WATER RESEARCH A Journal of the International Water Association
Editor-in-Chief Mogens Henze Institute of Environment & Resources Technical University of Denmark Bygningstorvet DK-2800 KGS Lyngby Denmark Tel: 45 4525 1477 Fax: 45 4593 2850 E-mail:
[email protected]
Editors J. Block Université H. Poincaré, Nancy I France David Dixon University of Melbourne Australia Hiroaki Furumai The University of Tokyo Japan Gregory Korshin University of Washington USA Anna Ledin Technical University of Denmark Denmark Eberhard Morgenroth University of Illinois Urbana-Champaign USA W. Rauch University Innsbruck Austria Maria Reis Universidade Nova de Lisboa/FCT Portugal Hang-Shik Shin Korea Advanced Institute of Science and Technology Korea Mark van Loosdrecht Delft University of Technology The Netherlands Thomas Ternes Bundesanstalt für Gewässerkunde Germany Stefan Wuertz Univ. of California, Davis USA Hanqing Yu University of Science & Technology of China China
Associate Editors Andrew Baker The University of Birmingham UK
Damien Batstone The University of Queensland Australia
S-L. Lo National Taiwan University Taiwan
G-H. Chen The Hong Kong University of Science & Technology Hong Kong China
Y. Matsui Hokkaido University Japan
Tom Curtis Univ. of Newcastle upon Tyne UK Ana Deletic Monash University USA Francis de los Reyes III North Carolina State University USA Rob Eldridge The University of Melbourne Australia Rosina Girones University of Barcelona Spain Stephen Gray Victoria University Australia Kate Grudpan Chiang Mai University Thailand Xiaodi Hao Beijing University of Civil Engineering and Architecture China E.E. Herricks University of Illinois - Urbana USA H-Y. Hu Tsinghua University China P.M. Huck University of Waterloo Canada Bruce Jefferson Cranfield University UK Sergey Kalyuzhnyi Moscow State University Russian Federation
Max Maurer EAWAG Switzerland How Yong Ng National University of Singapore Singapore Satoshi Okabe Hokkaido University Japan S.L. Ong National University of Singapore Singapore Jong M. Park Pohang University of Science & Technology Korea Miguel Salgot University of Barcelona Spain David Sedlak University of California, Berkeley USA Jean-Philippe Steyer LBE-INRA France M. Takahashi Hokkaido University Japan Kai Udert EAWAG Switzerland V.P. Venugopalan BARC Facilities India E. von Sperling Federal University of Minas Gerais Brazil A.I. Zouboulis Aristotle University Greece
Jaehong Kim Georgia Institute of Technology USA
Editorial Office
Jes La Cour Jansen Lund Institute of Technology Sweden
Water Research Radarweg 29 1043 NX Amsterdam The Netherlands Tel: 31 20 485 2701 Fax: 31 20 485 2521; 31 20 485 3264 E-mail:
[email protected]
G. Langergraber BOKU/Univ. of Natural Res. and Applied Life Scs. Austria
Mailing address
Publication information: Water Research (ISSN 0043-1354). For 2010, volume 44 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: Please visit this journal’s homepage at http://www.elsevier.com/locate/watres. You can track accepted articles at http://www.elsevier.com/ trackarticle and set up e-mail alerts to inform you of when an article’s status has changed. Also accessible from here is information on copyright, frequently asked questions and more. Contact details for questions arising after acceptance of an article, especially those relating to proofs, will be provided by the publisher. Orders, claims, and journal enquiries: Please contact the Elsevier Customer Service Department nearest you: St. Louis: Elsevier Customer Service Department, 11830 Westline Industrial Drive, St. Louis, MO 63146, USA; phone: (877) 8397126 [toll free within the USA]; (+1) (314) 4537076 [outside the USA]; fax: (+1) (314) 5235153; 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.
water research 44 (2010) 351
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Editorial
Editorial to special issue in Water Research Emerging contaminants in water The chemical pollution of natural waters is one of the big challenges of the 21st century. Based on the rapid evolution in analytical chemistry, with the possibility to detect more polar compounds a whole new suite of ‘‘emerging contaminants’’ such as pharmaceuticals, hormones and perfluorinated compounds has been identified in various compartments of the water cycle including both natural and technical aquatic systems. The discovery of these micropollutants in the aquatic environment has triggered research efforts to investigate sources and mitigation strategies with conventional and novel treatment processes and assess their importance with regard to ecotoxicology and human health. In this special issue on emerging contaminants, we have compiled 20 research articles and 2 reviews, which deal with these timely issues. As mentioned above, quantification of micropollutants in aquatic systems is a key requirement to assess their fate. Several studies in this issue address the determination of pharmaceuticals in archived biosolids (Halden et al.), in wastewater treatment (Lindberg et al.) and lagoon treatment (Wong et al.). Another study uses analytical data to evaluate the fraction of pharmaceutical residues in wastewater originating from hospitals (Ort et al.). Finally, a review covers the occurrence and fate of phytoestrogens in the environment (Liu et al.). To further elucidate the relevance of the micropollutants detected in various aquatic compartments, their (eco)toxicological potential has to be assessed. Several papers address related issues for individual compounds (lipid regulators, Fernandez-Pinas et al.) or classes of compounds (ionic liquids, Yun et al.) and for pharmaceuticals in advanced wastewater treatment systems such as powdered activated carbon and ozonation with in vivo and in vitro tests (Escher et al., Stalter et al.). Furthermore, one study is focused on toxicity nanotube suspensions (Tarabara et al.), which have been on the radar of emerging contaminants recently. Finally, one paper focuses on the toxicological relevance of emerging contaminants for drinking water (Schriks et al.). In recent years, municipal wastewater has been recognized as an important source of micropollutants to the
receiving water bodies. Therefore, mitigation strategies for the minimization of the discharge of these compounds play an increasingly important role in the urban water management. In this special issue, the removal of benzotriazoles (Reemtsma et al.) and pharmaceuticals, caffeine and DEET (Yu et al.) and other emerging contaminants (Rosal et al.) has been investigated for conventional wastewater treatment. Numerous papers address the oxidative removal of micropollutants from wastewater with chlorine, chlorine dioxide, ferrate, ozone, advanced oxidation processes, such as UV/ H2O2, (solar) photo-Fenton and non-thermal plasma (Lee et al., Malato et al., Reungoat et al., Gerrity et al., MendezArriaga). Other options for removal of micropollutants include membrane processes such as reverse osmosis (Hu et al.) and nanofiltration (Yangali-Quintanilla et al.) as well as sorption on sludge (Carrere et al.). The challenges caused by harmful algae producing toxins for desalination operations were reviewed by David Caron and co-workers. Finally, mitigation of micropollutants may also occur during managed aquifer recharge (Drewes et al.) or in biological Fenton-like processes (Vicent et al.). In short, we are very happy to provide you this Theme Issue on Emerging Contaminants. We appreciate the contributions by the authors, reviewers and editorial staff of Water Research to this project.
Thomas Ternes* Federal Institute of Hydrology (BFG), Am Mainzer Tor 1, 56068 Koblenz, Germany *Corresponding author. Tel.: þ49 261 1306 5560; fax: þ49 261 1306 5363. Urs von Gunten EAWAG, Ueberlandstrasse 133, Duebendorf CH-8600, Switzerland 0043-1354/$ – see front matter ª 2010 Published by Elsevier Ltd. doi:10.1016/j.watres.2010.01.015
water research 44 (2010) 352–372
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Review
Environmental fate and toxicity of ionic liquids: A review Thi Phuong Thuy Pham a, Chul-Woong Cho a, Yeoung-Sang Yun a,b,* a
Department of Bioprocess Engineering, Chonbuk National University, Jeonju, Chonbuk 561-756, Republic of Korea Division of Semiconductor and Chemical Engineering and Research Institute of Industrial Technology, Chonbuk National University, Chonbuk 561-756, Republic of Korea b
article info
abstract
Article history:
Ionic liquids (ILs) are organic salts with low melting point that are being considered as
Received 31 May 2009
green replacements for industrial volatile organic compounds. The reputation of these
Received in revised form
solvents as ‘‘environmental friendly’’ chemicals is based primarily on their negligible vapor
27 August 2009
pressure. Nonetheless, the solubility of ILs in water and a number of literature
Accepted 12 September 2009
documenting toxicity of ILs to aquatic organisms highlight a real cause for concern. The
Available online 24 September 2009
knowledge of ILs behavior in the terrestrial environment, which includes microbial degradation, sorption and desorption, is equally important since both soil and aquatic
Keywords:
milieu are possible recipients of IL contamination. This article reviews the achievements
Ionic liquids
and current status of environmental risk assessment of ILs, and hopefully provides
Toxicity
insights into this research frontier.
Degradation
ª 2009 Elsevier Ltd. All rights reserved.
Biodegradation Environmental fate Sorption
Contents 1. 2.
3.
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Toxicological aspect of ILs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1. Effects of ILs in an enzyme level . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2. Antibacterial activity of ILs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3. Toxicity of ILs to algae . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.4. Cytotoxicity of ILs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.5. Phytotoxicity of ILs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.6. Toxicity of ILs to invertebrates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.7. Inhibitory effects of ILs on vertebrates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Environmental fate of ILs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1. Chemical degradation of ILs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2. Biodegradability of ILs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3. Sorption of ILs in environmental systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
353 354 354 356 361 361 363 363 364 364 364 365 367
* Corresponding author. Division of Semiconductor and Chemical Engineering, Chonbuk National University, Jeonju, Chonbuk 561-756, Republic of Korea. Tel.: þ82 63 270 2308; fax: þ82 63 270 2306. E-mail address:
[email protected] (Y.-S. Yun). 0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2009.09.030
water research 44 (2010) 352–372
4.
1.
353
Concluding remarks and future directions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 368 Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 369 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 369
Introduction
Most of volatile organic compounds (VOCs) commonly used in industrial applications cause a major concern in the current chemical processing industry. The main problems are the toxicity of the organic solvents to both the process operators and the environment as well as the volatile and flammable nature of these solvents which make them a potential explosion hazard (Schmid et al., 1998). Recently, the deleterious effects of many solvents combined with serious environmental issues, such as atmospheric emissions and contamination of aqueous effluents are making their use prohibitive. Thus, many researchers have focused on the development of ‘‘green engineering’’ which represents research aimed at finding environmentally benign alternatives to harmful chemicals. Among the neoteric solvents applicable in ‘‘green technologies’’ ionic liquids (ILs) have garnered increasing attention over the others such as supercritical CO2 (Blanchard et al., 1999; Blanchard and Brennecke, 2001; Kazarian et al., 2000) and aqueous biphasic systems (Myasoedov et al., 1995; Rogers et al., 1995; Willauer et al., 1999). Ionic liquids, formerly known as molten salts, constitute one of the hottest areas in chemistry these days. Basically, they have melting points below 100 C, which can be achieved by incorporating a bulky asymmetric cation into the structure together with a weakly-coordinating anion (Ranke et al., 2004). The unique, highly solvating, yet non-coordinating environment of ILs provides an attractive medium for various types of chemical processes. Also, the physical properties of ILs can be tailored by a judicious variation in the length and branching of the alkyl chain and the anionic precursor (Fuler et al., 1997; Huddleston et al., 2001). In this way, ILs can be made taskspecific for a certain application. The almost limitless structural possibilities of ILs, as opposed to limited structural variations within molecular solvents, make them ‘‘designer
solvents’’ (Marsh et al., 2004; McFarlane et al., 2005; Sheldon, 2005). Some independent reports (Hagiwara and Ito, 2000; Olivier, 1999; Welton, 1999) and many reviews (Earle and Seddon, 2000; Rooney and Seddon, 2001) have highlighted ILs as representing a state-of-the-art, innovative approach to sustainable chemistry, with the argument that their vapor pressure is immeasurably low and they are not flammable. Recently, the application of these liquids as reaction media for organic synthesis, catalysis, or biocatalysis has been well documented (Earle and Seddon, 2000; Wasserscheid and Keim, 2000; Welton, 1999) (Fig. 1). Gordon (2001) pointed out that there is an obvious advantage in performing many reactions in ILs due to the improvement in process economics, reaction activity, selectivity and yield. Although ILs can lessen the risk of air pollution due to their insignificant vapor pressure, they do have significant solubility in water (Anthony et al., 2001; McFarlane et al., 2005; Wong et al., 2002). As a result, this is the most likely medium through which ILs will be released into the environment. Ionic liquids currently are not widely used in industrial applications; nonetheless, continued development and further use of these solvents may lead to accidental discharge and contamination. The properties that make them be the target of industrial interest (i.e. high chemical, thermal stability and non-volatility) suggest potential problems with degradation or persistence in the environment. In general, the deficiency of information and uncertainty surrounding the environmental impact of ILs is a major barrier to the utilization of these compounds by industry. Initial efforts have been made to overcome this drawback and offer a preliminary insight into the behavior of ILs in the aqueous environments. These studies provided extensive data sets, e.g. on (eco)toxicity, biodegradability, bioaccumulation and distribution of ILs in different environmental compartments. Therefore, it is necessary to consolidate all the available data in a single
Fig. 1 – Applications of ionic liquids.
354
water research 44 (2010) 352–372
review to lay the groundwork for more comprehensive community and ecosystem investigations. The overall objective of this review is to systematically gather and interpret existing information about the fate, removal options and (eco)toxicological assessment strategies of ILs.
2.
Toxicological aspect of ILs
The current literature represents a number of studies addressing the biological effects of ILs evaluated on the basis of toxicological test systems. The ILs toxicities towards these systems of different levels of biological complexity as well as several environmental compartments (Fig. 2) are successively discussed in the following subsections. All the structures of IL compounds discussed in this review were listed in Table 1. The acronyms used for these substances were adapted from Ranke et al. (2007a). In this way, the cation head groups were abbreviated as ‘‘IM’’ for imidazolium, ‘‘Py’’ for pyridinium, ‘‘Pyr’’ for pyrrolidinium, ‘‘Mor’’ for morpholinium, ‘‘Pip’’ for piperidinium, ‘‘Quin’’ for quinolinium, ‘‘N’’ for quaternary ammonium and ‘‘P’’ for quaternary phosphonium. The alkyl chains attached to the head group were given as numbers corresponding to the number of carbon in the alkyl residues. For example, the 1-butyl-3-methylimidazolium moiety was denoted as IM14. In case the carbon chain length equals or exceeds 10, the numbers were separated by a hyphen (e.g. IM1-10 indicated 1-decyl-3-methylimidazolium). Particularly, for pyridinium
entities, the carbon-bound alkyl chains were appended to the head group at different positions and the abbreviation was made by noting the position of attachment and a symbol for the attached group (e.g. Py4-2Me for 1-butyl-2-methylpyridinium). The anionic components were shortened as they are in the periodic table for the halides. For tetrafluoroborate, hexafluorophosphate, bis(trifluoromethylsulfonyl)imide, dicyanamide and hydrogen sulfate the abbreviations were BF4, PF6, (CF3SO2)2N, CN(N)2 and HSO4 in respective to their structural formula.
2.1.
Effects of ILs in an enzyme level
Enzyme inhibition data by ILs include those of the acetylcholinesterase from electric eel (Electrophorus electricus) (Arning et al., 2008; Jastorff et al., 2005; Matzke et al., 2007; Ranke et al., 2007b; Stasiewicz et al., 2008; Stock et al., 2004; Torrecilla et al., 2009; Zhang and Malhotra, 2005), the AMP deaminase (Sk1adanowski et al., 2005) and the antioxidant enzyme system of mouse liver (Yu et al., 2009a). The enzyme acetylcholinesterase plays the most important role in nerve response and function. Also, acetylcholinesterase catalyzes the hydrolysis of acetylcholinesters with a relative specificity for acetylcholine, which is a neurotransmitter common to many synapses throughout mammalian nervous systems (Fulton and Key, 2001; Massoulie´ et al., 1993). Thus, an inhibition of acetylcholinesterase leads to various adverse effects in neuronal processes, such as heart diseases or myasthenia
Fig. 2 – The flexible (eco)toxicological test battery considering aquatic and terrestrial compartments as well as different trophic levels including enzymes, luminescent marine bacteria, freshwater green algae, mammalian cells, duckweed, freshwater crustacean and zebrafish (Adapted from Matzke et al. (2007) by permission of the Royal Society of Chemistry).
355
water research 44 (2010) 352–372
Table 1 – Selection of cationic and anionic structures of commonly used ionic liquids. Head group
R2 N
N
+
Side chain R1 ¼ -C2H5, -C3H7, -C4H9, -C5H11, -C6H13, -C7H15, -C8H17, -C9H19, -C10H21, -C14H29, -C16H33, -C18H37, -C19H39 R2 ¼ -CH3, -C2H5
R1
Imidazolium (IM)
CH3 CH3
R1
+ N
R1 ¼ -C2H5, -C3H7, -C4H9, -C5H11, -C6H13, -C8H17
CH3
Pyridinium (Py)
CH3 N
+
R1 ¼ -C4H9, -C6H13, -C8H17
R1 Pyrrolidinium (Pyr)
O
N
CH3
+
R1 Morpholinium (Mor)
R1 ¼ -C4H9
Cation
R1
+
N
CH3
R1 ¼ -C4H9
Piperidinium (Pip)
+
N
R1 ¼ -C4H9, -C6H13, -C8H17
R1 Quinolinium (Quin)
R1
R2 +
N R4
R1-4 ¼ -CH3, -C2H5, -C3H7, -C4H9, -C6H13
R3
Quaternary ammonium (N)
R1
R2 +
P
R1-4 ¼ -C4H9, -C6H13, -C14H29
R3 (
R4
Quaternary phosphonium (P
(continued on next page)
356
water research 44 (2010) 352–372
Table 1 (continued) Head group Anion
Side chain Cl Br
Chloride Bromide
F Tetrafluoroborate [BF4]
B
F
-
F
F F
F
F
-
P
Hexafluorophosphate [PF6]
F
F F
F3 C
S
Bis(trifluoromethylsulfonyl)imide [(CF3SO2)2N]
O N
Dicyanamide [(CN)2N]
in humans (Chemnitius et al., 1999; Pope et al., 2005). Ranke et al. (2007b) published a comprehensive collection of acetylcholinesterase inhibition values for 292 compounds covering a large variety of ILs and closely related salts. Among these data, only those of the commonly tested ILs are summarized in Table 2 for the ease of comparison of ILs toxicity from molecular up to organism levels of biological complexity. It was found that all observed inhibitory effects on the enzyme could be exclusively accounted for the cationic moiety (Arning et al., 2008). In particular, the IL with pyridinium as cationic core structure inhibited the enzyme slightly stronger than the imidazolium analogue whereas the compounds based on phosphonium was less inhibitory. All anion species exerted no effect on the enzyme activity with only exception of the fluoride anion and the fluoride containing [SbF6] and [PF6] species. Both species are known to readily undergo hydrolysis in contact with moisture and thus the fluoride seems to be the active compound. The non-inhibiting effects of anion might be explained by their limited interactions with the active site of this enzyme (Matzke et al., 2007). In addition, a correlation between an increasing chain length of the side chains connected to the cationic head groups and an enhanced inhibitory potential of the ILs was found. It is believed that the mechanism involves the similarity of the positively charged imidazolium or pyridinium to the choline part that binds to the anionic site of the enzyme, such that the longer alkyl chain results in an improved fit (Stock et al., 2004). Sk1adanowski et al. (2005) discussed the usefulness of in vitro AMP deaminase inhibition assay as a potential molecular method in prospective risk analysis of imidazolium-based ILs. The results revealed that IM14 salts associated with [PF6], [BF4], p-tosylate and [Cl] demonstrated a dose-dependent inhibition of AMP deaminase activity. The IC50 values (concentration of ILs inhibiting 50% of enzyme activity) for those containing a fluorine compartment [PF6] and [BF4] are lower (5 mM) than those for [Cl] and p-tosylate (10 mM), which indicated the adverse effect of these fluoride-containing anions. The other study on enzyme inhibition assay dealt with
O
O -
CF3 S O
N -
N
N
the effects of acute exposure of intraperitoneal injection of aqueous IM18 Br on the antioxidant enzymes of the treated mouse liver (Yu et al., 2009a). The antioxidant enzymes tested included superoxide dismutase, catalase, glutathione peroxidase and glutathione-S-transferase. The results showed that administration of IM18 Br modified activities of these defense enzymes in mouse liver, and caused damage to livers of treated mice at median lethal dose (LD50) of 35.7 mg/kg. Though data published by these authors did not cover a large variety of ILs, the enzyme inhibition assays suggest the trend in which cationic moiety is the dominating factor influencing the toxicity of ILs, especially when substituted with a long alkyl side chain. Regarding the anion types, perfluoronated ions are of toxicological interest due to hydrolysis resulting in HF formation, while the others cause less prominent effect.
2.2.
Antibacterial activity of ILs
Bacteria serve as an ideal starting point for ILs toxicity estimations as they have short generation times. Preliminary toxicological investigations have shown quaternary ammonium and pyridinium compounds have critical inhibitory effects on a variety of bacteria and fungi (Babalola, 1998; Kelman et al., 2001; Li et al., 1998). In the studies of Pernak’s group (Cieniecka-Ros1onkiewicz et al., 2005; Pernak et al., 2001a; Pernak et al., 2001b; Pernak and Chwa1a, 2003; Pernak et al., 2003; Pernak et al., 2004a), they observed a trend of increasing toxicity with an increase in the alkyl chain length substituent in the pyridinium, imidazolium and quaternary ammonium salts to various bacteria including rods, cocci and fungi. As a measure of microbial activity of imidazolium and pyridinium ILs with varying alkyl chain lengths, Docherty and Kulpa (2005) also used a group of microorganisms possessing a variety of physiological and respiratory activities. It was found that imidazolium and pyridinium bromides incorporated hexyl- and octyl-chain had considerable antimicrobial effect to pure cultures of Escherichia coli,
Table 2 – Toxicity of ILs to different levels of biological complexity including enzyme, bacteria, algae, rat cell line, human cell lines, duckweed and invertebrate. Log10EC50 (mM)a
Compound Acetylcholin esterase IM12 Cl
2.0614
IM12 IM12 IM12 IM13 IM13 IM13 IM14
2.0514 2.0514 2.0314 2.2714 2.28 0.0318 2.2214 1.91 0.0411
BF4 PF6 (CF3SO2)2N Cl BF4 PF6 Cl
1.90 0.0218
IM14 BF4
1.98 0.01811
IM14 PF6
2.15 0.0518
IM14 (CF3SO2)2N IM14 (CN)2N
1.96 0.02111 1.95 0.0718
IM15 IM15 IM15 IM16
1.9614 1.8614 1.8514 1.9214
Cl BF4 PF6 Cl
IM16 Br
N.A.
IM16 IM16 IM16 IM17 IM17 IM17 IM18
1.8814 1.8814 2.1514 2.0714 2.1214 1.9114 1.6014
BF4 PF6 (CF3SO2)2N Cl BF4 PF6 Cl
IM18 Br
N.A.
4.5510 4.33 0.1119 N.A. N.A. N.A. N.A. 3.94 0.0613 N.A. 3.71 0.144 2.955 3.34 0.136 3.47 0.0419 4.01 0.054 3.07 0.0313 3.355 3.27 0.096 3.55 0.0413 3.10 0.176 3.12 0.3516 3.07 0.296 3.39 0.084 3.67 0.104 2.995 N.A. 3.14 0.0213 N.A. 1.9410 2.32 0.166 2.91 0.0913 1.42 0.124 0.815 3.18 0.0313 2.17 0.066 N.A. N.A. 2.44 0.0613 N.A. 1.19 0.116 1.01 0.0619 0.63 0.074 0.075
MCF7b
Lemna minor
N.A.
N.A.
N.A.
3.4414 3.9214 N.A. >4.3014 3.4714 >3.0014 3.5514
4.00 0.0420 N.A. 3.26 0.0420 N.A. N.A. N.A. N.A.
N.A. N.A. N.A. N.A. N.A. N.A. N.A.
N.A. N.A. N.A. N.A. N.A. N.A. 2.8211
N.A. N.A. N.A. N.A. N.A. N.A. 1.934 1.93 0.061
N.A.
3.4314
3.44 0.1120
N.A.
N.A.
1.574 1.56 0.071 1.85 0.0622
N.A.
2.1111
3.1214
3.72 0.0517 3.66 0.0820
N.A.
2.497
1.684 1.67 0.111
4.15 0.069
2.20 0.0421
N.A.
3.1014
4.14 0.2217
N.A.
N.A.
2.55 0.159 N.A.
1.80 0.0712 N.A.
1.81 0.1511 N.A.
2.6814 3.1514
3.07 0.0820 N.A.
N.A. N.A.
2.45 0.0811 N.A.
1.854 1.85 0.101 N.A. N.A.
N.A. N.A. N.A. N.A.
N.A. N.A. N.A. 1.92 0.0921
N.A. N.A. N.A. 0.0819
>3.0014 >3.0014 >3.0014 2.8514
N.A. N.A. N.A. N.A.
N.A. N.A. N.A. N.A.
N.A. N.A. N.A. N.A.
N.A. N.A. N.A. N.A.
N.A.
2.57 0.152
N.A.
N.A.
N.A.
N.A.
N.A.
N.A. 3.25 0.679 2.53 0.159 N.A. N.A. N.A. N.A.
N.A. N.A. N.A. N.A. N.A. N.A. 1.4621
N.A. N.A. N.A. N.A. N.A. N.A. 2.67 0.3719
2.9814 2.9114 2.2414 2.5314 2.5814 2.3014 2.0114
N.A. N.A. N.A. N.A. N.A. N.A. N.A.
N.A. N.A. 2.818 N.A. N.A. N.A. N.A.
N.A. N.A. N.A. N.A. N.A. N.A. N.A.
0.784 1.06 0.0422 N.A. N.A. N.A. N.A. N.A. N.A. N.A.
N.A.
1.65 0.252
N.A.
2.48 0.0420
N.A.
N.A.
Escherichia coli
Pseudo kirchneriella subcapitata
Scenedesmus vacuolatus
IPC-81
N.A.
N.A.
2.78 0.0619
N.A.
5.25 0.069 N.A. N.A. N.A. N.A. N.A. N.A.
N.A. N.A. N.A. N.A. N.A. N.A. 2.34 0.0121
N.A. N.A. N.A. N.A. N.A. N.A. 2.26 0.0811
N.A.
3.46 0.0623
4.60 0.029
N.A.
HeLa
Daphnia magnac N.A.
water research 44 (2010) 352–372
IM14 Br
Vibrio fischeri
1.334 0.54 0.1222
357
(continued on next page)
358
Table 2 (continued) Log10EC50 (mM)a
Compound Acetylcholin esterase 1.53 0.02511 2.0314 2.0314 N.A. 1.0914
IM1-10 BF4 IM1-10 PF6 IM1-14 Cl IM1-16 Cl IM1-18 Cl IM1-19 Cl IM1-19 BF4 IM1-19 PF6 IM22 Br IM23 Br IM24 BF4 IM25 BF4 IM26 Br IM26 BF4 IM2-10 Br Py Cl Py2 Cl Py3 Br Py3 (CF3SO2)2N Py4 Cl
1.10 0.0418 1.6814 0.5414 0.6814 0.9614 1.3614 1.4314 1.6214 2.0814 2.2114 2.03 0.0118 N.A. 1.7714 1.8414 0.9214 >3.0014 2.1014 2.2214 2.2114 1.7014
Py4 Br
1.7714
Py4 BF4 Py4 PF6 Py4 (CN)2N
1.8014 1.8414 N.A.
Py5 Br Py5 (CF3SO2)2N Py6 Cl Py6 Br Py6 PF6 Py6 (CF3SO2)2N Py8 Cl Py8 (CF3SO2)2N Py4-2Me Cl Py4-2Me BF4
1.5214 1.5514 1.7214 N.A. 1.7614 1.8514 1.6014 1.4014 0.7014 0.8214
Escherichia coli
Pseudo kirchneriella subcapitata
Scenedesmus vacuolatus
IPC-81
HeLa
MCF7b
Lemna minor
1.41 0.0713 0.95 0.126 N.A. 0.72 0.0413 0.50 0.0713 0.23 0.0619 0.18 0.0613 N.A. 0.15 0.0719 0.23 0.0819 1.45 0.0519 N.A. N.A. N.A. N.A. N.A. 2.8 0.0413 3.1413 N.A. 2.15 0.0513 N.A. N.A. N.A. N.A. N.A. 3.41 0.084 2.645 3.18 0.0619 3.40 0.014 2.735 N.A. N.A. 3.31 0.104 2.615 N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.
N.A. 2.64 0.159 N.A. N.A. N.A.
N.A. N.A. N.A. N.A. N.A.
2.3011 N.A. N.A. N.A. 3.57 0.0619
1.5914 1.9614 1.6414 N.A. 1.3414
2.48 0.0220 N.A. 2.28 0.0220 N.A. N.A.
2.848 N.A. N.A. N.A. N.A.
0.907 N.A. N.A. N.A. N.A.
N.A. N.A. N.A. N.A. N.A.
N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.
N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. 2.57 0.0621
N.A. N.A. 2.48 0.219 >2.0019 >2.0019 N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. 2.59 0.1119
0.7714 1.5014 0.4214 0.1914 0.0114 1.4014 1.6514 1.8514 >3.0014 >3.3014 3.2614 N.A. 2.0114 2.2614 0.5314 N.A. N.A. N.A. N.A. N.A.
N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. 4.36 0.0917 N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.
N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.
N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. 2.32 0.1819
N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.
N.A.
N.A.
N.A.
3.9014
3.50 0.0720
N.A.
N.A.
N.A.
N.A. N.A. N.A.
N.A. N.A. N.A.
N.A. N.A. N.A.
3.1814 N.A. N.A.
N.A. N.A. N.A.
N.A. N.A. N.A.
N.A. N.A. N.A.
N.A. N.A. N.A.
N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.
N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.
N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.
N.A. N.A. N.A. N.A. N.A. N.A. 1.2714 N.A. N.A. 3.2514
N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.
N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.
N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.
N.A. N.A. N.A. 1.074 N.A. N.A. N.A. N.A. N.A. N.A.
Daphnia magnac
water research 44 (2010) 352–372
IM18 BF4 IM18 PF6 IM18 (CF3SO2)2N IM19 BF4 IM1-10 Cl
Vibrio fischeri
1.1514 N.A.
Py4-3Me BF4 Py4-3Me PF6 Py4-3Me (CN)2N
1.53 0.0218 1.45 0.0218 1.2214
Py6-3Me Cl Py6-3Me Br
1.0614 N.A.
Py6-4Me Py6-4Me Py8-3Me Py8-3Me
Cl BF4 Cl Br
1.4414 1.4814 0.6414 N.A.
Py8-4Me Cl Py8-4Me BF4 Pyr14 Cl Pyr14 Br
N.A. 3.46 0.0623
N.A. N.A.
N.A. N.A.
N.A. N.A.
N.A. N.A.
N.A. N.A.
N.A. 1.764
N.A. N.A. N.A.
N.A. N.A. N.A.
N.A. N.A. N.A.
3.3014 N.A. N.A.
N.A. N.A. N.A.
N.A. N.A. N.A.
N.A. N.A. N.A.
N.A. N.A. N.A.
N.A. N.A.
N.A. N.A.
N.A. N.A.
N.A. N.A.
N.A. N.A.
N.A. N.A.
N.A. N.A.
N.A. 0.594
N.A. N.A. N.A. N.A.
N.A. N.A. N.A. N.A.
N.A. N.A. N.A. N.A.
N.A. 2.1714 N.A. N.A.
N.A. N.A. N.A. N.A.
N.A. N.A. N.A. 1.008
N.A. N.A. N.A. N.A.
N.A. N.A. N.A. 0.404
N.A. N.A. N.A. N.A.
N.A. N.A. N.A. 3.67 0.283
N.A. N.A. 3.37 0.1019 N.A.
1.6314 1.4914 >4.3014 3.7714
N.A. N.A. N.A. N.A.
N.A. N.A. 2.16 0.2519 N.A.
N.A. N.A. N.A. N.A.
Pyr14 BF4 Pyr14 (CF3SO2)2N Pyr14 (CN)2N Pyr16 Cl Pyr16 (CF3SO2)2N Pyr18 Cl Pyr18 BF4 Pyr66 Mor14 Cl Mor14 Br Mor14 (CF3SO2)2N Pip14 Br
1.9114 2.1314 1.9814 2.4814 2.6014 2.3614 2.0214 2.0814 N.A. 2.7114 2.7814 1.8314
N.A. N.A. N.A. 2.9910 N.A. N.A. N.A. N.A. >4.3019 N.A. N.A. 4.27 0.0919
N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.
N.A. >2.3812 N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.
N.A. 2.5319 N.A. N.A. N.A. N.A. N.A. N.A. N.A. >4.0019 2.0019 3.27 0.1219
2.9014 3.0114 4.2314 2.9114 N.A. 2.5914 1.8214 1.2314 N.A. >4.3014 3.4314 4.0314
N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.
N.A. 2.98 0.3219 N.A. N.A. N.A. N.A. N.A. N.A. N.A. 3.11 0.1319 3.15 0.1319 0.4719
N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.
Pip14 (CF3SO2)2N Quin4 Br Quin4 BF4 Quin6 BF4 Quin8 Br Quin8 BF4 N1111 Br N1114 (CF3SO2)2N N1123 (CF3SO2)2N N1124 Cl N1124 (CF3SO2)2N N2222 Cl N2222 Br N2226 Br N4444 Br P4444 Br P666-14 Br
1.7814 0.7914 0.6214 0.4814 N.A 0.3014 N.A. 2.6014 2.3414 2.0614 2.0314 2.8014 N.A. N.A. 2.3014 2.6114 2.8514
N.A. N.A. N.A. N.A. N.A. N.A. >5.004 N.A. N.A. N.A. N.A. N.A. >5.004 2.46 0.164 3.27 0.074 2.714 3.41 0.024
N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.
N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.
2.0819 N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. >4.0019 1.78 0.1719 N.A. N.A. N.A. N.A. N.A. N.A.
3.4114 2.3214 2.1614 1.0714 0.0314 0.1714 N.A. 3.6114 N.A. >4.3014 3.4313 >3.4814 N.A. N.A. 2.2514 1.6614 N.A.
N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. 4.26 0.0420 N.A. N.A. N.A. N.A.
N.A. N.A. N.A. 4.64 0.0215 4.258 N.A. 3.148 N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. 4.15 0.0415 4.158 2.938 N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A.
2.85 0.0719 N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. 0.83 0.6719 N.A. N.A. N.A. N.A. N.A. N.A. N.A.
N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. 1.474 0.954 N.A.
(continued on next page)
359
N.A. N.A.
1.1114 1.2214 1.9214 1.9314
N.A. 2.75 0.134 2.125 N.A. N.A. 2.66 0.054 1.995 1.4410 2.06 0.164 1.485 N.A. N.A. N.A. 0.79 0.054 0.255 N.A. N.A. >4.3019 N.A.
water research 44 (2010) 352–372
Py4-3Me Cl Py4-3Me Br
360
References: 1Bernot et al. (2005a); 2Cho et al. (2007); 3Cho et al. (2008a,b); 4Couling et al. (2006); 5Docherty and Kulpa (2005); 6Garcia et al. (2005); 7Jastorff et al. (2005); 8Kumar et al. (2009); 9Lee et al. (2005); 10 Luis et al. (2007); 11Matzke et al. (2007); 12Pretti et al. (2008); 13Ranke et al. (2004); 14Ranke et al. (2007b); 15Salminen et al. (2007); 16Samorı` et al. (2007); 17Stepnowski et al. (2004); 18Stock et al. (2004); 19 Stolte et al. (2007a); 20Wang et al. (2007); 21Wells and Coombe (2006); 22Yu et al. (2009). a N.A. means not available (not determined). b Toxicity of ILs is expressed as log10IC50 (mM) in case of MCF7 cell line. c Toxicity of ILs is expressed as log10LC50 (mM) in case of D. magna.
N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. 1.90 0.0520 N.A. 0.4814 N.A. 0.2414 N.A. N.A. N.A. N.A. N.A.
Scenedesmus vacuolatus Pseudo kirchneriella subcapitata
N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. N.A. 3.47 0.0818 >3.3017 >3.4814 3.40 0.218 BF4 PF6 (CF3SO2)2N (CN)2N P666-14 P666-14 P666-14 P666-14
Compound
Table 2 (continued)
Acetylcholin esterase
Vibrio fischeri
Escherichia coli
Log10EC50 (mM)a
IPC-81
HeLa
MCF7b
Lemna minor
Daphnia magnac
water research 44 (2010) 352–372
Staphylococcus aureus, Bacillus subtilis, Pseudomonas fluorescens and Saccharomyces cerevisiae. The anion performed nearly no effect on antimicrobial activity in the case of imidazolium analogues (Docherty and Kulpa, 2005; Garcia et al., 2005; Lee et al., 2005; Pernak et al., 2003; Pernak et al., 2004a) whereas this was not the case for phosphonium salts. Within the group of alkyltrihexylphosphonium ILs in the study of CienieckaRos1onkiewicz et al. (2005), both cation structure and the type of anion had effects on the biological activity. The antibacterial activity of ILs not only involves in hampering the growth rate of microbes but also interferes with their productivity. Matsumoto et al. (2004a) tested the toxicity of imidazolium-based ILs to lactic acid producing bacterium Lactobacillus rhamnosus to examine whether these compounds can replace conventional organic solvents in the extractive fermentation of lactate. The results showed that the bacterium L. rhamnosus grew, consumed glucose, and produced lactate in the presence of imidazolium-based ILs. A change of alkyl length in the imidazolium cation had little difference on the survival of the cells. In a similar study (Matsumoto et al., 2004b), they focused on hiochii bacteria, Lactobacillus homochiochii and Lactobacillus fructivorans and also found that the bacteria could produce lactic acid in the presence of ILs. Nonetheless, the lactic acid producing activities of these bacteria generally decreased with the extension of alkyl chain length in the imidazolium cation moiety. Water miscible ILs had various effects on the physiology of Clostridium sporogenes when tested as additives in culture media or reaction media for reduction of nitrobenzene (Dipeolu et al., 2008). In their study, 2-hydroxyethyltrimethylammonium dimethylphosphate and N,N-dimethylethanolammonium acetate increased the growth rate of C. sporogenes; by contrast, IM14 BF4 and IM12 EtSO4 inhibited growth. Although IM12 EtSO4 inhibited growth, it was sufficiently non-toxic to allow efficient reduction of nitrobenzene using harvested cells. Thus, it is recommended that both noninhibitory and partially inhibitory ILs should be screened for use in biotransformation. Nonetheless, Ganske and Bornscheuer (2006) referred that ILs could have substantial inhibitory effects on the growth of microorganisms when they explored the effects of the two most commonly used ILs IM14 BF4 and IM14 PF6 on the growth of E. coli, Pichia pastoris and Bacillus cereus. Regarding inhibition assays used in assessment of environmental potential risk of a compound in aquatic milieu, the bioluminescence assay using Vibrio fischeri (formerly known as Photobacterium phosphoreum) is one of the most applied (Kaiser and Palabrica, 1991; Steinberg et al., 1995). This is a rapid, costeffective, and well-established method for toxicity determination focusing on environmental issues, and also a standard ecotoxicological bioassay in Europe (DIN EN ISO 11348). The published data on ILs toxicity towards V. fischeri were listed on Table 2 and were comprehensively interpreted in the study of Peraccini et al. (2007). Although it has been claimed that modifications of the anion lead to changes in chemical and physical properties of ILs (Sheldon, 2001), no clear increase in toxicity caused by the anion could be observed, and toxicity seemed to be determined mainly by the cationic component (Ranke et al., 2004). This is likely explained by the fact that lipophilic part of the molecules can be intercalated into the
water research 44 (2010) 352–372
membrane, whereas their ionic head group is at least partially solvated in the aqueous solution, as suggested by Austin et al. (1998). The ILs toxicity was also observed to correlate directly with the length of the n-alkyl residues in the methylimidazolium cation (Romero et al., 2008). Interestingly, Ranke et al. (2004) noted a slight hormetic effect at concentrations below inhibitory concentrations. Concerning the anionic influence, compounds with [PF6] were found to be slightly more toxic than compounds with other anions in their study (Ranke et al., 2004). The anion [(CF3SO2)2N] showed no intrinsic toxicity to V. fischeri in the report of Matzke et al. (2007); in contrast, an increased in toxicity was found for all tested compounds combined with [(CF3SO2)2N] for V. fischeri (Stolte et al., 2007a). Couling et al. (2006) extended the bioluminescence inhibition assay to pyridinium derivatives and it was noted that the quaternary ammonium compounds seemed to be less toxic to V. fischeri than the pyridinium and imidazolium analogues. Also, the quantitative structureproperty relationship (QSPR) modeling suggested that imidazolium cations, with two nitrogen atoms, are predicted to be more toxic than pyridinium moieties, which only have one nitrogen atom in the structure. In addition, the QSPR correlation predicted that quaternary ammonium cations are less toxic than those with cations containing nitrogen-bearing rings, which was in agreement with the experimental results (Couling et al., 2006). However, in contrast to the cases of aromatic ILs and ammonium compounds, the authors were unsuccessful in modeling the behavior of phosphonium salts using the developed correlation.
2.3.
Toxicity of ILs to algae
As algae are primary producers, either directly or indirectly, of organic matter required by animals in freshwater food chains, their ecology is crucial in providing the energy for sustaining other higher trophic levels. The ubiquity of algae makes these organisms ideal for toxicological studies and, because they have a short life cycle they can respond quickly to environmental change (Blaise, 1993; Lewis, 1995). To date, several groups have focused their attention on the use of algal primary producers to assess the effects of ILs to aquatic environments (Cho et al., 2007; Cho et al., 2008a,b,c; Grabinska-Sota and Kalka, 2006; Kulacki and Lamberti, 2008; Lata1a et al., 2005; Matzke et al., 2007; Matzke et al., 2008; Pham et al., 2008a,b; Pretti et al., 2009; Stolte et al., 2007a; Wells and Coombe, 2006). Cho and co-workers used Pseudokirchneriella subcapitata (formerly known as Selenastrum capricornutum) to study the effect of different head groups, side chains and anions of ILs on algal growth rate and photosynthetic activity. The data revealed that the toxic influence of ILs on growth rates were more significant than those of photosynthetic performance (Pham et al., 2008b). Once again, the trend of increasing toxicity with increasing alkyl chain length was observed in their reports (Cho et al., 2007; Pham et al., 2008b). Regarding the anionic effects, P. subcapitata was sensitive to the anion moieties in the order: [SbF6] > [PF6] > [BF4] > [CF3SO3] > [C8H17OSO3] > [Br] z [Cl]. In particularly, it was found that with respect to IL incorporating perfluorinated anion (i.e. IM14 BF4), EC50 values (concentrations which lead to a 50% reduction of the exposed organisms
361
relative to control) of the previously prepared stock solution (6 months prior to experiment) were significantly lower compared to those of the freshly made one (Pham et al., 2008a). This might be due to hydrolytic effects of IM14 BF4 leading to fluoride formation, as confirmed by ion chromatography analysis. This implies that after ILs are released into the aqueous system; they can become more hazardous than expected by laboratory data with fresh ILs. In a detailed study on hydrolysis of fluoride-containing anions, Cho et al. (2008a) showed that IM14 SbF6 generated a greater amount of fluoride compared to IM14 BF4, but no fluoride formation occurred with the hexafluorophosphate. When only small amounts of fluoride ions were formed from IM14 SbF6 and IM14 BF4 within 96 h, the formed fluoride ion did not affect the algal growth rate. Nevertheless, the fluoride ion formation from IM14 BF4 increased with incubating time of the stock solution; thus, the toxicity might significantly increase according to the further formed fluoride ions. In view of cationic effect, Pyr14 Br was found to be the least toxic of all the ILs tested to P. subcapitata (Cho et al., 2008b). For the limnic green alga Scenedesmus vacuolatus, a severe toxicity was found for 1-butyl-4-(dimethylamino)pyridinium, whereas the quaternary ammonium and morpholinium compounds exhibited no toxicity (Stolte et al., 2007a). Despite the extensive studies on the toxicological impact of ILs towards freshwater phytoplankton, inhibition mechanism of both the growth rate and photosynthetic activity by ILs has not been described by the authors. Lata1a et al. (2005), who selected two marine algae Oocystis submarina (green algae) and Cyclotella meneghiniana (diatom) as testing organisms, found that the two species differed dramatically in their ability to recover from IL exposure. Additionally, it was discovered that IL toxicity declined with increasing salinity. The lower toxicity of IL in this case is probably due to the reduced permeability of IL cations through the algal cell walls. High amounts of chloride provide a good ion-pairing environment for imidazolium cations, which consequently compete with hydroxyl or silanol functional groups in the cell-wall structure of green alga and diatom, respectively. Though no information on EC50 values was described, the facts emerged from this work provide useful information in the further fate assessment of ILs in marine environments.
2.4.
Cytotoxicity of ILs
As a cellular test system, promyelotic leukemia rat cell line IPC-81 has been frequently used in cytotoxicity assays of ILs, with the reduction of the WST-1 dye as an indicator of cell viability (Matzke et al., 2007; Ranke et al., 2004; Ranke et al., 2007a; Stasiewicz et al., 2008; Stolte et al., 2006; Stolte et al., 2007b; Torrecilla et al., 2009). It was observed that ILs with polar ether, hydroxyl and nitrile functional groups within the side chains exhibited low cytotoxicity compared to those incorporated with ‘‘simple’’ alkyl side chains (Kumar et al., 2009; Stasiewicz et al., 2008; Stolte et al., 2007b). Those functional groups were thought to impede cellular uptake by membrane diffusion and reduce lipophilicity based interactions with the cell membrane (Stolte et al., 2007b). Taking a closer look at the effects of sub-structural elements of ILs, [(CF3SO2)2N] anion and 4-(dimethylamino)pyridinium cation
362
water research 44 (2010) 352–372
were described to have intrinsic effects of anion and head group on cytotoxicity, respectively (Stolte et al., 2007b). The well known side chain length effect (decrease in EC50 values with elongation of the alkyl side chain) could also be confirmed in these studies. To date many studies have analyzed the toxicity of ILs on human cell lines (Frade et al., 2007; Garcı´a-Lorenzo et al., 2008; Hassoun et al., 2002; Kumar et al., 2009; Salminen et al., 2007; Stepnowski et al., 2004; Wang et al., 2007). These in vitro systems have been extremely beneficial in studying the molecular basis of chemical’s biological activity, including its toxic mode of action (Blaauboer et al., 1998) and could facilitate extrapolation of in vitro data with regard to possible effects on humans (Malich et al., 1997). Most of studies dealt with HeLa cells exemplifying prototypical cells of the human epithelium which is normally the site of first contact of an organism with toxicants. According to Stepnowski et al. (2004), the cytotoxicity data implied that effects of IM14 cation coupled with chloride, tetrafluoroborate or hexafluorophosphate were probably dependent on the anionic moieties. The lowest effect concentrations for tetrafluoroborate species were found to be 0.63 mM, whereas hexafluorophosphate and chloride inhibited HeLa cell growth at comparably high concentrations of >10 mM. Surprisingly, when the anion effect was compared, the strongest inhibition was found for [PF6]. This might be due to hydrolysis affecting fluoride formation, thus causing serious toxicological consequences through the decomposition product. A similar phenomenon was observed by Ranke et al. (2004) in IPC-81 leukemia cells, where the lower toxicity of 1-n-butyl-3methylimidazolium hexafluorophosphate in comparison to the hexafluorophosphate anion alone was explained by reduced anion uptake due to the formation of an ion pair. The anion in this ion pair can, however, also be partially decomposed. This was shown in recent work by Swatloski et al. (2003), who identified traces of 1-n-butyl-3-methylimidazolium fluoride hydrate as a decomposition product formed during the purification of the 1-n-butyl-3-methylimidazolium hexafluorophosphate. As shown by Wang et al. (2007) the phosphonium bis(trifluoromethylsulfonyl)imide salts performed the highest inhibitory to HeLa cells, followed by alkylimidazolium, alkylpyridinium, alkyltriethylammonium and N-alkyl-N,Ndimethyl-N-(2-hydroxylethyl)ammonium salts, in decreasing order. For each cation class the toxicity increased with increasing chain length of the alkyl substituent for a given anion: 1-ethyl-3-methylimidazolium bromide yielded an EC50 of 8.4 mM, substituting the ethyl moiety for a butyl group led to an EC50 of 2.8 mM, and for an octyl moiety an EC50 of 0.3 mM. This result was consistent with what has been observed in other studies. Salts containing the tetrafluoroborate anion showed the highest EC50, followed closely by bromide and chloride. Bis(trifluoromethylsulfonyl)imide salts were significantly more toxic than their halide counterparts. However, the effect of changing the anion was smaller than that of changing the alkyl substituent, e.g. while 1-ethyl-3-methylimidazolium tetrafluoroborate was observed to have an EC50 of 9.9 mM, the corresponding bromide and bis(trifluoromethylsulfonyl)imide salts had EC50 of 8.4 and 1.8 mM, respectively – these all considerably less toxic than 1-octyl-3-methylimidazolium bromide.
The CaCo-2 cells were used in the study of Garcı´a-Lorenzo et al. (2008) with the aim of a convenient screening method for obtaining first rough estimates for the toxic potential of ILs. The obtained data showed that in general, ILs with longer alkyl chains were more lipophilic than those with shorter alkyl chains. The former can be presumed to have a tendency to be incorporated into the phospholipid bilayers of biological membranes. In this respect, some authors have indicated that the increased toxicity of longer ILs can be accounted for enhanced membrane permeability altering the physical properties of the lipid bilayer (Lata1a et al., 2005; Ranke et al., 2004; Stepnowski et al., 2004). Additionally, it has been proposed that the mode of toxic action for ILs takes place through membrane disruption because of the structural similarity of imidazolium-based ILs to detergent, pesticides and antibiotics able to cause membrane-bound protein disturbance (Docherty and Kulpa, 2005). Recently, Ranke et al. (2007a,b) have demonstrated that lipophilicity of ILs dominates their in vitro cytotoxicity over a wide range of structural variations. The contribution of the anionic part of the ILs to the observed biological effect was evaluated by comparing the EC50 values obtained for the cations IM16 and IM18, combined with two different anions [Cl] and [PF6]. For both cations, a stronger toxic effect was found for chloride derivatives, but not for fluoride containing hexafluorophosphate. A similar result was reported by Stock et al. (2004) where the inhibitory effects of IM14 Cl and IM14 PF6 on the acetylcholinesterase activity were compared. In addition, slightly higher cytotoxicity for the chloride derivative has also been observed when the cytotoxicity of IM14 Cl and IM14 PF6 on HeLa cells was tested (Stepnowski et al., 2004). This implies the effect of perfluorinated ions is not drastic to all but vary according to species of organisms tested. Several authors have pointed out that altering the anion has only minimal effects on the toxicity of several imidazolium compounds (Bernot et al., 2005a; Garcia et al., 2005; Ranke et al., 2004). This indicates that ILs toxicity seems to be related to the alkyl chain branching and to the hydrophobicity of the imidazolium cation but not to the various anions. In this respect, a recent study using the IPC-81 rat leukemia cell line with a large pool of anions demonstrated that most of the commercially available anions showed no or only marginal cytotoxic effects. However, anionic compartments with lipophilic and hydrolysable structural elements are likely to be of considerable relevance with respect to the toxicity of ILs (Stolte et al., 2006). In a recent study (Frade et al., 2007), the human cell lines such as HT-29 and CaCo-2 cells were utilized to estimate the inhibitory effect of ILs with several types of cations and anions. In both cells, IM14, IM12OH (1-(2-hydroxyethyl)-3methylimidazolium), IM12O2O1 (1-(2-(2-methoxyethoxy)ethyl)-3-methylimidazolium) and cholines were the least toxic cations independently of the anion. Within the studied combinations, it can be noted that IM14 PF6, IM14 acesulfame, IM12OH BF4/PF6, IM12O2O1 BF4/PF6, IM12OH acesulfame and IM12OH saccharine are not toxic and present good alternatives to organic solvents. Meanwhile, increasing the length of the substituent chain may contribute to a significant increasing of imidazolium toxicity. It was also noted that [(CF3SO2)2N] anion decreased the toxicity to a large extent,
water research 44 (2010) 352–372
independently of the cation and for both cell types, which was in accordance with Salminen et al. (2007).
2.5.
