental Protection Agency, Edition of the drinking water standards and health advisories (2006). [46] X. Zhang, J. Geffen, P. Zhang, J. Wang, Trend spatial & temporal distribution, and sources of the tropospheric SO2 over China based on satellite measurement during 2004–2009, in: Proceedings of the Dragon 2 Programme Mid-Term Results 2008–2010, Guilin City, People’s Republic of China, 17–21 May 2010 (ESA SP-684, October 2010), 2010. [47] D. Ji, Y. Wang, Y. Sun, Z. Ma, Characteristics of atmospheric SO2 in Beijing, Climatic and Environmental Research 14 (2009) 69–76 (in Chinese with English abstract).

Journal of Hazardous Materials 195 (2011) 318–323

Contents lists available at SciVerse ScienceDirect

Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat

Nitrogen removal via short-cut simultaneous nitriﬁcation and denitriﬁcation in an intermittently aerated moving bed membrane bioreactor Shuai Yang a,∗ , Fenglin Yang b a

East China Sea Environmental Monitoring Center, Shanghai 200137, PR China Key Laboratory of Industrial Ecology and Environmental Engineering, MOE, School of Environmental and Biological Science and Technology, Dalian University of Technology, Dalian 116024, PR China b

a r t i c l e

i n f o

Article history: Received 24 April 2011 Received in revised form 12 August 2011 Accepted 15 August 2011 Available online 22 August 2011 Keywords: Moving bed membrane bioreactor Nitrogen removal Short-cut simultaneous nitriﬁcation and denitriﬁcation Ammonia-oxidizing bacteria Nitrite-oxidizing bacteria

a b s t r a c t An intermittently aerated moving bed membrane bioreactor (MBMBR) was developed and crucial parameters affecting nitrogen removal from wastewater by simultaneous nitriﬁcation and denitriﬁcation via nitrite were investigated, without strict control of solids retention time. Changes in the microbiological community and distribution in the reactor were monitored simultaneously. The intermittent-aeration strategy proved effective in achieving nitrition and the chemical oxygen demand (COD) to total nitrogen (TN) ratio was an important factor affecting TN removal. In the MBMBR, the nitrite accumulation rate reached 79.4% and TN removal efﬁciency averaged at 87.8% with aeration 2 min/mix 4 min and an inﬂuent COD/TN ratio of 5. Batch tests indicated that under the intermittently aerated mode, nitrite-oxidizing bacteria (NOB) were not completely washed out from the reactor but NOB activity was inhibited. The intermittently aerated mode had no effect on the activities of ammonia-oxidizing bacteria. Fluorescence in situ hybridizations (FISH) results also suggested that NOBs remained within the system. © 2011 Elsevier B.V. All rights reserved.

1. Introduction Simultaneous nitriﬁcation and denitriﬁcation (SND) means that nitriﬁcation and denitriﬁcation occur concurrently in the same reaction vessel under identical operating conditions. SND has become an attractive technology for nitrogen removal, due to its potential to eliminate the need for separate tanks, required in conventional treatment plants, inducing a simpliﬁed and smaller design. The traditional biological nitrogen removal processes involve the oxidation of ammonium (NH4 + –N) to nitrate (NO3 − –N) (nitriﬁcation) and then reduction with an organic carbon source (chemical oxygen demand, COD) to nitrogen gas (N2 ) (denitriﬁcation). Both nitriﬁcation and denitriﬁcation involve nitrite (NO2 − –N) as an intermediate. Hence, if SND is accompanied by the inhibition of the second step of nitriﬁcation (oxidation of nitrite to nitrate), theoretically many advantages over conventional SND could be achieved, including: (1) a 25% reduction in aeration and 40% reduction of COD demand during denitriﬁcation, (2) 63% higher rate of denitriﬁcation, (3) 300% lower biomass yield during anaero-

∗ Corresponding author. E-mail address: [email protected] (S. Yang). 0304-3894/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.jhazmat.2011.08.045

bic growth, and (4) no apparent nitrite toxicity effects for the microorganisms in the reactor [1]. This is of particular interest when biologically removing nitrogen from wastewater with a low COD/TN ratio. However, the difﬁculty in removing nitrogen via nitrite lies in achieving speciﬁc inhibition or removal of the nitrite oxidizingbacteria (NOB; those that oxidize nitrite to nitrate) while retaining ammonia oxidizing-bacteria (AOB; those that oxidize ammonia to nitrite), thereby attaining nitrition. At present, most studies achieve nitrition by controlling a number of operational parameters, such as the free ammonia (FA) concentration, the free hydroxylamine (FH) concentration, the pH, the temperature, and the dissolved oxygen (DO) concentration, which have effects on the transient build-up of the nitrite ion [2,3]. Many studies have claimed to achieve nitrition but some crucial problems have not been resolved. For example, (1) the SHARON process was the most mature process for achieving nitrition, but because of its strict operational conditions (30–40 ◦ C, solids retention time (SRT) 1–3 d), it can only be used for a few special wastewater treatments (e.g., sludge-digestion liquid) and is not suitable for municipal wastewater and most industrial wastewater [4]. (2) Low DO concentration would not only affect the rate of nitriﬁcation but could also result in sludge bulking [5]. (3) If NOBs cannot be washed out rapidly from the system, they may adapt to the high FA level because of aberrance and, thus, SRT selection remains a problem.

S. Yang, F. Yang / Journal of Hazardous Materials 195 (2011) 318–323

DO concentration is a key factor affecting nitrition. Under low DO concentration, AOBs have been suggested to out-compete NOBs, based on the higher oxygen afﬁnity of AOBs compared with NOBs [6]. Cecen and Gonenc [7] reported that nitrite accumulation reached a considerable degree at DO to FA concentration ratios lower than 5 during nitriﬁcation, and the formation of nitrate was inhibited. No nitrite occurrence was encountered when this ratio exceeded 5, which implies that oxygen limitation leads to nitrite accumulation. Also, some processes allow for simultaneous nitrogen oxidation and reduction, likely to occur via nitrite at low DO concentrations, such as the OLAND process and the CANON process [8,9]. Many factors could affect the selection of DO, such as the ammonium–nitrogen concentration, COD/TN ratio, oxygen mass transfer resistance, amongst others. Too low DO may affect the rate of nitriﬁcation, while overly high DO may affect the accumulation of nitrite and result in energy waste. In recent years, some reports have indicated alternating aerobic and anoxic conditions resulting from intermittent aeration may induce nitrition [10,11]. Yoo et al. [12] achieved nitrogen removal utilizing SND via nitrite in a proposed intermittently aerated cyclic activated – sludge single-reactor process, and suggested some dominative parameters for effective operation. Nowak et al. [13] reported that under anoxic conditions, the decay rate of AOB was zero, while the decay rate of NOB was invariable, almost equaling that under aerobic conditions. In theory, intermittent aeration in the bioreactor has a high probability of resulting in a low DO condition, which would beneﬁt the multiplication of AOB and the accumulation of nitrite. The inhibition of the second step of nitriﬁcation (oxidation of nitrite to nitrate) was achieved because of the lag-time in nitration [1]. Also, the anoxic condition caused by mixing time would be beneﬁcial to denitriﬁcation via nitrite. The aim of this study was to achieve nitrogen removal by SND via nitrite in an intermittently aerated moving bed membrane bioreactor (MBMBR), without strict control of SRT. Some dominative parameters for effective operation were selected and the nitriﬁcation characteristics investigated. We adopt a two-step nitriﬁcation model to evaluate the behavior of both AOB and NOB in the intermittently aerated conditions. Temporal variations in the microbiological community and distribution in the reactor were simultaneously monitored. 2. Materials and methods 2.1. Experimental set-up and operating conditions Fig. 1 shows a schematic of the experimental apparatus. The reactor was made of plexiglas, with a working volume of

319

30 L. Temperature was thermostatically controlled at 25 ◦ C. A balance-box with a ﬂoat-ball valve was used to control the water level. Polypropylene hollow-ﬁber membranes (Hangzhou Kaihong, China) with a pore size of 0.1 ␮m and a ﬁltration area of 0.4 m2 were used. A piece of clapboard with holes was ﬁtted into the MBMBR to divide the reactor into two sections, with a volume ratio of 4:1. Carriers (30% v/v) were placed into the larger section and the membrane module was ﬁxed in the smaller one. A nonwoven carrier was used in the MBMBR. The density of the carriers was 0.27 g/cm3 and the effective speciﬁc surface area was 900 m2 /m3 . The clapboard was added to avoid the suspended carriers accumulating around the membrane module. The intermittently aerated mode was actualized through the modulation of the air pump and the agitator intermittently controlled by a timer-controlled power supply system. During the aerobic phase, the speciﬁc aeration demand per membrane area (SADm) was 0.75 m3 /m2 h. The MBMBR was maintained in continuous operation for about 6 months. The variations in operational parameters are summarized in Table 1. The SRT was maintained at 15 days by periodically removing sludge mixed liquor. The MBMBR was inoculated with activated sludge taken from the secondary settling tank of a municipal wastewater treatment plant (Chun-liu, China). Synthetic wastewater fed to the reactor consisted of sodium acetate, NH4 Cl, KH2 PO4 and mineral solution containing MgSO4 7H2 O (25 mg/L), CaCl2 ·2H2 O (22 mg/L), FeSO4 ·2H2 O (20 mg/L) and NaCl (25 mg/L). The initial inﬂuent contained 400 mg COD/L, 30 mg NH4 + –N/L and 4 mg PO4 3+ –P/L. The pH in the reactor was maintained at 7.6–8.5. 2.2. Analytical methods COD, ammonia nitrogen (NH4 + –N), nitrate nitrogen (NO3 − –N), nitrite nitrogen (NO2 − –N), mixed liquor suspended solid (MLSS), mixed liquor volatile suspended solid (MLVSS), and sludge volume index (SVI) were analyzed according to standard methods for the analysis of water and wastewater [14]. A certain amount of carriers were taken out from the bioreactor and placed into a beaker with deionized water of 500 mL. The carriers were then stirred with a magnetic stirrer for 60 min to wash out the biomass ﬁxed within the carriers. The suspension was dried and weighed to calculate the concentration of the bioﬁlm in the MBMBR. DO and pH in the reactor were measured by a DO meter (YSI 55/12 FT, USA) and a pH meter (Sartorius PB-10, Germany), respectively (Aqualytic). TN was determined based on the sum of NH4 + –N, NO2 − –N and NO3 − –N, rather than an independent TN test. 2.3. Fluorescence in situ hybridization The composition and spatial structure of the microbial community in the reactor (including the bioﬁlm and the suspended biomass) were analyzed by ﬂuorescence in situ hybridizations (FISH). FISH were performed according to the method described by Hibiya et al. [15]. The microbial samples were dispersed into individual cells by ultrasonication, and placed in a hybridization well on a gelatin-coated microscopic slide. NSO190 targeted halophilic and halotolerant ␤-proteobacterial AOB [16]. Ntspa662 and Nit3 are speciﬁcally used to target Nitrospira and Nitrobacter [17,18]. After hybridization, the microbial samples on the slides were examined using an epiﬂuorescence microscope (OlympusBX51, Japan) together with the standard software package supplied with the instrument (version 4.0). 2.4. Batch tests

Fig. 1. Schematic of the experimental apparatus. (1) Wastewater reservoir, (2) balance-box, (3) MBMBR, (4) membrane module, (5) air pump, (6) rotameter, (7) vacuum gauge, (8) peristaltic pump, (9) agitator and (10) timer.

A series of batch tests were conducted to assess the nitriﬁcation characteristics under the intermittently aerated mode. Here,

– 2 4 4

suspended biomass and bioﬁlm were used for batch tests together, with the ratio of the two fractions adopted according to the state at that time in the bioreactor. The activated sludge sample was centrifuged at 3000 rpm for 10 min, washed three times and then diluted to about 3000 mg/L with deionized water. Sodium acetate, NH4 Cl and KH2 PO4 were added to the mixed liquor to give the desired concentrations of COD, NH4 + –N and PO4 3+ –P. During batch tests, the solution pH was maintained at 7.5–8.5 using NaHCO3 . Liquid samples were intermittently removed to analyze COD, NH4 + –N, NO2 − –N and NO3 − –N. Test a: The initial COD and NH4 + –N concentrations were 125 mg/L and 50 mg/L, respectively. An appropriate amount of carbon source was added every 60 min to maintain the COD at about 100 mg/L to explore the characteristics of nitriﬁcation with the COD/TN ratio ranging from 2.5 to 5. Test b1: The initial NH4 + –N concentration was 40 mg/L, without carbon source addition. The intermittently aerated mode was adopted during the test in order to compare with test b2. Test b2: The initial NH4 + –N concentration was 40 mg/L without carbon source addition. The continuously aerated mode was adopted during the test.

Continuous 2 2 2 Standard deviation is given between parentheses; CODin , inﬂuent COD concentration; TNin , inﬂuent TN concentration.

2.5. Two-step nitriﬁcation model

a

Aerobic

0.86 0.90 0.90 0.92 3080 3174 3236 3428 42.2 (10.2) 42.4 (9.4) 42.8 (7.5) 57.5 (7.9) 235.8 (27.2) 229.7 (36.3) 215.2 (34.5) 214.2 (26.7) I II III IV

1–19 20–40 41–127 128–175

5.6 5.4 5.0 3.7

16 16 16 16

886 935 1062 1125

MLVSS/MLSS Suspended MLSS (mg/L) Bioﬁlm (mg/L) HRT (h) COD/TN TNin (mg/L) CODin (mg/L) Operational days (day) Phase

Table 1 Operational parameters in the experiment.a

Anaerobic

S. Yang, F. Yang / Journal of Hazardous Materials 195 (2011) 318–323

Cycle time (min)

320

A two-step nitriﬁcation model describing the batch tests (test b1 and test b2), with and without aerobic mode control, was adopted to evaluate the nitrition experiments [19]. Nitriﬁcation was split into two sub-processes: ammonium oxidation and nitrite oxidation, which is a different approach from other models (e.g., ASM [20]), where nitriﬁcation is considered as a single-step process. We use the model equations as follows: Monod-based growth kinetics for AOBs: AOB = MAX AOB

SNH4 So Ks,NH4 + SNH4 Ko,AOB + So

(a)

Monod-based growth kinetics for NOBs: NOB = MAX NOB

SNO2 So Ks,NO2 + SNO2 Ko,NOB + So

(b)

where AOB , NOB represent growth rate of AOB and NOB, respectively (d−1 ); SNH4 represents ammonium–nitrogen concentration (mg N/); SNO2 represents nitrite–nitrogen concentration (mg N/L); Ks,NH4 and Ks,NO2 represent substrate half saturation constants with respect to ammonium and nitrite, respectively (mg N/L); Ko,AOB and Ko,NOB represent oxygen half saturation constants of AOB and NOB, respectively (mg O2 /L); SO represents oxygen concentration (mg O2 /L). The key assumptions in the model include: (1) Denitriﬁcation was not considered in the model because no carbon source was added during the test. (2) Ammoniﬁcation was not included in the model; only inorganic nitrogen was considered. (3) The assimilation of ammonium for cellular growth was not included in the model because the growth of heterotrophic bacteria was not considered in the model and the inﬂuence of cellular assimilation of ammonium would be the same for both nitrifying bacterial groups, hence not favoring either group. (4) Decay or lysis of bacteria was not considered in the model because the decay rates of AOB and NOB were assumed to be identical, thus not favoring either group predicted by this model. The default model parameters are shown in Table 2. According to Blackburne et al. [21], in the model, the higher mass transfer affected the Ks value used for AOBs compared with the Ks value

S. Yang, F. Yang / Journal of Hazardous Materials 195 (2011) 318–323

321

Table 2 Default model parameters. Model parameter

Value

Reference

Ks,AOB Ks,NOB Ko,AOB Ko,NOB MAX AOB MAX NOB

0.75 mg N/L 0.15 mg N/L 0.03 mg O2 /L 0.4 mg O2 /L 0.65 days−1 0.65 days−1

[22] [23] [19] [19] [21] [21]

for NOBs, which was chosen as a value corresponding to the situation where mass transfer effects are largely ignored. Ammonium must diffuse from bulk liquid into ﬂocs before being used by AOBs, whereas, nitrite is produced within ﬂocs and therefore has a much smaller diffusion distance before reaching the NOBs. 3. Results and discussion 3.1. Performance of the MBMBR 3.1.1. Organic substance removal Fig. 2 presents the variations in COD concentration and its removal efﬁciencies in the intermittently aerated MBMBR for the entire experimental period. The bioreactor performed well on organic carbon removal. The efﬂuent COD concentration averaged 14.9 mg/L and COD removal efﬁciencies averaged 93.2%. The results indicate that changes in aerobic duration and COD/TN ratio in the inﬂuent exhibited virtually no inﬂuence on COD removal. 3.1.2. Nitrogen removal The long term operational period for the two systems comprised two steps. The ﬁrst step was the start-up period, with the MBMBR aerated continuously to evaluate the system performance in complete aerobic condition (phase I). During the second step, the intermittently aerated mode was adopted to investigate the effect of aerobic duration/anoxic duration on NH4 + –N and TN removal efﬁciencies (phases II–IV). In order to elucidate the extent of shortcut nitriﬁcation, the nitrite accumulation rate (NAR) was deﬁned as follows: NAR =

efﬂuent NO2 − –N efﬂuent NO2 − –N + efﬂuent NO3 − –N

× 100%

(c)

Fig. 3 illustrates the variations in NH4 + –N and TN concentrations as well as their removal efﬁciencies in the MBMBR throughout the experiment. It can be seen that in phase I, the NH4 + –N removal efﬁciency averaged 97.0% and the average efﬂuent NH4 + –N concentration was 1.25 mg/L. The inﬂuent NH4 + –N was almost completely removed, whereas the average TN removal efﬁciency was only

Fig. 2. COD concentrations and removal efﬁciencies in the MBMBR.

Fig. 3. NH4 + –N, NO2 − –N, NO3 − –N, TN concentrations and removal efﬁciencies in the MBMBR.

67.6% and the efﬂuent TN concentration averaged 13.67 mg/L. The NAR was only 4.5%, which indicates that the nitriﬁcation was full-range nitriﬁcation and the main product was nitrate. In phase II, the intermittently aerated mode (aeration 2 min/mix 2 min) was adopted, the efﬂuent NH4 + –N concentration showed a slight increase and averaged 2.53 mg/L, while the average NH4 + –N removal efﬁciency decreased to 94.0%. The average TN removal efﬁciency increased to 69.5% and the NAR increased to 49.1%. Although the TN removal was not signiﬁcantly improved, the nitrite accumulation was achieved gradually under the intermittently aerated mode, which indicates that the intermittently aerated mode was an effective approach to controlling the nitriﬁcation to stop at nitrition. With the adjustment of the intermittent time, for phase III (aeration 2 min/mix 4 min), the TN removal efﬁciency averaged 87.8% and the average efﬂuent TN concentration was 5.43 mg/L, which indicates that lengthening the mixing time is effective in improving TN removal. The NAR increased to 79.4% and the average efﬂuent NO3 − –N concentration was only 0.35 mg/L. At the same time, the change in anoxic duration did not have an obvious inﬂuence on NH4 + –N removal, with the NH4 + –N removal efﬁciency averaging 91.8% and the average efﬂuent NH4 + –N concentration being 3.49 mg/L. In order to explore the performance of the short-cut nitriﬁcation at a lower COD/TN ratio, the inﬂuent NH4 + –N concentration was increased to 57 mg/L during phase IV. The NH4 + –N removal efﬁciency decreased to 80.3% and the average efﬂuent NH4 + –N concentration was 11.32 mg/L. In this phase, aeration 2 min/mix 4 min could not supply sufﬁcient DO concentration for NH4 + –N removal. Simultaneously, the TN removal efﬁciency deceased to 65.5%, with an average efﬂuent TN concentration of 19.8 mg/L, which indicates that the carbon source was insufﬁcient for denitriﬁcation at a low COD/TN ratio of 3.8. The NAR increased to 93.3% due to the high nitrogen load. It can be concluded that the intermittent aeration time is an important factor for achieving short-cut nitriﬁcation and that the COD/TN ratio is another key factor for achieving TN removal. In the MBMBR, aeration 2 min/mix

322

S. Yang, F. Yang / Journal of Hazardous Materials 195 (2011) 318–323

Fig. 4. NH4 + –N, NO2 − –N, NO3 − –N and COD proﬁles during test a.

4 min is a suitable duration for simultaneous COD and TN removal at an inﬂuent COD/TN ratio of 5. 3.2. Speciﬁc nitriﬁcation rate test In order to explore the characteristics of nitriﬁcation and TN removal under the intermittently aerated mode, batch test a was carried out on day 137. It can be seen from Fig. 4 that the proﬁle of NH4 + –N declined progressively with time during the whole test and the average nitriﬁcation rate was 4.55 mg/L h. At the end of the test, the nitrite was 6.53 mg/L but nitrate was only 0.71 mg/L, which indicates that intermittent aeration is an effective approach for achieving nitrition. It is also noted that from 0 to 60 min, as the COD/TN ratio changed from 2.54 to 1.24, nitrite rapidly increased to 2.95 mg/L. At the 61st min, the COD/TN increased to 2.91 as carbon source was added. From 61 to 120 min, the COD/TN changed from 2.91 to 1.28 and the nitrite concentration increased to 4.28 mg/L. In the second 60 min reaction, the accumulated nitrite concentration was 1.33 mg/L, which was less than that during the initial 60 min reaction (2.95 mg/L). A similar situation could also be observed in the subsequent reaction. In other words, with a gradual decrease in the NH4 + –N concentration, the average COD/TN ratio increased and most of the nitriﬁcation product (nitrite) could be simultaneously removed. This observation suggests that under intermittent aeration, the COD/TN ratio remains an important factor affecting TN removal.

Fig. 5. NH4 + –N, NO2 − –N, NO3 − –N, COD, AOB and NOB proﬁles for test b1 (a) and test b2 (b).

3.3. Characteristics of nitriﬁcation with intermittent aeration To explore the cause for nitrite accumulation under intermittent aeration, two parallel batch experiments were carried out on day 141. Test b1 was operated under the intermittently aerated mode and test b2 was operated under the continuously aerated mode. It can be seen from Fig. 5 that for test b1, NO3 − –N was very low throughout the entire test and NO2 − –N accumulated gradually. At the end of the test, the NAR had increased to 92.2%. On the other hand, for test b2, the NO3 − –N remained below 2 mg/L during the ﬁrst 120 min, then NO3 − –N increased gradually during the subsequent 240 min. At the end of the test the NAR was reduced to 59.6%. This observation indicates that NOBs had not been washed out from the reactor but remained in the system. However, the activities of the NOBs were inhibited under the intermittently aerated mode.

Fig. 6. FISH micrographs of microbial samples, with a CY3-labeled NSO190 (red) probe, a FITC-labeled Ntspa662 (green) probe and a FITC-labeled Nit3 (green) probe. (A) Suspended biomass sample taken on day 12 and (B) suspended biomass sample taken on day 137. (For interpretation of the references to color in this ﬁgure legend, the reader is referred to the web version of the article.)

S. Yang, F. Yang / Journal of Hazardous Materials 195 (2011) 318–323

Hence, at the beginning of test b2, the NO3 − –N remained very low but with continuous aeration, the activities of the NOBs recovered gradually and the NO3 − –N concentration increased. The values of AOB and NOB during the batch tests are also presented in Fig. 5. It can be seen that in test b1, NOB slightly increased during the initial 120 min but it remained below 0.50. The NO3 − –N concentration was very low during the test b1. For test b2, during the initial 30 min, the value of NOB was below 0.50 and NO3 − –N did not accumulate. After about 60 min, the value of NOB increased to 0.53 and the NO3 − –N concentration began to accumulate simultaneously. Then the value of NOB increased smoothly and reached 0.55 by the end of the experiment. The value of NOB in test b2 was clearly higher than that in test b1, implying that the activities of the NOBs recovered under continuous aeration. In contrast, the values of AOB were almost constant during the whole experiment period for both tests, suggesting that the change in aeration could not affect the activities of the AOBs. Kornaros et al. [24] also reported that the AOBs did not exhibit any impact following the anoxic disturbance, while the NOBs were seriously inhibited. 3.4. The microbiological community and distribution in MBMBR Suspended biomass samples were taken from the reactor on days 12 (Fig. 6A) and 137 (Fig. 6B). To assess the composition of the bioﬁlm cultured on the non-woven materials in steady state, FISH was performed with the 16S rRNA targeting oligonucleotide probes NSO190, Ntspa665 and Nit3. NSO190 targeted halophilic and halotolerant ␤-proteobacterial AOB, Ntspa665 and Nit3 targeted Nitrospira and Nitrobacter, respectively. At the beginning of the experimental period, without control of aeration, the suspended biomass consisted of AOB reacting with NSO190, and NOB reacting with Ntspa665 and Nit3. AOB and NOB accounted for 54 ± 5% and 48 ± 5% of the total biomass, respectively. After long term control of intermittent aeration, the NOB percentage was signiﬁcantly reduced, accounting for about 26 ± 5% of the total biomass (Fig. 6B). The results indicate that under short SRT conditions, some NOBs could be washed out from the system but they could not be completely eliminated, as a small amount could be observed. Combined with the results described in Section 3.3, one may conclude that under the intermittently aerated mode, shortcut nitriﬁcation was achieved by inhibiting the activities of the NOBs, not by their removal. 4. Conclusion An intermittently aerated MBMBR was investigated to achieve SND via nitrite. Results demonstrated that intermittent aeration was an effective approach to achieve nitrition and the COD/TN ratio is another key factor affecting TN removal. Batch tests indicated that under the intermittently aerated mode, NOBs were not completely washed out from the reactor but remained in the system. However, the activities of NOBs were inhibited and their activities could recover under subsequent continuous aeration. The changes in aeration had no effect on the activities of AOBs. FISH results proved that NOBs could also be observed in the intermittently aerated bioreactor.

323

References [1] O. Turk, D.S. Mavinic, Preliminary assessment of a shortcut in nitrogen removal from wastewater, Can. J. Civil Eng. 13 (1986) 600–605. [2] R. Van Kempen, J.W. Mulder, C.A. Uijterlinde, Overview: full scale experience of the SHARON process for treatment of rejection water of digested sludge dewatering, Water Sci. Technol. 44 (1) (2001) 145–152. [3] Y. Peng, G. Zhu, Biological nitrogen removal with nitriﬁcation and denitriﬁcation via nitrite pathway, Appl. Microbiol. Biotechnol. 73 (1) (2006) 15–26. [4] B. Szatkowska, C. Cema, E. Plaza, One-stage system with partial nitritation and anammox processes in moving-bed bioﬁlm reactor, Water Sci. Technol. 54 (7) (2007) 49–58. [5] A.O. Sliekers, S.C.M. Haaijer, M.H. Stafsnes, Competition and coexistence of aerobic ammonium and nitrite oxidising bacteria at low oxygen concentrations, Appl. Microbiol. Biotechnol. 68 (2005) 808–817. [6] K.A. Third, J. Paxman, M. Schmid, Treatment of nitrogen-rich wastewater using partial nitriﬁcation and anammox in the CANON process, Water Sci. Technol. 52 (2005) 47–54. [7] F.I. Cecen, E. Gonenc, Nitrogen removal characteristics of nitriﬁcation and denitriﬁcation ﬁlters, Water Sci. Technol. 29 (10–11) (1994) 409–416. [8] K. Windey, I. de Bo, W. Verstraete, Oxygen-limited autotrophic nitriﬁcation–denitriﬁcation (OLAND) in a rotating biological contactor treating high-salinity wastewater, J. Biotechnol. 39 (2005) 4512–4520. [9] K.A. Third, A.O. Sliekers, J.G. Kuenen, The CANON system (completely autotrophic nitrogen-removal over nitrite) under ammonium limitation: interaction and competition between three groups of bacteria, Syst. Appl. Microbiol. 24 (2001) 588–596. [10] S.Y. Ip, J.S. Bridger, N.F. Mills, Effect of alternating aerobic and anaerobic conditions on the economics of the activated sludge system, Water Sci. Technol. 19 (1987) 911–918. [11] L.A. Lishman, R.L. Legge, G.J. Farquhar, Temperature effects on wastewater treatment under aerobic and anoxic conditions, Water Res. 34 (8) (2000) 2263–2276. [12] H. Yoo, K. Ahn, H. Lee, et al., Nitrogen removal from synthetic wastewater by simultaneous nitriﬁcation and denitriﬁcation via nitrite in an intermittentlyaerated reactor, Water Res. 33 (1) (1999) 145–154. [13] O. Nowak, K. Svardal, P. Schweighofer, The dynamic behavior of nitrifying activated sludge systems inﬂuenced by inhibiting wastewater compounds, Water Sci. Technol. 31 (2) (1995) 115–124. [14] APHA, Standard Methods for the Examination of Water and Wastewater, American Public Health Association, 1995. [15] K. Hibiya, A. Terada, S. Tsuned, Simultaneous nitriﬁcation and denitriﬁcation by controlling vertical and horizontal microenvironment in a membrane-aerated bioﬁlm reactor, J. Biotechnol. 100 (1) (2003) 23–32. [16] B.K. Mobarry, M. Wagner, V. Urbain, Phylogenetic probes for analyzing abundance and spatial organization of nitrifying bacteria, Appl. Environ. Microbiol. 62 (6) (1996) 2156–2162. [17] M. Wagner, G. Rath, R. Amann, In situ identiﬁcation of ammonia-oxidizing bacteria, Syst. Appl. Microbiol. 18 (1995) 251–264. [18] H. Daims, J.L. Nielsen, P.H. Nielsen, In situ characterization of Nitrospira-like nitrite-oxidizing bacteria active in wastewater treatment plants, Appl. Environ. Microbiol. 67 (2001) 5273–5284. [19] R. Blackburne, Z. Yuan, J. Keller, Partial nitriﬁcation to nitrite using low dissolved oxygen concentration as the main selection factor, Biodegradation 19 (2008) 303–312. [20] M. Henze, W. Gujer, T. Mino, Activated Sludge Models ASM1, ASM2, ASM2d and ASM3, IWA Publishing, 2000. [21] R. Blackburne, Z. Yuan, J. Keller, Demonstration of nitrogen removal via nitrite in a sequencing batch reactor treating domestic wastewater, Water Res. 42 (2008) 2166–2176. [22] J.E. Alleman, Elevated nitrite occurrence in biological wastewater treatment systems, Water Sci. Technol. 17 (1984) 409–419. [23] R. Manser, Population dynamics and kinetics of nitrifying bacteria in membrane and conventional activated sludge, Ph.D. thesis, Swiss Federal Institute for Environmental Science and Technology, Swiss Federal Institute of Technology, 2005. [24] M. Kornaros, S.N. Dokianakis, G. Lyberatos, Partial nitriﬁcation/denitriﬁcation can be attributed to the slow response of nitrite oxidizing bacteria to periodic anoxic disturbances, Environ. Sci. Technol. 44 (19) (2010) 7245–7253.

Journal of Hazardous Materials 195 (2011) 324–331

Contents lists available at SciVerse ScienceDirect

Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat

Effects of UV irradiation on humic acid removal by ozonation, Fenton and Fe0 /air treatment: THMFP and biotoxicity evaluation Ming-Chi Wei a , Kai-sung Wang b , Tung-En Hsiao b , I.-Chen Lin b , Hui-Ju Wu b , Yuh-Luan Wu c , Pey-Horng Liu c , Shih-Hsien Chang b,d,∗ a

Department of Food Science, Central Taiwan University of Science and Technology, Taichung, Taiwan, ROC Department of Public Health, Chung-Shan Medical University, Taichung 402, Taiwan, ROC c Green Energy and Environment Research Laboratories, Industrial Technology Research Institute, Taiwan, ROC d Department of Family and Community Medicine, Chung Shan Medical University Hospital, Taichung 402, Taiwan, ROC b

a r t i c l e

i n f o

Article history: Received 26 April 2011 Received in revised form 7 August 2011 Accepted 15 August 2011 Available online 22 August 2011 Keywords: Fenton Ozone Fe0 /air Vibrio ﬁscheri

a b s t r a c t Effects of UV irradiation on humic acid (HA) removal by Fe0 /air, ozonation and Fenton oxidation were investigated. The trihalomethane forming potential (THMFP) and toxicity of treated solutions were also evaluated. The experimental conditions were ozone of 21 mg min−1 , H2 O2 of 8 × 10−4 M, Fe0 of 20 g L−1 , air ﬂow of 5 L min−1 , and UVC of 9 W. Results indicated that Fe0 /air rapidly removed HA color (>99%) and COD (90%) within 9 min. 51–81% of color and 43–50% of COD were removed by ozonation and Fenton oxidation after 60 min. Both UV enhanced ozone and Fenton oxidation removed HA, but the Fe0 /air process did not. Spectrum results showed all processes effectively diminished UV–vis spectra, except for ozonation. The THMFP of Fe0 /air-treated solution (114 ␮g L−1 ) was much lower than those of Fenton- (226 ␮g L−1 ) and ozonation-treated solutions (499 ␮g L−1 ). Fe0 /air with UV irradiation obviously increased the THMFP of treated solution (502 ␮g L−1 ). The toxicity results obtained from Vibrio ﬁscheri light inhibition test indicated that the toxicity of Fe0 /air-treated solution (5%) was much lower than that of ozonation- (33%) and Fenton-treated solutions (31%). Chlorination increased the solution toxicity. The correlation between biotoxicity and chloroform in the chlorinated solution was insigniﬁcant. © 2011 Elsevier B.V. All rights reserved.

1. Introduction Humic substrates are mixtures of organic matters that often occur in surface waters. Humic acid (HA) is a major component of humic substrates. It comes from different sources including nature, landﬁll leachate, and pulp wastewater [1,2]. HA can react with active chlorine in water treatment plants, resulting in the formation of trihalomethanes (THMs). THMs are carcinogenic and toxic to human life and, more speciﬁc, the kidneys [3]. Besides, HA causes membrane fouling [4] and scavenges the free radicals in advanced oxidation processes [5]. Many approaches have been applied to remove HA, including GAC adsorption [1], membrane ﬁltration [1], coagulation [6], ozonation [5], and advanced oxidation processes [5,7]. However, either their low removal efﬁciency or high costs often limit their application. To increase the degradation ability, UV irradiation is often employed to enhance HA removal by chemical oxidation and AOPs [5]. Besides, chlorine reacts with the intermediates in the treated solution and generates toxic byproducts during disinfec-

∗ Corresponding author. Tel.: +886 4 24730022x11799; fax: +886 4 22862587. E-mail address: [email protected] (S.-H. Chang). 0304-3894/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.jhazmat.2011.08.044

tion processes. Factors that inﬂuence trihalomethane formation potentials (THMFPs) during disinfection processes include humic acid content [8,9] and properties like aromaticity, unsaturated bonds [10,11], molecular weight [8,12], and hydrophobicity [9,13]. Because each technique has its own speciﬁc removal mechanisms, the organic contents, intermediate properties, toxicity, and THMFP of the treated HA solutions might be different. Zero-valent iron (Fe0 ) has received wide attention recently because it is low-cost and effective for various pollutant removals [14–17]. Iron erodes in Fe0 /H2 O system and removes pollutants through coagulation of iron corrosion products [18–20]. In addition, zero-valent iron under oxic condition can generate Fenton-like reaction and degrade organic pollutants [21–23]. Fe0 /air methods have been applied to remove various pollutants including EDTA [14], chlorinated organic compounds [15], and dyes [16,17]. Recently, it has been reported that Fe0 irradiated with UV can produce Fenton-like reaction which enhances organic removal [24]. However, studies on HA removal by Fe0 /air and UV/Fe0 /air process are limited. Studies on the biotoxicity and THMFP of Fe0 /air- and UV/Fe0 /air-treated HA solutions are also rare. In this study, humic acid (HA) was selected as the model humic substrate. The aims of this study were to investigate (1) HA removal by ozonation, Fenton oxidation, and Fe0 /air processes, (2) effects

M.-C. Wei et al. / Journal of Hazardous Materials 195 (2011) 324–331

325

2.2.3. Fe0 /air and UV/Fe0 /air treatment The Fe0 /air experiments were conducted in a 275-mL glass reactor (5 cm diameter, 13 cm high) containing 100 mL of HA solution. An air ﬂow rate of 5 L min−1 was used to maintain the suspension of iron powder in the solution. The initial pH of the solution was adjusted with H2 SO4 and NaOH solutions. For the UV/Fe0 /air experiment, a UVC light tube was inserted in the solution (7 cm depth). Samples were withdrawn at speciﬁc time intervals during the Fe0 /air and UV/Fe0 /air treatment. Fig. 1. Molecular structure of humic acid.

2.3. THMFP

2. Materials and methods

The THMFP test was conducted according to Standard Method 5710B [26]. The samples were adjusted to 7 by phosphate buffer. The concentrated sodium hypochlorite was dosed and the ﬁnal NaOCl concentration was 50 mg L−1 . The sample was kept at 25 ◦ C and placed in a dark place. After 7 days, THMs of the samples were measured using purge and trap and GC/MS (Agilent-6890 GC/5973N MS).

2.1. Chemicals

2.4. Vibrio ﬁscheri light inhibition test

Humic acid (60% purity) was obtained from Sigma–Aldrich and was used as received (Fig. 1). H2 O2 (30% purity) was purchased from Hayashi Pure Chemical Ind. Ltd., Japan) and FeSO4 (FeSO4 ·7H2 O, >99% purity) was obtained from Showa (Japan). The zero-valent iron (analytical grade, 99% purity, 300 mesh) was purchased from Shimakyu Chemical Osaka, Japan). NaOCl (5% purity) was obtained from Hayashi Pure Chemical Ind. (Japan).

The marine luminescent bacterium, V. ﬁscheri (NRRL B-11117, obtained from DSMZ Germany), was employed to evaluate the biotoxicity of treated solutions. The cultivation of luminescent bacteria and toxicity evaluation procedure were according to ISO 11348-1 standard protocol (ISO, 1998). The solution sample was adjusted to pH 7.0 ± 0.2. V. ﬁscheri was exposed to the solution samples for 5 min as determined by a luminometer at 15 ◦ C. Phenol was used as the positive control with EC50 ranging from 13 to 26 mg L−1 . Toxicity was expressed as the light inhibition ratio and was calculated as follows (Eq. (1)) [27]:

of UV irradiation on HA removal by above three processes, and (3) the trihalomethane forming potential and biotoxicity of treated solutions.

2.2. Humic acid treatment 2.2.1. Ozonation and UV/ozone treatment The ozonation and UV/ozone experiments were conducted in a 275 mL glass reactor (5 cm diameter, 13 cm high) containing 100 mL of HA solution. The initial HA solution in this study was 50 mg L−1 . Ozone was provided by ozone generator (CHYF-3A, Company, Ltd., Taiwan). The ozone ﬂow rate and ozone dose were 3 L min−1 and 21 mg min−1 , respectively. To obtain the desired pH, the solution was adjusted using diluted H2 SO4 and NaOH solutions and measured by pH meter (Cyberscan 510, Taiwan). For the UV/ozonation experiments, the UVC light (9 W, Philips) was placed in the center of the rector (7 cm depth). All experiments were conducted at room temperature (20 ± 2 ◦ C). Samples were withdrawn at speciﬁc time intervals during the ozonation. The concentration of the treated HA solution was measured based on the constructed calibration curves at absorption wavelength of 400 nm. The UV–vis spectrum during HA degradation was measured at 200–800 nm using a UV–vis spectrophotometer (Shimadzu, UV-mini 1240, Japan). The sample was diluted with distilled water when the absorbance exceeded the range of calibration curve. COD was determined according to standard method for examination of water and wastewater [25]. 2.2.2. Fenton and UV/Fenton Appropriate amounts of stock HA solution and ferrous ion were added to a 300 mL beaker and diluted with distilled–deionized water to 100 mL. The initial solution pH was adjusted using diluted H2 SO4 and NaOH solutions. H2 O2 was then added to initiate the Fenton reaction. The H2 O2 :Fe2+ molar concentration ratio was kept at 10:1. The applied H2 O2 doses were 0, 2, 4, and 8 × 10−4 M. The initial solution pH was 3 and the reaction time was 60 min. The air ﬂow rate was 5 L min−1 . For the UV/Fenton experiment, a UVC light tube (9 W, Philips) was inserted in the solution (7 cm depth). The sample was withdrawn at 60 min, centrifuged at 13,000 rpm for 5 min and then analyzed.

Light inhibition (%) =

I0 × fkt − If I0 × fkt

× 100

(1)

where fk is the correction factor at t = 5 min, fk = Ikc /I0c . I0c and Ikc are the luminescence intensity of the control sample at t = 0 and 5 min, respectively, and I0 and If the luminescence intensity of the sample at t = 0 and 5 min, respectively. The bioluminescence intensity of V. ﬁscheri may decrease with exposure time. The correcting factor fk is used to correct the luminescence intensity of the test sample at t = 0 min. Therefore, in this study, the inhibition ratio (percentage) was used to represent the biotoxicity of pollutants instead of inhibition percentage of the bioluminescence. A toxicity experiment was conducted to investigate the effectiveness of Na2 S2 O3 of 20 g L−1 in eliminating the residual H2 O2 and chlorine in Fentontreated and chlorinated solution, respectively [28]. The selected doses of H2 O2 (8 × 10−4 M) and NaClO (6.7 × 10−4 M) were based on doses used for Fenton oxidation and chlorination treatments in this study. Experimental results showed that without Na2 S2 O3 addition, the light inhibition ratio was 53% and 32% for 8 × 10−4 M of H2 O2 and 6.7 × 10−4 M of Cl2 , respectively. After addition of Na2 S2 O3 of 20 g L−1 , the light inhibition ratios of both solutions were 0%. This suggests that the addition of Na2 S2 O3 of 20 g L−1 could effectively reduce the residual H2 O2 and chlorine in both Fenton-treated and chlorinated solution. 3. Results and discussion 3.1. Ozonation and UV/ozone treatment HA decolorization by ozonation and the UV/ozone process were investigated. The operating conditions were humic acid of 50 mg L−1 , ﬂow rate of 3 L min−1 , and ozone dose of 7 mg L−1 . First, the initial solution pH (pH0 ) was evaluated. Fig. 2 indicates that at pH0 of 3, 77% color removal was obtained after 60 min. An increase

326

M.-C. Wei et al. / Journal of Hazardous Materials 195 (2011) 324–331 100

100 O3, pH0 3 O3, pH0 7 O3, pH0 11

80

UV/O3, pH0 11

color removal (%)

color removal (%)

80

UV, pH0 11

60

40

60

40

20 20

0 0

10

20

30

40

50

60

0

Time(min)

UV

Fig. 2. Effects of UVC on humic acid decolorization by ozonation. Humic acid of 50 mg L−1 , O3 dose of 7 mg L−1 , ozone ﬂow rate 3 L min−1 , and UVC of 9 W.

in initial solution pH slightly increased decolorization. At solution pH0 of 11, 81% decolorization was achieved after 60 min. Inﬂuences of UV irradiation on HA decolorization by ozonation was also assessed. Fig. 2 indicates that at pH0 of 11, UV alone decolorized less HA than ozonation. However, UV rapidly enhanced HA removal by ozonation; 47% of HA was rapidly decolorized by UV/ozone at 3 min. After that, the HA decolorization became slow and 88% of color removal was achieved after 60 min. In this study, HA color removal increased with an increase in solution pH0 . This is because at acidic or neutral conditions, most of the ozone is in molecular phase (E0 = 2.08 V) [29], which decolorizes HA by breaking the unsaturated carbon–carbon bonds [9]. At alkaline solution, ozone reacts with OH− to generate strong oxidants like hydroxyl radical (OH• , E0 = 2.80 V) (Eq. (2)) [29,30]. Free radicals can destroy aromatic sites and unsaturated components of HA and increase HA decolorization [5,11,31]. O3 + OH− → OH• + (O2 • ↔ HO2 • )

(2)

In this study, UV/ozone had higher HA decolorization than ozonealone. This is because UV can enhance production of free radicals through UV photolysis of ozone (Eqs. (3) and (4)) and increase HA decolorization [30,32]. O3 + H2 O + h → O2 + H2 O2

hv < 310 nm

O2 + H2 O2 → 2OH• + 3O2

(3) (4)

3.2. Fenton and UV/Fenton treatment HA decolorization by Fenton oxidation was evaluated. The experimental conditions were H2 O2 :Fe2+ molar ratio of 10, pH0 of 3, and reaction time of 60 min. Fig. 3 shows that 40% color removal was obtained after 60 min at H2 O2 of 2 × 10−4 M. The HA decolorization increased with the increase of H2 O2 concentrations. 51% color removal was achieved at 8 × 10−4 M at 60 min. Fig. 3 also illustrates that UV noticeably enhanced HA decolorization by Fenton oxidation. For example, 93% of HA decolorization was obtained by photo-Fenton at 60 min. It has been reported that H2 O2 (E0 = 1.776 V) can react with catalyst Fe2+ to generate free radicals like hydroxyl radical (OH• ) (Eqs. (5)–(7)) [7,33], which can effectively destroy HA chromophore groups [11,31]. Wu et al. [7] also indicates that anions (Fe2+ and Fe3+ ) in the Fenton solution also act as coagulants to remove HA. Fe2+ + H2 O2 → Fe3+ + OH• + OH−

(5)

+

(6)

3+

Fe

+ H2 O2 → Fe

2+

+ OOH•

+H

V V +U +U -4 M -4 M -4 M x1 0 x1 0 x1 0 -4 M -4 M F2 F4 F8 x1 0 x1 0 8 4 F F

Fig. 3. Effects of UVC on decolorization of humic acid by Fenton oxidation. Humic acid of 50 mg L−1 , H2 O2 :Fe2+ molar ratio of 10:1, and pH0 of 3, ﬂow rate of 5 L min−1 , UVC of 9 W. F: Fenton.

Fe3+ + • OOH → Fe2+ + H+ + O2

(7)

In this study, HA color removal by photo-Fenton was much higher than by Fenton oxidation. This is because H2 O2 reacts with ferrous iron oxide to form complex compound Fe(H2 O)5 (OH)2 + (Eq. (8)), which can be irradiated by UV, producing hydroxyl radicals, which decolorize HA (Eq. (9)) [7]. Fe2+ + H2 O2 → Fe(OH)2+ + OH• 2+

Fe(OH)

+ h → Fe

2+

+ OH•

(8) (9)

3.3. Fe0 /air and UV/Fe0 /air treatment HA decolorization by Fe0 /air and UV/Fe0 /air processes was investigated. First, the effects of pH0 on HA color removal were evaluated at Fe0 of 10 g L−1 . Fig. 4a shows that when pH0 decreased from 11 to 3, HA decolorization increased from 19% to 63% at 9 min. Acidic solution is suitable for HA decolorization by the Fe0 /air process. Fig. 4b illustrates the effects of Fe0 doses on HA decolorization at pH0 3. HA color removal increased with Fe0 doses. Greater than 99% of HA color was removed after 9 min at Fe0 of 20 g L−1 . A further increase of the Fe0 dose to 30 g L−1 did not obviously enhance HA decolorization. Effects of UV irradiation on HA decolorization by Fe0 /air process were also evaluated. The operating conditions were Fe0 of 20 g L−1 and pH0 of 3. Fig. 4c shows that UV enhanced HA decolorization by Fe0 /air process and >99% HA decolorization was achieved at 7 min. In this study, Fe0 /air rapidly decolorized HA. This is possibly because iron corrodes in Fe0 /H2 O system and generates iron corrosion products, like Fe(OH)2 , Fe(OH)3 , FeOOH, Fe2 O3 , and Fe3 O4 or green rusts (Eqs. (10)–(13)) [18]. Fe0 → Fe2+ + 2e−

(10)

−

(11)

2+

Fe

→ Fe

3+

+e

−

+ 2OH → Fe(OH)2

(12)

Fe3+ + 3OH− → Fe(OH)3

(13)

2+

Fe

During Fe0 /air treatment, iron corrodes and a layer of iron oxides/hydroxide forms on the Fe0 surface [34]. The layer of iron oxides/hydroxide can rapidly sorb the humic acid through electrostatic attraction. Humic acid has been reported to generate complexes with iron and iron oxide through various types of carboxylate, phenolic, and carbonyl functional groups in the humic acid [35,15]. Thus, the humic acid decolorization may also be caused

M.-C. Wei et al. / Journal of Hazardous Materials 195 (2011) 324–331

(b) 100

100 pH 3 pH 7 pH 11

color removal (%)

80

color removal (%)

(a)

60

40

20

10 g/L 20 g/L 30 g/L

80

60

40

20

0

0 0

1

3

5

7

9

0

1

3

Time (min) 100

color removal (%)

60

40

20

0

60

40

0

1

3

9

80

0 mM 1 mM 3 mM 10 mM

20

0

7

2D Graph 2

(d) 100

UV Fe 10 g/L Fe 20 g/L UV/Fe 10 g/L UV/Fe 20 g/L

80

5

Time (min)

color removal (%)

(c)

327

5

7

9

0

2

4 6 Time (min)

Time (min)

8

Fig. 4. Decolorization of humic acid by Fe0 /air method, (a) effects of pH0 , Fe0 of 10 g L−1 , (b) effects of Fe0 dose, (c) effects of UVC irradiation at different Fe0 doses, (d) effects of t-butanol, Fe0 dose of 20 g L−1 . If not mentioned otherwise, the experimental conditions were humic acid of 50 mg L−1 , pH0 of 3, UVC of 9 W, air ﬂow rate of 5 L min−1 .

by the adsorption of HA onto the layer of iron oxides/hydroxide formed on the Fe0 surface. Especially, since the dose of Fe0 is high (20 g L−1 ) in this study, the fast elimination of humic acid may result from the adsorption of humic acid on the Fe0 surface, which can also lead to signiﬁcant COD removal and a low concentration of THMFPs. Besides, production of strong oxidants by Fe0 under oxic condition was also responsible for the enhancement of HA decolorization. Several studies have indicated that iron corrosion in the presence of dissolved oxygen can produce H2 O2 [22] (Eq. (14)). The H2 O2 can react with the Fe2+ to produce strong oxidants such as hydroxyl radical (OH• ) (Eq. (5)) and ferryl (iron (IV) iron species) [14] which can oxidize HA. Fe0 + O2 + 2H+ → H2 O2 + Fe2+

irradiates iron to produce iron corrosion coagulants (Eq. (15)), which remove 1,4 dioxane. (2) UVC photolyzes H2 O to generate H2 O2 , which reacts with Fe2+ to produce Fenton-like reaction (Eqs. (16)–(18)) [24]. Fe0 + h → Fe2+ + 2e− 2H2 O + h → HO2 • + 3H+

(15) (< 698 nm)

Fe2+ + HO2 • (O2 •− ) → Fe3+ + H2 •

HO2 (O2

•− )

+

+ H → H2 O2 + H2

(16) (17) (18)

3.4. Comparisons of COD removal and spectrum change

(14)

Fe2+ + H2 O2 → Fe3+ + OH• + OH− An experiment with addition of t-butanol (0–10 mM), an OH• scavenger, into the reaction solution was conducted to investigate the HA oxidation by hydroxyl radical (OH• ) during Fe0 /air treatment. Fig. 4d shows that an increase of t-butanol doses from 0 to 10 mM did not obviously inhibit the HA decolorization by Fe0 /air treatment. This suggests that the HA oxidation by hydroxyl radicals only played a minor role in humic acid removal by Fe0 /air treatment. In this study, the HA decolorization by Fe0 was faster under acidic solution than under alkaline solution. This is because the thickness of the iron oxide/hydroxide layer might obviously affect the HA decolorization rate. Mielczarski et al. [34] indicate that the iron oxide/hydroxide layer on Fe0 surface is thin and most iron oxides exist in the solution at pH of 3. However, when solution pH increases, most of the iron oxides accumulate on the Fe0 surfaces and reduce HA decolorization rates. In this study, HA decolorization by UV/Fe0 /air process was faster than that by Fe0 /air process. This might be because UV accelerates iron corrosion and enhances Fe0 to produce Fenton-like reaction. Son et al. [24] applied the UVC/Fe0 process to remove 1,4 dioxane. They indicated that UVC/Fe0 removed 1,4 dioxane through two mechanisms: (1) UVC

The COD removals of humic acid solution by different processes were compared. Experimental conditions were ozonation: pH0 11, ozone of 21 mg min−1 , t = 60 min; Fenton oxidation: H2 O2 of 8 × 10−4 M and pH0 3, t = 60 min; Fe0 /air treatment: Fe0 dose of 20 g L−1 , air ﬂow 5 L min−1 , t = 9 min. The UVC light was 9 W. Fig. 5 indicates that ozone-alone and Fenton treatments removed 43% and 50% COD, respectively, after 60 min. Fe0 /air treatment rapidly removed 91% COD at 9 min. Fig. 5 also shows that UV obviously enhanced COD removal by the ozonation and Fenton processes. COD removal by UV/ozone and photo-Fenton were 83% and 87%, respectively. However, UV did not obviously increase COD removal by Fe0 /air treatment. The UV–vis spectrum changes of HA solution during different treatments were also investigated. For HA solution, characteristic absorption peaks at wavelength 400 nm and 254 nm are attributed to chromophore groups and benzene components, respectively [10,11]. Fig. 6 a shows that the A400 and A254 of the original HA were 0.539 and 1.983, respectively. Although ozonation could decolorize HA, absorption spectrum still remained after 60-min treatment (A400 = 0.146 and A254 of 1.096). Fig. 6b illustrates the spectra of UV, ozone, and UV/ozonetreated HA solutions after 60-min treatment. UV alone only slightly reduced the UV–vis absorption bands. UV/ozone treatment

328

M.-C. Wei et al. / Journal of Hazardous Materials 195 (2011) 324–331 100

COD removal (%)

80

60

40

20

0 O3

Fenton

Fe0/air

UV/O3 UV/FentonUV/Fe0/air

Fig. 5. COD removal of humic acid solution by ozonation, Fenton, and Fe0 /air treatments. If not mentioned otherwise, conditions are humic acid of 50 mg L−1 , ﬂow rate of 5 L min−1 , UVC of 9 W, t = 60 min. Ozonation: pH0 of 11, ﬂow rate of 3 L min−1 , ozone dose of 21 mg min−1 ; Fenton oxidation: pH0 of 3, H2 O2 of 8 × 10−4 M, H2 O2 /Fe2+ of 10:1; Fe0 /air treatment: Fe0 of 20 g L−1 , pH0 of 3, t = 9 min.

effectively destroyed chromophore groups (A400 ) and benzene components (A254 ). Only a small amount of UV absorption spectrum (A254 = 0.406) remained after treatment. The spectrum changes of HA solution by Fenton and UV/Fenton were evaluated. Fig. 6c indicates that UV–vis spectrum still remained after Fenton oxidation (A400 = 0.229 and A254 = 1.018). In contrast, UV/Fenton effectively diminished HA UV–vis absorption bands at 235–600 nm. Fig. 6d illustrates the HA spectrum changes during Fe0 /air treatment. Fig. 6d shows that the absorption bands were reduced rapidly and uniformly during Fe0 /air treatment. Both A400 and A254 were smaller than 0.01 after 9 min. The changes of HA spectrum by UV/Fe0 /air treatment were similar to those treated by the Fe0 /air process. In this study, high benzene components (A254 ) and organic matter (COD) remained in the treated solution (Fig. 6a). This is because ozone only partially oxidized unsaturated carbon bonds and broke the benzene ring, resulting in the generation of HA intermediates like formic acid and oxalic acid [5,6,9]. Ozone could not effectively degrade the intermediates. In contrast, UV/ozone effectively decreased the spectrum area and COD. This might be because UV/ozone can produce free radicals (Eqs. (2), (3) and (8)) and effectively destroy HA (chromophore and benzene components) and oxidize intermediates [11,13]. Similarly, UV irradiation can enhance Fenton reaction to produce free radicals (Eqs. (7)–(8)) [11,13]. Therefore, photo-Fenton decreased the absorption spectrum and COD more effectively than Fenton oxidation (Figs. 5 and 6c). In this study, the HA spectrum bands diminished uniformly during Fe0 /air and UV/Fe0 /air treatments (Fig. 6e). This implies that HA removal by iron corrosion coagulants might be the major mechanism responsible for HA removal by the Fe0 /air and UV/Fe0 /air processes [16,18–20,36]. The small UV absorbance in the Fe0 /airtreated solution also suggests that the concentrations of residual benzene intermediates were low in the treated solution. 3.5. Trihalomethane formation potentials of treated HA solution Table 1 indicates the trihalomethane formation potentials (THMFPs) of different treated HA solutions. Because NaOCl was used for chlorination, chloroform (CHCl3 ) was the major THM detected. Results indicate that without UV irradiation, the THMFP was the highest in the ozonation-treated HA solution (499 ␮g L−1 ), followed by Fenton-treated solution (226 ␮g L−1 ). The THMFP of Fe0 /air-treated HA solution was the lowest (114 ␮g L−1 ). When UV

irradiation was used during treatment, the THMFP of the UV/ozonetreated solution (650 ␮g L−1 ) was higher than that of ozone-treated solution. The THMFP of UV/Fenton-treated solution (211 ␮g L−1 ) was close to that of Fenton-treated solution. However, THMFP of UV/Fe0 /air-treated solution (502 ␮g L−1 ) was much higher than that of Fe0 /air-treated solution. In this study, even though the decrease of COD and A254 by ozonation and that by Fenton oxidation was close (Table 1), the THMFP of ozone-treated solution (499 ␮g L−1 ) was much higher than that of Fenton-treated solution (226 ␮g L−1 ). This suggests that besides organic content (COD) and benzene components (A254 ), there should be other factors inﬂuencing the THMFPs of treated solution, such as different properties of intermediates in these two treated solutions (Eqs. (4)–(6)). Several studies have reported that after ozonation, HA forms more active sites, which react easily with free chlorine and generate THMs [37]. In contrast, Fenton reaction degrades HA through free radical oxidation [11,13]. In this study, the THMFP of the Fe0 /air-treated solution was low (114 ␮g L−1 ). This is possibly because the major mechanism responsible for HA removal by Fe0 /air process was chemical coagulation (which in fact removed most HA) instead of oxidation. The above results indicate that although UV/ozone removed much more COD and A254 than ozonation, the THMFP of solution treated by UV/ozone was higher than that treated by ozonation. This phenomenon also suggests that besides organic matter and A254 , there should be other factors responsible for the THMFP of treated solution. This is possibly because of the lower molecular weight of intermediates produced by UV/ozone than that produced by ozonation [11,13]. Zhao et al. [8] indicated that organic matter with a low molecular weight is a major precursor for THM production during disinfection. Similar results were also observed in both Fenton and UV/Fenton-treated solutions. Liu et al. [38] used UVA/TiO2 and UVA/TiO2 /H2 O2 to degrade HA and found that, although A254 was very low in both treated solutions, the treated solutions were still very reactive to chlorine and that high THMs were formed during chlorination. Table 1 also shows that although COD and A254 were low in both Fe0 /airand UV/Fe0 /air-treated solutions, the THMFP in UV/Fe0 /air-treated solution was much higher than that in the Fe0 /air-treated solution. This might be because UV photolysis and Fenton-like reaction occur in the UV/Fe0 /air process (Eqs. (16)–(18)) [24], which change the properties of intermediates or increase low molecular weight intermediates in the treated solution. However, the molecular weight distribution of intermediates was not analyzed in this study. Further studies on molecular weight distribution are suggested in order to investigate their relations with THMFPs. Besides, Results of this study indicated that Fe0 /air rapidly removed HA color (>99%) and COD (90%) within 9 min. This implies that most of the organics in the HA solution have been removed. Total organic carbon can help to understand the importance of intermediates generated during treatment. Further studies on TOC removal by different processes are suggested to explain the results obtained by COD and THMFP. 3.6. Biotoxicity evaluation The toxicity of treated solutions before and after chlorination was investigated by V. ﬁscheri light inhibition test. Na2 S2 O3 was used to reduce the residual H2 O2 and chlorine. Fig. 7a indicates that the biotoxicity of Fe0 /air-treated solution (light inhibition of 6.7%) was much lower than those of ozonation-treated (light inhibition of 32%) and Fenton treated (light inhibition of 44%) solutions. After addition of Na2 S2 O3 , the biotoxicity of Fenton-treated solution decreased to light inhibition 31%. The UV irradiation enhanced ozonation, Fenton, and Fe0 /air to the point of reducing the biotoxicity of their treated solutions.

M.-C. Wei et al. / Journal of Hazardous Materials 195 (2011) 324–331 4

(b) 0 min 3 min 6 min 9 min 12 min 15 min 30 min 60 min

absorbance

3

2

1

0 min UVC O UVC/O

2

1

0

200

4

3

absorbance

(a)

0

300

400

500

600

700

800

200

300

400

wavelength (nm.) 4

(d) 0 min UVC Fenton UVC/Fenton

absorbance

3

800

0 min 1 min 3 min 5 min 7 min 9 min

2

0

0

300

400

500

600

700

800

200

300

400

500

600

700

800

wavelength (nm.)

wavelength (nm.) 4

0 min UVC Fe /air UVC/Fe /air

3

absorbance

700

1

1

(e)

600

4

3

2

200

500

wavelength (nm.)

absorbance

(c)

329

2

1

0

200

300

400

500

600

700

800

wavelength (nm.) Fig. 6. UV–vis spectrum change of humic acid solution by ozonation, Fenton, and Fe0 /air treatments. If not mentioned otherwise, conditions are humic acid of 50 mg L−1 , ﬂow rate of 5 L min−1 , UVC of 9 W. (a) ozonation, pH0 of 11, ﬂow rate of 3 L min−1 , O3 dose of 21 mg min−1 ; (b) UVC, O3 and UV/O3 , t = 60 min; (c) Fenton and UVC/Fenton, pH0 of 3, H2 O2 of 8 × 10−4 M, H2 O2 /Fe2+ of 10:1; t = 60 min; (d) Fe0 /air method, Fe0 of 20 g L−1 , pH0 of 3, t = 9 min; (e) UVC, Fe0 /air and UV/Fe0 /air.

This might be because UV enhancement of ozonation, Fenton, and Fe0 /air generate free radicals, which would lead to more complete intermediate removal. Effects of chlorination on biotoxicity of the treated solutions were also assessed. Fig. 7b shows that the

biotoxicity of treated solutions (38–57%) after chlorination was higher than those without chlorination. After addition of Na2 S2 O3 to dechlorinate the residual chlorine, the toxicity of chlorinated solutions (24–43%) was still higher than those without chlorination.

Table 1 The THMFPs of HA solutions treated by different methods.

Without UV O3 Fenton Fe0 /air With UV UV/O3 UV/Fenton UV/Fe0 /air

Color removal (%)

COD removal (%)

A400

A254

CHCl3 (␮g L−1 )

CHBrCl2 (␮g L−1 )

CHBr2 Cl (␮g L−1 )

CHBr3 (␮g L−1 )

81 51 99

43 50 90

0.146 0.229 0.029

1.096 1.018 0.067

499 226 114

ND ND ND

ND ND ND

ND ND ND

88 93 99

83 87 91

0.052 0.008 0.01

0.406 0.088 0.039

650 211 502

ND ND ND

ND ND ND

ND ND ND

If not mentioned otherwise, conditions are humic acid of 50 mg L−1 , ﬂow rate of 5 L min−1 , UVC of 9 W, t = 60 min. Ozonation: pH0 of 11, ﬂow rate of 3 L min−1 , ozone dose of 21 mg min−1 ; Fenton oxidation: pH0 of 3, H2 O2 of 8 × 10−4 M, H2 O2 /Fe2+ of 10:1; Fe0 /air treatment: Fe0 of 20 g L−1 , pH0 of 3, t = 9 min.

330

M.-C. Wei et al. / Journal of Hazardous Materials 195 (2011) 324–331

(a)

100

(b)

80

light inhibition (%)

light inhibition (%)

80

100

60

40

60

40

20

20

0

0 O3

Fenton

Fe0/air

UV/O3

O3

UV/Fenton UV/Fe0/air

Fenton

Fe0/air

UV/O3

UV/Fenton UV/Fe0/air

Fig. 7. Biotoxicity of ozonation-, Fenton-, and Fe0 /air-treated humic acid solution. (a) Before chlorination and (b) after chlorination. Experimental conditions for different treatments, if not mentioned otherwise: humic acid of 50 mg L−1 , ﬂow rate of 5 L min−1 , UVC of 9 W, and t = 60 min. Ozonation: pH0 of 11, ﬂow rate of 3 L min−1 , O3 of 21 mg min−1 ; Fenton oxidation: pH0 of 3, H2 O2 of 8 × 10−4 M, and H2 O2 /Fe2+ of 10:1; Fe0 /air treatment: Fe0 of 20 g L−1 , pH0 of 3, and t = 9 min.

The relationship between biotoxicity and CHCl3 concentrations of the dechlorinated solutions was also evaluated. The correlations between biotoxicity and CHCl3 were insigniﬁcant (r = 0.128, p > 0.05). This implies that besides CHCl3 , there were other intermediates that contributed to the acute toxicity of the dechlorinated solution. Further study on intermediate identiﬁcation is suggested to investigate the relationship between intermediates and biotoxicity in dechlorinated solutions. As shown in Fig. 6d, the spectrum of Fe0 /air treatment shows that prolonged Fe0 /air treatment from 7 min to 9 min only slightly reduced the spectrum of HA solution. In this study, the V. ﬁscheri light inhibition test was used to evaluate the biotoxicity evolution of Fe0 /air-treated solution before and after chlorination. The biotoxicity of the Fe0 /air-treated solution after chlorination was higher than those without chlorination (Fig. 7b). This implies that the residual organic matter in the Fe0 /air-treated solution reacted with free chlorine and generated toxic byproducts like THMs. Since ZVI is a strong reducing agent that is capable of abiotically dehalogenating chlorinated solvents (e.g., trichloroethene), further study on dechlorination of trihalomethane byproducts by ZVI is suggested. 4. Conclusions The effects of UV irradiation on HA removal by different processes were investigated. In addition, the THMFP and biotoxicity of treated solutions were evaluated. The experimental conditions were ozone dose of 7 mg L−1 , UVC of 9 W, H2 O2 of 8 × 10−4 M, Fe0 of 20 g L−1 , and air ﬂow of 5 L min−1 . Fe0 /air rapidly and effectively removed HA color (>99%) and COD (91%) within 9 min. In contrast, 51–81% of decolorization and 43–50% of COD removal were obtained by ozonation and Fenton oxidation after 60 min. UV radiation enhanced HA removal by ozonation and Fenton oxidation. The effect of UV on HA removal by Fe0 /air process was slight. Spectrum results indicated that all methods effectively diminished UV–vis spectra, except for ozonation. The THMFP of solution treated by Fe0 /air was much lower than those treated by Fenton and ozonation. Fe0 /air process with UV obviously increased the THMFP of its treated solution. Chlorination increased the toxicity of treated solutions. The relationship between toxicity and chloroform in the chlorinated solution after Na2 S2 O3 addition was insigniﬁcant. Acknowledgement The authors gratefully acknowledge the National Science Council of the ROC (Taiwan) for ﬁnancial support under Project No. NSC 98-2622-E-040-003-CC2.

References [1] S. Ciputra, A. Antony, R. Phillips, D. Richardson, G. Leslie, Comparison of treatment options for removal of recalcitrant dissolved organic matter from paper mill efﬂuent, Chemosphere 81 (2010) 86–91. [2] Y. Wu, S. Zhou, F. Qin, H. Peng, Y. Lai, Y. Lin, Removal of humic substances from landﬁll leachate by Fenton oxidation and coagulation, Process Saf. Environ. Prot. 88 (2010) 276–284. [3] A. Imyim, E. Prapalimrungsi, Humic acids removal from water by aminopropyl functionalized rice husk ash, J. Hazard. Mater. 184 (2010) 75–781. [4] H. Zhu, X. Wen, X. Huang, Membrane organic fouling and the effect of preozonation in microﬁltration of secondary efﬂuent organic matter, J. Membr. Sci. 352 (2010) 213–221. [5] D. Imai, A.H.A. Dabwan, S. Kaneco, H. Katsumata, T. Suzuki, T. Kato, K. Ohta, Degradation of marine humic acids by ozone-initiated radical reactions, Chem. Eng. J. 148 (2009) 336–341. [6] C.S. Uyguner, S.A. Suphandag, A. Kerc, M. Bekbolet, Evaluation of adsorption and coagulation characteristics of humic acids preceded by alternative advanced oxidation techniques, Desalination 210 (2007) 183–193. [7] Y. Wu, S. Zhou, F. Qin, K. Zheng, X. Ye, Modeling the oxidation kinetics of Fenton’s process on the degradation of humic acid, J. Hazard. Mater. 179 (2010) 533–539. [8] Z. Zhao, J. Gu, X. Fan, H. Li, Molecular size distribution of dissolved organic matter in water of the Pearl River and trihalomethane formation characteristics with chlorine and chlorine dioxide treatments, J. Hazard. Mater. 134 (2006) 60–66. [9] A. Teksoy, U. Alkan, H.S. Bas¸kaya, Inﬂuence of the treatment process combinations on the formation of THM species in water, Sep. Purif. Technol. 61 (2008) 447–454. [10] H. Zhang, J. Qu, H. Liu, X. Zhao, Characterization of isolated fractions of dissolved organic matter from sewage treatment plant and the related disinfection byproducts formation potential, J. Hazard. Mater. 164 (2009) 1433–1438. [11] Y. Nie, C. Hu, L. Zhou, J. Qu, Q. Wei, D. Wang, Degradation characteristics of humic acid over iron oxides/Fe0 core–shell nanoparticles with UVA/H2 O2 , J. Hazard. Mater. 173 (2010) 474–479. [12] D. Gang, T.E. Clevenger, S.K. Banerji, Relationship of chlorine decay and THMs formation to NOM size, J. Hazard. Mater. A96 (2003) 1–12. [13] S. Park, T. Yoon, The effects of iron species and mineral particles on advanced oxidation processes for the removal of humic acids, Desalination 208 (2007) 181–191. [14] J.D. Englehardt, D.E. Meeroff, L. Echegoyen, Y. Deng, F. Raymo, T. Shibata, Oxidation of aqueous EDTA and associated organics and coprecipitation of inorganics by ambient iron-mediated aeration, Environ. Sci. Technol. 41 (2007) 270–276. [15] D.C.W. Tsang, N.J.D. Graham, I.M.C. Lo, Humic acid aggregation in zero-valent iron systems and its effects on trichloroethylene removal, Chemosphere 75 (2009) 1338–1343. [16] S.H. Chang, K.S. Wang, S.J. Chao, T.H. Peng, L.C. Huang, Degradation of azo and anthraquinone dyes by a low-cost Fe0 /air process, J. Hazard. Mater. 166 (2009) 1127–1133. [17] S.H. Chang, S.H. Chung, H.C. Li, H.H. Liang, L.C. Huang, Comparative study on the degradation of I.C. Remazol Brilliant Blue R and I.C. Acid Black 1 by Fenton oxidation and Fe0 /air process and toxicity evaluation, J. Hazard. Mater. 166 (2009) 1279–1288. [18] C. Noubactep, A critical review on the mechanism of contaminant removal in Fe0 –H2 O systems, Environ. Technol. 29 (2008) 909–920. [19] C. Noubactep, Characterizing the discoloration of methylene blue in Fe0 /H2 O systems, J. Hazard. Mater. 166 (2009) 79–87. [20] Q. Wang, N. Cissoko, M. Zhou, X. Xu, Effects and mechanism of humic acid on chromium (VI) removal by zero-valent iron Fe0 nanoparticles, Phys. Chem. Earth Parts A/B/C 36 (2011) 442–446.

M.-C. Wei et al. / Journal of Hazardous Materials 195 (2011) 324–331 [21] G. Roy, P. de Donato, T. Görner, O. Barres, Study of tropaeolin degradation by iron-proposition of a reaction mechanism, Water Res. 37 (2003) 4954–4964. [22] C.E. Noradoun, I.E. Cheng, EDTA degradation induced by oxygen activation in a zerovalent iron/air/water system, Environ. Sci. Technol. 39 (2006) 7158–7163. [23] K.S. Wang, C.L. Lin, M.C. Wei, H.H. Liang, H.C. Li, C.H. Chang, Y.T. Fang, S.H. Chang, Effects of dissolved oxygen on dye removal by zero-valent iron, J. Hazard. Mater. 182 (2010) 886–895. [24] H.S. Son, J.K. Im, K.D. Zoh, A Fenton-like degradation mechanism for 1,4-dioxane using zero-valent iron Fe0 and UV light, Water Res. 43 (2009) 1457–1463. [25] APHA, AWWA and WEF, Standard Methods for the Examination of Water and Wastewater, 20th ed., American Public Health Association, Washington, DC, 1998. [26] American Public Health Association, Standard Method for the Examination of Water and Wastewater, 5520 Chemical Oxygen Demand, 1995, pp. 5-12–5-16. [27] ISO, Water Quality-determination of the Inhibitory Effect of Water Samples on the Light Emission of Vibrio ﬁscheri (Luminescent Bacteria Test). ISO 11348-1, International Standard Organization, Geneva, Switzerland, 1998. [28] S.H. Chang, K.S. Wang, H.H. Liang, H.Y. Chen, H.C. Li, T.H. Peng, Y.C. Su, C.Y. Chang, Treatment of Reactive Black 5 by combined electrocoagulation–granular activated carbon adsorption–microwave regeneration process, J. Hazard. Mater. 175 (2010) 850–857. [29] T.E. Agustina, H.M. Ang, V.K. Vareek, A review of synergistic effect of photocatalysis and ozonation on wastewater treatment, J. Photochem. Photobiol. C 6 (2005) 264–273.

