WATER RESEARCH A Journal of the International Water Association
Editor-in-Chief Mark van Loosdrecht Delft University of Technology and KWR Watercycle Research The Netherlands E-mail:
[email protected]
Editors Jean-Claude Block Université H. Poincaré, Nancy I France David Dixon University of Melbourne Australia Hiroaki Furumai University of Tokyo Japan Xiaodi Hao Beijing University of Civil Engineering and Architecture China Gregory Korshin University of Washington USA Anna Ledin Formas Sweden Eberhard Morgenroth ETH Zürich and Eawag, Swiss Federal Institute of Aquatic Science and Technology Switzerland Wolfgang Rauch University of Innsbruck Austria Maria Reis Universidade Nova de Lisboa/FCT Portugal Hang-Shik Shin Korea Advanced Institute of Science and Technology Korea Thomas Ternes Bundesanstalt für Gewässerkunde Germany Stefan Wuertz University of California, Davis USA
Associate Editors Andrew Baker University of New South Wales Australia Damien Batstone University of Queensland Australia
Guang-Hao Chen Hong Kong University of Science & Technology China Tom Curtis University of Newcastle upon Tyne UK Francis de los Reyes III North Carolina State University USA Ana Deletic Monash University USA Rob Eldridge Glen Waverley Australia Monica B. Emelko University of Waterloo Canada Rosina Girones University of Barcelona Spain Stephen Gray Victoria University Australia Kate Grudpan Chiang Mai University Thailand Edwin Herricks University of Illinois at Urbana-Champaign USA Peter Hillis United Utilities Plc UK Hong-Ying Hu Tsinghua University China Bruce Jefferson Cranfield University UK Ulf Jeppsson Lund University Sweden Sergey Kalyuzhnyi Moscow State University Russian Federation Jae-Hong Kim Georgia Institute of Technology USA Günter Langergraber University of Natural Resources and Applied Life Sciences, Vienna (BOKU) Austria
Yoshihiko Matsui Hokkaido University Japan Armand Maul Université de Metz France Max Maurer Eawag Switzerland How Yong Ng National University of Singapore Singapore Satoshi Okabe Hokkaido University Japan Jong Moon Park Pohang University of Science & Technology Korea
Shang-Lien Lo National Taiwan University Taiwan Dionisis Mantzavinos Technical University of Crete Greece
Anastasios Zouboulis Aristotle University of Thessaloniki Greece
Susan Richardson US Environmental Protection Agency USA Miguel Salgot University of Barcelona Spain David Sedlak University of California, Berkeley USA Jean-Philippe Steyer LBE-INRA France Masahiro Takahashi Hokkaido University Japan Nathalie Tufenkji McGill University Canada Kai Udert Eawag Switzerland Vayalam Venugopalan BARC Facilities India Eduardo von Sperling Federal University of Minas Gerais Brazil Julie Zilles University of Illinois at Urbana-Champaign USA
Editorial Office E-mail:
[email protected]
Publication information: Water Research (ISSN 0043-1354). For 2011, volume 45 is 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/watres). 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. 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/watres. 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. 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 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] Application for membership of International Water Association should be made to: Executive Director, IWA, Alliance House, 12 Caxton Street, London SW1H 0QS, U.K. (Tel.: +44 207 654 5500; Fax: +44 207 654 5555; e-mail:
[email protected]; website: http://www.IWAhq.org.uk). Registered Charity (England) No. 289269. Individual membership is available from £30 upwards. For details contact IWA.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 2 8 7 e4 3 1 0
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Retrospective of ecological approaches to excess sludge reduction Anwar Khursheed*, A.A. Kazmi Department of Civil Engineering, Indian Institute of Technology Roorkee, India
article info
abstract
Article history:
The problem of excess sludge handling produced during wastewater treatment is unde-
Received 18 June 2010
niable reality of grave concern with increasingly stringent legislations. The sludge
Received in revised form
synthesis yield being 0.4e0.6 kgVSS/kgCOD (0.57e0.8 kgCODcell/kgCOD), results in high
26 February 2011
power consumption on its digestion and therefore taken considerable attention to achieve
Accepted 22 May 2011
sustainable strategies.
Available online 31 May 2011
Solids reduction by physico-chemical methods results in buildup of chemicals. This may present risk to the environment and may require further treatment to remove the
Keywords:
chemicals of concern in future. Wastewater sludge reduction upto 100% by biological,
Low sludge production
sustainable, non-hazardous, and environment friendly methods has been successfully
MBR
tested at different levels. Therefore, above reasons were sufficient driving forces to confine
Oligochaeta
this review to non-chemically assisted processes. Similarly, the thermally assisted
Filamentous fungi
processes result in high carbon footprint and excluded from the scope of this review.
Maintenance metabolism
Enough has been reviewed on sludge reduction, as numbers of articles on the same subject
High oxygenation
with different angles have been reported, still the progress in the last few years is missing; hence, special emphasis is given herewith to highlight the efforts of the last five years. ª 2011 Elsevier Ltd. All rights reserved.
Contents 1.
2.
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.1. Carbonaceous oxidation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.2. Nitrification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.3. Denitrification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Sludge reduction efforts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1. Lysis-cryptic growth . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1.1. High oxygenation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2. Uncoupling metabolism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2.1. Initial substrate to biomass ratio (So/Xo) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2.2. Oxic-settling-anaerobic (OSA) process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3. Maintenance metabolism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3.1. Cost and energy considerations in MBR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
* Corresponding author. Tel.: þ91 9319077764; fax: þ91 1332 275568. E-mail addresses:
[email protected] (A. Khursheed),
[email protected] (A.A. Kazmi). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.05.018
4288 4288 4288 4289 4289 4291 4291 4292 4292 4293 4293 4297
4288
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 2 8 7 e4 3 1 0
2.4.
3. 4.
1.
Eco-manipulation (predation on bacteria) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.4.1. Two-stage system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.4.2. Oligochaeta . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.4.3. Filamentous fungi (FF) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Where; fsa ¼ fsa fm ¼ Net fraction of electron donor used in synthesis
Introduction
Organic matter and nutrient removal through biological oxidation is generally regarded as the most economically viable means of wastewater treatment. But, only those systems removing inorganic nutrients simultaneously and generating fewer volumes of excess sludge biomass can be regarded comprehensive, so as to give real protection to streams against eutrophication and environment against sludges. A very complex microbial ecosystem consisting of bacteria, protozoa, metazoa, viruses, worms, helminthes etc is involved in this sequential degradation of organics followed by nitrification, denitrification and phosphorous removal. Prerequisites to successful removal of all these substrates are efficient growth of biomass for organics, low COD, high dissolved oxygen (DO) and long sludge retention time (SRT) for nitrification, while reverse is true for denitrification in the form of sufficient COD in the absence of DO. This modus vivendi is itself a difficult preposition to optimize or even achieve (Metcalf and Eddy, 2003). In summary, the entire process is conversion of different components of wastewater by oxidation and reduction to end products, namely CO2, H2O, N2, new cells and fractions of intermediateries remaining due to inefficiencies (Eq (1.1)). microbes
v1 ðorganicsÞ þ v2 O2 þ v3 NH3 þ v4 PO3 4 ! v5 ðnew cellsÞ þ v6 CO2 þ v7 H2 O
(1.1)
Where, vi ¼ stoichiometric coefficient Although the outcome of pollution caused by the municipal or industrial effluents is environmental degradation, nevertheless its dimensions are different. Since, municipal wastewaters pose a universal problem; therefore, generalized expressions regarding whole process, which also gives stoichiometry of sludge growth are summarized below (Rittmann and McCarty, 2001; Metcalf and Eddy, 2003);
fsao ¼ Max fraction of electron donor used in synthesis ¼ 0.6 fm ¼ Fraction of electron donor used in maintenance feo ¼ Max fraction of electron donor used to provide energy Net fraction of electron donor used to provide energy ¼ fe fe þ fsa ¼ 1 Microbial biodegradable fraction, fd ¼ 0.8 b ¼ 0.15/d at 20 C, SRT ¼ qx, 1 þ 1 f d b qx 1 þ b qx ð1 þ 0:03 qxÞ f sa ¼ 0:6 ð1 þ 0:15 qxÞ 1 1 Synthesis; Ya ¼ f sa C5 H7 O2 N= C10 H19 O3 N 20 50
f sa ¼ f sa
C10 H19 O3 N þ 12:5 O2 ¼¼ 9CO2 þ 7H2 O þ NHþ 4 þ HCO3
Ya ¼ f sa
1.2.
1 f sa 1 1 O2 þ ¼¼ C10 H19 O3 N þ 9 5 f sa CO2 4 50 25 1 f sa 9 f sa H2 O þ 1 NHþ þ 1 4 2 20 25 1 f sa f sa C5 H7 O2 N HCO þ 3 þ 20 50 20
113 400 = ¼ 0:70625f sa gVSS=gUBOD 20 50
(1.5)
1 f sn 1 f sn f sn f sn 1 O2 þ NHþ HCO CO2 ¼¼ NO þ 4 þ 3 þ 3 4 20 5 8 20 8 1 f sn 9 3 1 f sn þ C5 H7 O2 N þ H2 O þ Hþ þ 2 20 20 8 4 (1.6) Where fractions of electrons and biodegradable fraction are in similar meaning; Max fraction of electron donor used in nitrifiers synthesis, fsno ¼ 0.14 b ¼ 0.11/d at 20 C 1 þ 1 f d b qx 1 þ b qx
f sn ¼ 0:14
(1.4)
Nitrification
o
Carbonaceous oxidation
(1.3)
COD ¼ UBOD of C10 H19 O3 N ¼ 12:5 2 16 ¼ 400
f sn ¼ f sn
1.1.
4298 4298 4300 4302 4302 4305 4305 4305
ð1 þ 0:022 qxÞ ð1 þ 0:11 qxÞ
(1.7)
f sn 1 f sn C5 H7 O2 N= þ NHþ 4 20 8 20 f sn 1 f sn 8:07 f sn 113= þ 14 ¼ ¼ g VSS=gN 20 2:5 þ f sn 8 20
Synthesis; Yn ¼
(1.2)
(1.8)
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 2 8 7 e4 3 1 0
Since, 14 g N is equivalent to 64 g O2, therefore
Yn ¼
1:765 f sn g VSS=g UBOD 2:5 þ f sn
1.3.
(1.9)
Denitrification
þ 1 f sd 1 1 NO C10 H19 O3 N þ 9 5f sd CO2 3 þ 1 f sd H ¼¼ 5 50 25 3 1 f sd 1 f sd 9 N2 þ 1þ H2 O þ 5 10 25 1 f sd f sd HCO C5 H7 O2 N þ 3 þ 20 50 20 1 f sd þ NHþ ð1:10Þ 4 50 20 Where; Max fraction of electron donor used in synthesis, fsdo ¼ 0.52. b ¼ 0.05/d, Microbial biodegradable fraction, fd ¼ 0.8 o
f sd ¼ f sd
f sd
1 þ 1 f d b qx 1 þ b qx
Synthesis; Yd ¼ f sd
Yd ¼ f sd
(1.11)
1 1 C5 H7 O2 N= C10 H19 O3 N 20 50
113 400 = ¼ 0:70625 f sd g VSS=gUBOD 20 50
(1.12)
The net synthesis in terms of NO3 (14 g N ¼ 64 g O2) would be; Yd ¼ f sd ¼
113 1 f sd 113x5 f sd g VSS=g NO = 3 ¼ 5 20 20x14 1 f sd
2:01 f sd gVSS=gNO 3 N 1 f sd
(1.13)
Therefore, the combined growth of the mixed culture would be, Y ¼ Ya þ Yn þ Yd ¼ 0:70625 f a þ 0:70625 f sd þ
1:765 f n 2:5 þ f n
ð1 þ 0:03 qxÞ ð1 þ 0:01 qxÞ ¼ 0:70625 0:6 þ 0:70625 0:52 ð1 þ 0:15 qxÞ ð1 þ 0:05 qxÞ ð1 þ 0:022qxÞ ð1 þ 0:022qxÞ 1 0:14 þ1:765 0:14 ð1 þ 0:11qxÞ ð1 þ 0:11qxÞ Y¼
0:424ð1 þ 0:03 qxÞ 0:367ð1 þ 0:01 qxÞ þ ð1 þ 0:15 qxÞ ð1 þ 0:05 qxÞ þ
explained above. For old or slow-growing cultures fs would be less than maximum value and correspondingly same would be the overall synthesis. However, there are other reactions also taking place simultaneously and the net sludge synthesis depends on numerous factors, mainly energy requirement for cell maintenance, decay of cells, endogenous respiration, grazing by predators and lysis due to suboptimal environmental conditions and toxicity. Bio-oxidation in activated sludge or similar process configurations is highly complex due to mixed cultures and their heterogeneity under variable conditions. Being substrate limiting, cell decay and endogenous respiration accounts for considerable oxygen consumption, this has been quantified as more than 50% in many practical cases (van Loosdrecht and Henze, 1999). Another intricacy is the re-dissolution of phosphates accumulated into the sludge due to in-situ excess sludge reduction and production of soluble microbial products (SMP) within it, which are basically biomasseassociated products. These products provide electron for heterotrophs and increase their mass (Laspidou and Rittmann, 2002).
2.
ð1 þ 0:01 qxÞ ¼ 0:52 ð1 þ 0:05 qxÞ
0:247ð1 þ 0:022 qxÞ 2:5ð1 þ 0:11 qxÞ 0:14ð1 þ 0:022 qxÞ
(1.14)
Based on above Eq. (1.14) the overall combined growth of the mixed culture would vary from 0.418 to 0.55 at 20 to 10 days SRT. The combined effect in terms of overall sludge growth of all heterotrophs and autotrophs is its dependence on fs value in each case independently and consequently on SRT as
4289
Sludge reduction efforts
“Sustainable sludge handling may be defined as a method that meets requirements of efficient recycling of resources, without supply of harmful substances to humans or the environment” (Commission of European Communities, 1998). Of the constituents removed by wastewater treatment, sludge is by far the largest in volume, therefore, it’s handling methods and disposal techniques are a matter of great concern, because, without a reliable disposal method for the sludge the actual concept of water protection cannot sustain. At the time of above legislative amendment, average dry weight per capita production of sewage sludge resulting from primary, secondary and tertiary treatment was 90 g per person per day in 1998 in Europe. At the time of the implementation of the above cited Urban Waste Water Treatment Directive (UWWTD) it was expected that by the year 2005, the sludge production would increase by 50%, i.e. 10 million tons annually. Therefore, by the year 2010 and 2050 the European Union targeted to reduce final waste disposal by 20% and 50% compared to the amount of sludge waste disposed in 2000 respectively (Lundin et al., 2004). Similarly the prohibitions of sludges in landfills or surface impoundments by virtue of 1984 amendment to the Resource Conservation and Recovery Act of 1976 (RCRA, 1984) in the form of Hazardous and Solid Waste Amendment (HSWA, 1984) by the US Congress was a clear intent to move sludge management towards more acceptable technologies. The legislative concern of European and American communities has been translated into a surge of scientific studies on sludge handling, which can best be realized from the fact that average number of published articles, which was around 83 per annum till the end of last century rise to 239 per annum up till now (Engineering Village, http://www. engineeringvillage2.com) as shown graphically below in Fig. 1: No such unified legislative concerns have been shown by the Asian countries, nonetheless the situation argues for a vigorous regulatory mechanism. But, economic realities often dictate that environmental concerns not be permitted to
4290
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 2 8 7 e4 3 1 0
Fig. 1 e Details of annual publication of articles on sludge reduction.
stand in the way of development and as a result the issue is inadequately addressed or not dealt with at all. The problem of handling of excess sludge produced during wastewater treatment is an undeniable reality of grave concern and gaining large proportion with increasingly stringent environmental legislations by the respective nations at different level. This resulted in a situation that sludge reduction has almost taken a central stage in overall wastewater scenario as its treatment and disposal accounts considerably (Horan, 1990). Consequently, there is a noticeable shifting of priorities from higher volumetric rate and effluent quality to less biomass production during wastewater treatment on the basis of overall mass balance of the inputs and outputs. At the outset, it is important to identify the various components of municipal excess sludge, which may be containing 0.75% from secondary clarifier to 10% solids from thickener and rest is water (Metcalf and Eddy, 2003). The major components of these solids in the form of residues of the original wastewater and synthesized biomass are around 60% organic compounds (non-toxic), organic and inorganic nitrogen, phosphorous, heavy metals (mainly Zn, Ni, Cd, Cu, Cr, Hg, Pb, As etc. in varying concentrations), trace organic matters such as pesticides, dioxins, phenols, polychlorinated biphenyls and polycyclic aromatic hydrocarbon etc., and microbes such as bacteria, viruses, pathogens, predators etc. Some of these are beneficial and reusable, while others are hazardous merely by presence or by virtue of their concentration. The sustainability of the sludge handling ultimately depends not only on volume and mass reduction only but also on recovery of usable and containment of adverse effects of hazardous components. The average sludge synthesis yield being 0.4e0.5 kg VSS/ kg COD consumed, which amounts to 0.57e0.8 kg COD cell/
kg COD consumed, obviously results in more than 57% of the power consumption on sludge digestion as rightly mentioned by Horan (1990) in his book and Chen et al. (2000). This could be attributed to the fact that most of the existing sludge reduction technologies are capital intensive and process-wise complex. As mentioned above, excess sludge reduction has taken considerable attention in order to achieve more sustainable strategies to cope up present as well as future requirements. It is not so that technologies are not available, the recent advent of sludge reduction technologies has validated the potential to significantly change the methods by which wastewater treatment biosolids are treated and handled across the globe, but still the real impetus is to explore economically viable one. Ideally, the solution should be in-pipe rather than subsequent treatment. In addition the real environmental friendly approach is to keep the process biological in place of physicochemical or mechanical. Solids reduction technologies can be categorized in three major categories according to their treatment methods viz.; physico-chemical, mechanical, and biological. The physicochemical and mechanical methods are fairly easily understood in how they might function, i.e. through the oxidation of organic material or the lysis of microbial material, thus making the overall mass more degradable, but the biological systems of sludge reduction, and the mechanisms behind them are much less understood. Since, the volumetric percentage of water derived from sewage treatment plants in receiving waters has been increasing; thereby, their quality is greatly influenced by the treated wastewater and most of the time resulting in bioaccumulation of chemicals. Research is in progress to carry out a meaningful ecological risk assessment of certain groups of chemicals of concern in wastewater treatment plant
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 2 8 7 e4 3 1 0
effluent which may present a risk to the environment. So that upgradation of treatment processes to remove the chemicals of concern could be implemented in future. The inclusion of disinfection byproducts may be due to anticipated effect of water recycling, where the potential exists for a gradual accumulation of disinfection byproducts in both the recycled water and wastewater. Chlorinated phenols constitute an important class of pollutants because of their wide use in sludge reduction as protonophores. Due to their strong toxicity and odour emission, persistence in environment, and suspected carcinogen and mutagen to the livings, chlorophenol have posed serious ecological problem (Armenante et al., 1999). Similarly, sonification and all the thermally assisted processes result in high carbon footprint and excluded from the scope of this review Therefore, lesser the assistance of chemicals, better for the environment, Hence, a detailed account of various engineering/technological approaches to achieve the goal of minimum sludge generation/reduction through bio-oxidation and without any chemical assistance is given in this paper based on classification as; (a) Biological lysis-cryptic growth (High oxygenation), (b) Biological uncoupling of metabolism (High So/Xo, OSA), (c) Maintenance Metabolism (MBR), and (d) Eco-manipulation (Predation). Since only in-place oxidation of excess sludges being natural process without any external and extraneous intervention is taken for discussion, therefore chemical uncoupling, has been left out. Moreover, enough has been reviewed on sludge reduction, still the progress in the last few years is missing, and hence, special emphasis is given herewith to highlight the efforts of the last five years.
2.1.
Lysis-cryptic growth
All cells have a plasma membrane, a protein-lipid bilayer that forms a barrier separating cell contents from the extracellular environment. Lysis (Greek, means to separate) refers to the death of a cell by breaking of the cellular membrane through different mechanisms that compromise its integrity. A solution containing the contents of lysed cells is called a “lysate”. Cell lysis is the first step in cell fractionation, which will release cell contents into the medium, thus providing an autochthonous substrate that contributes to the organic loading, microbial metabolism and a portion of the carbon is liberated as products of respiration and results in a reduced overall biomass production. The biomass grew on organic lysate is different from growth on original substrate, and therefore termed as cryptic (Greek word 徒ryptos means hidden) growth as it is not present from the beginning (Mason et al., 1986). It consists of lysis and biodegradation, where the former does not occur under normal conditions, however once lysed, it becomes easy for the living cells to biodegrade the lysed cells, therefore lysis is the rate-limiting step of lysis-cryptic growth, and an increase of the lysis efficiency can lead to an overall reduction of sludge production. The concept of cryptic growth was first introduced by Ryan (1959), followed by Mason’s (1986) PhD thesis on microbial death, lysis and cryptic growth. Mason and Hamer (1987) and Mason et al. (1986) on cryptic growth in Klebsiella
4291
pneumoniae in batch culture characterized sludge production yield of 0.33 g biomass/gCOD as against 0.56 g biomass/gCOD without cell lysis. The biodegradation of the cell wall is cited as the rate-limiting step and to increase it, several physicochemical and biological processes individually or in combination can be used in downstream processing. Several methods which have been applied so far for sludge disintegration summarized by Wei et al. (2003a) are herewith reported; (i) thermal treatment in the temperature range from 40 C to 180 C (Kepp et al., 1999; Barjenbruch et al., 1999), (ii) chemical treatment using acids or alkali (Tanaka et al., 1997,), (iii) mechanical disintegration using ultrasounds, mills, and homogenizers (Baier and Schmidheiny, 1997; Kopp et al., 1997; Camacho et al., 2002; Nolasco et al., 2002; Tiehm et al., 1997, 2001; Chu et al., 2001 and Onyeche et al., 2002), (iv) freezing and thawing (Chu et al., 1999), (v) biological hydrolysis with enzyme addition (Guellil et al., 2001), (vi) advanced oxidation processes such as wet air oxidation, using hydrogen peroxide and ozone (Weemaes et al., 2000a, 2000b; Shanableh, 2000 and Neyens et al., 2003c), and (vii) combination ways such as thermo-chemical treatment (Saiki et al., 1999; Neyens et al., 2003a, 2003b), combination of alkaline and ultrasonic treatment (Chiu et al., 1997). Ozonation has been successfully used on full scale (Yasui et al., 1996; Ahn et al., 2002; Egemen et al., 1999, 2001 and Deleris et al., 2002), Chlorination (Chen et al., 2001; Saby et al., 2002), integration of thermal/ultrasonic treatment and membrane (Canales et al., 1994; Chu et al., 2001), integration of alkaline and heat treatment (Rocher et al., 1999, 2001) are the other techniques in application for reducing sludge production, but on laboratory or pilot scale. The sludge reduction potential by ozonation, chlorination, thermo-chemical treatment, and high DO is reported as 100, 65, 60 and 25% respectively (Wei et al., 2003a). In abstract, the success of application of lysis-cryptic growth for sludge reduction depends on efficiency and economy of lysate production and till date no commercial success has been achieved except application of ozonation at full scale (Wei et al., 2003a).
2.1.1.
High oxygenation
Aerobic excess sludge digestion as a separate unit process is in application since ages. The three most commonly proven variations are; conventional, high-purity oxygen based and autothermal assisted digestion. However, excess sludge reduction by aerobic digestion on its place of production itself has been conflictingly reported in the literature from nil to 66% (Humenick and Ball, 1974; Roques et al., 1984; Sengewein, 1989; Metcalf and Eddy, 2003; Mudrack and Kunst, 1991; Bitton, 1994). McWhirter (1978) and Boon and Burgess (1974) reported 54% and 60% less yield with pure oxygen in comparison to air system, respectively. Abbasi et al. (2000) showed that a rise of the DO concentration from 2 to 6 mg/L led to about 25% sludge reduction at the sludge loading of 1.7 mg BOD5/mg MLSS d. The increase of the DO in the bulk liquid led to a deep diffusion of oxygen, which subsequently caused an enlargement of the aerobic volume inside the flocs, as a result, the hydrolyzed microorganisms in the floc matrix could be degraded and thus sludge
4292
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 2 8 7 e4 3 1 0
quantity was reduced, has been cited as a strong reason of increased sludge reduction. This has been substantiated by quantifying floc break-up due to mixing and oxygen and substrate profile inside the core of the sludge particle of 0.35 mm diameter size. This was accomplished by inducing nitrogen gas to keep the constant amount of diffused gas and consequently velocity gradient while increasing DO from 2 to 6 mg/L. The results showed 10% of the excess sludge reduced due to break-up of flocs as a result of the degree of mixing and remaining 12% was the result of elevated DO concentration in the mixed liquor (Abbasi et al., 2000). In conventional activated sludge (CAS) processes the oxygen transfer yields range from 0.6 to 4.2 kgO2/kW h depending on the methods of aeration, moreover, maintaining high DO in bulk liquid would further increase aeration cost very sharply, it is needless to mention that aeration itself costs more than 50% of the total energy consumption (Low and Chase, 1999a). Contrary to above, Wilen and Balmer (1999) concluded compact flocs at high and low effluent turbidity while working on effect of DO on structure and size of sludge flocs. However, Liu and Tay (2001) in their review rightly mentioned that the cell surface hydrophobicity, microbial activity and exopolymer production are linked to DO level in the reactor (Roques et al., 1984; Mishima and Nakamura, 1991; Palmegren et al., 1998; Pena et al., 2000). The role of DO as an uncoupler between anabolism and catabolism cannot be ruled out subject to further research. It is also mentioned that low DO favours growth of filamentous organism that cause sludge bulking (Richard et al., 1985; Nowak et al., 1986). On the contrary, activated sludge process operating on high DO can efficiently repress development of filamentous organisms in the aeration tank (McWhirter, 1978; Bitton, 1994; Abbassi et al., 2000). The choice of using air or pure oxygen has also been excercized subjective to degradability of excess activated sludge from a wastewater treatment plant at different temperatures (Zupancic and Ros, 2008). Considering the overall input in the digestion, there are advantages and disadvantages to oxygen aeration. Benefield and Randall (1980) reported number of advantages offered by pure oxygen process as compared to conventional air aeration process, such as ability to maintain a higher MLVSS concentration in the aeration tank; better sludge settling and thickening; lower net sludge production; higher oxygen transfer efficiency per horsepower and more stable operation. McWhirter (1978) observed that the growth yield in purified oxygenation activated sludge process can be reduced by 54% compared to conventional system, even at high sludge loading rates. Boon and Burgess (1974) reported 60% reduction in growth yield under similar SRT, utilizing non-purified air. Wunderlich et al. (1985) showed reduced sludge production from 0.38 to 0.28 mg VSS/mg COD removed as the SRT increased from 3.7 to 8.7 days in high-purity oxygen activated sludge system. These results indicated that the pure oxygen aeration process operated at a relatively longer SRT is more efficient in reduction of excess sludge production. Consequently, high oxygen process shows great industrial potential for minimization of excess sludge production as well as in improvement of system operation. However, economicefficiency and energy-balance calculations should be taken
in to consideration as important tool for performing the costbenefit analysis of a disintegration process.
2.2.
Uncoupling metabolism
The basic law of conservation of energy holds well in the biochemical pathways too, where the energy is transferred from electron donor to the microbes to perform different kind of activities. The reaction often starts with formation of electron carriers’ flavin adenine dinucleotide and nicotinamide adenine dinucleotide. These carriers during the oxidative phosphorylation give up electron and release proton. The proton during its discharge outside the cell membrane creates a charge imbalance and a pH gradient across the membrane. This is called proton motive force (PMF). The chemical energy stored in the proton gradient is used by the cells for adenosine triphosphate (ATP) formation from adenosine di-phosphate (ADP) apart from other products. The ATP thus formed during catabolism transfer it to cells for anabolism namely synthesis, maintenance and motility. The PMF is basically the driving force for the transfer of energy from catabolism to anabolism and is termed as energy coupling through rate limiting respiration (Rittmann and McCarty, 2001). The uncoupling is therefore short circuiting the PMF to restrict ATP formation, whilst simultaneously substrate oxidation (Stryer, 1988). This result in declined observed growth yield of biomass without reducing the removal rates of organic pollutants in biological wastewater treatment and may therefore provide a direct mechanism for reducing sludge production. The documented conditions of uncoupled metabolism are the presence of inhibitory compounds, heavy metals, abnormal temperatures, limitation of nutrients, excess energy source, and exposure of sludge to cyclic change in ATP content (Liu and Tay, 2001; Abbasi et al., 2000; Stouthamer, 1979 and Chudoba et al., 1992a). As mentioned in the beginning only last two are within the scope of unassisted uncoupling and hence would be discussed later in detail.
2.2.1.
Initial substrate to biomass ratio (So/Xo)
Chudoba et al. (1991) and Liu et al. (1998) observed decline in sludge growth at high initial substrate concentration to the initial biomass concentration (So/Xo as COD/biomass) in batch cultivation of mixed cultures as detailed in Table 1. The concept though modelled and verified experimentally (Liu, 2000) has yet to find a place in its engineering application in wastewater treatment plants, where the actual So/Xo ratios are 0.01e0.13 mg COD/mg MLSS for sewage (Chudoba et al., 1991c). Moreover, research on the process initiated by Chudoba et al. (1991) has not been globalized, and was only followed by Liu (1996, 2000) and Liu et al. (1998). Wei et al. (2003a) concluded that maintaining high food to microorganism ratio to achieve low sludge production may cause deterioration in effluent quality and would necessitates further treatment of wastewater (post-treatment) to meet the desirable levels of effluent organic matter. This would result in high capital and operation costs, and therefore makes it fit only for the biological treatment of high strength industrial wastewaters. No further progress in research was made since then.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 2 8 7 e4 3 1 0
4293
Table 1 e Summarized experiences with biological uncoupling metabolism. Reference(s)
Process
Highlights
Chudoba et al. (1991)
High So/Xo
Liu (1996), Liu et al., 1998, Ghigliazza et al. (1996)
High So/Xo
Liu, 2000
High So/Xo
Chudoba et al. (1992a,b) Chen et al. (2003)
OSA OSA
Saby et al., 2003 and Chen and Leung, 1999; Chen et al., 2003
OSA
Jun et al., 2004a
OSA
Energy dissipation (at So/Xo ratio 5 mg COD/mg MLSS) by leakage of ions through the cell membrane, which weakens the potential across it and thus uncouples oxidative phosphorylation. Substrate sufficiency in terms of high So/Xo cited a reason of lower yield owing to energy spilling due to uncoupling by excess substrate. The reduced sludge growth is quantified in terms of energy uncoupling coefficient (Eu) on a scale of 0e1.0. The kinetic model based on carbon partition between energy spilling-associated and growth-associated metabolisms at different So/Xo ratios provides mechanism of sludge reduction. Sludge yield were significantly lower by about 38e54% compared to CAS. Ruled out energy uncoupling, domination of slow growers, and SMPs as possible cause of sludge reduction and, instead low ORP is cited as the sole reason for the same. In a modified OSA process with a membrane module in the aeration tank, at a controlled ORP of 250 mV in the anaerobic tank, the excess sludge was reduced by 36%, and 58% compared to þ100 mV ORP or compared to CAS process. Increased soluble COD in the anaerobic tank, causing cryptic growth during aerobic phase caused low sludge yield in the overall OSA process. Established symbiosis between aerobes and methanogens by controlling DO from 0 to 0.5 mg/L in the aeration tank to reduce sludge production and higher removal efficiencies of total COD of 93%. The lower observed yield coefficients of 0.28 gVSS/ gCOD was achieved as against 0.41 and 0.37 in intermittent aeration and CAS reactors.
2.2.2.
Oxic-settling-anaerobic (OSA) process
Oxic-settling-anaerobic (OSA) process has well established the fact that anaerobic followed by aerobic in a cyclic order reduces the sludge through promotion of catabolism and demotion of anabolism by uncoupling the two reactions. Aerobic microorganisms capture energy in the form of ATP released from oxidation of organic content. The same microorganisms are unable to produce required energy when exposed to anaerobic conditions under severe food limiting condition and as a result consume their conserved ATP. Again on return to aerobic condition, augmentation of their depleted ATP reserves become first priority as against synthesis of new cell mass or anabolism. This in other words promotes catabolism and demotes anabolism by uncoupling the two reactions, thereby inducing sludge reduction (Chudoba et al. (1992a,b). Similarly, altering anaerobic and aerobic environment causes death of obligate aerobic and anaerobic microorganisms. Thus produced lysed microorganisms release the intracellular matters, which can be degraded by various extracellular enzymes. The first concept of anaerobic exposure to returned aerobic sludge was probably given by Westgarth et al. (1964). It consists of three tanks namely aeration tank followed by a settling tank and an anaerobic tank situated in the return sludge line to create an alternative oxic and anaerobic cycle, creating thereby a fasting/feasting condition to the exposed biomass, which ultimately caused reduction in excess sludge production. The OSA process has subsequently been investigated by various researchers such as Chudoba et al. (1991, 1992a,b); Ghiglizza et al. (1996); Copp and Dold (1998) and Chen et al. (2000, 2001a,b, 2003) (Table 1). Similar to explanation given by Chudoba et al. (1992a,b) regarding the process of sludge fasting/feasting the same was given by Chen and Liu (1999). Chen et al. (2003, 2001a,b) and
Saby et al. (2003) systematically investigated energy uncoupling, domination of slow growers, and SMPs as possible cause of sludge reduction and ruled out all of them, instead low ORP is cited as the sole reason for the same. Encouraged by the simultaneous nitrificationedenitrification at lower rates by controlling DO level below 0.5 mg/L in oxidation ditches (Trivedi and Heinen, 2000), Jun et al. (2004a,) established a similar symbiosis between aerobes and methanogens, a major group of archaea. The experiments were performed in an intermittent aeration reactor (I/A), an intermittent aeration reactor dosed with archaea solution once a day (I/A-arch) and a conventional activated system (CAS) (Table 1). However, validation of these theories is needed to substantiate the mechanism of sludge reduction in the OSA process and symbiosis between aerobes and methanogens. Citing ORP as the sole reason behind reduced sludge production by Chen et al. (2003) lack supporting mechanism. Nevertheless, it appears OSA process can successfully handle the problem of high excess sludge production, which is particularly important when handling high strength waste streams in order to maintain the economical feasibility.
2.3.
Maintenance metabolism
The energy obtained and captured in the form of ATP during biological oxidation is used by the microbial cells for their maintenance followed by synthesis. Therefore, long sludge age results in increased energy consumption for maintenance, which leaves less energy for cell synthesis. In other words, long SRT causes reduction in sludge loading rate or low food to microorganism (F/M) ratio reduces sludge production (van Loosdrecht and Henze, 1999). In case of higher concentration of reactor biomass and limited substrate, to the extent that it is just sufficient to cater the energy requirement for
4294
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 2 8 7 e4 3 1 0
maintenance, than in a dynamic situation no energy would remain available to sustain the growth of microorganisms anymore. The decreased observed yield consequent upon increased energy requirement induced by high concentration of NaCl in a culture of Saccharomyces cerevisiae due to more cell rupture was observed by Watson in 1970. Contemporarily, Lawrence and McCarty (1970) described that how observed yield (Yobs) decreased with increased SRT (qc) with reference to maximum theoretical or true yield Yobs ¼
Ym ð1 þ kd qcÞ
(2.1)
Pirt (1975) subsequently related observed yield (Ys or Yobs) and maximum or true yield (YG) to SOUR for maintenance requirement (qm), where; 1 q 1 ¼ mþ m Ys YG
(2.2)
Low and Chase (1999b) quantified the biomass production per unit volume based on their proposal that the mass balance on the utilization of the energy source is the sum of the substrate utilized by anabolism and the biomass for satisfying maintenance requirements. This justified the assumption that cells preferentially satisfy the energy requirements associated with maintenance functions and that additional cell synthesis occurs using the remaining substrate available. It was also assumed that with a constant supply of substrate and a situation where growth is substrate limited at a constant level, substrate uptake rate (rs) is also constant; 1 rs ¼ $rx qm X YG
(2.3)
Thus the biomass production per unit volume may be represented by; rx ¼ YG $rs qm X
(2.4)
Therefore, the constant supply of limiting substrate condition preferentially satisfies the energy requirements associated with maintenance functions and that additional cell synthesis occurs using the remaining substrate available, the substrate uptake rate (rs) is also constant. Thus, the rate of substrate uptake specifically for cell synthesis per unit volume (rsG) is given as; rsG ¼ ð1=YGÞ$rx ¼ rs þ qm X
(2.5)
It seems logical that with Pirt’s (1975) allocation of energy under such a condition of biomass sedimentation and recycle, when the biomass concentration within the reactor is divorced from biomass production, the meaningful determination of specific growth rate (m) is difficult. Hence, Low and Chase’s (1999b) model which excludes the specific growth rate, but incorporates the biomass concentration, would provide a more suitable description of a system with partial biomass recycle, as the biomass concentration is a function of the sludge return rate and therefore is an accessible control parameter. The model was validated using data presented by Bouillot et al. (1990) (Table 2). As it is impossible to increase the sludge concentration significantly in CAS processes by means of settling even assisted with coagulants, therefore, similar results were
obtained when biomass was concentrated by membrane separation in membrane bioreactor (MBR). Catering the need of comprehensive treatment, onsite sludge handling is difficult, but essentially required. Therefore, the core idea is to artificially restrain the normal sludge growth within an acceptable level through establishing a preferable long SRT system independent of HRT, in order to approach zero excess sludge discharge. Membrane bioreactor (MBR) is the application of concept of increased energy consumption in cell maintenance, leaving little for growth, ideally to attain no sludge from wastewater bio-oxidation. Yamamoto et al. (1989); Mu¨ller et al. (1995); Wagner and Rosenwinkel (2000); Visvanathan et al. (2000); Rosenberger et al. (2002) and Witzig et al. (2002) showed that the MBR caused little/zero sludge production, but expensive in terms of energy requirements. It has also been successfully applied in full-scale plants (Churchouse and Wildgoose, 1999). All the claims of excess sludge free performance of MBR depend on physical and rheological properties of the retained sludge, namely diluted sludge volume index, the capillary suction time, the specific resistance to filtration and the compressibility. The capillary suction time test involves the measurement of time to move a volume of filtrate over a specified distance as a result of the capillary suction pressure of dry filter paper. It provides information regarding the ease of separating the water portion from the solids portion of sludge (EPA, 1987). Specific resistance to filtration test was performed by recording the volume of filtrate versus time when applying a negative pressure (0.5 bar) on a sludge sample, and filtering it through standard filter paper (EPA, 1987). A bench scale MBR with complete solid retention upto 22e23 g TSS/L was investigated and compared for about one year with CAS. The sludge showed substantial similarity in terms of dewaterability. The application of Ostwald model (Blair et al., 1939) indicated less increase in apparent viscosity in proportion to increasing solid concentrations. Other parameters, such as the diluted SVI (Giokas et al., 2003) and the reduced hysteresis area appeared to be scarcely dependent on the MLSS concentration, although larger hysteresis areas were observed in the period of rapid biomass buildup, suggesting a possible link to the sludge growth (hysteresis area provides an indication of the reversibility of the effects of a shear stress, in other words, the energy provided to the system by increasing the shear rate is used partly to overcome the resistance opposed by viscosity and partly to break the links between particles flocs). Based on calculated Reynolds number (3000e600), energy consumption for mixing resulted in a limited increase of energy requirements by 25e30% at increased solid concentration from 3 to 30 g TSS/L. These results obtained by Pollice et al. (2007) on account of similarities in the basic properties of MBR and CAS sludges pointed out that MBR can be efficiently adopted not only for providing high-quality effluents, but also for reducing sludge handling and disposal costs with respect to typical activated sludge processes. The attempt of Pollice et al. (2007) is interesting in view of limited information available on sludge characteristics. An energy optimization may not be obtained by a better understanding of the fouling phenomena of the membranes. The essential requirement of maintenance of higher MLSS at
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 2 8 7 e4 3 1 0
4295
Table 2 e Summarized experiences with biological maintenance metabolism. Reference(s) Low and Chase (1999b)
Process
Highlights
MBR
The biomass production is the sum of the substrate utilized by anabolism and the biomass for satisfying maintenance requirements. Increased biomass from 1.7 to 10.3 g/L, reduced biomass by 44% at qm ¼ 0.023 0.005 g substrate/g biomass.h and YG ¼ 0.361 0.022 g biomass/g substrate. Bouillot et al. (1990) Coagulant assisted Acetate fed chemostat resulted in maintenance quotient qm being 0.027 g substrate/g biomass h and sludge thickening 5% decrease in the observed yield. 12% decrease in the observed yield was found when the F/M decreased from 0.22 to 0.11 g substrate/g biomass h as the biomass increased from 3 to 6 g/L in a coagulant assisted thickened recycled sludge. Chaze and Huyard MBR Reduced sludge production at SRT 100 days compared to 50 days. (1991), Cicek et al. (2001) MBR Similar results at long SRTs ranging from 2 to 30 days in MBR treating synthetic wastewater without causing any significant influence on COD removal and nitrification. Mu¨ller et al. (1995) MBR Reduced sludge to just 6% of the carbon fed to a cross flow MBR when the sludge concentration increased to 40e50 g/L. However, inorganic fraction in sludge increased to 23.5% from 21.6%. Rosenberger et al. MBR Low sludge production (0.002e0.032 kg/d) without sludge discharge for one year at 15e23 g SS/L and (1999; 2000) low F/M ratio of 0.07 kg COD/kg MLSS d. These studies mentioned maintenance metabolism as the main cause and ruled out role of protozoa/metazoa. Deterioration of sludge properties mainly; fragile flocs, viscous sludge, high SVI, poor settling, difficult dewatering. High SRT caused poor oxygenation, increased aeration cost and membrane fouling. Laera et al. (2005). MBR Near zero sludge yield as mmax/b approaching unity (¼ 1.5 0.6) at 41000 d SRT in 180 days with constant VSS of 16e18 g/L, at an OLR below 0.1 g COD/g VSS d and SOUR of 2e3 mg O2/g VSS h, with complete nitrification and good COD removal. Lobos et al. (2008) MBR and OSA Clubbing of uncoupling high F/M ratio (upto 12 g COD/g MLVSS d) and OSA in two immersed MBRs, one operating in sequential and another in continuous mode. The biomass production was half in the continuous MBR operation, because of substrate limitation. Sun et al. (2007) MBR High cell maintenance resulted low sludge yield and high decay rate of 0.115 g VSS/g COD and 0.024/ day, at higher sludge concentration of 14.5 g/L at prolonged SRT for 300 days. 99% removal efficiencies of both COD and TOC were attributed to the bio-fouling layer, which played as a prefilter to the membranes. MLVSS/MLSS ratio of 0.9 indicated that no accumulation of inorganic compounds occurred at high SRT due to production of larger than pore size hydrolysis or enzymatic solubilization compounds. This was inconsistent with other reported MBR systems (Huang and Qing, 2001; Han et al., 2005). Low SOUR at high SRT indicated lower energy requirement by the microorganisms due to maintenance metabolism. It was considered a controversial phenomenon owing to no reduction in the biological capability of the sludge despite changes in composition of mixed liquor caused by prolonged SRT. Heran et al. (2008) MBR Complete sludge retention is susceptible to induce high biomass concentration and consequently low F/M ratio, fed with easily biodegradable organic substrate. High efficiency, no accumulation of mineral solids and no decline of the membrane performance were observed. The yield reduction was 0.041. The use of ASM model provided a good simulation without integrating inert soluble COD produced by microorganism activity ( fps ¼ 1.1%). But the model overestimated the oxygen demands due to high lysis products, at high SRT. Wang et al. (2006) Nylon mesh Study was conducted on SBR equipped with a 100 mm pore size nylon mesh on a stainless steel frame Bioreactor and spacer, capable to increase MLSS upto 32 g/L at 10 days HRT resulting in 83.9% decline in excess sludge. The results indicated low SS, COD and colour. However, the decantation period was influenced by the extracellular polymers of the microbes in the reactors. Xing et al. (2006) Inclined-plate Another version of MBR (0.4 mm pore size, Mitsubishi Rayon SUR134 membranes) used to improve the MBR (iPMBR) anoxic condition by confining high MLSS sludge. The performance of a pilot iPMBR treating municipal wastewater was investigated at an HRT of 6 h for. The respective average removals of COD, ammonia nitrogen and turbidity were 92.1, 93 and 99.9% at 6 h HRT without any sludge discharge in 123 days. Xie et al. (2008) MBR with simple A submerged flat metal aerobic membrane bioreactor performed reasonably well by online back sponge cleaning flushing or simple sponge scrubbing and the system ran stably about 115 days at the permeate flux of 0.8e1 m3/m2$d without changing the membranes. The results showed that the mean COD and TN removal efficiencies were 96.69% and 32.12% under aerobic MBR mode, and 92.71% and 72.44% under A/O-MBR mode. A/O-MBR mode also resulted in perfect sludge reduction, but with serious irreversible membrane fouling. Lin et al. (2009) Gravel contact The GCOR provided an integrated aerobic, anaerobic and anoxic environment to form the oxidation reactor substrateebacteriaealgaeeprotozoaemetazoan food chain. The reduced excess biomass took place (GCOR) due to grazing by protozoa and metazoan on dead bacteria.
high SRT in the MBR increases fouling and energy demand and decreases sludge production. While less SRT decreases MLSS and associated energy consumption but increases sludge production, which further require energy for its treatment.
Therefore, MBR operation could be optimized between these two parameters on account of energy consumption and membrane fouling. Apart from suspended and colloidal fractions of the waste, autochthonous (i.e. microbial in origin)
4296
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 2 8 7 e4 3 1 0
organic solutes either EPS or SMP are the key foulants. EPS are extremely heterogeneous, encompasses carbohydrates, proteins, nucleic acids, (phospho) lipids and other polymeric compounds found in the intercellular space of microbial aggregates. The carbohydrate components of the SMP in some MBR membrane fouling studies have been correlated to promote fouling under certain conditions (Simon Judd, 2007). However, the methods used for assaying carbohydrates and proteins during EPS fractionation are not universally agreed, and consequently different studies have identified different components with the highest fouling propensity. Rosenberger et al. (2006) suggested that at lower SRTs it is the polysaccharide colloidal matter in the SMP that is primarily responsible for fouling. But, Drews (2007) concluded that fouling cannot be attributed to any one specific constituent of the mixed liquor for higher SRTs (>20 days). Drew’s review (2010) linked fouling complexity on many other parameters such as Calcium concentration, pH, temperature, which affect properties of SMP compounds which are relevant for fouling, like size, shape, charge, gelling potential and hydrophobicity. It is not only the supernatant SMP concentration that appears to be a generally useful indicator for fouling propensity, but spatial distribution of polysaccharides in the biocake on membranes has also to be taken into account. In order to universal application of MBR technology, a standardization of methods and an agreement on “good operating practice” is highly desirable. Use of modified biopolymers with a net cationic charge could also be a possible remedy to fouling. Membrane Performance Enhancer (Nalco’s patented Perma care Product) was successfully investigated while treating different wastewaters viz. Leachate, Food industry, Dairy industry/cheese industry and Paper mill (Wozniak, 2010). The advantages offered by performance enhancer are; reduction of the transmembrane pressure, improvement in the filtration rate, increasing the permeability, reduction of foam, reduction of colloidal EPS, improvement in the effluent quality, reduction in cleaning frequency, increase in the life of the membranes and no impact on the oxygen consumption. Consequently, it can save about 15e20% of the investment costs. This technology is not to clean the membranes, but it keeps the membranes clean during operation and increases the filterability of the cake layer, however, first a chemical cleaning must be carried out (Wozniak, 2010). The foregone discussion indicates that the main stumbling block in sludge thickening and reduction appears to be the membrane structure and their fouling. An excellent review covering almost all the aspects of membranes materials and their fouling, which is an important key to their application for the intended use of wastewater treatment including sludge reduction has been done by Meng et al. (2009). It was one of the top 25 hottest article of Water Research, theEarth and Planetary Sciences journal from April to June 2009. Membrane fouling and the high cost of membranes are main obstacles for wider application of MBRs. Over the past few years, considerable investigations have been performed to understand MBR fouling in detail and to develop high-flux or low-cost membranes. In this paper, recent advances of research on membrane fouling and membrane material in
MBRs were reviewed. From the viewpoint of fouling components, the fouling in MBRs is classified into three major categories namely; bio-fouling, organic fouling and inorganic fouling. The results obtained from recent investigations on bound EPS, SMP, filamentous bacteria and hydrodynamic conditions are updated. The article is itself a review and hence cannot be pre´cised further, a detailed reading is advised, and however, the areas suggested for future study on membrane fouling are reproduced below: (1) Studies on membrane fouling mechanisms should focus on identification and characterization of membrane foulants (i.e. chemical and biological components of foulants, bacteria community of the foulants). Cake formation, pore blocking, and (EPS)/SMP adsorption on/within the membranes are all important, but emphasis could be given to the interaction and interrelation between these mechanisms and sludge characteristics. (2) Development of procedures for the visualization and characterization of membrane fouling in MBRs. Direct monitoring and in-situ techniques will offer more useful information about the formation of membrane foulants. (3) Standardization of fouling characterization methods and an agreement on “good operating practice” is highly desirable. (4) Development of more effective and easy methods to control and minimize membrane fouling. Generally, removable fouling is controlled by creating shear stress on the membrane surface. Although air bubbles are used to promote shear stress and to enhance the membrane flux, they also have strong impact on biomass characteristics. Moreover, enforced aeration will need more energy. Research should be directed to optimization of the current coarse aeration methods for submerged membrane modules. Lastly, alternative filtration concepts to limit the deposition of foulants onto the membrane surface should also be developed. (5) Study of the fouling behaviour in full-scale MBR plants in order to reflect the real fouling behaviour. (6) Development of novel membrane modules for MBRs to reduce their capital costs and enhance their hydrodynamic conditions. (7) Modelling of mass transfer and membrane fouling by mathematical approaches such as computational fluid dynamics, Monte Carlo simulation, fractal theory, artificial neural network. In other words, a comprehensive investigation should be performed to understand, control and reduce membrane fouling, especially avoiding severe fouling; it is just like a systematic physical examination on a person to understand his/her health condition and to avoid the occurrence of illness, especially fatal diseases. In recent years, there are considerable investigations about the impacts of membrane materials, pore size, hydrophilicity/ hydrophobicity, etc., on membrane fouling; however, most of the recent investigations are focused on the application of low-cost filters to substitute the membranes, modification of membranes to enhance their hydrophilicity and use of dynamic membranes to improve the performance of membranes or low-cost filters. In the future, to our knowledge, the study on membrane materials in MBRs should still focus on development of anti-fouling membranes or
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 2 8 7 e4 3 1 0
modification of current membranes and enhancement of the performance of low-cost filters by modifying their surface properties (Meng et al., 2009). Since number of variables is high in every version of MBR application, therefore, a technique was needed to assess the performance on a common basis. Madoni’s (1994) Sludge Biotic Index (SBI) is the one, which is widely used specific biotic indexes to assess and predict the performance of the treatment system, particularly in view of the abundance and diversity of microfauna (Are´valo et al., 2009). In this study, the SBI index was applied to a pilot-scale membrane bioreactor equipped with polyvinylidenefluoride hollow fibre membranes (0.04 mm pore size). Two different sludge retention times (SRT) of 25 and 35 days were assayed, with a constant hydraulic retention time of 30 h. The experimental results showed a constant predominance of small flagellates, carnivorous ciliated protozoa and rotifers in assays with 35-day SRT, independently of effluent quality. However, continuous changes in micro-biota were observed, with a clear tendency for the SBI to increase over time as the sludge became more stable. Therefore, although it was not possible to establish an association between SBI and effluent quality for MBR activated sludge, the stabilization of sludge may be related to SBI and also in the identification of problems arising in the process, such as incorrect oxygenation and the presence of toxic substances. For this reason, there is a need for further experimentation and adaptation of this controlling index, in order to guarantee its successful application to MBR systems (Are´valo et al., 2009). The gravel contact oxidation reactor (GCOR) filled with crushed stone globular aggregates as carriers, has been demonstrated capable of reducing the excess sludge effectively through maintenance metabolism in some pilot and small scale engineering studies. In order to evaluate the variation and structure of the microbial community and their functions to excess sludge reduction in GCOR, a conventional activated sludge reactor was studied as a comparison (Lin et al., 2009). The best part of the article is a clear picture of phylogenetic diversity and population of microorganism grew in the porous carriers and on their biofilm in a GCOR, which was little known in-situ until then. Out of the 30 species grew on GCOR media the most abundant bacteria were those related to the b-Proteobacteria group followed by those related to g-Proteobacteria and then those related to phylum CFB. In the CAS the order was g-Proteobacteria, followed by beProteobacteria and then CFB. Shannon’s diversity index (H ) was as higher as 3.41 for diversity of bacteria extracted from carrier samples in GCOR than 2.71 for sludge sample in CAS. Species evenness (E, the relative abundance or proportion of individuals among the species) for the isolates from GCOR and CAS samples was 0.97 and 0.96, respectively. The total bacterial DNA concentration at normal operation on the carriers of GCOR was 8.98 105 mg/mL, about two times more than in CAS 4.67 105 mg/mL, which exhibited higher enrichment of bacteria. At the same time, the most representative eukarya were protozoa both the reactors, which were 15 no. per 20 mL in GCOR and 15 no. per 20 mL in CAS, next abundant group were attachment plants 10 no. per 20 mL in GCOR and 4 no.
4297
20 mL in CAS, respectively. Rotifers and copepoda belonging to metazoan were only present in GCOR (8 no. per 20 mL for rotifers and 8 no. 20 mL for copepoda). MLSS in the sediment tank of GCOR was only 4.5 mg/L. It was 25 times less than 115.4 mg/L in CAS at the time of two reactors’ normal operation. Microbial diversity (H) and their population difference (E) both in the carriers of GCOR and sludge of CAS indicated that diverse microbes and a large amount of biomass attached on the carriers were probably one of the main reason of excess sludge reduction in GCOR, where microbial community varied at different stages of the microbe incubation and normal wastewater treatment (Lin et al., 2009). It appears more genomics related studies are needed to focus on the phylogenetic diversity and abundance at different stages of bioreactors, and the microbial growth rate in association with substrate degradation, so as to more accurate exploration of microbial ecology fundamentals on the subject of excess sludge reduction in the bioreactors. The predominance of bProteobacteria as major constituents of the microbial community structure was also cited by Zubair et al. (2007).
2.3.1.
Cost and energy considerations in MBR
Information on energy consumption and consequently cost analysis of MBR plants is scarce in scientific literature. The sludge treatment cost minimization increases the aeration cost, therefore, optimum point could be identified between these two parameters. A methodology to obtain the most economical operational condition of membrane bioreactor was developed by Yoon et al. (2004), through which sludge production rate can be quantitatively estimated as functions of HRT and MLSS. When either target MLSS in bioreactor or HRT increases, sludge production rate decreases and aeration requirement increase. By summing the decreasing sludge treatment cost and increasing aeration cost, total variable operational cost is obtained. The energy costs in a Kubota flat sheet MBR pilot plant located in Southeast Spain were monitored at two fluxes, 19 and 25 L/m2.h for one year (Gil et al., 2010). Based on electricity price of 0.0806 V/kW, the prices of the treated water for the two fluxes were 0.49 and 0.39 V/m3 and the total consumptions were 6.06 and 4.88 kW h/m3 for above fluxes respectively. This energy consumption was on a much side as compared to reported value of 0.6e0.8 kW h/m3 of treated water. Coarse bubble aerator, followed by the mixer consumed about 50% and 25% of the total energy. But, the basis of energy data and cost evaluation has not been properly spelt out and statistically correlated in the paper. Fenu et al. (2010) validated a calibrated dynamic biological ASM model of a full scale MBR and CAS reactors in the form of two lanes at the treatment plant of Schilde. The overall energy consumption found to be 0.64 kW h/m3 of permeates in MBR lane in comparison to 0.3 kW h/m3 in CAS lane. The higher MLSS maintained in MBR implied significant mixing energy costs and reduced oxygen transfer, the smaller floc size did not reflect in a significant aeration energy saving. The impact of the filtration process on the overall energy consumption could be reduced if the coarse aeration flow would be better integrated in the biological process scheme of submerged MBRs. However this work indicated a minimal contribution of the coarse aeration flow to the biological oxygen requirements. Roels et al. (2010) identified
4298
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 2 8 7 e4 3 1 0
the following main energy consumers of the CAS system; 35e50% of the total energy accounts for coarse bubble aeration, 15e20% for influent flow, 10e15% for sludge recycle 5e15% for mixers. The overall filtration process, including the permeate extraction, the electrical heating of the cleaning in-place tank, the compressors for activation of the valves, accounted for more than 56% of energy consumption. The average energy cost of some municipal full scale MBRs ranges between 0.8 and 1.2 kW h per m3 (Mulder, 2009). The CAS in Schilde (Belgium) consumes 0.19 kW h per m3, whereas the MBR consumes 0.64 kW h per m3. The lower energy consumption in Schilde might be explained by an intensive cost reduction program started in 2006 (Garces et al., 2007). Judd (2007) prescribed two given below options for reducing energy demand in MBRs while reviewing fundamental facets of the MBR process; (i) use of ceramic membranes and (ii) anaerobic operation. Ceramic membranes are more fouling-resistant but are currently high in cost, which may expectedly be reduced with time owing to advances in fabrication techniques (Bishop, 2004). Submerged anaerobic MBRs may offer advantages over aerobic treatment because anaerobic operation demands least energy input in addition to energy generation via methane gas recovery. However, the advantages offered by this coupling of an anaerobic process with a membrane have not been quantified (Jefferson, 2007; Hu and Stuckey, 2006; Jeison, and van Lier, 2007). On account of cost of membranes, frequent replacement of membranes due to fouling, increased aeration owing to poor oxygen transfer, and taking the costebenefit scale into consideration, it is not always feasible to operate MBR with complete sludge retention or zero sludge bleeding in practice, and there must be a minimal rate at which excess sludge is wasted in order to keep an optimal range of sludge concentration in MBR. Moreover, the full-scale application documentation is only limited to one experience and still more studies on cost benefit analysis would justify MBR application.
2.4.
Eco-manipulation (predation on bacteria)
The principle of using microfauna to reduce excess sludge comes from the food chain. Materials and energy are dissipated (or lost) when they flow in the food chain, and hence the microfauna’s predation leads to sludge reduction. Most of the protozoa are aerobic organisms feed on organic sources including bacteria by the process called phagocytosis. Initially, their role was identified in effluent polishing by consuming dispersed bacteria by Curds and Cockburn (1970); Learner, 1979 and Gaudy and Gaudy (1981). Subsequent researchers have shown reduction in sludge production due to predation of bacteria by exploiting higher organisms such as protozoa and metazoa in the different variants of activated sludge process (Ratsak et al., 1996; Ratsak, 2001; Welander and Lee, 1994; Lee and Welander, 1994, 1996a,b; Rensink et al., 1996; Janssen et al., 1998; Ghyoot, 1998; Eikelboom, 1988). Rensink and Rulkens (1997) showed the grazing fauna in aerobic reactors as an Eltonian pyramid of numbers (Elton, 1935). The ecosystem in a biological wastewater treatment plant consists of bacteria, protozoa, metazoa, larvae of insects and arachnida (Curd and Hawkes, 1975), where bacteria are
the primary consumers, which themselves are consumed by protozoa and metazoa. It creates a food chain between each trophic level, linked by a predatoreprey relationship. During energy transfer from low to high trophic levels, energy is lost due to inefficient biomass conversion. Under optimal conditions the total loss of energy will be maximal and the total biomass production will thus be minimal Ratsak et al. (1996). The protozoa can be classified as ciliates (free swimming, crawling and sessile), flagellates, amoeba and heliozoa (Eikelboom, 2000). The metazoa consist normally of rotifers and nematode and occasionally Aeolosomatidae and Naididae. The predator population is normally dominated by protozoa in activated sludge system and metazoa in trickling filters. The presence of predators suppresses the growth of dispersed bacteria and results in favour of floc formation. As a result, major portions of the sludge remain unaffected by predation activity. This was overcome by making it a two-stage process by providing a short SRT completely mixed pretreatment tank to encourage the growth of dispersed bacteria in the absence of predators. Contrary to the conditions of the first stage, the second tank was conventional aeration tank with very high SRT (may be membrane assisted) to encourage growth of predators and consumption of earlier produced dispersed bacteria (Ratsak et al., 1994; Lee and Welander, 1996a; Ghyoot and Verstraete, 2000). In addition to the use of protozoa and metazoa as important indicators of process performance and efficiency of biological wastewater treatment processes, recently, many researchers have explored the potential of different kind of predators by manipulating the ecosystem as a biotechnological tool (Eikelboom, 2000; Eikelboom et al., 2001; Zhang, 2000; Luxmy et al., 2001; Lapinski and Tunnacliffe, 2003 and Wei et al., 2003b).
2.4.1.
Two-stage system
As mentioned above about the requirement of two stages for effective sludge reduction by predators, Ratsak et al. (1994) reported 12e43% of biomass reduction by employing the ciliated Tetrahymena pyriformis to graze on Pseudomonas fluorescens. Lee and Welander (1994, 1996a,b) tested the decreasing trend in sludge production, which was mediated by protozoa and metazoa. The high sludge yield of 0.17 g TSS/g COD removed in the acetic acid fed system was observed as the bacteria formed zoogloeal flocs, which protected them from grazing in the predator stage. Whereas, low sludge yield of 0.05 g TSS/g COD removed was observed in the system fed methanol as dispersed bacteria were obtained that were easily grazed by the protozoa and metazoa in the predator stage. Due to the mineralization activities of the predators, a significant release of nitrate (>7 mg N/L) and phosphate (>2.5 mg P/L) was observed in the effluent (Lee and Welander, 1996b). Similar study was performed on pulp and paper industry wastewater in a two-stage arrangement in both activated sludge and biofilm reactors were tested in second stage (Lee and Welander, 1996a). Results showed lower sludge yields of 0.01e0.23 g TSS/g COD removed as compared to 0.2e0.4 g TSS/ g COD removed in CAS processes treating the same wastewater. Growth of fast growing filaments, such as Sphaeroltilus natan and the formation of aggregates were encountered in the first stage, despite that their numbers were under control
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 2 8 7 e4 3 1 0
4299
Table 3 e Summarized experiences of sludge reduction by predation on bacteria. Reference(s)
Process
Lapinski and Tunnacliffe (2003)
Predation
Luxmy et al. (2001)
Predation
Rensink and Rulkens (1997) Predation and Ratsak et al. (1993)
Ratsak (1994)
Predation
Zhang (2000) and Eikelboom et al. (2001)
Predation
Wei et al. (2003b)
Predation
Elissen et al. (2006)
Predation
Hendrickx et al. (2009a)
Predation
Hendrickx et al. (2009b)
Predation
Liang et al. (2006a)
Predation
Highlights Bdelloid rotifers (‘‘leech-like wheel-bearers’’) are metazoans, which are ubiquitous in most fresh and wastewater habitats. A corona of cilia around the mouth opening draws suspended particles into mouth of rotifer. During grazing they consume several times their body weight per day. In addition they cause enhanced ‘‘bioflocculation’’ or a combination of both. The sludge reduction capacity of metazoa in an MBR and found insignificant sludge reduction at a metazoan population of 1000e2000 ind. per mL, instead greater abundances of metazoa were required, but metazoans were found to play an important role in fouling control of the membranes. In the activated sludge plants the ciliated protozoa were the dominant microfauna group in both the treatment systems, whereas metazoa, particularly Lecanidae rotifera were more abundant in the starch-enriched system. No significant differences in sludge production were found between the two systems. This was probably because metazoa levels in either system never exceeded more than 1500 ind/mL. Although the higher levels of metazoa in the starch-enriched system did not affect sludge settleability or production, they did have a great effect on the nitrogen removal efficiency (Puigagut et al., 2007). Predominance of oligochaetes worms especially Nais elinguis and Aeolosoma sp. and darkness preferring Tubificidae was observed. However, uncontrollability of Aeolosomatidae and Naididae i.e. their washout in effluent led to selection of Tubificidae for sludge reduction on different carriers. Two trickling filters one with lava slag and another with plastic media caused sludge reduction by 10e50% and 10e45% in the trickling filters compared with 10e15% and 10% without worms. The resulting sludge yield with and without Tubificidae was to 0.15 g MLSS/g COD and 0.40 g MLSS/g COD respectively. 25e50% sludge reduction by predominantly Nais elinguis followed by Pristina sp. and Aeolosoma hemprichicii worms was observed in one and a half year study. The number of worms varied both seasonally and among the aeration tanks. A major worm bloom resulted in a low SVI, lower energy consumption for oxygen supply and less sludge disposal. Sludge reduction was achieved using membranes to increase worm density. They found three species of worms namely; N. elinguis, Pristina sp. and Aeolosoma hemprichicii, but Aeolosoma sp. was predominant. It was contrary to the finding of Ratsak (1994) obtained in CAS. Wei et al. (2003b) found high worm density i.e. 2600e3800 Aeolosoma/mL mixed liquor in the MBR system, and resulted in a low sludge yield (0.10e0.15 kg SS/kg COD removed). It was found that worm growth in the CAS reactor was much better than in the MBR as the average worm density in the CAS reactor was 71 worms/mg VSS than 10 worms/mg VSS in the MBR. The prevalence of worms was 30 worms/mg VSS in CAS upto 172 days, however, worms did not produce naturally in MBR and their presence depend on sludge inoculation from the CAS. The impact of TSS, HRT, SRT, F/M; recycle ratio, temperature, pH and DO on the growth of worms in both the reactors was also correlated. Only sludge loading rate and SRT had no impact on the growth of worms in the MBR and CAS reactor, respectively. The separation of waste sludge and worm faeces was made possible with a new reactor concept in which L. variegatus was immobilized in a carrier material. This also eliminated the need to separate the worms from the sludge. The reactor concept consisted of a sludge compartment containing both waste sludge and worms. The open bottom of the sludge compartment was covered with a carrier material (polyamide mesh, 300 mm; with a surface area of 7.5 cm2) was used, through which the worms can protrude their tails. It was placed partially submerged in the SBR. L. variegatus respires and defecates via its tail, as a result; the worms kept their heads in the sludge compartment and protruded their tails into the aeration tank. Out of the total consumed waste sludge, 75% reduction in the amount of TSS was observed in addition to the natural sludge breakdown. Thus, carrier material acted as both a support material for the worms and a separation layer between the waste sludge and the worm faeces, which was beneficial to further processing. The effect of changes in DO concentration, ammonia concentration, temperature and light exposure on sludge reduction by L. variegatus in same reactor configuration used by Elissen et al. (2006) was studied to optimize the reactor. Sludge consumption rate was four times higher at DO above 8.1 mg/L, as against below 2.5 mg/L. Sludge reduction was 36 and 77% at these respective DO concentrations. Similar study was repeated by applying in continuous operation on increased mesh size to 350 mm over sludge. TSS reduction of 16e26% by the worms was achieved (22e30% VSS reduction). Mass balances showed that the worms contributed 41e71% towards total VSS reduction. The rest was caused by natural sludge breakdown. Very small Aeolosoma hemprichi worm (about 1e2 mm in size, doubling time about 1.4 days) grew at constant rate when the sludge concentration was higher than 300 mg VSS/L, higher growth rates of A. hemprichi were observed in higher protein containing and smaller particle size sludges. The sludge yield had negative correlation with the density of A. hemprichi regardless of differences in SRTs or F/M ratios. The relative sludge reduction by A. hemprichi was about 39e65% and in terms of per unit weight of A. hemprichi it was 0.53e6.32 mg VSS/mg A. hemprichi day. Stabilize the sludge settleability and total phosphorus removal but did not affect COD and NH4þ-N removals in the process. The CAS reactor consisted of three parts namely; aeration tank, transition region and settling tank, any effect of transition region on worms other than as a buffer for settling zone has not been spelt out in the paper. (continued on next page)
4300
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 2 8 7 e4 3 1 0
Table 3 (continued ) Reference(s)
Process
Song and Chen (2009)
Predation
Huang et al., (2007)
Predation
Wei et al. (2009a)
Predation
Highlights Different results were obtained while stabilizing the growth of A. hemprichi and minimizing the sludge production in batch and continuous experiments. A. hemprichi reached the maximum density when available VSS concentration was more than 3000 mg/L. No obvious difference was found between the initial specific growth rates (m) of A. hemprichi at various available VSS concentrations. Sludge reduction rate was correlated with both the growth rate and the density of A. hemprichi. The results indicated the sludge reduction rate was maximum at the density of 315 ind./mL. SRT more than 15 days did not affect the growth. Similar CAS reactor configuration which Liang et al. (2006a) used with A. hemprichi, was inoculated with oligochaete Tubifex tubifex. The results showed that the sludge reduction rate by T. tubifex (in terms of per unit dry mass of T. tubifex) was from 0.18 to 0.81 mg-VSS/mg-Tubifex.d. The sludge reduction capacity of the recycled sludge reactor was from 650 to 1080 mgSS/L.d. The optimum density of T. tubifex was 2500 mg/L and the optimum sludge recycled ratio was 1. The existence of T. tubifex did not affect COD and NH4þ-N removals in the process, but led to a slight decrease in TP removal, contrary to the findings of Liang et al. (2006a). SVI almost did not change when the T. tubifex density was lower than 3300 mg/L. A combination of a Tubificidae (known as Naididae) reactor with an integrated oxidation ditch with vertical circle (IODVC), was put to investigate excess sludge reduction (first stage) and returned sludge stabilization (second stage) as a new integrated system, dominantly containing Branchnria Sowerbyi worm after inoculation of Branchnria Sowerbyi and Limnodrilns sp. The maximal volume density of wet Tubificidae in vessels of the Tubificidae reactor was 17600 g/m3. The results showed that the excess sludge reduction rate was 46.4% in the first stage and the average sludge yield of the integrated system was 6.19 105 kg SS/kg COD in the second stage. Though the sludge returned to IODVC via the Tubificidae reactor, it had little impact on the effluent and the sludge quality.
in the second stage as compared to CAS. Different microfauna were observed under different set of conditions. Therefore, use of protozoa and metazoa for full-scale application for sludge reduction demands more understanding of microbial ecology of the system in question to ensure better operation and process control. Cech et al. (1994) also related a concomitantly poor phosphorous removal in a single stage laboratory scale reactor for a mixed population with a marked increase in predator numbers. The performance of CAS reactor and a submerged MBR as second stage was compared by Ghyoot (1998) in his thesis and later publication (Ghyoot and Verstraete, 2000). The sludge yield of the two-stage submerged MBR system was 20e30% lower than that of the two-stage CAS system under similar condition of SRT and F/M ratio owing increased grazing of predators in the two-stage MBR than those in the CAS reactor. However, the increased grazing on nitrifiers not only decreased the capacity of nitrification, but also resulted in high N and P concentration built-up in the effluent. MBR configuration also resulted in higher levels of soluble but poorly biodegradable COD. Subsequent research encounters with predators are shared in Table 3. It has been observed that there is a shift on research related to sludge reduction by predators towards oligochaeta rather than protozoa and metazoa mainly due to process uncertainties.
2.4.2.
Oligochaeta
Possessed with advantages such as low cost and no secondary pollution, worm (Oligochaete) technology, based on using microfauna’s predation to reduce excess sludge has recently begun to receive increasing attention by the researchers. It is a subclass in the biological phylum Annelida and includes various earthworms. Worms are the largest organisms found
in activated sludge and are already in application to regulate sludge. However, Wei et al. (2003b) mentioned that a practical application is still uncontrollable as there is no clear relationship between process conditions (e.g. retention times, temperature, sludge loading rates and shear forces) and worm growth. They state that one of the challenges is to maintain high densities of worms for a long time, particularly in fullscale applications. The main types of worms present in activated sludge system and trickling filters are Naididae, Aeolosomatidae and Tubificidae. Ghyoot and Verstraete (2000), Ratsak et al. (1996), Ratsak and Verkuijlen (2006), Rensink and Rulkens (1997), and Salvado et al. (1995) have performed research on oligochaeta to put forward advancements in handling and reduction of the excess sludge. Wei et al. (2003b) proved the effectiveness of using microfauna to reduce excess sludge, but the results were negatively correlated based only on sludge yield and density of the microfauna. As the sludge yield is easily affected by environmental and operational factors such as temperature, classes of bacteria, SRT and F/M ratio, therefore, any correlation excluding all these factors could be misleading on reduction scales obtained by comparing the correlation of sludge yield and the microfauna’s density alone. However, Luxmy et al. (2001) suggested that metazoa could not reduce sludge production in an aeration tank. Elissen et al. (2006) opined that microfauna could reduce sludge, but argued about which species of microfauna could reduce sludge best. These contradictory opinions were related to the lack of effective methods of detecting the rate of sludge reduction caused by microfauna. Therefore, it is important to accurately determine the rate of sludge reduction so as to compare and choose types of microfaunas as sludge predators. Table 3 gave a precise account of oligochaetes worms.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 2 8 7 e4 3 1 0
A dynamic energy-budget model was extended to describe the growth of Nais elinguis, a common oligochaete species frequently occurring in sewage treatment plants. N. elinguis reproduces asexually by dividing into an anterior and a posterior naidid. The daughter naidids initially have different growth rates due to the differences in energy reserves. The parameters of the model, such as maintenance rate coefficient and energy conductance, which represent physiological features of the organisms, facilitate insight into the processes underlying naidid growth. A numerical method was developed to predict asexual reproduction. The presented model was an important step towards a more comprehensive knowledge and modelling of organisms in activated sludge plant (Ratsak et al., 1993). On comparing the results obtained from sequencing batch reactor (SBR) with Lumbriculus variegatus worms immobilized on a carrier material and without worms, it was almost three times less sludge production than in the blank experiment. The worm faeces that are produced after sludge predation have an SVI that was approximately half of the initial waste sludge, capable to settle much faster than the initial waste sludge. Since, conditions beneficial to predator growth may not be optimal for bacterial growth and overall process, Lee and Welander (1996a) separated them into two-stages. The same principle was also applied with aquatic worms by Elissen et al. (2006), Table 3 describe the conclusions of studies reported by Hendrickx et al. (2009a), Liang et al. (2006a), Song and Xiaofei, (2009), Huang et al., (2007) and Wei et al. (2009a) The TSS reduction of 16e26% (Hendrickx et al., 2009b) was much lower than the 75% (Elissen et al., 2006) and 36% (Hendrickx et al., 2009a) with sludge from a municipal sewage treatment plant mentioned in Table 3. This effect may be attributed to the origin of the sludge and timings (same system, but during different months of the year) due to varying nutritional value of sludge for the worms. Recently, the results described above were summed up to give meaningful rector design parameters of a sequencing batch worm rectors using carrier materials with 300 and 350 mm mesh sizes (Hendrickx et al., 2010). The surface specific consumption rates were 45 and 58 g TSS/m2 d, respectively. The author mentioned 29% smaller reactor compared to using a 300 mm mesh, which is in fact 22% mainly due to 25% higher worm density of 1.1 kg ww/m2. To avoid substrate (sludge) limitation, a sludge load above 100 mg TSS/g ww d is recommended in the paper. Worm biomass growth rates were 0.026/ d for free worms and 0.009e0.01/d for immobilized worms on carrier material. Similarly, the decay rates under free and immobilized conditions were 0.018/d and 0.023/d, respectively. The SOUR of the worms required to supply and maintain desired oxygen level in the reactor is mentioned as 4.9 mg O2/g ww d, which amounts to 0.5e1.0 mg O2/mg VSS digested. Still the worm reactor size depends on the type of receiving sludges. Parts of nutrients are incorporated into the biomass during synthesis and then withdrawn with excess sludge. During sludge mineralization, increase of phosphorus, nitrogen, CO2 and even dissolved COD in effluent in the form of SMP is always an even possibility. The same is reported in the recent researches of sludge reduction induced by oligochaeta. Batch test and radioisotope 32P tracer test were carried out to further
4301
investigate nutrients release and phosphorus distribution among supernatant, sludge and worm during predation of oligochaeta on sludge (Wei et al., 2009b). The radioisotope 32P tracer was transferred step by step from synthetic wastewater to worms through a food chain (activated sludge cultivated by synthetic wastewater with exogenous Na2H32PO4, and predation of worms on activated sludge). Radioisotope activities for the radioisotope 32P tracer in supernatant, activated sludge and worms were then determined. Results showed that more nutrients release into supernatant occurred in the tests of worms with sterilized sludge than that of worms with activated sludge and release of nitrogen and phosphorus was few in the tests of worms with activated sludge. After 24 h 32P concentration of supernatant in the test of sludge with worms was 9% higher than that in the test of sludge without worms and 32P concentration of worm increased by 2.7%. Additionally, the release rate of phosphorus into supernatant caused by worm’s predation on activated sludge was 0.1211 mg TP/ worm (wet weight) h (Wei et al., 2009b). The optimum growth parameters of a number of oligochaeta have been cited by Inamori et al. (1983) and Kuniyasu et al. (1997), which are important for their comparative assessment. A comparative study was conducted to assess the potential of sludge reduction of Aeolosoma hemprichi, Daphnia magna, Tubifex tubifex and Physa acuta (Liang et al., 2006b). Since, Carbon accounts for more than 50% of all the ingredients, therefore, the rate of sludge reduction was correlated with the rate at which solid carbon form were changed into liquid and gas carbon form. The rates of sludge reduction using the four microfaunas were 0.8, 0.18, 0.54 and 0.1 mgsludge/mg-Microfauna/d, respectively, changing with the microfauna’s phylum or class and body size. The results were similar to those produced using the direct measuring method. The sludge reduction rate by T. Tubifex was not significantly affected by the difference in sludge, which came from different wastewater treatment plants in different cities. The microfauna in the class of Clitellata or subclass of Oligochaeta reduced the sludge more efficiently than those in the Crustacea or the Gastropoda classes. The sludge reduction rates of microfauna were related to their individual body size, even though they belong to the same phylum or class. For example, the sludge reduction rate of A. hemprichi was much higher than that of T. Tubifex because of their difference in individual weight. The smaller microfauna had higher sludge reduction rates due to their rapid metabolism. In addition, the amount of sludge reduced by microfauna was correlated to both the sludge reduction rate and the quantity of microfauna in the reactor. Though, the sludge reduction rate of A. hemprichi was higher than that of T. Tubifex, the quantity of T. Tubifex was more than that of A. hemprichi in the aeration tank (results of other experiments indicated 10e20 times higher quantity of T. Tubifex than that of A. hemprichi). Therefore, T. Tubifex reduced more sludge than A. hemprichi (Liang et al., 2006b). Biologically sludge reduction processes, highlight a weakness in the currently accepted approaches to modelling. Neither of these treatment processes can be accurately predicted with currently accepted simulators and models, such as International Water Association’s ASM Model (van Loosdrecht and Henze, 1999) and Anaerobic Digestion Model (ADM) (Batstone et al., 2002) series of models. To better
4302
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 2 8 7 e4 3 1 0
understand the mechanisms behind biological sludge reduction processes, a mathematical model similar to the ASM models (Henze et al., 2000) describing the interaction between nitrifiers, heterotrophs and predators in wastewater treatment has been developed by van Loosdrecht and coworkers (Moussa et al., 2005). The model includes six soluble compounds (DO, nitrogen gas, ammonia, nitrite, nitrate and COD) and five types of biomass (ammonia oxidizers, nitrite oxidizers, heterotrophs, predators and inert biomass) as particulate compounds. The kinetic expressions in the model are based on Monod equations. Two decay rates were considered in the model: aerobic decay occurred when the bacteria starved in the presence of oxygen, while anoxic decay occurred when the bacteria starved in the absence of oxygen and presence of NO3 (Leenen et al., 1997; Siegrist et al., 1999). Moreover, to assess the predator activity, a simple procedure was developed. The respiration under starvation condition with and without the presence of the predators was measured to determine the activity of predators. A shock load of NaCl was used to eliminate the predators. The impact of salt on the presence of the predators (protozoa, rotifers and nematodes) was followed in phase contrast mode using a light microscope. The minimum dose of NaCl and the contact time required for full elimination of the predators without affecting the activity of ammonia and nitrite oxidizers and of heterotrophs were 5 g NaCl as Cl/L and 1 h. The respiration activity of the samples treated with salt was estimated by measuring the OUR in a batch biological oxygen monitor. However, the use of salt as a selective inhibitor for predators under practical conditions demanded further investigations (Moussa et al., 2005). The results were then calibrated and validated on two laboratory scale SBRs operated at different SRT of 30 and 100 days for a period of 4 years. The fraction of active biomass (nitrifiers, denitrifiers and heterotrophs) predicted by the proposed model was only 33% and 14% at SRT of 30 and 100 days, respectively. The high fraction of inert biomass predicted by the model was in accordance with the microscopic investigations of biomass viability in both reactors. The results of this study presented the possibility for increasing the nitrification activity by suppressing the growth of predators in a nitrifying system or other systems in which slow-growing bacteria play an important role. The model results showed the need for careful optimization of systems operated at long SRT (such as membrane bioreactors), to avoid accumulation of high amounts of inert biomass and to avoid high operational costs without gaining any volumetric improvement. The model showed its capacity to elucidate the biological processes in activated sludge systems by including the effect of the predators. The practical application of the developed model and assessments of predators’ activity called for verification under full scale activated sludge plant operation (Moussa et al., 2005). The results of most of the studies on sludge reduction through worms are encouraging (Table 3), but still the control parameters of the system have not been established. Therefore, further research to correlate all the known parameters of wastewater treatment with the worms’ growth, density etc. in demonstration followed by full-scale needs to be undertaken.
2.4.3.
Filamentous fungi (FF)
Application of filamentous fungi is another possible area of eco-manipulation to preclude sludge growth. Fungi are saprophytic organisms and they obtain their nourishment from the degradation of dead organic matter. A great variety of fungi has been reported to be found in sludge, FF have been recognized for sludge treatment and possibly these strains can be utilized for simultaneous bioflocculation, solids and pathogens reduction and, removal and degradation of toxic compounds, a detailed account is given by More et al. (2010). The fungi treatment of sludge is less energy consuming. The oxygen supply needs of the microfungi are approximately one third of the oxygen requirement by a bacterial population. In addition, the microfungi will use all the forms of oxygen supply available in order to optimize the degradation of the organic matter. The fungi have advantages over bacteria because of fungi having capability to degrade more complex and variety of substrates. Certain fungi strains can also be chosen for their beneficial effects on the plants if treated sludge intended in the use as a fertilizer. Different fungi used for sludge treatment are; Phanerocheate chrysosporium (Alam et al., 2001), Mixed culture of Aspergillus niger and Penicillium corylophilum (Alam et al., 2003a, 2003b; Alam and Fakhru’l-Razi, 2003), Mixed culture of P. corylophilum WWZA1003 and Aspergillus niger SCahmA103 (Fakhru’l-Razi et al., 2002b), Mucor hiemalis (Fakhru’l-Razi and Molla, 2007), Aspergillus (Jamal et al., 2005), Aspergillus niger; Penicillium corylophilum (Mannan et al., 2005), Mixed cultures of Trichoderma harzianum with Phanerochaete chrysosporium 2094 and T. harzianum with Mucor Hiemalis (Molla et al., 2004), Penicillium expansum BS30 (Subramanian et al., 2006), P. expansum BS30 (Subramanian et al., 2008) However, the role of FF in sludge treatment is not well established and limited to shake flasks, which lack semblance to actual operating conditions. The all important selection of useful fungi and the effect of parameters such as dissolved oxygen, pH, temperature, agitation, suspended solid concentration, operating conditions and incubation time on fungal activity to achieve maximum solids degradation are yet to be established and therefore need further exploration. The specific studies on selection of fungal strains and influence of different concentration require attention. Therefore, extensive laboratory followed by pilot scale studies are needed for future applications of filamentous fungi as an eco-tool to sludge reduction.
3.
Discussion
Ecological imbalances are the fallout of economic and industrial development, the problem is particularly pronounced where the populations are large and increasing exponentially, for example the recent spur of development in Asia has its own environmental and social ramifications. The treatment and disposal of excess sludge is one amongst them. It is expected that by the year 2015, there will be more than 1000 operational municipal wastewater treatment plants (WWTPs) in China. In 2006, 39% of the Chinese population was served by municipal WWTPs, registering an increase by 18.2%, since the year 2000; while the wastewater treatment capacity has
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 2 8 7 e4 3 1 0
increased by 135%. In the year 2005, 297 cities in China had no WWTPs for treating wastewater. These facts indicate the potential growth of sewage treatment capability and sludge production along with economic and industrial growth (Idris et al., 2004; He et al., 2006; Jun et al., 2004b). Bhardwaj (2005) mentioned that about 26254 million litres per day (MLD) of wastewater were generated in the 921 towns in India (housing more than 70% of urban population) in 2004. It was only 18312 MLD in 644 cities in 1995. This reflects 30% growth of both urbanization and wastewater generation owing to industrial growth of the country. The municipal wastewater treatment capacity developed so far in India is about 7044 MLD e accounting for 27% of wastewater generation in urban areas (Bhardwaj, 2005; CPCB, 2004). These facts indicate the potential growth of sewage treatment capability and sludge production along with economic and industrial growth of these two economic power houses of Asia. It is further expected that the problem of sludge production will continue to increase with time due to increased social awareness and legislative concerns. As conventional or traditional methods of sludge reduction and disposal are under stringent control and almost phased out the immediate task is to find a novel cost effective and sustainable mechanism. Broadly speaking there are two approaches; online and offline, both with relative advantages and disadvantages. Varieties of possible techniques and their configurations have been and under investigation at different level, however, the research is in its infancy and therefore it is too early to come to a consensus. At best the relative merits and demerits can be precisely tabulated (Table 4). Cell fractionation is basically the rate limiting step in lysiscryptic growth and ozonation is the only successfully applied full scale process, however, enough scope still presents for the application of enzymatic hydrolysis, wet air oxidation and alkaline-ultrasound treatment. High rate oxygenation has limited application and efficiency, but it does not require any additional unit, however, process economy and energybalance needs further exploration. Non-chemically assisted uncoupling by virtue of high So/Xo and OSA process are promising efficient options. Their fullscale validation is still awaited. MBR is another technique which may have scope in future; however, further studies on membrane life, fouling and overall cost-benefit determination would further justify their large scale application. Sludge reduction by predation and oligochaeta is certainly environmental friendlier than others, but completely lack operational control from engineering point of view due to extreme uncertainties associated with them and keeping environmentalists away to apply on full scale. In summary, OSA and MBR are two such processes which could be short listed owing to their merits on account of friendliness to environment, defined mechanism of energy consumption in increased cell maintenance and flexibility in operation due to easy control. However, still the most important limitation in application of in-place sludge reduction is undefined support of process mathematics and subsequent modelling which is keeping us guessing and inviting for further research. The emphasis on in-place sludge reduction creates a unique or rather conflicting situation in simultaneous removal of organics, nitrogen, phosphorus and biomass as;
4303
Physiology of nitrifiers depends on high DO, while for denitrifiers it is low DO. High SRT required for organics and ammonia oxidation and sludge reduction, while low SRT for phosphorus uptake. Requirement of Carbon source for nitrification, while absence of any donor other than PHA/PHB in enhanced phosphate uptake. pH and alkalinity reduces during nitrification and denitrification, while high pH and alkalinity required for chemical precipitation of phosphate. The synthesized biomass normally contains 2e3% P by its dry weight, stoichiometric inclusion of it into biomass yields it as C5H7O2NP0.1 giving 2.67% P (i.e. 3.1 100/116g ¼ 2.67), in case of enhanced biological phosphate removal it could be as high as 2e5 times the above (5.4e13.4% of the cell mass). Biomass uptake is the only mechanism of phosphate removal, although chemical precipitation during denitrification in the form of Hydroxyapatite {Ca5(OH)(PO4)3} and Struvite (Mg NH4PO4) has also been reported (Quintana et al., 2005). Whereas, nitrogen has other route of removal through denitrification in addition to biomass uptake which is around 12% of the cell mass. The pinnacle of all efforts of nutrient removal is practically put-down the moment biological cell are lysed, hydrolyzed or digested causing reappearance of nitrogen in general and phosphorus in particular as mentioned above. Therefore, the phenomenon of in-place sludge reduction along with simultaneous removal of organics, nitrogen, and phosphorus depends on extent of reconciliation of conflicts as cited above. The phenomenon of sludges in Europe and USA is almost under strict surveillance through regulations, but the scenario is grim and uncertain in Asia, in view of social, political and legal infirmities and legislative inadequacies. There are three laws in Chinese environmental legislations that contain restrictions on sludge management, namely; Control standard for pollutants in sludge for agricultural use, GB4284-84 (promulgated in 1984 but never amended), Criteria for controlling the discharge of sewage and sludge from municipal wastewater treatment plant, CJ3025-93, and Criteria for controlling the discharge of pollutants from municipal wastewater treatment plant, GB18918-2002. The last one concludes that all sludge should be stabilized and shall meet the control degradation indexes pathogen concentrations after stabilization (Tottie, 2007). Solid Waste Pollution Prevention and Control Law enacted in 1995 establishes a broad national framework for the management of industrial, municipal and hazardous waste, aiming to safeguard human health by means of preventing and controlling solid waste pollution (Beyer, 2006). But, the reality is that the construction of WWTPs often is built only to achieve emission standards and not sludge quality standards. Similarly, the Municipal Solid Wastes (Management and Handling) Rules (2000), Government of India, do not provide for specific eco friendly sludge treatment. Realizing the dynamic economic development and the evolving political system, India’s environmental future is difficult to determine. The situation is so grim that due the absence of any suitable legislation and implementation of existing provisions even basic data on sludge generation and its expected growth is
4304
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 2 8 7 e4 3 1 0
Table 4 e Relative merits and demerits of Sludge reduction techniques. Process Pure oxygen
MBR
Merits High DO-activated sludge process can repress development of filamentous organisms. Ability to maintain a higher MLVSS concentration in the aeration tank. Better sludge settling and thickening. Higher oxygen transfer efficiency. More stable operation. Upto zero sludge growth Very high MLSS can be maintained Small footprint.
Flexibility of operation. Applied in full-scale
Demerits The efficacy of the process is not clear.
54 and 60%
Major References McWhirter (1978), Boon and Burges (1974)
The mechanism is not fully known.
High aeration cost.
Sludge settling and dewatering is more Upto 100% difficult Poor oxygenation: increased aeration cost. Membrane fouling, responsible for high cost Not feasible to operate with complete sludge retention, minimum wasting is desired. Energy requirements Cost of membranes. New materials still under investigation. Further research is needed to understand 38e54% the process and establish the optimum operational conditions and improve the process operation.
It is relatively easy to introduce the anaerobic zone to the conventional activated sludge process. Control of the growth of filamentous organisms. No physical or chemical forces are needed. The OSA process improves the COD removal and the settleability of activated sludge. Capable of handling high strength organic pollutants without serious sludge associated problems. PREDATION Predation on bacteria are used The worms growth is still uncontrollable, especially in the full scale application in large experimental operations today Kinetics is still undefined. Environmental friendly. High DO High degree of stabilization Inability to maintain a high MLVSS in the aeration tank Stable and Reliable operation The mechanism is not fully known. Easy to implement Inability to maintain a higher MLVSS concentration Can repress development of High energy consumption. filamentous organisms. No application of uncoupling Yet to find a place in its engineering So/Xo chemicals application Lower reactor volume Poor effluent quality. Wastewater post-treatment required. High capital and operation costs. Only applicable to high strength industrial wastewaters. OSA
Reduction in Growth
missing preventing even formulation of future blue print in this regard. But the scenario is different in Japan as it has been highly reliant on incineration for waste disposal (i.e. a total of 1320 plants in 2005), which is extremely high (i.e. 74%) compared to
12e75%
0e66%
Low and Chase (1999b), Churchouse and Wildgoose (1999), Rosenberger et al. (1999; 2002), Judd, 2007.
Chudoba et al. (1992a), Chen et al. (2001a,b, 2003)
Lee and Welander (1994, 1996a,b), Ghyoot and Verstraete (2000), Wei et al. (2003a, 2009a). Moussa et al. (2005). Abbassi et al. (2000) Ichinari et al. (2008)
Around 70% Chudoba et al. (1991), Ghigliazza et al. (1996), Liu (2000)
that in other Asian countries (Horio et al., 2009). 1997 saw the amendment to the 1991 Promotion of Resource Recycling and Reuse Law and a new Waste Disposal and Public Cleansing Law enacted in Japan as basic statutory regulations for waste disposal and recycling. This Law serves to promote recycling
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 2 8 7 e4 3 1 0
at the production, distribution, and consumption stages, effective use of resources, decreased waste generation, and environmental conservation (Tanaka, 1999). However, the element of eco friendly disposal of sludges is still missing as their treatment is largely confined to landfills and incineration. And still the amount of night soil, which can be collected from areas where sewage system has not been developed yet (i.e. 5% of all areas) was around 1.11million tons/year (PERI, 2007). However, the policy of enactment of law for the installation of onsite wastewater treatment in Japan called Johkasou paid rich dividends as it reduces sludge by 35% resulting in an observed growth yield of only 0.18. It is a combination of “Jouka” means purification and “Sou” means tub or tank (Ichinari et al., 2008). The success of Johkasou owe to the provision of a national subsidy and bond issuance to home owners, which increased from 100 million yen in 1987 to 21 billion yen in 2000 (JECES, 2004). Therefore, with determination some hard decisions have to be taken by the respective national governments and their regulatory authorities on stringent legislations, policies and maintenance of standards in the near future, lest the economic growth built assiduously will shortly return to same stupor as before.
4.
Summary
There is a noticeable shifting of priorities from higher volumetric rate and effluent quality to less biomass production during wastewater treatment on the basis of overall mass balance of the inputs and outputs. Particularly during the last decades there has been a major change in the ways sludges are disposed. Prior to 1998, land application as a fertilizer for agricultural reuse (37%), incineration (11%), simply land filling (40%) and use in some other areas such as forestry, land reclamation, seawaters disposal (12%) etc. were the common alternatives of municipal sludge disposal. Since 1998, onwards, European legislation prohibits the sea disposal of sewage sludge, in order to protect the marine environment and sludge deposits in landfills were also proposed to be phased out. This led to generation of significant scientific interest in the latest trends in the field of sludge management, i.e. agricultural reuse, biological oxidation, digestion, combustion, wet oxidation, pyrolysis, gasification and co-combustion of sewage sludge with other materials for further use as energy source. In addition to inherent minerals sewage sludge also contains nitrogen and phosphorous, resulting especially from nitrification and phosphorous accumulation phases in wastewater treatment process, which possesses fertilizer value. At the same time it may contain various other elements, which can be harmful when entering in human food chain, such as heavy metals, protonophores, disinfectants, pathogens and organic pollutants. An essential prerequisite of agriculture use is its storage; as sludge is being produced all year round whereas its application on land takes place once or twice a year; in addition, social unacceptability remains the grey area in the picture. Therefore, agricultural use is increasingly regarded as an insecure handling route. The other conventional route, such as land filling is eliminated due to stringent legislation and increased costs. Incineration on the
4305
other hand provides a large volume reduction of sewage sludge and results in improved thermal efficiency. However, the scrubbing costs of the product gases for air pollution control are usually very high in addition to high carbon footprint. Therefore, there are several driving forces for the search of alternatives for safe handling of the sewage sludges and reduction or minimization at the place of its production (inpipe) through biological lysis-cryptic growth, biological uncoupling of metabolism, maintenance metabolism and ecomanipulation. These sludge reduction technologies have validated the potential to significantly change the methods by which wastewater treatment biosolids are treated and handled across the globe, but still the real impetus is to explore more and more economically viable and environmentally friendlier alternatives. Some of the recent advent of sludge reduction technologies are just being introduced and thus are not yet well tested as combustion or incineration. To be precise, no consensus on sludge reduction approach has been evolved so far and at best the choice of technique depends on their relative merits and demerits. However, all the above summed issues indicate the importance of investigating thoroughly these novel trends in sewage sludge handling. The only conclusion drawn is that the total oxidation of synthesized biomass during wastewater treatment is indeed valid, albeit difficult to affectuated by current engineering practices evolved so far. It is also expected that this review would serve its purpose as dissemination of knowledge on this subject would take place.
Acknowledgements We are thankful to Editor Water Research for review, support and communication. We are grateful to the two anonymous reviewers, this article would not have taken shape without critical comments and time spared by them, their efforts deserve due acknowledgement. Thanks are due to Dr. Vinay K. Tyagi, M/S Abid A. Khan, Rubia, Muntajeer and Akanksha, whose support was our ultimate strength.
references
Abbassi, B., Dullstein, S., Rabiger, N., 2000. Minimization of excess sludge production by increase of oxygen concentration in activated sludge flocs: experimental and theoretical approach. Water Research 34 (1), 139e146. Ahn, K.H., Park, K.Y., Maeng, S.K., Hwang, J.H., Lee, J.W., Song, K. G., Choi, S., 2002. Ozonation of wastewater sludge for reduction and recycling. Water Science and Technology 46 (10), 71e77. Alam, M.Z., Fakhru’l-Razi, A., 2003. Enhanced settleability and dewaterability of fungal treated domestic wastewater sludge by liquid state bioconversion process. Water Research 37, 1118e1124. Alam, M.Z., Fakhru’l-Razi, A., Molla, A.H., 2003a. Biosolids accumulation and biodegradation of domestic wastewater treatment plant sludge by developed liquid state bioconversion process using batch fermentation. Water Research 37 (15), 3569e3578.
4306
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 2 8 7 e4 3 1 0
Alam, M.Z., Fakhru’l-Razi, A., Molla, A.H., 2003b. Optimization of liquid state. Bioconversion process for microbial treatment of domestic wastewater sludge. Journal of Environmental Engineering and Science 2, 299e306. Alam, M.Z., Fakhru’l-Razi, A., Molla, A.H., Roychoudhury, P.K., 2001. Treatment of wastewater sludge by liquid state bioconversion process. Journal of Environmental Science Health A 36 (7), 1237e1243. Are´valo, J., Moreno, B., Pe´rez, J., Go´mez, M.A., 2009. Applicability of the sludge biotic index (SBI) for MBR activated sludge control. Journal of Hazardous Materials 167], 784e789. Armenante, P.M., Kafkewitz, D., Lewandowski, G.A., Jou, C.J., 1999. Anaerobiceaerobic treatment of halogenated phenolic compounds. Water Research 33, 681e692. Baier, U., Schmidheiny, P., 1997. Enhanced anaerobic degradation of mechanically disintegrated sludge. Water Science and Technology 36 (11), 137e143. Barjenbruch, M., Hoffmann, H., Tra¨nckner, J., 2000. Minimizing of foaming in digesters by pre-treatment of the surplus sludge. Water Science and Technology 42 (9), 235e242. Batstone, D.J., Keller, J., Angelidaki, I., Kalyuzhnyi, S.V., Pavlostathis, S.G., Rozzi, A., Sanders, W.T.M., Siegrist, H., Vavilin, V.A., 2002. Anaerobic Digestion Model No. 1 (ADM1). IWA Publishing, London, UK, ISBN 1900222787. Benefield, L.D., Randall, C.W., 1980. Biological Process Design for Wastewater Treatment. Prentice-Hall, Inc., Englewood Cliffs, NJ. Beyer, S., 2006. Environmental law and policy in the People’s Republic of China. Chinese Journal of International Law 5 (1), 185e211. Bhardwaj, R.M. (2005) Status of Wastewater Generation and Treatment in India. IWG-Env Joint Work Session on Water Statistics, 20e22 June 2005, Vienna. Bishop, B. (2004) Use of ceramic membranes in airlift membrane bioreactors. 8th International Conference on Inorganic Membranes (ICIM8); 2004 July 18e22, 2004, Cincinnati, USA. Bitton, G., 2005. Wastewater Microbiology, third ed. Wiley-Liss, Hoboken, N.J. Boon, A.G., Burgess, D.R., 1974. Treatment of crude sewage in two high-rate activated sludge plants operated in series. Journal of the Water Pollution Control 74, 382. Bouillot, P., Canales, A., Pareilleux, A., Huyard, A., Goma, G., 1990. Membrane bioreactors for the evaluation of maintenance phenomena in wastewater treatment. Journal of Fermentation and Bioengineering 69 (3), 178e183. Camacho, P., Geaugey, V., Ginestet, P., Paul, E., 2002. Feasibility study of mechanically disintegrated sludge and recycle in the activated-sludge process. Water Science and Technology 46 (10), 97e104. Canales, A., Pareilleux, A., Rols, J., Goma, G., Huyard, A., 1994. Decreased sludge production strategy for domestic wastewater treatment. Water Science and Technology 30 (8), 97e106. Cech, J.S., Hartman, P., Macek, M., 1994. Bacteria and protozoa population dynamics in biological phosphate removal systems. Water Science and Technology 19 (7), 109e117. Chaize, S., Huyard, A., 1991. Membrane bioreactor on domestic wastewater treatment sludge production and modelling approach. Water Science and Technology 23, 1591e1600. Chen, G.H., Leung, D.H.W., 1999. Utilization of oxygen in a sanitary gravity sewer. Water Research 34 (15), 3813e3821. Chen, G.H., Liu, Y., 1999. Modeling of energy spilling in substratesufficient cultures. Journal of Environmental Engineering e ASCE 125 (6), 508e513. Chen, G.H., Mo, H.K., Liu, Y., 2002. Utilization of a metabolic uncoupler, 3,30 ,40 ,5 tetrachlorosalicylanilide (TCS) to reduce sludge growth in activated sludge culture. Water Research 36 (8), 2077e2083. Chen, G.H., Mo, H.K., Saby, S., Yip, W.K., Liu, Y., 2000. Minimization of activated sludge production by chemically stimulated energy spilling. Water Science and Technology 42 (12), 189e200.
Chen, G.H., An, K.J., Saby, S., Brois, E., Djafer, M., 2003. Possible cause of excess sludge reduction in an oxic-settling anaerobic activated sludge process (OSA process). Water Research 37 (16), 3855e3866. Chen, G.H., Saby, S., Djafer, M., Mo, H.K., 2001a. New approaches to minimize excess sludge in activated sludge systems. Water Science and Technology 44 (10), 203e208. Chen, G.H., Yip, W.K., Mo, H.K., Liu, Y., 2001b. Effect of sludge fasting/feasting on growth of activated sludge cultures. Water Research 35 (4), 1029e1037. Chiu, Y.C., Chang, C.N., Lin, J.G., Huang, S.J., 1997. Alkaline and ultrasonic pretreatment of sludge before anaerobic digestion. Water Science and Technology 36 (11), 155e162. Chu, C.P., Chang, B.V., Liao, G.S., Jean, D.S., Lee, D.J., 2001. Observation on changes in ultrasonically treated wasteactivated sludge. Water Research 35 (4), 1038e1046. Chu, C.P., Feng, W.C., Chang, B.V., Chou, C.H., Lee, D.J., 1999. Reduction of microbial density level in wastewater activated sludge via freezing and thawing. Water Research 33 (16), 3532e3535. Chudoba, B., Morel, A., Capdeville, B., 1992b. The case of both energetic uncoupling and metabolic selection of microorganisms in the OSA activated sludge system. Environmental Technology 13, 761e770. Chudoba, P., Capdeville, B., Chudoba, J., 1992c. Explanation of biological meaning of the So/Xo ratio in batch cultivation. Water Science and Technology 26 (3e4), 743e751. Chudoba, P., Chevalier, J.J., Chang, J., Capdeville, B., 1991. Effect of anaerobic stabilization of activated sludge on its production under batch conditions at various So/Xo ratios. Water Science and Technology 23, 917e926. Chudoba, P., Chudoba, J., Capdeville, B., 1992a. The aspect of energetic uncoupling of microbial growth in the activated sludge process: OSA system. Water Science and Technology 26 (9e11), 2477e2480. Churchouse, S., Wildgoose, D., 1999. Membrane bioreactor from lab to full-scale application. In: The 2nd Symposium on Membrane Bioreactors for Wastewater Treatment. Cranfield University, UK. Cicek, N., Macomber, J., Davel, J., Suidan, M.T., Audic, J., Genestet, P., 2001. Effect of solids retention time on the performance and biological characteristics of a membrane bioreactor. Water Science and Technology 43 (11), 43e50. Commission of European Communities, 1998. Council Directive 91/271/EEC of 21 March 1991 Concerning Urban Wastewater Treatment amended by the 98/15/EC of 27 February 1998. Copp, J.B., Dold, P.L., 1998. Comparing sludge production under aerobic and anoxic conditions. Water Science and Technology 38, 285e294. CPCB, 2004. News Letter, Urbanisation & Wastewater Management in India. Ministry of Environment and Forests, Government of India. Curds, C.R., Hawkes, H.A., 1975. Ecological Aspects of Used-Water Treatment: The Organisms and their Ecology. Academic Press, New York. Curds, C.R., Cockburn, A., 1970. Protozoa in biological sewage treatment process-I: a survey of the protzoan fauna of British percolating filters and activated sludge plants. Water Research 4, 225e236. Deleris, S., Geaugey, V., Camacho, P., Debellefontaine, H., Paul, E., 2002. Minimization of sludge production in biological processes: an alternative solution for the problem of sludge disposal. Water Science and Technology 46 (10), 63e70. Drews, A., 2007. Does fouling in MBR depend on SMP?. In: Proceedings of the 4th IWA Membranes Conference, May 15e17, 2007. Cranfield University Press, Harrogate, UK. Drews, A., 2010. Membrane fouling in membrane bioreactors e characterisation, contradictions, cause and cures. Journal of Membrane Science 363, 1e28.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 2 8 7 e4 3 1 0
Egemen, E., Corpening, J., Nirmalakhandan, N., 2001. Evaluation of an ozonation system for reduced waste sludge generation. Water Science and Technology 44 (2e3), 445e452. Egemen, E., Corpening, J., Padilla, J., Brennan, R., Nirmalakhandan, N., 1999. Evaluation of ozonation and cryptic growth for biosolids management in wastewater treatment. Water Science and Technology 39 (10e11), 155e158. Eikelboom, D.H., 1988. Extra to epassingsmogelijkheden voor protozoa en metazoa bij de zuivering van afval water TNOrapport no. R88/286, The Netherlands. Eikelboom, D.H., 2000. Process Control of Activated Sludge Plants by Microscopic Investigation. IWA Publishing, UK. Eikelboom, D.H., Borger, A.R., van Houten, R.T., 2001. Reductiezuiveringsslib Door Borstelwormen. Fase 1. TNOrapport. Laboratoriumonderzoek, The Netherlands. Elissen, H.J.H., Hendrickx, T.L.G., Temmink, H., Buisman, C.J.N., 2006. A new reactor concept for sludge reduction using aquatic worms. Water Research 40, 3713e3718. Elton, C.S., 1935. Animal Ecology. Sidgwick and Jackson, London. engineeringvillage2.com, 2010. Web-based Information Service. http://www.engineeringvillage2.com. EPA, 1987. Design Manual-Dewatering Municipal Wastewater Sludge Washington, DC. Fakhru’l-Razi, A., Alam, M.Z., Idris, A., Abd-Aziz, S., Molla, A.H., 2002b. Domestic wastewater accumulation by liquid state bioconversion process for rapid composting. Journal of Environmental Science and Health 38 (8), 1533e1543. Fakhru’l-Razi., A., Molla, A.H., 2007. Enhancement of bioseparation and dewaterability of domestic wastewater sludge by fungal treated dewatered sludge. Journal of Hazardous Materials 147, 350e356. Fenu, A., Roels, J., Wambecq, T., De Gussem, K., Thoeye, C., De Gueldre, G., Van De Steene, B., 2010. Energy audit of a full scale MBR system. Desalination 262, 121e128. Garces, A., De Wilde, W., Thoeye, C., De Gueldre, G., 2007. Operational cost optimisation of MBR Schilde. In: Proceedings of the 4th IWA Membranes Conference, 15e17 May 2007. Harrogate Cranfield University Press, UK. Gaudy, A.F., Gaudy, E.T., 1981. Microbiology for Environmental Scientists and Engineers, First ed. McGraw Hill. Ghigliazza, R., Lodi, A., Converti, A., Nicolella, C., Rovatti, M., 1996. Influence of the ratio of the initial substrate concentration to biomass concentration on the performance of a sequencing batch reactor. Bioprocess and Biosystems Engineering 14 (3), 131e137. Ghyoot, W. (1998) Membrane-assisted bioreactor for anaerobic sludge digestion and low sludge production in aerobic wastewater treatment. PhD thesis, University of Gent, Belgium. Ghyoot, W., Verstraete, W., 2000. Reduced sludge production in a two-stage membrane-assisted bioreactor. Water Research 34 (1), 205e215. Gil, J.A., Tu´a, L., Rueda, A., Montan˜o, B., Rodrı´guez, M., Prats, D., 2010. Monitoring and analysis of the energy cost of an MBR. Desalination 250, 997e1001. Giokas, D.L., Daigger, G.T., von Sperling, M., Kim, Y., Paraskevas, P.A., 2003. Comparison and evaluation of empirical zone settling velocity parameters based on sludge volume index using a unified settling characteristics database. Water Research 37 (16), 3821e3836. Guellil, A., Boualam, M., Quiquampoix, H., Ginestet, P., Audic, J.M., Block, J.C., 2001. Hydrolysis of wastewater colloidal organic matter by extra-cellular enzymes extracted from activated sludge flocs. Water Science and Technology 43 (6), 33e40. Huang, X., Peng, L., Yi, Q., 2007. Excess sludge reduction induced by Tubifex tubifex in a recycled sludge reactor. Journal of Biotechnology 127, 443e451.
4307
Han, S.S., Bae, T.H., Jang, G.G., Tak, T.M., 2005. Influence of sludge retention time on membrane fouling and bioactivities in membrane bioreactor system. Process Biochem 40 (7), 2329e2400. Hazardous and Solid Waste Amendment Act, (1984) U.S. Environmental Protection Agency (EPA): P.L. 98e616, 98 Stat. 3221, Nov. 9, 1984. He, P.-J., Lu, F., Zhang, H., Shao, L.-M., Lee, D.-J., 2006. Sewage Sludge in China: Challenges toward a Sustainable Future. State Key Laboratory of Pollution Control and Resources Reuse. Tongji University, Shanghai and Department of Chemical Engineering, National Taiwan University. Hendrickx, T.L.G., Temmink, H., Elissena, H.J.H., Buisman, C.J.N., 2009a. The effect of operating conditions on aquatic worms eating waste sludge. Water Research 43, 943e950. Hendrickx, T.L.G., Temmink, H., Elissena, H.J.H., Buisman, C.J.N., 2009b. Aquatic worms eating waste sludge in a continuous system. Bioresource Technology 100, 4642e4648. Hendrickx, T.L.G., Temmink, H., Elissena, H.J.H., Buisman, C.J.N., 2010. Design parameters for sludge reduction in an aquatic worm reactor. Water Research 44 (3), 1017e1023. Henze, M., Gujer, W., Mino, M., van Loosdrecht, M.C.M., 2000. Activated Sludge Models ASM1, ASM2, ASM2d and ASM3. IWA Publishing, London, UK. Heran, M., Wisniewski, C., Orantes, J., Grasmick, A., 2008. Measurement of kinetic parameters in a submerged aerobic membrane bioreactor fed on acetate and operated without biomass discharge. Biochemical Engineering Journal 38, 70e77. Horan, N.J., 1990. Biological Wastewater Treatment Systems, Theory and Operation. John Wiley and Sons, Chichester, UK. Horio, M., Shigeto, S., Shiga, M., 2009. Evaluation of energy recovery and CO2 reduction potential in Japan through integrated waste and utility management. Waste Management 29 (7), 2195e2202. Hu, A.Y., Stuckey, D.C., 2006. Treatment of dilute wastewaters using a novel submerged anaerobic membrane bioreactor. Journal of Environmental Engineering 132, 190e198. Huang, X., Gui, P., Qing, Y., 2001. Effect of sludge retention time on microbial behaviour in a submerged membrane bioreactor. Process Biochemistry 36, 1001e1006. Humenick, M.J., Ball, J.E., 1974. Kinetics of activated sludge oxygenation. Journal of the Water Pollution Control Federation 46, 553e561. Ichinari, T., Ohtsubo, A., Ozawa, T., Hasegawa, K., Teduka, K., Oguchi, T., Kiso, Y., 2008. Wastewater treatment performance and sludge reduction properties of a household wastewater treatment system combined with an aerobic sludge digestion unit. Process Biochemistry 43, 722e728. Idris, A., Inanc, B., Hassan, M.N., 2004. Overview of waste disposal and landfills/dumps in Asian countries. Material Cycles and Waste Management 6, 104e110. Inamori, Y., Suzuki, R., Sudo, R., 1983. Mass culture of small aquatic oligochaeta. Research Report from the National Institute for Environmental Studies 47, 125e137 (in Japanese). Jamal, P., Alam, M.Z., Salleh, M.R.M., Akib, M.M., 2005. Sewage treatment plant sludge: a source of potential microorganism for citric acid production. American Journal of Applied Sciences 2 (8), 1236e1239. Janssen, P.M.J., Rulkens, W.H., Rensink, J.H., van der Poest, H.F., 1998. The potential for metazoa in biological wastewater treatment. Water Quality International, 25e27. JECES, 2004. Japanese Education Center for Environmental Sanitation. Jefferson, B. (2007) Low temperature municipal sewage treatment with anaerobic MBRs. The 6th International Membrane Science and Technology Conference: November 5e9, 2007; Sydney. Jeison, D., van Lier, J.B., 2007. Feasibility of thermophilic anaerobic submerged membrane bioreactors (AnSMBR) for
4308
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 2 8 7 e4 3 1 0
wastewater treatment. In: Proceedings of the 4th IWA Membranes Conference, May 15e17, 2007. Cranfield University Press, Harrogate, UK. Judd, S., 2007. The status of membrane bioreactor technology. Trends Biotechnology 26 (2), 109e116. Jun, H.B., Park, S.M., Park, N.B., Lee, S.H., 2004a. Nitrogen removal and sludge reduction in a symbiotic activated sludge system between anaerobic archaea and bacteria. Water Science and Technology 50 (6), 189e197. Jun, H.-J., Liu, D., Liang, P., 2004b. Sludge Treatment and Disposal. Qinghua Water Industry Technology Green Paper, First Series. Department of Qinghua University, China’s water network environment. Kepp, U., Machenbach, I., Weisz, N., Solheim, O.E., 1999. Enhanced stabilization of sewage sludge through thermal hydrolysis e three years of experience with a full scale plant. Water Science and Technology 42 (9), 89e96. Kopp, J., Muller, J., Dichtl, N., Schwedes, J., 1997. Anaerobic digestion and dewatering characteristics of mechanically disintegrated excess sludge. Water Science and Technology 36 (11), 129e136. Kuniyasu, K., Hayashi, N., Inamori, Y., Sudo, R., 1997. Effect of environmental factors on growth characteristics of oligochaeta. Japanese Journal of Water Treatment Biology 33, 207e214 (in Japanese). Laera, G., Pollice, A., Saturno, D., Giordano, C., Lopez, A., 2005. Zero net growth in a membrane bioreactor with complete sludge retention. Water Research 39, 5241e5249. Lapinski, J., Tunnacliffe, A., 2003. Reduction of suspended biomass in municipal wastewater using bdelloid rotifers. Water Research 37 (9), 2027e2034. Laspidou, C.S., Rittmann, B.E., 2002. A unified theory for extracellular polymeric substances, soluble microbial products, and active and inert biomass. Water Research 36, 2711e2720. Lawrence, A.W., McCarty, P.L., 1970. Unified basis for biological treatment design and operation. ASCE Journal of the Sanitary Engineering Division 96 (SA3), 757. Learner, M.A., 1979. The distribution and ecology of the Naididae (Oligochaeta) which inhabit the filter-beds of sewage-works in Britain. Water Research 13, 1291e1299. Lee, N.M., Welander, T., 1994. Influence of predator in nitrification in aerobic biofilm. Water Science and Technology 29 (7), 355e363. Lee, N.M., Welander, T., 1996a. Reducing sludge production in aerobic wastewater treatment through manipulation of the ecosystem. Water Research 30 (8), 1781e1790. Lee, N.M., Welander, T., 1996b. Use of protozoa and metazoa for decreasing sludge production in aerobic wastewater treatment. Biotechnology Letters 18 (4), 429e434. Leenen, E.J.T.M., Boogert, A.A., van Lammeren, A.A.M., Tramper, J., Wijffels, R.H., 1997. Dynamics of artificially immobilized Nitrosomonas europaea: effect of biomass decay. Biotechnology and Bioengineering 55, 630e641. Liang, P., Huang, X., Qian, Y., 2006a. Excess sludge reduction in activated sludge process through predation of Aeolosoma hemprichi. Biochemical Engg Journal 28, 117e122. Liang, P., Huang, X., Qian, Y., Wei, Y., Ding, G., 2006b. Determination and comparison of sludge reduction rates caused by microfaunas’ predation. Bioresource Technology 97, 854e861. Lin, S., Jin, Y., Fu, L., Quan, C., Yang, Y.S., 2009. Microbial community variation and functions to excess sludge reduction in a novel gravel contact oxidation reactor. Journal of Hazardous Materials 165, 1083e1090. Liu, Y., 1996. Bioenergetic interpretation on the So/Xo in substrate-sufficient batch culture. Water Research 30 (11), 2766e2770. Liu, Y., 2000. The So/Xo-dependent dissolved organic carbon distribution in substrate-sufficient batch culture of activated sludge. Water Research 34 (5), 1645e1651.
Liu, Y., Tay, J.H., 2001. Strategy for minimization of excess sludge production from the activated sludge process. Biotechnology Advances 19 (2), 97e107. Liu, Y., Chen, G.H., Paul, E., 1998. Effect of the So/Xo ratio on energy uncoupling in substrate-sufficient batch culture of activated sludge. Water Research 32 (10), 2833e2888. Lobos, J., Wisniewski, C., Heran, M., Grasmick, A., 2008. Sequencing versus continuous membrane bioreactors: effect of substrate to biomass ratio (F/M) on process performance. Journal of Membrane Science 317, 71e77. Low, E.W., Chase, H.A., 1999a. Reducing production of excess biomass during wastewater treatment. Water Research 33 (5), 1119e1132. Low, E.W., Chase, H.A., 1999b. The effect of maintenance energy requirements on biomass production during wastewater treatment. Water Research 33 (3), 847e853. Lundin, M., Olofsson, M., Pettersson, G., Zetterlund, H., 2004. Environmental and economic assessment of sewage sludge handling options. Resources, Conservation and Recycling 41, 255e278. Luxmy, B.S., Kubo, T., Yamamoto, K., 2001. Sludge reduction potential of metazoa in membrane bioreactors. Water Science and Technology 44 (10), 197e202. Madoni, P., 1994. A sludge biotic index (SBI) for the evaluation of the biological performance of activated sludge plants based on the microfauna analysis. Water Research 28, 67e75. Mannan, S., Fakhru’l-Razi, A., Alam, Z.Md, 2005. Use of fungi to improve bioconversion of activated sludge. Water Research 39, 2935e2943. Mason, C.A. (1986) Microbial death, lysis and Cryptic growth: fundamental and applied aspects. PhD Thesis, Swiss Fed. Inst. Tecnol. Zurich. Mason, C.A., Hamer, G., 1987. Cryptic growth in Klebsiella pneumonia. Applied Microbiology and Biotechnology 25, 577e584. Mason, C.A., Hamer, G., Bryers, J.D., 1986. The death and lysis of microorganism in environmental process. FEMS Microbiology Review 39, 373e401. McWhirter, J.R., 1978. Oxygen and Activated Sludge Process. The Use of High-Purity Oxygen in the Activated Sludge Process, vol. 1. CRC Press, Boca Raton, FL: USA. Meng, F., Chae, S., Drews, A., Kraume, M., Shin, H., Yang, F., 2009. Recent advances in membrane bioreactors (MBRs): membrane fouling and membrane material. Water Research 43 (6), 1489e1512. Metcalf and Eddy, Inc., 2003. Wastewater Engineering: Treatment, Disposal and Reuse, fourth rev. ed. Tata McGraw-Hill, New Delhi. Mishima, K., Nakamura, M., 1991. Self-immobilization of aerobic activated sludge: a pilot study of the aerobic upflow sludge blanket process in municipal sewage treatment. Water Science and Technology 23, 981e990. Molla, A.H., Fakhru’l-Razi, A., Alam, M.Z., 2004. Evaluation of solid-state bioconversion of domestic wastewater sludge as a promising environmental friendly disposal technique. Water Research 38 (19), 4143e4152. More, T.T., Yan, S., Tyagi, R.D., Surampalli, R.Y., 2010. Potential use of filamentous fungi for wastewater sludge treatment. Bioresource Technology 101, 7691e7700. Moussa, M.S., Hooijmans, C.M., Lubberding, H.J., Gijzen, H.J., van Loosdrecht, M.C.M., 2005. Modelling nitrification, heterotrophic growth and predation inactivated sludge. Water Research 39, 5080e5098. Mudrack, K., Kunst, S., 1991. Biologie der Abwasserreinigung (Biology of Wastewater Purification), third ed. Gustav Fischer Verlag, Stuttgart, Germany. Mulder, J.W. (2009) Operational experiences with the hybrid MBR Heenvliet, a smart way of retrofitting. Book of Proceedings of final MBR Network workshop, 31 Marche1 April, 2009, Berlin, Germany.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 2 8 7 e4 3 1 0
Muller, E.B., Stouthamer, A.H., van Verseveld, H.W., Eikelboom, D. H., 1995. Aerobic domestic waste water treatment in a pilot plant with complete sludge retention by cross-flow filtration. Water Research 29 (4), 1179e1189. Municipal Solid Wastes (Management and Handling) Rules, 2000. Ministry of Environment and Forests, Government of India. Neyens, E., Baeyens, J., Weemaes, M., De heyder., B., 2003a. Alkaline thermal sludge hydrolysis. Journal of Hazardous Materials 97 (1e3), 295e314. Neyens, E., Baeyens, J., Weemaes, M., De heyder, B., 2003b. Hot acid hydrolysis as a potential treatment of thickened sewage sludge. Journal of Hazardous Materials 98 (1e3), 275e293. Neyens, E., Baeyens, J., Weemaes, M., De heyder., B., 2003c. Pilot scale peroxidation (H2O2) of sewage sludge. Journal of Hazardous Materials 98 (1e3), 91e106. Nolasco, M.A., Campos, A.L.O., Springer, A.M., Pires, E.C., 2002. Use of lysis and recycle to control excess sludge production inactivated sludge treatment: bench scale study and effect of chlorinated organic compounds. Water Science and Technology 46 (10), 55e61. Nowak, G., Brown, G., Yee, A., 1986. Effects of feed pattern and dissolved oxygen on growth of filamentous bacteria. Journal of the Water Pollution Control Federation 58, 978e984. Onyeche, T.I., Schla¨fer, O., Bormann, H., Schro¨der, C., Sievers, M., 2002. Ultrasonic cell disruption of stabilized sludge with subsequent anaerobic digestion. Ultrasonics 40 (1e8), 31e35. Palmegren, R., Jorand, F., Nielsen, P.H., Block, J.C., 1998. Influence of oxygen limitation on the cell surface properties of bacteria from activated sludge. Water Science and Technology 37, 349e352. Pena, C., Trujillo-Roldan, M.A., Galindo, E., 2000. Influence of dissolved oxygen tension and agitation speed on alginate production and its molecular weight in cultures of Azotobacter vinelandii. Enzyme and Microbial Technology 27, 390e398. PEGASUS Research Institute (PERI), 2007. The Report of Questionnaire Survey on Waste Treatment Facilities in Japan (in Japanese). Pirt, S.J., 1975. Principles of Microbe and Cell Cultivation, first ed. Blackwell Scientific Publications, Oxford. Pollice, A., Giordano, C., Laera, G., Saturno, D., Mininni, G., 2007. Physical characteristics of the sludge in a complete retention membrane bioreactor. Water Research 41, 1832e1840. Puigagut, J., Salvado´, H., Tarrats, X., Garcı´a, J., 2007. Effects of particulate and soluble substrates on microfauna populations and treatment efficiency in activated sludge systems. Water Research 41, 3168e3176. Quintana, M., Sanchez, E., Colmenarejo, M.F., Barrera, J., ıa, G., Borja, R., 2005. Kinetics of phosphorus removal and Garc struvite formation by the utilization of by-product of magnesium oxide production. Chemical Engineering Journal 111, 45e52. Ratsak, C.H., Verkuijlen, J., 2006. Sludge reduction by predatory activity of aquatic oligochaetes in wastewater treatment plants: science or fiction? A review. Hydrobiologia 564, 197e211. Ratsak, C.H. (1994) Grazer induced sludge reduction in wastewater treatment. PhD thesis, Vrije University, The Netherlands. Ratsak, C.H., 2001. Effects of Nais elinguis on the performance of an activated sludge plant. Hydrobiologia 463, 217e222. Ratsak, C.H., Kooi, B.W., van Verseveld, H.W., 1994. Biomass reduction and mineralization increase due to the ciliate Tetrahymena pyriformis grazing on the bacterium Pseudomonas fluorescens. Water Science and Technology 29 (7), 119e128. Ratsak, C.H., Kooijman, S.A.L., Kooi, B.W., 1993. Modelling the growth of an oligochaete on activated sludge. Water Research 27 (5), 739e747. Ratsak, C.H., Maarsen, K.A., Kooijman, S.A.L., 1996. Effects of protozoa on carbon mineralization in activated sludge. Water Research 30 (1), 1e12.
4309
Rensink, J.H., Rulkens, W.H., 1997. Using metazoa to reduce sludge production. Water Science and Technology 36 (11), 171e179. Rensink, J.H., Corstanje, R., van der Pal, J.H. (1996) A new approach to sludge reduction by metazoa. 10th European Sewage and Reuse Symposium, IFAT 1996, Munchen. Resource Conservation and Recovery Act (RCRA of 1976), (1984) 42 USCx6901, 98e616, N. Richard, M., Hao, O., Jenkins, D., 1985. Growth kinetics of Sphaerotilus species and their significance in activated sludge bulking. Journal of the Water Pollution Control Federation 57, 68e81. Rittmann, B.E., McCarty, P.L., 2001. Environmental Biotechnology: Principles and Applications, first ed. McGraw-Hill, New York. Rocher, M., Goma, G., Begue, A.P., Louvel, L., Rols, J.L., 1999. Towards a reduction in excess sludge production inactivated sludge processes: biomass physicochemical treatment and biodegradation. Applied Microbiology and Biotechnology 51 (6), 883e890. Rocher, M., Roux, G., Goma, G., Begue, A.P., Louvel, L., Rols, J.L., 2001. Excess sludge reduction in activated sludge processes by integrating biomass alkaline heat treatment. Water Science and Technology 44 (2e3), 437e444. Roels, J., Wambecq, T., De Gussem, K., Fenu, A. (2010) LCA and nutrient removal. Neptune and Innowatech, End User Conference, Gent, Belgium. Roques, H., Capdeville, B., Seropian, J.C., Grigoropoulou, H., 1984. Oxygenation by hydrogen peroxide of the fixed biomass used in biological water treatment. Water Research 18, 103e110. Rosenberger, S., Kraume, M., Szewzyk, U., 1999. Sludge free management of membrane bioreactors. In: The 2nd Symposium on Membrane Bioreactor for Wastewater Treatment. The School of Water Sciences, Cranfield University, UK. Rosenberger, S., Kruger, U., Witzig, R., Manz, W., Szewzyk, U., Kraume, M., 2002. Performance of a bioreactor with submerged membranes for aerobic treatment of municipal wastewater. Water Research 36 (2), 413e420. Rosenberger, S., Laabs, C., Lesjean, B., Gnirss, R., Amy, G., Jekel, M. , Schrotter, J.C., 2006. Impact of colloidal and soluble organic material on membrane performance in membrane bioreactors for municipal wastewater treatment. Water Research 40, 710e720. Rosenberger, S., Witzig, R., Manz, W., Szewzyk, U., Kraume, M., 2000. Operation of different membrane bioreactors: experimental results and physiological state of the microorganisms. Water Science and Technology 41 (10e11), 269e277. Ryan, F.J., 1959. Bacterial mutation in a stationary phase and the question of cell turnover. Journal of General Microbiology 21, 530e549. Saby, S., Djafer, M., Chen, G.H., 2002. Feasibility of using a chlorination step to reduce excess sludge in activated sludge process. Water Research 36 (3), 656e666. Saby, S., Djafer, M., Chen, G.H., 2003. Effect of low ORP in anoxic sludge zone on excess sludge production in oxic-settling anoxic activated sludge process. Water Research 37 (1), 11e20. Saiki, Y., Imabayashi, S., Iwabuchi, C., Kitagawa, Y., Okumura, Y., Kawamura, H., 1999. Solubilization of excess activated sludge by self-digestion. Water Research 33 (8), 1864e1870. Salvado, H., Gracia, M.P., Amigo, J.M., 1995. Capability of ciliated protozoa as indicators of effluent quality in activated-sludge plants. Water Research 29, 1041e1050. Scott Blair, G.W., 1939. J. Phys. Chem. 43 (7), 853e864. Sengewein, H., 1989. Das Sauerstoffbelebungsverfahren, Abwasserreinigung mit reinem Sauerstoff. (The Activated Sludge Process on Oxygen, Waste Water Purification with Pure Oxygen). Academia Verlag Richarz, Sankt Augustin. Shanableh, A., 2000. Production of useful organic matter from sludge using hydrothermal treatment. Water Research 34 (3), 945e951.
4310
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 2 8 7 e4 3 1 0
Siegrist, H., Brunner, I., Koch, G., Phan, L.C., Le, V.C., 1999. Reduction of biomass decay rate under anoxic and anaerobic conditions. Water Sci. Technol 39, 129e137. Song, B., Xiaofei, C., 2009. Effect of Aeolosoma hemprichi on excess activated sludge reduction. Journal of Hazardous Materials 162, 300e304. Stouthamer, A.H., 1979. The search for correlation between theoretical and experimental growth yields. International Review of Biochemistry 21, 1e15. Stryer, L., 1988. Biochemistry, third ed. Freeman and Company, New York. Subramanian, B.S., Yan, S., Tyagi, R.D., Surampalli, R.Y. (2006) Isolation of extra cellular biopolymer producing microorganisms from wastewater sludge for sludge settling and dewatering. Proceedings of the 79th Annual Water Environment Federation Technical Exposition and Conference; Dallas, Texas Oct 21e25, 2006; Alexandria, Virginia, USA. Subramanian, B.S., Yan, S., Tyagi, R.D., Surampalli, R.Y., 2008. A new, pellet-forming fungal strain: its isolation, molecular identification, and performance for simultaneous sludgesolids reduction, flocculation, and dewatering. Water Environment Research 80 (9), 840e852. Sun, D.D., Khor, S.L., Hay, C.T., Leckie, J.O., 2007. Impact of prolonged sludge retention time on the performance of a submerged membrane bioreactor. Desalination 208, 101e112. Tanaka, M., 1999. Recent trends in recycling activities and waste management in Japan. Material Cycles and Waste Management 1, 10e16. Tanaka, S., Kobayashi, T., Kamiyama, K., Bildan, M.L., 1997. Effects of thermochemical pre-treatment on the anaerobic digestion of waste activated sludge. Water Science and Technology 36 (8), 209e215. Tiehm, A., Nickel, K., Neis, U., 1997. The use of ultrasound to accelerate the anaerobic digestion of sewage sludge. Water Science and Technology 36 (11), 121e128. Tiehm, A., Nickel, K., Zellhorn, M., Neis, U., 2001. Ultrasonic waste activated sludge disintegration for improving anaerobic stabilization. Water Research 35 (8), 2003e2009. Tottie, O. (2007) Evaluation of sludge management in Wuhan, China. Master Thesis, UPTEC W08 002, ISSN 1401e5765; Department of Microbiology, Swedish University of Agricultural Sciences, Uppsala University, Uppsala. Trivedi, H., Heinen, N., 2000. Simultaneous nitrification/ denitrification by monitoring NAOH Fluorescence in activated sludge. In: Proceedings of the Facility Operations II: Innovative Technology Forum; 73rd Annual Conference. WEF, Anaheim, CA. van Loosdrecht, M.C.M., Henze, M., 1999. Maintenance, endogenous respiration, lysis, decay and predation. Water Science and Technology 39 (1), 107e117. Visvanathan, C., Aim, R.B., Parameshwaran, K., 2000. Membrane separation bioreactors for wastewater treatment. Critical Reviews in Environmental Science and Technology 30 (1), 1e48. Wagner, J., Rosenwinkel, K.H., 2000. Sludge production in membrane bioreactors under different conditions. Water Science and Technology 41 (10e11), 251e258. Wang, W., Jung, Y., Kiso, Y., Yamada, T., Min, K., 2006. Excess sludge reduction performance of an aerobic SBR process equipped with a submerged mesh filter unit. Process Biochemistry 41, 745e751. Watson, T.G., 1970. Effects of sodium chloride on steady-state growth and metabolism of Saccharomyces cerevisiae. J. Genet. Microbiol. 64, 91e99. Weemaes, M., Grootaerd, H., Simoens, F., Huysmans, A., Verstraete, W., 2000a. Ozonation of sewage sludge prior to anaerobic digestion. Water Science and Technology 42 (9), 175e178.
Weemaes, M., Grootaerd, H., Simoens, F., Verstraetem, W., 2000b. Anaerobic digestion of ozonized biosolids. Water Research 34 (8), 2330e2336. Wei, Y., Van Houten, R.T., Borger, A.R., Eikelboom, D.H., Fan, Y., 2003b. Comparison performances of membrane bioreactor (MBR) and conventional activated sludge (CAS) processes on sludge reduction induced by Oligochaete. Environmental Science and Technology 37 (14), 3171e3180. Wei, Y., Van Houten, R.T., Borger, A.R., Eikelboom, D.H., Fan, Y., 2003a. Minimization of excess sludge production for biological wastewater treatment. Water Research 37, 4453e4467. Wei, Y., Wang, Y., Guo, X., Liu, J., 2009a. Sludge reduction potential of the activated sludge process by integrating an oligochaete reactor. Journal of Hazardous Materials 163, 87e91. Wei, Y., Zhu, H., Wang, Y., Li, J., Zhang, P., Hu, J., Liu, J., 2009b. Nutrients release and phosphorus distribution during oligochaetes predation on activated sludge. Biochemical Engineering Journal 43, 239e245. Welander, T, and Lee, N.M. (1994) Minimization of sludge production in aerobic treatment by use of predators. The Second International Symposium on Environmental Biotechnology, Brighton, UK. Westgarth, W.C., Sulzer, F.T., Okun, D.A. (1964) Anaerobiosis in the activated sludge process. Proceedings of the Second IAWPRC Conference. Tokyo, Japan. Wilen, B.M., Peter, B., 1999. The effect of dissolved oxygen concentration on the structure, size and size distribution of Activated sludge flocs. Water Research 33 (2), 391e400. Witzig, R., Manz, W., Rosenberger, S., Kruger, U., Kraume, M., Szewzyk, U., 2002. Microbiological aspects of a bioreactor with submerged membranes for aerobic treatment of municipal wastewater. Water Research 36 (2), 394e402. Wozniak, T., 2010. MBR design and operation using MPE-technology (Membrane Performance Enhancer). Desalination 250, 723e728. Wunderlich, R., Barry, J., Greenwood, D., Carry, C., 1985. Startup of a high-purity, oxygen-activated sludge system at the Los Angeles County Sanitation Districts’ Joint water pollution control plant. Journal of the Water Pollution Control Federation 57, 1012e1018. Xie, Y., Xu, C., Tong, Z., Kenji, F., Tsutomu, N., 2008. Operation Character of a Metal MBR in Treating the Domestic Sewage, vol. 29. Dongbei Daxue Xuebao/Journal of Northeastern University, Shengyang (Chinese)160e164. Xing, C.H., Yamamoto, K., Fukushi, K., 2006. Performance of an inclined-plate membrane bioreactorat zero excess sludge discharge. Journal of Membrane Science 275, 175e186. Yamamoto, K., Hiasa, M., Mahmood, T., Matsuo, T., 1989. Direct solideliquid separation using hollow fibre membrane in an activated sludge aeration tank. Water Science and Technology 21, 43e54. Yasui, H., Nakamura, K., Sakuma, S., Iwasaki, M., Sakai, Y., 1996. A full-scale operation of a novel activated sludge process without excess sludge production. Water Science and Technology 34 (3e4), 395e404. Yoon, S.-H., Kim, H.-S., Yeom, I.-T., 2004. The optimum operational condition of membrane bioreactor (MBR): cost estimation of aeration and sludge treatment. Water Research 38, 37e46. Zhang, S. (2000) Polluted water treatment by the combining processes of membrane separation and biodegradation. PhD thesis, Research Centre for Eco-Environmental Sciences, Chinese Academy of Sciences, China. Zubair, A., Chob, J., Lim, B., Song, K., Ahn, K., 2007. Effects of sludge retention time on membrane fouling and microbial community structure in a membrane bioreactor. Journal of Membrane Science 287, 211e218. Zupancic, G.D., Ros, M., 2008. Aerobic and two-stage anaerobiceaerobic sludge digestion with pure oxygen and air aeration. Bioresource Technology 99, 100e109.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 1 1 e4 3 4 0
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Bioassays as a tool for evaluating advanced oxidation processes in water and wastewater treatment Luigi Rizzo* Department of Civil Engineering, University of Salerno, via Ponte don Melillo 1, 84084 Fisciano (SA), Italy
article info
abstract
Article history:
Advanced oxidation processes (AOPs) have been widely used in water and wastewater
Received 5 April 2011
treatment for the removal of organic and inorganic contaminants as well as to improve
Received in revised form
biodegradability of industrial wastewater. Unfortunately, the partial oxidation of organic
24 May 2011
contaminants may result in the formation of intermediates more toxic than parent
Accepted 29 May 2011
compounds. In order to avoid this drawback, AOPs are expected to be carefully operated
Available online 12 June 2011
and monitored, and toxicity tests have been used to evaluate whether effluent detoxification takes place. In the present work, the effect of AOPs on the toxicity of aqueous
Keywords:
solutions of different classes of contaminants as well as actual aqueous matrices are
Antibiotics
critically reviewed. The dualism toxicityebiodegradability when AOPs are used as pre-
Biodegradability
treatment step to improve industrial wastewater biodegradability is also discussed. The
Drinking water
main conclusions/remarks include the followings: (i) bioassays are a really useful tool to
Dyes
evaluate the dangerousness of AOPs as well as to set up the proper operative conditions, (ii)
Emerging contaminants
target organisms for bioassays should be chosen according to the final use of the treated
Endocrine disruptors
water matrix, (iii) acute toxicity tests may be not suitable to evaluate toxicity in the
Industrial wastewater
presence of low/realistic concentrations of target contaminants, so studies on chronic
Oxidation intermediates
effects should be further developed, (iv) some toxicity tests may be not useful to evaluate
Pesticides
biodegradability potential, in this case more suitable tests should be applied (e.g., activated
Pharmaceuticals
sludge bioassays, respirometry).
Photocatalysis
ª 2011 Elsevier Ltd. All rights reserved.
Toxicity Urban wastewater Xenobiotics
Contents 1.
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.1. Bioassays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.1.1. Invertebrate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.1.2. Plants and algae . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.1.3. Microbial bioassays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.1.4. Fish bioassays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
* Tel.: þ39 089 969334; fax: þ39 089 969620. E-mail address:
[email protected]. 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.05.035
4312 4313 4313 4314 4314 4315
4312
2.
3.
4.
5.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 1 1 e4 3 4 0
1.2. Advanced oxidation processes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.3. Oxidation intermediates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Xenobiotics degradation by AOPs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1. Pharmaceuticals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2. Dyes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3. Pesticides . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Wastewater treatment by AOPs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1. Advanced treatment of urban wastewater . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2. Industrial wastewater treatment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2.1. Toxicity vs. biodegradability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2.2. Olive oil mill wastewater . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2.3. Textile wastewater . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2.4. Tannery wastewaters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2.5. Pesticides wastewaters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2.6. Pulp and paper mill wastewaters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Drinking water treatment by AOPs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1. NOM removal and DBPs control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2. Xenobiotics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.3. Microcystins . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusive remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Abbreviations AOC AOPs BOD5 BPA COD DBPs DOC EC EC50 EG FELST GAC ISO LC50 MBR MC
1.
assimilable organic carbon advanced oxidation processes biochemical oxygen demand after 5 days bisphenol A chemical oxygen demand disinfection by-products dissolved organic carbon emerging contaminants half maximal effective concentration energy gap fish early life stage toxicity test granular activated carbon international organization for standardization half maximal lethal concentration membrane biological reactor microcystins
Introduction
Advanced oxidation processes (AOPs) have been widely used in water and wastewater treatment for the removal of organic and inorganic contaminants as well as to improve biodegradability of industrial wastewater. In the last years different AOPs have been investigated in the removal of emerging contaminants from urban wastewater effluents (Baumgarten et al., 2007; Naddeo et al., 2009; Klamerth et al., 2010) and drinking water (Brose´us et al., 2009; Sanches et al., 2010). Unfortunately, the
4315 4316 4316 4316 4317 4317 4319 4319 4319 4319 4320 4323 4325 4325 4326 4329 4329 4330 4331 4334 4334 4334
NOAEL NOM OECD
no observed adverse effect level natural organic matter organization for economic cooperation and development OMW olive mill wastewater OUR oxygen uptake rate TOC total organic carbon TSS total suspended solids UF ultrafiltration membrane USEPA United States Environmental Protection Agency ultraviolet absorbance at xxx nm wavelength UVxxx UWWTP urban wastewater treatment plant YES yeast estrogen screen EDCs endocrine disrupting chemicals SUVA254 specific UV absorbance at 254 nm THMFP trihalomethanes formation potential
partial oxidation of organic contaminants may result in the formation of intermediates more toxic than parent compounds. In order to avoid this drawback, AOPs are expected to be carefully operated and monitored, and toxicity tests have been used to evaluate whether effluent detoxification takes place (Rizzo et al., 2009a; Klamerth et al., 2010). In the present work, the effect of AOPs on the toxicity of aqueous solutions of different classes of contaminants (such as dyes, pesticides and pharmaceuticals) as well as actual water (such as urban wastewater and drinking water) are critically reviewed.
4313
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 1 1 e4 3 4 0
AOPs have also been used as pre-treatment of industrial wastewater to improve biodegradability before the subsequent biological process (Arslan Alaton and Teksoy, 2007; Rizzo et al., 2008; Zapata et al., 2010). In this regard, sometime toxicity tests have been used to infer the behavior of treated wastewater in terms of biodegradability; but this approach may result in a misinterpretation of the effect of AOPs on the biodegradability of wastewater. Accordingly, some papers are also reviewed to elucidate the differences between toxicity and biodegradability when AOPs are used as pre-treatment step to improve industrial wastewater biodegradability. Moreover, the most used bioassays and AOPs are first introduced in this chapter. In particular it was believed important to provide the reader with some information about the main organisms used in the environmental applications as well as to summarize the corresponding standardized and most used methods.
1.1.
Bioassays
Bioassays rely on measuring the response of organisms exposed to contaminants relative to a control. They have been used to establish the toxicity levels of target contaminants and complex aqueous matrices (e.g., surface water, groundwater, wastewater) for aquatic organisms. The test organisms incorporated in these bioassays can be grouped in (Tothill and Turner (1996), Farre´ and Barcelo´ (2003)): microorganisms, plants and algae, invertebrates and fishes. Table 1 summarizes just (i) the most commonly used organisms to characterize toxicity of water and wastewater, (ii) the main organisms for each one group, (iii) the methods and (iv) some applications to water, wastewater and liquid waste (e.g., landfill leachate).
1.1.1.
Invertebrate
Invertebrates are widely used in the evaluation of toxic effects of pollutants in aqueous matrices. The most used organism to
Table 1 e Toxicity tests typically used in assessing risks to human health and aquatic life after water and wastewater treatment. Group Invertebrate
Plants and algae
Organism
Method
Daphnia magna
USEPA, 2002; ISO, 1996a
Sea urchin (Paracentrotus lividius and Sphaerechinus granularis) Brine shrimp (Artemia salina)
Pagano et al., 1982
Applications Disinfection of Hospital wastewater (Emmanuel et al., 2004) Drinking water treatment (Rizzo et al., 2005) Industrial wastewater treatment (Oral et al., 2007) Advanced treatment of urban wastewater (Rizzo et al., 2009a) Urban wastewater treatment (Hernando et al., 2005) Landfill leachate treatment (Marttinen et al., 2002) Industrial wastewater treatment (Meric¸ et al., 2005; Oral et al., 2007)
Persoone and Van Haecke, 1981; Migliore et al., 1997
Industrial wastewater treatment (Campos et al., 2002; Rodrigues et al., 2008; Pala´cio et al., 2009) Landfill leachate treatment (Silva et al., 2004)
Scenedesmus subspicatus Selenastrum capricornutum
ISO, 1989 ISO, 1989
Dunaliella tertiolecta Raphidocelis subcapitata Lettuce seeds (Lactuca sativa)
USEPA, 1988 Marttinen et al., 2002 OECD, 1984; USEPA, 1989
Industrial wastewater treatment (Tisler et al., 2004) Industrial wastewater treatment (Walsh et al., 1980; Tisler et al., 2004; Oral et al., 2007) Urban wastewater treatment (Hernando et al., 2005) Industrial wastewater treatment (Meric¸ et al., 2005) Landfill leachate treatment (Marttinen et al., 2002) Industrial wastewater treatment (Pala´cio et al., 2009)
Microrganisms Pseudomonas fluorescens strain P17 AOC method Drinking water quality and treatment and Spirillum sp. strain NOX (Van der Kooij et al., 1982; (Vrouwenvelder et al., 1998; Polanska et al., 2005; APHA/AWWA/WEF, 1998) Hammes et al., 2006; Lautenschlager et al., 2010) Wastewater reuse (Weinrich et al., 2010) Vibrio fischeri ISO, 1998 Disinfection of Hospital wastewater (Emmanuel et al., 2004) Industrial wastewater treatment (Tisler et al., 2004) Urban wastewater treatment (Hernando et al., 2005) Bacteria from activated sludge Respiration inhibition test Urban wastewater treatment (ISO, 2007; OECD, 2010) (Dalzell et al., 2002; Pagga et al., 2006) Industrial wastewater treatment (Burgess et al., 1999; Mert et al., 2010) Fishes
Zebrafish (Danio rerio) Rainbow trout (Oncorhynchus mykiss)
ISO, 1996b OECD, 1992
Industrial wastewater treatment (Tisler et al., 2004) Disinfected drinking water (Ferraris et al., 2005) Urban wastewater treatment plant effluent (Gagne´ et al., 2006) Advanced treatment of urban wastewater (Stalter et al., 2010)
4314
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 1 1 e4 3 4 0
characterize toxicity of water and wastewater treatment effluents is Daphnia magna. Acute lethality tests with Daphnia are well established and standardized (USEPA, 2002; ISO, 1996a). The organisms are exposed to target contaminants or aqueous matrices under controlled conditions and living (mobile) daphnias are counted after the required incubation period. Acute (24 and a 48 h) and chronic (21 days) toxicity tests with daphnids have also been reported (Tisler et al., 2004). The use of daphnids has many advantages for routine toxicity testing, such as high sensitivity to toxicants, short reproductive cycle and parthenogenetic reproduction (Tothill and Turner, 1996). Other tests have been developed such as those based on Artemia salina and sea urchin. A. salina serves as test organism in a wide range of toxicological assays and research; it was used in the screening of bioactive compounds in natural products, detection of cyanobacterial and algal toxicity in water, detection of anthropogenic chemicals in the environment and investigations into biochemical processes mediating acute toxic responses (Ruebhart et al., 2008). The Artemia bioassay is attractive for different reasons, including (i) the commercial availability of the cysts, (ii) Artemia can be maintained indefinitely in the laboratory in their cyst form and is easily induced to hatch, (iii) the assay is quick, simple, and performed at low cost, (iv) it requires small sample volume and can be performed with high sample throughput (microplates), and (v) it complies with animal ethics guidelines in many countries (Ruebhart et al., 2008). Sea urchin embryos have been either used to evaluate marine pollution or to test toxicity of specific pollutants and non-marine complex mixtures (Pagano et al., 2001; De Nicola et al., 2004). In addition, since tannery wastewater are typically characterized by high salinity, toxicity to sea urchin of tannery wastewater was also investigated (Meric¸ et al., 2005). Because of the high sensitivity of some invertebrates to high polluted aqueous matrices, such as industrial wastewater, these organisms may be not useful to characterize toxicity; in Section 3.2.1, the role of toxicity tests in industrial wastewater treatment by AOPs is better explained and discussed.
1.1.2.
Plants and algae
Bioassays based on plants are characterized by low maintenance cost and different assessment endpoints (e.g., germination rate, biomass weight, enzyme activity). For instance, Valerio et al. (2007) evaluated the sensitivity of lettuce in seed germination, root elongation, germination rate and root necrosis by exposing lettuce to different concentrations of soluble elements in soilewater solutions, as an alternative way to determine soil toxicity. Unfortunately, tests based on growth responses of plants require a long time, generally 4e6 days for root length measurements. Plant based bioassays have been used to evaluate the toxicity of organic and inorganic contaminants (Date et al., 2005; Wieczorek and Wieczorek, 2007; Di Salvatore et al., 2008), nanoparticles (Lin and Xing, 2007), contaminated soils (Robidoux et al., 2004; Zorrig et al., 2010), solid waste and sludges (Renoux et al., 2001). Due to their ubiquity and short life cycle, which make algae suitable for toxicological studies, toxicity tests based on these
organisms have also been developed (Joubert, 1980; Wong et al., 1995; Radix et al., 2000; Pehlivanoglu and Sedlak, 2004). At the end of the exposure time, the number of algae is assessed using an automatic particle counter and inhibition of algal growth is used as the indicator of toxicity. The main disadvantages of algal methods are difficulty in culturing and, sometimes, lack of reproducibility (Farre´ and Barcelo´, 2003).
1.1.3.
Microbial bioassays
A wide variety of microbial techniques has been developed and is used as toxicity screening procedures. These bioassays use different mechanisms based on (i) capacity of microorganisms to transform carbon, sulfur or nitrogen, (ii) enzymatic activity, (iii) growth, mortality or photosynthesis, (iv) glucose uptake activity, (v) oxygen consumption and (vi) luminescence output (Tothill and Turner, 1996). Among these assimilable organic carbon (AOC) test from one side and respiration inhibition test and luminescent microbial tests from the other find wide application in drinking water and wastewater toxicity characterization respectively. The AOC bioassay can be considered a measurement of the potential of a given water to support bacterial regrowth. Bacterial regrowth was found to be significantly limited for AOC values lower than a few decades of mg L1 (Van der Kooij, 1990; LeChevallier et al., 1991). In AOC test, water samples are inoculated with Pseudomonas fluorescens strain P17 and Spirillum sp. strain NOX up to achieve a given initial bacterial density. Inoculated samples are incubated and bacterial growth is periodically monitored by spread plating. The activated sludge respiration inhibition test has been established as an effective method for evaluating the toxicity of chemicals to activated sludge bacteria. This method was standardized since the late 80s by ISO and OECD, and today updated guidelines are available (OECD, 2010; ISO, 2007). Respiration inhibition test evaluates the effects of a substance on microorganisms from activated sludge by measuring their respiration rate in the presence of different concentrations of the test substance. The respiration rates of samples of activated sludge fed with synthetic sewage are measured in an enclosed cell containing an oxygen electrode after a contact time of 3 h (OECD, 2010). In order to investigate realistic exposure scenario, longer or shorter (e.g., in the presence of volatile substances or rapidly degraded abiotically via hydrolysis 30 min can be used) contact times could be appropriate. The sensitivity of activated sludge should be checked in parallel with a suitable reference substance. The test is typically used to determine the ECx (e.g. EC50) of the test substance and/or the NOAEL (No observed adverse effect level). Luminescent microorganisms have been used in the production of several toxicity test instruments. The most used microorganism is the marine bioluminescent bacterium, Vibrio fischeri. The test relies on the change in the bacterial luminescent when the microorganisms are exposed to toxic chemicals. The bioluminescence inhibition of the V. fischeri test has been standardized (ISO, 1998) and it is commercially available in different versions. The advantages of those toxicity tests include short time of analysis and simplicity of operation. The bacteria are provided by manufacturers in a lyophilized form and they can be stored for several months to be used “on demand”.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 1 1 e4 3 4 0
1.1.4.
Fish bioassays
The purpose of the acute toxicity test with fish species is to help in the assessment of possible risk to similar species in natural environments, which may be due to, for instance, the disposal of wastewater treatment plants effluents into aquatic environments. The rainbow trout (Oncorhynchus mykiss), and bluegill sunfish (Lepomis macrochirus), are preferred species to meet this requirement since they are sensitive indicator species and a large data base which characterizes the response to environmental contaminants is available, but other species as identified in USEPA guideline (USEPA, 1996) may be used. The goal of the test is to determine concentrationeresponse curves for fish mortality (LC50), for 1e4 days in a static, static-renewal, or flow-through system. Generally, fish bioassays show good sensitivities but they have some drawbacks such as standardization problems, time consuming and they need specialized equipment and operators with adequate skills (Farre´ and Barcelo´, 2003). In spite of in vitro screening methods have been preferred due to logistical, cost, time constraints and reliability, chronic in vivo tests (such as fish early life stage toxicity test with rainbow trout) have been also investigated to achieve a more comprehensive information about toxicity of oxidation intermediates formed upon ozonation of urban wastewater (Stalter et al., 2010).
1.2.
Advanced oxidation processes
AOPs can be broadly classified as oxidation methods which promote the formation of highly reactive species, such as hydroxyl radicals (the strongest oxidants (E0 ¼ 2.8 V) after fluorine), which allow the degradation of organic and inorganic pollutants. AOPs include a combination of oxidants (e.g., H2O2; ozone), UV radiation, catalysts (e.g., Fe2þ, TiO2) and ultrasounds. Among the most investigated AOPs in water and wastewater treatment there are the followings: heterogeneous and homogeneous photocatalysis based on near ultraviolet (UV) or solar visible irradiation, ozonation, Fenton reaction, ultrasound, electrochemical and wet-air oxidation. AOPs can be used in wastewater treatment in order to (Fig. 1): 1. improve the quality of urban wastewater treatment plant (UWWTP) effluent by removing residual xenobiotics in order to decrease final toxicity and make finished wastewater reusable (Fig. 1a); 2. disinfect biologically treated urban wastewater to be reused as alternative to conventional chemical disinfectants (such
4315
as chlorine, chlorine dioxide and ozone) which result in formation of toxic disinfection by-products (Fig. 1a); 3. increase industrial wastewater biodegradability before conventional biological process (Fig. 1b); 4. remove or convert in their nontoxic forms metals can be found in industrial wastewater. Industrial wastewater can be characterized by high chemical oxygen demand (COD) concentration. AOPs can be suitably used for the treatment of industrial wastewater with relatively small COD (<5.0 g L1), since higher COD values would require the consumption of too large amounts of expensive reactants (Andreozzi et al., 1999). Higher COD wastewater can be more conveniently treated by means of wet oxidation or incineration (Mishra et al., 1995). Some of the most used AOPs for water and wastewater treatment are shortly introduced. The combination of UV radiation with hydrogen peroxide results in the formation of hydroxyl radicals thus improving the degradation of target contaminants compared to photolysis as stand alone process (Ledakowicz et al., 2001; Andreozzi et al., 2003; Chen et al., 2006). The oxidation of organic and inorganic compounds during ozonation can occur via ozone, HO$ or their combination (Von Gunten, 2003). The oxidation pathway is determined by the ratio of ozone and HO$ concentrations and the corresponding kinetics. Unlike of acidic conditions which favor the development of direct ozone reactions, at higher pH the formation of hydroxyl radicals increases thus effecting process efficiency (Beltra´n et al., 2001; Muthukumar et al., 2004; Coca et al., 2005). Treatment efficiency may also be improved if ozone is used in combination with light irradiation (Amat et al., 2005; Rivas et al., 2009) and hydrogen peroxide ¨ tker, 2003; Rosal et al., 2008). (Balcıoglu and O Fenton reaction occurs in the presence of ferrous or ferric ions and hydrogen peroxide via a free radical chain reaction which produces hydroxyl radicals. Process efficiency was found to be strongly related to pH values, the optimum one being around 3, mainly because of Fe(III) speciation formed (Pignatello et al., 2006). Moreover, efficiency may be enhanced as well if the reaction takes place under UV irradiation (so-called photo-Fenton process) (Shemer et al., 2006; Rizzo et al., 2008; Kusic et al., 2009), as more hydroxyl radicals are produced. Photocatalysis is a photochemical reaction accelerated by the action of a catalyst. If the catalyst occurs in a different phase compared to the reaction medium, the process is
Fig. 1 e Application of AOPs to wastewater treatment: advanced treatment in UWWTP (a); pre-oxidation to increase biodegradability before conventional biological process in industrial wastewater treatment plants (b).
4316
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 1 1 e4 3 4 0
known as heterogeneous photocatalysis. When photocatalyst is irradiated by a light flux whose energy or a portion of, is equal to or higher than photocatalyst band-gap energy EG (hn EG), electrons move from valence to conduction band. Simultaneously, electron vacancies or holes are created in the valence band. The electron/hole pairs (eehþ) thus created migrate to the catalyst surface where they either recombine or participate in redox reactions. Although different catalyst have been investigated (e.g., ZnO, ZrO2, CdS, MoS2, Fe2O3, WO3) in heterogeneous photocatalysis applications to water and wastewater treatment, TiO2 has been the most widely used because it is that one better match photocatalyst proprieties: (i) photoactivity, (ii) ability to use visible and/or near UV radiation, (iii) not affected by photocorrosion, (iv) biologically and chemically inert, (v) nontoxic, and (vi) cheap. In ultrasound irradiation (or sonolysis) reactions take place upon high-intensity acoustic irradiation of liquids at frequencies that produce cavitation (typically in the range 20e1000 kHz). The cavitation process proceeds through the formation, growth and subsequent collapse of microbubbles or cavities occurring in extremely small intervals of time and releasing large amount of energy (Gogate, 2002). Cavitation results in a simultaneous release of reactive radicals too. Several factors may affect the degradation/oxidation of pollutants by cavitation, among these the most important ones are: the frequency and intensity of ultrasound, reactor geometry, type and nature of contaminant, bulk temperature and the water matrix (Gogate and Pandit, 2004a). The use of hydrogen peroxide or ozone is expected to enhance the rates of degradation because of (i) the generation of additional free radicals, (ii) enhanced ozone diffusion by mechanical effects of ultrasound, and (iii) the photolysis of ultrasound-generated H2O2 to produce hydroxyl radicals (Tezcanli-Guyer and Ince, 2004; Gogate and Pandit, 2004b; Kidak and Ince, 2007). Electrochemical oxidation involves the formation of hydroxyl radicals at the active sites of anode and has been used for the decontamination of various inorganic and organic pollutants (Comninellis et al., 2008; Klavarioti et al., 2009; Matilainen and Sillanpa¨a¨, 2010). Different anodes, in the presence of a suitable electrolyte (typically NaCl), have been investigated: graphite, Pt, TiO2, IrO2, PbO2, several Ti-based alloys and boron-doped diamond electrodes (Comninellis et al., 2008). The efficiency of electrochemical process depends on electrode and supporting electrolyte types, applied current, solution pH, nature of target contaminant/ water matrix and initial concentration of the pollutants. Wet-air oxidation is a thermochemical process where hydroxyl radicals and other active oxygen species are formed at high temperatures (i.e. 200e320 C) and pressures (i.e. 2e20 MPa) in the presence of air or pure oxygen. The basic assumption is that the reactivity of oxygen, or its capacity to oxidize, increases as temperature increases. At temperatures and pressures above the critical point of water (374 C, 22 MPa), the process is referred to as supercritical water oxidation, with its main feature being that gas and liquid phases form a homogeneous single phase. The process has great potential for the treatment of industrial wastewater with a moderate to high organic loading (i.e. 10e100 g L1 COD) (Levec and Pintar, 2007). Because of high temperature
and pressure conditions, wet-air oxidation is not an economically viable option.
1.3.
Oxidation intermediates
The advanced oxidation of complex organic contaminants typically does not result in a fast mineralization, with formation of carbon dioxide and inorganic species, but oxidation intermediate products form. These products may be more toxic than parent compounds so toxicity tests are a very useful tool to evaluate the operative AOP conditions to make the treated aqueous matrix safer. Some papers reviewed the formation of oxidation intermediates of pesticides (Konstantinou and Albanis, 2003), endocrine disrupting chemicals (EDCs), pharmaceuticals and personal care products (Esplugas et al., 2007), and dyes (Konstantinou and Albanis, 2004) over AOPs treatment. The main intermediates of pesticide degradation by TiO2 photocatalysis can be grouped in the following five products (Konstantinou and Albanis, 2003): (i) hydroxylated products and derivatives usually after dehalogenation of the parent pesticide, if halogen substituents are present (ii) products of oxidation of the side chain, if present (iii) ring opening products for aromatic pesticides (iv) decarboxylation products and (v) isomerization and cyclization products. Major identified intermediates in TiO2 photocatalytic degradation of azo dyes in aqueous solution were found to be hydroxylated derivatives, aromatic amines, naphthoquinone, phenolic compounds and several organic acids (Konstantinou and Albanis, 2004).
2.
Xenobiotics degradation by AOPs
AOPs have been investigated in the removal of a wide range of xenobiotics (particularly pharmaceuticals, dyes, and pesticides) from aqueous suspensions in order to characterize (i) degradation kinetics of target contaminants, (ii) oxidation intermediates and (iii) final toxicity. The oxidation of organic contaminants by AOPs may result in the formation of oxidation intermediates which may be more toxic than parent compounds. In the present chapter several works available in scientific literature, which deal with the degradation and final toxicity of xenobiotics and their oxidation intermediates, (in pure water solutions) are reviewed. The final toxicity depends on the (i) target contaminant and its concentration, (ii) organism selected for toxicity test, (iii) AOP, (iv) AOP operating conditions (e.g., oxidant concentration, catalyst loading, UV radiation source, etc.), and (v) oxidation intermediates. Among the investigated organisms V. fischeri has been the most used. The choice of organism is really important because some of them may be not enough sensitive to the target contaminant and its oxidation intermediates (Andreozzi et al., 2002; Rizzo et al., 2009b)
2.1.
Pharmaceuticals
Pharmaceuticals, unchanged or as metabolites, are continuously released into sewage systems through urine and feces. Several pharmaceuticals are not biodegradable therefore they cannot be effectively removed by conventional biologic
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 1 1 e4 3 4 0
process typically used in UWWTPs. This uncontrolled effluents discharge results in a serious threat to the aquatic environment related to (i) the development of antibiotic resistant bacteria, (ii) retardation of nitrite oxidation and methanogenesis, (iii) potential increased toxicity due to synergic effects of different chemicals and metabolites (Dalrymple et al., 2007). The investigation of pharmaceuticals removal by AOPs attracted a lot of attention in the last years. In Table 2, the effect of AOPs on final toxicity of pharmaceutical solutions and oxidation intermediates is summarized. Although the initial concentrations of target pharmaceuticals investigated are high compared to the concentrations typically detected in water and wastewater, the results are useful to characterize the relative toxicity. This means that, according to final toxicity, the AOPs operating conditions can be optimized as well as the best AOP can be selected among different AOPs comparatively investigated. Among the organisms investigated V. fischeri has been the most used because of its high sensitivity toward a wide range of pollutants. D. magna was also found to be sensitive enough to characterize process efficiency as well as oxidation intermediates formation. Pseudokirchneriella subcapitata (formerly Selenastrum capricornutum) was found to be more sensitive than D. magna to characterize toxicity of photocatalytically treated diclofenac solutions; on the opposite, A. salina was not found to be sensitive (Rizzo et al., 2009b). Moreover, when oxidation intermediates have been investigated in parallel, a relation with toxicity was observed (Calza et al., 2006; Sakkas et al., 2007; Me´ndez-Arriaga et al., 2008). Typically, AOPs treatment has been extended up to achieve a decrease of initial toxicity; moreover, an increase of toxicity in the early min of the treatment has been observed because of the formation of oxidation intermediates more toxic than parent compounds (Table 2). Photo-Fenton decreased toxicity of dipyrone and its main intermediates better than TiO2 photocatalysis, under solar radiation (Pe´rez-Estrada et al., 2007). Complete disappearance of dipyrone main metabolite (methylaminoantipyrine, MAA) was already observed during the dark Fenton reaction within 15 min after hydrogen peroxide was added. In parallel, TOC only decreased by around 10%, but as solar radiation started, TOC decreased very fast (around 60%) in the next 20 min. Accordingly, toxicity to V. fischeri was found to decrease below initial value. On the opposite, TiO2 photocatalytic treatment was less effective compared to photo-Fenton process: MAA removal increased up to 70% after 20 min irradiation (completely disappeared after 60 min), but TOC only decreased by around 30% in the same time. So, the main oxidation products generated by TiO2 photocatalytic treatment were still present at high concentrations thus resulting in an increased toxicity to V. fischeri compared to photo-Fenton treatment. Similar conclusions can be drawn by Andreozzi et al.’s (2004, 2006) works, where ozonation process was found to be less toxic when compared to other AOPs because it was more effective in the removal of the target pharmaceuticals and their oxidation intermediates.
2.2.
Dyes
Over 0.7 million tons of organic synthetic dyes are manufactured each year in the world mainly for use in textile, leather
4317
goods, industrial painting, food, plastics, cosmetics, and consumer electronic sectors (Rajeshwar et al., 2008). A significant amount of these is lost during the dying process and released into wastewater stream. According to application to textiles, dyes can be classified as acid, azo, reactive, metal complex, disperse, direct, vat, mordant, basic and sulfur dyes. Since reactive dyes represent an increasing market share, and taking into account that a large fraction is released into wastewater due to dye hydrolysis in the alkaline dyebath, a lot of works available in scientific literature deal with the oxidation of reactive dyes by AOPs. On the opposite, only few organisms have been investigated to characterize toxicity of dyes solutions, among these V. fischeri and Pseudomonas putida (Table 3). TiO2 photocatalytic degradation of two dyes (50 mg L1) was investigated in the presence of two different oxidants, hydrogen peroxide (H2O2) and potassium peroxydisulfate (K2S2O8) (Bizani et al., 2006). A complete decolorization was achieved in 120 min with 1 g TiO2 L1, but in the presence of H2O2 (3 103 mol L1) 60 min irradiation was enough. The decolorization process resulted in the formation of quite recalcitrant oxidation intermediates because mineralization was significant (85e90% DOC removal) only after 5-h irradiation time. According to decolorization and mineralization results, the lower toxicity to V. fischeri was observed in the presence of H2O2. UV/TiO2, electro-Fenton, wet-air oxidation and UV/electro-Fenton were compared for the degradation of Reactive Black 5 azo dye in aqueous solution (Kusvuran et al., 2005). In spite of wet-air (250 C) and UV/TiO2 (0.5 g TiO2 L1) processes were found to be the most effective methods for decolorization and mineralization (77 and 71% TOC removals respectively after 180 min treatment) of RB5 (100 mg L1), mineralization was not enough to achieve a decreased toxicity to P. putida compared to untreated solution. Typically, Fenton reaction based processes were found to be effective in the toxicity removal in the most of investigated dye solutions (Neamtu et al., 2004; Garcia-Montano et al., 2006) and when compared to others AOPs, a higher efficiency in decreasing toxicity was observed (Kusvuran et al., 2005; Arslan Alaton and Teksoy, 2007).
2.3.
Pesticides
Pesticides can be classified as herbicides, fungicides, acaricides, and insecticides according to their specific biological activity on target species. Moreover, they can also be classified by their chemical composition (e.g., phenylurea and phenoxyacid pesticides, chlorophenolic substances, triazines, triazole fungicides, carbamates, organophosphorous (OP) insecticides, and neonicotinoids). V. fischeri and D. magna have been the most used organisms to evaluate the toxicity of pesticides solutions treated by AOPs; anyway, Bacillus subtilis sp. and P. subcapitata have been successfully investigated in the evaluation of genotoxicity (Segura et al., 2008) and ecotoxicity (Fernandez-Alba et al., 2002) respectively. Several pesticides and initial concentrations were investigated and final toxicity was found to be strongly related to both of them (Table 4). In particular, the limits (long irradiation time, oxidants addition) of TiO2 photocatalysis to detoxify highly polluted aqueous solutions (some tens of mg L1 of
Target pharmaceutical
Investigated AOPs 1
1
1
Bezafibrate (0.2e0.5 mmol L )
O3 (1 mmol L , 0.38 L min
Carbamazepine (3.3 106 mmol dm3)
O3 (2.0% per volume, 36 dm3 h1 flowrate).
Diclofenac (0.8e9.2 mg L1)
TiO2/UV (0.22e0.5 g L1, Degussa P25, 1500 W sunlight simulator, up to 2 h contact time) TiO2/UV (0.2e1.6 g L1 Degussa P25, 125 W black-light fluorescent lamp, up to 2 h contact time)
Diclofenac (5e80 mg L1)
Diclofenac (200 mg L1)
flowrate)
Toxicity tests V. fischeri Ankistrodesmus braunii and S. capricornutum V. fischeri
D. magna, P. subcapitata and A. salina V. fischeri
Solar TiO2/UV (0.2 g TiO2 L1) and photo-Fenton (0.2e0.5 g H2O2 L1) pilot plant (35 L total volume, 22 L illuminated volume, 3.08 m2 irradiated surface).
V. fischeri
Lyncomicin (0.5 mM)
Sunlight irradiation, UV/H2O2 (254 nm low-pressure lamp) and O3.
Mixture of six pharmaceuticals (7.1 mg carbamazepine L1, 2.8 mg diclofenac L1, 11.2 mg clofibric acid L1, 0.6 mg ofloxacin L1, 2.2 mg sulfamethoxazole L1, 0.3 mg propranolol L1) Salbutamol (15 mg L1)
O3 (0.42 mM, 36 dm3 h1 flowrate), H2O2/UV (5 and 10 mM of H2O2) and TiO2/UV (suspended or immobilized TiO2, 0.3 g L1, Degussa, 300 W sunlight simulator, up to 48 h contact time)
P. subcapitata, Cyclotella meneghiniana, Synechococcus leopoliensis. S. leopoliensis, Brachyonus calyciflorus
Diclofenac (200 mg L1), Ibuprofen (25e200 mg L1), naproxen (200 mg L1) Dipyrone (50 mg L1)
Sulfamethoxazole (50 mg L1)
Sulfamethoxazole (30e80 mg L1)
Sulfamethoxazole (200 mg L1) Tetracycline (10e40 mg L1)
TiO2/UV (0.22e0.5 g L1, Degussa P25, 1500 W sunlight simulator, up to 3 h contact time) Solar photo-Fenton (up to 210 mg H2O2 L1, 2.6e10.4 mg Fe L1) pilot plant (35 L total volume, 22 L illuminated volume, 3.08 m2 irradiated surface). UVA radiation, O3, O3/TiO2, O3/UVA, O2/TiO2/UVA andO3/UVA/TiO2 (700 W high-pressure mercury lamp, 1.5 g TiO2 L1) Photo-Fenton (pyrex jacketed thermostatic 2 L vessel equipped with 8 W each three black-light blue lamps). Photolysis (500 W medium mercury lamp, up to 360 min irradiation)
V. fischeri
V. fischeri V. fischeri, D. magna
D. magna
V. fischeri V. fischeri
References
105 min of ozonation was necessary to decrease toxicity below the value of the initial BZF solution. Carbamazepine was found to be not toxic toward S. capricornutum and Ankistrodesmus braunii.
Dantas et al., 2007 Andreozzi et al., 2002
Toxicity first increased (up to 20 min) then almost totally disappeared after 120 min.
Calza et al., 2006
Toxicity on D. magna and P. subcapitata varied during the oxidation, probably due to the formation of intermediate products. A. salina was not found to be sensitive. Ozonation improved biodegradability (up to 0.19 BOD5/COD) and slightly reduced toxicity. Toxicity first increased (up to 120 min) then did not decrease significantly over 240 min.
Rizzo et al., 2009b
Toxicity was found to be below 40% inhibition for all the samples during the photo-Fenton treatment. A higher toxicity (50%) was initially detected for TiO2/UV treated samples. Ozonated samples were found to be less toxic on S. leopoliensis with respect to the parent compound even just for 1 h treatment.
Coelho et al., 2009 Me´ndez-Arriaga et al., 2008 Pe´rez-Estrada et al., 2007
Andreozzi et al., 2006
O3 and H2O2/UV decreased toxicity better than TiO2/UV
Andreozzi et al., 2004
Toxicity first increased (up to 15 min) then almost totally disappeared after 90 min. Photo-Fenton decreased toxicity to D. magna from 85% to 20%.
Sakkas et al., 2007 Trovo´ et al., 2009
O2/TiO2/UVA and O3/UVA/TiO2 resulted to be the most appropriate technologies to eliminate the toxicity of the water. No toxicity was detected after photo-Fenton treatment with 300 and 400 mg L1 of H2O2. The formation of photolysis intermediate products showed an increasing toxicity to V. fischeri.
Beltra´n et al., 2008 Gonza´lez et al., 2009 Jiao et al., 2008
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 1 1 e4 3 4 0
O3 (0.435 g h1 ozone production; best condition: 30 min reaction, ozone dose of 0.22 g L1) TiO2/UV (0.07e1.0 g L1, Degussa P25, 1000 W sunlight simulator, up to 4 h contact time)
Comments
4318
Table 2 e Toxicity tests used in evaluating the degradation of pharmaceuticals and their oxidation intermediates over AOPs treatment.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 1 1 e4 3 4 0
target pollutants) was confirmed by works of Evgenidou and co-workers, who investigated the degradation and final toxicity on V. fischeri of three pesticides. They found out that (i) toxicity increased under all investigated conditions in dichlorvos solutions (Evgenidou et al., 2005a), (ii) it was only slightly reduced after 6 h of treatment in dimethoate solutions (only peroxide addition resulted in a complete detoxification after 120 min treatment) (Evgenidou et al., 2005b), and (iii) a 50% reduction in the toxicity of prometryn solutions was observed (only peroxide addition resulted in a complete detoxification after 120 min treatment) (Evgenidou et al., 2007a). Moreover, according to others studies previously discussed on different target contaminants, when solar driven TiO2 photocatalysis process was compared to solar driven photo-Fenton process in the degradation of pesticides solutions TiO2 photocatalysis was found to be less effective to reduce the final toxicity (Fernandez-Alba et al., 2002; Pe´rezEstrada et al., 2007).
3.
Wastewater treatment by AOPs
3.1.
Advanced treatment of urban wastewater
Several xenobiotics such as pharmaceuticals, endocrine disruptors, personal care products and pesticides have been detected in waste-, surface- and even in drinking waters (Sacher et al., 2001; Kuch and Ballschmiter, 2001; Auriol et al., 2006; Castiglioni et al., 2006). One of the main sources of surface water contamination from xenobiotics is UWWTPs effluents, because several xenobiotics cannot be effectively removed by conventional biological processes (Auriol et al., et al., 2009). Their 2006; Castiglioni et al., 2006; Radjenovic continuous disposal results in a risk for aquatic systems and consequently for human health. In order to prevent this kind of water pollution an advanced treatment downstream of biological process effluent should be implemented. In the last years different AOPs have been investigated both to remove xenobiotics from wastewater and to decrease the final content of organic matter (measured as total organic carbon TOC, chemical oxygen demand COD, and biochemical oxygen demand BOD). Although several studies investigated the removal of xenobiotics from wastewater by AOPs, just a few studies investigated their effect on final toxicity (Table 5). Most of these deal with ozonation or ozone based AOPs (Baumgarten et al., 2007; Petala et al., 2008; Stalter et al., 2010; Reungoat et al., 2010), but homogeneous (Klamerth et al., 2010) and heterogeneous photocatalysis (Rizzo et al., 2009a), sonication (Naddeo et al., 2009) and gamma radiation (Sa´nchezPolo et al., 2009) were investigated too. According to the results available in scientific literature, AOPs either did not reduce toxicity or increase toxicity. A decreased toxicity was observed just when a post-treatment (activated carbon adsorption) was used after oxidation step (Reungoat et al., 2010). Unlike of pure water solutions, wastewaters are quite complex mixtures including a lot of pollutants which can react with either target contaminants or their oxidation intermediates thus resulting in an increased toxicity. Accordingly, although target contaminants can be effectively removed from urban wastewater effluents, it looks quite
4319
difficult to successfully operate AOPs to decrease final toxicity too. In our recent work, we investigated the degradation of 15 emerging contaminants (ECs) at low concentrations in a UWWTP effluent with photo-Fenton (at unchanged pH and low Fe concentration, 5 mg L1) in a pilot-scale solar compound parabolic collector (Klamerth et al., 2010). In spite of the investigated ECs were successfully degraded to negligible concentrations, the final toxicity to V. fischeri increased for real wastewater effluent (Fig. 2). The negative inhibition detected before photo-Fenton process started, was explained by availability of nutrients and salts, which in someway promoted the growth of V. fischeri until the toxic effect of ECs and their oxidation intermediates was neutralized. When photo-Fenton started, and up to 120 min irradiation (around 30 mg L1 H2O2 dose), the degradation of parent pollutants began forming more toxic intermediates, which drastically increased the inhibition rate (Fig. 2, Step 2). In the last step (3) inhibition with real wastewater effluent is quite constant, probably because toxic organic intermediates take longer time to mineralize. Compared to the results of ECs residual concentrations a high removal rate resulted in a correspondingly larger amount of more toxic intermediates.
3.2.
Industrial wastewater treatment
Industrial wastewater include biorecalcitrant and/or toxic molecules which make it refractory to biological treatment. On the opposite, AOPs can successfully remove these compounds as well as they have the potential to mineralize organic pollutants to carbon dioxide, water and mineral salts, without any sludge production. Unfortunately, the complete mineralization of organic pollutants may be really expensive because of the formation of organic oxidation intermediates which may be more refractory to oxidation treatment than parent compounds (Sarria et al., 2002; Mantzavinos and Psillakis, 2004; Zapata et al., 2010). These refractory oxidation intermediates include short-chain organic acids amenable to biological process (they can more easily enter cell wall and consequently they may be more readily biodegraded compared to parent compounds); accordingly, AOPs as pre-treatment step to biological process have been widely investigated (Mantzavinos and Psillakis, 2004). Industrial wastewaters also include heavy metals which are toxic to bacteria because of their valence state or concentration. In this case AOPs can decrease toxicity before biological process. In the following sub paragraphs, before to review the effect of AOPs on biodegradability and toxicity of different industrial wastewaters, some discussion about the dualism toxicity/ biodegradability is given.
3.2.1.
Toxicity vs. biodegradability
The dualism toxicity/biodegradability is particularly relevant when AOPs are investigated in industrial wastewater pretreatment because the main objective is to improve the biodegradability of a mixture of pollutants refractory to biological treatment. In this regard some authors use toxicity tests to infer the behavior of treated wastewater in terms of biodegradability in relation to a subsequent biological process as well as to set up optimum operating conditions of the
4320
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 1 1 e4 3 4 0
investigated AOPs. But this approach is incorrect because, depending on the organisms used in toxicity tests, the results may either underestimate or overestimate the effect of AOPs on the biodegradability of wastewater. Different organisms used for toxicity tests on the same aqueous matrix, because of their diversity, can give different results in terms of toxicity. This is particularly true if we compare organisms from different specie (plant seeds, bacteria, crustaceous, algae, fishes and so on). When a battery of 6 bioassays was used to evaluate non-specific toxicity (V. fischeri) and 5 specific toxic modes of action (estrogenic activity, arylhydrocarbon receptor response, neurotoxicity, phytotoxicity and genotoxicity) in the removal of 85 organic micropollutants by a full scale reclamation plant (including ozonation steps) treating secondary effluent, the process effect was found to vary from one bioassay to another (Reungoat et al., 2010). In our previous work we investigated the TiO2 photocatalytic oxidation of an urban wastewater treatment plant effluent contaminated with a mixture of pharmaceutical compounds through a set of bioassays (D. magna, P. subcapitata and Lepidium sativum) (Rizzo et al., 2009a). The toxicity behavior was found to be similar for D. magna and P. subcapitata up to 20 min treatment time (Fig. 3); but as irradiation time was increased up to 120 min, D. magna showed a still decreasing trend opposite to P. subcapitata which showed a quite constant growth inhibition. Differences in toxicity were also observed when P. subcapitata growth inhibition was compared with phytotoxicity to L. sativum (Fig. 4). Phytotoxicity decreased (increase in germination index) after wastewater sample was spiked with pharmaceutical mixture and up to 5 min treatment, then increased up to 20 min treatment (Fig. 4a); on the opposite, inhibition to P. subcapitata growth increased after spiking and decreased up to 10 min treatment (Fig. 4b). Accordingly, taking into account the different response of organisms to polluting load, we may fall in a mistake if we look for a relationship between toxicity and biodegradability, because we may find out that the investigated aqueous matrix is toxic to organisms used in the bioassay but no toxic to bacteria which promote biodegradation process. Gutie´rrez et al. (2002) compared V. fischeri toxicity tests with electrolytic respirometry to evaluate toxicity/biodegradability of seven organic and five inorganic compounds. In general, V. fischeri showed a higher sensitivity to the investigated contaminants but was less representative of effects on activated sludge compared to respirometry. For instance, bioassays carried out with linear sodium dodecylbenzene sulfonate, a biodegradable surfactant, showed a toxic effect on V. fischeri but good biodegradability and no toxicity in respirometry. The authors explained this result with the different nature of the organisms used for the bioassays: seawater organisms from one side (V. fischeri) and a bacterial consortium from activated sludge (respirometry) from the other. In a subsequent work, Farre´ et al. (2007) characterized biodegradability and toxicity of pesticides aqueous solutions following a combination of ozone and photo-Fenton processes in order to evaluate their suitability for a secondary biological treatment. After they found out BOD5/COD ratios in some disagreement with the toxicity analysis for alachlor and atrazine solutions (in spite of an increased toxicity to V. fischeri after treatment, pesticides solutions were conveniently
degraded with a secondary biological treatment) they carried out respirometric analysis too and they found out the following order of toxicity: pentachlorophenol > diuron > isoproturon > atrazine > alachlor, and the following order for respirometric data: pentachlorophenol > alachlor > atrazine > isoproturon ¼ diuron (the agreement was found only for pentachlorophenol). According to Gutie´rrez et al. (2002), the authors explained the disagreement considering the different nature of the organisms used. Zapata et al. (2009) investigated the effect of solar photo-Fenton treatment of commercial pesticides in water on toxicity (V. fischeri) and biodegradability (ZahneWellens test). In spite of biodegradability threshold was achieved as pesticides active ingredients were almost removed, toxicity was still over 50% inhibition (Fig. 5). Additionally, because of continuous changes in toxicity results along oxidation treatment, the authors inferred that V. fischeri toxicity test is not a suitable way to detect biodegradability threshold. Accordingly, Amat et al. (2009) found out that toxicity to V. fischeri was still high after solar photo-Fenton treatment of a mixture of four commercial pesticides, although the effluent might be compatible with biological process. They concluded that toxicity bioassays could be very useful as preliminary tests to find out the proper time for the application of long-time biodegradability tests. When Schrank and co-workers investigated different AOPs (UV, TiO2/UV, O3 and O3/UV) to improve biodegradability of tannery wastewater they found out that the removal of pollutants with aromatic structures did not increase biodegradability, in spite of toxicity to D. magna was found to be stabile or decreased (Schrank et al., 2004). Actually, already in 1995 Strotmann and co-workers suggested the use of both respiration inhibitory tests (such as heterotrophic respiration activity and nitrification activity tests) for monitoring the biological activity of wastewater treatment plant and toxicity tests (specifically luminescent bacteria based tests) for screening the toxicity of the effluent (Strotmann et al., 1995). They also investigated the effect of shock loading conditions by adding 2.3-dichlorophenol and 3.5-dichlorophenol and observed that nitrification and heterotrophic respiration activities as well as TOC degradation were significantly reduced. Moreover, an increased inhibition of luminescent bacteria in the effluent was observed during the shock loading experiments.
3.2.2.
Olive oil mill wastewater
Mediterranean countries (particularly Spain, Italy and Greece) are the main olive oil producers in the world with 2.5 106 t per year, the 95% of the annual world production (Brenes et al., 1999). The related amount of olive mill wastewaters (OMW) was estimated as high as 30 106 m3 per year (Mantzavinos and Kalogerakis, 2005). OMW are characterized by high organic and total suspended solids (TSS) loads as well as acidic pH (Beccari et al., 2002). The organic matter mainly consists of polysaccharides, sugars, phenols, polyalcohols, proteins, organic acids and oil (Cabrera et al., 1996). The high phenols concentrations results in phytotoxicity and toxicity to wastewater which make them difficult to be treated by biological processes (Borja et al., 1996; Gernjak et al., 2004). In order to solve this problem, several AOPs have been
Table 3 e Toxicity tests used in evaluating the degradation of dyes and their oxidation intermediates by AOPs. Target dye
Investigated AOPs
Toxicity tests V. fischeri
Reactive Black 5 (20e100 mg L1)
UV/TiO2, wet-air oxidation, electro-Fenton and UV/electro-Fenton.
P. putida
Reactive Black 5 (5e300 mg L1)
Two different ultrasound devices (high-frequency plate type (279 and 817 kHz), and a 20 kHz low-frequency probe system). ozonation, Fenton, UV/H2O2, and photo-Fenton
V. fischeri
Photo-Fenton (6 W Philips black-light fluorescent lamp, 1.38 109 Einstein s1 light intensity) Ultrasound (plate type piezoelectric transducer, 520 kHz, 100 W)
V. fischeri
Remazol Brilliant Blue R, Acid Black 1
Fenton oxidation and Fe0/air process
V. fischeri
Acid dyebath consisted of three different acid dyestuffs (C.I. Acid Yellow 242, C.I. Acid Red 360 and C.I. Acid Blue 264, 30 mg L1 each one) and two dye auxiliaries (a leveling agent (1500 mg L1) and an acid donor (500 mg L1))
Fenton
Activated sludge inhibition test
Reactive Red 120 (20e100 mg L1)
Disperse Red 354 (100 mg L1) Procion Red H-E7B (250 mg L1)
Reactive red 141, reactive black 5, basic brown 4, basic blue 3 (20e60 mg L1)
P. putida
V. fischeri
V. fischeri
Although DOC was almost 80% removed after 5 h of illumination, the toxicity of the solutions is slightly decreased. No bacterial growth inhibition was observed after 90 min treatment in all AOP investigated. The toxicity increased in the early 15 min of treatment; only electro-Fenton and UV/electro-Fenton (90 min treatment) decreased toxicity under initial value. No toxicity was detected at 20 mg L1 dye solution as well as no change in toxicity profile was observed, even after 6 h of ultrasound treatment (817 kHz, 100 W) The photo-Fenton process was found to result in the lowest toxicity after 15 min treatment. EC50 values were larger than the DOC content at every measurement time. RR141 and RBk5 were non-toxic at 20 and 30 mg/L, while BBl3 and BBr4 were toxic at both concentrations. Total toxicity removal was accomplished within shorter contact time than that necessary for total dye degradation. The toxicity of Fe0/air-treated solution was significantly lower than that of Fenton-treated solution; no toxicity was detected after treatment by the Fe0/air process. The inhibitory effect of acid dyebath toward sewage sludge can be completely eliminated after Fenton oxidation.
References Bizani et al., 2006 Kusvuran et al., 2004 Kusvuran et al., 2005 Vajnhandl and Le Marechal, 2007 Neamtu et al., 2004 Garcia-Montano et al., 2006 Tezcanli-Guyer and Ince, 2003
Chang et al., 2009
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 1 1 e4 3 4 0
UV/TiO2 (TiO2 Degussa P-25 and Hombikat UV-100, Philips HPK 125 W high-pressure mercury lamp). UV/TiO2, electro-Fenton, wet-air oxidation, and UV/electro-Fenton.
Cibacron Red FN-R and Cibacron Yellow FN2R (50 mg L1)
Comments
Arslan Alaton and Teksoy, 2007
4321
Target pesticide Pirimiphos-methyl (PMM)
Dichlorvos (10e50 mg L1)
Dimethoate (5e50 mg L1)
Prometryn (20 mg L1)
Pentachlorophenol (50 mg L1), chlorfenvinfos (50 mg L1), isoproturon (50 mg L1), diuron (42 mg L1), alachlor (50 mg L1) and atrazine (38 mg L1) Pentachlorophenol (50 mg L1), isoproturon (50 mg L1), diuron (42 mg L1), alachlor (50 mg L1) and atrazine (38 mg L1) A mixture of 5 pesticides (alachlor, atrazine, chlorfenvinphos, diuron, isoproturon) Methomyl (50 mg L1)
Methylparathion (10 mg L1) Methylparathion (10e20 mg L1)
Toxicity tests
Comments
References
TiO2 photocatalysis (simulation of solar radiation by 125 W high-pressure mercury lamp equipped with 335 nm cut-off filter transmitting wavelengths > 340 nm). Heterogeneous photocatalysis (125 W high-pressure mercury lamp) using TiO2 and ZnO as catalysts þ oxidants (H2O2 or K2S2O8). Heterogeneous photocatalysis (125 W high-pressure mercury lamp) using TiO2 and ZnO as catalysts þ oxidants (H2O2 or K2S2O8).
V. fischeri
The toxicity of PMM solution was found to increase in the early min than slowly decreased.
Herrmann et al., 1999
V. fischeri
Evgenidou et al., 2005a
Photo-assisted Fenton reaction (500 mL Pyrex reactor, 125 W high-pressure mercury lamp, 30e35 C temperature, 10e100 mg H2O2 L1, 0e6 mg Fe3þ L1) TiO2 photocatalysis (125 W high-pressure mercury lamp) þ oxidants (H2O2 or K2S2O8).
V. fischeri
Toxicity increased under all investigated conditions (6 h irradiation time, 0.1 g TiO2/L, 0.5 and 0.1 g ZnO/L) with and without oxidants. In the TiO2 system the toxicity was only slightly reduced after 6 h of treatment; only the addition of peroxide was able to achieve complete detoxification (120 min). Zinc dissolution and photodissolution increased toxicity in ZnO system. A complete and almost complete detoxification was achieved within 60 min of illumination for dimethoate and methylparathion respectively.
Photo-Fenton/ozone (PhFO) and TiO2photocatalysis/ozone (PhCO). 1.6 g h1 O3 dose and 6 W black-light lamp.
V. fischeri
O3 (1.75 g h1) photo-Fenton (6 W black-light)
V. fischeri
A decreased toxicity (increase of the EC50 value) for diuron, while even after 3 h treatment atrazine and alachlor solutions remain toxic.
Farre´ et al., 2007
Photo-Fenton (Hanau Suntest Simulator equipped with a xenon lamp with total radiant flux of 80 mW cm2). Solar driven photo-Fenton and TiO2 photocatalysis pilot plant
V. fischeri
A sharp decrease in toxicity was observed at the beginning of the photo-Fenton process, followed by a stable toxic level during all the rest of the photo-treatment. Although a clear decrease in water toxicity treated by both processes was observed, EC50 was not clearly beneath threshold after pesticide disappeared. Toxicity on V. fischeri and D. magna was reduced almost completely after 90 min photocatalytic treatment. Toxicity was significantly decreased (over 80%), faster for V. fischeri (30 min) than D. magna (90 min). The formation of malaoxon, isomalathion or trimethyl phosphate esters correlated well with the induced toxicity which was observed in photocatalysis of malathion and radotion. Photo-Fenton was found to decrease toxicity better than TiO2 photocatalysis.
Lapertot et al., 2007
TiO2 photocatalysis (six 20 W UV-A lamps, the light intensity of each one lamp was 2.4 mW cm2) Solar driven TiO2 photocatalysis
Malathion (14.7 mg L1), radotion (11.3 mg L1), malaoxon (18.0 mg L1) and isomalathion (15.7 mg L1)
TiO2 photocatalysis (six 20 W low-pressure mercury fluorescent lamps)
Dipyrone (50 mg L1)
Solar driven photo-Fenton and TiO2 photocatalysis pilot plant
V. fischeri
V. fischeri
V. fischeri, D. magna and S. capricornotum V. fischeri and D. magna V. fischeri and D. magna Inhibition of acetylcholinesterase
V. fischeri
After a slight increase in the toxicity a 50% reduction was observed after 6 h. In presence of H2O2 a 100% toxicity reduction was achieved after 2 h. For all investigated pesticides but alachlor, PhFO first increases toxicity than decreases (after 60e120 min treatment depending on specific pesticide).
Evgenidou et al., 2005b
Evgenidou et al., 2007b
Evgenidou et al., 2007a Farre´ et al., 2005
Fernandez-Alba et al., 2002 Kim et al., 2006 Zoh et al., 2006 Kralj et al., 2007
Pe´rez-Estrada et al., 2007
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 1 1 e4 3 4 0
Dimethoate and methylparathion (10 mg L1)
Investigated AOPs
4322
Table 4 e Toxicity tests used in evaluating the degradation of pesticides and their oxidation intermediates by AOPs.
Gonze et al., 1999 Toxicity to V. fischeri first decreased (2 h, 220 kW m3) due to PCP removal then increased (up to 10 h) because of oxidation intermediates and H2O2 formation. Similar behavior was observed for D. magna. D. magna and V. fischeri
Kim et al., 2007 TiO2 photocatalysis with UV-B radiation was the most effective method to remove PCP and decrease toxicity.
Sodium pentachlorophenate (PCP) (0.1 mM solutions)
Pentachlorophenol (10 mg L1)
D. magna and Bacillus subtilis sp.
Photo-Fenton (2 L reactor, three 6 W Philips black-light fluorescent lamps (I ¼ 5 106 Einstein s1), controlled temperature (25 C) Photolysis and photocatalytic indoor (six-columns continuous flow photoreactor equipped with 6 UV lamps) and outdoor solar (eight quartz tube modules) reactors Ultrasound (electrical generator: frequency 500 kHz, power 0e100 W). Temperature was maintained constant (20 C), pH in the range 6.8 7.5.
V. fischeri
Segura et al., 2008
V. fischeri
Phenoxy-acid pesticides 2,4-D (50 mg L1) and 3,6-dichloro-2methoxy-benzoic acid (110 mg L1) Imidacloprid (100 mg L1)
Ionizing radiation
Toxicity was found to increase at low irradiation doses because of the formation of more toxic oxidation intermediates, which can be decomposed at larger doses. Both acute toxicity to D. magna and genotoxicity to B. subtilis sp. remain detectable even after significant removal of the pesticide has been achieved.
Drzewicz et al., 2004
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 1 1 e4 3 4 0
4323
investigated either alone or in combination with biological processes (Beltra´n-Heredia et al., 2001; Amat et al., 2003; Rizzo et al., 2008) but just a few works investigated their effect on either toxicity and biodegradability (Table 6). The capacity of AOPs to decrease toxicity was found to depend on AOP investigated, OMW characteristics and dilution factor, as well as organisms used for toxicity tests. Although wet hydrogen peroxide oxidation implementing FeeBEA zeolites as heterogeneous catalysts, was found to decrease toxicity to V. fischeri at some extent (from 100 to 70%) (Najjar et al., 2009), other investigated AOPs, such as electrochemical oxidation, photo-Fenton and ozonation, did not significantly improved biodegradability neither reduced toxicity and phytotoxicity (Andreozzi et al., 2008; Chatzisymeon et al., 2009; Justino et al., 2009). In this regard it is point of discussion the proposed use of AOPs and toxicity tests. Chatzisymeon et al. (2009) investigated the effect of electrochemical oxidation on OMW biodegradability and used toxicity tests as support information. Justino et al. (2009) evaluated the effect of coupled photo-Fenton-biological processes to final toxicity. On the opposite, Andreozzi et al. (2008) investigated the effect of different AOPs to OMW toxicity and phytotoxicity in order to evaluate the risks related to OMW disposal on agricultural land as well as to verify whether fertirrigation could really represent a cost-effective solution to the OMW disposal problem. But, as previously discussed, the use of AOPs as sole treatment is expected to be really expensive because of the formation of organic oxidation intermediates which may be more refractory to oxidation treatment than parent compounds. Moreover, the oxidation intermediates may be more toxic than parent compounds, thus increasing final toxicity/phytotoxicity. Accordingly, toxicity and phytotoxicity tests should be used either as a support tool to evaluate biodegradability of industrial wastewater after AOP treatment or to evaluate final toxicity/phytotoxicity after coupled AOP-biological treatment.
3.2.3.
Textile wastewater
These wastewaters are typically characterized by: (i) strong color because of residual dyes, (ii) presence of recalcitrant compounds (such as dyes, surfactants and sizing agents), (iii) high salinity, (iv) high temperature and (v) highly variable pH. Table 7 summarizes data and information on the bioassays used in evaluating textile wastewater treatment by AOPs. Selcuk et al. (2006) investigated the effect of pre-ozonation treatment on COD fractions removal and detoxification of textile wastewater. The authors found out that, under optimum conditions (86e96% of color, 33e39% of soluble COD and 57e64% of total COD removals), toxicity to D. magna can be decreased at some extant in diluted samples, and they inferred that pre-ozonation process improved biodegradability too. Unfortunately, according to the previous discussion, the optimum operating conditions to decrease toxicity does not necessarily result in optimum conditions for biodegradability (Gutie´rrez et al., 2002; Farre´ et al., 2007). In this regard, in a recent study biodegradability and toxicity of pre-ozonated textile wastewater at pilot scale were investigated (Somensi et al., 2010). The authors found out that biodegradability slightly increased along ozonation time and a minimum acceptable biodegradability threshold (roughly 0.4 BOD5/COD)
4324
Table 5 e Bioassays used in evaluating final toxicity of UWWTP effluents after advanced treatment with AOPs. Wastewater characteristics Mixture of municipal and industrial wastewater pre-treated by MBR (COD 45 mg L1).
Target contaminants
Investigated AOPs 1
D. magna, Photobacterium phosphoreum, umu (genotoxicity) test
Solar driven photo-Fenton pilot plant (3 m2 total illuminated area, 35 L total volume, 22 L irradiated volume) 50 mg L1 of H2O2, 5 mg L1 of Fe2þ 3 pharmaceuticals: TiO2 photocatalysis, catalyst loading in amoxicillin (10 mg L1), the range of 0.2e0.8 g L1, 125 W blackcarbamazepine (5 mg L1), light fluorescent lamp, 120 min and diclofenac (2.5 mg L1). maximum irradiation time. 3 pharmaceuticals: Sonolysis, 20 kHz ultrasound amoxicillin, carbamazepine generator, power density 25e100 W L 1 , initial pH 3e11, and air sparging, and diclofenac in 200 mL of wastewater (WW) spiked at different initial various concentrations Ozonation (maximum applied ozone concentration 1 mg O3/mg DOC)
V. fischeri
Three different wastewater samples: effluent from secondary biological treatment (SBT) (COD 17 mg L1, pH 8.3), SBT þ ozonation (SBT-O3), SBTO3 þ sand filtration (SBT-O3-SF) (COD 15 mg L1, pH 8.3). Secondary treated wastewater samples (pH 7.8, HCO3 7.2 meq L1, TOC 17 mg L1) from a Motril (Granada, Spain).
Effluent from urban wastewater treatment plant (DOC varied from 14.2 to 19.7 mg L1)
Three nitroimidazoles: Metronidazole (MNZ), Dimetridazole (DMZ), Tinidazole, (TNZ), 65, 150 and 300 mM initial concentrations. 85 micropollutants
V. fischeri (toxicity) and Salmonella typhimurium (mutagenicity)
D. magna, P. subcapitata, L. sativum D. magna, P.subcapitata, L. sativum
FELST with rainbow trout (Oncorhynchus mykiss)
Comments Genotoxic potential could be reduced by O3 but no comment is provided about the effects of the investigated AOPs on toxicity tests. Ozonation may either increase or decrease the toxic potential of secondary effluents. In general, the toxicity and mutagenicity results were inconsistent. The high removal rate of EC probably resulted in a large amount of more toxic intermediates thus increasing toxicity. In general the photocatalytic treatment did not completely reduce toxicity under the investigated conditions. Sonication decreased the toxicity in single compounds and mixtures to some extent; however, this was not achieved at all the experimental conditions applied in this study. Ozonation resulted in a significant developmental retardation of tested organisms.
References Baumgarten et al., 2007
Petala et al., 2008
Klamerth et al., 2010
Rizzo et al., 2009a,b
Naddeo et al., 2009
Stalter et al., 2010
Gamma irradiation (high-level 60Co source) The initial activity 2.22 1014 Bq (6.000 Ci), 310 Gy h1 dose rate at 20 cm distance from the source.
V. fischeri
Although gamma irradiation removed Sa´nchez-Polo TOC (70% after 700 Gy of treatment) et al., 2009 MNZ degradation by-products were not significantly mineralized, and toxicity did not decrease.
Full scale reclamation plant using ozonation (2 mg L1 and 5 mg L1 of O3 in pre-oxidation and main oxidation step respectively) and activated carbon filtration
V. fischeri, estrogenic activity (E-SCREEN assay), arylhydrocarbon receptor response (CAFLUX assay), neurotoxicity (acetylcholinesterase inhibition assay), phytotoxicity (PSII inhibition I-PAM assay) and genotoxicity (umuC assay)
The effect of the investigated Reungoat processes varied from one bioassay to et al., 2010 another but their combination was almost totally responsible for the overall observed toxicity reduction.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 1 1 e4 3 4 0
Ciprofloxacin, enrofloxacin, O3 and O3/UV (40 mg L O3 dosage, 20 moxifloxacin and their and 40 min treatment), H2O2 and H2O2/ precursors at 50 mg L1 UV (6 mL L1 of 30% H2O2) each one. Ozonation (applied ozone 2.5e8.0 mg L1, 2e30 min treatment)
Effluent from secondary biological treatment, pH 7.8, 640 mg CaCO3 L1 alkalinity, 40.0 NTU turbidity, 0.17 cm1 of UV254, COD 60.0 mg L1, total nitrogen 7.0 mg L1. 15 EC at 100 mg L1 initial Effluent from secondary biological concentrations treatment, COD 60.0 mg L1, DOC 25 mg L1. Effluent from secondary biological treatment, pH 8.1, BOD5 10 mg L1, TOC 4.51 mg L1, UV280 0.110 cm1, TSS 11 mg L1 Effluent from secondary biological treatment, pH 7.5, BOD5 4 mg L1, COD 10.5 mg L1, TOC 4.4 mg L1, UV280 0.079 cm1, TSS 4 mg L1
Toxicity tests
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 1 1 e4 3 4 0
80 60
1
2
3
inhibition (%)
40 20 0 -100 -20
-50
0
50
100
150
200
250
300
t 30w (min)
-40 -60
Fig. 2 e Final toxicity to Vibrio fischeri in a real wastewater effluent spiked with 15 emerging contaminants at low concentrations, after advanced oxidation by photo-Fenton in a pilot-scale solar compound parabolic collector (revised from Klamerth et al., 2010, with permission from Elsevier).
was detected only after 240 min treatment. Although toxicity to V. fischeri was found to decrease, the authors did not explain if the results were achieved under same ozonation treatment conditions (particularly, ozonation time). In a previous work the effect of pre-oxidation treatment on biodegradability of simulated textile wastewater was evaluated in terms of inhibition of microbial growth of activated sludge (Ledakowicz and Gonera, 1999). The inhibitory effect of O3/UV and O3/UV/ H2O2 processes on microbial growth during subsequent biodegradation of textile wastewater accounted for only 10%.
3.2.4.
Tannery wastewaters
Wastewater from leather tanning industry includes high organic loads and priority pollutants such as sulphite, chromium, synthetic tannins, oils, resins, biocides, detergents (Jochimsen et al., 1997; Tisler et al., 2004). Because of their low
4325
biodegradability different AOPs have been investigated. Among these just a few works evaluated the effect of AOPs on final toxicity (Table 8). In particular, when UV, TiO2/UV, O3 and O3/UV were investigated, toxicity to D. magna was found to be stable or decreased, but no increase in biodegradability was observed (Schrank et al., 2004). The same research group also studied Fenton and H2O2/UV processes and did not find any significant effect on toxicity to D. magna and V. fischeri (Schrank et al., 2005). On the opposite when the effect of TiO2 photocatalysis to A. salina was investigated in lower organic loading tannery wastewater compared to Schrank et al.’s (2004) work, toxicity was found to increase in diluted and undiluted samples in spite of COD and BOD5 removals (Sauer et al., 2006), confirming that the organisms used for bioassay strongly effect toxicity final outcome. The authors generically related this result to the formation of toxic intermediates, but taking into account that they also detected an increase in ammonia concentration in parallel, probably related to the degradation of aromatic and aliphatic compounds containing nitrogen through the oxidative breakdown of the CeN bonds, the increased toxicity may just be due to higher ammonia concentration (Marttinen et al., 2002).
3.2.5.
Pesticides wastewaters
Because of their persistence, toxicity and no biodegradability, pesticides can bio-accumulate as well as result in serious health problem; some of them have been included in the list of persistent organic pollutants by EU members Countries (Directive 2000/60/EC, 2000). Pesticides contamination comes from point and non-point sources, such as field runoff, field drainage pipes, wastewater treatment plants effluents, sewer overflows and runoff from farmyards (Neumann et al., 2002). In the province of Almeria (southern Spain) pesticides containers are selectively collected to be recycled. The recycling process includes a washing step which results in the formation of highly pesticides polluted water (Malato et al.,
Fig. 3 e Toxicity to D. magna and P. subcapitata exposed to the mixture samples (M) during photocatalytic treatment using 0.4 g TiO2 LL1 (reprinted from Rizzo et al., 2009a, with permission from Elsevier).
4326
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 1 1 e4 3 4 0
pesticides mixture (Oller et al., 2007). Moreover, the high toxicity after pre-oxidation did not result in any toxic effect to biomass in the subsequent biological treatment. Subsequent studies confirmed the high efficiency of solar photo-Fenton process in the removal of pesticides mixtures. Amat et al. (2009) observed that initial active ingredients concentrations decreased under detection limits after 36 min (time normalized with respect to solar radiation) treatment but, as expected, mineralization was a more time-consuming process, due to both the formation of organic intermediates and the presence of solvents in the commercial pesticides investigated. When measurements based on activated sludge were performed, toxicity was found to decrease according to the removal of the active ingredients; on the opposite, toxicity to V. fischeri was still high after treatment. Table 9 summarizes additional information about some papers available in scientific literature where bioassays have been used in evaluating pesticides wastewater treatment by AOPs.
a
b
3.2.6.
Fig. 4 e Cell growth inhibition of P. subcapitata (%) (a) and germination index (%) of L. sativum (b) exposed to M-spiked wastewater (WW D M) samples before and after photocatalytic treatment at 0.8 g TiO2 LL1 loading (reprinted from Rizzo et al., 2009a, with permission from Elsevier).
2007). Homogeneous (photo-Fenton) and heterogeneous (TiO2) photocatalysis processes were investigated in parallel at pilot scale to evaluate the oxidation of a mixture of five pesticides commonly used in the province of Almeria, and photo-Fenton was found to be much more efficient than TiO2 photocatalysis process for the degradation and mineralization of the
100 S1
% Inhib. 30 min., without dilution % Inhib. 30 min., diluted 1:10
S2 S3
Samples for Zahn-Wellens test
150
S4
60
S5
Toxicity limit
40
S6
100
DOC (mg/L)
% Inhibition
200
DOC
80
50
20 Total elimination of active ingredients
0
0 0
50
100
150
200
t 30 W (min)
Fig. 5 e Vibrio fischeri inhibition after 30 min exposure to samples partially treated by photo-Fenton at 20 mg/L of Fe2D (reprinted from Zapata et al., 2009, with permission from Elsevier).
Pulp and paper mill wastewaters
Wastewater from pulp and paper mill is refractory to biological process because of high concentration of lignin, an irregular tri-dimensional hydrophobic and aromatic polymer causing brown color. Moreover, cellulose pulp bleaching step results in the formation of several chlorinated compounds via chlorination, and others toxic organic compounds, including lignin-derived refractory ones. Different AOPs have been investigated for the treatment of pulp and paper mill wastewater. Catalkaya and Kargi (2008) investigated the effect of UV, UV/H2O2, UV/TiO2 and UV/H2O2/TiO2 on TOC and toxicity removal from the effluent of biological wastewater treatment plant of a pulp and paper manufacturing. Among the investigated processes UV/TiO2 resulted in the highest TOC and toxicity removals. The effect of pH (3, 7, 11) was also investigated and the best performance in terms of toxicity removal (resazurin method) was detected at neutral pH (7). Unfortunately, it is not clear from the manuscript if pH was adjusted before toxicity tests, so it may be point of discussion if the toxicity effect at pH 3 and 11 is due to pH extreme conditions (pH not adjusted) or to the formation of higher toxicity oxidation intermediates. More than 250 chemicals have been identified in pulp and paper effluents among these resin acids, which are diterpenoid carboxylic acids with characteristic three-rings chemical structure, naturally occur in softwood cellulose raw material and are released into mill effluents by pulping processes (Ledakowicz et al., 2006). The concern related to the release of resin acids into the environment is due to their toxic effect to D. magna and rainbow trout (Salmo gairdneri) at quite low concentrations in water (24 h-LC50: 0.26e1.89 mg/L; 96 h-LC50 0.4e1.1 mg/L, respectively) (Peng and Roberts, 2000). Ledakowicz et al. (2006) investigated the effect of different O3 based AOPs (O3, O3 þ UV, O3 þ UV þ H2O2) on the removal of different resin acids mixtures as well as their effect on toxicity and biodegradability. O3 treatment was found to decrease toxicity to V. fischeri at the lower investigated doses, but a subsequent increase was observed when O3 dose was increased to completely remove resin acids. Moreover, based on final COD concentrations measured before and after treatment with O3 and O3 based AOPs of model wastewater,
Table 6 e Bioassays used in evaluating olive mill wastewater treatment by AOPs. Wastewater
Investigated AOPs 1
Toxicity/biodegradability tests
Pilot-scale plant combining electroFenton, anaerobic digestion and UF.
V. fischeri and L. sativum seeds.
Composite wastewater samples (pH 5.2, 115 g COD L1, 32 g SS/L, 5.6 g phenols L1) from an olive mill plant with a daily olive processing capacity of 30 tones in Bursa City, Turkey. Diluted and undiluted wastewater samples from a 3-phase process olive mill plant, North Portugal.
Fenton (FeSO4$7H2OeH2O2) and Fenton-like (FeCl3$6H2OeH2O2) processes; pH 3, 500 mg H2O2 L1, 500e3500 mg L1 of FeSO4 and FeCl3 doses.
Activated sludge inhibition tests (ISO 8192).
Photo-Fenton process before and after biological treatment by three species of fungi (P. sajor caju, T. versicolor and P. chrysosporium)
D. longispina.
Filtered wastewater sample (pH 4.4, 40 g COD mg L1, 0.6 g TSS L1, 3.5 g phenols L1) from three-phase olive oil mill company, in Chania, Crete, Greece.
Electrochemical oxidation over borondoped diamond (BDD) electrodes (70 cm2 area, distance between them 0.01 m).
V. fischeri and ZahneWellens biodegradability tests
Wastewater samples (pH 5.2, 64.4 g COD L1, 13.7 g BOD5 L1, 61.9 g TSS L1, 6.4 g phenols L1) from traditional discontinuous olive oil extraction plant in Sfax, southern Tunisia. Centrifuged wastewater sample (78.7 g COD mg L1, 4.4 g phenols mg L1) collected by a plant located in Bari (South Italy).
Wet hydrogen peroxide oxidation implementing FeeBEA zeolites as heterogeneous catalysts, differing in the amount of iron.
V. fischeri
Ozonation, solar photolysis, solar modified photo-Fenton, solar modified photo-Fentoneozonation.
Toxicity to P. subcapitata and phytotoxicity to seeds of R. sativus, C. sativus and L. sativa
References
Toxicity to V. fischeri decreased to 38% inhibition after anaerobic treatment and totally decreased after UF posttreatment. UF also increased germination index of L. sativum up to 124% compared to 12% of untreated OMW. Fenton and Fenton-like processes considerably removed inhibitory effect from untreated OMW.
Khoufi et al., 2009
Photo-Fenton process was found to preserve or even increase the effluent toxicity. Furthermore, when OMW was pre-treated by photo-Fenton, the treatment with fungi did not significantly decreased toxicity. BDD electrochemical oxidation did not improve biodegradability as assessed by the ZahneWellens test. All samples were found to be highly toxic to V. fischeri with EC50 values never exceeding 5%. Biodegradability increased and toxicity to V. fischeri decreased from 100 to 70%.
Justino et al., 2009
2 h ozonation treatment only moderately reduced phytotoxicity to seeds and the treated samples were still toxic to P. subcapitata. Both long-term aerated storage of OMW under sunlight and the combined process solar modified photo-Fentoneozonation did not significantly contribute to seed germination.
Mert et al., 2010
Chatzisymeon et al., 2009
Najjar et al., 2009
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 1 1 e4 3 4 0
Settled OMW (pH 5.2, 95 g COD L , 19 g BOD5 L1, 15 g TSS L1, 11.5 g polyphenols L1) from a discontinuous olive oil processing plant located in Sfax (southern Tunisia).
Remarks
Andreozzi et al., 2008
4327
Wastewater
Investigated AOPs
Toxicity/biodegradability tests
Remarks
References
Pre-treatment by O3, UV and UV/H2O2 had positive effect on subsequent biodegradation. Toxicity to V. fischeri was only tested on untreated single textile constituents. The inhibitory effect of acid dyebath toward sewage sludge can be completely eliminated after Fenton oxidation.
Ledakowicz et al., 2001
Simulated wastewater (pH 7.38, 2154 mg COD L1, 1050 mg BOD5 L1).
O3, UV, H2O2, O3/H2O2, UV/H2O2
Inhibition of microbial growth of activated sludge and toxicity tests to V. fischeri
Acid dyebath consisted of three different acid dyestuffs (C.I. Acid Yellow 242, C.I. Acid Red 360 and C.I. Acid Blue 264, 30 mg L1 each one) and two dye auxiliaries (a leveling agent (1500 mg L1) and an acid donor (500 mg L1)) Wastewater sample (330.5 mg DOC L1, pH 6) taken right after the dyebath and before entering the activated sludge treatment of textile industry located in Thessaloniki (Greece). Two raw textile wastewater samples (COD: 1600, 1560 mg L1; soluble COD: 950, 900 mg L1; BOD5, 150, 170 mg L1; TSS, 750, 250 mg L1; Color, 3300, 1000 PteCo; pH 8.8, 8.6) from the balancing tank of a textile industry located in Istanbul (Turkey) Raw (COD: 910 mg L1; filtered COD: 560 mg L1; BOD5: 150 mg L1; TSS: 150 mg L1; Color: 1570 PteCo unit; pH 10) and biologically treated (COD: 210 mg L1; filtered COD: 170 mg L1; BOD5: 20 mg L1; TSS: 90 mg L1; Color: 1450 PteCo unit; pH 8.5) textile wastewater samples from a textile finishing industry in Istanbul, Turkey. Raw wastewater (COD: 1505 mg L1; BOD5: 91 mg L1; Color: 0.754 as A455; pH 9.1) from the equalization storage tank of a textile mill in Blumenau (Brazil). Simulated textile wastewater diluted with 10% synthetic domestic sewage (pH: 7.38; COD: 2154 mg L1; BOD5: 1050 mg L1).
Fenton
Activated sludge inhibition test
TiO2 photocatalysis (1.5 L Pyrex reactor, diving Philips HPK 125 W highpressure mercury lamp jacked with a water-cooled Pyrex filter restricting the transmission of wavelengths below 290 nm). Ozonation (Pyrex glass reactor: 40 mm diameter, 1100 mm height; 18.5e24 mg L1 ozone dose; 0.566 m3 h1 air flowrate).
V. fischeri
Although a complete decolorization of the wastewater was achieved within 6 h, inhibition to V. fischeri only decreased to some extent (z35% for 6h treatment).
Bizani et al., 2006
D. magna
Ozone was effective for removing acute toxicity to D. magna from diluted textile wastewater.
Selcuk et al., 2006
Fenton (FeSO4 and H2O2 doses between 100 and 400 mg L1 and 600 to 1200 mg L1, respectively; pH 3.0 and temperature 40 C) and ozonation (1.4 g O3 L1 h1 applied dose to 1 L samples for 20 min without any pH adjustment (pH 10.0)).
D. magna
Both processes were effective in removing wastewater toxicity.
Meric¸ et al., 2005
Ozonation (20 g m3; flowrate 1 m3 h1)
BOD5/COD for biodegradability and V. fischeri for toxicity
Ozonation treatment reduced toxicity and improved biodegradability
Somensi et al., 2010
O3/UV, O3/UV/H2O2 (1.5 L stirred gaseliquid reactor; UV lamps: 150 W, l ¼ 254e578 nm, 15 W, l ¼ 254 nm)
Inhibition of microbial growth of activated sludge
The inhibitory effect of O3/UV and O3/ UV/H2O2 processes on microbial growth during subsequent biodegradation of textile wastewater accounts for only 10%.
Ledakowicz and Gonera, 1999
4328
Table 7 e Bioassays used in evaluating textile wastewater treatment by AOPs.
Arslan Alaton and Teksoy, 2007
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 1 1 e4 3 4 0
4329
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 1 1 e4 3 4 0
Table 8 e Bioassays used in evaluating tannery wastewater treatment by AOPs. Wastewater Effluent of a mechanical and chemical industrial wastewater treatment plant in Brazil (Bonato Couros SA) (COD: 2365 mg L1; BOD5:1010 mg L1; TOC: 820 mg L1; DOC: 720 mg L1). Coagulated/settled tannery wastewater (COD: 130 mg L1; BOD5:47 mg L1; TOC: 45 mg L1; DOC: 44 mg L1). Coagulated/settled wastewater (COD: 801 mg L1; BOD5: 349 mg L1; Chromium(VI) 0.1 mg L1; sulfate 1150 mg L1; pH adjusted at 2.5).
Investigated AOPs
Toxicity/ biodegradability tests
References
UV (15 W medium pressure mercury lamp), TiO2/UV (1 g TiO2 L1; 120 min), O3 and O3/UV (2.6 g O3 h1, 60 and 30 min respectively).
D. magna
In spite of toxicity to D. magna was found to stable or decreased, no increase in biodegradability was observed.
Schrank et al., 2004
Fenton (pH 3.5, 2 h, 50e240 mg L1 of FeSO4, 100e500 mg L1 of H2O2) and H2O2/UV (15 W medium pressure mercury lamp, different H2O2 and pHs) H2O2/UV, TiO2/H2O2/UV and TiO2/UV in a continuous operated annular reactor (15 W UV-lamp, 1 g TiO2 L1, 1 h).
D. magna and V. fischeri
No increase or only a slight decrease in toxicity was observed compared to the untreated wastewater
Schrank et al., 2005
A. salina
A. salina mortality increased after TiO2/UV treatment (1 h) for diluted and undiluted samples
Sauer et al., 2006
the authors concluded that the investigated processes do not improve biodegradation; unfortunately, since specific biodegradability tests were not performed, results cannot be considered conclusive. Yeber et al. (1999) investigated the effect of different AOPs (O3/UV, O3/UV/ZnO, O3/UV/TiO2, O2/ UV/ZnO, O2/UV/TiO2), on the biodegradability and toxicity of the effluent from bleaching process of Pinus radiata wood. Acute toxicity to V. fischeri was found to decrease between 30 and 60% after treatment, being the O2/UV/TiO2 system the most efficient process. Moreover, the investigated processes were found to increase biodegradability too in terms of BOD/ COD ratio and activated sludge dry weight. Their results found confirmation in the study from Rodrigues et al. (2008) which investigated the combined treatment of coagulationeflocculation followed by UV driven AOPs (UV/TiO2, UV/ H2O2 and UV/TiO2/H2O2). UV/TiO2/H2O2 treatment was found to decrease A. salina mortality and improve biodegradability in terms of COD/BOD ratio. Table 10 includes some additional information about the above discussed papers in relation to wastewater characteristics and AOPs operating conditions.
4.
Remarks
Drinking water treatment by AOPs
AOPs have been investigated over the past years in drinking water treatment to (i) decrease natural organic matter (NOM) typically found in waters, in order to prevent the formation of disinfection by-products (DBPs), (ii) disinfect water (iii) remove xenobiotics (particularly pharmaceuticals and endocrine disruptors compounds) and (iv) microcystins. Typically, these contaminants occur at low concentrations in natural waters and ecotoxicity tests may be not enough sensitive to characterize toxicity of both parent compounds and oxidation intermediates so, sometime, when pharmaceuticals and EDCs
have been investigated estrogenic activity was evaluated in parallel. In the following sub paragraphs the above mentioned topics are reviewed. Table 11 summarizes data and information (water characteristics, target contaminants, equipments and operating conditions for investigated AOPs, toxicity and/ or estrogenic activity tests implemented and related results) for the papers reviewed in the subsequent paragraphs.
4.1.
NOM removal and DBPs control
The formation of DBPs can be controlled by reducing NOM, which occurs in natural waters. NOM, particularly humic acids, can react with disinfectants (chlorine gas and hypochlorites) to form chlorination by-products (Nikolaou et al., 2007). Process efficiency can be characterized in terms of UV absorbance (typically measured at the wavelength of 254 nm, UV254), DOC, specific UV absorbance (UV254/DOC ratio, SUVA254) and trihalomethanes formation potential (THMFP) measurements. The effects of different treatment methods (ozonation, coagulation, ozonation followed by coagulation and TiO2 photocatalysis) on the removal of organic matter from surface waters taken from different regions of Istanbul (Turkey) and Salerno (Italy) were investigated in terms of UV254, SUVA254 and DBP formation potentials and toxicity to D. magna (Bekbolet et al., 2005). The related removal efficiencies as well as toxicity results were found to be site and treatment specific. The effect of pre-ozonation treatment on coagulation process was also investigated in terms of TOC and THMFP removals as well as toxicity to D. magna (Selcuk et al., 2007). The authors did not observe any doseeresponse relationship between the measured DBPs and toxicity to D. magna. Taking into account that more than 500 DBPs have been reported in scientific literature (Richardson, 1998), the toxicity may be due to one or more among the unmeasured DBPs. Moreover, DBPs
4330
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 1 1 e4 3 4 0
Table 9 e Bioassays used in evaluating pesticides wastewater treatment by AOPs. Wastewater A mixture of five pesticides (Methomyl, Dimethoate, Oxamyl, Cymoxanil and Pyrimethanil 50 mg L1 each one)
A mixture of four commercial pesticides including methyloxydemeton, methidathion, carbaryl, dimethoate as active ingredients (50 mg L1 each one). Wastewater (33700 2100 mg COD L1, 6100 700 mg BOD5 L1) from a pesticide factory in Hebei Province, North of China. A mixture of five pesticides (Methomyl, Dimethoate, Oxamyl, Iimidacloprid and Pyrimethanil, 40e100 mg DOC L1 each one).
Investigated AOPs Solar TiO2 photocatalytic experiments: 35-L compound parabolic collectors pilot plant, pH 6, 0.2 g TiO2 L1. Solar Photo-Fenton tests: 75 L pilot plant, pH adjusted to 2.7e2.9, 20 and 55 mg Fe2þ L1, 200e500 mg H2O2 L1 Solar Photo-Fenton tests: 40 L pilot plant, pH adjusted to 2.8, 20 mg Fe2þ L1, 200e500 mg H2O2 L1.
Toxicity/ biodegradability tests V. fischeri and biodegradability assay (ZahneWellens test)
V. fischeri and biodegradability assays (OUR inhibition test, ZahneWellens test, BOD5/ COD ratio)
Photo-Fenton was much more efficient than TiO2 photocatalysis for pesticide degradation and mineralization. The high toxicity after preoxidation did not result in any toxic effect to biomass in the subsequent biological treatment. Toxicity to V. fischeri was still high after the removal of pesticides active ingredients, although the effluent might be compatible with biological process.
References Oller et al., 2007
Amat et al., 2009
Fenton oxidation (optimum conditions: 97 mmol H2O2 L1 and 40 mmol Fe2þ L1 at initial pH 3).
BOD5/COD ratio
The biodegradability (BOD5/ COD) of the wastewater was enhanced from 0.18 to more than 0.47.
Chen et al., 2007
Solar Photo-Fenton tests: 50 L pilot plant (2.25 m2 irradiated surface, 22 L irradiated volume), pH adjusted to 2.7e2.9, 20 mg Fe2þ L1, 100 mg H2O2 L1
P. putida biodegradability assay
Biodegradability first increased with mineralization, then decreased at the end of the treatment because of the formation of less biodegradable by-products.
Ballesteros Martı´n et al., 2009
typically occur at low concentrations in drinking water (from ng L1 to mg L1), with no relevant acute toxic effect on the investigated organisms; therefore, more sensitive and/or reliable chronic bioassays should be developed to characterize possible toxic effects of AOPs in drinking water applications.
4.2.
Remarks
Xenobiotics
Conventional drinking water treatments such as coagulation, sedimentation, filtration and adsorption may not be effective for the removal of certain classes of xenobiotics (Westerhoff et al., 2005; Stackelberg et al., 2007). Moreover, chlorine, the most widely used oxidant/disinfectant in the treatment of ground and surface water, may not significantly improve the removal of these pollutants (Gibs et al., 2007). Although AOPs have been widely investigated in the removal of xenobiotic compounds (particularly pharmaceuticals and endocrine disruptors compounds) from aqueous solutions, only a few studies have been focused on the decontamination of natural waters (surface and ground waters) and AOPs effect on toxicity and estrogenic activity. While slurry heterogeneous photocatalytic reactors have been found to be effective in the removal of a wide range of pharmaceuticals and EDCs at bench-scale (Calza et al., 2006; Me´ndez-Arriaga et al., 2008; Rizzo et al., 2009a), a major limitation to scaling up this technology has been the difficulty in separating the TiO2 following treatment. Benotti et al. (2009)
investigated a patented TiO2 photocatalytic reactor membrane pilot system (Photo-Cat) for the removal of thirty-two pharmaceuticals and endocrine disrupting compounds from river water as well as for the effect on estrogenic activity. Higher than 70% removal was detected for twenty-nine investigated contaminants (the remaining three being removed at less than 50%), but no estrogenically active transformation products were detected after photocatalytic treatment. The degradation of EDCs by AOPs and the effect on estrogenic activity have also been investigated in the last years (Chen et al., 2006; Rosenfeldt et al., 2007). Among numerous EDCs detected into the environment, bisphenol A (BPA) has received a great deal of attention because of widespread use in the production of polycarbonates, epoxy resins, numerous plastic articles, and dental sealants (Staples et al., 1998) and effects on human health and environment (Segner et al., 2003; Kang et al., 2006). BPA has been detected into surface water in concentrations from ng to mg L1 (Belfroid et al., 2002; Jin et al., 2004). Since, conventional water treatment processes are not effective in the removal of most EDCs, Chen et al. (2006) investigated direct photolysis with low-pressure Hg UV lamps and UV/H2O2 processes for the degradation of BPA as well as changes in estrogenic activity. They found out that UV/ H2O2 significantly removed BPA and estrogenic activity in vitro and in vivo. Moreover, different sensitivities of the bioassays were observed because of removal rates of in vivo estrogenic activity were significantly lower than those observed in vitro.