Phytotoxicity of ILs
The studies on phytotoxic activity of ILs were conducted mostly on the duckweed, Lemna minor, a common aquatic vascular plant (Jastorff et al., 2005; Larson et al., 2008; Matzke et al., 2007; Stolte et al., 2007a). In general, 1-alkyl-3methylimidazolium compounds with longer alkyl chains were more toxic to L. minor than those with short alkyl chain lengths. Imidazolium and pyridinium cations with butyl groups had similar EC50s (the concentrations that produced a 50% reduction in root growth) (39.07 and 32.54 mM, respectively); while the equivalent ammonium cation had a much higher EC50 (101.48 mM; i.e., less toxic) (Larson et al., 2008). In consideration of anionic effect, [(CF3SO2)2N] was found to cause moderate toxicity to this duckweed (EC50 ¼ 6300 mM) (Matzke et al., 2007). On the other hand, this anion had no or even a positive influence on the observed effects on L. minor (Stolte et al., 2007a). Focusing on the terrestrial environment, Matzke et al. (2009a) investigated the influence of differently composed soils, with varying contents of the clay minerals smectite and kaolinite, on the toxicity of different anion species of imidazolium-based ILs towards the wheat Triticum aestivum. The data showed that IM14 (CF3SO2)2N appeared the most toxic, independently of the type and concentration of added clay. This is totally in contrast to the findings of Stolte et al. (2007a), who reported that [(CF3SO2)2N] caused no harm to L. minor, indicating the toxic effect of this anion is different between certain plants. The toxicity of 1-butyl-3-methylimidazolium incorporated chloride, tetrafluoroborate and hydrogen sulfate was mainly controlled by the cationic moiety. The observed effects varied according to the added clay type and clay concentration. An increase of clay content resulted in less inhibitory effects of these substances. On the contrary, for IM14 combined with bis(trifluoromethylsulfonyl)imide the addition of clay minerals led to higher toxicity compared to the reference soil. Since results are contradictious further study is necessary to unravel the underlying mechanism. Moreover, a detailed study on the effect of IM14 BF4 on the wheat T. aestivum seedlings (Wang et al., 2009) showed that IM14 BF4 was hazardous to the early development of wheat and had varying effects on different organs. At low concentrations, IM14 BF4 did not inhibit, and even promoted, wheat seedling growth. Nonetheless, at high concentrations, this IL inhibited wheat seedling growth significantly and decreased chlorophyll content, thereby reducing photosynthesis and plant growth. Therefore, the authors suggested that dilution could decrease the toxicity of IM14 BF4 to plants and would be a good method for remediating IL-polluted environments. In another research, the phytotoxicity tests of chiral ILs containing (-)-nopyl derivatives were carried out in a plant house using spring barley (Hordeum vulgare) which is a monocotyledonous plant, and a common radish (Raphanus sativus L. subvar. radicula Pers.) which is a dicotyledonous plant (Ba1czewski et al., 2007). According to the data obtained, increasing the concentration of ILs resulted in a systematic decrease in the crop fresh weight of total sprouts and the crop fresh
363
weight per plant, both for spring barley and for common radish. It could also be noted that common barley was a more resistant plant which fairly well tolerates test IL concentrations up to 200 mg kg1 of soil; whereas, for radish, the growth and development inhibiting concentration is 100 mg kg1 of soil. Using the same target plant (H. vulgare), Pernak et al. (2004b) reported that the 1,3-dialkoxymethylimidazolium tetrafluoroborate salts introduced to the soil at concentration of 1,000 mg kg1, or 100 mg kg1 dry mass of soil, were found to exert a phytotoxic effect on monocotyledonous plants. On the other hand, at a concentration of 10 mg kg1 no such effect on the growth of the roots was notified. Concerning phytotoxicity of ILs to garden cress (Lepidium sativum L.) in soil environment, Studzin´ska and Buszewski (2009) have proved that hazardous effects of imidazolium ILs are closely connected with organic matter content in soil. Soil with more organic carbon was observed to sorb IL cations more extensively than soil with little or no organic matter; hence, the more fertile in soil, the lower probability of hazardous effect of ILs to plants. On the other hand, the hazardous character of analyzed ILs was strongly connected with their hydrophobicity, indicating that the more hydrophobic IL, the higher decrease of seed germination. Although intensive work has not been conducted on phytotoxic influence of ILs, the available data offer initial hints for environmental scientists dealing with the potential impact of ILs towards aqueous and terrestrial plants.
2.6.
Toxicity of ILs to invertebrates
Ecotoxicological literature of ILs to invertebrates mainly focus on the use of Daphnia magna as a test organism (Bernot et al., 2005a; Couling et al., 2006; Garcia et al., 2005; Grabinska-Sota and Kalka, 2006; Luo et al., 2008; Nockemann et al., 2007; Pretti et al., 2009; Samorı` et al., 2007; Wells and Coombe, 2006; Yu et al., 2009b). Daphnia is an important link between microbial and higher trophic levels (McQueen et al., 1986), and has been the subject of hundreds of intensive ecological studies. The results of all studies again observed the well-established link between toxicity and alkyl chain length of the tested ILs containing imidazolium, pyridinium or quaternary ammonium as counter cations. The most toxic compound towards D. magna was found to be IM18 Br whereas the least toxic one was IM14 Cl with log10EC50 values of 1.33 and 1.93, respectively (Table 2). Also, the nature of the anion was suggested to have smaller effects compared to those of the cation. In a recent study, Luo et al. (2008) investigated the developmental toxicity of IM18 Br on D. magna. It was found that this compound exhibited toxicity on the development of three generation of D. magna with the decrease of number of offspring and average brood size correlated to increasing IM18 Br concentrations. This indicated that IM18 Br could cause deleterious effect to the population of Daphnia and indirectly disturb freshwater food webs. Couling et al. (2006) used experimental data to determine which part of the IL molecule is responsible for the observed toxic effects through a quantitative structure-property relationship (QSPR) modeling. In this respect, correlative and predictive equations were generated and proved that there was a distinct influence of the length of alkyl residues attached to the aromatic nitrogen atoms
364
water research 44 (2010) 352–372
towards D. magna. Moreover, the models predicted that the toxicity increased slightly with increasing number of aromatic nitrogen atoms in cation ring. This implies that ammonium salts are less toxic than pyridinium salts, which in turn are less toxic than imidazolium moieties. Interestingly, it was noted that methylating the aromatic carbons could be effective in reducing toxicity to D. magna, indicating that 1-nbutylpyridinium bromide can be more toxic than 1-n-butyl-3methylpyridinium bromide, which is more toxic than 1-nbutyl-3,5-dimethylpyridinium analogue. The QSPR, though is still in its infancy, has contributed initial guidelines for a rationale design of a new category of ILs with an acceptable environmental profile. Other studies include data on the snail Physa acuta (Bernot et al., 2005b), the spring tail Folsomia candida (a soil invertebrate) (Matzke et al., 2007), Caenorhabditis elegans (a soil roundworm) (Swatloski et al., 2004) and Dreissena polymorpha (zebra mussel) (Costello et al., 2009). It was also demonstrated a positive relationship between alkyl chain length and toxicity in these reports. In the research of Bernot et al. (2005b), the estimated LC50 (median lethal concentration) ranged from 3.50 to 1799.8 mM (0.54 to 3.26 in the logarithmic form), which implied that P. acuta are less sensitive to ILs than are D. magna (log10LC50 (mM) ranging from 1.33 to 1.93 (Table 2)). Also, the authors observed that at low concentrations, the IL may suppress snail movement, but concentrations above this threshold level trigger an escape response, causing the organism to move faster. Grazing patterns, nonetheless, showed that snails grazed less at higher IL concentrations. Physa spp. are key components of freshwater food webs, because they graze algae and are themselves important prey for fish and invertebrate predators (Bernot and Turner, 2001; Osenberg and Mittelbach, 1989). Thus, nonlethal IL concentrations affected P. acuta behaviors, potentially influencing individual fitness and good web interactions.
2.7.
being 42.4, 43.4 and 85.1 mg/L, respectively, indicating that the developmental toxicity of IM18 Br in the frog was stagesensitive. The number of dead embryos was also found to increase with the increasing concentrations of the IL IM18 Br. The developmental impact of IM18 Br was claimed not only in this finding but also in the work of Luo et al. (2008), who investigated on D. magna. Other work in the literature has focused on the acute toxicity of ILs on rats and mice (Bailey et al., 2008; Cheng et al., 2009; Landry et al., 2005; Pernak and Czepukowicz, 2001; Sipes et al., 2008). The values of acute toxicity of 3-hexyloxymethyl1-methylimidazolium tetrafluoroborate were found to be LC50 ¼ 1400 and 1370 mg kg1 for female and male Wistar rats, respectively (Pernak and Czepukowicz, 2001). Bailey et al. (2008) studied the effects of prenatal exposure of mice to IM14 Cl due to the potential for human exposure as a result of water or soil contamination from industrial effluent or accidental spills. As shown in the experimental data, after being contacted to the IL, fetal weight was considerably reduced at the two highest concentrations (169 and 225 mg kg1 d1). Malformations were also somewhat more numerous at the highest dosage, suggesting that IM14 Cl may be teratogenic. Maternal toxicity was also present, indicating that IM14 Cl appeared to be developmentally toxic at maternally toxic dosages. Also, IM14 Cl has been shown to cause thermal irritation when applied topically to rats, but produced only minimal contact sensitization when evaluated in the mouse local lymph node assay (Landry et al., 2005). Additionally, in this report, it is worth noting that the transdermal toxicity of IM14 Cl was influenced by the vehicle of administration. Use of the organic solvent, dimethylformamide, accentuated the acute toxicity. Very high concentrations of IM14 Cl (up to 95% IM14 Cl in water) applied to the rat skin were markedly less acutely toxic. This result may have a practical guideline that to reduce the acute toxicity, ILs can be handled in pure form with water as a co-solvent.
Inhibitory effects of ILs on vertebrates
Zebrafish (Danio rerio) plays an important role in ecotoxicology as a prominent model vertebrate. Concerning toxicity of ILs to the zebrafish, Pretti et al. (2006) revealed that ILs may cause a completely different effect on fish according to their chemical structures. As imidazolium, pyridinium and pyrrolidinium showed a LC50 (lethal effect) >100 mg L1, they could be regarded as non-highly lethal towards zebrafish. On the other hand, the ammonium salts showed LC50 remarkably lower than that reported for organic solvents and tertiary amines. In general, these data referred that fish are less sensitive to ILs toxicity compared to other species belonging to lower trophic levels. In a recent report, Li et al. (2009) used the frog Rana nigromaculata as an amphibian model for toxicity testing. Amphibians are often the main vertebrate group prone to contaminant exposure in aquatic systems mostly because their larvae live in water (Lahr, 1997; Mann and Bidwell, 2000). In their study, they evaluated the toxic effects of IM18 Br on the early embryonic development of the frog R. nigromaculata. The results demonstrated that the highest embryonic mortality occurred in the neural plate stage, followed by the early gastrula and early cleavage stages with the LC50 values
3.
Environmental fate of ILs
3.1.
Chemical degradation of ILs
Ionic liquids possess excellent chemical and thermal stability, which gives, unfortunately, a negative aspect for their treatment after usage prior to disposal. To assess the persistence of ILs in the environment as well as verify possibilities of their cleanup by chemical methods, several groups have focused their attention on oxidative and thermal degradation of ILs in aqueous media (Awad et al., 2004; Baranyai et al., 2004; Berthon et al., 2006; Itakura et al., 2008; Li et al., 2007; Morawski et al., 2005; Siedlecka and Stepnowski, 2009; Siedlecka et al., 2008a,b; Stepnowski and Zaleska, 2005). Pioneering work in the field of oxidative degradation was done by Stepnowski and Zaleska (2005) and Morawski et al. (2005) who showed that the greatest degradation efficiency for imidazolium ILs was achieved with a combination of UV light and a catalytic oxidant such as hydrogen peroxide or titanium dioxide. Subsequently, Li et al. (2007) studied the oxidative degradation of 1,3-dialkylimidazolium ILs in hydrogen peroxide/acetic acid medium assisted by ultrasonic chemical irradiation. It was
water research 44 (2010) 352–372
observed that 99% of tested compounds was degraded after 72 h. In addition, advanced oxidative degradation in the presence of reactive peroxides generated by Fenton reagent has been applied for the removal of ILs from water (Siedlecka and Stepnowski, 2009; Siedlecka et al., 2008a,b). According to the results, in a Fenton system with 1 mM of Fe(III) and 100 mM of H2O2, more than 97% of IM14 Cl was observed to degrade after 90 min. For Pyr14 Cl, IM16 Cl and IM18 Cl, the levels of degradation were 92%, 88% and 68%, respectively. Investigations of the degradation mechanisms indicated IM18 Cl was more resistant to oxidation by OH radicals cleaved from H2O2, suggesting that the oxidation rates of imidazolium ILs by OH are structure-dependent (Siedlecka and Stepnowski, 2009). The level of degradation was dependent on the alkyl chain length, consistent with Stepnowski and Zaleska (2005), who indicated that lengthening the alkyl chain lowered the rate of IL degradation. On contrast, the different length of the side chains and the type of anions did not affect the degradation process (Li et al., 2007). Regarding the thermal degradation studies of alkylimidazolium salts (Awad et al., 2004), extension of the alkyl chain enhanced the thermo-oxidative degradation of imidazolium salts. Interestingly, methyl substitution in the 2-position (i.e. between the two N atoms) was observed to decrease the oxidative decomposition of imidazolium ILs. The longer alkyl chain was also observed to induce an enhancement in photocatalytic decomposition of ILs (Morawski et al., 2005), which was not in the case of oxidative degradation (Stepnowski and Zaleska, 2005; Siedlecka and Stepnowski, 2009). Nonetheless, detailed account on the degradation of ILs by photocatalysis is required to verify this phenomenon.
3.2.
Biodegradability of ILs
In contrast to chemical degradation, which requires the assistance of a certain oxidant for catalysis, biodegradation is the microbial breakdown of chemical compounds. Biodegradation seems to be more environmentally friendly compared to chemical decomposition process. The initial attempt to examine the degradation potential of different IM14 cations combined with [Br], [BF4], [PF6], [N(CN)2], [(CF3SO2)2N] and octylsulfate as the counter ion was done using the Sturm and Closed-Bottle test protocols by the group of Scammells (Garcia et al., 2005; Gathergood and Scammells, 2002; Gathergood et al., 2004; Gathergood et al., 2006). Nonetheless, no compound showed significant degree of biodegradation with the exception of the octylsulfate-containing IL. The next step study on the biodegradation of ILs involved the design of ILs containing biodegradable side chains (Gathergood and Scammells, 2002). The design was done according to the principles of Boethling (Boethling, 1994, 1996; Howard et al., 1991) who identified three important parameters including the potential sites of enzymatic hydrolysis (for example, esters and amides) and oxygen in form of hydroxyl, aldehyde or carboxylic acid groups as well as unsubstituted linear alkyl chains (especially 4 carbons) and phenyl rings, which represent possible sites for attack by oxygenases. However, for a balance between chemical properties and biodegradability, not all of these factors were suitable for ILs. The
365
addition of oxygen containing functional groups such as alcohols, aldehyde and carboxylic acids was reported to restrict the ILs performance as reaction media whereas the incorporation of phenyl rings was known to increase the melting points of IL solvents (McGuinness and Cavell, 2000). Therefore, ester or amide group was selected to be coupled in alkyl side chain of ILs. The introduction of ester groups derived from a C2 acid and C4 or higher alcohol in the 3-N-substitutent was demonstrated to increase the biodegradation of imidazolium-based ILs (Gathergood et al., 2004). This can be explained by the fact that introduction of ester moiety probably provides a site susceptible to enzymatic attack (Gathergood et al., 2004; Gathergood et al., 2006) and hence, improves the biodegradation level. Though the addition of amide group is informed to improve the biodegradation of organic compounds (Boethling, 1994, 1996; Howard et al., 1991), no critical enhancement of biological degradation was noted when this group was appended into the imidazoliumbased ILs (Gathergood et al., 2004). However, no compound could be classified as ‘‘readily biodegradable’’ corresponding to Organization for Economic Cooperation and Development (OECD) standards (U.S. EPA, 1998), for which 60–70% or greater biodegradation by activated sludge microbial inoculate is required within a 10-day window in a 28-day period. Finally, the combination of the octylsulfate anion and imidazolium cation containing ester side chains resulted in readily biodegradable IL (Gathergood et al., 2006). Recently, Stolte et al. (2008) also paid their attention on investigation of functional groups incorporating alkyl chain ILs. Nonetheless, the introductions of terminal hydroxyl, carboxyl, ether and nitrile groups did not improve the biological degradation as expected. Kumar et al. (2006) investigated the fate of IM14 BF4 when in contact with soil-microorganisms, wastewater microorganisms, Pseudomonas putida and E. coli. Although IM14 BF4 was indicated to be recalcitrant in Sturm and Closed-Bottle test assays as mentioned above, it was observed in this study that P. putida was able to break down IM14 BF4 after 15 days of incubation. The breakdown products were monitored using GC-MS and identified to be 1-H-methylimidazole and 1-H-butylimidazole, which were in consistent with the theoretical metabolism scheme proposed by Jastorff et al. (2003). In case of bacteria from soil and wastewater, the metabolic intermediates appeared on the 12th day. It was also noted that different intermediate peaks were observed at different retention time with different microbes, indicating that the degradation mechanism of IM14 BF4 may vary in correspondence to certain microbes and metabolic pathways. In another study, the biodegradation pathway of IM18 moiety (Fig. 3) was proposed based on intermediate products via HPLC-MS analysis after 24-day period of incubation with activated sludge (Stolte et al., 2008). The metabolism of IM18 cation appeared to undergo oxidation reactions catalyzed probably by mono-oxygenases, e.g. the cytochrome P450 system on the terminal methyl group (u-oxidation). The alcohol formed was subsequently oxidized and converted into aldehydes, and then into carboxylic acids by dehydrogenases. The resulting carboxylic acids then might undergo b-oxidation and finally generated two carbon fragments that can enter the tricarboxylic acid cycle as acetyl Co-A (Fig. 3).
366
water research 44 (2010) 352–372
retention time in min.
m/z+
intensity
8.9
195
4*105
13.5 / 14.4
211
3*105 / 2*105
m/z+
N
N
+
N
+
N
OH
OH
12.2 / 12.7
209
1*106 / 2*106
N
+
N
N
+
N
211
O
10.0 – 12.5
225
2*104 – 6*104
N
H
+
N
O
O
19.5
183
1*105
OH
N
N
+
N
+
N
209
OH
16.2
197
3*10
5
N
+
N
OH
26.5
155
0.5*105
OH
N
N
+
N OH
24.2
169
4*106
O +
O
N
N
+
225
+
N
OH O
26.7
141
1*105
N
N
+
O OH
Fig. 3 – Biodegradation pathways of 1-octyl-3-methylimidazolium by activated sludge microbial community (Reproduced from Stolte et al. (2008) by permission of the Royal Society of Chemistry).
The proposed pathways provide basic information to both environmental scientists and chemical engineers; however, no studies have sought to examine the toxicity of metabolic products after degradation of ILs. This issue is of paramount importance since metabolism might not always end in less toxic products. Subsequently, Wells and Coombe (2006) extended the microbial degradation study with ammonium, imidazolium, phosphonium and pyridinium compounds by measuring the biological oxygen demand. The authors observed no biodegradability of cations incorporated short chains (C 4) within this test series, which was in agreement with Docherty et al. (2007) and Stolte et al. (2008). For longer alkyl chains (C12, C16 and C18) containing ILs, a strong inhibitory effect of these compounds on the inoculum used was found, indicating the active microbial consortium was significantly impacted by ILs toxicity. In recent studies (Docherty et al., 2007; GrabinskaSota and Kalka, 2004; Harjani et al., 2008; Stasiewicz et al., 2008), pyridinium-based ILs were reported to be fully catabolized by microbial community in activated sludge. This can be inferred from the fact that degradation pathways for pyridine –
the precursor of pyridinium-based compounds – under aerobic and anaerobic conditions were intensively investigated in the work of Kaiser et al. (1996). With respect to the common 1,3-dialkylpyridinium ILs, Pham et al. (2009) reported that after 21 days of incubation, microorganisms from activated sludge were able to break down Py4-3Me Br. Analyses of HPLC and MS/MS demonstrated that this biodegradation led to the formation of 1-hydroxybutyl-3-methylpyridinium, 1-(2-hydroxybutal)-3-methylpyridinium, 1-(2-hydroxyethyl)3-methylpyridinium and methylpyridine. Based on these intermediate products, biodegradation pathways were also suggested (Fig. 4), thereby providing the basic information which might be useful for assessing the factors related to the environmental fate and behavior of this commonly used pyridinium IL. Although this is the first report on biodegradation intermediates and pathway of pyridinium ILs, the authors have failed to systematically screen a single microorganism or a microbial consortium responsible for biodegradation of ILs. Therefore, it is needed to further investigate which type of microorganism is adaptable to ILs and which is responsible for degradation process. Also, the broken
367
water research 44 (2010) 352–372
II H3C
+
N
H3C
CH3
+
N
CH3 OH
I H3C
OH
+
N
H3C
H3C
H3 C
OH
+
N
+
N
+
H
+
OH
+
O
+
N
H3 C
OH
H3C
+
N
+
N
CH3 H3C
H
+
H3C OH
H3C O
H3C OH
Fig. 4 – Biodegradation pathways of 1-butyl-3-methylpyridinium entity by microorganisms from activated sludge (Reprinted with permission from Pham et al. (2009). Copyright 2009 American Chemical Society).
structures much further than methylpyridinium were not measured in the study. In particular, the possibility of cleavage of heterocyclic ring in ILs molecular structure (both pyridinium in this work and imidazolium in the study of Stolte et al. (2008)) has not been indentified. This issue should be clarified through further studies. Interestingly, Py4-3Me Br was not found to be metabolized by the activated sludge community (Docherty et al., 2007). This was attributed likely to the high IL concentration used in the study of Docherty’s group, which consequently inhibited the microbial consortium. Nonetheless, it was demonstrated that the structural manipulation of the pyridinium skeleton may lead to ILs with greater biodegradable extent compared to imidazolium-based compouds (Harjani et al., 2008). Concerning the anionic effect, ILs with halide counter ions were postulated to be more stable to degradation than perfluoronated ions (Awad et al., 2004; Gathergood and Scammells, 2002). In a preliminary study, Gathergood and Scammells (2002) confirmed this assumption and showed that the biodegradation efficiency decreased in the order [PF6] > [BF4] > [Br] with 60%, 59% and 48% of CO2 evolution values, respectively. In a later study (Gathergood et al., 2006), it was found that the octylsulfate anion is considerably more biodegradable than the other commonly used anions. The alkyl chain with C4, C6 or C8 was found to increase the rate of degradation (Docherty et al., 2007; Stolte et al., 2008); nonetheless, further increasing the chain length to C12, C16 or C18 was noted to cause toxic effect towards inoculum (Wells and Coombe, 2006). However, it was stated that the long octyl side chain was not a compulsory factor for biodegradation, but more important is a certain overall lipophilicity of the compound (Stolte et al., 2008).
3.3.
Sorption of ILs in environmental systems
Because the aquatic and terrestrial environments are possible recipients for contaminants, the distribution and behavior of ILs in soil are also extremely important. The retention and mobility of ILs in soils and sediments are strongly influenced by its tendency to be sorbed onto the various components of the soil matrix (Stepnowski, 2005). Since imidazolium-based ILs possess high electron acceptor potential of delocalized aromatic systems and hydrophobic components (e.g. the alkyl chain) (Stepnowski, 2005), they could be sorbed onto soils and sediments via several mechanisms. In a preliminary study, Stepnowski (2005) proved that electrostatic interactions contributed to the sorption of the imidazolium cations. Moreover, totally contrast trends were also observed demonstrating an extremely strong and practically irreversible sorption onto fine-textured marine sediments and a relatively weakly and reversibly binding to peaty soil (with the highest organic carbon content) (Stepnowski, 2005). This indicates the importance of the mineral component of the soil (sediment) in the sorption mechanism of ILs. Also, in this work, the author pointed out that compounds with longer alkyl chains were irreversibly bound to the soil component, which was in agreement with Stepnowski et al. (2007) and Matzke et al. (2009b). Interestingly, Beaulieu et al. (2008) did not find a positive effect of alkyl chain length on the sorption of alkylimidazolium-based ILs to aquatic sediments and suggested that hydrophobic interactions were not the most important sorption mechanism. The contrast results between these groups imply that sorption mechanisms of ILs may vary according to properties and composition of the environmental systems. The studies suggest that ILs may be retained by aquatic sediments;
368
water research 44 (2010) 352–372
nonetheless, the toxic action of these sorbed materials towards aquatic organisms has not been addressed. Also, further efforts should be continued to elucidate the reversibility of IL sorption on these sediments to give a better understanding of the ILs fate after being released into aqueous environment. Matzke et al. (2009b) investigated the influences of the two different clay minerals kaolinite and smectite as well as of organic matter on the cation sorption and desorption behaviors of three imidazolium based ILs including IM14 BF4, IM18 BF4 and IM14 (CF3SO2)2N in soil. The addition of organic matter and clays was observed to increase the sorption/decrease the desorption of all ILs tested, and in particular smectite had striking effects on the sorption efficiency of all substances. It is worth noting that not only the cationic moiety with different alkyl side chain lengths but the anionic compartments can also play an importance role in the sorption/ desorption processes. Imidazolium compounds with BF 4 as a counter anion showed higher sorption capacity compared to that of [(CF3SO2)2N], indicating the high potential of this type of IL to form ionic pairs in the soil matrix (Matzke et al., 2009b). However, further work with a variety of ILs incorporated different cationic and anionic moieties should be carried out to verify these phenomena. Gorman-Lewis and Fein (2004) examined the sorption behavior of IM14 Cl onto a range of surfaces which are commonly found in the near-surface environment. The results suggested that IM14 Cl could be minimally retarded by noninterlayer clay system and might lead to unimpeded transport through subsurface groundwater. Also, the adsorption capacity of this IL onto bacterial surfaces was not high, which might be due to the low hydrophobicity of IM14 Cl. Additionally, investigations of the adsorption of IM14 Cl towards different media carried out by our group (Vijayaraghavan et al., 2009) have shown that retardation of this compound was possible only by an ion-exchange resin and activated carbon, which was in consistency with the work of Anthony et al. (2001). However, no significant adsorption of IM14 Cl in the media of a fermentation waste (Corynebacterium glutamicum) and dried activated sludge was observed in our study. Conclusively, the data currently available demonstrated that ILs incorporated imidazolium cation can be sorbed to organic matter, whether found in aquatic sediments or terrestrial soils, and the presence of clays significantly enhanced the sorption capacity.
4.
Concluding remarks and future directions
Ionic liquids, of which the most often cited attribute is their negligible vapor pressure, have been suggested as a green alternative to traditional organic solvents with the desire to minimize diffusion to the atmosphere. Low volatility, however, does not completely eliminate potential environmental hazards and might pose serious threats to aquatic and terrestrial ecosystems. The studies of environmental fate and toxicity of ILs have shown that the ILs commonly used to date are toxic in nature and their toxicities vary considerably across organisms and trophic levels. In general, the effect of anionic moieties is not drastic as the alkyl length effect except for the case of [(CF3SO2)2N], which shows a clear (eco)toxicological
hazard potential. The other perfluorinated anions have been also proved to be hazardous due to hydrolytically unstable properties. In addition, the introduction of functional polar groups to the alkyl chain has been shown to reduce the toxicity of ILs and increase the biodegradation efficiency to some extent. This indicates the possibility of tailoring ILs by coupling suitable functional groups to their structure, which in turn leads to a more environmental friendly compound. The side chain length effect has been found to be consistent in all levels of biological complexity as well as different environmental compartments. Also, an increase in alkyl-chain length, or lipophilicity, was observed to be related to an increase in the rate of degradation as well as an increase in toxicity. This indicates a conflict of aims between minimizing the toxicity and maximizing the biodegradability of these neoteric solvents. Regarding the cationic compartment, pyridinium has been found to be more environmental friendly than imidazolium from both viewpoints of toxicology and microbial degradation. It can therefore be suggested that the structural manipulation of the pyridinium skeleton should be considered in design of a sustainable IL. From the currently available data, it is clear that some commonly used ILs are very far away from the image of green chemicals that are often cited in the literature. The uncertainties in their sustainable development hinder the applications of ILs under real conditions. Although some attempts have been made to give important hints in the prospective design and synthesis of inherently safer ILs, comprehensive studies dealing with the behaviors of ILs in aqueous media still await to be conducted. The important features required for the thorough insight into environmental fate of ILs include, but are not limited to: Providing more fundamental understanding into the mechanism for IL-induced toxicity to different levels of biological complexity. The underlying mechanisms of IL toxicity have rarely been studied. - Assessing the biodegradability of cationic and anionic compartments and toxicity of their degradation intermediates. This may provide useful information in consciously designing safer chemicals. - Investigating the aerobic and anaerobic biodegradation of ILs, which would suggest initial guidelines for the treatment of ILs waste by using the existing aerobic and anaerobic wastewater treatment facilities. Especially, anaerobic degradation awaits to be investigated. - Defining which organisms or enzymes may promote degradation pathways and determining specific microbial consortium or cultivatable communities capable of biotransformation of ILs. - Performing the ecotoxicity and biodegradation tests in real environmental conditions instead of controlled conditions of laboratory experiments, which would be advantageous in understanding the fate and behavior of ILs under real conditions. For this, the potential toxicological effects at population level and community level should be addressed. It must be encouraged to use tools such as experimental mesocosms to study the effects of ILs at higher levels of organization. - Creating database of environmentally benign structure moieties of ILs based upon their toxicological and -
water research 44 (2010) 352–372
biodegradation information, which would be practically useful as a reference for manufacturers and regulators to properly develop and regulate the use of ILs.
Acknowledgements This work was supported by NRF Grant funded by the Korean Government (KRF-2007-521-D00106, NRL 2009-0083194, and in part WCU R31-2008-000-20029-0).
references
Anthony, J.L., Maginn, E.J., Brennecke, J.F., 2001. Solution thermodynamics of imidazolium-based ionic liquids and water. J. Phys. Chem. B 105, 10942–10949. Arning, J., Stolte, S., Bo¨schen, A., Stock, F., Pitner, W.-R., WelzBiermann, U., Jastorff, B., Ranke, J., 2008. Qualitative and quantitative structure activity relationships for the inhibitory effects of cationic head groups, functionalized side chains and anions of ionic liquids on acetylcholinesterase. Green Chem. 10, 47–58. Austin, R.P., Barton, P., Davis, A.M., Manners, C.N., Stansfield, M.C., 1998. The effect of ionic strength on liposome-buffer and 1-octanol-buffer distribution coefficients. J. Pharm. Sci. 87, 599–607. Awad, W.H., Gilman, J.W., Nyden, M., Harris, R.H., Sutto, T.E., Callahan, J., Trulove, P.C., DeLong, H.C., Fox, D.M., 2004. Thermal degradation studies of alkyl-imidazolium salts and their application in nanocomposites. Thermochim. Acta 409, 3–11. Babalola, G.O., 1998. Anti-bacterial activity of synthetic N-heterocyclic oxidizing compounds. Lett. Appl. Microbiol. 26, 43–46. Bailey, M.M., Townsend, M.B., Jernigan, P.L., Sturdivant, J., HoughTroutman, W.L., Rasco, J.F., Swatloski, R.P., Rogers, R.D., Hood, R.D., 2008. Developmental toxicity assessment of the ionic liquid 1-butyl-3-methylimidazolium chloride in CD-1 mice. Green Chem. 10, 1213–1217. Ba1czewski, P., Bachowska, B., Bia1as, T., Biczak, R., Wieczorek, W. M., Balin´ska, A., 2007. Synthesis and phytotoxicity of new ionic liquids incorporating chiral cations and/or chiral anions. J. Agric. Food Chem. 55, 1881–1892. Baranyai, K.J., Deacon, G.B., MacFarlane, D.R., Pringle, J.M., Scott, J.L. , 2004. Thermal degradation of ionic liquids at elevated temperatures. Aust. J. Chem. 57, 145–147. Beaulieu, J.J., Tank, J.L., Kopacz, M., 2008. Sorption of imidazoliumbased ionic liquids to aquatic sediments. Chemosphere 70, 1320–1328. Bernot, R.J., Turner, A.M., 2001. Predator identity and traitmediated indirect effects in a littoral food web. Oecologia 129, 139–146. Bernot, R.J., Brueseke, M.A., Evans-White, M.A., Lamberti, G.A., 2005a. Acute and chronic toxicity of imidazolium-based ionic liquids on Daphnia magna. Environ. Toxicol. Chem. 24, 87–92. Bernot, R.J., Kennedy, E.E., Lamberti, G.A., 2005b. Effects of ionic liquids on the survival, movement, and feeding behavior of the freshwater snail, Physa acuta. Environ. Toxicol. Chem. 24, 1759–1765. Berthon, L., Nikitenko, S.I., Bisel, I., Berthon, C., Faucon, M., Saucerotte, B., Zorz, N., Moisy, Ph, 2006. Influence of gamma
369
irradiation on hydrophobic room-temperature ionic liquids [BuMeIm] PF6 and [BuMeIm] (CF3SO2)2N. Dalton Trans, 2526–2534. Blaauboer, B.J., Balls, M., Barratt, M., Casati, S., Coecke, S., Mohamed, M.K., Moore, J., Rall, D., Smith, K.R., Tennant, R., Schwetz, B.A., Stokes, W.S., Younes, M., 1998. Alternative testing methodologies and conceptual issues. Environ. Health Perspect 106, 413–418. Blaise, C.R., 1993. In: Richardson, M.L. (Ed.), Ecotoxicology Monitoring. VCH, Weinheim, pp. 83–108. Blanchard, L.A., Hancu, D., Beckman, E.J., Brennecke, J.F., 1999. Green processing using ionic liquids and CO2. Nature 399, 28–29. Blanchard, L.A., Brennecke, J.F., 2001. Recovery of organic products from ionic liquids using supercritical carbon dioxide. Ind. Eng. Chem. Res. 40, 287–292. Boethling, R.S., 1994. Cationic Surfactants, Surfactant Science Series, vol. 53. Marcel Dekker, New York, pp. 95–135. Boethling, R.S., 1996. Designing safer chemicals. ACS Symp. Ser. 640, 156. Chemnitius, J.-M., Sadowski, R., Winkel, H., Zech, R., 1999. Organophosphate inhibition of human heart muscle cholinesterase isoenzymes. Chem. Biol. Interact 120, 183–192. Cheng, Y., Wright, S.H., Hooth, M.J., Sipes, I.G., 2009. Characterization of the disposition and toxicokinetics of N-butylpyridinium chloride in male F-344 rats and female B6C3F1 mice and its transport by organic cation transporter 2. Drug Metab. Dispos. 37, 909–916. Cho, C.-W., Pham, T.P.T., Jeon, Y.-C., Vijayaraghavan, K., Choe, W.-S., Yun, Y.-S., 2007. Toxicity of imidazolium salt with anion bromide to a phytoplankton Selenastrum capricornutum: effect of alkylchain length. Chemosphere 69, 1003–1007. Cho, C.-W., Pham, T.P.T., Jeon, Y.-C., Yun, Y.-S., 2008a. Influence of anions on the toxic effects of ionic liquids to a phytoplankton Selenastrum capricornutum. Green Chem. 10, 67–72. Cho, C.-W., Jeon, Y.-C., Pham, T.P.T., Vijayaraghavan, K., Yun, Y.-S., 2008b. The ecotoxicity of ionic liquids and traditional organic solvents on microalga Selenastrum capricornutum. Ecotoxicol. Environ. Saf. 71, 166–171. Cho, C.-W., Pham, T.P.T., Jeon, Y.-C., Min, J., Jung, H.Y., Lee, D.S., Yun, Y.-S., 2008c. Microalgal photosynthetic activity measurement system for rapid toxicity assessment. Ecotoxicology 17, 455–463. Cieniecka-Ros1onkiewicz, A., Pernak, J., Kubis-Feder, J., Ramani, A., Robertson, A.J., Seddon, K.R., 2005. Synthesis, anti-microbial activities and anti-electrostatic properties of phosphoniumbased ionic liquids. Green Chem. 7, 855–862. Costello, D.M., Brown, L.M., Lamberti, G.A., 2009. Acute toxic effects of ionic liquids on zebra mussel (Dreissena polymorpha) surviving and feeding. Green Chem. 11, 548–553. Couling, D.J., Bernot, R.J., Docherty, K.M., Dixon, J.K., Maginn, E.J., 2006. Assessing the factors responsible for ionic liquid toxicity to aquatic organisms via quantitative structure-property relationship modeling. Green Chem. 8, 82–90. Dipeolu, O., Green, E., Stephens, G., 2008. Effects of watermiscible ionic liquids on cell growth and nitro reduction using Clostridium sporogenes. Green Chem. 11, 397–401. Docherty, K.M., Kulpa, C.F., 2005. Toxicity and antimicrobial activity of imidazolium and pyridinium ionic liquids. Green Chem. 7, 185–189. Docherty, K.M., Dixon, J.K., Kulpa, C.F., 2007. Biodegradability of imidazolium and pyridinium ionic liquids by an activated sludge microbial community. Biodegradation 18, 481–493. Earle, M.J., Seddon, K.R., 2000. Ionic liquids. Green solvents for the future. Pure Appl. Chem. 72, 1391–1398. Frade, R.F.M., Matias, A., Branco, L.C., Afonso, C.A.M., Duarte, C. M.M., 2007. Effect of ionic liquids on human colon carcinoma HT-29 and CaCo-2 cell lines. Green Chem. 9, 873–877. Fuler, J., Carlin, R.T., Osteryoung, R.A., 1997. The room temperature ionic liquids 1-ethyl-3-methylimidazolium tetrafluoroborate:
370
water research 44 (2010) 352–372
electrochemical couples and physical properties. J. Electrochem. Soc. 144, 3881–3886. Fulton, M.H., Key, P.B., 2001. Acetylcholinesterase inhibition in estuarine fish and invertebrates as an indicator of organophosphorus insecticide exposure and effects. Environ. Toxicol. Chem. 20, 37–45. Ganske, F., Bornscheuer, U.T., 2006. Growth of Escherichia coli, Pichia pastoris and Bacillus cereus in the presence of the ionic liquids [BMIM] [BF4] and [BMIM] [PF6] and organic solvents. Biotechnol. Lett. 28, 465–469. Garcia, M.T., Gathergood, N., Scammells, P.J., 2005. Biodegradable ionic liquids. part II: effect of the anion and toxicology. Green Chem. 7, 9–14. Garcı´a-Lorenzo, A., Tojo, E., Tojo, J., Teijeira, M., Rodrı´guezBerrocal, F.J., Gonza´lez, M.P., Martı´nez-Zorzano, V.S., 2008. Cytotoxicity of selected imidazolium-derived ionic liquids in the human Caco-2 cell line. Sub-structural toxicological interpretation through a QSAR study. Green Chem. 10, 508–516. Gathergood, N., Scammells, P.J., 2002. Design and preparation of room-temperature ionic liquids containing biodegradable side chains. Aust. J. Chem. 55, 557–560. Gathergood, N., Garcia, M.T., Scammells, P.J., 2004. Biodegradable ionic liquids: part I. Concept, preliminary targets and evaluation. Green Chem. 6, 166–175. Gathergood, N., Scammells, P.J., Garcia, M.T., 2006. Biodegradable ionic liquids. Part III: the first readily biodegradable ionic liquids. Green Chem. 8, 156–160. Gordon, C.M., 2001. New developments in catalysis using ionic liquids. Appl. Catal. A 222, 101–117. Gorman-Lewis, D.J., Fein, J.B., 2004. Experimental study of the adsorption of an ionic liquid onto bacterial and mineral surfaces. Environ. Sci. Technol 38, 2491–2495. Grabinska-Sota, E., Kalka, J., 2004. An assessment of the toxicity of pyridinium chlorides and their biodegradation intermediates. Environ. Int 28, 687–690. Grabinska-Sota, E., Kalka, J., 2006. Toxicity of imidazolium chlorides to aquatic organisms. Polish J. Environ. Stud. 15, 405–409. Hagiwara, R., Ito, Y., 2000. Room temperature ionic liquids of alkylimidazolium cations and fluoroanions. J. Fluorine Chem. 105, 221–227. Harjani, J.R., Singer, R.D., Garcia, M.T., Scammells, P.J., 2008. The design and synthesis of biodegradable pyridinium ionic liquids. Green Chem. 10, 436–438. Hassoun, E.A., Abraham, M., Kini, V., Al-Ghafri, M., Abushaban, A., 2002. Cytotoxicity of the ionic liquid, 1- N-butyl-3methylimidazolium chloride. Res. Commun, Pharmacol. Toxicol 7, 23–31. Howard, P.H., Boethling, R.S., Stiteler, W., Meylan, W., Beauman, J., 1991. Development of a predictive model for biodegradability based on BIODEG, the evaluated biodegradation data base. Sci. Total Environ. 110, 635–641. Huddleston, J.G., Visser, A.E., Reichert, W.M., Willauer, H.D., Broker, G.A., Rogers, R.D., 2001. Characterization and comparison of hydrophilic and hydrophobic room temperature ionic liquids incorporating the imidazolium cation. Green Chem. 3, 156–164. Itakura, T., Hirata, K., Aoki, M., Sasai, R., Yoshida, H., Itoh, H., 2008. Decompostition and removal of ionic liquid in aqueous solution by hydrothermal and photocatalytic treatment. Environ. Chem. Lett. doi:10.1007/s10311-008-0177-7. Jastorff, B., Sto¨rmann, R., Ranke, J., Mo¨lter, K., Stock, F., Oberheitmann, B., Hoffmann, W., Hoffmann, J., Nu¨chter, M., Ondruschka, B., Filser, J., 2003. How hazardous are ionic liquids? Structure-activity relationships and biological testing as important elements for sustainability evaluation. Green Chem. 5, 136–142. Jastorff, B., Mo¨lter, K., Behrend, P., Bottin-Weber, U., Filser, J., Heimers, A., Ondruschka, B., Ranke, J., Schaefer, M.,
Schro¨der, H., Stark, A., Stepnowski, P., Stock, F., Sto¨rmann, R., Stolte, S., Welz-Biermann, U., Ziegert, S., Tho¨ming, J., 2005. Progress in evaluation of risk potential of ionic liquids-basis for an eco-design of sustainable products. Green Chem. 7, 362–372. Kaiser, K.L.E., Palabrica, V.S., 1991. Photobacterium phosphoreum, toxicity data index. Water Poll. Res. J. Can. 26, 361–431. Kaiser, J.P., Feng, Y.C., Bollag, J.M., 1996. Microbial metabolism of pyridine, quinoline, acridine, and their derivatives under aerobic and anaerobic conditions. Microbiol. Rev. 60, 483–498. Kazarian, S.G., Briscoe, B.J., Welton, T., 2000. Combining ionic liquids and supercritical fluids: in situ ATR-IR study of CO2 dissolved in two ionic liquids at high pressures. Chem. Commun., 2047–2048. Kelman, D., Kashman, Y., Rosenberg, E., Ilan, M., Ifrach, I., Loya, Y., 2001. Antimicrobial activity of the reef sponge Amphimedon viridis from the Red Sea: evidence for selective toxicity. Aquat. Microb. Ecol 24, 9–16. Kulacki, K.J., Lamberti, G.A., 2008. Toxicity of imidazolium ionic liquids to freshwater algae. Green Chem. 10, 104–110. Kumar, R.A., Papaı¨conomou, N., Lee, J.-M., Salminen, J., Clark, D.S., Prausnitz, J.M., 2009. In vitro cytotoxicities of ionic liquids: effect of cation rings, functional groups, and anions. Environ. Toxicol. 24, 388–395. Kumar, S., Ruth, W., Sprenger, B., Kragl, U., 2006. On the biodegradation of ionic liquid 1-butyl-3-methylimidazolium tetrafluoroborate. Chim. Oggi 24, 24–26. Lahr, J., 1997. Ecotoxicology of organisms adapted to life in temporary freshwater ponds in arid and semi-arid regions. Arch. Environ. Contam. Toxicol 32, 50–57. Landry, T.D., Brooks, K., Poche, D., Woolhiser, M., 2005. Acute toxicity profile of 1-butyl-3-methylimidazolium chloride. Bull. Environ. Contam. Toxicol 74, 559–565. Larson, J.H., Frost, P.C., Lamberti, G.A., 2008. Variable toxicity of ionic liquid-forming chemicals to Lemna minor and the influence of dissolved organic matter. Environ. Toxicol. Chem. 27, 676–681. Lata1a, A., Stepnowski, P., Nedzi, ˛ M., Mrozik, W., 2005. Marine toxicity assessment of imidazolium ionic liquids: acute effects on the Baltic algae Oocystis submarina and Cyclotella meneghiniana. Aqua. Toxicol 73, 91–98. Lee, S.-M., Chang, W.-J., Choi, A.-R., Koo, Y.-M., 2005. Influence of ionic liquids on the growth of Escherichia coli. Korean. J. Chem. Eng 22, 687–690. Lewis, M.A., 1995. In: Rand, G.M. (Ed.), Fundamentals of Aquatic Toxicology: Effects, Environment Fate, and Risk Assessment, second ed. Taylor and Francis, Washington, DC, USA, pp. 135–170. Li, G., Shen, J., Zhu, Y., 1998. Study of pyridinium-type functional polymers. II. Antibacterial activity of soluble pyridinium-type polymers. J. Appl. Polym. Sci. 67, 1761–1768. Li, X., Zhao, J., Li, Q., Wang, L., Tsang, S.C., 2007. Ultrasonic chemical oxidative degradations of 1,3-dialkylimidazolium ionic liquids and their mechanistic elucidations. Dalton Trans, 1875–1880. Li, X.-Y., Zhou, J., Yu, M., Wang, J.-J., Pei, Y.C., 2009. Toxic effects of 1-methyl-3-octylimidazolium bromide on the early embryonic development of the frog Rana nigromaculata. Ecotox. Environ. Saf. 72, 552–556. Luis, P., Ortiz, I., Alcado, R., Irabien, A., 2007. A novel group contribution method in the development of a QSAR for predicting the toxicity (Vibrio fischeri EC50) of ionic liquids. Ecotoxicol. Environ. Saf 67, 423–429. Luo, Y.-R., Li, X.-Y., Chen, X.-X., Zhang, B.-J., Sun, Z.-J., Wang, J.-J., 2008. The developmental toxicity of 1-butyl-3-octylimidazolium bromide on Daphnia magna. Environ. Toxicol 23, 736–744. Malich, G., Markovic, B., Winder, C., 1997. The sensitivity and specificity of the MTS tetrazolium assay for detecting the in
water research 44 (2010) 352–372
vitro cytotoxicity of 20 chemicals using human cell lines. Toxicology 124, 179–192. Mann, R.M., Bidwell, J.R., 2000. Application of the FETAX protocol to assess the development toxicity of nonylphenol ethoxylate to Xenopus laevis and two Australian frogs. Aquat. Toxicol 51, 19–29. Marsh, K.N., Boxall, J.A., Lichtenthaler, R., 2004. Room temperature ionic liquids and their mixtures–a review. Fluid Phase Equilibr 219, 93–98. Massoulie´, J., Pezzementi, L., Bon, S., Krejci, E., Vallette, F.-M., 1993. Molecular and cellular biology of cholinesterases. Prog. Neurobiol. 41, 31–91. Matsumoto, M., Mochiduki, K., Fukunishi, K., Kondo, K., 2004a. Extraction of organic acids using imidazolium-based ionic liquids and their toxicity to Lactobacillus rhamnosus. Sep.Purif. Technol. 40, 97–101. Matsumoto, M., Mochiduki, K., Kondo, K., 2004b. Toxicity of ionic liquids and organic solvents to lactic acid-producing bacteria. J. Biosci. Bioeng 98, 344–347. Matzke, M., Stolte, S., Thiele, K., Juffernholz, T., Arning, J., Ranke, J., Welz-Biermann, U., Jastorff, B., 2007. The influence of anion species on the toxicity of 1-alkyl-3-methylimidazolium ionic liquids observed in an (eco)toxicological test battery. Green Chem. 9, 1198–1207. Matzke, M., Stolte, S., Bo¨schen, A., Filser, J., 2008. Mixture effects and predictability of combination effects of imidazolium based ionic liquids as well as imidazolium based ionic liquids and cadmium on terrestrial plants (Triticum aestivum) and limnic green algae (Scenedesmus vacuolatus). Green Chem. 10, 784–792. Matzke, M., Stolte, S., Arning, J., Uebers, U., Filser, J., 2009a. Ionic liquids in soils: effects of different anion species of imidazolium based ionic liquids on wheat (Triticum aestivum) as affected by different clay minerals and clay concentrations. Ecotoxicology 18, 197–203. Matzke, M., Thiele, K., Mu¨ller, A., Filser, J., 2009b. Sorption and desorption of imidazolium based ILs in different soil types. Chemosphere 74, 568–574. McFarlane, J., Ridenour, W.B., Luo, H., Hunt, R.D., Depaoli, D.W., Ren, R.X., 2005. Room temperature ionic liquids for separating organics from produced water. Sep. Sci. Technol 40, 1245–1265. McGuinness, D.S., Cavell, K.J., 2000. Donor-functionalized heterocyclic carbine complexes of palladium(II): efficient catalysts for C-C coupling reactions. Organometallics 19, 741–748. McQueen, D.J., Post, J.R., Mills, E.L., 1986. Trophic relationships in freshwater pelagic ecosystems. Can. J. Fish. Aqua. Sci. 43, 1571–1581. Morawski, A.W., Janus, M., Goc-Maciejewska, I., Syguda, A., Pernak, J., 2005. Decomposition of ionic liquids by photocatalysis. Pol. J. Chem. 79, 1929–1935. Myasoedov, B.F., Molochnikova, N.P., Shkinev, V.M., Spivakov, B.Y., 1995. The Behavior of Actinides in Two-Phase Aqueous Systems Based on Polyethylene Glycol. Plenum, New York, pp. 91. Nockemann, P., Thijs, B., Driesen, K., Janssen, C.R., Hecke, K.V., Meervelt, L.V., Kossmann, S., Kirchner, B., Binnemans, K., 2007. Choline saccharinate and choline acesulfamate: ionic liquids with low toxicities. J. Phys. Chem. B 111, 5254–5263. Olivier, H., 1999. Recent developments in the use of non-aqueous ionic liquids for two-phase catalysis. J. Mol. Cat. A: Chem. 146, 285–289. Osenberg, C.W., Mittelbach, G.G., 1989. Effects of body size on the predator–prey interaction between pumpkinseed sunfish and gastropods. Ecol. Monogr. 59, 405–432. Pernak, J., Czepukowicz, A., 2001. New ionic liquids and their antielectrostatic properties. Ind. Eng. Chem. Res. 40, 2379– 2383.
371
_ J., Mirska, I., 2001a. Synthesis and Pernak, J., Rogoza, antimicrobial activities of new pyridinium and benzimidazolium chlorides. Eur. J. Med. Chem. 36, 313–320. Pernak, J., Kalewska, J., Ksycin´ska, H., Cybulski, J., 2001b. Synthesis and anti-microbial activities of some pyridinium salts with alkoxymethyl hydrophobic group. Eur. J. Med. Chem. 36, 899–907. Pernak, J., Chwa1a, P., 2003. Synthesis and anti-microbial activities of choline-like quaternary ammonium chlorides. Eur. J. Med. Chem. 38, 1035–1042. Pernak, J., Sobaszkiewicz, K., Mirska, I., 2003. Anti-microbial activities of ionic liquids. Green Chem. 5, 52–56. Pernak, J., Goc, I., Mirska, I., 2004a. Anti-microbial activities of protic ionic liquids with lactate anion. Green Chem. 6, 323–329. Pernak, J., Sobaszkiewicz, K., Foksowicz-Flaczyk, J., 2004b. Ionic liquids with symmetrical dialkoxymethyl-substituted imidazolium cations. Chem. Eur. J 10, 3479–3485. Pham, T.P.T., Cho, C.-W., Vijayaraghavan, K., Min, J., Yun, Y.-S., 2008a. Effect of imidazolium-based ionic liquids on the photosynthetic activity and growth rate of Selenastrum capricornutum. Environ. Toxicol. Chem. 27, 1583–1589. Pham, T.P.T., Cho, C.-W., Min, J., Yun, Y.-S., 2008b. Alkyl-chain length effects of imidazolium and pyridinium ionic liquids on photosynthetic response of Pseudokirchneriella subcapitata. J. Biosci. Bioeng 105, 425–428. Pham, T.P.T., Cho, C.-W., Jeon, C.-O., Chung, Y.-J., Lee, M.-W., Yun, Y.-S., 2009. Identification of metabolites involved in the biodegradation of the ionic liquid 1-butyl-3-methylpyridinium bromide by activated sludge microorganisms. Environ. Sci. Technol 43, 516–521. Peraccini, D., Chiappe, C., Intorre, L., Pretti, C., 2007. In: Letcher, T. M. (Ed.), Thermodynamics, Solubility and Environmental Issues. Elsevier, UK, pp. 259–278. Pope, C., Karanth, S., Liu, J., 2005. Pharmacology and toxicology of cholinesterase inhibitors: uses and misuses of a common mechanism of action. Environ. Toxicol. Pharmacol 19, 433–446. Pretti, C., Chiappe, C., Pieraccini, D., Gregori, M., Abramo, F., Monni, G., Intorre, L., 2006. Acute toxicity of ionic liquids to the zebrafish (Danio rerio). Green Chem. 8, 238–240. Pretti, C., Chiappe, C., Baldetti, I., Brunini, S., Monni, G., Intorre, L., 2009. Acute toxicity of ionic liquids for three freshwater organisms: Pseudokirchneriella subcapitata, Daphnia magna and Danio rerio. Ecotoxicol. Environ. Saf. 72, 1170–1176. Ranke, J., Mo¨lter, K., Stock, F., Bottin-Weber, U., Poczobutt, J., Hoffmann, J., Ondruschka, B., Filser, J., Jastorff, B., 2004. Biological effects of imidazolium ionic liquids with varying chain length in acute Vibrio fischeri and WST-1 cell viability assays. Ecotox. Environ. Saf 58, 396–404. Ranke, J., Mu¨ller, A., Bottin-Weber, U., Stock, F., Stolte, S., Arning, J., Sto¨rmann, R., Jastorff, B., 2007a. Lipophilicity parameters for ionic liquid cations and their correlation to in vitro cytotoxicity. Ecotox. Environ. Saf 67, 430–438. Ranke, J., Stolte, S., Sto¨rmann, R., Arning, J., Jastorff, B., 2007b. Design of sustainable chemical products – the example of ionic liquids. Chem. Rev. 107, 2183–2206. Rogers, R.D., Bond, A.H., Bauer, C.B., Zhang, J., Rein, S.D., Chomko, R.R., Roden, D.M., 1995. Partitioning behavior of 99 Tc and, 129 I from simulated Handford tank wastes using polyethylene-glycol based aqueous biphasic systems. Solvent Extr. Ion Exch. 13, 689–713. Romero, A., Santos, A., Tojo, J., Rodrı´guez, A., 2008. Toxicity and biodegradability of imidazolium ionic liquids. J. Hazard. Mater 151, 268–273. Rooney, D.W., Seddon, K.R., 2001. In: Wypych, G. (Ed.), Handbook of Solvents. ChemTech Publishing, Toronto, pp. 1459–1484. Salminen, J., Papaiconomou, N., Kumar, R.A., Lee, J.-M., Kerr, J., Newman, J., Prausnitz, J.M., 2007. Physicochemical properties and toxicities of hydrophobic piperidinium and pyrrolidinium ionic liquids. Fluid Phase Equilibr 261, 421–426.