331

˜ [30] M.A. Alsheyab, A.H. Munoz, Reducing the formation of trihalomethanes THMs by ozone combined with hydrogen peroxide H2 O2 /O3 , Desalination 194 (2006) 121–126. [31] M. Fukushima, K. Tatsumi, S. Nagao, Degradation characteristics of humic acid during photo-Fenton processes, Environ. Sci. Technol. 35 (2001) 3683–3690. [32] M.S. Lucas, J.A. Peres, G.L. Puma, Treatment of winery wastewater by ozonebased advanced oxidation processes O3 , O3 /UV and O3 /UV/H2 O2 ) in a pilotscale bubble column reactor and process economics, Sep. Purif. Technol. 72 (2010) 235–241. [33] S. Yang, P. Wang, X. Yang, L. Shan, W. Zhang, X. Shao, R. Niu, Degradation efﬁciencies of azo dye Acid Orange 7 by the interaction of heat, UV and anions with common oxidants: persulfate, peroxymonosulfate and hydrogen peroxide, J. Hazard. Mater. 179 (2010) 552–558. [34] J.A. Mielczarski, G.M. Atenas, E. Mielczarski, Role of iron surface oxidation layers in decomposition of azo-dye water pollutants in weak acidic solutions, Appl. Catal. B: Environ. 56 (2005) 289–303. [35] L. Xie, C. Shang, Role of humic acid and quinone model compounds in bromate reduction by zerovalent iron, Environ. Sci. Technol. 39 (2005) 1092–1100. [36] P. Lai, H. Zhao, C. Wang, J. Ni, Advanced treatment of coking wastewater by coagulation and zero-valent iron processes, J. Hazard. Mater. 147 (2007) 232–239. [37] T.F. Marhaba, D. Van, The variation of mass and disinfection by-product formation potential of dissolved organic matter fractions along a conventional surface water treatment plant, J. Hazard. Mater. 74 (2000) 133–147. [38] S. Liu, M. Lim, R. Fabris, C. Chow, K. Chiang, M. Drikas, R. Amal, Removal of humic acid using TiO2 photocatalytic process—fractionation and molecular weight characterisation studies, Chemosphere 72 (2008) 263–271.

Journal of Hazardous Materials 195 (2011) 332–339

Contents lists available at SciVerse ScienceDirect

Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat

Chelating agent free-solid phase extraction (CAF-SPE) of Co(II), Cu(II) and Cd(II) by new nano hybrid material (ZrO2 /B2 O3 ) Özcan Yalc¸ınkaya a , Orhan Murat Kalfa b , Ali Rehber Türker a,∗ a b

Gazi University, Science Faculty, Department of Chemistry, 06500, Ankara, Turkey Dumlupınar University, Science and Art Faculty, Department of Chemistry, 43100, Kütahya, Turkey

a r t i c l e

i n f o

Article history: Received 17 May 2011 Received in revised form 15 August 2011 Accepted 15 August 2011 Available online 22 August 2011 Keywords: Nanosorbent Flame atomic absorption spectrometry Preconcentration Boron oxide Zirconia Heavy metal

a b s t r a c t New nano hybrid material (ZrO2 /B2 O3 ) was synthesized and applied as a sorbent for the separation and/or preconcentration of Co(II), Cu(II) and Cd(II) in water and tea leaves prior to their determination by ﬂame atomic absorption spectrometry. Synthesized nano material was characterized by scanning electron microscope, transmission electron microscope and X-ray diffraction. The optimum conditions for the quantitative recovery of the analytes, including pH, eluent type and volume, ﬂow rate of sample solution were examined. The effect of interfering ions was also investigated. Under the optimum conditions, adsorption isotherms and adsorption capacities have been examined. The recoveries of Co(II), Cu(II) and Cd(II) were 96 ± 3%, 95 ± 3%, 98 ± 4% at 95% conﬁdence level, respectively. The analytical detection limits for Co(II), Cu(II), and Cd(II) were 3.8, 3.3, and 3.1 ␮g L−1 , respectively. The reusability and adsorption capacities (32.2 mg g−1 for Co, 46.5 mg g−1 for Cu and 109.9 mg g−1 for Cd) of the sorbent were found as satisfactory. The accuracy of the method was conﬁrmed by analyzing certiﬁed reference material (GBW07605 Tea leaves) and spiked real samples. The method was applied for the determination of analytes in tap water and tea leaves. © 2011 Elsevier B.V. All rights reserved.

1. Introduction The accurate determination of trace elements in environmental samples is an important and challenging task in analytical chemistry. Direct determination of trace elements appears to be difﬁcult task as the concentration of them is close to or below the detection limits of most of the analytical techniques besides the real sample matrix may cause serious interference during their determination process. However, these problems can be solved by applying the preconcentration techniques which can simultaneously remove the sample matrix and increase the concentration of analytes [1,2]. Various preconcentration methods including liquid–liquid extraction [3,4], ion exchange [5,6], co-precipitation [7,8], ﬂotation [9], cloud point extraction [10,11] and solid phase extraction [12–17] have been used to remove the sample matrix and increase the concentration of analytes. Among the preconcentration techniques, solid phase extraction (SPE) technique has increasingly become a popular technique. It has several major advantages including operation simplicity, high preconcentration factor, minimum eluent volume, reduced disposal cost, shorter extraction time for sample preparation [18,19] and the availability of a wide

∗ Corresponding author. Tel.: +90 312 2021110; fax: +90 312 2122279. E-mail addresses: [email protected], [email protected] (A.R. Türker). 0304-3894/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.jhazmat.2011.08.048

variety of sorbent materials that mainly affect the extraction efﬁciency. In the past decade, nanometer solid materials have become more and more important due to their special properties [20]. It is known that diameter of nanoparticles is less than 100 nm, and with the decrease in diameter the number of atom on the surface and surface area increases rapidly. Consequently, nano-meter sized materials can adsorb many ions and reach equilibrium in a very short time. Moreover, the adsorption ability of nano materials compared with other solid phase extractor is rather high [21]. Recently nano materials have been used as a sorbent due to their improved intrinsic properties such as chemical activity and ﬁne grain size compared with classical sorbents (i.e., TiO2 , Al2 O3 ) [2,22,23]. Nano material can be prepared by various techniques such as chemical vapor deposition [24], and sol–gel method [25,26]. In recent years, many papers have been published for trace metal preconcentration by nano materials as sorbent which have been synthesized and used [27–34]. An actual tendency in the ﬁeld of solid phase extraction is related to synthesis of the new sorbent materials with good performance, such as high resistance to acids and bases, selective for analytes, large surface area and high adsorption capacity [28]. To increase the performance of a sorbent it is usually modiﬁed with chelating agents or microorganisms. As a modiﬁcation, chelating agents or microorganisms have been loaded or immobilized on a solid substrate [35,36]. However, the modiﬁcation or chelating agent addition may cause an increase in

Ö. Yalc¸ınkaya et al. / Journal of Hazardous Materials 195 (2011) 332–339

possible contamination and interferences due to the reagents used for modiﬁcation. The modiﬁcation also limits repeated use of sorbent due to the loss of chelating agents or microorganisms from the solid sorbent. To minimize possible contaminations and interfering effects and to increase repeated use of sorbent, chelating agent or microorganism free solid phase extractors have also been proposed [29,37–39]. In this study, hybrid nano ZrO2 /B2 O3 was synthesized as a new solid phase extractor and used without modiﬁcation for the preconcentration of Co(II), Cu(II) and Cd(II). To the best of our knowledge this is the new solid phase extractor used for this purpose. This hybrid material was ﬁrstly synthesized by us and used for this purpose. The synthesized material was ﬁrst characterized by using scanning electron microscope (SEM), transmission electron microscope (TEM) and X-ray diffraction (XRD) methods and then used as solid phase extractor. The procedure was validated by analyzing certiﬁed reference materials and applied for various real samples. In our previous paper [29], nano B2 O3 /TiO2 composite material had been used as a new solid phase extractor for the preconcentration and separation of cadmium. In this study, ZrO2 /B2 O3 was used for the preconcentration and separation of Co, Cu and also cadmium. These two materials have almost similar advantages for capturing heavy metals. The higher preconcentration factor was obtained for cadmium by using B2 O3 /TiO2 material. 2. Experimental 2.1. Apparatus A Technai G2 120 kV transmission electron microscope (Oregon, USA), a JEOL LV6060 model scanning electron microscope (Tokyo, Japan) with EDS apparatus and a BRUKER D8 discover X-ray diffractometer (Madison, USA) were used for the determination of morphology of the synthesized nano material. The experimental conditions of XRD measurement were as follows: CuK˛ radiation; tube voltage/current, 20 kV/30 mA; scanning range (2), 5–80◦ ; scanning rate, 5◦ /min. A Varian (Palo Alto, CA, USA) AA240FS model ﬂame atomic absorption spectrometer equipped with a deuterium-lamp background corrector, copper hallow cathode lamp (Varian), cadmium hallow cathode lamp (Varian), multi element (Co–Mo–Pb–Zn) hallow cathode lamp (Varian) and air acetylene ﬂame as the atomizer was used for the determination of Co, Cu and Cd under the conditions suggested by the manufacturer. The wavelength, lamp current, slit width and acetylene ﬂow rate were 240.7 nm, 7 mA, 0.2 nm and 2 L min−1 for Co, 324.8 nm, 4.0 mA, 0.5 nm and 2 L min−1 for Cu and 228.8 nm, 4 mA, 0.5 nm and 2 L min−1 for Cd, respectively. A Varian AA240Z graphite furnace atomic absorption spectrometer was used for cadmium and cobalt determination in Certiﬁed Reference Material (GBW-07605) using the instrumental parameters given in Table 1. All pH measurements were made with a WTW 720 model pH meter (Weilheim, Germany).

333

Table 1 Instrumental parameters of electrothermal atomic absorption spectrometry for Co and Cd. Parameters

Sample volume (␮L) Lamp current (mA) Argon ﬂow rate (mL min−1 ) Drying 1, ◦ C Ramp time (s) Hold time (s) Drying 1, ◦ C Ramp time (s) Hold time (s) Pyrolysis, ◦ C Ramp time (s) Hold time (s) Atomization, ◦ C Ramp time (s) Hold time (s) Cleaning, ◦ C Ramp time (s) Hold time (s)

Values Co

Cd

20 7 3 95 5 40 120 15 15 750 15 10 2300 1 3 2500 1 2

20 4 3 95 5 40 120 15 15 300 15 10 1800 1 3 2200 1 2

2.3. Preparation of hybrid nano zirconium dioxide–boron oxide sorbent Hybrid nano ZrO2 /B2 O3 was synthesized by modifying the procedure given in the literature [28], given for nano alumina synthesis. For this purpose, 10 g of H3 BO3 and 5 g of Zr(OCl)4 were weighed into a beaker containing about 50 mL of ethanol. Then, 1.5 mL of Triton X-114 was added in the mixture as surfactant and stirred for 1 h. The mixture was sonicated for about 30 min in ultrasonic bath. pH of the mixture was adjusted to 6 and then sonication was applied again for 15 min. The mixture was left for 12 h at room temperature and then dried in an oven at 70 ◦ C for about 1 h. The obtained solid material was then transferred into the mufﬂe furnace and heated at 850 ◦ C for about 2 h. Then, the material was ground in a Spex type ball mill to obtain powder product. 2.4. Column preparation A glass column (150 mm length and 8 mm i.d.) having a stopcock at the bottom and a tank of 250 mL on top of the column was used. A small amount of glass wool was placed over its stopcock in order to hold the sorbent. 200 mg dry hybrid nano material (ZrO2 /B2 O3 ) was made slurry in water and then placed into the column. Then, another small amount of glass wool was inserted onto the top of the sorbent to avoid disturbance of the adsorbent during sample passage. The column was preconditioned by passing blank solution having same pH with the sample solution prior to use. After each use, the nano material in the column was washed with dilute HCl (0.5 mol L−1 ) and water, respectively and stored in water until the next experiment. 2.5. Preconcentration and determination procedure

2.2. Reagents and solutions All reagents were of analytical grade, unless otherwise stated. All solutions were prepared in ultra pure water (18.3 ␮S cm−1 ). Multi element standards (in various concentrations) and model solutions were prepared by dilution of single element stock solutions (1000 ␮g mL−1 ) of Co(II), Cu(II) and Cd(II) ions purchased from Merck. Zr(OCl)4 (BDH), ethanol (99.5%, J.T. Baker), Al(NO3 )3 ·9H2 O (Merck), NaNO3 (Carlo Erba), KNO3 (Merck), Ca(NO3 )2 ·4H2 O (J.T. Baker), MgSO4 (Carlo Erba), Ni(NO3 )2 ·6H2 O (Carlo Erba), HNO3 (Merck, 65%), HCl (Merck, 37%), NH3 (Merck, 27%) were used.

The separation/preconcentration procedure based on the solid phase extraction via hybrid nano material (ZrO2 /B2 O3 ) was tested with model sample solutions prior to the determination of analytes in real samples. For this purpose, 50 mL of solution containing 10 ␮g Co(II), 10 ␮g Cu(II) and 10 ␮g Cd(II) were taken and the pH was adjusted to the optimum value determined experimentally with dilute HCl (0.1 mol L−1 ) or NH3 (0.1 mol L−1 ) solutions. The column was conditioned to the working pH by passing the aqueous solution having same pH value. Then the sample solution was passed through the column at a ﬂow rate adjusted to the optimum value (5 mL min−1 ) determined experimentally. The retained ana-

334

Ö. Yalc¸ınkaya et al. / Journal of Hazardous Materials 195 (2011) 332–339

lytes on the sorbent were then eluted with 10 mL of 1 mol L−1 HNO3 solution. This solution was aspirated into an air-acetylene ﬂame for the determination of analytes by FAAS. A blank solution was also run under the same conditions without adding the analytes. The nano ZrO2 /B2 O3 sorbent was used repeatedly after washing with 1 mol L−1 HNO3 solution and distilled water, respectively. The recovery of analytes was calculated from the ratio of the concentration found by FAAS to that calculated theoretically. 2.6. Collection and preparation of samples Tap water sample was collected from the laboratory of chemistry department. This water sample was ﬁltered from blue ribbon ﬁlter paper. Approximately 1.0 g of tea samples (Tea samples purchased from local market in the city of Ankara, Turkey and standard reference material, Tea leaves GBW-07605) were taken in a 250 mL beaker and 10 mL of concentrated nitric acid was added. The beaker was heated on a hot plate at about 130 ± 10 ◦ C for 3 h. After cooling to room temperature, 2 mL of hydrogen peroxide was added drop wise. The beaker was heated until complete decomposition of tea sample. The resulting solution was transferred into a 50 mL volumetric ﬂask by washing the interior surface of the beaker with small portions of ultra pure water, and the solution was diluted to the mark with ultra pure water. 3. Results and discussion Firstly characterization of synthesized material has been performed to demonstrate the nano character and to determine crystalline structure of synthesized material. Then, the applicability of the material as a solid phase extractor has been tested in detail. In order to obtain optimum separation and/or preconcentration conditions and maximum recoveries, some experimental parameters such as the pH of sample solution, type and concentration of eluent, volume of sample solution and ﬂow rate of sample solution have been optimized. Interfering effects, reusability of hybrid sorbent and adsorption isotherms have also been studied. The analytical parameters such as limit of detection (LOD), limit of quantitation (LOQ), precision, accuracy and linear working range have been determined at optimal experimental conditions. 3.1. Characterization of synthesized material

Fig. 1. SEM micrographs of synthesized hybrid nano ZrO2 /B2 O3 .

recovery of the analytes was determined by applying the general procedure (Section 2.5) by changing the pH of model solution in the range of 1–10. The variation in recovery of metal ions with pH is shown in Fig. 4. As can be seen, Co(II) and Cd(II) were quantitatively (above 95%) recovered at the pH range of 8–10 and Cu(II) was quantitatively recovered at the pH range 6–10. pH 8 is an appropriate pH for all of the analytes when they will be determined simultaneously. However, if they will be determined separately, as an optimum pH 8 should be selected for Co(II) and Cd(II) and 6 should be selected for Cu(II) due to advantages of acidic pH values. Hence, the following optimization work was carried out at these pHs. 3.3. Effect of eluent type and concentration The type, amount, and concentration of eluent are other important parameters for this kind of studies. In order to determine type and amount of elution solution, 50 mL of model solutions contain-

The synthesized material was characterized by scanning electron microscope (SEM), transmission electron microscope (TEM) and X-ray diffraction (XRD) method. Firstly, it was investigated that whether the material has nano size or not. For this purpose, SEM and TEM images of the material were obtained. As can be seen from the SEM and TEM images of hybrid nano ZrO2 /B2 O3 hybrid material (Figs. 1 and 2), the ZrO2 /B2 O3 particles are very ﬁne and the grain size is below 100 nm. From these results, it can be concluded that nano scale material, which is one of the purpose of this study, could be obtained by the method described above. Secondly, in order to characterize the nature of synthesized nano material, XRD pattern of the material was also investigated. According to the XRD pattern shown in Fig. 3, the nanoparticles are identiﬁed as crystalline B2 O3 and ZrO2 . It can be seen that the synthesized new nano material consisted mainly of B2 O3 and ZrO2 . 3.2. Effect of pH of sample solution The pH value plays an important role in adsorption of the ions onto sorbents. It strongly inﬂuences the sorption availability of the metal ions. Therefore, pH was the ﬁrst optimized parameter. The

Fig. 2. TEM micrographs of synthesized hybrid nano ZrO2 /B2 O3 .

Ö. Yalc¸ınkaya et al. / Journal of Hazardous Materials 195 (2011) 332–339

335

Fig. 3. XRD pattern of synthesized hybrid nano ZrO2 /B2 O3 .

3.4. Effect of the sample ﬂow rate

100 90 80 % Recovery

70 60

Cd

50

Cu

40

Co

30 20 10 0 1

2

3

4

5

6

7

8

9

10

The inﬂuences of ﬂow rates of samples were investigated by controlling the ﬂow rate of sample solution with peristaltic pump. The effect of ﬂow rate of sample solutions on the recoveries of analytes was examined in the range of 1–10 mL min−1 . Under optimum conditions (pH: 6 for Cu(II) and 8 for Co(II) and Cd(II); eluent: 10 mL of 1.0 mol L−1 HNO3 ), the analytes were quantitatively recovered up to 5 mL min−1 of the ﬂow rates. Above 5 mL min−1 the recoveries were decreased gradually. Because the eluent volume was very low, the effect of ﬂow rate of eluent has not been studied.

pH

3.5. Effect of volume of sample solution Fig. 4. Effect of pH of sample solution.

ing 10 ␮g Co(II), Cu(II) and Cd(II) were used. pH of the solutions was adjusted to 6 for Cu(II) and to 8 for Co(II) and Cd(II), and the general procedure was applied. For the elution process, HCl and HNO3 solutions having various concentrations and various volumes were tested (Table 2). As a result of experiments, 10 mL of 1.0 mol L−1 HNO3 and 10 mL of 2.0 mol L−1 HNO3 solutions that give maximum recovery was found as optimum eluents. However, for subsequent experiments, 10 mL of 1.0 mol L−1 HNO3 was used to desorption of Co(II), Cu(II) and Cd(II) from the column due to its lower concentration.

Table 2 The effect of eluent type and volume on the recovery of analytes. Eluent type and volume

Recoverya (%)

10 mL 1 M HCl 10 mL 2 M HCl 10 mL 3 M HCl 5 mL 1 M HCl 5 mL 2 M HCl 5 mL 3 M HCl 10 mL 0.5 M HNO3 10 mL 1 M HNO3 10 mL 2 M HNO3 5 mL 1 M HNO3 5 mL 2 M HNO3

95 94 95 75 78 80 91 98 98 78 82

Cd(II)

a

Mean of the three replicates.

± ± ± ± ± ± ± ± ± ± ±

3 4 4 5 5 4 5 4 3 4 5

Cu(II) 122 118 111 75 78 80 88 99 102 72 76

± ± ± ± ± ± ± ± ± ± ±

Co(II) 6 4 5 3 4 4 3 3 3 3 4

87 ± 85 ± 93 ± – – – 83 ± 95 ± 97 ± 66 ± 71 ±

3 4 4

4 3 3 4 3

Another parameter investigated to ﬁnd the best experimental conditions is the volume of sample solution and/or analyte concentration. In order to determine the maximum applicable sample solution (or minimum analyte concentration), the effect of the volume of sample solution on the recovery of the analytes were investigated by using model solutions and by applying general procedure mentioned in Section 2.5. For this purpose, Co(II), Cu(II) and Cd(II) were preconcentrated from sample volumes of 25, 50, 100, 150, 200 and 250 mL containing 10 ␮g Co(II), Cu(II) and Cd(II) corresponding to analyte concentration of 0.4, 0.2, 0.1, 0.067, 0.05 and 0.04 ␮g mL−1 , respectively. The recovery of analytes were quantitative (>95%) for sample volumes up to 100 mL for Co and Cu and 150 mL for Cd. After the preconcentration of 150 mL for Cd and 100 mL for Co and Cu sample solution, if 10 mL of eluent solution was used for the analysis, the preconcentration factor was found to be 15 for Cd and 10 for Co and Cu. As a result, it can be concluded that 0.1 ␮g mL−1 for Co(II) and Cu(II) and 0.067 ␮g mL−1 for Cd(II) could be determined by applying this preconcentration method; that cannot be determined directly by FAAS with sufﬁcient precision and accuracy. 3.6. Reusability of the sorbent The stability and potential reusability of the sorbent were assessed by monitoring the change in the recoveries of the analytes through several adsorption–elution cycles. The passage of 100 mL of sample (containing 200 ␮g L−1 analyte) solution, 10 mL of 1.0 mol L−1 HNO3 and 50 mL of ultra pure water through the column packed with 200 mg of hybrid sorbent, respectively, was considered one adsorption–elution cycle. The adsorbent was always stored in

336

Ö. Yalc¸ınkaya et al. / Journal of Hazardous Materials 195 (2011) 332–339

water when it was not in use. It was observed that the column could be reused up to 100 times without decrease in the recoveries of the analytes.

Table 3 The effect of some foreign ions on the recovery of the Co(II), Cu(II) and Cd(II) ions. Interfering ions

Concentration (mg L−1 )

3.7. Inﬂuence of foreign ions The preconcentration procedures and thus the recovery of trace metals may be affected by the other constituents of the samples. For this reason, the reliability of the proposed method should be examined in the presence of possible interfering ions of the samples. To investigate the effect of other constituent on the recovery of the analytes, the possible interfering elements were added to 50 mL of model solutions containing 10 ␮g Co(II), Cu(II) and Cd(II) ions as their nitrate or chloride salts. As the tolerance limit, ion concentration causing ±5% deviation in recovery of the analytes is considered. As can be seen from Table 3, foreign ions except Ni(II) have no signiﬁcant effect on the recovery of analyte ions. The results show that above 1 mg L−1 Ni(II) ions interfere with Co(II) signal. In the presence of Ni(II), absorbance of cobalt decreases signiﬁcantly. Preliminary experiments show that Ni(II) can also be retained by the sorbent at pH 8 and can be eluted by same eluent. As a result, it can be concluded that the interfering effect does not occur in preconcentration process but it occurs during atomic absorption measurements. Therefore, in order to eliminate the Ni(II) interference on Co, atomization conditions were optimized. To dissociate inter-metallic compound that could be occur between Ni and Co, higher temperature ﬂame, N2 O-acetylene ﬂame which was proposed by the cookbook of the instrument, was used for Co(II) determination. By using N2 O-acetylene ﬂame, the tolerance limit of Ni(II) increases up to 25 mg L−1 .

Na+

K+

Ca2+

Mg2+

Zn2+

3.8. Adsorption isotherm and adsorption capacity The adsorption isotherms were used to characterize the interaction of each analyte ions with the adsorbent. Among the several isotherm equations, Langmuir adsorption isotherm which is valid for monolayer adsorption onto a surface with a ﬁnite number of identical sites and based on the assumption of surface homogeneity such as equally available adsorption sites, and no interaction between adsorbed species [41], was investigated. Adsorption isotherm provides a relationship between the concentration of analyte ions in the medium and the amount of analyte ions adsorbed on the solid phase when the two phases are at equilibrium. The adsorption isotherms and adsorption capacity of the synthesized nano hybrid sorbent for cobalt(II), copper(II) and cadmium(II) were studied by using the batch method. To obtain adsorption isotherm and determine the adsorption capacity, 50 mL of sample solutions containing 20, 50, 100, 200, 300 and 400 ␮g L−1 Co(II) and Cd(II) were adjusted to pH 8, and 10, 20, 50 and 100 ␮g L−1 Cu(II) was adjusted to pH 6 and added to a beaker containing 100 mg of the sorbent. The solutions were shaken for 2 h at 120 rpm at room temperature to reach equilibrium. Then, 10 mL of solution was taken from each solution and amount of residual Co(II), Cu(II) and Cd(II) in the each solution was determined by ﬂame atomic absorption spectrometry. The proﬁle of adsorption isotherm for Co(II), Cu(II) and Cd(II) were drawn by plotting the milligrams of Co(II), Cu(II) and Cd(II) adsorbed per gram of the sorbent versus the equilibrium concentration of Co(II), Cu(II) and Cd (mg L−1 ) in solution. By using the data obtained from the adsorption isotherms, the most widely used linearized Langmuir equation [42] given below was obtained. From the slopes and intercepts of the linearized plot of (CE /QE ) versus CE (Figs. 5–7), Langmuir constants (Q0 and b) were calculated. CE 1 CE = + QE Q0 Q0 b

Ni2+

Fe3+

Co2+

Cu2+

Mn2+

Cd2+

a b

5 25 50 100 250 500 5 25 50 100 250 500 1000 1500 1 5 25 50 100 250 500 1 5 25 50 100 250 500 1 5 25 50 100 1 5 10 25 1 5 25 1 5 10 25 50 1 5 25 50 100 1 5 25 50 100 1 5 25 50 100

Recoverya (%)

Cd(II)

Cu(II)

Co(II)

99 ± 2 99 ± 2 95 ± 3 96 ± 3 97 ± 3 82 ± 3 99 ± 2 95 ± 3 96 ± 3 95 ± 4 97 ± 4 95 ± 4 95 ± 2 81 ± 4 99 ± 3 98 ± 2 97 ± 3 95 ± 4 98 ± 5 60 ± 4 51 ± 4 99 ± 2 95 ± 3 96 ± 4 98 ± 3 98 ± 4 98 ± 3 83 ± 2 97 ± 3 95 ± 2 95 ± 3 98 ± 5 64 ± 4 99 ± 3 95 ± 4 96 ± 3 58 ± 2 99 ± 4 100 ± 3 97 ± 4 98 ± 2 95 ± 3 99 ± 4 98 ± 5 85 ± 3 97 ± 3 95 ± 2 96 ± 4 95 ± 3 87 ± 3 100 ± 3 99 ± 4 101 ± 4 95 ± 3 74 ± 5 – – – – –

97 ± 2 95 ± 3 99 ± 4 95 ± 3 90 ± 2 78 ± 4 96 ± 3 99 ± 4 100 ± 4 98 ± 2 96 ± 4 95 ± 3 96 ± 5 85 ± 3 99 ± 3 98 ± 2 100 ± 4 95 ± 4 98 ± 3 67 ± 5 56 ± 6 96 ± 3 98 ± 2 97 ± 3 95 ± 4 98 ± 4 75 ± 5 66 ± 4 95 ± 3 99 ± 3 102 ± 3 95 ± 4 86 ± 4 96 ± 4 95 ± 3 88 ± 3 75 ± 4 97 ± 3 99 ± 2 97 ± 4 101 ± 3 98 ± 4 95 ± 3 97 ± 2 80 ± 5 – – – – – 98 ± 4 100 ± 5 95 ± 4 88 ± 3 79 ± 4 99 ± 4 95 ± 3 96 ± 4 90 ± 2 75 ± 4

99 96 99 88 78 65 94 98 95 97 95 87 73

± ± ± ± ± ± ± ± ± ± ± ± ±

3 2 4 4 3 4 3 3 4 3 4 3 4

97 ± 3 99 ± 2 95 ± 3 95 ± 2 88 ± 4 73 ± 5 54 ± 5 99 ± 4 96 ± 3 95 ± 3 90 ± 4 86 ± 5 70 ± 5 56 ± 3 98 ± 4 99 ± 3 95 ± 3 86 ± 4 84 ± 4 95 ± 3 (97 ± 2)b 88 ± 4 (95 ± 3) 76 ± 3 (96 ± 2) 51 ± 4 (95 ± 3) 98 ± 3 95 ± 2 94 ± 4 – – – – – 99 ± 4 102 ± 3 97 ± 3 95 ± 2 88 ± 2 99 ± 4 101 ± 4 85 ± 5 82 ± 3 71 ± 4 98 ± 3 95 ± 4 86 ± 4 80 ± 2 78 ± 4

Mean ± standard deviation for the ﬁve determinations. Values in parenthesis are obtained with N2 O-acetylene ﬂame.

where Q0 (mg g−1 ) is the maximum amount of the sorbed ions per unit mass of sorbent (capacity parameter) to form a complete monolayer coverage on the surface, CE (mg L−1 ) is the equilibrium concentration of analytes, QE (mg g−1 ) is the amount of analyte ions adsorbed per unit mass of sorbent at equilibrium and b (L mg−1 ) is the Langmuir constant related to the afﬁnity of binding sites and

Ö. Yalc¸ınkaya et al. / Journal of Hazardous Materials 195 (2011) 332–339

Table 5 Determination of Co(II), Cu(II) and Cd(II) in certiﬁed reference material (GBW-07605 Tea leaves).

14 y = 0.0311x + 1.3379 R2 = 0.9456

Ce/Qe, g/L

12 10

Analytes

Certiﬁed values (␮g g−1 )

Found valuesa (␮g g−1 )

Relative error (%)

RSDb (%)

Cd Cu Co

0.057 ± 0.008 17.3 ± 0.2 0.18 ± 0.02

0.060 ± 0.002 15.6 ± 0.3 0.183 ± 0.009

+5.3 −9.8 +1.7

3.3 1.9 4.9

8 6 4 2

a

0

b

0

50

100

150

200

250

300

337

350

400

Mean of ﬁve determinations at 95% conﬁdence level. Relative standard deviation (N = 5).

Ce, mg/L Fig. 5. Linearized Langmuir adsorption isotherm of Co(II) on nano ZrO2 /B2 O3 .

Table 6 Determination of Co(II), Cu(II) and Cd(II) in tap water. Analytes

Added (␮g mL−1 )

Founda (␮g mL−1 )

Relative error (%)

RSDb (%)

Cd

– 1 – 1 – 1

NDc 1.01 ± 0.01 NDc 1.01 ± 0.01 ND 0.91 ± 0.03

– +1 – +1 – −9

– 0.99 – 0.99 – 3.3

Cu Co a b c

Fig. 6. Linearized Langmuir adsorption isotherm of Cu(II) on nano ZrO2 /B2 O3 .

Fig. 7. Linearized Langmuir adsorption isotherm of Cd(II) on nano ZrO2 /B2 O3 .

a measure of the stability of the bond formed between metal ions and adsorbent under speciﬁed experimental conditions. The data of the isotherm reveal that the adsorption process conforms to Langmuir model. The Langmuir adsorption constants calculated from the corresponding isotherms with the correlation coefﬁcients are presented in Table 4. Langmuir adsorption capacities of sorbent for Co(II), Cu(II) and Cd(II) were found as 32.2, 46.5 and 109.9 mg g−1 , respectively. 3.9. Analytical ﬁgures of merits As analytical ﬁgures of merit, limit of detection (LOD), limit of quantitation (LOQ), precision and accuracy for the proposed method have been investigated. In order to determine the instrumental detection limit for each analyte, 50 mL of blank solution

Mean of ﬁve determinations at 95% conﬁdence level. Relative standard deviation (N = 5). Not detected.

was passed through the column under the optimum experimental conditions (pH = 6 for Cu(II) and pH = 8 for Co(II) and Cd(II); eluent, 1.0 mol L−1 HNO3 ; ﬂow rate, 5 mL min−1 ). Blank solutions were prepared by adding a minimum amount of the analytes to the water in order to obtain readable analyte signals. The sorbed analyte was eluted by 50 mL of 1.0 mol L−1 HNO3 solution (there is no preconcentration) and signal of this blank solution was measured about 20 times. The instrumental detection limits of the elements based on the ratio of three standard deviation of the blank signal to slope of the calibration curve (3 s/b) were found as 38, 33, 46.5 ␮g L−1 for Co(II), Cu(II) and Cd(II), respectively. The analytical detection limits calculated by dividing the instrumental detection limits [43,44] by the preconcentration factor 15 for Cd(II) and 10 for Co(II) and Cu(II) were 3.8, 3.3, 3.1 ␮g L−1 for Co(II), Cu(II)and Cd(II), respectively. The analytical limit of quantitation (LOQ) based on 10 s/m were 12.7, 11.0, 10.3 ␮g L−1 for Co(II), Cu(II) and Cd(II), respectively. The linear working ranges for the analytes were found as 12.7–5000 ␮g L−1 for Co(II), 11.0–5000 ␮g L−1 for Cu(II) and 10.3–3000 ␮g L−1 for Cd(II) by considering the LOQ values as lower limit of linear working range with a correlation coefﬁcient of about 0.999. The precision of proposed method evaluated as the standard deviations of recoveries obtained from ﬁve replicates under optimum experimental conditions (amount of Co, Cu and Cd, 10 ␮g; volume of model solution, 50 mL; pH 6.0 for Cu and pH 8.0 for Co and Cd; elution solution, 10 mL, 1.0 mol L−1 HNO3 ; ﬂow rate, 5 mL min−1 ) were 96 ± 3%, 95 ± 3%, 98 ± 4% for Co(II), Cu(II) and Cd(II), respectively.

Table 7 Determination of Co(II), Cu(II) and Cd(II) in tea leaves. Analytes Cd

Table 4 Langmuir constants and correlation coefﬁcients of isotherm models for the sorption of Co(II), Cu(II) and Cd(II) ions from aqueous solutions. Analytes

Q0 (mg g−1 )

b (L mg−1 )

r2

Cd(II) Co(II) Cu(II)

109.9 32.2 46.5

0.075 0.232 0.023

0.9984 0.9456 0.8562

Cu Co a b c

Added (␮g g−1 ) – 5 – 20 – 20

Founda (␮g g−1 ) c

ND 5.2 ± 0.3 15.4 ± 0.4 37.8 ± 0.7 ND 18.8 ± 0.8

Relative error (%)

RSDb (%)

– +4 – +7 – −7

– 5.8 2.6 1.9 – 4.3

Mean of ﬁve determinations at 95% conﬁdence level. Relative standard deviation (N = 5). Not detected.

338

Ö. Yalc¸ınkaya et al. / Journal of Hazardous Materials 195 (2011) 332–339

Table 8 Comparative data for various SPE preconcentration methods. Analytes

Preconcentration system

LOD (ng L−1 )

V, Cr, Mn, Co, Ni, Cu, Zn, Cd, Pb Cd Cd, Pb

Nanometer-sized alumina

6–79

Cr, Cu, Pb

Zn, Mn, Ni, Pb, Cd, Cu, Fe, Co Cu, Co, Cd

Nano B2 O3 /TiO2 Modiﬁed nano alumina Magnetic nanoparticles/bismuthiol II Amberlite XAD-16/DHP Nano ZrO2 /B2 O3

1440 150–170 43–85

Adsorption capacity (mg g−1 ) 1–18

49 11.1–16.4 5.4–9.4

PF

Determination techniquea

Ref.

ICP-MS

[19]

FAAS FAAS

[29] [31]

87–96

FAAS

[32]

5

100 250

2900–5020

6.7–37.9

200–300

FAAS

[40]

3100–3800

32.2–109.9

10–15

FAAS

This work

a

PF: preconcentration factor; ICP-OES: inductively coupled plasma optical emission spectrometry; ICP-MS: inductively coupled plasma mass spectrometry; FAAS: ﬂame atomic absorption spectrometry.

The accuracy of the proposed method was tested by determining the content of Co(II), Cu(II) and Cd(II) ions in the certiﬁed reference materials under optimal experimental conditions. As seen in Table 5, the determined values were in good agreement with the certiﬁed values at 95% conﬁdence level. The relative error was found <10% which is acceptable for analytical purposes at these trace metal determinations. 3.10. Application of proposed method The proposed preconcentration method was applied for determination of Co, Cu and Cd in tap water and tea leaves samples, under optimal experimental conditions. The accuracy of method was also checked by measuring the recovery of Co, Cu and Cd in spiked real samples. A good agreement was obtained between added and found value of the analyte. The results obtained are given in Tables 6 and 7. Relative errors below 10%, demonstrate the applicability of the method and independence from matrix constituents of the samples. 3.11. Comparison of the method with others The analytical performance of the nanosorbent comparable with the other conventional sorbent or/and nanosorbents. Some comparative data about sorption is summarized in Table 8. Limit of detection, adsorption capacity and preconcentration factors obtained are comparable to those presented by other methods. The present work has relatively high adsorption capacity when compared to other methods [19,29,31,32,40]. Other parameters, limit of detection and preconcentration factor is relatively lower than those of the others methods. However, the proposed method is simpler than the others. For example, there is a no need to use any complexing and/or chelating agent. 4. Conclusion Nano ZrO2 /B2 O3 hybrid material as a new solid phase extractor provides a simple, selective, accurate, economical, rapid and precise method for preconcentration and determination of Co, Cu and Cd. There is a no need to use any complexing and/or chelating agent for modifying the sorbent, or for adding to the sample solution to form complex compounds of the analytes before the preconcentration procedure to obtain quantitative recovery of the Pb, Co and Cu. The matrix effect appeared with the use of the proposed method is reasonably tolerable. The adsorbent is stable with a recycling period greater than 100 cycles, without major loss in its quantities and metal recovery property. The enrichment factor, detection limit and adsorption capacity of the new proposed sorbent for Co, Cu

and Cd are also satisfactory. The duration time (time required for passing of sample, elution of analytes and cleaning procedure of the column) is about 1 h for a 250 mL sample solution. The detection limits can be improved by using sensitive method such as ICP-OES and ICP-MS. Acknowledgement This work was supported by The Scientiﬁc and Technological Research Council of Turkey (TUBITAK, Project No. 106T668). References [1] J. Minczenski, J. Chwastowska, R. Dybezynski, Separation and Preconcentration Methods in Inorganic Analysis, Ellis Horwood, Chichester, 1982. [2] A.R. Türker, New sorbents for solid phase extraction for metal enrichment, Clean 35 (2007) 548–557. [3] A.B. Tabrizi, Development of a dispersive liquid–liquid microextraction method for iron speciation and determination in different water samples, J. Hazard. Mater. 183 (2010) 688–693. [4] X. Jia, Y. Han, X. Liu, T. Duan, H. Chen, Dispersive liquid–liquid microextraction combined with ﬂow injection inductively coupled plasma mass spectrometry for simultaneous determination of cadmium, lead and bismuth in water samples, Microchim. Acta 171 (2010) 49–56. [5] T. Sardohan, E. Kir, A. Gulec, Y. Cengeloglu, Removal of Cr(III) and Cr(VI) through the plasma modiﬁed and unmodiﬁed ion-exchange membranes, Sep. Purif. Technol. 74 (2010) 14–20. [6] L.N. Moskvin, M.YA. Kamentsev, G.L. Grigor’ev, N.M. Yakimova, Capillary electrophoretic determination of zinc and cadmium ions in aqueous solutions with ion exchange preconcentration, J. Anal. Chem. 65 (2010) 99–102. [7] V.N. Bulut, D. Arslan, D. Ozdes, M. Soylak, M. Tufekci, Preconcentration, separation and spectrophotometric determination of aluminium(III) in water samples and dialysis concentrates at trace levels with 8-hydroxyquinoline–cobalt(II) coprecipitation system, J. Hazard. Mater. 182 (2010) 331–336. [8] S¸. Sac¸macı, S¸. Kartal, Determination of some trace metal ions in various samples by FAAS after separation/preconcentration by copper(II)–BPHA coprecipitation method, Microchim. Acta 170 (2010) 75–82. [9] H. Karimi, M. Ghaedi, A. Shokrollahi, H.R. Rajabi, M. Soylak, B. Karami, Development of a selective and sensitive ﬂotation method for determination of trace amounts of cobalt, nickel, copper and iron in environmental samples, J. Hazard. Mater. 151 (2008) 26–32. [10] A. Shokrollahi, M. Shamsipur, F. Jalali, H. Nomani, Cloud point extraction preconcentration and ﬂame atomic absorption spectrometric determination of low levels of zinc in water and blood serum samples, Cent. Eur. J. Chem. 7 (4) (2009) 938–944. [11] S. Khana, T.G. Kazia, J.A. Baiga, N.F. Kolachia, H.I. Afridi, S.K. Wadhwaa, A.Q. Shaha, G.A. Kandhroa, F. Shaha, Cloud point extraction of vanadium in pharmaceutical formulations, dialysate and parenteral solutions using 8-hydroxyquinoline and nonionic surfactant, J. Hazard. Mater. 182 (2010) 371–376. [12] M. Ince, G. Kaya, M. Yaman, Solid phase extraction and preconcentration of cobalt in mineral waters with PAR-loaded Amberlite XAD-7 and ﬂame atomic absorption spectrometry, Environ. Chem. Lett. 8 (2010) 283–288. [13] G. Xiang, Y. Zhang, X. Jiang, L. He, L. Fan, W. Zhao, Determination of trace copper in food samples by ﬂame atomic absorption spectrometry after solid phase extraction on modiﬁed soybean hull, J. Hazard. Mater. 179 (2010) 521–525. [14] N. Pourreza, K. Ghanemi, Solid phase extraction of cadmium on 2mercaptobenzothiazole loaded on sulfur powder in the medium of ionic liquid 1-butyl-3-methylimidazolium hexaﬂuorophosphate and cold vapor

Ö. Yalc¸ınkaya et al. / Journal of Hazardous Materials 195 (2011) 332–339

[15]

[16]

[17]

[18]

[19]

[20]

[21]

[22]

[23]

[24]

[25]

[26]

[27]

[28]

generation–atomic absorption spectrometric determination, J. Hazard. Mater. 178 (2010) 566–571. N. Pourreza, J. Zolgharnein, A.R. Kiasat, T. Dastyar, Silica gel–polyethylene glycol as a new adsorbent for solid phase extraction of cobalt and nickel and determination by ﬂame atomic absorption spectrometry, Talanta 81 (2010) 773–777. H. Ciftci, Solid phase extraction method for the determination of cobalt in water samples on Duolite XAD-761 resin using 4-(2-pyridylazo) resorcinol by FAAS, Curr. Anal. Chem. 6 (2010) 154–160. A.S. Pereira, G. Ferreira, L. Caetano, M.A.U. Martines, P.M. Padilha, A. Santos, G.R. Castro, Preconcentration and determination of Cu(II) in a fresh water sample using modiﬁed silica gel as a solid phase extraction adsorbent, J. Hazard. Mater. 175 (2010) 399–403. R.K. Sharma, P. Pant, Preconcentration and determination of trace metal ions from aqueous samples by newly developed gallic acid modiﬁed Amberlite XAD16 chelating resin, J. Hazard. Mater. 163 (2009) 295–301. J. Yin, Z.C. Jiang, G. Chang, B. Hu, Simultaneous on-line preconcentration and determination of trace metals in environmental samples by ﬂow injection combined with inductively coupled plasma mass spectrometry using a nanometer-sized alumina packed micro-column, Anal. Chim. Acta 540 (2005) 333–339. Y. Zhai, X. Chang, Y. Cui, N. Lian, S. Lai, H. Zhen, Q. He, Selective determination of trace mercury(II) after preconcentration with 4-(2pyridylazo)-resorcinolmodiﬁed nanometer-sized SiO2 particles from sample solutions, Microchim. Acta 154 (2006) 253–259. L. Yang, B. Hu, Z. Jiang, H. Pan, On-line separation and preconcentration of Trace metals in biological samples using a microcolumn loaded with PAN-modiﬁed nanometer-sized titanium dioxide, and their determination by ICP-AES, Microchim. Acta 144 (2004) 227–231. V.A. Lemos, L.S.G. Teixeira, M.A. Bezerra, A.C.S. Costa, J.T. Castro, L.A.M. Cardoso, D.S. Jesus, E.S. Santos, P.X. Baliza, L.N. Santos, New materials for solid phase extraction of trace elements, Appl. Spectrosc. Rev. 43 (2008) 303–334. L. Zhang, T. Huang, M. Zhang, X. Guo, Z. Yuan, Studies on the capability and behavior of adsorption of thallium on nano-Al2 O3 , J. Hazard. Mater. 157 (2008) 352–357. M. Natali, G. Carta, V. Rigato, G. Rossetto, G. Salmaso, P. Zanella, Chemical, morphological and nano-techanical characterizations of Al2 O3 thin ﬁlms deposited by metal organic chemical vapour deposition on AISI 304 stainless steel, Electrochim. Acta 50 (2005) 4615–4620. Ö.B. Ergü, M. Gürü, C. Cabbar, Preparation and characterization of alumina–zirconia composite material with different acid ratios by the sol–gel method, Cent. Eur. J. Chem. 6 (2008) 482–487. M. Thiruchitrambalam, V.R. Paklar, V. Gopinathan, Hydrolysis of aluminum metal and sol–gel processing of nano alumina, Mater. Lett. 58 (2004) 3063–3066. J.S. Suleiman, B. Hu, X. Pu, C. Huang, Z. Jiang, Nanometer-sized zirconium dioxide micro column separation /preconcentration of trace metals and their determination by ICP-OES in environmental and biological samples, Microchim. Acta 159 (2007) 379–385. J. Yin, Z. Jiang, G. Chang, B. Hu, Simultaneous on-line preconcentration and determination of trace metals in environmental samples by ﬂow injection combined with inductively coupled plasma mass spectrometry using a nanometer-sized alumina packed micro-column, Anal. Chim. Acta 540 (2005) 333–339.

339

[29] O.M. Kalfa, Ö. Yalc¸ınkaya, A.R. Türker, Synthesis of nano B2 O3 /TiO2 composite material as a new solid phase extractor and its application to preconcentration and separation of cadmium, J. Hazard. Mater. 166 (2009) 455–461. [30] Y. Liu, P. Liang, L. Guo, Nanometer titanium dioxide immobilized on silica gel as sorbent for preconcentration of metal ions prior to their determination by inductively coupled plasma atomic emission spectrometry, Talanta 68 (2005) 25–30. [31] M. Ezoddin, F. Shemirani, K. Abdi, M.K. Saghezchi, M.R. Jamali, Application of modiﬁed nano-alumina as a solid phase extraction sorbent for the preconcentration of Cd and Pb in water and herbal samples prior to ﬂame atomic absorption spectrometry determination, J. Hazard. Mater. 178 (2010) 900–905. [32] J.S. Suleiman, B. Hu, H. Peng, C. Huang, Separation/preconcentration of trace amounts of Cr, Cu and Pb in environmental samples by magnetic solid phase extraction with bismuthiol-II-immobilized magnetic nanoparticles and their determination by ICP-OES, Talanta 77 (2009) 1579–1583. [33] S.H. Huang, D.H. Chen, Rapid removal of heavy metal cations and anions from aqueous solutions by an amino-functionalized magnetic nano-adsorbent, J. Hazard. Mater. 163 (2009) 174–179. [34] R.S. Amais, J.S. Ribeiro, M.G. Segatelli, I.V.P. Yoshida, P.O. Luccas, C.R.T. Tarley, Assessment of nanocomposite alumina supported on multi-wall carbon nanotubes as sorbent for on-line nickel preconcentration in water samples, Sep. Purif. Technol. 58 (2007) 122–128. [35] M.N. Nezhati, H.A. Panahi, E. Moniri, S.R. Kelahrodi, F. Assadian, M. Karimi, Synthesis, characterization and application of allyl phenol modiﬁed Amberlite XAD-4 resin for preconcentration and determination of copper in water samples, Korean J. Chem. Eng. 27 (2010) 1269–1274. [36] S. Ozdemir, R. Gul-Guven, E. Kilinc, M. Dogru, S. Erdogan, Preconcentration of cadmium and nickel using the bioadsorbent Geobacillus thermoleovorans subsp stromboliensis immobilized on Amberlite XAD-4, Microchim. Acta 169 (2010) 79–85. [37] O. Yalc¸ınkaya, O.M. Kalfa, A.R. Türker, Chelating agent free solid phase extraction (CAF-SPE) method for separation and/or preconcentration of iron(III) ions, Turk. J. Chem. 34 (2010) 207–217. [38] B. Saad, C.C. Chong, A.S.M. Ali, M.F. Bari, I.A. Rahman, N. Mohamad, M.I. Saleh, Selective removal of heavy metal ions using sol–gel immobilized and SPEcoated thiacrown ethers, Anal. Chim. Acta 555 (2006) 146–156. [39] I.G. Dakova, I.B. Karadjova, V.T. Georgieva, G.S. Georgiev, Polycarboxylic microsphere polymer gel for solid phase extraction of trace elements, Microchim. Acta 164 (2009) 55–61. [40] G. Venkatesh, A.K. Singh, Macromolecular chelator-Amberlite XAD-16 anchored with 2, 3-Dihydroxypyridine (DHP) for enrichment of metal ions before their determination by ﬂame atomic absorption spectrometry, Sep. Sci. Technol. 42 (2007) 3429–3446. [41] K.H. Chu, Removal of copper from aqueous solution by chitosan in prawn shell: adsorption equilibrium and kinetics, J. Hazard. Mater. 90 (2002) 77–95. [42] R.A.K. Rao, M.A. Khan, Removal and recovery of Cu(II), Cd(II) and Pb(II) ions from single and multimetal systems by batch and column operation on neem oil cake (NOC), Sep. Purif. Technol. 57 (2007) 394–402. [43] A.C. Sahayam, Speciation of Cr(III) and Cr(VI) in potable waters by using activated neutral alumina as collector and ET-AAS for determination, Anal. Bioanal. Chem. 372 (2002) 840–842. [44] M.A.A. Akl, I.M.M. Kenawy, R.R. Lasheen, Organically modiﬁed silica gel and ﬂame atomic absorption spectrometry: employment for separation and preconcentration of nine trace heavy metals for their determination in natural aqueous systems, J. Microchem. 78 (2004) 143–156.

Journal of Hazardous Materials 195 (2011) 340–345

Contents lists available at SciVerse ScienceDirect

Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat

Synthesis and adsorption properties of titanosilicates ETS-4 and ETS-10 from ﬂy ash Liying Liu a,b , Ranjeet Singh b,∗ , Gang Li b , Penny Xiao b , Paul Webley b , Yuchun Zhai a a b

School of Material and Metallurgy, Northeastern University, Shenyang, Liaoning 110004, PR China Cooperative Research Centre for Greenhouse Gas Technologies (CO2CRC), Department of Chemical Engineering, Monash University, Clayton, Victoria 3800, Australia

a r t i c l e

i n f o

Article history: Received 17 June 2011 Received in revised form 15 August 2011 Accepted 15 August 2011 Available online 22 August 2011 Keywords: Fly ash Titanosilicates ETS-4 ETS-10 Adsorption

a b s t r a c t ETS-4 and ETS-10 titanosilicates were prepared from ﬂy ash and anatase, as silica and titanium sources respectively, via a hydrothermal procedure for the ﬁrst time. The fusion of ﬂy ash by alkali was carried out at a relatively low temperature and the use potassium ﬂuoride salt was avoided in the synthesis of ETS. The by-product of this process is mainly NaCl, which is a useful source material for industry. The energy efﬁciency and yield of the synthesis process was improved by directly recycling the ﬁnal ﬁltrate after recovering the product viz ETS-4. All the ETS materials were characterized in terms of structural morphology, thermal stability and surface/pore properties. The properties of ETS-4 prepared from ﬂy ash by the ﬁltrate recycling method were comparable to that from commercial sources. The results show that ETS type materials can be prepared from cheaper resources, with good purity, comparable physicochemical properties as well as excellent adsorption properties with lower environmental impact. © 2011 Elsevier B.V. All rights reserved.

1. Introduction A large amount of ﬂy ash is produced from coal-ﬁred power stations annually. As a solid waste, more than 65% is disposed of in landﬁlls and ash ponds. Recycling coal ﬂy ash has received extensive attention due to increasing landﬁll costs and negative environmental impact. Fly ash usually is rich in Si and Al, hence, converting ﬂy ash into a useful commodity (e.g. cement) has a number of beneﬁts from both economic and environmental aspects. In particular, a large number of patents and technical articles have proposed different methods for zeolites synthesis from ﬂy ash [1,2], using hydrothermal processes. Various types of zeolites have been prepared from ﬂy ash including Na–A, Na–X, Na–P1, K-chabazite, ZSM-5, MCM-41, etc. These adsorbents can be used for removal of heavy metals from waste water or as adsorbents for the removal of SO2 , NH3 , and CO2 from industrial gas sources [3–10]. Titanosilicate materials possess zeolite-like properties and ﬁnd numerous applications in catalysis, adsorption and separation [11–14], since ﬁrst reported by Chapman and Rod [15]. Titanosilicates ETS-4, ETS-10 and ETS-14 were later developed and patented by Engelhard Corporation. ETS-4 has a mixed octahedral/tetrahedral structure, with small pores between 0.3 and 0.4 nm, which can be easily tuned by progressive dehydration

∗ Corresponding author. Tel.: +61 3 99052632; fax: +61 3 99055686. E-mail address: [email protected] (R. Singh). 0304-3894/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.jhazmat.2011.08.046

(Molecular Gate Effect). Hence, ETS-4 could be optimized as an adsorbent for separation of gases of close size, for example methane/nitrogen, oxygen/argon, etc. [16,17]. The Na form of ETS4 is known to be thermally unstable, however, ion exchanging Na+ with bivalent ions such as Sr2+ , Ba2+ , Ca2+ , Mg2+ , etc., results in improved thermal stability, which can be then be exploited practically for gas separation [14,18,19]. Pressure Swing Adsorption (PSA) processes using Sr-ETS-4 has already been commercialized for the important N2 /CH4 separation [20]. Additionally, adsorption based separation of CO2 , CH4 and C2 H5 as well as O2 and Ar have also been examined [21,22] by PSA with titanosilicates. ETS-10 on the other hand is a large pore titanosilicate with an effective pore size of approximately 8 A˚ with high thermal stability and good cation exchange capacity [23]. ETS-10 materials generally ﬁnd application in catalysis viz acid–base catalysis, photocatalysis, etc. ETS-10 has gained considerable attention as an ionic sieving material for heavy and radioactive metal ions [24–26]. Furthermore, cation exchanged ETS-10 can demonstrate separation selectivity and it has been reported as a suitable adsorbent for PSA-based CO2 separation from methane [21] and for the separation of ethylene/ethane or propylene/propane mixtures [27,28]. Additionally, ETS-4 and ETS-10 membranes have been prepared and their performance was tested in water: ethanol separation [29–32]. ETS-4 and ETS-10 can be synthesized by a hydrothermal process, wherein the titanium source can be TiCl3 or TiCl4 in HCl, anatase, rutile, etc. and the silica source is normally fumed silica, sodium silicate, rice husk, etc. in the presence of an alkali, ﬂuoride or chloride [27], as described in the literature. Additionally, the effect of seeds

L. Liu et al. / Journal of Hazardous Materials 195 (2011) 340–345 Table 1 Chemical composition of ﬂy ash in wt%. SiO2

Al2 O3

Fe2 O3

CaO

C

TiO2

MgO

53.1

39.5

3.37

1.76

0.43

1.32

0.37

on the particle size [33], synthesis time, pH [24,35], etc. in case of ETS-10, has also been studied. Although ETS-10 has been prepared using rice husk as silica source [37], to the best of our knowledge, synthesis of titanosilicate ETS-4 and ETS-10 using ﬂy ash as silica source has not been examined, perhaps due to the large portion of Al in the ﬂy ash (as shown in Table 1). Moreover, in the case of ETS4, we demonstrate that the supernatant liquid after recovering the product can be recycled to reproducibly synthesize ETS-4 thus minimizing waste. The ultimate aim of this work was to demonstrate that ETS-4 and ETS-10 can be prepared from low cost, eco-friendly starting materials (use of KF was avoided) and the un-reacted materials in the ﬁltrate can be reused and the produced material can be used for adsorption of green house gases such as CO2 , CH4 , etc. 2. Experimental 2.1. Synthesis Fly ash used for this study was provided by Shoutou electricity plant (China) and its composition is described in Table 1. The synthesis procedure for both ETS-4 and ETS-10 is summarized schematically in the ﬂow diagram (Fig. 1). In a typical procedure,

Fly ash

341

60 g of ﬂy ash and 48 g of NaOH (Ajax Fine Chemicals) were mixed with 120 g of water, followed by heating the slurry (in a Parr autoclave) at 120 ◦ C for 2 h under continuous stirring. The mixture was diluted by adding 80 g of water, followed by ﬁltration. Then, concentrated HCl solution (32 wt%) was added drop wise to the resulting ﬁltrate for pH adjustment. To synthesize ETS-4, 0.5 g of ETS-4 seeds and 2.5 g of commercial anatase (99.8% Sigma–Aldrich) was added to the above ﬁltrate with vigorously stirring for 30 min. To synthesize ETS-10, 3.6 g of anatase and 13.3 g of KCl (Ajax Fine Chemicals) were added to the ﬂy ash slurry. In both of the cases, the mixture was transferred into a 300 mL stainless steel autoclave (Parr Instruments, USA) and heated at 230 ◦ C for 24 h under static conditions. After cooling to room temperature, the resultant solid was ﬁltered, washed three times with deionized water, and dried at 70 ◦ C overnight. The samples prepared from ﬂy ash were labeled as ETS-4(FA) and ETS-10(FA), respectively. The ﬁltrate containing unreacted Na+ , Cl− , TiO4 − , SiO3 2− after the hydrothermal treatment stage in the case of ETS-4 was recycled for the next synthesis batch as demonstrated in the ﬂow diagram. In a typical procedure, the ﬁltrate was evaporated to 120 mL. Then, 30 g of ﬂy ash and 24 g of NaOH were added to the ﬁltrate, followed by the same procedure detailed above, and the resulting samples were labeled as ETS-4(FAR). For comparison purposes, ETS-4 and ETS-10 were also prepared by reported methods using commercial silica sources [34,35]. In a typical synthesis of ETS-4, 12.5 g of colloidal silica (30 wt.% SiO2 , Ludox) solution, 2.5 g of NaOH, 2 g of NaCl, 1 g of anatase were mixed with 12.5 g of water, followed by vigorously stirring for 30 min. To prepare ETS-10, 20 g of sodium silicate solution (8.5 wt% Na2 O,

NaOH(aq)

Silicon exraction at 120ºC for 1h Filtration -

Na +,Si O 32- , OH ,trace Al 3+

Recycle

HCl (aq)

Filter cake

pH adjustment

Na2 Si O3 , NaCl, NaOH

Anatase

Hydrothermal treatment at 230ºC for 24h

As materials for preparing Al 2O 3 KCl (only for ETS-10)

Filtration

Na +, (K+),Cl - ,Ti O4,- Si O32-

Filter cake Rinse

ETS

NaCl

Fig. 1. Flowchart for synthesis of ETS-4 and ETS-10 from ﬂy ash.

342

L. Liu et al. / Journal of Hazardous Materials 195 (2011) 340–345

c A

b A

a

10

20

30

40

b

Intensity(a.u.)

Intensity(a.u.)

A

a

A

50

0

10

20

30

2.2. Measurement and characterization The chemical composition of the ﬂy ash, as-synthesized ETS-4 and ETS-10 were determined by Inductively Coupled Plasmamass Spectrometry (ICP-AES, Ultratrace, Perth). The crystalline properties of the samples were examined by X-ray diffraction (XRD) using a Philips PW1140/90 diffractometer with Cu K␣ ˚ FE-SEM (Field Emission Scanning Elecradiation ( = 1.5404 A). tron Microscopy) analysis was conducted by employing a JEOL 6300F/7001F scanning electron microscope operated at 15 kV. All samples were platinum coated prior to measurement. BET and Langmuir surface areas measurements were conducted on an ASAP 2020 analyzer (Micromeritics, USA) using N2 and H2 as probe gases at 77 K in the relative pressure range of 0.05–0.25. The total pore volume was evaluated with a single point method at a relative pressure of 0.995. CO2 , N2 , CH4 adsorption isotherms were measured on an ASAP 2010 gas adsorption analyzer. Prior to measurements, all the samples were degassed under vacuum (90 ◦ C for ETS-4 and 330 ◦ C for ETS-10), for 8 h. 3. Results and discussion 3.1. Structure and morphology The X-ray diffraction pattern for ﬂy ash (Fig. 2a) indicates that the crystalline phases are mainly ␣-quartz (SiO2 ) and mullite (3Al2 O3 ·2SiO2 ), identiﬁed by the sharp peaks, with coexistence of an amorphous phase in the range of about 2 = 15–30◦ . XRD patterns of as-synthesized ETS-4(FA) at pH 12.5 and pH 11.6 are shown in Fig. 2b and c respectively. The overall X-ray diffraction patterns show distinct peaks, which are consistent with those reported for ETS-4. As expected, the ETS-4(FA) sample synthesized at pH > 12 shows better crystallinity [36]. However, in all samples, a small peak attributed to anatase phase is seen at about

Fig. 3. X-ray diffraction patterns for: (a) ETS-4(T), (b) ETS-4(FAR), “A” represents anatase impurity.

25◦ , which indicates incomplete conversion of anatase into the ETS4 product. The intensity of the peak corresponding to anatase can be reduced by decreasing the amount of the anatase in the gel, however, the yield of ETS-4 reduces consequently. Furthermore, the ETS-4(T) sample, prepared from colloidal silica source also shows anatase impurity (Fig. 3a), suggesting that when anatase is used as a Ti-source, the conversion is usually incomplete and it is difﬁcult to get pure ETS materials [35,37], although product purity can be improved by adjusting the ratio of Si/Ti. Hydrothermal reaction conditions are also critical in controlling the purity of the synthesized ETS materials [36]. The XRD results show that when the pH value is in the range of 12–12.5, high purity ETS-4 can be obtained. However, ETS-10 impurity is detected when the pH is around 11.6 (Fig. 2c) and when the pH is below 11, ETS-10 became the main product. Fig. 3b shows the X-ray diffraction patterns for ETS-4(FAR) sample prepared by recycling the efﬂuent. It is clear from the pattern that high-quality ETS-4 can be prepared by recycling the gel solution and suggests that there is no additional solid formed during this recycling process. As expected, the yield of ETS-4 increased with the number of cycles due to more efﬁcient utilization of the reactants. However, with the increase of recycle cycles, NaCl in the liquid phase accumulates and eventually reaches saturation and co-precipitates with ETS-4, which can be easily removed by rinsing the solid. Hence, from environmental 600

d Q

400

Intensity(a.u.)

26.5 wt% SiO2 , BDH) was diluted with 20 g of water. Then 6.9 g NaCl, 1.3 g KCl and 1.3 g anatase were added and the solution was homogenized by continuous stirring for 30 min. In both synthesis procedures, the gel was heated in an autoclave at 230 ◦ C for 24 h under static conditions. The ﬁnal products were ﬁltered, washed three times with deionized water and dried at 70 ◦ C overnight. The samples were labeled as ETS-4(T) and ETS-10(T), respectively.

50

2θ

2θ Fig. 2. X-ray diffraction patterns for: (a) ﬂy ash, (b) as-synthesized ETS-4(FA) obtained at pH 12.5, (c) as-synthesized ETS-4(FA) obtained at pH 11.6, circles represents anatase impurity, while squares indicate ETS-10 impurity.

40

A Q

c A

200

b

Q

Q

A

a 0 0

10

20

30

40

50

2θ Fig. 4. X-ray diffraction patterns for: (a) ETS-10 (T), ETS-10(FA) obtained at pH of (b) 11.2, (c) 10.6, (d) 11.6, A corresponds to unreacted anatase, Q represents quartz impurity, hollow square represents ETS-4 impurity.

L. Liu et al. / Journal of Hazardous Materials 195 (2011) 340–345

perspective, this process is eco-friendly because the only waste generated is NaCl (a by-product). Fig. 4a–d shows the X-ray diffraction patterns for both ETS-10 samples prepared using colloidal silica source as well as ﬂy ash. ETS-10(T) has the distinct peaks (Fig. 4a), corresponding with the characteristic signatures of ETS-10. However, an incomplete transformation of ETS-10 was observed, which can again be attributed to the commercial anatase source [35,37]. Similarly, all the ETS-

343

10(FA) samples contain a small amount of unreacted anatase in addition to quartz (Fig. 4b and c). According to the XRD results, the best ETS-10(FA) was obtained at pH 11.2, with a small amount of quartz (Fig. 4b). At pH 10.6, ETS-10(FA) was contaminated with a large amount of quartz (Fig. 4c), while at pH 11.6, ETS-10(FA) was heavily contaminated with ETS-4 impurity (Fig. 4d). In agreement with past observations in the literature we observe that pure ETS10 could not be obtained using commercial anatase as titanium

Fig. 5. FE-SEM images of: (a) original ﬂy ash, (b), ETS-4(T), (c) ETS-4(FA), (d) inset EDX spectra of ETS-4(FA), (e), ETS-10(T), (f) ETS-10 (FA).

344

L. Liu et al. / Journal of Hazardous Materials 195 (2011) 340–345 Table 3 The structural properties of ETS-10 samples using sodium silica solution and ﬂy ash.

Table 2 Chemical composition of as-synthesized ETS-4 and ETS-10. Component (wt%)

ETS-4(T)

ETS-4(FA)

ETS-10(T)

ETS-10(FA)

Samples

Surface area (m2 /g)

The total pore volume

Si Al Ti Na K

21.1 0.02 13.6 6.47 0.2

17.5 0.6 19 9.53 0.2

26.3 0.1 11.9 6.6 2.9

22.1 0.19 19.3 4.66 3.3

Fly ash ETS-4(T) ETS-4(FA) ETS-4(FA) ETS-10(T) ETS-10(FA)

2.2a 203b 5.2a 210b 313a 304a

0.005 0.08 0.01 0.10 0.20 0.19

a b

source, as the particle size of anatase is too large compared with other titanium sources such as P25 or nano-anatase [35,37]. The particle size of anatase is known to be the controlling factor in the purity of the resultant ETS materials [33]. The chemical composition by ICP of the as-synthesized ETS samples is displayed in Table 2. Fly ash contains a considerable amount of Al (20.9%), as shown in Table 1. ICP data reveals that ETS-4(FA) and ETS-10(FA) contain a very small amount of Al impurity, about 0.6% and 0.19%, respectively. This suggests that most of the alumina from the ﬂy ash is removed during the process. Fig. 5 shows the scanning electron micrographs of ﬂy ash and ETS samples. SEM for ﬂy ash (Fig. 5a) shows particles with a wide size distribution and spherical morphology. ETS-4(T) and ETS-4(FA) (Fig. 5b and c) samples show inter-grown aggregates of plate-like crystals with rectangular morphology, with approximately 5.0 (±2.0) × 1.8 (±0.2) ␮m in size. The elemental analysis using EDX (inset) gave high intensity signals for Ti, Si, Na and O, and does not show presence of any additional elements, suggesting that ETS-4(FA) contains only traces of elemental impurity of Al2 O3 . The morphology of as-synthesized ETS-10 samples is shown in Fig. 5e–f. Regardless of the silica source, both ETS-10(T) and ETS-10(FA) samples display truncated bipyramid structure suggesting that the morphology of ETS-10 materials is dependent on Ti source rather than silica source. The particles of ETS-10(FA) show uniform distribution approximately 1.5 (±0.2) × 0.6 (±0.1) ␮m in size. The elemental analysis using EDX (not shown) showed high intensity signals corresponding to Ti, Si, Na and O, and do not show the presence of any additional elements, suggesting that the trace impurities were removed during the fusion and extraction process.

Denotes BET surface area obtained by N2 adsorption. Denotes Langmuir surface area obtained by H2 adsorption.

3.2. Adsorption properties Speciﬁc BET surface area and total pore volume of the ETS-4(FA) sample measured with liquid nitrogen are quite low as expected, about 5.2 m2 /g and 0.01 cm3 /g, respectively as N2 has practically no access to the micropore system [38]. In order to investigate the details of textural properties of as-synthesized ETS-4(FA), smaller diameter H2 gas adsorption isotherm was measured at 77 K. A typical Type I isotherm (Fig. 6a) was measured, conﬁrming presence of only micropores. Single-point total pore volume estimated from the amount adsorbed at a relative pressure P/P◦ of 0.995 was 0.10 cm3 /g. The Langmuir surface areas of ETS-4(T) and ETS-4(FA) (Table 2) are 203 m2 /g and 210 m2 /g, respectively, and are slightly lower than the reported values (247 m2 /g) in the literature [39], which may be attributed to the small amount of impurity such as unreacted anatase. The N2 adsorption isotherm (77 K) of ETS-10(FA) shows a typical Type I isotherm and a narrow pore size distribution centered at the value of 0.7 nm (not shown). The speciﬁc BET surface area of ETS-10(T) and ETS-10(FA) shown in Table 3 were nearly similar viz 313 m2 /g and 304 m2 /g, respectively, well within experimental error. Single-point total pore volumes estimated from the amount adsorbed at a relative pressure P/P◦ of 0.995 was 0.20 cm3 /g and 0.19 cm3 /g, respectively. The single-component equilibrium isotherms for N2 and CH4 were also measured at 30 ◦ C for the ETS-4(T) and ETS-4(FA) and are shown in Fig. 6b and c. All the isotherms were reversible. The adsorption loadings of both gases on ETS-4(FA) and ETS-4(T) were in agreement suggesting that ETS-4 prepared from ﬂy ash was similar to ETS-4 prepared from traditional sources. CH4 adsorption capacity was higher as compared to N2 , which could be attributed

3.0

3 0.5

a

a

0.4

-1

Amount adsorbed(mmol.g )

Amount adsorbed (mmol/g)

2.5

2.0

c 0.3 1.5 0.2

1.0

b 0.1

0.5

0.0 0

200

400

600

800

200

400

600

800

Pressure(mmHg) Fig. 6. (a) Hydrogen adsorption isotherm of synthesized ETS-4(FA) at 77 K. Adsorption isotherms of (b) N2 and (c) CH4 in the ETS-4(T) and ETS-4(FA) at 30 ◦ C, where the square denotes ETS-4(T) and circles denotes ETS-4 (FA).

2

b

1

c

0 0

200

400

600

800

Pressure(mmHg) Fig. 7. Adsorption isotherms of: (a) CO2 , (b) CH4 , (c) N2 in ETS-10(T) and ETS-10(FA) at 25 ◦ C, where the square denotes ETS-10(T) and circle denotes ETS-10 (FA).

L. Liu et al. / Journal of Hazardous Materials 195 (2011) 340–345

to the induced octopole moment that CH4 exhibits and its large polarizability [40]. The representative adsorption isotherms measured at 25 ◦ C for pure CO2 , N2 and CH4 on the ETS-10(T) and ETS-10(FA) samples were shown in Fig. 7. The adsorption loadings of CO2 , CH4 and N2 on ETS-10(T) and ETS-10(FA) are comparable. CO2 adsorption capacity was the highest followed by methane and nitrogen. Furthermore, selectivity of methane over nitrogen was caused by the higher polarizability [40]. ETS-10(FA) showed good selectivity for CO2 over N2 at 25 ◦ C. Thus, prepared ETS-10 (FA) could be an appropriate adsorbent for CO2 separation in the future. Furthermore, the adsorption loadings of gases on ETS-10(FA) were comparable with the data published by Anson et al. [21], suggesting the ETS10 prepared with ﬂy ash has similar adsorption properties with counterparts obtained with commercial sources. 4. Conclusion Fly ash was successfully utilized for production of ETS for the ﬁrst time in the absence of an organic template. High purity ETS-4 and ETS-10 can be obtained at pH of 12.5 and 11.2, respectively. ETS-4 and ETS-10 obtained using ﬂy ash exhibit high surface areas and pore volumes, in spite of trace impurities. The properties of ETS prepared from ﬂy ash are comparable with those from commercial sources. The recycling of the efﬂuent after removal of solids ensures efﬁcient use of the reactant materials, while maintaining high product purity, and most importantly this process is extremely beneﬁcial from an environmental perspective. Acknowledgements The authors gratefully acknowledge CO2CRC for ﬁnancial support. Liying Liu would also like to acknowledge NNSF of China (51074205) for ﬁnancial support. References ˜ A. Alastuey, E. Hernández, A. López-Soler, F. [1] X. Querol, N. Moreno, J.C. Umana, Plana, Synthesis of zeolites from coal ﬂy ash: an overview, Int. J. Coal Geol. 50 (2002) 413–423. ˜ R. Juan, S. Hernández, C. Fernandez-Pereira, [2] X. Querol, N. Moreno, J.C. Umana, C. Ayora, M. Janssen, J. Garcıˇıa-Martıˇınez, A. Linares-Solano, D. Cazorla-Amoros, Application of zeolitic material synthesized from ﬂy ash to the decontamination of waste water and ﬂue gas, J. Chem. Technol. Biotechnol. 77 (2002) 292–298. [3] H. Tanaka, S. Furusawa, R. Hino, Synthesis, characterization, and formation process of Na–X zeolite from coal ﬂy ash, J. Mater. Synth. Process. 10 (2002) 143–148. [4] K.S. Hui, C.Y.H. Chao, Pure, single phase, high crystalline, chamfered-edge zeolite 4A synthesized from coal ﬂy ash for use as a builder in detergents, J. Hazard. Mater. 137 (2006) 401–409. [5] H.L. Chang, W.H. Shih, Synthesis of zeolites A and X from ﬂy ashes and their ionexchange behavior with cobalt ions, Ind. Eng. Chem. Res. 39 (2000) 4185–4191. [6] H. Tanaka, Y. Sakai, R. Hino, Formation of Na–A and –X zeolites from waste solutions in con version of coal ﬂy ash to zeolites, Mater. Res. Bull. 37 (2002) 1873–1884. [7] K.S. Hui, C.Y.H. Chao, Effects of step-change of synthesis temperature on synthesis of zeolite 4A from coal ﬂy ash, Microporous Mesoporous Mater. 88 (2006) 145–151. [8] H.J. Koroglu, A. Sarioglan, M. Tatlier, A. Erdem-Senatalar, O.T. Savasci, Effects of low-temperature gel aging on the synthesis of zeolite Y at different alkalinities, J. Cryst. Growth 241 (2002) 481–488. [9] O. Andac, M. Tather, A. Sirkecioglu, I. Ece, A. Erdem-Senatalar, Effects of ultrasound on zeolite: a synthesis, Microporous Mesoporous Mater. 79 (2005) 225–233. [10] M. Gross-Lorgouilloux, M. Soulard, P. Caullet, J. Patarin, E. Moleiro, I. Saude, Conversion of coal ﬂy ashes into faujasite under soft temperature and pressure conditions: inﬂuence of additional silica, Microporous Mesoporous Mater. 127 (2010) 41–49. [11] Y. Shiraishi, D. Tsukamoto, T. Hirai, Selective photocatalytic transformations on microporous titanosilicate ETS-10 driven by size and polarity of molecules, Langmuir 24 (2008) 12658–12663.