4331
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 1 1 e4 3 4 0
Table 10 e Bioassay used in evaluating pulp and paper mill wastewater treatment by AOPs. Wastewater Wastewater from bleaching process of Pinus radiata wood of pulp mill plant (VIII Region, Chile) (Color: 4510 mg Pt L1; pH 10.8; COD: 1550 mg L1; BOD: 534 mg L1; Cl: 47 mmol L1). Effluent of biological wastewater treatment plant of a pulp and paper manufacturing (Izmir, Turkey) (pH 7.08; COD: 592 mg L1; BOD: 240 mg L1) Resins acids based solutions (BOD5: 200e1300 mg L1; COD: 300e1900 mg L1; initial toxicity to Vibrio fischeri 30e96%)
Effluent of cellulose and paper industry (pH: 9.8; COD: 1303 mg L1; BOD: 148 mg L1).
Investigated AOPs
Toxicity/ biodegradability tests
O3/UV, O3/UV/ZnO, O3/UV/TiO2, V. fischeri O2/UV/ZnO, O2/UV/TiO2 (125 W high-pressure Hg lamp; 2 g O3 h1)
Remarks
References
All AOPs significantly improved Yeber biodegradability (BOD/COD) et al., 1999 and decreased toxicity to some extent.
Catalkaya and Kargi, 2008
UV, UV/H2O2, UV/TiO2, UV/ H2O2/TiO2 (16 W low-pressure mercury vapor lamp
Washed activated sludge UV/TiO2 resulted in a higher (Resazurin reduction toxicity removal compared to method) UV/H2O2/TiO2.
O3, O3 þ UV, O3 þ UV þ H2O2 (0.03e2.9 mgO3/mgCOD, 15 W low-pressure lamp, 2 ml H2O2 L1 of solution, temperature: 20, 50 and 80 C, initial pH: 9e10, after ozonation: 7e8). Combined treatment of coagulationeflocculation followed by photocatalysis (UV/ TiO2, UV/H2O2 and UV/TiO2/ H2O2); 250 W low-pressure mercury lamp.
V. fischeri
O3 treatment decreased toxicity Ledakowicz et al., 2006 in the range of 0.3e0.5 mgO3/ mgCOD (in parallel, for 3 of 4 resins mixtures the removal was >90%), then, for higher O3 doses toxicity increased.
A. salina and biodegradability assay (COD/BOD)
A. salina mortality decreased after coagulation and was improved by photocatalysis treatment. Biodegradability improved from 0.14 to 0.5 after coagulation, to 0.63 and 0.71 after 2 and 4 h photocatalytic treatment respectively.
Furthermore, the UV/H2O2 process was found to be effective for reducing larval lethality in treated BPA solutions, suggesting that the degradation process did not result in the production of acutely toxic oxidation intermediates. The same investigation group studied the degradation and the corresponding effect on estrogenic activity of other two EDCs, 17-bestradiol (E2) and 17-a-ethinyl estradiol (EE2) following direct UV photolysis and UV/H2O2 treatments (Rosenfeldt et al., 2007). UV/H2O2 process effectively decreased (90%) estrogenic activity of E2 and EE2 at environmentally relevant concentrations (roughly 3 mg L1). Furthermore, no statistically significant difference between removal rates of E2 and EE2 and the subsequent reduction in estrogenic activity was observed, implying that the oxidation products of E2 and EE2 are not as estrogenic as the parent compounds. Although drinking water mainly comes from groundwater in Italy, the increasing chemical contamination in some area makes surface water an important alternative supply. Ozone based AOPs (UV/O3 and UV/O3/H2O2) were compared with resin and granular activated carbon (GAC) adsorption to evaluate their effectiveness in the removal of toxic and mutagenic organic micropollutants from Como Lake waters (Italy) (Guzzella et al., 2002). The results showed a decrease of the mutagenic and toxic activities after adsorption on GAC and resins, while AOPs were generally found out to increase these parameters. When a GAC adsorption step was performed in addition to the AOPs, no toxicity and mutagenic activity was observed.
4.3.
Rodrigues et al., 2008
Microcystins
Microcystins (MCs) are a group of monocyclic heptapeptide hepatotoxins produced by different cyanobacteria such as Microcystis, Anabaena, Nostoc and Oscillatoria which can be found in fresh waters (Pe´rez and Aga, 2005). MCs exhibit acute and chronic effects on humans and wildlife as well as they can result in serious damage to the liver (Gupta et al., 2003; Weng et al., 2007). MC-LR is usually regarded as one of the most acutely toxic cyanobacteria toxins and WHO recommends a guideline value of 1 mg L1 for drinking water (WHO, 2008). Since MCs are stable against physicochemical and biological factors (such as temperature, sunlight and enzymes) and conventional water treatment processes have been proven to be unreliable for the removal of these toxins (Jurczak et al., 2005), advanced treatment such as AOPs have been investigated (Lawton et al., 2003; Al Momani et al., 2008; Miao et al., 2010). Liu et al. (2002) found out that TiO2 photocatalysis can effectively destroy MC-LR in aqueous solutions, confirming previous studies (Lawton et al., 1999), but it can also significantly reduce protein phosphatase 1 (PP1) inhibition, which is potentially one of the most harmful effects to humans who may consume water contaminated by MC. In spite of a rapid disappearance of the MC-LR after photocatalysis treatment, the PP1 inhibition only slightly decreased after 20 min reaction time. However, the inhibition was found to rapidly decrease after 30 min treatment and only 20% residual inhibition was detected after 60 min. Moreover, when H2O2 was added to photocatalytic
Water characteristics
Natural ground (pH 7.5, HCO3 8.8 meq L1, TOC 10.3 mg/L) and surface (pH 8.3, HCO3 6.4 meq L1, TOC 11.9 mg L1) water samples were collected from a drinking water treatment plant (Motril, Spain). Microcystins aqueous solutions
Surface water (Tai Lake, China).
Investigated AOPs
Toxicity/Estrogenic activity tests
Comments
References
Disinfection by-products
Ozone (10.5 mg L1 min1 for 5 min) and TiO2 photocatalysis (125 W black-light fluorescent lamp)
D. magna
The immobilization test results were in accordance with the DBPs distribution rather than NOM removal efficiency.
Bekbolet et al., 2005
Disinfection by-products
Ozone (10.5 mg L1 min1 for 5 min)
D. magna
No doseeresponse relationship was observed between DBPs and toxicity to D. magna. Spiked bromide just resulted in higher levels of brominated DBPs but no significant change was observed in the immobilization of D. magna.
Selcuk et al., 2007
Three relevant metabolites (10 and 50 mg L1 initial concentrations) of the analgesic and antipyretic drug dipyrone, 4-methylaminoantipyrine (4-MAA), 4-formylaminoantipyrine (4-FAA) and 4-acetylaminoantipyrine (4-AAA) Three nitroimidazoles: Metronidazole (MNZ), Dimetridazole (DMZ), Tinidazole, (TNZ), 65, 150 and 300 mM initial concentrations.
Simulated solar irradiation system (Suntest) equipped with a 1100 W xenon arc lamp and special filters (wavelength < 290 nm).
D. magna
24 and 48 h inhibition increased to 27e55% and 60e85% respectively after photolysis treatment. Since investigated metabolites were totally removed, the toxicity to D. magna is due to the formation of toxic intermediates.
Go´mez et al., 2008
Gamma irradiation (high-level 60 Co source) The initial activity 2.22 1014 Bq (6.000 Ci), 310 Gy h 1 dose rate at 20 cm distance from the source.
V. fischeri
Toxicity during MNZ gamma irradiation increased with longer treatment time up to an irradiation dose of 200e300 Gy for both source waters, but decreased at higher doses.
Sa´nchez-Polo et al., 2009
TiO2 photocatalysis (480 W xenon lamp)
Protein phosphatase inhibition (PP1) assay
Liu et al., 2002
Ozone (1e6 mg O3 mg1 MC).
In vitro (protein phosphatase inhibition method) and in vivo (mouse bioassay method) tests.
The photocatalytic process significantly decreased PP1 inhibition. Ozonation treatment drastically decreased toxicity.
Microcystin-LR extracted from a bloom of Microcystis aeruginosa Two Microcystins (MCs) MC-LR and MC-RR extracted from the natural cyanobacteria of Microcystis Aeruginosa
Miao et al., 2010
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 1 1 e4 3 4 0
Surface water samples from Buyukcekmece and Omerli (Istanbul, Turkey) and Alento (Salerno, Italy) (pH 7.65, 7.18, 7.97; alk. 150, 70, 170 mg CaCO3 L1; bromide 274, 95, nd mg L1; chloride 98, 45, 14 mg L1, TOC 3.61, 3.05, 2.47 mg L1 respectively) Surface water samples from Buyukcekmece and Omerli (Istanbul, Turkey) and Carmine (Salerno, Italy) (pH 7.65, 7.18, 7.73; alk. 150, 70, 111 mg CaCO3 L1; bromide 274, 95, nd mg/L; chloride 98, 45, 17 mg L1, TOC 3.61, 3.05, 2.05 mg L1 respectively) Synthetic fresh water
Target contaminants
4332
Table 11 e Toxicity/estrogenic activity in drinking water after treatment with AOPs.
Organic micropollutants
UV/O3 and UV/O3/H2O2 were compared with GAC adsorption, amberlite and ion exchange resins
V. fischeri
Model water
bisphenol A
Photolysis with low-pressure Hg UV lamps and UV/H2O2
Two natural water samples (DOC 3.06 and 1.39 mg L1; alkalinity 2.63 and 0.67 mM; pH 8.0 and 7.98; UV254 0.04 and 0.11 cm1) taken from water treatment plant inlets before any treatment and filtered through a 0.45 mm filter. Surface water from Colorado River water, Lake Mead, NV, USA (2.6 mg TOC L1; alkalinity 137 mg L1; pH 8.0; UV254 0.036 cm1).
17-b-estradiol (E2) and 17-aethinyl estradiol (EE2) (10 nMe5 mM).
Direct UV photolysis and UV/ H2O2 advanced oxidation. Two irradiation systems: 1 kW medium pressure UV lamp and four 15-W low-pressure UV lamps.
Estrogenic activity was evaluated by in vitro yeast estrogen screen (YES) and in vivo vitellogenin (VTG) assays with Japanese medaka fish (Oryzias latipes). Estrogenic activity was evaluated by YES assay.
Thirty-two pharmaceuticals and endocrine disrupting compounds (16e1300 ng L1).
A patented photocatalytic reactor membrane pilot system (Phot-Cat), employing UV/TiO2 photocatalysis.
Estrogenic activity was evaluated by YES assay.
Opposite to GAC and resin treatment, AOP processes generally increased toxicity and mutagenic activity (Ames assay). The absence of mutagenic activity was detected only when a GAC step was performed in addition to the AOP process. UV/H2O2 significantly removed BPA and estrogenic activity in vitro and in vivo.
Guzzella et al., 2002
90% removal of estrogenic activity of E2 and EE2 (3 mg L1) was achieved with 5 mg H2O2 L1.
Rosenfeldt et al., 2007
Twenty-nine of the targeted compounds in addition to total estrogenic activity were greater than 70%. No estrogenically active transformation products were formed during treatment.
Benotti et al., 2009
Chen et al., 2006
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 1 1 e4 3 4 0
Surface water from Como lake, Italy (2e10 mg TOC L1).
4333
4334
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 1 1 e4 3 4 0
Table 12 e The role of AOPs in the treatment of actual aqueous matrices and the corresponding role of the bioassays. Aqueous matrix Fresh/drinking water
Urban wastewater
Industrial wastewater
Role of AOPs Metal oxidation To improve coagulation Disinfection Advanced treatment downstream of biological process to remove toxic/emerging contaminants before discharge or reuse Pre-treatment to biological process to improve biodegradability of wastewater Post-treatment to biological process to improve the removal of residual contaminants
system, the destruction of MC-LR took place much faster and toxicity dropped more quickly (a rapid reduction in PP1 inhibition within 5 min of photocatalytic treatment with a complete disappearance in 20 min was observed). Miao et al. (2010) investigated MC-LR and MC-RR degradation and detoxification by ozone. The toxicity of the MCs ozonation by-products was evaluated by assaying the protein phosphatase inhibition in vitro and acute toxicity in vivo. The toxicity evaluation did not show any adverse effects in vivo and in vitro of ozonation end-products, moreover MCs toxicity was completely removed.
5.
Role of bioassays
Conclusive remarks
In conclusion, bioassays, when properly used, showed to be a really useful tool to evaluate the dangerousness of AOPs as well as to set up the proper operative conditions. Table 12 summarizes the role of AOPs in the treatment of a given actual aqueous matrix and the corresponding role of the bioassays. The text organism should be chosen according to the final use of the treated water matrix; for instance, inhibition tests with D. magna may be suitable to evaluate the toxicity of wastewater treatment plant effluent before its disposal; plants bioassay may be suitable to characterize toxicity of wastewater treatment plant effluent before its agricultural reuse. Sometime, acute toxicity tests may not be the most suitable to evaluate the ecotoxicological hazard of micropollutants because of the low concentrations (Rizzo et al., 2005; Baumgarten et al., 2007), although a chronic effect can be expected (Crane et al., 2006). Accordingly studies on chronic effects should be further developed. Finally, some care should be taken in the characterization of the effect of AOPs on the biodegradability of industrial wastewater. Toxicity tests may be not suitable to achieve this aim, so they may be used just as screening test before to use more suitable biodegradability tests (e.g., activated sludge bioassays, respirometry).
Acknowledgment The author wishes to thank the University of Salerno for funding the project entitled “Rimozione di composti xenobiotici
To evaluate the toxicity of oxidation intermediates
Toxicity tests: to evaluate if toxicity decrease after AOPs treatment. Phytotoxicity tests: to evaluate if the effluent is suitable for agricultural reuse As support to biodegradability tests To evaluate the toxicity of oxidation intermediates
e detossificazione di acque reflue urbane destinate al riutilizzo mediante processi di ossidazione avanzata” (FARB, 2009).
references
Al Momani, F., Smith, D.W., El-Din, M.G., 2008. Degradation of cyanobacteria toxin by advanced oxidation processes. Journal of Hazardous Materials 150, 238e249. Amat, A.M., Arques, A., Beneyto, H., Garcı´a, A., Miranda, M.A., Seguı´, S., 2003. Ozonisation coupled with biological degradation for treatment of phenolic pollutants: a mechanistically based study. Chemosphere 53, 79e86. Amat, A.M., Arques, A., Miranda, M.A., Lo´pez, F., 2005. Use of ozone and/or UV in the treatment of effluents from board paper industry. Chemosphere 60, 1111e1117. Amat, A.M., Arques, A., Garcıa-Ripoll, A., Santos-Juanes, L., Vicente, R., Oller, I., Maldonado, M.I., Malato, S., 2009. A reliable monitoring of the biocompatibility of an effluent along an oxidative pre-treatment by sequential bioassays and chemical analyses. Water Research 43, 784e792. Andreozzi, R., Caprio, V., Insola, A., Marotta, R., 1999. Advanced oxidation processes (AOP) for water purification and recovery. Catalysis Today 53, 51e59. Andreozzi, R., Marotta, R., Pinto, G., Pollio, A., 2002. Carbamazepine in water: persistence in the environment, ozonation treatment and preliminary assessment on algal toxicity. Water Research 36, 2869e2877. Andreozzi, R., Caprio, V., Marotta, R., Vogna, D., 2003. Paracetamol oxidation from aqueous solutions by means of ozonation and H2O2/UV system. Water Research 37, 993e1004. Andreozzi, R., Campanella, L., Fraysse, B., Garric, J., Gonnella, A., Lo Giudice, R., Maritta, R., Pinto, G., Pollio, A., 2004. Effects of advanced oxidation processes (AOPs) on the toxicity of a mixture of pharmaceuticals. Water Science and Technology 50 (5), 23e28. Andreozzi, R., Canterino, M., Lo Giudice, R., Marotta, R., Pinto, G., Pollio, A., 2006. Lincomycin solar photodegradation, algal toxicity and removal from wastewaters by means of ozonation. Water Research 40, 630e638. Andreozzi, R., Canterino, M., Di Somma, I., Lo Giudice, R., Marotta, R., Pinto, G., Pollio, A., 2008. Effect of combined physico-chemical processes on the phytotoxicity of olive mill wastewaters. Water Research 42, 1684e1692. APHA/AWWA/WEF, 1998. Standard Methods for the Examination of Water and Wastewater, 20th ed. APHA/AWWA/WEF, Washington, DC, USA. Arslan Alaton, I., Teksoy, S., 2007. Acid dyebath effluent pretreatment using Fenton’s reagent: process optimization,
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 1 1 e4 3 4 0
reaction kinetics and effects on acute toxicity. Dyes and Pigments 73, 31e39. Auriol, M., Filali-Meknassi, Y., Tyagi, R.D., Adams, C.D., Surampalli, R.Y., 2006. Endocrine disrupting compounds removal from wastewater, a new challenge. Process Biochemistry 41, 525e539. ¨ tker, M., 2003. Treatment of pharmaceutical Balcıoglu, I.A., O wastewater containing antibiotics by O3 and O3/H2O2 processes. Chemosphere 50, 85e95. Ballesteros Martı´n, M.M., Sa´nchez Pe´rez, J.A., Garcı´a Sa´nchez, J.L., Casas Lo´pez, J.L., Malato Rodrı´guez, S., 2009. Effect of pesticide concentration on the degradation process by combined solar photo-Fenton and biological treatment. Water Research 43, 3838e3848. Baumgarten, S., Schro¨der, H.Fr., Charwath, C., Lange, M., Beier, S., Pinnekamp, J., 2007. Evaluation of advanced treatment technologies for the elimination of pharmaceutical compounds. Water Science and Technology 56 (5), 1e8. Beccari, M., Carucci, G., Lanz, A.M., Magone, M., Petrangeli Papini, M., 2002. Removal of molecular weight fractions of COD and phenolic compounds in an integrated treatment of olive oil mill effluents. Biodegradation 13, 401e410. Bekbolet, M., Uyguner, C.S., Selcuk, H., Rizzo, L., Nikolaou, A.D., Meric¸, S., Belgiorno, V., 2005. Application of oxidative removal of NOM to drinking water and formation of disinfection byproducts. Desalination 176, 155e166. Belfroid, A., van Velzen, M., van der Horst, B., Vethaak, D., 2002. Occurrence of bisphenol A in surface water and uptake in fish: evaluation of field measurements. Chemosphere 49, 97e103. ´ lvar, P.M., 2001. PH sequential Beltra´n, F.J., Garcı´a-Araya, J.F., A ozonation of domestic and wine-distillery wastewaters. Water Research 35, 929e936. Beltra´n, F.J., Aguinaco, A., Garcı´a-Araya, J.F., Oropesa, A., 2008. Ozone and photocatalytic processes to remove the antibiotic sulfamethoxazole from water. Water Research 42, 3799e3808. Beltra´n-Heredia, J., Torregrosa, J., Garcı´a, J., Dominguez, J.R., Tierno, J.C., 2001. Degradation of olive mill wastewater by the combination of Fenton’s reagent and ozonation processes with an aerobic biological treatment. Water Science and Technology 44 (5), 103e108. Benotti, M.J., Stanford, B.D., Wert, E.C., Snyder, S.A., 2009. Evaluation of a photocatalytic reactor membrane pilot system for the removal of pharmaceuticals and endocrine disrupting compounds from water. Water Research 43, 1513e1522. Bizani, E., Fytianos, K., Poulios, I., Tsiridis, V., 2006. Photocatalytic decolorization and degradation of dye solutions and wastewaters in the presence of titanium dioxide. Journal of Hazardous Materials 136, 85e94. Borja, R., Banks, C.J., Maestro-Duran, R., Alba, J., 1996. The effects of the most important phenolic constituents of olive mill wastewater on batch anaerobic methanogenesis. Environmental Technology 17, 167e174. Brenes, M., Garcia, A., Garcia, P., Rios, J.J., Garrido, A., 1999. Phenolic compounds in Spanish olive oils. Journal of Agricultural and Food Chemistry 47, 3535e3540. Brose´us, R., Vincent, S., Aboulfadl, K., Daneshvar, A., Sauve´, S., Barbeau, B., Pre´vost, M., 2009. Ozone oxidation of pharmaceuticals, endocrine disruptors and pesticides during drinking water treatment. Water Research 43, 4707e4717. Burgess, J.E., Quarmby, J., Stephenson, T., 1999. Micronutrient supplements to enhance biological wastewater treatment of phosphorus-limited industrial effluent. Process Safety and Environmental Protection 77, 199e204. Cabrera, F., Lopez, R., Martinez-Bordiu`, A., de Lome, E.D., Murillo, J. M., 1996. Land treatment of olive oil mill wastewater. International Biodeterioration and Biodegradation 38, 215e225. Calza, P., Sakkas, V.A., Medana, C., Baiocchi, C., Dimou, A., Pelizzetti, E., Albanis, T., 2006. Photocatalytic degradation
4335
study of diclofenac over aqueous TiO2 suspensions. Applied Catalysis B: Environmental 67, 197e205. Campos, J.C., Borges, R.M.H., Oliveira Filho, A.M., Nobrega, R., Sant’Anna Jr., G.L., 2002. Oilfield wastewater treatment by combined microfiltration and biological processes. Water Research 36, 95e104. Castiglioni, S., Bagnati, R., Fanelli, R., Pomati, F., Calamari, D., Zuccato, E., 2006. Removal of pharmaceuticals in sewage treatment plants in Italy. Environmental Science and Technology 40, 357e363. Catalkaya, E.C., Kargi, F., 2008. Advanced oxidation treatment of pulp mill effluent for TOC and toxicity removals. Journal of Environmental Management 87, 396e404. Chang, Shih-Hsien, Chuang, Shun-Hsing, Li, Heng-Ching, Liang, Hsiu-Hao, Huang, Lung-Chiu, 2009. 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. Journal of Hazardous Materials 166, 1279e1288. Chatzisymeon, E., Xekoukoulotakis, N.P., Diamadopoulos, E., Katsaounis, A., Mantzavinos, D., 2009. Boron-doped diamond anodic treatment of olive mill wastewaters: statistical analysis, kinetic modeling and biodegradability. Water Research 43, 3999e4009. Chen, P.-J., Linden, K.G., Hinton, D.E., Kashiwada, S., Rosenfeldt, E.J., Kullman, S.W., 2006. Biological assessment of bisphenol A degradation in water following direct photolysis and UV advanced oxidation. Chemosphere 65, 1094e1102. Chen, S., Sun, D., Chung, J.-S., 2007. Treatment of pesticide wastewater by moving-bed biofilm reactor combined with Fenton-coagulation pretreatment. Journal of Hazardous Materials 144, 577e584. Coca, M., Pen˜a, M., Gonza´lez, G., 2005. Variables affecting efficiency of molasses fermentation wastewater ozonation. Chemosphere 60, 1408e1415. Coelho, A.D., Sans, C., Agu¨era, A., Go´mez, M.J., Esplugas, S., Dezotti, M., 2009. Effects of ozone pre-treatment on diclofenac: intermediates, biodegradability and toxicity assessment. Science of the Total Environment 407, 3572e3578. Comninellis, C., Kapalka, A., Malato, S., Parsons, S.A., Poulios, I., Mantzavinos, D., 2008. Advanced oxidation processes for water treatment: advances and trends for R&D. Journal of Chemical Technology and Biotechnology 83, 769e776. Crane, M., Watts, C., Boucard, T., 2006. Chronic aquatic environmental risks from exposure to human pharmaceuticals. Science of the Total Environment 367, 23e41. Dalrymple, O.K., Yeh, D.H., Trotz, M.A., 2007. Removing pharmaceuticals and endocrine disrupting compounds from wastewater by photocatalysis. Journal of Chemical Technology and Biotechnology 82, 121e134. Dalzell, D.J.B., Alte, S., Aspichueta, E., de la Sota, A., Etxebarria, J., Gutierrez, M., Hoffmann, C.C., Sales, D., Obst, U., Christofi, N., 2002. A comparison of five rapid direct toxicity assessment methods to determine toxicity of pollutants to activated sludge. Chemosphere 47, 535e545. Dantas, R.F., Canterino, M., Marotta, R., Sans, C., Esplugas, S., Andreozzi, R., 2007. Bezafibrate removal by means of ozonation: primary intermediates, kinetics, and toxicity assessment. Water Research 41, 2525e2532. Date, S., Terabayashi, S., Kobayashi, Y., Fujime, Y., 2005. Effects of chloramines concentration in nutrient solution and exposure time on plant growth in hydroponically cultured lettuce. Scientia Horticulturae 103, 257e265. De Nicola, E., Gallo, M., Iaccarino, M., Meric, S., Oral, R., Russo, T., Sorrentino, T., Tunay, O., Vuttariello, E., Warnau, M., Pagano, G., 2004. Hormetic vs. toxic effects of vegetable tannin in a multi-test study. Archives of Environmental Contamination and Toxicology 46, 336e344.
4336
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 1 1 e4 3 4 0
Directive 2000/60/EC of the Council and the European Parliament of 23 October 2000. OJ L 237, 22.12.2000. Di Salvatore, M., Carafa, A.M., Carratu`, G., 2008. Assessment of heavy metals phytotoxicity using seed germination and root elongation tests: a comparison of two growth substrates. Chemosphere 73, 1461e1464. Drzewicz, P., Na1ezcz-Jawecki, G., Gryz, M., Sawicki, J., Bojanowska-Czajka, A., G1uszewski, W., Kulisa, K., Wo1kowicz, S., Marek, T., 2004. Monitoring of toxicity during degradation of selected pesticides using ionizing radiation. Chemosphere 57, 135e145. Emmanuel, E., Keck, G., Blanchard, J.-M., Vermande, P., Perrodin, Y., 2004. Toxicological effects of disinfections using sodium hypochlorite on aquatic organisms and its contribution to AOX formation in hospital wastewater. Environment International 30, 891e900. Esplugas, S., Bila, D.M., Krause, L.G.T., Dezotti, M., 2007. Ozonation and advanced oxidation technologies to remove endocrine disrupting chemicals (EDCs) and pharmaceuticals and personal care products (PPCPs) in water effluents. Journal of Hazardous Materials 149, 631e642. Evgenidou, E., Fytianos, K., Poulios, I., 2005a. Semiconductorsensitized photodegradation of dichlorvos in water using TiO2 and ZnO as catalysts. Applied Catalysis B: Environmental 59, 81e89. Evgenidou, E., Fytianos, K., Poulios, I., 2005b. Photocatalytic oxidation of dimethoate in aqueous solutions. Journal of Photochemistry and Photobiology, A 175, 29e38. Evgenidou, E., Bizani, E., Christophoridis, C., Fytianos, K., 2007a. Heterogeneous photocatalytic degradation of prometryn in aqueous solutions under UVeVis irradiation. Chemosphere 68, 1877e1882. Evgenidou, E., Konstantinou, I., Fytianos, K., Poulios, I., 2007b. Oxidation of two organophosphorous insecticides by the photo-assisted Fenton reaction. Water Research 41, 2015e2027. Farre´, M., Barcelo´, D., 2003. Toxicity testing of wastewater and sewage sludge by biosensors, bioassays and chemical analysis. Trends Analytical Chemistry 22, 299e310. Farre´, M.J., Franch, M.I., Malato, S., Ayllo´n, J.A., Peral, J., Dome´nech, X., 2005. Degradation of some biorecalcitrant pesticides by homogeneous and heterogeneous photocatalytic ozonation. Chemosphere 58, 1127e1133. Farre´, M.J., Franch, M.I., Ayllo´n, J.A., Peral, J., Dome`nech, X., 2007. Biodegradability of treated aqueous solutions of biorecalcitrant pesticides by means of photocatalytic ozonation. Desalination 211, 22e33. Fernandez-Alba, A.R., Hernando, D., Aguera, A., Caceres, J., Malato, S., 2002. Toxicity assays: a way for evaluating AOPs efficiency. Water Research 36, 4255e4262. Ferraris, M., Chiesara, E., Radice, S., Giovara, A., Frigerio, S., Fumagalli, R., Marabini, L., 2005. Study of potential toxic effects on rainbow trout hepatocytes of surface water treated with chlorine or alternative disinfectants. Chemosphere 60, 65e73. Gagne´, F., Blaise, C., Andre, C., 2006. Occurrence of pharmaceutical products in a municipal effluent and toxicity to rainbow trout (Oncorhynchus mykiss) hepatocytes. Ecotoxicology and Environmental Safety 64, 329e336. Garcia-Montano, J., Torrades, F., Garcia-Hortal, J.A., Dome`nech, X. , Peral, J., 2006. Degradation of Procion Red H-E7B reactive dye by coupling a photo-Fenton system with a sequencing batch reactor. Journal of Hazardous Materials B134, 220e229. Gernjak, W., Maldonado, M.I., Malato, S., Caceres, J., Krutzler, T., Glaser, A., Bauer, R., 2004. Pilot-plant treatment of olive mill wastewater (OMW) by solar TiO2 photocatalysis and solar photo-Fenton. Solar Energy 77, 567e572. Gibs, J., Stackelberg, P.E., Furlong, E.T., Meyer, M., Zaugg, S.T., Lippincott, R.L., 2007. Persistence of pharmaceuticals and other
organic compounds in chlorinated drinking water as a function of time. Science of the Total Environment 373, 240e249. Gogate, P.R., 2002. Cavitation: an auxiliary technique in wastewater treatment schemes. Advances in Environmental Research 6, 335e358. Gogate, P.R., Pandit, A.B., 2004a. A review of imperative technologies for wastewater treatment I: oxidation technologies at ambient conditions. Advances in Environmental Research 8, 501e551. Gogate, P.R., Pandit, A.B., 2004b. A review of imperative technologies for wastewater treatment II: hybrid methods. Advances in Environmental Research 8, 553e597. Go´mez, M.J., Sirtori, C., Mezcua, M., Ferna´ndez-Alba, A.R., Agu¨era, A., 2008. Photodegradation study of three dipyrone metabolites in various water systems: identification and toxicity of their photodegradation products. Water Research 42, 2698e2706. Gonza´lez, O., Esplugas, M., Sans, C., Torres, A., Esplugas, S., 2009. Performance of a sequencing batch biofilm reactor for the treatment of pre-oxidized sulfamethoxazole solutions. Water Research 43, 2149e2158. Gonze, E., Fourel, L., Gonthier, Y., Boldo, P., Bernis, A., 1999. Wastewater pretreatment with ultrasonic irradiation to reduce toxicity. Chemical Engineering Journal 73, 93e100. Gupta, N., Pant, S.C., Vijayaraghavan, R., Rao, P.V.L., 2003. Comparative toxicity evaluation of cyanobacterial cyclic peptide toxin microcystin variants (LR, RR, YR) in mice. Toxicology 188, 285e296. Gutie´rrez, M., Etxebarria, J., de las Fuentes, L., 2002. Evaluation of wastewater toxicity: comparative study between Microtox and activated sludge oxygen uptake inhibition. Water Research 36, 919e924. Guzzella, L., Feretti, D., Monarca, S., 2002. Advanced oxidation and adsorption technologies for organic micropollutant removal from lake water used as drinking-water supply. Water Research 36, 4307e4318. Hammes, F., Salhi, E., Ko¨ster, O., Kaiser, H.-P., Egli, T., von Gunten, U., 2006. Mechanistic and kinetic evaluation of organic disinfection by-product and assimilable organic carbon (AOC) formation during the ozonation of drinking water. Water Research 40, 2275e2286. Hernando, M.D., Ferna´ndez-Alba, A.R., Tauler, R., Barcelo´, D., 2005. Toxicity assays applied to wastewater treatment. Talanta 65, 358e366. Herrmann, J.M., Guillard, C., Arguello, M., Agu¨era, A., Tejedor, A., Piedra, L., Ferna´ndez-Alba, A., 1999. Photocatalytic degradation of pesticide pirimiphos-methyl: determination of the reaction pathway and identification of intermediate products by various analytical methods. Catalysis Today 54, 353e367. ISO (International Organization for Standardization), 1989. Water QualitydFresh Water Algal Growth Inhibition Test with Scenedesmus subspicatus and Selenastrum capricornutum. ISO 8692, Geneva, Switzerland. ISO (International Organisation for Standardization), 1996a. Water Quality: Determination of the Inhibition of the Mobility of Daphnia magna Straus (Cladocera, Crustacea) e Acute Toxicity Test. ISO 6341, Geneva, Switzerland. ISO (International Organization for Standardization), 1996b. Water QualitydDetermination of the Acute Lethal Toxicity of Substances to a Freshwater Fish (Brachydanio rerio Hamilton Buchanan, Teleostei, Cyprinidae). Part 1: Static Method. ISO 7346-1, Geneva, Switzerland. ISO (International Organization for Standardization), 1998. Water QualitydDetermination of the Inhibitory Effect of Water Samples on the Light Emission of Vibrio fischeri (Luminescent Bacteria Test)dPart 2: Method Using Liquid-Dried Bacteria. ISO 11348-2, Geneva, Switzerland.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 1 1 e4 3 4 0
ISO (International Organization for Standardization), 2007. Water Quality-Test for Inhibition of Oxygen Consumption by Activated Sludge for Carbonaceous and Ammonium Oxidation. ISO 8192, Geneva, Switzerland. Jiao, S., Zheng, S., Yin, D., Wang, L., Chen, L., 2008. Aqueous photolysis of tetracycline and toxicity of photolytic products to luminescent bacteria. Chemosphere 73, 377e382. Jin, X., Jiang, G., Huang, G., Liu, J., Zhou, Q., 2004. Determination of 4-tert-octylphenol, 4-nonylphenol and bisphenol A in surface waters from the Haihe River in Tianjin by gas chromatographyemass spectrometry with selected ion monitoring. Chemosphere 56, 1113e1119. Jochimsen, J.C., Schenk, H., Jekel, M.R., Hegemann, W., 1997. Combined oxidative and biological treatment for separated streams of tannery wastewater. Water Science Technology 36, 209e216. Joubert, G., 1980. A bioassay application for quantitative toxicity measurements, using the green algae Selenastrum capricornutum. Water Research 14, 1759e1763. Jurczak, T., Tarczynska, M., Izydorczyk, K., Mankiewicz, J., Zalewski, M., Meriluoto, J., 2005. Elimination of microcystins by water treatment processesdexamples from Sulejow Reservoir, Poland. Water Research 39, 2394e2406. Justino, C.I., Duarte, K., Loureiro, F., Pereira, R., Antunes, S.C., Marques, S.M., Gonc¸alves, F., Rocha-Santos, T.A.P., Freitas, A.C., 2009. Toxicity and organic content characterization of olive oil mill wastewater undergoing a sequential treatment with fungi and photo-Fenton oxidation. Journal of Hazardous Materials 172, 1560e1572. Kang, J.-H., Kondo, F., Katayama, Y., 2006. Human exposure to bisphenol A. Toxicology 226, 79e89. Khoufi, S., Aloui, F., Sayadi, S., 2009. Pilot scale hybrid process for olive mill wastewater treatment and reuse. Chemical Engineering and Processing 48, 643e650. Kidak, R., Ince, N.H., 2007. Catalysis of advanced oxidation reactions by ultrasound: a case study with phenol. Journal of Hazardous Materials 146, 630e635. Kim, J.-K., Choi, K., Cho, I.-H., Son, H.-S., Zoh, K.-D., 2007. Application of a microbial toxicity assay for monitoring treatment effectiveness of pentachlorophenol in water using UV photolysis and TiO2 photocatalysis. Journal of Hazardous Materials 148, 281e286. Kim, T.-S., Kim, J.-K., Choi, K., Stenstrom, M.K., Zoh, K.-D., 2006. Degradation mechanism and the toxicity assessment in TiO2 photocatalysis and photolysis of parathion. Chemosphere 62, 926e933. Klamerth, N., Rizzo, L., Malato, S., Maldonado, M.I., Agu¨era, A., Ferna´ndez-Alba, A.R., 2010. Degradation of fifteen emerging contaminants at mg L1 initial concentrations by mild solar photo-Fenton in MWTP effluents. Water Research 44, 545e554. Klavarioti, M., Mantzavinos, D., Kassinos, D., 2009. Removal of residual pharmaceuticals from aqueous systems by advanced oxidation processes. Environment International 35, 402e417. Konstantinou, I.K., Albanis, T.A., 2003. Photocatalytic transformation of pesticides in aqueous titanium dioxide suspensions using artificial and solar light: intermediates and degradation pathways. Applied Catalysis B: Environmental 42, 319e335. Konstantinou, I.K., Albanis, T.A., 2004. TiO2-assisted photocatalytic degradation of azo dyes in aqueous solution: kinetic and mechanistic investigations: a review. Applied Catalysis B: Environmental 49, 1e14. Kralj, M.B., Cernigoj, U., Franko, M., Trebse, P., 2007. Comparison of photocatalysis and photolysis of malathion, isomalathion, malaoxon, and commercial malathiondproducts and toxicity studies. Water Research 41, 4504e4514. Kuch, H.M., Ballschmiter, K., 2001. Determination of endocrine disrupting phenolic compounds and estrogens in surface and
4337
drinking water by HRGCe(NCI)eMS in the pictogram per liter range. Environmental Science and Technology 35, 3201e3206. Kusic, H., Koprivanac, N., Horvat, S., Bakija, S., Loncaric Bozic, A., 2009. Modeling dye degradation kinetic using dark- and photo-Fenton type processes. Chemical Engineering Journal 155, 144e154. Kusvuran, E., Gulnaz, O., Irmak, S., Atanur, O.M., Yavuz, H.I., Erbatur, O., 2004. Comparison of several advanced oxidation processes for the decolorization of Reactive Red 120 azo dye in aqueous solution. Journal of Hazardous Materials B109, 85e93. Kusvuran, E., Irmak, S., Yavuz, H.I., Samil, A., Erbatur, O., 2005. Comparison of the treatment methods efficiency for decolorization and mineralization of Reactive Black 5 azo dye. Journal of Hazardous Materials B119, 109e116. Lapertot, M., Ebrahimi, S., Dazio, S., Rubinelli, A., Pulgarin, C., 2007. Photo-Fenton and biological integrated process for degradation of a mixture of pesticides. Journal of Photochemistry and Photobiology, A 186, 34e40. Lautenschlager, K., Boon, N., Wang, Y., Egli, T., Hammes, F., 2010. Overnight stagnation of drinking water in household taps induces microbial growth and changes in community composition. Water Research 44, 4868e4877. Lawton, L.A., Robertson, P.K.J., Cornish, B.J.P.A., Jaspars, M., 1999. Detoxification of microcystins (Cyanobacterial Hepatotoxins) using TiO2 photocatalytic oxidation. Environmental Science and Technology 33, 771e775. Lawton, L.A., Robertson, P.K.J., Cornish, B.J.P.A., Marr, I.L., Jaspars, M., 2003. Processes influencing surface interaction and photocatalytic destruction of microcystins on titanium dioxide photocatalysts. Journal of Catalysis 213, 109e113. LeChevallier, M.W., Schultz, W., Lee, R.G., 1991. Bacterial nutrients in drinking water. Applied Environmental Microbiology 57 (3), 857e862. Ledakowicz, S., Gonera, M., 1999. Optimisation of oxidants dose for combined chemical and biological treatment of textile wastewater. Water Research 33, 2511e2516. Ledakowicz, S., Solecka, M., Zylla, R., 2001. Biodegradation, decolourisation and detoxification of textile wastewater enhanced by advanced oxidation processes. Journal of Biotechnology 89, 175e184. Ledakowicz, S., Michniewicz, M., Jagiella, A., Stufka-Olczyk, J., Martynelis, M., 2006. Elimination of resin acids by advanced oxidation processes and their impact on subsequent biodegradation. Water Research 40, 3439e3446. Levec, J., Pintar, A., 2007. Catalytic wet-air oxidation processes: a review. Catalysis Today 124, 172e184. Lin, D., Xing, B., 2007. Phytotoxicity of nanoparticles: inhibition of seed germination and root growth. Environmental Pollution 150, 243e250. Liu, I., Lawton, L.A., Cornish, B., Robertson, P.K.J., 2002. Mechanistic and toxicity studies of the photocatalytic oxidation of microcystin-LR. Journal of Photochemistry and Photobiology A 148, 349e354. Malato, S., Blanco, J., Alarco´n, D.C., Maldonado, M.I., Ferna´ndezIba´n˜ez, P., Gernjak, W., 2007. Photocatalytic decontamination and disinfection of water with solar collectors. Catalysis Today 122, 137e149. Mantzavinos, D., Kalogerakis, N., 2005. Treatment of mill effluents Part I. organic matter degradation by chemical and biological processes e an overview. Environment International 31, 289e295. Mantzavinos, D., Psillakis, E., 2004. Enhancement of biodegradability of industrial wastewaters by chemical oxidation pre-treatment. Journal of Chemical Technology and Biotechnology 79, 431e454. Marttinen, K., Kettunen, R.H., Sormunen, K.M., Soimasuo, R.M., Rintala, J.A., 2002. Screening of physicalechemical methods for removal of organic material, nitrogen and toxicity from low strength landfill leachates. Chemosphere 46 (6), 851e858.
4338
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 1 1 e4 3 4 0
Matilainen, A., Sillanpa¨a¨, M., 2010. Removal of natural organic matter from drinking water by advanced oxidation processes. Chemosphere 80, 351e365. Me´ndez-Arriaga, F., Esplugas, S., Gime´nez, J., 2008. Photocatalytic degradation of non-steroidal anti-inflammatory drugs with TiO2 and simulated solar irradiation. Water Research 42, 585e594. Meric¸, S., De Nicola, E., Iaccarino, M., Gallo, M., Di Gennaro, A., Morrone, G., Warnau, M., Belgiorno, V., Pagano, G., 2005. Toxicity of leather tanning wastewater effluents in sea urchin early development and in marine microalgae. Chemosphere 61, 208e217. lu, K., 2010. Pre-treatment Mert, B.K., Yonar, T., Kilic¸, M.Y., Kestiog studies on olive oil mill effluent using physicochemical, Fenton and Fenton like oxidations processes. Journal of Hazardous Materials 174, 122e128. Miao, H.-F., Qin, F., Tao, G.-J., Tao, W.-Y., Ruan, W.-Q., 2010. Detoxification and degradation of microcystin-LR and -RR by ozonation. Chemosphere 79, 355e361. Migliore, L., Civitareale, C., Brambilla, G., Di Pelupis, G., 1997. Toxicity of several important agricultural antibiotics to Artemia. Water Research 31 (7), 1801e1806. Mishra, V.S., Mahajani, V.V., Joshi, J.B., 1995. Wet air oxidation. Industrial and Engineering Chemistry Research 34, 2e48. Muthukumar, M., Sargunamani, D., Selvakumar, N., Venkata Rao, J., 2004. Optimisation of ozone treatment for colour and COD removal of acid dye effluent using central composite design experiment. Dyes and Pigments 63, 127e134. Naddeo, V., Meric¸, S., Kassinos, D., Belgiorno, V., Guida, M., 2009. Fate of pharmaceuticals in contaminated urban wastewater effluent under ultrasonic irradiation. Water Research 43, 4019e4027. Najjar, W., Azabou, S., Sayadi, S., Ghorbel, A., 2009. Screening of FeeBEA catalysts for wet hydrogen peroxide oxidation of crude olive mill wastewater under mild conditions. Applied Catalysis B: Environmental 88, 299e304. Neamtu, M., Yediler, A., Siminiceanu, I., Macoveanu, M., Kettrup, A., 2004. Decolorization of disperse red 354 azo dye in water by several oxidation processesda comparative study. Dyes and Pigments 60, 61e68. Neumann, M., Schulz, R., Schafer, K., Muller, W., Mannheller, W., Liess, M., 2002. The significance of entry routes as point and non-point sources of pesticides in small streams. Water Research 36, 835e842. Nikolaou, A., Rizzo, L., Selcuk, H., 2007. Control of Disinfection ByProducts in Drinking Water Systems. Nova Science Publishers, Inc., 400 Oser Avenue, Suite 1600, Hauppauge, NY 11788, ISBN 88-7850-003-8. OECD (Organization for Economic Cooperation and Development), 1984. Terrestrial plants, growth test. In: Guidelines of the OECD for Testing Chemical Products. OECD Method 208, Paris. OECD (Organisation for Economic Cooperation and Development), 1992. Fish Early Life Stage Toxicity Test. OECD Guidelines for the Testing of Chemicals. OECD No. 210, Paris. OECD (Organization for Economic Cooperation and Development), 2010. Guidelines for the Testing of Chemicals: Activated Sludge, Respiration Inhibition Test (Carbon and Ammonium Oxidation). OECD No. 209, Paris. Oller, I., Malato, S., Sanchez-Perez, J.A., Maldonado, M.I., Gasso, R. , 2007. Detoxification of wastewater containing five common pesticides by solar AOPs e biological coupled system. Catalysis Today 129, 69e78. Oral, R., Meric¸, S., De Nicola, E., Petruzzelli, D., Della Rocca, C., Pagano, G., 2007. Multi-species toxicity evaluation of a chromium-based leather tannery wastewater. Desalination 211, 48e57. Pagano, G., Esposito, A., Giordano, G.G., 1982. Fertilization and larval development in sea urchins following exposure of
gametes and embryos to cadmium. Archives of Environmental Contamination and Toxicology 11, 47e55. Pagano, G., Iaccarino, M., De Biase, A., Meric, S., Trieff, N.M., 2001. Toxicity of the R6 fungicide mixture as related to the effects of its components, cymoxanil and copper oxychloride on sea urchin fertilization and development. Human and Experimental Toxicology 20, 404e411. Pagga, U., Bachner, J., Strotmann, U., 2006. Inhibition of nitrification in laboratory tests and model wastewater treatment plants. Chemosphere 65, 1e8. Pala´cio, S.M., Espinoza-Quin˜ones, F.R., Mo´denes, A.N., Oliveira, C.C. , Borba, F.H., Silva Jr., F.G., 2009. Toxicity assessment from electro-coagulation treated-textile dye wastewaters by bioassays. Journal of Hazardous Materials 172, 330e337. Pehlivanoglu, E., Sedlak, D.L., 2004. Bioavailability of wastewaterderived organic nitrogen to the alga Selenastrum capricornutum. Water Research 38, 3189e3196. Peng, G., Roberts, J.C., 2000. Solubility and toxicity of resin acids. Water Research 34, 2779e2785. Pe´rez-Estrada, Leonidas A., Malato, Sixto, Aguera, Ana, Fernandez-Alba, Amadeo R., 2007. Degradation of dipyrone and its main intermediates by solar AOPs e identification of intermediate products and toxicity assessment. Catalysis Today 129, 207e214. Pe´rez, S., Aga, D.S., 2005. Recent advances in the sample preparation, liquid chromatography tandem mass spectrometric analysis and environmental fate of microcystins in water. Trends in Analytical Chemistry 24, 658e670. Persoone, G., Van Haecke, P., 1981. Intercalibration exercise on a short-term standard toxicity test with Artemia naupli. Final Report. Contract CEE-ENV-396 B(N). Petala, M., SamarasZouboulis, P.A., Kungolos, A., Sakellaropoulos, G.P., 2008. Influence of ozonation on the in vitro mutagenic and toxic potential of secondary effluents. Water Research 42, 4929e4940. Pignatello, J.J., Oliveros, E., MacKay, A., 2006. Advanced oxidation processes for organic contaminant destruction based on the Fenton reaction and related chemistry. Critical Review in Environmental Science and Technology 36, 1e84. Polanska, M., Huysman, K., van Keer, C., 2005. Investigation of assimilable organic carbon (AOC) in flemish drinking water. Water Research 39, 2259e2266. Radix, P., Le´onard, M., Papantoniou, C., Roman, G., Saouter, E., Gallotti-Schmitt, S., Thie´baud, H., Vasseur, P., 2000. Comparison of four chronic toxicity tests using algae, bacteria, and invertebrates assessed with sixteen chemicals. Ecotoxicology and Environmental Safety 47, 186e194. , J., Petrovic , M., Barcelo´, D., 2009. Fate and distribution Radjenovic of pharmaceuticals in wastewater and sewage sludge of the conventional activated sludge (CAS) and advanced membrane bioreactor (MBR) treatment. Water Research 43, 831e841. Rajeshwar, K., Osugi, M.E., Chanmanee, W., Chenthamarakshan, C.R., Zanoni, M.V.B., Kajitvichyanukul, P., Krishnan-Ayer, R., 2008. Heterogeneous photocatalytic treatment of organic dyes in air and aqueous media. Journal of Photochemistry and Photobiology C: Photochemistry Reviews 9, 171e192. Renoux, A.Y., Tyagi, R.D., Samson, R., 2001. Assessment of toxicity reduction after metal removal in bioleached sewage sludge. Water Research 35, 1415e1424. Reungoat, J., Macova, M., Escher, B.I., Carswell, S., Mueller, J.F., Keller, J., 2010. Removal of micropollutants and reduction of biological activity in a full scale reclamation plant using ozonation and activated carbon filtration. Water Research 44, 625e637. Richardson, S.D., 1998. Encyclopedia of Environmental Analysis and Remediation. In: Meyers, R.A. (Ed.), Drinking water
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 1 1 e4 3 4 0
disinfection by-products, vol. 3. John Wiley & Sons, New York, pp. 1398e1421. Rizzo, L., Belgiorno, V., Gallo, M., Meric¸, S., 2005. Removal of THM precursors from a high-alkaline surface water by enhanced coagulation and behaviour of THMFP toxicity on D. magna. Desalination 176, 177e188. Rizzo, L., Lofrano, G., Grassi, M., Belgiorno, V., 2008. Pre-treatment of olive mill wastewater by chitosan coagulation and advanced oxidation processes. Separation and Purification Technology 63, 648e653. Rizzo, L., Meric, S., Guida, M., Kassinos, D., Belgiorno, V., 2009a. Heterogenous photocatalytic degradation kinetics and detoxification of an urban wastewater treatment plant effluent contaminated with pharmaceuticals. Water Research 43, 4070e4078. Rizzo, L., Meric, S., Kassinos, D., Guida, M., Russo, F., Belgiorno, V., 2009b. Degradation of diclofenac by TiO2 photocatalysis: UV absorbance kinetics and process evaluation through a set of toxicity bioassays. Water Research 43, 979e988. Rivas, J., Gimeno, O., Beltra´n, F., 2009. Wastewater recycling: application of ozone based treatments to secondary effluents. Chemosphere 74, 854e859. Robidoux, P.Y., Gong, P., Sarrazin, M., Bardai, G., Paquet, L., Hawari, J., Dubois, C., Sunahara, G.I., 2004. Toxicity assessment of contaminated soils from an antitank firing range. Ecotoxicology and Environmental Safety 58, 300e313. Rodrigues, A.C., Boroski, M., Shimada, N.S., Garcia, J.C., Nozaki, J., Hioka, N., 2008. Treatment of paper pulp and paper mill wastewater by coagulationeflocculation followed by heterogeneous photocatalysis. Journal of Photochemistry and Photobiology, A 194, 1e10. Rosal, R., Rodrı´guez, A., Perdigo´n-Melo´n, J.A., Mezcua, M., Hernando, M.D., Leto´n, P., Garcı´a-Calvo, E., Agu¨era, A., Ferna´ndez-Alba, A.R., 2008. Removal of pharmaceuticals and kinetics of mineralization by O3/H2O2 in a biotreated municipal wastewater. Water Research 42, 3719e3728. Rosenfeldt, E.J., Chen, P.J., Kullman, S., Linden, K.G., 2007. Destruction of estrogenic activity in water using UV advanced oxidation. Science of the Total Environment 377, 105e113. Ruebhart, D.R., Cock, I.E., Shaw, G.R., 2008. Brine shrimp bioassay: importance of correct taxonomic identification of Artemia (Anostraca) species. Environmental Toxicology 23, 555e560. Sacher, F., Lange, F.T., Brauch, H.-J., Blankenhorn, I., 2001. Pharmaceuticals in groundwaters: analytical methods and results of a monitoring program in Baden-Wu¨rttemberg, Germany. Journal of Chromatography A 938, 199e210. Sakkas, V.A., Calza, P., Medana, C., Villioti, A.E., Baiocchi, C., Pelizzetti, E., Albanis, T., 2007. Heterogeneous photocatalytic degradation of the pharmaceutical agent salbutamol in aqueous titanium dioxide suspensions. Applied Catalysis B: Environmental 77, 135e144. Sanches, S., Barreto Crespo, M.T., Pereira, V.J., 2010. Drinking water treatment of priority pesticides using low pressure UV photolysis and advanced oxidation processes. Water Research 44, 1809e1818. Sa´nchez-Polo, M., Lo´pez-Pen˜alver, J., Prados-Joya, G., FerroGarcı´a, M.A., Rivera-Utrilla, J., 2009. Gamma irradiation of pharmaceutical compounds, nitroimidazoles, as a new alternative for water treatment. Water Research 43, 4028e4036. Sarria, V., Parra, S., Adler, N., Pe´ringer, P., Benitez, N., Pulgarin, C., 2002. Recent developments in the coupling of photoassisted and aerobic biological processes for the treatment of biorecalcitrant compounds. Catalysis Today 76, 301e315. Sauer, T.K., Casaril, L., Oberziner, A.L.B., Jose´, H.J., Muniz Moreira, R.P., 2006. Advanced oxidation processes applied to tannery wastewater containing Direct Black 38delimination
4339
and degradation kinetics. Journal of Hazardous Materials B135, 274e279. Schrank, S.G., Jose´, H.J., Moreira, R.F.P.M., Schroder, H.Fr., 2004. Elucidation of the behavior of tannery wastewater under advanced oxidation conditions. Chemosphere 56, 411e423. Schrank, S.G., Jose´, H.J., Moreira, R.F.P.M., Schroder, H.Fr., 2005. Applicability of Fenton and H2O2/UV reactions in the treatment of tannery wastewaters. Chemosphere 60, 644e655. Segner, H., Caroll, K., Fenske, M., Janssen, C.R., Maack, G., Pascoe, D., Schafers, C., Vandenbergh, G.F., Watts, M., Wenzel, A., 2003. Identification of endocrine-disrupting effects in aquatic vertebrates and invertebrates: report from the European IDEA project. Ecotoxicology and Environmental Safety 54, 302e314. Segura, C., Zaror, C., Mansilla, H.D., Mondaca, M.A., 2008. Imidacloprid oxidation by photo-Fenton reaction. Journal of Hazardous Materials 150, 679e686. Selcuk, H., Eremektar, G., Meric, S., 2006. The effect of pre-ozone oxidation on acute toxicity and inert soluble COD fractions of a textile finishing industry wastewater. Journal of Hazardous Materials B137, 254e260. Selcuk, H., Rizzo, L., Nikolaou, A., Meric, S., Belgiorno, V., Bekbolet, M., 2007. DBPs formation and toxicity monitoring in different origin water treated by ozone and alum/PAC coagulation. Desalination 210, 31e43. Shemer, H., Kunukcu, Y.K., Linden, K.G., 2006. Degradation of the pharmaceutical metronidazole via UV, Fenton and photoFenton processes. Chemosphere 63, 269e276. Silva, A.C., Dezotti, M., Sant’Anna Jr., G.L., 2004. Treatment and detoxification of a sanitary landfill leachate. Chemosphere 55, 207e214. Somensi, C.A., Simionatto, E.L., Bertoli, S.L., Wisniewski Jr., A., Radetski, C.M., 2010. Use of ozone in a pilot-scale plant for textile wastewater pre-treatment: physico-chemical efficiency, degradation by-products identification and environmental toxicity of treated wastewater. Journal of Hazardous Materials 175, 235e240. Stackelberg, P.E., Gibs, J., Furlong, E.T., Meyer, M.T., Zaugg, S.D., Lippincott, R.L., 2007. Efficiency of conventional drinkingwater-treatment processes in removal of pharmaceuticals and other organic compounds. Science of the Total Environment 377, 255e272. Stalter, D., Magdeburg, A., Weil, M., Knacker, T., Oehlmann, J., 2010. Toxication or detoxication? In vivo toxicity assessment of ozonation as advanced wastewater treatment with the rainbow trout. Water Research 44, 439e448. Staples, C.A., Dorn, P.B., Klecka, G.M., O’Block, S.T., Harris, L.R., 1998. A review of the environmental fate, effects, and exposures of bisphenol A. Chemosphere 36, 2149e2173. Strotmann, U.J., Keinath, A., Hfittenhain, S.H., 1995. Biological test systems for monitoring the operation of wastewater treatment plants. Chemosphere 30, 327e338. Tezcanli-Guyer, G., Ince, N.H., 2003. Degradation and toxicity reduction of textile dyestuff by ultrasound. Ultrasonics Sonochemistry 10, 235e240. Tezcanli-Guyer, G., Ince, N.H., 2004. Individual and combined effects of ultrasound, ozone and UV irradiation: a case study with textile dyes. Ultrasonics 42, 603e609. Tisler, T., Zagorc-Koncan, J., Cotman, M., Drolc, A., 2004. Toxicity potential of disinfection agent in tannery wastewater. Water Research 38, 3503e3510. Tothill, I.E., Turner, A.P.F., 1996. Developments in bioassay methods for toxicity testing in water treatment. Trends in analytical Chemistry 75, 178e188. Trovo´, A.G., Nogueira, R.F.P., Agu¨era, A., Fernandez-Alba, A.R., Sirtori, C., Malato, S., 2009. Degradation of sulfamethoxazole
4340
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 1 1 e4 3 4 0
in water by solar photo-Fenton. Chemical and toxicological evaluation. Water Research 43, 3922e3931. USEPA, 1988. Methods for Toxicity Tests of Single Substances and Liquid Complex Wastes with Marine Unicellular Algae. Environmental Research Laboratory, Gulf Breeze, FL. EPA/600/04. USEPA, 1989. Lettuce seed germination (Lactuca sativa). In: Protocol for Short Term Toxicity Screening of Hazardous Waste Sites. Environmental Research Laboratory, Corvallis, OR EPA/600/3-88/029. USEPA, 1996. Ecological Effects Test Guidelines OPPTS 850.1075, Fish Acute Toxicity Test, Freshwater and Marine. Environmental Protection Agency, Prevention, Pesticides and Toxic Substances (7101). EPA 712-C-96-118. USEPA, 2002. Methods for Measuring the Acute Toxicity of Effluents and Receiving Waters to Freshwater and Marine Organisms, fifth ed. U.S. Environmental Protection Agency Office of Water (4303T), 1200 Pennsylvania Avenue, NW Washington, DC 20460. EPA-821-R-02-012. Vajnhandl, Simona, Le Marechal, Alenka Majcen, 2007. Case study of the sonochemical decolouration of textile azo dye Reactive Black 5. Journal of Hazardous Materials 141, 329e335. Valerio, M.E., Garcı´a, J.F., Peinado, F.M., 2007. Determination of phytotoxicity of soluble elements in soils, based on a bioassay with lettuce (Lactuca sativa L.). Science of the Total Environment 378, 63e66. Van der Kooij, D., Visser, A., Hijnen, W.A.M., 1982. Determination of easily assimilable organic carbon in drinking water. Journal of the American Water Works Association 74, 540e545. Van der Kooij, D., 1990. Assimilable organic carbon (AOC) in drinking water. In: McFeters, G.A. (Ed.), Drinking Water Microbiology, pp. 57e87. New York. Von Gunten, U., 2003. Ozonation of drinking water: Part I. Oxidation kinetics and product formation. Water Research 37, 1443e1467. Vrouwenvelder, H.S., van Paassen, J.A.M., Folmer, H.C., Hofman, J. A.M.H., Nederlof, M.M., van der Kooij, D., 1998. Biofouling of membranes for drinking water production. Desalination 118, 157e166. Walsh, G.E., Bahner, L.H., Horning, W.B., 1980. Toxicity of textile mill effluents to freshwater and estuarine algae, crustaceans and fishes. Environmental Pollution Series A, Ecological and Biological 21, 169e179. Weinrich, L.A., Jjemba, P.K., Giraldo, E., LeChevallier, M.W., 2010. Implications of organic carbon in the deterioration of water
quality in reclaimed water distribution systems. Water Research 44, 5367e5375. Weng, D., Lu, Y., Wei, Y.N., Liu, Y., Shen, P.P., 2007. The role of ROS in microcystin-LR-induced hepatocyte apoptosis and liver injury in mice. Toxicology 232, 15e23. Westerhoff, P., Yoon, Y., Snyder, S., Wert, E., 2005. Fate of endocrine-disruptor, pharmaceutical, and personal care product chemicals during simulated drinking water treatment processes. Environmental Science and Technology 39, 6649e6663. WHO, 2008. Guidelines for drinking-water quality: incorporating 1st and 2nd addenda, vol. 1, Recommendations, third ed. Wieczorek, J.K., Wieczorek, Z.J., 2007. Phytotoxicity and accumulation of anthracene applied to the foliage and sandy substrate in lettuce and radish plants. Ecotoxicology and Environmental Safety 66, 369e377. Wong, S.L., Wainwright, J.F., Piment, J., 1995. Quantification of total and metal toxicity in wastewater using algal bioassays. Aquatic Toxicology 31, 57e75. Yeber, M.C., Rodriguez, J., Freer, J., Baeza, J., Duran, N., Mansilla, H.D., 1999. Advanced oxidation of a pulp mill bleaching wastewater. Chemosphere 39, 1679e1688. Zapata, A., Oller, I., Rizzo, L., Hilgert, S., Maldonado, M.I., Sa´nchez-Pe´rez, J.A., Malato, S., 2010. Evaluation of operating parameters involved in solar photo-Fenton treatment of wastewater: interdependence of initial pollutant concentration, temperature and iron concentration. Applied Catalysis B: Environmental 97, 292e298. Zapata, A., Velegraki, T., Sa´nchez-Pe´rez, J.A., Mantzavinos, D., Maldonado, M.I., Malato, S., 2009. Solar photo-Fenton treatment of pesticides in water: effect of iron concentration on degradation and assessment of ecotoxicity and biodegradability. Applied Catalysis B: Environmental 88, 448e454. Zoh, K.-D., Kim, T.-S., Kim, J.-G., Choi, K., Yi, S.-M., 2006. Parathion degradation and toxicity reduction in solar photocatalysis and photolysis. Water Science and Technology 53 (3), 1e8. Zorrig, W., Rouached, A., Shahzada, Z., Abdelly, C., Davidiana, J.-C., Berthomieu, P., 2010. Identification of three relationships linking cadmium accumulation to cadmium tolerance and zinc and citrate accumulation in lettuce. Journal of Plant Physiology 167, 1239e1247.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 4 1 e4 3 5 4
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Review
Occurrence and control of nitrogenous disinfection by-products in drinking water e A review Tom Bond*, Jin Huang, Michael R. Templeton, Nigel Graham Department of Civil and Environmental Engineering, Imperial College London, London SW7 2AZ, United Kingdom
article info
abstract
Article history:
The presence of nitrogenous disinfection by-products (N-DBPs), including nitrosamines,
Received 10 November 2010
cyanogen halides, haloacetonitriles, haloacetamides and halonitromethanes, in drinking
Received in revised form
water is of concern due to their high genotoxicity and cytotoxicity compared with regu-
29 May 2011
lated DBPs. Occurrence of N-DBPs is likely to increase if water sources become impacted by
Accepted 30 May 2011
wastewater and algae. Moreover, a shift from chlorination to chloramination, an option for
Available online 7 June 2011
water providers wanting to reduce regulated DBPs such as trihalomethanes (THMs) and haloacetic acids (HAAs), can also increase certain N-DBPs. This paper provides a critical
Keywords:
review of the occurrence and control of N-DBPs. Data collated from surveys undertaken in
NDMA
the United States and Scotland were used to calculate that the sum of analysed haloni-
Nitrosamines
tromethanes represented 3e4% of the mass of THMs on a median basis; with Pearson
Haloacetonitriles
product moment correlation coefficients of 0.78 and 0.83 between formation of dihaloa-
Cyanogen halides
cetonitriles and that of THMs and HAAs respectively. The impact of water treatment
Halonitromethanes
processes on N-DBP formation is complex and variable. While coagulation and filtration are
Haloacetamides
of moderate efficacy for the removal of N-DBP precursors, such as amino acids and amines, biofiltration, if used prior to disinfection, is particularly successful at removing cyanogen halide precursors. Oxidation before final disinfection can increase halonitromethane formation and decrease N-nitrosodimethylamine, and chloramination is likely to increase cyanogen halides and NDMA relative to chlorination. ª 2011 Elsevier Ltd. All rights reserved.
Contents 1. 2.
Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Occurrence of N-DBPs in drinking water . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1. Interpreting N-DBP occurrence data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2. Haloacetonitriles (HANs) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3. Haloacetamides (HAcAms) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4342 4343 4343 4343 4344
* Corresponding author. Present address: Pollution Research Group, School of Chemical Engineering, University of KwaZulu-Natal, Durban 4041, South Africa. Tel.: þ27 31 260 3131; fax: þ27 31 260 3241. E-mail address:
[email protected] (T. Bond). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.05.034
4342
3. 4.
5.
1.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 4 1 e4 3 5 4
2.4. Cyanogen halides (CNX) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.5. Halonitromethanes (HNMs) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.6. Nitrosamines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.7. Other N-DBPs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Precursor sources and formation pathways . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Key factors in the formation of N-DBPs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1. Water quality parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1.1. The impact of pH . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1.2. The impact of bromide and iodide . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2. The impact of treatment and disinfection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2.1. The impact of treatment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2.2. The impact of pre-oxidation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2.3. The impact of chlorination and chloramination . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusions: minimising N-DBPs in water treatment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Background
Research into disinfection by-products (DBPs), an unintentional result of water treatment, stems from the mid-1970s, when the formation of trihalomethanes (THMs) was linked to reactions between chlorine and natural organic matter (NOM) in Dutch drinking water (Rook, 1974). By the turn of that decade the THMs were regulated in the USA at 100 mg L1 due to a cancer risk, and a second group of DBPs, the haloacetic acids (HAAs), had been identified in drinking water at comparable levels to THMs. Soon after, the haloacetonitriles (HANs), an important group of N-DBPs, were detected in chlorinated natural waters (Oliver, 1983; Trehy and Bieber, 1981). An enhanced risk of bladder cancer has been associated with exposure to DBPs (Villanueva et al., 2007), although the contribution of various DBPs to this association remains uncertain, given that no identified chlorination DBPs are believed to be plausible bladder carcinogens (Hrudey, 2009). Over 600 DBPs have been reported in drinking water or simulated laboratory disinfection tests, resulting from the use of chlorine and other disinfectants, notably chloramines, ozone and chlorine dioxide (Krasner et al., 2006). However, data for many nitrogenous DBPs (N-DBPs) remains relatively limited. In the United States (US) selected N-DBPs were analysed in water treatment plant (WTP) surveys undertaken in 1988e1989 (Krasner et al., 1989), 1997e98 (McGuire et al., 2002), 2000e2002 (Krasner et al., 2006; Weinberg et al., 2002) and 2006e2007 (Krasner et al., 2007; Mitch et al., 2009). The 2000e2002 study encompassed over 70 emerging DBPs and among the analysed N-DBPs were HANs, halonitromethanes (HNMs) and haloacetamides (HAcAms). There is less information available on the occurrence of N-DBPs in other countries, although relevant surveys have been carried out in Canada (Williams et al., 1995, 1997), Australia (Simpson and Hayes, 1998) and Scotland (Goslan et al., 2009). Another important N-DBP is N-nitrosodimethylamine (NDMA), formerly used in the production of rocket fuel and other industrial processes. Initially detected in Canadian drinking water in the 1980s (Jobb et al., 1994) NDMA has since been reported as a DBP produced from reactions
4345 4346 4346 4346 4348 4349 4349 4349 4350 4350 4350 4351 4351 4351 4352 4352
between monochloramine (NH2Cl) and dimethylamine (DMA) (Choi and Valentine, 2002). Several factors have seen a particular recent focus on NDBPs. Firstly, many N-DBPs are of greater perceived health risk than regulated DBP species. Comparison of data from in vitro mammalian cell tests demonstrated the HANs, HNMs and HAcAms are all far more cytotoxic and genotoxic than the non-nitrogenous THMs and HAAs, although the haloacetaldehydes also exhibit very high cytotoxicity and genotoxicity (Plewa and Wagner, 2009). Moreover, the nitrosamines may play a significant role in human carcinogenesis (Loeppky and Michejda, 1994) and the United States Office of Environmental Health Hazard Assessment (OEHHA) have issued a public health goal of 3 ng L1 for NDMA (OEHHA, 2006). At present, however, no N-DBPs are formally regulated by large governmental bodies anywhere in the world (Box 1). Secondly, water utilities are increasingly switching from chlorination to alternative disinfectants, particularly chloramines, in order to limit the formation of regulated THMs and HAAs (Seidel et al., 2005). Disinfectant type is a key factor in NDBP formation since, depending on the compound and reaction conditions, the nitrogen can derive either from the organic precursors, i.e. dissolved organic nitrogen (DON), or in the case of chloramination, from the disinfectant. Finally, the impact of human activity upon drinking water sources is increasingly being felt in the form of wastewater effluent and algal activity (Mitch et al., 2009). Since these are both enriched in DON, their presence is likely to lead to raised concentrations of many N-DBPs. Of the components of DON, amino acids are known to act as precursors of HANs, HAcAms and cyanogen halides (CNX) (Hirose et al., 1988; Ram, 1985; Reckhow et al., 2001; Trehy et al., 1986), while amine precursors of NDMA are believed to be largely anthropogenic in origin (Sacher et al., 2008; Schreiber and Mitch, 2006b), in contrast to the THMs and HAAs, where NOM, typically of terrestrial origin, is the main precursor pool. Hence, understanding and controlling the incidence of NDBPs is a contemporary challenge to the water industry. The objectives of this review are to highlight typical concentrations of identified N-DBPs drinking water, investigate
4343
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 4 1 e4 3 5 4
Box 1 Acronyms used. AOM CNBr CNCl CNMs CNX CNX DBPs BCAN DBAN DHAN DCAA DBAcAm DCAN DEA DHAN DHNMFP DHNMs DMA DON DPA EEM EfOM FP HAA9 HAAs HAcAms HAN4
Algal organic matter Cyanogen bromide Cyanogen chloride Chloronitromethanes Cyanogen halides CNX formation potential Disinfection by-products Bromochloroacetonitrile Dibromoacetonitrile Dihaloacetonitrile Dichloroacetic acid Dibromoacetamide Dichloroacetonitrile Diethylamine Dihaloacetonitrile Dihalonitromethane formation potential Dihalonitromethanes Dimethylamine Dissolved organic nitrogen Dipropylamine Excitation-emission matrix Effluent organic matter Formation potential Sum of nine surveyed HAAs Haloacetic acids Haloacetamides Sum of four HANs (DCAN, BCAN, DBAN and TCAN)
potential relationships between N-DBPs and other DBPs based on published data and to examine strategies for their mitigation. (Table 1).
2.
Occurrence of N-DBPs in drinking water
2.1.
Interpreting N-DBP occurrence data
A number of caveats should be kept in mind by those reviewing and comparing available data on N-DBPs. Firstly, important differences in disinfection practice between countries impact upon exposure of precursor material to disinfectants. Many US WTPs operate pre-chlorination or pre-chloramination (i.e. before or intermediate to other treatment) as well as post-disinfection, whereas in Europe post-chlorination/chloramination alone is typical. Furthermore, in the US 2000e2002 survey many selected WTPs had high bromide (median level 120 mg L1) and total organic carbon (median 5.8 mg L1) levels (Krasner et al., 2006) and were thus thought likely to generate relatively high DBP loads. A confounding factor when comparing relative effects of disinfectants on N-DBP formation is that it is often WTPs treating high bromide and/or organic carbon waters which switch from chlorination to chloramination in an attempt to lower formation of regulated DBPs. In the 2006e2007 US NDBP survey most WTP intake waters were impacted by upstream algal blooms and/or the discharge of treated wastewater (Mitch et al., 2009), where it can be expected that
HANs HNMs ICR MOR ND NDBA NDMA N-DBPs NDPA NEMA NMOR NOM NPOC NPYR NR OEHHA PYR TCAN TCNM THM4 THMs TMA TOC TOX UDMH US WTP WWTP
Haloacetonitriles Halonitromethanes Information collection rule Morpholine Not detected N-nitrosodibutylamine N-nitrosodimethylamine Nitrogenous disinfection by-products N-nitrosdipropylamine N-nitrosoethylmethylamine N-nitrosomorpholine Natural organic matter Non-purgable organic carbon N-nitrosopyrrolidine Not reported Office of Environmental Health Hazard Assessment Pyrrolidine Trichloroacetonitrile Trichloronitromethane (chloropicrin) Sum of four regulated THMs Trihalomethanes Trimethylamine Total organic carbon Total organic halogen Unsymmetrical 1,1-dimethylhydrazine United States Water treatment plant Wastewater treatment plant
N-DBP formation is above that found in more pristine water sources. Moreover, some N-DBPs have been identified in atypical waters: in the study where 2,3,5-tribromopyrrole was first detected (along with 3-bromopropanenitrile) bromide concentrations were 2 mg L1 (Richardson et al., 2003) and/or using sample concentration techniques followed by qualitative gas chromatography (GC) methods without commercially-available standards (Richardson et al., 1999, 2003). Meanwhile, disparate laboratory disinfection protocols are used for measuring DBPs formed from model compounds or isolates of NOM. To illustrate, in two studies testing the formation potential of NOM isolates the pH, chloramine dose and contact time were respectively 7.0, 45 mg of Cl2 per mg DOC (pre-formed monochloramine added) and 10 days (Lee et al., 2007) and 8.0, 3.0 mg Cl2 per mg DOC (ammonia added before chlorine at 1:3 weight ratio) and 3 days (Dotson et al., 2009).
2.2.
Haloacetonitriles (HANs)
Of the three major N-DBP groups captured by existing analytical methodologies e HANs, HAcAms and HNMs - in the 2000e2002 US survey, HANs occurred at the highest concentrations, with median and maximum levels of 3 and 14 mg L-1 respectively, and dichloroacetonitrile (DCAN) was the most prevalent species (Table 2) (Krasner et al., 2006; Weinberg et al., 2002). In the 2006e2007 US survey, median values for the sum of DCAN, bromochloroacetonitrile (BCAN), dibromoacetonitrile (DBAN) and trichloroacetonitrile (TCAN)
4344
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 4 1 e4 3 5 4
Table 1 e Important nitrogenous disinfection by-products (N-DBPs). Group/formula
Structure
Important species
R Haloacetonitriles (HANs) R3CCN
R
N R
Cyanogen halides (CNX) RCN
R NH2
R
R
N R
Halonitromethanes (HNMs) R3CNO2
N+
R
Dichloroacetamide (DCAcAm) (right), dibromoacetamide (DBAcAm), trichloroacetamide (TCAcAm)
Cyanogen chloride (CNCl) (right), cyanogen bromide (CNBr)
N Cl
N
Cl
O
Cl
NH2
H
Cl
O
N
O
N Cl
Trichloronitromethane (chloropicrin) (right), tribromonitromethane (bromopicrin), bromodichloronitromethane, dibromochloronitromethane
R
R
H
O-
R
Nitrosamines R2NNO
Cl
O
R Haloacetamides (HAcAms) R3CCONH2
Dichloroacetonitrile (DCAN) (right), bromochloroacetonitrile (BCAN), dibromoacetonitrile (DBAN), trichloroacetonitrile (TCAN) and tribromoacetonitrile (TBAN)
Structure
N-nitrosodimethylamine (NDMA) (right), N-nitrosopyrrolidine (NPYR), N-nitrosomorpholine (NMOR), N-nitrosodiethylamine (NDEA)
ON+
Cl
O
Cl
H3 C N
N
O
H3 C
R is typically Cl, Br, I, H or an alkyl group, though it can also be a larger aliphatic or aromatic group.
(collectively HAN4) were slightly higher at 4.0 mg L1 (Krasner et al., 2007), presumably a reflection of the mainly algal and wastewater-impacted waters chosen. In Australia, HAN4 levels up to 36 mg L1 have been observed, something most likely related to high organic content and bromide, which also resulted in THM levels up to 191 mg L1 (Simpson and Hayes, 1998). In contrast, HANs were lower in Scotland than the US, with median and maximum HAN4 concentrations of 1 mg L1 and 4 mg L1 respectively (Goslan et al., 2009). Median non-purgeable organic carbon (NPOC) and bromide levels were 3.6 mg L1 and 55 mg L1 respectively, in the Scottish waters, versus equivalent figures of 5.8 mg L1 and 120 mg L1 in the US survey (Goslan et al., 2009; Krasner et al., 2006). It has been proposed that the mass of HANs typically represents around 10% of the THMs (Krasner et al., 1989; Oliver, 1983). To investigate such rules the extensive DBP data from the US 2000e2002 survey was collated with that from the 2006e2007 US survey and Scotland and relevant ratios and correlations between the DBP groups calculated (Tables 3 and 4). At the 12 WTPs in the US 2000e2002 survey HAN4 as a proportion of the four regulated THMs (THM4) varied from 2% to 14%, with a median value of 8%, while median ratios for HAN4 and DHAN (HAN4 without TCAN) were respectively 2% and 7% in the Scottish and US 2006e2007 surveys (Table 3, Scottish ratios computed from median
values across the whole of the survey). Thus, the 10% value is an approximate guide to HAN formation. A good positive correlation (r ¼ 0.90) has previously been observed between HAN and THM formation (Krasner et al., 1989). For this review correlations between DHAN and THM4 was calculated as 0.78, with a correlation of 0.83 between HAN4 and the nine surveyed HAAs (HAA9) (Table 4, n ¼ 15). This indicates HAAs may be at least as good a predictor of HAN formation as THMs. As DCAN hydrolyses to 2,2-dichloroacetamide (DCAcAm) and consequently dichloroacetic acid (DCAA) in the presence of free chlorine or at alkaline pH (Reckhow et al., 2001) this is perhaps unsurprising. In the US 2000e2002 survey and Scotland median values of HAN4/DHAN in finished water accounted for 7% of HAA9 formation (Table 3).
2.3.
Haloacetamides (HAcAms)
HAcAms were reported for the first time during the 2000e2002 US survey, DCAcAm being the most prominent species, with a median concentration of 1.3 mg L1 (Table 2). The median and maximum concentrations of the sum of HAcAms were 1.4 mg L1 and 7.4 mg L1 respectively, though note that not all the possible brominated and chlorinated HAcAms were quantified. HAcAms were frequently identified in finished water from three sites where chlorine dioxide was applied prior to chlorine/chloramine. Krasner and co-workers noted
4345
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 4 1 e4 3 5 4
Table 2 e Summary of N-DBP occurrence in effluents of selected US WTPs (Krasner et al., 2006; Weinberg et al., 2002). Occurrence (mg L1)
Name
WTP of max occurrence
Median Maximum Haloacetonitriles Dichloroacetonitrile Bromochloroacetonitrile
Conditions of max occurrence TOC (mg L1)
Bromidea (mg L1)
pH
Disinfection
12.0 10 (opposite) 3.0e4.2 7.5e7.7 50e70 Cl2 - chloramines 3.0 2 (opposite), 1.6e3.0 7.9e8.7 120e400 Cl2 - chloramines 11 (opposite) and 12 (below) 2.9e4.2 7.4e7.5 160e210 ClO2eCl2 - chloramines Other HANs recorded: dibromoacetonitrile, trichloroacetonitrile, dibromochloroacetonitrile, chloroacetonitrile, bromoacetonitrile Sum of haloacetonitriles 3 14 10 (as above) Haloacetamides 2,2-dibromoacetamide 0.6 2.8 12 (opposite) 3.2e4.5 7.6e8.3 150e330 ClO2 - chloramines 2,2-dichloroacetamide 1.3 5.6 6 (opposite), 12 (as above) 3.5e4.5 5.8e7.0 39e80 ClO2eCl2 - chloramines Other HAcAms recorded: 2-bromoacetamide, 2,2,2-trichloroacetamide, 2-chloroacetamide Sum of haloacetamides 1.4 7.4 6 (as above) Halonitromethanes Chloropicrin 0.2 2.0 10 (as above) Bromodichloronitromethane 0.3 3.0 7 3.0e4.2 7.5e7.7 50e70 Cl2 - chloramines Dibromochloronitromethane ND 3.0 12 (as above) Bromopicrin ND 5.0 12 (as above) Other HNMs recorded: bromochloronitromethane, chloronitromethane, bromonitromethane, dichloronitromethane, dibromonitromethane Sum of halonitromethanes 1 10 12 (as above) 1 0.6
a ¼ Raw water bromide concentration.
that DCAcAm occurred at a similar level to DCAN (respective median values 1.3 mg L1 and 1.0 mg L1) and that HAcAm formation was w10% of HAAs, with dichloro representatives of the two groups found at higher levels than the trichloro species (Krasner et al., 2006).
2.4.
Cyanogen halides (CNX)
In the 1988e1989 US survey an association was noted between chloramination and CNCl formation, median values of CNCl in treatment works with free chlorine and chloramination were 0.4 mg L1 and 2.2 mg L1 respectively (Krasner et al., 1989). This finding has been re-confirmed by subsequent research (see Section 4.2.3). During the 2006e2007 NDBP survey CNX (i.e. CNCl plus CNBr) formation was generally
only observed at the plants with chloramination, in which the median and maximum formation was 2.6 mg L1 and 7.8 mg L1 respectively (Mitch et al., 2009). Nonetheless, CNX precursors were widely present, as shown by plant influent samples disinfected under laboratory conditions designed to maximise CNX (3 h pre-chlorination then chloramination for 21 h), which generated respective median and maximum levels of 12 and 34 mg L1 (Mitch et al., 2009). A low formation of CNX was found in plants where ozone was applied prior to biofiltration and chlorination/chloramination, suggesting biological treatment effectively removed formaldehyde and other CNX precursors resulting from ozonation (Krasner et al., 2007). In Australia CNCl has been recorded up to a level of 10 mg L1 from a WTP practicing monochloramination (Simpson and Hayes, 1998).
Table 3 e Ratios (mg/mg) between N-DBPs and other DBP groups in finished water samples from US and Scotland. Survey
US 2000e2002 Weinberg et al., 2002
US 2006e2007 a
Mitch et al., 2009
Ratios
Min
Median
Max
HAN4/THM4 DHAN/THM4 HAN4/HAA9 HAN4/DXAA TCNM/THM4 Sum of HNMs/THM4 Sum of HNMs/HAA9
0.02 0.02 0.02 0.02 0 0 0
0.08 0.08 0.07 0.13 0.00 0.03 0.03
0.14 0.14 0.12 0.2 0.01 0.23 0.19
Scotland b
Goslan et al., 2009 c
25th %ile
Median
90th %ile
0.07
0.07
0.16
Median 0.02 0.07
0 0
0.01 0.01
0.03 0.04
0.00
a ¼ Ratios calculated from mean values of 4e5 seasonal samples taken at each of 12 WTPs. Not reported taken as half the minimum reporting level; not detected taken as zero. Between four and nine HNMs quantified, depending on the sample (see Table 2). b ¼ Five HNMs quantified. c ¼ Ratios computed from median data across whole of survey. THM4 ¼ sum of four regulated THMs; HAA9 ¼ sum of nine surveyed HAAs. DHANs ¼ HAN4 e TCAN.
4346
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 4 1 e4 3 5 4
Table 4 e Correlationsa between DBP groups in USb and Scottishb DBP surveys.
HAA9 (n ¼ 16) DXAA (n ¼ 12) TXAA (n ¼ 12) HAN4 (n ¼ 16) DHAN (n ¼ 15) Sum of HNMs (n ¼ 15) TCNM (n ¼ 19)
THM4 (n ¼ 19)
HAA9
DXAA
TXAA
HAN4
DHANs
HNMs
0.81 0.76 0.80 0.30 0.78 0.02 0.09
0.84 0.89 0.71 0.83 0.14 0.54
0.42 0.59 0.61 0.56 0.44
0.83 0.79 0.21 0.62
0.99 0.06 0.42
0.18 0.67
0.30
a Correlations ¼ Pearson product moment correlation coefficients (r). b Bulk of data (n ¼ 12 for all analytes) from US 2000e2002 survey (mean values of 4e5 seasonal samples taken at each of 12 WTPs. Between four and nine HNMs quantified). Remainder (n ¼ 3) from US 2006e2007 survey (THM4, DHAN, sum of (five) HNMs and TCNM) and (n ¼ 4) from Scottish survey (THM4, HAA9 and TCNM). DHANs ¼ HAN4 e TCAN. TCNM ¼ trichloronitromethane.
2.5.
Halonitromethanes (HNMs)
For the sum of HNMs, median and maximum levels of 1 and 10 mg L1 were recorded during the 2000e2002 US survey. Bromopicrin and dibromochloronitromethane maxima were 5.0 and 3.0 mg L1, respectively (Table 2) at a site characterised by high bromide (150e330 mg L1), pre-oxidation with chlorine dioxide and post-chloramination. The sum of analysed HNMs represented respectively 3% and 1% of THM4 on a median basis in the US in 2000e2002 and 2006e2007(Table 3), although a maximum 23% of THM4 formation was recorded at one location where THM4 was relatively low (mean ¼ 8.5 mg L1). As intimated by this, HNM formation does not appear related to the THMs or HAAs, with no meaningful correlations calculated between the sum of HNMs and the two regulated groups (Table 4, n ¼ 15). The formation of chloropicrin (trichloronitromethane (TCNM)) in the 2006e2007 N-DBP study was higher, with median and maximum values of 0.5 and 7.6 mg/L, respectively (Krasner et al., 2007), again highlighting the importance of wastewater and algae as precursor sources. Ozonation before chlorination can dramatically enhance HNM formation (Hoigne and Bader, 1988) (see Section 4.2.2).
2.6.
Nitrosamines
NDMA is typically observed in the low ng L1 range in drinking water, although concentrations equal to or above 1000 ng L1 have been recorded in chloraminated raw water (Sacher et al., 2008) and chlorinated or chloraminated wastewater effluent (Krasner et al., 2009a; Mitch et al., 2003). Analysis of a Canadian drinking water supply in 1986 detected NDMA at concentrations between 5 and 115 ng L1 (Jobb et al., 1994), which stimulated a survey of 145 WTPs in Ontario, Canada. Finished water was under 5 ng L1 in the majority of cases. These and other studies highlighted associations between elevated NDMA occurrence and municipal and industrial wastewater input, chloramination, cationic polymers and ion exchange resins (Najm and Trussell, 2001). Various synthetic chemicals containing a DMA moiety have subsequently been identified as NDMA precursors. These include the pharmaceutical ranitidine (Sacher et al., 2008) and diuron, a herbicide (Chen and Young, 2008). Charrois et al. (2004) developed an ammonia positive chemical ionisation method which enabled detection of two
additional nitrosamines in drinking water: N-nitrosopyrrolidine (NPYR) and N-nitrosomorpholine (NMOR), at 2e4 ng L1 and 1 ng L-1 respectively, in addition to NDMA at 2e180 ng L1. Also evident were increased levels of NDMA in the distribution system (180 ng L1) relative to treated effluent (67 ng L1) of a plant using chloramination and UV disinfection. A recent study compared the formation of eight nitrosamines in finished water samples from six utilities using various treatments with raw water samples from the same sources chloraminated under laboratory conditions (Sacher et al., 2008) (Table 5). In contrast to the treated water samples, where NDMA peaked at 4.9 ng L1 and no other nitrosamines were reported, the laboratory disinfected samples generated NDMA up to 110 ng L1 and NPYR and Nnitrosoethylmethylamine (NEMA) at maxima of 7.6 and 3.4 ng L1, respectively, indicating precursors of these species were present in raw waters. Chloramination of 81 river and lake samples revealed a similar pattern: NDMA was always the dominant species, with median and maximum levels of 45 and 1000 ng L1 respectively, while other nitrosamines were periodically present, albeit always at least an order of magnitude lower than NDMA. NPYR and N-nitrosodiethylamine (NDEA) were the second and third most frequently recorded nitrosamines and reached respective peaks of 35 and 23 ng L1 (Table 5). Another nitrosamine, N-nitrosodibutylamine (NDBA), has been detected in one UK water distribution system at 6.4 ng L1 (Templeton and Chen, 2010).
2.7.
Other N-DBPs
Intermediates in the reactions schemes portrayed (Figs. 1e3) are either known or presumed to occur in drinking water. In particular, hydrazine is carcinogenic and has been detected at 0.5e2.6 ng L1 in chloraminated drinking water, though was not detected in chlorinated samples (Davis and Li, 2008). Other organic hydrazines presumably form during the unsymmetrical hydrazine pathway of nitrosamine formation (displayed for NDMA in Fig. 3) (Choi and Valentine, 2002). Organic chloramines form from chlorination or chloramination of DON, for example, amino acid chlorination (Fig. 1) and can lead to a w10% overestimation of disinfection capacity in chloraminated water systems (Lee and Westerhoff, 2009). There are additional N-DBPs for which very limited occurrence data exists. Benzeneacetonitrile, heptanenitrile and cyanoformaldehyde have been detected as ozone
4347
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 4 1 e4 3 5 4
Table 5 e Summary of nitrosamine levels in drinking water sources (Sacher et al., 2008) and of DBPs in wastewater (Krasner et al., 2009a and 2009b). Survey
Range of NDMA levels (ng/L)
Reference: Sacher et al., 2008 81 surface water samples (rivers and lakes) chloraminated under laboratory conditions (0.4 mM, 7 days). Eight nitrosamines. Original water samples from 6 water utilities. Samples from same 6 water utilities chloraminated under laboratory conditions (0.4 mM, 7 days). References: Krasner et al., 2009a and 2009b 23 US WWTPs at different seasons. Before chlorination/chloramination
Median (ng/L)
1e1000
45
<1e4.9
NR
8e110
NPYR: <1e7.6
ND e 34
2.7
8.5
ND e 8.2 ND e 3165
3 11
24
Range of HANs (mg/L)
Post-chlorination (Cl2/N > 10 mg/mg) Post-chloramination (Cl2/N < 10 mg/mg)
Median (mg/L)
0.9e30 ND e 12
COO-
R
NHCl
COO-
- CO2 + Cl-
NCl2
2 HOCl
-
HOCl
O
O
+
ND e 0.7 ND e 0.6
0.8
O
H 2O NH
R
R H Aldehyde
NCl
H
H H
NHCl - CO2 + Cl-
R C N
-HCl
NH
NH3
H2O
Nitrile
O NH3 H H Formaldehyde
NH3
O glycine
H R
- CO2 + Cl-
O -
Chloropicrin (mg/L)
case larger (and unidentified) nitriles and halonitroalkanes are perhaps expected to form during water treatment (Mitch et al., 2009). Since, on a median basis, halogenated DBPs quantified in the 2000e2002 US survey accounted for only around 30% of total organic halogen (TOX) formed after disinfection (Krasner et al., 2006), many unidentified DBPs occur in drinking water, some of which may contain nitrogen. The recent application of a total N-nitrosamine (TONO) assay to six recreational waters (swimming pool or hot tub) and their common tap water source produced a similar conclusion (Kulshrestha et al., 2010). This method established that NDMA accounted for only a mean of 13% (range 3e46%) of total nitrosamines present. Extrapolating the results of this study suggests that unidentified nitrosamines also occur in potable water.
COO-
R
NMOR: ND e 12700 (median ¼ 5.5; 75th %ile ¼ 17) NMOR: ND e w8 (median ¼ 5.6) NMOR: ND e 911. (median ¼ 3.3)
75th percentile (mg/L)
16 0.3
disinfection by-products (Richardson et al., 1999). However, cyanoformaldehyde was not detected in any of the treatment plants in the 2000e2002 US occurrence study. Other N-DBPs were recorded in Israeli water with very high bromide (2 mg L1) and chlorine dioxide disinfection: 2,3,5tribromopyrrole and 3-bromopropanenitrile (Richardson et al., 2003). The first of these compounds represented the first incidence of a halogenated pyrrole as a DBP. There are also various other nitrogenous compounds as yet unidentified in drinking water but which have been highlighted for future research on the basis of chemical and toxicological models (Bull et al., 2006). It has been suggested that degradation of larger compounds to one or two carbon amine precursors may be rate-limiting in the formation of many N-DBPs, in which
NH2 Amino acid
NDEA: <1e23
90
NEMA: <1e3.4
Other N-DBPs
R
Other nitrosamines (ng/L)
NPYR: <1e35
Post-chlorination (Cl2/N > 10 mg/mg) Post-chloramination (Cl2/N < 10 mg/mg)
HOCl
75th percentile (ng/L)
-
2 HOCl
O
H
NCl2 O
- CO2 + Cl-
H
HOCl NCl
H C N
Cl C N
-HCl Cyanogen chloride
Fig. 1 e Chlorination of amino acids (top), with specific reference to glycine (below). Based on Deborde and von Gunten (2008) and Joo and Mitch (2007).
4348
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 4 1 e4 3 5 4
O
OH NH2Cl
H3 C
H
-HCl
-H2O
H3C
NHCl
H 3C
H3 C
2HOCl N
NCl
Cl2HC N
-H2O DCAN
Acetaldehyde O
OH NH2Cl
H
H
-HCl
-H2O H
NHCl
H
H N
NCl
Cl
HOCl -H2O
N
Cyanogen chloride
Formaldehyde
Fig. 2 e The aldehyde pathway, which converts aldehydes to the equivalent nitrile and subsequently halonitrile (Mitch et al., 2009).
3. Precursor sources and formation pathways DON species such as amino acids, proteins, amino sugars, amides, nitriles, pyrroles, purines and pyrimidines are widespread in surface water (Westerhoff and Mash, 2002). They typically comprise a small portion of NOM by weight (0.5e10%) and derive from the activity of microorganisms, leaching from soil, or the influence of wastewater discharge (Westerhoff and Mash, 2002). The hydrophilic neutral, hydrophilic base and colloidal fractions of NOM are especially rich in nitrogenous matter (Leenheer et al., 2007). Selected formation pathways which account for the formation of aldehydes and nitriles from amino acid chlorination; formaldehyde and CNCl from glycine chlorination (both Fig. 1 (Deborde and von Gunten, 2008; Joo and Mitch, 2007)); DCAN and CNCl from monochloramination of respectively acetaldehyde and formaldehyde (Fig. 2 (Mitch et al., 2009)) and NDMA from DMA (Fig. 3 (Choi and Valentine, 2002; Keefer and Roller, 1973; Schreiber and Mitch, 2006a)) are portrayed. Note that HANs and CNCl can result from either chlorination of nitrogenous precursors
H3 C NH H3 C
NH2Cl
Dimethylamine (DMA)
H3 C NH H3 C
NHCl2
N
H3 C
H N
Oxidation
H3 C
N
N Cl
Chlorinated UDMH
Dimethylamine (DMA)
N
O
H3 C N-nitrosdimethylamine
H3 C
NH H3 C
O
N-nitrosdimethylamine
Nitrosating agent, e.g. NO+, HNO2, N2O4
H3 C
N
H3 C
H3 C H3 C
H3 C
NH2
UDMH
-HCl
Dimethylamine (DMA)
Oxidation
H3 C N
-HCl
(i.e. amino acids) or monochloramination of non-nitrogenous precursors (i.e. aldehydes). Wastewater and algal activity are linked to increased DON levels in drinking water (Krasner et al., 2008). To illustrate this, the average DON level in 28 US WTPs was 186 mg-N L1 (Lee et al., 2006), whereas the equivalent value in 16 WTPs subject to algal or wastewater influence was 290 mg-N L1 (Dotson and Westerhoff, 2009). Meanwhile water sources affected by algal blooms can have DON levels around 1 mg L1 as N, with the median concentration in effluents of wastewater treatment plants (WWTPs) from 1 to 4 mg L-1 as N (Pocernich and Litke, 1997). Because the formation of N-DBPs, including HANs and HNMs, was highest from nitrogen-rich fractions of dissolved organic matter, increased effluent organic matter (EfOM) and algal organic matter (AOM) in drinking water supplies will generally increase N-DBP formation (Dotson et al., 2009). Several recent DBP studies have focused on EfOM as a source of both DBPs and their precursors. During a campaign sampling 23 WWTPs at different seasons, both NDMA and NMOR were present before disinfection, with NMOR quantified up to 12,700 ng L1. However, these data probably relate to
N
N
O
H3 C N-nitrosdimethylamine
Fig. 3 e NDMA formation pathways, involving UDMH (top) (Choi and Valentine, 2002); chlorinated UDMH (middle) (Schreiber and Mitch, 2006a,b) and nitrosation (bottom) (Keefer and Roller, 1973).
4349
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 4 1 e4 3 5 4
contamination from an industrial or other anthropogenic process feeding the specific WWTPs. After postchloramination at the WWTPs NDMA median and maximum values were 11 and 3165 ng L1 respectively, and thus higher than equivalent concentrations post-chlorination (3 and 8.2 ng L1) (Krasner et al., 2009a, 2009b) (Table 5). In contrast, NMOR concentrations were generally lower postdisinfection. HANs were also higher in WWTP effluents with a chlorination stage than WTPs (e.g. Table 2), with respective median and maximum values 16 and 30 mg L1 postchlorination (Table 5). Conversely, chloropicrin only reached respective maxima of 0.7 and 0.6 mg L1 post-chlorination and post-chloramination (Table 5), lower than the maximum of 2 mg L1 in the nationwide WTP study (Table 2) and indicating treated wastewater was not a major source of chloropicrin precursors. Another factor impacting N-DBPs is the level of treatment in WWTPs: in those with nitrification lower amounts of NDMA and HAN precursors were present. Thus, median yields of HANs from samples classified as having treatment with no nitrification and good nitrification were 29 and 13 mg L1 respectively, and the corresponding data for NDMA were 878 and 440 ng L1 (Krasner et al., 2008, 2009b). These values were obtained via laboratory formation potential tests undertaken with chlorine for HANs and monochloramine for NDMA. Precursor concentrations are also liable to change after discharge into the environment. Sampling of an EfOM dominated Arizonan river fed by a WWTP featuring chlorination/dechlorination found the EfOM was biodegradable by up to 40% (over 38 km of river; estimated travel time 4.2 days) (Chen et al., 2009). Removal of HAN and nitrosamine precursors, as monitored by formation potential tests, correlated with that of EfOM. For chloropicrin precursors the situation was more complex, with peaks at intermediate sampling points tentatively related to variation in nitrite, a possible nitrogen source in chloropicrin (Choi and Richardson, 2004).
4.
Key factors in the formation of N-DBPs
4.1.
Water quality parameters
4.1.1.
The impact of pH
As a generalisation, there is a trade-off between operating disinfection at low pH to minimise THMs and using a higher
pH which disfavours other DBP groups (Stevens et al., 1989). As seen below, the TNMs are an exception to this rule (Joo and Mitch, 2007; Merlet et al., 1985) (Table 6), and DCAA is rather insensitive to pH change (Stevens et al., 1989). During a pilotplant investigation DCAN was stable only at pH 5, where its concentration increased over time. At pH 7 DCAN decreased over time and at pH 9.4 it barely formed at all (Stevens et al., 1989). Amongst the HANs, trihaloacetonitriles (THANs) had the highest rates of hydrolysis, followed by DHANs (Glezer et al., 1999; Oliver, 1983). Reaction rates with chlorine follow a similar trend, while all degradation rates increased with increasing pH (Reckhow et al., 2001). It has been noted that after chlorination of aspartic acid for 3 h at pH 5, 7 and 9, the DCAN yield was highest at pH 5 (Chu et al., 2010). However, DCAcAm formation was negligible at pH 5, 0.2% at pH 7 and 0.49% at pH 9 (Chu et al., 2010), hinting that DCAN hydrolysis may not be the only formation pathway. At the two sites of maximum DCAcAm, dibromoacetamide (DBAcAm) and total HAcAm occurrence (Table 2) in the US 2000e2002 survey, the pH ranged from 6.8 to 7.1 and 7.6e8.3 respectively. In chlorinated waters, decreased CNX presence with increasing pH values was attributed to its base-catalysed decomposition (Heller-Grossman et al., 1999), and the shift in equilibrium distribution of free chlorine is another factor (Xie and Reckhow, 1992). The hypochlorite ion (OCl), rather than hypochlorous acid (HOCl), is the reactive compound responsible for the decomposition of CNCl, and more OCl exists at pH 6 than at pH 5. This instability in the presence of free chlorine may explain more frequent appearance of CNCl in chloraminated waters. Fewer studies have examined pH-mediated effects on NDBPs in chloraminated waters. The net formation of DCAN and CNCl was only slightly reduced over a pH range of 7.5e9 (Pedersen et al., 1999). The highest formation of CNCl from formaldehyde was observed at pH 5 and the formation of HNMs upon chloramination was less influenced by pH as compared with chlorination (Joo and Mitch, 2007). Regarding NDMA, optimum conditions for its formation from UDMH occur at pH 7e8 (Mitch and Sedlak, 2002) (Table 6). Nitrosation of DMA by N2O3 is controlled by the formation of N2O3 and therefore nitrosation of DMA was most rapid at pH 3.4 and rather sluggish at neutral and basic pH (Mirvish, 1975). By a similar logic, free chlorine enhanced nitrosation was most rapid at neutral pH, when N2O4 formation was favoured (Choi and Valentine, 2003). Further, due to the postulated importance of dichloramine to NDMA formation, the pH
Table 6 e Effect of pH on occurrence of major N-DBP groups. Group Haloacetonitriles Haloacetamides Halonitromethanes Cyanogen halides Nitrosamines
pH effect
References
More stable at acidic pH, hydrolysed at alkaline pH. Uncertain but presumably hydrolysed at alkaline pH. See text. Chloropicrin formation increases with pH. Higher formation at acidic and neutral pH, unstable in presence of free chlorine. NDMA formation via UDMH peaks at pH 7e8. Chlorine enhanced nitrosation most rapid at neutral pH. Nitrosation itself increases with pH but normally limited by formation of nitrosating agent.
Glezer et al., 1999; Oliver, 1983; Stevens et al., 1989 Chu et al., 2010; Reckhow et al., 2001 Joo and Mitch, 2007; Merlet et al., 1985 Joo and Mitch, 2007; Mitch et al., 2009 Sacher et al., 2008; Mirvish, 1975; Choi and Valentine, 2003; Mitch and Sedlak, 2002
4350
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 4 1 e4 3 5 4
dependence of chloramines speciation is another factor to consider. Alkaline conditions, especially above pH 8.5, limit dichloramine formation and this can lead to reduced NDMA formation (Schreiber and Mitch, 2006a).
4.1.2.
The impact of bromide and iodide
Most brominated and especially iodinated DBPs are more cytotoxic and genotoxic than their chlorinated analogues (Plewa and Wagner, 2009). Bromine species (HOBr/OBr-) are known to be more effective substitution agents than the equivalent chlorine species (Symons et al., 1993) and molar DHAN formation was found to increase up to 101% for a largely autochthonous water source and 73% for a heavily allochthonous water source after spiking each with bromide (up to 30 mM) (Hua and Reckhow, 2007). Increasing bromide concentrations shifted the distribution of DHANs from DCAN to BCAN and then to DBAN. Elevated levels of bromide can also raise the CNX yield on a molar basis, although HOBr can catalyse cyanogen bromide (CNBr) hydrolysis, analogous to chlorine enhancing CNCl degradation, resulting in a lower net concentration of CNBr (Heller-Grossman et al., 1999). Bromamines have been suggested to play an important role in DBP formation (Diehl et al., 2000), with monobromamine and dibromamine formed during ozonation found to outcompete chloramines in forming CNBr (Lei et al., 2006). There is scant information about the presence of iodinated NDBP compounds in drinking water, although at least several potentially occur. The compound, 2-2-bromoiodoacetamide has been detected in chloraminated drinking water (Plewa et al., 2008; Richardson et al., 2007). Since many nonnitrogenous iodinated DBPs have also been most frequently detected following chloramination it can be predicted that the same will be found to apply for iodinated N-DBPs.
4.2.
The impact of treatment and disinfection
4.2.1.
The impact of treatment
Nitrogen-rich colloidal, hydrophilic neutral and hydrophilic base fractions tend to dominate N-DBP formation (Dotson et al., 2009). As THM and N-DBP precursors have different physicochemical properties, treatments effective at removing the former may have less success with N-DBP precursors. In particular, the hydrophobic acid fraction is typically a major source for THM precursors and is susceptible to removal by coagulation. While coagulation is also effective for removing colloids (a fraction holding a significant portion of N-DBP precursors) the same does not apply to the similarly reactive
hydrophilic base and neutral fractions. Recently, DON removal by coagulation was reported to be 21% on a median basis by WTPs which had suspended chlorination and/or chloramination, compared with 37% for bulk DOC (Mitch et al., 2009) (Table 7). Moderate removals of N-DBP precursors have been observed by coagulation, ranging from 18% to 52% for DHAN precursors as assessed by chlorine and chloramine formation potential (FP) tests, respectively (Mitch et al., 2009) (Table 7). In the same study, the removal of precursors by filtration proved more variable, ranging from 3% for DHAN precursors (chlorine FP test) to 61% for chloropicrin precursors (chlorine FP test). Filtration did not significantly impact upon CNX or DHAN formation, except where biological filtration removed aldehydes identified as probable CNX precursors (Mitch et al., 2009). Where filters were believed to be biologically active formaldehyde was not detected and median acetaldehyde concentration was 0.6 mg L1, whereas equivalent median values in WTPs without biological filtration were 22 and 5 mg L1. However, the importance of free amino acids as HAN, HAcAm and CNX precursors, as repeatedly highlighted by model compound work, is not consistent with full-scale data. Mean values for total amino acids and amino sugars accounted for 15% (median 10%) of DON in algal and wastewater impacted plant influents, but only 5% (median 4%) of the DON in filter effluent samples (intermediate treatments included ozone, chlorine dioxide, chlorine, chloramines, lime softening and coagulation, depending on the plant) (Mitch et al., 2009). Thus amino acids showed preferential removal relative to that of DOC and DON. In both cases the proportion of free amino acids was low compared with total (combined) species. Hence, it is thought amino acids found in AOM and EfOM do not fully account for recorded N-DBP formation and consequently that unidentified precursors exist as part of DON (Dotson and Westerhoff, 2009). For comparison, although model compound studies have determined selected free amino acids, as well as the nitrosamine precursors dimethylamine, diethylamine, morpholine and piperidine, to be ineffectively removed by coagulation (Bond et al., 2010; Pietsch et al., 2001), aliphatic members of these groups are typically susceptible to biodegradation (Bond et al., 2009; Pietsch et al., 2001). Meanwhile, the 43% increase in NDMA after coagulation reported from WTPs which had suspended chlorination and/ or chloramination (Table 7) was thought to be linked to polymer use (Mitch et al., 2009). Overall, Table 7 shows the limited efficacy of the common water treatment processes (coagulation and filtration) in removing N-DBP precursors, therefore more advanced processes such as membranes and
Table 7 e Removal of N-DBP precursors and water quality parameters from WTPs that suspended use of chlorine and/or chloramines (Mitch et al., 2009). Median results reported for samples chlorinated and/or chloraminated before and after treatment under laboratory conditions in formation potential (FP) tests. Treatment
Coagulation Ozonation Filtration
DOC
37% 4% 12%
DON
21% 3% 18%
DHAN
DHAN
TCNM
TCNM
CNX
NDMA
(Cl2)
(NH2Cl)
(Cl2)
(NH2Cl)
(Cl2eNH2Cl)
(NH2Cl)
18% 20% 3%
52% 28% 24%
49% 226% 48%
44% 133% 61%
48% 4% 17%
43% 10% 52%
Negative value indicates increased formation. TCNM ¼ chloropicrin (trichloronitromethane).
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 4 1 e4 3 5 4
biodegradation are likely to be more appropriate for reducing concentrations of nitrogenous precursors.
4.2.2.
The impact of pre-oxidation
Ozone can have extreme effects on HNM formation, as demonstrated by an increased median chloropicrin occurrence of 226% and 133% when followed by chlorination and chloramination formation potential tests, respectively, (Table 7) (Mitch et al., 2009). Ozone can also increase CNX formation, probably by increasing levels of formaldehyde or related precursors. Formaldehyde is known to be produced during drinking water treatment by chlorination and ozonation at respective levels of 0.14 and 0.33 mM (Krasner et al., 1989; Richardson et al., 1999; Weinberg et al., 1993), which points to chlorination or ozonation followed by chloramination as a likely scenario in producing CNCl. Meanwhile, relatively high concentrations of HANs and CNX were formed when chlorine dioxide was used in combination with chlorine (Heller-Grossman et al., 1999; Richardson et al., 1994). UV irradiation can also influence subsequent N-DBP formation, depending on the lamp type. Chloropicrin increased from 0.6 mg L1 at 0 mJ cm2 to 1.8 mg L1 following 140 mJ cm2 of medium pressure irradiation, whereas low pressure UV had no impact (Reckhow et al., 2010), while its formation was nearly doubled when post-chlorination rather than post-chloramination was coupled with medium pressure UV (Shah et al., 2011). This enhancement has been linked to photolysis of nitrate below a wavelength of 250 nm, in which region low pressure UV does not emit significantly (Reckhow et al., 2010). It was postulated that nitrite radicals produced in this manner nitrated aromatic NOM structures. In addition, due to the photolability of NDMA, UV irradiation is an effective degradation method (Sharpless and Linden, 2003). Oxidation before chloramination has proved an effective means of controlling NDMA formation. Chlorine, ozone, permanganate and simulated sunlight were all capable of reducing NDMA formation relative to chloramination alone (Chen and Valentine, 2008), presumably due to alteration of precursor sites. For example, a 50% reduction was measured when chloramination of a concentrated river water sample was preceded by 10 min free chlorine contact time at 0.08 M. Ozonation was effective at degrading aliphatic and particularly alicyclic nitrosamine precursors, with half-lives of respectively 17, 58, 15 and 0.19 min (ozone dose 1.0 mg L1 at pH 7) for dimethylamine, diethylamine, piperidine and morpholine (Pietsch et al., 2001), although it only reduced NDMA formation by 10% on a median basis in surveyed WTPs (Table 7).
4.2.3.
The impact of chlorination and chloramination
Chloramination normally involves addition of both chlorine and ammonia under controlled conditions, with an average chlorine contact time of 26 min before ammonia addition noted in the US (AWWA, 2000). For the HANs, comparative formation potential tests with chlorine/chloramines on samples from surveyed WTPs showed a concentration ratio of 1.2/1.0 on a median basis (Mitch et al., 2009), using the method described for Dotson et al., 2009. This pattern is comparable to the Scottish survey, where median concentrations of HAN4 were 1.7 mg L1 in chlorinated waters and 1.3 mg L1 in chloraminated waters (Goslan et al., 2009).
4351
Dotson et al. (2009) isolated NOM fractions from nitrogenrich sources and tested their propensity to form N-DBPs. On a central tendency basis DCAN yields were approximately twice as high after chlorination than after chloramination, with highest yields from the most nitrogen-rich fraction (hydrophilic bases). In the study by Lee et al. (2007), DCAN formation was on average approximately five times higher after chloramination than chlorination of 17 fractions from various water sources. Since disinfection protocols varied widely between these studies (see Section 2.1) and HAN precursors can be either nitrogenous or non-nitrogenous (Figs. 1 and 2) contradictory literature results are perhaps not unexpected. Another relevant factor is that, in the presence of free chlorine, DCAN degrades to DCAA (Reckhow et al., 2001) while appearing to be more stable in the presence of monochloramine (Lee et al., 2007). Equal amounts of chloropicrin were found upon chlorination and chloramination (Richardson, 2008), while yields of chloropicrin were slightly higher from chlorination than chloramination of nitrogen-rich isolates (Dotson et al., 2009). Unlike free chlorine, chloramines do not catalyse CNX decay (Na and Olson, 2004), correlating with the noted trend for CNX to be more widespread in chloramination WTPs (see Section 2.4). By adding pre-formed chloramines to wastewater less NDMA resulted than from addition of chlorine and ammonia in situ, a result attributed to lower dichloramine levels (Schreiber and Mitch, 2005). Unequal distribution of chlorine and ammonia, especially around the point of chlorine addition, can lead to localised formation of dichloramine and this can also be alleviated by adding free chlorine before ammonia (Schreiber and Mitch, 2005). Enhanced dichloramine formation close to the breakpoint (at wCl:NH3 1.7) was thought to underlie NDMA formation an order of magnitude higher than during monochloramination (i.e. at lower Cl:NH3 ratios) (Schreiber and Mitch, 2007), while reactive nitrosating intermediates were thought to contribute to heightened NDMA formation when small amounts of free chlorine were present. Thus, to limit NDMA it was recommended that chloramination be operated away from the breakpoint and with no free chlorine residual (Schreiber and Mitch, 2007).
5. Conclusions: minimising N-DBPs in water treatment From the foregoing review of published research, several conclusions can be made regarding methods of minimising NDBPs in water treatment. Due to identified N-DBP groups having disparate precursors and formation pathways certain control parameters are group specific and may have the opposite effect on other N-DBPs. Thus no single method can be recommended for all N-DBPs, instead control strategies should be based upon a water specific assessment of precursor sources and DBP formation. DHANs can be produced from both chlorination (e.g. of amino acids) and chloramination (e.g. of aldehydes). Thus, dependent on the availability of reactive precursors contradictory effects of chlorination versus chloramination
4352
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 4 1 e4 3 5 4
on HAN formation are possible, with most literature showing higher levels following chlorination. Pre-oxidation with chlorine dioxide and post-disinfection at acidic pH have been observed to enhance HAN formation, thus avoiding these conditions may provide mitigation strategies. HAcAms are a recently reported group of DBPs whose formation is believed to be related to that of HAAs and HANs. Assuming this association is relevant to drinking water treatment, much of the above description of HAN formation and control may apply to the HAcAms, although further investigation is needed to confirm this prediction. No meaningful correlations were found between HNM formation and the THMs or HAAs using available N-DBP data, indicating disparate precursors and/or formation routes. UV irradiation or ozonation of drinking water before chlorination can increase chloropicrin formation by at least double and this is presumably related to the nitration of aromatic NOM. Thus, use of alternative treatments can be an effective method to control HNMs. Elevated levels of CNX in chloraminated drinking water are consistent with the pre-eminence of a formation route such as the monochloramination of formaldehyde. Further, the potential for formaldehyde release from ozonation and chlorination of NOM indicates either of these treatments preceding post-chloramination are combinations likely to promote CNX in potable water. Thus, CNX formation can be minimised by operating disinfection without secondary chloramination. Factors involved in the formation of NDMA in drinking water are chloramination, the presence of EfOM and the use of certain ion exchange resins or polymers. The majority of NMDA precursors are believed to be of anthropogenic origin. Oxidation prior to chloramination, for example by free chlorine, can limit NDMA formation. Other nitrosamines, notably NPYR, NMOR and NDEA, have been identified in the low ng L-1 range in waters with significant NDMA, indicating common precursor sources (e.g. wastewater) and that strategies reducing NDMA may also be effective for the other named nitrosamines. However, more work is needed to increase understanding of the formation and occurrence of nitrosamines other than NDMA. Common water treatment processes such as coagulation and filtration are typically rather ineffective for removing DON in general and N-DBP precursors in particular, therefore advanced processes such as membranes and biodegradation are likely to be more appropriate for reducing concentrations of these groups.
Acknowledgements The Department for Environment, Food and Rural Affairs (Defra) is gratefully acknowledged for funding a study by the authors entitled, ‘Review of the Current Toxicological and Occurrence Information Available on Nitrogen-Containing Disinfection ByProducts’, on which this paper is based. The authors wish to thank Stuart Krasner, Michael Plewa and Susan Richardson for their assistance with the original study.
references
AWWA, 2000. AWWA water quality division disinfection systems committee. Committee report - disinfection at large and medium sized systems. J. AWWA 92, 32e43. Bond, T., Goslan, E.H., Jefferson, B., Roddick, F., Fan, L., Parsons, S.A., 2009. Chemical and biological oxidation of NOM surrogates and effect on HAA formation. Water Res. 43, 2615e2622. Bond, T., Goslan, E.H., Parsons, S.A., Jefferson, B., 2010. Disinfection by-product formation of natural organic matter surrogates and treatment by coagulation, MIEX and nanofiltration. Water Res. 44, 1645e1653. Bull, R.J., Reckhow, D.A., Rotello, V., Bull, O.M., Kim, J., 2006. Use of Toxicological and Chemical Models to Prioritize DBP Research. AWWA Research Foundation, Denver, CO, USA. Charrois, J.W.A., Arend, M.W., Froese, K.L., Hrudey, S.E., 2004. Detecting N-nitrosamines in drinking water at nanogram per liter levels using ammonia positive chemical ionization. Environ. Sci. Technol. 38, 4835e4841. Chen, B.Y., Nam, S.N., Westerhoff, P.K., Krasner, S.W., Amy, G., 2009. Fate of effluent organic matter and DBP precursors in an effluent-dominated river: a case study of wastewater impact on downstream water quality. Water Res. 43, 1755e1765. Chen, W.-H., Young, T.M., 2008. NDMA formation during chlorination and chloramination of aqueous diuron solutions. Environ. Sci. Technol. 42, 1072e1077. Chen, Z., Valentine, R.L., 2008. The influence of the pre-oxidation of natural organic matter on the formation of Nnitrosodimethylamine (NDMA). Environ. Sci. Technol. 42, 5062e5067. Choi, J., Richardson, S.D., 2004. Formation Studies of Halonitromethanes in Drinking Water. AWWA, Denver, CO, USA. AWWA Water Quality Technology Conference. Choi, J., Valentine, R.L., 2002. Formation of Nnitrosodimethylamine (NDMA) from reaction of monochloramine: a new disinfection by-product. Water Res. 36, 817e824. Choi, J., Valentine, R.L., 2003. N-nitrosodimethylamine formation by free-chlorine-enhanced nitrosation of dimethylamine. Environ. Sci. Technol. 37, 4871e4876. Chu, W.-H., Gao, N.-Y., Deng, Y., 2010. Formation of haloacetamides during chlorination of dissolved organic nitrogen aspartic acid. J. Hazard. Mater. 173, 82e86. Davis, W.E., Li, Y., 2008. Analysis of hydrazine in drinking water by isotope dilution gas chromatography/tandem mass spectrometry with derivatization and liquid-liquid extraction. Anal. Chem. 80, 5449e5453. Deborde, M., von Gunten, U., 2008. Reactions of chlorine with inorganic and organic compounds during water treatmentkinetics and mechanisms: a critical review. Water Res. 42, 13e51. Diehl, A.C., Speitel, G.E., Symons, J.M., Krasner, S.W., Hwang, S. J., Barrett, S.E., 2000. DBP formation during chloramination. J. AWWA 92, 76e90. Dotson, A., Westerhoff, P., 2009. Occurrence and removal of amino acids during drinking water treatment. J. AWWA 101, 101e115. Dotson, A., Westerhoff, P., Krasner, S.W., 2009. Nitrogen enriched dissolved organic matter (DOM) isolates and their affinity to form emerging disinfection by-products. Water Sci. Technol. 60, 135e143. Glezer, V., Harris, B., Tal, N., Iosefzon, B., Lev, O., 1999. Hydrolysis of haloacetonitriles: linear free energy relationship, kinetics and products. Water Res. 33, 1938e1948. Goslan, E.H., Krasner, S.W., Bower, M., Rocks, S.A., Holmes, P., Levy, L.S., Parsons, S.A., 2009. A comparison of disinfection by-
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 4 1 e4 3 5 4
products found in chlorinated and chloraminated drinking waters in Scotland. Water Res. 43, 4698e4706. Heller-Grossman, L., Idin, A., Limoni-Relis, B., Rebhun, M., 1999. Formation of cyanogen bromide and other volatile DBPs in the disinfection of bromide-rich lake water. Environ. Sci. Technol. 33, 932e937. Hirose, Y., Maeda, N., Ohya, T., Nojima, K., Kanno, S., 1988. Formation of cyanogen chloride by the reaction of amino acids with hypochlorous acid in the presence of ammonium ion. Chemosphere 17, 865e873. Hoigne, J., Bader, H., 1988. The formation of trichloronitromethane (chloropicrin) and chloroform in a combined ozonation/chlorination treatment of drinking water. Water Res. 22, 313e319. Hrudey, S.E., 2009. Chlorination disinfection by-products, public health risk tradeoffs and me. Water Res. 43, 2057e2092. Hua, G.H., Reckhow, D.A., 2007. Relationship between brominated THMs, HAAs, and total organic bromine during drinking water chlorination. Symposium on occurrence, formation, health effects and control of disinfection by-products in drinking water held at the 233rd ACS National Meeting American Chemical Society. Jobb, D.B., Hunsinger, R.B., Meresz, O., Taguchi, V., 1994. Removal of N-nitrosodimethylamine from the Oshweken (six nations) water supply. Energy Final Report Ontario Ministry of Environment and Energy. Joo, S.H., Mitch, W.A., 2007. Nitrile, aldehyde, and halonitroalkane formation during chlorination/chloramination of primary amines. Environ. Sci. Technol. 41, 1288e1296. Keefer, L.K., Roller, P.P., 1973. N-nitrosation by nitrite ion in neutral and basic medium. Science 181, 1245e1247. Krasner, S.W., McGuire, M.J., Jacangelo, J.G., Patania, N.L., Reagan, K.M., Marco Aieta, E., 1989. Occurrence of disinfection by-products in US drinking water. J. AWWA 81, 41e53. Krasner, S.W., Sclimenti, M.J., Mitch, W.A., Westerhoff, P., Dotson, A.A., 2007. Wastewater and Algal Derived N-DBPs. AWWA, Denver, CO, USA. AWWA Annual Conference. Krasner, S.W., Weinberg, H.S., Richardson, S.D., Pastor, S.J., Chinn, R., Sclimenti, M.J., Onstad, G.D., Thruston, A.D., 2006. Occurrence of a new generation of disinfection byproducts. Environ. Sci. Technol. 40, 7175e7185. Krasner, S.W., Westerhoff, P., Chen, B., Amy, G., Nam, S.-N., Chowdhury, Z.K., Sinha, S., Rittmann, B.E., 2008. Contribution of Wastewater to DBP Formation. AWWA Research Foundation, Denver, CO, USA. Krasner, S.W., Westerhoff, P., Chen, B., Rittmann, B.E., Amy, G., 2009a. Occurrence of disinfection byproducts in United States wastewater treatment plant effluents. Environ. Sci. Technol. 43, 8320e8325. Krasner, S.W., Westerhoff, P., Chen, B., Rittmann, B.E., Nam, S.-N., Amy, G., 2009b. Impact of wastewater treatment processes on organic carbon, organic nitrogen, and DBP precursors in effluent organic matter. Environ. Sci. Technol. 43, 2911e2918. Kulshrestha, P., McKinstry, K.C., Fernandez, B.O., Feelisch, M., Mitch, W.A., 2010. Application of an optimized total Nnitrosamine (TONO) assay to pools: placing Nnitrosodimethylamine (NDMA) determinations into perspective. Environ. Sci. Technol. 44, 3369e3375. Lee, W., Westerhoff, P., 2009. Formation of organic chloramines during water disinfection - chlorination versus chloramination. Water Res. 43, 2233e2239. Lee, W., Westerhoff, P., Croue´, J.-P., 2007. Dissolved organic nitrogen as a precursor for chloroform, dichloroacetonitrile, N-nitrosodimethylamine, and trichloronitromethane. Environ. Sci. Technol. 41, 5485e5490. Lee, W., Westerhoff, P., Esparza-Soto, M., 2006. Occurrence and removal of dissolved organic nitrogen in US water treatment plants. J. AWWA 98, 102e110.
4353
Leenheer, J.A., Dotson, A., Westerhoff, P., 2007. Dissolved organic nitrogen fractionation. Annals of Environ. Science 1, 45e46. Lei, H.X., Minear, R.A., Marinas, B.J., 2006. Cyanogen bromide formation from the reactions monobromamine and dibromamine with cyanide ion. Environ. Sci. Technol. 40, 2559e2564. Loeppky, R.N., Michejda, C.J., 1994. Nitrosamines and Related NNitroso Compounds, ACS Symposium Series. American Chemical Society, Washington, DC. McGuire, M.J., McLain, J.L., Obolensky, A., 2002. Information Collection Rule Data Analysis. AwwaRF and AWWA, Denver, CO, USA. Merlet, N., Thibaud, H., Dore, M., 1985. Chloropicrin formation during oxidative treatments in the preparation of drinking water. Sci. Total Environ. 47, 223e228. Mirvish, S.S., 1975. Formation of N-nitroso compounds chemistry, kinetics, and invivo occurrence. Toxicol. Appl. Pharmacol. 31, 325e351. Mitch, W.A., Krasner, S.W., Westerhoff, P., Dotson, A., 2009. Occurrence and Formation of Nitrogenous Disinfection ByProducts. Water Research Foundation, D., CO, USA. Mitch, W.A., Sedlak, D.L., 2002. Formation of Nnitrosodimethylamine (NDMA) from dimethylamine during chlorination. Environ. Sci. Technol. 36, 588e595. Mitch, W.A., Sharp, J.O., Trussell, R.R., Valentine, R.L., AlvarezCohen, L., Sedlak, D.L., 2003. N-nitrosodimethylamine (NDMA) as a drinking water contaminant: a review. Environ. Eng. Sci. 20, 389e404. Na, C.Z., Olson, T.M., 2004. Stability of cyanogen chloride in the presence of free chlorine and monochloramine. Environ. Sci. Technol. 38, 6037e6043. Najm, I., Trussell, R.R., 2001. NDMA formation in water and wastewater. J. AWWA 93, 92e99. OEHHA, 2006. Public Health Goal for N-Nitrosodimethylamine in Drinking Water. Office of Environmental Health Hazard Assessment, CA. Oliver, B.G., 1983. Dihaloacetonitriles in drinking-water - algae and fulvic-acid as precursors. Environ. Sci. Technol. 17, 80e83. Pedersen, E.J., Urbansky, E.T., Marinas, B.J., Margerum, D.W., 1999. Formation of cyanogen chloride from the reaction of monochloramine with formaldehyde. Environ. Sci. Technol. 33, 4239e4249. Pietsch, J., Sacher, F., Schmidt, W., Brauch, H.J., 2001. Polar nitrogen compounds and their behaviour in the drinking water treatment process. Water Res. 35, 3537e3544. Plewa, M.J., Muellner, M.G., Richardson, S.D., Fasano, F., Buettner, K.M., Woo, Y.-T., McKague, A.B., Wagner, E.D., 2008. Occurrence, synthesis, and mammalian cell cytotoxicity and genotoxicity of haloacetamides: an emerging class of nitrogenous drinking water disinfection byproducts. Environ. Sci. Technol. 42, 955e961. Plewa, M.J., Wagner, E.D., 2009. Mammalian Cell Cytotoxicity and Genotoxicity of Disinfection By-Products. Water Research Foundation, D., CO, USA. Pocernich, M., Litke, D.W., 1997. Nutrient concentrations in wastewater treatment plant effluents, South Platte River Basin. J. Am. Water Resour. Ass 33, 205e214. Ram, N.M., 1985. A review of the significance and formation of chlorinated N-organic compounds in water supplies including preliminary studies on the chlorination of alanine, tryptophan, tyrosine, cytosine, and syringic acid. Environ. Int. 11, 441e451. Reckhow, D.A., Linden, K.G., Kim, J., Shemer, H., Makdissy, G., 2010. Effect of UV treatment on DBP formation. J. AWWA 102, 100e113. Reckhow, D.A., Platt, T.L., MacNeill, A.L., McClellan, J.N., 2001. Formation and degradation of dichloroacetonitrile in drinking waters. J. Water Supply Res. Technol. 50, 1e13.
4354
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 4 1 e4 3 5 4
Richardson, S.D., 2008. Environmental mass spectrometry: emerging contaminants and current issues. Anal. Chem. 80, 4373e4402. Richardson, S.D., Plewa, M.J., Wagner, E.D., Schoeny, R., DeMarini, D.M., 2007. Occurrence, genotoxicity, and carcinogenicity of regulated and emerging disinfection byproducts in drinking water: a review and roadmap for research. Mutat. Res. Rev. Mutat. Res. 636, 178e242. Richardson, S.D., Thruston, A.D., Caughran, T.V., Chen, P.H., Collette, T.W., Floyd, T.L., Schenck, K.M., Lykins, B.W., Sun, G.R., Majetich, G., 1999. Identification of new ozone disinfection byproducts in drinking water. Environ. Sci. Technol. 33, 3368e3377. Richardson, S.D., Thruston, A.D., Collette, T.W., Patterson, K.S., Lykins, B.W., Majetich, G., Zhang, Y., 1994. Multispectral identification of chlorine dioxide disinfection by-products in drinking-water. Environ. Sci. Technol. 28, 592e599. Richardson, S.D., Thruston, A.D., Rav-Acha, C., Groisman, L., Popilevsky, I., Juraev, O., Glezer, V., McKague, A.B., Plewa, M.J., Wagner, E.D., 2003. Tribromopyrrole, brominated acids, and other disinfection byproducts produced by disinfection of drinking water rich in bromide. Environ. Sci. Technol. 37, 3782e3793. Rook, J.J., 1974. Formation of haloforms during chlorination of natural water. Water Treat. Examination 23, 234e243. Sacher, F., Schmidt, C., Lee, C., von Gunten, U., 2008. Strategies for Minimizing Nitrosamine Formation during Disinfection. AWWA Research Foundation, Denver, CO, USA. Schreiber, I.M., Mitch, W.A., 2005. Influence of the order of reagent addition on NDMA formation during chloramination. Environ. Sci. Technol. 39, 3811e3818. Schreiber, I.M., Mitch, W.A., 2006a. Nitrosamine formation pathway revisited: the importance of chloramine speciation and dissolved oxygen. Environ. Sci. Technol. 40, 6007e6014. Schreiber, I.M., Mitch, W.A., 2006b. Occurrence and fate of nitrosamines and nitrosamine precursors in wastewaterimpacted surface waters using boron as a conservative tracer. Environ. Sci. Technol. 40, 3203e3210. Schreiber, I.M., Mitch, W.A., 2007. Enhanced nitrogenous disinfection byproduct formation near the breakpoint: implications for nitrification control. Environ. Sci. Technol. 41, 7039e7046. Seidel, C.J., McGuire, M.J., Summers, R.S., Via, S., 2005. Have utilities switched to chloramines? J. AWWA 97, 87e97. Shah, A.D., Dotson, A.A., Linden, K.G., Mitch, W.A., 2011. Impact of UV disinfection combined with chlorination/chloramination on the formation of halonitromethanes and haloacetonitriles in drinking water. Environ. Sci. Technol. 45, 3657e3664.
Sharpless, C.M., Linden, K.G., 2003. Experimental and model comparisons of low- and medium-pressure Hg Lamps for the direct and H2O2 assisted UV photodegradation of Nnitrosodimethylamine in simulated drinking water. Environ. Sci. Technol. 37, 1933e1940. Simpson, K.L., Hayes, K.P., 1998. Drinking water disinfection byproducts: an Australian perspective. Water Res. 32, 1522e1528. Stevens, A.A., Moore, L.A., Miltner, R.J., 1989. Formation and control of non-trihalomethane disinfection by-products. J. AWWA 81, 54e60. Symons, J.M., Krasner, S.W., Simms, L.A., Sclimenti, M., 1993. Measurement of THM and precursor concentrations revisited the effect of bromide ion. J. AWWA 85, 51e62. Templeton, M., Chen, Z., 2010. NDMA and seven other nitrosamines in Selected UK drinking water supply systems. J. Water Supply: Res. Technol. AQUA 2010, 277e283. Trehy, M.L., Bieber, T.I., 1981. Detection, identification, and quantitative analysis of dihaloacetonitriles in chlorinated natural waters. In: Keith, L.H. (Ed.), Advances in the Identification and Analysis of Organic Pollutants in Water. Abb Arbor Science, Ann Arbor, MI, pp. 941e975. Trehy, M.L., Yost, R.A., Miles, C.J., 1986. Chlorination byproducts of amino acids in natural waters. Environ. Sci. Technol. 20, 1117e1122. Villanueva, C.M., Cantor, K.P., Grimalt, J.O., Malats, N., Silverman, D., Tardon, A., Garcia-Closas, R., Serra, C., Carrato, A., Castan˜o-Vinyals, G., Marcos, R., Rothman, N., Real, F.X., Dosemeci, M., Kogevinas, M., 2007. Bladder cancer and exposure to water disinfection by-products through ingestion, bathing, showering, and swimming in pools. Am. J. Epidemiol. 165, 148e156. Weinberg, H.S., Glaze, W.H., Krasner, S.W., Sclimenti, M.J., 1993. Formation and removal of aldehydes in plants that use ozonation. J. AWWA 85, 72e85. Weinberg, H.S., Krasner, S.W., Richardson, S.D., Thruston, A.D., 2002. The Occurrence of Disinfection By-Products (DBPs) of Health Concern in Drinking Water: Results of a Nationwide DBP Occurrence Study Athens, GA. Westerhoff, P., Mash, H., 2002. Dissolved organic nitrogen in drinking water supplies: a review. J. Water Supply Res. T. 51, 415e448. Williams, D.T., LeBel, G.L., Benoit, F.M., 1995. A National Survey of Chlorinated Disinfection By-products in Canadian Drinking Water Canada, H. Williams, D.T., LeBel, G.L., Benoit, F.M., 1997. Disinfection byproducts in Canadian drinking water. Chemosphere 34, 299e316. Xie, Y., Reckhow, D.A., 1992. Stability of Cyanogen Chloride in the Presence of Sulfite and Chlorine. AWWA, Denver, CO, USA. AWWA Water Quality Technology Conference.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 5 5 e4 3 6 6
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Foam in the aquatic environment Katerina Schilling a,b,*, Matthias Zessner a,b a b
Institute for Water Quality, Resources and Waste Management, Vienna University of Technology, Karlsplatz 13/226, 1040 Vienna, Austria The Centre for Water Resource Systems, Vienna University of Technology, Karlsplatz 13/226, 1040 Vienna, Austria
article info
abstract
Article history:
Foams are ubiquitous in the environment, commonly seen as discoloured patches on
Received 10 January 2011
streams, rivers, lakes and sea water. They often are assumed to be anthropogenic in origin
Received in revised form
as they are aesthetically unpleasant, yet they frequently appear in pristine environments
1 June 2011
indicating a natural origin. In contrast to “hidden” chemical pollution, e.g. heavy metals,
Accepted 1 June 2011
pesticides etc. the visibility of foam alarms the public.
Available online 28 June 2011
To derive more information on foam in freshwaters and marine ecosystems, a literature review was performed. Alongside with some basic considerations on the formation of
Keywords:
foam, on methods to measure foam formation and on the legal aspects of foam on surface
Foam formation
waters, the ecological importance of foam in the aquatic environment is discussed in this
Surface microlayer
paper.
Foam enrichment
ª 2011 Elsevier Ltd. All rights reserved.
Sea foam
Contents 1. 2.
3. 4.
5. 6.
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Foam characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1. Foam formation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2. Surface active compounds . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3. Surface microlayer (SML) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.4. Foam stability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Methods to quantify foam formation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Legal aspects of foam formation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1. Regulation of instream foam formation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2. Regulation of foam emitted by point sources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Foam formation in wastewater systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Relevance of foam on surface waters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.1. Enrichment of foams in lakes and rivers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.1.1. Metals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.1.2. Carbon and nutrients . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.1.3. Lipids and hydrocarbons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.1.4. Polychlorinated biphenyl (PCB) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4356 4356 4356 4357 4357 4357 4358 4359 4359 4359 4359 4360 4360 4360 4361 4361 4361
* Corresponding author. Institute for Water Quality, Resources and Waste Management, Vienna University of Technology, Karlsplatz 13/226, 1040 Vienna, Austria. Tel.: þ43 1 58801 22613; fax: þ43 1 58801 22699. E-mail address:
[email protected] (K. Schilling). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.06.004
4356
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 5 5 e4 3 6 6
6.2. 6.3.
7.
1.
Occurrence and composition of sea foam . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Ecological importance of foam . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.3.1. Toxicity of foam . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.3.2. Foam as food resource . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.3.3. Foam as habitat . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.3.4. Other roles of foam in the environment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Introduction
Since the implementation of biodegradable detergents and the proper operation of wastewater treatment plants, it has been believed that the occurrence of foam on surface waters was eliminated in Austria and other countries with a welldeveloped waste water treatment infrastructure. But still, the formation of foam in freshwater ecosystems such as rivers and lakes has been reported several times. In this context foam is linked to point source emissions (Defrain and SchulzeRettmer, 1989; Madrange et al., 1992; Ruzicka et al., 2009) as well as natural occurring surfactants derived from algae and plants (Harrington, 1997; Lu et al., 2006; Shi et al., 2005; Wegner and Hamburger, 2002). Considerable literature is available on the occurrence of “sea foam” in marine ecosystems (Ba¨rlocher et al., 1988; Ba¨tje and Michealis, 1986; Craig et al., 1989; Desroy and Denis, 2004; Eberlein et al., 1985; Maynard, 1968; Seuront et al., 2006; Velimirov, 1982). Nevertheless information on foam in the aquatic environment is limited and sometimes out-dated compared to the existing literature on the foams linked to various industrybranches, e.g. food industry (Dale et al., 1999; Piazza et al., 2008; Vanrell et al., 2007), fire-fighting-industry (Figueredo and Sabadini, 2003; Tafreshi and Marzo, 1999), cement-industry (Pedersen et al., 2007), cosmetic- and detergent-industry (Buzzacchi et al., 2006; Regismond et al., 1998; Zhang et al., 2004). Other sources of information are investigations on foam in wastewater treatment processes due to the occurrence of filamentous bacteria (Blackall et al., 1991; Blackall and Marshall, 1989; Hladikova et al., 2002; La´nsky´ et al., 2005; Lemmer et al., 2005; Petrovski et al., 2010; Torregrossa et al., 2005). Today quality and pollution of surface water is a not only an issue for scientists, but also policymakers, politicians and the public in general. Although foam can be used in environmental remediation processes, e.g. in aquifers (Vikingstad, 2006), in water science foam is known to be the polluter, not the pollution remover. The public exhibits considerable interest in foams and usually associates them with detergents or some form of pollution (Mills et al., 1996; Pojasek and Zajicek, 1978). A psychological investigation demonstrated that the public associated the presence of foam on surface waters with pollution which results in a reduced preference to use the water for recreational purposes or as drinking water sources (Wilson et al., 1995). A good example on how foam can provoke public concern is the formation of foam on an Austrian river close to the border with Hungary that resulted in an impaired political
4361 4362 4362 4362 4363 4363 4363 4364 4364
relationship due to massive protests from Hungarian locals. A research study related the foam to the effluent of three tanning factories, although their wastewater treatment complied with best available technology and met the legal requirements (Ruzicka et al., 2009). As a result of the public concern, the Austrian emission directive for tanneries was adapted and the implementation of quaternary chemical treatment became a requirement to prevent instream foam formation. This example highlights the increasing public awareness in terms of water pollution. Certainly, the visibility of foam addresses the ecological ‘‘conscience” of people more than ‘‘hidden” chemical pollution. In fact, foam is not necessarily associated with pollution, but can occur naturally in very pristine environments, e.g. humic waters from rainforests. The increasing environmental awareness and regulatory pressures demand a better understanding of the composition, origin and significance of foam (Napolitano and Richmond, 1995). In consequence, a literature review was carried out to identify existing information on foam itself, reasons for its formation and its behaviour in the aquatic environment. This paper will also deal with methods to measure foam formation, the existing legislation as well as with the importance and dangers of foam formation from an ecological perspective.
2.
Foam characteristics
2.1.
Foam formation
Foam is a dispersion of a gas in a liquid or solid separated by thin liquid films or lamellae (Heard et al., 2008; Holmberg et al., 2003). A pure liquid cannot foam unless a surface-active material is present. A gas bubble introduced below the surface of a pure liquid will burst immediately as soon as the liquid drains away. In fact, purity of water can be estimated from the bubble persistence time when shaking the water sample in a closed container. A bubble persistence time of even 1 s is an indication of the presence of surface-active impurities (Pugh, 1996). Consequently, foams are always formed from mixtures, where one component must be surface active. A measure of the surface activity is the decrease in surface tension upon adding the surface-active component to a pure liquid (Holmberg et al., 2003). Surface-active foaming materials include particles, polymers, specific absorbed cations or anions from organic salts etc., most of them cause foaming at extremely low concentrations (Pugh, 1996).
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 5 5 e4 3 6 6
A second precondition for foam formation is a source of gas bubbles (Heard et al., 2008) that must be injected at a higher rate than the rate at which the liquid between the bubbles can drain away (Napolitano and Cicerone, 1999). In aquatic ecosystems mechanical impact is needed to introduce air bubbles into the water (Poremba, 1991), which can be caused by cascades or hydraulic structures such as weirs, dams, spillways and discharge pipes, due to stormy conditions and the pounding of waves, in areas of strong currents or in areas, where fresh and salt water mixes (Craig et al., 1989; Ettema et al., 1989; Fisenko, 2004).
2.2.
Surface active compounds
Surface active components or surfactants have a great influence on the surface or interfacial properties in a solution. A surfactant is a molecule that has a hydrophobic hydrocarbon chain and a hydrophilic head group. The surfactant can be anionic, cationic or zwitterionic. Above the critical micelle concentration (cmc) the surfactant molecules aggregate in micelles. The micelles have an ordered structure that is dependent on the hydrophilic and hydrophobic properties of the surfactant (Vikingstad, 2006). According to literature several surface-active compounds can cause foam in the aquatic environment. Natural foams are usually linked to humic and fulvic acid substances (Ettema et al., 1989; Napolitano and Richmond, 1995), fine colloidal particles (Ettema et al., 1989), lipids and proteins originating from aquatic or terrestrial plants leaching from soil by precipitations events (Napolitano and Richmond, 1995), saponins representing a family of plant glycosides (Pojasek and Zajicek, 1978; Wegner and Hamburger, 2002), the exudation or decomposition products of phytoplankton containing carbohydrates and proteins (Ba¨tje and Michealis, 1986; Eberlein et al., 1985; Seuront et al., 2006) and the natural reservoir of organic matter occurring in sediments (Napolitano and Richmond, 1995). Man-made foam is linked to phosphates from farm fertilizers (Ettema et al., 1989) and organic and inorganic (detergents) pollution discharged by point sources especially from the paper and leather industry (Ettema et al., 1989; Fisenko, 2004; Madrange et al., 1992; Ruzicka et al., 2009). In summary, the literature provides a wide range of foam causing substances, which have been investigated to greater and lesser extents. Considering the fact that foam formation is a sum effect of all surface-active compounds present in the water, in most cases not one single substance, but a mixture of various components is responsible for foam (Wegner and Hamburger, 2002).
2.3.
Surface microlayer (SML)
In marine and freshwater ecosystems, the airewater boundary, known as the surface microlayer (SML), constitutes an important interface between the troposphere and the underlying water (Ho¨rtnagl et al., 2010). It is a vehicle for the transport of inorganic and organic materials between the atmosphere and the water column (Napolitano and Cicerone, 1999). Substances and particles temporarily or permanently incompatible with purely aquatic or atmospheric ecosystems, e.g. natural oils, organic acids and proteins will accumulate in
4357
this zone near the air-water interface. SMLs may also include components derived from human activities, such as petroleum compounds, synthetic surfactants, long-chain alcohols, synthetic pesticides and herbicides (Napolitano and Richmond, 1995; Parker and Barsom, 1970). These materials accumulate in the SML by adsorbing onto bubbles as they rise through the water column (Harden and Williams, 1989). At the water surface the bubbles may burst ejecting aerosol droplets into the air. Under certain conditions emerging bubbles may not burst instantly, but accumulate on the water surface producing foam. The aqueous foam phase contains the surface-active substances accumulated in the SML. Assuming that a typical foam consists of 90% air and that the mean thickness of the SML is about 50 mm, 1 L of foam water would represent 2 m2 of SML (Napolitano and Cicerone, 1999). Foam may also be produced by the compression of the SML generated by wind or as a result of breaking waves at the shoreline (Ba¨rlocher et al., 1988; Eisenreich et al., 1978; Harden and Williams, 1989).
2.4.
Foam stability
Due to their high interfacial energy foams are thermodynamically unstable. The instability has been classified into two types, which are (1) unstable or transient foams with lifetime of seconds and (2) metastable or so called permanent foams with lifetimes measured in hours to days (Pugh, 1996). Collapsing foams are the result of the bubble coalescence defined as joining together of two bubbles in a fluid to form one larger bubble. It can be described as a three-step process: bubble approach and creation of a thin film, film drainage and film rupture. In pure water no stable film is formed and coalescence takes only a few milliseconds, whereas surfactants stabilise bubbles against coalescence leading to foam lifetimes from seconds to days (Henry, 2010). The most obvious force acting on foam is gravity causing drainage of the liquid between the air bubbles. Drainage can be reduced by either increasing the viscosity of the bulk liquid or by adding particles (Heard et al., 2008; Holmberg et al., 2003; Pugh, 1996). Such systems give very stable foams (Holmberg et al., 2003; Wegner and Hamburger, 2002). The influence of solid particles on the formation and stability of foam is dependent on the surfactant type, the particle size and concentration. Hydrophilic particles present in the aqueous phase of the foam films are able to enhance foam stability by slowing down the film drainage. On the other hand, hydrophobic particles entering the air-water surfaces of the foam can cause destabilisation via the bridging-dewetting mechanism (Binks, 2002). According to Pugh (1996) partially hydrophobic particles can cause an increase or decrease in foam stability. Small particles, if not fully wetted, may become attached to the interface and give some mechanical stability to the lamella. If completely dispersed, they may cause an increase in bulk viscosity and stabilise foams. On the other hand, larger particles having a higher degree of hydrophobicity (coal dust, sulphur, non-wetting quartz) may cause destabilisation. In wastewater treatment plants, the stability of foams is often associated with hydrophobic particles such as bacteria cells (Blackall and Marshall, 1989; Heard et al., 2008; Petrovski et al., 2010) Particles can show a second
4358
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 5 5 e4 3 6 6
stabilizing mechanism by being surface active themselves and having a high affinity for the liquideair interface. Eisenreich et al. (1978) found surface-active proteinaceous matter as well as small Si, Ca and Fe particles to be responsible for foam stability. Addition of salt is another factor promoting foam stability. Craig et al. (1993) found that bubble coalescence was inhibited by some salts, whereas others had no effect. According to Holmberg et al. (2003) salt increases the surfactant critical packing parameter (CPP), which indicates how close surfactants are packed together at the air-water surface. Thus an increase in CPP will result in closely packed surfactants, in an increased surface elasticity and viscosity and thus lead to high foaming ability and foam stability. Henry (2010) argued that electrolytes inhibit bubble coalescence in water, but the inhibition is ion specific. Her experiments on thin film drainage in electrolyte solutions showed that electrolytes affect both film drainage rate and film rupture thickness. According to her results, electrolyte coalescence inhibition is a nonequilibrium effect that acts upon the dynamic film drainage process, through ion specific interfacial partitioning. Pugh (1996) describes the increase of foam stability due to a mixture of surfactants, e.g. a mixture of tannin and heptanoic acid in aqueous solution, whereas much lower foam stability is observed from the two constituents separately. In summary, bubble coalescence is the result of bubble approach, film drainage and finally film rupture. Foam stability can be increased by adding various substances, such as particles, salts or a mixture of surfactants.
3.
Methods to quantify foam formation
Literature on foam formation and stability mainly originates from the wastewater sector and in sectors dealing with foams in industry (Blackall et al., 1991; Blackall and Marshall, 1989; Heard et al., 2008; Hladikova et al., 2002; Paris, 2004; Vikingstad et al., 2005). Table 1 provides an overview of several methods used to measure foam. In the wastewater sector, tests on foam stability are related to the occurrence of bulking sludge. The most cited method was developed by Blackall and Marshall (1989). They designed
Table 1 e Overview on methods to measure foam. Method
Measured parameters
Aeration
Foam generation, Foam stability Bubble size and stability, foam height Scum index Height of foam column as function of time Foam height, foam stability Foaming factor Foam index
Alka-Seltzer
Aeration Mixing
Handshaking Shaking Webcam
Cited in Blackall and Marshall (1989) Paris (2004)
Hladikova et al. (2002) Vikingstad et al. (2006)
Madrange et al. (1992) Ruzicka et al. (2009) Ruzicka et al. (2009)
a foaming apparatus consisting of a glass cylinder and a sintered glass disc with a maximum pore size of 40e90 mm. In the cylinder 50 ml of sample are aerated with compressed air (200 ml/min) via the glass disc. Foam generation and stability are recorded and assessed according to a classification system rated in terms of foam volume, bubble size, speed of formation and time taken for the foam to collapse after aeration ceased. Heard et al. (2008) slightly modified the method of Blackall and Marshall (1989) and applied 20 ml of sample which is aerated with an air flow rate of 100 ml/min. They use the time at which bubbles collapse as a measure of foam stability. The AlkaeSeltzer test was developed by Ho und Jenkins and modified by Kopplow und Barjenbruch (Paris, 2004). This test uses two tablets of AlkaeSeltzer (AlkaeSeltzer classic, Bayer, is a drug containing sodium bicarbonate amongst others) which are added to a beaker with 250 ml of sample. Dissolution of the tablets creates bubbles that lead to foam formation. Size and stability of the bubbles are noted and the height of foam is measured at intervals of one, three and 5 min after complete dissolution of the tablet, in order to calculate the “foam value” (Kopplow and Barjenbruch, 2002). The scum index (SI) is a method to estimate foaming severity and was primarily proposed by Pretorius and Laubscher (1987) and modified by Hladikova et al. (2002). The first step is the determination of the concentration of suspended solids in the mixed liquor. Then two litres of mixed liqor are aerated in a laboratory cylinder with a flow rate of 480 ml/min via a sintered silica sand diffusor. After an aeration time of 4 h, the dry mass of recovered scum is determined. The scum index is calculated by dividing the mass of suspended solids initially present by the mass of the stored scum, multiplied by 100%. Vikingstad et al. (2005) applied foam tests to assess foam stability in the oil-water-interface. Foam tests were made by mixing air into the surfactant solution. Air was dispersed into the 300 ml test solution with a pedal connected to a mixer at a speed of 2000 rounds per minute for 5 min. The mixer was a polymix obtained from Kinematica, type RW20 S12. In all the experiments the height of the foam column above the liquid phase was measured as a function of time after mixing ceased. Limited information exists on the foaming capacity of effluent from wastewater treatment plants. Madrange et al. (1992) proposed a method to determine the foaming capacity of industrial effluents. 250 ml of sample were handshaken for 5 min and the height of the resulting foam was measured with an accuracy of 0.2 cm. Foam stability, defined as the time until the foam cover breaks, was measured. Ruzicka et al. (2009) introduced the foaming factor and the foam potential to estimate the capability of an effluent to cause foam in the receiving river. The foaming factor is derived by shaking 250 ml of effluent in Erlenmeyer flasks with baffles on a laboratory Shaker (Type Ceromat-U) for 3 min at a speed of 300 rpm. Samples are diluted with dilution media (in this case unpolluted river water), until no more foam appears. The dilution factor, at which minimal foam occurred, is defined as the “foaming factor”. For the calculation of the “foam potential”, the foaming factor of an effluent is multiplied with the discharge of the effluent. The calculated foam potential of an effluent is defined as the volume of river
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 5 5 e4 3 6 6
water which can potentially become foamed by the effluent’s discharge, if laboratory conditions are applied. The study of Ruzicka et al. (2009) also developed a sevenstage “foam index” (FI) based on webcam pictures of a river weir, where foam occurred. This parameter was developed to characterise instream foam formation and assess the amount of foam on the surface of the river. The index does not quantify the foam, but allows a semi-quantitative differentiation between the varying foaming conditions. Ruzicka et al. (2009) successfully correlate the resulting “foam index” in the river to the emitted “foam potential” of the major point sources in the catchment. As a result, all methods to quantify foam (apart from the foam index) are strongly dependent on the introduction of a gas source into the liquid, which is provided by (i) aerating, (ii) mixing or (iii) shaking the sample. As most of the methods originate from the wastewater sector as a result of the bulking sludge problems, a uniform approach would be beneficial and should be developed to make future investigations comparable.
4.
Legal aspects of foam formation
The need to act against degradation of aquatic ecosystems has been acknowledged by politicians and legislation has been adopted to stop further deterioration and to restore aquatic ecosystems to a healthy state (Carstensen, 2007). Examples of legislation in Europe are the European Water Framework Directive and the Regulation for Water Pollution Control in Switzerland; in the USA the US Clean Water Act is the leading standard. In all these legislations the regulation of foam is an almost neglected topic.
4.1.
Regulation of instream foam formation
The Clean Water Act (CWA) establishes the basic structure for regulating discharges of pollutants into the waters of the United States and regulates quality standards for surface waters. It proclaims that “all surface waters should be free of scum in unsightly amounts”. The phrase “unsightly” is not defined any further and its interpretation is subject to the evaluator (Federal Water Pollution Control Act, 2002). In the European Union the European Water Framework Directive (EU WFD) provides the legal standard. It applies the definition of ‘‘good chemical and ecological status” for river water bodies, which means only a minor deviation from the reference (natural) status. The chemical status is regulated by environmental quality standards for priority substances, which contain also foam related parameters such as tensides (e.g. nonylphenols). Thus an indirect criterion for foam exists by the regulation of tensioactive compounds, although specific regulation on formation of foam in surface waters is not available (Water Framework Directive, 2000760/EC, 2000). The Austrian state monitoring network is administered according to the requirements of the EU WFD and so far does not include any monitoring of foam formation (Bundesministerium fu¨r Land- und Forstwirtschaft, 2006). The Regulation for Water Pollution Control in Switzerland states that “treated wastewater discharge must not lead to
4359
foam formation after advanced mixing in the river with exception from rainfalls”. The evaluation of instream foam formation adopts a three-step scale, i.e. “no foam”, “minor to medium foam” and “lots of foam”. It is a visual survey and the resulting rating is based on comparison with pictures (Gewa¨sserschutzverordnung Schweiz, 1998).
4.2.
Regulation of foam emitted by point sources
In terms of point source pollution, the regulation of foam is weaker than with regards to instream foam formation. In the CWA, regulation of foam from point sources is not available (Federal Water Pollution Control Act, 2002). In the EU, the emissions from municipal wastewater treatment plants are regulated in the Urban Wastewater Treatment Directive, and this does not include any parameters regarding foam (Council Directive 91/271/EEC concerning urban wastewater treatment, 1991). In Austria, the Directive on emissions from municipal wastewater treatment plants lacks regulation on foam emissions, although emission criteria for foam related parameters such as tensioactive compounds, e.g. the sum parameter for anionic and non-ionic surfactants exist (Bundesministerium fu¨r Land- und Forstwirtschaft, 1991). Another directive indirectly dealing with foam is the Directive on emissions from tanneries, which includes the parameter “surface tension” in the effluent to avoid foam formation in the river (Bundesministerium fu¨r Land- und Forstwirtschaft, 2007). The Swiss Regulation for Water Pollution Control has no emission based legislation for foam, but does include the assessment of instream foam formation below point sources (Gewa¨sserschutzverordnung Schweiz, 1998). In summary, the regulation of foam in the aquatic environment has been neglected entirely, although the occurrence of foam is often prominent in the media and is frequently cited as a reason for public concern. In consequence, proper legislation is necessary for instream foam formation as well as for foam originating from point sources. In this context, the regulation of a sum parameter representing foam formation such as surface tension or foam potential would offer the greatest feasibility.
5.
Foam formation in wastewater systems
As foam on surface waters is often linked to point source emissions (Defrain and Schulze-Rettmer, 1989; Madrange et al., 1992; Ruzicka et al., 2009) and a considerable amount of literature exists on this subject, foam formation in wastewater systems is discussed in this section. The formation of stable foam which reduces oxygen transfer, decreases the quality of the effluent and therefore increases maintenance costs, is a widely observed phenomenon in wastewater treatment plants and was first noted in 1969 (Heard et al., 2008). It is a common feature of activated sludge systems around the world and considerable effort has been directed to enhance understanding of the microbial ecology of foaming (Petrovski et al., 2010). Microscopic examination of foams has identified a wide range of bacteria, with the filamentous bacteria Microthrix parvicella and actinomycetes among the most common (Blackall et al., 1991; Heard et al.,
4360
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 5 5 e4 3 6 6
2008). The results of Lemmer et al. (2005) highlighted that bacteria assembled in foam comprise morphotypes beside Microthrix parvicella and nocardioform actinomycetes, which belong to a variety of species, genera and bacteria groups. Lemmer et al. (2005) argues that the immense number of species prevent the recommendation of specific troubleshooting measures. Heard et al. (2008) indicated that bacteria can enhance the persistence of foams, but cannot in themselves cause foaming in the absence of a surfactant. Their findings are of importance, as much of the current literature suggests that foaming in wastewater treatment plants is caused directly by the presence of bacteria (Heard et al., 2008). Nonetheless, the earlier investigations of Blackall and Marshall (1989) highlighted that both surfactants and cells were necessary for stable foam formation, since only unstable foams were formed in the absence of bacteria, and no foams were formed in the absence of a surfactant. Interestingly, the surfactant needed to initialize foam formation must not be a synthetic surfactant from the influent of the treatment plants, but may be a biotenside, e.g. glycolipids produced by various sludge bacteria such as actinomycetes, Pseudomonas or Acinetobacter species (Lemmer et al., 2005). Foam stability seems to be dependent on a large biomass of filamentous, hydrophobic organisms such as Microthrix parvicella or actinomycetes (Blackall et al., 1991; Blackall and Marshall, 1989; Heard et al., 2008; Lemmer et al., 2005). According to Petrovski et al. (2010) foam is generated by a selective enrichment of these hydrophobic bacteria by a process of flotation, that requires three components: (i) gas bubbles surrounded by liquid films, generated by the aeration system, (ii) surfactants which reduce the surface tensions and thus prevent liquid drainage from gas bubble walls and (iii) small hydrophobic particles (bacterial cells) responsible for the long-term stabilisation. With insufficient hydrophobic cells, but in the presence of the other two, large amounts of unstable foam will be generated. These findings strongly support the investigations of Blackall and Marshall (1989). Furthermore, the authors demonstrated that the nonhydrophobic Bacillus subtilis might be an important contributor to stable foams due to the production of the powerful surfactant surfactin, supporting the biotenside theory of Lemmer et al. (2005). In the experiments of Blackall and Marshall (1989) the prevention of stable foams has been achieved by the addition of colloidal, hydrophilic clay particles. Although a variety of hydrophilic materials was tested, only a 2:1 lattice clay, montmorillonite, was able to inhibit stable foams. The authors postulate that a salt-dependent, reversible bacteriaemontmorillonite complex is formed, which confers hydrophilicity to the otherwise hydrophobic actinomycetes. This property prevents cells from entering and stabilising the foam phase. Other measures to fight stable foams include spraying with water to destroy the foam (Lemmer et al., 2005) or dosing with chemical agents such as polymers (Hwang and Tanaka, 1998). In conclusion, the widely observed appearance of stable foams on wastewater treatment plants results from the occurrence of surfactants or biotensides in the mixed liquor, whereas the hydrophobic bacterial cells are responsible for
the long-term stabilisation. Because of the high diversity of filamentous bacteria the effective control of foam in wastewater treatment plants is hard to achieve.
6.
Relevance of foam on surface waters
6.1.
Enrichment of foams in lakes and rivers
Foams occurring on rivers, lakes and in the sea are basically collections of materials normally present in the SML (Napolitano and Richmond, 1995). Similar to the foam fractionation techniques used in industry or in wastewater treatment, the formation of foam on surface waters induces the transfer and concentration of surface-active substances from the SML into the foam (Johnson et al., 1989). Depending on the type of aquatic environment various components are enriched in the resulting foam. Fisenko (2004) hypothesized that nature “uses” this “foam fractionation and flotation technique” as a process for self-purification. In his study he demonstrated the self-restoration of the Etobicoke river after a toxic waste spill. Within three months the river was selfpurified which he attributed to the production of large amounts of foam enriched in substances degrading water quality, such as cyanide and heavy metals. Some information is available regarding the enrichment of certain substances in foams of various rivers and lakes. The existing information deals with the accumulation of nutrients, lipids, heavy metals, hydrocarbon and pesticides in “freshwater foam” (Baier et al., 1974; Eisenreich et al., 1978; Johnson et al., 1989; Napolitano and Richmond, 1995; Parker and Barsom, 1970; Pojasek and Zajicek, 1978). The enrichment is calculated via the fractionation or enrichment ratio, which is defined as the concentration of a parameter in the foam divided by its concentration in the underlying water.
6.1.1.
Metals
The analysis of wind-generated lake foam (Lake Mendota, Wisconsin) highlighted an enrichment with metals as compared to the underlying water (Eisenreich et al., 1978). Table 2 provides an overview on the fractionation ratios for several metals (total) in Lake Mendota compared to other studies cited in Eisenreich et al. (1978) as well as a study for foam below the Niagara Falls (Johnson et al., 1989). Although the fractionation ratios show huge variations, an enrichment of the foam with metals is obvious. Differences in metal fractionation ratios between various studies are most likely to result from the use of different collection techniques and bulk water concentrations than foam accumulation mechanisms (Eisenreich et al., 1978). The observed metal enrichment in foam of Lake Mendota is caused by metal scavenging by surface-active or particulate material in surface films, bubble adsorption and atmospheric deposition. Furthermore, a significant portion of the dissolved metal was complexed by organic matter or associated with colloidal material accumulated in the foam (Eisenreich et al., 1978). Pojasek and Zajicek (1978) analysed the metal carrying capacity of natural SMLs and foams of several streams using iron and manganese as indicator metals (both dissolved). They found fractionation ratios varying between 10 and 40 for
4361
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 5 5 e4 3 6 6
Table 2 e Fractionation ratios for heavy metals in “freshwater foams” e modified from Eisenreich et al. (1978) Metal (total)
Fractionation ratio (average) Lake Mendota
Delaware Bay
Lake Michigan
Niagara Falls
Eisenreich et al., 1978
Szekielda et al., 1972
Elzerman, 1976
Johnson et al., 1989
293 544 1110 448 240
w10,000
14 67 271 98
80 20
Zn Cd Pb Cu Fe
w10,000 w10,000
manganese and 30 to 74 for iron depending on the preanalyses treatment applied. Half of the metals were strongly bound to organic terrestrial decomposition products; the rest were present as inorganic species or weakly bound organic complexes.
6.1.2.
Carbon and nutrients
Parker and Barsom (1970) investigated the SML of three freshwater habitats in the vicinity of St. Louis. Total nitrogen and orthophosphate showed a two- to threefold concentration in the SML compared to the underlying water. Unfortunately, no absolute numbers are available for this study. In Niagara Falls the fractionation ratio for DOC is about 8 (Johnson et al., 1989) compared to an average ratio of 40 for DOC and 48 for TOC in the foam of Lake Mendota (Eisenreich et al., 1978). The ratio for organic nitrogen was 95 in Lake Mendota, indicating the presence of large quantities of proteinaceous matter. About 80% of the total phosphorus occurred in dissolved form and was evenly distributed between dissolved reactive phosphorus (DRP) and dissolved organic phosphorus (DOP) with average ratios of 11 and 84, respectively (Eisenreich et al., 1978).
6.1.3.
Lipids and hydrocarbons
Lipids are one of the major organic constituents of SMLs and foams because of their hydrophobicity, a low relative density and low vapour pressure (Napolitano and Cicerone, 1999). The lipid concentrations of foam samples from rivers in eastern Tennessee show some variation, but were higher than in the underlying water with fractionation ratios spanning two orders of magnitude. According to the authors, the varying concentrations between the streams could be attributed to differences in the particle load of the foam, foam age and the extent of colonization by microbes (Napolitano and Richmond, 1995). Fractionation ratios of foam below the Niagara falls ranged from 15 for total fatty acids to 370 for total sterol depending on the degree of polarity (Johnson et al., 1989). Napolitano and Richmond (1995) regenerated foam in the laboratory to measure the enrichment under controlled conditions (8l of stream water plus 200 ml of corresponding foam). The resulting fractionation ratios for phospholipids and hydrocarbons were much lower than those found in the field. These findings suggest that high concentrations of lipids in natural foams are not a direct consequence of the foam formation, but indicate a secondary enrichment of lipids in the foam after its formation due to microbial growth or entrapment of suspended solids.
300
Natural hydrocarbons typically account for only a small proportion of the total lipids (typically <5%). The presence of hydrocarbons at 20%e30% in river foams in eastern Tennessee and at 10%e15% in foam below the Niagara Falls indicates oil contamination (Johnson et al., 1989; Napolitano and Richmond, 1995). Eisenreich et al. (1978) reported high concentrations of chlorinated hydrocarbons, such as dieldrin and DDT-group pesticides (4e360 ng/l) in foam of Lake Mendota compared to the underlying water (<1 ng/l). Total DDT (sum of DDT, DDE, DDD) in the foam ranged from 75 to 767 ng/l. Although the resulting fractionation ratios are significantly high, they are subject to some uncertainty due to very low levels in the underlying water.
6.1.4.
Polychlorinated biphenyl (PCB)
Polychlorinated biphenyls are a class of organic compounds with 1e10 chlorine atoms attached to biphenyl, which is a molecule composed of two benzene rings. Due to PCB’s toxicity and classification as a persistent organic pollutant, PCB production was banned by the Stockholm Convention on Persistent Organic Pollutants in 2001. In Lake Mendota the PCBs were detected in the foam samples with fractionation ratios in the range of 100e1000, which is far higher than in the river foam in eastern Tennessee with ratios between 5 and 9 (Eisenreich et al., 1978; Napolitano and Richmond, 1995). In summary, various substances enriched in the SML will be transferred into and concentrated in the foam. The presence of a substance in foam does not necessarily mean that it causes foam. Often substances, e.g. metals are complexed and concentrated in the foam, even though other surface-active compounds are responsible for the foam formation. The effect of the enriched substances on the aquatic environment will be discussed in section 6.3.
6.2.
Occurrence and composition of sea foam
The occurrence of sea foam is a widely known phenomenon around the world (Ba¨tje and Michealis, 1986; Craig et al., 1989; Eberlein et al., 1985; Kesaulya et al., 2008; Seuront et al., 2006). According to Baier et al. (1974) stable sea foams usually include a major silica component, which can be associated with diatom remnants. Ba¨tje and Michaelis (1986) report unusual amounts of sea foam in the North Sea in 1978 caused by a bloom of the planktonic algae Phaeocystis pouchetii. During mass production of this organism the water turns reddish-brown and great amounts of carbohydrates and proteins are released by the
4362
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 5 5 e4 3 6 6
mucilaginous cell colonies at the peak of the bloom and during the breakdown. Through wave action this solution is whisked and washed ashore, where layers of foam cover the beaches up to several meters. In Guzma´n et al. (1990) a severe bloom of Cochlodinium catenatum in the eastern Pacific is reported to be responsible for copious amounts of viscous foams and mucus in the water column. According to Velimirov (1980) many seaweeds exude water-soluble mucilage, which provides enough surfaceactive agents to induce foam formation. A further important contributor to foam formation are broken phytoplankton cells, which release organic matter causing foam. Velimirov’s (1980) analysis of sea foam collected near kelp beds at Oudekraal (South Africa) demonstrated that total protein was the dominant component (21% of the total freeze-dried weight), followed by total lipids (6.1%) and carbohydrates (2.4%). Within the protein fraction the trichloroacetic-acid (TCA)-precipitated protein was the most common (15%), which is an easily metabolizable protein available for consumers (Velimirov, 1980). In a further study, Velimirov (1982) investigated the amount of individual sugar and lipid components in foam near kelp beds. The results indicate a dominance of aldoses and deoxy sugars with b-mannose (32% of total carbohydrates) being the prevalent component followed by b-glucose (19%) and b-galactose (16%). The most important lipid class is represented by triglycerides, which amount to more than 50% of total lipids. Amongst the remaining lipid classes the free fatty acids (22%) and polar lipids (7%) seem worth mentioning. In contrast to foams from rivers and streams the hydrocarbons were only present in trace concentrations or totally absent from foam. Sea foam collected at two sites in New Brunswick was studied by Ba¨rlocher et al. (1988). The authors found phenols to be the major constituent, whereas approximately 13% of the organic carbon content was present in amino acids and carbohydrates. Protein concentration was two to four times higher than that of carbohydrates. Up to 75% of the organic carbon content remained unidentified and could partly be accounted for by lipids. Subsequent investigations by Craig et al. (1989) showed a strong correlation between organic carbon in the sea foam and phenolics. As phenolics are more common in higher plants than in marine algae, the authors argue that vascular plants detritus is the dominant source of the observed sea foam. Their results confirm findings from an earlier study by Coffey (1986), who determined stable C isotope ratios of sea foam and concluded that local Spartina marshes and terrestrial plants are the major contributors of sea foam carbon. In summary, sea foam results mainly from the enrichment of surface-active substances exuded by (i) phytoplankton blooms, (ii) seaweed or (iii) even terrestrial plants in the SML. The enriched material is whisked into foam by the action of waves and washed ashore, where unaestethic foam layers accumulate.
6.3.
Ecological importance of foam
Due to its composition and capacity to accumulate various components, the literature suggests foam has an ecological relevance.
6.3.1.
Toxicity of foam
SMLs and foams are subject to concentrations of materials transported by bubbles. Dissolved and particulate materials adsorbed on a bubble surface are forced into intimate contact as the bubble rises resulting in surface coagulation. This process provides a mechanism for the conversion of dissolved and colloidal materials into particulate form and can produce aggregates rich in surface-active toxins. Foams enriched with those surface-active aggregates represent regions in which organisms experience accelerated rates of accumulated toxic material (Johnson et al., 1989). According to Napolitano and Richmond (1995) various pollutants might be concentrated in the SML. As a consequence, the neustonic organisms (representing the biocoenosis of the SML) could be exposed to far higher concentrations of contaminants in the SML and the resulting foam than those in underlying waters. Sea foam collected in New Brunswick was shown to have a toxic effect on the amphipod Corophium volutator (macroinvertebrate) which was attributed to either the phenolic content of the foam or the levels of heavy metals and pesticides (Craig et al., 1989). Eberlein et al. (1985) report that sea foam produced by a Phaeocystis pouchetii bloom contained acrylic acid which is known to be an antibiotic. Guzma´n et al. (1990) highlight the suffocation of scleractinian corals by mucilagic substances causing sea foam in the eastern Pacific during a bloom of dinoflagellates. Their results indicate that the production of the polysaccharides may be an indirect mechanism of mortality. Harmful effects caused by an external coating of marine birds with a proteinaceous foam derived from a red tide bloom is reported by Jessup et al. (2009). Although this red tide bloom was ostensibly nontoxic, the seabird feathers dipped in the foam lost their normal water repellence and became soaked resulting in feather fouling, reduced mobility and hypothermia. The breaking of foam bubbles carrying surface-active pollutants produce aerosols, which become widely distributed and may cause an increased risk of human exposure to toxins and pathogens (Johnson et al., 1989; Maynard, 1968). Eisenreich et al. (1978) suggested that foam was both a sink for inorganic and organic material and a source of chemical input to the atmosphere by bubble breaking and wind-suspension processes. Furthermore the authors highlighted the crucial role of foam in the transfer of toxic pollutants into the food web, as bacteria and plankton abundant in the foam are ingested by fish and waterfowl are commonly observed to feed on foam. Finally, the enrichment of metals, such as Zn, Pb, Cu, Fe, Mn and chlorinated hydrocarbons can cause an environmental problem, if these foams are transported to drinking water supplies (Harden and Williams, 1989).
6.3.2.
Foam as food resource
The enrichment of foam with various substances is not only a threat to the aquatic biocoenosis, but also represents a potential food resource for organisms either living in the sea or at the shoreline (Ba¨rlocher et al., 1988; Craig et al., 1989; Velimirov, 1980, 1982). The calorific content for freshly formed marine foam near Kelp beds is reported to be 15 kJ/g ash-free dry weight, which
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 5 5 e4 3 6 6
4363
demonstrates the importance of foam as potential energy source for the marine environment Velimirov (1980). According to the author the mucilage excreted by the kelp provides enough surface-active agents to induce foam formation and to improve the foam stability. Long lasting foams are a prerequisite for sufficient movement of the foam to various sites in and around the kelp beds, enabling the energy pool to be available to the consumer chain. Although mucilage consisting mainly of the sugar mannitol, which is likely to be an important component of the foam, none of the foam samples investigated by Velimirov (1982) contained quantifiable amounts of this sugar type. One explanation for the lack of mannitol could be the presence of heterotrophic organisms in the foam, which rapidly utilize the released sugar. Ba¨rlocher et al. (1988) and Craig et al. (1989) investigated the capacity of sea foam to serve as a food source for the amphipod Corophium volutator. In laboratory experiments they proved that C. volutator has a wide variety of enzymes to digest sea foam. The released sugars and amino acids could potentially satisfy 70% of the nutritional requirements of the amphipod. However, the large spectrum of fatty acids in foam samples, suggest that sea foam concentrates and distributes essential dietary components which most consumers are unable to synthesize (Velimirov, 1982). River foam collected from five western North American rivers contained the Bradford-reactive soil protein (BRSP), a glycoprotein of soil and of arbuscular mycorrhizal fungal origin, which was also present in alluvial soils (Harner et al., 2004). Laboratory experiments proved that the protein can be leached and washed from the soil and accumulate in the foam due to its hydrophobic properties. The authors speculate that BRSP might act as a nutrient source for aquatic food webs by contributing carbon and nitrogen from terrestrial sources to water via erosion and leaching of floodplain soils containing the protein.
various aquatic habitats (freshwater and marine) and found heavy concentrations of diatoms, lesser numbers of dinoflagellates and green and blue-green algae. An outstanding fact derived from her study was that the majority of species present in foam was either benthic (attached to sediments) or periphytic (attached to plants e aufwuchs) rather than planktonic (free floating in the water column). She considered foam to be an important habitat which has been ignored previously.
6.3.3.
7.
Foam as habitat
Foam is not only a food resource for organisms, but also acts as an important habitat for various species. Regardless of their source, coastal bio foams have been described as enclosing marine metazoan fauna of several taxa and larvae, e.g. polychaete, mussel and crustacean (Castilla et al., 2007). Eberlein et al. (1985) reported massive occurrences of saprophytic bacteria on sea foam collected in the German Bight. Wilson (1959) demonstrated that sea foam contained 25% solid matter, much of which consisted of bacteria and diatoms. Velimirov (1980) found high densities of bacteria in foam with significantly higher bacterial densities in old foam than in fresh foam. Relatively large amounts of phospholipids measured in river foams indicate that bacteria, protozoa and/or algae are major contributors to organic matter in foams (Napolitano and Richmond, 1995). In studies dealing with aquatic hyphomycetes, also known as amphibious fungi, foam samples show that hyphomycetes’ spores are stored in foams in a viable, ungerminated state for considerable periods of time (Harrington, 1997; Pascoal et al., 2005). The most comprehensive investigation on foam as habitat was a study by Maynard (1968) who collected foam from
6.3.4.
Other roles of foam in the environment
Ettema et al. (1989) discussed the role of foam in the initiation of ice covers. The observations reported indicate that small and medium rivers convey chemicals resulting either from the decay of organic substances or from manmade pollution that generates foam. The resulting foam possibly causes ice covers to form more rapidly than in streams without foam formation and at water temperatures above freezing temperature. This is particularly likely in small rivers where complete ice-cover reduces air supply resulting in decreasing oxygen concentrations. However, the importance of foam as an ice-cover initiation mechanism diminishes with increasing size and width of a river, as wind fetch increases leading to foam dispersal. Apart from the physical effects, foam is known to ensure egg retention and larval development in organisms (Castilla et al., 2007). According to the authors, foam nests enhance the fertilization success and the retention of eggs and larvae in freshwater nest-building fish, tropical aquatic frogs and tunicates. In summary, the formation of foam either in freshwater systems or as sea foam is not imperatively disadvantageous for the environment. Although foams may show toxic effects depending on the substances enriched within, they also can serve as a food resource and as a habitat for organisms.
Conclusions
This review on foam in the aquatic environment has highlighted that minor amounts of information are available in literature, and that most studies dealing with that subject are at least twenty to thirty years old. Several significant references on surface microlayers (SML) in freshwater and marine ecosystems were identified. SML represent the air-water boundary where surface-active components accumulate. As a consequence of the SML, the introduction of air and other gases into the water by turbulences, cascades etc. will lead to foam formation. In the presence of organic (living or dead organisms) and inorganic (silt, sand) particles very stable foams can occur. Foam formation is observed in nearly every aquatic environment, such as rivers, lakes and oceans. Although the majority of studies show that foam is the product of natural processes and factors, the public tends to associate foam formation with manmade pollution. Public concern is likely to be enhanced due to the visibility of foam, which lead to it being more obvious than “hidden” chemical pollution. Surface-active components causing foam include the degradation products of organic material (e.g. humic substances), lipids and proteins originating from aquatic
4364
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 5 5 e4 3 6 6
plants or the terrestrial environment, and exudation or decomposition products of phytoplankton. Manmade foam is often the result of point source pollution (particularly from industry) or diffuse pollution originating mainly from agriculture. There is no regulation regarding instream foam formation in the United States, nor the EU or Switzerland. Although emissions from point sources are strictly governed in general, the foam causing potential of point sources is not regulated. Methods to assess foaming and foam stability are found mainly in the wastewater sector. In wastewater treatment plants stable foam is mainly observed as a result of filamentous bacteria such as Microthrix parvicella or nocardioform actinomycetes which occur in the water. The role of the bacteria in foam formation is on the one hand the production of surface-active biotensides, and on the other hand the stabilisation of foam due to their hydrophobicity. Apart from the aesthetic aspects of foam, foam formation involves several other ecological aspects. Due to its enrichment capacity, carbon, nutrients, metals, hydrocarbons and even pesticides accumulate in foam. According to Napolitano and Richmond (1995) foams are relatively isolated microenvironments, in which the inhabiting organisms are exposed to higher concentrations of contaminants. Toxic effects of foam were demonstrated for the amphipod Corophium volutator and the transfer of toxic substances from foam into organisms via the food web seems to be a potential danger. The break of foam bubbles carrying surface-active pollutants may even cause an increased risk of human exposure to toxins and pathogens. Foam is also believed to be an important food resource and a site of nutrient recycling which transfers energy to the consumer level. Some literature is available on the importance of foam as a habitat, especially for benthic and periphytic organisms. The review of the existing literature demonstrated both the crucial role of foam with regards to pollution and the ecological aspects of foam in the aquatic environment. Considering this, it is even more surprising that information on foam in the aquatic environment is still scarce and that this topic appears to have been ignored over the last decade.
Acknowledgement The authors would like to thank the Austrian Science Funds FWF for financial support as part of the Doctoral program DKplus W1219-N22 on Water Resource Systems.
references
Baier, R.E., Goupil, D.W., Perlmutter, S., King, R., 1974. Dominant chemical composition of sea-surface films, natural slicks and foams. Journal de Recherches Atmospheriques 8, 571e600. Ba¨rlocher, F., Gordon, J., Ireland, R.J., 1988. Organic composition of seafoam and its digestion by Corophium volutator (Pallas). Journal of Experimental Marine Biology and Ecology 115, 179e186.
Ba¨tje, M., Michealis, H., 1986. Phaeocystis pouchetii blooms in the East Frisian coastal waters (German Bight, North Sea). Marine Biology 93, 21e27. Binks, B.P., 2002. Particles as surfactantsesimilarities and differences. Current Opinion in Colloid & Interface Science 7, 21e41. Blackall, L.L., Harbers, A.E., Greenfield, P.F., Hayward, A.C., 1991. Activated sludge foams: effects of environmental variables on organism growth and foam formation. Environmental Technology 12, 241e248. Blackall, L.L., Marshall, K.C., 1989. The mechanism of stabilization of actinomycete foams and the prevention of foaming under laboratory conditions. Journal of Industrial Microbiology & Biotechnology 4, 181e187. Bundesministerium fu¨r Land- und Forstwirtschaft, U.u.W, 1991. Allgemeine Abwasseremissionsverordnung Republik ¨ sterreich. O Bundesministerium fu¨r Land- und Forstwirtschaft, U.u.W, 2006. ¨ sterreich, R. (Ed.), 479. Verordnung In: O ¨ V. Gewa¨sserzustandsu¨berwachungsverordnung GZU Bundesministerium fu¨r Land- und Forstwirtschaft, U.u.W, 2007. ¨ nderung der AEV Gerberei, 3 pp. 61. Verordnung - A Buzzacchi, M., Schmiedel, P., von Rybinski, W., 2006. Dynamic surface tension of surfactant systems and its relation to foam formation and liquid film drainage on solid surfaces. Colloids and Surfaces A: Physicochemical and Engineering Aspects 273, 47. Carstensen, J., 2007. Statistical principles for ecological status classification of water framework directive monitoring data. Marine Pollution Bulletin 55, 3e15. Castilla, J.C., Manriquez, P.H., Delgado, A.P., Gargallo, L., Leiva, A., Radic, D., 2007. Bio-foam enhances larval retention in a freespawning marine tunicate. Proceedings of the National Academy of Sciences of the United States of America 104, 18120e18122. Coffey, W.L.P., 1986. Organic Matter in Seafoam. Mount Allison University, Sackville, New Brunswick. Council Directive 91/271/EEC, 1991. Concerning Urban Wastewater Treatment. European Council, 16 pp. Craig, D., Ireland, R.J., Ba¨rlocher, F., 1989. Seasonal variations in the organic composition of seafoam. Journal of Experimental Marine Biology and Ecology 130, 71e80. Craig, V.S.J., Ninham, B.W., Pashley, R.M., 1993. Effect of Electrolytes on Bubble Coalescence. Nature Publishing Group, London (ROYAUME-UNI). Dale, C., West, C., Eade, J., Rito-Palomares, M., Lyddiatt, A., 1999. Studies on the physical and compositional changes in collapsing beer foam. Chemical Engineering Journal 72, 83. Defrain, M., Schulze-Rettmer, R., 1989. Schaumentwicklung in biologisch gereinigtem Abwasser und im Gewa¨sser. Vom Wasser 73, 251e257. Desroy, N., Denis, L., 2004. Influence of spring phytodetritus sedimentation on intertidal macrozoobenthos in the eastern English Channel. Marine Ecology-Progress Series 270, 41e53. Eberlein, K., Leal, M.T., Hammer, K.D., Hickel, W., 1985. Dissolved organic substances during a Phaecystis pouchetii bloom in the German Bight (North Sea). Marine Biology 89, 311e316. Eisenreich, S.J., Elzerman, A.W., Armstrong, D.E., 1978. Enrichment of micronutrients, heavy metals, and chlorinated hydrocarbons in wind-generated lake foam. Environmental Science & Technology 12, 413e417. Elzerman, A.W., 1976. Surface microlayer-microcontaminant interactions in freshwater lakes. PhD thesis. University of Wisconsin, Madison, Wis. Ettema, R., Johnson, J.K., Schaefer, J.A., 1989. Foam-initiated ice covers on small rivers and streams: an observation. Cold Regions Science and Technology 16, 95e99.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 5 5 e4 3 6 6
Federal Water Pollution Control Act, 2002. In: Congress, U.S. (Ed.), Senate and House of Representatives of the United States of America. Figueredo, R.C.R., Sabadini, E., 2003. Firefighting foam stability: the effect of the drag reducer poly(ethylene) oxide. Colloids and Surfaces A: Physicochemical and Engineering Aspects 215, 77. Fisenko, A.I., 2004. A new long-term on site clean-up approach applied to non-point sources of pollution. Water, Air, & Soil Pollution 156, 1. Gewa¨sserschutzverordnung Schweiz, 1998. In: Bundesrat, S. (Ed.). Schweizer Bundesrat. Guzma´n, H.M., Corte´s, J., Glynn, P.W., Richmond, R.H., 1990. Coral mortality associated with dino-flagellate blooms in the eastern Pacific (Costa Rica and Panama). Marine Ecology Progress Series 60, 299e303. Harden, S., Williams, D., 1989. Stable carbon isotopic evidence for sources of particulate organic carbon found in sea foam. Estuaries and Coasts 12, 49e56. Harner, M.J., Ramsey, P.W., Rillig, M.C., 2004. Protein accumulation and distribution in floodplain soils and river foam. Ecology Letters 7, 829e836. Harrington, T.J., 1997. Aquatic hyphomycetes of 21 rivers in southern Ireland. Biology and Environment-Proceedings of the Royal Irish Academy 97B, 139e148. Heard, J., Harvey, E., Johnson, B.B., Wells, J.D., Angove, M.J., 2008. The effect of filamentous bacteria on foam production and stability. Colloids and Surfaces B: Biointerfaces 63, 21e26. Henry, C.L., 2010. Bubbles, Thin Films and Ion Specificity. Department of Applied Mathematics, Research School of Physical Sciences and Engineering. The Australian National University. Hladikova, K., Ruzickova, I., Klucova, P., Wanner, J., 2002. An investtgation into studying of the activated sludge foaming potential by using physicochemical parameters. Water Science and Technology 46, 525e528. Holmberg, K., Jo¨nsson, B., Kronberg, B., Lindman, B., 2003. Foaming of Surfactant Solutions. John Wiley & Sons, Ltd. Ho¨rtnagl, P., Pe´rez, M.T., Zeder, M., Sommaruga, R., 2010. The bacterial community composition of the surface microlayer in a high mountain lake. FEMS Microbiology Ecology 73, 458e467. Hwang, Y., Tanaka, T., 1998. Control of Microthrix parvicella foaming in activated sludge. Water Research 32, 1678e1686. Jessup, D.A., Miller, M.A., Ryan, J.P., Nevins, H.M., Kerkering, H.A., Mekebri, A., Crane, D.B., Johnson, T.A., Kudela, R.M., 2009. Mass Stranding of marine birds caused by a surfactantproducing red tide. PLoS ONE 4, e4550. Johnson, B.D., Zhou, X., Parrish, C.C., Wangersky, P.J., Kerman, B. R., 1989. Fractionation of particulate matter, the trace metals Cu, Cd, and Zn, and lipids in foam and water below Niagara falls. Journal of Great Lakes Research 15, 189e196. Kesaulya, I., Leterme, S.C., Mitchell, J.G., Seuront, L., 2008. The impact of turbulence and phytoplankton dynamics on foam formation, seawater viscosity and chlorophyll concentration in the eastern English Channel. Oceanologia 50, 167e182. Kopplow, O., Barjenbruch, M., 2002. Beurteilung und Weiterentwicklung von Methoden zur Erfassung des Schaumpotentials. Institut fu¨r Kulturtechnik und Siedlungswasserwirtschaft, Universita¨t Rostock. La´nsky´, M., Ruzickova´, I., Bena´kova´, A., Wanner, J., 2005. Effect of Coagulant dosing on phyiscochemical and microbiological characteristics of activated sludge and foam formation. Acta Hydrochimica Et Hydrobiologica 33, 266e269. Lemmer, H., Lind, G., Mu¨ller, E., Schade, M., 2005. Non-famous scum bacteria: biological characterization and troubleshooting. Acta hydrochimica et hydrobiologica 33, 197e202. Lu, K.H., Jin, C.H., Dong, S.L., Gu, B.H., Bowen, S.H., 2006. Feeding and control of blue-green algal blooms by tilapia (Oreochromis niloticus). Hydrobiologia 568, 111e120.
4365
Madrange, L., Chaboury, P., Ferrandon, O., Mazet, M., Rodeaud, J., 1992. Study of the formation and stability of chemical foam on the Vienne river between Limoges and Confolens. Revues des Sciences de L’eau 6, 315e334. Maynard, N.G., 1968. Aquatic foams as an acological habitat. Zeitschrift fu¨r Allgemeine Mikrobiologie 8, 119e126. Mills, M.S., Thurman, E.M., Ertel, J., Thorn Kevin, A., 1996. Organic Geochemistry and Sources of Natural Aquatic Foams, Humic and Fulvic Acids. American Chemical Society. 151e192. Napolitano, G.E., Cicerone, D.S., 1999. Lipids in water-surface microlayers and foams. In: Arts, M.T., Wainmann, B.C. (Eds.), Lipids in Freshwater Ecosystems, pp. 235e262. Napolitano, G.E., Richmond, J.E., 1995. Enrichment of biogenic lipids, hydrocarbons and PCBs in stream-surface foams. Environmental Toxicology and Chemistry 14, 197e201. Paris, S., 2004. Beka¨mpfung von Schwimmschlamm verursacht durch Microthrix parvicella. Lehrstuhl fu¨r Wassergu¨te- und Abfallwirtschaft. Technische Universita¨t Mu¨nchen, Mu¨nchen, 138 pp. Parker, B., Barsom, G., 1970. Biological and chemical significance of surface microlayers in aquatic ecosystems. BioScience 20, 87e93. Pascoal, C., Marvanova, L., Cassio, F., 2005. Aquatic hyphomycete diversity in streams of Northwest Portugal. Fungal Diversity 19, 109e128. Pedersen, K.H., Andersen, S.I., Jensen, A.D., Dam-Johansen, K., 2007. Replacement of the foam index test with surface tension measurements. Cement and Concrete Research 37, 996. Petrovski, S., Dyson, Z.A., Quill, E.S., McIlroy, S.J., Tillett, D., Seviour, R.J., 2010. An examination of the mechanisms for stable foam formation in activated sludge systems. Water Research 45, 2146e2154. Piazza, L., Gigli, J., Bulbarello, A., 2008. Interfacial rheology study of espresso coffee foam structure and properties. Journal of Food Engineering 84, 420e429. Pojasek, R.B., Zajicek, O.T., 1978. Surface microlayers and foamsesource and metal transport in aquatic systems. Water Research 12, 7e10. Poremba, R., 1991. Untersuchung zur Schaumbildung auf Fließgewa¨ssern. Fachhochschule fu¨r Druck, Stuttgart, 59 pp. Pretorius, W.A., Laubscher, C.J.P., 1987. Control of biological scum in activated sludge plants by means of selective flotation. Water Science and Technology 19, 1003e1011. Pugh, R.J., 1996. Foaming, foam films, antifoaming and defoaming. Advances in Colloid and Interface Science 64, 67. Regismond, S.T.A., Winnik, F.M., Goddard, E.D., 1998. Stabilization of aqueous foams by polymer/surfactant systems: effect of surfactant chain length. Colloids and Surfaces A: Physicochemical and Engineering Aspects 141, 165. Ruzicka, K., Gabriel, O., Bletterie, U., Winkler, S., Zessner, M., 2009. Cause and effect relationship between foam formation and treated wastewater effluents in a transboundary river. Physics and Chemistry of the Earth, Parts A/B/C 34, 565. Seuront, L., Vincent, D., Mitchell, J.G., 2006. Biologically induced modification of seawater viscosity in the Eastern English Channel during a Phaeocystis globosa spring bloom. Journal of Marine Systems 61, 118. Shi, H.X., Qu, J.H., Wang, A.M., Li, G.T., Lei, P.J., Liu, H.J., 2005. Primary investigation of blue-green algae and their toxins from water blooms in Guanting Reservoir of Beijing. Chemical Journal of Chinese Universities-Chinese 26, 1653e1655. Szekielda, K.-H., Kupperman, S.L., Klemas, U., Polis, D.F., 1972. Elemental enrichment in organic films and foam associated with aquatic frontal systems. Journal of Geophysical Research 77, 5278e5282. Tafreshi, A.M., Marzo, M.D., 1999. Foams and gels as fire protection agents. Fire Safety Journal 33, 295.
4366
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 5 5 e4 3 6 6
Torregrossa, M., Viviani, G., Vinci, V., 2005. Foaming estimation tests in activated sludge systems. Acta Hydrochimica Et Hydrobiologica 33, 240e246. Vanrell, G., Canals, R., Esteruelas, M., Fort, F., Canals, J.M., Zamora, F., 2007. Influence of the use of bentonite as a riddling agent on foam quality and protein fraction of sparkling wines (Cava). Food Chemistry 104, 148. Velimirov, B., 1980. Formation and potential trophic significance of marine foam near kelp beds in the benguela upwelling system. Marine Biology 58, 311e318. Velimirov, B., 1982. Sugar and lipid components in Sea foam near kelp beds. Marine Ecology 3, 97e107. Vikingstad, A.K., 2006. Static and Dynamic Studies of Foam and Foam-oil Interactions. Department of Chemistry. University of Bergen, Bergen, 88 pp. Vikingstad, A.K., Skauge, A., Høiland, H., Aarra, M., 2005. Foam-oil interactions analyzed by static foam tests. Colloids and Surfaces A: Physicochemical and Engineering Aspects 260, 189e198.
Water Framework Directive 2000760/EC, 2000. In: Council, E. (Ed.), Directive of the European Parliament and of the Council 2000/60/EC Establishing a Framework for Community Action in the Field of Water Policy. European Union, Luxembourg. Wegner, C., Hamburger, M., 2002. Occurrence of stable foam in the Upper Rhine River caused by plant-derived surfactants. Environmental Science & Technology 36, 3250e3256. Wilson, A.T., 1959. Surface of the ocean as a source of air-borne nitrogenous material and other plant nutrients. Nature 184, 99e101. Wilson, M.I., Robertson, L.D., Daly, M., Walton, S.A., 1995. Effects of visual cues on assessment of water quality. Journal of Environmental Psychology 15, 53. Zhang, H., Miller, C.A., Garrett, P.R., Raney, K.H., 2004. Defoaming effect of calcium soap. Journal of Colloid and Interface Science 279, 539.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 6 7 e4 3 8 0
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Leaching of human pathogens in repacked soil lysimeters and contamination of potato tubers under subsurface drip irrigation in Denmark Anita Forslund a,*, Finn Plauborg b, Mathias Neumann Andersen b, Bo Markussen c, Anders Dalsgaard a a
Department of Veterinary Disease Biology, Faculty of Life Sciences, University of Copenhagen, Groennegaardsvej 15, DK 1870 Frederiksberg C, Denmark b Crop Production Group, Department of Agroecology and Environment, Faculty of Agricultural Sciences, University of Aarhus, Denmark, Blichers Alle´ 20, P.O. BOX 50, DK 8830 Tjele, Denmark c Department of Basic Sciences and Environment, Faculty of Life Sciences, University of Copenhagen, Thorvaldsensvej 40, DK-1871 Frederiksberg C, Denmark
article info
abstract
Article history:
The risk for contamination of potatoes and groundwater through subsurface drip irrigation
Received 7 December 2010
with low quality water was explored in 30 large-scale lysimeters containing repacked
Received in revised form
coarse sand and sandy loam soils. The human pathogens, Salmonella Senftenberg,
11 May 2011
Campylobacter jejuni and Escherichia coli O157:H7, and the virus indicator Salmonella Typhi-
Accepted 11 May 2011
murium bacteriophage 28B, were added weekly through irrigation tubes for one month
Available online 28 June 2011
with low irrigation rates (8 mm per week). In the following six months lysimeters were irrigated with groundwater free of pathogens. Two weeks after irrigation was started,
Keywords:
phage 28B was detected in low concentrations (2 pfu ml1) in leachate from both sandy
Human pathogens
loam soil and coarse sand lysimeters. After 27 days, phage 28B continued to be present in
Leaching
similar concentrations in leachate from lysimeters containing coarse sand, while no phage
Contaminated water
were found in lysimeters with sandy loam soil. The added bacterial pathogens were not
Potatoes
found in any leachate samples during the entire study period of 212 days. Under the study
Subsurface irrigation
conditions with repacked soil, limited macropores and low water velocity, bacterial
Repacked soil lysimeters
pathogens seemed to be retained in the soil matrix and died-off before leaching to groundwater. However, viruses may leach to groundwater and represent a health risk as for some viruses only few virus particles are needed to cause human disease. The bacterial pathogens and the phage 28B were found on the potato samples harvested just after the application of microbial tracers was terminated. The findings of bacterial pathogens and phage 28 on all potato samples suggest that the main risk associated with subsurface drip irrigation with low quality water is faecal contamination of root crops, in particular those consumed raw. ª 2011 Elsevier Ltd. All rights reserved.
* Corresponding author. Tel.: þ45 35 332725; fax: þ45 35 332755. E-mail address:
[email protected] (A. Forslund). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.05.009
4368
1.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 6 7 e4 3 8 0
Introduction
Clean freshwater is a limited resource and its use for crop irrigation is in competition with the demand for household and industrial consumption. Further, water availability is of importance to the external environment, e.g. groundwater sources, as stated by the European Water Framework Directive (EC, 2000). On top of this, the problem with limited clean freshwater resources will be amplified by changes in climate and precipitation patterns reducing groundwater recharge as a consequence of decreased precipitation across Europe (IPCC, 2007). Lack of clean freshwater have already forced agriculture, especially in Central Europe and the Mediterranean area, to search for alternative water sources and irrigation strategies to sustain food production. Even in humid areas irrigated agriculture may foresee reduction in water availability as climate change scenarios forecast a decrease in summer precipitation (IPCC, 2007). Hence, low quality water, e.g. treated/untreated wastewater or surface water run-off, will be increasingly used for irrigation in agriculture. Already today, low quality waters are used to irrigate food crops in Australia, Mediterranean countries and elsewhere, e.g. Israel has for decades used treated wastewater in irrigated agriculture (Lazarova et al., 2000). Irrigation of agricultural land with low quality water, in particular subsurface irrigation, can potentially also lead to contamination of groundwater when irrigation water contains high numbers of faecal microorganisms and human pathogens like Salmonella, Campylobacter, Shigella, enteric viruses, and protozoan parasites (Calci et al., 1998; Nwachuku and Gerba, 2008; U.S. EPA, 2004). Waterborne illness associated with consumption of contaminated groundwater is common in United States and Europe, but often these outbreaks are related to faecal contamination in the distribution system or from surface run-off water, e.g. contamination of wells (Abbaszadegan et al., 2003; Craun et al., 2006; Kramer et al., 2001). It is unknown to what extent groundwater aquifers are contaminated due to irrigation with faecal contaminated water and the subsequent transport of pathogens through the soil to the groundwater. Water scarcity requires different measures to save water and increase productivity in irrigated agriculture. Therefore, there is a need for water-saving irrigation practices to be explored. The efficiency of crops to take up water is significantly increased by the use of subsurface drip irrigation, mainly due to reduced soil evaporation, but also because the requirements of plants for water can be met more precisely (Ayars et al., 1999; Shahnazari et al., 2007). In Denmark, surface drip irrigation is presently used for irrigation of onethird of strawberries fields and more than 50% of apple and pear orchards (approx. 3000 ha), but farmers’ advisory service has recently initiated experiments on subsurface drip irrigation of potatoes and lettuce as quotas on irrigation water are under implementation. However, subsurface soil application of treated wastewater, which often still contains human pathogens, may potentially increase pathogen survival by preventing their exposure to the harmful effects of UV-light and desiccation. Pathogens in protected soil environments may subsequently be transferred to root crops and could
therefore pose a food safety risk for consumers, in particular when such products are consumed raw, e.g. radishes and other vegetables (Natvig et al., 2002). In 2006, illness associated with the consumption of fruits and vegetables due to contamination with bacteria and viruses accounted for 8% of reported cases of illness in United States (CDC, 2009). The farm environment, including irrigation water, has been the likely source of contamination with Escherichia coli O157:H7 (So¨derstro¨m et al., 2008; Wendel et al., 2006) and hepatitis A (Herna´ndez et al., 1997) in disease outbreaks associated with consumption of spinach and lettuce. The survival of microorganisms in soil depends on parameters such as temperature, moisture content, pH, soil composition and inhibitory competition from the indigenous microflora (Abu-Ashour et al., 1994; Chu et al., 2003; Mawdsley et al., 1995), as well as the time the microorganisms are able to survive outside a natural host. Pathogen numbers will show a temporal decrease even at low temperatures, if the conditions are unfavorable (Maule, 1999). Survival of bacterial pathogens in soil have been reported for up to one month after they were applied to grassland soils (Nicholson et al., 2005), while several studies have reported prolonged survival of viruses in soil (Feachem et al., 1983; Rzezutka and Cook, 2004). In the present study human bacterial pathogens were studied rather than faecal indicators like thermotolerant coliforms and E. coli as few studies have examined their fate and transport in natural soil system (Bech et al., 2010). This was done as differences in cell surface, but also other properties of microorganisms may affect their transport through and survival in soil and water (Castro and Tufenkji, 2007; Long et al., 2009). Bolster et al. (2006) observed a greater transport of Campylobacter jejuni compared to E. coli. Even though E. coli is used as an indicator for the presence of pathogenic bacteria, Salmonella has shown better survival in the soil environment compared to E. coli (Winfield and Groisman, 2003). Bacteriophages have been suggested as model organisms for virus transport to predict human enteric viral behavior and risks for their environmental transmission (Havelaar, 1991). Salmonella Typhimurium bacteriophage 28B was used in this study and it has previously been used as a surrogate for human enteric viruses, like adenovirus and rotavirus (Leclerc et al., 2000). The current investigation was carried out in repacked soil lysimeters with coarse sand and sandy loam soils where potatoes were irrigated with water spiked with the microbial tracers Salmonella, Campylobacter, pathogenic E. coli and bacteriophage 28B (a virus indicator). The objective of the study was to determine the occurrence of the microbial tracers on potatoes and in leachate following subsurface drip irrigation with artificially contaminated water at low irrigation rate.
2.
Materials and methods
2.1.
Study site and climatic conditions
The study was carried out in Jutland, Denmark at the Research Centre Foulum (56 300 N, 9 350 E). The irrigation with microbial tracers was initiated in August 2007 and the study terminated in March 2008. Details on the irrigation strategy and frequency
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 6 7 e4 3 8 0
are provided in section 2.4. The climate is temperate with an annual average rain fall of approximately 800 mm. Weather data were collected at the local climate station located 500 m from the study site (Fig. 1). Reference evapotranspiration was calculated using a modified Makkink formula (Aslyng and Hansen, 1982). Total precipitation was 372 mm during the study and temperatures were 2 C higher than normal in January and February. Groundwater used for irrigation was received from Research Centre Foulum Waterworks. The analytical report from the local Waterworks showed that the used groundwater had pH 7.9, electrical conductivity of 36 mS m1, sodium concentration of 9.1 mg L1 and sodium adsorption ratio (SAR) of 0.31 meq L1. Concentrations of E. coli and coliform bacteria were below detection limit of 1 cell per 100 ml in the groundwater.
2.2.
Facility and experimental lay-out
The experiment was carried out in large concrete lysimeters measuring 2.70 m 1.60 m with a soil depth of 1.40 m totaling 6.05 m3. The lysimeters were accessible from underground basement corridors where leachate was collected (Fig. 2). During the experiment a mobile roof covered the lysimeters during periods of rain to prevent unregulated irrigation of the soil. Volumes and frequency of irrigation of the individual lysimeters was controlled by a computer. The lysimeters were irrigated with groundwater by subsurface drippers to reduce soil evaporation. Potato plants of the cultivar Folva were planted on April 17, 2007 and each lysimeter contained 20 potato plants distributed equally (Fig. 3). The top of potato plants was removed just before irrigation with microbial tracers was initiated on August 1, 2007 to prevent further evaporation and transpiration from the crop. Tubers were harvested on August 28, 2007. The experiment included 30 lysimeters of which 15 contained coarse sand and 15 sandy loam soil.
2.3.
Soil types and irrigation system
Two different soil types, coarse sand (Orthic Haplohumod, coarse sand, siliceous, mesic, pH 7.20, mean grain size (d50)
4369
240 mm) and sandy loam soil (Typic Agrudalf, loam, mixed, mesic, calcareous, pH 7.15, d50 27 mm), typical for Danish agriculture and climatic conditions were studied (Table 1). Textural composition was determined after removal of C by hydrogen peroxide and CaCO3 by hydrochloric acid. Samples were dispersed in sodium pyrophosphate solution and clay and silt contents were measured with a hydrometer. Sand fractions were separated using wet-sieving technique. Organic matter (OM) content was determined by combustion of a sample in O2, followed by IR-measurement of the evolved CO2, and using a conversion factor of 1.724 between OC and OM contents. Saturated hydraulic conductivity was measured by the constant-head method on 20 20 cm (diameter length) undisturbed soil cores and bulk density on 100 cm3 core-samples dried at 105 C. Both soil types had been used for agricultural production for many years before placement in the lysimeter in 1993. During the establishment of the lysimeters, each soil type was separated into three diagnostic horizons, which were homogenized and vibrated back to their original dry bulk density (Nielsen and Møberg, 1985). In the previous 14 years, the lysimeters have been grown with cereal crops, including oat and barley. In the spring, the tubers were ridged with 15-cm of soil and the drip lines placed on the soil. In each lysimeter four drip lines (Netafim, Tel Aviv, Israel) each containing six drip emitters were installed halfway between seed potatoes, i.e., 15-cm from each potato plant (Fig. 3). The drip emitters could deliver 1 L h1. Drip lines were finally ridged with 10-cm soil giving a total height of the ridge of 35 cm. The drip lines in each lysimeter were regulated individually by the irrigation system.
2.4.
Irrigation strategy and frequency
All lysimeters were irrigated with groundwater during the potato-growing season by subsurface drip irrigation. On July 31, soil water deficits were measured in all lysimeters by Time Domain Reflectometry (Ahmadi et al., 2009) and rewetted to field capacity according to the soil water deficit of individual lysimeters by application of 10e40 mm of irrigation water. The
Fig. 1 e Daily average temperature, precipitation and reference evapotranspiration during the study period.
4370
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 6 7 e4 3 8 0
Fig. 2 e The cross-section of the lysimeters shows the four rows which each consist of 25 lysimeters filled with the same soil type. Leachate samples were collected through the drainage pipe in the basement corridor.
volumes and frequency of irrigation applied were estimated to reflect average rain fall conditions for the study period (Fig. 4). From August 1e31, 8 mm irrigation water was applied weekly on the first day of the week. From September 1 to November 19, 3 mm irrigation water was applied monthly to create a slow drainage reflecting conditions during late summer. Then initiated on November 28, 6 mm irrigation water was applied daily to create a steady state water flow and leaching at a soil water status close to saturated water content. Groundwater was not applied to lysimeters from December 16 to January 5 due to frost in irrigation tubes. The occurrence and survival of microbial tracers in leachate from lysimeters were studied during 212 days. During the first 28 days, irrigation water containing microbial tracers was applied at different frequencies as described below. For the remaining study period, only groundwater was used for irrigation. Lysimeters received in triplicate between
1e4 applications with groundwater added microbial tracers during the first month of the study period. All 24 lysimeters received one application four weeks before harvest (August 1); 18 lysimeters received another application three weeks before harvest (August 8); 12 lysimeters received a third application two weeks before harvest (August 15); and six lysimeters received a fourth application one day before harvest (August 27). The six control lysimeters were irrigated with groundwater without microbial tracers (Table 2). With the chosen irrigation strategy, the distribution of microbial tracers would cover a depth of approximately 32 mm if macropore flow did not occur. The four applications with microbial tracers were done with a total of 30 L applied to each lysimeter at a flow rate of 5.5 mm h1. The microbial tracers were added continuously to the irrigation water through a dispenser pump during the pulse of irrigation. Details on the preparation of the microbial
Drip line
30 cm 17 cm
75 cm
7 cm
35 cm
8 cm
Field level (0 cm)
Potato Potato plants
Drip lines
Drip emitters
Fig. 3 e Location of potato plants in a lysimeter with drip lines (left) and cross-section of potato ridge system with drip lines installed (right).
4371
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 6 7 e4 3 8 0
Table 1 e Physical and chemical properties of soil types in the lysimeters. Soil type
Coarse sand
Sandy loam soil
Horizon
Ap Bhs C Ap EB Cg
Depth
Organic Claya Silta2e20 mm Fine sanda Coarse sanda 20e200 mm 200e2000 mm matter <2 mm
(cm)
(%)
(%)
(%)
(%)
(%)
0e30 30e70 70e140 0e30 30e70 70e140
1.9 0.7 0.2 2.3 0.5 0.3
5.8 5.9 5.2 17.6 21.6 21.7
2.1 0.5 0.7 12.9 13.4 15.8
17.8 14.1 19.0 48.0 43.9 40.2
72.3 78.6 74.9 19.2 20.6 22.0
Saturated Bulk Total hydraulic density porosity conductivity (%) (mm/h) (g/cm3) 212 242 89 95 76 36b/10c
1.41 1.46 1.50 1.44 1.53 1.55
47.0 45.0 44.0 47.0 44.0 43.0
a The size classes of primary particles are according to the Danish Soil Classification (Landbrugsministeriet, 1976). b Depth 60e90 cm. c Depth 90e140 cm.
tracer solution are described in section 2.7. After the last application of microbial tracers, the potatoes were harvested on August 28. Monthly irrigation with groundwater (3 mm, 13 L per lysimeter) continued for 11 weeks. Since only a few leachate samples contained low concentrations of the phage 28B, the frequency of irrigation was increased to 6 mm (26 L per lysimeter) daily for a three month period. This strategy was chosen to have almost the entire pore volume exchanged in the lysimeters during the study period. During this last irrigation period, water samples collected every second and fourth week were analyzed. This strategy allowed an assessment of both the transport of pathogens and phage in leachate and food safety aspects associated with their occurrence on potatoes.
2.5.
Microbial tracers
The human pathogenic bacteria C. jejuni (C. jejuni) (NCTC 11168), E. coli serotype O157:H7 (E. coli O157:H7) (ATCC 43888, nontoxigenic) and Salmonella enterica serotype Senftenberg 775W (S. Senftenberg) (resistant to nalidixic acid) were used in this study. Salmonella Typhimurium bacteriophage 28B
(Lilleengen, 1948) was included as a model organism for virus transport. Salmonella Typhimurium bacteriophage 28B (hereafter referred to as phage 28B) has a documented resistance toward high temperatures, changes in pH and high NH3-levels (Vinnera˚s et al., 2003). Phage 28B has not been shown to occur in environmental samples or in faeces and has been used to model the transport of viral contaminants in groundwater resources (Carlander et al., 2000; Johansson et al., 1998) as well as an indicator of viral survival (Vinnera˚s et al., 2008). Initial laboratory analysis using plaque assay method (Adams, 1959) confirmed that phage 28B did not infect S. Senftenberg 775W.
2.6.
Sampling of leachate, soil and potatoes
Initially, leachate and soil samples from all lysimeters were analyzed for the presence of the three human bacterial pathogens and phage 28B before irrigation with the microbespiked groundwater. During the study, all leachate from the individual lysimeters was collected in sterile 30-L bottles, which were weighed and mixed well before a subsample of 300 ml was taken for microbial analyses. The 30-L bottles were emptied, cleaned and sterilized after each sampling time.
Fig. 4 e Volume of accumulated irrigation water and leachate from sandy loam soil and coarse sand lysimeters collected during the study period. P is average precipitation in Denmark from 1961 to 1990 shown as accumulated monthly values. A shows date of application of microbial tracers in irrigation water.
4372
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 6 7 e4 3 8 0
Table 2 e Application of microbial tracers with irrigation water during the study period (August 2007eMarch 2008). Number of applications
Number of lysimeters
Details on the application of microbial tracers
0
6
Control lysimeters. Groundwater was used for irrigation during the study.
1
6
2
6
3
6
4
6
Microbial tracers added to the irrigation water at the start of the study (first day). Microbial tracers added to the irrigation water at the start of the study and again one week after. Microbial tracers were added to the irrigation water at the start of the study and in the second and third week. Microbial tracers were added to the irrigation water at the start of the study in addition to once in the second, third and fourth week.
Type of samples
Total number of samplesa
Duration of sampling (days)
Leachate Soil Potatoes Leachate
72 12 12 72
212 28 28 212
Leachate
24
41
Leachate
24
41
Leachate Soil Potatoes
72 12 12
212 28 28
a Each application was done on both sandy loam soil and coarse sand with three replicate lysimeters per soil type.
During the first six weeks leachate from all lysimeters were analyzed weekly. Thereafter, leachate from lysimeters that had received microbial tracers one and four times, as well as from control plots, was analyzed every second and fourth week (Table 2). The 300-ml subsample was stored at 5 C and analysis initiated within a maximum of 12 h after collection. Each microbial tracer was enumerated in triplicate one ml subsample by the methods described below. At harvest, soil and potato samples were collected from control lysimeters and from lysimeters that had received four applications of microbial tracers with the last application done one day before harvest. Pooled samples of six 30e50 g representative soil samples were collected on the ridge near the drip emitter from each lysimeter. Soil samples were stored for 11 weeks at 5 C and then analyzed for the presence of bacterial pathogens and phage 28B. Potatoes were collected from two different potato plants in each lysimeter and the surface of the potato, including any soil attached to the potato, was analyzed for bacterial pathogens and phage 28B. Potato samples were stored at 5 C and analyzed within three weeks of storage for bacterial pathogens and phage present on the surface of the potatoes including soil adhering to the potato.
2.7.
Preparation of microbial tracers for irrigation
A microbial tracer solution of 200 ml Maximum Recovery Diluent (MRD) (CM0733; Oxoid, Basingstoke, England) containing each of the microbial tracers was prepared for each lysimeter to be added to 30-L of irrigation water. E. coli O157:H7, C. jejuni and S. Senftenberg 775W were grown over night at 37 C in Brain Heart Infusion broth (CM1135; Oxoid). For growth of S. Senftenberg, 40 mg ml1 nalidixic acid (N-8878-5G; Sigma, Germany) was added. C. jejuni was incubated microaerofilic (5% O2, 10% CO2, and 85% N2) generated by CampyGenenvelope (Oxoid) in anaerobic jars. Concentrations of bacterial tracer organisms were measured by OD600 and diluted to OD600 1.0 with Maximum Recovery Diluent (Oxoid) giving a concentration of 5 106 colony forming units (cfu) ml1. A maximum of two ml growth medium containing the individual microbial tracer was added to the microbial tracer solution. Preparation
of a stock solution containing phage 28B was done by infecting Salmonella Typhimurium type 5 host strain with phage 28B as described previously by Ho¨glund et al. (2002). Four ml of phage 28B stock solution with a concentration of 5 1010 plaque forming units (pfu) ml1 was additionally added to the microbial tracer solution. This microbial tracer solution was then transported in a cooling box containing cooling elements to the field site. A dispenser pump continuously mixed the microbial tracer solution with the irrigation water (total volume of 30-L) during the pulse of irrigation. With this high level of dilution of the growth medium it was assumed that the low concentration of growth medium would not significantly affect survival and transport of the microbial tracers. During each of the four applications of microbial tracers, the total numbers of the individual bacterial pathogens and phage 28B applied to each lysimeter were 1 109 cfu and 2 1011 pfu, respectively. The concentrations of bacterial pathogens and phage 28B in irrigation water were accordingly 3 104 cfu ml1 and 7 106 pfu ml1, respectively. It should be noted that these concentrations are 3e5 Log10 higher that those found for human enteric viruses (500e700 pfu L1) in domestic sewage (Oron et al., 1995) and for Salmonella in wastewater (few cells to 8000 cfu per 100 ml) (Bitton, 2005). The present study would therefore represent a worst case scenario.
2.8.
Enumeration of microbial tracers
Leachate samples were 10-fold serial diluted in MRD (Oxoid) and one ml sample was analyzed. The analysis of leachate was done in triplicates. E. coli O157:H7, S. Senftenberg and C. jejuni in leachate were enumerated by direct plating on selective agar plates with a detection limit of 1 cfu ml1. Enumeration of phage 28B was done by the double-layer-agar method and the detection limit was 1 pfu ml1. Samples of 10 g of soil was added to 90 ml MRD (Oxoid) and treated in ultra sound bath (Metason 60; Struers, Copenhagen, Denmark) for 30 s to remove microorganisms from the surface of the soil particles. The soil solution was subsequently serially diluted before enumeration. Detection limit for microbial tracers in soil samples was 10 cfu g1 and 10 pfu g1, respectively. Approximately 600 g potatoes
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 6 7 e4 3 8 0
(5e7 potatoes) were analyzed for the concentration of microbial tracers attached directly to the surface of potatoes. This included tracers present in the soil attached to the surface of the potatoes. To the plastic bag containing the potatoes there was added 200 ml MRD (Oxoid). The plastic bag was subsequently gently massaged for 30 s by hand to release the microbial tracers and the solution was serially diluted before enumeration of the microbial tracers. Concentration of microbial tracers on potato surfaces was express as units per surface area (cfu per cm2 or pfu per cm2). The length and equatorial diameter of the potatoes was measured and the surface area calculated according to Eqs. (1) and (2) in section 2.9. E. coli O157:H7 was enumerated on Sorbitol MacConkey Agar (1.09207; Merck, Darmstadt, Germany) added CTsupplement (1.09202; Merck). After incubation at 37 C for 24 h colorless colonies were counted and confirmed by agglutination with E. coli O157 diagnostic antiserum (Statens Serum Institut, Copenhagen, Denmark). S. Senftenberg 775W was enumerated on MacConkey Agar (1.05465; Merck) added 40 mg ml1 nalidixic acid (Sigma). Agar plates were incubated at 37 C for 24 h and colorless colonies were counted and verified by agglutination with Salmonella diagnostic antiserum (Statens Serum Institut). C. jejuni was isolated on Campylobacter Blood-free Selective Agar (CM0739; Oxoid) with supplement (SR0115E; Oxoid). Incubation was done in microaerobic atmosphere conditions (5% O2, 10% CO2 and 85% N2) (CampyGen CN0025; Oxoid) at 37 C for 48 h and gray colored colonies were counted. Colonies were further studied under microscope and were confirmed as Campylobacter if typically curved motile bacteria were present. Phage 28B was enumerated by a double-agar layer method (Adams, 1959). The host strain Salmonella Typhimurium Type 5 was grown in Nutrient broth (CM67; Oxoid) at 37 C for 4 h. From the 10-fold diluted samples, one ml was taken and mixed with one ml broth culture of the host strain and three ml of soft agar consisting of 70% Blood agar base (CM55; Oxoid) and 30% Nutrient broth (Oxoid). The mixture was spread on a well-dried Blood agar base plate (Oxoid) which was incubated at 37 C for 18 h. Clear zones (plaques) were counted as plaque forming units (pfu). When a high bacterial background flora was expected (mainly at the lower dilutions), the samples were filtered through 0.45 mm pore size filters (Minisart; Sartorius, Goettingen, Germany) before mixed with the soft agar.
2.9.
where C0 is the concentration of the phage added to the irrigation water, Cmax is the maximum concentration of the phage 28B detected in the leachate (peak concentration) and h is the height of lysimeter. In situations where quantitative measurements have many non-detectable results, special methods are required in the statistical analysis. In this paper we have combined a logarithmic regression on the binary outcomes, i.e. non-detected vs. detected, with a normal regression on the logarithm of the positive measurements. The probability p that a measurement is above detection limit is modeled with a linear model in log( p), i.e. the logarithm is used as link function. When a measurement is above detection limit the actual measurement x is modeled with a normal regression in log(x). For measurements below detection limit, this additional information is not available. The two models can be combined by multiplying their likelihood functions, and likelihood ratio test can be performed. This approach has several benefits. Firstly, all information is used in a single analysis hence enhancing the statistical power. Secondly, using the logarithm both for the binary and for the normal regression has the advantage that the parameters in these two models have the same biological interpretation, namely in terms of relative differences between contrasted subgroups. Hence it also makes sense to test whether the regression parameters are the same in the two models. The intercept parameters model the scales in the two models, and there is no reason for these to be the same. Finally, the populations mean C/C0 was estimated using the formula l ¼ logðC=C0 Þ=h ¼ logðPðX > 0Þ meanðXjX > 0ÞÞ 1=h ¼ logðPÞ=h logmeanðXjX > 0Þ=h where log meanðXjX > 0Þ is found using the logarithmic normal distribution. Since the right hand side of the above formula is linear in the model parameters confidence intervals on the populations mean can be found. The concentration (C ) of microbial tracers in leachate, soil and on potatoes were normalized with the concentration of the microbial tracers added to the irrigation water (C0). All the statistical model testing was completed in the statistic computer programme SAS, version 9.2 (SAS Institute Inc., Cary, USA).
3.
Data analyses and statistical design
4373
Results
3 1=2 3 p Apotato ¼ 2 ð2p=3pÞ D2 =4 þ p2
(1)
Initial analyses of the soils and leachate from each of the 30 lysimeters before microbial tracers were added showed no presence of E. coli O157, Campylobacter spp., Salmonella spp. or phage 28B. Leachate, soil and potatoes from control lysimeters irrigated solely with groundwater did not contain bacterial pathogens and phage 28B.
p ¼ D2 =4H
(2)
3.1.
The surface area of the potato was assumed to be double paraboloid and was calculated according to Eqs. (1) and (2) (Battilani, 2010):
where D is the potato equatorial diameter and H is the length of potato. The removal rate, l, was calculated according to Eq. (3) (Pang, 2009): l ¼ log10 ðCmax =C0 Þ=h
(3)
Microbial tracers in leachate
Volumes of irrigation water and collected leachate were similar during the study period (Fig. 4). This was in particular due to a low evapotranspiration as a consequence of low temperature and removal of potato plants (Fig. 1). The total amount of groundwater used for irrigation and leachate
4374
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 6 7 e4 3 8 0
collected during the study period was 450 mm corresponding to 7% more than the precipitation (Fig. 4). Low concentrations of the phage 28B, i.e. 2 pfu ml1, were detected in leachate from 2/12 lysimeters of sandy loam or coarse sand soils two weeks after the initial application of microbial tracers. Irrespective of soil type, these lysimeters had received irrigation water containing the phage 28B once and the other lysimeters received such irrigation water twice. After 27 days, phage 28B was detected in leachate from three lysimeters containing coarse sand at a concentration of 2 pfu ml1, whereas no phage were found in sandy loam soil lysimeters. Although identical concentrations of the phage 28B were found in these three lysimeters, they were originally irrigated once, twice and three times, respectively, with irrigation water containing the phage. Breakthrough of the phage 28B was seen at 0.019e0.030 pore volume in sandy loam soil and 0.019e0.028 for coarse sand (Fig. 5). The mean recovery rate of the phage in sandy loam soil was only 3.9 106% and for coarse sand 1.7 105% showing a high retention of the phage in the soil system. Phage 28B leached from 29% (7/24) of the lysimeters with two of the lysimeters containing sandy loam soil but no significant difference could be detected between the two soil types ( p ¼ 0.1958). After the four weekly applications of irrigation water containing bacterial pathogens and the phage, phage 28B were not detected during the remaining 185 days of the study period. During the entire study period there were not isolated any bacterial pathogens in the leachate. It should be noted that 0.75 of the total pore volume was exchanged when the experiment was terminated. The reduction in the concentrations of the phage through the lysimeters expressed as the removal rate (Pang, 2009) was found to be 5.06 log unit m1 for sandy loam soil and 4.96 log unit m1 for coarse sand.
3.2.
Microbial tracers on potatoes and in soil
A large variation was observed in the frequency as well as in the concentration of the bacterial pathogens and phage 28B
Fig. 5 e Breakthrough curves for Salmonella Typhimurium phage 28B added to repacked soil lysimeters containing coarse sand (CS) and sandy loam soil (SL). Phage was found in leachate on day 15 and day 27, respectively. C/C0 is the normalized effluent concentration of the phage. T1. T2 and T3 are referring to the number of application with spiked irrigation water in the particular lysimeter.
recovered from the surface of the potatoes (Table 3). Phage 28B was found on potatoes from lysimeters containing coarse sand at a concentration of 6500 pfu per cm2 potato surface. However, the phage was not detected on potatoes grown in sandy loam soil. The bacterial pathogens were found on potatoes from both coarse sand and sandy loam soil lysimeters. Salmonella Senftenberg was isolated as the only bacterial pathogen in three potato samples grown in sandy loam soil. One potato sample contained only E. coli O157:H7 and all bacterial pathogens were isolated from another sample of potatoes grown in sandy loam soil. From coarse sand lysimeters, two potato samples contained E. coli O157:H7 together with the phage, one potato sample contained S. Senftenberg and E. coli O157:H7 and on two potato samples the phage, S. Senftenberg and E. coli O157:H7 were detected. All bacterial pathogens and the phage were detected simultaneously in one potato sample. Significantly more potatoes grown in coarse sand lysimeters ( p ¼ 0.0083) were contaminated with microbial tracers compared to potatoes harvested from the sandy loam soil lysimeters and significant differences were detected between the four microbial tracers ( p ¼ 0.02367). All sandy loam soil and 83% of coarse sand samples, respectively, contained the phage (Table 3), but no significant difference was found between the two soil types. No bacterial pathogens could be isolated from the soil samples.
4.
Discussion
4.1.