372
water research 44 (2010) 352–372
Samorı`, C., Pasteris, A., Galletti, P., Tagliavini, E., 2007. Acute toxicity of oxygenated and nonoxygenated imidazoliumbased ionic liquids to Daphnia magna and Vibrio fischeri. Environ. Toxicol. Chem. 26, 2379–2382. Schmid, A., Kollmer, A., Mathys, R.G., Withot, B., 1998. Development toward large-scale bacterial bioprocesses in the presence of bulk amounts of organic solvents. Extremophiles 2, 249–256. Sheldon, R., 2001. Catalytic reactions in ionic liquids. Chem. Commun, 2399–2407. Sheldon, R.A., 2005. Green solvents for sustainable organic synthesis: state of the art. Green Chem. 7, 267–278. Siedlecka, E.M., Stepnowski, P., 2009. The effect of alkyl chain length on the degradation of alkylimidazolium- and pyridinium-type ionic liquids in a Fenton-like system. Environ. Sci. Pollut. Res. 16, 453–458. Siedlecka, E.M., Mrozik, W., Kaczyn´ski, Z., Stepnowski, P., 2008a. Degradation of 1-butyl-3-methylimidazolium chloride ionic liquid in a Fenton-like system. J. Hazard. Mater 154, 893–900. Siedlecka, E.M., Go1ebiowski, ˛ M., Kumirska, J., Stepnowski, P., 2008b. Identification of 1-butyl-3-methylimidazolium chloride degradation products formed in Fe(III)/H2O2 oxidation system. Chem. Anal. (Warsaw) 53, 943–951. Sipes, I.G., Knudsen, G.A., Kuester, R.K., 2008. The effects of dose and route on the toxicokinetics and disposition of 1-butyl3-methylimidazolium chloride in male F-344 rats and female B6C3F1 mice. Drug Metab. Dispos. 36, 284–293. Sk1adanowski, A.C., Stepnowski, P., Kleszczyn´ski, K., Dmochowska, B., 2005. AMP deaminase in vitro inhibition by xenobiotics. A potential molecular method for risk assessment of synthetic nitro- and polycyclic musks, imidazolium ionic liquids and N-glucopyranosyl ammonium salts. Environ. Toxicol. Phar. 19, 291–296. Stasiewicz, M., Mulkiewicz, E., Tomczak-Wandzel, R., Kurmirska, J., Siedlecka, E.M., Go1ebiowski, M., Gajdus, J., Czerwicka, M., Stepnowski, P., 2008. Assessing toxicity and biodegradation of novel, environmentally benign ionic liquids (1-alkoxymethyl-3-hydroxypyridinium chloride, saccharine and acesulfamates) on cellular and molecular level. Ecotox. Environ. Saf 71, 157–165. Steinberg, S.M., Poziomek, E.J., Engelmann, W.H., Rogers, K.R., 1995. A review of environmental applications of bioluminescence measurements. Chemosphere 30, 2155–2197. Stepnowski, P., Sk1adanowski, A.C., Ludwiczak, A., qaczyn´ska, E., 2004. Evaluating the cytotoxicity of ionic liquids using human cell line HeLa. Hum. Exp. Toxicol 23, 513–517. Stepnowski, P., 2005. Preliminary assessment of the sorption of some alkyl imidazolium cations as used in ionic liquids to soils and sediments. Aust. J. Chem. 58, 170–173. Stepnowski, P., Zaleska, A., 2005. Comparison of different advanced oxidation processes for the degradation of room temperature ionic liquids. J. Photoch. Photobio. A 170, 45–50. Stepnowski, P., Mrozik, W., Nichthauser, J., 2007. Adsorption of alkylimidazolium and alkylpyridinium ionic liquids onto natural soils. Environ. Sci. Technol 41, 511–516. Stock, F., Hoffmann, J., Ranke, J., Sto¨rmann, R., Ondruschka, B., Jastorff, B., 2004. Effects of ionic liquids on the acetycholinesterase–a structure-activity relationship consideration. Green Chem. 6, 286–290. Stolte, S., Arning, J., Bottin-Weber, U., Matzke, M., Stock, F., Thiele, K., Uerdingen, M., Welz-Biermann, U., Jastorff, B., Ranke, J., 2006. Anion effects on the cytotoxicity of ionic liquids. Green Chem. 8, 621–629. Stolte, S., Matzke, M., Arning, J., Bo¨schen, A., Pitner, W.-R., Welz-Biermann, U., Jastorff, B., Ranke, J., 2007a. Effects of
different head groups and functionalized side chains on the aquatic toxicity of ionic liquids. Green Chem. 9, 1170–1179. Stolte, S., Arning, J., Bottin-Weber, U., Mu¨ller, A., Pitner, W.-R., Welz-Biermann, U., Jastorff, B., Ranke, J., 2007b. Effects of different head groups and functionalized side chains on the cytotoxicity of ionic liquids. Green Chem. 9, 760–767. Stolte, S., Abdulkarim, S., Arning, J., Blomeyer-Nienstedt, A.-K., Bottin-Weber, U., Matzke, M., Ranke, J., Jastorff, B., Tho¨ming, J., 2008. Primary biodegradation of ionic liquid cations, identification of degradation products of 1-methyl-3-octylimidazolium chloride and electrochemical wastewater treatment of poorly biodegradable compounds. Green Chem. 10, 214–224. ´ ska, S., Buszewski, B., 2009. Study of toxicity of imidazolium Studzin ionic liquids to watercress (Lepidium sativum L.). Anal. Bioanal. Chem. 393, 983–990. 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, 361–363. Swatloski, R.P., Holbrey, J.D., Memon, S.B., Caldwell, G.A., Caldwell, K.A., Rogers, R.D., 2004. Using Caenorhabditis elegans to probe toxicity of 1-alkyl-3-methylimidazolium chloride based ionic liquids. Chem. Commun, 668–669. Torrecilla, J.S., Garcı´a, J., Rojo, E., Rodrı´guez, R., 2009. Estimation of toxicity of ionic liquids in leukemia rat cell line and acetylcholinesterase enzyme by principal component analysis, neural networks and multiple linear regressions. J. Hazard. Mater 164, 182–194. U.S. Environmental Protection Agency 1998. Fate, transport, and transformation test guidelines. OPPTS 835.3110. Readily Biodegradability. Vijayaraghavan, K., Pham, T.P.T., Cho, C.-W., Won, S.W., Choi, S. B., Mao, J., Kim, S., Kim, Y.-R., Chung, B.W., Yun, Y.-S., 2009. An assessment on the interaction of a hydrophilic ionic liquid with different sorbents. Ind. Eng. Chem. Res. 48, 7283–7288. Wang, L.-S., Wang, L., Wang, L., Wang, G., Li, Z.-H., Wang, J.-J., 2009. Effect of 1-butyl-3-methylimidazolium tetrafluoroborate on the wheat (Triticum aestivum L.) seedlings. Environ. Toxicol 24, 296–303. Wang, X., Ohlin, C.A., Lu, Q., Fei, Z., Hu, J., Dyson, P.J., 2007. Cytotoxicity of ionic liquids and precursor compounds towards human cell line HeLa. Green Chem. 9, 1191–1197. Wasserscheid, P., Keim, W., 2000. Ionic liquids–New ‘‘solutions’’ for transition metal catalysis. Angew. Chem. Int. Ed. 39, 3772–3789. Wells, A.S., Coombe, V.T., 2006. On the freshwater ecotoxicity and biodegradation properties of some common ionic liquids. Org. Pro. Res. Dev. 10, 794–798. Welton, T., 1999. Room-temperature ionic liquids, solvents for synthesis and catalysis. Chem. Rev. 99, 2071–2083. Willauer, H.D., Huddleston, J.G., Griffin, S.T., Rogers, R.D., 1999. Partitioning of aromatic molecules in aqueous biphasic systems. Sep. Sci. Technol 34, 1069–1090. Wong, D.S.H., Chen, J.P., Chang, J.M., Chou, C.H., 2002. Phase equilibria of water and ionic liquids [emim][PF6] and [bmim][PF6]. Fluid Phase Equilibr 194–197, 1089–1095. Yu, M., Li, S.-M., Li, X.-Y., Zhang, B.-J., Wang, J.-J., 2009a. Acute effects of 1-octyl-3-methylimidazolium bromide ionic liquid on the antioxidant enzyme system of mouse liver. Ecotox. Environ. Saf. 71, 903–906. Yu, M., Wang, S.-H., Luo, Y.-R., Han, Y.-W., Li, X.-Y., Zhang, B.-J., Wang, J.-J., 2009b. Effects of the 1-alkyl-3-methylimidazolium bromide ionic liquids on the antioxidant defense system of Daphnia magna. Ecotox. Environ. Saf. 72, 1798–1804. Zhang, C., Malhotra, S.V., 2005. Increased paraoxon detection by acetylcholinesterase inactivation with ionic liquid additives. Tatanta 67, 560–563.
water research 44 (2010) 373–384
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
A QSAR model for predicting rejection of emerging contaminants (pharmaceuticals, endocrine disruptors) by nanofiltration membranes Victor Yangali-Quintanillaa,b,*, Anwar Sadmania, Megan McConvillea,b, Maria Kennedya, Gary Amya,b a
UNESCO-IHE, Institute for Water Education, Westvest 7, 2611AX Delft, The Netherlands Delft University of Technology, Stevinweg 1, 2628CN Delft, The Netherlands
b
article info
abstract
Article history:
A quantitative structure activity relationship (QSAR) model has been produced for pre-
Received 25 February 2009
dicting rejection of emerging contaminants (pharmaceuticals, endocrine disruptors,
Received in revised form
pesticides and other organic compounds) by polyamide nanofiltration (NF) membranes.
4 June 2009
Principal component analysis, partial least square regression and multiple linear regres-
Accepted 26 June 2009
sions were used to find a general QSAR equation that combines interactions between
Available online 3 July 2009
membrane characteristics, filtration operating conditions and compound properties for predicting rejection. Membrane characteristics related to hydrophobicity (contact angle),
Keywords:
salt rejection, and surface charge (zeta potential); compound properties describing
Pharmaceuticals
hydrophobicity (log Kow, log D), polarity (dipole moment), and size (molar volume, molec-
Endocrine disruptors
ular length, molecular depth, equivalent width, molecular weight); and operating condi-
Nanofiltration
tions namely flux, pressure, cross flow velocity, back diffusion mass transfer coefficient,
Modelling
hydrodynamic ratio (Jo/k), and recovery were identified as candidate variables for rejection
QSAR
prediction. An experimental database produced by the authors that accounts for 106 rejection cases of emerging contaminants by NF membranes as result of eight experiments with clean and fouled membranes (NF-90, NF-200) was used to produce the QSAR model. Subsequently, using the QSAR model, rejection predictions were made for external experimental databases. Actual rejections were compared against predicted rejections and acceptable R2 correlation coefficients were found (0.75 and 0.84) for the best models. Additionally, leave-one-out cross-validation of the models achieved a Q2 of 0.72 for internal validation. In conclusion, a unified general QSAR equation was able to predict rejections of emerging contaminants during nanofiltration; moreover the present approach is a basis to continue investigation using multivariate analysis techniques for understanding membrane rejection of organic compounds. ª 2009 Elsevier Ltd. All rights reserved.
* Corresponding author: UNESCO-IHE, Institute for Water Education, Westvest 7, 2611AX Delft, The Netherlands. Tel.: þ31 15 215 1745; fax: þ31 15 215 2921. E-mail addresses:
[email protected],
[email protected] (V. Yangali-Quintanilla). 0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2009.06.054
374
1.
water research 44 (2010) 373–384
Introduction
Nanofiltration (NF) and reverse osmosis (RO) are technologies that provide medium to high rejections of organic compounds present as emerging contaminants (micropollutants) in water, namely endocrine disrupting compounds (EDCs), pharmaceutically active compounds (PhACs), and pesticides (Kiso et al., 2001; Scha¨fer et al., 2003; Kimura et al., 2003, 2004; Nghiem et al., 2004). The presence of micropollutants has been identified in surface water bodies, sewage treatment plant effluents, and stages of drinking water treatment plants and even at trace-levels in finished drinking water (Kolpin et al., 2002; Heberer, 2002; Castiglioni et al., 2006). The possible effects on aquatic organisms and human health, associated with the consumption of water containing low concentrations of single compounds, have been presented in toxicology studies (Pomati et al., 2006; Escher et al., 2005; Vosges et al., 2008). The studies demonstrate that researchers do not yet understand the exact risks from decades of persistent exposure to random combinations of low levels of pharmaceuticals, EDCs, and other organic compounds; hence, the long-term effects of consumption of water containing low concentrations of micropollutants will remain as an unanswered question for the foreseeable future. Meanwhile, water treatment facilities are implementing monitoring programs, research organizations dealing with water reuse have published reports, and studies have addressed the topic (Drewes et al., 2006; Verliefde et al., 2007). An important aspect to deal with the problem has been the identification of compound physicochemical properties and membrane characteristics to explain transport, adsorption and removal of micropollutants by different mechanisms, explicitly size/ steric exclusion, hydrophobic adsorption and partitioning, and electrostatic repulsion (Kiso et al., 2001; Scha¨fer et al., 2003; Kimura et al., 2003; Nghiem et al., 2004, Ozaki and Li, 2002; Van der Bruggen and Vandecasteele, 2002; Bellona and Drewes, 2005; Xu et al., 2005). A number of articles have proposed a mechanistic understanding of the interaction between membranes and organic compounds; others have tried to apply fitting parameter models to model rejection (Cornelissen et al., 2005; Kim et al., 2007; Verliefde et al., 2008). However, there have been few models to ‘‘predict’’ the rejection of compounds. To overcome that status, our objective was to create a general QSAR model to predict rejections based on an integral approach that considers membrane characteristics, filtration operating conditions and physicochemical compound properties. A quantitative structure activity relationship (QSAR) is a method that relates an activity of a set of compounds quantitatively to chemicals descriptors (structure or property) of those compounds (Sawyer et al., 2003). QSAR has the objective of prediction but maintaining a relationship to mechanistic interpretation. Applications of QSAR for the development of models to find relationships between membranes and organic compounds have been presented in journals related to drug discovery and medicinal chemistry for analysis of permeability of membranes to organic compounds (Ren et al., 1996; Fujikawa et al., 2007). The study of reverse
osmosis membranes has also experienced the application of QSAR principles. Campbell et al. (1999) performed a QSAR analysis of surfactants influencing attachment of a mycobacterium to cellulose and aromatic polyamide reverse osmosis membranes; their objective was to understand the relationship between surfactant molecular properties and activity on the membrane surface to inhibit bacterial attachment to the membrane in order to reduce biofilm formation and to increase permeate production. More recently, Libotean et al. (2008) developed an artificial neural network model based on quantitative structure-property relations, the model claim to predict organic solute passage through reverse osmosis membranes considering simultaneous correlation of organic solutes (molecular descriptors) and membrane properties. Alike, our study uses the concept of QSAR analysis to quantify an activity, compound rejection by a membrane, in terms of organic compound physicochemical properties, membrane characteristics (salt rejection, pure water permeability, molecular weight cut-off, charge, hydrophobicity) and operating conditions (pressure, flux, cross flow velocity, back diffusion mass transfer coefficient, recovery). In this work a QSAR model was constructed with internal experimental data used for training. The model was internally validated using measures of goodness of fit and prediction. Subsequently, after identification of a relationship in form of an equation, estimation of rejections for an external dataset for different compounds and membranes were used to externally validate the model. Similarly, rejections of more emerging organic contaminants can be predicted in advance, before nanofiltration or reverse osmosis applications. Nevertheless, the QSAR model is applicable in the range of boundary experimental conditions that will be defined in the experimental section of this publication (Section 3).
2.
Theory
2.1.
Principal component analysis
Principal component analysis (PCA) is a method that allows simplification of many variables into a group of a few variables that might be measuring the same principles of a system. It may occur that a system considers an abundance of variables to explain a process; in this case principal component analysis reduces the redundancy of information. The general objectives of PCA are data reduction and interpretation. Although p variables (components) are required to reproduce the total system variability, often much of this variability can be accounted for by a small number k of principal components. The k is the number of components (reduced) that represent the initial p variables. In general, PCA is concerned with whether the covariances or correlations between a set of observed variables x1, x2, ., xp can be explained in terms of a smaller number of unobservable components, c1, c2, ., ck, where k < p. Thus, there is as much information in the k components as there is in the original p variables. Comprehensive details about the theory of PCA can be found elsewhere (Jolliffe, 2002; Johnson and Wichern, 2007; Everitt and Dunn, 2001).
water research 44 (2010) 373–384
2.2.
Multiple linear regression
Multiple linear regression is a method of analysis for assessing the strength of the relationship between a set of explanatory variables known as independent variables, and a single response or dependent variable. Applying multiple regression analysis to a set of data results in what are known as regression coefficients, one for each explanatory variable. The multiple regression model for a response variable, y, with observed values, y1, y2, ., yn (where n is the sample size) and q explanatory variables, x1, x2, ., xq with observed values, x1i, x2i, ., xqi for i ¼ 1, ., n, is yi ¼ b0 þ b1 x1i þ b2 x2i þ / þ bq xqi þ 3i
(1)
The regression coefficients, b0, b1, ., bq, are generally estimated by least squares. The term 3i is the residual or error for individual i and represents the deviation of the observed value of the response for this individual from that expected by the model. These error terms are assumed to have a normal distribution with variance s2. The fit of a multiple regression model can be judged with calculation of the multiple correlation coefficient, R, defined as the correlation between the observed values of the response variable and the values predicted by the model. The squared value of R (R2) gives the proportion of the variability of the response variable accounted for by the explanatory variables. Analysis of variance (ANOVA) will provide an F-test of the null hypothesis that each of b0, b1, ., bq, is equal to zero, or in other words that R2 is zero (Landau and Everitt, 2004).
2.3. Principal component regression and partial least squares regression Principal component regression (PCR) is a method in which the components from the principal component method are used for regression. Hence, the principal components of the matrix X are used as regressors of a dependent Y. The orthogonality of the principal components eliminates the multicollinearity problem. But, the problem of choosing an optimum subset of predictors remains. A possible strategy is to keep only a few of the first components. But they are chosen to explain X rather than Y, and therefore, nothing guarantees that the principal components, which ‘‘explain’’ X, are relevant for Y. Problems may arise, however, if there is a lot of variation in X. PCR finds, somewhat uncritically, those latent variables that describe as much as possible of the variation in X. But sometimes the variable itself gives rise to only small variations in X, and if the interferences vary a lot, then the latent variables found by PCR may not be particularly good at describing Y. In the worst case important information may be hidden in directions in the X-space that PCR interprets as disturbance, and therefore leaves out. Partial least squares regression (PLS) is able to cope better with this problem, by forming variables that are relevant for describing Y. By contrast, PLS regression finds components from X that are also relevant for Y. Specifically, PLS regression searches for a set of components (called latent vectors) that performs a simultaneous decomposition of X and Y with the constraint that these components explain as much as possible of the
375
covariance between X and Y. This step generalizes PCA. The goal of PLS regression is to predict Y from X and to describe their common structure (Abdi, 2003; Jørgensen and Goegebeur, 2006).
3.
Experimental
3.1. Chemicals, membranes, materials and experimental conditions The list of organic compounds representing emerging contaminants is presented in Table 1. The compounds were selected considering: their occurrence in surface water and drinking water, their identification as priority emerging contaminants, the availability of analytical methods, their quantitative and qualitative representation of physicochemical properties. The pharmaceutical compounds (caffeine, sulfamethoxazole, acetaminophen, phenacetin, phenazone, carbamazepine, naproxen, ibuprofen, metronidazole), endocrine disrupting compounds (17b-estradiol, estrone, bisphenol A, nonylphenol, atrazine) and sodium alginate were purchased from Sigma–Aldrich (Sigma–Aldrich, Schnelldorf, Germany). Potassium chloride, sodium hydroxide, hydrochloric acid and magnesium sulphate anhydrous were purchased from J.T.Baker (J.T.Baker, Deventer, Netherlands). Two thin film composite NF membranes were selected for this study (NF-200 and NF-90, Dow-Filmtec, Dow Chemical Co., Midland, MI). The experimental setup consisted of two filtration SEPA CF II (GE Osmonics, Minnetonka, MN) cells and cell holders in parallel, in order to increase permeate production and achieve hydrodynamic conditions, two hydraulic pumps (Power Team, Bega Int. BV, Netherlands), a 60 litres stainless steel tank (Tummers, Netherlands), a positive displacement pump (Hydra-Cell pump, Wanner Eng. Inc., Minneapolis, MN), a frequency converter (VLT microdrive, Danfoss, SA), a chiller/heater (Julabo, Germany), control needle valves, pressure gauges, flow meters, a proportional pressure relief valve and stainless steel tubings (Swagelok BV, Netherlands), a digital balance (Sartorius, Germany) and, a computer for flow rate data acquisition. A piece of membrane was compacted with deionised water for 6 h at a pressure of 276 kPa before performing an experiment with the membrane. The experiments were conducted in a recycle mode in which permeate and concentrate were recirculated into the feed tank for the first 72 h (a pre-equilibration period); then, permeate was collected within the next 24 h. The feed solution of all the experiments contained a cocktail of 14 compounds (concentration ranging from 6.5 to 65 mg/L). The main reason for conducting experiments at concentrations of mg/L was to accelerate steady state (after membrane adsorption) conditions in a limited time (3 days). At very low concentrations more time would be needed to achieve steady state conditions and at short time tests low concentrations may lead to over-estimation of rejections. A specific flux decline of 15% of initial flux was targeted to foul the membranes using a feed solution containing w10 mg/L DOC of sodium alginate. The study of Lee et al. (2004) concluded that polysaccharides were important membrane foulants. Sodium alginate was used as
376
water research 44 (2010) 373–384
Table 1 – List of compounds and physicochemical properties. Name
Acetaminophen Phenacetine Caffeine Metronidazole Phenazone Sulfamethoxazole Naproxen Ibuprofen Carbamazepine Atrazine 17b-estradiol Estrone Nonylphenol Bisphenol A a b c d e f
Molecular Acid Log Kowb Log Da (pH 7) weight pKa 20 Ca (g/mol) 151 179 194 171 188 253 230 206 236 216 272 270 220 228
10.2 N/A N/A N/A N/A 5.7 4.3 4.3 N/A N/A 10.3 10.3 10.3 10.3
0.46 1.58 0.07 0.02 0.38 0.89 3.18 3.97 2.45 2.61 4.01 3.13 5.71 3.32
0.23 1.68 0.45 0.27 0.54 0.45 0.34 0.77 2.58 2.52 3.94 3.46 5.88 3.86
Dipole moment (debye)c
Molar volumed (cm3/ mol)
Molec. length (nm)e
Molec. width (nm)e
Molec. depth (nm)e
4.55 4.05 3.71 6.30 4.44 7.34 2.55 4.95 3.66 3.43 1.56 3.45 1.02 2.13
120.90 163.00 133.30 117.80 162.70 173.10 192.20 200.30 186.50 169.80 232.60 232.10 236.20 199.50
1.14 1.35 0.98 0.93 1.17 1.33 1.37 1.39 1.20 1.26 1.39 1.39 1.79 1.25
0.68 0.69 0.87 0.90 0.78 0.71 0.78 0.73 0.92 1.00 0.85 0.85 0.75 0.83
0.42 0.42 0.56 0.48 0.56 0.58 0.75 0.55 0.58 0.55 0.65 0.67 0.59 0.75
Equiv. Classificationf width (nm)e 0.53 0.54 0.70 0.66 0.66 0.64 0.76 0.64 0.73 0.74 0.74 0.76 0.66 0.79
HL-neutral HL-neutral HL-neutral HL-neutral HL-neutral HL-ionic HB-ionic HB-ionic HB-neutral HB-neutral HB-neutral HB-neutral HB-neutral HB-neutral
ADME/Tox Web Software. Experimental database: SRC PhysProp Database. Chem3D Ultra 7.0. ACD/ChemSketch Properties Batch. Molecular Modeling Pro. HL ¼ hydrophilic, HB ¼ hydrophobic.
surrogate of polysaccharides. Alginate is frequently used as a model for organic matter of algae origin (Henderson et al., 2008). Experiments were carried out for both clean and fouled NF-90 and NF-200 membranes. The selection of membranes was based on a qualitative rejection assessment of emerging contaminants with molecular weight of more than 150 Da by membranes with a MWCO between 200 and 300 Da. A total of eight experiments were performed; at hydrodynamic ratio of pure water permeation flux to back diffusion mass transfer coefficient (J0/k) of w1 and recovery 3% (NF-90 clean, NF-200 clean), at J0/k of w1 and recovery 8% (NF-200 clean), at J0/k of w2 and recovery 3 (NF-90 clean), and at J0/k of w2 and recovery 8 (NF-90 clean and fouled, NF-200 clean and fouled). The calculation method of k (back diffusion mass transfer coefficient) was presented in a previous publication (YangaliQuintanilla et al., 2009). All the experiments were carried out at a controlled temperature of 20 C, an ionic strength of 10 mM as KCl and a pH of 7. The transmembrane pressures were in the range of 276–483 kPa. The fluxes were between 4.3 and 30.2 L/m2 per day; cross flow velocities were between 0.5 and 4.5 cm/s. The experiments produced a total internal dataset of 106 rejection cases; the dataset can be accessed as supplementary data. The boundary experimental conditions of the internal dataset are presented in Table 2. The internal dataset was used to develop the model. An external dataset that gathered three different datasets was used for validation of the model. The external dataset is presented as supplementary data. Experimental conditions for the first part of the external dataset can be obtained from Kim et al. (2007) and Yangali-Quintanilla et al. (2008). Experimental conditions for the second and third parts can be found in Verliefde et al. (2008); the data correspond to filtration experiments using synthetic water solutions.
3.2. Analytical equipment, analyses of compounds and membranes The pharmaceuticals and endocrine disruptors (with the exception of atrazine) were analyzed by Technologiezentrum Wasser (TZW, Karlsruhe, Germany). The detection limit was 10 ng/L per compound. The uncertainty of estimates was of 15% according to a validation method of the analysis protocol and due to high concentrations of the samples (mg/L). The analyses of pharmaceuticals were performed according to
Table 2 – Data range of membrane characteristics, operating conditions and rejections. Variable Molecular weight cut-off (MWCO) Pure water permeability (PWP) Salt rejection (SR)a Zeta potential (ZP) Contact angle (CA) Pressure (P) Cross flow velocity (v) Back diffusion mass transfer coefficient (k) Flux (J ) Hydrodynamic ratio (J0/k) Recovery (recov) Rejection (rejection)
Unit
Min. value Max. value
Da
200
300
L/m2 per day/kPa – MV
0.86
2.23
kPa cm/s cm/s
0.96 48.04 39.3 276 0.5 2.70E-04
0.98 10.78 58.0 483 4.5 5.99E-04
L/m2 per day –
4.3 1
30.2 2
% %
3 17.7
8 99.0
a 2000 mg/L MgSO4, 25 C, recovery 15%, pressure 1034 kPa, pH 8.
water research 44 (2010) 373–384
the protocols described by Sacher et al. (2001, 2008). Analyses of 17b-estradiol, estrone, nonylphenol and bisphenol A were done by gas chromatography/mass spectrometry (GC/MS) after automated solid-phase extraction onto a polymeric material and subsequent silylation of the analytes. First, 10 ml of a 50 ng/ml solution of 4-n-nonylphenol in acetone which was used as internal standard for the overall procedure were added to an aliquot of the water sample (1000 ml). Automated solid-phase extraction (Tekmar AutoTrace, Germany) was done on plastic cartridges filled with 200 mg of bondelut material (Fa. Varian, Darmstadt, Germany). After the enrichment step the solid-phase material was dried in a gentle stream of nitrogen. Elution was done with 4 ml of acetone. The acetone was evaporated to 100 ml in a stream of nitrogen and to dryness in a drying oven at 80 C. The dry residue was reconstituted with 100 ml of a silylation reagent mixture (N-methyl-N-trimethylsilyltrifluoro acetamide (MSTFA)/2% trimethyliodo silane). After a reaction time of 20 min at 80 C (drying oven), determination of the derivatives was done by GC/MS using a 6890 GC/MS system from Agilent Technologies (Waldbronn, Germany). Concentrations of atrazine were determined using microplate enzyme-linked immunoabsorbent assay (ELISA) kits (Abraxis LLC, Warminster, PA). Atrazine was determined with a detection limit of 0.04 mg/L, and uncertainty of 15%. To determine the hydrophobicity of membranes, contact angles of clean and fouled membrane surfaces were measured with CAM200 optical contact angle and surface tension meter (KSV Instruments, Finland) at Delft University of Technology; to measure contact angle, the sessile drop method was used. Surface charge, in terms of zeta potential, of clean and fouled membranes was quantified using ELS-8000 zeta potential analyzer (Otsuka Electronics, Japan). The zeta potential analyses were determined using a Milli-Q water solution at pH 7 and ionic strength of 10 mM KCl. The zeta potential was determined using the electrophoresis method using a cell consisting of membrane and quartz cells. The zeta potential was calculated from the electrophoretic mobility using the Smoluchowski formula, a detailed explanation of calculation was provided in a previous publication (Shim et al., 2002). The pH of the solutions was measured using a calibrated Metrohm 691 pH-meter (Metrohm AG, Herisau, Switzerland); the electrical conductivity and temperature were measured with a WTW Cond 330i (WTW GmbH, Weilheim, Germany) portable conductivity meter. Clean and fouled membranes were characterized to determine magnesium sulphate salt rejection at standard conditions specified by manufacturers, a pure water solution containing 2000 mg/L of magnesium sulphate at 25 C and pH 8 was filtrated at pressure of 1034 kPa and recovery of 15%.
3.3.
Characterization and classification of compounds
The acid dissociation constant as log Ka (pKa) was used to determine the speciation of the organic compound in ionic species at pH 7. For hydrophobicity determination, log Kow and log D were used; log Kow is the octanol–water partition coefficient and log D is the ratio of the equilibrium concentrations of all species (unionized and ionized) of a molecule in octanol to the same species in the water phase. Values of pKa and log D
377
were calculated by ADME/Tox web software. Solubility and log Kow values were obtained from SRC Physprop experimental database. The value of the molecular dipole moment was equal to the vector sum of the individual bond dipole moments. Dipole moment was calculated by Chem3D Ultra 7.0, Cambridgesoft. Size descriptors included molar volume (MV), molecular length, molecular width, molecular depth and equivalent molecular width. The molecular length is defined as the distance between the two most distant atoms. The molecular width and molecular depth (width > depth) are measured by projecting the molecule on the plane perpendicular to the length axis and the equivalent molecular width is defined as the geometric mean of width and depth (Santos et al., 2006). Molar volumes were calculated using the program ACD/ChemSketch Properties Batch, ACD/Labs; and Molecular Modeling Pro, ChemSW, was used to compute size descriptors after optimization geometry of a molecule from the interaction of conformational analysis and energy minimization with a semi-empiric method MOPAC-PM3. Based on pKa and log Kow values, the compounds were classified as hydrophilic neutral, hydrophilic ionic, hydrophobic ionic and hydrophobic neutral (see Table 1). Compounds with log Kow 2 were referred to as hydrophobic; therefore those with log Kow < 2 were hydrophilic. The classification was based on an early reference (Connell, 1990). Although the value of 2 may seem low to consider hydrophobicity, the classification was not used in constructing the models and therefore the magnitude of log Kow or log D became more important. Table 1 shows the calculated values of molecular weight, pKa, log Kow, log D, dipole moment, molar volume, molecular length, molecular width, molecular depth and equivalent width.
4.
Results and discussion
4.1.
QSAR methodology
The procedure to find a general QSAR equation to describe rejection was performed in four phases. The first phase was the organization of data from the experimental part. The data comprised of 106 rejection cases. The database showing the rejection cases is presented as supplementary data. A total of 21 initial variables were used. The variables considered as compound descriptors were molecular weight (MW), solubility, log Kow, log D, dipole moment, molar volume, molecular length, molecular width, molecular depth and equivalent width; variables describing membrane characteristics were molecular weight cut-off (MWCO), pure water permeability (PWP), magnesium sulphate salt rejection (SR), charge of the membrane as zeta potential (ZP), and hydrophobicity as contact angle (CA); variables describing operating conditions were operating pressure (P), cross flow velocity (v), back diffusion mass transfer coefficient (k), flux (J ), ratio of pure water permeation flux J0 and back diffusion mass transfer coefficient (J0/k) and recovery. The range of values for membrane characteristics, operating conditions and rejections is presented in Table 2. The second phase was dedicated to the process of variables reduction using a correlation matrix and factor analysis with principal component analysis. The third phase corresponded to the regression analysis. In
378
water research 44 (2010) 373–384
the third phase three methodologies were implemented; the first was principal component analysis (PCA), with sequential application of multiple linear regressions (MLR). The second method was the use of partial least squares (PLS) regression and MLR; and the third method was the use of MLR only. The last phase was the validation process. The model was internally validated using measures of goodness of fit (regression coefficients) and prediction (leave-one-out cross-validation); Section 4.5 gives details about the validation process. External validation of the general QSAR model was implemented by predicting rejections for an external dataset of experiments performed with different compounds and membranes, and with comparable operating conditions. PCA and PLS were performed using the research and statistical package SPSS Statistics 16.0. Leave-one-out cross-validations of the models were performed with MobyDigs (Talete, Milano, Italy).
4.2. Variables reduction with principal component analysis and QSAR model The correlation matrix of the initial 21 variables was scrutinized in order to obtain a not positive definite matrix as a requisite of PCA; the matrix is accompanied as supplementary data. A matrix is called not positive definite when there are both positive and negative eigenvalues. In the case of symmetric matrices, such as a correlation matrix, positive definiteness will only hold if the matrix and every principal submatrix have a positive determinant. A non-positive definite input matrix may signal a perfect linear dependency of one variable on another, known as collinearity. This was the case for MWCO and salt rejection (SR) that were perfectly linearly correlated. Therefore application of PCA considering independently one variable or the other will give the same results of variables reduction and number of components. In other words, MWCO will not be excluded with PCA, the variable will be separated in advance and the results obtained for SR may be replaced by the variable MWCO, or vice versa. Once an appropriate matrix was defined, the variables were analyzed in terms of how significant their correlations with rejection were; those correlations are also shown as an additional row and column of the 21 21 variables matrix. Rejection is only a reference variable to evaluate correlation with the rest of variables. After a sequential implementation of PCA, three components were extracted; they defined the initial database of 21 variables with 11 variables describing three relations namely membrane/operating-conditions (comp. 1: flux, pure water permeability, salt rejection, zeta potential, mass transfer coefficient, cross flow velocity), hydrophobicity/size (comp. 2: length, log Kow, log D) and size (comp. 3: equivalent width, depth). The final three components accounted for 89.3% explanation of total variance. It is important to mention that these results were produced for the experimental dataset. The next step was the implementation of multiple linear regression (MLR) using the new set of variables. The use of MLR after PCA presents the advantage of a more simplified modelling approach. Moreover, the analysis of data before MLR may help to identify variables that are similar in response, which was the case of SR and MWCO. The dependent variable for all regression analyses was rejection. Two
methods of linear regression were used, the first method is called enter (forced) method; which performs a regression with the contribution of all variables entered to model the dependent variable. The second method is stepwise regression; which is a more sophisticated method. Each variable is entered in sequence and its contribution is assessed according to an F-test. In the present study an F-test with a statistical significance >0.10 implied removal of the variable, and F-test with a significance <0.05 implied entry of the variable in the model. If adding the variable contributes to the model then it is retained, but all other variables in the model are then retested to see if they are still contributing to the success of the model, otherwise an elimination process is carried out to remove variables that are no longer judged to improve the model. Therefore, the method should ensure that the model contains a set of appropriate predictor variables. Considering the previous explanation, the variables log Kow, log D, length, depth, eqwidth, PWP, SR, ZP, v, k, J and the 106 rejection cases were used to model rejection. The final regression resulted in variables SR, eqwidth, log D, length and depth. The model presented a correlation coefficient R2 of 0.75 with an F-test of 60.2, with a statistical significance of w0%. Besides, all coefficients of the model variables showed very acceptable significances (<0.001). Therefore the QSAR linear equation model for rejection can be written as rejection ¼
265:150 eqwidth 117:356 depth þ 81:662 length 5:229 log D þ 1358:090 SR 1447:817 (2)
Where, the units of the variables were specified in Tables 1 and 2. Eq. (2) can be mechanistically interpreted; rejection will increase by the effect of size, which is explained by the positive coefficients of length and equivalent width. The mechanism of steric hindrance due to size exclusion has been recognized as a main cause of rejection in many studies (Kiso et al., 2001; Van der Bruggen et al., 1999; Ozaki and Li, 2002; Nghiem et al., 2005). Contrarily, the negative coefficient of log D will decrease rejection, which clearly states that the effect of hydrophobicity lessens rejection due to adsorption and subsequent partitioning mechanisms. Indeed, partitioning is a combined effect that is not only compound property dependent (size, hydrophobicity) but also is related to membrane characteristics. It is important to mention that log D is assuming the role of hydrophobicity for neutral and ionic compounds; compounds with high log D values will adsorb to the membrane and may partition after saturation. Contrarily, ionic compounds will present log D values depending on the pH of the solution, thus may not adsorb when log D is very low or negative, and may adsorb at greater values of log D. Hydrophobicity influences rejection after adsorptive interactions with the membrane; this fact has been documented in some studies (Kiso et al., 2001; Kimura et al., 2004). The role of depth in the equation will compensate size exclusion contributions of length and equivalent width in a final rejection. The equation also shows that salt rejection (SR) is a parameter incorporating steric/size hindrance and electrostatic repulsion effects related to charge of the membrane and operating conditions. This effect may be related to cake-enhanced concentration polarization affecting the salt rejection of clean and fouled membranes (Hoek and
water research 44 (2010) 373–384
Elimelech, 2003). Thus, SR is ultimately serving as a comparison parameter between membranes of the same type (aromatic polyamide) but with narrow differences in pore size, and possibly with differences in charge. The QSAR equation merges information about interaction of membrane characteristics, filtration operating conditions and organic compounds properties to predict rejections during nanofiltration. According to Eq. (2), contact angle and zeta potential as measurements of hydrophilicity and membrane surface charge, respectively, were not part of the equation and therefore did not contribute quantitatively to model rejection. However, size/steric hindrance effects related to salt rejection, and hydrophobicity of the solutes were part of the model equation. In conclusion, rejection increased by size/steric hindrance effects; but hydrophobicity decrease rejection due to adsorption and partitioning mechanisms. The results of PCA and the model cannot be generalized to experimental conditions outside of the boundary experimental conditions defined in Section 3. For instance, excessive changes of pH affect the ionic speciation of charged compounds, obviously pKa values of solutes and pH of feed waters will determine boundary conditions for applicability of the model. Changes of membrane properties such as charge and pore size due to swelling will also influence the model. Other considerations for application of the model are the type of membrane used (aromatic polyamide), fluxes, pressures and cross flow velocities. Cellulose acetate or even different membrane chemistry than NF-90 or NF-200 will influence the PCA and the model. Nonetheless, the approach is valid and can be generalized under certain conditions in upscale NF applications.
4.3. QSAR model after partial least squares regression and MLR The following variables: MW, solubility, MV, MWCO, ZP, CA, P, v, J, Jo/k, recovery and width, were removed after partial least squares (PLS) regression. Therefore, the final PLS model is defined by variables named log Kow, log D, dipole, length, depth, eqwidth, PWP, SR and k. The main advantage of PLS is the ability to handle collinearity among the independent variables. In contrast, principal component analysis can not control collinearity. Another advantage of PLS was that the calculation process was simpler. However, PLS is used more as a predictive technique and not as an interpretive technique such as MLR. Therefore, in this exploratory analysis, PLS regression serves as a variable selection process and as prelude for implementation of MLR. Once again as occurred with PCA, a reduced number of variables simplified the implementation of MLR. After applying stepwise regression to the reduced number of variables, the model result obtained was also Eq. (2).
4.4.
QSAR model after multiple linear regression
To finalize the model development under various statistical application scenarios, the implementation of direct multiple linear regression was performed. The R2 is calculated for all possible subset models. Using this technique, the model with the largest R2 is declared the best linear model. However, this
379
technique has some disadvantages. First, the R2 increases with each variable included in the model. Therefore, this approach encourages including all variables in the best model although some variables may not significantly contribute to the model. This approach also contradicts the principal of parsimony that encourages as few parameters in a model as possible. Thus, the application of MLR without previous data analysis is a possibility when the number of variables is limited. Another disadvantage during the present study was that MLR was not able to distinguish collinearity between variables. This was the case for variables MWCO and SR; MWCO can replace the role of SR in Eq. (2). The new equation presented an R2 of 0.75; the F-test was 52.5 with a significance of w0%. The equation was the following rejection ¼
265:150 eqwidth 117:356 depth þ 81:662 length 5:229 log D 0:272 MWCO 62:565 (3)
It is important to mention that Eq. (3) may have been equally defined in Section 4.2 if the variable selected before implementing PCA was MWCO and not SR as explained before; therefore it is not surprising that Eqs. (2) and (3) only show differences in two coefficients. However, considering practical or operational facts, it may be more difficult to determine changes of MWCO when fouling occurs; on the other hand, salt rejections tests are part of monitoring practices. In addition, the variable of MWCO may be difficult to define when a range of MWCO exists for a membrane rather than an approximate single MWCO value if compared to salt rejection, although that fact is not desirable for a membrane because will greatly influence rejection of solutes with sizes close to or in the range of MWCO. It may be argued that a simple stepwise MLR will produce the same results of a PCA or PLS followed by MLR, however, the application of MLR without PCA or PLS has the disadvantage of removing or adding appropriate variables during the iterative process of stepwise MLR. The stepwise MLR is only dependent on the fulfillment of a statistical condition; it even may happen that different combination of variables defines a good equation. However, the advantage of PCA or PLS is that only important variables are considered, and only those will be part of the final MLR implementation.
4.5.
Internal and external validation of the QSAR model
Actual (measured) rejection values (106 rejection cases) versus modelled (fitted) rejections of the data used to generate the model are shown in Fig. 1, the dataset is provided as supplementary data; a 95% confidence interval shows that very few modelled rejections were out of that interval. Besides of a good fit of a model, it is necessary an assessment of the predictive power of the model, i.e. appropriate robustness (Eriksson et al., 2003). The R2 is the most widely used measure of the ability of a QSAR model to reproduce the internal data in the training (goodness of fit), but does not explain its robustness and prediction power. One technique to evaluate prediction is the leave-one-out cross-validation technique, in which one case at a time is iteratively held-out from the training set and the rest is used for model development and the excluded case is predicted by the developed model (Gramatica, 2007).
380
water research 44 (2010) 373–384
b HB-ion HB-neu HL-ion HL-neu
95% confidence interval
length, eqwidth, depth, log D, SR R² = 0.75
modeled rejection (%)
measured rejection (%)
measured rejection (%)
a
HB-ion HB-neu HL-ion HL-neu
95% confidence interval
length, eqwidth, depth, log D, MWCO R² = 0.75
modeled rejection (%)
Fig. 1 – QSAR model of experimental database: (a) Eq. (2); and (b) Eq. (3).
According to Gramatica, the predictive power of a model may be estimated by the goodness of prediction parameter Q2 leave-one-out (1-PRESS/TSS, where PRESS is the predictive error sum of squares and TSS is the total sum of squares). In general, a Q2 > 0.5 is regarded as good and Q2 > 0.9 as excellent (Eriksson et al., 2003). For the developed QSAR models, the model with SR (Eq. (2)) presented a Q2 leave-one-out of 0.72, and the model with MWCO (Eq. (3)) presented a Q2 leave-oneout of 0.72. After internal cross-validation it was demonstrated that Eqs. (2) and (3) were valid to model rejection, however, an adjustment must be made to the equation before using it to compare measured vs. predicted rejections for external databases. This adjustment was necessary to overcome the mathematical structure of the equation. Using a physical interpretation, it was evident that size parameters referring to variables length and equivalent width may be large enough to cause rejection predictions over 100%, which can be explained after observing positive coefficients for equivalent width and length. This situation may also be detrimental for rejection predictions of ionic compounds of medium to large size (0.6–1.2 nm as equivalent width) that mostly are rejected due to electrostatic repulsion and less steric hindrance. Therefore Eqs. (2) and (3) can be transformed to the following conditional equation rejection ¼
100 if QSAR model 100 QSAR model
(4)
An external dataset (that gathered three different datasets) was selected for external validation of the QSAR model. The external dataset is presented as supplementary data. The first part of the external dataset corresponds to membrane Filmtec NF-90. The second part corresponds to NF membrane Trisep TS-80 and the third part corresponds to NF membrane Desal HL. Desal HL membrane has a main difference with NF-90 and Trisep TS-80 membranes, viz. the MWCO of Desal HL is in the range of 150–300 Da, while NF-90 and Trisep TS-80 were reported to have a MWCO of 200 Da. Therefore an average MWCO of 225 Da was assumed for Desal HL during the application of Eq. (3). It is worthwhile to mention that the
second and third parts of the external dataset were generated using spiral wound membrane elements instead of flat sheet membranes. Fig. 2a show plotted results of measured rejections vs. predicted rejections after calculations with Eq. (4), for QSAR model with Eq. (2) (SR). According to Fig. 2a an R2 of 0.75 was obtained after considering all compounds of the external dataset. However, after observing the rejection cases by NF-90 for bromoform (BF) and trichloroethene (TCE), it appeared that BF and TCE may be considered atypical results because their rejections did not correspond to their size and hydrophobicity when compared to other compounds with comparable molecular descriptors. According to Table 3 BF has approximately the same hydrophobicity and polarity as CF, but BF is bigger than CF; therefore, measured rejection for BF was expected to be higher than rejection of CF (0%) due to size exclusion. The measured rejection of TCE (3%) was not comparable to the rejection of perchloroethene (39%) although they have the same length but a very small difference in equivalent width. In an experiment conducted by Kim et al. (2007) rejections of BF and TCE were of 50 and 33%, respectively, for a low pressure reverse osmosis membrane. It was also observed that the measured rejection (53%) of linuron (LNU) may be considered as atypical observation because a higher rejection was expected. According to Table 3, monolinuron is close to LNU in size and polarity; thus measured rejection of monolinuron was of 77%, higher than 53%. Also according to Table 3, carbamazepine is close to LNU in size and hydrophobicity; thus measured rejection of carbamazepine was of 94%, higher than 53%. A similar explanation was given for the low rejection (70%) of N-acetylL-tyrosine (NAT) by Desal HL compared to 94% rejection by TS-80 as can be seen in Table 3. Other non-expected rejection was that of 2-(1H)-quinoline (QNL), with a rejection of 22% when compared to 2-methoxyethanol and perchloroethene with rejections of 32 and 39%, respectively; even though the length of QNL is greater than the length of 2-methoxyethanol and perchloroethene, a lower rejection of QNL was observed. However, rejections of methacetin (MTC) and NDPA may be influenced by their small equivalent width as can be seen in
381
water research 44 (2010) 373–384
b 100
100
80
80
measured rejection (%)
measured rejection (%)
a
NAT
MTBE
60 LNU NDPA
40 ETH
MTC QNL
20
60
40
20
NDPA
R2 = 0.84
2
BF
0
0
20
MTC
TCE
R = 0.75 0
40
60
80
100
0
20
predicted rejection (%)
NF-90
TS-80
HL
40
60
80
100
predicted rejection (%)
Linear ( )
NF-90
TS-80
HL
Linear ( )
Fig. 2 – Predicted rejections for external dataset using magnesium sulphate SR: (a) all external data; and (b) selected external data.
Table 3. The response of the model for rejection of those particular compounds (with 2*equivalent width < or w length) was of over prediction, but the model presented better response for rejections of metribuzin, atrazin and N-acetyl-Ltyrosine (all with length >2*equivalent width) as observed in Table 3. Ethanol (ETH) and MTBE with rejections of 38 and 60%, respectively, were expected to be lower for Desal HL membrane because it was observed that rejection of ETH was of 9% for TS-80; and MTBE was expected to have a rejection compared to that of 2-methoxyethanol (32%) or perchloroethene (39%) due to proximities in size. After selection
and justified separation of the mentioned rejection cases, Fig. 2b presents the predictions of the external dataset with an R2 of 0.84. In a similar explanatory scenario, Figs. 3a,b show measured rejections vs. predicted rejections after calculations with Eq. (4), for QSAR model with equation 3 (MWCO). The main difference between Figs. 2b and 3b was that the model with MWCO (Fig. 3b) showed a lower R2 (0.80) than the model with SR (R2 ¼ 0.84), meaning that the latter had a better goodness of fit for external prediction response. Moreover, the characterization of magnesium sulphate salt rejection for a membrane may be preferred instead of MWCO, particularly
Table 3 – Partial list of rejections and compound properties of external dataset. Compound Chloroform Ethanol Ethanol Carbontetrachloride Bromoform MTBE Trichloroethene Perchloroethene 2-methoxyethanol 2-(1H)-quinoline NDPA NDPA Metribuzin Carbamazepine Linuron Monolinuron Atrazin Methacetin Methacetin N-acetyl-L-tyrosine N-acetyl-L-tyrosine
Abb.
ETH BF MTBE TCE
QNL NDPA NDPA
LNU
MTC MTC NAT NAT
Length
Eqwidth
Depth
Log D
Dipole
Measured rejection
Predicted rejection
0.53 0.64 0.64 0.64 0.69 0.77 0.78 0.78 0.87 1.00 1.16 1.16 1.17 1.20 1.21 1.22 1.26 1.28 1.28 1.33 1.33
0.42 0.52 0.52 0.6 0.56 0.63 0.49 0.59 0.52 0.52 0.60 0.60 0.74 0.73 0.69 0.69 0.74 0.52 0.52 0.71 0.71
0.35 0.51 0.51 0.57 0.48 0.59 0.36 0.45 0.51 0.36 0.53 0.53 0.64 0.58 0.53 0.65 0.55 0.42 0.42 0.60 0.60
1.97 0.31 0.31 2.83 2.40 0.94 2.29 3.40 0.77 1.26 1.36 1.36 0.47 2.45 3.20 2.30 2.61 1.03 1.03 2.18 2.18
1.12 1.55 1.55 0.30 1.00 1.37 0.95 0.11 0.25 3.38 3.40 3.40 0.52 3.66 2.11 2.02 3.43 2.20 2.20 3.45 3.45
0 9 38 35 0 60 3 39 32 22 45 19 97 88 53 79 91 38 5 94 70
0 14 0 26 33 25 36 46 34 53 69 55 99 94 84 77 100 72 58 100 100
Membrane NF-90 TS-80 Desal HL NF-90 NF-90 Desal HL NF-90 NF-90 TS-80 TS-80 TS-80 Desal HL TS-80 TS-80 TS-80 TS-80 TS-80 TS-80 Desal HL TS-80 Desal HL
382
water research 44 (2010) 373–384
a
b 100
80 NAT
MTBE
60 LNU NDPA ETH
40
MTC QNL NDPA
20
measured rejection (%)
measured rejection (%)
100
80
60
40
20
R2 = 0.80
2
MTC BF
0 0
R = 0.74
TCE
20
0 40
60
80
100
0
20
predicted rejection (%)
NF-90
TS-80
HL
40
60
80
100
predicted rejection (%)
Linear ( )
NF-90
TS-80
HL
Linear ( )
Fig. 3 – Predicted rejections for external dataset using MWCO: (a) all external data; and (b) selected external data.
for nanofiltration and low pressure reverse osmosis membranes; besides, the effect of fouling in membranes can also be quantified by salt rejection experiments. We can state that the QSAR model with SR demonstrated to be acceptable for the external dataset of NF-90, Trisep TS-80 and Desal HL with an R2 of 0.75 and 0.84 for the total external dataset and justified selected external dataset, respectively. Although the model can be valid with limitations related to boundary experimental conditions mentioned in Section 3, its applicability and approach can be of value for the construction of a model with combined datasets organized in training and testing groups.