345

[12] N. Ismail, I.H.A. El-Maksod, H. Ezzat, Synthesis and characterization of titanosilicates from white sand silica and its hydrogen uptake, Int. J. Hydrogen Energy 35 (2010) 10359–10365. [13] A. Thangaraj, S.M. Kuznicki, G.S. Koermer, Titanium silicate molecular sieve oxidation catalysts and the production thereof, U.S. Patent 6087514 (2000). [14] S.M. Kuznicki, I. Petrovich, B.T. Desai, Small-pored crystalline titanium molecular sieve zeolites and their use in gas separation processes, U.S. Patent 6068682 (1990). [15] D.M. Chapman, A.L. Roe, Synthesis, characterization and crystal chemistry of microporous titanium–silicate materials, Zeolites 10 (1990) 730–737. [16] S.M Kuznicki, V.A. Bell, S. Nair, H.W. Hillhouse, R.M. Jacubinas, C.M. Braunbarth, B.H. Toby, M. Tsapatsis, A titanosilicate molecular sieve with adjustable pores for size-selective adsorption of molecules, Nature 412 (2001) 720–724. [17] S.M. Kuznicki, Preparation of small-pored crystalline titanium molecular sieve zeolites, U.S. Patent 4938939 (1990). [18] S.M. Kuznicki, V.A. Bell, I. Petrovic, P.W. Blosser, Separation of nitrogen from mixtures thereof with methane utilizing barium exchanged ETS-4, U.S. Patent 5989316 (1999). [19] S. Cavenati, C.A. Grande, F.V.S. Lopes, A.E. Rodrigues, Adsorption of small molecules on alkali-earth modiﬁed titanosilicates, Microporous Mesoporous Mater. 121 (2009) 114–120. [20] K.F. Butwell, B.B. Dolan, S.M. Kuznicki, Selective removal of nitrogen from natural gas by pressure swing adsorption, U.S. Patent 6315817 (2001). [21] A. Anson, C.C.H. Lin, S.M. Kuznicki, J.A. Sawada, Adsorption of carbon dioxide, ethane, and methane on titanosilicate type molecular sieves, Chem. Eng. Sci. 64 (2009) 3683–3687. [22] R.S. Pillai, S.A. Peter, R.V. Jasra, Adsorption of carbon dioxide, methane, nitrogen, oxygen and argon in NaETS-4, Microporous Mesoporous Mater. 113 (2008) 268–276. [23] S.M. Kuznicki, Large-pore crystalline titanium molecular sieve zeolites, U.S. Patent 4853202 (1989). [24] J. Rocha, P. Brandao, J.D. Pedrosa de Jesus, A. Philippou, M.W. Anderson, Synthesis and characterization of microporous titanoniobosilicate ETNbS-10, Chem. Commun. (1999) 471–472. [25] Y. Goa, H. Yoshitake, P. Wu, T. Tatsumi, Controlled detitanation of ETS-10 materials through the post-synthetic treatment and their applications to the liquid-phase epoxidation of alkenes, Microporous Mesoporous Mater. 70 (2004) 93–101. [26] C.C. Pavel, K. Popa, N. Bilba, A. Cecal, D. Cozma, A. Pui, The sorption of some radiocations on microporous titanosilicate ETS-10, J. Radioanal. Nucl. Chem. 258 (2003) 243–248. [27] S.M. Kuznicki, A. Ansón, T. Segin, C.C.H. Lin, Modiﬁed ETS-10 zeolites for oleﬁn separation, U.S. Patent 20090187053 (2009). [28] I. Tiscornia, S. Irusta, P. Pradanos, C. Tellez, J. Coronas, J. Santamaria, Preparation and characterization of titanosilicate Ag-ETS-10 for propylene and propane adsorption, J. Phys. Chem. C 111 (2007) 4702–4709. [29] C.M. Braunbarth, L.C. Boudreau, M. Tsapatis, Synthesis of ETS-4/TiO2 composite membranes and their pervaporation performance, J. Membr. Sci. 174 (2000) 31–42. [30] G.Q. Guan, K. Kusakabe, S. Morooka, Synthesis and permeation properties of ion-exchanged ETS-4 tubular membranes, Microporous Mesoporous Mater. 50 (2001) 109–120. [31] Z. Lin, J. Rocha, A. Navajas, C. Tellez, J. Coronas, J. Santamaria, Synthesis and characterization of titanosilicate ETS-10 membranes, Microporous Mesoporous Mater. 67 (2004) 79–86. [32] A. Nalaparaju, Z.Q. Hu, X.S. Zhao, J.W. Jiang, Exchange of heavy metal ions in titanosilicate Na-ETS-10 membrane from molecular dynamics simulations, J. Membr. Sci. 335 (2009) 89–95. [33] C. Casado, Z. Amghouz, J.R. Gracia, K. Boulahya, J.M. Gonsalez, C. Tellez, J. Coronas, Synthesis and characterization of microporous titanosilicate ETS-10 obtained with different Ti sources, Mater. Res. Bull. 44 (2009) 1225–1231. [34] X. Liu, J.K. Thomas, Synthesis of microporous titanosilicates ETS-10 and ETS-4 using solid TiO2 as the source of titanium, Chem. Commun. (1996) 1435–1436. [35] J. Rocha, A. Ferreira, Z. Lin, M.W. Anderson, Synthesis of microporous titanosilicate ETS-10 from TiCl3 and TiO2 : a comprehensive study, Microporous Mesoporous Mater. 23 (1998) 253–263. [36] C.C. Pavel, D. Vuono, L. Catanzaro, P.D. Luca, N. Bilba, A. Nastro, J.B. Nagy, Synthesis and characterization of the microporous titanosilicates ETS-4 and ETS-10, Microporous Mesoporous Mater. 56 (2002) 227–239. [37] Y. Ng, C. Jei, M. Shamsuddin, Titanosilicate ETS-10 derived from rice husk ash, Microporous Mesoporous Mater. 122 (2009) 195–200. [38] T. Armaroli, G. Busca, F. Milella, F. Bregani, G.P. Toledo, A. Nastro, P. De Luca, G. Bagnasco, M. Turco, a study of ETS-4 molecular sieves and their adsorption of water and ammonia, J. Mater. Chem. 10 (2000) 1699–1705. [39] K.P. Prasanth, H.C. Bajaj, H.D. Chung, K.Y. Choo, T.H. Kim, R.V. Jasra, Hydrogen sorption in transition metal modiﬁed, J. Alloys Compd. 480 (2009) 580–586. [40] J.A. Delgado, M.A. Uguina, V.I. Agueda, A. Garcia-Sanz, Adsorption and diffusion parameters of methane and nitrogen on microwave-synthesized ETS-4, Langmuir 24 (2008) 6107–6115.

Journal of Hazardous Materials 195 (2011) 346–354

Contents lists available at SciVerse ScienceDirect

Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat

Rose-like monodisperse bismuth subcarbonate hierarchical hollow microspheres: One-pot template-free fabrication and excellent visible light photocatalytic activity and photochemical stability for NO removal in indoor air Fan Dong a,b,c , S.C. Lee b,∗ , Zhongbiao Wu c , Yu Huang b , Min Fu a , Wing-Kei Ho d , Shichun Zou e , Bo Wang b a

College of Environmental and Biological Engineering, Chongqing Technology and Business University, Chongqing, 400067, PR China Department of Civil and Structural Engineering, Research Center for Environmental Technology and Management, The Hong Kong Polytechnic University, Hong Kong, PR China c Department of Environmental Engineering, Zhejiang University, Hangzhou 310027, PR China d Nano and Advanced Materials Institute Limited, Hosted by The Hong Kong University of Science and Technology, Hong Kong, PR China e School of Marine Sciences, Sun Yat-Sen University, Guangzhou, 510275, PR China b

a r t i c l e

i n f o

Article history: Received 29 May 2011 Received in revised form 5 August 2011 Accepted 16 August 2011 Available online 22 August 2011 Keywords: (BiO)2 CO3 Hollow microsphere Visible light Photocatalytic Indoor air

a b s t r a c t Rose-like monodisperse hierarchical (BiO)2 CO3 hollow microspheres are fabricated by a one-pot template-free method for the ﬁrst time based on hydrothermal treatment of ammonia bismuth citrate and urea in water. The microstructure and band structure of the as-prepared (BiO)2 CO3 superstructure are characterized in detail by X-ray diffraction, Raman spectroscopy, Fourier transform-infrared spectroscopy, transmission electron microscopy, scanning electron microscopy, N2 adsorption–desorption isotherms, X-ray photoelectron spectroscopy and UV–vis diffuse reﬂectance spectroscopy. The monodisperse hierarchical (BiO)2 CO3 microspheres are constructed by the self-assembly of single-crystalline nanosheets. The aggregation of nanosheets result in the formation of three dimensional hierarchical framework containing mesopores and macropores, which is favorable for efﬁcient transport of reaction molecules and harvesting of photo-energy. The result reveals the existence of special two-band-gap structure (3.25 and 2.0 eV) for (BiO)2 CO3 . The band gap of 3.25 eV is intrinsic and the formation of smaller band gap of 2.0 eV can be ascribed to the in situ doped nitrogen in lattice. The performance of hierarchical (BiO)2 CO3 microspheres as efﬁcient photocatalyst are further demonstrated in the removal of NO in indoor air under both visible light and UV irradiation. It is found that the hierarchical (BiO)2 CO3 microspheres not only exhibit excellent photocatalytic activity but also high photochemical stability during long term photocatalytic reaction. The special microstructure, the high charge separation efﬁciency due to the inductive effect, and two-band-gap structure in all contribute to the outstanding photocatalytic activities. The discovery of monodisperse hierarchical nitrogen doped (BiO)2 CO3 hollow structure is signiﬁcant because of its potential applications in environmental pollution control, solar energy conversion, catalysis and other related areas. © 2011 Elsevier B.V. All rights reserved.

1. Introduction Many materials in the nature, such as lotus leaf and seashell, consisting of ordinary composition, exhibit fascinating properties owing to their special structural characteristics [1,2]. Such intricate natural designs have inspired materials scientists to fabricate morphology and structure controlled materials, with expectations to obtain novel or enhanced properties [2–4]. Hierarchical hollow structured materials have been a subject of intensive research in the past decade because of their novel physicochemical properties, which differ markedly from those

∗ Corresponding author. Tel.: +852 27666011; fax: +852 27666011. E-mail address: [email protected] (S.C. Lee). 0304-3894/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.jhazmat.2011.08.050

of bulk materials, and their potential applications as nanoscale chemical reactors, (photo)catalysts, sensing, lithium batteries, solar energy conversion, photonic building blocks and environmental applications, to name a few [5–10]. The fabrication of such structures usually relies on templating approaches, in which hard templates [11–14] (e.g., monodisperse polymer latex, carbon, silica spheres, and reducing metal nanoparticles) or soft sacriﬁcial templates [15–19] (e.g., micelles, microemulsions, macromolecules, oil droplets, and gas bubbles) were used to direct the growth of hierarchical or hollow structure. The template approach can be easily conducted for a speciﬁc structure. The capability of constructing complicated structure, however, is usually limited by the availability of templates. Disadvantages related to high cost and tedious synthetic procedures have also impeded scale-up of these template methods for

F. Dong et al. / Journal of Hazardous Materials 195 (2011) 346–354

applications. In comparison with these methods, which involve multistep procedures, a one-pot template-free approach for the controlled synthesis of hierarchical hollow sphere is highly attractive and desirable based on different mechanisms. Recently, a number of one-pot template-free methods for generating hollow inorganic micro- and nanostructures have been developed by using some well-known physical phenomena, for example, Ostwaldripening [20], Kirkendall-effect [21] and oriented attachment [22]. Recently, there has been great interest in developing semiconductor photocatalysts with high activities (especially the visible light driven photocatalysts) for both energetic and environmental applications, such as photocatalytic hydrogen evolution, creation of self-cleaning surfaces, disinfection of water, degradation of organic contaminants and conversion of carbon dioxide into hydrocarbon fuels [23–32]. The key to the application of photocatalysis technology is to develop photocatalytic materials with efﬁcient activity and high stability. Photocatalytic materials with hollow or hierarchical structure, such as TiO2 [33,34], ZnS [35], WO3 [36], BiVO4 [37], Bi2 WO6 [38] and Cu2 PO4 OH [39] have been fabricated by different methods, and are proved to be high-performance in environmental pollutants degradation due to their special morphological structure with respect to pore structure, light harvesting, and charge separation and so on. Development of new visible light driven photocatalytic materials may also overcome the limitations of transitional TiO2 (reduced sensitivity to sunlight and limited visible light photocatalytic activity) [25]. Controlling the shape and morphology of new photocatalytic materials may, therefore provide new opportunities to explore their novel structural properties and photocatalytic activity. In spite of these advances, realization of controlled one-pot fabrication of novel hierarchical hollow structure by facile template-free method remains a great challenge. The bismuth subcarbonate (BiO)2 CO3 is ﬁrst reported in 1984 [40]. It has an orthorhombic crystal structure with cell param˚ belonging to Imm2 eters of a = 3.865, b = 3.865, and c = 13.675 A, space group. After the ﬁrst report in 1984, there is little investigation on the fabrication and properties of (BiO)2 CO3 . Until very recently, synthetic (BiO)2 CO3 is reported to display promising results in antibacterial and environmental applications [41–43]. Although some advances has been made, much is unknown on the controlled fabrication, morphological structure, especially the visible light photocatalytic properties of uniform hierarchical (BiO)2 CO3 hollow microspheres. It will be, therefore of fundamental and technological interest to develop facile and effective methods for the fabrication of such novel structured (BiO)2 CO3 with novel properties for energetic and environmental applications [28]. In this study, we developed a one-pot template-free method for the fabrication of uniform monodisperse hierarchical (BiO)2 CO3 hollow microspheres for the ﬁrst time based on hydrothermal reaction between ammonia bismuth citrate and urea. The asprepared hierarchical (BiO)2 CO3 microspheres were analyzed by various characterization tools to fully understand the structural properties and were used as visible light photocatalyst for indoor air NO removal. It was found that the attractive hierarchical (BiO)2 CO3 microspheres were constructed by self-assembly of single-crystalline nanosheets. Interestingly, nitrogen was in situ doped into the lattice of (BiO)2 CO3 , which modiﬁed the band structure signiﬁcantly. For the ﬁrst time, the monodisperse hierarchical (BiO)2 CO3 hollow microspheres were found to exhibit excellent photocatalytic activity and photochemical stability for indoor air NO removal under both visible and UV light irradiation. The attractive hierarchical (BiO)2 CO3 hollow microspheres will also ﬁnd wide application in other areas, such as solar energy conversion, aqueous pollution control and production of fuels by fully using solar energy.

347

2. Experimental 2.1. Fabrication All chemicals used in this study were analytical grade (Sigma–Aldrich) and were used without further puriﬁcation. Distilled water was used in all experiments. In a typical synthesis, appropriate amounts of ammonia bismuth citrate (1.66 g) and urea (0.72 g) were mixed with 75 ml of H2 O in a 100 ml autoclave Teﬂon vessel and stirred for 30 min. The resulted transparent precursor solution was then hydrothermally treated at 180 ◦ C for 12 h. The sample obtained was ﬁltered, washed with water and ethanol for four times and dried at 60 ◦ C to get ﬁnal (BiO)2 CO3 with no further treatment. For comparison, C-doped TiO2 were prepared by reported hydrothermal method [29] and the commercial Degussa P25 was also used as reference sample. 2.2. Characterization The crystal phases of the sample were analyzed by X-ray diffraction with Cu K␣ radiation (XRD: model D/max RA, Rigaku Co., Japan). The accelerating voltage and the applied current were 40 kV and 150 mA, respectively. Raman spectra were recorded at room temperature using a micro-Raman spectrometer (Raman: RAMANLOG 6, USA) with a 514.5 nm Ar + laser as the excitation source in a backscattering geometry. The incident laser power on the samples was less than 10 mW. X-ray photoelectron spectroscopy with Al K␣ X-rays (h = 1486.6 eV) radiation operated at 150 W (XPS: Thermo ESCALAB 250, USA) was used to investigate the surface properties and to probe the total density of the state (DOS) distribution in the valence band (VB). The shift of the binding energy due to relative surface charging was corrected using the C1s level at 284.8 eV as an internal standard. FT-IR spectra were recorded on a Nicolet Nexus spectrometer on samples embedded in KBr pellets. A scanning electron microscope (SEM, JEOL model JSM-6490, Japan) was used to characterize the morphology of the obtained products. The morphology, structure and grain size of the samples were examined by transmission electron microscopy (TEM: JEM-2010, Japan). The UV–vis diffuse reﬂection spectra were obtained for the dry-pressed disk samples using a Scan UV–vis spectrophotometer (UV–vis DRS: TU-1901, China) equipped with an integrating sphere assembly, using BaSO4 as reﬂectance sample. The spectra were recorded at room temperature in air ranged from 250 to 800 nm. Nitrogen adsorption–desorption isotherms were obtained on a nitrogen adsorption apparatus (ASAP 2020, USA). All the samples were degassed at 200 ◦ C prior to measurements. 2.3. Photocatalytic activity evaluation The photocatalytic activity of the resulting samples was investigated by oxidation of NO at ppb levels in a continuous ﬂow reactor at ambient temperature. The volume of the rectangular reactor, which was made of stainless steel and covered with Saint-Glass, was 4.5 L (30 cm × 15 cm × 10 cm). A 300 W commercial tungsten halogen lamp (General Electric) was vertically placed outside the reactor above the reactor. Four mini-fans were ﬁxed around the lamp to avoid the temperature rise of the ﬂow system. Adequate distance was also kept from the lamp to the reactor for the same purpose to keep the temperature at a constant level. The distance between the lamp and the sample is 30 cm. For the visible light photocatalytic activity test, UV cutoff ﬁlter (420 nm) was adopted to remove UV light in the light beam. For photocatalytic activity test under simulated solar light, the UV cutoff ﬁlter was removed. For UV light photocatalytic activity test, two 6 W UV lamps (Cole–Parmler), emitting a primary wavelength at 365 nm was used. For each photocatalytic activity test experiment, one

348

F. Dong et al. / Journal of Hazardous Materials 195 (2011) 346–354

sample dish (with a diameter of 12 cm) containing the photocatalyst powders was placed in the center of the reactor. The photocatalyst samples were prepared by coating an aqueous suspension of the samples onto the glass dish. The weight of the photocatalysts used for each experiment was kept at 0.15 g. The dishes containing the photocatalyst were pretreated at 70 ◦ C for several hours until complete removal of water in the suspension and then cooled to room temperature before the photocatalytic test. The NO gas was acquired from a compressed gas cylinder at a concentration of 100 ppm of NO (N2 balance, BOC gas) with traceable National Institute of Standards and Technology (NIST) standard. The initial concentration of NO was diluted to about 400 ppb by the air stream supplied by a zero air generator (Thermo Environmental Inc., model 111). The desired relative humidity (RH) level of the NO ﬂow was controlled at 70% by passing the zero air streams through a humidiﬁcation chamber. The gas streams were premixed completely by a gas blender, and the ﬂow rate was controlled at 3.3 L/min by a mass ﬂow controller. After the adsorption–desorption equilibrium among water vapor, gases, and photocatalysts was achieved, the lamp was turned on. The concentration of NO was continuously measured by a chemiluminescence NO analyzer (Thermo Environmental Instruments Inc., model 42c), which monitors NO, NO2 , and NOx (NOx represents NO + NO2 ) with a sampling rate of 0.7 L/min. The removal efﬁciency (␩) of NO was calculated as (%) = (1 – C/C0 ) × 100%, where C and C0 are concentrations of NO in the outlet steam and the feeding stream, respectively. 3. Results and discussion 3.1. Crystal phase Fig. 1 shows the XRD pattern of the as-prepared hierarchical (BiO)2 CO3 hollow microspheres compared with standard PDF card (JCPDS-ICDD Card No. 41-1488). All the diffraction peaks can be indexed to (BiO)2 CO3 . No peaks from other phases have been detected, indicating the phase purity of the product. The crystal structure of (BiO)2 CO3 is shown in Fig. S1 (supporting information). The (Bi2 O2 )2+ layers and CO3 2− layers are inter-grown with the plane of the CO3 2− group orthogonal to the plane of the (Bi2 O2 )2+ layer. The large cation Bi3+ with 8-coordination shows stereo active lone-pair behaviors that results in the Bi–O polyhedron. The internal layered structure would guide the lower growth rate along certain axis to form nanosheet morphologies [44]. 3.2. Vibrational spectra The vibrational spectra of FT-IR and Raman obtained are shown in Fig. 2. The “free” carbonate ion (point group symmetry D3h )

Fig. 1. XRD pattern of the as-prepared hierarchical (BiO)2 CO3 microspheres.

possessed four internal vibrations: symmetric stretching mode 1 , the corresponding antisymmetric vibration 3 , the out-of-plane bending mode 2 and the in-plane deformation 4 . Four characteristic bands in FT-IR at 1 (1067 cm−1 ), 2 (846 and 820 cm−1 ), 3 (1468 and 1391 cm−1 ), 4 (670 cm−1 ), 1 + 4 (1756 and 1730 cm−1 ) are observed (Fig. 2a). The Raman bands at 1070, 1407 and 690 cm−1 are attributed to 1 , 3 and 4 of the carbonate ion CO3 2− vibration, respectively (Fig. 2b). The 2 mode is absent in the Raman spectra, indicating 2 is only IR active. The Raman band at 512 cm−1 can be assigned to the Bi O stretching. The low bands at 417 and 162 cm−1 in Raman are due to the external vibration, which were also observed in other reports [45]. 3.3. Morphological structure The hydrothermal treatment of ammonia bismuth citrate and urea mixture produces a population of hierarchical microspheres with an interesting rose-like shape, each containing a concavity on its center, as shown in the typical SEM images in Fig. 3. The as-prepared (BiO)2 CO3 microspheres have an average diameter of 1.0 ␮m and thickness of 0.5 ␮m. No other morphologies can be observed, indicating a high yield of such (BiO)2 CO3 superstructure. All the microspheres in Fig. 3a and b show almost the same size and are separated to each other, indicating the produced microspheres are highly uniform and monodisperse. The higher magniﬁcation SEM image shown in Fig. 3c reveals that each microsphere is

Fig. 2. FT-IR (a) and Raman (b) spectra of the as-prepared hierarchical (BiO)2 CO3 microspheres.

F. Dong et al. / Journal of Hazardous Materials 195 (2011) 346–354

349

Fig. 3. SEM (a–c), TEM (d, e) the as-prepared hierarchical (BiO)2 CO3 microspheres, SAED (f) of a single nanosheet.

composed of many layers of nanoscale sheet-like structures. These nanosheets are arranged at progressively increasing angles to the radial axis and are highly directed to form arrays in a hierarchical fashion. The side view of an individual rose-like structure (Fig. 3c) supports the conclusion that such a microsphere is composed of regularly packed nanosheets with an average thickness of about 20 nm. The morphology and size of the product was further investigated by TEM. As shown in Fig. 3d, the entire microsphere is composed of self-assembled nanosheets. The nanosheets are very thin and therefore relatively transparent to the electron beam. The microsphere is hollow in the center. An HRTEM image of a single nanosheet in the microsphere is shown in Fig. 3e. The lattice spacing is measured

to be 0.271 nm, matching with the spacing of the (1 1 0) crystal plane of Bi2 O2 CO3 . The SAED pattern in Fig. 3f displays an array of clear and regular diffraction spots of one single nanosheet, indicating that the nanosheet is well-deﬁned single-crystalline in nature. Thus, the hierarchical Bi2 O2 CO3 microspheres originates from the self-assembly of the single-crystalline nanosheets with preferred orientation. In this fabrication approach, (BiO)2 CO3 with particle morphology would be produced when urea was replaced by the same molar amount of sodium carbonate (see Fig. S2 and S3 in the supporting information). That is, the addition of urea plays a crucial role in forming hierarchical microsphere structure. It is known that urea can be decomposed in aqueous solution at an elevated temperature.

350

F. Dong et al. / Journal of Hazardous Materials 195 (2011) 346–354

also well suited for harvesting photo-energy and introducing reactive molecules into the interior space of hierarchical (BiO)2 CO3 microspheres [48]. 3.5. XPS analysis

Fig. 4. N2 adsorption–desorption isotherms and pore-size distribution (inset) of hierarchical (BiO)2 CO3 microspheres.

The following reaction may take place in the hydrothermal process. CO(NH2 )2 + H2 O → CO2 + 2NH3 +

NH3 + H2 O → NH4 + OH −

CO2 + 2OH → CO3

2−

−

+ H2 O

Bi3+ (citrate) + H2 O → BiO+ + 2H+ +

2BiO + CO3

2−

→ (BiO)2 CO3 (s)

(1) (2) (3) (4) (5)

It is worthwhile and interesting to elucidate the growth mechanism of hierarchical (BiO)2 CO3 hollow microspheres. As the growth mechanism is complicated, the precise understanding is in currently progress.

The XPS measurements were carried out to determine the chemical state of the elements and total density of states distribution (DOS) of the valence band in as-prepared (BiO)2 CO3 , as shown in Fig. 5. Two strong peaks at 159.2 and 164.5 eV in the high-resolution spectra (Fig. 5a) are assigned to Bi4 f7/2 and Bi4 f5/2 , respectively, which is characteristic of Bi3+ in (BiO)2 CO3 [49]. Fig. 5b shows the C1s spectra of (BiO)2 CO3 . A broad energy range from 291 to 282 eV can be observed. The peaks at 284.8, 286.2 and 288.0 eV can be assigned to adventitious carbon species from XPS measurement [29], while the peak at 289.0 eV can be ascribed to carbonate ion in (BiO)2 CO3 . Meanwhile, the O1s spectra are also recorded (Fig. 5c), which can be ﬁtted by three peaks at binding energies of 530.3, 531.3 and 532.4 eV, respectively. The peak at 530.3 eV is characteristic of Bi–O binding energy in (BiO)2 CO3 [49], and the other two peaks at around 531.3 and 532.4 eV can be assigned to carbonate species and adsorbed H2 O on the surface. Nitrogen is detected by XPS with N1s binding energy centered at 400 eV as shown in Fig. 5d. This fact indicates that nitrogen was in situ doped into the (BiO)2 CO3 during hydrothermal treatment. The N1s peak at 400 eV was also observed in the widely investigated nitrogen doped TiO2 , where nitrogen substituted for oxygen in TiO2 [50–52]. In our case, nitrogen may also be doped into the lattice of (BiO)2 CO3 and substitute for oxygen atom. The DOS of VB is shown in Fig. 5e. The valence band maximum (VBM) is determined to be 1.7 eV. Interestingly, additional diffusive electronic states above the VBM are observed above the valence band edge, indicating the existence of mid-gap above the valence band [53,54]. Such mid-gap can be ascribed to the nitrogen doping, similar to that of nitrogen doped TiO2 [53,54]. The newly formed mid-gap between valence band and conduction band could make (BiO)2 CO3 microspheres absorb visible light. 3.6. UV–vis DRS and band structure

3.4. BET surface areas and pore structure Fig. 4 shows nitrogen adsorption–desorption isotherms and the corresponding pore-size distribution (PSD) curves (inset in Fig. 4) for the as-prepared hierarchical (BiO)2 CO3 microspheres. Nitrogen adsorption–desorption isotherm show hysteresis loops at relative pressures close to unity, indicating the presence of large mesopores (about 23.0 nm), which categorize it as type IV according to IUPAC classiﬁcation [46]. The shape of adsorption branch at relative pressures is close to unity resemblances somewhat type II, indicating the presence of macropores [46]. The shapes of hysteresis loops are of type H3, associated with mesopores present in hierarchical nanosheet particles, giving rise to slit-like pores, which is consistent with the TEM results (Fig. 3). The isotherm also shows high adsorption at relative pressures (P/P0 ) approaching 1.0, suggesting the formation of large mesopores and macropores [46]. In fact, the mesopores and macropores are formed due to aggregation of nanosheets, because the single-crystal nanosheets do not contain mesopores and macropores. As can be seen from this inset the PSD curve is quite broad (from 2 to 100 nm) and bimodal with small mesopores (∼3.6 nm) and larger ones (∼23.0 nm). The smaller mesopores reﬂect porosity within nanosheets, while larger mesopores can be related to the pores formed between stacked nanosheets. This result further conﬁrmed the existence of mesopores and macropores. Such organized porous structures might be extremely useful in photocatalysis because they possess efﬁcient transport pathways to reactant and product molecules [33,47]. The unique three dimensional macroporous hierarchical framework is

The as-prepared hierarchical (BiO)2 CO3 microspheres show a light yellow color, suggesting their ability to absorb light in the visible region. The UV–vis DRS of the materials in Fig. 6a shows two band-edge absorptions, one at UV region and another at visible light region. The (BiO)2 CO3 can, thus absorb both UV and visible light. The band gap energy can be estimated from the intercept of the tangents to the plots of (˛h)1/2 vs. photon energy, as shown in Fig. 6b. Very interesting, the as-prepared (BiO)2 CO3 shows two band gaps of 3.25 and 2.0 eV, consistent with additional diffusive electronic states observed with VB-XPS [25]. Khan et al. also observed the two-band-gap structure (3.0 and 2.32 eV) for carbon doped TiO2 , where the band gap of 3.0 eV was intrinsic for TiO2 and the smaller band gap of 2.0 eV caused by the doped carbon in lattice [55]. In our case, the band gap of 3.25 eV is intrinsic for (BiO)2 CO3 , similar to report by Xie and Huang [42,43]. The smaller band gap of 2.0 eV can be due to the in situ doped nitrogen in lattice forming mid-gap. The as-prepared (BiO)2 CO3 is, therefore sensitive to both UV and visible light. It is noteworthy that CO3 2− ions having a large negative charge, maintains a large dipole in the (BiO)2 CO3 , which prefers the photogenerated charge separation [56]. This effect is called an inductive effect, which can be described as the action of one group to affect the electron distribution in another group though electrostatic force [56]. The inductive effect prefers to attract holes and repel electrons, thus enhancing the photogenerated charge separation. It is thus expected that the hierarchical (BiO)2 CO3 microspheres may exhibit high photocatalytic activity under both visible and UV light

F. Dong et al. / Journal of Hazardous Materials 195 (2011) 346–354

351

Fig. 5. XPS spectra of the as-prepared hierarchical (BiO)2 CO3 , (a) Bi4 f, (b) C1s, (c) O1s, (d) N1s and (e) VB.

irradiation, fully using solar light for degradation of environmental pollutants. 3.7. Photocatalytic activity and photochemical stability 3.7.1. Photocatalytic activities under visible and UV light irradiation The as-prepared hierarchical (BiO)2 CO3 microspheres were used to photocatalytic removal of NO in gas phase in order

to demonstrate their potential ability for indoor air puriﬁcation (Fig. 7). Fig. 7a shows the variation of NO concentration (C/C0 %) with irradiation time over the hierarchical (BiO)2 CO3 microspheres under visible light irradiation. Here, C0 is the initial concentration of NO, and C is the concentration of NO after photocatalytic reaction for t. As a comparison, photocatalytic oxidation of NO over C-doped TiO2 and (BiO)2 CO3 particles are also performed under identical conditions. As previously proved, NO could not be photolyzed under light irradiation ( > 420 nm) [57]. As shown in Fig. 7a, after

352

F. Dong et al. / Journal of Hazardous Materials 195 (2011) 346–354

Fig. 6. UV–vis DRS (a) and plots of (˛h)1/2 vs. photon energy (b) of the as-prepared hierarchical (BiO)2 CO3 microspheres.

Fig. 7. Photocatalytic activities of hierarchical (BiO)2 CO3 microspheres, (BiO)2 CO3 particles, C-doped TiO2 and TiO2 -P25 under visible (a) and UV (b) light irradiation for NO removal in indoor air.

60 min irradiation, the photodegradation rate of NO over (BiO)2 CO3 particles is merely 8.0%. For C-doped TiO2 , the NO removal rate reaches at 20.5% due to its well known good visible light activity. Interestingly, the as-prepared hierarchical (BiO)2 CO3 microspheres exhibits higher photocatalytic activity (42.0% of NO removal rates) under visible light irradiation than that of C-doped TiO2 , although

Fig. 8. Long term photocatalytic activity of hierarchical (BiO)2 CO3 microspheres.

the BET surface area and the pore volume are much smaller than that of C-doped TiO2 (Table 1). The higher photocatalytic activity of hierarchical (BiO)2 CO3 microspheres could mainly be attributed to the hierarchical structure. The hierarchical structure was favorable for the diffusion of reaction intermediates and the efﬁcient utilization of photo-energy, thus enhancing the photocatalytic activity [2,33,48,58].

Fig. 9. SEM image of hierarchical (BiO)2 CO3 microspheres after long time irradiation.

F. Dong et al. / Journal of Hazardous Materials 195 (2011) 346–354

353

Fig. 10. XRD pattern (a) and FTIR spectra (b) of hierarchical (BiO)2 CO3 microspheres after long time irradiation.

There are two band gaps of 2.0 and 3.25 eV for the hierarchical (BiO)2 CO3 microspheres. The excellent visible light photocatalytic activity corresponds to the suitable band gap energy of 2.0 eV. For the band gap of 3.25 eV, we test the photocatalytic activity under UV irradiation, as shown in Fig. 7b. P25 is a famous excellent UV light photocatalyst, which naturally shows decent activity (removal rate of 35.1%) under UV irradiation. Interestingly, the hierarchical (BiO)2 CO3 microspheres exhibit even higher activity (removal rate of 43.5%) than that of P25 and (BiO)2 CO3 particles (removal rate of 42.5%) and under UV irradiation. Although the BET surface areas of hierarchical (BiO)2 CO3 microspheres is much smaller than that of P25, the pore volume is larger than that of P25 (see Table 1). Large pore volume is favorable for the adsorption of reactant and diffusion of reaction intermediates. Compared to (BiO)2 CO3 particles, the hierarchical structure of (BiO)2 CO3 microspheres contributed to the enhanced photocatalytic activity under UV irradiation [2,33,48,58]. 3.7.2. Long term photocatalytic activity and photochemical stability The stability of a photocatalyst under irradiation should be considered for its practical application [57]. It would be most desirable if the photocatalyst can maintain durable activity so that the catalyst can be used for a long time. Some photocatalysts often suffer from instability for air puriﬁcation applications, because the intermediates generated during photocatalytic reaction can accumulate on the catalyst surface, which possibly deactivates the photocatalyst. Fig. 8 shows long term photocatalytic activity of hierarchical (BiO)2 CO3 microspheres under simulated solar irradiation. It can be seen that the photocatalytic activity is very stable during the photocatalytic oxidation of NO without deactivation. SEM image of hierarchical (BiO)2 CO3 microspheres after long time irradiation is shown in Fig. 9. The morphology is almost the same as the SEM in Fig. 3b, indicating the microstructure doesn’t change after long irradiation. The stability of the hierarchical (BiO)2 CO3 microspheres is further conﬁrmed by XRD pattern and FI-TR spectra of hierarchical (BiO)2 CO3 microspheres after long time irradiation, as shown in Table 1 BET speciﬁc surface areas and pore parameters of as-prepared (BiO)2 CO3 , C-doped TiO2 and P25. Samples

SBET (m2 /g)

Total volume (cm3 /g)

Peak diameter (nm)

(BiO)2 CO3 C-doped TiO2 P25

16.5 122.5 51.2

0.114 0.248 0.090

3.6/23.0 3.5 22.0

Fig. 10. The XRD pattern in Fig. 10a of the used (BiO)2 CO3 microspheres shows that the crystal structure was not changed after the photocatalytic reaction, suggesting its phase stability. It can be seen from Fig. 10b that the reaction intermediates and products during photocatalytic oxidation of NO (such as HNO2 and HNO3 ) cannot be observed except for the vibration mode of (BiO)2 CO3 , which suggest that the reaction intermediates and products could diffuse rapidly due to the hierarchical structure. This is another reason for the excellent photochemical stability of hierarchical (BiO)2 CO3 microspheres. As the solar light consists of both UV and visible light, our hierarchical (BiO)2 CO3 hollow microspheres could fully use the solar light to remove the environmental pollutants by photocatalysis. 4. Conclusions Uniform monodisperse hierarchical (BiO)2 CO3 hollow microspheres were successfully fabricated by one-pot template-free method. The hierarchical (BiO)2 CO3 superstructure was constructed by the self-assembly of single-crystalline nanosheets. The aggregation of nanosheets produced three dimensional hierarchical structures containing mesopores and macropores, which was favorable for efﬁcient transport of reaction molecules and harvesting of photo-energy. The results conﬁrmed the existence of two-band-gap structure of 3.25 and 2.0 eV for (BiO)2 CO3 . In situ doped nitrogen contributed to the band gap of 2.0 eV by formation of mid-gap energy through substitution for oxygen atom in lattice. Owing to the attractive hierarchical hollow microstructure, and the high charge separation efﬁciency and special two-band-gap structure, the as-prepared (BiO)2 CO3 microspheres not only exhibited excellent photocatalytic activity under both visible and UV light irradiation but also excellent photochemical stability during long term photocatalytic reaction for removal of NO in indoor air. The monodisperse hollow microspheres with hierarchical structure prepared by a facile method should ﬁnd wide applications in environmental pollution control, solar energy conversion, catalysis and other related areas. Acknowledgements This research is ﬁnancially supported by the National Natural Science Foundation of China (51108487), National High Technology Research and Development Program (863 Program) of China (2010AA064905), Research Grants of Chongqing Technology and Business University (2010-56-13), the Research Grants Council of Hong Kong (PolyU 5204/07E and PolyU 5175/09E), the Hong Kong

354

F. Dong et al. / Journal of Hazardous Materials 195 (2011) 346–354

Polytechnic University (GU712, GYX75 and GYX0L), the Program for Chongqing Innovative Research Team Development in University (KJTD201020) and the Chongqing Key Natural Science Foundation (CSTC, 2008BA4012). Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.jhazmat.2011.08.050. References [1] T. Douglas, A bright bio-inspired future, Science 299 (2003) 1192–1193. [2] H.X. Li, Z.F. Bian, J. Zhu, D.Q. Zhang, G.S. Li, Y.N. Huo, H. Li, Y.F. Lu, Mesoporous titania spheres with tunable chamber stucture and enhanced photocatalytic activity, J. Am. Chem. Soc. 129 (2007) 8406–8407. [3] Y.F. Lu, H.Y. Fan, A. Stump, T.L. Ward, T. Rieker, J.C. Brinker, Aerosol-assisted self-assembly of mesostructured spherical nanoparticles, Nature 398 (1999) 223–226. [4] J.N. Cha, G.D. Stucky, D.E. Morse, T.J. Deming, Biomimetic synthesis of ordered silica structures mediated by block copolypeptides, Nature 403 (2000) 289–292. [5] X.W. Lou, L.A. Archer, Z.C. Yang, Hollow micro-/nanostructures: synthesis and applications, Adv. Mater. 20 (2008) 3987–4019. [6] Y. Zhao, L. Jiang, Hollow micro/nanomaterials with multilevel interior structures, Adv. Mater. 21 (2009) 3621–3638. [7] L.P. Zhu, H.M. Xiao, W.D. Zhang, G. Yang, S.Y. Fu, One-Pot template-free synthesis of monodisperse and single-crystal magnetite hollow spheres by a simple solvothermal route, Cryst. Growth Des. 8 (2008) 957–963. [8] Y.H. Zheng, Y. Cheng, Y.S. Wang, L.H. Zhou, F. Bao, C. Jia, Metastable ␥-MnS hierarchical architectures: synthesis, characterization, and growth mechanism, J. Phys. Chem. B 110 (2006) 8284–8288. [9] J.G. Yu, H.G. Yu, H.T. Guo, M. Li, S. Mann, Spontaneous formation of a tungsten trioxide sphere-in-shell superstructure by chemically induced selftransformation, Small 4 (2008) 87–91. [10] L.S. Zhang, W.Z. Wang, L. Zhou, H.L. Xu, Bi2 WO6 nano- and microstructures: shape control and associated visible-light-driven photocatalytic activities, Small 3 (2007) 1618–1625. [11] M. Yang, J. Ma, C.L. Zhang, Z.Z. Yang, Y.F. Lu, General synthetic route toward functional hollow spheres with double-shelled structures, Angew. Chem. Int. Ed. 44 (2005) 6727–6729. [12] J.G. Yu, X.X. Yu, Hydrothermal synthesis and photocatalytic activity of zinc oxide hollow spheres, Environ. Sci. Technol. 42 (2008) 4902–4907. [13] S.W. Kim, M. Kim, W.Y. Lee, T. Hyeon, Fabrication of hollow palladium spheres and their successful application to the recyclable heterogeneous catalyst for Suzuki coupling reactions, J. Am. Chem. Soc. 124 (2002) 7642–7643. [14] J.H. Gao, B. Zhang, X.X. Zhang, B. Xu, Magnetic dipolar interaction induced selfassembly affords wires of cobalt selenide hollow nanoscrystals, Angew. Chem. Int. Ed. 45 (2006) 1220–1223. [15] A.D. Dinsmore, M.F. Hsu, M.G. Nikolaides, M. Marquez, A.R. Bausch, D.A. Weitz, Colloidosomes: selectively permeable capsules composed of colloidal particles, Science 298 (2002) 1006–1009. [16] Y.S. Li, J.L. Shi, Z.L. Hua, H.R. Chen, M.L. Ruan, D.S. Yan, Hollow spheres of mesoporous aluminosilicate with a three-dimensional pore network and extraordinarily high hydrothermal stability, Nano Lett. 3 (2003) 609–612. [17] H.G. Yang, H.C. Zeng, Creation of intestine-like interior space for metal-oxide nanostructures with a quasi-reverse emulsion, Angew. Chem. Int. Ed. 43 (2004) 5206–5209. [18] D.H.M. Buchold, C. Feldmann, Nanoscale ␥-AlO(OH) hollow spheres: synthesis and container-type functionality, Nano Lett. 7 (2007) 3489–3492. [19] J. Zhou, W. Wu, D. Caruntu, M.H. Yu, A. Martin, J.F. Chen, C.J. O’Connor, W.L. Zhou, Synthesis of porous magnetic hollow silica nanospheres for nanomedicine application, J. Phys. Chem. C 111 (2007) 17473–17477. [20] H.C. Zeng, Curr, Ostwald ripening: a synthetic approach for hollow nanomaterials, Curr. Nanosci. 3 (2007) 177–181. [21] H.J. Fan, U. Gösele, M. Zacharias, Formation of nanotubes and hollow nanoparticles based on Kirkendall and diffusion processes: a review, Small 3 (2007) 1660–1671. [22] J. Zhang, F. Huang, Z. Lin, Progress of nanocrystalline growth kinetics based on oriented attachment, Nanoscale 2 (2010) 18–34. [23] X.B. Chen, S.H. Shen, L.J. Guo, S.S. Mao, Semiconductor-based photocatalytic hydrogen generation, Chem. Rev. 110 (2010) 6503–6570. [24] G. Zhang, W. Choi, S.H. Kim, S.B. Hong, Selective photocatalytic degradation of aquatic pollutants by titania encapsulated into FAU-type zeolites, J. Hazard. Mater. 188 (2011) 198–205. [25] F. Dong, W.R. Zhao, Z.B. Wu, Characterization and photocatalytic activities of C N and S co-doped TiO2 with 1D nanostructure prepared by nano-conﬁnement effect, Nanotechnology 19 (2008) 365607. [26] C.C. Chen, W.H. Ma, J.C. Zhao, Semiconductor-mediated photodegradation of pollutants under visible-light irradiation, Chem. Soc. Rev. 39 (2010) 4206–4219. [27] S. Rehman, R. Ullah, A.M. Butt, N.D. Gohar, Strategies of making TiO2 and ZnO visible light active, J. Hazard. Mater. 170 (2009) 560–569.

[28] P.L. Zhang, S. Yin, T. Sato, The inﬂuence of synthesis method on the properties of iron contained N doped TiO2 photocatalysts, Appl. Catal. B 103 (2011) 462–469. [29] F. Dong, H.Q. Wang, S. Guo, Z.B. Wu, S.C. Lee, Enhanced visible light photocatalytic activity of novel Pt/C-doped TiO2 /PtCl4 three-component nanojunction system for degradation of toluene in air, J. Hazard. Mater. 187 (2011) 509–516. [30] X.L. Hu, G.S. Li, J.C. Yu, Design, Fabrication, and modiﬁcation of nanostructured semiconductor materials for environmental and energy applications, Langmuir 26 (2010) 3031–3039. [31] J. Kim, C.W. Lee, W. Choi, Platinized WO3 as an environmental photocatalyst that generates OH radicals under visible light, Environ. Sci. Technol. 44 (2010) 6849–6854. [32] J.H. Mo, Y.P. Zhang, Q.J. Xu, Y.F. Zhu, J.J. Lamson, R.Y. Zhao, Determination and risk assessment of by-products resulting from photocatalytic oxidation of toluene, Appl. Catal. B 89 (2009) 570–576. [33] J.G. Yu, Y.R. Su, B. Cheng, Template-free fabrication and enhanced photocatalytic activity of hierarchically macro/mesoporous titania, Adv. Funct. Mater. 17 (2007) 1984–1990. [34] Y.J. Wang, R. Shi, J. Lin, Y.F. Zhu, Signiﬁcant photocatalytic enhancement in methylene blue degradation of TiO2 photocatalysts via graphene-like carbon in situ hybridization, Appl. Catal. B 100 (2010) 179–183. [35] X.X. Yu, J.G. Yu, B. Cheng, B.B. Huang, One-pot template-free synthesis of monodisperse zinc sulﬁde hollow spheres and their photocatalytic properties, Chem. Eur. J. 15 (2009) 6731–6739. [36] D. Chen, J.H. Ye, Hierarchical WO3 hollow shells: dendrite sphere, dumbbell, and their photocatalytic properties, Adv. Funct. Mater. 18 (2008) 1922–1928. [37] W.Z. Yin, W.Z. Wang, M. Shang, L. Zhou, S.M. Sun, L. Wang, BiVO4 hollow nanospheres: anchoring synthesis, growth mechanism, and their application in photocatalysis, Eur. J. Inorg. Chem. 437 (2009) 9–4384. [38] L.S. Zhang, W.Z. Wang, Z.G. Chen, L. Zhou, H.L. Xu, W. Zhu, Fabrication of ﬂower-like Bi2 WO6 superstructures as high performance visible-light driven photocatalysts, J. Mater. Chem. 17 (2007) 2526–2532. [39] I.S. Cho, D.W. Kim, S.W. Lee, C.H. Kwak, S.T. Bae, J.H. Noh, S.H. Yoon, H.S. Jung, D.W. Kim, K.S. Hon, Synthesis of Cu2 PO4 OH hierarchical superstructures with photocatalytic activity in visible light, Adv. Funct. Mater. 18 (2008) 2154–2162. [40] P. Aylor, S. Sundek, V. Lopata, Structure, spectra, and stability of solid bismuth carbonates, Can. J. Chem. 62 (1984) 2863–2873. [41] R. Chen, M.H. So, J. Yang, F. Deng, C.M. Che, H.Z. Sun, Fabrication of bismuth subcarbonate nanotube arrays from bismuth citrate, Chem. Commun. 226 (2006) 5–2267. [42] Y. Zheng, F. Duan, M.Q. Chen, Y. Xie, Synthetic Bi2 O2 CO3 nanostructures: novel photocatalyst with controlled special surface exposed, J. Mol. Catal. A: Chem. 317 (2010) 34–40. [43] H.F. Cheng, B.B. Huang, K.S. Yang, Z.Y. Wang, X.Y. Qin, X.Y. Zhang, Y. Dai, Facile template-free synthesis of Bi2 O2 CO3 hierarchical microﬂowers and their associated photocatalytic activity, ChemPhysChem 11 (2010) 2167–2173. [44] C.Z. Wu, Y. Xie, Controlling phase and morphology of inorganic nanostructures originated from the internal crystal structure, Chem. Commun. 594 (2009) 3–5957. [45] G.E. Tobon-Zapata, S.B. Etcheverry, E.J. Baran, Vibrational spectrum of bismuth subcarbonate, J. Mater. Sci. Lett. 16 (1997) 656–657. [46] K.S.W. Sing, D.H. Everett, R.A.W. Haul, L. Moscou, R.A. Pierotti, J. Rouquerol, T. Siemieniewska, Reporting physisorption data for gas/solid systems, Pure Appl. Chem. 57 (1985) 603–619. [47] X.C. Wang, J.C. Yu, C.M. Ho, Y.D. Hou, X.Z. Fu, Photocatalytic activity of a hierarchically macro/mesoporous titania, Langmuir 21 (2005) 2552–2559. [48] F. Dong, W.R. Zhao, Z.B. Wu, S. Guo, Band structure and visible light photocatalytic activity of multi-type nitrogen doped TiO2 nanoparticles prepared by thermal decomposition, J. Hazard. Mater. 162 (2008), 763-760. [49] S. Shamaila, A.K.L. Sajjad, F. Chen, J.L. Zhang, Study on highly visible light active Bi2 O3 loaded ordered mesoporous titania, Appl. Catal. B 94 (2010) 272–280. [50] X.B. Chen, C. Burda, Photoelectron spectroscopic investigation of nitrogendoped titania nanoparticles, J. Phys. Chem. B 108 (2004) 15446–15449. [51] Y. Cong, J.L. Zhang, F. Chen, M. Anpo, D.N. He, Preparation, photocatalytic activity, and mechanism of nano-TiO2 co-doped with nitrogen and iron (III), J. Phys. Chem. C 111 (2007) 10618–10623. [52] Z.B. Wu, F. Dong, W.R. Zhao, S. Guo, Visible light induced electron transfer process over nitrogen doped TiO2 nanocrystals prepared by oxidation of titanium nitride, J. Hazard. Mater. 157 (2008) 57–63. [53] X.B. Chen, C. Burda, The Electronic origin of the visible-light absorption properties of C-, N- and S-doped TiO2 nanomaterials, J. Am. Chem. Soc. 130 (2008) 5018–5019. [54] X.B. Chen, P.A. Glans, X. Qiu, S. Dayal, W.D. Jennings, K.E. Smith, C. Burda, J. Guo, X-ray spectroscopic study of the electronic structure of visible-light responsive N-, C- and S-doped TiO2 , J. Electron. Spectrosc. 162 (2008) 75–81. [55] S.U.M. Khan, M. Al-Shahry, W.B. Ingler Jr., Efﬁcient photochemical water splitting by a chemically modiﬁed n-TiO2 , Science 297 (2002) 2243–2245. [56] C.S. Pan, Y.F. Zhu, New type of BiPO4 oxy-acid salt photocatalyst with high photocatalytic activity on degradation of dye, Environ. Sci. Technol. 44 (2010) 5570–5574. [57] Z.H. Ai, W.K. Ho, S.C. Lee, L.Z. Zhang, Efﬁcient photocatalytic removal of NO in indoor air with hierarchical bismuth oxybromide nanoplate microspheres under visible light, Environ. Sci. Technol. 43 (2009) 4143–4150. [58] G.S. Li, D.Q. Zhang, J.C. Yu, M.K.H. Leung, An efﬁcient bismuth tungstate visiblelight-driven photocatalyst for breaking down nitric oxide, Environ. Sci. Technol. 44 (2010) 4276–4281.

Journal of Hazardous Materials 195 (2011) 355–364

Contents lists available at SciVerse ScienceDirect

Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat

Assessment of heavy metal contamination in sediments of the Tigris River (Turkey) using pollution indices and multivariate statistical techniques Memet Varol ∗ Ministry of Agriculture and Rural Affairs, Province Control Laboratory, 21010 Diyarbakır, Turkey

a r t i c l e

i n f o

Article history: Received 19 April 2011 Received in revised form 25 July 2011 Accepted 17 August 2011 Available online 22 August 2011 Keywords: Heavy metals Sediment pollution indices Sediment quality guidelines Multivariate statistical techniques Tigris River

a b s t r a c t Heavy metal concentrations in sediment samples from the Tigris River were determined to evaluate the level of contamination. The highest concentrations of metals were found at the ﬁrst site due to metallic wastewater discharges from copper mine plant. Sediment pollution assessment was carried out using contamination factor (CF), pollution load index (PLI), geoaccumulation index (Igeo) and enrichment factor (EF). The CF values for Co, Cu and Zn were >6 in sediments of the ﬁrst site, which denotes a very high contamination by these metals. The PLIs indicated that all sites except the ﬁrst site were moderately polluted. Cu, Co, Zn and Pb had the highest Igeo values, respectively. The mean EF values for all metals studied except Cr and Mn were >1.5 in the sediments of the Tigris River, suggesting anthropogenic impact on the metal levels in the river. The concentrations of Cr, Cu, Ni and Pb are likely to result in harmful effects on sediment-dwelling organisms which are expected to occur frequently based on the comparison with sediment quality guidelines. PCA/FA and cluster analysis suggest that As, Cd, Co, Cr, Cu, Mn, Ni and Zn are derived from the anthropogenic sources, particularly metallic discharges of the copper mine plant. © 2011 Elsevier B.V. All rights reserved.

1. Introduction Metal contamination in aquatic environments has received huge concern due to its toxicity, abundance and persistence in the environment, and subsequent accumulation in aquatic habitats. Heavy metal residues in contaminated habitats may accumulate in microorganisms, aquatic ﬂora and fauna, which, in turn, may enter into the human food chain and result in health problems [1,2]. Heavy metals discharged into a river system by natural or anthropogenic sources during their transport are distributed between the aqueous phase and bed sediments [3]. Because of adsorption, hydrolysis and co-precipitation only a small portion of free metal ions stay dissolved in water and a large quantity of them get deposited in the sediment [4]. Sediments are ecologically important components of the aquatic habitat and are also a reservoir of contaminants, which play a signiﬁcant role in maintaining the trophic status of any water body [5]. The measurements of pollutants in the water only are not conclusive due to water discharge ﬂuctuations and low resident time. The same holds true for the suspended material [6]. The study of sediment plays an important role as they have a long residence time. River sediments, therefore, are important sources for the assessment of man-made contamination in rivers. Sediments, not only

act as the carrier of contaminants, but also the potential secondary sources of contaminants in aquatic system [7,8]. Therefore, the analysis of river sediments is a useful method to study the metal pollution in an area [9]. The Tigris River is one of the most important rivers in Turkey. Some reports have been published on the heavy metal levels in sediment samples from the upper regions of the Tigris River [10,11]. In this paper, we report the ﬁrst comprehensive study on distribution of heavy metals in sediments of the Tigris River that was accomplished through regular monitoring of the river during a period of one year at seven different sites spread over the river stretch of about 500 km. The objectives of this study were (i) to determine the spatial and temporal distributions of heavy metals in surface sediments of the Tigris River, (ii) to deﬁne the natural and/or anthropogenic sources of these metals using multivariate statistical techniques, (iii) to explore the degree of heavy metal contamination in the river using contamination indices, (iv) to assess environmental risks of these metals in the study area by comparison with sediment quality guidelines (SQGs). 2. Materials and methods 2.1. Study area

∗ Tel.: +90 412 2266046; fax: +90 412 2266052. E-mail address: [email protected] 0304-3894/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.jhazmat.2011.08.051

The Tigris has been an important river throughout history and was one of the main water sources of the ancient Mesopotamian

356

M. Varol / Journal of Hazardous Materials 195 (2011) 355–364

2.2. Sampling sites Fig. 1 shows the locations of the sampling sites. Surface sediment samples were collected from seven sites, namely Maden (Site-1), E˘gil (Site-2), Diyarbakır (Site-3), Bismil (Site-4), Batman (Site-5), Hasankeyf (Site-6) and Cizre (Site-7), along the Tigris River. Uncontaminated sediment samples (USS) were also collected from a mountain stream in the study area for background studies. The brief description of sampling sites selected for this study is recorded in Table 1. 2.3. Sample collection

Fig. 1. Map showing sampling sites on the Tigris River.

Surface sediment samples were collected at monthly intervals between February 2008 and January 2009. The samples collected from each site consisted of 4–5 composite samples. Composite sediments (top 2 cm of surface) were taken by using a self-made sediment core sampler with an inner diameter of 6 cm and length of 60 cm. After sampling, the sediment samples were sealed in clean polyethylene bags, placed in a cooler at 4 ◦ C, and transported to the laboratory immediately for further analysis. 2.4. Chemical analysis

civilizations. The Tigris River originates in the Toros mountains of the Eastern Anatolia region of Turkey and follows a southeastern route to Cizre, where it forms the border between Turkey and Syria for 32 km before entering Iraq. The total length of the river is approximately 1900 km, of which 523 km is within Turkey. It drains a catchment area of about 57,614 km2 [12]. Currently, there are two major dams in operation on the Tigris River in Turkey: the Kralkızı and Dicle. Maximum ﬂows occur from February through April, whereas minimum ﬂows occur from August through October. The annual mean ﬂow of the river in Diyarbakır (upstream) and Cizre (downstream) was calculated to be 28.3 m3 /sn and 211.8 m3 /sn, respectively [13]. The continental climate of the Tigris Basin is a subtropical plateau climate. The annual mean air temperature varied between 14.6 ◦ C (Maden) and 21.8 ◦ C (Cizre) with the highest and the lowest temperature of 35.9 ◦ C and 0 ◦ C, respectively. Annual total precipitation ranged from 294.1 mm Cizre (downstream) to 611.1 mm in Maden (upstream), of which 82% concentrated during the time period of October to April [14].

2.4.1. Reagents and standards All reagents were of analytical grade or of Suprapure quality (Merck, Darmstadt, Germany). Double deionized water (Milli-Q System, Millipore) was used for the preparation of all solutions. The element standard solutions used for calibration were prepared by diluting stock solutions of 1000 mg/l of each element. Stock standard solutions were Merck Certiﬁcate AA standard (Merck). All glasswares used were cleaned by soaking in dilute acid for at least 24 h and rinsed abundantly in deionized water before use. 2.4.2. Analysis of sediment samples Sediment samples were air dried; then, stones and plant fragments were removed by passing the dried sample through a 2 mm sieve. The sieved sample was powdered and ﬁnally passed through a 500 ␮m sieve and stored in acid washed and deionized water rinsed glass bottles. For heavy metal content determinations, 0.25 g sediment subsamples were digested in teﬂon vessels with 12 ml HNO3 (65%):HCl (37%) (3:1) mixture in a microwave oven (MARSXpress, CEM) [15]. After microwave digestion, the sample solutions

Table 1 Locations and description of sampling sites along the Tigris River. Sites

Coordinates

Altitude (m)

Description

S-1

38◦ 20’N–39◦ 41’E

860

S-2

38◦ 06’N–40◦ 08’E

616

S-3

37◦ 53’N–40◦ 13’E

576

S-4

37◦ 50’N–40◦ 39’E

538

S-5

37◦ 54 N–41◦ 05’E

540

Site-1 is located about 3 km downstream of copper mine plant in Maden Township. Wastewaters containing heavy metal from plant discharge into the river before this site. Site-2 is located about 2 km downstream of Dicle Dam in E˘gil Township. Agricultural runoff and irrigation return ﬂow are pollution sources at this site. Site-3 is just near On Gözlü Köprü (Ten-Eyed Bridge) in Diyarbakır Province. Some wastewater drains that collect mixed domestic and industrial wastewater empty into the river before this site. Site-4 is situated just near Bismil Bridge. Wastewaters from the Diyarbakır wastewater treatment plant discharge into the river between the site-3 and site-4. In addition, domestic wastewaters from Bismil Township discharge into the river just before it reaches site-4. Agricultural runoff and irrigation return ﬂow are other pollution sources at this site. Site-5 is situated just near Batman Bridge. Agricultural runoff is pollution source at this site. Site-6 is located near Hasankeyf Bridge. Animal manure wastes and municipal wastewater discharges from Hasankeyf Township are pollution source at this site. Site-7 is located just near Cizre Bridge. Wastewater drains from Cizre Township empty into the river directly before this site. Additional pollution sources at this site are the sand pits near the river.

◦

◦

S-6

37 42’N–41 24’E

471

S-7

37◦ 19’N–42◦ 11’E

371

M. Varol / Journal of Hazardous Materials 195 (2011) 355–364

were ﬁltered, adjusted to a suitable volume with double deionized water. The sediment extracts were analyzed for Co, Cr, Cu, Fe, Mn, Ni and Zn by a ﬂame atomic absorption spectrometry (FAAS) equipped with deuterium background correction (AA240FS, Varian). As, Cd and Pb in extracts were measured by using a graphite furnace atomic absorption spectrometry (GFAAS) with Zeeman background correction (AA240Z, Varian).

2.4.3. Quality control The analytical data quality was guaranteed through the implementation of laboratory quality assurance and quality control methods, including the use of standard operating procedures, calibration with standards, analysis of reagent blanks, recovery of spiked samples and analysis of replicates. The accuracy and precision of the analytical procedures were tested by recovery measurements on spiked sediment samples. The sediment samples collected as uncontaminated sediment samples were spiked with metals and digested with the same procedure as the samples. The percentage recoveries of the metals in the spiked samples ranged from 91.4% (Fe) to 105.2% (Pb). The precision of the analytical procedures, expressed as the relative standard deviation (RSD), ranged from 5 to 10%. The precision for the analysis of standard solution was better than 5%. All analyses were carried out in duplicate, and the results were expressed as the mean.

2.5. Assessment of sediment contamination In the interpretation of geochemical data, choice of background values plays an important role. Many authors have used the average shale values or the average crustal abundance data as reference baselines. The best alternative is to compare concentrations between contaminated and mineralogically and texturally comparable, uncontaminated sediments [16–18]. Since there were no data on background concentrations for the investigated Tigris sediment and soils of close areas, the background values in this paper were calculated from the mean concentrations of heavy metals in uncontaminated sediments of the study area. In this study, four different indices were used to assess the degree of heavy metal contamination in sediments of the Tigris River.

2.5.1. Contamination factor (CF) The CF is the ratio obtained by dividing the concentration of each metal in the sediment by baseline or background value (concentration in uncontaminated sediment): CF =

Cheavy metal Cbackground

CF values were interpreted as suggested by Hakanson [19], where: CF < 1 indicates low contamination; 1 < CF < 3 is moderate contamination; 3 < CF < 6 is considerable contamination; and CF > 6 is very high contamination.

2.5.2. Pollution load index (PLI) For the entire sampling site, PLI has been determined as the nth root of the product of the n CF: PLI = (CF1 × CF2 × CF3 × · · · × CFn )1/n This empirical index provides a simple, comparative means for assessing the level of heavy metal pollution. When PLI > 1, it means that a pollution exists; otherwise, if PLI < 1, there is no metal pollution [20].

357

2.5.3. Geoaccumulation index (Igeo) The geoaccumulation index (Igeo) is deﬁned by the following equation: Igeo =

Log2 (Cn ) 1.5(Bn )

where Cn is the concentration of metals examined in sediment samples and Bn is the geochemical background concentration of the metal (n). Factor 1.5 is the background matrix correction factor due to lithospheric effects. The geoaccumulation index consists of seven classes [21]. Class 0 (practically unpolluted): Igeo ≤ 0; Class 1 (unpolluted to moderately polluted): 0 < Igeo < 1; Class 2 (moderately polluted): 1 < Igeo < 2; Class 3 (moderately to heavily polluted): 2 < Igeo < 3; Class 4 (heavily polluted): 3 < Igeo < 4; Class 5 (heavily to extremely polluted): 4 < Igeo < 5; Class 6 (extremely polluted): 5 > Igeo [22]. 2.5.4. Enrichment factor (EF) Enrichment factor (EF) is a useful tool in determining the degree of anthropogenic heavy metal pollution [16]. The EF is computed using the relationship below: EF =

(Metal/Fe) Sample (Metal/Fe) Background

In this study, iron (Fe) was used as the reference element for geochemical normalization because of the following reasons: (1) Fe is associated with ﬁne solid surfaces; (2) its geochemistry is similar to that of many trace metals and (3) its natural concentration tends to be uniform [22]. EF values were interpreted as suggested by Sakan et al. [16], where: EF < 1 indicates no enrichment; <3 is minor enrichment; 3–5 is moderate enrichment; 5–10 is moderately severe enrichment; 10–25 is severe enrichment; 25–50 is very severe enrichment; and >50 is extremely severe enrichment. 2.6. Sediment quality guidelines Sediment quality assessment guidelines (SQGs) are very useful to screen sediment contamination by comparing sediment contaminant concentration with the corresponding quality guideline [23]. These guidelines evaluate the degree to which the sediment-associated chemical status might adversely affect aquatic organisms and are designed to assist in the interpretation of sediment quality. Such SQGs have been used in numerous applications, including designing monitoring programs, interpreting historical data, evaluating the need for detailed sediment quality assessments, assessing the quality of prospective dredged materials, conducting remedial investigations and ecological risk assessments, and developing sediment quality remediation objectives [23]. The consensus-based sediment-quality guidelines (SQGs) were used in this study to assess possible risk arises from the heavy metal contamination in sediments of the study area. The consensus-based SQGs were developed from the published freshwater sediment-quality guidelines that have been derived from a variety of approaches [23]. These synthesized guidelines consist of a threshold effect concentration (TEC) below which adverse effects are not expected to occur and a probable effect concentration (PEC) above which adverse effects are expected to occur more often than not. An apparent advantage of the consensus-based guidelines is that MacDonald et al. [23] evaluated the reliability of the TECs and PECs for assessing sedimentquality conditions by determining their predictive ability that is, the ability of the guidelines to correctly classify ﬁeld-collected sediments as nontoxic or toxic to one or more aquatic organisms.

358

M. Varol / Journal of Hazardous Materials 195 (2011) 355–364

Table 2 Maximum, minimum, mean and standard deviation values of heavy metals in sediments of all sites studied in the Tigris River. Sites

Maden

E˘gil

Diyarbakır

Bismil

Batman

Hasankeyf

Cizre

Metal concentrations (mg/kg)

Max Min Mean SD Max Min Mean SD Max Min Mean SD Max Min Mean SD Max Min Mean SD Max Min Mean SD Max Min Mean SD

As

Cd

Co

Cr

Cu

Mn

Ni

Pb

Zn

18.0 5.0 8.9 4.0 4.9 2.0 3.3 0.9 6.6 3.5 4.8 1.0 5.2 2.4 3.5 0.8 6.0 2.3 3.6 1.0 3.6 2.2 2.9 0.5 8.5 2.9 5.4 1.8

4.9 1.4 2.4 1.2 2.4 1.4 1.8 0.3 3.0 1.4 1.8 0.4 2.6 1.0 1.6 0.5 1.6 0.7 1.2 0.3 2.5 0.8 1.6 0.5 2.7 1.8 2.2 0.3

389.8 55.6 155.9 107.7 30.6 20.9 25.7 2.9 39.7 23.2 30.3 4.9 25.6 12.3 15.9 3.7 12.2 5.4 9.0 1.8 16.1 5.4 10.0 2.6 19.0 11.1 14.1 2.7

151.7 76.4 119.0 21.8 96.0 56.4 76.4 11.3 163.4 98.1 115.4 16.9 113.2 67.7 83.8 13.6 65.9 35.8 50.5 9.5 90.2 28.4 54.6 15.7 124.4 65.7 93.6 20.2

5075.6 673.1 1941.9 1592.3 131.6 91.6 117.0 12.4 297.2 117.5 189.7 53.4 136.3 50.8 73.9 25.9 36.4 17.2 24.1 5.7 64.2 11.2 28.5 15.5 59.2 27.7 37.3 10.4

1657.0 822.0 1233.3 268.5 752.0 540.5 629.8 63.9 787.5 556.3 663.2 70.7 1228.0 528.7 641.3 189.8 590.8 282.2 420.2 92.2 791.0 329.5 489.7 130.6 982.9 529.8 702.5 124.4

288.0 151.9 216.8 44.9 144.4 113.4 132.0 9.1 174.5 153.3 162.3 5.9 172.8 137.0 149.6 10.1 109.7 79.3 93.9 8.9 125.7 74.0 91.0 15.9 244.7 135.5 173.7 35.9

566.6 144.4 393.9 121.7 358.1 89.1 255.5 104.0 387.7 89.0 250.3 102.7 392.4 185.5 274.3 59.0 299.2 62.3 163.7 95.6 344.6 73.9 221.8 86.9 387.6 93.7 297.3 79.8

2396.6 191.3 530.4 597.9 190.6 134.0 165.2 14.5 247.0 136.0 178.2 27.6 220.8 107.3 146.1 31.7 183.1 84.8 129.6 30.3 189.0 60.1 120.5 30.6 191.2 123.1 152.1 21.0

2.7. Statistical analyses Analysis of variance (ANOVA) was performed to analyze the signiﬁcant spatial and temporal differences (p < 0.05). Relationships among the considered variables were tested using Pearson’s coefﬁcient with statistical signiﬁcance set at p < 0.05. Multivariate analysis of the river data set was performed using cluster analysis (CA) and principal component analysis/factor analysis (PCA/FA) techniques. The above statistical analyses were applied to experimental data standardized through z-scale transformation to avoid misclassiﬁcation due to wide differences in data dimensionality. Kaiser–Meyer–Olkin (KMO) and Bartlett’s sphericity tests were performed to examine the suitability of the data for PCA/FA [24]. KMO is a measure of sampling adequacy that indicates the proportion of variance that is common, i.e., variance that may be caused by underlying factors. A high value (close to 1) generally indicates that principal component/factor analysis may be useful, as was the case in this study, where KMO = 0.76. Bartlett’s test of sphericity indicates whether a correlation matrix is an identity matrix, which would indicate that variables are unrelated. The signiﬁcance level of 0 in this study (less than

0.05) indicated that there were signiﬁcant relationships among the variables. Hierarchical agglomerative cluster analysis (CA) was performed on the normalized data set using Ward’s method with squared Euclidean distances as a measure of similarity. Factor analysis (FA) was conducted after principal component analysis (PCA). PCA of the normalized variables (data set) was performed to extract signiﬁcant principal components (PCs) and to further reduce the contribution of variables with minor signiﬁcance; these PCs were then subjected to varimax rotation (raw) to generate varifactors (VFs).