Transport of microbial tracers
The fate of the microorganisms in soil is determined mainly by inactivation and transport processes (Yates and Yates, 1990). The processes determining the vertical movement of microorganisms through soil depends on several factors. Microorganisms show higher degree of retention in and adherence to the clay matrix compared to a sandy soil matrix, but on the other hand, cracks that facilitate preferential water movement, e.g. through roots and earthworm channels, occur more frequently in clay-rich soil (McMurry et al., 1998; Stevik et al., 2004). Naturally occurring cracks will be diminished in repacked or tilled soils since the original structure of the soil is disrupted (Young and Ritz, 2000) and the preferential flow paths subsequently disturbed (Jamieson et al., 2002; McMurry et al., 1998). Transport of microorganisms in non-structured soils is dominated by matrix flow (McGechan and Lewis, 2002) and this will slow the leaching of microorganisms through the soil profile (McMurry et al., 1998). In our study, 0.75 of the total soil pore volume was exchanged at the end of the experiment, but the phage 28B was already detected at 0.02e0.03 pore volume indicating spatial variability in the soil structure which could have developed during or after the establishment of the lysimeters and facilitated transport of microorganisms. A possible formation of macropores would result in large variations of the pore water velocity and enable the phage to be transported by macropore flow in some of the lysimeters. Larger pores, e. g. fractures and cracks, have in general a high flow velocity and microorganisms may therefore be excluded from the smaller soil pores because of the
4375
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 6 7 e4 3 8 0
Table 3 e Occurrence and concentration of microbial tracers in soil and on potatoes applied through irrigation water four times. S. Senftenberg E. coli O157:H7 C. jejuni Sample Soil type Phage 28B type (PFU g1 or PFU per cm2) (CFU g1 or CFU per cm2) (CFU g1 or CFU per cm2) (CFU g1 or CFU per cm2) P/Ta Soil
Potato
Sandy loam soil Coarse sand Sandy loam soil Coarse sand
Mean [minemax]
P/T
Mean [minemax]
P/T
Mean [minemax]
P/T
Mean [minemax]
6/6
417 [160e947]
0/6
ndb
0/6
nd
0/6
nd
5/6
6942 [6.7e20000]
0/6
Nd
0/6
nd
0/6
nd
0/6
nd
4/6
36 [2.7e163]
2/6
7.4 [6.5e38]
1/6
76
5/6
6500 [447e31737]
4/6
33 [4.3e174]
6/6
28 [2.2e51]
1/6
4.2
a P/T, positive samples/total samples. b nd, not detected.
high flow velocity. Pang et al. (2008) found the maximum concentration of bacteria and phage leaching through intact soil columns at 0.26e0.37 pore volume in sandy soil (clay content 6%) whereas maximum concentration of microbial tracers in silt loam soils (approximate clay contents 20%) were observed at 0.03 to 0.11 pore volume which is similar to our findings. Carlander et al. (2000) studied transport of phage 28B in 1.2 m clay soil lysimeters under moderate irrigation and detected the phage in leachate 2e24 h after application. In comparable lysimeters containing sandy loam (8% clay content) the breakthrough time was considerably longer, i.e. 126e159 days. Carlander et al. (2000) explained the fast transport in the clay lysimeters by rapid flow of soil water in macropores created over the eight years since establishment of the lysimeters due to e.g. physical weathering. But they did not find a similar fast transport in sandy loam lysimeters even though they were established in the same way. In the present study where the lysimeters were established 14 years ago, a limited number of small-sized macropores were likely present that facilitated a fast transport of the phage, but not the bacterial pathogens. These limited numbers of macropores could be a consequence of the low clay content in the lysimeters as drying-rewetting events are known to affect the soil structure through shrinking and swelling and the formation of cracks which may be of very fine size (Tessier et al., 1990). In addition, these macropores may also be created by roots of crops grown in the lysimeters and by earthworms (Beven and Germann, 1982). It is therefore likely that the predominant flow of water in the lysimeters was matrix flow due to the repacked soil and destruction of natural pores, but preferential flow may have occurred in lysimeters that leached the phage. Transport of bacteria through sieved and mixed soil columns have been reported to be minimal compared to intact soil cores (Smith et al., 1985), findings which are supported by Dean and Foran (1992) who found that shearing of macropores in the surface soil by tillage retained E. coli. The use of repacked soil in our study may therefore explain why bacterial pathogens were not detected in the leachate. On the other hand, Stoddard et al. (1998) reported that tillage of soil did not affect the concentration of faecal bacteria leached through soil. In addition, transport of E. coli O157:H7 through intact and repacked soil cores did not result
in a significant difference in concentration of E. coli found in the drainage water (Gagliardi and Karns, 2000). The irrigation rate in the present study was adjusted to simulate natural rain events of approximately 1 mm/day when bacterial pathogens and the phage 28B were applied. Other studies have reported a much faster breakthrough of phage 28B, e.g. already after 2e24 h when 6e120 times larger volumes of irrigation water was applied (Carlander et al., 2000; McLeod et al., 2001; Pang et al., 2008). Similar fast breakthrough of bacteriophage PRD1 was also reported during natural rain conditions as a consequence of macropore flow (Nicosia et al., 2001). The soil moisture content in our lysimeters was lower compared to those used by Carlander et al. (2000) and this could on the other hand have decreased the deeper penetration of phage 28B. Decreasing the saturation has been reported to increase the likelihood of viruses attaching to soil particles (Chu et al., 2003; Torkzaban et al., 2006) whereas spreading of microorganisms on a wet soil can lead to a very high leaching of contaminants (McGechan and Vinten, 2003). According to the colloid filtration theory, the decrease of water velocity in soil increases the number of collisions of colloids, e.g. microorganisms, with soil particles. This results in increased colloid retention due to attachment to the soil particles (Guber et al., 2005). In addition, other physical factors, e.g. straining, soil surface roughness and water content, have been reported to have a significant role in colloid retention in soil (Bradford et al., 2006; Dı´az et al., 2010; Torkzaban et al., 2008). Straining is size depending and larger microorganisms like bacteria are therefore retained at higher rates as compared to the smaller viruses (Bradford et al., 2006). A ratio of microorganism size to the median size of soil particles greater than 0.5% has been proposed as guideline criterion for straining (Bradford et al., 2003). According to this, recovery of phage 28B in our study, but not any of the bacterial pathogens, could be explained by the straining of the larger sized bacterial cells (0.5e2 mm) compared to the smaller size of phage (60 nm) in the repacked soil used in the experiments. This is corroborated by other studies showing an extended transport of bacteriophages through soil when compared to E. coli (Hijnen et al., 2005; Sinton et al., 1997). However, small particles with similar sizes as viruses can be retained in the fine pores due to diffusion into the soil matrix (Cumbie and McKay, 1999). Our
4376
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 6 7 e4 3 8 0
findings of phage 28B in leachate 27 days after it was applied in lysimeters containing coarse sand, but not in any lysimeters with sandy loam soil, are supported by an enhanced retention and adsorption of microorganisms in fine-textured soil compared to larger sand particles (Huysman and Verstraete, 1993). In general, clay minerals provide better adsorption sites due to their small size, their platy shapes, surface charge and the large surface area per given volume for microorganisms (Gerba and Bitton, 1984). Several physical properties of soil have been reported to influence the ability of the soil to retain microorganisms, e.g. bulk density, mean grain size and hydraulic conductivity. It is frequently reported that bulk density is not a good measure of the degree of compactness of soils (Dexter et al., 2007) but a relative bulk density (ratio of natural bulk density to a reference bulk density) incorporating the clay content of the soil is a useful quantity for comparison of compactness of different soils (Asgarzadeh et al., 2010). Artz et al. (2005) reported a decreasing leaching rate of E. coli O157:H7 with increasing bulk density and at a relative bulk density of 0.6 (bulk density of 1.15 g cm3) they found that less than 0.4% of added bacterial cells were recovered in repacked soil cores. The relative bulk density in our study was in the ranges 0.73e0.77 (coarse sand) and 0.79e0.86 (sandy loam soil) which are in support of the low leaching rate found. Knappett et al. (2008) reported substantial decreases in effluent concentration of both MS-2 bacteriophage (25 nm) and carboxylated microsphere (1.5 mm) by decreasing mean grain size from medium sand (d50 700 mm) to fine sand (d50 340 mm). In our study, d50 for coarse sand and sandy loam soil was 240 mm and 27 mm, respectively, which would lead to an enhanced retainment of phage and bacterial pathogens in the repacked lysimeters. Intact soils have a higher hydraulic conductivity facilitating microbial transport compared to repacked soil (Dec et al., 2008). The leaching of Rhodotorula sp. was increased in sand columns filled with 368 mm-sized sand particles (hydraulic conductivity of 0.137 cm s1) compared to columns filled with 240 mm-sized sand particles (hydraulic conductivity of 0.056 cm s1) (Marlow et al., 1991). Textural discontinuities between soil layers were present in the lysimeter. In addition, relative bulk density increased and saturated hydraulic conductivity decreased by depth. Sandy loam soil had a higher relative bulk density and a lower hydraulic conductivity and was therefore more efficient in retaining microbial tracers compared to the coarse sand (Table 1). This could have induced lateral redistribution and decreased the leaching of microbial tracers through the layers in the lysimeters. Virus results indicated the presence of preferential flow conditions that produced irregular concentration profiles with depth due to layers of contrasting texture at an experimental site (Powelson et al., 1993). Thus, even under our experimental conditions with repacked soil, the removal rate of phage 28B was 5.06 log m1 for sandy loam soil and 4.96 log m1 for coarse sand which corroborates findings in a similar study of a Salmonella phage in loamy sand and faecal coliforms in sandy loam soil (Pang, 2009).
4.2.
Survival of microbial tracers
The long duration of the experiment totaling 212 days was a consequence of the irrigation strategy to apply water
volumes at a frequency similar to normal precipitation during winter time together with the large volume of the lysimeters. The survival time of different pathogens in soil has been shown to vary from 4 to 160 days (Abu-Ashour et al., 1994; Sjogren, 1994), e.g. Salmonella may survive in coarse sand up to 64 days under unsaturated conditions (Parker and Mee, 1982). Due to the long duration of the experiment it is likely that the bacterial pathogens were inactivated because of stress and detrimental impacts of various factors in the soil. The exposure to the soil environment may also have induced the so-called VBNC (Viable-But-Non-Culturable) stage of bacterial cells which is a response to stress and way of survival for many bacteria. In the VBNC stage, the bacterial cells cannot be cultured by traditionally cultured-based methods, but only identified by direct detection, including DNA-based methods (Colwell, 2000). Reports and opinions vary regarding the possibility of VBNC bacterial cells to resuscitate as well as initiating infection (Winfield and Groisman, 2003). It seems likely that bacterial cells in the VBNC stage seldom can initiate disease, but if such cells are resuscitated to the metabolic active stage then their virulence and ability to cause infection is retained (Oliver, 2010). However, it should be noted that the VBNC stage of E. coli O157:H7, V. cholerae and V. vulnificus have been associated with infections in humans and animals (Oliver, 2010). Only bacterial pathogens that could be cultured were enumerated in the present study and we do therefore not know if a VBNC stage of the bacterial pathogens were present in leachate or soil. The VBNC state does not seem to promote persistence of bacterial cells in soil (Mascher et al., 2000). This is supported by the findings of equal numbers of total and culturable Salmonella cells in the leachate from soil monoliths indicating that most cells leached were viable (Bech et al., 2010). It should be noted that potatoes were stored at 4e5 C for three weeks and soil samples for 11 weeks before being analyzed. Although this prolonged storage, which was caused by logistical problems with transport due to long distance from the field site to the laboratory, may have caused a die-off of the microbial tracers during storage, bacterial pathogens were still found on potatoes and phage 28B was recovered in both soil types and on potatoes grown in coarse sand. The potatoes therefore still could represent a source of cross contamination in handling and preparation of foods in the kitchen.
4.3.
Food safety
The main concern based on our study is that bacterial pathogens accumulated on the root crops in both soil types, even though none were evident in the effluent. The use of subsurface drip irrigation protects pathogens from the lethal exposure of ultraviolet light and desiccation at the soil surface (Beard, 1940). Subsurface irrigation further minimizes the contact between irrigation water and crops like fruits and plants with edible parts located above the soil (Oron et al., 2001); however, root crops are at risk of direct faecal contamination by this irrigation method. As a consequence of low irrigation rate, bacterial pathogens have an increased retention time at the root zone level thereby elevating the risk for root crop contamination or root internalization (Bernstein et al., 2007). Drip emitters were placed right between two potato plants. No phage was detected on potato samples
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 6 7 e4 3 8 0
grown in sandy loam soil indicating possible adsorption to clay minerals before reaching the potatoes. This is supported by the findings of Straub et al. (1992) who observed an increase in sorption of poliovirus with increased clay content. Phage 28B was found in both types of soil when soil samples were taken close to the drip emitter. This corresponds with findings of the highest concentration of bacteriophages in the proximity of the drip emitter (Assadian et al., 2005; Enrı´quez et al., 2003). In subsurface drip irrigated column containing repacked clay loam soil, Kouznetsov et al. (2004) only detected F-specific RNA phages and faecal coliforms close to the emitter 24 h after irrigation but 72 h after irrigation, F-specific RNA phages were no longer detectable. E. coli have been found on the surface of cleaned potatoes grown in soil treated with wastewater sludge but could not be detected in the potato core (Chale-Matsau and Snyman, 2006). As our study only analyzed the surface of potatoes, further studies are needed to reveal whether the microbial tracers will be removed by washing the soil of the potatoes or if they are able to penetrate the potato peel.
5.
Conclusion
The main risk of virus contamination through subsurface drip irrigation using low quality water seems to be the possible contamination of groundwater. Low numbers of phage 28B was found in this study and as only few viruses are needed to cause disease in humans such low numbers represent a health risk. Our study suggests that in repacked soils with a low water velocity, human bacterial pathogens will be retained in the soil matrix and inactivated before leaching to groundwater. For bacterial pathogens the main risk associated with subsurface drip irrigation seems to be contamination of root crops. Thus, guidelines for safe use of low quality water for subsurface irrigation should take into consideration the food safety risks associated with subsurface irrigation and consumption of root crops eaten raw, e.g. carrots, asparagus, and radish. Future experiments on pathogen transport should be conducted in lysimeters with intact soil as this will better reflect natural field conditions, i.e. effect of intact macropore structures. The low water velocity should be maintained for simulation of water-saving irrigation practices. Further, rather than potatoes as used in the current study, the contamination and safety of root crops eaten raw should be assessed. It would also be interesting to confirm if crop types, where the edible parts are located above the soil surface, e.g. lettuce, can be safely irrigated with contaminated water through subsurface drip irrigation.
Acknowledgment We would like to thank Annika Holmqvist for the provision of Salmonella Senftenberg 775W and the phage host strain S. Typhimurium type 5 and Jacob Ottoson for providing us with the Salmonella Typhimurium phage 28B. The technical assistance of Igor Kljujev, Diana Karapetian and Finn Christensen was very much appreciated. The researchers would also like to thanks the anonymous reviewers for their constructive comments that significantly improved the quality of this paper.
4377
The study was supported by the “Safe and High Quality Food Production using Low Quality Waters and Improved Irrigation Systems and Management” project (SAFIR, EU, FOOD-CT-2005023168) (www.safir4eu.org) funded by European Commission and PathOrganic (http://pathorganic.coreportal.org).
references
Abbaszadegan, M., LeChevallier, M., Gerba, C., 2003. Occurrence of viruses in U.S. groundwaters. Journal of American Water Works Association 95 (9), 107e120. Abu-Ashour, J., Joy, D.M., Lee, H., Whiteley, H.R., Zelin, S., 1994. Transport of microorganisms through soil. Water. Air and Soil Pollution 75 (1e2), 141e157. Adams, M.H., 1959. Bacteriophages. Interscience Publishers Inc, New York, USA. Ahmadi, S.H., Andersen, M.N., Poulsen, R.T., Plauborg, F., Hansen, S., 2009. A quantitative approach for developing more mechanistic gas exchange models for field grown potato: a new insight into chemical and hydraulic signalling. Agricultural and Forest Meteorology 149 (9), 1541e1551. Artz, R.R.E., Townend, J., Brown, K., Towers, W., Killham, K., 2005. Soil macropores and compaction control the leaching potential of Escherichia coli O157:H7. Environmental Microbiology 7 (2), 241e248. Asgarzadeh, H., Mosaddeghi, M.R., Mahboubi, A.A., Nosrati, A., Dexter, A.R., 2010. Soil water availability for plants as quantified by conventional available water, least limiting water range and integral water capacity. Plant and Soil 335 (1e2), 229e244. Aslyng, H.C., Hansen, S., 1982. Water Balance And Crop Production Simulation: Model WATCROS for Local And Regional Application. Hydrotechnical Laboratory. The Royal Veterinary and Agricultural University, Copenhagen, Denmark. 200. Assadian, N.W., Di Giovanni, G.D., Enciso, J., Iglesias, J., Lindemann, W., 2005. The transport of waterborne solutes and bacteriophage in soil subirrigated with a wastewater blend. Agriculture, Ecosystems and Environment 111 (1e4), 279e291. Ayars, J.E., Phene, C.J., Hutmacher, R.B., Davis, K.R., Schoneman, R.A., Vail, S.S., Mead, R.M., 1999. Subsurface drip irrigation of row crops: a review of 15 years of research at the Water Management Research Laboratory. Agricultural Water Management 42 (1), 1e27. Battilani, A., 2010. Personal communication. Consorzio di Bonfica di secondo grado per il Canale Emilliano Romagnolo -CER, Area Agronomico e ambientale Via E. Masi 8, Ie40137 Bologna, Italy. Beard, P.J., 1940. Longevity of Eberthella Typhosus in various soils. American Journal of Public Health 30 (9), 1077. Bech, T.B., Johnson, K., Dalsgaard, A., Laegdsmand, M., Jacobsen, O.H., Jacobsen, C.S., 2010. Transport and distribution of Salmonella enterica serovar Typhimurium in loamy and sandy soil monoliths with applied liquid manure. Applied and Environmental Microbiology 76 (3), 710e714. Bernstein, N., Sela, S., Neder-Lavon, S., 2007. Effect of irrigation regimes on persistence of Salmonella enterica serovar Newport in small experimental pots designed for plant cultivation. Irrigation Science 26 (1), 1e8. Beven, K., Germann, P., 1982. Macropores and water flow in soils. Water Resources Research 18 (5), 1311e1325. Bitton, G., 2005. Pathogens and parasites in domestic wastewater. In: Bitton, G. (Ed.), Wastewater Microbiology, third ed. John Wiley & Sons, New York, USA, pp. 109e151. Chapter 4.
4378
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 6 7 e4 3 8 0
Bradford, S.A., Simunek, J., Bettahar, M., van Genuchten, M.T., Yates, S.R., 2003. Modeling colloid attachment, straining and exclusion in saturated porous media. Environmental Science and Technology 37 (10), 2242e2250. Bradford, S.A., Simunek, J., Bettahar, M., van Genuchten, M.T., Yates, S.R., 2006. Significance of straining in colloid deposition: evidence and implications. Water Resources Research 42 (W12S15), 1e16. Bolster, C.H., Walker, S.L., Cook, K.L., 2006. Comparison of Escherichia coli and Campylobacter jejuni transport in saturated porous media. Journal of Environmental Quality 35 (4), 1018e1025. Calci, K.R., Burkhardt III, W., Watkins, W.D., Rippey, S.R., 1998. Occurrence of male-specific bacteriophage in feral and domestic animal wastes, human feces, and human-associated wastewaters. Applied and Environmental Microbiology 64 (12), 5027e5029. Carlander, A., Aronsson, P., Allestam, G., Stenstrom, T.A., Perttu, K., 2000. Transport and retention of bacteriophages in two types of willow-cropped lysimeters. Journal of Environmental Science and Health, Part A 35 (8), 1477e1492. Castro, F.D., Tufenkji, N., 2007. Relevance of nontoxigenic strains as surrogates for Escherichia coli O157:H7 in groundwater contamination potential: role of temperature and cell acclimation time. Environmental Science and Technology 41 (12), 4332e4338. CDC (Center for Disease Control and Prevention), 2009. Surveillance for foodborne diseases outbreak, United States, 2006. Morbidity and Mortality Weekly Report 58 (22), 609e615. Chale-Matsau, J.R.B., Snyman, H.G., 2006. The survival of pathogens in soil treated with wastewater sludge and in potatoes grown in such soil. Water Science and Technology 54 (5), 163e168. Chu, Y., Jin, Y., Baumann, T., Yates, M.V., 2003. Effect of soil properties on saturated and unsaturated virus transport through columns. Journal of Environmental Quality 32 (6), 2017e2025. Colwell, R.R., 2000. Viable but nonculturable bacteria: a survival strategy. Journal of Infection and Chemotherapy 6 (2), 121e125. Craun, M.F., Craun, G.F., Calderon, R.L., Beach, M.J., 2006. Waterborne outbreaks reported in the United States. Journal of Water and Health 4 (2), 19e30. Cumbie, D.H., McKay, L.D., 1999. Influence of diameter on particle transport in a fractured shale saprolite. Journal of Contaminant Hydrology 37 (1e2), 139e157. Dean, D.M., Foran, M.E., 1992. The effect of farm liquid waste application on tile drainage. Journal of Soil and Water Conservation 47 (5), 368e369. Dec, D., Do¨rner, J., Becker-Fazekas, O., Horn, R., 2008. Effect of bulk density on hydraulic properties of homogenized and structured soils. Journal of Soil Science and Plant Nutrition 8 (1), 1e13. Dexter, A.R., Czyz, E.A., Gate, O.P., 2007. A method for prediction of soil penetration resistance. Soil and Tillage Research 93 (2), 412e419. Dı´az, J., Rendueles, M., Dı´az, M., 2010. Straining phenomena in bacteria transport through natural porous media. Environmental Science and Pollution Research 17 (2), 400e409. Enrı´quez, C.E., Absar, A., Suarez-Rey, E.M., Choi, C.Y., Oron, G., Gerba, C.P., 2003. Survival of bacteriophages MS-2 and PRD-1 in turfgrass irrigated by subsurface drip irrigation. Journal of Environmental Engineering 129 (9), 852e857. EC, European Water Framework Directive (2000/60/EC), 2000. http://ec.europa.eu/environment/water/water-framework/ index_en.html (accessed November 16, 2010). Feachem, R.G., Bradley, D.J., Garelick, H., Mara, D.D., 1983. In: Feachem, R.G., Bradley, D.J., Garelick, H., Mara, D.D. (Eds.), Sanitation and Disease: Health Aspects of Excreta and
Wastewater Management. John Wiley & Sons, Chichester, England, pp. 133e164. Gagliardi, J.V., Karns, J.S., 2000. Leaching of Escherichia coli O157:H7 in diverse soils under various agricultural management practices. Applied and Environmental Microbiology 66 (3), 877e883. Gerba, C.P., Bitton, G., 1984. Microbial pollutants: their survival and transport pattern to groundwater. In: Bitton, G., Gerba, C. P. (Eds.), Groundwater Pollution Microbiology. John Wiley & Sons, New York, USA, pp. 65e88. Guber, A.K., Shelton, D.R., Pachepsky, Y.A., 2005. Transport and retention of manure-borne coliforms in soil. Vadose Zone Journal 4 (3), 828e837. Havelaar, A.H., 1991. Bacteriophages as model viruses in water quality control. Water Research 25 (5), 529e545. Herna´ndez, F., Monge, R., Jime´nez, C., Taylor, L., 1997. Rotavirus and Hepatitis A virus in market lettuce (Latuca sativa) in Costa Rica. International Journal of Food Microbiology 37 (2e3), 221e223. Hijnen, W.A.M., Brouwer-Hanzens, A.J., Charles, K.J., Medema, G. J., 2005. Transport of MS2 phage, Escherichia coli, Clostridium perfringens, Cryptosporidium parvum, and Giardia intestinalis in a gravel and a sandy soil. Environmental Science and Technology 39 (20), 7860e7868. Huysman, F., Verstraete, W., 1993. Water-facilitated transport of bacteria in unsaturated soil columns: influence of cell surface hydrophobicity and soil properties. Soil Biology and Biochemistry 25 (1), 83e90. Ho¨glund, C., Ashbolt, N., Stenstrom, T.A., Svensson, L., 2002. Viral persistence in source-separated humane urine. Advances in Environmental Research 6 (3), 265e275. IPCC, 2007. Climate Change 2007: Synthesis Report. Contribution of Working Groups I, II and III to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change. IPCC, Geneva, Switzerland. Jamieson, R.C., Gordon, R.J., Sharples, K.E., Stratton, G.W., Madani, A., 2002. Movement and persistence of fecal bacteria in agricultural soils and subsurface drainage water: a review. Canadian Biosystems Engineering 44 (1), 1e9. Johansson, P.-O., Espeby, B., Nilsson, B., Allestam, G., 1998. Artificial groundwater recharge in Stockholm e II Column test design and tracer tests. In: Peters, J.H. (Ed.), Artificial Recharge of Groundwater, Proceedings of the 3. International Symposium on Artificial Recharge of Groundwater, TISAR 98. A.A. Balkema, Rotterdam, Netherland, pp. 383e385. Knappett, P.S.K., Emelko, M.B., Zhuang, J., McKay, L.D., 2008. Transport and retention of a bacteriophage and microspheres in saturated, angular porous media: effect of ionic strength and grain size. Water Research 42 (16), 4368e4378. Kouznetsov, M.Y., Pachepsky, Y.A., Gillerman, L., Gantzer, C.J., Oron, G., 2004. Microbial transport in soil caused by surface and subsurface drip irrigation with treated wastewater. International Agrophysics 18 (3), 239e247. Kramer, M.H., Quade, G., Hartemann, P., Exner, M., 2001. Waterborne diseases in Europe 1986e1996. Journal of American Water Works Association 93 (1), 48e53. Landbrugsministeriet, 1976. The Danish Soil Classification (Den Danske Jordklassificering). Teknisk Redegørelse, København, Danmark. 88. Lazarova, V., Cirelli, G., Jeffrey, P., Salgot, M., Icekson, N., Brissaud, F., 2000. Enhancement of integrated water management and water reuse in Europe and the Middle East. Water Science and Technology 42 (1e2), 193e202. Leclerc, H., Edberg, S., Pierzo, V., Delattre, J.M., 2000. Bacteriophages as indicators of enteric viruses and public health risk in groundwaters. Journal of Applied Microbiology 88 (1), 5e21. Lilleengen, K., 1948. Typing of Salmonella Typhimurium by Means of a Bacteriophage. PhD Thesis. The Bacteriological and
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 6 7 e4 3 8 0
Hygienical Department of the Royal Veterinary College, Stockholm, Sweden. Long, G., Zhu, P., Shen, Y., Tong, M., 2009. Influence of extracellular polymeric substances (EPS) on deposition kinetics of bacteria. Environmental Science and Technology 43 (7), 2308e2314. Marlow, H.J., Duston, K.L., Wiesner, M.R., Tomson, M.B., Wilson, J.T., Ward, C.H., 1991. Microbial transport through porous media: the effects of hydraulic conductivity and injection velocity. Journal of Hazardous Materials 28 (1e2), 65e74. Mascher, F., Hase, C., Moe¨nne-Loccoz, Y., De´fago, G., 2000. The viable-but-nonculturable state induced by abiotic stress in biocontrol agent Pseudomonas fluorescens CHA0 does not promote strain persistence in soil. Applied and Environmental Microbiology 66 (4), 1662e1667. Maule, A., 1999. Environmental aspect of E. coli O157. International Food Hygiene Journal 9, 21e23. Mawdsley, J.L., Bardgett, R.D., Merry, R.J., Pain, B.F., Theodorou, M. K., 1995. Pathogens in livestock waste, their potential for movement through soil and environmental pollution. Applied Soil Ecology 2 (1), 1e15. McGechan, M.B., Lewis, D.R., 2002. Transport of particulate and colloid-sorbed contaminants through soil, Part 1: general principles. Biosystems Engineering 83 (3), 255e273. McGechan, M.B., Vinten, A.J.A., 2003. Simulation of transport through soil of E. coli derived from livestock slurry using the MACRO model. Soil Use and Management 19 (4), 321e330. McLeod, M., Aislabie, J., Smith, J., Fraser, R., Roberts, A., Taylor, M., 2001. Viral and chemical tracer movement through contrasting soils. Journal of Environmental Quality 30 (6), 2134e2140. McMurry, S.W., Coyne, M.S., Perfect, E., 1998. Fecal coliform transport through intact soil blocks amended with poultry manure. Journal of Environmental Quality 27 (1), 86e92. Natvig, E.E., Ingham, S.C., Ingham, B.H., Cooperband, L.R., Roper, T.R., 2002. Salmonella enterica Serovar Typhimurium and Escherichia coli contamination of root and leaf vegetables grown in soils with incorporated bovine manure. Applied and Environmental Microbiology 68 (6), 2737e2744. Nicholson, F.A., Groves, S.J., Chambers, B.J., 2005. Pathogen survival during livestock manure storage and following land application. Bioresource Technology 96 (2), 135e143. Nicosia, L.A., Rose, J.B., Stark, L., Stewart, M.T., 2001. A field study of virus removal in septic tank drainfields. Journal of Environmental Quality 30 (4), 1933e1939. Nielsen, J.D., Møberg, J.P., 1985. Classification of soil profiles at Danish Research stations (Klassificering af jordprofiler fra forsøgsstationer i Danmark). Danish Journal of Plant Soil Science 89, 157e167. Nwachuku, N., Gerba, C.P., 2008. Occurrence and persistence of Escherichia coli O157:H7 in water. Reviews in Environmental Science and Biotechnology 7 (3), 267e273. Oliver, J.D., 2010. Recent findings on the viable but nonculturable state in pathogenic bacteria. FEMS Microbiology Reviews 34 (4), 415e425. Oron, G., Goemans, M., Manor, Y., Feyen, J., 1995. Poliovirus distribution in the soil-plant system under reuse of secondary wastewater. Water Research 29 (4), 1069e1078. Oron, G., Armon, R., Mandelbaum, R., Manor, Y., Campos, C., Gillerman, L., Salgot, M., Gerba, C., Klein, I., Enriquez, C., 2001. Secondary wastewater disposal for crop irrigation with minimal risks. Water Science and Technology 43 (10), 139e146. Pang, L., McLeod, M., Aislabie, J., Simunek, J., Close, M., Hector, R., 2008. Modeling transport of microbes in ten undisturbed soils under effluent irrigation. Vadose Zone Journal 7 (1), 97e111. Pang, L., 2009. Microbial removal rates in subsurface media estimated from published studies of field experiments and
4379
large intact soil cores. Journal of Environmental Quality 38 (1), 1531e1559. Parker, W.F., Mee, B.J., 1982. Survival of Salmonella adelaide and fecal coliforms in coarse sands of the Swan Coastal Plain, Western Australia. Applied and Environmental Microbiology 43 (5), 981e986. Powelson, D.K., Gerba, C.P., Yahya, M.T., 1993. Virus transport and removal in wastewater during aquifer recharge. Water Research 27 (4), 583e590. Rzezutka, A., Cook, N., 2004. Survival of human enteric viruses in the environment and food. FEMS Microbiology Reviews 28 (4), 441e453. Shahnazari, A., Fulai, F., Andersen, M.N., Jacobsen, S.E., Jensen, C. R., 2007. Effects of partial root-zone drying on yield, tuber size and water use efficiency in potato under field conditions. Field Crops Research 100 (1), 117e124. Sinton, L.W., Finlay, R.K., Pang, L., Scott, D.M., 1997. Transport of bacteria and bacteriophages in irrigated effluent into and through an alluvial gravel aquifer. Water, Air and Soil Pollution 98 (1e2), 17e42. Sjogren, R.E., 1994. Prolonged survival of an environmental Escherichia coli in laboratory soil microcosms. Water, Air and Soil Pollution 75 (3e4), 389e403. Smith, M.S., Thomas, G.W., White, R.E., Ritonga, D., 1985. Transport of Escherichia coli through intact and disturbed soil columns. Journal of Environmental Quality 14 (1), 87e91. Stevik, T.K., Aa, K., Ausland, G., Hanssen, J.F., 2004. Retention and removal of pathogenic bacteria in wastewater percolating through porous media: a review. Water Research 38 (6), 1355e1367. Stoddard, C.S., Coyne, M.S., Grove, J.H., 1998. Fecal bacteria survival and infiltration through a shallow agricultural soil: timing and tillage effects. Journal of Environmental Quality 27 (6), 1516e1523. Straub, T.M., Pepper, I.L., Gerba, C.P., 1992. Persistence of viruses in desert soils amended with anaerobically digested sewage sludge. Applied and Environmental Microbiology 58 (2), 636e641. ¨ sterberg, P., Lindqvist, A., Jo¨nsson, B., So¨derstro¨m, A., O Lindberg, A., Blide Ulander, S., Welinder-Olsson, C., Lo¨fdahl, S., Kaijser, B., de Jong, B., Kuhlman-Berenzon, S., Boqvist, S., Eriksson, E., Szanto, E., Allestam, G., Hedenberg, I., Ledet Muller, L., Andersson, Y., 2008. A large Escherichia coli O157 outbreak in Sweden associated with locally produced lettuce. Foodborne Pathogens and Disease 5 (3), 339e349. Tessier, D., Beaumont, A., Pedro, G., 1990. Influence of clay mineralogy and rewetting rate on clay microstructure. Developments in Soil Science 19, 115e121. Torkzaban, S., Bradford, S.A., van Genuchten, M.T., Walker, S. L., 2008. Colloid transport in unsaturated porous media: the role of water content and ionic strength on particle straining. Journal of Contaminant Hydrology 96 (1e4), 113e127. Torkzaban, S., Hassanizadeh, S.M., Schijven, J.F., de Bruin, H.A.M., de Roda Husman, A.M., 2006. Virus transport in saturated and unsaturated sand columns. Vadose Zone Journal 5 (3), 877e885. U.S. EPA, 2004. Guidelines for Water Reuse. United States Environmental Protection Agency, Washington. Vinnera˚s, B., Holmqvist, A., Bagge, E., Albihn, A., Jo¨nsson, H., 2003. The potential for disinfection of separated faecal matter by urea and by peracetic acid for hygienic nutrient recycling. Bioresource Technology 89 (2), 155e161. Vinnera˚s, B., Nordin, A., Niwagaba, C., Nyberg, K., 2008. Inactivation of bacteria and viruses in human urine depending on temperature and dilution rate. Water Research 42 (15), 4067e4074. Wendel, A.M., Johnson, D.H., Sharapov, U., Grant, J., Archer, J.R., Monson, T., Koschmann, C., Davis, J.P., 2006. Multistate
4380
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 6 7 e4 3 8 0
outbreak of Escherichia coli O157:H7 infection associated with consumption of packaged spinach, AugusteSeptember 2006: the Wisconsin investigation. Clinical Infectious Diseases 48 (8), 1079e1086. Winfield, M.D., Groisman, E.A., 2003. Role of nonhost environments in the lifestyle of Salmonella and Escherichia
coli. Applied and Environmental Microbiology 69 (7), 3687e3694. Yates, M.V., Yates, S.R., 1990. Modeling microbial transport in soil and groundwater. AMS News 56 (6), 324e327. Young, I.M., Ritz, K., 2000. Tillage, habitat space and function of soil microbes. Soil and Tillage Research 53 (3e4), 201e213.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 8 1 e4 3 8 9
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Oxygen consumption by a sediment bed for stagnant water: Comparison to SOD with fluid flow Makoto Higashino Dept. of Civil and Environmental Engineering, Oita National College of Technology, 1666 Maki, Oita 870-0152, Japan
article info
abstract
Article history:
A model of sedimentary oxygen demand (SOD) for stagnant water in a lake or a reservoir is
Received 8 July 2010
presented. For the purposes of this paper, stagnant water is defined as the bottom layer of
Received in revised form
stratified water columns in relatively unproductive systems that are underlain by silt and
18 April 2011
sand-dominated sediments with low-organic carbon (C) and nitrogen (N). The modeling
Accepted 29 April 2011
results are compared to those with fluid flow to investigate how flow over the sediment
Available online 11 May 2011
surface raises SOD compared to stagnant water, depending on flow velocity and biochemical activity in the sediment. SOD is found to be substantially limited by oxygen
Keywords:
transfer in the water column when water is stagnant. When flow over the sediment surface
Biochemical oxygen uptake
is present, SOD becomes larger than that for stagnant water, depending on flow velocity
Dissolved oxygen
and the biochemical oxygen uptake rate in the sediment. Flow over the sediment surface
Fluid flow
causes an insignificant raise in SOD when the biochemical oxygen uptake rate is small. The
Silt and sand-dominated sediment
difference between SOD with fluid flow and SOD for stagnant water becomes significant as
bed
the biochemical oxygen uptake rate becomes larger, i.e. SOD is 10e100 times larger when
Sediment/water interaction
flow over the sediment surface is present. ª 2011 Elsevier Ltd. All rights reserved.
Stagnant water
1.
Introduction
The dissolved oxygen (DO) balance is one of the most important components of water quality and ecosystems in natural and manmade water systems such as rivers, channels, lakes and reservoirs. The DO balance in water depends on oxygen sinks and sources. Both oxygen exchange through the water surface (re-aeration) and photosynthesis are typical examples of the oxygen sources. Oxygen uptake by bottom sediments can be an oxygen sink as well as respiration in the water column. Thus, it is indispensable to quantify the oxygen uptake rate by the sediments for management of those water systems. The oxygen uptake rate by sediments depends on the condition of water over the sediment surface, i.e. either stagnant or flowing over the sediment surface. On one hand, a large number of studies have been conducted for stagnant water (e.g. Pamatmat, 1971; Hargrave, 1972; Smith, 1978;
Barcelona, 1983). In most cases, researchers have been concerned about quantifying oxygen uptake rate for each unique sediment. Those results, therefore, can hardly be used for general prediction of water quality, and hence, water quality management. On the other hand, water motion, i.e. flow velocity, over the sediment surface has a significant effect on the DO flux (from flowing water to the sediment surface) at the sediment/water interface (Sedimentary Oxygen Demand: SOD) (e.g. Boudreau and Joergensen, 2001). The significance of flow over the sediment surface on the DO concentration profile, and corresponding SOD was first demonstrated using an oxygen microprobe (Joergensen and Revsbech, 1985; Joergensen and DesMarais, 1990). The experimental results from the microprobe showed that the DO profile varies rapidly within a thin layer which is on the order of a few millimeters above the sediment surface, i.e. the diffusive boundary layer. The thickness of the diffusive
E-mail address:
[email protected]. 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.04.051
4382
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 8 1 e4 3 8 9
boundary layer and corresponding DO concentration gradient, and thereby, the DO flux at the sediment/water interface (SOD) varies with flow velocity. The effect of flow velocity on SOD has been studied theoretically (e.g. Rahm and Svensson, 1989; Dade, 1993; Nakamura and Stefan, 1994) and experimentally (e.g. Mackenthun and Stefan, 1998; Josiam and Stefan, 1999; Steinberger and Hondzo, 1999; House, 2003; Roy et al., 2004; Hondzo et al., 2005; O’Connor et al., 2008; O’Connor and Hondzo, 2008). The author studied oxygen transfer in the diffusive boundary layer and oxygen consumption due to microbial and chemical processes in sediments, and developed a model to quantify SOD (Higashino et al., 2004). The model incorporates the effect of turbulence above the sediment/water interface, and describes the oxygen uptake rate as a function of biochemical activity. This paper considers SOD mainly in manmade reservoirs that do not have a continuous inflow or outflow, but rather periodically emptied by operating the gate (Fig. 1), and refilled. A lot of dams have been constructed for flood control and water power generation in Japan, however, some of those reservoirs have water quality problems such as oxygen depletion in the bottom layer. In response to these countermeasures, e.g. flushing or installment of aerators in the bottom, has been conducted to improve water quality. Considering stratified reservoirs, water can be apparently stagnant (flow over the sediment can be negligible) when the inflow or outflow rate is relatively smaller than the size of reservoirs. When water is stagnant, SOD may be much smaller than that with fluid flow. However, the question of how flow over the sediment surface raises stagnant water SOD rates depending on flow velocity and oxygen uptake in the sediment still remains open. Meanwhile, SOD may vary associated with the DO concentration profile near the sediment/water interface with time. Then, the detention time of the water body (the interval between flushing) can be significant for the total oxygen uptake by the sediment, and hence, oxygen balance in water (see Fig. 1). The detention time (hydraulic resident time) may be controlled e.g. by an operation of gates in manmade reservoirs. For water quality management it is important to know how SOD changes with the detention time, under which conditions turbulence exerts a significant effect on SOD, and how much flow over the sediment surface raises SOD compared to stagnant water. Such a study is, therefore, warranted. The purposes of this study are (1) to develop a model of SOD for stagnant water, (2) to demonstrate a significance of
Fig. 1 e Schematic of water exchange and oxygen uptake by the sediment in a manmade reservoir.
biochemical oxygen uptake and of the detention time of water on SOD for stagnant water, (3) to compare SOD for stagnant water to SOD with fluid flow, and (4) to investigate under which conditions the effect of turbulence above the sediment surface on SOD becomes significant. The model presented in this paper is for sediments that are relatively low in organic carbon (C) and nitrogen (N) content, such as relatively clean clay, silt, and sand (e.g. House, 2003) with insignificant nitrogenous SOD (and also relatively low carbonaceous SOD). Since diagenesis reactions are not taken into account, the model may not be suitable for sediments rich in C and N with significant diagenesis reactions in e.g. a eutrophic lake. Diagenesis models (e.g. Ditoro, 2001; Chapra, 1997) can be used for sediments that are relatively rich in organic C and N and contain anaerobic pore water in sediment zones where diagenesis reactions occur, overlain by a relatively thin oxidized layer.
2.
Model
Consider a manmade reservoir as shown in Fig. 1. Water stays over a period of the detention time (T) depending on the interval between flushing. Since water motion over the sediment surface is assumed to be negligible DO transfer from the water column to the sediment surface is apparently molecular. Then, the DO concentration profile and corresponding DO flux at the sediment/water interface (SOD) vary with time during the detention time. In this paper, the detention time will be varied over a range from 1 h to 3 months. When the detention time is short (less than 1 day), water motion may be present due to e.g. inflow and outflow. Even for short detention time, the assumption of stagnant water still can be valid for density stratified lakes or reservoirs. Oxygen uptake in sediment depends on the utilization of oxygen by bacteria decomposing organic matter and oxidation of reduced metabolites. In this study, Monod kinetics is used to describe oxygen uptake in the sediment because experimental results using a microprobe showed that both zero-order kinetics and the Monod equation are sufficient to express oxygen uptake due to microbial and chemical processes in the sediment (House, 2003). The oxygen balances in the water column above the sediment/water interface and in the sediment are, therefore, written as vC v2 C ¼ Dw 2 ðoverlying waterÞ vt vy
(1)
vC v2 C mC ðinside the sedimentÞ ¼ Ds 2 vt vy Ko2 þ C
(2)
in which C is DO concentration, t is time, Dw is the molecular diffusivity for DO in water, Ds is the effective diffusion coefficient for DO within the sediment, Ko2 is the half-saturation constant that is taken to be 0.2 mg l1 based on previous results (Higashino et al., 2004), m is the maximum oxidation rate, and y is the vertical coordinate from the sediment/water interface (positive upward) (see Fig. 2). The model (Eqs. (1) and (2)) is similar to that with fluid flow. When flow (turbulence) is
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 8 1 e4 3 8 9
y Water
C
Boundary layer thickness
w
Cw
C
0 C/ C =0.99
s
Sediment/water interface Penetration depth Sediment Fig. 2 e Schematic dissolved oxygen concentration profile near the sediment/water interface.
present over the sediment surface, turbulent diffusion of DO in the water column (in the diffusive boundary layer just above the sediment/water interface) is incorporated in Eq. (1) (Higashino et al., 2004). Since the half-saturation constant in Eq. (2) is relatively small compared to DO in the sediment pore water, the DO dependence becomes largely zero-order, beside when DO is very low. The limitation of microbial DO uptake in the sediment by the availability of organic substrate could be expressed by adding a Monod expression for dissolved organic carbon, similar to that for DO, to the last term of Eq. (2). Recent experimental results support that it is not necessary when the detention time is short. House (2003) measured the oxidation rate over a period of 38 days for silt under fluid flow condition. Although measured oxidation rate, i.e. the zero-order rate constant, seemed to decrease slightly with time (see Table 2 and Fig. 2(b) in House (2003)), it can substantially be constant except first 24 h. This suggests that organic substrate availability is not likely vary with depth in the sediment layer and that available organic substrate concentration that is included implicitly in the maximum oxidation rate (m) in Eq. (2) can substantially be constant over a few days or weeks. Therefore, Eq. (2) can be sufficient for short time periods (less than 38 days). This approach is different from that of geochemists who study long time periods (years), while this paper considers a few days to 3 months duration. Geochemists assume that microbial populations can be starved and can replicate rapidly. The long-term temporal variability of organic matter availability to the micro-organisms is then important, and the
4383
balance of organic substrate concentration in the sediment needs to be tracked by a separate mass balance equation. Based on the experimental results by House (2003), it can be assumed that organic matter is available at a constant when the duration of stagnant conditions is relatively short. The limitation and applicability of the model depend on the characteristics of sediments. Equation (2) can be used for relatively clean silt and sand-dominated sediments with low-organic C and N like the sediments used in House’s experiments (2003), i.e. oxygen uptake by diagenesis reactions is insignificant. Oxygen demand caused by benthic flux of dissolved organic carbon (CBOD) and ammonia (NBOD), and by deposition and diagenesis of organic C and N, can be significant for sediments rich in C and N (highly organic sediments), and in this case diagenesis models (e.g. Ditoro, 2001; Chapra, 1997) need to be used. The effective diffusion coefficient for DO in the sediment pore space (Ds) is related to the sediment characteristics (Boudreau and Joergensen, 2001), and is taken to be 50% of molecular diffusion in water, i.e. Ds y1=2$Dw . This value was chosen because only the pore space fraction is available to diffusion (Bear, 1972), and molecules must travel a longer path around the sediment grains. Several empirical relationships between Ds and the sediment porosity (f) have been proposed (e.g. Boudreau, 1997; Iversen and Jørgensen, 1993). The factor by which Dw has to be reduced from its value in water is typically from 0.3 to 0.7 based on typical sediment porosities. The value of 1/2 has been chosen because it has been used in other investigations (e.g. Higashino et al., 2004; Gantzer and Stefan, 2003). The interstitial water motion and its effect on mass transport are not included. The basic equations are normalized using the sediment depth as the length scale (d), and kinematic viscosity (n) by algebraic procedures, as is common practice in physics and engineering: vC 1 v2 C ¼ vt Sc vy2
(3)
vC 1 v2 C m C ¼ $ vt 2Sc vy2 Ko2 þ C
(4)
where non-dimensional parameters are 9 C n y> ; t ¼ 2 t; y ¼ > = CN d d 2 Ko2 md > > ; ; m ¼ Ko2 ¼ CN nCN C ¼
(5)
where CN is DO concentration in the bulk water (see Fig. 2). A characteristic parameter, the Schmidt number defined by Eq. (6), appears in Eqs. (3) and (4). Sc ¼
n Dw
(6)
The Schmidt number for DO in water is a function of the temperature, e.g. Sc ¼ 1240 at 4 C (hypolimnion of deep lake or colder stream) and Sc ¼ 300 at 30 C (tropical and some shallower systems in summer). In this study, the Schmidt number is taken to be 500 at 20 C (Denny, 1993).
4384
3.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 8 1 e4 3 8 9
Solution
The non-dimensional basic equations (3) and (4) were expressed in the Crank-Nicolson finite-difference form with a spatial step (Dy*) of 0.01 and a time step (Dt*) of 0.01. Equations (3) and (4) are subject to two boundary conditions, respectively. The DO concentration at y ¼ N (far away from the sediment/water interface) is taken to be CN (the DO concentration in the bulk water) for Eq. (3). Although the continuous inflow is considered, the inflow rate is assumed to be much smaller than the size of the reservoir. For Eq. (4), the DO concentration at the bottom of the sediment (y ¼ -d) is take to be zero, i.e. C ¼ 0. Both equations (3) and (4) also require the DO concentration at the sediment/water interface (Cw). Its value is calculated explicitly. First, the DO concentration C1nþ1 next to the sediment/water interface in the water column is calculated as ¼ Cn1 þ Cnþ1 1
Dt 1 n $ C 2Cn1 þ Cnw Dy2 Sc 2
(7)
Fig. 3 e Variation of DO concentration profile with time for Sc [ 500, Ko2 [ 0.2 mg lL1, and m [ 100 mg lL1 dL1.
where superscript (n) denotes the time step. Second, the DO concentration C1nþ1 next to the sediment/water interface in the sediment for the next time step is calculated as n Cnþ1 1 ¼ C1 þ
1 Dt n m $Dt $ 2 C2 2Cn1 þ Cnw Cn1 n Ko2 þ C1 2Sc Dy
(8)
Using C1nþ1and C1nþ1, the DO concentration Cwnþ1 at the sediment/water interface is obtained as ¼ Cnþ1 w
Cnþ1 þ Cnþ1 1 1 2
(9)
The DO concentration profile as a function of time is determined by solving equations (3) and (4) implicitly. The computational domain went from y ¼ d ¼ 10 mm to y ¼ 100 cm.
4. Time variation of DO profile and DO flux at the sediment/water interface The time variation of the DO concentration profile near the sediment/water interface obtained by the model is shown in Fig. 3 for the Schmidt number Sc ¼ 500 and the maximum oxidation rate m ¼ 100 mg l1 d1. The normalized DO concentration (C*) was taken to be zero everywhere in the sediment and to be CN ¼ 10 mg l1 in the water column, initially. The DO profile varies substantially with time in the water column, but does not change significantly in the sediment. The normalized DO concentration at the sediment/ water interface (Cw/CN) is less than 0.1, and is close to zero regardless of biochemical activity characterized by the value of m. When flow is present over the sediment surface, the normalized DO concentration at the sediment/water interface (Cw/CN) and the oxygen penetration depth (ds) were reported to be 0.50e0.96, and 0.44e0.56 cm, respectively, for pffiffiffiffiffiffiffiffiffiffi m ¼ 100 mg l1 d1 depending on the shear velocity (U* ¼ s0 =r, where s0 is the boundary shear stress and r is the fluid density) (Higashino et al., 2004). Figs. 3 and 6 show that the DO
penetration depths are close to zero beside the maximum oxidation rate is small (m ¼ 10 mg l1 d1). This indicates that except highly organic sediment (m ¼ 10,000 mg l1 d1) organic matter can be oxidized sufficiently in an aerobic layer which is on the order of a few millimeter just below the sediment/ water interface, i.e. dissolved organic carbon is hardly transferred from the sediment to overlying water. The time variation of the DO flux at the sediment/water interface (SOD) obtained by the model is shown in Fig. S1 for m ¼ 100 mg l1 d1 and the detention time T ¼ 1 h. When the water body was replaced by a new one, the normalized DO concentration in the water column everywhere was taken to be unity again, i.e. C ¼ CN. The oxygen flux at the sediment/ water interface (J), i.e. SOD, can be calculated from the DO concentration profile by Fick’s law as
Fig. 4 e SODave for stagnant water versus detention time (T ) for Sc [ 500 and Ko2 [ 0.2 mg lL1.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 8 1 e4 3 8 9
4385
stagnant, the time-averaged SOD (SODave) over a period of the detention time (T) defined by Eq. (11) is introduced. SODave ¼
1 T
ZtþT JðtÞdt
(11)
t
Since the DO profile and corresponding DO flux at the sediment/water interface (SOD) vary with time when water is
Fig. 4 illustrates the SODave calculated by Eq. (11) for the Schmidt number Sc ¼ 500, and biochemical oxygen uptake rate m ¼ 10, 100, 1000, and 10,000 mg l1 d1. As shown in Eq. (1), the model assumes that there is no oxygen sink in the water column. Considering the short time periods (days to weeks) the model can be applicable when the sediment is not highly organic, e.g. m 1000 mg l1 d1 because an aerobic zone which is on the order of a few millimeters is present just below the sediment/water interface (see Fig. 6). Highly organic sediments in e.g. eutrophic lakes can be anoxic, and then, the mass transport of dissolved organic carbon from the sediment to overlying water may play an important role in determining the oxygen balance in the water column, and hence, DO concentration near the sediment/water interface. The detention time (T) is over a range from 1 h to 3 months. Since effects of benthic fluxes of dissolved organic carbon and ammonia from the sediment are not taken into account, the model may underestimate SOD for long time periods (T ¼ 2 and 3 months). The SODave is large when the detention time (T) is short, and diminishes as the detention time (T) becomes longer. The SODave decreases to 21, 8, and 4% of the SODave for T ¼ 1 h with increasing detention time, i.e. T ¼ 1 day, 1 week, and 1 month, respectively, for m ¼ 100 mg l1 d1. The SODave increases as the biochemical oxygen uptake rate (m) becomes larger. But dependence of the SODave on oxygen consumption rate (m) in the sediment is not as strong as the SOD with fluid flow (see Figs. 5 and 8 in Higashino et al., 2004). As can be expected, the water column DO next to the sediment/water interface gets lower with increasing detention time (T), and thus, the SODave becomes smaller depending on the microbial activity in the sediment described by m. For long time periods (more than several months), re-aeration,
Fig. 6 e Oxygen penetration depth (ds) versus detention time (T) for Sc [ 500 and Ko2 [ 0.2 mg lL1.
Fig. 7 e Dependence of SOD for stagnant water versus SOD with fluid flow (SODave/SODN) on detention time (T).
Fig. 5 e Boundary layer thickness (dw) versus detention time (T) for Sc [ 500 and Ko2 [ 0.2 mg lL1.
J ¼ Dw
vC vC ¼ D s vy y¼þ0 vy y¼0
(10)
The value of SOD gets smaller with time resulted from decreasing DO concentration gradient at the sediment/water interface (see Fig. 3). When the water body is replaced by a new one, the SOD increases discontinuously due to a large DO concentration gradient at the sediment/water interface. The change in the SOD during the detention time (T) shown in Fig. S1 repeats the same pattern every period of the detention time (T).
5.
Time-averaged SOD
4386
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 8 1 e4 3 8 9
longer. The DO penetration depth (ds) depends also on biochemical oxygen uptake. The DO penetration depth (ds) is large when the value of m is small, i.e. m ¼ 10 mg l1 d1, whereas decreases as the biochemical oxygen uptake rate (m) becomes larger. The DO penetration depth (ds) can close to zero for m > 1000 mg l1 d1. When flow is present over the sediment surface, the DO penetration depth was reported to be 0.44e0.56, and 0.10e0.18 cm for m ¼ 100 and 1000 mg l1 d1, respectively (Higashino et al., 2004). These are much larger than those for stagnant water.
6. Comparison of SOD for stagnant water to SOD with fluid flow
Fig. 8 e Dependence of SOD with fluid flow versus SOD for stagnant water on shear velocity.
photosynthesis, and the oxygen demand due to benthic flux of dissolved organic carbon (CBOD) and ammonia (NBOD) may have a significant effect on oxygen balance in the water column and on SOD. Since this paper considers relatively short time periods (days to weeks), these are not taken into account. Fig. 4 also shows the dependence of SOD on the detention time (T), and is useful for water quality management in manmade reservoirs when the detention time is short (less than 1 month). The diffusive boundary layer thickness was defined as the distance from the sediment/water interface (y ¼ 0) to the point where C/CN ¼ 0.99 when turbulence is present over the sediment surface (Higashino et al., 2004). Similarly, the boundary layer thickness above the sediment/water interface can be defined as the distance from the sediment/water interface (y ¼ 0) to the point where C/CN ¼ 0.99. For stagnant water the boundary layer starts to grow when the water body is replaced (t ¼ 0). The boundary layer thickens with time, and becomes thickest at t ¼ T (detention time). Simulated boundary layer thickness at t ¼ T is illustrated in Fig. 5. The boundary layer thickness (dw) depends on the detention time (T), i.e. the boundary gets thicker as the detention time (T) becomes longer, and is independent of the value of m. Fig. 5 also indicates how the oxygen depletion zone grows with increasing detention time (T) in lakes and reservoirs due to oxygen uptake by the sediment. When water is stagnant and respiration in the water column is ignored, SOD is a weak oxygen sink, and the oxygen depletion zone grows slowly. Flow above the sediment surface makes SOD a more significant oxygen sink, i.e. SOD increases as the velocity of water flowing above the sediment/water interface becomes larger (Higashino et al., 2004). Both flow over the sediment surface and respiration in the water column make the oxygen depletion zone grow faster. The DO penetration depth (ds) can be defined as the distance from the sediment/water interface to the point where C* ¼ 0.001, and is shown in Fig. 6. The DO penetration depth (ds) diminishes as the detention time (T) becomes
The effect of flow over the sediment surface on oxygen uptake by the sediment can be demonstrated by comparing SOD for stagnant water to SOD with fluid flow. SOD with fluid flow was previously analyzed using the diffusive boundary layer theory and zero-order kinetics, and was given as a function of the shear velocity (U*), and the zero-order rate constant m0 (Higashino et al., 2004) as vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 0 pffiffiffi u 2 12 u u 3 3 2 nSc 3 U A Ds m0 þ tðDs m0 Þ þ2m0 Ds CN @ 2p SOD ¼
pffiffiffi 2 3 3 nSc 3 U 2p
(12)
in which n(¼0.1) is a constant. Fig. S2 illustrates the analytical solution (Eq. (12)) of the SOD with fluid flow (SODflow) for the zero-order rate constant m0 ¼ 10, 100, 1000, and 10,000 mg l1 d1. The SODflow depends on the shear velocity (U*), i.e. the SODflow increases as the shear velocity (U*) increases, and approaches a constant value (SODN) given by Eq. (13) when the shear velocity (U*) is large enough (U*/N in Eq. (12)). SODN ¼
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2m0 Ds CN
(13)
The SODflow also depends on the biochemical oxygen uptake rate, i.e. the zero-order rate constant, in the sediment (m0). The value of SODN obtained by Eq. (13) for m0 ¼ 1000 mg l1 d1 is 10 times larger than that for m0 ¼ 10 mg l1 d1, which is different from the SODave for stagnant water (Fig. S2). SODN can be a good measure for oxygen uptake by the sediment when flow over the sediment surface is present. SODN is related to biochemical oxygen uptake measured by the value m0 in the sediment because when the shear velocity (U*) is large, oxygen transfer through the diffusive boundary layer is fast, and the DO flux at the sediment/water interface (SOD) is limited only by the oxygen consumption rate in the sediment (Higashino et al., 2004). The SODave for stagnant water in Fig. 4 is normalized by the SODN, and is shown in Fig. 7. When water is stagnant SOD is limited by oxygen transfer from the water column to the sediment surface. For weak biochemical activity in the sediment, i.e. m ¼ 10 mg l1 d1, SOD can be sediment-side controlled when the water body is replaced by a new one, and then, the SODave becomes larger than the SODN. This results from a high oxygen transfer rate induced by a large DO concentration gradient near
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 8 1 e4 3 8 9
the sediment/water interface (see Fig. 3). The DO concentration gradient becomes smaller with time, resulting in decreasing oxygen transfer rate, and finally, oxygen transfer in the water column becomes limiting when the detention time (T) is long. When biochemical activity in the sediment is strong, i.e. m > 100 mg l1 d1, SOD is completely controlled by oxygen transfer in the water column. The ratio SODave/SODN decreases as the detention time (T) becomes longer. The decrease in water column and sediment DO concentration causes reduction in SOD due to attenuation of the oxidation rate described by the Monod function in Eq. (2) with increasing detention time (T). Fig. S3 demonstrates that the oxygen attenuation factor (C/ (Ko2 þ C)) in Eq. (2) decreases as the detention time (T) becomes longer at the sediment/water interface for the biochemical oxygen uptake rate m ¼ 10, 100, 1000 and 10,000 mg l1 d1. When the detention time is longer than 10 days the ratio SODave/SODN decreases below 101 for m ¼ 100 and 1000 mg l1 d1, and 102 for m ¼ 10,000 mg l1 d1, respectively. When flow over the sediment surface is present, SOD can be much larger than that for stagnant water (SOD with fluid flow [SOD for stagnant). The SODflow is obtained by Eq. (12), and is normalized by the SODave for the detention time T ¼ 1 day (SODstagnant) as the reference value. Fig. S4 illustrates that the ratio SODflow/SODstagnant depends on the biochemical oxygen uptake rate in the sediment (m). The ratio SODflow/ SODstagnant is about 2, 7, 20 and 60 for m ¼ 10, 100, 1000, and 10,000 mg l1 d1, respectively, when the shear velocity (U*) is large enough. The dependence of the ratio SODflow/SODstagnant on the biochemical oxygen uptake rate (m) is attributed to the fact that SODstagnant is limited by oxygen transfer in the water column, whereas biochemical oxygen consumption controls SODflow when the shear velocity (U*) is large enough. This suggests that flow over the sediment surface hardly raises SOD much larger than that for stagnant water when the biochemical oxygen uptake rate (m) is small. The difference between SOD with fluid flow and SOD for stagnant water becomes significant as the biochemical oxygen uptake rate (m) becomes larger, i.e. SOD is 10e100 times larger when flow is present over highly organic sediments.
7.
SOD for hyporheic exchange
This paper considers impermeable sediments. When sediments consist of permeable materials, e.g. sand and gravel, “hyporheic exchange” can significantly raise the SOD. “Hyporheic exchange” occurs when water flows over the permeable sediment surface, and is due to interstitial flow driven by a spatially and temporally variable pressure distribution at the sediment/water interface. Interstitial flow can be driven by surface waves (Huettel and Webster, 2001; Qian et al., 2009), bedforms (e.g. Elliott and Brooks, 1997; Packman et al., 2004; Cardenas and Wilson, 2006), or near-bed coherent motions (Higashino et al., 2009). When the interstitial water motion is present due to pressure fluctuations at the sediment/water interface induced by near-bed coherent motions, the SOD for larger oxidation rate e.g. m ¼ 1000 and 2000 mg l1 d1 is almost 5 times that with no pore water flow (Higashino and Stefan, 2011).
8.
4387
Conclusions
This paper considers sedimentary oxygen demand (SOD) for stagnant water, and investigates how flow over the sediment surface raises SOD compared to stagnant water, depending on flow velocity and biochemical activity in the sediment mainly in manmade reservoirs (Fig. 1). A model was developed for stagnant water, and results were compared to those with fluid flow (Higashino et al., 2004). Specific conclusions are as follows: 1. The DO profile varies with time in the water column, but does not change significantly in the sediment. The normalized DO concentration at the sediment/water interface (Cw/CN) is less than 0.1, and is close to zero regardless of the biochemical oxygen uptake rate (m) in the sediment. The normalized DO concentration at the sediment/water interface (Cw/CN) with fluid flow is greater than 0.9 when biochemical oxygen uptake rate is small, i.e. m < 200 mg l1 d1 (Higashino et al., 2004). 2. Since the DO profile near the sediment/water interface, and corresponding DO flux at the sediment/water interface (SOD) varies with time when water is stagnant, the timeaveraged SOD (SODave) over a period of the detention time (T) was introduced (Eq. (11)). The value of SODave is large when the detention time (T) is short, and diminishes as the detention time becomes longer. The SODave decreases to 21, 8, and 4% of that for T ¼ 1 h with increasing detention time, i.e. T ¼ 1 day, 1 week, and 1 month, respectively. 3. The SODave increases as the biochemical oxygen uptake rate (m) becomes larger. But the dependence of SODave on oxygen consumption in the sediment is not as strong as SOD with fluid flow (see Figs. 5, and 8 in Higashino et al., 2004). 4. The boundary layer thickness, i.e. a distance from the sediment/water interface to the point where C/CN ¼ 0.99, gets thicker as the detention time (T) becomes longer, and is independent of the biochemical oxygen uptake rate (m) in the sediment. 5. The DO penetration depth, i.e. a distance from the sediment/water interface to the point where C/CN ¼ 0.001, diminishes with increasing detention time (T). The DO penetration depth decreases as the biochemical oxygen uptake rate (m) becomes larger. The DO penetration depth (ds) is much smaller than that with fluid flow (Higashino et al., 2004). 6. SOD is controlled by oxygen transfer through the diffusive boundary layer and/or by biochemical oxygen uptake in the sediment when flow is present over the sediment surface (Higashino et al., 2004). However, SOD is substantially limited by oxygen transfer in the water column when water is stagnant. 7. When flow over the sediment surface is present, SOD becomes larger than that for stagnant water depending on biochemical oxygen uptake rate (m) in the sediment, i.e. SOD with fluid flow (SODN, when the shear velocity is large) is about 2, 7, 20 and 60 times larger than that for stagnant water (SODave for T ¼ 1day) for m ¼ 10, 100, 1000, and 10,000 mg l1 d1, respectively. 8. Flow over the sediment surface hardly raises SOD much larger than that for stagnant water when biochemical
4388
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 8 1 e4 3 8 9
oxygen uptake rate (m) is small. The difference between SOD with fluid flow and SOD for stagnant water becomes significant as the biochemical oxygen uptake rate (m) becomes larger, i.e. SOD is 10e100 times larger when flow over the sediment surface is present for highly organic sediments.
Acknowledgments This work was supported by the Japan Society for the Promotion of Science (Young Researcher Overseas Visit Program, Number 21-5018), and by JSPS Grant-in-Aid for Scientific Research (22560522). Two anonymous reviewers provided helpful comments and suggestions on the manuscript. The author is grateful to these individuals and organizations for their support.
Appendix. Supplementary data Supplementary data related to this article can be found online at doi:10.1016/j.watres.2011.04.051.
references
Barcelona, M.J., 1983. Sediment oxygen demand fractionation, kinetics and reduced chemical substances. Water Research 17, 1081e1093. Bear, J., 1972. Dynamics of fluids in porous media. American Elsevier Publishing Campany, New York. Boudreau, B.P., Joergensen, B.B. (Eds.), 2001. The Benthic Boundary Layer: Transport Processes and Biogeochemistry. Oxford University Press, UK. Boudreau, B.P., 1997. A one-dimensional model for bed boundary layer particle exchange. Journal of Marine Systems 11, 279e303. Cardenas, M.B., Wilson, J.L., 2006. The influence of ambient groundwater discharge on exchange zones induced by current-bedform interactions. Journal of Hydrology 331, 103e109. Chapra, S.C., 1997. Surface Water Quality Modeling. McGraw Hill. Dade, W.B., 1993. Near-bed turbulence and hydrodynamic control of diffusional mass transfer at the sea floor. Limnology and Oceanography 38 (1), 52e69. Denny, M.W., 1993. Air and Water. Princton University Press, Princton, N.J. Ditoro, D.M., 2001. Sediment Flux Modeling. Wiley, New York. Elliott, A.H., Brooks, N.H., 1997. Transfer of non-sorbing solutes to a streambed with bed forms: theory. Water Resources Research 33 (1), 123e136. Gantzer, C.J., Stefan, H.G., 2003. A model of microbial activity in lake sediments in response to periodic water column mixing. Water Research 37, 2833e2846. Hargrave, B.T., 1972. Aerobic decomposition of sediment and detritus as a function of particle surface area and organic content. Limnology and Oceanography 17, 583e596. Higashino, H.G., Stefan, M., 2011 . Dissolved oxygen demand at the sediment-water interface of a stream: near-bed turbulence and pore water flow effects. Journal of Environmental Engineering, ASCE 137 (7) pp. Higashino, M., Clark, J.J., Stefan, H.G., 2009. Porewater flow due to near-bed turbulence and associate solute transfer in a stream
or lake sediment bed. Water Resources Research 45 (12), W12414. doi:10.1029/2008WR007374. Higashino, M., Gantzer, C.J., Stefan, H.G., 2004. Unsteady diffusional mass transfer at the sediment/water interface: theory and significance for SOD measurements. Water Research 38, 1e12. Hondzo, M., Feyaerts, T., Donovan, R., O’Connor, B.L., 2005. Universal scaling of dissolved oxygen distribution at the sediment-water interface: a power law. Limnology and Oceanography 50 (5), 1667e1676. House, W.A., 2003. Factors influencing the extent and development of the oxic zone in river-bed sediment. Biogeochemistry 63, 317e333. Huettel, M., Webster, I.T., 2001. Porewater flow in permeable sediment. In: Boudreau, B.P., Joergensen, B.B. (Eds.), The Benthic Boundary Layer: Transport Processes and Biogeochemistry. Oxford University Press, UK, pp. 144e179. Iversen, N., Jørgensen, B.B., 1993. Diffusion coefficients of sulfate and methane in marine sediment. Geochimica et Cosmochimica Acta 57, 571e578. Josiam, R., Stefan, H.G., 1999. Effect of flow velocity on sediment oxygen demand: comparison of theory and experiments. Journal of the American Water Resources Association 35 (2), 433e439. Joergensen, B.B., Revsbech, N.P., 1985. Diffusive boundary layers and the oxygen uptake of sediment and detritus. Limnology and Oceanography 30 (1), 111e122. Joergensen, B.B., DesMarais, D.J., 1990. Diffusive boundary layer of sediments: oxygen microgradients over a microbial mat. Limnology and Oceanography 35 (6), 1343e1355. Mackenthun, A., Stefan, H.G., 1998. Effect of flow velocity on sediment oxygen demand: laboratory measurements. Journal of Environmental Engineering, ASCE 12 (3), 222e230. Nakamura, Y., Stefan, H.G., 1994. Effect of flow velocity on sediment oxygen demand: theory. Journal of Environmental Engineering, ASCE 120 (5), 996e1016. O’Connor, B.L., Hondzo, M., 2008. Dissolved oxygen transfer to sediments by sweep and eject motions in aquatic environment. Limnology and Oceanography 53 (2), 566e578. O’Connor, B.L., Hondzo, M., Harvey, J.W., 2008. Incorporating both physical and kinetic limitations in quantifying dissolved oxygen flux to aquatic sediments. Journal of Environmental Engineering, ASCE 135 (12), 1304e1314. Packman, A.I., Salehin, M., Zaramella, M., 2004. Hyporheic exchange with gravel beds: basic hydrodynamic interactions and bedform-induced advective flows. Journal of Hydraulic Engineering, ASCE 130 (7), 647e656. Pamatmat, M.M., 1971. Oxygen consumption by the seabed. 4. Shipboard and laboratory experiments. Limnology and Oceanography 16, 536e550. Qian, Q., Clark, J.J., Voller, V.R., Stefan, H.G., 2009. Depthdependent dispersion coefficient for modeling of vertical solute exchange in a lake bed under surface waves. Journal of Hydraulic Engineering, ASCE 135 (3), 187e197. doi:10.1061/ (ASCE)0733-9429. Rahm, L., Svensson, U., 1989. On the mass transfer properties of the benthic boundary layer with an application to oxygen fluxes. Netherlands Journal of Sea Research 24 (1), 27e35. Roy, H., Huttel, M., Jorgensen, B.B., 2004. Transition of oxygen concentration fluctuations through the diffusive boundary layer overlying aquatic sediments. Limnology and Oceanography 49, 686e692. Smith Jr., K.L., 1978. Benthic community respiration in the N.W. Atlantic Ocean: in situ measurements from 40 to 5200 m. Marine Biology 47, 337e347. Steinberger, N., Hondzo, M., 1999. Diffusional mass transfer at the sediment-water interface. Journal of Environmental Engineering, ASCE 125 (2), 192e200.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 8 1 e4 3 8 9
Nomenclature C: dissolved oxygen concentration (mg l1) CN: dissolved oxygen concentration in the bulk water (mg l1) Cw: dissolved oxygen concentration at the sediment/water interface (mg l1) C*: normalized dissolved oxygen concentration Dw: molecular diffusivity of dissolved oxygen in water (cm2 s1) Ds: effective diffusion coefficient for oxygen in the sediment (cm2 s1) Ko2: half-saturation constant for dissolved oxygen (mg l1) Sc: the Schmidt number SOD: sedimentary oxygen demand (g m2 d1) SODave: time-averaged SOD (g m2 d1) SODflow: SOD with fluid flow (g m2 d1)
4389
SODstagnant: SOD for stagnant water (g m2 d1) T: detention time of water in a lake or a reservoir (h) U*: bed shear velocity (cm s1) y: vertical coordinate (cm) d: sediment layer thickness (cm) dw: boundary layer thickness above the sediment/water interface (cm) ds: dissolved oxygen penetration depth in the sediment (cm) m: maximum oxidation rate (mg l1 d1) m0: zero-order rate constant (mg l1 d1) q: sediment tortuosity r: fluid (water) density (g cm3) s: bed shear stress (N cm2) 4: sediment porosity
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 9 0 e4 3 9 8
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Analysis of the bacterial community in a laboratory-scale nitrification reactor and a wastewater treatment plant by 454-pyrosequencing Lin Ye a, Ming-Fei Shao a, Tong Zhang a,*, Amy Hin Yan Tong b, Si Lok b a b
Environmental Biotechnology Laboratory, The University of Hong Kong, Hong Kong SAR, China Genome Research Center, The Li Ka Shing Faculty of Medicine, The University of Hong Kong, Hong Kong SAR, China
article info
abstract
Article history:
For full understanding of the microbial community in the wastewater treatment bioreac-
Received 25 January 2011
tors, one of the feasible and effective ways is to investigate the massive genetic informa-
Received in revised form
tion contained in the activated sludge. In this study, high-throughput pyrosequencing was
20 March 2011
applied to analyze the 16S rRNA gene of bacteria in a laboratory-scale nitrification reactor
Accepted 22 May 2011
and a full-scale wastewater treatment plant. In total, 27,458 and 26,906 effective sequence
Available online 31 May 2011
reads of the 16S rRNA gene were obtained from the Reactor and the wastewater treatment plant activated sludge samples respectively. The taxonomic complexities in the two
Keywords:
samples were compared at phylum and genus levels. According to the pyrosequencing
Activated sludge
results, even for a laboratory-scale reactor as simple as that in this study, a small size clone
Bacterial community
library is far from enough to reflect the whole profile of the bacterial community. In
Cloning
addition, it was found that the commonly used informatics tool “RDP classifier” may
454 High-throughput
drastically assign Nitrosomonas sequences into a wrong taxonomic unit resulting in
pyrosequencing
underestimation of ammonia-oxidizing bacteria in the bioreactors. In this paper the reasons for this mistakenly assignment were analyzed and correction methods were proposed. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
Nitrification is an important step for removing ammonium nitrogen from wastewater. The key role that bacterial communities play in wastewater treatment has been intensively studied in the past decades in laboratory-scale and fullscale bioreactors by the use of various molecular methods. PCR-DGGE (polymerase chain reaction - denaturing gradient gel electrophoresis) and T-RFLP (terminal restriction fragment length polymorphism) were used to investigate the diversity of the bacterial communities of activated sludge from different wastewater treatment plants (WWTPs) (Boon et al., 2002; Regan et al., 2002). More recently, analysis of clone
library (Matsumoto et al., 2009) as well as direct visualization of bacterial species by FISH (fluorescent in situ hybridization) (Hao et al., 2009) were employed to determine the community compositions. Although bacterial species involved in nitrification have also been characterized by the aforementioned approaches, the extraordinary diversity of microorganisms in the activated sludge exceeds the sensitivity and dynamic range of those molecular methods and precludes a complete characterization of the interplay of various components of the microbial community in nitrification. Pyrosequencing developed by Roche 454 Life Science (Branford, CT, USA) is a high-throughput analytical method that can generate huge amounts of DNA reads through
* Corresponding author. Fax: þ86 852 2559 5337. E-mail address:
[email protected] (T. Zhang). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.05.028
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 9 0 e4 3 9 8
a massively parallel sequencing-by-synthesis approach (Margulies et al., 2005). This technology have been used widely to analyze the microbial community in various environmental samples, such as marine water (Qian et al., 2011), soil (Roesch et al., 2007), human distal intestine (Claesson et al., 2009), wastewater treatment plant influent (McLellan et al., 2010) et al.. However, few studies have been conducted on activated sludge by this method. Kwon and colleagues (Kwon et al., 2010) investigated the microbial diversity in an integrated fixed-film activated sludge system. Their results showed the bacterial abundances were quite high, totally 3034 and 1451 operational taxonomic units (OTUs) were identified at the 3% cutoff for the suspended and attached samples, respectively. In the present study, we characterized and compared the bacterial communities in a laboratory-scale nitrification reactor with that from activated sludge of a wastewater treatment plant by 454-pyrosequencing. The diversity and abundance of the nitrifiers in these two samples were also investigated. Additionally, we found that RDP Classifier, a commonly used informatics tool for pyrosequencing data analysis, may have phylogenetically assigned Nitrosomonas sequences into a wrong order. The key nucleotides leading to the mistaken assignment have been identified and a method to overcome this problem was proposed based on the BLAST results.
4391
added to the influent to get the ammonia nitrogen concentration of 200 mg/L, In addition, 20 mg/L of KH2PO4 was added into the influent to provide sufficient phosphorus for the growth of microorganisms in the Reactor. Without adding organic matter, the total organic carbon (TOC) of the influent was as low as 0.64 0.05 mg/L. The hydraulic retention time (HRT) of the Reactor was 18.4 h. The Reactor was shielded with aluminum foil to avoid exposure to light. Prior to the present study, the Reactor was ran continuously for more than 500 days to investigate the ammonia-oxidizing bacteria (AOB) and nitrite-oxidizing bacteria (NOB) under different conditions (Jin et al., 2010; Ye and Zhang, 2010). Shatin WWTP is a full-scale wastewater treatment plant in Hong Kong. This WWTP treats saline sewage (salinity 1.2%) with a four-stage process (anoxiceaerobiceanoxiceaerobic) that may simultaneously remove organic compounds and nitrogen. The seed sludge of the Reactor described above was taken from Tank No.16 of the first aerobic stage. Activated sludge sample used to perform 454-pyrosequencing analysis was also taken from the same tank.
2.2.
Chemical analysis
2.
Materials and methods
Concentrations of ammonium, nitrite and nitrate were measured according to the Standard Methods (Eaton and Franson, 2005) by Nesslerizaion Method, Colorimetric Method and Ultraviolet Spectrophotometric Screening Method, respectively.
2.1.
Reactor operation and WWTP description
2.3.
In the present study, a fermentor (Sartorius Biostat A plus) (Goettingen, Germany) with a working volume of 2.6 L was configured for continuous operation to conduct nitrification studies. A pH value of 7.5 was held by automated addition of sodium bicarbonate. Dissolved Oxygen (DO) was maintained at 0.5 mg/L by stirring and aeration. The influent was made with deionized water (67%) and seawater (33%) to simulate the typical salinity of sewage found in Hong Kong. NH4Cl was
DNA extraction and PCR
Sludge samples of Day 165, 178, 190, and 201 (as indicated by the inverted triangle in Fig. 1) were taken from the Reactor for DNA extraction using FastDNA SPIN Kit for Soil (MP Biomedicals, Illkirch, France). For clone library construction, the above DNA mixture was pooled and amplified by PCR using primer set EUB8F (50 -AGAGTTTGATCMTGGCTCAG-30 ) (Heuer et al., 1997) and UNIV1392R (50 -ACGGGCGGTGTGTRC30 ) (Ferris et al., 1996). 30 ml PCR mixture contained 0.2 ml of Ex
Fig. 1 e Nitrogen concentration in the influent and effluent (The 4 inverted open triangles indicate the time points of sludge samples for cloning and pyrosequencing analysis).
4392
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 9 0 e4 3 9 8
Taq TM(TaKaRa, Dalian, China), 3 ml of 10 Ex Taq Buffer, 3 ml of dNTP mixture, 0.2 mM of each primer, and 20e50 ng of genomic DNA. The thermocycling steps were as follows: 95 C for 7 min, followed by 35 cycles at of 95 C for 1 min, 55 C for 1 min, 72 C for 1 min and a final extension step at 72 C for 10 min. For pyrosequencing, the above DNA mixture of the four samples was amplified with a set of primers targeting the hypervariable V4 region of the 16S rRNA gene (RDP’s Pyrosequencing Pipeline: http://pyro.cme.msu.edu/pyro/help.jsp). The forward primer is 50 -AYTGGGYDTAAAGNG-30 and the reverse primers are the mixture of four primers, i.e. 50 TACCRGGGTHTCTAATCC-30 , 50 -TACCAGAGTATCTAATTC-30 , 50 -CTACDSRGGTMTCTAATC-30 , and 50 -TACNVGGGTATCTAATCC-30 (Claesson et al., 2009). Barcodes that allow sample multiplexing during pyrosequencing were incorporated between the 454 adapter and the forward primers.
2.4.
Cloning
PCR products were purified using PCRquick-spinTM PCR Product Purification Kit (iNtRON Biotechnology, SangdaewonDong, Korea). The purified PCR products were cloned using the InsTAclone PCR Cloning Kit (Fermentas, Burlingtong, Ontario, Canada) following the instructions of the vendor. White colonies were selected for whole-cell PCR amplification with the M13F (50 -TGTAAAACGACGGCCAGT-30 ) and M13R (50 CAGGAAACAGCTATGAC-30 ) primer set. The PCR products were purified and sequenced on the ABI 3730xl capillary sequencer (Applied Biosystems, Foster City, CA, USA) using M13F or M13R primers. The clone library sequences in this study have been deposited in GenBank under accession numbers HM117160 to HM117171.
2.5.
qblast function in Biopython to run BLAST and search against “nr” database through the internet automatically for all the above-mentioned extracted sequences; 3) parse the BLAST results to check the top 10 hits whether there is a hit containing Nitrosomonas or Comamonas in the title; and 4) record the maximum identities between the query sequences and the subject sequences identified in last step. On the other hand, all sequences obtained from pyrosequencing in this study were compared with Greengenes 16S rRNA gene database (DeSantis et al., 2006) using NCBI’s BLASTN tool (Altschul et al., 1990) and the default parameters except for the maximum hit number of 100 (Claesson et al., 2009). Then the sequences were assigned to NCBI taxonomies with MEGAN (Huson et al., 2007) by using the Lowest Common Ancestor (LCA) algorithm and the default parameters, i.e. absolute cutoff: BLAST bitscore 35, and relative cutoff: 10% of the top hits.
3.
Results and discussion
3.1.
Reactor performance
Prior to this study, the Reactor was operated under a very low oxygen concentration condition (DO 0.15 mg/L). Accordingly, the bulk of the ammonium was partially oxidized to nitrite. In present study, the DO level was increased to 0.5 mg/L, the nitrite was gradually reduced and several molecular methods (DGGE, T-RFLP and Cloning) have been used to confirm that nitrite-oxidizing bacteria (Nitrospira) proliferated intensively in this period (Ye and Zhang, 2010). The operational condition and the performance of the Reactor were described in detail in our previous paper (Ye and Zhang, 2010).
High-throughput 454 pyrosequencing 3.2.
The composition of the PCR products of V4 region of 16S rRNA gene was determined by pyrosequencing using the Roche 454 FLX Titanium sequencer (Roche 454 Life Sciences, Branford, CT, USA). Samples in this study were individually barcoded to enable multiplex sequencing. The results are deposited into the NCBI short reads archive database (Accession Number: SRA026842.2).
2.6.
Sequence analysis and phylogenetic classification
Following pyrosequencing, Python scripts were written to: 1) remove sequences containing more than one ambiguous base (‘N’); 2) check the completeness of the barcodes and the adapter; 3) remove sequences shorter than 150 bps. The “RDP Align” tool in RDP’s Pyrosequencing Pipeline was used to align the effective sequences. A cluster file was generated with “RDP Complete Linkage Clustering” tool. From the cluster file, the rarefaction curve was generated using the “RDP Rarefaction” tool. Taxonomic classification of the sequences was performed using the RDP Classifier (Version 2.2) with a set confidence threshold of 50%. Python and Biopython (Cock et al., 2009) were used to create scripts to: 1) extract all sequences that were assigned into the order of Burkholderiales according to the result of RDP Classifier (the downloaded assignment detail text file); 2) use
Cloning results
Twelve OTUs were obtained with 3% nucleotide cutoff from 61 clones that were sequenced in 16S rRNA gene clone library. According to RDP Classifier, the 61 sequences examined by Sanger-dideoxy based sequencing can be assigned to 3 phyla. Based on both RDP Classifier and BLAST analysis (Table 1), OTU-2 and OTU-4 are Nitrosomonas and Nitrospira species, which accounted for 21.3% and 3.2% in total bacterial community, respectively, suggesting that these species represent the dominant AOB and NOB in the Reactor. It should be noted that according to the BLAST results, the most dominant OTU, OTU-1, is probably a heterotrophic species that is similar (max identity 99%) to an uncultured bacteria reported in the marine sediment.
3.3.
Taxonomic complexity of the bacterial community
Pyrosequencing of the Reactor sample and the WWTP sludge sample yielded 27,458 and 26,906 effective sequence tags respectively. The amount of sequences was comparable to those of other studies (Kwon et al., 2010; Lee et al., 2010; McLellan et al., 2010). RDP Classifier was firstly used to assign these sequence tags into different phylogenetic bacterial taxa. Fig. 2 and Supplementary Table S1 show the relative bacterial community abundances on the phylum level. Except
4393
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 9 0 e4 3 9 8
Table 1 e The affiliation and closest match of the bacterial OTUs. OTU 1 2 3 4 5 6 7 8 9 10 11 12
Clones
Percentage
Genus assignment based on RDP classifier [Probability]
Closest match from BLAST [Max Identity]
34 13 3 2 2 1 1 1 1 1 1 1
55.7% 21.3% 4.9% 3.2% 3.2% 1.6% 1.6% 1.6% 1.6% 1.6% 1.6% 1.6%
Coxiella [35%] Nitrosomonas [56%] Oleiphilus [10%] Nitrospira [100%] Loktanella [40%] Aminobacter [56%] Azoarcus [56%] Muricauda [100%] Adhaeribacter [31%] Roseivirga [74%] Roseivirga [45%] Marinicola [21%]
Uncultured bacterium [99%] Nitrosomonas sp. [99%] Uncultured geproteobacterium [99%] Uncultured Nitrospira sp. [98%] Uncultured bacterium [94%] Uncultured bacterium [94%] Denitromonas indolicum [97%] Muricauda sp. [98%] Uncultured bacterium [91%] Uncultured bacterium [98%] Uncultured bacterium [97%] Uncultured bacterium [91%]
for a small number of minor phyla accounting for no more than 0.6% of total community found only in WWTP sludge, the numbers of tabulated phylum present in the Reactor and the WWTP sludge were nearly identical. Proteobacteria, Firmicutes and Bacteroidetes were three phyla that were abundant in both samples. These three phyla were also ubiquitous in soil (Fierer et al., 2007). However, Actinobacteria and Chloroflexi were much less abundant in the Reactor than in the WWTP. Heterotrophic bacteria in these phyla may be depleted in the Reactor under such an oligotrophic environment. It was also observed that Nitrospira phylum in the Reactor was significantly more abundant than that in the WWTP sludge, suggesting that the elevated level of nitrite and oxygen present in the Reactor may have favored the propagation of these nitrite-oxidizing bacteria. Notably, although the organic matter in the influent of the nitrification reactor was very low, the heterotrophic bacteria were still dominant over the autotrophic bacteria. The carbon source of these heterotrophic bacteria were probably from soluble microbial products (SMP) in the Reactor (Barker and Stuckey, 1999; Rittmann and McCarty, 2001). 55%
454-Pyprosequencing provides at least three logs or more sensitivity over conventional Sanger-dideoxy based sequencing for assessing microbial diversity. Fig. 3 showed the diversity of bacteria in the Reactor was significantly reduced after 500 days’ operation compared with the seed sludge from WWTP. The diversity reduction from seeding sludge to lab-scale reactor was usually investigating by DGGE and cloning previously (Liu et al., 2002). Heterotrophic bacteria were greatly depleted in the Reactor. Especially, the phylum Actinobacteria, where most members are heterotrophs (Servin et al., 2008), were dramatically reduced (Fig. 2). While the absence of light under Reactor conditions avoids perturbing the growth of the nitrifiers (Sinha and Annachhatre, 2007), light dependent bacteria were depleted. A marked decrease of the phototropic Chloroflexi phylum (Holt et al., 1994) (Fig. 2) in the Reactor was observed compared with the seed sludge. In order to further compare the microbial communities of the two samples from the Reactor and WWTP, all-against-all comparison was conducted by using the MEGAN software. The sequences in each of the samples were normalized before doing the comparison. The tree created by MEGAN was shown in Fig. 4. The pie charts beside the leaves of the tree indicate the relative abundance of the genus in the two samples. From
50%
Reactor WWTP
45% 40%
3500
Percentage
35% 30%
3% Reactor 5% Reactor 3% WWTP 5% WWTP
3000
25% 2500
20%
OTUs
15% 10%
2000 1500
5% 1000
0% i us ia e M7 lex es ria tes etes pira e ria etes eria b ia f u te et o te T t t rm yd s ro ac Bac m ic r oid itro bac my c ba c ic r h e ha o m l b c r la e o co m s -T N ino c to Fi e o i ed Ch piro ct id t u Ch ot f n S Ba Ac erru c c Ac Pl a Pr ss i o V a c l o c in e Un D ria
Phylum name
Fig. 2 e Bacterial community compositions at phylum level revealed by pyrosequencing.
500 0 0
5000
10000
15000
20000
25000
30000
Number of sequences
Fig. 3 e Rarefaction curves of OTUs defined by 3% and 5% distances in Reactor and WWTP sludge samples.
4394
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 9 0 e4 3 9 8
Fig. 4 e Sequences from the Reactor and WWTP assigned into NCBI taxonomies with BLAST and MEGAN. (Pie charts indicate the relative abundance for each genus. The ratio of gray color area to dark color area in each pie represented the ratio of the relative abundance of the corresponding genus in WWTP to that in the Reactor.)
4395
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 9 0 e4 3 9 8
Fig. 4, it could be seen that at the genus level the microbial communities of the Reactor and the WWTP were quite different. There were some genera (such as Marinobacter, Pseudomonas, Aequorivita, Muricauda, etc.) appearing only in the Reactor. According to the previous reports (Bowman and Nichols, 2002; Gauthier et al., 1992; Yoon et al., 2005), these genera are usually halotolerant bacteria and exist in the marine environment. These bacteria may come from the seawater of the influent and could adapt themselves to the conditions of the Reactor. Many other genera, which were marked by gray color in Fig. 4, exist only in the WWTP. That indicates the diversity of the bacteria in the WWTP was much more complex than that in the Reactor. Also, there are some genera, including Phyllobacteriaceae, Rhodobacteraceae, Chromatiales, Comamonadaceae, Nitrospira, Rhodococcus, etc., exist both in the Reactor and the WWTP. According to the cluster files produced by the “RDP Complete Linkage Clustering” tool, there were 494 and 381 OTUs in Reactor sludge using two cutoffs levels of 3% and 5%, respectively. By contrast, WWTP sludge had 1986 and 1648 OTUs using the same cutoffs. According to the rarefaction curve (Fig. 3), the species complexity in the Reactor was sixfold less than that in WWTP. Judging from the numbers of OTUs obtained by pyrosequencing in this study, not to mention the WWTP, even for a laboratory reactor as simple as that in our study, a small scale clone library was not sufficient to reflect the whole profile of the bacterial community, especially for those minor populations. Furthermore, it can be deduced that the commonly used molecular methods (such as DGGE and T-RFLP) in environmental biotechnology may also have insufficient resolutions to characterize the microbial communities in the wastewater treatment bioreactors. High-throughput sequencing methods have the potential to be effective means for better understanding of the microorganisms in various environmental engineering facilities.
3.4.
Diversity and abundances of AOB and NOB
It was found that in the results of RDP Classifier (Table 2), the sequences that were assigned into Nitrosomonadales order were quite few. For the Reactor sludge sample, only 0.65% of the sequences were classified into this order, which was inconsistent with the limited results from Sangerdideoxy sequencing of the clone library. Such low abundance of AOB also conflicted with the performance (high ammonium removing rate) of the Reactor. A further check showed that there were large numbers of Nitrosomonadales
sequences that were wrongly assigned by the RDP Classifier into the order of Burkholderiales, a neighbor of Nitrosomonadales in Proteobacteria phylum. Most of these misassigned sequences were closely related (identity>97%) to Nitrosomonas species as shown by BLAST analysis. Following reclassification, 15.54% of the sequences from the reactor sludge sample were Nitrosomonas related, indicating that Nitrosomonas-like AOB were remarkably enriched in the reactor (Table 2). In both the Reactor and the WWTP sludge, the dominant AOB species was Nitrosomonas and the dominant NOB species was Nitrospira, which were affiliated to Nitrosomonadales and Nitrospirales order, respectively. This result was consistent with the previous reports of activated sludge from other researchers (Layton et al., 2005; Logemann et al., 1998). Except Nitrosomonas and Nitrospira, the other AOB and NOB genus were very rare in the reactor. Nitrosospira, another genus belonging to b-Proteobacteria AOB, accounted for only 0.6% and 0.056% in the reactor and the WWTP sludge samples, respectively. Only one sequence and seven sequences of Nitrosococcus were found in 27,458 and 26,906 sequences in the Reactor and WWTP sludge samples, respectively. For NOB, only 13 Nitrobacter sequences were found in the WWTP sludge and none was found in the reactor. The present results suggested that except Nitrosomonas and Nitrospira, all other species of bacterial nitrifiers play only a very small role in nitrification process in the wastewater treatment reactors.
3.5.
Mistakenly classified Nitrosomonas sequences
In this study, as aforementioned, RDP Classifier may wrongly assign sequences in Nitrosomonadales order into another order, consequently, leading to an underestimation of the abundance of AOB in the samples. Except for the samples in this study, we also investigated the pyrosequencing results of other 14 samples from WWTPs in another study, a lot of sequences were mistakenly assigned to Burkholderiales. Thirty sequences, including 10 from mistakenly assigned Nitrosomonas sequences, 10 from correctly assigned Nitrosomonas sequences, and 10 from Comamonas (a genus in Burkholderiales) sequences, were selected from the sequences obtained by the pyrosequencing, and used to draw a Neighbor-joining phylogenetic tree using Jukes-Cantor model (Fig. 5). It was found that the Nitrosomonas sequences and Comamonas sequences can be clearly classified into different groups, indicating that there are marked differences
Table 2 e Relative abundance of dominant AOB and NOB in the Reactor and WWTP sludge. Sample
Reactor WWTP
AOB/NOB
Nitrosomonadales Nitrospirales Nitrosomonadales Nitrospirales
RDP Classifier
Clone Library
BLAST-corrected RDP Classifier
MEGAN Assignment
Reads
%
Reads
%
Reads
%
Reads
%
179 1782 14 273
0.65% 6.49% 0.05% 1.01%
13 2 e e
21.3% 3.2% e e
4059 1782 14 273
14.78% 6.49% 0.05% 1.01%
4268 1817 13 273
15.54% 6.61% 0.05% 1.01%
4396
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 9 0 e4 3 9 8
Nitro_M-GPQYMAQ01D2WQ6 Nitro_M-GPQYMAQ01CPALL Nitro_M-GPQYMAQ01EE6QD Nitro_M -GPQYMAQ01DCDM8 Nitro_M-GPQYMAQ01BBFHN 40
I
Nitro_M-GPQYMAQ01B7TNH Nitro_M-GPQYMAQ01D96JO Nitro_M-GPQYMAQ01A2KAW
43
Nitro_M-GPQYMAQ01EIO8Z Nitro_M-GPQYMAQ01EI4S3
72
Nitro-GPQYMAQ01A9K6V Nitro-GPQYMAQ01E0QA6 Nitro-GPQYMAQ01C68KE Nitro-GPQYMAQ01DEH7U Nitro-GPQYMAQ01CCBG2
90
Nitro-GPQYMAQ01AHLH1
98
II
Nitro-GPQYMAQ01AYECJ
60
Nitro-GPQYMAQ01CXQ2B
86
67 Nitro-GPQYMAQ01BQZAW 31 75 82
Nitro-GPQYMAQ01BEAJB Comam-GPQYMAQ01EM28M Comam-GPQYMAQ01CGVBO
Comam-GPQYMAQ01BYP7R Comam-GPQYMAQ01EUM1H
100
Comam-GPQYMAQ01C0W9M 62
III
Comam-GPQYMAQ01ELIAF
33
Comam-GPQYMAQ01B2DQT
33
Comam-GPQYMAQ01DJK4R
25 Comam-GPQYMAQ01DPQIB
Comam-GPQYMAQ01AYTQ6
0.01
Fig. 5 e Neighbor-joining phylogenetic tree using Jukes-Cantor model of Nitrosomonas sequences and Comamonas sequences based on V4 region of 16S rRNA gene sequences (I - Mistakenly assigned Nitrosomonas sequences, II e Correctly assigned Nitrosomonas sequences, III e Comamonas sequences).
Fig. 6 e The key nucleotides that caused the mistakenly assigned Nitrosomonas sequences (I - Mistakenly assigned Nitrosomonas sequences, II e Correctly assigned Nitrosomonas sequences).
in the sequences of these two groups. It indicates the possible unrecognized deficiencies of the RDP Classifier. Further examination of these sequences reveals that the correct assignment of Nitrosomonas related sequences into the Nitrosomonas genus is dependent on a key dinucleotide position, as show in Fig. 6. If the ‘GC’ dincucleotide at position 106 of group I was changed to ‘AT’, as it is found in group II, the RDP Classifier would then assign group I correctly into Nitrosomonas. It is therefore advised that the phylogenetic results of Nitrosomonas species obtained using the RDP Classifier should be cross-validated by other independent tools such as Greengenes’ classification tool (DeSantis et al., 2006) and GAST (Huse et al., 2008). Accordingly, we have developed a batch BLAST method, which has been described in the materials and methods part, to confirm the suspicious sequences.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 9 0 e4 3 9 8
4.
Conclusions
The diversity of the bacterial community in the nitrification reactor was significantly reduced compared with seeding sludge from the WWTP after 500 days’ operation. While, the bacteria both in the reactor and the seeding sludge distributed almost over the same phyla. RDP classifier is a powerful tool for pyrosequencing data analysis but it appears to misassign Nitrosomonas sequences into wrong taxonomic rank. Some other tools, such as Blast and Greengenes classification tool, can be used to correct the results of the RDP classifier. We also developed a batch Blast method to confirm the suspicious sequences. According to the pyrosequencing results, for such a reactor, a small scale clone library is not enough to reflect the profile of the bacterial community. Although the influent of the nitrification reactor contained nearly no organic matter, the heterotrophic bacteria were still dominant and much more than autotrophic bacteria in the reactor.
Acknowledgments Dr. Ming-Fei Shao thanks HKU for the postdoctoral fellowship. Lin Ye thanks HKU for the postgraduate studentship. We also would like to thank Hong Kong General Research Fund (HKU7197-08E) for financial support of this study and W Chan and CK Wong for technical help in pyrosequencing.
Appendix. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.watres.2011.05.028.
references
Altschul, S., Gish, W., Miller, W., Myers, E., Lipman, D., 1990. Basic local alignment search tool. J. Mol. Biol. 215, 403e410. Barker, D.J., Stuckey, D.C., 1999. A review of soluble microbial products (SMP) in wastewater treatment systems. Water Res. 33, 3063e3082. Boon, N., Windt, W., Verstraete, W., Top, E., 2002. Evaluation of nested PCReDGGE (denaturing gradient gel electrophoresis) with group-specific 16S rRNA primers for the analysis of bacterial communities from different wastewater treatment plants. FEMS Microbiol. Ecol. 39, 101e112. Bowman, J., Nichols, D., 2002. Aequorivita gen. nov., a member of the family Flavobacteriaceae isolated from terrestrial and marine Antarctic habitats. Int. J. Syst. Evol. Microbiol. 52, 1533. Claesson, M., O’Sullivan, O., Wang, Q., Nikkila, J., Marchesi, J., Smidt, H., De Vos, W., Ross, R., O’Toole, P., 2009. Comparative analysis of pyrosequencing and a phylogenetic microarray for exploring microbial community structures in the human distal intestine. PloS One 4, e6669. Cock, P., Antao, T., Chang, J., Chapman, B., Cox, C., Dalke, A., Friedberg, I., Hamelryck, T., Kauff, F., Wilczynski, B., 2009. Biopython: freely available Python tools for computational molecular biology and bioinformatics. Bioinformatics 25, 1422.
4397
DeSantis, T., Hugenholtz, P., Larsen, N., Rojas, M., Brodie, E., Keller, K. , Huber, T., Dalevi, D., Hu, P., Andersen, G., 2006. Greengenes, a chimera-checked 16S rRNA gene database and workbench compatible with ARB. Appl. Environ. Microbiol. 72, 5069. Eaton, A., Franson, M., 2005. Standard methods for the examination of water & wastewater. Am Public Health Assn. Ferris, M., Muyzer, G., Ward, D., 1996. Denaturing gradient gel electrophoresis profiles of 16S rRNA-defined populations inhabiting a hot spring microbial mat community. Appl. Environ. Microbiol. 62, 340e346. Fierer, N., Bradford, M.A., Jackson, R.B., 2007. Toward an ecological classification of soil bacteria. Ecology 88, 1354e1364. Gauthier, M., Lafay, B., Christen, R., Fernandez, L., Acquaviva, M., Bonin, P., Bertrand, J., 1992. Marinobacter hydrocarbonoclasticus gen. nov., sp. nov., a new, extremely halotolerant, hydrocarbon-degrading marine bacterium. Int. J. Syst. Evol. Microbiol. 42, 568. Hao, X., Wang, Q., Zhang, X., Cao, Y., Mark Loosdrecht, C.M., 2009. Experimental evaluation of decrease in bacterial activity due to cell death and activity decay in activated sludge. Water Res. 43, 3604e3612. Holt, J., Krieg, N., Sneath, P., Staley, J., Williams, S., 1994. Bergey’s Manual of Systematic Bacteriology, nineth ed. Williams and Wilkins. Heuer, H., Krsek, M., Baker, P., Smalla, K., Wellington, E., 1997. Analysis of actinomycete communities by specific amplification of genes encoding 16S rRNA and gelelectrophoretic separation in denaturing gradients. Appl. Environ. Microbiol. 63, 3233e3241. Huse, S., Dethlefsen, L., Huber, J., Welch, D., Relman, D., Sogin, M., 2008. Exploring microbial diversity and taxonomy using SSU rRNA hypervariable tag sequencing. PLoS Genet. 4, e1000255. Huson, D., Auch, A., Qi, J., Schuster, S., 2007. MEGAN analysis of metagenomic data. Genome Res. 17, 377e386. Jin, T., Zhang, T., Yan, Q., 2010. Characterization and quantification of ammonia-oxidizing archaea (AOA) and bacteria (AOB) in a nitrogen-removing reactor using T-RFLP and qPCR. Appl. Microbiol. Biotechnol. 87, 1167e1176. Kwon, S., Kim, T., Yu, G., Jung, J., Park, H., 2010. Bacterial community composition and diversity of a full-scale integrated fixed-film activated sludge system as investigated by pyrosequencing. J. Microbiol. Biotechnol. 20, 1717e1723. Layton, A., Dionisi, H., Kuo, H., Robinson, K., Garrett, V., Meyers, A., Sayler, G., 2005. Emergence of competitive dominant ammonia-oxidizing bacterial populations in a fullscale industrial wastewater treatment plant. Appl. Environ. Microbiol. 71, 1105. Lee, T.K., Van Doan, T., Yoo, K., Choi, S., Kim, C., Park, J., 2010. Discovery of commonly existing anode biofilm microbes in two different wastewater treatment MFCs using FLX Titanium pyrosequencing. Appl. Microbiol. Biotechnol., 1e9. Liu, W.T., Chan, O.C., Fang, H.H.P., 2002. Microbial community dynamics during start-up of acidogenic anaerobic reactors. Water Res. 36, 3203e3210. Logemann, S., Schantl, J., Bijvank, S., Loosdrecht, M., Kuenen, J., Jetten, M., 1998. Molecular microbial diversity in a nitrifying reactor system without sludge retention. FEMS Microbiol. Ecol. 27, 239e249. Margulies, M., Egholm, M., Altman, W., Attiya, S., Bader, J., Bemben, L., Berka, J., Braverman, M., Chen, Y., Chen, Z., 2005. Genome sequencing in microfabricated high-density picolitre reactors. Nature 437, 376e380. Matsumoto, S., Katoku, M., Saeki, G., Terada, A., Aoi, Y., Tsuneda, S., Picioreanu, C., van Loosdrecht, M., 2009. Microbial community structure in autotrophic nitrifying granules characterized by experimental and simulation analyses. Environ. Microbiol. 12, 192e206.
4398
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 9 0 e4 3 9 8
McLellan, S., Huse, S., Mueller Spitz, S., Andreishcheva, E., Sogin, M., 2010. Diversity and population structure of sewage derived microorganisms in wastewater treatment plant influent. Environ. Microbiol. 12, 378e392. Qian, P., Wang, Y., Lee, O., Lau, S., Yang, J., Lafi, F., AlSuwailem, A., Wong, T., 2011. Vertical stratification of microbial communities in the Red Sea revealed by 16S rDNA pyrosequencing. ISME J. 5, 507e518. Regan, J.M., Harrington, G.W., Noguera, D.R., 2002. Ammonia-and nitrite-oxidizing bacterial communities in a pilot-scale chloraminated drinking water distribution system. Appl. Environ. Microbiol. 68, 73e81. Rittmann, B., McCarty, P., 2001. Environmental Biotechnology: Principles and Applications. McGraw-Hill, New York. Roesch, L., Fulthorpe, R., Riva, A., Casella, G., Hadwin, A., Kent, A., Daroub, S., Camargo, F., Farmerie, W., Triplett, E., 2007.
Pyrosequencing enumerates and contrasts soil microbial diversity. ISME J. 1, 283e290. Servin, J., Herbold, C., Skophammer, R., Lake, J., 2008. Evidence Excluding the Root of the tree of Life from the Actinobacteria. Mol. Biol. Evol. 25, 1e4. Sinha, B., Annachhatre, A., 2007. Partial nitrificationdoperational parameters and microorganisms involved. Rev. Environ. Sci. Biotechnol. 6, 285e313. Ye, L., Zhang, T., 2010. Estimation of nitrifier abundances in a partial nitrification reactor treating ammonium-rich saline wastewater using DGGE, T-RFLP and mathematical modeling. Appl. Microbiol. Biotechnol. 88 (6), 1403e1412. Yoon, J., Lee, M., Oh, T., Park, Y., 2005. Muricauda flavescens sp. nov. and Muricauda aquimarina sp. nov., isolated from a salt lake near Hwajinpo Beach of the East Sea in Korea, and emended description of the genus Muricauda. Int. J. Syst. Evol. Microbiol. 55, 1015.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 9 9 e4 4 0 8
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Characterization of bottom sediments in lakes using hydroacoustic methods and comparison with laboratory measurements Michael A. Anderson*, Porfirio Pacheco Department of Environmental Sciences, University of California, Riverside, CA 92521, USA
article info
abstract
Article history:
The acoustical properties of bottom sediments in two lakes were shown to be strongly
Received 29 December 2010
correlated with clay content, organic C and total N concentrations, and other important
Received in revised form
sediment properties. The fractal dimension of the bottom echo was more strongly corre-
28 April 2011
lated with sediment physical and chemical properties than energy-based measures. The
Accepted 22 May 2011
fractal dimension was also related to rates of PO4-P and NH4-N release from intact sedi-
Available online 31 May 2011
ment cores and sediment oxygen demand. Measurements made at 430-kHz were more sensitive to differences in sediment properties than 201- or 38-kHz. Hydroacoustic
Keywords:
measurements allow rapid assessment of properties important in lake restoration and
Hydroacoustic methods
water resource management. ª 2011 Elsevier Ltd. All rights reserved.
Sediment properties Fractal dimension
1.
Introduction
Bottom sediments are a critical component of lake and reservoir systems. Bottom sediments serve as habitat for benthic invertebrates (Peeters et al., 2004), influence macrophyte distribution (Duarte and Kalff, 1986), regulate nutrient recycling rates in lakes (Sondegaard et al., 2003), control concentrations of dissolved oxygen (DO), H2S and other constituents in bottom waters (Hatcher, 1986; Reese et al., 2008), accumulate contaminants such as metals, pesticides and other hydrophobic organic compounds (Baudo et al., 1989; Forstner and Wittmann, 1979; Karickhoff et al., 1979), and provide a record of past conditions in the lake, airshed, watershed and climate (Lehman, 1975; Kirby et al., 2007). Sediment properties can vary widely over space and time, however. Wave action tends to limit deposition of fine sediments and leave coarse-textured sediments in place in shallower waters, while fine organic sediments are often focused
into deeper regions of a lake or reservoir basin (Lehman, 1975; Hakanson and Jansson, 1983; Anderson et al., 2008). Organic sediments are generally zones of intense microbial activity, rapidly depleting DO concentrations, promoting denitrification, reduction of Fe and Mn oxyhydroxides and often resulting in sulfate reduction or methanogenesis (Hargrave, 1972; Reese et al., 2008). Reducing conditions often result in very high rates of release of phosphate and ammonium from bottom sediments (Holdren and Armstrong, 1980). In contrast, coarser textured sediments that are lower in organic carbon content support much slower rates of sediment oxygen demand and internal nutrient recycling (Hargrave, 1972). The capacity for carboxylic acid and other functional groups on organic matter to strongly bind metals can result in enrichment of organic sediments to very high concentrations of metals relative to the water column (Forstner and Wittmann, 1979). The potential for toxicity effects on benthic invertebrates and other organisms has led to development of sediment
* Corresponding author. Tel.: þ1 951 827 3757; fax: þ1 951 827 3993. E-mail address:
[email protected] (M.A. Anderson). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.05.029
4400
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 9 9 e4 4 0 8
quality criteria, and high contaminant concentrations can also lead to management and disposal issues. At the same time, sediments enriched in organic carbon exhibit very high partitioning coefficients for hydrophobic organic contaminants (Karickhoff et al., 1979), potentially leading to food-chain transfer and biomagnification (Lundgren et al., 2002). Understanding the properties and distribution of sediments in a lake or reservoir, then, can be critical to management and restoration. The distribution of sediments and their properties are typically determined using sediment grab samples or cores from a finite number of sites. The numbers of sites sampled are generally limited, since such sampling is time-consuming, and thus often fails to capture the detailed distribution and heterogeneities present in the lake (Downing and Rath, 1988). Models have been developed to understand sediment distribution in lake basins. For example, Hakanson (1982) developed the dynamic ratio model, where lake area and mean depth could be used to estimate the area of erosion and transport, and therefore also the area of deposition. Rowan et al. (1992) used wave theory to predict the depth that separates the erosional zone from transitional and depositional zones in lakes (the so-called mud energy boundary depth). Such models provide only coarse-scale predictions about depositional and erosional zones and do not make specific predictions about sediment properties, however (Anderson et al., 2008). Hydroacoustic measurements offer a way to rapidly collect information about sediment properties at a very large number of sites. Hydroacoustic methods involve emission of soundwaves at known energy and wavelength from a transducer; the soundwave is then propagated through the water column and reflected off of objects at a strength proportional to their size, shape and density and soundspeed contrasts with water. Bottom sediments represent the dominant source of acoustic backscatter in a lake, although aquatic plants (Zhu et al., 2007), fish (Kubecka and Duncan, 1998), zooplankton (Holliday and Pieper, 1980), larval insects (Knudsen et al., 2006) and gas bubbles (Ostrovsky et al., 2008) also scatter sound. Fisheries remain the most common application of hydroacoustic methods, although it is increasingly being used to identify different sediment types within nearshore ocean and estuarine studies (Tegowski, 2005; Freitas et al., 2005; Anderson et al., 2002). An early such application of hydroacoustic methods was conducted by Pouliquen and Lurton (1992), who found that cumulative energy curves of bottom echoes were related to sediment physical properties. Other acoustical properties have been used to distinguish sediments with different hardness, roughness and grain size, including the ratio of the first and second bottom echo intensities (Orlowski, 1984; Chivers et al., 1990) and the first and second parts of the first bottom echo (Sternlicht and de Moustier, 2003; BioSonics, 2008). The fractal dimension of the bottom echo has also been proposed (Tegowski and Lubniewski, 2000). These studies have, to this point, generally been conducted in nearshore and open ocean environments with relatively low frequency transducers (e.g., Freitas et al., 2005; Anderson et al., 2008). Characterization of bottom sediments in inland lakes and reservoirs using hydroacoustic methods has received much less attention.
The overall objective for this study was to evaluate the capability of hydroacoustic methods to reliably estimate bottom sediment properties in lakes. Specific objectives were to: (i) quantify the acoustical response of different types of sediment using 3 different frequencies; (ii) assess the capacity of different acoustical properties to quantify sediment characteristics; and (iii) test the extensibility of findings to other lakes.
2.
Materials and methods
2.1.
Hydroacoustic measurements
Hydroacoustic measurements were made using a BioSonics Inc. (Seattle, WA) DTX echosounder multiplexed with 430-kHz single beam, 201-kHz split beam and 38-kHz single beam transducers (Table 1). A JRC 212W real-time differential globalpositioning satellite (DGPS) receiver was mounted directly over the transducers and recorded differentially-corrected positions every 1-s. BioSonics VisualAcquisition 4.0 software on a Dell ATG laptop was used to set up transmit and receive information and acquire data during calibration and measurement. Data were acquired at 5 pings per sec at a pulse duration of 0.4 ms on each frequency (Table 1). Field calibration was conducted using tungsten-carbide spheres of known target strength. Echograms were analyzed using BioSonics VBT v.1.12 software with 30 log R time-varied gain (TVG) and depthnormalization to lake mean depth (Dommisse et al., 2005). The VBT software was used to extract 4 different attributes of the bottom echo envelope, i.e., the time-dependent amplitude of a received pulse (Horne, 2000). A soundwave emitted by the transducer propagates through the water at approximately 1500 m s1 and spreads radially depending upon the transducer half-beam angle (Table 1). The soundwave is then reflected off of the lake bottom and returned to the transmitter as the first bottom echo. The shape and energy of the reflected soundwave carries with it information about the bottom sediment. The increasing (first) part of the bottom echo (E10 ) contains information about hardness, while the decreasing (second) part of the bottom echo (E1) contains information about volume scattering and roughness (BioSonics, 2008; Ostrovsky and Tegowski, 2010). Sound reflected off of the bottom sediments is also reflected from the lake surface, and is subsequently reflected a second time off of the bottom sediments which can also be recorded by the transducer (the second bottom echo, E2). The ratio of the first and second parts of the first bottom echo (E10 /E1) can resolve different types of
Table 1 e Transducer configurations used in this study. Property Frequency (kHz) Beam angle ( ) Source level (dB/mPa) Receive sensitivity (dBC/mPa) Pulse length (ms) Pings per second (pps)
DTX-38
DTX-200
DTX-420
38 10.0 217.03 41.1 0.4 5
201 6.6 221.3 57.6 0.4 5
430 7.0 220.0 62.9 0.4 5
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 9 9 e4 4 0 8
bottom sediments and can serve as an indicator of hardness (Ostrovsky and Tegowski, 2010) The ratio of the second part of the first bottom echo and second bottom echo (E1/E2) has also been used to classify different types of sediments (Chivers et al., 1990). In addition to these energy-based measures, the shape and structure of the bottom echo also contains information about the sediments properties (Van Walree et al., 2005). VBT characterizes the shape of the bottom echo as a fractal dimension (FD) using the box-dimension approach (Van Walree et al., 2005; BioSonics, 2008). The FD of the bottom echo is thus a measure of the echo fluctuation and describes heterogeneity of the bottom (Tegowski, 2005).
2.2.
Study sites
Hydroacoustic surveys were conducted at Lake Elsinore and San Dieguito Reservoirs in southern California, USA. Lake Elsinore is the largest natural lake in southern California and is located in southwestern Riverside County about 110 km southeast of Los Angeles. The lake is nominally about 1200 ha in surface area, and has a mean depth near 4 m, although historical bathymetric data were derived from depth measurements made at about 50 locations across the lake. Lake Elsinore is highly eutrophic, with frequent algal blooms and periodic fish kills. A 270 km survey, on an orthogonal grid with 100 m spacing, was conducted over 6 days from June 27e30 and July 12e14, 2010. Sediment surface grab samples were collected from 28 sites on a staggered-start sampling grid matched to the hydroacoustic survey grid over several days in late MayeJune 2010 using an Ekman dredge (Fig. 1a, solid circles). An additional 6 sites were sampled from the lake on September 21, 2010 (Fig. 1a, open circles). Measurements were also made on San Dieguito Reservoir, a small (w20 ha) reservoir that serves as a source drinking
4401
water reservoir for the Santa Fe Irrigation District (SFID). Two 20 km surveys were conducted on April 24 and August 12, 2010. Sediment surface grab samples were collected from 5 sites on the lake during the summer sampling (Fig. 1b).
2.3.
Sediment analyses
Sediment grab samples collected with an Ekman dredge were homogenized and subsampled into individual 500-mL widemouth glass jars with Teflon-lined lid. Samples were stored on ice in a cooler and returned to the lab. Upon return to the lab, sediment was promptly rehomogenized and subsampled for sediment characterization and porewater analysis. Water content was determined on subsamples that were heated overnight at 105 C. This higher temperature, commonly used in soil analyses, was chosen over the lower temperatures (60e80 C) used by Hakanson and Jansson (1983) and others because of the high amount of hydrous clays and moderate organic matter content found in these sediments, although weight differences were minimal (often <1%) for the two temperatures. Loss-on-ignition (LOI) was measured on ovendried samples by heating to 550 C for 2 h (Heiri et al., 2001). Particle size was determined on samples mechanically dispersed in sodium hexametaphosphate without CaCO3 or organic matter removal using the hydrometer method (Gee and Bauder, 1986). Total C, N, and S were measured by drycombustion methods using a Thermo Flash EA NC soil analyzer (Nelson and Sommers, 1982). Inorganic C and CaCO3 were determined manometrically following Loeppert and Suarez (1996). Organic C was taken as the difference between total C and inorganic C. Total and inorganic P was determined following Aspila et al. (1976). Organic P was taken as the difference between total and inorganic P (Aspila et al., 1976). Porewater was extracted by centrifugation, filtered
Fig. 1 e Study sites, showing bathymetry developed from hydroacoustic surveys and sediment sampling locations (solid circles): (a) Lake Elsinore and (b) San Dieguito Reservoir (Open circles and “X”s show supplemental grab sampling and sediment core sites on Lake Elsinore.).
4402
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 9 9 e4 4 0 8
through 0.45 mm polycarbonate filters fitted to disposable plastic syringes, and acidified with reagent-grade H2SO4. Porewater NH4-N and dissolved orthophosphate (PO4-P) and extracted P were analyzed using a Technicon autoanalyzer following standard methods (APHA, 1998). Rates of nutrient release and sediment oxygen demand (SOD) were measured on triplicate cores collected in September 2010 from 3 sites on Lake Elsinore that varied in depth and organic C contents based upon the hydroacoustic measurements (Fig. 1a, sites designated by X). Cores were collected by inserting 6.3 cm diameter by 30.5 cm length Lucite core tubes approximately 10 cm into sediment retrieved with an Ekman dredge (Anderson, 2004). Cores were topped off with lake water with no headspace, sealed with no. 13 rubber stoppers, and transported back to the lab. Approximately 4 L of lake water was also collected. Upon return to the lab, 10-mL samples of water overlying the sediment cores were collected from each of the cores, filtered through 0.45 mm polycarbonate syringe filters, and acidified to pH < 2 with 1 drop 50% H2SO4. Overlying water was very gently mixed by bubbling with N2 or air at a gas flow rate of about 50 cm3 h1 to maintain dissolved oxygen concentrations present at the time of sampling, approximately 0.5, 1 and 4 mg L1 for the deep (7.5 m), intermediate (4.5 m) and shallow (2.8 m) sediment samples, respectively. 10-mL water samples were taken daily for 5 days, and analyzed for PO4-P and NH4-N on a Technicon Autoanalyzer. The mass of nutrients accumulated in the overlying water, normalized to sediment cross-sectional area, was used to determine nutrient flux rates (Holdren and Armstrong, 1980). Immediately following the final sampling for nutrient flux measurements, waters overlying sediment cores were gently aerated to bring dissolved oxygen (DO) concentrations to near saturation. Initial DO concentrations were recorded, and cores and water samples were then sealed with no. 13 stoppers without headspace and incubated in the dark at known temperature (25 1 C). Dissolved oxygen concentrations were measured twice a day for 3 days following gentle mixing of water using a YSI Model 55 DO meter. Concentrations of DO in separate triplicate lake water samples (without sediments) were also measured under equivalent conditions. DO concentrations decreased linearly over time in waters overlying sediment cores and in lake water alone. Dissolved oxygen losses in water overlying the sediment cores were corrected for DO loss measured in water-only samples and normalized to sediment surface area to yield estimates of the rate of sediment oxygen demand (SOD).
3.
Results
3.1.
Bathymetry
(Lawson and Anderson, 2007). The impellers have scoured away sediment under the arrays, resulting in well-defined depressions at the locations of each of the 5 axial flow pump arrays (Fig. 1a). Also evident from the bathymetric map is the very flat bottom over much of the lake area (excluding the lake margin where somewhat greater slopes were found, the bottom slope averaged 0.14%). A bathymetric map for San Dieguito Reservoir was also developed from the hydroacoustic survey (Fig. 1b). The maximum depth of the lake was found adjacent to the dam, while a large shallow region was present in the northeast part of the lake. The reservoir has a comparable mean and maximum depth as Lake Elsinore (approximately 4 and 9 m, respectively).
3.2.
Sediment properties
Sediment samples collected from the initial 28 sites on Lake Elsinore were found to have a wide range of properties (Table 2). Samples were collected from 1.3 to 7.5 m depth (mean depth 4.6 1.9 m) and varied strongly in texture (from 0 to 88% sand and 3.3 to 65.4% clay) (Table 2). Water content and organic C content both varied with texture; water content averaged 58.5% and ranged from 15.6 to 83.1%, while organic C content averaged 1.93% (range 0.04e4.99%) (Table 2). The sediments also tend to be strongly enriched in CaCO3. Acoustic properties measured at each of the sites were then correlated with these sediment properties (Table 3). Correlation coefficients varied widely across the different sediment and acoustical properties, with r-values >0.50 statistically significant at p ¼ 0.01. Thus, a large number of statistically significant correlations were present (e.g., 11 out of 36 correlations were significant at 0.01 for the 430-kHz data) (Table 3). Of the acoustical properties, the fractal dimension (FD) and energy ratio of the first and second part of the first bottom echo (E10 /E1) were both statistically significantly correlated with a number of sediment properties, especially at the 430- and 201-kHz frequencies (Table 3). While statistical significance is important, r-values near 0.5 capture a relatively small fraction (25%) of the variance in the sediment properties; meaningful predictive capabilities require much higher rvalues. On that basis, one notes that generally very strong correlations exist between the fractal dimension of the first
Table 2 e Summary of sediment properties from Lake Elsinore (n [ 28). Property
The hydroacoustic survey on Lake Elsinore was first used to develop a more detailed bathymetric map for the lake at a surface elevation of 379.0 m above MSL, relative to an earlier one constructed at a lower surface elevation and from a limited number of measurements (Lawson and Anderson, 2007) (Fig. 1a). The region of maximum depth is located in the northeastern part of the lake, reaching about 9 m below one of the axial flow pump arrays installed on the lake
Depth Sand Silt Clay H2O content Organic C LOI CaCO3 Total N Total P
Units
Mean
Std dev
Min
Max
m % % % % % % % % mg kg1
4.6 30.8 38.8 30.5 58.5 2.63 8.31 10.44 0.35 788
1.9 35.9 19.6 23.4 25.1 1.93 5.69 6.90 0.20 475
1.33 0.0 8.4 3.3 15.6 0.04 0.36 0.11 0.01 106
7.5 88.0 72.4 65.4 83.1 4.99 16.04 18.17 0.59 1456
4403
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 9 9 e4 4 0 8
Table 3 e Correlation table showing correlation coefficient values between sediment properties and corresponding acoustical attributes (r-value >0.5 significant at 0.01). Sand 430-kHz FD E1/E2 E10 /E1 E1/FD 201-kHz FD E1/E2 E10 /E1 E1/FD 38-kHz FD E1/E2 E10 /E1 E1/FD
Silt
Clay
% H2O
Org C
LOI
CaCO3
Total N
Total P
0.83 0.35 0.51 0.21
0.42 0.34 0.40 0.18
0.92 0.24 0.45 0.17
0.86 0.35 0.56 0.18
0.91 0.26 0.48 0.14
0.91 0.32 0.51 0.17
0.84 0.34 0.55 0.17
0.85 0.26 0.41 0.13
0.13 0.06 0.09 0.10
0.76 0.16 0.65 0.65
0.34 0.22 0.54 0.41
0.89 0.06 0.54 0.65
0.81 0.16 0.66 0.71
0.87 0.14 0.56 0.72
0.87 0.16 0.61 0.76
0.79 0.11 0.60 0.71
0.82 0.15 0.40 0.67
0.08 0.18 0.16 0.06
0.89 0.39 0.14 0.72
0.69 0.19 0.10 0.52
0.79 0.44 0.14 0.68
0.89 0.48 0.22 0.74
0.85 0.45 0.09 0.78
0.88 0.46 0.14 0.80
0.88 0.59 0.25 0.71
0.75 0.44 0.02 0.70
0.17 0.26 0.15 0.03
bottom echo and a number of biogeochemically relevant sediment properties. For example, the FD at 430-kHz was strongly correlated with % organic C content (r ¼ 0.91), % clay content (r ¼ 0.92) and total N (r ¼ 0.85) (Table 3). The FD of the first bottom echo thus accounted for 83% of the variance in measured organic C content in the sediments of Lake Elsinore. Unlike the other sediment properties, the FD was negatively correlated with % sand content (r ¼ 0.83). Interestingly, total P concentration in sediments was not well-correlated with any acoustical property and, by extension, not correlated with other sediment properties as well (Table 3). The other, energy-based acoustical properties of bottom sediments, such as the energy ratio of the second part of the first bottom echo and second bottom echo (E1/E2), ratios of the first and second parts of the first bottom echo (E10 /E1) were less strongly correlated with physical and chemical characteristics than the FD or “shape” of the bottom echo (Table 3). The energy ratio of the first and second part of the first bottom echo (E10 /E1) at 430-kHz was nonetheless significantly correlated (at p ¼ 0.01) with % water content, loss-on-ignition and CaCO3 content, but provides much lower predictive power than FD. Correlation analysis yielded broadly similar results for most of the acoustical attributes at 201-kHz (Table 3). The FD was again the strongest correlate with sediment properties, with very strong but slightly lower r-values compared with 430-kHz. The ratio of E10 /E1 was somewhat more strongly correlated with % organic C and other properties at 201-kHz than 430-kHz (e.g., r-value of 0.56 vs. 0.48 for 201- and 430kHz, respectively, with organic C), but still only accounted for 31% of variance in organic C content. The ratio of E1/FD was more strongly correlated with most sediment properties at 201-kHz than 430-kHz (r-values up to 0.76 and accounting for up to 58% of variance), but also remained below values for FD (Table 3). Interestingly, r-values were often of the opposite sign at 38kHz when compared with the higher frequencies (Table 3). Fractal dimension was again the strongest correlate, but in this case, FD values generally declined with increasing clay content, organic C content (r-values of 0.79 to 0.85, respectively) and other properties. The exception was % sand
content that exhibited a positive r-value (opposite of that witnessed at 430- and 201-kHz) (Table 3). Frequency-dependent acoustic response has been used in plankton and fish studies to help resolve different types or size classes of scatterers (e.g., Jurvelius et al., 2008) and proposed to improve classification of bottom sediments (Anderson et al., 2008). Differences and ratios of E10 , E1, E2 and FD at 38-, 201- and 430-kHz respectively were not found to offer greater predictive power than single frequency measurements made at 430-kHz, however (data not shown). While r-values as provided in Table 3 provide a convenient summary of the strength of the relationship between acoustical and sediment physical and chemical properties, they can mask more subtle non-linear relationships. Plots of the FD of sediment bottom echo against physical properties indicate that strong linear relationships were, in fact, present for a number of properties, including clay content, organic C content, and total N content (Fig. 2aec) and loss-on-ignition (not shown), although a non-linear relationship between FD and water content was present (Fig. 2d). These latter two properties are often used as convenient surrogates for other more physically and chemically relevant properties (Hakanson and Jansson, 1983). Linear regressions of the FD of the bottom echo at 430-kHz with sediment properties from the initial 28 sites on Lake Elsinore (Fig. 2, solid circles) yielded the following equations: %Clay ¼ 528:5 FD 471:1
r2 ¼ 0:85
%Organic C ¼ 43:0 FD 38:2 %LOI ¼ 126:7 FD 111:9 %Total N ¼ 4:25 FD 3:68
r2 ¼ 0:83
r2 ¼ 0:83 r2 ¼ 0:72
(1) (2) (3) (4)
Included on these figures are sediment physicochemical properties and the FD of the bottom echo at 430-kHz from 6 additional sites from Lake Elsinore and the 5 sites sampled from San Dieguito Reservoir (Fig. 1). One sees that the additional 6 sites from Lake Elsinore (shown as open circles, Fig. 2) fell within the range of values found for the original 28 sites
4404
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 9 9 e4 4 0 8
a
100
b L. Elsinore Jul 2010 L. Elsinore Oct 2010 San Dieguito Res.
60
40
20
0 0.88
L. Elsinore Jul 2010 L. Elsinore Oct 2010 San Dieguito Res.
8
Organic C Content (%)
Clay Content (%)
80
10
6
4
2
0.92
0.96
1
1.04
0 0.88
1.08
0.92
Fractal Dimension
c
1.2
d L. Elsinore Jul 2010 L. Elsinore Oct 2010 San Dieguito Res.
0.8 0.6 0.4
1.04
1.08
100
60 L. Elsinore Jul 2010 L. Elsinore Oct 2010 San Dieguito Res.
40
20
0.2 0 0.88
1
80
Water Content (%)
Total N Content (%)
1
0.96
Fractal Dimension
0.92
0.96
1
1.04
1.08
Fractal Dimension
0 0.88
0.92
0.96
1
1.04
1.08
Fractal Dimension
Fig. 2 e Fractal dimension (FD) of bottom echo at 430-kHz vs. (a) clay content, (b) organic C content, (c) total N content and (d) % water content.
(shown at solid circles, Fig. 2). Thus sampling later in the summer did not substantively alter the relationship between FD and clay content, organic carbon content, total N or water content of the bottom sediments within this relatively narrow, transitional range of properties (Fig. 2). Also shown on these figures are the physicochemical and acoustical properties of sediments in San Dieguito Reservoir (Fig. 2). The five sites from the reservoir ranged in depth from 3.3 to 9.2 m, and varied strongly in clay, organic C, total N and water contents (Fig. 2). The highest of these components were found in the deepest water, consistent with the focusing of fine organic sediments there. It is noteworthy that these high concentrations exceeded by 50e100% the highest values found in Lake Elsinore, although the linear trends seen between clay, organic C and total N contents with FD in Lake Elsinore appear to hold for sediments collected from a very different lake (Fig. 2aec). The non-linear relationship between water content and FD also holds (Fig. 2d). The capability of FD to predict sediment properties can be quantitatively explored with regression equations fit to the original Lake Elsinore data using attributes measured at
additional sites from Lake Elsinore and at San Dieguito Reservoir. As illustrated in Fig. 2a, Eq. (1) yielded low predicted clay contents, on the order of 10e20%, in the additional Lake Elsinore samples collected in September in reasonable agreement with measured values (Fig. 2a, open circles). The regression model between FD and % clay content also reasonably predicted clay content in the sediments of San Dieguito Reservoir (Fig. 2a, shaded squares). The mean error in predicted clay content (predictedeobserved concentrations) was 1.2 7.4% (Table 3). The regression model for clay content was thus close to neutral bias, and accurate to within about 7% for these samples. The regression model developed from the original Lake Elsinore sites also reasonably predicted organic C contents of sediments collected from the lake in September, as well as those collected from San Dieguito Reservoir (Fig. 2b). The mean error estimate was biased slightly low, but generally within 1.0% of the measured values (Table 4). Similarly, %LOI and total N were reasonably well-described by the regression models, with mean errors of 0.14 1.97% and 0.02 0.16%,
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 9 9 e4 4 0 8
Table 4 e Summary of estimate errors (predictedeobserved) from regression equations when fit to San Dieguito Reservoir and additional Lake Elsinore samples (n [ 11). Estimate error
% Clay
% Organic C
Mean s.d. 1.2 7.4 0.21 1.00 Range 10.7e16.7 1.9e1.8
% LOI
% Total N
0.14 1.97 0.02 0.16 2.7e4.4 0.29e0.22
respectively (Table 4), although regression equation fitted to the initial Lake Elsinore samples was observed to underpredict total N in the organic-rich San Dieguito Reservoir sediments (Fig. 2c). The capability of predicting sediment properties from their hydroacoustic properties allows one to develop detailed spatial representations of sediment properties across a lake basin. For example, using the regression equation developed from the initial sampling of Lake Elsinore and validated with subsequent samples collected there as well as at San Dieguito Reservoir (Eq. (2)), the survey reveals that organic C concentrations in Lake Elsinore are highest in the central-northern part of the lake (Fig. 3a), coinciding with greatest depth there (Fig. 1a). Sediments with very low organic C contents (and low clay and high sand contents) were present in a shallow bay at the south end of the lake, consistent with wave-induced resuspension owing to the strong afternoon winds from the northwest. Distribution of organic C (Fig. 3a) thus broadly followed that of depth in this simple basin (Fig. 1a), although heterogeneities are clearly evident (e.g., note the 3e4% organic C sediment that intrudes into the south-central part of the lake basin where 4e5% organic C was more broadly found). The area of the lake basin with very low sediment organic C concentrations (<1%) comprised about 160 ha, or 13.7% of the total area, while higher organic C sediments (e.g., >4%) comprised a larger fraction of the lake bottom (28.9%). The finescale heterogeneities (on a scale less than the transect spacing) that are present in the figure are a result of the interpolation process (kriging) used to develop these contour plots.
4405
The distribution of organic C in San Dieguito Reservoir exhibited a weaker relationship with depth, with high organic C sediments present both in deep water near the dam, and also in the shallower water on the east side of the lake, while low organic C content sediments were present in the south part of the reservoir (Fig. 3b). San Dieguito Reservoir differs from Lake Elsinore in the presence of a well-developed littoral zone; in contrast, the seasonally and annually fluctuating surface elevation of Lake Elsinore (Lawson and Anderson, 2007) limits the development of an aquatic macrophyte community there. Thus, focusing of organic matter into the deeper part of San Dieguito Reservoir results in increased organic C concentrations there, although organic matter content of the sediments is also influenced by the presence of emergent and submerged aquatic macrophytes. Two additional factors are also thought to have influenced the distribution of organic C in San Dieguito Reservoir. First of all, filter backwash from the R.E. Badger Filtration plant is discharged to the shallow northeastern part of the lake that may have increased the total organic C content of sediments there. Secondly, prior to impoundment in 1918, Escondido Creek flowed through the site; as a result, native wetland soils would have also been present near and within some stretches of the creek. The region of maximum organic C content (reaching or exceeding 10% by weight) (Fig. 3b) was found in intermediate depth (about 6e7 m) near or within the creek channel (Fig. 1b). Since SOD and internal nutrient recycling are both correlated with the organic C content of sediments (Hargrave, 1972), it follows that hydroacoustic methods can also be used to infer the rates of these important processes in sediments. As found with clay content, organic C and other properties, the fractal dimension of the bottom echo was also linearly related to rates of release of PO4-P (r2 ¼ 0.98) and NH4N (r2 ¼ 0.96) from bottom sediments as well as SOD (r2 ¼ 0.94) (Fig. 4). Nonetheless, the FD values of bottom sediment represent the echo fluctuations due to heterogeneities in sediment physicochemical properties, and are thus indirectly related to rates of nutrient recycling or SOD, although depending upon one’s particular interests, the method allows
Fig. 3 e Distribution of organic C concentrations across: (a) Lake Elsinore and (b) San Dieguito Reservoir.
4406
b
16
NH4-N Flux (mg m-2 d-1)
PO4-P Flux (mg m-2 d-1)
12
8
4
0
-4 0.88
0.92
0.96
1
Fractal Dimension
1.04
c
160
120
80
40
0 0.88
1
0.8
SOD (g m-2 d-1)
a
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 9 9 e4 4 0 8
0.6
0.4
0.92
0.96
1
Fractal Dimension
1.04
0.2 0.88
0.92
0.96
1
1.04
Fractal Dimension
Fig. 4 e Fractal dimension (FD) at 430-kHz vs. (a) PO4-P flux, (b) NH4-N flux and (c) sediment oxygen demand (SOD) (Lake Elsinore).
estimation of physicochemical properties as well as (autocorrelated) biogeochemical processes.
4.
Discussion
The acoustical properties of bottom sediments from Lake Elsinore were significantly correlated with a number of physical and chemical properties, including textural information, water content, and organic C and total N contents. The fractal dimension of the first bottom echo was the acoustic property most strongly correlated with these sediment attributes (correlation coefficient values typically greater than 0.8) (Table 3). A survey from the North Sea that included mean grain size analysis on 50 grab samples also found that the fractal dimension of the bottom echo increased with smaller average particle size, and was more strongly correlated than other acoustical attributes (e.g., echo energy) at an echosounder frequency of 66 kHz (r ¼ 0.84) (Van Walree et al., 2005). A slightly weaker correlation was found at 150 kHz. The fractal dimension of the bottom echo was also found to effectively distinguish between different types of bottom sediments (silt, clay, fine-grained sand, and coarse sand þ gravel) in the southern Baltic Sea (Tegowski, 2005) and in Lake Kinneret in Israel (Ostrovsky and Tegowski, 2010). Coarse-textured sediment, often found in shallow regions of lakes, is frequently more homogeneous vertically than finer textured sediments with higher organic matter contents within depositional zones than can include varves, gas pockets, and other fine-scale features (Ostrovsky and Tegowski, 2010). That greater heterogeneity within the bottom sediments can alter the shape and increase the fractal dimension of the reflected soundwave. The slightly stronger correlations found at 430-kHz than either 201- or 38-kHz result from differences in penetration depths, and ability to resolve small spatial structures within the sediments (e.g., laminations, gas pockets). The lower
frequencies penetrate bottom sediments more effectively (Dunbar et al., 1999), but the longer wavelength reduces the ability to resolve fine-scale heterogeneities within the sediment. In a very simple way, assuming resolution is approximated by l/2 and the speed of sound is near 1500 m s1, the 430-kHz frequency would be able to resolve features on the scale of about 2 mm, while the 38-kHz frequency would only be able to resolve features 10 larger, or about 2 cm. In a direct comparison of acoustic sediment classes with ground-truth measurements using both 66-kHz and 150-kHz frequencies, Van Walree et al. (2005) also found the higher frequency was more effective at resolving sediment types in a North Sea survey. The negative correlations found for FD at 38-kHz for most of the measured sediment properties in this study (Table 3) are thought to result from greater variance in penetration depths in combination with the longer wavelength and lower resolution at this frequency. Results from this study demonstrate that the hydroacoustic characterization of bottom sediments, especially the fractal dimension of the bottom echo, can provide valuable insights into the properties and distribution of sediments in lakes and reservoirs. Beyond mean grain size and, in some cases, loss-on-ignition that has been evaluated in most coastal studies, the fractal dimension of the bottom echo can provide information about organic C and total N concentrations in bottom sediments, as well as estimates of rates of internal nutrient recycling and sediment oxygen demand. Findings here indicate that there is some general extensibility of results from measurements made on Lake Elsinore in JuneeJuly 2010 to those made on the lake somewhat later in the year, as well as to San Dieguito Reservoir. This suggests that extensive ground-truthing may not be necessary for similar lakes in this region if qualitative estimates of organic C and other physical and chemical properties are sufficient, although ground-truthing remains necessary to maximize the accuracy of the estimates. Ground-truthing is needed to understand rates of biogeochemical processes in sediments,
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 9 9 e4 4 0 8
however, owing to their sensitivity to water column properties, such as temperature and DO. Incorporation of the additional sample results from Lake Elsinore, as well as those from San Dieguito Reservoir, into the regression models provide somewhat improved correlations relative to the initial Lake Elsinore dataset (increasing r2 values by 0.04e0.11): %Clay ¼ 516:9 FD 460:4
r2 ¼ 0:89
%Organic C ¼ 46:6 FD 41:5 %LOI ¼ 128:5 FD 113:7 %Total N ¼ 5:15 FD 4:53
r2 ¼ 0:87
r2 ¼ 0:88 r2 ¼ 0:83
(5) (6) (7)
2. The fractal dimension of the bottom echo was most strongly correlated with sediment properties, with r2 values of 0.83e0.85 at 430-kHz. 3. The fractal dimension of the bottom echo was also linearly related with rates of PO4-P and NH4-N flux and sediment oxygen demand of bottom sediments. 4. Somewhat stronger correlations between sediment properties and fractal dimension of the bottom echo were found at 430-kHz relative to 201- and 38-kHz. 5. Hydroacoustic methods offer a rapid way to determine sediment physical and chemical properties and rates of important biogeochemical processes across a lake, and are thus a valuable tool for habitat assessment, water quality management and lake restoration.
(8)
Inclusion of additional sample results increased the r2 value for % total N more so than the other properties (r2 value increased from 0.72 to 0.83); the additional samples lowered the response at low FD, resulting in a higher slope than found from the initial sample set (Fig. 2c). Additional work would be needed to test the applicability of these regression equations to other lakes and to possible seasonal differences (e.g., due to sediment gas accumulation, production and water level fluctuations) (Ostrovsky and Tegowski, 2010), basin morphometries and water quality. Fine lateral scale heterogeneities are also thought to have influenced observations reported herein. That is, the Ekman dredge necessarily samples a small cross-sectional are of sediment (about 230 cm2). In contrast, the transducers acoustically sampled a much larger cross-sectional area that is a function of beam angle and sediment depth. For example, with a beam halfangle of 7 (Table 1), the 430-kHz transducer would acoustically sample a sediment area over 0.7 m2 (or 33 greater than the Ekman dredge) at the mean depth of the lake. This larger sampling area is advantageous in the context of sediment characterization within a lake or basin survey, but does introduce spatial scale differences during ground-truthing. As a result, sediment surface grab samples were collected from a relatively large number of sites during this study. Notwithstanding, the capacity to map sediment properties across a lake can provide critical information for use in habitat assessment, water quality management and lake restoration. For example, being able to map the distribution of PO4-P flux from bottom sediments based upon FD (Fig. 4a) allows targeted alum applications to the regions where most benefit can be attained. Similarly, understanding the distribution of SOD across a lake basin can aid in the design of hypolimnetic oxygenation and diffused aeration systems and placement of diffuser lines. The value of hydroacoustic mapping of sediments for benthic habitats has already been recognized in coastal, seafloor and mid-shelf systems (Hewitt et al., 2004; Freitas et al., 2006).
5.
4407
Conclusions
1. Sediment physical and chemical properties were significantly correlated with acoustical properties of bottom sediments.
Acknowledgments Thanks to the Lake Elsinore-San Jacinto Watersheds Authority (LESJWA) and the Santa Fe Irrigation District (SFID) for their support of this study. Special thanks to Pat Kilroy, City of Lake Elsinore, and Tim Bailey and Cor Shaffer, SFID, for their support and assistance, and to David Thomason and Ed Betty, University of California-Riverside, for assistance in the collection and analysis of the sediment samples.
references
Anderson, J.T., Gregory Gregory, R.S., Collins, W.T., 2002. Acoustic classification of marine habitats in coastal Newfoundland. ICES J. Mar. Sci. 59, 156e167. Anderson, J.T., Van Holliday, D., Kloser, R., Reid, D.G., Simard, Y., 2008. Acoustic seabed classification: current practice and future directions. ICES J. Mar. Sci. 65, 1004e1011. Anderson, M.A., Whiteaker, L., Wakefield, E., Amrhein, C., 2008. Properties and distribution of sediment in the Salton Sea, California: an assessment of predictive models. Hydrobiologia 604, 97e110. Anderson, M.A., 2004. Impacts of metal salt addition on the chemistry of Lake Elsinore, California: 2. Calcium salts. Lake Reservoir Manage. 20, 270e279. APHA, 1998. Standard Methods for the Examination of Water and Wastewater, twentieth ed. American Public Health Association, Washington, D.C. Aspila, K.I., Agemian, H., Chau, A.S.Y., 1976. A semi-automated method for determination of inorganic, organic and total phosphate in sediments. Analyst 101, 187e197. Baudo, R., Amantini, L., Bo, F., Cenci, R., Hannaert, P., Lattanzio, A., marengo, G., Muntau, H., 1989. Spatial-distribution patterns of metals in the surface sediments of Lake Orta (Italy). Sci. Total Environ. 87, 117e128. BioSonics, Inc., 2008. User Guide: Visual Bottom Typer 1.10. BioSonics, Inc., Seattle, WA, 113 pp. Chivers, R.C., Emerson, N., Burns, D.R., 1990. New acoustic processing for underway surveying. Hydrographic J. 56, 9e17. Dommisse, M., Urban, D., Finney, B., Hills, S., 2005. Potential depth biasing using BioSonics VBT seabed classification software. Mar. Technol. Soc. J. 39, 90e93. Downing, J.A., Rath, L.C., 1988. Spatial patchiness in the lacustrine sedimentary environment. Limnol. Oceanogr. 33, 447e458. Duarte, C.M., Kalff, J., 1986. Littoral slope as a predictor of the maximum biomass of submerged macrophyte communities. Limnol. Oceanogr. 31, 1072e1080.
4408
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 9 9 e4 4 0 8
Dunbar, J.A., Allen, P.M., Higley, P.D., 1999. Multifrequency acoustic profiling for water reservoir sedimentation studies. J.Sediment. Res. 69, 521e527. Forstner, U., Wittmann, G.T., 1979. Metal Pollution in the Aquatic Environment. Springer-Verlag, Berlin. Freitas, R., Sampaio, L., Rodrigues, A.M., Quintino, V., 2005. Seabottom classification across a shallow-water bar channel and near-shore shelf, using single-beam acoustics. Estuar. Coast. Shelf Sci. 65, 625e632. Freitas, R., Sampaio, L., Oliveira, J., Rodrigues, A.M., Quintino, V., 2006. Validation of soft bottom benthic habitats identified by single-beam acoustics. Mar. Pollut. Bull. 53, 72e79. Gee, G.W., Bauder, J.W., 1986. Particle-size analysis. In: Klute, A. (Ed.), Methods of Soil Analysis. Part 1Agronomy Monographs, second ed., vol. 9. ASA and SSSA, Madison, WI, pp. 383e411. Hakanson, L., Jansson, M., 1983. Principles of Lake Sedimentology. Springer-Verlag, New York, NY. Hakanson, L., 1982. Lake bottom dynamics and morphometry: the dynamic ratio. Water Resour. Res. 18, 1444e1450. Hargrave, B.T., 1972. Aerobic decomposition of sediment and detritus as a function of particle surface area and organic content. Limnol. Oceanogr. 17, 583e596. Hatcher, K.J., 1986. Sediment Oxygen Demand: Processes, Modeling & Measurement. Institute of Natural Resources, Univ. of Georgia, Athens, GA. Heiri, O., Lotter, A.F., Lemcke, G., 2001. Loss on ignition as a method for estimating organic and carbonate content in sediments: reproducibility and comparability of results. J. Paleolimnol. 25, 101e110. Hewitt, J.E., Thrush, S.F., Legendre, P., Funnel, G.A., Ellis, J., Morrison, M., 2004. Mapping of marine soft-sediment communities: integrated sampling for ecological interpretation. Ecol. Appl. 14, 1203e1216. Holdren, G.C., Armstrong, D.E., 1980. Factors affecting phosphorus release from intact sediment cores. Environ. Sci. Technol. 14, 79e87. Holliday, D.V., Pieper, R.E., 1980. Volume scattering strengths and zooplankton distributions at acoustic frequencies between 0.5 and 3 MHz. J. Acoust. Soc. Am. 67, 135e146. Horne, J.K., 2000. Acoustic approaches to remote species identification: a review. Fish. Oceanogr. 9, 356e371. Jurvelius, J., Knudsen, F.R., Balk, H., Marjomaki, T.J., Peltonen, H., Taskinen, J., Tuomaala, A., Viljanen, M., 2008. Echo-sounding can discriminate between fish and macroinvertebrates in fresh water. Freshwater Biol. 53, 912e923. Karickhoff, S.W., Brown, D.S., Scott, T.A., 1979. Sorption of hydrophobic pollutants on natural sediments. Water Res. 13, 241e248. Kirby, M.E., Lund, S.P., Anderson, M.A., Bird, B.W., 2007. Insolation forcing of Holocene climate change in Southern California: a sediment study from Lake Elsinore. J. Paleolimnol. 38, 395e417. Knudsen, F.R., Larsson, P., Jakobsen, P.J., 2006. Acoustic scattering from a larval insect (Chaoborus flavicans) at six echosounder frequencies: implication for acoustic estimates of fish abundance. Fish. Res. 79, 84e89. Kubecka, J., Duncan, A., 1998. Acoustic size vs. real size relationships for common species of riverine fish. Fish. Res. 35, 115e125. Lawson, R., Anderson, M.A., 2007. Stratification and mixing in Lake Elsinore, California: an assessment of axial flow pumps
for improving water quality in a shallow eutrophic lake. Water Res. 41, 4457e4467. Lehman, J.T., 1975. Reconstructing the rate of accumulation of lake sediment: the effect of sediment focusing. Quaternary Res. 5, 541e550. Loeppert, R.H., Suarez, D.L., 1996. Carbonate and gypsum. In: Sparks, D.L. (Ed.), Methods of Soil Analysis. Part 3Agronomy Monographs, third ed., vol. 9. ASA and SSSA, Madison, WI, pp. 437e474. Lundgren, K., Tysklind, M., Ishaq, R., Broman, D., Van Bavel, B., 2002. Polychlorinated naphthalene levels, distribution, and biomagnification in a benthic food chain in the Baltic Sea. Environ. Sci. Technol. 36, 5005e5013. Nelson, D.W., Sommers, L.E., 1982. Total carbon, organic carbon, and organic matter. In: Page, A.L., Miller, R.H., Keeney, D.R. (Eds.), Methods of Soil Analysis, Part 2Agronomy Monographs, second ed., vol. 9. ASA and SSSA, Madison, WI, pp. 539e580. Orlowski, A., 1984. Application of multiple echoes energy measurements for evaluation of sea bottom type. Oceanologia 19, 61e78. Ostrovsky, I., Tegowski, J., 2010. Hydroacoustic analysis of spatial and temporal variability of bottom sediment characteristics in Lake Kinneret in relation to water level fluctuation. Geo-Mar. Lett. 30, 261e269. Ostrovsky, I., McGinnis, D.F., Lapidus, L., Eckert, W., 2008. Quantifying gas ebullition with echosounder: the role of methane transport by bubbles in a medium-sized lake. Limnol. Oceanogr. Meth. 6, 105e118. Peeters, E.T.H.M., Gylstra, R., Vos, J.H., 2004. Benthic macroinvertebrate community structure in relation to food and environmental variables. Hydrobiologia 519, 103e115. Pouliquen, E., Lurton, X., 1992. Identification of the nature of the seabed using echosounders. J. Phys. IV (2), 941e944. Reese, B.K., Anderson, M.A., Amrhein, C., 2008. Hydrogen sulfide production and volatilization in a polymictic eutrophic saline lake, Salton Sea, California. Sci. Total Environ. 406, 205e218. Rowan, D.J., Kalff, J., Rasmussen, J.B., 1992. Estimating the mud deposition boundary depth in lakes from wave theory. Can. J.Fish. Aquat. Sci. 49, 2490e2497. Sondegaard, M., Jensen, J.P., Jeppesen, E., 2003. Role of sediment and internal loading in shallow lakes. Hydrobiologia 506e509, 135e145. Sternlicht, D.D., de Moustier, C.P., 2003. Time-dependent seafloor acoustic backscatter (10e100 kHz). J. Acoust. Soc. Am. 114, 2709e2725. Tegowski, J., Lubniewski, Z., 2000. The use of fractal properties of echo signal for acoustical classification of bottom sediments. Acoustica/Acta Acoust. 86, 276e282. Tegowski, J., 2005. Acoustical classification of the bottom sediments in the southern Baltic Sea. Quatern. Int. 130, 153e161. Van Walree, P.A., Tegowski, J., Laban, C., Simons, D.G., 2005. Acoustic seafloor discrimination with echo shape parameters: a comparison with ground truth. Cont. Shelf Res. 25, 2273e2293. Zhu, B., Fitzgerald, D.G., Hoskins, S.B., Rudstam, L.G., Mayer, C.M., Mills, E.L., 2007. Quantification of historical changes of submerged aquatic vegetation cover in two bays in Lake Ontario with three complementary methods. J. Great Lakes Res. 33, 122e135.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 0 9 e4 4 1 8
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Transport and deposition of CeO2 nanoparticles in water-saturated porous media Zhen Li a, Endalkachew Sahle-Demessie b, Ashraf Aly Hassan b, George A. Sorial a,* a
Environmental Engineering Program, School of Energy, Environmental, Biological, and Medical Engineering, University of Cincinnati, P.O. Box 210012, Cincinnati, OH 45221-0012, USA b U.S. Environmental Protection Agency, Office of Research and Development, NRMRL, 26 W. Martin Luther King Drive (MS 443), Cincinnati, OH 45268, USA
article info
abstract
Article history:
Ceria nanoparticles are used for fuel cell, metal polishing and automobile exhaust catalyst;
Received 16 March 2011
however, little is known about the impact of their release to the environment. The stability,
Received in revised form
transport and deposition of engineered CeO2 nanoparticles through water-saturated
16 May 2011
column packed with sand were studied by monitoring effluent CeO2 concentration. The
Accepted 23 May 2011
influence of solution chemistry such as ionic strength (1e10 mM) and pH (3e9) on the
Available online 31 May 2011
mobility and deposition of CeO2 nanoparticles was investigated by using a three-phase (deposition-rinse-reentrainment) procedure in packed bed columns. The results show
Keywords:
that water chemistry governs the transport and deposition of CeO2 nanoparticles. Trans-
Ceria
port is significantly hindered at acidic conditions (pH 3) and high ionic strengths (10 mM
Flow through porous media
and above), and the deposited CeO2 particles may not be re-entrained by increasing the pH
Modeling transport and deposition
or lowering the ionic strength of water. At neutral and alkaline conditions (pH6 and 9), and
Nanoparticles
lower ionic strengths (below 10 mM), partial breakthrough of CeO2 nanoparticles was observed and particles can be partially detached and re-entrained from porous media by changing the solution chemistry. A mathematical model was developed based on advection-dispersion-adsorption equations and it successfully predicts the transport, deposition and re-entrainment of CeO2 nanoparticles through a packed bed. There is strong agreement between the deposition rate coefficients calculated from experimental data and predicted by the model. The successful prediction for attachment and detachment of nanoparticles during the deposition and re-entrainment phases is unique addition in this study. This work can be applied to access the risk of CeO2 nanoparticles transport in contaminated ground water. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
Manufactured nanoparticles have been developed and used in a diverse range of products and industries in the past few decades due to their unique and novel physicochemical properties. While their applications can benefit medicine, textiles, electronics, agriculture, cosmetics, new
materials and environmental remediation; the potential impact of engineered nanomaterials on human health and aquatic animals is not fully understood once they are released to the environment (Aitken et al., 2006; Baun et al., 2008; Bystrzejewska-Piotrowskaa et al., 2009; Dunphy Guzma´n et al., 2006; Navarro et al., 2008; Nowack and Bucheli, 2007).
* Corresponding author. Tel.: þ1 513 556 2987; fax: þ1 513 556 4162. E-mail address:
[email protected] (G.A. Sorial). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.05.025
4410
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 0 9 e4 4 1 8
Several researchers have investigated the transport and retention of nanomaterials through water-saturated porous media in packed bed columns, which represents the behavior of nanomaterials in ground water environments or engineered granular filtration systems. Experimental and mathematical modeling studies of the transport and deposition of Fullerene (C60) nanoparticles have been conducted under varying flow conditions and electrolyte species in glass beads, quartz sands and Ottawa sand (Li et al., 2008). The mobility of Fullerene and metal oxides nanoparticles in aquatic system were evaluated and compared via column study, and the role of factors such as water velocity, electrolyte species and ionic strength was investigated (Lecoanet et al., 2004). In these studies, aqueous suspensions of nanoparticles were injected into columns packed with granular media and the behavior of the nanoparticles was commonly interpreted with clean-bed filtration theory, which describes the transport by mechanisms, interception on the particle by the media, sedimentation caused by gravity, and diffusion due to Brownian motion (Yao et al., 1971). CeO2 nanoparticles are used to make novel nanomaterials, and are widely applied for polishing materials (Kosynkin et al., 2000), automobile exhaust catalysts (Fu et al., 2001), fuel cell materials (Corma et al., 2004), and additives in glass and ceramic application (Livingston and Helvajian, 2005). However, CeO2 nanoparticles have significant chronic toxicity for algae (Thill et al., 2006) and large uptake of nanoscale cerium was found in the liver of zebrafish exposed via ingestion (Johnston et al., 2010), and human lung fibroblast cells fast absorb nanoCeO2 even at a low concentration (100 ppbe100 ppm) (Limbach et al., 2005). CeO2 nanoparticles produce significant oxidative stress in human lung cells, indicating lipid peroxidation and cell membrane damage (Lin et al., 2006). On the other hand, there is limited work on the stability, mobility, transport and deposition of CeO2 nanoparticles, which determines their potential exposure and bioavailability of environmentally released particles. Hence, understanding the transport and fate of CeO2 nanoparticles is of particular interest to fill this knowledge gap. In this study, we explored the transport and deposition of CeO2 nanoparticles in water-saturated sand columns, a process that is relevant to both ground water movement and the treatment of potable water by sand filtration methods. The objective of the present experimental and modeling study is to investigate the stability, transport and deposition of commercial manufactured CeO2 nanoparticles through watersaturated pre-cleaned sand columns, and to evaluate the effect of solution chemistry (ionic strength and pH) on the mobility of CeO2 nanoparticles.
2.
Theoretical consideration
For steady-state fluid flow, the transport and retention of CeO2 nanoparticles in a homogeneous porous medium can be described by the traditional 1-dimensional advectiondispersion-sorption/desorption kinetics (Eq. (1)). The kinetics of CeO2 attachment was expressed similar to the clean-bed filtration theory (Kuhnen et al., 2000; Li et al., 2008; Saiers et al., 1994) as a function of two coefficients: One is kmod, which is
the modeled rate of nanoparticle attachment along the depth, while the other Kdet estimates the detachment rates. Smax is the maximum retention capacity of CeO2 nanoparticles within the bed. Equations (1) and (2) given below are solved according to the initial and boundary conditions given in equation (3). vC v2 C vC r vS ¼ D 2 vp b vt vx vx qw vt
(1)
rb vS Smax S r kmod C b kdet S ¼ qw vt qw Smax
(2)
CðX; t ¼ 0Þ ¼ 0 Sðx; t ¼ 0Þ ¼ 0 B:C CðX ¼ 0; tÞ ¼ C0 vC ðX ¼ L; tÞ ¼ 0 vX
(3)
I:C
where, C is the concentration of CeO2 nanoparticles in solution, S is the concentration of CeO2 nanoparticles associated with the solid phase, t is time elapsed, x is the distance parallel to the flow, rb is the sand bulk density, D is the hydrodynamic dispersion coefficient, qwis the volumetric water content and vp is pore water velocity. Prior to the introduction of CeO2 nanoparticles, a nonreactive tracer test was conducted to assess water flow and hydrodynamic dispersion in the columns (data not shown). The hydrodynamic dispersion coefficient, D, used later in the model was calculated using the tracer study. The deposition and detachment rate coefficients can be estimated by fitting the transport equations (1) and (2) and initial and boundary conditions, equation (3), to the experimental breakthrough curve using a nonlinear least square technique. The mathematical model was solved by using Mathematica software (Wolfram, 1991). The deposition of nanoparticles in porous media is limited, by the frequency of collision between the colloids and the matrix surface, and by the collision efficiency. The frequency of collision is controlled by the characteristics of nanoparticles and the matrix, and hydrodynamics of the flow. The collision efficiency is the fraction of collisions resulting in the attachment and sorption of the nanoparticles on the matrix surface. In the analysis of experimental data, attachment efficiency, a, is defined as the ratio of the rate of particle deposition on a collector to the rate of collisions with that collector that has been used. For column studies, a can be expressed as a function of the relative effluent concentration C/C0, where C is the effluent concentration at time t, and C0 is the influent concentration, shown in Eq. (4) (Yao et al., 1971): 2dc a¼ lnðC=C0 Þ 3ð1 eÞh0 L
(4)
Where dc is the median diameter of the porous media, e is the bed porosity, L is the length of the bed, and h0 is the theoretical clean-bed single collector efficiency, which describes the particle transport to an individual collector before particle accumulation alters the collector geometry. Single collector efficiency can be calculated as the sum of individual contributions each transport mechanism, and the overall correlation equation (Tufenkji and Elimelech, 2003). The experimental particle deposition rate coefficient, kexp, which
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 0 9 e4 4 1 8
represents the rate of physical chemical filtration, is related to the single-collector efficiency, h0, when the transport of colloids in saturated porous media is within the advectiondispersion range. The expression for particle deposition rate is given as (Tufenkji and Elimelech, 2003): kexp ¼
3 ð1 eÞ Uah0 2 dc e
(5)
Where U is the Darcy velocity of flow.
3.
Experimental methods
3.1. Preparation and characterization of CeO2 nanoparticles High grade CeO2 nanoparticles were obtained from Alfa Aesar (Ward Hill, MA) in suspension form (18 wt% in water with dispersants and polishing additives) and were used as is. Stock solutions of CeO2 nanoparticles were prepared by diluting the original suspension. The electrolyte concentration and pH of stock solutions were adjusted to the desired levels by adding appropriate amount of standard solutions: NaCl solution (1 M), NaOH (1 M), and HCl (1 N). Both laser diffraction particle size analyzer (Beckman Coulter LS230) and dynamic light scattering (DLS) system (ZetaSizer Nano ZS, Malvern) were used to determine the size distribution of CeO2 nanoparticles. Laser Doppler velocimetry in conjunction with phase analysis light scattering (ZetaSizer Nano, Malven) was used to measure the zeta potential of CeO2 nanoparticles. For size and zeta potential measurements, CeO2 nanoparticles suspensions were prepared with various initial ionic strengths (1e100 mM) over a pH range of 2.83e10.23. Transmission electron microscopy (JEOL, JEM2010) images were obtained to confirm the primary size distribution of the particles. Suspended CeO2 nanoparticles were sonicated for 20 min before deposited onto a copper mesh grid and left to air-dry prior to TEM analysis. Powder XRD analysis was performed to determine the chemical composition and crystallographic structure of the nanoparticles. XRD data were collected with Cu Ka radiation (l ¼ 0.15059 nm) with a 0.02 2q step size at the speed of 2 min1. UVevis absorbance spectroscopy of CeO2 nanoparticles were measured by HP 8453 over wavelength range of 200e800 nm. Calibration was based on maximum absorbance of l ¼ 309 nm.
3.2.
Transport and deposition study
Glass columns (45 cm length 2.54 cm diameter) were packed with the 20 30 mesh size fraction (geometric mean diameter ¼ 0.717 mm) of industrial mineral silica sand that contains 98.2% SiO2 and trace amount of metal oxides (AGSCO Co., IL). To remove metal and organic impurities the packing sand was thoroughly cleaned by sequential washing (1 M HNO3) water rinsing, and oven-drying (55 C, 12 h).The columns were packed with pretreated sand using a wet packing method by adding 1 cm depth at a time, yielding a bed porosity of 0.34. Schematic diagram of experimental setup is shown in Supplementary Material (Fig. 1(A)). A
4411
three-phase procedure was applied to evaluate the role of solution chemistry on the transport, deposition and retention of CeO2 nanoparticles following the procedure reported earlier (McDowell-Boyer, 1992). In Phase I (Deposition Phase), six pore volumes of CeO2 nanoparticle suspension was introduced continuously to the column, followed by Phase II (Rinse Phase), during which the column was rinsed with four pore volumes of the particle free solution which has the same solution chemistry as the nanoparticle suspensions used in Phase I. The ionic strength and pH was kept the same during these two phases but varied for different tests. In Phase III (Re-entrainment Phase), MilliQ water was fed to the column to lower the ionic strength. The flow rate was kept constant at 20 mL/min for all three phases, providing NRe of 0.844. A sample of 3 mL was collected from the effluent stream every 1 min and was analyzed by UVevis spectrometer at the wavelength of l ¼ 309 1 nm. Calibration of UVevis spectrometer based on multi-concentration standard samples of CeO2 nanoparticles were used for quantitative analysis as shown in Fig. B-1 in supplementary material.
4.
Results
4.1.
Characterization of CeO2 nanoparticles
The particle size distribution of the CeO2 NPs in 1 mM NaCl solution was measured by both DLS and laser diffraction particle analyzer (Fig. 1(a)). DLS intensity based The average hydrodynamic diameter as determined by DLS analyzer was 152.7 nm, and the laser diffraction particle analyzer provided a mean diameter of 62.6 nm. DLS measurements also gave a wider size distribution. This difference can be accounted to differences in the techniques of the two methods. DLS measures the hydrodynamic diameter, which refers to how a particle diffuses in fluid, where a laser beam measures time-dependent fluctuations in scattered intensity caused by particles undergoing Brownian motion. The intensity fluctuations are auto correlated by particle size distribution according to Doppler Effect. Whereas, for laser diffraction particle analyzer, particle size distribution is obtained by measurements of low angle light scattering intensity as a function of the scattering angle, the wavelength and polarization of light measured based on applicable scattering models. Correlations for both measurement techniques are based on the assumption that the particles are spherical. For a monodispersed, spherical latex particle standard (Standard, L300, Nominal 300 nm Latex Particles, Beckman Coulter), the two techniques give identical size distribution. Hence the difference in size distribution of the two techniques (Fig. 1(a)) indicates that the CeO2 used in this study are non-spherical particles, which is confirmed by TEM image. A representative TEM image of CeO2 nanoparticles is shown in Fig. 1-b, which shows the variance in shape and size of the primary particles and provides evidence for aggregation. The images show that the 2dimensional projects of the primary particles range from ca.5 nm diameter to ca.60 nm diameter and appear to have triangle, pentagonal and hexagonal shapes similar to self-
4412
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 0 9 e4 4 1 8
Fig. 1 e Characterization of CeO2 nanoparticles. (a) Size distribution of CeO2 in MilliQ water, measured by DLS and Laser Diffraction pattern, with TEM image inserted, (b) XRD of CeO2 nanoparticles.
assembled CeO2 nanostructure reported in literature (Wang and Feng, 2003). XRD analysis verified the purity and crystallinity of CeO2 (Fig. 1(b)). The peaks of diffraction angle (28.6 , 33.1 , 47.6 , 56.5 , 76.9 , 79.2 , and 88.3 ) match the maximum diffraction angles and relative intensities of ceria from the Trace database (PDF number 75-0390) with primary particle size of 54 nm, confirming the CeO2 nanoparticles are ceria particles with nanocrystalline structure.
4.2.
Stability of CeO2 nanoparticles
The electrophorotic stability of CeO2 was studied by measuring the zeta potential across a range of pH (2.83e10.23) at ionic strength of 1 and 10 mM NaCl (Fig. 2-a). These data show decrease in the magnitude of the zeta potential with increasing ionic strength due to the compression of electrical double layers. As the pH increased from 2.8 to 10.23, zeta potential changed from 4.41 mV to 28.8 mV in 1 mM NaCl solution, and from 1.98 mV to 39.7 mV in 10 mM NaCl solution. Aggregation of CeO2 nanoparticles was studied by measuring the hydrodynamic size distribution after adding
Fig. 2 e (a) Zeta potential and electrophoretic mobility of CeO2 nanoparticles as a function of pH at two levels of ionic strength (1 mM and 10 mM) (b) Intensity-weighted size distribution of CeO2 in suspension under variable electrolyte concentration and pH.
electrolyte or changing pH followed by sonication for 15 min and samples were left undisturbed up to 24 h. At pH 6, the size distribution of CeO2 nanoparticles did not show significant change, indicating little aggregation occurred (Fig. 2-b). However, at pH 3, which approaches the point of zero charge (pHZPC), size distribution become wider with multiple peaks, indicating that aggregation and polydispersity occurred due to the formation of bi- and trimodal systems.
4.3.
Transport and deposition of CeO2 in porous media
4.3.1.
Screening study
A factorial screening experimental design was first conducted to study the influence of water chemistry (NaCl concentration of 1, 10 and 100 mM, and pH 3, 6, and 9). Effluent stream concentration of CeO2 nanoparticles were monitored and breakthrough curves are presented in Fig. 3. The normalized effluent nanoparticle concentration (C/C0) is shown as a function of cumulative volume normalized to pore volume. The breakthrough curves of CeO2 nanoparticle suspensions through porous media differ distinctively for different solution chemistry. The results are summarized into three cases. Case (1) Breakthrough during deposition phase, where sharp breakthrough curves were observed at one pore volume for two of the test conditions at pH ¼ 6, 1 mM NaCl, and pH 9, 10 mM NaCl. The normalized value of
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 0 9 e4 4 1 8
4413
breakthrough occurred during deposition phase, indicating all the nanoparticles were captured by the sand. However, sharp peaks were observed during the re-entrainment phase, when the ionic strength of the column was lowered by introducing MilliQ water and the pH was gradually brought back to 5.6 0.5 (the pH of MilliQ water). This indicates the detachment of weakly attached nanoparticles that were previously captured during the deposition phase by the porous media. Case (3) There were no breakthrough during deposition phase and no peaks were observed during reentrainment phase. For test runs at pH of 3, no CeO2 nanoparticles were detected in the effluent streams during deposition phase or re-entrainment phase, indicating stronger attachment of CeO2 to sand bed at lower that was not reversed at neutral pH or lowering the ionic strength.
4.3.2.
Fig. 3 e A profile of experimental breakthrough curves on an experimental design of 9 runs at three pH levels (pH 3, 6, and 9) and under ionic strength of a) 1 mM, b) 10 mM, and c) 100 mM. The lines represent model predictions.
C/C0 showed a step increase from the initial value of zero to a plateau of 0.9 0.5. During the rinse phase, the normalized effluent stream concentration decreased sharply to almost zero when particle free solution was fed to the column, indicating little or no deposition occurring at these two conditions. Case (2) No breakthrough was observed during deposition phase, but peaks were observed during reentrainment phase: For test runs operated at conditions of pH ¼ 6, 10 or 100 mM NaCl, pH ¼ 9, 10 or 100 mM NaCl, no
The role of ionic strength
The screening study has shown significant difference between 1 mM and 10 mM NaCl (Fig. 3a and 3b), while higher levels of ionic strength showed little effects (Fig. 3c). The results of rigorous test runs made at pH 6 with ionic strengths of 1, 2, 3, 5, and 10 mM for understanding the effects of ionic strength are presented in Fig. 4. During the deposition phase, fast and complete breakthrough curves were observed at ionic strength of 1 mM and 2 mM, where the normalized effluent concentration rapidly reached to 0.96 0.4. Whereas partial breakthroughs and gradual increase in C/C0 were observed with decreasing slope and plateau values as the ionic strength changed from 2, to 3 and 5 mM. Instead of fast, steep breakthrough as observed before, the normalized nanoparticle concentrations increased gradually to a plateau. The specific shape of the partial breakthrough can be explained by the blocking effect (Chen et al., 2002; Kuhnen et al., 2000; Song and Elimelech, 1994). As seen in Fig. 2, at pH 6, CeO2 NPs were negatively charged from 1 to 10 mM NaCl, and the size distribution of CeO2 NPs didn’t widen for 24 h duration, indicating that stable particleeparticle interaction predominates in these conditions. In this case, only a monolayer of deposited particles is formed on the sand surface and multi layer deposition could be neglected. Decrease in ceria deposition rate was reflected by the gradual increase in C/C0 with time, forming partially breakthrough. The maximum surface coverage was increased due to the increased screening of particle surface charge as could be seen in table with values of Smax of 0.1, 0.2, 1.2 mg/g corresponding to increased ionic strength of 3, 5 and 10 mM in runs 11, 12 and 6, respectively. During the rinse phase, C/C0 values decreased sharply to zero as particle free solution was fed to the column and during the re-entrainment phase a pulse increased with narrow, sharp peaks were observed. The peak height increased with increase in ionic strength of the influent nanoparticle suspensions. The attachment efficiency, a, and the deposition rate coefficient kexp were determined using equation (4) and 5 for each transport experiment. Mathematical expressions based on mass balance were developed to evaluate the mass fraction of nanoparticles recovered during the rinse phase, phase 2, and the re-entrainment phase, phase 3 (Franchi and O’Melia, 2003). The fractions of nano-CeO2 recovered during the two phases, FRC2 and FRE3 are defined as:
4414
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 0 9 e4 4 1 8
P Mass of NPs Recovered During Phase 2 FRC2 ¼ P Mass of NPs Deposited During Phase 1
ð5Þ
P
FRC3 ¼ P
Mass of NPs Recovered during Phase 3 Mass of NPs Deposited During Phase 1 Mass of NPs Recovered during Phase 2
Fig. 5-a shows attachment efficiency versus ionic strength at two distinct regions. At low ion concentrations (1e5 mM), where the attachment efficiency and deposition rate coefficient increased proportionally with increasing NaCl concentration, whereas at high ion concentrations above 10 mM deposition rate coefficient and the attachment efficiency were independent of NaCl concentration. Fig. 5-b presents logarithmic values of FRC2 and FRE3 as a function of log concentrations of NaCl. FRC2 values are in the range of 0.803w0.946 at ionic strength of 3 mM and lower, indicating almost complete detachment of nanoparticles during the rinse phase without changing the water chemistry. When the ionic strength increase to 5 mM and higher, there was a sharp drop in the FRC2 value from 0.9 to 0.105, indicating large retention of nanoparticles occurred in Phase 1. The values of FRE3 decreased as ionic strength increased. When the ionic strength was 1 mM and 2 mM, the corresponding values of FRE3 were0.694 and 0.390, respectively, indicating the nanoparticles deposited in phase 1 were subsequently reentrained. As ionic strength increased to 3 mM and above, FRE3 values dropped to below 0.033. Batch CeO2 nanoparticle suspensions left undisturbed showed little aggregation even after 24 h at pH 6, 1e100 mM NaCl. This suggests that, although particle aggregation and deposition kinetics are closely related, the hindered transport at higher ionic strength in this study is not caused by nanoparticle aggregation. Unlike other similar studies, such as fullerence C60 nanoparticles (Brant et al., 2005), where
(6)
deposition is related to particle aggregation induced by the screening effect of electrolyte concentration, these effects were less obvious with CeO2 nanoparticles. The increase of electrolyte concentration, at pH of 6, significantly increased nanoparticles attachment and deposition to the bed matrix, although there are limited aggregation of CeO2 nanoparticles. Derjaguin-Landau-Verwey-Overbeek (DLVO) theory (Derjaguin and and Landau, 1941; Verwey and Overbeek, 1948) was used to explain the particle-sand grain interaction energy at each ionic strength. The classical DLVO theory of colloidal stability describes the total interaction energy experienced by nanoparticles when approaching a collector surface as the sum of van der Waals (VDW) and electrical double layer (EDL) repulsion. Theoretical analysis was conducted with a spheareplate interaction to calculate the total interaction energy as a nanoparticle approaches sand surface (Derjaguin and Landau, 1941; Elimelech and O’Melia, 1990; Gregory, 1981; Healy and White, 1978; Hogg et al., 1966). The variation of the DLVO interaction energy with separation distance at different ionic strengths at pH 6 is shown an insert in Fig. 4. Transport behavior of CeO2 nanoparticles at various levels of ionic strength is in qualitative agreement with the DLVO theory (DLVO theory calculation is provided in the Supplimentary Material). At low ionic strength, diffusion layer surrounding nanoparticles and collector surfaces cause the screening effect of the salt is smaller than the electrostatic repulsion between particles and sand surface. For solutions with ionic strength of 1 and 2 mM, calculated values predict
Fig. 4 e The influence of ionic strength on the breakthrough curves at pH [ w6. The lines represent model prediction.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 0 9 e4 4 1 8
Fig. 5 e Effects of ionic strength on deposition and reentrainment of CeO2 nanoparticles, showing (a) attachment efficiency a, rate constant, and kexp, (b) mass fractions of recovered CeO2 during phase II (FRC2,) and Phase III (FRE3).
the presence of a substantial repulsive energy barrier to deposition ranging from 33 to 19 kT at 1 mM and 2 mM, respectively. The high-energy barrier suggested that only a small fraction of the particles can overcome the primary energy barrier and deposit onto sand surface. At conditions when complete breakthrough was observed, the attachment efficiency and deposition rate constants increased with the increase of ionic strength and strong electrostatic repulsion effectively hindered attachment of nanoparticles to the sand surface. Low attachment efficiency results smaller deposition rate and the deposition kinetic becomes reaction-limited. As the ionic strength increased, the diffuse layers of sand surface and nanoparticles are progressively compressed and consequently, electrostatic repulsion is reduced. At solution ionic strength of 3e5 mM, the interaction energy calculations indicate small energy barrier to deposition, also the secondary minimum is deeper and located at a closer distance of separation between sand surface, suggesting that mechanism can be a combination of disposition in both primary minimum and secondary minimum. This is supported by the observation that only a portion of the particles remaining in the porous media after Phase 2 attached to the sand surface. Partial breakthrough of the column studies suggests as CeO2 nanoparticles approach silica sand, they experience attractive
4415
force and deposition occurred at the shallow secondary minimum well, where the nanoparticles are continuously captured and released by the sand surface. The slope and plateau value of the breakthrough curves decrease with the increase of the ionic strength because the depth of decreasing secondary minimum. The deposition rate coefficient and the attachment efficiency increased when ionic strength increased from 3 to 5 mM, and became independent of ionic strength at 10 mM and above. The deposition rate coefficient is independent of the ionic strength at high ion concentrations duet to the reduction of electrostatic repulsion between nanoparticles and porous matrix surface. At these conditions, more collisions result in attachment, and the process was fast and the kinetic is limited by mass transfer. Analysis of effluent stream during Phase 2 indeed deposited in secondary energy-minimum. During Phase 3, sharp peaks of pulse release of nanoparticles were observed, which is vident of the deposition in secondary minimum. The introduction of MilliQ water to the column during Phase 3 eliminated the presence of secondary minimum, resulting rapid release of nanoparticles previously deposited at higher ionic strengths from sand surface. Fig. 5(b) shows the fraction of deposited particles that was eluted during Phase 3, given as FRE3, decreased with the increase of ionic strength. This indicates that as the ionic strength increased, the contribution of secondary minimum became smaller and more fraction of nanoparticles were deposited in the primary energy well. While the role of the calculated secondary minimum provided theoretically explanation for the deposition and reentraiment of CeO2, it needs to be noted that certain limitations of the DLVO calculation here. For instance, the zeta potential was used in the calculation as estimation for surface charge. Also sphere-to-plate interaction energy profiles are sensitive to changes in particle size. The height of the repulsive energy barrier and the depth of the secondary energy well both change with changing particle diameter. In this study the energy barrier height and the depth of secondary minimum calculation are based on the average diameter of the NPs. Due to the wide size distribution and polydispersitvity of the NPs, as mentioned before, the energy interaction profile can only represent the majority of the NPs in the suspension and a small fraction of NPs that are either very small or large cannot be represented here. Calculations of the theoretical predictions assume that the particles are spherical, while both the particles and collectors selected for this study are likely to be considerably more spherical than the majority of those encountered in realistic situations, neither the CeO2 NPs nor the sand grains are perfectly spherical. The angularity of the collector grain shape could also contribute to the removal of NPs by physical straining effect. These limitations could cause the deviation between the DLOV theoretical prediction and the actual deposition behaviors.
4.3.3. model
Predicting transport of nano-CeO2 with mathematical
The mathematical model for the transport and retention of CeO2 nanoparticles was optimized to estimate the breakthrough under corresponding experimental conditions. The values of kmod and Smax were determined by solving Eq. (1) and Eq. (2) and the necessary initial and boundary conditions,
4416
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 0 9 e4 4 1 8
Table 1 e Attachment and detachment coefficients at different operating conditions.
Run Run Run Run Run Run Run Run Run Run Run Run
1 2 3 4 5 6 7 8 9 10 11 12
Ionic strength
pH
Initial CeO2 Concentration mg/L
Smax (mg/g)
kmod (1/Hr)
kexp (1/Hr)
kdet (1/Hr)
10 mM 1 mM 100 mM 1 mM 10 mM 10 mM 1 mM 100 mM 100 mM 2 mM 3 mM 5 mM
9 6 9 9 3 6 3 6 3 6 6 6
10 10 50 50 50 50 10 50 50 50 50 50
0.1 1.0 1.2 1.0 1.0 1.2 0.9 1.1 1.0 1.0 0.1 0.2
24.5 0.3 34.9 0.1 27.0 25.1 20.2 25.6 27 0.5 2.3 8.4
26.1 0.70 30.7 0.1 30.7 23.37 22.5 27.6 31.4 0.77 1.91 11.62
0.7 0.0 1.2 0.1 0.0 0.6 0.0 0.8 0.0 0.1 0.1 0.1
Eq. (3), for the Deposition Phase, while assuming kdet is negligible. The effect of kdet was then considered for the whole column run while optimizing the value of kdet in the Reentrainment Phase. The initial conditions presented in Eq.(3) apply only for the first phase. For the second and third phases, the initial conditions used are the final distributions calculated for both CeO2 concentrations in solution (C ) and associated to the solid phase (S ) from the earlier phase. The model successfully captured the sharp increase and decline of the CeO2 concentration in the first two phases. There was a strong agreement of the deposition rate coefficients calculated from transport experimental data, kexp, and predicted by the model, kmod,(see Fig. 5-a), confirming successful simulation of the deposition process. The dispersion term in Equation (1) accounts for broadening the elution curve. For conditions where breakthrough of the nano-CeO2 occurred this term was insignificant (Chen et al., 2006). The volumetric water content was determined independently from physical measurements in the column. The results are plotted together with the experimental data in Figs. 3 and 4. Model predictions of effluent concentration fitted well with experimental data for Phases 1 and 2, however, the model parameter optimization failed to converge and did not fully capture the sharp increases and drop of effluent curve during Phase 3. Although, previous studies have shown steady-state condition of the first deposition phase (Liu et al., 2009) both first phases (Hydutsky et al., 2007; Li et al., 2008) or a pulse injection in the column (Hanna et al., 2010), the modeling effort for attachment and detachment of nanoparticles with changes in the fluid chemistry during the deposition and reentrainment phase is a unique contribution of this study. The mathematical modeling efforts have the advantage in predicting the amount and distribution of CeO2 held within the columns during all the phases. Particle deposition continues to grow within the column during the test run of pH 6, and ionic strength of 3 mM. As shown in Table 1, the maximum attachment concentration occurred at the column inlet and the amount increased during the deposition phase. The attachment distribution continued to grow linearly along the length of the column with the highest deposition observed near the inlet. Deposition within the column redistributed suspended particles to sand surface during Phase 2. The same behavior was observed in run 12 (pH 6, ionic strength 5 mM),
where higher deposition rates were observed. The maximum attachment at the inlet calculated for this run was 0.085 mg/g maintaining a ratio of 10% of the maximum deposition Smax (0.2 mg/g) similar to the previous run. For the runs without breakthrough the deposition was also maximum near the inlet. The deposited mass of nanoparticles dropped rapidly by an order of magnitude at half the length of the column. For instance, in run 9 the maximum attachment calculated was 0.35 mg/g which is 35% of the total Smax. On the other hand, with full breakthrough, i.e. run 2, the deposition was almost negligible. Experimental effort was made to validate the distribution of deposited nanoparticles in the column by taking portions of the bed along the length of the column, washing and sonicating in MilliQ water to detach deposited nano-CeO2. Unfortunately, desorbed silica interfered strongly with the analysis. A selective and quantitative technique is needed that can give 3-D distribution of deposited particles.
5.
Conclusion
Accidental or deliberate introduction of CeO2 nanoparticles into subsurface environments may lead to contamination of drinking water supplies and can act as colloidal carriers for sorbed contaminants. This study highlights the implication of CeO2 nanoparticle and provides important insights to confirm the mobility of CeO2 nanoparticle under typical ground water movements. It is not straightforward to precisely predict the behavior of CeO2 nanoparticle upon their environmental release because of the compliance of various environmental parameters. However, the values of attachment efficiency, deposition rate coefficients, fraction recovered/re-entrained at varying ionic strength and the modeling results provided in this study can be used to further estimate the relative mobility and evaluate the potential exposure and risk of CeO2 nanoparticles. Results of column studies clearly showed that the water chemistry governs the transport, deposition and reentrainment of nanoparticles. The increase of ionic strength decreased the mobility of CeO2 nanoparticles due to the compression of electrostatic double layer repulsion, which is in general agreement with DLVO theory prediction. The three phase flow method allowed better understanding of the
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 0 9 e4 4 1 8
influence of ionic strength in determining the condition whether the capture-release dynamics is reaction-limited or transport limited. A mathematical model was developed and successfully simulates both complete and partial breakthrough and re-entrainment of the CeO2 breakthrough curves. The model is also capable of predicting the distribution of nanoparticle deposition within the porous media.
Acknowledgments The authors are grateful for the financial support from U.S. Environmental Protection Agency under contract No. PRC108-1170.
Appendix. Supplementary data Supplementary data related to this article can be found online at doi:10.1016/j.watres.2011.05.025.
references
Aitken, R.J., Chaudhry, M.Q., Boxall, A., Hull, M., 2006. Manufacture and use of nanomaterials: current status in the UK and global trends. Occupational Medicine 56 (5), 300e306. Baun, A., Hartmann, N., Grieger, K., Kusk, K., 2008. Ecotoxicity of engineered nanoparticles to aquatic invertebrates: a brief review and recommendations for future toxicity testing. Ecotoxicology 17 (5), 387e395. Brant, J., Lecoanet, H., Wiesner, M.R., 2005. Aggregation and deposition characteristics of fullerene nanoparticles in aqueous systems. Journal of Nanoparticle Research 7 (4), 545e553. Bystrzejewska-Piotrowskaa, G., Golimowskib, J., Urbana, P.L., 2009. Nanoparticles: their potential toxicity, waste and environmental management. Waste Management 29, 2587e2595. Chen, J.Y., Klemic, J.F., Elimelech, M., 2002. Micropatterning microscopic charge heterogeneity on flat surfaces for studying the interaction between colloidal particles and heterogeneously charged surfaces. Nano Letters 2 (4), 393e396. Chen, K.L., Mylon, S.E., Elimelech, M., 2006. Aggregation kinetics of alginate-coated hematite nanoparticles in monovalent and divalent electrolytes. Environmental Science and Technology 40 (5), 1516e1523. Corma, A., Atienzar, P., Garcia, H., Chane-Ching, J.-Y., 2004. Hierarchically mesostructured doped CeO2 with potential for solar-cell use. Nature Materials 3 (6), 394e397. Derjaguin, B.V., Landau, L., 1941. Acta Physicochimca URSS. Dunphy Guzma´n, K.A., Taylor, M.R., Banfield, J.F., 2006. Environmental risks of nanotechnology: National nanotechnology initiative funding, 20002004. Environmental Science and Technology 40 (5), 1401e1407. Elimelech, M., O’Melia, C.R., 1990. Effect of particle size on collision efficiency in the deposition of Brownian particles with electrostatic energy barriers. Langmuir 6 (6), 1153e1163. Franchi, A., O’Melia, C.R., 2003. Effects of natural organic matter and solution chemistry on the deposition and reentrainment of colloids in porous media. Environmental Science and Technology 37 (6), 1122e1129.
4417
Fu, Q., Weber, A., Flytzani-Stephanopoulos, M., 2001. nanostructured AueCeO2 catalysts for low-temperature wateregas shift. Catalysis Letters 77 (1), 87e95. Gregory, J., 1981. Approximate expressions for retarded Van der waals interaction. Journal of Colloid and Interface Science 83 (1), 138e145. Hanna, K., Rusch, B., Lassabatere, L., Hofmann, A., Humbert, B., 2010. Reactive transport of gentisic acid in a hematite-coated sand column: experimental study and modeling. Geochimica et Cosmochimica Acta 74 (12), 3351e3366. Healy, T.W., White, L.R., 1978. Ionizable surface group models of aqueous interfaces. Advances in Colloid and Interface Science 9 (4), 303e345. Hogg, R., Healy, T.W., Fuerstenau, D.W., 1966. Mutual coagulation of colloidal dispersions. Transactions of the Faraday Society 62, 1638e1651. Hydutsky, B.W., Mack, E.J., Beckerman, B.B., Skluzacek, J.M., Mallouk, T.E., 2007. Optimization of nano- and microiron transport through sand columns using polyelectrolyte mixtures. Environmental Science and Technology 41 (18), 6418e6424. Johnston, B.D., Scown, T.M., Moger, J., Cumberland, S.A., Baalousha, M., Linge, K., van Aerle, R., Jarvis, K., Lead, J.R., Tyler, C.R., 2010. Bioavailability of nanoscale metal oxides TiO2, CeO2, and ZnO to Fish. Environmental Science and Technology 44 (3), 1144e1151. Kosynkin, V.D., Arzgatkina, A.A., Ivanov, E.N., Chtoutsa, M.G., Grabko, A.I., Kardapolov, A.V., Sysina, N.A., 2000. The study of process production of polishing powder based on cerium dioxide. Journal of Alloys and Compounds 303-304, 421e425. Kuhnen, F., Barmettler, K., Bhattacharjee, S., Elimelech, M., Kretzschmar, R., 2000. Transport of iron oxide colloids in packed quartz sand media: monolayer and multilayer deposition. Journal of Colloid and Interface Science 231 (1), 32e41. Lecoanet, H.F., Bottero, J., Wiesner, M.R., 2004. Laboratory assessment of the mobility of nanomaterials in porous media. Environmental Science and Technology 38 (19), 5164e5169. Li, Y., Wang, Y., Pennell, K.D., Abriola, L.M., 2008. Investigation of the transport and deposition of Fullerene (C60) nanoparticles in quartz sands under varying flow conditions. Environmental Science and Technology 42 (19), 7174e7180. Limbach, L.K., Li, Y., Grass, R.N., Brunner, T.J., Hintermann, M.A., Muller, M., Gunther, D., Stark, W.J., 2005. Oxide nanoparticle uptake in human lung fibroblasts: effects of particle size, agglomeration, and diffusion at low concentrations. Environmental Science and Technology 39 (23), 9370e9376. Lin, W., Huang, Y., Zhou, X., Ma, Y., 2006. Toxicity of cerium oxide nanoparticles in human lung cancer cells. International Journal of Toxicology 25 (6), 451e457. Liu, X., Wazne, M., Christodoulatos, C., Jasinkiewicz, K.L., 2009. Aggregation and deposition behavior of boron nanoparticles in porous media. Journal of Colloid and Interface Science 330 (1), 90e96. Livingston, F.E., Helvajian, H., 2005. Variable UV laser exposure processing of photosensitive glass-ceramics: maskless microto meso-scale structure fabrication. Applied Physics A: Materials Science and Processing 81 (8), 1569e1581. McDowell-Boyer, L.M., 1992. Chemical mobilization of micron-sized particles in saturated porous media under steady flow conditions. Environmental Science and Technology 26 (3), 586e593. Navarro, E., Baun, A., Behra, R., Hartmann, N., Filser, J., Miao, A.-J., Quigg, A., Santschi, P., Sigg, L., 2008. Environmental behavior and ecotoxicity of engineered nanoparticles to algae, plants, and fungi. Ecotoxicology 17 (5), 372e386. Nowack, B., Bucheli, T.D., 2007. Occurrence, behavior and effects of nanoparticles in the environment. Environmental Pollution 150 (1), 5e22. Saiers, J.E., Hornberger, G.M., Liang, L., 1994. First- and secondorder kinetics approaches for modeling the transport of
4418
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 0 9 e4 4 1 8
colloidal particles in porous media. Water Resource Research 30 (9), 2499e2506. Song, L., Elimelech, M., 1994. Transient deposition of colloidal particles in heterogeneous porous media. Journal of Colloid and Interface Science 167 (2), 301e313. Thill, A., Zeyons, O., Spalla, O., Chauvat, F., Rose, J., Auffan, M., Flank, A.M., 2006. Cytotoxicity of CeO2 nanoparticles for Escherichia coli. Physico-chemical insight of the cytotoxicity mechanism. Environmental Science and Technology 40 (19), 6151e6156. Tufenkji, N., Elimelech, M., 2003. Correlation equation for predicting single-collector efficiency in physicochemical
filtration in saturated porous media. Environmental Science and Technology 38 (2), 529e536. Verwey, E., Overbeek, J., 1948. Theory of the Stability of Lyophobic Colloids. Elsevier, Amsterdam. Wang, Z.L., Feng, X., 2003. Polyhedral shapes of CeO2 nanoparticles. The Journal of Physical Chemistry B 107 (49), 13563e13566. Wolfram, S., 1991. Mathematica: A System for Doing Mathematics by Computer, second ed. Addison Wesley Longman Publishing Co., Inc, Redwood City, CA, USA. Yao, K., Habibian, M.T., O’Melia, C.R., 1971. Water and waste water filtration. Concepts and applications. Environmental Science and Technology 5 (11), 1105e1112.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 1 9 e4 4 2 7
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Kinetics of model high molecular weight organic compounds biodegradation in soil aquifer treatment Peter Fox, Roshan Makam* Department of Civil and Environmental Engineering, Arizona State University, PO Box 5306, Tempe, AZ 85287-5306, USA
article info
abstract
Article history:
Soil Aquifer Treatment (SAT) is a process where treated wastewater is purified during
Received 6 December 2010
transport through unsaturated and saturated zones. Easily biodegradable compounds are
Received in revised form
rapidly removed in the unsaturated zone and the residual organic carbon is comprised of
14 May 2011
primarily high molecular weight compounds. This research focuses on flow in the satu-
Accepted 22 May 2011
rated zone where flow conditions are predictable and high molecular weight compounds
Available online 15 June 2011
are degraded. Flow through the saturated zone was investigated with 4 reactors packed with 2 different particle sizes and operated at 4 different flow rates. The objective was to
Keywords:
evaluate the kinetics of transformation for high molecular weight organics during SAT.
Soil aquifer treatment
Dextran was used as a model compound to eliminate the complexity associated with
High molecular weight organics
studying a mixture of high molecular weight organics. The hydrolysis products of dextran
Hydrolysis
are easily degradable sugars. Batch experiments with media taken from the reactors were
Dextran
used to determine the distribution of microbial activity in the reactors. Zero-order kinetics
Kinetics
were observed for the removal of dextran in batch experiments which is consistent with
Saturated
hydrolysis of high molecular weight organics where extracellular enzymes limit the substrate utilization rate. Biomass and microbial activity measurements demonstrated that the biomass was independent of position in the reactors. A Monod based substrate/ biomass growth kinetic model predicted the performance of dextran removal in the reactors. The rate limiting step appears to be hydrolysis and the overall rate was not affected by surface area even though greater biomass accumulation occurred as the surface area decreased. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
Water Reuse has become important in arid areas throughout the world including regions such as the southwestern United States and the Mideast. Soil aquifer treatment (SAT) is a technology where natural systems have been employed to treat wastewater for indirect potable reuse (Bouwer and Rice, 1984; Bouwer, 1985; Amy et al., 1993; Drewes and Jekel, 1998; Wilson et al., 1995, Wild and Reinhard, 1999). SAT involves
the water quality benefits derived during percolation through vadose zone sediments and subsequent ground water transport. The organic matter in treated wastewater is a complex mixture of simple carbohydrates, amino acids, alcohols, volatile fatty acids mixed with polymers and heteropolymers including proteins (1/3 of COD), polysaccharides (1/5 of COD) and lipids (1/3 of COD) along with Natural Organic Matter (NOM) and Soluble Microbial Products (SMPs) (Raunkjaer et al., 1994). In the unsaturated zone, easily biodegradable low
* Corresponding author. Biotechnology Department, PES Institute of Technology, 82 East End ’B’ Main Road, Jayanagar 9th Block, Bangalore, Karnataka 560069, India. Tel.: þ91 80 26633721. E-mail addresses:
[email protected] (P. Fox),
[email protected] (R. Makam). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.05.023
4420
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 1 9 e4 4 2 7
molecular weight organics are biodegraded while residual high molecular weight organics are biodegraded over longer time periods in the saturated zone (Fox et al., 2001; Amy et al., 2006). The removal of dissolved organic carbon may be modeled as a first order reaction where the organic components are divided into fractions with different biodegradation rates. Dissolved organic carbon (DOC) is often used as a surrogate to monitor the removal of organic compounds without consideration for the characteristics of individual organic compounds. During saturated flow, the majority of organic carbon transformations are associated with high molecular weight organics. The high molecular weight fraction cannot be assimilated directly by microorganisms (Levine et al., 1985). Bacteria assimilate high molecular weight compounds via hydrolysis by extracellular enzymes. These extracellular enzymes are either bound to the cell surface (ecto-enzymes) (Chrost, 1991) or released (exoenzymes) (Vetter and Deming, 1999; Wetzel, 1991) into the medium so as to hydrolyze high molecular weight compounds. Cadoret et al. (2002) has demonstrated the utilization of low and high molecular weight substrates as a consequence of extracellular enzymes in whole and dispersed activated sludges (e.g.: azocasein MW ¼ 26,000, amulose azure 32,000 < MW < 86,000). Slow hydrolysis of organic matter by exoenzymes has been demonstrated by Eliosov and Argaman (1995) and Guellil et al. (2001). When substrate utilization kinetics for high molecular weight DOC are dominated by hydrolysis of complex substrates, zero-order kinetics with respect to substrate concentration may be observed. This research focuses on understanding the rate limiting step for the biodegradation of DOC in the saturated zone during SAT. These transformations are vital for the sustainability of SAT and they support microbial activity for long sub-surface travels times allowing for potential co-metabolic reactions of trace anthropogenic compounds (Nalinakumari et al., 2010; Wild and Reinhard, 1999). The objective of this research is to evaluate the kinetics of biodegradation of a model high molecular weight compound during flow through a saturated media. A model compound was chosen to avoid the complexities associated with a mixture of many compounds and provide insight into the mechanisms of removal for a high molecular weight compound. To accomplish this, two different particle size sands were used to provide two different surface areas for microbial attachment and the flow rates were varied. The experimental design maintained aerobic conditions to eliminate complexities from varying redox conditions.
2.
Methods
2.1.
Dextran as a model compound
Dextran was chosen as the model compound because it is a high molecular weight compound (MW ¼ 10,000), mimics polysaccharide soluble microbial products and is readily biodegradable. The hydrolysis products of dextran are sugars that easily biodegradable. Analysis of carbohydrates can be used for monitoring dextran and its hydrolysis products. When combined with dissolved organic carbon measurements, a mass balance on dextran can be done.
2.2.
Reactors
The experimental apparatus was designed to simulate saturated flow in a sand aquifer. The experimental apparatus consisted of 4 cylindrical reactors packed with with two different particle sizes of sand. The reactors were 0.915 m tall with a 0.076 m inner diameter and were constructed of Plexiglass. Two different clean silica sand sieve sizes were used as packing material (Border Products, Arizona). One size was US standard sieve 16 x 30 with a geometric mean diameter of 0.6 mm and the second size was US standard sieve 40 x 60 with a geometric mean diameter of 0.353 mm. The sands were washed with de-ionized water to remove any residual fines and dried before packing. The sands were packed in the reactors to achieve an average dry packing density of 1.4 g/ cm3. Reactors 1 and 4 were packed with 0.6 mm silica sand and Reactors 2 and 3 were packed with 0.353 mm silica sand. The reactors were operated in an upflow mode. The 4 reactors were seeded by feeding filtered nitrified/ denitrified effluent from the Mesa Northwest Reclamation Plant, Arizona for a period of 50 days. After seeding, the reactors were operated with synthetic feed containing dextran (average M.W. ¼ 10,000) (Sigma, St. Louis, Missouri, USA) as the substrate. The synthetic feed had a nominal concentration of 6.8 mg DOC/L added to dechlorinated drinking water. The background DOC of the dechlorinated tap water used to formulate the influent feed was found to be at 1.26 þ 0.4 mg DOC/L. During previous studies using the same tap water in soil columns, less than 0.2 mg/L of the DOC was removed over a 30 day retention time (Nalinakumari et al., 2010). During this study, the retention times were much lower and therefore the natural organic matter in the tap water should not significantly influence the results. Based on the stoichiometry, the tap water contained sufficient nutrients and there was no need add supplemental nutrients. Weekly monitoring of turbidity, UVA254 and dissolved organic carbon was done. The flow rates chosen for simulating sub-surface transport provided a Reynolds Number less than 1 and the Peclet Number ranged from 0.2 to 6. Ground water recharge sites have saturated flow conditions with Reynolds Numbers less than 1 and for most aquifer materials the Peclet Numbers range from 0.2 to 36. The reactors were operated in two phases with different flow rates to evaluate a range of conditions that occur during saturated flow in SAT systems. During Phase 1, Reactors 1 and 2 were operated at 0.5 L/day and Reactors 3 and 4 were operated at 4 L/day. All 4 reactors were operated a minimum of 150 days under saturated aerobic conditions to achieve steady state. Once Phase I was completed, Phase II was initiated. During Phase II, the flow rate to Reactors 1 and 2 was increased 4 fold to 2 L/day and the flow rate to Reactors 3 and 4 was decreased by a factor of 4 to 1 L/day. All 4 reactors were run for at least 50 days under saturated aerobic conditions during Phase II. Table 1 shows the operating conditions for the four reactors during Phases 1 and 2. In this experimental design, the empty bed contact in Reactors 1 and 2 was the always identical and the empty bed contact time in Reactors 3 and 4 was always the same. Table 1 also shows the mean effluent concentrations under these conditions along with the student’s t-test results for comparing the reactors.
4421
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 1 9 e4 4 2 7
Table 1 e Effluent DOC Concentrations during Phase I and Phase II. Reactor 3
Particle Size (m 10 ) 3
3
1
2
3
4
0.600
0.353
0.353
0.600
Phase I
Flow rate (m /day 10 ) Empty Bed Contact Time (days) Reynolds Number (103) Peclet Number Mean Effluent Concentration (mg/L)
0.500 8.340 0.762 0.762 1.980
0.500 8.340 0.448 0.448 1.740
4.000 1.040 3.584 3.584 2.060
4.000 1.040 6.091 6.091 2.070
Phase II
Flow rate (m3/day 103) Empty Bed Contact Time (days) Reynolds Number (103) Peclet Number Mean Effluent Concentration (mg/L)
2.000 2.080 3.046 3.046 1.950
2.000 2.080 1.792 1.792 1.860
1.000 4.170 0.896 0.896 1.940
1.000 4.170 1.523 1.523 1.930
Student’s t-test
p-value (based on particle size) p-value (based on phase flow rate)
0.183 0.320
0.500 0.172
0.500 0.172
0.183 0.320
2.3.
Batch kinetic tests
A modified biodegradable dissolved organic carbon (BDOC) reactor method was used to measure the kinetics of dextran biodegradation using native sand originally acclimated to tertiary effluent. The native sand used was from the Agua Fria River Basin (passed through a 2 mm screen) at a planned recharge site in Arizona, USA. In addition batch tests were completed with the clean silica sands used in the four reactors. The BDOC sand reactors were 500 ml Erlenmeyer Flasks containing 100 g of biologically active sand in each of the reactors. The biologically active sand was seeded initially with the same Mesa Tertiary Effluent used to seed the reactors. The BDOC reactors were seeded with 300 ml aliquots of effluent for three sequential batch tests where each test was 5 days. The reactors were then acclimated with 300 ml of dextran at a target concentration of 5 mg-C/L. After each 5 day reaction period, the solution was decanted and the acclimated sand in the reactors was washed with 100 ml of solution containing 0.15 M Sodium Chloride with 1 mM Magnesium Chloride. The washing procedure was repeated 3 times before new feed was added to initiate a new batch experiment. The reactors were then incubated with 300 ml of dextran at a nominal concentration of 9.66 mg-C/L to simulate the synthetic feed used in the column studies. The initial concentration in the batch tests was 40% higher than the concentration used in the column studies to provide kinetic information over a larger concentration range. Initial samples were taken to measure the concentration of DOC at the beginning of each experiment after mixing with the sand and biomass. The DOC and UVA254 were monitored every day for a period of 5 days consistent with other BDOC tests (Cha et al., 2004). Also, 5 g of sand was removed daily from the batch reactors and the sand was analyzed for biomass using organic nitrogen and carbohydrate analysis. The results were normalized to the quantity of sand and biomass after each sampling event to account for the removal of the sand and biomass. The data obtained from DOC measurements were used for analyzing the kinetics of dextran utilization for each reactor. The data from organic nitrogen analysis was then used to calculate a yield coefficient. By normalizing the substrate utilization rate data to the biomass content, a specific substrate utilization ratio was
calculated. This value was then compared to values determined for the reactors.
2.4.
Biomass characrerization in the columns
Immediately after Phase II was completed, the sand from the column reactors was extruded and divided into 11 to 12 sections to provide a profile of the sand as a function of reactor length. Each section was approximately 0.076 m in thickness and the sand was analyzed for biomass composition by determining the organic nitrogen, carbohydrate and volatile suspended solid attached to the sand.
2.5.
Analytical methods
2.5.1.
UVA254
The ultraviolet absorbance (UVA) at 254 nm was routinely measured using a Model 8452 A Hewlett Packard Diode Array Spectrophotometer. A 1 cm pathlength was used with a quartz cuvette.
2.5.2.
Dissolved organic carbon
The dissolved organic carbon (DOC) was routinely monitored using a Shimadzu TOC 5000 A Total Organic Carbon Analyzer in accordance with Method 5310 in Standard Methods for Examination of Water and Wastewater (Andrew et al., 2007). The minimum detection level was 0.5 mg/L.
2.5.3.
Biomass extraction and Quantification
The extraction of attached biomass from column reactor media or BDOC reactor media was done using the following procedure. Approximately 5 g of wet media was weighed and transferred into a 10 ml volumetric flask and 3 ml of 20% (w/v) of tricholoroacetic acid (TCA) was added. The flasks were placed in a VWR Scientific Aqua Sonic Model 150 T ultrasonic cleaner and sonicated for 10 min. After sonication, the solution was decanted from the flask and transferred to a second 10 ml volumetric flask. The sand remaining in the flask was rinsed once with 10 ml of water and the rinse water was transferred to the second volumetric flask. The sonication and transfer steps were repeated two more times to complete the separation of attached biomass. The extracts were analyzed
4422
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 1 9 e4 4 2 7
for organic nitrogen and carbohydrate to characterize the biomass. The sand used in each analysis was then dried and weighed.
2.5.4.
Organic nitrogen analysis
The extracted biomass was quantified for organic nitrogen using a Hach Total Nitrogen kit with the Persulfate Digestion Method in accordance with Method 4500 in Standard Methods for Examination of Water and Wastewater (Andrew et al., 2007) and a Hach DR/4000 Spectrophotometer.
2.5.5.
Carbohydrate analysis
The extracted biomass was also quantified for carbohydrate content. A 1 ml sample was mixed with 1 ml of 5% (w/v) of aqueous phenol solution. Then 5 ml of concentrated sulfuric acid was added (Herbert et al., 1971). A calibration standard with 0e100 mg glucose was used. The carbohydrate was measured at 488 nm using a Model 8452 A Hewlett Packard Diode Array Spectrophotometer. A 1 cm pathlength was used with a quartz cuvette.
2.5.6.
Indirect dextran measurements
Dextran was measured indirectly using both DOC and carbohydrate analyses. The carbohydrate analysis measured both simple and complex carbohydrates and can provided a direct measurement of dextran before biodegradation. There was less than þ3% difference between dextran measurements using carbohydrate analysis as compared to dextran measurements using DOC analysis. The correlation between DOC and carbohydrates was found to be valid throughout the studies even though the carbohydrate analysis will measure sugars in biomass such as riboses. This correlation also demonstrated that measuring carbohydrate on biomass extracted from the sand media was primarily due to biomass and the influence of dextran was negligible.
2.5.7.
Data analysis
And the specific rate of substrate utilization is q ¼ qmax
S Ks þ S
(4)
Where qmax is the maximum specific substrate utilization rate (g-C/g-biomass-t) Assuming hydrolysis is the rate limiting step, Ks is very small since there are not many extracellular enzymes and the concentration of extracellular enzymes is constant (Guellil et al., 2001). The hydrolysis products of dextran are sugars that should be rapidly degraded at a rate faster than the hydrolysis rate. The specific growth rate (Eq. (1)) and specific rate of substrate utilization (Eq. (4)) becomes (5) m ¼ mmax q ¼ qmax ¼ k
(6)
where k ¼ zero-order substrate utilization rate constant (g-C/ g-biomass-day) Considering a fixed-bed and assumingplug flow, steady state conditions with zero-order kinetics with respect to substrate utilization and first order kinetics with respect to biomass growth, we get using the following definitions in the model: L ¼ Length of reactor (m), U velocity in direction of flow (m/day), S ¼ Substrate concentration (g-C/m3), X Biomass concentration (g-biomass/m3), Y Yield Coefficient (g-C/ g-biomass), k zero-order substrate utilization rate constant (g-C/g-biomass-day), b decay coefficient (day1), z axial direction (m), Xf Biofilm density (g/m3), Lf Biofilm thickness (m), q ¼ residence time (day), a specific surface area (m2/m3). Where the biomass concentration may be expressed in terms of the biofilm density and thickness as X ¼ Xf Lf a
(7)
Biofilm kinetics may be analyzed using a classic biofilm model (Fig. 1) where external mass transfer resistance and diffusion through the biofilm are considered. However, if
2.5.7.1. Monod substrate utilization/biomass growth kinetic model. A model was developed for substrate utilization and
Y
biomass growth in the column reactors studied. Monod in 1942 gave an empirical model (Eq. (1)) for microbial growth kinetics introducing the concept of a growth limiting substrate.
dX dS
YX=S dS ¼ YX=S q m¼ X dt
Lf
Bulk Liquid
Diffusion Layer
(1)
where m ¼ specific growth rate (day1), mmax ¼ maximum specific growth rate (day1), S ¼ substrate concentration (g-C/m3), Ks ¼ substrate saturation constant i.e., substrate concentration at half mmax (g-C/m3) In addition to this, Monod also related the Yield Coefficient Yx/s (Eq. (2)) to the specific biomass growth rate and the specific rate of substrate utilization q (Eq. (3)) YX=S ¼
Z
Biofilm
S Ks þ S
Xf
Sand Particle
m ¼ mmax
X
S
L
(2)
(3)
Fig. 1 e Biofilm Model e Flow is In the Z-Direction while Diffusion is in the X-Direction. As Dextran Approaches the Biofilm Extracellular Enzymes Can Hydrolyze the Dextran.
4423
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 1 9 e4 4 2 7
hydrolysis by extracellular enzymes is the rate limiting step, diffusion into the biofilm is only important for the products of hydrolysis. For dextran, the hydrolysis products are rapidly degraded and diffusion into the biofilm was assumed to not be rate limiting. A Steady State Substrate Mass Balance may be written assuming external mass transfer to the biofilm is negligible.
0¼V
dS ¼ dt
! dS U kXf Lf a V dz
DS=Dt ¼ ks whereDS ¼ mass substrate utilized over the time Dt
(13)
The yield of biomass per mass of substrate utilized substrate (YX/S) is obtained using equation 14 where DXX is the biomass growth over the time Dt. YX=S ¼ DXX=DS
(8)
Steady State Biomass Mass Balance 0 ¼ aV
The rate of substrate (S) utilization is obtained using Eq. (13).
dXf Lf ¼ ðYk bÞXf Lf aV dt
(9)
(14)
where YX/S yield of biomass on substrate (mg-X/mg-C) The specific substrate utilization rate (kbatch) for the batch experiment is obtained using Eq. (15). kbatch ¼ ðks =XXÞ ð300ml=1000ml=LÞ
(15)
Recognizing that the substrate utilization rate-k Xf is the rate of substrate utilization, then Eq. (9) reduces to Eq. (10) where J is the flux into the biofilm in gC/m2ed.
where kbatch specific substrate utilization rate (mg-C/mgbiomass-day) The specific growth rate is then obtained according to Eq. (16).
JY ¼ bXf Lf
m ¼ YX=S kbatch
(10)
which is the classic result for a steady state biofilm where growth equals decay. Solving Eq. (8) for X where X XfLfa yields X¼
ðS0 SÞU Lk
(11)
Since hydrolysis limits substrate utilization to zero-order kinetics, the model predicts biomass content becomes independent of position.
2.6.
(16) 1
where m specific growth rate (day ) For each particle size of soil, the kinetic parameters are tabulated in Table 2. The specific substrate utilization rate constants for column experiments (kcolumn) are then obtained according to Eq. 17 and listed in Table 3. kcolumn ¼ kbatch particle surface area 5=residence time (17) The predicted biomass from the model is then obtained using Eq. (18) and is listed in Table 3.
Batch experiments data analysis
Predicted Biomass ¼ X V Batch experiments were analyzed for substrate utilization and microbial growth kinetics. Analysis was done to determine the zero-order substrate utilization rate (ks). The specific substrate utilization rate (kbatch) was determined by normalizing to biomass attached to the sand which was measured in terms of organic nitrogen per gram of sand. The quantity of attached biomass (XX) in mg was calculated using organic nitrogen data assuming an empirical cell composition of C5H7O2N according to Eq. (12). In Eq. (12), Org-N is the measured organic nitrogen in mg-N/g-sand. XX ¼ ðOrg N=g sandÞ 100 g sand 113 mg C5 H7 O2 N=14 mgN
(12)
where XX ¼ amount of biomass formed (mg-X)
(18)
where V volume of reactor (L), X concentration of biomass according to Eq. (11) (mg-X/m3)
3.
Results and discussion
3.1.
Column/Reactor experiments with dextran
The influent and effluent DOC concentrations were monitored as a function of time for the four reactors and the average concentrations were used to evaluate the reactors for the two phases of operation. Effluent DOC was used as a surrogate to understand the removal of dextran in the soil columns since these values correlated with carbohydrate analysis as
Table 2 e Kinetic parameters from Batch Kinetic Experiments. Reactor
1 2 3 4
Particle Size (m 103)
Substrate rate constant ks (mg-C/L/day)
Final Biomass-N per gram of sand DN (mg-N/g-sand)
Amount of Biomass formed X (mg-X)
Amount of Substrate utilized S (mg-C)
Yield of Biomass on Substrate YX/S (mg-X/mg-C)
Specific substrate utilization rate kbatch (mg-C/mg-X/day)
Specific Growth Rate m (day1)
0.600 0.353 0.353 0.600
0.4426 0.8856 0.8856 0.4426
0.023 0.041 0.041 0.023
18.564 33.093 33.093 18.564
0.664 1.328 1.328 0.664
27.958 24.920 24.920 27.958
0.0072 0.0080 0.0080 0.0072
0.2 0.2 0.2 0.2
4424
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 1 9 e4 4 2 7
Table 3 e Biomass according to substrate utilization/microbial growth model and experimentation Reactor
1 2 3 4
Experimental Experimental Predicted Predicted Particle Flux Flow rate Particle Biofilm Size (m 103) (m3/day 103) surface area (mg-C/m2/day) Thickness (mm) (m2 106) 0.600 0.353 0.353 0.600
0.500 0.500 4.000 4.000
1.133 0.392 0.392 1.133
1.550 0.910 7.290 12.390
mentioned above. Phase I was run for a period of 150 days such that the DOC concentration became constant with time reflecting the development of an acclimated population and pseudo-steady state conditions. The reactors were operated under saturated and aerobic conditions. The feed dissolved oxygen (DO) was maintained near equilibrium with atmospheric oxygen with an approximate concentration of 6 mg/L. Aerobic conditions in the reactors were maintained as 0.7 mg of oxygen was needed per mg of dextran according to stoichiometry (Eq. (19)) and the average feed dextran concentration was 6.7 mg/L resulting in an effluent dissolved oxygen concentration greater than 2 mg/L. þ 0:042C6 H10 O5 þ 0:02HCO 3 þ 0:02NH4 þ 0:15O2 /0:02C5 H7 O2 N
þ 0:25CO2 þ 0:188H2 O (19) Routine monitoring of DO, Turbidity, UV254 and DOC was performed on the reactors. The mean feed concentration which includes the background DOC of tap water was 6.7 þ 0.5 mg DOC/L and the average effluent concentrations ranged from 1.70 to 2.07 mg DOC/L for Phase 1 (Table 1). During Phase 2, the effluent concentrations in all four reactors were very similar and ranged only from 1.86 to 1.95 mg DOC/L (Table 1). Also, listed in Table 1 are the Reynolds Number and Peclet Number corresponding to the mean effluent concentrations during Phase 1 and Phase 2. A Student’s t-test was performed on the mean effluent concentrations in the reactors based on particle size and flow rate for both the Phases
Fig. 2 e BDOC Kinetic Experiment with different sands.
0.004 0.001 0.010 0.028
Column Predicted Experiment Rate Biomass Biomass Constant (Eq (18)) (mg-X) kcolumn (mg-X) (mg-C/mg-X/L/day) 0.0176 0.0068 0.0541 0.1406
508.48 1363.38 1287.20 501.12
393.110 1185.680 1196.140 498.060
and was found to have no statistical difference between the various mean effluent concentrations.
3.1.1.
Batch kinetic experiments with dextran as substrate
Batch experiments with the modified BDOC reactors were completed using Agua Fria sand and two clean silica sands sieved to the same geometric diameter as used in the column reactors which was 0.353 mm and 0.6 mm. Prior to running the kinetic experiments the BDOC reactors were acclimated with Mesa Tertiary Effluent as described above. Acclimatization was determined complete when the final DOC concentration after each test was the same as the previous test. After acclimation, the kinetic experiments were carried out with a nominal initial dextran concentration 40% higher than the influent to the reactors. The results are summarized in Fig. 2. Fig. 2 shows that the substrate concentration decreases linearly as a function of time which is consistent with zeroorder kinetics for the different types of sand studied. i.e., Agua Fria sand, 0.353 mm silica sand and 0.6 mm silica sand. The UVA254 in the effluent increased from zero to 0.4 cm1 for 0.353 mm silica sand and from zero to 0.18 cm1 for 0.6 mm silica sand during the batch tests. The increase in UVA254 is consistent with the presence of aromatic compounds as the microorganisms were producing or desorbing soluble microbial products. The larger increase with the 0.353 mm silica sand is consistent with the larger surface area and biomass content. The increase in UVA254 was not enough to significantly impact the DOC concentration. The organic nitrogen data used to evaluate the growth of biomass
Fig. 3 e Organic Nitrogen data during the kinetic experiment with different sands.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 1 9 e4 4 2 7
4425
Biomass (C5H7O2N) (mg-X/g-sand)
Reactor Biomass 0.40 Reactor Reactor Reactor Reactor
0.35 0.30
3 4 2 1
0.25 0.20 0.15 0.10 0.05 0.00
0
2
4 6 8 Reactor Section (Bottom-Top)
10
12
Fig. 4 e Reactor Biomass versus Reactor Sections.
on the sand is shown in Fig. 3. The gradual increase with reaction time for the different types of sand is consistent with microbial growth. From the kinetic experiments with the BDOC reactors, the rate of DOC biodegradation kinetics was observed to be zero-order with respect to substrate concentration. The rate for the smaller particle size was found to be approximately twice that of the larger particle size. The smaller particle size had approximately three times the surface area of the larger particle size. Assuming the microbial population was primarily attached, the surface area appeared to directly influence the removal rate. The BDOC reactor experiments support a zero-order relationship for substrate utilization which might be expected when hydrolysis of high molecular weight compounds is the rate limiting step. The specific substrate utilization rate, yield of biomass on substrate and the specific growth rate were calculated according to the Batch Experiments Data Analysis and are listed in Table 2. The rate constants for the column experiments were then obtained in accordance with Eq. (17).
3.1.2.
Biomass data from columns
Once, the column experiments were completed the sand in the reactors were cut into 11 or 12 sections and were analyzed for biomass using Org-N, carbohydrate and volatile suspended solids. Fig. 4 shows the biomass profiles for the 4 reactors based on the Org-N measurements. The Organic-N profiles,
Fig. 5 e Activity Kinetics for Reactor 1.
Fig. 6 e Activity Kinetics for Reactor 2.
which represent a measure of the protein content of the cells was relatively constant along the length of the reactors. The organic-N could also represent extracellular polymers, nucleic acids and the organic-N does not distinguish between active cells and cellular debris. We see from Fig. 4 that the biomass growth was similar for reactors 1 and 4 and for reactors 2 and 3. This indicates that the biomass growth was independent of the applied substrate flux and was primarily affected by the surface area and hence the particle size. Table 3 lists the quantity of biomass according to the biomass growth model (Eq. (18)) and from the experimental investigation. We see from Table 3 that the biomass is proportional to the surface area in the reactors. The biomass increases by a factor of 3 for a 3-fold increase in individual particle surface area when comparing the reactors with 0.6 mm sand to reactors with 0.353 mm sand. Table 3 also shows that the biofilm thickness is less than one micron. Therefore, diffusion in the biofilm can be assumed and this verifies the assumption made in model development. A fully-penetrated biofilm with the traditional Monod model for substrate utilization/biomass growth agrees with experimental results.
3.1.3.
Activity kinetics
The sand samples extruded as sections from the reactors were analyzed for microbial activity using the modified BDOC
Fig. 7 e Activity Kinetics for Reactor 3.
4426
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 1 9 e4 4 2 7
Fig. 8 e Activity Kinetics for Reactor 4.
kinetic test. Sand extruded from every 4th section of each of the reactors was tested to determine the removal of dextran over a period of 5 days. The activity test data for sand samples taken from reactors 1 through 4 is presented in Figs. 5e8, respectively. The kinetics were found to be zero-order with respect to substrate removal for samples from all the reactors. For each section of the reactors, the rate was independent of the distance from the influent to the effluent. Hence, the overall rate for a reactor could be calculated based on the results for the average of sections from a reactor. The individual rate for reactor 1 was found to be equal to 0.1734 mg-C/L-day, for reactor 2 it was 0.2849 mg-C/L-day, for reactor 3 it was 0.3989 mg-C/L-day and for reactor 4 it was 0.3257 mg-C/L-day. The overall rates for the column reactors were calculated using the ratio of total mass of sand in the columns divided by the mass used for batch testing. These values were found to be equal to 10.404 mg-C/L-day for reactor 1, 17.094 mg-C/L/day for reactor 2, 23.934 mg-C/L/day for reactor 3 and 19.542 mg-C/ L/day for reactor 4. Reactors 1 and 2 always had the same hydraulic retention time and Reactors 3 and 4 always had the same hydraulic retention time. We see that the reactors with smaller particle size had higher rates than that of reactors with larger particle diameter for the same flow rates (i.e. Reactor 2 Reactor 1 and Reactor 3 Reactor 4). The smaller difference between Reactors 3 and 4 could be from an accumulation of biomass during Phase 1 when the loading rate was 8 times the loading rate in Phase II. Similar data was observed when the BDOC reactors were used to evaluate kinetics using a single dextran feed concentration. This is interesting since the BDOC reactors were acclimated in a batch system while the sand from the reactors was acclimated in a column, yet the effects of surface area were similar.
4.
Summary
Dextran (average MW 10,000 Da) was observed to biodegrade with zero-order substrate utilization kinetics during batch kinetic experiments. This is consistent with the expected kinetics when hydrolysis is the rate limiting step. During continuous flow column experiments, the biomass distribution did not vary significantly based on activity measurements
and organic nitrogen analyses. Four column experiments were completed with two different particle sizes (hence surface area contact) and 4 different flow rates. The removal of substrate was independent of flow rate and particle size. However, the surface area had a positive relationship with biomass accumulation. The ratio of organic-N (hence biomass) was a factor of 3 higher for 0.353 mm particle size when compared with 0.6 mm particle size in one set of paired columns (Columns 1 and 2) and a factor of 2.4 higher for 0.353 mm particle size when compared with 0.6 mm particle size in the second set of paired columns (Columns 3 and 4). The surface area for the columns with a particle size of 0.353 mm was 3 times greater than the columns with 0.6 mm particle size. The experimental results demonstrated that biological removal of a biodegradable high molecular weight compound was robust during flow over a porous media. The Monod based substrate/growth kinetic model does predict the removal of a single type of biodegradable high molecular weight and the biofilm thickness was insufficient to cause diffusion limitations. The rate limiting step appears to be hydrolysis. Hydrolysis may be the rate limiting step for mixtures of high molecular weight compounds that are present in actual systems. The mixtures contain compounds with different biodegradabilities and their hydrolysis products might not be easily biodegradable resulting in higher order kinetics.
references
Amy, G., Wilson, L.G., Conroy, A., Chahbandour, J., Ahai, W., Siddiqui, M., 1993. Fate of chlorination byproducts and nitrogen species during effluent recharge and soil aquifer treatment. Water and Environmental Research 65, 726e734. Amy, G.L., Drewes, J., Westerhoff, P., 2006. Organic matter in soilaquifer treatment systems. Journal of Environmental Engineering 132 (11), 1447e1458. Andrew, D.E., Rice, E.W., Baird, R.B., 2007. In: Standard Methods for the Examination of Water and Wastewater, first ed. American Public Health Association. Bouwer, H., 1985. Renovation of Wastewater with Rapidinfiltration Land Treatment Systems. In: Asano, T. (Ed.), Artificial Recharge of Groundwater. Butterworth, Boston, pp. 249e282. Bouwer, H., Rice, R.C., 1984. Renovation of wastewater at the 23rd avenue rapid infiltration project. Journal of Water Pollution Control Federation 36, 76e83. Cadoret, Aurore, Conrad, Arnaud, Block, Jean-Claude, 2002. Availability of low and high molecular weight substrates to extracellular enzymes in whole and dispersed activated sludges. Enzyme and Microbial Technology 31, 179e186. Cha, W., Choi, H.C., Fox, P., 2004. Abiotic and biotic removal mechanisms for organic carbon during soil aquifer treatment. Water Environment Research 76, 756e804. Chrost, R.J., 1991. Environmental control of the synthesis and activity of aquatic microbial ecto-enzymes. In: Chrost, R.J. (Ed.), Microbial Enzymes in Aquatic Environments. Springer, New York, pp. 29e59. Drewes, J.E., Jekel, M., 1998. Behavior of DOC and AOX using advanced treated wastewater for groundwater recharge. Water Research 32, 3125e3133. Eliosov, B., Argaman, Y., 1995. Hydrolysis of particulate organics in activated sludge systems. Water Research 29, 155e163.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 1 9 e4 4 2 7
Fox, P., Narayanaswamy, K., Genz, A., Drewes, J.E., 2001. Water quality transformations during soil aquifer treatment at the nesa Northwest water reclamation plant USA. Water Science and Technology 43 (10), 343e350. Guellil, A., Boualam, M., Quiquampoix, H., Ginsestet, P., Audic, J.M., Block, J.C., 2001. Hydrolysis of wastewater colloidal organic matter by extracellular enzymes extracted from activated sludge flocs. Water Science and Technology 43 (6), 33e40. Herbert, D., Phipps, P.J., Strange, R.E., 1971. Carbohydrate analysis. vol. 5B. In: Norris, J.R., Ribbons, D.W. (Eds.), Methods in Microbiology. Academic Press, New York, pp. 265e301. Levine, A.D., Tchobanoglous, G., Asano, T., 1985. Characterization of the size distribution of contaminants in wastewater: treatment and reuse implications. Journal of Water Pollution Control Federation 57, 805e816. Nalinakumari, B., Cha, W., Fox, P., 2010. Effects of primary substrate concentration on N-nitrosodimethylamine (NDMA) during simulated aquifer recharge. ASCE Journal of Environmental Engineering 136 (4), 373e380.