5.
Acknowledgements The authors acknowledge Delft Cluster and EU Techneau Project for funding this project. The authors also acknowledge Tae-Uk Kim and Arne Verliefde for providing data and details of their research. The authors thank Filmtec (Dow Chemical Co.) for donating the membranes, Dr. Jaeweon Cho of GIST (Korea), Dr. Frank Sacher of TZW (Germany) and Steven Mookhoek of TU Delft (Netherlands) for contributing with analytical results and facilities.
Conclusions – A general QSAR model equation was developed to merge information about interaction of membrane characteristics, filtration operating conditions and solute properties to predict rejections of emerging contaminants during nanofiltration. – The QSAR model identified that the most important variables that influence rejection of organic solutes were log D, salt rejection, equivalent width, depth and length. – Rejection increased by size/steric hindrance effects, solute hydrophobicity decreased rejection due to adsorption and partitioning mechanisms. – Salt rejection incorporated steric hindrance and electrostatic repulsion effects that were related to the membrane structure and operating conditions. – The use of MWCO was acceptable for modelling purposes; however NF membranes with a broad range of MWCO (pore size and distribution) may difficult estimation of rejection of contaminants, thus magnesium sulphate salt rejection may be more appropriate.
Appendix. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.watres.2009.06.054
references
Abdi, H., 2003. Partial least squares (PLS) regression. In: LewisBeck, M., Bryman, A., Futing, T. (Eds.), Encyclopaedia of Social Sciences Research Methods. Sage, Thousand Oaks, CA. Bellona, C., Drewes, J.E., 2005. The role of membrane surface charge and solute physico-chemical properties in the rejection of organic acids by NF. Journal of Membrane Science 249 (1–2), 227–234. Campbell, P., Srinivasan, R., Knoell, T., Phipps, D., Ishida, K., Safarik, J., Cormack, T., Ridgway, H., 1999. Quantitative structure-activity relationship (QSAR) analysis of surfactants influencing attachment of a Mycobacterium sp. to cellulose
water research 44 (2010) 373–384
acetate and aromatic polyamide reverse osmosis membranes. Biotechnology and Bioengineering 64 (5), 527–544. Castiglioni, S., Bagnati, R., Fanelli, R., Pomati, F., Calamari, D., Zuccato, E., 2006. Removal of pharmaceuticals in sewage treatment plants in Italy. Environmental Science & Technology 40 (1), 357–363. Connell, D.W., 1990. Bioaccumulation of Xenobiotic Compounds. CRC Press, Boca Raton, FL, pp. 75-110. Cornelissen, E.R., Verdouw, J., Gijsbertsen-Abrahamse, A.J., Hofman, J.A.M.H., 2005. A nanofiltration retention model for trace contaminants in drinking water sources. Desalination 178 (1–3), 179–192. Drewes, J.E., Amy, G., Kim, T.-U., Xu, P., Bellona, C., Oedekoven, M., Macalady, D., 2006. Rejection of WastewaterDerived Micropollutants in High-Pressure Membrane Applications Leading to Indirect Potable Reuse Effect of Membranes and Micropollutant Properties. WateReuse, Alexandria, VA. Eriksson, L., Jaworska, J., Worth, A.P., Cronin, M.T.D., McDowell, R.M., Gramatica, P., 2003. Methods for reliability and uncertainty assessment and for applicability evaluations of classification and regression-based QSARs. Environmental Health Perspectives 111 (10), 1361–1375. Escher, B.I., Bramaz, N., Eggen, R.I.L., Richter, M., 2005. In vitro assessment of modes of toxic action of pharmaceuticals in aquatic life. Environmental Science & Technology 39 (9), 3090–3100. Everitt, B.S., Dunn, G., 2001. Applied Multivariate Data Analysis. Arnold, London. Fujikawa, M., Nakao, K., Shimizub, R., Akamatsu, M., 2007. QSAR study on permeability of hydrophobic compounds with artificial membranes. Bioorganic & Medicinal Chemistry 15 (11), 3753–3767. Gramatica, P., 2007. Principles of QSAR models validation: internal and external. QSAR & Combinatorial Science 26 (5), 694–701. Heberer, T., 2002. Occurrence, fate, and removal of pharmaceutical residues in the aquatic environment: a review of recent research data. Toxicology Letters 131 (1–2), 5–17. Henderson, R.K., Baker, A., Parsons, S.A., Jefferson, B., 2008. Characterisation of algogenic organic matter extracted from cyanobacteria, green algae and diatoms. Water Research 42 (13), 3435–3445. Hoek, E.M.V., Elimelech, M., 2003. Cake-enhanced concentration polarization: a new fouling mechanism for salt-rejecting membranes. Environmental Science & Technology 37 (24), 5581–5588. Johnson, R.A., Wichern, D.W., 2007. Applied Multivariate Statistical Analysis. Pearson Prentice Hall, Upper Saddle River, NJ. Jolliffe, I.T., 2002. Principal Component Analysis. Springer-Verlag, New York. Jørgensen, B., Goegebeur, Y., 2006. Multivariate Data Analysis and Chemometrics. Department of Statistics, University of Southern Denmark. Kim, T.-U., Drewes, J.E., Summer, R.S., Amy, G., 2007. Solute transport model for trace organic neutral and charged compounds through nanofiltration and reverse osmosis membranes. Water Research 41 (17), 3977–3988. Kimura, K., Amy, G., Drewes, J., Heberer, T., Kim, T., Watanabe, Y., 2003. Rejection of organic micropollutants (disinfection byproducts, endocrine disrupting compounds, and pharmaceutically active compounds) by NF/RO membranes. Journal of Membrane Science 227 (1–2), 113–121. Kimura, K., Toshima, S., Amy, G., Watanabe, Y., 2004. Rejection of neutral endocrine disrupting compounds (EDCs) and pharmaceutical active compounds (PhACs) by RO membranes. Journal of Membrane Science 245 (1–2), 71–78.
383
Kiso, Y., Sugiura, T., Kitao, T., Nishimura, K., 2001. Effects of hydrophobicity and molecular size on rejection of aromatic pesticides with nanofiltration membranes. Journal of Membrane Science 192 (1–2), 1–10. 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, 1999–2000: a National Reconnaissance. Environmental Science & Technology 36 (6), 1202–1211. Landau, S., Everitt, B., 2004. A Handbook of Statistical Analysis using SPSS. Chapman & Hall/CRC, Boca Raton, FL. Lee, N., Amy, G., Croue, J.-P., Buisson, H., 2004. Identification and understanding of fouling in low-pressure membrane (MF/UF) filtration by natural organic matter (NOM). Water Research 38 (20), 4511–4523. Libotean, D., Giralt, J., Rallo, R., Cohen, Y., Giralt, F., Ridgway, H., Rodriguez, G., Phipps, D., 2008. Organic compounds passage through RO membranes. Journal of Membrane Science 313 (1-2), 23–43. Nghiem, L.D., Scha¨fer, A.I., Elimelech, M., 2004. Removal of natural hormones by nanofiltration membranes: measurement, modelling, and mechanisms. Environmental Science & Technology 38 (6), 1888–1896. Nghiem, L.D., Scha¨fer, A.I., Elimelech, M., 2005. Pharmaceutical retention mechanisms by nanofiltration membranes. Environmental Science & Technology 39 (19), 7698–7705. Ozaki, H., Li, H., 2002. Rejection of organic compounds by ultralow pressure reverse osmosis membrane. Water Research 36 (1), 123–130. Pomati, F., Castiglioni, S., Zuccato, E., Fanelli, R., Vigetti, D., Rossetti, C., Calamari, C., 2006. Effects of a complex mixture of therapeutic drugs at environmental levels on human embryonic cells. Environmental Science & Technology 40 (7), 2442–2447. Ren, S., Das, A., Lien, E.J., 1996. QSAR analysis of membrane permeability to organic compounds. Journal of Drug Targeting 4 (2), 103–107. 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 (1–2), 199–210. Sacher, F., Ehmann, M., Gabriel, S., Graf, C., Brauch, H.-J., 2008. Pharmaceutical residues in the river Rhinedresults of a onedecade monitoring programme. Journal of Environmental Monitoring 10 (5), 664–670. Santos, J.L.C., de Beukelaar, P., Vankelecom, I.F.J., Velizarov, S., Crespo, J.G., 2006. Effect of solute geometry and orientation on the rejection of uncharged compounds by nanofiltration. Separation and Purification Technology 50 (1), 122–131. Sawyer, C.N., McCarty, P.L., Parkin, G.F., 2003. Chemistry for Environmental Engineering and Science. McGraw-Hill, New York, pp. 479-520. Scha¨fer, A.I., Nghiem, L.D., Waite, T.D., 2003. Removal of the natural hormone estrone from aqueous solutions using nanofiltration and reverse osmosis. Environmental Science & Technology 37 (1), 182–188. Shim, Y., Lee, H.-J., Lee, S., Moon, S.-H., Cho, J., 2002. Effects of natural organic matter and ionic species on membrane surface charge. Environmental Science & Technology 36 (17), 3864–3871. Van der Bruggen, B., Vandecasteele, C., 2002. Modeling of the retention of uncharged molecules with nanofiltration. Water Research 36 (5), 1360–1368. Van der Bruggen, B., Schaep, J., Wilms, D., Vandecasteele, C., 1999. Influence of molecular size, polarity and charge on the retention of organic molecules by nanofiltration. Journal of Membrane Science 156 (1), 29–41. Verliefde, A., Cornelissen, E., Amy, G., Van der Bruggen, B., van Dijk, H., 2007. Priority organic micropollutants in water
384
water research 44 (2010) 373–384
sources in Flanders and the Netherlands and assessment of removal possibilities with nanofiltration. Environmental Pollution 146 (1), 281–289. Verliefde, A.R.D., Cornelissen, E.R., Heijman, S.G.J., Verberk, J.Q.J.C., Amy, G.L., Van der Bruggen, B., van Dijk, J.C., 2008. The role of electrostatic interactions on the rejection of organic solutes in aqueous solutions with nanofiltration. Journal of Membrane Science 322 (1), 52–66. Vosges, M., Braguer, J.-C., Combarnous, Y., 2008. Long-term exposure of male rats to low-dose ethinylestradiol (EE2) in drinking water: effects on ponderal growth and on litter size of their progeny. Reproductive Toxicology 25 (2), 161–168.
Xu, P., Drewes, J.E., Bellona, C., Amy, G., Kim, T.-U., Adam, M., Heberer, T., 2005. Rejection of emerging organic micropollutants in nanofiltration-reverse osmosis membrane applications. Water Environment Research 77 (1), 40–48. Yangali-Quintanilla, V., Kim, T.-U., Kennedy, M., Amy, G., 2008. Modeling of RO/NF membrane rejections of PhACs and organic compounds: a statistical analysis. Drinking Water Engineering and Science 1 (1), 7–15. Yangali-Quintanilla, V., Sadmani, A., McConville, M., Kennedy, M., Amy, G., 2009. Rejection of pharmaceutically active compounds and endocrine disrupting compounds by clean and fouled nanofiltration membranes. Water Research 43 (9), 2349–2362.
water research 44 (2010) 385–416
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Harmful algae and their potential impacts on desalination operations off southern California David A. Carona,*, Marie-E`ve Garneaua, Erica Seuberta, Meredith D.A. Howarda,b, Lindsay Darjanya, Astrid Schnetzera, Ivona Cetinic´a, Gerry Filteauc, Phil Laurid, Burton Jonesa, Shane Trusselle a
Department of Biological Sciences, University of Southern California, 3616 Trousdale Parkway, Los Angeles, CA 90089-0371, USA Southern California Coastal Water Research Project, 3535 Harbor Blvd., Suite 110, Costa Mesa, CA 92626, USA c Separation Processes, Inc., 3156 Lionshead Avenue, Suite 2, Carlsbad, CA 92010, USA d West Basin Municipal Water District, 17140 Avalon Blvd., Suite 210, Carson, CA 90746, USA e Trussell Technologies, Inc., 6540 Lusk Boulevard, Suite C175, San Diego, CA 92121, USA b
article info
abstract
Article history:
Seawater desalination by reverse osmosis (RO) is a reliable method for augmenting
Received 2 April 2009
drinking water supplies. In recent years, the number and size of these water projects have
Received in revised form
increased dramatically. As freshwater resources become limited due to global climate
12 June 2009
change, rising demand, and exhausted local water supplies, seawater desalination will play
Accepted 23 June 2009
an important role in the world’s future water supply, reaching far beyond its deep roots in
Available online 30 June 2009
the Middle East. Emerging contaminants have been widely discussed with respect to wastewater and freshwater sources, but also must be considered for seawater desalination
Keywords:
facilities to ensure the long-term safety and suitability of this emerging water supply.
Harmful algal blooms
Harmful algal blooms, frequently referred to as ‘red tides’ due to their vibrant colors, are
Desalination
a concern for desalination plants due to the high biomass of microalgae present in ocean
Red tides
waters during these events, and a variety of substances that some of these algae produce.
Phytoplankton
These compounds range from noxious substances to powerful neurotoxins that constitute
Phytotoxins
significant public health risks if they are not effectively and completely removed by the RO membranes. Algal blooms can cause significant operational issues that result in increased chemical consumption, increased membrane fouling rates, and in extreme cases, a plant to be taken off-line. Early algal bloom detection by desalination facilities is essential so that operational adjustments can be made to ensure that production capacity remains unaffected. This review identifies the toxic substances, their known producers, and our present state of knowledge regarding the causes of toxic episodes, with a special focus on the Southern California Bight. ª 2009 Elsevier Ltd. All rights reserved.
* Corresponding author. Tel.: þ1 213 740 0203; fax: þ1 213 740 8123. E-mail address:
[email protected] (D.A. Caron). 0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2009.06.051
386
1.
water research 44 (2010) 385–416
Introduction
1.1. General overview of harmful algal blooms: a growing global concern Microscopic algae constitute an essential component of all aquatic food webs. Photosynthetic production of organic material by this diverse group of species comprises the primary source of nutrition for all heterotrophic forms of life in much of the world’s ocean and freshwater ecosystems. Microalgae can reach high abundances in the plankton during periods of optimal growth and reduced grazing pressure by herbivores. Such localized mass proliferations are known as algal (or phytoplankton) blooms. In addition, a small proportion of microalgal species are capable of producing a number of noxious or toxic compounds that cause a variety of adverse effects on ecosystem structure and function. These substances pose the potential for ecosystem damage, food web disruption and marine animal mortality, and present a significant human health risk through the consumption of contaminated seafood and, in at least one case, direct exposure to water or aerosols containing these toxic compounds. Additionally, the algal biomass and the associated organic load cause significant desalination operational issues, impacting the pretreatment system and possibly forcing the treatment plant to be taken off-line (Petry et al., 2007). Countless human deaths resulting from the consumption of seafood contaminated with algal toxins have been avoided through rigorous monitoring programs, but sea life has not been so fortunate. Approximately one half of all unusual marine mammal mortality incidents are now attributable to the ingestion of food or prey contaminated by harmful algal blooms (Ramsdell et al., 2005). Losses in revenue due to the direct contamination of seafood products and indirect effects on tourism and other uses of coastal areas have been estimated in the tens of millions of dollars annually in the U.S. states along the Pacific coast (Trainer et al., 2002). There is now convincing evidence that harmful algal bloom (HAB) events are increasing at local, regional and global scales worldwide (Smayda, 1990; Hallegraeff, 1993, 2003; Anderson et al., 2002; Glibert et al., 2005a) and along the North American west coast in particular (Horner et al., 1997; Trainer et al., 2003). This increased occurrence may be due in part to better detection of HAB episodes in recent years or the global dispersal of toxic algal species via the transport of resting spores in ships’ ballast waters (Hallegraeff and Bolch, 1992; Burkholder et al., 2007), but another very likely cause is the increasing impact of anthropogenic activities on coastal ecosystems (Smayda, 1990; Anderson et al., 2002; Glibert et al., 2005b, 2006; Howard et al., 2007; Cochlan et al., 2008; Kudela et al., 2008a). Recent reports reveal extensive and, in some cases, newly emerging occurrences of HABs along the coasts of the U.S. (Fig. 1). These incidents engender a variety of noxious impacts on ecosystems and public health, including direct effects on organisms due to the production of acutely toxic substances, and indirect effects such as reduced availability of dissolved oxygen in the water column resulting from the decomposition of the extensive amounts of organic substances usually produced during such blooms. The
Fig. 1 – Distribution of some well-known regional HAB issues along U.S. shores, including (a) Alaska and (b) Hawaii. Causes and impacts of these poisoning events are defined in Tables 1–3. Summarized from information presented on the Harmful Algae webpage (http://www. whoi.edu/redtide/).
dramatic increases in biomass and organic load that accompany these events pose a significant threat to seawater desalination facilities (Gaid and Treal, 2007).
1.2.
Regional HAB issues along U.S. coastlines
Harmful algae are present throughout U.S. coastal waters, but not all species are of equal concern in all regions (Fig. 1). For example, toxic species of the dinoflagellate genus Alexandrium are common over vast stretches of the U.S. coastline, but coastal regions of the northeastern and northwestern U.S. appear to experience particularly high rates of occurrence of toxic ‘red tides’ caused by these species. The neurotoxins produced by Alexandrium, called saxitoxins, cause paralytic shellfish poisoning (PSP) in humans when ingested through contaminated seafood (particularly filter-feeding shellfish). Similarly, several toxic species of the diatom genus Pseudonitzschia occur along the entire U.S. coastline but significant concentrations of the neurotoxin, domoic acid, produced by these species have historically constituted a health threat primarily in the northeastern and northwestern U.S. (Bates et al., 1989) where it has been documented as the cause of amnesic shellfish poisoning (ASP) in humans. However, high concentrations of domoic acid in the plankton and in diverse planktivorous organisms have been recently documented along the entire Pacific coast of the U.S. (Scholin et al., 2000; Trainer et al., 2002; Schnetzer et al., 2007), as well as in the Gulf of Mexico (Pan et al., 1998). Domoic acid has been attributed to numerous marine animal mortalities along the U.S. west coast. In the Gulf of Mexico, primarily along the west coast of Florida, extensive and recurrent blooms of the dinoflagellate Karenia brevis produce a suite of toxins, known as brevetoxins, that can be aerosolized by breaking waves and
water research 44 (2010) 385–416
induce neurotoxic shellfish poisoning (NSP) in people inhaling the aerosols (see review of Kirkpatrick et al., 2004). The Tampa Bay seawater desalination facility is the only operating seawater desalination treatment plant of significant size in the United States. It is located along the west coast of Florida and is likely to encounter algal blooms that contain brevetoxin. Less toxic blooms also take place with regional specificity. The pelagophyte Aureococcus anophagefferens causes ‘brown tides’ in coastal waters of Rhode Island, near Long Island (NY) and southward along the mid-Atlantic coast of the U.S. since 1985. No specific toxins have been identified from A. anophagefferens, and no human fatalities have been directly attributed to these blooms. Nevertheless, this species appears to be unpalatable or inhibitory to many filter-feeding mollusks and has caused substantial mortality among these populations, including commercially valuable species (Bricelj and Lonsdale, 1997). Other microalgal species can disrupt food webs or cause reductions in water quality without producing acutely toxic conditions. Among these are the ‘colorful’ red tides of the dinoflagellate Lingulodinium polyedrum, a yessotoxin producer, that have occurred periodically throughout several decades along the south and central Californian coasts (Horner et al., 1997; Gregorio and Pieper, 2000). These blooms have so far been found to be relatively innocuous in these waters but massive accumulations of these cells could have significant impact on desalination plants because of increased turbidity, high suspended solids and organic loading of influent water. Furthermore, accumulations of cells in protected harbors can cause fish mortality by depleting oxygen dissolved in the water, further challenging influent screening and pretreatment systems at desalination plants. Other taxa, such as species of the prymnesiophyte genus Phaeocystis, produce substances that can lead to enormous buildups of sea foam along coasts (Armonies, 1989).
1.3.
Desalination, plankton and water quality issues
Large research programs have developed within different geographic areas throughout the U.S. to address regional HAB issues. These programs are designed to study the species, toxins and environmental causes of HAB outbreaks. These efforts, as well as local, county, state and federal monitoring programs provide basic information for marine resource use and have focused almost exclusively on threats to human health via the consumption of contaminated seafood. Unfortunately, few if any of these programs provide sufficient information on appropriate temporal and spatial resolution for thoroughly assessing the potential impact of HAB events on reverse osmosis desalination operations. Moreover, toxin analyses have primarily examined the presence of these substances in particulate material (plankton or animal tissue, particularly shellfish and finfish), and therefore may be poor predictors for the amount of toxins that might occur in seawater in the dissolved state during algal blooms, which would be most likely to be loaded onto reverse osmosis membranes during desalination. There are two potential impacts that HABs may have on seawater desalination facilities: (1) algal toxins in ocean water
387
pose a significant treatment challenge for the reverse osmosis system to ensure that these molecules are effectively removed and (2) increased turbidity, total suspended solids and total organic content resulting from algal biomass and growth challenge the entire desalination facility’s treatment train. The significance of these issues will depend on the specific algae forming a bloom and the toxin(s) or other substances that they produce, the magnitude and duration of the bloom, and the specific desalination process conducted. For example, multi-effect distillation and multi-flash distillation might be susceptible to (2) but would be much less affected by toxins in the water (1). Desalination using reverse osmosis presumably would be vulnerable to both issues. Therefore, for the latter desalination approach, a thorough understanding of HAB episodes in terms of incidence and seasonality, vertical and horizontal spatial distribution, as well as biological aspects such as algal composition within a geographical region could help optimize the design and operational efficiency of desalination plants employing reverse osmosis. This paper provides an overview of HABs occurring along the continental U.S. coastline with special emphasis on the southwestern U.S., and provides some insight on the potential impacts that these events may have on the seawater desalination process. In recent years, this geographical area has become a focal point of discussions regarding desalination (Cooley et al., 2006) because of its sizable population and the particularly tenuous nature of the water supply to this region. Although numerous issues involving the desalination process are now being examined (Separation Processes Inc., 2005; Gaid and Treal, 2007; Pankratz, 2008, 2009), very limited information exists on the risks that algal blooms pose to seawater desalination facilities. A review of the major species producing harmful blooms, the substances they produce, and information on the spatial and temporal distributions of blooms are presented along with some conclusions on their potential impacts. This paper also provides some general guidelines on how early detection may help prevent or minimize the impact of HABs on a desalination facility’s production capacity or its water quality.
2. Toxin producers and toxin concentrations of the west coast A variety of toxins including several powerful neurotoxins are produced by microalgae, and a number of these toxins and potentially toxic algal species have been detected on the U.S. west coast (Table 1). The ability to rapidly detect and quantify toxic algae in natural water samples is problematic at this time. Many of these species are difficult to identify using light microscopy. For this reason, new genetic and immunological methods for species identification and enumeration have been appearing rapidly in the literature (Miller and Scholin, 1998; Bowers et al., 2000, 2006; Coyne et al., 2001; Caron et al., 2003; Galluzzi et al., 2004; Anderson et al., 2005; Mikulski et al., 2005, 2008; Ahn et al., 2006; Handy et al., 2006; Moorthi et al., 2006; Iwataki et al., 2007, 2008; Demir et al., 2008; Matsuoka et al., 2008). Moreover, many toxin-producing algal species exhibit variable toxin production in response to environmental conditions, and among different strains of the same species
388
water research 44 (2010) 385–416
Table 1 – Planktonic species occurring along the west coast of the U.S. that are potential concerns for reverse osmosis operations. Microalgae
Toxin(s)
Poisoning Event
References
Amnesic Shellfish Poisoning (ASP)
Subba Rao et al. (1988), Bates et al. (1989), Martin et al. (1990), Buck et al. (1992), Garrison et al. (1992), Rhodes et al. (1996), Horner et al. (1997), Lundholm et al. (1997), Rhodes et al. (1998), Trainer et al. (2000, 2001), Baugh et al. (2006)
Diatoms Pseudo-nitzschia spp. P. australisb P. cuspidatab P. delicatissimab P. fraudulentab P. multiseriesb P. pungensb P. pseudodelicatissimab P. seriataa
Domoic acid (DA)
Dinoflagellates Alexandrium spp. A. acatenellaa A. catenellab A. fundyensea A. hiranoia A. ostenfeldiia A. tamarensea
Saxitoxins (STXs)
Dinoflagellates Lingulodinium polyedrumb Gonyaulax spiniferaa Protoceratium reticulatuma,c
Yessotoxins (YTXs)
Human and ecosystem effects None reported
Holmes et al. (1967), Satake et al. (1997, 1999), Draisci et al. (1999a), Paz et al. (2004, 2007), Armstrong and Kudela (2006), Rhodes et al. (2006), Howard et al. (2007)
Dinoflagellates Dinophysis spp. D. acuminataa D. acutaa D. caudate D. fortiia D. norvegicaa D. rotundataa D. triposa
Okadaic acid (OA) Dinophysistoxins (DTXs) Pectenotoxins (PTXs)
Diarrhetic Shellfish Poisoning (DSP)
Holmes et al. (1967), Yasumoto et al. (1980), Murata et al. (1982), Yasumoto et al., (1985), Cembella (1989), Lee et al. (1989), Horner et al. (1997), Cembella (2003), Miles et al. (2004), Shipe et al. (2008), Sutherland (2008)
Human effects Gastro-intestinal symptoms Neurologic symptoms Death Ecosystem effects Marine mammal mortalities Bird mortalities
Paralytic Shellfish Poisoning (PSP) Human effects Gastro-intestinal symptoms Paralysis Death Ecosystem effects Marine mammal mortalities
Human effects Gastro-intestinal symptoms Ecosystem effects None reported
Sommer and Meyer (1937), Gaines and Taylor (1985), Steidinger (1993), Scholin et al. (1994), Taylor and Horner (1994), Jester (2008)
Prorocentrum spp. P. micans P. minimuma,d Raphidophytes Chattonella marinaa Fibrocapsa japonicaa Heterosigma akashiwoa
a b c d
Brevetoxins (PbTxs)
Neurotoxic Shellfish Poisonining (NSP) Human effects Gastroenteritis Neurologic symptoms Respiratory irritation and/or failure Ecosystem effects Marine mammal mortalities Fish mortality events
Loeblich and Fine (1977), Hershberger et al. (1997), Gregorio and Connell (2000), Hard et al. (2000), Tyrell et al. (2002), O’Halloran et al. (2006)
Reported to produce toxin. Reported to produce toxin on the west coast of the United States. Conflicting reports on toxicity of P. reticulatum cultures isolated from California, Washington and Florida. Reported to be present on the west coast of Mexico.
even when isolated from the same geographic region (Smith et al., 2001; Trainer et al., 2001; Kudela et al., 2004). Laboratory experiments have revealed a wide range of physico-chemical factors that increase or decrease toxin
production by harmful species of algae, and which appear to be species-specific (see review of Grane´li and Flynn, 2006). Reports of factors inducing toxin production have sometimes been conflicting, presumably indicating that multiple factors, or
water research 44 (2010) 385–416
perhaps generally stressful conditions, may stimulate toxin production. Factors affecting toxin production include: (1) temperature (Ono et al., 2000); (2) light intensity (Ono et al., 2000); (3) salinity (Haque and Onoue, 2002a,b); (4) trace metal availability, especially iron (Ladizinsky and Smith, 2000; Rue and Bruland, 2001; Maldonado et al., 2002; Wells et al., 2005; Sunda, 2006) but also copper (Maldonado et al., 2002) and selenium (Mitrovic et al., 2004, 2005); (5) macronutrient availability including silicate (Pan et al., 1996b; Fehling et al., 2004; Kudela et al., 2004), phosphate (Pan et al., 1996a, 1998; Fehling et al., 2004), nitrogen (Bates et al., 1991; Pan et al., 1998; Kudela et al., 2004) and combinations of nutrient limitation (Anderson et al., 1990; Flynn et al., 1994; John and Flynn, 2000); (6) cellular elemental ratios of nutrients and physiological stress (Grane´li and Flynn, 2006; Schnetzer et al., 2007); (7) growth phase (Anderson et al., 1990; Bates et al., 1991; Flynn et al., 1994; Johansson et al., 1996; Maldonado et al., 2002; Mitrovic et al., 2004). The precise combination(s) of environmental factors that select for population growth of particular algal species within diverse natural assemblages, and the specific conditions that induce toxin production, are poorly understood for most harmful algae. This present state of knowledge makes it difficult to predict the timing, duration or spatial extent of the vast majority of HAB events and the toxic events resulting from them. Our ability to thoroughly characterize HABs is also complicated by the complex array of toxins produced by algae. Marine algal species produce a suite of toxic components (Yasumoto and Murata, 1993), and unidentified toxins undoubtedly remain to be described. Additionally, most toxins are actually composed of families of closely related compounds. Slightly different forms of a toxin can exhibit very different levels of toxicity, or may be characterized differently by some detection methods and analytical approaches (Garthwaite et al., 2001; Lefebvre et al., 2008). Such complexity and variability can sometimes yield vague or contradictory conclusions regarding the exact source of toxicity in a natural sample (Bates et al., 1978; Paz et al., 2004, 2007). Finally, characterization of HAB events is complicated by inherent difficulties associated with linking specific toxins measured in natural water samples to a specific algal species in a complex, natural phytoplankton assemblage and, as noted above, the presence of toxic species in a water sample does not necessarily indicate the presence of toxins. Despite these shortcomings, there is considerable knowledge of many of the major algal toxins and their producers in U.S. coastal waters that constitute the most important potential concerns for desalination activities because they are the most likely to be encountered in ocean water intakes.
2.1.
Domoic acid
2.1.1.
Toxin description and activity
Domoic acid (Fig. 2; Table 3) is an amino acid derivative belonging to the kainoid class of compounds containing three carboxyl groups and one secondary amino group (Wright et al., 1990; Jeffery et al., 2004). All four groups are charged at neutral pH, and the carboxyl groups become successively protonated as pH decreases, yielding five possible protonated forms of domoic acid (Quilliam, 2003; Jeffery et al., 2004). There
389
are currently ten known isomers of domoic acid, including the isodomoic acids A through H and the domoic acid 5’ diestereomer (Jeffery et al., 2004). Domoic acid and other members of the kainoid class are glutamate analogues that interfere with neurochemical pathways by binding to glutamate receptors of brain neurons (Wright et al., 1990; Quilliam, 2003). The resulting effect of these neuroexcitants, or excitotoxins, is a continuous stimulation of the neurons, which can lead to rupture and/or eventual formation of lesions (Wright et al., 1990). Depolarized neurons result in short-term memory loss (Clayden et al., 2005), which has led to the common name for the illness related to the consumption of seafood contaminated with domoic acid: amnesic shellfish poisoning (ASP). Symptoms of ASP include gastroenteritis (vomiting, diarrhea, abdominal cramps) that can be experienced in humans within 24 h after ingestion, and neurological symptoms of confusion, memory loss, disorientation, seizures, coma and/or cranial nerve palsies that are typically experienced within 48 h (Perl et al., 1990; Wright et al., 1990). The number of human illnesses resulting from domoic acid poisoning has been few (Horner et al., 1997), likely due to active monitoring of fisheries. However, cultured blue mussels (Mytilus edilus) contaminated with domoic acid poisoned 107 people and killed three during the first major documented ASP outbreak in 1987 on Prince Edward Island, Canada (Perl et al., 1990). ASP poses a serious threat to marine wildlife, and the deaths of thousands of marine mammals and sea birds have been attributed to domoic acid intoxication (Bates et al., 1989; Scholin et al., 2000; Gulland et al., 2002; Caron et al., unpublished data). The first documented poisoning episode of marine animals related to domoic acid on the U.S. west coast was attributed to Pseudo-nitzschia australis and occurred in September 1991 off central California (Table 2; Buck et al., 1992; Fritz et al., 1992). High concentrations of domoic acid were also detected in Washington and Oregon in the 1990s (Wekell et al., 1994; Adams et al., 2000; Trainer et al., 2002), and a decade later in coastal waters off southern California (Schnetzer et al., 2007). The frequency and severity of these toxic events appears to be increasing (Trainer et al., 2007).
2.1.2.
Producers
The production of domoic acid and its isomers is confined to approximately a dozen chain-forming marine pennate diatom species within the genus Pseudo-nitzschia (Bates and Trainer, 2006), a genus containing species that form long chains of cells attached at their ends (Fig. 3a and b). The main toxin producing species that have been documented on the U.S. west coast include: P. australis, P. delicatissima, P. fraudulenta, P. multiseries, P. pungens, P. pseudodelicatissima, P. seriata and P. cuspidata (Tables 1 and 2). These species are distinguished based on fine morphological features of their silica frustules (Fig. 3a and b). These distinctions are subtle and require careful electron microscopical analysis and elaborate taxonomic training. As a consequence, historical misidentifications are not unusual and debates regarding some species descriptions are still unresolved. It is surprising that the first reports of ASP on the west coast of the U.S. were not recorded until the 1990s, even though Pseudo-nitzschia species have been recorded in surveys of
390
water research 44 (2010) 385–416
Fig. 2 – Chemical structures of commonly encountered toxins produced by microalgae in U.S. coastal waters.
phytoplankton species in the Southern California Bight since 1917 (Allen, 1922, 1924, 1928, 1936, 1940, 1941; Reid et al., 1970, 1985; Lange et al., 1994; Fryxell et al., 1997; Thomas et al., 2001). Given that these species generally comprise a significant portion of the total diatom assemblage in these waters, it can be surmised that either toxin production has increased in these west coast species, or that poisoning events prior to the 1990s have occurred but have not been attributed to these diatoms. Historical accounts of ‘unusual animal mortality events’ along the U.S. west coast tend to support the latter hypothesis. There have been increasing numbers of toxic events recorded along the U.S. west coast (Table 2), notably in Puget Sound (Trainer et al., 2003, 2007), Monterey Bay (Vigilant and Silver, 2007; R. Kudela, unpubl. data), Santa Barbara Channel (Trainer et al., 2000; Anderson et al., 2006; Mengelt, 2006), San Pedro Channel (Busse et al., 2006; Schnetzer et al., 2007), Newport Beach (Busse et al., 2006) and San Diego (Lange et al., 1994; Busse et al., 2006). Most recently, toxic blooms of Pseudonitzschia in the Long Beach-Los Angeles Harbor and San Pedro Channel have been particularly toxic, with some of the highest domoic acid concentrations recorded for the U.S. west coast (Caron et al., unpublished data). The increased incidence and severity of these toxic episodes off the western U.S. coast parallels the increase in frequency and intensity of harmful
algal blooms observed globally (Smayda, 1990; Hallegraeff, 1993, 2003; Anderson et al., 2002; Glibert et al., 2005b).
2.2.
Saxitoxins
2.2.1.
Toxin description and activity
Saxitoxin is a complex guanidine-based alkaloid that exists as more than 30 identified analogues in nature (Llewellyn, 2006). It is the most powerful marine toxin currently known and among the most dangerous poisons on Earth, except for some venoms and bacterial toxins (Schantz et al., 1957). Due to its acute toxicity, saxitoxin is currently listed as a chemical weapon in Schedule 1 of the Chemical Weapons Convention (Llewellyn, 2006). Saxitoxins display a rigid tricyclic core (Fig. 2; Table 3) and are very stable in biological and physiological solutions (Rogers and Rapoport, 1980). This nitrogenrich molecule and its chemical relatives are polar and have a positive charge at pH 7.7 (Shimizu et al., 1981). Consequently, they are soluble in water and alcohols, and insoluble in organic solvents (Schantz et al., 1957). Saxitoxins are known to disrupt the flow of ions through voltage gate sodium channels (Catterall, 1992; Cestele and Catterall, 2000). It has also been recently discovered that they have the ability to bind to calcium (Su et al., 2004) and
Table 2 – Distribution and concentrations of marine toxins in plankton of confirmed toxin producers in U.S. west coast waters. Toxin(s) Domoic acid
Location and year
Causative species
Particulate mg L1 (nmol L1)
Cellular pg cell1 b.d.–4.6
Washington coast and Juan de Fuca Eddy, WA (1997, 1998)
P. pseudodelicatissima Pseudo-nitzschia spp.
b.d.–2.7 3.6–8.7
Penn Cove, WA (1997)
P. pungens P. multiseries P. australis P. pseudodelicatissima
b.d.–0.8
Washington coast, WA (2001)
P. australis
b.d.–0.03
Pseudo-nitzschia spp.
(0.4–15)
Puget Sound, WA (2005)
P. pseudodelicatissima Pseudo-nitzschia spp.
b.d.–14
Central Oregon coast, OR (1998)
P. australis
Pt. An˜o Nuevo, San Francisco, CA (1998)
References Adams et al. (2000), Trainer et al. (2001, 2002) Trainer et al. (1998)
Marchetti et al. (2004) 4
2 10 –0.3 0.1–94.4
a
(b.d.–4.3) (1–5)
a
Baugh et al. (2006)
Trainer et al. (2007)
0.5
35
Trainer et al. (2001)
P. pungens P. multiseries
0.1–0.7
0.3–6.3
Trainer et al. (2000)
Bolinas Bay, San Francisco, CA (2003)
P. australis
0.15–9.4
Monterey Bay, CA (1991, 1998)
P. australis
b.d.–12.3 0.1–6.7
3–37 7.2–75
Monterey Bay, CA (1998)
P. pseudodelicatissima P. multiseries
0.1–0.4 0.67
0.8–1.2 6
Monterey Bay, CA (2000)
Pseudo-nitzschia spp. P. australis
Monterey Bay, CA (2002–2003)
Pseudo-nitzschia spp.
Morro Bay, CA (1998)
P. australis
1.3–7.4
37–78
Trainer et al. (2000, 2001)
San Luis Obispo, CA (2003–2005)
P. australis P. multiseries
1.5–7.6
9–38
Mengelt (2006)
Point Conception, CA (1998)
P. australis
2.2–6.3
15–22
Trainer et al. (2000)
Santa Barbara, CA (1998)
P. australis P. pungens P. pseudodelicatissima
0.5–1.2
0.1–0.9
Trainer et al. (2000)
Howard et al. (2007)
b.d.–24
24
Buck et al. (1992), Garrison et al. (1992), Walz et al. (1994), Scholin et al. (2000) Trainer et al. (2000, 2001) b.d.–8491
water research 44 (2010) 385–416
Washington coast, WA (2003)
Dissolved pg mL1 (nmol L1)
Bargu et al. (2002, 2008)
Vigilant and Silver (2007)
391
(continued on next page)
392
Table 2 (continued ) Toxin(s)
Saxitoxins
Location and year
Causative species
Particulate mg L1 (nmol L1)
Cellular pg cell1
0.03–1.7
0.14–2.1
Anderson et al. (2006)
6–12
b.d.–80
Mengelt (2006)
b.d.–117
Schnetzer et al. (2007)
P. australis
Santa Barbara (Santa Rosa Island and north San Miguel) (2004)
P. australis P. multiseries
Southern California Bight, CA (2003, 2004)
Pseudo-nitzschia spp. P. australis P. cuspidata
5.6–12.7
San Diego and Orange counties, CA (2004)
P. australis P. multiseries
b.d.–2.33
Sequim Bay, WA (2004–2007)
Alexandrium spp.
0.02–0.5
Oregon coast, OR (2004)
Alexandrium spp.
0.004–0.028
Humboldt Bay, CA (2004)
A. catenella
San Mateo County coast, CA (2004) Monterey Bay, CA (2004)
A. catenella
Monterey Bay, CA (2003–2005)
A. catenella
Morro Bay, CA (2004)
A. catenella
Yessotoxin
La Jolla, CA (1993)
Lingulodinium polyedrum
Brevetoxins
Indian Inlet, Bald Eagle Creek and Torque Canal, DE (2000)
Chattonella cf. verruculosa
b.d.: Below detection limit. a Toxin concentration from cells in culture.
Busse et al. (2006)
150–800
Jester (2008)
2.1–62.6
a
Jester (2008)
0.6–31.3
a
Jester (2008)
1.4–16.6
a
b.d.–0.962
0.008–<0.2
Lefebvre et al. (2008) Jester et al. (unpublished data)
1.6–19a
A. catenella
References
Jester et al. (2009b) Jester (2008)
0.002–0.02a 0–0.05a
Armstrong and Kudela (2006) Howard et al. (2008)
6
Bourdelais et al. (2002)
water research 44 (2010) 385–416
Santa Barbara Channel, CA (2003)
Dissolved pg mL1 (nmol L1)
water research 44 (2010) 385–416
potassium channels (Wang et al., 2003) and to be a weak inhibitor of neuronal nitric oxide synthase (reviewed in Llewellyn, 2006). These activities directly affect the nervous system, and the consumption of seafood containing saxitoxin can result in serious human illness and death, commonly referred to as paralytic shellfish poisoning (PSP). Minor symptoms of PSP, such as burning or tingling sensation of the lips and face, dizziness, headache, salivation, intense thirst and perspiration, vomiting, diarrhea and stomach cramps, can be experienced within 30 min after the consumption of contaminated seafood (Llewellyn, 2006). The consumption of a lethal dose can result in death within hours due to muscular paralysis and respiratory difficulty followed by complete respiratory arrest. PSP outbreaks result in more than 2000 illnesses worldwide each year, with a 5–10% mortality rate (Hallegraeff, 2003). PSP toxins also have adverse effects on marine wildlife that can cause mortalities among fish, marine mammal and seabird populations (Geraci et al., 1989; Montoya et al., 1996; Shumway et al., 2003). There are presently no records of unusual animal mortality events along the Californian coast that are attributable to saxitoxin poisoning (Jester et al., 2009b), but occurrence of the toxins in species consumed by humans is sufficient to warrant year-round monitoring.
2.2.2.
Producers
Saxitoxins are biosynthesized by dinoflagellates in marine ecosystems, most notably species within the genus Alexandrium (Fig. 3c), as well as Gymnodinium catenatum, Pyrodinium bahamense var. copressum, and by some cyanobacteria in freshwater ecosystems (Hallegraeff, 2003). Blooms of toxic and noxious dinoflagellates are often referred to as ‘red tides’ because of the red discoloration of water created by the accessory pigments of the cells. However, toxic levels of saxitoxins can be attained at dinoflagellate abundances that do not significantly discolor the water because of the exceptionally high potency of saxitoxins (Burkholder et al., 2006). This situation exists for Alexandrium in that it does not typically reach ‘bloom’ abundances on the U.S. west coast, and constant toxin monitoring is necessary to identify toxic conditions (Langlois, 2007; IOC HAB Programme, 2008; Jester et al., 2009a). Alexandrium species and measurable saxitoxin concentrations are common along the U.S. west coast (Table 1), although concentrations reported for this toxin typically have not been as high as noted along the U.S. east coast (Table 2). Thus, few west coast studies have contributed to our understanding the dynamics of Alexandrium abundances and saxitoxin production while ongoing research in the Gulf of Maine represents the most comprehensive regional study of this dinoflagellate (Anderson et al., 2005; McGillicuddy et al., 2005). Combined field observations, laboratory studies and modeling efforts have led to a scenario for toxic events along the northeastern coast of the U.S. that involve an interplay between river runoff, resuspension of dinoflagellate cysts from coastal sediments, favorable offshore growth conditions, and winds that generate onshore flow into coastal shellfish areas. A monitoring study of PSP in Puget Sound (WA) from 1993 to 2007 underscores that the timing and location of PSP outbreaks and high Alexandrium abundances are highly variable and not easily predicted from local or large-scale climate data (Moore et al., 2009). However, the study points out that
393
periods of warm air and low stream flow may favor saxitoxin accumulation in sentinel mussels (Moore et al., 2009).
2.3.
Brevetoxins
2.3.1.
Toxin description and activity
Brevetoxins are polyether, non-polar compounds that depolarize cell membranes by opening voltage gate sodium ion channels and induce enhanced inward flux of ions into cells (Lin et al., 1981; Baden, 1983, 1989; Purkerson et al., 1999). Brevetoxins exist as two structural types and multiple analogs possessing various levels of potency (Baden, 1989; Cembella, 2003; Kirkpatrick et al., 2004). The types differ in their ladderframe polycyclic ether structural backbones and are designated type A and type B (Fig. 2). The brevetoxin derivatives found in the marine environment (PbTx-2, PbTx-3 and PbTx-9; Table 3) are produced most commonly by dinoflagellate and raphidophyte algae and are of the structural type B (Baden, 1989; Baden et al., 2005). Brevetoxins bind to site 5 of the voltage-sensitive sodium channel in neurons, causing these channels to remain open and fire repeatedly (Catterall, 1992; Cestele and Catterall, 2000). Brevetoxin poisoning in humans is referred to as neurotoxic shellfish poisoning (NSP), and includes gastrointestinal symptoms of nausea, diarrhea and abdominal pain, neurologic symptoms of paresthesia, and respiratory irritation and/or failure (Kirkpatrick et al., 2004). The effects of brevetoxins on human health are well documented along the western coast of Florida where severe, nearly annual red tides caused by the dinoflagellate Karenia brevis release large amounts of brevetoxins into the air when the fragile cells are broken in breaking waves at the water’s edge (Kusek et al., 1999; Kirkpatrick et al., 2004). The aerosolized toxins constitute a significant health risk when they are inhaled and, as a result, K. brevis blooms are one of the most intensively studied and best-understood regional HABs. Red tides caused by K. brevis have been implicated in marine mammal fatalities, fish kills and human illnesses. Brevetoxins have not been reported from the U.S. west coast, and therefore no known human fatalities or health issues have yet been attributed to brevetoxins from that region.
2.3.2.
Producers
Several dinoflagellate species and a few raphidophyte species produce a suite of brevetoxins (Baden, 1989). K. brevis, the most notorious brevetoxin producer within the Gulf of Mexico, has not been observed on the west coast of the U.S., but several species of raphidophytes that are potential brevetoxin producers have been documented (Tables 1 and 2). Heterosigma akashiwo (Fig. 3e), Chattonella marina (Fig. 3f) and Fibrocapsa japonica have been isolated and cultured from coastal waters off southern California. In general there are few reports of significant blooms of these species on the west coast, although blooms of raphidophytes in San Francisco Bay and Delaware Inland Bays have been observed with cell abundances in excess of 108 cells L1 (Herndon et al., 2003; Coyne et al., 2005). In part, this lack of information is a consequence of the fact that raphidophyte species are notoriously difficult to identify using traditional microscopical techniques because they preserve poorly (Hallegraeff and
394
water research 44 (2010) 385–416
water research 44 (2010) 385–416
Hara, 1995; Throndsen, 1997). Recently developed genetic approaches for the identification and quantification of some raphidophytes are beginning to provide much-needed tools for studying the distributions and ecology of these HAB species (Handy et al., 2006; Demir et al., 2008). Despite the difficulties of characterizing these blooms, fish kills have been attributed to raphidophyte blooms on the west coast of the U.S. although these studies have not quantified brevetoxins (Hershberger et al., 1997; Hard et al., 2000).
2.4.
Diarrhetic shellfish toxins
2.4.1.
Toxin description and activity
Toxins that cause diarrhetic shellfish poisoning (DSP) include okadaic acid, dinophysistoxins and pectenotoxins (Ramsdell et al., 2005). Okadaic acid is a monocarboxylic acid named for the marine sponge Halichondria okadai from which it was first isolated (Tachibana et al., 1981). Okadaic acid can also be found in natural water samples in polar and non-polar esteric forms (Prassopoulou et al., 2009). The first dinophysistoxin described was isolated from the mussel M. edilus and was found to be a methyl form of okadaic acid (Murata et al., 1982). Okadaic acid and dinophysistoxins are linear polyethers (Fig. 2) and the mode of action is the inhibition of protein phosphatases (Takai et al., 1987; Bialojan and Takai, 1988; Haystead et al., 1989), enzymes that play a key role in dephosphorylation in many biological processes including cell cycle regulation. The pectenotoxins are lipid soluble and differ structurally from other diarrhetic toxins in that they possess a lactone ring (Fig. 2), and not considered to be protein phosphatase inhibitors, but have a high actin-depolarizing action (Hori et al., 1999). There is speculation that pectenotoxins may not produce diarrhetic effects (Cembella, 2003). DSP toxins (Table 3) were named for the human symptoms resulting from the ingestion of contaminated shellfish, including inflammation of the intestinal tract, diarrhea, abdominal cramps, vomiting and nausea beginning 30 min to a few hours after ingestion (Hallegraeff, 2003). In addition to the symptoms listed above, okadaic acid is known to be a strong tumor promoter (Suganuma et al., 1988), although the potential health implications of this activity due to the ingestion of contaminated seafood is unknown. There are presently no documented cases of DSP resulting from okadaic acid, dinophysistoxins or pectenotoxins on the U.S. west coast, and thus these toxins are not regularly monitored in the marine environment. DSP toxins have been detected in mussels and water samples from California (Sutherland, 2008), so it is possible that DSP has occurred on the U.S. west coast but has been attributed to other sources of contamination.
2.4.2.
395
Producers
Okadaic acid and dinophysistoxins are produced by a few species of the dinoflagellate genus Prorocentrum (Cembella, 2003) and most commonly by species of the genus Dinophysis. Dinophysis species present on the western U.S coast include D. acuminata, D. acuta, D. caudata, D. fortii, D. norvegica, D. rotundata, and D. tripos (Table 2). D. acuminata and D. fortii have been documented in Californian waters for many years (Bigelow and Leslie, 1930). D. acuminata produces okadaic acid (Yasumoto et al., 1985), D. fortii produces okadaic acid, dinophysistoxins and pectenotoxins (Yasumoto et al., 1980; Murata et al., 1982) while D. rotundata and D. tripos produce dinophysistoxin-1 (Lee et al., 1989). Dinophysis species are technically not phytoplankton, but heterotrophic protists that retain chloroplasts acquired from their prey. Dinophysis acquires its chloroplasts by preying on ciliates, which in turn prey on cryptophyte algae. This complex trophic relationship has made the culture of these species unsuccessful until recently (Park et al., 2006), and therefore no information on the environmental conditions that influence toxin production by these species presently exists. However, genus members are easily distinguished by light microscopy because they possess a pronounced ‘keel’ and other unique morphological aspects (Fig. 3d). These species are often encountered in plankton samples. Abundances of 103 cells L1 are commonly encountered (Nishitani et al., 2005), and occasionally they may reach abundances in excess of 105 cells L1 (Carpenter et al., 1995).
2.5.
Yessotoxin
2.5.1.
Toxin Description
Yessotoxin (Fig. 2; Table 3) is a chiral molecule of high polarity due to the presence of two sulfate groups. The molecule consists of fused polyether rings organized into a ladder shaped skeleton (Yasumoto and Murata, 1993; Wright and Cembella, 1998), a structure similar to other ladder-like polyether toxins such as the ciguatera toxin complex (ciguatoxins and maitotoxin), gambieric acids and brevetoxins (Yasumoto and Murata, 1993; Wright and Cembella, 1998). There are nearly 100 analogs of yessotoxin that have been identified to date (Satake et al., 1997, 1999; Ciminiello et al., 1998, 2000, 2001; Daiguji et al., 1998; Miles et al., 2004, 2005a,b, 2006; Paz et al., 2006). The yessotoxin class was named for the species of scallop, Patinopecten yessoensis, in which it was initially detected (Murata et al., 1987). Yessotoxin was originally classified in the DSP-toxin class because it was detected with other DSP toxins, but it appears that it does not induce
Fig. 3 – Major algal toxin producers occurring along the U.S. west coast. (a) Pseudo-nitzschia australis, a producer of domoic acid; (b) Scanning electron micrograph of Pseudo-nitzschia australis; (c) Alexandrium catenella, a producer of saxitoxin; (d) Dinophysis sp., a producer of okadaic acid; (e) Heterosigma akashiwo, a producer of brevetoxins; (f) Chattonella marina, a producer of brevetoxins; (g) Cochlodinium sp.; (h) Lingulodinium polyedrum, a producer of yessotoxin; (i) Phaeocystis globosa colony; (j) foam produced by the prymnesiophyte, Phaeocystis accumulating along the shore; (k) Unconcentrated seawater from King Harbor, City of Redondo Beach, with significant discoloration due to an algal bloom; (l) Prorocentrum sp., the dominant organism in (k); and (m) higher magnification of Prorocentrum sp. from (l). Scale bars [ 10 mm. Photo (b) courtesy of Peter Miller, (c) courtesy of Carmelo Tomas, (j) courtesy of Cindi Heslin.