3. Results and discussion 3.1. Heavy metals in sediments of the Tigris River The basic statistics for all of the metal parameters measured during the sampling period of one year at seven different sites are summarized in Table 2. During the study period, all heavy metals showed signiﬁcant spatial variations (ANOVA, p < 0.05). The ranges of metals

Table 3 Heavy metal concentrations reported for previous studies conducted in the Tigris River. Sites

Maden E˘gil Diyarbakır Bismil Maden E˘gil Diyarbakır Bismil Diyarbakır Diyarbakır Bismil

Metal concentrations (mg/kg)

References

Cd

Co

Cu

Mn

Ni

Pb

Zn

– – – – – – – – – nd nd

– – – – 503 118 21 4 32.01 43.13 32.4

3433 1213 904 991 3433 1213 904 991 728.96 137 92.5

– – – – – – – – – 622.9 497.7

– – – – 403 305 50 41 66.35 124.5 99.51

– – – – 102 83 31 24 – nd nd

891 456 405 716 891 456 405 716 369.14 30 42.7

[27] [27] [27] [27] [10] [10] [10] [10] [28] [11] [11]

[16] [29] [30] [5] [31] [32] [33] [34] [35] [36] [37] [38] [39] [40] 54–567 24.7–45.5 178–645 8.47–343.47 86.1–708.8 78–2010 42–271 68–5280 11–221 82–3700 404–1920 160–8076 54–4380 21.09–25.66 11–123 3.3–17.3 32–98.5 4.86–156.2 39.3–189 14.7–541.8 11–140 17–13400 3.5–23.2 19–270 522–6880 62–2281 11–681 8.65–38.29 –

17–55 15.4–79.2 52.3–161 4.75–76.08 – 17.5–173.3 19–188 1.6–36 2–112 – – 9–23 6–78 – 490–2316 221–446 – 65.73–834.7 – 442–1655 – – 75–2810 410–6700 936–5240 750–14000

32–162 13.1–38.7 31.7–90.1 3.6–245.33 71.6–420.8 31.1–8088 14–93 22–2700 10–81 18–480 19.5–76.9 51–796 9–1739 6.47–178.61 –

7–23 – – 2.4–88.7 84.4–23.4 26.5–556.5 39–180 11–151 8–274 33–71 – 24–71 29–240.5 28.7–152.73 – – – 1.5–23.4 – – 6.8–42 – – – 13–24

2.08–12.90

0.12–0.55 1.08–3.7 0.09–17.83 1.0–4.3 1.1–32.9 1–11 0.13–12 – 0.1–22 0.95–5.95 0.5–31 0.5–24.8 0.03–0.37 – – – – 8.1–388 1–40 – – 2.8–31 – 21–1543

[3] 32–2200 126–345 – – 207–1660 13–66 22–47 –

–

This study This study Zn

191.3–2396 60.1–247 144.4–566.6 62.3–392.4

Pb Ni

151.9–288 74–244.7 822.0–1657 282.2–1228

Mn Cu

673.1–5075.6 11.2–297.2 76.4–151.7 28.4–163.4

Cr Co

55.6–389.8 5.4–39.7 1.4–4.9 0.7–3 5–18 2–8.5

Cd As

Metal concentrations (mg/kg)

359

Tigris River (site-1), Turkey Tigris River (the rest of sites), Turkey Shing Mun River, Hong Kong Tisza River, Serbia Yes¸ilırmak River, Turkey River Po, Italy Gomti River, India Almendares River, Cuba Danube River, Europa Axios River, Greece Tinto River, Spain Nile River, Egypt South Platte River, USA Tees River, UK Rimac River, Peru Sea Scheldt River, Belgium Luan River, China

The results of contamination factors (CFs) and pollution load index (PLI) are presented in Table 5. The highest CF values for all metals studied were found at site-1 (Maden), which receives

Locations

3.2. Indices of sediment contamination

Table 4 Heavy metal concentrations in sediment samples from the Tigris River and other selected rivers from the literature.

in sediments were: 2.0–18.0 mg/kg for As, 0.7–4.9 mg/kg for Cd, 5.4–389.8 mg/kg for Co, 28.4–163.4 mg/kg for Cr, 11.2–5075.6 mg/kg for Cu, 282.2–1657.0 mg/kg for Mn, 74.0–288.0 mg/kg for Ni, 62.3–566.6 mg/kg for Pb and 60.1–2396.6 mg/kg for Zn. The highest concentrations of heavy metals were found at site-1 (Maden) due to metallic wastewater discharges from copper mine plant in Maden Township. Site-3 (Diyarbakır) which receives untreated domestic and industrial wastewaters from Diyarbakır province, site-4 (Bismil) which receives partially treated domestic wastewater from Diyarbakır wastewater treatment plant, untreated domestic wastewater from Bismil Township and agricultural runoff, and site-7 (Cizre) which receives untreated domestic wastewater from Cizre Township had also high metal concentrations. The lowest mean values of As, Ni and Zn were found at site-6 (Hasankeyf), while the lowest mean values of Cd, Co, Cr, Cu, Fe, Mn and Pb were calculated at site-5 (Batman). In this study, total metal concentrations followed the order of site-1 > site-7 > site-4 > site-3 > site > 2 > site > 6 > site-5. During the study, all metals studied did not show signiﬁcant temporal differences (ANOVA, p > 0.05). In this study, heavy metal concentrations in assessed sediment samples from the Tigris River were compared with previous studies (Table 3). The mean values of Co, Cu, Ni and Zn except Pb at site-1 (Maden) were lower when compared with an earlier study conducted in 1990 [10] due to reduction of the activity of the copper mine plant. The mean values of Co, Cu, Ni and Zn at site-2 (E˘gil) were found signiﬁcantly lower than those at the same site reported for the Tigris River owing to the construction of two dams on the river over the last 10 years: Kralkızı and Dicle. It is well known that concentrations of suspended solids and heavy metals in the reservoir water will be decreased signiﬁcantly due to sediment deposition. The water leaving the reservoir can be clearer, and this could affect the river downstream of the dam. However, the mean values of Co, Ni and Pb except Cu and Zn at site-3 (Diyarbakır) and site-4 (Bismil) were higher than those reported by Gümgüm et al. [10]. In this study, the mean values of Cd, Cu, Mn, Ni, Pb and Zn except Co at site-3 were found higher, while the mean values of Cd, Mn, Ni, Pb and Zn except Co and Cu at site-4 were higher when compared with a previous study conducted in 2000 [11]. The increase in some metal concentrations at site-3 (Diyarbakır) and site-4 (Bismil) may be due to increased anthropogenic activities in the Diyarbakır province which has the largest urban settlement in Tigris Basin. It may have contributed large amounts of heavy metals into the river. Total heavy metal concentrations in the sediment samples from the Tigris River followed the order: Fe > Mn > Cu > Pb > Zn > Ni > Cr > Co > As > Cd. The results were not compatible with previous studies [10,11] conducted in the Tigris River. Karadede-Akin and Ünlü [11] found that Fe was the most abundant in the sediment, followed by Mn, Cu and Co, and the least was Zn, while Cd and Pb were not recorded. Gümgüm et al. [10] reported that the accumulation order of heavy metals in the sediment samples was Cu > Zn > Ni > Co > Pb. Comparison of metal contamination data of the Tigris River with the published data of other rivers (Table 4) reveals that the sediments of site-1 are severely polluted with heavy metals, while sediments of the rest of sites are slightly polluted. The extent of metal pollution in the Tigris River was not much more serious than that in the Tinto River, Danube River and Rimac River, and much worse than the Yes¸ilırmak River, River Po, Luan River, Nile River and Axios River (Table 4).

References

M. Varol / Journal of Hazardous Materials 195 (2011) 355–364

360

M. Varol / Journal of Hazardous Materials 195 (2011) 355–364

Table 5 Metal contamination factors (CFs) and pollution load indices (PLIs) for sediments of all sites studied in the Tigris River. Sites

Contamination factors (CFs)

Site-1 Site-2 Site-3 Site-4 Site-5 Site-6 Site-7 Mean Min Max

PLI

As

Cd

Co

Cr

Cu

Fe

Mn

Ni

Pb

Zn

3.42 1.27 1.85 1.35 1.38 1.12 2.08 1.78 1.12 3.42

2.86 2.14 2.14 1.90 1.43 1.90 2.62 2.14 1.43 2.86

8.66 1.43 1.68 0.88 0.50 0.56 0.78 2.07 0.50 8.66

1.86 1.19 1.80 1.31 0.79 0.85 1.46 1.32 0.79 1.86

34.68 2.09 3.39 1.32 0.43 0.51 0.67 6.16 0.43 34.68

1.11 0.99 1.02 1.08 0.88 0.98 1.10 1.02 0.88 1.11

2.37 1.21 1.28 1.23 0.81 0.94 1.35 1.31 0.81 2.37

2.93 1.78 2.19 2.02 1.27 1.23 2.35 1.97 1.23 2.93

5.79 3.76 3.68 4.03 2.41 3.26 4.37 3.90 2.41 5.79

6.80 2.12 2.28 1.87 1.66 1.54 1.95 2.60 1.54 6.80

a huge amount of metallic discharge from copper mine plant in Maden Township. The CF values for Co, Cu and Zn were >6 in sediments of site-1, which denotes a “very high contamination” by these metals. The CF values for As and Pb in sediments of site1 showed a “considerable contamination”, while the CF values for Cd, Cr, Fe, Mn and Ni indicated a “moderate contamination”. The CF values for metals studied except Pb at other sites showed “moderate contamination”. In this study, Cu had the highest and lowest CF values among the ten metals studied. However, Pb had the highest CF values among the ten metals studied at all sites except site-1. Site-3 (Diyarbakır) which receives municipial and industrial wastewater discharges from Diyarbakır and site7 (Cizre) which receives municipial wastewater discharges from Cizre showed high CF values. Total contamination factors followed the order of site-1 > site-3 > site-7 > site-2 > site-4 > site-6 > site-5. The pollution load index (PLI) ranged from 1.02 to 4.19 (Table 5). According to the mean PLI value (1.88), the Tigris River was moderately polluted. Site-1 had the highest PLI (4.19) within the study area, indicating that the sediments of site-1 were strongly polluted by investigated heavy metals. Other sites where PLI was between 1 and 2 must be classiﬁed as moderately polluted. The PLI followed the order of site-1 > site-3 > site-7 > site-2 > site-4 > site-6 > site-5. Table 6 presents Igeo and EF values of the metals studied. The Igeo values of As at sites 2, 4, 5 and 6, Cd at site-5, Co at sites 2, 4, 5, 6 and 7, Cr and Mn at all sites except site-1, Cu at sites 4, 5, 6 and 7, Fe at all sites, and Ni at sites 5 and 6 were less than zero, suggesting that these sites were not polluted by these metals. The Igeo values for Cd, Cr, Mn and Ni were under 1 in the sediments of all sites which usually had “unpolluted to moderately polluted” class. Among ten metals studied, Cu, Co, Zn and Pb had the highest Igeo values, respectively. The highest Igeo values of metals studied were found in the sediments of site-1. The Igeo class of Cu was “extremely polluted” for sediments of site-1. The Igeo class of As and Pb were “moderately polluted” for sediments of site-1, while

4.19 1.67 1.99 1.55 1.02 1.11 1.62 1.88 1.02 4.19

the Igeo class of Co and Zn were “moderately to heavily polluted”. Total Igeo values followed the order of site-1 > site-3 > site-2 > site7 > site-4 > site-6 > site-5. According to Zhang and Liu [25], EF values between 0.05 and 1.5 indicate that the metal is entirely from crustal materials or natural processes, whereas EF values higher than 1.5 suggest that the sources are more likely to be anthropogenic. In this study, the mean EF values for all metals studied except Cr and Mn were >1.5 in the sediments of the Tigris River, suggesting anthropogenic impact on the metal levels in the river. The highest EF values were found at site-1 (Maden) due to metallic wastewater discharges from the copper mine plant in Maden Township. The EF value for Cu in the sediments of site-1 was 31.34, showing “very severe enrichment”, while the EF values for Co, Pb and Zn were between 5 and 10, indicating “moderately severe enrichment”. However, the EF values for As, Cd, Cr, Mn and Ni at site-1 indicated “minor enrichment”. Cu had the highest and lowest EF values among the ten metals studied. Co had the second highest EF value. Pb at all sites except site-1 had the highest EF values among the ten metals studied. The EF values for metals studied in sediments of other sites showed “minor to moderate enrichment”. Total EF values followed the order of site-1 > site-3 > site-2 > site-7 > site-4 > site-6 > site-5. 3.3. Application of sediment quality guidelines It is important to determine whether the concentrations of heavy metals in sediments pose a threat to aquatic life. In this study, heavy metal concentrations in assessed sediment samples were compared with consensus-based TEC and PEC values (Table 7). As, Cu and Zn were lower than the TEC in 96.4%, 28.6% and 16.7% of samples, respectively. Cd, Cr, Cu and Zn were between the TEC and PEC in 95.2%, 71.4%, 46.4% and 79.7% of samples, respectively. Ni and Pb exceeded the PEC in 100% and 83.3% of samples, respectively. Cr exceeded the PEC in 9 of samples at site-1, 6 of samples at site-3, 1 of samples at site-4 and 3 of samples at site-7. Cu exceeded the PEC

Table 6 Geoaccumulation indices (Igeo) and enrichment factors (EF) of heavy metals for sediments of all sites studied in the Tigris River. Sites

Site-1 Site-2 Site-3 Site-4 Site-5 Site-6 Site-7 Mean Min Max

As

Cd

Co

Cr

Cu

Fe

Mn

Ni

Pb

Zn

Igeo

EF

Igeo

EF

Igeo

EF

Igeo

EF

Igeo

EF

Igeo

EF

Igeo

EF

Igeo

EF

Igeo

EF

Igeo

EF

1.19 −0.24 0.30 −0.16 −0.12 −0.43 0.47 0.14 −0.43 1.19

3.09 1.29 1.81 1.25 1.57 1.14 1.90 1.72 1.14 3.09

0.93 0.51 0.51 0.34 −0.07 0.34 0.80 0.48 −0.07 0.93

2.58 2.17 2.10 1.77 1.62 1.95 2.39 2.08 1.62 2.58

2.53 −0.07 0.17 −0.76 −1.58 −1.43 −0.94 −0.30 −1.58 2.53

7.83 1.45 1.65 0.82 0.57 0.57 0.72 1.94 0.57 7.83

0.31 −0.33 0.27 −0.20 −0.93 −0.81 −0.04 −0.25 −0.93 0.31

1.68 1.21 1.77 1.21 0.90 0.87 1.34 1.28 0.87 1.77

4.53 0.48 1.18 −0.18 −1.80 −1.56 −1.17 0.21 −1.80 4.53

31.34 2.12 3.32 1.22 0.49 0.52 0.61 5.66 0.49 31.34

−0.44 −0.60 −0.56 −0.48 −0.77 −0.62 −0.45 −0.56 −0.77 −0.44

1 1 1 1 1 1 1 1 1 1

0.66 −0.31 −0.23 −0.28 −0.89 −0.67 −0.15 −0.27 −0.89 0.66

2.14 1.23 1.25 1.14 0.92 0.96 1.23 1.27 0.92 2.14

0.97 0.25 0.55 0.43 −0.24 −0.29 0.65 0.33 −0.29 0.97

2.65 1.81 2.15 1.87 1.44 1.26 2.14 1.90 1.26 2.65

1.95 1.32 1.30 1.43 0.68 1.12 1.54 1.33 0.68 1.95

5.23 3.81 3.61 3.74 2.74 3.34 3.99 3.78 2.74 5.23

2.18 0.50 0.61 0.32 0.15 0.04 0.38 0.60 0.04 2.18

6.15 2.15 2.24 1.74 1.89 1.58 1.78 2.50 1.58 6.15

M. Varol / Journal of Hazardous Materials 195 (2011) 355–364

361

Table 7 Comparison between sediment quality guidelines (SQGs) with heavy metal concentrations (mg/kg) of all sites studied in the Tigris River.

SQGs Measured values in this study Site-1

Site-2

Site-3

Site-4

Site-5

Site-6

Site-7

Total

TEC PEC Minimum Maximum Average Samples < TEC Samples between TEC and PEC Samples > PEC Samples < TEC Samples between TEC and PEC Samples > PEC Samples < TEC Samples between TEC and PEC Samples > PEC Samples < TEC Samples between TEC and PEC Samples > PEC Samples < TEC Samples between TEC and PEC Samples > PEC Samples < TEC Samples between TEC and PEC Samples > PEC Samples < TEC Samples between TEC and PEC Samples > PEC Samples < TEC Samples between TEC and PEC Samples > PEC

As

Cd

Cr

Cu

Ni

Pb

Zn

9.79 33 2.0 18.0 4.6 9 3 0 12 0 0 12 0 0 12 0 0 12 0 0 12 0 0 12 0 0 81 (96.4%) 3 (3.6%) 0 (0%)

0.99 4.98 0.7 4.9 1.8 0 12 0 0 12 0 0 12 0 0 12 0 2 10 0 2 10 0 0 12 0 4 (4.8%) 80 (95.2%) 0 (0%)

43.4 111 28.4 163.4 84.8 0 3 9 0 12 0 0 6 6 0 11 1 2 10 0 3 9 0 0 9 3 5 (6%) 60 (71.4%) 19 (22.6%)

31.6 149 11.2 5075.6 344.6 0 0 12 0 12 0 0 3 9 0 12 0 11 1 0 9 3 0 4 8 0 24 (28.6%) 39 (46.4%) 21 (25%)

22.7 48.6 74.0 288.0 145.6 0 0 12 0 0 12 0 0 12 0 0 12 0 0 12 0 0 12 0 0 12 0 (0%) 0 (0%) 84 (100%)

35.8 128 62.3 566.6 265.3 0 0 12 0 3 9 0 2 10 0 0 12 0 6 6 0 2 10 0 1 11 0 (0%) 14 (16.7%) 70 (83.3%)

121 459 60.1 2396.6 203.1 0 9 3 0 12 0 0 12 0 3 9 0 5 7 0 6 6 0 0 12 0 14 (16.7%) 67 (79.7%) 3 (3.6%)

in all of samples at site-1 and 9 of samples at site-3. Ni exceeded the PEC in all of samples. Pb exceeded the PEC in all of samples at site-1 and site-4, 9 of samples at site-2, 10 of samples at site-3 and site6, 6 of samples at site-5 and 11 of samples at site-7. Zn exceeded the PEC in 3 of samples at site-1. These results indicate that the concentrations of Cr, Cu, Ni and Pb are likely to result in harmful effects on sediment-dwelling organisms which are expected to occur frequently. An index of toxicity risk, PEC quotients, was also evaluated in this study. PEC quotients were calculated using the methods of MacDonald et al. [23]. Sediment samples are predicted to be not toxic if PEC quotients are <0.5. In contrast, sediment samples are predicted to be toxic when PEC quotients exceed 1.5 [23]. In this study, PEC quotients varied from 0.09 to 13.03 (Table 8). The lowest value of PEC quotients was calculated at site-6, while the highest value was calculated for the sediments of site-1. The total PEC quotients followed the order of site-1 > site-3 > site-7 > site-4 > site2 > site-6 > site-5. PEC quotients of Cu at site-1, Ni and Pb at all sites exceeded 1.5, suggesting a potential toxicity of these metals in sediments of the river. Conversely, the toxicity risks were much lower for As and Cd at all sites, Cr and Cu at sites 5 and 6 and Zn at all sites except site-1, with PEC quotients <0.5.

3.4. Multivariate statistical analyses 3.4.1. Principal component analysis/factor analysis PCA/FA was performed on the normalized data to compare the compositional pattern between the sediment samples and to identify the factors inﬂuencing each one. PCA of the entire data set (Table 2) revealed three PCs with eigenvalues >1 that explained about 83.9% of the total variance in the sediment quality data set. The ﬁrst PC accounting for 54.8% of the total variance was correlated (loading >0.70) with Cd, Co, Cu, Mn and Ni. The second PC accounting for 17.4% of total variance was correlated with Fe. Whereas the third PC accounted for the total variance of 11.7%, it correlated (loading >0.70) with none of the metal parameters. Three VFs were obtained through FA performed on the PCs. The corresponding VFs, variable loadings and the explained variance are presented in Table 9. The loading plots of the ﬁrst two VFs are presented in Fig. 2. VF coefﬁcients having a correlation greater than 0.70 were considered signiﬁcant (strong). VF1, which explained 56.5% of the total variance, had strong positive loadings (>0.70) on Cd, Co, Cu and Mn, and a moderate positive loading on Ni. This VF represents anthropogenic sources. In Maden Township (upstream), there is a copper mine plant that discharges

Table 8 PEC quotients of heavy metals for sediments of all sites studied in the Tigris River.

As Cd Cr Cu Ni Pb Zn Mean Min Max

Site-1

Site-2

Site-3

Site-4

Site-5

Site-6

Site-7

0.27 0.48 1.07 13.03 4.46 3.08 1.16 3.36 0.27 13.03

0.10 0.36 0.69 0.78 2.72 2.00 0.36 1.00 0.10 2.72

0.14 0.37 1.04 1.27 3.34 1.96 0.39 1.22 0.14 3.34

0.11 0.33 0.76 0.50 3.08 2.14 0.32 1.03 0.11 3.08

0.11 0.23 0.45 0.16 1.93 1.28 0.28 0.63 0.11 1.93

0.09 0.31 0.49 0.19 1.87 1.73 0.26 0.71 0.09 1.87

0.16 0.44 0.84 0.25 3.57 2.32 0.33 1.13 0.16 3.57

362

M. Varol / Journal of Hazardous Materials 195 (2011) 355–364

Table 9 Loadings of experimental variables (10) on signiﬁcant principal components for the Tigris River data set* . VF1 As Cd Co Cr Cu Fe Mn Ni Pb Zn Eigenvalue % Total variance Cumulative % variance *

VF2

Apr. Oct.

VF3

−0.096 0.263 0.102 0.258 0.141 0.931 0.304 0.414 0.908 −0.198 1.707 17.065 73.611

0.458 0.726 0.938 0.031 0.945 −0.016 0.732 0.507 0.342 0.561 5.655 56.546 56.546

Feb.

0.792 0.165 0.234 0.911 0.130 0.090 0.533 0.669 0.143 0.547 1.158 11.584 85.196

Nov. Mar. Jul. Jan. May Dec. Jun. Aug. Sep. 0

20

40

60

80

100

120

(Dlink/Dmax)*100

Bold and italic values indicate strong and moderate loadings, respectively. Fig. 4. Dendrogram showing clustering of sampling periods. 1,2 Fe

1,0

Pb

VF2 (17.065%)

0,8 0,6

Ni

0,4

Mn Cd

Cr

Cu Co

0,2 0,0

As

Zn

-0,2 -0,4 -0,1

0,0

0,1

0,2

0,3

0,4

0,5

0,6

0,7

0,8

0,9

1,0

VF1 (56.546%) Fig. 2. Loading plots of the ﬁrst two VFs obtained for the data set.

metallic wastewaters containing high levels of Co, Cu and Ni into the Tigris River [10,26]. VF2, which accounted for 17.0% of the total variance, had strong positive loadings on Fe and Pb. This factor represents lithogenic sources. VF3 (11.5% of total variance) had strong positive loadings on As and Cr, and moderate positive loadings on Mn, Ni and Zn. This VF represents anthropogenic sources. The elements are derived from municipal and industrial wastewaters, and metallic wastewaters of the copper mine plant. 3.4.2. Cluster analysis Cluster analysis (CA) was applied to the river sediment quality data set to group the similar sampling sites (spatial variability). Spatial CA rendered a dendrogram (Fig. 3) where all seven sampling sites on the river were grouped into three statistically signiﬁ-

cant clusters at (Dlink /Dmax ) × 100 < 40. Cluster 1 (Maden) site was located in a high pollution region, which receives metallic wastewater discharges from copper mine plant. Cluster 2 (E˘gil, Diyarbakır, Bismil, Hasankeyf and Cizre) sites were in a moderate pollution region. Cluster 3 (Batman) site was in a region of relatively low pollution. Temporal CA generated a dendrogram (Fig. 4) that grouped the 12 months into two clusters at (Dlink /Dmax ) × 100 < 60, and the difference between the clusters was signiﬁcant. Cluster 1 included February, April, October, November, March, July, January, May and December roughly corresponding to the wet season in Turkey (October to April). In this study, about 82% annual total precipitation was concentrated in the time period from October to April. Cluster 2 included the remaining months (June, August and September), closely corresponding to the dry season (May to September). However, if 12 months had been empirically divided into spring (March to May), summer (June to August), autumn (September to November) and winter (December to February), or into dry/wet seasons, a mistake in grouping could have been made. In fact, Fig. 4 shows that the temporal patterns in water quality were not purely consistent with the four seasons or the dry/wet season. Similarly, CA was performed to group the analyzed parameters. CA rendered a dendrogram (Fig. 5) where all ten metal parameters were grouped into three statistically signiﬁcant clusters at (Dlink /Dmax ) × 100 < 85. Cluster 1 includes As and Zn which were identiﬁed as contaminants derived from anthropogenic sources (wastewater discharges of copper mine plant). Cluster 2 contains Cd, Mn, Ni, Cr, Co and Cu derived from anthropogenic sources

As Zn

Maden (1)

Cd

Eğil (2)

Mn

Diyarbakır (3)

Ni Cr

Bismil (4)

Co

Cizre (7)

Cu

Hasankeyf (6)

Fe

Batman (5)

Pb

0

20

40

60

80

100

120

(Dlink/Dmax)*100 Fig. 3. Dendrogram showing clustering of sampling sites on the Tigris River.

0

20

40

60

80

100

(Dlink/Dmax)*100 Fig. 5. Dendrogram showing clustering of the analyzed parameters.

120

M. Varol / Journal of Hazardous Materials 195 (2011) 355–364

363

Table 10 Pearson correlation matrix of heavy metals in the Tigris River.

As Cd Co Cr Cu Fe Mn Ni Pb Zn

As

Cd

Co

Cr

Cu

Fe

Mn

Ni

Pb

Zn

1 0.277b 0.404a 0.583a 0.383a −0.013 0.575a 0.473a 0.120 0.681a

1 0.660a 0.331a 0.682a 0.214 0.714a 0.699a 0.466a 0.260b

1 0.317a 0.973a 0.219b 0.835a 0.717a 0.488a 0.393a

1 0.259b 0.267b 0.585a 0.732a 0.398a 0.316a

1 0.213 0.796a 0.673a 0.502a 0.389a

1 0.363a 0.425a 0.853a −0.068

1 0.862a 0.608a 0.528a

1 0.629a 0.342a

1 0.088

1

Bold values represent correlation with signiﬁcance. a Signiﬁcance at the 0.01 probability level (2-tailed). b Signiﬁcance at the 0.05 probability level (2-tailed).

(wastewater discharges of copper mine plant, and industrial and domestic wastewaters). Cluster 3, which contains Fe and Pb, are derived from lithogenic sources.

3.4.3. Correlation matrix In order to establish relationships among metals and determine the common source of metals in the Tigris River, a correlation matrix was calculated for heavy metals in the sedimens. According to the values of Pearson correlation coefﬁcients (Table 10), a significant positive correlation existed among the metals studied. In this study, Fe did not show signiﬁcant correlation with As, Cd, Cu and Zn, and Pb did not show signiﬁcant correlation with Zn. Fe was signiﬁcantly correlated with Pb (r = 0.853, p < 0.01), indicating that the elements were derived from lithogenic sources. The signiﬁcantly positive correlation of As (r = 0.383, p < 0.01), Cd (r = 0.682, p < 0.01), Co (r = 0.973, p < 0.01), Cr (r = 0.259, p < 0.01), Mn (r = 0.796, p < 0.01), Ni (r = 0.673, p < 0.01), and Zn (r = 0.389, p < 0.01) with Cu showed that the elements were derived from wastewater discharges of copper mine plant and also moving together.

4. Conclusion Different useful tools, methods, guidelines and indices have been employed for evaluation of sediment pollution in the Tigris River, Turkey. The highest concentrations of heavy metals were found at site-1 (Maden) due to metallic wastewater discharges from copper mine plant in Maden Township. Site-3 (Diyarbakır), site-4 (Bismil) and site-7 (Cizre) had also high metal concentrations due to domestic and industrial wastewaters. Total heavy metal concentrations in the sediment samples from the Tigris River followed the order: Fe > Mn > Cu > Pb > Zn > Ni > Cr > Co > As > Cd. The highest values of contamination factor (CF), pollution load index (PLI), geoaccumulation index (Igeo) and enrichment factor (EF) for all metals studied were found at site-1 (Maden), which receives a huge amount of metallic discharge from copper mine plant in Maden Township. Heavy metal concentrations in assessed sediment samples were compared with consensus-based TEC and PEC values. The results have indicated that the concentrations of Cr, Cu, Ni and Pb are likely to result in harmful effects on sediment-dwelling organisms which are expected to occur frequently. Multivariate analysis (PCA/FA, CA) and correlation matrix were used in this study. The PCA/FA applied on the investigated heavy metals identiﬁed three varifactors (VFs). VF1 and VF3, which were loaded with As, Cd, Co, Cr, Cu, Mn, Ni and Zn, were related to the anthropogenic sources. The CA classiﬁed all the sampling sites into three main groups of spatial similarities. A signiﬁcant positive correlation is observed among As, Cd, Co, Cr, Cu, Mn, Ni and Zn, indicating that these metals were derived from similar sources and also moving together.

Acknowledgements The author thanks three anonymous reviewers for their valuable comments and constructive suggestions.

References [1] J.A. Cook, S.M. Andrew, M.S. Johnson, Lead, zinc, cadmium and ﬂuoride in small mammals from contaminated grass-land established on ﬂuorspar tailings, Water Air Soil Poll. 51 (1990) 43–54. [2] J. Deniseger, J. Erickson, A. Austin, M. Roch, M.J.R. Clark, The effects of decreasing heavy metal concentrations on the biota of Buttle Lake, Water Res. 24 (1990) 403–416. [3] S.N. Sin, H. Chua, W. Lo, L.M. Ng, Assessment of heavy metal cations in sediments of Shing Mun River, Hong Kong Environ. Int. 26 (2001) 297–301. [4] V.K. Gaur, S.K. Gupta, S.D. Pandey, K. Gopal, V. Misra, Distribution of heavy metals in sediment and water of River Gomti, Environ. Monit. Assess. 102 (2005) 419–433. [5] M. Singh, A.A. Ansari, G. Muller, I.B. Singh, Heavy metals in freshly deposited sediments of Gomti river (a tributary of the Ganga river): effects of human activities, Environ. Geol. 29 (1997) 246–252. [6] U. Förstner, G.T.W. Wittman, Metal Pollution in Aquatic Environment, SpringerVerlag, New York, 1983. [7] J.A. Thornton, A.J. McComb, S.O. Ryding, The role of sediments, in: A.J. McComb (Ed.), Eutrophic Shallow Estuaries and Lagoons, CRC Press, Boca Raton, 1975, pp. 205–223. [8] W. Calmano, W. Ahlf, U. Forstner, Exchange of heavy metals between sediment components and water, in: J.A.C. Broekaert, S¸. Güc¸er, F. Adams (Eds.), Metal Speciation in the Environment, NATO ASI Series, vol. G 23, Springer-Verlag, Berlin, 1990, pp. 503–522. [9] G.E. Batley, Trace Element Speciation: Analytical Methods and Problems, CRC Press, Boca Raton, Florida, 1989. [10] B. Gümgüm, E. Ünlü, Z. Tez, Z. Gülsün, Heavy metal pollution in water, sediment and ﬁsh from the Tigris River in Turkey, Chemosphere 29 (1994) 111–116. [11] H. Karadede-Akin, E. Ünlü, Heavy metal concentrations in water, sediment, ﬁsh and some benthic organisms from Tigris River, Turkey, Environ. Monit. Assess. 131 (2007) 323–337. [12] E.N. Akbulut, S. Bayarı, A. Akbulut, Y. S¸ahin, Rivers of Turkey, in: K. Tockner, C.T. Robinson, U. Uehlinger (Eds.), Rivers of Europa, Academic Press, New York, 2009, pp. 643–672. [13] Anonymous, Mean Monthly Flow Values of the Tigris River in the Period 2008–2009, General Directorate of Electrical Power Resources Survey and Development Administration, Ankara, 2009. [14] Anonymous, Mean Monthly Values of Temperature and Rainfall in Maden, Ergani, Diyarbakır, Batman and Cizre in the Period 2008–2009, Turkish State Meteorological Service, Ankara, 2009. [15] USEPA, Method 3051a: Microwave Assisted Acid Dissolution of Sediments, Sludges, Soils, and Oils, Revision 1, United States Environmental Protection Agency, Washington, DC, 2007. [16] S.M. Sakan, D.S. Djordjevic, D.D. Manojlovic, P.S. Polic, Assessment of heavy metal pollutants accumulation in the Tisza river sediments, J. Environ. Manage. 90 (2009) 3382–3390. [17] B. Rubio, M.A. Nombela, F. Vilas, Geochemistry of major and trace elements in sediments of the Ria de Vigo (NW Spain): an assessment of metal pollution, Mar. Pollut. Bull. 40 (2000) 968–980. [18] S. Covelli, G. Fontolan, Application of a normalization procedure in determining regional geochemical baselines, Environ. Geol. 30 (1/2) (1997) 34–45. [19] L. Hakanson, Ecological risk index for aquatic pollution control. A sedimentological approach, Water Res. 14 (1980) 975–1001. [20] D.C. Tomlinson, J.G. Wilson, C.R. Harris, D.W. Jeffery, Problems in the assessment of heavy metals levels in estuaries and the formation of a pollution index, Helgol. Wiss. Meeresunters 33 (1980) 566–575.

364

M. Varol / Journal of Hazardous Materials 195 (2011) 355–364

[21] G. Müller, Die Schwermetallbelstung der sedimente des Neckars und seiner Nebenﬂusse: eine Bestandsaufnahme, Chem. Zeitung 105 (1981) 157–164. [22] M.A.H. Bhuiyan, L. Parvez, M.A. Islam, S.B. Dampare, S. Suzuki, Heavy metal pollution of coal mine-affected agricultural soils in the northern part of Bangladesh, J. Hazard. Mater. 173 (2010) 384–392. [23] D.D. MacDonald, C.G. Ingersoll, T.A. Berger, Development and evaluation of consensus-based sediment quality guidelines for freshwater ecosystems, Arch. Environ. Con. Tox. 39 (2000) 20–31. [24] M. Varol, B. S¸en, Assessment of surface water quality using multivariate statistical techniques: a case study of Behrimaz Stream, Turkey Environ. Monit. Assess. 159 (2009) 543–553. [25] J. Zhang, C.L. Liu, Riverine composition and estuarine geochemistry of particulate metals in China-weathering features, anthropogenic impact and chemical ﬂuxes, Estuar. Coast. Shelf Sci. 54 (2000) 1051–1070. [26] F. Tümen, M. Bildik, M. Baybay, M. Cici, B. Solmaz, Pollution potential of solid wastes from Ergani Copper Plant, Do˘ga Tr. J. Eng. Env. Sci. 16 (1992) 43–55. [27] E. Ünlü, B. Gümgüm, Concentrations of copper and zinc in ﬁsh and sediments from the Tigris River in Turkey, Chemosphere 26 (1993) 2055–2061. [28] B. Gümgüm, G. Öztürk, Chemical speciation of heavy metals in the Tigris River sediment, Chem. Spec. Bioavailab. 13 (2001) 25–29. [29] D. Mendil, Ö.F. Ünal, M. Tüzen, M. Soylak, Determination of trace metals in different ﬁsh species and sediments from the River Yes¸ilırmak in Tokat, Turkey Food Chem. Toxicol. 48 (2010) 1383–1392. [30] A. Farkas, E. Claudio, L. Vigano, Assessment of the environmental signiﬁcance of heavy metal pollution in surﬁcial sediments of the River Po, Chemosphere 68 (2007) 761–768.

[31] S. Olivares-Rieumont, D. de la Rosa, L. Lima, D.W. Graham, K. D’ Alessandro, J. Borroto, F. Martinez, J. Sanchez, Assessment of heavy metal levels in Almendares River sediments-Havana City, Cuba Water Res. 39 (2005) 3945–3953. [32] P. Woitke, J. Wellmitz, D. Helm, P. Kube, P. Lepom, P. Litheraty, Analysis and assessment of heavy metal pollution in suspended solids and sediments of the river Danube, Chemosphere 51 (2003) 633–642. [33] A.P. Karageorgis, N.P. Nikolaidis, H. Karamanos, N. Skoulikidis, Water and sediment quality assessment of the Axios River and its coastal environment, Cont. Shelf Res. 23 (2003) 1929–1944. [34] J. Morillo, J. Usero, I. Gracia, Heavy metal fractionation in sediments from the Tinto River (Spain), Int. J. Environ. Anal. Chem. 82 (2002) 245–257. [35] A.E. Rifaat, Major controls of metals’ distribution in sediments off the Nile Delta Egypt, Egypt J. Aquat. Res. 31 (2005) 16–28. [36] J.S. Heiny, C.M. Tate, Concentration distribution comparison of selected trace elements in bed sediment ﬁsh tissue in the South Platte River Basin U.S.A. 1992–1993, Arch. Environ. Con. Tox 32 (1997) 246–259. [37] K. Hudson-Edwards, M. Macklin, M. Taylor, Historic metal mining inputs to Tees river sediment, Sci. Total Environ. 194/195 (1997) 437–445. [38] W. Mendez, Contamination of Rimac River Basin Peru, due tomining tailings. TRITA-LWR Master Thesis, Environmental Engineering and Sustainable Infrastructure, The Royal Institute of Technology (KTH), Stockholm, 2005. [39] B. Vandecasteele, B. De Vos, F.M.G. Tack, Temporal-spatial trends in heavy metal contents in sediment-derived soils along the Sea Scheldt river (Belgium), Environ. Pollut. 122 (2003) 7–18. [40] J. Liu, Y. Li, B. Zhang, J. Cao, Z. Cao, J. Domagalski, Ecological risk of heavy metals in sediments of the Luan River source water, Ecotoxicology 18 (2009) 748–758.

Journal of Hazardous Materials 195 (2011) 365–370

Contents lists available at SciVerse ScienceDirect

Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat

Molecular mechanism of kidney injury of mice caused by exposure to titanium dioxide nanoparticles Suxing Gui a,1 , Zengli Zhang b,1 , Lei Zheng a,1 , Yaling Cui a,1 , Xiaorun Liu c,d,1 , Na Li a,e , Xuezi Sang a , Qingqing Sun a , Guodong Gao a , Zhe Cheng a , Jie Cheng a , Ling Wang a , Meng Tang c,d,∗ , Fashui Hong a,∗ a

Medical College, Soochow University, Suzhou 215123, People’s Republic of China Public Health School, Soochow University, Suzhou 215123, People’s Republic of China c Key Laboratory of Environmental Medicine and Engineering, Ministry of Education; School of Public Health, Southeast University, Nanjing 210009, China d Jiangsu key Laboratory for Biomaterials and Devices; Southeast University, Nanjing 210009, China e General Hospital of Jincheng Anthracite Mining Group Co. Ltd., Jincheng 048006, People’s Republic of China b

a r t i c l e

i n f o

Article history: Received 30 June 2011 Received in revised form 17 August 2011 Accepted 17 August 2011 Available online 24 August 2011 Keywords: Titanium dioxide nanoparticules Kidney Inﬂammatory response Cytokines

a b s t r a c t Numerous studies have demonstrated that damage of kidney of mice can be caused by exposure to titanium dioxide nanoparticles (TiO2 NPs). However, the molecular mechanism of TiO2 NPs-induced nephric injury remains unclear. In this study, the mechanism of nephric injury in mice induced by an intragastric administration of TiO2 NPs was investigated. The results showed that TiO2 NPs were accumulated in the kidney, resulting in nephric inﬂammation, cell necrosis and dysfunction. Nucleic factor-␬B was activated by TiO2 NPs exposure, promoting the expression levels of tumor necrosis factor-␣, macrophage migration inhibitory factor, interleukin-2, interleukin-4, interleukin-6, interleukin-8, interleukin-10, interleukin18, interleukin-1␤, cross-reaction protein, transforming growth factor-␤, interferon-␥ and CYP1A1, while heat shock protein 70 expression was inhibited. These ﬁndings implied that TiO2 NPs-induced nephric injury of mice might be associated with alteration of inﬂammatory cytokine expression and reduction of detoxiﬁcation of TiO2 NPs. Crown Copyright © 2011 Published by Elsevier B.V. All rights reserved.

1. Introduction In the development of nanotechnology, nanomaterials are recognized to have potential applications due to their larger surface area to volume ratio, which enhances chemical reactivity and easier penetration into cells. Among the various nanomaterials, customarily titanium dioxide nanoparticles (TiO2 NPs) are regarded as chemical inert, nontoxic and biocompatible material [1–3], they have been widely used in the sunscreen ingredient, pharmaceutical, and paint industries as a colouring material [4–7]. In over ten years, however, TiO2 NPs toxicology has attracted considerable attention owing to their small sizes, large surface per mass and high reactivity. A number of investigations have deﬁnitely showed that TiO2 NPs exposure are able to cause injuries in various animal organ types, including lung, liver, spleen, and brain [8–24]. Recently, the toxicity of TiO2 NPs to kidneys has been reported. Scown et al. had found that TiO2 NPs were accumulated in the kidney, but had minimal effects on renal functions in rainbow trout [25]. In contradiction, Wang et al. had observed that TiO2 NPs exposure to mice resulted

in higher blood urea nitrogen and creatinine levels and the renal tubule was ﬁlled with proteinic liquids [10]. Chen et al. had also observed renal glomerulus dilatation and proteinic liquids ﬁlled renal tubule, but no kidney dysfunction was found with TiO2 NPstreated mice [26]. Furthermore, TiO2 NPs were also suggested to induce nephric inﬂammation and impair nephric functions, which exerted its toxicity through ROS accumulation [27]. However, the molecular mechanism of TiO2 NPs-induced nephric inﬂammation remains unclear. Nephric inﬂammation and dysfunction are due to altered in kidney regardless of the cause of these diseases. Thus TiO2 NPs induced nephric inﬂammation and dysfunction are able to be monitored through inﬂammatory cytokine expression levels in kidney. To conﬁrm the above hypothesis, mice were continuously exposed to TiO2 NPs for 90 days by an intragastric administration. The inﬂammatory cytokine expression in the mouse kidney was determined and the possible mechanism of the TiO2 NPs induced nephric pathogenesis in mice was discussed. 2. Materials and methods 2.1. Chemicals, preparation and characterization

∗ Corresponding authors. Tel.: +86 0512 61117563; fax: +86 0512 65880103. E-mail addresses: [email protected] (M. Tang), Hongfsh [email protected] (F. Hong). 1 Contributed equally to this work.

Nanoparticulated anatase TiO2 was prepared via controlled hydrolysis of titanium tetrabutoxide. The details of the synthesis

0304-3894/$ – see front matter. Crown Copyright © 2011 Published by Elsevier B.V. All rights reserved. doi:10.1016/j.jhazmat.2011.08.055

366

S. Gui et al. / Journal of Hazardous Materials 195 (2011) 365–370

and characterization TiO2 NP were previously described by our previous reports [21,28]. The average particle sizes of powder suspended in 0.5% w/v hydroxypropylmethylcellulose K4M (HPMC, K4M) solvent after 12 h and 24 h incubation ranged from 5 to 6 nm. The mean hydrodynamic diameter of TiO2 NPs in HPMC solvent ranged between 208 and 330 nm (mostly 294 nm), and the zeta potential after 12 h and 24 h incubation was 7.57 mV and 9.28 mV [21]. 2.2. Animal and treatment It has been previously demonstrated by Wang et al. that sensitivity to TiO2 exposure was higher in CD-1 (ICR) female mice than CD-1 (ICR) male mice [1]. Therefore, CD-1 (ICR) female mice were used in this study. 80 CD-1 (ICR) female mice (24 ± 2 g) were purchased from the Animal Center of Soochow University (China). All mice were housed in stainless steel cages in a ventilated animal room. Room temperature of the housing facility was maintained at 24 ± 2 ◦ C with a relative humidity of 60 ± 10% and a 12-h light/dark cycle. Distilled water and sterilized food were available for mice ad libitum. Prior to dosing, the mice were acclimated to this environment for 5 days. All animals were handled in accordance with the guidelines and protocols approved by the Care and Use of Animals Committee of Soochow University (China). All procedures used in animal experiments conformed to the U.S. National Institutes of Health Guide for the Care and Use of Laboratory Animals [29]. The mice were randomly divided into four groups (N = 20), including a control group treated with 0.5% w/v HPMC and three experimental groups treated with 2.5, 5, and 10 mg/kg BW TiO2 NPs, respectively). The mice were weighed, and the TiO2 NP suspensions were administered to the mice by an intragastric administration every day for 90 days. Any symptom or mortality was observed and recorded carefully everyday during the 90 days. After 90 days, all mice were weighed ﬁrstly, and then sacriﬁced after being anesthetized using ether. Blood samples were collected from the eye vein by removing the eyeball quickly. Serum was collected by centrifuging blood at 2500 rpm for 10 min. Kidneys were collected and weighed. 2.3. Coefﬁcient of kidney After weighing the body and kidneys, the coefﬁcient of kidney to body weight was calculated as the ratio of kidney (wet weight, mg) to body weight (g). 2.4. Titanium content analysis Kidneys were removed from the −80 ◦ C and then thawed, and roughly 0.3 g of the kidney was weighed, digested and analyzed for titanium content. Inductively coupled plasma-mass spectrometry (ICP-MS, Thermo Elemental X7, Thermo Electron Company) was used to analyze the titanium concentration in the samples. For the analysis, an Indium concentration of 20 ng/mL was utilized as an internal standard element, and the detection limit of titanium was 0.074 ng/mL. The data were expressed as nanograms per gram fresh tissue. 2.5. Biochemical analysis of kidney functions Kidney functions were determined by uric acid (UA), blood urea nitrogen (BUN), creatinine (Cr), calcium (Ca) and phosphonium (P). All biochemical assays were performed using a clinical automatic chemistry analyzer (Type 7170A, Hitachi, Japan).

2.6. Histopathological examination of kidney For pathological studies, all histopathological tests were performed using standard laboratory procedures [30]. The kidneys were embedded in parafﬁn blocks, then sliced into 5 ␮m in thickness and placed onto glass slides. After hematoxylin–eosin (HE) staining, the slides were observed and the photos were taken using an optical microscope (Nikon U-III Multi-point Sensor System, USA), and the identity and analysis of the pathology slides were blind to the pathologist. 2.7. Expression assay of inﬂammatory cytokines The level of mRNA expression of nucleic factor-␬B (NF-␬B), NF-␬B-inhibiting factor (I␬B), tumor necrosis factor-␣ (TNF-␣), macrophage migration inhibitory factor (MIF), interleukin-2 (IL2), interleukin-4 (IL-4), interleukin-6 (IL-6), interleukin-8 (IL-8), interleukin-10 (IL-10), interleukin-18 (IL-18), interleukin-1␤ (IL1␤), cross-reaction protein (CRP), transforming growth factor-␤ (TGF-␤), interferon-␥ (INF-␥), cytochrome p450 1A (CYP1A) and heat shock protein 70 (HSP70) in the mouse kidney was determined using real-time quantitative RT polymerase chain reaction (RT-PCR) [31–33], respectively. Synthesized cDNA was used for the real-time PCR by employing primers that were designed using Primer Express Software according to the software guidelines, and PCR primer sequences are available upon request. To determine NF␬B, I␬B, TNF-␣, MIF, IL-2, IL-4, IL-6, IL-8, IL-10, IL-18, IL-6, IL-1␤, CRP, TGF-␤, INF-␥, Bax, Bcl-2, CYP1A1 and HSP-70 levels in the mouse kidney, an enzyme linked immunosorbent assay (ELISA) was performed using commercial kits that are selective for each respective protein (R&D Systems, USA). Manufacturer’s instruction was followed. The absorbance was measured on a microplate reader at 450 nm (Varioskan Flash, Thermo Electron, Finland), and the concentrations of NF-␬B, I␬B, TNF-␣, MIF, IL-2, IL-4, IL-6, IL-8, IL-10, IL-18, IL-6, IL-1␤, CRP, TGF-␤, INF-␥, CYP1A1 and HSP-70 were calculated from a standard curve for each sample. 2.8. Statistical analysis Statistical analyses were conducted using SPSS 11.7 software. Data were expressed as means ± standard deviation (SD). One-way analysis of variance (ANOVA) was carried out to compare the differences of means among multi-group data. Dunnett’s test was performed when each dataset was compared with the solventcontrol data. Statistical signiﬁcance for all tests was judged at a probability level of 0.05 (p < 0.05). 3. Results 3.1. Coefﬁcient of kidney and titanium accumulation Signiﬁcant increases of the coefﬁcients of kidney (p < 0.05 or p < 0.01) were caused by TiO2 NPs exposure for consecutive 90 days (Fig. 1). Furthermore, with increasing TiO2 NPs dose, the obvious accumulation of titanium in the kidney occurred (p < 0.01) (Fig. 2). These results show that the accumulation of titanium in the kidney was associated with the coefﬁcients of kidney of mice. The increase of kidney indices caused by TiO2 NPs exposure may be related to the nephric dysfunction and tissue injury, which are conﬁrmed by the further assays of biochemical parameters and histopathological observation of kidney of mice. 3.2. Biochemical parameters in serum of kidney The changes of biochemical parameters in the blood serum of mice kidney caused by TiO2 NPs exposure are presented in Table 1.

S. Gui et al. / Journal of Hazardous Materials 195 (2011) 365–370

367

Table 1 The changes of biochemical parameters in the blood serum of mice after intragastric administration with TiO2 NPs for 90 days. Indexes

TiO2 NPs (mg/kg, BW) 0

UA (␮mol/L) Cr (␮mol/L) BUN (mmol/L) Ca (mmol/L) P (mmol/L)

222.56 8.81 9.28 2.43 3.28

2.5 ± ± ± ± ±

11.13 0.44 0.46 0.12 0.16

160.21 9.75 8.11 2.48 3.33

5 ± ± ± ± ±

8.01* 0.49* 0.41* 0.12 0.16

110.88 11.68 7.05 2.51 3.46

10 ± ± ± ± ±

5.54** 0.58** 0.35** 0.13 0.17

96.76 13.19 6.32 2.71 3.52

± ± ± ± ±

4.84** 0.66** 0.32** 0.14 0.18

Ranks marked with an asterisk or double asterisks means it is statistically signiﬁcant different from the control (unexposed mice) at the 5% or 1% conﬁdence level, respectively. Values represent means ± SD, N = 10.

With TiO2 NPs dose increased, the contents of Ca and P of nephric function parameters were not signiﬁcant compared with the control group (p > 0.05). However, the Cr was increased, and the UA, and BUN were decreased gradually (p < 0.05 or p < 0.01), respectively; demonstrating that long-term exposure to low dose TiO2 NPs impaired nephric functions in mice. 3.3. Histopathological evaluation of kidney Fig. 3 presents the histopathological changes of kidneys in mice treated by TiO2 NPs exposure. In the 2.5 mg/kg BW TiO2 NPs treated group, the nephric tissue is signiﬁcantly showed to inﬂammatory cell inﬁltration and congestion of mesenchyme blood vessel (Fig. 3b). In the 5 mg/kg BW TiO2 NPs treated group, inﬂammatory cell inﬁltration, congestion of mesenchyme blood vessel and spotty necrosis of renal tubular epithelial cells were observed (Fig. 3c). Furthermore, a large area of necrosis of renal tubular epithelial cells

was detected in the 10 mg/kg BW TiO2 NPs treated group (Fig. 3d). The ﬁndings indicate that the kidney injury was related to a dosedependent manner of TiO2 NPs exposure. 3.4. Cytokine expression To further conﬁrm the role of molecular mechanism in the TiO2 NPs-induced kidney injury, the changes of the inﬂammationrelated genes or detoxiﬁcation-related genes and their proteins expression in mice caused by TiO2 NP exposure were detect using real time RT-PCR and ELISA (Tables 2 and 3). The mRNA expression levels of NF-␬B, TNF-␣, MIF, IL-2, IL-4, IL-6, IL-8, IL-10, IL-18, IL1␤, CRP, TGF-␤, INF-␥, and CYP1A1 were increased signiﬁcantly in the TiO2 NP treated groups (p < 0.05 or 0.01). Interestingly, I␬B and HSP-70 expression levels were decreased signiﬁcantly compared with control group (p < 0.05 or 0.01). 4. Discussion

Fig. 1. The coefﬁcients of kidney of mice by an intragastric administration with TiO2 NPs for consecutive 90 days. Bars marked with an asterisk or double asterisks means it is signiﬁcantly different from the control (unexposed mice) at the 5% or 1% conﬁdence level, respectively. Values represent means ± SD, N = 20.

Fig. 2. The contents of titanium in the mouse kidney by an intragastric administration with TiO2 NPs for 90 days. Bars marked with an asterisk or double asterisks means it is statistically signiﬁcant different from the control (unexposed mice) at the 5% or 1% conﬁdence level, respectively. Values represent means ± SD, N = 5.

In this study, effects of TiO2 NPs on the mouse kidney were studied. After intragastric administrations with 2.5, 5, and 10 mg/kg BW of TiO2 NPs for 90 consecutive days, signiﬁcant increases of the kidney indices (Fig. 1) and titanium accumulation in mouse kidneys (Fig. 2) were observed, coupled with increase of Cr level, decrease of BUN, UA excretion (Table 1), induced inﬂammatory response and necrosis of kidneys (Fig. 3). Previous study indicated that abnormal pathological changes of the mouse kidney and the nephric dysfunction were not able to be triggered by intraperitoneal injection with 5 mg/kg BW TiO2 NPs for 14 days, but with 50, 100 and 150 mg/kg BW TiO2 NPs exposure, impairment of kidney functions and severe inﬂammatory response of kidney were observed [27]. Wang et al. also observed that the 2-week exposure to the 5 g/kg BW TiO2 NPs by a gavage caused nephric dysfunction and tissue damage of mice [10]. In this study, molecular evidences were provided to prove TiO2 NPs induced nephric dysfunction and inﬂammation of mice by alteration of gene expression levels of the cytokines involved in inﬂammatory response or detoxiﬁcation. NF-␬B is known as a critical intracellular mediator of the inﬂammatory cascade, and it binds to inhibitory proteins (I␬Bs) which prevent NF-␬B from migrating to the nucleus from cytoplasm. When an appropriate inducer existed, I␬Bs are phosphorylated and degraded, allowing nuclear uptaking of NF-␬B and initiating gene transcriptions, including MIF, the proinﬂammatory cytokines of TNF-␣, IL-1␤, IL6, IL-8, IL-18, CRP, and anti-inﬂammatory cytokines of IL-2, IL-4, and IL-10 [34]. TGF-␤ is proved to be involved with a dual-role as an anti-inﬂammatory and a proﬁbrotic cytokine. IFN-␥ and TNF are essential for primary defense against infection [35,36], and mice that lack these two cytokines or their cognate receptors succumb to infection rapidly [37]. In response to TiO2 NPs stimulation, our results suggested that TiO2 NPs exposure for 90 consecutive days could signiﬁcantly up-regulate mRNA expression levels of several relative inﬂammatory cytokines genes, including NF-␬B, TNF-␣, MIF, IL-2, IL-4, IL-6, IL-8, IL-10, IL-18, IL-1␤, CRP, TGF-␤, and

368

S. Gui et al. / Journal of Hazardous Materials 195 (2011) 365–370

Table 2 Effect of TiO2 NPs on the ampliﬁcation of regulating the inﬂammation, immune and oxidative stress gene mRNA of the mouse kidney by real-time PCR analysis after intragastric administration with TiQ2 NPs for consecutive 90 days. Ratio of gene/actin

TiO2 NPs (mg/kg, BW) 0

NF-B/actin IB/actin TNF-␣/actin MIF/actin IL-2/actin IL-4/actin IL-6/actin IL-8/actin IL-10/actin IL-18/actin IL-1ˇ/actin CRP/actin TGF-ˇ/actin INF-/actin CYP1A1/actin HSP-70/actin

0.30 0.71 0.08 0.21 0.06 0.07 0.09 0.15 0.12 0.28 0.21 0.42 0.26 0.20 0.28 0.41

2.5 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ±

0.015 0.036 0.004 0.011 0.003 0.004 0.005 0.008 0.006 0.014 0.011 0.021 0.013 0.010 0.014 0.021

5

0.36 ± 0.0l8 0.58 ± 0.029* 0.11 ± 0.006* 0.32 ± 0.016* 0.09 ± 0.005* 0.09 ± 0.005* 0.13 ± 0.007* 0.19 ± 0.010* 0.17 ± 0.009* 0.31 ± 0.016 0.33 ± 0.017** 0.53 ± 0.027* 0.38 ± 0.019* 0.26 ± 0.013* 0.36 ± 0.018* 0.32 ± 0.016* *

0.57 0.42 0.18 0.49 0.15 0.16 0.18 0.31 0.24 0.46 0.42 0.65 0.53 0.41 0.66 0.25

10 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ±

**

0.029 0.021** 0.009** 0.025* 0.008** 0.008** 0.009** 0.016** 0.012** 0.023** 0.021** 0.033** 0.027** 0.021** 0.033** 0.013**

0.88 0.32 0.31 0.76 0.23 0.27 0.31 0.51 0.38 0.61 0.58 0.88 0.70 0.58 1.01 0.11

± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ±

0.044** 0.016** 0.016** 0.038** 0.012** 0.014** 0.016** 0.026** 0.019** 0.031** 0.029** 0.044** 0.035** 0.029** 0.051** 0.006**

Ranks marked with an asterisk or double asterisks means it is statistically signiﬁcant different from the control (unexposed mice) at the 5% or l% conﬁdence level, respectively. Values represent means ± SD, N = 5.

INF-␥, and decrease I␬B expression. The obvious alterations of these cytokines’ expression indicated the involvement of inﬂammatory responses in TiO2 NPs-induced kidney toxicity. Studies had showed that TiO2 NPs promoted the expression of inﬂammatory cytokines in the lung, liver, spleen and brain of rat and mice [9,12,38–41]. Increases of NF-␬B expression in the mouse liver were also detected due to the signiﬁcant increases of NF-␬B-inducible kinase and I␬B kinase expression and decrease of I␬B expression after treated with TiO2 NPs for 60 days [17]. In this study, signiﬁcant increase of the CYP1A expression and reduction of HSP70 expression were observed (Tables 2 and 3). CYP1A and HSP70 were selected since they represent different processes that the cells follow to detoxify and/or defend against environmental toxicants [42]. Differences of gene expression of CYP1A and HSP70 were then used to explain the toxic characteristic

signatures of TiO2 NPs. It is well known that CYP1A induction is activated by the aryl hydrocarbon receptor (AHR) pathway, and its protein plays an essential function in the biotransformation and detoxiﬁcation of endogenous and exogenous compounds. It is a widely accepted environmental biomarker, useful for monitoring the biological effects of several xenobiotic groups, including heavy metals [42]. De Jongh et al. showed that administration of 2,3,7,8tetrachlorodibenzo-p-dioxin (TCDD) to male C57BL/6J mice had caused the increases of both CYP 1A1 and CYP 1A2 hepatic protein levels [43]. In this study the high level expression of this gene and its protein products indicated that TiO2 NPs may cause kidney intoxication in mice. Likewise, higher level expression of HSP70 is often associated with a cellular response to a harmful stress or to adverse life conditions. The reduction of the HSP70 expression in the kidney by exposure to TiO2 NPs indicated a slow biotransformation or

Fig. 3. Histopathological observation of kidney caused by an intragastric administration with TiO2 NPs for consecutive 90 days. (a) Control group (unexposed mice) presents integrated glomerulars and normal kidney tubulars; (b) 2.5 mg/kg TiO2 NPs group presents inﬂammatory cell inﬁltration (yellow cycle) and congestion of mesenchyme blood vessel (blue arrow); (c) 5 mg/kg TiO2 NPs group indicates inﬂammatory cell inﬁltration (yellow cycle), congestion of mesenchyme blood vessel (blue arrow) and spotty necrosis of renal tubular epithelial cell (green cycle); (d) 10 mg/kg TiO2 NPs group indicates severe necrosis of renal tubular epithelial cell (green cycles). (For interpretation of the references to color in this ﬁgure legend, the reader is referred to the web version of the article.) The scale bar presented at the upside of each photomicrograph indicated 100 ␮m.

S. Gui et al. / Journal of Hazardous Materials 195 (2011) 365–370

369

Table 3 Effect of TiO NPs on the inﬂammatory cytokine protein levels of the mouse kidhey by ELISA analysis after intragastric administration with TiO NPs for consecutive 90 days. Protein expression

TiO2 NPs (mg/kg, BW) 0

NF-␬B (ng/g tissue) I␬B (g tissue) TNF-␣ (ng/g tissue) MIF (ng/g tissue) IL-2 (ng/g tissue) IL-4 (ng/g tissue) IL-6 (ng/g tissue) IL-8 (ng/g tissue) IL-10 (ng/g tissue) IL-18 (ng/g tissue) IL-1␤ (ng/g tissue) CRP (␮g/g tissue) TGF-␤ (ng/g tissue) TGF-␥ (ng/g tissue) CYP1A1 (ng/g tissue) HSP-70 (ng/g tissue)

34.62 18.71 72.83 269 66.45 44.99 6.95 32.93 5.96 6.17 88.94 38.68 21.69 19.50 12.02 11.95

2.5 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ±

1.73 0.94 3.64 13 3.32 2.25 0.35 1.65 0.30 0.31 4.45 1.93 1.08 0.98 0.60 0.60

38.13 14.26 81.66 582 73.28 49.39 8.23 38.99 7.12 8.29 105.77 53.91 34.71 28.57 21.17 7.14

5 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ±

1.91 0.71* 4.08* 29** 3.66* 2.47 0.41* 1.95* 0.36* 0.41* 5.29* 2.70* 1.74** 1.43** 1.06** 0.36*

52.95 10.77 171.26 2749 87.39 57.78 9.67 42.98 8.35 12.25 168.81 74.85 51.99 47.92 36.59 5.56

10 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ±

2.65** 0.54** 8.56** 137** 4.37** 2.89* 0.48** 2.15** 0.42** 0.61** 8.44** 3.74** 2.60** 2.40** 1.83** 0.28**

89.96 7.86 327.79 3129 94.46 71.19 13.99 53.37 10.66 19.99 196.42 95.93 72.48 63.81 52.88 3.78

± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ±

4.50** 0.39** 16.39** 156** 4.72** 3.56** 0.70** 2.67** 0.53** 1.00** 9.82** 4.80** 3.62** 3.19** 2.64** 0.19**

Ranks marked with an asterisk or double asterisks means it is statistically signiﬁcant different from the control (unexposed mice) at the 5% or l% conﬁdence level, respectively. Values represent means ± SD, N = 5.

detoxiﬁcation and decreased response to the adverse effects experienced in the kidney [44–47]. About the dose selection in this study, we consulted the report of World Health Organization in 1969. According to the report, LD50 of TiO2 for rats is larger than 12,000 mg/kg BW after oral administration. In the present study, we selected 5, 10, and 50 mg/kg BW TiO2 NPs exposed to mice every day. They were equal to about 0.15–0.7 g TiO2 NPs of 60–70 kg body weight for humans with such exposure, which were relatively safe doses. However, we think, attention should be aroused on the application of TiO2 NPs and their potential long-term exposure effects especially on human beings. In conclusion, the present study shows that mice treated with 2.5, 5 and 10 mg/kg BW TiO2 NPs for 90 consecutive days resulted in signiﬁcant increases of NF-␬B, TNF-␣, MIF, IL-2, IL-4, IL-6, IL-8, IL-10, IL-18, IL-1␤, CRP, TGF-␤, INF-␥, CYP1A expression and signiﬁcant decrease of HSP70 expression, leading to the increase of kidney indices, inﬂammatory responses and cell necrosis in mouse kidney.

Acknowledgements This work was supported by a Project Funded by the Priority Academic Program Development of Jiangsu Higher Education Institutions, the National Natural Science Foundation of China (grant No. 30901218), the Major State Basic Research Development Program of China (973 Program) (grant No. 2006CB705602), National Important Project on Scientiﬁc Research of China (grant No. 2011CB933404), National Natural Science Foundation of China (grant No. 30671782, 30972504). References [1] R.W. Tennant, B.H. Margolin, M.D. Shelby, E. Zeiger, J.K. Haseman, J. Spalding, et al., Prediction of chemical carcinogenicity in rodents from in vitro genetic toxicity assay, Science 236 (1987) 933–941. [2] S. Huda, S.K. Smoukov, H. Nakanishi, B. Kowalczyk, K. Bishop, B.A. Grzybowski, Antibacterial nanoparticle monolayers prepared on chemically inert surfaces by cooperative electrostatic adsorption (CELA), Appl. Mater. Interfaces 2 (2010) 1206–1210. [3] J.F. Li, Y.F. Huang, Y. Ding, Z.L. Yang, S.B. Li, X.S. Zhou, et al., Shell-isolated nanoparticle-enhanced raman spectroscopy, Nature 464 (2010) 392–395. [4] M. Hedenborg, Titanium dioxide induced chemiluminescence of human polymorphonuclear leukocytes, Int. Arch. Occup. Environ. Health 61 (1988) 1–6. [5] C. Gelis, S. Girard, A. Mavon, M. Delverdier, N. Paillous, P. Vicendo, Assessment of the skin photoprotective capacities of an organo-mineral broad-spectrum sunblock on two ex vivo skin models, Photoimmunol. Photomed. 19 (2003) 242–253.

[6] K. Kakinoki, K. Yamane, M. Igarashi, M. Yamamoto, R. Teraoka, Y. Matsuda, Evaluation of titanium dioxide as a pharmaceutical excipient for preformulation of a photo-labile drug: effect of physicochemical properties on the photostability of solid-state nisoldipine, Chem. Pharm. Bull. 53 (2005) 811–815. [7] J. Gurr, C. Ass Wang, K.J. Chen, Ultraﬁne titanium dioxide particle in the absence of photoactivation can induce oxidative damage to human bronchial epithelial cells, Toxicology 213 (2005) 66–73. [8] F. Afaq, P. Abidi, R. Matin, Q. Rahman, Cytotoxicity, pro-oxidant effects and antioxidant depletion in rat lung alveolar macrophages exposed to ultraﬁne titanium dioxide, J. Appl. Toxicol. 18 (1998) 307–312. [9] H.W. Chen, S.F. Su, C.T. Chien, W.H. Lin, S.L. Yu, C.C. Chou, et al., Titanium dioxide nanoparticles induce emphysema-like lung injury in mice, FASEB J. 20 (2006) 1732–1741. [10] J.X. Wang, G.Q. Zhou, C.Y. Chen, H.W. Yu, T.C. Wang, Y.M. Ma, et al., Acute toxicity and biodistribution of different sized titanium dioxide particles in mice after oral administration, Toxicol. Lett. 168 (2007) 176–185. [11] H.T. Liu, L.L. Ma, J.F. Zhao, J. Liu, J.Y. Yan, J. Ruan, et al., Biochemical toxicity of mice caused by nano-anatase TiO2 particles, Biol. Trace Elem. Res. 129 (2009) 170–180. [12] L.L. Ma, J.F. Zhao, J. Wang, Y.M. Duan, J. Liu, N. Li, et al., The acute liver injury in mice caused by nano-anatase TiO2 , Nanosacle Res. Lett. 4 (2009) 1275–2128. [13] Y.M. Duan, J. Liu, L.L. Ma, N. Li, H.T. Liu, J. Wang, et al., Toxicological characteristics of nanoparticulate anatase titanium dioxide in mice, Biomaterials 31 (2010) 894–899. [14] N. Li, L.L. Ma, J. Wang, J. Liu, Y.M. Duan, H.T. Liu, et al., Interaction between nano-anatase TiO2 and liver DNA from mice in vivo, Nanoscale Res. Lett. 5 (2010) 108–115. [15] Y.L. Cui, X.L. Gong, Y.M. Duan, N. Li, R.P. Hu, H.T. Liu, et al., Hepatocyte apoptosis and its molecular mechanisms in mice caused by titanium dioxide nanoparticles, J. Hazard. Mater. 183 (2010) 874–880. [16] H.T. Liu, L.L. Ma, J. Liu, J.F. Zhao, J.Y. Yan, F.S. Hong, Toxicity of nano-anatase TiO2 to mice: liver injury, oxidative stress, Toxicol. Environ. Chem. 92 (2010) 175–186. [17] Y.L. Cui, H.T. Liu, M. Zhou, Y.M. Duan, N. Li, X.L. Gong, et al., Signaling pathway of inﬂammatory responses in the mouse liver caused by TiO2 nanoparticules, J. Biomed. Mater. Res. A 96 (2011) 221–229. [18] N. Li, Y.M. Duan, M.M. Hong, L. Zheng, M. Fei, X.Y. Zhao, et al., Spleen injury and apoptotic pathway in mice caused by titanium dioxide nanoparticules, Toxicol. Lett. 195 (2010) 161–168. [19] L.L. Ma, J. Liu, N. Li, J. Wang, Y.M.i Duan, J.Y. Yan, et al., Oxidative stress in the brain of mice caused by translocated nanoparticulate TiO2 delivered to the abdominal cavity, Biomaterials 31 (2010) 99–105. [20] R.P. Hu, X.L. Gong, Y.M. Duan, N. Li, Y. Che, Y.L. Cui, et al., Neurotoxicological effects and the impairment of spatial recognition memory in mice caused by exposure to TiO2 nanoparticles, Biomaterials 31 (2010) 8043–8050. [21] R.P. Hu, L. Zheng, T. Zhang, Y.L. Cui, G.D. Gao, Z. Cheng, et al., Molecular mechanism of hippocampal apoptosis of mice following exposure to titanium dioxide nanoparticles, J. Hazard. Mater. 191 (2011) 32–40. [22] J. Wu, W. Liu, C.B. Xue, S.C. Zhou, F.L. Lan, L. Bi, Toxicity and penetration of TiO2 nanoparticles in hairless mice and porcine skin after subchronic dermal exposure, Toxicol. Lett. 191 (2009) 1–8. [23] J.A. Shin, E.J. Lee, S.M. Seo, H.S. Kim, J.L. Kang, E.M. Park, Nanosized titanium dioxide enhanced inﬂammatory responses in the septic brain of mouse, Neuroscience 165 (2010) 445–454. [24] Y. Yu, W. Ren, B. Ren, Nanosize titanium dioxide cause neuronal apoptosis: a potential linkage between nanoparticle exposure and neural disorder, Neurol. Res. (2008), doi: 10. 1179/17431 3208X 305391.

370

S. Gui et al. / Journal of Hazardous Materials 195 (2011) 365–370

[25] T.M. Scown, R.Y. van Aerle, B.D. Johnston, S. Cumberland, J.R. Lead, O. Richard, et al., High doses of intravenously administered titanium dioxide nanoparticles accumulate in the kidneys of rainbow trout but with no observable impairment of renal function, Toxicol. Sci. 109 (2009) 372–380. [26] J.Y. Chen, X. Dong, J. Zhao, G.P. Tang, In vivo acute toxicity of titanium dioxide nanoparticles to mice after intraperitoneal injection, J. Appl. Toxicol. 29 (2009) 330–337. [27] J.F. Zhao, J. Wang, S.S. Wang, X.Y. Zhao, J.Y. Yan, J. Ruan, et al., The mechanism of oxidative damage in nephrotoxicity of mice caused by nano-anatase TiO2 , J. Exp. Nanosci. 5 (2010) 447–462. [28] P. Yang, C. Lu, N. Hua, Y. Du, Titanium dioxide nanoparticles codoped with Fe3+ and Eu3+ ions for photocatalysis, Mater. Lett. 57 (2002) 794–801. [29] National Institutes of Health (NIH), Guide for the Care and Use of Laboratory Animals, National Academy Press, Washington D.C., 1996. [30] A.B. Tarabishy, B. Aldabagh, Y. Sun, Y. Imamura, P.K. Mukherjee, J.H. Lass, et al., MyD88 regulation of Fusarium keratitis is dependent on TLR4 and IL-1R1 but not TLR2, J. Immunol. 181 (2008) 593–600. [31] K.J. Livak, T.D. Schmittgen, Analysis of relative gene expression data using realtime quantitative PCR and the 2(-delta delta C(T)) method, Methods 25 (2001) 402–408. [32] W.H. Liu, D.A. Saint, Validation of a quantitative method for real time PCR kinetics, Biochem. Biophys. Res. Commun. 294 (2002) 347–353. [33] L.D. Ke, Z. Chen, A reliability test of standard-based quantitative PCR: exogenous vs. endogenous standards, Mol. Cell. Probes 14 (2000) 127–135. [34] S. Ghosh, M.J. May, E.B. Kopp, NF-␬B and Rel proteins: evolutionarily conserved mediators of immune responses, Annul. Rev. Immunol. 16 (1998) 225–260. [35] N.A. Buchmeier, R.D. Schreiber, Requirement of endogenous interferon-gamma production for resolution of listeria monocytogenes infection, Proc. Natl. Acad. Sci. U.S.A. 82 (1985) 7404–7408. [36] E.A. Havell, Evidence that tumor necrosis factor has an important role in antibacterial resistance, J. Immunol. 143 (1989) 2894–2899.

[37] J.T. Harty, M.J. Bevan, Speciﬁc immunity to Listeria monocytogenes in the absence of IFN-␥, Immunity 3 (1995) 109–117. [38] D. Hohr, Y. Steinfartz, R.P. Schins, A.M. Knaapen, G. Martra, B. Fubini, et al., The surface area rather than the surface coating determines the acute inﬂammatory response after instillation of ﬁne and ultraﬁne TiO2 in the rat, Int. J. Hyg. Environ. Health 205 (2002) 239–244. [39] J.X. Wang, C.Y. Chen, Y. Liu, F. Jiao, W. Li, F. Lao, et al., Potential neurological lesion after nasal instillation of TiO2 nanoparticles in the anatase and rutile crystal phases, Toxicol. Lett. 183 (2008) 72–80. [40] J.X. Wang, Y. Liu, F. Jiao, F. Lao, W. Li, Y.Q. Gu, et al., Time-dependent translocation and potential impairment on central nervous system by intranasally instilled TiO2 nanoparticles, Toxicology 254 (2008) 82–90. [41] J. Wang, N. Li, L. Zheng, Y. Wang, Y.M. Duan, S.S. Wang, et al., P38-Nrf-2 signaling pathway of oxidative stress in mice caused by nanoparticulate TiO2 , Biol. Trace Elem. Res. 140 (2011) 186–197. [42] N.A. Lewis, T.D. Williams, J.K. Chipman, Functional analysis of a metal response element in the regulatory region of ﬂounder cytochrome P450 1A and implications for environmental monitoring of pollutants, Toxicol. Sci. 92 (2006) 387–393. [43] J. De Jongh, M. DeVito, R. Nieboer, L. Birnbaum, M. Van den Berg, Induction of cytochrome P450 isoenzymes after toxico-kinetic interactions between 2, 3,7,8-tetrachlorodibenzo-p-di-oxin and 2,2 ,4,4 ,5,5 -hexachlorobiphenyl in the liver of the mouse, Fund. Appl. Toxicol. 25 (1995) 264–270. [44] M.P. Mayer, B. Bukau, Hsp70 chaperones: cellular functions and molecular mechanism, Cell. Mol. Life Sci. 62 (2005) 670–684. [45] H.M. Beere, B.B. Wolf, K. Cain, D.D. Mosser, A. Mahboubi, T. Kuwana, et al., Heatshock protein 70 inhibits apoptosis by preventing recruitment of procaspase-9 to the Apaf-1 apoptosome, Nat. Cell Biol. 2 (2000) 469–475. [46] H.M. Li, T. Niki, T. Taira, S.M. Iguchi-Ariga, H. Ariga, Association of DJ-1 with chaperones and enhanced association and colocalization with mitochondrial Hsp70 by oxidative stress, Free Radic. Res. 39 (2005) 1091–1099. [47] H.N. Hong, H.N. Kim, K.S. Park, S.K. Lee, M.B. Gu, Analysis of the effects diclofenac has on Japanese medaka (Oryzias latipes) using real-time PCR, Chemosphere 67 (2007) 2115–2121.