4427
Raunkjaer K, Hvitved-Jacobsen T., Nielsen P.H. 1994. Measurement of Pools of Protein, Carbohydrate and Lipid in Domestic Wastewater. 28(2), 251e262. Vetter, Y.A., Deming, J.W., 1999. Growth rates of marine bacterial isolates on particulate organic substrates solubilized by freely released extracellular enzymes. Microbial Ecology 37, 86e94. Wetzel, R.G., 1991. Extracellular enzymatic interactions: storage, redistribution and inter specific communication. In: Chrost, R.J. (Ed.), Microbial Enzymes in Aquatic Environments. Springer, New York, pp. 6e28. Wild, D., Reinhard, M., 1999. Biodegradation residual of 4-Octylphenoxyacetic acid in laboratory columns under groundwater recharge conditions. Environ. Sci. Technol. 33 (24), 4422e4426. Wilson, L.G., Amy, G.L., Gerba, C.P., Gordon, H., Johnson, B., Miller, J., 1995. Water quality changes during soil aquifer treatment of tertiary effluent. Water and Environmental Research 67, 371e376.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 2 8 e4 4 3 6
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Accuracy and precision of Legionella isolation by US laboratories in the ELITE program pilot study Claressa E. Lucas*, Thomas H. Taylor Jr., Barry S. Fields Division of Bacterial Diseases, Centers for Disease Control and Prevention, 1600 Clifton Rd NE MS G03, Atlanta, GA 30333, USA
article info
abstract
Article history:
A pilot study for the Environmental Legionella Isolation Techniques Evaluation (ELITE)
Received 9 December 2010
Program, a proficiency testing scheme for US laboratories that culture Legionella from
Received in revised form
environmental samples, was conducted September 1, 2008 through March 31, 2009.
28 March 2011
Participants (n ¼ 20) processed panels consisting of six sample types: pure and mixed
Accepted 26 May 2011
positive, pure and mixed negative, pure and mixed variable. The majority (93%) of all
Available online 7 June 2011
samples (n ¼ 286) were correctly characterized, with 88.5% of samples positive for Legionella and 100% of negative samples identified correctly. Variable samples were incorrectly
Keywords:
identified as negative in 36.9% of reports. For all samples reported positive (n ¼ 128),
Proficiency testing
participants underestimated the cfu/ml by a mean of 1.25 logs with standard deviation of
Environmental sampling
0.78 logs, standard error of 0.07 logs, and a range of 3.57 logs compared to the CDC re-test
Bacteria enumeration
value. Centering results around the interlaboratory mean yielded a standard deviation of
Legionella monitoring
0.65 logs, standard error of 0.06 logs, and a range of 3.22 logs. Sampling protocol, treatment regimen, culture procedure, and laboratory experience did not significantly affect the accuracy or precision of reported concentrations. Qualitative and quantitative results from the ELITE pilot study were similar to reports from a corresponding proficiency testing scheme available in the European Union, indicating these results are probably valid for most environmental laboratories worldwide. The large enumeration error observed suggests that the need for remediation of a water system should not be determined solely by the concentration of Legionella observed in a sample since that value is likely to underestimate the true level of contamination. Published by Elsevier Ltd.
1.
Introduction
Legionellaceae are ubiquitous in moist environments and a frequent contaminant of building water systems (Fields et al., 2002). Inhalation of aerosolized water contaminated with legionellae by susceptible individuals results in legionellosis that may present as the acute pneumonia, Legionnaires’ disease, or the less severe Pontiac Fever. Legionella pneumophila is the most common etiological agent, however all species of legionellae are presumed to be capable of causing disease (Alli et al., 2003; Palusinska-Szysz and Cendrowska-Pinkosz, 2009). * Corresponding author. Tel.: þ1 404 639 3564; fax: þ1 866 638 0199. E-mail address:
[email protected] (C.E. Lucas). 0043-1354/$ e see front matter Published by Elsevier Ltd. doi:10.1016/j.watres.2011.05.030
Legionellosis cannot be spread person-to-person and is only acquired from environmental sources. The widespread presence of legionellae precludes their removal from the environment so that disease prevention methods instead focus on reducing transmission of bacteria to susceptible hosts (Fields et al., 2002; Sehulster and Chinn, 2003; Freije, 2004; Tablan et al., 2004; Fields and Moore, 2006). Monitoring levels of legionellae in building water systems by routine environmental sampling has been employed by some as a means of controlling transmission, though the relationship between the presence of legionellae and
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 2 8 e4 4 3 6
incidence of disease remains unclear (O’Neill and Humphreys, 2005; Den Boer et al., 2007). Environmental sampling for Legionella spp. as an approach for primary control can provide useful information to institutions that house persons at high risk for disease, such as chronic care and transplant facilities (Anonymous, 1997; Butler et al., 1997; Yu, 1997; Fiore et al., 1999; Fields and Moore, 2006). However, routine culture in the absence of documented cases of legionellosis is an area of considerable controversy, mostly because there is no generally accepted protocol for the choice of sampling sites, frequency of sampling events, or endpoints for remediation (Den Boer et al., 2007; Stout et al., 2007; Ditommaso et al., 2010). Recommendations that suggest the use of routine sampling for primary control acknowledge that the acceptable limits of contamination, either measured in cfu/ml or as a percentage of positive sites sampled, are based on limited data (Force, 1997; Sehulster and Chinn, 2003; Stout et al., 2007). Another confounding factor that is not frequently mentioned in the discussion on routine sampling is the considerable variability in recovery of legionellae from repeated sampling of sites or even seeded tap water suggesting that interlaboratory enumeration error may be high (Boulanger and Edelstein, 1995; Bentham, 2000; Napoli et al., 2009). The Centers for Disease Control and Prevention (CDC) guidelines for the reduction of legionellosis advise that there is no acceptable level of legionellae contamination and that if Legionella spp. are detected a plan to prevent transmission to susceptible individuals should be employed (1997). Domestic and international organizations have published procedures for the recovery of Legionella bacteria from environmental samples but the application of these techniques can still require considerable expertise from the laboratorian (Anonymous, 1998, 2000, 2004, 2005a,b, 2008). Numerous variables contribute to effective recovery including properly taking and transporting samples, the application of sample pre-treatments intended to increase the representation of viable legionellae, and the use of suitable selective media (Reeves et al., 1981; Tesh and Miller, 1981; Shahamat et al., 1991; Lee et al., 1993; Ta et al., 1995; Leoni and Legnani, 2001; Wiedenmann et al., 2001; Bartie et al., 2003; Luck et al., 2004). Characterization of isolates entails the ability to distinguish Legionella colony morphology from autochthonous microbiota and serological methods for confirmation of identity, both of which determinations are subjective and sometimes difficult to interpret without considerable prior experience by the laboratorian (Benson and Fields, 1998; Helbig et al., 2007; Wagner et al., 2007). PCR-based methods of isolate identification offer greater sensitivity and specificity but are often cost prohibitive for laboratories with a low volume of tests for legionellae and do not yet provide enough discrimination between strains for epidemiological surveillance (Thurmer et al., 2009; Tronel and Hartemann, 2009). In the European Union (EU) laboratories that culture environmental samples for Legionella spp. are required to participate in a proficiency testing (PT) scheme to ensure baseline quality standards throughout the industry (2005). The EU scheme, in operation since June 2004, requires participants to process PT products as would be performed for potable water and report results to the program administrators for scoring
4429
according to ISO standards (1997; 1997). The EU scheme has been successful in assessing the status of the industry and publishes quarterly reports of their trends and performance distributions over time. A new PT scheme, the Environmental Legionella Isolation Techniques Evaluation (ELITE) program was created by CDC to capture similar information for US laboratories. A Pilot Program was performed September 1, 2008eMarch 31, 2009 to determine industry limits of detection and ascertain the accuracy of legionellae enumeration by participating laboratories. The results indicate that US laboratories are generally capable of a qualitative assessment of environmental samples for the presence of legionellae but that quantitation displays significant inter- and intralaboratory variability.
2.
Methods and materials
2.1. Sample creation, panel composition, quality control, and re-tests Legionellae and heterotrophic bacteria were grown from freezer stocks of less than three passages on BCYE media (BBL agar base plus 10 g/L L-cysteine) for 3e5 days at 35 C with 2.5% (v/v) CO2. Samples distributed for PT were removed from plates by suspension in sterile de-ionized water. Bacterial suspensions were diluted in 10% (v/v) AYE broth prior to lyophilization. The formula for full strength AYE broth is as follows: 5 g bovine serum albumin fraction V, 10 g ACES, 10 g yeast extract, 0.4 g L-cysteine, 0.25 g iron pyrophosphate. Bring volume to 1 L with sterile de-ionized water, adjust pH to 6.9 with 1 N potassium hydroxide, and filter to sterilize. The AYE broth helped to stabilize lyophilized cells during storage and formed a consistent pellet regardless of sample composition. Samples were made in batches monthly with 20e25 aliquots per lot. A representative sample from each lot was examined for quality control (QC) one week after lyophilization by plating serial dilutions from the vial reconstituted in 1 ml sterile de-ionized water on BCYE in triplicate at each serial dilution. Plates were incubated 5 days at 35 C with 2.5% (v/v) CO2 prior to being read. The total bacterial count and Legionella-specific count per plate were determined. At least one representative Legionella-like colony per plate was confirmed as a cysteine auxotroph. PT panels consisted of 6 samples, one of each sample type: pure positive, mixed positive, pure negative, mixed negative, pure variable, and mixed variable (see Table 1). Positive samples were either Legionella in pure culture or mixed with a low ratio of heterotrophs. Negative samples contained no viable legionellae, though they might contain heat killed legionellae. Variable samples consisted of either low levels of Legionella in pure culture or a mixture of organisms with a high ratio of heterotrophs to legionellae. Laboratories were scored on a pass/fail basis, required to correctly identify both positive and both negative samples for a passing score. Variable samples did not count toward the score but were included in each panel to assess lower limits of detection. Panels identical to those sent to participants were shipped back to the program administrators for re-tests. Re-test samples were first reconstituted in the vial with 1 ml of sterile,
Sample ID no
A21-08052203 A48-08071517 A65-08091607 A67-08091619 A62-08091601 A20-08052201 A28-08052204 A6508091615 A48-08071518 A6708091617 A62-08091609 A50-08081904 A71-08111906 A38-08061703 A25-08052218 A84-08120201 A88-08120215 A50-08081908 A28-08052207 A89-08120213 A14-08041609 A39-08061716 A86-08120217 A31-08061717 A89-08120211 A62-08091607 A32-08061706
Negative e media Variable e mixed Variable e low Positive e pure Negative e mixed Positive e mixed Negative e media Negative e mixed Variable e mixed Positive e mixed Positive e pure Variable e pure Negative e pure nonLegionella Negative e mixed Variable e mixed Variable e pure Positive e mixed Positive e pure Negative e media Negative e mixed Variable e pure Variable e mixed Positive e mixed Positive e pure Negative e media Negative e mixed Variable e pure Positive e mixed Positive e pure Variable e mixed Negative e media Negative e mixed Variable e pure Positive e mixed Variable e mixed Positive e pure Negative e mixed Variable e mixed Variable e pure Positive e mixed
Panel
Sample composition
QC total cfu/ml SD
QC Legionella cfu/ml SD
Re-testa total cfu/ml SD
Re-test Legionella cfu/ml
Re-test protocol
I I I I I, A I, A II II II II II II, A III
Blank media L. pneumophila Sg1 þ HTs L. pneumophila Sg3 L. pneumophila Sg8 HTs L. bozemani þ HTs Blank media HTs L. pneumophila Sg1 þ HTs L. pneumophila Sg3 þ HTs L. pneumophila Sg8 L. cherrii HTs
0 5 0.3 Pure Pure 1370 374 4400 712 0 703 9 41 583 26 Pure Pure 3330 471
0 0.6 0.4 53 13 211 14 0 170 11 0 0 0.7 4 48 4 254 17 74 0
0 7 0.3 Pure Pure 861 5 3000 0 300 41 300 Pure Pure ND
0 1 52 139 0 170 0 0 1 35 344 2 0
D, BCYE D, GPCV D, BCYE SD, BCYE SD, BCYE SD, PCV D, BCYE SD, BCYE D, GPCV SD, A, PCV SD, BCYE D, BCYE SD, BCYE
III III III III III, A IV IV IV IV IV IV V V V V V V, A VI VI VI VI VI VI VII VII VII VII
HTs L. pneumophila Sg1 þ HTs L. pneumophila Sg1 L. pneumophila Sg8 þ HTs L. pneumophila Blank media HTs L. pneumophila Sg8 L. pneumophila Sg1 þ HTs L. pneumophila Sg1 þ HTs L. pneumophila Sg8 Blank e broth only HTs L. rubrilucens L. dumoffii HTs L. feelei L. pneumophila Sg1 þ HTs Blank e broth only HTs L. pneumophila Sg10 L. rubrilucens þ HTs L. rubrilucens þ HTs L. dumoffii HTs L. rubrilucens þ HTs L. pneumophila Sg1 L. pneumophila Sg3 þ HTs
83,000 3740 7870 309 Pure 4730 624 Pure 0 15, 200 2380 Pure 883 58 18, 7000 1250 Pure 0 15,600 1130 Pure 39,000 5720 Pure 600 36 0 13,600 8740 Pure 16,200 9420 70,700 7040 Pure 7870 141 205 23 Pure 11,700 5350
0 150 39 593 170 2000 630 7600 1280 0 0 61 7 257 38 4000 1410 3300 804 0 0 10 0.4 1670 125 27,300 5910 33 5 0 0 183 24 1290 173 2400 57 74,200 17,000 0 35 4 59 17 1480 398
ND ND Pure ND Pure 0 ND Pure ND ND Pure ND ND Pure ND Pure ND 0 ND Pure ND ND Pure ND ND Pure ND
0 13 120 311 3000 0 0 16 193 412 3000 0 0 3 1000 16500 12 0 0 102 540 2000 34300 0 17 60 200
SD, BCYE A, SD, PCV SD, BCYE A, SD, GPCV SD, BCYE D, BCYE SD, BCYE D, BCYE A, SD, BCYE A, SD, PCV SD, BCYE D, BCYE SD, BCYE D, BCYE A, SD, GPCV SD, BCYE A, SD, BCYE D, BCYE SD, BCYE SD, BCYE A, SD, PCV A, SD, BCYE SD, BCYE SD, BCYE SD, BCYE D, BCYE A, SD, GPCV
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 2 8 e4 4 3 6
A50-08081901 A49-08071519 A32-08061701 A12-08041602 A31-08061718 A25-08052219 A50-08081905 A30-08061719 A49-08071520 A32-08061707 A12-08041604 A32-08061702 A31-08061704
Status
4430
Table 1 e PT sample characteristics: All PT samples used during the Pilot Program (Panel column: Panels number IeVIII and Accelerated) are listed here with sample composition, QC, and re-test results. Samples were categorized into sample types according to QC results. Re-test treatments specific to each sample are abbreviated: D [ direct plating, SD [ serial dilutions, A [ 15 min acid.
4431
VIII VIII VIII VIII VIII Negative e mixed Variable e mixed Variable e pure Positive e mixed Positive e pure A71-08111907 A89-08120212 A65-08091608 A32-08061708 A62-08091604
VIII Negative e media A90-09010509
a Heterotrophic plate counts were enumerated by re-test only for the first round (Panels I and II). Heterotrophic cfu/ml were not determined (ND) by re-test for panels III through VIII.
SD, BCYE A, GPCV SD, BCYE A, SD, BCYE SD, BCYE 0 5 45 200 835 7670 858 1130 17 Pure 4430 377 Pure
0 23 12 300 183 37 1330 557
ND ND Pure ND Pure
D, BCYE 0 0 0
0
1410 0 VII VII, A A62-08091603 A90-09010505
Positive e pure Negative e media
L. pneumophila Sg1 L. pneumophila Sg1 [heat killed] L. pneumophila Sg10 [heat killed] HTs L. rubrilucens þ HTs L. pneumophila Sg1 L. pneumophila Sg3 þ HTs L. pneumophila Sg1
Pure 0
3100 804 0
Pure 0
SD, BCYE D, BCYE
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 2 8 e4 4 3 6
de-ionized water then diluted 1:1000 in a bottle of sterile, deionized water to yield a 1 L Test Sample. All Test Samples were filtered. Each Test Sample was then processed according to an individual protocol determined by sample composition, which was designed for maximum recovery and enumeration of legionellae (Table 1). Three replicates of each dilution were plated but only the highest count was used as the re-test value for data analysis. All re-test values were within one standard deviation of the mean of the re-test replicate plate counts (data not shown). Test Samples were subjected to combinations of pre-treatment with acid (18 parts 0.2 M KCl to 1 part 0.2 M HCl), serial dilutions prior to plating, and/or plating on selective media PCV (BCYE plus 13.22 mg/L polymixin B, 80 mg/L cyclohexamide, and 5 mg/L vancomycin) or GPCV (PCV plus 2 g/L glycine). QC results, re-test results, and re-test treatment protocols are listed for all samples in Table 1.
2.2.
Pilot participants
Because current industry capabilities were unknown at the inception of the ELITE Program, a pilot study was conducted September 1, 2008 through March 31, 2009 to generate baseline values. The Pilot Program was limited to ten participating laboratories. A total of 20 laboratories enrolled in the program prior to the August 31, 2008 deadline for inclusion in the Pilot Program. Laboratories that could not be accommodated in the Pilot Program due to space limitations were enrolled as Accelerated Members. Commercial laboratories comprised the majority of enrollees but the Pilot Program was designed to include at least one of each type of laboratory as listed in Table 2. CDC reference laboratory personnel separate from those administering the ELITE Program participated in the pilot. To avoid confusion in the text, CDC reference laboratory Pilot Program participants are designated the “federal” laboratory in the text and figures, while administrators’ results are labeled “CDC”. For data analysis purposes, State, County, and Hospital laboratories were grouped into the category “Local Public Health.” Participants were asked to process the panels according to their standard in-house protocol and return results as they would be given to clients. Participants were directed to rehydrate the lyophilized pellets in 1 ml sterile water then dilute the suspension 1:1000 in sterile water to yield a test sample of the volume dictated by the participant’s in-house protocol. Supplement 1 contains the complete Sample Handling Instructions available to all participants. Pilot Members processed four panels shipped September 3, 2008, November 3,
Table 2 e Pilot and accelerated program demographics: the number and type of each participant in the pilot or accelerated program.
Commercial Federal State County Hospital Total
Pilot
Accelerated
Total
6 1 1 1 1 10
6 0 3 0 1 10
12 1 4 1 2 20
4432
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 2 8 e4 4 3 6
2008, December 29, 2008, and March 2, 2009. Accelerated Members processed a single panel consisting of samples already tested by Pilot Members that was shipped March 20, 2009. All laboratories enrolled during this time were given two months to process a panel and report results.
2.3.
Data collection and analysis
Combined results from Pilot and Accelerated Members were captured from the CDC SQL 2005 database June 12, 2009. Reported laboratory concentrations were tabulated and summarized. Differences were computed and tabulated between the reported concentrations and (a) the CDCvalidated concentration (re-test values) and (b) the intralaboratory mean among all laboratories reporting a value. In addition, tabulations of un-quantified identifications were made. Quantitative results were analyzed directly and through log transformations. Reporting laboratories were categorized, respectively, by: type of laboratory (state, local, federal, hospital, commercial); annual volume of Legionella tests performed; how long (years) the laboratory has tested for Legionella; number of full-time and Legionella-dedicated employees; and by protocol-related factors such as incubation time and temperature, water volume, use of CO2, and standard protocol reference (ASM, ISO, or CDC). State, county, and hospital laboratories were grouped into the category “Local Public Health” for further analyses. For each categorization of laboratories or procedures, Chi-square tests were computed for dichotomous outcomes such as finding Legionella or not or being within a specified error, such as one log, of the respective (re-test or interlaboratory mean) reference concentration. In addition, for continuous measures, t-tests were performed comparing the errors of mutually-exclusive groups of laboratories, such laboratories which used less than 750 ml of water versus those who used 750 ml or more.
3.
Results
3.1.
PT sample qualitative identification
Concordance between expected positive or negative results, as determined by QC, and results returned by Pilot or Accelerated Members was high regardless of laboratory type (Table 3). The majority of positive samples (88.5%) were correctly identified. No false positives from a negative sample were reported by any member. In contrast, 36.9% of all variable samples tested (n ¼ 103) were incorrectly identified (Table 4). The federal laboratory was most often correct, followed by commercial laboratories, and then local public health laboratories. Correct identification of positive samples depended on the Legionella concentrations in the samples. Samples with less than 10 cfu/ml, as determined by re-test, were identified as negative in 93.1% of reports while samples with 10 cfu/ml or more were reported positive in 85.3% of reports. These results were independent of whether the sample was pure or mixed, positive or variable indicating that 10 cfu/ml is at or near the lower limit of detection. In addition to whether a sample was positive or negative for Legionella growth, participating
Table 3 e Concordance of positive and negative sample reports by laboratory type: displays the agreement between expected results and participants’ reported results for all positive or negative samples, both mixed and pure. Variable samples’ reported results are not included in this table. Definitions of column headings are: True Negative [ expected negative and reported negative, False Negative [ expected positive but reported negative, True Positive [ expected positive and reported positive.
Commercial Federal Local public health Total results (n)
True negative
False negative
True positive
Total results (n)
56 (100%) 8 (100%) 32 (100%)
3 (6.3%) 0 (0%) 7 (21.9%)
45 (93.8%) 7 (100%) 25 (78.1%)
104 15 64
96 (100%)
10 (11.5%)
77 (88.5%)
183
laboratories had the option to report species and serogroup of legionellae recovered. All laboratories that provided optional results correctly identified the species and serogroup of recovered Legionella.
3.2.
Accuracy and precision of Legionella quantitation
Pilot and Accelerated Members were encouraged to also provide results for their observed concentration of legionellae in samples (n ¼ 128). The expected concentration for each sample was determined by re-test (Table 1). The log error for each report is displayed graphically in Fig. 1. The majority (99.5%) of reported results underestimated the expected concentration by an average of 1.25 logs with a standard deviation of 0.78 logs, standard error of 0.07 logs, and range of 3.57 (3.30 to 0.27) logs. Centering the results on the interlaboratory mean forced an overall net difference of zero logs (Fig. 2). The standard deviation was 0.62 logs, standard error was 0.06 logs and the range was 3.22 (1.89 to 1.33) logs. The standard deviation from the interlaboratory mean of the results from laboratories that entered at least three concentrations for evaluation (n ¼ 11) ranged from 0.53 to 0.96 and standard error from 0.14 to 0.58. No sampling protocol, treatment regimen, incubation procedure, or organizational structure type analyzed produced significantly greater accuracy when compared to re-test values or precision when compared to the interlaboratory mean concentration (Table 5).
Table 4 e Accuracy of variable sample reports by laboratory type: Displays the reported results for all variable samples, both mixed and pure. No positive or negative sample results are included in this table.
Commercial Federal Local public health Total results (n)
Reported positive
Reported negative
Total results (n)
39 (62.9%) 7 (77.8%) 19 (59.4%)
23 (37.1%) 2 (22.2%) 13 (40.6%)
62 9 32
65 (63.1%)
38 (36.9%)
103
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 2 8 e4 4 3 6
4433
Fig. 1 e Accuracy of quantitative results relative to re-test values for all samples reported positive: Displays the log error of the difference between re-test values and reported concentrations for all samples reported positive. Each marker represents a sample reported positive by a participant (n [ 190). Markers are coded by laboratory type: commercial (green circle), federal (red square), local public health (blue triangle). Enumeration error [Error(Log(CFU))] was calculated by taking the difference between the log value of the expected concentration in cfu/ml and the log value of the concentration reported by a participant in cfu/ml. Participant responses are plotted as enumeration error ( y-axis) vs. expected concentration derived from re-test values (x-axis). Samples reported positive without a concentration value (n [ 62) are indicated by markers connected with the blue line ( y [ Lx) and are not included in calculations measuring accuracy. The green line depicts the mean enumeration error across the range of reported concentrations (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
4.
Discussion
The results from the ELITE Pilot Program indicate that most participating US laboratories are capable of qualitatively
identifying Legionella spp. from a water sample and that the lower limit of detection is approximately 10 cfu/ml. According to 2009 EU Health Protection Agency external quality assessment for Legionella isolation from water samples reports, the accuracy of the European PT participants (Shah et al., 2009a,
Fig. 2 e Precision of quantitative results relative to interlaboratory means for all samples reported positive: Displays the log error of the difference between interlaboratory mean values and reported concentrations for all samples reported positive. Each marker represents a sample reported positive by a participant (n [ 190). Markers are color coded by laboratory type: commercial (green circle), federal (red square), local public health (blue triangle). Enumeration error [Error(Log(CFU))] was calculated by taking the difference between the log value of the interlaboratory mean cfu/ml and the log value of the concentration reported by a participant in cfu/ml. Participant responses are plotted as enumeration error ( y-axis) vs. expected interlaboratory mean (x-axis). Samples reported positive without a concentration value (n [ 62) are indicated by markers connected with the blue line (y [ -x) and were not included in calculations to determine the interlaboratory mean (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
4434
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 2 8 e4 4 3 6
Table 5 e Statistical analysis of accuracy as a function of procedural variables: The enumeration errors associated with treatment, protocol, and experience categories were compared with t-tests. The p-values are given when the baseline is calculated from either re-test values or the interlaboratory mean. Measurement
Incubation temperature Incubation with CO2 Water volume Incubation time Sample composition Sampling protocol Years of testing Number of dedicated employees Number of samples processed annually
Categories
p Value compared to re-test
p Value compared to interlaboratory mean
35.0 C or >35.0 C <2.5% CO2 or 2.5e5.0% CO2 (v/v) <750 ml or 750e1000 ml 7 days or >7 days Pure or mixed sample CDC or other 10 or >10 years 2 or >2 employees 200 or >200
0.297 0.537 0.454 0.318 0.065 0.221 0.672 0.993 0.919
0.311 0.332 0.476 0.562 >0.999 0.216 0.159 0.960 0.565
2009b, 2009c, 2009d) averaged 93% concordance (range 78e98%). US laboratories in this study averaged 86% concordance with positive samples, well within the range of EU performance. On the other hand, with regard to precision, US laboratories demonstrated more extreme variability in enumerating bacteria in the PT samples both within and between participants. US labs underestimated the concentration of Legionella by an average of 1.25 (0.78) logs when compared to re-test values. Further, US labs exhibited 0.62 logs variance around the interlaboratory mean. In contrast, EU laboratory results from 2009 clustered very close to the interlaboratory mean and FEPTU median (a reference standard comparable to re-test values), usually within a 0.5 log range of either score, with a mean standard deviation of 0.44 logs (Shah et al., 2009a, 2009b, 2009c, 2009d). Thus, EU and US laboratories demonstrate similar accuracy in identifying positive samples but US laboratories appear to be less precise than EU laboratories in enumerating viable bacteria. The apparent greater precision of enumeration by EU laboratories compared to US laboratories is partially due to the bacterial concentrations used in the two sources of PT samples. EU sample concentrations spanned 0.2e110 cfu/ml while US PT samples contained between 1 and 34,300 cfu/ml (as determined by re-test), allowing the possibility for a larger range of enumeration error within US samples. However, a greater contribution to the disparity in variance between EU and US reported results was the different methodologies used to generate reference standards for comparison. US re-test values were calculated from the highest count recovered from a representative sample after treatment with an individualized protocol that took sample concentration and composition into account. Plating serial dilutions from the PT samples in triplicate, a feature of the majority of individualized re-test protocols (Table 1) but no referenced standard protocol (Anonymous, 1998, 2004, 2005a, 2005b, 2008, 2010), resulted in plates that had reduced numbers of obscuring heterotrophs and increased physical distance between colonies, making Legionella cfu easier to determine. In contrast, EU FEPT medians were generated from the results of 10 samples treated according to a single standard protocol without regard to sample concentration or composition, making the results inherently less variable. Both reference standards are valuable
in assessing participant results but address different laboratory capabilities. Re-tests return the most accurate count of viable bacteria within a sample, independent of protocol bias while the EU FEPT median and ELITE interlaboratory mean values address the precision with which a sample can be enumerated using a (set of) standard protocol(s). Thus, comparison to the re-test value indicates how close a laboratory can come to a ‘true’ answer while FEPT medians and interlaboratory means illustrate how reproducible the results are between and within a laboratory. Laboratories that submitted more than three concentration reports were statistically indistinguishable from each other in enumeration error and also demonstrated large intralaboratory variance (0.53e0.96). Two PT samples were quantified with significantly different log error from re-test values compared to the rest. These observations can be explained by sample composition. A62-08091603 (0.71 vs. 1.27, p ¼ 0.002, n ¼ 5) was a pure sample with a concentration within the optimum range for allowing distinct colony growth on media from a single 10-fold dilution and so could be measured more precisely than other samples. In contrast, sample A3208061706 (2.19 vs. 1.21, p ¼ 0.006, n ¼ 5) was heavily mixed with heterotrophs that had a spreading colony morphology, obscuring legionellae growth, resulting in a sizeable difference between QC and re-test results (Table 1), and a substantial interlaboratory enumeration error. Enumeration of the other 46 samples was statistically indistinguishable. No sampling protocol, treatment, incubation, or experience level analyzed in this study affected the accuracy or precision of enumeration (Table 5) at the 0.05 significance level. Taken together, these results indicate that intralaboratory and interlaboratory variance in precision and accuracy were similar in degree and magnitude for all pilot study participants. Two obvious caveats to these results are the small number of participating laboratories and their method of selection. ELITE Program participation is voluntary rather than mandated by legal statutes, as in the EU, and there was little promotion for the creation and implementation of the program. Most Pilot and Accelerated Members discovered the program through scientific meetings and/or word of mouth. Pilot Members were also likely to have a previous relationship with CDC, multiple years experience with Legionella isolation,
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 2 8 e4 4 3 6
and/or a large number of well-trained personnel dedicated to Legionella testing. Thus, it is possible that Pilot Members do not accurately reflect the capabilities of all US laboratories that culture environmental samples but are a subset of those whose primary focus is Legionella.
5.
Conclusions
Overall, Legionella PT qualitative and quantitative results were similar between US and EU laboratories in 2009 despite differences in sample composition, delivery, and reference standard determination. The observed variability in enumeration by both US and EU laboratories is probably due to the inherent inconsistency in assessing a sample by culture techniques. Given these data no protocol can be recommended to yield more accurate or precise results than any other. Responses from ELITE Program participants will continue to be monitored and analyzed to determine if more data can illuminate practices that contribute to increased accuracy and precision. However, the current findings have several implications for the use of routine sampling as a primary method of legionellosis control. Agreement between EU and US PT schemes suggest these results are applicable worldwide to environmental sampling laboratories. Since a sample qualitatively identified as positive could represent a 3 log cfu/ml range of viable legionellae it would be in the best interests of public health to consider any detectable level a hazard. Therefore, primary legionellosis prevention should consider the risk posed by an individual water system, assessing the likelihood of transmission and population affected, to determine if remediation is required rather than relying on a contamination cutoff level to take action.
Acknowledgments The authors would like to thank the Safe Water Program of the Division of Environmental Hazards and Health Effects, NCEH for funding the ELITE Program implementation and website administration October 1, 2007 through September 30, 2009. The ELITE Program website URL is https://wwwn.cdc.gov/ elite/Public/EliteHome.aspx.
Appendix. Supplementary data Supplementary data associated with this article can be found in the online version, at doi:10.1016/j.watres.2011.05.030.
references
Alli, O.A., Zink, S., von Lackum, N.K., Abu-Kwaik, Y., 2003. Comparative assessment of virulence traits in Legionella spp. Microbiology 149, 631e641. Anonymous, 1997. Guidelines for prevention of nosocomial pneumonia. MMWR Recomm Rep 46, 1e79. Centers for Disease Control and Prevention.
4435
Anonymous, 1998. Water Quality e Detection and Enumeration of Legionella Edited by I.S. Organization. Rule no. 11731:1998 TC no. 147/SC4. Anonymous, 2000. Minimizing the Risk of Legionellosis Associated with Building Water Systems. American Society of Heating, Refrigerating, and Air-Conditioning Engineers, Inc, Atlanta, GA. Document No. 12-2000:1-6. Anonymous, 2004. In: I.S. Organization (Ed.), Water Quality e Detection and Enumeration of Legionella e Part 2: Direct Membrane Filtration Method for Waters with Low Bacterial Counts Rule no. ISO 11731-2:2004. Anonymous, 2005a. European Guidelines for Control and Prevention of Travel Associated Legionnaires’ Disease. European Parliament and the Council. January 2005. 2119/ 98/EC. Anonymous, 2005b. In: Fields, B.S. (Ed.), Procedures for the Recovery of Legionella from the Environment. US Department of Health and Human Services, Public Health Service, Centers for Disease Control and Prevention, Atlanta, GA, pp. 1e13. http://www.cdc.gov/legionella/files/LegionellaProcedures-508. pdf. Recommendation January 2005. Anonymous, 2008. Standard Guide for The Inspection of Water Systems for Legionella and The Investigation of Possible Outbreaks of Legionellosis (Legionnaires’ Disease or Pontiac Fever) ASTM Rule no. D5952-08, 2008. Anonymous, 2010. Legionella Isolation Scheme. European Union Health Protection Agency, London, UK. www.hpa.org.uk/web/ HPAwebFile/HPAweb_C/1254510409973. Bartie, C., Venter, S.N., Nel, L.H., 2003. Identification methods for Legionella from environmental samples. Water Research 37, 1362e1370. Benson, R.F., Fields, B.S., 1998. Classification of the genus Legionella. Seminars in Respiratory Infections 13, 90e99. Bentham, R.H., 2000. Routine sampling and the control of Legionella spp. in cooling tower water systems. Current Microbiology 41, 271e275. Boulanger, C., Edelstein, P., 1995. Precision and accuracy of recovery of Legionella pneumophila from seeded tap water by filtration and centrifugation. Applied and Environmental Microbiology 61, 1805e1809. Butler, J., Fields, B.S., Breiman, R.F., 1997. Prevention and Control of Legionellosis. Infectious Disease and Clinical Practice 6, 458e464. Den Boer, J.W., Verhoef, L., Bencini, M.A., Bruin, J.P., Jansen, R., Yzerman, E.P., 2007. Outbreak detection and secondary prevention of Legionnaires’ disease: a national approach. International Journal of Hygiene and Environmental Health 210, 1e7. Ditommaso, S., Giacomuzzi, M., Gentile, M., Moiraghi, A.R., Zotti, C.M., 2010. Effective environmental sampling strategies for monitoring Legionella spp contamination in hot water systems. American Journal of Infection Control 38 (5), 341e343. Fields, B.S., Benson, R.F., Besser, R.E., 2002. Legionella and Legionnaires’ disease: 25 years of investigation. Clinical Microbiology Reviews 15, 506e526. Fields, B.S., Moore, M.R., 2006. Control of legionellae in the environment: a guide to the US guidelines. ASHRAE Transactions 112, 691e699. Fiore, A.E., Butler, J.C., Emori, T.G., Gaynes, R.P., 1999. A survey of methods used to detect nosocomial legionellosis among participants in the National Nosocomial Infections Surveillance System. Infection Control and Hospital Epidemiology 20, 412e416. Force, A.C.H.D.L.T., 1997. Approaches to Prevention and Control of Legionella Infection in Allegheny County Health Care Facilities. Allegheny County Health Department, Pittsburgh, PA, pp. 1e15.
4436
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 2 8 e4 4 3 6
Freije, M.R., 2004. The word on water: new CDC guidelines recommend a proactive approach to Legionella. Health Facilities Management 17, 33e34. 36. Helbig, J.H., Benson, R.F., Pelaz, C., Jacobs, E., Luck, P.C., 2007. Identification and serotyping of atypical Legionella pneumophila strains isolated from human and environmental sources. Journal of Applied Microbiology 102, 100e105. Lee, T.C., Vickers, R.M., Yu, V.L., Wagener, M.M., 1993. Growth of 28 Legionella species on selective culture media: a comparative study. Journal of Clinical Microbiology 31, 2764e2768. Leoni, E., Legnani, P.P., 2001. Comparison of selective procedures for isolation and enumeration of Legionella species from hot water systems. Journal of Applied Microbiology 90, 27e33. Luck, P.C., Igel, L., Helbig, J.H., Kuhlisch, E., Jatzwauk, L., 2004. Comparison of commercially available media for the recovery of Legionella species. International Journal of Hygiene and Environmental Health 207, 589e593. Napoli, C., Iatta, R., Fasano, F., Marsico, T., Montagna, M.T., 2009. Variable bacterial load of Legionella spp. in a hospital water system. The Science of The Total Environment 408, 242e244. O’Neill, E., Humphreys, H., 2005. Surveillance of hospital water and primary prevention of nosocomial legionellosis: what is the evidence? The Journal of Hospital Infection 59, 273e279. Palusinska-Szysz, M., Cendrowska-Pinkosz, M., 2009. Pathogenicity of the family Legionellaceae. Archivum Immunologiae et Therapiae Experimentalis 57, 279e290. Reeves, M.W., Pine, L., Hutner, S.H., George, J.R., Harrell, W.K., 1981. Metal requirements of Legionella pneumophila. Journal of Clinical Microbiology 13, 688e695. Sehulster, L., Chinn, R.Y., 2003. Guidelines for environmental infection control in health-care facilities. Recommendations of CDC and the Healthcare Infection Control Practices Advisory Committee (HICPAC). MMWR Recommendations and Reports 52, 1e42. Shah, H., Lang, N.L., Russell, J.E., 2009a. External quality assessment for: Legionella isolation from water samples. In: Russell, J.E. (Ed.), Summary of Results, G64. Center for Infections/Food and Environmental Proficiency Testing Unit, London, UK, pp. 1e15. http://www.hpa.org.uk/web/ HPAwebFile/HPAweb_C/1194947324524. Shah, H., Lang, N.L., Russell, J.E., 2009b. External quality assessment for: Legionella isolation from water samples. In: Russell, J.E. (Ed.), Summary of Results, G65. Center for Infections/Food and Environmental Proficiency Testing Unit, London, UK, pp. 1e15. http://www.hpa.org.uk/web/ HPAwebFile/HPAweb_C/1245581545311. Shah, H., Lang, N.L., Russell, J.E., 2009c. External quality assessment for: Legionella isolation from water samples. In: Russell, J.E. (Ed.), Summary of Results, G66. Center for Infections/Food and Environmental Proficiency Testing Unit,
London, UK, pp. 1e15. http://www.hpa.org.uk/web/ HPAwebFile/HPAweb_C/1249454476861. Shah, H., Lang, N.L., Russell, J.E., 2009d. External quality assessment for: Legionella isolation from water samples. In: Russell, J.E. (Ed.), Summary of Results, G67. Center for Infections/Food and Envinronmental Proficiency Testing Unit, London, UK, pp. 1e15. http://www.hpa.org.uk/web/ HPAwebFile/HPAweb_C/1259151931733. Shahamat, M., Paszko-Kolva, C., Keiser, J., Colwell, R.R., 1991. Sequential culturing method improves recovery of Legionella spp. from contaminated environmental samples. Zentralbl Bakteriol 275, 312e319. Stout, J.E., Muder, R.R., Mietzner, S., 2007. & other authors Role of environmental surveillance in determining the risk of hospital-acquired legionellosis: a national surveillance study with clinical correlations. Infection Control and Hospital Epidemiology 28, 818e824. Ta, A.C., Stout, J.E., Yu, V.L., Wagener, M.M., 1995. Comparison of culture methods for monitoring Legionella species in hospital potable water systems and recommendations for standardization of such methods. Journal of Clinical Microbiology 33, 2118e2123. Tablan, O.C., Anderson, L.J., Besser, R., Bridges, C., Hajjeh, R., 2004. Guidelines for preventing health-careeassociated pneumonia, 2003: recommendations of CDC and the Healthcare Infection Control Practices Advisory Committee. MMWR Recommendations and Reports 53, 1e36. Tesh, M.J., Miller, R.D., 1981. Amino acid requirements for Legionella pneumophila growth. Journal of Clinical Microbiology 13, 865e869. Thurmer, A., Helbig, J.H., Jacobs, E., Luck, P.C., 2009. PCR-based ’serotyping’ of Legionella pneumophila. Journal of Medical Microbiology 58, 588e595. Tronel, H., Hartemann, P., 2009. Overview of diagnostic and detection methods for legionellosis and Legionella spp. Letters in Applied Microbiology 48, 653e656. Wagner, C., Kronert, C., Luck, P.C., Jacobs, E., Cianciotto, N.P., Helbig, J.H., 2007. Random mutagenesis of Legionella pneumophila reveals genes associated with lipopolysaccharide synthesis and recognition by typing monoclonal antibodies. Journal of Applied Microbiology 103, 1975e1982. Wiedenmann, A., Langhammer, W., Botzenhart, K., 2001. A case report of false negative Legionella test results in a chlorinated public hot water distribution system due to the lack of sodium thiosulfate in sampling bottles. International Journal of Hygiene and Environmental Health 204, 245e249. Yu, V.L., 1997. Prevention and control of Legionella: an idea whose time has come. Infectious Disease and Clinical Practice 6, 420e421.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 3 7 e4 4 4 8
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Refining the estimation of illicit drug consumptions from wastewater analysis: Co-analysis of prescription pharmaceuticals and uncertainty assessment Foon Yin Lai a, Christoph Ort b,c,*, Coral Gartner d, Steve Carter e, Jeremy Prichard f, Paul Kirkbride g, Raimondo Bruno h, Wayne Hall d, Geoff Eaglesham a,e, Jochen F. Mueller a a
The University of Queensland, The National Research Centre for Environmental Toxicology (Entox), 39 Kessels Road, Coopers Plains, QLD 4108, Australia b The University of Queensland, Advanced Water Management Centre (AWMC), St. Lucia, QLD 4072, Australia c Eawag, Swiss Federal Institute of Aquatic Science and Technology, CH 8600 Du¨bendorf, Switzerland d The University of Queensland, UQ Centre for Clinical Research, Royal Brisbane and Women’s Hospital, Herston, QLD 4029, Australia e Queensland Health Forensic Scientific Services (QHFSS), Queensland Government, 39 Kessels Road, Coopers Plains, QLD 4108, Australia f Law Faculty, University of Tasmania, Private Bag 89, Hobart TAS 7001, Australia g Australian Federal Police, Forensic and Data Centers, GPO Box 401, Canberra, ACT 2601, Australia h School of Psychology, University of Tasmania, Private Bag 30, Hobart TAS 7001, Australia
article info
abstract
Article history:
Wastewater analysis is a promising monitoring tool to estimate illicit drug consumption at
Received 21 February 2011
the community level. The advantage of this technique over traditional surveys and other
Received in revised form
surveillance methods has been emphasized in recent studies. However, there are meth-
21 May 2011
odological challenges that can affect reliability. The objectives of this study were to
Accepted 28 May 2011
systematically reduce and assess uncertainties associated with sampling (through a strin-
Available online 14 June 2011
gent optimization of the sampling method) and the back calculation of per capita drug consumption (through a refined estimation of the number of people actively contributing
Keywords:
to the wastewater in a given period). We applied continuous flow-proportional sampling to
Normalization
ensure the collection of representative raw wastewater samples. Residues of illicit drugs,
Estimated population
opioids, prescription pharmaceuticals and one artificial sweetener were analyzed by liquid
Error propagation
chromatography coupled with tandem mass spectrometry. A parameter estimating the
LC-MS/MS
number of people actively contributing to wastewater over a given period was calculated
Australia
from the measured loads of prescription pharmaceuticals, their annual consumption and relative excretion data. For the calculation of substance loads in sewage, uncertainties were propagated considering five individual components: sampling, chemical analysis, flow measurements, excretion rates and the number of people contributing to the wastewater. The daily consumption per 1000 inhabitants was estimated to be almost 1000 mg for cannabis and several hundred mg for cocaine, methamphetamine and ecstasy. With the best sampling practice and current chemical analysis, we calculated the remaining uncertainty to be in the range of 20e30% (relative standard deviation, RSD) for the estimation of consumed drug masses in the catchment; RSDs for the per capita consumption were lower (14e24%), as one of the biggest uncertainty components (i.e. error in flow measurements) cancels out in the proposed method for the estimation of the number of
¨ berlandstrasse 133, P.O. Box 611, CH 8600 Du¨bendorf, Switzerland. Tel.: þ41 (0) 58 765 * Corresponding author. Present address: Eawag, U 52 77; fax: þ41 (0) 58 765 53 89. E-mail address:
[email protected] (C. Ort). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.05.042
4438
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 3 7 e4 4 4 8
people contributing to the daily wastewater volume. In this study, we provide methodological improvements that substantially enhance the reliability of the estimation method e a prerequisite for the application of this technique to meaningfully assess changes in drug consumption and the success of drug intervention strategies in future studies. ª 2011 Elsevier Ltd. All rights reserved.
List of abbreviations A
annual consumption of prescription pharmaceutical C concentration of chemical in raw wastewater E average excretion fraction of a given drug residue ENCP estimated number of contributing people (i.e. the number of people that actively contributed to the daily wastewater volume which was sampled) EDDP 2-ethylidene-1,5-dimethyl-3,3diphenylpyrrolidine F total flow in a day load of chemical in sewer Lsewer per capita consumption of a given illicit drug in MENCP the catchment normalized with ENCP Mcatchment consumed mass of chemical in the catchment MDA 3,4-methylenedioxyamphetamine MDEA 3,4-methylenedioxyethamphetamine MDMA 3,4-methylenedioxymethamphetamine P population
1.
Introduction
The use of illicit drugs in Australia and many other developed countries adversely affects population health, social order and the economy (Collins and Lapsley, 2008; Degenhardt et al., 2004). It is difficult to estimate the prevalence of an illegal, highly stigmatized and clandestine activity. To date, the extent and trends of illicit drug use at the national level have been monitored through indirect methods, such as socioepidemiological studies, mortality data from drug-related causes (e.g. opioid overdoses) and monitoring of border seizures and seizures from domestic traffickers and users (Shand et al., 2003; UNODC, 2010). In addition to these methods, Daughton (2001) proposed estimating illicit drug consumption through the analysis of drug residues in raw wastewater. Zuccato et al. (2005) applied this method by quantifying excreted parent drugs and their key metabolite(s) in wastewater samples. Their data provided the first direct estimates for illicit drug use (Zuccato et al., 2005, 2008). They also reported predictable temporal variations, with the highest loads of certain drugs occurring on weekends. A number of research teams have since applied this approach in cities in Europe, North America and Australia (Boleda et al., 2009; Bones et al., 2007; Chiaia et al., 2008; Huerta-Fontela et al., 2007, 2008; Irvine et al., in press; Karolak et al., 2010; Metcalfe et al., 2010; Postigo et al., 2010; Terzic et al., 2010; van Nuijs et al., 2009a,b,c; Zuccato et al., 2008). A potential key application is to estimate and assess changes in illicit drug
R ratio of molar mass of parent drug to its metabolite RSD relative standard deviation STP sewage treatment plant THC Δ9-tetrahydrocannabinol THCeCOOH 11-nor-9-carboxy-D9-tetrahydrocannabinol Total population in Australia in 2009 (i.e. 22 TPAUS million) uncertainty due to chemical analysis UC uncertainty of the estimated number of UENCP contributing people uncertainty of the fraction of metabolite to parent UE compound uncertainty of flow meter UF uncertainty of chemical loads in sewer ULsewer UMcatchment uncertainty of illicit drug loads consumed in the catchment uncertainty of illicit drug loads normalized to 1000 UMENCP people (ENCP) sampling uncertainty US
consumption in a given population. This approach provides valuable data for health and law enforcement agencies regarding, for example, the impact of strategies that attempt to reduce the supply of, or the demand for, illicit drugs. In these contexts, it is essential to recognize and minimize uncertainties related to the parameters used for the calculation of illicit drug consumption based on wastewater analysis. Many of the previous wastewater studies provided sophisticated analytical methods, explained the back calculation (i.e. estimation of drug consumption per capita) and demonstrated the applicability and potential of these methods to estimate illicit drug consumption in a given population. However, there are numerous uncertainties affecting these estimates. To date, uncertainties related to sample collection and the number of people contributing to the wastewater have been insufficiently reported or not reported at all. Recently, Ort et al. (2010c) noted that these studies typically have not presented evidence that the analyzed samples were collected in an appropriate manner; proper sampling is fundamental for the accurate calculation of chemical loads (Ort et al., 2010b). Depending on the characteristics of the sewer system and the exact sampling location, a flow-proportional, high-frequency sampling method is required, since common sampling modes and frequencies can result in systematic and random artifacts ranging from “insignificant” to “100% or more” (Ort et al., 2010c). This becomes even more important if samples are collected farther upstream in small sub-catchments or in the effluent of
4439
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 3 7 e4 4 4 8
individual institutions to evaluate intervention strategies (Prichard et al., 2010). Furthermore, the back calculation requires reliable information on the size of the population that has contributed to the sampled wastewater. Census data or design capacities of sewage treatment plants (STPs) have typically been used for these calculations. However, in reality this may not reflect the actual number of people in the contributing population, and may be variable due to daily commuting patterns, seasonal variability and/or special events. Zuccato et al. (2008) recognized the limitation of assuming one fixed population size and suggested that the accuracy of the back calculation could be improved by measuring human biomarkers in wastewater to estimate the number of people in a catchment. One candidate marker is creatinine (Chiaia et al., 2008). Recently van Nuijs et al. (2011) proposed using the average of loads of nitrogen, phosphorus and chemical and biological oxygen demand to estimate a day-specific number of inhabitants in the catchment. However, the loads of these conventional water quality parameters may be significantly and systematically influenced by other sources such as industrial discharges. To our knowledge, no practical correction methods using indicators that are specific to humans have been used to date to estimate the number of people who have contributed to sampled wastewater. In this study, we aimed to enhance the reliability of the back calculation method for illicit drug consumption through: (a) the collection of representative wastewater samples using a continuous flow-proportional sampling technique at the influent of a municipal sewage treatment plant; (b) the evaluation of daily variations in the number of people that actively contributed to a given sample by conducting analysis of prescription pharmaceuticals in wastewater; and (c) the assessment of the remaining uncertainty of the estimates.
2.
Material and methods
2.1.
Sample collection
Sampling was conducted during a dry weather period from 20th November to 1st December in 2009 at the influent of a municipal STP located in South-East Queensland. The STP serves a growing population of approximately 300,000 to 350,000 (Regional Council data). Sewage volumes, from 8:00 AM to 8:00 AM the following day, ranged from 52.6 to 55.5 ML day1. Daily 24-h composite samples from the raw influent were collected before the primary clarifier, refrigerated at 4 C during collection and kept on ice during transportation. The online signal of the flow sensor in the inlet of the STP was used to control the speed of a peristaltic sampling pump: a continuous flow-proportional side stream (a few mL min1) ensured representative samples (Ort et al., 2010b). Samples were preserved at pH 2 with 2M hydrochloric acid on site and stored at 20 C until analysis in amber glass bottles which were prerinsed with dichloromethane and methanol.
2.2.
Chemicals of interest
We measured residues of ten illicit drugs (five parent compounds and five primary metabolites) and four psychoactive
opioids that are most frequently used in Australia according to the National Drug Strategy Household Survey 2007 (AIHW, 2008). These included cocaine, benzoylecgonine, ecgonine methyl ester, amphetamine, methamphetamine, 3,4-methylenedioxymethamphetamine (MDMA), 3,4-methylenedioxyethamphetamine (MDEA), 3,4-methylenedioxyamphetamine (MDA), Δ9-tetrahydrocannabinol (THC), 11-nor-9-carboxy-D9tetrahydrocannabinol (THC-COOH), methadone, 2-ethylidene1,5-dimethyl-3,3-diphenylpyrrolidine (EDDP), codeine and morphine. Furthermore, we included and analyzed five prescription pharmaceuticals, i.e. atenolol, carbamazepine, gabapentin, hydrochlorothiazide and venlafaxine, to assess their suitability to refine the estimation of the number of people that contributed to wastewater in our catchment. Additionally, we also analyzed acesulfame, an artificial sweetener contained in beverages and food that are widely consumed. Acesulfame is almost completely excreted as parent compound, persistent in wastewater and can be measured accurately.
2.3.
Chemical analysis
We adopted and validated our analytical method to analyze illicit drugs and opioids from those reported in the literature (e.g., Boleda et al., 2007; Castiglioni et al., 2006, 2008). Briefly, acidified and filtered samples (200 mL) were spiked with the deuterated analogs of targeted chemicals (10e100 ng) and then loaded onto Oasis MCX cartridges (6 cc, 150 mg, 30 mm) preconditioned with methanol (6 mL), Milli-Q water (4 mL) and HCl-acidified Milli-Q water at pH 2 (4 mL). Subsequently, cartridges were dried by centrifugation at 3000 rpm for 3 min. The target analytes were eluted into two different fractions with methanol and then 2% ammonia hydroxide in methanol. Both fractions were concentrated using a gentle stream of high purity nitrogen gas. Final extracts of the first and second fractions were reconstituted with methanol and an aqueous solution of 5% acetonitrile and 0.1% formic acid, respectively, and then measured by high-performance liquid chromatography coupled to a triple quadrupole tandem mass spectrometer. For prescription pharmaceuticals and acesulfame, filtered samples (400 mL) were spiked with deuterated standards (8e40 ng) and then directly analyzed (for details see Tables S1 and S2, in the Supplementary Information). The mean recovery of the reported illicit drug and opioid residues in Milli-Q water and wastewater samples were in the range of 89e106% (inter-day RSD 4e19%) and 83e113% (inter-day RSD 2e19%), respectively. The inter-day precision of the quantification of prescription pharmaceuticals and acesulfame was 3e20% (for details see Table S2).
2.4.
Back calculation
The common approach to estimate the daily consumption of a drug normalized per 1000 inhabitants in a given catchment by means of wastewater analysis can be found e.g. in Zuccato et al. (2008): 0
1 mg Ri Ci $F$ day Ei B C Daily drug consumptioni @ A ¼ P 1000 people
4440
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 3 7 e4 4 4 8
where Ci is the concentration of a given drug residue i (parent drug or metabolite) measured in raw wastewater samples, F is the total flow during the sampling period (typically 24 h), P is the number of people in the catchment, Ri is the ratio of molar mass of parent drug to its metabolite and Ei is the average excretion rate of a drug residue i (see Table 1). Here we present a set of equations (see Table 2) to consequently assess the uncertainties associated with each step in a transparent way, from the drug load in the sewer (Lsewer), over the back calculation to the consumed mass in the catchment (Mcatchment) to the consumed mass normalized per 1000 people (MENCP). The uncertainties (U) affecting the accuracy of Lsewer are US (sampling), UC (chemical analysis) and UF (flow measurement). Additionally, for Mcatchment, UE (excretion rate) is considered; UB (biodegradation in sewer) was not assessed in our study. The magnitude of uncertainty was determined either experimentally in the laboratory (UC), adopted from the literature (UE), estimated based on a model (US) or specified by STP operators (UF). While the values in our equation may not be independent (e.g. higher flow due to rain may imply lower concentrations of illicit drugs), the tools and methods to quantify each value are independent. Hence, we can apply Gaussian error propagation to calculate the total uncertainty (for multiplications and divisions of independent uncertainties the squared RSDs can be summed). The uncertainty of ENCP is calculated based on the four individual uncertainty components introduced above.
2.4.1.
Load in sewer (Lsewer)
Three factors of uncertainty influence the calculation of Lsewer (Eq. 1.2): US and UC, both affecting the accuracy of the average concentration (C), and UF. In theory, with the sampling mode we applied, US would be zero. However, we conservatively assign US a value of 5% to account for unforeseen or unknown uncertainties; as we will see later, this value will not contribute substantially to the total uncertainty. The values for UC were determined based on the inter-day precision (RSD) in the wastewater matrix which is 2e19% for the illicit drug and
Table 1 e Pharmacokinetic data of targeted illicit drug residues. Parent drug Cocaine Methamphetamine MDMA THC
Illicit drug residue Cocaine Benzoylecgonine Methamphetamine Amphetamine MDMA THCeCOOH
Ei (range) (%) b,c,e,i
7.5 (0.08e15) 35 (14e55)b,c,e,i 39 (2e76)d,g 5.5 (2e76)d,g 15 (6e25)a,j 0.6d,h(20e35)f,k
Ri 1.0 1.1 1.0 1.1 1.0 0.9
a Abraham et al., 2009. b Ambre et al., 1984. c Ambre et al., 1988. d Baselt, 2008. e Cone et al., 2003. f Grotenhermen, 2003. g Postigo et al., 2008. h Zuccato et al., 2008. i From 26 subjects over different administration routes (intravenous, intranasal and smoke). j From 32 subjects in oral dose. k Acid metabolites in urine.
opioid residues and 3e20% for the prescription pharmaceuticals and acesulfame (for details see Table S2). Usually, when asking the STP staff for a value for UF, only precision (factory settings) can be obtained. These random errors are typically very small and cancel each other out over the course of a day when integrating the flow to calculate the daily wastewater volume. However, systematic errors are difficult to recognize and quantify (e.g. offset from installation or deterioration over time). Through dialogue with the staff of the STP sampled in this study, a conservative estimate of 20% was made, which seems reasonable in view of other studies (e.g. Thomann, 2008).
2.4.2.
Consumed mass in the catchment (Mcatchment)
Two additional uncertainty factors are involved in the back calculation to estimate Mcatchment based on Lsewer: UE and UB. The effect of UB was not assessed but is discussed in section 3.4. The variable (uncertain) excretion rate (UE) of a user is compoundspecific and was estimated from literature (Table 1). Values for excretion rates are reported sparsely in the literature and sometimes only ranges instead of individual values are specified; it is difficult to meaningfully derive a distribution and average from a small number of values or a minimum to maximum range. To avoid excluding values outside the reported min to max range, we assumed a normal distribution instead of a uniform distribution for this study and calculated the average as (min þ max)/2. The standard deviation of the normal distribution was calculated as ((maxmin)/2)/O(3). The estimated standard deviation was then used to calculate the relative standard deviation (i.e. UE) for the error propagation. Taking the excretion of cocaine as an example, the reported range is 0.08e15% (Table 1). The average is 7.5% [(0.08 þ 15)/2] and the standard deviation is 4% [((15e0.08)/2)/ O(3)]. For the error propagation UE needs to be a RSD, which in this example is about 57% [4/7.5]. It has to be noted that this value for UE is for one single user. The combined excretion rate of many users will tend toward the mean and the uncertainty UE for n users decreases by O(n) (see Eq. 2.2). The consumed mass of cocaine was back calculated from both cocaine and benzoylecgonine measurements, the mass for methamphetamine was back calculated from both methamphetamine and amphetamine measurements, the mass for MDMA was back calculated only from MDMA measurements and the mass for THC was back calculated only from THC-COOH measurements (Table 1)
2.4.3.
Estimated number of contributing persons (ENCP)
According to Eq. 3.1, the number of people (ENCP) that are present in the catchment and actively contributed to the sampled wastewater can be derived from measured mass loads of prescription pharmaceuticals in the sewer relative to the expected consumption (national audit data, PBS 2009) after correcting for excretion rates (see Table 3). This idea holds true if changes in levels of prescription pharmaceuticals that need to be consumed regularly in consistent amounts are assumed to reflect changes in the overall number of people in the catchment. A similar approach has been proposed in a recent review to reduce errors and thus advance the utility of the back calculation (Daughton, 2011). If no local or regional consumption data are available for the catchment and prescription drug of interest one must assume a homogenous
4441
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 3 7 e4 4 4 8
Table 2 e Equations used to back calculate consumed drug masses and assess effects of uncertainties. Eq.
Formula
Eq.
Lsewer;i ¼ Ci $F
1.1
Uncertainty qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ULsewer;i ¼ U2Si þ U2Ci þ U2F vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi !2 u u UE UMcatchment;i ¼ tU2Si þ U2Ci þ U2F þ pffiffiffiffii ffi þ ½U2B;i ni vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi !2 u u UEj UENCPj ¼ tU2Si þ U2Ci þ U2F þ pffiffiffiffiffi þ ½U2B;j nj vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ffi !2 !2 u u U U E E j t 2 2 2 2 i UMENCPi;j ¼ U2Si þ U2Ci þ pffiffiffiffiffi þ ½UB;i þ USi þ UCj þ pffiffiffiffiffi þ ½UB;j nj ni
1.2 Ci $F$REii
2.2
R
3.2
2.1
Mcatchment;i ¼
3.1
ENCPj ¼ Cj $F$Ejj $TPAAUS j 365
4.1 MENCPi;j ¼
Ci $
Ri Ei
4.2
R
Cj $Ejj $TPAAUS j 365
Eq. 1.1e1.2: Ci is the measured concentration of a given drug residue i (parent drug or metabolite) measured in raw wastewater samples; F is the total flow during the sampling period (typically 24 h); Lsewer: chemical load in sewer; USi is the uncertainty of sampling; UCi is the uncertainty of chemical analysis; UF is the uncertainty of flow measurement; ULsewer is the uncertainty of Lsewer. To the consumed mass of the parent drug; Eq. 2.1e2.2: Ri is the ratio of molar mass of parent drug to its metabolite; Ei is the average excretion rate of a drug residue i; UEi is the uncertainty of excretion rate; ni is the number of users in the catchment; USi is the uncertainty of chemical biodegradation; Mcatchment;i is the consumed mass of the parent drug; UMcatchment;i is the uncertainty of calculating the consumed mass of the parent drug; Eq. 3.1e3.2: Cj is the measured concentration of the prescription pharmaceutical j (parent drug) in raw wastewater samples; TPAUS is the total population in Australia in 2009 (22 million) (ABS, 2009); Aj is the annual consumption of the prescription pharmaceutical; UENCPj is the uncertainty of calculating ENCP; Eq. 4.1e4.2: MENCPi;j is per capita consumption of the illicit drug in the catchment normalized with ENCP; UMENCPi;j is the uncertainty of MENCPi;j .
distribution of the annual national consumption. It is then necessary to demonstrate that the demographics in the catchment do not significantly differ from the national average. As shown in Eq. 3.2, UENCP is equal to that of Mcatchment (Eq. 2.2) for prescription pharmaceuticals.
2.4.4.
Per capita consumption in the catchment (MENCP)
The consumed mass of a drug in the catchment (Mcatchment) is normalized with ENCP to calculate the consumption expressed per 1000 people (MENCP). The remaining uncertainty is calculated according to Eq. 4.2. The values for the individual uncertainty components are reported in Table 4. It should be noted that the uncertainties due to systematic flow measurement errors cancel each other out: for example, an erroneously high flow reading would result in an overestimation of Mcatchment of any illicit drug, but also for any load of prescription pharmaceuticals, resulting in an equivalent overestimation of ENCP.
2.5.
Statistical analysis
The relationships among the measured loads of illicit drug and opioid residues, prescription pharmaceuticals and acesulfame in the samples were examined using non-parametric Spearman’s rank correlation because the dataset did not follow a normal distribution (KolmogoroveSmirnov test). The statistical analysis was carried out in XLSTAT e Pro 7.1 software. The results are shown in Table S3 (Supplementary Information).
3.
Results and discussion
3.1.
Estimated drug loads in sewer (Lsewer)
Table 5 summarizes the measured daily loads of illicit drug and opioid residues, prescription pharmaceuticals and acesulfame in the sewer over 12 monitoring days (see also Figs. S1aec,
Table 3 e Total annual consumption and pharmacokinetic data of five prescription pharmaceuticals and one artificial sweetener. Chemicals Atenolol (beta-blocker) Gabapentin (anti-convulsant) Hydrochlorothiazide (cardiac agent, diuretic/kidney treatment, etc) Methadone (addiction treatment, etc) Venlafexine (anti-depressant) Acesulfame (artificial sweetener) a Data from Australia PBS (2009). b average excretion fraction as parent drug in urine. c Australian Medicines Handbook (2010). d Baselt (2008). e Buerge et al. (2009). f Howell et al., 1993 g Lienert et al. (2007). h eMIMS Version 5.01.0100, 2010.
Total consumptiona (kg/year)
Eb (range) (%)
Dosec,h (mg)
6906 6718 3748 107 10709 Not available
37 (33e40)g 78.5 (76e81)d 82 (68e95)g 27.5 (5e50) d 4.7 (SD ¼ 3.1)f 100.5 (100e101)e
25, 50, 100 100, 300 12.5, 25, 200 10, 60, 80 37.5, 300 Not available
4442
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 3 7 e4 4 4 8
Table 4 e Evaluated uncertainty values (RSD) related to sampling, chemical analysis, flow measurement and excretion rate. Illicit drugs and prescription pharmaceuticals
Cocaine CocaineBenzoylecgonine Methamphetamine MethamphetamineAmphetamine MDMA THC Methadone Atenolol Gabapentin Venlafaxine Hydrochlorothiazide Acesulfame
US
5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0
UC
1.6 3.5 2.7 6.0 6.8 4.0 4.2 12 5.3 20 8.9 3.3
UF
20 20 20 20 20 20 20 20 20 20 20 20
ULsewer
21 21 21 21 22 21 21 24 21 29 22 21
UE (n¼1)
57 34 55 16 35 16 47 5.5 1.8 62 10 0.3
UMcatchment
UMcatchmentðnÞ
if n ¼ 1
if n ¼ 100
n
61 40 59 27 41 26 52 24 21 68 24 21
21 21 22 22 22 21 22 24 21 29 23 21
880 670 2960 3330 460 2350 126 6370 1850 9760 1450 nd
21 21 21 21 22 21 21 24* 21* 29* 22* 21*
UMENCP
14 14 14 15 16 14 15 18 15 24 17 nd
CocaineBenzoylecgonine and MethamphetamineAmphetamine: back calculation using the metabolites benzoylecgonine for cocaine and amphetamine for methamphetamine; n: estimated average number of users per day in the catchment when assuming that one person takes one dose per day (average dose: cocaine 100 mg, methamphetamine 20 mg, MDMA 100 mg, THC 125 mg, methadone 50 mg, atenolol 50 mg, gabapentin 200 mg, venlafaxine 170 mg, hydrochlorothiazide 80 mg); nd: not determined. * ¼ UENCP.
Supplementary Information). Of the ten illicit drugs and four opioid residues targeted in the analyses, codeine was found at the highest daily loads (up to 185 g day1). Cocaine, benzoylecgonine, methamphetamine, MDMA and morphine were also measured at relatively high daily loads (3.5 g day1
(minimum for MDMA) to 75 g day1 (maximum for morphine)). Loads of amphetamine, MDA, THCeCOOH, methadone and EDDP were approximately one order of magnitude lower than the other illicit drug and opioid residues (0.65 g day1 (minimum for MDA) to 5.1 g day1 (maximum for EDDP)).
Table 5 e Results in range (median) of the illicit drug and psychoactive opioid residue, prescription pharmaceutical and an artificial sweetener in raw wastewater over the 12 monitoring days. Chemicals
Measured daily loads in sewer (Lsewer) [g/day]
Illicit drug and psychoactive opioid residue Cocaine 3.96e11.9 (5.83) Benzoylecgonine 10.9e45.1 (20.3) Ecgonine methyl ester nd Amphetamine 2.40e4.66 (3.18) Methamphetamine 17.1e34.9 (21.9) MDMA 3.52e13.6 (6.73) MDA 0.653e4.61 (1.78) MDEA nd THC nd THC-COOH 1.23e2.70 (1.90) Morphine 39.2e75.1 (46.9) Codeine 116e185 (136) EDDP 2.54e5.13 (3.49) Prescription pharmaceutical Atenolol 73.6e199 (109) Gabapentin 213e412 (291) Hydrochlorothiazide 72.5e160 (83.3) Methadone 1.35e2.36 (1.64) Venlafaxine 65.1e131 (71) Artificial sweetener Acesulfame 1360e3180 (1760)
Parent drug consumption (census based) [mg/day/1000 peoplea]
Parent drug consumption (ENCP based) [mg/day/1000 peopleb]
176e531 (259) 109e450 (222)d nd 161e312 (213)e 146298 (187) 76.6e297 (147)f nd nd nd 623e1370 (964)g nd nd 37.6e75.9 (51.6) 663e1790 (986) 905e1750 (1240) 295e649 (338) 16.4e28.6 (19.9) 4340e8750 (4720) 4520e10600 (5870)
nd: not determined. a P in Eq. (1) is a constant value estimated from council’s census data with P ¼ 300,000 persons. b P ¼ ENCP evaluated from atenolol. c calculated from its annual consumption of 22 million Australians. Predicted parent illicit drug consumption. d cocaine. e methamphetamine. f MDMA. g THC.
129e614 (202) 98.8e521 (160)d nd 123e360 (176)e 110e345 (145) 44.7e343 (116)f nd nd nd 364e1410 (908)g nd nd 34.8e60.8 (44.0) 860c 837c 467c 13.0c 1330c Not available
4443
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 3 7 e4 4 4 8
The results clearly reflect different patterns among illicit drug and opioids, prescription pharmaceuticals and acesulfame residues in the wastewater over the 12 monitoring days (Figs. S1aec, Supplementary Information). Overall, we found a positive correlation among the loads of different prescription pharmaceuticals, opioids and acesulfame over the sampling period. However, no significant correlation between illicit drug residue loads and prescription pharmaceutical loads or acesulfame loads could be identified (Table S3, Supplementary Information). This finding implies that a higher number of people in the catchment does not necessarily relate to a higher level of illicit drug consumption. While this is not relevant when assessing Lsewer, it must be taken into consideration when determining per capita drug consumption; i.e. applying an ENCP instead of using one number for the population over the monitoring period. Ecgonine methyl ester, MDEA and THC were not quantifiable in the samples. In laboratory tests we found that ecgonine methyl ester was only efficiently recovered from the acidified Milli-Q water (Table S2) but not from the wastewater samples along the whole analytical method, which suggests that ecgonine methyl ester may not be stable in the acidified wastewater matrix. The concentrations of THC-COOH in the samples were found to be between the limits of detection and quantification over the sampling period. Prescription pharmaceuticals were measured in greater daily loads than all illicit drug residues in the samples (Table 5): atenolol (73.6e199 g day1), gabapentin (213e412 g day1), hydrochlorothiazide (72.5e160 g day1) and venlafaxine (65.1e131 g day1). The measured loads of methadone were considerably lower than those of the other prescription pharmaceuticals. Carbamazepine was consistently below the limit of quantification (1 mg L1) and therefore not reported here. Acesulfame, an artificial sweetener, was found in the greatest mass loads among all the analyzed chemicals, at about 9.3 2.5 mg person1 day1 which is comparable to the data from nine STPs in Switzerland (10 3.4 mg person1 day1) (Buerge et al., 2009).
3.2.
Estimated number of contributing persons (ENCP)
3.2.1.
Selection of a suitable marker
The selection of prescription pharmaceuticals as suitable markers to calculate ENCP was based on the following criteria:
1) wide and common usage; 2) known consumption data; 3) frequently detected at relatively high concentrations in wastewater samples in Australia (Ort et al., 2010a); 4) assumed not to significantly biodegrade in sewers; and 5) readily measurable without sample extraction. Gabapentin, hydrochlorothiazide, methadone and venlafaxine were found to be inappropriate because at least one of the above criteria was not met: (a) a relatively small number of patients consume these prescription pharmaceuticals (methadone and gabapentin); (b) the annual PBS consumption data did not capture a substantial proportion of some of the drug use (i.e. in the case of methadone when dispensed through the Methadone Program in hospitals and drug treatment facilities); (c) inaccurate prescription and/or excretion data (i.e. in the application of venlafaxine for which the ENCP resulted in a substantial overestimation of the population); or (d) non-representative or non-homogenous usage (i.e. the database suggests that the use of many of the hydrochlorothiazide analogs is much lower in Queensland than in the rest of Australia (PBS 2009)). This is the reason why mean values for ENCP based on different prescription pharmaceuticals vary significantly (Table 6). Atenolol was found to fulfill the criteria best since (a) the prevalence of atenolol use is relatively high (based on daily consumption and daily doses about 1e3% of the Australian population use atenolol (Table. 6), which needs to be consumed on a regular basis by patients); (b) it is expected to be relatively persistent and mobile in sewers since it is not effectively removed in wastewater treatment plants (Castiglioni et al., 2006; Mie`ge et al., 2009; Onesios et al., 2009); and (c) the per capita atenolol consumption in Queensland is representative for Australia. Twenty percent of the total Australian population lived in Queensland (ABS, 2009) and about 22% of the total atenolol usage in Australia is in Queensland (PBS Item Reports, 2009; Medicare Australia Statistics). Furthermore, the age demographic for atenolol usage (greater than 65 years of age, ABS, 2006), in Australia’s and Queensland’s populations is about 13% and 12% respectively, which is comparable to that in our study catchment of 14% in 2009 (ABS, 2009). This implies that this age group in our study area is representative for Queensland and Australia and so is the average atenolol consumption.
Table 6 e Daily estimated number of contributing people (ENCP) for 12 days according to different prescription pharmaceuticals. Prescription pharmaceuticals Atenolol Gabapentin Hydrochlorothiazide Methadone Venlafaxine
Rangea of ENCPs (1000 inhabitants)
Mean of ENCP (1000 inhabitants)
CVb (total observed variation, n ¼ 12)
Average percentage of users [% of total population]c
240e620 340e620 190e410 340e640 930e1950
370 440 250 470 1200
29% 19% 31% 20% 27%
3, 2, 1 0.8, 0.3 3.7, 1.9, 0.2 0.02, 0.02, 0.13 3.6, 0.4
Estimated population: 300,000e350,000 (based on census data). a minimumemaximum, for details see Table S4. b Coefficient of variation. As a normalized standard deviation, it covers approximately the range/accuracy, one would/could expect based on the complete uncertainty analysis of the whole method if the population was assumed to be constant. However, it is unknown how much the number of people actually varied (inter- and intra-day) due to commuting or any special events during sampling. Therefore, individual values of the ENCP can plausibly be significantly larger or smaller on a daily basis c references for doses are reported in Table 3.
4444
3.2.2.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 3 7 e4 4 4 8
ENCP based on atenolol loads
The average consumed mass of atenolol in the catchment over 12 days was 299 g day1 and results in an average ENCP of approx. 370,000 with an uncertainty of 24% (UENCP). This is in line with the estimate from the Regional Council (growing population of 300,000e350,000). The total observed variability of the ENCP based on measured atenolol loads over the 12 monitoring days was 29% (Table 6). If the methodological uncertainty (UMcatchment ¼ UENCP for atenolol ¼ 24%, Table 4) is subtracted from the total observed variation, a value of 16% [ ¼ O(292e242)] remains, which can be attributed to ‘real’ variation of the number of people in the catchment. This value seems to be reasonable in view of the fact that the sampling period coincided with the end of the school term where considerable movement of people and subsequent variation from the resident population could be expected (i.e. recreational visitors and seasonal tourists, festivals or events). If the total observed variation was smaller than the complete uncertainty estimate of the method for an individual value, none of the observed variation could be attributed to ‘real’ variation.
3.3.
Estimated drug consumption per 1000 people
Table 5 and Fig. 1 show the estimated drug consumption per 1000 people based on both census data and ENCP over the 12-day monitoring period. The rank order of estimated drug consumption was: THC > methamphetamine z cocaine > MDMA. Consumption of cocaine, methamphetamine and MDMA generally increases during weekends and declines during weekdays. This pattern is more distinct when calculated with ENCP (Fig. 1B) instead of a single value based on census data (Fig. 1A) since the latter cannot account for day-to-day variation of the people in the catchment.
3.3.1.
Cocaine
The daily consumption of cocaine estimated from cocaine loads was slightly higher than the estimate based on its primary metabolite benzoylecgonine (average difference about 15%). Benzoylecgonine is expected to result in a more reliable estimation of cocaine consumption because of its higher persistence in wastewater and because it originates only from human excretion (van Nuijs et al., in press). The high market value of cocaine and the strong correlation with the benzoylecgonine pattern suggest that dumping of high levels of cocaine into the wastewater is unlikely. The observed difference may indicate that the fraction of excreted benzoylecgonine and/or cocaine used in our study was slightly elevated and/or declined respectively to some extent. This may be related to the route of administration (Cone et al., 1998). For example, users in Australia typically snort or inject cocaine (Stafford and Burns, 2010), whereas there is more ‘crack’ (smoked) cocaine use in North America (DASIS, 2007).
3.3.2.
Amphetamine and methamphetamine
We recognize that both amphetamine and methamphetamine consumption results in the excretion of amphetamine (Baselt, 2008). Police seizure data indicate that the use of amphetamine is negligible in Australia (ACCR, 2010). Hence, we assumed that the measured trace amounts of amphetamine
in the samples arise from methamphetamine consumption. This is substantiated by the fact that methamphetamine loads are consistent when back calculated from either amphetamine or methamphetamine.
3.3.3.
THC
THC showed much less daily variation over the period. We also found that its metabolite THCeCOOH did not correlate with other illicit drug and opioid residues, opioids or even prescription pharmaceuticals and acesulfame (Table S3). The lack of inter-day variability in THC in the wastewater results could be more related to the long excretion half-life of THC and its metabolites (range 0.8e9.8 days, average 3 days) (Baselt, 2008) than the actual weekly usage pattern of THC.
3.3.4.
Comparison with other studies
To our knowledge this study provides the first published data on the consumption of illicit drugs estimated from wastewater analysis in Queensland, Australia. Irvine et al. (in press) have recently measured three illicit drug residues in metropolitan and rural areas in the State of South Australia. While MDMA and methamphetamine loads in our catchment were similar, benzoylecgonine loads were about five to six times higher. In general the observed weekly pattern of illicit drug use was similar to the ones described in other studies in Europe and North America (e.g., Huerta-Fontela et al., 2008; Karolak et al., 2010; Terzic et al., 2010; van Nuijs et al., 2009b; Zuccato et al., 2008). However, the absolute average levels of individual illicit drug loads per 1000 people vary (e.g. lower benzoylecgonine).
3.4.
Uncertainty assessment
To evaluate the results it is important to communicate the expected uncertainty associated with the estimations (Ort et al., 2010a). As can be seen from Table 4 the uncertainty due to unknown systematic errors in flow measurements (UF) dominate the uncertainty estimates for the chemical loads in sewers in our study. This uncertainty can only be reduced with a conscientious calibration of flow meters. The uncertainty due to a wide range of reported excretion rates for individual users may suggest that the back calculation for consumed masses in the catchment is subject to even higher uncertainty. This uncertainty decreases with the number of users. In our catchment we estimate more than 100 users for each illicit drug (Table 4) which implies a reduction by at least a factor of 10 [O(100)] for the uncertainty of the ‘combined’ excretion rate by many users. However, systematic errors are not compensated by a large number of users. The limitation of the excretion data available in the literature have been recently reviewed and highlighted (van Nuijs et al., in press). For example, the published clinical data originate from urinary analysis of a very limited number of young or healthy men and adults (Ambre et al., 1984, 1988; Baselt, 2008; Cone et al., 2003). Furthermore, poly-consumption of drugs (i.e. drugedrug interactions) and different administration routes such as intravenous, intranasal, smoked and/or the combination with alcohol can result in a higher or lower excretion rate which is also not taken into account. To date comprehensive data on biodegradation of illicit drugs in sewers is missing. It must be assumed that
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 3 7 e4 4 4 8
4445
Fig. 1 e Daily consumption of illicit drugs per 1000 people over a 12-day monitoring period. A: using census data (300,000 persons) to normalize for population size (error bars in Fig. 1A indicate total uncertainty for the estimated per capita consumption (UMcatchment)) and B: using ENCP (estimated number of contributing people based on atenolol loads (see text for more details). The error bars in Fig. 1B indicate the total uncertainty for the estimated per capita consumption (UMENCP, see Table 4 and Eq. 4.2). Gray shadows indicate weekend days. COC: cocaine; COCBE: cocainebenzoylecgonine; MA: methamphetamineamphetamine; MAAM: methamphetamineamphetamine. The subscripts in cocainebenzoylecgonine and methamphetamineamphetamine imply that the consumed load of cocaine was back calculated from the metabolites benzoylecgonine and methamphetamine from its metabolite amphetamine. biodegradation leads to a systematic underestimation of Mcatchment. The effects of temperature, pH, hydraulic residence time, biofilm, total suspended solids, etc. have not been assessed yet and therefore we cannot assign a justified value
for UB at this stage. However, we assume that the effect of biodegradation is more or less constant within a given sewer system and over a short sampling period (i.e. days) and that inter-day variability is negligible. This may not hold true when
4446
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 3 7 e4 4 4 8
data among different locations or within a location over a longer time span (i.e. year, seasonal effects) is compared. The uncertainty of the estimated number of people contributing to the wastewater is in our case the uncertainty of the estimated mass of consumed atenolol (UENCP ¼ 24%). Since census data only accounts for the permanent residential population and not for tourists or commuters, a meaningful, independent direct comparison is difficult. However, the census data (300,000e350,000 people) is within the 95% confidence interval of observed ENCP (370,000) and observed coefficient of variation. Another systematic underestimation might be caused by exfiltration: through leaky sewers causing a part of the wastewater collected in the catchment to never arrive at the inlet of the STP under investigation (Rieckermann et al., 2007).
3.5.
Potential strengths, limitations and alternatives
The advantage of estimating the number of people in the catchment based on a parameter in the collected wastewater samples is two-fold: 1) it can account for day-to-day variation; and 2) systematic errors due to flow measurement uncertainties can be avoided with this normalization approach (Table 2, Eq. 4.1, 4.2). This also holds true qualitatively for effects due to wastewater losses (exfiltration through leaky sewers) and potentially even to biodegradation. With atenolol we suggest a compound for which sales data are known and which is regularly taken by a relatively large fraction in the Australian population (1 in 30e100). To our knowledge this is the first paper objectively assessing methodological uncertainty vs. observed variation: over the 12-day period all observed variation in illicit drug use was larger than the methodological uncertainty (see Table S5, Supplementary Information). We recognize two aspects as limitations of this approach: 1) degree of patient compliance; and 2) stability of compounds. Furthermore, sampling errors must not only be minimized for the determination of illicit drug loads, but also for any substance used to calculate the ENCP to keep the overall uncertainties low. In view of these strengths and limitations we suggest using multiple wastewater quality parameters (beyond other prescription pharmaceuticals), which might be adequate in other countries, to calculate the ENCP: acesulfame is one of them. Unfortunately, the consumption data were not available. However, we still could use it as a normalization factor to assess day-to-day variation, but not from month to month or between different locations, since consumption could be different and change over these time frames. The advantage of using traditional water quality parameters such as biological or chemical oxygen demand (BOD, COD), total nitrogen (Ntot) or phosphorus (P) is that they are often measured routinely by the STP operators. In addition to the limitations outlined above it has to be noted that human excretion is not the only source for BOD/COD/Ntot and P. A study in Switzerland showed an average industrial contribution in the range of 10e40% (depending on the parameter and assumed per capita excretion, unpublished data). Besides knowing the type and level of industry in the catchment one would also have to know their discharge pattern. If it is not accounted
for, it is expected to result in an underestimation of the per capita consumption of illicit drugs during the week (because increased industrial pollutant loads would suggest a higher number of people in the catchment) and a relative overestimation of the weekend consumption if industry is not active during the weekend. NH4eN, which is more exclusively from human excretion might be the parameter which is least affected by industry.
4.
Conclusions
Estimating the number of people contributing to a sample: Among five prescription pharmaceuticals atenolol was found to be the most appropriate candidate to estimate the number of people contributing to the sampled wastewater in the studied catchment. Acesulfame concentration could be used to estimate population and may be a better basis for calculations given its demographic relevance to illicit drug consumption. Unfortunately national consumption figures for this artificial sweetener were not available. We suggest the development of a method based on a combination of prescription pharmaceutical loads, acesulfame loads and conventional parameters for the more precise and reliable estimation of the number of people contributing to a specific wastewater sample. Estimating illicit drug use: We estimated the daily consumption per 1000 persons using both census data and the estimated number of contributing people in the wastewater derived from atenolol loads, respectively: 623e1370 mg and 364e1410 mg (cannabis (THC)); 176e531 mg and 129e614 mg (cocaine predicted based on cocaine), 109e450 mg and 98.8e521 mg (cocaine predicted based on benzoylecgonine); 146e298 mg and 110e345 mg (methamphetamine predicted by methamphetamine), 161e312 mg and 123e360 mg (methamphetamine predicted by amphetamine); 76.6e297 mg and 44.7e343 mg (ecstasy (MDMA)) in the studied catchment. Higher consumption was found at weekends, which was similar to other studies. For both cocaine and methylamphetamine estimations of the daily consumption based on either the load of the parent drug itself or its metabolites were in good agreement. Uncertainty assessment: Our study describes an approach to estimate the number of people contributing to the wastewater stream and how to quantify the methodological uncertainty. The uncertainty components include sampling, chemical analysis, flow measurement, excretion fraction and estimated number of people that contributed to the wastewater. With current best practice we determined a remaining uncertainty applicable to an estimate of consumed chemical mass to be about 20e30% at the study catchment while that of per capita drug consumption to be about 14e24% (using our method to estimate the number of contributing people). In other words, any difference or change in drug loads smaller than this uncertainty cannot be considered to be significant. This is important for all kinds of studies (i.e. screening, policy, intelligence, operational purposes) and particularly to evaluate the success of intervention strategies aiming at a reduction of drug consumption in a given community or institution (e.g. prison, school).
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 3 7 e4 4 4 8
Acknowledgments The authors sincerely thank Mr. Jake O’Brien (Entox) and Mr. Jack Thompson (Entox) for assisting with sampling and SouthEast Queensland Council for sampling opportunities. We also acknowledge the assistance of Dr. Shalona Anuj-Outten (QHFSS) for instrumental training. Vanna Mabbott and Chris Raymond from The Drug Utilisation Sub-Committee of the Pharmaceutical Benefits Advisory Committee, Department of Health and Ageing, Commonwealth of Australia (providing Australian drug-use statistics). Financial assistance for this research was from the QHFSS/Entox Collaborative Research Funds and the Australian Future Forensics Innovation Network as supported by the Department of Employment, Economic Development and Innovation, through the National and International Research Alliance Program. Wayne Hall is funded by an NHMRC Australia Fellowship. Entox is a joint venture of The University of Queensland and QHFSS. We also thank Dr. Margaret Murphy (City University of Hong Kong) and Dr. Jo¨rg Rieckermann (Eawag) for valuable input and discussion.
Appendix. Supplementary material Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.watres.2011.05.042.
references
Abraham, T.T., Barnes, A.J., Lowe, R.H., Kolbrich Spargo, E.A., Milman, G., Pirnay, S.O., Gorelick, D.A., Goodwin, R.S., Huestis, M.A., 2009. Urinary MDMA, MDA, HMMA, and HMA excretion following controlled MDMA administration to humans. J. Anal. Toxicol. 33, 439e446. ABS (Australian Bureau of Statistics), 2006. 4821.0.55.001 e Cardiovascular Disease in Australia: a Snapshot, 2004e05. ABS, Canberra. Available. http://www.abs.gov.au/ausstats/
[email protected]/mf/4821.0.55.001. ABS (Australian Bureau of Statistics), 2009. 3201.0 e Population by Age and Sex, Australian States and Territories, Jun 2009. ABS, Canberra. Available. http://www.abs.gov.au/Ausstats/abs@. nsf/mf/3201.0. ACCR (Australian Crime Commission Report), 2010. Illicit Drug Data Report 2008e09. Available. http://www.crimecommissi on.gov.au/publications/iddr/index.htm. AIHW (Australian Institute of Health and Welfare), 2008. 2007 National Drug Strategy Household Survey: Detailed Findings. Drug Statistics Series No. 22. Cat. No. PHE 107. AIHW, Canberra. Available. http://www.aihw.gov.au/publications/ index.cfm/title/10674. Ambre, J., Fischman, M., Ruo, T., 1984. Urinary excretion of ecgonine methyl ester, a major metabolite of cocaine in humans. J. Anal. Toxicol. 8, 23e25. Ambre, J., Ruo, T., Nelson, J., Belknap, S., 1988. Urinary excretion of cocaine, benzoylecgonine and ecgonine methyl ester in humans. J. Anal. Toxicol. 12, 301e306. AMH (Australian Medicines Handbook), 2010. The Australasian Society of Clinical and Experimental Pharmacologists and Toxicologists, the Pharmaceutical Society of Australia and the
4447
Royal Australian College of General Practitioners. July 2010 Edition. Available 1998. http://amh.hcn.net.au/view.php. Baselt, R.C., 2008. Disposition of Toxic Drugs and Chemicals in Man, eighth ed., Biomedical Publications, Foster City, CA. Boleda, M.R., Galceran, M.T., Ventura, F., 2009. Monitoring of opiates, cannabinoids and their metabolites in wastewater, surface water and finished water in Catalonia, Spain. Water Res. 43, 1126e1136. Boleda, M.R., Galceran, M.T., Ventura, F., 2007. Trace determination of cannabinoids and opiates in wastewater and surface waters by ultra-performance liquid chromatographytandem mass spectrometry. J. Chromatogr. A 1175, 38e48. Bones, J., Thomas, K.V., Paull, B., 2007. Using environmental analytical data to estimate levels of community consumption of illicit drugs and abused pharmaceuticals. J. Environ. Monit. 9, 701e707. Buerge, I.J., Buser, H.R., Kahle, M., Mu¨ller, M.D., Poiger, T., 2009. Ubiquitous occurrence of the artificial sweetener acesulfame in the aquatic environment: an ideal chemical marker of domestic wastewater in groundwater. Environ. Sci. Technol. 43, 4381e4385. Castiglioni, S., Zuccato, E., Chiabrando, C., Fanelli, R., Bagnati, R., 2008. Mass spectrometric analysis of illicit drugs in wastewater and surface water. Mass Spectrom. Rev. 27, 378e394. Castiglioni, S., Zuccato, E., Crisci, E., Chiabrando, C., Fanelli, R., Bagnati, R., 2006. Identification and measurement of illicit drugs and their metabolites in urban wastewater by liquid chromatographyetandem mass spectrometry. Anal. Chem. 78, 8421e8429. Chiaia, A.C., Banta-Green, C., Field, J., 2008. Eliminating solid phase extraction with large volume injection LC/MSMS: analysis of illicit and legal drugs and human urine indicators in US wastewaters. Environ. Sci. Technol. 42, 8841e8848. Collins, D.J., Lapsley, H.M., 2008. The Costs of Tobacco, Alcohol and Illicit Drug Abuse to Australian Society in 2004/05, ISBN 1-74186437-2 Available. http://www.health.gov.au/internet/drugstrate gy/publishing.nsf/Content/34F55AF632F67B70CA2573F60 005D42B/$File/mono64.pdf Online. Cone, E.J., Sampson-Cone, A.H., Darwin, W.D., Huestis, M.A., Oyler, J.M., 2003. Urine testing for cocaine abuse: metabolic and excretion patterns following different routes of administration and methods for detection of false-negative results. J. Anal. Toxicol. 27, 386e401. Cone, E.J., Tsadik, A., Oyler, J., Darwin, W.D., 1998. Cocaine metabolism and urinary excretion after different routes of administration. Ther. Drug Monit. 20, 556e560. Daughton, C.G., 2001. Illicit drugs in municipal sewage: proposed new non-intrusive tool to heighten public awareness of societal use of illicit/abused drugs and their potential for ecological consequences. In: Daughton, C.G., Jones-Lepp, T.L. (Eds.), Pharmaceuticals and Personal Care Products in the Environment. Scientific and Regulatory Issues. American Chemical Society, Washington, DC, pp. 348e364. Symposium Series 791. Daughton, C.G., 2011. Illicit drugs: contaminants in the environment and utility in forensic epidemiology. Rev. Environ. Contam. Toxicol. 210, 59e110. Degenhardt, L., Hall, W., Lynskey, M., Warner-Smith, M., 2004. Illicit drug use. In: Ezzati, M., Lopez, A.D., Rodgers, A., Murray, R. (Eds.)Comparative Quantification of Health Risks: Global and Regional Burden of Disease Attributable to Selected Major Risk Factors, second ed., chapter 13. World Health Organization, Geneva, pp. 1109e1176. DASIS (Drug and Alcohol Services Information System), 2007. The DASIS Report: Cocaine Route of Administration Trends: 1995e 2005. Substance Abuse and Mental Health Services Administration, Office of Applied Studies, Rockville, MD. Available. http://www.oas.samhsa.gov/2k7/crackTX/crackTX.pdf.
4448
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 3 7 e4 4 4 8
Grotenhermen, F., 2003. Pharmacokinetics and pharmacodynamics of cannabinoids. Clin. Pharmacokinet. 42, 327e360. Howell, S.R., Husbands, G.E.M., Scatina, J.A., Sisenwine, S.F., 1993. Metabolic disposition of 14C-venlafaxine in mouse, dog, rhesus money and man. Xenobiotica 23, 349e359. Huerta-Fontela, M., Galceran, M.T., Ventura, F., 2007. Ultraperformance liquid chromatography-tandem mass spectrometry analysis of stimulatory drugs of abuse in wastewater and surface waters. Anal. Chem. 79, 3821e3829. Huerta-Fontela, M., Galceran, M.T., Martin-Alonso, J., Ventura, F., 2008. Occurrence of psychoactive stimulatory drugs in wastewaters in north-eastern Spain. Sci. Total Environ. 397, 31e40. Irvine, R.J, Kostakis, C, Felgate, P.D, Jaehne, E.J, Chen, C, White, J. M. Population drug use in Australia: a wastewater analysis. Forensic Sci. Int. in press, doi:10.1016/j.forsciint.2011.01.037. Karolak, S., Nefau, T., Bailly, E., Solgadi, A., Levi, Y., 2010. Estimation of illicit consumption by wastewater analysis in Paris area. Forensic Sci. Int. 200, 153e160. Lienert, J., Gudel, K., Escher, B.I., 2007. Screening method for ecotoxicological hazard assessment of 42 pharmaceuticals considering human metabolism and excretory routes. Environ. Sci. Technol. 41, 4471e4478. Metcalfe, C., Tindale, K., Li, H., Rodayan, A., Yargeau, V., 2010. Illicit drugs in Canadian municipal wastewater and estimates of community drug use. Environ. Pollut. 158, 3179e3185. Mie`ge, C., Choubert, J.M., Ribeiro, L., Euse`be, M., Coquery, M., 2009. Fate of pharmaceuticals and personal care products in wastewater treatment plants e conception of a database and first results. Environ. Pollut. 157, 1721e1726. Onesios, K.M., Yu, J.T., Bouwer, E.J., 2009. Biodegradation and removal of pharmaceuticals and personal care products in treatment systems: a review. Biodegradation 20, 441e466. Ort, C., Lawrence, M.G., Reungoat, J., Eaglesham, G., Carter, S., Keller, J., 2010a. Determining the fraction of pharmaceutical residues in wastewater originating from a hospital. Water Res. 44, 605e615. Ort, C., Lawrence, M.G., Reungoat, J., Mueller, J.F., 2010b. Sampling for PPCPs in wastewater systems: comparison of different sampling modes and optimization strategies. Environ. Sci. Technol. 44, 6289e6296. Ort, C., Lawrence, M.G., Rieckermann, J., Joss, A., 2010c. Sampling for pharmaceuticals and personal care products (PPCPs) and illicit drugs in wastewater systems: are your conclusions valid? A critical review. Environ. Sci. Technol. 44, 6024e6035. Prichard, J., Ort, C., Bruno, R., Gartner, C., Kirkbride, P., Hall, W., Lai, F.Y., Carter, S., Mueller, J., Salinas, A., 2010. Developing a method for site-specific wastewater analysis: implications for prisons and other agencies with an interest in illicit drug use. J. Law Inf. Sci. 20, 15e27. Postigo, C., Lopez de Alda, M.J., Barcelo, D., 2008. Analysis of drugs of abuse and their human metabolites in water by LC-MS/MS: a non-intrusive tool for drug abuse estimation at the community level. Trends Anal. Chem., 1053e1069.
Postigo, C., Lopez de Alda, M.J., Barcelo, D., 2010. Drugs of abuse and their metabolites in the Ebro river basin: occurrence in sewage and surface water, sewage treatment plants removal efficiency, and collective drug usage estimation. Environ. Int. 36, 75e84. Rieckermann, J., Bare s, V., Kracht, O., Braun, D., Gujer, W., 2007. Estimating sewer leakage from continuous tracer experiments. Water Res. 41 (9), 1960e1972. Shand, F., Topp, L., Darke, S., Makkai, T., Griffiths, P., 2003. The monitoring of drug trends in Australia. Drug Alcohol Rev. 22, 63e74. Stafford, J., Burns, L., 2010. Australian Drug Trends 2009. Findings from the Illicit Drug Reporting System (IDRS). Australian Drug Trend Series No. 37. National Drug and Alcohol Research Centre, University of New South Wales, Sydney. Available. http://www.med.unsw.edu.au/NDARCWeb.nsf/page/National %25%20Reports. Terzic, S., Senta, I., Ahel, M., 2010. Illicit drugs in wastewater of the city of Zagreb (Croatia) e estimation of drug abuse in a transition country. Environ. Pollut. 158, 2686e2693. Thomann, M., 2008. Quality evaluation methods for wastewater treatment plant data. Water Sci. Technol. 57 (10), 1601e1609. UNODC (United Nations Office on Drugs and Crime), 2010. World Drug Report. Methodology. Available. http://www.unodc.org/ unodc/en/data-and-analysis/WDR-2010.html. van Nuijs, A.L.N., Pecceu, B., Theunis, L., Dubois, N., Charlier, C., Jorens, P.G., Bervoets, L., Blust, R., Neels, H., Covaci, A., 2009a. Cocaine and metabolites in waste and surface water across Belgium. Environ. Pollut. 157, 123e129. van Nuijs, A.L.N., Pecceu, B., Theunis, L., Dubois, N., Charlier, C., Jorens, P.G., Bervoets, L., Blust, R., Neels, H., Covaci, A., 2009b. Spatial and temporal variations in the occurence of cocaine and benzoylecgonine in waste- and surface water from Belgium and removal during wastewater treatment. Water Res. 43, 1341e1349. van Nuijs, A.L.N., Pecceu, B., Theunis, L., Dubois, N., Charlier, C., Jorens, P.G., Bervoets, L., Blust, R., Meulemans, H., Neels, H., Covaci, A., 2009c. Can cocaine use be evaluated through analysis of wastewater? A nation-wide approach conducted in Belgium. Addiction 104, 734e741. van Nuijs, A.L.N., Castiglioni, S., Tarcomnicu, I., Postigo, C., de Alda, M.L., Neels, H., Zuccato, E., Barcelo, D., Covaci, A. Illicit drug consumption estimations derived from wastewater analysis: a critical review. Sci. Total Environ., in press. van Nuijs, A.L.N., Mougel, J.F., Tarcomnicu, I., Bervoets, L., Blust, R., Jorens, P.G., Neels, H., Covaci, A., 2011. Sewage epidemiology e a real-time approach to estimate the consumption of illicit drugs in Brussels, Belgium. Environ. Int. 37, 612e621. Zuccato, E., Chiabrando, C., Castiglioni, S., Bagnati, R., Fanelli, R., 2008. Estimating community drug abuse by wastewater analysis. Environ. Health Persp. 116, 1027e1032. Zuccato, E., Chiabrando, C., Castiglioni, S., Calamari, D., Bagnati, R., Schiarea, S., et al., 2005. Cocaine in surface waters: a new evidence-based tool to monitor community drug abuse. Environ. Health 4, 14e20.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 4 9 e4 4 5 8
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Fulvic acid mediated photolysis of ibuprofen in water Laura E. Jacobs a,1, Ryan L. Fimmen b, Yu-Ping Chin b,*, Heath E. Mash c, Linda K. Weavers d a
Environmental Science Graduate Program, The Ohio State University, Columbus, OH 43210, USA School of Earth Sciences, The Ohio State University, 125 South Oval Mall, 275 Mendenhall Laboratory, Columbus, OH 43210, USA c United States Environmental Protection Agency, 26 W. Martin Luther King Drive, Cincinnati, OH 45268, USA d Department of Civil and Environmental Engineering and Geodetic Science, The Ohio State University, 470 Hitchcock, 2070 Neil Avenue, Columbus, OH 43210, USA b
article info
abstract
Article history:
Photolysis of the non-steroidal anti-inflammatory drug ibuprofen was studied by exposure
Received 22 February 2011
to a solar simulator in solutions of fulvic acid (FA) isolated from Pony Lake, Antarctica;
Received in revised form
Suwannee River, GA, USA; and Old Woman Creek, OH, USA. At an initial concentration of
22 May 2011
10 mM, ibuprofen degrades by direct photolysis, but the presence of FA significantly increases
Accepted 28 May 2011
reaction rates. These reactions proceeded up to 6 faster in FA solutions at lower ibuprofen
Available online 7 June 2011
concentrations (0.1 mM), but the rates are highly dependent upon DOM composition. Incomplete quenching of the reaction in the presence of isopropanol suggests that the
Keywords:
hydroxyl radical is only partially responsible for ibuprofen’s photodegradation in FA solu-
Pharmaceuticals
tions, and other reactive transients likely play an important role. Liquid chromatography-
Ibuprofen
quadrupole time-of-flight mass spectrometry and NMR spectroscopy reveal the formation of
Photolysis
multiple photoproducts, with three byproducts identified as 1-(4-isobutylphenyl)ethanol,
Dissolved organic matter
isobutylacetophenone, and a phenol derivative. Pony Lake FA significantly increases the
Photodecarboxylation
production of the major byproduct relative to yields produced by direct photolysis and the
Fulvic acids
other two FA. Thus, the photolytic fate of ibuprofen in sunlit waters is affected by its initial concentration and the source of dissolved organic matter present. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
Pharmaceuticals and personal care products (PPCPs) have been detected in natural waters (e.g., wetlands, lakes, and rivers) throughout the world and are widely recognized as emerging environmental contaminants (Ternes, 1998; Kolpin et al., 2002; Daughton and Ternes, 1999; Nikolaou et al., 2007; plus many others). While many PPCPs (but not all) can be efficiently removed in wastewater treatment plants (from <50% for the more refractory compounds to >90%) these compounds can enter waterways from treated sewage effluent, septic systems and combined sewer overflows (Winker et al., 2008; Lindqvist et al., 2005; Nakada et al., 2006,
2007). Most of these compounds exist at low levels (sub part per billion) and exhibit relatively low acute toxicity (EC50 > 1 mg/L), but little is known about synergistic effects i.e., mixtures and long-term chronic exposure of these substances to aquatic organisms (Fent et al., 2006). Moreover, the toxicological impacts of their unidentified derivatives have not been well studied. In 1999/2000 a United States Geological Survey study detected the pharmaceutical ibuprofen in w10% of 139 streams sampled across the United States at concentrations as high as 5 nM (Kolpin et al., 2002). Ibuprofen is a non-steroidal anti-inflammatory drug (NSAID) and is one of the most widely used over-the-counter medications. It is an alkylbenzene with a carboxylic acid
* Corresponding author. Tel.: þ1 614 292 6953; fax: þ1 614 292 7688. E-mail address:
[email protected] (Y.-P. Chin). 1 Present address: The National Academies, 500 5th St. NW, Washington DC 20001, USA. 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.05.041
4450
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 4 9 e4 4 5 8
functional group, a pKa of 4.8, and a log n-octanol/water partition coefficient (Kow) of 2.48 at pH 7 (Scheytt et al., 2005). Ibuprofen is chiral and its pharmacological activity is attributable to the S-isomer, but is administered as a racemic mixture (Hutt and Caldwell, 1983). Over 70% of ibuprofen is metabolized in the body and excreted in urine. Hydroxylated and carboxylated compounds are the predominant metabolized forms (Hutt and Caldwell, 1983). Despite this loss mechanism, levels of ibuprofen as high as 0.015 mM have been detected in wastewater influent (Buser et al., 1999). NSAIDs, as well as many other classes of pharmaceuticals, are known to degrade by both photochemical and biological processes (Boreen et al., 2003; Packer et al., 2003). With respect to ibuprofen, it is rather efficiently removed in wastewater treatment plants (Winker et al., 2008; Lindqvist et al., 2005; Nakada et al., 2006, 2007), but the rates of ibuprofen degradation vary depending upon the microbial consortia and processes involved e.g., biofilms, membranes, etc. Nonetheless our interpretation of the literature reveals that the biodegradation kinetics of ibuprofen is similar to or slower than photochemical processes (Zwiener and Frimmel, 2003; Quintana et al., 2005; Xu et al., 2009; Yu et al., 2006). Packer et al. (2003) showed that ibuprofen is susceptible to degradation by both direct and indirect photolytic pathways, whereby dissolved organic matter (DOM) was demonstrated to be an important photosensitizer. DOM and in particular the fulvic acid (FA) fraction can generate reactants such as reactive oxygen species (ROS) and triplet DOM (3DOM) (Gerecke et al., 2001; Miller and Chin, 2002; Vaughan and Blough, 1998), which can subsequently transform ibuprofen. Fulvic acids represent the chromophoric portion of DOM and as a result are the most photochemically active fraction (Aguer et al., 1997; Brown et al., 2004). Packer et al. (2003) also showed that the indirect photolysis of ibuprofen occurs in part through reaction with hydroxyl radicals (OH) generated from excited-state DOM. The OH mediated pathway, however, does not explain all the observed reactivity and an additional reactive transient of unknown origin also contributes to the photodegradation of ibuprofen in natural waters (Packer et al., 2003). Several studies have identified a number of PPCP derivatives resulting from photochemical reactions in aqueous media and proposed chemical mechanisms that form these derivatives (Edhlund et al., 2006; Boreen et al., 2005; Pe´rezEstrada et al., 2005). Indeed, light-induced photodecarboxylation of NSAIDs and the derivatives formed from this pathway have been studied in both non-aqueous (Castell et al., 1987; Budac and Wan, 1992; Bosca´ et al., 1994, 2001; Monti et al., 1997) and aqueous systems (Pe´rez-Estrada et al., 2005). This process is a mechanism whereby a benzylic radical is formed followed by the subsequent ejection of CO2 and a solvated electron. This electron can then be scavenged by triplet molecular oxygen (O2), generating superoxide (,O 2) leading to a series of oxygenated byproducts (Bosca´ et al., 1994). Castell et al. (1987) proposed the same mechanism for the decomposition of ibuprofen when photolyzed by a medium pressure Hg lamp in methanol. This irradiation gave rise to several byproducts via light-induced excitation of ibuprofen, which cleaves the CeC bond a to the carboxyl group generating a benzylic radical intermediate followed by
hydrogen abstraction, dimerization, methanol addition, or reaction with dioxygen to generate various byproducts. This study examines the mechanism(s) responsible for the photochemical degradation of ibuprofen in the aquatic environment and the effect of DOM (specifically aquatic fulvic acids) on this process. We studied the direct and indirect photolysis of ibuprofen in the presence of fulvic acids that represent the continuum of DOM derived from terrestrial and microbial processes. We chose to use fulvic acids rather than whole water DOM because the manner in which they were processed (by XAD chromatography from a single sampling event) rendered them temporally stable i.e., experiments can be repeated using the same batch of freeze-dried FA. In addition we examined how the initial concentration of ibuprofen influences indirect photolysis in the presence of DOM. Finally, we examine the chemical mechanism of ibuprofen degradation through the identification of its photoderivatives and an analysis of byproduct production in the presence of various fulvic acids.
2.
Methods
2.1.
Chemicals and fulvic acids
Milli-Q (18 MU-cm, Millipore) water was used for all aqueous solutions. All solvents (acetonitrile, hexane, methanol, isopropanol and ether) and chemicals (hydrochloric acid, isobutylacetophenone, etc.) were reagent-grade or higher. CDCl3 (99.8% -D) and ibuprofen (racemic mixture) with a purity of 99% were obtained from Acros Organics. Suwannee River fulvic acid (SRFA) was obtained from the International Humic Substances Society and Pony Lake fulvic acid (PLFA) provided by Dr. Diane McKnight. Additionally, we isolated a fulvic acid from Old Woman Creek, a National Estuarine Research Reserve wetland in Huron, OH, using the XAD-8 protocol (Leenheer, 1981; Thurman and Malcolm, 1981), which we refer to in this paper as OWCFA.
2.2.
Photolytic reactions
Ibuprofen stock solutions were prepared in acetonitrile. Experimental solutions of 10 mM or 0.1 mM were prepared by addition of an appropriate volume of stock, complete evaporation of the acetonitrile, and re-constitution into the desired aqueous matrix. The pH of the solutions was adjusted to 7 (0.1) with HCl and/or NaOH. Some solutions were prepared anoxically by either sparging the sample with argon for 1 min/mL and transferring it to photolysis tubes within a glove box (95% N2/5% H2) or subjecting solutions to 3 freeze-pump-thaw cycles in a Schlenk line followed by cannula transfer into photolysis tubes. After this treatment, oxygen levels in photolysis solutions were below the limit of detection as determined by both a dissolved oxygen probe and chemical indicators. To investigate the role of DOM on the indirect photolysis of ibuprofen, each solution was spiked with the same amount of the desired fulvic acid (by weight) and an aliquot was used for total organic carbon analysis (Shimadzu TOC-5000). Airtight quartz tubes (1 cm path length capped with Teflon-lined quartz lids) were used for the photolysis experiments.
4451
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 4 9 e4 4 5 8
Photolysis of ibuprofen was conducted using a solar simulator (Atlas Suntest CPSþ) fitted with a 500 W Xenon arc lamp at a temperature of 25 C and a duration equivalent to at least one half-life. Selected photolysis experiments were evaluated over three half-lives in order to determine the appropriate rate model for degradation kinetics. For experiments where the photolytic rates were very slow e.g., direct photolysis and indirect experiments in the presence of scavengers we did not exceed 50 h irradiation time. Chemical actinometry ( p-nitroanisole/pyridine) (Dulin and Mill, 1982) was conducted in both the solar simulator and under natural sunlight in Columbus, OH (40 40 41" N, 83 40 24" W) in June at noon. Actinometry showed no significant change in the light intensity of our solar simulator over the course of experiments and the difference in p-nitroanisole rate constants between our light source and sunlight under the conditions described above was a factor of 3.7 i.e., our light source was 3.7 times more intense than June sunlight at noon in the absence of clouds in Columbus, OH. Dark controls were run concurrently in foil-covered tubes. To ascertain the role of OH in our experiments, we used isopropanol (20 mM) as a hydroxyl radical scavenger (Packer et al., 2003). Temperature, pH, and radiometer readings were monitored during each experiment, with all three parameters remaining consistent. Finally, the procedure of Zhou and Mopper (1990) was used to quantify hydroxyl radical production in sample solutions, and is further described in the Supplemental Information (SI) file.
2.3.
Analysis of ibuprofen and its photoderivatives
Ibuprofen and its photolysis byproducts were analyzed by UVeVisible spectrophotometry (Varian Cary 13) and by highpressure liquid chromatography (HPLC) using a UVeVisible dual wavelength detector at l ¼ 223 nm and l ¼ 264 nm respectively (Waters Corp. 2487, Breeze 3.3 software). Aqueous
samples were directly injected into the HPLC (50 mL) and analytes separated using a Restek C-18 reverse-phase chromatography column. Each sample was eluted isocratically at 1 ml min1 using a 60% acetonitrile and 40% water mobile phase (v/v) buffered with phosphate to pH 3. Photolysis byproducts from experiments ([Ibuprofeno] ¼ 10 mM) were further analyzed using proton nuclear magnetic resonance (1H NMR) spectroscopy in both 1-D and 2-D mode, liquid chromatography-quadrupole time-of-flight mass spectrometry (LC-Q-TOFMS), and gas chromatography mass spectrometry (GCeMS). A detailed description of the methods employed is described in the SI.
3.
Results and discussion
3.1.
Photolysis experiments
Ibuprofen (10 mM initial concentration) degrades very slowly by direct photolysis, with a pseudo first order rate constant of 0.0025 h1 (Table 1). We observed no statistical difference between ibuprofen as a racemic mixture or its “S” isomer (Table 1). This translates to a half-life greater than 200 h, and is nearly identical to the half-life reported by Lin and Reinhard (2005) who used the same light source and faster than in natural sunlight (600 h) as reported by Yamamoto et al. (2009). Ibuprofen degrades approximately 5 times faster in the presence of each fulvic acid than by direct photolysis (Table 1: Fig. 1). We attribute this slow direct photolytic rate to its low absorbance at wavelengths present in sunlight (>290 nm: Fig. 2), and the observed increase in its reaction kinetics in the presence of DOM can be attributed to indirect pathways. Surprisingly, the photolytic rate constants in the presence of each fulvic acid are statistically identical even though they are derived from different precursor materials i.e., higher plants vs. bacteria and phytoplankton. Small differences do occur if one normalizes the rate constants to DOC whereby the
Table 1 e 10 mM and 0.1 mM racemic and S-(D) (when indicated) ibuprofen degradation rate constants (kobs hL1) in the presence of simulated sunlight, OWCFA (Old Woman Creek Fulvic Acid), SRFA (Suwannee River Fulvic Acid), and PLFA (Pony Lake Fulvic Acid) at indicated dissolved organic carbon levels (DOC), in the absence of molecular oxygen, and in the presence of isopropanol. (L) Denotes not applicable. [Racemic Ibuprofen]0 ¼ 10 mM
Direct Direct (S-(þ)-Ibuprofen) OWCFA SRFA PLFA SRFA low O2 OWCFA low O2 [Racemic Ibuprofen]0 ¼ 0.1 mM
OWCFA SRFA PLFA SRFA low O2 OWCFA low O2 PLFA low O2
[DOC] mgC/L
kobs (h1)
Half-lives (hours)
Fulvic Acid þ 20 mM Isopropanol kobs (h1)
e e
0.0025 0.001 0.0031 0.001 0.015 0.001 0.016 0.001 0.014 0.001 0.017 0.002 0.019 0.001
277 224 46 43 50 41 36
e e 0.009 0.001 0.007 0.001 0.016 0.001 e e
5.56 7.20 5.45 7.20 5.56
0.18 0.20 0.19 0.20 0.18
[DOC] mgC/L
5.56 7.20 5.45 7.20 5.56 5.45
0.18 0.20 0.19 0.20 0.18 0.19
kobs (h1) 0.019 0.029 0.079 0.015 0.022 0.056
0.001 0.004 0.013 0.002 0.006 0.009
Fulvic Acid þ 20 mM Isopropanol kobs (h1) 36 24 9 46 32 12
0.011 0.002 0.020 0.004 0.013 0.006 e e e
4452
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 4 9 e4 4 5 8
Fig. 1 e Pseudo first order (ln(Concentration/Initial Concentration)) degradation of Ibuprofen (10 mM) direct photolysis degradation and in the presence of Old Woman Creek Fulvic Acid (OWCFA) with and without isopropanol.
transformation of ibuprofen by PLFA w OWCFA > SRFA. Nonetheless we believe that this observation is coincidental as experiments using chemical scavengers reveal significantly different photolytic pathways that appear to be dependent upon DOM composition. DOM can produce a plethora of reactive species, which varies with the fulvic acid composition (White et al., 2003; Hoigne´ et al., 1989; Cooper et al., 1989; Canonica et al., 1995; Canonica and Hoigne´, 1995; Canonica and Freiburghaus, 2001; Halladja et al., 2007, plus many others). Our choice of fulvic acids represents a wide diversity of compositional differences ranging from SRFA (derived from predominantly higher plant detritus) to PLFA (a microbially derived fulvic acid isolated from a hyper-eutrophic lake in Antarctica). Finally, OWCFA’s composition is comprised of organic matter derived from a combination of microbial and terrestrial precursors. The differences in chemical composition of all three FA are well documented in the literature
Fig. 2 e UVeVisible absorption spectrum of a 10 mM solution of ibuprofen.
(Thurman and Malcolm, 1981; Brown et al., 2004; Cawley et al., 2009) and are not presented in detail in this paper. As stated previously, we believe that the actual transformation pathways differed depending upon the fulvic acid used even though the degree of photosensitization appears to be composition independent. By using scavengers or altering experimental conditions we were able to elucidate specific pathways. To probe the role of hydroxyl radicals formed from photo-irradiated fulvic acids, the known OH scavenger, isopropanol, was added to photolysis experiments. While isopropanol is not the most ideal OH scavenger (due to its ability to react with other ROS) these data provide us with a rough estimate of the hydroxyl radical’s role in ibuprofen’s photofate. We observed significant decreases in the degradation rate of ibuprofen for OWCFA (kobs ¼ 0.009 0.001 h1; 46% reduction: see Fig. 1) and SRFA (k ¼ 0.007 0.001 h1; 64% reduction), corroborating observations by Packer et al. (2003) who used the same scavenger. Surprisingly, in PLFA, the addition of isopropanol does not change the rate of ibuprofen degradation (Fig. 3). Measurements of the steady state hydroxyl radical concentration ([OH]ss) corroborate our observations because photolyzed PLFA solutions (8.80 1017 M) yielded OH levels that are a factor of three lower than those measured for OWCFA (2.42 1016 M). Thus, other pathways are more important than OH for PLFA. Details for the determination of [OH]ss are found in SI. The role of ROS produced by irradiated organic matter was studied by photolyzing anoxic ibuprofen solutions in the presence of each fulvic acid. While the rate of ibuprofen degradation in the presence of SRFA remains statistically the same as oxic solutions, its kinetics was enhanced in OWCFA solutions (Table 1). We speculate that in the anoxic OWCFA system ibuprofen may also react via 3DOM oxidation in addition to OH pathways. Ibuprofen has an electron donating branched alkyl group that renders it susceptible to oxidation by 3DOM (Canonica and Freiburghaus, 2001). In the absence of dioxygen more 3DOM would be made available to react with ibuprofen thereby increasing its degradation rate.
Fig. 3 e Photoinduced pseudo first order (ln(Concentration/ Initial Concentration)) degradation of Ibuprofen (10 mM) in Milli-Q (Direct) photolysis and in the presence of Pony Lake Fulvic Acid (PLFA) with and without 20 mM isopropanol.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 4 9 e4 4 5 8
Unfortunately, the PLFA results were inconclusive (due to possible air contamination in the solutions and our limited amounts prevented us from repeating the experiment for this study) and are not included in this paper. Environmental levels of ibuprofen in surface waters are significantly smaller than the concentrations used in this and other studies. Because of ibuprofen’s electron donating alkyl group we investigated whether its indirect photolysis kinetics are influenced by its initial concentration. Canonica and Freiburghaus (2001) observed concentration effects for other electron rich probe molecules (alkyl and methoxy phenols) whereby the reaction kinetics of these compounds increases significantly (w 2 to 3) at low concentrations (0.1 mM vs. 5 mM) when irradiated in the presence of fulvic acids. We conducted similar experiments for all three fulvic acids at the lowest detectable concentration for ibuprofen that can be measured by our HPLC (Co ¼ 0.1 mM with an average detection limit of w0.02 mM). For all three fulvic acids the degradation rate constant increased compared to a Co of 10 mM concentration (Fig. 4). The increase in reaction kinetics varied among the fulvic acids where respective PLFA and SRFA rate constants increased by a factor of 5 and 2, while ibuprofen photodegradation in the presence of OWCFA exhibited the smallest enhancement (Table 1, Fig. 4). Canonica and Freiburghaus (2001) were able to show that electron rich aromatic compounds reacted with DOMgenerated photo-oxidants of various lifetimes. At high concentrations degradation is dominated by short-lived species such as 3DOM while at low concentrations reaction kinetics reflect the combined effect of short-lived (2 ms) species and long-lived species (>>2 ms, but variable by several orders of magnitude depending upon the species), which may include peroxyl, oxyl, or phenoxyl radicals, and excited states of DOM chromophoric constituents. Our results for SRFA and PLFA are consistent with those reported by others (Canonica and Hoigne´, 1995; Richard and Canonica, 2005; Canonica and Freiburghaus, 2001; Halladja et al., 2007). The increase in
Fig. 4 e 10 mM and 0.1 mM ibuprofen pseudo first order degradation rate constants (kobs hL1) in the presence of Pony Lake Fulvic Acid (PLFA), Suwannee River Fulvic Aci (SRFA), and Old Woman Creek Fulvic Acid (OWCFA).
4453
degradation rates at low ibuprofen concentrations (Fig. 4) reflects the influence of long-lived and short-lived reactive transients generated by the fulvic acids on ibuprofen photodegradation. As observed by us and others (Canonica and Freiburghaus, 2001; Canonica and Laubscher, 2008; Cawley et al., 2009) fulvic acids derived from algal/microbial sources e.g., PLFA appears to be more reactive by these pathways relative to DOM that includes significant amounts of organic matter derived from higher plants. Our results show that the degree to which these long-lived photo-transients are formed is dictated by the composition of the fulvic acid given that the effect is most pronounced for PLFA and is smallest in the presence of OWCFA. To further probe reactive transients at low concentrations, anoxic ibuprofen photolytic experiments were performed in the presence of each fulvic acid. In contrast to the experiments conducted at the higher ibuprofen concentration, degradation rates in the anoxic PLFA and SRFA solutions decreased (Table 1 and Fig. 5). This result is not unexpected given that fulvic acid- generated long-lived transients such as peroxyl and oxyl radicals require dioxygen to form (Richard and Canonica, 2005). Conversely, low initial concentration ibuprofen kinetics remained unchanged in the presence of OWCFA under suboxic conditions (Fig. 5). This fulvic acid was also the least sensitive to changes in the initial concentration of ibuprofen, which suggests that it is either producing longlived radicals that can form in the absence of dioxygen e.g., phenoxyl radicals (Mvula et al., 2001) and/or is an exceptionally good scavenger of these oxidants (Canonica and Laubscher, 2008). Thus, whatever long-lived transients are responsible for the increased degradation of ibuprofen at low concentrations, their effectiveness appears to be highly dependent upon the composition of the fulvic acid. The addition of isopropanol decreased ibuprofen’s degradation rates at low initial concentrations in the presence of all the fulvic acids. Similar observations made at the higher initial ibuprofen concentration (10 mM) were observed for SRFA and OWCFA, but not PLFA where isopropanol had no
Fig. 5 e 0.1 mM ibuprofen pseudo first order degradation rate constants (kobs hL1) in the presence of Pony Lake Fulvic Acid (PLFA), Suwannee River Fulvic Acid (SRFA), and Old Woman Creek Fulvic Acid (OWCFA) under oxic and suboxic systems and with 20 mM isopropanol (Iso).
4454
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 4 9 e4 4 5 8
effect. Thus, it appears that isopropanol reacts with some of the unknown long-lived radicals capable of reacting with ibuprofen at these lower initial concentrations. This study demonstrates that DOM composition plays an important role in the type and amount of reactive transients formed when irradiated. Work conducted by us (Cawley et al., 2009; Guerard et al., 2009) showed that compounds that were transformed by different pathways were highly sensitive to the photosensitizer composition. While ibuprofen reacts with both OH and 3DOM it appears that the former pathway is more important for terrestrially-derived DOM, while the later is the dominant mechanism for DOM derived from microbial (algal and bacterial) precursors. This corroborates recent work by Wenk et al. (2011) who showed that terrestrially-derived DOM is effective at inhibiting the oxidation of organic compounds by the 3DOM pathway, but not by the hydroxyl radical.
3.2.
Photolysis byproduct formation
Multiple byproducts were formed and accumulated in irradiated ibuprofen solutions in the absence and presence of SRFA, OWCFA, and PLFA. In particular one byproduct dominates in the HPLC chromatograms, while another shows up as a redshifted peak in our UVeVis spectrophotometry scans (264 nm absorbance vs. 223 nm for ibuprofen) (Fig. 6a and b). Further analysis of these photo-derivatives using 1H-NMR spectroscopy (Figs S.1 and S.2 in the SI: Note that all figures in SI will have an “S” prefix) and LC-QTOF-MS (Fig. 7a) of unreacted ibuprofen and the photolyzed solution revealed multiple compounds, consistent with our HPLC data. LC-QTOF-MS in positive ion mode identified an unknown compound with a molecular weight of 176.2 that could be two possible derivatives in the photolysis solution: isobutylacetophenone (IBAP), which is an intermediate in the synthesis of ibuprofen (Fig. 7b and c) and/or 1-(4isobutylphenyl) ethanol (4IBPE). Our GCeMS data corroborates these results (Table S.1). The identification of IBAP agrees with the results of Castell et al. (1987), who identified this compound as a byproduct of ibuprofen irradiated in methanol (w8% IBAP is formed at an [Ibuprofen]o ¼ 16 mM). A UVeVis spectrum of an IBAP standard also confirms its existence as its absorbance maximum occurs at the same absorbance (264 nm) of one of the photo-derivatives (Fig. 6 and Fig S.3), and the HPLC retention time for IBAP is also identical to one of our byproduct peaks. Finally parts of our 1H NMR spectrum of the IBAP standard are similar to the photolyzed solution (Figs S.2 and S.4) further confirming this structural assignment (note the quartet at 2.5 ppm and doublet at 7.8 ppm in both spectra). Indeed, isobutylacetophenone has been detected at concentrations of 0.00023 mM in waters in urban areas, citing ibuprofen degradation as its environmental precursor (Zorita et al., 2007). We were able to isolated one of the byproduct peaks (Fig. 6b; peak retention time w7 min) from the HPLC waste line effluent and analyzed it using both 1-D and 2-D correlation spectroscopy (COSY) 1H NMR. Surprisingly, the HPLC isolated peak revealed 1-D spectra that are quite complex and lack the methyl group quartet at 2.5 ppm and doublet at 7.8 ppm (Fig S.5) and is thus not IBAP. The 1-D NMR spectra’s complexity
Fig. 6 e a. UVeVis scan of photolyzed ibuprofen solution at various time points (hours) demonstrating byproduct accumulation at 264 nm; b. HPLC chromatogram (l [ 264 nm) of solution (Co [ 10 mM) photolyzed for 48 h (ibuprofen retention time w5 min; “major” photolysis byproduct retention time w7 min, additional byproduct retention time w8 min).
leads us to speculate that multiple compounds are co-eluting from the HPLC column at 7 min. 2-D 1H NMR (COSY) of the HPLC isolated fraction reveals structural connectivity between the quartet at 4.1 ppm and a doublet at 1.2 ppm (Fig S.6). This quartet at 4.1 ppm, when compared to the ibuprofen scan (Fig S.1), is shifted upfield from 3.7 ppm after photolysis, but is still well resolved. This shift, combined with the observed COSY connectivity, indicates the presence of a more electronegative functional group at the carboxylic acid group of ibuprofen, which has presumably been altered due to photolysis. This was not observed in the 2-D 1H NMR spectrum for IBAP (Fig S.7). Taken together, the NMR data reveal the structure of the major byproduct observed in our HPLC chromatograms as 4IBPE plus other co-eluting compounds and confirms our mass spectrometry results. Our analysis of the byproducts revealed other substances in addition to IBAP and 4IBPE. We observed aromatic doublets in the 1-D 1H NMR data that shifted from the left of the solvent peak (CHCl3 w 7.4 ppm) prior to irradiation to the right after
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 4 9 e4 4 5 8
4455
Fig. 7 e Liquid Chromatography Electrospray Ionization Quadrupole Time of Flight Mass Spectrometer (LC-ESI-QTOF-MS) of 10 mM ibuprofen solution photolyzed 48 h and proposed byproduct structures.
photolysis (Fig S.5). This shift suggests the attachment of a more electronegative moiety to the benzene ring or more specifically, hydroxylation of the aromatic spin system. The NMR spectra taken together with the GCeMS and LC-QTOFMS data suggest a phenol derivative of the parent compound. Photodecarboxylation pathways of NSAIDs in nonaqueous systems have been extensively probed (Castell et al., 1987; Budac and Wan, 1992; Bosca´ et al., 1994, 2001; Monti et al., 1997). Castell et al. (1987) demonstrated that IBAP and 4IBPE are the result of reactions involving dioxygen. These investigators also report equal percent yield of IBAP and 4IBPE, upon irradiation of ibuprofen in methanol in contrast to our systems, which demonstrated that the 4IBPE derivative is the dominant photoproduct. Because Castell et al. (1987) conducted their experiments in methanol, where the solubility of oxygen is significantly higher than in water the yields
of IBAP is presumably higher. Our identification of both these byproducts strongly suggests that photodecarboxylation of ibuprofen can also occur in aqueous systems (Scheme 1). Finally, we have also shown that hydroxylation of the benzene ring takes place (Fig S.5). Fulvic acid composition can also affect photo-induced byproduct formation. At [Ibuprofen]o ¼ 10 mM in the presence of SRFA and OWCFA, HPLC peak area units of the major byproduct do not significantly change when compared to direct photolysis (Fig S.8). In the presence of PLFA, however, the byproduct peak at retention time w7 min (see Fig. 6b) increases relative to direct photolysis and in the presence of the other FA (Fig S.8: filled triangles). We have no explanation as to why PLFA is able to increase the production of this byproduct. Formation of the byproduct peak (7 min retention time) in the presence of fulvic acids at the low initial ibuprofen
4456
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 4 9 e4 4 5 8
Scheme 1 e Proposed reactions: Ibuprofen photodecarboxylation (1) followed by oxygen addition to carbon centered radical and subsequent rearrangement resulting in the formation of isobutylacetophenone (2) and the hydroxylation of carbon centered radical to form 1-(4-isobutylphenyl)ethanol.
concentrations (0.1 mM) does not show the same trend (Fig S.9). Indeed, no correlation is apparent between byproduct peak area generation and the presence of the fulvic acids. Moreover, the linear increase in photo-derivative abundance demonstrated at 10 mM (Fig S.8) is not replicated at the lower concentration (Fig S.9). Our low concentration experiments demonstrate that reaction of ibuprofen with both long- and short-lived reactive transients at this concentration most probably results in the formation of yet-to-be identified derivatives.
4.
Conclusions
The presence of dissolved organic matter significantly enhances the photolytic fate of ibuprofen in sunlit natural waters. Even though our results show that the composition of DOM does not appear to influence the rate of ibuprofen photolysis at high (10 mM) initial concentrations, controlled scavenging and anoxic studies reveal that it undergoes phototransformation by different mechanisms. Further the type of DOM influences the relative abundance of byproducts formed. We were able to identify three important byproducts of ibuprofen photolysis; isobutylacetophenone (a precursor in the synthesis of ibuprofen), 1-(4-isobutylphenyl)ethanol, and a phenol derivative. We observed a significant enhancement in photolytic reaction rates in the presence of DOM at lower initial ibuprofen concentrations. Further DOM composition plays
a more important role at these lower concentrations whereby the microbially derived Pony Lake fulvic acid proved to be the most reactive toward ibuprofen. We attribute these differences in reaction rates to the ability of the DOM phase to form “long-lived” reactive photo-transients of unknown origin when irradiated. Thus, at environmentally relevant concentrations the photolytic transformation of ibuprofen is significantly enhanced by DOM, which may in part explain its extremely low concentrations in sunlit surface waters.
Acknowledgments We thank the members of the Chin research group for helping us isolate the Old Woman Creek fulvic acid and especially Collin Ward for helping LEJ with the photolysis experiments. We also thank Silvio Canonica for his helpful discussions with LEJ as well as comments provided by two anonymous reviewers. This work was partially supported by NOAA/NERR Fellowship awarded to LEJ and by a grant from the National Science Foundation CBET 0504434.
Appendix. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.watres.2011.05.041.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 4 9 e4 4 5 8
references
Aguer, J., Richard, C., Andreux, F., 1997. Comparison of the photoinductive properties of commercial, synthetic and soilextracted humic substances. J. Photochem. Photobiol. A: Chem. 103 (1e2), 163e168. Boreen, A.L., Arnold, W.A., McNeill, K., 2003. Photodegradation of pharmaceuticals in the aquatic environment: a review. Aquat. Sci. 65 (4), 320e341. Boreen, A.L., Arnold, W.A., McNeill, K., 2005. Triplet-sensitized photodegradation of sulfa drugs containing six-membered heterocyclic groups: identification of an SO2 extrusion photoproduct. Environ. Sci. Technol. 39 (10), 3630e3638. Bosca´, F., Marı´n, M.L., Miranda, M.A., 2001. Photoreactivity of the non-steroidal anti-inflammatory 2-arylpropionic acids with photosensitizing side effects. Photochem. Photobiol. 74 (5), 637e655. Bosca´, F., Miranda, M.A., Carganico, G., Mauleo´n, D., 1994. Photochemical and photobiological properties of ketoprofen associated with the benzophenone chromophore. Photochem. Photobiol. 60 (2), 96e101. Brown, A., McKnight, D.M., Chin, Y.P., Roberts, E.C., Uhle, M., 2004. Chemical characterization of dissolved organic matter in Pony Lake, a saline coastal pond in Antarctica. Mar Chem. 89 (1e4), 327e337. Budac, D., Wan, P., 1992. Photodecarboxylation: mechanism and synthetic utility. J. Photochem. Photobiol. A: Chem. 67 (2), 135e166. Buser, H., Poiger, T., Muller, M.D., 1999. Occurrence and environmental behavior of the chiral pharmaceutical drug ibuprofen in surface waters and in wastewater. Environ. Sci. Technol. 33 (15), 2529e2535. Canonica, S., Freiburghaus, M., 2001. Electron-rich phenols for probing the photochemical reactivity of freshwaters. Environ. Sci. Technol. 35 (4), 690e695. Canonica, S., Hoigne´, J., 1995. Enhanced oxidation of methoxy phenols at micromolar concentration photosensitized by dissolved organic matter. Chemosphere 30 (12), 2365e2374. Canonica, S., Jans, U., Stemmler, K., Hoigne´, J., 1995. Transformation kinetics of phenols in water: photosensitization by dissolved natural organic material and aromatic ketones. Environ. Sci. Technol. 29 (7), 1822e1831. Canonica, S., Laubscher, H., 2008. Inhibitory effect of dissolved organic matter on triplet-induced oxidation of aquatic contaminants. Photochem. Photobiol. Sci. 7 (5), 547e551. Castell, J.V., Gomez, M.J., Miranda, M.A., Morera, I.M., 1987. Photolytic degradation of ibuprofen. Toxicity of the isolated photoproducts on fibroblasts and erythrocytes. Photochem. Photobiol. 46 (6), 991e996. Cawley, K.M., Hakala, J.A., Chin, Y.P., 2009. Evaluating the triplet state photoreactivity of dissolved organic matter isolated by chromatography and ultrafiltration using an alkylphenol probe molecule. Limnol. Ocean. Meth. 7, 391e398. Cooper, W.J., Zika, R.G., Petasne, R.G., Fischer, A.M., 1989. Sunlight induced photochemistry of humic substances in natural waters: major reactive species. ACS Symp. Ser. 219, 333e362. Daughton, C.G., Ternes, T.A., 1999. Pharmaceutical and personal care products in the environment: agents of subtle change? Environ. Health Persp 107 (6), 907e938. Dulin, D., Mill, T., 1982. Development and evaluation of sunlight actinometers. Environ. Sci. Technol. 16 (11), 815e820. Edhlund, B.L., Arnold, W.A., McNeill, K., 2006. Aquatic photochemistry of nitrofuran antibiotics. Environ. Sci. Technol. 40 (17), 5422e5427. Fent, K., Weston, A.A., Caminada, D., 2006. Ecotoxicology of human pharmaceuticals. Aquat. Tox. 76 (2), 122e159.
4457
Gerecke, A.C., Canonica, S., Muller, S.R., Scharer, M., Schwarzenbach, R.P., 2001. Quantification of dissolved organic matter (DOM) mediated phototransformation of phenylurea herbicides in lakes. Environ. Sci. Technol. 35 (19), 3915e3923. Guerard, J.J., Miller, P.L., Trouts, T.D., Chin, Y., 2009. The role of fulvic acid composition in the photosensitized degradation of aquatic contaminants. Aquat. Sci. 71 (2), 160e169. Halladja, S., Ter Halle, A., Auger, J.P., Boukamh, A., Richard, C., 2007. Inhibition of humic substances mediated photooxygenation of furfuryl alcohol by 2,4,6-trimethylphenol evidence for reactivity of the phenol with humic triplet excited states. Environ. Sci. Technol. 41 (17), 6066e6073. Hoigne´, J., Faust, B.C., Haag, W.R., Scully, F.E., Zepp, R.G., 1989. Aquatic humic substances and sinks of photochemically produced transients reactants. ACS Symp. Ser. 219, 363e381. Hutt, A.J., Caldwell, J., 1983. The metabolic chiral inversion of 2arylpropionic acids- a novel route with pharmacological consequences. J. Pharm. Pharmacol. 35 (11), 693e704. Kolpin, D.W., Furlong, E.T., Meyer, M.T., Thurman, E.M., Zaugg, S. D., Barber, L.B., Buxton, H.T., 2002. Pharmaceuticals, hormones, and other organic, wastewater contaminants in U. S. streams, 1999e2000: a national reconnaissance. Environ. Sci. Technol. 36 (6), 1202e1211. Leenheer, J.A., 1981. Comprehensive approach to preparative isolation and fractionation of dissolved organic carbon from natural waters and wastewaters. Environ. Sci. Technol. 15 (5), 578e587. Lin, A.Y., Reinhard, M., 2005. Photodegradation of common environmental pharmaceuticals and estrogens in river water. Environ. Toxicol. Chem. 24 (6), 1303e1309. Lindqvist, N., Tuhkanen, T., Kronberg, L., 2005. Occurrence of acidic pharmaceuticals in raw and treated sewages and in receiving waters. Wat. Res. 39 (11), 2219e2228. Miller, P.L., Chin, Y.P., 2002. Photoinduced degradation of carbaryl in a wetland surface water. J. Agric. Food Chem. 50 (23), 6758e6765. Monti, S., Sortino, S., De Guidi, G., Marconi, G., 1997. Photochemistry of 2-(3-benzoylphenol)proponic acid (ketoprofen) part 1: a picosecond and nanosecond time resolved study in aqueous solution. J. Chem. Soc. Faraday Trans. 93 (13), 2269e2275. Mvula, E., Schuchman, M.N., Sonntag, C., 2001. Reactions of phenol-OH-adduct radicals. Phenoxyl radical formation by water elimination vs. oxidation by dioxygen. J. Chem. Soc. Perkin Trans. 2, 264e268. Nakada, N., Tanishima, T., Shinohara, H., Kiri, K., Takada, H., 2006. Pharmaceutical chemicals and endocrine disrupters in municipal wastewater in Tokyo and their removal during activated sludge treatment. Wat. Res. 40 (17), 3297e3303. Nakada, N., Shinohara, H., Murata, A., Kiri, K., Managaki, S., Sato, N., Takada, H., 2007. Removal of selected pharmaceuticals and personal care products (PPCPs) and endocrine-disrupting chemicals (EDCs) during sand filtration and ozonation at a municipal sewage treatment plant. Wat. Res. 41 (19), 4373e4382. Nikolaou, A., Meric, S., Fatta, D., 2007. Occurrence patterns of pharmaceuticals in water and wastewater environments. Anal. Bioanal. Chem. 387 (4), 1225e1234. Packer, J.L., Werner, J.J., Latch, D.E., McNeill, K., Arnold, W.A., 2003. Photochemical fate of pharmaceuticals in the environment: naproxen, diclofenac, clofibric acid, and ibuprofen. Aquat. Sci. 65 (4), 342e351. Pe´rez-Estrada, L.A., Malato, S., Gernjak, W., Agu¨era, A., Thurman, E.M., Ferna´ndez-Alba, A.R., 2005. Photo-Fenton degradation of diclofenac: identification of main intermediates and degradation pathway. Environ. Sci. Technol. 39 (21), 8300e8306. Quintana, J.B., Weiss, S., Reemtsma, T., 2005. Pathways and metabolites of microbial degradation of selected acidic
4458
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 4 9 e4 4 5 8
pharmaceutical and their occurrence in municipal wastewater treated by a membrane bioreactor. Wat. Res. 39 (12), 2654e2664. Richard, C., Canonica, S., 2005. Aquatic phototransformation of organic contaminants induced by coloured dissolved natural organic matter. In: The Handbook of Environmental Chemistry, vol. 2, Part M, pp. 299e323. Scheytt, T., Mersmann, P., Lindsta¨dt, R., Heberer, T., 2005. 1-octanol/water partition coefficients of 5 pharmaceuticals from human medical care: carbamazepine, clofibric acid, diclofenac, ibuprofen, and propyphenazone. Water Air Soil Poll. 165 (1e4), 3e11. Ternes, T.A., 1998. Occurrence of drugs in German sewage treatment plants and rivers. Wat. Res. 32 (11), 3245e3260. Thurman, E.M., Malcolm, R.L., 1981. Preparative isolation of aquatic humic substances. Environ. Sci. Technol. 15 (4), 463e466. Vaughan, P.P., Blough, N.V., 1998. Photochemical formation of hydroxyl radical by constituents of natural waters. Environ. Sci. Technol. 32 (19), 2947e2953. Wenk, J., von Gunten, U., Canonica, S., 2011. Effect of dissolved organic matter on the transformation of contaminants induced by excited triplet states and the hydroxyl radical. Environ. Sci. Technol. 45 (4), 1334e1340. White, E.M., Vaughan, P.P., Zepp, R.G., 2003. Role of the photoFenton reaction in the production of hydroxyl radicals and photobleaching of colored dissolved organic matter in a coastal river of the southeastern United States. Aquat. Sci. 65 (4), 402e414.
Winker, M., Faika, D., Gulyas, H., Otterpohl, R., 2008. A comparison of pharmaceutical concentrations in raw municipal wastewater and yellow water. Sci. Total Env. 399 (1e3), 96e104. Xu, J., Wu, L., Chang, A.C., 2009. Degradation and adsorption of selected pharmaceuticals and personal care products (PPCPs) in agricultural soils. Chemosphere 77 (10), 1299e1305. Yamamoto, H., Nakamura, Y., Moriguchi, S., Nakamura, Y., Honda, Y., Tamura, I., Hirata, Y., Hayashi, A., Sekizawa, J., 2009. Persistence and partitioning of eight selected pharmaceuticals in the aquatic environment: laboratory photolysis, biodegradation, and sorption experiments. Wat. Res. 43 (2), 351e362. Yu, J.T., Bouwer, E.J., Coelhan, M., 2006. Occurrence and biodegradability studies of selected pharmaceuticals and personal care products in sewage effluent. Agric. Water Manag. 86 (1e2), 72e80. Zhou, X., Mopper, K., 1990. Determination of photochemically produced hydroxyl radicals in seawater and freshwater. Mar. Chem. 30, 71e88. Zorita, S., Barri, T., Mathiasson, L., 2007. A novel hollow-fiber microporous membrane liquid-liquid extraction for determination of free 4-isobutylacetophenone concentration at ultra trace level in environmental aqueous samples. J. Chromatogr. 1157 (1e2), 30e37. Zwiener, C., Frimmel, F., 2003. Short-term tests with a pilot sewage plant and biofilm reactors for the biological degradation of the pharmaceutical compounds clofibric acid, ibuprofen, and diclofenac. Sci. Total Environ. 309 (1e3), 201e211.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 5 9 e4 4 6 9
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Intrinsic biodegradation potential of aromatic hydrocarbons in an alluvial aquifer e Potentials and limits of signature metabolite analysis and two stable isotope-based techniques Barbara Morasch a,*, Daniel Hunkeler a, Jakob Zopfi b, Brice Temime c, Patrick Ho¨hener c a
Center for Hydrogeology, University of Neuchaˆtel, Rue Emile Argand 11, 2009 Neuchaˆtel, Switzerland Laboratory of Microbiology, University of Neuchaˆtel, Rue Emile Argand 11, 2009 Neuchaˆtel, Switzerland c Laboratoire Chimie Provence, Universite´ de Provence, CNRS, Place Victor Hugo, F-13331 Marseille Cedex 3, France b
article info
abstract
Article history:
Three independent techniques were used to assess the biodegradation of monoaromatic
Received 21 February 2011
hydrocarbons and low-molecular weight polyaromatic hydrocarbons in the alluvial aquifer
Received in revised form
at the site of a former cokery (Fle´malle, Belgium).
12 May 2011
Firstly, a stable carbon isotope-based field method allowed quantifying biodegradation
Accepted 28 May 2011
of monoaromatic compounds in situ and confirmed the degradation of naphthalene. No
Available online 14 June 2011
evidence could be deduced from stable isotope shifts for the intrinsic biodegradation of larger molecules such as methylnaphthalenes or acenaphthene. Secondly, using signature
Keywords:
metabolite analysis, various intermediates of the anaerobic degradation of (poly-) aromatic
Groundwater contamination
and heterocyclic compounds were identified. The discovery of a novel metabolite of ace-
Natural attenuation
naphthene in groundwater samples permitted deeper insights into the anaerobic biode-
(Poly-) aromatic hydrocarbons
gradation of almost persistent environmental contaminants. A third method, microcosm
Signature metabolites
incubations with
Stable isotopes
techniques one and two by providing quantitative information on contaminant biodegra-
Biodegradation rates
dation independent of molecule size and sorption properties. Thanks to stable isotope
13
C-labeled compounds under in situ-like conditions, complemented
labels, the sensitivity of this method was much higher compared to classical microcosm studies. The
13
constants for
13
C-microcosm approach allowed the determination of first-order rate C-labeled benzene, naphthalene, or acenaphthene even in cases when
degradation activities were only small. The plausibility of the third method was checked by comparing 13C-microcosm-derived rates to field-derived rates of the first approach. Further advantage of the use of
13
C-labels in microcosms is that novel metabolites can be linked
more easily to specific mother compounds even in complex systems. This was achieved using alluvial sediments where
13
C-acenaphthyl methylsuccinate was identified as trans-
formation product of the anaerobic degradation of acenaphthene. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
Biodegradation is claimed to be the key process leading to decontamination of many abandoned industrial sites impaired
with coal- and tar oil-derived compounds. In practice, it is often difficult to judge whether degradation is taking place or not. Under environmental conditions, particularly in the absence of oxygen, mono- and polyaromatic contaminants are
* Corresponding author. Present address: Environmental Mineralogy and Chemistry, Center for Applied Geoscience (ZAG), University of Tuebingen, Sigwartstrasse 10, 72076 Tuebingen, Germany. Tel.: þ49 7071 2973135. E-mail address:
[email protected] (B. Morasch). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.05.040
4460
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 5 9 e4 4 6 9
biodegraded at low rates or are supposedly persisting (Zamfirescu and Grathwohl, 2001; Foght, 2008). Consequently, solid information is needed on the in situ biodegradation of coal- and tar oil-derived pollutants. Due to low solubility in water and tendencies to sorb, turnover of (poly-) aromatic compounds (PAHs) may often be limited by mass transfer and not by microbial activity (Bosma et al., 1997). It is controversial whether biodegradation of aromatic hydrocarbons can be distinguished at all from partitioning when only a small decrease in concentration is measured (Foght, 2008). Even though it is considered as an issue of increasing importance, no universal technique exists to measure biodegradation at contaminated sites. In this study, three independent approaches were used in combination to assess the intrinsic biodegradation of aromatic and heterocyclic environmental contaminants. Particular attention was paid to useful approaches for the quantification of in situ biodegradation of compounds that show small degradation activities or have the tendency to sorb. The potentials and limits of the three techniques are compared and discussed.
1.1. Compound-specific stable isotope analysis of individual pollutants in groundwater samples of contaminated sites This technique has proven appropriate for the direct assessment of biodegradation at those field sites where a contaminant plume has established. Along the centerline of a contaminant plume, in situ biodegradation is resolved over distance in the stable isotope shifts of individual groundwater pollutants. Stable isotope fractionation during degradation of monoaromatic hydrocarbons, naphthalenes, alkanes, chlorinated solvents, and gasoline additives was studied under various redox conditions (Hunkeler and Morasch, 2010). Larger molecules of high environmental relevance (e.g. three and more ring PAHs) have been investigated much less extensively and no enrichment of heavier isotopes above the analytical error was reported (Mazeas et al., 2002). Furthermore, compound-specific stable isotope analysis (CSIA) was used in field studies to calculate the percentage of in situ biodegradation and first-order rate constants (Richnow et al., 2003b; Batlle-Aguilar et al., 2009).
1.2. Screening for signature metabolites indicating contaminant biodegradation in groundwater samples Signature metabolites are highly specific reaction intermediates produced only during biodegradation of target contaminants. They need to be excluded as contaminants themselves in order to be indicative of on-site remediation (Phelps et al., 2002). In previous studies, various signature metabolites were either identified in batch culture experiments or were extracted from groundwater samples directly. Recognized molecules include intermediates of the anaerobic degradation of alkylated aromatic compounds, e.g. benzylsuccinates that are formed in addition reactions with fumarate (Elshahed et al., 2001; Beller, 2002). Apart from methylated benzenes and naphthalenes, methylated heterocyclic compounds are anaerobically degraded also via initial fumarate addition (Annweiler et al., 2000; Safinofski et al., 2006).
1.3. Incubations of field material in the lab under in situ-like conditions adding isotopically labeled contaminants The intrinsic biodegradation potential of individual compounds can be assessed by incubating field material in the laboratory under in situ-like conditions. Even in complex systems, biodegradation can be specifically tracked when isotopically labeled compounds are supplied as markers. Isotope labels circumvent difficulties related to classical microcosm experiments with field material that potentially contains organic background contaminants. Briefly, the method is based on the recovery of isotope labels in CO2 produced from mineralization of isotope-labeled substrates. Complete oxidation of naphthalene under anoxic conditions e.g. was shown by adding 14C-labeled naphthalene to aquifer material and to marine harbor sediments (Chapelle et al., 1996; Coates et al., 1996; Langenhoff et al., 1996). Recently, naphthalene was also used in 13C-labeled form to confirm biodegradation in sedimentegroundwater microcosms (Morasch et al., 2007). All techniques mentioned above can provide evidence for natural attenuation of contaminants according to the three lines of evidence established by the National Research Council (2000). By definition, conditions are met if (I) the loss of contaminant is documented at the field scale, (II) the presence of degrading microorganisms is confirmed by means of incubation experiments using field material, and (III) the direct evidence for microbial activity in situ can be provided. Techniques one (CSIA) and two (signature metabolite analysis) meet criterion (III). They have already been applied in combination, e.g. at a former gasworks site with a longstanding contamination with mono- and polyaromatic hydrocarbons and in a controlled release experiment of benzene, toluene, and o-xylene at a US Air Force base (Griebler et al., 2004; Beller et al., 2008). However, the combination of CSIA and signature metabolite analysis has its limitations. CSIA has the potential to quantify the biodegradation of smaller contaminants, while its sensitivity decreases with increasing molecular size of the compounds of interest. Since frequently the biodegradation rates also get lower with molecular size the method is not applicable for larger contaminants. Signature metabolite analysis can e in a qualitative way e provide information on the biodegradation of these larger compounds. Hence, degradation kinetics of larger contaminant molecules remains unrevealed. In this study we use a third technique (microcosms with 13C-labeled compounds) to overcome these limitations and to provide additional quantitative information on in situ biodegradation independent of molecular size. Results of all three techniques are compared and their potentials and limits evaluated.
2.
Materials and methods
2.1.
Field site
The study site was a former coke and gas factory that was dismantled in 1984. The property of 400 m 250 m was
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 5 9 e4 4 6 9
situated in an industrial environment at 25 m distance from the Meuse River upstream of the city of Lie`ge, Belgium (Fig. 1; geographical location: þ50 360 19.7600 , þ5 290 13.5200 , www. google.com/maps). The topmost 4 m below the surface consisted of backfill material followed by silt, sand, and clay deposits of about 2 m thickness, and 8 m of fine gravel above the carboniferous shale bedrock. The water table was located in the alluvial gravel layer at 5.5e7 m depth. Groundwater flow was in eastern direction. Previous field characterization evidenced a hydrological gradient of 0.3% and a saturated hydraulic conductivity that varied between 105 and 103 m/s (Batlle-Aguilar et al., 2009). Severe contamination with heavy metals, cyanides, mineral oils, as well as with mono- and polyaromatic hydrocarbons reached down to 11 m depth. The contamination was of unknown horizontal extent according to various measurements by the Walloon Environmental Protection Agency. In the north-western part of the site (Fig. 1), Eh values of the groundwater were at 300 mV and nitrate was almost depleted. Strictly reducing conditions prevailed up to 100 m in southeastern direction toward the Meuse River before the Eh rose to þ100 mV and nitrate concentrations of up to 15 mg L1 were observed. Sulfate concentrations in groundwater were between 500 and 2100 mg/L, hence sulfate was assumed to be the major electron acceptor in the degradation of organic contaminants after O2 had been depleted (For a redox zonation map, see Batlle-Aguilar et al., 2009).
2.2.
Sampling
Sampling campaigns took place in March 2005 and July 2006 where 17 and 23 groundwater wells were sampled, respectively (Table S1). Water was pumped with submersible pumps at a rate of 1e5 L/min. Groundwater table, temperature,
4461
conductivity, pH, and dissolved O2 were recorded using specific field probes (WTW; Weilheim, Germany). Subsequently, water was sampled to determine alkalinity, Mn2þ, Fe2þ, HS-, and methane. Additional samples were taken for the quantification of aromatic contaminants, CSIA, and the identification of signature metabolites. Water samples were conserved immediately on site by adding 0.1% (vol/vol) of NaOH (5 M). Upgradient from the contamination, groundwater wells E6p and F4 served as references (Fig. 1). In 2005, freshly drilled core material from location U13 adjacent to the source zone was sampled every 0.5 m for microcosm experiments.
2.3.
Microcosm set-up and sampling
Aquifer material was collected during drillings, filled into brown-glass bottles of 120 mL with Teflon-sealed caps, cooled immediately, and stored at 5 C until microcosms were prepared. Sediments of core U13 were pooled to four different depth layers (5e6, 7e8, 9, and 12e13 m). Working under an atmosphere of N2 in a glove bag (SigmaeAldrich), 53 microcosms were set up in culture bottles of 50 mL volume. Every microcosm consisted of 25 g of sediment and 15 mL of O2-free groundwater that had been sterile-filtered and diluted fivefold with autoclaved nanopure water, to achieve nitrate concentrations that better resembled groundwater concentrations of the strictly reducing zone. The diluted groundwater contained 6 mg/L of nitrate and 140 mg/L of sulfate. Benzene, naphthalene, or acenaphthene labeled with 13C at six positions (99% purity, Cambridge Isotope Laboratories) was dissolved in the anoxic groundwater that was added to the culture bottles (Morasch et al., 2007). The headspace either was air for sediments from the unsaturated zone or N2 for microcosms with sediments from the saturated zone. Microcosms were closed with non-absorptive Viton rubber stoppers. Controls were
Fig. 1 e Site map of the former cokery of Fle´malle, Belgium, located in the direct vicinity of the river Meuse. Locations of groundwater (GW) samples are shown as circles, the sediment sampling location U13 is depicted as square. The arrow roughly describes the groundwater flow direction. The map has been modified from Batlle-Aguilar et al. (2009).
4462
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 5 9 e4 4 6 9
prepared without addition of substrate or microcosms were autoclaved twice at days 1 and 4 of the experiment. Incubations took place in the dark at 16 C.
2.4.
Analysis of organic contaminants
Concentrations of benzene, toluene, ethylbenzene, and mxylene (BTEX), and of low-molecular weight PAHs were analyzed using a gas chromatograph (Varian 3800) with a CP8410 autoinjector for solid phase microextraction (SPME). The aromatic compounds were extracted from the headspace of 2 mL vials filled with 0.5 mL of groundwater and 0.3 g of NaCl using polydimethylsiloxane fibers (100 mm film thickness, Supelco). Extraction and analysis were performed as described previously (Morasch et al., 2007).
2.5.
Stable carbon isotope analysis
Groundwater for 13C/12C isotope analysis was sampled in volumes of 1e2 L. CSIA of aromatic hydrocarbons was performed using a Trace GC coupled to a Delta Plus XP isotope ratio mass spectrometer (IRMS) via a GC combustion III interface (Thermo Finnigan). Complete protocols of sample preparation and analysis are provided in Supplementary material. For stable carbon isotope analysis of CO2, 2 mL headspace samples were taken from microcosms through the Viton stoppers with a gas tight syringe working under an atmosphere of N2. Sampling intervals were between three and seven days at the beginning of the incubation and three-monthly toward the end of the experiment (Table S2). Concentration measurements of CO2 and stable carbon isotope measurements were performed connecting a headspace autosampler (Tekmar Dohrmann 7000) to the GC-IRMS using a previously described protocol (Morasch et al., 2007). CO2 concentrations were quantified based on five-point calibration curves with an average correlation of r ¼ 0.98. Stable isotope ratios were determined relative to an external CO2 reference gas and reported as d [&] deviation to the VPDB standard d13 C ½& ¼
Rsample 1 1000 Rstd
(1)
where Rsample and Rstd are the carbon stable isotope ratios of the sample and of the standard, respectively.
2.6.
Metabolite extraction, analysis, and identification
Putative degradation intermediates of aromatic or heterocyclic contaminants were extracted with dichloromethane from 1 L groundwater samples acidified to pH 1e2 with HCl (37%). The extraction was repeated and dichloromethane fractions were pooled. The complete protocol is provided in Supplementary material. After the completion of degradation experiments, potential metabolites were extracted from the water phase and the sediment of microcosms. For metabolite extraction from the sediment fraction, 20 mL of acetone was added into the microcosm bottles and placed into an ultrasonic bath for 10 min. The acetone phase was decanted from the microcosm bottle and the procedure was repeated with 20 mL of
dichloromethane. Then, extracts from the water phase and the sediment fraction were combined. The complete extraction protocol may be found in Supplementary material. Analysis by GCeMS was performed using a Trace GC coupled to a Polaris Q Ion Trap Mass spectrometer (Finnigan). Identity of substances was confirmed by co-elution with reference compounds and by comparison of mass spectra with published data.
2.7.
Calculations
In situ biodegradation of BTEX was calculated using the approach of Richnow et al. (2003b): B½% ¼
ct 100 1 c0
(2)
where B is the percentage of biodegradation of the substrate; c0 and ct are substrate concentrations at the source and at a downgradient monitoring point. Independently from concentration measurements, ct was obtained from c0 in combination with the stable isotope ratios R0 at the source and Rt at a downgradient monitoring point, and substrate-specific stable isotope enrichment factors (3) derived from laboratory studies:
Rt ct ¼ c0 R0
1000 3
(3)
First-order biodegradation rates l were calculated from field isotope data according to l ¼ Dd13 C=ð3 tc Þ
(4)
13
where Dd C is the shift in the carbon isotope ratio between the source and a downgradient monitoring point, and tc is the travel time of the contaminant (Hunkeler et al., 2002; Blum et al., 2009). Travel times of the contaminants were estimated based on an intermediate groundwater flow velocity of 0.29 m d1 (Batlle-Aguilar et al., 2009) and compound-specific retardation factors 1f FR ¼ 1 þ r Kd f
(5)
that were obtained assuming a sediment density of r ¼ 2500 kg/m3 and a mobile porosity of f ¼ 0.2. Solidewater distribution coefficients Kd ¼ fOC KOC were calculated assuming a fraction of organic carbon in the sediments of fOC ¼ 0.001. Organic carbon-normalized distribution coefficients KOC were predicted from compound class-specific log KOC log KOW relationships (Schwarzenbach et al., 2003) taking octanolewater coefficients KOW from the Physical Properties Database (SRC Inc., 2009). Half-life times t1/2 and half-concentration distances x1/2 were defined as follows: t1=2 ¼
lnð2Þ l
(6)
x1=2 ¼
v t1=2 FR
(7)
where v is the intermediate groundwater flow velocity in the aquifer.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 5 9 e4 4 6 9
2.8.
3.2. Qualitative assessment of in situ biodegradation using CSIA
Inorganic carbon mass balance
To assess the intrinsic biodegradation potential of the 13Clabeled contaminants, an inorganic carbon mass balance was applied to sedimentegroundwater microcosms (a detailed description is provided in the Supplementary material). The initial amount of inorganic carbon M0 was approximated from the sum of CO2 (g) as determined by GC-IRMS at the beginning of the microcosm experiment, H2CO3 (aq), and the concentration of HCO3 (aq) of 2 mM determined by alkalinity titration for the groundwater added to microcosm bottles at the beginning of the experiment (see Supplementary material for determination of alkalinity). CO2 dissolved in the water phase was calculated from CO2 concentrations in the gas phase using Henry’s law according to ½H2 CO3 ðaqÞ ¼ ½CO2 ðgÞ Kh with a Henry coefficient of Kh ¼ 0.77 (SRC Inc., 2009). The amount of inorganic carbon produced upon the biodegradation of 13C substrates at time t, is depicted MS. Compared to MS, the isotope signature of 13 C-CO2 originating from the degradation of undefined background carbon sources (MBG) was considered equal to the isotope signature M0 at the beginning of the experiment. The biodegradation rate was calculated using a first-order type equation: l$t MSð0Þ MS ¼ MSð0Þ e FR
(8)
with MS(0) being the amount of 13C [mmol] added to the bottle at t ¼ 0 d, the first-order rate constant l, and the reciprocal value of the retardation factor 1/FR designating the fraction of contaminant present in the water phase of the microcosms (Schwarzenbach et al., 2003). For hydrophobic compounds such as BTEX and PAH, the correction by FR is necessary to account for sorption to the sediment matrix since only the dissolved compound fraction is readily available for biodegradation. 13C-CO2-based biodegradation rates were validated previously by comparing the changes in the 13C-aniline concentrations in the water phase to the concomitant evolution of 13C-CO2 in the gas phase (Morasch et al., 2011).
3.
4463
Results and discussion
3.1. Abundance of aromatic hydrocarbons in groundwater Based on a systematic assessment of contaminant concentrations in the aquifer of the Fle´malle site, one major source zone was identified in the north-western part around piezometers D2bis, D1p, and D3p (Fig. 1). In that zone, groundwater concentrations of benzene, toluene, and m-xylene were in the mg/L range; ethylbenzene was in the mg/L range (Table S3). Also the highest concentrations of naphthalene (up to 25 mg/L) were detected in D2bis, D1p, and D3p and decreased along the groundwater flow path toward the Meuse River in eastern direction (Fig. 2a). Concentrations of low-molecular weight PAHs were elevated in piezometer 14 located 39 m downgradient of well D2bis. At well 15, 133 m east of the source zone, there was an additional point of increased contamination with the three-ring compounds acenaphthene and fluorene (Table S3).
13
C/12C isotope signatures of residual groundwater contaminants were measured using CSIA in 2005 and 2006. Improved protocols (Purge&Trap for BTEX, SPME for PAHs) resulted in a higher number of isotope signatures that could be determined in 2006 (Table S4). Most negative d-values of BTEX were found in the major source zone (D2bis, D3p, D1p). Then, 13C became progressively enriched in the residual BTEX along the groundwater flow path in the methanogenic to sulfate-reducing aquifer section between the source and well U13. Less reducing groundwater sampled further toward the river showed lower 13C/12C ratios in the residual contaminants. Representative for all aromatic hydrocarbons detected at the Fle´malle site, stable carbon isotope ratios at the Fle´malle site are displayed for benzene, naphthalene, and acenaphthene (Fig. 2bed). According to CSIA, no naphthalene degradation took place within the first 90 m of the contaminant plume. However, 13 C-enriched naphthalene (with a signature of 19.7&) in the groundwater of well 15, suggested biodegradation beyond the strictly reducing zone in direction toward the river. No conclusive stable isotope shift was obtained for acenaphthene. For comparison, at a former gas manufacturing plant in Southern Germany d13C shifts of 3.3 and 3.6& provided evidence for intrinsic anaerobic biodegradation of benzene and naphthalene (Griebler et al., 2004). At that site, acenaphthene formed long contaminant plumes with almost constant concentrations and insignificant stable carbon isotope shifts over a distance of more than 135 m (Zamfirescu and Grathwohl, 2001; Steinbach et al., 2004). This lack of evidence for in situ biodegradation of acenaphthene is in agreement with our findings (Fig. 2d).
3.3.
Quantitative assessment of in situ biodegradation
The BTEX plume originating from the source zone around piezometer D2b was delineated using the piezometric lines (as established by Batlle-Aguilar et al., 2009) in combination with 13 C ratios of the residual BTEX of the campaign 2006 (Fig. S1). Due to the continuous increase in the d13C values of residual benzene, toluene, ethylbenzene, and m-xylene with distance in the strongly reducing part of the aquifer, piezometers D2bis, D1p, U4, 11, and U13 were attributed to the same contaminant plume. Significantly more negative d-values in wells U6 and 12 (partially even below the isotope signatures at the source D2bis) suggested local secondary sources of contaminants with BTEX concentrations that were orders of magnitude below the concentration in the major source zone (well D1p). Wells U6 and 12 were consequently excluded from the quantitative assessment of in situ biodegradation. Mixing of contaminants originating from secondary sources with the main contaminant plume that is more enriched in 13C, would lead to slight underestimations of the in situ degradation. In the following moderately reducing to oxic aquifer section 90e160 m from the major source zone, no further 13C-enrichment in residual BTEX was observed. Occasionally, the residual monoaromatic contaminants showed more negative d13C ratios (Fig. S1). Well 15 was not considered for quantitative evaluation, since it was surrounded by a less reducing to oxic zone (Batlle-Aguilar et al., 2009).
4464
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 5 9 e4 4 6 9
Fig. 2 e aed: Concentrations of benzene, naphthalene, and acenaphthene and corresponding stable carbon isotope ratios determined in groundwater of the Fle´malle site in 2006. Benzene (outer circle), naphthalene (middle circle), and acenaphthene (inner circle) were chosen as representatives of mono-, di-, and tri-ring aromatic compounds. b was modified from Batlle-Aguilar et al. (2009).
100 Biodegradation [%]
The percentage of biodegradation was based on the 13Cenrichment in residual groundwater contaminants with distance from the source (Eqs. (2) and (3)). The results confirmed that a considerable part of benzene, toluene, and m-xylene was already biodegraded within the strictly reducing zone (Fig. 3). According to CSIA, the percentage of biodegradation was highest for m-xylene (>90% relative to D2bis); 80% of the initial benzene and toluene, and 40% of the initial ethylbenzene were removed by biodegradation. Due to decreasing d-values in the residual BTEX at 90e160 m distance from the source zone, it was not possible to quantify the biodegradation in this aquifer section using the stable isotopebased approach (Fig. 3). This is comparable to another anoxic aquifer, where 99% of the decrease in toluene and o-xylene could be attributed to intrinsic biodegradation based on the observed stable isotope shifts (Richnow et al., 2003a). Apart from the percentage of biodegradation, the kinetics of the anaerobic in situ biodegradation was determined. Firstorder rate constants (l) were calculated based on the continuous 13C-enrichment downgradient the source zone (Eq. (4)). For this, distances were converted into travel times using an intermediate groundwater flow velocity of 0.26 m/d. For the single stretches from D2bis to the individual piezometers
80
U13
60
8
7
U6
40
11 D1p
20 0
Reducing to oxic
Strongly reducing
0 25 (D2bis)
P5
U4 15
50
75
100
125
150
175
Distance from source [m]
Fig. 3 e Percentage of biodegradation of benzene (A), toluene (C), ethylbenzene (:), and m-xylene (-) along the groundwater flow path. Filled symbols indicate strictly reducingd, open symbols mildly reducing to oxic groundwater samples. Large symbols represent samples, which were incorporated in the first-order rate calculations (see text for details).
4465
240
178 3.9 10 R2 ¼ 0.96
3
2.9 103
187 3.7 103
770 0.9 10 R2 ¼ 0.80
3
1.4 103
491
4.6 727 1.0 103
161 187
4.3 10 R2 ¼ 0.99
3
3.7 10 R2 ¼ 0.74
3
175 4.0 103 178 3.9 103
165 4.2 103 2.6
1.3
337 2.1 103 0.0 1.1 371 1.9 103 0.0 1.1 185 3.7 103 0.0 1.0
152 389 117 208 4.6 103 1.8 103 5.9 103 3.3 103
Linear regression
t1/2 [d] l [/d] l [/d]
t1/2 [d]
Dd [&]
l [/d]
t1/2 [d]
Dd [&]
l [/d]
t1/2 [d]
Dd [&]
m-Xylene Ethylbenzene
Dd [&]
0.0 1.3 0.9 3.9 3.8 Average
Acenaphthyl methylsuccinate was not commercially available as reference compound; therefore, the potential first
D2bis D1p U4 11 U13
3.5. Tentative identification of acenaphthene methylsuccinate
Toluene
Screening campaigns for metabolites in groundwater samples were performed in 2005 and 2006. In the first year, a series of carboxylic acid and alcohol derivatives of aromatic compounds were detected in groundwater samples of wells D2bis and C3bis, however, no signature metabolites in the strict sense could be identified. Groundwater of more than 94 m distance from the source mostly was free of contamination with aromatic hydrocarbons (wells 1, 2, 7, 8, P4, and P6), and no potential metabolites of BTEX or PAHs were detected. In order to improve the sensitivity, larger volumes of groundwater were extracted in the second campaign. Methylsuccinyl-adducts of five different aromatic hydrocarbons and carboxylic acids of twelve different aromatic and heterocyclic compounds were identified in form of their methyl esters in groundwater samples of ten different wells (Table S5). The fumarate adduct of toluene was identified by comparison with an authentic benzylsuccinate standard and had a GC-retention time of 28.0 min. Methyl esters of the fumarate adducts of xylene, methyl- and dimethylnaphthalene, were tentatively identified based on published reference spectra without distinguishing between the different isomers (Table S5). Fumarate adducts of BTEX were only detected in samples of well D1p (source zone), whereas methylsuccinates of naphthalene and methylnaphthalene were extracted from several wells in the strictly reduced zone of the aquifer. Putative acenaphthyl methylsuccinate was present in groundwater from D2bis, D1p, and well 14 (for the identification of this compound, see paragraph below). Carboxylic acids that were potentially related to the degradation of BTEX, biphenyl, or PAHs occurred in groundwater of the source zone and close by, as well as in piezometers 7 and 15, both in the reducing section of the aquifer in 117 and 133 m distance from the source, respectively. Carboxylic acids of the heterocyclic compounds benzothiophene, benzofuran, indane, and indene were equally present in the source zone and in the strictly reducing groundwater sampled further downgradient (Table S5).
Benzene
3.4. Screening for signature metabolites of contaminant biodegradation in groundwater
Well
downgradient of the source, anaerobic in situ biodegradation rate constants were in the range of 1.4 103/d for ethylbenzene and 4.0 103/d for toluene (Table 1). Assessing firstorder biodegradation between D2bis and U13 via linear regression resulted in comparable rate constants. Recently, the stable isotope-based model was applied to determine first-order rate constants for o-xylene and naphthalene degradation at a former wood preservation plant (Blum et al., 2009). Under strictly reducing conditions, biodegradation of o-xylene proceeded at a rate of 2 103/d which was in the same range as the rate constants for monoaromatic compounds at the Fle´malle site. At the other site, it was possible to determine first-order rate constants of naphthalene of 4 103/d and 3 103/d for the anoxic and oxic sections of the plume, respectively (Blum et al., 2009).
Table 1 e First-order rate constants l and corresponding half-life times t1/2 of anaerobic in situ biodegradation of monoaromatic groundwater contaminants at the site of the former cokery of Fle´malle. Values were calculated between isotope signatures in groundwater of D2bis (source) and individual piezometers located along the groundwater flow path in the strongly reducing part of the aquifer. An intermediate groundwater flow velocity of 0.26 m/d was assumed (Batlle-Aguilar et al., 2009). Calculations were based on average isotope enrichment factors taken from the literature (Hunkeler and Morasch, 2010). For comparison, rate constants from linear regression analysis retrieved from the stable isotope-based first-order biodegradation model are shown.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 5 9 e4 4 6 9
4466
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 5 9 e4 4 6 9
Fig. 4 e a and b: Mass spectra of dimethyl esters of putative acenaphthyl methylsuccinate (A), and naphthyl methylsuccinate (B) tentatively identified by comparison with a reference spectrum from Annweiler et al. (2000).
fragmentation pattern of acenaphthyl methylsuccinate was compared to commercially available acenaphthene succinate. An absolute difference of mass fragments in 14 m/z was observed between the two respective dimethyl esters (Fig. 4a, Fig. S2a). Their GC-retention times were 41.7 and 40.5 min. Thirdly, several microcosms that had been incubated with 13 C-acenaphthene were solvent-extracted at the end of the incubation period in search of the metabolite additionally bearing six 13C-atoms. Operating the GCeMS in single ion mode for higher sensitivity, putative 13C6-acenaphthyl methylsuccinyl dimethyl ester was detected at 42.2 min (Fig. S2b). Although, several studies reported on the anaerobic degradation of acenaphthene in microcosm experiments, the degradation pathway has remained unknown (Mihelcic and Luthy, 1988; Rothermich et al., 2002; Chang et al., 2003; Yuan and Chang, 2007). The metabolite that was tentatively identified in this study suggested the introduction of a methyl group in the acenaphthene skeleton and a subsequent fumarate addition in analogy to the anaerobic degradation of naphthalene (Safinofski and Meckenstock, 2006; Foght, 2008).
115 133
165 127 152
141
178
193
207
195 221
153
20
213 239
40
286
60
181
80
167
226 252
100
312
Relative abundance [%]
metabolite of methyl acenaphthene was tentatively identified based on three independent lines of evidence. Firstly, mass spectra of putative acenaphthyl methylsuccinate were compared to those of naphthyl methylsuccinate; both converted to their dimethyl esters (Fig. 4a and b). Due to the additional ethylene bridge of the acenaphthene skeleton, both molecules had an absolute difference of 26 m/z in their respective mass peaks (312 and 286 m/z). This shift of 26 m/z was reflected in all major peaks of both MS-fractionation patterns. Comparing the relative abundances of the major fragments of naphthyl methylsuccinate- and putative acenaphthyl methylsuccinate dimethyl ester, a correlation coefficient of R ¼ 0.879 was obtained (Fig. 5). Secondly, the MS-
0
Eliminated fragment [m/z] Fig. 5 e Comparison of mass fragmentation patterns of naphthyl methylsuccinate (gray bars) and putative acenaphthyl methylsuccinate (black bars). On the x-axis displayed is the size of the eliminated fragment (m/z), in case of deviating fragment size, the first value refers to naphthyl methylsuccinate and the second to acenaphthyl methylsuccinate. Values on top of the bars are major mass fragments of the original GCeMS spectra.
3.6. Microcosms under in situ-like conditions with labeled substrates
13
C-
The intrinsic biodegradation potential was studied in sedimentegroundwater microcosms that were spiked with benzene, naphthalene, or acenaphthene in 13C-labeled form and incubated under in situ-like conditions. Sediments originated from four different depth layers of drilling location U13 in the strictly reducing part of the aquifer (Fig. 1). The content in d13C-CO2 in the headspace of all alive and dead controls stayed constant over the whole incubation experiment (Table S2). The aerobic degradation of benzene and naphthalene started within the first day, and the aerobic degradation of acenaphthene within less than one week after microcosm set-
4467
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 5 9 e4 4 6 9
Table 2 e Intrinsic first-order rate constants l derived from sedimentegroundwater microcosms at the site of the former cokery of Fle´malle. Drilling material from different depth layers of location U13 was used. The table shows mean values of duplicate incubations, except for 13C-benzene at 5 m, where one microcosm did not grow. n.d.: not determined. Depth [m]
5 7 9 12
Conditions
Oxic Anoxic Anoxic Anoxic
13
13
C-Benzene
13
C-Naphthalene
C-Acenaphthene
l [/d]
t1/2 [d]
l [/d]
t1/2 [d]
l [/d]
t1/2 [d]
2.1 102 1.2 103 1.6 103 0.9 103
33 581 422 771
1.5 103 0.3 104 0.7 104 1.1 104
457 28,250 9412 6104
8.7 103 3.4 104 n.d. n.d.
80 2056 n.d. n.d.
up. Under anoxic conditions, the mineralization of 13Cbenzene started within five days and reached a plateau after 48 days (158 days for the less permeable sediment from 7 to 8 m depth). Comparably, the anaerobic mineralization of naphthalene started within the first week of the incubation period but proceeded much slower than under oxic conditions. In sediments from two of the three examined anoxic depth layers, acenaphthene biodegradation started after a lag phase of more than 100 days. In anoxic sediment from 12 to 13 m depth, no intrinsic biodegradation potential for acenaphthene was detected over the whole duration of the microcosm experiment of 327 days (Table S2). Compared to Morasch et al. (2007), we applied an advanced, quantitative approach where first-order rate constants (l) of intrinsic contaminant degradation were determined based on an inorganic carbon mass balance (Eq. (8)). For sediments of the
unsaturated zone incubated in presence of air, l values were 2.1 102/d, 1.5 103/d, and 8.7 103/d for benzene, naphthalene, and acenaphthene (Table 2). These rates were equivalent to half-life times (t1/2) of 0.1, 1.3, and 0.2 years, respectively (Eq. (6)). In anoxic microcosms that contained sediments from the saturated zone, the 15e30 times lower rate constants corresponded to mean half-life times of 1.6, 40, and 5.6 years. Comparison of microcosm-derived first-order rate constants of benzene with the field-derived rates of the first approach revealed rates that were on average three times lower.
3.7.
Implications for the field site
Based on stable isotope generated first-order rate constants (Eq. (7)), predicted half-concentration distances (x1/2) of BTEX under anoxic conditions were between 15 and 32 m which is
Table 3 e Comparison of the three different approaches applied in this study summarizing their potentials and limits proving the in situ biodegradation of benzene, naphthalene, and acenaphthene. The quantitative and qualitative assessment of the three approaches is divided by a slash; D designates appropriate and L designates inappropriate approaches, ± stands for limited applicability. Stable isotopes (field)
13
C-microcosms
Signature metabolites
þ/þ
/b
þ/þ
þ/a
/þ
þ/þ
/
/c
þ/d
Restrictions
Fast Multi compound For smaller molecules
Slow, long term Single compound Time-consuming
Needs
Experimental 3 value
Prospects
Determination of reaction mechanisms from isotope effect
Fast Multi compound For compounds with signature metabolites Knowledge of degradation pathway Identification of new metabolites
Assessment
Gray fields mark the most suitable options. a Method successful in another study. b Metabolites are ambiguous. c New metabolites postulated based on this method. d Duration of experiment in parts too short for rate determination. e Stable isotope probing e for a review, see Madsen, 2006.
13C substrates Detection of metabolites Microcosms can be used for SIPe
4468
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 5 9 e4 4 6 9
in agreement with the flow and transport model of benzene (Batlle-Aguilar et al., 2009). In contrast, predictions based on the half-life times of benzene, naphthalene, and acenaphthene in anoxic microcosms resulted in x1/2 of 100, 412, and 16 m, respectively. The groundwater flow path from the contaminant source D2bis to the river bank is approximately 220 m long and the first 100 m are under anoxic conditions. For the following oxic plume interval, shorter half-concentration distances of 6, 13, and 1 m for benzene, naphthalene, and acenaphthene were predicted based on microcosmderived rate constants that were corrected for contaminant retardation in groundwater. Event-based infiltration of surface water supplies additional O2 in this part of the aquifer (Batlle-Aguilar et al., 2009). Predictions on the basis of stable isotopes matched the observations from the field: neither BTEX nor PAHs were ever detected in piezometers close to the river bank (Table S3).
4. Synthesis e complementarity of approaches In a recent review on the assessment of in situ biodegradation, Bombach et al. (2010) recommended using several approaches in combination according to the local conditions. In the present study, we employed three approaches in order to gain qualitative and quantitative information on the in situ biodegradation of BTEX and PAHs (Table 3). What distinguishes our combination from many others is its applicability to a wider variety of contaminants e independent of molecule size and hydrophobicity. CSIA of groundwater pollutants, the first technique that we applied, is useful for collecting data on the anaerobic in situ biodegradation of several BTEX compounds at once. Using CSIA, biodegradation rates of BTEX can be determined and the fate of naphthalene can be assessed in a qualitative way. However, substituted naphthalenes and larger PAHs cannot be examined. In practice, the CSIA-based field approach is barely applicable to study the intrinsic biodegradation potential of polyaromatic compounds because the bulk isotope effect is below the detection limit of the method (Elsner, 2010). Signature metabolite analysis, the second approach, bears the potential to identify new degradation intermediates and pathways, as presented in this study for the anaerobic degradation of acenaphthene and elsewhere for heterocyclic compounds (Safinofski et al., 2006). Nevertheless, its biggest potential lies in a reliable detection of the anaerobic biodegradation of (methylated) aromatic- and aliphatic hydrocarbons as well as heterocyclic compounds. The signature metabolite approach thus provides additional qualitative insights into the biodegradation of larger compounds where CSIA is not applicable. Microcosms with 13C-labeled substrates, the third approach, allow the quantitative assessment of biodegradation for any 13C-labeled compound of interest (Table 3). Their substrate specificity combined with very sensitive detection, makes 13C-microcosms a particularly interesting option for compounds that sorb, are rather recalcitrant, or cannot be studied by CSIA. Moreover, novel 13C-labeled metabolites of
the specific substrate may be extracted from the microcosms and provide new insights into degradation pathways. Even though 13C-microcosms may also be used as stand-alone technique, combination with CSIA and signature metabolite analysis in the field overcomes the limitation of substrate specificity and allows conclusions on a wider spectrum of contaminants.
Acknowledgements This work was funded by the EU integrated project Aquaterra. We thank M. Aragno for providing lab space and J. BatlleAguilar and S. Brouye`re for coordinating work on site. F. Chatelain is acknowledged for technical assistance and D. Grandjean for assistance at the GCeMS. We thank three anonymous reviewers for their valuable comments and suggestions.
Appendix. Supplementary data Supplementary data associated with this article can be found in the online version, at 10.1016/j.watres.2011.05.040.
references
Annweiler, E., Materna, A., Safinofski, M., Kappler, A., Richnow, H.H., Michaelis, W., Meckenstock, R.U., 2000. Anaerobic degradation of 2-methylnaphthalene by a sulfate-reducing enrichment culture. Applied and Environmental Microbiology 66, 5329e5333. Batlle-Aguilar, J., Brouye`re, S., Dassargues, A., Morasch, B., Hunkeler, D., Ho¨hener, P., Diels, L., Vanbroekhoven, K., Seuntjens, P., Halen, H., 2009. Benzene dispersion and natural attenuation in an alluvial aquifer with strong interactions with surface water. Journal of Hydrology 369, 305e317. Beller, H.R., 2002. Analysis of benzylsuccinates in groundwater by liquid chromatography/tandem mass spectrometry and its use for monitoring in situ BTEX biodegradation. Environmental Science and Technology 36, 2724e2728. Beller, H.R., Kane, S.R., Legler, T.C., McKelvie, J.R., Sherwood Lollar, B., Pearson, F., Balser, L., MacKay, D.M., 2008. Comparative assessments of benzene, toluene, and xylene natural attenuation by quantitative polymerase chain reaction analysis of a catabolic gene, signature metabolites, and compound-specific isotope analysis. Environmental Science and Technology 42, 6065e6072. Blum, P., Hunkeler, D., Weede, M., Beyer, C., Grathwohl, P., Morasch, B., 2009. Quantification of biodegradation for oxylene and naphthalene using first order decay models, MichaeliseMenten kinetics and stable carbon isotopes. Journal of Contaminant Hydrology 105, 118e130. Bombach, P., Richnow, H.H., Kastner, M., Fischer, A., 2010. Current approaches for the assessment of in situ biodegradation. Applied Microbiology and Biotechnology 86, 839e852. Bosma, T.N.P., Middeldorp, P.J.M., Schraa, G., Zehnder, A.J.B., 1997. Mass transfer limitation of biotransformation: quantifying bioavailability. Environmental Science and Technology 31, 248e252. Chang, B.V., Chang, S.W., Yuan, S.Y., 2003. Anaerobic degradation of polycyclic aromatic hydrocarbons in sludge. Advances in Environmental Research 7, 623e628.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 5 9 e4 4 6 9
Chapelle, F.H., Bradley, P.M., Lovley, D.R., Vroblesky, D.A., 1996. Measuring rates of biodegradation in a contaminated aquifer using field and laboratory methods. Ground Water 34, 691e698. Coates, J.D., Anderson, R.T., Lovley, D.R., 1996. Oxidation of polycyclic aromatic hydrocarbons under sulfate-reducing conditions. Applied and Environmental Microbiology 62, 1099e1101. Elshahed, M.S., Gieg, L.M., McInerney, M.J., Suflita, J.M., 2001. Signature metabolites attesting to the in situ attenuation of alkylbenzenes in anaerobic environments. Environmental Science and Technology 35, 682e689. Elsner, M., 2010. Stable isotope fractionation to investigate natural transformation mechanisms of organic contaminants: principles, prospects and limitations. Journal of Environmental Monitoring 12, 2005e2031. Foght, J., 2008. Anaerobic biodegradation of aromatic hydrocarbons: Pathways and prospects. Journal of Molecular Microbiology and Biotechnology 15, 93e120. Griebler, C., Safinofski, M., Vieth, A., Richnow, H.H., Meckenstock, R.U., 2004. Combined application of stable carbon isotope analysis and specific metabolites determination for assessing in situ degradation of aromatic hydrocarbons in a tar oil-contaminated aquifer. Environmental Science and Technology 38, 617e631. Hunkeler, D., Ho¨hener, P., Zeyer, J., 2002. Engineered and subsequent intrinsic in situ bioremediation of a diesel fuel contaminated aquifer. Journal of Contaminant Hydrology 59, 231e245. Hunkeler, D., Morasch, B., 2010. Isotope fractionation during transformation processes. In: Aelion, C.M., Ho¨hener, P., Hunkeler, D., Aravena, R. (Eds.), Environmental Isotopes in Biodegradation and Bioremediation. Taylor & Francis Group, Boca Raton, pp. 79e125. Langenhoff, A.A.M., Zehnder, A.J.B., Schraa, G., 1996. Behaviour of toluene, benzene and naphthalene under anaerobic conditions in sediment columns. Biodegradation 7, 267e274. Madsen, E.L., 2006. The use of stable isotope probing techniques in bioreactor and field studies on bioremediation. Current Opinion in Biotechnology 17, 92e97. Mazeas, L., Budzinski, H., Raymond, N., 2002. Absence of stable carbon isotope fractionation of saturated and polycyclic aromatic hydrocarbons during aerobic bacterial biodegradation. Organic Geochemistry 33, 1259e1272. Mihelcic, J.R., Luthy, R.G., 1988. Microbial degradation of acenaphthene and naphthalene under denitrification conditions in soilewater systems. Applied and Environmental Microbiology 54, 1188e1198. Morasch, B., Ho¨hener, P., Hunkeler, D., 2007. Evidence for in situ degradation of mono- and polyaromatic hydrocarbons in alluvial sediments based on microcosm experiments with 13 C-labeled contaminants. Environmental Pollution 148, 739e748.
4469
Morasch, B., Ho¨hener, P., Hunkeler, D., 2011. Determination of in situ biodegradation rates using 13C-labeled aniline. In: Schirmer., M., Hoehn, E., Vogt, T. (Eds.), GQ10: Groundwater Quality Management in a Rapidly Changing World. IAHS, Zu¨rich, Switzerland, pp. 287e290. National Research Council (NRC), 2000. Natural Attenuation for Groundwater Remediation. In: Remediation, C.o.I., Board, W.S. a.T., Management, B.o.R.W., Commission on Geosciences, E., Resources (Eds.). National Academy Press, Washington, D.C. Phelps, C.D., Battistelli, J., Young, L.Y., 2002. Metabolic biomarkers for monitoring anaerobic naphthalene biodegradation in situ. Environmental Microbiology 4, 532e537. Richnow, H.H., Annweiler, E., Michaelis, W., Meckenstock, R.U., 2003a. Microbial in situ degradation of aromatic hydrocarbons in a contaminated aquifer monitored by carbon isotope fractionation. Journal of Contaminant Hydrology 65, 101e120. Richnow, H.H., Meckenstock, R.U., Ask, L., Baun, A., Ledin, A., Christensen, T.H., 2003b. In situ biodegradation determined by carbon isotope fractionation of aromatic hydrocarbons in an anaerobic landfill leachate plume (Vejen, Denmark). Journal of Contaminant Hydrology, 59e72. Rothermich, M.M., Hayes, L.A., Lovley, D.R., 2002. Anaerobic, sulfate-dependent degradation of polycyclic aromatic hydrocarbons in petroleum-contaminated harbor sediment. Environmental Science and Technology 36, 4811e4817. Safinofski, M., Griebler, C., Meckenstock, R.U., 2006. Anaerobic cometabolic transformation of polycyclic and heterocyclic aromatic hydrocarbons e evidence from laboratory and field studies. Environmental Science and Technology 40, 4165e4173. Safinofski, M., Meckenstock, R.U., 2006. Methylation is the initial reaction in anaerobic naphthalene degradation by a sulfatereducing enrichment culture. Environmental Microbiology 8, 347e352. Schwarzenbach, R.P., Gschwend, P.M., Imboden, D.M., 2003. Environmental Organic Chemistry. John Wiley & Sons, New York. SRC Inc., 2009. SRC Physical Properties Database. Available from: http://www.syrres.com. Steinbach, A., Seifert, R., Annweiler, E., Michaelis, W., 2004. Hydrogen and carbon isotope fractionation during anaerobic biodegradation of aromatic hydrocarbons: a field study. Environmental Science and Technology 38, 609e616. Yuan, S.Y., Chang, B.V., 2007. Anaerobic degradation of five polycyclic aromatic hydrocarbons from river sediment in Taiwan. Journal of Environmental Science and Health Part B e Pesticides Food Contaminants and Agricultural Wastes 42, 63e69. Zamfirescu, D., Grathwohl, P., 2001. Occurrence and attenuation of specific organic compounds in the groundwater plume at a former gasworks site. Journal of Contaminant Hydrology 53, 407e427.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 7 0 e4 4 8 2
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Determination of sorption of seventy-five pharmaceuticals in sewage sludge Maritha Ho¨rsing a,1,*, Anna Ledin a,1, Roman Grabic b,2, Jerker Fick b, Mats Tysklind b, Jes la Cour Jansen c, Henrik R. Andersen a a
Department of Environmental Engineering, Technical University of Denmark, B113, DK-2800 Kgs Lyngby, Denmark Department of Chemistry, Umea˚ University, SE-901 87 Umea˚, Sweden c Water and Environmental Engineering at Department of Chemical Engineering, Lund University, P.O. Box 124, SE-221 00 Lund, Sweden b
article info
abstract
Article history:
Sorption of 75 active pharmaceutical ingredients (APIs) to three different types of sludge
Received 14 December 2010
(primary sludge, secondary sludge with short and long sludge age respectively) were
Received in revised form
investigated. To obtain the sorption isotherms batch studies with the APIs mixture were
26 May 2011
performed in four nominal concentrations to water containing 1 g of sludge. The range of
Accepted 28 May 2011
APIs concentrations was between ng L1 to mg L1 which are found in the wastewater
Available online 14 June 2011
effluents. Isotherms were obtained for approximately 45 of the APIs, providing distribution coefficients for linear (Kd), Freundlich (Kf) and Langmuir (KL) isotherms. Kd, Kf and KL ranging
Keywords:
between 7.1 104 and 3.8 107, 1.1 102 and 6.1 104 and 9.2 103 and 1.1 L kg1,
Distribution coefficients (Kd)
respectively. The obtained coefficients were applied to estimate the fraction of APIs in the
Pharmaceuticals
water phase (see Abstract Graphic). For 37 of the 75 APIs, the predicted presence in the
Sorption isotherms
liquid phase was estimated to >80%. 24 APIs were estimated to be present in the liquid
Sludge
phase between 20 and 80%, and 14 APIs were found to have <20% presence in the liquid
Wastewater treatment plant
phase, i.e. high affinity towards sludge. Furthermore, the effect of pH at values 6, 7 and 8
(WWTP)
was evaluated using one way ANOVA-test. A significant difference in Kds due to pH changes were found for 6 of the APIs (variation 10e20%). ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
The presence of human and veterinary pharmaceuticals in the environment has been recognized as a potential environmental threat (Ternes et al., 2002). After their use, pharmaceuticals are either excreted unchanged or as metabolites via urine and faeces or are washed off and subsequently reach a wastewater treatment plant (WWTP) via the sewage system. The active pharmaceutical ingredients (APIs) are designed to
have pharmacological effects at low concentrations, which have lead to concerns regarding their distribution in aquatic environments and potential non-wanted biological effects in different organisms. APIs have been found in several environmental compartments such as waste, surface and ground waters (Calisto and Esteves, 2009; Fick et al., 2009; Gabet-Giraud et al., 2010; Lindberg et al., 2010), as well as in sludge and sediments (Ternes et al., 2002; Andersen et al., 2003). Sorption of APIs to sludge, during
* Corresponding author. Tel.: þ46 462228212. E-mail address:
[email protected] (M. Ho¨rsing). 1 Present address: Water and Environmental Engineering at Department of Chemical Engineering, Lund University, P.O. Box 124, SE221 00 Lund, Sweden. 2 Present address: University of South Bohemia in Ceske Budejovice, Faculty of Fisheries and Protection of Waters, South Bohemian any, Czech Republic. Research Center of Aquaculture and Biodiversity of Hydrocenoses, Za´ti sı´ 728/II CZ 389 25 Vodn 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.05.033
4471
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 7 0 e4 4 8 2
wastewater treatment processes in which sludge is separated from the wastewater stream may be an important factor for the removal of non-biodegradable pharmaceuticals from WWTPs. Solid matter is separated from wastewater during treatment. After the separation of incoming particles, referred to as primary sludge, the remaining organic material is generally biologically removed in an activated sludge system. After treatment, sludge is separated from the treated water by a second sedimentation step, secondary sludge. pH may have impact on the sorption behavior. This could be due to structural changes of the surface of the sludge, such as changes in surface charge or tertiary structure. Surveys in which pKa (acid-ionization constant) have been determined for sludge, reported pKa values of 5.3 and 9.5 for dissolved organic matter (DOM) in the sludge (Wang et al., 1998) and 6.1 for sludge particles (Wang et al., 2000). It could also be due to protolysation of the APIs. The effect of pH on the sorption of APIs has been discussed in the literature (e.g. Ternes et al., 2004; Urase and Kikuta, 2005; Jones et al., 2006; Carballa et al., 2008) but without firm conclusions. Temperature may also have impact on the sorption behavior. Comprehensive studies where this has been studied are limited. In a study by Zeng et al. (2009) the difference in Kd for 17a-ethinylestradiol (EE2) between 10, 20 and 30 C was found to be 20e25%, i.e. the lower temperature having the highest Kd. Also Ifelebuegu et al. (2010) showed a 20% increase in Kd for EE2 by decreasing temperature from 20 to 15 C. The objective of this study was to experimentally determine Kd values for 75 selected APIs to different types of sewage sludge. The Kd values were obtained from sorption isotherms in three different types of sludge, a) primary sludge, and b) secondary sludge with long and c) short sludge age, respectively. The obtained Kd values were applied in order to estimate the removal of APIs by sorption to sludge in WWPTs. Furthermore, the influence on sorption caused by normal pH variations (pH 6e8) found in WWTPs was investigated.
2.
Materials and methods
2.1.
Sludge
Sludge was collected from three WWTPs which were representative of WWTPs industrialised countries for treatment of municipal wastewater.
Primary sludge was collected from Avedøre WWTP, which treats wastewater corresponding to 275 000 person equivalents (PE) from 10 municipalities in the southern part of Greater Copenhagen, Denmark. This plant was chosen since it was possible to collect primary sludge without contamination from internal recirculation of secondary sludge. Suspended solids (SS) and volatile suspended solid (VSS) were 43 and 35 g L1, respectively. Secondary sludge with a long sludge age, approximately 10 days, and nitrification was collected from Klagshamn WWTP. This plant treats municipal wastewater corresponding to about 60 000 PE from the southern part of Malmo¨ City, Sweden and a nearby municipality. The plant uses primary precipitation and nitrification in activated sludge (sludge content 3 g L1). SS and VSS were 4 and 3 g L1, respectively. Denitrification takes place in a post-denitrifying, moving bed biofilm reactor. Biological sludge with short sludge age, 2e3 days, was collected from Sjo¨lunda WWTP. This is typical for plants in the north and in many inland cities in Sweden. The plant treats wastewater corresponding to approximately 315 000 PE from Malmo¨ city, Sweden and surrounding municipalities. The plant was operated with combined primary and simultaneous precipitations. Removal of organic matter take place in a highly loaded activated sludge plant where the sludge was collected (sludge content 6 g L1). SS and VSS were 8 and 6 g L1, respectively. Nitrification takes place in a fixed bed trickling filter and denitrification in a moving bed biofilm reactor. Each portion of sludge was washed twice with tap water, decanted in order to remove water soluble constituents and frozen at 18 C. The sludge was gently freeze-dried in order to preserve the structure and sterilized by heating at 103 C for minimum 3 h.
2.2.
Experimental design
Nominal concentrations of the 75 APIs (see Table 1) included in the experiments were chosen based on the solubility of the APIs, limit of quantification (LOQ) and linear range of the analytical method (Grabic et al., unpublished data; Fick et al., 2009). Volatilisation was considered (KH; Table S2 supplementary data) by calculations in EPI Suite software HenryWin v3.10. All APIs had a KH < 9 107 Atm m3 mol1, therefore volatilisation would be insignificant, since only compounds with KH > 3 103 are considered to be volatile (Ternes and Joss, 2008). The aim was to get four API equilibriums with water concentration (Cw; g L1) in the range of
Table 1 e The experimental set up, including API concentrations, pH sludge concentration and the number of bottles per blank/zero sludge/sludge. From each bottle triplicate solid phase extractions (SPE) were made. API conc. (mgL1) Blank for each sludge No sludge Primary sludge Avedøre Secondary sludge short sludge age Sjo¨lunda Secondary sludge long sludge age Klagshamn
0 0.08 0.08 0.08 10 10 10 0.08 10
0.4 0.4 0.4
2 2 2
10 10 10
0.4
2
10
pH
Sludge conc. (gL1)
No. of bottles
7 7 7 7
1 0 1 1 10 50 1
3 4 4 4 1 1 1 4 1
6 7 8
4472
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 7 0 e4 4 8 2
Table 2 e Sorption isotherms obtained during sorption to primary sludge. P is the significance that the linear model has a better fit than another model tested, the model tested was Freundlich and Langmuir. Linear model 1
Kd (L kg ) Afluzosin Amitryptiline Atenolol Atracurium Azelastine Biperiden Bupropion Chloprothixene Citalopram Clomipramine Clonazepam Clotrimazol Cyproheptadine Desloratidine Dicycloverin Donepezil Duloxetine Etonogestrel Ezetimibe Fexofenadine Fluoxetine Flutamide Glibenclamide Glimepiride Haloperidol Hydroxyzine Irbesartan Ketoconazole Loperamide Maprotiline Megesterol Mianserin Nefazodone Oxazepam Paroxetine Pizotifen Progesterone Repaglinide Risperidone Sertraline Sulfamethoxazol Telmisartan Tramadol Trimethoprim Verapamil
3
1.8 10 4.1 103 4.6 102 3.5 102 6.4 103 8.2 102 85 3.8 104 5.4 102 1.7 104 5.7 102 3.2 104 1.1 104 3.7 103 1.4 103 3.6 103 1.3 104 No fit 2.3 103 2.7 103 1.0 104 1.5 103 3.6 103 2.1 103 1.0 104 1.2 103 7.0 102 9.7 103 1.4 104 6.7 103 No fit 3.0 103 1.4 104 7.9 102 1.4 104 4.7 103 7.5 102 1.7 102 1.9 103 3.5 104 3.2 102 1.3 103 1.1 102 3.9 102 1.8 103
Freundlich model
2
R (%)
1
Kf (L g )
n
2
R (%)
Langmuir model P (%)
99 98 88 100 82 88 98 98 76 99 96 91 98 99 92 96 77
2.7 3.1 4.0 0.24 46 1.3 4.8 103 8.3 0.97 7.6 0.10 19.0 2.2 1.9 1.2 10.4 30.8
1.1 0.96 1.3 0.96 1.4 1.1 0.76 0.77 1.1 0.88 0.84 0.92 0.81 0.92 0.98 1.2 1.2
99 98 91 100 86 88 99 99 76 99 96 92 91 100 94 96 78
42 70 17 40 11 83 0.99 0.98 83 8.5 33 59 30 17 90 21 53
97 82 99 88 77 92 76 98 92 88 98 99
4.9 102 1.1 105 2.7 0.13 No fit 1.9 103 60.9 3.2 1.5 102 1.6 0.60 1.7
0.68 0.38 0.83 0.77
99 94 99 90
1.1 0.15 3.6 27
0.53 1.4 1.1 0.70 0.79 0.69 0.83
97 80 99 93 90 99 100
0.8 21 14 19 30 0.04 0.05
81 98 90 96 100 98 94 99 97 77 100 94 98 99
12.5 2.7 0.80 5.6 No fit 0.18 1.4 3.1 4.1 0.91 0.86 4.9 104 0.43 1.1
1.2 0.79 1.0 0.87
83 99 90 97
36 4.0 99 28
0.86 1.3 1.1 0.72 1.1 0.95 0.63 1.0 0.94
99 95 99 99 77 100 96 98 99
14 19 36 0.4 79 51 3.3 91 26
90% of the starting concentration (C0; g L1) and the LOQ. The water solubility should be larger than the starting concentration; Cw C0; LOQ < Cw. Sludge concentration in the experiments was 1 g L1, with exception for one sludge where it was increased to 10 and 50 g L1 (Table 1). The 1 g L1 sludge density was about 5- to 10-fold more than the realistic sludge production that can occur in a WWTP which ensures that APIs that adsorb significantly to sludge in a real WWTP also would be removed significantly from the water phase in the batch experiment which was the basis for determining a Kd value, while low sorbing APIs Kds will not be determined.
1
smax (L g ) 5.5 No fit 5.7 No fit 2.0 2.4 No fit No fit 1.0 No fit No fit No fit No fit No fit 6.2 1.9 1.6 No fit No fit No fit No fit No fit No fit No fit 1.5 1.9 No fit No fit No fit No fit No fit 1.6 No fit 2.1 No fit No fit No fit 7.6 3.2 No fit 8.6 No fit No fit 3.7 No fit
R2 (%)
KL
P (%)
5
99
31
103
1.4 104
90
18
104 104
5.8 104 4.6 105
89 89
3.5 46
104
6.7 105
104 104 104
2.7 105 2.6 104 1.1 103
94 97 79
66 7.3 37
104 104
1.2 103 8.3 105
81 99
12 16
104
2.9 104
85
17
104
4.3 105
90
71
103 104
2.8 105 7.0 105
94 99
53 17
103
5.4 105
78
58
104
1.1 105
98
73
10
4
3.7 10
7.7
59
A stock solution including the 75 API of 0.1 g L1 was prepared in MeOH from which the four MeOH stock solutions of 0.4, 2.0, 10.0 and 50.0 mg L1 were prepared, respectively. These stock solutions suited the design (Table 1) and were theoretically determined to fit the criteria for the present study. Bo¨hm and During (2010) showed that there was no significant difference between determination of the distribution coefficient KDOC for single compounds or mixtures. Artificial media was used in the study for the water phase (Berg and Nyholm, 1996), modified as described by Andersen et al. (2005). The artificial sewage was a phosphate-buffered mineral media containing Ca2þ, Kþ, Mg2þ, Naþ, Cl and SO42.
4473
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 7 0 e4 4 8 2
Table 3 e Sorption isotherms obtained during sorption to secondary sludge long sludge age P is the significance that the linear model has a better fit than another model tested, the model tested was Freundlich and Langmuir. Linear model 1
Kd (L kg ) Afluzosin Alprazolam Amitryptiline Atenolol Atracurium Azelastine Biperiden Bisoprolol Bupropion Chloprothixene Citalopram Clomipramine Clotrimazol Cyproheptadine Desloratidine Dicycloverin Donepezil Duloxetine Eprosartan Estradiol Etonogestrel Ezetimibe Fexofenadine Fluoxetine Flutamide Glibenclamide Glimepiride Haloperidol Hydroxyzine Irbesartan Ibuprofen Ketoconazole Levonorgestrel Loperamide Maprotiline Medroxyprogesterone Megesterol Mianserin Nefazodone Orphenadrine Oxazepam Paroxetine Pizotifen Progesterone Repaglinide Risperidone Sertraline Sotalol Sulfamethoxazol Telmisartan Tramadol Trimethoprim Venlafaxine Verapamil
3
10 102 103 103 102 103 102 102 102 104 102 103 104 103 103 103 102 103 71 No fit No fit 3.0 103 3.6 102 6.0 103 7.5 102 1.3 103 9.6 102 2.9 103 7.2 102 9.4 102 3.6 102 8.5 103 2.6 102 5.5 103 4.5 103 1.2 102 5.9 102 9.1 102 8.3 103 6.4 102 1.1 103 8.3 103 3.1 103 1.1 103 2.1 102 6.5 102 1.7 104 3.6 102 3.7 102 No fit 1.9 102 4.2 102 1.0 102 4.0 102 1.2 7.4 2.8 1.6 4.7 2.0 7.5 1.1 1.4 2.0 2.1 6.7 3.4 3.6 2.9 1.7 9.7 2.9
2
R (%)
Freundlich model 1
Kf (L kg )
100 94 99 94 100 99 98 4.4 99 98 94 100 96 100 100 99 99 98 93
2.2 1.6 102 1.9 22 0.37 3.2 0.13 46 0.35 9.4 0.90 3.4 4.5 3.7 2.1 0.50 4.6 2.8 0.43
96 95 99 90 93 99 98 98 94 91 91 61 97 99 70 82 99 96 99 87 97 100 87 17 98 92 99 96
3.0 0.2 1.3 1.5 0.2 0.2 3.5 0.54 5.3 1.6 0.70 18 0.53 0.65 5.2 6.7 1.4 0.96 0.12 11 0.63 1.9 8.6 10 2.8 1.3 0.56 5.2
99 99 85 84
102
104
104 102
103
9.0 102 0.18 0.77 20
2
n
R (%)
Langmuir model P (%)
1
smax(L kg ) 4
4.0 4.6 40 0.0 40 31 7.7 0.14 4.2 14 25 1.4 2.2 93 40 13 0.0 98 38
3.7 No No 7.8 No 5.8 No 9.9 5.3 No 3.8 No No No No No 1.0 No 1.8
10 fit fit 103 fit 104 fit 102 103 fit 103 fit fit fit fit fit 104 fit 103
0.63 0.93 0.83 0.50 0.82 0.83 1.0 0.97 0.54 0.73 0.73 1.9 103 0.77 0.79 1.8 1.4 1.1 0.76 0.84 1.4 0.73 0.94 1.3 1.8 1.2 0.71 1.1 0.68
100 96 100 99 100 99 98 68 100 99 95 100 98 100 100 99 100 98 94 0.0 0.0 99 96 100 96 99 99 98 98 97 94 95 81 98 100 81 85 99 98 98 91 98 100 90 98 99 97 99 98
0.0 74 0.80 0.13 8.2 11 78 68 1.2 38 0.8 1.4 0.83 0.05 0.08 16 45 1.6 2.6 9.2 0.07 26 17 0.0 1.1 1.3 35 5.6
No No No No No No 1.1 No No No No 1.9 No No 9.5 5.3 2.6 No No 6.5 No No 7.4 2.7 8.9 No 1.3 No
fit fit fit fit fit fit 105 fit fit fit fit 103 fit fit 102 103 104 fit fit 103 fit fit 103 103 103 fit 104 fit
0.92 0.91 1.3 1.8
99 99 86 96
39 16 32 0.0
No No 2.1 3.3
fit fit 103 103
1.1 0.69 0.95 1.5 0.97 1.1 0.84 3.0 1.1 0.89 1.2 0.91 0.73 1.0 0.96 0.88 1.2 1.0 1.2
Sludge was added to 1 L borosilicate glass bottles. In order to inhibit microbial growth oxygen was removed by purging with N2(g) for 1 min and Na2SO3 was added to a final concentration of 50 mg L1 to each bottle. The bottles were left on stirring in the dark at þ4 C in order for the sludge to
KL
R (%)
P (%)
5
100
1.3
4.9 104
99
0.0
3.9 105
99
1.3 103 3.5 105
86 100
8.1 105
95
1.5 104
100
6.6 105
94
32
2.8 105
98
76
6.6 104
87
4.4 104 2.4 104 4.0 105
85 87 99
3.8 7.1 29
3.6 104
92
5.2
2.9 104 2.8 104 1.2 104
92 99 99
5.4 0.0 0.15
3.3 105
99
14
7.9 105 4.3 104
0.0 98
25 0.0
4.0 10
22 0.0 1.4 20
0.0
0.23
rehydrate. After 12 h, 200 mL of the API stock solutions of 0.4, 2.0, 10.0 or 50.0 mg L1 were added to the bottles using a Hamilton syringe giving the final concentrations presented in Table 1. The bottles were left on stirring in the dark at þ4 C for 12 h. Experiments were performed at three pH values
4474
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 7 0 e4 4 8 2
Table 4 e Sorption isotherms obtained during sorption to secondary sludge short sludge age. P is the significance that the linear model has a better fit than another model tested, the model tested was Freundlich and Langmuir. Linear model 1
Kd (L kg ) Alprazolam Amitryptiline Atenolol Atracurium Azelastine Biperiden Bisoprolol Bupropion Chloprothixene Clomipramine Clotrimazol Cyproheptadine Desloratidine Dicycloverin Diltiazem Duloxetine Estradiol Etonogestrel Ezetimibe Fexofenadine Fluoxetine Flutamide Glibenclamide Glimepiride Haloperidol Hydroxyzine Irbesartan Ibuprofen Ketoconazole Levonorgestrel Loperamide Maprotiline Medroxyprogesterone Megesterol Mianserin Nefazodone Orphenadrine Oxazepam Paroxetine Pizotifen Progesterone Repaglinide Risperidone Sertraline Sotalol Sulfamethoxazol Telmisartan Trimethoprim Verapamil
2
4.3 10 2.8 103 1.9 103 6.1 102 1.4 103 8.4 102 94 2.0 102 No fit 7.3 103 No fit 5.3 103 3.2 103 1.7 103 4.4 102 3.2 103 2.3 102 2.4 102 8.5 103 6.7 102 5.7 103 1.2 103 2.3 103 2.6 103 1.7 103 6.0 102 No fit 2.0 102 No fit 2.0 102 1.1 104 3.9 103 2.5 102 8.3 102 5.2 102 8.9 103 5.4 102 1.6 103 8.6 103 3.1 103 1.1 103 5.1 102 3.3 102 No fit 7.4 102 2.8 102 No fit 2.8 102 6.3 102
2
Freundlich model 1
R (%)
Kf (L kg )
80 96 100 98 96 96 64 97
No fit 4.2 102 No fit 2.9 102 1.7 102 0.13 7.5 1.9 102
88 97 96 97 100 80 72 73 87 94 91 93 97 94 98 95
Langmuir model 1
smax(L kg )
n
R (%)
P (%)
0.66
98
0.27
0.73 0.65 0.82 2.0 0.79
99 99 96 79.6 98
6.0 103
0.49
97
0.0
0.23 0.10 0.11 0.65 1.7 0.16 0.11 0.72 2.5 1.4 1.6 0.16 9.1 0.90 0.27
0.70 0.69 0.75 1.0 0.37 0.97 0.92 0.98 0.60 0.47 0.49 0.75 0.70 1.0 0.91
99 99 98 100 99 72 73 87 98 99 99 98 94 96 95
0.0 0.22 3.1 35 0.0 93 87 91 0.54 0.0 0.0 1.2 2.0 75 56
No fit No fit No fit 5.2 104 No fit 1.6 104 No fit 9.4 104 No fit No fit No fit No fit No fit No fit No fit
81
62 61 0.88 3.1 65 66 22 0.0 25 40 0.0 0.0 20 28 45
105
103 103 104 102
0.1 0.0 26 2.0 6.2
80
1.1
1.3
86 91 91 96 86 94 92 98 97 82 98 96 98 98
0.40 0.16 9.3 0.46 0.30 6.6 3.0 0.17 3.1 5.6 0.20 0.21 1.3 0.15
1.1 0.62 0.66 1.1 0.89 0.80 0.54 0.88 1.1 0.37 0.74 0.83 1.1 0.92
83 96 95 96 86 95 99 98 97 98 1005 96 98 98
99 95
0.22 No fit
0.87
99
6.3
98 96
1.3 102 0.90
0.75 1.0
100 96
0.15 75
102
102 102
10
5
(Table 1). The pH 7.0 was chosen based on a typical pH in WWTPs and pH 6.0 and 8.0 were based on low and high values from WWTPs. Control batches without APIs in the sludge and standard APIs solutions were prepared (Table 1).
2.3.
2
Extraction and chromatography
Triplicate extractions were made from each 1 L borosilicate glass bottle. After 12 h of stirring, the samples were allowed to
KL
R (%)
P (%)
3.4 104
81
1.2
9.2 106
100
27
1.6 105
72
85
1.0 104
87
65
2.4 103
1.3 104
82
56
1.1 104 No fit No fit 9.7 103 No fit No fit No fit No fit 2.3 104 No fit No fit No fit 2.5 104 No fit
1.9 105
83
74
3.0 105
96
59
8.7 105
98
33
2.4 105
98
44
2.1 105
96
63
No fit
No fit No fit 1.1 103 No fit No fit
No fit No fit No fit 3.3 104
stand for 30 min in order to let the sludge settle. In order to remove particles, the liquid phase was decanted and filtered through a glass microfiber filters (GC/F; VWR Denmark). In the filtered samples (100 g) a surrogate standard mixture was added followed by solid phase extraction using OASIS HLB (6cc, Waters, Sweden). A detailed description of sample preparation and analyses employing the LC-MS/MS methodology reported in Grabic et al., (unpublished data) and Fick et al. (2009) may be found as Supplementary data S1.
4475
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 7 0 e4 4 8 2
2.4.
Data analysis
The measured concentrations of APIs in the water phase (C0) where no sludge was added and when sludge was added (Cw) and SS were used to calculate Cs as follows in each experimental replicate:
Removed fraction ðFR Þ ¼
Cs ¼ ðC0 Cw Þ=SS
(1)
The sorption isotherms define the equilibrium between the concentration of a chemical in aqueous and solid phases (Schwarzenbach et al., 2003). With batch sorption experiments including multiple concentrations, sorption isotherms may be constructed, from which the solidewater distribution coefficients can be determined. Three equations are used here to describe the sorption isotherms; linear (Eq. (2)), Freundlich (Eq. (3)) and Langmuir (Eq. (4)). Cs is the concentration sorbed to the sludge (g kg1) and Kd is the linear sorption constant. Kf is the Freundlich coefficient and n is the Freundlich exponent. s represents the total number of surface sites per mass of sorbent and KL is the Langmuir coefficient. Linear Cs ¼ Kd Cw Freundlich Cs ¼ Kf Cw
(3)
smax KL Cw 1 þ KL Cw
(6)
The sludge production from treatment of municipal sewage can be considered reasonably constant irrespective of the methods of treatment. Based on Henze et al. (2002) the typical amounts of sludge removed from a WWTP can be calculated for primary and secondary sludge. For a WWPT with 2 h of settling time, the removal of primary sludge was estimated to 210 g m3 of treated wastewater. In order to estimate the removal of secondary sludge, the yield coefficient for a low load treatment plant was employed giving a removal of 110 g m3 for the secondary sludge. The corresponding value for a high load treatment plant would be 165 g m3.
Estimation of the sorbed fraction
The estimation of the sorbed fraction of each API was made by employing the obtained Kd-values. In those cases where both Freundlich and Langmuir isotherms were found to have a better fit than the linear, the one with the best significance was chosen as the best fit. For the APIs where the Freundlich or Langmuir isotherm gave the best fit (see Tables 2e4), Kdvalues were calculated for a water concentration of 1 mg L1. The fraction of the APIs at equilibrium for a given sludge concentration (SS; kg L1) was calculated using the Kd-values (L kg1) according to Eq. (5). Sorbed fraction ðFS Þ ¼ 1 Fw ¼
3.
Results and discussion
3.1.
Sludge properties
(4)
The linear isotherm is the simplest case where the affinity of the API remains constant over the concentration interval. The Langmuir isotherm may have the best fit in cases where the sorbent becomes saturated at higher concentrations of API. Freundlich, is commonly employed to describe experimentally obtained sorption data (Schwarzenbach et al., 2003). The software GraphPad Prism 5 for Windows (GraphPad Software, Inc.) was used for data evaluation, using a 95% confidence interval for the best-fit sorption isotherms. The two hypotheses tested were whether the linear isotherm was a better fit than the Freundlich isotherm, and whether the linear was a better fit compared to Langmuir isotherm. Furthermore, in order to qualify as the best fit the R2-value for the curve should be >0.7, otherwise no fit was made.
2.5.
RESS Kd 1 þ RESS Kd
(2) 1 n
Langmuir Cs ¼
which would not be lost either by degradation or stripping, but that will be removed at equilibrium can be calculated as shown in Eq. (6). Thus, the fraction of APIs in the water phase can be calculated after removal of sludge and the APIs sorbed to the sludge removed.
CS SS Kd ¼ CW þ SS CS 1 þ SS Kd
(5)
Furthermore, if the mass of the sludge removed from the WWTP per volume of treated sewage (RESS; kg L1) is known, the fraction of the total APIs load into the activate sludge tank
An ocular inspection of the freeze-dried sludge showed that the primary sludge may be described as wadding, whereas the two types of secondary sludge had an appearance as instant coffee. The ocular differences between the different types of sludge may be due to their origin. Primary sludge is mainly settable particles of wastewater including faeces, toilet paper and particles of food, while secondary sludge consists of bacterial biomass and biopolymers created by bacteria. It is likely that the mainly plant/wood derived primary sludge has different densities of functional groups and aromatic rings compared to the bacteria derived secondary sludge.
3.2.
Sorption isotherms
Due to analytical limitations and the experimental conditions it was possible to determine sorption isotherms for 4452 APIs out of the 75 APIs (Table S2). The linearly obtained Kd values ranged from 85 to 38 400, 199 to 11 340 and 71 to 34 050 L kg1 for primary sludge, secondary sludge with short sludge age and secondary sludge with long sludge age, respectively. Examples of the obtained isotherms are shown in Fig. 1. Tables 2e4 presents for 1 g L1 sludge the obtained distribution coefficients, Freundlich coefficient, Freundlich exponent, smax and Langmuir coefficient in the cases where the isotherms fitted these isotherm descriptions. The sorption isotherm with the best fit was predominantly linear followed by the Freundlich isotherm within the studied concentration range (0.08e10 mg L1). Tables 2e4 exhibit the order of significance for each hypothesis tested. The isotherms coefficients did not change significantly even at the higher sludge densities (Table S3). The average difference between the Kd’s obtained from 1 g L1 sludge
4476
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 7 0 e4 4 8 2
Fig. 1 e Example of the obtained sorption isotherms. From the top Pizotifen linear isotherm, second Maprotiline Freundlich isotherm and at the bottom Bisoprolol Langmuir isotherm.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 7 0 e4 4 8 2
compared to the Kd obtained, when 10 and 50 g L1 sludge was 4%.
3.3.
Sorption results and literature comparison
It was not possible to obtain sorption isotherms under the chosen experimental conditions for all APIs included. For Amiodiarone, Bromocriptine, Chlorpromazine, Clemastine, Dihydroergotamin, Fluphenazine, Levomepromazine, Meclozine, Miconazole, Perphenazine, Prometazine, Roxithromycine and Tamoxifen, the possible reason was too strong sorption to the glass surface of the bottle and/or the water surface. Thirteen of the APIs exhibited so low sorption on the sludge that a sorption isotherm was impossible to determine. The low affinity to sludge obtained for Buprenorphine, Cilazapril, Carbamazepine, Codeine, Diclofenac, Estrone, Flecainide, Fluconazole, Metoprolol, Naloxone and Rosuvastatine, Tramadol and Venlafaxine in all three sludges demonstrated the limit of the design in the opposite direction. The experimental design of the present study and the LOQ in the present matrixes for the individual APIs measured, provides the limits of Kd. The highest Kd that could be obtained within the concentration range 0.08e10 mg L1 and given the median LOQ, 8 ng L1 was 1.2 106 L kg1. However, if the LOQ was low (1 ng L1) or high (170 ng L1), the highest Kd could be expected within the range from 5 104 to 1 107 L kg1. Furthermore, assuming that at least 10% was required to be sorbed in order to determine a Kd value the lowest Kd value obtained within the present study was 100 L kg1. Some of the APIs were on the border between water soluble and low sorption. Bisoprolol, Clonazepam, Diltiazem, Eprosartan, Estradiol, Ibuprofen, Levonorgestrel, Medroxyprogesterone, Orphenadrine and Sotalol were the APIs which specifically exhibited low sorption for the primary sludge and no sorption isotherm could be obtained, which it were for the secondary sludges (Tables 3 and 4). Possible explanation for this behaviour could be the different surface properties of the sludges due to different origin. It should be noticed that Eprosartan exhibited low sorption in the secondary sludge long sludge age (Table 3) and that the sorption in the secondary sludge short sludge age was extremely low (i.e. could not be obtained). Diltiazem, an API for which sorption was obtained in the secondary sludge short sludge age but not in the secondary sludge long sludge age or the primary sludge, it could be speculated that some moieties in the sludge enhance the sorption. Alfuzosin, Citalopram, and Donepezil, the APIs were sorption was obtained in the primary sludge (Table 2) and the a secondary sludge long sludge age (Table 3) but not in the secondary sludge short sludge age this behavior give rise to speculations like there was something within the sludge which prevented sorption to occur or enhanced sorption based on unknown similarities and differences within the sludges, however no clear explanation could be found. However, the difference between the secondary sludge was not larger than 6 percentage points with the median at 4 percentage points. Greater difference (ca 20%) was found between aerobic and anaerobic sludge for EE2 (Zeng et al., 2009). Several of the Kd values reported in the literature would in comparison with the present study be below or around the lowest border for Kd values that could be obtained. Examples
4477
of such APIs are Codeine (Wick et al., 2009) and Estrone, (Andersen et al., 2005; Carballa et al., 2008). Further, compared to the present study, low Kd values were reported by Ternes et al. (2004) with Carbamazepine in primary and secondary sludge <20 and 1.2 L kg1 and Ibuprofen <20 and 7.1 L kg1, respectively. Results from Carballa et al. (2008) and Joss et al. (2005) supported these findings. Abegglen et al. (2009) reports Kd values from experiments with secondary sludge from membrane bioreactors for Carbamazepine and Ibuprofen which both were found to have low sorption (<75 L kg1) in compare to the present study. Calculated Kd values for primary and secondary sludge were in the same range for Ibuprofen 9.5 and 0 L kg1, respectively, whereas for Carbamazepine they were 10e100 times higher (314 and 135 L kg1; et al., 2009). Urase and Kikuta (2005) evaluated the Radjenovic sorption and degradation of APIs. Two of the APIs included were Ibuprofen and Carbamazepine, which were found to have Kd values of 80 and 66 L kg1, respectively at pH 6.7 (Urase and Kikuta, 2005). In the present study Ibuprofen had a weak sorption in the two secondary sludge, which resulted in low Kd values, 200 and 360 L kg1, obtained from the short and long sludge age, respectively, the reason may be due to inherent differences in the sludges. Contrary to the present study Ternes et al. (2004) report for Diclofenac the Kd value 459 L kg1 for primary sludge which is within the set up limits of the present study but no sorption isotherm could be obtained, whereas for secondary sludge the authors report 16 L kg1 which would be to low too be obtained in the present study. Also Joss et al. (2005) report values for Diclofenac in the same range for both sludges. Carballa et al. (2008) reported values ranging between 19 and 158 L kg1 which would be out of the et al. (2009) calculated limits for the present study. Radjenovic the Kd value for Diclofenac for primary and secondary sludge at 194 and 118, respectively. Based on this, Diclofenac appears to have higher affinity for the primary sludge. One explanation could be the difference in pH values (Ternes et al. (2004)). However, in the present study Diclofenac was not among those APIs for which pH was affecting its sorption (see below Section 3.4 and Table 5). The Kd values obtained in this study for Sulfamethoxazole (Tables 2e4) were in the same order as those reported in Go¨bel et al. (2005) for activated sludge. Abegglen et al. (2009) and Carballa et al. (2008) found the Kd of Sulfamethoxazole to be one order of magnitude lower et al. (2009) compared to the present study, and Radjenovic reported two order of magnitude lower Kd values. Another contradictory study of Sulfamethoxazole come from Wu et al. (2009) whom claimed sorption to be too weak, but also reported Kd for other antibiotics in the same level as found here for Sulfamethoxazole. Sulfamethoxazole has been reported to be photosensitive (e.g. Zhou and Moore, 1994; Trovo´ et al., 2009; Ryan et al., 2010). In the present study precautions were taken against photo-degradation. Depending on how different sorption studies have been carried out and precautions taken against photo-degradation throughout the whole experiment, including the extracts, photodegradation might be an explanation to the diverging Kd results. The Kd values obtained in this study for Trimetoprim (Tables 2e4) were in the same order as those reported in Go¨bel et al. (2009), et al. (2005) for activated sludge. Also Radjenovic and Abegglen et al. (2009) found Kd for Trimetoprim to be in
4478
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 7 0 e4 4 8 2
Table 5 e Evaluating the effect of pH within the pH rang for a WWTP, pH 6-8. Average one point Kd values with standard deviation (n [ 3) obtained in sludge from secondary sludge long sludge age at pH 6, 7 and 8 for the concentration 10 mg LL1. By employing one way ANOVA with a 95% confidence interval was the following question asked; are the means significantly different? P < 0,05?