396
Table 3 – Summary of toxins that can be present in Southern California waters. MW: molecular weight. Toxin
Properties
Formula
MW
Mode of action
References
Hydrosoluble At pH 7: DA3
C15H21NO6
311.14
Binds to glutamate receptors in the brain disrupting normal neurochemical transmission
Wright et al. (1990), Quilliam (2003)
Saxitoxins (STXs)
Hydrosoluble pH 7: Stable
C10H17N7O4
299.3
Bind to site 1 of voltage-sensitive sodium channels and block sodium conductance; bind to calcium and potassium channels
Wong et al. (1971), Wang et al. (2003), Su et al. (2004), Catterall (1992), Cestele and Catterall (2000)
Brevetoxins (PbTxs) Brevetoxin 2 (PbTx 2) Brevetoxin 3 (PbTx 3) Brevetoxin 9 (PbTx 9) Diarrhetic shellfish toxins Okadaic acid (OA)
Liposoluble C50H70O14 C50H72O15 C50H74O14
895.1 897.1 899.1
Bind to site 5 of voltage-sensitive sodium channels, shifting activation to more negative membrane potentials and block channel activation
Lin et al. (1981), Baden (1983, 1989), Purkerson et al. (1999)
C44H68O13
805
Inhibits protein phosphatases, inhibits dephosphorylation of proteins
Tachibana et al. (1981), Murata et al. (1982), Yasumoto et al. (1985), Takai et al. (1987), Bialojan and Takai (1988), Haystead et al. (1989), Hori et al. (1999)
Dinophysistoxins (DTXs) Pectenotoxins (PTXs) Yessotoxins (YTXs)
Liposoluble Hydrosoluble
High actin-depolarizaing action C55H80O21S2Na2
1187.3
Activation of phosphodiesterase in the presence of external Ca2þ; Disruption of the E-cadherin–catenin system in epithelial cells and potentially disrupting its tumour suppressive functions
Murata et al. (1987), Takahashi et al. (1996), Alfonso et al. (2003), Ronzitti et al. (2004)
water research 44 (2010) 385–416
Domoic acid (DA)
water research 44 (2010) 385–416
diarrhetic effects (Ogino et al., 1997). Accordingly, the regulation of the European Commission on marine biotoxins now considers yessotoxins separately from DSP toxins (European Commission, 2002). The great number of yessotoxin analogs complicates toxicity studies, and may explain the sometimes contradictory reports of its mode of action. Studies have shown that lysosomes, the immune system and the thymus (with tumorigenic implications) are the biological targets of yessotoxin (Franchini et al., 2004; Malagoli et al., 2006), while other reports have indicated cardiotoxic effects (Terao et al., 1990; Ogino et al., 1997; Aune et al., 2002). The cardiotoxicity of yessotoxin might be attributed to phosphodiesterase activation in the presence of external calcium ions (Alfonso et al., 2003). Unlike the other marine toxins mentioned above, there have been no reported human health issues or marine mammal deaths associated with yessotoxins.
2.5.2.
Producers
There are three known yessotoxin-producing dinoflagellates, Protoceratium reticulatum, Lingulodinium polyedrum, and Gonyaulax spinifera, and they have all been observed in coastal waters off the western U.S. (Table 1). According to phylogenetic analyses of available rRNA gene sequences, the capacity for yessotoxin production appears to be restricted to the order Gonyaulacales (Howard et al., 2009). However, toxin production among strains within each species appears to be highly variable. Yessotoxin has been detected in L. polyedrum isolates cultured from around the globe (Tubaro et al., 1998; Draisci et al., 1999a; Strom et al., 2003; Paz et al., 2004), including isolates from Californian coastal waters (Armstrong and Kudela, 2006). The reported cellular concentrations in the latter cells ranged from below detection to 1.5 pg cell1, indicating that L. polyedrum is significantly less toxic than P. reticulatum or G. spinifera. Yessotoxin has been recorded in blue mussels at low concentrations along the U.S. west coast (Table 2) during red tides caused by L. polyedrum, as well as during non-bloom conditions, but yessotoxin production by this dinoflagellate appears to be less than toxin levels produced by isolates from other geographical regions (see Table 2 in Howard et al., 2008). Expansive and dense blooms of L. polyedrum (Fig. 3 h) have been reported in California since 1901, but there have only been anecdotal reports of health problems associated with the red tides caused by L. polyedrum (Kudela and Cochlan, 2000). Yessotoxin production by isolates of P. reticulatum has been confirmed (Satake et al., 1997; Boni et al., 2002; Miles et al., 2002; Riobo et al., 2002; Stobo et al., 2003; Samdal et al., 2004; Eiki et al., 2005), but concentrations ranging from below detection to 79 pg cell1 have been reported for isolates from Washington, California and Florida (Paz et al., 2004, 2007). Isolates of G. spinifera appear to be the most prolific yessotoxin producers. Concentrations in New Zealand isolates ranged from below detection to 200 pg cell1 (Rhodes et al., 2006), more than 20-fold higher than the per-cell toxicity of P. reticulatum and 600-fold higher than L. polyedrum. G. spinifera does not generally bloom in high densities on the U.S. west coast, but it has been frequently observed at low abundances (Howard et al., 2008; M. Silver, pers. comm.) and has reached bloom concentrations in
397
Tomales Bay, north of San Francisco (G. Langlois pers. comm.). Yessotoxins are monitored in New Zealand, Europe and Japan but they are not routinely measured on the U.S. west coast. Howard et al. (2008) was the first study to confirm yessotoxins in California and Washington coastal waters, albeit at very low concentrations.
2.6.
Toxin detection and quantification
A wide variety of methodologies and technologies exist to detect, characterize and quantify the major toxins produced by microalgae. These approaches can be broadly divided into those used to characterize biological activity (toxicity assays) and those used to identify specific chemical structure(s) (immunological, various analytical techniques). Because of the highly variable approaches employed, and in most cases the highly diverse set of compounds comprising a toxin class, the methods provide somewhat different estimates of absolute toxin concentrations or presumed toxicity. It is also noteworthy that the majority of the protocols used to measure algal toxins have focused on the analysis of tissue samples that may be a source of human contamination (e.g. shellfish or finfish tissue) or plankton material filtered from seawater samples. Relatively few studies have examined the concentrations of toxins dissolved in seawater (Table 2). For this reason, limits of detection for toxins in seawater are poorly known for most approaches. However, based on analytical approaches presently available, a practical limit of detection for a few of the major concerns (domoic acid and saxitoxins) is in the range of 0.1 mg per liter of seawater for immunological approaches, but a detection limit of approximately 0.01 mg per liter for dissolved domoic acid in seawater using highperformance liquid chromatography has been reported (Pocklington et al., 1990). Detection limits for toxins in particulate material in the water can be significantly lower because particles can be concentrated by filtration prior to extraction. Knowledge of the concentrations of dissolved toxins would be preferable from the perspective of reverse osmosis desalination operations because dissolved toxins (rather than cell-bound toxins) would most likely impact these membranes. Domoic acid has been detected and quantified in seawater, plankton, shellfish extract and homogenate, as well as sea bird and mammalian body fluid (e.g. blood, urine, amniotic fluid, cerebral spinal fluid). Analytical approaches for these measurements include commercially available immunological techniques (enzyme-linked immunosorbent assay; ELISA) (Garthwaite et al., 1998, 2001), high pressure liquid chromatography (HPLC) with ultraviolet (UV) diode array detection (DAD) (Quilliam et al., 1989; Quilliam, 2003), receptor binding assay (RBA) (Van Dolah et al., 1994), mouse bioassay (MBA) or liquid chromatography–mass spectrometry (LC–MS). The results obtained by these various approaches are not yet completely compatible or comparable, and therefore comparisons across studies using different analytical methods can be problematic. In general, the choice of an approach is a compromise between cost of analysis (or access to costly equipment), sample throughput, sensitivity and analytical goal (e.g. thorough chemical characterization versus overall
398
water research 44 (2010) 385–416
toxicity, or human health risk versus scientific understanding of bloom dynamics). Saxitoxins can be rapidly detected and quantified with commercial ELISA kits, but the high specificity of these tests precludes the recognition of certain members of the saxitoxin family, especially the neo-saxitoxin (Garthwaite et al., 2001). Saxitoxin detection and quantification is often accomplished by HPLC, RBA, LC–MS, and MBA. Regulatory programs for seafood consumption are still based on ‘lethal mouse dosage’. Similarly, detection and quantification of brevetoxin and its derivatives in seawater, shellfish, and mammalian body fluids can be accomplished using commercially available ELISA kits (Naar et al., 2002), by HPLC, RBA (Van Dolah et al., 1994), LC–MS and MBA. A comparative study also quantified brevetoxins by radioimmunoassay (RIA) and a neuroblastoma (N2A) cytotoxicity assay (Twiner et al., 2007). Quantification of the DSP toxin suite can be underestimated by ELISA because commercial ELISA assays are usually optimized to detect okadaic acid and not the dinophysistoxins (Garthwaite et al., 2001). HPLC with fluorimetric detection (HPLC-FLD) has been commonly used to detect and quantify okadaic acid, its polar and non-polar esters, as well as dinophysistoxins (Lee et al., 1987). The MBA method has also been routinely used for the detection of DSP toxins. The detection and quantification of yessotoxin is problematic because of the extensive suite of derivatives that may exist. HPLC-FLD analysis (Yasumoto and Takizawa, 1997), MBA, LC-MS (Draisci et al., 1999b; Paz et al., 2006, 2008) and ELISA (Samdal et al., 2004, 2005) have been employed.
2.7. Other potentially toxic, noxious and nuisance organisms Reports of newly occurring HAB species, or recognition of extant issues that have gone previously undocumented, are
increasing our awareness of other potentially harmful bloomforming algae along the U.S. west coast. For example, blooms of an emerging potentially toxic organism, Cochlodinium sp. (Fig. 3g) off central California have recently been reported (Curtiss et al., 2008; Iwataki et al., 2008; Kudela et al., 2008b). This species is difficult to identify using light microscopy, and therefore researchers have begun to use gene sequences to provide accurate identification (Iwataki et al., 2007, 2008; Matsuoka et al., 2008). While this organism has only recently reached sufficient abundances to discolor Californian waters, there has already been one reported abalone loss in central California that appears to be associated with a bloom of Cochlodinium (R. Kudela pers. comm). Species of the dinoflagellate Prorocentrum (Fig. 3k–m) occasionally bloom in Californian coastal waters where they can attain very high abundances periodically, causing discolorations of the water and nuisance accumulations of algae (Holmes et al., 1967; Shipe et al., 2008). The Prorocentrum species known to occur in Californian waters have not yet demonstrated DSP toxin production, but species from elsewhere in the world are known to produce theses toxins (Table 1). Similarly, massive blooms of the dinoflagellate Akashiwo sanguinea are common in coastal waters of southern and central California and have recently been the cause of seabird mortality due to surfactant-like proteins (Jessup et al., 2009). Although they may not be overtly toxic, these blooms can cause animal mortalities, deplete oxygen, and result in an increased organic and biomass loading to a seawater desalination facility. The prymnesiophyte Phaeocystis globosa infrequently attains high abundances off the Californian coast (Armonies, 1989). This species produces single cells that are <10 mm in size, but it also forms fluid-filled colonies several millimeters in diameter in which individual cells are embedded in the polysaccharide skin of the colony (Fig. 3i). Single cells of
Fig. 4 – Time series of chlorophyll a fluorescence, temperature and dissolved oxygen in King Harbor, City of Redondo Beach, CA. Note the short-term temporal fluctuations in these parameters that are a result of tide, wind and biological interactions. These measurements were collected using autonomously recording sensors that provide high resolution observations of chemical and physical properties that might indicate an algal bloom, or environmental factors that might stimulate a bloom.
water research 44 (2010) 385–416
Phaeocystis are consumed by many zooplankton species but the colonies are typically a poor food source. Selective feeding on single cells appears to favor colony formation and the accumulation of colonies in the water column (Netjstgaard et al., 2007). When released via colony destruction or algal life cycle events, the colony matrix material is easily worked into a ‘sea foam’ that can form layers many centimeters (even meters) thick at the ocean surface or along the coastline over fairly extensive regions (Fig. 3j).
3. Spatial and temporal patterns of harmful algae A fundamental aspect of the biology of harmful algal blooms, and of vital importance for desalination operations, is the tendency for rapid and dramatic changes in the spatial and temporal distributions of these species. These changes occur rapidly across a wide range of scales, and pose challenges for documenting and predicting these distributions. Numerous approaches and instruments have been developed to characterize the dynamics of phytoplankton communities. These approaches presently have significant limitations on their abilities to identify species composition of a bloom, but they provide crucial information on the emergence and longevity of bloom events as well as their vertical and geographical extent. This information can help seawater desalination facilities adjust operations to ensure reliable production for the duration of the bloom.
3.1.
Temporal variability
Significant temporal variations in the abundances of phytoplankton take place on time scales ranging from hours to decades. Short-term temporal variability (hours to a few weeks) can be a consequence of rapid population growth or consumption by herbivores, sinking of senescent populations, diel vertical migration, tidal movement, and aggregation or dispersal by physical processes such as water mass convergences or divergences. Diel vertical migration of several dinoflagellates has been attributed to geotaxis, phototaxis and nutrient status (Eppley et al., 1968; Blasco, 1978; Cullen and Horrigan, 1981; Levandowsky and Kaneta, 1987). Classic responses involve nighttime sinking out of surface waters to deeper water where nutrient concentrations are greater, and rising into surface waters for photosynthesis during daytime. Shifts in nutrient cell quotas that accompany these migrations may have significant implications for toxin production because cell toxicity can be related to nutrient status of the cells (Anderson et al., 1990; Flynn et al., 1994; John and Flynn, 2000; Flynn, 2002). Diel-to-weekly variations in phytoplankton abundance can be characterized using self-contained or wirelessly networked sensor packages. Data collected in King Harbor of the City of Redondo Beach, CA (Fig. 4) demonstrate the efficacy of these instruments for providing high temporal resolution of chlorophyll a fluorescence (which approximates phytoplankton biomass) and pertinent environmental factors (e.g. dissolved oxygen and temperature) that provide insight into the factors
399
controlling the pattern. These data reveal a 4-fold variation in phytoplankton standing stock over a two-day period. In addition, changes in water quality criteria were easily and rapidly identified (e.g. decrease in dissolved oxygen concentration near noontime on September 13). Daily variations in the latter parameter can be extreme at night during algal blooms. High resolution, short-term monitoring approaches allow rapid detection of sentinel parameters, and in turn provide information for the development of predictive models of HAB events. Chlorophyll a fluorescence sensors provide valuable information on the short-term temporal patterns of total algal biomass, but these instruments cannot identify noxious algal species within a phytoplankton assemblage. More sophisticated instruments now coming online offer that possibility. For example, the Environmental Sample Processor (http:// www.mbari.org/ESP) is an in situ instrument capable of performing real-time identifications of HAB and other microbial species, as well as toxin analyses such as domoic acid using on-board molecular analyses (Greenfield et al., 2006). Similarly, handheld devices now exist for the detection of some HAB species and toxins in the field (Casper et al., 2007). These rapid and highly specific analyses are becoming valuable tools for quick determinations of toxin presence resulting from algal blooms. These more costly instruments can be used to improve the information available on bloom composition once the cheaper and more readily available sensors identify an emerging bloom event. Seasonal variability of HAB species and their toxins along the Californian coast can be gleaned from the records of the Marine Biotoxin Monitoring Program (MBMP) of the California Department of Public Health (CDPH), a program started after a major domoic acid outbreak in the fall of 1991. At present, the annual effort involves the analysis of approximately 300 shellfish samples for domoic acid and >1000 samples for PSP toxins from all fifteen Californian coastal counties (CDPH, 2007). Shellfish toxin information provides a reasonable representation of toxins in the upper water column over seasonal and annual scales such as demonstrated in studies on toxic dinoflagellates (Montojo et al., 2006; Jester et al., 2009a; Moore et al., 2009). Detailed information on the sampling effort is provided in the MBMP annual reports (Langlois, 2007). Distributions of domoic acid and PSP toxins along the Californian coast during the period 2002–2007 based on the MBMP data reveal both seasonal and geographical trends (Figs. 5 and 6). Monthly averages for the 6-year period (histograms) and maximal concentrations (triangles and lines) showed detectable concentrations of PSP toxins and domoic acid in shellfish in nearly all months for all Californian coastal counties (Fig. 5). Domoic acid concentrations showed pronounced seasonality, with very high peaks during spring and much smaller peaks during fall. This temporal trend is concordant with previous observations along the southern Californian coast (Lange et al., 1994; Walz et al., 1994; Schnetzer et al., 2007; Shipe et al., 2008). Fall domoic acid peaks correspond to minor blooms of toxic Pseudo-nitzschia occasionally noted on the west coast of the U.S. (Bolin and Abbott, 1960; Buck et al., 1992; Walz et al., 1994; Trainer et al., 2002; Schnetzer et al., 2007).
400
water research 44 (2010) 385–416
Domoic acid (µg/100g)
70000 60000 50000 40000 30000 20000
Mean of monthly averages Absolute maximum
15000
10000
5000
0 1600 1400 1200 1000 800 600 400
PSP toxins (µg/100g)
300
200
100
0 Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
Dec
Fig. 5 – Seasonal variability of domoic acid (upper panel) and PSP toxin (lower panel) concentrations along the Californian coast. Monthly averages of data collected from 2002 and 2007 summarized for fifteen Californian coastal counties are shown as histograms. Also shown are the maximal values recorded during each month over the entire study period (triangles and solid lines). Toxin concentrations were derived from shellfish tissue. Data from CDPH (Langlois, 2007).
Seasonal peaks in domoic acid along the U.S. west coast differ along a latitudinal gradient. Highest domoic acid concentrations in Washington have been observed during the fall (Trainer, 2002; Office of Shellfish and Water Protection, 2008; Moore et al., 2009) compared to spring blooms that are common in southern California (Schnetzer et al., 2007). This north-south trend in seasonality is also evident on a smaller latitudinal scale. Blooms along the Californian coast have been more frequently observed later in the year in northern counties (Humboldt versus Santa Barbara counties in Fig. 6). The time lag between bloom periods observed from south to north may be related to the timing of the California Current System (CCS) upwelling maximum, which brings nutrients into surface waters and promotes phytoplankton growth. The CCS upwelling occurs in early spring in southern California, in June off Washington,
and throughout summer in northern California and Oregon (Reid et al., 1956; Landry, 1989). Monthly averages as well as maximal PSP toxin concentrations showed less pronounced seasonal variability than domoic acid during the period 2002–2007, although the highest concentrations were recorded from July to September (Fig. 5). This seasonal pattern of maximal PSP is in agreement with the last 25 years of monitoring results (Langlois, 2007). It is also consistent with the 1927–1989 observations on the Californian coast indicating that most significant concentrations of the toxin take place between May and October (Price et al., 1991). The highest PSP toxin concentrations in shellfish have also been observed in the summer and the fall periods off the Washington and Oregon coasts (Trainer et al., 2002; Determan, 2003).
401
water research 44 (2010) 385–416
Domoic acid
PSP toxins Marin County 140
12000
120
10000
100
PSP toxins (µg/100 g)
North
Domoic acid (µg/100 g)
Humboldt County 14000
8000 6000 4000 2000
****
*
** **
**
2002 2003 2004 2005 2006 2007 * not detected
80 60 40 20
**
0
0 April
May
June
July
140
12000
120
10000
100
8000 6000 4000 2000
**
September
San Luis Obispo County
14000
PSP toxins (µg/100 g)
South
Domoic acid (µg/100 g)
Santa Barbara County
August
* * **
80 60 40 20
**
**
**
* *
**
*
0
0 April
May
June
July
August
September
Fig. 6 – Interannual variation in toxin concentration in four Californian coastal counties from 2002 to 2007 during three months exhibiting high concentrations of domoic acid (April, May, June) and saxitoxin (July, August, September). Humboldt and Marin counties are located north of Santa Barbara and San Luis Obispo counties. Toxin concentrations were measured from shellfish tissues. Data from CDPH (Langlois, 2007).
Diel to seasonal temporal patterns in ASP and PSP outbreaks off the U.S. west coast are augmented by large interannual variability in the intensity and frequency of HABs. Interannual variations in HABs presumably are related, at least in part, to changes in atmospheric and hydrographic features modulated by the 3–7 year cycles of El Nin˜o-Southern Oscillations (ENSOs) (Price et al., 1991; Horner et al., 1997), but the exact relationship between HABs and these climatic events is not clear because there are still relatively few observations spanning these temporal regimes. Warming off California during El Nin˜o episodes reduces seasonal upwelling, enhances physical stratification in the CCS and lowers the nutricline in the water column (the depth at which nutrient concentrations increase rapidly). It is clear that these changes in water stability and nutrient availability have significant impacts on plankton productivity and community structure, but the specific responses of the phytoplankton communities vis-a`-vis HAB events are not yet predictable (Barber and Chavez, 1983). ASP outbreaks occurred during El Nin˜o episodes of 1991 and 1997–98 along the coasts of Oregon, Washington and California (Table 2; Moore et al., 2008), but the specific factors contributing to toxic Pseudo-nitzschia blooms during these events could not be clearly identified (Horner and Postel, 1993; Trainer et al., 2000). Interannual variability in ASP and PSP concentrations in coastal waters along the Californian coast was evident and substantial in the MBMP dataset during the period 2002–2007 (Fig. 6). ASP and PSP outbreaks were frequent, but
they varied in intensity and the timing of peak concentrations between the years within a single geographical location, and between southern and northern Californian counties in the same year and season. Notably, the magnitude of this variability is on the same order of magnitude as the variability observed on short-term (daily) or seasonal time scales. Little is known regarding the longer time scale fluctuations in HABs along the U.S. west coast. Multi-decadal fluctuations in ocean temperature are known to provoke shifts in the biological regime (Chavez et al., 2003), and it is anticipated that climatic shifts would affect the timing, intensity or frequency of phytoplankton blooms. Long-term regime shifts may affect the occurrence of blooms of L. polyedrum, a producer of yessotoxin, along the coast of southern California (Tables 1 and 2). The Pacific Ocean was cooler in the years preceding 1976, and red tides dominated by L. polyedrum commonly developed along the Southern California Bight during fall (Gregorio and Pieper, 2000). During a recent warm regime (1976-mid 1990s), red tides occurred during winter and spring and persisted until summer in the region of the Los Angeles River mouth (Gregorio and Pieper, 2000). Recent massive blooms of L. polyedrum during fall may indicate a return to the pre-1976 conditions (Moorthi et al., 2006). The generality surmised from data depicting short- to longterm temporal variability in phytotoxin dynamics is that variability can be high at all scales. Given our present state of understanding regarding the specific combination of forcing factors that give rise to this high variability, it is difficult to
402
water research 44 (2010) 385–416
Fig. 7 – Two-dimensional presentations of chemical and physical data collected by an autonomous vehicle (Webb Slocum glider, Teledyne Webb Research, East Falmouth, MA) along a nearshore–offshore transect at Newport Beach, CA shown in (f) (indicated on map by the blue line). Contour plots are shown for (a) temperature, (b) chlorophyll a fluorescence, (c) salinity, (d) backscattered light and (e) water density. The Webb Slocum glider is an autonomous vehicle commonly employed in coastal ecosystems. This buoyancy-driven underwater vehicle generates horizontal motion by ascending and descending with pitched wings (Schofield et al., 2007). A rudder directs heading while buoyancy is controlled by pumping seawater into and out of the nose of the vehicle. This long-lived, low-power glider achieves horizontal velocities of approximately 25– 30 cm sL1 with vertical velocities of 10–15 cm sL1.
accurately predict the timing and magnitude of toxic blooms. For these reasons, monitoring at multiple temporal scales is necessary to adequately characterize plankton dynamics.
3.2.
Spatial variability
Spatial variability of HABs is considerable at multiple scales, analogous (and strongly related) to the temporal variability described above. Blooms can be highly localized (10s of meters) or expansive (100s of kilometers), and distributions vertically within the water column are heterogeneous over scales of centimeters to meters. The geographical extent and heterogeneous nature of the U.S. west coast results from differences in local hydrography that are manifested in smalland large-scale differences in spatial patterns of toxic blooms. Regional-scale variations in HAB distributions are illustrated by MBMP data during 2002–2007 for ASP and PSP concentrations observed in counties along northern and central California (Fig. 6). ASP events (frequency and level of
toxicity) were generally lower in northerly Humboldt County during this period relative to Santa Barbara County nearly 1000 km to the south. More recently, high domoic acid concentrations have been observed within the Southern California Bight, presumably indicating a continuing southward movement of the Pseudo-nitzschia spring blooms (Langlois, 2007; Schnetzer et al., 2007). On the other hand, Marin County (north of San Francisco) exhibited higher monthly PSP values during most months than San Luis Obispo County located nearly 500 km to the south. This general latitudinal trend in PSP events is consistent with findings that the three southernmost counties (Los Angeles, Orange and San Diego) generally experience low concentrations of PSP relative to northern California (Price et al., 1991; Langlois, 2007). Regional-scale and geographical differences in ASP and PSP events have also been reported along the coastline of Oregon (Trainer et al., 2002). Small-scale spatial variability (horizontally and vertically) can also be dramatic. HAB events can be highly restricted
water research 44 (2010) 385–416
geographically (e.g. relegated to a protected embayment). Even within spatially extensive blooms, phytoplankton biomass is often highly discontinuous over very small spatial scales because of differences in local circulation, and wind or wave forcing. Considerable spatial variability in the abundance of P. australis within Monterey Bay has been observed, a phenomenon that has been attributed to advective forces (Buck et al., 1992). Discontinuities within a water column, such as thermoclines, haloclines, nutriclines and light absorption often leads to the establishment of subsurface microlayers where phytoplankton biomass can be many-fold elevated relative to algal standing stocks only centimeters above or below (Dekshenieks et al., 2001; Rines et al., 2002). Characterization of phytoplankton spatial distributions include approaches ranging from shore-based sampling, to the use of oceanographic ships, to remote sensing of largescale patterns using satellite imagery. Vertical profiling of phytoplankton assemblages can be accomplished using over-the-side, ship-based instrument packages and more recently autonomous vehicles equipped with a variety of sensor packages. The use of autonomous vehicles to provide synoptic measurements of phytoplankton biomass from chlorophyll a fluorescence and pertinent chemical/physical parameters is state-of-the-art for obtaining two-dimensional cross-sections or three-dimensional patterns in the water column (Fig. 7). Autonomous vehicles provide time- and depth-stamped measurements of a variety of parameters that can be optimized for a specific mission. Such an instrument deployed off Newport Beach, CA (blue line in Fig. 7f) during May–June 2008 yielded detailed patterns of chemical/biological parameters that provided information on the extent and vertical distribution of phytoplankton biomass (Fig. 7a–e). Evidence of upwelling in this
33°45’N
33°40’N
33°35’N
118°10’W
118°00’W
0.01-0.50 0.51-1.00 1.01-1.50 1.51-2.00 2.01-2.50 2.51-3.00 3.01-3.50 3.51-4.00 4.01-4.50 4.51-5.00
Particulate domoic acid (µg L-1)
Fig. 8 – Spatial variability in domoic acid concentrations contained in the total particulate material (algal cells, other microbes and detritus) in surface waters within the San Pedro Bay area including the Long Beach-Los Angeles harbor area. Data were collected at 20 sampling stations (locations indicated by filled circles).
403
spring deployment was apparent from the upward-pointing isotherms (temperature) and isopycnals (density) on the shoreward end of the transect (left sides of Fig. 7a, e). This was particularly evident in the temperature plot 5–10 km from shore between the surface and 20 m (Fig. 7a, red circle). Nutrients (not shown) were generally depleted in surface waters, and increased with decreasing temperature. This physico-chemical structure is concordant with a significant subsurface maximum in chlorophyll a concentration, indicative of a response of the phytoplankton assemblage to elevated nutrient concentrations at this depth (red circle on left in Fig. 7b), as previously observed (Jones et al., 2002). The cross-sectional picture provided by the autonomous vehicle indicated that the phytoplankton community had a patchy structure on the scale of meters (vertically) and kilometers (horizontally). Two major horizontal patches were observed at 6–12 km and at 15–21 km from shore (red circles in Fig. 7b). Measurements in addition to chlorophyll a fluorescence can add information on the observed general patterns such as particle size distributions derived from the optical backscatter spectrum. The backscatter profile obtained at a wavelength of 532 nm indicated a patch of particles that were not of algal nature (Fig. 7d, red circle). The wavelength-dependent slope of the backscatter, which is dependent on the particle size distribution, can also be mapped to indicate the size-class of phytoplankton particles that dominate the chlorophyll a maxima. This information is particularly important for a seawater desalination facility, where the incoming particle size distribution is known to impact the source water filterability. Autonomous vehicles allow nearly synoptic measurements of the spatial distribution of phytoplankton and ancillary parameters. Many of these instruments operate for significant periods of time (weeks) and thereby supply temporal as well as spatial coverage. Knowledge of the temporal evolution and spatial organization of coastal marine systems enables a better understanding of the linkages between physical processes and the biological responses that contribute to the formation of algal blooms. Moreover, data from these instruments can be telemetered to the laboratory in near-real time and used to direct costly efforts such as shipboard sampling, or plan operations for land-based activities and measurements. Broad-scale, horizontal distributions can also be acquired via shipboard sampling programs (Fig. 8). Shipboard work is time and labor intensive, but onboard sample processing enables more sophisticated analyses than autonomous vehicles are presently capable of providing. Moreover, ships and other manned platforms permit time series studies at a single study site. Shipboard sampling conducted in April 2008 in the Long Beach-Los Angeles harbor area and the adjacent San Pedro Channel demonstrated considerable spatial variability in the distribution of HAB species and their toxins within this relatively small area (approximately 500 km2). Results indicated a patchy distribution of domoic acid in particulate material (i.e. within phytoplankton cells) collected near the surface with highest concentrations in the vicinity of the harbor breakwater and at several offshore locations (Fig. 8). Intermediate regions exhibited toxin concentrations that were more than an order of magnitude less than these maxima.
404
Environmental driving factors
45
7
40
Relative fluorescence TMP
6
35
5
30
4
25
3
20 15
2
10
1
5
Transmembrane pressure (TMP, psi)
Characterizing the factors that lead to the stimulation of harmful algae and the production of toxins by these algae has been an area of very active research for decades. These studies involve field observations to document the spatiotemporal extent of blooms and toxin concentrations in plankton and marine life, and laboratory experiments aimed at understanding the key environmental factors leading to HAB events and toxin production. The overall results gleaned from many years of work group into three basic categories: (1) factors and conditions leading to phytoplankton blooms in general, (2) factors leading specifically to the growth of HAB species, and (3) factors leading to toxin production. Numerous factors have been implicated as contributors to the observed global expansion of HABs (Smayda, 1990; Hallegraeff, 1993, 2003; Anderson et al., 2002; Glibert et al., 2005b). Phytoplankton blooms occur naturally as a consequence of the vertical mixing of deep, nutrient-rich waters into lighted surface waters. This process occurs seasonally in temperate environments due to winter storm events, and due to coastal upwelling events caused by appropriate regional wind conditions. There is no a priori reason why these ‘natural’ sources of nutrients cannot lead to HAB events, but the global increase in frequency and severity of HABs implies that human activities may be an underlying reason for this escalation. Eutrophication of coastal ecosystems is a growing global concern that has clear consequences for blooms of nearshore algal populations (Anderson et al., 2008; Heisler et al., 2008; Howarth, 2008). Nutrient enrichment has been implicated in harmful blooms occurring in some protected bays, but the linkage between nutrient discharges mediated by human activities and many HAB events is still unconfirmed. For example, field studies have shown that coastal upwelling of nitrate-rich waters can be a driving factor leading to toxigenic Pseudo-nitzschia blooms along the U.S. west coast (Horner et al., 2000; Scholin et al., 2000; Anderson et al., 2006) but the specific role of river/coastal runoff in domoic acid production is unclear (Scholin et al., 2000; Schnetzer et al., 2007). The importance of nutrient discharge into coastal waters is, of course, dependent on the amount of nutrients available for phytoplankton growth from natural sources but the latter term is poorly defined in most situations. Constructing nutrient budgets for coastal ecosystems is an area ripe for future work. In the meantime, it has been speculated that anthropogenic nutrient sources, such as elevated nutrient concentrations in river discharge, coastal runoff from agricultural land, and sewage discharge may significantly increase the total amounts of nutrients available for the growth of coastal phytoplankton (Scholin et al., 2000; Glibert et al., 2005a,b, 2006; Howard et al., 2007; Kudela et al., 2008a). There now exists a basic understanding of the general conditions that favor the growth of phytoplankton per se (Allen et al., 2008). Despite this basic understanding, there is still only limited information on the specific conditions that selectively stimulate the growth of harmful algal bloomforming species of phytoplankton. As a result, mathematical models that attempt to predict HAB events tend to be more correlative than deterministic (i.e., they identify the
conditions that may promote a HAB, rather than the conditions that will promote a bloom). One generality is that rarely can one identify a ‘silver bullet’, a single parameter or set of circumstances that provide an accurate prediction of the occurrence of a particular HAB species. The environmental circumstances leading to the dominance of a HAB population over all other species of algae in a given locale are composed of a complex set of physical, chemical and biological conditions with poorly known variances, and these conditions appear to be species-specific. Biological factors contributing to the success or demise of individual HAB taxa include allelopathy among competing phytoplankton, mixotrophy by HAB species, and the deterrence of potential consumers via the production of noxious or toxic compounds (Strom et al., 2003; Burkholder et al., 2008; Buskey, 2008; Flynn, 2008; Smayda, 2008). These biological interactions presuppose a period of stable environmental conditions in order that the scenarios of allelopathy, grazer deterrence or phagotrophic activity of HAB species can play themselves out. This requirement may explain why the formation of a stable water mass appears to play a role in the development of some HAB events (Scholin et al., 2000). Models predicting the population growth of potentially toxic algae are necessary for understanding bloom dynamics, but these models must also integrate information on toxin induction. Many toxins do not appear to be constitutively produced by algae, but are induced by a variety of specific environmental conditions that are not completely understood. Silica, phosphorus, nitrogen and trace metal limitations, and nutrient or elemental ratios (in addition to or instead of absolute concentrations) have all been implicated in toxin induction (Pan et al., 1996a, 1998; Rue and Bruland, 2001; Fehling et al., 2004; Wells et al., 2005; Grane´li and Flynn, 2006; Schnetzer et al., 2007). Again, species-specific differences (and perhaps strain-specific differences) may exist in the factors promoting toxin production. Physical aspects such as temperature and light intensity may stimulate toxin production by some harmful algae (Ono et al., 2000).
Relative fluorescence
3.3.
water research 44 (2010) 385–416
0
0 0
5
10
15
20
25
30
35
40
45
Days Fig. 9 – Correlation between chlorophyll a fluorescence (open squares, in relative fluorescence units) and transmembrane pressure (filled circles) in a pilot-scale reverse osmosis desalination system. Increased loading of phytoplankton biomass resulted in greater transmembrane pressure.
405
water research 44 (2010) 385–416
Pseudo-nitzschia spp. (cells L-1)
2.0x105
1.5x105
1.0x105
5.0x104
0
Particulate domoic acid (µg L-1)
4.0 3.0 2.0 1.0 0.5
0.0 3.0
Dissolved domoic acid (µg L-1)
2.5 2.0 1.5 1.0 0.5
6 ov -0 01 -N
6 01 -Se p-0
01 -Ju l-0 6
ay -06 01 -M
06 ar01 -M
n-0 6 01 -Ja
01 -N ov -05
5 01 -Se p-0
01 -Ju l-0 5
ay -05 01 -M
01 -M
ar-
05
0.0
Fig. 10 – Time series of abundances of Pseudo-nitzschia spp. cells (top), domoic acid concentrations in particulate material (middle) and dissolved in seawater (bottom) at a coastal monitoring site in El Segundo, CA.
4.
Desalination operations and HAB events
A thorough understanding of the specific factors and conditions giving rise to harmful algal blooms and toxin production in coastal waters will require a great deal of additional research before accurate models for predicting these toxic events will be readily available. Until then, appropriate monitoring strategies to detect imminent bloom events and the ability to track the evolution of an active bloom, coupled with an understanding of the potential toxins being produced,
the toxin chemistry, and their rejection by seawater reverse osmosis membranes, provide a seawater desalination facility with the best strategy for making operational adjustments to ensure that the treatment plant capacity or product water quality remains unaffected.
4.1. Concern with harmful algal blooms and their toxin production As seawater desalination has continued to become more costeffective and less energy intensive (Al-Sahlawi, 1999), many
406
water research 44 (2010) 385–416
communities are planning or implementing seawater desalination facilities (Al-Sahlawi, 1999; Burbano et al., 2007). The selected pretreatment procedures and the process engineering that determines the ultimate facility design is entirely dependent upon the source water quality (e.g. suspended solids, turbidity, organic material content, algal cell content, etc.) and its variations, particularly for facilities incorporating open intakes (Bonnelye et al., 2004b; Gaid and Treal, 2007). When the seawater desalination process is performed by reverse osmosis membranes, the selection and proper operation of a pretreatment system is paramount to the success of the downstream desalination process (Tenzer et al., 1999; Bonnelye et al., 2004b; Separation Processes Inc., 2005; Gaid and Treal, 2007; Petry et al., 2007). Algal blooms are known to have significant negative impacts on reverse osmosis desalination facilities. A variety of pretreatment trains have been considered to address the difficult source water quality associated with algal blooms, where the organic and biomass load increase dramatically (Adin and Klein-Banay, 1986; Al Arrayedhy, 1987; Hasan AlSheikh, 1997; Watson, 1997; Abdul Azis et al., 2000; Bonnelye et al., 2004a,b; Burbano et al., 2007; Gaid and Treal, 2007; Petry et al., 2007; Peleka and Matis, 2008). Recently, microfiltration and ultrafiltration membrane pretreatment has been identified as a component of a preferred pretreatment train due to the consistent, high quality water produced by membrane filtration, especially when compared to conventional processes (Wilf and Schierach, 2001). However, one significant drawback to implementation of these modern pretreatment technologies is that they are as susceptible, or possibly more so, to significant algal blooms (Bonnelye et al., 2004a). An early warning system can provide information to a seawater desalination facility so that functional changes can be made to efficiently maintain operations even as source water quality deteriorates. Turbidity sensors offer a rapid measurement of the total amount of suspended particles in the intake water for making these decisions. On in situ instruments such as the Slocum glider, backscatter measurements yield this type of information (Fig. 7d). Measurements of chlorophyll a fluorescence can augment this surveillance approach by estimating the degree to which algal bimoass contributes to the total load of suspended particles. Responses to deteriorating water quality may include chemical additions, use of additional pretreatment equipment, or additional staff preparations (e.g. maintenance activities, guaranteeing all membranes and filters are clean in preparation for an event) to continuously deliver a high quality feedwater to the reverse osmosis system that will produce the desalinated drinking water.
4.2. Addressing spatiotemporal variability in HAB abundance and early detection As detailed above, it is essential that a desalination facility incorporate a means of rapid algal bloom detection so that, when necessary, proper process changes can be made to maintain the production capacity. Sensors for detecting an eminent algal bloom can be located at the desalination facility to inform personnel regarding changes in water quality that are directly observed on the source water. Fig. 9 presents the
transmembrane pressure (TMP) of a microfiltration system that serves as pretreatment to a pilot-scale reverse osmosis desalination system along with the levels of chlorophyll a fluorescence observed in the feedwater. It is clear from this figure that increased membrane fouling rates (e.g. faster daily rise in the TMP) were associated with increasing chlorophyll a fluorescence (i.e. increased algal biomass) in the source water. It is well known that higher concentrations of algae cause increased membrane fouling rates in microfiltration systems that are frequently incorporated, or considered, in today’s desalination facilities (Gijsbertsen-Abrahamse et al., 2006; Lee and Walker, 2006; Reiss et al., 2006). A more complete approach might include a monitoring system located offshore that measures some of the primary factors influencing algal blooms, such as nutrient monitoring in near-real time using new in situ sensor technology (Glibert et al., 2008). Such information would be useful to both the desalination facility and HAB researchers who are continually improving their understanding of the causative factors that produce HABs and their associated toxins. Using the information provided by offshore sensors, the desalination facility personnel could note trends and shifts in driving factors that generate algal blooms and make any chemical orders or perform maintenance procedures that have significant lead times. The same offshore sensor might also incorporate real-time monitors of sentinel parameters for changes in algal biomass, such as turbidity and chlorophyll a, allowing the facility to prepare for changes in chemical additions and redundant equipment service. Monitoring of basic chemical parameters of seawater (e.g. chlorophyll a concentrations) will provide valuable information for facility operations, but this activity is not sufficient to fully assess potentially toxic conditions that might arise from algae that do not require high standing stocks to constitute a significant toxic threat, such as Alexandrium species. Species-specific approaches, such as automated in situ instruments or laboratory-based methods, as well as chemical/immunological analyses that identify and quantify specific algal toxins are necessary to more thoroughly characterize the potential hazards posed by HAB species. The consistent removal of these potentially toxic substances through the reverse osmosis process is both a function of size (e.g. molecular weight) and charge (e.g. zeta potential) (Amy et al., 2005). Depending on the size and charge of the contaminant of concern, the rejection, or removal, by the reverse osmosis process will differ. It is important that we continue to broaden our knowledge on potentially toxic substances excreted by algal stock and their associated blooms. The approach for obtaining this information would be best complemented with knowledge of the species that are present regionally, the potential problems they pose (e.g. specific toxins and the amounts of soluble microbial products and extracellular polymeric substances excreted), the spatial extent of HAB episodes, and their seasonality. The seasonality of Pseudo-nitzschia spp. and domoic acid near the intake of a pilot desalination plant in El Segundo, CA, exemplifies the usefulness of routine monitoring for identifying potentially toxic conditions in coastal waters adjacent to a plant (Fig. 10). Abundances of Pseudo-nitzschia spp. and concentrations of domoic acid contained in the
water research 44 (2010) 385–416
algal cells or dissolved in the intake water exhibited a springtime peak. Knowledge of the seasonality of this toxic bloom-forming species allows intensive sampling of coastal waters during spring when toxic events are more common, improving the overall effectiveness of the monitoring effort and making it more cost effective. Historical and real-time information on the spatial distribution of HABs can provide information vital for optimizing design and performance of desalination operations. Local/ regional hydrography, and resultant algal blooms can differ dramatically. When constructing a new intake pipeline, the selection of its location (e.g. depth and distance from shore) can be greatly enhanced through the use of offshore monitoring devices and efforts to take into account the presence of any local accumulations of algal biomass due to currents, water mass convergences/divergences or internal waves, and also subsurface maxima in algal abundance. Properly locating offshore monitoring can provide significant information that will allow optimal location of a new intake pipeline or identification of issues that might affect an existing one, thereby significantly reducing the organic and suspended solid loads present in the feedwater during algal bloom events. These considerations will ease pretreatment operations, reduce the cost of water production, and help improve the facility’s longevity.
5.
Conclusions
5.1. Potential impacts, unresolved issues and research prospectus The presence of harmful algae in coastal waters that might be employed in reverse osmosis desalination pose potential problems for these operations that have been known to even cause desalination facilities to temporarily cease production (Tenzer et al., 1999; Pankratz, 2008). As the number of seawater desalination facilities continues to grow with lower costs and increasing demand, it is essential that these operating facilities develop the tools necessary to allow process changes and ensure capacity objectives continue to be met. Regardless of the pretreatment configuration, changes in source water quality require adjustments and these changes need to carefully coordinate to ensure that the reverse osmosis membranes are not irreversibly fouled or damaged in the process. Benchmark work is required to establish the effectiveness of the seawater reverse osmosis process in dealing with HAB toxins and other phytoplankton-derived substances. Even if advanced pretreatment technologies such as microfiltration are implemented upstream of the reverse osmosis process, passage of transparent extracellular material produced by the algal bloom (Alldredge et al., 1993; Hong et al., 1997) may affect reverse osmosis membrane performance. Additionally, the physical durability of phytoplankton varies greatly and the pretreatment process might disrupt cells and create significantly higher concentrations of dissolved organic substances, including toxins, than were originally present in the source water. For example, dissolved domoic acid has been observed in the seawater passing through the prefiltration process (Fig. 10, bottom panel) but it is unclear if these values are
407
higher due to cell breakage as a result of the prefiltration process. Therefore, it is important that the international desalination community carefully characterize these potential contaminants and their removal to improve treatment approaches in seawater desalination. To our knowledge, there are no published reports on the effectiveness of reverse osmosis for removing dissolved algal toxins from seawater. Some of these toxin molecules (e.g. domoic acid) are near the theoretical molecular size of molecules rejected by reverse osmosis membranes, but experimental studies are required to validate the effective of this process on toxin removal. In addition, more information will be needed to understand the potential impact of discharged brine and pretreatment backwash water resulting from the reverse osmosis desalination process on the ecology of coastal ecosystems. The use of ferric sulfate or ferric chloride as a pretreatment coagulant would concentrate toxic algae and their associated toxins if they are present in the intake water. Similarly, the discharge of brine resulting from the reverse osmosis process would contain elevated concentrations of dissolved algal toxins relative to unfiltered seawater. The degree of concentration of these toxins would not be expected to be large, but the significance of these processes will depend on the starting concentrations in the raw intake or prefiltered water and the degree of concentration due to treatment. There is presently no information on algal toxins in these discharges. HABs on the U.S. west coast exhibit significant generalities across geographical and temporal scales (e.g. many of the same species occur throughout the region), but the details of bloom dynamics differ with geographic location, depth and season (and perhaps on interannual and decadal scales). The high degree of variability associated with these events makes constant monitoring of HABs in intake water for desalination a vital issue. Regional HAB programs and regulatory agencies along the U.S. west coast presently provide useful information for some known potential problems (e.g. ASP and PSP toxins) for end users that need information on coastal water quality. Awareness (and augmentation) of this information could improve planning and safe operation of desalination facilities. Monitoring of newly emerging HAB concerns (e.g. Cochlodinium spp.), or HABs and toxins that are presently poorly characterized (e.g. NSP, DSP, and yessotoxin poisoning) should also be implemented in the future to allow evaluation of their potential impacts on desalination processes. New technologies for toxin detection and quantification, and in situ monitoring of biological and chemical parameters are rapidly improving our ability to monitor coastal ecosystems and identify potentially problematic situations involving HABs. Advances in in situ observing technologies (sensor networks, autonomous sensor-equipped vehicles) provide the capability for obtaining unprecedented resolution in the spatial and temporal distributions of chemical and physical parameters, and some biologically important features (Sukhatme et al., 2007). New approaches and instruments for toxin detection help identify contaminated seafood products, and constitute sentinels for the threats of HABs to marine animal populations. Future uses of coastal waters for desalination will also benefit from, and contribute to, these activities.
408
water research 44 (2010) 385–416
Acknowlegements The authors are grateful to Mr. Mark Donovan at Separation Processes, Inc. for providing the data for Fig. 9. The preparation of this manuscript was supported in part by funding from a contract between the West Basin Municipal Water District, Department of Water Resources and the University of Southern California, National Oceanic and Atmospheric Administration grants NA05NOS4781228 and NA07OAR4170008, Sea Grant NA07OAR4170008, National Science Foundation grants CCR0120778 (Center for Embedded Networked Sensing; CENS), DDDAS-0540420, MCB-0703159, and a NASA Earth and Space Science Fellowship Grant NNX06AF88H. M.-E`. Garneau was supported by a fellowship from the Fonds que´be´cois de recherche sur la nature et les technologies (FQRNT).
references
Abdul Azis, P.K., Al-Tisan, I., Al-Daili, M., Green, T.N., Dalvi, A.G.I., Javeed, M.A., 2000. Effects of environment on source water for desalination plants on the eastern coast of Saudi Arabia. Desalination 132 (1–3), 29–40. Adams, N.A., Lesoing, M., Trainer, L., 2000. Environmental influences on domoic acid accumulation in razor clams on the Washington coast. Journal of Shellfish Research 19 (2), 1007–1015. Adin, A., Klein-Banay, C., 1986. Pretreatment of seawater by flocculation and settling for particulates removal. Desalination 58 (3), 227–242. Ahn, S., Kulis, D.M., Erdner, D.L., Anderson, D.M., Walt, D.R., 2006. Fiber-optic microarray for simultaneous detection of multiple harmful algal bloom species. Applied and Environmental Microbiology 72 (9), 5742–5749. Al Arrayedhy, M., 1987. Pre- and post-treatment at the RO plant at RA’s Abu Jarjur, Bahrain. Desalination 63 (C), 81–94. Al-Sahlawi, M.A., 1999. Seawater desalination in Saudi Arabia: economic review and demand projections. Desalination 123 (2–3), 143–147. Alfonso, A., de la Rosa, L., Vieytes, M.R., Yasumoto, T., Botana, L.M., 2003. Yessotoxin, a novel phycotoxin, activates phosphodiesterase activity: effect of yessotoxin on cAMP levels in human lymphocytes. Biochemical Pharmacology 65 (2), 193–208. Alldredge, A.L., Passow, U., Logan, B.E., 1993. The abundance and significance of a class of large, transparent organic particles in the ocean. Deep-Sea Research Part I Oceanographic Research Papers 40 (6), 1131–1140. Allen, W.E., 1922. Observations on surface distribution of marine diatoms between San Diego and Seattle. Ecology 3 (2), 140–145. Allen, W.E., 1924. Surface catches of marine diatoms and dinoflagellates made by U.S.S. Pioneer between San Diego and Seattle in 1923. In: University of California Publications in Zoology, vol. 26, pp. 243–248. Allen, W.E., 1928. Review of five years of studies on phytoplankton at Southern California piers, 1920–1924, inclusive. Bulletin of the Scripps Institution of Oceanography Technical Services 1, 357–401. Allen, W.E., 1936. Occurrence of marine plankton diatoms in a tenyear series of daily catches in Southern California. American Journal of Botany 23 (2), 60–63. Allen, W.E., 1940. Summary of results of twenty years of researches on marine phytoplankton. Proceedings of the 6th Pacific Scientific Congress 3, 576–583.