Journal of Hazardous Materials 195 (2011) 371–377

Contents lists available at SciVerse ScienceDirect

Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat

Ozonation kinetics for the degradation of phthalate esters in water and the reduction of toxicity in the process of O3 /H2 O2 Gang Wen a , Jun Ma a,∗ , Zheng-Qian Liu b,∗∗ , Lei Zhao c a b c

State Key Laboratory of Urban Water Resource and Environment, Harbin Institute of Technology, Harbin 150090, People’s Republic of China School of Environmental Science and Engineering, Huazhong University of Science and Technology, Wuhan 430074, People’s Republic of China School of Civil Engineering, Harbin Institute of Technology, Harbin 150090, People’s Republic of China

a r t i c l e

i n f o

Article history: Received 12 May 2011 Received in revised form 11 July 2011 Accepted 18 August 2011 Available online 24 August 2011 Keywords: Phthalate esters Rate constant Ozone (O3 ) Hydroxyl radical (• OH) Hydroxyl radical/ozone ratio (Rct ) Toxicity assessment

a b s t r a c t The oxidation kinetics of four phthalate esters (PAEs) with ozone alone and hydroxyl radical (• OH) were investigated. The toxicity reduction in the process of O3 /H2 O2 was evaluated. The second order rate constants for the reaction of four PAEs with ozone and • OH were determined by direct oxidation method and competition kinetics method in bench-scale experiment, and found to be 0.06–0.1 M−1 s−1 and (3–5) × 109 M−1 s−1 , respectively. The oxidation kinetic rate constant of the selected PAEs (diethyl phthalate, DEP) was conﬁrmed using Song Hua-jiang river water as the background. The results indicated that DEP degradation in this river water was close to the simulated value based on the determined rate constants. The toxicity test performed with bioluminescence test, showed that the toxicity expressed as the inhibition rate changed from 36% to below detection limit in the process of O3 /H2 O2 , which means that catalytic ozonation is an efﬁcient way for DEP degradation and toxicity reduction, but an ineffective method for DEP minimization on the basis of the total organic carbon determination. © 2011 Elsevier B.V. All rights reserved.

1. Introduction Phthalate esters (PAEs) are frequently used as plasticizers for cellulosic and vinyl ester resins to improve their ﬂexibility and softness, and also in ceramic, paper, cosmetic and paint industries [1]. The production of PAEs has reached 3.5 million tons per year [2]. PAEs have been detected in surface and groundwater in ng L−1 –mg L−1 concentration range and associated with birth defects, organ damage, infertility, as well as testicular cancer, and are also known to be among the major endocrine disrupter chemicals (EDCs) [3,4]. Recently, it has been revealed that di-butyl phthalate (DBP) exhibits antagonistic thyroid receptor activity [5]. Drinking water treatment plant is the most important barrier to prevent the organic matters from human being contact. Previous investigation on 13 EDCs removal from traditional waterworks in China has demonstrated that four types of PAEs occurred almost in all samples with concentrations ranging from 20 to 163,760 ng L−1 , which was inefﬁciently removed during traditional drinking water treatment processes [6]. Advanced treatment processes are required to attenuate the PAEs contamination. Various protocols are explored to enhance PAEs removal during water

∗ Corresponding author. Tel.: +86 451 86282292/86283010; fax: +86 451 82368074. ∗∗ Corresponding author. E-mail addresses: [email protected] (J. Ma), [email protected] (Z.-Q. Liu). 0304-3894/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.jhazmat.2011.08.054

treatment processes, including biotransformation [7], adsorption [8] and advanced oxidation processes [9–14]. The biotransformation of phthalates under both aerobic and anaerobic conditions has been investigated, but it is not suitable for applying in drinking water treatment due to less biomass existing in waterworks and requirement of long hydraulic retention time. Adsorption is an efﬁcient way to remove PAEs due to its higher n-octanol/water partition coefﬁcients (Kow ) (Table 1). However, it is only a method to shift the contaminations, but not to minimize them. Advanced oxidation processes would be the most powerful way for PAEs degradation and minimization. Several studies have been conducted for the elimination of PAEs by the processes of TiO2 /UV [9], H2 O2 /UV [10], electro-coagulation [11], ozonation [12] and catalytic ozonation processes [13,14]. Ozone is widely used in drinking water treatment for organic matter decomposition and microbiology disinfection all over the world. Several researchers conducted experiments on PAEs removal in the processes of ozonation and catalytic ozonation [12–14], with more attention to the removal efﬁciency and the way to improve it. However, the results of these studies illustrated that these processes could not be applied into practical water utilities due to the inﬂuence of natural water background. Elovitz and von Gunten [15] developed the hydroxyl radical/ozone ratio (Rct ) concept, which allows the prediction of the transformation of contaminations in natural water background combined with rate constants and oxidant behavior. In fact, ozonation and catalytic ozonation processes always involve in two active species: ozone

372

G. Wen et al. / Journal of Hazardous Materials 195 (2011) 371–377

Table 1 Chemical characteristics of four selected PAEs.

O

O

R

O

R

O Compound

Structure (–R)

CAS number

Molecular weight

Solubility (mg L−1 )

Log Kow

Dimethyl phthalate (DMP) Diethyl phthalate (DEP) Dipropyl phthalate (DPrP) Dibutyl phthalate (DBP)

–CH3 –CH2 CH3 –CH2 CH2 CH3 –CH2 CH2 CH2 CH3

131-11-3 84-66-2 131-16-8 84-74-2

194.2 222.2 250.3 278.4

4200 1100 108 11.2

1.66 2.65 3.27 4.50

and hydroxyl radical (• OH) [16]. The knowledge about second order rate constants of oxidation processes involving in both ozone alone and • OH would provide a powerful tool to optimize the degradation of PAEs. Unfortunately, a few kinetic data are available for ozone alone and • OH with PAEs. David Yao [17] reported the second order rate constant of DMP and DEP for the ﬁrst time with ozone alone by means of pseudo-ﬁrst order reaction in excess of ozone. However, there are not any reported rate constants of DBP and DPrP with ozone alone. Haag and Yao [18] estimated the second rate constant of DMP and DEP with • OH using a model method based on structure–activity relationship, but no experimental result has been reported on the second order constant rate of PAEs with • OH. Therefore, detailed kinetic constants of PAEs with ozone alone and • OH are required for further studies. The aim of this study is to determine the second-order rate constants of ozone alone and • OH with four PAEs by bench scale experiments in pure aqueous solution and to validate its applicability through oxidation and simulation of DEP degradation in river water background, where only DEP was selected as a representative due to the similarity of kinetics constant and the structures of 4 types of PAEs. Furthermore, bioluminescence test was performed to evaluate the acute toxicity change in the process of O3 /H2 O2 because of its higher degradation efﬁciency.

DMP, DEP, DPrP, DBP (Guang Fu Chemical Inc., China, 99.5% purity), p-chloro benzoic acid (pCBA, Sigma–Aldrich Chemicals, USA, 98% purity), high performance liquid chromatography (HPLC) grade methanol (Fisher, American), H2 O2 (30% w/w), ZnO powder (diameter 100 ␮m) and all other chemicals were of analytical grade and were used without further puriﬁcation. Milli-Q water (Millipore Q Biocel system) was used for sample preparation. Ozone stock solutions were produced by sparging O3 /O2 air into Milli-Q water. The chemical characteristics of four selected PAEs are listed in Table 1.

bench-scale glass reactor. Ozone delivered into the reactor via a medium porosity ceramic was kept constant concentration across the whole experiment process, which was supplied from an ozone generator (DHX-IIB model, Harbin Jiujiu Co.) with the inlet ozone concentration is 0.7 mg min−1 . When the ozone concentration in reactor was reaching a constant value (4–5 mg L−1 ), a small aliquot (10 ml of DMP, DEP, DPrP and 25 ml of DBP) of PAEs stock solution (100 ␮M of DMP, DEP, DPrP and 40 ␮M of DBP) was injected into the reactor, followed by starting the reaction. An aliquot of 0.1 mol L−1 sodium thiosulfate solution was used to quench the reaction after sampling at various intervals. Meanwhile, the ozone concentration in liquid was determined with indigo method [20]. The experiment was carried out in Milli-Q water using tert-butyl alcohol (TBA 10 mM) as the • OH scavenger and was adjusted to pH 2 with perchlorate (1 M). If not stated otherwise, the experiments were controlled at 25 ◦ C by using cooling water. To determine the activation energy for the reaction of ozone with PAEs, the same experiments were also performed at 5, 10, and 15 ◦ C. The experiment was repeated at least three times and the errors given were 95% conﬁdence intervals. Because of rapid reaction of • OH with organic matters, the rate constant of PAEs with • OH is difﬁcult to determine directly, but can be measured by using competition kinetics method [16,18]. Considering the structure of PAEs, the rate constant of • OH with PAEs was constant throughout the pH range evaluated. These experiments were carried out with Milli-Q water at 25 ◦ C and the pH was kept at 10 for ozone decomposition into • OH. The reference compound was pCBA exhibiting a rate constant of k• OH 5 × 109 M−1 s−1 [17]. Under alkaline conditions, the half-life time of ozone is about several seconds, and the dominant reaction is between • OH and organic matters. Therefore, the reaction with molecular ozone can be ignored due to its short half-life time and lower second rate constant [19]. The equal concentrations of the compounds (1 ␮M pCBA and 1 ␮M PAEs) were spiked into the water samples. Thereafter different under-stoichiometric concentration levels of ozone (ranging from 0.1 to 1 ␮M) were added. After ozone injection, the solutions were vigorously stirred. The residual concentrations of target and referenced compound in the ﬂask were analyzed by HPLC.

2.2. Determination of rate constants for the reaction of PAEs with ozone alone and • OH

2.3. Degradation and simulation of DEP decomposition in different ozonation processes in river water

Determination of the second-order rate constant of 4 types of PAEs with ozone was conducted by direct oxidation method under the condition of excessive ozone concentration. A semicontinuous ﬂow reaction model was used to determine the rate constant of PAEs with ozone, which was described by a previous study [19]. Brieﬂy, experiments were performed in a 1 L

Song Hua-jiang river water (water quality parameters are shown in Table 2) was used as the background for PAEs degradation. Due to the similarity of structure and kinetic properties of 4 types of PAEs, DEP was selected as the representative chemical to investigate its degradation in river water. River water was quickly ﬁltered (0.45 ␮m, cellulose acetate) within 8 h after sampled and stored at

2. Experimental 2.1. Materials and reagents

G. Wen et al. / Journal of Hazardous Materials 195 (2011) 371–377

373

Table 2 Water quality characteristics of the ﬁltered Song Hua-jiang river. pH

UV254 (cm−1 )

UV215 (cm−1 )

DOC (mg L−1 )

SUVA (L cm−1 mg−1 )

Alkalinity (mM)

7.9

0.0875

0.5248

3.5

3.36

4.7

4 ◦ C before use. During DEP degradation, river water was buffered to pH 8 by adding 10 mM borate buffer. The low concentration of DEP (1 ␮M) was spiked into river water, and 40 ml cold ozone stock solution (50 mg L−1 ) was injected to start the chemical reaction. Samples were taken at presumed intervals to determine the DEP concentration with HPLC. As described by Elovitz and von Gunten [15], Rct value can be calculated from the extent of the decrease of a probe compound (pCBA) concentration, which reacts fast with • OH but slowly with ozone, and a simultaneous determination of the ozone concentration. Once the Rct value is known, the elimination of a compound (M), which reacts with both oxidants, can be calculated by secondorder kinetics and expressed as a function of Rct , kO3 , kOH , and the ozone exposure ( [O3 ] dt) according to Eq. (1); ln

[M] [M]0

= −(

[O3 ] dt)(kOH Rct + kO3 )

(1)

In order to simulate the degradation of DEP, another experiment was designed as follows. pCBA was added as a probe compound (0.5 ␮M) to determine Rct , H2 O2 and ZnO were used as catalyst in catalytic ozonation experiments, respectively, with a ratio of 0.34 mg of H2 O2 /mg of O3 , and ZnO added with a concentration of 0.1 g L−1 . The other part of this experiment was the same as the DEP degradation described as above. Before DEP and pCBA analysis, the samples were ﬁltered with a ﬁlter (0.45 ␮m in pore size, cellulose acetate). 2.4. Toxicity test The acute toxicity of DEP and its degradation intermediate were tested with luminescent bacterium bioassay following the Chinese standard method (GB/T 15441-1995, 1996) [21,22]. It was performed using gram negative luminescent bacteria of the species Vibrio qinghaiensis sp. Nov (Q67). Due to the higher decomposition efﬁciency of O3 /H2 O2 , which was selected to decrease the toxicity of DEP and its intermediates products, a semi-continuous experiment (volume 1 L) was taken for DEP degradation in Milli-Q water with initial concentration of DEP of 20 ␮M, ozone concentration of 0.7 mg min−1 and H2 O2 concentration of 0.3 mM. Different samples at each interval were taken for toxicity test and TOC determination. The samples chosen for toxicity test were concentrated in Milli-Q water and adjusted to pH 7 before the analysis. Starting from a concentration factor of 300 times, eight double consecutive elution were tested (dilution factor 1:2), and the bioluminescence was then measured with Glomax illumination equipment (Turner Biosystems) [23]. The toxicity variation is expressed as Eq. (2). I(%) =

LB − LS × 100 LB

(2)

where I represents the inhibition of the concentrated sample to luminescent bacteria, LB is the luminescent intensity of blank, and LS is the luminescent intensity of sample. 2.5. Analytical methods The concentration of dissolved ozone in water was determined by the indigo method [20]. The concentration of ozone in gas phase was analyzed by iodometric method [24]. The concentration of PAEs were analyzed using HPLC equipped with an automatic Waters

717 plus autosampler injector and a Waters 1525 binary pump, using a waters symmetry C18 column (4.6 mm × 150 mm, 5 ␮m particle size) and methanol/water (50/50 for DMP, 60/40 for DEP, 70/30 for DPrP, 80/20 for DBP, v/v) as the mobile phase with a rate of 1 mL min−1 . The water sample was detected by a UV detector (Waters 2487 dual absorbance detector) at 230 nm and injected volume was 100 ␮L. The condition for pCBA analysis was as follows: an eluent with a rate of 1.0 mL min−1 consisting of 60/40 (v/v) methanol/water (adjusted to pH 2 with H3 PO4 ), UV wavelength of 240 nm was used. The pH in aqueous solution was measured by pH acidometer (Delta 320, Shanghai Leici Apparatus Fac., China). The TOC was analyzed by a TOC Analyzer (Analytik jena Multi N/C 3100). A Cary 500 UV–Vis spectrophotometer was used to measure UV254 value. The alkalinity of water was analyzed by titration method according to standard method [25]. 3. Results and discussion 3.1. Determination of rate constant for reaction of PAEs with ozone alone Several methods have been reported to determine second-order rate constant between ozone with compounds, including pseudoﬁrst order reaction (ozone in excess or chemicals in excess) [16,17], competition kinetics [17] and direct oxidation method [19]. The second-order rate constant was determined by direct kinetics in semi-continuous batch reactor, and the exact experiment step can be found as above mentioned. Both • OH inhibition and ozone decomposition were considered in this study. The reaction between PAEs with • OH was expelled by using 10 mM TBA as a way for scavenging it, ozone concentration was kept constant during the whole process, so the second-order reaction could be transformed in pseudo-ﬁrst order reaction. Table 1 shows that PAEs do not dissociate within all the range of pH, indicating that the second rate constant of PAEs is pH-independent. So, the experiment was carried out in pH 2 by using perchlorate for minimizing the reaction of • OH with PAEs. Under the condition of acidic pH and higher concentration of TBA in the ozonation system, molecular ozone reacts with PAEs predominantly, and the inﬂuence of • OH can be ignored. So, the rate of PAEs degradation can be written as follows: −

d[PAEs] = kO3 [O3 ][PAEs] dt

(3)

−

d[PAEs] = kO3 [O3 ]dt [PAEs]

(4)

Because of the constant ozone concentration, Eq. (4) can be converted to Eq. (5): ln

[PAEs]t = −kO3 [O3 ]t [PAEs]0

(5)

According to the above Eq. (5), kO3 can be concluded from the plots of PAE degradation versus time. The result of experiment is shown in Fig. 1. Table 3 lists all the results of the present experiment and the previous study. The rate constant of DMP, DEP, DPrP and DBP are 0.072, 0.085, 0.11 and 0.092 M−1 s−1 , respectively. Comparing the rate constants of four PAE compounds, there is no obvious relationship in the oxidation rate constants with the increase of PAE

374

G. Wen et al. / Journal of Hazardous Materials 195 (2011) 371–377 4.5

0.22

4.0

0.20

0.12 0.10

2.0

0.08

1.5

0.06 1.0

0.04

0.5

1.0 0.8 0.6 0.4

0.02

0.0 0

2000

4000

6000

8000

0.2

0.00 10000 12000 14000 16000

Time (s) Fig. 1. Variation of ozone concentration with time and the second-order plots of four types of PAEs degradation by ozone alone. Experimental conditions: [PAEs]0 = 1 ␮M, [TBA] = 10 mM, [O3 ] = 3.85 mg L−1 , pH = 2.0, T = 25 ◦ C.

carbon chain. The rate constants of DMP and DEP are close to those reported in previous study [17]. As can be seen from Table 3, PAEs are difﬁcult to be oxidized by ozone alone. Previous study shows that ozone behaves as an electrophilic species, which reacts only with some electronrich organic moieties, such as phenols, anilines, oleﬁns, and deprotonated-amines [26]. Advanced oxidation process, especially catalytic ozonation, produces more • OH and would be the option for PAEs elimination. The reaction rate constant between PAEs and • OH would be discussed afterwards. In order to ﬁgure out the activation energies, four different reaction temperatures were performed. Activation energy can be calculated according to the slope of the Arrhenius plot of ln k against T−1 to ﬁt Eq. (6). ln k = ln A −

Ea RT

(6)

where K is second order rate constant, Ea is reaction activation energy, T is reaction temperature, R is gas constant. DMP, DEP, DPrP and DBP exhibited 66 ± 8, 73 ± 4, 54 ± 10 and 58 ± 9 kJ mol−1 , respectively. They are endothermic reaction, enhanced reaction rate as the temperature increase. The results show no trends between the substitution groups with the activation energies. 3.2. Determination of rate constant for reaction of PAEs with • OH Rate constants for the reaction of 4 types of PAEs with • OH were determined by competition kinetics methods, which is the frequently used. Primary methods for generating • OH include UV/H2 O2 , ␥-radiolysis, O3 /H2 O2 [16], Fenton method, Walling’s method [18] and O3 /OH− [19]. In the present study, competition kinetics with pCBA as reference compound was adopted and the O3 /OH− method was selected for producing • OH. The second-order rate constants of the four types of PAEs compounds (M) with • OH were determined using Eq. (7) by plotting Table 3 Second-order rate constants for the reaction between ozone and 4 types of PAEs. Compound

(T = 25 ◦ C) (M−1 s−1 ) Measured

DMP DEP DPrP DBP

DBP DPP DMP DEP

1.2

0.14

ln(PAEs)0/(PAEs)

2.5

0.16

-ln(C/C0)

-1

O3 (mg L )

3.0

1.4

0.18

O3 DEP DBP DPP DMP

3.5

1.6

0.072 0.085 0.110 0.092

± ± ± ±

0.012 0.021 0.033 0.042

0.0 0.0

Reference

0.99 0.98 0.98 0.98

0.2 ± 0.1 [17] 0.14 ± 0.05 [17] No report No report

0.8

1.2

1.6

ln(pCBA)0/(pCBA) Fig. 2. The second-order plots of the degradation of four types of PAEs by • OH. Experimental conditions: [PAEs]0 = 1 ␮M, [pCBA] = 1 ␮M, [O3 ] = 0.1–1 ␮M, pH = 10.0, T = 25 ◦ C.

the decrease of M versus the reference compound (C), which can be seen from Fig. 2, a typical plot of relative rate experiment for the degradation of two compounds. k•MOH =

ln ([M]0 /[M]∞ ) C k• ln ([C]0 /[C]∞ ) OH

(7)

where k•MOH and k•COH are the rate constant for PAEs and reference compound, respectively. The measured second-order rate constants for the reaction of the PAEs with • OH are summarized in Table 4. The rate constant of DMP, DEP, DPrP and DBP is 2.67 × 109 , 3.98 × 109 , 4.47 × 109 and 4.64 × 109 M−1 s−1 , respectively. The rate constant of DMP is lower than that calculated by Haag, whereas DEP’s rate constant is comparative to Haag’s result [18]. Additionally, it is the ﬁrst time to report the second order rate constant of DPrP and DBP. As seen from the results in Table 4, the rate constant increases as the carbon chain extends. It is well-known that • OH attacks compounds by abstracting a hydrogen atom (H-abstraction), electron transfer reaction or by addition to an unsaturated bonds (such as C C bond). Seen from PAEs molecular structure, the attack to PAEs mainly depends on Habstraction, and the electron donor capacity of the substitute group follows the sequence: C(CH3 )3 > CH3 CH2 CH2 > CH3 CH2 > CH3 > H, which can explain the phenomenon of the increasing rate constant of PAEs as the increase of carbon chain length. Those second rate constants of PAEs would be further validated to its applicability through PAEs elimination in the processes of ozonation and catalytic ozonation processes and are discussed below. 3.3. Degradation and simulation of DEP decomposition in different ozonation processes in river water In natural water, the reactions of PAEs with ozone and • OH have to be considered together and it is essential to know their exposures Table 4 Second-order rate constants for the reaction between • OH and 4 types of PAEs. Compound

R2

0.4

KOH (T = 25 ◦ C) (M−1 s−1 ) Measured

DMP DEP DPrP DBP

(2.67 (3.98 (4.47 (4.64

± ± ± ±

0.26)E+9 0.21)E+9 0.35)E+9 0.41)E+9

R2

Reference

0.98 1.0 0.99 0.99

4E+9 [18] 4E+9 [18] No report No report

G. Wen et al. / Journal of Hazardous Materials 195 (2011) 371–377

2.0

0.5

375

1.0

1.0

0.8

0.9

O3 alone

0.8

O3/H2O2

O3

-1

-1.0

model of O3 alone

-1.5 -2.0 -2.5

1.2

0.6

-3.0 0

4

8

12

16

20

24

28

32

Time (min)

0.8

0.4

0.4

0.2

0.0

0.0

[pCBA]/ [pCBA]0

O3 (mg L )

pCBA

-0.5

C/C0

1.6

ln ([O3]/ [O3]0)

0.0

model of O3/H2O2 O3/ZnO

0.7

model of O3/ZnO

0.6

0.5 0

4

8

12

16

20

24

28

32

Time (min)

0.4 0

Fig. 3. Variation of ozone concentration with time and the plot of pCBA degradation in the process of ozonation. Experimental condition: [O3 ]0 = 2 mg L−1 , T = 25 ◦ C, [pCBA] = 0.5 ␮M, pH = 8, inset is the decomposition of ozone.

[27]. In natural river water background, a part of matrixes exists as • OH initiator, whereas another part of them serves as • OH scavengers. Rct , can be used to forecast the decomposition of compound in ozonation or catalytic ozonation processes in association with compound’s second-order rate constant of ozone and • OH. 3.3.1. Quantiﬁcation of Rct of ozonation and catalytic ozonation processes in river water Ozone decomposition in natural water can be divided into an initial and a second phase. During the second phase (>20 s), ozone decomposition follows an apparent ﬁrst-order rate law, the rate constant in the second phase is 10–100 times smaller than that during the initial phase [28]. As seen from Eq. (8), Rct describes that the ratio of • OH exposure to O3 -exposure, which can be calculated from the decrease in concentration of pCBA and O3 . The • OH exposure can be calculated by means of Eq. (9). The ozone exposure can be calculated from the integral of the ozone concentration versus time. Substitution of Eq. (9) into Eq. (8) gives the result of Rct , shown as Eq. (10). Rct =

[OH]dt

[OH] dt = − Rct = −

(8)

[O3 ]dt

ln([pCBA]/[pCBA]0 ) kOH,pCBA

(9)

ln([pCBA]/[pCBA]0 ) kOH,pCBA ·

(10)

[O3 ] dt

The concentration of pCBA (• OH probe) and O3 were detected, and typical results of Rct are shown in Fig. 3. The results show that the decomposition of O3 followed ﬁrst order rate reaction. The calculated Rct of three oxidation processes are shown in Table 5, indicating that the O3 /H2 O2 process presents the most powerful capability to produce • OH, followed by O3 /ZnO process in generating • OH. As was reported previously, the mechanism of H2 O2 and ZnO for improving organic matter removal is the enhancement of ozone decomposition and conversion of ozone into • OH [29,30]. H2 O2 generates plenty of HO2 − which enhances ozone decompoTable 5 Measured and calculated Rct in different processes. Different process

O3 alone

O3 /H2 O2

O3 /ZnO

Rct (measured) Rct (calculated)

1.20E−08 6.06E−08

1.37E−07 3.19E−07

3.23E−08 6.06E−08

5

10

15

20

25

30

Time (min) Fig. 4. Simulation and degradation of DEP removal during ozonation or catalytic ozonation in river water (dots mean experimental data and lines mean modeled results). Experimental condition: [O3 ]0 = 2 mg L−1 , T = 25 ◦ C, [pCBA] = 0.5 ␮M, [DEP] = 1 ␮M, pH = 8.

sition, meanwhile hydroxyl group on ZnO surface improves the decomposition of ozone and the formation of • OH [31]. Recently, a method for predicting Rct in different natural water background was developed [32], which showed that Rct was dependent on water quality characteristics and could be simulated (R2 = 0.92), using water quality characteristics and experimental conditions (Eq. (11)). log Rct = −10.12 + 2.04 DH2 O2 − 0.325 DOC + 0.747 pH −11.47 UV254 − 0.143 URI

(11)

URI = UV relative index (calculated as the ratio of UV215 over UV254 ); DH2 O2 = peroxide dosage (mg H2 O2 /mg O3 ); UV254 = UV absorbance at 254 nm (cm−1 ); DOC = dissolved organic carbon (mg C L−1 ). In our present study, the model was used to predict the Rct in Song Hua-jiang river water, which can be seen in Table 5. Comparing the experimental results with the simulated results, the model data has the same order as experimental result. However, it cannot differentiate the processes of ozonation alone and O3 /ZnO catalytic ozonation because there is no consideration of the inﬂuence of heterogeneous catalyst on ozone decomposition. Therefore, the model is not appropriate for predicting the Rct in heterogeneous catalytic ozonation processes. 3.3.2. Simulation of DEP decomposition in river water Natural water matrixes have important impact on the ozonation process of organic matter, where natural organic matter may promotes or prohibits the radical chain reaction, acting as an initiator or scavenger, while, the alkalinity competes with organic matter for • OH as a scavenger [33]. Batch experiments with river water as background were performed to examine the removal efﬁciency of DEP in O3 , O3 /H2 O2 , and O3 /ZnO processes. The water quality parameters are given in Table 2. As seen from the experimental results in Fig. 4, the oxidation of DEP was mainly determined in reactions with • OH. The oxidation efﬁciencies increases with the increase of Rct , namely O3 /H2 O2 process with higher Rct removes DEP much faster than that of O3 alone or O3 /ZnO process.

376

G. Wen et al. / Journal of Hazardous Materials 195 (2011) 371–377 0.5

5

1.0 4

0.8

TOC

3

0.6

DEP

0.4

0.1

0.0 0

5

10

15 Time (min)

20

25

2

0.2

1

0.0

0

-1

0.2

TOC (mg L )

0.3

C/C 0

Inhibitation rate (%)

0.4

Acknowledgements

30

Fig. 5. Toxicity assessment of DEP oxidation and TOC change in the process of O3 /H2 O2 .

Furthermore, it is possible to predict the DEP removal with a function of ozone exposure, Rct , kOH , and kO3 according to Eq. (12). ln

[M] t

[M]0

= −(

Rct were conducted. Three different ozonation processes including ozone alone, O3 /H2 O2 and O3 /ZnO were evaluated in Song Huajiang river water background, which showed that the O3 /H2 O2 process had the highest capacity in degrading DEP. The DEP transformation was successfully simulated with the determined second-order rate constant in combination with Rct . The acute toxicity in the process of O3 /H2 O2 was assessed by luminescent bacteria, which showed that O3 /H2 O2 was an efﬁcient way for PAEs degradation and toxicity reduction, but it was a less efﬁcient method for DEP mineralization.

This research was supported by State Key Laboratory of Urban Water Resource and Environment, Harbin Institute of Technology (2010DX10), 863 Hi-tech Research and Development Program of China (Grant No. 2009AA06Z310) and the National Key Special Founding for Water Pollution Control and Management (Grant No. 2008ZX07421-002, 2009ZX07424-005, 2009ZX07424-006) and NSFC (50821002). References

[O3 ] dt)(kOH Rct + kO3 )

(12)

Fig. 4 presents the data for the predicted and measured oxidation of DEP in the three oxidation processes. The results indicated that DEP removal efﬁciency in river water followed the sequence, O3 /H2 O2 > O3 /ZnO > O3 , which is in accord with the Rct , meaning that O3 /H2 O2 shows the most powerful capability to produce • OH among the three processes. It was calculated with experimental Rct and model Rct for DEP removal with Eq. (12), which showed poor simulation results compared with the degradation data using the calculated Rct (data not shown here). Using the experimental Rct , the simulated results were well in consistence with the experimental data. From the consistence of experimental data with the simulated data, it can be concluded that above determined secondorder rate constants of DEP with O3 and • OH are applicable in natural water. 3.4. Toxicity assessment Fig. 5 shows the variations of toxicity and TOC in the process of O3 /H2 O2 . It is seen from the results that after 30 min oxidation, the acute toxicity expressing inhabitation rate decreases to below the detection limit, namely no acute toxicity was detected. The decrease in acute toxicity of DEP mainly occurs in the ﬁrst 10 min, which coincides with the decrease of DEP concentration. Therefore, the primary toxicity to luminescent bacteria may come from the DEP parent molecular, while the intermediates contribute a little to the toxicity for Q67. After 30 min oxidation, the TOC decreases from 3.76 to 3.22 mg L−1 , indicating that only 14.4% of TOC is mineralized and the intermediates are accumulated in the oxidation process, which illustrated furthermore that the decrease of toxicity to Q67 was due to DEP parent molecular removal, but not the formation of intermediates. Through continuous oxidation of ozone and • OH, the parent of DEP had been decomposed into nontoxic intermediate. However, the mineralization of DEP requires more powerful oxidation process or in combination with biotransformation. 4. Conclusions The second-order rate constants for the reaction of 4 PAEs with ozone and • OH were determined by direct oxidation method and competition kinetics method by bench-scale experiments. Degradation of DEP in river and simulation decomposition based on

[1] M. Vitali, M. Guidotti, G. Macilenti, C. Cremisini, Phthalate esters in freshwaters as markers of contamination sources – a site study in Italy, Environ. Int. 23 (1997) 337–347. [2] J. Sundell, C.G. Bornehag, Presences of phthalate esters in homes, Epidemiology 17 (2006) S86. [3] C.A. Stales, D.R. Peterson, T.F. Parkerton, W.J. Adams, The environmental fate of phthalate esters: a literature review, Chemosphere 35 (1997) 667–749. [4] C.A. Harris, P. Henttu, M.G. Parker, J.P. Sumpter, The estrogenic activity of phthalate esters in vitro, Environ. Health Perspect. 105 (1997) 80–811. [5] N. Li, D. Wang, Y. Zhou, M. Ma, J. Li, Z. Wang, Dibutyl phthalate contributes to the thyroid receptor antagonistic activity in drinking water processes, Environ. Sci. Technol. 44 (2010) 6863–6868. [6] X.L. Shao, J. Ma, G. Wen, Investigation of endocrine risrupting chemicals in a drinking water work located in Songhua River basin, Environ. Sci. 129 (2008) 2723–2728 (In Chinese). [7] B.V. Chang, C.M. Yang, C.H. Cheng, S.Y. Yuan, Biodegradation of phthalate esters by two bacteria strains, Chemosphere 55 (2004) 533–538. [8] S. Venkata Mohan, S. Shailaja, M. Rama Krishna, P.N. Sarma, Adsorptive removal of phthalate ester (di-ethyl phthalate) from aqueous phase by activated carbon: a kinetic study, J. Hazard. Mater. 146 (2007) 278–282. [9] X.R. Xu, S.X. Li, X.Y. Li, J.D. Gu, F. Chen, X.Z. Li, H.-B. Li, Degradation of n-butyl benzyl phthalate using TiO2 /UV, J. Hazard. Mater. 164 (2009) 527–532. [10] B. Xu, N.Y. Gao, X.F. Sun, S.J. Xia, M. Rui, M.-O. Simonnot, C. Causserand, J.F. Zhao, Photochemical degradation of diethyl phthalate with UV/H2 O2 , J. Hazard. Mater. 139 (2007) 132–139. [11] I. KabdaslI, A. Keles, T. Ölmez-HancI, O. Tünay, I. Arslan-Alaton, Treatment of phthalic acid esters by electrocoagulation with stainless steel electrodes using dimethyl phthalate as a model compound, J. Hazard. Mater. 171 (2009) 932–940. [12] H.Y. Li, J.H. Qu, H.J. Liu, Removal of a type of endocrine disruptors-di-n-butyl phthalate from water by ozonation, J. Environ. Sci. 18 (2006) 845–851. [13] Y.H. Chen, N.C. Shang, D.C. Hsieh, Decomposition of dimethyl phthalate in an aqueous solution by ozonation with high silica zeolites and UV radiation, J. Hazard. Mater. 157 (2008) 260–268. [14] M. Hammad Khan, J.Y. Jung, Ozonation catalyzed by homogeneous and heterogeneous catalysts for degradation of DEHP in aqueous phase, Chemosphere 72 (2008) 690–696. [15] M.S. Elovitz, U. von Gunten, Hydroxyl radical/ozone ratios during ozonation processes I. The Rct concept, Ozone Sci. Eng. 21 (1999) 239–260. [16] U. von Gunten, Ozonation of drinking water: part I. Oxidation kinetics and product formation, Water Res. 37 (2003) 1443–1467. [17] C.C. David Yao, W.R. Haag, Rate constants for direct reactions of ozone with several drinking water contaminants, Water Res. 25 (1991) 761–773. [18] W.R. Haag, C.C.D. Yao, Rate constants for reaction of hydroxyl radicals with several drinking water contaminants, Environ. Sci. Technol. 26 (1992) 1005–1013. [19] J.M. Shen, Z.L. Chen, Z.Z. Xu, X.Y. Li, B.B. Xu, F. Qi, Kinetics and mechanism of degradation of p-chloronitrobenzene in water by ozonation, J. Hazard. Mater. 152 (2008) 1325–1331. [20] H. Bader, J. Hoigné, Determination of ozone in water by the indigo method, Water Res. 15 (1981) 449–456. [21] M. Ma, Z. Tong, Z. Wang, W. Zhu, Acute toxicity bioassay using the freshwater luminescent bacterium; Vibrio qinghaiensis sp. Nov.—Q67, Bull. Environ. Contam. Toxicol. 62 (1999) 247–253. [22] X.W. Zhu, S.S. Liu, H.L. Ge, Y. Liu, Comparison between the short-term and the long-term toxicity of six triazine herbicides on photobacteria Q67, Water Res. 43 (2009) 1731–1739.

G. Wen et al. / Journal of Hazardous Materials 195 (2011) 371–377 [23] S. Monarca, D. Feretti, C. Collivignarelli, L. Guzzella, I. Zerbini, G. Bertanza, R. Pedrazzani, The inﬂuence of different disinfectants on mutagenicity and toxicity of urban wastewater, Water Res. 34 (2000) 4261–4269. [24] K. Rakness, G. Gordon, B. Langlais, W. Masschelein, N. Matsumoto, Y. Richard, C.M. Robson, I. Somiya, Guideline for measurement of ozone concentration in the process gas from an ozone generator, Ozone Sci. Eng. 18 (1996) 209–229. [25] A.E. Greenberg, L.S. Clesceri, A.D. Eaton, Standard Methods for the Examination of Water and Wastewater, 20 ed., 1998, Washington, DC, pp. 1325. [26] Y. Lee, U. von Gunten, Oxidative transformation of micropollutants during municipal wastewater treatment: comparison of kinetic aspects of selective (chlorine, chlorine dioxide, ferrateVI, and ozone) and non-selective oxidants (hydroxyl radical), Water Res. 44 (2010) 555–566. [27] M.M Huber, S. Canonica, G.Y. Park, U. von Gunten, Oxidation of pharmaceuticals during ozonation and advanced oxidation processes, Environ. Sci. Technol. 37 (2003) 1016–1024.

377

[28] M.O. Bufﬂe, U. Von Gunten, Phenols and amine induced • OH generation during the initial phase of natural water ozonation, Environ. Sci. Technol. 40 (2006) 3057–3063. [29] E.J. Rosenfeldt, K.G. Linden, S. Canonica, U. von Gunten, Comparison of the efﬁciency of OH radical formation during ozonation and the advanced oxidation processes O3 /H2 O2 and UV/H2 O2 , Water Res. 40 (2006) 3695–3704. [30] M. Muruganandham, J.J. Wu, Synthesis, characterization and catalytic activity of easily recyclable zinc oxide nanobundles, Appl. Catal. B: Environ. 80 (2008) 32–41. [31] J. Nawrocki, B. Kasprzyk-Hordern, The efﬁciency and mechanisms of catalytic ozonation, Appl. Catal. B: Environ. 99 (2010) 27–42. [32] S. Vincent, A. Kotbi, B. Barbeau, Predicting hydroxyl radical activity and trace contaminants removal in ozonated water, Ozone Sci. Eng. 32 (2010) 244–251. [33] J. Staehelin, J. Hoigne, Decomposition of ozone in water in the presence of organic solutes acting as promoters and inhibitors of radical chain reactions, Environ. Sci. Technol. 19 (1985) 1206–1213.

Journal of Hazardous Materials 195 (2011) 378–382

Contents lists available at SciVerse ScienceDirect

Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat

Inﬂuence of microbial adaption and supplementation of nutrients on the biodegradation of ionic liquids in sewage sludge treatment processes Marta Markiewicz a , Stefan Stolte b , Zoﬁa Lustig a , Justyna Łuczak a , Michał Skup c , Jan Hupka a , Christian Jungnickel a,∗ a

Department of Chemical Technology, Chemical Faculty, Gda´ nsk University of Technology, ul. Narutowicza 11/12, 80-233 Gda´ nsk, Poland Center for Environmental Research and Sustainable Technology, University of Bremen, Leobener Strasse UFT, D-28359 Bremen, Germany c Department of Environmental Analysis, Faculty of Chemistry, University of Gda´ nsk, Sobieskiego 18, 80-233 Gda´ nsk, Poland b

a r t i c l e

i n f o

Article history: Received 8 June 2011 Received in revised form 17 August 2011 Accepted 18 August 2011 Available online 24 August 2011 Keywords: Ionic liquids Biodegradation OECD 301 Adaptation Supplementation

a b s t r a c t As ionic liquids are winning more attention from industry as a replacement of more hazardous chemicals, some of their structures have the potential to become persistent pollutants due to high stability towards abiotic and biotic degradation processes. Therefore it is important to determine the hazard associated with the presence of ILs in the environment, for example biodegradation under real conditions. Standard biodegradation testing procedures generally permit pre-conditioning of inoculum but do not allow for pre-exposition to the test substance. These are usually conducted in a mineral medium which does not provide additional organic nutrients. Though very valuable, as a point of reference, these tests do not fully represent real conditions. In in situ conditions, for example in wastewater treatment plants or natural soils and water bodies, the presence of readily available sources of energy and nutrients as well as the process of adaptation may often alter the fate and metabolic pathways of xenobiotics. Our results have shown that these are the opposing processes inﬂuencing the biodegradation rate of ILs in sewage sludge. The results have signiﬁcant practical implications with respect to the assessment of biodegradability and environmental fate of ILs and other xenobiotics in environmental conditions and their potential remediation options. © 2011 Elsevier B.V. All rights reserved.

1. Introduction Industrial development during the last decades resulted in increased pollution of the environment by xenobiotics. Due to this, the need for understanding the impact of toxic compounds on microbial populations and the catabolic degradation pathways of xenobiotics has arisen. Thus standardized biodegradability and toxicity test were developed to allow for classiﬁcation of xenobiotics according to the environmental hazard they pose. Bearing in mind the deﬁnition of xenobiotics, as man-made chemicals foreign to organisms which inhabit the environment, their biodegradation rate in natural soils and waters is in most cases much lower than that of natural compounds. Nevertheless structural similarities to biomolecules can result in relatively high biodegradation rates if enzymes of low substrate speciﬁcity are present. Factors which may inﬂuence this rate, among others, include microbial adaptation and availability of additional nutrients [1]. Ionic liquids (ILs) as a non-conventional class of novel solvents are becoming increasingly important owning to a number

∗ Corresponding author. Tel.: +48 58 3472334. E-mail address: [email protected] (C. Jungnickel). 0304-3894/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.jhazmat.2011.08.053

of desirable characteristics including negligible volatility, nonﬂammability, high thermal stability, low melting point, broad liquid range and controlled miscibility with organic compounds or water [2–5]. The negligible volatility limits their impact on air quality, but their release to the environment may affect soil and water. Moreover, some IL structures have the potential to become persistent pollutants due to their high stability towards abiotic and biotic degradation processes. Therefore it is important to determine the hazard associated with the presence of ILs in the environment. Adaptation is deﬁned as a change in the microbial community that leads to an increase in the biodegradation rate, or maximal biodegradable concentration of a given xenobiotic as a result of previous exposure. Examples of such adaptation processes are e.g. rapid degradation of p-nitrophenol by aquatic microorganisms [6] and enhanced degradation rates after elongated exposure of subsurface soil communities to m-cresol, m-aminophenol and aniline [7]. Mechanisms of adaptation usually involve processes such as genetic mutation or horizontal gene transfer, induction of speciﬁc enzymes which enhance the degradative capacity of the entire community, and population change such as selective growth of certain strains [8]. All mechanisms may take place simultaneously, or one may dominate and exact prediction of which will occur is not possible [9]. Furthermore it should be noted, that adaptation does

M. Markiewicz et al. / Journal of Hazardous Materials 195 (2011) 378–382

B

1

100

0.9

90

0.8

80

0.7

70

biodegradation [%]

concentration [mM]

A

0.6 0.5 0.4 0.3

60 50 40 30

0.2

20

0.1

10

0 0

5

10

15

20

25

time [days]

379

0

0

5

10

15

20

25

time [days]

), Fig. 1. (A) Biodegradation curves of [OMIM][Cl] as a sole source of carbon (—), with supplementation of glucose (- - -), with supplementation of synthetic feed ( ). (B) Biodegradation (%) normalized for sorption to sewage sludge ﬂocs of [OMIM][Cl] as a sole source of carbon (—), with supplementation of glucose sorption control ( ). (- - -), with supplementation of synthetic feed (

not necessarily occur in every case. Aelion et al. did not observe any adaptation of subsurface microbial communities to chloro- and trichlorobenzene after an eight months adaptation period [7]. Similarly Nyhoim et al. did not note any increase in the biodegradation rate after pre-exposure of activated sewage sludge to aniline and pentachlorophenol [9]. The speciﬁc reason for this remains unclear though many theories exist. The most probable reasons include lack of complete enzyme systems within the population, accumulation of toxic degradation products, binding with enzymes causing inactivation or insufﬁcient cell density of inoculum [10]. One of the few papers which discusses the adaptation of soil microorganisms to ionic liquids proposes that the electron-donor ability of the IL effect the biodegradability [11]. A number of research groups have performed biodegradation tests with alkyl substituted imidazolium cations using activated sewage sludge [12–15]. In most cases ILs were used as a sole source of organic carbon and organic nitrogen. This is especially important, because it should be remembered that in wastewater treatment plants or natural environments, other organic substrates are present, which might be preferentially degraded or co-metabolized with the primary contaminant resulting in lower biodegradation rates [16]. Romero et al. [17], discussed the biodegradability of imidazolium ILs in the presence of additional carbon source. It was found that the ILs tested were not biodegradable when D-glucose was available. However, ILs with no additional carbon were also not degraded (2–10%), which is in contrast to other research where, e.g. complete primary biodegradation of 1-methyl-3-octylimidazolium chloride [OMIM][Cl] was shown [12,18]. Results of Romero et al., though very interesting, should be treated with caution due to the very short duration of the test (ﬁve days) as well as lack of collaborating results in literature. Standard biodegradation testing procedures generally permit pre-conditioning of inoculum (aeration in the presence of a mineral medium) but do not allow for pre-exposition to the test substance. The purpose of this is to provide repeatable results enabling comparison and standardization of biodegradation rates of different chemicals usually for regulatory purposes. Though very valuable, as a point of reference, these tests do not fully represent real conditions [19]. Therefore, to more accurately predict biodegradation under real conditions it is beneﬁcial to take adaptation into account especially if biodegradation requires induction of speciﬁc metabolic pathways, e.g. aromatic ring break-down [6,20]. One of the few works which discuss the adaptation of microorganisms to ILs, conducted by Stolte et al., found a sixfold increase in the biodegradation rate of [OMIM][Cl] over a period of 31 days [12]. Additionally, Docherty et al. observed complete biodegradation of

hexyl-methylimidazolium bromide after extending duration of the test and concluded that though IL could not be classiﬁed as readily biodegradable it is not expected to persist in the environment [20]. The aim of this paper is to describe the effect of additional substrates and pre-exposition of bacteria to IL on the rate of biodegradation, and thereby discuss the relevance of including preexposition in standardized tests.

2. Experimental methodology 2.1. Modiﬁed OECD 301A DOC Die-Away test – supplementation The ionic liquid used in the test was [OMIM][Cl] provided by Merck KGA, Darmstadt, Germany. The sewage sludge (dry mass 6.5 g L−1 ) was taken from the aeration chamber of the ´ “Gdansk – Wschód” municipal wastewater treatment plant, ´ Poland. Primary degradation was detected by direct Gdansk, determination of the substrate by HPLC – UV. Eight test ﬂasks containing 0.5 L of sewage sludge ﬂocs and mineral medium composed of: 8.5 mg L−1 KH2 PO4 , 21.75 mg L−1 K2 HPO4 , 22.3 mg L−1 Na2 HPO4 ·2H2 O, 1.7 mg L−1 NH4 Cl, 27.5 mg L−1 CaCl2 , 22.5 mg L−1 MgSO4 ·7H2 O and 0.25 mg L−1 FeCl3 dissolved in water were prepared as recommended by OECD procedure [21]. Subsequently, a solution of [OMIM][Cl] was added to yield the concentration of 1 mM and the amount of test solution was made up to 1 L. Each test concentration was conducted in duplicate. Two test ﬂasks were additionally supplemented with glucose and two with synthetic sewage feed (16 g of peptone, 11 g of meat extract, 3 g of urea and 0.7 g NaCl dissolved in 1 L of water). Nutrients were added three times a week, 0.36 g and 2.5 mL, respectively. Also blank samples (without test substance) and chemically sterilized negative controls were prepared. All test vessels were aerated and analytical samples were collected in duplicate at speciﬁc time intervals. Mass loss due to evaporation was compensated at every collection interval.

2.2. Modiﬁed OECD 301A DOC Die-Away test – adaptation The test vessels were prepared as previously described. Sewage sludge from the same source was used (dry mass 5.5 g L−1 ). Increasing concentrations of [OMIM][Cl] (1 mM, 1.5 mM, 2 mM, 2.5 mM) were added every fortnight. The total time for the adaptation test was two months. All vessels were aerated and analytical samples were collected in duplicate at speciﬁc time intervals.

380

M. Markiewicz et al. / Journal of Hazardous Materials 195 (2011) 378–382

conditions were set in accordance to [22] with a capillary voltage of 3500 V, drying gas ﬂow-rate of 5 L min−1 , drying gas temperature at 300 ◦ C and nebulizer at 70 psi.

6

concentraon [mM]

5

3. Results and discussion

4

3.1. Supplementation

3

2

1

0

1

4

7

11 13 17 18 22 25 29 30 32 36 39 44 45 46 50 53 58

me [days] Fig. 2. Biodegradation of [OMIM][Cl] by adapted sewage sludge community (black bars) and abiotic control (grey bars).

2.3. HPLC analysis Analytical samples were centrifuged to remove solids and a supernatant was taken for the HPLC-UV analysis. A Perkin Elmer Series 200 HPLC consisting of a chromatographic interface (Link 600) binary pump, UV/VIS detector, vacuum degasser and Rheodyne injection valve were used. For IL’s cation separation C6-Phenol (Phenomenex) 150 × 4.6 mm column was used in conjunction with detection by UV adsorption at 218 nm. As a mobile phase 27% acetonitrile/water + 0.1% (v/v) triﬂuoroacetic acid at the ﬂow rate of 0.8 mL min−1 was applied. For preparation of HPLC mobile phase HPLC – grade acetonitrile, Lab – Scan (Dublin, Ireland) and spectrophotometric grade triﬂuoroacetic acid (Sigma–Aldrich, Germany) were used. 2.4. Metabolites analysis Additional analytical samples were taken from vessels containing live and chemically sterilized inocula at the end of adaptation tests. Samples were centrifuged to remove solids and the supernatant was diluted hundredfold (biotic sample) or thousandfold (abiotic sample) with a 9:1 methanol–water mixture resulting in approximately 5 ␮M concentration of parent compound in all samples. Subsequently samples were analyzed for the parent compound and metabolites by electrospray ionization mass spectrometry equipped with ion trap detector (Brucker-Daltonic GmbH, Germany). Mass spectra for cations were acquired in the positive ion mode in the scan range of m/z+ 50–300. The ESI source

concentration [mM]

2.5 2 1.5 1 0.5 0

0

5

10

15

20

25

30

35

time [days] Fig. 3. Comparison of biodegradation rate of 2 mM [OMIM][Cl] conducted by adapted (solid line) and non-adapted (dashed line) activated sewage sludge community.

The initial, fast decrease in [OMIM][Cl] concentration was due to sorption of [OMIM][Cl] onto ﬂocs of activated sewage sludge (shown in Fig. 1A). The sewage sludge organic matter (especially extracellular polymeric substances) can act as a ‘buffer’ for the IL, initially decreasing the bioavailable concentration and thereby mitigating its toxicity. The supplementation with glucose and synthetic feed increased the time of [OMIM][Cl] primary biodegradation (Fig. 1B). It can be observed that after approximately 14 days the biodegradation of 0.2 mM [OMIM][Cl] remaining after sorption was completed only when no supplements were added. After more than 20 days biodegradation was accomplished in the sample where synthetic feed was added. In the vessel with glucose supplementation complete biodegradation was not observed within the test timeframe. [OMIM][Cl] biodegradation with synthetic feed cannot be explained using the diauxie effect [23], as the feed was added continuously for the duration of the test. For [OMIM][Cl] as a nominal source of organic carbon and organic nitrogen complete primary biodegradation is achieved within 14 days. When synthetic feed is present the biodegradation rate is clearly reduced, even though complete primary degradation is achieved within 23 days. When supplemented with glucose, providing easily available source of organic carbon, [OMIM][Cl] is utilized in less than 20%. Therefore the presence of other nutrients in the sewage or within the environmental media in general can have a strong inﬂuence on the biodegradability of ionic liquids. Also compounds which have been classiﬁed as “readily biodegradable” might present recalcitrance towards biodegradation under real environmental conditions which has to be taken into account when evaluating their fate in the environment. Generally, the reduced biodegradation rate of [OMIM][Cl] in the presence of glucose is consistent with the research conducted by Lewis et al. [24] where the addition of organic carbon signiﬁcantly decreased the degradation of p-cresol. The addition of synthetic feed containing organic carbon and nitrogen decreases the rate of biodegradation (relative to non-supplemented tests). Similar results were obtained by Swindoll et al. for p-nitrophenol [25]. On the other hand, Piekarska et al. showed that the addition of other sources of organic carbon and nitrogen increased the efﬁciency of degradation of diesel oil [26]. This can be explained by the difference in the chemical structure of the primary substrates. Nitrophenol is a pure aromatic compound, biodegradation of which requires the induction of a speciﬁc metabolic pathway. Diesel oil is a mixture containing mostly linear or branched hydrocarbons, which are degraded through the ␤-oxidation pathway. The enzymatic systems for this pathway are relatively common in most soil and sewage microorganisms. To summarize, addition of easily available organic carbon and nitrogen sources seems to facilitate biodegradation when the primary pollutant itself is relatively easy to degrade. It can be anticipated that xenobiotics containing structures which are commonly recognized as poorly biodegradable will not be metabolized if microorganisms can obtain carbon from other sources. This hypothesis seems to hold true when the secondary substrate is organic. Inorganic supplements were proven to facilitate biodegradation. The addition of inorganic carbon (NaHCO3 ) and inorganic nitrogen (NH4 Cl) has previously been shown to increase the rate of biodegradation of xenobiotic compounds [9,24]. Utilization of

M. Markiewicz et al. / Journal of Hazardous Materials 195 (2011) 378–382

m/z +

381

intensity biotic sample (0.5 mM OMIM; 1:100 diluted)

abiotic sample (5 mM OMIM; 1:1000 diluted)

195

3*105

3*105

211

1*104

<10 3

N

N

N

+

+

N

OH

209

1*104

<10 3

N

+

N

O H

225

2*104

<103

N

+

N

O

The fragmentation of these isolated ions in ESI-MS/MS mode gave a fragment of a protonated methylimidazole (m/z + = 83) which is characteristic for 3-methylimidazolium based ILs N

N

+

H

OH

169

3*104

<103

N

+

N

OH O

Fig. 4. Mass-to-charge ratio (positive mode), intensity of signals within the mass spectra and proposed chemical structures of degradation products.

organic carbon/nitrogen requires cleavage of these elements from the carbon skeleton before it can be assimilated by the cell, therefore it is a less energetically favorable source of these elements [27]. If both primary and secondary substrates are organic usually a similar set of reactions is needed for their break down which results in a competition for enzymes. It should be mentioned that the concentration of [OMIM][Cl] degraded in the non-supplemented test was higher than in any previously published work [18]. Thus it can be assumed that in wastewater treatment plant operating conditions this substance might be treatable as deﬁned by OECD standards. The high level of sorption of [OMIM][Cl] will also allow for enhanced removal.

3.2. Microbial adaptation Municipal sewage sludge was exposed to gradually increasing levels of [OMIM][Cl] added in fortnightly intervals. Progressive accumulation of the xenobiotic can be observed in abiotic control (Fig. 2). Theoretical total concentration anticipated in all samples at the end of the test, excluding biodegradation and sorption processes, is 7 mM. Concentrations measured in aqueous phase of test media in biotic and abiotic control are presented in Fig. 2. In non-sterilized sample sewage sludge was able to adapt to the IL. An initial lag phase of around two weeks was observed. The biodegradation rate for each addition increased from 1% day−1 to 6% day−1 , to 8% day−1 after the ﬁnal addition (percentage degraded relative to the amount added for each period). The degradation data was ﬁt using ﬁrst order kinetics according to Paul and Clark [28], the corresponding rate constants where calculated to be 0.065, 0.13, 0.27, and 0.13 s−1 . The half-life of the contaminant in the test for each addition of IL was therefore 10.5, 5.3, 2.6, and 5.5 days. It should be noted that with each addition the concentration of IL is also raised, which might account for the lowered calculated rate constant of the last addition. The adaptation has allowed the sewage sludge to degrade concentrations of [OMIM][Cl] previously reported to be too high [18]. A small decrease in [OMIM][Cl] con-

centration in sorption control was observed possibly indicating that sterilizing agent did not inhibit biodegradation completely, however sufﬁciently to allow for distinction from sorption. Considering the fact that within the ﬁnal sample (day 58, Fig. 2) a total of 7 mM OMIM was degraded and no transformation products were detected via HPLC-UV (the imidazolium core is responsible for UV absorption) it is reasonable to believe that the whole structure, including the imidazolium core, was biodegraded. Fig. 3 shows biodegradation curves for adapted and nonadapted communities. No biodegradation of [OMIM][Cl] in concentration of 2 mM was observed for the non-adapted community. No adaption was observed in this case, probably due to toxicity of the IL at this concentration. The initial decrease in concentration is due to sorption on sewage sludge ﬂocs, as previously shown. The adapted community, however, was able to utilize the IL. Complete primary biodegradation was achieved within 15 days. The results from ESI-MS analysis have been used to identify biological transformation products. Therefore, the biotic and abiotic samples after the course of biodegradation (53 days, Fig. 2) have been used for a qualitative analysis. The ESI-MS analysis indicates that several biological transformation products were formed such as compounds with hydroxylated or carboxylated side chains (Fig. 4, for mass spectra see supplementary data). The identiﬁed metabolites correspond to the recently proposed degradation pathway of the OMIM cation [12] starting with an ␻-oxidation (introduction of a terminal hydroxyl group) and a subsequently degradation of the alkyl side chain via ␤-oxidation. Also in the abiotic sample some biodegradation products could be identiﬁed with very low peak intensities (not present in the OMIM standard) conﬁrming that inhibition of inoculum was not complete.

4. Conclusions In in situ conditions, for example in wastewater treatment plants or natural soils and water bodies, the presence of readily available sources of energy and nutrients (e.g. sugars, fats or proteins) may

382

M. Markiewicz et al. / Journal of Hazardous Materials 195 (2011) 378–382

often alter the fate and metabolic pathways of xenobiotics. Our results have shown that there are two opposing processes inﬂuencing the biodegradation rate of ILs in sewage sludge. The results have signiﬁcant practical implications with respect to the assessment of biodegradability and environmental fate of ILs and other xenobiotics in environmental conditions and their potential remediation options. In the activated sewage sludge process, the excess sewage is sequentially removed, however adapted communities of microorganisms are expected to remain in the reactor. The signiﬁcance of adsorption on sewage sludge ﬂocs as a way of xenobiotic removal during the wastewater treatment cannot be overstated. The typical time of hydraulic retention of sewage in the bioreactor is approximately one day and is too short for the biodegradation of most xenobiotics to occur. However these contaminants usually do not persist in the wastewater treatment plant efﬂuents, and therefore the removal by physical adsorption on ﬂocs which are subsequently removed is of a great importance [19]. In present case, adaptation promotes the removal of ILs. It should be noted that not only is it increasing the rate of biodegradation, but also increasing the maximum biodegradable concentration. It was shown hereby that through the process of adaptation a nearly 30-fold increase of the biodegradation rate can be obtained. Additionally concentrations of ILs previously reported to be too high to be metabolized by non-adapted communities were degraded. Moreover complete degradation of imidazolium ring was observed. It can be anticipated that during wastewater treatment of IL in an activated sewage sludge process similar phenomena will occur. On the contrary, the supplementation with organic carbon or nitrogen diminished the rate of biodegradation. The microbial community preferentially utilized secondary supplements and ILs persisted. This contradicts the description of ready biodegradability by OECD which states that “it is assumed that such [readily biodegradable] compounds will rapidly and completely biodegrade in aquatic environments under aerobic conditions”. It was shown that supplying both organic carbon and nitrogen has a less detrimental effect than supplying only carbon. In the ﬁrst case the observed lag phase was long but was eventually followed by degradation whereas in the latter complete inhibition of biodegradation was observed. It should be noted that the biodegradation of every xenobiotic is determined with standardized laboratory procedures. The results obtained from these tests provide information allowing for comparison and classiﬁcation of chemicals according to their biodegradability. However the results of these tests cannot be easily extrapolated to environmental conditions. According to OECD guidelines if a substance reaches thresholds of ready biodegradability in 301 series of tests it should be easily and rapidly degraded in the environment. We have shown hereby that this is not always the case. Acknowledgements Financial support was provided by the Polish Ministry of Science and Higher Education grants Nos.: N N305 317040, N N305 359138 and N N305 320636. This research was also supported by the European Union within the European Social Fund in the framework of the project “InnoDoktorant – Scholarships for PhD students, 2nd edition”. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.jhazmat.2011.08.053.

References [1] P.-G. Rieger, H.-M. Meier, M. Gerle, U. Vogt, T. Groth, H.-J. Knackmuss, Xenobiotics in the environment: present and future strategies to obviate the problem of biological persistence, J. Biotechnol. 94 (2002) 101–123. [2] N.V. Plechkova, K.R. Seddon, Ionic liquids: “designer” solvents for green chemistry, in: A. Perosa, F. Zecchini, P. Tundo (Eds.), Methods and Reagents for Green Chemistry: an Introduction, Wiley-Interscience, 2007. [3] N.V. Plechkova, K.R. Seddon, Applications of ionic liquids in the chemical industry, Chem. Soc. Rev. 37 (2008) 123–150. [4] P.R. Wasserscheid, V.A. Bösmann Hal, 1-n-Butyl-3-methylimidazolium ([bmim]) octylsulfate – an even ‘greener’ ionic liquid, Green Chem. 4 (2002) 400–404. [5] S.T. Handy, Greener solvents: room temperature ionic liquids from biorenewable sources, Chem. Eur. J. 9 (2003) 2938–2944. [6] J.C. Spain, P.A.V. Veld, Adaptation of natural microbial communities to degradation of xenobiotic compounds: effects of concentration, exposure time, inoculum, and chemical structure, Appl. Environ. Microbiol. 45 (1983) 428–435. [7] M. Aelion, M. Swindoll, F.K. Pfaender, Adaptation to and biodegradation of xenobiotic compounds by microbial communities from a pristine aquifer, Appl. Environ. Microbiol. 53 (1987) 2212–2217. [8] M. Devers, N. Rourad, F. Martin-Laurent, Fitness drift of an atrazine-degrading population under atrazine selection pressure, Environ. Microbiol. 10 (2008) 676–684. [9] N. Nyhoim, F. lngerslev, U.T. Berg, J.P. Pedersen, H. Frimer-Larsen, Estimation of kinetic rate constants for biodegradation of chemicals in activated sludge wastewater treatment plants using short term batch experiments and ␮g/L range spiked concentrations, Chemosphere 33 (1996) 851–864. [10] M. Kulkarni, A. Chaudhari, Microbial remediation of nitro-aromatic compounds: an overview, J. Environ. Manage. 85 (2007) 496–512. [11] A. Modelli, A. Sali, P. Galletti, C. Samori, Biodegradation of oxygenated and nonoxygenated imidazolium-based ionic liquids in soil, Chemosphere 73 (2008) 1322–1327. [12] S. Stolte, S. Abdulkarim, J. Arning, A.-K. Blomeyer-Nienstedt, U. Bottin-Weber, M. Matzke, J. Ranke, B. Jastorff, J. Thöming, Primary biodegradation of ionic liquid cations, identiﬁcation of degradation products of 1-methyl-3-octylimidazolium chloride and electrochemical waste water treatment of poorly biodegradable compounds, Green Chem. 10 (2008) 214–224. [13] M.T Garcia, N. Gathergood, P.J. Scammells, Biodegradable ionic liquids. Part II. Effect of the anion and toxicology, Green Chem. 7 (2005) 9–14. [14] N. Gathergood, M.T. Garcia, P.J. Scammells, Biodegradable ionic liquids. Part I. Concept preliminary targets and evaluation, Green Chem. 6 (2004) 166–175. [15] N. Gathergood, P.J. Scammells, M.T. Garcia, Biodegradable ionic liquids. Part III. The ﬁrst readily biodegradable ionic liquids, Green Chem. 8 (2006) 156–160. [16] R.M. Goldstein, L.M. Mallory, M. Alexander, Reasons for possible failure of inoculation to enhance biodegradation, Appl. Environ. Microbiol. 50 (1985) 977–983. [17] A. Romero, A. Santos, J. Tojo, A. Rodrıguez, Toxicity and biodegradability of imidazolium ionic liquids, J. Hazard. Mater. 151 (2008) 268–273. [18] M. Markiewicz, C. Jungnickel, A. Markowska, U. Szczepaniak, M.J. Hupka Paszkiewicz, 1-Methyl-3-octylimidazolium chloride-sorption and primary biodegradation analysis in activated sewage sludge, Molecules 14 (2009) 4396–4405. [19] P. Reuschenbach, U. Pagga, U. Strotmann, A critical comparison of respirometric biodegradation tests based on OECD 301 and related test methods, Water Res. 37 (2003) 1571–1582. [20] K.M. Docherty, J.K. Dixon, C.F. Kulpa Jr., Biodegradability of imidazolium and pyridinium ionic liquids by an activated sludge microbial community, Biodegradation 18 (2007) 481–493. [21] OECD, OECD guideline for testing of chemicals 301 – Ready Biodegradability, 1992. [22] P. Stepnowski, A. Müller, P. Behrend, J. Ranke, J. Hoffmann, B. Jastorff, Reversedphase liquid chromatography of selected room-temperature ionic cations, J. Chromtogr. A 993 (2003) 173–178. [23] W. Den, C. Huang, C.-H. Li, Effects of cross-substrate interaction on biotrickling ﬁltration for the control of VOC emissions, Chemosphere 57 (2004) 697–709. [24] D.L. Lewis, H.P. Kollig, R.E. Hodson, Nutrient limitation and adaptation of microbial populations to chemical transformations, Appl. Environ. Microbiol. 51 (1986) 598–603. [25] C.M. Swindoll, C.M. Aelion, F.K. Pfaender, Inﬂuence of inorganic and organic nutrients on aerobic biodegradation and on the adaptation response of subsurface microbial communities, Appl. Environ. Microbiol. 54 (1987) 212–217. [26] K. Piekarska, J. Pawlak, Ocena skuteczno´sci osadu czynnego w oczyszczaniu ´ 1 (1994) 21–26. s´ cieków zaolejonych, Ochrona Srodowiska [27] U.L. Zweifel, B. Norrman, A. Hagström, Consumption of dissolved organic carbon by marine bacteria and demand for inorganic nutrients, Mar. Ecol. Prog. Ser. 101 (1993) 23–32. [28] E.A. Paul, F.E. Clark, Soil Microbiology and Biochemistry, Academic Press, San Diego, 1996.

Journal of Hazardous Materials 195 (2011) 383–390

Contents lists available at SciVerse ScienceDirect

Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat

A comparative study of experimental optimization and response surface optimization of Cr removal by emulsion ionic liquid membrane Rahul Kumar Goyal, N.S. Jayakumar, M.A. Hashim ∗ Department of Chemical Engineering, University of Malaya, Malaysia

a r t i c l e

i n f o

Article history: Received 24 March 2011 Received in revised form 21 July 2011 Accepted 19 August 2011 Available online 26 August 2011 Keywords: Emulsion ionic liquid membrane Chromium Response surface methodology [BMIM]+ [NTf2 ]− Optimization

a b s t r a c t A comparative study on the optimization of process parameters of an emulsion ionic liquid membrane (EILM) by experimental work and response surface methodology (RSM) has been carried out. EILM was prepared by using kerosene as solvent, Span 80 as surfactant, NaOH as internal reagent, a hydrophobic ionic liquid 1-butyl-3-methylimidazolium bis(triﬂuoromethylsulfonyl)imide ([BMIM]+ [NTf2 ]− ) as a stabilizer and a second ionic liquid tri-n-octylmethylammonium chloride (TOMAC) as a carrier. The prepared EILM was used to separate and concentrate Cr from wastewaters. The comparison between the experimentally optimized and the RSM optimized values was accomplished by optimizing the following parameters: homogenization speed, carrier concentration, internal phase concentration, agitation speed, treat ratio, internal to membrane phase ratio, surfactant concentration and pH of the feed phase. The comparison showed that all the values were in good agreement except for the internal phase concentration and the treat ratio. It was observed that the stability provided by [BMIM]+ [NTf2 ]− decreased as the extraction progressed due to its high density. Nevertheless, a good stability could be obtained by the combination of [BMIM]+ [NTf2 ]− and Span 80 during extraction process. © 2011 Published by Elsevier B.V.

1. Introduction Hexavalent Cr is a very toxic form of the Cr metal compared to trivalent Cr (III) due to its oxidizing nature. Cr (VI) is not only carcinogenic but causes many serious health problems to the biological system such as nose bleeding, respiratory problems, skin rashes, etc. Cr (VI) also affects wild life in a signiﬁcant amount that has raised serious concerns in our eco-system. World Health Organization has declared Cr (VI) removal from wastewater as a serious and prime research topic in 1998 meeting held in Geneva. The development of new technologies for the removal of Cr is still progressing day by day [1–4]. Emulsion liquid membrane (ELM) has been proven to be a competent extraction technology for the removal of Cr from industrial wastewaters. In this process, a prepared emulsion is dispersed into a feed phase which contains the solute to be removed. Single stage operation and less power consumption make this technology more favorable over pressure-driven membrane process and solvent extraction [5–8]. There are various parameters which affect the efﬁciency of ELM such as homogenization speed, carrier concentration, internal phase concentration, agitation speed, treat ratio, internal to membrane phase ratio, surfactant concentration, pH of

∗ Corresponding author. Fax: +60 3 79675319. E-mail address: [email protected] (M.A. Hashim). 0304-3894/$ – see front matter © 2011 Published by Elsevier B.V. doi:10.1016/j.jhazmat.2011.08.056

the feed phase and the concentration of stabilizer. The maximum percentage removal of Cr is obtained when all these parameters have optimized values. These values can be estimated either by experiments involving the variation in the value of one parameter at a time while keeping other parameters at constant values or by using response surface methodology [9–11]. Recent advances in the applications of ionic liquids have drawn a signiﬁcant attention of researchers all over the world. Room temperature ionic liquids are composed of organic cations and organic or inorganic anions. The physical and chemical properties of RTILs can be modiﬁed by changing the cation or anion or both to facilitate a particular task. These properties include negligible vapour pressure, inﬂammability, thermal stability even at high temperatures and application based adjustable miscibility/immiscibility in chemical processes [12–17]. Ionic liquids possess a very negligible vapour pressure that has enabled them to be used as a “green solvent” in synthesis [15,16,18–20], separation and puriﬁcation [21–27] and electrochemical application [27]. In our recent work, we have shown that ionic liquid [BMIM]+ [NTf2 ]− acted as a stabilizer in ELM system when kerosene as solvent, Span 80 as surfactant, NaOH (0.1N) as internal phase and TOMAC as carrier was used (article in press, Journal of Hazardous material). It was found out that 3% (w/w) of [BMIM]+ [NTf2 ]− provided substantial amount of stability during the extraction process with the least possible resistance to the mass transfer of Cr. In this study, EILM (emulsion ionic liquid membrane) has been used to extract Cr from wastewaters.

384

R.K. Goyal et al. / Journal of Hazardous Materials 195 (2011) 383–390

Response surface methodology (RSM) has attracted a substantial deal of interest in the past 2–3 decades. It is a statistical tool that is used nowadays very commonly to optimize experimental parameters, to model and to analyze a response of interest. RSM not only provides the optimum level for each variable but also estimates interactions among them and their impact on one or more measured responses [28,29]. This methodology cuts down the number of experiments by a substantial amount without affecting the interactions among the parameters. RSM has been utilized to optimize the parameters of several biotechnological processes such as fermentation, biosorption of metals, oxidation and catalyzed reaction conditions [30–32]. It has also been a useful tool to optimize widely employed parameters such as temperature, pH, stirring speed, concentration of various phases, and aeration rate in several processes [33–36]. In literature, there is not much work available on the comparison between the experimental and RSM optimization of Cr removal by EILM. Therefore, a statistical optimization of the parameters such as homogenization speed, carrier concentration, internal phase concentration, agitation speed, treat ratio, internal to membrane phase ratio, surfactant concentration, pH of the feed phase was carried out to study the individual and interactive effects of parameters on the extraction of Cr using ELM, and the optimized values were compared with the values obtained from the individual experimental optimization of the parameters. 2. Materials and methods 2.1. Chemicals All the chemicals described here were used without any further puriﬁcation. Ionic liquids [BMIM]+ [NTf2 ]− and TOMAC were procured from Merck (Germany). Span 80 a non-ionic surfactant was kindly supplied by Merck (Malaysia). ACROS (USA) kindly provided kerosene of boiling point ranged from 180 to 280 ◦ C. Sodium hydroxide pellets, potassium dichromate and hydrochloric acid were obtained from R&M Chemicals (UK). The solution of sodium hydroxide of desired normality was prepared by dissolving appropriate weight of pellets in de-ionized water. Similarly, Cr solution of 100 mg/L was prepared by mixing suitable amount of potassium dichromate in de-ionized water. 2.2. Analytical instruments The preparation of emulsion was carried out using a high speed homogenizer (IKA, model: T25 digital Ultra Turrax) and subsequently it was dispersed in the feed phase by using a stirrer (IKA, model: RW11 Lab Egg). An ICP-spectrophotometer (PerkinElmer, model: Optima 7000 DV) was used for the measurement of the Cr concentration. pH of the feed phase was quantiﬁed using a CyberScan 510 pH meter.

Table 1 Factors and their corresponding values. Factors

Levels

Homogenization speed (rpm), X1 Internal phase concentration (%, w/w), X2 Carrier concentration (%, w/w), X3 Surfactant concentration (%, w/w), X4 pH of the feed phase, X5 Agitation speed (rpm), X6 Treat ratio, X7 Internal to organic phase ratio, X8

−1

0

4000 0.05 0.1 1 0.2 100 1 0.25

7000 0.525 0.3 3 1.1 250 2 0.625

1 10,000 1 0.5 5 2 400 3 1

mechanical stirrer with the speed in the range of 200–400 rpm. The pH of the feed phase was initially taken lower than 2 to establish a pH difference between the stripping and the feed phase. Samples were taken on a regular interval by using syringes and the syringes were kept undisturbed for some time until the emulsion and the feed phase were separated. The feed phase was then taken out, ﬁltered and analyzed using ICP-spectrophotometer. 3. Results and discussion Design-Expert 7.16 software was used to analyze the data and to estimate the coefﬁcients of the regression equation. An orthogonal 24 Box–Behnken design (BBD) having ﬁve replicates at the centre point was used to estimate the coefﬁcients of the response function which is a second order polynomial as given by Eq. (1). It resulted in 29 experiments for four variables. Y = ˇ0 +

k i=1

ˇi Xi +

k i=1

ˇi Xi2 +

k k ii<j

j

ˇij Xi Xj + · · ·

(1)

where Y is the predicted response, ˇi , ˇj , and ˇij are the coefﬁcients estimated from regression, Xi is the uncoded value of the ith variable, i is the linear coefﬁcient, j is the quadratic coefﬁcient, and k is the number of factors. Due to the large number of variables, two sets of four variables each were formed, and are described as following: Set-I Homogenization speed (rpm), X1 Internal phase concentration (%, w/w), X2 Carrier concentration (%, w/w), X3 Surfactant concentration (%, w/w), X4

Set-II pH of the feed phase, X5 Agitation speed (rpm), X6 Treat ratio, X7 Internal to organic phase ratio, X8

The effect of ionic liquid [BMIM]+ [NTf2 ]− concentration on the removal of Cr was found to be insigniﬁcant experimentally. Therefore, it has been excluded from the parameter list. The coded and uncoded values, different levels and the range of the variables are given in Table 1

3.1. Interactions and optimization of the variables of set-I 2.3. Procedure 2.3.1. Preparation of emulsion The emulsion was prepared in a 100 mL beaker by mixing organic solvent, an appropriate amount of Span 80, TOMAC and [BMIM]+ [NTf2 ]− . The mixture was homogenized for up to 5 min. The drops of NaOH were added into the mixture by using a syringe, keeping the whole mixture homogenized for the next 5 min. Hence, we obtained the emulsion which appears as a milky white homogeneous solution. 2.3.2. Extraction of chromium The prepared emulsion was poured into another 250 mL beaker containing the feed phase. The mixture was stirred gently by using a

Box–Behnken design matrix for the set-I is given in Table 2 along with the experimental and the predicted response. The percentage removal efﬁciency of Cr was considered as the predicted response in DoE. Experiments were conducted in the same sequence as they are provided in the table, keeping all other four variables at constant values. pH of the feed phase was maintained at 0.5 while agitation speed of 300 was ﬁxed to achieve the maximum surface area. Treat ratio (F/E) and internal to organic phase ratio (I/O) were kept at the value of 2 and 1/3, respectively. The corresponding coefﬁcients of Eq. (1) for this set were obtained by regression analysis of the experimental data, and the equation in decoded form is shown in Eq. (2). In the equation, only the signiﬁcant variables are included.