Average Kd pH 6e8 Alfuzosin Amitryptiline Atenolol Atracurium Azelastine Biperiden Bisoprolol Bupropion Chloprothixene Clomipramine Clotrimazol Cyproheptadine Desloratidine Diclofenac Dicycloverin Donepezil Duloxetine Eprosartan Estradiol Ezetimibe Fexofenadine Fluoxetine Flutamide Glibenclamide Glimepiride Haloperidol Hydroxyzine Irbesartan Levomepromazine Levonorgestrel Loperamide Maprotiline Medroxyprogesterone Megestrol Mianserin Nefazodone Orphenadrine Oxazepam Paroxetine Pizotifen Progesterone Repaglinide Risperidone Sertraline Sotalol Sulfamethoxazol Telmisartan Trimetoprim Venlafaxine Verapamil
1.0 2.8 1.6 4.1 1.7 0.8 1.2 1.4 1.9 6.6 3.6 3.4 2.9 0.8 1.8 0.8 3.0 0.6 0.3 3.3 2.7 6.2 0.8 1.2 9.4 2.3 0.5 0.3 2.5 0.1 5.4 4.5 0.1 0.5 0.6 6.4 0.7 1.5 8.2 3.1 1.0 0.1 0.6 1.8 3.0 0.3 0.8 3.5 0.5 4.0
e0.2 10 1.0 103 0.2 103 0.6 102 0.7 103 0.2 103 0.9 102 0.1 102 0.8 104 2.4 103 0.5 104 1.0 103 0.5 103 0.4 103 0.6 103 0.3 103 1.2 103 0.7 102 0.2 103 0.6 103 1.4 102 2.3 103 0.1 103 0.2 103 0.8 102 0.8 103 0.2 103 0.5 103 1.5 103 0.2 103 2.0 103 1.7 103 0.1 103 0.3 103 0.2 103 2.2 103 0.2 103 0.4 103 3.0 103 1.0 103 0.4 103 0.1 103 0.1 103 0.7 104 0.6 102 0.1 103 0.4 103 0.7 102 0.6 102 0.8 102
pH 6 3
12 2.8 1.6 4.7 2.0 7.6 0.6 1.4 2.0 6.7 3.6 3.6 2.9 8.0 1.7 9.6 2.9 0.5 0.4 3.2 3.6 6.1 0.8 1.4 9.6 2.9 7.1 0.9 2.4 2.5 5.7 4.6 0.2 0.6 9.1 8.8 6.5 1.1 8.5 3.1 1.1 0.1 6.5 1.8 4.0 0.3 1.0 4.3 1.0 3.8
pH 7 2
0.4 10 0.2 103 0.2 103 0.2 102 0.1 103 0.9 102 0.6 102 0.1 102 0.1 104 0.2 103 0.5 104 0.2 103 0.2 103 0.6 102 0.2 103 0.1 102 0.1 103 0.4 102 0.3 103 0.3 103 0. 8 102 0.4 103 0.1 103 0.1 103 0.8 102 0.2 103 0.9 102 0.2 103 70. 103 0.6 102 0.6 103 0.4 103 0.1 103 0.2 103 0.9 102 1.1 103 0.5 102 0.2 103 1.2 103 0.2 103 0.3 103 0.1 103 0.6 102 0.2 104 0.3 102 0.2 103 0.1 103 0.3 102 0.4 102 0.7 102
the same order of magnitude. The Kd obtained by Radjenovic et al. (2009) for Atenolol and Glibenclamide was one order of magnitude lower compared to those presented in this study (Tables 2e4). In addition, Maurer et al. (2007) obtained even lower Kd values below 40 L kg1 for Atenolol, Sotalol, Metoprolol. Roxythromycin was included in the study conducted by Abegglen et al. (2009) where the authors determined Kd to
7.3 1.8 1.6 413 8.7 0.6 2.1 1.4 10 3.9 3.3 2.1 2.3 479 1.2 4.8 1.8 0.6 0.2 3.0 3.5 3.7 0.8 1.1 0.9 1.2 5.3 1 1.0 7 3.0 2.7 0.2 0.3 404 6.3 5.0 1.9 4.9 2.0 0.6 0.2 4.2 9.8 2.4 0.3 0.9 348 6 3.6
pH 8 2
0.3 10 0.3 103 0.2 103 20 103 0.5 102 0.1 103 0.9 102 0.2 102 0.7 103 0.2 103 0.4 104 0.1 103 0.2 103 81 102 0.3 103 0.3 102 0.3 103 0.9 102 0.9 102 0.6 103 0.5 102 0.7 103 0.1 103 0.1 103 0.1 103 0.1 103 1.0 102 0 0.3 103 193 0.3 103 0.4 103 0.8 102 0.2 103 0.3 102 0.9 103 0.5 102 0.2 103 0.8 103 0.2 103 0.1 103 0.1 103 0.4 102 0.7 103 0.5 102 0.1 103 0.4 103 9 21 0.8 102
11 4.0 16 3.4 2.3 1.0 0.8 1.5 2.8 9.3 4.0 4.4 3.4 1.2 2.4 10 4.4 0.8 4.2 3.9 1.0 8.7 0.6 1.1 9.6 28 0.3 14 4.2 0.2 7.3 6.2 0.2 0.7 6.2 4.2 8.2 1.6 11 4.3 1.3 0.7 6.2 2.6 3.0 2.7 0.4 2.8 0.6 0.4
0.5 0.7 0.9 0.2 0.1 0.2 0.4 0.2 4.6 1.1 0.4 0.3 0.3 0.6 0.5 0.4 0.7 1.0 0.7 0.7 0.9 1.6 0.1 0.1 0.1 0.3 0.1 2 0.1 0.1 1.2 1.2 0.1 0.2 0.3 0.9 0.9 0.3 2.0 0.6 0.2 1.2 0.3 0.4 0.2 0.8 0.2 0.2 0.4 0.1
pH 6e8 2
10 103 102 102 103 103 102 102 104 103 104 103 103 103 103 102 103 102 102 103 102 103 103 103 102 102 103 103 103 103 103 103 103 102 103 102 103 103 103 103 102 102 104 102 102 103 102 102 103
No Yes No Yes Yes No No No Yes Yes No Yes No No No No Yes No No No Yes Yes No Yes No No Yes Yes Yes No No No No No Yes Yes Yes No No Yes No No No Yes No No No Yes No Yes
be 570 L kg1 which is almost 6 times higher than the one reported by Ternes et al., (2004). In the present study Roxythromycin was among those APIs for which no sorption isotherm could be obtained. However, the main differences were found between primary and secondary sludge and not between the two secondary types of sludge. Based on the different surface characteristics due to different origin of the
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 7 0 e4 4 8 2
4479
sludge it is likely that the sorption of different APIs structures can vary between the sludge types. The Kd values obtained in the present study may be slightly overestimated since the experiment were performed at 4 C, which is even below the temperature in WWTPs and consequently show somewhat higher Kd-values in comparison with other studies conducted at 10 or 20 C.
3.4.
Effect of pH on sorption
In the present study, 20 out of 50 APIs were found to have a significant difference in their mean Kd values obtained in the secondary sludge with long sludge age as a function of pH (6, 7 and 8; Table 5). By employing one way ANOVA-test, with Tukeys post-test at 95% confidence level, the significant difference within the pH range of WWTPs was determined and identified. All structures of the APIs, where the ANOVAtest indicated a significant difference between Kds included a nitrogen atom and in many case an amine functional group and hence had basic properties. Even though the Kd values were significantly different in the pH range 6e8, it will not have any significant effect on the removal via sorption to sludge (in a subsequent appendix Table A1). However, Chlorpramine, Chloprothixene, Duloxetine, Fluoxetine, Levomepromazine, Loperamide, Nefazedone and Sertraline were exceptions. For these APIs the pH variation from 6 to 8 affected the fraction in the liquid phase by 10e20% (in a subsequent appendix Table A1). These APIs are to be weak bases. In the current study the variation of pH did not significantly affect the Kd. for Diclofenac even though pH was 2e4 pH units higher than the pKa. Fluoxetine (pKa 9.6) was in the current study shown to be affected by the pH, the difference between pKa and the investigated pH was 2e4 pH units. Even though the differences between pKa for Diclofenac and Fluoxetine was in the same range the two APIs showed different influence by changes of the pH. An investigation of the effect of pH on Naproxen and Carbamazepine, found the highest sorption
Fig. 2 e The plot illustrates that for majoring of the APIs in this study it was not possible estimate their sorption behavior based on log Dow.
was at pH 4, and no significant difference occur between pH values 6 and 8 (Maoz and Chefetz, 2010). Further investigation of the particulate fraction of the sludge found the overall pKa to be 6.1 (Wang et al., 2000). The present study was conducted at a pH value above the pKa for sludge. It cannot be excluded that the Kd was affected by changes caused by pH both in the API structure and the structure of the sludge since among the studied compounds were APIs containing N-groups. The APIs which did not show a significant difference in the Kd-values within the pH range 6e8 can be expected to be stable in their ionized and neutral form, respectively.
3.5.
Correlation of sorption with Kow
An illustration of the absence of correlation between log Dow (calculated) and log Kd-values is shown in Fig. 2. The calculation of log Dow was based on calculated values of log Kow using
Fig. 3 e Estimated fraction, i.e. the fraction of the total APIs load into the activated sludge tank which isn’t lost either by degradation or stripping, of the API in the water phase based on experimentally determined sorption isotherms obtained by using primary and secondary sludge. For further information see supplementary data Figure S1.
4480
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 7 0 e4 4 8 2
KOWIN and calculated pKa values presented by Manallack (2009). Log Dow was for acidic compounds calculated according to Dow ¼ Kow þ 1/(1 þ 10(pHpKa)) and for basic compounds Dow ¼ Kow þ (1/(1 þ 10(pKapH))). To estimate the sorption behavior based only on log Dow gives an incorrect sequence when compared and put in relation to other compounds. A few examples to illustrate the difficulties include Atenolol log Kd 3.6 and log Dow 2.6 and Bisoprolol log Kd 3.1 and log Dow 0.7, with the calculated pKa 9.6 and 9.5, respectively. However, using log Dow to predict the tendency of sorption for Glimpiride and Ibuprofen with the log Dow 2.7 and 2.0, respectively, using pKa 5.0 and 4.3, respectively would give the impression that Glimperide could be expected to have higher affinity to sludge than Ibuprofen. The experimentally determined log Kd (3.0 and 2.6, respectively) values obtained in the present study confirm such a hypothesis. However, applying the same hypothesis for Estradiol and Ezetimibe with the same log Kow 3.94 but with respect to pKa (9.72 and 10.3) the log Dow would be 0.67 and 1.22, respectively, the tendency would be that Ezetimibe would have higher affinity to sludge. But the results from the present study imply the opposite, log Kd 2.47 and log Kd 1.98, respectively. Even though efforts have been made to calculate the Kd based on Kow with respect to pKa values, the inherent properties of the APIs and the sludge will need multiple descriptors to get close to experimentally derived Kd-values. The APIs different structures together with the character of the sludge polar, un-polar, charged moieties will give possibilities for different interactions such as ionic, hydrophobic, etc. Therefore to predict sorption based on the APIs log Dow and not consider the properties of the sludge will not provide with reliable Kd values for all APIs. Correlation could however be expected among un-polar APIs where the main interaction would be hydrophobic.
3.6.
Consequence of sorption for the fate of APIs
In order to estimate the removal (see Materials and methods (2.5) Eq.(6)) of APIs that were not lost from degradation or stripping but due to sorption in the WWTP process, the obtained Kd values were employed to estimate the fraction of API in the water phase (see Fig. 3; Table S4). The Kd values obtained from the primary sludge and secondary sludge with long sludge age were used for the estimation. Therefore, the results can be considered to be similar. Fig. 3 presents the predicted distribution of APIs between the water and sludge based on the determined Kds and some selected typical primary and secondary sludge outputs from the WWTP. Fig. 3 shows that 37 APIs will mainly be present in the water phase to an extent of >80%. For 15 of the APIs, the sorption was so strong that the fraction in the water phase is predicted to be 20% or less. Earlier investigations of APIs removal in activated sludge (Go¨bel et al., 2005) distinguished between different treatments steps in the WWTP. For Sulfamethoxazole and Trimetoprim the API removed by sorption can be compared with the present study. Go¨bel et al. (2005) found 1.5% and 4% as total removal of these compounds to primary and secondary sludge, respectively. The removal of those compounds was in this study approximately 10% (see Fig. 3) for both APIs. Ra et al. (2008) presents the removal for Estradiol, Diclofenac and Ibuprofen which were in the same range as in the present study, see Fig. 3.
4.
Conclusion
In this study experimentally derived sorption isotherms are presented along with the corresponding obtained Kd values. The obtained Kd values were used in order to estimate the removal of the APIs in the WWTPs due to sorption to sludge. The major findings from this study are: Experimentally derived Kd values and sorption isotherms for 52 APIs. For 13 APIs sorption to sludge was stronger than 1.2 106 L kg1. For 10 APIs sorption to sludge was less than 100 L kg1. Demonstration of the reduction of APIs due to sorption to sludge showed that for 31 of the APIs <20% would be recovered in the liquid phase.15 APIs have high affinity towards the sludge, i.e. <20% of the initial concentration would be found in the liquid phase. These APIs will therefore mainly be removed from the wastewater with the sludge, unless they are biodegraded significantly during treatment.
Acknowledgment We gratefully acknowledge Mrs. Marika Wennberg at Klagshamn WWTP, Mrs. Maria Mases at Sjo¨lunda WWTP and Mr. Thomas Guildal at Avedøre WWTP, for information regarding the WWTPs and assistance when sludge was collected. The present study was part of the MistraPharma project, and was supported by the Foundation for Strategic Environmental Research (MISTRA) and for the post doc scholarship to Maritha Ho¨rsing granted by The Swedish Research Council for Environment, Agricultural Sciences and Spatial Planning (FORMAS). The authors are grateful to Dr. Maria G. Antoniou (DTUEnvironment) for the editorial assistance of the manuscript.
Appendix. Supplementary material Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.watres.2011.05.033.
Appendix
Table A1 e The water fraction after sorption to secondary sludge with long sludge age at different pH. API Alfuzosin Amitryptiline Atenolol Atracurium Azelastine Biperiden Bisoprolol Bupropion
Fw pH6 (%)
Fw pH7(%)
Fw pH8 (%)
88 77 85 95 82 92 99 98
93 84 85 96 91 94 98 99
89 70 85 96 80 90 99 98
(continued on next page)
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 7 0 e4 4 8 2
Table A1 (continued ) API Chloprothixene Clomipramine Clotrimazol Cyproheptadine Desloratidine Diclofenac Dicycloverin Donepezil Duloxetine Eprosartan Estradiol Ezetimibe Fexofenadine Fluoxetine Flutamide Glibenclamide Glimepiride Haloperidol Hydroxyzine Irbesartan Levomepromazine Levonorgestrel Loperamide Maprotiline Medroxyprogesterone Megestrol Mianserin Nefazodone Orphenadrine Oxazepam Paroxetine Pizotifen Progesterone Repaglinide Risperidone Sertraline Sotalol Sulfamethoxazol Telmisartan Trimetoprim Venlafaxine Verapamil
Fw pH6 (%)
Fw pH7(%)
Fw pH8 (%)
31 57 20 72 76 92 84 90 76 99 96 74 96 60 92 87 90 76 93 91 79 97 61 66 98 94 91 51 93 89 52 74 89 99 93 33 96 97 90 96 99 96
46 70 22 81 80 95 89 95 83 99 100 75 96 71 92 89 91 88 94 100 90 100 75 77 100 97 96 59 95 83 65 82 94 97 96 48 97 97 91 96 100 96
25 49 18 67 73 88 79 90 67 99 96 70 99 51 93 89 90 76 96 100 68 98 55 59 98 92 94 68 92 85 45 68 88 99 94 26 97 97 95 97 99 95
references
Abegglen, C., Joss, A., McArdell, C.S., Fink, G., Schlu¨sener, M.P., Ternes, T.A., Siegrist, H., 2009. The fate of selected micropollutants in a single-house MBR. Environ. Int. 43 (7), 2036e2046. Andersen, H.R., Siegrist, H., Halling-Sørensen, B., Ternes, T., 2003. Fate of esterogens in a municipal sewage treatment plant. Environ. Sci. Technol. 37 (18), 4021e4026. Andersen, H.R., Hansen, M., Kjølholt, J., Stuer-Lauridsen, F., Ternes, T., Halling-Sørensen, B., 2005. Assessment of the importance of sorption for steroid estrogens removal during activated sludge treatment. Chemosphere 61 (1), 139e146. Berg, U.T., Nyholm, N., 1996. Biodegradability simulation studies in semicontinuous activated sludge reactors with low (mg/L)
4481
and standard (ppm range) chemical concentrations. Chemosphere 33 (4), 711e735. Bo¨hm, L., During, R.-A., 2010. Partitioning of polycyclic musk compounds in soil and aquatic environment e experimental determination of KDOC. J. Soil. Sedim. 10 (4), 708e713. Calisto, V., Esteves, V.I., 2009. Psychiatric pharmaceuticals in the environment. Chemosphere 77 (10), 1257e1274. Carballa, M., Fink, G., Omil, F., Lema, J., Ternes, T., 2008. Determination of the solid-water distribution coefficient (Kd) for pharmaceuticals, estrogens and musk fragrances in digested sludge. Water Res. 42 (1e2), 287e295. Fick, J., So¨derstro¨m, H., Lindberg, R.H., Phan, C., Tysklind, M., Larsson, D.G.J., 2009. Contamination of surface, ground and drinking water from pharmaceutical production. Environ. Toxicol. Chem. 28 (12), 2522e2527. Gabet-Giraud, V., Mie`ge, C., Choubert, J.M., Martin Ruel, S., Coquery, M., 2010. Occurrence and removal of estrogens and beta blockers by various processes in wastewater treatment plants. Sci. Total. Environ. 408 (19), 4257e4269. Grabic, R., Fick, J., Lindberg, R.H., Fedorova, G., Tysklind, M. Multiresidue method for trace level determination of pharmaceuticals in environmental samples by liquid chromatography coupled to triple quadrupole mass spectrometry, unpublished data. Go¨bel, A., Thomasen, A., Mcardell, C.S., Joss, A., Giger, W., 2005. Occurrence and sorption behavior of sulfonamides, macrolides, and Trimetoprim in activated sludge treatment. Environ. Sci. Technol. 39 (11), 3981e3989. Henze, M., Harremoe¨s, P., Arvin, E., Jansen, J. la C., 2002. In: Fo¨rstner, U., Murphy, R.J., Rulkens, W.H. (Eds.), Wastewater Treatment: Biological and Chemical Processes. Springer, Berlin, p. 430. Ifelebuegu, A.O., Theophilus, S.C., Bateman, M.J., 2010. Mechanistic evaluation of the sorption properties of endocrine disrupting chemicals in sewage sludge biomass. Int. J. Environ. Sci. Technol. 7 (4), 617e622. Jones, O.A.H., Voulvoulis, N., Lester, J.N., 2006. Partitioning behavior of five pharmaceutical compounds to activated sludge and river sediment. Arch. Environ. Contam. Toxicol. 50 (3), 297e305. Joss, A., Keller, E., Alder, A.C., Go¨bel, A., McArdell, C.S., Ternes, T., Siegrist, H., 2005. Removal of pharmaceuticals and fragrances in biological wastewater treatment. Water Res. 39 (14), 3139e3152. Lindberg, R., Fick, J., Tysklind, M., 2010. Screening of antimycotics in Swedish sewage treatment plants e Waters and sludge. Water Res. 44 (2), 649e657. Manallack, D.T., 2009. The acid-base profile of a contemporary set of drugs: implications for drug discovery. SAR QSAR Environ. Res. 20 (7e8), 611e655. Maoz, A., Chefetz, B., 2010. Sorption of the pharmaceuticals carbamazepine and naproxen to dissolved organic matter: role of structural fractions. Water Res. 44 (3), 981e989. Maurer, M., Escher, B.I., Richle, P., Schaffner, N., Alder, A.C., 2007. Elimination of b-blockers in sewage treatment plants. Water Res. 41 (3), 1614e1622. Ra, J.S., Oh, S.-Y., Lee, B.C., Kim, S.D., 2008. The effect of suspended particles coated by humic acid on the toxicity of pharmaceuticals, estrogens, and phenolic compounds. Environ. Int. 34 (2), 184e192. , J., Petrovic , M., Barcelo´, D., 2009. Fate and Radjenovic distrubution of pharmaceuticals in wastewater and sewage sludge of the conventional activated sludge (CAS) and advanced membrane bioreactor (MBR) treatment. Environ. Toxicol. Chem. 43 (3), 831e841. Ryan, C.C., Tan, D.T., Arnold, W.A., 2010. Direct and indirect photolysis of sulfamethoxazole and trimetoprim in wastewater treatment plant effluent. Water Res. 45 (3), 1280e1286.
4482
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 7 0 e4 4 8 2
Schwarzenbach, R.P., Gschwend, P.M., Imboden, D.M., 2003. Environmental Organic Chemistry. Wiley & Sons, Inc, Hoboken, New Jersey. Ternes, T., Meisenheimer, M., McDowell, D., Sacher, F., Brauch, H.-J., Haist-Gulde, B., Preuss, G., Wilme, U., ZuleiSeibert, N., 2002. Removal of Pharmaceuticals during drinking water treatment. Environ. Sci. Technol. 36 (17), 3855e3863. Ternes, T., Herrmann, N., Boners, M., Knacker, T., Siegrist, H., Joss, A., 2004. A rapid method to measure the soliddwater distribution coefficient (Kd) for pharmaceuticals and fragrances in sewage sludge. Water Res. 38 (19), 4075e4084. Ternes, T., Joss, A. (Eds.), 2008. Human Pharmaceuticals, Hormones and Fragrances: The Challenge of Micropollutants in Urban Water Management. IWA Publishing. Trovo´, A.G., Nogueira, R.F.P., Agu¨era, A., Frenandez-Alba, A.R., Sirtori, C., Malato, S., 2009. Degradation of sulfamethoxazole in water by solar photo-Fenton. Chemical and toxicological evaluation. Water Res. 43 (16), 3922e3931. Urase, T., Kikuta, T., 2005. Separate estimation of adsorption and degradation of pharmaceutical substances and
estrogens in the activated sludge process. Water Res. 39 (7), 1289e1300. Wang, J., Huang, C.P., Allen, H.E., Takiyama, L.R., Poesponegoro, I., Poesponegoro, H., Pirestani, D., 1998. Acid characteristics of dissolved organic matter in wastewater. Water Environ. Res. 70 (5), 1041e1048. Wang, J., Huang, C.P., Allen, H.E., 2000. Surface physicalechemical characteristics of sludge particulates. Water Environ. Res. 72 (5), 545e553. Wick, A., Fink, G., Joss, A., Siegrist, H., Ternes, T.A., 2009. Fate of beta blockers and psycho-active drugs in conventional wastewater treatment. Water Res. 43 (4), 1060e1074. Wu, C., Spongberg, A.L., Witter, J.D., 2009. Sorption and biodegradation of selected antibiotics in biosolids. J. Environ. Sci. Health Part A 44 (5), 454e461. Zeng, Q., Li, Y., Gu, G., 2009. Effect of temperature on the sorption of 17a-ethinylestradiol to aerobic and anaerobic sludge. International Conference on Energy and Environmental Technology. Zhou, W., Moore, D.E., 1994. Photochemical decomposition of sulfamethoxazole. Int. J. Pharm. 110 (1), 55e63.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 8 3 e4 4 9 0
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Perfluorinated compounds in surface waters and WWTPs in Shenyang, China: Mass flows and source analysis Hongwen Sun a,*, Fasong Li a, Tao Zhang a, Xianzhong Zhang a, Na He a, Qi Song a, Lijie Zhao a, Lina Sun b, Tieheng Sun b,c a
MOE Key Laboratory of Pollution Processes and Environmental Criteria, College of Environmental Science and Engineering, Nankai University, 94 Weijin Street, Tianjin 300071, China b MOE Key Laboratory of Regional Environment and Eco-Remediation, Shenyang University, Shenyang 110044, China c Key Laboratory of Terrestrial Ecological Process, Institute of Applied Ecology, Chinese Academy of Sciences, Shenyang 110016, China
article info
abstract
Article history:
Concentrations of 10 perfluorinated chemicals (PFCs) were investigated in the Hun River
Received 1 January 2011
(HR), four canals, ten lakes, and influents and effluents from four main municipal waste-
Received in revised form
water treatment plants (WWTPs) in Shenyang, China. Mass flows of four main PFCs were
18 May 2011
calculated to elucidate the contribution from different sections of the HR. Overall, per-
Accepted 30 May 2011
fluorooctanoic acid (PFOA) and perfluorohexanoic acid (PFHxA) were the major PFCs in the
Available online 7 June 2011
HR, with ranges of 2.68e9.13 ng/L, and 2.12e11.3 ng/L, respectively, while perfluorooctane sulfonate (PFOS) was detected at lower levels, ranging from 0.40 to 3.32 ng/L. The PFC
Keywords:
concentrations in the HR increased after the river passes through two cities (Shenyang and
PFCs
Fushun), indicating cities are an important contributor for PFCs. Mass flow analysis in the HR
PFOA
revealed that PFC mass flows from Fushun are 1.65e5.50 kg/year for C6-C8 perfluorinated
PFOS
acids (PFCAs) and 1.29 kg/year for PFOS, while Shenyang contributed 2.83e5.18 kg C6-C8
Surface water
PFCAs/year, and 3.65 kg PFOS/year. The concentrations of PFCs in four urban canals were
Mass flow
higher than those in the HR, with the maximum total PFCs of 240 ng/L. PFOA and PFOS
Hun River
showed different trends along these canals, suggesting different sources for the two PFCs.
Shenyang
Total PFCs in ten lakes from Shenyang were at low levels, with the greatest concentration
China
(56.2 ng/L) detected in a heavily industrialized area. The PFC levels in WWTP effluents were higher than those in surface waters with concentrations ranging from 18.4 to 41.1 ng/L for PFOA, and 1.69e3.85 ng/L for PFOS. Similar PFC profiles between effluents from WWTPs and urban surface waters were found. These results indicate that WWTPs are an important PFC source in surface water. Finally, we found that the composition profiles of PFCs in surface waters were similar to those in tap water, but not consistent with those in adult blood from Shenyang. The calculation on total daily intake of PFOS by adults from Shenyang showed that the contribution of drinking water to human exposure was minor. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
Perfluorinated compounds (PFCs) are a class of manmade chemicals which have been widely used as surfactants, surface
protectors and performance chemicals in products for over 50 years (Kissa, 2001; OECD, 2002). The PFCs have been detected in environmental samples globally (Toms et al., 2009; Ahrens et al., 2010; Loos et al., 2010a, 2010b; Pan and You, 2010; Zhang et al.,
* Corresponding author. Tel.: þ86 22 23509241; fax: þ86 22 23508807. E-mail address:
[email protected] (H. Sun). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.05.036
4484
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 8 3 e4 4 9 0
2010a). Due to the strong CF bond, many PFCs can resist biological and chemical degradation, and are persistent in the environment. Moreover, toxic and bioaccumulative effects of two representative PFCs, perfluorooctane sulfonate (PFOS) and perfluorooctanoic acid (PFOA) have been confirmed by extensive studies (Lau et al., 2007; Fenton et al., 2009; US EPA, 2010). PFOS has been listed as a Persistent Organic Pollutant under the Stockholm Convention since May 2009 (Stockholm Convention, 2009). In 2002, 3M Company voluntarily phased out the perfluorosulfonyl fluoride based chemistries, however, a variety of related PFCs are still being produced due to the lack of suitable substitutes (Wang et al., 2010). It has been proposed that PFCs could be transported globally within the thermohaline circulation system (Armitage et al., 2006), and water is a major medium for their transport (Prevedouros et al., 2006). The global occurrence of PFCs in openocean waters has been described with concentration ranges of some tens to hundreds of pg/L (Yamashita et al., 2005; Wei et al., 2008; Ahrens et al., 2009, 2010; Busch et al., 2010). Rivers are an important pathway for the transport of contaminants from land to oceans, and PFC levels in rivers are up to thousands of ng/L (Skutlarek et al., 2006; So et al., 2007). One previous study showed that the total discharge of PFOA from European rivers is in reasonable agreement with reported emission estimates from this area before (Mclachlan et al., 2007). In addition, mass flows in the Switzerland’s Glatt Valley watershed were analyzed for PFC contributions. The PFCs emanating from wastewater treatment plants (WWTPs) were not significantly removed in 35 km of the Glatt River (Huset et al., 2008). Similar results were found in Lake Ontario (Boulanger et al., 2005a). On the other hand, variability of PFC concentrations in the Cape Fear River Drainage basin in North Carolina suggested that series of sources exist throughout the basin (Nakayama et al., 2007). All these results indicate that PFCs can be conserved in aqueous phase, and concentrations and mass flows of PFCs can be used for analyzing their sources. PFCs have been detected in influents and effluents from WWTPs around the world (Loganathan et al., 2007; Becker et al., 2008; Guo et al., 2010). However, PFCs in municipal WWTPs from China have scarcely been reported previously (Li et al., 2010). The previous reports revealed that PFCs could not be removed efficiently in traditional WWTPs, and the discharge of WWTP effluent is a significant source for PFCs in the aquatic environment. This is the first study to compare the distribution of PFCs in WWTPs and their ambient aquatic environment. In recent years, various industries, such as textiles, electronics, and packaging products, develop rapidly in China, and China has become the largest manufacturer of PFCcontaining products in the world. Therefore, it is worthwhile to investigate the environmental distribution and sources of PFCs in China. Shenyang is the biggest city in northeastern China with an area of 13 000 km2 and a population of over 7.2 million. One of the largest PFC manufacturing plants in China is located in Fuxin city, near Shenyang. A previous study (Yeung et al., 2006) on PFC levels in human blood in China reported the greatest mean PFOS concentration in human blood serum samples from Shenyang (153.2 ng/mL with a range of 31.7e310 ng/mL) among nine studied cities. However, the main exposure pathway for human is still unknown. In this study, the Hun River (HR) was sampled around Shenyang for analyzing PFC levels, and mass
flows of PFCs at key sections of the river were calculated. In addition, surface water samples from two urban canals, two drainage canals, ten lakes, and influents and effluents from four WWTP were measured to analyze the potential sources of PFCs in environmental waters in Shenyang. Furthermore, adult human exposure to PFOS in Shenyang was primarily discussed.
2.
Materials and methods
2.1.
Sample collection
All water samples were collected from July to September, 2009 from Shenyang and its neighborhood, China. Sampling sites were selected based on the distribution of surface waters in Shenyang, which included the HR and its tributaries (n ¼ 12), North Urban Canal (NUC) (n ¼ 8), South Urban Canal (SUC) (n ¼ 4), Weigong Drainage Canal (WDC) (n ¼ 6), Xihe Drainage Canal (XDC) (n ¼ 8), and ten lakes (n ¼ 10). Furthermore, grab samples of influents and effluents from four main WWTPs in Shenyang were also investigated (n ¼ 8). All the water samples were collected in the middle of rivers or lakes, and three travel blanks for water samples were checked for sampling events. The sampling locations are shown in Fig. 1, and geographic information of the sampling sites is given in Table S1 (Supporting Information). The HR is a primary upstream of Daliao River (a main water system in China) which is the main water resource for Shenyang and other neighboring cities. In this study, the sampling campaign in the HR started from the Dahuofang Reservoir and its tributaries, which is the main drinking source for Fushun city, and ended at Liaozhong County downstream from Shenyang (Fig. 1). Detailed information of the studied WWTPs is showed in Table S2. Temperature of the surface water samples was in the range of 22e25 C. Duplicate water samples were collected and fulfilled in 500 mL polypropylene (PP) bottles. All samples were stored at 20 C before extraction. All Teflon-containing laboratory materials were avoided to prevent possible contamination of the samples.
2.2.
Sample preparation and analysis
The water samples were filtered by glass fiber filters before spiked with three internal standards (13C PFOA, 13C PFOS, and 13 C 8:2 fluorotelomer unsaturated acid (8:2 FTUCA)). The PFCs in water samples were extracted and cleaned up using SPE by methods similar to that described by Taniyasu et al. (2005) with some modifications (Li et al., 2011). Concentrations of 10 PFCs were determined with a Waters Alliance 2695 high performance liquid chromatograph (HPLC) equipped with a Quattro Micro triple quadrupole mass spectrometer (MS/ MS). The MSeMS was operated in electrospray negative ionization (ESI-) mode. Details regarding reagents and chemicals, sample preparation and extraction procedures, and instrumental methods are given in the Supporting Information.
2.3.
Quality control and quality assurance
Matrixspike recoveries of individual PFCs through the analytical procedure were obtained by spiking of 10 target analytes into
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 8 3 e4 4 9 0
4485
Fig. 1 e Sampling locations in the Hun River and surface waters, WWTPs in Shenyang, Liaoning, China.
500 mL of randomly selected wastewaters (n ¼ 5, 20.0 ng each) and river samples (n ¼ 5, 10.0 ng each). Recoveries of PFCs (Table S3) spiked into samples ranged from 63 12% to 110 10% for wastewaters, and ranged from 82 8% to 105 5% for river waters. Method precision was good, with relative standard deviations (RSDs) ranging from 3% to 13%. Calibration curves were prepared by spiking fixed levels of the three internal standards into Milli-Q water with varying concentrations of the target PFCs. Linear calibrations were acquired in the range of 100 to 80 000 ng/L with correlation coefficients more than 0.99. After calibrated by internal standards, the recoveries of the 10 PFCs were improved to 68e116% (average) for wastewater and 91e109% (average) for river samples (Table S3). Method and instrumental blanks were checked for the presence of target PFCs, and quality control standards were injected for each batch of 20 samples. The limit of detection (LOD) was defined as the concentration that yielded an signal to noise ratio (S/N) of 3,while the limit of quantification (LOQ) was defined as the concentration that yielded an S/N ratio of 10 or the lowest point at calibration curve calculated to be with 30% of its actual value. The LOD and LOQ for analyzed PFCs ranged from 0.28 to 0.75 ng/L, and 0.66e2.60 ng/L, respectively (Table S3). Concentrations in the range of LOD to LOQ were assigned a value twice that of the LOD, and concentrations at or lower than the LOD were assigned a value of zero. Procedural blanks (500 mL of MilliQ water) were extracted in the same manner as the samples. No detectable PFCs were found in the procedural blanks and travel blanks.
3.
Results and discussion
3.1. Concentrations of PFCs in the Hun River and its tributaries Only PFOA and perfluorohexanoic acid (PFHxA) were detected in HR’s upstreams, i.e., Qinyuan River (sampling location, In1),
Suzi River (In 2), and the mouth of Dahuofang Reservoir (HR1), with the maximum concentrations of 2.68 and 2.70 ng/L, respectively. The concentrations of other PFCs were very low (
C8) were also found at Tr2, ranging from 1.20 (PFUnA) to 2.48 ng/L (PFDoA), while these longchain PFCAs were seldom detected at other sampling sites along the HR. To our understanding, many industries, such as petroleum refinery, chemical engineering, machinery, and textile industry are located in the Dongzhou District in Fushun city. The discharges from these industries might contribute to the high PFC levels in the Dongzhou River (i.e., Tr2), which flows through this area (http://www.fushun.gov.cn/). The PFOA and PFHxA were the most prevalent PFCs at the sampling sites of HR1-8 (Table 1), with the concentration ranges of 2.68e9.13 ng/L, and 2.12e11.3 ng/L, respectively. The PFC levels in the HR downstream were generally higher than those in upstream. The greatest total PFC concentration (26.7 ng/L) was found at HR7, indicating some extra contributions of PFCs from Shenyang (Table 1). PFOS was detected at low levels (<0.66e3.32 ng/L) in the HR, and 8:2 FTUCA was not detected in any samples collected from the HR. The PFOA in the HR is at moderate level as compared to other surface waters in China (Table S4). It is comparable to those in Pearl River Delta and higher than those in Yangtze River nearby Nanjing and Yichang (So et al., 2004, 2007); but
4486
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 8 3 e4 4 9 0
Table 1 e Measured concentrations (ng/L) of perfluorinated compounds and flux (108 m3/year) in the Hun River and its selected tributaries. Sites (see Fig. 1) Qingyuan (In1) Suzi River (In2) The mouth of Reservoir (HR1) Zhangdang River (Tr1) Dongzhou River (Tr2) Gebu Bridge (HR2) Qijianfang (HR3) Dongling Bridge (HR4) Shashan (HR5) Yutai (HR6) Qitaizi (HR7) Yujiafang (HR8)
Flux 6.94 6.43 14.2 0.68 0.22 15.1 15.1 12.1 13.1 NA NA 15.2
PFHxA 2.70 1.26a 2.12 4.31 37.6 11.3 3.08 3.67 2.37 9.55 6.51 5.18
PFHpA a
0.56 0.56a 0.56a 1.59 0.97 4.00 2.17 2.86 3.63 1.96 3.44 2.55
PFOA 2.63 1.83 2.68 4.03 10.7 4.66 6.15 7.23 7.67 8.94 9.13 8.41
PFNA
PFDA
ND ND 0.78a 0.78a ND 0.78a 0.78a 0.78a 0.78a 0.78a 1.55 1.33
PFUnA
ND ND ND ND 0.66a ND 0.66a ND ND 0.66a 0.66a 0.66a
PFDoA
ND ND ND ND 1.20 ND ND ND ND 0.56a 0.56a 0.56a
ND ND ND ND 2.48 ND ND ND ND ND ND ND
PFHxS ND ND ND ND 1.50a ND ND ND ND 1.50a 1.50a ND
PFOS a
0.40 0.40a 0.40a 1.29 0.81 0.74 1.23 0.89 0.82 2.68 3.32 3.20
8:2 FTUCA ND ND ND ND ND ND ND ND ND ND ND ND
ND ¼ not detected (
lower than those in Yangtze River nearby Chongqing and Shanghai, Haihe River in Tianjin (So et al., 2007; Ju et al., 2008; Li et al., 2011). The PFOS in the HR is lower than those in Pearl River Delta, Yangtze River nearby Shanghai (So et al., 2004, 2007) and Haihe River in Tianjin (Li et al., 2011), but higher than Yangzi River nearby Chongqing, Yichang and Nanjing (So et al. 2004, 2007), Globally, both PFOA and PFOS in the HR were lower than those in Rhine River nearby Ruhr area and Moehne River in Europe (Skutlarek et al., 2006; Ahrens et al., 2010).
Mass flow of PFCs in Hun River
It has been shown that many PFCs, especially those of environmental concern, are resistant to biotransformation and photodegradation (Kissa, 2001; Giesy et al., 2006). Furthermore, PFCs exist mainly in aqueous phase due to their unique properties, such as high water solubility (Boulanger et al., 2005a; Huset et al., 2008; Bao et al., 2009) and low volatility (Boulanger et al., 2005a). Hence, the analysis of PFC mass flows could lead to a good understanding of PFC sources in the environment. To evaluate the PFC contributions from different sections of the HR, we calculated the mass flows of PFHxA, perfluoroheptanoic acid (PFHpA), PFOA and PFOS (the four most frequently detected PFCs) based on their average concentrations and river flux according to Eq. (1): Mass flow ¼ Cwater Fwater
(1)
where Cwater is PFC concentration in water (ng/L) and Fwater is the river flux at selected sections (108 m3/year, in Table 1). Mass flow of In1 þ In2 represents the sum of inflows from Qinyuan River (In1) and Suzi River (In2) to Dahuofang Reservoir (Fig. 2). No significant difference was observed between the PFC inflows and outflow for Dahuofang Reservoir (HR1). This indicates that the target PFCs are quite stable in this reservoir, and outflow to the HR is the main pathway for PFC removal in Dahuofang Reservoir. Mass flows of the selected PFCs at HR3 were significantly higher than those at HR1, the mass flows at HR1 and HR3 were 3.01 and 4.66 kg PFHxA/year, 0.80 and 3.29 kg PFHpA/year, 3.01 and 9.31 kg PFOA/year, and 0.57 and 1.86 kg PFOS/year,
16 14
Ma ss flo w (Kg /y ea r)
3.2.
respectively. This suggests a significant PFC pollution contribution from Fushun city (Fig. 2), which is located between sampling sites, HR1 and HR3. Fushun is the fourth biggest city in Liaoning Province, China, with an estimated population of 1.40 million. In this traditional industrial city, the capacity for crude oil processing is over 9.5 million tons, and the outputs of other major products (coal, iron ore, steel, synthetic detergent, and plastic) are also substantial (http://www.fszwgk.gov.cn/). The mass flows at HR8 were 8.92 kg PFHxA/year, 6.12 kg PFHpA/year, 15.0 kg PFOA/year, and 5.51 kg PFOS/year, respectively, and these values were much higher than those at HR3 and HR4. Shenyang is located between sampling sites, HR8 and HR4, which means that this city contributed significant loads of PFCs to the HR (Fig. 2). Moreover, the PFC contributions from Shenyang are obviously greater than those from Fushun. This could be explained by the population and economic level (as indicated by GDP) of the two cities, which are 7.2 and 1.4 million, and 64.1 and 10.6 billion US dollar in 2009 for Shenyang and Fushun, respectively. Compared to PFC mass flows from other rivers in the world, the mass flows from the HR were at relatively low levels (Table S5). Though PFOA concentration is relatively high (8.41 ng/L) in the HR, the water flux of the HR (48.2 m3/S) is rather low, which leads to the relatively low mass flow of PFOA (15.0 kg/ year) in the HR. By contrast, large rivers, such as the Ganges in
12
PFHxA PFHpA
10 PFOA 8 6
PFOS
4 2 0 In1 + In2
HR1
HR3
HR4
HR8
Sampling stations
Fig. 2 e Mass flows of main perfluorinated compounds at different sections along the Hun River.
4487
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 8 3 e4 4 9 0
India and the Yangtze in China, discharge huge water flux (12400 and 47000 m3/S, respectively), and they contribute more PFC flows (24.1 and 3780 kg/year for PFOA, respectively) to the ocean even though PFOA concentration in these rivers (0.062 and 2.55 ng/L, respectively) are not high. Hence, mass flow is a more suitable index than concentration when considering the contribution for pollutant flux of river basins.
3.3. Concentrations of PFCs in surface waters in Shenyang The mean, median and range of PFC concentrations in four canals (i.e., NUC, SUC, WDC and XDC) in Shenyang are listed in Table 2. Compared to the HR, the total PFCs in these canals were at relatively high levels with the concentration ranges of 16.8e240 ng/L. The PFOA, PFHxA and 8:2 FTUCA were the dominant PFCs in NUC with concentration ranges of 12.6e24.5 ng/L, 5.88e14.5 ng/L, 4.34e5.41 ng/L, respectively, while PFOS was detected at relatively low levels (1.20e2.38 ng/ L). The PFC levels in SUC were lower than those in NUC, and the mean concentrations of PFOA and PFOS were 9.39 and 0.83 ng/L in SUC, and 19.0 and 1.77 ng/L in NUC, respectively. Furthermore, the variation in concentration of each PFC in SUC was not as great as that in NUC. This may be due to that several lakes are located along SUC, the water exchange is slow, and no point source of PFCs is located in this region. Relatively high concentrations of PFHxA and PFOA were found in WDC with ranges of 8.58e36.7 ng/L, and 13.2e20.4 ng/ L, respectively. The concentration of PFOS in WDC ranged from 0.75 to 6.31 ng/L, and other PFCs were detected at levels < LOQs. The PFHpA was the dominant PFC in XDC with mean concentration of 38.1 ng/L, followed by Perfluorohexanesulfhonate (PFHxS, 34.1 ng/L), PFHxA (19.2 ng/L), PFOA (14.5 ng/L), PFOS (10.7 ng/L), and 8:2 FTUCA (5.85 ng/L). In addition, longchain PFCAs were less frequently detected in XDC, and the concentrations were near the LOQs. The WDC and XDC are the wastewater drainage system in Shenyang, and PFC concentrations were higher in these drainage canals than those in urban canals (SUC and NUC). The PFC composition profiles are different among the four canals (Fig. S1), though they are connected to each other. PFOA, PFHxA and PFHpA were the main PFCs in the two urban canals and WDC, accounting for 2261%, 065% and 223% of
the total PFCs, respectively, while PFOS and other PFCs contributed only 210% and 226% of the total PFCs. PFHxA, PFOA and PFHpA were also the predominant PFCs in XDC but with relative lower contributions, accounting for 8e27%, 6e25%, and 10e28% of the total PFCs, respectively, except for PFHpA at the sampling site of XDC2 (146 ng/L, 61%). In XDC, PFHxS, PFOS and other PFCs contributed 16e39%, 3e16% and 4e15% of the total PFCs, respectively. The spatial distributions of PFOS and PFOA in the four canals are shown in Fig. 3. The PFOA did not vary dramatically, and its highest concentration (24.5 ng/L) was found at the sampling site of NUC2 (Figs. 1 and 3). The PFOS levels (0.66e2.38 ng/L) were similar among NUC, SUC, and WDC (except WDC6), and gradually increased in XDC (up to 16.3 ng/L). As far as we know, there are many small factories (e.g., electroplating industry, paper manufacture, chemical industry and machinery manufacture) along XDC. Majority of these industries are of small scale, and some do not treat their wastewater at all and directly discharge wastewater into surrounding water bodies. Hence, we delude that there might be some extra sources from industry for the elevated level of PFCs in XDC. The total PFC concentrations in Shenyang lakes ranged from 10.5 to 56.2 ng/L (Table 3), which is slightly lower than those in the canals. PFOA was the most abundant PFC with the range of 4.41e15.6 ng/L. Other PFCs were detected at low levels, with the exception of PFHxA, whose concentrations are 25.0 ng/L in Xihu Lake and 18.7 ng/L in Wetland Lake, respectively. The highest total PFC level was found in Wetland Lake (56.2 ng/L), followed by Xihu Lake (49.0 ng/L). These elevated concentrations may be due to the fact that both lakes are located in the west of Shenyang where an important industrial park (Tiexi District) is located (http://www.tiexi.gov.cn/). The PFOA and PFOS levels in the present study are less than those measured in other lakes in China and other countries (Table S4) (Boulanger et al., 2004; Sinclair et al., 2006; Naile et al., 2010).
3.4.
Concentrations of PFCs in WWTPs in Shenyang
Measured concentrations of PFCs in grab samples of influents and effluents from four main WWTPs in Shenyang are shown in Table 3. PFOA was the major PFCs in these WWTPs, with the concentrations ranging from 26.2 to 71.1 ng/L in influents, and from 18.4 to 41.1 ng/L in effluents; followed by PFHxA and
Table 2 e Measured concentrations (ng/L) of perfluorinated compounds in water samples collected from urban canals and drainage canals in Shenyang. Canals a NUC SUC WDC XDC
PFHxA M (M) range M (M) range M (M) range M (M) range
9.48 (8.81) 5.88e14.5 1.59 (1.26)b 1.26e2.58 24.3 (28.4) 8.58e36.7 19.2 (20.6) 8.67e32.5
PFHpA 3.89 (3.82) 2.88e4.90 3.23 (3.30) 2.56e3.77 2.32 (2.13) 1.28e4.39 38.1 (21.5) 6.34e146
PFOA 19.0 (18.1) 12.6e24.5 9.39 (9.39) 9.25e9.55 16.2 (16.2) 13.2e20.4 14.5 (14.8) 9.23e21.5
PFNA
PFDA b
0.78 (0.78) NDe1.64 0.78 (0.78)b 0.78b 0.78 (0.78)b NDe5.17 0.78 (0.78)b NDe1.83
PFUnA b
0.66 (0.66) NDe0.66b 0.66 (0.66)b NDe0.66b 0.66 (0.66)b NDe0.66b 0.66 (0.66)b NDe1.19
PFDoA b
PFHxS b
0.56 (0.56) NDe0.56b ND
0.60 (0.60) NDe0.60b ND
0.56 (0.56)b NDe0.56b 0.56 (0.56)b NDe0.56b
0.60 (0.60)b NDe0.60b 0.60 (0.60)b NDe0.60b
PFOS b
1.50 (1.50) NDe5.64 1.50 (1.50)b 1.50b 1.50 (1.50)b NDe2.44 34.1 (31.2) 11.3e76.1
1.77 (1.80) 1.20e2.38 0.83 (0.81) 0.66e1.09 2.06 (1.34) 0.75e6.31 10.7 (9.22) 7.84e16.3
M (M) ¼ mean (median) concentration; ND ¼ not detected (
8:2 FTUCA 4.87 (4.89) 4.34e5.41 ND 1.56 (1.56)b NDe1.56 5.09 (5.85) 2.75e7.11
4488
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 8 3 e4 4 9 0
Fig. 3 e Concentrations of PFOS (a) and PFOA (b) along the urban canals and drainage canals in Shenyang. Abscissa represents the distance between sampling stations and the confluence of the Hun River and North Urban Canal.
PFHpA with the concentration ranges of 15.7e33.4 ng/L and 3.14e17.9 ng/L in influents, and 10.7e11.3 and 1.60e24.2 ng/L in effluents, respectively. PFOS was detected at relatively low levels, ranging from 2.88 to 12.8 ng/L in influents, and from 1.69 to 3.85 ng/L in effluents. Similar to PFCs in surface waters, longchain PFCAs were detected with less frequency and at lower concentrations. The reason for this pattern may be that shortchain PFCAs (C 8) are produced and consumed more frequently than longchain ones (Boulanger et al., 2005b). In addition, the solubility of longchain PFCAs is low. 8:2 FTUCA was not analyzed for the samples from WWTPs. Compared to the influents, the concentrations of most individual PFCs decreased significantly in effluents, by 28%e 100%. Two exceptions occurred, PFOA in Beibu WWTP and PFHxA in Mantanghe WWTP, where the concentrations increased by 122% (from 26.6 to 32.5 ng/L) and 701% (from 3.45 to 24.2 ng/L) in effluents as compared to the influents,
respectively. The reduction in PFC concentration in water phase could be attributed to the sorption onto sludge (Sun et al., 2011). The increase in PFOA and PFHxA in effluents of WWTPs as compared to influents was also reported by other literatures (Loganathan et al., 2007; Becker et al., 2008), and this was ascribed to incomplete degradation of precursors, such as fluorotelomer alcohols, perfluoroalkyl phosphates or fluorotelomer sulfonates. It should be noted that great variation in PFC concentration in wastewaters may exist (Thompson et al., 2011) during a day, and grab sampling used in the present study may also account for the variation in PFC levels between the influent and effluent. Recently, Thompson et al. (2011) found that the variation of PFCs concentrations in grab samples from WWTP influents was larger than that in composite grab samples. The level of PFCs follows the sequence of WWTPs > drainage canals > urban canals > lakes > the HR,
Table 3 e Measured concentrations (ng/L) of perfluorinated compounds in water samples collected from WWTPs and lakes. Stations WWTPs
Beibu Xiannvhe Shengshuiwan Mantanghe
Lakes
Dingxiang Belin Laodong Xiannv South Qinnian Wanquan Wanliutang Xihu Wetland Totalb
PFHxA influent effluent influent effluent influent effluent influent effluent
mean median
24.0 11.3 33.4 11.2 15.7 11.3 15.8 10.7 ND 3.92 1.26a 2.29 1.26a 1.26a 2.58 1.26a 25.0 18.7 5.78 1.26a
PFHpA 17.9 5.82 3.14 1.60 5.93 2.66 3.45 24.2 3.92 3.32 2.18 1.48 2.56 3.77 3.07 3.53 3.71 5.83 3.34 3.43
PFOA 26.6 32.5 34.5 28.0 71.1 41.1 26.2 18.4 12.4 10.3 9.17 4.41 9.52 9.25 9.26 9.55 15.6 19.6 10.9 9.54
PFNA ND ND ND ND 4.53 ND ND ND 0.78a 0.78a 0.78a 0.78a 0.78a 0.78a 0.78a 0.78a 1.41 8.66 1.60 0.78a
PFDA ND 1.36 2.59 ND ND ND ND ND 0.66a 0.66a ND ND 0.66a 0.66a 0.66a ND 0.66a 0.66a ND 0.66a
PFUnA a
0.56 0.56a 0.56a 0.56a 0.56a 0.56a 0.56a 0.56a ND 0.56a ND ND ND ND ND ND 0.56a 1.26 ND ND
PFDoA a
0.60 0.60a 0.60a 0.60a 0.60a 0.60a 0.60a 0.60a ND ND ND ND ND ND ND ND 0.60a ND ND ND
PFHxS
PFOS
8.06 4.32 19.9 ND 9.27 ND 22.3 ND 1.50a 1.50a 1.50a 1.50a 1.50a 1.50a 1.50a 1.50a 1.50a 1.50a 1.50a 1.50a
5.05 2.86 6.36 2.56 12.8 3.85 2.88 1.69 ND ND 0.40a ND 0.98 1.09 0.62 0.63 ND ND ND ND
ND ¼ not detected; NA ¼ unavailable. a Concentration values greater than or equal to the LOD but less than the LOQ were assigned a value twice that of the LOD. b Mean (median) concentrations were calculated based on PFC levels in all lakes. n ¼ 2.
8:2FTUCA NA NA NA NA NA NA NA NA ND 1.56 ND ND ND 1.56 1.56 ND ND ND ND ND
a
a a
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 8 3 e4 4 9 0
which indicates that municipal WWTP is a major source of PFCs into surface waters in this area, some other sources from the industrial region may also contribute to PFC contamination, especially for some individual PFCs, such as PFOS in XDC.
3.5.
Sources analysis based on PFC profiles in Shenyang
A recent study (Mak et al., 2009) reported that PFOA (2.6 ng/L) was the predominant PFC, and accounted for > 40% of the total PFCs in tap water from Shenyang, and PFOS was at low level (0.39 ng/L). The composition profiles and concentrations of PFCs in surface waters found in this study are consistent with the above results. Similar composition profiles between tap water and surface water have been found in other areas, such as Hong Kong, Nanjing and Shanghai, China (Mak et al., 2009) and Neheim, Germany (Skutlarek et al., 2006). These indicate that PFCs are not successfully removed by bank filtration or artificial recharge. Moreover, poor removal efficiency of PFCs was reported for most water treatment technologies (Takagi et al., 2008; Quinones and Snyder, 2009). Hence, the risk of PFC pollution in surface water to human health should be paid further attention. A contemporaneous study (Liu et al., 2009) reported PFC levels in human blood from adults in Shenyang. PFOS was the most abundant PFC, accounted for 65% of the total PFCs, followed by PFHxS (19%), PFOA (9%), and other PFCs (7%). In our previous study (Zhang et al., 2010b), we estimated the total daily intake (TDI) of PFOS by adults from Shenyang using a one-compartment toxicokinetic model (Fromme et al., 2007; Thompson et al., 2010). The modeled TDI was 0.92 ng/kg bw/ d when calculated using 60 kg of body weight. Comparison of daily intake of PFOS via drinking water (0.013 ng/kg bw/d) (Mak et al., 2009) with modeled TDI showed that contribution for PFOS in human blood from drinking water was minor (<2%).Therefore, further study is needed to reveal the major human exposure sources to PFOS and other PFCs in Shenyang.
4.
Conclusions
In summary, concentrations of 10 PFCs were investigated in the Hun River (HR), four canals, ten lakes, and four WWTPs from Shenyang, China. PFOA, PFHxA were the major PFCs in surface waters, PFOS and longchain PFCAs (>C8) were less frequently detected and at relatively low levels. The mass flows of PFCs along the HR were calculated to determine PFC contribution from different sections. Major cities (Shenyang and Fushun) are the main contributors for PFC contamination in surface water, which appears to be influenced by population and economic level. Inside Shenyang city, the level of PFCs followed the sequence of WWTPs > drainage canals > urban canals > lakes > the HR, which indicates that the municipal WWTPs were the main point sources of PFCs into surface waters. Finally, we found that the composition profiles PFCs in surface waters (this study) was similar to those in tap water (Mak et al., 2009), but not consistent with those in adult blood samples (Liu et al., 2009) from Shenyang, which suggests a poor removal efficiency of PFCs between surface water and tap water. The total daily intake calculation of PFOS by adults from Shenyang showed that the contribution of drinking water to human exposure of PFOS was minor.
4489
Acknowledgments This paper was supported by Ministry of Science and Technology (No. 2009DFA92390) of China and Nature Science Foundation (No. 20877043).
Appendix. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.watres.2011.05.036.
references
Ahrens, L., Barber, J.L., Xie, Z., Ebinghaus, R., 2009. Longitudinal and latitudinal distribution of perfluoroalkyl compounds in the surface water of the Atlantic Ocean. Environ. Sci. Technol. 43, 3122e3127. Ahrens, L., Xie, Z., Ebinghaus, R., 2010. Distribution of perfluoroalkyl compounds in seawater from Northern Europe, Atlantic Ocean, and Southern Ocean. Chemosphere 78, 1011e1016. Armitage, J., Cousins, I.T., Buck, R.C., Prevedouros, K., Russell, M.H., Macleod, M., Korzeniowski, S.H., 2006. Modeling globalscale fate and transport of perfluorooctanoate emitted from direct sources. Environ. Sci. Technol. 40, 6969e6975. Bao, J., Jin, Y., Liu, W., Ran, X., Zhang, Z., 2009. Perfluorinated compounds in sediments from the Daliao River system of northeast China. Chemosphere 77, 652e657. Becker, A.M., Gerstmann, S., Frank, H., 2008. Perfluorooctane surfactants in waste waters, the major source of river pollution. Chemosphere 72, 115e121. Boulanger, B., Vargo, J., Schnoor, J.L., Hornbuckle, K.C., 2004. Detection of perfluorooctane surfactants in Great Lakes water. Environ. Sci. Technol. 38, 4064e4070. Boulanger, B., Peck, A.M., Schnoor, J.L., Hornbuckle, K.C., 2005a. Mass budget of perfluorooctane surfactants in Lake Ontario. Environ. Sci. Technol. 39 1920-1920. Boulanger, B., Vargo, J.D., Schnoor, J.L., Hornbuckle, K.C., 2005b. Evaluation of perfluorooctane surfactants in a wastewater treatment system and in a commercial surface protection product. Environ. Sci. Technol. 39, 5524e5530. Busch, J., Ahrens, L., Xie, Z., Sturm, R., Ebinghaus, R., 2010. Polyfluoroalkyl compounds in the East Greenland Arctic Ocean. J. Environ. Monitor 12, 1242e1246. Fenton, S.E., Reiner, J.L., Nakayama, S.F., Delinsky, A.D., Stanko, J. P., Hines, E.P., White, S.S., Lindstrom, A.B., Strynar, M.J., PetropoulouS-S, E., 2009. Analysis of PFOA in dosed CD-1 mice. Part 2: Disposition of PFOA in tissues and fluids from pregnant and lactating mice and their pups. Reprod. Toxicol. 27, 365e372. Fromme, H., Schlummer, M., Mller, A., Gruber, L., Wolz, G., Ungewiss, J., Bhmer, S., Dekant, W., Mayer, R., Liebl, B., 2007. Exposure of an adult population to perfluorinated substances using duplicate diet portions and biomonitoring data. Environ. Sci. Technol. 41, 7928e7933. Giesy, J.P., Mabury, S.A., Martin, J.W., Kannan, K., Jones, P.D., Newsted, J.L., Coady, K., 2006. Perfluorinated Compounds in the Great Lakes. In: The Handbook of Environmental Chemistry, vol. 5, pp. 391e438. Guo, R., Sim, W.-J., Lee, E.-S., Lee, J.-H., Oh, J.-E., 2010. Evaluation of the fate of perfluoroalkyl compounds in wastewater treatment plants. Water Res. 44, 3476e3486. Huset, C.A., Chiaia, A.C., Barofsky, D.F., Jonkers, N., Kohler, H.P.E., Ort, C., Giger, W., Field, J.A., 2008. Occurrence and mass flows
4490
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 8 3 e4 4 9 0
of fluorochemicals in the Glatt Valley watershed, Switzerland. Environ. Sci. Technol. 42, 6369e6377. Ju, X., Jin, Y., Sasaki, K., Saito, H., 2008. Perfluorinated surfactants in surface, subsurface water and microlayer from Daliao coastal waters in China. Environ. Sci. Technol 42, 3538e3542. Kissa, E., 2001. Fluorinated Surfactants and Repellents, second ed. Marcel Dekker, Inc., New York. Lau, C., Anitole, K., Hodes, C., Lai, D., Pfahles-Hutchens, A., Seed, J., 2007. Perfluoroalkyl acids: a review of monitoring and toxicological findings. Toxicol. Sci. 99, 366e394. Li, F.S., Sun, H.S., He, N., Hao, Z.N., Zhao, L.J., Zhang, T., Sun, T.H., 2011. Perfluorinated compounds in Haihe River and Dagu Drainage Canal in Tianjin, China. Chemsphere 84, 265e271. Li, F., Zhang, C.J., Qu, Y., Chen, J., Chen, L., Liu, Y., Zhou, Q., 2010. Quantitative characterization of short- and long-chain perfluorinated acids in solid matrices in Shanghai. China. Sci. Total Environ. 408, 617e623. Liu, J., Li, J., Luan, Y., Zhao, Y., Wu, Y., 2009. Geographical distribution of perfluorinated compounds in human blood from Liaoning Province, China. Environ. Sci. Technol. 43, 4044e4048. Loganathan, B., Sajwan, K., Sinclair, E., Senthilkumar, K., Kannan, K., 2007. Perfluoroalkyl sulfonates and perfluorocarboxylates in two wastewater treatment facilities in Kentucky and Georgia. Water Res. 41, 4611e4620. Loos, R., Locoro, G., Comero, S., Contini, S., Schwesig, D., Werres, F., Balsaa, P., Gans, O., Weiss, S., Blaha, L., 2010a. Pan-European survey on the occurrence of selected polar organic persistent pollutants in ground water. Water Res. 44, 4115e4126. Loos, R., Locoro, G., Contini, S., 2010b. Occurrence of polar organic contaminants in the dissolved water phase of the Danube River and its major tributaries using SPE-LC-MS2 analysis. Water Res. 44, 2325e2335. Mak, Y.L., Taniyasu, S., Yeung, L.W.Y., Lu, G., Jin, L., Yang, Y., Lam, P.K.S., Kannan, K., Yamashita, N., 2009. Perfluorinated compounds in tap water from china and several other countries. Environ. Sci. Technol. 43, 4824e4829. Mclachlan, M.S., Holmstrom, K.E., Reth, M., Berger, U., 2007. Riverine discharge of perfluorinated carboxylates from the European continent. Environ. Sci. Technol. 41, 7260e7265. Nakayama, S., Strynar, M.J., Helfant, L., Egeghy, P., Ye, X., Lindstrom, A.B., 2007. Perfluorinated compounds in the Cape Fear Drainage Basin in North Carolina. Environ. Sci. Technol. 41, 5271e5276. Naile, J.E., Khim, J.S., Wang, T., Chen, C., Luo, W., Kwon, B., Park, J., Koh, C., Jones, P.D., Lu, Y., Giesy, J.P., 2010. Perfluorinated compounds in water, sediment, soil and biota from estuarine and coastal areas of Korea. Environ. Pollut. 158, 1237e1244. OECD, 2002. Hazard Assessment of Perfluorooctane Sulfonate (PFOS) and Its Salts. Organisation for Economic Co-operation and Development. Prevedouros, K., Cousins, I.T., Buck, R.C., Korzeniowski, S.H., 2006. Sources, fate and transport of perfluorocarboxylates. Environ. Sci. Technol. 40, 32e44. Pan, G., You, C., 2010. Sedimentewater distribution of perfluorooctane sulfonate (PFOS) in Yangtze River Estuary. Environ. Pollut. 158, 1363e1367. Quinones, O., Snyder, S.A., 2009. Occurrence of perfluoroalkyl carboxylates and sulfonates in drinking water utilities and related waters from the United States. Environ. Sci. Technol. 43, 9089e9095. Sinclair, E., Mayack, D.T., Roblee, K., Yamashita, N., Kannan, K., 2006. Occurrence of perfluoroalkyl surfactants in water, fish, and birds from New York State. Arch. Environ. Contam. Toxicol. 50, 398e410. Skutlarek, D., Exner, M., Fa¨rber, H., 2006. Perfluorinated surfactants in surface and drinking waters. Environ. Sci. Pollut. Res. 13, 299e307.
So, M.K., Taniyasu, S., Yamashita, N., Giesy, J.P., Zheng, J., Fang, Z., Im, S.H., Lam, P.K.S., 2004. Perfluorinated compounds in coastal waters of Hong Kong, South China, and Korea. Environ. Sci. Technol 38, 4056e4063. So, M.K., Miyake, Y., Yeung, W.Y., Ho, Y.M., Taniyasu, S., Rostkowski, P., Yamashita, N., Zhou, B.S., Shi, X.J., Wang, J.X., Giesy, J.P., Yu, H., Lam, P.K.S., 2007. Perfluorinated compounds in the Pearl River and Yangtze River in China. Chemosphere 68, 2085e2095. Stockholm Convention, 2009. http://chm.pops.int/Convention/ Pressrelease/OP4Geneva9May2009/tabid/542/language/en-US/ Default.aspx (accessed 10.1.2009). Sun, H.W., Gerecke, A.C., Giger, W., Alder, A.C., 2011. Long-chain perfluorinated chemicals in digested sewage sludges in Switzerland. Environ. Pollut 159, 654e662. Takagi, S., Adachi, F., Miyano, K., Koizumi, Y., Tanaka, H., Mimura, M., Watanabe, I., Tanabe, S., Kannan, K., 2008. Perfluorooctanesulfonate and perfluorooctanoate in raw and treated tap water from Osaka, Japan. Chemosphere 72, 1409e1412. Taniyasu, S., Kannan, K., Soc, M.K., 2005. Analysis of fluorotelomer alcohols, fluorotelomer acids, and short- and long-chain perfluorinated acids in water and biota. J. Chromatogr. A 1093, 89e97. Thompson, J., Lorber, M., Toms, L.L., Kato, K., Calafat, A.M., Mueller, J.F., 2010. Use of simple pharmacokinetic modeling to characterize exposure of Australians to perfluorooctanoic acid and perfluorooctane sulfonic acid. Environ. Int. 36, 390e397. Thompson, J., Eaglesham, G., Reungoat, J., Poussade, Y., Bartkow, M., Lawrence, M., Mueller, J.F., 2011. Removal of PFOS, PFOA and other perfluoroalkyl acids at water reclamation plants in South East Queensland Australia. Chemosphere 82, 9e17. Toms, L.-M.L., Calafat, A., Kato, K., Thompson, J., Harden, F., Hobson, P., Sjodin, A., Mueller, J.F., 2009. Polyfluoroalkyl chemicals (PFCs) in human blood serum from children and adults in Australia. Environ. Sci. Technol. 43, 4194e4199. US EPA, 2010. Long-chain perfluorinated chemicals (PFCs)Action Plan summary. http://www.epa.gov/opptintr/ existingchemicals/pubs/actionplans/pfcs.html. Wang, Y., Fu, J., Wang, T., Liang, Y., Pan, Y., Cai, Y., Jiang, G., 2010. Distribution of perfluorooctane sulfonate and other perfluorochemicals in the ambient environment around a manufacturing facility in China. Environ. Sci. Technol. 44, 8062e8067. Wei, S., Chen, L.Q., Taniyasu, S., So, M.K., Murphy, M.B., Yamashita, N., Yeung, L.W.Y., Lam, P.K.S., Wei, S., Chen, L.Q., 2008. Distribution of perfluorinated compounds in surface seawaters between Asia and Antarctica. Mar. Pollut. Bull. 54, 1813e1838. Yamashita, N., Kannan, K., Taniyasu, S., Horii, Y., Petrick, G., Gamo, T., 2005. A global survey of perfluorinated acids in oceans. Mar. Pollut. Bull. 51, 658e668. Yeung, L.W.Y., So, M.K., Jiang, G., Taniyasu, S., Yamashita, N., Song, M., Wu, Y., Li, J., Giesy, J.P., Guruge, K.S., Lam, P.K.S., 2006. Perfluorooctanesulfonate and related fluorochemicals in human blood samples from China. Environ. Sci. Technol. 40, 715e720. Zhang, T., Sun, H.W., Wu, Q., Zhang, X.Z., Yun, S.H., Kannan, K., 2010a. Perfluorochemicals in meat, eggs and indoor dust in China: assessment of sources and pathways of human exposure to perfluorochemicals. Environ. Sci. Technol. 44, 3572e3759. Zhang, T., Wu, Q., Sun, H.W., Zhang, X.Z., Yun, S.H., Kannan, K., 2010b. Perfluorinated compounds in whole blood samples from infants, children, and adults in China. Environ. Sci. Technol. 44, 4341e4347.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 9 1 e4 5 0 0
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Effect of temperature shocks on membrane fouling in membrane bioreactors Paula van den Brink a,b,*, On-Anong Satpradit c, Andre´ van Bentem d, Arie Zwijnenburg b, Hardy Temmink b,e, Mark van Loosdrecht a a
Department of Biotechnology, Delft University of Technology, Julianalaan 67, 2628 BC Delft, The Netherlands Wetsus, Centre for Sustainable Water Technology, PO Box 113, 8900 CC Leeuwarden, The Netherlands c van Hall Larenstein, University of Applied Sciences, PO Box 1528, 8901 BV Leeuwarden, The Netherlands d DHV Water, PO Box 1132, 3800 BC Amersfoort, The Netherlands e Department of Environmental Technology, Wageningen University, PO Box 8129, 6700 EV Wageningen, The Netherlands b
article info
abstract
Article history:
Temperature is known to influence the biological performance of conventional activated
Received 13 March 2011
sludge systems. In membrane bioreactors (MBRs), temperature not only affects the
Received in revised form
bioconversion process but is also shown to have an effect on the membrane performance.
16 May 2011
Four phenomena are generally reported to explain the higher resistance for membrane
Accepted 31 May 2011
filtration found at lower temperatures: (1) increased mixed liquor viscosity, reducing the
Available online 12 June 2011
shear stress generated by coarse bubbles, (2) intensified deflocculation, reducing biomass floc size and releasing EPS into the mixed liquor, (3) lower backtransport velocity and (4)
Keywords:
reduced biodegradation of COD. Although the higher resistance at low temperatures has
Membrane fouling
been reported in several papers, the relation with supernatant composition has not been
Temperature
investigated before. In this paper, the composition of the soluble fraction of the mixed
Membrane bioreactor
liquor is related to membrane performance after exposing the sludge to temperature
Particle size
shocks. Flux step experiments were performed in an experimental system at 7, 15, and 25
Flux step method
Celsius with sludge that was continuously recirculated from a pilot-scale MBR. After cor-
Polysaccharides
recting the permeate viscosity for temperature, higher membrane fouling rates were
Fouling mechanisms
obtained for the lower temperature in combination with low fouling reversibility. The soluble fraction of the MBR mixed liquor was analysed for polysaccharides, proteins and submicron particle size distribution. At low temperature, a high polysaccharide concentration was found in the experimental system as compared to the MBR pilot. Upon decreasing the temperature of the mixed liquor, a shift was found in particle size towards smaller particles. These results show that the release of polysaccharides and/or submicron particles from sludge flocs could explain the increased membrane fouling at low temperatures. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
Temperature is known to influence the biological performance of conventional activated sludge systems (Farrell and
Rose, 1967; McClintock et al., 1993; Metcalf and Eddy, 2004). Both the rate of treatment and the microbial composition are affected by temperature (Chiemchaisri and Yamamoto, 1994; Wile´n et al., 2000a, 2000b). Temperature also influences the
* Corresponding author. Department of Biotechnology, Delft University of Technology, Julianalaan 67, 2628 BC Delft, The Netherlands. E-mail address: [email protected] (P. van den Brink). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.05.046
4492
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 9 1 e4 5 0 0
structure of sludge flocs, although conflicting results have been reported. Experiments in which the sludge settleability was studied at different temperatures, showed an increased tendency for sludge bulking at lower temperatures (5 C) (Henry and Salenieks, 1980). Long-term monitoring of a fullscale municipal MBR revealed that settleability was positively influenced by higher temperatures (Lyko et al., 2008). In other studies, higher temperatures showed the opposite effect. An increase of sludge volume index (SVI) was found at higher temperatures (Su¨ru¨cu¨ and C ¸ etin, 1990). Poor settling was also observed at elevated temperatures (35 C as compared to 15 C) for SBR reactors running at low SRTs (Krishna and van Loosdrecht, 1999). In MBR’s, temperature has a similar effect on the biological processes as in conventional activated sludge systems. However, because temperature also has an impact on sludge morphology, the membrane filtration process can be affected. Seasonal variations in membrane fouling, with increased membrane fouling at lower temperatures, have been mentioned for several full scale systems (Rosenberger et al., 2006; Lyko et al., 2008; Miyoshi et al., 2009; Moreau, 2010; Wang et al., 2010b). Temperature has an impact on membrane filtration because it determines permeate viscosity. However, even after correcting for this change in permeate viscosity, increased membrane fouling has been quantified before for a small temperature shift (17e18 C to 13e14 C) (Jiang et al., 2005). Four phenomena, induced by low temperatures, were mentioned that could explain this effect: (1) increased mixed liquor viscosity, reducing the stress generated by coarse bubbles; (2) more severe deflocculation, reducing biomass floc size and releasing extracellular polymeric substances (EPS) into the mixed liquor that may cause membrane fouling; (3) lower particle backtransport velocity because Brownian diffusion is linearly related to temperature; (4) reduced biodegradation of COD, resulting in higher concentrations of soluble and particulate COD in the reactor (Tian et al., 1994; Lishman et al., 2000), including potential foulants. Apart from these hypotheses, other explanations or phenomena could be of importance as well. Changes in the cake layer thickness and/or porosity were possible explanations for the increased membrane fouling found at low temperatures for a laboratory scale MBR treating domestic wastewater, in which the temperature was lowered every two weeks in steps of 5 C from 25 C to 5 C (Chiemchaisri and Yamamoto, 1994). Temperature also has an effect on bubble size: for pure water, bubble coalescence and consequently bubble size was increased as the liquid temperature was raised from 10 C to 30 C (Ribeiro jr. and Mewes, 2006). Coarse bubbles are generally accepted to be more effective in fouling mitigation than smaller bubbles (Li et al., 1997; Judd, 2005; Zhang et al., 2009). Therefore, low temperature might induce membrane fouling. Although more severe membrane fouling at low temperatures has been reported in several papers, the relation with supernatant composition has not been investigated before. In the experiments presented in this paper, the composition of the
supernatant with respect to organic carbon concentrations and submicron particle size distribution and its relation to membrane fouling was investigated in short term temperature experiments. The focus on submicron particle size was chosen, because these particles are more likely to cause irreversible fouling (Geilvoet, 2010). By changing the sludge temperature, while keeping the influent composition constant, the individual temperature experiments could be more properly compared than data from full scale MBRs run at different temperatures. Operational data from a full scale MBR were used to inventorise long term temperature effects on membrane fouling. A better understanding of both short and long term temperature effects on membrane fouling in MBRs can help to develop more effective membrane fouling control strategies.
2.
Materials and methods
2.1.
Experimental set-up
A pilot-scale MBR (Fig. 1) with a working volume of 85 L and fed with municipal wastewater was used as a source of activated sludge for the experiments. The wastewater was screened (5 mm) before entering the biological reactors. The COD of the wastewater was 350 mg L1 on average, but fluctuated between 150 and 600 mg L1. The overall hydraulic retention time (HRT) of the MBR was 6 h. The dissolved oxygen concentration in the aerobic tank was measured on-line and controlled at 1.5 mg L1 with a fine-bubble diffuser. The membrane tank was equipped with 5 PE flat sheet membranes with a nominal pore size of 0.3 mm and a surface area of 0.063 m2 each (Kubota, Japan). The channel width between the membranes was 7 mm. Coarse bubble aerators placed below the membranes provided a specific aeration demand (SAD) of 1.4 m3 m2 h1 (equivalent to 0.1 cm s1) to scour the membrane surface and in this manner reduce fouling. With a peristaltic pump, permeate was continuously extracted at a flux of 47.9 L m2 h1. The pilot reactor was operated at an HRT of 6.3 h and SRT of 25 days. The average mixed liquor suspended solid (MLSS) concentration was around 4.0 g L1 in the aerobic tank of the MBR. This low MLSS concentration was due to the low strength of the municipal wastewater. The MBR pilot had good COD removal and full nitrification. To measure filterability, sludge from the aerobic reactors of the MBR was continuously recirculated over a vessel with a volume of 5 L and a retention time of 1 h (Fig. 1). The vessel contained two homemade submerged flat-sheet PVDF membranes, with a nominal pore size of 0.1 mm and a surface area of 0.014 m2. The temperature in the aerated tank of the pilot was continuously monitored. For temperature control, a cooling spiral connected to a water bath (Thermo Haake DC30-K10, US) was placed in the vessel. The filterability set-up (also called “experimental system”) was equipped with a coarse bubble aerator to scour the membrane surface at a flow rate of about 28 m3 m2 h1 (equivalent to 2.1 cm s1). This SAD was very high compared to practice to compensate for the broad channel width between the membrane sheets. Also, the air flow was divided by a relatively small surface area of the membranes used in these tests, resulting in a high
4493
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 9 1 e4 5 0 0
Fig. 1 e Pilot-scale MBR and filterability set-up (Remy et al., 2010).
specific air flow. Although the applied membranes and hydrodynamics are not the same in the pilot reactor and the filterability set-up, the individual experiments performed in the filterability set-up can be properly compared. The transmembrane pressure (TMP) was monitored on-line (Endress þ Hauser, Cerabar M PMC 41), both in the pilot MBR and in the experimental system.
2.2.
Experimental procedure
Membrane fouling can be quantified using different methods. In the past few years, efforts have been made to quantify membrane fouling, and more specifically the critical flux, in a reproducible way by developing the so-called “flux step method” (Le-Clech et al., 2003). The critical flux concept was originally developed by Field et al. (1995) and described as such: “the critical flux hypothesis for microfiltration is that on start-up there exists a flux below which a decline of flux with time does not occur; above it, fouling is observed”. The improved flux step method gives, apart from the critical flux value, additional information on reversibility and irreversibility of the membrane fouling (van der Marel et al., 2009). Flux step experiments were performed at three different temperatures in the filterability set-up: 7, 15 (reference temperature) and 25 C. These temperatures were selected because they are representative for operating conditions in winter and summer times: values from 8.4 to 26.8 C have been reported in an extensive report on filterability in pilot and full-scale MBRs all over Europe (Moreau, 2010). In order to properly compare the lower and higher temperatures with the reference temperature, experiments were performed in the following sequence: 15 e 7 e 15 e 25 e 15 C, using the same membrane for all 5 experiments. The filtrations were performed in duplicate, i.e. two membrane sheets were filtering the same mixed liquor but had separate TMP measurements. The complete sequence of filtrations was carried out 3 times to minimise effects of varying wastewater composition. After reaching the required temperature (taking 1e3 h) in the filtration set-up, flux step experiments lasting 5 h were
performed. Samples of the mixed liquor in both the filtration set-up and the aerated tank of the pilot were taken after the first flux step cycle, i.e. one filtration and one relaxation step. The filtrations were performed according to the flux step method as described by van der Marel et al. (2009), applying 5 membrane fluxes for JH: 20, 40, 60, 80, and 100 L m2 h1. Each relaxation step of 15 min was followed by a filtration step of 45 min. Before starting the experiment, a clean water filtration was performed for 30 min to determine the clean water permeability of the membranes. After each experiment, the membranes were mechanically cleaned by thorough rinsing with a nozzle to remove all accumulated fouling. The total filtration resistance was calculated from the TMP, membrane flux and permeate viscosity, according to the following expression: Rt ¼ TMP=ðhJÞ
(1)
where Rt is total resistance (m1), h the permeate viscosity (Pa s) and J the membrane flux (m3 m2 s1). A temperature correction was performed on permeate viscosity, according to the following relation (Roorda and van der Graaf, 2001) h ¼ 0:497ðT þ 42:5Þ1:5
(2)
with temperature T in degrees Celsius. This relation agrees with the values as reported in the CRC Handbook of Chemistry and Physics and first published by Sengers and Watson (1986). Total fouling rate, i.e. the increase of the total filtration resistance in time, was calculated at each flux step from the slope of the resistance in time (R(t)) after 30 min of filtration. Reversibility upon relaxation for 15 min at a membrane flux of 5 L m2 h1 was expressed as a percentage of the total resistance increase for one membrane flux step: % Reversibility ¼ Rhigh n Rlow n
Rhigh n Rlow n1 100
(3)
Rhigh,n represents the resistance at the end of one filtration step, Rlow,n represents the average resistance at the
4494
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 9 1 e4 5 0 0
corresponding relaxation step and Rlow,n1 represents the average resistance at the previous relaxation step. For the schematic representation of this calculation, the reader is referred to van den Brink et al. (2009).
2.3.
Sampling and fractionation
As mentioned above, samples of the mixed liquors of the aerobic tank of the MBR pilot and of the experimental system were taken after the first flux step cycle. Mixed liquor supernatant was obtained with a centrifuge (Sigma, 2e16) at 3260 g for 10 min. After centrifugation, the supernatant was subsequently paper filtered (Whatman Black Ribbon 589/1, 12e25 mm) and membrane filtered (Cronus PTFE syringe filter, nominal pore size of 0.45 mm). The membrane filtrate, which is referred to as the soluble fraction of the supernatant, was used for all analyses.
2.4.
Analyses
Concentrations of polysaccharides and proteins were measured according to the methods of Dubois et al. (1956) and Biorad based on Bradford (1976), respectively using glucose and bovine serum albumin (BSA) as standards. Apart from chemical techniques, fouling potential can also be assessed by measuring the particle size distribution of the mixed liquor. Particle size measurements are often used in membrane fouling studies (Wisniewski and Grasmick, 1998; Wang et al., 2008, 2010a; Zhang et al., 2010; Remy et al., 2010). However, usually particle sizes of 1 mm and higher are measured, while this is not the interesting range for a direct relation to membrane fouling. Particle size measurements in the micrometer range could be interesting to, for example, study flocculation and deflocculation. For assessing the fouling potential of a solution, rather the submicron range should be considered. Particle size distribution of the soluble fractions (<0.45 mm) was analysed with a particle size analyser based on dynamic light scattering (Coulter N4MD, US). This system is able to measure particle size down to 10 nm. Each sample was measured in three different size ranges in order to reach highest sensitivity over the entire measuring range. The minimum measured intensity was 50.000 counts per second. These data were used by the particle size analyser to create a frequency distribution. From this distribution, a Gaussian curve was created.
3.
Results
3.1.
Membrane fouling
Although also other parameters have varied considerably during this time period, full scale data on 6 years operation of the Varsseveld municipal wastewater treatment plant (the Netherlands) clearly indicate higher transmembrane pressures at lower temperatures (Fig. 2A). This effect is still apparent after correcting the trendlines to a temperature of 15 C (Fig. 2B). It can also be observed that this effect is more distinct at higher flux values. Similarly, differences in
Fig. 2 e Effect of temperature on transmembrane pressure at different flux values in full scale MBR Varsseveld, NL (A) uncorrected data (B) data corrected for permeate viscosity.
permeability of 50% were found between summer and winter periods in the full scale MBR Monheim, Germany (Wedi et al., 2009). The significant changes in microbial composition detected during the year for this plant were suggested as an explanation for these permeability differences. In order to uncouple temperature from seasonal variations, the effect of temperature on membrane fouling was tested with sludge from a pilot-scale MBR. In a test set-up the sludge of the pilot MBR was subjected to different temperatures and the impact on membrane fouling was assessed using the flux step method. After correcting the permeate viscosity for temperature, fouling resistance was observed to build up faster at lower temperatures. This effect was reproducible for each of the three sequences of filtrations that were carried out. A representative result of the filtrations performed according to the improved flux step method is provided in Fig. 3. Resistance was building up faster at the higher fluxes. Resistance increased faster and to higher values at 7 C, while the results for 15 and 25 C were similar. The relaxation steps were not very effective for the first flux steps, as the resistance during relaxation stayed at the same level as for the filtration
4495
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 9 1 e4 5 0 0
Table 1 e Average end resistances and reversibilities for different flux steps at three different temperatures. Flux step (L m2 h1) 20 40 60 80 100
Fig. 3 e Membrane resistance for sludge filtration at three different temperatures in a flux step experiment. Permeate viscosity, used in the calculations, was corrected for temperature.
steps. At the higher flux steps, a part of the fouling was removed during relaxation. Average values for the fouling rate at every flux step and temperature were calculated from three independent experiments at 7 and 25 C and from 7 experiments at 15 C (Fig. 4). The large standard deviation for the fouling rates can be explained by the sensitivity of the method as has been explained before (van den Brink et al., 2009). Fouling rates at each flux step were significantly higher when the temperature of the mixed liquor decreased from 15 to 7 C, while an increase from 15 to 25 C only gave a small decrease in fouling rate for most flux steps. Differences in fouling rates between 15 and 5 C were observed at all flux steps, but can be most clearly seen at the higher flux steps, starting at 60 L m2 h1. Apart from fouling rates, also end resistances after every filtration step and fouling reversibilities were compared (Table 1). Average values for the end resistance at every flux step and temperature were determined from three independent experiments at 7 and 25 C and from 7 experiments at 15 C. End resistances showed the same trend as fouling rates: severe fouling at 7 C, and only a small decrease in fouling when comparing 25 C to 15 C. Differences were most clearly observed at the higher flux steps from 60 L m2 h1.
7 C
15 C
25 C
11.5 14.0 21.0 39.8 85.6
9.90 11.3 14.1 19.9 29.7
10.8 11.9 13.7 16.6 21.6
Reversibility (%) 7 C 15 C 25 C 4.25 36.4 65.1 63.6 84.0
86.2 76.5 29.7 59.1 43.9
100 85.5 100 96.2 67.9
Fouling reversibility shows an interesting trend: the fouling obtained at 7 C has an increasing reversibility at higher fluxes. The opposite seems to be true for the first three flux steps at 15 C with decreasing reversibilities at increasing fluxes. At 25 C, the little fouling that was measured is also very reversible for most flux steps. To test whether these effects of temperature on membrane fouling were not (partly) caused by differences in intrinsic membrane resistance, membrane resistances were measured in dead-end filtrations of MilliQ water using Amicon cells at different temperatures. Intrinsic membrane resistances of the homemade PVDF membranes were found not to change with temperature (data not shown), and could therefore not explain the temperature effect that was found when filtering sludge.
3.2.
Particle size distribution
The average particle size for the sludge supernatant in the MBR pilot could have differed from day to day. Therefore, particle sizes in the experimental system were compared with the particle sizes of the sludge supernatant in the MBR pilot at the same time. Throughout this article, the word “particle” is also used for the fraction 0e0.45 mm, although this fraction is referred to as the “soluble fraction”. The presented graphs represent the trend found for all performed experiments. For the low temperature (7 C) (Fig. 5A), a shift can be observed from larger particles (average particle size 190 nm) to smaller particles (average particle size 130 nm) when comparing the soluble fractions of the MBR pilot (15 C) and the experimental system mixed liquors. The average particle size in the experimental system at 7 C was very close to the nominal pore size of the membranes (100 nm). Taking into account the higher membrane fouling at this low temperature, these smaller particles could be important foulants. For the other temperatures (15 and 25 C) (Fig. 5B and C), the MBR pilot peak covers the peak for the experimental system. Average particle sizes of the MBR pilot and the experimental system were more similar for 15 and 25 C than for 7 C. In all Figures, the average particle size of the MBR pilot is similar.
3.3.
Fig. 4 e Average fouling rates for different flux steps at three different temperatures.
End resistance (1010 m1)
Chemical analyses mixed liquor
The ratio between the polysaccharide concentration in the soluble fraction of the experimental system and the MBR pilot was high for 7 C, while it was around 1 for the other two temperatures (Fig. 6). For proteins, the ratios between the experimental system and the MBR pilot were around 1 for all
4496
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 9 1 e4 5 0 0
Fig. 6 e Ratio between organic carbon concentrations in the soluble fraction of the experimental system and MBR pilot at the following temperatures: 7 (n [ 3), 15 (n [ 7) and 25 (n [ 3) C.
conclusions on the relation between specific organic carbon groups and membrane fouling.
Fig. 5 e Particle size distribution in the soluble fraction of the MBR pilot (filled symbols) and experimental system (open symbols) at 7 C (A), 15 C (B) and 25 C (C).
tested temperatures. From these results, low temperature was shown to increase polysaccharide concentrations in the soluble fraction of the mixed liquor. Taking into account the higher membrane fouling at this low temperature, polysaccharides could play an important role here (Kimura et al., 2005; Rosenberger et al., 2005; Rosenberger et al., 2006). It should be noted that the standard deviations for the polysaccharide method were high, while the applied method is the best available one. More reliable methods should be developed to measure polysaccharides in order to more accurately draw
4.
Discussion
4.1.
Temperature effect membrane fouling
Both full scale data shown in Fig. 2 and experiments performed with the filterability set-up (Figs. 3e6) give a strong indication of a negative effect of temperature on membrane fouling. In the filterability set-up, this effect was found separately from seasonal influences. Apart from fouling rates also reversibility was determined, which is important for identifying the membrane fouling mechanisms involved. As mentioned in the introduction, four explanations for increased membrane fouling at low temperatures were mentioned by Jiang et al. (2005): (1) increased mixed liquor viscosity, reducing the stress generated by coarse bubbles; (2) more severe deflocculation, reducing biomass floc size and releasing extracellular polymeric substances (EPS) into the mixed liquor that may cause membrane fouling; (3) lower particle back transport velocity because Brownian diffusion is linearly related to temperature; (4) reduced biodegradation of COD, resulting in higher concentrations of soluble and particulate COD in the reactor. Both reduced stress by coarse bubbles due to increased mixed liquor viscosity and smaller bubble size, and a lower particle back transport velocity at low temperatures could not be measured in the filterability set-up. Back transport by Brownian motion is not the only force acting on the foulants. In case fouling is controlled by bubble aeration, inertial lift forces and shear induced diffusion could be more important (Belfort et al., 1994). Techniques commonly used to study bubble size are mostly optical (Yeo et al., 2006; Yamanoi and Kageyama, 2010), which makes it difficult to measure in an
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 9 1 e4 5 0 0
activated sludge mixed liquor. Non-optical techniques include intrusive and non-intrusive techniques but are all quite advanced (Yang et al., 2007). Therefore, bubble size and back transport velocity unfortunately could not be tested. Indications of stronger deflocculation were found in the short term temperature effect experiments, because of the release of EPS and submicron particles into the soluble fraction of the mixed liquor. Relatively high levels of polysaccharides were found at a low temperature in the soluble fraction. Also, a shift towards smaller particles was observed at low temperature. Short term experiments, considered as shock experiments, were performed to rule out differences in biodegradation or population changes. Longer experiments would be required to capture these phenomena. A reduction of strictly aerobic bacteria was found upon decreasing the temperature of the MBR from 25 C down to 5 C in two weeksteps (Chiemchaisri and Yamamoto, 1994). A difference in microbial composition of the sludge was observed between summer and winter periods in the full scale MBR of Monheim, Germany (Wedi et al., 2009). Decreased concentrations of protozoa and metazoa were also found upon controlled temperature shifts, both from 35 C to 45 C and from 45 C to 35 C (Morgan-Sagastume and Allen, 2003). Therefore, it seems that temperature shocks in general might negatively influence microfauna. Although the effluent quality was still high, an absence of microfauna might induce membrane fouling, as the concentration of dispersed cells will increase and there might be an accumulation of certain substrates that can foul the membrane.
4.2.
Particle size distribution soluble fraction
Apart from characterising the dissolved organic carbon in the mixed liquor, measuring the particle size distribution is of interest as well to get to know the fouling potential of this mixed liquor. When measuring particle size distribution, the aim of the measurement should always be taken into account. In case the floc size of the activated sludge is monitored, particle sizes in the order of (tens of) micrometers need to be measured. Because the cake layer that is formed by flocs is mostly reversible, focus has shifted to particles in the submicron range, which are more likely to cause irreversible fouling (Geilvoet, 2010). In order to relate particle size distribution of a solution to its fouling potential, measurements in this range are to be performed. Often, particle sizes are only measured accurately down to 0.4e0.45 mm, while these measured particles are then claimed to foul membranes with pore sizes that are several times smaller (Wisniewski et al., 2000; Jiang et al., 2007; Geilvoet, 2010). Reported particle size distributions and associated conclusions on membrane fouling potential of solutions should therefore be carefully evaluated for their validity. Decreased temperature was reported to cause deflocculation (Wile´n et al., 2000a). Also in our experiments, a shift in particle size distribution to smaller particles was observed upon a temperature decrease to 7 C, while this was not the case for 15 and 25 C (Fig. 5). For non-interacting particles, smaller particles are known to cause a less permeable cake layer, because they reduce the void fraction of the cake and thereby increase the specific cake resistance according to the
4497
CarmaneKozeny equation (Carman, 1938). Small particles are also back transported less efficiently from the membrane surface (Lahoussine-Turcaud et al., 1990). In addition, the closer the particle size distribution of the mixed liquor soluble fraction gets to the pore size distribution, the more likely it is that pore blocking and pore narrowing become important. Pore blocking and pore narrowing are known to be irreversible fouling mechanisms. This agrees with the low fouling reversibility values found for the first flux steps at low temperature (Table 1). Apart from increased deflocculation at lower temperatures, also decreased (re)flocculation might be expected. For coagulation of kaolinite suspensions, especially the slow growth of small flocs was severely influenced by temperature: at 2 C, the slow coagulation rate was greatly reduced as compared to 22 C (Xiao et al., 2008). This was explained by reduced Brownian diffusion.
4.3.
Organic carbon characterisation of soluble fractions
Protein and polysaccharide measurements showed different behaviour: low temperature was shown to increase polysaccharide concentrations in the soluble fraction of the mixed liquor, while no effect of temperature was observed on protein levels. This is in line with the smaller particles found at lower temperatures (Fig. 5A). The nature of these submicron particles in the pore size range could well be polysaccharides. For example, for the alginate molecule radii of gyration have been reported ranging from 50 7 to 137 nm (Wedlock et al., 1986; Bowen and Cooke, 1991; Windhues and Borchard, 2002). Increased membrane fouling was linked before to polysaccharide concentration in the mixed liquor (Kimura et al., 2005; Rosenberger et al., 2005; Rosenberger et al., 2006). Also, low temperature could be as an environmental stress factor to the sludge flocs, causing the release of EPS or SMP into the solution (Barker and Stuckey, 1999; Rosenberger et al., 2006). It should be noted that the polysaccharide and protein methods are not very sensitive (Mehrez et al., 2007), which makes it hard to measure small differences in concentrations. Furthermore, the applied method (DuBois) is known to give a higher response to smaller polysaccharide molecules and monosaccharides. It would be interesting to use more specific organic carbon analyses than polysaccharide and protein measurements in order to study the relation between mixed liquor composition and membrane fouling. This was also mentioned by Kimura et al. (2005) who performed more specific FTIR and 13C NMR analyses on the organic carbon. While concentrations of dissolved organic matter in the reactors and mixed liquor viscosity did not show a clear relationship with membrane fouling in their study, the food-microorganisms ratio turned out to be important for membrane fouling because it influences the type of organic carbon molecules found in the membrane fouling layer.
4.4.
Fouling mechanisms
In order to identify the fouling mechanisms that are playing a role in these experiments, it is important to look at the reversibility values (Table 1). At a lower temperature, a higher
4498
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 9 1 e4 5 0 0
amount of submicron particles in the pore size range was found (Fig. 5A), in combination with a lower reversibility for the first number of fluxes. This indicates a fast and irreversible fouling at the start of filtration, caused by direct pore blocking or narrowing of the membrane pores (Belfort et al., 1994). Also adsorption of soluble compounds might play a role here. At the later (higher) flux steps, after the initial fouling, cake layer formation will become dominant as the membrane surface and pores are already fouled. Cake layer formation is generally regarded as a reversible fouling mechanism, removable by both relaxation and backwashing (van der Marel et al., 2010). This explains the higher reversibility at the later flux steps. Submicron particles influence cake layer formation in two ways: 1. Because of their small size their backtransport velocity is smaller because the physical forces acting on the submicron particle during membrane filtration, such as inertial lift, are smaller than for larger particles (Altmann and Ripperger, 1997; Bhattacharjee and Hong, 2005). This results in a relatively high amount of small particles in the cake. 2. The combination of smaller and bigger particles results in a less permeable cake as the void fraction of the cake decreases (Carman, 1938; Le-Clech et al., 2006). As a result of these two phenomena, smaller particles are retained in the cake layer and thereby accumulate in the cake. Preferential deposition of the smaller particles in a polydisperse suspension was shown by theoretical work and experimentally (Baker et al., 1985; Lu and Ju, 1989). Moreover, the smallest particles in a poly-dispersed suspension often determine the filtration rate (Kim et al., 2002; Kromkamp et al., 2002). For a temperature of 15 C, reversibility was high at the beginning of filtration and decreased towards the highest fluxes. This information combined with the particle size distribution data would suggest the development of a cake or gel that is getting thicker during the filtration. Towards the last flux steps, compression of the cake might have taken place (van den Brink et al., 2009), explaining the decreasing reversibility. At 25 C, very little fouling was observed with a reversibility of close to 100%. Because there is so little fouling at this temperature, compression could not have played a role here. Moreover, when there is little fouling the TMP stays low and the fouling can therefore be more easily back transported into the mixed liquor. This can explain the high fouling reversibility at 25 C.
4.5.
Practical relevance
Studying the effect of temperature on membrane fouling in MBRs is very relevant for practice. As shown in Fig. 2, even after correction for permeate viscosity, lower temperatures caused higher TMPs and thus resistances at the full scale plant of Varsseveld, the Netherlands. Similar results were obtained by the Delft Filtration Characterisation method that was applied in several pilot and full scale installations all over Europe (Moreau, 2010). Unfortunately, no full scale data were available of plants with a relatively constant temperature all
year round. Perhaps, overall MBR performance would be higher in warmer countries. A pilot study with a submerged MBR in Singapore, run at an average mixed liquor temperature of 30 C, showed stable operation at net fluxes of 25e29 L m2 h1 (Qin et al., 2009). The 5-mgd demo plant at Ulu Pandan (Singapore) has been in operation since December 2006 and had stable operation at membrane flux of 25.3 L m2 h1 (website PUB, 2008). When comparing a number of large scale pilot trials in which several MBR technologies were tested in Europe, USA and Singapore, net fluxes were found to be higher in the warmer countries (van der Roest et al., 2002; Adham et al., 2004; Tao et al., 2005). It should be noted that in many full scale systems conservative average fluxes are applied (Judd, 2006). As shown in this study, temperature influences membrane fouling in MBRs. Therefore, temperature should be taken into account in the design of such systems. In countries with seasonal temperature variations, extra membrane area should be installed for the colder periods. By using the extra membrane area during winter time, fluxes can be lowered in order to control membrane fouling. In countries with constant temperature during the year, fewer membranes need to be installed. Higher and more constant fluxes should be reached in this case, and these places are favourable for applying MBRs. In order to react on temperature shocks it is important to develop control strategies. Addition of low amounts of Powdered Activated Carbon (PAC) or other inert particles to the activated sludge was shown to make the sludge flocs stronger and less susceptible to factors that induce deflocculation such as shear (Lee et al., 2001; Remy et al., 2009, 2010). Perhaps PAC addition could also prevent the deflocculation of the sludge flocs when it is subjected to a temperature decrease. In winter periods, this could offer a rather straightforward solution to decreased filterability. Another option, focused on cleaning instead of fouling prevention, is to apply (chemically) enhanced backwashing, a method that is applied in both traditional membrane filtration processes and membrane bioreactor systems (MBR) (Teodosiu et al., 1999; LeClech et al., 2006). Backwashing is not applicable when flat sheet membranes are used. Other possibilities for controlling membrane fouling are for example decreasing the flux or increasing membrane aeration.
5.
Conclusions
Flux step experiments showed increased membrane fouling at lower temperatures. Resistances calculated with a temperature corrected permeate viscosity gave increased fouling rates at lower temperatures. The combination with low fouling reversibility yielded a detrimental effect of low temperature on permeability. Analyses of the mixed liquor soluble fraction showed a high polysaccharide concentration in the experimental system as compared to the MBR pilot. Upon decreasing the temperature of the mixed liquor, a shift was found in particle size distribution of the soluble fraction towards smaller particles. Because these smaller particles are very similar in size to the membrane pore size, fast pore narrowing or pore blocking could explain the low fouling reversibility.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 9 1 e4 5 0 0
The results showed that the release of polysaccharides or submicron particles from sludge flocs could explain the increased membrane fouling at low temperatures. Consequently, temperature should be taken into account when designing a MBR.
Acknowledgements This work was performed in the TTIW-cooperation framework of Wetsus, Centre of Excellence for Sustainable Water Technology (www.wetsus.nl). Wetsus is funded by the Dutch Ministry of Economic Affairs, the European Union Regional Development Fund, the Province of Fryslaˆn, the City of Leeuwarden and the EZ/Kompas program of the ‘Samenwerkingsverband Noord-Nederland”. The authors like to thank the participants of the research theme “Membrane Bioreactors” for the discussions and their financial support.
references
Adham, S., DeCarolis, J.F., Pearce, W., 2004. Optimisation of Various MBR Systems for Water Reclamation e Phase II Desalination and Water Purification Research and Development Program Final Report, No. 103. Altmann, J., Ripperger, S., 1997. Particle deposition and layer formation at the crossflow microfiltration. J. Mem. Sci. 124 (1), 119e128. Baker, R.J., Fane, A.G., Fell, C.J.D., Yoo, B.H., 1985. Factors affecting flux in crossflow filtration. Desalination 53 (1e3), 81e93. Barker, D.J., Stuckey, D.C., 1999. A review of soluble microbial products (SMP) in wastewater treatment systems. Water Res. 33 (14), 3063e3082. Belfort, G., Davis, R.H., Zydney, A.L., 1994. The behavior of suspensions and macromolecular solutions in crossflow microfiltration. J. Mem. Sci. 96 (1e2), 1e58. Bhattacharjee, S., Hong, S., 2005. Fundamentals of particle fouling in membrane processes. Korean Mem. J. 7 (1), 1e18. Bowen, W.R., Cooke, R.J., 1991. Properties of microfiltration membranes: computer automated determination of the electrokinetic properties of polycarbonate membranes. J. Coll. Int. Sci. 141 (1), 280e287. Bradford, M., 1976. A rapid and sensitive method for the quantification of microgram quantities of protein utilizing the principle of protein-dye binding. Anal. Biochem. 72, 248e254. Carman, P.C., 1938. Fundamental principles of industrial filtration. Trans. Inst. Chem. Eng. 16, 168e187. Chiemchaisri, C., Yamamoto, K., 1994. Performance of membrane separation bioreactor at various temperatures for domestic wastewater treatment. J. Mem. Sci. 87 (1e2), 119e129. Dubois, M., Gilles, K.A., Hamilton, J.K., Rebers, P.A., Smith, F., 1956. Colorimetric method for determination of sugars and related substances. Anal. Chem. 28 (3), 350e356. Farrell, J., Rose, A.H., 1967. Temperature effects on microorganisms. In: Rose, A.H. (Ed.), Thermobiology. Academic Press, London, pp. 147e218. Field, R.W., Wu, D., Howell, J.A., Gupta, B.B., 1995. Critical flux concept for microfiltration fouling. J. Mem. Sci. 100 (3), 259e272. Geilvoet, S.P., (2010). The Delft filtration characterisation method. assessing membrane bioreactor activated sludge filterability. PhD thesis, Delft University of Technology, Delft, The Netherlands.
4499
Henry, J.G., Salenieks, E.E., 1980. Variation in sludge settleability with temperature. Water Poll. Res. J. Can. 15 (1), 73e82. http://www.pub.gov.sg/research/Key_Projects/Pages/ Membrane2.aspx Jiang, T., Kennedy, M.D., Guinzbourg, B.F., Vanrolleghem, P.A., Schippers, J.C., 2005. Optimising the operation of a MBR pilot plant by quantitative analysis of the membrane fouling mechanism. Water Sci. Technol. 51 (6e7), 19e25. Jiang, T., Kennedy, M.D., Yoo, C., Nopens, I., van der Meer, W., Futselaar, H., Schippers, J.C., Vanrolleghem, P.A., 2007. Controlling submicron particle deposition in a side-stream membrane bioreactor: a theoretical hydrodynamic modelling approach incorporating energy consumption. J. Mem. Sci. 297 (1e2), 141e151. Judd, S., 2005. Fouling control in submerged membrane bioreactors. Water Sci. Technol. 51 (6e7), 27e34. Judd, S., 2006. The MBR Book: Principles and Applications of Membrane Bioreactors for Water and Wastewater Treatment. Elsevier, Oxford. Kim, S., Cho, S.-H., Park, H., 2002. Effects of particle size distribution on the cake formation in crossflow microfiltration. Water Sci. Technol. Water Supply 2 (2), 305e311. Kimura, K., Yamato, N., Yamamura, H., Watanabe, Y., 2005. Membrane fouling in pilot-scale membrane bioreactors (MBRs) treating municipal wastewater. Environ. Sci. Technol. 39 (16), 6293e6299. Krishna, C., van Loosdrecht, M.C.M., 1999. Effect of temperature on storage polymers and settleability of activated sludge. Water Res. 33 (10), 2374e2382. Kromkamp, J., van Domselaar, M., Schroen, K., van der Sman, R., Boom, R., 2002. Shear-induced diffusion model for microfiltration of polydisperse suspensions. Desalination 146 (1e3), 63e68. Lahoussine-Turcaud, V., Wiesner, M.R., Bottero, J.-Y., 1990. Fouling in tangential-flow ultrafiltration: the effect of colloid size and coagulation pretreatment. J. Mem. Sci. 52 (2), 173e190. Le-Clech, P., Chen, V., Fane, A.G., 2006. Fouling in membrane bioreactors used in wastewater treatment e A review. J. Mem. Sci. 284 (1e2), 17e53. Le-Clech, P., Jefferson, B., Chang, I.S., Judd, S.J., 2003. Critical flux determination by the flux-step method in a submerged membrane bioreactor. J. Mem. Sci. 227 (1e2), 81e93. Lee, J.C., Kim, J.S., Kang, I.J., Cho, M.H., Park, P.K., Lee, C.H., 2001. Potential and limitations of alum or zeolite addition to improve the performance of a submerged membrane bioreactor. Water Sci. Technol. 43 (11), 59e66. Li, Q.Y., Cui, Z.F., Pepper, D.S., 1997. Effect of bubble size and frequency on the permeate flux of gas sparged ultrafiltration with tubular membranes. Chem. Eng. J. 67 (1), 71e75. Lishman, L.A., Legge, R.L., Farquhar, G.J., 2000. Temperature effects on wastewater treatment under aerobic and anoxic conditions. Water Res. 34 (8), 2263e2276. Lu, W.-M., Ju, S.-C., 1989. Selective particle deposition in crossflow filtration. Sep. Sci. Technol. 24 (7e8), 517e540. Lyko, S., Wintgens, T., Al-Halbouni, D., Baumgarten, S., Tacke, D., Drensla, K., Janot, A., Dott, W., Pinnekamp, J., Melin, T., 2008. Long-term monitoring of a full-scale municipal membrane bioreactor-Characterisation of foulants and operational performance. J. Mem. Sci. 317 (1e2), 78e87. McClintock, S.A., Randall, C.W., Pattarkine, V.M., 1993. Effects of temperature and mean cell residence time on biological nutrient removal processes. Water Environ. Res. 65 (5), 110e118. Mehrez, R., Ernst, M., Jekel, M., 2007. Development of a continuous protein and polysaccharide measurement method by sequential injection analysis for application in
4500
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 9 1 e4 5 0 0
membrane bioreactor systems. Water Sci. Technol. 56 (6), 163e171. Metcalf and Eddy, 2004. Wastewater Engineering: Treatment and Reuse (4th International Edition). McGraw-Hill, New York, USA. Miyoshi, T., Tsuyuhara, T., Ogyu, R., Kimura, K., Watanabe, Y., 2009. Seasonal variation in membrane fouling in membrane bioreactors (MBRs) treating municipal wastewater. Water Res. 43 (20), 5109e5118. Moreau, A.A., (2010), Filterability assessment of membrane bioreactors at European scale. PhD thesis, Delft University of Technology, Delft, The Netherlands. Morgan-Sagastume, F., Allen, D.G., 2003. Effects of temperature transient conditions on aerobic biological treatment of wastewater. Water Res. 37 (15), 3590e3601. accessed 20.01.11. Qin, J.J., Oo, M.H., Tao, G., Kekre, K.A., Hashimoto, T., 2009. Pilot study of a submerged membrane bioreactor for water reclamation. Water Sci. Technol. 60 (12), 3269e3274. Remy, M., Potier, V., Temmink, H., Rulkens, W., 2010. Why low powdered activated carbon addition reduces membrane fouling in MBRs. Water Res. 44 (3), 861e867. Remy, M., van der Marel, P., Zwijnenburg, A., Rulkens, W., Temmink, H., 2009. Low dose powdered activated carbon addition at high sludge retention times to reduce fouling in membrane bioreactors. Water Res. 43 (2), 345e350. Ribeiro Jr., C.P., Mewes, D., 2006. On the effect of liquid temperature upon bubble coalescence. Chem. Eng. Sci. 61 (17), 5704e5716. Roorda, J.H., van der Graaf, J.H.J.M., 2001. New parameter for monitoring fouling during ultrafiltration of WWTP effluent. Water Sci. Technol. 43 (10), 241e248. Rosenberger, S., Evenblij, H., te Poele, S., Wintgens, T., Laabs, C., 2005. The importance of liquid phase analyses to understand fouling in membrane assisted activated sludge processes e six case studies of different European research groups. J. Mem. Sci. 263 (1e2), 113e126. Rosenberger, S., Laabs, C., Lesjean, B., Gnirss, R., Amy, G., Jekel, M., Schrotter, J.-C., 2006. Impact of colloidal and soluble organic material on membrane performance in membrane bioreactors for municipal wastewater treatment. Water Res. 40 (4), 710e720. Sengers, J.V., Watson, J.T.R., 1986. Improved international formulations for the viscosity and thermal conductivity of water substance. J. Phys. Chem. Ref. Data 15 (4), 1291e1314. Su¨ru¨cu¨, G., C¸etin, F.D., 1990. Effects of temperature, pH and D.O. concentration on settleability of activated sludge. Env. Technol. 11 (3), 205e212. Tao, G., Kekre, K., Wei, Z., Lee, T.C., Viswanath, B., Seah, H., 2005. Membrane bioreactors for water reclamation. Water Sci. Technol. 51 (6e7), 431e440. Teodosiu, C.C., Kennedy, M.D., van Straten, H.A., Schippers, J.C., 1999. Evaluation of secondary refinery effluent treatment using ultrafiltration membranes. Water Res. 33 (9), 2172e2180. Tian, S., Lishman, L.A., Murphy, K.L., 1994. Investigations into excess activated sludge accumulation at low temperatures. Water Res. 28 (3), 501e509. van den Brink, P., Zwijnenburg, A., Smith, G., Temmink, H., van Loosdrecht, M.C.M., 2009. Effect of free calcium concentration and ionic strength on alginate fouling in cross-flow membrane filtration. J. Mem. Sci. 345 (1e2), 207e216. van der Marel, P., Zwijnenburg, A., Kemperman, A.J.B., Wessling, M., Temmink, H., van der Meer, W.G.J., 2009. An improved flux-step method to determine the critical flux and
the critical flux for irreversibility in a membrane bioreactor. J. Mem. Sci. 332 (1e2), 24e29. van der Marel, P., Zwijnenburg, A., Kemperman, A.J.B., Wessling, M., Temmink, H., van der Meer, W.G.J., 2010. Influence of membrane properties on fouling in submerged membrane bioreactors. J. Mem. Sci. 348 (1e2), 66e74. van der Roest, H.F., Lawrence, D.P., van Bentem, A.G.N., 2002. Membrane Bioreactors for Municipal Wastewater Treatment. IWA Publishing. Wang, Z., Wang, P., Wang, Q., Wu, Z., Zhou, Q., Yang, D., 2010a. Effective control of membrane fouling by filamentous bacteria in a submerged membrane bioreactor. Chem. Eng. J. 158 (3), 608e615. Wang, Z., Wu, Z., Tang, S., 2010b. Impact of temperature seasonal change on sludge characteristics and membrane fouling in a submerged membrane bioreactor. Sep. Sci. Technol. 45 (7), 920e927. Wang, Z., Wu, Z., Yin, X., Tian, L., 2008. Membrane fouling in a submerged membrane bioreactor (MBR) under sub-critical flux operation: membrane foulant and gel layer characterization. J. Mem. Sci. 325 (1), 238e244. Wedi, D., Bleisteiner, S., Wild, W., 2009. Seasonal changes in filterability and permeability shown by the example of MBRMonheim. In: Proceedings of the 8. Aachener Tagung Wasser und Membranen, Aachen, Germany A26-1eA26-11. Wedlock, D.J., Fasihuddin, B.A., Phillips, G.O., 1986. Comparison of molecular weight determination of sodium alginate by sedimentation-diffusion and light scattering. Int. J. Biol. Macromol. 8 (1), 57e61. Wile´n, B.-M., Keiding, K., Nielsen, P.H., 2000a. Anaerobic deflocculation and aerobic reflocculation. Water Res. 34 (16), 3933e3942. Wile´n, B.-M., Nielsen, J.L., Keiding, K., Nielsen, P.H., 2000b. Influence of microbial activity on the stability of activated sludge flocs. Colloid Surf. B. 18 (2), 145e156. Windhues, T., Borchard, W., 2002. Temperature depending light scattering measurements of aqueous gelatin and alginate solutions and their mixtures. Eur. Polym. J. 38 (6), 1219e1227. Wisniewski, C., Grasmick, A., 1998. Floc size distribution in a membrane bioreactor and consequences for membrane fouling. Coll. Surf. A 138 (2e3), 403e411. Wisniewski, C., Grasmick, A., Leon Cruz, A., 2000. Critical particle size in membrane bioreactors: case of a denitrifying bacterial suspension. J. Mem. Sci. 178 (1e2), 141e150. Xiao, F., Ma, J., Yi, P., Huang, J.-C.H., 2008. Effects of low temperature on coagulation of kaolinite suspensions. Water Res. 42 (12), 2983e2992. Yamanoi, I., Kageyama, K., 2010. Evaluation of bubble flow properties between flat sheet membranes in membrane bioreactor. J. Mem. Sci. 360 (1e2), 102e108. Yang, G.Q., Du, B., Fan, L.S., 2007. Bubble formation and dynamics in gas-liquid-solid fluidization-A review. Chem. Eng. Sci. 62 (1e2), 2e27. Yeo, A.P.S., Law, A.W.K., Fane, A.G., 2006. Factors affecting the performance of a submerged hollow fiber bundle. J. Membr. Sci. 280 (1e2), 969e982. Zhang, K., Cui, Z., Field, R.W., 2009. Effect of bubble size and frequency on mass transfer in flat sheet MBR. J. Mem. Sci. 332 (1e2), 30e37. Zhang, Y.P., Fane, A.G., Law, A.W.K., 2010. Critical flux and particle deposition of fractal flocs during crossflow microfiltration. J. Mem. Sci. 353 (1e2), 28e35.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 0 1 e4 5 1 0
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Iron oxide amended biosand filters for virus removal Ian Bradley, Anthony Straub, Peter Maraccini 1, Sheila Markazi, Thanh H. Nguyen* Department of Civil and Environmental Engineering, The Center of Advanced Materials for the Purification of Water with Systems, University of Illinois at Urbana-Champaign, Urbana, IL 61801, United States
article info
abstract
Article history:
Laboratory studies were performed to determine if the addition of iron oxides throughout
Received 25 March 2011
biosand filter (BSF) media would increase virus removal due to adsorption. The proposed
Received in revised form
mechanism is electrostatic adsorption of negatively charged virion particles to positively
25 May 2011
charged iron oxides formed during the corrosion of zerovalent iron. Initial tests conducted
Accepted 31 May 2011
using continuous flow, small-scale glass columns showed high MS2 bacteriophage removal
Available online 12 June 2011
in an iron-amended sand column (5log10) compared to a sand-only column (0.5log10) over 20 pore volumes. Additionally, two experiments with a column containing iron particles
Keywords:
revealed 4log10 and 5log10 removal of rotavirus in the presence of 20 mg/L total organic
Drinking water
carbon. Full-scale BSFs with iron particles removed >4log10 MS2 for the duration of the
Point-of-use technologies
experiment (287 days), while BSF with steel wool removed >4log10 MS2 for the first 160
Physical and chemical processes
days. Plug flow for the BSF was shown to depend on uniformity between the iron oxide
Virus
material and sand media grains. The results suggest that the duration of effective virus removal by iron-amended biosand filtration depends on source water conditions and the quantity and composition of iron material added. Overall, this study provides evidence that iron-amended BSFs may advance the field of point-of-use technologies and bring relief to millions of people suffering from waterborne diseases. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
An estimated 884 million people e13% of the world population e lack access to safe drinking water (UNICEF/WHO, 2008). As a result, millions of people die each year from water-related diseases (WHO, 2008). While it is not possible to quantify the proportion of deaths directly due to unsafe drinking water and not attributed to other fecaleoral transmission routes (Curtis et al., 2000), access to clean drinking water and proper sanitation can provide substantial improvements in health (Logsdon et al., 2002; Nelson and Murray, 2008). Point-of-use (POU) water treatments, which allow the purification of water at the point of consumption rather than at a centralized location, allow water quality to be improved at the household scale (Sobsey, 2002).
Already widespread in their usage, as of 2007, 19 million people are estimated to use POU water treatment, in addition to the 350 million people who boil their water (Clasen et al., 2007). Studies indicate that the improvement of water quality through the use of POU technologies results in 30e40% reductions in diarrheal disease (Clasen et al., 2007; Esrey et al., 1985, 1991; Fewtrell et al., 2005). One of the most promising and widespread POU technologies is the biosand filter (BSF), a household-scale, intermittently operated slow sand filter, in which the upper layers of sand media remain saturated in between operations to allow the formation of a biologically active layer (Sobsey et al., 2008). The BSF consists of a plastic or concrete hollow chamber that tapers slightly toward the bottom (CAWST, 2010). A drainage gravel layer is laid at the bottom of the chamber,
* Corresponding author. Tel.: þ1 217 244 5965; fax: þ1 217 333 6968. E-mail address: [email protected] (T.H. Nguyen). 1 Present address: Department of Civil and Environmental Engineering, Stanford University, United States. 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.05.045
4502
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 0 1 e4 5 1 0
covered by a separating gravel layer and a filtration sand layer. Approximately 5 cm above the filtration sand layer sits a diffuser. An outlet tube collects water from bottom of the chamber, passes the water vertically, and discharges the water at the outlet located at a height between the diffuser and the top of the filtration sand layer. During 24-h cycle of filter usage, water is poured into the inlet reservoir. As a result, the hydraulic head pushes the water downwards through the sand filtration layer and into the drainage gravel layer, where it is collected by the outlet tube and ultimately discharged. As the water level drops in the inlet reservoir, the flow rate decreases. Flow ceases when the standing water within the inlet reservoir is at a height equal to the height of the outlet. The top portion of the filtration sand layer collects the suspended solids, organic constituents, and microorganisms of the source water. Since the filtration sand remains saturated during and in between operation, a biological zone, wherein the sand grains are covered in a biofilm, develops in the top 10e20 cm of the filtration sand layer. The biofilm is credited with the enhanced removal of suspended solids and pathogens through mechanical trapping, adsorption, predation, and natural death. Development of the biofilm results in greater removal of suspended solids and pathogens, but its development also reduces flow rates (Elliott, 2010; Elliott et al., 2008; Weber-Shirk and Dick, 1997). Dr. David Manz developed the first BSFs in the 1990s at the University of Calgary as a way of improving water quality for low-income families in rural areas with restricted access to safe drinking water (Manz et al., 1993). Since then, BSFs have been chosen by hundreds of humanitarian groups as the best method for improving water quality in developing countries and, as of 2009, it is estimated that over 200,000 BSFs have been implemented in over 70 countries (CAWST, 2010). Surveys reveal its wide acceptance by users due to the improved appearance, smell, and taste of the treated water (Ngai et al., 2007). Considering the criteria of water quantity produced, water quality, ease of use, and ease of access, BSFs have been identified as the point-of-use technology having the most potential to deliver sustainable potable water treatment to the developing world (Sobsey et al., 2008). Both laboratory and field studies have documented improved microbiological water quality through the use of the BSF. BSFs remove greater than 99.9% of Giardia cysts and Cryptosporidium oocysts (Palmateer et al., 1999). Bacterial concentrations are reduced 70e99.99%, depending on biofilm development and time of sampling (Baumgartner et al., 2007; Buzunis, 1995; Elliott et al., 2008; Stauber, 2006). The improved water quality has been attributed to at least 20% reductions in frequency of diarrheal illness in studies conducted in the Dominican Republic (Stauber, 2006) and Kenya (Tiwari et al., 2009). However, both field and laboratory researches have identified a critical shortcoming: BSFs are not highly effective in removing viruses (Elliott et al., 2008). Viruses cause approximately 40% of diarrheal illnesses in developing countries (Ramani and Kang, 2009), with rotavirus being the leading cause of childhood diarrhea hospitalizations worldwide (Parashar et al., 2006). In natural water conditions of pH 6e8, sand and most viruses are negatively charged, causing a net repulsion and reducing virus removal efficacy by sand filtration (Jin et al., 2000). Thus, water contaminated with
pathogenic viruses is not yet potable after passing through a BSF, and a form of virus removal is required to treat the effluent (CAWST, 2010). The addition of zerovalent iron to the sand media results in filters that more effectively remove viruses from water (You et al., 2005) as corrosion on the iron surface generates a positively charged oxide layer (Lukasik et al., 1999) to which the viruses may electrostatically adsorb (Ryan et al., 1999). The primary objectives of this study were to determine: (1) the efficacy of virus removal during the daily operation of the iron-amended BSF; (2) the efficacy of virus removal using different iron oxide sources; (3) the duration for which ironamended biosand filtration effectively removed viruses. This study is unique to other studies for two reasons. First, unlike previous iron-amended BSFs, the iron source was added to the top half of the sand media rather than in the diffuser basin to increase the contact time between viruses and iron oxides. Second, to our knowledge, this is the longest BSF study conducted examining virus removal in both iron-amended and unmodified BSFs. Both small-scale columns and householdscale BSFs were tested using bacteriophage MS2. Rotavirus was used for select tests with small-scale columns, due to difficulty in propagating the virus.
2.
Methods
2.1.
Virus selection and assay
Bacteriophage male specific type 2 (MS2) was selected as a model virus because of its structural resemblance to many human enteric viruses and its ease of use. MS2 was replicated and purified as described previously (Gutierrez et al., 2009, 2010; Kitis et al., 2003; Page et al., 2009; Sirikanchana et al., 2008) with the following modifications. Briefly, Escherichia coli (ATCC 15597) grown in tryptic soy broth solution was inoculated with MS2 and incubated, followed by the separation of MS2 via centrifugation. After cell lysis and virus release, debris was removed via microfiltration through 0.2mm and 0.05-mm low-protein-binding polycarbonate tracketched membranes (Whatman Nucleopore, USA). Virus was concentrated on a 100-kDa membrane surface (Koch Membranes, USA) in a Millipore ultra-microfiltration unit (Whatman Nucleopore, USA). The virus accumulated on the membrane surface was washed extensively with sterilized 1 mM NaCl solution to remove nutrients, microbial products and dissolved organic matter. The final MS2 stock was stored at 4 C at a concentration of 1011 PFU/mL. MS2 was enumerated by the double agar layer procedure USEPA Method 1602. Briefly, plaques formed due to the inoculation of E. coli with MS2 at 37 C for 16 h, and plates with between 20 and 200 plaques were used for calculating the concentration of MS2. Any plates containing more than 200 plaques were quantified from a higher dilution plate. Select experiments were also performed using rotavirus (RV) to verify MS2 results. Group A porcine rotavirus OSU strain was obtained from the American Type Culture Collection (catalog # VR892). Rotavirus was propagated in embryonic African green monkey kidney cells (MA-104 cells) and extracted from culture as described by Rolsma et al. (1994).
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 0 1 e4 5 1 0
The purification protocol was the same as for MS2, except with an additional filtration step in which a 0.05-mm membrane was used. Rotavirus was cleaned and stored in 1 mM sodium chloride (NaCl) plus 0.1 mM calcium chloride (CaCl2) during the 100 kDa ultrafiltration to prevent the dissociation of the outer capsid proteins (Ruiz et al., 1996). The final RV stock was stored at 4 C at a concentration of >105 focus forming unit (FFU) per mL. Rotavirus infectivity assays were carried out following the procedures described by Rolsma et al. (1998).
2.2.
Solution chemistry
Newmark groundwater (NGW) was used as the background solution. Collected from a natural aquifer underneath the Newmark Civil Engineering Laboratory (205 N. Matthews, Urbana, Illinois, 61801), NGW is greensand-filtered for manganese and iron removal. NGW has been characterized and used in previous studies (Li et al., 2003). Content of total organic carbon (TOC) was measured using a Phoenix 8000 TOC Analyzer (Dohrmann, USA) in previous studies and found to be 2.35 mg/L (Gutierrez et al., 2009). Turbidity of the groundwater and the effluent were 0.25 NTU and 0.70 NTU, respectively, as measured by a Hach Turbidity Meter 43900. Newmark groundwater was chosen over available surface waters for the following reasons: (1) Nearby river water would be impacted by the seasonal runoff from agricultural fields in the surrounding and upstream area, resulting in shifting water chemistry throughout the course of the study. (2) Stagnant water bodies would freeze over in the wintertime, causing inconsistent water collection over the duration of the experiment, which spans multiple months. (3) Newmark groundwater has already been documented and provides a consistent water quality that better suites the longevity of the project. Pasteurized primary wastewater effluent from a local wastewater treatment plant was used to establish a biofilm in the MS2 experiments, as previously suggested by Elliott et al. (2008). Treated wastewater effluent from the same treatment plant was used to conduct rotavirus experiments with high levels (20 mg/L TOC) of natural organic matter (NOM). This wastewater treatment plant uses conventional activated sludge treatment.
2.3.
Column experiments
2.3.1.
Continuous flow test for MS2
Small-scale column tests were conducted to compare results with previous research examining MS2 removal through sand and sand/iron oxide columns (You et al., 2005), and to determine what effect, if any, orientation of iron particles in the sand column has on virus removal. Results were used to determine how household-scale filters would be incorporated with iron as an addition to the sand media. In addition, smallscale columns were used to examine the removal of rotavirus in the proposed design. Due to the difficulty of propagating the virus and the maximum concentrations necessary for use in the BSF, experiments on the household-scale filter with rotavirus were not possible. The glass columns had the following dimensions: 3.8 cm tapered ends, 8.9 cm main body length,
4503
2.5 cm inner diameter, and a total volume between 146.7 cm3 and 161.5 cm3 (Fig. 1). The control column was dry packed with 224.1 g clean quartz sand. The iron column was packed with a 10% iron by volume mixture with sand (30.4 g iron; 182.7 g sand) and a layer of sand only (42 g) at the influent. Zerovalent iron particles (ETI8/50) used in the column studies were obtained from Peerless Metal Powders & Abrasive (Detroit, MI). The iron was used without pretreatment. Sand used in the column studies was an industrial quartz obtained from Unimin Corporation (Le Sueur, MN) described as 5020 (20% retained on 50 mesh or coarser) and 0.15 mm effective size in filtration. The sand was washed with deionized water until the supernatant was no longer cloudy. Each column was flushed using a peristaltic pump for 10 pore volumes (PVs) of NGW at a flow rate of 1.36 mL/min to establish a steady-state flow. Held vertically, the bottom end of the column was used as the influent entrance (Fig. 1). Plug flow condition was verified by conducting tracer tests using 3 pore volumes of a 0.1 M NaCl solution. Each column was then flushed with another 10 pore volumes of NGW. A solution containing w109 PFU/mL of MS2 was introduced and samples were taken in 1.5 mL centrifuge tubes until breakthrough was well-established. Samples were taken from the control column every 5 min for 7 pore volumes, while samples were taken from the iron column every 15 min for 20 pore volumes. The effluent concentrations of MS2 were determined by plaque assay.
2.3.2.
Non-continuous flow test for MS2
To simulate the normal operation of a BSF, 4 additional columns (with the same dimensions and packing procedure as described in Section 2.3.1) were charged daily with 1 pore volume NGW containing w108 PFU/mL of MS2. Primary effluent (PE) obtained from the local wastewater treatment plant, was also added to the solution (2.5% PE) to stimulate biofilm growth at the influent sand surface, located at the
Fig. 1 e (Left) Breakthrough curves of NaCl tracer from columns packed with sand and zerovalent iron. (Right) Diagram of glass columns used for all small-scale experiments. For non-continuous tests, the column was inverted after each daily charge and the influent exposed to air to allow biofilm growth. The sketch for the column is not made to scale.
4504
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 0 1 e4 5 1 0
bottom of the column during charges. The columns were stored between charges in an inverted position, so that the influent entrance was located on top and exposed to air (Fig. 1). Three of the columns were packed with different orientations of iron particles (10% iron by total volume): an even mixture throughout the sand, a middle layer, and a layer at the influent. One column was packed with only sand.
2.3.3.
Non-continuous flow test for rotavirus
Two different columns (sand only and 10% iron by total volume) were then flushed with aquifer water according to the procedures described in Section 2.3.2. Columns were charged with NGW and 2.5% PE for 6 weeks to establish a biofilm. Two sets of experiments were conducted. For the first set, the influent water was Newmark groundwater with 2.35 mg/L TOC. About 30 days after the first set of experiments, the second set of experiments was conducted, for which the influent water was effluent from a local wastewater treatment plant. The treated wastewater effluent was not disinfected and had a TOC of 20 mg/L. For each set of experiments, influent water was seeded with rotavirus (w104 FFU/mL or w105 FFU/mL) and each column was charged with 1 pore volume of solution. After 24 h, effluent samples were collected and determined by rotavirus infectivity tests.
2.4.
Biosand filter experiments
2.4.1.
Filter and media preparation
Two 60 L capacity plastic BSFs were obtained from HydrAid (Grand Rapids, MI). The filters were packed according to the four layer system established by Manz (2007). Each filter contained 5 cm of under drain gravel (6.25e12.5 mm), 5 cm of medium sized support gravel (3.125e6.25 mm) to separate the drainage layer from the filtration sand, and 40 cm of filter media (effective size 0.4 mm) with a 5 cm top layer of fine sand (effective size 0.15 mm). Sand was manually sieved using the appropriate meshes. The effective sand size was similar to that used previously in research and practice (CAWST, 2010; Elliott et al., 2008). One filter was amended with iron by adding 5.54 kg of a mild steel nail (40 mm length, 2 mm diameter, “bright” finish) mixed evenly throughout the top 20 cm of the filter media, excluding the layer of fine sand. The initial maximum flow rates for the unmodified plastic BSF and iron-amended plastic BSF following the first 20 L charges were 0.64 L/min (0.70 m/h) and 0.93 L/min (1.01 m/h), respectively. Every 24 h each filter was charged with 20 L of a solution containing w108 PFU/mL MS2 and 2.5% PE to establish biofilm growth. PE was not added after day 20, when biofilm activity was well-established. Three concrete BSFs were constructed using a steel mold built to specifications (McCarroll, 2009) provided by the Centre for Affordable Water and Sanitation Technology (CAWST) and were representative of BSFs in use worldwide (CAWST, 2010, 2011). Each concrete filter was packed using the same specifications provided for the plastic filter. One filter was amended with iron by adding 5.54 kg of zerovalent iron particles (ETI8/ 50, same as used in column experiments) mixed evenly throughout the top 20 cm of the filter media, excluding the layer of fine sand. Another filter was packed with extra fine
steel wool (Red Devil, Inc. #0000), which was mixed evenly throughout the top 20 cm of filter media. Due to the steel wool’s low weight and large volume, only 0.26 kg of steel wool was used. The initial maximum flow rates for the unmodified concrete BSF, iron-particle-amended concrete BSF, and steelwool-amended BSF following the first 20 L charges were 0.64 mL/min (0.70 m/h), 0.59 mL/min (0.64 m/h), 0.50 mL/min (0.54 m/h), respectively. Every 24 h each filter was charged with 20 L of a solution containing w107 PFU/mL MS2 and 2.5% PE to establish biofilm growth. PE was not added after day 20, when biofilm activity was well-established. X-ray diffraction (XRD) experiments were performed on rusted iron particle samples to determine the mineral composition of the rust. A Rigaku D/Max-b diffractometer with a copper X-ray source controlled by MDI’s DataScan was used. The following parameters were used: 45 kV and 20 mA, the scanning angle range (2q) was 15e80 , and the scanning rate was 0.6 /min with a step increment of 0.05.
2.4.2.
Water chemistry
Influent and effluent samples were collected in 15 mL tubes for analysis of pH, dissolved oxygen (DO) content, alkalinity, NO3, NH4þ, Cl. Samples for MS2 were taken in 1.5 mL sterilized centrifuge tubes and held in a 4 C refrigerator until analysis within 24 h. Trace metal concentrations were determined using inductively coupled plasma mass spectrometry (ICPeMS) with an ELAN Dynamic Reaction Cell instrument (PerkinElmer, Norwalk, USA). Samples were diluted to a total dissolved solid concentration of 0.25%, and the light wavelength intensity from excited atom species was used to determine analyte concentrations. Turbidity of the groundwater and the final effluent was 0.25 and 0.7 NTU as measured by a Hach Turbidity Meter 43900.
3.
Results
3.1. Tracer tests for column experiments with noncontinuous flow Four NaCl breakthrough curves through sand and iron columns with varying orientations of zerovalent iron are plotted in Fig. 1. The three zerovalent iron orientations (10% iron by volume) were an even mixture throughout the sand (‘mixed’), a middle layer (‘band’), and a layer at the influent (‘top’). Independent estimates of the pore volumes (sand only/top placement: 50.0 0.1 mL, mixed/band: 54.9 0.1 mL) were found by measuring the water volume necessary to fill each column completely. All columns had a resulting porosity of 34%. The NaCl tracer response suggests uniform plug flow with a sharp incline in conductivity after one pore volume and a sharp decline after a negative step input was introduced. The Morrill Dispersion Index (MDI) was calculated for all four tracer tests using the method defined by Tchobanoglous et al. (2003) to confirm plug flow. The MDI for each test was approximately 1.4. The USEPA classifies flow with an MDI of less than 2.0 as effective plug flow, while an ideal plug flow reactor has an MDI of 1.0 (USEPA, 1986).
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 0 1 e4 5 1 0
3.2.
Column experiments
3.2.1.
Short term removal using continuous flow (MS2)
The log10 reductions obtained by continuous flow through clean quartz sand with no iron particles and sand mixed with iron (10% iron by volume) are shown in Fig. 2. For the sand with no iron particles, the breakthrough concentration of MS2 was achieved after one pore volume, with approximately half of the MS2 being retained (49.9%). MS2 concentrations in the effluent of the column of sand mixed with iron particles were under the limit of quantification (20 PFU/plate of undiluted sample) until 6 pore volumes continuous flow had passed, with the breakthrough concentration around 5log10 removal (99.999%). In addition, overnight (i.e. 18-h) batch experiments using 15 mL tubes containing 10 mL of solution and 5 mL of iron particles or rusted iron nails revealed complete removal of 4 106 PFU/mL of MS2. Thus, MS2 removal was due to adsorption onto iron oxides. Sorption of MS2 in the iron column most likely occurred due to electrostatic interactions between positively charged iron oxides formed during iron corrosion (e.g. hematite and magnetite) and negatively charged virion particles. A number of studies have also reported the adsorption and inactivation of viruses by iron oxides (Gutierrez et al., 2009; Moore et al., 1981; Rao et al., 1968; Sagripanti et al., 1993). You et al. found complete breakthrough of MS2 by sand columns in a similar study (You et al., 2005). Although 49.9% removal is higher than observed in previous studies, overall removal by the quartz sand was limited. Differences in sand composition may have led to the increased reduction. The quartz sand in this study was cleaned and dried, but it was not treated with a citrate solution to remove existing metal ions and oxides. Trace levels of metal oxides may have been present, leading to heightened MS2 removal. You et al. used a solution containing sodium citrate and citric acid to decrease iron levels in the sand columns from 32.5 mg iron/kg sand to below the detection limit (0.02 mg iron/kg sand) before testing (You et al., 2005). The higher reduction may also be due to the pH of the Newmark groundwater (pH 6.2), which is lower than that of artificial groundwater (AGW; pH 7.5) used in other experiments. The isoelectric point (IEP), the pH at which the surface charge of the virus is neutral or zero, is 3.5e3.9 for MS2
Fig. 2 e Log10 reduction of bacteriophage MS2 in groundwater through clean quartz sand and iron particles (10% by volume) mixed evenly with quartz sand.
4505
(Gutierrez et al., 2009; Overby et al., 1966; Penrod et al., 1995; Zerda and Gerba, 1984). As the pH of the aqueous solution decreases and approaches the IEP, the net negative charge of the MS2 decreases, resulting in the reduction of electrostatic repulsive interactions between the negatively charged virion particles and the negatively charged sand particles (Logsdon et al., 2002). A lower pH in the source water would result in less repulsion and, therefore, greater virus removal by filtration.
3.2.2. Long term removal of MS2 in columns with biofilm growth The long term (72 day) log10 removal for four glass columns simulating a daily, 1 pore volume charged BSF is shown in Fig. 3. While the sand column averaged only 1log10 (90%) removal, all three iron columns had greater than 5log10 removal for the duration of the experiment. Reduction of MS2 by the iron columns ranged from 5log10 to complete removal (>8log10). In comparison, the virus reduction by the sand column ranged from no removal to 3log10 (99.9%). After about 1 week, biofilm growth could be seen in all 4 columns, but removal rates did not increase as the biofilm developed. Biofilm formation occurred only at the influent entrance, which was tapered down significantly from the maximum diameter of the column. Since biofilm formed only on a relatively small area, it made minimal contribution to virus removal. In contrast, Elliot et al. found that the reduction of bacteriophages MS2 and PRD-1 in BSFs increased from 0 to 1.3log10 as the biofilm developed (Elliott et al., 2008). Previous studies have also shown that bacteriophage adsorption (MS2) is reduced in the presence of natural organic matter (NOM) and phosphates that compete for and fill iron oxide adsorption sites (Chiew et al., 2009). Influent and effluent total organic carbon levels for all four columns were measured as an indicator of NOM. The influent concentration of TOC for each column was approximately 3.1 mg/L. Although each iron column retained more TOC (0.5e0.3 mg/L) than the sand column (0.1 mg/L), virus removal was not adversely affected. The extended contact time (24 h) most likely contributed to the increased removal, combating any negative effects caused by the introduction of NOM. It was
Fig. 3 e Log10 reductions of MS2 by three iron columns (different orientations) and one sand-only column.
4506
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 0 1 e4 5 1 0
expected that the iron column with a layer of iron particles at the influent entrance (the “top” column) would experience reduced adsorption because the absence of sand media and established biofilm above it would lead to increased exposure to NOM. However, as discussed previously, relatively small biofilm development occurred. Therefore, no significant differences in removal rates between columns of different iron particle orientations were observed.
3.2.3.
Removal of rotavirus in columns with biofilm growth
For the first set of rotavirus experiments (described in Section 2.3.3), the iron column was charged on two separate days with w104 FFU/mL of rotavirus and obtained removal below the detection limit on each daily charge. The second set of experiments run on the same iron and sand columns used treated wastewater effluent with 10 times higher TOC to evaluate the effect of NOM on virus removal. With the treated wastewater effluent, we obtained 5.2log10 removal of rotavirus by the iron column and 1.1log10 removal by the sand-only column. Note that even with higher TOC, the iron column still allowed substantial rotavirus removal. The rotavirus results are consistent with the MS2 experiments performed in these column studies and rotavirus experiments performed by Gutierrez et al. (2009). In a study using flow-through columns packed with hematite coated glass fibers, it was found that 3 104 FFU of rotavirus was removed per gram of hematite nanoparticles. Furthermore, Gutierrez et al. (2009) found that only 2% of adsorbed rotavirus remained infectious after attachment. Previous research suggests that electrostatic interactions between viruses and iron oxides are so strong that they can cause viruses to disintegrate (Gutierrez et al., 2009; Ryan et al., 2002). With an IEP of 4.5 (Gutierrez et al., 2009), the highly negative potential of rotavirus in source water leads to favorable conditions for viruses to adsorb and be rendered non-infectious.
3.3.
Biosand filter experiments
3.3.1.
Head-loss development over time
As the BSF accumulates bacteria and organic particulate matter to develop a biofilm, the head-loss of the system increases, and users will notice the effluent flow rate decreasing over time as the BSF is used repeatedly. The buildup of organic matter and development of the biological layers (known as “ripening”) results in a more effective filter (Elliott et al., 2008; Stauber, 2006). The flow rates for the two plastic filters are plotted in Fig. 4. To promote biofilm growth, 2.5% primary effluent (PE) was added to both filters until day 20. The BSF without iron nails behaved as expected: as time progressed and the filter ripened, the flow rate decreased and removal efficacy increased. The iron nail filter, however, behaved differently. The flow rate did not decrease as expected, indicating a problem in establishing uniform flow across the entire cross-sectional area of the filter. Although the filter with iron nails distributed evenly throughout the sand established uniform flow early in the experiment, independent flow paths eventually developed. Water bypassed the iron nails at an elevated flow rate, despite the continued growth of biofilm.
Fig. 4 e Flow rates for sand BSF (plastic), iron nail BSF (plastic), steel wool BSF (concrete), and iron particle BSF (concrete).
Fig. 4 also shows the flow rate of the concrete filter packed with iron particles compared to the flow rate of the plastic BSF packed with iron nails. Flow rate problems were effectively eliminated by using an iron material closer in size to the filtration sand media grain size, and the flow rate for the iron particle BSF behaved as desired. As organic matter accumulated, the filter ripened and the flow rate decreased. The steel wool filter displayed similar results.
3.3.2.
Water characteristics
Analytical parameters including pH, DO, alkalinity, and NO3 remained consistent throughout each BSF experiment. Influent and effluent pH was between 7.1 and 8.2 with an average alkalinity of 22.5 mg/L as CaCO3. DO was reduced in the filters but always remained above 1.01 mg/L. Iron oxides formed during iron corrosion were filtered out by the sand media, and iron was not present in the effluent (ICPeMS detection limit of 0.0059 mg/L). Leaching of other trace metals from the iron sources was not seen, and heavy metals such as chromium, lead, cadmium, and arsenic were not present in the effluent.
3.3.3.
MS2 removal
Results for the removal of MS2 through the sand/iron nail BSF are shown in Fig. 5. The iron nail filter started with a 6.5log10 (99.99997%) removal, but adsorption of virus particles quickly declined as flow paths short-circuited around the iron nails. Subsequent filter charges did not achieve plug flow, causing
Fig. 5 e MS2 reduction in the BSF containing iron nails.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 0 1 e4 5 1 0
reduced contact time with iron particles. After 10 days, removal for the iron nail BSF was less than 4log10 (99.99%), and successive samples were between 3 and 4log10 (99.9e99.99%). These results indicate that the iron nails used are not appropriate for use in the BSF; smaller iron material must be used to obtain a more evenly graded media that will promote uniform flow and not detract from the original (without iron) efficacy of the BSF. The flow paths created also highlight the importance of contact time, a point further illustrated in Fig. 6. Water that has been allowed to sit overnight in the filter (24 h) achieves additional removal of w1log10 (90%). The concrete sand filter achieved between 4log10 removal and no removal, with an average of about 2log10 (99%) removal (Fig. 7) for the first 150 days. In comparison, the two filters packed with iron (particles and steel wool) achieved between 7log10 (99.99999%) and >5log10 removal, with an average of 6log10 (99.9999%) reduction of virion particles for the first 170 days of filter use. The iron particles and steel wool resulted in filters with a more even media grade than in the filters containing nails. This promoted uniform, steady flow throughout the entire cross-sectional area of the filter, allowing sufficient contact time between each filter charge and the iron media. The corresponding increase in MS2 removal shows that ironamended BSFs can effectively remove select viruses, with removal rates that exceed U.S. Environmental Protection Agency (USEPA) drinking standards (4log10 removal) (USEPA, 1991). Iron samples from the iron particle BSF were taken for XRD analysis after the study ended. Iron particle composition after corrosion was 48% iron, 40% magnetite, 9% hematite, and 3% other. With 48% of the particles still present in the form of iron after 376 days of operation, further iron oxide generation was possible. However, removal in the steel wool filter decreased significantly after day 170, dropping to an average of 2log10 removal. XRD analysis revealed that the steel wool present in the steel wool BSF was completely oxidized. This indicates that virus sorption to the steel wool stopped once its adsorption sites were exhausted, and remaining removal of viruses was due to the sand media. The reduction in removal capacity to levels below those observed in the sand-only filter may indicate that the addition of steel wool to the sand media adversely affected the traditional filter’s removal efficacy. The sand-only BSF saw a steady increase in removal from 2log10 to >4log10. Previous research has shown that biofilm ripening and media aging contribute significantly to the MS2
Fig. 6 e Comparison of MS2 reduction at two different pause periods for the BSF containing iron nails.
4507
Fig. 7 e Long term MS2 reduction by three different concrete filters: sand, steel wool, and iron particles.
removal capacity of the traditional sand BSF (Elliott et al., 2008), but long term studies (>2 months) have not been previously conducted. However, it is well-known that filter ripening plays a crucial role in particle removal during granular filtration (Kim et al., 2008; Kretzschmar et al., 1997; Tobiason and Omelia, 1988). The results from this study suggest that traditional BSFs may experience significant virus removal (>2log10) once filters have been in use for several months. If this is the case, filters would only need to be amended with an appropriate iron source, preferably zerovalent iron particles, for the initial ripening period.
4.
Discussion
Previous testing in small-scale columns has shown that MS2 has minimal sorption to sand media. In particular, studies have demonstrated lower adsorption of MS2 when compared to bacteriophages ( phiX-174) and human pathogenic viruses (rotavirus, echovirus-12, and poliovirus). Data from smallscale saturated flow studies showed that MS2 had no sorption compared to phiX-174 with about 80% removal (Jin et al., 1997). Higher removal of phiX-174 was attributed to a higher isoelectric point (IEP ¼ 6.6) than that of MS2 (IEP ¼ 3.5e3.9). Virus sorption is largely governed by electrostatic interactions and van der Waal’s forces. Because quartz sand is negatively charged in source water (IEP ¼ w2.2), and MS2 is more negatively charged than phiX-174, MS2 experiences higher repulsion at pH ¼ 7. Bales et al. (1991, 1993) also demonstrated that phiX-174 and poliovirus exhibit similar removal through sand columns due to their shared IEP of 6.6. Additionally, Bales et al. (1993) showed that poliovirus sorbed to silica beads much more readily than MS2. Small-scale studies with sand columns have also shown higher adsorption of rotavirus (IEP ¼ 4.5) than MS2 (Gutierrez et al., 2009). It is also important to note that hydrophobic interactions may be important for bacteriophage adsorption to hydrophobic surface due to the presence of hydrophobic areas on the surface of bacteriophage such as MS2 and PRD-1 (Bales et al., 1991). In large-scale studies using BSFs similar to those used in this study, Elliott et al. (2008) saw greater removal of echovirus-12 (>2log) than bacteriophages MS2 and PRD-1
4508
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 0 1 e4 5 1 0
(w1log). These differences were also attributed to echovirus12 having a higher IEP (echoviruses have IEP of 5e6.4) than MS2 and PRD-1, which have similar IEPs of 3.5e3.9 and 4.2, respectively (Michen and Graule, 2010). When positively charged iron oxides are added throughout the sand media, it is expected that all viruses that are negatively charged at pH ¼ 7 will be able to readily adsorb to iron oxide surfaces. At the same time, MS2 experiences high repulsion from the sand particles due to its low isoelectric point, and based on past studies, is a more difficult virus, compared to echovirus type 12, to remove through conventional sand filtration (Elliott et al., 2008). Having a low IEP would make a virus relatively negative at pHs near neutral. This charge leads to less removal in a sand-only BSF, due to the repulsion of negatively charged sand particles, and greater removal in an iron-amended BSF, due to stronger electrostatic attraction to positively charged iron oxides. While MS2 bacteriophage can be used as a surrogate for rotavirus because these two virus have similar interfacial properties (Brady-Estevez et al., 2010; Gutierrez et al., 2009, 2010; Mylon et al., 2010), other enteric viruses may require different surrogates. When a virus attaches to particles in water, the overall size of the aggregate is greater than that of a monodispersed virus. As a result, virus-particle aggregates would be easier to remove by filtration (Kretzschmar et al., 1999; Petosa et al., 2010). Experiments presented here were conducted as a worst case scenario for monodispersed viruses. In addition, a recent publication showed that norovirus, which has a similar size as the studied virus MS2, associated with 0.45e180 mm particles, and attachment and settling was not an effective removal mechanism for viruses in waste stabilization ponds (Da Silva et al., 2008). Similarly, recent study with bacteriophage MS2 and rotavirus has found that steric repulsion prevents viruses from aggregating and adsorbing to silica and organic matter surfaces (Gutierrez et al., 2010; Mylon et al., 2010). It is expected that biofilm development is highly dependent on source water characteristics, including concentration and identity of native microorganisms that attach to the sand, nutrients from which the microbes derive an energy source, NOM that could shelter or cover the microbes, DO levels, and even water temperature. In our household-scale studies, MS2 removal in the sand-only BSF increased with biofilm development in a trend that matches previous studies (Elliott et al., 2008), despite the utilization of different water sources. While investigating biofilms developed from different water sources was not a focus during this phase of the research, the currently on-going research involves collecting data from BSFs installed and operating at different locations in Guatemala.
5.
Conclusions
The following conclusions resulting from this study indicate a potential advancement in household water treatment technologies by amending the BSF with iron materials: Both MS2 and rotavirus were treated to USEPA standards for virus removal, greater than 99.99% removal, through the
adsorption to positively charged iron oxides in small-scale studies. Untreated iron material distributed uniformly in sand media of a BSF will oxidize and effectively remove >4log10 MS2 and rotavirus from a natural source water. The duration of effective virus removal by iron-amended biosand filtration depends on both source water conditions and quantity and composition of iron material added, all of which need to be researched further. After 200 days in operation, the unmodified BSF underwent significant ripening and was able to provide more than 4log10 of MS2 virus removal. Further research is needed to determine the effects of competitive adsorption with other water constituents on the efficacy of virus removal by iron-amended biosand filtration. For relatively little cost (approximately 4 USD per filter based on initial cost estimates), BSFs can be amended with local iron materials, thereby providing a substantially improved barrier against waterborne viruses and, hopefully, bringing relief to millions of lives in the process.
Acknowledgments This work was partially supported by the Center of Advanced Materials for the Purification of Water with Systems (WaterCAMPWS), a Science and Technology Center under the National Science Foundation (NSF) Award No. CTS-0120978, under the United States Environmental Protection Agency (USEPA) People, Prosperity, and the Planet (P3) Phase 1 (SU834296) and Phase 2 (SU834754) grants, the NSF Career grant (0954501) to THN, and the University of Illinois College of Engineering International Programs in Engineering (IPENG). Leo Gutierrez and Ofelia Romero were acknowledged for preparing MS2 and rotavirus stocks and assays. BSFs were run daily by IB and AS with the assistance of Kevin Swanson and other members of the Engineers Without Borders (EWB) University of Illinois at Urbana-Champaign Chapter.
references
Bales, R.C., Hinkle, S.R., Kroeger, T.W., Stocking, K., Gerba, C. P., 1991. Bacteriophage adsorption during transport through porous-media e chemical perturbations and reversibility. Environmental Science & Technology 25 (12), 2088e2095. Bales, R.C., Li, S.M., Maguire, K.M., Yahya, M.T., Gerba, C.P., 1993. MS-2 and poliovirus transport in porous-media e hydrophobic effects and chemical perturbations. Water Resources Research 29 (4), 957e963. Baumgartner, J., Murcott, S., Ezzati, M., 2007. Reconsidering ‘appropriate technology’: the effects of operating conditions on the bacterial removal performance of two household drinking-water filter systems. Environmental Research Letters 2 (2), 6. Brady-Estevez, A.S., Nguyen, T.H., Gutierrez, L., Elimelech, M., 2010. Impact of solution chemistry on viral removal by a single-walled carbon nanotube filter. Water Research 44 (13), 3773e3780.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 0 1 e4 5 1 0
Buzunis, B.J., 1995. Intermittently Operated Slow Sand Filtration: A New Water Treatment Process. University of Calgary, Calgary. CAWST, 2010. Biosand Filter Manual: Design, Construction, Installation, Operation, and Maintenance. Centre for Affordable Water and Sanitation Technology, Calgary, Alberta, Canada, p. 59. CAWST, 2011. About Us. http://www.cawst.org/en/about-us. Chiew, H., Sampson, M.L., Huch, S., Ken, S., Bostick, B.C., 2009. Effect of groundwater iron and phosphate on the efficacy of arsenic removal by iron-amended bios and filters. Environmental Science & Technology 43 (16), 6295e6300. Clasen, T., Schmidt, W.P., Rabie, T., Roberts, I., Cairncross, S., 2007. Interventions to improve water quality for preventing diarrhoea: systematic review and meta-analysis. British Medical Journal 334, 7597e7782. Curtis, V., Cairncross, S., Yonli, R., 2000. Review: domestic hygiene and diarrhoea e pinpointing the problem. Tropical Medicine & International Health 5 (1), 22e32. Da Silva, A.K., Le Guyader, F.S., Le Saux, J.C., Pommepuy, M., Montgomery, M.A., Elimelech, M., 2008. Norovirus removal and particle association in a waste stabilization pond. Environmental Science & Technology 42 (24), 9151e9157. Elliott, M.A., 2010. Mechanisms of Microbial Reduction and Implications for Design and Operation of the Biosand Water Filter PhD thesis. University of North Carolina at Chapel Hill, Chapel Hill. Elliott, M.A., Stauber, C.E., Koksal, F., DiGiano, F.A., Sobsey, M.D., 2008. Reductions of E. coli, echovirus type 12 and bacteriophages in an intermittently operated household-scale slow sand filter. Water Research 42 (10e11), 2662e2670. Esrey, S.A., Feachem, R.G., Hughes, J.M., 1985. Interventions for the control of diarrhoeal diseases among young children: improving water supplies and excreta disposal facilities. Bulletin of the World Health Organization 63 (4), 757e772. Esrey, S.A., Potash, J.B., Roberts, L., Shiff, C., 1991. Effects of improved water supply and sanitation on ascariasis, diarrhoea, dracunculiasis, hookworm infection, schistosomiasis, and trachoma. Bulletin of the World Health Organization 69 (5), 609e621. Fewtrell, L., Kaufmann, R.B., Kay, D., Enanoria, W., Haller, L., Colford Jr., J.M., 2005. Water, sanitation, and hygiene interventions to reduce diarrhoea in less developed countries: a systematic review and meta-analysis. Lancet Infectious Diseases 5 (1), 42e52. Gutierrez, L., Li, X., Wang, J.W., Nangmenyi, G., Economy, J., Kuhlenschmidt, T.B., Kuhlenschmidt, M.S., Nguyen, T.H., 2009. Adsorption of rotavirus and bacteriophage MS2 using glass fiber coated with hematite nanoparticles. Water Research 43 (20), 5198e5208. Gutierrez, L., Mylon, S.E., Nash, B., Nguyen, T.H., 2010. Deposition and aggregation kinetics of rotavirus in divalent cation solutions. Environmental Science & Technology 44 (12), 4552e4557. Jin, Y., Yates, M.V., Thompson, S.S., Jury, W.A., 1997. Sorption of viruses during flow through saturated sand columns. Environmental Science & Technology, 548e555. Jin, Y., Chu, Y., Li, Y., 2000. Virus removal and transport in saturated and unsaturated sand columns. Journal of Contaminant Hydrology 43 (2), 111e128. Kim, J., Nason, J.A., Lawler, D.F., 2008. Influence of surface charge distributions and particle size distributions on particle attachment in granular media filtration. Environmental Science & Technology 42 (7), 2557e2562. Kitis, M., Lozier, J.C., Kim, J.H., Mi, B.X., Marinas, B.J., 2003. Microbial removal and integrity monitoring of RO and NF membranes. Journal of American Water Works Association 95 (12), 105e119. Kretzschmar, R., Barmettler, K., Grolimund, D., Yan, Y.D., Borkovec, M., Sticher, H., 1997. Experimental determination of
4509
colloid deposition rates and collision efficiencies in natural porous media. Water Resources Research 33 (5), 1129e1137. Kretzschmar, R., Borkovec, M., Grolimund, D., Elimelech, M., 1999. Advances in Agronomy, vol. 66. Academic Press Inc, San Diego, pp. 121e193. Li, Q., Snoeyink, V., Marinas, B., Campos, C., 2003. Elucidating competitive adsorption mechanisms of atrazine and NOM using model compounds. Water Research 37, 773e784. Logsdon, G.S., Kohne, R., Abel, S., LaBonde, S., 2002. Slow sand filtration for small water systems. Journal of Environmental Engineering & Science 1 (5), 339e348. Lukasik, J., Cheng, Y.F., Lu, F.H., Tamplin, M., Farrah, S.R., 1999. Removal of microorganisms from water by columns containing sand coated with ferric and aluminum hydroxides. Water Research 33 (3), 769e777. Manz, D.D.H., 2007. Preparation of Media for the BioSand Water Filter: Four Layer System. Manz, D.H., Buzunis, B., Morales, C., 1993. Nicaragua Household Water Supply and Testing Project. University of Calgary, Calgary, p. 10. McCarroll, R. (Ed.), 2009. The All Natural Biosand Filter Manual: Design, Construction, and Installation. The Centre for Affordable Water and Sanitation Technologies, Calgary, Alberta, Canada. Michen, B., Graule, T., 2010. Isoelectric points of viruses. Journal of Applied Microbiology 109 (2), 388e397. Moore, E.S., Tylor, D.H., Sturman, L.S., Reddy, M.M., 1981. Poliovirus adsorption by 34 minerals and soils. Applied and Environmental Microbiology 42, 963e975. Mylon, S.E., Rinciog, C.I., Schmidt, N., Gutierrez, L., Wong, G.C.L., Nguyen, T.H., 2010. Influence of salts and natural organic matter on the stability of bacteriophage MS2. Langmuir 26 (2), 1035e1042. Nelson, K.L., Murray, A., 2008. Sanitation for unserved populations: technologies, implementation challenges, and opportunities. Annual Review of Environment and Resources 33, 119e151. Ngai, T.K.K., Shrestha, R.R., Dangol, B., Maharjan, M., Murcott, S.E., 2007. Design for sustainable development e household drinking water filter for arsenic and pathogen treatment in Nepal. Journal of Environmental Science and Health e Part A Toxic/Hazardous Substances and Environmental Engineering 42 (12), 1879e1888. Overby, L.R., Barlow, G.H., Doi, R.H., Jacob, M., Spiegel, S., 1966. Comparison of two serologically distinct ribonucleic acid bacteriophages. Journal of Bacteriology 91 (1), 442e448. Page, M.A., Shisler, J.L., Marinas, B.J., 2009. Kinetics of adenovirus type 2 inactivation with free chlorine. Water Research 43 (11), 2916e2926. Palmateer, G., Manz, D., Jurkovic, A., McInnis, R., Unger, S., Kwan, K.K., Dutka, B.J., 1999. Toxicant and parasite challenge of Manz intermittent slow sand filter. Environmental Toxicology 14 (2), 217e225. Parashar, U.D., Gibson, C.J., Bresee, J.S., Glass, R.I., 2006. Rotavirus and severe childhood diarrhea. Emerging Infectious Diseases 12 (2), 304e306. Penrod, S.L., Olson, T.M., Grant, S.B., 1995. Whole particle microelectrophoresis for small viruses. Journal of Colloid and Interface Science 173 (2), 521e523. Petosa, A.R., Jaisi, D.P., Quevedo, I.R., Elimelech, M., Tufenkji, N., 2010. Aggregation and deposition of engineered nanomaterials in aquatic environments: role of physicochemical interactions. Environmental Science & Technology 44 (17), 6532e6549. Ramani, S., Kang, G., 2009. Viruses causing childhood diarrhoea in the developing world. Current Opinion in Infectious Diseases 22 (5), 477e482.
4510
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 0 1 e4 5 1 0
Rao, V.C., Sullivan, R., Read, R.B., Clarke, N.A., 1968. A simple method for concentrating and detecting viruses in water. Journal of American Water Works Association 60, 1288e1294. Rolsma, M.D., Gelberg, H.B., Kuhlenschmidt, M.S., 1994. Assay for evaluation of rotavirusecell interactions e identification of an enterocyte ganglioside fraction that mediates group-A porcine rotavirus recognition. Journal of Virology 68 (1), 258e268. Rolsma, M.D., Kuhlenschmidt, T.B., Gelberg, H.B., Kuhlenschmidt, M.S., 1998. Structure and function of a ganglioside receptor for porcine rotavirus. Journal of Virology 72 (11), 9079e9091. Ruiz, M., Charpilienne, A., Liprandi, F., Gajardo, R., Michelangeli, F., Cohen, J., 1996. The concentration of Ca21 that solubilizes outer capsid proteins from rotavirus particles is dependent on the strain. Journal of Virology 70 (8), 4877e4883. Ryan, J.N., Elimelech, M., Ard, R.A., Harvey, R.W., Johnson, P.R., 1999. Bacteriophage PRD1 and silica colloid transport and recovery in an iron oxide-coated sand aquifer. Environmental Science & Technology 33 (1), 63e73. Ryan, J.R.H., Metge, D., Elimelech, M., Pieper, A., 2002. Field and laboratory investigations of inactivation of viruses (PRD1 and MS2) attached to iron oxide-coated quartz sand. Environmental Science & Technology 36, 2403e2413. Sagripanti, J.L., Routson, L.B., Lytle, C.D., 1993. Virus inactivation by copper or iron ions alone and in the presence of peroxide. Applied and Environmental Microbiology 59, 4374e4376. Sirikanchana, K., Shisler, J.L., Marinas, B.J., 2008. Effect of exposure to UV-C irradiation and monochloramine on adenovirus serotype 2 early protein expression and DNA replication. Applied and Environmental Microbiology 74 (12), 3774e3782. Sobsey, M.D., 2002. Managing Water in the Home: Accelerated Health Gains from Improved Water Supply, Geneva. Sobsey, M.D., Stauber, C.E., Casanova, L.M., Brown, J.M., Elliott, M. A., 2008. Point of use household drinking water filtration: a practical, effective solution for providing sustained access to
safe drinking water in the developing world. Environmental Science & Technology 42 (12), 4261e4267. Stauber, C.E., 2006. Characterisation of the biosand filter for E. coli reductions from household drinking water under controlled laboratory and field use conditions. Water Science and Technology 54 (3), 1e7. Tchobanoglous, G., Burton, F.L., Stensel, H.D., 2003. Wastewater Engineering: Treatment and Reuse. McGraw-Hill Higher Education, New York. Tiwari, S.S.K., Schmidt, W.P., Darby, J., Kariuki, Z.G., Jenkins, M. W., 2009. Intermittent slow sand filtration for preventing diarrhoea among children in Kenyan households using unimproved water sources: randomized controlled trial. Tropical Medicine and International Health 14 (11), 1374e1382. Tobiason, J.E., Omelia, C.R., 1988. Physicochemical aspects of particle removal in depth filtration. Journal of American Water Works Association 80 (12), 54e64. UNICEF/WHO, 2008. Progress on Drinking Water and Sanitation: Special Focus on Sanitation. USEPA, 1986. Design Manual, Municipal Wastewater Disinfection, Cincinnati, OH. USEPA, 1991. Guidance Manual for Compliance with the Filtration and Disinfection Requirements for Public Water Systems Using Surface Water Sources. Prepared for the Office of Drinking Water, Science and Technology Branch, Criteria and Standards Division, Washington, DC. Weber-Shirk, M.L., Dick, R.D., 1997. Biological mechanisms in slow sand filters. Journal of American Water Works Association 89 (2), 72e83. WHO, 2008. Safer Water, Better Health: Costs, Benefits, and Sustainability of Interventions to Protect and Promote Health. You, Y., Han, J., Chiu, P.C., Jin, Y., 2005. Removal and inactivation of waterborne viruses using zerovalent iron. Environmental Science & Technology 39 (23), 9263. Zerda, K.S., Gerba, C.P., 1984. Agarose isoelectrofocusing of intact virions. Journal of Virological Methods 9 (1), 1e6.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 1 1 e4 5 2 1
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
A new dynamic model for bioavailability and cometabolism of micropollutants during anaerobic digestion Liliana Delgadillo-Mirquez*, Laurent Lardon, Jean-Philippe Steyer, Dominique Patureau INRA, UR050, Laboratoire de Biotechnologie de l’Environnement, Avenue des Etangs, Narbonne F-11100, France
article info
abstract
Article history:
Organic micropollutants (OMPs) are present in wastewater and sludge. Their possible
Received 8 February 2011
impact to the environment contributes to their increasing scientific and social interest.
Received in revised form
Anaerobic digestion has been shown as a potential biological process for removal of these
25 May 2011
compounds. An accurate description of OMP distribution in the environmental system can
Accepted 31 May 2011
be used to better understand which compartment is used for degradation and to improve
Available online 12 June 2011
their depletion in conventional wastewater treatment technologies. In this work, we proposed a dynamical model with a four-compartment distribution to describe the Poly-
Keywords:
cyclic Aromatic Hydrocarbons (PAHs) fate during anaerobic digestion. The model is cali-
Biodegradation
brated and validated using experimental data obtained from two continuous reactors fed
Methanogenic conditions
with primary and secondary sludge operated under mesophilic conditions. A non-linear
PAHs
least square method was used to optimize the model parameters. The resulted model is
Sorption
in accordance with the experimental data. The PAH biodegradation rate is well modeled
Xenobiotic
when considering the aqueous fraction (including free and sorbed to dissolved/colloidal matter PAHs) as the bioavailable compartment. It was also demonstrated in the simulations that the PAHs biodegradation is linked to a mechanism of cometabolism. The model proposed is potentially useful to better understand the micropollutant distribution, predict the fate of PAHs under anaerobic condition and help to optimize the operation process for their depletion. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
Organic micropollutants (OMPs) have become an important environmental topic in recent years due to the risk they pose on aquatic environment and on human health e.g. endocrine disrupting effects (Press-Kristensen et al., 2007; Couillard et al., 2008) and to the development of highly accurate analytical methodologies with lower detection limits (Trably et al., 2004). OMPs are frequently detected in different environmental compartments (rivers, lakes, groundwaters,
sediments, wastewaters, drinking waters) at low concentration (ng to mg/L and mg to mg/kg dry matter). In wastewater treatment plants (WWTPs), OMPs are partially removed by abiotic and biotic processes, including volatilization, stripping, sorption to sludge and biological and/or chemical transformation (Alder et al., 1997; Byrns, 2001; Lindblom et al., 2009). However, the conventional treatment technologies have not been specifically designed for removing OMPs but they can reduce OMPs concentrations as well as their potential environmental impact. Furthermore, these removals are
* Corresponding author. E-mail addresses: [email protected] (L. Delgadillo-Mirquez), [email protected] (L. Lardon), [email protected] (J.-P. Steyer), [email protected] (D. Patureau). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.05.047
4512
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 1 1 e4 5 2 1
dependant of the OMPs physico-chemical properties, the sludge characteristics and the WWTPs operational conditions (Clara et al., 2005; Dionisi et al., 2006; Joss et al., 2006). Several mathematical models described the fate and the distribution of OMPs between the aqueous and the solid phase (Byrns, 2001; Dionisi et al., 2006; Joss et al., 2004, 2006; Lindblom et al., 2009; Plo´sz et al., 2010). In general, these models used the solidewater partition coefficient to describe equilibrium condition and assumed that the aqueous phase is available for microbial biodegradation activity and the solid phase is bioaccessible and can be transferred to aqueous phase during the process (Artola-Garicano et al., 2003). Nevertheless, Barret et al. (2010b) have demonstrated that the sorption phenomena also occur onto the aqueous phase containing dissolved and colloidal matter. In this study, the sludge has been considered as a three-compartment system with two equilibrium constants. The presence of this third compartment can thus influence the distribution and the pollutants bioavailability. In fact, the distribution in three compartments can help to find the real bioavailable fraction of OMPs. Furthermore, sorption onto the solid phase and dissolved/colloidal matter of sludge is a very fast mechanism in comparison with biological anaerobic kinetics (Chang et al., 2003; Dionisi et al., 2006; Barret et al., 2010d). The bioavailability is influenced by a variety of factors including (i) sorption-desorption processes that could be ratelimiting for biodegradation, (ii) irreversibility or sequestration phenomena due to physical and/or chemical interactions and (iii) presence of other compounds that might compete for sorption sites. Moreover, the biodegradation of multiple substrates can also take place the cometabolism process i.e. a compound of interest does not function as a growth substrate (Criddle, 1993). Previous publications have shown the cometabolism as a mechanism approach in the transformation of some recalcitrant contaminants (Chang et al., 1993, 2003; Criddle, 1993; Tiehm and Fritzsche, 1995; Yuan et al., 2001; Clara et al., 2005; Plo´sz et al., 2010; Barret et al., 2010c). This study aimed to propose, analyze and validate a dynamic model for the Polycyclic Aromatic Hydrocarbons (PAHs) fate under anaerobic condition considering sludge as a four-compartment system. To this end, two hypotheses were evaluated. The first hypothesis consists in modeling the PAHs biodegradation kinetics with a cometabolism kinetics and to compare it with a Monod-type kinetics. The second hypothesis tests which one of the compartments is really available for degradation: the free dissolved one, the aqueous one or the sum of all compartments. This approach should improve the prediction of PAHs distribution, bioavailability and biodegradation.
2.
Material and methods
2.1.
Sludge source
All experiments were performed using activated sludge from an urban wastewater treatment plant. The primary sludge sample (PS) was collected at the outlet of a primary settling tank of a domestic wastewater treatment plant treating 33,000 PE (Person Equivalent). The secondary sludge sample (SS) came from the biological aerobic unit of another domestic plant treating 250,000 PE with a very low hydraulic retention time (0.4 day). Prior to their direct use, PS and SS were stored at 20 C. All these samples were finally diluted with deionized water to reach 24 5 gCOD/L and spiked at 5 mg/gDM for each PAH except for indeno(1,2,3,c,d)pyrene (1 mg/gDM). Table 1 shows the main characteristics of these primary and secondary sludge.
2.2.
Micropollutants
Polycyclic Aromatic Hydrocarbons (PAHs) were selected as model micropolluants (Table 2). All solvents were purchased from J.T.Baker. Mixtures are indicated in volume percentage. PAH powders were obtained from Dr Ehrenstorfer GmbH. Each PAH was dissolved in dichloromethane at 1 g/L. The spiking mix was prepared from these individual concentrated solutions, adding 5 mL of each, evaporating solvent under gentle nitrogen flow and dissolving in 50 mL of acetonitrile. Final concentrations were 100 mg/L for each PAH. The 10 mg/L standard solution of PAHs in acetonitrile was provided by Dr Ehrenstorfer GmbH. Dilution factors from 10 to 1000 were applied to obtain 6 calibration levels. Standards were stored at 20 C.
2.3.
Experimental set-up
Two continuous reactors have been operated at a constant organic load of 1.2 0.2 gCOD/L.d and a hydraulic retention time of 20 days. Temperature was regulated at 35 C using hot water circulation in the external jacket. The reactors were fed with primary (PS) and secondary sludge (SS). The feed was stored at 4 C. Six times a day, it was pumped into the reactor just after pumping out the digested sludge, collected in tanks at 4 C. For the start-up, reactors were filled with methanogenic sludge coming from an anaerobic mesophilic reactor adapted to PAHs-polluted sludge, and directly fed at the normal operating conditions. The pH and the biogas volumetric production were monitored on line. Seven days
Table 1 e Primary (PS) and secondary sludge (SS) characteristics (average value and standard deviation from 5 measurements performed at steady state). Sludge
PS SS
COD
DM
DCM
Proteins
Carbohydrates
Lipids
VFA
gO2/L
gDM/L
% of total DM
geqBSA/gDM
geqGlu/gDM
gPEEM/gDM
gVFA/gDM
28 4 23 2
22 2 19 1
51 44
0.27 0.03 0.25 0.02
0.29 0.09 0.30 0.10
0.13 0.05 0.10 0.04
0.03 0.02 0.04 0.04
Source: Barret et al., 2010d. Chemical oxygen demand (COD), dry matter (DM), dissolved/colloidal matter (DCM) and volatile fatty acids (VFAs).
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 1 1 e4 5 2 1
Table 2 e Physicochemical characteristics of the PAHs. Kp (mL/gCODLpart) and KDCM(mL/gCODLDCM), equilibrium constants of sorption determined from a threecompartment methodology (Barret et al., 2010b) for PS and SS. PAH
M (g/mol)
Fluorene Phenanthrene Anthracene Fluoranthene Pyrene Benzo(a)anthracene Chrysene Benzo(b)fluoranthene Benzo(k)fluoranthene Benzo(a)pyrene Dibenzo(a,h)anthracene Benzo(g,h,i)perylene Indeno(1,2,3,c,d)pyrene
166 178 178 202 202 228 228 252 252 252 278 276 276
PS
SS
log Kp
log KDCM
log Kp
log KDCM
0.098 0.398 0.398 0.398 0.698 0.798 0.898 0.898 0.998 0.898 1.198 1.098 0.898
0.481 0.681 0.681 0.781 0.981 1.281 1.281 1.381 1.281 1.281 1.481 1.481 1.681
0.283 0.683 0.583 0.603 0.883 0.983 1.083 1.083 1.183 1.083 1.383 1.283 1.083
0.902 1.102 1.102 1.202 1.302 1.302 1.502 1.602 1.802 1.702 1.902 1.902 1.702
composite samples were taken once a week from the feed tank, the outlet tank and the gaseous phase.
2.4.
Analytical methods
Inlet and outlet composite samples were analyzed for their chemical oxygen demand (COD) in both soluble and particulate fraction, dry matter (DM), organic matter (OM), proteins, carbohydrates, organic carbon in particles (POC) and in dissolved/colloidal matter (DCOC) and volatile fatty acids (VFAs), according to Barret et al. (2010a). The percentage of methane (CH4) and carbon dioxide (CO2) in the biogas were measured using a gas chromatograph (Shimadzu GC-8A), with argon as the carrier gas and equipped with a thermal conductivity detector. PAHs were extracted from the ORBO cartridge using a Soxhlet setup, operated during 16 h at 60 C, with 200 mL of hexane/acetone (50:50 v:v). Extraction from inlet and outlet sludge samples were performed according to Trably et al. (2004). Equilibrium constants (Table 2) of sorption to particles (Kp) and to dissolved/colloidal matter (KDCM) were determined according to the experimental methodology proposed by Barret et al. (2010b) for the thirteen PAHs and the two sludge (PS and SS).
2.5.
4513
distributions of these compounds influence their bioavailability and it may be the main limiting factor of OMP biodegradation in sludge. OMPs are assumed to be able to sorb onto either particles (P) or dissolved/colloidal matter (DCM) (Barret et al., 2010b). In the model, OMPs is thus assumed to be distributed between four physical compartments (Fig. 1): the free dissolved (Cf, mg/L), the gas (Cg, mg/L), the sorbed to DCM (cDCM, mg/gCOD-DCM) and the sorbed to particles (cp, mg/gCOD-p) compartments. At equilibrium, the four-compartment system can be described by the three following equations: Kp ¼
cp Cf
KDCM ¼
(1) cDCM Cf
(2)
and KH ¼
Cg Cf
(3)
where Kp is the equilibrium constant of OMPs sorption to particle (L/gCOD-p), KDCM is the equilibrium constant of OMPs sorption to DCM (L/gCOD-DCM) and Henry constant (KH, dimensionless) describes the equilibrium between gas phase and free dissolved concentration of OMPs. The total mass of OMP can be expressed as total liquid concentration (Ct,liq, mg/L) and gas concentration (Cg, mg/L) of OMP: Ct;liq ¼ Cf þ cp Sp þ cDCM SS
(4)
Cg ¼ KH Cf
(5)
where Sp is the particulate substrate (particulate concentration, gCOD-p/L) and SS is the soluble substrate (dissolved and colloidal concentration, gCOD-DCM/L). Thus, based on the experimental measurement of total liquid concentration and on Equations (1), (2) and (4), the concentrations in the different compartments and initial condition can be estimated from Equations (6)e(8) for each
Simulation software
Simulations presented in this work have been developed in MatLab-Simulink. Optimization toolbox for solving nonlinear least square problems has been used to estimate the model parameters.
3.
Model description and assumptions
3.1.
The four-compartment model
The model describes the physical exchanges of PAHs between compartments in the reactor. Hence, the physical
Fig. 1 e Representation of the four-compartment model of an OMP.
4514
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 1 1 e4 5 2 1
Table 3 e Matrix representation the fate of OMP in the four-compartment model.
a
pollutant. Indeed, the Table 2 shows the equilibrium constants Kp and KDCM calculated for the thirteen PAHs and the two sludge (PS and SS), according to previously developed methodology (Barret et al., 2010b). Cf ¼
Ct;liq 1 þ Kp Sp þ KDCM SS
(6)
Ct;liq Kp cp ¼ 1 þ Kp Sp þ KDCM SS cDCM ¼
(7)
Ct;liq KDCM 1 þ Kp Sp þ KDCM SS
(8)
Kinetics of sorption and desorption of a pollutant between the free dissolved compartment and the particle one and between the free dissolved compartment and the DCM one are described by Equations (9) and (10): part rsorp=desorp
¼ k1
cp
cp ¼ k1 Kp Cf cp
(9)
(10)
where cp and cDCM are the OMPs particle and DCM concentrations at equilibrium, respectively. k1 and k2 are the first-order kinetic constants of sorption to particle and DCM, respectively.
3.2.
khyd
Sp /SS m
SS /X þ CH4 þ CO2 Table 3 represents the mathematical model with seven components and nine processes. The metabolism of substrate growth is incorporated into the three first processes. Hydrolysis is described with a first order kinetics to represent the enzymatic degradation of particulate substrate in soluble substrate. Beside, decay is assumed in the transformation of activated biomass into particulate substrate. Biomass (X, gCOD/ L) growth is linked to soluble substrate uptake and modeled with Monod-type kinetics: m ¼ mmax
rDCM sorp=desorp ¼ k2 cDCM cDCM ¼ k2 KDCM Cf cDCM
gCOD/L) and then soluble substrate (SS, gCOD/L) biodegradation to biogas.
Biodegradation
A two-steps model has been used to describe the anaerobic digestion of sludge: first hydrolysis to particulate matter (Sp,
SS KS þ SS
(11)
where mmax (1/d) is the maximum bacterial growth rate and KS (gCOD/L) is the half-saturation constant associated with the soluble substrate SS. OMP biodegradation may be considered as a classical metabolism with a specific OMP-degrader (Fig. 2b) and modeled with a Monod-type kinetics (Equation (11): mmax,OMP, KS,OMP). Nevertheless, the removal of pollutant present only in trace levels (ng/L or mg/L) could not result in any significant biomass growth (Clara et al., 2005). Thus, we assumed that cometabolism is the main OMP biodegradation mechanism (Fig. 2a). Criddle (1993) proposed a cometabolism model between a growing substrate and a non-growing substrate in simple
4515
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 1 1 e4 5 2 1
b
a
Fig. 2 e Scheme illustrating (a) cometabolism and (b) classic metabolism of OMP.
ecosystem. This equation is based on the assumption that the cometabolic degradation rate is enhanced by the generation of reductants caused by the degradation of growth substrate (SS) and, in its absence, the cometabolic transformation is linked to endogenous decay. Moreover, the cometabolic model includes competitive inhibition between growth and nongrowth substrate and negative effect of the toxic products. However, we did not include those two last terms in this work because of the low concentration of OMPs in the system and a large number of kinetic parameters can complicate the modeling effort under current conditions. In our case, PAHs are the non-growing substrate and Criddle’s equation has to be modified. In particular, the bioavailability limitation can be accounted for by replacing the total concentration by the bioavailable one (Cbioav):
1 tf
(12)
where Tc is the OMPs transformation capacity (mgOMP/gCODSs ) standing for cometabolic interaction between the soluble substrate metabolism and the OMPs metabolism, kc is the maximum specific rate of OMPs biodegradation in absence of primary substrate (mgOMP/gCOD-X.d) and KSC is the half saturation constant of OMPs in the Monod formalism (mgOMP/L). The kc and KSC parameters are representative of the response of the OMP metabolic route to the OMP bioavailable fraction including transporters and enzymes affinity for their substrate. m is the growth rate (1/d) and Y is the growth yield (gCOD-X/gCODSs ). Furthermore, the four-compartment distribution may help us to find which compartment is the real bioavailable fraction (Cbioav) to be biodegraded. Beside, the four-compartment model can be modified. By this way, the OMPs biodegraded fraction can be assumed to be the free dissolved fraction (Cf e only process 7) or the aqueous fraction (Cf and CDCM e processes 7 and 8), or the sum of all fractions (Cf, CDCM and Cp e processes 7, 8 and 9) as proposed by Fountoulakis et al. (2006).
3.3.
sq ¼
Sensitivity analysis
The model has a cascade structure, which means that the variables X, SS and Sp are not influenced by the other variables and, then, by the parameters associated with the other state variables. This cascade structure is an advantage to find the parameter set. First of all, we can estimate the parameters of biomass and substrates (mmax, KS, Y, b and khyd), and then, the parameters linked with the biodegradation of each OMP (Tc, kc and KSC).
Ztf 0
zqþDq zq dt zq
(13)
where tf is the test duration, zq is the variable z associated with base value of parameter q, and zqþDq is the variable z when the parameter q is changed an amount Dq. The sensitivity coefficient is presented in Fig. 3. Growth yield, Y, is the most sensitive parameter. A strong influence of parameters linked to soluble substrate uptake (Y, mmax, KS) can be noted. This is in agreement with cometabolism concept, where the micropollutant fate is associated to growth substrate degradation. Half saturation constant of OMP KSC and the specific biodegradation rate of OMP kc show a relative sensibility. These parameters are representative of the OMP metabolic route and depend on the type of microbial
Y kc
μmax Parameter
rbio
m C bioav X ¼ Tc þ kc KSC þ Cbioav Y
A sensitivity analysis for OMP total concentration was conducted to identify the most sensitive parameters in the four-compartment model with free dissolved compartment (Cf) as the available fraction. In reference to a given set of parameter values, initial condition and characteristics of pollutants and reactors, eleven parameters were changed over 10%, 20%, 30% and 50% of their based values. In steady state, nine simulations were run at each of these values to generate nine concentrations profiles of compartments (Cf, Cp, CDCM, Cg) for each parameter. A sensitivity coefficient, sq, of the variable z to the parameter q, defined by Equation (13) (Bernard et al., 2001; Myint et al., 2007), was calculated to quantify the average spread for each parameter.
KS KSC khyd b Tc KLa k1 k2 0
0,1
0,2
0,3
Sensitivity coefficient, σq Fig. 3 e Sensitivity coefficient of OMP concentration for the model parameters.
4516
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 1 1 e4 5 2 1
Model calibration
The model simulations were compared with experimental data of anaerobic digestion of PAHs for two mesophilic continuous reactors fed with primary (PS) and secondary sludge (SS). The influent and effluent macroscopic performances of reactors and biogas production are represented in Table 4. Volatile fatty acids (VFAs) did not accumulate and the methane content in biogas was about 70% in the two mesophilic reactors. The overall removals in COD, dry matter and organic matter were higher than 60%. The variation of PAHs removal between reactors was slightly different while between PAHs in one reactor was high. The reported overall PAHs removal rates were from 32 to 74% and from 38 to 73% for PS and SS reactor, respectively. In the model, the set of parameters were estimated using a non-linear least square method between simulated values
Table 4 e Anaerobic performances of PS and SS reactors. Average and standard deviation calculated from 5 measurements performed at steady state. Parameter COD input (g/L) COD output (g/L) % COD removal DM input (g/L) DM output (g/L) % DM removal VFAs input (gCOD/L) VFAs output (gCOD/L) % VFAs removal % CH4 L CH4/d
PS 28.1 13.4 52.4 22.4 12.1 46.0 0.70 0.11 86.2 0.67 0.80
3.6 1.7 7.5 1.7 2.0 8.4 0.31 0.12 12.9 0.23 0.21
SS 23.0 1.9 9.1 2.6 60.4 6.5 19.5 1.4 10.2 1.8 52.9 6.9 0.90 0.47 0.20 0.24 89.4 9.6 0.68 0.25 0.61 0.10
Parameter
Meaning
Unit
Value PS
1
mmax KS
Maximum growth rate d Half saturation of growth gCODSs/L substrate Growth yield gCODX/gCOD-Ss First-order endogenous d1 decay First-order kinetic of d1 hydrolysis
Y b khyd
SS
0.62 0.63 3.25 5.10 0.75 0.75 0.05 0.05 0.07 0.13
and measurements. We take here advantage of the cascade structure of the model with the identification of a first set of parameters. Table 5 summarizes the values of the parameters linked to biomass and substrates for PS and SS. The KS values suggest that the metabolism and the implied microbial population could be different in PS and SS biodegradation. Fig. 4 shows the behavior of soluble and particulate substrate in PS and SS reactors. The simulations closely followed the dynamic evolution of the soluble substrate. Note the difference of the particulate substrate concentration between PS and SS, as well as, the fast decline of the particulate substrate in SS reactor. It could explain the dissimilarity of the hydrolysis coefficients found for both digesters. However, the particulate substrate in PS reactor is not well predicted by the model. This may be due to the first order kinetics used in the hydrolysis step which may be different between a non-stabilized sludge (PS) by comparison to a stabilized one (SS). Moreover, it is well known that the hydrolysis is the rate-limiting step in the anaerobic digestion for particulate substrate and the first order kinetics may be inaccurate to describe the hydrolysis of certain complex substrates (Vavilin et al., 2008). The set of parameters estimated in this section were used in the model calibration for evaluation of cometabolism and bioavailability hypotheses.
4.1.
Cometabolism evaluation
The physical and chemical characteristics of the OMP, as well as environmental factors, may influence their biodegradability. There are numerous references reporting that 4
20
3
15
Sp (gCOD/L)
4.
Table 5 e Estimated values of the biomass and substrates parameters.
SS (gCOD/L)
consortium. Moreover, hydrolysis step (khyd) has little influence on the compartments concentration despite it is the ratelimiting step in the anaerobic digestion. As pointed out by the results, the least sensitive parameters are Tc, KLa, k1 and k2, and therefore they will be less precisely estimated. In fact, the volatilization of PAHs is negligible, and then the OMP gas concentrations are small, as a result KLa does not influence the OMP concentration. On the other hand, hydrophobic character of PAHs makes easy their sorption onto sludge. Therefore, the first order kinetic constants (k1, k2) have high values (Dionisi et al., 2006). Beside, Kordel et al. (1997), Dionisi et al. (2006) and Barret et al. (2010b) have demonstrated that the sorption equilibrium state for PAHs was achieved after 1 or 2 h shaking. This sorption mechanism is faster compared to the biodegradation of these compounds under anaerobic condition (Chang et al., 2003). As a consequence for hydrophobic pollutant, this little influence of the first-order kinetic constants of sorption to particle and DCM (k1, k2) on the model showed that the sorption kinetics are not the rate-limiting steps and that the equilibrium state can be sufficient to represent the sorption phenomenon in the OMPs removal. Indeed, the processes 5 and 6 can be replaced by their equilibrium in the model in the case of hydrophobic compounds.
2 1 0
0
20
40
60
10 5 0
0
20
40
60
Time (d) Fig. 4 e Behavior of soluble (SS) and particulate (Sp) substrate for reactors PS (gray) and SS (black). Circles (C): experimental data and solid line: model.
4517
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 1 1 e4 5 2 1
4.2.
recalcitrant compounds may be transformed in the presence of another compound used as carbon and energy source, i.e. cometabolism (Chang et al., 1993, 2003; Criddle, 1993; Tiehm and Fritzsche, 1995; Yuan et al., 2001; Clara et al., 2005; Plo´sz et al., 2010; Barret et al., 2010c). It is usually assumed that the cometabolism may occur relatively slower than metabolism of growth substrate. Consequently, the general scheme adopted for the cometabolism of an OMP is shown on Fig. 2a. Total biomass consortium synthesizes enzymes for soluble substrate uptake and OMP degradation. Likewise, we propose the free dissolved compartment of OMP as bioavailable fraction. The behavior of the model with cometabolism has been compared to a classic metabolism of OMP (Fig. 2b). In this case, a specific degrader uptake OMP (XOMP) with Monod-type kinetics. Besides, non-linear least square optimization method has been used to estimate PAH parameters in the cometabolism kinetics (Tc, kc and KSC) and classic kinetics (i.e. Monod kinetics for an OMP: mmax,OMP, KS,OMP, YOMP) for each reactor (PS and SS) and thirteen PAHs. Fig. 5 displays the simulated (with both models) and experimental results for the fluoranthene in PS and SS reactors. Similar behaviors have been obtained for the other PAHs (data not shown). The model simulations with cometabolism closely follow the dynamic evolution of the total fluoranthene concentration and its compartments. In contrast, the model with a classic metabolism resulted in an overestimation of the experimental data. The residual values evaluated are 50.1 and 0.82 for metabolism and cometabolism, respectively, in Cf compartment for fluoranthene and PS. Moreover, cometabolic route of our results can be reinforced by following facts: (i) Barret et al. (2010c) reported a strong correlation between PAH and dry matter removal rates, it agrees with the results of Trably et al. (2003) and (ii) under anaerobic condition, Chang et al. (2003) and Trably et al. (2003) shown no growth with PAH as source of carbon.
PS 200
100
100
50
0
0
20
40
60
0
0
20
40
60
Various definitions of bioavailability are used across many disciplines (Semple et al., 2004). In this paper, a bioavailable compound is the chemical fraction that can be freely transformed by a microorganism. From a general point of view, a sorbed micropollutant is not available for microbial degradation; while its biodegradation occurs predominantly in the bulk aqueous phase (Byrns, 2001; Artola-Garicano et al., 2003; Urase and Kikuta, 2005; Dionisi et al., 2006; Plo´sz et al., 2010; Barret et al., 2010c). However, few studies have concluded that at least some microorganisms are capable of degrading compounds directly from the sorbed phase (Haws et al., 2006; Fountoulakis et al., 2006). In order to find the real OMP bioavailable fraction, the fourcompartment model was modified into the biotic process matrix for Cf, CDCM and Cp compartments (Table 3 and processes 7, 8 and 9). To this end, we have tested three hypotheses in the model (Table 3). Hypothesis 1 (processes 1e7): the bioavailable fraction was assumed to only be the free dissolved compartment, given that it is possible to separate the free dissolved fraction and sorbed to DCM of the aqueous phase in the model. Hypothesis 2 (processes 1e8): aqueous fraction is available for the microbial degradation activity. This in concordance with the widespread assumption that the aqueous fraction of OMP corresponds to their bioavailable compartment (Chang et al., 2003; Artola-Garicano et al., 2003; Dionisi et al., 2006; Barret et al., 2010c). Indeed, two mechanisms were proposed for the degradation of micropollutant sorbed to DCM: (i) low molecular weight DCM might be able to cross the microbial membrane in form of micropollutant-DCM complex, because some molecules up to a few kDa were shown to cross bacterial membrane (Nikaido, 2003) and (ii) recently it had been demonstrated that particles might transport the sorbed micropollutant directly or to vicinity of
Free-dissolved Compartment ( g/L)
Effluent HAP ( g/L)
Influent HAP ( g/L)
Bioavailability evaluation
Sorbed to DCM Compartment ( g/L)
Sorbed to particle Compartment ( g/L)
3
30
60
2
20
40
1
10
20
0
0
20
40
60
0
0
20
40
60
0
0
20
40
60
0
20
40
60
SS 200
100
100
50
0
0
20
40
60
0
0
20
40
60
3
30
60
2
20
40
1
10
20
0
0
20
40
60
0
0
20
40
60
0
Time (d) Fig. 5 e Fluoranthene behavior in PS and SS reactors: (gray line) influent concentration of PAH, (black circles) experimental data, (white circles) values estimated from equilibrium constants, (black line) model with cometabolism and (dashed line) model with classic metabolism.
4518
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 1 1 e4 5 2 1
PS 150
Free-dissolved Compartment ( g/L)
Effluent HAP ( g/L)
Influent HAP ( g/L)
Sorbed to DCM Compartment ( g/L)
80
1
20
40
0.5
10
60
100
40
50 0
Sorbed to particle Compartment (μg/L)
20
0
20
40
60
0
0
20
40
60
0
0
20
40
60
0
0
20
40
60
0
0
20
40
60
0
20
40
60
SS 150
80
1
20
40
0.5
10
60
100
40
50 0
20
0
20
40
60
0
0
20
40
60
0
0
20
40
60
0
0
20
40
60
0
Time (d) Fig. 6 e Chrysene behavior in PS and SS reactors: (gray line) influent concentration of PAH, (black circles) experimental data, (white circles) values estimated from equilibrium constants, (black line) hypothesis 1, (dark-gray line) hypothesis 2 and (light-gray line) hypothesis 3.
Rate: rSs (gCOD/gCOD.d), rOMP ( g/gCOD.d)
the cell surface (Smith et al., 2009), prior to the diffusion of free micropollutant throughout cell membrane. Finally, the hypothesis 3 (processes 1e9): All compartments are bioavailable as proposed by Fountoulakis et al. (2006). This is probably a mechanism of transport of sorbed micropollutant to vicinity of microorganism. Non-linear least square optimization method has been used to estimate PAH parameters (Tc, kc and KSC) for each case, thirteen PAHs and two sludge. The residuals value taken up to quantify the best fitting do not present a high variation between cases, for example the residual values are 0.11, 0.10 and 0.12 for free dissolved, aqueous phase and all compartments as available fraction, respectively, for chrysene in Cf compartment and PS. Fig. 6 shows the comparison of model predictions (three hypotheses) for chrysene in PS and SS reactors. Similar behaviors have been obtained for the other PAHs (data not shown). Such results could suggest that the PAH degradation occurs at the same time into free, aqueous and solid fraction, i.e sorbed fractions into particle and dissolved colloidal matter could be bioavailable to degraders. However, the half
saturation constants (KSC) of PAHs estimated in the first case (the free dissolved fraction is the bioavailable one) are ten times and hundred times lower than for the two other hypotheses (aqueous fraction and the sum of all compartments, respectively) as shown on Fig. 7. Consequently, in the hypothesis 1, the degradation rate of OMP is faster than that of soluble substrate as shown on Fig. 7a. This disagrees with the cometabolism studies demonstrating that the cometabolism is relatively slow by comparison to the metabolism of growth substrate (Chang et al., 1993; Haws et al., 2006). In addition, half saturation value estimated in the hypothesis 3 is higher than the OMP concentration in the free compartment (KSC>>Cf). This implies that the free compartment degradation is negligible compared to the particle compartment one and suggests high affinity for the OMP sorbed to particle. This case is really atypical, since it is generally considered that the sorbed chemicals are unavailable for microorganisms unless desorption occurs first. Indeed, Feng et al. (2000) have demonstrated that some bacteria can degrade sorbed chemical but it is not more important than the OMP aqueous phase
b
KSC = 0.7
a
KSC = 8.8
c
KSC = 63
Concentration: Ss (gCOD/L), OMP (μg/L) Fig. 7 e Comparison between substrate degradation rate (solid line) and OMP degradation rate (dashed line) for three hypotheses. (a) hypothesis 1: Cbioav [ Cf, (b) hypothesis 2: Cbioav [ Cf, CDCM and (c) hypothesis 3: Cbioav [ Cf, CDCM, Cp.
4519
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 1 1 e4 5 2 1
0.16
0.008
kc/KSC (L/gCOD.d)
Tc/KSC (L/gCOD)
0.010
R2 = 0.50
0.006 0.004 2 R = 0.67
0.002 0
150
200
250
0.12
0.08
2 R = 0.49
0.04
0
300
R2 = 0.55
150
M (g/mol)
200
250
300
M (g/mol)
Fig. 8 e Tc/KSC and kc/KSC as a function of molecular masses of PAHs. PS (gray) and SS (black).
degradation. In contrast, hypothesis 2 presents a cometabolism slower than metabolism of soluble substrate (Fig. 7b) and the affinity for free compartment and DCM compartment are comparable. Finally, based on the three-compartment model, Barret et al. (2010c) reported a strong correlation between PAHs aqueous fraction degradation and the dry matter removal and shown that the PAH biodegradation depended on a combination of bioavailability and cometabolism. Thus, the widespread assumption that the aqueous fraction of PAHs corresponds to their bioavailable compartment (Chang et al., 2003; Artola-Garicano et al., 2003; Dionisi et al., 2006; Barret et al., 2010c) was validated by our results. It is worth noting that PAH volumetric gas fraction does not exceed 0.05% of micropollutant total concentration for both low and higher molecular weight PAHs (data not shown). Moreover, hydrophobic character of PAH proves strong PAH affinity (higher to 98%) for both particle and DCM while PAH free concentration is hardly detectable (1.5%). However, PAH sorption to particle in PS reactor (85 5%) was higher than SS (65 5%) one and PAH affinity for DCM was 13 5% and 33 5% for PS and SS, respectively. As a consequence in this study, higher biological removal of individual PAH was observed in SS reactor in contrast with PS one.
4.3.
Kinetic parameters
The results suggest that the three cometabolism parameters (Tc, kc and KSC) could explain the different biodegradation rates between PAHs and between bioreactors. This is valid under the assumption that the aqueous fraction (sum of free and sorbed to DCM compartments) is the bioavailable compartment. Fig. 8 shows the variation of Tc/KSC and kc/KSC as function of molecular weight of PAHs. The transformation capacity values Tc did not present differences between PAHs and reactors (value close to 0.05 mgOMP/gCODSs ). As a result, the Tc value could correlate with a molecular structure family. Indeed, this term links PAHs degradation to soluble substrate utilization, so it might play a role in the different fates of PAHs in the reactor fed with PS and SS. However, previous PS and SS characterizations presented similar composition and reported slight differences in proteins and lipids content (Table 1). As a result, Tc can present trifling variation between substrates (PS and SS). Half saturation constant KSC is likely to vary as a function of PAH molecular weight. KSC values indeed increase when PAHs
molecular weight increases. This is in accordance with the idea that high molecular weight PAHs are less efficiently removed (Chang et al., 2003). Moreover, specific biodegradation rate, kc was shown to vary between reactors in a similar range (PS: 0.70e0.85 and SS: 0.60e0.90 mgOMP/gCOD-X.d). Therefore, half saturation constant KSC and kc probably depend on microbial consortium. In this study, it was shown that different consortia exhibit different KSC and kc. This microbial effect could account for biodegradation differences reported when bioaugmentation strategy has been developed (Trably et al., 2003).
5.
Conclusion
A four-compartment model of the fate of thirteen PAHs during anaerobic digestion of contaminated sludge was developed and confronted with experimental data. The model includes abiotic and biotic processes: volatilization, sorption, biodegradation as metabolism or cometabolism. Furthermore, in the case of hydrophobic pollutants, the sorption process can be represented by the equilibrium state. The model helps in elucidating which fraction of the PAHs distribution at equilibrium state is the real bioavailable compartment. Thus, the simulation carried out in this study validated the accepted assumption that the aqueous phase is bioavailable. Indeed, biodegradation affinity for OMP free dissolved and OMP sorbed to DCM are comparable. Furthermore, PAH biodegradation rate was coupled to cometabolism. The PAH removal was linked to soluble substrate uptake in the anaerobic digestion of sludge. The modified cometabolism model predicted well the relation between bioavailability and cometabolism of OMP. As a result of the numerical simulation, the three cometabolism parameters (Tc, kc and KSC) were shown to be molecule-dependant. These estimated parameter values could explain the different biodegradation rates between PAHs and between reactors. Nevertheless, the applied methodology for the parameters identification may converge toward several values but this study can be considered as a starting point, given that the parameter values comparison with previously published data was hardly feasible. A limitation in this model is that it does not include the OMPs inhibition and toxic effect, which could be considered in future work. However, the model proposed is potentially useful to better understand the micropollutant distribution, predict the
4520
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 1 1 e4 5 2 1
fate of PAHs under anaerobic condition and help to optimize the operation process for their removal.
Acknowledgments We would like to thank the Ibague University and COLCIENCIAS (Administrative Department of Science, Technology and Innovation) e Colombia for their financial support.
references
Alder, A.C., Siegrist, H., Fent, K., Egli, T., Molnar, E., Poiger, T., Schaffner, C., Giger, W., 1997. The fate of organic pollutants in wastewater and sludge treatment: significant processes and impact of compound properties. CHIMIA International Journal of Chemistry 51 (12), 922e928. Artola-Garicano, E., Borkent, I., Damen, K., Jager, T., Vaes, W.H.J., 2003. Sorption kinetics and microbial biodegradation activity of hydrophobic chemicals in sewage sludge: model and measurements based on free concentrations. Environmental Science & Technology 37 (1), 116e122. Barret, M., Carrere, H., Latrille, E., Wisniewski, C., Patureau, D., 2010a. Micropollutant and sludge characterization for modeling sorption equilibria. Environmental Science & Technology 44 (3), 1100e1106. Barret, M., Patureau, D., Latrille, E., Carre`re, H., 2010b. A threecompartment model for micropollutants sorption in sludge: methodological approach and insights. Water Research 44 (2), 616e624. Barret, M., Carre`re, H., Delgadillo, L., Patureau, D., 2010c. PAH fate during anaerobic digestion of contaminated sludge: do bioavailability and/or cometabolism limit their biodegradation? Water Research 44 (13), 3797e3806. Barret, M., Cea Barcia, G., Guillon, A., Carrere, H., Patureau, D., 2010d. Influence of feed characteristics on the removal of micropollutants during the anaerobic digestion of contaminated sludge. Journal of Hazardous Materials 181 (1e3), 241e247. Bernard, O., Hadj-Sadok, Z., Dochain, D., Genovesi, A., Steyer, J.P., 2001. Dynamical model development and parameter identification for an anaerobic wastewater treatment process. Biotechnology and Bioengineering 75 (4), 424e438. Byrns, G., 2001. The fate of xenobiotic organic compounds in wastewater treatment plants. Water Research 35 (10), 2523e2533. Chang, M.K., Voice, T.C., Criddle, C.S., 1993. Kinetics of competitive inhibition and cometabolism in the biodegradation of Benzene, Toluene, and pXylene by two Pseudomonas isolates. Biotechnology and Bioengineering 41 (11), 1057e1065. Chang, B.V., Chang, S.W., Yuan, S.Y., 2003. Anaerobic degradation of polycyclic aromatic hydrocarbons in sludge. Advances in Environmental Research 7 (3), 623e628. Clara, M., Kreuzingera, N., Strenna, B., Gansb, O., Kroissa, H., 2005. The solids retention time-a suitable design parameter to evaluate the capacity of wastewater treatment plants to remove micropollutants. Water Research 39 (1), 97e106. Couillard, C.M., Courtenay, S.C., Macdonald, R.W., 2008. Chemical-environment interactions affecting the risk of impacts on aquatic organisms: a review with a Canadian perspective - interactions affecting vulnerability. Environmental Reviews 16, 19e44. Criddle, C.S., 1993. The kinetics of cometabolism. Biotechnology and Bioengineering 41 (11), 1048e1056.
Dionisi, D., Bertin, L., Bornoroni, L., Capodicasa, S., Petrangeli Papini, M., Fava, F., 2006. Removal of organic xenobiotics in activated sludges under aerobic conditions and anaerobic digestion of the adsorbed species. Journal of Chemical Technology & Biotechnology 81 (9), 1496e1505. Feng, Y., Park, J.H., Voice, T., Boyd, S., 2000. Bioavailability of soilsorbed Biphenyl to Bacteria. Environmental Science & Technology 34 (10), 1977e1984. Fountoulakis, M.S., Stamatelatou, K., Batstone, D.J., Lyberatos, G., 2006. Simulation of DEHP biodegradation and sorption during the anaerobic digestion of secondary sludge. Water Science & Technology 54 (4), 119e128. Haws, N.W., Ball, W.P., Bouwer, E.J., 2006. Modeling and interpreting bioavailability of organic contaminant mixtures in subsurface environments. Journal of Contaminant Hydrology 82 (3e4), 255e292. Joss, A., Zabczynski, S., Go¨bel, A., Hoffmann, B., Lo¨ffler, D., McArdell, C.S., Ternes, T.A., Thomsen, A., Siegrist, H., 2006. Biological degradation of pharmaceuticals in municipal wastewater treatment: proposing and classification scheme. Water Research 40 (8), 1686e1696. Joss, A., Andersen, H., Ternes, T., Richle, P.R., Siegrist, H., 2004. Removal of estrogens in municipal wastewater treatment under aerobic and anaerobic condition: consequences for plant optimization. Environmental Science & Technology 38 (11), 3047e3055. Kordel, W., Hennecke, D., Franke, C., 1997. Determination of the adsorption-coefficients of organic substances on sewage sludges. Chemosphere 35 (1e2), 107e119. Lindblom, E., Press-Kristensen, K., Vanrolleghem, P.A., Mikkelsen, P.S., Henze, M., 2009. Dynamic experiments with high bisphenol-A concentrations modelled with an ASM model extended to include a separate XOC degradation microorganism. Water Research 43 (13), 3169e3176. Myint, M., Nirmalakhandan, N., Speece, R.E., 2007. Anaerobic fermentation of cattle manure: modeling of hydrolysis and acidogenesis. Water Research 41 (2), 323e332. Nikaido, H., 2003. Molecular basis of bacterial outer membrane permeability revisited. Microbiology and Molecular Biology Reviews 67 (4), 593e656. Plo´sz, B.G., Leknes, H., Thomas, K.V., 2010. Impacts of competitive inhibition, parent compound formation and partitioning behavior on the removal of antibiotics in municipal wastewater treatment. Environmental Science & Technology 44 (2), 734e742. Press-Kristensen, K., Ledin, A., Schmidt, J.E., Henze, M., 2007. Identifying model pollutants to investigate biodegradation of hazardous XOCs in WWTPs. Science of the Total Environment 373 (1), 122e130. Semple, K.T., Doick, K.J., Jones, K.C., Burauel, P., Craven, A., Harms, H., 2004. Defining bioavailability and bioaccessibility of contaminated soil and sediment is complicated. Environmental Science & Technology 38 (12), 228e231. Smith, K.E.C., Thullner, M., Wick, L.Y., Harms, H., 2009. Sorption to humic acids enhances polycyclic aromatic hydrocarbon biodegradation. Environmental Science & Technology 43 (19), 7205e7211. Tiehm, A., Fritzsche, C., 1995. Utilization of solubilized and crystalline mixtures of polycyclic aromatic hydrocarbons by a mycobacterium sp. Applied Microbiology and Biotechnology 42 (6), 964e968. Trably, E., Patureau, D., Delgene`s, J.P., 2003. Enhancement of polycyclic aromatic hydrocarbons (PAH) removal during anaerobic treatment of urban sludge. Water Science & Technology 48 (4), 53e60. Trably, E., Delgenes, N., Patureau, D., Delgenes, J.P., 2004. Statistical tools for the optimization of a highly reproducible method for the analysis of polycyclic aromatic hydrocarbons
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 1 1 e4 5 2 1
in sludge samples. International Journal of Environmental Analytical Chemistry 84, 995e1008. Urase, T., Kikuta, T., 2005. Separate estimation of adsorption and degradation of pharmaceutical substances and estrogens in the activated sludge process. Water Research 39 (7), 1289e1300.
4521
Vavilin, V.A., Fernandez, B., Palatsi, J., Flotats, X., 2008. Hydrolysis kinetics in anaerobic degradation of particulate organic material: an overview. Waste Management 28 (6), 939e951. Yuan, S.Y., Chang, J.S., Yen, J.H., Chang, B.V., 2001. Biodegradation of phenanthrene in river sediment. Chemosphere 43 (3), 273e278.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 2 2 e4 5 3 0
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Micellar enhanced ultrafiltration process for the treatment of olive mill wastewater Abdelilah El-Abbassi a, Mohamed Khayet b, Abdellatif Hafidi a,* a b
Laboratory of Food Sciences, Faculty of Sciences-Semlalia, Bd. My Abdellah, PB: 2390, 40090 Marrakech, Morocco Department of Applied Physics I, Faculty of Physics, University Complutense of Madrid, Av. Complutense s/n, 28040 Madrid, Spain
article info
abstract
Article history:
Olive mill wastewater (OMW) is an important environmental pollution problem, especially
Received 30 March 2011
in the Mediterranean, which is the main olive oil production region worldwide. Environ-
Received in revised form
mental impact of OMW is related to its high organic load and particularly to the phytotoxic
24 May 2011
and antibacterial action of its phenolic content. In fact, polyphenols are known as powerful
Accepted 31 May 2011
antioxidants with interesting nutritional and pharmaceutical properties. In the present
Available online 14 June 2011
work, the efficiency of OMW Micellar Enhanced Ultrafiltration (MEUF) treatment for removal and concentration of polyphenols was investigated, using an anionic surfactant
Keywords:
(Sodium Dodecyl Sulfate salt, SDS) and a hydrophobic poly(vinyldene fluoride) (PVDF)
Olive mill wastewater
membrane. The effects of the process experimental conditions on the permeate flux were
Polyphenols
investigated, and the secondary membrane resistance created by SDS molecules was
Micellar enhanced ultrafiltration
evaluated. The initial fluxes of OMW processing by MEUF using SDS were 25.7 and
Treatment
44.5 l/m2 h under transmembrane pressures of 3.5 and 4.5 bar, respectively. The rejection
Secondary resistance
rate of polyphenols without using any surfactant ranged from 5 to 28%, whereas, it reached
Rejection rate
74% when SDS was used under optimum pH (pH 2). The MEUF provides a slightly colored permeate (about 88% less dark), which requires clearly less chemical oxygen demand (COD) for its oxidation (4.33% of the initial COD). These results showed that MEUF process can efficiently be applied to the treatment of OMW and for the concentration and recovery of polyphenols. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
Olive oil extraction processes produce large quantities of olive mill wastewater (OMW), which exhibit phytotoxic properties, mainly due to natural phenolic compounds. In Morocco about 5.8 105 m3 of OMW are produced yearly (Achkari-Begdouri and Klimm, 2004). The traditional oil extraction system produces about 0.5 m3 of liquid waste per ton of olives. The three-phase system produces 1.1e1.5 times the weight of milled olives (Paraskeva et al., 2007). Besides, OMW chemical composition depends on the cultivar, climate, soil nature, olives ripeness and oil extraction process (Parinos et al., 2007).
To protect the environment and the crops from possible damage, a specific law prohibiting direct discharge of OMW without pre-treatment has been recently adopted in Morocco. Similar situations can be found in other Mediterranean countries, where olive oil is produced. Under mechanical processing, only about 1% of the total polyphenols present in olives can be found in oil. Most part of the olive polyphenols remain in OMW and also in solid wastes (Niaounakis and Halvadakis, 2004). Several studies demonstrate the negative impact of a direct discharge of OMW on soil properties (Rinaldi et al., 2003). The antibacterial and phytotoxic properties of polyphenols were
* Corresponding author. Tel.: þ212(0)661412030; fax: þ212(0)524437412. E-mail address: [email protected] (A. Hafidi). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.05.044
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 2 2 e4 5 3 0
List of symbols Cf Cp Cg COD J Lp OD ΔP R Rmem
polyphenols concentration in the feed (g/l) polyphenols concentration in the permeate (g/l) gelation concentration (M) Chemical Oxygen Demand (g of O2/l) permeate flux (l/m2 h) hydraulic membrane permeability (l/m2 h Pa) optical density () transmembrane pressure (bar) rejection rate of phenolic compounds (%) intrinsic membrane resistance (m1)
confirmed as well as the negative effects of the acidic pH and high salinity, brought by traditional salting practices of olives prior to oil extraction, on soil (Sierra et al., 2001). However, when diluted, OMW are used for fertilizing, long term effects on micro-organisms, fertility and soil properties are still not completely understood. Numerous researches have tried to develop efficient technologies for OMW treatment. These technologies are mainly based on biological degradation (anaerobic and aerobic) (Ammary, 2005) or advanced oxidation process, such as ozonation, Fenton’s reagent, electrochemical oxidation, etc., (Marques et al., 1997), as well as on various combinations of them (Khoufi et al., 2006). It should be noted that the efficiency of the process, the complexity and the costs involved may vary significantly. High costs are often the main reason for not adopting these OMW treatment methods. Almost all the above cited treatments aim the destruction of the organic matter present in OMW although it contains high valuable compounds such as phenolic compounds. It is worth quoting that wastewaters from olive oil mills, can be interesting biological sources of high added value compounds, such as hydroxytyrosol or other antioxidant phenolic compounds. In fact, the phenolic compounds are known as important natural antioxidants with nutritional and pharmaceutical properties (Tuck and Hayball, 2002). Several medicinal and pharmaceutical scientists have demonstrated that olive phenols, especially hydroxytyrosol (3,4-di-hydroxyphenyl-ethanol), were effective in preventing and curing some important diseases (Visioli et al., 1995). Nevertheless, hydroxytyrosol is not commercially available in large quantities like other food additives. Several methods have been proposed for the production of hydroxytyrosol by means of chemical (Visioli et al., 1998) or enzymatic synthesis (Tuck et al., 2000). Such protocols are usually slow and expensive, resulting in few numbers of commercially available products containing pure hydroxytyrosol. Various methods have been proposed to recover phenols from olive mill wastewater such as solvent extraction, resin chromatography, solideliquid or liquideliquid extraction, and supercritical fluid extraction. Unfortunately, almost all tested processes are expensive and/or inappropriate (Ferna´ndezBolan˜os et al., 2002). Membrane technology has been proposed as a promising tool for oil wastewaters treatment. Paraskeva et al. (2007) tested different combinations of three membrane processes (ultrafiltration (UF), nanofiltration (NF),
Rs Rcp Rf Rtot s t vp vr hw
4523
membrane secondary resistance (m1) resistance related to the concentration polarization (m1) resistance related to the fouling phenomenon (m1) total resistance of the membrane (m1) effective membrane area (cm2) operating time (min) permeate volume (ml) retentate volume (ml) viscosity (Pa s)
and/or reverse osmosis (RO)) for olive mill wastewater fractionation. The combination of centrifugation and UF allow a chemical oxygen demand (COD) reduction of about 90% (Turano et al., 2002). The treatment consists in a preliminary centrifugation step, in which the suspended solids are removed in a selective separation phase by UF of the centrifuge supernatant. Stoller and Bravi (2010) studied the effect of particle size distributions in OMW on the critical flux and membrane fouling when different pre-treatment processes are applied prior to membrane processing. The reported results showed that optimal pre-treatment process, which allows the highest critical flux, should reduce significantly the solute concentration and increase the particle size over the membrane pore sizes. A treatment of olive mill wastewater (OMW) by combining an UF technique and an advanced oxidation process using UV/ H2O2 was applied to remove a large part of total solids and organic carbon (Drouiche et al., 2004). A combined application of RO and adsorption processes was also applied to the treatment of OMW (Canepa et al., 1988), and allowed a COD reduction of about 99%. Nevertheless, the main limitation of these proposed treatments is the high involved costs. Recently, membrane distillation (MD) has been efficiently applied for OMW treatment and concentration for polyphenols recovery (El-Abbassi et al., 2009). However, the limitation of such process was the obtained low permeate fluxes compared to pressure-driven membrane technologies. The efficiency of micellar enhanced ultrafiltration (MEUF) to recover phenolic derivatives from model mixtures have been evaluated and have shown a high rejection of such compounds (Purkait et al., 2005). So far, MEUF has been used to separate different organic and inorganic compounds, using various surfactants (Purkait et al., 2006). However, research studies using MEUF for the treatment of multi-solute systems such as OMW are very scarce within membrane literature. The concentration of OMW may reduce significantly the costs of the process used for polyphenols recovery. The natural phenolic compounds can be used as food additives; antimicrobials, pesticides; cosmetics and pharmaceutical compounds (Visioli et al., 1999). MEUF is a separation process in which surfactants are added to a waste stream to promote the removal of smaller molecules. Beyond a certain concentration level called the Critical Micellar Concentration (CMC), the surfactant
4524
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 2 2 e4 5 3 0
molecules will aggregate into structures known as micelles and will then solubilize the organic compound (OC) to form large OC-surfactant structures. The solution may be then subjected to UF using an appropriate membrane to concentrate the waste and produce a high quality permeate. The aim of the present work is to study the efficiency of MEUF when treating OMW and concentrating its phenolic compounds. The effects of some operating parameters such as pH and SDS concentration on the permeate flux and the rejection rate as well as on the membrane fouling were investigated.
2.
Materials and methods
2.1.
Olive oil mill wastewater samples
residue was dissolved in 10 ml of pure methanol. This final extract was called hereafter methanolic extract.
2.3.1.2. Determination of total phenolic content. The total phenolic content was determined following the Folin-Ciocalteau spectrophotometeric method according to Singleton et al. (1999) and using Tyrosol as a standard. The methanolic extract (0.2 ml) was diluted with distilled water (6.8 ml). Folin-Ciocalteau reagent (0.5 ml) was added and the contents of flask were mixed thoroughly. After 3 min, 1 ml of a sodium carbonate anhydrous solution (20 wt%) was added, and then the mixture was allowed to stand for 1 h in the dark. The optical density of the bluecolored samples was measured at 765 nm. The total phenolic content was determined as tyrosol equivalents (TYE) and values are expressed as mg of tyrosol/l of OMW. 2.3.2.
Membrane preparation
2.3.
Procedures
PVDF flat sheet membranes were prepared by phase inversion technique, using PEG as an additive and DMAc as a solvent. First, 80 ml of DMAc was mixed with 3 g of PEG. The mixture was allowed to stand for at least 12 h at ambient temperature. After complete dissolution of PEG in DMAc, 17 g of the PVDF polymer was added to the mixture and heated at 45 C during 24 h. The formed polymer solution was degassed at a room temperature before casting the polymer solution into a glass plate. The cast film together with the glass plate, were immersed in distilled water for coagulation. The contact of the cast film with distilled water induces a diffusional mass exchange, solvent-out and water-in the nascent membrane. During gelation, it was observed that the membrane peeled out from the glass plate spontaneously. After 12 h in distilled water (coagulant bath), the formed membrane was subjected to solvent exchange method. First, the membrane was immersed in 50% aqueous methanol solution for 6 h. Subsequently, the membrane was dried at ambient conditions for at least 24 h.
2.3.1.
Physico-chemical characterization of OMW samples
2.4.
OMW samples were obtained from semi-modern units based on press-based oil extraction process located in region of Marrakech (Morocco). The samples were freshly used for the physico-chemical analysis and UF experiments.
2.2.
Chemicals
The anionic surfactant SDS (Sodium Dodecyl Sulfate salt), the poly(vinyldene fluoride) (PVDF, MW ¼ 180 kD), the additive poly(ethylene glycol) (PEG, 1 kD), the solvent N,N-Dimethyl acetamide (DMAc), methanol, ethyl acetate, Folin-Ciocalteu reagent, Tyrosol and sodium bicarbonate were procured from SigmaeAldrich Co. Hydrochloric acid and n-hexane were purchased from Rieden-deHae¨n Co. All chemicals were of analytical grade. Tyrosol was used as standard for the determination of total phenolic content. All chemical solutions were prepared using distilled water.
The electrical conductivity (EC) and pH of the OMW samples were measured directly using WTW Multilab (P5 Germany). The humidity and dry extract were determined before and after drying the sample overnight at 105 C. The chemical oxygen demand (COD) was measured using closed refluxcolorimetric method. The total organic carbon (TOC) was determined by the Anne method described by Aubert (1978). Proteins and sugars were measured using Bradford reagent (Bradford, 1976). Total phosphorus was measured colorimetrically as a molybdovanadate phosphoric acid (APHA, 1981). The total phenolic content of both the feed and permeate samples was determined colorimetrically by the FolinCiocalteau reagent (Folin and Ciocalteau, 1927), after the liquideliquid extraction detailed below:
2.3.1.1. Liquideliquid extraction of polyphenols from OMW. About 10 ml of each sample (feed or permeate) were acidified to pH 2 with HCl (6 N) and washed twice with 10 ml of nhexane to eliminate traces of lipids. The water phase was extracted three times with 10 ml of ethyl acetate by centrifugation at 3000 g for 5 min. The three organic phases were recuperated and evaporated under reduced pressure and the
MEUF experiments
UF experiments were carried out at room temperature in a stirred UF cell (AMICON 8200, Millipore USA) with a 200 ml volume. The effective area (s) of the membrane is 28.7 cm2. A schematic diagram of experimental apparatus is shown in Fig. 1. Before each run, pure water was employed to determine the permeability of the membrane. Then, the surfactant solutions were prepared by solubilizing sodium dodecyl sulfate in distilled water to reach concentrations 10 times the critical micellar concentration (CMC). The CMC of the used surfactant (SDS) in distilled water is 9.7 mM (Rosen, 2004). The feed solutions were mixed adequately for at least 5 h before UF tests. Stirring was set at 500 rpm and the transmembrane pressure was regulated with a manometer. To evaluate the membrane separation efficiency for removal of natural phenolic compounds from the feed solution, the following equation is used:
Rð%Þ ¼
Cp 100 1 Cf
(1)
4525
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 2 2 e4 5 3 0
COD removal ð% ¼
CODpermeate 100 1 CODfeed
(3)
3.
Results and discussion
3.1.
Physico-chemical characterization of OMW
The physico-chemical properties of the used OMW are shown in Table 1. This effluent is relatively dense, meanly acidic and charged in salt with a high organic load that reaches values as high as 156 g/l COD. It contains large amounts of dry matter up to about 90 g/l. Amongst other organic constituents, OMW contain high concentrations of phenolic compounds up to about 4 g/l, which exceeds widely the environment legislation limit, set at 0.5 ppm for phenol (Pinto et al., 2005).
3.2.
Fig. 1 e Schematic diagram of the stirred ultrafiltration cell and experimental set-up used for the MEUF: (1) Permeate outlet, (2) Membrane disc, (3) Feed tank, (4) Stirrer, (5) Blowoff valve, (6) Nitrogen inlet pressure, (7) Valve, (8) Manometer, (9) Nitrogen cylinder, (10) Magnetic stirrer and (11) graduated cylinder.
where R is the rejection rate, Cp is the permeate concentration and Cf is the feed concentration, of the phenolic compounds, respectively.
2.4.1.
It was observed that UF permeate volumes (v) of distilled water show a straight linear relationship with time (t) (v ¼ at þ b). Thus, the permeate flux could be directly obtained as: J¼
dv a ¼ sdt s
2.4.1.1. Decolorization. The color of OMW was monitored by measuring the absorbency at different wavelengths. Measurements at 395 nm and 465 nm after 100 times dilution against distilled water were carried out (Jaouani et al., 2003). The removal of color was estimated using the following equation. ODpermeate 100 % of decolorization ¼ 1 ODfeed
(2)
where OD is the optical density.
2.4.1.2. Chemical oxygen demand (COD). COD of OMW feed, permeate and retentate was determined using the dichromate method. An appropriate amount of sample was introduced in a previously prepared digestion solution containing potassium dichromate, sulfuric acid and mercuric sulfate (LaPara et al., 2000). The mixture was then incubated for 120 min at 150 C in a COD reactor (Thermoreactor CR3000). COD concentration was measured colorimetrically using UVeVisible spectrophotometer. The process efficiency in reducing COD was expressed as follows:
(4)
where s is the membrane area. Fig. 2 shows the obtained water flux (Jw) for different transmembrane pressures (ΔP). A linear relationship was obtained between Jw and ΔP with a high correlation factor (0.9993). The hydraulic membrane permeability (Lp) was determined according to the following equation (Cassano et al., 2008):
Pollution removal efficiency of the MEUF technique
To determine the efficiency of the MEUF technique in reducing the recalcitrant pollutants, the color and the chemical oxygen demand (COD) were considered.
Ultrafiltration of distilled water
Lp ¼
Jw DP
(5)
Moreover the membrane resistance (Rmem) was estimated as follows (Cassano et al., 2008): Rmem ¼
1 hw LP
(6)
Table 1 e Physico-chemical characteristics of OMW sample. Parameters Humidity (%) pH Conductivity (mS/cm) Dry extract (g/l) Oil (g/l) Total Organic Carbon (g/l) COD (g of O2/l) Total Phosphorus (g/l) Total polyphenols (g of TYE/l) Sugar (g/l) Proteins (g/l)
Average value Standard deviation 90 5.3 24 90 7 25 156 0.6 4.1
15 0.3 8 36 1 7 27 0.2 0.6
4.3 0.3 1.8 0.4
4526
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 2 2 e4 5 3 0
Fig. 2 e Effect of transmembrane pressure on permeate fluxes of distilled water ultrafiltration.
Fig. 3 e Permeate volume versus UF operation time for the surfactant solution (10 CMC of SDS) tested under different transmembrane pressures.
where hw is the viscosity of the permeate. The water viscosity is equal to 0.00083 Pa s (at 28 C and under the atmospheric pressure). The obtained values are given in Table 2. The hydraulic permeability (Lp) of the membrane is found to be 11.2 104 0.8 l/m2 h Pa, and the membrane resistance (Rmem) is 3.87 0.3 1012 m1. It is worth quoting that the membrane permeability and the membrane resistance decrease slightly with the increase of the transmembrane pressure.
3.3.
Ultrafiltration of surfactant solutions
As it was stated earlier, a surfactant solution, 10 CMC of SDS in distilled water (pH ¼ 2) was tested by UF. The results are shown in Fig. 3 for different transmembrane pressures. It can be seen that the UF permeate volume does not follow a straight line as a function of time. Therefore the results were fitted to a second order polynomial equation (v ¼ at2 þ bt þ c) and the permeate flux was calculated as follows: JðtÞ ¼
dv 2at þ b ¼ sdt s
(7)
As it was expected, it was found that the permeate flux increased with the enhancement of the transmembrane pressure. However, the transmembrane pressure effect on the permeate flux was found to be less significant compared to distilled water flux (Fig. 2). Moreover, a decrease of the
Table 2 e Permeate flux (Jw), membrane permeability (Lp) and membrane resistance (Rmem) under different transmembrane pressures. Lp (104 m)
permeate flux was observed during processing time. This reduction of flux may depend on four phenomena: pore diameter reduction due to the solute sorption on the pore wall, formation of a dense solute layer with low permeability directly on the membrane surface, formation of a compressible filter cake and presence of a concentration polarization layer. This fouling phenomenon is known to be the main limitation of the pressure-driven membrane separation processes, and different procedures were adopted to limit this membrane fouling (feed solution pre-treatment, membrane cleaning, membrane surface modification, etc.). By changing sensibly the particle size distributions of the suspended solids, flocculation pre-treatment was reported to reduce membrane fouling effects and the membrane’s performances demonstrated to be strictly dependant on the coagulant type used (Stoller, 2009). Effective mixing is efficient to reduce the influence of the concentration polarization. Therefore, the observed reduction of flux could be attributed to the formation of a gel, as was already observed by many other researchers (Urbanski et al., 2002). A linear relationship of permeate flux versus the logarithm of the surfactant concentration (by CMC) in the retentate could be expected. Since only monomeric molecules of SDS can pass through the membrane, and consequently the surfactant concentration in the MEUF permeate will not exceed 1 CMC (Talens-Alesson, 2007). The surfactant concentration in the retentate (CR) could be approximated from the mass balance. nsp þ nsr ¼ nso
(8)
Rmem (1012 m1)
ΔP (bar)
Jw (l/m2.h)
2 2.5 3 3.5 4
200.32 271.94 340.31 409.17 488.20
10 10.9 11.3 11.7 12.2
4.31 3.97 3.81 3.69 3.54
Mean value
11.2 0.8
3.87 0.3
where nsP is the quantity of the surfactant in the permeate ðnsP ¼ vp $CMCÞ, nsR is the quantity of the surfactant in the retentate ðnsR ¼ ðvo vp Þ$CR Þ and nsO is the initial quantity of the surfactant in the feed solution ðnso ¼ vo $10 CMCÞ, where vo and vp refer to the feed initial volume and the permeate volume, respectively. Therefore, the concentration of the surfactant in the retentate (CR) is expressed as follows:
4527
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 2 2 e4 5 3 0
10vo vp CMC CR ¼ vo vp
(9)
If ð10vo vp Þ=ðvo vp Þ is replaced by a, Eq. (9) is written as: CR ¼ a$CMC
(10)
where a is a dimensionless coefficient and the subindexes S, P and R denote the surfactant, the permeate and the retentate, respectively. Fig. 4 shows the permeate flux as a function of the logarithm of the coefficient a for different transmembrane pressures. Linear relationships were observed for each transmembrane pressure. The permeate fluxes will be zero at more or less similar values. The corresponding concentration in the retentate is the so called the gelation concentration Cg which can be deduced at J ¼ 0 as (Cg ¼ a$CMC). The obtained values of Cg are depicted in Table 3. The average SDS gelation concentration was found to be 28 CMC (0.27 M). This value is small compared to the values reported in literature, 0.87 M (Urbanski et al., 2002) and 0.57 M (Christian and Scamehorn, 1989). This difference can be justified by the various experimental conditions affecting the concentration of the surfactant at the membrane surface (e.g. membrane characteristics, temperature, pH, ionic strength of the solution). Moreover, PVDF has ferroelectric properties and it is electrostatically charged. Consequently, some interactions between the PVDF membrane and charged solutes may occur favoring a rapid formation of a gel. The secondary resistance (Rs) caused by the SDS layer formation (gelation) could be calculated as follow: RS ¼
DP Rmem hw JðtÞ
(11)
When considering the resistance-in-series model (Kim et al., 2002), the total resistance can be written as: Rtot ¼ Rmem þ Rcp þ Rf ¼ Rmem þ Rs
(12)
where the subscripts tot, mem, cp and f are the total, membrane, concentration polarization and fouling resistances. Rs is the secondary resistance. The most significant parameter affecting permeate flux decline is fouling resistance (Rf), which contributes to a great extent to the secondary
Fig. 4 e Permeate flux versus Ln(a) for distilled water ultrafiltration under different transmembrane pressures.