Allen, W.E., 1941. Twenty years’ statistical studies of marine plankton dinoflagellates of Southern California. American Midland Naturalist 26, 603–635. Allen, J.I., Smyth, T.J., Siddom, J.R., Holt, M., 2008. How well can we forecast high biomass algal biomass events in a eutrophic coastal sea? Harmful Algae 8 (1), 70–76. Amy, G., Kim, T.U., Yoon, J., Bellona, C., Drewes, J., Pellegrino, J., Heberer, T., 2005. Removal of micropollutants by NF/RO membranes. Water Science and Technology: Water Supply 5 (5), 25–33. Anderson, C.R., Brzezinski, M.A., Washburn, L., Kudela, R., 2006. Circulation and environmental conditions during a toxigenic Pseudo-nitzschia australis bloom in the Santa Barbara Channel, California. Marine Ecology Progress Series 327, 119–133. Anderson, D.M., Kulis, D.M., Sullivan, J.J., Hall, S., Lee, C., 1990. Dynamics and physiology of saxitoxin production by the dinoflagellates Alexandrium spp. Marine Biology 104 (3), 511–524. Anderson, D.M., Glibert, P.M., Burkholder, J.M., 2002. Harmful algal blooms and eutrophication: nutrient sources, composition, and consequences. Estuaries 25 (4B), 704–726. Anderson, D.M., Kulis, D.M., Keafer, B.A., Gribble, K.E., Marin, R., Scholin, C.A., 2005. Identification and enumeration of Alexandrium spp. from the Gulf of Maine using molecular probes. Deep-Sea Research Part II 52 (19–21), 2467–2490. Anderson, D.M., Burkholder, J.M., Cochlan, W.P., Glibert, P.M., Gobler, C.J., Heil, C.A., Kudela, R.M., Parsons, M.L., Rensel, J.E.J., Townsend, D.W., Trainer, V.L., Vargo, G.A., 2008. Harmful algal blooms and eutrophication: examining linkages from selected coastal regions of the United States. Harmful Algae 8 (1), 39–53. Armonies, W., 1989. Occurrence of meiofauna in Phaeocystis seafoam. Marine Ecology Progress Series 53, 305–309. Armstrong, M., Kudela, R., 2006. Evaluation of California isolates of Lingulodinium polyedrum for the production of yessotoxin. African Journal of Marine Science 28 (2), 399–401. Aune, T., Sørby, R., Yasumoto, T., Ramstad, H., Landsverk, T., 2002. Comparison of oral and intraperitoneal toxicity of yessotoxin towards mice. Toxicon 40 (1), 77–82. Baden, D.G., 1983. Marine food-borne dinoflagellate toxins. International Review of Cytology 83, 99–150. Baden, D.G., 1989. Brevetoxins: unique polyether dinoflagellate toxins. The FASEB Journal 3 (7), 1807–1817. Baden, D.G., Bourdelais, A.J., Jacocks, H., Michelliza, S., Naar, J., 2005. Natural and derivative brevetoxins: historical background, multiplicity and effects. Environmental Health Perspectives 113 (5), 621–625. Barber, R.T., Chavez, F.P., 1983. Biological consequences of El Nin˜o. Science 222 (4629), 1203–1210. Bargu, S., Powell, C.L., Coale, S.L., Busman, M., Doucette, G.J., Silver, M.W., 2002. Krill: a potential vector for domoic acid in marine food webs. Marine Ecology Progres Series 237, 209–216. Bargu, S., Powell, C.L., Wang, Z., Doucette, G.J., Silver, M.W., 2008. Note on the occurrence of Pseudo-nitzschia australis and domoic acid in squid from Monterey Bay, CA (USA). Harmful Algae 7 (1), 45–51. Bates, H.A., Kostriken, R., Rapoport, A., 1978. The occurrence of saxitoxin and other toxins in various dinoflagellates. Toxicon 16 (6), 595–601. Bates, S.S., Trainer, V.L., 2006. The ecology of harmful diatoms. In: Grane´li, E., Turner, J.T. (Eds.), Ecology of Harmful Algae. Springer, Berlin, pp. 81–93. Bates, S.S., Bird, C.J., Defreitas, A.S.W., Foxall, R., Gilgan, M., Hanic, L.A., Johnson, G.R., McCulloch, A.W., Dodense, P., Pocklington, R., Quilliam, M.A., Sim, P.G., Smith, J.C., Rao, D.V. S., Todd, C.D., Walter, J.A., Wrigh, J.L.C., 1989. Pennate diatom Nitzschia pungens as the primary source of domoic acid, a toxin in shellfish from eastern Prince Edwards Island, Canada. Canadian Journal of Fisheries and Aquatic Sciences 46 (7), 1203–1215.
water research 44 (2010) 385–416
Bates, S.S., DeFreitas, A.S.W., Milley, J.E., Pocklington, R., Quilliam, M.A., Smith, J.C., Worms, J., 1991. Controls on domoic acid production by the diatom Nitzschia pungens f. multiseries in cultures: nutrients and irradiance. Canadian Journal of Fisheries and Aquatic Sciences 48 (7), 1136–1144. Baugh, K.A., Bush, J.M., Bill, B.D., Lefebvre, K.A., Trainer, V.L., 2006. Estimates of specific toxicity in several Pseudo-nitzschia species from the Washington coast, based on culture and field studies. African Journal of Marine Science 28 (2), 403–407. Bialojan, C., Takai, A., 1988. Inhibitory effect of a marine-sponge toxin, okadaic acid, on protein phosphatases. Specificity and kinetics. Biochemical Journal 256 (1), 283–290. Bigelow, H.B., Leslie, M., 1930. Reconnaissance of the waters and plankton of Monterey Bay July 1928. Bulletin of the Museum of Comparative Zoology 70 (5), 430–581. Blasco, D., 1978. Observations on the diel migration of marine dinoflagellates off the Baja Californian coast. Marine Biology 46 (1), 41–47. Bolin, R.L., Abbott, D.P., 1960. Studies on the marine climate and phytoplankton of the central coastal area of California, 1954–1960. California Cooperative Oceanic Fisheries Investigations 9, 23–45. Boni, L., Ceredi, A., Guerrini, F., Milandri, A., Pistocchi, R., Poletti, R., Pompei, M., 2002. Toxic Gonyaulax Grindley Reinecke in the North-Western Adriatic Sea (Italy). In: Tenth International Conference on Harmful Algal Blooms, St. Petersburg, FL (abstract). Bonnelye, V., Snaz, M.A., Comte, C., Colas, F., Plasse, L., Gueguen, F., 2004a. Ultrafiltration versus conventional pre-treatment upstream RO: advantages and limits, Seoul, Korea. Bonnelye, V., Sanz, M.A., Durand, J.P., Plasse, L., Gueguen, F., Mazounie, P., 2004b. Reverse osmosis on open intake seawater: pre-treatment strategy. Desalination 167 (1–3), 191–200. Bourdelais, A.J., Tomas, C.R., Naar, J., Kubanek, J., Baden, D.G., 2002. New fish-killing alga in coastal Delaware produces neurotoxins. Environmental Health Perspectives 110 (5), 465–470. Bowers, H.A., Tengs, T., Glasgow Jr., H.B., Burkholder, J.M., Rublee, P.A., Oldach, D.W., 2000. Development of real-time PCR assays for rapid detection of Pfiesteria piscicida and related dinoflagellates. Applied and Environmental Microbiology 66 (11), 4641–4648. Bowers, H.A., Tomas, C., Tengs, T., Kempton, J.W., Lewitus, A.J., Oldach, D.W., 2006. Raphidophyceae (Chadefaud ex Silva) systematics and rapid identification: sequence analyses and real-time PCR assays. Journal of Phycology 42 (6), 1333–1348. Bricelj, V.M., Lonsdale, D.J., 1997. Aureococcus anophagefferens: causes and ecological consequences of brown tides in U.S. mid-Atlantic coastal waters. Limnology and Oceanography 42 (5), 1023–1038. Buck, K.R., Uttal-Cooke, L., Pilskaln, C.H., Roelke, D.L., Villac, M.C., Fryxell, G.A., Cifuentes, L., Chavez, F.P., 1992. Autecology of the diatom Pseudo-nitzschia australis, a domoic acid producer, from Monterey Bay, California. Marine Ecology Progress Series 84 (3), 293–302. Burbano, A.A., Adham, S.S., Pearce, W.R., 2007. The state of fullscale RO/NF desalination - Results from a worldwide survey. Journal of American Water Works Association 99 (4), 116–127. Burkholder, J.M., Azanza, R.V., Sako, Y., 2006. The ecology of harmful dinoflagellates. In: Grane´li, E., Turner, J.T. (Eds.), Ecology of Harmful Algae. Springer, Berlin, pp. 53–66. Burkholder, J.M., Hallegraeff, G.M., Melia, G., Cohen, A., Bowers, H.A., Oldach, D.W., Parrow, M.W., Sullivan, M.J., Zimba, P.V., Allen, E.H., Kinder, C.A., Mallin, M.A., 2007. Phytoplankton and bacterial assemblages in ballast water of U.S. military ships as a function of port of origin, voyage time, and ocean exchange practices. Harmful Algae 6 (4), 486–518.
409
Burkholder, J.M., Glibert, P.M., Skelton, H.M., 2008. Mixotrophy, a major mode of nutrition for harmful algal species in eutrophic waters. Harmful Algae 8 (1), 77–93. Buskey, E.J., 2008. How does eutrophication affect the role of grazers in harmful algal bloom dynamics? Harmful Algae 8 (1), 152–157. Busse, L.B., Venrick, E.L., Antrobus, R., Miller, P.E., Vigilant, V., Silver, M.W., Mengelt, C., Mydlarz, L., Prezelin, B.B., 2006. Domoic acid in phytoplankton and fish in San Diego, CA, USA. Harmful Algae 5 (1), 91–101. Caron, D.A., Dennett, M.R., Moran, D.M., Schaffner, R.A., Lonsdale, D. J., Gobler, C.J., Nuzzi, R., McLean, T.I., 2003. Development and application of a monoclonal-antibody technique for counting Aureococcus anophagefferens, an alga causing recurrent brown tides in the Mid-Atlantic United States. Applied and Environmental Microbiology 69 (9), 5492–5502. Carpenter, E.J., Janson, S., Boje, R., Pollehne, F., Chang, J., 1995. The dinoflagellate Dinophysis norvegica: biological and ecological observations in the Baltic Sea. European Journal of Phycology 109. Casper, E.T., Patterson, S.S., Bhanushali, P., Farmer, A., Smith, M., Fries, D.P., Paul, J.H., 2007. A handheld NASBA analyzer for the field detection and quantification of Karenia brevis. Harmful Algae 6 (1), 112–118. Catterall, W.A., 1992. Cellular and molecular biology of voltagegated sodium channels. Physiological Reviews 72 (Suppl. 4.), S15–S48. CDPH, 2007. Preharvest Shellfish Protection and Marine Biotoxin Monitoring Program. California Department of Public Health. Cembella, A.D., 1989. Occurrence of okadaic acid, a major diarrheic shellfish toxin, in natural populations of Dinophysis spp. from the eastern coast of North America. Journal of Applied Phycology 1 (4), 307–310. Cembella, A.D., 2003. Chemical ecology of eukaryotic microalgae in marine ecosystems. Phycologica 42 (4), 420–447. Cestele, S., Catterall, W.A., 2000. Molecular mechanisms of neurotoxin action on voltage-gated sodium channels. Biochimie 82 (9–10), 883–892. Chavez, F.P., Ryan, J., Lluch-Cota, S.E., Niquen, M., 2003. From anchovies to sardines and back: multidecadal change in the Pacific ocean. Science 299 (5604), 217–221. Ciminiello, P., Fattorusso, E., Forino, M., Poletti, R., Viviani, R., 2000. Structure determination of carboxyhomoyessotoxin, a new yessotoxin analogue isolated from Adriatic mussels. Chemistry Research in Toxicology 13 (8), 770–774. Ciminiello, P., Fattorusso, E., Forino, M., Magno, S., Poletti, R., 2001. 42,43,44,45,46,47,55-Heptanor-41-oxohomoyessotoxin, a new biotoxin from mussels of the Northern Adriatic Sea. Chemical Research in Toxicology 14 (5), 596–599. Ciminiello, P., Fattorusso, E., Forino, M., Magno, S., Poletti, R., Viviani, R., 1998. Isolation of adriatoxin a new analogue of yessotoxin from mussels of the Adriatic Sea. Tetrahedron Letters 39 (48), 8897–8900. Clayden, J., Read, B., Hebditch, K., 2005. Chemistry of domoic acid, isodomoic acids, and their analogues. Tetrahedron 61 (24), 5713–5724. Cochlan, W.P., Herndon, J., Kudela, R.M., 2008. Inorganic and organic nitrogen uptake by the toxigenic diatom Pseudo-nitzschia australis (Bacillariophyceae). Harmful Algae 8 (1), 111–118. Cooley, H., Gleick, P.H., Wolff, G., 2006. Desalination, with a grain of salt: a California perspective. In: Hart, I. (Ed.). Pacific Institute for Studies in Development, Environment, and Security, Hayward, CA, p. 88. Coyne, K.J., Hutchins, D.A., Hare, C.E., Cary, S.C., 2001. Assessing temporal and spatial variability in Pfiesteria piscicida distributions using molecular probing techniques. Aquatic Microbial Ecology 24 (3), 275–285.
410
water research 44 (2010) 385–416
Coyne, K.J., Handy, S.M., Demir, E., Whereat, E.B., Hutchins, D.A., Portune, K.J., Doblin, M.A., Cary, S.C., 2005. Improved quantitative real-time PCR assays for enumeration of harmful algal species in field samples using an exogenous DNA reference standard. Limnology and Oceanography Methods 3, 381–391. Cullen, J.J., Horrigan, S.G., 1981. Effects of nitrate on the diurnal vertical migration, carbon to nitrogen ratio, and the photosynthetic capacity of the dinoflagellate Gymnodinium splendens. Marine Biology 62 (2–3), 81–89. Curtiss, C.C., Langlois, G.W., Busse, L.B., Mazzillo, F., Silve, M.W., 2008. The emergence of Cochlodinium along the California Coast (USA). Harmful Algae 7 (3), 337–346. Daiguji, M., Satake, M., Ramstad, H., Aune, T., Naoki, H., Yasumoto, T., 1998. Structure and fluorometric HPLC determination of 1-desulfoyessotoxin, a new yessotoxin analog isolated from mussels from Norway. Natural Toxins 6 (6), 235–239. Dekshenieks, M.M., Donaghay, P.L., Sullivan, J.M., Rines, J.E.B., Osborn, T.R., Twardowski, M.S., 2001. Temporal and spatial occurrence of thin phytoplankton layers in relation to physical processes. Marine Ecology Progress Series 223, 61–71. Demir, E., Coyne, K.J., Doblin, M.A., Handy, S.M., Hutchins, D.A., 2008. Assessment of microzooplankton grazing on Heterosigma akashiwo using a species-specific approach combining quantitative real-time PCR (QPCR) and dilution methods. Microbial Ecology 55 (4), 583–594. Determan, T., 2003. Paralytic shellfish poisoning (PSP) patterns in Puget Sound shellfish in 2001-A report for the Puget Sound Ambient Monitoring Program Office of Food Safety and Shellfish Programs. Washington State Department of Health, Olympia. Draisci, R., Ferretti, E., Palleschi, L., Marchiafava, C., Poletti, R., Milandri, A., Ceredi, A., Pompei, M., 1999a. High levels of yessotoxin in mussels and presence of yessotoxin and homoyessotoxin in dinoflagellates of the Adriatic Sea. Toxicon 37, 1187–1193. Draisci, R., Palleschi, L., Giannetti, L., Lucentini, L., James, K.J., Bishop, A.G., Satake, M., Yasumoto, T., 1999b. New approach to the direct detection of known and new diarrhetic shellfish toxins in mussels and phytoplankton by liquid chromatography-mass spectrometry. Journal of Chromatrography 847 (1–2), 213–221. Eiki, K., Satake, M., Koike, K., Ogata, T., Mitsuya, T., Oshima, Y., 2005. Confirmation of yessotoxin production by the dinoflagellate Protoceratium reticulatum in Mutsu Bay. Fisheries Science 71 (3), 633–638. Eppley, R.W., Holm-Harisen, O., Strickland, J.D.H., 1968. Some observations on the vertical migration of dinoflagellates. Journal of Phycology 4 (4), 333–340. European Commission, 2002. Directive Commission Decision, No. C2002 1001. Official Journal of the European Communities March 15, 2002, 1.75/63, 16.3. Fehling, J., Davidson, K., Bolch, C.J., Bates, S.S., 2004. Growth and domoic acid production by Pseudo-nitzschia seriata (Bacillariophyceae) under phosphate and silicate limitation. Journal of Phycology 40 (4), 674–683. Flynn, K.J., 2002. Toxin production in migrating dinoflagellates: a modelling study of PSP producing Alexandrium. Harmful Algae 1 (2), 147–155. Flynn, K.J., 2008. Attack is not the best form of defense: lessons from harmful algal bloom dynamics. Harmful Algae 8 (1), 129–139. Flynn, K., Franco, J.M., Fernandez, P., Reguera, B., Zapata, M., Wood, G., Flynn, K.J., 1994. Changes in toxin content, biomass and pigments of the dinoflagellate Alexandrium minutum during nitrogen refeeding and growth into nitrogen or phosphorus stress. Marine Ecology Progres Series 111, 99–109. Franchini, A., Marchesini, E., Poletti, R., Ottaviani, E., 2004. Lethal and sub-lethal yessotoxin dose-induced morpho-functional
alterations in intraperitoneal injected Swiss CD1 mice. Toxicon 44 (1), 83–90. Fritz, L., Quilliam, M.A., Wright, J.L.C., Beale, A.M., Work, T.M., 1992. An outbreak of domoic acid poisoning attributed to the pennate diatom Pseudo-nitzschia australis. Journal of Phycology 28 (4), 439–442. Fryxell, G.A., Villac, M.C., Shapiro, L.P., 1997. The occurrence of the toxic diatom genus Pseudo-nitzschia (Bacillariophyceae) on the West Coast of the USA, 1920–1996: a review. Phycologia 36 (6), 419–437. Gaid, K., Treal, Y., 2007. Le dessalement des eaux par osmose inverse: l’expE`rience de VE`olia Water. Desalination 203 (1–3), 1–14. Gaines, G., Taylor, F.J.R., 1985. An exploratory analysis of PSP patterns in British Columbia: 1942–1984. In: Anderson, D.M., White, A.W., Baden, D.G. (Eds.), Toxic dinoflagellates. Elsevier Science, New York, pp. 439–444. Galluzzi, L., Penna, A., Bertozzini, E., Vila, M., Garces, E., Magnani, M., 2004. Development of a real-time PCR assay for rapid detection and quantification of Alexandrium minutum (a dinoflagellate). Applied and Environmental Microbiology 70 (2), 1199–1206. Garrison, D.L., Conrad, S.M., Eilers, P.P., Waldron, E.M., 1992. Confirmation of domoic acid production by Pseudo-nitzschia australis (Bacillariophyceae) cultures. Journal of Phycology 28 (5), 604–607. Garthwaite, I., Ross, K.M., Miles, C.O., Hansen, R.P., Foster, D., Wilkins, A.L., Towers, N.R., 1998. Polyclonal antibodies to domoic acid, and their use in immunoassays for domoic acid in sea water and shellfish. Natural Toxins 6 (3–4), 93–104. Garthwaite, I., Ross, K.M., Miles, C.O., Briggs, L.R., Towers, N.R., Borrell, T., Busby, P., 2001. Integrated Enzyme-Linked Immunosorbent Assay screening system for amnesic, neurotoxic, diarrhetic, and paralytic shellfish poisoning toxins found in New Zealand. Journal of AOAC International 84 (5), 1643–1648. Geraci, J.R., Anderson, D.M., Timperi, R.J., St. Aubin, D.J., Early, G.A., Prescott, J.H., Mayo, C., 1989. Humpback whales (Megaptera novaeangliae) fatally poisoned by dinoflagellate toxin. Canadian Journal of Fisheries and Aquatic Sciences 46 (11), 1895–1898. Gijsbertsen-Abrahamse, A.J., Schmidt, W., Chorus, I., Heijman, S.G.J., 2006. Removal of cyanotoxins by ultrafiltration and nanofiltration. Journal of Membrane Science 276 (1–2), 252–259. Glibert, P.M., Anderson, D.M., Gentien, P., Grane´li, E., Sellner, K.G., 2005a. The global, complex phenomena of harmful algal blooms. Oceanography 18 (2), 137–147. Glibert, P.M., Seitzinger, S., Heil, C.A., Burkholder, J.M., Parrow, M.W., Codispoti, L.A., Kelly, V., 2005b. The role of eutrophication in the global proliferation of harmful algal blooms. Oceanography 18 (2), 198–209. Glibert, P.M., Harrison, J., Heil, C., Seitzinger, S., 2006. Escalating worldwide use of urea – a global change contributing to coastal eutrophication. Biogeochemistry 77 (3), 441–463. Glibert, P.M., Kelly, V., Alexander, J., Cosdispoti, L.A., Boicourt, W.C., Trice, T.M., Michael, B., 2008. In situ nutrient monitoring: a tool for capturing nutrient variability and the antecedent conditions that support algal blooms. Harmful Algae 8 (1), 175–181. Grane´li, E., Flynn, K., 2006. Chemical and physical factors influencing toxin content. In: Grane´li, E., Turner, J.T. (Eds.), Ecology of Harmful Algae. Springer, Berlin, pp. 229–241. Greenfield, D.I., Marin, R., Jensen, S., Massion, E., Roman, B., Feldman, J., Scholin, C., 2006. Application of the Environmental Sample Processor (ESP) methodology for quantifying Pseudonitzschia australis using ribosomal RNA-targeted probes in sandwich and fluorescent in situ hybridization. Limnology and Oceanography Methods 4, 426–435.
water research 44 (2010) 385–416
Gregorio, D.E., Connell, L., 2000. Range of Heterosigma akashiwo (Raphidophyceae) expanded to include California, USA. In: Ninth International Conference on Harmful Algal Blooms, Tasmania, Australia (abstract). Gregorio, D.E., Pieper, R.E., 2000. Investigations of red tides along the southern California. Bulletin of Southern California Academy of Sciences 99 (8), 147–160. Gulland, F.M.D., Haulena, M., Fauquier, D., Langlois, G., Lander, M.E., Zabka, T., Duerr, R., 2002. Domoic acid toxicity in California sea lions (Zalophus californianus): clinical signs, treatment and survival. Veterinary Record 150 (15), 475–480. Hallegraeff, G., Hara, Y., 1995. Taxonomy of harmful marine raphidophytes. In: Hallegraef, G., Anderson, D.M., Cembella, A.D. (Eds.), Manual on Harmful Marine Microalgae. IOC UNCESCO, Paris, pp. 365–371. Hallegraeff, G.M., 1993. A review of harmful algal blooms and their apparent global increase. Phycologia 32 (2), 79–99. Hallegraeff, G.M., 2003. Harmful algal blooms: a global overview. In: Hallegraeff, G.M., Anderson, D.M., Cembella, A.D. (Eds.), Manual on Harmful Marine Microalgae. Monographs on Oceanographic Methodology. UNESCO Publishing, Paris pp. 25–49. Hallegraeff, G.M., Bolch, C.J., 1992. Transport of diatom and dinoflagellate resting spores in ships’ ballast water: implications for plankton biogeography and aquaculture. Journal of Plankton Research 14 (8), 1067–1084. Handy, S.M., Hutchins, D.A., Cary, S.C., Coyne, K.J., 2006. Simultaneous enumeration of multiple raphidophyte species by multiprobing and mutiplexing: capabilities and limitations. Limnology and Oceanography: Methods 4, 193–204. Haque, S.M., Onoue, Y., 2002a. Effects of salinity on growth and toxin production of a noxious phytoflagellate, Heterosigma akashiwo (Raphidophyceae). Botanica Marina 45 (4), 356–363. Haque, S.M., Onoue, Y., 2002b. Variation in toxin compositions of two harmful raphidophytes, Chattonella antiqua and Chattonella marina, at different salinities. Environmental Toxicology 17 (2), 113–118. Hard, J.J., Connell, L., Hershberger, W.K., Harrell, L.W., 2000. Genetic variation in mortality of chinook salmon during a bloom of the marine alga Heterosigma akashiwo. Journal of Fish Biology 56 (6), 1387–1397. Hasan Al-Sheikh, A.H., 1997. Seawater reverse osmosis pretreatment with an emphasis on the Jeddah Plant operation experience. Desalination 110 (1–2), 183–192. Haystead, T.A.J., Sim, A.T.R., Carling, D., Honnor, R.C., Tsukitani, Y., Cohen, P., Hardie, D.G., 1989. Effects of the tumour promoter okadaic acid on intracellular protein phosphorylation and metabolism. Nature 337, 78–81. Heisler, J., Glibert, P.M., Burkholder, J.M., Anderson, D.M., Cochlan, W., Dennison, W.C., Dortch, Q., Gobler, C.J., Heil, C.A., Humphries, E., Lewitus, A., Magnien, R., Marshall, H.G., Sellner, K., Stockwell, D.A., Stoecker, D.K., Suddleson, M., 2008. Eutrophication and harmful algal blooms: a scientific consensus. Harmful Algae 8 (1), 3–13. Herndon, J., Cochlan, W.P., Horner, R., 2003. Heterosigma akashiwo blooms in San Francisco Bay. Interagency Ecological Program for the San Francisco Estuary Newsletter 16, 46–48. Hershberger, P.K., Rensel, J.E., Postel, J.R., Taub, F.B., 1997. Heterosigma bloom and associated fish kill. Harmful Algae News 16 (1), 4. Holmes, R.W., Williams, P.M., Eppley, R.W., 1967. Red water in La Jolla, 1964–1966. Limnology and Oceanography 12 (3), 503–512. Hong, Y., Smith Jr., W.O., White, A.M., 1997. Studies on transparent exopolymer particles (TEP) produced in the Ross Sea (Antarctica) and by Phaeocystis antarctica (Prymnesiophyceae). Journal Of Phycology 33 (3), 368–376.
411
Hori, M., Matsuura, Y., Yoshimoto, R., Ozaki, H., Yasumoto, T., Karaki, H., 1999. Actin depolymerizing action by marine toxin, pectenotoxin-2. Folia Pharmacologica Japonica 114 (Suppl. 1), 225–229. Horner, R.A., Postel, J.R., 1993. Toxic diatoms in western Washington waters (U.S. west coast). Hydrobiologia 269/270 (1), 197–205. Horner, R.A., Garrison, D.L., Plumley, F.G., 1997. Harmful algal blooms and red tide problems on the U.S. west coast. Limnology and Oceanography 42 (2), 1076–1088. Horner, R.A., Hickey, B.M., Postel, J.R., 2000. Pseudo-nitzschia blooms and physical oceanography off Washington state. South African Journal of Marine Science 22 (1), 299–308. Howard, M.D.A., Cochlan, W.P., Ladizinsky, N., Kudela, R.M., 2007. Nitrogenous preference of toxigenic Pseudo-nitzschia australis (Bacillariophyceae) from field and laboratory experiments. Harmful Algae 6 (2), 206–217. Howard, M.D.A., Silvera, M., Kudela, R.M., 2008. Yessotoxin detected in mussel (Mytilus californicus) and phytoplankton samples from the U.S. west coast. Harmful Algae 7 (5), 646–652. Howard, M.D., Smith, G.J., Kudela, R.M., 2009. Phylogenetic relationships of yessotoxin-producing dinoflagellates, based on the large subunit and internal transcribed spacer ribosomal DNA domains. Applied and Environmental Microbiology 75 (1), 54–63. Howarth, R.W., 2008. Coastal nitrogen pollution: a review of sources and trends globally and regionally. Harmful Algae 8 (1), 14–20. IOC HAB Programme, 2008. IOC Harmful Algal Bloom Website. Intergovernmental Oceanic Commission of UNESCO. Iwataki, M., Kawami, H., Matsuoka, K., 2007. Cochlodinium fulvescens sp nov (Gymnodiniales, Dinophyceae), a new chainforming unarmored dinoflagellate from Asian coasts. Phycological Research 55 (3), 231–239. Iwataki, M., Kawami, H., Mizushima, K., Mikulski, C.M., Doucette, G.J., Relox, J.R., Anton, A., Fukuyo, Y., Matsuoka, K., 2008. Phylogenetic relationships in the harmful dinoflagellate Cochlodinium polykrikoides (Gymnodiniales, Dinophyceae) inferred from LSU rDNA sequences. Harmful Algae 7 (3), 271–277. Jeffery, B., Barlow, T., Moizer, K., Paul, S., Boyle, C., 2004. Amnesic shellfish poison. Food and Chemical Toxicology 42 (4), 545–557. 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. Jester, R., 2008. An investigation into the prevalence of Alexandrium derived toxins in marine food webs. Ph.D., University of California, Santa Cruz, Santa Cruz, pp. 116. Jester, R., Lefebvre, K.A., Langlois, G., Vigilant, V., Baugh, K.A., Silver, M.W., 2009a. A shift in the dominant toxin-producing algal species in central California alters phycotoxins in food webs. Harmful Algae 8 (2), 291–298. Jester, R.J., Baugh, K.A., Lefebvre, K.A., 2009b. Presence of Alexandrium catenella and paralytic shellfish toxins in finfish, shellfish and rock crabs in Monterey Bay, California, USA. Marine Biology 156 (3), 493–504. Johansson, N., Grane´li, E., Yasumoto, T., Carlsson, P., Legrand, C., 1996. Toxin production by Dinophysis acuminata and D. acuta cells grown under nutrient sufficient and deficient conditions. In: Yasumoto, T., Oshima, Y., Fukuyo, Y. (Eds.), Harmful and Toxic Algal Blooms. Intergovernmental Oceanographic Commission of UNESCO, Paris, pp. 277–279. John, E.H., Flynn, K.J., 2000. Growth dynamics and toxicity of Alexandrium fundyense (Dinophyceae): the effect of changing N:
412
water research 44 (2010) 385–416
P supply ratios on internal toxin and nutrient levels. European Journal of Phycology 35 (1), 11–23. Jones, B.H., Noble, M.A., Dickey, T.D., 2002. Hydrographic and particle distributions over the Palos Verdes continental shelf: spatial, seasonal and daily variability. Continental Shelf Research 22 (6–7), 945–965. Kirkpatrick, B., Fleming, L.E., Squicciarini, D., Backer, L.C., Clark, R., Abraham, W., Benson, J., Cheng, Y.S., Johnson, D., Pierce, R., Zaias, J., Bossart, G.D., Baden, D.G., 2004. Literature review of Florida red tide: implications for human health effects. Harmful Algae 3 (2), 99–115. Kudela, R.M., Cochlan, W.P., 2000. Nitrogen and carbon uptake kinetics and the influence of irradiance for a red tide bloom off southern California. Aquatic Microbial Ecology 21 (1), 31–47. Kudela, R.M., Roberts, A., Armstrong, M., 2004. Laboratory analyses of nutrient stress and toxin production in Pseudonitzschia spp. from Monterey Bay, California. In: Steidinger, K., Landsberg, J., Tomas, C., Vargo, G. (Eds.). Florida Environmental Research Institute and UNESCO, St. Petersberg, pp. 136–138. Kudela, R.M., Lane, J.Q., Cochlan, W.P., 2008a. The potential role of anthropogenically derived nitrogen in the growth of harmful algae in California, USA. Harmful Algae 8 (1), 103–110. Kudela, R.M., Ryan, J.P., Blakely, M.D., Lane, J.Q., Peterson, T.D., 2008b. Linking the physiology and ecology of Cochlodinium to better understand harmful algal bloom events: a comparative approach. Harmful Algae 7 (3), 278–292. Kusek, K.M., Vargo, G., Steidinger, K., 1999. Gymnodinium breve in the field in the lab and in the newspaper - scientific and journalistic analysis of Florida red tides. Contributions in Marine Science 34, 1–229. Ladizinsky, N., Smith, G.J., 2000. Accumulation of domoic acid by the coastal diatom Pseudo-nitzschia multiseries: a possible copper complexation strategy. Journal of Phycology 36, 41. Landry, M.R., 1989. Broad-scale distributional patterns of hydrographic variables on the Washington/Oregon shelf. In: Landry, M.R., Hickey, B.M. (Eds.), Coastal Oceanography of Washington and Oregon. Elsevier, New York, pp. 1–40. Lange, C.B., Reid, F.M.H., Vernet, M., 1994. Temporal distribution of the potentially toxic diatom Pseudo-nitzschia australis at a coastal site in Southern California. Marine Ecology Progress Series 104 (3), 309–312. Langlois, G., 2007. Marine Biotoxin Monitoring Program Annual Report. California Department of Public Health for California Department of Fish and Game, Sacramento. Lee, J., Walker, H.W., 2006. Effect of process variables and natural organic matter on removal of microcystin-LR by PAC-UF. Environmental Science and Technology 40 (23), 7336–7342. Lee, J.S., Yanagi, T., Kenma, R., Yasumoto, T., 1987. Fluorometric determination of diarrhetic shellfish toxins by highperformance liquid chromatography. Agricultural and Biological Chemistry 51 (3), 877–881. Lee, J.S., Igarashi, T., Fraga, S., Dah, E., Hovgaard, P., Yasumoto, T., 1989. Determination of diarrhetic toxins in various dinoflagellate species. Journal of Applied Phycology 1 (2), 147–152. Lefebvre, K.A., Bill, B.D., Erickson, A., Baugh, K.A., O’Rourke, L., Costa, P.R., Nance, S., Trainer, V.L., 2008. Characterization of intracellular and extracellular saxitoxin levels in both field and cultured Alexandruim spp. samples from Sequim Bay, Washington. Marine Drugs 6, 103–116. Levandowsky, M., Kaneta, P.J., 1987. Behaviour in dinoflagellates. In: Taylor, F.J.R. (Ed.), The Biology of Dinoflagellates. Blackwell, Oxford, pp. 360–397. Lin, Y.Y., Risk, M., Ray, S.M., Van Engen, D., Clardy, J., Golik, J., James, J.C., Nakanishi, K., 1981. Isolation and structure of brevetoxin B from the ‘red tide’ dinoflagellate Ptychodiscus
brevis (Gymnodinium breve). Journal of American Chemical Society 103 (22), 6773–6775. Llewellyn, L.E., 2006. Saxitoxin, a toxic marine natural product that targets a multitude of receptors. Natural Product Reports 23 (2), 200–222. Loeblich III, A.R., Fine, K.E., 1977. Marine chloromonads: More widely distributed in neritic environments than previously thought. Proceedings of the Biological Society of Washington 90 (2), 388–399. ˜ ., 1997. Studies Lundholm, N., Skov, J., Pocklington, R., Moestrup, A on the marine planktonic diatom Pseudo-nitzschia. 2. Autecology of P. pseudodelicatissima based on isolates from Danish coastal waters. Phycologia 36, 381–388. Malagoli, D., Marchesini, E., Ottaviani, E., 2006. Lysosomes as the target of yessotoxin invertebrate and vertebrate cell lines. Toxicology Letters 167 (1), 75–83. Maldonado, M.T., Hughes, M.P., Rue, E.L., Wells, M.L., 2002. The effect of Fe and Cu on growth and domoic acid production by Pseudo-nitzschia multiseries and Pseudo-nitzschia australis. Limnology and Oceanography 47 (2), 515–526. Marchetti, A., Trainer, V.L., Harrison, P.J., 2004. Environmental conditions and phytoplankton dynamics associated with Pseudo-nitzschia abundance and domoic acid in the Juan de Fuca eddy. Marine Ecology Progress Series 281, 1–12. Martin, J.L., Haya, K., Burridge, L.E., Wildish, D.J., 1990. Nitzschia pseudodelicatissima - a source of domoic acid in the Bay of Fundy, eastern Canada. Marine Ecology Progress Series 67 (2), 177–182. Matsuoka, K., Iwataki, M., Kawami, H., 2008. Morphology and taxonomy of chain-forming species of the genus Cochlodinium (Dinophyceae). Harmful Algae 7 (3), 261–270. McGillicuddy Jr., D.J., Signell, R.P., Stock, C.A., Keafer, B.A., Keller, M.D., Ehtland, R.D., Anderson, D.M., 2005. A mechanisms for offshore initiations of harmful algal blooms in the coastal Gulf of Maine. Journal of Plankton Research 25, 1131–1138. Mengelt, C., 2006. Ultraviolet Photoecology, Dark Survival, and Seasonal Abundance of Pseudo-nitzschia australis and P. multiseries in Coastal Waters of Central California. Dissertation, University of California, Santa Barbara. Mikulski, C.M., Morton, S.L., Doucette, G.J., 2005. Development and application of LSU rRNA probes for Karenia brevis in the Gulf of Mexico, USA. Harmful Algae 4 (1), 49–60. Mikulski, C.M., Park, Y.T., Jones, K.L., Lee, C.K., Lim, W.A., Lee, Y., Scholin, C.A., Doucette, G.J., 2008. Development and field application of rRNA-targeted probes for the detection of Cochlodinium polykrikoides Margalef in Korean coastal waters using whole cell and sandwich hybridization formats. Harmful Algae 7 (3), 347–359. Miles, C.O., Aasen, J., Dahl, E., Samdal, I., Aune, T., Briggs, L.R., Olseng, C.D., Edvardsen, B., Naustvoll, L.J., 2002. Yessotoxin in Norwegian Blue Mussels caused by Protoceratium reticulatum. In: Tenth International Conference on Harmful Algal Blooms, St. Petersburg, FL (abstract). Miles, C.O., Wilkins, A.L., Samdal, I.A., Sandvik, M., Petersen, D., Quilliam, M.A., Nausvoll, L.J., Rundberget, T., Torgersen, T., Hovgaard, P., Jensen, D.J., Cooney, J.M., 2004. A novel pectenotoxin, PTX-12, in Dinophysis spp. and shellfish from Norway. Chemical Research and Toxicology 17 (11), 1423–1433. Miles, C.O., Wilkins, A.L., Hawkes, A.D., Selwood, A.I., Jensen, D.J., Munday, R., Cooney, J.M., Beuzenberg, V., 2005a. Polyhydroxylated amide analogs of yessotoxin from Protoceratium reticulatum. Toxicon 45 (1), 61–71. Miles, C.O., Samdal, I.A., Aasen, J.A.G., Jensen, D.J., Quilliam, M.A., Petersen, D., Briggs, L.M., Wilkins, A.L., Rise, F., Cooney, J.M., MacKenzie, A.L., 2005b. Evidence for numerous analogs of
water research 44 (2010) 385–416
yessotoxin in Protoceratium reticulatum. Harmful Algae 4 (6), 1075–1091. Miles, C.O., Wilkins, A.L., Hawkes, A.D., Selwood, A.I., Jensen, D.J., Cooney, J.M., Beuzenberg, V., MacKenzie, L., 2006. Identification of 45-hydroxy-46,47-dinoryessotoxin, 44-oxo45,46,47-trinoryessotoxin, and 9-methyl-42,43,44,45,46,47,55hepta-nor-38-en-41-oxoyessotoxin, and partial characterization of some minor yessotoxins, from Protoceratium reticulatum. Toxicon 47 (2), 229–240. Miller, P.E., Scholin, C.A., 1998. Identification and enumeration of cultured and wild Pseudo-nitzschia (Bacillariophyceae) using species-specific LSU rRNA-targeted fluorescent probes and filter-based whole cell hybridization. Journal of Phycology 34 (2), 371–382. Mitrovic, S.M., Ferna´ndez Amandi, M., MacKenzie, L., Furey, A., James, K.J., 2004. Effects of selenium, iron and cobalt addition to growth and yessotoxin production of the toxic marine dinoflagellate Protoceratium reticulatum in culture. Journal of Experimental Marine Biology and Ecology 313 (2), 337–351. Mitrovic, S.M., Hamilton, B., MacKenzie, L., Furey, A., James, K.J., 2005. Persistence of yessotoxin under light and dark conditions. Marine Environmental Research 60 (3), 397–401. Montojo, U.M., Sakamoto, S., Cayme, M.F., Gatdula, N.C., Furio, E. F., Relox, J.R.J., Sato, S., Fukuyo, Y., Kodam, M., 2006. Remarkable difference in accumulation of paralytic shellfish poisoning toxins among bivalve species exposed to Pyrodinium bahamense var. compressum bloom in Masinloc bay. Philippines Toxicon 48 (1), 85–92. Montoya, N.G., Akselman, R., Franco, J.M., Carreto, J.I., 1996. Paralytic shellfish toxins and mackerel (Scomber japonicus) mortality in the Argentine sea. In: Yasumoto, T., Oshima, Y., Fukuyo, Y. (Eds.), Harmful and Toxic Algal Blooms. Intergovernmental Oceanographic Community, UNESCO, Paris, pp. 417–420. Moore, S.K., Mantua, N.J., Kellogg, J.P., Newton, J.A., 2008. Local and large-scale climate forcing of Puget Sound oceanographic properties on seasonal to interdecadal timescales. Limnology and Oceanography 53 (5), 1746–1758. Moore, S.K., Mantua, N.J., Hickey, B.M., Trainer, V.L., 2009. Recent trends in paralytic shellfish toxins in Puget Sound, relationships to climate, and capacity for prediction of toxic events. Harmful Algae 8 (3), 463–477. Moorthi, S.D., Countway, P.D., Stauffer, B.A., Caron, D.A., 2006. Use of quantitative real-time PCR to investigate the dynamics of the red tide dinoflagellate Lingulodinium polyedrum. Microbial Ecology 52 (1), 136–150. Murata, M., Shimatani, M., Sugitani, Y., Oshima, Y., Yasumoto, T., 1982. Isolation and structural elucidation of the causative toxin of the diarrhetic shellfish poisoning. Bulletin of the Japanese Society of Scientific Fisheries 48 (4), 549–552. Murata, M., Kumagai, M., Lee, J.S., Yasumoto, T., 1987. Isolation and structure of Yessotoxin, a novel polyether compound implicated in diarrhetic shellfish poisoning. Tetrahedron Letters 28 (47), 5869–5872. Naar, J., Bourdelais, A., Tomas, C., Kubanek, J., Whitney, P.L., Flewelling, L., Steidinger, K., Lancaster, J., Baden, D.G., 2002. A competitive ELISA to detect brevetoxins from Karenia brevis (formerly Gymnodinium breve) in seawater, shellfish, and mammalian body fluid. Environmental Health Perspectives 110 (2), 179–185. Netjstgaard, J.C., Tang, K.W., Steinke, M., Dutz, J., Koski, M., Antajan, E., Long, J.D., 2007. Zooplankton grazing on Phaeocystis: a quantitative review and future challenges. Biogeochemistry 83 (1–3), 147–172. Nishitani, G., Yamaguchi, M., Ishikawa, A., Yanagiya, S., Mitsuya, T., Imai, M., 2005. Relationships between occurrences
413
of toxic Dinophysis species (Dinophyceae) and small phytoplanktons in Japanese coastal waters. Harmful Algae 4 (4), 755–762. O’Halloran, C., Silver, M., Holman, T., Scholin, C., 2006. Heterosigma akashiwo in central California waters. Harmful Algae 5 (2), 124–132. Office of Shellfish and Water Protection, 2008. 2007 Annual Report: Commercial and Recreational Shellfish Areas in Washington State Olympia. Office of Shellfish and Water Protection. Ogino, H., Kumagai, M., Yasumoto, T., 1997. Toxicologic evaluation of Yessotoxin. Natural Toxins 5 (6), 255–259. Ono, K., Khan, S., Onoue, Y., 2000. Effects of temperature and light intensity on the growth and toxicity of Heterosigma akashiwo (Raphidophyceae). Aquaculture Research 31 (5), 427–433. Pan, Y., Subba Rao, D.V., Mann, K.H., 1996a. Changes in domoic acid production and cellular chemical composition of the toxigenic diatom Pseudo-nitzschia muliseries under phosphate limitation. Journal of Phycology 32 (3), 371–381. Pan, Y., Subba Rao, D.V., Mann, K.H., Li, K.W., Harrison, W.G., 1996b. Effects of silicate limitation on production of domoic acid, a neurotoxin, by the diatom Pseudo-nitzschia multiseries. I. Batch culture studies. Marine Ecology Progress Series 131 (1–3), 225–233. Pan, Y., Bates, S.S., Cembella, A.D., 1998. Environmental stress and domoic acid production by Pseudo-nitzschia: a physiological perspective. Natural Toxins 6 (3–4), 127–135. Pankratz, T., 2008. Red tides close desal plants. Water Desalination Report 44, 1. Pankratz, T., 2009. Speaking of red tides. Water Desalination Report 45, 2. Park, M.G., Kim, S., Kim, H.S., Myung, G., Kang, Y.G., Yih, W., 2006. First successful culture of the marine dinoflagellate Dinophysis. Aquatic Microbial Ecology 45 (2), 101–106. Paz, B., Riobo, P., Fernandez, A.L., Fraga, S., Franco, J.M., 2004. Production and release of yessotoxins by the dinoflagellates Protoceratium reticulatum and Lingulodinium polyedrum in culture. Toxicon 44 (3), 251–258. Paz, B., Riobo, P., Souto, M.L., Gil, L.V., Norte, M., Fernandez, J.J., Franco, J.M., 2006. Detection and identification of glycoyessotoxin A in a culture of the dinoflagellate Protoceratium reticulatum. Toxicon 48 (6), 611–619. Paz, B., Daranas, A.H., Cruz, P.G., Franco, J.M., Pizarro, G., Souto, M.L., Norte, M., Ferna´ndez, J.J., 2007. Characterization of okadaic acid related toxins by liquid chromatography coupled with mass spectrometry. Toxicon 50 (2), 225–235. Paz, B., Daranas, A.H., Norte, M., Riobo, P., Franco, J.M., Ferna´ndez, J.J., 2008. Yessotoxins, a group of marine polyether toxins: an overview. Marine Drugs 6 (6), 73–102. Peleka, E.N., Matis, K.A., 2008. Application of flotation as a pretreatment process during desalination. Desalination 222 (1–3), 1–8. Perl, T.M., Bedard, L., Kosatsky, T., Hockin, J.C., Todd, E.C.D., McNutt, L.A., Remis, R.S., 1990. Amnesic shellfish poisoning: a new clinical syndrome due to domoic acid. Canadian Diseases Weekly Report 16 (Suppl 1E), 7–8. Petry, M., Sanz, M.A., Langlais, C., Bonnelye, V., Durand, J.P., Guevara, D., Nardes, W.M., Saemi, C.H., 2007. The El Coloso (Chile) reverse osmosis plant. Desalination 203 (1–3), 141–152. Pocklington, R., Milley, J.E., Bates, S.S., Bird, C.J., deFreitas, A.S.W., Quilliam, M.A., 1990. Trace determination of domoic acid and phytoplankton by high-performance liquid chromatography of the fluorenylmethoxycarbonyl (FMOC) derivative. International Journal of Environmental Analytical Chemistry 38, 351–358. Prassopoulou, E., Katikou, P., Georgantelis, D., Kyritsakis, A., 2009. Detection of okadaic acid and related esters in mussels
414
water research 44 (2010) 385–416
during diarrhetic shellfish poisoning (DSP) episodes in Greece using the mouse bioassay, the PP2A inhibition assay and HPLC with fluorimetric detection. Toxicon 53 (2), 214–227. Price, D.W., Kizer, K.W., Hansgen, K.H., 1991. California’s paralytic shellfish poisoning prevention program, 1927–89. Journal of Shellfish Research 10 (11), 119–145. Purkerson, S.L., Baden, D.G., Fieber, L.A., 1999. Brevetoxin modulates neuronal sodium channels in two cell lines derived from rat brain. Neurotoxicology 20 (6), 909–920. Quilliam, M., 2003. Chemical methods for domoic acid, the amnesic shellfish poisoning (ASP) toxin. In: Hallegraeff, G., Anderson, D., Cembella, A. (Eds.), Manual on Harmful Marine Microalgae, Monographs on Oceanographic Methodology. Intergovernmental Oceanographic Commission (UNESCO), Paris, pp. 247–266. Quilliam, M.A., Sim, P.G., McCulloch, A.W., McInnes, A.G., 1989. High-performance liquid chromatography of domoic acid, a marine neurotoxin, with application to shellfish and plankton. International Journal of Environmental Analytical Chemistry 36 (3), 139–154. Ramsdell, J.S., Anderson, D.M., Glibert, P.M., 2005. HARRNESS, Harmful Algal Research and Response: a National Environmental Science Strategy 2005–2015. Ecological Society of America, Washington, DC, 96 pp. Reid, J.L., Roden, G.I., Wyllie, J.G., 1958. Studies of the California Current system. California Cooperative Oceanic Fisheries Investigations Progress Report, 1 July 1956–1 January 1958, pp. 27–56. Reid, F.M.H., Fuglister, E., Jordan, J.B., 1970. The ecology of the plankton off La Jolla, California, in the period April through September, 1967. Part 5: phytoplankton taxonomy and standing crop. Bulletin of the Scripps Institute of Oceanography Techical Series 17, 51–66. Reid, F.M.H., Lange, C.B., White, M.M., 1985. Microplankton species assemblages at Scripps Pier from March to November 1983 during the 1982–1983 El Nin˜o Event. Botanica Marina 28 (10), 443–452. Reiss, C.R., Robert, C., Owen, C., Taylor, J.S., 2006. Control of MIB, geosmin and TON by membrane systems. Journal of Water Supply: Research and Technology – AQUA 55 (2), 95–108. Rhodes, L., White, D., Syhre, M., Atkinson, M., 1996. Pseudonitzschia species isolated from New Zealand coastal waters: domoic acid production in vitro and links with shellfish toxicity. In: Yasumoto, T., Oshima, Y., Fukuyo, Y. (Eds.), Harmful and Toxic Algal Blooms. IOC of UNESCO, Paris pp. 155–158. Rhodes, L., Scholin, C., Garthwaite, I., Haywood, A., Thomas, A., 1998. Domoic acid producing Pseudo-nitzschia species deduced with whole cell DNA probe-based and immunochemical assays. In: Reguera, B., Blanco, J., Ferna´ndez, M.L., Wyatt, T. (Eds.), Harmful Algae. Xunta de Galicia and Intergovernmental Oceanographie Commission of UNESCO, pp. 274–277. Rhodes, L., McNabb, P., de Salas, M., Briggs, L., Beuzenberg, V., Gladstone, M., 2006. Yessotoxin production by Gonyaulax spinifera. Harmful Algae 5 (2), 148–155. Rines, J.E.B., Donaghay, P.L., Dekshenieks, M.M., Sullivan, J.M., Twardowski, M.S., 2002. Thin layers and camouflage: hidden Pseudo-nitzschia spp. (Bacillariophyceae) populations in a fjord in the San Juan Island, Washington, USA. Marine Ecology Progress Series 225, 123–137. Riobo, P., Paz, B., Fernandez, M., Fraga, S., Franco, J.M., 2002. Lipophylic toxins of different strains of Ostreopsidaceae and Gonyaulacaceae. In: Tenth International Conference on Harmful Algal Blooms, St. Petersburg, FL (abstract). Rogers, R.S., Rapoport, H., 1980. The pKas of saxitoxin. Journal of American Chemical Society 102 (24), 7335–7339.
Ronzitti, G., Callegari, F., Malaguti, C., Rossini, G.P., 2004. Selective disruption of the E-cadherin–catenin system by an algal toxin. British Journal of Cancer 90 (5), 1100–1107. Rue, E.L., Bruland, K.W., 2001. Domoic acid binds iron and copper: a possible role for the toxin produced by the marine diatom Pseudo-nitzschia. Marine Chemistry 76 (1–2), 127–134. Samdal, I.A., Naustvoll, L.J., Olseng, C.D., Briggs, L.R., Miles, C.O., 2004. Use of ELISA to identify Protoceratium reticulatum as a source of yessotoxin in Norway. Toxicon 44 (1), 75–82. Samdal, I.A., Aesen, J.A.G., Briggs, L.R., Dahl, E., Miles, C.O., 2005. Comparison of ELISA and LC–MS analyses for yessotoxins in blue mussels (Mytilus edulis). Toxicon 46, 7–15. Satake, M., Mackenzie, L., Yasumoto, T., 1997. Identification of Protoceratium reticulatum as the biogenetic origin of yessotoxin. Natural Toxins 5 (4), 164–167. Satake, M., Ichimura, T., Sekiguchi, K., Yoshimatsu, S., Oshima, Y., 1999. Confirmation of yessotoxin and 45, 46, 47trinoryessotoxin production by Protoceratium reticulatum collected in Japan. Natural Toxins 7 (4), 147–150. Schantz, E.J., Mold, J.D., Stanger, D.W., Shavel, J., Riel, F.J., Bowden, J.P., Lynch, J.M., Wyler, R.S., Riegel, B., Sommer, H., 1957. Paralytic shellfish poison. VI. A Procedure for the isolation and purification of the poison from toxic clam and mussel tissues. Journal of American Chemical Society 79 (19), 5230–5235. Schnetzer, A., Miller, P.E., Schaffner, R.A., Stauffer, B.A., Jones, B.H., Weisberg, S.B., DiGiacomo, P.M., Berelson, W.M., Caron, D.A., 2007. Blooms of Pseudo-nitzschia and domoic acid in the San Pedro Channel and Los Angeles harbor areas of the Southern California Bight, 2003-2004. Harmful Algae 6 (3), 372–387. Schofield, O., Kohut, J., Aragon, D., Creed, L., Graver, J., Haldeman, C., Kerfoot, J., Roarty, H., Jones, C., Webb, D., Glenn, S., 2007. Slocum gliders: robust and ready. Journal of Field Robotics 24 (6), 473–485. Scholin, C.A., Herzog, M., Sogin, M., Anderson, D.M., 1994. Identification of group-and strain-specific genetic markers for globally distributed Alexandrium (Dinophyceae). Part II. Sequence analysis of a fragment of the LSU rRNA gene. Journal of Phycology 30 (6), 999–1011. Scholin, C.A., Gulland, F., Doucette, G.J., Benson, S., Busman, M., Chavez, F.P., Cordaro, J., DeLong, R., De Vogelaere, A., Harvey, J., Haulena, M., Lefebvre, K.A., Lipscomb, T., Loscutoff, S., Lowenstine, L.J., Marin 3rd, R., Miller, P.E., McLellan, W.A., Moeller, P.D., Powell, C.L., Rowles, T., Silvagni, P., Silver, M., Spraker, T., Trainer, V.L., Van Dolah, F.M., 2000. Mortality of sea lions along the central California coast linked to a toxic diatom bloom. Nature 403 (6765), 80–84. Separation Processes Inc., 2005. Investigation of Microfiltration and Reverse Osmosis Seawater Desalination. National Water Research Institute. Shimizu, Y., Hsu, C.P., Genenah, A.A., 1981. Structure of saxitoxin in solutions and stereochemistry of dihydrosaxitoxins. Journal of American Chemical Society 103, 605–609. Shipe, R.F., Leinweber, A., Gruber, N., 2008. Abiotic controls of potentially harmful algal blooms in Santa Monica Bay, California. Continental Shelf Research 28 (18), 2584–2593. Shumway, S.E., Allen, S.M., Boersma, P.D., 2003. Marine birds and harmful algal blooms: sporadic victims or under-reported events? Harmful Algae 2 (1), 1–17. Smayda, T.J., 1990. Novel and nuisance phytoplankton blooms in the sea: evidence for a global epidemic. In: Grane´li, E., Gundstro¨m, B., Edler, L., Anderson, D.M. (Eds.), Toxic Marine Phytoplankton. Elsevier, New York, pp. 29–40. Smayda, T.J., 2008. Complexity in the eutrophication–harmful algal bloom relationship, with comment on the importance of grazing. Harmful Algae 8 (1), 140–151.
water research 44 (2010) 385–416
Smith, G., Ladizinsky, N., Miller, P., 2001. Amino acid profiles in species and strains of Pseudo-nitzschia from Monterey Bay, California: insights into the metabolic role(s) of domoic acid. In: Hallegraef, G., Blackburn, S., Bolch, C., Lewis, R. (Eds.). IOC, UNESCO, Hobart, Tasmania, pp. 324–327. Sommer, H., Meyer, K.F., 1937. Paralytic shellfish poisoning. Archives of Pathology 24, 560–598. Steidinger, K.A., 1993. Some taxonomic and biologic aspects of toxic dinoflagellates. In: Falconer, I. (Ed.), Algal Toxins in Seafood and Drinking Water. Academic Press, London, pp. 1–28. Stobo, L.A., Lewis, J., Quilliam, M.A., Hardstaff, W.R., Gallacher, S., Webster, L., Smith, E.A., McKenzie, M., 2003. Detection of yessotoxin in UK and Canadian isolates of phytoplankton and optimization and validation of LC–MS methods. Canadian Technical Report of Fisheries and Aquatic Sciences 2498, 8–14. Strom, S., Wolfe, G.V., Slajer, A., Lambert, S., Clough, J., 2003. Chemical defenses in the microplankton II: inhibition of protist feeding by B-dimethylsulfoniopropionate (DMSP). Limnology and Oceanography 48 (1), 230–237. Su, Z., Sheets, M., Ishida, H., Li, F., Barry, W.H., 2004. Saxitoxin blocks L-type ICa. Journal of Pharmacology and Experimental Therapeutics 308 (1), 324–329. Subba Rao, D.V., Quilliam, M.A., Pocklington, R., 1988. Domoic acid – a neurotoxic amino acid produced by the marine diatom Nitzschia pungens in culture. Canadian Journal of Fisheries and Aquatic Sciences 45 (12), 2076–2079. Suganuma, M., Fujiki, H., Suguri, H., Yoshizawa, S., Hirota, M., Nakayasu, M., Ojika, M., Wakamatsu, K., Yamada, K., Sugimura, T., 1988. Okadaic acid: an additional non-phorbol12-tetradecanoate-13-acetate-type tumor promoter. Proceedings of the National Academy of Sciences, United States of America 85 (6), 1768–1771. Sukhatme, G.A.D., Zhang, B., Oberg, C., Stauffer, B.A., Caron, D.A., 2007. The design and development of a wireless robotic networked aquatic microbial observing system. Environmental Engineering Science 24, 205–215. Sunda, W., 2006. Trace metals and harmful algal blooms. In: Grane´li, E., Turner, J.T. (Eds.), Ecology of Harmful Algae. Springer, Berlin, pp. 203–214. Sutherland, C.M., 2008. Diarrhetic Shellfish Toxins Linked to Local Dinophysis Populations in the California Coastal Waters of Monterey Bay. Dissertation, University of California, Santa Cruz, Santa Cruz. Tachibana, K., Scheuer, P.J., Tsukitani, Y., Kikuchi, H., Engen, D.V., Clardy, J., Gopichand, Y., Schmitz, F.J., 1981. Okadaic acid, a cytotoxic polyether from two marine sponges of the genus Halichondria. Journal of American Chemical Society 103 (9), 2469–2471. Takahashi, H., Kusumi, T., Kan, Y., Satake, M., Yasumoto, T., 1996. Determination of the absolute configuration of yessotoxin, a polyether compound implicated in diarrhetic shellfish poisoning, by NMR spectroscopic method using a chiral anisotropic reagent, methoxy-(2-naphthyl) acetic acid. Tetrahedron Letters 37 (39), 7087–7090. Takai, A., Bialojan, C., Troschka, M., Ru¨egg, J.C., 1987. Smooth muscle myosin phosphatase inhibition and force enhancement by black sponge toxin. FEBS Letters 217 (1), 81–84. Taylor, F.J.R., Horner, R.A., 1994. Red tides and other problems with harmful algal blooms in Pacific Northwest coastal waters. In: Review of the Marine Environment and Biota of Strait of Georgia, Puget Sound and Juan de Fuca Strait. Canadian Fisheries and Aquatic Science Technical Report 1948, pp. 175–186. Tenzer, B., Adin, A., Priel, M., 1999. Seawater filtration for fouling prevention under stormy conditions. Desalination 125 (1–3), 77–88.