R.K. Goyal et al. / Journal of Hazardous Materials 195 (2011) 383–390

385

Table 2 Box–Behnken design matrix for set-I. Run

X1

X2

X3

X4

% Removal of Cr Experimental

Predicted

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29

0 −1 0 −1 0 0 0 1 −1 0 −1 0 −1 1 1 0 0 0 0 −1 1 0 1 0 0 0 1 0 0

−1 0 0 −1 0 0 0 0 0 −1 0 0 1 0 −1 −1 −1 1 0 0 0 1 1 1 0 1 0 0 0

0 0 0 0 0 0 1 0 1 −1 0 0 0 0 0 1 0 0 −1 −1 −1 −1 0 0 −1 1 1 1 0

−1 −1 0 0 0 0 1 −1 0 0 1 0 0 1 0 0 1 −1 1 0 0 0 0 1 −1 0 0 −1 0

75.56 74.34 93.86 84.56 93.43 93.44 76.45 83.76 82.45 80.43 83.54 93.3 78.61 77.45 81.63 76.32 75.43 75.89 76.34 78.34 79.01 70.76 85.1 76.67 68.98 84.89 86.35 80.56 93.94

75.90 74.33 93.59 83.91 93.59 93.59 75.91 84.00 82.72 80.02 83.22 93.59 78.86 77.37 81.11 76.43 76.57 75.11 76.12 78.81 79.10 70.57 85.48 76.69 69.25 85.21 86.24 80.51 93.59

Fig. 2. Response surface plot for the interaction between the homogenization speed and the internal phase concentration.

100 Experimental vs Predicted

Predicted values

90

80

70

60 60

70

80

90

100

Experimental values Fig. 1. Predicted vs. experimental values for set-I.

Y = 93.59 + 0.95 ∗ X1 + 2.76 ∗ X3 + 0.57 ∗ X4 − 3.80 ∗ X12

3.2. Signiﬁcant interactions of the variables of set-I

− 7.46 ∗ X22 − 8.08 ∗ X32 − 10.07 ∗ X42 + 2.36 ∗ X1 ∗ X2 + 0.81 ∗ X1 ∗ X3 + 3.88 ∗ X1 ∗ X4 + 4.56 ∗ X2 ∗ X3 − 2.87 ∗ X3 ∗ X4

The Model F-value of 304.56 implies that the model is signiﬁcant. There is only a 0.01% chance that a “Model F-value” this large could occur due to noise. Values of “prob > F” less than 0.0500 indicate model terms are signiﬁcant and the rest are considered as insigniﬁcant.In this case X1 , X3 , X4 , X12 , X22 , X32 , X42 , X1 X2 , X1 X3 , X1 X4 , X2 X3 , X3 X4 are signiﬁcant model terms. Values greater than 0.1000 indicate the model terms are not signiﬁcant. The “Lack of Fit Fvalue” of 5.08 implies that it is not siginiﬁcant in comparison with the pure error. There is a 6.56% chance that a “Lack of Fit F-value” this large could occur due to noise. It is always necessary to have the value of “Lack of Fit F-value” non signiﬁcant to make the model best ﬁt. Predicted R2 represnts the prediction of a respose value estimated by the model. The difference between adjusted R2 and predicted R2 is always wanted to be in the range of 0–0.200 for the adequacy of the model. In this case, the differecne between them is .0113 which implies that both the values are in good agreement. Adequate Precision is an estimation of the signal to noise ratio. A ratio greater than 4 is desirable. The ratio of 59.787 implies an adequate signal. Hence, this model can be used to navigate the design space. Coefﬁcient of variation indicates the error expressed as a percentage of the mean. The response surface curves indicate the interaction of the variables and also determine the optimum level of variables for maximum response. The response surface plots for signiﬁcant interaction between two variables against % removal efﬁciency of Cr by EILM are, as shown in Figs. 2–6.

(2)

where Y is the percentage removal efﬁciency of Cr by ELM, X1 is the homogenization speed, X2 is the internal phase concentration, X3 is the carrier concentration and X4 is the surfactant concentration. The predicted values calculated from Eq. (2) were in very good agreement with the experimental values, as shown in Fig. 1. Hence, this quadratic model is well suited for this experimental set up. However, the signiﬁcance and the ﬁtness of the model was veriﬁed by using statistical test widey known as ANOVA (analysis of variance). ANOVA also helps to check the validity of the equation. ANOVA results are illustrated in Table 3.

The contour plot in Fig. 2 illustrates the interaction between the homogenization speed and the internal phase concentration/NaOH concentration. In the ﬁgure, the parabolic nature of contours implies that the interaction between both the variables is signiﬁcant. The homogenization speed and the internal phase concentration, both cause the increase in the % removal efﬁciency when their values were increased from lower level to up to a certain point. After this point, the % removal efﬁciency decreases, as illustrated in Fig. 2. The optimized values are provided in Section 3.3. The interaction between the homogenization speed and the internal phase concentration can be explained by the fact that the viscosity is directly proportional to internal phase concentration, and the internal phase droplet size is dependent on the viscosity and the homogenization speed. Thus, they are inter-related.

386

R.K. Goyal et al. / Journal of Hazardous Materials 195 (2011) 383–390

Table 3 ANOVA for response surface model of set-I. Source

Sum of squares

DF

Mean square

F-value

P-value, prob > F

Model X1 X2 X3 X4 X12 X22 X32 X42 X1 X2 X1 X3 X1 X4 X2 X3 X2 X4 X3 X4 Residual Lack of Fit Pure error Cor Total

1366.76 10.94 0.34 91.63 3.84 93.52 360.81 423.43 657.53 22.18 2.61 60.14 83.17 0.21 32.89 4.49 4.16 0.33 1371.25

14 1 1 1 1 1 1 1 1 1 1 1 1 1 1 14 10 4 28

97.63 10.94 0.34 91.63 3.84 93.52 360.81 423.43 657.53 22.18 2.61 60.14 83.17 0.21 32.89 0.32 0.42 0.082

304.56 34.14 1.05 285.86 11.99 291.74 1125.63 1320.96 2051.30 69.21 8.14 187.62 259.48 0.65 102.61

<0.0001 <0.0001 0.3228 <0.0001 0.0038 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 0.0128 <0.0001 <0.0001 0.4350 <0.0001

5.08

0.0656

signiﬁcant

not signiﬁcant

Std. deviation: 0.57; mean: 81.43; coefﬁcient of variation: 0.70; R-squared: 0.9967; adjusted R-squared: 0.9935; predicted R-squared: 0.9822; adequate precision: 59.787

Fig. 3. Response surface plot for the interaction between homogenization speed and the carrier concentration.

Fig. 5. Response surface plot for the interaction between the internal phase concentration and the carrier concentration.

The interaction between the homogenization speed and the carrier concentration is, as shown in Fig. 3. Parabolic contours signify that the interaction between them is signiﬁcant. The curve illustrates that both the values increase the % removal efﬁciency upon

increment from lower level, but after certain values, the % removal efﬁciency tend to decline until the higher level. The optimum values can be found out easily since the contours are parabolic. The reason for signiﬁcant interaction can be given as the viscosity of EILM is dependent on the carrier concentration, and the small sized globules formation is dependent on the viscosity and

Fig. 4. Response surface plot for the interaction between the homogenization speed and the surfactant concentration.

Fig. 6. Response surface plot for the interaction between the carrier concentration and the surfactant concentration.

R.K. Goyal et al. / Journal of Hazardous Materials 195 (2011) 383–390

homogenization speed. TOMAC increases the viscosity of EILM that hinders the effect of the homogenization speed in order to form the small size globules. Fig. 4 demonstrates the interaction between the homogenization speed and the surfactant concentration. % removal of Cr increases upon increasing the concentration of Span 80 up to a certain value. However, it decreases with further increment. Similar kind of results were obtained for the homogenization speed meaning that both are having optimum values at which maximum % removal can be achieved. Since contours are parabolic, the interaction between them is signiﬁcant. The reason for the signiﬁcant interaction can be provided by the nature of the surfactant which helps to protect the internal droplets from high shear and stress caused by homogenization speed. Surfactant concentration also contributes in making ﬁne droplets of internal phase under a reasonable homogenization speed. Interaction between the internal phase concentration and the carrier concentration is illustrated in Fig. 5. It can be observed from the ﬁgure that the contours are parabolic, which indicates that the interaction is signiﬁcant. % removal of Cr increases upon increasing the values of both the parameters from the lower range up to certain values. After these values, the % removal of Cr decreases. The signiﬁcance of the interaction can be explained by the reaction between TOMAC and NaOH which takes place at the inner interface. The reaction not only affects the stripping reactions but also the overall removal efﬁciency. Another important interaction between them is the ﬁght for the limiting reagent, since carrier and internal phase are involved in the extraction and stripping reaction, respectively. Hence, they are highly interactive parameters in EILM process. Fig. 6 shows the interaction between the carrier concentration and the surfactant concentration. Circular contours of the ﬁgure imply that the interaction between the parameters is not signiﬁcant. Hence, the optimum values of the variables are not easy to ﬁnd. However, increasing the surfactant concentration and carrier concentration increases the % removal efﬁciency up to certain values. After that, the removal efﬁciency decreases upon further increment in both the values. The trend of the curve is more or less similar to the ones described earlier for other variables’ interactions except the contours.

3.3. Optimization of the variables of set-I The statistical optimization of all four parameters was done by design of experiments (DoE) 7.04. DoE resulted in 10 different solutions with having % removal of Cr almost same in all of them. However, the solution with minimum carrier concentration was selected, since it is the most expensive chemical in comparison with the rest. The coded values of the homogenization speed, the internal phase concentration, the carrier concentration and the surfactant concentration are 0.22, 0.05, 0.18 and −0.03, respectively. The % removal efﬁciency was predicted at the value of 93.94% under these optimized values. The uncoded values for the homogenization speed and the internal phase concentration were calculated from coded values are 7660 rpm and 0.548 M, respectively. Similarly, the carrier concentration and the surfactant concentration are 0.336 and 3.06% (w/w), respectively. All the optimum values are in good agreement with the experimental optimization except for the value of the internal phase concentration. The reason for this discrepancy may be given as the large value range chosen (0.05–1 M) for the internal phase concentration. However, model results are signiﬁcant and well suited with the experimental results.

387

Table 4 Box–Behnken design matrix for set-II. Run

X1

X2

X3

X4

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29

−1 0 0 1 0 0 1 0 1 0 0 0 0 0 0 −1 1 0 1 0 0 −1 0 0 0 −1 −1 −1 1

0 0 −1 −1 −1 0 1 0 0 1 0 0 0 0 −1 −1 0 1 0 0 1 0 −1 0 1 0 1 0 0

0 0 −1 0 0 1 0 −1 0 −1 0 1 0 0 1 0 −1 1 1 0 0 1 0 −1 0 0 0 −1 0

1 0 0 0 −1 1 0 1 −1 0 0 −1 0 0 0 0 0 0 0 0 1 0 1 −1 −1 −1 0 0 1

% Removal of Cr Experimental

Predicted

85.43 94.69 69.45 50.23 73.48 89.92 48.29 70.69 55.45 78.43 95.02 78.43 93.7 95.32 78.64 79.41 44.98 84.12 56.32 93.72 84.26 85.78 76.68 78.33 82.89 84.01 89.87 80.32 52.23

87.20 94.49 68.97 50.36 75.13 89.02 50.83 70.95 53.21 77.38 94.49 78.35 94.49 94.49 79.22 77.05 44.83 84.13 56.27 94.49 82.90 86.22 77.15 79.41 82.71 83.78 89.91 80.66 51.99

3.4. Interactions and optimization of the variables of set-II Similarly, Box–Behnken design matrix for the set-II was obtained by DoE 7.04 and is given in Table 4. Experiments were conducted in the same sequence as they are provided in the table, keeping all other four variables at constant optimized values as predicted by set-I analysis. The homogenization speed was maintained at 7660 rpm while the internal phase concentration of the value of 0.548 M was ﬁxed to achieve the optimum stripping reagent. The carrier concentration and the surfactant concentration of the values of 0.336 and 3.06% (w/w) were taken respectively, to carry out the experiments. The corresponding coefﬁcients of Eq. (1) for this set were obtained by regression analysis of the experimental data, and it is shown in Eq. (3) in decoded form. In the equation, insigniﬁcant terms are discarded and only signiﬁcant terms have been included: Y = 94.49 − 16.44 ∗ X5 + 3.33 ∗ X6 + 4.25 ∗ X7 − 18.94 ∗ X52 − 8.51 ∗ X62 − 8.55 ∗ X72 − 6.50 ∗ X82 − 3.10 ∗ X5 ∗ X6 + 4.78 ∗ X7 ∗ X8

(3)

where Y is the percentage removal efﬁciency of Cr by ELM, X5 is the pH of the feed phase, X6 is the agitation speed, X7 is the treat ratio and X8 is the internal to membrane phase ratio. The predicted values calculated from Eq. (2) were in very good agreement with the experimental values, as shown in Fig. 7. Hence, this quadratic model is well suited for this experimental set up. Furthermore, the validation and the ﬁtness of the model were afﬁrmed by ANOVA (analysis of variance). ANOVA results are shown in Table 5. The Model F-value of 202.94 implies that the model is signiﬁcant. There is only a 0.01% chance that a “Model F-value” this large could occur due to noise. Values of “prob > F” less than 0.0500 indicate model terms are signiﬁcant and the rest are considered as insigniﬁcant.In this case X5 , X6 , X7 , X52 , X62 , X72 , X82 , X5 X6 and X7 X8

388

R.K. Goyal et al. / Journal of Hazardous Materials 195 (2011) 383–390

Table 5 ANOVA for response surface model of set-II. Source

Sum of squares

DF

Mean square

F-value

P-value, prob > F

Model X5 X6 X7 X8 X52 X62 X72 X82 X5 X6 X5 X7 X5 X8 X6 X7 X6 X8 X7 X8 Residual Lack of Fit Pure error Cor Total

6374.46 3244.60 133.13 216.84 3.65 2326.86 469.89 474.59 274.48 38.44 8.64 5.38 3.06 0.84 91.49 31.41 29.18 2.23 6405.87

14 1 1 1 1 1 1 1 1 1 1 1 1 1 1 14 10 4 28

455.32 3244.60 133.13 216.84 3.65 2326.86 469.89 474.59 274.48 38.44 8.64 5.38 3.06 0.84 91.49 2.24 2.92 0.56

202.94 1446.14 59.34 96.64 1.63 1037.10 209.43 211.53 122.34 17.13 3.85 2.40 1.36 0.37 40.78

<0.0001 <0.0001 <0.0001 <0.0001 0.2228 <0.0001 <0.0001 <0.0001 <0.0001 0.0010 0.0699 0.1437 0.2622 0.5511 <0.0001

5.24

0.0622

signiﬁcant

not signiﬁcant

Std. deviation: 50; mean: 76.90; coefﬁcient of variation: 1.95; R-squared: 0.9951; adjusted R-squared: 0.9902; predicted R-squared: 0.9732; adequate precision: 46.096.

are signiﬁcant model terms. Values greater than 0.1000 indicate the model terms are not signiﬁcant. The “Lack of Fit F-value” of 5.24 implies that it is not siginiﬁcant in comparison with the pure error. There is a 6.22% chance that a “Lack of Fit F-value” this large could occur due to noise. It is always necessary to have the value of “Lack of Fit F-value” non signiﬁcant to make the model best ﬁt. Predicted R2 represents the prediction of a respose value estimated by the model. The difference between adjusted R2 and predicted R2 is always wanted to be in the range of 0–0.200 for the adequacy of the model. In this case, the difference between them is 0.0173 which implies that both the values are in good agreement. Adequate precision is an estimation of the signal to noise ratio. A ratio greater than 4 is desirable. The ratio of 46.096 implies an adequate signal. Hence, this model can be used to navigate the design space. Coefﬁcient of variation indicates the error expressed as a percentage of the mean. Fig. 8. Response surface plot for the interaction between the pH of the feed phase and the agitation speed.

3.5. Signiﬁcant interactions of the variables of set-II There were only two signiﬁcant interactions for set-II, as previously shown by ANOVA. One was the interaction between the pH of the feed phase and the agitation speed and the other one was between the treat ratio and the internal to membrane phase ratio. The response curves of both are shown in Figs. 8 and 9. Fig. 8 describes the interaction of pH of the feed phase and the agitation speed. Since contours of the curve are parabolic, it implies

that the interaction is signiﬁcant and the optimum values can be calculated easily. % removal of Cr increases upon decreasing the pH of the feed phase but after a certain value, the removal decreases with the further decrease in pH. Similar kind of trend can also be observed for agitation speed. It implicates that both are having

100 Experimental vs Predicted

Predicted values

90 80 70 60 50 40 40

50

60

70

80

90

100

Experimental values Fig. 7. Predicted vs. experimental values for set-II.

Fig. 9. Response surface plot for the interaction between the treat ratio and the internal to membrane phase ratio.

R.K. Goyal et al. / Journal of Hazardous Materials 195 (2011) 383–390 Table 6 Comparison of experimentally optimized values and RSM optimized values. Parameter

Experimentally optimized

RSM optimized

Homogenization speed (rpm) Internal phase concentration (M) Carrier concentration (%, w/w) Surfactant concentration (%, w/w) pH of the feed phase Agitation speed (rpm) Treat ratio Internal to membrane phase ratio

8000 0.1 0.3 3.0 0.5 300 2 0.337

7660 0.548 0.336 3.06 0.425 296.5 2.60 0.715

certain optimum values on which maximum removal of Cr can be achieved. Physically, the interaction between the pH of the feed phase and the agitation speed can be explained by the fact that the swelling in EILM is caused by the pH of the feed phase and by the agitation speed also. Therefore, swelling acts as a mediator to regulate the interaction between them. The osmotic pressure difference between the internal and the feed phase is also dependent on the pH of the feed phase and the agitation speed. This may also act as a connecting parameter for this interaction. Fig. 9 shows the interaction between the treat ratio and the internal to membrane phase ratio. Parabolic contours signify that the interaction is quite signiﬁcant. From the ﬁgure it can be predicted that the increment in the treat ratio increases the % removal of Cr up to the upper level while internal to membrane phase ratio achieves maxima in between the lower and the upper level. The possible reason for the interaction can be provided as the direct proportionality of the internal to membrane phase ratio to the treat ratio. An increment in the internal to membrane phase ratio increases the volume of ELM, and hence decreases the treat ratio. Therefore, they have a good interaction for ELM process.

3.6. Optimization of the variables of set-II The statistical optimization of all four parameters was done by design of experiments (DoE) 7.04. DoE resulted in 10 different solutions with having % removal of Cr up to 97.5. However, the solution with the maximum treat ratio has been selected to economize the ELM process. The coded values of the pH of the feed phase, the agitation speed, the treat ratio and the Internal to membrane phase ratio are −0.75, 0.31, 0.60 and 0.24, respectively. The % removal efﬁciency was predicted at the value of 96.38% under these optimized values. The uncoded values for the pH of the feed phase and the agitation speed were calculated from coded values are 0.425 and 296.5 rpm, respectively. Similarly, the treat ratio and the internal to membrane phase ratio of the values of 2.60 and 0.715 respectively were obtained after converting coded values to uncoded values. All the above optimum values were found to be in good agreement with the experimental optimization, except for the value of the treat ratio. However, model results are signiﬁcant and well suited with the experimental results. Table 6 shows that comparison of the optimized values calculated experimentally and by RSM. It shows that the all the values are in good agreement except for the internal phase concentration and the treat ratio. This difference is dependent on the range and levels deﬁned for RSM, swelling of ELM and other experimental errors. However, the model can be used to predict the % removal of Cr by ELM.

389

4. Conclusion An emulsion ionic liquid membrane (EILM) was prepared using two ionic liquids, i.e. TOMAC as an extractant and [BMIM]+ [NTf2 ]− as a stabilizer for the separation of Cr. The parameters such as carrier concentration, internal phase concentration, agitation speed, treat ratio, internal to membrane phase ratio, surfactant concentration and pH of the feed phase were optimized separately by individual experimental work and by response surface methodology. The comparison between two optimized values showed that they were in good agreement except for the internal phase concentration and the treat ratio. Interaction between two parameters suggested their inter-dependence on each other and their effect on the ﬁnal percentage removal of Cr. An effort to lessen the use of organic solvent was made to economize ELM process and to make it more environmentally friendly. Ionic liquid [BMIM]+ [NTf2 ]− hinders mass transfer of Cr by its polymeric form but the effect was quite insigniﬁcant. List of symbols feed to emulsion phase ratio F/E I/O internal to organic phase ratio X1 homogenization speed (rpm) X2 internal phase concentration (%, w/w) carrier concentrations (%, w/w) X3 X4 surfactant concentration (%, w/w) X5 pH of the feed phase X6 agitation speed (rpm) treat ratio X7 X8 internal to organic phase ratio Acknowledgement The authors are grateful to University of Malaya, Malaysia for providing the fund to carry out this research work. References [1] M. Chiha, M.H. Samar, O. Hamdaoui, Extraction of chromium (VI) from sulphuric acid aqueous solutions by a liquid surfactant membrane (LSM), Desalination 194 (2006) 69–80. [2] A. Zouhri, B. Ernst, M. Burgard, Bulk liquid membrane for the recovery of chromium(VI) from a hydrochloric acid medium using dicyclohexano-18crown-6 as extractant-carrier, Sep. Sci. Technol. 34 (1999) 1891–1905. [3] G. Arslan, A. Tor, H. Muslu, M. Ozmen, I. Akin, Y. Cengeloglu, M. Ersoz, Facilitated transport of Cr(VI) through a novel activated composite membrane containing Cyanex 923 as a carrier, J. Membr. Sci. 337 (2009) 224–231. [4] A. Agrawal, C. Pal, K.K. Sahu, Extractive removal of chromium (VI) from industrial waste solution, J. Hazard. Mater. 159 (2008) 458–464. [5] N.N. Li, Separating Hydrocarbons with Liquid Membranes, U.S. Patent 3,410,794 (1968). [6] R. Ali Kumbasar, Extraction of chromium (VI) from multicomponent acidic solutions by emulsion liquid membranes using TOPO as extractant, J. Hazard. Mater. 167 (2009) 1141–1147. [7] R.A. Kumbasar, Selective separation of chromium (VI) from acidic solutions containing various metal ions through emulsion liquid membrane using trioctylamine as extractant, Sep. Purif. Technol. 64 (2008) 56–62. [8] S.A. Cavaco, S. Fernandes, M.M. Quina, L.M. Ferreira, Removal of chromium from electroplating industry efﬂuents by ion exchange resins, J. Hazard. Mater. 144 (2007) 634–638. [9] P.S. Kulkarni, S. Mukhopadhyay, M.P. Bellary, S.K. Ghosh, Studies on membrane stability and recovery of uranium (VI) from aqueous solutions using a liquid emulsion membrane process, Hydrometallurgy 64 (2002) 49–58. [10] C.C. Lin, R.L. Long, Removal of nitric acid by emulsion liquid membrane: experimental results and model prediction, J. Membr. Sci. 134 (1997) 33–45. [11] H.R. Mortaheb, H. Kosuge, B. Mokhtarani, M.H. Amini, H.R. Banihashemi, Study on removal of cadmium from wastewater by emulsion liquid membrane, J. Hazard. Mater. 165 (2009) 630–636. [12] J.D. Holbrey, K.R. Seddon, Ionic Liquids, Clean Products and Processes, vol. 1, 1999, pp. 223–237. [13] M.J. Earle, K.R. Seddon, Ionic liquids. Green solvents for the future, Pure Appl. Chem. 72 (7) (2000) 1391–1398. [14] K. Mikami, Green Reaction Media in Organic Synthesis, ﬁrst ed., Blackwell, UK, 2005.

390

R.K. Goyal et al. / Journal of Hazardous Materials 195 (2011) 383–390

[15] T. Welton, Room-temperature ionic liquids. Solvents for synthesis and catalysis, Chem. Rev. 99 (1999) 2071–2083. [16] P. Wasserscheid, T. Welton, Ionic Liquids in Synthesis, second ed., Wiley-VCH, Weinheim, 2008. [17] M. Freemantle, New horizons for ionic liquids, Chem. Eng. News 79 (2001) 21–25. [18] C.J. Adam, M.J. Earle, G. Robert, K.R. Seddon, Friedel–Crafts reactions in room temperature ionic liquids, Chem. Commun. 19 (1998) 2097–2098. [19] P.J. Dyson, D.J. Ellis, D.C. Parker, T. Welton, Arene hydrogenation in a roomtemperature ionic liquid using a ruthenium cluster catalyst, Chem. Commun. 1 (1999) 25–26. [20] C.E. Song, E.J. Roh, Practical method to recycle a chiral (salen)Mn epoxidation catalyst by using an ionic liquid, Chem. Commun. 10 (2000) 837–838. [21] A.E. Visser, R.P. Swatloski, W.M. Reichert, R. Mayton, S. Sheff, A. Wierzbicki, J.H. Davis Jr., R.D. Rogers, Task-speciﬁc ionic liquids for the extraction of metal ions from aqueous solutions, Chem. Commun. 1 (2001) 135–136. [22] A.P. de los Rios, F.J. Hern ndez-Fern ndez, L.J. Lozano, S.S. nchez, M.J.I.C. God nez, Removal of metal ions from aqueous solutions by extraction with ionic liquids, J. Chem. Eng. Data 55 (2010) 605–608. [23] A.E. Visser, R.P. Swatloski, W.M. Reichert, S.T. Grifﬁn, R.D. Rogers, Traditional extractants in nontraditional solvents: group 1 and 2 extraction by crown ethers in room-temperature ionic liquids, Ind. Eng. Chem. Res. 39 (10) (2000) 3596–3604. [24] S. Dai, Y.H. Ju, C.E. Barnes, Solvent extraction of strontium nitrate by a crown ether using room temperature ionic liquids, J. Chem. Soc. Dalton Trans. (1999) 1201–1202. [25] L.C. Branco, J.G. Crespo, C.A.M. Afonso, High selective transport of organic compounds by using supported liquid membranes based on ionic liquids, Angew. Chem. Int. Ed. 41 (2002) 2771–2773. [26] A.G. Fadeev, M.M. Meagher, Opportunities for ionic liquids in recovery of biofuels, Chem. Commun. 3 (2001) 295–296.

[27] I.W. Sun, C.L. Hussey, Electrochemistry of niobium chloride and oxide chloride complexes in the basic aluminum chloride-1-methyl-3-ethylimidazolium chloride room-temperature ionic liquid, Inorg. Chem. 28 (1989) 2731–2737. [28] T. Ölmez, The optimization of Cr(VI) reduction and removal by electrocoagulation using response surface methodology, J. Hazard. Mater. 162 (2009) 1371–1378. [29] M.S. Bhatti, A.S. Reddy, A.K. Thukral, Electrocoagulation removal of Cr(VI) from simulated wastewater using response surface methodology, J. Hazard. Mater. 172 (2009) 839–846. [30] E. Kristo, C.G. Biliaderis, N. Tzanetakis, Modelling of the acidiﬁcation process and rheological properties of milk fermented with a yogurt starter culture using response surface methodology, Food Chem. 83 (2003) 437–446. [31] L.-S.T. Lai, C.-C. Pan, B.-K. Tzeng, The inﬂuence of medium design on lovastatin production and pellet formation with a high-producing mutant of Aspergillus terreus in submerged cultures, Process Biochem. 38 (2003) 1317–1326. [32] E.L. Soo, A.B. Salleh, M. Basri, R.N.Z.A. Rahman, K. Kamaruddin, Response surface methodological study on lipase-catalyzed synthesis of amino acid surfactants, Process Biochem. 39 (2004) 1511–1518. [33] S. Mannan, A. Fakhru’l-Razi, M.Z. Alam, Optimization of process parameters for the bioconversion of activated sludge by Penicillium corylophilum, using response surface methodology, J. Environ. Sci. 19 (2007) 23–28. [34] M. Rajasimman, R. Sangeetha, P. Karthik, Statistical optimization of process parameters for the extraction of chromium(VI) from pharmaceutical wastewater by emulsion liquid membrane, Chem. Eng. J. 150 (2009) 275–279. [35] S. Nosrati, N.S. Jayakumar, M.A. Hashim, Extraction performance of chromium(VI) with emulsion liquid membrane by Cyanex 923 as carrier using response surface methodology, Desalination 266 (2010) 286–290. [36] M. Rajasimman, R. Sangeetha, Optimization of process parameters for the extraction of chromium (VI) by emulsion liquid membrane using response surface methodology, J. Hazard. Mater. 168 (2009) 291–297.

Journal of Hazardous Materials 195 (2011) 391–397

Contents lists available at SciVerse ScienceDirect

Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat

Treatment and recycling of asbestos-cement containing waste F. Colangelo a , R. Ciofﬁ a,∗ , M. Lavorgna b , L. Verdolotti b , L. De Stefano c a

Department of Technology, University Parthenope, Naples, Italy Institute for Biomedical and Composite Materials – CNR, Naples, Italy c Institute for Microelectronics and Microsystems – CNR, Naples, Italy b

a r t i c l e

i n f o

Article history: Received 27 April 2011 Received in revised form 1 August 2011 Accepted 19 August 2011 Available online 26 August 2011 Keywords: Asbestos waste Milling Inertization Recycling Mortars

a b s t r a c t The remediation of industrial buildings covered with asbestos-cement roofs is one of the most important issues in asbestos risk management. The relevant Italian Directives call for the above waste to be treated prior to disposal on landﬁll. Processes able to eliminate the hazard of these wastes are very attractive because the treated products can be recycled as mineral components in building materials. In this work, asbestos-cement waste is milled by means of a high energy ring mill for up to 4 h. The very ﬁne powders obtained at all milling times are characterized to check the mineralogical and morphological transformation of the asbestos phases. Speciﬁcally, after 120 min of milling, the disappearance of the chrysotile OH stretching modes at 3690 cm−1 , of the main crystalline chrysotile peaks and of the ﬁbrous phase are detected by means of infrared spectroscopy and X-ray diffraction and scanning electron microscopy analyses, respectively. The hydraulic behavior of the milled powders in presence of lime is also tested at different times. The results of thermal analyses show that the endothermic effects associated to the neoformed binding phases signiﬁcantly increase with curing time. Furthermore, the technological efﬁcacy of the recycling process is evaluated by preparing and testing hydraulic lime and milled powder-based mortars. The complete test set gives good results in terms of the hydration kinetics and mechanical properties of the building materials studied. In fact, values of reacted lime around 40% and values of compressive strength in the range of 2.17 and 2.29 MPa, are measured. © 2011 Elsevier B.V. All rights reserved.

1. Introduction Asbestos-cement was extensively produced in Italy between 1904 and 1985, with the result that a huge quantity of such materials have been used all over the country and, according to the latest estimates, very large quantities of these are still on site [1,2]. Recent Italian directives (Environment Ministry Decrees 13 March 2003 and 3 August 2005) classify all asbestos-containing waste (ACW) as hazardous in line with the European Waste Catalogue code 170605* and requires its treatment prior to disposal in controlled landﬁlls [3,4]. The Italian Environment Ministry Decree n. 248/2004 lays down guidelines for the treatment, disposal and recycling of ACW and recommends that preference be given to those stabilization and inertization processes that favour recycling in order to reduce ACW-related hazards [5–8]. In particular, high energy milling (HEM) is mentioned as being able to ensure waste amorphisation through the mineralogical and morphological transformation of asbestos phases [9].

∗ Corresponding author. Tel.: +39 0815476732; fax: +39 0815476777. E-mail address: raffaele.ciofﬁ@uniparthenope.it (R. Ciofﬁ). 0304-3894/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.jhazmat.2011.08.057

HEM is a mechanical process that brings about the deformation, fracture and local welding of particles caused by high-energy collisions between grinding media (rings, rollers, balls hammers, etc.) and the asbestos-containing powders. The mechanical transfer of high energy to the powders may determine the destruction of the crystalline lattice as well as a major increase in speciﬁc surface area [10,11]. HEM was originally developed in the mining industry in order to obtain ﬁne powders but it has also been used for mechanical alloying, for instance in the scale production of copper and bronze ﬂakes [12]. Recently, HEM has been applied to the treatment of organic pollutants, such as the destruction of DDT in presence of calcium oxide [13], the dehalogenation of chlorobenzenes over calcium hydride [14,15], the detoxiﬁcation of PCB-polluted soils mixed with NaH and NaBH4 [16] and the destruction of n-C16 H34 in model systems containing alumina and silica [17]. Finally, following studies on the amorphisation of aluminosilicate during the micronization process [18], the HEM process has been applied to the treatment of pure asbestos minerals and ACW [19]. The results of spectrophotometric and diffractometric analyses have reported a complete transformation of the chemical and crystalline structure of asbestos. The chemical composition of asbestos-free and partially amorphous powders makes them suitable for use in numerous civil engineering applications and, indeed,

392

F. Colangelo et al. / Journal of Hazardous Materials 195 (2011) 391–397

Table 1 Chemical composition of asbestos containing powders, wt.%. Oxides

Without treatment

MP30

MP120

MP240

SiO2 Al2 O3 Fe2 O3 CaO Na2 O K2 O MgO MnO SO3 Loss of ignition

18.76 4.75 1.46 23.98 0.57 1.62 2.63 0.50 20.56 29.11

18.64 4.83 1.45 23.59 0.60 1.59 2.59 0.53 19.67 28.72

18.95 4.91 1.47 24.00 0.61 1.61 2.64 0.54 20.00 29.21

18.60 4.82 1.45 23.55 0.60 1.58 2.59 0.53 19.63 28.67

Table 2 Composition of hydratory mixtures. Milled powders

Mixtures Milled powder/calcium hydroxide weight ratio

MP30 MP120 MP240

50/50

60/40

70/30

A30 A120 A240

B30 B120 B240

C30 C120 C240

Table 3 Mixture proportion of mortars, g. Mortars

Lime

MP120

Sand

NP

H2 O

RM 50 M 60 M 70 M

120 150 120 90

– 150 180 210

1200 1200 1200 1200

180 –

150 150 150 150

Fig. 2. XRD spectrograms of asbestos-cement waste as received and after 120 and 240 min milling.

the nature of hydration products are determined. Finally, hydraulic mortars, containing 4 h milled powders, are cast, cured and tested to evaluate their mechanical properties. 2. Experimentals 2.1. ACW handling and milling treatment

it is well known that huge amounts of aluminosilicate amorphous powders can be recycled as a pozzolanic addition in the production of cement and mortars. Furthermore, the very high ﬁneness of these powders means that they can be used as a ﬁne artiﬁcial aggregate in the preparation of special mortars and concrete [20]. This work reports the results of HEM treatment of ACW and the recycling of the asbestos-free powders obtained. The pozzolanic activity of the powders is tested through the hydration of different binding mixtures containing commercial lime and 30%, 40% and 50% of milled material. Chemically bonded water, reacted lime and

Fig. 1. FT-IR spectrograms of asbestos-cement waste as received and after 30 and 120 min milling.

The ACW, partially wet as it had been treated prior to removal from the roof of the industrial building, was received sealed in double polyethylene bags (2 mm thick), according to the Italian Environment Ministry Decree 6 September 1994 and related acts. Each bag was opened inside a laminar ﬂow hood to prevent any ﬁbre dispersion into the laboratory. Some of the ACW, measuring about 10 cm in size, was dry crushed to a millimetric size in an agate mortar inside the hood. Crushed samples were carefully transferred to the ball mill enclosed in a vial. During all the above operations, the air was monitored by ﬁltration through cellulose ﬁlters and subsequent ﬁbre counting by means of scanning electron microscopy (SEM) analysis, according to the Italian Environment Ministry Decree 6 September 1994. The measured ﬁbre concentration never exceeded the units per litre threshold limit. The technicians in charge of these operations wore protective disposable full-body overalls and prescribed facial masks. The ACW, containing about 12% in weight of asbestos chrysotile, was milled using a FRITSCH Pulverisette 9 ring-mill, with a total ring mass of 3637 g, 350 cm3 volumetric capacity and 750 rpm ring rotation speed. Weighted amounts of waste were milled for 30, 120 and 240 min using a rings/material weight ratio of 60/1. After each treatment the obtained ﬁne powders were analyzed by Fourier transform infrared spectroscopy (FT-IR) and X-ray diffraction analysis (XRD). Nitrogen-BET analysis was also employed to evaluate the speciﬁc surface area of the powders. The chemical composition of the asbestos containing powders was determined by means of inductively coupled plasma (ICP) spectrophotometry analysis. The results relating to before milling powders (BMP) and after 30 (MP30), 120 (MP120) and 240 (MP240) min of treatment are reported in Table 1. It can be seen that the amounts of each component are very similar before and after the different milling times.

F. Colangelo et al. / Journal of Hazardous Materials 195 (2011) 391–397

393

Fig. 3. SEM micrographs of ACW, before (a) and after (b) 120 min of milling treatment at 2500× magniﬁcation.

Fig. 4. DTA thermograms of the mixtures B30, B120 and B240.

The very high content of SO3 may be due to weathering, as the roof was exposed to the attack of SO2 emissions present in the industrial area of Naples, but the increased SO3 content may also be attributable to prolonged acid rain attack.

2.3. Hardened mortars

2.2. Hydration behavior The milled materials were also characterized in terms of their hydraulic behavior in presence of hydrated lime. To this end, hydraTable 4 Chemically bonded water and reacted lime of the systems A, B and C at different curing time. Systems

A1 B1 C1 A2 B2 C2 A3 B3 C3

tory mixtures containing different amounts of milled powder and lime were prepared and tested. The compositions of all systems studied are reported in Table 2. The hydration process was carried out with a water to solid ratio equal to 0.5 at a temperature of 25 ◦ C and relative humidity of 100% for curing times of 7, 14, 28 and 56 days. Cylindrical specimens, 2 cm in diameter and 3 cm in height, were prepared for each experimental condition. After each curing time, specimen hydration was stopped by grinding under acetone and then the powders were dried with diethyl ether. The hydration kinetic was evaluated by means of quantitative determinations of chemically bonded water (CBW) and reacted lime. The former was measured by loss of ignition at 750 ◦ C, while the latter by extraction and titration of residual free lime, according to the Franke method [21]. The quantitative determination of CBW content was performed as follows. A sample of mass W0 at time 0 acquires a mass Wt after t days hydration. Of course, if X0 and Xt are the fraction weight loss (at 750 ◦ C) at time 0 and time t, respectively, it holds: W0 (1 − X0 ) = Wt (1 − Xt ). Then, W0 = Wt (1 − Xt )/(1 − X0 ). In conclusion, the percentage of CBW is: % CBW = (Wt − W0 )/W0 × 100. In this way, all the mix components that contribute to ignition loss are taken into account (CaCO3 , Ca(OH)2 , . . .). The neo-formed products’ composition and morphology were evaluated by means of differential thermal analysis (DTA) and SEM analyses, respectively.

Chemically bonded water (wt.%)

Reacted lime (wt.%)

Hydration time (days)

Hydration time (days)

7

14

28

56

7

14

28

56

3.1 8.0 11.7 8.1 3.1 8.1 4.1 9.1 9.5

6.2 7.2 12.1 7.0 3.2 9.3 9.3 11.5 8.8

11.5 7.0 13.6 4.8 4.6 10.7 12.3 16.4 12.0

11.8 9.7 22.6 3.5 7.1 14.8 18.7 24.4 6.3

5.8 18.5 24.1 14.5 13.4 26.4 18.7 18.9 17.5

11.5 23.8 28.0 16.7 13,6 31.3 30.1 31.3 23.8

20.8 27.1 29.1 27.7 15.3 39.3 36.7 38.4 32.0

21.7 30.2 35.3 29.1 18.8 41.3 41.5 42.1 41.1

Powder milled for 120 min (MP120) was employed to prepare different kinds of mortars in which 50/50 (50 M), 60/40 (60 M) and 70/30 (70 M) MP120 to lime ratios were used. In addition, a reference mortar (RM) containing 60% natural pozzolana (NP) and 40% lime was also prepared. The natural pozzolana was supplied by Italiana Zeoliti S.r.l. (Prignano sulla Secchia, Modena, Italy) and its chemical composition was as follows: SiO2 = 58.82%; Al2 O3 = 19.10%; Fe2 O3 = 4.60%; CaO = 3.10%; MgO = 1.11%; Na2 O = 3.44% and K2 O = 9.39%. According to EN 196-1 standard [22], normalized natural sand with maximum diameter less than 4 mm was employed as aggregate. The compositions of all mortars prepared are shown in Table 3. Three prismatic specimens (4 cm × 4 cm × 16 cm) of each kind of mortar were cast, cured and tested, according to UNI EN 196-1 standard [22]. The time, temperature and humidity of curing were 90 days, 20 ◦ C and 100% RH, respectively. Compressive strength was evaluated for all the hardened specimens by means of a Controls 50-C5600 apparatus.

394

F. Colangelo et al. / Journal of Hazardous Materials 195 (2011) 391–397

Fig. 5. SEM micrograph and EDS analysis of the B240 system after 28-day curing at 300× magniﬁcation.

3. Results and discussion 3.1. High energy milling Fig. 1 shows the result of FT-IR investigation carried out on the asbestos-cement as received and the two powders obtained by 30 min and 120 min milling. The curves show the disappearance of the chrysotile OH stretching modes at 3690 cm−1 after 120 min milling. The absorption bands centered at about 1445 and 880 cm−1 are due to ␯3 and ␯2 modes of CO3 −2 , those centered at about 1000 and 450 cm−1 are relative to ␯3 and ␯2 modes of SiO4 −4 and, ﬁnally, those centered at about 1150 and 700 cm−1 are given to ␯3 and ␯4 modes of SO4 −2 . The presence of silicates, carbonates and sulphates is obviously due to the nature of ACW which also contains set cement. As the external measured mill temperature, at equilibrium, never exceeds 80 ◦ C during the treatment, the OH bond disruption cannot in any way be attributed to a thermal action. On the

contrary, it is due to direct energy transfer between the grinding bodies and the processed materials through multiple impacts. The observed results are in line with previous ﬁndings obtained for silicate and aluminosilicate minerals [11,12]. Indeed, it is well known that these minerals undergo amorphization during micronization or pressure rising up to a few tens of GPa. From a structural point of view, the amorphization is due to a rapid fragmentation of sheets: the direct OH bonds between the layers are destroyed or opened. The net effect of this dehydroxylation is the collapse of the crystalline long range order and a consequent amorphization of materials. The water formed during this process may well be retained as adsorbed molecules on the solid surface which is extended and made more energetic by milling. The results shown in Fig. 1 reveal that even if asbestos mineral phases are present in a complex matrix (asbestos-cement) their behavior is the same. The results of XRD investigations are shown in Fig. 2. The reported traces refer to untreated and 120 min and 240 min high energy milled ACW, respectively. It is immediately clear that the

Fig. 6. SEM micrograph and EDS analysis of the B240 system after 28-day curing at 8000× magniﬁcation.

F. Colangelo et al. / Journal of Hazardous Materials 195 (2011) 391–397

395

Fig. 7. SEM micrograph of the B240 system after 28-day curing at 10,000× magniﬁcation.

main crystalline chrysotile peaks fall sharply (2 = 12.1◦ and 24.5◦ ) as a function of time during the micronization process. Indeed, after 120 min of milling these peaks practically disappear. Moreover, there is a clear and progressive increase in the background baseline between 5◦ 2 and 20.2◦ 2 which is a characteristic indication of a quantitative amorphous phases increase. Fig. 3 shows the SEM micrographs of ACW, before (a) and after (b) 120 min of milling treatment. It is well evident that the treated ACW does not contain any ﬁbres. It is also worth noting that the high energy milling process acts selectively in destroying the crystalline structure of asbestos. The CaCO3 XRD main characteristic peak (2 = 29.45), shown in Fig. 2, is almost the same after 240 min of milling. This effect is also conﬁrmed by the loss on ignition values reported in Table 1, which remain almost the same after 30, 120 and 240 min of treatment. Results of nitrogen-BET analysis show that the speciﬁc surface area does not increase proportionally to milling time: the measured values are: 16.18 m2 /g, 23.95 m2 /g and 18.92 m2 /g, after 30, 120 and 240 min of milling, respectively. After 120 min of milling, a 48% increase is observed compared to the powder milled for 30 min. The speciﬁc surface area then falls sharply in the following 120 min of treatment (240 min in total). This suggests a competition between particle size reduction and agglomeration. The destruction of asbestos ﬁbres due to the impact with milling bodies initially produces smaller particles so that a net increase in the speciﬁc surface area can be observed. Subsequently, these micrometric particles tend to agglomerate through electrostatic attraction and Van der Waals force, so that the speciﬁc surface area drastically decreases. From an economic point of view, the investment cost of an industrial ball milling plant can be derived from that of similar plants used in other environmental applications. This cost is orders of magnitude less than of thermal or chemical based inertization systems [23–26]. 3.2. Hydration behavior Fig. 4 reports the thermograms relative to the B30, B120 and B240 systems (see Table 2). The traces are relative to the uncured and to the 28 and 56-day cured samples for each system.

It is possible to observe that the endothermic peaks at about 80–100 ◦ C and 200–240 ◦ C related to the dehydration of neoformed phases, such as calcium silicate hydrates (CaO–SiO2 –H2 O, C–S–H) and calcium aluminate hydrates (CaO–Al2 O3 –H2 O, C–A–H), increase with curing time [27,28]. At the same time the endothermic peak at 450 ◦ C, dependent on dehydration of Ca(OH)2 reactant, decreases [29,30]. The endothermic peak at about 120–140 ◦ C, due to the dehydration of the neo-formed ettringite (6CaO·Al2 O3 ·3SO3 ·32H2 O), is less evident because it overlaps with the calcium silicate hydrated thermal effect. The presence of ettringite is more evident in the B120 and B240 systems where a shoulder is present at about 120–140 ◦ C. Endothermic effects present in the 600–800 ◦ C range are due to the decarboxylation of CaCO3 and remain almost the same even when the milling time of powders and the curing time of mixtures increase. This evidence conﬁrms the selectivity of HEM treatment as revealed by the XRD analysis. Other detected endothermic peaks are due to the unreacted mixture as they are already present before curing. The results of SEM analysis carried out on the B240 system after 28-days curing are shown in Figs. 5–7. In Fig. 5, areas with very different porosity are present and EDS analysis carried out on denser areas indicates that these parts are mainly composed of calcium carbonate. Figs. 6 and 7 report micrographs at a higher magniﬁcation in which the presence of ﬂaky C–S–H crystals (Fig. 6) and needle-like ettringite crystals (Fig. 7) are shown. In Fig. 5 the results of EDS analysis are also reported. The spectrum conﬁrms the typical C–S–H chemical composition of the ﬂaky zone. In Table 4 the weight percentages of chemically bonded water and reacted lime measured on systems A, B and C (see Table 1) are reported. The values are relative to 7, 14, 28 and 56 days curing time. Analysis of the values of chemically bonded water makes it possible to observe that a continuous rate is not present. This is due to the continuous change in the typology of the neo-formed hydration products. The amount of chemically bonded water in the various forms of the main hydrated phases, such as C–S–H; C–A–H and ettringite, and their hydration kinetics vary considerably. The stoichiometry of calcium silicate, calcium aluminate

396

F. Colangelo et al. / Journal of Hazardous Materials 195 (2011) 391–397

Table 5 Compressive strength of mortars after 90 day curing. Systems

Strength (MPa)

RM 50 M 60 M 70 M

1.10 2.26 2.29 2.17

and calcium trisulpho aluminate hydrates changes over the curing time, determining a non-continuous variation in the amount of non-evaporable water with time. In the case of reacted lime determination, it can be observed that the related amounts increase for all the systems when the curing time increases. Speciﬁcally, the values rapidly increase up to 14 days of curing, then almost constant values are reached after 28 days of the hydration time. These phenomena can be explained if we consider the very high speciﬁc surface area of the powders, which accelerates the hydration kinetic during the ﬁrst few days of hydration. Values of reacted lime around 40% conﬁrm that HEM powders are able to give hydration reactions that involve considerable amounts of lime. This is very useful for the formation of new phases with a good hydraulic behavior. 3.3. Mechanical characterization The compressive strength values measured on the cast mortars after 90 days curing are reported in Table 5. In this table values of 2.26 MPa, 2.29 MPa and 2.17 MPa are reported for the 50 M, 60 M and 70 M systems, respectively. These values are more than twice that measured for the reference mortar prepared with natural pozzolana: in the latter case a compressive strength value of 1.10 MPa, was determined. The good mechanical performance of the mortars containing milled powders are almost certainly due to the formation of ettringite together with C–S–H and C–A–H. This is in agreement with previous ﬁndings in which various hydratory systems able to form both ettringite and C–S–H have been studied [31–33]. Therefore, the presence of about 20% SO3 in the composition of milled powders gives these systems an added value compared to lime pozzolana-based standard hydraulic mortars. 4. Conclusions This study leads to the conclusion that powders obtained through the high energy milling of asbestos-cement waste are asbestos-free and can be proﬁtably recycled in the ﬁeld of building materials. Speciﬁcally, it can be drawn that: a) Two hours of high energy milling is sufﬁcient to ensure total amorphization of waste. In fact, the typical chrysotile X-ray diffraction peaks, spectroscopy infrared bands and ﬁbrous phase completely disappear after this mechanical treatment. This ﬁnding is of great interest considering the very high energy amount needed for the mineralogical transformation of asbestos waste by means of a much more expensive thermal treatment. b) The hydration processes of the lime-inert milled powders mixtures are very complex. The simultaneous formation of calcium silicate and aluminate hydrates and ettringite can be determined by means of differential thermal analyses and scanning electron microscopy observations. c) The mechanical characteristics of the hydraulic mortars prepared with the above mixtures are better than those limepozzolana based ones. These results prove the good pozzolanic activity of the employed asbestos-free powders and conﬁrm the

possibility of being recycled as a mineral addition in the manufacture of building materials.

References [1] C. Bianchi, T. Bianchi, Asbestos, A Century of Experimentation on Humans, ISTAT (Italian Institute for Statistic), Hammerle, Trieste, Italy, 2002, ISBN 88-7209017-2. [2] G. Chapman, World Mineral Statistics, Mineral Resources, vol. 28–29, British Geological Survey, 2000, www.bgs.ac.uk. [3] EU Commission Decision of 3 May 2000 replacing Decision 94/3/EC establishing a list of wastes pursuant to Article 1(a) of Council Directive 75/442/EEC on waste. [4] EU Council Decision 94/904/EC establishing a list of hazardous waste pursuant to Article 1(4) of Council Directive 91/689/EEC on hazardous waste. [5] A.F. Gualtieri, A. Tartaglia, Thermal decomposition of asbestos and recycling in traditional ceramics, J. Eur. Ceram. Soc. 20 (2000) 1409–1418. [6] C. Leonelli, P. Veronesi, D.N. Beccaccini, M.R. Rivasi, L. Barbieri, F. Andreola, I. Lancellotti, D. Rabitti, G.C. Pellicani, Microwave thermal inertization of asbestos containing waste and its recycling in traditional ceramics, J. Hazard. Mater. B135 (2006) 149–155. [7] A.F. Gualtieri, C. Cavenati, I. Zanatto, M. Meloni, G. Elmi, M. Lassinantti Gualtieri, The transformation sequences of cement-asbestos slates up to 1200 ◦ C and safe recycling of the reaction product in stone ware tile mixtures, J. Hazard. Mater. 152 (2008) 563–570. [8] Y.M. Chan, P. Agamauthu, R. Mahalingam, Solidiﬁcation and stabilization of asbestos waste from an automobile brake manufacturing facility using cement, J. Hazard. Mater. B77 (2000) 209–226. [9] P. Plescia, D. Gizzi, S. Benedetti, L. Camilucci, C. Fanizza, P. De Simone, F. Paglietti, Mechanochemical treatment to recycling asbestos-containing waste, Waste Manage. 23 (2003) 209–218. [10] S. Palaniandy, N.H. Jamil, Inﬂuence of milling conditions on the mechanochemical synthesis of CaTiO3 nanoparticles, J. Alloys Compd. 476 (2009) 894–902. [11] P. Billik, M. Èaplovièová, Synthesis of nanocrystalline SnO2 powder from SnCl4 by mechanochemical processing, Powder Technol. 191 (2009) 235–239. [12] V.V. Boldyrev, Mechanochemistry and Mechanical Activation of Solids, Imperial College Press, London, 2004. [13] K. Hall, M. Harrowﬁeld, J. Hart, G. McCormick, Mechanochemical reaction of DDT with calcium hydride, Environ. Sci. Technol. 30 (1996) 3401–3407. [14] S. Loiselle, M. Branca, G. Mulas, G. Cocco, Selective mechanochemical dehalogenation of chlorobenzenes over calcium hydride, Environ. Sci. Technol. 31 (1996) 261–265. [15] G. Cocco, M. Monagheddu, G. Mulas, S. Doppiu, S. Raccanelli, Reduction of polychlorinated dibenzodioxins and dibenzofurans in contaminated muds by mechanically induced combustion reactions, Environ. Sci. Technol. 33 (1999) 2485–2488. [16] M. Aresta, P. Caramuscio, L. De Stefano, T. Pastore, Solid state dehalogenation of PCBs in contaminated soil using NaBH4 , Waste Manage. 23 (2003) 315–319. [17] R. Ciofﬁ, L. De Stefano, R. Lamanna, F. Montagnaro, L. Santoro, S. Senatore, A. Zarrelli, TG, FT-IR and NMR characterization of n-C16 H34 contaminated alumina and silica after mechanochemical treatment, Chemosphere 70 (2008) 1068–1076. [18] S. Milosevic, M. Tomasevic-Canovic, R. Dimitrijevic, M. Petrov, M. Djuricic, B. Zivanovic, Amorphization of aluminosilicate minerals during micronization process, Ceram. Bull. 71 (1992) 771–775. [19] M. Lavorgna, L. De Stefano, R. Ciofﬁ, Reuse of asbestos-waste milled materials, vol. 1, in: M. Pelino (Ed.), Proceed. III Intern. Conf. Valorization and Recycling of Industrial Waste, L’Aquila, Italy, 2001, pp. 159–164. [20] R. Ciofﬁ, F. Colangelo, D. Caputo, B. Liguori, in: V.M. Malhotra (Ed.), Inﬂuence of High Volumes of Ultra-Fine Additions on Self-Compacting Concrete, Superplasticizers and Other Chemical Admixtures in Concrete, American Concrete Institute SP-239, Farmington hills, MI, USA, 2006, pp. 117–136. [21] L. Franke, K. Sisomphon, A new chemical method for analyzing free calcium hydroxide content in cementing material, Cem. Concr. Res. 34 (2004) 1161–1165. [22] UNI EN 196-1: Standard Methods of Testing Cement – Part 1: Determination of Strength, 2005. [23] F. Saito, Dehalogenation from organic compounds by mechanochemical method – in the case of polyvinyl chloride. Annual Report of Hosokawa Powder Technol, Foundation 6 (1998) 120–123. [24] F. Saito, Mechanochemical effects by grinding and its application to materials processing, Ceramics 34 (1999) 844–847. [25] F. Saito, Engineering application of mechanochemistry – materials synthesis and wastes disposal, Mater. Life 11 (1999) 152–156. [26] F. Saito, Mechanochemistry and resources wastes processing, Kinzoku 69 (1999) 1082–1088. [27] G.L. Valenti, L. Santoro, G. Volpicelli, Hydration of granulated blast furnace slag in the presence of phosphogypsum, Thermochim. Acta 78 (1984) 101–112. [28] L. Santoro, I. Aletta, G.L. Valenti, Hydration of mixture containing ﬂy ash, lime and phosphogypsum, Thermochim. Acta 96 (1986) 71–80. [29] V.S. Ramachandran, Application of DTA in Cement Chemistry, Chemical Publishing Co. Inc., New York, 1969.

F. Colangelo et al. / Journal of Hazardous Materials 195 (2011) 391–397 [30] M. Murat, Stabilité thermique des aluminates de calcium hydratés et phases apparentées. Caractérisation par les méthodes thermoanalytiques, vol. 5, in: Int. Semin. Aluminates Calcium, Turin, Italy, 1982, pp. 9–65. [31] G.L. Valenti, R. Ciofﬁ, L. Santoro, S. Ranchetti, Inﬂuence of chemical and physical properties of Italian ﬂy ashes on reactivity towards lime, phosphogypsum and water, Cem. Concr. Res. 18 (1988) 91–102.

397

[32] J. Beretka, R. Ciofﬁ, L. Santoro, G.L. Valenti, Cementitious mixtures containing industrial process wastes suitable for the manufacture of preformed building materials, J. Chem. Technol. Biotechnol. 59 (1994) 243–247. [33] R. Ciofﬁ, M. Marroccoli, G. Martone, L. Santoro, Utilization of zeolite-rich tuff for the manufacture of building materials based on calcium silicate and trisulphoaluminate hydrates, Thermochim. Acta 306 (1997) 93–98.

Journal of Hazardous Materials 195 (2011) 398–404

Contents lists available at SciVerse ScienceDirect

Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat

Removal of arsenic from water by Friedel’s salt (FS: 3CaO·Al2 O3 ·CaCl2 ·10H2 O) Danni Zhang a,c , Yongfeng Jia a,∗ , Jiayu Ma b,c , Zhibao Li b,∗ a

Key Laboratory of Pollution Ecology and Environmental Engineering, Institute of Applied Ecology, Chinese Academy of Sciences, Shenyang 110016, China Institute of Process Engineering, Chinese Academy of Sciences, Beijing 110016, China c Graduate School, Chinese Academy of Sciences, Beijing 100049, China b

a r t i c l e

i n f o

Article history: Received 4 May 2011 Received in revised form 30 July 2011 Accepted 19 August 2011 Available online 26 August 2011 Keywords: Arsenic Removal Water Friedel’s salt

a b s t r a c t Low levels of arsenic can be effectively removed from water by adsorption onto various materials and searching for low-cost, high-efﬁciency new adsorbents has been a hot topic in recent years. In the present study, the performance of Friedel’s salt (FS: 3CaO·Al2 O3 ·CaCl2 ·10H2 O), a layered double hydroxide (LDHs), as an adsorbent for arsenic removal from aqueous solution was investigated. Friedel’s salt was synthesized at lower temperature (50 ◦ C) compared to traditional autoclave methods by reaction of calcium chloride with sodium aluminate. Kinetic study revealed that adsorption of arsenate by Friedel’s salt was fast in the ﬁrst 12 h and equilibrium was achieved within 48 h. The adsorption kinetics are well described by second-order Lageren equation. The adsorption capacity of the synthesized sorbent for arsenate at pH 4 and 7 calculated from Langmuir adsorption isotherms was 11.85 and 7.80 mg/g, respectively. Phosphate and silicate markedly decreased the removal of arsenate, especially at higher pH, but sulfate was found to suppress arsenate adsorption at lower pH and the adverse effect was disappeared at pH ≥ 6. Common metal cations (Ca2+ , Mg2+ ) enhanced arsenate adsorption. The results suggest that Friedel’s salt is a potential cost-effective adsorbent for arsenate removal in water treatment. © 2011 Elsevier B.V. All rights reserved.

1. Introduction Arsenic is a primary concern of water contamination due to its high toxicity and carcinogenicity and is regarded as the ﬁrst priority issue among the toxic substances [1]. Millions of people in developing countries, e.g. Bangladesh, Vietnam and China are suffering serious health problems such as cancer, skin lesions, metabolic and cardiac disorders due to chronic exposure to arsenic contaminated drinking water [2]. The maximum contaminant level (MCL) for arsenic in public drinking water in Europe, China and the United States is 10 ␮g/L, which means that some arsenic removal processes have to be adopted by public water system in order to meet the lately released drinking water standard. Various technologies are currently available to remove arsenic from aqueous solution, such as ion exchange [3], coagulation (coprecipitation) [4], reverse osmosis [5], bioremediation [6], and adsorption [7,8]. Among these methods, adsorption technique using various adsorbents (elemental iron, iron (hydr)oxides, modiﬁed active carbons, silicotinphosphate molecular sieve, active alumina and aluminosilicate clay materials) has been widely investigated [7–13], all showing some advantages and disadvantages on performance, cost-effectiveness and easiness of implementation.

∗ Corresponding authors. E-mail addresses: [email protected] (Y. Jia), [email protected] (Z. Li). 0304-3894/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.jhazmat.2011.08.058

Hence, investigation of new adsorbents with high effectiveness and low cost in order to meet large numbers of demands from industrial and civil community is attracting increasing attentions. To this end, layer double hydroxide (LDHs), especially, Friedel’s salt (FS: 3CaO·Al2 O3 ·CaCl2 ·10H2 O) may be one of the potential candidates. Layer double hydroxides (LDHs) are a group of anionic clay minerals consisting of cationic brucite layers and exchangeable interlayer anions. These compounds can be nominally expressed with a chemical formula [M1−x 2+ Mx 3+ (OH)2 ]X+ (An− )x/n ·mH2 O, where M2+ and M3+ are divalent and trivalent metal cations, respectively, with an ionic radius similar to that of M2+ ; An− is an n-valent exchangeable anion (inorganic or organic) and x typically ranges from 0.17 to 0.33 [14]. Most of LDHs have relatively weak interlayer bonding and, as a consequence, exhibit excellent ability to capture inorganic anions by both surface adsorption and anion exchange [15]. Various types of LDHs with easily exchangeable anions, such as chloride, nitrate and carbonate-LDHs and their corresponding calcined product have been explored as adsorbents to remove arsenic contaminants in recent years [16–23]. Friedel’s salt was ﬁrst observed in the work of Friedel, who studied the reactivity of lime with aluminum chloride. This compound is formed in cements such as Portland cement which contains large amounts of tricalcium aluminate [24] and belongs to the layer double hydroxide (LDHs) family. Due to its ion exchange characteristic, FS has received considerable attention in recent years by various groups. For example, Dai et al. examined the adsorption behaviors

D. Zhang et al. / Journal of Hazardous Materials 195 (2011) 398–404

of FS for Cr (VI) from aqueous solution at different concentrations and various initial pHs [25]. Recently, Wu et al. found that Friedel phase can rapidly adsorb large amounts of selenate from aqueous solutions [26]. The FS was usually synthesized using autoclaves at elevated temperatures. Most recently, FS was prepared using calcium chloride generated from soda ash industry at low temperature and found to have desilication capacity in sodium aluminate solution [27]. To the best of our knowledge, there are no published studies on the adsorption behavior of arsenic on Friedel’s salt from aqueous solution. The objective of this study was to test the performance of Friedel’s salt as an adsorbent for arsenic removal from water. The effects of various parameters such as pH and common ions on arsenic removal were investigated. 2. Experimental All chemicals were of analytical grade and used without further puriﬁcation. Distilled water was used for all experiments. All glassware was cleaned by soaking in 5% HNO3 and rinsed three times with distilled water. Arsenate and arsenite stock solution was prepared from Na3 AsO4 ·12H2 O and NaAsO2 , respectively. The container of arsenite stock solution was purged with N2 and sealed tightly to prevent from oxidation of As(III). 2.1. Synthesis of Friedel’s salt Friedel’s salt was synthsized by precipitation at ambient pressure. The preparation experiments were conducted by pre-heating 0.5 M CaCl2 (300 mL) to a desired temperature (50 ◦ C) and 0.25 M NaAlO2 with equal volume was added using a peristaltic pump at a rate of 5 mL/min and the mixture was stirred at 300 rpm. After the mixture was reacted for 1 h, the white precipitate was collected by ﬁltration, washed three times with distilled water to remove residual substances and dried in a vacuum oven at 50 ◦ C for 10 h. 2.2. Characterization of Friedel’s salt The structure and morphology of the synthesized samples were examined by using powder X-ray diffraction (XRD) and scanning electron microscopy (SEM). The powder X-ray diffraction patterns were recorded on a diffractometer (X’Pert PRO MPD, PANalytical, The Netherlands) using Cu K␣ radiation, operating at 40 kV/30 mA. A scanning rate of 0.02◦ /s was applied to record the patterns in the 2 angle range from 10◦ to 90◦ . The morphology of the synthesized sample was examined by scanning electron microscopy (SEM, JEOL-JSM-6700F). The FTIR spectrum was obtained on a Shimadzu IR Prestige-21 spectrometer with the resolution of 4 cm−1 . The KBr/sample discs were prepared by mixing 1% of samples in KBr. 2.3. Arsenic adsorption kinetics 0.1 g Friedel’s salt adsorbent was added to the conical ﬂasks containing 500 mL of 2 mg/L arsenic solution. The initial pH of the solution was adjusted to 7 ± 0.2 with 1 mol/L HNO3 or 1 mol/L NaOH and maintained constant throughout the adsorption process. The mixture was shaken for 72 h at 25 ◦ C. At regular intervals an aliquot of supernatant was sampled and ﬁltered through 0.22␮m membrane ﬁlter (Millipore) for the analysis of aqueous arsenic concentration. 2.4. Arsenic adsorption isotherms In the tests of adsorption isotherms, 0.2 g/L of Friedel’s salt was used. To obtain adsorption isotherms, a series of arsenate solutions

399

with initial arsenic concentration from 0.4 to 4.1 mg/L were prepared. The pH of As(V) solution was preadjusted to 4 and 7 prior to adsorption tests. 100 mL of As(V) solution and 0.02 g of Friedel’s salt were added into a 250-mL conical ﬂask and the mixture was equilibrated at 25 ◦ C for 48 h on a reciprocating platform shaker. The slurries were maintained at pH 4 ± 0.2 and pH 7 ± 0.2 by addition of 0.1 N HCl or NaOH. The solution was ﬁltered through 0.22-␮m membrane ﬁlter for the concentration analysis of aqueous arsenic. 2.5. Arsenic adsorption envelopes The adsorption envelopes were obtained by equilibrating 0.02 g of Friedel’s salt with 100 mL of 2 mg/L arsenic solution for 48 h at pH 3–10, 25 ◦ C. For As(III), the conical ﬂasks were covered with aluminum foil to prevent from light irradiation and purged with N2 before capping and during pH adjustment. Each experiment was run in duplicate and the mean value was reported. 2.6. Effect of coexisting ions 0.02 g of Friedel’s salt was added to 100 mL of 2 mg/L arsenic solutions containing additional cations (calcium and magnesium) or anions (phosphate, silicate and sulfate). The slurry was adjusted to various pHs ranging from 3 to 10 and the molar ratio of arsenic to the added ions was 1:10. The mixtures were mildly agitated for 48 h at 25 ◦ C and the concentration of arsenic remaining in solution was analyzed. 2.7. Determination of the concentration of As, Al, Ca The concentration of arsenic in solution was determined on an atomic ﬂuorescence spectrophotometer (AFS-2202E, Haiguang Corp., Beijing) coupled with a hydride generator. The detection limit of the instrument for arsenic was 0.1 ␮g/L. The concentration of aluminum and calcium in solution was determined on inductively coupled plasma-atomic emission spectroscopy (ICP-AES) (Perkin Elmer Optima 3000) with the detection limit of 0.2 and 0.02 mg/L, respectively. 3. Results and discussion 3.1. Characterization of the adsorbent The SEM images of the synthesized Friedel’s salt are shown in Fig. 1. Typical ﬂat hexagonal (or pseudohexagonal) crystal morphology was observed. Some aggregation and small non-uniform growths were also present in this compound. The XRD patterns of the synthesized Friedel’s salt were sharp and consistent with the literature values (JCPDS 78-2051) (see Fig. 2). The infrared spectrum of the synthesized Friedel’s salt was also presented in Fig. 2. The features at 528 cm−1 and 789 cm−1 are ascribed to the bending and stretching vibration of Al–OH [26]. The peak at 1624 cm−1 was due to the H–O–H bending vibration of the interlayer water molecule (␯2 H2 O). The stretching vibration of lattice water and structural OH groups (␯OH ) in the Friedel’s salt were also reﬂected by the strong overlapping bands at 3473 and 3630 cm−1 , respectively. The band at 1420 cm−1 was attributed to CO3 2− due to incorporation of CO2 during synthesis of the compound. 3.2. Adsorption kinetics As shown in Fig. 3, the kinetics of arsenic adsorption by Friedel’s salt included two steps: a fast initial sorption followed by a much slower sorption process. At pH 7, approximately 66.2% of arsenic was removed from As(V) solution in the ﬁrst 12 h and adsorption equilibrium was reached within 48 h. Thus, the equilibration time

400

D. Zhang et al. / Journal of Hazardous Materials 195 (2011) 398–404

Fig. 1. SEM microimages of the synthesized Friedel’s salt.

(611)

(334)

(406) (208) (226) (118) (413) (219) (019)

(116) (025) (404) (223)

(114)

(024)

(112)

Intensity

ally, a fast initial oxyanion adsorption followed by a slower process to reach equilibrium was the characteristic of oxyanion adsorption kinetics using various forms of LDHs as adsorbents [15]. Similar phenomenon was observed here for the adsorption of arsenate on Friedel’s salt. The kinetics of adsorption reaction was well described by second-order Lageren equation: dQt = k2 (Qe − Qt )2 dt

(1)

where Qe (mg/g) is the amount of arsenate adsorbed on Friedel’s salt at equilibrium, Qt (mg/g) is the amount of arsenate adsorbed at time t, k2 (g/mg h) is the adsorption rate constant. Eq. (1) can be simpliﬁed by using initial condition and boundary condition (t = 0, Qt = 0; t = t, Qt = Qe ) as follows:

The original Friedel's salt (316)

(006)

A

(222)

of 48 h was applied in the determination of adsorption isotherms and envelopes. In a previous study on the application of uncalcined MgAl–CO3 –LDHs for arsenic removal from water, it took 48–72 h to reach equilibrium [17]. In comparison, Lazaridis et al. reported that it needed only 8 h to reach equilibrium when the same mineral was used as adsorbent for the removal of arsenic and the fast kinetics was ascribed by the authors to the positive inﬂuence of potassium nitrate used for regulating the solution ionic strength [16]. Gener-

t 1 t = + Qt Qe k2 Qe2

The solid after arsenic adsorption

(2)

According to Eq. (2), the rate constant can be obtained as k2 = 0.13 g/mg h. 3.3. Adsorption envelopes

0

20

40

60

80

The adsorption envelopes of arsenite and arsenate on Friedel’s salt at pH 3–10 are illustrated in Fig. 4. Apparently, arsenate

100

Two theta (Degrees) 2.0

B

10

t/q t ((h ·g)/mg)

Transmittance %

As concentration (mg/L)

8

1.5

6 4 2

1.0

0 -10

0

10

20

30

40

50

60

70

80

Time (h)

0.5

0.0 4000

3500

3000

2500

2000

1500

1000

500

0

10

20

30

40

50

60

70

80

Time (h)

Wavenumber (cm -1) Fig. 2. (A) XRD patterns of the Friedel’s salt prepared at 50 ◦ C and the solid after adsorption of arsenate at pH 7 and (B) FTIR spectrum of the prepared Friedel’s salt.

Fig. 3. The kinetics of arsenic adsorption by Friedel’s salt. Reaction conditions: 2.0 mg/L As(V), 0.2 g/L Friedel’s salt, pH 7, 25 ◦ C. Inset shows modeling of the kinetics by second-order Lageren equation.