Table 3 e Gelation concentration of SDS for the feed solution prepared with a surfactant solution of pH 2 and an SDS concentration of 10 CMC. ΔP (bar) 2 3.5 4
Ln(a) (at J ¼ 0)
a
Cg (M)
3.29 3.33 3.35
26.91 27.89 28.37
0.26 0.27 0.28
Mean value
27.73 0.74
0.27 0.01
resistance. Nevertheless, Rf can be reduced by appropriate methods such as cross-flow ultrafiltration. The membrane fouling can be divided into reversible and irreversible fouling depending on the attachment strength of solutes to the membrane surface. The secondary resistance caused by SDS molecules was calculated at t0 using Eq. (11). The permeate flux measured under a transmembrane pressure of 4 bar was used. The SDS molecules were found to contribute to about 91% of the total resistance (Rtot ¼ 41.67 1012 m1 vs Rs ¼ 38.13 1012 m1).
3.4.
Micellar enhanced ultrafiltration of OMW
3.4.1.
pH effect
As many natural phenolic compounds show weak acid properties, the pH of the solutions can affect their solubilization into SDS micelles. Therefore, first the optimal pH under which SDS shows high efficiency for polyphenols solubilization was determined. The variation of polyphenols rejection by MEUF of OMW is shown in Fig. 5 for different pH values. The pH of the feed was adjusted by mean of HCl and NaOH (2 N) to the desired values and the initial surfactant concentration was maintained the same in all experiments (10 CMC). The results show that polyphenols rejection (R) decreases with the increase of the pH and R is the highest (>70%) when the pH value was adjusted to 2. For pH values higher than 8, R decreases with the increase of the pH to reach values under 30%. A reduction of the pH considerably lowers the SDS Critical Micellar Concentration from 0.9 mM at pH 6 to 0.34 mM at pH 2.9 (Paulenovfi et al., 1998) and from 4.6 mM at pH 7 to
Fig. 5 e Polyphenols retention of OMW by MEUF applying different pH values (Transmembrane pressure: 4 bar, SDS concentration: 10 CMC).
4528
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 2 2 e4 5 3 0
1.26 mM at pH 2 (Macisek and Danihlik, 1998). The increase of pH may cause deprotonation of phenols and as a result all types of interactions with the polar head of the surfactant will be reduced. These two considerations may explain the high rejection of the phenolic compounds at low pH values. The pH value 2 was considered as an optimal value and used in all the subsequent experiments.
3.4.2.
Effect of SDS concentration
The variation of polyphenols rejection (R) at different surfactant concentrations (0; 5 and 10 CMC) is presented in Fig. 6. In the case of OMW treated without surfactant (0 CMC of SDS), it was observed that the rejection is small (<30%), and decreases with time. In this case more solutes will be deposited on the membrane surface leading to an increase of polyphenols concentration at the membrane surface (concentration polarization). Consequently, the convective transport of phenolic compounds to the permeate side increases and thereby increasing the permeate concentration of polyphenols. As a result, R decreases with time. However, the rejection of polyphenols during MEUF of OMW with surfactant concentrations (5 and 10 CMC) was found to be significantly higher and decreases marginally with time. The surfactant concentration 10 CMC gives the best rejection which reaches about 74%. This clearly indicates that polyphenols are solubilized on/in the surfactant micelles which are subsequently retained by the ultrafiltration membrane.
3.4.3.
MEUF limits for OMW treatment and concentration
The main limitation of MEUF process, like in other membrane processes, is membrane fouling and concentration polarization. To evaluate these effects that occur during OMW treatment by MEUF, the following experiments were performed: SDS concentration of 10 CMC, pH 2, and a transmembrane pressure of 3.5 bar and 4.5 bar. The normalized permeate fluxes (ratio of the permeate flux at different time: J(t) to the initial permeate flux Jo) versus operating time are reported in Fig. 7. It can be seen that for both transmembrane pressures the normalized permeate flux decreases with time due to membrane fouling, concentration polarization and the increased solute concentration in the feed membrane side.
Fig. 7 e MEUF normalized permeate flux of OMW versus operation time.
The decrease of the normalized fluxes was fitted to Eq. (13) using LAB fit (Curve fitting software V 7.2.43): JðtÞ ¼ A þ B exp ðC tÞ Jo
The curves fitted to the experimental data are shown in Fig. 7 and the corresponding equations are as follows: JðtÞ ¼ 0:4096 þ 0:5941 exp ð0:0439 ðtÞÞ; Jo R2 ¼ 0:9881; ðDP ¼ 3:5 barÞ JðtÞ ¼ 0:3383 þ 0:6854 exp ð0:0287 ðtÞÞ; Jo R2 ¼ 0:9894; ðDP ¼ 4:5 barÞ The initial fluxes were 25.7 and 44.5 l/m2 h under the transmembrane pressures 3.5 bar and 4.5 bar, respectively. The measured permeate flux after 160 min shows lower values when applying 4.5 bar than that corresponding to 3.5 bar. A relatively severe fouling occur at the two tested transmembrane pressure. Stoller (2008) reported a critical pressure of 2.6 bar or below where fouling did not trigger, when treating olive wastewaters by cross-flow ultrafiltration using spiral-wounded modules.
3.5.
Fig. 6 e Polyphenols retention versus MEUF operating time using different concentration of SDS, (-) 10 CMC, (:) 5 CMC, (C) 0 CMC, (Transmembrane pressure: 4 bar).
(13)
Pollution removal efficiency of MEUF
Dark color of OMW is an indicator of its pollutant potential. The OMW processing by MEUF allows the rejection of a great part of the organic matter and pigments, which contribute to the OMW dark color. Our results demonstrated that MEUF process reduced effectively more than 87% of OMW color as shown in Table 4. This value is higher than that reported by Jarbaoui et al. (2008) after the infiltration of OMW through a soil using a laboratory-scale column. In another study (Jaouani et al., 2003) a decolorization of about only 30% of the OMW dark color was obtained after a biological treatment with Pleurotus sajor caju. However, a treatment of OMW by electrocoagulation has shown a decolorization rate of about 95%, for an OMW sample with only 57.8 g of O2/l as a COD (Adhoum and Monser, 2004). Furthermore, it is worth quoting
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 2 2 e4 5 3 0
Table 4 e MEUF efficiency for OMW decolorization using 10 CMC of the SDS and adjusting the feed pH to 2. l 395 nm 465 nm
Absorbance () Retentate Permeate Retentate Permeate
1.276 0.143 1.046 0.134
0.12 0.03 0.09 0.04
Decolorization (%) 88.79 3.75 87.19 5.40
that MEUF has considerably reduced chemical oxygen demand in the range of 95.7% (CODfinal ¼ 5.88 vs. CODinitial ¼ 136 g of O2/l). The infiltration of OMW through soil removes only 60% of the COD (Jarbaoui et al., 2008). Thus, MEUF efficiency for reducing COD of OMW is better than that obtained with other treatments. The MEUF allowed the rejection of more than 70% of phenolic compounds, which exceed the reported dephenolization (60%) by oxidative catalysts after 48 h of treatment (Iamarino et al., 2009). Furthermore, the phenolic compounds rejection from OMW by MEUF is much higher than the obtained rejection when using classical ultrafiltration on different membranes. The rejection did not exceed 17% (Cassano et al., 2011). This demonstrates clearly that micellar enhanced ultrafiltration is a powerful tool for the depollution of OMW. Moreover, the obtained polyphenols rich concentrate can be used for the extraction of polyphenols or other high added value compounds.
4.
Conclusion
Under our experimental conditions, the average SDS gelation concentration (Cg) in distilled water was found to be equal to about 28 CMC (0.27 M). The SDS was found to contribute to about 91% of the total membrane resistance. MEUF has allowed polyphenols to be concentrated from an aqueous waste, using a PVDF membrane. The MEUF process reduces considerably the color and the COD of OMW. In addition, it enables removal of other contaminants and could be an efficient method for olive mill wastewater treatment and valorization by subsequent extraction of natural phenolic compounds from the concentrate using solvents or surfactant-aided methods. The best conditions obtained for OMW processing by Micellar Enhanced Ultrafiltration on PVDF membranes were at 4 bar as transmembrane pressure and 10 CMC as the anionic surfactant concentration in an acidic medium (pH 2). The membrane fouling caused by surfactant gelation and adsorption should be minimized by reducing the membraneesurfactant interactions. Thus, the choice of the surfactant and membrane must be further optimized before a large scale application.
Acknowledgments The authors gratefully acknowledge the financial support from International Foundation of Science (Stockholm, Sweden) Grant N : W/4749-1.
4529
references
Achkari-Begdouri, A., Klimm, E., 2004. L’industrie ole´icole au Maroc et son impact sur l’environnement, propositions d’actions de lutte contre la pollution ge´ne´re´e par les huileries d’olives: cas de la province de Taounate. Secre´tariat d’Etat Charge´ de l’Environnement, Maroc. Adhoum, N., Monser, L., 2004. Decolourization and removal of phenolic compounds from olive mill wastewater by electrocoagulation. Chemical Engineering and Processing 43, 1281e1287. Ammary, B.Y., 2005. Treatment of olive mill wastewater using an anaerobic sequencing batch reactor. Desalination 177, 157e165. APHA American Public Health Association, 1981. Standard Methods for the Examination of Water and Wastewater, fifteenth ed.. Washington, DC, USA. Aubert, G., 1978. Me´thodes d’analyses des sols. C.R.D.P, Marseille, France. Bradford, M.M., 1976. A rapid and sensitive method for the quantification of microgram quantities of protein utilizing the principle of protein-dye binding. Analytical Biochemistry 12, 248e254. Canepa, P., Marignetti, N., Rognoni, U., Calgari, S., 1988. Olive mills wastewater treatment by combined membrane processes. Water Research 22 (12), 1491e1494. Cassano, A., Conidi, C., Drioli, E., 2011. Comparison of the performance of UF membranes in olive mill wastewaters treatment. Water Research 45, 3197e3204. Cassano, A., Mecchia, A., Drioli, E., 2008. Analyses of hydrodynamic resistances and operating parameters in the ultrafiltration of grape must. Journal of Food Engineering 89 (2), 171e177. Christian, S.D., Scamehorn, J.F., 1989. In: Scamehorn, J.F., Harwell, J.H. (Eds.), Surfactant-based Separation Processes, Surfactant Science Series, vol. 33. Dekker, New York, USA, p. 3. Drouiche, M., Le Mignot, V., Lounici, H., Belhocine, D., Grib, H., Pauss, A., Mameri, N., 2004. A compact process for the treatment of olive mill wastewater by combining OF and UV/ H2O2 techniques. Desalination 169, 81e88. El-Abbassi, A., Hafidi, A., Garcı´a-Payo, M.C., Khayet, M., 2009. Concentration of olive mill wastewater by membrane distillation for polyphenols recovery. Desalination 246, 297e301. Ferna´ndez-Bolan˜os, J., Rodrı´guez, G., Rodrı´guez, R., Heredia, A., Guille´n, R., Jime´nez, A., 2002. Production in large quantities of highly purified hydroxytyrosol from liquidesolid waste of two-phase olive oil processing or ‘‘Alperujio’’. Journal of Agricultural and Food Chemistry 50, 6804e6811. Folin, O., Ciocalteau, V., 1927. On tyrosine and tryptophan determination in protein. Journal of Biological Chemistry 73, 627e649. http://www.minenv.gov.ma/fodep/traitement_des_ margines.asp (accessed 09.08.09.). Iamarino, G., Rao, M.A., Gianfreda, L., 2009. Dephenolization and detoxification of olive-mill wastewater (OMW) by purified biotic and abiotic oxidative catalysts. Chemosphere 74, 216e223. Jaouani, A., Sayadi, S., Vanthournhout, M., Pennickx, M.J., 2003. Potent fungi for decolourisation of olive oil mill wastewaters. Enzyme and Microbial Technology 33, 802e809. Jarbaoui, R., Sellami, F., Kharroubi, A., Gharsallah, N., Ammar, E., 2008. Olive mill wastewater stabilization in open-air ponds: impact on clayesandy soil. Bioresource Technology 99, 7699e7708. Khoufi, S., Aloui, F., Sayadi, S., 2006. Treatment of olive oil mill wastewater by combined process electro-Fenton reaction and anaerobic digestion. Water Research 40, 2007e2016.
4530
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 2 2 e4 5 3 0
Kim, J.O., Shin, E.B., Bae, W., Kim, S.K., Kim, R.H., 2002. Effect of intermittent back ozonation for membrane fouling reduction in microfiltration using a metal membrane. Desalination 143, 269e278. LaPara, T.M., Alleman, J.E., Greg Pope, P., 2000. Miniaturized closed reflux, colorimetric method for the determination of chemical oxygen demand. Waste Management 20, 295e298. Macisek, F., Danihlik, A., 1998. Solvent extraction enhanced microfiltration of strontium and technetium with di-2ethylhexyphosphoric acid and quaternary ammonium salt. Solvent Extraction and Ion Exchange 16 (2), 587e596. Marques, P.A.S.S., Rosa, M.F., Mendes, F., Collares, M.P., Blanco, J., Malato, S., 1997. Wastewater detoxification of organic and inorganic toxic compounds with solar collectors. Desalination 108, 213e220. Niaounakis, M., Halvadakis, C.P., 2004. Olive-mill Waste Management. Literature review and patent survey TypothitoGeorge Dardanos, Atene. Paraskeva, C.A., Papadakis, V.G., Tsarouchi, E., Kanellopoulou, D.G., Koutsoukos, P.G., 2007. Membrane processing for olive mill wastewater fractionation. Desalination 213, 218e229. Parinos, C.S., Stalikas, C.D., Giannopoulos, Th.S., Pilidis, G.A., 2007. Chemical and physicochemical profile of wastewaters produced from the different stages of Spanish-style green olives processing. Journal of Hazardous Material 145, 339e343. Paulenovfi, A., Rajec, P., Adamik, P., 1998. Micellar ultrafiltration preconcentration of strontium by anionic micellar solution. Journal of Radioanalytical and Nuclear Chemistry 228 (12), 11e17. Pinto, R.T.P., Lintomen, L., Luz, L.F.L., Wolf-Maciel, M.R., 2005. Strategies for recovering phenol from wastewater: thermodynamic evaluation and environmental concerns. Fluid Phase Equilibria 228e229, 447e457. Purkait, M.K., DasGupta, S., De, S., 2005. Micellar enhanced ultrafiltration of phenolic derivatives from their mixtures. Journal of Colloid and Interface Science 285, 395e402. Purkait, M.K., DasGupta, S., De, S., 2006. Micellar enhanced ultrafiltration of eosin dye using hexadecyl pyridinium chloride. Journal of Hazardous Materials B136, 972e977. Rinaldi, M., Rana, G., Introna, M., 2003. Olive-mill wastewater spreading in southern Italy: effects on a durum wheat crop. Field Crops Research 84, 319e326. Rosen, M.J., 2004. Surfactant and Interfacial Phenomena, third ed. Wiley, New York.
Sierra, J., Marti, E., Montserrat, G., Cruanas, R., Garau, M.A., 2001. Characterization and evolution of a soil affected by olive oil mill wastewater disposal. Science of the Total Environment 279, 207e214. Singleton, V.L., Orthofer, R., Lamuela-Raventos, R.M., 1999. Analysis of total phenols and other oxidation substrates and antioxidants by means of folin-ciocalteu reagent. Methods in Enzymology 299, 152e178. Stoller, M., 2008. Technical optimization of a dual ultrafiltration and nanofiltration pilot plant in batch operation by means of the critical flux theory: a case study. Chemical Engineering and Processing: Process Intensification 47, 1165e1170. Stoller, M., 2009. Particle size distribution for membrane fouling reduction. Desalination 240, 209e217. Stoller, M., Bravi, M., 2010. Critical flux analyses on differently pretreated olive vegetation waste water streams: some case studies. Desalination 250, 578e582. Talens-Alesson, F.I., 2007. Behaviour of SDS micelles bound to mixtures of divalent and trivalent cations during ultrafiltration. Colloids and Surfaces A: Physicochemical and Engineering Aspects 299, 169e179. Tuck, K.L., Hayball, P.J., 2002. Major phenolic compounds in olive oil: metabolism and health effects. Journal of Nutritional Biochemistry 13, 636e644. Tuck, K.L., Tan, H.W., Hayball, P.J., 2000. Synthesis of tritiumlabeled hydroxytyrosol, a phenolic compound found in olive oil. Journal of Agricultural and Food Chemistry 48, 4087e4090. Turano, E., Curcio, S., De Paola, M.G., Calabro, V., Iorio, G., 2002. An integrated centrifugationeultrafiltration system in the treatment of olive mill wastewater. Journal of Membrane Sciences 209, 519e531. Urbanski, R., Goralska, E., Bart, H.-J., Szymanowski, J., 2002. Ultrafiltration of surfactant solutions. Journal of Colloid and Interface Science 253, 419e426. Visioli, F., Bellomo, G., Galli, C., 1998. Free radical-scavenging properties of olive oil polyphenols. Biochemical and Biophysical Research Communications 247, 60e64. Visioli, F., Bellomo, G., Montedoro, G.F., Galli, C., 1995. Low density lipoprotein oxidation is inhibited in vitro by olive oil constituents. Atherosclerosis 117, 25e32. Visioli, F., Romani, A., Mulinacci, N., Zarini, S., Conte, D., Vincieri, F.F., Galli, C., 1999. Antioxidant and other biological activities of olive mill waste waters. Journal of Agricultural and Food Chemistry 47, 3397e3401.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 3 1 e4 5 4 3
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Transformation kinetics of biochemically active compounds in low-pressure UV Photolysis and UV/H2O2 advanced oxidation processes Carolina Baeza a, Detlef R.U. Knappe b,* a
Environmental Science Center EULA-Chile, University of Concepcio´n, P.O. Box 160-C, Concepcio´n, Chile North Carolina State University, Department of Civil, Construction, and Environmental Engineering, Campus Box 7908, Raleigh, NC 27695-7908, USA b
article info
abstract
Article history:
Factors controlling photolysis and UV/H2O2 photooxidation rates of the biochemically
Received 27 January 2011
active compounds (BACs) sulfamethoxazole, sulfamethazine, sulfadiazine, trimethoprim,
Received in revised form
bisphenol A, and diclofenac were determined. Experiments were conducted with a quasi-
24 May 2011
collimated beam apparatus equipped with low-pressure UV lamps. The effects of pH, H2O2
Accepted 31 May 2011
concentration, and background water matrix (ultrapure water, lake water, wastewater
Available online 7 June 2011
treatment plant effluent) on BAC transformation rates were evaluated. For the sulfa drugs, solution pH affected direct photolysis rates but had little effect on the hydroxyl radical
Keywords:
oxidation rate. For sulfamethoxazole, the neutral form photolyzed more easily than the
Antibiotics
anionic form while the reverse was the case for sulfamethazine and sulfadiazine. For
Endocrine disrupting chemicals
trimethoprim, the hydroxyl radical oxidation rate was higher for the cationic form (pH 3.6)
Pharmaceuticals
than for the neutral form (pH 7.85). Quantum yields and second order rate constants
Photooxidation
describing the reaction between the hydroxyl radical and BACs were determined and used
Quantum yield
together with background water quality data to predict fluence-based BAC transformation rate constants (k0 ). For both the lake water and wastewater treatment plant effluent matrices, predicted k0 values were generally in good agreement with experimentally determined k0 values. At typical UV/H2O2 treatment conditions (fluence ¼ 540 mJ cm2, H2O2 dose ¼ 6 mg L1), BAC transformation percentages in North Carolina lake water ranged from 43% for trimethoprim to 98% for diclofenac. In wastewater treatment plant effluent, BAC transformation percentages were lower (31e97%) at the same treatment conditions because the hydroxyl radical scavenging rate was higher. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
The presence of biochemically active compounds (BACs) such as endocrine disrupting chemicals (EDCs) and antimicrobial compounds in the aquatic environment continues to be an issue of concern. BACs are commonly detected in surface and
ground water, and concentrations of some BACs can reach low mg L1 levels (e.g., Alexy and Ku¨mmerer, 2006; Petrovic et al., 2004). While EDC concentrations in some surface water bodies are sufficiently high to cause gender bending in fish (e.g., Kidd et al., 2007), it is still debated whether the evolution of antibiotic-resistant bacteria is facilitated at such
* Corresponding author. Tel.: þ1 919 515 8791; fax: þ1 919 515 7908. E-mail address: [email protected] (D.R.U. Knappe). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.05.039
4532
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 3 1 e4 5 4 3
Table 1 e Molecular structures and characteristics of targeted compounds. BAC
Chemical structure
Properties
O H2N
S
NH
O
Sulfamethoxazole (SMX)
CH3
N
O
Sulfonamide (antibiotic) Molecular weight ¼ 278.3 Da pKa,1 ¼ 2.26a pKa,2 ¼ 7.65b
O H2N
S
NH N
O N
Sulfamethazine (SMZ)
Sulfonamide (antibiotic) Molecular weight ¼ 253.3 Da pKa,1 ¼ 1.74a pKa,2 ¼ 5.65b
CH3
H3C
O H2N
S
NH N
O
Sulfadiazine (SDZ)
Sulfonamide (antibiotic) Molecular weight ¼ 280.2 Da pKa,1 ¼ 2.02a pKa,2 ¼ 6.43b
N
CH3
NH2
O
Trimethoprim (TMP)
H3C
N
O H3C
DHFR inhibitor (antibiotic) Molecular weight ¼ 290.3 Da pKa,1 ¼ 3.23c pKa,2 ¼ 6.76c NH2
N O
CH3 HO
OH
Bisphenol A (BPA)
CH3
Cl OH NH
Endocrine disrupting chemical Molecular weight ¼ 228.3 Da pKa,1 ¼ 9.78d pKa,2 ¼ 10.53d
Analgesic Molecular weight ¼ 296.15 Da pKa ¼ 4.15e
O
Diclofenac (DCL) Cl
a b c d e
Lin et al., 1997a. Lin et al., 1997b. Qiang and Adams, 2004. Calculated from SPARC v.4.5 (http://ibmlc2.chem.uga.edu/sparc/). Database of experimental values in EPI SUITE v.4 (http://www.epa.gov/opptintr/exposure/pubs/episuitedl.htm).
concentrations. Also, the effects of chronic human exposure to different pharmaceuticals at trace levels in drinking water are not known (Snyder et al., 2005). Wastewater treatment plants (WWTPs) represent one important entry point for BACs into the environment (e.g. Go¨bel et al., 2005), and BACs are not effectively removed by conventional drinking water processes (Adams et al., 2002; Westerhoff et al., 2005).
The use of UV disinfection processes has increased dramatically in drinking water and wastewater treatment. While the transformation of organic compounds by direct photolysis at disinfection doses is limited (Adams et al., 2002; Canonica et al., 2008), advanced oxidation processes (AOPs) present a potentially effective treatment alternative for micropollutants. Several studies have evaluated UV/H2O2
4533
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 3 1 e4 5 4 3
processes for EDC and pharmaceutical removal. Rosenfeldt and Linden (2004) showed that bisphenol A transformation with a UV fluence of 1000 mJ cm2 and the addition of 15 mg L1 of H2O2 was w90% in ultrapure water and w60% in river water. Using surface water at pH 7, a low-pressure UV lamp, and a H2O2 dose of 10 mg L1, Pereira et al. (2007) found that 99% transformation of carbamazepine, naproxen, and clofibric acid required a UV fluence > 1500 mJ cm2 while iohexol required a UV fluence of 720 mJ cm2. In a pilot plant treating pre-treated lake water, pharmaceutical transformation ranged from 67 to 98% with a medium pressure UV fluence of 540 mJ cm2 and a H2O2 dose of 6 mg L1 (Kruithof et al., 2007). This research was conducted to provide a more detailed understanding of factors controlling BAC transformation rates in UV photolysis and UV/H2O2 processes. The BACs tested in this study were the antimicrobial compounds sulfamethoxazole (SMX), sulfamethazine (SMZ), sulfadiazine (SDZ), and trimethoprim (TMP), the EDC bisphenol A (BPA), and the analgesic diclofenac (DCL). Specific objectives were to (1) determine the effects of solution pH on photolysis and photooxidation rates of BACs for which the degree of ionization can change considerably over the pH range typically encountered in water treatment, and (2) describe both experimentally and mathematically the influence of lake water (LW) and wastewater treatment plant effluent (WWTPE) matrices on BAC transformation rates.
2.
Materials and methods
2.1.
Experimental approach
Batch photolysis and UV/H2O2 oxidation experiments were carried out in a bench scale quasi-collimated beam (QCB) apparatus (Bolton and Linden, 2003). The purpose of the QCB apparatus is to ensure that UV rays reaching the sample are perpendicular to the water surface, which permits the accurate measurement of UV irradiance at the water surface and thus the accurate determination of the UV fluence (dose) delivered to the sample. The QCB was equipped with four low-pressure (LP) UV lamps, and a UV radiometer (UVX Radiometer, Upland, CA, USA) was used to measure the UV irradiance at the surface of the sample. An iodide/iodate actinometer was used to calibrate the radiometer readings (Rahn et al., 2006). The delivered UV fluence to the sample was calculated with the method described by Bolton and Linden (2003). Photolysis and UV/H2O2 oxidation experiments were conducted at initial BAC concentrations of 4 (1) mM, and the parent compound transformation was monitored as a function of UV dose (mJ cm2). SMX, SMZ, SDZ, TMP, BPA, and DCL served as target compounds in this study, and their characteristics are shown in Table 1. The effects of the following factors on BAC photooxidation rates were evaluated: (1) pH, (2) H2O2 concentration, and (3) background water matrix composition (hydroxyl radical scavenging due to background organic matter, alkalinity and other inorganic constituents). To determine the pH-dependent quantum yield of SMX, SMZ, SDZ, and TMP, photolysis experiments were conducted in
ultrapure water (UPW) buffered at pH values at which either the neutral or ionic form of the antimicrobial compounds was dominant (neutral form of sulfonamides at pH 3.6, anionic form of SMX at pH 7.85, anionic forms of SMZ and SDZ at pH 9.7, neutral form of TMP at pH 9.7, and cationic form of TMP at pH 3.6). For the UV/H2O2 degradation of SMX, SMZ, SDZ and TMP in UPW, only experiments at pH 3.6 and 7.85 were conducted to evaluate pH effects on oxidation rates. At pH 9.7, the reaction between the hydroxyl radical (OH) and the BAC would be affected by elevated carbonate concentrations due to enhanced dissolution of atmospheric CO2. Solution pH effects were not evaluated for BPA and DCL, because their pKa values (9.78 and 4.15, respectively) are distant from typical water treatment pH values; at the tested pH of 7.85, the neutral form of BPA and the anionic form of DCL was dominant. To quantify BAC oxidation rates in the UV/H2O2 process, QCB experiments were conducted with H2O2 concentrations of 2, 6, and 10 mg L1. Experiments in the presence of background organic matter were conducted at pH 7.85 in LW collected from Lake Wheeler (Raleigh, NC) and in WWTPE (Cary, NC). Both LW and WWTPE were filtered through a 0.45-mm nylon membrane (Magna-R, MSI, Westboro, MA) prior to use. Characteristics of LW and WWTPE are shown in Table 2. Photolysis and oxidation rate data were evaluated using two approaches. First, a fluence-based pseudo-first order reaction rate approach was used to evaluate the effects of pH, H2O2 concentration, and background water matrix on BAC transformation rates. Furthermore, the quantum yield for each BAC was determined from the fluence-based pseudofirst order photolysis rate. Second, the second order rate constant (kOH) describing BAC oxidation by OH in ultrapure water at pH 7.85 was obtained through competition kinetics (Huber et al., 2003; Pereira et al., 2007). In this study, pchlorobenzoic acid ( p-CBA) was selected as the reference compound because it is not measurably degraded by direct photolysis. Benitez et al. (2004) reported a quantum yield of 0.0030 mole Einstein1 for p-CBA at pH 7 and a wavelength of 254 nm. The second order rate constant describing the oxidation of p-CBA by OH has a value of 5 109 M1 s1 in the pH range of 6e9.4 (Buxton et al., 1988). In addition, kOH of the target compound was corrected by the percentage of the pseudo-first order reaction rate that accounted for the direct
Table 2 e Representative water quality parameters for the lake water (LW) and wastewater treatment plant effluent (WWTPE) used in this study.
DOC (mg L1) A254 (cm1) Alkalinity (mg L1as CaCO3) Nitrate (mg L1) Nitrite (mg L1) Sulfate (mg L1) Chloride (mg L1) Bromide (mg L1)
LWa
WWTPEa
5.1 0.130 24.4 1.9 <0.25 4.6 5.8 <0.25
7.3 0.146 74.3 4.9 <0.25 47.5 59.2 <0.25
a filtered through a 0.45-mm membrane.
4534
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 3 1 e4 5 4 3
photolytic degradation of each BAC. Batch experiments for the determination of kOH were conducted with equimolar concentrations of BAC and p-CBA (4 1 mM) and an initial H2O2 concentration of 10 mg/L.
2.2.
minimize photodegradation. Catalase was obtained from Sigma Chemical Corporation. Acetonitrile used for HPLC analysis was HPLC grade (Fisher Scientific, Pittsburgh, PA, USA). All BACs were dissolved in the water matrix of interest without the use of a solvent carrier.
Reagents
The six targeted BACs (Table 1) were purchased from Sigma Chemical Corporation (St. Louis, MO, USA). SMX, SDZ, BPA and DCL, were stored at ambient temperature, while SMZ and TMP were stored at 4 C. All compounds were stored in the dark to
2.3.
Analytical methods
2.3.1.
BAC concentration
Aqueous BAC concentrations were quantified by highperformance liquid chromatography (HPLC). Samples
a
b
c
d
e
f
Fig. 1 e Decadic molar absorption coefficients e as a function of wavelength for (a) SMX, (b) SMZ, (c) SDZ, (d) TMP, (e) BPA and (f) DCL.
4535
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 3 1 e4 5 4 3
collected in the presence of H2O2 were quenched with catalase (1% v/v of a 0.2 mg L1 stock solution) and filtered with a 0.22 mm membrane prior to HPLC analysis. The HPLC system (Breeze, Waters, Milford, MA) was equipped with a C18-AQ HPLC column (5 mm, 4.6 250 mm, Alltima HP, Grace) and a dual-wavelength UV detector. All samples were analyzed by direct injection, and the injection volume was 200 mL. The mobile phase flow rate was 1.0 mL min1. For SMX and SMZ analyses, the mobile phase was composed of 24% v/v A (acetonitrile) and 76% v/v B (25 mM ammonium acetate buffer at pH 5). The mobile phase was 20% v/v A and 80% v/v B for SDZ and TMP, and it was 42% v/v A and 58% v/v B for BPA and DCL. The detector wavelength was set at 266 nm for SMX, SMZ and SDZ, 238 nm for TMP, 225 nm for BPA, and 220 nm for DCL.
2.3.2.
Hydrogen peroxide
H2O2 concentrations were quantified with the Ghormley method (Klassen et al., 1994). This method is based on the spectrophotometric determination of I 3 that is produced when H2O2 reacts with I.
2.3.3.
Background water matrix
Dissolved organic carbon (DOC) was quantified with a Total Organic Carbon Analyzer (Model TOC-5000A, Shimadzu, Columbia, MD). Alkalinity was measured by titration with 0.02 N H2SO4 to the methyl orange endpoint. Chloride, nitrate, nitrite, sulfate and bromide were measured by ion chromatography (DIONEX ICS2500, Sunnyvale, CA).
3.
Results and discussion
3.1.
Photolysis rate and quantum yields
Photochemical reactions occur when a photon is absorbed by a molecule, and the likelihood of a compound to absorb light
at a specific wavelength is defined by the decadic molar absorption coefficient, e (Schwarzenbach et al., 2003). Values of e were determined as a function of wavelength (l) for each BAC, and the resulting absorption spectra are summarized in Fig. 1. For the antibiotics SMX, SMZ, SDZ and TMP, spectra are depicted at pH values at which the neutral and ionic forms (anionic for the sulfonamides, cationic for TMP) of the antibiotics dominated. Values of e254 nm for the six BACs are summarized in Table 3. The three sulfonamides exhibited higher e254 nm values than TMP, BPA, and DCL. Furthermore, the ionic forms of SMX, SMZ, SDZ, and TMP exhibited higher e254 nm values than their respective neutral forms. For BPA, the e254 nm value was less than a 1000 M1 cm1, and the degradation of BPA by direct photolysis can therefore be expected to be minor. In addition, quantum yields were calculated for each BAC. The quantum yield (f) determines the efficiency of photolysis and is defined as the moles of a compound that are transformed per mole of photons that were absorbed by the compound (Bolton, 2001). Bolton and Stefan (2002) derived an expression of the photolysis rate as a function of the average fluence rate; thus, f can be determined directly from the fluence-based pseudo-first order rate constant (kd0 ) describing the direct photolytic degradation of a compound; i.e., kd 0 ¼
f254nm e254nm lnð10Þ U254nm
where kd0 is the experimentally determined fluence-based pseudo-first order photolysis rate constant (cm2 mJ1), f254 nm is the quantum yield at 254 nm (mol Einstein1), e254 nm is the decadic molar absorption coefficient at 254 nm (M1 cm1), and U is the molar photon energy at 254 nm (4.72 105 J Einstein1). Values for kd0 and f254 nm are summarized in Table 3 for each BAC in UPW at different pH values. As shown in Table 3, the neutral form of SMX (pH 3.6) photolyzed more readily than
Table 3 e Decadic molar absorption coefficients, quantum yields, and pseudo-first order photolysis rate constants describing the photochemical behavior of BACs (all at 254 nm). pH SMX
SMZ
SDZ
TMP
BPA
DCL
1
1
e (M cm ) f (mol Einstein1) kd0 (cm2 mJ1)a e (M1 cm1) f (mol Einstein1)a kd0 (cm2 mJ1)a e (M1 cm1) f (mol Einstein1)a kd0 (cm2 mJ1)a e (M1 cm1) f (mol Einstein1)a kd0 (cm2 mJ1)a e (M1 cm1) f (mol Einstein1)a kd0 (cm2 mJ1)a e (M1 cm1) f (mol Einstein1)a kd0 (cm2 mJ1)a
3.6 11,130 0.180 0.00976 16,196 0.00282 0.000223 13,590 0.00430 0.000284 4956 0.00059 0.000014
(0.0073)a
(0.00017) (0.000013) (0.00045) (0.000030) (0.00029) (0.000007) e e e e e e
7.85
9.7
16,580 0.0297 (0.00086) 0.00240 18,525 0.00870 (0.00022) 0.000787 (0.000020) 20,150 0.00581 (0.00054) 0.000572 (0.000054) 2942 0.00118 (0.00011) 0.000017 (0.000002) 750 0.00460 (0.00043) 0.000017(0.000002) 5202 0.213(0.0047) 0.00533(0.00013)
e e e 20,538 0.00849 0.000851 20,660 0.00378 0.000382 2635 0.00149 0.000019
(0.00021) (0.000021) (0.000036) (0.0000036)
e e e e e e
a Values in parentheses represent one standard deviation; means and standard deviations were determined from experiments conducted in duplicate or triplicate.
4536
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 3 1 e4 5 4 3
the anionic form (pH 7.6), but the same was not observed for SDZ and SMZ. In addition, the molar absorptivity (e) of each compound was not related to the compound’s photolysis rate; i.e., e of SMX at pH 7.6 was higher than at pH 3.6, but the neutral form exhibited a higher quantum yield and faster photolysis rate. For SMX, the same pH dependence of e and f was obtained by Boreen et al. (2004) when solar photolysis conditions (l > 290 nm) were used. Also, the e and f values determined here are similar to those determined by Canonica et al. (2008) at l ¼ 254 nm for the neutral and anionic forms of SMX; i.e., Table 3 values are within 7 and 36% of the e and f values presented by Canonica et al. (2008), respectively. For SMZ and SDZ, the molar absorptivities of the anionic species were also higher than those of the neutral species, but quantum yields were similar for both the neutral and anionic forms of SDZ or slightly lower for the neutral form of SMZ. The net result was that the anionic forms of SMZ and SDZ photolyzed more rapidly than their neutral forms. Trends for SMZ and SDZ obtained here at 254 nm differ from those obtained with natural sunlight (l > 290 nm), for which no pH effect was observed (Boreen et al., 2005). The photolysis data for the sulfa compounds suggest that the N-bond substituent on the sulfonamide moiety (five-member heterocyclic ring for SMX, six-member heterocyclic ring for SMZ and SDZ) is responsible for their unrelated photochemical behavior. TMP and BPA had very low quantum yields in addition to their low molar absorptivity; as a result, both TMP and BPA photolysis rates were very slow. Although DCL exhibited a lower e than the sulfonamides, it had a high f value and a fast photolysis rate. Prior studies have shown that DCL readily photolyzes at 254 nm (Vogna et al., 2004; Canonica et al., 2008) and in natural sunlight (Packer et al., 2003).
3.2. Second order rate constants describing BAC oxidation by the hydroxyl radical Competition kinetics experiments were conducted at pH 7.85 to determine second order rate constants (kOH) that describe the oxidation of BACs by OH. Values of kOH ranged from 5e10 109 M1 s1 (Table 4), and this range agrees with values reported previously for different pharmaceutical compounds (e.g., Dodd et al., 2006; Pereira et al., 2007). Representative second order constants that were found in literature are shown in Table 4 for comparison. The kOH value for each BAC was corrected by the percentage of the pseudo-first order reaction rate that accounted for direct
photolytic degradation. SMX and DCL transformation in the UV/ H2O2 process was strongly influenced by direct photolysis; i.e., direct UV photolysis explained 39% and 55% of the overall UV/ H2O2 pseudo-first order reaction rate, respectively. In contrast, SMZ, SDZ, TMP and BPA transformation in the UV/H2O2 process was dominated by OH oxidation, and only 18%, 12%, 0.45% and 0.44%, respectively, of the overall UV/H2O2 pseudo-first order reaction rate was explained by direct photolysis.
3.3. Background water matrix effects on BAC transformation rates Experiments were conducted to quantify the effects of (1) solution pH and (2) background water matrix (LW, WWTPE) on fluence-based pseudo-first order rate constants describing BAC transformation rates in the UV/H2O2 process.
3.3.1.
Effect of solution pH
As shown in Fig. 2, pseudo-first order oxidation rates for SMX, SMZ, SDZ and TMP at pH 3.6 and 7.85 increased linearly with the initially applied H2O2 concentration. SMX transformation rates were higher when the neutral form dominated (pH 3.6), while SMZ and SDZ transformation rates were higher when the anionic form dominated (pH 7.85). The same pH effect for SMZ and SDZ was expected due to their structural similarity (Table 1). Another interesting observation is the uniformity within the slopes of the lines for the sulfonamides at different pH values. Similar slopes suggest that OH reacts at a similar rate with both the neutral and anionic forms of the three sulfonamides. The effect of solution pH on the UV/H2O2 transformation rates of sulfonamides is therefore mainly due to differences between the photolysis rates of the neutral and anionic species of SMX, SMZ and SDZ. For TMP, the UV/H2O2 oxidation rate of the cationic form (pH 3.6) was faster. At both pH 3.6 and 7.85, TMP photolysis was negligible, and TMP degradation occurred primarily via OH oxidation. The different slopes obtained at pH 3.6 and 7.85 therefore suggest that the second order rate constant for the reaction between TMP and OH is pH dependent and that the protonated form reacts more readily than the neutral form. This trend differs from that found for TMP oxidation by ozone, which reacts more readily with the neutral form of TMP (Dodd et al., 2006). However, the trend observed here is consistent with that found for TMP oxidation by potassium permanganate, for which the TMP oxidation rate also increased with decreasing solution pH (Hu et al., 2010).
Table 4 e Second order rate constants (kOH, ML1 sL1) describing BAC oxidation by the hydroxyl radical. kOH (M1 s1) this study at pH 7.85a SMX SMZ SDZ TMP BPA DCL
5.56(0.04) 5.65(0.05) 5.30(0.09) 5.70(0.03) 5.80(0.08) 9.26(0.26)
109 109 109 109 109 109
kOH (M1 s1) from literature 5.50 109 at pH 7.0, Huber et al., 2003 5.0 109 at pH 3.0, Boreen et al., 2005 3.7 109 at pH 3.0, Boreen et al., 2005 6.9 109 at pH 7.0, Dodd et al., 2006 1.02 1010 at pH 7.35, Rosenfeldt and Linden, 2004 7.5 109 at pH 7.0, Huber et al., 2003
a Values in parentheses represent one standard deviation; means and standard deviations were determined from experiments conducted in triplicate.
4537
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 3 1 e4 5 4 3
0.006
0.005
UPW - buffered at pH 3.6 UPW - buffered at pH 7.85 0.000 0
2
8
-1
10
2
-1 2
0.002
UPW - buffered at pH 3.6 UPW - buffered at pH 7.85
0.000
12
c
0.005
-1
rate constant (cm mJ )
6
[H2O2] mg L
0.003
0
2
4
6
-1
[H2O2] mg L
8
10
12
0.006
0.006
Fluence-based pseudo first order
4
0.004
0.001
0.004 0.003 0.002
UPW - buffered at pH 3.6 0.001
UPW - buffered at pH 7.85
0.000 0
2
4
6
-1
[H2O2] mg L
8
10
12
Fluence-based pseudo first order rate 2 -1 constant (cm mJ )
2
0.010
b
0.005
constant (cm mJ )
Fluence-based pseudo first order rate
a
-1
rate constant (cm mJ )
Fluence-based pseudo first order
0.015
d
0.005
0.004
0.003
0.002
UPW - buffered at pH 3.6
0.001
UPW - buffered at pH 7.85 0.000 0
2
4
6
-1
[H2O2] mg L
8
10
12
Fig. 2 e Effect of pH on fluence-based pseudo-first order rate constants for (a) SMX, (b) SMZ, (c) SDZ and (d) TMP. Note different y-axis scale for SMX.
3.3.2.
Effect of background water matrix
To determine background water matrix effects on BAC oxidation rates, QCB experiments were conducted with UPW, LW, and WWTPE at pH 7.85. The dependence of the fluencebased pseudo-first order oxidation rate constants (k’) on the applied H2O2 concentration is summarized in Fig. 3 for the six studied BACs. For all compounds, k0 increased linearly with the applied H2O2 concentration (R2 ranged from 0.977 to 1.000), and the regression equations relating k0 to the initial H2O2 concentration are shown in Table 5. As seen in Fig. 3, BAC transformation rates were slower in LW and WWTPE than in buffered UPW. For TMP, BPA and DCL, the photolysis rates in LW and WWTPE appear to be slightly higher than in UPW. Although enhanced photolytic transformation due to the presence of photosensitizers in LW and WWTPE cannot be completely ruled out (Canonica et al., 2008), the observed differences are likely due to experimental uncertainty. Even though the mean values of the photolysis rates of TMP, BPA, and DCL in LW and WWTPE were higher than in UPW, they were not significantly different at the 95% confidence level; therefore no case can be made that photosensitizers in LW and WWTPE enhanced photolysis rates. For the sulfonamides, photolysis rates in LW and WWTPE were lower than in UPW. In general, the water matrix affected the photolysis rate less than the OH oxidation rate, suggesting that the presence of OH scavengers in LW and WWTPE was the dominant factor that decreased BAC transformation rates. OH scavengers include dissolved organic carbon (DOC), alkalinity, chloride, sulfate and nitrate. As shown in Table 2, concentrations of
these scavengers in WWTPE were higher than in LW, which explains why BAC transformation rates were lower in WWTPE than in LW. The percentage contribution of each water quality parameter to OH scavenging was calculated by the reaction rate of each individual scavenger with OH and is described in detail below. It should be noted that SMX and DCL transformation rates in WWTPE did not increase substantially with the addition of H2O2 (Fig. 3, panels a and f). SMX and DCL in WWTPE were primarily transformed by direct photolysis while OH oxidation played only a minor role because of the strong scavenging effect of the WWTPE matrix. Therefore, SMX and DCL transformation in WWTPE benefits only marginally from the addition of H2O2. Fluence-based pseudo-first order rate constants can be used to determine the required UV fluence to achieve a desired treatment goal. With the linear regression equations derived from Fig. 3 and shown in Table 5, the UV doses required to achieve 90% BAC transformation with photolysis and UV/H2O2 at H2O2 concentrations of 2, 6, and 10 mg L1 were calculated (Fig. 4). For SMX and DCL, UV doses of <860 and <330 mJ cm2, respectively, were needed in the presence of 10 mg/L H2O2 to transform 90% of the parent compound in all three background water matrices. In contrast, for SDZ, SMZ, TMP and BPA, 90% transformation in LW and WWTPE required UV doses >900 mJ cm2 when the H2O2 dose was 10 mg L1. A UV dose of 900 mJ cm2 is high compared to those typically used in disinfection applications (w40e140 mJ cm2). At 40 mJ cm2 (without H2O2 addition), SMX and DCL transformations percentages would be w8% and w20%,
4538
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 3 1 e4 5 4 3
a
b
c
d
e
f
Fig. 3 e Effect of background water matrix on fluence-based pseudo-first order rate constants for (a) SMX, (b) SMZ, (c) SDZ, (d) TMP, (e) BPA and (f) DCL. Note different y-axis scale for DCL.
respectively, while SDZ, SMZ, TMP and BPA transformation would be negligible (3%, Table 5). At 140 mJ cm2, transformation percentages would increase to w25% and w55% for SMX and DCL, respectively, but would not exceed 10% for the remaining BACs (Table 5). Another interesting reference point
is a full-scale UV/H2O2 water treatment plant in the Netherlands. This plant operates at a UV dose of 540 mJ cm2 and a H2O2 concentration of 6 mg L1 (Kruithof et al., 2007). At these conditions, transformation percentages would range from 43 to 98% in the tested LW and from 31 to 97% in the
4539
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 3 1 e4 5 4 3
Table 5 e Regression equations relating the fluence-based pseudo-first order constant to the initial H2O2 concentration and predicted BAC transformation percentages in UPW, LW, and WWTPE for typical UV fluences used in disinfection applications (40 and 140 mJ cmL2) and conditions used in a full-scale UV/H2O2 water treatment plant. Compound
Water
Linear regression equation for k0
R2
% BAC transformation 2
40 mJ cm SMX
SMZ
SDZ
TMP
BPA
DCL
UPW LW WWTPE UPW LW WWTPE UPW LW WWTPE UPW LW WWTPE UPW LW WWTPE UPW LW WWTPE
k0 k0 k0 k0 k0 k0 k0 k0 k0 k0 k0 k0 k0 k0 k0 k0 k0 k0
¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼
0.000360 [H2O2]0 þ 0.00238 0.000154 [H2O2]0 þ 0.00200 0.000055 [H2O2]0 þ 0.00215 0.000360 [H2O2]0 þ 0.000781 0.000165 [H2O2]0 þ 0.000727 0.000106 [H2O2]0 þ 0.000767 0.000411 [H2O2]0 þ 0.000581 0.000144 [H2O2]0 þ 0.000471 0.0000890 [H2O2]0 þ 0.000431 0.000336 [H2O2]0 þ 0.000117 0.000164 [H2O2]0 þ 0.0000445 0.000109 [H2O2]0 þ 0.0000404 0.000358 [H2O2]0 þ 0.0000252 0.000180 [H2O2]0 þ 0.0000560 0.000104 [H2O2]0 þ 0.0000624 0.000437 [H2O2]0 þ 0.00559 0.000217 [H2O2]0 þ 0.00576 0.000130 [H2O2]0 þ 0.00578
0.990 0.996 0.977 0.997 0.999 0.998 0.999 1.000 0.998 0.986 0.999 0.999 1.000 1.000 0.994 0.991 1.000 0.997
140 mJ cm2
540 mJ cm2 þ 6 mg L1 H2O2
28 24 26 10 9.7 10 7.8 6.4 5.9 1.6 2.8 0.56 0.35 0.78 0.87 54 55 55
91 79 74 80 60 53 81 51 41 68 43 31 69 46 31 99 98 97
9.1 7.7 8.2 3.1 2.9 3.0 2.3 1.9 1.7 0.47 0.83 0.16 0.10 0.22 0.25 20 21 21
k0 : fluence-based pseudo-first order constant (cm2 mJ1); [H2O2]0: initial H2O2 concentration (mg L1).
tested WWTPE (Table 5). Using a pilot plant equipped with medium pressure UV lamps and operating at the same conditions as the full-scale Dutch UV/H2O2 plant, Kruithof et al. (2007) found similar transformation percentages for SMX (w75%) and DCL (w98%) in pre-treated Dutch lake water (Ijssel Lake). The close agreement between the results of Kruithof et al. (2007) and those calculated for the LW used in this study is likely fortuitous given differences in UV lamp type, UV reactor configuration, and background water quality.
3.4.
Mathematical modeling
The parameters obtained in this study (quantum yields, molar absorptivities and second order rate constants) were incorporated into a mathematical model to predict BAC transformation rates in waters with different background matrices (Sharpless and Linden, 2003; Rosenfeldt and Linden, 2004; Pereira et al., 2007). Background matrix effects that affect BAC transformation rates include UV absorbance (or transmittance) and OH scavenging by background organic matter and inorganic species such as HCO3-. Contaminant degradation rates in a UV/H2O2 process are described by the sum of the direct photolysis rate (kd0 ) and the OH oxidation rate or indirect photolysis rate (ki0 ); i.e., d½BAC ¼ kd0 þ ki0 ½BAC dt The direct photolysis fraction of the overall reaction rate is described by kd0 ¼ Ks; BAC at 254nm fBAC at 254nm where fBAC at 254 nm is the quantum yield for the BAC at 254 nm (Table 3) and Ks,BAC at 254nm is the specific rate of light
absorption of the target compound that is calculated with the following expression: Ks;BAC at 254nm ¼
Eo254nm eBAC at 254nm 1 10a254nm Z a254nm Z
whereEo254nm is the incident photon irradiance, eBAC at 254 nm is the decadic molar absorption coefficient of the targeted BAC at 254 nm (M1 cm1), Z is the solution depth (equal to 3.51 cm for the experiments in this study) and a254 nm is equal to the absorbance of the solution (comprised of the water background, H2O2, and the BAC); i.e., a254nm ¼ awater background þ eH2 O2
at 254nm ½H2 O2
þ eBAC at 254nm ½BAC
The indirect photolysis (or OH oxidation) fraction of the overall reaction rate is described by ki0 ¼ kOH=BAC ½ OHss where kOH=BAC is the second order rate constant describing the reaction between OH and a given BAC (Table 2) and ½ OHss is the steady state concentration of OH that is formed via H2O2 photolysis and is calculated from: ½ OHss ¼
KH2 O2 ;254nm fH2 O2 ;254nm ½H2 O2 ; and P k OH=Scavenger ½Scavenger i
KH2 O2
at 254nm
¼
Eo254nm eH2 O2
1 10a254nm Z a254nm Z
at 254nm
where all terms have been defined previously except for the P scavenging factor ðk OH=Scavenger ½ScavengerÞ. To determine the i scavenging factor, the water quality parameters shown in Table 2 for LW and WWTPE were used in combination with k OH=Scavenger values obtained from the literature. For the
4540
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 3 1 e4 5 4 3
Fig. 4 e UV fluence required to achieve 90% transformation of (a) SMX, (b) SMZ, (c) SDZ, (d) TMP, (e) BPA and (f) DCL in UPW, LW and WWTPE (all at pH 7.85).
inorganic scavengers, the following values were used: 8.5 106 M1 s1 for HCO3 (Buxton et al., 1988), 3.9 108 M1 s1 for CO32 (Buxton et al., 1988), 2.7 107 M1 s1 for H2O2 (Buxton et al., 1988), 1.5 106 M1 s1 for SO42 (Nakatani et al., 2007), 2 107 M1 s1 for Cl (Nakatani et al., 2007) and 5 105 M1 s1 for NO3 (Nakatani et al., 2007). For the background organic matter in LW and WWTPE, a second order rate
constant of 1.9 104 L (mg-C)1 s1 was used (Westerhoff et al., 2007). With the above model, a time-based pseudo-first order rate constant was obtained. This result was multiplied by the average fluence rate in a given collimated beam test to compare the model results with the experimentally determined fluence-based pseudo-first order rate constants (Fig. 5).
4541
0.001
0.000 1.95 mg/L H2O2
a
10.34 mg/L H2O2
2.03 mg/L H2O2
W W TPE
5.34 mg/L H2O2
10.94 mg/L H2O2
LW
Fluence-based pseudo first order rate 2 -1 constant (cm mJ )
0.004
Experimental Model
0.003
0.002
0.001
0.000
c
2.06 mg/L 6.13 mg/L 9.87 mg/L 1.96 mg/L 5.97 mg/L 9.86 mg/L H2O2 H2O2 H2O2 H2O2 H2O2 H2O2 WWTPE
LW
0.004
Experimental Model 0.003
2
-1
constant (cm mJ )
Fluence-based pseudo first order rate
5.23 mg/L H2O2
0.002
0.001
0.000 2.04mg/L H2O2
e
6.04 mg/L H2O2 W W TPE
10.22 mg/L H2O2
1.96 mg/L H2O2
6.17 mg/L H2O2
10.05 mg/L H2O2
LW
Experimental Model 0.003
2
-1
constant (cm mJ )
0.002
Fluence-based pseudo first order rate
0.003
Experimental Model
0.004
0.002
0.001
0.000 2.05 mg/L H2O2
b
10.09 mg/L 2.06 mg/L H2O2 H2O2
6.25 mg/L H2O2
10.07 mg/L H2O2
LW
0.004
Experimental Model 0.003
0.002
0.001
0.000
d
f
6.54 mg/L H2O2 W W TPE
Fluence-based pseudo first order rate -1 2 constant (cm mJ )
0.004
Fluence-based pseudo first order rate constant (cm2 mJ-1)
Fluence-based pseudo first order rate constant (cm2 mJ-1)
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 3 1 e4 5 4 3
2.04mg/L 5.91 mg/L 9.73 mg/L 2.10 mg/L 6.22 mg/L 9.90 mg/L H2O2 H2O2 H2O2 H2O2 H2O2 H2O2 W W TPE
LW
0.010 Experimental 0.008
Model
0.006
0.004
0.002
0.000 2.00mg/L H2O2
6.07 mg/L H2O2 WWTPE
9.62 mg/L H2O2
2.09 mg/L H2O2
6.17 mg/L H2O2
10.19 mg/L H2O2
LW
Fig. 5 e Comparison of experimentally determined fluence-based pseudo-first order rate constants with model predictions for (a) SMX, (b) SMZ, (c) SDZ, (d) TMP, (e) BPA and (f) DCL in UPW, LW and WWTPE (all at pH 7.85). Note different y-axis scale for DCL.
Good agreement between the model and the experimental values was obtained for SMX, SMZ, SDZ and DCL, while larger discrepancies were obtained for BPA and TMP (up to 41%), especially at higher H2O2 doses. Uncertainties in the second order rate constants for the BACs as well as the scavengers could have contributed to the differences between the model and the experimental values. The dominant contributors to OH scavenging in LW and WWTPE were DOC, alkalinity, and the added H2O2. For experiments at an H2O2 concentration of 10 mg L1, the DOC represented 86% and 71% of the overall scavenging rate in LW and WWTPE, respectively. For three wastewater
matrices, Rosario-Ortiz et al. (2010) found that the scavenging contribution of effluent organic matter in wastewater ranged from 76 to 93% of the overall OH scavenging rate. Considering the ions sulfate, chloride and nitrate in WWTPE, the overall scavenging factor is w18% higher than that obtained when only DOC, H2O2 and alkalinity are considered. For LW, sulfate, chloride, and nitrate contributed only w3% to the overall scavenging factor. In both waters, the dominant scavenging contribution among sulfate, chloride and nitrate ions came from the chloride ion. Even though the sulfate concentration in WWTPE was relatively high, kOH for sulfate (1.5 106 M1 s1) is one order of magnitude smaller
4542
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 3 1 e4 5 4 3
than that for chloride (2 107 M1 s1). As a result, OH scavenging due to sulfate accounted for <0.4% in WWTPE.
4.
Conclusions
Transformation rates of SMX, SMZ, SDZ, TMP, BPA and DCL in low-pressure UV photolysis and UV/H2O2 oxidation systems were determined. In particular, the effects of solution pH and background organic matter matrix composition on photooxidative BAC transformation rates were evaluated. Key findings were: For sulfonamides, pH-related differences in transformation rates were mainly due to differences in the photolysis rate between the neutral and anionic species For TMP, the reaction rate between TMP and OH was pH dependent, and the protonated form reacted more readily than the neutral form. Second order rate constants describing the OH oxidation of the six BACs ranged from 5.3 109 to 9.3 109 M1 s1. The ranking of the investigated BACs in terms of UV/H2O2 transformation efficiency was DCL > SMX > SMZ > SDZ> BPA z TMP. For a H2O2 dose of 10 mg L1, the required UV dose to achieve 90% SMX and DCL transformation was <860 and <330 mJ cm2, respectively. For the remaining BACs, UV doses >900 mJ cm2 were required to achieve 90% transformation in LW and WWTPE.
Acknowledgments This research was supported by the North Carolina Water Resources Research Institute, the National Science Foundation through a Graduate Research Fellowship, and a National Water Research Institute (NWRI) fellowship. In addition, the authors would like to thank Erik Rosenfeldt and Martin Srb for their assistance.
references
Adams, C., Wang, Y., Loftin, K., Meyer, M., 2002. Removal of antibiotics from surface and distilled water in conventional water treatment processes. Journal of Environmental Engineering 128 (3), 253e260. Alexy, R., Ku¨mmerer, K., 2006. Antibiotics for human use. In: Reemtsma, T., Jekel, M. (Eds.), Organic Pollutants in the Water Cycle. Wiley VCH, Weinheim, Germany, pp. 65e86. Benitez, F.J., Acero, J.L., Real, F.J., May, C., 2004. Modeling of photooxidation of acetamide herbicides in natural waters by UV radiation and the combinations UV/H2O2 and UV/O3. Journal of Chemical Technology and Biotechnology 79 (9), 987e997. Bolton, J.R., Linden, K.G., 2003. Standardization of methods for fluence (UV dose) determination in bench-scale UV experiments. Journal of Environmental Engineering 129 (3), 209e215. Bolton, J.R., Stefan, M.I., 2002. Fundamental photochemical approach to the concepts of fluence (UV dose) and
electrical energy efficiency in photochemical degradation reactions. Research on Chemical Intermediates 28 (7e9), 857e870. Bolton, J.R., 2001. Ultraviolet Applications Handbook, second ed. Bolton Photosciences Inc., Edmonton, AB, Canada. Boreen, A.L., Arnold, W.A., McNeill, K., 2005. Triplet-sensitized photodegradation of sulfa drugs containing six-membered heterocyclic groups: identification of an SO2 extrusion photoproduct. Environmental Science & Technology 39 (10), 3630e3638. Boreen, A.L., Arnold, W.A., McNeill, K., 2004. Photochemical fate of sulfa drugs in the aquatic environment: sulfa drugs containing five-membered heterocyclic groups. Environmental Science & Technology 38 (14), 3933e3940. Buxton, G., Greenstock, C.L., Helman, W.P., Ross, A.B., 1988. Critical review of rate constants for reactions of hydrated electrons, hydrogen atoms and hydroxyl radicals (OH/O) in aqueous solution. Journal of Physical and Chemical Reference Data 17 (2), 513e886. Canonica, S., Meunier, L., von Gunten, U., 2008. Phototransformation of selected pharmaceuticals during UV treatment of drinking water. Water Research 42 (1e2), 121e128. Dodd, M.C., Buffle, M.O., von Gunten, U., 2006. Oxidation of antibacterial molecules by aqueous ozone: moiety-specific reaction kinetics and application to ozone-based wastewater treatment. Environmental Science & Technology 40 (6), 1969e1977. Go¨bel, A., Thomsen, A., Mcardell, C.S., Joss, A., Giger, W., 2005. Occurrence and sorption behavior of sulfonamides, macrolides, and trimethoprim in activated sludge treatment. Environmental Science & Technology 39 (11), 3981e3989. Hu, L., Martin, H.M., Strathmann, T.J., 2010. Oxidation kinetics of antibiotics during water treatment with potassium permanganate. Environmental Science & Technology 44 (16), 6416e6422. Huber, M.M., Canonica, S., Park, G.Y., von Gunten, U., 2003. Oxidation of pharmaceuticals during ozonation and advanced oxidation processes. Environmental Science & Technology 37 (5), 1016e1024. Kidd, K.A., Blanchfield, P.J., Mills, K.H., Palace, V.P., Evans, R.E., Lazorchak, J.M., Flick, R.W., 2007. Collapse of a fish population after exposure to a synthetic estrogen. Proceedings of the National Academy of Sciences of the United States of America 104 (21), 8897e8901. Klassen, N.V., Marchington, D., McGowan, H.C.E., 1994. H2O2 determination by the I3e method and by KMnO4 titration. Analytical Chemistry 66 (18), 2921e2925. Kruithof, J.C., Kamp,., P.C., Martijn, B.J., 2007. UV/H2O2 treatment: a practical solution for organic contaminant control and primary disinfection. Ozone Science & Engineering 29 (4), 273e280. Lin, C.E., Chang, C.C., Lin, W.C., 1997a. Migration behavior and separation of sulfonamides in capillary zone electrophoresis. II. Positively charged species at low pH. Journal of Chromatography A 759 (1e2), 203e209. Lin, C.E., Lin, W.C., Chen, Y.C., Wang, S.W., 1997b. Migration behavior and selectivity of sulfonamides in capillary electrophoresis. Journal of Chromatography A 792 (1e2), 37e47. Nakatani, N., Hashimoto, N., Shindo, H., Yamamoto, M., Kikkawa, M., Sakugawa, H., 2007. Determination of photoformation rates and scavenging rate constants of hydroxyl radicals in natural waters using an automatic light irradiation and injection system. Analytica Chimica Acta 581 (2), 260e267. Packer, J.L., Werner, J.J., Latch, D.E., McNeill, K., Arnold, W.A., 2003. Photochemical fate of pharmaceuticals in the
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 3 1 e4 5 4 3
environment: naproxen, diclofenac, clofibric acid, and ibuprofen. Aquatic Sciences 65 (4), 342e351. Pereira, V.J., Weinberg, H.S., Linden, K.G., Singer, P.C., 2007. UV degradation kinetics and modeling of pharmaceutical compounds in laboratory grade and surface water via direct and indirect photolysis at 254 nm. Environmental Science & Technology 41 (5), 1682e1688. Petrovic, M., Eljarrat, E., de Alda, M.J.L., Barcelo, D., 2004. Endocrine disrupting compounds and other emerging contaminants in the environment: a survey on new monitoring strategies and occurrence data. Analytical and Bioanalytical Chemistry 378 (3), 549e562. Qiang, Z.M., Adams, C., 2004. Potentiometric determination of acid dissociation constants (pKa) for human and veterinary antibiotics. Water Research 38 (12), 2874e2890. Rahn, R.O., Bolton, J., Stefan, M.I., 2006. The iodide/iodate actinometer in UV disinfection: determination of the fluence rate distribution in UV reactors. Photochemistry and Photobiology 82 (2), 611e615. Rosario-Ortiz, F.L., Wert, E.C., Snyder, S.A., 2010. Evaluation of UV/H2O2 treatment for the oxidation of pharmaceuticals in wastewater. Water Research 44 (5), 1440e1448. Rosenfeldt, E.J., Linden, K.G., 2004. Degradation of endocrine disrupting chemicals bisphenol A, ethinyl estradiol, and estradiol during UV photolysis and advanced oxidation processes. Environmental Science & Technology 38 (20), 5476e5483.
4543
Sharpless, C.M., Linden, K.G., 2003. Experimental and model comparisons of low- and medium-pressure Hg lamps for the direct and H2O2 assisted UV photodegradation of N-nitrosodimethylamine in simulated drinking water. Environmental Science & Technology 37 (9), 1933e1940. Schwarzenbach, R.P., Gschwend, P.M., Imboden, D.M., 2003. Photochemical transformation reactions. In: Environmental Organic Chemistry, second ed. John Wiley & Sons, Inc., New York, pp. 611e686. Snyder, E.M., Pleus, R.C., Snyder, S.A., 2005. Pharmaceuticals and EDCs in the US water industry - an update. Journal American Water Works Association 97 (11), 32e36. Vogna, D., Marotta, R., Napolitano, A., Andreozzi, R., d’Ischia, M., 2004. Advanced oxidation of the pharmaceutical drug diclofenac with UV/H2O2 and ozone. Water Research 38, 414e422. Westerhoff, P., Mezyk, S.P., Cooper, W.J., Minakata, D., 2007. Electron pulse radiolysis determination of hydroxyl radical rate constants with Suwannee river fulvic acid and other dissolved organic matter isolates. Environmental Science & Technology 41 (13), 4640e4646. Westerhoff, P., Yoon, Y., Snyder, S., Wert, E., 2005. Fate of endocrine-disruptor, pharmaceutical, and personal care product chemicals during simulated drinking water treatment processes. Environmental Science & Technology 39 (17), 6649e6663.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 4 4 e4 5 5 0
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Adsorption of geosmin and 2-methylisoborneol onto powdered activated carbon at non-equilibrium conditions: Influence of NOM and process modelling Kristin Zoschke*, Christina Engel, Hilmar Bo¨rnick, Eckhard Worch Institute of Water Chemistry, Technische Universita¨t Dresden, 01062 Dresden, Germany
article info
abstract
Article history:
The adsorption of the taste and odour (T&O) compounds geosmin and 2-methylisoborneol
Received 7 July 2010
(2-MIB) onto powdered activated carbon (PAC) has been studied under conditions which are
Received in revised form
typical for a drinking water treatment plant that uses reservoir water for drinking water
19 May 2011
production. The reservoir water as well as the pre-treated water (after flocculation)
Accepted 3 June 2011
contains NOM that competes with the trace compounds for the adsorption sites on the
Available online 21 June 2011
carbon surface. Although the DOC concentrations in the reservoir water and in the pretreated water were different, no differences in the competitive adsorption could be seen.
Keywords:
By using two special characterisation methods for NOM (adsorption analysis, LC/OCD) it
Taste and odour compounds
could be proved that flocculation removes only NOM fractions which are irrelevant for
Powdered activated carbon
competitive adsorption.
Adsorption modelling Non-equilibrium conditions
Different model approaches were applied to describe the competitive adsorption of the T&O compounds and NOM, the tracer model, the equivalent background compound model, and the simplified equivalent background compound model. All these models are equilibrium models but in practice the contact time in flow-through reactors is typically shorter than the time needed to establish the adsorption equilibrium. In this paper it is demonstrated that the established model approaches can be used to describe competitive adsorption of T&O compounds and NOM also under non-equilibrium conditions. The results of the model applications showed that in particular the simplified equivalent background compound model is a useful tool to determine the PAC dosage required to reduce the T&O compounds below the threshold concentration. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
The occurrence of taste and odour (T&O) compounds in water reservoirs has a strong impact on the drinking water quality and is one of the major causes of consumer complains. Most T&O events in water reservoirs worldwide are caused by secondary metabolites of cyanobacteria, mainly geosmin and 2-methylisoborneol (2-MIB). Although these compounds are
non-toxic they can be detected by consumers at levels as low as 10 ng L1 as an earthy/musty odour. To preserve the aesthetic quality of drinking water, the concentration of the T&O compounds has to be reduced below the threshold odour concentration by an effective elimination process. The treatment with activated carbon has proved to be efficient for the removal of geosmin and 2-MIB. Powdered activated carbon (PAC) can be applied in the early stages of the conventional
* Corresponding author. Tel.: þ49 351 463 34967; fax: þ49 351 463 37271. E-mail address: [email protected] (K. Zoschke). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.06.006
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 4 4 e4 5 5 0
treatment (coagulation/flocculation, filtration, and disinfection) when algal episodes are seasonal or infrequent. If blooms of cyanobacteria regularly occur it is more economic to include granular activated carbon (GAC) filters following the rapid sand filtration step in the treatment process (Drikas et al., 2009). Several studies dealt with the adsorption of T&O compounds on PAC (Lalezary-Craig et al., 1988; Gillogly et al., 1999; Cook et al., 2001; Bruce et al., 2002; Newcombe et al., 2002a, b, c) and GAC (Chen et al., 1997; Ridal et al., 2001; Drikas et al., 2009). A general problem of the adsorption process consists in the reduction of the adsorption capacity due to the presence of dissolved natural organic matter (NOM). Typical NOM concentrations in drinking water sources are in the range of 2e10 mg L1 dissolved organic carbon (DOC) whereas the concentration of the micropollutants is in the range of 10e100 ng L1. In general, two mechanisms are described to explain the reduction of the adsorption capacity by NOM: larger NOM molecules can block the pores of the activated carbon and restrict the access for smaller molecules and NOM molecules can directly compete with the micropollutant for adsorption sites (Newcombe et al., 1997, 2002a). The general aim of the presented study was to investigate the adsorption of geosmin and 2-MIB on PAC for a typical situation in a water treatment plant in Saxony/Germany that uses reservoir water as raw water. The main treatment steps in this waterworks are flocculation, sand filtration, hardening, and disinfection. In case of the seasonal occurrence of T&O compounds in the raw water due to the growth of benthic cyanobacteria in the reservoir a PAC suspension is added after the flocculation basin. The activated carbon is then removed together with the residual flocs/particles from the flocculation step by sand filtration. The contact time between PAC and water is approx. 30 min which is not long enough to reach the state of adsorption equilibrium. This could be shown from preliminary kinetic experiments. Such non-equilibrium conditions are frequently found if PAC is added during drinking water treatment. On the other hand, the typical competitive adsorption models are equilibrium models. Therefore, the applicability of equilibrium models to non-equilibrium conditions was particularly examined in this study. The subtasks of the study were i) to determine short-term (30 min) and equilibrium adsorption isotherms for geosmin and 2-methylisoborneol in pure water, reservoir water, and pre-treated water and to quantify the NOM influence on T&O compound adsorption, ii) to characterize the NOM and to identify the NOM fractions which are responsible for competition with the T&O compounds, and iii) to check the applicability of competitive adsorption models to non-equilibrium data.
2. Modelling the competitive adsorption of NOM and T&O compounds In principle, there are two different approaches that can be used to describe the competitive adsorption of NOM and trace organic compounds, the tracer model and the equivalent background compound (EBC) model. Both models are based on the ideal adsorbed solution theory (IAST) which is
4545
a thermodynamic model for predicting mixture isotherms on the basis of single-solute adsorption parameters. The tracer model uses the IAST together with a preliminary NOM characterisation by the so-called adsorption analysis. The adsorption analysis is a fictive-component approach that allows for describing the NOM as a mixture of a limited number of differently adsorbable fractions. Using the adsorption analysis, the concentration distribution of the NOM fractions can be found from an experimental DOC isotherm (total isotherm) by a mathematical method based on a backward application of the IAST (Johannsen and Worch, 1994). To predict the trace compound adsorption in presence of NOM, the IAST has to be applied to the mixture system consisting of the NOM fraction and the trace compound. However, it was found that this conventional application of the IAST to such NOM/trace compound systems fails in many cases. There are different possible reasons for this failure. One of them is the need to use mass concentrations instead of the thermodynamically required molar concentrations, because DOC can be only measured as mass concentration (mg/L). Other possible reasons are non-ideal behaviour of the adsorbed phase, different accessibility of the micropores for NOM and trace compounds, and pore blocking effects. The tracer model was developed to overcome these difficulties. The tracer model includes a correction of the trace compound isotherm parameters for using in the IAST based on a fitting procedure with a measured trace isotherm in presence of NOM. After estimating the corrected isotherm parameters, they can be used for all further IAST calculations in the considered system. Details of the tracer model and the reasons for the IAST corrections are discussed by Burwig et al. (1995), Rabolt (1998), and Worch (2010). The EBC model (Najm et al., 1991) is based on the assumption that only one fraction of the NOM (the so-called equivalent background compound, EBC) is able to compete with the trace compound. The concentration and isotherm parameters of the EBC can be found from a measured isotherm of the tracer compound in presence of NOM by using a fitting procedure on the basis of the IAST. The EBC parameters together with the trace compound parameters can then be used to describe the competitive adsorption. A detailed description of the tracer model and the EBC model including their benefits and limitations was given in a previous paper (Worch, 2010). Knappe et al. (1998) and more generally Qi et al. (2007) have shown that the EBC model can be simplified for the special case of batch adsorption processes and under the condition that the adsorbed amount of the EBC is much higher than the adsorbed amount of the trace compound. The simplified EBC model gives the theoretical justification for the experimental finding that the relative removal c/c0 of the trace compound for a given adsorbent dose becomes independent from the trace compound’s initial concentration. Therefore, for a given NOM-containing water and a given adsorbent only one removal curve c/c0 ¼ f(adsorbent dose) has to be determined. For the considered system, this curve is then generally valid, independent of the trace compound’s initial concentration. It follows from the simplified EBC model that the general removal curve can be described mathematically by a twoparameter equation. The parameters can be found from the
4546
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 4 4 e4 5 5 0
removal
c1;0 1 mA ln 1 ¼ ln lnðAÞ c1 V n1
curve
by
linear
(1)
where c1,0 is the initial concentration of the trace compound, c1 is the (equilibrium) concentration of the trace compound, mA is the adsorbent mass, V is the volume of the solution, n1 and A are the characteristic parameters of the equation. n1 is the Freundlich exponent of the trace compound and A summarizes parameters of the trace compound and the EBC. Regardless of their physical meaning both parameters can be simply used as fitting parameters for the removal curve. If the parameters n1 and A are once known predictions of the removal efficiency for other initial concentrations are possible. The adsorbent dose necessary for a requested trace compound removal can be calculated by Eq. (2). n1 mA c1;0 ¼ An1 1 V c1
4.
Results and discussion
(2)
It has to be noted that all models developed for batch processes can also be applied to completely mixed flowthrough reactors if the adsorbent mass and the solution volume are substituted by the respective fluxes, mass per time and volume per time. All models described above are equilibrium models. They are based on the IAST which is derived from fundamental thermodynamic equilibrium relationships. On the other hand, all models contain fitting parameters which have to be determined from an experimental mixture isotherm. If using short-term (non-equilibrium) instead of equilibrium data within the models it can be assumed that only the values of the fitting parameters are affected, but the mathematical methods themselves should remain applicable. It was one of the goals of this study to prove this hypothesis.
3.
The odour standards (geosmin or 2-MIB, both from Supelco) were added to an initial concentration of 100 ng L1. After the contact time the samples were filtered (0.45 mm) to remove the PAC, and the odour compounds were analysed by headspace-solid phase microextraction-gas chromatography/ mass spectrometry (HS-SPME-GC/MS) outlined in Kutschera et al. (2009). To characterize the composition of the NOM, the LC/OCD method (Huber and Frimmel, 1992) was used. For the LC/OCD measurement the NOM compounds were separated in a Toyopearl HW-50S column (Tosoh Biosep, Stuttgart, Germany) and detected by an IR-detector after oxidation by a UV lamp (DOC-Labor Dr. Huber, Karlsruhe, Germany). Adsorption analysis was carried out according to Johannsen and Worch (1994).
Materials and methods
The adsorption experiments were conducted as batch experiments with 550 mL of reservoir water (DOC: 2.7 mg L1), water from the waterworks after flocculation (herein referred to as pre-treated water, DOC: 1.8 mg L1) or pure water from a Millipore system (DOC: 0.1 mg L1) at 20 C with a stirring rate of 400 rpm. The adsorption flasks were sealed with a glass stopper to avoid volatilization and a blank (without PAC addition) was included with each isotherm experiment to evaluate adsorbate losses by mechanisms other than adsorption. The results showed that no volatilization losses occurred during the studied contact times. The PAC SA Super from Norit which is frequently applied in the waterworks was selected for the experiments. Prior to use it was dried at 105 C. The PAC revealed the following pore volume distribution: 0.41 cm3 g1 micropores (<2 nm diameter) and 0.48 cm3 g1 mesopores (<50 nm diameter). For determining the adsorption isotherms, the PAC was added as a suspension to the water samples. The adsorption after contact times of 30 min and 24 h were studied. Kinetic studies conducted over 24 h contact time showed that equilibrium was reached within 8 h.
4.1. Comparison of short-term and equilibrium isotherms The adsorption isotherms of geosmin and 2-MIB were determined for a contact time that is sufficient for establishing the equilibrium (24 h) and for the contact time realized in the waterworks (30 min). Fig. 1 shows the isotherms for the adsorption on SA Super in pure water for both contact times. The adsorption data at equilibrium conditions as well as at non-equilibrium conditions can be described by the Freundlich isotherm. After 30 min adsorption time the isotherms of geosmin and 2-MIB are quite similar, but after 24 h the amount of geosmin adsorbed on the PAC has increased. Due to its flatter structure and lower molecular weight geosmin can enter smaller pores, this diffusion controlled transport process causes the increase in the adsorbed amount after 24 h compared to 30 min. Considering the error range of the isotherms, there is no difference between the isotherms of 2-MIB after 30 min and 24 h apparent. The isotherm data corresponding to the isotherms shown in Fig. 1 are given in Table 1. In the following sections it should be investigated if the competitive adsorption models developed for equilibrium conditions can also be applied to non-equilibrium (shortterm) isotherms because the contact time realized under practical conditions is not long enough to reach the state of equilibrium. 1.E-02
-1
determined
q [mmol g ]
experimentally regression:
1.E-03
1.E-04
1.E-05 1.E-09
geosmin, 30 min
geosmin, 24 h
2-MIB, 30 min
2-MIB, 24 h
1.E-08
1.E-07
1.E-06
-1
c [mmol L ]
Fig. 1 e Short-term and equilibrium adsorption isotherms of geosmin and 2-MIB on SA Super in pure water.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 4 4 e4 5 5 0
Table 1 e Freundlich isotherm parameters found for the adsorption of geosmin and 2-MIB on SA Super after contact times of 24 h (equilibrium) and 30 min (practical conditions). T&O compound Geosmin 2-MIB
Contact time
K [mmol1n g1 Ln]
n
R2
30 min 24 h 30 min 24 h
2.490 2.120 0.453 0.032
0.538 0.382 0.423 0.285
0.950 0.896 0.869 0.777
4.2. Influence of NOM on the adsorption of geosmin and 2-methylisoborneol Fig. 2 shows the elimination of the T&O compounds geosmin and 2-MIB from pure water, pre-treated water, and reservoir water in dependence on the PAC dose for a contact time of 30 min. Although the adsorption of the studied T&O compounds within 30 min from pure water is quite similar (see Fig. 2), geosmin is much better adsorbable than 2-MIB in NOMcontaining water. Given that geosmin has a lower molecular weight and a flatter structure than 2-MIB it can enter smaller micropores and shows therefore less direct competition with the NOM molecules. However, the adsorption efficiency of both compounds is decreased in natural water (containing NOM) compared to pure water. This effect is due to the direct competition between NOM molecules and the target compounds for adsorption sites and pore blocking by larger NOM molecules. Some NOM molecules are small enough to enter the micropores and, thus, compete directly with geosmin and 2-MIB for adsorption sites. Additionally, the NOM molecules may block access to even smaller micropores (Newcombe 2002a, b). As suggested by Kilduff et al. (1998), pore blocking and direct site competition become indistinguishable as the competing and target compounds become closer in size. Interestingly, the adsorption from pre-treated water and reservoir water is similar although the DOC concentration is quite different (pre-treated water: 1.8 mg L1, reservoir water: 2.7 mg L1). This effect was found for both compounds and suggests that the NOM fraction that is removed during the
Fig. 2 e Elimination of geosmin and 2-MIB in dependence on the dose of the powdered activated carbon SA Super after 30 min contact time (c0 [ 100 ng LL1).
4547
flocculation process has no effect on the adsorption of the T&O compounds. In order to corroborate this assumption, the NOM composition of the reservoir water and of the pre-treated water was further characterized by adsorption analysis and liquid chromatography with organic carbon detection (LC/OCD). The adsorption analysis is a fictive-component approach that can be used to characterize the NOM composition in view of differently adsorbable fractions. It is based on experimentally determined DOC isotherms. Details of this method can be found elsewhere (Sontheimer et al., 1988; Johannsen and Worch, 1994; Worch, 2010). For comparison, the adsorption analysis was carried out for the reservoir water and the pretreated water. In order to define four differently adsorbable fractions, the Freundlich K values were set to 0, 15, 55, and 150 mg1n g1 Ln. According to Johannsen and Worch (1994) the Freundlich exponents were set to 0.2 for all adsorbable fractions. The concentration distributions of the fractions were found by using a search routine based on the IAS theory and are depicted in Fig. 3. In both water samples only a small portion of the NOM is non-adsorbable (K ¼ 0). The comparison of the adsorption analyses of the water samples reveals that the flocculation process removes mainly low (K ¼ 15) and medium adsorbable (K ¼ 55) compounds and has no effect on highly adsorbable (K ¼ 150) compounds which are particularly relevant for the competition effect. The LC/OCD analysis can be used to describe the size of the NOM molecules and the amount of the different fractions. Fig. 4 shows the LC/OCD chromatograms of the raw water and the pre-treated water. It can be seen that the flocculation results in a reduction of mainly humic substances and building blocks and has nearly no influence on the lowmolecular fractions of NOM. Assuming that only smaller molecules can enter the activated carbon pore system, it can be expected that the low-molecular fraction of the NOM is the most important in view of competitive adsorption. In summary, the results of the adsorption analyses and the LC/OCD chromatography show that the studied reservoir
Fig. 3 e Results of the adsorption analyses of the reservoir water and the pre-treated water (adsorbent SA Super), K in mg1Ln gL1 Ln.
4548
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 4 4 e4 5 5 0
Fig. 4 e LC/OCD chromatogram of the reservoir water and pre-treated water (peak labelling according to Huber and Frimmel (1996)).
water and the pre-treated water contain a similar amount of NOM compounds which are relatively small and highly adsorbable. Since no difference in the adsorption of geosmin and 2-MIB from reservoir water and pre-treated water can be found, these fractions are obviously the most competitive for the T&O compounds. These results are in accordance with the findings of Newcombe et al. (2002a, b, c) and Hepplewhite et al. (2004) who stated that the low-molecular weight NOM (<600 g mol1) are most competitive. Furthermore, it can be concluded that the DOC removal from 2.7 mg L1 to 1.8 mg L1 during flocculation affects only the larger and less adsorbable NOM components which are irrelevant for competitive adsorption. Nevertheless, the extension of the contact time by the addition of the PAC before the flocculation step is not considered as alternative by the operator of the waterworks. The addition of the PAC after the flocculation/coagulation obviates adverse effects of the coagulant or the pH adjustment and prevents sedimentation of the PAC in the plant components before the filter system.
4.3.
Modelling the competitive adsorption
To describe the competitive adsorption of T&O compounds and NOM in pre-treated water, the tracer model and the EBC model were used. Additionally, the simplified EBC model was tested for applicability. The results of the modelling for nonequilibrium conditions using the tracer model and the EBC model are shown in Fig. 5. The tracer model fits very well the experimental data. The tracer model uses the results of the adsorption analysis and
Fig. 5 e Experimental data for the adsorption of geosmin and 2-MIB in pre-treated water and modelling by the tracer model and the EBC model (c0 [ 100 ng LL1, contact time 30 min).
modifies the isotherm parameters of the micropollutant. The modified values are determined by fitting the isotherm data of the adsorption of the micropollutant in natural water on basis of the IAST. The data of the adsorption analysis and the modified isotherm parameters of the micropollutant are used to predict the competitive adsorption. The isotherm parameters found from curve fitting and the original isotherm parameters are given in Table 2. These data together with the results from the adsorption analysis can be used for modelling the adsorption of the micropollutants as well as the NOM in the given system. It is a singularity of the tracer model that it allows for simultaneously describing the adsorption of NOM and micropollutants with a single parameter set. As further shown in Fig. 5 the EBC model can also describe the adsorption of the T&O compounds in NOM-containing water. For the curve fitting the isotherm parameters K and n of the EBC were adopted from the micropollutant assuming that EBC and micropollutant show nearly the same adsorption characteristics. With this database the EBC concentration was found to be 7.8$106 mmol L1 for the system geosmin/NOM and 5.0$105 mmol L1 for the system 2-MIB/NOM. To get an impression of the order of magnitude of the EBC concentration in terms of DOC, the DOC concentration was calculated for an assumed NOM molecular weight of 500 g mol1 and a carbon content of 50%. Under these conditions the DOC concentration of the EBC would be 2.0$103 mg L1 and 1.3$102 mg L1 for the geosmin/NOM and 2-MIB/NOM system, respectively. Compared to the total DOC of the pre-treated water of
Table 2 e Original isotherm parameters and modified isotherm parameters from the tracer model for the adsorption of geosmin and 2-MIB in pre-treated water with the PAC SA Super. Original isotherm parameters
K [(mg C/g)/(mg C/L)n] n
Modified isotherm parameters (tracer model)
Geosmin
2-MIB
Geosmin
2-MIB
15.6 0.50
7.5 0.42
137 0.15
363 0.48
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 4 4 e4 5 5 0
1.8 mg L1 this assumption shows that only a very small fraction of the NOM competes with the odour compounds for adsorption sites. This approves the conclusions drawn from the LC/OCD analysis and the adsorption analysis as well as the statements of Newcombe et al. (1997, 2002a, b, c). The results show that the models that were developed for the competitive adsorption at equilibrium can also be applied to non-equilibrium (short-term) isotherms. This is a very important finding in view of the practical conditions typically found in waterworks. On the other hand, the application of these models is time-consuming because they both require the experimental measurement of two isotherms. In case of the tracer model the DOC isotherm (for adsorption analysis) and the micropollutant isotherm in presence of NOM are necessary whereas in case of the EBC model the micropollutant isotherm in presence and absence of NOM have to be measured. Qi et al. (2007) evolved a simplified model approach based on the IAST and the EBC for the special case of batch reactors. The presumption of this approach is that the removal of the trace compound at a given PAC dose becomes independent from the initial concentration. This approach is based on the fitting of two parameters to only one measured removal curve according to Equation (1). These parameters can then be used to estimate the removal efficiency for any PAC dose and any initial concentration of the micropollutant. Fig. 6 shows for geosmin as an example that the removal curves for different initial concentrations coincide. The same result was found for 2-MIB and this effect was also described by Knappe et al., (1998), Gillogly et al., (1999), Cook et al., (2001), Bruce et al., (2002) and Matsui et al., (2003). Taking into account that the contact time in all cases was only 30 min, it can be concluded that the simplified EBC model is also applicable to non-equilibrium systems. This will be demonstrated by the following example. Using the data for pre-treated water at c0 ¼ 100 ng L1 the parameters n1 and ln A were found to be n1 ¼ 0.55 and ln A ¼ 0.81 for geosmin and n1 ¼ 1.14 and ln A ¼ 1.69 for 2-MIB, respectively. These parameters can subsequently be used to predict the removal curves for other initial concentrations and PAC concentrations as shown in Fig. 7. The removal curve predicted by the simplified model of Qi et al. (2007) reflects well the experimental data. For this model approach only one experimental curve has to be determined and for this reason the model approach is easily applicable for practical operations.
4549
Fig. 7 e Adsorption of geosmin and 2-MIB in pre-treated water on the powdered activated carbon SA Super after 30 min contact time (c0 [ 100 ng LL1), experimental data and modelling by the simplified EBC model.
5.
Conclusions
The batch adsorption experiments showed that the T&O compound geosmin is better adsorbable than 2-MIB and that the relatively adsorbed amount is independent of the initial concentration of the trace compounds. Comparison of the adsorption isotherms under equilibrium and non-equilibrium conditions revealed that the data can be described by Freundlich isotherms. The adsorption efficiency for both compounds was decreased in waters containing NOM. The results of the adsorption in reservoir water and the pre-treated water, the adsorption analysis, and the LC/OCD analysis indicated that small, highly adsorbable NOM compounds are the most competitive for adsorption sites. This NOM fraction seems to directly compete with the micropollutants for adsorption sites. The modelling of the competitive adsorption revealed that it is possible to describe the adsorption process for conventional equilibrium conditions as well as for non-equilibrium (shortterm) conditions that could be found for practical applications. The reason for the applicability under non-equilibrium conditions is that all models contain one or more fitting parameters. Errors resulting from the fact that the equilibrium is not reached are included in the value of the fitting parameters. These fitting parameters are therefore only valid for the given contact time. The conventional models, the tracer model and the EBC model, as well as the simplified EBC model allow good descriptions of the isotherm data in presence of NOM. For practical applications the simplified model approach is a useful tool to predict the required PAC dose to reduce the concentration of geosmin and 2-MIB below the threshold odour concentration.
Acknowledgements
Fig. 6 e Adsorption of geosmin from pre-treated water on SA Super at different initial concentrations (contact time: 30 min).
The authors thank the Institute of Water Resources and Water Supply of the Technical University Hamburg-Harburg for the LC/OCD measurements. We also want to acknowledge our cooperators, the DREWAG Stadtwerke Dresden GmbH, and the Federal Ministry of Education and Research for the financial support. The authors thank the European Social Fund (ESF) and the Free State of Saxony for the financial support.
4550
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 4 4 e4 5 5 0
references
Bruce, D., Westerhoff, P., Brawley-Chesworth, A., 2002. Removal of 2-methylisoborneol and geosmin in surface water treatment plants in Arizona. J. Water Supply Res. Technol. AQUA 51 (4), 183e197. Burwig, G., Worch, E., Sontheimer, H., 1995. A new method to calculate the adsorption behaviour of organic trace compounds in mixtures (Eine neue Methode zur Berechnung des Adsorptionsverhaltens von organischen Spurenstoffen in Gemischen). Vom Wasser 84, 237e249. Chen, G., Dussert, B., Suffet, I., 1997. Evaluation of granular activated carbons for removal of methylisoborneol to below odor threshold concentration in drinking water. Water Res. 31 (5), 1155e1163. Cook, D., Newcombe, G., Sztajnbok, P., 2001. The application of powdered activated carbon for MIB und geosmin removal: predicting PAC doses in four raw waters. Water Res. 35 (5), 1325e1333. Drikas, M., Dixon, M., Morran, J., 2009. Removal of MIB and geosmin using granular activated carbon with and without MIEX pre-treatment. Water Res. 43 (20), 5151e5159. Gillogly, T.E.T., Snoeyink, V.L., Newcombe, G., Elarde, J.R., 1999. A simplified method to determine the powdered activated carbon dose required to remove methylisoborneol. Wat. Sci. Tech. 40 (6), 59e64. Hepplewhite, C., Newcombe, G., Knappe, D.R.U., 2004. NOM and MIB, who wins in the competition for activated carbon adsorption sites? Wat. Sci. Tech. 49 (9), 257e265. Huber, S.A., Frimmel, F.H., 1992. A liquid chromatographic system with multi-detection for the direct analysis of hydrophilic organic compounds in natural waters. Fresenius J. Anal. Chem. 342, 198e200. Huber, S.A., Frimmel, F.H., 1996. Size-exclusion chromatography with organic carbon detection (LC-OCD): a fast and reliable method for the characterization of hydrophilic organic matter in natural waters (Gelchromatographie mit Kohlenstoffdetektion (LC-OCD): Ein rasches und aussagekra¨ftiges Verfahren zur Charakterisierung hydrophiler organischer Wasserinhaltsstoffe). Vom Wasser 86, 277e290. Johannsen, K., Worch, E., 1994. A mathematical model for realizing adsorption analyses (Eine mathematische Methode zur Durchfu¨hrung von Adsorptionsanalysen). Acta Hydroch. Hydrob 22 (5), 225e230. Kilduff, J.E., Karanfil, T., Weber, W.J., 1998. Competitive effects of nondisplaceable organic compounds on trichloroethylene uptake by activated carbon. I. Thermodynamic predictions and model sensitivity analyses. J. Colloid Interf. Sci. 205, 271e279. Knappe, D.R.U., Matsui, Y., Snoeyink, V.L., 1998. Predicting the capacity of powdered activated carbon for trace organic compounds in natural waters. Environ. Sci. Technol. 31, 1694e1698.
Kutschera, K., Bo¨rnick, H., Worch, E., 2009. Photoinitiated oxidation of geosmin and 2-methylisoborneol by irradiation with 254 nm and 185 nm UV light. Water Res. 43 (8), 2224e2232. Lalezary-Craig, S., Pirbazari, M., Dale, M., Tanaka, T., McGuire, M., 1988. Optimising the removal of geosmin andmethylisoborneol by powdered activated carbon. J. Am. Water Works Assoc. 80 (3), 73e80. Matsui, Y., Fukuda, Y., Inoue, T., Matsushita, T., 2003. Effect of natural organic matter on powdered activated carbon adsorption of trace contaminants: characteristics and mechanism of competitive adsorption. Water Res. 37 (18), 4413e4424. Najm, I.N., Snoeyink, V.L., Richard, Y., 1991. Effect of initial concentration of a SOC in natural water in its adsorption by activated carbon. J. Amer. Water Works Assoc. 83 (8), 57e63. Newcombe, G., Drikas, M., Hayes, R., 1997. Influence of characterised natural organic material on activated carbon adsorption: II. Effect on pore volume distribution and adsorption of 2-Methyisoborneol. Water Res. 31 (5), 1065e1073. Newcombe, G., Morrison, J., Hepplewhite, C., Knappe, D.R.U., 2002a. In the (adsorption) competition between NOM and MIB, who is the winner, and why? Water Sci. Technol. Water Supply 2, 59e67. Newcombe, G., Morrison, J., Hepplewhite, C., 2002b. Simultaneous adsorption of MIB and NOM onto activated carbon. I. Characterisation of the system and NOM adsorption. Carbon 40, 2135e3146. Newcombe, G., Morrison, J., Hepplewhite, C., Knappe, D.R.U., 2002c. Simultaneous adsorption of MIB and NOM onto activated carbon. II. Competitive effects. Carbon 40, 2147e3156. Qi, S., Schideman, L., Marin˜as, B.J., Snoeyink, V.L., Campos, C., 2007. Simplification of the IAST for activated carbon adsorption of trace organic compounds from natural water. Water Res. 41 (2), 440e448. Rabolt, B., 1998. Investigation of the Competitive Adsorption of Micropollutants and Natural Organic Constituents (Untersuchungen zur konkurrierenden Adsorption von Mikroverunreinigungen und natu¨rlichen organischen Wasserinhaltsstoffen), PhD thesis Technische Universita¨t Dresden, Germany. Ridal, J., Brownlee, B., McKenna, G., Levac, N., 2001. Removal of taste and odour compounds by conventional granular activated carbon filtration. Water Qual. Res. J. Can. 36 (1), 43e54. Sontheimer, H., Crittenden, J.C., Summers, R.S., 1988. Activated Carbon for Water Treatment. DVGW-Forschungsstelle. EnglerBunte-Institut, Karlsruhe University, Germany. Worch, E., 2010. Competitive adsorption of micropollutants and NOM: a comparative study of different model approaches. J. Water Supply Res. Technol. AQUA 59 (5), 285e297.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 5 1 e4 5 6 1
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Fate of hazardous aromatic substances in membrane bioreactors I. Mozo a,b,c, M. Stricot a,b,c, N. Lesage d, M. Spe´randio a,b,c,* a
Universite´ de Toulouse, INSA, UPS, INP; LISBP, 135 Avenue de Rangueil, F-31077 Toulouse, France INRA, UMR792 Inge´nierie des Syste`mes Biologiques et des Proce´de´s, F-31400 Toulouse, France c CNRS, UMR5504, F-31400 Toulouse, France d TOTAL Petrochemicals France e PRDML, RN117 BP 47, F-64170 Lacq, France b
article info
abstract
Article history:
In this work, the removal of hazardous aromatic compounds was investigated in two types
Received 22 February 2011
of membrane bioreactors (MBRs), based on cross-flow and semi dead-end filtration
Received in revised form
systems. BTEX and PAH were efficiently eliminated from wastewater during treatment via
27 May 2011
a membrane bioreactor (90e99.9%) but non-biotic processes, i.e. volatilisation and sorption,
Accepted 3 June 2011
contributed significantly. The semi dead-end MBR showed slightly better removal effi-
Available online 14 June 2011
ciencies than the cross-flow MBR. However, non-biotic processes were more significant in the first process and, finally, degradation rates were higher in the cross-flow MBR. Higher
Keywords:
degradation rates were explained by a higher bio-availability of pollutants. Differences in
MBR
shear stress imposed in cross-flow and semi dead-end filtration systems radically modify
Hazardous substances
the sludge morphology. High shear stress (cross-flow filtration) generates dispersed
PAH
bacteria and larger quantities of dissolved and colloidal matter. Sorption of hydrophobic
BTEX
compounds (PAHs) on suspended solid was less marked in disaggregated sludge. The
Bio-availability
results suggest new strategies for improving micro-pollutant degradation in MBRs. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
The removal of hazardous substances from domestic and industrial wastewater is now necessary to protect water resources. At the same time, more stringent constraints concerning sludge disposal and atmospheric pollution are being imposed on treatment plants. For industrial wastewaters, membrane bioreactors (MBR) are considered as the best available technologies (BREF, 2003), the conventional activated sludge process (CAS) being insufficient to reach emission limits in some cases. Basically, membrane bioreactors achieve very good organics (COD, BOD) and nutrient removal
(N, P) as well as perfect retention of suspended solids. However, the removal of hazardous aromatic compounds in MBRs has been little investigated (Bernhard et al., 2006; Cirja et al., 2007; De Wever et al., 2007; Lesjean et al., 2004; Schonerklee et al., 2009). Moreover, different MBR configurations have been proposed, based on external or internal submerged membranes, cross-flow or dead-end filtration, but there is still no comparison of the advantages or disadvantages of these options for treating specific pollutants. This work focuses on wastewaters from the chemical, petrochemical and petroleum industries, which commonly contain the following substances: poly-aromatic hydrocarbons
* Corresponding author. Universite´ de Toulouse, INSA, UPS, INP; LISBP, 135 Avenue de Rangueil, F-31077 Toulouse, France. Tel.: þ33 (0) 561559755; fax: þ33 (0) 561559760. E-mail addresses: [email protected] (I. Mozo), [email protected] (M. Stricot), [email protected] (N. Lesage), [email protected] (M. Spe´randio). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.06.005
4552
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 5 1 e4 5 6 1
(PAH), benzene, toluene, ethylbenzene, xylene (BTEX) and phenols. In activated sludge processes, aromatic compounds are removed by three main processes: biodegradation, volatilisation and sorption. The respective contributions of these processes depend on the physicalechemical properties of the molecules. The fate of these compounds in wastewater treatment plants has been simulated with steady state models (Byrns, 2001; CEMC, 2006; Clark et al., 1995) and the concepts have been discussed for mono and poly-aromatic compounds in conventional activated sludge processes (Manoli and Samara, 2008; Wang et al., 2007). However, little attention has been paid to their operation in membrane bioreactors. In most of the studies concerning aerobic biological processes, non-biotic transformations of aromatic compounds are generally insufficiently assessed because the authors focus on removal from the aqueous phase, evaluating the gas and solid phases only briefly (Farhadian et al., 2008). Based on the model of Byrns (2001), volatilisation of monoaromatic compounds (whose Henry’s constant is higher than 100 Pa m3 mol1) can reach up to 30% of the overall removal rate in activated sludge treatment. Similarly, considering the fate of hydrophobic compounds in activated sludge (example: PAHs) reveals that removal by sorption varies from 10% to 90% for molecules whose coefficients Kow vary from 103.5 to 106 respectively (Byrns, 2001; Manoli and Samara, 2008; Wang et al., 2007). The contribution of sorption to micropollutant removal in MBRs has been less studied (Joss, 2005). On the one hand, sludge sorption concepts developed for CAS can be transposed to MBR to some extent. However, on the other hand, modification of sludge properties due to shear stress and membrane retention modifies particulate and colloidal matters, and then could modify transport and partitioning phenomena. The adsorption of PAHs on dissolved and colloidal matter has been shown to improve the PAHs’ bioaccessibility in an anaerobic digester and a similar phenomenon could occur in MBRs (Barret et al., 2010). In addition, concerning the concentration of PAHs in the water discharged from an MBR, it is reasonable to think that the membrane gives an advantage by retaining the small particles containing adsorbed PAHs, which would pass through a conventional settling tank. But this contribution has not been quantified in an MBR yet. As non-biotic processes compete with biotic transformations, assessing and predicting the biodegradation rate of a specific aromatic compound is finally a critical task. The solids retention time (SRT) is considered as the most suitable design parameter to evaluate micro-pollutant removal in CAS and MBR (Byrns, 2001; Clara et al., 2005; Joss, 2005; Lesjean et al., 2004). As micro-pollutant degradation is generally considered to increase with increasing SRT, MBRs may have an advantage because they can work at higher SRT than CAS for a similar footprint. Additional specificities should be considered in MBRs: aeration with coarse bubbles, hydrodynamic constraints due to liquid circulation, total retention of small particles and accumulation of colloids. Consequently, various amounts of dispersed bacteria and extracellular polymeric substances are generally observed in MBR sludge. These specificities vary with the MBR configuration and operation. Basically, membranes can be submerged in the bioreactor or operated in an external element (side-stream
configuration). In the latter case, the system can be operated with cross-filtration (high liquid velocity is then imposed at the membrane surface) or semi dead-end filtration with submerged membranes (coarse bubble aeration is then used to control fouling). Both systems generate shear stress which reduces the floc size to a greater or lesser extent (Kim et al., 2001; Stricot et al., 2010). This phenomenon reduces mass transfer resistance, which improves the accessibility of bacteria to pollutants and modifies the apparent biokinetic parameters (Fenu et al., 2010). But the relation between sludge structure and biokinetics is complex and still controversial. Most aromatic compounds become toxic at a given concentration and aggregation is also a microbial protection from this. For example, based on short-term experiments Sponza (2002) and Henriques et al. (2005) observed that floc disaggregation increased the inhibition of biomass for a given inhibitor concentration. However, there is still no information on the long-term consequences of disaggregation in MBR as acclimatisation of the microbial population plays a major role in resistance development and biodegradation kinetics for xenobiotic compounds (Rezouga et al., 2009). Consequently, the aim of this study is to evaluate the fate of hazardous aromatic compounds in two MBR configurations, one using a submerged membrane in an external reactor (semi dead-end filtration), which generates low shear stress, the other an external cross-flow filtration unit which generates high shear stress. Specific attention is paid to the contribution of non-biotic phenomena as well as biodegradation and inhibition.
2.
Materials and methods
2.1.
Experimental set-up
Two lab-scale pilots were operated in parallel. Both MBRs provided a permeate rate of 1 L h1 but different configurations led to differences in terms of liquid velocity and aeration. The first membrane bioreactor worked with a cross-flow filtration unit, using a tubular ceramic membrane (inside/out ultra-filtration membrane Novasep/Carbosep 40, total surface area 0.01 m2). It was operated in a side-stream loop with a high liquid velocity (5 m s1) which induced a strong shear stress (72 N m2). A mean filtration flux of 100 L m2 h1 was maintained by imposing a trans-membrane pressure (TMP) of 2 bar. The second MBR was a semi dead-end filtration system using hollow fibres submerged in an external cell (outside/in polysulphone membrane, external diameter 1.4 mm, mean pore diameter 0.1 mm, total surface area 0.1 m2, Polymem). The sludge recirculation flow rate was equal to the reactor feeding flow rate, which induced low liquid velocity and low shear stress (0.07 N m2). Aeration was applied under the membrane bundle to prevent fouling (flow rate of 200 NL h1). The filtration flux was fixed at 10 L m2 h1 by a suction peristaltic pump. The hydrodynamic characterisation of these processes is detailed in Stricot et al. (2010). Both bioreactors, similar in volume (18 L), were continuously fed with synthetic wastewater, the composition of which was inspired by the analysis of a petrochemical wastewater. It was composed of readily biodegradable compounds on the one hand (methanol,
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 5 1 e4 5 6 1
4553
ethanol, acetate, each contributing 1/3 of the easily degradable COD fixed at 2000 mgO2 L1) and aromatic xenobiotic compounds on the other hand: benzene, toluene, ethylbenzene, xylene, 2.4-dichloro-phenol, 3.5-dimethyl-phenol, naphthalene, acenaphthylene, acenaphthene, phenanthrene, anthracene, fluoranthene, benzo(a)pyrene, with respective concentrations of 60.4, 61.3, 24.8, 24.3, 39.4, 23.7, 6.14, 3.82, 1.02, 0.309, 0.107, 0.0587, 0.00192 mg L1. Analyses were focused only on BTEX and PAH removal in this work but all these aromatic compounds (including the two phenol derivates) were included in order to use a synthetic wastewater that was as representative as possible of a real petrochemical wastewater. Concentrations were chosen considering the worst scenario which could be observed in a real petrochemical industry wastewater treatment plant. Only readily biodegradable compounds were fed to the MBRs during a starting period of 18 days. Then aromatic compounds were added during an acclimatising period (days 19e55) and the last period was considered as a steady state period (from day 56 to day 83). During the second and the last periods, the influent total COD reached 2800 mg L1. COD/N/P was fixed at 100/7/1.5 with sodium phosphate and ammonium sulphate. Temperature was controlled at 20 1 C, pH was naturally regulated between 7 and 8. Both processes were operated at the same hydraulic retention time (18 h) and similar sludge retention times (20 days) imposed by a daily sludge wastage (0.9 L) from each reactor. Dissolved oxygen was maintained at 2.5 0.5 mg L1. The aeration flow rate imposed by a fine bubble diffuser in each bioreactor was controlled and measured.
volatile compounds (benzene, toluene, ethylbenzene, xylene, naphthalene, acenaphthylene, acenaphthene) and solid phase micro-extraction (SPME) with 7-mm polydimethylsiloxane (PDMS) SPME fibres for most PAHs (phenanthrene, anthracene, fluoranthene, benzo(a)pyrene). Calibration was performed with deuterated PAHs semivolatile Internal Standard Mix reaching a quantification limit of 0.01 mg L1. Sludge samples were first frozen and dried before automatic solvent extraction was performed with dichloromethane (Soxtec, HT2 Tecator (Foss)) on 1 g of solids. Sludge samples were collected in duplicate. Before starting the study, a non-biotic test was performed by continuous feeding of the filtration systems for one week with synthetic wastewater and sodium diazide. This confirmed that PAH and BTEX were not significantly adsorbed or retained when passing through membranes. Membrane permeability was also measured before and after the test and no significant difference was observed (less than 5%).
2.2.
where C is the concentration of the compound measured in the aqueous phase of the bioreactor (mg L1), CIN is the influent concentration (mg L1), COUT is the outlet concentration or permeate (mg L1), QIN is the influent flow rate (L day1), QOUT is the outlet flow rate or permeate (L day1), QW is the sludge wastage flow rate (L day1), V is the reactor volume (L), rADS is the rate of advection in the sorbed phase (mg L1 day1), rVOL is the rate of volatilisation to the gas phase (mg L1 day1), rDEG is the rate of degradation (mg L1 day1). The volatilisation rate can be obtained from the following expression, similarly to Byrns (2001):
Analytical methods
Daily common measurements (suspended solids, volatile suspended solid, chemical oxygen demand) were performed using the standard method. Soluble protein concentration was determined according to the bicinchoninic acid method (Smith et al., 1985). Soluble polysaccharides were measured by the anthrone method (Dreywood, 1946). Floc size distribution was measured with a laser particle size analyser (Mastersizer 2000, Malvern Instruments S.A.). The resistance of biomass to inhibition was determined through the standard measurement of EC50 with 2.4-dichloro-phenol. It consists of respirometric analysis (OUR measurement) in closed cells with various inhibitor concentrations, EC50 being defined as the concentration that induces a 50% reduction of the microbial respiration. For hazardous aromatic compounds, mass balances were performed during the steady state period. Analysis was triplicated on each sample, and about 10 samples were collected over three weeks so as to obtain a mean value for each compound and each different phase: effluent (membrane permeate), suspended solids of the sludge, and sludge supernatant (centrifuged at 4200 g for 15 min at 20 C). Analyses were performed with a GCeMS analyser (Gas chromatograph (Varian 3900) equipped with a Mass Spectrometer detector (Varian 2100 SaturnT) allowing electron impact and operating in the “Full Scan” acquisition mode). Liquid samples were analysed by the automatic headspace technique using a QHSS40 (Quma Headspace Sampler) for
2.3.
Mass balance and removal rates
Removal rates were calculated on the basis of analysis performed in the liquid and sludge samples. Contributions of non-biotic processes, i.e. volatilisation and sorption, were also estimated. A global mass balance for a given compound in the liquid phase at steady state can be expressed as: QIN QOUT QW $CIN $COUT $C rADS rVOL rDEG ¼ 0 V V V
rVOL ¼
Qair $H$C R$T$V
(1)
(2)
where H is the Henry’s law constant for the compound (Pa m3 mol1), Qair is the air flow rate (NL day1), R is the universal gas constant and T is the absolute temperature. At steady state, a mass balance on the adsorbed compound indicates that the rate of advection in the sorbed phase is equal to the rate of compound wasted with the excess sludge: QW $X$CADS (3) V where X is the suspended solid concentration of wasted sludge (mg L1) and CADS is the amount of compound adsorbed on the sludge phase (mg g1). At steady state, the degradation rate (rDEG) can be deduced from the mass balance in Eq. (1) as all the other terms are known. rADS ¼
4554
3.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 5 1 e4 5 6 1
Results
3.1. Overall process performance and acclimatisation of microbial system The overall performances of the two MBRs are summarised in Table 1, indicating the mean values of parameters for three periods (starting, acclimatising, steady state). Although both MBRs were operated with a similar volumetric loading rate and similar solid retention time, the measured mixed liquor suspended solids (MLSS) were stabilised at slightly different values, i.e. 8 0.5 g L1 in the semi dead-end MBR and 6 0.5 g L1 respectively in the semi dead-end MBR in the cross-flow MBR. This result could support the idea that the high shear stress imposed in the cross-flow MBR generated suspended solid disintegration which reduced the measured sludge production (20e30% lower). However, MLSS measured by the standard method with glass fibre filtration (or even 3000 g common centrifugation) cannot catch all the dispersed bacteria. Consequently, the measured suspended solids are only a fraction of the total biomass production. For this reason, it is probable that less biomass was quantifiable by MLSS in the system with high shear stress, which does not mean that the biomass sludge yield was greatly reduced. The results found in this study (with continuously fed systems) are in accordance with batch tests performed previously by Stricot et al. (2010), who observed a significant reduction of measured MLSS with the same equipment, explained by a transfer of organic matter from the flocculated to the aqueous phase. Clearly, the reduction of sludge production due to shear stress would need more detailed investigation in a dedicated work, but it was not the objective of the present study. Consequently the estimated F:M (food to micro-organism) ratio was slightly higher in the cross-flow MBR than the F:M
ratio stabilised in the dead-end MBR. COD removal efficiency was good in both systems but it was slightly higher in the dead-end MBR (with submerged membranes) compared to the cross-flow system. This difference was not really significant during the first period, when only readily biodegradable substances were received in the systems (98.8% and 99.9%). During the second and third periods, when hazardous aromatic substances were introduced, COD removal efficiency deteriorated significantly in both systems but more significantly in the cross-flow MBR. In both bioreactors, dissolved and colloidal matter (DCM) accumulated in the sludge supernatant as indicated by COD, as did polysaccharides and proteins (Table 1). This accumulation was more significant in the cross-flow MBR as higher shear stress was imposed. As previously shown by Kim et al. (2001) and Stricot et al. (2010), shear stress in cross-flow filtration generated floc disintegration and release of exopolymeric substances. In both systems, release of soluble and colloidal organic matters increased dramatically after introduction of the hazardous aromatic compounds. COD of the supernatant was multiplied by 15 in both MBRs, varying from 700 to 1300 mg L1. A large increase of proteins (multiplied by 10) was measured in the supernatant. This phenomenon can be attributed to the response of the microbial consortium to the inhibiting xenobiotic molecules, i.e. bacteria could excrete additional metabolites in addition to the release of intracellular organic material due to cell lysis. It should be noted that more than 90% of polysaccharides were retained by the membrane but only 50 10% of proteins were retained. The loss of soluble microbial products (i.e. proteins) significantly contributed to the deterioration of effluent quality (Table 1). It should be stressed that COD removal efficiency decreased after the introduction of hazardous substances but progressively improved during the acclimatisation phase.
Table 1 e Mean data collected for each phase on membrane bioreactors (concentration in mg LL1) and standard deviation. Period Cross-flow MBR
Semi dead-end MBR
Inlet COD COD removal Effluent COD outlet Proteins Polysaccharides Sludge supernatant COD Proteins Polysaccharides Sludge EC50 (DCP) Inlet COD COD removal Effluent COD outlet Proteins Polysaccharides Sludge supernatant COD Proteins Polysaccharides Sludge EC50 (DCP)
Average 2604
2604
Starting (0e18 d), 9 samples
Acclimatising (19e55 d), 6 samples
Steady state (56e83 d), 7 samples
2000 98.8%
2804 77%
2728 93%
23.5 18 53 63
566 200 71 24 31
184 90 n.d. n.d.
84 63 20 5 10 5 40 3 2000 99.9%
1330 321 200 50 60 15 225 20 2804 83%
1253 302 n.d. n.d. 400 30 2728 96%
85 4 1.5 54
393 219 58 28 2 1.5
84 95 n.d. n.d.
63 84 84 31 70 5
979 410 150 50 20 5 240 20
710 119 n.d. n.d. 410 31
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 5 1 e4 5 6 1
In the steady state period, removal efficiencies of 93% and 96% were observed in the cross-flow MBR and semi dead-end MBR respectively. Acclimatisation was also clearly demonstrated by respirometric activity, which was monitored in both bioreactors (not shown). 58% and 63% reductions of oxygen uptake rates (OUR) were observed during the 24 h following the feeding of aromatic substances into the semi dead-end MBR and cross-flow MBR respectively. But activities were partly recovered after 3 days and OUR progressively increased until the end of the working period, returning to initial values. As expected, particle size distribution and morphology of biological flocs were very different in the two MBRs (Fig. 1). Properties were similar at the start of the study (both reactors being inoculated with the same activated sludge) but major modifications were observed after two weeks of operation. The high shear stress imposed in the cross-flow filtration system generated notable floc breakage: mean floc size (d0.5) stabilised around 20 mm in this reactor (ranging from 2 to 100 mm). In contrast, mean floc size (d0.5) was around 100 mm in the semi dead-end MBR (ranging from 10 to 1000 mm), which was comparable to the conventional activated sludge process. Microscopic observation indicated that bacterial clusters of 15e20 mm were present in the flocs of the semi dead-end MBR with filamentous bacteria whereas only isolated bacteria were observed in the sludge from cross-flow MBR, embedded in a matrix of exopolymeric substances. Sludge samples were collected at different times in order to characterise the resistance of the microbial system to a given dose of inhibiting substance (2.4-dichloro-phenol, DCP). The concentration of DCP leading to 50% reduction of oxygen uptake rate was then measured (EC50 given in Table 1). During the first period, after 13 days of operation, it was observed that EC50 was lower for the sludge of the cross-flow MBR (40 mg L1) compared to those measured on the sludge from the dead-end MBR (70 mg L1). This result confirmed that short-term floc disaggregation negatively impacted the resistance of the biomass to inhibition, which is thought to be due to less diffusion resistance within the flocs as suggested by Henriques et al. (2005). However, after long-term operation and exposure of biomass to hazardous substances (second and last periods) EC50 increased greatly, reaching a similar value of 400 mg L1 in both MBRs at the end of the experiment. This result indicated impressive acclimatisation of the
a
4555
microbial consortium. Hence, despite the differences in floc morphology, the sensitivity to inhibition progressively became similar in both systems.
3.2.
Aromatic substances concentrations
Average concentrations of aromatic substances measured in the influent, the sludge supernatant and the permeate (outlet) are compared between the two MBRs in Fig. 2. Firstly, it should be stressed that the removal efficiencies were very high in both membrane bioreactors. Residual concentrations in the treated water (permeate) ranged from 0.1 to 2 mg L1 for fluoranthene, phenanthrene and anthracene, 1e200 mg L1 for low molecular weight PAHs (naphthalene, acenaphthene, acenaphthylene), and 10e200 mg L1 for BTEX. Total removal efficiencies determined from measurement of the inlet and outlet liquid phase varied from 97% to 99.99% for the semi dead-end MBR and ranged from 90% to 99.9% in the cross-flow MBR (Table 2). Benzo(a)pyrene was not detected in sample analyses (<0.01 mg L1, <0.5 mg kg1) and the mass balance has not been performed for this compound. The data from the two MBRs are compared in Fig. 3. Concentrations in the effluent (Fig. 3a) were slightly lower in the semi dead-end MBR than in the cross-flow MBR for the most volatile molecules (BTEX, naphthalene, acenaphthene, acenaphthylene). In contrast, for other PAHs, outlet concentrations were similar (phenanthrene, anthracene) or even lower (fluoranthene) in the cross-flow MBR. Surprisingly, the inverse trend was observed for the bioreactor supernatant (Fig. 3b), where the concentrations of the most hydrophobic compounds (phenanthrene, anthracene, fluoranthene) were higher in the cross-flow MBR whereas the concentrations of the most volatile compounds (BTEX, naphthalene, acenaphtene, acenaphtylene) were similar in both MBRs. Concerning concentrations measured in the sludge suspended solids (Fig. 3c), only PAH concentrations were detected, which is logical as BTEX are not sufficiently hydrophobic to adsorb significantly on suspended solids. Except for naphthalene, Fig. 3c indicates that the concentrations of PAHs sorbed on suspended solids were systematically half as high in the sludge extracted from the cross-flow MBR compared to the dead-end MBR. These results demonstrate the importance of overall mass balance to determine the fate of these compounds. Without a global mass balance, very different
b
Fig. 1 e Comparison of floc size evolution in cross-flow MBR (a) and dead-end MBR (b).
4556
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 5 1 e4 5 6 1
10000
1000
1000 µg/L
O U
FL
E
T
BE N TO L
AN
FL U O
E
T AN
E
PH
Y
AC
AC
L
AP N
XY
BE
ET H
0.01 N
0.1
0.01
PH
1
0.1
AC
10
1
E
100
10
L N AP AC Y
100
influent sludge supernatant permeate
100000
10000
TO L
µg/L
100000
b 1000000
ET H
influent sludge supernatant permeate
XY
a 1000000
Dead-end
Crossflow
Fig. 2 e Concentration of hazardous substances in the influent, sludge supernatant, and permeate (outlet) for the cross-flow MBR (a) and the semi dead-end MBR (b).
conclusions could be drawn from aromatic compound analysis in the different phases. As shown by Fig. 2, most of the hazardous molecule removal took place in the bioreactor. Although the overall contribution of filtration to compound removal was small, membrane filtration regularly caused a decrease in compound concentration in the system effluents as well as in COD (Fig. 4). For example, the filtration step seems to have played a statistically significant role in compound removal for anthracene in the cross-flow MBR, and naphthalene and acenaphthene in the dead-end system (Fig. 2). The decrease in compound concentration may be related to the decrease in effluent COD as a result of membrane filtration, but also to volatilisation in the case of the dead-end system (aerated membranes). When compound concentrations in the bioreactor supernatant and in the permeate are compared, aromatic compounds show a mean decrease of 80% through a submerged membrane module (dead-end MBR), whereas a reduction of 37% is observed in the cross-flow module (Fig. 4a). The high rejection in the submerged membrane module is especially observed for the most volatile compounds, which can be explained by the aeration of the module (to prevent fouling) playing a significant role in total volatilisation rate in this MBR. For the most hydrophobic molecules, which are also the least volatile, removal in the membrane units was more variable (from 10 to
50%). As PAHs were not significantly retained by clean membrane filtration, it is probable that PAH sorption on dissolved or colloidal matter (DCM), as well as organic deposit on the membrane, can explain this retention in MBR (definition of the dynamic membrane). As shown in Fig. 4b, similar retention efficiency was observed on dissolved and colloidal COD in both membrane units (about 40%), as was comparable retention of supernatant proteins (about 50%). Consequently, the better COD removal efficiency in the dead-end submerged MBR may be related not only to volatilisation but also to the build up of a fouling layer on the membranes. With less shear and more biomass, there would be a thicker fouling layer on the dead-end MBR. This layer may include a significant amount of biomass, contributing to degradation, and also serve as a “dynamic membrane”, potentially enhancing removals across the membrane unit.
3.3.
Aromatic compound mass balances
Sorption removal rate (rADS), volatilisation rate (rVOL), degradation rate (rDEG) and residual effluent rate (QOUT$COUT/V) are presented in Fig. 5 for each substance. Data are given simultaneously in mass load and in percentage of inlet load. Fig. 4 has already shown that the amount of PAHs was higher in the solids from the semi dead-end MBR than those
Table 2 e Removal efficiency and residual percentage. MW (g mol1)
Benzene Toluene Ethylbenzene Xylene Naphthalene Acenaphthylene Acenaphthene Phenanthrene Anthracene Fluoranthene
BEN TOL ETH XYL NAP ACY ACE PHE ANT FLUO
78 92 106 106 128 152 154 178 178 202
Log Kow
2.13 2.65 3.15 3.15 3.37 4.07 4.33 4.46 4.45 5.33
Log H
2.75 2.83 2.90 2.89 1.69 1.05 1.20 0.94 0.82 0.20
Dead-end
Cross-flow
Removed from liquid
Degraded
Removed from liquid
Degraded
99.95% 99.96% 99.95% 99.91% 99.99% 97.20% 99.84% 99.20% 99.91% 99.68%
70.4% 55.3% 92.9% 60.1% 64.5% 79.1% 61.7% 74.0% 95.9% 44.3%
99.62% 99.74% 99.84% 99.76% 98.63% 90.61% 90.90% 99.16% 99.92% 99.83%
68.0% 63.1% 88.9% 75.6% 87.9% 80.1% 70.2% 91.0% 98.5% 86.8%
4557
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 5 1 e4 5 6 1
100 10
BEN TOL ETH XYL NAP ACY ACE PHE ANT FLU
1 0.1 0.01 0.01
0.1
1
10
100
c
1000 100 BEN TOL ETH XYL NAP ACY ACE PHE ANT FLU
10 1 0.1 0.01 0.01
1000
C ADS cross-flow (µg/g)
b
1000
C cross-flow (µg/l)
COUT cross-flow (µg/L)
a
0.1
1
10
100
NAP ACY ACE PHE ANT FLU
400 300 200 100 0
1000
C dead-end (µg/L)
COUT dead-end (µg/L)
500
0
100
200
300
400
500
CADS dead-end (µg/g)
Fig. 3 e Comparison of residual concentration of hazardous aromatic substances for cross-flow and dead-end MBR (average values), (a): measured in the effluent, (b): measured in the liquid phase of the bioreactor, (c): measured in the solid phase of the bioreactor.
from the cross-flow MBR. Consequently, extraction of PAHs by sludge wastage was much lower in the cross-flow MBR (Fig. 5a). It accounted for 0.5e55% of PAH removal in the semi dead MBR but only 0.5e13% of PAH removal in the cross-flow MBR, depending on the molecule. It was negligible for monoaromatic compounds. Fig. 5b indicates that volatilisation was responsible for 10e45% of most of aromatic compound removal and was negligible for high molecular weight PAHs (phenanthrene, anthracene, fluoranthene). Highest values of rVOL were observed for benzene, toluene, xylene, and naphthalene. In addition, for most molecules, the volatilisation rate was higher in the dead-end MBR (working with submerged membrane) than in the cross-flow MBR. In the dead-end system, volatilisation was due not only to reactor aeration but also to membrane aeration imposed in the filtration unit (air flow rate imposed in bioreactor and membrane unit were comparable). The rate of release of molecules in the effluent (Fig. 5c) was very low for most species, making up less than 0.5% of the inlet loading rate for mono-aromatic compounds (most volatile) and the heaviest PAHs (most hydrophobic). The residual fluxes were slightly higher for acenaphthene, naphthalene, acenaphthylene, and phenanthrene. The highest residual
Discussion
4.1.
Importance of non-biotic processes in MBRs
Protecting the environment from hazardous aromatic compounds needs an integrated evaluation of the liquid, solid
b
500
1500 Cross-flow
Cross-flow
400 COUT (µg/L)
4.
COD permeate (mg/L)
a
percentages were observed in the cross-flow MBR, where about 10% of acenaphthene and acenaphthylene were lost in the effluent, versus 0.17 and 2.9% respectively for the same molecule in the dead-end MBR. The degradation rate of aromatic compounds was estimated with Eq. (1) considering the different losses of molecule in the solid, liquid and gas phases. As a consequence of the lower volatilisation rate and lower transfer by sorption, the degradation rates of aromatic substances were higher in the cross-flow MBR (from 63 to 98.5%) compared to dead-end MBR (from 44 to 95.9%). Finally the results indicate that the deadend MBR showed better results in terms of removal from the liquid phase whereas the cross-flow MBR allowed better degradation of molecules. This result clearly stresses the importance of taking non-biotic and biotic transformation into account in MBRs.
dead end
300 200 y = 0.629x
100
dead end 1000
500
y = 0.201x
0
0
0
100
200
300
C (µg/L)
400
500
0
500 1000 1500 COD sludge supernatant (mg/L)
Fig. 4 e Average concentrations measured in effluent (COUT) versus concentrations measured in sludge supernatant (C). Aromatic substances (a) of the two periods and COD (b) of the three periods.
4558
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 5 1 e4 5 6 1
b
60%
0.1
40%
40
40%
0.05
20%
20
20%
0%
0
0%
Volatilisation
Sorption 1
100%
d
100
100%
40%
40
40%
0.2
20%
20
20%
0%
0
0%
Residual in effluent
BEN
FLUO
ANT
PHE
ACE
ACY
NAP
XYL
ETH
TOL
BEN
0
FLUO
0.4
ANT
60%
PHE
60
ACE
60%
ACY
0.6
NAP
80%
XYL
80
ETH
80%
TOL
0.8
mg/L.d
mg/L.d
c
ACE
FLUO
ANT
PHE
ACY
ACE
NAP
XYL
ETH
TOL
BEN
0
ANT
60
FLUO
60%
PHE
0.15
ACY
80%
NAP
80
XYL
80%
ETH
100%
TOL
100
0.2
mg/L.d
100%
rate dead-end rate Cross-flow % dead-end % cross-flow
BEN
0.25
mg/L.d
a
Degraded
Fig. 5 e Removal by adsorption (a), volatilisation (b), final residual effluent (c), and biodegradation (d).
and gas phases in treatment systems. This aspect has been poorly evaluated in previous MBR studies. In the present work, results confirm that volatilisation and sorption play a significant role in aromatic compound removal. Additionally, the dead-end MBR shows higher volatilisation and sorption rates than the cross-flow MBR. Volatilisation rate was logically linked to the value of Henry’s constant (Fig. 6a). A comparison with the literature indicated that the volatilisation rates obtained in the crossflow MBR were in the same range as those obtained in conventional activated sludge systems (Byrns, 2001, Farhadian et al., 2008). For example, Byrns (2001) estimated that 20e30% volatilisation was obtained for molecules with a Henry’s constant in the range of 100e200 Pa m3 mol1 in CAS. Exactly the same range was obtained in the cross-flow MBR in this study (Fig. 6a). In addition, in the case of dead-end MBR, the air flow imposed through the hollow fibre bundle led to an increase in volatilisation (and a decrease in degradation). This stressed the negative impact of membrane aeration on atmospheric pollution during the treatment of industrial effluent with a high mono-aromatic solvent content as well as low molecular weight PAHs. In the case of the submerged membrane system, the benefit of better removal from the liquid phase is obtained with a detrimental effect on
atmospheric pollution and, finally, a reduction of molecule availability for degradation. In both MBRs, the contribution of adsorption depends on the molecule hydrophobicity (Fig. 6b). Sorption was significant for PAHs with Log Kow higher than 4 and increased proportionally to Log Kow (except for anthracene). This was not surprising as partitioning coefficients (Kp) are commonly correlated to Log Kow (Byrns, 2001, Manoli and Samara, 2008). In the conventional activated sludge process, Byrns (2001) found that the sorption percentage varied from 10 to 50% for PAHs with Log Kow varying from 4 to 5.2. Very similar results were obtained in our study for the dead-end MBR (Fig. 6b). However, the cross-flow MBR showed lower values, indicating that sludge disintegration imposed by shear stress modified either the partitioning correlation (sorption isotherm) or the biodegradation rate. These constitute two important assumptions that need to be investigated. Concerning the first assumption (modification of partition), our results lead us to think that the transfer of organic exopolymers from the solid phase to the supernatant probably modifies the distribution of PAH between the aqueous and solid phases. Moreover, recent works on MBRs show that differences in shear conditions influence the characteristics of the bound EPS, high shear stress generating a decrease of
4559
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 5 1 e4 5 6 1
Dead end
50% removal percentage
b
60%
40% 30% 20% 10% 0%
60% Cross flow
40% 30% 20% 10%
0
1
2
0%
3
Log H L1
Fig. 6 e Removal by volatilisation as a function of Log H in Pa m mol
bound EPS and a modification of the EPS size distribution (McAdam et al., 2011; Menniti et al., 2009). These differences may play a role in differentiating the sorption behaviours observed in the current study. For instance, if compounds are absorbed to EPS in the sludge matrix, the dead-end MBR (low shear stress) is likely to contain a greater fraction of EPS, potentially increasing sorption. As demonstrated by Barret et al. (2010), three compartments can be defined in complex sludge with a high dissolved and colloidal matter (DCM) content: free PAH, PAH adsorbed on DCM, and PAH adsorbed on solids. Sludge disintegration generates an increase of COD in the supernatant (Table 1), which increases the fraction of PAHs adsorbed on DCM. This can also accentuate the reduction of PAHs adsorbed on solid and the increase of aqueous PAHs. It suggests that the three-compartment theory developed in an anaerobic digester for PAHs can probably be applied to membrane bioreactors. This is an important perspective of this study.
Role of bio-availability on degradation rate in MBR
Degradation of hazardous aromatic compounds in bioprocesses depends on their bio-availability. Bio-availability decreases as transfer rate to the gas phase increases. Additionally, bio-availability may decrease as transfer to the solid phase increases, even if there is still some uncertainty about whether adsorption to bacterial surfaces hinders or aids biodegradation (Barret et al., 2010; Stringfellow and AlvarezCohen, 1999). These paradigms are probably valuable for most aerobic bio-processes including MBR, even though the exact laws governing the relation between accessibility and biodegradation rates are still controversial. Specific degradation rate rDEG/X (where X is the MLSS concentration) is plotted against molecular weight of pollutants in Fig. 7a and against aqueous concentration of pollutant measured in the bioreactor in Fig. 7b. Degradation rates decrease with increasing molecular weight, as already observed in an anaerobic reactor by Barret et al. (2010). In addition, degradation rates were systematically higher in the cross-flow MBR compared to the dead-end submerged MBR. This improvement of biodegradation can be explained
2
3
4
5
6
Log Kow 3
4.2.
Dead end
50%
Cross flow
removal percentage
a
(a). Removal by sorption as a function of Log Kow (b).
by: (1) an increase in pollutant bio-availability due to lower transfer rates to the gas and solid phases and (2) reduction of transfer limitation due to biomass dispersion. The degradation rate could also be expected to be influenced by the active biomass concentration. Whereas the cross-flow MBR had 25% less MLSS than the dead-end MBR, a similar oxygen uptake rate was continuously measured in both reactors. This means that either the concentration of active bacteria was probably not very different in the two reactors, whereas less inert suspended matter accumulated in the cross-flow MBR (notably bound exopolymeric substances) or that the specific activity of bacteria was much higher in cross-flow, perhaps due to less transfer limitation. As non-biotic phenomena were lower in the cross-flow MBR, it could be assumed that the increase in biodegradation rate partly resulted from an increase in molecule availability. Fig. 7b shows that concentration of pollutant in the bioreactor and degradation rates are simultaneously higher in the crossflow MBR whatever the aromatic compound. Assuming that bacteria degradation rate is related to the pollutant concentration by first order kinetics, it seems logical that an increase in dissolved concentration should lead to an increased degradation rate. Sorption of PAH is commonly considered as a limiting factor for biodegradation but the exact effect of sorption on biodegradation is still unclear. PAH removal rate is then taken to be related to the aqueous fraction of PAH (Barret et al., 2010), which is increased by the presence of dissolved and colloidal matter. As shown by Stringfellow and AlvarezCohen (1999), sorption can decrease the rate of PAH biodegradation in the short-term but it can also result in the increase of PAH retention in the treatment system, where it may be ultimately biodegraded. In this study, as results suggest a better accessibility of pollutant to bacteria in the cross-flow MBR, it is possible that PAHs were less adsorbed on inert biomass or bound EPS, and so a smaller quantity of adsorbed PAHs accumulated in the sludge. A specificity of the MBR compared to conventional activated sludge is that membrane retention can increase the retention time of some molecules in the bioreactor. Our results demonstrate that about 40% of PAHs, probably linked to polymeric substances, are retained by the membrane. This
4560
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 5 1 e4 5 6 1
100 10
rDEG /X (mg/g.d)
b
cross-flow dead end
rDEG /X (mg/g.d)
a
1 0.1 0.01
100 10
cross-flow dead end
1 0.1 0.01
0.001 50
100
150
200
250
0.001 0.001
M (g/moL)
0.1
10
1000
C (µg/L)
Fig. 7 e Specific degradation rate for each substance rDEG/X (mg gSSL1 dL1), according to: (a) its molecular mass, (b) its aqueous concentration.
results in a higher retention time than in the conventional activated sludge process and probably in higher degradation efficiency. This confirms the benefit of MBRs for improving the removal of hydrophobic micro-pollutants. In addition, pollutant accessibility for bacteria is increased thanks to biomass dispersion. This process reduces the transfer resistance into bio-aggregates, which controls biokinetic parameters, especially at low concentration (Fenu et al., 2010). Consequently, another explanation of the positive results in the cross-flow MBR could be that sludge disintegration generated by hydrodynamic constraints in the cross-flow MBR led to an increase in the specific degradation rate of PAHs. This idea was confirmed by the higher specific oxygen uptake rate measured in the disintegrated sludge. In parallel, respirometric experiments showed that long-term operation with disintegrated sludge did not penalise the acclimatisation or the resistance of biomass. In opposition with short-term analysis, which confirmed a rapid decrease of EC50 as previously shown by Henriques et al. (2005), long-term results indicated that the EC50 increased similarly in both types of process after the acclimatisation phase. For all these reasons, it can be considered that biomass disintegration in MBR can play a synergetic role in improving the biodegradation of hazardous aromatic compounds.
concentration of PAH in the aqueous phase (free or adsorbed on DCM). Whereas shear stress transiently reduced the resistance of biomass to inhibition, no difference was observed between the two MBRs in terms of sensitivity to an inhibiting compound after acclimatisation. Sorption of hydrophobic compounds (PAH) on suspended solid was less significant in disintegrated sludge (cross-flow MBR) whereas PAHs preferentially accumulated in the sludge of the dead-end MBR (with low shear stress). This result suggests that sludge disaggregation in MBR could improve the bio-availability of hydrophobic micro-pollutants. Hence this work provides important information for the choice of side-stream MBR technology in industrial wastewater treatment. It improves our understanding of the fate of hazardous aromatic substances in membrane bioreactors, suggesting possible ways of optimising MBR performance.
Acknowledgements The authors thank Evrard Mengelle, Gerard Cancel, Mansour Bounouba and Slim Ellouze for their contributions to this work.
references
5.
Conclusion
BTEX and PAH were efficiently eliminated from wastewater during treatment via membrane bioreactor (90e99.9%). From aqueous effluent analysis, it results that the semi dead-end MBR showed slightly better efficiencies than the cross-flow MBR. However, non-biotic processes (volatilisation and sorption) were more significant in this process, and, finally, the degradation rate was higher in the cross-flow MBR. The difference in degradation rates seems to be explained by a higher bio-availability of pollutants. High shear stress (imposed in cross-flow filtration) generates dispersed bacteria, increases the amount of dissolved or colloidal organic matter (microbial products) and reduces flocculated biomass. This reduces the amount of PAHs adsorbed on suspended solid phase and increases the
Barret, M., Carrere, H., Delgadillo, L., Patureau, D., 2010. PAH fate during the anaerobic digestion of contaminated sludge: do bioavailability and/or cometabolism limit their biodegradation? Water Research 44 (13), 3797e3806. Bernhard, M., Mu¨ller, J., Knepper, T.P., 2006. Biodegradation of persistent polar pollutants in wastewater: comparison of an optimised lab-scale membrane bioreactor and activated sludge treatment. Water Research 40 (18), 3419e3428. BREF, 2003. Reference Document on Best Available Techniques in Common Waste Water and Waste Gas Treatment/Management Systems in the Chemical Sector. Integrated Pollution Prevention and Control (IPPC), European Commission. Byrns, G., 2001. The fate of xenobiotic organic compounds in wastewater treatment plants. Water Research 35 (10), 2523e2533. Cemc, C.E.M.C, 2006. STP Model Version 2.11. Cirja, M., Zuehlke, S., Ivashechkin, P., Hollender, J., Schaffer, A., Corvini, P.F.X., 2007. Behavior of two differently radiolabelled
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 5 1 e4 5 6 1
17 alpha-ethinylestradiols continuously applied to a laboratory-scale membrane bioreactor with adapted industrial activated sludge. Water Research 41 (19), 4403e4412. Clara, M., Kreuzinger, N., Strenn, B., Gans, O., Kroiss, H., 2005. The solids retention time e a suitable design parameter to evaluate the capacity of wastewater treatment plants to remove micropollutants. Water Research 39 (1), 97e106. Clark, B., Henry, G.L.H., Mackay, D., 1995. Fugacity analysis and model of organic chemical fate in a sewage treatment plant. Environmental Science & Technology 29 (6), 1488e1494. De Wever, H., Weiss, S., Reemtsma, T., Vereecken, J., Mu¨ller, J., Knepper, T., Ro¨rden, O., Gonzalez, S., Barcelo, D., Dolores Hernando, M., 2007. Comparison of sulfonated and other micropollutants removal in membrane bioreactor and conventional wastewater treatment. Water Research 41 (4), 935e945. Dreywood, R., 1946. Qualitative test for carbohydrate material. Industrial & Engineering Chemistry Analytical Edition 18 (8), 499. Farhadian, M., Duchez, D., Vachelard, C., Larroche, C., 2008. Monoaromatics removal from polluted water through bioreactors e a review. Water Research 42 (6e7), 1325e1341. Fenu, A., Guglielmi, G., Jimenez, J., Sperandio, M., Saroj, D., Lesjean, B., Brepols, C., Thoeye, C., Nopens, I., 2010. Activated sludge model (ASM) based modelling of membrane bioreactor (MBR) processes: a critical review with special regard to MBR specificities. Water Research 44 (15), 4272e4294. Henriques, I.D.S., Holbrook, R.D., Kelly, R.T., Love, N.G., 2005. The impact of floc size on respiration inhibition by soluble toxicants e a comparative investigation. Water Research 39 (12), 2559e2568. Joss, A., 2005. Removal of pharmaceuticals and fragrances in biological wastewater treatment. Water Research 39 (14), 3139e3152. Kim, J.S., Lee, C.H., Chang, I.S., 2001. Effect of pump shear on the performance of a crossflow membrane bioreactor. Water Research 35 (9), 2137e2144. Lesjean, B., Gnirss, R.B.H., Keller, S., Tazi-Pain, A., Luck, F., 2004. Outcomes of a 2-Year Investigation on Enhanced Biological Nutrients Removal and Trace Organics Elimination in Membrane Bioreactor (MBRs). IWA Publishing, London, UK. Marrakech, Morocco, 19e24 September.
4561
Manoli, E., Samara, C., 2008. The removal of polycyclic aromatic hydrocarbons in the wastewater treatment process: experimental calculations and model predictions. Environmental Pollution 151 (3), 477e485. McAdam, E.J., Cartmell, E., Judd, S.J., 2011. Comparison of deadend and continuous filtration conditions in a denitrification membrane bioreactor. Journal of Membrane Science 369 (1e2), 167e173. Menniti, A., Kang, S., Elimelech, M., Morgenroth, E., 2009. Influence of shear on the production of extracellular polymeric substances in membrane bioreactors. Water Research 43 (17), 4305e4315. Rezouga, F., Hamdi, M., Sperandio, M., 2009. Variability of kinetic parameters due to biomass acclimation: case of paranitrophenol biodegradation. Bioresource Technology 100 (21), 5021e5029. Schonerklee, M., Peev, M., De Wever, H., Reemtsma, T., Weiss, S., 2009. Modelling the degradation of micropollutants in wastewater: parameter estimation and application to pilot (laboratory-scale) MBR data in the case of 2,6-NDSA and BTSA. Water Science and Technology 59 (1), 149e157. Smith, P.K., Krohn, R.I., Hermanson, G.T., Mallia, A.K., Gartner, F. H., Provenzano, M.D., Fujimoto, E.K., Goeke, N.M., Olson, B.J., Klenk, D.C., 1985. Measurement of protein using bicinchoninic acid. Analytical Biochemistry 150 (1), 76e85. Sponza, D.T., 2002. Extracellular polymer substances and physicochemical properties of flocs in steady- and unsteadystate activated sludge systems. Process Biochemistry 37 (9), 983e998. Stricot, M., Filali, A., Lesage, N., Sperandio, M., Cabassud, C., 2010. Side-stream membrane bioreactors: influence of stress generated by hydrodynamics on floc structure, supernatant quality and fouling propensity. Water Research 44 (7), 2113e2124. Stringfellow, W.T., Alvarez-Cohen, L., 1999. Evaluating the relationship between the sorption of PAHs to bacterial biomass and biodegradation. Water Research 33 (11), 2535e2544. Wang, J., McPhedran, K.N., Seth, R., Drouillard, K.G., 2007. Evaluation of the STP model: comparison of modelled and experimental results for ten polycyclic aromatic hydrocarbons (PAHs). Chemosphere 69 (11), 1802e1806.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 6 2 e4 5 7 0
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Role of mass transfer in overall substrate removal rate in a sequential aerobic sludge blanket reactor treating a non-inhibitory substrate Ju-Sheng Huang*, Chun-Wen Tsao, Yen-Chun Lu, Hsin-Hsien Chou Department of Environmental Engineering, Kun Shan University, Tainan City 710, Taiwan
article info
abstract
Article history:
A laboratory study was undertaken to explore the role of mass transfer in overall substrate
Received 18 January 2011
removal rate and the subsequent kinetic behavior in a glucose-fed sequential aerobic
Received in revised form
sludge blanket (SASB) reactor. At the organic loading rates (OLRs) of 2e8 kg chemical
1 June 2011
oxygen demand (COD)/m3-d, the SASB reactor removed over 98% of COD from wastewater.
Accepted 3 June 2011
With an increase in OLR, the average granule diameter (dp ¼ 1.1e1.9 mm) and the specific
Available online 12 June 2011
oxygen utilization rate increased; whereas biomass density of granules and solids reten-
Keywords:
ated using break-up and intact granules, respectively. The calculated COD removal
Sequential aerobic sludge blanket
efficiencies using the kinetic model (incorporating intrinsic kinetics) and empirical model
reactor
(incorporating apparent kinetics) agreed well with the experimental results, implying that
Granule kinetics
both models can properly describe the overall substrate removal rate in the SASB reactor.
Intra-granular diffusion
By applying the validated kinetic model, the calculated mass transfer parameter values and
Kinetic model
the simulated substrate concentration profiles in the granule showed that the overall
Empirical model
substrate removal rate is intra-granular diffusion controlled. By varying different dp within
tion time decreased (13e32 d). The intrinsic and apparent kinetic parameters were evalu-
a range of 0.1e3.5 mm, the simulated COD removal efficiencies disclosed that the optimal granular size could be no greater than 2.5 mm. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
Anaerobic reactors, including upflow anaerobic sludge bed, expanded granular sludge bed, and anaerobic fluidized bed, are generally regarded as high efficient biological processes because these reactors when treating various kinds of wastewaters can not only maintain high biomass concentrations [20,000e60,000 mg volatile suspended solids (VSS)/L] but also high volumetric loading rates [5e20 kg chemical oxygen demand (COD)/m3-d] (Lettingna and Pol, 1991; Schmidt and Ahring, 1996; Huang et al., 2003; Chou and
Huang, 2005; Chou et al., 2008). Developed about one decade ago, sequential aerobic sludge blanket (SASB) reactors were increasingly drawing interest of researchers in the field of bioprocess engineering. Morgenroth et al. (1997) first demonstrated that aerobic granules can be formed in SASB reactors. Thereafter, many researchers reported that the SASB reactors can effectively treat non-inhibitory substrates (e.g., glucose) and inhibitory substrates (e.g., phenol) at high organic loading rates (OLRs) of 1e15 kg COD/m3-d (Beun et al., 1999, 2000; Etterer and Wilderer, 2001; Tay et al., 2001; Tay et al., 2002; Moy et al., 2002; Tay et al., 2004) and
* Corresponding author. Tel.: þ886 6 2051331. E-mail address: [email protected] (J.-S. Huang). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.06.003
4563
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 6 2 e4 5 7 0
1.4e6 kg COD/m3-d (Jiang et al., 2002, 2004; Tay et al., 2005), respectively. An OLR that can be operated in the SASB reactor is generally 5e10-fold of the conventional activated-sludge reactor, making the SASB reactor a promising, highefficiency technique for biological wastewater treatment. The main results of the afore-mentioned literature are briefly summarized as follows: the aerobic granules formed in SASB reactors generally have a compact microbial structure with granule diameter (dp) of 0.5e4.6 mm, high biomass concentrations of 4e17 g VSS/L, and high specific oxygen utilization rates (SOURs) of 25e144 mg O2/g VSS-h. Till now, studies of SASB reactors were mostly focused on treatment performance (operating OLR and COD removal efficiency), microbial activity (expressed in SOUR) and granule characteristics (biomass concentration, biomass density, and granule size). Nonetheless, studies on the role of mass transfer in the overall substrate removal rate as well as the kinetic behavior in substrate degradation process in the SASB reactor are rather limited. The biomass granule is generally treated as a biocatalyst when a biofilm model is formulated. The substrate is first transferred from the bulk liquid via the diffusion layer to the surface of the granule (i.e., external mass transfer), followed by successive intra-granular mass transfer and biochemical reaction within the granule (Wu and Hickey, 1997). To explore the role of external and intra-granular mass transfer in the overall substrate removal rate in the SASB reactor, four mass transfer parameters can be used, including Thiele modulus (4), Biot number (Bi), diffusion layer thickness (L), and overall effectiveness factor (h) (Grady et al., 1999; Chou and Huang, 2005). In general, if 4 is less than 0.3 and h approaches to unity, the overall substrate removal rate is reaction-controlled instead of diffusion-controlled (Bailey and Ollis, 1986). Chou and Huang (2005) and Chou et al. (2008) reported that the resistance to intra-granule mass transfer in both upflow anaerobic sludge blanket reactors (dp ¼ 0.9e2.1 mm; 4 ¼ 3.8e5.0; h ¼ 0.75e0.88) and expanded granular sludge bed reactors (dp ¼ 1.4e2.4 mm; 4 ¼ 3.2e6.1; h ¼ 0.64e0.75) should be taken into consideration. Accordingly, the principal objective of this study was to explore the role of external and intra-granular mass transfer in the overall substrate removal rate as well as the subsequent kinetic behavior in the SASB reactor treating a non-inhibitory substrate. Therefore, an SASB reactor was used to treat glucose-based synthetic wastewater (by varying different OLRs ranging from 2 to 8 kg COD/m3-d) to generate experimental data. A kinetic model (incorporating intrinsic kinetics) and an empirical model (incorporating apparent kinetics) for the SASB reactor were also formulated and validated by experiments. Batch experiments were carried out to evaluate intrinsic and apparent kinetic parameters which are essential in model calculation. The validated model was also applied to calculate mass transfer parameters (4, Bi, L, and h), to simulate substrate concentration profiles in the granule, and to simulate the effects of granule size on SASB-reactor performances. Moreover in this article, physical characteristics (biomass concentration, bed porosity, granule diameter, and biomass density), and SOUR of biomass granules are discussed as well.
2.
Model formulation
2.1.
Kinetic model
The following assumptions are made for the formulation of a kinetic model that can be used for simulating variations in substrate concentration with different operating conditions in the SASB reactor: 1. The granule is spheroid-shaped. 2. The microbial growth and detachment rates are in equilibrium. Also, a steady-state granule has no net increase in mass per granule (Rittmann and McCarty, 1980). 3. According to the literature mentioned previously, the aerobic granules have a compact microbial structure with granule diameter (dp) of 0.5e4.6 mm. Thus, the dispersed biomass in the diffusion layer is neglected, and Fick’s law follows. 4. The complete-mix flow regime in the liquid phase of the SASB reactor is assumed to prevail because in the present study a superficial gas velocity of as high as 0.0138e0.0277 m/s did induce rigorous mixing in the reactor, resulting in a slight variation (2e3 mg COD/L) in the measured substrate concentrations along the sludge-bed height. 5. The aerobic degradation rate of a non-inhibitory substrate glucose (expressed in COD) follows Monod kinetics
2.1.1.
Solid phase
When steady state is reached in the SASB reactor, the diffusion rate of substrate from the bulk liquid to the granule equals the utilization rate of glucose. By selecting the halfsaturation constant (Ks) and the granule radius (R) as a substrate characteristic concentration and a characteristic length, respectively, the solid-phase model (Rittmann and McCarty, 1980; Jih et al., 2003) in dimensionless form can be derived, as shown below: d2 Sf 2 dSf S þ ¼ f2 f dr2 r dr 1 þ Sf
(1)
where f2 ¼ kXf R2 =Df Ks
(2)
The boundary conditions for Eq. (1) are dSf dr dSf dr
¼0
at r ¼ 0 ðthe center of granuleÞ
(3)
¼ Bi Sb Ss
at r ¼ 1 ðthe surface of granuleÞ
(4)
where Bi ¼ Dw R=Df L
(5)
The thickness of the diffusion layer (L) in Eq. (5) is a function of hydraulic condition and bed porosity (e) of the SASB reactor (Hines and Maddox, 1985): 2=3 Re0:51 L ¼ 1:1 Dw u1 s Sc
(6)
4564
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 6 2 e4 5 7 0
where
2.3.
us Re ¼ Reynolds number ¼ na n Sc ¼ Schmidt number ¼ Dw a ¼ 6ð1 eÞ=dp
(7) (8) (9)
To account for the influence of external and intra-granular mass transfer resistances on the overall substrate removal rate, the overall effectiveness factor for substrate uptake (h) is defined by the ratio of the observed reaction rate for substrate removal to the reaction rate which would be obtained without external and intra-granular mass transfer resistances (Fink et al., 1973; Grady et al., 1999; Jih et al., 2003): Z1 f2
1þ
0
h¼
Z1 f
2 0
2.1.2.
Sf Sf
r2 dr
Sb 2 r dr 1 þ Sb
3 1 þ Sb dSf ¼ dr r ¼1 f2 Sb
2.2.
From an engineering perspective, it should be more acceptable for environmental engineers if the kinetic model can be replaced with an empirical model that only contains an algebraic equation. Accordingly, apparent kinetic parameters were evaluated by using intact granules removed from the steady-state SASB reactor. Referring to Lawrence and McCarty (1969), an empirical model can be formulated, based on a material-balance equation for substrate entering and leaving the complete-mix SASB reactor: QðSi Sb Þ k0 Sb ¼ 0 Ks þ Sb Mx
(12)
The validity of the empirical model [Eq. (12)] is based on the following assumptions: (10)
Liquid-phase
Steady-state material balance for the substrate in the complete-mix SASB reactor gives QðSi Sb Þ kSb ¼h Ks þ Sb Mx
Empirical model
(11)
1. The flow regimes in the liquid phase of the SASB reactor approach complete mix. 2. Biomass measured in the sludge-bed zone represents total biomass in the SASB reactor. 3. The aerobic degradation rate of glucose (expressed in COD) follows Monod kinetics. To calculate Sb using the empirical model, only a set of data including Q, Si, Mx, and apparent kinetic parameter values (k0 and Ks 0 ) needs to be input.
Model calculation
Details of calculations for h using the solid-phase model are described as follows: 1. A set of data including substrate concentration in bulk liquid (Sb), R, biological and physical parameter values (k, Ks, Dw, and Df), biomass density (Xf), dynamic viscosity in bulk liquid (n), superficial velocity (us), and e is input. 2. L is calculated using Eq. (6). 3. Together with the fourth-order RungeeKutta method and backward shooting algorithm (Press et al., 1986), the governing continuity equation [Eq. (1)] and boundary conditions [Eqs. (3) and (4)] are used to calculate the dimensionless substrate concentration within the granule ðSf Þ (i.e., from the center to the surface of the granule layer by layer). 4. Calculate h using Eq. (10). Details of calculations for Sb using the liquid-phase model are described as follows: 1. A set of data including inflow rate (Q), influent substrate concentration (Si), biological parameter values (k and Ks), and biomass in the sludge-bed zone of the SASB reactor (Mx) is input. 2. Within the range of 0.001e50% of Si, 51 sets of Sb with an equal interval are set. The solid-phase model is used to calculate h for each Sb to generate a mass transfer function relating h to Sb. The h in Eq. (11) is then substituted for this mass transfer function, resulting in a nonlinear algebraic equation with a single variable Sb. Thereafter, the bisection iteration method is used to calculate Sb using Eq. (11).
3.
Materials and methods
3.1.
Bioreactor operation
A Plexiglas SASB reactor with six-equal distance sampling ports along the reactor height was used, as shown in Fig. 1. The SASB reactor had dimensions of 6.0 (length) 6.0 (width) 105 cm (height) and a reactor volume (VR) of 3.8 L. The SASB reactor was inoculated with approximately 0.7 L of aerobic sludge taken from a secondary clarifier of a local fullscale wastewater treatment plant treating domestic sewage. The inoculated aerobic sludge had the following characteristics: suspended solids (SS), 13,000 mg/L; volatile suspended solids (VSS), 7000 mg/L; and specific oxygen utilization rate (SOUR), 40 mg O2/g VSS-h. The SASB reactor was fed with a glucose-based synthetic wastewater solution, as shown in Table 1. The organic loading rates of the SASB reactor were respectively maintained at 2, 4, 6 and 8 kg COD/m3-d (test runs 1e4) by controlling the same inflow rate of 11.3 L/d but with an increase of influent COD concentration from 667 up to 2668 mg COD/L. Referring to Beun et al. (1999) and Jiang et al. (2002), the SASB reactor was operated sequentially in 4-h cycles (2 min of feeding, 231 min of aeration, 5 min of settling, and 2 min of discharging). A sequential operation of the SASB reactor (i.e., feeding, aeration, settling, and discharging) was automatically controlled by timers together with a motor valve and a solenoid valve. A mass flow controller system was used to keep airflow constant. During the period of aeration, the dissolved oxygen (DO) concentration and the superficial gas velocity were respectively
4565
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 6 2 e4 5 7 0
Fig. 1 e A schematic diagram of SASB reactor. maintained at greater than 5 mg/L and 0.0138e0.0277 m/s by introducing air at the bottom of the reactor using a fine bubble aerator. Effluent was discharged at the middle part of the reactor (a volumetric exchange ratio of 50%). The temperature was maintained at 28 1 C by using an outer-water jacket and a thermostat bath; the pH was maintained at 7.5 0.2 by adding alkali NaHCO3. Prior to carrying out the first test run, seed sludge in the SASB reactor had been acclimated for a three-month period of operation. To ensure that the SASB reactor reached steady state, the reactor in each test run was continuous for at least two months. COD and SS in the effluent and VSS in the reactor were monitored. Steady state was assumed to be reached after test results of the three parameters were within 10% deviation for three consecutive samples (sampling three times weekly).
3.2. Evaluation of intrinsic and apparent kinetic parameters Dispersed sludge and intact granules were used to evaluate intrinsic kinetic parameters (for kinetic-model calculation) and apparent kinetic parameters (for empirical-model
calculation), respectively. Dispersed sludge was prepared by placing the intact granules (removed from near half height of the sludge-bed zone of the SASB reactor) together with a few glass beads in a serum vial, followed by placing the serum vial on a mechanical shaker (28 1 C; 80e100 rpm) for over 1 h to break up granules. The size distribution of the obtained dispersed sludge was determined using a laser scattering particle size distribution analyzer (Horiba, model LA-920, Japan). More than 95% of the particles were found within the colloidal size range (0.26e101 mm). Thus, the resistance to external mass transfer and diffusion within tiny microbial cells in batch culture was considered to be negligible. Batch reactors (1 L serum vial) were used to determine the intrinsic and apparent kinetic parameters (k, Ks; k0 , Ks 0 ). The serum vial was added with an adequate amount of dispersed sludge or intact granules (approximately 334e580 mg VSS/L) together with synthetic wastewater containing glucose (approximately initial concentrations of 176e464 mg COD/L), nitrogen and phosphorus nutrients (COD: N: P 100: 5: 1), trace elements (Table 1), and a phosphate buffer solution (pH ¼ 7.5 0.1). In the serum vial the DO concentration was maintained at greater than 5 mg/L using a fine bubble diffuser; the temperature was maintained at 28 1 C using water bath. During the batch experiments, samples were analyzed for the initial biomass concentration in the serum vial and for COD remaining in the solution for every 30 min. Thereafter, nonlinear regression with the LevenbergeMarquardt algorithm (Press et al., 1986) was applied to search for a set of kinetic parameters which would fit in experimental data.
3.3.
Analytical procedures
Approximately 5e10 mL of granules was randomly removed from the lower-part, middle-part, and upper-part of the sludge-bed zone of the SASB reactor. Then the granule diameter was determined using the method of image analysis, as previously described (Chou et al., 2004). Biomass density is defined as the density of granules in the sludge-bed zone after the exclusion of water in between granules. The procedures for the determination of biomass density are briefly described as follows. The granules were first placed on filter paper underneath facial tissue with water-absorption ability was placed beforehand. After taking water away from the granules, the water displacement method was used to measure the volume of granules. Then, biomass density was calculated by dividing the biomass weight (expressed in VSS) by the volume of granules.
Table 1 e Chemical composition of glucose-based synthetic wastewater (diluted with tap water). Test run
1 2 3 4
Substrate as COD
NH4Cl
KH2PO4
FeSO4
CaCl2
Alkali NaHCO3
(mg/L)
(mg/L)
(mg/L)
(mg/L)
(mg/L)
(mg CaCO3/L)
667 1334 2000 2668
127 255 382 510
29 58 88 117
17 33 50 66
60 60 60 60
400 400 400 400
Trace metals: Ni2þ, 0.5; Fe3þ, 0.5; Co2þ, 0.3; Mo6þ, 0.6; Zn2þ, 0.5; Mn2þ, 0.5 mg/L.
4566
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 6 2 e4 5 7 0
Table 2 e Operating conditions and performance of SASB reactor.a
1 2 3 4 a b c d e f
Biomass Xic Weighted Mean
OLRb
Test Influent run COD
3
(mg/L)
(kg COD/m -d)
(mg VSS/L)
667 1334 2000 2668
2 4 6 8
7900 8660 9430 10950
Effluent COD
SS
(g VSS) (mg/L) (mg/L) 30 33 36 41
13 13 12 17
82 122 155 290
COD qcd Removal
SSURe
SOURf
(%)
(d) (mg COD/mg VSS-d) (mg O2/g VSS-h)
98 98 99 99
32 24 20 13
0.25 0.46 0.63 0.73
62 73 81 96
Operating temperature ¼ 28 1 C, VR ¼ 3.8 L, Q ¼ 11.3 L/d, HRT ¼ 8 h, Cycle time ¼ 240 min. OLR ¼ organic loading rate. Xi ¼ biomass concentration. qc ¼ solids retention time. SSUR ¼ specific substrate utilization rate. SOUR ¼ specific oxygen utilization rate.
COD, SS, VSS, and SOUR were analyzed according to Standard Methods (APHA, 1995).
4.
Results and discussion
4.1.
Bioreactor performance
The operating conditions and main results of the glucose-fed SASB reactor are presented in Table 2. The COD removal efficiency reached over 98% when the OLR of the SASB reactor was increased from 2, 4, 6, to 8 kg COD/m3-d. The obtained biomass concentration and the specific substrate utilization rate were 7900e10,950 mg VSS/L and 0.25e0.73 mg COD/mg VSS-d, respectively. The effluent SS concentration in the SASB reactor (82e290 mg/L) increased with increasing OLR, resulting in a decrease in solids retention time (SRT ¼ 13e32 d). Similarly, Tay et al. (2004) reported that the SASB reactor when operated at an OLR of 4 kg COD/m3-d (SRT ¼ 31 d) for treating a mixture of glucose, peptone and meat extract can maintain stable granules, reaching a constant biomass concentration of 12,000 mg VSS/L; whereas the reactor when operated at an OLR of 8 kg COD/m3-d (SRT declined to1.5 d) could not maintain stable granules, resulting in fluctuations of biomass
concentration (1500e6300 mg VSS/L) and washout of granules from the reactor. Accordingly, a proper OLR should be maintained in SASB reactors to give stable granules and a low effluent SS concentration. In the present study, the COD concentration monitored in the effluent of the SASB reactor was quite low (12e17 mg/L). Thus prior to carrying out batch experiments for the measurement of SOUR, glucose was supplemented to make an initial COD concentration of approximately 30 mg/L. As shown in Table 2, the SOUR values (62e96 mg O2/g VSS-h) increased with increasing OLR because a higher OLR gave a shorter SRT. The measured SOUR values in the present study are close to that (69 mg O2/g VSS-h) reported by Tay et al. (2002).
4.2.
Granule physical characteristics
The granule characteristics (granule diameter, biomass density, bed porosity and granule’s specific gravity) of the glucose-fed SASB reactor are presented in Table 3. With an increase in OLR, the average granule diameter and the average granule’s specific gravity increased (dp ¼ 1.1e1.9 mm; 1.05e1.07) whereas the average biomass density of granules and bed porosity decreased (Xf ¼ 53500e63100 mg VSS/L; e ¼ 0.80e0.87). This implied that a lower OLR tended to result
Table 3 e Granule characteristics of SASB reactor. Test run
OLR
Xfa Weighted mean
1 2 3 4
Upper
(kg COD /m3-d)
(mg VSS/L)
()
2 4 6 8
63100 58600 57200 53500
0.87 0.85 0.84 0.80
a Xf ¼ Biomass density of granules. b Porosity ¼ 1e (Xi/Xf) c dp ¼ average diameter of granules.
dpc
Porosityb Middle
Granule’s sp. gr
Lower
Weighted mean
Upper
Middle
(mm) 1.0 1.1 1.3 1.4
1.1 1.4 1.8 2.0
1.2 1.9 2.5 2.5
Lower
Weighted mean
() 1.1 1.5 1.8 1.9
1.04 1.06 1.06 1.07
1.05 1.07 1.07 1.07
1.05 1.07 1.07 1.07
1.05 1.07 1.07 1.07
4567
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 6 2 e4 5 7 0
40
OLR = 4 kg COD/m3-d
run 2) in the lower-part, middle-part, and upper-part of the sludge-bed zone in the SASB reactor is presented in Fig. 2. Over 67% of granules (based on volume) in the SASB reactor fell in granule size distribution ranging from 1.0 to 3.0 mm; larger sizes of granules tended to be retained in the lower-part of the sludge bed zone, which is similar to the results of upflow anaerobic sludge bed and expanded granular sludge bed reactors (Huang et al., 2003; Chou et al., 2008). The bed porosity achieved in the present study was similar to that (e ¼ 0.86) reported by Tay et al. (2001), indicating that a channeling effect would not occur in the SASB reactor.
Upper
30
20
10
40 0
% of biomass volume
Middle 30
4.3.
20
10
0 40
Lower 30
20
10
0
0
1
2
3
4
Granule diameter (mm) Fig. 2 e Granule size distribution in upper-part, middlepart, and lower-part of SASB reactor.
in a small but compact granule, whereas a higher OLR tended to give a large but fluffy granule. The biomass densities of granules were relatively high compared with that (Xf ¼ 41100 mg VSS/L) reported by Tay et al. (2001) because the SRTs in the present study (13e32 d) were longer than that in their study (SRT ¼ 7 d). The dp values obtained in the present study fell in a typical range of 1.0e4.6 mm (Etterer and Wilderer, 2001; Tay et al., 2001, 2002; Moy et al., 2002). A typical case of granule size distribution (0.5e4.0 mm; test
Intrinsic and apparent kinetic parameters
Monod kinetics is often used to describe the degradation rate of a non-inhibitory substrate. The intrinsic kinetic parameters were determined using the break-up granules (dispersed sludge); whereas the apparent kinetic parameters were determined using the intact granules to adequately represent the real granule characteristics. As shown in Table 4, the intrinsic parameters k0 and Ks0 were 12.4e23.9 mg COD/mg VSS-d; 77e92 mg COD/L, respectively; whereas the apparent parameters k0 and Ks0 were 8.9e14.2 mg COD/mg VSS-d; 100e140 mg COD/L, respectively. In addition, by varying three different initial COD concentrations (test run 4), the obtained intrinsic parameters k (23.8 0.1 mg COD/mg VSS-d), Ks (90 2 mg COD/L), k0 (13.9 0.4 mg COD/mg VSS-d) and KS0 (137 5 mg COD/L) varied slightly. At the OLR of 2e8 kg COD/ m3-d, the k value (i.e., without intra-granular mass transfer resistance) was significantly higher than the k0 value (i.e., with intra-granular mass transfer resistance). Both k and k0 increased with increasing OLR because an increasing OLR gave a short SRT (i.e., an increase in microbial activity). With an increase in OLR, Ks varied slightly, whereas KS0 increased markedly. This can be explained by that an increasing OLR gave a larger granule, resulting in a decrease in contact area for the affinity of substrate to the granule. Similar results were also reported by Wu and Hickey (1997) and Chou and Huang (2005).
4.4.
Model simulation and validation
In the present study, the measured biomass density (Xf) and granule size (dp) along the sludge-bed height varied. If the
Table 4 e Intrinsic and apparent kinetic parameter values. Test run
qc
OLR
Batch test Intrinsic
3
(kg COD/m -d)
1 2 3 4 4 4
2 4 6 8 8 8
Intrinsic
Apparent
Intrinsic
Apparent 0
Biomass
Initial COD
k
Ks
k0
Ks
(mg VSS/L)
(mg COD/L)
(d1)
(mg/L)
(d1)
(mg/L)
12.4 14.8 16.7 23.7 23.9 23.7
82 77 90 92 90 88
8.9 11.3 12.7 13.9 13.5 14.2
100 107 126 140 131 140
(d)
32 24 20 13 13 13
Apparent
Kinetic parameters
385 512 384 476 461 385
560 580 510 462 431 402
410 464 325 420 248 176
400 408 445 408 256 188
4568
variations in substrate degradation rate, Xf and dp along the sludge-bed height are incorporated to model simulation, the simulated substrate concentrations in the liquid phase of the SASB reactor will not be homogeneous. Thus, the weightedmean values of Xf and dp (Table 3) were used in model simulation. The rest of biological and physical parameter values used in model calculation are as follows: k, Ks, k0 , and KS0 ¼ variable (Table 4), e ¼ variable (Table 3), Mx ¼ variable (Table 2), Dw ¼ 3.3 105 m2/d (Perry and Chilton, 1973), Df ¼ 2.6 105 m2/d (Williamson and McCarty, 1976), and n ¼ 8.48 107 m2/s (water at 28 C). By using the kinetic and the empirical models together with biological and physical parameter values, the residual substrate concentrations in the SASB reactor (Sb, expressed in COD) were calculated. As shown in Table 5, the calculated COD removal efficiencies using the kinetic and empirical models were all in good agreement with the experimental results (within 1.5% deviations), implying that the two respective models can properly describe the overall substrate removal rate in the SASB reactors. Moreover, the relative percentage deviation (RPD) of COD removal efficiencies for kinetic-model and empiricalmodel simulation only fell in a small range of 0.8e1.0%. Thus in engineering practice, the application of the empirical model for process design of SASB reactors should be acceptable. The success of the empirical model for the simulation of SASBreactor performances may be attributed to the following two factors: the flow regimes in the liquid phase of the SASB reactor approached complete mix; the intact granules that can adequately represent the real granule characteristics in the sludge-bed zone of the SASB reactor were used to evaluate apparent kinetic parameters.
4.5.
Mass transfer versus overall substrate removal rate
By using the validated kinetic model, the mass transfer parameters [4, Bi, L, and h; Eq. (2), Eq. (5), Eq. (6), and Eq. (7)] were calculated and the substrate concentration profiles in 1.00 0.75
1.00 OLR = 2 kg COD/m 3-d
0.75
0.50
0.50
0.25
0.25
0.00 0.75 Sf/Sb (-)
a R ¼ COD removal efficiency. b Deviation of removal efficiency ¼ (calculated COD removal efficiency e experimental COD removal efficiency)/experimental COD removal efficiency. c RPD (relative percentage deviation) for model simulation ¼ absolute value of the difference (of COD removal efficiencies respectively calculated using kinetic and empirical models) divided by their average value.
0.02 0.02 0.02 0.02 0.20 0.19 0.16 0.12 27 49 55 64 107 216 326 504 0.9 0.8 1.0 0.9 1.5 0.6 0.3 0.3 3 5 7 8 9 16 27 31 98.1 99.0 99.4 99.4 1 2 3 4
667 1334 2000 2668
2 4 6 8
13 13 12 17
(%) (mg COD/L) (mg/L)
(kg COD/m3-d) (mg COD/L)
(%)
98.7 98.8 98.7 98.8
0.6 0.2 0.8 0.5
99.6 99.6 99.7 99.7
(%) (%) (mg COD/L) (%)
(%)
Deviation of removal efficiencyb Ra Sb Deviation of removal efficiencya Ra Sb Ra Sb
Calculated (Kinetic model) Experimental OLR Test Influent run COD
Table 5 e Model verification results and calculated mass transfer parameter values.
Calculated (Empirical model)
RPD for model simulation c
42
Bi
() ()
h
()
L
(mm)
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 6 2 e4 5 7 0
0.00 OLR = 4 kg COD/m 3-d
0.75
0.50
0.50
0.25
0.25
0.00 0.75
0.00 OLR = 6 kg COD/m 3-d
0.75
0.50
0.50
0.25
0.25
0.00 0.75
0.00 3
OLR = 8 kg COD/m -d
0.75
0.50
0.50
0.25
0.25
0.00 0.0
0.2
0.4
0.6
0.8
0.00 1.0
Dimensionless radial distance form center of granule (-)
Fig. 3 e Simulated substrate concentration profiles in granule.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 6 2 e4 5 7 0
the granule were simulated to explore the role of external and intra-granular mass transfer in the overall substrate removal rate in the SASB reactor. All the biological and physical parameter values used in model simulation are the same as previously described (Section 4.4). Because 4 and dp (R) are linearly dependent [Eq. (2)], the calculated 42 value (107e504; Table 5) increased with increasing dp (1.1e1.9 mm). Both 42 and dp values increased with increasing OLR, indicating that a higher OLR/larger dp gave a lower intra-granular mass transfer rate. The calculated Bi value (27e64; Table 5) increased with increasing OLR and dp, whereas the L value (0.02 mm; Table 5) remained nearly unchanged. Generally, a larger Bi value gives a larger external mass transfer rate than the intra-granular mass transfer rate. The calculated h value (0.12e0.20; Table 5) decreased with increasing OLR; a larger 42 value resulted in a smaller h value. The decreasing pattern of h (0.20e0.12) for increasing values of 42 (107e504) indicated that the overall substrate removal rate in the SASB reactor was diffusion-controlled (Bailey and Ollis, 1986). Based on the above mass transfer parameter values, the influence of intra-granular mass transfer on the overall substrate removal rate in the SASB reactors should not be neglected. To further clarify the role of intra-granular mass transfer in the overall substrate removal rate in the SASB reactor, the simulated substrate concentration profiles in the granule are presented in Fig. 3. Obviously, at the OLR of 2e8 kg COD/m3-d, Sf declines rapidly from the granule surface toward the inside of the granule. Thus, the overall substrate removal in the SASB reactor is evidently intra-granular diffusion-controlled. Moreover, by varying different granule sizes (dp) within a range of 0.1e3.5 mm, the simulated treatment efficiencies of the SASB reactor can be calculated (Fig. 4). At the OLRs of 2e8 kg COD/m3-d, the COD removal efficiency of the SASB reactor decreases slightly with increasing dp of up to 2.5 mm. Nonetheless, with a further increase in dp, the COD removal efficiency declines markedly at the OLRs of 2e4 kg COD/m3-d, whereas the COD removal efficiency declines moderately at the OLRs of 6e8 kg COD/m3-d. This can be explained by the following two reasons: a higher OLR gave a higher specific substrate utilization rate (Table 2); an increase of OLR from 2 to 8 kg COD/m3-d in the present study was controlled by increasing influent COD concentration. Accordingly, the
COD removal (%)
100.0
100.0
95.0
d c b
95.0
90.0
a
90.0
85.0
85.0
80.0
80.0
75.0 0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
75.0 4.0
dp (mm) Fig. 4 e Influence of granule size on treatment efficiency in SASB reactor (a [ 2.0 kg COD/m3-d, b [ 4.0 kg COD/m3-d, c [ 6.0 kg COD/m3-d, and d [ 8.0 kg COD/m3-d).
4569
optimal granular size could be no greater than 2.5 mm when the SASB reactor is used to treat non-inhibitory substrates.
5.
Conclusions
The present study demonstrated that the glucose-fed SASB reactor when operated at the OLRs of 2e8 kg COD/m3-d can effectively removed over 98% of COD from wastewater. The calculated residual substrate concentrations using the kinetic and empirical models agreed well with the experimental results (within 1.5% deviations). The relative percentage deviation of substrate removal efficiencies for kinetic-model and empirical-model calculation only fell in a small range (0.8e1.0%). Therefore, a kinetic model that is rather cumbersome for process design of SASB reactors can be replaced with an empirical model. The obtained intrinsic parameters k and Ks were 12.4e23.9 mg COD/mg VSS-d; 77e92 mg COD/L, respectively; whereas the apparent parameters k0 and Ks0 were 8.9e14.2 mg COD/mg VSS-d; 100e140 mg COD/L, respectively, revealing that the apparent kinetic parameters (with intragranular mass transfer resistance) determined using the intact granules can adequately represent the real granule characteristics. By applying the validated kinetic model, the calculated mass transfer parameter values (4, Bi, L, and h) and the simulated substrate concentration profiles in the granule showed that the overall substrate removal in the SASB reactor is evidently intra-granular diffusion-controlled. The simulated COD removal efficiency of the SASB reactor decreases slightly with increasing dp of up to 2.5 mm. Nonetheless, with a further increase in dp, the simulated COD removal efficiency declines markedly and moderately at the OLRs of 2e4 and 6e8 kg COD/m3-d, respectively. Thus, the optimal granular size could be no greater than 2.5 mm when the SASB reactor is used to treat a non-inhibitory substrate.
Acknowledgment Financial support of this research (Grant number: NSC 982211-E-168-012-MY2) from the National Science Council of the Republic of China (Taiwan) is gratefully acknowledged.
Nomenclature
A Bi Sb; Sb Sf; Sf Ss; Ss Si dp Df
specific surface area (mm1) Biot number (dimensionless) COD concentration in bulk liquid (mg/L; dimensionless) COD concentration within granule (mg/L; dimensionless) COD concentration at liquid/granule interface (mg/L; dimensionless) COD influent concentration (mg/L) average diameter of granules based on surface area (mm) diffusion rate of COD within granule (m2/d)
4570
Dw k; k0 Ks; Ks 0 L Mx Q r; r* R us VR Xf
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 6 2 e4 5 7 0
diffusion rate of COD in diffusion layer (m2/d) intrinsic and apparent maximum specific COD utilization rate (mg COD/mg VSS-d) intrinsic and apparent half-saturation constant (mg COD/L) thickness of diffusion layer (mm) biomass (g VSS) inflow rate (L/d) radial distance from center of granule (mm; dimensionless) granule radius (mm) superficial velocity (m/h) reactor volume (L) biomass density of granules (mg VSS/L)
Greek letters 4 Thiele modulus (dimensionless) e porosity of SASB reactor (dimensionless) v dynamic viscosity (m2/s) h effectiveness factor (dimensionless) Superscript * dimensionless
references
American Public Health Association (APHA), American Water Works Association (AWWA), and Water Environment Federation (WEF), 1995. Standard Methods for the Examination of Water and Wastewater, nineteenth ed. American Public Health Association, Washington, DC. Bailey, J.E., Ollis, D.F., 1986. Biochemical Engineering Fundamentals, second ed. McGraw-Hill, New York. Beun, J.J., Hendriks, A., van Loosdrecht, M.C., 1999. Aerobic granulation in a sequencing batch reactor. Water Res. 33, 2283e2290. Beun, J.J., van Loosdrecht, M.C., Heijnen, J.J., 2000. Aerobic granulation. Water Sci. Tech. 41, 41e48. Chou, H.H., Huang, J.S., 2005. Role of mass transfer resistance in overall substrate removal rate in upflow anaerobic sludge bed reactors. J. Envir. Eng. (ASCE) 131, 548e556. Chou, H.H., Huang, J.S., Hong, W.F., 2004. Temperature dependency of granule characteristics and kinetic behavior in UASB reactors. J. Chem. Technol. Biotechnol. 79, 797e808. Chou, H.H., Huang, J.S., Jheng, J.H., Ohara, R., 2008. Influencing effect of intra-granule mass transfer in expanded granular sludge bed reactors treating an inhibitory substrate. Bioresour. Technol. 99, 3403e3410. Etterer, T., Wilderer, P.A., 2001. Generation and properties of aerobic granular sludge. Water Sci. Technol. 43, 19e26. Fink, D.J., Na, T.Y., Schultz, J.S., 1973. Effectiveness factor calculations for immobilized enzyme catalysts. Biotechnol. Bioeng. 15, 879e888.
Grady, C.P.L., Daigger, G.T., Kim, H.C., 1999. Biological Wastewater Treatment (Revised and Expended), second ed. Marcel Dekker, New York. Huang, J.S., Jih, C.G., Lin, S.D., Ting, W.H., 2003. Process kinetics of UASB reactors treating non-inhibitory substrate. J. Chem. Technol. Biotechnol. 78, 762e772. Hines, A.L., Maddox, R.N., 1985. Mass Transfer Fundamentals and Applications. Prentice-Hall, New Jersey. Jiang, H.L., Tay, J.H., Tay, S.T.L., 2002. Aggregation of immobilized activated sludge cells into aerobically grown microbial granules for the aerobic biodegradation of phenol. Lett. Appl. Microbiol. 35, 439e445. Jiang, H.L., Tay, J.H., Tay, S.T.L., 2004. Changes in structure, activity and metabolism of aerobic granules as a microbial response to high phenol loading. Appl. Microbiol. Biotechnol. 63, 602e608. Jih, C.G., Huang, J.S., Huang, S.Y., 2003. Process kinetics of UASB reactors treating inhibitory substrate. Water Environ. Res. 75, 5e14. Lawrence, A.W., McCarty, P.L., 1969. Kinetics of methane fermentation in anaerobic treatment. J. Water Pollut. Control Fed 41 (2), R1eR17. Lettingna, G., Pol, L.H., 1991. UASB-process design for various types of wastewaters. Water Sci. Technol. 24, 87e107. Morgenroth, E., Sherden, T., van Looserecht, M.C.M., Heijnen, J.J., Wilderer, P.A., 1997. Aerobic granular sludge in a sequencing batch reactor. Water Res. 31, 3191e3194. Moy, B.Y.P., Tay, J.H., Toh, S.K., Liu, Y., Tay, S.T.L., 2002. High organic loading influences the physical characteristics of aerobic sludge granules. Lett. Appl. Microbiol. 34, 407e412. Perry, R.A., Chilton, C.C., 1973. Chemical Engineering Handbook, fifth ed. McGraw-Hill, Toronto. Press, W.H., Flannery, B.P., Tenkolsky, S.A., Vetterling, W.T., 1986. Numerical Recipes: the Art of Scientific Computing. Cambridge University Press, London, UK. Rittmann, R.E., McCarty, P.L., 1980. Model of steady-state-biofilm kinetics. Biotechnol. Bioeng. 22, 2343e2357. Schmidt, J.E., Ahring, B.K., 1996. Granule sludge formation in upflow anaerobic sludge blanket (UASB) reactors. Biotechnol. Bioeng. 49, 229e246. Tay, J.H., Liu, Q.S., Liu, Y., 2001. Microscopic observation of aerobic granulation in sequential aerobic sludge blanket reactor. J. Appl. Microbiol. Biotechnol. 91, 168e175. Tay, J.H., Liu, Q.S., Liu, Y., 2002. Aerobic granulation in sequential blanket reactor. Water Sci. Technol. 46, 13e18. Tay, J.H., Pan, S., He, Y., Tay, S.T.L., 2004. Effect of organic loading rate on aerobic granulation. I: reactor performance. J. Envir. Eng. (ASCE) 10, 1094e1101. Tay, S.T.L., Moy, B.Y.P., Jiang, H.L., Tay, J.H., 2005. Rapid cultivation of stable aerobic phenol-degrading granules using acetate-fed granules as microbial seed. Biotechnology 115, 387e395. Williamson, K., McCarty, P.L., 1976. A model of substrate utilization by bacterial films. J. Water Pollut. Control Fed. 48, 9e24. Wu, M.M., Hickey, R.F., 1997. Dynamic model for UASB reactor including reactor hydraulics, reaction, and diffusion. J. Envir. Eng. (ASCE) 123, 244e252.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 7 1 e4 5 8 2
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Of microparticles and bacteria identification e (resonance) Raman micro-spectroscopy as a tool for biofilm analysis Ann-Kathrin Kniggendorf*, Merve Meinhardt-Wollweber 1 Gottfried Wilhelm Leibniz University of Hannover, Institute of Biophysics, Herrenha¨user Str. 2, 30149 Hannover, Germany
article info
abstract
Article history:
Confocal resonance Raman microscopy is a powerful tool for the non-invasive analysis of
Received 22 March 2011
complex biological aggregates without preparation and prior knowledge of the samples.
Received in revised form
We present the capabilities of confocal resonance Raman microscopy with a spatial
28 May 2011
resolution of 350 nm2 2.0 mm and excitation times of 1 s and less per recorded spectrum.
Accepted 4 June 2011
Granules sampled from two sequencing batch reactors (SBR) for anaerobic ammonium
Available online 14 June 2011
oxidization (anammox) were regularly mapped in vivo for three months after SBR startup. Uncultured microorganisms and mineral particles were tracked throughout operation and
Keywords:
identified in situ by their (resonance) Raman spectra. Co-existing microcolonies of Nitro-
Confocal Raman microscopy
somonae formed the outer layer of anammox granules. Polymorph TiO2 microparticles were
Imaging
found embedded in the outer layer of granules overgrown with purple bacteria, indicating
Native samples
bacterial response to the variant toxicity of the mineral phase. ª 2011 Elsevier Ltd. All rights reserved.
Biofilms Mineral particles Uncultured microorganisms
1.
Introduction
The analysis of complex microbial aggregates and multispecies biofilms is still a challenge even for advanced imaging and identification techniques. These aggregates are formed of dense, highly hydrated, highly structured clusters of bacterial cells bound together by extracellular polymeric substances (EPS). Such clusters e so called microcolonies e differ in the bacterial species or phenotype of the species clustered together. In addition, a majority of the bacteria associated in biofilms cannot be grown in a pure culture, deferring most e if not all e standard-techniques for bacterial identification. Even those species for which pure cultivation is
possible often show distinctly different phenotypic traits in a biofilm than seen in their planctonic culture (Steward and Franklin, 2008). Understanding these complex communities requires advanced non-destructive imaging techniques providing information on spatial structure and distribution of e as well as tracking and identification capabilities for e the microorganisms associated in them, preferably without requiring invasive preparations or prior knowledge of the sample. Raman micro-spectroscopy e successfully used for the in vivo mapping of biofilms as well as identification and taxonomy of bacteria e is among the most promising techniques proposed for these tasks (Sandt et al., 2007; Harz et al., 2008). However, it
* Corresponding author. Present address: Gottfried Wilhelm Leibniz University of Hannover, Institute of Quantumoptics, Welfengarten 1, 30167 Hannover, Germany. Tel.: þ49 511 762 5125; fax: þ49 511 762 19432. E-mail addresses: [email protected] (A.-K. Kniggendorf), [email protected] (M. MeinhardtWollweber). 1 Present address: Gottfried Wilhelm Leibniz University of Hannover, HOT - Hannoversches Zentrum fu¨r Optische Technologien, Nienburger Str. 17, 30167 Hannover, Germany. 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.06.007
4572
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 7 1 e4 5 8 2
suffers from long measurement times and a high sensitivity to environmental (or culture) conditions, making it difficult to apply on extended segments of biofilm and bacteria in various habitats. To overcome these limitations, Raman microspectroscopy has been combined with several other methods, such as confocal laser scanning microscopy (CLSM) (Wagner et al., 2009), environmental scanning electron microscopy (ESEM) (Schwartz et al., 2009), and even fluorescent in situ hybridization (FISH) (Huang et al., 2007). However, by employing molecular resonances of inherent and ubiquitous but specific chromophores in confocal Raman microscopy, it is possible to achieve an in-depth analysis of the targeted microbial communities without additional, more invasive techniques or prior knowledge of the sample being required. The omnipresent autofluorescent background seen in biofilms at visible excitation wavelengths returns structural and morphologic information even in the rare case of absent chromophores thus preventing false negatives. In the presence of chromophores, information about structure, water content, bacteria (tracking, identification, and distribution), and non-biological components such as mineral microparticles is accessible on the scale of single cells (approx. 1 mm3). As an example, we analyzed the outer layer of bacterial aggregates sampled over three months from two identical sequencing batch reactors (SBR) for anaerobic ammonium oxidization (anammox) e one showing nominal anammox operation and one severely infected with purple bacteria e with confocal (resonance) Raman micro-spectroscopy at 532 nm excitation.
2.
Materials and methods
2.1.
Sequencing batch reactors (SBR)
Sample granules were provided by two sequencing batch reactors for the anammox process, each with a 12.8 l volume of which 10 l was used in operation. The SBRs were started with UASB pellets from an industrial IC reactor in Germany, treating the wastewater of an ethanol distillery, and active sludge from the municipal wastewater treatment plant in Hanover, Germany, in a ratio of 1:1. SBR-1 was additionally inoculated with seeding sludge from an operational anammox-IC in Rotterdam (Van der Star et al., 2008). Construction and operation of the SBRs were otherwise identical. SBR-1 was previously described and characterized by Wesoly (2009). It showed steady anammox activity after a startup phase of approx. two weeks. The main anammox organism present was candidatus Brocadia anammoxidans, tested with the FISH probe Ban162 (Wesoly, 2009). SBR-0 showed an increasing population of not otherwise specified purple bacteria after three weeks of operation. SBR0’s operation was discontinued after three months when no anammox occurred, while SBR-1 was kept in operation and monitored for another six months, showing steady anammox activity. Both SBRs were monitored in 48 h intervals with resonance Raman measurements as described in 2.3.1. A 3D image as described in 2.3.2 was recorded every 96 h. Identical
measurements were performed on granules of the seeding sludge prior to inoculation. Samples for subsequent resonance Raman measurements were taken from the SBRs always at the same phase of the operation cycle (during stirring). 50 ml of reactor water including microbial granules were sampled with a wide syringe at half-height from each SBR immediately prior to measurements. A single granule of average size (diameter: w1 mm) was taken from the sampled volume and placed free-floating in ample bulk liquid on an uncoated 1.2 mm indentation slide. The cover slip was sealed with acrylic lacquer to the slide to avoid dehydration of the sample. Measurements were started 2 min after sealing (to allow for drying of the sealing lacquer).
2.2.
Bacterial cultures and reference chemicals
Four strains of Nitrosomonae expressing cytochrome-c e Nitrosomonas communis Nm-02, two strains of Nitrosomonas europaea (Nm-50, Nm-53), and Nitrosomonas eutropha Nm-57 provided as liquid cultures by Dr. Pommerening-Ro¨ser of the Biozentrum Klein Flottbek, Hamburg, Germany e were cultured using media and conditions as described by Koops et al. (1991). Rhodobacter sphaeroides DSM 158T producing spheroidene and cytochrome-cbb3 and its carotenoids-free mutant DSM 2340T expressing only cytochrome-cbb3 were obtained as freeze-dried cultures from the German Collection of Microorganisms and Cell Cultures (Deutsche Sammlung von Mikroorganismen und Zellkulturen [DSMZ]), Braunschweig, Germany and cultured as described in Kniggendorf et al. (2011). Carotenoid production e or the lack thereof e was tested spectrometrically (Kontron Uvikon 932). Samples for subsequent Raman measurements were prepared and measured as previously described in Kniggendorf et al. (2011a). 99% pure cytochrome-c from bovine heart was obtained from SigmaeAldrich Chemie GmbH, Munich, Germany.
2.3.
Confocal resonance Raman measurements
Resonance Raman measurements were performed at room temperature with a confocal Raman microscope (CRM200, by WITec GmbH, Ulm, Germany), equipped with a waterimmersion objective (Nikon CFI Fluor) with a magnification of 60, and a Numerical Aperture of 1.0. A stabilized, frequencydoubled continuous-wave Nd:YAG laser at 532 nm was used for excitation. Photons from Rayleigh scattering were blocked with an edge filter, covering the range from 120 to 120 rel. cm1. The system has an ellipsoid measurement volume of approx. 1 mm3 defined by the lens properties (assumed refractive index within the sample: 1.33 (water)). The spatial resolution in the horizontal plane was 350 nm and 2.0 mm perpendicular to it, tested experimentally on glass/ chromophore-sample transitions. Slit width was 50 mm, realized by a multimode fiber connecting the Raman microscope with the spectrometer (Acton 308-SP). The spectrometer was equipped with two computer-controlled detection units e a backthinned CCD camera (Andor DU401-BV), electrically
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 7 1 e4 5 8 2
cooled to 70 C for spectrally resolved detection, and a singlephoton sensitive Avalanche photodiode (APD; SPCM-AQR) e selectable via an adjustable mirror. The used grating had 600 lines per millimeter. Laser intensity was adjusted to 9 mW, giving 2.3 MW/cm2 on the sample within the measurement volume.
2.3.1.
Spectral imaging
Measurement time per spectrum (or recorded pixel) was set to 1.0 s, sufficient for high-contrast resonance Raman spectra of cytochrome-c. Recorded images consisted of 140 80 spectra screening an area of 70 mm horizontally and 80 mm vertically (into the granule). The spectral resolution of the setup was 4 cm1 with a spectral accuracy of 2 cm1. Recorded spectrum range was 125 to 3694 rel. cm1.
2.3.2.
3D imaging
Measurement time per recorded pixel was set to 0.1 s. The setup recorded a section of approx. 4 cm1 around the denoted center frequency with a spectral accuracy of 2 cm1, allowing for specifically targeting resonantly enhanced Raman lines. Recorded image stacks consisted of 80 images distanced 1 mm along the z-axis (vertically into the granule). Each image consisted of 300 300 spectra, covering an area of 100 100 mm2 in the xy-plane.
2.4.
Spectral image analysis
Spectral images were recorded with ScanCTRL plus and digitally analyzed with WITec Project 1.94, both by WITec GmbH, Ulm, Germany. Image dimensions mimic the scanned micrometer range, not the number of recorded pixels. Since each recorded pixel contains full spectral information e autofluorescent background as well as possible Raman
4573
spectra e two different sets of filters were used to extract information and generate informative images from the recorded data. Broadband image filters are independent of whether or not Raman active chromophores are present. They display a sectioned superposition of the always present background and possible Raman spectra, allowing to mark areas of pure autofluorescence and/or high water content. For this, the recorded spectrum range was divided into three disjoint regions with region I going from 200 to 1300 cm1 (colored red), region II from 1300 to 2400 cm1 (yellow), and region III from 2400 to 3500 cm1 (blue). Fig. 1(a) gives the spectral sectioning for broadband filters superimposed on an exemplary set of (resonant) Raman spectra. Maximum and minimum intensity were defined per filtered image as the brightest and darkest pixel, with the darkest pixel being set to “black” (maximum contrast), resulting in a superimposition of the three filtered images (“broadband image”) giving white in case of plain fluorescence (same intensity in all three regions), blue in case of high water content due to the wide Raman band consisting of unresolved OH-stretching modes above 3000 cm1, and monochrome, unstructured red in case of glasses with homogenous density (for example microscope slides or cover slips). Microbial content shows as structured red and purple in a broadband image, due to the main intensity of bacterial spectra being primarily in broadband section I (red) and less in II (yellow). The water content of low density biofilm matter gives a low intensity in III (blue), resulting in a dark blue coloring (or purple in case of bacteria being present). Mineral microparticles, having Raman spectra with comparatively low intensity and no fluorescent background do not show in broadband images, except as dark areas with little to no intensity.
Fig. 1 e Filter concepts for spectral imaging. Broadband image filters (a) are independent of a chromophore being detected. The full spectrum as recorded is divided into three equal regions: 200e1300 cmL1 (I e red), 1300e2400 cmL1 (II e yellow), and 2400e3500 cmL1 (III e blue). Included (resonant) Raman spectra are glass of a microscope slide (1) and a cover slip (2), water (3), Nitrosomonas communis Nm-02 (chromophore: cytochrome-c) (4), and Rhodobacter sphaeroides DSM 158T (chromophore: spheroidene) (5). Fingerprint image filters (b) return the distribution and intensity of a specific (resonance) Raman spectrum within the recorded spectral image. For their construction the shaded parts of the spectrum are cut and the autofluorescent background (hatched) is subtracted. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article).
4574
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 7 1 e4 5 8 2
Fingerprint image filters give the distribution of a specific Raman spectrum within the recorded spectral image. Fingerprint image filters were constructed from the fingerprint region of Raman spectra found recurrently in the spectral images (including various types of unidentified bacteria and microparticles). Raman spectra of mineral microparticles were identified via RRUFF (Downs, 2006) and used in their entirety due to their negligible background and low number of Raman lines. Fingerprint image filters for more complex Raman spectra with noticeable autofluorescent background e as for example seen in spectra of bacteria expressing cytochrome-c as main chromophore e were prepared as follows: The spectra were cut to the fingerprint region (600e1800 cm1 throughout this work), followed by a full background reduction (9th order polynomial fit) over the same region (Fig. 1(b)). Intensity of a specific Raman line has to exceed the random noise by at least a factor of two to be considered as signal. Maximum intensity was set for the pixel holding the strongest Raman spectrum with respect to the four most intense Raman lines within the fingerprint region. Minimum intensity (“black”) was set for the four strongest Raman lines of the respective filter not exceeding background intensity. Fingerprint images were constructed by superimposing the fingerprint filtered images of interest. Colors were chosen with respect to visibility. The number of fingerprint filters usable on a recorded spectral image is virtually unlimited, but the number of fingerprint filtered images that can be included in a single superimposition (“fingerprint image”) is limited by the number of available colors and the limitations of additive color blending. However, an unlimited number of different fingerprint images can be created for a recorded spectral image.
2.5. Raman spectra comparison (OPUS IDENT) and hierarchical cluster analysis (HCA) Preparation and analysis of the Raman spectra was done with commercial spectra analysis software (OPUS version 5.5 including IDENT by Bruker). Spectrum properties were determined and bacterial references compiled as previously described in Kniggendorf et al. (2011a). The critical random noise property is additionally illustrated in the inset of Fig. 1(b). As a standard preparation of spectral HCA, the allowed spectral-to-noise ratio (S/N) was limited. Only Raman spectra with an S/N between 15 and 25 were used for further analysis to prevent differences in spectrum quality from exceeding variations caused by the actual bacteria. All spectra were cut to the fingerprint region of 600e1800 cm1 (the unshaded area in Fig. 1(b)), holding the most prominent peaks of the resonantly enhanced Raman spectrum of cytochrome-c, and vector-normalized to the same region. Spectral distances were calculated with the Euclidian Distance measurement. The initial vectornormalization limits the possible spectral distance to 2 (diameter of the unit ball) with a distance of 2 indicating an inverse spectrum and 0 indicating absolute identity (a spectrum being compared to itself).
Hierarchical cluster analyses were performed as described in Kniggendorf et al. (2011a) for single spectra with the Weighted-Average-Linkage algorithm.
3.
Results
Fig. 2 shows a typical set of three spectral images for SBR0 (purple bacteria) recorded after one month of operation: full spectral intensity as recorded (a), broadband image (b), and fingerprint image (c). The corresponding Raman spectra with the color-coding for Fig. 2(c) are given in Fig. 3. A similar set of spectral images for SBR-1 (anammox) recorded after one month of operation is given in Fig. 4 with the color-coded Raman spectra for Fig. 4(c) being given in Fig. 5.
3.1.
Chromophore identification
The main chromophores found in Raman spectra from the outer layer of granules sampled from SBR-0 were identified as carotenoids with a spheroidene backbone (most intense Raman line at 1519 cm1; bacterial spectrum given in Fig. 3, spectrum (2), magenta) or a neurosporene backbone (most intense Raman line at 1526 cm1; bacterial spectrum given in Fig. 3, spectrum (1), blue). Whereas the main chromophore found in the Raman spectra recorded in the outer layer of granules sampled from SBR-1 was heme-c as part of cytochrome-c. Cytochrome-c, resonantly enhanced in the Q-band, allows for the tracking (described below in 3.3) and identification (3.4) of bacteria in the biofilm. See spectra (1) to (3) in Fig. 5 for examples. These substances were initially identified by their Raman spectra. Spheroidene was additionally confirmed by measuring a culture of R. sphaeroides DSM 158T (data not shown), producing spheroidene as confirmed by absorption spectroscopy (data not shown). Cytochrome-c was additionally confirmed by comparison to a measurement of the pure substance (99% pure cytochrome-c from bovine heart) (data not shown). Unexpectedly, the Raman spectra of two phases of mineral TiO2 were found recurrently: Rutile (Fig. 3, spectrum (3), red) forming acicular or prismatic crystals, and anatase (Fig. 3, spectrum (4), yellow) in its typical form of dipyramidal or planar crystals. Raman spectra occurring only isolated and in low numbers in the three months of measurements were considered natural impurities and ignored in further analysis.
3.2. Spatially resolved information: structure and content 3.2.1.
Structure of the outer layer
As can be seen already in the full intensity images (Fig. 2(a) and Fig. 4(a)), the outer layers of the two SBRs are distinctly different from each other, with the outer layer of the granules from SBR-0 (Fig. 2(a)) being less regularly structured and of smaller overall width than that found in anammox granules sampled from SBR-1 (Fig. 4(a)). The microcolonies in granules sampled from SBR-0 form a haphazard monolayer with wide, irregular gaps, while the microcolonies seen in SBR-1 granules
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 7 1 e4 5 8 2
4575
Fig. 2 e Spectral Imaging based on (resonance) Raman spectra (SBR-0 infested with purple bacteria). Set of three spectral images of the outer layer of a granule sampled after one month of operation from SBR-0 (infested with purple bacteria): full spectral intensity as recorded (a), broadband image (b), and fingerprint image (c). The broadband image (b) shows wide, water-rich gaps in a single layer of elongated microcolonies with intense autofluorescence (white) at their core covered by a thin layer of Ramanactive bacteria (purple). The wide gaps contain substances with very low spectral intensity (black) and are filled and covered by a thick layer of low density biomass (dark blue). Yellow is indicative of the main spectral intensity being between 1300 and 2400 cmL1, typically caused by Raman spectra with additional autofluorescence in the spectrum. The fingerprint image (c) reveals anatase microparticles (yellow) in the gaps and microcolonies of purple bacteria holding spheroidene (magenta), growing at and around rutile microparticles (red) and a thin cover of neurosporene producing bacteria (blue) especially on surfaces exposed to the bulk water. The green “streaks” in the bulk water are caused by free-floating cells without active chromophores, dragged along by the laser focus. The (resonant) Raman spectra in the respective color-coding are given in Fig. 3. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article).
are ordered in a dense multilayer with well-defined borders and few distinct canals into the depth of the granule. The width of these canals in SBR-1 seldom exceeded 3 mm. 3D images, recorded as described in 2.3.2, revealed the distance
between adjacent bacterial microcolonies to seldom exceed 0.5 mm in granules sampled from SBR-1 (see Fig. 6), whereas microcolonies in granules from SBR-0 were distanced 1 mm and more from one another (Fig. 2).
3.2.2.
Discriminating autofluorescence, water, and biomass
The broadband image of the granule sampled from SBR0 (Fig. 2(b)) revealed strong autofluorescence (white) at the core of the microcolonies and large gaps of approx. 8e10 mm width in the outer layer, filled with water (blue) and watersaturated, low density biomass (dark blue). Fig. 4(b) shows the broadband image recorded from the outer layer of the granule sampled from SBR-1. Only a comparatively small area on the surface shows slight autofluorescence (white). The low intensity of the fluorescent area can be seen by comparison with the image of full intensity as recorded (Fig. 4(a)). The canals of approx. 3 mm width are filled with water (blue). The outer layer traversed by these canals is formed of microbial colonies (red; purple colonies have a noticeable water content, confirmed via the respective Raman spectra). Fig. 3 e (Resonant) Raman spectra color-coded as in Fig. 2(c): (1) Purple bacteria producing neurosporene (Fig. 2(c) blue), (2) purple bacteria producing spheroidene (Fig. 2(c) magenta), (3) TiO2 in rutile phase (Fig. 2(c) red), (4) TiO2 in anatase phase (Fig. 2(c) yellow), and (5) a bacterial cell containing no chromophores (Fig. 2(c) green). Spectra are presented as recorded. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article).
3.2.3.
Analyzing microbial communities e who’s where?
The elongated microcolonies seen already in the full intensity images of both SBRs (see 3.2.1) are distinctly different from each other in their resonant Raman signatures. The main chromophores found in the bacteria forming the microcolonies in the outer layer of SBR-0 granules were carotenoids of the spheroidene and neurosporene group. As seen in Fig. 2(c), individual bacterial colonies typically had a width of approx. 2e3 mm in the xy-plane and a length of
4576
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 7 1 e4 5 8 2
Fig. 4 e Spectral Imaging based on (resonance) Raman spectra (SBR-1 with nominal anammox activity). Set of three spectral images of the outer layer of a granule sampled after one month of operation from SBR-1 (anammox): full spectral intensity as recorded (a), broadband image (b), and fingerprint image (c). The outer layer of the granule sampled from SBR-1 shows a dense multilayer of microcolonies crossed centrally by a vertical canal (a second canal is touched near the right border of the image) already visible in the full spectral information (a). The broadband filtering (b) revealed water (blue) within the canal and the top third of the multilayer as well as weak autofluorescence (white; yellow in the presence of additional Raman lines) to the left of the bulk water opening of the canal. Fingerprint filtering (c) showed two types of Nitrosomonas microcolonies: N. communis (green) and a second species, probably N. europaea, (red). The corresponding resonance Raman spectra in the respective color-coding are given in Fig. 5. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article).
approx. 6e15 mm in the z-direction of the image. Only one layer of colonies was typically observed, spanning the whole width of the outer layer. The strong autofluorescent background made it impossible to detect Raman signals in the core of these microcolonies. However, all microcolonies showed
Fig. 5 e (Resonant) Raman spectra color-coded as in Fig. 4(c): (1) seed fingerprint type-I identified as Nitrosomonas communis (Fig. 4(c) green), (2) seed fingerprint type-II, probably N. europaea, (Fig. 4(c) red), (3) cytochromec spectrum 80 mm below the surface (Fig. 4(c) brown), (4) weak autofluorescence (Fig. 4(c) white), and (5) weakest autofluorescence (Fig. 4(c) yellow). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article).
purple bacteria producing primarily spheroidene (magenta, spectrum (2) in Fig. 3) on their surfaces, and often an additional thin layer of bacteria producing neurosporene (blue, spectrum (1) in Fig. 3) on surfaces exposed directly to bulk water. The latter were also seen on the walls of the wide, water-filled gaps. In contrast, the nitrification zone of the granules sampled from SBR-1 was formed by multiple, interwoven layers of two types of bacterial colonies of similar appearance (width: 2e3 mm, length: up to 5 mm in the image), distinct only in their resonant Raman fingerprint: Type-I (green, spectrum (1) in Fig. 5) and the much rarer Type-II (red, spectrum (2) in Fig. 5). Bacteria of Type-I formed the main bulk of the microcolonies (green), while Type-II (red) was much rarer and always surrounded by densely packed Type-I colonies without direct contact to the granule surface or water-filled canals (Fig. 4(c)). The main chromophore of both types was cytochrome-c, allowing for species-sensitive tracking and identification of the bacteria (see 3.3 and 3.4 below). Additional bacterial Raman spectra of cytochrome-c were detected underneath the nitrification zone (dark red, spectrum (3) in Fig. 5) and could be traced as deep as 80 mm into the granule. However, the quality of these spectra was not sufficient for a reliable comparison to the bacteria found within the nitrification layer. Variant thickness of the nitrification zone was typically caused by increased distance between the microcolonies of the outer layer, i.e. the space between the microcolonies widened and filled with water and low density biomass (blue and dark blue in Fig. 7(b)). This can be seen by comparing Figs. 4, 6, and 7. Also note that Type-II colonies (red) are virtually absent from the thick, water-rich nitrification zone seen in Fig. 7(c).
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 7 1 e4 5 8 2
3.2.4.
4577
Microparticles
A noticeable amount of microparticles e TiO2 in rutile and anatase phase e of a size up to 10 mm (average diameter was approx. 3 mm) were unexpectedly found in the outer layers of granules sampled from SBR-0 infested with purple bacteria. It is noteworthy, that these mineral particles have characteristic Raman lines at low wavenumbers, making them easily detectable even against increased autofluorescence as seen in the cores of purple bacteria colonies. The Raman spectrum of rutile (spectrum (3) in Fig. 3, red) consists of two lines at 447 and 613 cm1, while the spectrum of anatase (spectrum (4) in Fig. 3, yellow) has four lines with the strongest line as low as 144 cm1 (the others being at 394, 514, and 638 cm1 respectively). In SBR-0 (Fig. 2(c)), pyramidal and tabular anatase crystals (yellow) were found only on the surface of the outer layer, often partially embedded in low density matter (visible as less intense water content (dark blue) in Fig. 2(b)) and without direct contact to bacterial microcolonies. In contrast, short sections of rutile prisms and needles (red) were typically found embedded in the outer layer and often thickly packed with microcolonies of spheroidene producing purple bacteria. The outer layer of granules sampled from SBR-1 held only a few rutile microparticles and virtually no anatase was detected throughout the monitoring, discounting free particles in the bulk water (data not shown).
3.3.
Fig. 6 e Excerpt of an image stack covering 100 3 100 3 20 mm3 of the outer layer of an anammox granule showing a canal (D). Every second image is shown; images below L20 mm were ignored; altitude was set to 0 mm for the first focused layer of cells; center frequency: 749 cmL1 (pyrrole breathing, the strongest line of cytochrome-c at 532 nm excitation); excitation time per pixel: 0.1 s.
Bacterial tracking e who stays?
The two distinct Raman fingerprints of bacteria e Type-I and Type-II e found in the nitrification zone of granules sampled from SBR-1, were already found in abundance in the anammox seed granules obtained from Rotterdam and tracked successfully throughout operation of SBR-1. As seen in 3.2.3, these bacteria formed the majority of microcolonies in the outer layer of the granules sampled from SBR-1 (Fig. 4(c) and 7(c)) and of the seed granules obtained from Rotterdam (data not shown). Bacterial fingerprint Type-I (green) corresponded to spectrum (1) and Type-II (red) to spectrum (2) in Fig. 5. The respective Raman spectra recorded from the seed granules are given in Fig. 8. Fig. 9 shows the hierarchical cluster analysis of a typical set of bacterial Raman single spectra extracted from various measurements made at different times during the operation of SBR-1 together with two sample spectra obtained from the seed granules (seed I and II). Resonant Raman spectra were identified with the fingerprint image filters (see 2.4) for Types-I and eII and exported from the respective spectral images. Spectra with a suitable signal-to-noise ratio were subsequently subjected to HCA as described in 2.5. Spectra were labeled with the date of the measurement (mmdd) and the type as identified by the fingerprint image filters. As can be seen in Fig. 9, the two types of bacterial Raman spectra, classified as Type-I and eII by the fingerprint filters, were separated properly with a spectral distance larger than 0.7 between the branches I and II. Spectral distances between the components of a branch ranged from 0.25 to 0.46, with branch I forming a homogenous structure (spectrum 0903 has a stronger similarity to the seed 0801 and spectrum 0820 than
4578
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 7 1 e4 5 8 2
Fig. 7 e Spectral images based on resonance Raman spectra of an anammox granule swollen with water: full spectral intensity as recorded (a), broadband filtered image (b) categorizing bacteria (red, magenta), water (blue), low density biomass (dark blue), and autofluorescence (white), and fingerprint image (c) showing primarily green microcolonies of the tracked bacteria type-I (Nitrosomonas communis), very few bacteria of type-II (red), and weak autofluorescence (white) at the surface and right beneath the outer layer on the left. Filtered spectra are the same as in Fig. 4(c) and are given in Fig. 5. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article).
to 1022, as indicated by the corrective numbers underneath the respective clusters) and branch II being divided into a starting cluster (0801, 0813, 0830) including the seed and spectra recorded within the first month of operation and a cluster (0917, 0920, 1015, 1101) containing spectra recorded after 45 days of operation and later. This later cluster had a spectral distance of 0.45 from the starting cluster. This may indicate a change in the bacteria of Type-II between the 30th and the 45th day of operation. In order to gauge the significance of the spectral distances seen in Fig. 9, another HCA was performed with resonant Raman spectra of individual cells from N. communis (Nm-02), two strains of N. europaea (Nm-50, Nm-53), and the carotenoids-free mutant DSM 2340T of R. sphaeroides in pure culture (data not shown). The resulting dendrogram showed
spectral distances of 0.16e0.19 between spectra belonging to the same strain, a spectral distance of 0.44 between strains of the same species (N. europaea), and a spectral distance of 0.58 between different species holding the same chromophore (N. communis and N. europaea). For comparison, the spectra of bacteria holding a chromophore slightly variant from heme-c e like heme-cbb3 in cytochrome-cbb3 expressed by the carotenoids-free mutant of R. sphaeroides e are at a spectral distance larger than 1 to those of bacteria expressing heme-c.
3.4.
Bacterial identification e who’s who?
The first bacterial fingerprint (Type-I) recorded from the seed mass was identified with 94% certainty as N, communis Nm-02 grown in planctonic culture by OPUS IDENT as described
Fig. 8 e Resonant Raman spectra of bacteria containing cytochrome-c as main chromophore: type-I (a1) and type-II (a2) as recorded in the seed granules obtained from Rotterdam and Nitrosomonas communis Nm-02 (b1) and Nitrosomonas europaea Nm-50 (b2) as recorded from pure planctonic cultures.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 7 1 e4 5 8 2
4579
Fig. 9 e Hierarchical cluster analysis as described in 2.4 of resonant Raman spectra extracted from spectral images of the outer layer of anammox granules. Spectra are marked by date of measurement (month-day) and type as identified by the fingerprint filter. Seed I and II were found in the outer layer of the anammox seed mass.
in 2.5. Spectrum comparison between the second bacterial fingerprint (Type-II) and the other available reference spectra found a best match with a spectral similarity of approx. 80% to several N. europaea references (Nm-50, Nm-53), indicating a different strain of N. europaea or a phenotype variant from the phenotype observed in planctonic culture. Bacteria of this type were typically found surrounded by N. communis as identified by reference from planctonic culture (Type-I) and seldom or never had contact to water-rich areas of the active layer. For comparison, resonant Raman spectra recorded from individual cells of a pure culture as analyzed for the reference dendrogram in 3.3 are identified with 96 4% certainty when compared to spectra of the same strain and with 84 5% certainty when compared to spectra of a different strain of the same species. The spectral similarity to bacteria expressing heme-cbb3 instead of heme-c was typically 60% or less. The uncertainties decrease significantly with stricter limitation of the spectrum quality (i.e. allowing a narrower band of S/N ratios in spectra submitted to HCA).
4.
Discussion
The identity of the chromophores found in the samples were confirmed by their Raman spectra as reported by Koyama (1995) (carotenoids), Hu et al. (1993) (cytochrome-c), and Varotsis et al. (1995) (cytochrome-cbb3; solely in reference cultures of purple bacteria).
4.1.
Outer layers
The characteristics observed by confocal resonance Raman micro-spectroscopy without prior knowledge about the
samples are in very good agreement with the results of several other groups investigating similar bacterial aggregates and biofilms with conventional methods (Sliekers et al., 2002; Gieseke et al., 2003; Nielsen et al., 2005; Okubo et al., 2006; Tsushima et al., 2007). In detail:
4.1.1.
SBR-1: nitrifying layer on anammox granules
The existence of a structured nitrifying layer of approx. 20e30 mm thickness (up to 100 mm) e as seen in Figs. 4 and 7 e on the surface of anammox granules of similar size was already reported by Nielsen et al. (2005), testing indiscriminant for Nitrosomonas (N. europaea, N. eutropha, Nitrosococcus mobilis, Nitrosomonas halophila) and betaproteobacterial aerobic ammonium oxidizers in general. While Nielsen et al. did not discuss canals and microcolonies present, both are clearly visible in their presented data obtained with FISH of paraformaldehyde (PFA)-fixed cryosections of sampled aggregates. In addition, confocal (resonance) Raman microspectroscopy revealed the nitrifying layer of granules sampled from our sequencing batch reactor with nominal anammox activity (SBR-1) as a dense multilayer of at least two different nitrifying bacteria species e N. communis and a second species (probably N. europaea) e forming microcolonies in distinct spatial relation, with the second species typically not having contact to bulk liquid or water-filled canals. Co-existing microcolonies of different nitrifying bacteria (N. europaea/eutropha, N. mobilis) associated on the microscale were previously reported for example by Gieseke et al. (2003), who employed CLSM in combination with FISH on PFA-fixed thin sections. HCA of the obtained Raman spectra indicated that a change occurred in the second species of Nitrosomonas between 30 and 45 days after inoculation, coinciding with the
4580
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 7 1 e4 5 8 2
time after which denitrification continuously exceeded anammox in SBR-1 according to Wesoly (2009). Similar time frames for biomass adaptation to changed environmental conditions were reported for example by Tsushima et al. (2007) for the simultaneous removal of ammonium and nitrite being detectable after startup, and Sliekers et al. (2002) for the formation of biomass containing 45% aerobic ammonium oxidizers, primarily Nitrosomonas, after aeration of a previously strict anoxic SBR. At this point in time, we may only speculate about the observed change detected in the second Nitrosomonas species to be the result of phenotypic adaptation to the by then established local microenvironment (Steward and Franklin, 2008) between the dense colonies of N. communis separating them from the bulk water.
4.1.2.
SBR-0: purple bacteria
Confocal resonance Raman micro-spectroscopy distinguished two different types of bacteria in the outer layer of SBR-0. Bacteria producing carotenoids with a neurosporene backbone were found typically in contact with bulk water or on the walls of water-filled gaps, while bacteria within the microcolonies produced carotenoids with a spheroidene backbone. Carotenoids are ubiquitous chromophores, which in phototrophic bacteria are mostly associated with the light harvesting complexes (Cogdell et al., 2006). Since both carotenoids are intermediates of the spheroidene pathway for the biosynthesis of carotenoids typical in phototrophic purple bacteria (Takaichi, 2008), we cannot decide at this point in time whether or not the variant chromophores are indicative of two different species of purple bacteria (Okubo et al., 2006), or are in fact indicative of an adaptation of the same species to different habitat conditions in the biofilm (Steward and Franklin, 2008), e.g. stopping the carotenoid synthesis at neurosporene when in contact to bulk water, instead of continuing down the synthesis pathway towards spheroidene. However, Pudney et al. (2011) successfully traced the molecular status of multiple carotenoids in tomatoes and tomato products with confocal Raman micro-spectroscopy. Given the complexity of the carotenoid composition found in many bacteria (Takaichi, 2008a), a species-sensitive detection based on present carotenoids, carotenoid associations and conformations may well be possible.
4.1.3.
TiO2 microparticles and bacteria
The exact origin of the TiO2 microparticles found in the samples is unknown as they were not specifically added to the SBRs. However, it is highly likely that municipal wastewater used in the startup process and in the cyclic feeding of the SBRs contained noticeable traces of them, given that TiO2 in rutile phase is widely used as a white pigment even in foods (E171) and anatase is used for example as optical brightener in plastics (Fink, 2010). The location of the TiO2 microparticles with respect to the purple bacteria in the outer layer of granules sampled from SBR-0 appears to be mineral phase dependent. Microparticles of TiO2 in anatase phase were not in direct contact to bacterial microcolonies and often stuck in thick swaths of low density matter, which might be comparable to the gelatinous matrix seen in mats of phototrophic non-sulfur purple bacteria reported by Okubo et al. (2006). Rutile microparticles were
typically densely populated with purple bacteria appearing almost embedded in large microcolonies. We are not aware of a study reporting anything comparative to this for small microparticles in the range of 1e10 mm (or anything larger). Nothing of this kind was observed in the nitrification layer of granules sampled from SBR-1. However, Fang et al. (2010) reported anatase nanoparticles to cause cell shrinkage and membrane damage in N. europaea, independent of particle size (25 nm, 200 nm), while Liu et al. (2010) showed membrane damage caused in Escherichia coli by TiO2, confirming bactericidal properties even in absence of UV irradiation. In addition, Johnston et al. (2009) reviewed the toxicity of TiO2 in anatase and rutile phase on tissues and eukaryotic cells, finding the toxic potency of anatase considerably larger than that of rutile in most of the studies, some of which covered prokaryotic cells as well. This may explain while anatase microparticles embedded in the outer layer were surrounded by low density matter keeping them separated from the purple bacteria microcolonies, while rutile crystals were found overgrown. Considering the slow growth of Nitrosomonas in comparison to most purple bacteria and the aforementioned results of Fang et al., the toxic effects of rutile e and esp. anatase e microparticles may have a more severe impact on microcolonies of Nitrosomonas. This may explain why there were few rutile and virtually no anatase microparticles embedded in the nitrifying layer of granules sampled from SBR-1.
4.2.
Quantification and digital image analysis
Sandt et al. (2008) proposed a quantitative analysis of the water-to-biomass ratio of biofilms based on the CH-stretching modes seen in EPS and biomass and the OH-stretching modes of water (centered around 2840 cm1 and 3400 cm1 respectively). However, in our case the water-to-biomass ratio of a single spectrum already depended highly on whether or not a bacterial cell was within the measurement volume. In addition, a statistical approach by averaging multiple spectra would have been highly arbitrary, given that discriminating between extra-cellular water in the EPS and intra-cellular water of the bacteria is not possible. While we did not use digital image analysis for quantification on our data, the nature of the data allows straightforward employment of the technique. The bias caused by manually set thresholds (or thresholding algorithms) as described by Daims and Wagner (2007) and Merod et al. (2007) does not exist due to the inherent threshold definition of the major known Raman lines in a fingerprint spectrum to exceed the background of autofluorescence and random noise by at least a given intensity, e. g. for the S/N to satisfy an adapted Rose criterion (Bright et al., 1998).
4.3. Resonance Raman micro-spectroscopy in comparison to other methods Very few techniques for blind, label-free imaging of native, undisturbed biofilm have been reported to date: most notably, infrared absorption and normal Raman spectroscopy, often in combination with confocal microscopy, have been employed for several biofilm and bacteria studies. However, long exposure times (100 s and more) and high variability of bacterial
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 7 1 e4 5 8 2
Raman spectra due to variant environmental conditions and metabolic variants of the cells pose significant challenges for mapping biofilms as well as identifying bacteria (Sandt et al., 2007; Harz et al., 2008). In contrast, resonance Raman spectroscopy relies primarily on specific ubiquitous molecules (chromophores) produced by the bacteria. The resonant excitation of these chromophores allows for significantly shorter exposure times (1 s and less) and reduces the variability of the spectra to metabolic changes of the bacterial cell directly affecting the chromophores. However, the reliance on chromophores restricts specificity to the specificity of the chromophore within the bacteria and also dictates the excitation wavelength, thus affecting the compatibility with methods such as FISH, which require fluorescent staining (Neu et al., 2010). Most, if not all, available fluorescent stains do not allow for subsequent (resonant) Raman measurements with visible excitation. The combination of Raman and FISH reported by Huang et al. (2007) relied on the detection of 13C labeled cells, trading most e if not all e of the advantages of Raman micro-spectroscopy in regard of non-invasive measurements of native, undisturbed samples. Ivleva et al. (2010) used rapid surface enhanced Raman scattering (SERS) in combination with normal Raman for the in situ chemical characterization of complex biofilms grown on marked glass slides, mapping several SERS fingerprints indicative of polysaccharides and proteins. The presented raster maps of 60 60 mm2 with a spatial resolution of 3 mm were recorded with an excitation time of 1 s per SERS spectrum, bringing this technique in the same range of spatial resolution and recording time as resonance Raman measurements. However, the requirement of silver nanoparticles as SERS substrate may limit this technique to endpoint analyses, given that silver nanoparticles of similar size have been reported by Liang et al. (2010) to have a severe negative impact on nitrifying bacteria in activated sludge, leading to significant nitrification inhibition and changes in the communal composition of the present bacteria. Haisch and Niessner (2007) reported on optical coherence tomography (OCT) for online, in vivo and in situ visualization of three-dimensional biofilm density structures with a 10 mm resolution and transient processes with an impressive temporal resolution between 1 s and a few minutes for complete images. However, while discrimination between bacterial matter and larger microparticles (diameter >10 mm) and even different types of microparticles may be possible due to different particle densities, the discrimination between different types of bacteria and thus the subsequent identification of bacteria as shown with resonance Raman microspectroscopy is not possible with OCT. A combination of OCT and CLSM as reported by Wagner et al. (2010) for analyzing comparatively large swaths of biofilm (sample volumes of 4 4 3 mm3 at a resolution below 20 mm), gives hope for the successful combination of OCT with (resonance) Raman micro-spectroscopy in the future.
5.
Conclusion
We successfully demonstrated the suitability of confocal (resonance) Raman micro-spectroscopy for blind in vivo
4581
analysis of the first 80 mm of undisturbed biofilms in water. The analysis covered structural information and distribution, tracking, and identification on the scale of single bacterial cells. Microparticles embedded in the outer layer were also successfully analyzed with respect to the mineral phase and the surrounding bacterial microcolonies. Confocal (resonance) Raman micro-spectroscopy has the following key features: non-invasive optical analysis of living, undisturbed wet samples (no fixation, probes, labels, or stains required) of up to 80 mm thickness blind measurements (no advance knowledge required; chromophores can be identified after the measurement; background analysis prevents false negatives) spatial resolution at cell size (0.35 mm2 2 mm) or (0.35 mm2 1.2 mm), allowing quantification at absolute cell numbers if required, with excitation times of 0.5e1.0 s for a single spectrum (0.1 s for single wavelength detection in 3D) identification of bacteria in genus, strain and/or phenotype (dependent on chromophore), distribution and tracking of bacteria independent from identification biological and mineral components can be analyzed in the same measurement
Acknowledgments This work was kindly supported by the German Research Foundation (DFG, grant no. AN 712/1-5). The authors would like to thank Dr. Andreas Pommerening-Ro¨ser of the Biozentrum Klein Flottbek, Mikrobiologie und Biotechnologie, in Hamburg for providing us with the Nitrosomonas references and invaluable information about their cultivation. The authors would like to express special thanks to Dr. Ilona Wesoly for detailed information on the starting conditions and operation of the SBRs. The help of Dipl.-Biol. Tobias William Gaul in cultivating the Nitrosomonas references is kindly acknowledged.
references
Bright, D.S., Newbury, D.E., Steel, E.B., 1998. Visibility of objects in computer simulations of noisy micrographs. Journal of Microscopy 189 (1), 25e42. Cogdell, R.J., Gall, A., Ko¨hler, J., 2006. The architecture and function of the light-harvesting apparatus of purple bacteria: from single molecules to in vivo membranes. Quarterly Reviews of Biophysics 39 (3), 227e324. Daims, H., Wagner, M., 2007. Quantification of uncultured microorganisms by fluorescence microscopy and digital image analysis. Applied Microbiology Biotechnology 75, 237e248. Downs, R.T., 2006. The RRUFF Project: an Integrated Study of the Chemistry, Crystallography, Raman and Infrared Spectroscopy of Minerals Program and Abstracts of the 19th General Meeting of the International Mineralogical Association in Kobe, Japan03e13.