415
Terao, K., Ito, E., Oarada, M., Murata, M., Yasumoto, T., 1990. Histopathological studies on experimental marine toxin poisoning: the effects in mice of yessotoxin isolated from Patinopecten yessoensis and of a desulfated derivative. Toxicon 28 (9), 1095–1104. Thomas, W.H., Zavoico, T., Hewes, C., 2001. Historical Phytoplankton Species Time-Series Data (1917–1939) from the North American Pacific Coast. Scripps Institution of Oceanography. Ref. no. 01-12. Throndsen, J., 1997. The plankton marine flagellates. In: Tomas, C.R. (Ed.), Identifying marine phytoplankton. Academic Press, San Diego, pp. 591–729. Trainer, V.L., 2002. Harmful algal blooms on the U.S. west coast. In: Taylor, F.J., Trainer, V.L. (Eds.), Harmful Algal Blooms in the PICES Region of the North Pacific, pp. 89–118. PICES Scientific Report 23. Trainer, V.L., Adams, N.G., Bill, B.D., Anulacion, F., Wekell, J.C., 1998. Concentration and dispersal of a Pseudo-nitzschia bloom in Penn Cove, Washington, USA. Natural Toxins 6 (3–4), 113–126. Trainer, V.L., Adams, N.G., Bill, B.D., Stehr, C.M., Wekell, J.C., Moeller, P., Busman, M., Woodruff, D., 2000. Domoic acid production near California coastal upwelling zones, June 1998. Limnology and Oceanography 45 (8), 1818–1833. Trainer, V.L., Adams, N.G., Wekell, J.C., 2001. Domoic acid producing Pseudo-nitzschia species off the U.S. west coast associated with toxification events. In: Hallegraeff, G.M., Blackburn, S.I., Bolch, C.J., Lewis, R.J. (Eds.), Harmful Algal Blooms 2000. Intergovernmental Oceanographic Commission of UNESCO, Paris, pp. 46–49. Trainer, V.L., Hickey, B.M., Horner, R.A., 2002. Biological and physical dynamics of domoic acid production off the Washington Coast. Limnology and Oceanography 47 (5), 1438–1446. Trainer, V.L., Eberhart, B.T.L., Wekell, J.C., Adams, N.G., Hanson, L., Cox, F., Dowell, J., 2003. Paralytic shellfish toxins in Puget Sound, Washington State. Journal of Shellfish Research 22 (1), 213–223. Trainer, V.L., Cochlan, W.P., Erickson, A., Bill, B.D., Cox, F.H., Borchert, J.A., Lefebvre, K., 2007. Recent domoic acid closures of shellfish harvest areas in Washington State inland waterways. Harmful Algae 6 (3), 449–459. Tubaro, A., Sidari, L., Della Loggia, R., Yasumoto, T., 1998. Occurrence of yessotoxin in phytoplankton and mussels from Northern Adriatic Sea. In: Proceedings of the VIII International Conference on Harmful Algae, Vigo, Spain (abstract). Twiner, M.J., Dechraoui, M.Y.B., Wang, Z., Mikulski, C.M., Henry, M. S., Pierce, R.H., Doucette, G.J., 2007. Extraction and analysis of lipophilic brevetoxins from the red tide dinoflagellate Karenia brevis. Analytical Biochemistry 369 (1), 128–135. Tyrell, J.V., Connell, L.B., Scholin, C.A., 2002. Monitoring for Heterosigma akashiwo using a sandwich hybridization assay. Harmful Algae 1 (2), 205–214. Van Dolah, F.M., Finley, E.L., Haynes, B.L., Doucette, G.J., Moeller, P.D., Ramsdell, J.S., 1994. Development of rapid and sensitive high throughput pharmacologic assays for marine phycotoxins. Natural Toxins 2 (4), 189–196. Vigilant, V.L., Silver, M., 2007. Domoic acid in benthic flatfish on the continental shelf of Monterey Bay, California, USA. Marine Biology 151 (6), 2053–2062. Walz, P.M., Garrison, D.L., Graham, W.M., Cattey, M.A., Tjeerdema, R.S., Silver, M.W., 1994. Domoic acid-producing diatom blooms in Monterey Bay, California: 1991–1993. Natural Toxins 2 (5), 271–279. Wang, J., Salata, J.J., Bennett, P.B., 2003. Saxitoxin is a gating modifier of hERG Kþ channels. Journal of General Physiology 121 (6), 583–598.
416
water research 44 (2010) 385–416
Watson, M., 1997. Control of Membrane Fouling Due to Algae. Applied Research Department. Technology Transfer Note. No.12. Wekell, J.C., Gauglitz, E.J.J., Barnett, H.J., Hatfield, C.L., Simons, D., Ayres, D., 1994. Occurrence of domoic acid in Washington state razor clams (Siliqua patula) during 1991–1993. Natural Toxins 2 (4), 197–205. Wells, M.L., Trick, C.G., Cochlan, W.P., Hughes, M.P., Trainer, V.L., 2005. Domoic acid: the synergy of iron, copper, and the toxicity of diatoms. Limnology and Oceanography 50 (6), 1908–1917. Wilf, M., Schierach, M.K., 2001. Improved performance and cost reduction of RO seawater systems using UF pretreatment. Desalination 135 (1–3), 61–68. Wong, J.L., Oesterlin, R., Rapoport, H., 1971. Struture of saxitoxin. Journal of American Chemical Society 93 (26), 7344–7345. Wright, J.L.C., Cembella, A.D., 1998. Ecophysiology and biosynthesis of polyether marine biotoxins. In: Anderson, D.M., Cembella, A.D., Hallegraeff, G.M. (Eds.), The Physiological
Ecology of Harmful Algal Blooms. Springer-Verlag, Heidelberg, pp. 427–451. Wright, J.L.C., Bird, C., de Freitas, A., Hampson, D., McDonald, J., Quilliam, M., 1990. Chemistry, biology, and toxicology of domoic acid and its isomers. Canadian Diseases Weekly Report 16 (Suppl. 1E), 21–26. Yasumoto, T., Murata, M., 1993. Marine toxins. Chemical Reviews 93, 1867–1909. Yasumoto, T., Takizawa, A., 1997. Fluorometric measurement of yessotoxins in shellfish by high-pressure liquid chromatography. Bioscience, Biotechnology and Biochemistry 61 (10), 1775–1777. Yasumoto, T., Oshima, Y., Sugawara, W., Fukuyo, Y., Oguri, H., Igarashi, T., Fujita, N., 1980. Identification of Dinophysis fortii as the causative organism of diarrhetic shellfish poisoning. Bulletin of the Japanese Society of Scientific Fisheries 46 (11), 1405–1411. Yasumoto, T., Murata, M., Oshima, Y., Sano, M., Matsumoto, G.K., Clardy, J., 1985. Diarrhetic shellfish toxins. Tetrahedron 41 (6), 1019–1025.
water research 44 (2010) 417–426
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Occurrence and removal of pharmaceuticals, caffeine and DEET in wastewater treatment plants of Beijing, China Qian Sui, Jun Huang, Shubo Deng, Gang Yu*, Qing Fan POPs Research Centre, Department of Environmental Science & Engineering, Tsinghua University, Beijing 10084, China
article info
abstract
Article history:
The occurrence and removal of 13 pharmaceuticals and 2 consumer products, including
Received 14 January 2009
antibiotic, antilipidemic, anti-inflammatory, anti-hypertensive, anticonvulsant, stimulant,
Received in revised form
insect repellent and antipsychotic, were investigated in four wastewater treatment plants
5 July 2009
(WWTPs) of Beijing, China. The compounds were extracted from wastewater samples by
Accepted 8 July 2009
solid-phase extraction (SPE) and analyzed by ultra-performance liquid chromatography
Available online 15 July 2009
coupled with tandem mass spectrometry (UPLC–MS/MS). Most of the target compounds were detected, with the concentrations of 4.4 ng L1–6.6 mg L1 and 2.2–320 ng L1 in the influents
Keywords:
and secondary effluents, respectively. These concentrations were consistent with their
Pharmaceuticals
consumptions in China, and much lower than those reported in the USA and Europe. Most
Wastewater
compounds were hardly removed in the primary treatment, while their removal rates ranging
Removal efficiency
from 12% to 100% were achieved during the secondary treatment. In the tertiary treatment,
Advanced treatment
different processes showed discrepant performances. The target compounds could not be
Risk assessment
eliminated by sand filtration, but the ozonation and microfiltration/reverse osmosis (MF/RO)
China
processes employed in two WWTPs were very effective to remove them, showing their main contributions to the removal of such micro-pollutants in wastewater treatment. ª 2009 Elsevier Ltd. All rights reserved.
1.
Introduction
With the progress of sensitive analytical techniques, the frequent detection of various pharmaceuticals in the aquatic environment has received global concerns of both the academic community and the public (Daughton and Ternes, 1999; Jones et al., 2005). After intake by humans or animals, the pharmaceuticals will be partially converted to metabolites, however, partially excreted unchanged or as conjugates, and finally delivered to the wastewater treatment plants (WWTPs). As there is no unit specifically designed to remove these compounds, the elimination by most WWTPs seems to be inefficient (Ternes, 1998; Castiglioni et al., 2006; Lishman et al., 2006; Nakada et al., 2006; Santos et al., 2007; Vieno et al., 2007b; Xu et al., 2007; Gulkowska et al., 2008; Paxeus, 2004). Together with treated wastewater, these compounds are
released to the aquatic environment, and consequently found to contaminate the receiving water bodies (Lindqvist et al., 2005; Kasprzyk-Hordern et al., 2009), or even raw water sources of drinking water treatment plant (Ternes et al., 2002; Vieno et al., 2007a; Radjenovic et al., 2008). Meanwhile, results of toxicology studies have revealed that some pharmaceuticals are suspected to have direct toxicity to certain aquatic organisms (Ferrari et al., 2003; Jjemba, 2006; Grung et al., 2008; Quinn et al., 2008). Besides, their continual but undetectable effects could accumulate slowly, and finally lead to irreversible change on wildlife and human beings (Daughton and Ternes, 1999). Therefore, the occurrence and behavior of pharmaceuticals in the WWTPs, which are both the sink and source of the compounds, should be focused on. So far, concentrations of pharmaceuticals from various therapeutic classes in the WWTPs have been well documented in the
* Corresponding author. Tel.: þ86 10 62787137; fax: þ86 10 62794006. E-mail address:
[email protected] (G. Yu). 0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2009.07.010
418
water research 44 (2010) 417–426
North America (Thomas and Foster, 2005; Lishman et al., 2006), Japan (Nakada et al., 2006) and some European countries (Ternes, 1998; Castiglioni et al., 2006; Santos et al., 2007; Vieno et al., 2007b; Jones et al., 2007; Paxeus, 2004). Reported species and concentrations of pharmaceuticals varied from country to country, and plant to plant, owing to the different usage patterns. Meanwhile, the removal efficiencies of pharmaceuticals also varied much (Nakada et al., 2006; Gulkowska et al., 2008), indicating that the removal could be affected by both the compound-specific properties, and the factors concerning specific WWTPs, such as types of treatment processes, solids retention time (SRT), hydraulic retention time (HRT), temperature, etc. In recent years, very few studies about the situation in China have been reported. Only one specific therapeutic class, antibiotics, has been investigated by limited previous studies (Xu et al., 2007; Gulkowska et al., 2008; Chen et al., 2008). Therefore, it is necessary and important to investigate the occurrence and removal of pharmaceuticals from different therapeutic classes in the WWTPs of China. Due to the low efficiency of conventional wastewater treatment processes, some advanced treatment technologies have been evaluated. Ozonation was found to be effective to remove pharmaceuticals in real municipal WWTPs of Japan (Nakada et al., 2007; Okuda et al., 2008) and Germany (Ternes et al., 2003). Nanofiltration (NF) and reverse osmosis (RO) membrane filtration, the well-proven technologies to remove pharmaceuticals from different kinds of waters, have also been applied at bench, pilot and full scale (Khan et al., 2004; Nghiem et al., 2005; Drewes et al., 2005; Al-Rifai et al., 2007; Watkinson et al., 2007; Comerton et al., 2008; Radjenovic et al., 2008). Retention behavior of pharmaceuticals during the processes associated with physicochemical properties of pharmaceuticals, membranes as well as the solution chemistry, and mechanisms of pharmaceutical rejection have been discussed in Kimura et al. (2004), Nghiem et al. (2005), Nghiem and Coleman (2008) and Comerton et al. (2008). Recently, considering the requirement of reclaimed water, several advanced treatment facilities have been installed in the WWTPs of Beijing. However, the removal efficiency of micro-pollutants, such as pharmaceuticals, has not been evaluated yet. In the present study, we investigated the contamination levels of 13 pharmaceuticals and 2 consumer products from 8 classes (i.e. antibiotic, antilipidemic, anti-inflammatory, antihypertensive, anticonvulsant, stimulant, insect repellent and antipsychotic) in four WWTPs of Beijing, China, which have different advanced treatment units, and evaluated the elimination efficiencies of the target pharmaceuticals. To the best of our knowledge, this is the first report on the occurrence and removal of pharmaceuticals and consumer products from multiple classes in the WWTPs of China, especially for the situation during the advanced treatment processes.
2.
Materials and methods
2.1.
Chemicals
All the standards including chloramphenicol (CP), nalidixic acid (NA), trimethoprim (TP), bezafibrate (BF), clofibric acid (CA), gemfibrozil (GF), diclofenac (DF), indometacin (IM),
ketoprofen (KP), mefenamic acid (MA), metoprolol (MTP), carbamazepine (CBZ), caffeine (CF), N,N-diethyl-meta-toluamide (DEET) and sulpiride (SP) (Appendix) were of analytical grade (>90%), and purchased from Sigma–Aldrich (Steinheim, Germany). Isotopically labeled compounds, used as internal standards, were 13C-phenacetin obtained from Sigma– Aldrich, and 3D-mecoprop from Dr. Ehrenstorfer (Augsburg, Germany). HPLC grade methanol, acetone, dichloromethane, hexane, as well as formic acid were provided by Dikma (USA), and ultra-pure water was produced by a Milli-Q unit (Millipore, USA). Stock solutions of individual compound were prepared in methanol and mixture standards with different concentrations were prepared by diluting the stock solutions before each analytical run. All the solutions were stored at 4 C in the dark.
2.2.
Sample collection
Four full-scale municipal WWTPs, referred as A, B, C and D, were selected in our study. These WWTPs employ similar conventional treatment processes: primary treatment to remove particles coupled with secondary biological treatment. For the secondary biological treatment processes, WWTPs A and D employ anaerobic/anoxic/oxic (A2/O) activated sludge process, anoxic/oxic (A/O) activated sludge process is adopted in WWTP B, and WWTP C employs oxidation ditch (OD). Other detailed information on each WWTP, such as inhabitants served, daily flow, HRT and SRT are shown in Table 1. Part of the secondary effluents was further treated in WWTPs A, B and D, by the processes of ultrafiltration (UF)/ozone, sand filtration (SF) and microfiltration/reverse osmosis (MF/RO), respectively. In WWTP A, a dead-end ultrafiltration system (Zenon GE) is used. The whole system has 6 trains of Zee-Weed 1000 membrane. Each train contains 9 cassettes of 57–60 modules per cassette. The membrane, with the pore size of 0.02 mm, is made by PVDF. The module is operated in an outside/in configuration at a constant flow of 23 L (m2 h)1 and the total treatment capacity reaches 80,000 m3 d1. The membrane is hydraulically backwashed at a constant flow rate of 34 (m2 h)1, and 29 times per day. The backwash phase lasts for 1 min. Maintenance cleaning is conducted once per day. Membranes are soaked in the sodium hypochlorite solution (50 mg L1) for 25 min. For the ozonation process, gaseous ozone is generated from an ozone generator (Mitsubishi Electric). The ozone dosage and contact time in the reaction tank is 5 mg L1 and 15 min, respectively. The pH of the wastewater before ozonation ranges 6.5–8.0 and shows no significant change after ozonation. As the heart of the advanced treatment in WWTP D, a spiral-wound crossflow module is employed for the reverse osmosis (RO) membrane filtration. The RO membrane (Filmtec, DOW) is made from a thin-film composite polyamide material. Each module is designed to operate at a water flux of 1.3 m3 h1, and a product water recovery of 75–80%. The trans-membrane pressure is between 0.04 and 0.06 MPa, and the salt rejection remained at the level of 99%. Every 3–6 months, normally when the transmembrane pressure reaches above 0.06 MPa, the membrane is cleaned with 0.1%(w) sodium hydroxide solution (for organic foulants), 2%(w) citric acid (for inorganic foulants) and 0.5%(w) formaldehyde (as biocide). Schematic diagram of treatment processes in the four WWTPs is shown in Fig. 1.
419
water research 44 (2010) 417–426
Table 1 – Information of the WWTPs investigated. WWTP A B C D
Inhabitants served 103
Daily flow (103 m3)
HRT (h)
SRT (d)
Secondary treatment
Tertiary treatment
814 2400 480 2415
400 1000 200 600
11 11 15 10.67
12–15 20 12–16 15
A2/O A/O OD A2/O
UF/ozone SF – MF/RO
The samples were collected once from the four WWTPs during June and July 2008, with no compensation for HRT. All of them were collected as grab samples in duplicate (500 mL for influents and 1000 mL for the others) in prewashed amber glass bottles, kept in the cooler and transported to the laboratory. Immediately after delivery to the laboratory, they were filtered through prebaked (400 C, >4 h) glass microfiber filters (GF/F, Whatman) to remove particles and stored at 4 C before extraction.
2.3.
conditioned, wastewater samples, added with internal standards and adjusted to pH ¼ 7, were introduced to the cartridge via a PTFE tube, at a flow rate of 5–10 mL min1. After washing by 5 mL of 5.0% methanol solution, the cartridge was dried under vacuum for 2 h and eluted with 5 mL of methanol. The extract was then concentrated to 0.4 mL under a gentle nitrogen stream and stored at 4 C for analysis. Concentrations of the target compounds were analyzed using ultraperformance liquid chromatography coupled with tandem mass spectrometry (UPLC–MS/MS). Analytes were separated using Waters Acquity UPLC system (Waters Corporation, USA) equipped with Acquity UPLC BEH C18 column (50 2.1 mm, particle size of 1.7 mm), and detected by Quattro Premier XE tandem quadrupole mass spectrometry (Waters Corp., USA) equipped with an electrospray ionization source. The analysis was carried out in multiple reaction monitoring (MRM) mode,
Sample extraction and analysis
The method for the extraction and analysis of pharmaceuticals and consumer products is presented elsewhere (Sui et al., in press) and briefly described here. After the solid-phase extraction (SPE) cartridges (Oasis, HLB, 200 mg, 6 mL) were
a
Ozonation
Ultrafiltration
WWTP A
Tertiary Effluent Influent Grit Removal
A/O treatment
Secondary Clarifier
Secondary Effluent
Screen
b
Sand Filtration
WWTP B
Tertiary Effluent Influent Grit Removal
A2/O treatment
Primary Clarifier
Secondary Clarifier Secondary Effluent
Screen
c
WWTP C
Influent
Grit Removal
Oxidation Ditch
Screen
Secondary Clarifier
Secondary Effluent
MF
d
WWTP D
RO
Tertiary Effluent
`
Influent Grit Removal
Primary Clarifier
A2/O treatment
Secondary Clarifier
Secondary Effluent
Screen
Fig. 1 – Schematic diagram of the treatment processes in the four WWTPs and sampling site location (C).
420
water research 44 (2010) 417–426
and in general, two precursor ion/product ion transitions were monitored for one compound with the purpose of quantification and confirmation.
concentrations of some target compounds (i.e. caffeine, DEET, carbamazepine) resulted in slightly higher deviations.
2.4.
3.
Result and discussion
3.1.
Influents
Quality control
For each sampling, 500 mL Milli-Q water in an amber glass bottle as a field blank was brought to the WWTPs, exposed to the environment where the samples were taken from, and then delivered back to the laboratory with samples. For each set of samples (normally 10 samples), at least one procedural blank was prepared from ultra-pure water in the laboratory. Both the field blanks and procedural blanks were run identically to the wastewater samples, and the concentrations of target compounds were below the limit of quantification (LOQ). The absolute recoveries, calculated by comparing the concentrations of target compounds in spiked and unspiked wastewaters, were proved to be 73–102% and 50–95% in the effluent and influent for most compounds, respectively. While for several compounds (i.e. sulpiride, gemfibrozil, mefenamic acid), the absolute recoveries were not satisfactory. However, 13 C-phenacetin and 3D-mecoprop, the surrogate standards used for positive and negative ion mode respectively were able to compensate for the loss of most analytes, and relative recoveries were 67–130% for all the analytes in the effluent and 79–140% in the influents except mefenamic acid (251%) and nalidixic acid (178%). Therefore, the concentrations of these two compounds in the wastewater influents were not quantitatively determined and reported. The LOQs were 0.3–5.5 ng L1 and 0.7–20 ng L1 in the effluent and influent, respectively. Detailed information about the calibration, recoveries, LOQ, matrix effects, etc. were described in Sui et al. (in press), and briefly listed in Table 2. As duplicate samples were collected at each sampling site, mean concentrations were adopted. In most cases, deviations of duplicate samples were less than 20%. For some tertiary effluent samples, low
As shown in Fig. 2, 12 target compounds were detected in all the influent samples from the four WWTPs, while ketoprofen was below LOQ in all wastewater samples. The most abundant compounds detected were the consumer products, caffeine (3.4–6.6 mg L1) and N,N-diethyl-meta-toluamide (0.6–1.2 mg L1), probably due to the large consumption of drinks containing caffeine (i.e. coffee, tea, etc.) and wide application of insect repellent during the summer time when we sampled. Diclofenac, trimethoprim, sulpiride, carbamazepine, indometacin and metoprolol showed relatively high concentrations (Fig. 2). A similar composition distribution was observed among all the influents of the four WWTPs. The concentrations of target pharmaceuticals except diclofenac and trimethoprim, were much lower than those reported in the European and North American countries (Thomas and Foster, 2005; Lishman et al., 2006; Vanderford and Snyder, 2006; Santos et al., 2007; Gomez et al., 2007; Vieno et al., 2007b; Huerta-Fontela et al., 2008). For instance, the concentrations of ketoprofen in the wastewater influents were recorded to be 2.0 0.6 mg L1 in Finland (Lindqvist et al., 2005), 200 ng L1 in Australia (Al-Rifai et al., 2007), and 300–1360 ng L1 in Spain (Santos et al., 2007), while in the influents of four WWTPs in Beijing, it could not be detected. Concerning gemfibrozil, which is used to lower cholesterol and triglyceride levels in the blood, the contamination level found in the present study was 24–140 ng L1, even 1 or 2 order of magnitude lower than those in the USA (4770 ng L1, Vanderford and Snyder, 2006) and Canada (418 ng L1,
Table 2 – Instrumental quantification limit (IQL), limit of quantification (LOQ), absolute recovery (AR), relative recovery (RR) and matrix suppression of target compounds. Compounds
BF CA CBZ CF CP DEET DF GF IM KP MA MTP NA SP TP
IQL (pg)
0.5 10 2.5 2.5 2.5 2.5 10 10 2.5 10 10 2.5 2.5 0.5 2.5
LOQ (ng L1) Effluent
Influent
0.3 5.4 1.0 1.7 1.0 1.3 4.7 4.2 1.3 5.5 5.5 1.1 1.1 0.4 1.0
0.7 16 2.8 3.3 2.3 3.2 9.4 20 2.8 18 5.2 3.3 2.1 1.0 2.7
AR (n ¼ 6, %) Effluent
a Value in the brackets refers to the deviation of the recovery.
74 74 100 60 98 76 86 95 80 69 73 88 91 51 102
(3) (4) (4) (3) (7) (3) (12) (12) (12) (9) (11) (3) (9) (5) (7)
a
RR (n ¼ 6, %) a
Influent 58 50 71 61 86 63 85 40 73 44 154 60 95 42 74
(5) (4) (12) (37) (9) (6) (19) (11) (13) (3) (24) (3) (9) (3) (16)
a
Effluent 94 94 130 77 124 99 109 120 101 87 92 115 118 67 132
(4) (6) (6) (4) (10) (5) (16) (16) (16) (12) (15) (5) (12) (7) (9)
Matrix effect (%) a
Influent 94 82 133 114 140 118 139 65 119 72 251 113 178 79 139
(10) (8) (23) (70) (17) (14) (32) (18) (22) (7) (41) (9) (20) (7) (31)
27 30 6 49 39 24 4 28 1 46 13 15 3 63 5
421
water research 44 (2010) 417–426
a
10000
Concentration (ng L-1)
Influents
1000
100
10
Cpred ¼ SP
CBZ
MTP
IM
DF
GF
CA
BF
TP
CP
CF
DEET
1
10000
Concentration (ng L-1)
Pharmaceuticals & Consumer Products
b
1000
Lishman et al., 2006). The low levels of target pharmaceuticals were probably due to the lower per capita consumption in China than in the countries with higher socioeconomic statuses, where medical care is more prevalent (Thomas and Foster, 2005). The per capita consumption rate of gemfibrozil in China is estimated to be 0.036 mg person1 d1 (Table 3), lower than those in Germany (0.2 mg person1 d1, Ternes, 1998) and Canada (0.2 mg person1 d1, Lishman et al., 2006). Since the levels of target pharmaceuticals were somewhat different from those of European and North American countries, we theoretically calculate the concentration of pharmaceuticals in the wastewater influent by the following equation (Lindqvist et al., 2005; Nakada et al., 2006)
Secondary effluents
100
10
SP
CBZ
MTP
MA
IM
DF
GF
CA
BF
TP
CF
CP
DEET
1
Pharmaceuticals & Consumer Products Fig. 2 – Concentrations of target pharmaceuticals in wastewater influents (a) and secondary effluents (b) of four WWTPs in Beijing.
T e% I 1012 365 P Q
(1)
where Cpred is the predicted concentration of the pharmaceutical in wastewater influent (ng L1); T is the total production of a pharmaceutical both for human and animal use in China per year (ton year1), P is the population of China, e% is the amount of the pharmaceutical excreted unchanged, I is the number of inhabitants served and Q is the influent flow (m3 d1). The predicted concentrations of gemfibrozil, diclofenac, indometacin, ketoprofen, carbamazepine, and sulpiride were comparable to those measured in the influents (Table 3). Much lower measured concentration than predicted concentration of chloramphenicol was probably because it had been forbidden for use in food and aquaculture in China since 2005, and the available data about the production of pharmaceuticals were based on the year of 2004. It should be noticed that since there is no available data on the total consumption of any pharmaceutical, we used figures for total production of individual pharmaceutical instead. Therefore, the differences between the amounts actually produced and applied as well as the amount used in human and veterinary medicine could not be distinguished, which might result in overestimation of the theoretical concentration. Nevertheless, the comparability between the predicted concentrations and measured concentrations illustrates the overall reasonability of the approach.
Table 3 – Outputs, per capita consumption, predicted concentrations (PECs) and measured concentrations (MECs) of some pharmaceuticals in the wastewater influents of WWTPs investigated. Compound
CP TP GF DF IM KP CBZ SP
Outputa (tons year1)
Per capita consumption (mg person1 d1)
Excreted unchangedb (%)
PEC (ng L1)
MEC (mean, ng L1)
1929 2352 17 328 277 92 395 98
4.051 4.939 0.036 0.689 0.582 0.193 0.830 0.206
5–10 45 76 15 10–20 2.7 2–3 15–70
844 6173 76 287 242 15 58 243
31 400 60 318 129 n.d.c 113 157
a From CMEIN (2005). b From Bolton and Null (1981), Ternes (1998), Khan and Ongerth (2004), Niwa et al. (2005), Nakada et al. (2006), Jjemba (2006). c n.d. ¼ Not detected.
422
water research 44 (2010) 417–426
Compound
MECeff (ng L1)
PNECa (ng L1)
MEC/PNEC
15 140 4.7 13 204 n.d. 79 108
182,000 16,000 6000 42,000 100 306,000 31,000 420
0.0001 0.0087 0.0007 0.0003 2.0440 – 0.0025 0.2574
CF TP BF CA DF KP MTP CBZ
a From Santos et al. (2007), Lindqvist et al. (2005), Grung et al. (2008), Ferrari et al. (2003), Huschek et al. (2004).
the dominant contributor in the wastewater effluent, the risk quotient was higher than 1, implicating a risk to the aquatic environment.
3.3.
Removal efficiency of conventional treatment
The removal efficiency during the primary treatment was low, indicating no significant adsorption of target compounds to the particles removed in this stage (Fig. 4). Most of the pharmaceuticals and consumer products have log Kow values of less than 3.0, so they are not expected to adsorb significantly to the particles. Other pharmaceuticals with higher Kow values, such as gemfibrozil, have much lower pKa values than the pH of wastewater. Therefore, they are dissociated and expected to be >98% in the aqueous phase (Thomas and Foster, 2005), and not bound to the particles. During the secondary treatment, the average removal rate for different compounds ranged from 12% to 100% (Fig. 4). Caffeine, bezafibrate, trimethoprim and DEET were effectively removed, with the average efficiency of 100 0%, 88 12%, 76 24% and 69 21%, respectively. These results were comparable with those found in the previous studies (Ternes, 1998; Okuda et al., 2008; Thomas and Foster, 2005; Castiglioni et al., 2006). Caffeine was proved to be readily biodegradable (Okuda et al., 2008; Thomas and Foster, 2005; Huerta-Fontela
120
Primary treatment Secondary treatment
100
A
40 20
-40 0
20
40
60
80
100
Pharmaceutical Compostion (%) Fig. 3 – Composition profiles of target pharmaceuticals in secondary effluent samples from four WWTPs in Beijing.
-60
Pharmaceutical Fig. 4 – Removal efficiencies of target pharmaceuticals during the conventional treatment.
SP
CBZ
GF
CA
-20
BF
0 TP
B
60
CF
C
80
DEET
Others GF MTP IM CBZ SP TP DF DEET
Removal efficiency (%)
WWTP
D
MTP
Similar to the influent samples, ketoprofen was below the LOQ in all the secondary effluent samples. Nalidixic acid and chloramphenicol were detected only in one WWTP, with the concentration of 8.1 and 19 ng L1, respectively. The mean concentrations of the other 12 compounds ranged from 5 to 200 ng L1 (Fig. 2). Diclofenac, N,N-diethyl-meta-toluamide, trimethoprim, sulpiride and indometacin showed high concentrations in the secondary effluents. Carbamazepine and metoprolol followed, with the concentrations ranging from 69 to 120 ng L1, and 60 to 108 ng L1, respectively. Other compounds, such as caffeine, gemfibrozil and mefenamic acid, occurred at the lowest levels. Despite of a wide variation of trimethoprim from different WWTPs, the composition profiles of target pharmaceuticals in secondary effluents from the four WWTPs were quite similar (Fig. 3). The concentration levels of most pharmaceuticals and consumer products detected in the secondary effluent were also lower than those reported in the Europe. They were over 100 ng L1, in some cases even up to 500 ng L1 in the wastewater effluents of the European countries (Santos et al., 2007; Gomez et al., 2007; Ternes, 1998; Vieno et al., 2007b). While in the present study, 10 out of 15 compounds were less than 100 ng L1, and none of them exceeded 400 ng L1 in any effluent samples (Fig. 2). Our results were in agreement with those in Japan (Nakada et al., 2006), Korea (Kim et al., 2007) and some other cities of China (Xu et al., 2007; Gulkowska et al., 2008; Chen et al., 2008). For instance, the concentrations of chloramphenicol in the effluents of 4 WWTPs in Guangzhou were
Table 4 – Measured concentrations in the effluent samples, predicted no-effect concentrations and risk quotients (MEC/PNEC) of target compounds.
IM
Secondary effluent
DF
3.2.
423
water research 44 (2010) 417–426
et al., 2008; Gomez et al., 2007). The removal rate of bezafibrate was found to be 87% in six Italian WWTPs in summer time (Castiglioni et al., 2006), very similar to that observed in our study. The concentrations of DEET were decreased by more than 80% during the biological treatment in the WWTP of Japan (Okuda et al., 2008), slightly better than our results. A second group of pharmaceuticals, including three antiinflammatory drugs, clofibric acid, gembrozil, metoprolol and sulpiride, had lower removal rates with large variation in different WWTPs studied. For instance, 28–53% of diclofenac, a representative of the anti-inflammatory drugs, was removed by secondary treatment in the WWTPs, which was between 26% in Finland (Lindqvist et al., 2005) and 69% in Germany (Ternes, 1998). The elimination of these compounds may be highly dependent on the configurations and operation conditions of individual WWTP as well as wastewater characteristics, and thus no definitive conclusion could be reached. Higher load of carbamazepine was found in the secondary effluent than in the primary effluent, indicating negative removal efficiency during the secondary treatment. Some carbamazepine was found to be excreted as the form of conjugates (Vieno et al., 2007b), which was biodegraded to carbamazepine by enzymatic processes during the secondary treatment, resulting in additional amounts of carbamazepine in the secondary effluent. However, as the calculations of all the removal efficiencies were based on grab samples that were not sampled with a hydraulic lag in the present study, some error might be brought in due to diurnal variation of the concentration. Therefore, the present study only provided a snapshot of the removal of pharmaceuticals and consumer products in the WWTPs of Beijing. To better illustrate that, 24-h composite samples that are lagged by HRT should be collected and analyzed in further studies. It has been reported that high HRT (>12 h) and SRT (>10 d) may contribute to an increased removal rate of pharmaceuticals (Jones et al., 2007; Vieno et al., 2007b). In the present study, the WWTP C, in which the HRT was higher than the others, was the best in removing these compounds, due to increased contact time of target compounds and the microorganisms. On the other hand, the different SRTs did not have significant effects on the removal efficiency, probably because the SRTs in all the four WWTPs were relatively high (>10 d), and without large differences. In addition, it is noteworthy that the WWTP C employed oxidation ditches, which showed better removal of natural estrogens and estrogenic activity than A/O (Hashimoto et al., 2007). It also could be the reason for the higher removal efficiencies in the WWTP C. Further investigation for different types of WWTPs is necessary to confirm the results mentioned above.
3.4.
Removal efficiency in advanced treatment processes
The removal efficiencies of the pharmaceuticals during the SF, UF/ozonation, as well as MF/RO treatment in three corresponding WWTPs are listed in Table 5. Generally, sand filtration was not effective for these compounds. Only trimethoprim, DEET and gemfibrozil were removed slightly during this treatment process. It should be noticed that these compounds were efficiently removed in the secondary treatment, indicating that the biodegradation on
Table 5 – Removal efficiencies (%) of target pharmaceuticals and consumer products by advanced treatment processes in studied WWTPs. Compound
DEET CF TP BF CA GF DF IM MA MTP CBZ SP
WWTP A
WWTP B
WWTP D
UF
Ozone
SF
MF/RO
0–50 <0 0–50 0–50 <0 0–50 0–50 0–50 0–50 0–50 <0 0–50
50–80 50–80 >90 0–50 50–80 80–90 >90 >90 80–90 80–90 >90 >90
0–50 <0 80–90 0–50 <0 50–80 <0 <0 <0 <0 0–50 0–50
>90 50–80 >90 >90 80–90 >90 >90 >90 0–50 >90 >90 >90
the biofilm present on the sand particle, rather than the removal with particles, may be the main reason for their elimination (Gobel et al., 2007). The results showed that ozonation is effective in removing most of the target compounds, probably due to the operation conditions employed in WWTP A (ozone dosage: 5 mg L1, contact time: 15 min). Carbamazepine, diclofenac, indomethacin, sulpiride and trimethoprim were significantly eliminated, with the removal rates of above 95%. The double bond in the azepine ring of carbamazepine and pyrrole ring of indomethacin, and the non-protonated amine of diclofenac and trimethoprim were susceptible to ozone attack (Vieno et al., 2007a; Nakada et al., 2007; Westerhoff et al., 2005). The removal efficiencies of DEET and metoprolol were modest. The amide group, which is not reactive with ozone, could be the reason for the modest removal of DEET (Nakada et al., 2007). Low removal efficiencies were found for bezafibrate, clofibric acid, as well as caffeine. Only 14% of bezafibrate disappeared in the ozone process, consistent with its low rate constants with ozone (590 50 M1 S1, Huber et al., 2003). The reaction site of bezafibrate is the R-oxysubstituent (–O–C(CH3)2COOH) on one of the aromatic rings. However, as the pKa of bezafibrate is 3.6, the R-oxysubstituent cannot be deprotonated and consequently the overall rate constant at pH > 4 is much lower (Huber et al., 2003). It should be noticed that during the ozonation, most of the pharmaceuticals were not mineralized but transformed to the oxidation products. For instance, three oxidation products containing quinazoline-based functional groups were identified during the ozonation of CBZ (Mcdowell et al., 2005). The good performance of ozonation in the present study was consistent with Ternes et al. (2003), Huber et al. (2005) and Okuda et al. (2008). When 5 mg L1 ozone was applied to the effluent of a municipal WWTP in Germany (contact time: 18 min), target compounds, such as trimethoprim, carbamazepine, indomethacin, clofibric acid, were removed by more than 50% (Ternes et al., 2003). Huber et al. (2005) conducted a pilot study on the oxidation of pharmaceuticals during ozonation of conventional activated sludge (CAS) and membrane bio-reactor (MBR) effluents with various ozone dosages, and found that macrolide and sulfonamide antibiotics, estrogens, and acidic pharmaceuticals diclofenac,
water research 44 (2010) 417–426
a
Tertiary treatment Conventional treatment
120 100 80 60 40 20
SP
CBZ
MTP
IM
DF
GF
CA
BF
TP
CP
-20
CF
0 DEET
Removal Contribution (%)
Pharmaceutical
b 180
Tertiary treatment Secondary treatment Primary treatment
160 140 120 100 80 60 40 20
-40
SP
CBZ
MTP
IM
DF
GF
CA
BF
TP
CF
0 -20
DEET
naproxen and indomethacin were oxidized by more than 90–99% for ozone doses 2 mg L1 in all effluents. The elimination by ultrafiltration in the WWTP A was low for all the investigated compounds. The molecular weight cutoff (MWCO) of UF membranes was much higher than 1000 Da, thus UF membranes showed poor retention of all the investigated pharmaceuticals, of which the molecular weight are less than 400 Da. The removal of individual target compound was less than 50%, and might be due to the adsorption onto the membrane. It has been also demonstrated that UF membrane typically had less than 40% retention of 27 PPCPs, and the mass balances calculated based on the concentration of each compound in feed, permeate and retentate showed the observed retention was significantly governed by adsorption (Yoon et al., 2006). In contrast, MF/RO employed in WWTP D was very effective. In the effluent of MF/RO, all the target compounds except caffeine were not detected. Generally, one or combination of three basic mechanisms could be involved during the rejection of solute by NF/RO membrane: steric effect, charge exclusion and adsorption (Radjenovic et al., 2008). For most pharmaceuticals, the rejections were considered to be dominated by steric interaction in ‘‘tight’’ NF or RO membrane filtration (Nghiem et al., 2005; Radjenovic et al., 2008). As most investigated compounds have molecular weights about 200–400 Da, smaller than MWCO of RO membrane applied, excellent rejection of most pharmaceuticals by RO membrane was observed in this study as well as in previous studies (Kimura et al., 2004; Al-Rifai et al., 2007; Radjenovic et al., 2008). Besides, membrane fouling and the presence of organic matter in the wastewater effluents likely contributed to higher rejections of pharmaceuticals, especially for some hydrophobic ionogenic compound (Nghiem and Coleman, 2008; Comerton et al., 2008). Nevertheless, the rejections of two compounds, caffeine and mefenamic acid were slightly lower (i.e. 50–80% and 0–50%, respectively). The concentration of mefenamic acid in feed wastewaters of MF/RO membrane process was very low, only a bit higher than its LOQ in the wastewater effluent, which could be the reason for the low rejection rate. The low retention of caffeine in the present study was inaccordance with Drewes et al. (2005). They found that in two full-scale RO facilities, target EDCs and PPCPs were efficiently rejected to below detection limit except for caffeine, still detected in the permeates. The physiochemical properties might explain the low rejection rate of caffeine. As a representative of hydrophilic and non-ionic compounds, the rejection driven by charge exclusion and adsorption is negligible, and steric exclusion is solely responsible for the retention of caffeine (Nghiem et al., 2005). However, the molecular weight of caffeine is 195 Da, smaller than other target compounds, and might result in the decreased removal efficiency during the RO membrane filtration process. Compared to the other two, the WWTPs employing ozone and RO membrane filtration as advanced treatment were more efficient in removing pharmaceuticals. For these WWTPs, the advanced treatment made a significant contribution to the total elimination of most pharmaceuticals (Fig. 5). Therefore, the utility of efficient advanced treatment could be considered as a tool to reduce pharmaceuticals in the
Contribution to removal efficiency (%)
424
-60
Pharmaceutical Fig. 5 – Contributions of primary, secondary (or conventional treatment) and tertiary treatment to the total elimination of selected pharmaceuticals in WWTP A (a) and WWTP D (b).
municipal wastewater treatment plants. However, the problems of membrane fouling and further treatment or disposal of retentate challenge the application of RO membrane filtration (Van der Bruggen et al., 2008). For ozonation, as most of the pharmaceuticals could not be mineralized, and oxidation products are formed from parent pharmaceutical compounds (Mcdowell et al., 2005), more research is required to identify the oxidation products and their potential toxicity during the partial oxidation process (Nakada et al., 2007). Besides, economic feasibility should be evaluated by estimating the energy consumption and investment and operation costs for both advanced treatment processes (Joss et al., 2008).
4.
Conclusion
13 out of 15 pharmaceuticals and consumer products from eight classes were detected at four WWTPs in Beijing, China. The concentrations of most compounds in the influent and secondary effluent were lower than those reported in the USA and Europe, but consistent with the production profile of the
water research 44 (2010) 417–426
pharmaceuticals in China. According to the result of risk assessment for the secondary effluent, only diclofenac might pose a risk to the aquatic environment. The removal efficiencies by the conventional treatment varied for different compounds, depending on their chemical structures, physiochemical properties, as well as the specific treatment processes utilized at each WWTP. Further removal could be achieved by adopting some advanced treatment processes, such as ozonation and MF/RO. However, others, such as sand filtration, showed low efficiency in removing these compounds from secondary effluent.
Acknowledgement This study was supported by the National Science Fund for Distinguished Young Scholars (No. 50625823).
Appendix. Supplementary data Supplementary information related to this article can be found at doi:10.1016/j.watres.2009.07.010.
references
Al-Rifai, J.H., Gabelish, C.L., Schafer, A.I., 2007. Occurrence of pharmaceutically active and non-steroidal estrogenic compounds in three different wastewater recycling schemes in Australia. Chemosphere 69 (5), 803–815. Bolton, S., Null, G., 1981. Caffeine: psychological effects, use and abuse. Orthomolecular Psychiatry 10 (3), 202–211. 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 (1), 357–363. Chen, H.C., Wang, P.L., Ding, W.H., 2008. Using liquid chromatography–ion trap mass spectrometry to determine pharmaceutical residues in Taiwanese rivers and wastewaters. Chemosphere 72 (6), 863–869. China Medicine Economic Information Net (CMEIN), 2005. Chinese Medical Statistical Yearbook. Products section. China Medicine Economic Information Net (CMEIN), Beijing, China. Comerton, A.M., Andrews, R.C., Bagley, D.M., Hao, C.Y., 2008. The rejection of endocrine disrupting and pharmaceutically active compounds by NF and RO membranes as a function of compound and water matrix properties. Journal of Membrane Science 313 (1–2), 323–335. Daughton, C.G., Ternes, T.A., 1999. Pharmaceuticals and personal care products in the environment: agents of subtle change? Environmental Health Perspectives 107 (S6), 907–938. Drewes, J.E., Bellona, C., Oekekoven, M., Xu, P., Kim, T.U., Amy, G., 2005. Rejection of wastewater-derived micropollutants in high-pressure membrane applications leading to indirect potable reuse. Environmental Progress 24 (4), 400–409. European Environment Agency, 1998. Environmental Risk Assessment – Approaches, Experiences and Information Source. European Environment Agency, London, pp. 68–86. Ferrari, B., Paxeus, N., Giudice, R.L., Pollio, A., Grric, J., 2003. Ecotoxicological impact of pharmaceuticals found in treated
425
wastewaters: study of carbamazepine, clofibric acid, and diclofenac. Ecotoxicology and Environmental Safety 55 (3), 359–370. Gobel, A., Mcardell, C.S., Joss, A., Siegrist, H., Giger, W., 2007. Fate of sulfonamides, macrolides, and trimethoprim in different wastewater treatment technologies. Science of the Total Environment 372 (2–3), 361–371. Gomez, M.J., Bueno, M.J.M., Lacorte, S., Fernandez-Alba, A.R., Aguera, A., 2007. Pilot survey monitoring pharmaceuticals and related compounds in a sewage treatment plant located on the Mediterranean coast. Chemosphere 66 (6), 993–1002. Grung, M., Kallqvist, T., Sakshaug, S., Skurtveit, S., Thomas, K.V., 2008. Environmental assessment of Norwegian priority pharmaceuticals based on the EMEA guideline. Ecotoxicology and Environmental Safety 71 (2), 328–340. Gulkowska, A., Leung, H.W., So, M.K., Taniyasu, S., Yamashita, N., Yeung, L.W.Y., Richardson, B.J., Lei, A.P., Giesy, J.P., Lam, P.K.S., 2008. Removal of antibiotics from wastewater by sewage treatment facilities in Hongkong and Shenzhen, China. Water Research 42 (1–2), 395–403. Hashimoto, T., Onda, K., Nakamura, Y., Tada, K., Miya, A., Murakami, T., 2007. Comparison of natural estrogen removal efficiency in the conventional activated sludge process and the oxidation ditch process. Water Research 41 (10), 2117–2126. Huber, M.M., Canonica, S., Park, G.Y., Gunten, U.V., 2003. Oxidation of pharmaceuticals during ozonation and advanced oxidation processes. Environmental Science and Technology 37 (5), 1016–1024. Huber, M.M., Gobel, A., Joss, A., Hermann, N., Loffler, D., Mcardell, C.S., Ried, A., Siegrist, H., Ternes, T.A., Gunten, U.V., 2005. Oxidation of pharmaceuticals during ozonation of municipal wastewater effluents: a pilot study. Environmental Science and Technology 39 (11), 4290–4299. Huerta-Fontela, M., Galceran, M.T., Martin-Alonso, J., Ventura, F., 2008. Occurrence of psychoactive stimulatory drugs in wastewaters in north-eastern Spain. Science of the Total Environment 397 (1–3), 31–40. Huschek, G., Hansen, P.D., Maurer, H.H., Krengel, D., Kayser, A., 2004. Environmental risk assessment of medicinal products for human use according to European Commission recommendations. Environmental Toxicology 19 (3), 226–240. Jjemba, P.K., 2006. Excretion and ecotoxicity of pharmaceutical and personal care products in the environment. Ecotoxicology and Environmental Safety 63 (1), 113–130. Jones, O.A.H., Voulvoulis, N., Lester, J.N., 2005. Human pharmaceuticals in wastewater treatment processes. Critical Reviews in Environmental Science and Technology 35 (4), 401–427. Jones, O.A.H., Voulvoulis, N., Lester, J.N., 2007. The occurrence and removal of selected pharmaceutical compounds in a sewage treatment works utilising activated sludge treatment. Environmental Pollution 145 (3), 738–744. Joss, A., Siegrist, H., Ternes, T.A., 2008. Are we about to upgrade wastewater treatment for removing organic micropollutants. Water Science and Technology 57 (2), 251–255. Kasprzyk-Hordern, B., Dinsdale, R.M., Guwy, A.J., 2009. The removal of pharmaceuticals, personal care products, endocrine disruptors and illicit drugs during wastewater treatment and its impact on the quality of receiving waters. Water Research 43 (2), 363–380. Khan, S.J., Wintgens, T., Sherman, P., Zaricky, J., Schafer, A.I., 2004. Removal of hormones and pharmaceuticals in the advanced water recycling demonstration plant in Queensland, Australia. Water Science and Technology 50 (5), 15–22. Khan, S.J., Ongerth, J.E., 2004. Modelling of pharmaceutical residues in Australian sewage by quantities of use and fugacity calculations. Chemosphere 54 (3), 355–367.