D. Zhang et al. / Journal of Hazardous Materials 195 (2011) 398–404

100 As (V) As (III)

90

Arsenic removal (%)

80 70 60 50 40 30 20 10 0 2

3

4

5

6

7

8

9

10

11

pH Fig. 4. Adsorption envelopes of As(V) and As(III) on Friedel’s salt at initial arsenic concentration of 2 mg/L, the solid/liquid ratio of 0.2 g/L and equilibration time of 48 h.

removal efﬁciency is strongly dependent on media pH, while arsenite removal appeared not to be signiﬁcantly affected by pH. The percentage of arsenate removal by Friedel’s salt went down all the way from 92% to <30% when pH increasing from 3 to 10, whereas the removal of arsenite appears unchanged at ∼12% at pH 3–8 and decreased at pH ≥ 9. It was also reported in previous work that adsorption of arsenate on ferrihydrite and nano-sized zerovalent iron decreased from mildly acidic to alkaline media, while adsorption of arsenite increased in acidic solution but decreased in alkaline solution [28,29]. Compared to iron-based adsorbents, the removal efﬁciency of arsenite by Friedel’s salt was much lower than that of arsenate at all pH. Similar results were also found by Yang et al. who reported that uncalcined MgAl–CO3 –LDH adsorbed almost no arsenite [17]. The possible reason for low adsorption capacity of Friedel’s salt for arsenite is that its major chemical compositions are aluminum and calcium. It is commonly observed in water treatment literatures that aluminum oxides and oxyhydroxides or aluminosilicate minerals are ineffective for the removal of As(III) from potable water but can remove As(V) efﬁciently [30,31]. The pH dependent behavior of arsenic adsorption by Friedel’s salt is the coeffect of several competing factors controlling the adsorption reaction. It was proposed that other types of LDHs can uptake contaminants from aqueous medium by three different mechanisms: (1) surface adsorption, (2) interlayer anion exchange, and (3) intercalation by reconstruction of the structure of the calcined LDHs [15]. Grover et al. suggested that the underlying mechanism for arsenate removal by hydrocalcumitetype LDH appears to be anion exchange as well as partial dissolution–precipitation [22]. Dai et al. reported that the removal and ﬁxation of chromate from aqueous solution by Friedel’s salt involves the adsorption/exchange process [25]. Since arsenate and chromate were both oxyanions with varying aqueous species at different pH, they probably show similar adsorption mechanism by Friedel’s salt. Oxyanions migration to the surface of the adsorbent is the prerequisite of the adsorption reaction and largely controlled by electrostatic attraction or repulsion of the aqueous arsenate or arsenite species with the surface of the adsorbent [32]. Hence, pH of zero point charge (pHzpc ) of the adsorbent and the speciation of aqueous arsenate and arsenite are governing factors. The degree of protonation of arsenate and arsenite anions in aqueous solution is a function of pH. The dissociation constants of aqueous arsenate are pKa1 = 2.3, pKa2 = 6.8, pKa3 = 11.6, resulting in arsenate species varying from H2 AsO4 − , HAsO4 2− , to AsO4 3− when pH increases from acidic region to alkaline region [31]. It is well

401

known that solid surface is positively charged at pH below pHzpc and negatively charged at pH above pHzpc , resulting in increased electrostatic attraction or repulsion with anionic arsenic species, hence leading to more or less readily adsorption. The pHzpc for the uncalcined LDH was reported to be in the range 6.8–8.9 [33], below which the surface of the adsorbent is positively charged and normally beneﬁcial for the adsorption of the negatively charged anionic species. The surface of the adsorbent becomes less positively charged when pH increased hence shows less attraction towards anionic arsenate species. Therefore, at higher pH range, the adsorption of arsenate decreased signiﬁcantly. The adverse effect of pH on arsenate adsorption at higher pH range may be further compounded by the increasing competitive effect of OH− adsorption on Friedel’s salt. By referring to the mechanism proposed by Dai et al. and Wu et al. for chromate and selenate adsorption on Friedel’s salt [25,26], arsenate adsorption may be schematically depicted by following reactions: Ca4 Al2 (OH)12 Cl2 (H2 O)4 ·xH2 O + H2 AsO4 − → Ca4 Al2 (OH)12 (H2 AsO4 − )2 (H2 O)4 ·xH2 O + 2Cl−

(3)

Ca4 Al2 (OH)12 Cl2 (H2 O)4 ·xH2 O + HAsO4 2− → Ca4 Al2 (OH)12 HAsO4 (H2 O)4 ·xH2 O + 2Cl−

(4)

When the pH was increased to 10, the following two reactions may also be involved. Firstly, the freshly formed OH− could probably be intercalated into the interlayer spacing by replacing the original interlayer ion (Cl− in Friedel’s salt) prior to arsenate adsorption. Ca4 Al2 (OH)12 (OH)2 (H2 O)4 ·xH2 O + HAsO4 2− → Ca4 Al2 (OH)12 HAsO4 (H2 O)4 ·xH2 O + 2OH−

(5)

Secondly, the solution can readily adsorb CO2 from air and transfer to CO3 2− and trigger the following exchange of HAsO4 2− with the original carbonate hydrocalmite to happen: Ca4 Al2 (OH)12 CO3 (H2 O)4 ·xH2 O + HAsO4 2− → Ca4 Al2 (OH)12 HAsO4 (H2 O)4 ·xH2 O + CO3 2−

(6)

The mechanism of arsenate adsorption on Friedel’s salt may be different at lower pH and higher pH. After contacting with lowerpH arsenate solution, the adsorbent has been partly transformed to aluminum oxides (see Fig. 2). In this case, the bidentate binuclear interaction mode of arsenate anions with aluminum oxide may also be involved. 3.4. Adsorption isotherms Fig. 5 shows the adsorption isotherms of arsenate by Friedel’s salt at pH 4 and 7. Langmuir and Freundlich models are usually used to describe adsorption isotherms of compounds from liquid onto solid. Langmuir model assumes monolayer adsorption onto homogeneous surface with a ﬁnite number of identical sites, while the Freundlich model is empirical in nature. The results of arsenate adsorption were found to ﬁt well with both Freundlich and Langmuir isotherm models (r2 > 0.95) and the adsorption constants evaluated from the isotherms are listed in Table 1. The adsorption capacity of Friedel’s salt for arsenate calculated from the isotherms obtained at arsenic equilibrium concentration of <2 mg/L (Fig. 6) was 11.85 mg/g (pH 4) and 7.80 mg/g (pH 7), respectively. This is higher than the adsorption capacity of some reported adsorbents: activated carbon (3.1 mg/g) [9], nano-iron (hydr)oxide impregnated granulated activated carbon (0.263 mg/g) [34] and iron-containing ordered mesoporous carbon (7.0 mg/g)

402

D. Zhang et al. / Journal of Hazardous Materials 195 (2011) 398–404 Table 2 The concentration of aluminum and calcium released at various equilibrium concentration of arsenate after adsorption by Friedel’s salt at pH 4 and 7, respectively.

15

Arsenic adsorbed (mg/g)

pH 4 pH 7

pH 4

10

5

pH 7

Aseq (mg/L)

Ca (mg/L)

Al (mg/L)

Aseq (mg/L)

Ca (mg/L)

Al (mg/L)

0.026 0.043 0.162 0.172 0.439 0.740 0.904 1.046 1.721

5.19 5.14 5.55 5.34 5.47 5.45 5.67 5.88 5.62

b.d. b.d. b.d. b.d. b.d. b.d. b.d. b.d. b.d.

0.013 0.018 0.047 0.067 0.139 0.297 0.655 1.348 1.412

5.69 5.52 5.84 5.50 5.22 5.31 5.57 5.24 5.25

b.d. b.d. b.d. b.d. b.d. b.d. b.d. b.d. b.d.

b.d.: below detection limit.

0

0.0

0.5

1.0

1.5

2.0

Equilibrium concentration of arsenic (mg/L) Fig. 5. Adsorption isotherms of As(V) by Friedel’s salt at pH 4 or 7, the solid/liquid ratio of 0.2 g/L, 25 ◦ C and equilibration time of 48 h. Table 1 Langmuir and Freundlich adsorption isotherm parameters of arsenate on Friedel’s salt at pH 4 and 7, respectively. Langmuir model: qeq =

pH

4 7

bQmax Ceq 1+bCeq

1/n

Freundlich model: qeq = KCeq

b

Qmax (mg/g)

r2

K

n

r2

20.59 13.78

11.85 7.800

0.999 0.998

11.31 7.071

6.460 4.045

0.953 0.964

[35]. While for other common types of adsorbents, activated alumina has a comparable arsenate adsorption capacity of 15.9 mg/g at pH 5.2 while bead cellulose loaded with iron oxyhydroxide has a higher capacity as 33.2 mg/g at pH 7 [7,12]. It has been reported that the capacities of other forms of LDHs for arsenate adsorption range from 4.5 mg/g to 615 mg/g [15], however, it should be noticed that the relatively higher arsenate adsorption capacities such as 615 mg/g and 87.5 mg/g were obtained using calcined forms of LDHs as adsorbents. It has been observed in previous studies that arsenate adsorption by calcined LDHs was considerably higher than the adsorption by uncalcined LDHs [16]. As for the uncalcined form of LDHs which was the case in our investigation, Liu et al. reported that arsenate adsorption 100

without additives 3with PO4 2-

with SO4

80

Arsenic removal (%)

3-

with SiO4 60

capacity was 24.2 mg/g at pH 5 for LiAl–Cl–LDHs at initial arsenate concentration of 150 mg/L and solid/liquid ratio of 2.5 g/L [18]. Lazaridis et al. found arsenate adsorption capacity of 15.8 mg/g and 32.6 mg/g by MgAl–CO3 –LDHs at neutral pH value, ion strength of 0.001 and 0.1 respectively [16]. These results as well as our data indicated that different chemical composition or types of anions present in the interlayers of LDHs as well as using different experimental procedures (different initial arsenate concentration, adsorption pH, ionic strength as well as solid liquid ratio) may have an effect on arsenate adsorption capacity. The stability of an adsorbent is one of the most important considerations in its practical applications. If metals in the brucite-like sheets are released in the solution, it would be important that the amount of metals keeps below the levels that are harmful, or otherwise to choose cations with low toxicity [15]. In fact, few works report adsorbent stability when using other type of LDHs or iron-based adsorbents for removal of contaminant from water. Ferreira et al. found that the amount of magnesium ions released from the Mg–Fe/Al–LDHs increased with decreasing pH during removal of boron from water [36]. Yang et al. observed signiﬁcant dissolution of aluminum and magnesium from the LDHs during removal arsenic and selenate contaminants [17]. In our investigation, it was observed that during the adsorption of arsenate, little aluminum was released from the adsorbent especially at neutral pH. The concentration of aluminum in water after adsorption was below the detection limit (i.e. <0.2 mg/L), which meets the drinking water standard for aluminum in some countries (e.g. 0.2 mg/L in China). Calcium release was remained at a relative stable level (5.1–5.8 mg/L) irrespective of the amount of arsenate uptake (Table 2). The structure of the adsorbent after adsorption of arsenate was characterized by XRD and the result was presented in Fig. 2. As can be observed, the structure of Friedel’s salt was altered after contacting with arsenate solution. The XRD patterns were dominated by aluminum oxides features, while the features of some Ca-related compounds which constitute the relatively unactive adsorption site of the adsorbent disappeared [26]. Similar results have also been reported by Wu et al. who found that at lower pH, the framework ‘Ca(OH)2 ’ was partially dissolved [26].

40

3.5. Effect of coexisting ions 20

0 2

3

4

5

6

7

8

9

10

pH Fig. 6. Effect of coexisting anions on arsenate removal by Friedel’s salt at pH 3–10. Initial arsenate concentration was 2 mg/L and the solid/liquid ratio of 0.2 g/L; the molar ratio of As to coexisting anions was 1:10.

Drinking water or industrial efﬂuent usually contains various anions and cations which may negatively or positively inﬂuence the adsorption of arsenic. The effects of some common anions (PO4 3− , SiO4 3− , SO4 3− ) and cations (Ca2+ , Mg2+ ) on arsenate adsorption by Friedel’s salt were investigated (see Figs. 6 and 7). The removal of arsenate was adversely affected in the presence of phosphate, silicate and sulfate. Among the oxyanions considered in this work, phosphate showed most adverse effect on arsenate removal across the whole pH range. The suppressing effect of phosphate

D. Zhang et al. / Journal of Hazardous Materials 195 (2011) 398–404

removal efﬁciency for arsenate than arsenite across wide pH range from acidic to alkaline media. The adsorption capacity for arsenate at pH 4 and 7 was 11.85 and 7.80 mg/g, respectively. Arsenic removal was markedly decreased in the presence of phosphate or silicate, however, sulfate was found to suppress arsenate adsorption at pH < 6 and its effect was negligible at pH > 6. The presence of common metal cations (Ca2+ , Mg2+ ) in water enhanced arsenate removal.

100 90 80

Arsenic removal (%)

403

70 60 50 40 30 20

with Ca

2+

with Mg

2+

without M

Acknowledgement The authors thank National Natural Science Foundation of China for the ﬁnancial supports (40925011, 40803032 and 41073086).

2+

10 0 2

3

4

5

6

7

8

9

10

References

pH Fig. 7. Effect of coexisting cations on arsenate removal by Friedel’s salt at pH 3–10. Initial arsenate concentration was 2 mg/L and the solid/liquid ratio of 0.2 g/L; the molar ratio of As to coexisting cations was 1:10.

and silicate on arsenate removal became more pronounced with increasing pH. Dadwhal et al. examined the adsorption isotherms of arsenic on MgAl–CO3 –LDH in the presence of various competing ions and found that sulfate and phosphate severely affected arsenic adsorption [37]. Violante et al. also demonstrated that arsenate adsorption on MgAl–CO3 –LDH was decreased in the presence of phosphate [20]. Arsenate, phosphate and silicate are speciﬁc adsorbing anions while sulfate ions can be sorbed both speciﬁcally and non-speciﬁcally. They would compete for similar binding sites, hence decreasing arsenate sorption. The detrimental effect of these anions on arsenate removal may reduce treatment efﬁciency of arsenate-contaminated water by Friedel’s salt. The effect of sulfate on arsenate removal by Friedel’s salt is surprising. At pH < 5, arsenate removal was more signiﬁcantly reduced by sulfate at lower pH, while at pH ≥ 6 the negative effect of sulfate was negligible. This indicates that for the treatment of arsenic-contaminated groundwater using Friedel’s salt as adsorbent at environmental relevant pH, arsenate removal may not be reduced by sulfate. The removal of arsenate by most ironcontaining adsorbents was not appreciably affected by sulfate [38]. High concentration of sulfate was found to even improve arsenate removal by zero-valent iron [39]. The reason for the signiﬁcantly adverse effect of sulfate on arsenate adsorption on Friedel’s salt at acidic pH is unknown. Common divalent metal cations such as Mg2+ and Ca2+ were found to enhance the adsorption of arsenate on the synthesized adsorbent. The positive effect increased slightly with increasing pH. This is beneﬁcial for the treatment arsenate contaminated groundwater that usually contains dissolved magnesium and calcium ions. The enhancing effect of metal cations on arsenate adsorption was reported for iron (hydr)oxides [29,40]. This is probably due to that the presence of metal cations in the solution shifted the surface of the adsorbent to more positively charged nature, which in turn enabled the adsorbent to show higher afﬁnity for arsenate anions. 4. Conclusions Friedel’s salt with good crystallinity was prepared from sodium aluminate solution using calcium chloride at 50 ◦ C, ambient pressure and its performance for arsenic removal from water was investigated by batch adsorption experiments. The results showed that the synthesized adsorbent was effective for the removal of arsenate with relatively fast kinetics. The adsorbent showed higher

[1] S.L. Lo, H.T. Jeng, C.H. Lai, Characteristics and adsorption properties of iron coated sand, Water Sci. Technol. 35 (1997) 63–70. [2] A. Basu, J. Mahata, S. Gupta, A.K. Giri, Genetic toxicology of a paradoxical human carcinogen, arsenic: a review, Mutation Res. 488 (2001) 171–194. [3] J. Kim, M.M. Benjamin, Modeling a novel ion exchange process for arsenic and nitrate removal, Water Res. 38 (2004) 2053–2062. [4] A. Zouboulis, I. Katsoyiannis, Removal of arsenates from contaminated water by coagulation-direct ﬁltration, Sep. Sci. Technol. 37 (2002) 2859–2873. [5] R.Y. Ning, Arsenic removal by reverse osmosis, Desalination 143 (2002) 237–241. [6] T.M. Gihring, G.K. Druschel, R.B. McCleskey, R.J. Hamers, J.F. Banﬁeld, Rapid arsenite oxidation by thermus aquaticus and thermus thermophilus: ﬁeld and laboratory investigations, Environ. Sci. Technol. 35 (2001) 3857–3862. [7] X. Guo, F. Chen, Removal of arsenic by bead cellulose loaded with iron oxyhydroxide from groundwater, Environ. Sci. Technol. 39 (2005) 6808–6818. [8] X.H. Guan, J. Wang, C.C. Chusuei, Removal of arsenic from water using granular ferric hydroxide: macroscopic and microscopic studies, J. Hazard. Mater. 156 (2008) 178–185. [9] C.L. Chuang, M. Fan, M. Xu, R.C. Brown, S. Sung, B. Saha, C.P. Huang, Adsorption of As(V) by activated carbon prepared from oat hulls, Chemosphere 61 (2005) 478–483. [10] Z. Gu, J. Fang, B. Deng, Preparation and evaluation of GAC-based iron containing adsorbents for arsenic removal, Environ. Sci. Technol. 39 (2005) 3833–3843. [11] D. Chandra, N.K. Mal, A. Bhaumik, Novel mesoporous silicotinphosphate molecular sieve with high anion exchange capacity, J. Mol. Catal. A: Chem. 247 (2006) 216–221. [12] T.F. Lin, J.K. Wu, Adsorption of arsenite and arsenate within activated alumina grains: equilibrium and kinetics, Water Res. 35 (2001) 2049–2057. [13] B.A. Manning, S. Goldberg, Adsorption and stability of arsenic(III) at the clay mineral–water interface, Environ. Sci. Technol. 31 (1997) 2005–2011. [14] T. Kwon, G.A. Tsigdinos, T.J. pinnavaia, Pillaring of layered double hydroxides (LDH’s) by polyoxometalate anions, J. Am. Chem. Soc. 110 (1988) 3653–3654. [15] K.H. Goh, T.T. Lim, Z. Dong, Application of layered double hydroxides for removal of oxyanions: a review, Water Res. 42 (2008) 1343–1368. [16] N.K. Lazaridis, A. Hourzemanoglou, K.A. Matis, Flotation of metal-loaded clay anion exchangers. Part II: the case of arsenates, Chemosphere 47 (2002) 319–324. [17] L. Yang, Z. Shahrivari, P.K.T. Liu, M. Sahimi, T.T. Tsotsis, Removal of trace levels of arsenic and selenium from aqueous solutions by calcined and uncalcined layered double hydroxides (LDH), Ind. Eng. Chem. Res. 44 (2005) 6804–6815. [18] Y.T. Liu, M.K. Wang, T.Y. Chen, P.N. Chiang, P.M. Huang, J.F. Lee, Arsenate sorption on lithium/aluminum layered double hydroxide intercalated by chloride and on gibbsite: sorption isotherms, envelopes, and spectroscopic studies, Environ, Sci. Technol. 40 (2006) 7784–7789. [19] G.P. Gillman, A simple technology for arsenic removal from drinking water using hydrotalcite, Sci. Total Environ. 366 (2006) 926–931. [20] A. Violante, M. Pucci, V. Cozzolino, J. Zhu, M. Pigna, Sorption/desorption of arsenate on/from Mg–Al layered double hydroxides: inﬂuence of phosphate, J. Colloid Interface Sci. 333 (2009) 63–70. [21] K. Grover, S. Komarneni, H. Katsuki, Uptake of arsenite by synthetic layered double hydroxides, Water Res. 43 (2009) 3884–3890. [22] K. Grover, S. Komarneni, H. Katsuki, Synthetic hydrotalcite-type and hydrocalumite-type layered double hydroxides for arsenate uptake, Appl. Clay Sci. 48 (2010) 631–637. [23] A. Bhaumik, S. Samanta, N.K. Mal, Efﬁcient removal of arsenic from polluted ground water by using a layered double hydroxide exchanger, Indian J. Chem. A. 44 (2005) 1406–1409. [24] A.K. Suryavanshi, J.D. Scantlebury, S.B. Lyon, Mechanism of Friedel’s salt formation in cements rich in tricalcium aluminate, Cement Concrete Res. 26 (1996) 717–727. [25] Y. Dai, G. Qian, Y. Gao, Y. Chi, Y. Xu, J. Zhou, Q. Liu, Z.P. Xu, S. Qiao, Effective removal and ﬁxation of Cr(VI) from aqueous solution with Friedel’s salt, J. Hazard. Mater. 170 (2009) 1086–1092.

404

D. Zhang et al. / Journal of Hazardous Materials 195 (2011) 398–404

[26] Y. Wu, Y. Chi, H. Bai, G. Qian, Y. Gao, J. Zhou, Y. Xu, Q. Liu, Z.P. Xu, S. Qiao, Effective removal of selenate from aqueous solutions by the Friedel phase, J. Hazard. Mater. 176 (2010) 193–198. [27] J. Ma, Z. Li, Y. Zhang, G.P. Demopoulos, Desilication of sodium aluminate solution by Friedel’s salt (FS: 3CaO·Al2 O3 ·CaCl2 ·10H2 O), Hydrometallurgy 99 (2009) 225–230. [28] K.P. Raven, A. Jain, R.H. Loeppert, Arsenit and arsenate adsorption on ferrihydrite: kinetics, equilibrium, and adsorption envelopes, Environ. Sci. Technol. 32 (1998) 344–349. [29] H.J. Zhu, Y.F. Jia, X. Wu, H. Wang, Removal of arsenic from water by supported nano zero-valent iron on activated carbon, J. Hazard. Mater. 172 (2009) 1591–1596. [30] Y.J. Arai, E.J. Elzinga, D.L. Sparks, X-ray absorption spectroscopic investigation of arsenite and arsenate adsorption at the aluminum oxide–water interface, J. Colloid Interface Sci. 235 (2001) 80–88. [31] S. Goldberg, C.T. Johnston, Mechanisms of arsenic adsorption on amorphous oxides evaluated using macroscopic measurements, vibrational spectroscopy, and surface complexation modeling, J. Colloid Interface Sci. 234 (2001) 204–216. [32] A. Jain, K.P. Raven, R.H. Loeppert, Arsenite and arsenate adsorption on ferrihydrite: surface charge reduction and net OH− release stoichiometry, Environ. Sci. Technol. 33 (1999) 1179–1184. [33] D.P. Das, J. Das, K. Parida, Physicochemical characterization and adsorption behavior of calcined Zn/Al hydrotalcite-like compounds (HTLc) towards

[34]

[35]

[36]

[37]

[38]

[39]

[40]

removal of ﬂuoride from aqueous solution, J. Colloid Interface Sci. 261 (2003) 213–220. K.D. Hristovski, P.K. Westerhoff, T. möller, P. Sylvester, Effect of synthesis conditions on nano-iron (hydr)oxide impregnated granulated activated carbon, Chem. Eng. J. 146 (2009) 237–243. Z. Gu, B. Deng, J. Yang, Synthesis and evaluation of iron-containing ordered mesoporous carbon (FeOMC) for arsenic adsorption, Microporous Mesoporous Mater. 102 (2007) 265–273. O.P. Ferreira, S.G. Moraes, N. Durán, L. Cornejo, O.L. Alves, Evaluation of boron removal from water by hydrotalcite-like compounds, Chemosphere 62 (2006) 80–88. M. Dadwhal, M. Sahimi, T.T. Tsotsis, Adsorption isotherms of arsenic on conditioned layered double hydroxides in the presence of various competing ions, Ind. Eng. Chem. Res. 50 (2011) 2220–2226. H.J. Shipley, S.Y. Yean, A.T. Kan, M.B. Tomson, A sorption kinetics model for arsenic adsorption to magnetite nanoparticles, Environ. Sci. Pollut. Res. 17 (2010) 1053–1062. H. Sun, L. Wang, R. Zhang, J. Sui, G. Xu, Treatment of groundwater polluted by arsenic compounds by zero valent iron, J. Hazard. Mater. 129 (2006) 297–303. Y. Jia, G.P. Demopoulos, Adsorption of arsenate onto ferrihydrite from aqueous solution: inﬂuence of media (sulfate vs. nitrate), added gypsum, and pH alteration, Environ. Sci. Technol. 39 (2005) 9523– 9527.

Journal of Hazardous Materials 195 (2011) 405–413

Contents lists available at SciVerse ScienceDirect

Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat

Treatment of airborne asbestos and asbestos-like microﬁber particles using atmospheric microwave air plasma A. Averroes a,∗ , H. Sekiguchi a , K. Sakamoto b a b

Department of Chemical Engineering, Tokyo Institute of Technology, 2-12-1 O-okayama, Meguro-ku, Tokyo 152-8552, Japan Street Design Corporation, 6-9-30 Shimo odanaka, Kawasaki-shi, Kanagawa 211-0041, Japan

a r t i c l e

i n f o

Article history: Received 24 May 2011 Received in revised form 20 August 2011 Accepted 22 August 2011 Available online 27 August 2011 Keywords: Fiber particle Airborne asbestos Microwave air plasma Atmospheric non-equilibrium plasma Particle shape index

a b s t r a c t Atmospheric microwave air plasma was used to treat asbestos-like microﬁber particles that had two types of ceramic ﬁber and one type of stainless ﬁber. The treated particles were characterized via scanning electron microscopy (SEM) and X-ray diffraction (XRD). The experiment results showed that one type of ceramic ﬁber (Alumina:Silica = 1:1) and the stainless ﬁber were spheroidized, but the other type of ceramic ﬁber (Alumina:Silica = 7:3) was not. The conversion of the ﬁbers was investigated by calculating the equivalent diameter, the aspect ratio, and the ﬁber content ratio. The ﬁber content ratio in various conditions showed values near zero. The relationship between the normalized ﬁber vanishing rate and the energy needed to melt the particles completely per unit surface area of projected particles, which is deﬁned as , was examined and seen to indicate that the normalized ﬁber vanishing rate decreased rapidly with the increase in . Finally, some preliminary experiments for pure asbestos were conducted, and the analysis via XRD and phase-contrast microscopy (PCM) showed the availability of the plasma treatment. © 2011 Elsevier B.V. All rights reserved.

1. Introduction Asbestos refers to natural ﬁbrous silicate minerals with several types based on their composition. They fall into two groups: serpentines and amphiboles [1]. Serpentines are structured with sheets of silicates and essentially consist of chrysotile. Amphiboles have a structure with double-chain silicates such as amosite, crocidolite, and anthophyllite [2]. Asbestos minerals show several good performance aspects such as incombustibility, low thermal conductivity, high electrical resistance, and resistance to alkali, acid, and microorganisms. Because of such properties, asbestos has been widely used in construction. Asbestos, however, has serious problems such as its needle shape, lightweight, and easy inhalation. When accumulated in the lungs, it is considered to induce two kinds of cancer – mesothelioma and lung cancer – and two nonmalignant conditions: asbestosis and diffuse pleural thickening [3]. Because of these toxicities, the import and fabrication, and use of asbestos have been banned, except for limited uses under the Labor Safety Law Enforcement Ordinance No. 9 since October 1, 2004. The total amount of asbestos imported to Japan until 2004 was about 10 billion tons, 90% of which was used in construction [4]. Therefore, much asbestos can still be expected

∗ Corresponding author. Tel.: +81 3 5734 2110. E-mail address: [email protected] (A. Averroes). 0304-3894/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.jhazmat.2011.08.062

from the renovation of old buildings that contain asbestos. Some processes aimed at achieving full crystallochemical transformation of asbestos have been studied such as vitriﬁcation (1200–1600 ◦ C), ceramization (700–1000 ◦ C), and chemical attack (∼50 ◦ C) [1,5,6]. These processes have led to the stabilization of asbestos. Airborne asbestos is very harmful, and a person leaving a disposal station for asbestos-containing materials needs efﬁcient treatment. Therefore, this study proposes the treatment of airborne asbestos from the exhaust gas of disposal stations with microwave air plasma to prevent its spread. Microwave plasma is an electrical discharge with microwave as its source energy. It has some advantageous features such as electrode-less discharge, high gas temperature (about 3000 ◦ C), and compact size [7]. Its electrode-less discharge enables it to be unaffected by oxygen. The apparatus has almost the same size as a kitchen microwave oven, so it can be easily attached to the disposal station. Moreover, it is possible to convert air into stable plasma with a small amount of electricity. Applying this high gas temperature will melt asbestos and then spheroidize it and change its composition. Thus, the hazard of asbestos is deemed to be decreased. In this study, atmospheric microwave air plasma treatment of asbestos-like microﬁber materials was investigated under several experiment conditions. These microﬁber materials were chosen because they have similar needle shaped and close melting points to asbestos. The plasma-treated ﬁber materials were observed with scanning electron microscopy (SEM). The diameter of the

406

A. Averroes et al. / Journal of Hazardous Materials 195 (2011) 405–413 Table 1 Experiment conditions.

Nomenclature S Da W L AR V t1 t2 h p Vp Ap Cp k Dp Tf T0 Tm −V −V/V0

area of the projected particle (m2 ) diameter of the equivalent circular area (m) minor axis of the approximate ellipse of the projected particle (m) major axis of the approximate ellipse of the projected particle (m) aspect ratio (−) ﬁber content ratio (−) time required to increase the temperature of the particle until its melting point (s) time required to melt the particle completely at its melting point (s) heat transfer coefﬁcient of plasma (W m−2 K−1 ) density of the solid particle (kg m−3 ) volume of the solid particle (m3 ) surface area of the solid particle (m2 ) melting heat of the solid particle (J kg−1 ) speciﬁc heat (J kg−1 K−1 ) thermal conductivity of plasma (W m−1 K−1 ) diameter of the solid particle (m) temperature of plasma (K) initial temperature of the particle (K) melting point of the particle (K) ﬁber vanishing rate (−) normalized ﬁber vanishing rate (−)

equivalent circular area, the aspect ratio, and the ﬁber content ratio were used as particle shape indices. The relationship between the heat needed to melt the particle per surface area and the ﬁber vanishing rate were also considered in evaluating the applicability of the plasma method. Once the methodology was optimized, three types of pure asbestos were plasma-treated, and the content of the asbestos after its plasma treatment was evaluated. 2. Experiment 2.1. Experimental methodology Fig. 1 shows the experimental setup of the plasma treatment. To generate plasma, air streams were fed tangentially through a nozzle into a quartz torch (I.D.: 9.5 mm). Microwave power (2.45 GHz, maximum power 1.5 kW, IDX Co., Ltd.) was coupled to the gas as

Fig. 1. Experiment apparatus.

Reaction tube (mm) Swirl air ﬂow rate (L/min) Carrier gas ﬂow rate (L/min) Input power (W) Type of ﬁber particle: IBI wool (IW) SMF300UE (SMF) Fibermax (FM) Pure chrysotile (CHR) Pure amosite (AMO) Pure crocidolite (CRO) Particle density (kg/m3 )

Melting point (K) Melting enthalpy (kJ/mol) Fiber average diameter (␮m) Fiber average length (␮m)

I.D. = 9.5, O.D. = 11.6, L = 400 8.1, 11.4, 14.0 3.5, 5.1 1000 Al2 O3 = 0.46, SiO2 = 0.51 Fe = 0.83, Cr = 0.16, Ferritic stainless steel = 0.01 Al2 O3 = 0.72, SiO2 = 0.28 Mg3 Si2 O5 (OH)4 = 0.95 (Fe,Mg)7 (Si8 O22 )(OH)2 = 0.99 Na2 Fe3 2+ Fe2 3+ (Si8 O22 )(OH)2 = 0.99 IW = 2700; SMF = 7700; FM = 2900 CHR = 2550; AMO = 3475 CRO = 3250 IW = 2033; SMF = 1742; FM = 2143 CHR = 1794; AMO = 1572; CRO = 1466 IW = 436; SMF = 272; FM = 588 CHR = 1012; AMO = 359; CRO = 529 IW = 1.8–3, SMF = 5–10, FM = 4–6 CHR = 25.5, AMO = 24, CRO = 29 IW = 34, SMF = 15–50, FM = 48 CHR = 53, AMO = 96, CRO = 98.5

it passed through a rectangular waveguide. Pure argon gas was introduced for the plasma ignition before creating totally pure air plasma. Raw particles were fed constantly into the air plasma from the top of the quartz tube with a particle feeder (type: MF, Technoserve Co., Ltd.) and air as a carrier gas. The plasma-treated particles were trapped in water and ﬁltered. Then the samples were dried and analyzed via SEM (KEYENCE, VE-8800) to observe their melting condition and shape change. They were also analyzed via X-ray diffraction (RIGAKU, RINT-2200) to investigate their phase alteration. Three kinds of microﬁber particles were used as precursor materials: IBI wool (IW, Ibiden Co.); stainless ﬁber (SMF300UE, JFE Techno-research Co.); and Fibermax (FM, ITM Co.). The experiment was conducted by varying the particle feed rate and the swirl air gas ﬂow rate, which are the experiment conditions listed in Table 1. Once the methodology was optimized, experiment results on pure asbestos of the Japan Association for Working Environment Measurement (JAWE) were described. Due to the handling of hazardous materials, the experiments were conducted in an area cleared by the Ministry of Environment of Japan. The layout of the workplace is shown in Fig. 2. It was isolated with a double plastic sheet. A Negative Air Machine with HEPA ﬁlter was used as an air ventilator. A security zone was set at the entrance that was equipped with an air shower machine. Safety equipment such as dust masks, coveralls, gloves, and shoe covers were provided to prevent exposure to airborne asbestos while undressing in the workplace.

Fig. 2. Layout of the workplace.

A. Averroes et al. / Journal of Hazardous Materials 195 (2011) 405–413

407

Table 2 Average values of particle shape indices before treatment. Type of ﬁber particle

IBI wool

SMF300UE

Fibermax

Pure chrysotile

Pure amosite

Pure crocidolite

Da [␮m] AR [−]

16.7 0.32

25.0 0.43

17.4 0.19

35.8 0.485

42.9 0.25

48.1 0.29

2.2. Two-dimensional analysis 100–200 samples of both the treated and untreated particles were captured via SEM, after which an image analysis software (Scion Image) was used to measure the area of the projected particle (S) and the major axis (L) and minor axis (W) of an approximate ellipse of the projected particle. The diameter of the equivalent circular area (Da ) and the aspect ratio (AR) were used as the particle shape indices [Eqs. (1) and (2)], with the initial values shown in Table 2. The diameter of the equivalent circular area was

determined using Eq. (1), from the diameter of a circle that had the same area as the projected particle. The aspect ratio (AR) shows the ratio of the major axis to the minor axis of an approximate ellipse of the projected particles, as shown in Eq. (2).

Da = 2

AR =

W L

S

(1)

(2)

Fig. 3. SEM photographs of the microﬁber particles before (left) and after (right) their air plasma treatment for IBI wool (top), SMF300-UE (center), and Fibermax (bottom).

408

A. Averroes et al. / Journal of Hazardous Materials 195 (2011) 405–413

Fig. 4. XRD patterns of the microﬁber particles before and after their treatment for (a) IBI wool, (b) SMF300UE, and (c) Fibermax.

3. Results

that the phase change occurred during the plasma treatment of the stainless ﬁber.

3.1. Treatment of the microﬁber particles 3.1.1. Change in the shape of the particles after their plasma treatment Fig. 3 shows the SEM pictures of the microﬁber particles before and after their plasma treatment at 1 kW microwave power, a 11.4 L/min swirl air ﬂow rate, and a 3.5 L/min carrier gas ﬂow rate. At 200× magniﬁcation, all three types of microﬁber particles contained many ﬁbrous particles before their plasma treatment. Moreover, there was no spherical particle in the untreated microﬁber, but many spherical particles emerged from the treated ﬁber particles, except for Fibermax. Under several experiment conditions, the spheroidized particles were mixed with the agglomerated and unconverted particles.

3.1.2. Investigation by X-ray diffraction XRD analysis of both the untreated and treated microﬁber particles was conducted to detect the phase change after the treatment (Fig. 4). IBI wool and Fibermax showed no alteration in their phases, which showed the typical peaks of amorphous ceramic ﬁber and mullite ﬁber, respectively. The SMF300UE of the stainless ﬁber had different peaks before and after its plasma treatment, though. The peaks describe the X-ray diffraction (XRD) pattern of the ferrite stainless steel (2 = 44.6, 64.93, and 82.1) for the particles before the treatment [8], but the peaks of chromite and/or magnetite (2 = 35.5) appeared after the treatment [9]. It can thus be said

3.1.3. Effects of the particle feed rate on the particle shape indices The particle shape indices were used to investigate the results of the plasma treatment. The difference between the particle shape indices before and after the treatment was calculated and described with a delta (). The effects of the particle feed rate on the average difference in the equivalent diameter and the difference in the aspect ratio are shown in Fig. 5(a) and (b) at a 1 kW input power and a 11.1 L/min air ﬂow rate, with the values at point zero indicating the conditions before the treatment. In the case of IBI wool, Da had a large value at a small particle feed rate, then decreased and tended to be constant with the increase in the particle feed rate. Da seems to have had a peak with the increase in the particle feed rate for SMF300UE and Fibermax. Actually, due to the high standard deviation, the tendency that appeared was weak. Although, this could have indicated that almost all the particles were agglomerated since almost of Da was greater than zero. This is important because the bigger size will decrease the harmful effect of ﬁber. It showed, however, that AR had a different tendency for all types of microﬁber particles. The average AR of IBI wool was 0.4 and tended to remain constant, and the values of both SMF300UE and Fibermax tended to decrease as the particle feed rate increased. 3.1.4. Effects of the swirl air ﬂow rate on the particle shape indices Fig. 6(a) and (b) shows the behavior of the average difference in the equivalent diameter and the difference in the aspect ratio at the swirl air ﬂow rates of 8 L/min, 11.1 L/min, and 14 L/min. As the

Fig. 5. Effects of the particle feed rate on the (a) equivalent diameter and (b) average aspect ratio for an input power of 1000 W and a swirl air ﬂow rate of 11.1 L/min).

A. Averroes et al. / Journal of Hazardous Materials 195 (2011) 405–413

409

Fig. 6. Effects of the swirl air ﬂow rate on the (a) equivalent diameter and (b) average aspect ratio for an input power of 1000 W and a carrier gas ﬂow rate of 3.5 L/min.

swirl air ﬂow rate increased, the average Da decreased for IBI wool and Fibermax, but not for SMF300UE. This is interesting because the volume of plasma is considered to be lower at a higher swirl air ﬂow rate [9], which should result in a lower melting condition. The average Da of SMF300UE peaked at 14.0 L/min, though. This suggests that at this rate, the ﬁber particles of SMF300UE received a higher gravity force due to their density and tended to have a short transit time inside the plasma. As a result, SMF300UE melted only at the surface, attached to each other, and agglomerated to form a bigger size. On the other hand, the average AR indicates the lowest point for all types of microﬁber particles at a swirl air ﬂow of 14.0 L/min, as shown in Fig. 5(b), which was a reasonable result because the volume of the plasma plume was considered to have been lowest at that condition [10]. Thus, fewer particles passed through the hightemperature plasma region. The arithmetic average of the aforementioned Da and AR has a wide standard deviation that makes it difﬁcult to clearly conclude the tendency of the treatment results. By assuming, however, that each case had a normal distribution, the amount of particles left untreated was predicted.

3.1.5. Adoption of the ﬁber content rate As noted in the previous section, it was difﬁcult to clearly conclude the results tendency using the arithmetic average. Besides, the average value often fails to express the real conditions with

Fig. 8. Relationship between and the normalized ﬁber vanishing rate (−V/V0 ) of (a) the microﬁber particle only (IP = 1000 W and AFR = 11.4 L/min) and (b) of the microﬁber particle with pure asbestos (IP = 1000 W and AFR = 12.5 L/min).

respect to energy. Even if two particles have the same aspect ratio, the bigger particle needs more heat to melt. The arithmetic average is still important from the aspect of safety. According to the World Health Organization (WHO), asbestos ﬁber has an aspect ratio that is greater than 3:1 [11]. Based on this, the ﬁber content ratio is

Fig. 7. Behavior of the ﬁber content rate for an input power of 1000 W due to the change in the (a) particle feed rate and (b) swirl air ﬂow rate.

410

A. Averroes et al. / Journal of Hazardous Materials 195 (2011) 405–413

Fig. 9. SEM photographs of the pure asbestos before (left) and after (right) its air plasma treatment for chrysotile (top), amosite (center), and crocidolite (bottom).

deﬁned as the ratio of the projected area of the particles with an AR value below 0.33 to the total area of the projected particles, as follows.

V=

projected area of particle with AR ≤ 0.33

projected area of particle

× 100%

the ﬁber content ratio under several experiment conditions. These results indicate that this method would be useful for asbestos, the melting point of which is roughly similar to that of these microﬁbers.

(3)

The effects of the particle feed rate and the swirl air ﬂow rate on the ﬁber content ratio are shown in Fig. 7. The values at zero on the x axis represent the ﬁber content ratio of the microﬁber particles before the plasma treatment. The ﬁber content ratio of IBI wool and SMF300UE decreased to almost zero at a lower particle feed rate and a lower swirl air ﬂow rate. The ﬁber content ratio of Fibermax decreased only to about 0.7 for the experiments with the particle feed rate as a parameter and to 0.5 at the lowest value of the swirl air ﬂow rate. By adopting this new particle shape index, it can be concluded that IBI wool and SMF300UE were better treated than Fibermax and showed an approximately zero value for

3.1.6. Heat transfer between the plasma and the ﬁber particle Although the ﬁber content ratio can describe the tendency of the results under each experiment condition, the particle shape index, which can evaluate the potential of this method, is necessary. In this section, the heat balance between the plasma and the microﬁber particles is discussed. The mechanism of the plasma treatment can be described as follows. First, the particles receive heat from the plasma and melt instantaneously, and then the melted particles rapidly cool at the plasma afterglow. Thus, by assuming that there is no temperature gradient inside the

A. Averroes et al. / Journal of Hazardous Materials 195 (2011) 405–413

411

particle and that the effect of radiation can be neglected, the complete melting time is described in Eq. (4). 1 t1 + t2 = h

Vp Cp ln Ap

Vp Cp h (t1 + t2 ) = ln Ap

Tf − T0 Tf − Tm Tf − T0

Tf − Tm

Vp + Ap (Tf − Tm ) +

Vp = Ap (Tf − Tm )

(4)

(5)

It was assumed that the plasma temperature was 3000 K [12,13] and that all the particles before the treatment were cylindrical. Then the new parameter in Eq. (5) was introduced and calculated. In the equation, is the product of the multiplication of the heat coefﬁcient of the plasma and the complete melting time, which indicates the amount of energy needed to melt the particles completely per unit surface area. The relationship between and the normalized ﬁber vanishing rate (−V/V0 ) is shown in Fig. 8. The ﬁber vanishing rate (−V) is the difference between the ﬁber content ratio before and after the treatment, and V0 is the ﬁber content ratio before the treatment. As shown in Fig. 8, the normalized ﬁber vanishing rate had a downward tendency with an increase in , which has a rational tendency. The same experiment conditions brought about nearly the same normalized ﬁber vanishing rate. Thus, the ﬁber vanishing rate is suggested as a proper index for evaluating the result of this treatment method. By assuming that the correlation in Fig. 8 is valid, an approximate curve line can be drawn, and it becomes possible to estimate the value of that makes −V/V0 equal to unity. Moreover, by introducing the natures of pure asbestos in Eq. (5), the normalized ﬁber vanishing rate is expected to reach unity for a cylinder diameter equal to or less than 10 ␮m. Actually, it is still necessary to verify this relationship by calculating a numerical model. 4. Prove treatment of pure asbestos 4.1. SEM analysis of the plasma treatment of pure asbestos Fig. 9 shows SEM pictures of the pure asbestos before and after its plasma treatment at 1 kW input power, a 12.5 L/min swirl air ﬂow rate, and a 2.5–5.3 L/min carrier gas ﬂow rate. At 500× magniﬁcation, some ﬁbers were still observed in amosite and crocidolite, but not in chrysotile, as shown in Fig. 8. The bigger size of the plasma-treated particles indicates the agglomeration of the particles during the treatment. 4.2. Investigation of the plasma content in the plasma-treated pure asbestos To determine the effects of plasma treatment of pure asbestos, the XRD patterns of the pure asbestos after the treatment were compared with the XRD patterns before the treatment (Fig. 10). Asbestos has some typical peaks that represent its content. Asbestos peaks (chrysotile = 12 and 24; amosite = 10.7, 27.3, and 29.1; and crocidolite = 10.5, 28.6, and 32.8) were conﬁrmed for all the untreated pure asbestos samples. The peaks of chrysotile mostly disappeared after its treatment and new crystalline peaks of forsterite (2 = 17.4, 23 and 32.4) and enstatite (2 = 31) appeared. The formation of forsterite and enstatite will occur as chrysotile heated more than 850 ◦ C [14,15]. However, the typical peaks of amosite and crocidolite remained, though they signiﬁcantly decreased. In the case of amosite, the formation of oxy-amosite occurs at relatively low temperature (∼500 ◦ C). The breakdown of oxy-amosite begins at = 800 ◦ C produces the decomposition of hematite, magnetite, and possibly quartz [15]. Some new peaks of ␣-quartz (2 =21 and 27) and magnetite (2 = 18) appeared as

Fig. 10. XRD patterns of the pure asbestos before its plasma treatment for (a) chrysotile, (c) amosite, and (e) crocidolite; and after its treatment for (b) chrysotile, (d) amosite, and (f) crocidolite. 䊉, chrysotile; , amosite; , crocidolite.

shown in Fig. 10(d). For transformation of crocidolite, the formation of oxy-crocidolite occurs at temperatures up to 650 ◦ C. Above 650 ◦ C the oxy-crocidolite begins to break down with formation of acmite, hematite, cristobalite, and possibly some magnetite [15]. Some new peaks of acmite (2 = 30), hematite (2 = 33), and magnetite (2 = 18) were detected after treatment as shown in Fig. 10(f). The plasma-treated pure asbestos was analyzed with phasecontrast polarized light microscopy to conﬁrm the results of the XRD analysis. This was because XRD cannot distinguish between asbestiform and nonasbestiform materials with the same mineral phase [16]. This is why the qualitative analysis of asbestos via XRD must be conﬁrmed via either phase-contrast microscopy (PCM) or SEM. Actually, the dispersion staining methods of light microscopy can uniquely identify asbestos ﬁbers with diameters as small as 1 ␮ m. This technique involves suspending liquids of known refractive indices (Cargille refractive index liquids) and observing their color display by means of a dispersion staining objective [17]. Asbestos minerals have double refractive indices. By immersing the ﬁber in appropriate liquids, the optical properties of asbestos signals can be detected from a unique color combination with

412

A. Averroes et al. / Journal of Hazardous Materials 195 (2011) 405–413

Fig. 11. Pictures from a phase-contrast microscope (100×) of the pure asbestos before its treatment (left) and after its treatment (right) for chrysotile (top), amosite (center), and crocidolite (bottom).

the polarized light. This identiﬁcation was conducted via phasecontrast microscopy (Nikon, Eclipse 80i) on pure asbestos before and after the treatment. The asbestos dispersion-staining colors (chrysotile = blue-violet, amosite = peach, and crocidolite = orange) for the particles after the treatment were not detected, as shown in Fig. 11, unlike the particles before the treatment. From the aforementioned results, asbestos peaks were detected via XRD but were undetected via PCM. Many needle-shaped particles remained in the treated particles of amosite and crocidolite which was suggested due to the cohesion of pure asbestos particles, especially for amosite and crocidolite. This somehow causes bunch formation of the ﬁber particles, which affects their steady feeding. This condition affected the temperature of the air plasma when many particles were fed into the quartz tube in one burst. Moreover, referring to Fig. 8 in Section 3.1.6, the value of the normalized ﬁber vanishing rate of the pure asbestos is small. It can be considered that the experiment results for pure asbestos did

not reach unity because its ﬁber structure formed a bunch shape that increased its average diameter to more than 10 ␮m, as shown in Table 1. Pure asbestos is practically never used, however, and airborne asbestos is usually dispersed in the air, with a smaller diameter of about 1–2 ␮m [18]. 5. Conclusion Atmospheric microwave air plasma was used to treat microﬁber particles, followed by the conduct of the prove test for pure asbestos-containing materials by altering the particle feed rate and the swirl air ﬂow rate as the experiment parameters. The SEM results showed that the plasma-treated microﬁber particles were spheroidized and agglomerated. The XRD analysis results showed that the peaks of the pure amosite and the pure crocidolite remained, but not that of the pure chrysotile. The PCM results indicated, however, that there was no asbestos in the plasma-treated

A. Averroes et al. / Journal of Hazardous Materials 195 (2011) 405–413

pure asbestos particle. Therefore, this method was shown to be suited for pure chrysotile and needs some improvement especially for pure amosite and pure crocidolite. Regarding energy use, the ﬁber content ratio and the ﬁber vanishing rate were used as the particle shape indices. A rational correlation between the amount of energy needed per unit surface area () and the normalized ﬁber vanishing rate was observed. Finally, it is suggested that this atmospheric microwave air plasma treatment method has potential for the treatment of airborne building materials that contain asbestos. That means this technology will be efﬁcient to treat airborne asbestos that contained in exhaust gas of asbestos disposal station. Airborne building materials such as cement and concrete will also be treated due to the high temperature of plasma. Acknowledgments The authors are grateful to the Ministry of Education, Culture, Sports, Science, and Technology of Japan, and to the Japan Science and Technology Agency, for their ﬁnancial support for this research. References [1] F. Turci, M. Tomatis, S. Mantegna, G. Cravotto, B. Fubini, A new approach to the decontamination of asbestos-polluted waters by treatment with oxalic acid under power ultrasound, Ultrason. Sonochem. 15 (2008) 420–427. [2] C. Leonelli, P. Veronesi, D.N. Boccaccini, M.R. Rivasi, L. Barbieri, F. Andreola, I. Lancellotti, D. Rabitti, G.C. Pellacani, Microwave thermal inertisation of asbestos containing waste and its recycling in traditional ceramics, J. Hazard. Mater. B 135 (2006) 149–155. [3] A.J. Darnton, D.M. McElvenny, J.T. Hodgson, Estimating the number of asbestosrelated lung cancer death in Great Britain from 1980 to 2000, Ann. Occup. Hyg. 50 (2006) 29–38.

413

[4] M. Fujishige, R. Sato, A. Kuribara, I. Karasawa, A. Kojima, Low-temperature decomposition of sprayed-on asbestos, J. Ceram. Soc. Jpn. 114 (2006) 1133–1137. [5] P. Plescia, D. Gizzi, S. Benedetti, L. Camilucci, C. Fanizza, P.D. Simone, F. Paglietti, Mechanochemical Treatment to recycling asbestos-containing waste, Waste Manage. 23 (2002) 209–218. [6] E. Gomez, D. Amutha Rani, C.R. Cheeseman, D. Deegan, M. Wise, A.R. Boccaccini, Thermal plasma technology for the treatment of wastes: a critical review, J. Hazard. Mater. 161 (2009) 614–626. [7] S. Nakanishi, H. Sekiguchi, Comparison of reforming behaviors of hexane and isooctane in microwave steam plasma, J. Jpn. Petrol. Inst. 48 (2005) 22–28. [8] H. Tsujimura, T. Goto, Y. Ito, Electrochemical surface nitriding of SUS 430 ferritic stainless steel, Mater. Sci. Eng. A355 (2003) 315–319. [9] D. Majuste, M.B. Mansur, Leaching of the ﬁne fraction of the argon oxygen decarburization with lance (AOD-L) sludge for the prefential removal of iron, J. Hazard. Mater. 162 (2009) 356–364. [10] K.M. Green, M.C. Borras, P.P. Woskov, G.J. Flores, K. Hadidi, P. Thomas, Electronic excitation temperature proﬁles in an air microwave plasma torch, IEEE Trans. Plasma Sci. 29 (2001) 399–406. [11] K. Donaldson, C.L. Tran, An introduction to the short-term toxicology of respirable industrial ﬁbres, Mutat. Res. 553 (2004) 5–9. [12] S.Y. Moon, W. Choe, Parametric study of atmospheric pressure microwaveinduced Ar/O2 plasmas and the ambient air effect on the plasma, Phys. Plasmas 13 (2006), 103503:1-6. [13] J. Happold, P. Lindner, R. Roth, Spatially resolved temperature measurement in an atmospheric plasma torch using the A2 + , v = 0 → X2 , v = 0 OH band, J. Phys. D: Appl. Phys. 39 (2006) 3615–3620. [14] T. Zaremba, A. KArzakala, J. Piotrowski, D. Garcczorz, Study on the thermal decomposition of chrysotile asbestos, J. Therm. Anal. Calorim. 101 (2010) 479–485. [15] M. Jeyaratnam, N.G. West, A Study of heat-degraded chrysotile, amosite, and crocidolite by X-ray diffraction, Ann. Occup. Hyg. 38 (1994) 137–148. [16] D.A. Vallero, M.E. Beard, Selecting appropriate analytical methods to characterize asbestos in various media, the practice periodical of hazardous, toxic, and radioactive, Waste Manage. 13 (2009) 249–260. [17] W.C. McCrone, Detection and Identiﬁcation of asbestos by microscopical dispersion staining, Environ. Health Perspect. 9 (1974) 57–61. [18] S. Hashimoto, H. Takeda, A. Okuda, A. Kambayashi, S. Honda, H. Awaji, K. Fukuda, Detoxiﬁcation of asbestos-containing building material waste and its application to cement product, J. Ceram. Soc. Jpn. 115 (2007) 290–293.

Journal of Hazardous Materials 195 (2011) 414–421

Contents lists available at SciVerse ScienceDirect

Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat

Study of cyanotoxins presence from experimental cyanobacteria concentrations using a new data mining methodology based on multivariate adaptive regression splines in Trasona reservoir (Northern Spain) P.J. Garcia Nieto a,∗ , F. Sánchez Lasheras b , F.J. de Cos Juez c , J.R. Alonso Fernández d a

Department of Mathematics, Faculty of Sciences, University of Oviedo, 33007 Oviedo, Spain Department of Construction and Manufacturing Engineering, University of Oviedo, 33204 Gijón, Spain c Mining Exploitation and Prospecting Department, University of Oviedo, 33004 Oviedo, Spain d Water Planning Ofﬁce, Cantabrian Basin Authority, 33071 Oviedo, Spain b

a r t i c l e

i n f o

Article history: Received 9 June 2011 Received in revised form 13 August 2011 Accepted 22 August 2011 Available online 27 August 2011 Keywords: Statistical learning techniques Cyanobacteria Cyanotoxins Multivariate adaptive regression splines (MARS)

a b s t r a c t There is an increasing need to describe cyanobacteria blooms since some cyanobacteria produce toxins, termed cyanotoxins. These latter can be toxic and dangerous to humans as well as other animals and life in general. It must be remarked that the cyanobacteria are reproduced explosively under certain conditions. This results in algae blooms, which can become harmful to other species if the cyanobacteria involved produce cyanotoxins. In this research work, the evolution of cyanotoxins in Trasona reservoir (Principality of Asturias, Northern Spain) was studied with success using the data mining methodology based on multivariate adaptive regression splines (MARS) technique. The results of the present study are two-fold. On one hand, the importance of the different kind of cyanobacteria over the presence of cyanotoxins in the reservoir is presented through the MARS model and on the other hand a predictive model able to forecast the possible presence of cyanotoxins in a short term was obtained. The agreement of the MARS model with experimental data conﬁrmed the good performance of the same one. Finally, conclusions of this innovative research are exposed. © 2011 Elsevier B.V. All rights reserved.

1. Introduction Cyanobacteria also known as blue-green algae, blue-green bacteria, and cyanophyta is a phylum of bacteria that obtain their energy through photosynthesis. Cyanobacteria can be found in almost every conceivable environment: in oceans, lakes and rivers as well as on land. Even they ﬂourish in Arctic and Antarctic lakes [1], hotsprings and wastewater treatments plants. Aquatic cyanobacteria is probably best known for the extensive and highly visible blooms that can form in both freshwater and the marine environment. The association of toxicity with such blooms has frequently led to the closure of recreational waters when blooms are observed. Some cyanobacteria produce toxins, called cyanotoxins [2], and in freshwater ecosystems are the most common cause of eutrophication. The blooms are not always green [3]. They can be blue, and some cyanobacteria species are coloured brownish-red. The water can become malodorous when the cyanobacteria in the bloom die.

∗ Corresponding author. Tel.: +34 985 103417; fax: +34 985 103354. E-mail address: [email protected] (P.J. Garcia Nieto). 0304-3894/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.jhazmat.2011.08.061

Cyanotoxins are an important environmental problems in reservoirs [4]. Water is never perfectly clean and polluted water is also a continuing threat to human health and welfare [5]. The toxins include potent neurotoxins, hepatotoxins, cytotoxins, and endotoxins [6]. Most reported incidents of poisoning by microalgal toxins have occurred in freshwater environments, and they are becoming more common and widespread [7]. Generally these blooms are harmless, but if not they are called harmful algal blooms (HABs) [8]. HABs can contain toxins which result in ﬁsh kill and can also be fatal to humans [9]. The aim of this research is to construct a multivariate adaptive regression splines (MARS) model to identify spatial cyanotoxins in waterways in the Trasona reservoir (Principality of Asturias, Northern Spain) (see Fig. 1(a) and (b)). Multivariate adaptive regression splines (MARS) technique is a form of regression analysis introduced by Friedman in 1991 [10–13]. It is a non-parametric regression technique and can be seen as an extension of linear models that automatically models non-linearities and interactions as those analyzed in this innovative research work successfully. The Trasona reservoir, which was initially destined to the industrial supply, is complemented at present with a recreational utilization as a high performance training centre of canoeing. It is an eutrophic ecosystem, which has been characterized for cyanobacteria

P.J. Garcia Nieto et al. / Journal of Hazardous Materials 195 (2011) 414–421

415

Fig. 1. (a) Aerial photograph of the city of Avilés (Northern Spain) (2) and Trasona reservoir (1); and (b) an aerial photograph of Trasona reservoir in great detail (lower).

outcrops in certain periods, which sometimes has produced variable concentrations of cyanotoxinas, mainly mycrocistins. This innovative research work is structured as follows. In the ﬁrst place, the necessary materials and methods are described to carry out this study. Next the obtained results are shown and discussed. Finally, the main conclusions drawn from the results are exposed. 2. Materials and methods 2.1. Experimental data set The data used for the MARS analysis were collected over ﬁve years (2006–2010) from lots of samples in Trasona reservoir and the total number of data processed was about ﬁve hundred and eleven values. The supplementary site-speciﬁc experimental data associated with this article can be found at the following online link: http://dl.dropbox.com/u/36679320/Trasona reservoir data.xls. The information is quantitative on the abundance of phytoplankton species. Speciﬁcally, this reservoir was sampled several times a month from January 1, 2006 to December 31, 2010, following

the sampling protocols for lakes and reservoirs of the Spanish Ministry of Environment and Rural and Marine Affairs, which are consistent with the guidelines established by the European Union and international agencies dealing with these issues [4–9]. In practice, a single point of sampling is taken into account in the place of greater depth of the reservoir, which is determined with a depth gauge [9]. The samples were taken with a Niskin hydrographic bottle (see Fig. 2(a)) at different depths in the zone corresponding to the depth of the water in the reservoir that is exposed to sufﬁcient sunlight for photosynthesis to occur called the euphotic zone [5]. This zone is determined from the Secchi depth which is the depth at which the pattern on the Secchi disk (see Fig. 2(b)) is no longer visible and it is taken as a measure of the transparency of the water in lakes, reservoirs and oceans. The values of phytoplankton and concentrations of cyanotoxins and chlorophyll were determined from a sample composed of ﬁve homogeneous subsamples obtained with the hydrographic bottle at various equidistant depths in the euphotic zone [14–16]. The main goal of this research work is to obtain the dependence relationship of the cyanotoxins (output variable) of the Trasona reservoir as a function of the following input variables [17]:

416

P.J. Garcia Nieto et al. / Journal of Hazardous Materials 195 (2011) 414–421

Fig. 2. (a) A Niskin hydrographic bottle about to be lowered into the water; and (b) different kinds of Secchi disks.

• Microcystis aeruginosa: Is a type of harmful blue-green algae which is also referred to as colonial cyanobacteria. • Woronichinia naegeliana: Is a kind of cyanobacteria present in waters of a lower trophic status. • Other cyanobacteria: They represent the rest of cyanobacteria excluding the two previous ones. • Diatoms: Are a major group of algae, and are one of the most common types of phytoplankton. • Chrysophytes: Are small ﬂagellates that are a yellowish brown colour. They can also be found singly or in a colony. • Chlorophytes: Refer to a highly paraphyletic group of all the green algae within the green plants. • Other species of the phytoplankton: They represent the rest of the phytoplankton excluding all the previous ones. All the input variables are measured in number of cells per milliliter and the output variable (cyanotoxins) in micrograms per liter. To ﬁx ideas, in this research work, physical–chemical parameters normally used in limnological studies have been measured [7]. Analyses of chlorophyll have been carried out to study of the phytoplankton. Fig. 3(a) shows the evolution of chlorophyll concentration and cyanobacteria cell number per milliliter in the Trasona reservoir from January of 2006 to December of 2010. Higher levels of both variables are observed at certain periods of the years 2006–2008, which are signiﬁcantly greater than the values obtained in the years 2009 and 2010. The peaks in Fig. 3(a) correspond to the cyanobacteria blooms: summer and fall of those years. However, there are no cyanobacteria blooms in years 2009 and 2010. Fig. 3(b) shows the evolution of cyanotoxins

concentration and cyanobacteria cell number per milliliter in the Trasona reservoir from January of 2006 to December of 2010. Similarly, the peaks in Fig. 3(b) correspond to the cyanobacteria blooms and large concentrations of cyanotoxins. Speciﬁcally, cyanobacteria cell number per milliliter was less than 50,000 and cyanotoxins concentration was always zero in 2009 and 2010. In fact, the Trasona reservoir is an eutrophic ecosystem [15] which has been characterized for the presence of cyanobacteria. These last ones sometimes have produced variable concentrations of cyanotoxins, mainly microcystins [16]. Microcystins are cyclic nonribosomal peptides produced by cyanobacteria. They are cyanotoxins and can be very toxic for plants and animals including humans [17]. Their hepatotoxicity may cause serious damage to the liver [18]. Once the problem has been identiﬁed, civil works have been carried out in order to diminish the nutrients contributions to the reservoir although a part of spillages still reaches the same one. The guideline values for safe recreational water quality raises alert level 2 [19] with values greater than 100,000 cells per milliliter and a microcystin concentration greater than 20.0 ␮g/l (see Fig. 3(a) and (b)). The inventories of cells were taken through an inverted microscope on settled samples. The cyanotoxins have been analyzed by means of the high-performance liquid chromatography (HPLC) technique [20]. High-performance liquid chromatography (or highpressure liquid chromatography) is a chromatographic technique that can separate a mixture of compounds and is used in biochemistry and analytical chemistry to identify, quantify and purify the individual components of the mixture. With the HPLC technique, a pump (rather than gravity) provides the higher pressure required to move the mobile phase and analyte through the densely packed column. The increased density arises from smaller particle sizes. This allows for a better separation on columns of shorter length when compared to ordinary column chromatography. The Trasona reservoir is located near the industrial city of Avilés (Principality of Asturias, Northern Spain). Practically chained to the Trasona reservoir, it is possible to observe a wetland created artiﬁcially in order to shelter one changeable aquatic avifauna. This lagoon is able to store approximately 50,000 m3 of water and the almost constant level of the water sheet of this lagoon allows the building of nests of different species of birds. Both the Trasona reservoir and the wetland belong to a ZEPA (zone of special protection for the birds) area [21–23]. 2.2. Multivariate adaptive regression splines (MARS) method Multivariate adaptive regression splines (MARS) is a multivariate nonparametric classiﬁcation/regression technique introduced by Friedman [10–13,24,25]. The theoretical model that is explained below has already been presented by the authors in previous researches [26,27]. In spite of this fact and due to its interest for the reader in order to achieve a full understanding of the research that is presented in this paper. Its main purpose is to predict the values of a continuous dependent variable, y (n × 1), from a set of × p). The MARS model can independent explanatory variables, X(n be represented as: + e y = f (X)

(1)

and where f is a weighted sum of basis functions that depend on X e is an error vector of dimension (n × 1). MARS does not require any a priori assumptions about the underlying functional relationship between dependent and independent variables. Instead, this relation is uncovered from a set of coefﬁcients and piecewise polynomials of degree q (basis func y ). The tions) that are entirely “driven” from the regression data (X, MARS regression model is constructed by ﬁtting basis functions to

P.J. Garcia Nieto et al. / Journal of Hazardous Materials 195 (2011) 414–421

417

Fig. 3. (a) Evolution of chlorophyll concentration and cyanobacteria cell number per milliliter as a function of time in the Trasona reservoir from January of 2006 to December of 2010; and (b) evolution of cyanotoxins concentration and cyanobacteria cell number per milliliter as a function of time in the Trasona reservoir from January of 2006 to December of 2010.

distinct intervals of the independent variables. Generally, piecewise polynomials, also called splines, have pieces smoothly connected together. In MARS terminology, the joining points of the polynomials are called knots, nodes or breakdown points. These will be denoted by the small letter t. For a spline of degree q each segment is a polynomial function. MARS uses two-sided truncated power functions as spline basis functions, described by the following equations [10–13]:

[−(x

q − t)]+

=

[+(x

q − t)]+

=

(t − x)q

if x < t

0

otherwise

(t − x)q

if x ≥ t

0

otherwise

(2)

(3)

where q(≥ 0) is the power to which the splines are raised and which determines the degree of smoothness of the resultant function estimate. When q = 1, which is the case in this study, only simple linear splines are considered.

The MARS model of a dependent variable y with M basis functions (terms) can be written as [24–27]: yˆ = fˆM (x ) = c0 +

M

cm Bm (x )

(4)

m=1

where yˆ is the dependent variable predicted by the MARS model, c0 is a constant, Bm (x ) is the mth basis function, which may be a single spline basis functions, and cm is the coefﬁcient of the mth basis functions. Both the variables to be introduced into the model and the knot positions for each individual variable have to be optimized. For a containing n objects and p explanatory variables, there data set X are N = n × p pairs of spline basis functions, given by Eqs. (2) and (3), with knot locations xij (i = 1, 2, . . ., n ; j = 1, 2, . . ., p). A two-step procedure is followed to construct the ﬁnal model. First, in order to select the consecutive pairs of basis functions of the model, a two-at-a-time forward stepwise procedure is implemented [25,28,29]. This forward stepwise selection of basis function leads to a very complex and overﬁtted model. Such a model, although it ﬁts the data well, has poor predictive abilities

418

P.J. Garcia Nieto et al. / Journal of Hazardous Materials 195 (2011) 414–421 Table 3 Evaluation of the importance of the variables that form the model according to criteria Nsubsets, GCV and RSS.

Table 1 Set of input variables used in this study. Input variables (cell/ml)

Name of the variable

Microcystis aeruginosa Woronichinia naegeliana Other cyanobacteria Diatoms Chrysophytes Chlorophytes Other species of the phytoplankton Microcystis aeruginosa × Woronichinia naegeliana (synergistic interaction variable)

Microcystis aeruginosa Woronichinia naegeliana Other species Cyanobacteria Diatoms Chrysophytes Chlorophytes Other phyto Microcys × Worochinia

Variable

Nsubsets

GCV

RSS

Microcys × Worochinia Other species Cyanobacteria Microcystis aeruginosa Woronichinia naegeliana

15 13 12 10

100 51.25753 23.66865 12.10512

100 52.41336 24.81192 13.28461

which each variable is included) and the residual sum of squares RSS [32]. 3. Analysis of results and discussion

for new objects. To improve the prediction, the redundant basis functions are removed one at a time using a backward stepwise procedure. To determine which basis functions should be included in the model, MARS utilizes the generalized cross-validation (GCV) [10–13,25,30,31]. In this way, the GCV is the mean squared residual error divided by a penalty dependent on the model complexity. The GCV criterion is deﬁned in the following way [10–13,25]: GCV (M) =

(1/n)

n i=1

(yi − fˆM (xi ))

(1 − C(M)/n)

2

2

(5)

where C(M) is a complexity penalty that increases with the number of basis functions in the model and which is deﬁned as [10–13]: C(M) = (M + 1) + dM

(6)

where M is the number of basis functions in Eq. (4), and the parameter d is a penalty for each basis function included into the model. It can also be regarded as a smoothing parameter. Large values of d lead to fewer basis functions and therefore smoother function estimates. For more details about the selection of the d parameter, see the references [10–13,25]. In our studies, the parameter d equals 2, and the maximum interaction level of the spline basis functions is restricted to 3. 2.3. The importance of the variables in the MARS model Once the MARS model is constructed, it is possible to evaluate the importance of the explanatory variables used to construct the basis functions. Establishing predictor importance is in general a complex problem which in general requires the use of more than one criterion. In order to obtain reliable results, it is convenient to use the GCV parameter explained before together with the parameters Nsubsets (criterion counts the number of model subsets in

The list of input variables taken into account is the research work shown in Table 1 [33–35]. As it can be observed one of the variables is formed by the product of the variable M. aeruginosa multiplied by the variable W. naegeliana due to the coexistence of these two species of cyanobacteria in order to reproduce their dynamics without interference from external factor. This mathematical formulation adds a multiplicative additional term to account for the two species’ interactions according to a more realistic mathematical modelling in Biology [36,37]. This kind of interaction (synergistic interaction) will be explained later in more detail. All the input variables are measured in number of cells per milliliter and the output variable (cyanotoxins) in micrograms per liter. The total number of prediction variables used to build the MARS model was 8. In this work, a second-order MARS model has been used, so that the basis functions of the model consist of linear and second-order splines and the maximum number of terms was not limited (no pruning). The results of the MARS model computed using all the available data observations is shown in Table 2. Table 2 shows a list of the 16 main basis functions of the MARS models and their coefﬁcients. Please note that h(x) = x if x > 0 and h(x) = 0 if x ≤ 0. Therefore, the MARS model is a form of non-parametric regression technique and can be seen as an extension of linear models that automatically models non-linearities and interactions as a weighted sum of basis functions called hinge functions [10–13]. The predicted response or cyanotoxins presence is now a better ﬁt to the original values since the MARS model has automatically produced a kink in the predicted dependent variable to take into account non-linearities. A graphical representation of the terms that constitute the model can be seen in Fig. 4. In this research work, the ﬁtted MARS model has a coefﬁcient of determination R2 equal to 0.84 and a correlation coefﬁcient equal to 0.91. These results indicate an important goodness of ﬁt, that is

Table 2 List of basis functions of the MARS model and their coefﬁcients ci . Bi

Deﬁnition

ci

B1 B2 B3 B4 B5 B6 B7 B8 B9 B10 B11 B12 B13 B14 B15 B16

1 h(0,Microcystis aeruginosa-135,000) h(0,135,000-Microcystis aeruginosa) h(0,Microcys × Worochinia-16,900) h(0,16,900-Microcys × Worochinia) h(0,Microcystis aeruginosa-135,000) × h(0,Woronichinia naegeliana-110,000) h(0,Microcystis aeruginosa-135,000) × h(0,110,000-Woronichinia naegeliana) h(0,Microcystis aeruginosa-135,000) × h(0,Woronichinia naegeliana-120,000) h(0,Microcystis aeruginosa-70,000) × h(0,16,900-Microcys × Worochinia) h(0,70,000-Microcystis aeruginosa) × h(0,16,900-Microcys × Worochinia) h(0,Microcystis aeruginosa-140,000) × h(0,Microcys × Worochinia-16,900) h(0,Woronichinia naegeliana-130,000) × h(0,Microcys × Worochinia-10,377) h(0,130,000-Woronichinia naegeliana) × h(0,Microcys × Worochinia-10,377) h(0,Other species Cyanobacteria-39,617) × h(0,Microcys × Worochinia-10,377) h(0,Other species Cyanobacteria-60,000) × h(0,Microcys × Worochinia-10,377) h(0,60,000-Other species Cyanobacteria) × h(0,Microcys × Worochinia-10,377)

1.8 × 103 0.042 0.025 1.9 0.011 −5.2 × 10−6 −8.6 × 10−7 1.1 × 10−5 3.3 × 10−6 −1.5 × 10−6 2.5 × 10−5 −7.4 × 10−6 −7.9 × 10−6 2.6 × 10−5 1.4 × 10−5 2.2 × 10−5

P.J. Garcia Nieto et al. / Journal of Hazardous Materials 195 (2011) 414–421

419

Fig. 4. Graphical representation of the terms that constitute the MARS model: (a) ﬁrst order term of the variable Microcystis aeruginosa; (b) ﬁrst order term of the product of the variables Microcystis aeruginosa and Woronichinia naegeliana; (c) second order term of the variables Microcystis aeruginosa and the synergistic interaction variable Microcystis aeruginosa Woronichinia naegeliana; (d) second order term of the variables synergistic interaction variable Microcystis aeruginosa Woronichinia naegeliana and Woronichinia naegeliana; (e) second order term of the variables synergistic interaction variable Microcystis aeruginosa Woronichinia naegeliana and other species of cyanobacteria; and (f) second order term of the variables Woronichinia naegeliana and Microcystis aeruginosa.

to say, a good agreement is obtained between our model and the observed data. It must be taken into account that the goodness of ﬁt should not be considered as a proof of the predictive ability of the MARS model. According to the results shown in Table 3, the most important variables for the prediction of the cyanotoxins (output variable) are as follows: M. aeruginosa multiplied by W. naegeliana (Microcys × Worochinia), Other species of cyanobacteria, the Microcystis aeruginosa and ﬁnally the Woronichinia naegeliana on its own. In order to guarantee the ability prediction of the MARS model an exahustive cross-validation algorithm is used. The referred algorithm consists on the creation of 511 different MARS model (one model for each observation). Each of this model was trained using all the data except the observarion for which it was created as the validation was performed predicting its corresponding output value. The results obtained by means of this procedure are shown in Fig. 5. The main ﬁnding of this study is the interaction between input variables M. aeruginosa and W. naegeliana not considered in previous works [38–40] and it is the result of the exhaustive work carried out on the Trasona reservoir for ﬁve years and presented here. This led to the consideration of a new input variable equal to

the product of the concentrations of the two above input variables in addition to other input variables empirically measured in Trasona reservoir. The consideration of this interaction is known as synergy or synergistic behavior and it has not been considered in previous research works. It is well known that M. aeruginosa is potentially toxic and produces a type of toxin known as microcystin. Up to now, there is no evidence of the toxicity of the W. naegeliana in Spain and there is only a partial evidence of its toxicity outside Spain [38]. This synergistic behavior is the result of joint action of two or more causes, but characterized by having a greater effect than that resulting from the sum of these causes, that is to say, the production of cyanotoxins from M. aeruginosa can be increased by combined presence of both species: M. aeruginosa and W. naegeliana. Synergy has been advanced as a hypothesis on how complex systems operate. Environmental systems may react in a nonlinear way to perturbations, so that the outcome may be greater than the sum of the individual component alterations. Synergistic responses are a complicating factor in environmental modelling. Finally, this research work was able to estimate the presence of cyanobacteria blooms from 2006 to 2010 in agreement to the actual cyanobacteria blooms observed with great accurateness and success (see Fig. 5).