4582
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 7 1 e4 5 8 2
Fang, X., Yu, R., Li, B., Somasundaran, P., Chandran, K., 2010. Stresses exerted by ZnO, CeO2, and anatase TiO2 nanoparticles on the Nitrosomonas europaea. Journal of Colloid and Interface Science 348, 329e334. Fink, J.K., 2010. A Concise Introduction to Additives for Thermoplastic Polymers. Scrivener Publishing, LLC, Salem, Massachusetts. Gieseke, A., Bjerrum, L., Wagner, M., Amann, R., 2003. Structure and activity of multiple nitrifying bacterial populations co-existing in a biofilm. Environmental Microbiology 5 (5), 355e369. Haisch, C., Niessner, R., 2007. Visualisation of transient processes in biofilms by optical coherence tomography. Water Research 41, 2467e2472. Harz, M., Ro¨sch, P., Popp, J., 2008. Vibrational spectroscopy e a powerful tool for the rapid identification of microbial cells at the single-cell Level. Cytometry Part A 75A, 104e113. Hu, S., Morris, I.K., Singh, J.P., Smith, K.M., Spiro, T.G., 1993. Complete assignment of cytochrome c resonance Raman spectra via enzymic reconstitution with isotopically labeled hemes. Journal of the American Chemical Society 115 (26), 12446e12458. Huang, W.E., Stoecker, K., Griffiths, R., Newbold, L., Daims, H., Whiteley, A.S., Wagner, M., 2007. Raman-FISH: combining stable-isotope Raman spectroscopy and fluorescence in situ hybridization for the single cell analysis of identity and function. Environmental Microbiology 9 (8), 1878e1889. Ivleva, N.P., Wagner, M., Szkola, A., Horn, H., Niessner, R., Haisch, C., 2010. Label-free in situ SERS imaging of biofilms. Journal of Physical Chemistry B 114, 10184e10194. Johnston, H.J., Hutchison, G.R., Christensen, F.M., Peters, S., Hankin, S., Stone, V., 2009. Identification of the mechanisms that drive the toxicity of TiO2 particulates: the contribution of physicochemical characteristics. Particle and Fibre Toxicology 6 (33). doi:10.1186/1743-8977-6-33. Kniggendorf, A.-K., Gaul, T.W., Meinhardt-Wollweber, M., 2011. Effects of ethanol, formaldehyde, and gentle heat fixation in confocal resonance Raman microscopy of purple nonsulfur bacteria. Microscopy Research and Technique 74, 177e183. Kniggendorf, A.-K., Gaul, T.W., Meinhardt-Wollweber, M., 2011a. Hierarchical cluster analysis (HCA) of microorganisms: an assessment of algorithms for resonance Raman spectra. Applied Spectroscopy 65 (2), 165e173. Koops, H.-P., Bo¨ttcher, B., Mo¨ller, U.C., Pommerening-Ro¨ser, A., Stehr, G., 1991. Classification of eight new species of ammonia-oxidizing bacteria: Nitrosomonas communis sp. nov., Nitrosomonas ureae sp. nov., Nitrosomonas aestuarii sp. nov., Nitrosomonas marina sp. nov., Nitrosomonas nitrosa sp. nov., Nitrosomonas eutropha sp. nov., Nitrosomonas oligotropha sp. nov. and Nitrosomonas halophila sp. nov. Journal of General Microbiology 137, 1689e1699. Koyama, Y., 1995. In: Britton, G., Liaaen-Jensen, S., Pfander, H. (Eds.), Carotenoids Volume 1B: Spectroscopy, (chapter 5): Resonance Raman Spectroscopy. Birkha¨user Verlag, Basel, pp. 135e146. Liang, Z., Das, A., Hu, Z., 2010. Bacterial response to a shock load of nanosilver in an activated sludge treatment system. Water Research 44, 5432e5438. Liu, P., Duan, W., Wang, Q., Li, X., 2010. The damage of outer membrane of Escherichia coli in the presence of TiO2 combined with UV light. Colloids and Surfaces B: Biointerfaces 78, 171e176. Merod, R.T., Warren, J.E., McCaslin, H., Wuertz, S., 2007. Toward automated analysis of biofilm architecture: bias caused by extraneous confocal laser scanning microscopy images. Applied and Environmental Microbiology 73 (15), 4922e4930. Neu, T.R., Manz, B., Volke, F., Dynes, J.J., Hitchcock, A.P., Lawrence, J.R., 2010. Advanced imaging techniques for
assessment of structure composition and function in biofilm systems. FEMS Microbiology Ecology 72, 1e21. Nielsen, M., Bollmann, A., Sliekers, O., Jetten, M., Schmid, M., Strous, M., Schmidt, I., Hauer Larsen, L., Nielsen, L.P., Revsbech, N.P., 2005. Kinetics, diffusional limitation and microscale distribution of chemistry and organisms in a CANON reactor. FEMS Microbiology Ecology 51, 247e256. Okubo, Y., Futamata, H., Hiraishi, A., 2006. Characterization of phototrophic purple nonsulfur bacteria forming colored microbial mats in a swine wastewater ditch. Applied and Environmental Microbiology 72 (9), 6225e6233. Pudney, P.D.A., Gambelli, L., Gidley, M.J., 2011. Confocal Raman microspectroscopic study of the molecular status of carotenoids in tomato fruits and foods. Applied Spectroscopy 65 (2), 127e134. Sandt, C., Smith-Palmer, T., Pink, J., Brennan, L., Pink, D., 2007. Confocal Raman microspectroscopy as a tool for studying the chemical heterogeneities of biofilms in situ. Journal of Applied Microbiology 103, 1808e1820. Sandt, C., Smith Palmer, T., Pink, J., Pink, D., 2008. Quantification of local water and biomass in wild type PA01 biofilms by confocal Raman microspectroscopy. Journal of Microbiological Methods 75, 148e152. Schwartz, T., Jungfer, C., Heißler, S., Friedrich, F., Faubel, W., Obst, U., 2009. Combined use of molecular biology taxonomy, Raman spectrometry, and ESEM imaging to study natural biofilm grown on filter materials at waterworks. Chemosphere 77, 249e257. Sliekers, A.O., Derwort, N., Campos Gomez, J.L., Strous, M., Kuenen, J.G., Jetten, M.S.M., 2002. Completely autotrophic nitrogen removal over nitrite in one single reactor. Water Research 36, 2475e2482. Steward, P.S., Franklin, M.J., 2008. Physiological heterogeneity in biofilms. Nature Reviews Microbiology 6, 199e210. Takaichi, S., 2008. In: Hunter, C.N., Daldal, F., Thurnauer, M.C., Beatty, J.T. (Eds.), Advances in Photosynthesis and Respiration Vol. 28: The Purple Phototrophic Bacteria, (Chapter 6)-II: Carotenogenesis. Springer, Dordrecht, The Netherlands, pp. 101e111. Takaichi, S., 2008a. In: Hunter, C.N., Daldal, F., Thurnauer, M.C., Beatty, J.T. (Eds.), Advances in Photosynthesis and Respiration vol. 28: The Purple Phototrophic Bacteria, (Chapter 6)-III: Carotenoids in Purple Bacteria. Springer, Dordrecht, The Netherlands, pp. 111e117. Tsushima, I., Ogasawara, Y., Kindaichi, T., Satoh, H., Okabe, S., 2007. Development of high-rate anaerobic ammoniumoxidizing (anammox) biofilm reactors. Water Research 41, 1623e1634. Van der Star, W.R.L., Miclea, A.I., van Dongen, U.G.J.M., Muyzer, G. , Picioreanu, C., van Loosdrecht, M.C.M., 2008. The membrane bioreactor: a novel tool to grow anammox bacteria as free cells. Biotechnology Bioengineering 101, 286e294. Varotsis, C., Babcock, G.T., Garcia-Horsman, J.A., Gennis, R.B., 1995. Resonance Raman spectroscopy of the heme groups of cytochrome cbb3 in rhodobacter sphaeroides. Journal of Physical Chemistry 99 (46), 16817e16820. Wagner, M., Ivleva, N.P., Haisch, C., Niessner, R., Horn, H., 2009. Combined use of confocal laser scanning microscopy (CLSM) and Raman microscopy (RM): investigation on EPS-Matrix. Water Research 43, 63e76. Wagner, M., Taherzadeh, D., Haisch, C., Horn, H., 2010. Investigation of the mesoscale structure and volumetric features of biofilms using optical coherence tomography. Biotechnology and Bioengineering 107 (5), 844e853. Wesoly, I. 2009. Operating Strategies for the anammox process in sequencing batch reactors (German). Doctoral thesis. Technical University of Berlin, Germany.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 8 3 e4 5 9 1
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Adsorption of ciprofloxacin on surface-modified carbon materials S.A.C. Carabineiro a,*, T. Thavorn-Amornsri a,b, M.F.R. Pereira a, J.L. Figueiredo a a
Laborato´rio de Cata´lise e Materiais, LSRE/LCM e Laborato´rio Associado, Faculdade de Engenharia, Universidade do Porto, Rua Dr. Roberto Frias, s/n, 4200-465 Porto, Portugal b International School of Engineering, Faculty of Engineering, Chulalongkorn University, Pathumwan, Bangkok 10330, Thailand
article info
abstract
Article history:
The adsorption capacity of ciprofloxacin (CPX) was determined on three types of carbon-
Received 21 February 2011
based materials: activated carbon (commercial sample), carbon nanotubes (commercial
Received in revised form
multi-walled carbon nanotubes) and carbon xerogel (prepared by the resorcinol/formal-
31 May 2011
dehyde approach at pH 6.0). These materials were used as received/prepared and func-
Accepted 6 June 2011
tionalised through oxidation with nitric acid. The oxidised materials were then heat
Available online 16 June 2011
treated under inert atmosphere (N2) at different temperatures (between 350 and 900 C).
Keywords:
the point of zero charge and by temperature programmed desorption. High adsorption
Ciprofloxacin
capacities ranging from approximately 60 to 300 mgCPx gC1 were obtained (for oxidised
Adsorption
carbon xerogel, and oxidised thermally treated activated carbon Norit ROX 8.0, respec-
Activated carbon
tively). In general, it was found that the nitric acid treatment of samples has a detrimental
Carbon xerogel
effect in adsorption capacity, whereas thermal treatments, especially at 900 C after
Carbon nanotubes
oxidation, enhance adsorption performance. This is due to the positive effect of the surface
The obtained samples were characterised by adsorption of N2 at 196 C, determination of
basicity. The kinetic curves obtained were fitted using 1st or 2nd order models, and the Langmuir and Freundlich models were used to describe the equilibrium isotherms obtained. The 2nd order and the Langmuir models, respectively, were shown to present the best fittings. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
Ciprofloxacin (CPX) is a synthetic antibiotic (Fig. 1) marketed and used worldwide for the treatment of several bacterial infections in humans and animals. It can be released into water sources due to incomplete metabolism in humans or coming from effluents of drug manufacturers (Bhandari et al., 2008; Diwan et al., 2009, 2010; Giger et al., 2003; Karthikeyan and Meyer, 2006; Larsson et al., 2007). In wastewater treatment plants CPX, and other antibiotics, are only partially eliminated and residual amounts can reach surface water or
groundwater. The presence of CPX in wastewater and surface water, even in low concentrations, can lead to the development of antibiotic resistant bacteria (Bhandari et al., 2008; Diwan et al., 2009, 2010; Giger et al., 2003; Golet et al., 2002; Karthikeyan and Meyer, 2006; Larsson et al., 2007). CPX has been measured in water and wastewater at concentrations typically <1 mg L1 (Bhandari et al., 2008; Carmosini and Lee, 2009; Golet et al., 2002; Karthikeyan and Meyer, 2006; Larsson et al., 2007; Renew and Huang, 2004); however, much higher concentrations have been measured in effluents from hospitals (3e87 mg L1) (Carmosini and Lee, 2009) and
* Corresponding author. Fax. þ351 22 5081449. E-mail address: [email protected] (S.A.C. Carabineiro). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.06.008
4584
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 8 3 e4 5 9 1
Thus, oxidation in the gas or liquid phase can be used to increase the concentration of surface oxygen groups, while heating under inert atmosphere may be used to selectively remove some of these functional groups (Carabineiro et al., 2011; Figueiredo et al., 1999, 2007; Serp and Figueiredo, 2009). This approach was previously used in our group to study the adsorption of phenol, aniline and nitrobenzene on activated carbons (Villacan˜as et al., 2006). In the present work, the adsorption of CPX was studied on different carbon materials (not only activated carbon, but also carbon xerogel and carbon nanotubes) with different surface treatments. The objective is to compare different types of carbon materials, and also to determine the effect of their surface chemistry on the adsorption of this antibiotic. To the best of our knowledge, no such comparative work was carried out before.
Fig. 1 e Molecular structure of ciprofloxacin (CPX) with views of the molecule with two perspectives in perpendicular planes. Atoms are labelled on the top view.
1
drug production facilities (31 mg L ) (Carmosini and Lee, 2009; Larsson et al., 2007), therefore its removal from water sources has become an increasingly important subject. Even so, there are not many studies dealing with the elimination of CPX, when compared with other antibiotics. A few studies in literature deal with CPX adsorption in activated charcoal and talc (Ibezim et al., 1999), sorption by dissolved organic carbon (Carmosini and Lee, 2009), photodegradation (Belden et al., 2007), photo-Fenton oxidation processes (Sun et al., 2009), oxidation by chlorine and chlorine dioxide (Dodd et al., 2005; Wang et al., 2010) and ozonation (De Witte et al., 2010, 2008). Carbon materials (namely activated carbons) are wellknown “universal” adsorbents and present unique advantages due to their low cost, high adsorption capacity and easy disposal (Carabineiro et al., 2011; Villacan˜as et al., 2006). Activated carbons have been widely used in studies of adsorption of pollutants from wastewater (Cabrita et al., 2010; Faria et al., 2008; Moreno-Castilla, 2004; Radovic et al., 2000; Villacan˜as et al., 2006). Similar works have been reported for carbon nanotubes (Cho et al., 2008; Li et al., 2011; Sheng et al., 2010). The performance of carbon materials depends greatly on their texture and surface chemistry. The presence of oxygen atoms originates a variety of functional groups on the surface of carbon materials (Carabineiro et al., 2011; Figueiredo et al., 1999, 2007; Serp and Figueiredo, 2009). The effect of surface oxygen groups on the adsorption of organic compounds has been studied by preparing oxidised activated carbons (Faria et al., 2008; Villacan˜as et al., 2006) or carbon nanotubes (Cho et al., 2008; Li et al., 2011; Sheng et al., 2010), and comparing them with the parent materials. In general, it has been found that increasing the oxygen content of samples (i.e., turning the carbon surface more acidic), has detrimental effects on adsorption (Cho et al., 2008; Faria et al., 2008; Li et al., 2011; Sheng et al., 2010; Villacan˜as et al., 2006). The nature and concentration of surface functional groups can be modified by suitable thermal or chemical treatments.
2.
Experimental
2.1.
Preparation of carbon materials
Three types of carbon materials were used in this work: activated carbon, polymer-based carbon xerogel, and carbon nanotubes. A commercial activated carbon was used, namely NORIT ROX 0.8 (sample AC), which is an extruded acid washed activated carbon with cylindrical pellets of 0.8 mm diameter and 5 mm length, prepared from peat by steam activation. A carbon xerogel was synthesized by the polycondensation of resorcinol and formaldehyde, as described in earlier publications (Mahata et al., 2008; Samant et al., 2004). The xerogel was made at pH 6 (sample CX). Commercial Nanocyl-3100 multiwalled carbon nanotubes were supplied by Nanocyl, Belgium (sample CNT). This material has an average diameter of 9.5 nm, an average length of 1.5 mm (with average inner diameter of 4 nm) and a carbon purity higher than 95%. Further details on this material can be found elsewhere (Gonc¸alves et al., 2010; Tessonnier et al., 2009).
2.2. Modification of surface chemistry of carbon materials AC and CX (9 g) were oxidised with nitric acid (HNO3 5 M, 200 ml), at the boiling temperature of 130 C, for 3 h, with stirring, then washed with distilled water to neutral pH and dried in an oven at 120 C for 24 h, as described in earlier publications (Figueiredo et al., 1999, 2007; Mahata et al., 2008; Samant et al., 2004). Samples ACa and CXa were obtained. Sample CNT was also treated by a similar procedure, but the nitric acid concentration used was 7 M (sample CNTa), as described in a previous paper (Gonc¸alves et al., 2010), since carbon nanotubes are generally more resistant to oxidation than activated carbons. Further thermal treatments were carried out on samples ACa, CXa and CNTa in order to modify their surface chemistry: samples ACa and CXa were heated to 350 C at 10 C min1 under a flow of N2 (100 cm3 min1) and kept at this temperature for 1 h (ACb and CXb samples were obtained) or to 900 C (samples ACc and CXc). CNTa was subjected to a similar
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 8 3 e4 5 9 1
4585
procedure, but heated to 400 C and 900 C (samples CNTb and CNTc were obtained, respectively). Gas phase activation was also employed on CX to generate oxygen groups on the surface by heating to 400 C at the heating rate of 10 C min1 using 5% O2 (in N2) and keeping isothermally until 25% burn-off was achieved (sample CXd).
caused the adsorption process to be too fast to be monitored. Volumes of 50 and 75 mL of 20 mg L1 of aqueous solutions of CPX were tested with 10 and 5 mg carbon samples, respectively, but CPX was still removed too fast to be monitored. Finally, the suitable experimental conditions were set as described below.
2.3.
2.4.2.
2.3.1.
Characterisation of carbon materials Textural characterisation
Nitrogen adsorption isotherms at 196 C were determined in a Quantachrom NOVA 4200e apparatus. In a typical experiment, around 100 mg of sample were used and degasification was carried out for 3 h at 160 C. The specific surface area (SBET) was calculated by the BrunauereEmmetteTeller (BET) equation (Brunauer et al., 1938), the total pore volume (Vp) was determined at P/P0 ¼ 0.98, the average mesopore width (L) by the BarretteJoynereHalenda (BJH) method (Barrett et al., 1951), and micropore volume (Vmicro) and the external area (Sexternal) were determined by the t-method using an appropriate standard isotherm (Rodrı´guez-Reinoso and Linares-Solano, 1989).
2.3.2.
Surface characterisation
2.3.2.1. Temperature programmed desorption. Temperature programmed desorption (TPD) experiments were performed in a fully automated AMI-200 Catalyst Characterisation Instrument (Altamira Instruments), equipped with a quadrupole mass spectrometer (Dymaxion 200 amu, Ametek). In a typical TPD experiment, around 100 mg of sample was placed in a U-shaped quartz tube located inside an electrical furnace and subjected to a 5 C min1 heating rate up to 1100 C, under a Helium flow of 30 cm3 min1. Desorbed CO and CO2 were monitored by mass spectrometry.
2.3.2.2. Determination of the point of zero charge. The point of zero charge (pHPZC) was determined following a method proposed in the literature (Rivera-Utrilla et al., 2001). Portions of 20 mL of 0.01 M NaCl solution were prepared in different flasks. Their pH was adjusted with addition of 0.1 M solutions of NaOH or HCl to the desired values between 2 and 11. When the pH value was constant, 20 mg of carbon sample was added to each flask. The flasks were then shaken for 20 h to reach equilibrium. Blank tests were also made without carbon samples to eliminate other influences. After shaking, the pH of the blank test was measured and designated as pHinitial. The pHpzc value is the point where the curve pHfinal vs pHinitial crosses the line pHinitial ¼ pHfinal.
2.4.
Adsorption
2.4.1.
Preliminary tests
Ciprofloxacin (CPX) was supplied by Fluka Chemie GmbH. Some experiments were conducted to relate the optimal concentration of adsorbate (CPX) and adsorbent (carbon materials) with the appropriate equilibrium time. It was observed that CPX is easily adsorbed by carbon materials. Immediate adsorption was observed when 20 mL of an aqueous solution of CPX with 20 mg L1 was put in contact with 20 mg of carbon samples. The use of lower concentrations
Kinetics of adsorption
The kinetics of adsorption was determined for the untreated carbon samples (AC, CX and CNT). The solutions of CPX were sonicated to enhance solubility. About 50 mg of carbon samples (particle size < 100 mm, in order to eliminate mass transfer limitations), were introduced into closed flasks containing 1 L of the aqueous solutions and shaken for 72 h to reach equilibrium, at 25 C, with a pH of 5 (pH of water). Small volumes of samples (with no carbon suspension), obtained by decantation (or in some cases, like with the CNT samples, by filtration), were periodically taken from the flasks. This set of experiments was repeated with different initial amounts of CPX (varying from 3 to 30 mg L1). The concentrations were measured with a Jasco V-550 UVevisible spectrophotometer. The wavelength where the absorbance of CPX solution is the maximum was previously found to be at 270 nm. The decrease of this peak showed the reduction of CPX concentration in solution, indicating that it was being adsorbed on the carbon materials. By varying the concentration, a calibration curve was obtained. No considerable changes in the pH were found with time. Selected experiments were repeated and results were found to be consistent. Also blank tests were carried out showing that there was no degradation during the adsorption time, that is, the solution absorbance at 270 nm did not decrease in the absence of an adsorbent. The non-linear fitting for the kinetic models was carried out in the Origin software using the Levenberg-Marquardt routine. The 1st order and 2nd order kinetic models (expressed by Eq. (1) and Eq. (2)) were fitted to the experimental points. dq ¼ k1 qe q dt
(1)
2 dq ¼ k2 qe q dt
(2)
where q (mgCPx gC1) is the concentration of CPX in the solid phase, qe (mgCPx gC1) is the equilibrium concentrations of CPX in the solid phase, t (h) is time, k1 (h1) and k2 (gC mgCPx1 h1) are rate constants for the 1st order and 2nd order models, respectively. The concentration of CPX in the solid, q (mgCPx gC1), can be obtained by Eq. (3): q¼
ðc0 cÞV mc
(3)
where c (mgCPx L1) is CPX concentration in the solution, c0 (mgCPx L1) is the initial concentration, V (L) is the volume of adsorptive solution, and mC (mg) is the carbon sample mass. Upon integration, Eqs. (1) and (2) may be written as: q ¼ qe 1 ek1 t
(4)
4586
q¼
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 8 3 e4 5 9 1
k2 q2e t 1 þ k2 qe t
2.4.3.
(5)
Adsorption isotherms
Isotherms were obtained, for each carbon sample, by measuring the CPX concentration at equilibrium. The modified carbon materials were also tested, in order to analyze the influence of the surface chemistry on the adsorption capacity. About 2.5 mg of carbon materials and 50 cm3 of adsorptive aqueous solution of CPX were introduced into well-closed flasks and shaken for 72 h, at 25 C. Then, the absorbance at 270 nm was measured, from the decanted or filtered solutions, in order to obtain the CPX concentrations. About 2e3 days were needed in order to reach equilibrium. The classic equations of Langmuir (Eq. (6)) and Freundlich (Eq. (7)) were used to fit the experimental equilibrium adsorption data (through the linear fitting to their linearized forms): qe ¼ qm
bce 1 þ bce
qe ¼ Kce1=n
(6)
(7)
These equations can be linearized to achieve Eqs. (8) and (9) respectively: ce 1 1 þ ¼ ce qe qm b qm
(8)
1 ln qe ¼ ln K þ ln ce n
(9)
where ce (mgCPx L1), qe (mgCPx gC1) are the equilibrium concentrations of CPX in the liquid and solid phases, respectively, qm (mgCPx gC1) is the maximum adsorption capacity according to the Langmuir model, b is a constant associated with the adsorption intensity for Langmuir model, while K and n are constants associated with the adsorption capacity and adsorption intensity for Freundlich model.
3.
Results and discussion
3.1.
Textural characterisation of carbon materials
The textural parameters of the carbon materials are shown in Table 1. AC shows the highest surface area among the starting materials. The xerogels exhibited mesoporous properties with large pore sizes. Carbon nanotubes have a cylindrical structure and the pores result from the free space in the bundles, so they show lower surface areas than the other carbon materials. By comparing the parameters of oxidised activated carbon (ACa) and carbon xerogel (CXa) samples with those of the parent materials (AC and CX, respectively), it was observed that liquid phase activation slightly decreased the surface areas and micropore volumes, probably due to some pore wall collapse or to the presence of numerous oxygen-containing surface groups, which might partially block the access of N2 molecules to the micropores. Nitric acid consumes large numbers of carbon atoms and changes the structure of pores, merging some of them together. On the other hand, gas phase activation significantly resulted in more exposed surface areas and volumes as observed for CXd. This behavior is expected, as formation of new pores and opening of otherwise inaccessible pores are possible (Mahata et al., 2008). Smaller changes were observed following the thermal treatment (at 400 and 900 C), but an increase in the micropore volume was observed, which resulted from the thermal decomposition of the oxygen surface groups that might be either inside the micropores or blocking the entrances of some of them. However, these observations are not valid for the carbon nanotube samples tested. A possible reason is that oxidation may increase the amount of surface groups on the defects of the carbon basal planes, or expose the inside of some CNTs by opening some of their endcaps. Therefore, nitric acid treatments on CNTs can significantly increase the surface areas of the samples (Gonc¸alves et al., 2010; Gotovac et al., 2007).
Table 1 e Description and characterisation of carbon samples: surface area (SBET), total pore volume (Vp), average mesopore width (L), micropore volume (Vmicro), external area (Sexternal), point of zero charge (pHPZC), and amounts of CO and CO2 desorbed, as determined by temperature programmed desorption (TPD). Vp L Vmicro Sexternal pHPZC CO CO2 Sample Description SBET AC ACa ACb ACc CX CXa CXb CXc CXd CNT CNTa CNTb CNTc
Microporous activated carbon NORIT ROX 0.8 AC oxidised (with nitric acid under reflux) ACa heat treated at 350 C in N2 ACa heat treated at 900 C in N2 Carbon xerogel prepared at pH 6 CX oxidised (with nitric acid under reflux) CXa heat treated at 350 C in N2 CXa heat treated at 900 C in N2 CX heat treated at 400 C in 5% O2 (and 95% N2) Multi-walled carbon nanotubes Nanocyl-3100 CNT oxidised (with nitric acid under reflux) CNTa heat treated at 400 C in N2 CNTa heat treated at 900 C in N2
(m2/g)
(cm3/g)
(nm)
(cm3/g)
(m2/g)
974 914 960 1013 604 570 605 786 932 257 400 432 456
0.67 0.62 0.66 0.68 0.91 0.80 0.81 1.07 1.21 2.89 1.89 1.77 2.13
e e e e 13.7 18.8 18.4 18.8 13.7 e e e e
0.348 0.324 0.337 0.350 w0 0.035 0.046 0.051 0.084 w0 w0 w0 w0
260 247 267 283 604 512 529 705 789 257 400 432 456
7.50 4.80 6.45 7.45 6.10 3.00 4.21 7.45 6.14 7.30 4.21 6.89 7.40
(mmol/g)
(mmol/g)
740 3558 3417 658 564 4790 4976 602 6643 194 1686 1451 205
205 1703 865 80 118 3113 1702 99 1560 70 846 260 19
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 8 3 e4 5 9 1
3.2.
Surface chemistry characterisation
3.2.1.
Temperature programmed desorption
Figs. 2e4 show the TPD spectra of the different samples corresponding to original materials (activated carbon, carbon xerogel and carbon nanotubes, respectively). The total amounts of CO and CO2 released, determined by curve integration, are presented in Table 1. Comparing the surface-modified AC samples with the corresponding starting material (samples AC, ACa, ACb, and ACc), different amounts of CO and CO2 were released upon TPD, corresponding to the decomposition of surface groups, such as carboxylic acids, anhydrides, lactones, phenol, and carbonyls formed during the activation process. An increase in carboxylic acid groups was observed with the nitric acid treatment (liquid phase activation), for sample ACa. As for the heat treated samples (ACb and ACc), the desorbed amounts of CO and CO2 decreased with temperature increase. The reason is that the surface groups are selectively decomposed at different temperatures. Therefore, on the samples treated at higher temperatures (ACc, and also CXc and CNTc) only some basic groups (like a few carbonyls and pyrone and chromenetype structures) are retained. According to Fig. 3, the TPD profiles of carbon xerogels are similar to those of activated carbon. However, the amounts of CO and CO2 released are much larger than those observed with an activated carbon oxidised to the same extent (Figueiredo et al., 1999). This shows the higher reactivity of the polymer-based carbon, which is due to the presence of a larger
Fig. 2 e TPD spectra of activated carbon samples: (top) CO evolution, (bottom) CO2 evolution.
4587
Fig. 3 e TPD spectra for carbon xerogels samples: (top) CO evolution, (bottom) CO2 evolution.
Fig. 4 e TPD spectra for carbon nanotubes samples: (top) CO evolution, (bottom) CO2 evolution.
4588
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 8 3 e4 5 9 1
number of active sites created during its synthesis. Distinctive results between gas phase and liquid phase activation were also observed. The oxygen activated xerogel (CXd) released extensive amounts of CO and CO2 (as seen in Table 1), at temperatures higher than 400 C (Fig. 3), which correspond to phenol, lactones, anhydrides and carbonyls. CO2 peaks at lower temperature were almost absent, since the sample was treated at 400 C. On the other hand, liquid phase activation induced the formation of carboxylic acid groups. Thus, the liquid phase treatments increased the CO2 evolution at low temperatures (from 100 to 400 C), while oxidation in the gas phase induced a CO peak at higher temperatures. TPD profiles of carbon nanotubes are shown in Fig. 4. Lower amounts of CO and CO2 were detected, in comparison to the other materials. The reason most likely lies in the structure of the carbon nanotubes themselves: the surface areas available to be oxidised are smaller. Nevertheless, activation gave similar results for the expected peaks. Carboxylic and phenol groups were the majority of surface groups present, as suggested by the obtained peaks of CO2 at 300 C and CO at 600 C.
3.2.2.
Point of zero charge
Table 1 also shows the pHPZC values obtained. AC is relatively basic and the value obtained for CX was according to expected, corresponding to the pH of preparation. A neutral surface was observed on CNT, since there was no significant amount of surface groups on CNTs. As most of the oxygen-containing groups have acid character, samples ACa, CXa, and CNTa show lower pHPZC values, since they have large amounts of carboxylic groups. Since the thermal treatments remove these groups, the surface becomes less acidic as the treatment temperature increases. The highest values of pHPZC were observed for ACc, CXc, and CNTc, as expected. In samples ACc, CXc, and CNTc, groups released as CO2 were almost totally eliminated and only a small amount of oxygen remained on the carbon surface, which was released as CO at higher temperature. This can be assigned to some carbonyl, pyrone and chromene groups that have a basic character, conferring basic properties to these samples (Figueiredo et al., 1999). The other factor contributing to this basic character may be ascribed to the delocalized p electrons on the basal planes of the carbon materials (Villacan˜as et al., 2006). The gas phase activated materials showed pHPZC values similar to those of the original samples.
3.3.
Adsorption kinetics studies
The adsorption kinetics of CPX on the starting carbon-based materials was evaluated. The concentration decreased drastically in the first 5 h then a slow but gradual removal of CPX was observed until equilibrium was reached after 3 days. Usually, a longer period of time is needed for equilibrium; however, small adsorbent particle sizes were used in the experiments to eliminate mass transfer limitations, accelerating the adsorption process, since the rate of adsorption is inversely proportional to the square of the adsorbent particle ´ rfa˜o et al., 2006). diameter (O Both 1st order and 2nd order adsorption kinetics were determined using Eqs. (4) and (5). Some results of kinetic
fittings are shown in Fig. S1 of Supplementary data, as plots of q vs contact time. The parameters found for the different models are listed in Tables S1 and S2 of Supplementary data. The qe values obtained by both models are in agreement. It was shown that the 2nd order model fits better most of the adsorption data, since the line corresponding to the model fitting is closer to the experimental points than that from the 1st order model, as can be seen in some examples of Fig. S1, and consequently the R2 values (shown in Tables S1 and S2) are higher in the case of 2nd order fitting. For the same initial concentration of CPX, the equilibrium concentration in solid phase (qe) increases in the order of CX < CNT < AC. These observations will be further discussed in Section 3.4, comparing with the results from surface-modified carbon materials. Since the 2nd order fitting curves were more consistent with the experimental data, only the 2nd order model parameters will be discussed here. High values of k were apparently found for c0 < 10 mgCPx L1. Generally, the rate constant associated with the adsorption kinetics decreases for larger concentrations due to limited adsorption sites on carbon materials. A constant value of k would supposedly be observed if higher concentrations of CPX were used.
3.4.
Equilibrium adsorption isotherms
The adsorption isotherms of surface-modified carbon materials were investigated, along with the samples discussed in Section 3.2. The classic equations of Langmuir (Eq. (6)) and Freundlich (Eq. (7)) were used to fit the experimental equilibrium adsorption data. All the parameters are listed in Table S3, while the fitting curves are shown in Figs. S2eS4 of supplementary data. It was shown that the Langmuir equation offers the best fit to the adsorption isotherms for most samples, since the values of R2 obtained are higher in the case of the Langmuir model (Table S3). Only sample CX showed a higher R2 using the Freundlich model (0.996) when compared with the Langmuir model (0.984). Fig. 5 shows that the maximum adsorption capacity (qm) increases as CX < CNT < AC, in terms of untreated carbon samples. For the activated carbons, the order is: ACa < ACb < AC < ACc. For carbon xerogels CXa < CX < CXb < CXc < CXd while for carbon nanotubes is CNT < CNTa z CNTb z CNTc. High adsorption capacities ranging from approximately 60 to 300 mgCPx gC1 were obtained (for CXa and ACc, respectively). Being carbon materials able to adsorb such high amounts of CPX, they are also certainly adequate to treat real effluents, that contain concentrations of pollutant ranging from < 1 mg L1 to around 31 mg L1 (Carmosini and Lee, 2009), as explained in Section 1. Similar values were obtained for the adsorption of acetaminophen on activated carbons derived from several wastes, i.e., 113 to 267 mg gC1, for a sample obtained from plastic waste and commercial material produced from the physical activation of wood, respectively (Cabrita et al., 2010). However, lower values, ranging from 35 to 177 mg gC1 were obtained for the adsorption of aniline on commercial activated carbons, and from 48 to 103 mg gC1 for sulfanilic acid using the same materials (Faria et al., 2008). The amounts of aminetriazol adsorbed on activated carbon fibres and activated carbons
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 8 3 e4 5 9 1
4589
Fig. 5 e Adsorption capacity (qm) of carbon samples.
ranged from approximately 15 to 20 mg gC1 (Moreno-Castilla, 2004), which is much lower than what we obtained for CPX (Fig. 5). Previous works reported values ranging from 85 to 339 mg gC1 for phenol adsorption, 93 to 419 mg gC1 for aniline and 209 to 3119 mg gC1 for nitrobenzene (Villacan˜as et al., 2006) on activated carbons. In general, larger adsorption values were obtained at pH 7 (Faria et al., 2008; Villacan˜as et al., 2006). Considering that we did not make any pH correction (since we used the pH of water), our values can be considered quite high. Other authors used Fenton and photo-Fenton processes and achieved a degradation efficiency of 91% by Fenton oxidation and 100% by photo-Fenton within 120 min for 1.5 mgCPx (100 mL of a concentration of 15 mgCPx L1) (Sun et al., 2009). In the present work, we would be able to achieve the same immediate results on water treatment with 100 mg of carbon (a very cheap material), as these authors got after 2 h. Other authors also showed that activated carbon had higher adsorptive capacity for CPX than talc (Ibezim et al., 1999). Photodegradation processes, although being able to immediately reduce the amount of CPX in surface water, have the inconvenient to ultimately cause the sorption to organic material to prolong the environmental half-life of CPX (Belden et al., 2007). Degradation by ozonation seems to be promising, but further optimisation is still needed (De Witte et al., 2010, 2008). HOCl also reacts very rapidly with CPX, forming a chloramine intermediate that spontaneously decays in aqueous solution by concerted piperazine fragmentation, being promising for CPX degradation (Dodd et al., 2005). Transformation of fluoroquinolones like CPX was found under typical ClO2 disinfection conditions, however antibacterial activity was not eliminated since little destruction at the quinolone ring was found (Wang et al., 2010). The advantage of adsorption processes is that CPX is efficiently removed from water. In this work, the adsorption capacity of the microporous sample AC is high due to the large surface area and pore volume (see Table 1). Fig. 1 shows two views of the CPX molecule. Since it has a planar configuration and approximate dimensions of
13.5 A3 A 7.4 A, in principle it should be able to penetrate into the pores of AC. As the micropores are normally “slit” shaped, the molecule should be able to enter “sideways”. CPX can also enter the pores of CX (which are cylindrical, but their dimensions are larger than those of the molecule) and CNT openings (as their internal diameter is approximately 40 A (Gonc¸alves et al., 2010)). The pKa values for CPX are 5.90 0.15 (for the carboxylic acid group) and 8.89 0.11 for the basic-N-moiety, so it can exist as a cation, zwitterion and anion under typical soil and water pH conditions (Carmosini and Lee, 2009). Protonation of the basic-N can facilitate cation exchange with negatively charged sorbents, and complexation with positively charged sorbents can occur with the deprotonated carboxylate and the keto-O groups (Carmosini and Lee, 2009). The results suggest though that CPX adsorbs better on a less acidic surface with a small amount of carboxylic groups. CPX has many electron-rich site groups such as carboxylic, ketone, and cyclic amide groups, which might form complexes with electron-withdrawing groups upon deprotonation. CX showed the lowest adsorption capacity among the starting materials. From the results obtained, it seems that the nitric acid treatment decreases the adsorption capacity (comparing AC vs ACa and CX vs CXa) due to the molecular interaction discussed above, as observed with different pollutants (Faria et al., 2008; Villacan˜as et al., 2006). Adsorption on basic materials is favored, mainly due to the dispersive interactions between the delocalized p electrons in the carbon basal planes and the free electrons in the CPX molecules (namely those of the aromatic rings). The surface of the samples containing carboxylic acid groups, resulting from liquid phase activation, are negatively charged at the experimental pH around 5 (pH > pHPZC), thus repelling the partially deprotonated CPX, resulting in lower adsorption. On the other hand, the gas activation of sample CXd led to a drastic increase in the adsorption capacity, since the surface was functionalised mainly with phenol and carbonyl groups that are weak electron-acceptors. In addition, as pH is lower than the pHPZC of the samples, the surface will be positively charged and some attractive electrostatic forces
4590
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 8 3 e4 5 9 1
will be involved. Moreover, CXd had relatively larger surface area and pore volume available for adsorption than other CX materials (Table 1). The thermal treatments provide more active sites for adsorption, as described in Section 3.2.1. Therefore, when some surface groups are removed through these treatments, there is an increase of the pHPZC (Table 1), favoring the electrostatic attraction between the positively charged carbon surface and the deprotonated CPX (as observed for ACa vs ACb vs ACc, and CXa vs CXb vs CXc). Moreover, the thermal treatments increase the surface area, which may also contribute to higher adsorption. In contrast, carbon nanotubes behave differently, since the acid treatment increases the adsorption capacity, unlike what was found by other authors for other pollutants (Cho et al., 2008; Li et al., 2011; Sheng et al., 2010). Due to the unique molecular structure of CNTs, liquid phase oxidation increases the surface groups on the defects of the carbon nanotubes basal planes, yielding more open tubes by opening some of the CNTs endcaps. As a result, the adsorption capacity increases after oxidation due to the substantial increase of the surface area. On the other hand, a thermal treatment did not significantly enhance adsorption because the surface groups originated by oxidation were very few, as seen in TPD profiles. In this case, results can be explained by the textural properties. pHPZC also influences the adsorption capacity. It is known that pHPZC is related to the contents of surface groups. Therefore, its contribution is well explained above. In summary, the mechanism of adsorption is related with the interaction/hindrance of surface groups, and with the surface area, where CPX can participate in electronic interactions with the carbon basal planes.
4.
Conclusions
The adsorption capacity of ciprofloxacin on three different types of carbon-based materials (activated carbons, carbon nanotubes and carbon xerogels), subjected to different surface treatments (in order to test the influence of surface chemistry) was determined. High adsorption capacities ranging from approximately 60 to 300 mgCPx gC1 were obtained (for oxidised carbon xerogel, and oxidised and thermally treated activated carbon Norit ROX 8.0, respectively). Being carbon materials able to adsorb such high amounts of CPX, they are also certainly adequate to treat real effluents, that contain concentrations of pollutant ranging from <1 mg L1 to around 31 mg L1. Further studies dealing with the use of the best adsorbent samples, simulating real conditions, or even in samples from real effluents, in continuous operation, would be the last step to validate this technology. In general, it was found that the nitric acid treatment of samples has a detrimental effect in their adsorption capacities, whereas the thermal treatments, especially at 900 C after oxidation, enhance the adsorption performance. This is assigned to the positive effect of the surface basicity. The kinetic curves obtained were fitted using 1st or 2nd order models, and the Langmuir and Freundlich models were used to describe the isotherms obtained. The 2nd order and the
Langmuir models, respectively, were shown to provide the best fittings.
Acknowledgements Authors thank Fundac¸a˜o para a Cieˆncia e a Tecnologia (FCT), for financial support: CIENCIA 2007 program and project PTDC/QUI-QUI/100682/2008, financed by FCT and FEDER in the context of Programme COMPETE. The International Association for the Exchange of Students for Technical Experience (GS1) (IAESTE) Portugal and FCT are acknowledged for supporting the internship of Thanakrit Thavorn-amornsri (ref. PT/2010/49) at the LA LSRE/LCM of the University of Porto.
Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.watres.2011.06.008.
references
Barrett, E.P., Joyner, L.G., Halenda, P.P., 1951. The determination of pore volume and area distributions in porous substances. I. Computations from nitrogen isotherms. Journal of the American Chemical Society 73 (1), 373e380. Belden, J.B., Maul, J.D., Lydy, M.J., 2007. Partitioning and photodegradation of ciprofloxacin in aqueous systems in the presence of organic matter. Chemosphere 66 (8), 1390e1395. Bhandari, A., Close, L.I., Kim, W., Hunter, R.P., Koch, D.E., Surampalli, R.Y., 2008. Occurrence of ciprofloxacin, sulfamethoxazole, and azithromycin in municipal wastewater treatment plants. Practice Periodical of Hazardous, Toxic, and Radioactive Waste Management 12, 275e281. Brunauer, S., Emmett, P.H., Teller, E., 1938. Adsorption of gases in multimolecular layers. Journal of the American Chemical Society 60 (2), 309e319. Cabrita, I., Ruiz, B., Mestre, A.S., Fonseca, I.M., Carvalho, A.P., Ania, C.O., 2010. Removal of an analgesic using activated carbons prepared from urban and industrial residues. Chemical Engineering Journal 163 (3), 249e255. ´ rfa˜o, J.J.M., Figueiredo, J.L., Carabineiro, S.A.C., Pereira, M.F.R., O 2011. Activated Carbon: Classifications, Properties and Applications. Nova Publishers. Carmosini, N., Lee, L.S., 2009. Ciprofloxacin sorption by dissolved organic carbon from reference and bio-waste materials. Chemosphere 77 (6), 813e820. Cho, H.-H., Smith, B.A., Wnuk, J.D., Fairbrother, D.H., Ball, W.P., 2008. Influence of surface oxides on the adsorption of naphthalene onto multiwalled carbon nanotubes. Environmental Science & Technology 42 (8), 2899e2905. De Witte, B., Van Langenhove, H., Demeestere, K., Saerens, K., De Wispelaere, P., Dewulf, J., 2010. Ciprofloxacin ozonation in hospital wastewater treatment plant effluent: effect of pH and H2O2. Chemosphere 78 (9), 1142e1147. De Witte, B., Dewulf, J., Demeestere, K., Van De Vyvere, V., De Wispelaere, P., Van Langenhove, H., 2008. Ozonation of ciprofloxacin in water: HRMS identification of reaction products and pathways. Environmental Science & Technology 42 (13), 4889e4895.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 8 3 e4 5 9 1
Diwan, V., Tamhankar, A.J., Aggarwal, M., Sen, S., Khandal, R.K., Lundborg, C.S., 2009. Detection of antibiotics in hospital effluents in India. Current Science 97 (12), 1752e1755. Diwan, V., Tamhankar, A.J., Khandal, R.K., Sen, S., Aggarwal, M., Marothi, Y., Iyer, R.V., Sundblad-Tonderski, K., Lundborg-, C.S., 2010. Antibiotics and antibiotic-resistant bacteria in waters associated with a hospital in Ujjain, India. BioMedCentral Public Health 10, 414e422. Dodd, M.C., Shah, A.D., Von Gunten, U., Huang, C.H., 2005. Interactions of fluoroquinolone antibacterial agents with aqueous chlorine: reaction kinetics, mechanisms, and transformation pathways. Environmental Science & Technology 39 (18), 7065e7076. ´ rfa˜o, J.J.M., Figueiredo, J.L., Pereira, M.F.R., 2008. Faria, P.C.C., O Adsorption of aromatic compounds from the biodegradation of azo dyes on activated carbon. Applied Surface Science 254 (11), 3497e3503. ´ rfa˜o, J.J.M., 1999. Figueiredo, J.L., Pereira, M.F.R., Freitas, M.M.A., O Modification of the surface chemistry of activated carbons. Carbon 37 (9), 1379e1389. ´ rfa˜o, J.J.M., 2007. Figueiredo, J.L., Pereira, M.F.R., Freitas, M.M.A., O Characterization of active sites on carbon catalysts. Industrial & Engineering Chemistry Research 46 (12), 4110e4115. Giger, W., Alder, A.C., Golet, E.M., Kohler, H.-P.E., McArdell, C.S., Molnar, E., Siegrist, H., Suter, M.J.-F., 2003. Occurrence and fate of antibiotics as trace contaminants in wastewaters, sewage sludges, and surface waters. Chimia 57 (9), 485e491. Golet, E.M., Alder, A.C., Giger, W., 2002. Environmental exposure and risk assessment of fluoroquinolone antibacterial agents in wastewater and river water of the glatt valley watershed, switzerland. Environmental Science & Technology 36 (17), 3645e3651. ´ rfa˜o, J.J.M., Pereira, M.F.R., 2010. Gonc¸alves, A.G., Figueiredo, J.L., O Influence of the surface chemistry of multi-walled carbon nanotubes on their activity as ozonation catalysts. Carbon 48 (15), 4369e4381. Gotovac, S., Yang, C.M., Hattori, Y., Takahashi, K., Kanoh, H., Kaneko, K., 2007. Adsorption of polyaromatic hydrocarbons on single wall carbon nanotubes of different functionalities and diameters. Journal of Colloid and Interface Science 314 (1), 18e24. Ibezim, E.C., Ofoefule, S.I., Ejeahalaka, C.N., Orisakwe, O.E., 1999. In vitro adsorption of ciprofloxacin on activated charcoal and talc. American Journal of Therapeutics 6 (4), 199e201. Karthikeyan, K.G., Meyer, M.T., 2006. Occurrence of antibiotics in wastewater treatment facilities in wisconsin, USA. Science of The Total Environment 361 (1e3), 196e207. Larsson, D.G.J., de Pedro, C., Paxeus, N., 2007. Effluent from drug manufactures contains extremely high levels of pharmaceuticals. Journal of Hazardous Materials 148 (3), 751e755. Li, X., Zhao, H., Quan, X., Chen, S., Zhang, Y., Yu, H., 2011. Adsorption of ionizable organic contaminants on multi-walled
4591
carbon nanotubes with different oxygen contents. Journal of Hazardous Materials 186 (1), 407e415. Mahata, N., Pereira, M.F.R., Sua´rez-Garcı´a, F., Martı´nez-Alonso, A., Tasco´n, J.M.D., Figueiredo, J.L., 2008. Tuning of texture and surface chemistry of carbon xerogels. Journal of Colloid and Interface Science 324 (1, 2), 150e155. Moreno-Castilla, C., 2004. Adsorption of organic molecules from aqueous solutions on carbon materials. Carbon 42 (1), 83e94. ´ rfa˜o, J.J.M., Silva, A.I.M., Pereira, J.C.V., Barata, S.A., Fonseca, I.M. O , Faria, P.C.C., Pereira, M.F.R., 2006. Adsorption of a reactive dye on chemically modified activated carbons-influence of pH. Journal of Colloid and Interface Science 296 (2), 480e489. Radovic, L.R., Moreno-Castilla, C., Rivera-Utrilla, J., 2000. In: Radovic, L.R. (ed.) Marcel Dekker, New York. Renew, J.E., Huang, C.-H., 2004. Simultaneous determination of fluoroquinolone, sulfonamide, and trimethoprim antibiotics in wastewater using tandem solid phase extraction and liquid chromatography-electrospray mass spectrometry. Journal of Chromatography A 1042 (1, 2), 113e121. Rivera-Utrilla, J., Bautista-Toledo, I., Ferro-Garcı´a, M.A., MorenoCastilla, C., 2001. Activated carbon surface modifications by adsorption of bacteria and their effect on aqueous lead adsorption. Journal of Chemical Technology & Biotechnology 76 (12), 1209e1215. Rodrı´guez-Reinoso, F., Linares-Solano, A., 1989. Chemistry and Physics of Carbon. Dekker, M., New York. Samant, P.V., Gonc¸alves, F., Freitas, M.M.A., Pereira, M.F.R., Figueiredo, J.L., 2004. Surface activation of a polymer based carbon. Carbon 42 (7), 1321e1325. Serp, P., Figueiredo, J.L. (Eds.), 2009. Carbon Materials for Catalysis. Wiley, Hoboken, New Jersey. Sheng, G.D., Shao, D.D., Ren, X.M., Wang, X.Q., Li, J.X., Chen, Y.X., Wang, X.K., 2010. Kinetics and thermodynamics of adsorption of ionizable aromatic compounds from aqueous solutions by as-prepared and oxidized multiwalled carbon nanotubes. Journal of Hazardous Materials 178 (1e3), 505e516. Sun, S.-P., Guo, H.-Q., Ke, Q., Sun, J.-H., Shi, S.-H., Zhang, M.-L., Zhou, Q., 2009. Degradation of antibiotic ciprofloxacin hydrochloride by photo-fenton oxidation process. Environmental Engineering Science 26 (4), 753e759. Tessonnier, J.-P., Rosenthal, D., Hansen, T.W., Hess, C., Schuster, M.E., Blume, R., Girgsdies, F., Pfa¨nder, N., Timpe, O., Su, D.S., Schlo¨gl, R., 2009. Analysis of the structure and chemical properties of some commercial carbon nanostructures. Carbon 47 (7), 1779e1798. ´ rfa˜o, J.J.M., Figueiredo, J.L., 2006. Villacan˜as, F., Pereira, M.F.R., O Adsorption of simple aromatic compounds on activated carbons. Journal of Colloid and Interface Science 293 (1), 128e136. Wang, P., He, Y.L., Huang, C.H., 2010. Oxidation of fluoroquinolone antibiotics and structurally related amines by chlorine dioxide: reaction kinetics, product and pathway evaluation. Water Research 44 (20), 5989e5998.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 9 2 e4 6 0 0
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Enhanced trace phosphate removal from water by zirconium(IV) loaded fibrous adsorbent Md. Rabiul Awual a,b,*, Akinori Jyo a,*, Toshihiro Ihara a, Noriaki Seko c, Masao Tamada c, Kwon Taek Lim d a
Department of Applied Chemistry and Biochemistry, Kumamoto University, Kurokami 2-39-1, Kumamoto 860-8555, Japan National Institute for Materials Science, Exploratory Materials Research Laboratory for Energy and Environment, 1-2-1 Sengen, Tsukuba, Ibaraki-ken 305-0047, Japan c Takasaki Advanced Radiation Research Institute, Japan Atomic Energy Agency, 1233 Watanuki-machi, Takasaki, Gunma 370-1292, Japan d Division of Image and Information Engineering, Pukyong National University, Busan 608-739, Republic of Korea b
article info
abstract
Article history:
This study was investigated for the trace phosphate removal at high feed flow rate by
Received 25 February 2011
ligand exchange fibrous adsorbent. The zirconium(IV) loaded bifunctional fibers containing
Received in revised form
both phosphonate and sulfonate were used as a highly selective ligand exchange adsorbent
7 June 2011
for trace phosphate removal from water. The precursory fiber of the bifunctional fibers was
Accepted 7 June 2011
co-grafted by polymerization of chloromethylstyrene and styrene onto polyethylene
Available online 16 June 2011
coated polypropylene fiber and then bifunctional fibers were prepared by Arbusov reaction followed by phosphorylation and sulfonation. Phosphate adsorption experimental work
Keywords:
was carried out in column approach. Phosphate adsorption increased with decreasing the
Bifunctional fiber
pH of feed solutions. An increase in the feeds flow rate brings a decrease in both break-
Phosphate removal
through capacity and total adsorption. The effect of competing anions on phosphate
Graft polymerization
adsorption systems was investigated. The experimental findings reveal that the phosphate
Highly selective adsorbent
adsorption was not affected in the presence of competing anions such as chloride and
Regeneration
sulfate despite the enhancement of the breakthrough points and total adsorption. Due to high selectivity to phosphate species, low concentration level of phosphate (0.22 mg/L) was removed at high feed flow rate of 450 h1 in space velocity. The adsorbed phosphate on the Zr(IV) loaded fibrous column was quantitatively eluted with 0.1 M NaOH solution and then the column was regenerated by 0.5 M H2SO4 for the next adsorption operation. During many adsorptioneelutioneregeneration cycles, no measurable Zr(IV) was found in the column effluents. Therefore, the Zr(IV) loaded bifunctional fibrous adsorbent is to be an effective means to treat wastewater to prevent eutrophication in the receiving water bodies for long time without any deterioration. ª 2011 Elsevier Ltd. All rights reserved.
* Corresponding authors. Department of Applied Chemistry and Biochemistry, Kumamoto University, Kurokami 2-39-1, Kumamoto 860-8555, Japan. Tel.: þ81 96 342 3871; fax: þ81 96 342 3679. E-mail addresses: [email protected], [email protected] (Md.R. Awual), [email protected] (A. Jyo). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.06.009
4593
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 9 2 e4 6 0 0
1.
Introduction
Phosphorus is an essential element for living organisms. However, huge volume of wastewater production containing high concentration of phosphate will deteriorate natural ecosystems and natural water quality and also responsible for eutrophication problem of rivers, lakes and inland seas (Ryther and Dustan, 1971; Conley et al., 2009; Dennison et al., 1993; Morse et al., 1998; Mullan et al., 2006; Onyango et al., 2007; Kabayama et al., 2003). Therefore, phosphate removal from water containing nutrients at high concentration levels is important to conserve natural aquatic environment. The USEPA has recommended a maximum level of phosphorus in water to be less than 50 mg/L to prevent eutrophication problem and Florida Everglades Forever Act also recommended a new mandate of 10 mg/L of phosphorus in water (USEPA, 1986; Florida Everglades Forever Act, 1994). Several chemical and biological methods have been proposed for phosphate removal from water and wastewater. Coagulation and biological methods are used in industries but these treatment processes are unable to satisfy the mandate level or reduce it to near zero or below 10 mg/L (Stante et al., 1997; Zhao and SenGupta, 1998; Bektas et al., 2004; Vasudevan et al., 2008; Namasivayam and Prathap, 2005; Babatunde et al., 2008; Yildiz, 2004; de-Bashan and Bashan, 2004). Therefore, extensive researches have been investigated to develop suitable methods based on stable running, simple operation technique, high selectivity and excellent kinetic performances. Ion exchange adsorbents, electrodialysis, and specially ligand exchangers are possessing all of these characteristics. In ligand exchange adsorbents, hard Lewis metal ions have been immobilized on chelating resins or fibers; here
CH 2
Electron irradiation
Zr(IV), Fe(III), Al(III), Cu(II), Co(II), La(III), Mo(VI) and etc. have been used as hard Lewis acid cation (Helfferich, 1962; Zhao and SenGupta, 1998; Babatunde et al., 2008; Kuzawa et al., 2006; Seko et al., 2004; Zhu and Jyo, 2005; Chimenos et al., 2003; Seida and Nakano, 2002; Lee et al., 2003; Chitrakar et al., 2006; Wu et al., 2007; Jyo et al., 2008; Genz et al., 2004). Among the possible separation and purification techniques, adsorption by selective adsorbent shows promise in being among the most attractive and efficient methods for purification and separation of trace anion contaminants. Moreover, these prefer phosphate in the presence of competing anions such as chloride, nitrate, bicarbonate, and sulfate (Zhao and SenGupta, 1998; Jyo et al., 2008; Hamoudi et al., 2007). Fibrous adsorbents show excellent kinetic performances than granular adsorbents and are able to take up phosphate, arsenic and several cations at high feed flow rate (Awual et al., 2008). In addition, these are completely regenerated in initial form and usable for many cycles without loss of their original significant performances (Awual et al., 2011). Undoubtedly, ligand exchange adsorbents are highly selective to phosphate but they are unable to take up phosphate at high feed flow rate more than 40 h1 in low concentration level (Zhao and SenGupta, 1998; Zhu and Jyo, 2005). However, we are interested to remove low concentration phosphate in the presence of competing anions at high feed flow rate. There are several multi-functional resins/fibers prepared but their preparations, higher thermal stability, specific selectivity are different (Hamabe et al., 2009; Zhang et al., 2008; Jyo et al., 2010). Here, we have developed a bifunctional cation exchange fiber having both phosphonate and sulfonate groups (Scheme 1) (Jyo et al., 2005). Therefore, this work was planned to evaluate the behavior of Zr(IV) loaded bifunctional fiber in column-
HC
CH2 CH HC Liquid phase graft copolymerization with CH 2 CH 2 chloromethylstyrene and styrene CH 2
x
CH2 CH y
CH 2
CH 2 CH 2
H2C
CH 2 PPPE
CH2Cl
H2C Precursory fiber
Radical generation
Arbusov reaction
HC
CH2 CH x
CH2 CH y
CH2 CH y
CH 2
CH 2 1) Sulfonation with ClSO 3H
CH 2 H2C
CH2 CH x
HC
CH2
O
P
OC 2H5
OC 2H5
2) Hydrolysis With Conc. HCl H2C
CH 2
SO3H
CH2
O
P
OH
OH
Bifunctional fiber (FPS)
Scheme 1 e Synthetic route for the preparation of bifunctional fiber containing both phosphonate and sulfonate groups (FPS).
4594
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 9 2 e4 6 0 0
mode phosphate adsorption under various conditions. The main objectives of this study were to evaluate (i) selectivity to phosphate over competing anions such as chloride and sulfate in neutral pH region, (ii) loss of Zr(IV) in the column effluents both in adsorption and elution operations, (iii) kinetic performance to take up phosphate at high feed flow rate, and (iv) the regeneration method for long-term repeated use.
2.
Materials and methods
2.1.
Materials
All materials and chemicals were of analytical grade and used as purchased without further purification. Ultra-pure water prepared by a Milli-Q Academic-A10 (Nippon Millipore Co., Tokyo, Japan) was used throughout. Phosphoric acid (85 wt%), Zr(SO4)2/4H2O and Na2H2PO4/12H2O are guaranteed grade and were obtained from Wako Pure Chemical Industries Ltd., Osaka, Japan. Sulfuric acid, nitric acid, perchloric acid and sodium hydroxide were also purchased from Wako Pure Chemical Industries Ltd., Osaka, Japan. Chloromethylstyrene (CMS) and Styrene (ST) were co-grafted onto polyethylene coated polypropylene fibers (diameter 12 mm, length 3.8 cm) by means of electron irradiation induced liquid phase graft polymerization technique in which equimolar mixture of CMS and ST was used. Scheme 1 shows the major steps for the preparation of bifunctional cation exchange fiber (FPS) by means of radiation induced graft polymerization technique and followed by Arbusov reaction, chlorosulfonation and refluxed with concentrated HCl acid. The detailed procedures for the graft polymerization and additional properties of FPS are available elsewhere (Jyo et al., 2005). The acid capacity, phosphorus and sulfur contents were determined by reported methods (Maeda and Egawa, 1984). Here, we simply described the acid capacity and phosphorus content measurement for reader understanding. Acid capacity was measured in batch approached. The FPS sample (0.20 g) was taken into an Erlenmeyer flask, and a solution (50 ml) containing NaOH (0.1 M) and NaNO3 (1.0 M) was added to the flask. The flask was shaken at 30 C for 24 h with a mechanical shaker at a constant agitation speed of 110 rpm. Then an aliquot (10 ml) of the supernatant was titrated with a standardized 0.1 M HCl. The acid capacity was calculated from the decrease in the concentration of hydroxide ion in the supernatant. The phosphorus content was measured by a colorimetric method. The FPS sample (50 mg) was completely digested with nitric acid and perchloric acid at elevated temperature and the resulting transparent solution was exactly diluted to be 100 mL. The concentration of phosphorus in this solution was determined by measurement of absorbance at 440 nm after developing yellow color with the addition of ammonium metavanadate and ammonium molybdate solutions. The acid capacity of FPS was 4.22 meq/g and its phosphorus and sulfur contents were 0.94 and 1.91 mmol/g, respectively.
2.2.
Zr(IV) loaded bifunctional fiber
Zr(IV) was immobilized onto FPS in column method. A dried sample of FPS (0.45 g) was packed into polyethylene column
(inner diameter 1.3 cm) and equilibrated for 24 h in water. Then, the fiber bed was carefully pressed with a glass rod with flat end until the height of the bed became constant. The volume of wet fiber bed in the column was 2.12 mL, which was used as the reference volume to convert the flow rate in mL/h into the space velocity (SV) in h1 as well as volumes of supplied solutions or water to the column in mL into bed volumes (BV) in mL/mL-fiber. Flow rates of all solutions and water were expressed by SV, which is designated by the ratio F/Vbed in h1; here F is flow rate of solutions or water in mL/h and Vbed is a volume of the fiber bed. All solutions or water volumes supplied to the column in adsorptioneelutioneregeneration operations were expressed by BV, which is designated by the ratio of Vsupplied/Vbed; here Vsupplied is the volume of a feed supplied to the column in mL. The Zr(IV) solution was prepared by dissolving Zr(SO4)2$4H2O into an aqueous H2SO4 (0.5 M). The Zr(IV) solution was feed to the column at a flow rate of 4 h1 in SV. Then, acidic 0.01 M Zr(IV) solution (0.5 M H2SO4) (80 mL) was fed to the column. After washing the column with 0.5 M H2SO4 (25 mL) and water (20 mL) in successive, 0.2 M NaOH solution (30 mL) was fed to the column. The total amount of immobilized Zr(IV) onto FPS was calculated from the difference of Zr(IV) concentrations between the feed solutions and column effluents. After the column was washed with water (20 mL) and 0.5 M H2SO4 (25 mL), acidic 0.01 M Zr(IV) solution (0.5 M H2SO4) (80 mL) was fed to the column for more loading of Zr(IV) onto FPS. To investigate stability and leakage of loaded Zr(IV) onto FPS, the sulfuric acid solution (0.5 M) and 0.2 M sodium hydroxide solution were used to wash column, respectively. The loading operations of Zr(IV) onto FPS (ZreFPS) were repeated three times. All effluents were collected on a fraction collector and concentration of Zr(IV) in each fraction was measured by ICPeAES (IRIS, Nippon Jarrell Ash Co. Ltd., Kyoto, Japan). The total immobilized Zr(IV) was sum of a loaded amount of Zr(IV) in each run.
2.3.
Column adsorption and elution of phosphate
Phosphate feed solutions were prepared by dissolving sodium dihydrogenphosphate (NaH2PO4$2H2O) in water and the pH of feed solutions was adjusted by dilute sulfuric acid and sodium hydroxide solution, as needed. Before phosphate adsorption operation, the column was conditioned to be the same pH as that of phosphate feed solutions. Phosphate feed solutions were supplied to the column at a given flow rate. After adsorption operation, the column was washed with water (15 bed volumes) at flow rate of 5 h1 and then elution operation was carried out by feeding 0.1 M NaOH to the column at a flow rate of 4 h1. Phosphate feed solutions were remaining above the fiber bed after adsorption operation. Therefore, column was washed with water to adsorb the remaining/floating phosphate solution by ZreFPS. After washing the column with water, the ZreFPS column was regenerated with 0.5 M H2SO4 for the next use. All column effluents were collected on a fraction collector and phosphorus and zirconium concentrations in each fraction were determined by ICPeAES. The detailed procedures of phosphorus analysis method are available elsewhere (Awual et al., 2011). The total adsorption (TA) of phosphate in the adsorption operations was calculated from the following equation:
4595
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 9 2 e4 6 0 0
TA ðmmol=gÞ ¼
Cf Vf
n X
!, m
Ci V i
(1)
i¼1
where Cf, Vf, and m are the concentrations of phosphorus (P) in the feed, the volume of the supplied feed, and amount of loaded bifunctional fiber, respectively; also Ci, Vi, and n are the concentration of P, the volume of the ith fraction, and the number of the last fraction in the adsorption operation. Breakthrough capacity (BC) of exchangers for P was calculated from the following relation: Cf Vp
BC ðmmol=gÞ ¼
n0 X
SO3
!, m
Ci V i
CH2 (2)
O O
i¼1
Here, Vp is the volume of the feed up to the breakthrough point and n0 is number of the last fraction up to the breakthrough point. Here, the breakthrough point was designated when effluent concentration was 10 mg/L or 0.01 mg/L of P. The eluted amount of phosphate (EMP) was calculated from Eq. (3) EMP ðmmol=gÞ ¼ ð1=mÞ
n00 X
OH 2 Zr4+
P O
X X
OH2 Scheme 2 e Bonding mechanism of Zr(IV) with bifunctional fibers (FPS).
! Ct Vt
(3)
t¼1
00
Here, Ct, Vt and n are the concentration of P and volume of the tth fraction, and number of the last fraction in the elution operations.
3.
Results and discussion
3.1.
Loading and stability of Zr(IV) onto FPS
Zirconium(IV) is an effective metal ion in ligand exchange adsorbent for phosphate, which is typical Hard Lewis base and can be immobilized on chelating resin even in strongly acidic media (Zhu and Jyo, 2001). The bifunctional fibers containing phosphonate and sulfonate have high affinity and immobilization capacity to Zr(IV). The immobilized Zr(IV) was strongly fixed on the fiber phase over the wide pH ranges. In the first run, the immobilized amount of Zr(IV) onto FPS was 0.56 mmol/g, which was much lower than its cation exchange capacity. Then the loading procedure was repeated three times and the amount of loaded Zr(IV) increased up to 1.18 mmol/g. The loaded amount of Zr(IV) onto FPS in the 2nd, and 3rd was 0.39, 0.23 mmol/g, respectively. Scheme 2 shows one of possible structures around the loaded Zr(IV) onto bifunctional fibers of FPS, in which Zr(IV) coordinates with phosphonate group of bifunctional fibers and water or chloride or oxygen (Suzuki et al., 2000; Balaji et al., 2005). The symbol ‘X’ represents a water molecule or a chloride ion or oxygen or sulfate. In addition, Zr(IV) solution was prepared in acidic media for maximum Zr(IV) adsorption onto FPS. The poor adsorption of Zr(IV) onto FPS was possible above pH 4.0 due to the hydrolysis of Zr(IV) as zirconium hydroxide (Balaji et al., 2005). The ZreFPS column was treated with H2SO4 solution or NaOH solution to check the stability of immobilized Zr(IV) onto FPS and did not observe any measurable loss of the loaded Zr(IV). This was clarified that Zr(IV) strongly retained onto FPS to behave as a good ligand exchanger.
3.2.
Effect of pH on phosphate adsorption
Ligand exchange process is highly pH sensitive. Therefore, the feed solutions acidity is one of the important factors affecting the performance of phosphate adsorption. Because the distribution of phosphate species changes with different pH area and the different phosphate species have different affinities toward the ZreFPS. Then, the behavior of phosphate adsorption was investigated by changing pH of feed solutions from 2.01 to 7.01. The acid dissociation constant values of phosphoric acid are pKa1 ¼ 2.16, pKa2 ¼ 7.21 and pKa3 ¼ 12.32 (Ringbom, 1963). However, pH of the natural water and municipal wastewater normally ranges from 6.5 to 7.3 and the ionic strength is ca. 0.01 mol/L (Zhao and SenGupta, 2000). Therefore, monovalent phosphate (H2PO4) is the major phosphate species in the aqueous phase under such conditions. Fig. 1 shows the breakthrough profiles of phosphate to determine the effect of feed solutions of pH and Table 1 (entry nos. 1e5) lists detailed experimental conditions and numerical results. The pH increased in feeds brought the decreasing in breakthrough points and breakthrough capacities as well as total adsorption of phosphate as judged from data of entry nos. 1e5 in Table 1. An increase in pH, hydroxide concentration increases so that interference of hydroxide with the hardest Lewis acid anion in water increases resulting in decrease in phosphate adsorption. However, this adsorbent was able to purify 470 BV of water containing 0.4 mg/L of P down to less than 0.004 mg/L of P even at neutral pH area where the dominant species was divalent phosphate (HPO42) in pH from 4.6 to 7.2. The results given in Table 1 (entry nos. 1e5) imply that univalent phosphate or divalent phosphate was taken up by ZreFPS by ligand exchange complexation reaction mechanism. Here, the phosphate was adsorbed with ZreFPS by coordination to Zr(IV). The weak Lewis base sulfate is coordinated to Zr(IV) just after the regeneration operation with treatment of aqueous sulfuric acid (Helfferich, 1962). Therefore, the weak Lewis base is easily replaced with
4596
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 9 2 e4 6 0 0
metal oxide loaded polymeric ligand exchange adsorbent (Zhao and SenGupta, 1998).
Effluent concentration (mg of P/L)
0.3 pH 2.01 1
0.25
pH 3.00 2
3.3.
Effect of flow rate on phosphate adsorption
pH 4.02 3
0.2
pH 5.12 4 pH 7.01 5
0.15
0.1
0.05
0 0
200
400
600
800
1000
Feed volume (BV, mL/mL-fiber) Fig. 1 e Effect of pH of feed solutions on breakthrough profiles of phosphate. Figure after pH values in inset corresponds to entry nos. 1e5 in Table 1. For detailed conditions and numerical results refer to entry nos. 1e5 in Table 1.
phosphate species such as monovalent (H2PO4) and divalent (HPO42) according to the following reaction (4): 2 F SO þ H2 PO PO3 ZrðIVÞ SO2 3 4 4# 1=2 2 2 F SO3 PO3 ZrðIVÞ H2 PO4 þ ð1=2Þ SO4
(4)
The hydroxide ion (OH) is the hardest Lewis base, the following ligand exchange reaction (5) shifts to right hand side with increasing the pH:
The ZreFPS has high selectivity to phosphate at neutral pH area which is approximate pH of lake and river waters. Then, the neutral pH feed solution was supplied to the column at three different feed flow rates (50e250 h1). These feed flow rates are much higher than the feed flow rates in granular resin packed columns, where the flow rates were usually used from 10 to 40 h1 to take up target anions. Fig. 2 shows the breakthrough profiles of phosphate for the determination of feed flow rate effect. Breakthrough points of phosphate were decreased with an increased in feed flow rate. The detailed experimental conditions and numerical results are summarized in Table 1 (entry nos. 5e7). At flow rate 50 h1, the breakthrough point was 470 BV and it was 340 BV when the flow rate was 250 h1. Namely, an increase in the flow rate by 5 times brings the decrease in the breakthrough point by 38%. Similarly, an increase in the flow rate from 50 to 100 h1 reduced the breakthrough points from 470 to 415 BV. This means that an increase in the flow rate by 2 times decreases the breakthrough point by 13%. These results indicate that the ligand exchange reaction is slower than conventional anion exchange (polyallylamine resin) and hydrogen bonding interaction mechanism (Awual et al., 2008; Awual and Jyo, 2009). In case of weak-base and polyallylamine fiber/resin adsorbents, an increase in flow rate ca. 8e13% decreases the breakthrough point only ca. 22e40%, indicating high kinetic performances (Awual et al., 2011; Awual and Jyo, 2009). It is probably that the anion exchange and hydrogen bonding interactions mechanism are much faster than ligand exchange complexation reaction mechanism.
3.4.
2 F SO PO3 ZrðIVÞ H2 PO 3 4 þ OH # 2 F SO3 PO3 ZrðIVÞ OH þ H2 PO 4
(5)
Several researchers also reported the same trend for metal ions loaded adsorbents to take up target ions (Zhu and Jyo, 2005; Wu et al., 2007). Similarly, high breakthrough points and total adsorption are also observed in neutral pH area by
Effect of competing anions on phosphate adsorption
Many ions present in river water and these are chloride, sulfate, nitrate sodium, potassium, magnesium and calcium. For example, the amount of chloride, nitrate and sulfate (Shirakawa River, Kumamoto, Japan) is 0.44, 0.02 and 0.97 mM, respectively (Zhu and Jyo, 2001). However, the concentrations of these ions are not same in all rivers water. Among all the
Table 1 e Detailed experimental conditions and numerical data for the effects of pH of feed solutions and feed flow rates in column-mode phosphate adsorption by ZreFPS. Entry no.
Feed Solution Flow rate, C0 of phosphate pH h1 in SV (mg of P/L)
1 2 3 4 5 6 7
2.01 3.00 4.02 5.12 7.01 6.98 6.95
50 50 50 50 50 100 250
0.411 0.397 0.385 0.396 0.405 0.418 0.411
Volume of wet fiber bed: 2.12 mL. Feed volume: 943 BV.
Breakthrough Breakthrough Total uptake Eluted amount Recovery point (BV) capacity (mmol/g) (mmol/g) (mmol/g) (%)
625 590 560 515 470 415 340
0.0390 0.0356 0.0327 0.0309 0.0289 0.0264 0.0212
0.0559 0.0532 0.0501 0.0479 0.0468 0.0464 0.0406
0.0571 0.0536 0.0499 0.0468 0.0483 0.0465 0.0422
102 101 99.6 97.7 103 100 104
4597
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 9 2 e4 6 0 0
0.3
0.35 50 h
-1
Effluent concentration (mg og P/L)
Effluent concentration (mg of P/L)
0.4
5
0.3 -1
6
-1
7
100 h
0.25
150 h
0.2 0.15 0.1
0.25
Control
5
-
Cl (10 mg/L)
8
2-
SO 4 (10 mg/L)
0.2
9
Cl-(10 mg/L) & 2-
SO 4 (10 mg/L) 10
0.15
0.1
0.05
0.05 0 0
0
200
400
600
800
0
1000
200
400
600
800
1000
Feed volume (BV, mL/mL-fiber)
Feed volume (BV, mL/mL-fiber) Fig. 2 e Breakthrough profiles of phosphate adsorption by ZreFPS under different feed flow rates. Figure after flow rate in inset corresponds to entry nos. 5e7 in Table 1. For detailed conditions and numerical results refer to entry nos. 5e7 in Table 1.
Fig. 3 e Effect of competing anions on phosphate adsorption by ZreFPS. Figure for each symbol in inset corresponds to entry nos. 5 and 8e10 in Table 2. For detailed conditions and numerical results refer to entry nos. 5 and 8e10 in Table 2.
commonly present competing anions, sulfate is divalent and has greater competition through enhanced electrostatic interaction (Blaney et al., 2007). Then competing anions’ effect was investigated using phosphate solutions containing high concentration of chloride and sulfate. Fig. 3 shows the breakthrough profiles of phosphate in evaluation of the effect of competing anions. Here, the flow rate of the feed was 50 h1 and phosphate feeds concentration was from 0.405 to 0.419 mg of P/L. The pH of feed solutions was also in neutral area. The detailed conditions and numerical results are summarized in Table 2 (entry nos. 5 and 8e10). In the presence of 10 mg/L chloride in phosphate feed, phosphate adsorption and its breakthrough point were not adversely affected but slightly enhanced; the breakthrough point and total adsorption increased from 470 to 490 BV and from 0.0468 to 0.0497 mmol/g, respectively. Similarly, in the presence of 10 mg/L sulfate in phosphate feed, breakthrough point and total adsorption also increased from 470 to 525 BV and from 0.468 to 0.0529 mmol/g, respectively (entry nos. 5, 8 and 9 in Table 2). In the presence of both chloride and sulfate at 10 mg/L in each, total phosphate adsorption (0.0515 mmol/g)
is nearly equal to that (0.0529 mmol/g) in the presence of 10 mg/L sulfate only. This enhancement comes from the phosphate adsorption by the co-ion (sodium ion) effect because Zr(IV) very strongly binds phosphate that increasing of sodium ion enters the exchanger phase through ion pair formation with phosphate and maintains electroneutrality through the Donnan invasion mechanism (Helfferich, 1962; Donnan, 1995). Therefore, an increase in the sodium ion concentration in the aqueous phase shifts the equilibrium toward the right hand side as shown in Scheme 3 (Donnan, 1995). Several researchers are also reported that phosphate adsorption by metal ion or metal oxide loaded ligand exchange chelating resins is not adversely affected with common anions because of ion pair or outer sphere complex formation (Zhao and SenGupta, 1998; Zhu and Jyo, 2005).
3.5.
Elution of adsorbed phosphate and reuses
It is reported that hydroxide is the hardest Lewis base among common inorganic anions. Therefore aqueous NaOH behaves as an effective eluting reagent (Zhu and Jyo, 2005). Fig. 4(a)
Table 2 e Phosphate adsorption by ZreFPS under different concentrations of competing anions and high feed flow rate. Entry no.
5 8 9 10 11
Feed C0 of P (mg/L)
HCO3
0.405 0.412 0.419 0.415 0.216
e e e e 15.0
(mg/L)
Cl (mg/L) e 10.0 10.0 10.0
SO42 (mg/L)
Flow rate, h1 in SV
e e 10.0 10.0 20.0
50 50 50 50 450
Breakthrough point (BV)
Breakthrough capacity (mmol/g)
Total uptake (mmol/g)
Eluted amount (mmol/g)
Recovery (%)
470 490 525 535 640
0.0289 0.0307 0.0334 0.0337 0.0210
0.0468 0.0497 0.0529 0.0515 0.0268
0.0483 0.0507 0.0545 0.0525 0.0271
103 102 103 102 101
Wet fibers bed: 2.12 mL. pH of feed solutions: 6.96e7.02. Volume of feed: 943 BV.
4598
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 9 2 e4 6 0 0
2-
SO3 CH2 O O P O
+
2Na+(aq) + HPO4
OH2
CH2 O O P O
Zr4+ X
OH2
+
SO3
(aq)
X
OH2
Na+(aq)
+
X (aq)
Zr4+ X 2HPO4
OH2
Na+
Scheme 3 e Possible mechanism of phosphate adsorption by ZreFPS.
phosphate, the shorter the time required for purification of contaminated water. However, phosphate species are coexisted with some anions in wastewater. Therefore, the feed solution was prepared with three competing anions (Cl, HCO3 and SO42) and supplied to the column at feed flow rate of 450 h1. Fig. 4(b) shows the breakthrough profile of phosphate and detailed conditions and numerical results are summarized in Table 2 (entry no. 11). The breakthrough point of phosphate was 640 BV, which corresponds to the breakthrough capacity of 0.0210 mmol/g. This means that 1.4 L of water containing phosphorus at 0.216 mg/L can be purified to less than 10 mg/L by ZreFPS (0.45 g FPS), which is lower than the permissible limit recommended by USEPA and Florida Everglades Forever Act (USEPA, 1986; Florida Everglades Forever Act, 1994). Here, the bifunctional fiber was prepared by radiation induced graft polymerization technique which is useful technique due to applicability into various shapes and sizes of existing polymers and allows a high degree of penetration of graft chains to the polymer matrix and ability to produce uniform active radical sites on the polymer during electron irradiation (Lee et al., 1996). Although the feed concentration of phosphate was as low as 0.216 mg of P/L and the flow rate was as high as 450 h1, the breakthrough capacity was not decreased markedly compared with the cases of feed flow rate at 250 h1. The flow rate of 450 h1 is much higher than the metal loaded granular resin packed column (Zhao and SenGupta, 1998; Seko et al., 2004; Zhu and Jyo, 2005). Our findings revealed that Zr(IV) loaded
shows an elution profile of phosphate. Phosphate adsorbed on ZreFPS was quantitatively eluted with 0.1 M aqueous NaOH. Almost 95% adsorbed phosphate was eluted within 15 BV of the eluent. In complete elution, ca. 35 BV of eluent was needed as shown in Fig. 4(a). In addition, sodium hydroxide at 0.1 M gave sharp peak in phosphate elution. The sharp long tail peak means the phosphate species strongly binds with ZreFPS adsorbent. During elution operations, no measurable Zr(IV) was found in the column effluents. After the elution operation, the column was regenerated with 0.5 M H2SO4 solution at a flow rate of 4 h1. The regenerated column was conditioned for the next adsorption operation and then adsorptioneelutioneregeneration operations were repeated. The Zr(IV) loaded fibrous adsorbent is able to retaining functionality and adsorption capability after multiple regenerations cycles without any lost of original efficiency. However, the preparation of FPS required sophisticated instrument which is quite expensive. Due to its high selectivity to phosphate in the presence of large amount competing anions and reversibility/ reusability over many cycles of operations without noticeable deterioration, relatively high cost of FPS will not a big barrier in practical application in the chemical industry.
3.6.
Rapid removal of trace phosphate
To remove huge volume of water containing phosphate at very low concentration by column-mode, kinetically excellent adsorbent is preferred. The higher the adsorption rate of
0.16
a
Effluent concentration (mg of P/L)
Concentration of phosphorus (mg/L)
60
50
40
30
20
10
0 0
5
10
15
20
25
Eluent volume (BV, mL/mL-fiber)
30
35
b
0.14 Run 1
0.12
Run 2
0.1 0.08 0.06 0.04 0.02 0 0
200
400
600
800
1000
Feed volume (BV, mL/mL-fiber)
Fig. 4 e (a) Elution of adsorbed phosphate from ZreFPS column with 0.1 M NaOH at 4 hL1. For detailed conditions and numerical data of the adsorption operation refer to entry no. 5 in Table 1. (b) Breakthrough profiles of phosphate adsorption from feed solution containing phosphate (0.216 mg of P/L), ClL (10 mg/L), HCO3L (15 mg/L) and SO42L (20 mg/L) at pH 7.02 under the feed flow rate at 450 hL1. For detailed conditions and numerical results refer to entry no. 11 in Table 2.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 9 2 e4 6 0 0
bifunctional fibrous adsorbent able to purify low concentration level of phosphate even in the presence of high concentration of competing anions at high feed flow rate. On the other hand, fibrous adsorbents based on preparation of radiation induced graft polymerization techniques are shown excellent kinetic performances to take up toxic anions even flow rate at 1000e2500 h1 (Awual et al., 2011, 2008). All experimental results in the column study were obtained using the same column for 5 months without renewing of ZreFPS column or reloading Zr(IV) onto FPS. No measurable Zr(IV) was found in the column effluents during many adsorptioneelutioneregeneration cycles. Each entry number of adsorptioneelutioneregeneration operation was repeated at least two times to check the reproducibility. Average data of listed runs are indicated in Tables 1 and 2.
4.
Conclusions
Phosphate adsorbent was prepared by loading Zr(IV) onto a bifunctional cation exchange fiber FPS containing phosphonate and sulfonate. In this study, an extensive laboratory work was demonstrated by ligand exchange fibrous adsorbent for the removal of trace phosphate based on selectivity and kinetics performance. Phosphate adsorption was tested under various conditions by using a column packed with Zr(IV) loaded onto FPS. The maximum phosphate adsorption was observed in acidic media. Phosphate adsorption decreased with an increase in pH of the feeds. However, at neutral pH, the ZreFPS worked well in taking up phosphate. Therefore, the main benefit of the system is that it can achieve desired levels of phosphate removal regardless of site conditions such as pH and temperature. Phosphate adsorption by ZreFPS was not adversely affected in presence of Cl and SO42 but slightly increased, indicating no interference of these competing anions despite enhancement of total adsorption. However, phosphate adsorption was affected by the feed flow rate from 50 to 250 h1; breakthrough capacity and total adsorption were gradually decreased with an increase in feed flow rate. ZreFPS was able to take up phosphate from a dilute solution as low as 0.216 mg of P/L even at high feed flow rate of 450 h1. Adsorbed phosphate on ZreFPS column was quantitatively eluted with 0.1 M NaOH solution, and then the column was regenerated by 0.5 M H2SO4 solution for next adsorption operation. No measurable Zr(IV) was detected in column effluents during repeated cycles of adsorption and elution operations. Therefore, Zr(IV) loaded ligand exchange adsorbent is to be an effective means to remove trace phosphate from water for long time.
Acknowledgment The authors wish to thank the anonymous reviewers and editor for their helpful suggestions and enlightening comments to improve the paper.
4599
references
Awual, M.R., Urata, S., Jyo, A., Tamada, M., Katakai, A., 2008. Arsenate removal from water by a weak-base anion exchange fibrous adsorbent. Water Res. 42 (3), 689e696. Awual, M.R., Jyo, A., 2009. Rapid column-mode removal of arsenate from water by crosslinked poly(allylamine) resin. Water Res. 43 (5), 1229e1236. Awual, M.R., Jyo, A., El-Safty, S.A., Tamada, M., Seko, N., 2011. A weak-base fibrous anion exchanger effective for rapid phosphate removal from water. J. Hazard. Mater. 188 (1e3), 164e171. Babatunde, A.O., Zhao, Y.Q., Yang, Y., Kearney, P., 2008. Reuse of dewatered aluminum-coagulated water treatment residual to immobilize phosphorus: batch and column trials using a condensed phosphate. Chem. Eng. J. 136 (2e3), 108e115. Balaji, T., Yokoyama, T., Matsunaga, H., 2005. Adsorption and removal of As(V) and As(III) using Zr-loaded lysine diacetic acid chelating resin. Chemosphere 59 (8), 1169e1174. Bektas, N., Akbulut, H., Inan, H., Dimoglo, A., 2004. Removal of phosphate from aqueous solutions by electro-coagulation. J. Hazard. Mater. 106 (2e3), 101e105. Blaney, L.M., Cinar, S., SenGupta, A.K., 2007. Hybrid anion exchanger for trace phosphate removal from water and wastewater. Water Res. 41 (7), 1603e1613. Chimenos, J.M., Fernandez, A.I., Villalba, G., Segarra, M., Urruticoechea, A., Artaza, B., Espiell, F., 2003. Removal of ammonium and phosphates from wastewater resulting from the process of cochineal extraction using MgO-containing byproduct. Water Res. 37 (7), 1601e1607. Chitrakar, R., Tezuka, S., Sonoda, A., Sakane, K., Ooi, K., Hirotsu, T., 2006. Selective adsorption of phosphate from seawater and wastewater by amorphous zirconium hydroxide. J. Colloid Interf. Sci. 297 (2), 426e433. Conley, D.J., Paerl, H.W., Howarth, R.W., Boesch, D.F., Seitzinger, S.P., Havens, K.E., Lancelot, C., Likens, G.E., 2009. Controlling eutrophication: nitrogen and phosphorus. Science 323 (5917), 1014e1015. de-Bashan, L.E., Bashan, Y., 2004. Recent advances in removing phosphorus from wastewater and its future use as fertilizer (1997e2003). Water Res. 38 (19), 4222e4246. Dennison, W.C., Orth, R.J., Moore, K.A., Stevenson, J.C., Carter, V., Kollar, S., Bergstrom, P.W., Batiuk, R.A., 1993. Assessing water quality with submersed aquatic vegetation. Habitat requirements as barometers of Chesapeake Bay health. Bioscience 43 (2), 86e94. Donnan, F.G., 1995. Theory of membrane equilibria and membrane potentials in the presence of non-dialysing electrolytes. A contribution to physicalechemical physiology. J. Membr. Sci. 100 (1), 45e55. Florida Everglades Forever Act, 1994. Florida State Legislature, Tallahassee, FL. Genz, A., Kornmuller, A., Jekel, M., 2004. Advanced phosphorus removal from membrane filtrates by adsorption on activated aluminium oxide and granulated ferric hydroxide. Water Res. 38 (16), 3523e3530. Hamabe, Y., Matsuura, R., Jyo, A., Tamada, M., Katakai, A., 2009. Properties of bifunctional phosphonate fibers derived from chloromethylstyrene grafted polyolefin fibers. React. Funct. Polym. 69 (1), 1e8. Hamoudi, S., Saad, R., Belkacemi, K., 2007. Adsorptive removal of phosphate and nitrate anions from aqueous solutions using ammonium-functionalized mesoporous silica. Ind. Eng. Chem. Res. 46 (25), 8806e8812. Helfferich, F., 1962. Ion Exchange. McGraw-Hill, New York, NY.
4600
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 9 2 e4 6 0 0
Jyo, A., Okada, K., Tamada, M., Kume, T., Sugo, T., Tazaki, M., 2005. Bifunctional cation exchange fibers having phosphonic and sulfonic acid groups. In: Environment Science Research. Chemistry for the Protection of the Environment 4, vol. 59, pp. 49e62. Jyo, A., Awual, M.R., Kobayashi, K., 2008. Behavior of crosslinked polyallylamine resin PAA in uptake of inorganic anions. Soc. Chem. Ind., 487e494. Jyo, A., Hamabe, Y., Matsuura, H., Shibata, Y., Fujii, Y., Tamada, M., Katakai, A., 2010. Preparation of bifunctional chelating fiber containing iminodi(methylphosphonate) and sulfonate and its performances in column-mode uptake of Cu(II) and Zn(II). React. Funct. Polym. 70 (8), 508e515. Kabayama, M., Sakiyama, T., Kawasaki, N., Nakamura, T., Araki, M., Tanada, S., 2003. Characteristics of phosphate ion adsorptionedesorption onto aluminum oxide hydroxide for preventing eutrophication. J. Chem. Eng. Jpn. 36 (4), 499e505. Kuzawa, K., Jung, Y.J., Kiso, Y., Yamada, T., Nagai, M., Lee, T.G., 2006. Phosphate removal and recovery with a synthetic hydrotalcite as an adsorbent. Chemosphere 62 (1), 45e52. Lee, W., Oshikiri, T., Saito, K., Sugita, K., Sugo, T., 1996. Comparison of formation site of graft Chain between nonporous and porous films prepared by RIGP. Chem. Mater. 8 (11), 2618e2621. Lee, S.I., Weon, S.Y., Lee, C.W., Koopman, B., 2003. Removal of nitrogen and phosphate from wastewater by addition of bittern. Chemosphere 51 (4), 265e271. Maeda, H., Egawa, H., 1984. Preparation of macroreticular chelating resins containing mercapto groups from 2, 3epithiopropyl methacrylate/divinylbenzene copolymer beads and their adsorption capacity. Anal. Chim. Acta 162, 339e346. Morse, G.K., Brett, S.W., Guy, J.A., Lester, J.N., 1998. Review: phosphorus removal and recovery technologies. Sci. Total Environ. 212 (1), 69e81. Mullan, A., McGrath, J.W., Adamson, T., Irwin, S., Quinn, J.P., 2006. Pilot-scale evaluation of the application of low pH-inducible polyphosphate accumulation to the biological removal of phosphate from wastewaters. Environ. Sci. Technol. 40 (1), 296e301. Namasivayam, C., Prathap, K., 2005. Recycling Fe(III)/Cr(III) hydroxide, an industrial solid waste for the removal of phosphate from water. J. Hazard. Mater. B123 (1e3), 127e134. Onyango, M.S., Kuchar, D., Kubota, M., Matsuda, H., 2007. Adsorptive removal of phosphate ions from aqueous solution using synthetic zeolite. Ind. Eng. Chem. Res. 46 (3), 894e900. Ringbom, A., 1963. Complexation in Analytical Chemistry. Interscience Publishers, New York/London.
Ryther, J.H., Dustan, W.M., 1971. Nitrogen, phosphorus, and eutrophication in the coastal marine environment. Science 171 (3975), 1008e1013. Seida, Y., Nakano, Y., 2002. Removal of phosphate by layered double hydroxides containing iron. Water Res. 36 (5), 1306e1312. Seko, N., Basuki, F., Tamada, M., Fumio, Y., 2004. Rapid removal of arsenic(V) by zirconium(IV) loaded phosphoric chelate adsorbent synthesized by radiation induced graft polymerization. React. Funct. Polym. 59 (3), 235e241. Stante, L., Cellamare, C.M., Malaspina, F., Bortone, G., Tilche, A., 1997. Biological phosphorus removal by pure culture of Lampropedia spp. Water Res. 31 (6), 1317e1324. Suzuki, T.M., Tanaka, D.A.P., Tanco, M.A.L., Kanesato, M., Yokoyama, T., 2000. Adsorption and removal of oxo-anions of arsenic and selenium on the zirconium(IV) loaded polymer resin functionalized with diethylenetriamine-N, N, N0 , N0 polyacetic acid. J. Environ. Monitor. 2, 550e555. U.S. Environmental Protection Agency (USEPA), 1986. Quality Criteria for Water 1986. Office of Water, Regulation and Standard, Washington, DC 20460. EPA 440/5-86-001. Vasudevan, S., Sozhan, G., Ravichandran, S., Jayaraj, J., Lakshmi, J., Sheela, S.M., 2008. Studies on the removal of phosphate from drinking water by electrocoagulation process. Ind. Eng. Chem. Res. 47 (6), 2018e2023. Wu, R.S.S., Lam, K.H., Lee, J.M.N., Lau, T.C., 2007. Removal of phosphate from water by a highly selective La(III)-chelex resin. Chemosphere 69 (2), 289e294. Yildiz, E., 2004. Phosphate removal from water by fly ash using cross-flow microfiltration. Sep. Purif. Technol. 35 (3), 241e252. Zhang, S., Chen, S., Zhang, Q., Li, P., Yuan, C., 2008. Preparation and characterization of an ion exchanger based on semicarbonized polyacrylonitrile fiber. React. Funct. Polym. 68 (4), 891e898. Zhao, D., SenGupta, A.K., 1998. Ultimate removal of phosphate from wastewater using a new class of polymeric ion exchangers. Water Res. 32 (5), 1613e1625. Zhao, D., SenGupta, A.K., 2000. Ligand separation with a copper(II)-loaded polymeric ligand exchanger. Ind. Eng. Chem. Res. 39 (2), 455e462. Zhu, X., Jyo, A., 2001. Removal of arsenic(V) by zirconium(IV)loaded phosphoric acid chelating resin. Sep. Sci. Technol. 36 (14), 3175e3189. Zhu, X., Jyo, A., 2005. Column-mode phosphate removal by a novel highly selective adsorbent. Water Res. 39 (11), 2301e2308.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 6 0 1 e4 6 1 4
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Ionic liquids for extraction of metals and metal containing compounds from communal and industrial waste water Lisa Fischer a, Thomas Falta a, Gunda Koellensperger a, Anja Stojanovic b, Daniel Kogelnig b, Markus Galanski b, Regina Krachler b, Bernhard K. Keppler b, Stephan Hann a,* a
Department of Chemistry, Division of Analytical Chemistry, University of Natural Resources and Life Sciences, Muthgasse 18, A-1190 Vienna, Austria b University of Vienna, Department of Inorganic Chemistry, Waehringer Strasse 42, A-1090 Vienna, Austria
article info
abstract
Article history:
In a fundamental study the potential of ionic liquids based on quaternary ammonium- and
Received 17 January 2011
phosphonium cations and thiol-, thioether-, hydroxyl-, carboxylate- and thiocyanate-
Received in revised form
functionalized anions has been assessed for future application in advanced sewage
9 June 2011
treatment. The elimination of the metal(oid)s Ag, As, Cd, Cr, Cu, Hg, Ni, Pb, Pt, Sn, Zn and
Accepted 13 June 2011
the cancerostatic platinum compounds cisplatin and carboplatin was screened using
Available online 21 June 2011
a liquid phase micro-extraction set-up. The analytical tool-set consisted of ICP-SFMS and LC-ICP-MS for quantification of metal(oid)s and cancerostatic platinum compounds,
Keywords:
respectively. The purity of the ILs was assessed for the investigated metal(oid)s on the base
Water treatment
of present EU environmental quality standards and was found to be sufficient for the
Ionic liquids
intended use. In model solutions at environmental relevant concentrations extraction
Liquid phase micro-extraction
efficiencies 95% could be obtained for Ag, Cu, Hg and Pt with both phosphonium- and
Water framework directive
ammonium-based ILs bearing sulphur functionality in the form of thiosalicylate and 2-(methylthiobenzoate) anions, as well as with tricaprylmethylammonium thiocyanate within an extraction time of 120 min. All other metals were extracted to a lower extent (7e79%). In the case of cancerostatic platinum compounds a phosphoniumbased IL bearing thiosalicylate functionality showed high extraction efficiency for monoaquacisplatin. For the first time, liquid phase micro extraction with ionic liquids was applied to industrial and communal waste water samples. The concentration of all investigated metal(oid)s could be significantly reduced. The degree of elimination varied with the initial concentration of metals, pH and the amount of suspended particulate matter. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
The recent European Union Water Framework Directive (WFD) (2000/60/EC) sets high Environmental Quality Standards (EQS) for priority substances in surface water regulating the annual averages and maximum allowable concentrations in surface
water (Directive 2000/60/EC). The list of priority substances includes 33 organic and inorganic compounds, which have become a serious problem in the aquatic environment due to their toxicity, bioaccumulation and persistence. The metals Ni, Cd, Hg, und Pb and their compounds belong to this list of priority substances (Directive 2000/60/EC, Annex I). Their
* Corresponding author. Tel.: þ43 1 47654 6086; fax: þ43 1 47654 6059. E-mail address: [email protected] (S. Hann). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.06.011
4602
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 6 0 1 e4 6 1 4
emission causes pollution of the water cycle leading to poor water quality, insufficient supply of drinking water and complicate drinking water pre-treatment (Deutsches Umweltbundesamt, 2003). Additional negative effects are the generation of a high toxicological risk for the aquatic organisms and the bioaccumulation of the toxic compounds in the food chain. Several other metal compounds have been recently identified as hazardous contaminants of the aquatic environment. Among them, the cancerostatic platinum compounds cisplatin (cis-diamminedichloroplatinum(II), carboplatin (cisdiammine(1-1-cyclobutanedicarboxylato)platinum(II) and oxaliplatin ([(1R,2R)-cyclohexanediamine-N,N0 ]oxalate(2-)-O,O0 platinum)), which are successfully and widely used in chemotherapy (Hann et al., 2005; Kelland, 2007) pose an environmental problem. After administration, considerable amounts of the drugs are excreted via the patients’ urine, thus reaching the waste water system (Ku¨mmerer and Helmers, 1997; Lenz et al., 2005). Since cisplatin is classified as carcinogenic to animals and all other planar platinum complexes are supposed to be carcinogenic as well (IARC, 1987; Hann et al., 2005), efforts have been made to eliminate CPC from hospital waste water. However, it has been recognized that a significant fraction of these compounds is not removed via conventional sewage treatment (Lenz et al., 2007). As a matter of fact, technologies for elimination of heavy metals from waste water and prevention of their emission into surface water is of particular importance. The employment of ionic liquids (ILs) as extracting agents for elimination of those compounds via waste water treatment could be an important alternative to advanced sewage procedures based on adsorption and irradiation. ILs are generally defined as salts that are liquid below 100 C and consist entirely of ions. They show extraordinary properties such as an extremely low vapour pressure, high thermal stability and their physico-chemical properties can be tuned by modifying their chemical structure (Marsh et al., 2004; Zhao et al., 2005; Pandey, 2006; Han and Armstrong, 2007). Several studies have investigated the use of room temperature ILs based on imidazolium-, pyridinium-, pyrrolidinium- or phosphonium cations for the extraction and separation of organic as well as inorganic substances from aqueous media (Wei et al., 2003; Liu et al., 2003; Papaiconomou et al., 2008; Regel-Rosocka, 2009; Lertlapwasin et al., 2010; Rios et al., 2010). Further, efficient extraction procedures for metals bound to complexing ligands (e.g. crown-ethers, calixarenes, dithizone) into imidazoliumbased ILs were depicted (Dietz and Dzielawa, 2001; Shimojo and Goto, 2004; Luo et al., 2004; Domanska and Rekawek, 2009). Modifying their ionic composition by appending metal-ion ligating functional groups, selective extraction of solute metals by ILs can be adjusted. Rogers et al. (Visser et al., 2001, 2002) first investigated the potential of “task specific” ionic liquids as extractants for Hg and Cd from water using imidazolium cations with thioether-, urea-, thioureaderivatized side chains (TSILs). However, the main drawback of TSILs is that their hydrophobicity is achieved by incorporation the harmful fluorine containing anion hexafluorophosphate. Therefore, they pose a severe environmental risk due to hydrolysis and formation of hydrofluoric acid in the presence of water or air moisture (Swatloski et al., 2003). On
the other hand, some research groups achieved the taskspecificity of ILs via anchoring a functional group onto the anion (Kalb et al., 2006; Egorov et al., 2010; Stojanovic et al., 2010), gaining TSILs suitable for the extraction of different metals from aqueous solutions. For example, an ionic liquid based on the trioctylmethylammonium cation with thiosalicylate as anion, prepared via a halide free synthesis route, is commercially available and has been evaluated as extracting agent for heavy metals (Kalb, 2005; Kalb et al., 2006). We have recently shown that low cost ILs based on the quaternary phosphonium- and ammonium ionic liquids CyphosIL101 (Cytec) and Aliquat336 (Henkel) are suitable to eliminate platinum from aqueous solution by simple replacement of the chloride anion with functionalized aromatic anions bearing thiol- and thioether groups (Stojanovic et al., 2010). Egorov et al. (2010) successfully applied Aliquat-based TSIL with salicylate anion as extracting agent for iron and copper from a model matrix, suggesting the formation of salicylate complexes with metal species. On the other hand, tricaprylammonium thiocyanate, [A336][SCN], is a well known extracting agent for actinides from acid solutions via anion exchange (Moore, 1964). We have recently evaluated this ILs as potential extracting agent for pre-concentration of uranium from real water matrix (Srncik et al., 2009). In this study we have evaluated the potential of anion functionalized ILs as extracting agents for the priority metals Cd, Ni, Hg, Pb as well as for As, Cr, Cu, Pt, Sn, Zn. Considering typical contact time and volume ratios during waste water treatment a liquid phase micro-extraction set-up was implemented. The analytical methods allowed rapid and accurate high-throughput screening of the extraction efficiencies. Our experiments included model solutions and, for the first time, industrial waste waters from different sources considering variable chemical and physical properties. As a further novelty, cancerostatic platinum compounds were included in the context of waste water treatment by ILs.
2.
Experimental section
2.1.
Synthesis of ionic liquids
All evaluated ILs were synthesized according to protocols published in the literature. Ammonium and phosphoniumbased ILs were prepared via a metathesis reaction using tricaprylmethylammonium chloride (Aliquat336) and trihexyl(tetradecyl)phosphonium chloride (Cyphos IL 101) and corresponding Brønsted acids or sodium salts as precursors (Visser et al., 2002; Kogelnig et al., 2008). One- and two dimensional NMR experiments, FTIR, elemental analysis and electrospray ionization mass spectrometry (ESI-MS) analysis confirmed the composition and purity of the prepared ILs. In Table 1 the structure of ILs and their characteristic physicochemical parameters are displayed.
2.2. IL-extraction experiments of metals and metal containing compounds from aqueous solutions Batch experiments were performed by liquid phase microextraction (LPME) based on a set-up of Liu et al. (2004, 2005).
4603
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 6 0 1 e4 6 1 4
Table 1 e Physico-chemical properties of studied ionic liquids. Ionic liquid
Structure
Water c Density Viscosity Decomposition Chloride Water h(cP) temperature TD content content saturation wt (Cl) g/cm3 (25 C) (25 C) [ C] wt% wt% % [ppm]a MiliQ- 10 mM H2O CaCl2
Tricaprylmethylammonium thiosalicylate, [A336][TS]
0.95b
3220b
270b
0.11b
1.54b
4.1b
4.0b
Tricaprylmethylammonium 2-(methylthio) benzoate, [A336] [MTBA]
0.94b
5242b
260b
0.39b
0.23b
15.1b
14.8b
Tricaprylmethylammonium benzoate, [A336] [BA]
0.94b
3860b
240b
0.21b
0.21b
17
18.1
456
Tricaprylmethylammonium hexanoate, [A336] [Hex]
0.88c
e
148c
0.59c
0.1
24.5
23.8
523
Tricaprylmethylammonium thiocyanate, [A336] [SCN]
e
1017d
e
0.08
2.5
4.7
182
Trihexyl(tetradecyl) phosphonium thiosalicylate, [PR4] [TS]
0.93b
3875b
390b
0.56b
0.56b
11.2b
10.8b
Trihexyl(tetradecyl) phosphonium 2(methylthio) benzoate, [PR4] [MTBA]
0.94b
875b
350b
0.26b
0.26b
10.6b
10.5b
Trihexyl(tetradecyl) phosphonium salicylate, [PR4][Sal]
0.92b
567b
340b
0.66b
0.80b
6.5
6.2
a b c d
<0.03
50b
55.7b
13.1b
1
231
Decrease of chloride concentration in aqueous phase (10 mM CaCl2) after extraction with ILs. Stojanovic et al., 2010. Kogelnig et al., 2008. Kulkarni et al., 2007.
A schematic diagram of the modified set-up is shown in Fig. 1. A defined volume of model solution containing mg L1 concentrations of various metals, cisplatin or carboplatin, respectively, was transferred into a 50 mL polyethylene- (PE-) flask. All solutions contained 10 mmol L1 CaCl2, which had
been adjusted to pH 7.5 using suprapure o-phosphoric acid and sodium hydroxide. For LPME, approximately 25 mL of the investigated IL was drawn into a piece of PEEK-capillary (0.02” ID, 1/16” OD PEEK, 5 cm length, BESTA-Technik GmbH, Wilhelmsfeld, Germany) utilizing a 1 mL polypropylene- (PP-)
4604
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 6 0 1 e4 6 1 4
Table 2 e Chemical/physical properties of waste water samples used for extraction studies (Total metal concentration in mg LL1).
PP- syringe + connector PEEK-capillary (0.02” ID, 1/16” OD, 5 cm) IL drop Sample solution Stir bar Fig. 1 e Schematic diagram of the IL-based liquid phase micro-extraction procedure.
syringe (Terumo Europe, Leuven, Belgium). The plunger was depressed to expose a 10 mL IL drop. Extraction started by immersion of the IL drop into the stirred sample solution. To study the effect of contact time on metal uptake by ILs, 1 mL sample aliquots were taken after 30, 60, 90 and 120 min. Three repetitive extraction experiments were performed for each IL. Extraction experiments of metals from model solutions were performed by LPME of a 20 mL volume of model solution containing a metal concentration of 7.5 mg L1 Ag, As, Cd, Hg, Pb, Pt, Sn and 75.0 mg L1 Cr, Cu, Ni and Zn (approx.). To investigate the leaching of metals from the different ILs into the water phase LPME was applied employing blank model solutions with a contact time of 120 min. After extraction, the collected aliquots were immediately transferred into acid-cleaned 10 mL PP-vials, acidified with 1% concentrated nitric acid for stabilization (final pH < 1.5), capped and stored at 4 C until measurement. Quantification of the diluted samples (factor 10) was carried out by ICP-SFMS via external calibration and internal standardization using indium (0.5 mg L1 in the final solution). Several waste water samples were collected for studying metal extraction from real matrixes. Total metal concentrations, pH, electrical conductivity (EC) and suspended solid content (xss) of the selected samples are summarized in Table 2. Gravimetric determination of the xss in real waste water samples was performed according to DIN 38409-2 (1987).The extraction experiments of unfiltered waste waters were performed with the LPME set-up as described above. Dilution of waste water samples prior to measurement was adjusted to match the working range of the ICP-SFMS method. The certified reference material TM 28.3 (Low level fortified standard prepared from Lake Ontario water, National Water Research Institute) was used for controlling the trueness of results. All measured values agreed with the certified values within the 2s confidence limits. To investigate the extraction efficiency of ILs for intact cisplatin and carboplatin, model solutions were freshly prepared by spiking cisplatin stock solutions (10 mg L1) into the 10 mmol L1 CaCl2 solution (pH 7.5). 30 mL of model solution containing 50 mg L1 cisplatin or carboplatin were used for the extraction experiments, which were performed according to the method described above. Extraction studies concerning the cisplatin degradation products monoaqua- and diaquacisplatin/ monohydroxo- and dihydroxocisplatin were performed by
Waste water
Ag As Cd Cr Cu Hg Ni Pb Pt S Sn Zn pH ECa xssb
1
2
3
4
5
6
24.3 2.05
36.2 2.68 0.04 187 181 0.18 106 0.74
0.23 0.18
using aged model solutions (incubation time of 48 h at 20 C). During all extraction experiments aliquots of the model solutions were sampled after 30, 60, 90 and 120 min into HPLC-vials, placed in a cooled metal free autosampler (5 C) and analyzed by HPLC-ICP-MS as described below. Quantification of CPC was performed by external calibration in a working range of 0.01e50 mg L1. Extraction efficiencies (E ) were calculated by: Eð%Þ ¼ c0aq c1aq
c0aq 100
where c0aq and c1aq are the total metal concentrations in the aqueous phase before and after the respective extraction time.
2.3.
Chemicals and standards
All reagents used throughout the study were of ultra-pure grade. Ultrapure HNO3 was prepared by double sub-boiling distillation of 65% nitric acid of p.a. grade (Merck) using a duoPUR quartz sub-boiling unit (MLS Lab Systems GmbH, Leutkirch, Germany). Ultra-pure water was used for the preparation of standards and model solutions by sub-boiling distillation of purified water (18.2 MU cm) obtained from an ultra clear system (SG water GmbH, Barsbu¨ttel, Germany). Quantitative element standards (Ag, As, Cd, Cr, Cu, Hg, In, Ni, Pb, Pt, S, Sn and Zn) were certified single element ICP standards for trace analysis, purchased from Merck KGaA (Darmstadt, Germany). For quantification of the metals in the samples obtained from LPME an aqueous multi-element stock solution of 10.0 mg L1 (Cr, Cu, Ni, S, Zn) and 1.0 mg L1 (Ag, As, Hg, Pb, Pt, Sn) was prepared by diluting the 1000 mg L1 single element ICP-MS-standard solutions in 5% nitric acid solution and stored in PFA-bottles at 4 C. Working solutions were prepared immediately before use in 1% HNO3.
4605
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 6 0 1 e4 6 1 4
For HPLC formic acid (Suprapur quality) and Methanol (HPLC gradient grade) were obtained from Merck (Darmstadt, Germany) and sub-boiled ammonium hydroxide was obtained from Aldrich (Miliwaukee, WI, USA). Solid cisplatin (purity 100%) and carboplatin were purchased from Sigma. Stock solutions (10 mg L1) for the time-dependent elimination studies were prepared in case of cisplatin in 150 mmol L1 NaCl, while for carboplatin 5% glucose-solution was used as solvent. Polypropylene (PP) and Polyethylen (PE) materials (bottles, vials, pipette tips) used for sample preparation were acidleached according to a routine cleaning protocol involving incubation in nitric acid baths (10 and 1% HNO3) and rinsing with ultra-pure water. Bottles for preparation and storage of elemental standards were made of perfluorated polymer (PFA) and pre-cleaned with an acid steam system (MLS Lab Systems). To avoid contamination, all sample preparation steps and measurements were carried out under clean room conditions (class 100000 and class 10000 with clean benches class 100, respectively) with temperature control (20 C) and overpressure (þ5 Pa).
2.4. Determination of total metal(oid) concentrations by inductively coupled plasma sector field mass spectrometry (ICP-SFMS) The concentrations of the investigated metal(oid)s were measured using an Element 2 HR-ICP-SFMS (Thermo Fisher, Bremen, Germany). The instrument is specified for three fixed resolution settings (R ¼ m/Dm at 10% peak valley): low resolution (LR, R ¼ 300), medium resolution (MR, R ¼ 4500) and high resolution (HR, R ¼ 10000). For the introduction of particle free samples from model experiments a self-aspirating set-up consisting of a PFA-ST micro-flow nebulizer (Elemental Scientific Inc., Cuming, Omaha, USA) with a sample uptake of 100 mL min1 was used. The nebulizer was combined with a PC3 cyclonic quartz chamber (ESI) operated at 4 C, a quartz injector pipe and torch as well as aluminium sampler and skimmer cones (Thermo Fisher). A slurry-type nebulization set-up employing a V-groove nebulizer (Glas Expansion, Melbourne, Australia) with an i. d. of 145 mm was used for introduction of unfiltered waste water samples. Sample transport to the nebulizer was enabled by a peristaltic pump with a sample uptake rate of 1.0 mL min1. All instrumental operating conditions and the selected isotopes for interferencefree ICP-SFMS measurements are listed in Table 3.
2.5. Speciation of cancerostatic platinum compounds and degradation products by liquid chromatography (LC) ICP-MS Chromatographic separation of carboplatin, cisplatin and the major degradation compounds monoaqua- and diaquacisplatin was performed on a Discovery HS F5 column (3 mm particle diameter, 150 2.1 mm, Supelco, Bellefonte, PA, USA), applying the conditions described in detail elsewhere (Hann et al., 2005). ICP-MS instrument settings and measurement parameters are listed in Table 3. RF-power and gas flows were daily optimized by a tuning procedure.
Table 3 e ICP-MS operating parameters used for multielement analysis and LC-ICP-MS. Tuning parameters Plasma power Sample gas flow Auxiliary gas flow Plasma gas flow Sample uptake rate PFA microflow nebulizer V-groove nebulizer Isotopes measured R ¼ 300 R ¼ 4500
R ¼ 10000
2.6.
ICP-SFMS
LC-ICP-MS
1350e1400 W 16 L min1
1300e1400 W 16 L min1
0.9e1.1 L min1
0.9e1.1 L min 1
0.9e1.3 L min1
0.9e1.2 L min 1
0.35 ml min1
1.0 ml min1
195 107
111
52
65
202
Pt
60
Ag, Cd, Cr, Cu, Hg, Ni, 208 Pb, 195Pt, 32S, 118Sn, 66Zn, 115In (internal standard) 75 As, 115In (internal standard)
Data evaluation
Generation and export of transient signals (chromatograms from LC-ICP-MS) was performed using Chromlink (Version 2.1, Perkin Elmer SCIEX) in combination with Totalchrom (Version 6.2, Perkin Elmer SCIEX). Chromeleon software (Version 6.70, Dionex, Sunnyvale, CA, USA) was used for integration and evaluation of all chromatographic data.
3.
Results and discussion
In this work two classes of ILs based on quaternary ammonium- or phosphonium cations with functionalized anions were comprehensively studied with regard to their potential as novel extractants for waste water treatment. In fundamental experiments, the extraction efficiency for Ag, As, Cu, Cr, Hg, Ni, Pb, Pt, Sn, Zn and cancerostatic platinum compounds (CPC) from model solutions was assessed at environmentally relevant concentration levels. The chemical form of chromium, platinum and arsenic was Cr(III), [Pt(IV) Cl6]2- and As(III), respectively. For the first time, elimination of metals and metal compounds via ILs from real waste water matrices was addressed. Table 1 shows the structures and properties of the ILs, which have been included in this study. As a pre-requisite, all prepared ILs exhibited sufficient purity with residual chloride content of max. 0.7% wt. (a fact proving quantitative anion exchange during preparation). Furthermore several physico-chemical properties are summarized. As can be readily observed all ILs of the selected panel exhibited comparable density values at room temperature (0.88e0.95 g cm3). However, the enhanced thermal stability of phosphonium-based ILs compared to their ammonium
4606
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 6 0 1 e4 6 1 4
analoga enabled their application in a broader temperature range. The experimental set-up for testing the extraction capacity was based on liquid phase micro-extraction (LPME) shown in Fig. 1. This set-up has been introduced by Liu et al. (2005), studying the elimination of organometallic compounds and other environmental pollutants from aqueous samples at high concentrations (0.01e1 mg L1). The method, which used an IL/water volume ratio of 1:2000 (5 mL þ 10 mL sample) was adopted for this work for studying time depended elimination of metal(oid)s and metal compounds from model solutions and waste water.
ILs showed significant leaching of the metals Cd, Hg, Ni and Pb, belonging to the priority substances listed in the Water Framework Directive (2000/60/EC) except [A336][BA] and [A336][Hex]. Cd concentrations determined in the water phase were 0.11 and 0.23 mg L1, respectively and exceeded the maximum allowable concentration, which is 0.08e1.15 mg L1, depending on the water hardness class. Significant leaching of Hg could be excluded for all of the tested ILs. Maximum concentrations of Ni and Pb found in the aqueous phase were 0.78 and 0.12 mg L1 respectively, being significantly lower as the annual average EU-quality standards of surface water. (20 mg L1 for Ni and 7.2 mg L1 for Pb). As there are no uniform environmental quality standards for other pollutants (As, Cr, Cu, Ag, Zn), the ILs were evaluated regarding the maximum permissible addition (MPA) published by the Commissie Integraal Waterbeheer (2000). In this context only [PR4][Sal] showed a significant release of Zn (39.3 5.9 mg L1) exceeding the MPA of 7.8e52.0 mg L1. These results indicate that [A336] [TS], [A336][MTBA], [A336][SCN], [PR4][TS] and [PR4][MTBA] are of sufficient purity for the intended use. However, scaling up of IL synthesis for technical use should be carefully evaluated concerning possible sources of contamination i.e. the precursors of IL synthesis as well as technical equipment and materials. Additionally, sulphur leaching from ILs bearing S-containing anions was determined. The obtained sulphur
3.1. Extraction efficiencies of ILs for metal(oid) extraction 3.1.1. phase
Leaching of metal impurities from ILs into the water
In a first step the ILs were studied regarding the leaching of metal impurities employing LPME of blank model solutions. Table 4 lists the concentration of metals present in the water phase after a contact time of 120 min. The given uncertainty represents the standard deviation (SD) of the results of three independently prepared samples. The limits of detection (LOD) are expressed as the threefold SD of the noise determined in the blank model solution (n ¼ 6). None of the studied
Table 4 e Purity of ionic liquids. [A336][TS]
Conc. Ag As Cd Cr Cu Hg Ni Pb Pt Sn Zn
[A336][BA]
0.25 0.08 0.01 0.003 0.23
[PR4][TS] Conc. Ag As Cd Cr Cu Hg Ni Pb Pt Sn Zn
Conc.
0.01
[A336][MTBA]
Conc. 0.06 0.56
0.002 0.01 0.003
0.045 0.005 0.0002 0.04
[PR4][Sal]
0.01
0.04
0.001
0.03
5.90 1
Conc.
0.0006 0.01
0.004 0.0008
0.01
[PR4][MTBA]
Conc.
[A336][Hex]
Conc.
0.10 0.002 0.003
0.01 0.004 0.0009 0.07
LOD
[A336][SCN] Conc.
0.03
0.01 0.01 0.004
0.008
0.24
MAC-EQS/AA-EQSa
0.01 0.02 0.06
0.34
0.005 0.005 0.0006 0.002 0.017 0.015 0.022 0.022 0.0004 0.045 0.027
0.08e1.5*/0.08e0.25*
0.07/0.05 n. a./20 n. a./7.2
Concentrations and standard deviations (2s) are in mg L ; Instrumental limits of detection (LOD) [mg L1] expressed as three times the standard deviation of the noise determined in 1% HNO3. a Environmental Quality Standards (EQS) for priority substances in surface water (Directive 2006/60/EC - COM 2006 397 final), maximum allowable concentration (MAC) and annual average (AA) in mg L1 (n. a. not applicable; *depending on water hardness classes).
4607
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 6 0 1 e4 6 1 4
concentrations in the range of 0.08e3.57 mg L1 can be attributed to sulphur containing impurities and intermediate products (maximum of 1% sulphur impurity according to elemental analysis) on the one hand and to the partial solubility of ILs in the aqueous phase under neutral conditions on the other hand. In the latter case, obtained sulphur concentrations of 2.6.106 to 1.1.104 mol L1 would correspond to a leaching of 0.5e13% wt. As a consequence, we conclude that prior to their industrial use ILs have to be critically assessed regarding solubility and toxicity. In order to eliminate this possible disadvantage of ILs we are currently developing procedures for immobilization of ILs onto different surfaces e.g. in membranes (e.g. with PSF, PP, PE) or encapsulated in biomaterials as “backbones” and on supporting materials. The use of such materials as extracting agents can prevent the transfer of ILs into the aqueous phase. However, before such steps liquideliquid extraction studies are necessary in order to evaluate the potential of ILs regarding metal extraction. The biological effect of ILs similar to those investigated within this work are provided by http://www.il-eco.uft.unibremen.de/. However, it is known from the literature (Pham et al., 2010) that the toxicity of ILs is strongly dependent on the structure of the utilized anions and cations and a general conclusions cannot be made without experimental data. Nevertheless, within this study, we have implemented longchain quaternary ammonium cations with thiosalicylate anion in order to avoid toxic fluorine containing anions, such as PF6.
3.1.2.
Screening of IL extraction efficiency
For rapid assessment of the potential of the investigated ILs, a LPME-screening was performed employing model solutions and an extraction time of 120 min. Table 5 lists the extraction efficiencies of the different quaternary ammonium- and phosphonium-based ILs. It can be clearly seen that all investigated ILs revealed high extraction efficiencies for Ag, Hg and Pt, whereas only a limited fraction of As, Cr and Ni could be eliminated from the CaCl2 solutions after 120 min. Extraction efficiencies for Hg ranged from 80 to 100% for all tested ILs without obvious trends regarding IL-functionality. This corresponds to distribution ratios >1500 and is in accordance with a generally high solubility of Hg in both functionalized and unfunctionalized ILs, observed by several research groups e.g. Visser et al. (2002); Papaiconomou et al. (2008) and Germani et al. (2007). The similar tendency can be observed for Ag, which is highly soluble in all studied ILs, independent of the functional group. Obviously, sulphur containing ILs with thiosalicylate [TS], methylthiobenzoate [MTBA] and thiocyanate [SCN] anions, respectively, showed higher efficiencies, ranging from 82 to 100%, while ILs with benzoate [BA], hexanoate [HEX] or salicylate [Sal] anions showed a lower efficiency with a maximum of 66%. Corresponding distribution ratios >1000 are significantly higher compared to distribution ratios (approx. 40) observed from Papaiconomou et al. (2008) for ILs bearing nitrile functionality. For Pt [A336][SCN] and [PR4][MTBA] revealed excellent extraction efficiencies (95 and 97%), whereas the extraction rate of the corresponding quaternary ammonium IL [A336] [MTBA] was only 40% after 120 min. Again, the efficiencies obtained for the thiosalicylate anion containing ILs [A336][TS]
Table 5 e Extraction efficiency [%] (n [ 3) of evaluated IL for the extraction of metals from model solutions with an extraction time of 120 mina. [A336] [A336] [A336] [A336] [A336] [PR4] [PR4] [PR4] [TS] [BA] [MTBA] [Hex] [SCN] [TS] [Sal] [MTBA] Ag As Cd Cr Cu Hg Ni Pb Pt Sn Zn
94 e e e 95 83 e e 85 79 e
56 e 5 e e 84 e 5 69 e e
87 e 8 12 11 95 10 7 40 5 10
66 e e e e 80 e e 54 e e
97 7 15 16 17 93 7 14 95 17 43
82 e 14 e 81 93 12 41 78 64 24
47 5 e 11 19 100 e 8 64 48 e
100 8 38 e 37 91 e e 97 64 e
e: Extraction efficiency <5%; the relative total combined uncertainty of the given results is 7% (coverage factor 2) (ISO Guide to the Expression of uncertainty in measurement, 1993). a Aqueous phase 0.01 M CaCl2 (pH 7.5) with metals Ag, As, Cd, Hg, Pb, Pt, Sn c0 7.5 and Cr, Cu, Ni, Zn 75.0 mg L1 (approx.).
and [PR4][TS] were significantly higher as those of [BA], [HEX] or [Sal] containing anions. Regarding Cu and Sn, the thiosalicylate containing ammonium IL [A336][TS] showed high elimination rates for Cu (95%) and Sn (82%) The distribution ratios for Cu (>1900) are comparable with the results of Papaiconomou et al. (2008) for thioether egroup containing ILs, where significantly lower distribution ratios could be achieved with trioctylmethylammonium salicylate IL by Egorov et al. (2010) (approx. 30). While both [TS] containing ILs showed a remarkable extraction for Cu and Sn, [A336][BA] did not show any affinity for these two metals. It is interesting to mention that the viscosity of these three aromatic ILs is in the same range (see Table 1). Interestingly, Cd, Pb and Zn could be eliminated partly from the model solution by [PR4][TS], but there was no effect on the corresponding ammonium cation with the same anion. [A336][MTBA] showed maximal 10% extraction efficiency for those metals and no effect was obtained with the analogue [PR4][MTBA] for Pb and Zn while 38% of Cd could be eliminated. In average, 40% of those metals could be eliminated from the model matrix as [PR4][MTBA] shows the highest potential for Cd, [PR4][TS] for Pb and the highest extraction efficiency for Zn could be obtained by the thiocyanate containing IL [A336][SCN]. The results indicate that not only the functionality appended to the anion, but also the physicochemical properties of the investigated ILs exhibit a remarkable impact on extraction efficiency. Our results are in accordance with Papaiconomou et al. (2008), who have evaluated the elimination of numerous metals (e.g. Ag, Hg, Pd, Cu) from aqueous solutions with nitrile- and thioether functionalized ILs appended to pyridinium- as well as piperidinium cations and fluor-containing anions. Although it was shown that the extraction efficiency was predominantly governed by the functional group, the cation ring as well as the anion was strongly influencing the metal uptake of the investigated ILs. In general, extraction properties of the ILs based both on the quaternary ammonium- and phosphonium cation with [TS], [MTBA] or [SCN] functionality were significantly higher than
4608
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 6 0 1 e4 6 1 4
those of the ILs containing [BA], [Hex] or [Sal], indicating that the sulphur containing functional groups strongly affect the extraction efficiency. However, the presence of the sulphur functionality cannot be regarded as the sole explanation for the extraction efficiency. Indeed, the cationeanion combination, and therefore the IL as a whole (and resulting physicochemical properties), seems to have a great impact on the extraction behaviour.
3.1.3.
Time dependent extraction efficiency
To investigate the influence of contact time on IL extraction, time-resolved LPME experiments were performed. Fig. 2 shows the effect of contact time (0e120 min, measurement increments of 30 min, three independent replicates) of the tested ILs on elimination of Ag, Cd, Cu, Hg, Pb, Pt, Sn and Zn from model solutions. The results demonstrate that the concentration of Ag, Hg and Pt in the model solutions decreased with increasing contact time in the case of most ILs. An extrapolation of the observed elimination curves for the last named metals results in an estimated elimination efficiency of >90% for all investigated ILs except [A336] [MTBA] and in case of Pt for [A336][Hex] as well as they reached a steady state at approximately 40 and 50% respectively. Interestingly, both the quaternary ammonium and phosphonium ILs bearing [MTBA] functionality showed a fast uptake of metals within 30 min as >80% to 95% of Ag, Hg and >90% of Pt could be extracted with [PR4][MTBA] and no significantly decrease was observed within the remaining extraction time. In case of Cu and Sn the cations bearing [TS] functionality showed successful elimination potential within 120 min. It is observable that the extraction efficiency of [A336][TS] is higher than that of [PR4][TS] with the same contact time. These results may be explained with the significantly lower water content of [A336][TS] after equilibration (approx. 4.1% wt.) compared to the phosphonium analogue (approx. 10.6% wt.), which is in accordance with the observations of Visser et al. (Visser et al., 2002). Their results regarding Cd and Hg extraction with different TSILs dissolved in the hydrophobic IL 1-butyl-3-methylimidazolium hexafluorophosphate, indicated a strong impact of the decreased water content on increasing metal-ion distribution ratios. On the other hand, regarding [PR4][Sal], an unexpected low extraction efficiency for Cu was achieved, whereas Egorov et al. (2010) presented an efficient Cu extraction (distribution ratio approx. 30) in the bulk liquid/liquid extraction experiments with the Aliquatbased analogue [A336][Sal]. Extraction efficiencies of the investigated ILs for Zn, Cd and Pb are low. This may be attributed to the pH dependence of the extraction efficiencies leading to higher extraction ranges at higher pH (see results for communal and industrial waste waters below) and could also explain the higher extraction efficiencies for Cd reported in experiments using surface water (Kogelnig et al., 2008). In contrast to the results observed for Cu and Sn, a significant elimination potential could be observed with the more hydrophobic [PR4] based ILs. [PR4][MTBA] was the only IL which showed a reasonable extraction efficiency for Cd while moderate extraction of Zn and Pb was investigated with [PR4] [TS]. In fact, high extraction efficiencies for those metals can only be obtained at elevated contact times.
3.1.4.
Waste water extraction experiments
The experiments with waste water samples aimed at the evaluation of the suitability of ILs for (i) advanced treatment of the effluent of communal sewage treatment plants and (ii) the first cleaning step of industrial waste waters containing high amounts of different metal(oid)s. For this purpose the waste waters were sampled from six different sources, i.e. two effluents from communal sewage treatment plants (waste water 1 and 2) and four untreated industrial waste waters (waste water 3e6). The sampling sites were chosen considering pH value, concentration of metals and the content of suspended solids. Physical and chemical properties describing the waste water samples at the time of the metal extraction experiments are summarized in Table 2. To simulate the procedure in a waste water treatment plant, the waste water samples were subjected to LPME without filtration or centrifugation. [A336][TS] and [A336][SCN] have been chosen for the elimination studies of waste water 1 - 3 because of the excellent extraction potential for Ag, Hg and Cu. Waste water 4 - 6 showed a different composition regarding metal contamination and pH as the concentrations of Cu and Ag were low, but the samples contained high concentrations of Pb, Cd and Zn (see Table 2). As [PR4][TS] was the only IL with an extraction potential for Pb (approx. 40% from model solutions) and satisfactory extraction efficiencies for Cd and Zn, this substance was included in the waste water extraction experiments. Results of the extraction studies are shown in Table 6. No general conclusion can be made for evaluation of the extraction potential of tested ILs for metal(oid)s from model matrixes and real samples as various factors are influencing the mechanism of metal extraction. Several authors (Kalb et al., 2006; Egorov et al., 2010; Stojanovic et al., 2010) have evaluated a possible mechanism for task specific ILs based on quaternary ammonium and phosphonium cations and thiolas well as hydroxy group containing anions. Kalb et al. (2006) have proposed that trioctylmethylammonium thiosalicylate, [TOMATS], can extract heavy metals by forming metalthiolates and additionally a complex-bond with the carboxylate group of the thiosalicylate anion. Egorov et al. (2010) have successfully demonstrated that tricaprylylmethylammonium salicylate could extract Fe(III) and Cu(II) from aqueous solutions in form of salicylate complexes. Furthermore, Preston (1983) has shown that Co(II) and Ni(II) could be eliminated with tricaprylylmethylammonium thiocyanate. In this case, the extent of extraction was shown to depend upon the identity of the counter anion of the metal present in the aqueous phase. Stojanovic et al. (2010) have evaluated phosphonium-based ILs with thiol- and thioether containing anions for the elimination of Pt(IV). Obtained results clearly demontrated that physicochemical parameters (e.g. viscosity) have also a remarkable influence for metal distribution. Therefore it is clear for the results obtained within this study that not only a simple “task specifity” of the evaluated ILs is responsible for the metal distribution, but also their physico-chemical properties as well as the composition of the matrix. The initial concentration of metal(oid)s in the samples was observed to be an important factor influencing extraction efficiency. Pb uptake from waste water 4 (c0 ¼ 412 mg L1) was
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 6 0 1 e4 6 1 4
4609
Fig. 2 e Effect of contact time on metal extraction (n [ 3) from model solution by ammonium- and phosphonium-based ILs compared to a reference sample. Aqueous phase: 0.01 M CaCl2 (pH 7.5) with c0 7.5 mg LL1 (Ag, Cd, Hg, Pb, Pt, Sn) and c0 75 mg LL1 (Cu, Zn) respectively, Vaq/VIL 2000:1. The total combined uncertainty of the given concentrations is 7%, scaled by a coverage factor of 2 to give a confidence level of approx. 95% (ISO Guide to the expression of uncertainty in measurement; ISO Guide to the Expression of uncertainty in measurement, 1993).
4610
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 6 0 1 e4 6 1 4
Fig. 2 e (continued).
significantly higher up to 90% than from model solutions containing Pb concentrations in the low mg L1 range. Ni uptake from waste water 1 and 2 with the highest Ni content was in the range of 86e92% and decreased with decreasing Ni
content of the water samples. The same effect was found for Ag, as up to 75% could be extracted from a waste water sample containing concentrations, while the extraction efficiency was less than 45% in a model solution with an initial concentration
4611
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 6 0 1 e4 6 1 4
Table 6 e Extraction efficiencies [%] of metals from waste waters by ionic liquids with an extraction time of 120 min (n [ 3). Waste water 1
Waste water 2
[A336][TS]
[A336][SCN]
[A336][TS]
[A336][SCN]
[A336][TS]
[A336][SCN]
68% e * e 73% 89% 86% * * * 65%
86% e * e 73% 52% 86% * * * 69%
75% 10% * 14% 92% * 96% 14% * * 79%
72% 7% * 8% 61% * 90% e * * 73%
43% 5% * 9% 60% * 36% * * * 69%
25% 9% * 6% 24% * e * * * 44%
Ag As Cd Cr Cu Hg Ni Pb Pt Sn Zn
Waste water 4
Ag As Cd Cr Cu Hg Ni Pb Pt Sn Zn
Waste water 3
Waste water 5
Waste water 6
[A336][TS]
[A336][SCN]
[PR4][TS]
[A336][TS]
[A336][SCN]
[PR4][TS]
[A336][TS]
[A336][SCN]
[PR4][TS]
* 10% e 13% 45% * e 5% * * 57%
* 8% e 20% 44% * 43% 79% * * 68%
* e e 13% 53% * 48% 90% * * 86%
* e * 18% 22% * 18% e 31% * 13%
* e * 10% 15% * 8% 9% 33% * e
* e * 12% 20% * 14% e e * 8%
* e e 26% 10% * e 7% * * 9%
* e e 3% e * e e * * e
* e e 5% e * e e * * e
e: Extraction efficiency < 5%. *: Concentration in reference sample < LOD within 120 min; the relative total combined uncertainty of the given results is 7% (coverage factor 2) (ISO Guide to the Expression of uncertainty in measurement, 1993).
<1 mg L1. Elimination of Zn from waste water 1e4 was insignificant dependent on the concentration which ranged from 3.49 to 40.6 mg L1. With the tested IL 44e70% of Zn could be extracted from those waste waters, which gave significantly higher extraction efficiencies compared to the model solution. Low elimination rates of Zn from waste water 5 and 6 indicate the poor extraction from Zn strongly bound to suspended solids as discussed below (Popp et al., 2008). Generally, highly contaminated waste water yielded superior extraction efficiencies as waters with low level contaminations. Another factor influencing extraction efficiency is the pH of the samples. It is well known that the elimination efficiencies of metal (complexes) are controlled by pH and ionic strength of the system (Visser et al., 2002; Wei et al., 2003., Egorov et al., 2010; Lertlapwasin et al., 2010). This effect might be caused by a higher stability of the complexes in a more basic milieu and would therefore explain the increased elimination of Ni, Zn and Pb from waste water 1e4 as the pH of those samples was in the range of 7.9e9.4 compared to the model matrix with a pH of 7.5. We found that the concentration of suspended particulate matter represents a further important characteristic affecting extraction efficiency. As published elsewhere (Popp et al., 2008), metals are specifically distributed between suspended particles and the water phase, which implicates that the particles may act as competitive adsorbent (“extractant”) in the water/IL system. This assumption is supported by the fact
that waste waters 5 and 6 containing high fractions of suspended solids showed generally low extraction rates for metals by ILs. Moreover, those samples showed a high electrical conductivity of 8470 in waste water 5 and 23000 mS cm1 in waste water 6, which suggests high contamination of total
Fig. 3 e HPLC-ICP-MS chromatogram obtained from an aged cisplatin solution (initial concentration of cisplatin 50 mg LL1, incubation time 48 h) showing the signals of cisplatin and the degradation products monoaquacisplatin and diaquacisplatin (full line). The dashed line shows the effect of [PR4][TS] after an extraction time of 120 min indicating a high and moderate extraction efficiency for monoaquacisplatin and diaquacisplatin, respectively.
4612
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 6 0 1 e4 6 1 4
Fig. 4 e Extraction efficiencies of phosphonium-based ILs for the extraction of inorganic Pt, cisplatin and its degradation products mono- and diaquacisplatin (incubation for 48 h at 20 C) from model solution with an extraction time of 120 min. Aqueous phase 0.01 M CaCl2 (pH 7.5) with Pt c0 7.5 mg LL1 and intact cisplatin c0 50 mg LL1, respectively. The inset shows the extraction efficiencies of the different ILs for extraction of monoaquacisplatin (dashed lines) and diaquacisplatin (full lines).
dissolved solids (TDS). It can be concluded that both, the interaction of metal(oid)s with suspended particulate matter and TDS, strongly influenced the mechanism of extraction leading to decreased extraction efficiencies. Nevertheless, the results indicate that a variety of toxic metals could be efficiently extracted from contaminated water by the use of small amounts of ILs. We can also assume that several selective TSILs might be used in parallel to separate multiple contaminants from a mixed waste water stream. As evaluated ILs are miscible with each other it is absolutely realistic that the efficiency for the extraction process might also be increased, when choosing the correct composition of the IL extracting system. As a pre-requisite, the selection of appropriate ILs or task specific IL mixtures for waste water treatment certainly requires a preliminary characterisation of the waste water.
3.2. Extraction efficiencies of ILs for extraction of cancerostatic platinum compounds from model solutions The potential of ILs regarding the elimination of the cancerostatic platinum compounds (CPC) from model solutions was tested, as recent investigations have shown that hazardous platinum containing compounds are only partially removed by conventional sewage treatment (Lenz et al., 2007). Since oxaliplatin is supposed to be exclusively present in waste water treatment plants in the form of its biotransformation products and various adducts with biomolecules, our elimination experiments were restricted to cisplatin and carboplatin. In waste water, carboplatin is mainly present as intact drug, while cisplatin is degraded e in dependence on pH, chloride concentration and age of the waste water e to the two major aquation products cis-[PtCl(H2O)(NH3)2]þ (monoaquacisplatin) and cis[Pt(H2O)2(NH3)2]2þ (diaquacisplatin) as well as to the neutral hydroxocomplexes cis-[PtCl(OH)(NH3)2] (monohydroxocisplatin) and cis-[Pt(OH)2(NH3)2] (dihydroxocisplatin). In order to detect
these species in the model solutions used for the elimination experiments, speciation analysis by HPLC-ICP-MS was performed. Fig. 3 shows the chromatographic separation of cisplatin, monoaquacisplatin and diaquacisplatin in an aged standard solution (incubation time 48 h) with an initial cisplatin concentration of 50 mg L1. For investigation of the IL extraction efficiencies for intact platinum drugs, LPME experiments were performed with freshly prepared model solutions spiked with cisplatin and carboplatin at concentration levels of 50 mg L1. Compared to the excellent results obtained for inorganic Pt employing phosphonium-based IL bearing thiosalicylate and -2-(methylthio)benzoate functionality, neither intact cisplatin nor carboplatin were significantly extracted by these ILs. In order to investigate the potential of the tested IL regarding the elimination of cisplatin degradation products, extraction experiments on aged model solutions (48 h) were performed. For phosphonium-based ILs increased extraction efficiencies could be observed for the cisplatin metabolites. The phosphonium-based IL bearing thiosalicylate functionality showed a significantly higher extraction efficiency for monoaquacisplatin as the -2-(methylthio)benzoate functionality. In the case of diaquacisplatin no significant difference in the extraction efficiency of these two phosphonium-based IL could be observed (see Fig. 4). These results are of particular relevance for the elimination of cisplatin, since it is known from a previous study that more than 75% of this drug enters sewage treatment as highly active monoaquacisplatin/monohydroxocisplatin (Hann et al., 2003).
4.
Conclusion
This study reports the evaluation of various anion functionalized ammonium- and phosphonium-based ILs regarding purity and extraction efficiencies for metal(oid)s
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 6 0 1 e4 6 1 4
and cancerostatic platinum compounds from model matrixes and from communal and industrial waste water samples applying liquid-liquid extraction procedures. Concerning partial solubility of ILs in water, an appropriate immobilisation strategy is a pre-requisite to suppress the loss of ILs into the water phase. Current investigations deal with the development of appropriate immobilisation procedures. According to our results, task specific ILs represent an auspicious procedure for selective cleaning of contaminated waste water. In particular, the application of thiol- and thioether- functionalized ILs [A336][TS], [A336] [MTBA], [PR4][TS] and [PR4][MTBA] for communal waste water treatment and the use of [A336][SCN] for industrial waste water with high level Zn contamination is recommended. As a matter of fact the variation of ILs and IL mixtures is only possible in laboratory experiments. The application of ILs on a large industrial scale needs the selection of a low number of low cost ILs, which have to be proposed for authorization by REACH legislation. Due to the fact that functionalized ILs incorporating complexing thiol-groups did not indicate significant affinity towards planar intact platinum complexes, another approach should be investigated in the future. Since the hydrophobicity and polarizability of ILs seem to have a remarkable impact on the extraction processes, a design of strongly hydrophobic, apolar ILs, and hence extraction of apolar platinum complexes from the aqueous phase as a whole, may be a useful approach. Future research should concern procedures for stripping and recovery of metals and metal compounds from ILs as well as the recycling of ILs for a potential use on a larger scale. Kalb et al. (2006) could strip extracted heavy metals from the commercially available anion functionalized IL trioctylmethylammonium thiosalicylate [TOMATS] via oxidation of the thiol-group with HNO3. This was also successfully achieved for uranium with tricaprylylmethylammonium thiosalicylate [A336][TS], as demonstrated in a previous study (Srncik et al., 2009). For analytical purposes - such as selective pre-concentration of desired metals or facilitating sample preparation methods e this is a prosperous result. However, due to the oxidation the IL is destroyed, making recycling impossible. Therefore we are currently evaluating different back-extracting agents in order to establish continuous application of ILs. Preliminary experiments using EDTA as back-extracting agent turned out to be efficient for the recovery of e.g. Pb(II) from the IL [A336][TS], making recycling possible. Furthermore, electro deposition of metals would also be an interesting field of further research, especially for noble metals such as Pt. In our opinion, with an appropriate immobilization strategy to avoid water miscibility of ILs and an effective back extraction procedure which enables the reuse of ILs a positive life-cycle analysis may be expected.
Acknowledgements This work was supported by grants from the Austrian Federal Ministry of Agriculture, Forestry, Environment and Water
4613
Management (BMLFUW, Project Title: “Elimination of priority substances from waste water“). Maria Fuerhacker (University of Natural Resources and Life Sciences e BOKU Vienna, Department of Water, Atmosphere and Environment) is gratefully acknowledged for providing the waste water samples.
references
Commissie integraal waterbeheer, May 2000. Normen voor het waterbeheer (In Dutch). Dietz, M.L., Dzielawa, J.A., 2001. Ion-exchange as a mode of cation transfer into room-temperature ionic liquids containing crown ethers: implications for the “greenness” of ionic liquids as diluents in liquid-liquid extraction. Chem. Commun., 2124e2125. DIN 38409-2, 1987. German Standard Methods for the Examination of Water, Waste Water and Sludge; Parameters Characterizing Effects and Substances (Group H); Determination of Filterable Matter and the Residue on Ignition (H2). Beuth Verlag, Berlin, Germany (In German). Directive 2000/60/EC of the European Parliament and of the Council of 23 October 2000 establishing a framework for community action in the field of water policy. Off. J. Eur. Communities L 327/1, 22.12.2000. Deutsches Umweltbundesamt, 2003. Produktion, Verwendung und Emissionsquellen ausgewa¨hlter priorita¨rer Stoffe in der EU. Deutsches Umweltbundesamt (In German). Domanska, U., Rekawek, A., 2009. Extraction of metal ions from aqueous solutions using imidazolium based ionic liquids. J. Sol. Chem. 38, 739e751. Egorov, V.M., Djigailo, D.I., Momotenko, D.S., Chernyshov, D.V., Torocheshnikova, I.I., Smirnova, S.V., Pletnev, I.V., 2010. Taskspecific ionic liquid trioctylmethylammonium salicylate as extracting solvent for transition metal ions. Talanta 80, 1177e1182. Germani, R., Mancini, M.V., Savalli, G., Spreti, N., 2007. Mercury extraction by ionic liquids: temperature and alkyl chain length effect. Tetrahedron Lett. 48 (10), 1767e1769. Han, X., Armstrong, D.W., 2007. Ionic liquids in separation. Acc. Chem. Res. 40, 1079e1086. Hann, S., Koellensperger, G., Stefa´nka, Zs., Stingeder, G., Fu¨rhacker, M., Buchberger, W., Mader, R.M., 2003. Application of HPLC-ICP-MS to speciation of cisplatin and its degradation products in water containing different chloride concentrations and in human urine. J. Anal. Spectrom. 18, 1391e1395. Hann, S., Stefa´nka, Zs., Lenz, K., Stingeder, G., 2005. Novel separation method for highly sensitive speciation of cancerostatic platinum compounds by HPLC-ICP-MS. Anal. Bioanal. Chem. 381, 405e412. International Agency for Research on Cancer (IARC), 1987. Monographs on the Evaluation of Cancerogenic Risks to Humans Lyon, supp7. ISO Guide to the Expression of Uncertainty in Measurement, 1993 Geneva, Switzerland. Kalb, R.S., 2005. Method for Producing Ionic Liquids, Ionic Solids or Mixtures thereof WO 2005/021484. Kalb, R.S., Krachler, R., Keppler, B.K., 2006. Determination of heavy metal polluted process water, waste water and filter cake with high performance. In: Ho¨flinger, W. (Ed.), Chemical Industry and Environment V, vol. I Vienna, Austria. Kelland, L., 2007. The resurgence of platinum-based cancer chemotherapy. Nat. Reviews/Cancer 7, 573e584.
4614
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 6 0 1 e4 6 1 4
Kogelnig, D., Stojanovic, A., Galanski, M., Groessl, M., Jirsa, F., Krachler, R., Keppler, B.K., 2008. Greener synthesis of new ammonium ionic liquids and their potential as extraction agents. Tetrahedron Lett. 49, 2782e2785. Kulkarni, P.S., Branco, L.C., Crespo, J.G., Nunes, M.C., Raymundo, A., Afonso, C.A.M., 2007. Comparison of physicochemical properties of new ionic liquids based on imidazolium, quaternary ammonium, and guanidinium cations. Chem. Eur. J. 13, 8478e8488. Ku¨mmerer, K., Helmers, E., 1997. Hospital effluents as a source for platinum in the environment. Sci. Total. Environ. 193, 179e184. Lenz, K., Hann, S., Koellensperger, G., Stefa`nka, Zs., Stingeder, G., Weissenbacher, N., Mahnik, S.N., Fuerhacker, M., 2005. Presence of cancerostatic platinum compounds in hospital wastewater and possible elimination by adsorption to activated sludge. Sci. Total. Environ. 345, 141e152. Lenz, K., Koellensperger, G., Hann, S., Weissenbacher, N., Mahnik, S.N., Fuerhacker, M., 2007. Fate of cancerostatic platinum compounds in biological wastewater treatment of hospital effluents. Chemosphere 69, 1765e1774. Lertlapwasin, R., Bhawawet, N., Imyim, A., Fuangswasdi, S., 2010. Ionic liquid extraction of heavy metal ions by 2aminothiophenol in 1-butyl-3-methylimidazolium hexafluorophosphate and their association constants. Sep. Purif. Tech. 72, 70e76. Liu, J., Jiang, G., Chi, Y., Cai, Y., Zhou, Q., Hu, J., 2003. Use of ionic liquids for liquid-phase micro extraction of polycyclic aromatic hydrocarbons. Anal. Chem. 75, 5870e5876. Liu, J., Chi, Y., Jiang, G., Tai, C., Peng, J., Hu, J., 2004. Ionic liquidbased liquid-phase micro extraction, a new sample enrichment procedure for liquid chromatography. J. Chrom. A. 1026, 143e147. Liu, J., Chi, Y., Jiang, G., 2005. Screening the extractability of some typical environmental pollutants by ionic liquids in liquidphase micro extraction. J. Sep. Sci. 28, 87e91. Luo, H., Dai, S., Bonnesen, P.V., 2004. Solvent extraction of Sr2þ and Csþ based on room-temperature ionic liquids containing monoaza-substituted crown ethers. Anal. Chem. 76, 2773e2779. Marsh, K.N., Boxall, J.A., Lichtenthaler, R., 2004. Room temperature ionic liquids and their mixtures e a review. Fluid Phase Equilibria 219, 93e98. Moore, F.L., 1964. New approach to separation of trivalent actinide elements from lanthanide elements-selective liquidliquid extraction with tricaprylmethylammonium thiocyanate. Anal. Chem. 36, 2158e2162. Pandey, S., 2006. Analytical applications of room-temperature ionic liquids: a review of recent efforts. Anal. Chim. Acta 556, 38e45.
Papaiconomou, N., Lee, J., Salminen, J., von Stosch, M., Prausnitz, J.M., 2008. Selective extraction of copper, mercury, silver and palladium ions from water using hydrophobic ionic liquids. Ind. Eng. Chem. Res. 47, 5080e5086. Preston, J.S., 1983. Solvent extraction of Cobalt(II) and Nickel(II) by a quaternary ammonium thiocyanate. Sep. Sci. Technol. 17, 1697e1718. Popp, M., Koellensperger, G., Stingeder, G., Hann, S., 2008. Novel approach for determination of trace metals bound to suspended solids in surface water samples by inductively coupled plasma sector field mass spectrometry (ICP-SFMS). J. Anal. Spectrom. 23, 111e118. Pham, T.P., Cho, C., Yun, Y.S., 2010. Environmental fate and toxicity of ionic liquids: a review. Water Res. 44 (2), 352e372. Regel-Rosocka, M., 2009. Extraction removal of zinc(II) from chloride liquors with phosphonium ionic liquids/toluene mixtures as novel extractants. Sep. Purif. Tech. 66, 19e24. Rios de los, A.P., Herna´ndez-Ferna´ndez, L.J., Lozano, L.J., Sa´nchez, S., Moreno, J.I., Godinez, C., 2010. Removal of metal ions from aqueous solutions by extraction with ionic liquids. J. Chem. Eng. Data 55, 605e608. Shimojo, K., Goto, M., 2004. Solvent extraction and stripping of silver ions in room-temperature ionic liquids containing calixarenes. Anal. Chem. 76, 5039e5044. Srncik, M., Kogelnig, D., Stojanovic, A., Koerner, W., Krachler, R., Wallner, G., 2009. Uranium extraction from aqueous solutions by ionic liquids. Appl. Radiat. Isot. 67, 2146e2149. Stojanovic, A., Kogelnig, D., Fischer, L., Hann, S., Galanski, M., Groessl, M., Krachler, R., Keppler, B.K., 2010. Phosphonium and ammonium ionic liquids with aromatic anions: synthesis, properties and platinum extraction. Aust. J. Chem. 63, 511e524. Swatloski, R.P., Holbrey, J.D., Rogers, R.D., 2003. Ionic liquids are not always green: hydrolysis of 1-butyl-3-methylimidazolium hexafluorophosphate. Green Chem. 5, 361e363. Visser, A.E., Swatloski, R.P., Reichert, W.M., Mayton, R., Sheff, S., Wierzbicki, A., Davis Jr., J.H., Rogers, R.D., 2001. Task-specific ionic liquids for the extraction of metal ions from aqueous solutions. Chem. Commun., 135e136. Visser, A.E., Swatloski, R.P., Reichert, W.M., Mayton, R., Sheff, S., Wierzbicki, A., Davis Jr., J.H., Rogers, R.D., 2002. Task-specific ionic liquids incorporating novel cations for the coordination and extraction of Hg2þ and Cd2þ: synthesis, characterization and extraction studies. Environ. Sci. Technol. 36, 2523e2529. Wei, G., Yang, Z., Chen, C., 2003. Room temperature ionic liquid as a novel medium for liquid/liquid extraction of metal ions. Anal. Chim. Acta 488, 183e192. Zhao, H., Xia, S., Ma, P., 2005. Use of ionic liquids as "green" solvents for extractions. J. Chem. Technol. Biotechnol. 80, 1089e1096.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 6 1 5 e4 6 2 2
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Influence of attached bacteria and biofilm on double-layer capacitance during biofilm monitoring by electrochemical impedance spectroscopy Taeyoung Kim a, Junil Kang a, Joon-Hee Lee b, Jeyong Yoon a,* a
World Class University (WCU) program of Chemical Convergence for Energy & Environment (C2E2), School of Chemical and Biological Engineering, College of Engineering, Seoul National University (SNU), 599 Gwanak-ro, Gwanak-gu, Seoul 151-744, Republic of Korea b Laboratory of Microbiology, Department of Pharmacy, College of Pharmacy, Pusan National University, Research Building 532, San 30, Jangjun-Dong, Geumjung-Gu, Busan 609-735, Republic of Korea
article info
abstract
Article history:
Development of an effective strategy for biofilm control in water-related system has
Received 6 January 2011
become a matter of significant concern nowadays. Electrochemical monitoring, especially
Received in revised form
electrochemical impedance spectroscopy (EIS), is one of the efficient approaches to dealing
21 April 2011
with biofilm-related issues. However, currently used EIS methods without a redox probe
Accepted 13 June 2011
intend to detect all effects generated from media components, bacteria, and bacterial
Available online 24 June 2011
metabolites, which used to make the signals from the attached bacteria and biofilm weakened. In this study, we tried improved EIS measurement to monitor bacterial adhe-
Keywords:
sion and biofilm maturation using a double-layer capacitance. In this improved method,
Biofilm monitoring
we minimized background signal by subtracting the interference of electrolyte caused by
Bacterial adhesion
bacterial metabolism. Pseudomonas aeruginosa PA14 wild type and wspF mutant that form
Biofilm maturation
the biofilm of distinct nature were used for the model strains to test our method. During
Electrochemical
bacterial adhesion and biofilm maturation, EIS data were collected and equivalent circuit
impedance spectroscopy
analysis was carried out to obtain constant phase element (CPE) values representing
Double-layer capacitance
double-layer capacitance. Since the influence by the bacterial growth-related culture media condition was eliminated by adopting fresh electrolyte at the measurement, the contribution of attached bacteria and biofilm was exclusively measured. As a result, the bacterial adhesion at the early stage of biofilm development was specifically monitored from reduction in double-layer capacitance. Particularly, the plateau in double-layer capacitance appeared upon biofilm maturation, indicating that biofilm maturation could be expected beyond this point. In conclusion, this study found that measurement of double-layer capacitance based on EIS could provide a monitoring parameter suggesting bacterial adhesion and the initiation point of biofilm maturation. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
Biofilm is a microbial community in which microbial cells adhere to surfaces as microcolonies surrounded by
a polysaccharide matrix. Biofilm growth in diverse environments causes various problems such as bacterial contamination in distribution systems, biofouling and biocorrosion in industrial water systems, and persistent bacterial infections
* Corresponding author. Tel.: þ82 2 880 8927; fax: þ82 2 876 8911. E-mail address: [email protected] (J. Yoon). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.06.010
4616
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 6 1 5 e4 6 2 2
in animal body (LeChevallier et al., 1987; Mattila-Sandholm and Wirtanen, 1992; Costerton et al., 1999; Beech and Sunner, 2004; Dexter and LaFontaine, 1998). Therefore, development of an appropriate and effective strategy to control the biofilm formation is an important issue in many fields. So far, the antimicrobial agents have been widely used and almost only choice for this purpose. However, it has been known to be unsuccessful to treat mature biofilms with antimicrobials due to strong antibiotic resistance of the cells in biofilm (Mah and O’Toole, 2001). An alternative solution for this problem has been to develop an appropriate biofilm monitoring method that enables us to act ahead of the biofilm growth (Flemming, 2003). In order to achieve this goal, turbidity measurement (Klahre and Flemming, 2000), fiber optical device (Tamachkiarow and Flemming, 2003), and photoacoustic spectroscopy (Schmid et al., 2004) have been applied. Among the various monitoring techniques, electrochemical methods have attracted more attention because in-situ, real-time, and on-line monitoring are possible as well as monitoring sensors are convenient for installation and operation. In particular, electrochemical impedance spectroscopy (EIS) is suitable for biofilm monitoring since small potential perturbations have small or no effect on bacteria (Franklin et al., 1991). Common approaches that utilize EIS without a redox probe have been mainly focused on culture media. Its changes in components caused by bacterial metabolism during growth were typically reflected to the EIS data such as solution resistance and double-layer capacitance (Yang et al., 2003; Felice et al., 1999; Flint and Brooks, 2001). The main purpose of these studies was to quickly detect a presence of bacteria; hence the monitoring was conducted during initial bacterial growth. In particular, the changes in culture media dominantly influenced the signals so that contributions of attached bacteria to the signals were marginally evaluated. Though a portion of such limitations have been overcome recently (Dheilly et al., 2008; Munoz-Berbel et al., 2008), further studies on correlating the electrochemical signals with the extent of the bacterial adhesion and biofilm maturation are still need to be investigated. In this study, the effects of bacterial adhesion and biofilm maturation on double-layer capacitance were investigated by EIS analysis on platinum electrodes using Pseudomonas aeruginosa PA14 wild type and PA14 wspF mutant. In order to improve this method to understand the contribution of attached bacteria and biofilm, we excluded interference resulting from the culture media by introduction of fresh electrolyte. In particular, fluorescent images were analyzed together with the EIS to confirm the relationship between double-layer capacitance and the extent of the bacterial adhesion and biofilm maturation.
2.
Materials and methods
2.1.
Bacteria and culture media
Pseudomonas aeruginosa PA14 wild type and PA14 wspF mutant strains were grown in 1/10-strength Tryptic Soy Broth (TSB, Becton, Dickinson and Company, USA) at 37 C for 18 h. For bacterial adhesion and biofilm experiments, the cells were
harvested by centrifugation and washed twice with M9 minimal salts (BD, USA) supplemented with 20 ml of 20% glucose and 2 ml of 1.0 M MgSO4 per liter. The cell density was adjusted to approximately 106 CFU/ml using UV/visible spectrophotometer (Agilent Technologies, USA) in which an optical density of 0.1 at 600 nm indicated 108 CFU/ml.
2.2.
Electrode preparation
Platinum disk electrodes (CH instrument, USA), 2 mm in diameter and insulated with polychlorotetrafluoroethylene (PCTFE), were polished using 0.05-mm alumina powder (CH instrument, USA). Afterward, the electrodes were mildly washed and sonicated with ultrapure water for 15 min followed by nitrogen blow. Sterilization was carried out under ultraviolet ramp (254 nm) for 15 min and using 70% ethanol for the platinum part and the other part, respectively. Prior to the bacterial adhesion and biofilm formation experiments, the electrodes were kept immersed overnight in sterile M9 minimal salts with stirring (100 rpm).
2.3.
Bacterial adhesion and biofilm formation
A cylindrical glass reactor, with an output feed line located on the side wall was manufactured with a residence volume of 120 ml, which is 1/3-scale of conventional CDC biofilm reactor (CBR, BioSurface Technologies Corp., USA). Three platinum electrodes were inserted into a Teflon lid, which was then installed on the reactor. For the initial attachment of bacterial cells, the reactor was filled with a cell suspension of 106 CFU/ ml in M9 minimal salts supplemented with glucose and MgSO4 and agitated with magnetic stirrer at 100 rpm at room temperature. After 24 h of bacterial adhesion, fresh nutrients (1/10-strength M9 minimal salts containing glucose and MgSO4) were supplied for additional 48 h through the input feed line for further development of the attached bacteria into the mature biofilm. The flow rate was adjusted using a peristaltic pump (Gilson, Inc., USA) to attain a residence time of 30 min. Additionally, a control experiment excluding bacteria was carried out with M9 minimal salts in batch mode.
2.4.
Electrochemical impedance spectroscopy (EIS)
All electrochemical experiments were carried out in a threeelectrode cell with a platinum disk electrode as the working electrode (CH instrument, USA), a platinum wire (Princeton Applied Research, USA) as the counter electrode, and Ag/AgCl (saturated KCl) as the reference electrode (Princeton Applied Research, USA). For the electrochemical analysis, the platinum disk electrodes containing bacteria or biofilm were transferred from the reactor filled with the cell suspension to an electrochemical cell filled with fresh M9 minimal salts (pH 6.8), in which EIS analysis was conducted at predetermined cultivation times of 0, 6, 24, 48, and 72 h. The platinum disk electrodes were transferred to the reactor filled with the cell suspension after every measurement for further bacterial adhesion or biofilm maturation. EIS data were recorded using PARSTAT 2273 (Princeton Applied Research, USA) and ac 10 mV at open circuit potential between 1e100,000 Hz was applied via PowerSuite software (Princeton Applied Research,
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 6 1 5 e4 6 2 2
USA). Then, the impedance data after analyzing with an equivalent circuit utilizing software (ZView, Scribner Associates Inc., USA) was represented by the percent change in double-layer capacitance (ΔCdl) according to Eq. (1), DCdl ¼ ðCPEt CPE0 Þ=CPE0 100%
(1)
where CPEt and CPE0 are the constant phase element (CPE) representing double-layer capacitance in the modeling procedure at the specific and initial time of measurement, respectively. Representative data of averaged ΔCdl and its standard deviation obtained from the three electrodes were plotted versus time.
2.5.
Protein adsorption
Protein solution was prepared by adding 200 mg of tryptone (Becton, Dickinson and Company, USA) per liter of 0.1 M phosphate buffer solution (pH 7.1). Polished platinum disk electrodes were inserted into a reactor lid, which was installed on a cylindrical glass reactor. EIS data were recorded every 10 min during first 20 min in the 0.1 M phosphate buffer solution without tryptone. Then, the electrodes were transferred to the protein solution followed by EIS analysis in fresh 0.1 M phosphate buffer solution at predetermined times of 2, 4, 6, 8, 10, 20, 30, and 40 min. After every measurement, the electrodes were transferred back to the protein solution for further adsorption. The recorded EIS data were converted into ΔCdl as described in the previous Section 2.4. Representative data with standard deviation obtained from the three electrodes was used in this study.
2.6.
Image analysis
The electrode used in the EIS analysis was unsuitable for microscopic observation since the structure of the electrode did not allow the platinum part to face the lens. To solve this problem and allow visualization of the attached bacteria and biofilm, the same experiment described in the Section 2.3 was conducted on a platinum surface (1 cm 1 cm), which was prepared by physical deposition of platinum onto Si wafer (SITECH, INC., USA) using an e-gun evaporator (Maestech Co., Ltd., Korea). Consequently, we obtained a 100-nm-thick platinum surface on the 100-nm-thick SiO2 wafer, and Cr was deposited before platinum evaporation for better adhesion. Bacteria on this surface were stained with a fluorescent dye (FilmTracer FM 1-43 Green Biofilm Cell Stain, Invitrogen, USA) and observed via confocal laser scanning microscope (CLSM, eclipse 90i, Nikon, Japan). The data obtained from the five different positions were reconstructed into top-down images utilizing Imaris software (Bitplane, Switzerland) and surface coverage was calculated using i-solution (IMT technology, USA).
3.
Results and discussion
Fig. 1(a) demonstrates an equivalent circuit for the monitoring of bacterial adhesion and biofilm maturation representing electrode/electrolyte interface. It is composed of solution resistance (Rsol), charge-transfer resistance (Rct), and double-
4617
layer capacitance (Cdl). The solution resistance was determined between the working electrode and reference electrode and charge-transfer resistance was adopted to describe the leakage current. In the case of double-layer capacitance, constant phase element (CPE) was introduced to represent the inhomogenieties of the electrode/electrolyte interface (Brug et al., 1984). This equivalent circuit was selected to describe the system where no faradaic reaction occurs, which is consistent with the measurement conditions we made. At 0.3 V vs. Ag/AgCl in saturated KCl, an approximate open circuit potential of our system, the attached bacteria were unable to participate in the electrochemical reaction since gram-negative bacteria are known to show peak current around 0.70e0.74 V vs. SCE (Matsunaga and Nakajima, 1985). In addition, the selected electrolyte and electrode are hardly oxidized or reduced so that we excluded any faradaic reactions. Among the three parameters in the equivalent circuit, the CPE value selected for a monitoring parameter denoting double-layer capacitance was obtained by fitting procedure. The representative fitting results of PA14 wild type are shown in Fig. 1(b)e(d), where each of them demonstrates Nyquist, Bode, and ColeeCole plot. The ColeeCole plot has relation to the impedance by Eq. (2), C ¼ 1=juZ
(2)
where C is the real and imaginary part of the capacitance in ColeeCole plot, Z is the real and imaginary part of the impedance, j ¼ (1)1/2, u ¼ 2pf, and f is the frequency. During the fitting procedure, we focused on obtaining the CPE value which can be taken from the semi-circle in the ColeeCole plot. The accuracy of the charge-transfer resistance (Rct) was poor in this regard and its meaning was obscure to interpret our system so that this value was not used in this study. In addition, it should be noted that the CPE was not converted to double-layer capacitance in this study (Eq. (1)). In this respect, we correlated the fitted data to the degree of bacterial adhesion and biofilm maturation by the change in CPE values (ΔCdl). Representative impedance spectra of P. aeruginosa PA14 wild type are shown in Fig. 2(a) and (b), which correspond to Bode and ColeeCole plot, respectively. In the Bode plot, impedance increases over time in the low frequency region (1e100 Hz) where double-layer capacitance provides main contribution. The change is more clearly observed in the ColeeCole plot (Fig. 2(b)), and quantitative changes in ΔCdl values over P. aeruginosa PA14 wild type, wspF mutant, and control are shown in Fig. 2(c). The wspF mutant was chosen as it overproduces extracellular polymeric substance (EPS) to high level, thus shows fast bacterial adhesion and yields thick biofilm compared to the wild type. The ΔCdl was clearly reduced in both strains when compared to the control. Whereas the ΔCdl of control was reduced by 0.1% at 6 h and 1.7% at 24 h, those of PA14 wild type and wspF mutant were reduced by 6.8% and 8.9% at 6 h and by 13.3% and 14.7% at 24 h, respectively. This result implies that the amount of bacteria attached to the electrode surface can be correlated to the double-layer capacitance which is defined by Eq. (3), Cdl ¼ 330 A=d
(3)
4618
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 6 1 5 e4 6 2 2
Fig. 1 e Equivalent circuit and representative impedance spectra with fitting result. (a) Equivalent circuit used in this study to describe electrode/electrolyte interface for monitoring bacterial adhesion and biofilm maturation. Rsol represents the resistance between the working and reference electrodes, CPE represents double-layer capacitance, and Rct represents the charge-transfer resistance related to the leakage current. Representative impedance spectra (P. aeruginosa PA14 wild type) and fitting results are provided as (b) Nyquist, (c) Bode, and (d) ColeeCole plot, respectively.
where 3 is the dielectric constant of the electrolyte, 30 (8.854 1012 F/m) is the permittivity of free space, d (m) is the thickness of the double-layer, and A (m2) is the electrode area (Bard and Faulkner, 1980). Since EIS measurement was conducted in fresh electrolyte, which provided a constant electrolyte composition to every measurement, double-layer capacitance of Eq. (3) would be mainly dependent on the electrode area. The effects of the changes in culture media on the double-layer capacitance caused by bacterial metabolism were eliminated in this manner. Therefore, the decrease in the double-layer capacitance thought to predominantly come from attached bacteria and bacteria-associated materials which obstructs the double-layer charging by blocking the electrode surface. Fig. 3 provides its evidence in the form of fluorescent images, which are those of PA14 wild type and wspF mutant. Attached bacteria of both strains were observed after 24 h of cultivation (Fig. 3(a), (b)). The comparison of these images with the changes in double-layer capacitance (Fig. 2) confirms that the reduction in double-layer capacitance is closely related to the bacterial adhesion. In the case of wspF mutant, more
bacteria were observed on the platinum surface compared to wild type, which is the main characteristic of this strain (Chung et al., 2008). After the bacterial adhesion, an operation mode was changed to supply fresh nutrients for further development of the adhered bacteria into mature biofilm and resulting biofilms were observed via CLSM (Fig. 3(c), (d)). As can be seen in both strains, mature biofilms were formed upon population growth of attached bacteria. Especially, wspF mutant formed thick biofilm compared to the wild type, which arose from hyper-adherence, EPS overproduction, and impaired motility (Chung et al., 2008). This remarkable change is noteworthy for comparing to the change in double-layer change, in which the plateau appeared since 24 h (Fig. 2). In this stage, where the biofilm maturation initiated, the relationship between the amount of biofilm on the electrode surface and the change in double-layer capacitance was no more maintained, which can be expected from the case of bacteria adhesion stage and from the Eq. (3). Therefore, this difference was examined in detail to make better estimation of biofilm from the monitoring parameter.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 6 1 5 e4 6 2 2
4619
Fig. 2 e Representative impedance spectra (P. aeruginosa PA14 wild type) and percent change in double-layer capacitance. (a) Bode plot shows that impedance in low frequency range (inset figure) increases over time, indicating decrease in doublelayer capacitance. (b) In this ColeeCole plot, the size of semi-circle representing double-layer capacitance decreases over time. (c) Percent change in double-layer capacitance (ΔCdl) during bacterial adhesion and biofilm maturation. EIS data of control (C), P. aeruginosa PA14 wild type (:), and P. aeruginosa PA14 wspF mutant (-) were collected in an electrochemical cell filled with fresh M9 electrolyte (pH 6.8) at predetermined times of 0, 6, 24, 48, and 72 h.
Fig. 4 demonstrates a quantitative comparison between surface coverage and the ΔCdl in Fig. 2. Surface coverage was calculated from the fluorescent images showing the attached bacteria and biofilm on the electrode surface. As shown in Fig. 4, the ΔCdl values of the bacterial adhesion stage (11.1e15.5%, 12.9e16.1%) and the biofilm maturation stage (12.1e14.1%, 14.5e14.9%) were similar or slightly changed in both strains. However, surface coverage of the biofilm maturation stage (77.1e83.7%, 95.8e98.8%) was greatly increased compared to that of the bacterial adhesion stage (7.3e11.3%, 20.5e33.3%). We suspected that two factors may play a key role during in biofilm maturation stage, a gap between bacteria and a surface as well as diffuse layer of the electrical double-layer. Bacterial cell body, which is visualized by staining procedure,
attaches to a surface with a gap of 10e20 nm (Yang et al., 2004). Since the thickness of the diffuse layer in our study (w0.1 M of electrolyte concentration) was less than 1 nm, the bacterial cell body may reside outside the diffuse layer. An electrical double-layer is generated at the electrode/electrolyte interface by electronic charge in an electrode surface and ionic charge in a diffuse layer (Bard and Faulkner, 1980), thus the attached bacterial cell body may not obstruct the movement of ionic charge in electrolyte, which has a minor effects on doublelayer capacitance. Instead, other components accessible to diffuse layer, such as physical appendages of bacteria including flagella, fimbriae, and pili (Garrett et al., 2008), molecular components outside of the bacterial cell body, and various metabolites including EPS, might affect the doublelayer capacitance.
4620
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 6 1 5 e4 6 2 2
Fig. 3 e CLSM images of attached bacteria and biofilm on the platinum surface. P. aeruginosa PA14 wild type (left side) and P. aeruginosa PA14 wspF mutant (right side) were obtained after 24 h (a, b) and 48 h (c, d) of the cultivation. At 24 h of cultivation, attached bacteria are observed, which indicates bacterial adhesion. After 48 h of cultivation, the additional population growth was observed which indicates development of the biofilm.
To determine how these factors contribute to the doublelayer capacitance, tryptone was selected as a model compound representing possible components other than bacterial cell body and its effect on ΔCdl was obtained over time (Fig. 5). Tryptone has a wide molecular weight distribution (<3000 Da); consequently its effect on an electrode surface could be used to describe the interaction between the electrode surface and various factors of the bacteria and biofilm. As shown in Fig. 5, the ΔCdl was sharply reduced upon contact with the tryptone (20 min), since the protein adsorption to a solid surface occurs instantaneously and also is detectable by changes in double-layer capacitance (Wahlgren and Arnebrant, 1991; Moulton et al., 2003). The result implies that various bacterial and biofilm associated components were able to instantaneously affect the double-layer capacitance after
generation and contact with the electrode surface. Usually, the supply of nutrients and subsequent production of various metabolites during bacterial growth are inevitable for developing biofilm, so that they actively participate in the monitoring procedure. During the monitoring, the bacteriaassociated components together with the bacteria would possibly occupy the most adsorption sites of the electrode. The double-layer capacitance became less responsive with the biofilm maturation and even with further maturation (72 h, Fig. 2) in this respect. It properly reveals that factors derived from sources other than bacterial cell body might have affected double-layer capacitance during monitoring. However, it must be mentioned that the model compound may have insufficiently represented all of the suspected factors, so that the effects of these factors must be elucidated in a future study.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 6 1 5 e4 6 2 2
4621
under clean condition, every system has possibility to be contaminated by bacteria and subsequent development of biofilm. Therefore, periodic monitoring by means of reduction in double-layer capacitance on such systems could provide useful information when to react against contamination.
4.
Fig. 4 e Comparison between surface coverage and percent change in double-layer capacitance (LΔCdl). At the bacterial adhesion stage (24 h, left side), the surface coverage of wspF mutant (3rd bar) was higher than that of the wild type (1st bar) due to hyper-adherence. At the biofilm maturation stage (48 h, right side), percent change in both strains (2nd and 4th bar) had hardly changed, although the surface coverage of wild type and wspF mutant (1st and 3rd bar) was increased compared to the bacterial adhesion stage.
Conclusions
An electrochemical approach based on measurement of the double-layer capacitance was performed for monitoring bacterial adhesion and subsequent biofilm maturation. Meanwhile, the effects of bacteria and biofilm attached to a surface were analyzed apart from the change in culture media which can disturb accurate analysis of monitoring parameters. As a result, bacterial adhesion and initiation point of biofilm maturation were successfully monitored by the reduction in double-layer capacitance, which was confirmed by CLSM images. This study reports the possibility of double-layer capacitance as a monitoring parameter indicating bacterial adhesion and biofilm maturation in water-related systems such as medical devices and water purifier. Future studies are required for differentiation of bacteria from other contaminants and an application this method to a pilot system.
Acknowledgments From an application perspective, this method could be employed as a monitoring sensor to health and clinical apparatuses such as a waterline in medical devices and a water purifier. Though these apparatuses are generally maintained
This research was supported by WCU (World Class University) program through the National Research Foundation of Korea funded by the Ministry of Education, Science and Technology (R31-10013) and also supported by Korea Ministry of Environment as “The Eco-technopia 21 project” (no. 102-081-067). For Joon-Hee Lee, this research was supported by Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Education, Science and Technology (grant number; 2010-0006622).
references
Fig. 5 e Percent change in double-layer capacitance (ΔCdl) of tryptone adsorption in 0.1 M phosphate buffer (pH 7.1). ΔCdl was reduced immediately following tryptone adsorption (20 min) and became saturated after a short time (w40 min), showing typical adsorption characteristics.
Bard, A., Faulkner, L., 1980. Electrochemical Methods: Fundamentals and Applications. Wiley New York. Beech, I.B., Sunner, J., 2004. Biocorrosion: towards understanding interactions between biofilms and metals. Current Opinion in Biotechnology 15 (3), 181e186. Brug, G., Van Den Eeden, A., Sluyters-Rehbach, M., Sluyters, J., 1984. The analysis of electrode impedances complicated by the presence of a constant phase element. Journal of Electroanalytical Chemistry 176 (1e2), 275e295. Chung, I., Choi, K., Heo, Y., Cho, Y., 2008. Effect of PEL exopolysaccharide on the wspF mutant phenotypes in Pseudomonas aeruginosa PA14. Journal of Microbiology and Biotechnology 18 (7), 1227e1234. Costerton, J., Stewart, P., Greenberg, E., 1999. Bacterial biofilms: a common cause of persistent infections. Science 284 (5418), 1318e1322. Dexter, S.C., LaFontaine, J., 1998. Effect of natural marine biofilms on galvanic corrosion. Corrosion 54 (11), 851e861. Dheilly, A., Linossier, I., Darchen, A., Hadjiev, D., Corbel, C., Alonso, V., 2008. Monitoring of microbial adhesion and biofilm
4622
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 6 1 5 e4 6 2 2
growth using electrochemical impedancemetry. Applied Microbiology and Biotechnology 79 (1), 157e164. Felice, C., Madrid, R., Olivera, J., Rotger, V., Valentinuzzi, M., 1999. Impedance microbiology: quantification of bacterial content in milk by means of capacitance growth curves1. Journal of Microbiological Methods 35 (1), 37e42. Flemming, H., 2003. Role and levels of real-time monitoring for successful anti-fouling strategies e an overview. Water Science & Technology 47 (5), 1e8. Flint, S., Brooks, J., 2001. Rapid detection of Bacillus stearothermophilus using impedance-splitting. Journal of Microbiological Methods 44 (3), 205e208. Franklin, M., Nivens, D., Guckert, J., White, D., 1991. Effect of electrochemical impedance spectroscopy on microbial biofilm cell numbers, viability, and activity. Corrosion 47 (7), 519e522. Garrett, T., Bhakoo, M., Zhang, Z., 2008. Bacterial adhesion and biofilms on surfaces. Progress in Natural Science 18 (9), 1049e1056. Klahre, J., Flemming, H., 2000. Monitoring of biofouling in papermill process waters. Water Research 34 (14), 3657e3665. LeChevallier, M., Babcock, T., Lee, R., 1987. Examination and characterization of distribution system biofilms. Applied and Environmental Microbiology 53 (12), 2714e2724. Mah, T., O’Toole, G., 2001. Mechanisms of biofilm resistance to antimicrobial agents. Trends in Microbiology 9 (1), 34e39. Matsunaga, T., Nakajima, T., 1985. Electrochemical classification of gram-negative and gram-positive bacteria. Applied and Environmental Microbiology 50 (2), 238e242.
Mattila-Sandholm, T., Wirtanen, G., 1992. Biofilm formation in the industry: a review. Food Reviews International 8 (4), 573e603. Moulton, S., Barisci, J., Bath, A., Stella, R., Wallace, G., 2003. Investigation of protein adsorption and electrochemical behavior at a gold electrode. Journal of Colloid and Interface Science 261 (2), 312e319. Munoz-Berbel, X., Garcia-Aljaro, C., Munoz, F., 2008. Impedimetric approach for monitoring the formation of biofilms on metallic surfaces and the subsequent application to the detection of bacteriophages. Electrochimica Acta 53 (19), 5739e5744. Schmid, T., Panne, U., Adams, J., Niessner, R., 2004. Investigation of biocide efficacy by photoacoustic biofilm monitoring. Water Research 38 (5), 1189e1196. Tamachkiarow, L., Flemming, H., 2003. On-line monitoring of biofilm formation in a brewery water pipeline system with a fibre optical device. Water Science and Technology: a Journal of the International Association on Water Pollution Research 47 (5), 19e24. Wahlgren, M., Arnebrant, T., 1991. Protein adsorption to solid surfaces. Trends in Biotechnology 9 (1), 201e208. Yang, L., Ruan, C., Li, Y., 2003. Detection of viable Salmonella typhimurium by impedance measurement of electrode capacitance and medium resistance. Biosensors and Bioelectronics 19 (5), 495e502. Yang, L., Li, Y., Griffis, C., Johnson, M., 2004. Interdigitated microelectrode (IME) impedance sensor for the detection of viable Salmonella typhimurium. Biosensors and Bioelectronics 19 (10), 1139e1147.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 6 2 3 e4 6 3 3
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Persistence of microbial and chemical pig manure markers as compared to faecal indicator bacteria survival in freshwater and seawater microcosms O. Solecki a,b, L. Jeanneau c, E. Jarde´ c, M. Gourmelon d, C. Marin d, A.M. Pourcher a,b,* a
Cemagref, 17, avenue de Cucille´, 35044 Rennes cedex, France Universite´ Europe´enne de Bretagne, France c CNRS UMR 6118 Ge´osciences Rennes, France d Ifremer, EMP, Laboratoire de Microbiologie, Plouzane´, France b
article info
abstract
Article history:
Natural seawater and freshwater microcosms inoculated with pig manure were set up to
Received 8 December 2010
determine the persistence of pig faecal microbial and chemical markers in these two types
Received in revised form
of surface water. The concentrations of Lactobacillus amylovorus, the Bacteroidales Pig-2-Bac
1 June 2011
16S rRNA genetic marker, five stanols and the evolution of two ratios of stanols, R1 (cop-
Accepted 13 June 2011
rostanol to the sum of coprostanol and 24-ethylcoprostanol) and R2 (sitostanol to copro-
Available online 21 June 2011
stanol) were analyzed during two months along with the concentration of Faecal Indicator Bacteria (FIB). Pig manure was inoculated to unfiltered water microcosms incubated
C in the dark. The faecal contamination load represented by the
Keywords:
aerobically at 18
Faecal source tracking
concentrations of culturable Escherichia coli and/or enterococci remained for two months in
Decay rates
the freshwater and seawater microcosms water column. These concentrations followed
Faecal stanols
a biphasic decay pattern with a 97% reduction of the initial amount during a first rapid
Biphasic kinetic
phase (<6 days) and a remaining proportion undergoing a slower or null second decline.
Lactobacillus amylovorus
The L. amylovorus marker and five stanols persisted as long as the indicators in both
Bacteroidales marker
treatments. The Pig-2-Bac marker persisted 20 and 27 days in seawater and freshwater, respectively. The ratios R1 and R2 were in the range specific to pig manure until day 6 in both types of water. These results indicate that Pig-2-Bac, L. amylovorus and stanol ratios might be used in combination to complement FIB testing to determine the pig source of fecal pollution. However, stanol ratios are to be used when the time point of the discharge is known. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
Each year, the microbial analyses of the 540 freshwater and seawater bathing areas located in Brittany, North West of France, reveal that these areas might be sporadically contaminated by high faecal loads. Among the three main
sources of faecal contamination (human, pig and cow) which can lead to the downgrading of these areas (Soller et al., 2010), pig manure spreading is considered as a potential factor of water pollution (Thurston-Enriquez et al., 2005). Brittany supports more than half of the national pigs’ livestock while representing only 5% of the French territory and
* Corresponding author. Cemagref, 17, avenue de Cucille´, 35044 Rennes cedex, France. Tel.: þ33 223 48 21 37. E-mail address: [email protected] (A.M. Pourcher). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.06.012
4624
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 6 2 3 e4 6 3 3
this high concentration generates from 8 to 10 million tons of pig manure each year. To protect public health whilst bathing and to improve management practises at bathing waters, the revised Bathing Water European Directive (2006/7/EC) requires to establish bathing water profiles to identify the source(s) of a faecal pollution upstream a targeted water body. To track pollution to livestock facilities or diffuse non point sources, animal-specific markers have been proposed. They include DNA molecules from intestinal bacteria (Dick et al., 2005; Ufnar et al., 2007), endogenous eukaryotic cells (Balleste et al., 2010) and faecal stanols (Tyagi et al., 2009). Among pig genetic markers, studies have focused essentially on bacteria from the Bacteroidales order as potential markers preferred for their abundance in the gastrointestinal tract and host specificity occurrence (Dick et al., 2005; Mieszkin et al., 2009). Among chemical markers, faecal stanols are considered as direct markers because they occur in the faeces (Leeming et al., 1996). The distribution of those compounds in animal faeces depends on three host factors: (i) the animal’s diet, (ii) the ability to biosynthesize endogenous sterols and (iii) the occurrence of anaerobic bacteria able to biohydrogenate sterols to stanols of various isomeric configurations. The combination of those three factors determines the “sterol fingerprint” that is characteristic of each animal faeces (Leeming et al., 1996). In a previous study from our teams, several microbial and chemical markers detected in river waters impacted by faecal pollutions were proposed to be used in a toolbox as Faecal Source Tracking methods (FST) (Gourmelon et al., 2010). Two of which were microbial pig-specific markers previously validated for their host specificity, the genetic Bacteroidales marker Pig-2-Bac and the bacterial species Lactobacillus amylovorus (Marti et al., 2010; Mieszkin et al., 2009). Five faecal stanols found in pig faeces and two steroids ratios were also selected to be part of the toolbox. Specific markers should exhibit high host specificity and represent the load of faecal pollution (Field and Samadpour, 2007). In this scope, the evaluation of markers to be used as FST must consider whether the degradation and transport characteristics of the markers are similar to that of one or several pathogens or to the traditional faecal indicator bacteria (FIB) such as Escherichia coli (E. coli) and enterococci. The present study examines the detection of Pig-2-Bac marker, L. amylovorus and faecal stanols as compared to the detection of FIB in natural surface water microcosm. The faecal stanols selected comprised coprostanol (5b-cholestan-3b-ol) and 24-ethylcoprostanol (24-ethyl-5b-cholestan-3b-ol) dominant in fresh pig manure (Leeming et al., 1996; Shah et al., 2007). We also included epicoprostanol (5b-cholestan-3a-ol) and campestanol (24-methyl-5a-cholestan-3b-ol) since the land spreading of pig manure as a soil fertilizer results in high concentration of those compounds (Jarde et al., 2009). Two ratios of concentrations of steroids were calculated along the experiment: coprostanol to the sum of coprostanol and 24-ethylcoprostanol (R1) and sitostanol to coprostanol (R2). In the previous study to this work, pig manure was characterized by a R1 ratio of 0.57 0.02 and a R2 ratio of 0.3 0.1 (Gourmelon et al., 2010). The objective of this study was to estimate the persistence and decay rates of pig genetic markers, faecal stanols, and FIB
in fresh and marine water during two months. The evaluation to which extent the detection of tested markers and ratios correlates with that of FIB and hence, represents faecal load contamination should allow to validate their usefulness as markers of pig faecal pollution in these two types of surface water.
2.
Materials and methods
2.1.
Microcosms design
Microcosms consisted of 6 one hundred litres inert glass aquariums placed in a dark room, protected from sunlight and fluctuating temperature. Three of which were filled with seawater and the remaining aquariums with freshwater. Both waters were not filtered to study the persistence of FIB and specific markers in presence of protozoa. Waters were seeded with pig liquid manure. Constant mixing was achieved with the aid of a helix agitator and oxygen saturation with air pumped in throughout the experiment. A plastic film cover limited evaporation of water. The type of surface water was the changing parameter. Ambient temperature (around 18 C) corresponded with surface water temperature during the warmer months in Brittany (France). Nine hundred mL of untreated liquid pig manure was added to 90 L of water (1:100 dilution). This ratio was chosen to represent a high faecal load contamination likely to remain for two months. However, the turbidity resulting from this parameter hindered light treatment. Sampling took place on the starting day, then on day 2, 6, 13, 20, 27, 34, 41, 48 and 55. Both unseeded initial types of water were kept in the same conditions during the whole experiment to use as controls, sampling took place on the starting day and on day 55. Culturable E. coli (cEC), culturable enterococci (cENT), bacterial genetic markers and stanol concentrations were measured at each sampling point. Dissolved O2 concentration and temperature were measured every 3e4 days. Although constant mixing was achieved in the middle of the aquarium, sedimentation occurred during the course of the experiment and biofilms formation occurred on the walls. Samples were drawn from the water column.
2.2.
Water and pig manure samples
The seawater was collected in the end of January 2010 in the Atlantic Ocean from Landunvez, in the NW Brittany region of France (lat long: 48.540819 4.751587). Salinity was 33 g/L, total dissolved organic carbon measurement was 0.4 mg C/L, total dissolved nitrogen was 4.0 mg N/L. The freshwater was sampled from a lake in Commana in Brittany (lat long: 48.3887488 4.0177564) on the same day, the total dissolved organic carbon was 2.7 mg C/L, the total dissolved nitrogen was 9.3 mg N/L. Pig manure was collected from a farm located in Brittany and samples were taken from a storage tank after homogenisation with propeller agitator for 20 min, the total dissolved carbon and total dissolved nitrogen were 6.5 g C/L, and 2.9 g N/L, respectively.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 6 2 3 e4 6 3 3
2.3.
Enumeration of FIB
Depending on sample turbidity, FIB counts were achieved either by serial dilution in buffered peptoned water (Oxoid, Basingstoke, England) or by filtration of 100 mL of sample on a 0.45 mm cellulose membrane (Whatman, Dassel, Germany). Filters or 0.1 mL of the dilution was plated on TBX agar (Oxoid) and on Slanetz and Bartley agar (Biokar Diagnostics, Beauvais, France). TBX plates were incubated for 24 h at 44 C. Blue colonies (glucuronidase positive) were counted to determine the concentration of E. coli. After incubation at 37 C for 48 h, membranes on Slanetz and Bartley agar were transferred onto Bile-Esculin-Azide agar (BEA) (Biokar Diagnostics) and incubated for 2 h at 44 C. Black colonies on BEA were counted as enterococci. The detection limit of both methods was 1 colony forming units (CFU) per 100 mL.
2.4.
Microbial markers analyses
2.4.1.
Samples preparation and DNA extraction
Two hundreds mL of samples were either centrifuged (9000 g for 15 min) or filtered onto 0.2 mm polycarbonate membrane (Sartorius, Goettingen, Germany) depending on suspended matter density. Filtration was the preferred method to recover DNA. However, filtration was not possible until day 13. On that day, samples from one microcosm of both types of water were treated by both methods. Since qPCR results were similar, it was decided to carry on with filtration. From 0.30 to 250 mg of solid matter were recovered. DNA extraction was performed on sample solid matter with the aid of the FastDNA SPIN for Soil kit (MP Biomedicals, Illkirch, France), following manufacturer’s instructions. The elution volume was 100 mL.
2.4.2.
Real-time PCR
2.4.2.1. Oligonucleotide primers and probes. The pig-specific Bacteroidales 16S rRNA gene marker (Pig-2-Bac) and the L. amylovorus marker were quantified with the primers and probe described by Mieszkin et al. (2009) and Konstantinov et al. (2005), respectively.
2.4.2.2. DNA standard curves. For the quantification of the Bacteroidales marker Pig-2-Bac, standard curves were generated from serial dilutions of known concentration of plasmid DNA ranging from 5 107 to 5 100 copies per reaction. Linear plasmids were extracted with the QIAquick Miniprep Extraction Kit (Qiagen), following the manufacturer’s instructions. The linear form of plasmid was obtained with NotI enzyme (Roche Diagnostics, Meylan, France) in a final volume of 50 mL for 3 h at 37 C. The PCR standard curve for the L. amylovorus markers was prepared by 10-fold dilution of bacterial genomic DNA extracted from one mL of a pure culture of L. amylovorus DSM16698, with the Wizard genomic DNA purification Kit (Promega, Madison, USA) according to the manufacturer’s instructions. Dilutions ranged from 4.5 106 to 4.5 100 CFU equivalent from direct plating count considering that 100% of the DNA from the culture was recovered. Standard curves were generated by plotting threshold cycles (Ct) against 16S rRNA genes or CFU equivalent, depending on the marker.
4625
Standard curves were obtained by means of 3 replicates per point.
2.4.2.3. Real-time PCR assays. For Pig-2-Bac marker, amplification was performed using the Chromo4 real-time detection system associated with Bio-Rad Opticon Manager software version 3.1 (Bio-Rad, Hercules, CA). Real-time PCR was performed using the TaqMan Brilliant II QPCR Master Mix kit (Agilent technologies, Massy, France). Each reaction was run in triplicate. The cycle conditions were 1 cycle at 95 C for 10 min, followed by 40 cycles at 95 C for 15 s and 60 C for 1 min. Reactions were carried out in a final volume of 25 mL with primers and probe final concentration being 300 nM and 200 nM respectively. Quantification limit was 1250 16S rRNA gene copies per 100 mL. The presence/absence of PCR inhibitors was verified using an Internal Positive Control (IPC; Applied Biosystem, France). Samples were diluted if inhibitors were present. Concerning the L. amylovorus marker, PCR was performed on the CFX96 real-time system (Bio-Rad), with the software Opticon Monitor version 3.1.32 and CFX manager version 1.1 (Bio-Rad), using the IQ SYBR-Green Supermix (Bio-Rad). The cycle conditions were 1 cycle at 95 C for 3 min, followed by 40 cycles at 95 C for 15 s and 60 C for 45 s. Reactions were carried out in a final volume of 25 mL with primers final concentration being 200 nM. Quantification limit was 112.5 CFU equivalents per 100 mL. No filtration, extraction and template positive controls from a known concentration of a control DNA fragment were included in the method to evaluate the yield of DNA recovery.
2.5.
Faecal stanols analysis
Five faecal stanols, namely coprostanol, epicoprostanol, 24ethylcoprostanol, campestanol and sitostanol were investigated in this study. Analyses were performed on 1 L of non seeded initial waters and 500 mL of microcosm waters. Faecal stanols were extracted from the dissolved phase (<0.7 mm) by solid phase extraction and quantified by gas chromatography e mass spectrometry (GCeMS) according to the protocol developed by Jeanneau et al. (2011). Cholesterol d6 (2,2,3,4,4,6-2H6-5-cholesten-3b-ol) was used as a recovery standard and was added to water samples prior to the extraction step. 5a-cholestane was used as an internal standard and was added prior to the GCeMS analysis. Faecal stanols were quantified by the internal standard method using a five-point calibration curve. The resulting calibration ranged from 0.01 to 40 mg/L for stanols in water.
2.6.
Decay rate calculations
The decay rates of faecal stanols were calculated based on a first order decay model (Chick model): CðtÞ ¼ C0 ekt or ln CðtÞ =C0 ¼ kt Where C0 is the average initial concentration of the target in mg/L, C(t) is the target average concentration at time t in mg/L, t is the time in days and k the decay constant or rate in days1.
4626
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 6 2 3 e4 6 3 3
The model describes a linear regression, k is the slope of the regression line and R2 the regression coefficient. The biphasic model described by Lee et al. (2001) was used to calculate decay rates for every microbial target in both treatments. Two constants were calculated from a biphasic first order decay model (Cerf model): CðtÞ ¼ C0 f ek1 t þ ð1 f Þ ek2 t ¼ ln f ek1 t þ ð1 f Þ ek2 t
or ln CðtÞ =C0
Where f is the proportion of C0 that declined during the first phase, k1 is the decay constant of the first phase and k2 the decay constant of the second phase. C(t) and C0 are expressed in CFU/100 mL for FIB, in CFU equivalent/100 mL for L. amylovorus and in DNA copies/100 mL for Pig-2-Bac. The biphasic model and associated parameters were obtained with the aid of XLSTAT 2010.4 using the non linear regression modelling. Decay rates were calculated until the day concentrations were below quantification limit or until day 55 when the detection limit was not reached. The length of time (expressed in days) needed to obtain a 90% reduction in initial concentration of stanols or bacteria was calculated as follows: T90 ¼ ln(0.1)/k (Chick model) or, if f was >90%, T90 ¼ ln(0.1)/k1 (Cerf Model).
2.7.
Regression tests
In order to compare occurrence and concentration of FIB and stanols with time, regressions were performed by plotting average concentrations of each stanol against average concentrations of FIB. Regression lines were drawn between four or several time points. The regression coefficient R2 values illustrate the relationship between the two variables compared.
2.8.
Statistical analysis
2.8.1.
Decay rates distribution
The validity of parametric tests is limited to samples following a normal distribution. When the distribution is unknown non parametric tests should be preferred. To infer on normality, a large number of samples is required, therefore to check whether the decay rates obtained in this study were normally distributed, we performed an additional experiment where 20 microcosms were investigated. These latter consisted of 2Lpolypropylene bottles placed in the same conditions as in the study, filled with the same freshwater and the same pig manure added to 1:100. For practical and economical reasons, only, culturable E. coli concentrations were measured. The concentrations of cEC were followed every week for 6 weeks. Decay constants k1 and k2 were found to be normally distributed with the Normality tests from XLSTAT 2010.4 (n ¼ 20) (data not shown). As a result parametric tests were performed to compare decay constants of FIB and microbial markers. However, regression coefficients R2 were found not to follow a normal distribution, hence all R2 were compared with the Mann and Whitney test.
2.8.2.
Tests of significance
Two null hypotheses were posed (i) no difference exists between decay rates of one target in either treatments, (ii) no
difference exists between decay rates of either indicator (cEC or cENT) and marker in a particular treatment (seawater or freshwater). The risk a to reject the null hypothesis while it might be true was set at 0.05. All statistical analyses were calculated from the regression coefficients (R2) and decay rates (k) of three independent experimental replicates. Concerning microbial markers, an Ftest was performed to determine variance equality between set of samples prior to a two tailed Student’s t-test assuming equal or unequal variance depending on the F-test results. Analyses were achieved by Microsoft Office Excel 2003. It was not possible to compare decay rates from FIB and faecal stanols because the former decay was described with a non linear regression model whilst the latter with a linear model. Only the first null hypothesis could be considered. The distribution of stanols decay rates could not be determined. As a consequence, to compare decay rates of stanols in freshwater and seawater microcosms. A non parametric test for small samples of unknown distribution (ManneWhitney) was performed on XLSTAT 2010.4. Decay rates were compared only if the R2 of the linear regressions they originated from were not statistically different. Otherwise, the test is stated to be non applicable (NA).
3.
Results
In this study, we examined the persistence of pig faecal markers in three independent controlled unfiltered water microcosms under aerobic condition. Temperature oscillated between 16 and 20 C and dissolved O2 ranged from 8.3 to 9.7 mg/L. The waters used to constitute microcosms were free or slightly contaminated with FIB and stanols. In the seawater, FIB were not detected. Among stanols, only coprostanol and sitostanol were quantifiable with a concentration of 0.02 and 0.05 mg/L, respectively. In the freshwater, cEC concentration was 60 CFU/100 mL and cENT were not detected. The concentrations of coprostanol, epicoprostanol, 24-ethylcoprostanol and sitostanol were 0.04, 0.02, 0.02 and 0.07 mg/L, respectively whereas campestanol was not detected. Pig-specific DNA markers Pig-2-Bac and L. amylovorus were not detected in both types of water. At the end of the experiment FIB were not detected in both types of water controls in 100 mL. We thus inferred that microflora and stanols present in the microcosms at day 0 arose from the pig manure.
3.1.
Decay curves
3.1.1.
FIB and microbial markers
FIB and microbial markers followed a biphasic first order decay kinetic (Fig. 1). We, therefore, used the Cerf biphasic decay model to determine the decay rates of the first and second phase. An example of this model, illustrated by the behaviour of cENT in seawater is shown on Fig. 2A. The model shows a sharper slope for the regression of the first phase than for the second phase. Hence k1 is always higher than k2 in this model (Table 1). All decay curves fitted to the model (R2 > 0.90). As indicated by the value of the f parameter e proportion of the initial concentration of the target that declined during the first phase-, more than 97% of the initial inoculum was lost during
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 6 2 3 e4 6 3 3
4627
Fig. 1 e Decay curves of pig-specific microbial markers in water microcosms seeded with liquid pig manure (1:100 dilution), (A) seawater, (B) freshwater. Conditions: dark, O2 saturation, around 18 C. The uncertainties represent standard errors. Limit of quantification was 1 CFU/100 mL for FIB, 1250 gene copies/100 mL for Pig-2-Bac and 112 CFU equivalent/100 mL for L. amylovorus.
this phase. The decimal reduction time occurred thus during the first phase soon after the beginning of the experiment and was reached in less than 6 days independently of the bacteria or the microcosm conditions. However, in many instances a small remaining proportion persisted until the end of the experiment. Culturable E. coli were recovered until day 20 in seawater whilst it was still detected on day 55 in freshwater microcosms. Culturable enterococci and L. amylovorus concentrations were over the limit of quantification in both types of water on the last day of the experiment. In seawater, Pig-2-Bac marker followed the same trend as E. coli, it reached the quantification limit by day 13. On day 20, it was detected just below quantification limit in the three microcosms. It was decided to take this point into account for calculations (Fig. 1A). The week after Pig-2-Bac was still detected in two microcosms. In freshwater, it was found over quantification limit until day 27 in the three microcosms (Fig. 1B), but remained detectable until day 34 in two microcosms.
3.1.2.
Faecal stanols
Faecal stanols followed a monophasic first order decay kinetic as illustrated by Fig. 2B. In seawater microcosms, initial concentrations of coprostanol, epicoprostanol, 24ethylcoprostanol, campestanol and sitostanol were 16.7, 3.1, 12.1, 2.6 and 4.7 mg/L, respectively. Their final concentrations reached 0.15, 0.05, 0.18, 0.03 and 0.10 mg/L, respectively (Fig. 3A). The concentrations observed after 55 days represented a degradation of 98e99% of the initial amounts of the five stanols. Nevertheless, they remained higher than the initial concentrations in seawater before the addition of pig manure. In freshwater microcosms, initial concentrations of coprostanol, epicoprostanol, 24-ethylcoprostanol, campestanol and sitostanol were higher than in seawater microcosms and were 33.4, 9.3, 25.5, 6.9 and 11.8 mg/L, respectively whereas their final concentrations were 0.49, 0.15, 0.53, 0.10 and 0.20 mg/L, respectively. At the end of the experiment, they were in average 13 6 times higher than the initial
Fig. 2 e Modelisation of the mean concentration of cENT (A) by Cerf biphasic decay model (R2 [ 0.97) and of 24ethylcoprostanol (B) by Chick first order decay model (R2 [ 0.98) from the seawater microcosms.
4628
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 6 2 3 e4 6 3 3
Table 1 e Decay rates and T90 of FIB and pig-specific microbial and chemical markers in water microcosms seeded with liquid pig manure (1:100 dilution). p-values represent comparison between decay rates in seawater and freshwater. Conditions: dark, O2 saturation, around 18 C. Marker
Seawater 1
cE. coli cENT Pig-2-Bac L. amylovorus Coprostanol Ethylcoprostanol Epicoprostanol Campestanol Sitostanol
1
Freshwater
k1 (d )
k2 (d )
f (%)
T90 (d)
R
k1 (d )
k2 (d1)
f (%)
T90 (d)
R2
p-value k1
p-value k2
1.291 0.605 1.100 0.406 0.092 0.086 0.078 0.078 0.072
0.388 0.016 0.075 0.031
99.7 99.9 99.9 99.9
1.8 3.8 2.1 5.7 23.7 26.2 25.0 25.6 32.0
0.96 0.97 0.99 0.97 0.98 0.98 0.91 0.88 0.82
0.428 1.015 1.247 0.737 0.082 0.075 0.075 0.083 0.079
0.021 0.048 0.001 0.014
97.8 99.7 99.9 99.4
5.4 2.3 1.9 3.1 29.5 32.0 29.1 25.6 29.1
0.94 0.94 0.95 0.94 0.92 0.93 0.98 0.93 0.91
0.0006 0.0377 0.7488 0.0295 0.1000 0.2000 0.3000 1.0000 0.3000
0.0004 0.6714 0.0412 0.3089
2
concentrations in freshwater before the addition of pig manure. Furthermore, as observed in the seawater microcosms, they represented a degradation of 98e99% of the initial amount of the five stanols (Fig. 3B).
3.2.
1
between coprostanol and 24-ethylcoprostanol (Table 1). The changes in the values of R2 were also similar in freshwater and in seawater. They were most marked than those of R1 ratio as R2 ranged mainly from 0.20 to 0.91 with however a maximum value of 1.55 on day 20 (Fig. 4B).
Progression of stanol ratios 3.3.
At the beginning of the experiment, the initial values of R1 (coprostanol/(coprostanol þ 24-ethylcoprostanol)) and R2 (sitostanol/coprostanol) ratios ranged between 0.57 and 0.58 and between 0.20 and 0.23, respectively. Regardless the microcosms, the ratio R1 progressively decreased to reach a value of 0.47 (Fig. 4A), due to the difference of decay rate
Decay rates statistical analysis
3.3.1. Decay rates comparisons from seawater and freshwater microcosms The regression coefficients from non linear regressions were not statistically different. As a consequence every k1 and k2 drawn from microbial markers kinetic models could be
Fig. 3 e Decay curves of stanols in seawater (A) and freshwater (B) microcosms. The uncertainties represent standard errors. The scale of Y-axes are different.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 6 2 3 e4 6 3 3
4629
Fig. 4 e Evolution of stanols ratios R1 (A; coprostanol/coprostanolD24-ethylcoprostanol) and R2 (B; sitostanol/coprostanol). Grey areas correspond to the range of values characteristic of pig manure. SWP: seawater, FWP: freshwater.
compared. The results concerning the first null hypothesis are shown in Table 1. L. amylovorus decay first constants (k1) were statistically different whilst their seconds (k2) were not. The opposite was noted for Pig-2-Bac, its first decay constant was not different in seawater or freshwater microcosms whilst they were significantly different during the second phase. For stanols, the regression coefficients from linear regressions were also not statistically different. Thus, every k drawn from faecal stanols kinetics could be compared. According to the Mann and Whitney non parametric test, stanols decay rates were not significantly different in freshwater or seawater microcosms (Table 1).
3.3.2. Decay rates comparisons from FIB and microbial markers In seawater, the first phase constant from Pig-2-Bac decay did not significantly differed from that of FIB. Contrary to this pigspecific marker, the decrease of L. amylovorus was slower than the ones of cEC ( p < 0.0001) and of cENT ( p ¼ 0.027) during the first phase. During the second phase, the decays of both genetic markers were not significantly different from that of cENT whilst they were lower than that of cEC ( p < 0.001). In freshwater, Pig-2-Bac and L. amylovorus decays did not significantly differed from that of cENT in the first phase while they were faster than cEC ( p ¼ 0.001). During the second phase, decay rates of both markers were not statistically different from those of FIB.
3.4.
Regression tests
In seawater microcosms, the changes in concentrations of cEC from day 0 to day 13 were well correlated with those of coprostanol (R2 ¼ 0.92) and 24-ethylcoprostanol (R2 ¼ 0.91). However the decrease in concentrations of cEC was not correlated to concentrations of epicoprostanol, campestanol and sitostanol (R2 < 0.40) due to their increases between day 6 and day 13. The correlation between the decrease of concentrations of cENT and stanols showed a same trend with R2 > 0.95 for coprostanol and 24-ethylcoprostanol and R2 < 0.60 for epicoprostanol, campestanol and sitostanol. In freshwater microcosms, the five stanols were better correlated to the change in concentrations of cEC (R2 > 0.85) than to
cENT with R2 ranging from 0.73 (sitostanol) to 0.78 (epicoprostanol).
4.
Discussion
The objective of this research was to evaluate the persistence of bacterial and chemical markers as compared to the survival of indicator organisms that are measured currently to assess water microbial quality. It is expected that a microbial load added to a water body by a faecal pollution for instance would decline with time due to the effects of several parameters including sunlight, sedimentation, dilution, transport or grazing by biological agents (Barcina et al., 1997; Easton et al., 1999). Although here, conditions of natural water bodies were not fulfilled, since the experiment was performed in a closed environment, a decline was observed for every target: molecular and living organisms. This observation is consistent with other recent microcosm studies in freshwater and seawater (Dick et al., 2010; Walters et al., 2009).
4.1.
Decay curves
The kinetics of the FIB and microbial markers followed a biphasic curve (Fig. 1). The first phase occurred within 6 days on average. During this phase, a high proportion of the starting inoculum decayed. The remaining proportion persisted in the water column and seemed to be more resistant to decline, since the second phase decline was slow or even null. This was reported before by Easton et al. (1999) who showed, using in-situ chambers that faecal microorganisms did not die-off at a constant rate, and this was only true for the initial decline. Their experiment demonstrated that the die-off rate slowed down as the organism level approached equilibrium with the environment. They found that the initial rapid die-off occurred, generally during the first seven days of the experiment which is consistent with our findings. They proposed two hypotheses to explain this observation (1) the microorganism die-off at a rapid rate until the carrying capacity of the environment is reached, (2) microorganisms would use quorum sensing to regulate their numbers and adapt to
4630
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 6 2 3 e4 6 3 3
their new environment. Although here other parameters than the genetic programming of organisms might have triggered the decline such as sedimentation (Hartz et al., 2008), grazing by protozoa (Bell et al., 2009) or loss of culturability (Barcina et al., 1997), it seems that bacteria can regulate their numbers in a microcosm. Thus, Hellweger et al. (2009) established the biphasic decay kinetic of a pure strain of culturable E. coli inoculated in sterile phosphate buffered saline. They observed a decline of the initial inoculum during two days followed by a slight increase of E. coli densities. Their experiment established that the resistant fraction was not a population or strain dependent parameter since they used a pure strain. They proposed as possible explanation that this resistant fraction was made of mutants growing on nutrients released by dead cells. In agreement with our results, Bae and Wuertz (2009) demonstrated the biphasic persistence of Bacteroidales gene markers and Enterococcus 23S rRNA gene from human, cattle and dog faecal samples in seawater. Furthermore, Dick et al. (2010) also observed a decay with a biphasic pattern for cEC and Bacteroidales genetic markers from human waste water in freshwater. They also noted that 99% of the initial inoculum was inactivated during the first phase regardless the microcosm conditions. It is noteworthy that in the mentioned studies, although the type of water, the polluting matrix and the physical conditions differed, the biphasic pattern of every different microbial target was observed. This is consistent with our results as each bacterial target in both treatments followed a biphasic decay trend. Furthermore, it has been reported that a high level of prey would be reduced by predators to an equilibrium density that would ensure the survival of the predators (Marino and Gannon, 1991; Menon et al., 2003). It appears that numerous factors are involved in the biphasic decay and the mechanisms responsible for this trend. Results from our study would probably depend on both predation, as the waters were not filtered, and on intrinsic characteristic of the studied bacteria. The persistence pattern of the DNA markers tended to follow the survival pattern of the living organisms. We could then speculate that the DNA we quantified arose from living cells. This is somehow illustrated by the positive second decay rate from the L. amylovorus marker (Table 1). This very low rate could be due to sampling or measurement variations or to a multiplication of the marker. However, it was not possible to verify this hypothesis because no medium enables the isolation of L. amylovorus from a complex matrix. Stanol decay results were concordant with the evolution of coprostanol in seawater during microcosm experiment performed in darkness at 19 C (Thoumelin et al., 1990). The increase of the concentration of sitostanol, campestanol and epicoprostanol between day 6 and day 13 could be due to the death of living organisms inherited from the pig manure. As a consequence the sterols that constitute those organisms were liberated in the dissolved phase (Marty et al., 1996) and further hydrogenated into sitostanol, campestanol and epicoprostanol (Pratt et al., 2008). This explanation would agree with the observed microbial decay.
4.2.
FIB survival and markers persistence
In seawater, Pig-2-Bac presented a rapid decay rate close to that of cEC (1.1 and 1.3 d1, respectively) during the first phase, twice faster than that of cENT. It has been described that protozoa eliminate Gram positive bacteria at lower rates than Gram negative bacteria (Barcina et al., 1997; Davies et al., 1995). Furthermore, in a recent study Balleste and Blanch (2010) proved that Bacteroides fragilis survival was highly hindered by grazing predators in warm conditions in a river. Additionally, Jin et al. (2005) established in a natural slightly salted water storm event experiment that the percentage of E. coli attached to suspended particles was 21.8e30.4% compared to 8.3e11.5% for enterococci. However, in salty water, Gram positive bacteria would try to protect themselves from the osmotic pressure by attaching to organic matter (Hartz et al., 2008). In this experiment, turbidity was not measured, but it was clear since we could filtered the water from day 13, that suspended organic matter had settled and thus the water column was poorer in organic matter from that day. As a consequence, the concentration of organic matter might have accounted for in the survival of E. coli and Pig-2-Bac marker in seawater. Predation and sedimentation might explain the more rapid decay of Gram negative compared to Gram positive markers in seawater. Another factor that might explain the sensibility of the Bacteroidales marker compared to the three other organisms in freshwater, is their sensibility to oxygen. It is well established that Bacteroidales cells are negatively affected by increased dissolved oxygen in water (Bae and Wuertz, 2009; Balleste and Blanch, 2010). In freshwater, the Bacteroidales marker was the only target not detected until the end of the experiment although, Pig-2-Bac decay constant was not different from cENT decay constants (Table 1), thus the shorter relative persistence time period might also be explained by the higher detection limit of the method. This could be improved by filtering higher volumes of water especially, this would be possible in natural environment normally less concentrated in suspended organic matter, or improving DNA extraction yield. No microcosm studies have yet reported on the persistence of L. amylovorus. It is thus interesting to note that the pigspecific L. amylovorus marker followed the same trend as cENT until day 55 in seawater and in freshwater, although the detection limit was elevated (112 CFU equivalent/100 mL) compared to that of FIB (1 CFU/100 mL). However, the marker follows cEC evolution only in freshwater as it was inactivated more rapidly in seawater. This is not surprising as it has been demonstrated that unlike Enterococcus faecalis or Lactobacillus casei, E. coli does not harbour resistance mechanisms to high osmotic pressure (Lee et al., 1977). However, as stated before, attachment to organic matter aids the bacteria to resist to this pressure. The progressions of the amount of FIB and of the concentration of stanols followed the same trends except for cEC in seawater. Coprostanol and ethylcoprostanol were correlated with cEC until it was no more detected in seawater. Regression coefficients were higher between the five stanols and cENT than with cEC in seawater. On the other hand in freshwater microcosms the five stanols concentrations were more closely related to the amount of
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 6 2 3 e4 6 3 3
cEC than of cENT. However the regression coefficients were still high i.e. R2 > 0.70 and defined a close relationship between the occurrence and concentrations of stanols and FIB.
4.3.
Effect of the type of water
To determine whether the type of water influenced the persistence of markers in the conditions of the experiment, we compared decay rates observed from both microcosms for each target. Pig-2-Bac was not significantly influenced by the type of water during the first phase maybe due to the presence of suspended particles but salinity or other factors from the seawater accelerated the decay during the second phase (Table 1) when the particles had settled. As a consequence, the marker was detected one more week in freshwater microcosms. These results are somehow different from those from Okabe and Shimazu (2007) who exposed their Bacteroidales pig marker Pig-Bac2 to different salinities and observed no real difference in the persistence or decay of the marker to 0, 10, 20 and 30 g/L under dark conditions; however their experiments were performed at 10 C which decreases the decay. The lactobacillus marker was affected by the type of water at the beginning of the experiment, but the remaining proportion of the population was not affected and persisted for the same length of time in freshwater and seawater microcosms. It has been shown that L. amylovorus can multiply in NaCl concentration of 5e10 g/L and can survive in concentrations from 20 to 40 g/L (Neysens et al., 2003). This new genetic marker, which belongs to a genus phylogenetically close to the Enterococcus genus, might be useful to monitor marine waters. The concentrations of stanols after the addition of pig manure were twice to three times higher in the dissolved phase (<0.7 mm) of freshwater than of seawater. This observation was probably due to the increased salinity in seawater that induced an aggregation of dissolved macromolecules instead of allowing them to suspend in the dissolved phase as colloids. In spite of slight differences observed on decay curves (Fig. 3), the decay rates calculated for stanols in seawater and freshwater microcosms did not exhibit significant differences (Table 1).
4.4.
MST toolbox validation
Pig-2-Bac and L. amylovorus markers were selected to be part of a toolbox to identify sources of faecal pollution in water in our previous study (Gourmelon et al., 2010). We wanted to estimate how their detection is representative of a faecal load. In seawater and freshwater, in the conditions of the study, according to the concentrations in culturable enterococci, the faecal contamination was present until the end of the experiment. L. amylovorus followed globally the same trend as cENT and its detection was thus representative of the faecal load during two months. Pig-2-Bac followed also the same trend as cENT but only until day 20 and day 27 in seawater and freshwater, respectively. Thus it was no longer detected whilst the faecal contamination was still present. However, although Pig-2-Bac appeared less persistent than L. amylovorus, both markers were detected for a long period of time (at least 20
4631
days) that would allow water managers to take necessary actions in cases of important discharges. The development of the MST toolbox in our previous study has highlighted two ratios allowing the discrimination between human, pig and cow faeces. Thus, coprostanol to the sum of coprostanol and 24-ethylcoprostanol (R1) allows the differentiation between human (>0.71), porcine (0.55e0.59) and herbivore (<0.41) contributions and the second ratio, sitostanol to coprostanol (R2) exhibits values > 1 for bovine manures and <0.4 for pig manures and waste water treatment plant effluents (Gourmelon et al., 2010). In seawater and freshwater microcosms, between day 0 and day 13, R1 exhibited values characteristic of pig manure. However, over the course of the experiment, R1 remained above the specific values of bovine manure, suggesting that slight changes in R1 values over time do not lead to misinterpretation of the origin of the contamination. In both microcosms, R2 was characteristic of pig manure from day 0 to day 6. As a consequence, the combination of stanol ratios R1 and R2 can be investigated in order to indicate a faecal contamination from pig manure up to six days from the beginning of the discharge, which allows time to enumerate FIB in order to determine if further analyses are necessary.
5.
Conclusion
This laboratory microcosm study aimed to compare the decay rates of FIB, Bacteroidales Pig-2-Bac and L. amylovorus pig genetic markers and stanol ratios in both freshwater and seawater inoculated with pig manure. Regardless the microcosm conditions and the target, the persistence or survival profiles of the two genetic markers and of the FIB followed a biphasic curve whereas the five stanols followed a monophasic first order decay kinetic. The persistence of genetic and chemical markers were similar in freshwater and in seawater. According to the values of the ratio R1 and R2, and as the persistence of L. amylovorus and Pig-2-Bac Bacteroidales marker was relatively close to that of cultivable E. coli and enterococci, all these markers can be used to complement E. coli and enterococci detection method to identify a source of pig pollution at least during 6 days when the time point of the contaminating discharge is known. Furthermore, they should prove useful to trace a pig faecal pollution from bathing areas and shellfish farming waters that are sporadically classified as “non satisfactory” in terms of microbial quality.
Acknowledgement This work was done in a French project entitled Marquopoleau, recognised by the economic competitiveness cluster Poˆle Mer Bretagne and supported by the Fond Unique Interministeriel, Regional council of Brittany, County council of Morbihan, County council of Finiste`re and Brest Me´tropole Oce´ane. The authors wish to thank Emmanuelle Guillerm from IDHESA Bretagne Oce´ane for her technical assistance.
4632
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 6 2 3 e4 6 3 3
references
Bae, S., Wuertz, S., 2009. Rapid decay of host-specific faecal bacteroidales cells in seawater as measured by quantitative PCR with propidium monoazide. Water Research 43, 4850e4859. Balleste, E., Blanch, A.R., 2010. Persistence of bacteroides species populations in a river as measured by molecular and culture techniques. Applied and Environmental Microbiology 76, 7608e7616. Balleste, E., Bonjoch, X., Belanche, L.A., Blanch, A.R., 2010. Molecular indicators used in the development of predictive models for microbial source tracking. Applied and Environmental Microbiology 76, 1789e1795. Barcina, I., Lebaron, P., VivesRego, J., 1997. Survival of allochthonous bacteria in aquatic systems: a biological approach. FEMS Microbiology Ecology 23, 1e9. Bell, A., Layton, A.C., Mckay, L., Williams, D., Gentry, R., Sayler, G. S., 2009. Factors influencing the persistence of fecal bacteroides in stream water. Journal of Environmental Quality 38, 1224e1232. Davies, C.M., Long, J.A.H., Donald, M., Ashbolt, N.J., 1995. Survival of fecal microorganisms in marine and fresh-water sediments. Applied and Environmental Microbiology 61, 1888e1896. Dick, L.K., Bernhard, A.E., Brodeur, T.J., Domingo, J.W.S., Simpson, J.M., Walters, S.P., Field, K.G., 2005. Host distributions of uncultivated faecal bacteroidales bacteria reveal genetic markers for faecal source identification. Applied and Environmental Microbiology 71, 3184e3191. Dick, L.K., Stelzer, E.A., Bertke, E.E., Fong, D.L., Stoeckel, D.M., 2010. Relative decay of bacteroidales microbial source tracking markers and cultivated Escherichia coli in freshwater microcosms. Applied and Environmental Microbiology 76, 3255e3262. Easton, J. H., Lalor, M., Gauthier, J. J., Pitt, R., Newman, D. E., Meyland, S., 1999. Determination of survival rates for selected bacterial and Protozoan pathogens from Wet Weather Discharges. WEFTEC 1999. Water Environment Federation 72nd Annual Conference and Exposition. New Orleans. Field, K.G., Samadpour, M., 2007. Fecal source tracking, the indicator paradigm, and managing water quality. Water Research 41, 3517e3538. Gourmelon, M., Caprais, M.P., Mieszkin, S., Marti, R., We´ry, N., Jarde´, E., Derrien, M., Jadas-Hecart, A., Communal, P.Y., Jaffrezic, A., Pourcher, A.M., 2010. Development of microbial and chemical MST tools to identify the origin of the faecal pollution in bathing and shellfish harvesting waters in France. Water Research 44, 4812e4824. Hartz, A., Cuvelier, M., Nowosielski, K., Bonilla, T.D., Green, M., Esiobu, N., McCorquodale, D.S., Rogerson, A., 2008. Survival potential of Escherichia coli and enterococci in subtropical beach sand: implications for water quality managers. Journal of Environmental Quality 37, 898e905. Hellweger, F.L., Bucci, V., Litman, M.R., Gu, A.Z., Onnis-Hayden, A., 2009. Biphasic decay kinetics of faecal bacteria in surface water not a density effect. Journal of Environmental Engineering-Asce 135, 372e376. Jarde, E., Gruau, G., Jaffrezic, A., 2009. Tracing and quantifying sources of fatty acids and steroids in amended cultivated soils. Journal of Agricultural and Food Chemistry 57, 6950e6956. Jeanneau, L., Jarde, E., Gruau, G., 2011. Influence of salinity and natural organic matter on the solid phase extraction of sterols and stanols: application to the determination of the human sterol fingerprint in aqueous matrices. Journal of Chromatography A 1218, 2513e2520.
Jin, G., Jeng, H.W., Bradford, H., Englande, A.J., 2005. Comparison of E. coli, Enterococci, and fecal coliform as indicators for brackish water quality assessment (vol 76, pg 245, 2004). Water Environment Research 77, 433. Konstantinov, S.R., Smidt, H., de Vos, W.M., 2005. Representational difference analysis and real-time PCR for strain-specific quantification of Lactobacillus sobrius sp nov. Applied and Environmental Microbiology 71, 7578e7581. Lee, D.U., Heinz, V., Knorr, D., 2001. Biphasic inactivation kinetics of Escherichia coli in liquid whole egg by high hydrostatic pressure treatments. Biotechnology Progress 17, 1020e1025. Lee, S.K., Calcott, P.H., MacLeod, R.A., 1977. Relationship of cytochrome content to the sensitivity of bacteria to NaCl on freezing and thawing. Canadian Journal of Microbiology 23, 413e419. Leeming, R., Ball, A., Ashbolt, N., Nichols, P., 1996. Using faecal sterols from humans and animals to distinguish faecal pollution in receiving waters. Water Research 30, 2893e2900. Marino, R.P., Gannon, J.J., 1991. Survival of faecal coliforms and faecal streptococci in storm drain sediment. Water Research 25, 1089e1098. Marti, R., Dabert, P., Ziebal, C., Pourcher, A.M., 2010. Evaluation of Lactobacillus sobrius/L. amylovorus as a new microbial marker of pig manure. Applied and Environmental Microbiology 76, 1456e1461. Marty, Y., Quemeneur, M., Aminot, A., LeCorre, P., 1996. Laboratory study on degradation of fatty acids and sterols from urban wastes in seawater. Water Research 30, 1127e1136. Menon, P., Billen, G., Servais, P., 2003. Mortality rates of autochthonous and fecal bacteria in natural aquatic ecosystems. Water Research 37, 4151e4158. Mieszkin, S., Furet, J.P., Corthier, G., Gourmelon, M., 2009. Estimation of pig faecal contamination in a river catchment by real-time PCR using two pig-specific bacteroidales 16S rRNA genetic markers. Applied and Environmental Microbiology 75, 3045e3054. Neysens, P., Messens, W., De Vuyst, L., 2003. Effect of sodium chloride on growth and bacteriocin production by Lactobacillus amylovorus DCE 471. International Journal of Food Microbiology 88, 29e39. Okabe, S., Shimazu, Y., 2007. Persistence of host-specific Bacteroides-Prevotella 16S rRNA genetic markers in environmental waters: effects of temperature and salinity. Applied Microbiology and Biotechnology 76, 935e944. Pratt, C., Warnken, J., Leeming, R., Arthur, J.M., Grice, D.I., 2008. Degradation and responses of coprostanol and selected sterol biomarkers in sediments to a simulated major sewage pollution event: a microcosm experiment under sub-tropical estuarine conditions. Organic Geochemistry 39, 353e369. Shah, V.G., Dunstan, R.H., Geary, P.M., Coombes, P., Roberts, M.K., Von Nagy-Felsobuki, E., 2007. Evaluating potential applications of faecal sterols in distinguishing sources of faecal contamination from mixed faecal samples. Water Research 41, 3691e3700. Soller, J.A., Schoen, M.E., Bartrand, T., Ravenscroft, J.E., Ashbolt, N.J., 2010. Estimated human health risks from exposure to recreational waters impacted by human and nonhuman sources of faecal contamination. Water Research 44, 4674e4691. Thoumelin, G., Marty, Y., Lecorre, P., Aminot, A., 1990. Laboratory investigation of the degradation of organic-matter in estuarine and coastal waters - sterols variations. Oceanologica Acta 13, 53e60. Thurston-Enriquez, J.A., Gilley, J.E., Eghball, B., 2005. Microbial quality of runoff following land application of cattle manure and swine slurry. Journal of Water and Health 3, 157e171.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 6 2 3 e4 6 3 3
Tyagi, P., Edwards, D.R., Coyne, M.S., 2009. Fecal sterol and bile acid biomarkers: runoff concentrations in animal waste-amended pastures. Water Air and Soil Pollution 198, 45e54. Ufnar, J.A., Ufnar, D.F., Wang, S.Y., Ellender, R.D., 2007. Development of a swine-specific faecal pollution marker
4633
based on host differences in methanogen mcrA genes. Applied and Environmental Microbiology 73, 5209e5217. Walters, S.P., Yamahara, K.M., Boehm, A.B., 2009. Persistence of nucleic acid markers of health-relevant organisms in seawater microcosms: Implications for their use in assessing risk in recreational waters. Water Research 43, 4929e4939.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 6 3 4 e4 6 4 0
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Specific detection of viable Listeria monocytogenes in Spanish wastewater treatment plants by Fluorescent In Situ Hybridization and PCR Yolanda Moreno, Lorena Ballesteros, Jorge Garcı´a-Herna´ndez, Paula Santiago, Ana Gonza´lez, M. Antonia Ferru´s* Departamento de Biotecnologı´a, Universidad Polite´cnica de Valencia, Camino de Vera 14, 46022 Valencia, Spain
article info
abstract
Article history:
Listeria monocytogenes detection in wastewater can be difficult because of the large amount
Received 15 October 2010
of background microbiota and the presence of viable but non-culturable forms in this
Received in revised form
environment. The aim of this study was to evaluate a Fluorescent In Situ Hybridization
15 June 2011
(FISH) assay combined with Direct Viable Count (DVC) method for detecting viable L.
Accepted 15 June 2011
monocytogenes in wastewater samples, as an alternative to conventional culture methods.
Available online 24 June 2011
16S rRNA sequence data were used to design a specific oligonucleotide probe. In order to assess the suitability of the method, the assays were performed on naturally (n ¼ 87) and
Keywords:
artificially (n ¼ 14) contaminated samples and results were compared to those obtained
L. monocytogenes
with the isolation of cells on selective media and with a PCR method. The detection limit of
FISH
FISH and PCR assays was 104 cells/mL without enrichment and 10 cells/mL after enrich-
PCR
ment. A total of 47 samples, including 3 samples from effluent sites, yielded FISH positive
Wastewater
results for L. monocytogenes. Using DVC-FISH technique, the presence of viable L. mono-
Detection
cytogenes cells was detected in 23 out of these 47 FISH positive wastewater samples. PCR and culture methods yielded 27 and 23 positive results, respectively. According to these results, FISH technique has the potential to be used as a sensitive method for the detection and enumeration of L. monocytogenes in environmental wastewater samples. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
Listeria monocytogenes is the causal agent of one of the most important foodborne diseases worldwide, with a case-fatally rate of about 30% (Newell et al., 2010). Infection has been associated with the ingestion of a great variety of food products (Lunden et al., 2004). L. monocytogenes is widely distributed in the natural environment (Fenlon, 1999) and can be found in wastewater at high levels (Paillard et al., 2005; Odjadjare et al., 2010). Some authors have observed that L. monocytogenes is able to survive wastewater physical secondary treatment
(Paillard et al., 2005). Therefore, the organism could re-enter the human food chain via sludge applications to land as fertilizer or by irrigating crops with treated water. Although when present in low concentrations it should not pose a risk to human health, several cases of Listeria outbreaks associated with raw and treated wastewater have been reported (Paillard et al., 2005) and there is a need for more extensive studies to assess the real risk for public health (Odjadjare et al., 2010). Isolation of L. monocytogenes may require about 4e5 days. Besides, under unfavourable environmental conditions Listeria cells can enter into a Viable But Nonculturable (VBNC)
* Corresponding author. Tel.: þ34 963877423; fax: þ34 963879429. E-mail address: [email protected] (M.A. Ferru´s). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.06.015
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 6 3 4 e4 6 4 0
state, in which they cannot be detected by traditional culture methods (Bersnard et al., 2000). Molecular techniques such as PCR have been applied as a rapid alternative to conventional detection methods (Liu, 2008; Shannon et al., 2007). However, when applying PCR-based methods to environmental samples a complication arises: inhibitory substances, such as humic acids, can have a significant effect on the activity of the Taq polymerase enzyme (Lemarchand et al., 2005). Ribosomal RNA probe hybridization without culturing (Fluorescent in situ Hybridization, FISH) is less sensitive to inhibitory substances than PCR and has shown to be a very useful tool for phylogenetic, ecological, diagnostic and environmental studies in microbiology (Bottari et al., 2006). It has been successfully used for the detection and identification of different pathogens in foods, surface water, drinking water and wastewater (Moreno et al., 2007; Piqueres et al., 2006; Schmid et al., 2003). FISH and PCR techniques do not allow discrimination between dead and viable cells (Okoh et al., 2007), which is a limitation that can lead to false positive results. The Direct Viable Count (DVC) procedure (Kogure et al., 1979) involves exposing bacterial cells to a revival medium which contains antibiotics that prevent cellular division; elongated cells are then enumerated as viable cells (Servais et al., 2009). The combination of DVC, which rises intracellular rRNA levels and increases cell length (Doudu and Colquhoun, 2010), with FISH has been proposed for monitoring viable cells in different environments (Baudart et al., 2002; Piqueres et al., 2006). In this work, the development of a rapid DVC- in situ hybridization combined protocol using a 16S rRNA probe to detect and identify viable L. monocytogenes in raw and treated wastewater samples for investigating the occurrence and the survival to wastewater treatments of this pathogen has been reported. Results were compared with those obtained by a culture on selective media and by a PCR method.
2.
Materials and methods
2.1.
Bacterial strains and culture conditions
Seven reference Listeria strains, i.e. 5 L. monocytogenes (CECT 911, CECT 4031, CECT 4032, CECT 933 and CECT 936), 1 Listeria ivanovii (CECT 913), 1 Listeria innocua (CECT 910) were used in experiments. Fourteen non-Listeria strains (Vibrio vulnificus CECT 529, Vibrio parahaemolyticus CECT 511, Staphylococcus aureus CECT 240, Micrococcus luteus CECT 245, Citrobacter freundii CECT 401, Escherichia coli CECT 349, Salmonella enterica CECT 915, Enterobacter cloacae CECT 194, Pseudomonas aeruginosa ATCC 10145, Enterobacter faecalis DSMZ 20478, Campylobacter jejuni NCTC 11828, Campylobacter coli NCTC 11366, Helicobacter pylori NCTC 11637 and Arcobacter butzleri NCTC 12481) were also used to evaluate the specificity of the assays. All of them were rehydrated and cultured according to their Culture Collections instructions (CECT; Spanish Type Culture Collection, Valencia, Spain. ATCC; American Type Culture Collection, Rockville, Md. USA. DSMZ; German Type Culture Collection, Braunschweig, Germany. NCTC; Health Protection Agency Culture Collections, Salisbury, UK). Listeria strains were grown on TSA (Casein e peptone soy meal e peptone agar for microbiology, Merck, Darmstadt, Germany)
4635
for 24 h at 37 C and subcultured overnight in Brain-HeartInfusion (BHI) broth (Merck, Darmstadt, Germany) for specificity and sensitivity assays.
2.2.
Fluorescent in situ hybridization assays
An oligonucleotide probe complementary to a 16S rRNA region of L. monocytogenes was designed (Lmon probe: 50 -CTATCCATTGTAGCACGTG-30 ). The probe targeted positions 1242e1260 in L. monocytogenes 16S rRNA. The specificity of the probe was assessed by a gapped BLAST search (Altschul et al., 1997). Specificity of L. monocytogenes probe was also evaluated by in situ hybridization with the Listeria and non-Listeria reference strains included in this study, by whole-cell hybridization according to Wagner et al. (1998). The probe was synthesized and labelled by MGW Biotech (Mannheim, Germany) with 5 (6)-carboxyfluorescein-N-hydroxysuccinimide ester (FLUOS) and CY3. For FISH analysis, a volume of 1 mL of each sample was centrifuged (8000 rpm, at 4 C for 10 min). Resulting pellet was resuspended in PBS buffer (130 mM sodium chloride, 10 mM sodium phosphate, [pH 7.2]), and then fixed with ethanol/PBS (50:50) mixture (Amann et al., 1995). Fixed samples were stored at 20 C until their hybridization. Ten mL of each fixed sample were spotted on a gelatincoated slide, allowed to air dry, treated with lysozyme and dehydrated (50, 80, 100% ethanol) as previously described (Wagner et al., 1998). A concentration of 20% formamide in the hybridization buffer (0.9 M NaCl, 0.01% SDS, 20 mM TriseHCl, pH 7.6) was enough to reach the specificity of the whole-cell hybridization. In the case of wastewater FISH analysis, the percentage of formamide was increased to 30% to avoid crosshybridization with other non-culturable genus of bacteria present in the samples. The EUB338 universal probes mixture, complementary to a region of 16S rRNA of the domain Bacteria (Amann et al., 1995) was used as a positive control. Slides were mounted in FluoroGuard Antifade Reagent (Bio-Rad, Madrid, Spain) and visualized by epifluorescence Olympus microscopy BX50 with 460e490 nm (U-MWIB) and 520e550 nm (U-MWIG) exciter filters.
2.3.
DVC-FISH assay optimization
An overnight culture of L. monocytogenes was serially diluted from 101 to 108 CFU/mL. An aliquot of 1 mL of each dilution was added to 9 mL of BHI broth supplemented with yeast extract (2.5 mg/mL) and different concentrations (0.8, 1, 1.5 mg/ mL) of the antimicrobial agent Ciprofloxacin (SIGMA Chemical Co., St. Louis, Mo.), as described by Bersnard et al. (2000). DVC broths were incubated at 37 C in aerobic conditions and aliquots from each dilution were immediately taken after inoculation and after 7, 16 and 24 h and fixed for FISH as described above.
2.4.
PCR analysis
Cells were harvested by centrifugation at 14,000 g for 30 s, washed twice with sterile phosphate buffer and suspended in TE buffer (10 mM Tris-Cl, pH 7.5, 1 mM EDTA). DNA isolation was performed using Realpure genomic DNA isolation kit
4636
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 6 3 4 e4 6 4 0
(Durviz, Paterna, Spain) according to the manufacturer’s instructions. Concentrated DNA was stored at 20 C. L. monocytogenes detection was performed by multiplex PCR using the primers described by Bansal et al. (1996), which amplify a genus specific 938 bp fragment of 16S rDNA of Listeria sp. and a species specific 750 bp fragment of hlyA gene of L. monocytogenes, according to Zamora et al. (2000). A final reaction volume of 30 ml was obtained by addition of 3 ml of sample, 20 pmol of each primer, 0.2 mM of each deoxynucleotide, 2.5 mM MgCl2 and 2 U of Taq polymerase (New England Biolabs, UK). The amplification consisted of an initial DNA denaturing step at 95 C for 1 min, followed by a 40-cycle reaction (94 C for 30 s, 51 C for 20 s and 74 C for 20 s) and a final extension step at 72 C for 2 min. PCR reactions were performed with an automatic thermal cycler (PHC-3 Thermal Cycler, Techne Corporation, Cambridge, UK). PCR products were analyzed by electrophoresis at 100 V for 1 h through 1% (wt/vol) SeaKem LE agarose (FMC Bioproducts, Denmark) gels. Amplimers were visualized by ethidium bromide staining under UV light. A 100 bp DNA ladder was used as a molecular weight marker.
2.5. Detection of L. monocytogenes in inoculated samples An overnight culture of L. monocytogenes CECT 911 was serially diluted from 100 to 108 CFU/mL, and used to inoculate 300 mL of PBS buffer and 300 mL of two Listeria-free (Listeria negative by PCR and by culture) wastewater samples with different origins (one from each plant, A and B). The number of cells in each dilution was calculated from the colony count on Tryptone Soy Agar plates (Merck, Darmstadt, Germany). For direct detection, cells from 100 mL of each inoculated water sample were harvested by centrifugation (1000 g) and the pellet was resuspended in 3 mL of PBS buffer. For detection after enrichment, one hundred mL of each inoculated water sample were filtered through 0.45 mm cellulose nitrate membrane filters (Whatman, Maidstone, England), the membranes were aseptically rolled, transferred to 100 mL of Modified Fraser broth (Garrec et al., 2003) and incubated in aerobic conditions at 37 C for 24 h. One mL aliquots of each PBS suspension and each enrichment broth were fixed, inoculated for DVC incubation and processed for PCR analysis respectively.
culture, FISH, DVC-FISH and PCR analysis within 6 h of collection. All the assays, except the DVC-FISH one, were performed with and without enrichment, as described for inoculated samples. To confirm the results, each sample was tested twice in different experiments. Aliquots of 100 mL of each sample, before and after enrichment, were plated onto selective Palcam media (OXOID SA, Madrid, Spain) and Chromogenic Listeria Agar (OCLA, OXOID) and incubated at 37 C for 24 h. Presumptive colonies were purified and identified by API-Lis biochemical system (Biomerieux, Mercy L’Etoile, France) and PCR, as described above.
3.
Results
3.1.
FISH control assays
Alignment of the Lmon designed probe with the sequences deposited in GenBank for Listeria and other related organisms showed that the studied probe completely and exclusively matched the target region of L. monocytogenes 16S rRNA. The Lmon probe hybridized with all the L. monocytogenes strains, whereas none of the other tested species showed any fluorescent signal (Fig. 1). The detection limit of the FISH assays for L. monocytogenes in inoculated PBS buffer was 104 CFU/mL. In inoculated wastewaters, it was ten-fold higher than in PBS (105 CFU/mL). However, the 24-h enrichment step enabled a detection limit of 10 CFU/mL in both PBS and wastewaters (Fig. 2). The optimal concentration of Ciprofloxacin for DVC analysis was 2 mg/L in both pure bacterial cultures and spiked samples. A 7-h incubation generated a significant elongation of viable cells. DVC-FISH specific analysis showed elongated and not elongated L. monocytogenes cells in the inoculated wastewater samples but, in accordance with Bersnard et al. (2000),
2.6. Detection of L. monocytogenes in natural wastewater samples A total of 261 samples were collected in 87 sampling campaigns from two secondary wastewater treatment plants during a 2-year period: 45 sample collections (M1eM45) were taken from plant A (899,000 population equivalents), and 42 (M46eM87) from plant B (262,000 population equivalents). Both plants, located in Valencia, Spain, collect urban and industrial wastewater and apply biological secondary treatment (activated sludge tank), and tertiary UV disinfection treatment. Final effluent is discharged into the sea or used for irrigation purposes. On each occasion, samples were taken from three sites: the influent, the secondary treatment effluent and after the tertiary treatment. All samples were placed into sterile glass bottles, refrigerated and processed for
Fig. 1 e FISH micrographs showing hybridization of a mixed L. monocytogenes CECT 911 and L. innocua CECT 910 culture by simultaneous application of probes Lmon-CY3 and EUB338mix-FLUOS. Green fluorescence signifies hybridization with EUBmix-FLUOS and orange signifies hybridization with both Lmon-CY3 and EUB338mix-FLUOS.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 6 3 4 e4 6 4 0
4637
Fig. 2 e FISH-DVC specific detection of L. monocytogenes CECT 911 in inoculated wastewater samples.
only those elongated at least twice their original size were estimated as viable.
3.2.
PCR control assay
Alignment of the sequences of L. monocytogenes deposited in GenBank with those of other related organisms showed that the two sets of primers used in this study were suitable for multiplex PCR detection of L. monocytogenes species. All the reference Listeria strains tested were positive by multiplex PCR (mPCR) reaction. No amplification was obtained for the remaining bacterial genera included in this work. The detection limit of mPCR assay for L. monocytogenes in both inoculated wastewater samples and PBS buffer without enrichment was 104 CFU/mL and after 24 h of incubation it decreased to 10 CFU/mL. Incubation periods longer than 24 h did not improve the detection limit in any type of sample.
3.3. L. monocytogenes detection in wastewater samples Forty-eight out of the 261 samples tested were found to be positive for the presence of L. monocytogenes using any of the assays, including one PCR positive sample, which was Listeria negative by FISH and culture detection (M44). By FISH analysis, 47 samples yielded positive results for Lmon probe hybridization (Fig. 3), 37 of them allowing L. monocytogenes detection prior to enrichment (Table 1). Twenty-three out of these 37 FISH positive samples were also positive by DVC-FISH (presence of elongated viable L. monocytogenes cells). Positive PCR results for L. monocytogenes were obtained in 27 samples, although in 25 only after the enrichment step. By culture, 23 samples were positive, although in 25 cases, only after enrichment. Biochemical tests and multiplex PCR identified all the isolates as L. monocytogenes. When compared, 15 samples yielded concordant results for all three methods. FISH and culture results were concordant for 23 samples and discordant for 24 samples (Table 2). FISH and PCR yielded concordant results for 26 samples, while discordances were observed in 22 cases. Finally, culture and PCR results were concordant for 31 samples and discordant for 17. Thirteen FISH positive samples yielded negative results by both, PCR and culture techniques. In eleven cases, samples were positive by FISH and PCR, but not by culture. PCR was negative for eight FISH and culture positive samples.
Fig. 3 e FISH-DVC detection of L. monocytogenes by simultaneous application of probes Lmon-CY3 and EUB338mix-FLUOS in wastewater samples.
When comparing results from the two wastewater treatment plants, percentage of positive samples by any technique was 49% (22/45) in plant A and 62% (26/42) in B. Although there were two positive samples after secondary treatment in plant A and 6 in B, only three samples (2 in plant A and one in plant B) were Listeria-positive after tertiary treatment.
4.
Discussion
In this study, a specific probe for L. monocytogenes FISH detection has been developed. The FISH method has the advantage of not being inactivated by inhibitors, even when a large amount of sample is processed (Moreno et al., 2001). Besides, a protocol to obtain the DNA from bacteria is not necessary, and positive results may be directly observed in the sample without culture. A combination of FISH with a DVC step allowed the direct in situ identification and visualization of viable cells of L. monocytogenes within the sample. A positive response to the DVC procedure indicates that a cell is substrate-responsive in the incubation conditions used for DVC. This does not mean that a DVC-positive cell is active in the natural environment, but it demonstrates that its cellular machinery is intact and that it can be considered viable and, therefore, potentially infective (Servais et al., 2009). According to other authors (Moreno et al., 2001; Liu, 2008), we included an enrichment step in order to reduce the use of large volume samples to reach the detection limit of pathogens in low contaminated samples. In contrast with other works (Piqueres et al., 2006) in which enrichment diminished the effectiveness of detection, enrichment increased the level of L. monocytogenes cells present in the samples, thus avoiding interference of competitive microbiota. The increase of positive samples after an enrichment step shows the possible underestimation of this pathogen when samples are directly analyzed by FISH. The FISH assay was shown to be more sensitive than culture: of the 48 samples positive for the presence of
4638
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 6 3 4 e4 6 4 0
Table 1 e Detection of L. monocytogenes in natural wastewater samples. Only positive samples for any assay are shown. Results were obtained prior to enrichment, unless indicated. Samples M1 M2 M3 M4 M5 M6 M7 M8 M9 M10 M11 M12 M13 M14 M15 M16 M17 M18 M19 M20 M21 M44 M48 M49 M51 M52 M53 M53 M55 M56 M56 M58 M59 M59 M61 M63 M65 M66 M67 M67 M67 M70 M70 M73 M76 M79 M85 M85
Treatment plant A A A A A A A A A A A A A A A A A A A A A A B B B B B B B B B B B B B B B B B B B B B B B B B B
Origin Influent Influent Influent Influent Influent Influent Influent Influent Influent Influent Influent After secondary treatment Influent After tertiary treatment Influent After secondary treatment Influent Influent After secondary treatment Influent After secondary treatment After tertiary treatment Influent Influent Influent Influent Influent After secondary treatment Influent Influent After secondary treatment Influent Influent After secondary treatment Influent Influent Influent Influent Influent After secondary treatment After tertiary treatment Influent After secondary treatment Influent Influent Influent Influent After secondary treatment
Culturea þb þb þb þb þ þ þb þ þb þb þb þ þb þ þb þb þ þ þb þb þb þ þb
FISH b
þ þ þ þ þ þb þ þ þ þ þ þ þb þ þb þb þb þ þb þ þ þ þ þ þ þ þ þ þ þb þ þb þ þb þ þ þ þ þ þ þ þ þ þ þ þ þ
DVC-FISH
mPCR
þ þ þ þ þ þ þ þ þ þ þ þ þ þ þ þ þ þ þ þ þ þ þ
þb þb þb þb þb þb þb þb þb þb þb þb þb þb þb þb þb þb þb þb þb þb þb þ þ þb þ
a Isolate identified as L. monocytogenes. b Positive results obtained only after enrichment.
L. monocytogenes detected by any technique, half were positive by the FISH test and negative by culture. Thus, as expected, DVC-FISH yielded more positive results than culture before enrichment. There were four DVC-negative samples in which culture was positive, but only after enrichment, indicating that the number of cells in the sample was too low to be directly detected by this technique. However, in two samples (M53 and M61) culture was positive without enrichment, while DVC-FISH yielded false-negative results. It must be considered that differentiation between live and dead Listeria based on
the elongation is a subjective test, which contributes to increase the subjectivity of any fluorescent assay (Coallier et al., 1994). Furthermore, fluoroquinolone MIC can vary according to the strain and subMIC concentrations for one isolate can inhibit and even kill other ones. The comparison of results obtained using the molecular methods described in this study shows that FISH allowed direct detection of this pathogen in a high percentage of naturally contaminated wastewater samples, although in inoculated samples it showed the same detection limit as the
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 6 3 4 e4 6 4 0
Table 2 e Comparison of L. monocytogenes detection results obtained by the three methods used in this study. Number of samples PCR FISH
PCR þ þ þ
26 1
Culture
þ
21 0
18 5 23 0
12 13 24 1
PCR technique. It has been reported that FISH sensibility can be affected by some environmental factors, such as contamination (Bouvier and del Giorgio, 2003). However, in our work, background fluorescent signals due to non-specific probe attachment to flocks microbiota did not interfere with the FISH specific signal. In contrast, microbial background represented a problem for L. monocytogenes isolation from this complex bacterial community due to overgrowth in selective media. In six samples, viable Listeria cells were directly detected from the sample, while culture was negative, even after enrichment. This could indicate the presence of viable but non-culturable (VBNC) cells. This fact may be important from a sanitary point of view; some authors (Pommepuy et al., 1996; Asakura et al., 2002; Jolivet-Gougeon et al., 2006) have suggested that pathogenic VBNC bacteria can maintain their virulence, becoming a potential reservoir of disease. Although PCR method is considered to be more sensitive than FISH or culture methods, more L. monocytogenes positive samples were obtained by FISH technique than by culture or PCR analysis, even when an enrichment step was performed. Similar results have been previously found for the detection of other pathogens such as H. pylori in wastewater (Piqueres et al., 2006). Lepeuple et al. (2003) also reported the problem of cultural techniques and the higher number of positive samples by FISH techniques to enumerate E. coli in water. In our work, an enrichment step was necessary for obtaining multiplex PCR positive results in most samples. To confirm this, each sample was tested twice and, for all samples, repeated PCR analysis yielded consistent results. Some authors have suggested that this could be due to inhibitory substances present in wastewater as humic acids that can have a significant effect on the activity of the Taq polymerase enzyme, yielding false-negative results (Lemarchand et al., 2005). An enrichment step dilutes inhibitors of the sample, thus improving detection rates (Liu, 2008). Although FISH seems to be more effective than PCR for L. monocytogenes detection in wastewater samples, and including a DVC step has allowed us to distinguish between live and dead cells, its combination with PCR could be an excellent tool to avoid false-negative results. However, for detection purposes in environmental samples, the costebenefit ratio must be taken into account. Thirty out of 87 samplings of wastewater showed the presence of L. monocytogenes when directly examined by FISH, what means that concentrations were at least 104 CFU/mL. Percentage of positive samples by any technique was similar for both wastewater treatment plants (49% in A and 62% in B). Although there were more positive samples after secondary
4639
treatment in plant B, differences were not significant. These results are in agreement with those reported by other authors, which have previously shown a great prevalence of L. monocytogenes in wastewater (Paillard et al., 2005). Despite an official analysis to detect the presence of L. monocytogenes on final effluents is not usually required, the spread of this pathogen in the environment should be controlled since these treated waters could be used in agriculture, reaching humans or animals.
5.
Conclusions
Viable L. monocytogenes cells are present in wastewater samples, including final treated water. These findings indicate that L. monocytogenes is able to survive tertiary wastewater treatment. The DVC-FISH combined method developed in this work is a quick and specific tool for the detection, identification and direct visualization of viable L. monocytogenes in complex mixed communities such as wastewater samples.
Acknowledgments This work was supported by the grant AGL2008-05275-C03-02 (national and FEDER funds) from Ministerio de Ciencia e Innovacio´n, Spain.
references
Amann, R.I., Ludwig, W., Schleifer, K.H., 1995. Phylogenetic identification and in situ detection of individual microbial cells without cultivation. Microbiol. Rev. 59, 143e169. Altschul, S.F., Madden, T.L., Scha¨ffer, A.A., Zhang, J., Zhang, Z., Miller, W., Lipman, D.J., 1997. Gapped BLAST and PSI-BLAST: a new generation of protein database search programs. Nucleic Acids Res. 1 25 (17), 3389e3402. Asakura, H., Watarai, M., Shirahata, T., Makino, S., 2002. Viable but nonculturable salmonella species recovery and systemic infection in morphine-treated mice. J. Infect. Dis. 186 (10), 1526e1529. Bansal, N.S., McDonell, F.H.Y., Smith, A., Arnold, G., Ibrahim, G.F., 1996. Multiplex PCR assay for the routine detection of Listeria in food. Int. J. Food Microbiol. 33, 293e300. Baudart, J., Coallier, J., Laurent, P., Pre´vost, M., 2002. Rapid and sensitive enumeration of viable diluted cells of members of the family Enterobacteriaceae cells in freshwater and drinking waters. Appl. Environ. Microbiol. 68, 5057e5063. Bersnard, J., Federighi, M., Cappelier, J.M., 2000. Evidence of Viable But Non-Culturable state in Listeria monocytogenes by direct viable count and CTC-DAPI double staining. Food Microbiol. 17, 697e704. Bottari, B., Ercolini, D., Gatti, M., Neviani, E., 2006. Application of FISH technology for microbiological analysis: current state and prospects. Appl. Microbiol. Biotechnol. 73, 485e494. Bouvier, T., del Giorgio, P.A., 2003. Factors influencing the detection of bacterial cells using fluorescence in situ hybridization (FISH): a quantitative review of published reports. FEMS Microbiol. Ecol. 44, 3e15. Coallier, J., Prevots, M., Rompre, A., 1994. The optimization and application of two direct viable count methods for bacteria in distributed drinking water. Can. J. Microbiol. 40, 830e836.
4640
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 6 3 4 e4 6 4 0
Doudu, S., Colquhoun, D., 2010. Monitoring the survival of fishpathogenic Francisella in water microcosms. FEMS Microbiol. Ecol. 74, 534e541. Fenlon, D.R., 1999. Listeria monocytogenes in the natural environment. In: Ryser, E.T., Marth, E.H. (Eds.), Listeria, Listeriosis and Food Safety, pp. 21e37. New York, N.Y., U.S.A. Garrec, N., Picard-Bonnaud, F., Pourcher, A.M., 2003. Occurrence of Listeria sp. and L. monocytogenes in sewage sludge used for land application: effect of dewatering, liming and storage in tank on survival of Listeria species. FEMS Immunol. Med. Microbiol. 35, 275e283. Jolivet-Gougeon, A., Sauvager, F., Bonnaure-Mallet, M., Colwell, R.R., Cormier, M., 2006. Virulence of viable but nonculturable S. thyphimurium LT2 after peracetic acid treatment. Int. J. Food Microbiol. 112 (2), 147e152. Kogure, K., Simidu, U., Taga, N., 1979. A tentative direct microscopic method for counting living marine bacteria. Can. J. Microbiol. 25, 415e420. Lemarchand, K., Berthiaume, T., Maynard, C., Harel, J., Payment, P., Bayardelle, P., Masson, L., Brousseau, R., 2005. Optimization of microbial DNA extraction and purification from raw wastewater samples for downstream pathogen detection by microarrays. J. Microbiol. Methods 63, 115e126. Lepeuple, S., Delabre, K., Gilouppe, S., Intertaglia, L., de Roubin, M. R., 2003. Laser scanning detection of FISH-labelled Escherichia coli from water samples. Water Sci. Technol. 47, 123e129. Liu, D., 2008. Preparation of Listeria monocytogenes specimens for molecular detection and identification. Int. J. Food Microbiol. 122 (3), 229e242. Lunden, J., Tolvanen, R., Korkeala, H., 2004. Human listeriosis outbreaks linked to dairy products in Europe. J. Dairy Sci. 87 (E. Suppl.), E6eE11. Moreno, Y., Piqueres, P., Alonso, J.L., Jime´nez, A., Gonza´lez, A., Ferru´s, M.A., 2007. Survival and viability of Helicobacter pylori after inoculation into chlorinated drinking water. Water Res. 41, 3490e3496. Moreno, Y., Herna´ndez, M., Ferru´s, M.A., Alonso, J.L., Botella, S., Montes, R., Herna´ndez, J., 2001. Direct detection of thermotolerant campylobacters in chicken products by PCR and in situ hybridization. Res. Microbiol. 152, 577e582. Newell, D.G., Koopmans, M., Verhoef, L., Duizer, E., AidaraKane, A., Sprong, H., Opsteegh, M., Langelaar, M., Threfall, J., Scheutz, F., van der Giessen, J., Kruse, H., 2010. Food-borne
diseases e the challenges of 20 years ago still persist while new ones continue to emerge. Int. J. Food Microbiol. 139 (Suppl. 1), S3eS15. Odjadjare, E.E.O., Obi, L.C., Okoh, A.I., 2010. Municipal wastewater effluents as a source of Listerial pathogens in the aquatic milieu of the eastern Cape province of South Afrique: a concern of public health importance. Int. J. Environ. Res. Public Health 7, 2376e2394. Okoh, A.I., Odjadjare, E.E., Igbinosa, E.O., Osode, A.N., 2007. Wastewater treatment plants as a source of microbial pathogens in receiving watersheds. Afr. J. Biotechnol. 6, 2932e2944. Paillard, D., Dubois, V., Thiebaut, R., Nathier, F., Hoogland, E., Caumette, P., Quentin, C., 2005. Occurrence of Listeria spp. In: Effluents of French Urban Wastewater Treatment Plants. Appl. Envir. Microbiol., vol. 71, pp. 7562e7566. Piqueres, P., Moreno, Y., Alonso, J.L., Ferrus, M.A., 2006. A combination of direct viable count and fluorescent in situ hybridization for estimating Helicobacter pylori cell viability. Res. Microbiol. 157, 345e349. Pommepuy, M., Butin, M., Derrien, A., Gourmelon, M., Colwell, R. R., Cormier, M., 1996. Retention of enteropathogenicity by viable but nonculturable E. coli exposed to seawater and sunlight. Appl. Environ. Microbiol. 62, 4621e4626. Shannon, K.E., Lee, D.Y., Trevors, J.T., Beaudette, L.A., 2007. Application of real-time quantitative PCR for the detection of selected bacterial pathogens during municipal wastewater treatment. Sci. Total Environ. 382, 121e129. Schmid, M., Walcher, M., Bubert, A., Wagner, M., Wagner, M., Schleifer, K.H., 2javascript:AL_get(this, ‘jour’, ‘FEMS Immunol Med Microbiol.’), 2003. Nucleic acid-based, cultivation-independent detection of Listeria spp and genotypes of L. monocytogenes. FEMS Immunol. Microbiol. 35 (3), 215e225. Servais, P., Prats, J., Passerat, J., Garcia-Armisen, T., 2009. Abundance of culturable versus viable Escherichia coli in freshwater. Can. J. Microbiol. 55 (7), 905e909. Wagner, M., Schmid, M., Juretschk, S., Trebesius, K.H., Bubert, A., Goebel, W., Schleifer, K.H., 1998. In situ detection of a virulence factor mRNA and 16S rRNA in Listeria monocytogenes. FEMS Microbiol. Lett. 160, 159e168. Zamora, A., Ossa, H., Carrascal, A., Poutou, R., Jimenez, D., 2000. Identificacio´n preliminar de Listeria monocytogenes por PCR. Laboratorio Actual 17 (33), 38e41.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 6 4 1 e4 6 5 0
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Occurrence and treatment of wastewater-derived organic nitrogen Baiyang Chen a,*, Youngil Kim b, Paul Westerhoff c a
School of Civil and Environmental Engineering, Harbin Institute of Technology Shenzhen Graduate School, Shenzhen, PR China Chungcheongnamdo Watershed Management Research Center, Chungnam Development Institute, Republic of Korea c Arizona State University, School of Sustainable Engineering and the Built Environment, Engineering Center (G-Wing), Room ECG-252, Tempe, AZ 85287-5306, USA b
article info
abstract
Article history:
Dissolved organic nitrogen (DON) derived from wastewater effluent can participate in
Received 4 October 2010
reactions that lead to formation of nitrogenous chlorination by-products, membrane
Received in revised form
fouling, eutrophication, and nitrification issues, so management of DON is important for
12 April 2011
both wastewater reuse applications and nutrient-sensitive watersheds that receive
Accepted 16 June 2011
discharges from treated wastewater. This study documents DON occurrence in full-scale
Available online 24 June 2011
water/wastewater (W/WW) treatment plant effluents and assesses the removal of wastewater-derived DON by several processes (biodegradation, coagulation, softening, and
Keywords:
powdered activated carbon [PAC] adsorption) used for advanced treatment in wastewater
Dissolved organic nitrogen
reuse applications. After varying levels of wastewater treatment, the dominant aqueous
Nitrogen speciation
nitrogenous species shifts from ammonia to nitrate after aerobic processes and nitrate to
Organic matter characterization
DON in tertiary treatment effluents. The fraction of DON in total dissolved nitrogen (TDN)
Biodegradability
accounts for at most 52% in tertiary treated effluents (median ¼ 13%) and 54% in surface
Bioavailability
waters impacted by upstream wastewater discharges (median ¼ 31%). The 5-day biode-
Wastewater reuse
gradability/bioavailability of DON (39%) was higher, on average, than that of dissolved organic carbon (DOC, 26%); however, upon chlorination, the DON removal (3%) decreased significantly. Alum coagulation (with 8 mg/L alum per mg/L DOC) and lime softening (with pH 11.3e11.5) removed <25% of DON and DOC without selectivity. PAC adsorption preferentially removed more DOC than DON by 10% on average. The results provided herein hence shed light on approaches for reducing organic nitrogen content in treated wastewater. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
Reclaimed, recycled, and reused wastewaters are now perceived as a valuable alternative water supply in some arid areas to meet the demands of growing population and industrial development. In 2004, the United States Environmental Protection Agency (USEPA) upgraded its “Guidelines for Water
Reuse” (EPA/625/R-04/108) to promote wastewater reuse for urban, industrial, agricultural, and even potable purposes. However, concerns about possible adverse effects of compounds persisting in treated wastewater may hinder potable reuse practices (Servais et al., 1999). These constituents of health concern include but are not limited to pharmaceuticals (Heberer, 2002), endocrine-disrupting compounds (Snyder
* Corresponding author. Tel.: þ86 134 8072 7605; fax: þ86 755 2603 3509. E-mail addresses: [email protected] (B. Chen), [email protected] (Y. Kim), [email protected] (P. Westerhoff). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.06.018
4642
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 6 4 1 e4 6 5 0
et al., 2001), and disinfection by-products (DBPs) (Krasner et al., 2009). Earlier studies also found that dissolved organic nitrogen (DON) in water and wastewater has several adverse implications, including that it serves as a precursor of nitrogenous DBPs (Lee et al., 2007; Krasner et al., 2009); alters the speciation of carbonaceous DBPs (Hureiki et al., 1994); and supports microbial survival and growth, which may cause membrane fouling (Her et al., 2004), eutrophication (Pehlivanoglu and Selak, 2004), and nitrification issues (Zhang et al., 2009). Systematic evaluation of the occurrence and control of DON is therefore important to enable better decisions for wastewater reuse practices and receiving water protections. The occurrence and treatment of organic nitrogen in algaeimpacted drinking water supplies can shed light, potentially, on DON removal in wastewater effluents. Studies in DON is present in drinking water supplies at levels typically less than 0.3 mg-N/L, although some highly eutrophic surface waters contain up to 10 mg-N/L (Westerhoff and Mash, 2002; Lee and Westerhoff, 2006; Pehlivanoglu-Mantas and Sedlak, 2006; Dotson et al., 2008; Dotson and Westerhoff, 2009). The composition of DON in these water supplies is not well known, but can include small amounts of free (FAAs) and combined (CAAs) amino acids, which consist of FAAs, hydrolyzable proteins and polypeptides. For example, a survey of several algal-influenced drinking water supplies found that, on average, FAAs and CAAs accounted for 0.5% and 15% of DON, respectively, with the other portion of DON remaining unclassified (Dotson and Westerhoff, 2009). In treated wastewater effluent, DON can account up to 2.5 mg/LN in activated sludge effluent. FAAs account for 0.1e2% of DON and CAAs composed less than 13% of DON (Burleson et al., 1980; Parkin and McCarty, 1981; Dignac et al., 2000). Fractionation of organic matter from reservoirs and wastewaters concluded that colloidal, basic and neutral organic matter fractions were nitrogen enriched relative to acidic fractions; a significant portion of DON is present in the colloidal fraction of poorly nitrified effluents (e.g., trickling filters); and a portion of the acidic fractions (proteins) were more nitrogen enriched from the terpenoid acid fractions (Leenheer et al., 2007). Such fractionation provides a framework for potential DON treatment. Coagulation of surface waters appears not to selectively remove nitrogen containing organic matter. Enhanced coagulation during drinking water treatment removed equal or slight lower amounts of DON (35%) as compared to dissolved organic carbon (DOC) (Lee and Westerhoff, 2006), and this value was on average 30% in algae-influenced waters (Dotson and Westerhoff, 2009). DON from raw wastewater was biodegradable (50e60%) in during activated sludge treatment, and advanced treatment of treated effluents achieved 72% DON removal using high levels of powdered activated carbon (PAC) dosage, or 33e56% removal by the use of cation exchange resins targeting to basic organic matter fractions (Parkin and McCarty, 1981). Even for low TN effluents (TN ¼ 4e5 mg/L), which represent well-nitrified and denitrified waters, algae and bacteria can utilize 18e61% of the DON (Urgun-Demirtas et al., 2008). While advanced treatment of wastewaters have not historically considered DON removal as a major goal, it may be increasingly important because of the reasons mentioned above.
Current analytical methods for DON have been developed for surface waters (Lee and Westerhoff, 2005; Vandenbruwane et al., 2007) but may require further refinement to deal with wastewater samples which often contain much higher levels of dissolved inorganic nitrogen (DIN). The two most commonly-used DON analysis methods involve: 1) subtracting ammonia from total Kjeldahl nitrogen (TKN) or 2) subtracting DIN, which includes ammonia, nitrite, and nitrate, from total dissolved nitrogen (TDN). Waters with low ammonia (e.g., well-nitrified or denitrified waters) often contain TKN levels at or below common detection limits, between 0.5 and 2 mg-N/L, and thus are often not sensitive enough for wastewater effluents intended for reuse. For the second method, the accuracy of the DON measurement relies strongly on the accuracies of the methods used to determine the amount of each DIN species and also is dependent on the fraction of DIN in TDN. It was reported that when DIN/TDN is higher than 0.6, the variance in DIN measurements can be greater than actual DON levels (Lee and Westerhoff, 2005). Therefore, addition of a dialysis pretreatment step was recommended to remove DIN prior to determination of DON. Dialysis pretreatment decreased DIN levels in the sample while not allowing larger DON molecules to permeate through the dialysis membrane, thus reducing the DIN/TDN ratio and facilitating accurate DON (Lee and Westerhoff, 2005; Vandenbruwane et al., 2007). Wastewater effluent is high in DIN (typically >5 mg/L) which results in elevated DIN/TDN ratios (typically >0.9). Consequently, wastewater effluents often require greater DIN removal during pretreatment to obtain reliable DON values. This study had two aims: 1) the occurrence of DON in the United States at various types of full-scale water/wastewater (W/WW) treatment plants, focusing on its relative magnitudes with TDN and DOC; and 2) DON treatment by certain commonly-used W/WW treatment processes, including biodegradation, coagulation, softening, and PAC adsorption, especially the factors affecting DON further treatment. It was hypothesized that DON contributes to a significant portion of the TDN in highly-treated wastewater effluent and that commonly employed advanced reuse processes can reduce DON levels. Evidence to support the aims were obtained from both field and laboratory tests, including a USA nationwide survey for full-scale plants, a series of bench-scale experiments, and two monitoring events of an effluent-dominated stream.
2.
Materials and methods
2.1.
Survey of plants
A survey examined 32 full-scale W/WW treatment plants utilizing a wide variety of treatment technologies across the USA, which are described in detail elsewhere (Krasner et al., 2008). The samples included 100 effluent samples from 23 wastewater treatment plants (WWTPs) using aerated lagoon, activated sludge, biofilter, nitrification, denitrification, membrane bioreactor, reverse osmosis, softening, PAC, or sand filtration processes; 30 samples from 9 drinking water treatment plants (DWTPs) equipped with conventional
4643
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 6 4 1 e4 6 5 0
Table 1 e Qualities of effluents for biodegradation tests. Parameters pH DOC DON TDN NHþ 4 NO 2 NO 3 UVA SUVA
Unit
Effluent from AL
Effluent from AS
Effluent from ND
Effluent from MBR
unitless mg/L-C mg/L-N mg/L-N mg/L-N mg/L-N mg/L-N cm1 L/mg-m
7.98 12.85 1.21 30.45 28.14 BDL 0.11 0.192 1.49
7.93 17.97 1.49 14.96 11.2 BDL 0.12 0.169 0.94
8.05 12.21 1.56 9.39 0.09 BDL 7.33 0.15 1.23
7.88 7.16 0.69 7.59 3.21 BDL 2.28 0.207 2.89
AL: aerated lagoon, AS: activated sludge, ND: nitrification and denitrification, MBR: membrane bioreactor, BDL: below detection limit.
coagulation, filtration, softening, ozonation, and chlorination processes; and 21 samples from 10 monitoring wells downstream of WWTPs with varying levels of soil aquifer treatments. The water quality differed significantly in terms of DOC (0.2e23 mg/L), DON (0.03e2.44 mg/L) concentrations, and ultraviolet absorbance at 254 nm wavelength (UVA254, 0.01e1.3 cm1); more details regarding the sampling sites, seasons, and the geological variations were archived in the project report (Krasner et al., 2008).
2.2.
Monitoring of effluent-dominated river
The Santa Cruz River, AZ served as an effluent-dominated stream for the study of DON fate and transport under natural conditions (Chen et al., 2009), and potentially represents DON transformations which could occur between upstream WWTP discharges and downstream DWTP intakes in other river systems. During dry periods, the Santa Cruz River (SCR) in Arizona (USA) consists entirely of wastewater effluent discharged from the Nogales International Wastewater Treatment Plant (NIWWTP). The contents of conservative ions, including chloride and sulfate, varied by less than 10% over 14 miles, confirming that the stream had no significant inflow from unexpected sources. The samples were collected in two seasons: in summer (June 1e3, 2004) and in winter (February 2, 2005). During the summer event, water was collected three times per day for three days at each of five sites over 14.3 miles (w23 km) along the river; during the winter event, each site was sampled only once. The NIWWTP treated approximately 10 million gallons per day of domestic wastewater during the study and employed aerated lagoon treatment. Our previous work showed that over a 14-mile reach below the NIWWTP there was a decrease in organic matter (DOC and DON) along with a shift from ammonia to nitrite and nitrate, indicating active biological mechanisms within the stream (Chen et al., 2009).
3.
Bench-scale experiments
3.1.
Biodegradation
Laboratory biodegradation experiments were carried using biologically active sand (BAS) reactors. The bioreactor protocol was adopted from an earlier study (Allgeier et al., 1996) in which
fine sand (Mesh #50) was acclimated with return activated sludge for more than two months. The feed sludge was obtained from a full-scale WWTP at a point immediately after nitrification and denitrification treatment but prior to the settling tank. After acclimation, each 1-L amber bioreactor was fed with 100 mL of BAS and 400 mL of target effluent. The bioreactors were kept in the dark at room temperature (w20 C) with continuous stirring, and supernatants were collected 1, 3, and 5 days after the start of the experiment. Dissolved oxygen concentrations were higher than 3 mg/L during the experiment period. The study employed four representative WWTP effluents to evaluate the effect of biological pretreatment methods on DON further biodegradability (Table 1). One effluent was obtained at the effluent of NIWWTP (named AL sample); one effluent was collected from an effluent of conventional AS treatment (named AS sample); another effluent named ND was retrieved from an effluent with nitrification and denitrification treatment (note: the target plant was different from the one at which the BAS feed sludge was collected); and the fourth effluent was obtained from an effluent after membrane bioreactor treatment (named MBR sample). The influence of chlorination on changes in organic matter biodegradability was also evaluated by dosing the same effluents with free chlorine (Cl2:DOC ¼ 3:1 in weight basis) at room temperature (20 C) and buffered pH (8.2). For effluents AL and AS, chloramine residuals were found after three days; for effluents ND and MBR, free chlorine was maintained for 24 h; all tests were followed by quenching of residual disinfectants via sodium sulfite.
3.2.
Physical/chemical treatment processes
The experimental methods used for coagulation, softening, and powdered activated carbon adsorption treatment have been described elsewhere (Westerhoff et al., 2005) and also in the Supplementary Information (SI). The target waters came from four sources (Table 2): one effluent was obtained from NIWWTP behind the aerated lagoon (AL sample); one sample was an AS treatment effluent from a full-scale WWTP (AS sample); the third sample was a nitrified/denitrified effluent (ND sample); and a fourth sample was an artificial soluble microbial product (SMP) generated by a 20-gallon activated sludge (AS) reactor, which was acclimated by glucose and inorganic nutrients (ammonia, iron, etc) to produce SMPs without other refractory organics present in full-scale wastewater effluents (Krasner et al., 2008).
4644
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 6 4 1 e4 6 5 0
Table 2 e Qualities of effluents for coagulation, softening, and PAC adsorption tests. Parameters pH DOC DON TDN NHþ 4 NO 2 NO 3 UVA SUVA
Units
Sample from AL
Sample from AS
Sample from ND
Sample from Lab-AS
unitless mg/L mg/L-N mg/L-N mg/L-N mg/L-N mg/L-N cm1 L/mg-m
7.7 12.95 1.47 25.33 25.60 2.09 0.16 0.157 1.21
7.0 10.95 1.28 20.78 BDL 0.60 20.17 0.264 2.41
7.3 10.41 2.44 11.80 0.35 0.78 8.23 0.173 1.66
7.3 3.90 0.68 25.86 1.06 BDL 24.14 0.075 1.91
AL: aerated lagoon, AS: activated sludge, ND: nitrification and denitrification, Lab-AS: laboratory-generated activated sludge, BDL: below detection limit.
3.3.
Analytical methods
All samples were filtered using 0.45-mm filters (polyethersulfone, GE Osmonics) prior to chemical analysis. DOC was detected by the catalytic combustion method at 720 C using an organic carbon analyzer (TOC-VCSH, Shimadzu Scientific Instruments). TDN was analyzed by a coupled TOCVCSH and nitrogen analyzer (TNM-1, Shimadzu Scientific Instruments) without acidification pretreatment of the samples, which intended to retain volatile nitrogen species, such as nitrous acid (pKa ¼ 3.25, Henry’s law constant ¼ 2.45 102 atm-m3/mole), and to achieve a high N recovery. Ammonia was measured via the salicylate method by a continuous-flow wet chemistry analyzer (TrAAcs 800 Autoanalyzer, Bran-Luebbe). Nitrite and nitrate were analyzed by Dionex DX-120 ion chromatography. UVA254 was measured by a spectrophotometer (Shimadzu). Methods of other miscellaneous parameters (chloride, sulfate, pH, temperature, free chlorine, chloramine, oxygen, etc) were documented elsewhere (Krasner et al., 2008). The DON values reported here are based on the differential method using TDN minus DIN (Equation (1)) after dialysis pretreatment (Lee and Westerhoff, 2005). þ DON ¼ TDN DIN ¼ TDN NO 3 NO2 NH4
(1)
Due to the characteristic high DIN level (typically >5 mg/L) and DIN fraction in TDN (typically >90%, or DON/TDN <10%) in wastewater effluent, the dialysis period was extended to 48 h (versus 24 h for drinking water) to increase the separation degree of DIN from DON and minimize the subtractionmagnified error. The three analytical methods for computing DON, TDNeDIN with dialysis versus TDNeDIN without dialysis versus TKNeNH4 without dialysis, were compared in the SI.
4.
Results and discussion
4.1.
Survey of biologically treated effluent
4.1.1.
Sample classification
Typical wastewater effluent organic matter (EfOM) consists of refractory natural organic matter (NOM) originating from
drinking water (Fox et al., 2001), soluble microbial products produced by bacteria and algae growth and decay (Rittmann et al., 1987), and synthetic organic chemicals of anthropogenic heritage (Daughton and Ternes, 1999). DOC and DON are two measures of EfOM. The performance of biological treatment processes in WWTPs is determined by many factors, such as sludge retention time, aeration intensity, organic loading, and mixed liquor concentration (Rittmann and McCarty, 2001). It is possible that a plant equipped with extended aeration facilities does not achieve complete nitrification as intended, and a plant that claims to use a conventional aeration using an activated sludge process may achieve partial denitrification too. Evidence existed that nitrification and denitrification can occur simultaneously (Rittmann and Langeland, 1985; Bertanza, 1997). Therefore, to better reflect the degrees of biological treatment in various types of processes, WWTP samples were classified into three groups according to the inorganic nitrogen concentration and speci ation: 1) if TDN >5 mg/L-N and NHþ 4 > NOx , the sample was binned as non-nitrified (NN); 2) if TDN > 5 mg/L-N but NHþ 4 > NOx , the sample was classified as well-nitrified (WN); and 3) if TDN < 5 mg/L-N, the sample was considered a tertiary treatment effluent (TE) that had undergone denitrification or beyond (e.g., membrane filtration, PAC, ozonation, etc). For samples collected outside of WWTPs, the waters were categorized as soil aquifer treated (SAT) samples; wastewateraffected DWTP influents (DWI) samples, and DWTP effluents (DWE) samples.
4.1.2.
DON, DIN, and TDN
Fig. 1 summarizes the levels of TDN, ammonia, nitrite, nitrate, and DON in several types of waters. The median TDN values of wastewater effluents were 24, 13, 2.8, 4.5 mg-N/L for NN, WN, TE, and SAT samples, respectively. NN samples contained the highest concentrations of ammonia (median ¼ 21.3 mg/L; average ¼ 20.7 mg/L) and negligible nitrate (median ¼ 0.3 mg/ L; average ¼ 0.75 mg/L), whereas WN effluents contained the highest nitrate concentrations (median ¼ 10.0 mg/L; average ¼ 11.7 mg/L) and the least ammonia amounts (median ¼ 0.3; average ¼ 0.8 mg/L). TDN and nitrate concentrations in SAT samples were higher than those in TE samples, because some effluents were not tertiary treated prior to recharge. Due to the dilution effect of surface water and
4645
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 6 4 1 e4 6 5 0
50
75% 50%
20 25% 10
30
10% 5%
NN(41)
WN(35)
TE(24)
SAT(20)
DWI(13) DWE(17) 25
4
20
3
15
NO3 (mg/L-N)
5
2
NN(41)
WN(32)
TE(23)
SAT(11)
DWI(13) DWE(13)
NN(40)
WN(35)
TE(24)
SAT(20)
DWI(11) DWE(15)
NN(41)
WN(35)
TE(24)
SAT(19)
DWI(13) DWE(17)
10
-
-
NO2 (mg/L-N)
10
0
0
1
5
0
0
NN(30)
WN(15)
TE(6)
SAT(0)
DWI(4)
DWE(2) 80
2.5
2.0
60
1.5
DON/TDN (%)
DON (mg/L-N)
20
+
TDN (mg/L-N)
90% 30
40
NH4 (mg/L-N)
NN: Non-nitrified effluent WN: Well-nitrified effluent TE: Tertiary effluent SAT: Soil aquifer treatment sample DWI: Drinking water plant influent DWE: Drinking water plant effluent
95% 40
1.0
40
20
.5 0
0.0
NN(41)
WN(35)
TE(24)
SAT(19)
DWI(13) DWE(17)
Fig. 1 e Occurrences of nitrogen species in full-scale water/wastewater treatment plants (note: numbers of detectable samples are shown in brackets; dotted lines represent average values).
multiple transformation mechanisms (e.g., biodegradation, photolysis, and hydrolysis) in the watershed, the TDN and nitrate contents in the intake sites of DWTP were lowered to less than 4 mg/L. DWI samples had a lower median TDN of 1.0 mg/L (average ¼ 1.44 mg/L). Nitrite occurred in most NN samples (median ¼ 1.0; average ¼ 1.1 mg/L; 30 of 42 samples contained detectable nitrite) but rarely in SAT, DWI and DWE samples. The percentage of DON in TDN ranged from <1% to 54% with a median of 6.0% (average ¼ 10.8%) for all samples collected; DIN/TDN ratios varied from 0.36 to >0.99. The percentages of DON in TDN were small (median ¼ 4.7%, average ¼ 5.3%) for NN and WN samples, but reached up to
52% for TE samples (median ¼ 13.3%; average ¼ 18.7%). This result was in line with earlier findings that DON can dominate the TDN of nitrificationedenitrification effluent (Pehlivanoglu-Mantas and Sedlak, 2006). NN and WN samples contained high DIN/TDN ratios (median ¼ 0.94, average ¼ 0.89), which justified the need for extended dialysis pretreatment to achieve accurate DON measurements. The percentage of DON in TDN for DWI and DWE (highest ¼ 54%, median ¼ 24.2%, average ¼ 25.3%) were much higher than those in WWTP effluents, showing DON to be an important or even a dominant portion of the nitrogen in wastewaterimpacted drinking water supplies. The percentages decreased from DWI(median ¼ 31.4%, average ¼ 29.3%) to
4646
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 6 4 1 e4 6 5 0
DWE (median ¼ 22.55%, average ¼ 22.3%), suggesting that in most DWTPs, DON was more subject to treatment than DIN by most physical and chemical treatment processes used within drinking water treatment plants. Overall, the dominant nitrogen species in NN effluents was ammonia, in WN effluents was nitrate/nitrite, and in TE effluents and DWTP samples was DON (SI Fig. 2).
4.1.3.
DON and DOC
Fig. 2 presents the DOC and UVA levels. DOC concentrations ranged from 0.2 to 24 mg-C/L, and the median DOC values of wastewater effluents were 10.5, 6.5, 3.1, 1.1 mg-C/L for NN, WN, TE, and SAT samples, respectively. The change in the DOC:DON ratio is an indicator of the removal selectivity of treatment processes: increase in DOC:DON ratio means that DON is preferentially removed while decrease in DOC:DON ratio means that DOC is preferentially removed. The median ratios of DOC to DON were increased from non-nitrified (NN) samples (median ¼ 9.8 mg-C/mg-N; average ¼ 12.5 mg-C/mg-N) to wellnitrified (WN) samples (median ¼ 13.8 mg-C/mg-N; average ¼ 14.7 mg-C/mg-N), implying that DON was more biodegraded than DOC under aerobic biodegradation processes. The DOC:DON ratios, however, decreased from samples in WN to those in TE and SAT, which means that DON was not preferentially removed during the tertiary treatment processes or may due to an input of DON from the release of
soluble microbial products during biomass decay (Rittmann and McCarty, 2001). In general, the median DOC:DON ratio ranged from 8 to 11 mg-C/mg-N in WWTP effluents, significantly below the ratios (median ¼ 19 mg-C/mg-N) of natural waters (Lee and Westerhoff, 2006; Dotson et al., 2008), but close to the samples (median ¼ 12.6 mg-C/mg-N) influenced by algal activity or WWTP discharges (Dotson and Westerhoff, 2009). UVA254 is an indicator of the hydrophobicity and aromatic content of organic matter. Specific UVA (SUVA), calculated as UVA254 per unit of DOC, is a parameter allowing classification of humic (e.g., >4 L/mg-m) and non-humic matter (e.g., <2 L/ mg-m) (Edzwald and Van Benschoten, 1990). SUVA was lower in NN (median ¼ 1.5 L/mg-m, average ¼ 1.5 L/mg-m) than in WN samples (median ¼ 1.8 L/mg-m, average ¼ 1.9 L/mg-m) (Fig. 2), which indicates that aerobic biodegradation favors removal of non-UVA254 absorbing organic matter. In contrast, DWTP facilities tended to remove more UV-absorbing materials than DOC. As a result, SUVA in DWI samples (median ¼ 1.8 L/mg-m, average ¼ 2.6 L/mg-m) was greater than that in effluent samples (median ¼ 1.2, average ¼ 1.8 L/mg-m).
4.2.
Increasingly, natural systems are viewed as potential “treatment systems” or “natural buffers”. Similar to the results of treatment in WWTPs, the organic matter concentrations in
25
40
95%
15
90% 75%
10
50% 25% 10% 5%
5
NN: Non-nitrified effluent WN: Well-nitrified effluent TE: Tertiary effluent SAT: Soil aquifer treatment sample DWI: Drinking water plant influent DWE: Drinking water plant effluent
DOC:DON (mg/L-C:mg/L-N)
DOC (mg C/L)
20
Transformations in DON in natural systems
0
30
20
10
0
NN(41)
W N(35)
TE(24)
SAT(21)
DW I(13) DW E(17)
NN(41)
W N(35)
TE(24)
SAT(19)
DW I(13) DW E(17)
NN(36)
W N(33)
TE(19)
SAT(19)
DW I(13) DW E(17)
10 .3
8
SUVA (mg/L-cm)
-1
UVA (cm )
.2
.1
6
4
2
0.0
0
-.1 NN(36)
W N(33)
TE(20)
SAT(18)
DW I(13) DW E(17)
Fig. 2 e Water quality changes in full-scale water/wastewater treatment plants (note: numbers of detectable samples are shown in brackets; dotted lines represent average values).
4647
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 6 4 1 e4 6 5 0
5. Laboratory testing on wastewater effluents for organic nitrogen removal 5.1.
Biodegradation
Biodegradation of DON and DOC in four effluents was studied with and without chlorination pretreatment (Fig. 4). The average percentage reduction in DON or DOC over the five-day test occurred in the following rank-order: DON without chlorination had the highest removal (39%) > DOC without
Amount of Parameter Remaining (%)
160 140 120 100 80 60 40 DON w/o Cl2
20
0
2 3 4 Biogradation Time (days)
5
6
100 80 60 40 20 DOC
DON
SUVA
0 0
20
40
60
80
100
120
Alum Dose (mg/L)
DON-Winter DOC-Winter SUVA-Winter
140
120
100
80
60
40 0.0
1
120
Amount of Parameter Remaining (%) .. .
Percent of Parameter Relative to Zero Distance (%).
140
SUVA w/ Cl2
Fig. 4 e Water quality changes during biodegradation of effluents with (w/) and without (w/o) chlorination pretreatment (note: data points are average values; error bars indicate standard deviations of multiple samples).
160 DON-Summer DOC-Summer SUVA-Summer
DON w/ Cl2 DOC w/ Cl2
DOC w/o Cl2 SUVA w/o Cl2
0
Amount of Parameter Remaining (%) ..
the SCR decreased constantly along the length of the river (Fig. 3). In summer, only 17% of DON and 37% of DOC was removed within the 14.3-mile river length. In winter, however, the DON concentration was reduced by 35% and DOC by only 27%. The reduced DOC removal in winter (median ambient water temperature ¼ 14 C) relative to in summer (median temperature ¼ 29 C) could be related to reduced microbial activity at water lower temperatures (Rittmann and McCarty, 2001). However, the elevated DON biodegradability in winter was unexpected. After investigating the processes used in the NIWWTP, we found that free chlorine (average dose ¼ 3.9 mg/l Cl2) was applied in summer but not in winter. The treatment plant discontinued the chlorination process during winter because of the low risks of human contact with wastewater. It was speculated that the chlorination pretreatment altered the biodegradability of DON. This was further investigated in laboratory biodegradation tests (see below). Reductions of 10%e20% in UVA values were observed along the river reach, but these were less than the observed reductions in DOC, leading to rising SUVA values along the river length. The observation that SUVA increases with extended biological treatment here, and with NN and WN samples above, is consistent with the framework that biodegradation favors non-aromatic carbon structures which tend to have low UVA254 (e.g., carbohydrates).
120 100 80 60 40 20
DOC
DON
SUVA
0 3.0
6.0
9.0
12.0
15.0
Distance (miles)
Fig. 3 e Water quality changes in an effluent-dominated river (Santa Cruz River) during two seasons (note: data points are average values; error bars indicate standard deviations of multiple samples).
pH-raw
pH-9.3~9.5
pH-11.3~11.5
Fig. 5 e Effects of alum coagulation (top) and lime softening (bottom) processes on water quality changes in wastewater effluents (note: data points are average values; error bars indicate standard deviations of multiple samples).
4648
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 6 4 1 e4 6 5 0
chlorination (26%) > DOC with chlorination (16%) > DON with chlorination (3%). The observation that chlorination decreases DOC and DON biodegradation in the five-day tests is consistent with the reduced degradation observed within the Santa Cruz River (above). Chlorination has many potential impacts on wastewater effluents. One of the effects of Chlorination is to oxidize organic matter into lower molecular weight organic matter with higher carboxylic acid content (Westerhoff et al., 2004; Swietlik et al., 2009). However, the dominant cause may be the fact that chlorination oxidizes proteins and amino acids to organic chloramines, some of which are relatively stable and have disinfecting capabilities (Donnermair and Blatchley, 2003), so that chlorine substituted organics are more difficult to aerobically degrade than non-substituted analogs. Biological pretreatment levels affected DOC and DON biodegradability as well (SI Fig. 3). In the absence of disinfectant, the activated sludge effluent had the highest DON (43%) and DOC (46%) removals, whereas the membrane bioreactor effluent had the smallest reductions (DON ¼ 34%; DOC ¼ 8%) with 5 day tests. In all cases, <10 percent of UVA254 was removed in all effluents, much lower than DOC and DON removals, resulting in increases in SUVA (27% in summer, 10% in winter). The observation of rising SUVA values is in line with the observations of the full-scale survey and effluent-dominated river.
consistent with data from drinking water treatment plants (Lee and Westerhoff, 2006; Dotson and Westerhoff, 2009). Dosages of 8 mg/L alum per mg/L DOC or lime softening at pH of 11.3e11.5 resulted in <25% removals of DON and DOC, indicating it is difficult to remove organic matter in these low SUVA waters. This is consistent with the premise behind the Enhanced Coagulation concept of the USEPA Disinfection/ Disinfection Byproduct rule where waters with SUVA <2 L/ mg-m can be exempt from the mandate to remove organic carbon through coagulation. SUVA itself changed little during treatment, varying by <10%, exhibiting no selectivity for UVA and non-UVA absorbing matter. For the nine full-scale wastewater-impacted DWTPs surveyed in this study, similar results were observed although it is a combined effect of coagulation/softening, filtration, chlorination, etc. The median removals of DON and DOC from plant influent to effluent were 23% and 21%, respectively. A full-scale wastewater recycling plant was equipped with lime softening and sand filtration processes. For two sampling seasons, the average removals of DOC (11%) and DON (17%) at pH of 9.5 in this plant also fell within the range of those achieved in the bench-scale tests.
5.2.
DOC concentrations decreased with increasing PAC dosages (SI Fig. 4). DOC removals were greater than DON removals by 10% on average. The percentage of UVA removal was much greater than both DOC and DON removals. Consequently, the SUVA values decreased considerably (>50%). To further quantify and
Alum coagulation and lime softening
For wastewater effluents, DON removal efficiencies were nearly equal to or slightly greater than those of DOC during alum coagulation and lime softening tests (Fig. 5), which is
5.3.
PAC adsorption
Fig. 6 e Freundlich isotherms of DOC and DON in equilibrium with PAC. AL: aerated lagoon, AS: activated sludge, ND: nitrification and denitrification, Lab-AS: laboratory-generated activated sludge.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 6 4 1 e4 6 5 0
compare the dose-dependent removal of DOC and DON, data were fit by Freundlich adsorption isotherms (Fig. 6). For DOC, the adsorption capacity (K) ranged from 2.5 to 11 (mg-C or N/g) [L/mg-PAC]1/n, and the adsorption intensity (1/n) ranged from 1.1 to 1.8. In comparison, K (from 5.8 to 65 (mg-C or N/g)[L/mgPAC]1/n;) and 1/(1/n: from 1.5 to 4.6) values for DON were more variable than Freundlich parameters for DOC across the different waters tested. This could indicate greater variability in organic nitrogen composition among water sources, because organic nitrogen is part of larger and more complex organic matter, or greater competition between nitrogen enriched and nitrogen deficient organic matter fractions. All 1/ n values were greater than unity, indicating unfavorable adsorption trends with continued loading to the PAC. qe ¼
K$Ce1=n
Acknowledgments The authors are grateful to AwwaRF (currently the Water Research Foundation) for its financial, technical, and administrative assistance (Project #2948). The comments and views detailed herein may not necessarily reflect the views of the AwwaRF, its officers, directors, affiliates, agents, or the U.S. government. Thanks go to Drs. Stuart Krasner and Mario Esparza-Soto who coordinated the sampling event, and staff in the participating plants for their support.
references
(2)
where Ce is the equilibrium concentration after experiment (mg/L); qe is the adsorbed amount of organic matter per gram of PAC at equilibrium (mg-C or N/g-PAC)and K and n are the Freundlich constants characteristic of the system. K is the adsorption capacity factor (mg-C or N/g)[L/mg-PAC]1/n, and n is the intensity factor (unitless). The 1/n values of DON and DOC decreased from the AL sample to the AS sample to the ND sample, indicating biological treatment removes non-adsorbing organic matter fractions and the efficiency of PAC improves with increased levels of biological activity, as indicated by better removal of dissolved inorganic nitrogen. Biodegradation processes tend to remove hydrophilic substances, leaving behind more hydrophobic organic matter, which happens to be more absorbable to PAC than hydrophilic organic matter (Karanfil, 2006).
6.
4649
Summary and conclusions
Based on the studies of full-scale plants, an effluent-dominated river, and bench-scale tests, our dataset provides important insights into the occurrence and treatment of dissolved organic nitrogen. The DON fractions in TDN were low (<10%) for most wastewater effluents, which justified the need for a longer dialysis pretreatment time to obtain accurate DON detection. The DON fractions in TDN were as high as 52% (median ¼ 13%) in tertiary treated effluents and 54% (median ¼ 31%) in wastewater-affected waters, indicating that DON can sometimes be a considerable or dominant part of TDN. Removal of DON from wastewater effluents via coagulation and softening processes, as part of reuse planning or protection of the environment, can be difficult. Because wastewater effluents have low SUVA values (<2 L/mg-m), they exhibit poor DON adsorption onto alum floc (coagulation) or calcium carbonate solids (lime softening). Likewise, DON exhibits low adsorption capacities onto activated carbon. In-situ biological treatment using soil systems or rivers does seem to remove part of the DON. Chlorination appears quite detrimental to the efficiency of these natural systems, possibly due to the formation of organic chloramines. Thus, as direct or unintentional potable reuse of wastewater for drinking water expands, the focus for controlling DON should focus at improved removal at WWTPs.
Allgeier, S.C., Summers, R.S., Jacangelo, J.G., Hatcher, V.A., Moll, D.M., Hooper, S.M., et al., 1996. A simplified and rapid method for biodegradable dissolved organ carbon measurement. In: Proceedings of the AWWA Water Quality Technology Conference Boston, MA. Bertanza, G., 1997. Simultaneous nitrification-denitrification process in extended aeration plants: pilot and real scale experiences. Water Science and Technology 35 (6), 53e61. Burleson, J., Peyton, G., Wh, G., 1980. Gas chromatographic mass spectrometric analysis of derived amino acids in municipal wastewater products. Environmental Science & Technology 14 (11), 1354e1359. Chen, B., Nam, S.-N., Westerhoff, P.K., Krasner, S.W., Amy, G., 2009. Fate of effluent organic matter and DBP precursors in an effluent-dominated river: a case study of wastewater impact on downstream water quality. Water Research 43 (6), 1755e1765. Daughton, C.G., Ternes, T.A., 1999. Pharmaceuticals and personal care products in the environment: agents of Subtle change? Environmental Health Perspectives Supplements 107 (S6), 907e938. Dignac, M.F., Ginestet, P., Rybacki, D., Bruchet, A., Urbain, V., Scribe, P., 2000. Fate of wastewater organic pollution during activated sludge treatment: nature of residual organic matter. Water Research 34 (17), 4185e4194. Donnermair, M.M., Blatchley, E.R., 2003. Disinfection efficacy of organic chloramines. Water Research 37 (7), 1557e1570. Dotson, A., Westerhoff, P., 2009. Occurrence and removal of amino acids during drinking water treatment. Journal AWWA 101 (9), 101e115. Dotson, A., Westerhoff, P., Chen, B., Lee, W., 2008. Organic nitrogen occurrence and Characterization. In: Disinfection ByProducts in Drinking Water. American Chemical Society, Washington, DC, pp. 274e288. Edzwald, J.K., Van Benschoten, J.E., 1990. Aluminum coagulation of natural organic matter. In: Hahn, H.H., Klute, R. (Eds.), Chemical Water and Wastewater Treatment. Springer-Verlag, Berline. Fox, P., Houston, S., Westerhoff, P., Drewes, J.E., Nellor, M., Yanko, W., et al., 2001. Soil Aquifer Treatment for Sustainable Water Reuse. Denver, Co. Heberer, T., 2002. Occurrence, fate, and removal of pharmaceutical residues in the aquatic environment: a review of recent research data. Toxicology Letters 131 (1e2), 5e17. Her, N., Amy, G., Park, H.-R., Song, M., 2004. Characterizing algogenic organic matter (AOM) and evaluating associated NF membrane fouling. Water Research 38 (6), 1427e1438. Hureiki, L., Croue, J.P., Legube, B., 1994. Chlorination studies of free and combined amino-acids. Water Research 28 (12), 2521e2531.
4650
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 6 4 1 e4 6 5 0
Karanfil, T., 2006. In: Teresa, J.B. (Ed.), Activated Carbon Adsorption in Drinking Water Treatment. Interface Science and Technology, vol. 7. Elsevier, pp. 345e373. Krasner, S.W., Westerhoff, P., Chen, B., Amy, G., Nam, S.N., Chowdhury, Z.K., et al., 2008. Contribution of wastewater to DBP formation. AwwaRF. Krasner, S.W., Westerhoff, P., Chen, B., Rittmann, B.E., Amy, G., 2009. Occurrence of disinfection byproducts in United States wastewater treatment plant effluents. Environmental Science & Technology 43 (21), 8320e8325. Lee, W., Westerhoff, P., 2005. Dissolved organic nitrogen measurement using dialysis pretreatment. Environmental Science & Technology 39 (3), 879e884. Lee, W., Westerhoff, P., 2006. Dissolved organic nitrogen removal during water treatment by aluminum sulfate and cationic polymer coagulation. Water Research 40 (20), 3767e3774. Lee, W., Westerhoff, P., Croue´, J.-P., 2007. Dissolved organic nitrogen as a precursor for chloroform, dichloroacetonitrile, N-nitrosodimethylamine, and trichloronitromethane. Environmental Science & Technology 41 (15), 5485e5490. Leenheer, J.A., Aaron, D., Westerhoff, Paul, 2007. Dissolved organic nitrogen fractionation. Annals of Environmental Science 1, 45e56. Parkin, G.F., McCarty, P.L., 1981. A comparison of the characteristics of soluble organic nitrogen in untreated and activated sludge treated wastewaters. Water Research 15, 139e149. Pehlivanoglu-Mantas, E., Sedlak, D.L., 2006. Wastewater-derived dissolved organic nitrogen: analytical methods, characterization, and effectsdA review. Critical Reviews in Environmental Science and Technology 36 (3), 261e285. Pehlivanoglu, E., Selak, D.L., 2004. Bioavailability of wastewaterderived organic nitrogen to the alga Selenastrum Capricornutum. Water Research 38, 3189e3196. Rittmann, B., McCarty, P., 2001. Environmental Biotechnology Principles and Applications. Ohio McGraw-Hill, Columbus. Rittmann, B.E., Bae, W., Namkung, E., Lu, C.J., 1987. A criticalevaluation of microbial product formation in biological
processes. Water Science and Technology 19 (3e4), 517e528. Rittmann, B.E., Langeland, W.E., 1985. Simultaneous denitrification with nitrification in single-channel oxidation ditches. Journal of the Water Pollution Control Federation 57 (4), 300e308. Servais, P., Garnier, J., Demarteau, N., Brion, N., Billen, G., 1999. Supply of organic matter and bacteria to aquatic ecosystems through waste water effluents. Water Research 33 (16), 3521e3531. Snyder, S.A., Villeneuve, D.L., Snyder, E.M., Giesy, J.P., 2001. Identification and quantification of estrogen receptor agonists in wastewater effluents. Environmental Science & Technology 35 (18), 3620e3625. Swietlik, J., Raczyk-Stanislawiak, U., Nawrocki, J., 2009. The influence of disinfection on aquatic biodegradable organic carbon formation. Water Research 43 (2), 463e473. Urgun-Demirtas, M., Sattayatewa, C., Pagilla, K.R., 2008. Bioavailability of dissolved organic nitrogen in treated effluents. Water Environment Research 80 (5), 397e406. Vandenbruwane, J., De Neve, S., Qualls, R.G., Salomez, J., Hofman, G., 2007. Optimization of dissolved organic nitrogen (DON) measurements in aqueous samples with high inorganic nitrogen concentrations. Science of the Total Environment 386 (1e3), 103e113. Westerhoff, P., Chao, P., Mash, H., 2004. Reactivity of natural organic matter with aqueous chlorine and bromine. Water Research 38 (6), 1502e1513. Westerhoff, P., Mash, H., 2002. Dissolved organic nitrogen in drinking water supplies: a review. AQUA 51 (8), 415e448. Westerhoff, P., Yoon, Y., Snyder, S., Wert, E., 2005. Fate of endocrine-disruptor, pharmaceutical, and personal care product chemicals during simulated drinking water treatment processes. Environmental Science & Technology 39 (17), 6649e6663. Zhang, Y., Love, N., Edwards, M., 2009. Nitrification in drinking water systems. Critical Reviews in Environmental Science and Technology 39 (3), 153e208.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 6 5 1 e4 6 6 0
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Sedimentation of helminth eggs in water Mita E. Sengupta a, Stig M. Thamsborg a, Thorbjørn J. Andersen b, Annette Olsen a, Anders Dalsgaard a,* a
Department of Veterinary Disease Biology, Faculty of Life Sciences, University of Copenhagen, Groennegaardsvej 15, 2, DK-1870 Frederiksberg C, Denmark b Department of Geography and Geology, Faculty of Science, University of Copenhagen, DK-1350 Copenhagen C, Denmark
article info
abstract
Article history:
Helminth parasite eggs in low quality water represent health risks when used for irrigation
Received 12 April 2011
of crops. The settling velocities of helminth eggs (Ascaris suum, Trichuris suis, and Oeso-
Received in revised form
phagostomum spp.) and wastewater particles were experimentally determined in tap water
31 May 2011
and in wastewater using Owen tubes. The settling velocities of eggs in tap water was
Accepted 16 June 2011
compared with theoretical settling velocities calculated by Stoke’s law using measure-
Available online 24 June 2011
ments of size and density of eggs as well as density and viscosity of tap water. The mean settling velocity in tap water of 0.0612 mm s1 found for A. suum eggs was significantly
Keywords:
lower than the corresponding values of 0.1487 mm s1 for T. suis and 0.1262 mm s1 for
Helminth eggs
Oesophagostomum spp. eggs. For T. suis and Oesophagostomum spp. eggs the theoretical
Water
settling velocities were comparable with the observed velocities in the Owen tubes, while it
Sedimentation
was three times higher for A. suum eggs. In wastewater, the mean settling velocity for A.
Stokes’ law
suum eggs (0.1582 mm s1) was found to be different from T. suis (0.0870 mm s1), Oeso-
Particles
phagostomum spp. (0.1051 mm s1), and wastewater particles (0.0474 mm s1). This strongly
Settling velocity
indicates that in low quality water the eggs are incorporated into particle flocs with different settling velocities and that the settling velocity of eggs and particles is closely associated. Our results document that there is a need to differentiate the sedimentation of different types of helminth eggs when assessing the quality of low quality water, e.g. for irrigation usage. The results can also be used to improve existing models for helminth egg removal. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
Freshwater sources are becoming increasingly scarcer in many parts of the world and use of wastewater has become an attractive option for conserving and expanding available water supplies. Wastewater use in agriculture has a long history and it has been estimated that about 20 million hectares of agricultural land worldwide is irrigated with treated or untreated wastewater (Jimenez and Asano, 2008). Especially when untreated wastewater is used for crop irrigation, it
poses substantial risks to human health, not only for farmers, but also for surrounding communities and consumers of the crops, in particular when crops are eaten uncooked. The major health risks from irrigation with low quality water are associated with viral, bacterial, and parasite pathogens that are usually present in untreated, partially treated, and occasionally treated wastewater (Feachem et al., 1983; Shuval et al., 1986). Helminth parasite eggs in wastewater are of particular health concern due to the high burden of helminthic diseases
* Corresponding author. Tel.: þ45 35332720; fax: þ45 35332755. E-mail address: [email protected] (A. Dalsgaard). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.06.017
4652
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 6 5 1 e4 6 6 0
in low-income and lower-middle-income countries where use of wastewater is most prominent. The most important intestinal worms are Ascaris lumbricoides (the human roundworm), Trichuris trichiura (the human whipworm), Ancylostoma duodenale and Necator americanus (the two human hookworms), and Taenia saginata and T. solium (the beef and pork tapeworms) (Feachem et al., 1983; Shuval et al., 1986). It is estimated that ascariasis globally affects more than 1.2 billion people while trichuriasis and hookworm each affect 700e800 million people (Bethony et al., 2006; de Silva et al., 2003). The helminth eggs are shed by infected humans within whom the female worms produce vast quantities of eggs (up to 200,000 per day) (Feachem et al., 1983). The eggs enter the wastewater through direct fecal input, discharge of treated and untreated sewage, and surface water runoff from agricultural lands (Kirby et al., 2003). Since helminth parasites are extremely resistant to environmental stress, a high degree of helminth egg removal is required by wastewater treatment processes if treated wastewater is to be used safely in irrigated agriculture (Stott, 2003). In its recent guidelines the World Health Organization (WHO) therefore recommends that wastewater used for irrigation should contain less than one helminth egg per liter when no other risk reduction options are available (WHO, 2006). The occurrence and concentration of helminth eggs in wastewater depend on a variety of factors, i.e. the parasite species, the number of infected persons serving as the source, and the volume and concentration of sewage (Shuval et al., 1986). The concentration of Ascaris in raw wastewater may vary from 10e100 eggs per L in endemic areas to 100e1000 eggs per L in hyperendemic areas (Kamizoulis, 2008; Mara and Sleigh, 2010). Wastewater can be treated by a variety of methods involving highly mechanized processes, from secondary and tertiary treatment, to simple waste stabilization ponds (WSP; anaerobic, facultative, and maturation). In most of these treatments helminth eggs are removed by sedimentation processes, although there is a wide variation in removal efficiency (Shuval et al., 1986). Tertiary treatment can significantly reduce helminth egg concentrations, but it is costly, uses advanced technology and requires a high level of maintenance. Such treatment technologies therefore often fail when installed in low-income countries (Kim and Stolzenbach, 2004). Waste stabilization ponds on the other hand are well-documented appropriate wastewater treatment technologies in low-income countries and have demonstrated high rates of helminth egg removal, although hookworm larvae have been found in the final pond effluent (Mara, 2004; Ellis et al., 1993; Stott et al., 1994). Among the wastewater treatment systems, WSPs have been designed specifically for helminth egg removal. A model was developed by Ayres et al. (1992) for predicting nematode egg removal as a function of hydraulic retention time, when effluent is required for restricted irrigation only. Based on the initial concentrations of nematode eggs in raw wastewater this model can be used to determine the retention time in different pond types (anaerobic, facultative and maturation) needed to treat wastewater so it meets the WHO water quality guideline value of less than one helminth egg per L (Ayres et al., 1992). Using this model it was estimated that a retention time of nine days would remove up to 99% of the eggs, but the model does not include parameters such as water depth, temperature, and
turbulence, which are all likely to influence the sedimentation rate of parasite eggs. However, series of pond types are more efficient than fewer ponds with long retention times. Since helminth eggs remain viable for several months and even years in the environment (Feachem et al., 1983), accumulated eggs in sludge and other types of sediments represent a health hazard when the sludge is handled and used as crop fertilizer. Further, resuspension of parasite eggs in sediments, e.g. in wastewater and irrigation canals, during increased water flow and disturbance events like storm water and when pumping water, will lead to increased concentration of eggs in water and associated health risks when such water is used in irrigated agriculture. However, little is known about resuspension processes of helminth eggs under different hydrologic conditions in aquatic environments. Sedimentation of particles in water is expected to follow Stokes’s law which implies that settling velocity depends on particle size (defined as spherical), difference in density between particles and water, and the viscosity of water. This is also assumed for pollen (Sosnoskie et al., 2009), bacterial cells (Wan et al., 1995), protozoa (oo)cysts (Medema et al., 1998), and helminth eggs (Shuval et al., 1986). Theoretical calculations of the settling velocities of Ascaris, Trichuris, and hookworm eggs in clean water have been made by using Stoke’s law, i.e. 0.65 m h1, 1.53 m h1, and 0.39 m h1, respectively (Shuval, 1978). However, these calculations do not take into account differences in morphologies and densities of the different types and development stages of helminth eggs and have not been confirmed experimentally. Further, high particle concentrations in wastewater typically results in flocculation of the particles (Droppo, 2001) and this may lead to attachment and entrapment of helminth eggs to these flocs affecting the settling velocity of the eggs. Flocculation generally increases settling velocity of the suspended particles due to the increase in apparent particle diameter. This has been documented for protozoan parasites attached to wastewater particles (Medema et al., 1998). However, floc density at the same time decreases when water makes up part or most of the flocs and if these have very low densities, the outcome may actually be a decrease in settling velocity. This effect of decreased settling velocity of particle flocs in water, however, has not yet been studied. The pig helminths, Ascaris suum, Trichuris suis, and Oesophagostomum spp. (from now on referred to as Ascaris, Trichuris and Oseophagostomum) are often used as model organisms for the previous mentioned human helminth parasites (Boes and Helwigh, 2000). The eggs of pig helminths are virtually identical in morphology and size to the corresponding human parasite eggs and are relative easy to obtain in high numbers from infected pigs. The conformation of Ascaris, Trichuris and Oesophagostomum eggs are very different from spherical particles (Fig. 1). The Ascaris egg is round to elliptical in shape, covered with a proteinaceous external layer which is irregularly mammillated. Trichuris eggs are lemon shaped with a knob at each pole (polar plugs) and have a smooth surface whereas the egg of Oesophagostomum is elliptical with a smooth surface (Alicata, 1935). The eggs of Ascaris and Trichuris are quoted to be sticky and adhere to surfaces and suspended matter, e.g. in wastewater (Gaspard et al., 1994; Roepstorff, 2003). A small negative
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 6 5 1 e4 6 6 0
4653
Fig. 1 e Morphology of the three helminth eggs used in the sedimentation experiments. a. Ascaris suum; b. Trichuris suis; c. Oesophagostomum spp. Not to scale. Photo ª Mita E. Sengupta.
surface charge has been reported for Ascaris eggs (Capizzi and Schwartzbrod, 2001; Dunn, 1991), but its effect on particle attachment is not clear. In the WHO guidelines (WHO, 2006), the recommendation for less than one helminth egg per L of treated wastewater is based on epidemiological evidence. However, dose-response data for A. lumbricoides are now available (Navarro et al., 2009), thus it is possible to conduct quantitative microbial risk analysis (QMRA) simulations for Ascaris to determine required minimum reduction for restricted and unrestricted irrigation (Mara and Sleigh, 2010). In the process of determining how to achieve the required helminth reduction (sedimentation, filtering, etc.), information such as settling velocities of eggs are crucial for e.g. dimensioning of settling tanks (Chancelier et al., 1998). Until now theoretical calculations of the settling velocities for parasite eggs in water as proposed by Shuval (1978) have been used, whereas a direct determination of the actual settling velocities of helminth eggs remains to be done as has been reported for protozoan parasites (Medema et al., 1998). Additionally, no data exists on the settling mechanism and velocity of helminth eggs in wastewater where some degree of attachment to particles is expected. The aim of this study therefore was to experimentally determine the settling velocity of different helminth egg types in clean water (no particles) and in wastewater (with particles), and to investigate the use of Stokes’ law as a predictive model of the settling velocity of helminth eggs in water.
2.
Materials and methods
2.1.
Recovery of helminth eggs
Eggs of Ascaris suum, Trichuris suis, and Oesophagostomum spp. (Fig. 1) were recovered from feces of naturally infected pigs in Denmark. Eggs were isolated by sieving fresh feces through a series of sieves using cold tap water. The mesh sizes used in declining order were 500, 212, 90 and 38 mm for Ascaris eggs; 500, 212, 90, 38, 35, 31.5 and 30 mm for Trichuris eggs; and 500, 212, 90, and 53 mm for Oesophagostomum eggs. These mesh sizes were used to optimize egg recovery (Jørgensen, 1978; Oksanen et al., 1990). All eggs were then collected on a 20-mm
sieve and transferred to a 500 ml glass beaker and left overnight at 5 C. The following day the liquid upper phase in the beaker was discarded and a number of centrifuge tubes were each filled with 10 ml of the egg suspension left in the beaker. Each tube was resuspended with flotation fluid (50 g glucose monohydrate/100 ml saturated NaCl solution yielding a specific gravity of 1.27 g/ml) to a total volume of 50 ml and subsequently centrifuged for 7 min at 253 g (modified from Larsen and Roepstorff, 1999; Roepstorff and Nansen, 1998). The supernatant containing the eggs was drawn off and collected in a new centrifuge tube. The resuspension, centrifugation and collection of the supernatant were repeated three times, and the eggs in the combined supernatant were collected on a 20mm mesh size sieve and subsequent washed with demineralized cold water. The eggs of Ascaris and Trichuris were removed from the sieve and stored in H2SO4 (0.05 M, pH 1, 24 eggs/ml), and Oesophagostomum eggs were stored in demineralized water (24 eggs/ml) at 5 C (modified from Eriksen, 1990).
2.2.
Characteristics of tap and wastewater
The tap water used had the following characteristics: pH, 7.15e7.45; conductivity, 81e168 mS/m; iron (Fe), <0.01e0.09 mg/L; and oxygen (O2), 7.6e10.7 mg/L (water quality characteristics according to the Frederiksberg Municipality Water Supply). The wastewater used for the sedimentation experiments was obtained from the inlet to the primary settling tanks at the wastewater treatment plant “Lynetten”, Copenhagen, Denmark. Characteristics of the wastewater used were as follows: total suspended solids, 240 mg/L; total dissolved solids, 1620 mg/L; pH 7.95; specific density, 0.9987 g/ml at 23 C. Initial analyses showed that the majority (>87%) of the particles in the wastewater was <20 mm and only a small fraction was larger than 250 mm (results not shown). Examinations of the raw wastewater by sedimentation and flotation showed no occurrence of helminth eggs.
2.3. Sedimentation of helminth eggs in tap and wastewater Sedimentation experiments in tap water were done with suspensions of 500e600 eggs of each individual parasite
4654
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 6 5 1 e4 6 6 0
whereas in wastewater the egg suspensions consisted of 500e600 eggs of each parasite together. Sedimentation experiments were done in new Owen tubes (Fig. 2) made of acrylic plastic, and with an internal diameter of 50 millimeters (Owen, 1976). According to Huisman (1995) the small diameter of the Owen tube does not interfere with sedimentation since the upward flow of displaced fluid does not significantly hinder the sedimentation if the particle is smaller than approximately 1/45 of the column diameter. Prior to a sedimentation experiment and after the termination of an experiment, the glass cylinders used for mixing tap and wastewater with parasite eggs as well as the Owen tubes were rinsed well with 0.01% Tween20 solution (Merck, Hohenbrunn, Germany) to minimize egg adhesion to the tube wall,
and to remove helminth eggs that adhered to the glass and plastic surfaces to prevent contamination with possible leftovers of eggs to subsequent experiments. Two L of tap water or wastewater kept at room temperature (21e23 C) were added the suspensions of parasite eggs and then stirred well to secure a homogeneous distribution of eggs and particles, where after the solution was poured into the Owen tubes and allowed to settle. After predetermined time intervals (in tap water: 4, 8, 15, 30, 60, 120, 240, 600, 1320, 2640 min; and in wastewater: 2, 4, 8, 16, 24, 32, 48, 64, 128, 260 min), ten 200-ml subsamples were collected from the bottom of the Owen tubes by a funnel with a clamp into 250-ml glass beakers. Each 200-ml subsample solution was transferred to centrifuge tubes and centrifuged for 7 min at 253 g. The supernatant was discarded with a vacuum pump leaving about 10 ml of egg sample suspension in the bottom of each tube. The egg suspensions from these tubes were then transferred into a single tube and each tube was rinsed with 0.01% Tween20 solution to ensure that all eggs were transferred. The combined egg solution was centrifuged and the supernatant discarded leaving 1.5 ml of combined egg suspension. Wastewater samples were further processed to remove particulate matter to ensure that eggs were not trapped in particle flocs, and hence not counted. This was done by centrifuging the single tube containing all the egg suspensions for 7 min at 253 g and discarding the supernatant leaving 5 ml of combined egg suspension. Then, 35 ml of 0.01% Tween20 solution was added and the suspension vortexed for 1 min. This solution was then poured through one layer of fine gauze (Cutisoft Cotton, BSN-Medical, Hamburg, Germany) and the filtrate collected in a glass beaker. The filtrate was transferred back to same tube and centrifuged before the supernatant was discarded as described above. The gauze was rinsed with 40 ml of 0.01% Tween20 in the glass beaker and poured into the centrifuge tube with the remaining 5 ml filtered egg suspension. The sample was centrifuged and the supernatant discarded leaving 1.5 ml of combined egg suspension as was done with the tap water samples. To each combined egg suspension obtained from the wastewater and tap water samples, 6 ml of the above mentioned flotation fluid was added and the entire volume of eggs and flotation fluid transferred with a glass Pasteur pipette to 7e8 McMaster slides. Eggs present both inside and outside the normal gridded areas of the McMaster chamber were enumerated. For each type of helminth egg, the sedimentation experiment was repeated five times in tap water as well as in wastewater.
2.4.
Fig. 2 e Schematic illustration of the sedimentation tube (Owen tube).
Sedimentation of wastewater particles
Experiments with wastewater without added helminth eggs were done to determine the sedimentation rate of the particles contained in the water. Wastewater from the same batch used for the egg sedimentation experiments was poured into a 2000 ml glass cylinder, stirred well and then poured into the Owen tube and allowed to settle. After predetermined time intervals (2, 4, 8, 16, 24, 32, 48, 64, 128, 260 min), ten 200-ml subsamples were collected from the bottom of the Owen tubes by a funnel into 250 ml glass beakers. Each subsample was filtered through filters with mesh sizes of 250 mm (polyamide, Sintab Product AB, Oxie, Sweden), 50 mm (polyamide,
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 6 5 1 e4 6 6 0
Sintab Product AB) and 0.45 mm (cellulose nitrate, Sartorius Stedim Biotech, Aubagne Cedex, France) with a vacuum pump. Prior to filtering the samples, the filters were moistured and dried in an oven for two hours at 60 C, left on the table for one hour to adjust to temperature and air humidity and weighted with a precision of 0.1 mg. After filtration of the wastewater samples, all the filters were processed and weighted as mentioned above allowing the determination of the dry weight of the particles on the filter. As controls, three sets of filters for each sedimentation experiment were processed with water and handled as described above and used to correct the weight of the filters containing wastewater particles. The sedimentation experiments with particles in wastewater were repeated five times.
2.5. Calculation of settling velocity and statistical analyses The median settling velocity of eggs and particles in wastewater was calculated using the method described by Owen (Owen, 1976). For eggs, the number of eggs in each sub sample was used in the calculation, and for particles their dry weight in each sub sample was used. The calculation of the settling velocities is based on the difference in egg numbers or particle weights in the ten sub samples obtained from one Owen tube experiment, taken at different time intervals corresponding to a known settling height in the tube (total height 1 m). Thus, one median settling velocity is obtained for each Owen tube experiment. Based on the time interval selected for wastewater sampling (see sections 2.3 and 2.4) from the Owen tubes, the lower limit for estimating the median settling velocity was 0.023 mm s1. In a few cases, more than 50% of the eggs and particles were settling with a velocity less than 0.023 mm s1 and hence a median settling velocity could not be calculated. In these cases, the settling velocity was set to 0.023 mm s1. Median settling velocities were log-transformed in order to stabilize variances. One-way ANOVA analyses were performed on the three egg types in tap water and in wastewater separately, to determine if there was a difference in settling velocity between the eggs in each water type. In wastewater, pair-wise comparisons of the settling velocity of each egg type with the particles were performed. Additionally, the settling of each egg type in tap water and wastewater were compared by T-tests.
2.6.
Theoretical settling velocity of eggs
The settling velocity of particles is anticipated to follow Stokes’ law, which is described by the following equation: Vs ¼ g=18 d rp rl h1
4655
eggs of each helminth with a Leica microscope using the software program Leica IM500 (version4.0; Leica, Cambridge, United Kingdom). The mean of length and width for each of 100 eggs was used to calculate a mean egg size of each helminth type to be used in Stoke’s equation. Confidence intervals (CI) were also calculated. The egg size distribution was also measured with laser particle analyzer (LISST 100C, Sequoia, Seattle, USA) (Agrawal and Pottsmith, 2000) by analyzing the light-diffraction of the egg solutions (0.5e1.5 million eggs in tap water) introduced into the standard 160 170 190 mm laboratory-chamber of the instrument. The egg densities (rp) were determined by density gradient centrifugation in a sucrose solution (David and Lindquist, 1982). The following sucrose solutions were used to prepare the gradients: 3% (sp. density 1.0092 g/ml), 13% (1.0551 g/ml), 22% (1.0922 g/ml), 30% (1.1316 g/ml), 35% (1.1535 g/ml), and 54% (1.2516 g/ml). Two ml of each solution was carefully layered using a Pasteur pipette in a 12-ml tube from the lightest (3%) to the heaviest (54%) sucrose solution. Five hundred parasite eggs were layered on top of the gradient and centrifuged at 800 g for 20 min and the centrifuge allowed to come to a standstill with the break set to zero. Following centrifugation, each layer in the gradient was carefully transferred to clean McMaster tubes with a Pasteur pipette and suspended in demineralized water to a total of 10 ml. The tube was then centrifuged at 253 g for 7 min and the supernatant removed with a vacuum pump leaving 0.5 ml suspension. To each egg suspension, 4 ml of the above mentioned flotation fluid was added and the total number of eggs in the entire volume was counted in McMaster chambers as described earlier. Egg specific density was calculated as weighted mean, and the mean of three determinations for each helminth egg type was used in Stokes’ equation. The specific density (rl) of tap water and wastewater were measured at 23 C with a glass pycnometer. The mean of three measurements was used in Stokes’ law. The kinematic viscosity of tap water (h) was determined at 23 C by a calibrated viscometer with a viscosity range of 0.9 to 3 mm2 s1. The dynamic viscosity (h) was calculated from the kinematic viscosity (v) by the following equation: h ¼ vr where h is dynamic viscosity (kg m1 s1), v is kinematic viscosity (m2 s1), and r is density of the medium (kg m3). The mean of three viscosity determinations was used in Stokes’ equation.
3.
Results
2
where Vs is settling velocity (m s1), g is gravitational acceleration (9.81 m s2), d is particle diameter (m), rp is specific density of the particle (kg m3), rl is specific density of the liquid (kg m3), and h is dynamic viscosity of the liquid (kg m1 s1). The parameters in Stokes’ equation were determined as follows with regards to egg size, egg density, and liquid density and viscosity. The egg sizes (d ) were determined by measuring the largest diameter (length) and smallest diameter (width) of 100
3.1. Settling velocity of helminth eggs in tap water and wastewater The results of the sedimentation experiments in tap water are shown in Fig. 3A. The mean settling velocity calculated based on the outcome of the five determinations for each egg type, was 0.0612 0.0053 mm s1 (SD) for Ascaris eggs, 0.1487 0.0602 mm s1 for Trichuris eggs, and 0.1262 0.0219 mm s1 for Oesophagostomum eggs. The settling velocity of Ascaris was significant lower (P < 0.05) than the velocity of the
4656
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 6 5 1 e4 6 6 0
Fig. 3 e Settling velocity in five experiments carried out with Ascaris (Asc), Trichuris (Tri) and Oesophagostomum (Oes) in tap water (A) and in wastewater, including natural occurring particles (B).
two other helminth egg types. There was very little variation in the five experimental repeats for Ascaris compared to the data for Trichuris and Oesophagostomum. The overall recovery of the three helminth eggs was 70e80% in tap water. The results of the sedimentation experiments with helminth eggs in wastewater are shown in Fig. 3B. The mean settling velocity calculated based on results obtained from the five determinations for each egg type, was 0.1593 0.0883 mm s1 for Ascaris eggs, 0.0866 0.0672 mm s1 for Trichuris eggs, and 0.1062 0.0875 mm s1 for eggs of Oesophagostomum. There was significant difference between the settling velocities of Ascaris and the two other helminth egg types in wastewater (P < 0.05). The wastewater particles had a mean settling velocity of 0.0474 0.0292 mm s1 which was significantly different from the settling velocity of Ascaris eggs (P < 0.05), but not significantly different from Trichuris and Oesophagostomum. Ascaris eggs had a significantly higher (P < 0.05) settling velocity compared to tap water. In contrast, Trichuris and Oesophagostomum eggs seemed to have decreased settling velocities in wastewater compared to tap water. The recovery of helminth eggs in wastewater was generally lower than in tap water: 57e80%, 54e59%, and 27e47% for Ascaris, Trichuris and Oesophagostomum, respectively.
3.2.
width of the eggs was used in Stoke’s equation as an approximation of the dynamic diameter of helminth eggs. The mean size used for Ascaris eggs was 61.31 3.58 mm, for Trichuris eggs 46.47 2.03 mm, and for Oesophagostomum eggs 63.25 1.11 mm. A random orientation of the eggs was reflected in the egg size distribution, as measured by the LISST laser particle analyzer (Fig. 5), showing a very narrow peak for close-to-spherical Ascaris eggs and a wider distribution of Oesophagostomum and Trichuris with ellipsoid eggs. Densities for Ascaris, Trichuris and Oesophagostomum eggs were 1.12 0.007 kg m3, 1.10 0.010 kg m3, and 1.07 0.002 kg m3, respectively. In tap water, the mean specific density was 0.9978 0.0011 kg m3; the mean kinematic viscosity was 0.9424 0.0068 mm2 s1 and the dynamic viscosity was calculated as 0.000940 Pa s. The theoretical settling velocity of the helminth eggs was calculated using the above mentioned equation parameters in Stokes’ law, and is shown in Table 2. The theoretical settling velocity for Trichuris (95% CI, 0.1270e0.1314) shows overlap in confidence intervals with the experimentally observed velocity (95% CI, 0.0959e0.2015). However, for Ascaris the theoretical settling velocity (95% CI, 0.2686e0.2813) was three times higher than the observed settling velocity (95% CI, 0.0565e0.0658) and for Oesophagostomum the theoretical
Theoretical settling velocity
To allow for a comparison of the experimental and theoretical (Stokes’ equation) settling velocities, the values of the equation parameters (egg size and density; liquid density and viscosity; and settling velocity) and their variations were determined for the conditions used in the experiments. The measured eggs sizes (width, length and mean diameter) together with the mean eccentricity (ratio between length and width) are listed in Table 1. Overall, Trichuris had the smallest eggs and Oesophagostomum the largest sized eggs. None of the eggs were spherical (eccentricity of 1) which is assumed by Stokes’ law for calculating the settling velocity, nevertheless Ascaris eggs were more spherical compared to eggs of Trichuris and Oesophagostomum (Fig. 4). Since none of the eggs are spherical, it is likely that they settle with different but unknown orientations due to Brownian motions in the Owen sedimentation tube. Thus, the mean of length and
Table 1 e Measured sizes of helminth eggs used in the experiments. Parameter
Egg size (mm) Mean (SD)
Min
Max
Ascaris Length Width Eccentricity
67.20 (5.33) 55.41 (3.91) 1.22 (0.12)
52.16 46.80 1.00
84.07 64.64 1.80
Trichuris Length Width Eccentricity
62.16 (2.60) 30.78 (2.72) 2.03 (0.16)
54.25 26.67 1.56
68.02 38.29 2.38
Oesophagostomum Length Width Eccentricity
76.17 (5.52) 50.33 (6.63) 1.53 (0.15)
62.74 38.35 1.24
85.67 60.23 1.99
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 6 5 1 e4 6 6 0
Fig. 4 e Egg size distribution of helminth eggs. The solid line represents an eccentricity of 1 (ratio between length and width).
(95% CI, 0.1504e0.1613) was only slightly higher than the experimental settling velocity (95% CI, 0.1069e0.1454).
4.
Discussion
The results of the sedimentation experiments in tap water showed that Ascaris eggs had the slowest settling velocity compared to the settling velocities observed for Trichuris and Oesophagostomum eggs. Our findings are different when compared with the theoretical calculations made by Shuval (1978) for human helminth eggs which estimated that Trichuris eggs (0.425 mm s1) settle the fastest, followed by Ascaris eggs (0.181 mm s1) with hook worm eggs (0.108 mm s1) settling with the slowest velocity. As also documented in our study, Stokes’ law (see formula above) has some assumptions and limitations (Gordon et al., 2004). A falling particle is assumed to be spherical and have a smooth surface, which is generally not the case for helminth eggs (see Fig. 1 and Table 1). Despite the non-spherical shape of
Fig. 5 e The egg size distribution measured with LISST 100C laser particle analyzer. Note the log-scale on X-axis.
4657
helminth eggs, our predicted settling velocities for Trichuris and Oesophagostomum eggs showed good agreement with the observed velocities found in the Owen tube experiments (Table 2). For Ascaris eggs, the theoretical settling velocity was much higher than the observed. This could be explained by the fact that the surface of Ascaris eggs are mammillated and not smooth as is the surface of the other two eggs (Fig. 1). Stokes’ law is only valid for laminar flow (fluid layers gliding over one another) which is seen only for particles with smooth surfaces. Thus, the mammillated surface structure of Ascaris eggs may cause a turbulent flow so the fluid layers move in random irregular paths (Gordon et al., 2004) which decrease the settling velocity. Other studies (e.g. Sosnoskie et al., 2009) support our findings that a turbulent flow around particles will lead to an overestimation of settling velocity using Stokes’ law. Variations in egg density and egg size have a major impact on settling velocity as calculated by Stokes’ law. In our study, the settling velocities of the different egg types was calculated using measured values of egg density, egg size and liquid density and viscosity. Settling velocities obtained were 0.2749 mm s1 for Ascaris, 0.1292 mm s1 for Trichuris and 0.1581 mm s1 for Oesophagostomum (Table 2). These settling velocities are higher for Ascaris and Oesophagostomum eggs and lower for Trichuris eggs than the settling velocities calculated by Shuval (1978) who used density and size information on the corresponding human parasites (Faust et al., 1970). The egg densities used by Shuval (1978) were similar to the densities measured in this study, only Shuval (1978) found that Trichuris eggs had higher density (1.15 kg m3). David and Lindquist (1982) reported egg densities of the three helminth egg types collected from pigs and dogs (Ancylostoma caninum) that were similar to our findings, but found a higher density of Trichuris (1.13 kg m3) eggs. It is unknown if there exist host-dependent variations in density of different helminth egg types. However, it is documented that helminth eggs in the environment exhibit individual density variations depending on age or stage of development, e.g. degree of embryonation (Magat et al., 1972; Sawitz, 1942). We were surprised to find a lower mean density for Trichuris than Ascaris eggs, as it is widely recognized that the former is more difficult to detect and count in the laboratory using standard flotation techniques. The difficulties in floating Trichuris eggs may be associated with a much larger variation in egg density as compared to Ascaris and Oesophagostomum eggs as suggested by our findings that a fraction of 5e10% showed a density above 1.25 kg m3 (data not shown). If a proportion of the eggs do not float in standard flotation fluids, e.g. saturated salt solutions, this will inevitably lead to an underestimation of egg numbers. The egg sizes that Shuval (1978) used (A. lumbricoides, 55 40 mm; T. trichiura, 50 22 mm; hookworm 60 40 mm) for calculating settling velocities are considerably smaller than the sizes measured in this study which was based on 100 measurements of individual eggs of Ascaris, Trichuris and Oesophagostomum (Table 1). In our calculations of settling velocities we used the mean of length and width of the eggs as an approximation of the dynamic diameter of the eggs. Shuval (1978) in contrast did not describe which egg diameters he used in Stokes’ equation. The differences in egg densities, especially for Trichuris eggs, and sizes for all three egg types, result in marked
4658
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 6 5 1 e4 6 6 0
Table 2 e Experimental and theoretical (using Stokes’ law) mean settling velocities for helminth eggs in tap water. Experimental settling velocity (mm s1) [m h1]
Ascaris Trichuris Oesophagostomum
Theoretical settling velocity (mm s1) [m h1]
Mean
95% CI
Mean
95% CI
0.0612 [0.22] 0.1487 [0.54] 0.1262 [0.45]
0.0565e0.0658 0.0959e0.2015 0.1069e0.1454
0.2749 [0.99] 0.1292 [0.47] 0.1581 [0.57]
0.2686e0.2813 0.1270e0.1314 0.1504e0.1613
differences in egg settling velocities as demonstrated in this study and by Shuval (1978). Our Owen tube sedimentation experiments showed that Trichuris and Oesophagostomum eggs had decreased settling velocities in wastewater, whereas the velocity of Ascaris eggs increased compared to the settling velocities found for eggs in tap water (Fig. 3). Trichuris and Oesophagostomum eggs were found to have similar mean settling velocity as the particles in the wastewater where as Ascaris eggs settled faster. It should be noted that the range of velocities of the eggs where broader than for the particles. The decreased settling velocity of Trichuris and Oesophagostomum eggs is likely due to entrapment of eggs to the flocculated particles. The entrapment of eggs may also explain the lower egg recovery in wastewater than in tap water. Unfortunately, it was impossible to determine the size distribution of the particle flocs with the LISST particle analyzer due to high water turbidity. The very low settling velocity measured for flocculated particles in wastewater as compared to settling velocities reported in saline water indicates that the water content of the wastewater flocs was very high (Gibbs, 1983). The similar settling velocity of wastewater particles and Trichuris and Oesophagostomum eggs therefore strongly suggests that the eggs were incorporated into flocs with low density. This means that the behavior and settling velocity of the particles in wastewater determines the settling velocity of these two helminth egg types. It is generally anticipated that particle flocs settle faster than individual suspended particles (Kim and Stolzenbach, 2004). The increased settling velocity of Ascaris eggs is most likely due to their adhesion to particles which then increase the floc size and hence enhances the settling velocity. We can only speculate on the reason for the different behavior of these eggs, but it could be that the mammilated surface of the Ascaris egg (and the higher density of these eggs) causes entrapment in other types of particle flocs that settle faster than the smooth eggs of Trichuris and Oesophagostomum. This particular aspect could be the subject for future studies on settling of Ascaris eggs. Helminth eggs from Trichuris and Oesophagostomum entrapped in flocs settled slower compared to eggs in tap water. It should be noted that Stokes’ law assumes that eggs settle in isolation (Gordon et al., 2004) which seems to only occur in clean water. Thus, because of the observed interaction between eggs and other particles, our study demonstrates that Stokes law is only suitable for calculating settling velocities of helminth eggs in clean water types, e.g. drinking water, and not in low quality water like wastewater. Aquatic recipients contaminated with wastewater containing helminth eggs, e.g. rivers used as sources of irrigation water, and waste stabilization ponds do also often have a high concentration of suspended particles resulting in
flocculation of the suspended material, including the eggs. It is therefore likely that helminth eggs in different types of surface waters do not settle as single entities but rather as part of flocs consisting of fine-grained suspended material (Droppo, 2001). Future studies should assess how different types of wastewater (e.g. treated/un-treated, composition of organic materials, particle concentration and particle size) influence the flocculation and settling of helminth eggs. Such studies should also consider the water salinity as it is welldocumented that flocculation increases significantly in saline waters (Gibbs, 1983). Changes in concentration of suspended material are expected to increase settling velocity with increasing concentration of such material leading to more flocculation (van Leussen, 1999). Variations in other particle parameters can also be expected to significantly influence flocculation in complex, but poorly understood associations (Droppo, 2004). Our observed low settling velocities of the flocculated suspensions, including eggs, in low quality water are unlikely to result in any significant sedimentation in most aquatic habitats where turbulence events caused by water flow, wind, rain, temperature, and human disturbances seem to be more important than gravitational settling in determining the sedimentation of eggs in water. When low quality water, e.g. treated wastewater, is assessed for its safe use for irrigation of crops, Ascaris is usually used as the main indicator organism due to the long survival time of the eggs and low infectious dosage compared with other parasite, bacterial and viral pathogens (Feachem et al., 1983). As the major factor determining helminth egg removal is sedimentation, the efficiency of different aquatic environments, including wastewater treatment systems, in removing eggs should be based on the helminth egg type with the lowest settling velocity. This study clearly documents that in clean water Ascaris eggs should be used as an indicator of fecal pollution with helminth eggs due to its slow settling velocity (Fig. 3A), whereas in low quality water like wastewater Ascaris eggs settle with the fastest velocity and this egg type is therefore not a good indicator of the removal rate of other helminth types by sedimentation (Fig. 3B). Trichuris and Oesophagostomum eggs settle slower and there is a need to differentiate the type of helminth eggs when assessing the quality of low quality water, e.g. when implementing the WHO guidelines of less than one helminth egg per liter (WHO, 2006). Our findings that Trichuris and Oesophagostomum eggs settle with the same velocity as wastewater particles indicates that measurements of particle settling velocity in treated low quality water, e.g. in waste stabilization ponds, may be used as an indicator of helminth egg settling velocity. Further studies are needed to document how these settling velocities may be used to model helminth
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 6 5 1 e4 6 6 0
egg removal. It is also proposed that existing models of helminth egg removal, such as the model of Ayres et al. (1992), incorporate our findings to refine the models.
5.
Conclusion
Our study documents that not all helminth eggs settle in clean water according to Stokes’ law which sometimes overestimates the settling velocity. This must be taken into consideration when Stokes’ law is used to model the helminth egg removal in water, e.g. when designing water treatment facilities. Furthermore, the settling velocity and behavior of Trichuris and Oesophagostomum eggs in water is determined by the presence of particles in water, whereas Ascaris eggs show a faster settling velocity. Therefore further studies are needed to assess how different types of particles and concentrations of suspended materials in different types of low quality water influence the flocculation and settling of helminth eggs. Additionally, at which concentration of suspended materials (ranging from clean water to wastewater) Stokes’ law becomes inadequate as a predictive model must be explored. A main implication of our findings is that there is a need to differentiate between the settling velocities of different helminth egg types when assessing the safety of low quality water, e.g. for use in irrigation.
Acknowledgements This study received financial support from the “Safe and High Quality Food Production using Low Quality Waters and Improved Irrigation Systems and Management” project (SAFIR, EU, FOOD-CT-2005-023168) funded by the European Commission and the Faculty of Life Sciences at the University of Copenhagen through the research school RECETO. We greatly appreciate Lise-Lotte Christiansen and Kurt Madsen for their immense help in constructing the sedimentation and filtration equipment for this study.
references
Agrawal, Y.C., Pottsmith, H.C., 2000. Instruments for particle size and settling velocity observations in sediment transport. Marine Geology 168 (1e4), 89e114. Alicata, J.E., 1935. Early developmental stages of nematodes occuring in swine. Technical Bulletin 489, 1e96. U.S. Department of Agriculture, Washington DC. Ayres, R.M., Alabaster, G.P., Mara, D.D., Lee, D.L., 1992. A design equation for human intestinal nematode egg removal in waste stabilization ponds. Water Research 26 (6), 863e865. Bethony, J., Brooker, S., Albonico, M., Geiger, S.M., Loukas, A., Diemert, D., Hotez, P.J., 2006. Soil-transmitted helminth infections: ascariasis, trichuriasis, and hookworm. Lancet 367 (9521), 1521e1532. Boes, J., Helwigh, A.B., 2000. Animal models of intestinal nematode infections of humans. Parasitology 121, 97e111. Capizzi, S., Schwartzbrod, J., 2001. Surface properties of Ascaris suum eggs: hydrophobic potential and Lewis acidebase
4659
interactions. Colloids and Surfaces B: Biointerfaces 22 (2), 99e105. Chancelier, J.P., Chebbo, G., Lucas-Aiguier, E., 1998. Estimation of settling velocities. Water Research 32 (11), 3461e3471. David, E.D., Lindquist, W.D., 1982. Determination of the specific gravity of certain helminth eggs using sucrose density gradient centrifugation. The Journal of Parasitology 68 (5), 916e919. de Silva, N.R., Brooker, S., Hotez, P.J., Montresor, A., Engels, D., Savioli, L., 2003. Soil-transmitted helminth infections: updating the global picture. Trends in Parasitology 19 (12), 547e551. Droppo, I.G., 2004. Structural controls on floc strength and transport. Canadian Journal of Civil Engineering 31 (4), 569e578. Droppo, I.G., 2001. Rethinking what constitutes suspended sediment. Hydrological Processes 15, 1551e1564. Dunn, A.J., 1991. The development of a predictive model for the removal of helminth eggs during rapid sand filtration. PhD thesis, Faculty of Engineering and Applied Science, University of Southampton, Southampton, UK. Ellis, K.V., Rodrigues, P.C.C., Gomez, C.L., 1993. Parasite ova and cysts in waste stabilization ponds. Water Research 27 (9), 1455e1460. Eriksen, L., 1990. Ascaris suum: Influence of egg density on in vitro development from embryonated egg to infective stage. Acta Veterinaria Scandinavica 31 (4), 489e491. Faust, E.C., Russel, P.F., Jung, R.C., 1970. Craig and Faust’s Clinical Parasitology. Lea & Febiger, Philadelphia, pp. 1e890. Feachem, R.G., Bradley, D.J., Garelick, H., Mara, D.D., 1983. Sanitation and Disease: Health Aspects of Excreta and Wastewater Management. John Wiley & Sons Edition, New York. Gaspard, P.G., Wiart, J., Schwartzbrod, J., 1994. Experimental study of the helminth eggs adhesion (Ascaris suum): analysis of the environmental implications. Revue des Sciences de L’Eau 7, 367e376. Gibbs, R.J., 1983. Coagulation rates of clay minerals and natural sediments. Journal of Sedimentary Petrology 53 (4), 1193e1203. Gordon, N.D., McMahon, T.A., Finlayson, B.L., Gippel, C.J., Nathan, R. J., 2004. In: Anonymous (Ed.), Stream Hydrology, An Introduction for Ecologists. John Wiley & Sons, Chichester, pp. 127e168. Huisman, L., 1995. Sedimentation and flotation and mechanical filtration. 1e166. Jimenez, B., Asano, T., 2008. In: Anonymous (Ed.), Water reuse: An International Survey of Current Practice, Issues and Needs. IWA Publishing, London,. Jørgensen, R.J., 1978. Isolation of Ascaris suum eggs for experimental purposes. Acta Veterinaria Scandinavica 19, 147e149. Kamizoulis, G., 2008. Setting health based targets for water reuse (in agriculture). Desalination 218 (1e3), 154e163. Kim, A.S., Stolzenbach, K.D., 2004. Aggregate formation and collision efficiency in differential settling. Journal of Colloid and Interface Science 271 (1), 110e119. Kirby, R.M., Bartram, J., Carr, R., 2003. Water in food production and processing: quantity and quality concerns. Food Control 14 (5), 283e299. Larsen, M.N., Roepstorff, A., 1999. Seasonal variation in development and survival of Ascaris suum and Trichuris suis eggs on pastures. Parasitology 119, 209e220. Magat, W.J., Hubbard, W.J., Jeska, E.L., 1972. Ascaris suum: quantitative chemical analysis of eggs and larvae. Experimental Parasitology 32 (1), 102e108. Mara, D.D., 2004. Domestic Wastewater Treatment in Developing Countries. Earthscan Publications, London. Mara, D., Sleigh, A., 2010. Estimation of Ascaris infection risks in children under 15 from the consumption of wastewaterirrigated carrots. Journal of Water & Health 8 (1), 35e38. Medema, G.J., Schets, F.M., Teunis, P.F., Havelaar, A.H., 1998. Sedimentation of free and attached Cryptosporidium oocysts and Giardia cysts in water. Applied and Environmental Microbiology 64 (11), 4460e4466.
4660
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 6 5 1 e4 6 6 0
Navarro, I., Jime´nez, B., Cifuentes, E., Lucario, S., 2009. Application of Helminth ova infection dose curve to estimate the risks associated with biosolid application on soil. Journal of Water & Health 7 (1), 31e44. Oksanen, A., Eriksen, L., Roepstorff, A., Ilsoe, B., Nansen, P., Lind, P., 1990. Embryonation and infectivity of Ascaris-suum eggs e a comparison of eggs collected from worm uteri with eggs isolated from pig feces. Acta Veterinaria Scandinavica 31 (4), 393e398. Owen, M.W., 1976. Determination of the Settling Velocities of Cohesive Muds. IT 161. Hydraulics Research Station, Wallingford, England, pp. 1e8. Roepstorff, A., 2003. Ascaris suum in pigs: Population Biology and Epidemiology. Danish Centre for Experimental Parasitology, The Royal Veterinary and Agricultural University, Copenhagen, pp. 1e113. Roepstorff, A., Nansen, P., 1998. Epidemiology, Diagnosis and Control of Helminth Parasites of Swine. Animal Health Manual no 3. Food and Agriculture Organization of the United Nations, Rome, Italy, pp. 1e161. Sawitz, W., 1942. The buoyancy of certain nematode eggs. The Journal of Parasitology 28 (2), 95e102. Shuval, H.I., Adin, A., Fattal, B., Rawitz, E., Yekutiel, P., 1986. Wastewater Irrigation in Developing Countries: Health Effects and Technical Solutions. World Bank, Washington DC.
Shuval, H.I., 1978. Parasitic disease and waste-water irrigation. In: Pacey, A. (Ed.), Sanitation in Developing Countries. John Wiley & Sons, Chichester, pp. 210e223. Sosnoskie, L.M., Webster, D.D., Rains, G.C., Grey, T.L., Culpepper, A.S., 2009. Pollen grain size, density and settling velocity for Palmer Amaranth (Amaranthus palmeri). Weed Science 57, 404e409. Stott, R., Ayres, R., Lee, D., Mara, D., 1994. An Experimental Evaluation of Potential Risks to Human Health from Parasitic Nematodes in Wastewater Treated in Waste Stabilization Ponds and Used for Crop Irrigation. Research Monograph No. 6. Department of Civil Engineering, University of Leeds, Leeds, England, pp. 1e73. Stott, R., 2003. Fate and behaviour of parasites in wastewater treatment systems. In: Mara, D., Horan, N. (Eds.), The Handbook of Water and Wastewater Microbiology. Academic Press, London, pp. 491e521. van Leussen, W., 1999. The variability of settling velocities of suspended fine-grained sediment in the Ems estuary. Journal of Sea Research 41 (1e2), 109e118. Wan, J.M., Tokunaga, T.K., Tsang, C.F., 1995. Bacterial sedimentation through a porous-medium. Water Resources Research 31 (7), 1627e1636. WHO, 2006. Guidelines for the safe use of wastewater, excreta and greywater. In: Wastewater Use in Agriculture, vol. 2. World Health Organization, Geneva, p. 222.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 6 6 1 e4 6 7 1
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Characterization of the size-fractionated biomacromolecules: Tracking their role and fate in a membrane bioreactor Fangang Meng a,*, Zhongbo Zhou a, Bing-Jie Ni b, Xing Zheng c, Guocheng Huang a, Xiaoshan Jia a, Shiyu Li a, Ya Xiong a, Matthias Kraume d a
School of Environmental Science and Engineering, Sun Yat-sen University, Guangzhou 510275, PR China Department of Environmental Engineering, Technical University of Denmark, Miljøvej Building 113, 2800 Kongens Lyngby, Denmark c Department of Water Quality Control, Technische Universita¨t Berlin, Sekr. KF4, Straße des 17, Juni 135, 10623 Berlin, Germany d Chair of Chemical and Process Engineering, Technische Universita¨t Berlin, Str. des 17. Juni 135, MA 5-7, 10623 Berlin, Germany b
article info
abstract
Article history:
This article presents a study aimed at the fractionation and characterization of what is
Received 19 April 2011
thought to be one of the most complex organic mixtures produced by activated sludge: bio-
Received in revised form
macromolecules (BMM). Photometric quantification combined with excitationeemission
18 June 2011
matrix (EEM) fluorescence spectroscopy and nuclear magnetic resonance (NMR) measure-
Accepted 21 June 2011
ments were used to characterize BMM in a membrane bioreactor (MBR) from a chemical
Available online 26 June 2011
perspective. Overall, the BMM in sludge supernatant were mainly present in three fractions:
Keywords:
molecular weight (MW) fraction (<5 kDa). The analysis of fluorescence regional integration
Membrane bioreactors (MBRs)
(FRI) showed that the organics in membrane permeate and those in the low-MW fraction of
Membrane fouling
sludge supernatant were of similar chemical composition. The characterization by NMR
Biopolymers
suggested that the BMMc fraction had similar carbon content of proteins and poly-
Soluble microbial products (SMP)
saccharides. In contrast, the BMMb and the low-MW BMM were proved to be carbonaceous
Extracellular polymeric substances
and aromatics, respectively. Moreover, because of the high MW and gelling property, poly-
(EPS)
saccharides were found to have a high potential to accumulate on the membranes. In
colloidal BMM (BMMc, >0.45 mm), biopolymeric BMM (BMMb, 0.45 mme100 kDa) and low
addition, the lipids present in the BMMb of the sludge supernatant were demonstrated to be another important foulant due to their large size. Our results also indicated that aromatic proteins had a higher fouling propensity than tryptophan proteins though they were of similar size nature. This work could be useful for better understanding of the chemical nature of BMMs in MBRs. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
Biomacromolecules (BMM) are a pool of complex mixture produced by microorganisms as a result of substrate metabolism and biomass growth/decay during biological wastewater treatment (Laspidou and Rittmann, 2002; Ni et al., 2010). The major chemical constituents of BMM are proteins,
polysaccharides, humic substances, low molecular weight organic acids, etc. The generation and presence of BMM in wastewater treatment systems are one of the major reasons determining the quality of the treated water. In particular, BMM are believed to the primary compounds leading to membrane fouling in membrane-based technologies used for water and wastewater treatments such as membrane bioreactors (MBRs)
* Corresponding author. Tel.: þ86 20 39335060. E-mail address: [email protected] (F. Meng). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.06.026
4662
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 6 6 1 e4 6 7 1
(Huang et al., 2000; Fan et al., 2006; Jarusutthirak and Amy, 2006; Ramesh et al., 2007). In MBRs, the BMM compounds larger than membrane pore size can be rejected, in which biodegradable BMM can be used as substrates for microorganisms, and the small-size ones can be discharged along with membrane permeate (Song et al., 2007; Meng et al., 2009a). Finally, the refractory and large-size BMM compounds are accumulated in the bioreactor or on membranes. Considering the significance of BMM in MBRs, numerous research efforts have been made in recent years (Meng et al., 2009b; Drews, 2010). The fouling propensity of BMM strongly depended on their size (Laabs et al., 2006; Rosenberger et al., 2006), composition (Chen et al., 2006; Rosenberger et al., 2006; Drews et al., 2008; Ng et al., 2006; Ng and Ng, 2010), and hydrophobic/hydrophilic nature (Liang et al., 2007; Yamamura et al., 2008; Tian et al., 2009). For instance, the hydrophilic substances with a molecular size above approx. 120 kDa, called biopolymers, were found to be the predominant foulants in MBRs (Rosenberger et al., 2006). Similarly, Liang et al. (2007) also showed that hydrophilic neutrals (e.g., polysaccharides) were likely responsible for the high fouling potential of BMM. Using a field flow fractionation (FFF), it was found that the lower molar-mass compounds in bacterial extracellular polymeric substances was dominated by protein-like substances, whereas the higher molar-mass fraction was rich in exoproteins and exopolysaccharides (Alasonati and Slaveykova, 2011). Based on the fractionation with centrifugation, Teychene et al. (2008) found that the MBR fouling is mainly governed by soluble BMM rather than colloidal fraction. Nonetheless, opposite conclusions were drawn as well. Lee et al. (2003), for example, reported that the hydrophobic proteins played an important role in membrane fouling due to their hydrophobic interactions with membranes. Indeed, hydrophobic and rough membranes have higher potential to be fouled by BMM than hydrophilic and smooth ones (Zhang et al., 2008). Such contradictions in previous research efforts are mostly due to the poor understanding of BMM, e.g., the BMM in previous studies may have different chemical or structural composition. In order to better understand the membrane fouling in MBRs, it is essential to characterize the BMM more comprehensively. In fact, a number of characterization methods focussing on component or size fractionation have been used in numerous studies (Leenheer, 1981; Liang et al., 2007). For example, XAD resin was used to separate the BMM into hydrophobic, hydrophilic, and transphilic constituents (Leenheer, 1981), which can aid in understanding the BMM in terms of hydrophobic/hydrophilic nature. Using a field flow fractionation (FFF), it was found that the lower molar-mass compounds in bacterial extracellular polymeric substances was dominated by protein-like substances, whereas the higher molar-mass fraction was rich in exoproteins and exopolysaccharides (Alasonati and Slaveykova, 2011). Based on the fractionation with centrifugation, Teychene et al. (2008) found that the MBR fouling is mainly governed by soluble BMM rather than colloidal fraction. In addition, the BMM can also be fractionated by filtering the samples through a series of membranes with different molecular weight cut-offs (MWCOs). Unfortunately, the characterization of the size-fractionated BMM was currently limited to the quantification of dissolved organic
carbon (DOC) (Liang et al., 2007), which can only provide rough information about the chemical nature of BMM. Indeed, a detailed characterization of the size-fractionated BMM can lead to a deeper understanding on BMM compounds, which can hopefully clear up the misunderstanding dealing with the contradictions in previous studies. More importantly, it is of high interest to track the origins of membrane foulants and the fate of BMM in the MBR process by an in-depth characterization of BMM. To date, however, only a few investigations have been conducted to identify the true chemical composition of the size-fractionated BMM. The objective of this study, therefore, was to characterize the individual chemical components of the size-fractionated BMM in order to better understand the membrane fouling in MBRs. The size fractionation was achieved by filtering samples with a series of membranes with different MWCOs. Different BMM components were identified and quantified using photometric methods. Furthermore, excitationemission matrix (EEM) fluorescence spectroscopy and 13C nuclear magnetic resonance (NMR) were used to reveal the chemical/structure nature of these BMM components.
2.
Materials and methods
2.1.
The MBR setup
A lab-scale MBR with an effective volume of 50 L was used in this work, which consisted of anoxic, aerobic, anoxic, anaerobic, and membrane tank (see Fig. S1 of Supporting Materials (SM)). The MBR was designed for simultaneous removal of organic carbon, nitrogen and phosphorous. A 0.23 m2 flatsheet membrane module (PVDF, 0.1 mm, Sinap Corp., Shanghai, China) was submerged in the membrane tank. This membrane is commercially used for industrial and municipal wastewater treatments particularly in China. The membrane module was constantly sucked with a peristaltic pump, and physical or chemical backwashing was not performed. As the transmembrane pressure (TMP) reached 0.025 MPa, the membrane module was washed with high pressure water and then was soaked in 0.3% NaClO solution for about 12 h. A complex synthetic wastewater simulating municipal wastewater was used as feedwater (see Table S1 of SM). The COD, TN, and TP in the feedwater were in the range of 310e400, 62e81, and 8e15 mg/L, respectively. Hydraulic retention time (HRT) and solid retention time (SRT) were set at 12e14 h and 20 days, respectively. Concentrations of the mixed liquid suspended solids (MLSS) in the membrane tank were in a range of 6000e7000 mg/L. Weekly, the BMM of feedwater, sludge supernatant and membrane permeate were sampled and measured.
2.2.
Sampling of membrane foulants
Due to the fact that the use of chemicals such as acids or alkaline solution may damage the original composition of membrane foulants, the membrane foulants were obtained by washing the fouled membrane module with high pressure water. Thus, only the removable compounds on the fouled membranes were obtained and analyzed. In fact, the foulants
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 6 6 1 e4 6 7 1
in MBRs are mainly dominated by cake layer (i.e., removable fouling) (Lee et al., 2001; Meng et al., 2007), thus, analysis of these removable compounds are representative. To have comparable data regarding the concentration of BMM compounds obtained at different times, the washed liquor was adjusted to 10 L every time with pure water.
2.3.
Fractionation of BMM compounds
The BMM compounds in sludge supernatant, membrane foulants and membrane permeate were all fractionated and analyzed in different size ranges. Strictly speaking, only the organics in soluble form can be considered as BMM. In the MBR study, researchers usually prepare the BMM solution by removing the sludge flocs with a filter paper (Lesjean et al., 2005; Drews et al., 2007), which includes both colloidal and soluble compounds. Probably, due to the fact that the colloids also play an important role in membrane fouling, the BMM solution was prepared by filtering the mixed liquor through a filter paper (10 mm). The BMM solution was subsequently filtered through a series of membranes with different MWCOs (PVDF material, 0.45 mm, 100 kDa, 30 kDa, 5 kDa). The filtration was performed in a stirred dead-end membrane filtration cell (MSC300, Mosu corp., Shanghai, China). Here, the BMM fraction in the range of 0.45e10 mm, and 100 kDae0.45 mm were defined as colloidal BMM (BMMc), and biopolymeric BMM (BMMb), respectively. It is well-known that the fouling-causing substances in most MBRs are usually dominated by organics rather than inorganics. This work was therefore focused on the analysis of compounds in organic form, and the dissolved inorganic matter such as silica was not studied.
2.4.
Photometric quantification of BMM
Concentrations of polysaccharide and protein were determined by photometric methods according to Dubois et al. (1956) and Lowry et al. (1951), respectively. The influence of nitrate and nitrite on polysaccharide measurements was corrected according to Drews et al. (2007). Humic substances were characterized by the UV254 (UNICO, UV-2000, USA).
2.5.
Spectroscopic analysis of BMM
EEM spectra of the BMM samples were analyzed by a fluorescence spectrometry system (F-4500, Hitachi, Japan). Use of the EEM spectrometry can be found elsewhere (Huang et al., 2011). Volumetric integration of the EEM peak was performed to calculate the content of a given BMM component semiquantitatively (Chen et al., 2003). Fluorescence regional integration (FRI) was analyzed according to the method of Chen et al. (2003), which was used to describe the volumetric percentage of a given EEM peak in all the EEM spectra. Solidstate cross polarization magic angle spinning carbon-13 (13C CPMAS) NMR spectra of the BMM samples were recorded on a Brucker Avance 400 MHz NMR spectroscopy. As the NMR analysis requires a considerable amount of dry samples (approx. 10e30 mg), three major fractions (BMMc, BMMb, and the fraction smaller than 100 kDa) of the BMM in sludge supernatant were analyzed. The fractionation step was performed by a tangential flow ultrafiltration (TFUF) system
4663
(Cogent, Millipore, USA), which allows for the separation of a large amount of BMM solution. Here, the BMM in membrane permeate and membrane foulants were not fractionated. In addition, the salts in all the samples were removed by filtering the samples in the TFUF system with a PVDF pellicon cassette filter (1 kDa). Prior to NMR analysis, all samples were dried by a vacuum freezing dryer (ALPHA, Marin Christ, Germany). To our best knowledge, this is the first work to study the structural components of size-fractionated BMM in MBRs by using EEM and CPMAS NMR.
3.
Results
3.1.
Photometric quantification of BMM
Over one year of operation, the MBR could achieve a steady performance on nutrient removal, and the overall removal efficiencies for COD, NHþ 4 -N, TN, and TP reached approx. 95%, 100%, 80%, and 90%, respectively. Fig. 1a shows the concentrations of proteins, polysaccharides, and humic substances in sludge supernatant, membrane foulants, and membrane permeate, respectively. The average concentrations of protein and polysaccharide in the sludge supernatant were 12.3 mg/L and 10.1 mg/L, respectively. The average rejection degree by the membranes was 44% for proteins, 27% for humic substances, and 89% for polysaccharides, respectively, indicating that polysaccharides were of large size and can thus be retained by membranes more readily. This finding is similar to a previous investigation (Meng et al., 2009a). In addition to the large size, the polysaccharides are expected to be captured by the fouling layer due to their glutinous nature (Okamura et al., 2009), which can also influence their own rejection degree. This is one reason why the fouling layer was rich in polysaccharides (see Fig. 1a). It suggests that the components of the fouling layer were likely to have a close relation with the rejection propensity of BMM compounds. As a result of the low rejection degree of humic substances, the membrane permeate is expected to be more aromatic (Meng et al., 2009a). Note that for a long-term operated MBR the deposited BMM on membranes could be degraded by the microorganisms in the fouling layer, and the microorganisms could also produce some BMM as a result of their metabolism. Thus, here it is difficult to make a mass balance on the BMM that transported to the permeate and that deposited on the membranes. Although the photometric measurements can quantify all BMM compounds no matter large or small, characterizing the size-fractionated BMM gives more details about the chemical nature of BMM components. To this point, the total BMM in sludge supernatant, membrane foulants and membrane permeate were further fractionated into several fractions by membranes with different MWCOs. Subsequently, the chemical composition of these fractions was analyzed in this work. As shown in Fig. 1b, the BMM compounds had a broad MW distribution, ranging widely from low MW (<5 kDa) to a large size (>0.45 mm). Overall, three major fractions were obtained: (i) BMMc (>0.45 mm), (ii) BMMb (100 kDae0.45 mm), and (iii) low-MW BMM (<5 kDa). As a result, the content of medium-MW BMM (5e100 kDa) was low. This bimodal MW distribution behaviour of BMM is
4664
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 6 6 1 e4 6 7 1
Fig. 1 e A schematic illustration showing the transport and deposition of BMM in MBRs (left image of (a)) and BMM concentration in sludge supernatant (n [ 30, right-top image of (a)), membrane foulants (n [ 3, right-middle image of (a)), and membrane permeate (n [ 30, right-bottom image of (a)); and MW distribution of BMM (b). PN and PS are the abbreviations of proteins and polysaccharides, respectively. The n [ 30 or 3 in the parentheses means the sample numbers analyzed in this work. Fig. 1a shows the quantitative values of BMM, and Fig. 1b presents the average value of MW distribution (their standard deviations are not shown, but are similar with those present in Fig. 1a).
consistent with that reported by Liang et al. (2007). It is interesting to note that most of polysaccharides are found in the fractions of BMMb (approx. 58%) and BMMc (approx. 22%). Thus, the majority of polysaccharides had an MW lager than 100 kDa, which give rise to a higher rejection degree by the membranes. In contrast, a considerable
amount of proteins were monitored in the range of 0e5 kDa (approx. 31%). The humic substances, likewise, were mostly composed of low molecules (i.e., 0e5 kDa), which accounted for more than 65% of the total. Therefore, the size nature of polysaccharides, proteins and humic substances were significantly different.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 6 6 1 e4 6 7 1
Due to the membrane rejection, the BMM in membrane permeate was mainly composed of low-MW compounds smaller than 5 kDa (see Fig. 1b). In contrast, BMMc and BMMb were the major constituents of membrane foulants, whereas the low-MW fraction (i.e., <100 kDa) can be neglected, indicating that the high-MW fractions were the major players of membrane fouling in MBRs. Again, the higher concentrations of the BMMc and BMMb in the sludge supernatant can directly contribute to the development of a fouling layer.
3.2.
Size nature of fluorescence active compounds
In comparison with the photometric methods, the EEM measurements can allow to get insights into more specific proteins (i.e., aromatic proteins and tryptophan proteins) and humic substances (i.e., humic acids and fulvic acids) in the BMM (Teychene et al., 2008; Galinha et al., 2011). Fig. 2 shows the EEM fluorescence spectra of feedwater, sludge supernatant, membrane foulants and permeate. The peaks located at the excitation/emission wavelengths (Ex/Em) of 225.0/ 350.0 nm (Peak A) and 275.0/350.0 nm (Peak B) indicate aromatic protein-like substances and tryptophan protein-like substances, respectively (Sheng and Yu, 2006). The peaks at 335.0/420.0 nm (Peak C) and 235.0/415.0 nm (Peak D) are assigned to humic-like acids and fulvic-like acids, respectively. As shown in Fig. 2, both feedwater and membrane foulants had a strong fluorescence intensity in Peak A and B, but had weak Peak C and D. The Peak C and D, however, were much stronger for the samples of sludge supernatant, indicating the production of humic acids and fulvic acids during the biological process. On the other hand, the fact that the
4665
Peak A and B of sludge supernatant were much weaker than those of feedwater suggests that the protein-like substances derived from feedwater can be partially biodegraded. In addition, the intensities of Peak A and B of the permeate were found to be around 50% smaller than those of the sludge supernatant (see Table S2 of SM), suggesting that a significant amount of aromatic and tryptophan proteins in the sludge supernatant were rejected by the membranes or captured by the fouling layer. Moreover, the sludge supernatant and the membrane permeate had similar fluorescence intensities for Peak C and D, confirming a small size of humic acids and fulvic acids. Similarly, Wang et al. (2009) also showed a small rejection degree of humic acids indicated by Peak C. The size-fractionated BMM were also characterized by the EEM spectroscopy (see Fig. S2, S3 S4, S5 of SM). Volumetric integration of the peaks in the EEM spectra was done according to Chen et al. (2003). The MW distribution of the EEM-detected compounds can be obtained (see Fig. 3a). Similar to the photometric measurements, both proteins in the sludge supernatant were present in three major fractions: BMMc, BMMb and low-MW fraction. Additionally, consistent with the results obtained by the photometric measurements, the majority of humic acids (63%) and fulvic acids (59%) in the sludge supernatant had an MW smaller than 5 kDa, whereas their proportion in the size range of 5e100 kDa was small. Moreover, consistent with the photometric measurements, the EEM data reveal that the organics in membrane foulants were mainly composed of BMMc and BMMb fractions. In fact, the contribution of humics to membrane foulants could be ignored despite the fact that they were also detected. This can be verified by the data shown in Fig. S4 of SM.
Fig. 2 e EEM spectra of feedwater, sludge supernatant, membrane permeate and membrane foulants.
4666
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 6 6 1 e4 6 7 1
a
100 <5kDa 80
% of total
30-5kDa 60 100-30kDa 40 BMMb 20 BMMc 0 APN TP T N HA FA AP A N TP T N HA FA AP A N TP T N HA FA Sludge supernatant
Membrane permeate
Membrane foulants
b 100%
F FA 80%
HA
FRI
60%
40%
TPN P
20%
APN P 0% B BMM c
B Mb BMM
Low-MW M (<100 kDa)
Membrane foula l nts
Membrane Permeat ae
Sludge supernatant Fig. 3 e MW Distribution EEM-detected compounds in different size ranges (a) and FRI of EEM-detected compounds in a given size-based BMM fraction (b). APN, TPN, HA and FA are the abbreviations of aromatic proteins, tryptophan proteins, humic acids and fulvic acids, respectively.
To have a deeper understanding on the chemical composition of BMM, FRI was applied to depict the relative content of a given component in the whole BMM. As shown in Fig. 3b, the FRI of the three size-fractionated BMM in the sludge supernatant is different, elucidating the differences in the chemical composition of the three size-fractionated samples. Apparently, both BMMc and BMMb of the sludge supernatant contained a great portion of aromatic (25% and 27%, respectively) and tryptophan (59% and 64%, respectively) proteins. On the contrary, the low-MW fraction of the sludge supernatant contained approx. 58% of humics, including humic acids and fulvic acids, in total. Thus, the content of aromatic compounds including aromatic proteins in the low-MW
fraction was up to 70%. According to the FRI data, the organics in the membrane permeate had a quite similar chemical composition to the low-MW fraction of the sludge supernatant. Thus, the low-MW BMM (<100 kDa) of the sludge supernatant was the major organics in the membrane permeate. In addition, the organics in the membrane foulants had similar content in both aromatic (46%) and tryptophan (43%) proteins. By comparison the FRI of the membrane foulants and that of BMMc and BMMb fractions of the sludge supernatant, it can be concluded that the aromatic proteins were more prone to accumulate within the fouling layer than tryptophan proteins, which is possibly due to the hydrophobic nature of aromatic proteins.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 6 6 1 e4 6 7 1
3.3. BMM
Structural composition of the size-fractionated
Unfortunately, EEM technique can only detect fluorescence active compounds, whereas other components such as polysaccharides were not detectable. Therefore, a characterization by 13C CPMAS NMR was performed to supplement the limitations of EEM. NMR spectra of the size-fractionated BMM of the sludge supernatant (i.e., BMMc, BMMb, and low-MW fraction), the membrane foulants and the membrane permeate are presented in Fig. 4. In the spectra, the peak near 175 ppm is an indicator of proteinaceous nature. The peak near 165 ppm is assigned to the presence of aromatic carbon (Kimura et al., 2005). The presence of carbonates in samples can also lead to such a peak. It should be mentioned that all the samples used for the NMR characterization were salt-free, and thus the peak at 165 ppm cannot be attributed to carbonates in this work. The peaks at 103 and 73 ppm are attributed to the secondary alcohols and the glycosidic carbon of carbohydrates or polysaccharides, respectively (Jiao et al., 2010). The glycosidic carbon is an indicator of the linkage nature of polysaccharides (e.g., alpha oxygen linkage, beta oxygen linkage) (Jiao et al., 2010). The peaks present at 0e45 ppm are usually assigned to long-chain aliphatics or lipids (i.e., methylene groups) (Metzger et al., 2009). The NMR spectra of the size-fractionated BMM showed in Fig. 4a reveal the different abundance and structure of organics present in the three samples. The most obvious difference is that the BMMb fraction had more and stronger peaks than the other two fractions, e.g., this fraction was more abundant in polysaccharides and lipids. One major peak for the low-MW fraction (<100 kDa) of sludge supernatant was detected at 165 ppm, but other peaks were weak. It indicates that the organics in the low-MW fraction was dominated by aromatic carbon. The fact that the NMR spectrum (i.e., signal pattern and intensity) of the membrane foulants was similar to that of the BMMb of the sludge supernatant (see Fig. 4a, b) suggests that they had similar structural composition. Therefore, the BMMb of the sludge supernatant should be a major origin of membrane foulants. Nevertheless, the possibility that the NMR signals of the BMMc in the membrane foulants might be overlapped by those of the BMMb cannot be ruled out. The results obtained from photometric measurements and EEM characterization can support such a possibility. Our work demonstrates that a size-based characterization avoids the overlapping phenomenon within different fractions, which leads to a clear picture of the chemical/structural nature of BMM in MBRs. A similar NMR spectrum between the low-MW BMM of the sludge supernatant and the membrane permeate was also found, indicating that their structural composition was somewhat analogical. By integrating the individual peak area of 13C NMR spectra, the relative carbon content of BMM compounds (i.e., polysaccharides, proteins, aromatics, and lipids) can be determined (Lankes et al., 2008). As shown in Fig. 4c, the BMMc fraction of the sludge supernatant was mostly and equally dominated by polysaccharides and proteins, which is consistent with the results obtained by the photometric methods. Interestingly, both the BMMb of the sludge supernatant (64%) and the membrane foulants (56%) and had a high carbon
4667
content of polysaccharides. In addition, they contained considerable lipids, e.g., 14% for the BMMb of the sludge supernatant and 21% for the membrane foulants. It indicates that both the BMMb and the membrane foulants contained considerable polysaccharides and lipids. As a consequence, the BMMb of the sludge supernatant was expected to be an important contributor of fouling layer. The lipid/polysaccharide ratio in the BMMb of the sludge supernatant and the membrane foulants was determined to be 0.25 and 0.33, respectively. It shows that lipids had a higher potential to accumulate in the fouling layer than polysaccharides. This might be due to the fact that the lipids were of both high MW and strong hydrophobicity (Lankes et al., 2008). The polysaccharides, on the contrary, were hydrophilic in addition to their high MW. On the other hand, both the membrane permeate (90%) and the low-MW fraction of the sludge supernatant (74%) mainly consisted of aromatics such as aromatic proteins, and humic substances. This result verified our previous assumption that the organics in membrane permeate were mostly aromatics. Of the three BMM fractions of the sludge supernatant, the BMMb had the lowest content of aromatics (13%) and hence the lowest aromaticity. The NMR analysis in conjunction with EEM measurements suggests that the aromatics in BMMc and BMMb were mainly attributed to the presence of aromatic proteins.
4.
Discussion
Our ultimate aim was to reveal the chemical or structural nature of BMM by a size-based characterization. The results obtained in this work can provide a deeper understanding on BMM in MBRs, which can be helpful for the control of membrane fouling. Of all the components in BMM, polysaccharides were found to be the most important compound affecting membrane fouling. Firstly, there was a considerable amount of polysaccharides in the sludge supernatant. Secondly, the results of NMR characterization showed that polysaccharides were mostly present in the fraction of BMMb with high MW. Previously, the polysaccharides produced by anaerobic sludge and aerobic granules were found to have high MW (Schiener et al., 1998; Seviour et al., 2010a). By using NMR analysis, the high-MW compounds (0.1e0.45 mm) in the natural organics of river water was also observed to be predominated by polysaccharides (Lankes et al., 2008). These previous studies, together with our current data, show that the polysaccharides are of high MW regardless of their different origins. In our study, it is interestingly noted that the freeze-dried samples of both membrane foulants and the BMMb in sludge supernatant had a crosslinked structure (see Fig. S6 of SM). However, the dried samples of the BMMc and low-MW fraction of the sludge supernatant and the organics in the permeate water were in a powdered form. The crosslinked structure was mainly due to the gelling propensity of the polysaccharides (Wang and Waite, 2009; Seviour et al., 2010a). The polysaccharides produced by sludge could be a complex heteropolysaccharide with a repeated sequence of some sugars such as galactose, glucose, heptose, rhammnos, mannose, etc (Seviour et al., 2010b). Thus, in addition to the
4668
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 6 6 1 e4 6 7 1
Fig. 4 e NMR spectra of size-based fractions of BMM in sludge supernatant (a), membrane foulants and membrane permeate (b); and their composition determined by 13C CPMAS NMR (given in % carbon) (c).
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 6 6 1 e4 6 7 1
large pool and high MW of polysaccharides in the sludge supernatant, the linkage nature of polysaccharides should be the third reason that leads to their strong fouling propensity. As can be seen from FRI analysis, the aromatic proteins had a much higher fouling potential than the tryptophan proteins. By using NMR analysis, lipids or fatty acids could be further important fouling-causing matter, which is likely derived from the BMMb of the sludge supernatant. Because of their hydrophobic nature (Lankes et al., 2008; Al-Halbouni et al., 2009), the lipids seem to accumulate on the membranes more readily than the polysaccharides. The hydrophobic interactions between lipids or aromatic proteins and membranes, in particular hydrophobic membranes, could accelerate membrane fouling greatly. Therefore, lipids and fatty acids should be paid enough attention to in future research of MBR fouling. A number of investigations were performed to find a relation between the overall quantity of BMM in sludge supernatant and the fouling rate of membranes. But, the chemical nature of BMM was not fully considered, which can largely lead to the contradictions of some reported findings. The results of our work suggest that the size, chemical/structural nature, and hydrophobic nature of BMM indeed play important roles in MBR fouling. Crucially, the fouling mechanisms of different BMM compounds may vary. Based on our current study, the fouling of polysaccharides was due to their high MW and gelling propensity. In comparison, the fouling of lipids was attributed to their high MW and hydrophobic nature. Likewise, the higher fouling propensity of aromatic proteins over tryptophan proteins could also be due to the high hydrophobicity. The humics in sludge supernatant however had low fouling potential in the long-term operated MBRs, and they were the main constituents of the organics in membrane permeate. Therefore, if all the nature of BMM (i.e., concentration, size nature, chemical nature and structural nature) could be fully taken into account, a universal fouling theory of MBRs can hopefully be reached in the future. It should be pointed out that temporal variation of membrane foulants can take place during the long-term operation of MBRs, which is expected to cause the change of chemical nature of membrane foulants such as the degradation and production of extracellular polymeric substances (EPS) by the deposited bacteria (Hwang et al., 2008). Additionally, the nature of feed wastewater is another factor affecting BMM. Different types of wastewater such as synthetic wastewater, municipal wastewater and specific industrial wastewater might have different potential to produce BMM. For instance, the average PS/PN ratio in the BMM of sludge supernatant in this MBR fed with synthetic wastewater was 0.82, which is much higher than that (PS/ PN ¼ 0.50) in a investigation based on a pilot-scale MBR which was fed with a mixture of domestic wastewater, rainwater and industrial wastewater (Meng et al., 2009a). Again, the BMM leading to irreversible fouling are of interest for the long-term operation of MBRs. However, the collection of these foulant species is limited because the use of chemicals could alter the original nature of foulants such as protein denaturation and hydrolysis. Development or use of in-situ and non-destructive characterization techniques such as confocal laser Raman spectroscopy or surface-enhanced
4669
Raman spectroscopy (Ivleva et al., 2008, 2009; Cui et al., 2011) would be an opportunity to address the limitations of current approach. For a long-term operated MBR, there is no doubt that the variation of operating conditions such as aeration intensity, SRT, HRT and imposed flux can also cause the change of BMM composition and hence their fouling propensity (Kimura et al., 2005; Ng et al., 2006). The focus of our current work was to understand the chemical nature of BMM compounds at a fundamental level. The findings of this study are expected to complement the conclusions obtained in previous literature and update the knowledge of BMM in wastewater treatment processes, which can be easily extended to the study of the influence of operating conditions on the chemical/structural nature of BMM.
5.
Concluding remarks
In this work, the chemical and structural characteristics of size-fractionated BMM in an MBR were investigated. The composition of BMM compounds in different size ranges was found to be chemically and structurally different. Both the photometric measurements and NMR characterization showed that the BMMc fraction in sludge supernatant had equal content of polysaccharides and proteins. In addition, the BMMb and the low-MW BMM (<100 kDa) were proved to be carbonaceous and aromatics, respectively. Moreover, the results obtained from all the measurements suggest that an MW of 100 kDa was a critical size for the deposition and transport of BMM in the MBR. The FRI analysis of the EEM spectra can also support the similarity in the protein and humics composition of membrane permeate and the low-MW BMM (<100 kDa) in sludge supernatant. The detailed characterization of BMM compounds revealed that polysaccharides and lipids originated from the BMMb of the sludge supernatant were of high significance for membrane fouling during the long-term operation of MBRs. Aromatic proteins had a higher fouling propensity than tryptophan proteins though they were of similar size nature. The role of low-MW BMM (<100 kDa) in MBR fouling could be ignored. Apparently, characterization of the size-fractionated BMM can not only get more information about the nature of BMM, but also help in tracking the fate of BMM in MBRs.
Acknowledgements This work was supported by Specialized Research Fund for the Doctoral Program of Higher Education of China (No. 20100171120014) and Fundamental Research Funds for the Central Universities (No. 2010380003161541).
Appendix. Supplementary material Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.watres.2011.06.026.
4670
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 6 6 1 e4 6 7 1
references
Al-Halbouni, D., Dott, W., Hollender, J., 2009. Occurrence and composition of extracellular lipids and polysaccharides in a full-scale membrane bioreactor. Water Research 43 (1), 97e106. Alasonati, E., Slaveykova, V.I., 2011. Composition and molar mass characterisation of bacterial extracellular polymeric substances by using chemical, spectroscopic and fractionation techniques. Environmental Chemistry 8 (2), 155e162. Chen, M.Y., Lee, D.J., Yang, Z., Peng, X.F., Lai, J.Y., 2006. Fluorecent staining for study of extracellular polymeric substances in membrane biofouling layers. Environmental Science & Technology 40 (21), 6642e6646. Chen, W., Westerhoff, P., Leenheer, J.A., Booksh, K., 2003. Fluorescence excitation e emission matrix regional integration to quantify spectra for dissolved organic matter. Environmental Science & Technology 37 (24), 5701e5710. Cui, L., Yao, M., Ren, B., Zhang, K.S., 2011. Sensitive and versatile detection of the fouling process and fouling propensity of proteins on polyvinylidene fluoride membranes via surfaceenhanced Raman spectroscopy. Analytical Chemistry 83 (5), 1709e1716. Drews, A., 2010. Membrane fouling in membrane bioreactors e characterisation, contradictions, cause and cures. Journal of Membrane Science 363 (1e2), 1e28. Drews, A., Mante, J., Iversen, V., Vocks, M., Lesjean, B., Kraume, M., 2007. Impact of ambient conditions on SMP elimination and rejection in MBRs. Water Research 41 (17), 3850e3858. Drews, A., Vocks, M., Bracklow, U., Iversen, V., Kraume, M., 2008. Does fouling in MBRs depend on SMP? Desalination 231 (1e3), 141e149. Dubois, M., Gilles, K., Hamilton, J., Rebers, P., Smith, F., 1956. Colorimetric method for determination of sugars and related substances. Analytical Chemistry 28 (3), 350e356. Fan, F., Zhou, H., Husain, H., 2006. Identification of wastewater sludge characteristics to predict critical flux for membrane bioreactor processes. Water Research 40 (2), 205e212. Galinha, C.F., Carvalho, G., Portugal, C.A.M., Guglielmi, G., Oliveira, R., Crespo, J.G., Reis, M.A.M., 2011. Real-time monitoring of membrane bioreactors with 2D-fluorescence data and statistically based models. Water Science and Technology 63 (7), 1381e1388. Huang, G.C., Meng, F.G., Zheng, X., Wang, Y., Wang, Z.G., Liu, H.J., Jekel, M., 2011. Biodegradation behavior of natural organic matter (NOM) in a biological aerated filter (BAF) as a pretreatment for ultrafiltration (UF) of river water. Applied Microbiology and Biotechnology 90 (5), 1795e1803. Huang, X., Liu, R., Qian, Y., 2000. Behaviour of soluble microbial products in a membrane bioreactor. Process Biochemistry 36 (5), 401e406. Hwang, B.K., Lee, W.N., Yeon, K.M., Park, P.K., Lee, C.H., Chang, I.S., Drews, A., Kraume, M., 2008. Correlating TMP increases with microbial characteristics in the bio-cake on the membrane surface in a membrane bioreactor. Environmental Science & Technology 42 (11), 3963e3968. Ivleva, N.P., Wagner, M., Horn, H., Niessner, R., Haisch, C., 2008. situ surface-enhanced Raman scattering analysis of biofilm. Analytical Chemistry 80 (22), 8538e8544. Ivleva, N.P., Wagner, M., Horn, H., Niessner, R., Haisch, C., 2009. Towards a nondestructive chemical characterization of biofilm matrix by Raman microscopy. Analytical and Bioanalytical Chemistry 393 (1), 197e206. Jarusutthirak, C., Amy, G., 2006. Role of soluble microbial products (SMP) in membrane fouling and flux decline. Environmental Science and Technology 40 (3), 969e974.
Jiao, Y.Q., Cody, G.D., Harding, A.K., Wilmes, P., Schrenk, M., Wheeler, K.E., Banfield, J.F., Thelen, M.P., 2010. Characterization of extracellular polymeric substances from acidophilic microbial biofilms. Applied and Environmental Microbiology 76 (9), 2916e2922. Kimura, K., Yamato, N., Yamamura, H., Watanabe, Y., 2005. Membrane fouling in pilot-scale membrane bioreactors (MBRs) treating municipal wastewater. Environmental Science and Technology 39 (16), 6293e6299. Laabs, C.N., Amy, G.L., Jekel, M., 2006. Understanding the size and character of fouling-causing substances from effluent organic matter (EfOM) in low-pressure membrane filtration. Environmental Science and Technology 40 (14), 4495e4499. Lankes, U., Ludemann, H.D., Frimmel, F.H., 2008. Search for basic relationships between “molecular size” and “chemical structure” of aquatic natural organic matter e answers from 13C and 15N CPMAS NMR spectroscopy. Water Research 42 (4e5), 1051e1060. Laspidou, C.S., Rittmann, B.E., 2002. A unified theory for extracellular polymeric substances, soluble microbial products, and active and inert biomass. Water Research 36 (11), 2711e2720. Lee, J., Ahn, W.Y., Lee, C.H., 2001. Comparison of the filtration characteristics between attached and suspended growth microorganisms in submerged membrane bioreactor. Water Research 35 (10), 2435e2445. Lee, W., Kang, S., Shin, H., 2003. Sludge characteristics and their contribution to microfiltration in submerged membrane bioreactors. Journal of Membrane Science 216 (1e2), 217e227. Leenheer, J.A., 1981. Comprehensive approach to preparative isolation and fractionation of dissolved organic carbon from natural waters and wastewaters. Environmental Science and Technology 15, 578e587. Lesjean, B., Rosenberger, S., Laabs, C., Jekel, M., Gnirss, R., Amy, G., 2005. Correlation between membrane fouling and soluble/ colloidal organic substances in membrane bioreactors for municipal wastewater treatment. Water Science and Technology 51 (6e7), 1e8. Liang, S., Liu, C., Song, L., 2007. Soluble microbial products in membrane bioreactor operation: behaviors, characteristics, and fouling potential. Water Research 41 (1), 95e101. Lowry, O.H., Rosebrough, N.J., Farr, A.L., Randall, R.J., 1951. Protein measurement with the folin phenol reagent. Journal of Biological Chemistry 193, 265e275. Meng, F., Drews, A., Mehrez, R., Iversen, V., Ernst, M., Yang, F., Jekel, M., Kraume, M., 2009a. Occurrence, source, and fate of dissolved organic matter (DOM) in a pilot-scale membrane bioreactor. Environmental Science and Technology 43 (23), 8821e8826. Meng, F., Zhang, H., Yang, F., Liu, L., 2007. Characterization of cake layer in submerged membrane bioreactor. Environmental Science and Technology 41 (11), 4065e4070. Meng, F.G., Chae, S.R., Drews, A., Kraume, M., Shin, H.S., Yang, F.L., 2009b. Recent advances in membrane bioreactors (MBRs): membrane fouling and membrane material. Water Research 43 (6), 1489e1512. Metzger, U., Lankes, U., Fischpera, K., Frimmel, F.H., 2009. The concentration of polysaccharides and proteins in EPS of Pseudomonas putida and Aureobasidum pullulans as revealed by C-13 CPMAS NMR spectroscopy. Applied Microbiology and Biotechnology 85 (1), 197e206. Ng, H.Y., Tan, T.W., Ong, S.L., 2006. Membrane fouling of submerged membrane bioreactors: impact of mean cell residence time and the contributing factors. Environmental Science and Technology 40 (8), 2706e2713. Ng, T.C.A., Ng, H.Y., 2010. Characterisation of initial fouling in aerobic submerged membrane bioreactors in relation to
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 6 6 1 e4 6 7 1
physico-chemical characteristics under different flux conditions. Water Research 44 (7), 2336e2348. Ni, B.J., Rittmann, B.E., Fang, F., Xu, J.A., Yu, H.Q., 2010. Long-term formation of microbial products in a sequencing batch reactor. Water Research 44 (13), 3787e3796. Okamura, D., Mori, Y., Hashimoto, T., Hori, K., 2009. Identification of biofoulant of membrane bioreactors in soluble microbial products. Water Research 43 (17), 4356e4362. Ramesh, A., Lee, D.J., Lai, J.Y., 2007. Membrane biofouling by extracellular polymeric substances or soluble microbial products from membrane bioreactor sludge. Applied Microbiology and Biotechnology 74 (3), 699e707. Rosenberger, S., Laabs, C., Lesjean, B., Gnirss, R., Amy, G., Jekel, M., Schrotter, J.C., 2006. Impact of colloidal and soluble organic material on membrane performance in membrane bioreactors for municipal wastewater treatment. Water Research 40 (4), 710e720. Schiener, P., Nachaiyasit, S., Stuckey, D.C., 1998. Production of soluble microbial products (SMP) in an anaerobic baffled reactor: composition, biodegradability, and the effect of process parameters. Environmental Technology 19 (4), 391e399. Seviour, T., Donose, B.C., Pijuan, M., Yuan, Z.G., 2010a. Purification and conformational analysis of a key exopolysaccharide component of mixed culture aerobic sludge granules. Environmental Science & Technology 44 (12), 4729e4734. Seviour, T., Lambert, L.K., Pijuan, M., Yuan, Z.G., 2010b. Structural determination of a key exopolysaccharide in mixed culture aerobic sludge granules using NMR spectroscopy. Environmental Science & Technology 44 (23), 8964e8970.
4671
Sheng, G.P., Yu, H.Q., 2006. Characterization of extracellular polymeric substances of aerobic and anaerobic sludge using three-dimensional excitation and emission matrix fluorescence spectroscopy. Water Research 40 (6), 1233e1239. Song, L.F., Liang, S., Yuan, L.Y., 2007. Retarded transport and accumulation of soluble microbial products in a membrane bioreactor. Journal of Environmental Engineering 133 (1), 36e43. Teychene, B., Guigui, C., Cabassud, C., Amy, G., 2008. Toward a better identification of foulant species in MBR processes. Desalination 231 (1e3), 27e34. Tian, J.y., Liang, H., Nan, J., Yang, Y.l., You, S.j., Li, G.b., 2009. Submerged membrane bioreactor (sMBR) for the treatment of contaminated raw water. Chemical Engineering Journal 148 (2e3), 296e305. Wang, X.M., Waite, T.D., 2009. Role of gelling soluble and colloidal microbial products in membrane fouling. Environmental Science & Technology 43 (24), 9341e9347. Wang, Z., Wu, Z., Tang, S., 2009. Characterization of dissolved organic matter in a submerged membrane bioreactor by using three-dimensional excitation and emission matrix fluorescence spectroscopy. Water Research 43 (6), 1533e1540. Yamamura, H., Kimura, K., Okajima, T., Tokumoto, H., Watanabe, Y., 2008. Affinity of functional groups for membrane surfaces: implications for physically irreversible fouling. Environmental Science and Technology 42 (14), 5310e5315. Zhang, G.J., Ji, S.L., Gao, X., Liu, Z.Z., 2008. Adsorptive fouling of extracellular polymeric substances with polymeric ultrafiltration membranes. Journal of Membrane Science 309 (1e2), 28e35.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 6 7 2 e4 6 8 2
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
The role of nitrite and free nitrous acid (FNA) in wastewater treatment plants Yan Zhou a,*, Adrian Oehmen b, Melvin Lim c, Vel Vadivelu d, Wun Jern Ng a a
Advanced Environmental Biotechnology Centre (AEBC), Nanyang Environment and Water Research Institute (NEWRI), Nanyang Technological University, School of Biological Science, Level N-B2-01, 60 Nanyang Avenue, Singapore 639798, Singapore b REQUIMTE e CQFB, Departamento de Quı´mica, Faculdade de Cieˆncias e Tecnologia, Universidade Nova de Lisboa, 2829-516 Caparica, Portugal c School of Applied Science, Republic Polytechnic, 9 Woodlands Avenue 9, Singapore 738964, Singapore d School of Chemical Engineering, Universiti Sains Malaysia, 14300 Nibong Tebal, Penang, Malaysia
article info
abstract
Article history:
Nitrite is known to accumulate in wastewater treatment plants (WWTPs) under certain
Received 27 April 2011
environmental conditions. The protonated form of nitrite, free nitrous acid (FNA), has been
Received in revised form
found to cause severe inhibition to numerous bioprocesses at WWTPs. However, this
17 June 2011
inhibitory effect of FNA may possibly be gainfully exploited, such as repressing nitrite
Accepted 21 June 2011
oxidizing bacteria (NOB) growth to achieve N removal via the nitrite shortcut. However, the
Available online 28 June 2011
inhibition threshold of FNA to repress NOB (w0.02 mg HNO2-N/L) may also inhibit other bioprocesses. This paper reviews the inhibitory effects of FNA on nitrifiers, denitrifiers,
Keywords:
anammox bacteria, phosphorus accumulating organisms (PAO), methanogens, and other
Nitrite
microorganisms in populations used in WWTPs. The possible inhibition mechanisms of
Free nitrous acid (FNA)
FNA on microorganisms are discussed and compared. It is concluded that a single inhi-
Bioprocesses
bition mechanism is not sufficient to explain the negative impacts of FNA on microbial
Inhibition mechanisms
metabolisms and that multiple inhibitory effects can be generated from FNA. The review
Enzymes
would suggest further research is necessary before the FNA inhibition mechanisms can be
Wastewater treatment
more effectively used to optimize WWTP bioprocesses. Perspectives on research directions,
plants (WWTPs)
how the outcomes may be used to manipulate bioprocesses and the overall implications of FNA on WWTPs are also discussed. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
Biological nutrient removal (BNR) is by far the most economic and environmentally friendly way to achieve nitrogen and phosphorus removal from wastewater (Tchobanoglous et al., 2003). Compared to chemical treatment methods, BNR reduces chemical consumption and cost, reduces the production of waste solids, and has lower energy requirement. In the nitrification process, ammonium is converted to nitrite
by ammonium oxidizing bacteria (AOB, nitritation) and nitrite is oxidized to nitrate by nitrite oxidizing bacteria (NOB, nitratation). Almost all nitrifying bacteria are autotrophic. For each carbon-atom fixed, autotrophic bacteria consume 80% of the energy generated from substrate oxidation, which results in a very low growth yield (Kelly, 1978). Denitrification is the process of nitrate reduction into nitrite and then into molecular nitrogen, which is performed by a functional group of heterotrophs that use oxidized nitrogen (NO 3 , NO2 , NO and
* Corresponding author. Tel.: þ65 6592 1832; fax: þ65 6515 5981. E-mail addresses: [email protected] (Y. Zhou), [email protected] (A. Oehmen), [email protected] (M. Lim), [email protected] (V. Vadivelu), [email protected] (W.J. Ng). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.06.025
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 6 7 2 e4 6 8 2
N2O) as the electron acceptor in respiration. This is usually coupled with oxidation of organic compounds as the electron donor for energy generation. Nitrite oxidation is typically considered to be the rate limiting step under normal operating conditions for nitrification. However, many environmental conditions can cause ammonia oxidation rates to exceed those of nitrite oxidation, leading to nitrite accumulation. For example, high temperature will cause AOB to out-compete NOB because of their greater growth rate at temperatures higher than 25 C (Hellinga et al., 1998). NOB also has been shown to have lower affinity for oxygen than AOB (Ciudad et al., 2006). Hence, a low dissolved oxygen (DO) conditions will favour nitrite accumulation. Nitrite accumulation can also occur under anoxic conditions, due to differences in the denitrification kinetics between nitrate and nitrite reduction (Glass and Silverstein, 1998). Imbalanced reduction rates can be attributed to many environmental factors such as, high temperature, the presence of high DO anoxically, low pH, types of carbon source (e.g. volatile fatty acids, and methanol) and low carbon to nitrogen ratios. It can also be due to the absence of certain denitrifying species or enzymes in the sludge (Meng et al., 2010). Further, it has been reported that the presence of phosphate (can depress nitrite reduction markedly), heavy metals, sunlight (NOB is more sensitive to sunlight than AOB) and reactor operation (e.g. length of sludge retention time (SRT) and hydraulic retention time (HRT)) can also cause nitrite accumulation (Philips et al., 2002). The presence of a high concentration of nitrite has been reported to be a severe inhibitor on a wide range of microorganisms. Depending on the concentration and operating pH and temperature, this inhibition can slow, or even completely cease microbial activities and reconfigure the microbial community structure. While nitrite accumulation can be viewed as an unfavourable occurrence in conventional wastewater treatment plants (WWTP), there are advanced processes which exploit the use of nitrite as an intermediate to achieve “shortcut” nitrogen removal (Hellinga et al., 1998; Kuai and Verstraete, 1998; Strous et al., 1999). Such processes have a number of potential advantages including: lower carbon source requirements in denitrification, lower consumption of oxygen in nitrification, reduction of reactor volumes due to lower HRT requirements, higher denitrification rates and smaller sludge production (Turk and Mavinic, 1986). These advantages are even more notable where a wastewater contains high ammonium or low organic carbon contents. Due to the inhibitory effects nitrite can impose on biomass, a relevant current challenge is to understand when such inhibition can occur, which in turn can contribute to the optimisation of advanced nitrogen removal processes. This review analyzes nitrite inhibitory effects on commonly found microbial communities in WWTPs, to yield better understanding of the inhibitory mechanisms involved. Strategies to reduce these inhibitory effects will also be discussed.
2. Free nitrous acid (FNA) e the true inhibitor on bacteria in BNR systems FNA, the protonated form of nitrite, has been shown in numerous cases to be the cause of inhibition for bacteria in
4673
WWTPs, rather than the nitrite anion itself. FNA can be determined through the nitrite concentration, pH and temperature, by the formula SNeNO2 =ðKa 10pH Þ, where Ka ¼ e2300=ð273þTð CÞÞ (Anthonisen et al., 1976). The studies that have investigated the factors affecting this inhibition and its severity at different FNA levels on the primary organisms responsible for BNR are summarised below.
2.1.
FNA inhibition on nitrifiers
Anthonisen et al. (1976) had first reported inhibition in nitrification processes is a function of nitrite concentration and pH. As nitrite is produced, pH decreases due to the release of hydrogen ions. The nitrite produced thus will exist in equilibrium with the unionized form (FNA). They had concluded that the inhibition on nitrification is related to the concentration of unionized nitrous acid (FNA) rather than the nitrite anion concentration, and the inhibition on nitrification will be initiated at an FNA concentration of 0.22e2.8 mg HNO2-N/L. Later studies have found variable inhibitory threshold levels, where an FNA concentration range of 0.42e1.72 mg HNO2-N/L has resulted in a 50% reduction in AOB activity (Anthonisen et al., 1976; Stein and Arp, 1998; Hellinga et al., 1999; Fux and Siegrist, 2004; Vadivelu et al., 2006a; Tan et al., 2008; Tora et al., 2010). A combination of process factors and the microbial populations within the sludge of these different studies could have been responsible for this wide range of inhibitory threshold levels. Tora et al. (2010) found that situations of inorganic carbon limitations enhanced the inhibitory effect of FNA in their AOB enriched culture, however, this finding contrasted with the study of Vadivelu et al. (2006a). It is noteworthy that the dominant organism in the case of Tora et al. (2010) bound to the more general Nso190 probe (81%), while in the case of Vadivelu et al. (2006a), the sludge was dominated by organisms binding the NEU probe (82%), and no binding to the Nso190 probe was observed. This difference in microbial community could explain the contrasting FNA inhibition effects. Moreover, different species and strains within a genus could possess distinct tolerances towards FNA. The genome of Nitrosomonas europaea was suggested to have some different metabolic properties as compared to Nitrosomonas eutropha, which notably includes abilities to use nitrite as an electron acceptor anaerobically (Zart and Bock, 1998; Chain et al., 2003; Stein et al., 2007). This suggested possibility of a difference in their capacity to tolerate FNA. FNA also affects multiple metabolic pathways in AOBs. By decoupling their energy generation and growth processes, Vadivelu et al. (2006a) demonstrated that FNA has different inhibitory effects on Nitrosomonas’ anabolic and catabolic activities, with the biosynthesis process being the most sensitive. It should be noted that FNA inhibition on nitritation can be reversible; Yang et al. (2003) found that the recovery period of nitritation from FNA inhibition lasted about 12 days. The range of FNA concentrations affecting NOB activity has been found to start from 0.011e0.07 mg HNO2-N/L, where complete inhibition was observed at 0.026e0.22 mg HNO2-N/L (Prakasam and Loehr, 1972; Anthonisen et al., 1976; Vadivelu et al., 2006b; Zhang et al., 2010). Nitrobacter was found to be the dominant community in most of these studies, and indeed have been commonly found in laboratory scale reactors,
4674
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 6 7 2 e4 6 8 2
although they are rarely found in full-scale plants, where Nitrospira is the main NOB commonly present (Burrell et al., 1999; Daims et al., 2001). Nitrospira has been reported to be more sensitive to FNA than Nitrobacter (Blackburne et al., 2007), where complete inhibition of oxygen uptake activity was observed at a FNA concentration of less than 0.03 mg HNO2-N/L. This may explain why it seems to be more difficult to culture Nitrospira in enrichment than Nitrobacter (Blackburne et al., 2007), since Nitrobacter is less inhibited by the substrate fed for their enrichment, and would be more active until very low FNA concentrations. In order to achieve nitritation, and thus shortcut N removal, the key point is to maintain the maximum simultaneous difference between AOB and NOB growth rates, thereby washing out NOBs from the sludge. The range of inhibition thresholds reported above show that NOB is more sensitive to FNA than AOB by approximately one order of magnitude. While very high concentrations of FNA may cause the wash out of both AOB and NOB from nitrifying systems, a lower FNA concentration range (e.g. 0.02e0.03 mg HNO2-N/L) is appropriate to wash out NOB in nitritation-based systems, such as the treatment of sludge dewatering liquors and leachates.
2.2.
FNA inhibition on denitrifiers
Numerous reports have indicated that nitrite accumulation (in the range of 140e700 mg NO 2 -N/L) inhibits the growth of denitrifying bacteria (Vanverseveld et al., 1977; Rake and Eagon, 1980). These reports often indicated that the toxicity of nitrite was strongly modulated by pH. Beccari et al. (1983) found that the biomass concentration together with the pH and nitrite concentration regulated the inhibitory effect on denitrification. At a pH level of 8.0e8.9, a nitrite concentration up to 2000 mg NO 2 -N/L showed no inhibitory effect on a denitrifying biofilm (Chen et al., 1991). Comparing all of these results, Abeling and Seyfried (1992) concluded that FNA was the controlling factor of the inhibition. The toxicity threshold in their study was reported to be 0.04 mg HNO2-N/L on activated sludge. Since then, other FNA toxicity threshold levels have been reported. Using a pure culture of Pseudomonas fluorescens, Almeida et al. (1995a) showed that cell growth was completely stopped at the FNA concentration of 0.066 mg HNO2-N/L. However, no inhibition of the primary metabolism including nitrate reduction, carbon source consumption and nitrite reduction was observed. They suggested that nitrite acted as a growth uncoupler, and particularly as a proton uncoupler. A lower FNA inhibitory threshold of 0.02 mg HNO2-N/L on denitrification, was found by Glass et al. (1997) using activated sludge. Similar to the case of nitrifiers, it is likely that the FNA inhibition threshold is a function of the microbial community characteristics as well as other operational factors. In addition to direct inhibition by FNA, under co-existence of nitrite and nitrate, the denitrification rate of each substrate is influenced by the competition between nitrite and nitrate reductases for electron donors (Wilderer et al., 1987; Almeida et al., 1995b; Glass and Silverstein, 1998). Recently, Ma et al. (2010) studied the inhibitory effect of FNA on nitrate and nitrite reduction by activated sludge. The authors revealed that nitrate reduction activity was reduced to 40% in the FNA
concentration range of 0.01e0.025 mg HNO2-N/L and completely stopped at the FNA concentration of 0.2 mg HNO2-N/L. Further, nitrite reduction or denitritation was also significantly inhibited at the same range of FNA concentrations. Thus, it appears FNA exhibited a similar inhibitory effect on nitrate and nitrite reducers.
2.3.
FNA and N2O emission
N2O is a by-product of denitrification, and can be produced through heterotrophic denitrification by denitrifiers as well as autotrophic denitrification by nitrifiers (Chung and Chung, 2000; Fux and Siegrist, 2004; Kishida et al., 2004; Shiskowski and Mavinic, 2006). The production of N2O may be affected by many parameters, such as sludge age, organic loading, types of organic carbon source, DO, pH and temperature (Brenner and Argaman, 1990). It has been frequently observed that the accumulation of nitrite leads to an increase in the N2O emission in both nitrification and denitrification processes (von Schulthess et al., 1994; Kampschreur et al., 2009). Betlach and Tiedje (1981) showed that the addition of nitrite does not always cause N2O accumulation, and the two denitrifying cultures in their study were not inhibited by the addition of nitrite at the concentration of 3.86 mg NO 2 -N/L, although the pH was not mentioned in this study. Likewise, Thorn and Sorensson (1996) showed that N2O production by denitrifying activated sludge was negligible at the nitrite concentration of 5e10 mg NO 2 -N/L, while low pH had a strong inhibitory effect, although it was not tested if nitrite accumulation accompanied the N2O production at low pH. von Schulthess et al. (1994) recommended that nitrite concentrations higher than 2 mg NO 2 -N/L should be avoided to prevent N2O emissions, although they did highlight that N2O production was triggered by the presence of oxygen in their denitrifying culture, in combination with nitrite. Similarly, nitrite inhibition also affected nitrifying sludge more severely at low DO levels; although it has been observed the addition of 10 mg NO 2 -N/L had significantly increased the N2O emission in a DO range of 0.1e2.0 mg O2/L (Tallec et al., 2006). Zhou et al. (2008b) reported that FNA was responsible for N2O emission from a denitrifying P-removal culture, not nitrite or pH alone. They revealed that inhibition occurred at a very low concentration of FNA inhibition at 0.0007e 0.001 mg HNO2-N/L or a nitrite concentration of approximately 3e4 mg NO 2 -N/L at pH 7. It is possible that FNA, rather than the nitrite anion, was the main factor affecting N2O production by nitrifiers and ordinary denitrifiers as well, a hypothesis which requires further confirmation.
2.4. FNA inhibition on enhanced biological phosphorus removal (EBPR) bacteria 2.4.1. FNA inhibition on polyphosphate accumulating organisms (PAOs) EBPR is carried out by polyphosphate accumulating organisms (PAOs), which take up organic carbon and release phosphorus under anaerobic conditions and take up phosphorus in excess under aerobic and anoxic conditions. EBPR plants are typically designed for nitrogen removal as well. Due to the limited carbon sources in sewage and large costs associated with
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 6 7 2 e4 6 8 2
aeration in BNR processes, combining nitritation and denitritation processes with phosphorus removal may be advantageous. However, many studies have reported that nitrite in the form of FNA has a negative impact on aerobic P-uptake. A severe inhibition on aerobic P-uptake by an enriched PAO culture was observed at FNA concentration as low as 0.5 103 mg HNO2-N/L (Saito et al., 2004). The same FNA concentration was reported to inhibit aerobic P-uptake by 50% in a highly enriched Candidatus Accumulibacter phosphatis (Accumulibacter e a well-known PAO) culture (Pijuan et al., 2010). Additionally, Pijuan et al. (2010) had shown that both anabolic (growth, P-uptake and glycogen production) and catabolic (PHA oxidation) processes were negatively affected by FNA to different extents. Long-term inhibition from FNA may cause the failure of EBPR systems (Soejima et al., 2006; Pijuan et al., 2010; Zeng et al., 2011). Anoxic phosphorus uptake by PAOs was also significantly inhibited by the presence of a high level of FNA (Zhou et al., 2008a). Zhou et al. (2007) reported that the anoxic P-uptake activity of PAO was completely ceased at a FNA concentration of 0.02 mg HNO2-N/L, which is substantially higher than that observed for aerobic P-uptake (0.004 mg HNO2-N/L; see Pijuan et al., 2010). Further, it has also been reported that the denitrification ability of PAOs was less affected by FNA as compared to anoxic P-uptake (Zhou et al., 2007), while aerobic denitrification rates increased as FNA levels increased (Pijuan et al., 2010). Thus, PAO are more sensitive to FNA under aerobic as compared to anoxic conditions, and likely denitrify aerobically as a detoxification mechanism rather than for energy generation (Pijuan et al., 2010). Anaerobic metabolism of P-release without external carbon source was observed by both Meinhold et al. (1999) and Zhou et al. (2007) when higher FNA concentrations were used. Faster P-release occurred with increased FNA concentrations. It was confirmed that this P-release was not due to cell lysis given the fact that PHA consumption rates were constant in the high FNA range (Zhou et al., 2010). It is possible that the energy generation processes were severely affected and P-release would provide additional energy for maintenance.
2.4.2. FNA inhibition on glycogen accumulating organisms (GAOs) and their competition with PAOs GAOs are able to proliferate under alternating anaerobic and aerobic conditions, and compete with PAOs for organic carbon uptake anaerobically. GAOs’ metabolism is very similar to that of PAOs’, except that they do not cycle polyphosphate, and thus do not contribute to P-removal. According to Ye et al. (2010), FNA also affected the performance of a highly enriched culture of Candidatus Competibacter phosphatis (Competibacter e a well-known GAO). However, compared to the effect on PAOs, FNA has a much lesser inhibitory effect on GAOs, notably with respect to their aerobic glycogen synthesis rate (Pijuan et al., 2010; Ye et al., 2010), their key storage polymer used to generate energy for VFA uptake under anaerobic conditions. Overall, results obtained thus far suggest FNA provides an advantage for GAOs to out-compete PAOs in EBPR plants and leads to the failure of P-removal (Pijuan et al., 2010). Nevertheless, it is noteworthy that the PAOs and GAOs in these studies were not adapted to nitrite. Yoshida et al. (2006) reported that PAOs that have been acclimatised to anoxic conditions are more able to withstand nitrite inhibition. The
4675
inhibition was observed to be stronger on aerobic P-uptake from an anaerobic/aerobic sludge (72%) than on an anaerobic/anoxic sludge (20%) at an equivalent FNA concentration of 0.7 103 mg HNO2-N/L. The tolerance of nitrite-adapted GAOs to FNA inhibition and their competition with PAOs in EBPR processes therefore remains unclear and requires further study.
3. Nitrite and FNA effects on other microorganisms in WWTPs 3.1.
Anammox bacteria
The anaerobic ammonia oxidation (anammox) process has attracted much attention in recent years, since it achieves N removal from wastewater with reduced aeration requirements, no external carbon addition and lower sludge production (Jetten and Van Loosdrecht, 1998). Anammox is considered preferable for high-strength NHþ 4 wastewater (especially with low C) including the supernatant from sludge digesters, landfill leachates and industrial wastewater (Wett, 2006; van der Star et al., 2007; Ganigue et al., 2009; Waki et al., 2010; Tokutomi et al., 2011). Two aspects are essential to achieve a functioning anammox process. Firstly, NOB must be continuously suppressed or completely removed from the sludge, and secondly the nitrite/ammonia ratio produced must be about 1.3 (Fux and Siegrist, 2004). It has not yet been conclusively established that FNA, rather than nitrite, is the true inhibitor on anammox bacteria. Strous et al. (1999) showed that nitrite inhibition was independent of pH over the relatively narrow range of 7.0e7.8. Meanwhile, Egli et al. (2001) showed that the anammox activity at a constant nitrite feeding level was completely inhibited at pH 6.0 and 6.5, and optimal at higher pH (7.5 and 8.0), which suggests that FNA was in fact the inhibitor. Further research would be valuable to clarify this point. FNA inhibition on NOB can be helpful to produce a desirable mixture of nitrite and ammonium for the anammox process. However, even very low concentrations of nitrite/FNA can be an inhibitor to anammox bacteria. The activity of an enriched anammox culture containing Candidatus Brocadia anammoxidans was completely inhibited by FNA at a very low concentration of 6.0 103 mg HNO2-N/L (Strous et al., 1999). Another enriched culture containing the anammox bacterium Kuenenia stuttgartiensis, lost its activity completely at a higher FNA concentration of 0.04 mg HNO2-N/L (Egli et al., 2001). The low inhibitory threshold of FNA on at least some anammox bacteria indicates that they are likely more sensitive to FNA than NOB. Hence, stable anammox performance will rely on the FNA concentration in the influent and its feeding rate, which should be well controlled to not exceed the inhibitory threshold of anammox bacteria.
3.2.
Methanogenic bacteria
In the anaerobic digester, excess activated sludge is degraded to produce biogas through methanogenesis. Methanogenic bacteria is very sensitive to environmental conditions, e.g. long chain fatty acid concentration, free ammonia (FA) concentration, pH, and temperature (Zeeman et al., 1985;
4676
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 6 7 2 e4 6 8 2
Rinzema et al., 1994; Chen et al., 2008). Although nitrite and FNA are not generated in the anaerobic digester, the mixed liquor transferred from the BNR process may bring in certain levels of N-compounds. It has been reported that nitrite can inhibit methanogenesis’ methane production in both reversible and irreversible ways, depending on the type of methanogenic bacteria and the nitrite concentrations (Kluber and Conrad, 1998). The activity of Methanosarcina barkeri and Methanobacterium bryantii was reduced to 50% by 0.7 and 14 mg NO 2N/L of nitrite, respectively. Methane production of a mixed methanogenic culture was inhibited by 80% at a nitrite concentration of 50 mg NO 2 -N/L and the recovery of methanogenesis only occurred 7 days after complete nitrite reduction once exposed to a nitrite concentration of 250 mg NO 2 -N/L (Tugtas and Pavlostathis, 2007). Whether this inhibitory effect is also attributable to FNA, rather than the nitrite anion, requires further research. It should be noted that FNA has been found to be the main cause of reducing methanogenic activity as well as sulfate reduction in sewer systems, reducing both greenhouse gas emissions from sewers as well as the production of corrosive sulfides (Jiang et al., 2011).
3.3.
Pathogens and yeast
On a potentially positive note, FNA has been reported to have an inhibitory effect on pathogens, which can be a useful way for removal of pathogens from wastewater discharges. Carlsson et al. (2002) concluded that urinary pathogens were markedly inhibited by FNA. Both Lambert and Bidlas (2007) and Schlag et al. (2007) showed that FNA inhibited pathogens and its biofilm formation, due to the link between nitrite accumulation and environmental pH on the inhibition. In addition, FNA inhibition has been shown to inhibit yeasts, which may cause sludge bulking problems in WWTPs (Hinze and Holzer, 1985, 1986; Mortensen et al., 2008). The inhibitory thresholds of FNA on pathogens and yeasts and the potential to exploit this effect for controlling sludge bulking and improving wastewater disinfection require further investigation.
4.
FNA inhibition mechanisms
electron transfer), stimulating ATP hydrolysis and inhibiting various exchange reactions catalyzed by the ATPase (Rottenberg, 1990). A protonophore is an uncoupler that increases proton permeability through cell membranes, which counteracts the proton pumping effect of the ATPase, inhibiting ATP synthesis through oxidative phosphorylation. Almeida et al. (1995a) and Sijbesma et al. (1996) had confirmed FNA acted as a protonophore using P. fluorescens. They proposed that the decreased growth is due to the expenditure of a higher proportion of energy to pump protons out of the cells after FNA had diffused through the cell membrane, due to the increased proton permeability (Fig. 1). Mortensen et al. (2008) had found that the intracellular pH of Debaryomyces hansenii decreased drastically within 1 min of exposure to nitrite. The plasma membrane ATPase activity of the cells was much lower compared to cases where there was nitrite absence. They explained that the intracellular acidification could be due to, in part, the uncoupling of energy generation and growth caused by FNA. This uncoupling seemingly occurs when the amount of energy for maintenance purposes has exceeded a certain minimum value. This higher maintenance energy requirement agrees well with the study of Sijbesma et al. (1996), and is likely related to the increased proton permeability caused by FNA. Thus, maintenance energy is a function of the FNA concentration, affecting the growth rate after reaching a threshold level. The protonophore hypothesis agrees well with other studies showing that internal ATP depletion increases as a function of FNA. Rake and Eagon (1980) earlier proposed that the toxicity of nitrite may completely inhibit ATP synthesis and cause rapid loss of intracellular ATP pools. Increased cell death caused by nitrite is probably a consequence of the ATP depletion as well (Schimz, 1980). Zhou et al. (2007, 2010) reported that the intracellular ATP level of Accumulibacter was severely affected by the exposure to FNA. With the incubation time increasing, the intracellular ATP level decreased and the decreasing rates depended on the FNA concentrations. The electron transport activity was reported to be inhibited as well (Zhou et al., 2010), further supporting this hypothesised mechanism.
Experimental data has shown FNA has an inhibitory effect on a very broad range of microorganisms (Vadivelu et al., 2006a, b; Glass et al., 1997; Zhou et al., 2007, 2008a, b, 2010; Ye et al., 2010; Zhang et al., 2010). The inhibition can affect numerous metabolic processes, including the active transport of substrates across the cell membrane, oxygen uptake, as well as oxidative phosphorylation. It has also been reported that FNA has a stronger inhibitory effect on the energy-consuming anabolic processes of many organisms, rather than on the catabolic processes that generate ATP, although both can be influenced (Vadivelu et al., 2006a, b; Zhou et al., 2010; Ye et al., 2010; Pijuan et al., 2010). While inhibitory mechanisms require further clarification, the following biochemical explanations have been proposed.
4.1.
FNA as an uncoupler, affecting energy generation
An uncoupler is an agent that stimulates basal electron transport, inhibiting ATP synthesis (without inhibiting
Fig. 1 e Illustration of ATP production and proton transportation in a denitrifying bacterial cell. E.T.C., electron transport chain; pmf, proton motive force.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 6 7 2 e4 6 8 2
4677
Nevertheless, as Sijbesma et al. (1996) point out, this FNA inhibition mechanism does not necessarily occur in every organism. Alefounder et al. (1983) showed that there was no significant uncoupling affect on the proton motive force (pmf) in Paracoccus denitrificans. Using Nitrobacter, Vadivelu et al. (2006b) reported that although FNA inhibited the growth severely, the energy generation capacity of Nitrobacter was not affected. This suggested FNA may not inhibit ATP generation metabolism of all microorganisms and the FNA uncoupling effect alone may not be the sole mechanism of FNA inhibition.
4.2.
FNA effect on enzymes
FNA may force bacteria to turn on or off particular enzymes to defend against its toxicity. Beaumont et al. (2004) had reported the concentration of FNA was the main environmental variable that controlled the expression of nitrite reductase NirK in N. europaea. The NsrR gene that regulates NirK expression is very sensitive to FNA, rather than pH alone (Beaumont et al., 2004). FNA may also affect the gene transcriptional process and further mislead the enzyme assemblage. Baumann et al. (1997) suggested that FNA inhibition on denitrification activity could either be because the translation of nitrite reductase mRNA was inhibited, or the enzyme was translocated and folded improperly. Although the nitrite reductase gene was induced properly, the enzyme could not be detected at sufficient amounts in the culture. The nitrite reductase concentration at lower pH (6.8) was 10e15 times lesser as compared to that at pH 7.5. Therefore, they proposed that the inhibition could be due to the inactivation of the already synthesized enzyme and/or a conformational change by the high concentration of FNA. FNA may directly react with the enzymes involved in the metabolic processes as well. For example, N2O reductase contains two metal centres, a binuclear copper centre, CuA, that serves to receive electrons from soluble donors, and a tetranuclear copper-sulfide centre, Cuz, at the active site (Rasmussen et al., 2005). HNO2 could bind to the active sites of copper-containing enzymes, causing competitive inhibition to N2O reduction. Sulfhydryl (SH)-containing enzymes are key regulators of the tricarboxylic acid (TCA) cycle, and have also been found to be inhibited through reaction with FNA (O’Leary and Solberg, 1976; Park, 1993) (Fig. 2). This reaction would negatively impact the energy generation process of cells, since the NADH produced through the TCA is then converted to ATP. FNA can also react with and inactivate glyceraldehyde-3phosphate dehydrogenase, an enzyme involved in both glycolysis and gluconeogenesis (Hinze and Holzer, 1986), which could potentially explain the findings of Zhou et al. (2010). They observed that glycogen production (i.e. through gluconeogenesis) was severely inhibited under anoxic conditions and glycogen degradation (i.e. through glycolysis) was not fully recovered under anaerobic conditions even after FNA was removed from the bulk media. Furthermore, the Calvin cycle, which is used for carbon fixation in AOBs (Chain et al., 2003; Stein et al., 2007; Norton et al., 2008), yields glyceraldehyde-3-phosphate that is then consumed during biosynthesis. It is possible that this anabolic pathway is also affected by FNA due to enzymatic reaction, resulting in inhibition of Nitrosomonas growth (Vadivelu et al., 2006a).
Fig. 2 e FNA reacting with SH groups to form S-nitrosothiols, resulting in the decrease of free SH groups in the liquid phase (Park, 1993).
4.3.
Nitric oxide (NO) inhibition
von Schulthess et al. (1995) had reported that when a denitrifying activated sludge reactor was saturated with nitrite, NO accumulation was observed. This was hypothesised to be due to the inactivation of the NO reductase by nitrite, which consequently resulted in NO accumulation, subsequently causing inhibition to both the nitrite and nitrous oxide reductase enzymes. NO, which is the product of HNO2 reduction, is thought to react directly with heme and metal centres of proteins, forming nitrosyl complexes (Reddy et al., 1983) (Fig. 3). Because of the function of the copper-sulfide proteins in electron transport and ATP generation in anaerobic bacteria, the formation of the complex, and hence destruction of the catalytic site on the enzymes, would almost certainly inhibit electron transport and growth. Nitric oxide also reacts with the oxygen respiration reductase and its reaction product inhibits oxygen respiration (Casey et al., 1999). Carlsson et al. (2002) concluded that the bacteriostatic effect of acidified nitrite (FNA) is likely related to the release of NO and other toxic reactive nitrogen intermediates. It has been reported that polyphosphate glucokinase and adenylate kinase activity was severely inhibited by NO during the anaerobic polyphosphate degradation process in a P-removing sludge (van Niel et al., 1998). Nevertheless, while
Fig. 3 e Interconversion of different redox forms of nitric oxide in reactions with biologically relevant reducers and oxidants (M [ Fe, Cu, Co). Adapted from Stamler et al. (1992).
4678
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 6 7 2 e4 6 8 2
NO can be directly inhibitory, it does not necessarily accumulate at high FNA concentrations. Under anoxic conditions, Zhou et al. (2008b) reported that there was no nitric oxide produced in their study of a P-removing sludge, suggesting that other inhibitory mechanism(s) triggered by FNA were involved during polyphosphate formation. The extent to which each inhibition mechanism plays a role during FNA inhibition of the different processes relevant to wastewater treatment is still unclear and requires further research. It is, however, clear that the inhibitory effect of FNA may not be due to a single mechanism or pathway and varies according to the microorganism.
5. FNA inhibition implications to the operation of WWTPs The impact of FNA inhibition is clearly dependent on the microbial community present in the sludge. Table 1 summarizes the lowest FNA and nitrite (pH 7) concentrations that have been reported to inhibit the key microbial populations involved in WWTPs by 50% and 100%, respectively. It shows that nitriteadapted microorganisms, i.e. AOB, NOB, denitrifiers and PAO (anoxically) are generally more tolerant to FNA inhibition. It appears that acclimation to FNA is necessary in order to build up tolerance to high FNA concentrations. The presence of FNA may reconfigure the functional structure of the microbial community, and can potentially provide advantages towards undesirable competitor microorganisms. For example, while denitrifiers are less susceptible to FNA, high levels could induce P-removal failure through competition with e.g. ordinary denitrifiers or GAOs. Table 1 also indicates that the lowest reported FNA concentration to inhibit NOB by 100% is 0.02 mg HNO2-N/L. Given the sensitivity of N2O reduction as well as anammox, P-removal, and methanogenic populations to FNA, this FNA level could inhibit their activity completely as well. It is clear that in most sewage treatment plants, using FNA inhibition alone to achieve the nitrite shortcut is not feasible, since the ammonia concentration in municipal sewage is normally below 50 mg NHþ 4 -N/L. The accumulated nitrite concentration would hardly reach the inhibition threshold of NOB (Table 1, assuming the operating pH is close to 7). Therefore, FNA inhibition should be combined with other approaches, e.g. implementation of an on-line DO control system, in order to
limit the growth of NOB. Selecting against NOB through FNA inhibition alone is more readily feasible when treating highstrength wastewater. Therefore, designing wastewater treatment plants that incorporate shortcut N removal as well as avoiding FNA inhibition represents a significant challenge. Future research should determine if these previously observed FNA threshold limits can be lowered through acclimation to FNA, as well as other operational strategies, thereby mitigating these potentially negative effects. While FNA adapted microbes may have stronger tolerance to FNA, it would require a relatively longer startup period to acclimate the microbial communities, a situation not always feasible in WWTPs due to the variable nature of the influent and the potential for disturbances (e.g. from industrial discharges). Such shock loads can also lead to situations of FNA-induced inhibition (Freitas et al., 2009). Other process control and optimisation strategies are also needed to reduce the negative impacts of FNA. The most obvious way to reduce the FNA concentration in WWTPs with nitrite accumulation is to increase the pH. By increasing the pH from 7 to 8, the FNA concentration can be reduced by 90%. Changing the pH will also have a simultaneous effect on the microbial population selection and efficiency of the different functional groups of microorganisms present in the sludge. Nevertheless, high pH (w8.0) has been reported to be beneficial for nitrification (Han et al., 2010) and also in favouring PAOs over GAOs in EBPR systems (Oehmen et al., 2005). It should be noted that excessively high pH (e.g. >8.5) is harmful to many bacteria in WWTPs and that the concentration of FA increases with increasing pH. While pH control is a beneficial strategy to reduce FNA inhibition and improve other sludge characteristics, it is only rarely applied in full-scale plants, likely due to the associated costs of adding pH controlling agents. To a lesser extent as compared to pH, temperature also impacts the concentration of FNA, where lower temperature increases FNA levels. FNA concentrations are approximately 20e25% higher at 10 C as compared to 20 C, and likewise for 20 C vs. 30 C. This implies that colder climates or winter conditions are more likely to lead to FNA inhibition. Temperature control of WWTPs is not practically feasible, however, due to the associated cost. Other potential strategies of minimising the inhibitory effects of FNA include the feeding rate. It has been reported
Table 1 e FNA inhibition thresholds on functional microbial consortiums in wastewater treatment plants. The corresponding nitrite concentrations at pH [ 7 are also shown. Culture AOB NOB Denitrifiers Anammox bacteria PAO (AnO2-Ox) PAO (AnO2-Ax) GAO (AnO2-Ox) N2O reduction (PAO AnO2-Ax) Methanogens
FNA (100%) (mg HNO2-N/L)
Nitrite at pH 7 (mg NO 2 -N/L)
FNA (50%) (mg HNO2-N/L)
Nitrite at pH 7 (mg NO 2 -N/L)
Reference
e 0.023 0.2 0.006 0.004 0.037 0.02 (70%) 0.004 0.0003
e 100 900 27 18 160 90 18 1.3
0.2 0.0175 0.025 e 0.0005 0.01 0.01 0.0007 0.00015
900 76 108 e 2 45 45 3 0.65
Hellinga et al. (1999) Vadivelu et al. (2006b) Ma et al. (2010) Strous et al. (1999) Pijuan et al. (2010) Zhou et al. (2010) Ye et al. (2010) Zhou et al. (2008b) Kluber and Conrad (1998)
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 6 7 2 e4 6 8 2
that continuous and step-feeding modes can greatly reduce the FNA inhibition effect as compared to dump-feeding. Zhou et al. (2008a) demonstrated for a denitrifying P-removal system treating abattoir wastewater that P-removal and denitrification were significantly improved using a continuous nitrite feeding strategy as compared with a dump-feed. By continuously feeding the 170 mg NO 2 -N/L influent at a controlled flow rate, nitrite was hardly detected in the anoxic phase. Meanwhile, the N2O emission problem was also solved due to the undetected level of FNA. Vargas et al. (2011) reported a P-removal system that was successfully transformed from using oxygen to nitrite as sole electron acceptor for P-removal and denitrification by introducing nitrite through separated pulses to reduce the inhibition from FNA. Therefore, by controlling the feed flow and mode, it is possible to eliminate FNA inhibition. In full-scale continuous-flow systems, a decrease in FNA concentration can also be achieved through increasing the recycle ratio. Zhou et al. (2007) also proposed that granular P-removal sludge was less inhibited by FNA as compared to flocculant Premoval sludge, due to mass transfer limitations. Therefore, granular sludge or biofilm-based systems may be able to tolerate higher FNA concentrations. However, FNA, as a chemical uncoupler of energy production and growth, could severely inhibit granule and biofilm formation as shown for other uncouplers (Jiang and Liu, 2010; Xu and Liu, 2011). The threshold of FNA inhibition in granulation and biofilm formation processes as well as their resistance to FNA in already existing granules and biofilms requires further research. Furthermore, the potential use of FNA to control biofouling in Membrane BioReactor (MBR) systems should be investigated, since other energy and growth uncouplers have recently been shown to be beneficial for this purpose (Xu and Liu, 2011). Mathematical modelling is a useful tool for predicting the inhibition effect of FNA on microbial communities and the population shifts caused by the inhibition. This can be helpful in controlling WWTP processes. Using modelling and process control, Bernet et al. (2005), Jubany et al. (2009) and Park et al. (2010) identified relevant parameters to control (e.g. pH, DO, air flow rate) and monitor the nitrite and FNA accumulation in order to minimise NOB communities and achieve short-cut nitrogen removal. It was shown that the impact of other factors alone (the effect of DO or the effect of DO plus direct pH inhibition) could not provide a strong enough selection for AOB over NOB in a practical setting, while the effects of FA and FNA were required in order to effectively promote this selection (Park et al., 2010). Nitrite and FNA inhibition modelling has also been included in EBPR models (Yoshida et al., 2009; Oehmen et al., 2010). The NO inhibition mechanism was incorporated into Yoshida’s nitrite-complex model, to describe the tolerance of PAO to nitrite inhibition as well as the inhibition after nitrite was fully removed. Oehmen et al. (2010) described the metabolism of PAOs and GAOs under anaerobic/anoxic/aerobic conditions and predicted the microbial community dynamics in EBPR systems, where FNA was considered as one of the important parameters to affect the PAOeGAO competition. According to the state of the art, some sustainable and practically applicable ways to reduce the impact of FNA
4679
inhibition in WWTPs would be the reduction of the nitrite concentration and/or gradient, particularly through controlling the feed and/or recycle flowrates, and implementing granular or biofilm systems. As mentioned above, the fundamental mechanisms of FNA inhibition vary according to the microbial community, their acclimation to FNA and other operational factors. A comprehensive understanding of the mechanisms involved during each bioprocess would further enable development of operating strategies to reduce inhibition and exploit the potential for using FNA to reduce the growth of undesirable microorganisms in WWTPs.
6.
Conclusions
This paper reviews the impact of FNA on WWTPs. Some bioprocesses exploit the impact of FNA in order to optimize the treatment performance, while the sensitivity to FNA inhibition depends on the functional microbial community. The hypothesised mechanisms of FNA inhibition are discussed, as well as strategies to reduce the negative impacts of FNA inhibition. The following aspects require researchers’ attention in the future: - When one considers the inhibition effect from nitrite, it would be helpful to correlate the effect with FNA as well. Further proof of FNA inhibition and inhibition thresholds on anammox bacteria, methanogens, pathogens and yeasts are needed. Aside from nitrite concentration, pH and/or temperature should be varied to investigate the combined effects (FNA) on those microbes. - Most of the inhibition studies are based on microbial communities cultured in the laboratory. Some of these microbes are not commonly found in real WWTPs. Identification of the microbial communities and investigation of the FNA inhibition on those communities abundant in WWTPs is essential. This will involve culture isolation or enrichment from full-scale plants and identifying the response of each culture to FNA inhibition. - It is important to investigate the adaptability of key functional microorganisms to FNA inhibition. Tracking the longterm operational data and changes in community structure would help to monitor the adaption process. Modelling tools could be particularly useful for this research direction. - Continuous feeding or decreased feed rates can greatly reduce the inhibition, however, some processes utilise FNA as the substrate and electron acceptor, thus reducing feeding rates could result in limited reaction kinetics. The balance between maximising loading rate and minimising inhibition should be maintained. - Detailed correlation between FNA with granulation and biofilm formation has not yet been established, though granules/biofilms may be favourable for coping with high FNA concentrations. It would be beneficial to study FNA effects on granule and biofilm formation and disaggregation, particularly with respect to inhibition at the enzyme activity and quorum sensing levels. - The potential exists for FNA to be useful in numerous other beneficial applications in WWTPs besides achieving the nitrite pathway. These include controlling sludge bulking,
4680
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 6 7 2 e4 6 8 2
reducing the growth of pathogens, and biofouling control in MBR systems. - The FNA inhibition mechanisms are inconclusive for most situations. Further study on the single or multiple inhibition effects of FNA for the different relevant microorganisms would help researchers to understand, apply and/or reduce its effect on wastewater treatment processes. For this purpose, pure or highly enriched cultures with identified metabolic pathways and functional genes would be required.
references
Abeling, U., Seyfried, C.F., 1992. Anaerobic-aerobic treatment of high-strength ammonium wastewater e nitrogen removal via nitrite, Washington, DC, USA, pp. 1007e1015. Alefounder, P.R., Greenfielda, A.J., McCarth, J.E.G., Ferguson, S.J., 1983. Selection and organisation of denitrifying electrontransfer pathways in Paracoccus denitrificans. Biochimica et Biophysica Acta (BBA) e Bioenergetics 724 (1), 20e39. Almeida, J.S., Julio, S.M., Reis, M.A.M., Carrondo, M.J.T., 1995a. Nitrite inhibition of denitrification by pseudomonas fluorescens. Biotechnology and Bioengineering 46 (3), 194e201. Almeida, J.S., Reis, M.A.M., Carrondo, M.J.T., 1995b. Competition between nitrate and nitrite reduction in denitrification by Pseudomonas fluorescens. Biotechnology and Bioengineering 46 (5), 476e484. Anthonisen, A.C., Loehr, R.C., Prakasam, T.B.S., Shinath, E.G., 1976. Inhibition of nitrification by ammonia and nitrous acid. Journal Water Pollution Control Federation 48 (5), 835e852. Baumann, B., vanderMee, J.R., Snozzi, M., Zehnder, A.J.B., 1997. Inhibition of denitrification activity but not of mRNA induction in Paracoccus denitrificans by nitrite at a suboptimal pH. Antonie Van Leeuwenhoek International Journal of General and Molecular Microbiology 72 (3), 183e189. Beaumont, H.J.E., Lens, S.I., Reijnders, W.N.M., Westerhoff, H.V., van Spanning, R.J.M., 2004. Expression of nitrite reductase in Nitrosomonas europaea involves NsrR, a novel nitrite-sensitive transcription repressor. Molecular Microbiology 54 (1), 148e158. Beccari, M., Passino, R., Ramadori, R., Tandoi, V., 1983. Kinetics of dissimilatory nitrate and nitrite reduction in suspended growth culture. Journal of Water Pollution Control Federation 55 (1), 58e64. Bernet, N., Sanchez, O., Cesbron, D., Steyer, J.P., Delgenes, J.P., 2005. Modeling and control of nitrite accumulation in a nitrifying biofilm reactor. Biochemical Engineering Journal 24 (2), 173e183. Betlach, M.R., Tiedje, J.M., 1981. Kinetic explanation for the accumulation of nitrite, nitric oxide, and nitric oxide. Applied and Environmental Microbiology 42 (6), 1074e1084. Blackburne, R., Vadivelu, V.M., Yuan, Z.G., Keller, J., 2007. Kinetic characterisation of an enriched Nitrospira culture with comparison to Nitrobacter. Water Research 41 (14), 3033e3042. Brenner, A., Argaman, Y., 1990. Effect of feed composition, aerobic volume fraction and recycle rate on nitrogen removal in the single-sludge system. Water Research 24 (8), 1041e1049. Burrell, P., Keller, J., Blackall, L.L., 1999. Characterisation of the bacterial consortium involved in nitrite oxidation in activated sludge. Water Science and Technology 39 (6), 45e52. Carlsson, S., Wiklund, N.P., Engstrand, L., Weitzberg, E., Lundberg, J.O.N., 2002. Effects of pH, nitrite, and ascorbic acid on nonenzymatic nitric oxide generation and bacterial growth in urine. Nitric Oxide-Biology and Chemistry 5 (6), 580e586.
Casey, T.G., Wentzel, M.C., Ekama, G.A., 1999. Filamentous organism bulking in nutrient removal activated sludge systems. Paper 9: review of biochemistry of heterotrophic respiratory metabolism. Water S.A. 25 (4), 409e424. Chain, P., Lamerdin, J., Larimer, F., Regala, W., Lao, V., Land, M., Hauser, L., Hooper, A., Klotz, M., Norton, J., Sayavedra-Soto, L., Arciero, D., Hommes, N., Whittaker, M., Arp, D., 2003. Complete genome sequence of the ammonia-oxidizing bacterium and obligate chemolithoautotroph Nitrosomonas europaea. Journal of Bacteriology 185 (9), 2759e2773. Chen, S.K., Juaw, C.K., Cheng, S.S., 1991. Nitrification and denitrification of high-strength ammonium and nitrite wastewater with biofilm reactors. Water Science and Technology 23 (7e9), 1417e1425. Chen, Y., Cheng, J.J., Creamer, K.S., 2008. Inhibition of anaerobic digestion process: a review. Bioresource Technology 99 (10), 4044e4064. Chung, Y.C., Chung, M.S., 2000. BNP test to evaluate the influence of C/N ratio on N2O production in biological denitrification. Water Science and Technology 42 (3), 23e27. Ciudad, G., Werner, A., Bornhardt, C., Munoz, C., Antileo, C., 2006. Differential kinetics of ammonia- and nitrite-oxidizing bacteria: a simple kinetic study based on oxygen affinity and proton release during nitrification. Process Biochemistry 41 (8), 1764e1772. Daims, H., Nielsen, J.L., Nielsen, P.H., Schleifer, K.H., Wagner, M., 2001. In situ characterization of Nitrospira-like nitrite oxidizing bacteria active in wastewater treatment plants. Applied and Environmental Microbiology 67 (11), 5273e5284. Egli, K., Fanger, U., Alvarez, P.J.J., Siegrist, H., van der Meer, J.R., Zehnder, A.J.B., 2001. Enrichment and characterization of an anammox bacterium from a rotating biological contactor treating ammonium-rich leachate. Archives of Microbiology 175 (3), 198e207. Freitas, F., Temudo, M.F., Carvalho, G., Oehmen, A., Reis, M.A.M., 2009. Robustness of sludge enriched with short SBR cycles for biological nutrient removal. Bioresource Technology 100 (6), 1969e1976. Fux, C., Siegrist, H., 2004. Nitrogen removal from sludge digester liquids by nitrification/denitrification or partial nitritation/ anammox: environmental and economical considerations. Water Science and Technology 50 (10), 19e26. Ganigue, R., Gabarro, J., Sanchez-Melsio, A., Ruscalleda, M., Lopez, H., Vila, X., Colprim, J., Balaguer, M.D., 2009. Long-term operation of a partial nitritation pilot plant treating leachate with extremely high ammonium concentration prior to an anammox process. Bioresource Technology 100 (23), 5624e5632. Glass, C., Silverstein, J., 1998. Denitrification kinetics of high nitrate concentration water: pH effect on inhibition and nitrite accumulation. Water Research 32 (3), 831e839. Glass, C., Silverstein, J., Oh, J., 1997. Inhibition of denitrification in activated sludge by nitrite. Water Environment Research 69 (6), 1086e1093. Han, X.L., Hai, R.T., Wang, W.X., 2010. Effect of COD/N ratio and pH on nitrification in a laboratory-scale constructed wetlands treating septic tank wastewater. Environmental Engineering and Management Journal 9 (5), 615e621. Hellinga, C., Schellen, A.A.J.C., Mulder, J.W., Van Loosdrecht, M.C. M., Heijnen, J.J., 1998. The SHARON process: an innovative method for nitrogen removal from ammonium-rich waste water. Water Science and Technology 37 (9), 135e142. Hellinga, C., Van Loosdrecht, M.C.M., Heijnen, J.J., 1999. Model based design of a novel process for nitrogen removal from concentrated flows. Mathematical and Computer Modelling of Dynamical Systems 5 (4), 351e371. Hinze, H., Holzer, H., 1985. Effect of sulfite or nitrite on the ATP content and the carbohydrate-metabolism in yeast. Zeitschrift
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 6 7 2 e4 6 8 2
Fur Lebensmittel-Untersuchung Und-Forschung 181 (2), 87e91. Hinze, H., Holzer, H., 1986. Analysis of the energy-metabolism after incubation of Saccharomyces-cerevisiae with sulfite or nitrite. Archives of Microbiology 145 (1), 27e31. Jetten, M.S.M., Van Loosdrecht, M.C.M., 1998. Method of treating ammonia-comprising waste water, Netherlands, Patent PCT/ NL1997/000482. Jiang, B., Liu, Y., 2010. Energy uncoupling inhibits aerobic granulation. Applied Microbiology and Biotechnology 85 (3), 589e595. Jiang, G, Gutierrez, O, Yuan, Z, 2011. The strong biocidal effect of free nitrous acid on anaerobic sewer biofilms. Water Research 45 (12), 3735e3743. Jubany, I., Lafuente, J., Baeza, J.A., Carrera, J., 2009. Total and stable washout of nitrite oxidizing bacteria from a nitrifying continuous activated sludge system using automatic control based on oxygen uptake rate measurements. Water Research 43 (11), 2761e2772. Kampschreur, M.J., Temmink, H., Kleerebezem, R., Jetten, M.S.M., van Loosdrecht, M.C.M., 2009. Nitrous oxide emission during wastewater treatment. Water Research 43 (17), 4093e4103. Kelly, D.P., 1978. Bioenergetics of chemolithotrophic bacteria. In: Bull., A.T., Meadow, P.M. (Eds.), Companion to Microbiology. Longman, New York, pp. 363e386. Kishida, N., Kim, J.H., Kimochi, Y., Nishimura, O., Sasaki, H., Sudo, R., 2004. Effect of C/N ratio on nitrous oxide emission from swine wastewater treatment process. Water Science and Technology 49 (5, 6), 359e365. Kluber, H.D., Conrad, R., 1998. Inhibitory effects of nitrate, nitrite, NO and N2O on methanogenesis by Methanosarcina barkeri and Methanobacterium bryantii. Fems Microbiology Ecology 25 (4), 331e339. Kuai, L., Verstraete, W., 1998. Ammonium removal by the oxygenlimited autotrophic nitrificationedenitrification system. Applied and Environmental Microbiology 64 (11), 4500e4506. Lambert, R.J.W., Bidlas, E., 2007. Gamma study of pH, nitrite, and salt inhibition of Aeromonas hydrophila. Applied and Environmental Microbiology 73 (7), 2239e2246. Ma, J., Yang, Q., Wang, S., Wang, L., Takigawa, A., Peng, Y., 2010. Effect of free nitrous acid as inhibitors on nitrate reduction by a biological nutrient removal sludge. Journal of Hazardous Materials 175 (1e3), 518e523. Meinhold, J., Arnold, E., Isaacs, S., 1999. Effect of nitrite on anoxic phosphate uptake in biological phosphorus removal activated sludge. Water Research 33 (8), 1871e1883. Meng, X.Z., Qian, D., Cao, X.S., 2010. Nitrite accumulation during wastewater denitrification. International Conference on Electrical and Control Engineering, pp. 4721e4724. Mortensen, H.D., Jacobsen, T., Koch, A.G., Arneborg, N., 2008. Intracellular pH homeostasis plays a role in the tolerance of Debaryomyces hansenii and Candida zeylanoides to acidified nitrite. Applied and Environmental Microbiology 74 (15), 4835e4840. Norton, J.M., Klotz, M.G., Stein, L.Y., Arp, D.J., Bottomley, P.J., Chain, P.S.G., Hauser, L.J., Land, M.L., Larimer, F.W., Shin, M. W., Starkenburg, S.R., 2008. Complete genome sequence of Nitrosospira multiformis, an ammonia-oxidizing bacterium from the soil environment. Applied and Environmental Microbiology 74 (11), 3559e3572. O’Leary, V., Solberg, M., 1976. Effect of sodium nitrite inhibition on intracellular thiol groups and on the activity of certain glycolytic enzymes in Clostridium perfringens. Applied and Environmental Microbiology 31 (2), 208e212. Oehmen, A., Lopez-Vazquez, C.M., Carvalho, G., Reis, M.A.M., van Loosdrecht, M.C.M., 2010. Modelling the population dynamics and metabolic diversity of organisms relevant in anaerobic/ anoxic/aerobic enhanced biological phosphorus removal processes. Water Research 44 (15), 4473e4486.
4681
Oehmen, A., Vives, M.T., Lu, H., Yuan, Z., Keller, J., 2005. The effect of pH on the competition between polyphosphateaccumulating organisms and glycogen-accumulating organisms. Water Research 39 (15), 3727e3737. Park, J.W., 1993. S-nitrosylation of sulfhydryl groups in albumin by nitrosating agents. Archives of Pharmacal Research 16 (1), 1e5. Park, S., Bae, W., Rittmann, B.E., 2010. Operational boundaries for nitrite accumulation in nitrification based on minimum/ maximum substrate concentrations that include effects of oxygen limitation, pH, and free ammonia and free nitrous acid inhibition. Environmental Science and Technology 44 (1), 335e342. Philips, S., Laanbroek, H.D., Verstraete, W., 2002. Origin, causes and effects of increased nitrite concentrations in aquatic environments. Reviews in Environmental Science and Biotechnology 1 (2), 115e141. Pijuan, M., Ye, L., Yuan, Z., 2010. Free nitrous acid inhibition on the aerobic metabolism of poly-phosphate accumulating organisms. Water Research 44 (20), 6063e6072. Prakasam, T.B.S., Loehr, R.C., 1972. Microbial nitrification and denitrification in concentrated wastes. Water Research 6 (7), 859e869. Rake, J.B., Eagon, R.G., 1980. Inhibition, but not uncoupling, of respiratory energy coupling of three bacterial species by nitrite. Journal of Bacteriology 144 (3), 975e982. Rasmussen, T., Brittain, T., Berks, B.C., Watmough, N.J., Thomson, A.J., 2005. Formation of a cytochrome c-nitrous oxide reductase complex is obligatory for N2O reduction by Paracoccus pantotrophus. Dalton Transactions 21, 3501e3506. Reddy, D., Lancaster, J.R., Cornforth, D.P., 1983. Nitrite inhibition of Clostridium botulimun: electron spin resonance detection of iron-nitric oxide complexes. Science 221 (4612), 769e770. Rinzema, A., Boone, M., Van Knippenberg, K., Lettinga, G., 1994. Bactericidal effect of long chain fatty acids in anaerobic digestion. Water Environment Research 66 (1), 40e49. Rottenberg, H., 1990. Decoupling of oxidative-phosphorylation and photophosphorylation. Biochimica Et Biophysica Acta 1018 (1), 1e17. Saito, T., Brdjanovic, D., Van Loosdrecht, M.C.M., 2004. Effect of nitrite on phosphate uptake by phosphate accumulating organisms. Water Research 38 (17), 3760e3768. Schimz, K.L., 1980. The effect of sulfite on the yeast SaccharomycesCerevisiae. Archives of Microbiology 125 (1, 2), 89e95. Schlag, S., Nerz, C., Birkenstock, T.A., Altenberend, F., Gotz, F., 2007. Inhibition of staphylococcal biofilm formation by nitrite. Journal of Bacteriology 189 (21), 7911e7919. Shiskowski, D.M., Mavinic, D.S., 2006. The influence of nitrite and pH (nitrous acid) on aerobic-phase, autotrophic N2O generation in a wastewater treatment bioreactor. Journal of Environmental Engineering and Science 5 (4), 273e283. Sijbesma, W.F.H., Almeida, J.S., Reis, M.A.M., Santos, H., 1996. Uncoupling effect of nitrite during denitrification by pseudomonas fluorescens: an in vivo 31P-NMR study. Biotechnology and Bioengineering 52 (1), 176e182. Soejima, K., Oki, K., Terada, A., Tsuneda, S., Hirata, A., 2006. Effects of acetate and nitrite addition on fraction of denitrifying phosphate-accumulating organisms and nutrient removal efficiency in anaerobic/aerobic/anoxic process. Bioprocess and Biosystems Engineering 29 (5, 6), 305e313. Stamler, J.S., Singel, D.J., Loscalzo, J., 1992. Biochemistry of nitric oxide and its redox-activated forms. Science 258 (5090), 1898e1902. Stein, L.Y., Arp, D.J., 1998. Loss of ammonia monooxygenase activity in Nitrosomonas europea upon exposure to nitrite. Applied and Environmental Microbiology 64 (10), 4098e4102. Stein, L.Y., Arp, D.J., Berube, P.M., Chain, P.S.G., Hauser, L., Jetten, M.S.M., Klotz, M.G., Larimer, F.W., Norton, J.M., den Camp, H., Shin, M., Wei, X.M., 2007. Whole-genome analysis
4682
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 6 7 2 e4 6 8 2
of the ammonia-oxidizing bacterium, Nitrosomonas eutropha C91: implications for niche adaptation. Environmental Microbiology 9 (12), 2993e3007. Strous, M., Kuenen, J.G., Jetten, M.S.M., 1999. Key physiology of anaerobic ammonium oxidation. Applied and Environmental Microbiology 65 (7), 3248e3250. Tallec, G., Garnier, J., Billen, G., Gousailles, M., 2006. Nitrous oxide emissions from secondary activated sludge in nitrifying conditions of urban wastewater treatment plants: effect of oxygenation level. Water Research 40 (15), 2972e2980. Tan, N.C.G., Kampschreur, M.J., Wanders, W., van der Pol, W.L.J., van de Vossenberg, J., Kleerebezem, R., van Loosdrecht, M.C.M., Jetten, M.S.M., 2008. Physiological and phylogenetic study of an ammonium-oxidizing culture at high nitrite concentrations. Systematic and Applied Microbiology 31 (2), 114e125. Tchobanoglous, G., Burton, F.L., Stensel, H.D., 2003. In: Wastewater Engineering: Treatment and Reuse, fourth ed. Metcalf & Eddy Inc., McGraw-Hill Science Engineering, New York, USA. Thorn, M., Sorensson, F., 1996. Variation of nitrous oxide formation in the denitrification basin in a wastewater treatment plant with nitrogen removal. Water Research 30 (6), 1543e1547. Tokutomi, T., Yamauchi, H., Nishimura, S., Yoda, M., Abma, W., 2011. Application of the nitritation and anammox process into inorganic nitrogenous wastewater from semiconductor factory. Journal of Environmental Engineering 137 (2), 146e154. Tora, J.A., Lafuente, J., Baeza, J.A., Carrera, J., 2010. Combined effect of inorganic carbon limitation and inhibition by free ammonia and free nitrous acid on ammonia oxidizing bacteria. Bioresource Technology 101 (15), 6051e6058. Tugtas, A.E., Pavlostathis, S.G., 2007. Inhibitory effects of nitrogen oxides on a mixed methanogenic culture. Biotechnology and Bioengineering 96 (3), 444e455. Turk, O., Mavinic, D.S., 1986. Preliminary assessment of a shortcut in nitrogen removal from wastewater. Canadian Journal of Civil Engineering 13 (6), 600e605. Vadivelu, V.M., Keller, J., Yuan, Z., 2006a. Effect of free ammonia and free nitrous acid concentration on the anabolic and catabolic processes of an enriched Nitrosomonas culture. Biotechnology and Bioengineering 95 (5), 830e839. Vadivelu, V.M., Yuan, Z., Fux, C., Keller, J., 2006b. The inhibitory effects of free nitrous acid on the energy generation and growth processes of an enriched Nitrobacter culture. Environmental Science and Technology 40 (14), 4442e4448. van der Star, W.R.L., Abma, W.R., Blommers, D., Mulder, J.W., Tokutomi, T., Strous, M., Picioreanu, C., van Loosdrecht, M.C. M., 2007. Startup of reactors for anoxic ammonium oxidation: experiences from the first full-scale anammox reactor in Rotterdam. Water Research 41 (18), 4149e4163. van Niel, E.W.J., Appeldoorn, K.J., Zehnder, A.J.B., Kortstee, G.J.J., 1998. Inhibition of anaerobic phosphate release by nitric oxide in activated sludge. Applied and Environmental Microbiology 64 (8), 2925e2930. Vanverseveld, H.W., Meijer, E.M., Stouthamer, A.H., 1977. Energy conservation during nitrate respiration in Paracoccus denitrificans. Archives of Microbiology 112 (1), 17e23. Vargas, M., Guisasola, A., Artigues, A., Casas, C., Baeza, J.A., 2011. Comparison of a nitrite-based anaerobic-anoxic EBPR system with propionate or acetate as electron donors. Process Biochemistry 46 (3), 714e720. von Schulthess, R., Kuhni, M., Gujer, W., 1995. Release of nitric and nitrous oxides from denitrifying activated sludge. Water Research 29 (1), 215e226. von Schulthess, R., Wild, D., Gujer, W., 1994. Nitric and nitrous oxides from denitrifying activated sludge at low oxygen concentration. Water Science and Technology 30 (6), 123e132.
Waki, M., Yasuda, T., Suzuki, K., Sakai, T., Suzuki, N., Suzuki, R., Matsuba, K., Yokoyama, H., Ogino, A., Tanaka, Y., Ueda, S., Takeuchi, M., Yamagishi, T., Suwa, Y., 2010. Rate determination and distribution of anammox activity in activated sludge treating swine wastewater. Bioresource Technology 101 (8), 2685e2690. Wett, B., 2006. Solved upscaling problems for implementing deammonification of rejection water. Water Science and Technology 53 (12), 121e128. Wilderer, P.A., Jones, W.L., Dau, U., 1987. Competition in denitrification systems affecting reduction rate and accumulation of nitrite. Water Research 21 (2), 239e245. Xu, H.J., Liu, Y., 2011. Control and cleaning of membrane biofouling by energy uncoupling and cellular communication. Environmental Science and Technology 45 (2), 595e601. Yang, W., Vollertsen, J., Hvitved-Jacobsen, T., 2003. Nitrite accumulation in the treatment of wastewaters with high ammonia concentration. Water Science and Technology 48 (3), 135e141. Ye, L., Pijuan, M., Yuan, Z., 2010. The effect of free nitrous acid on the anabolic and catabolic processes of glycogen accumulating organisms. Water Research 44 (9), 2901e2909. Yoshida, Y., Kim, Y., Saito, T., Tanaka, K., 2009. Development of the modified activated sludge model describing nitrite inhibition of aerobic phosphate uptake. Water Science and Technology 59 (4), 621e630. Yoshida, Y., Takahashi, K., Saito, T., Tanaka, K., 2006. The effect of nitrite on aerobic phosphate and denitrifying activity of phosphate-accumulating organisms. Water Science and Technology 53 (6), 21e27. Zart, D., Bock, E., 1998. High rate of aerobic nitrification and denitrification by Nitrosomonas eutropha grown in a fermentor with complete biomass retention in the presence of gaseous NO2 or NO. Archives of Microbiology 169 (4), 282e286. Zeeman, G., Wiegant, W.M., Koster-Treffers, M.E., Lettinga, G., 1985. Influence of total ammonia concentration on the thermophilic digestion of cow manure. Agricultural Wastes 14 (1), 19e35. Zeng, W., Li, L., Yang, Y.Y., Wang, X.D., Peng, Y.Z., 2011. Denitrifying phosphorus removal and impact of nitrite accumulation on phosphorus removal in a continuous anaerobic-anoxic-aerobic (A2O) process treating domestic wastewater. Enzyme and Microbial Technology 48 (2), 134e142. Zhang, L., Yang, J., Furukawa, K., 2010. Stable and high-rate nitrogen removal from reject water by partial nitrification and subsequent anammox. Journal of Bioscience and Bioengineering 110 (4), 441e448. Zhou, Y., Ganda, L., Lim, M., Yuan, Z., Kjelleberg, S., Ng, W.J., 2010. Free nitrous acid (FNA) inhibition on denitrifying polyphosphate accumulating organisms (DPAOs). Applied Microbiology and Biotechnology 88 (1), 359e369. Zhou, Y., Pijuan, M., Yuan, Z., 2007. Free nitrous acid inhibition on anoxic phosphorus uptake and denitrification by polyphosphate accumulating organisms. Biotechnology and Bioengineering 98 (4), 903e912. Zhou, Y., Pijuan, M., Yuan, Z., 2008a. Development of a 2-sludge, 3-stage system for nitrogen and phosphorous removal from nutrient-rich wastewater using granular sludge and biofilms. Water Research 42 (12), 3207e3217. Zhou, Y., Pijuan, M., Zeng, R.J., Yuan, Z., 2008b. Free nitrous acid inhibition on nitrous oxide reduction by a denitrifyingenhanced biological phosphorus removal sludge. Environmental Science and Technology 42 (22), 8260e8265.