420

P.J. Garcia Nieto et al. / Journal of Hazardous Materials 195 (2011) 414–421

Fig. 5. Comparison between the three blooms of cyanobacteria observed and predicted by the model MARS on the Trasona reservoir from 2006 to 2010.

4. Conclusions

References

In the ﬁrst place, the main purpose of this research was to build a cyanotoxin diagnostic model by using MARS technique in Trasona reservoir with the site-speciﬁc experimental data and this goal was achieved in this work successfully. Future researches may aim at collecting more important physical, chemical and biological variables that will increase the calculation accuracies. Secondly, the predicted results for the MARS model have demonstrated to be consistent with the observed actual cyanobacteria blooms history from 2006 to 2010. In this way, this original and innovative methodology can be applied to other reservoirs with similar or different sources of pollutants, but it is always necessary to take into account the speciﬁcities of each location. Finally, one of the main ﬁndings of this study is the existence of synergistic behavior between two cyanobacteria: speciﬁcally, M. aeruginosa and W. naegeliana. Synergy, in general, may be deﬁned as two or more things functioning together to produce a result not independently obtainable. This performance is the result of joint action of the two cyanobacteria in the production of cyanotoxins: the efﬁciency can be increased by combined action of both species.

[1] L. Spoof, K.A. Berg, J. Rapala, K. Lahti, L. Lepisto, J.S. Metclaf, G.A. Codd, J. Meriluoto, First observation of cylindrospermopsin in Anabaena lapponica isolated from the boreal environment (Finland), Environ. Toxicol. 21 (6) (2006) 552–560. [2] A. Quesada, E. Moreno, D. Carrasco, T. Paniagua, L. Wormer, C. de Hoyos, A. Sukenik, Toxicity of Aphanizomenon ovalisporum (cyanobacteria) in a spanish water reservoir, Eur. J. Phycol. 41 (1) (2006) 39–45. [3] M.J. Smith, G.R. Shaw, G.K. Eaglesham, L. Ho, J.D. Brookes, Elucidating the factors inﬂuencing the biodegration of cylindrospermopsin in drinking water sources, Environ. Toxicol. 23 (3) (2008) 413–421. [4] V. Vasconcelos, Eutrophication, toxic cyanobacteria and cyanotoxins: when ecosystems cry for help, Limnetica 25 (1–2) (2006) 425–432. [5] M.J. Dasí, M.R. Miracle, A. Camacho, J.M. Soria, E. Vicente, Summer phytoplankton assemblages across trophic gradients in hard-water reservoirs, Hydrobiologia 369–370 (1998) 27–43. [6] A. Dixit, R.K. Dhaked, S.I. Alam, L. Singh, Military potential of biological neurotoxins, Toxin Rev. 24 (2) (2005) 175–207. [7] A.I. Negro, C. de Hoyos, J.C. Vega, Phytoplankton structure and dynamics in Lake Sanabria and Valparaíso reservoir (NW Spain), Hydrobiologia 424 (2000) 25–37. [8] G.E. Fogg, W.D.P. Stewart, P. Fay, A.E. Walsby, The Blue-green Algae, London, Academic Press, 1973. [9] C. de Hoyos, A. Negro, J.J. Aldasoro, Cyanobacteria distribution and abundance in the Spanish water reservoirs during thermal stratiﬁcation, Limnetica 23 (1-2) (2004) 119–132. [10] J.H. Friedman, Multivariate adaptive regression splines (with discussion), Ann. Stat. 19 (1991) 1–141. [11] T. Hastie, R. Tibshirani, J. Friedman, The Elements of Statistical Learning, Springer-Verlag, New York, 2003. [12] J.H. Friedman, C.B. Roosen, An introduction to multivariate adaptive regression splines, Stat. Methods Med. Res. 4 (1995) 197–217. [13] S.S. Sekulic, B.R. Kowalski, MARS: a tutorial, J. Chemom. 6 (1992) 199–216. [14] C. Pérez-Martínez, P. Sánchez-Castillo, Temporal occurrence of Ceratium hirundinella in spanish reservoirs, Hydrobiologia 452 (1–3) (2004) 101–107. [15] A. Quesada, D. Sanchis, D. Carrasco, Cyanobacteria in spanish reservoirs. How frequently are they toxic? Limnetica 23 (1-2) (2004) 109–118. [16] M. Alvarez Cobelas, M. Arauzo, Phytoplankton responses to varying time scales in a eutrophic reservoir, Arch. Hydrobiol. Ergebn. Limnol. 40 (2006) 69–80. [17] C.S. Reynolds, Ecology of Phytoplankton, Cambridge University Press, New York, 2006. [18] I. Chorus, J. Bartram, Toxic Cyanobacteria in Water. A Guide to their Public Health Consequences, Monitoring and Management, Word Health Organization, London, 1999.

Acknowledgements Authors wish to acknowledge the computational support provided by the Departments of Mathematics, Construction and Mining Exploitation at University of Oviedo as well as pollutant data in the Trasona Reservoir of Avilés (Northern Spain) supplied by the Cantabrian Basin Authority (Spanish Ministry of Environment, Rural and Marine Affairs). Furthermore, authors would like to express their gratitude to the Department of Education and Science of the Principality of Asturias for its partial ﬁnancial support (Grant reference FC-11-PC10-19). The English grammar and spelling of the manuscript have been revised by a native person.

P.J. Garcia Nieto et al. / Journal of Hazardous Materials 195 (2011) 414–421 [19] World Health Organization, Guidelines for Drinking-water Quality: Health Criteria and Other Supporting Information, vol. 2, World Health Organization, Geneva, 1998. [20] American Public Health Association, American Water Works Association, Water Environment Federation, Standard Methods for the Examination of Water and Wastewater, no. 20, APHA/AWWA/WEF, Washington, 1998. [21] C. Brönmark, L.-A. Hansson, The Biology of Lakes and Ponds, Oxford University Press, New York, 2005. [22] M. Scheffer, Ecology of Shallow Lakes, Springer, New York, 2005. [23] A.G. van der Valk, The Biology of Freshwater Wetlands, Oxford University Press, New York, 2006. [24] Q.S. Xu, M. Daszykowski, B. Walczak, F. Daeyaert, M.R. de Jonge, J. Heeres, L.M.H. Koymans, P.J. Lewi, H.M. Vinkers, P.A. Janssen, D.L. Massart, Multivariate adaptive regression splines—studies of HIV reverse transcriptase inhibitors, Chemom. Intell. Lab. 72 (1) (2004) 27–34. [25] F.J. De Cos Juez, F. Sánchez Lasheras, P.J. García Nieto, M.A. Suárez Suárez, A new data mining methodology applied to the modelling of the inﬂuence of diet and lifestyle on the value of bone mineral density in post-menopausal women, Int. J. Comput. Math. 86 (10) (2009) 1878–1887. [26] D. Guzmán, F.J. de Cos Juez, R. Myers, A. Guesalaga, F. Sánchez Lasheras, Modeling a MEMS deformable mirror using non-parametric estimation techniques, Opt. Express 18 (20) (2010) 21356–21369. [27] J. De Andrés, P. Lorca, F.J. de Cos Juez, F. Sánchez-Lasheras, Bankruptcy forecasting: a hybrid approach using Fuzzy c-means clustering and multivariate adaptive regression splines (MARS), Expert Syst. Appl. 38 (3) (2011) 1866–1875. [28] V. Vapnik, The Nature of Statistical Learning Theory, Springer, New York, 1995. [29] V. Vapnik, Statistical Learning Theory, Wiley-Interscience, New York, 1998.

421

[30] F.J. de Cos Juez, P.J. García Nieto, J. Martínez Torres, J. Taboada, Analysis of lead times of metallic components in the aerospace industry through a supported vector machine model, Math. Comput. Model. 52 (7–8) (2010) 1177–1184. [31] J.R. Leathwick, J. Elith, T. Hastie, Comparative performance of generalized additive models and multivariate adaptive regression splines for statistical modelling of species distributions, Ecol. Model. 199 (2006) 188–196. [32] D. Guzmán, F.J. de Cos Juez, F. Sánchez Lasheras, R. Myers, L. Young, Deformable mirror model for open-loop adaptive optics using multivariate adaptive regression splines, Opt. Express 18 (2010) 6492–6505. [33] P.M. Gault, H.J. Marler, Handbook on Cyanobacteria: Biochemistry, Biotechnology and Applications, Nova Science Publishers, New York, 2009. [34] B.A. Whitton, M. Potts, The Ecology of Cyanobacteria: Their Diversity in Time and Space, Springer, New York, 2000. [35] J. Huisman, H.C.P. Matthijs, P.M. Visser, Harmful Cyanobacteria, Springer, New York, 2010. [36] E.S. Allman, J.A. Rhodes, Mathematical Models in Biology: An Introduction, Cambridge University Press, New York, 2003. [37] D.J. Barnes, D. Chu, Introduction to Modeling for Biosciences, Springer, New York, 2010. [38] R. Willame, T. Jurckzak, J.F. Ifﬂy, T. Kull, J. Meriluoto, L. Hoffman, Distribution of hepatotoxic cyanobacterial blooms in Belgium and Luxembourg, Hydrobiologia 551 (2005) 99–117. [39] I. Chorus, J. Bartram, Toxic Cyanobacteria in Water: A Guide to Their Public Health Consequences, Monitoring and Management, Spon Press, New York, 1999. [40] J. Seckbach, Algae and Cyanobacteria in Extreme Environments, Springer, New York, 2007.

Journal of Hazardous Materials 195 (2011) 422–431

Contents lists available at SciVerse ScienceDirect

Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat

Time of ﬂight secondary ion mass spectrometry and high-resolution transmission electron microscopy/energy dispersive spectroscopy: A preliminary study of the distribution of Cu2+ and Cu2+ /Pb2+ on a Bt horizon surfaces B. Cerqueira a , F.A. Vega a , C. Serra b , L.F.O. Silva c , M.L. Andrade a,∗ a

Department of Plant Biology and Soil Science, Faculty of Biology, University of Vigo, Lagoas, Marcosende, 36310 Vigo, Pontevedra, Spain CACTI, University of Vigo, Campus Lagoas-Marcosende, 36310 Vigo, Spain c Environmental Science and Nanotechnology Department, Catarinense Institute of Environmental Research and Human Development – IPADHC, Capivari de Baixo, Santa Catarina, Brazil b

a r t i c l e

i n f o

Article history: Received 13 June 2011 Received in revised form 1 August 2011 Accepted 22 August 2011 Available online 26 August 2011 Keywords: Soil Heavy metal Time-of-ﬂight secondary ion mass spectrometry (TOF-SIMS) Field emission scanning electron microscopy (FE-SEM) Energy-dispersive X-ray spectrometer (EDS)

a b s t r a c t Relatively new techniques can help in determining the occurrence of mineral species and the distribution of contaminants on soil surfaces such as natural minerals and organic matter. The Bt horizon from an Endoleptic Luvisol was chosen because of its well-known sorption capability. The samples were contaminated with Cu2+ and/or Pb2+ and both sorption and desorption experiments were performed. The preferential distribution of the contaminant species (63 Cu and 208 Pb) to the main soil components and their associations were studied together with the effectiveness of the surface sorption and desorption processes. The results obtained were compared with non-contaminated samples as well as with previous results obtained by different analytical techniques and advanced statistical analysis. Pb2+ competes favorably for the sorption sites in this soil, mainly in oxides and the clay fraction. Cu2+ and Pb2+ were mainly associated with hematite, gibbsite, vermiculite and chlorite. This study will serve as a basis for further scientiﬁc research on the soil retention of heavy metals. New techniques such as spectroscopic imaging and transmission electron microscopy make it possible to check which soil components retain heavy metals, thereby contributing to propose effective measures for the remediation of contaminated soil. © 2011 Elsevier B.V. All rights reserved.

1. Introduction The bioavailability of heavy metals and other soil pollutants and the risk that they reach surface or underground waters mainly depend on the sorption and desorption capacity of soil components. The term “sorption” is used to encompass adsorption, precipitation on soil particle surfaces, and ﬁxation and “desorption” for the release of sorbed species to the surrounding environment [1–3]. The sorption and desorption of cations predominantly involve negatively charged soil surfaces such as organic matter, clays and metallic oxides or hydroxides [3]. The distribution of metals among the soil components depends on the intrinsic properties of the metal species involved, the properties of the soil, and the amounts of metal added [4,5]. The immobilization of copper and lead in soils occur as a result of several processes, such as adsorption, chemisorption, ion exchange,

∗ Corresponding author. Tel.: +34 986 812630; fax: +34 986 812556. E-mail address: [email protected] (M.L. Andrade). 0304-3894/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.jhazmat.2011.08.059

or surface precipitation. Soil organic matter is an important sorbent for lead and, to a lesser extent, for copper [6]. Both ions are also bound by speciﬁc adsorption and precipitation in calcareous soils on calcite surfaces [7,8]. Soil iron minerals and hydrous iron and manganese oxide surfaces also play an important role in copper and lead sorption [9–11]. Clays such as smectite and vermiculite are well known as important sorbents of copper and lead. A large number of articles refer to heavy metal sorption on soils and soil components, and the experiments involved included techniques such as chemical extractions, batch and soil column sorption experiments [11–14]. Few of the papers that have been published have focused on advanced electron microscopy techniques to study the association of metals to different soil particles [15]. In previous publications [3,16–19] several soils were characterized and sorption and desorption isotherms were determined in order to evaluate both monometal and competitive sorption capacities. The inﬂuence of soil properties was established by means of advanced statistical analyses [19]. The analyses revealed that the soils with the highest pH, effective cation exchange capacities

B. Cerqueira et al. / Journal of Hazardous Materials 195 (2011) 422–431

(CECe) and Mn and Fe oxide contents, as well as those with the greatest mineralogical variety in the clay fraction, have the highest capacity for the individual and competitive sorption of Cu2+ and Pb2+ . The sorption of Cu2+ and Pb2+ appears to take place almost exclusively through cation exchange [19]. Since the surface chemistry exerts a major inﬂuence on chemical species and their eventual fate, the characterization of different chemicals on mineral and organic surfaces of natural samples is an important topic. The soil surface chemistry can vary considerably on a microscopic scale, and this heterogeneity or variability can lead to misinterpretation, introducing confusion in the characterization process. Relatively new spectroscopic imaging techniques can help in determining the distribution of contaminants on soil surfaces such as natural minerals and organic matter. The secondary ion mass spectrometry (SIMS) technique was developed about 60 years ago [20]. Later, the idea was to provide spatially resolved information with SIMS and that it should be possible to build an ion-optical collection system, analogous to a lens used in the light microscope, preserving the spatial relationship of the desorbed ions, as reviewed by Benninghoven et al. [21] and Benninghoven [22]. The principle of TOF (time of ﬂight) coupled with SIMS is based on the fact that ions with different masses travel at different speeds. Basically, desorbed secondary ions from the target surface are accelerated by an electrostatic ﬁeld and they travel to the detector. The TOF-SIMS technique makes it possible to determine the presence of a wide range of air, soil and water pollutants. Owing to an application of TOF-SIMS, processes involving adsorption and the migration of toxic substances in different environments can be observed, as well as pollution mechanisms. Generally, TOF-SIMS experiments concentrate on measuring the surface composition of analyzed materials and the distribution of particular components on the surface of the investigated samples [23]. This technique is one of the most surface-sensitive analytical techniques and has several advantages over alternative imaging methods, such as its sensitivity to all elements, detection of their isotopes, fast data acquisition due to a parallel detection system, very low sampling depth (1–2 nm), sensitivity in the part per million (ppm) to part per billion (ppb) in most of the species range, and the ability to successfully analyze insulators. Also, the occurrence of mineral species can be investigated by means of X-ray diffraction (XRD), environmental ﬁeld emission scanning electron microscopy (FE-SEM) with energy-dispersive X-ray spectrometer (EDS) capabilities for chemical analyses of individual particles, and high-resolution transmission electron microscope (HR-TEM) with selected area electron diffraction (SAED) and/or microbeam diffraction (MBD), and scanning transmission electron microscopy (STEM). The aim of this study is to detect Cu2+ and Pb2+ in soil surfaces applying TOF-SIMS and HR-TEM/EDS techniques, comparing noncontaminated and contaminated samples (treated with one or both ions, and after sorption and desorption experiments). The results obtained are compared with previous results obtained by different analytical techniques [3,18]. The preferential distribution of contaminant species (63 Cu and 208 Pb) at the main soil components and their associations were studied, together with the effectiveness of the surface sorption and desorption processes.

2. Material and methods 2.1. Soil samples The Bt horizon from an Endoleptic Luvisol (EL) [24] was chosen to perform this work because of its well-known high sorption and retention capacities and soil properties [3]. Statistical techniques

423

Table 1 Soil characteristics. pH Organic carbon (g kg−1 )

6.40 5.82

Effective cation exchange capacity and exchangeable cation content (cmol(+) kg−1 )

CECe Na K Ca Mg Al

62.16 0.69 0.22 2.70 58.35 0.20

Oxide content (g kg−1 )

Al2 O3 Fe2 O3 MnO

3.25 30.07 0.36

Particle size distribution (%)

Sand Silt Clay

21.30 12.60 66.10

Clay content (% of clay fraction)

Vermiculite Gibbsite Chlorite Kaolinite

60 12 20 8

Na2 O MgO Al2 O3 SiO2 P2 O5 K2 O CaO TiO2 Fe2 O3

78 0.7 16.6 7.1 37.3 ul 0.3 3.9 0.3 18.7

Speciﬁc surface area (m2 g−1 ) Major elements (total content, %)

Trace elements (total content, mg kg−1 )

V Cr Mn Co As Cd Ni Cu Zn Rb Sr Ba Pb Cl

87 331 1710 125 ul 1 233 47 29 14 84 96 24 ul

ul, undetectable level.

such as regression and correlation analysis were used to assess the inﬂuence of their components and properties on the sorption and desorption of monometal (Cu2+ ) and competitive (Cu2+ and Pb2+ ) species [18]. Sorption and retention capacities mainly depend on pH, CECe, and organic matter, Mn and Fe oxides and clay contents. Vermiculite and chlorite are the clay minerals that most inﬂuence these ﬁxation capacities. Table 1 shows the main soil horizon properties of Bt.EL. 2.2. Sample treatment Sorption solutions of a single-metal solution (3 mM Cu(NO3 )2 ) and a bi-metal solution (3 mM Cu(NO3 )2 and 3 mM Pb(NO3 )2 ) with 0.01 M NaNO3 as background electrolyte were prepared. Next, 6 g of soil and 100 ml of each sorption solution were placed in polyethylene jars and shaken in a rotary shaker for 24 h at 25 ◦ C, then centrifuged at 5000 rpm and ﬁltered through Whatman 42 paper. The resulting ﬁltrate was analyzed by inductively coupled plasma optical emission spectrometry (ICP-OES) in a Perkin–Elmer Optima 4300 DV device. The amount of each metal that had been sorbed was calculated from the difference between its concentrations in solution before the addition of soil and after equilibration (sorbed metal = concentration of the added solution − concentration in solution at equilibrium).

424

B. Cerqueira et al. / Journal of Hazardous Materials 195 (2011) 422–431

Fig. 1. Untreated soil (CS sample). (A) TOF-SIMS image of Al, C2 H3 , Si, Fe and Mn, and overlapping all the signals with 63 Cu and 208 Pb, there is no signal of the last two. (B) Image of overlapping Si and Al showing the concordance between both signals. (C) Image of overlapping Fe, Si and Mn showing the concordance between all signals.

The desorption experiments were performed using the pellets obtained after the sorption stage. They were dried at 45 ◦ C and weighed to check if any of the soil sample had been lost during the sorption ﬁltration stage. 100 ml of 0.01 M NaNO3 solution were added to each sample and then shaken for 24 h. The following procedure is the same as in the sorption experiments. The amount of metal retained after the sorption experiments was calculated by subtracting the concentration of the metal in solution following desorption from the previously sorbed amount.

2.3. Analyzing untreated soil samples and samples after sorption and desorption of Cu2+ and Cu2+ + Pb2+ by TOF-SIMS Time of ﬂight secondary ion mass spectrometry (TOF-SIMS IV instrument from Ion-TOF GmbH of Münster, Germany) was used to investigate the elemental and molecular structure of the samples and to obtain a clearer understanding of the chemical composition [25], location and relative abundance of the species present at the surface of the soil sample.

B. Cerqueira et al. / Journal of Hazardous Materials 195 (2011) 422–431

425

Fig. 2. Soil treated with Cu2+ and Pb2+ after sorption process (CS sample). TOF-SIMS images of overlapping all the signals with 63 Cu and 208 Pb showing the concordance.

The TOF analyzer separates the ions according to the time they take to travel through the length of the ﬁeld-free ﬂight-tube. This time interval is related to the mass and charge of the accelerated particles. The energy and angular dispersion of the secondary ions can be compensated using focusing elements such as a reﬂectron. The lighter secondary ions arrive before the heavier ones whereby a mass spectrum can be recorded. TOF-SIMS works by focusing and scanning a narrow pulsed ion beam on the surface. This process leads to the emission of charged secondary ions in a sputtering process from the outermost surface of the sample. Further analysis of the secondary ions provides information on the molecular and elemental species and their isotopes present on the surface. The secondary ions collected and represented in the mass spectra can be attributed to complete molecules, large fragments of molecules that have only lost functional groups. TOF-SIMS analysis was performed with Bt.EL untreated soil samples and those samples obtained after sorption and desorption experiments. During the TOF-SIMS experiment the corresponding sample was bombarded with a pulsed bismuth ion beam. The secondary ions generated were extracted with a 10 kV voltage and their time of ﬂight from the sample to the detector was measured in a reﬂection mass spectrometer. The analysis conditions were: 25 keV pulsed Bi3+ beam at 45◦ incidence, rastered over 304 ␮m × 304 ␮m; at a square-pixel density of 256 × 256, and 50 accumulative scans in each analyzed area. The operating pressure in the main chamber was 5 × 10-10 mbar. An Electron Flood Gun (low energy electrons) was used to compensate the surface charge build-up process during the experiment. Positive secondary ion mass spectra were acquired over a mass range from m/z = 0 to m/z = 1000. Negative ion TOF-SIMS spectra were not

considered in this study. The mass resolution (m/m) of the secondary ion peaks in the positive spectra was typically between 3600 and 6000. Before further analysis the positive spectra were calibrated using CH3 + , C2 H3 + , C3 H5 + , and C7 H7 + ions. To obtain two-dimensional imaging (chemical surface maps), polyatomic bismuth projectiles (Bi3+ ) were focused onto the surface in a rastered mode. The detected intensities for secondary ion signals were color-coded according to a color scale. The chemical maps produced by TOF-SIMS represent the ions that reached the detector rather than the ions that were present on the surface. Because each solid has its own ability to release ions, the intensities cannot be used to derive absolute surface concentrations. However, the chemical maps are very useful for indicating the relative surface abundance and how it changes as a function of time or sample treatment. The studied ions (Table 2) are Al, Fe, Mn, Si and C3 H5 + (as representative elements of main soil components) and 63 Cu and 208 Pb (as soil pollutants).

Table 2 Ions represented in the maps. Ion

Centre mass

Resolution

Al Si C3 H5 + Mn Fe 63 Cu 208 Pb

26.9798 27.9723 41.0389 54.9364 55.9299 62.9277 207.9558

4204 4232 3633 3687 6024 4168 5527

426

B. Cerqueira et al. / Journal of Hazardous Materials 195 (2011) 422–431

Fig. 3. Soil treated with Cu2+ and Pb2+ after desorption process (CD sample). TOF-SIMS images of overlapping all the signals with 63 Cu and 208 Pb showing the concordance.

2.4. Soil analysis by advanced electronic beam The occurrence of mineral and nanoparticles species was investigated by means of XRD, FE-SEM with EDS and HR-TEM with SAED or MBD, and STEM. Analysis of XRD was used to determine crystalline phases in the studied samples (Siemens model D5005 X-ray diffractometer). The samples were ground by hand in a ceramic mortar and pestle, dry mounted in aluminum holders, and scanned at 8–6◦ 2 with Cu K-␣ radiation [26,27]. The morphology, structural distribution, and particle chemical composition of samples containing ultraﬁne particles and minerals (crystalline and/or amorphous) were examined using a Zeiss Model ULTRA plus FE-SEM (with charge compensation for all applications on conductive as well as non-conductive samples) and a 200 kV Zeiss-LIBRA® 200FE HR-TEM (information limit < 0.08 nm at 200 kV) equipped with high efﬁcient ﬁeld emission cathode and energy omega-ﬁlter for high-accuracy measurements of structure and atomic composition of nano-sized minerals at ultimate resolution. The precession interface module with HR-TEM makes it possible to work at a maximum precession angle of up to 5◦ , and saves different alignments for different precession electron diffraction (PED) angles in memory; the module can correct on-line spot size aberration effects caused by high precession angles. Structure determination of nanominerals with PED has been proved to be highly successful in TEM for a variety of nanocrystals ranging from oxides, sulphates, and other minerals [28–31]. The FE-SEM was equipped with an EDS and the mineral identiﬁcations were made on the basis of morphology and grain composition using both secondary electron and back-scattered electron modes [32,33]. Geometrical aberrations were measured by HR-TEM and

controlled to provide less than a /4 phase shift of the incoming electron wave over the probe-deﬁning aperture of 14.5 mrad [34,35]. EDS spectra were recorded in FE-SEM images mode and then quantiﬁed using ES Vision software that uses the thin-foil method to convert X-ray counts of each element into atomic or weight percentages [36]. Electron diffraction patterns of the crystalline phases were recorded in SAED or MBD mode, and the d spacings were compared to the International Center for Diffraction Data [37] inorganic compound powder diffraction ﬁle (PDF) database to identify the crystalline phases. The soil samples were analyzed after the desorption process due to the importance of knowing the distribution between soil particles, especially in the case of high sorption hysteresis such as that found in the soil studied for both metals [17,18]. 2.5. Statistical analyses All analyses were performed in triplicate and all statistical calculations were performed using SPSS for Windows, version 14.0. 3. Results and discussion The sorbed and retained concentrations of Cu2+ and Pb2+ in each treatment are shown in Table 3. This soil has a greater afﬁnity for Pb2+ than for Cu2+ because Pb2+ competes favorably for sorption sites on soil components [3]. This is because the amount of Pb2+ (both sorbed and retained after sorption) is higher than the amount of Cu2+ when they are added together. Iron and manganese oxides and clay contents (particularly vermiculite and chlorite) are well correlated with the amounts of sorbed and retained ions after using statistical analysis [19]. Cu2+ and Pb2+ sorption was almost irre-

B. Cerqueira et al. / Journal of Hazardous Materials 195 (2011) 422–431

427

Fig. 4. Soil treated with Cu2+ after sorption process (MS sample). TOF-SIMS images of overlapping all the signals with 63 Cu showing the concordance.

versible (Table 3), which indicates the high ﬁxation capacity of this soil horizon. 3.1. TOF-SIMS experiments The TOF-SIMS technique makes it possible to determine the spatial ion distribution of contaminant species (63 Cu and 208 Pb) and to verify the association of Cu2+ and Pb2+ to soil components.

The results from soil samples, before and after sorption experiments (uncontaminated soil, and soil contaminated with one or both cations after sorption and desorption processes) are discussed below. Figs. 1–5 show the map distribution of the surface ions from TOF-SIMS. These images show the lateral distribution of speciﬁc chemical species on the sample surface, within a randomly chosen area (304 ␮m × 304 ␮m) on the sample.On the left of each image is

Table 3 Sample name and Cu2+ and Pb2+ soil content. Studied soil

Bt horizon from Endoleptic Luvisol developed over serpentinized amphibolite

Treatment, samples, nomenclature and Cu2+ and Pb2+ sorbed and retained Untreated sample (US) Monometallic (Cu2+ ) sorption (MS) Sorbed Cu2+ : 47.55 ␮mol g−1 Monometallic (Cu2+ ) desorption (MD) Retained Cu2+ : 46.77 ␮mol g−1 Competitive (Cu2+ + Pb2+ ) sorption (CS) Sorbed Cu2+ : 42.21 ␮mol g−1 Competitive (Cu2+ + Pb2+ ) desorption (CD) Retained Cu2+ : 40.04 ␮mol g−1 )

Sorbed Pb2+ : 43.33 ␮mol g−1 Retained Pb2+ : 42.32 ␮mol g−1

428

B. Cerqueira et al. / Journal of Hazardous Materials 195 (2011) 422–431

Fig. 5. Soil treated with Cu2+ after desorption process (MD sample). TOF-SIMS images of overlapping all the signals with 63 Cu showing the concordance.

the intensity bar (the signal intensity of each ion). The dark color means no signal; in these points the detection limit for this ion was reached. The lightest color represents the maximum signal intensity and shows the points of maximum ion abundance. The corresponding ion is indicated in the bottom left-hand side of each image. Below in the graph is the TC (number of total counts of each ion). By overlaying the three images with red, green and blue colors, it is possible to identify matching distributions of ions on the sample surface. In the overlaid image, areas can appear with a single color (red, green or blue) but also areas with different colors. These areas with colors other than red, green or blue are those where more than one ion is located. These new colors are the product of mixing the ﬁrst. Thus, red + green = orange; red + blue = purple and green + blue = yellow. If the three main colors (red, green and blue) occur at the same point, the ﬁnal result is a white region or point. Fig. 1 shows the results from untreated soil. There is no signal for Cu2+ and Pb2+ , but when soils are treated with one or both ions (Figs. 2–5) the signal is clear and more intense before the desorption process (Figs. 2 and 3). The maps also show that 63 Cu and 208 Pb (when possible) have the same distribution in terms of spatial intensity, and they also show good concordance with the distribution of iron (56 Fe) and manganese (55 Mn). There is a slightly poorer relation between the spatial intensity distribution of both 63 Cu and

208 Pb and that of silicon and aluminum ions. The association of Pb2+

with C3 H5+ is not very intense, but is higher than that of Cu2+ . TOF-SIMS illustrates the coincidence of Al-, Fe-, Mn-, Si- and Cu2+ and Pb2+ bearing areas (Figs. 2–5). This can be explained due to an association of clay and oxides, but does not provide any additional information on the dominance of any single Cu2+ or Pb2+ association. On the other hand, the desorption process in both types of treatments leads to a reduction in the extent to which the soil is contaminated. Nevertheless, there are differences in the effectiveness of the desorption process between both types of treatments. For example, in the monometal sorption (MS) sample, the effectiveness of copper desorption is lower than in the competitive sorption (CS) sample, because there is no competition with Pb2+ and Cu2+ is more strongly retained in MS. The close relationship found between the metals and soil components that were studied (particularly iron and manganese oxides, clay and organic matter, the latter especially for Pb2+ ) is in line with previous results [18]. Image analysis by TOF-SIMS is an excellent method for completing and verifying the results of sorption and desorption studies.Fig. 6 shows the spectrum comparison of untreated and treated soil: (I) Pb isotope group: 208 Pb (amu: 208), 207 Pb (amu: 207), 206 Pb (amu: 206), and (II) zone 208 Pb (amu: 208). The TOF-SIMS peaks between the region of mass 205.5 amu and

B. Cerqueira et al. / Journal of Hazardous Materials 195 (2011) 422–431

429

Fig. 8. Soil treated with Pb2+ : EDS spectrum of gibbsite containing Pb. Table 4 Comparison of the calculated ratios from normalized intensity of ions between the total counts of Si and the total counts of Cu2+ and Pb2+ (I = ISilicon/IPollutant).

Fig. 6. TOF-SIMS spectra of Pb isotopes of untreated and treated soil (I). Spectra comparison/positive polarity: 208 Pb zone (amu: 208) (II).

Fig. 7. TOF-SIMS spectra of 63 Cu of untreated and treated soil. Spectrum comparison/positive polarity. 63 Cu zone (amu: 63).

Sample

Ratio I 28 Si/I 208 Pb

Ratio I 28 Si/I 63 Cu

US (untreated soil) MS (copper sorbed soil) MD (copper desorbed soil) CS (copper and lead sorbed soil) CD (copper and lead desorbed soil)

229 91.03 220.5 4.05 6.04

75.8 4.11 4.24 3.28 4.95

208.5 amu are shown in Fig. 6. In this region, we detected 208 Pb, 207 Pb and 206 Pb, but the most abundant and representative type in the Pb2+ contaminated samples is 208 Pb. Fig. 7 also shows the comparison of spectra of untreated and treated soil for the zone 63 Cu (amu: 63). Although TOF-SIMS is not a quantitative technique, the ratios between the total counts of silicon (Si, amu = 28) (chosen as element representative of the soil) and the total counts of the contaminant elements (Cu2+ and Pb2+ ) were calculated by comparing the spectra. These ratios (Table 4) were taken as a qualitative reference for the effectiveness of the desorption process. The higher the ratio value in desorbed samples (compared to that from non-contaminated samples), the more effective the desorption process. The ratios (Table 4) are also consistent with the results obtained in previous studies. The advantages of the TOF-SIMS technique in

Fig. 9. EDS soil treated with Cu2+ and Pb2+ : typical gibbsite (containing 6% Cu2+ and 4% Pb2+ ) and kaolinite.

430

B. Cerqueira et al. / Journal of Hazardous Materials 195 (2011) 422–431

Fig. 10. Chlorite containing Cu2+ particles.

verifying the soil ion spatial distribution of heavy metals such as Pb2+ and Cu2+ are conﬁrmed in this study.

3.2. Electron beam microscopy The HR-TEM/EDS/SAED results combined with FE-SEM/EDS (Figs. 8–11) show that gibbsite plays a very important role in the ﬁxation of both metals (Pb2+ and Cu2+ ). These results are in agreement with those of previous studies [3]. It was found that both metals (Cu2+ and Pb2+ ) are sorbed in the gibbsite, whose presence and abundance in the soil signiﬁcantly increases the sorption capacity of both metals [3,17]. It was found that the gibbsite from contaminated soil samples contains 6% Cu2+ and 4% Pb2+ . The samples that were only treated with Pb2+ clearly show the sorption of this ion in the gibbsite (Fig. 8), proving the previously found high correlation [18]. Similar results were observed in the chlorite (Fig. 10). It is also important to note that the association of these metals with hematite has been proved (by HR-TEM/EDS) (Fig. 11). Several studies have shown that iron oxides signiﬁcantly contribute to Cu2+ and Pb2+ sorption [38,39]. Moreover, the results obtained by FE-SEM and HR-TEM analysis revealed that when soil samples were contaminated with both Cu2+ and Pb2+ they have a higher afﬁnity for Pb2+ sorption than for Cu2+ which is in agreement with the results of Harter and Naidu [40].

Fig. 11. Hematite ultraﬁne particle containing Cu2+ and Pb2+ in soil treated with Pb2+ and Cu2+ .

4. Conclusions The soil has higher afﬁnity for Pb2+ than for Cu2+ . The Pb2+ competes favorably for the sorption sites in these soils, mainly in the oxides and clay fraction. Cu2+ and Pb2+ were mainly associated with hematite, gibbsite, vermiculite and chlorite. High-resolution microscopy studies combined with analysis by TOF-SIMS, batch experimental studies and statistical analysis are an effective tool to check the afﬁnity of the soil components for cations, as well as competition between them for sorption sites. This study will serve as a basis for further scientiﬁc research on the soil retention of heavy metals, as high-resolution microscopy makes it possible to check which soil components retain heavy metals (with high toxicity), thereby contributing to propose effective measures for the remediation of contaminated soil. Acknowledgements This research was supported by the Project CGL2010-16765 (Spanish Ministry of Science and Innovation, co-funded with ˜ FEDER). F.A. Vega was awarded an “Ángeles Alvarino” research grant (Xunta de Galicia–University of Vigo). Luis F.O. Silva beneﬁted from a scholarship ﬁnanced by CNPq, Brazil – Ref: 382954/20114 and Processo: 380649/2011-0. We would like to thank Emma F. Covelo for her comments on the ﬁnal version of the manuscript. References [1] D.L. Sparks, A.M. Scheidegger, D.G. Strawn, K.G. Scheckel, Kinetics and mechanisms of metal sorption at the mineral–water interface, in: D.L. Sparks, T.J. Grundl (Eds.), Mineral–Water Interfacial Reactions, ACS Symp., Series 715, Am. Chem. Soc., Washington, DC, 1999, pp. 108–135. [2] R. Apak, Adsorption of heavy metal ions on soil surfaces and similar substances, in: A. Hubbard (Ed.), Encyclopedia of Surface and Colloid Science, Dekker, New York, 2002, pp. 385–417. [3] F.A. Vega, E.F. Covelo, M.L. Andrade, A versatile parameter for comparing the capacities of soils for sorption and retention of heavy metals dumped individually or together: results for cadmium, copper and lead in twenty soil horizons, J. Colloid Interface Sci. 327 (2008) 275–286. [4] F.X. Han, A. Banin, Long-term transformations and redistribution of potentially toxic heavy metals in arid-zone soils. II. Under the ﬁeld capacity regime, Water Air Soil Pollut. 114 (1999) 221–250. [5] F.X. Han, A. Banin, G.B. Triplett, Redistribution of heavy metals in arid-zone soils under a wetting–drying soil moisture regime, Soil Sci. 166 (2001) 18–28. [6] E.F. Covelo, F.A. Vega, M.L. Andrade, Sorption and desorption of Cd, Cr, Cu, Ni, Pb and Zn by a Fibric Histosol and its organo-mineral fraction, J. Hazard. Mater. 159 (2008) 342–347. [7] L.M. Dudley, J.E. McLean, T.H. Furst, J.J. Jurinak, Sorption of cadmium and copper from an acid mine waste extract by two calcareous soils: column studies, Soil Sci. 151 (1991) 121–135. [8] P. Sipos, T. Németh, V. Kovács Kis, I. Mohai, Sorption of copper, zinc and lead on soil mineral phases, Chemosphere 73 (2008) 461–469.

B. Cerqueira et al. / Journal of Hazardous Materials 195 (2011) 422–431 ´ J. Balík, Copper [9] M. Komárek, J. Száková, M. Rohoˇsková, H. Javorská, V. Chrastny, contamination of vineyard soils from small wine producers: a case study from the Czech Republic, Geoderma 147 (2008) 16–22. ´ M. Mihaljeviˇc, Lead isotopes in environmen[10] M. Komárek, V. Ettler, V. Chrastny, tal sciences: a review, Environ. Int. 34 (2008) 562–577. [11] T. Németh, J. Jiménez-Millán, P. Sipos, I. Abad, R. Jiménez-Espinosa, Z. Szalaic, Effect of pedogenic clay minerals on the sorption of copper in a Luvisol B horizon, Geoderma 160 (2011) 509–516. [12] N. Cavallaro, M.B. McBride, Zinc and copper sorption and ﬁxation by an acid soil clay: effect of selective dissolutions, Soil Sci. Soc. Am. J. 48 (1984) 1050–1054. [13] L.M. Shuman, Effect of organic matter on the distribution of manganese, copper, iron, and zinc in soil fractions, Soil Sci. 146 (1988) 192–198. [14] P. Sipos, T. Németh, I. Mohai, I. Dódony, Effect of soil composition on sorption of lead as reﬂected by a study on a natural forest soil proﬁle, Geoderma 124 (2005) 363–374. [15] P. Sipos, T. Németh, V. Kovács Kis, I. Mohai, Association of individual soil mineral constituents and heavy metals as studied by sorption experiments and analytical electron microscopy analyses, J. Hazard. Mater. 168 (2009) 1512–1520. [16] E.F. Covelo, F.A. Vega, M.L. Andrade, Simultaneous sorption and desorption of Cd, Cr, Ni, Cu, Pb and Zn in acid soils. I. Selectivity sequences, J. Hazard. Mater. 147 (3) (2007) 852–861. [17] E.F. Covelo, F.A. Vega, M.L. Andrade, Simultaneous sorption and desorption of Cd, Cr, Ni, Cu, Pb and Zn in acid soils. II. Soil ranking and inﬂuence of soil characteristics, J. Hazard. Mater. 147 (3) (2007) 862–870. [18] B. Cerqueira, E.F. Covelo, L. Andrade, F.A. Vega, The inﬂuence of soil properties on the individual and competitive sorption and desorption of Cu and Cd, Geoderma 162 (2011) 20–26. [19] F.A. Vega, M.L. Andrade, E.F. Covelo, Inﬂuence of soil properties on the sorption and retention of cadmium, copper and lead, separately and together, by 20 soil horizons: comparison of linear regression and tree regression analyses, J. Hazard. Mater. 174 (2010) 522–533. [20] R.F.K. Herzog, F. Viehboeck, Ion source for mass spectrography, Phys. Rev. 76 (1949) 855–856. [21] A. Benninghoven, B. Hagenhoff, E. Niehuis, M.S. Surface, Probing real-world samples, Anal. Chem. 65 (1993) 630–640. [22] A. Benninghoven, Chemical analysis of inorganic and organic surfaces and thin ﬁlms by static time-of-ﬂight secondary ion mass spectrometry (TOF-SIMS), Angew. Chem. Int. Ed. 33 (1994) 1023–1043. [23] J. Grams, New Trends, Potentialities of TOF-SIMS in Surface Studies, Nova Science Publishers Inc., New York, 2007, p. 273. [24] WRBSR, World reference base for soil resources, A framework for international classiﬁcation correlation and communication, IUSS-ISRIC-FAO, World Soil Resources Reports 103, Rome, 2006, 146 pp. [25] M.C. Rodríguez-Argüelles, R. Villalonga, C. Serra, R. Cao, M.A. Sanromán, M.A. Longo, A copper(II) thiosemicarbazone complex built on gold for the immobilization of lipase and laccase, J. Colloid Interface Sci. 348 (2010) 96– 100.

431

[26] J. Ribeiro, D. Flores, C.R. Ward, L.F.S. Silva, Identiﬁcation of nanominerals and nanoparticles in burning coal waste piles from Portugal, Sci. Total Environ. 408 (2010) 6032–6041. [27] L.F.O. Silva, T. Moreno, X. Querol, An introductory TEM study of Fenanominerals within coal ﬂy ash, Sci. Total Environ. 407 (2009) 4972–4974. [28] S. Hayashi, T. Takahashi, K. Kanehashi, N. Kubota, K. Mizuno, S. Kashiwakura, T. Sakamoto, T. Nagasaka, Chemical state of boron in coal ﬂy ash investigated by focused-ion-beam time-of-ﬂight secondary ion mass spectrometry (FIB-TOF-SIMS) and satellite-transition magic angle spinning nuclear magnetic resonance (STMAS NMR), Chemosphere 80 (2010) 881–887. [29] S.B. Abreu, W. Skinner, Predicting the surface chemistry contribution to the ﬂotation recovery of chalcopyrite by TOF-SIMS, Miner. Eng. 24 (2011) 160–168. [30] L.F.O. Silva, J.C. Hower, M. Izquierdo, X. Querol, Complex nanominerals and ultraﬁne particles assemblages in phosphogypsum of the fertilizer industry and implications on human exposure, Sci. Total Environ. 408 (2010) 5117–5122. [31] L.F.O. Silva, C.R. Ward, J.C. Hower, M. Izquierdo, F. Waanders, M.L.O. Oliveira, Z. Li, R.S. Hatch, X. Querol, Mineralogy and leaching characteristics of coal ash from a major Brazilian power plant, Coal Combust. Gasiﬁcation Prod. 2 (2010) 51–65. [32] L.F.O. Silva, X. Querol, K.M. da Boit, S.F.O.D. Vallejuelo, J.M. Madariaga, Brazilian coal mining residues and sulphide oxidation by Fenton’s reaction: an accelerated weathering procedure to evaluate possible environmental impact, J. Hazard. Mater. 186 (2011) 516–525. [33] L.F.O. Silva, M. Wollenschlager, M.L.S. Oliveira, A preliminary study of coal mining drainage and environmental health in the Santa Catarina region, Brazil, Environ. Geochem. Health (N.Y.) 33 (2011) 55–65. [34] L.F.O. Silva, M.L.S. Oliveira, E.R. Neace, J.M.K. O’Keefe, K.R. Henke, J.C. Hower, Nanominerals, ultraﬁne particles in sublimates from the Ruth Mullins coal ﬁre, Perry County, Eastern Kentucky, USA, Int. J. Coal Geol. 85 (2011) 237–245. [35] L.F.O. Silva, F. Macias, M.L.S. Oliveira, K.M. da Boit, F. Waanders, Coal cleaning residues and Fe-minerals implications, Environ. Monit. Assess. 172 (2011) 367–378. [36] L.F.O. Silva, K. DaBoit, Nanominerals and nanoparticles in feed coal and bottom ash: implications for human health effects, Environ. Monit. Assess. 174 (2011) 187–197. [37] ICDD, International Center for Diffraction Data, http://www.icdd.com (accessed 20.11.10). [38] A. Kabata-Pendias, Trace Elements in Soils and Plants, CRC Press, Boca Raton, FL, USA, 2001. [39] A.M. Moreno, J.R. Quintana, L. Pérez, J.G. Parra, Factors inﬂuencing lead sorption–desorption at variable added metal concentrations in Rhodoxeralfs, Chemosphere 64 (5) (2006) 758–763. [40] R.D. Harter, R. Naidu, An assessment of environmental and solution parameter impact on trace-metal sorption by soils, Soil Sci. Soc. Am. J. 65 (2001) 597–612.

Journal of Hazardous Materials 195 (2011) 432–439

Contents lists available at SciVerse ScienceDirect

Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat

Can bubble columns be an alternative to ﬁbrous ﬁlters for nanoparticles collection? A. Charvet a,b,∗ , N. Bardin-Monnier a,b , D. Thomas a,b a b

CNRS, Laboratoire Réactions et Génie des Procédés (LRGP), UPR 3349, 1 rue Grandville BP 20451, 54001 Nancy, France Nancy-Université, UHP, LRGP, 54000 Nancy, France

a r t i c l e

i n f o

Article history: Received 19 July 2011 Received in revised form 24 August 2011 Accepted 25 August 2011 Available online 31 August 2011 Keywords: Bubble column Collection Nanoparticles

a b s t r a c t The most effective and widely used dedusting techniques to separate nanoparticles of a carrier ﬂuid are ﬁbrous media. The main problem is the clogging of the ﬁlter that induces a pressure drop increase over time and thus requires a regular cleaning of the media (or its replacement). Following these observations, this study proposes to investigate the potential of bubble columns for nanoparticles collection. Despite collection efﬁciencies lower than those of ﬁbrous ﬁlters, experimental results show that bubble columns present likely performances for the collection of nanoparticles and have collection efﬁciency even more important when the liquid height is high and bubbling oriﬁces have low diameters. Experiments have also revealed the presence of a most penetrating particle size for a particle diameter range between 10 and 30 nm. The model developed in this article highlights a good agreement between the theoretical collection efﬁciency by Brownian diffusion and experimental collection efﬁciencies for particles lower than 20 nm. Nevertheless, the modelling may be extended to other collection mechanisms in order to explain the collection efﬁciency increase for particles higher than 20 nm and to conﬁrm or inﬁrm that electrostatic effects can be the cause of this efﬁciency increase. © 2011 Elsevier B.V. All rights reserved.

1. Introduction Nanometric aerosols localized in workplace environments mainly stem from two sources: on the one hand, nanoparticles are manufactured in more and more industrial applications and on the other hand, nanoparticles are generated by different processes: fumes, oil mists and metallization. The latter technique, which involves projecting (by compressed air) ﬁne metallic particles on a surface, is a very productive source of ultraﬁne toxic particles. Indeed, the metals most commonly used are zinc, tin and Zn/Al alloys. Coatings based on most dangerous metals as chromium or nickel are also employed [1]. Ultraﬁne particle emissions in the environment are increasingly regulated and the appearance of limit values imposes to act on the particle rejections. Currently, the most effective and widely used dedusting techniques to separate nanoparticles of a carrier ﬂuid are ﬁbrous media (often cartridges). These very high efﬁciency ﬁlters are mainly constituted of micronsized ﬁbres. The main problem with the ﬁltration of such particles is the rapid clogging of the ﬁlter. This clogging induces a fast pressure drop increase over time and thus requires a regular cleaning of

∗ Corresponding author at: CNRS, Laboratoire Réactions et Génie des Procédés (LRGP), UPR 3349, 1 rue Grandville BP 20451, 54001 Nancy, France. Tel.: +33 0 3 83 17 53 33. E-mail address: [email protected] (A. Charvet). 0304-3894/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.jhazmat.2011.08.064

the media or its replacement. Despite their widespread presence, there is still a lack of models able to predict the time evolution of the pressure drop and the lifetime of such ﬁlters and as a consequence these devices are not optimized yet. Moreover, the dust unclogging may be the cause of many problems affecting the performance of ﬁlters like a possible re-suspension of nanoparticles previously collected by the media [2,3]. This operation can also induce a deterioration of the ﬁlter structure which causes leaks and consequently a severe decrease of ﬁltration efﬁciency [4]. This media cleaning may thus run counter to the general objective of process and individual safety. The problem is most pronounced in the case of metallization processes as particle concentrations are very high (109 particles per cubic centimetre [1]). As a guide, the particle concentration of a polluted environment as can be found near dense road trafﬁc is one thousand times less. In addition, the number size distribution of particles emitted by metallization processes exhibits a population of ultraﬁne particles (more than 95% of particles of diameter < 100 nm). Nanoparticles tend to form agglomerates which favour the ﬁlter clogging. Following these observations, this study proposes to investigate the potential of other dedusting methods that could be applied to metallization fumes. The idea is to test wet scrubbers as an alternative to ﬁbrous ﬁlters. One advantage of these separators is that they operate at constant pressure drop and consequently, do not require dust unloading and therefore do not present any risk of nanoparticles re-suspension. This study examines the relevance of wet

A. Charvet et al. / Journal of Hazardous Materials 195 (2011) 432–439

scrubbers in terms of collection efﬁciency and more precisely, the absorption of nanoparticles by bubbling through a liquid will be investigated. 2. Background A survey of the literature revealed that different wet scrubbers, and in particular bubble and spray columns, have been used for collecting particles from gas streams. Bubble columns offer many advantages such as little maintenance requirement due to simple construction, easy temperature control and low initial costs of installation. However, few studies on scrubbing of particles have been reported. Meikap and Biswas [5] studied the performance of a bubble column with a diameter of 0.19 m and a height of 2 m. The gas velocities were selected to generate bubbles in the size range of 2–5 mm (determined by visual observations). The collection efﬁciencies for particle sizes between 2 and 50 ␮m were relatively large (above 97%) and the performance of the wet scrubber increased with the gas ﬂow and the particle concentration entering the column. Moreover, the authors showed that, between 0.2 and 1.3 m, a higher liquid level in the bubble column induced an increase in particle collection. This least observation conﬁrms that of Bandyopadhyay and Biswas [6] who studied the performance of a bubble column for the simultaneous treatment of SO2 and submicron particles (soot and ash with median diameters of 1.4 and 9.8 ␮m, respectively). Yuu et al. [7] studied the absorption of submicron particles (with diameters between 1.5 and 3.2 ␮m) when they are bubbled through water. They showed that the particle capture is almost totally explainable by the mechanism of impaction. They also highlighted that particles are collected with extremely small values of inertia parameter in bubble dust collection compared with ﬁbrous media collection. Nevertheless, bubbling has low collection efﬁciency compared with other methods such as bag ﬁlters. Finally, they emphasized that the collection efﬁciency increased exponentially with the water height and consequently with the residence time. More recently, Hermeling and Weber [8] studied the collection of carbon nanoparticles by bubbling through different liquids. These authors highlighted that an addition of surfactant in the liquid induces a decrease in bubble size and hence an increase in collection efﬁciency of nanoparticles. They also observed that the collection efﬁciency increased exponentially with the liquid level in the bubble column and that an increase in particle size resulted in a decrease in collection efﬁciency (efﬁciency of about 95%, 60% and

433

50% for particles of 18, 55 and 170 nm, respectively). The authors concluded that these results demonstrated the predominance of the diffusion mechanism in the collection of nanoparticles by bubbling through a liquid. The ﬁrst theory of absorption of particles in gas bubbles during their rise through a liquid has been developed by Fuchs [9]. Indeed, particles transported in a stable rising bubble may be collected by the surrounding liquid due to different deposition mechanisms. The most predominant ones are Brownian diffusion, inertial deposition and gravitational settling. In his theory, Fuchs characterizes each mechanism of absorption thanks to a corresponding coefﬁcient and considers that each of them depends on the rising velocity of the bubble and consequently on the bubble size, the inlet airﬂow rate and the bubbling oriﬁce size (the description of Fuchs’ theory, modiﬁed by Pich and Schütz [10], is detailed in Appendix A). From each absorption coefﬁcient, the total collection efﬁciency and the efﬁciency due to each of the single mechanisms can be calculated and plotted for a given bubble diameter (db ) and liquid height (h) and for each particle size (Fig. 1). Because each of these mechanisms is most effective in a given size range, the collection by a rising bubble presents a particle size that leads to a minimum efﬁciency (Most Penetrating Particle Size) in the range 100–500 nm. Brieﬂy, the collection of particles in bubble columns has been studied by several authors but most of these studies have focused on the collection of coarse and micron-sized particles. Finally, only Hermeling and Weber [8], in a short paper, focused on collection of nanoparticles. However, these collection efﬁciencies have been calculated from mean particle diameters and they did not mention the size distribution of generated particles. Consequently, it is difﬁcult to deﬁnitively conclude on the inﬂuence of the particle diameter and on the mechanisms responsible for the collection of nanoparticles in a bubble rising through a liquid. Consequently, results fail to verify the separation efﬁciency of bubble columns towards these ultraﬁne particles. 3. Experimental set-up Our study aims to determine the performance of a bubble column for nanoparticles collection. The test bench, dedicated to the study and presented below (Fig. 2), is divided into 3 main parts: • The generation of nanoparticles is performed using a PALAS® GFG 1000 generator, operated by electric discharges in an argon

Fig. 1. Collection efﬁciency calculated with the Pich’s model (h = 20 cm; db = 4 mm).

434

A. Charvet et al. / Journal of Hazardous Materials 195 (2011) 432–439

Fig. 2. Schematic view of the experimental set-up.

stream. This generation system can produce polydisperse solid particles with diameters ranging between 5 and 150 nm (Fig. 3). • The characterization of the particle size distribution is performed upstream and downstream of the column with either a TSI® Fast Mobility Particle Sizer (FMPS) or a TSI® Scanning Mobility Particle Sizer, which both measure the electrical mobility diameter of nanoparticles. Contrary to a SMPS which associates a Condensation Particle Counter (CPC) with a Differential Mobility Analyzer (DMA), a FMPS spectrometer uses multiple, low-noise electrometers for particle detection and operates at a high ﬂow rate (10 L/min) in order to minimize diffusion losses of ultraﬁne particles. For each particle size, the collection efﬁciency is determined from these concentration measurements upstream and downstream of the bubble column. • The collection of nanoparticles takes place in the bubble column whose main characteristics (liquid height, bubble diameter, air ﬂow at the column inlet) can be controlled. An experimental study was conducted to determine the performance of a bubble column in terms of capture efﬁciency of carbon nanoparticles. Bubble formation occurs thanks to a perforated stainless steel plate (70 mm in diameter) with 12 oriﬁces. The choice of demineralised water as collection ﬂuid was done in order to avoid the formation of secondary solid particles when bubbles

explode at the liquid surface. The collection efﬁciency of nanoparticles has been determined for liquid heights ranging from 5 to 30 cm, for various oriﬁce diameters (330, 440 and 500 ␮m) and for different ﬂows of air laden with nanoparticles (between 1 and 8 L/min). Air ﬂow rates are always expressed at a temperature of 273 K and a pressure of 101,325 Pa. By setting the mass ﬂow controller FM1, it is possible to adjust the particle concentration at the inlet of the bubble column. The opening of the valves V2 and V3 allows to adjust the airﬂow injected into the column. The mass ﬂow controller FM2 allows to know the bypassed ﬂow and thus to determine the ﬂow injected into the column by balance. This airﬂow entering the column can also be measured with the mass ﬂow controller FM4 (V5 and V7 closed). To determine the particle concentration upstream of the bubble column, the valve V4 is opened so that the particle sizer (SMPS or FMPS) can pump a suitable ﬂow. Downstream of the column, the valve V6 is closed and the air is ﬁltered before being released into the room. For the determination of the downstream concentration, the valve V5 is closed and valve V6 is opened. But, when the FMPS granulometer is used, the suction ﬂow is higher than the ﬂow in the column and a previously ﬁltered air must be pumped outside. The ﬂowmeter FM3 allows to determine the dilution of the downstream efﬂuent and therefore going back to its concentration. In order to overcome the eventual collection of nanoparticles on the

A. Charvet et al. / Journal of Hazardous Materials 195 (2011) 432–439

435

Fig. 3. Size distribution of carbon particles generated by the PALAS® GFG 1000.

walls of the column or in the pipes upstream and downstream of the column, the initial concentration, C0 , used to determine collection efﬁciency, is measured downstream of the column, without liquid. It is important to note that visual observations show that the bubbling regime is homogeneous at the 12 oriﬁces for a gas ﬂow rate of approximately 1 L/min. Indeed, the bubbles are almost uniformly distributed and they rise uniformly through the column, without interaction between each other. At higher gas ﬂow rates, the homogeneous gas-in-liquid dispersion cannot be maintained and an unsteady ﬂow pattern with channelling occurs [11]. Instead of single bubbles forming, a jet appears at the oriﬁce and it disintegrates at its top to form a lot of small bubbles [12].

of 1 L/min and liquid heights of 20 and 30 cm. Concentration measurements upstream and downstream of the bubble column were performed with the FMPS spectrometer for perforated plates with 12 oriﬁces of 330 or 440 ␮m (Fig. 4). This oriﬁce diameter appears to inﬂuence the performance of the bubble column. Indeed, small oriﬁces induce the formation of bubbles with smaller diameters as shown in the Gaddis and Vogelpohl relationship below [12]. This correlation is valid for all gas ﬂow rates in the bubbling regime and is considered by Kulkarni and Joshi [13] as the most suitable for the estimation of the bubble size in stagnant liquids.

4. Results and discussion

where db is the bubble diameter (m), do the oriﬁce diameter (m), the surface tension (N/m), the collecting liquid density (kg/m3 ), g the gravity acceleration (m/s2 ), the liquid kinematic viscosity (m2 /s) and Qo the gas ﬂow rate through the bubbling oriﬁce (m3 /s). Thus if the bubble size decreases, the distance between ultraﬁne particles and the gas/liquid interface inside each bubble decreases

The aim of this study is to identify the inﬂuence of some operating parameters (liquid height, oriﬁce diameter). Thus, experiments were conducted to test the inﬂuence of bubbling oriﬁce diameter on the collection efﬁciency of ultraﬁne particles at a gas ﬂow rate

db =

6 · d · 4/3 o ·g

+

81 · · Q 135 · Q 2 4/5 1/4 o o ·g

Fig. 4. Inﬂuence of oriﬁce diameter on collection efﬁciency of a bubble column (Q = 1 L/min).

+

4 · 2 · g

(1)

436

A. Charvet et al. / Journal of Hazardous Materials 195 (2011) 432–439

Fig. 5. Inﬂuence of liquid height on collection efﬁciency of a bubble column (do = 500 ␮m; Q = 5 L/min).

and the particle collection consequently increases. It is important to note that theoretical bubble diameters obtained with the correlation of Gaddis and Vogelpohl (db = 3.86 and 3.93 mm for oriﬁce diameters of 330 and 440 ␮m; respectively) are close to experimental ones determined by visual observations and estimated at 4 mm. Tests were also conducted in order to study the inﬂuence of the liquid height in the bubble column (Fig. 5). For a constant gas ﬂow rate of 5 L/min and constant oriﬁce diameters of 500 ␮m (this corresponds to a bubbling velocity of 35.4 m/s), the four curves obtained for different liquid heights have the same shape and highlight that the collection efﬁciency of nanoparticles increases with the liquid height in the column. This observation can be explained by an increase of the particles residence time in the column and thus an improvement of the particles transfer from the gas to the liquid phase. All the experiments have also revealed the presence of a most penetrating zone for particle diameters between 10 and 30 nm. The efﬁciency decrease between 6 and 15 nm may be due to a decrease in collection efﬁciency by the diffusional mechanism and seems consistent with the modelled results of Pich and Schütz [10]

(Fig. 1). In order to explain the increase in collection efﬁciency for particles higher than 20 nm, the presence of another mechanism has to be studied. It cannot correspond to particle sedimentation as sedimentation becomes effective only for particles higher than 200–500 nm. This efﬁciency increase may be caused by electrostatic effect between charged particles. This phenomenon may contribute to the particle transport towards the gas/liquid interface of the bubble. In order to conﬁrm this hypothesis, tests were conducted with highly positively charged carbon nanoparticles being injected in the bubble column (Fig. 6). The particle charging was performed with a charger which uses corona discharge to produce a high quantity of ions that can transfer their charge to particles passing by the charger. As a result of the charging process, particles will have a stable high positive charge state. This charging of the tested aerosol induces an increase of the collection efﬁciency and conﬁrms that electrostatic effects have a real inﬂuence on the performance of a bubble column. However, this result has to be conﬁrmed by other experiments with particles of perfectly known charge. Figs. 5 and 6 also reveal an efﬁciency decrease for a particle range between 40 and 70 nm. For sake of readability, we did not represent

Fig. 6. Inﬂuence of particle charging on collection efﬁciency of a bubble column (h = 20 cm; do = 500 ␮m; Q = 3 L/min).

A. Charvet et al. / Journal of Hazardous Materials 195 (2011) 432–439

437

Fig. 7. Inﬂuence of the supplied power on the total collection efﬁciency in a bubble column.

error bars on the graphs, but for these particle sizes, collection efﬁciencies are subject to considerable uncertainties. Indeed, the number of particles in this size range is very low and consequently upstream and downstream measurements by the Scanning Mobility Particle Sizer are less reliable. Thus, the current results do not allow to conclude on the particle collection efﬁciency in a size range between 40 and 70 nm. Additional experiments with less ﬁne aerosols may be performed in order to remove these uncertainties. Calculations of the total collection efﬁciency (in number) in other operating conditions conﬁrm conclusions on the inﬂuence of liquid height (Table 1). These data also highlight an inﬂuence of the ﬂow of air laden with nanoparticles and this inﬂuence seems to depend on the bubbling regime. In a jetting regime (between 3 and 8 L/min), the total collection efﬁciency seems to increase with the airﬂow despite a decrease of the residence time of the air in the bubble column. Visualizations of images acquired with a fast readout camera show that a high airﬂow rate induces the formation and the re-entrainment of micro-bubbles in the bottom of the column. This size decrease of some bubbles and the increase of their residence time (by the re-entrainment) may explain the increase of collection efﬁciency at high airﬂows. Pressure drop measurements were carried out during the various tests. These values are higher than those typically generated by a ﬁbrous media but are not excessive and will not increase over time as that of a ﬁbrous ﬁlter during its clogging. From the data of pressure drop and using Eq. (2), we calculated the power supplied during each test and compared this parameter with the total efﬁciency for each operating condition (Fig. 7). P = Q · P

(2)

where P is supplied power (W), Q the gas ﬂow rate entering the bubble column (m3 /s) and P the pressure drop between the entry and the output of the bubble column (Pa).

Air ﬂow rate 1 L/min

Liquid height

5 cm 10 cm 20 cm 30 cm

0.312 0.524 0.683 0.799

3 L/min 0.135 0.362 0.622 0.719

5 L/min 0.170 0.333 0.606 0.750

5. Collection modelling in diffusional regime Fig. 8 highlights that the bubble size greatly affects the theoretical collection efﬁciency. Despite the order of magnitude seems to be accurate, Pich’s model underestimates experimental collection efﬁciencies in diffusional regime (for a bubble diameter of 4 mm which corresponds to experimental bubble diameter). Moreover, for particles smaller than 15 nm, the shape of the theoretical collection efﬁciency curve does not ﬁt experimental results. Consequently, we propose a different approach from Pich’s one (described previously) in order to model the particle collection by Brownian diffusion inside bubbles. 5.1. Hypotheses • The particle concentration and the particle size distribution in each bubble are considered to be similar to measurements upstream of the column. • Bubbles are supposed to be spherical. • Each particle size is supposed to be homogeneously distributed in each bubble. 5.2. Calculation Knowing the fractional particle concentration (Cp,j ) in each bubble of diameter db,i , it is possible to determine the number of particles (of each diameter dp,j ) initially present inside a bubble.

Table 1 Total collection efﬁciency for various operating conditions (do = 500 ␮m). Total efﬁciency (in number)

Fig. 7 shows an efﬁciency minimum for low supplied powers due to the modiﬁcation of the bubbling regime. In a jetting regime (between 3 and 8 L/min), the collection efﬁciency of nanoparticles increases with the power supplied to the system. In the development of a new method for nanoparticles separation, energy aspects and collection efﬁciency must be taken into account. Consequently, the best operating conditions seem to be low airﬂow rates and high water levels.

8 L/min 0.195 0.690 0.982 –

Np,j,i,0 =

3 · db,i

6

· Cp,j

(3)

By calculating the Brownian diffusion coefﬁcient for each particle size DB,j , the root-mean-square net displacement (lj ) of a particle

438

A. Charvet et al. / Journal of Hazardous Materials 195 (2011) 432–439

Fig. 8. Comparison between experimental data and Pich’s model results.

(inside a bubble) during a time ts,b (which corresponds to the bubble residence time in the bubble column) can be evaluated. lj =

2 · DB,j · ts,b

(4) D =

where DB,j =

ts,b =

kB · T · Cu 3 · · G · dp,j

(6)

Ub

db,i − lj 2

Np,j,i,0

(8)

5.3. Comparison between experimental and theoretical results

hliq

4 = ·· 3

Np,j,i,0 − Np,j,i,f

(5)

where kB is the Boltzmann’s constant (J/K), T the gas temperature (K), Cu the Cunningham correction factor, G the dynamic viscosity of the gas (Pa s), hliq the liquid height (m) and Ub the bubble rise velocity calculated with Eq. (A.7) (m/s). We assume that if the Brownian displacement of a particle during the bubble time residence is higher than the bubble radius, the considered particle is collected by the liquid. The particles remaining (Np,j,i,f ) in the bubble at the end of the time ts,b are those which were initially at a distance from the interface higher than lj , i.e. in the sphere of radius (rb,i − lj ). Np,j,i,f

Thus we are able to determine the collection efﬁciency by Brownian diffusion:

3 · Cp,j

(7)

Results obtained with our modelling in the diffusional regime (for different bubble sizes) were compared with our experimental data for liquid heights of 20 and 30 cm (Fig. 9). Our theoretical and experimental collection efﬁciencies are in good agreement in the diffusional regime (for particles ﬁner than 15 nm) and for bubble size of 4 mm which is close to our visual observations. The curves shape of the present model is especially closer to experimental data than with Pich’s model. Nevertheless, the modelling of the collection efﬁciency by Brownian diffusion may be improved by a more precise determination of the experimental bubble size and by taking into account the imperfect spherical shape of the bubbles. Moreover, the modelling may be extended to other collection mechanisms (and not only Brownian diffusion) in order to conﬁrm or inﬁrm electrostatic effects and to explain the collection efﬁciency increase for particles higher than 20 nm.

Fig. 9. Comparison between experimental data and present model results.

A. Charvet et al. / Journal of Hazardous Materials 195 (2011) 432–439

6. Conclusion Despite collection efﬁciencies lower than those of ﬁbrous ﬁlters, the results of this exploratory study show that the bubble columns present likely performances. An optimization of operating parameters (bubble diameter, liquid height, gas ﬂow rate, etc.) could soon bring this technique to become a viable alternative to ﬁbrous media in terms of collection efﬁciency and energy consumption. Other experiments, with particles previously neutralized or charged may also be performed in order to conﬁrm electrostatic effects, the inﬂuence of the initial particle charge and the inﬂuence of the interactions between charged nanoparticles on collection efﬁciency. A more complete numerical analysis must be also developed in order to highlight the inﬂuence of the different collection mechanisms in a bubble column and to verify assumptions based on experimental results. Finally, the development of a bubble column capable of treating several hundred cubic meters per hour is also being studied in order to verify the relevance (in terms of energy expenditure and collection efﬁciency) of this separation process on an industrial scale. Acknowledgments

Appendix A. Appendix A

(A.1)

where Cp is the particle concentration inside the bubble, h the liquid height in the bubble column and a the absorption coefﬁcient of particles. After integration, the previous equation becomes: Cp = Cp,0 · e−a·h

Cp = 1 − e−a·h Cp,0

(A.3)

Consequently, the particle collection exclusively depends on the absorption coefﬁcient which is function of the intensity of the individual deposition mechanisms. Fuchs [9] proposed three different absorption coefﬁcients for three deposition mechanisms: sedimentation, inertia and diffusion respectively: aS = aI =

3·g·

4 · Ub · rb

(A.4)

9 · Ub ·

(A.5)

2 · rb2

aD = 1.8 ·

DB Ub · rb3

1/2

= 2.4 · rb

4 · 1/6

·

(A.7)

3

The particle relaxation time and the particle diffusion coefﬁcient which appear in previous equations can be expressed as:

=

2 · p · rp2 m = 6 · · G · rp 9 · G

DB =

(A.8)

kB · T 6 · · G · rp

(A.9)

where m, p and rp are the particle mass (kg), density (kg/m3 ) and radius (m), respectively, G the gas dynamic viscosity (Pa s), kB the Boltzmann’s constant (J/K) and T the gas temperature (K). Thus, by substituting Eqs. (A.7), (A.8) and (A.9) into Eqs. (A.4), (A.5) and (A.6), expressions of absorption coefﬁcients for the three deposition mechanisms become:

aI =

31/6 · p · g 3/2

2.4 · (4)1/6 · 6 · G · rb 2.4 · p · (4)1/6 3/2

31/6 · G · rb

· rp2 = CS · rp2

· rp2 = CI · rp2

1.8 · 31/12 · (kB · T ) (4)

1/12

· (6)

(A.10)

1/2

1/2

· 2.4

(A.11)

1/2

1/2 · G

7/4 · rb

·

1 1/2 rp

−1/2

= CD · rp

(A.12)

Supposing independence and consequently additivity of the three collection mechanisms, the total absorption coefﬁcient becomes: −1/2

a = (CI + CS ) · rp2 + CD · rp

(A.13)

References

(A.2)

where Cp,0 is the initial particle concentration inside the bubble. The total collection efﬁciency can be deﬁned as: =1−

1/6

Ub = 2.4 · Vb

aD =

Deposition of aerosols in a rising gas bubble is described by the differential equation: dCp = −a · Cp dh

Pich and Schütz [10] used a bubble rising velocity that only depends on the bubble volume (Vb ). The authors considered spherical bubbles and deﬁne an expression for the bubble rising velocity which can be included in the above equations:

aS =

Authors are grateful with the French National Research and Safety Institute (INRS) for its technical and ﬁnancial assistance and with the French Environment and Energy Management Agency (ADEME) for the ﬁnancial support of our project (agreement no. 11-81-C0084).

439

1/2 (A.6)

where g is gravity acceleration (m2 /s), the relaxation time of the particle (s), Ub the bubble rising velocity (m/s), rb the bubble radius (m) and DB the particle diffusion coefﬁcient (m2 /s).

[1] D. Bémer, R. Régnier, I. Subra, B. Sutter, M.-T. Lecler, Y. Morele, Ultraﬁne particles emitted by ﬂame and electric arc guns for thermal spraying of metals, Ann. Occup. Hyg. 54 (2010) 607–614. [2] X. Simon, D. Bémer, S. Chazelet, D. Thomas, R. Régnier, Consequences of high transitory airﬂows generated by segmented pulse-jet cleaning of dust collector ﬁlter bags, Powder Technol. 201 (2010) 37–48. [3] J. Binnig, J. Meyer, G. Kasper, Origin and mechanisms of dust emission from pulse-jet cleaned ﬁlter media, Powder Technol. 25 (2009) 108–114. [4] G. Mouret, D. Thomas, S. Chazelet, J.-C. Appert-Collin, D. Bémer, Penetration of nanoparticles through ﬁbrous ﬁlters perforated with deﬁned pinholes, J. Aerosol Sci. 40 (2009) 762–775. [5] B.C. Meikap, M.N. Biswas, Fly-ash removal efﬁciency in a modiﬁed multi-stage bubble column scrubber, Sep. Purif. Technol. 36 (2004) 177–190. [6] A. Bandyopadhyay, M.N. Biswas, Particulate-laden-SO2 scrubbing in a tapered bubble column, Environ. Prog. Sustainable Energy 28 (2009) 212–225. [7] S. Yuu, T. Yotaki, K. Abe, Investigation of the collection mechanism in absorption of aerosols by bubbling through water, Powder Technol. 17 (1977) 115–122. [8] M. Hermeling, A.P. Weber, Nanoparticle separation from rising bubbles in aqueous solutions, in: Proceedings of World Congress on Particle Technology, Nuremberg, Germany, 2010. [9] N.A. Fuchs, The Mechanics of Aerosols, Pergamon Press, 1964. [10] J. Pich, W. Schütz, On the theory of particle deposition in rising gas bubbles: the absorption minimum, J. Aerosol Sci. 22 (1991) 267–272. [11] G. Hébrard, D. Bastoul, M. Roustan, Inﬂuence of gas sparger on the hydrodynamic behaviour of bubble columns, Chem. Eng. Res. Des. 74 (1996) 406–414. [12] E. Gaddis, A. Vogelpohl, Bubble formation in quiescent liquids under constant ﬂow conditions, Chem. Eng. Sci. 41 (1986) 97–105. [13] A.A. Kulkarni, J.B. Joshi, Bubble formation and bubble rise velocity in gas-liquid systems: a review, Ind. Eng. Chem. Res. 44 (2005) 5473–5931.

No title

No title

No title

No title

No title

No title

No title

No title

No title

No title

No title

No title

No title

No title

No title

No title

No title

No title

No title

No title

No title

No title

No title

No title

No title

No title

No title

No title

No title

No title

Recommend Documents