426
water research 44 (2010) 417–426
Kim, S.D., Cho, J., Kim, I.S., Vanderford, B.J., Snyder, S.A., 2007. Occurrence and removal of pharmaceuticals and endocrine disruptors in South Korean surface, drinking, and waste waters. Water Research 41 (5), 1013–1021. Kimura, K., Toshima, S., Amy, G., Watanabe, Y., 2004. Rejection of neutral endocrine disrupting compounds (EDCs) and pharmaceutical active compounds (PhACs) by RO membranes. Journal of Membrane Science 245 (1–2), 71–78. Lishman, L., Smyth, S.A., Sarafin, K., Kleywegt, S., Toito, J., Peart, T., Lee, B., Servos, M., Beland, M., Seto, P., 2006. Occurrence and reductions of pharmaceuticals and personal care products and estrogens by municipal wastewater treatment plants in Ontario, Canada. Science of the Total Environment 367 (2–3), 544–558. Lindqvist, N., Tuhkanen, T., Kronberg, L., 2005. Occurrence of acidic pharmaceuticals in raw and treated sewages and in receiving waters. Water Research 39 (11), 2219–2228. Mcdowell, D.C., Huber, M.M., Wagner, M., Gunten, U.V., Ternes, T.A., 2005. Ozonation of carbamazepine in drinking water: identification and kinetic study of major oxidation products. Environmental Science and Technology 39 (20), 8014–8022. 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. Water Research 40 (17), 3297–3303. 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. Water Research 41 (19), 4372–4382. Nghiem, L.D., Schafer, A.I., Elimelech, M., 2005. Pharmaceutical retention mechanisms by nanofiltration membrane. Environmental Science and Technology 39 (19), 7698–7705. Nghiem, L.D., Coleman, P.J., 2008. NF/RO filtration of the hydrophobic ionogenic compound triclosan: transport mechanisms and the influence of membrane fouling. Separation and Purification Technology 62 (3), 709–716. Niwa, T., Inoue, S., Shiraga, T., Takagi, A., 2005. No inhibition of cytochrome P450 activities in human liver microsomes by sulpiride, an antipsychotic drug. Biological and Pharmaceutical Bulletin 28 (1), 188–191. Okuda, T., Kobayashi, Y., Nagao, R., Yamashita, N., Tanaka, H., Tanaka, S., Fujii, S., Konishi, C., Houwa, I., 2008. Removal efficiency of 66 pharmaceuticals during wastewater treatment process in Japan. Water Science and Technology 57 (1), 65–71. Paxeus, N., 2004. Removal of selected non-steroidal antiinflammatory drugs (NSAIDs), gemfibrozil, carbamazepine, beta-blockers, trimethoprim and triclosan in conventional wastewater treatment plants in five EU countries and their discharge to the aquatic environment. Water Science and Technology 50 (5), 253–260. Quinn, B., Gagne, F., Blaise, C., 2008. An investigation into the acute and chronic toxicity of eleven pharmaceuticals (and their solvents) found in wastewater effluent on the cnidarian, Hydra attenuata. Science of the Total Environment 389 (2–3), 306–314. Radjenovic, J., Petrovic, M., Ventura, F., Barcelo, D., 2008. Rejection of pharmaceuticals in nanofiltration and reverse osmosis
membrane drinking water treatment. Water Research 42 (14), 3601–3610. Santos, J.L., Aparicio, I., Alonso, E., 2007. Occurrence and risk assessment of pharmaceutically active compounds in wastewater treatment plants. A case study: Seville city (Spain). Environment International 33 (4), 596–601. Sui, Q., Huang, J., Deng, S. B., Yu, G. Rapid determination of pharmaceuticals from multiple therapeutic classes in wastewater by solid-phase extraction and ultra-performance liquid chromatography tandem mass spectrometry. Chinese Science Bulletin, in press. Ternes, T.A., 1998. Occurrence of drugs in German sewage treatment plants and rivers. Water Research 32 (11), 3245–3260. Ternes, T.A., Meisenheimer, M., Mcdowell, D., Sacher, F., Brauch, H.J., Haist-gulde, B., Preuss, G., Wilme, U., Zulei-seibert, N., 2002. Removal of pharmaceuticals during drinking water treatment. Environmental Science and Technology 36 (17), 3855–3863. Ternes, T.A., Stuber, J., Herrmann, N., Mcdowell, D., Ried, A., Kampmann, M., Teiser, B., 2003. Ozonation: a tool for removal of pharmaceuticals, contrast media and musk fragrances from wastewater? Water Research 37 (8), 1976–1982. Thomas, P.M., Foster, G.D., 2005. Tracking acidic pharmaceuticals, caffeine, and triclosan through the wastewater treatment process. Environmental Toxicology and Chemistry 24 (1), 25–30. Vanderford, B.J., Snyder, S.A., 2006. Analysis of pharmaceuticals in water by isotope dilution liquid chromatography/tandem mass spectrometry. Environmental Science and Technology 40 (23), 7312–7320. Van der Bruggen, B., Manttari, M., Nystrom, M., 2008. Drawbacks of applying nanofiltration and how to avoid them: a review. Separation and Purification Technology 63 (2), 251–263. Vieno, N.M., Harkki, H., Tuhkanen, T., Kronberg, L., 2007a. Occurrence of pharmaceuticals in river water and their elimination in a pilot-scale drinking water treatment plant. Environmental Science and Technology 41 (14), 5077–5084. Vieno, N., Tuhkanen, T., Kronberg, L., 2007b. Elimination of pharmaceuticals in sewage treatment plants in Finland. Water Research 41 (5), 1001–1012. Watkinson, A.J., Murby, E.J., Costanzo, S.D., 2007. Removal of antibiotics in conventional and advanced wastewater treatment: implications for environmental discharge and wastewater recycling. Water Research 41 (18), 4164–4176. 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 (17), 6649–6663. Xu, W.H., Zhang, G., Li, X.D., Zou, S.C., Li, P., Hu, Z.H., Li, J., 2007. Occurrence and elimination of antibiotics at four sewage treatment plants in the Pearl River Delta, South China. Water Research 41 (19), 4526–4534. Yoon, Y., Westerhoff, P., Snyder, S.A., Wert, E.C., 2006. Nanofiltration and ultrafiltration of endocrine disrupting compounds, pharmaceuticals and personal care products. Journal of Membrane Science 270 (1–2), 88–100.
water research 44 (2010) 427–438
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Application of the combination index (CI)-isobologram equation to study the toxicological interactions of lipid regulators in two aquatic bioluminescent organisms Ismael Rodea-Palomaresa,1, Alice L. Petreb,1, Karina Boltesb, Francisco Legane´sa, Jose´ Antonio Perdigo´n-Melo´nb, Roberto Rosalb, Francisca Ferna´ndez-Pin˜asa,* a
Departamento de Biologı´a, Facultad de Ciencias, Universidad Auto´noma de Madrid, 2 Darwin Street, Cantoblanco, 28049 Madrid, Spain Departamento de Ingenierı´a Quı´mica, Universidad de Alcala´, Alcala´ de Henares, E-28871 Madrid, Spain
b
article info
abstract
Article history:
Pharmaceuticals in the aquatic environment do not appear singly and usually occur as
Received 26 March 2009
complex mixtures, whose combined effect may exhibit toxicity to the aquatic biota. We
Received in revised form
report an environmental application of the combination index (CI)-isobologram equation,
13 July 2009
a method widely used in pharmacology to study drug interactions, to determine the nature
Accepted 18 July 2009
of toxicological interactions of three fibrates toward two aquatic bioluminescent organ-
Available online 25 July 2009
isms, Vibrio fischeri and the self-luminescent cyanobacterial recombinant strain Anabaena CPB4337. The combination index-isobologram equation method allows computerized
Keywords:
quantitation of synergism, additive effect and antagonism. In the Vibrio test, the fibrate
Antagonism
combinations showed antagonism at low effect levels that turned into an additive effect or
Combination index-isobologram
synergism at higher effect levels; by contrast, in the Anabaena test, the fibrate combinations
equation
showed a strong synergism at the lowest effect levels and a very strong antagonism at high
Cyanobacterium
effect levels. We also evaluated the nature of the interactions of the three fibrates with a real
Fibrates
wastewater sample in the cyanobacterial test. We propose that the combination index-
Synergism
isobologram equation method can serve as a useful tool in ecotoxicological assessment. ª 2009 Elsevier Ltd. All rights reserved.
Vibrio fischeri
1.
Introduction
Fibrates and statins (HMG-CoA reductase inhibitors) are the main lipid-lowering drugs prescribed either alone or in combination therapy in order to decrease plasma cholesterol levels and reduce the incidence of coronary heart disease. Although partially displaced by statins, the total number of fibrate prescriptions is in constant increase in the United States (Holoshitz et al., 2008). Fibric acids are the active forms
of fibrates and belong to the nuclear receptor superfamily of ligand-activated transcription factors. Gemfibrozil and fenofibrate are the fibrates currently marketed in the US, whereas bezafibrate is also available in Europe and other developed countries (Lambropoulou et al., 2008). Fenofibric acid, 2-[4-(4chlorobenzoyl)phenoxy]-2-methylpropanoic acid, is the active metabolite of fenofibrate, the inactive prodrug marketed and dispensed. Gemfibrozil, 5-(2,5-dimethylphenoxy)-2,2-dimethylpentanoic acid and bezafibrate, p-[4-[chlorobenzoylamino-
* Corresponding author. Tel.: þ34 9 1497 8176; fax: þ34 9 1497 8344. E-mail address:
[email protected] (F. Ferna´ndez-Pin˜as). 1 Both authors contributed equally to this work. 0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2009.07.026
428
water research 44 (2010) 427–438
ethyl]-phenoxy]-b-methylpropionic acid, are also fibric acid derivatives with similar pharmacokinetic behaviour (Miller and Spence, 1998). The occurrence of lipid regulators in the discharge of treated urban and municipal wastewater has been relatively well documented. Bezafibrate has been detected in effluents of two British STP with averages up to 230 ng/L (KasprzykHordern et al., 2009). Metcalfe et al. (2003) found around 1 mg/L of gemfibrozil in effluents of Canadian STP, whereas fenofibrate has been reported in concentrations up to 0.5 mg/L in the influent of several Brazilian STP (Stumpf et al., 1999). Andreozzi et al. (2003) found lipid regulators in the effluent of several European STP at concentrations up to 4.76 mg/L (gemfibrozil), 1.07 mg/L (bezafibrate) and 0.16 mg/L (fenofibrate). Rosal et al. (2008), reported the occurrence of bezafibrate and gemfibrozil at levels of 139 and 608 ng/L respectively in the effluent of a Spanish STP. In the same plant Rodrı´guez et al. (2008) found 165 ng/L of fenofibric acid, 61 ng/L of bezafibrate and 143 ng/L of gemfibrozil. It is also significant that removal efficiencies observed in current STP are not always high. Fent et al. (2006) reported maximum removal rates of 50–75% for fenofibric acid and gemfibrozil and somewhat higher for bezafibrate, although for the later, efficiencies below 15% have also been reported. Stumpf et al. (1999) reported a 45% removal of fenofibric acid by an activated sludge conventional treatment. KasprzykHordern et al. (2009) encountered an average degradation of bezafibrate not higher than 67%. On the other hand, Castiglioni et al. (2006) reported that the removal efficiency of bezafibrate during an activated sludge treatment greatly varied from 15% in winter to 87% in summer. At measured environmental concentrations as those reported above (mostly in the ng/L and mg/L range), many studies have shown that the risk of acute toxicity is unlikely (Fent et al., 2006; Han et al., 2006; Borgmann et al., 2007); however, there is a lack of data on chronic toxicity effects. Moreover, pharmaceuticals in the aquatic environments occur as complex mixtures from different classes, not as single contaminants (Gros et al., 2007); thus, although the concentration of individual pharmaceuticals is low, their mixture could prove ecotoxicologically significant (Brain et al., 2004). Current methods of risk assessment usually focus on the assessment of single chemicals, which may underestimate the risk associated with toxic action of mixtures; probably for this reason, in the last years there is an increasing number of studies dealing with complex mixtures of pharmaceuticals (Cleuvers, 2003, 2004; Crane et al., 2006; Han et al., 2006; Borgmann et al., 2007; Christensen et al., 2007; Pomati et al., 2008; Quinn et al., 2009). However, assessment of combined toxicities is not an easy issue. Basically, two different models are in use for the prediction of mixture toxicity, i.e., concentration addition, when pharmaceuticals have a similar mode of toxic action, and response addition or independent action, when pharmaceuticals have different modes of toxic action (Cleuvers, 2003; Teuschler, 2007). However, toxicological interactions, synergisms or antagonisms, between the pharmaceuticals and their effects can occur independently of mode of action; moreover, in most cases, the pharmacological mechanisms of action is known but the toxic mode of action may remain unknown (Cleuvers, 2003; Chou, 2006). In an effort
to overcome this limitation, we report an environmental application of a method widely used in pharmacology to interpret drug interactions; this method, termed as the median-effect/combination index (CI)-isobologram equation (Chou, 2006) allows quantitative determinations of chemical interactions where CI <1, ¼1 and >1 indicate synergism, additive effect and antagonism, respectively. One important property of the method is that previous knowledge of the mechanisms of action of each chemical is not required. Besides, the method takes into account both the potency and the shapes of the dose-effect curve of each chemical. The method has been computerized allowing an automated simulation of synergism and antagonism at different concentrations and at different effect levels of the chemicals in a mixture. The aim of our study was to assess the nature of the toxicological interactions of three fibrates, gemfibrozil, bezafibrate and fenofibric acid, by the method of combination index (CI)-isobologram equation. The three pharmaceuticals were used singly or in two- and three-drug combinations. As toxicity endpoint we have chosen the bioluminescent response of two prokaryotes, the naturally luminescent Vibrio fischeri and the recombinant bioluminescent cyanobacterium Anabaena sp. PCC 7120 CPB4337 (hereinafter, Anabaena CPB4337), both bioluminescent organisms have proved very useful in evaluating the toxicity of individual fibrates in a previous study (Rosal et al., 2009). For Anabaena CPB4337, we also evaluated the nature of the interactions of the three fibrates with a wastewater sample from a local STP, which already proved very toxic to the cyanobacterium (Rosal et al., 2009).
2.
Materials and methods
2.1.
Materials
Gemfibrozil (þ99%) and bezafibrate (þ98%) were purchased from Sigma–Aldrich. Fenofibric acid was produced from fenofibrate (Sigma–Aldrich, þ99% purity) by hydrolysis. A suspension of fenofibrate in isopropanol (30 wt.%, 400 mL) was refluxed during 4 h with an aqueous sodium hydroxide solution (2.0 M, 200 mL). After cooling to less than 70 C, a solution of hydrochloric acid (1.0 M, 325 mL) was slowly added while keeping the temperature over 60 C. The product crystallized after cooling and keeping at room temperature during 4 or more h. The product was filtered and rinsed with water and dried overnight at 60 C under nitrogen. The purity of the product was over 97% checked by HPLC. Solubility of acidic drugs in water is strongly pH dependent with few data considering this variable. Comerton et al. (2007) reported a solubility of 10.9 mg/L of gemfibrozil in water, but we could solve over 125 mg/L in 2 mM MOPS (3-[N-morpholino] propanesulfonic acid) at pH 6 and higher quantities for the pH at which V. fischeri bioassays were performed. In all cases, we avoided the use of solvents and the upper limit for the concentrations of the studied compounds was their solubility in pure water or wastewater at the pH of the bioassay. Wastewater samples were collected from the secondary clarifier of a STP located in Alcala´ de Henares (Madrid) that
429
water research 44 (2010) 427–438
receives domestic wastewater with a minor contribution of industrial effluents from facilities located near the city. This STP used a conventional activated sludge treatment and has been designed for a total capacity of 375,000 equivalent inhabitants with a maximum flow rate of 3000 m3/h. In a recent previous study (Rosal et al., 2009), we found that this wastewater was very toxic to Anabaena cells with a wastewater dilution as low as 0.11 causing 50% luminescence inhibition (wastewater EC50).
2.2.
Toxicity tests
Bioassays with the photo-luminescent bacteria Vibrio fischeri were carried out according to ISO 11348-3 standard protocol (ISO, 2007). This bioassay measures, during the prescribed incubation period, the decrease in bioluminescence induced in the cell metabolism due to the presence of a toxic substance. The bacterial assay used the commercially available Biofix Lumi test (Macherey-Nagel, Germany). The bacterial reagent is supplied freeze-dried (Vibrio fischeri NRRL-B 11177) and was reconstituted and incubated at 3 C for 5 min before use. The desired pH was set by using NaOH or HCl. The analysis media was 0.34 M NaCl (2% w/v) and tests were performed at 15 C and the measurements of light were made using a luminometer (Optocomp I). The effect of toxicants or toxicant mixtures (i.e., fibrates or fibrate combinations) was measured as percent inhibition with respect to the light emitted under test conditions in the absence of any toxic influence. Toxicity values were routinely obtained after 30 min exposure. Phenol and ZnSO4 7 H2O have been used as toxicity standards and all tests have been replicated to ensure reproducibility. The bioassays using the recombinant bioluminescent cyanobacterium Anabaena CPB4337 were based on the inhibition of constitutive luminescence caused by the presence of any toxic substance (Rodea-Palomares et al., 2009; Rosal et al., 2009). Anabaena CPB4337 was routinely grown at 28 C in the light, ca. 65 mmol photons m2 s1 on a rotary shaker in 50 mL AA/8 (Allen and Arnon, 1955) supplemented with nitrate (5 mM) in 125 ml Erlenmeyer flasks and 10 mg/mL of neomycin sulphate (Nm). Luminescence inhibition-based toxicity assays were performed as follows: 160 mL from five to seven serial dilutions of each tested toxicant or toxicant mixture (i.e.; fibrates or fibrate combinations) plus a control (ddH2O buffered with MOPS at pH 5.8) were disposed in an opaque white 96-well microtiter plates. 40 mL cells, grown as described, were washed twice and resuspended in ddH2O buffered with MOPS at pH 5.8 and were added to the microtiter plate wells to reach a final cell density at OD750 nm of 0.5. The luminescence of each sample was recorded every 5 min for up to 1 h in the Centro LB 960 luminometer. Three independent experiments with duplicate samples were carried out for all Anabaena toxicity assays. CuSO4 has been used as toxicity standard and all tests have been replicated to ensure reproducibility.
2.3.
Experimental design of fibrate combinations
Solutions of gemfibrozil (Gm), bezafibrate (Bz) and fenofibric acid (Fn) prepared as described above were used singly and in two (Bz þ Gm; Fn þ Gm; Fn þ Bz) and three (Fn þ Gm þ Bz)
combinations. Anabaena and Vibrio fischeri cells were treated with serial dilutions of each fibrate individually and with a fixed constant ratio (1:1), based on the individual EC50 values, in their binary and ternary combinations. Five dilutions (serial dilution factor ¼ 2) of each fibrate and combination plus a control were tested in three independent experiments with replicate samples. For evaluating the nature of the interaction of fibrates with wastewater, binary combinations of each fibrate plus wastewater (Fn þ WW; Gm þ WW; Bz þ WW) and a quaternary combination of the three fibrates plus wastewater (Fn þ Gm þ Bz þ WW) were also prepared and tested for Anabaena CPB4337. Anabaena cells were treated with serial dilutions of each fibrate and wastewater individually and with a fixed constant ratio (1:1), based on the individual EC50 values, in their binary and quaternary combinations. Five dilutions (serial dilution factor ¼ 2) of each fibrate and wastewater and their combinations plus a control were tested in three independent experiments with replicate samples. The experimental design is shown in Table 1. All individual fibrate, wastewater and their combination assays were carried out at the same time as recommended by Chou (2006) to maximize computational analysis of data.
2.4. Median-effect and combination index (CI)isobologram equations for determining combined fibrate interactions The results were analyzed using the median-effect/combination index (CI)-isobologram equation by Chou (2006) and Chou and Talalay (1984) which is based on the median-effect principle (mass-action law) (Chou, 1976) that demonstrates that there is an univocal relationship between dose and effect independently of the number of substrates or products and of the mechanism of action or inhibition. This method involved plotting the dose-effect curves for each compound and their combinations in multiple diluted concentrations by using the median-effect equation: fa ¼ fu
m
D Dm
(1)
Where D is the dose, Dm is the dose for 50% effect (e.g., 50% inhibition of bioluminescence or EC50), fa is the fraction affected by dose D (e.g., 0.75 if cell bioluminescence is inhibited by 75%), fu is the unaffected fraction (therefore, fa ¼ 1 fu), and m is the coefficient of the sigmoidicity of the dose-effect curve: m ¼ 1, m > 1, and m < 1 indicate hyperbolic, sigmoidal, and negative sigmoidal dose-effect curve, respectively. Therefore, the method takes into account both the potency (Dm) and shape (m) parameters. If Eq. (1) is rearranged, then: D ¼ Dm½fa=ð1 faÞ
1=m
(2)
The Dm and m values for each fibrate are easily determined by the median-effect plot: x ¼ log (D) versus y ¼ log ( fa/fu) which is based on the logarithmic form of Eq. (1). In the medianeffect plot, m is the slope and log (Dm) is the x-intercept. The conformity of the data to the median-effect principle can be ready manifested by the linear correlation coefficient (r) of the data to the logarithmic form of Eq. (1) (Chou, 2006).
Pure fibrate experiments
Fibrates plus wastewater experiments
Vibrio fischeri Dilutions
1
⁄4 (EC50) 1 ⁄2 (EC50) 1 (EC50) 2 (EC50) 4 (EC50)
Gm
(D)2 (D)1 0.4 8.75 0.8 17.5 1.6 35 3.2 70 6.4 140 Two toxicant combo (D)1 þ (D)2 (1.6:35)
Single toxicant Bz
Fn
(D)3 37.5 75 150 300 600
(D)1 (D)2 2.5 2.5 5 5 10 10 20 20 40 40 Two toxicant combo (D)1 þ (D)2 (1:1)
⁄4 (EC50) ⁄2 (EC50) 1 (EC50) 2 (EC50) 4 (EC50)
0.4 8.75 0.8 17.5 1.6 35 3.2 70 6.4 140 (D)1 þ (D)3 (1.6:150)
1
⁄4 (EC50) ⁄2 (EC50) 1 (EC50) 2 (EC50) 4 (EC50)
0.4 0.8 1.6 3.2 6.4 (D)2 þ (D)3 (35:150)
37.5 75 150 300 600
2.5 5 10 20 30* (D)2 þ (D)3 (1:5)
1
8.75 17.5 35 70 140 Three toxicant combo (D)1 þ (D)2 þ (D)3 (1.6:35:150)
37.5 75 150 300 600
0.4 0.8 1.6 3.2 6.4
37.5 75 150 300 600
1
1
⁄4 (EC50) ⁄2 (EC50) 1 (EC50) 2 (EC50) 4 (EC50) 1
1
⁄4 (EC50) ⁄2 (EC50) 1 (EC50) 2 (EC50) 4 (EC50) 1
8.75 17.5 35 70 140
Anabaena CPB4337
2.5 5 10 20 25* (D)1 þ (D)3 (1:5)
Gm
Single toxicant Bz (D)3 12.5 25 50 100 200
Fn
1
⁄4 (EC50) 1 ⁄2 (EC50) 1 (EC50) 2 (EC50) 4 (EC50)
1
2.5 5 10 20 25*
Gm
(D)1 (D)2 2.5 2.5 5 5 10 10 20 20 40 40 Two toxicant combo (D)1 þ (D)4 (1:0.01)
Bz
WW**
(D)3 12.5 25 50 100 200
(D)4 0.025 0.05 0.1 0.2 0.4
⁄4 (EC50) ⁄2 (EC50) 1 (EC50) 2 (EC50) 4 (EC50)
2.5 5 10 20 40 (D)2 þ (D)4 (1:0.01)
0.025 0.05 0.1 0.2 0.4
12.5 25 50 100 150*
1
2.5 5 10 20 40 (D)3 þ (D)4 (1:0.002)
0.025 0.05 0.1 0.2 0.4
2.5 5 10 20 40 Three toxicant combo (D)1 þ (D)2 þ (D)3 (1:1:5)
12.5 25 50 100 200
1
2.5 5 10 20 40
12.5 25 50 100 200
2.5 5 10 20 40
1
⁄4 (EC50) ⁄2 (EC50) 1 (EC50) 2 (EC50) 4 (EC50)
1
⁄4 (EC50) ⁄2 (EC50) 1 (EC50) 2 (EC50) 4 (EC50)
12.5 25 50 100 200
1
0.025 0.05 0.1 0.2 0.4
Four toxicant combo (D)1 þ (D)2 þ (D)3 þ (D)4 (1:1:5:0.01) 1/8 (EC50) ⁄4 (EC50) 1 ⁄2 (EC50) 1 (EC50) 2 (EC50) 1
1.25 2.5 5 10 20
1.25 2.5 5 10 20
6.25 12.5 25 50 100
0.0125 0.025 0.05 0.1 0.2
For the Anabaena test, the design for the experiment with the wastewater [WW (D4)] sample is also included. The experimental design is based on EC50 ratios as proposed by Chou and Talalay (1984). EC50 is the effective concentration of a toxicant which caused a 50% bioluminescence inhibition. The combination ratio was approximately equal to the EC50 ratio of the combination components (i.e., close to their equipotency ratio). *Upper maximal possible dose due to the solubility limit of fibrates in pure water. **EC50 for wastewater is the dilution which caused 50% luminescence inhibition. (D)1, (D)2 and (D)3 in mg/L, (D)4 is the dilution of wastewater in ddH2O.
water research 44 (2010) 427–438
1
Anabaena CPB4337
Single toxicant Fn
430
Table 1 – Experimental design for determining toxicological interactions of fenofibric acid [Fn (D)1], gemfibrozil [Gm (D)2], bezafibrate [Bz (D)3] and their binary and ternary combinations for Vibrio fischeri and Anabaena CPB4337 bioluminescence tests.
431
water research 44 (2010) 427–438
These parameters were then used to calculate doses of the fibrates and their combinations required to produce various effect levels according to Eq. (1); for each effect level, combination index (CI) values were then calculated according to the general combination index equation for n chemical combination at x% inhibition (Chou, 2006):
n
ðCIÞx ¼
n X
ðDÞj
j¼1
ðDx Þj
¼
P ðDx Þ1n ½Dj = n1 ½D
n X j¼1
ðDm Þj
(3)
n h io1=mj fax j = 1 fax j
where n(CI)x is the combination index for n chemicals (e.g., fibrates) at x% inhibition (e.g., bioluminescence inhibition); (Dx)1n is the sum of the dose of n chemicals that exerts x% P inhibition in combination, {[Dj]/ n1 ½D} is the proportionality of the dose of each of n chemicals that exerts x% inhibition in combination; and (Dm)j {( fax)j/[1 ( fax)j]}1/mj is the dose of each drug alone that exerts x% inhibition. From Eq. (3), CI <1, ¼1 and >1 indicates synergism, additive effect and antagonism, respectively.
2.5.
versus fa, the fraction affected by a particular dose; see Eq. (1)) and polygonograms (a polygonal graphic representation depicting synergism, additive effect and antagonism for three or more drug combinations). Linear regression analyses were computed using MINITAB Release 14 for Windows (Minitab Inc; USA).
Analysis of results
Computer program CompuSyn (Chou and Martin, 2005, Compusyn Inc, USA) was used for calculation of dose-effect curve parameters, CI values, fa-CI plot (plot representing CI
3.
Results
3.1. Toxicological interactions of fibrate combinations in Vibrio fischeri and Anabaena CPB4337 bioluminescence tests Applying the combination index-isobologram method, we evaluated the nature of gemfibrozil (Gm), fenofibric acid (Fn) and bezafibrate (Bz) interactions both in Vibrio fischeri and Anabaena CPB4337 bioluminescence tests. Table 2 shows the dose-effect curve parameters (Dm, m and r) of the three fibrates singly and their binary and ternary combinations, as well as mean combination index (CI) values of fibrate combinations. Dm was the dose required to produce the medianeffect (analogous to the EC50); Dm values for Fn were the lowest both, in Vibrio and Anabaena tests, Dm values for Gm were in the same range for both Vibrio and Anabaena while Bz
Table 2 – Dose-effect relationship parameters and mean combination index (CI) values (as a function of fractional inhibition of luminescence) of gemfibrozil (Gm), fenofibric acid (Fn), and bezafibrate (Bz) individually and of their binary and ternary combinations on Vibrio fischeri and Anabaena CPB4337 bioluminescence tests. Drug combo
Vibrio fischeri Dose-effect parameters Dm
Fn Gm Bz Gm þ Bz Fn þ Bz Fn þ Gm Fn þ Gm þ Bz
mg/L
(mM)
1.45 20.58 252.07 78.20 153.79 9.84 55.69
(4.01) (82.11) (696.46) (234.20) (424.93) (38.74) (166.69)
CI values
m
r
0.78 1.53 1.15 1.54 1.09 1.15 1.23
0.989 0.966 0.975 0.991 0.981 0.973 0.993
EC10
– – – 1.13 0.13 2.98 0.15 0.99 0.17 1.46 0.06
EC50
Add Ant Add Ant
– – – 0.97 0.04 1.71 0.03 0.75 0.05 1.01 0.02
EC90
Add Ant Syn Add
– – – 0.86 0.05 1.17 0.06 0.86 0.08 0.99 0.03
Syn Ant Syn Add
Anabaena CPB4337 Dose-effect parameters Dm mg/L Fn Gm Bz Gm þ Bz Fn þ Bz Fn þ Gm Fn þ Gm þ Bz
8.53 10.69 12.56 19.17 13.92 12.26 6.62
CI values
m
r
0.96 0.81 1.08 0.84 0.76 0.46 0.53
0.971 0.959 0.990 0.972 0.965 0.955 0.960
EC10
EC50
EC90
(mM) (23.62) (42.67) (34.70) (56.88) (38.49) (41.45) (19.45)
– – – 1.06 0.15 0.55 0.06 0.13 0.02 0.09 0.01
Add Syn Syn Syn
– – – 1.57 0.06 1.19 0.04 1.29 0.05 0.57 0.02
Ant Ant Ant Syn
– – – 2.5 0.22 2.59 0.14 12.9 2.33 3.92 0.19
Ant Ant Ant Ant
The parameters m, Dm and r are the antilog of x-intercept, the slope and the linear correlation coefficient of the median-effect plot, which signifies the shape of the dose-effect curve, the potency (EC50), and conformity of the data to the mass-action law, respectively (Chou, 1976; Chou and Talalay, 1984; Chou, 2006). Dm and m values are used for calculating the CI values (Eq. (3)); CI <1, ¼1, and >1 indicate synergism (Syn), additive effect (Add), and antagonism (Ant), respectively. EC10, EC50 and EC90, are the doses required to inhibit bioluminescence 10, 50 and 90%, respectively. Computer software CompuSyn was used for automated calculation and simulation.
432
water research 44 (2010) 427–438
Dm values were an order of magnitude higher in the Vibrio test (Rosal et al., 2009); m was the Hill coefficient used to determine the shape of the dose-response curve, hyperbolic (m ¼ 1), sigmoidal (m > 1) or negative sigmoidal (m < 1); also shown in the table, linear regression correlation coefficients (r-values) of the median-effect plots were >0.95 in all cases, indicating the conformity of the data to the median-effect principle which qualifies for further studies using this method. The Dm and m values for single fibrates and for their combination mixtures were used for calculating synergism or antagonism based on the CI Eq. (3) (Chou, 2006). Fig. 1 shows the fa-CI plot of fibrate interactions both for Vibrio (Fig. 1a) and Anabaena tests (Fig. 1b); the fa-CI plot depicts the CI value
a
3.0
Combination index, CI
2.5
2.0
1.5
1.0
0.5
0.0 0.0
0.2
0.4
0.6
0.8
1.0
Fraction affected, fa
b
3.0
Combination index, CI
2.5
2.0
1.5
versus fa (effect level or fraction of luminescence inhibited by a fibrate singly or in combination with respect to the control) for two (Fn þ Bz; Fn þ Gm and Bz þ Gm) and three fibrate (Fn þ Gm þ Bz) combinations. The fa-CI plot is an effectoriented plot that shows the evolution of the kind of interaction (synergism, antagonism, additive effect) as a function of the level of the effect ( fa) of a particular toxicant on the reference organism ( fa, where ECa ¼ fa 100; i.e., EC10 ¼ f10 100). In the Vibrio test (Fig. 1a), the Bz þ Gm and Fn þ Gm binary combination showed a slight antagonism at very low fa values and slight synergism (Fn þ Gm) or nearly additive effects (Bz þ Gm) at the highest fa values, the Fn þ Bz combination showed a strong antagonism at low effect levels but the antagonism decreased and approached an additive kind of interaction at the highest fa levels; the ternary combination (Fn þ Gm þ Bz) showed a moderate antagonism at low fa values that also turned into a nearly additive effect at fa values above 0.4. Correlation analyses were made between CI values of the fibrate ternary combination and CI values of each of the fibrate binary combinations to determine which binary combination interaction was predominant in the ternary mixture (Table 3); the highest correlation coefficient was found for the Fn þ Bz combination (r ¼ 0.91), suggesting that this combination interaction predominated in the three fibrate mixture. The fa-CI plot of the Anabaena test (Fig. 1b) showed the opposite pattern of interactions as the three binary and the ternary combinations showed from slight to strong synergism at the lowest fa values that turned into a very strong antagonism at fa values over 0.5; the ternary combination (Fn þ Gm þ Bz) closely followed the interaction pattern of the binary Fn þ Gm combination, this is confirmed by the highest correlation coefficient found between the CI values of the ternary combination and the CI values of the Fn þ Gm combination (r ¼ 0.996) which suggests that in the Anabaena test, this particular combination seemed to be the predominant in the ternary toxicological interaction. Selected average CI values for both Vibrio fischeri and Anabaena CPB4337 tests at three representative dose levels (EC10, EC50 and EC90) and the combined effects are summarized in Table 2.
1.0
3.2. Toxicological interactions of wastewater and fibrate combinations in the Anabaena CPB4337 bioluminescence test
0.5
0.0 0.0
0.2
0.4
0.6
0.8
1.0
Fraction affected, fa
Fig. 1 – Combination index plot ( fa-CI plot) for a set of three fibrate combinations: Fn D Bz (–6–), Bz D Gm (–B–), Fn D Gm (–,–) and Fn D Gm D Bz (–;–) for Vibrio fischeri test (a) and Anabaena CPB4337 test (b). CI values are plotted as a function of the fractional inhibition of bioluminescence ( fa) by computer simulation (CompuSyn) from fa [ 0.10 to 0.95. CI <1, [1 and >1 indicates synergism, additive effect and antagonism, respectively. At least three independent experiments with two replicates were used. The vertical bars indicate 95% confidence intervals for CI values based on sequential deletion analysis (SDA) (Chou and Martin, 2005). Fn [ fenofibric acid, Bz [ bezafibrate and Gm [ gemfibrozil.
In a recent previous study (Rosal et al., 2009), we found that a wastewater sample collected from a local STP was very toxic to Anabaena cells with a wastewater dilution of 0.11 causing 50% luminescence inhibition (wastewater EC50). The observed toxicity was attributed to the combined toxicities of over thirty micropollutants, which included fibrates as well as other pharmaceuticals (Rosal et al., 2008). We sought to investigate the nature of the interaction between the wastewater (WW) and the three fibrates in binary (Fn þ WW; Bz þ WW and Gm þ WW) and quaternary (Fn þ Gm þ Bz þ WW) combinations; for these experiments, the wastewater itself was regarded as a toxicant; the experimental design was analogous to the one for the three fibrate interactions and is also shown in Table 1. The r-values of the median-effect plots were >0.95 in all cases, indicating that the data conformed to the medianeffect principle (not shown). Fig. 2 shows the fa-CI plot for each
433
water research 44 (2010) 427–438
Table 3 – Correlation analyses between CI values of fibrate ternary and fibrate D wastewater quaternary combinations ( y) and their binary combinations (x) for Vibrio fischeri and Anabaena CPB4337 tests. Test organism
Combinations
V. fischeri
Fn þ Gm þ Bz
versus
Anabaena CPB4337
Fn þ Gm þ Bz
versus
WW þ Fn þ Gm þ Bz
versus
Regression parameters
Gm þ Bz Fn þ Bz Fn þ Gm Gm þ Bz Fn þ Bz Fn þ Gm Gm þ Bz Fn þ Bz Fn þ Gm Fn þ WW Gm þ WW Bz þ WW
xo
m
r
0.614 0.594 0.067 5.876 2.716 0.282 0.079 0.199 0.464 0.253 2.131 0.003
1.77 0.281 1.40 4.39 3.00 0.247 0.372 0.246 0.017 1.31 2.41 0.865
0.83 0.91 0.81 0.91 0.941 0.996 0.999 0.998 0.897 0.999 0.89 0.999
Fn ¼ fenofibric acid, Bz ¼ bezafibrate, Gm ¼ gemfibrozil, WW ¼ wastewater. The parameters of linear regression equations: x0 (value of y when x ¼ 0); m (slope) and r (correlation coefficient) with all p-values of 0.001. Analyses were computed using MINITAB Release 14 for Windows.
of the binary fibrate-wastewater combination and the quaternary combination; as can be observed, in a broad range of fa values, the binary combinations showed a strong synergism; however, at fa values above 0.8, the binary Fn þ WW and Bz þ WW combinations approached an additive effect and at fa values above 0.95, these two combinations yielded antagonism; by contrast, the Gm þ WW combination became even more synergistic. The quaternary combination interaction showed a strong synergism through a broad range of fa values but also turned into slight antagonism at fa values above 0.95,
Combination index, CI
2.0
1.5
1.0
0.5
0.0 0.0
0.2
0.4
0.6
0.8
1.0
Fractions affected, fa
Fig. 2 – Combination index plot ( fa-CI plot) for a set of three fibrates and toxic wastewater sample in their binary and quaternary combinations: Gn D WW (–6–), Fn D WW (–B–), Bz D WW (–,–) and Fn D Gm D Bz D WW (–;–) for the Anabaena CPB4337 test. CI values are plotted as a function of the fractional inhibition of bioluminescence ( fa) by computer simulation (CompuSyn) from fa [ 0.10 to 0.95. CI <1, [1 and >1 indicates synergism, additive effect and antagonism, respectively. At least three independent experiments with two replicates were used. The vertical bars indicate 95% confidence intervals for CI values based on sequential deletion analysis (SDA) (Chou and Martin, 2005). Fn [ fenofibric acid, Bz [ bezafibrate, Gm [ gemfibrozil, WW [ wastewater.
closely resembling the pattern of the Fn þ WW and Bz þ WW interactions which is confirmed by the highest r value (r ¼ 0.999) in the correlation analyses (Table 3), which suggests a predominant effect of Fn and Bz in the quaternary interaction. The computer software CompuSyn (Chou and Martin, 2005) displays a type of graphic termed polygonogram, which is a semiquantitative method of representing interactions between three or more compounds at a determined fa value. This graphic allows a simplified visual presentation of the overall results. Fig. 3 shows the polygonogram for the three fibrates and the wastewater at four fa values; synergism is indicated by solid lines and antagonism by broken ones; the thickness of the lines indicates the strength of the interaction. The polygonogram clearly shows the synergistic interaction of wastewater in combination with each of the three fibrates at low fa values and the antagonistic interaction that appeared at the highest fa value, 0.99, for the Fn þ WW and the Bz þ WW combinations. The same wastewater sample collected from a local STP was proved as responsible of stimulation of the bioluminescence activity of Vibrio fischeri to 110–120% of that of the control. Moreover, the EC50 values for the fibrates in the wastewater were higher than those for fibrates in pure water (Rosal et al., 2009). The same trend was observed comparing the dose-effect curve parameters (Dm, m and r) for the ternary combination (Fn þ Gm þ Bz) of fibrates in ddH2O and wastewater. The dose required to produce the median-effect (Dm) in Vibrio fischeri test when (Fn þ Gm þ Bz) were solved in wastewater was 131.936 compared to 55.6951 mg/L required when ddH2O was employed. CI values could not be calculated for Vibrio fischeri due to the fact that the wastewater itself was not toxic to this bacterium; synergism or antagonism could not be properly estimated (Chou, 2006).
4.
Discussion
The three fibrates that we have used in our study are lipid modifying agents that are effective in lowering elevated
434
water research 44 (2010) 427–438
Fig. 3 – Polygonograms showing the toxicological interactions of three fibrates and a toxic wastewater sample in their binary combinations (Fn D Bz, Bz D Gm, Fn D Gm, Gm D WW, Fn D WW, Bz D WW) as calculated by CompuSyn (Chou and Martin, 2005) for Anabaena CPB4337 test at four effect levels: fa [ 0.1 (a), fa [ 0.5 (b), fa [ 0.9 (c) and fa [ 0.99 (d). Solid lines indicate synergism, broken lines indicate antagonism. The thickness of the line represents the strength of synergism or antagonism. Figure generated by CompuSyn (Chou and Martin, 2005).
plasma triglycerides and cholesterol in humans (Staels et al., 1998). These pharmaceuticals are highly used, ubiquitous and persistent (Daughton and Ternes, 1999), they are found at ng/L to mg/L levels in many STP effluents, surface waters, estuaries of rivers and even in sea water (for a review, see Hernando et al., 2007). Although non-target organisms; the continuous release of these substances into the environment may cause acute or chronic toxicity to the aquatic biota. Regarding fibrates, in the recent literature there are many reports dealing with individual toxicity of different fibrates in a range of aquatic organisms from primary producers to consumers; a great variability has been found in the sensitivity of the different test organisms toward these pharmaceuticals (Hernando et al., 2007). However, pharmaceuticals such as fibrates do not occur singly in a polluted environment and are usually found as mixtures, therefore, for risk assessment strategies it is important to know the combined effects of pharmaceuticals in non-target organisms (Teuschler, 2007). There are two concepts widely used for the prediction of mixture toxicity: concentration addition (CA) and independent action (IA) (Backhaus et al., 2003; Vighi et al., 2003; Backhaus et al., 2004; Junghans et al., 2006). CA is used for mixtures whose components act in a similar mode of action while IA is based on the idea of dissimilar action, meaning that the compounds have different mechanisms of action; however, as discussed by Cleuvers (2003) the terms similar/
dissimilar action may be misleading. Pharmaceuticals such as fibrates may have the same pharmacological mechanism of action [i.e., interaction with the binding peroxisome proliferator-activated receptor a (PPARa)] in their target organism, humans; however, if fibrates released in the aquatic environments prove toxic to different non-target organisms, the exact mechanism of toxicity (probably different to the pharmacological mode of action) should be investigated in depth before choosing which approach, CA or IA, to use. In fact, only if toxicity is regarded as non-specific at all, the concept of CA may be used although it may also have limitations. Cleuvers (2003) found that two totally different pharmaceuticals, a fibrate and an anti-epileptic drug, followed the concept of CA in the Daphnia toxicity test and the concept of IA in an algal test; both pharmaceuticals apparently shared the same nonspecific toxic mode of action for both organisms; so it appeared that the concept of CA or IA did not depend on a similar/dissimilar mode of action but on the tested organism. The author also discussed that by definition, when using CA, substances applied below their individual noneffect concentration (NOEC) will contribute to the total effect of the mixture while when using IA, substances applied below their NOEC will not contribute to the total effect of the mixture, meaning that any combination effect will probably be higher if the substances follow the concept of CA and this may be misleading when considering the terms synergism or
water research 44 (2010) 427–438
antagonism because as also discussed by Chou (2006), synergism or antagonism may occur independently of a similar or dissimilar mode of action. In this context, Fent et al. (2006) tested mixtures of different kinds of pharmaceuticals (including fibrates) that might have estrogenic activity in a yeast reporter system; they applied the CA model and found that it had severe limitations when the dose-response curves of the individual pharmaceuticals were not identical or at low effect concentrations. As pharmaceuticals released in the environment may have such diverse dose-effect relationships, the lack of appropriate prediction suggests limitation of the CA mixtures concept. To study the nature of the combined fibrate interactions (synergism, additive effect, antagonism) for the Vibrio fischeri and Anabaena CPB4337 bioluminescence tests, we have followed the combination index (CI)-isobologram equation method of Chou (2006) and Chou and Talalay (1984); a method widely used to study drug interactions in pharmacology. This method may be considered a fractional analysis technique for drug interactions (Berenbaum, 1981; Bovill, 1998) that is independent of the mode of action and considers both the potency (EC50, Dm) and the shape (m) of the dose-effect curve for each drug. The method allows prediction of synergism/ antagonism at all effect levels ( fa) for a combination of n drugs; in contrast with the classical graphical isobologram method (Berenbaum, 1981; Bovill, 1998) that cannot be used for more than three compounds and have also graphical limitations to show all effect levels. By using this method, we have been able to determine the nature of interactions for a wide range of effect levels of three fibrates in binary and ternary combinations in two different bioluminescent organisms. However, the nature of these interactions was not uniform along the fa levels range in any of the two organisms. In Vibrio fischeri, antagonism predominated at low and intermediate fa levels but at the highest effect levels, interactions became additive or slightly synergistic. In Anabaena, a dual synergistic/antagonistic behaviour was observed with synergism predominating at fa levels below 0.4–0.5 and strong antagonism above these fa values. It is difficult to give an explanation to this phenomenon because the combination index method only allows quantitative determination of synergism or antagonism and the elucidation of the mechanism by which synergism or antagonism occurs is a separate issue that needs a different kind of approach. However, tentatively, antagonism, which could be considered the predominant interaction in Vibrio fischeri and Anabaena, might be explained by the structural similarity of fibrates which are related pharmaceuticals that share a common structural motif, a cyclic head and a hydrophobic tail (Rosal et al., 2009); at the fa levels where antagonism is found in both organisms, fibrates may compete with one another for the same target/ receptor sites. The slight synergism found at very high levels in Vibrio fischeri could perhaps be explained by the fact that at very high concentrations, fibrates may somehow combine to increase toxicity by an unspecific way of action that is probably not related to their pharmacological mechanism. Perhaps, the most puzzling interaction is the observed high synergism at very low fa levels in Anabaena; the mechanism of such synergistic interaction is not readily apparent. One could speculate that these fibrates at very low concentrations could
435
involve what Jia et al. (2009) in their extensive review of mechanisms of drug combinations call ‘‘facilitating actions’’ that means that secondary actions of one drug enhances the activity or level of another drug in the mixture or alternatively ‘‘complementary actions’’ when drugs act at the same target at different sites, at overlapping sites or at different targets of the same pathway. However, in the literature there are very few reports on possible targets of fibrates on the prokaryotic cell; English et al. (1994) reported that peroxisome proliferators such as fibrates have been shown to induce cytochrome P450BM-3 which catalyzes the hydroxylation of fatty acids, in Bacillus megaterium. Garbe (2004) reported that fibrates induced methyltransferase Rv0560c with a function in the biosynthesis of isoprenoid compounds in Mycobacterium tuberculosis; Garbe (2004) suggested that both effects may act on the plasma membrane, modulating its properties. In mitochondria, which have significant features that resemble those of prokaryotes, fibrates have been found to inhibit respiratory complex I (NDH-1 complex) and to interfere with mitochondrial fatty acid oxidation (Scatena et al., 2007). Whether fibrates may exert similar effects in Vibrio fischeri and Anabaena to those observed in Bacillus or mitochondria needs further research. In this context, we have found that, as the faCI plots show, fibrate interactions do not follow the same pattern in both bacteria, this may be due to the different origin and position in the food web of Vibrio fischeri, a heterotrophic marine prokaryote and Anabaena CPB4337, a recombinant strain of an obligate phototrophic freshwater prokaryote; in fact, Anabaena presents intracellular photosynthetic membranes called thylakoids where several functionally distinct NDH-1 complexes have been found with roles both in respiration and photosynthesis (Battchikova and Aro, 2007). If fibrates are also affecting NDH-1 complexes in Anabaena, their effects might be very different to those in Vibrio fischeri; so, although we have measured the same toxicity endpoint in both bacteria, i.e., luminescence inhibition, the combined effects of fibrates seem to depend on the test organism. Ince et al. (1999) assessed toxic interactions of heavy metals in binary mixtures on Vibrio fischeri and the freshwater aquatic plant Lemna minor and found that most binary metal mixtures exhibited only antagonistic interactions in the plant opposed to fewer antagonistic and some synergistic interactions in the heterotrophic bacterium. These authors also found that in the bacterium, the nature of the interaction (synergism or antagonism) also changed with the effect level of the binary metal combinations, although the authors did not provide a mechanistic explanation for this variability. Cheng and Lu (2002) made a comparison of joint interactions of organic toxicants in binary mixtures in Escherichia coli and Vibrio fischeri and found that toxicants with the same mechanisms of toxicity displayed mostly additive or antagonistic interactions in E. coli and Vibrio fischeri; however a synergistic interaction was found between glutardialdehyde and butyraldehyde in Vibrio. Synergistic effects in both bacteria were mostly associated with toxicants with different mechanisms of toxicity, although antagonism clearly predominated. They also found that for a total of 44 organic binary mixtures, only six mixtures resulted in identical type of interaction in both bacteria. From our results and those of other authors’ (Ince et al., 1999; Cheng and Lu, 2002; Cleuvers, 2003) one may
436
water research 44 (2010) 427–438
conclude that previous knowledge of the mechanism of toxic action of a compound is not useful enough to predict which kind of interactions it will display when combined with other toxicants with the same or different toxic mechanism; also, as we have shown, the nature of the interaction may depend on the effect level of the mixture. In addition, different types of organisms will show completely different responses to mixtures of potential toxicants. We previously found that a local wastewater was very toxic for the Anabaena CPB4337 test but non-toxic at all for the Vibrio fischeri or Daphnia magna tests. This wastewater is a mixture of over thirty micropollutants, mostly pharmaceuticals of different therapeutics groups that, besides the fibrates used in this study, included antibiotics, analgesics/anti-inflammatories, b-blockers, antidepressants, anti-epileptics/psychiatrics, ulcer healing compounds, diuretics and bronchodilators; personal care products and some priority organic pollutants are also present (Rosal et al., 2008). The method of Chou allows to combine one drug mixture with another drug mixture and determine their interactions; therefore, we studied the nature of the interaction of fibrates and wastewater in the Anabaena bioluminescence test; interestingly, we found that in a wide range of effect levels, the interaction of wastewater and the three fibrate combination was synergistic; particularly, at very low fa values which means that fibrates are at low concentrations and the wastewater is diluted several-fold, the method predicted a strong synergism; this may be due, as discussed above, to the observed synergistic interactions of fibrates with one another as well as interactions with some of the detected micropollutants when present at very low concentrations. This observed synergism may be environmentally relevant since most pharmaceuticals such as fibrates do not usually show acute toxicity on non-target organisms when tested at real environmental concentrations (Hernando et al., 2007) but in a mixture, if they act synergistically, they could prove toxic for a test organism even at low concentrations; these results agree with those found by Hernando et al. (2004) who reported synergistic toxic effects for Daphnia magna test when wastewater was spiked with environmental concentrations of several pharmaceuticals including fibrates. By contrast, our results show that at high fa values ( fa > 0.8), the combined interaction of the quaternary fibrates þ wastewater combination, the binary Fn þ WW and Bz þ WW combinations approached an additive effect and eventually became antagonistic; in our previous study, the wastewater itself decreased Anabaena bioluminescence by 84% with a lower confidence limit of 76% and an upper confidence limit of 91%; when the wastewater was spiked with increasing concentrations of each fibrate we found that, with the exception of gemfibrozil, the EC50 values for the fibrates in the wastewater were higher than those for fibrates in pure water; this was attributed either to reduced bioavailability or to antagonistic effects of fibrates with other chemicals present in the wastewater; although we did not use the method of Chou, we obtained similar results to the ones we report in this study; that is, at high effect levels (>84% luminescence inhibition) the interaction of fibrates with wastewater, except the Gm þ WW combination, showed antagonism. Based on our results, we propose that the combination index (CI)-isobologram equation, a method widely used in
pharmacology both for in vitro and in vivo bioassays, may also be applied in environmental toxicology as a general method to define interactions of potential toxicants in mixtures in any test organism and/or toxicological endpoint of interest and could be especially useful for risk assessment strategies that take into account the toxicological interactions of substances in a mixture.
5.
Conclusions
We report an environmental application of the combination index (CI)-isobologram equation to study the nature of the interactions of fibrate combinations in two bioluminescent aquatic organisms. The method allowed calculating synergism or antagonism of binary and ternary fibrate combinations at all effect levels simultaneously; we could also test the method with a real wastewater sample in binary and quaternary combination with the fibrates, finding that at very low effect levels, the fibrates acted synergistically with the wastewater in the Anabaena test. The proposed method may be used with other test organisms and/or toxicological endpoints and could be particularly useful for risk assessment approaches to toxicity of complex mixtures.
Acknowledgements The research was funded by the Spanish Ministry of Education through grants CTM2005-03080/TECNO and CSD2006-00044 and the Comunidad de Madrid, grants 0505/AMB-0395 and 0505/MB/0321.
references
Allen, M.B., Arnon, D.I., 1955. Studies on nitrogen-fixing blue green algae. I Growth and nitrogen fixation by Anabaena cylindrica Lemm. Plant Physiol 30 (4), 366–372. Andreozzi, R., Raffaele, M., Nicklas, P., 2003. Pharmaceuticals in STP effluents and their solar photodegradation in aquatic environment. Chemosphere 50 (10), 1319–1330. ˚ ., Blanck, H., Faust, M., Backhaus, T., Altenburger, R., Arrhenius, A Finizio, A., Gramatica, P., Grote, M., Junghans, M., Meyer, W., Pavan, M., Porsbring, T., Scholze, M., Todeschini, R., Vighi, M., Walter, H., Horst Grimme, L., 2003. The BEAM-project: prediction and assessment of mixture toxicities in the aquatic environment. Continental Shelf Res. 23 (17–19), 1757–1769. Backhaus, T., Arrhenius, A., Blanck, H., 2004. Toxicity of a mixture of dissimilarly acting substances to natural algal communities: predictive power and limitations of independent action and concentration addition. Environ. Sci. Technol 38 (23), 6363–6370. Battchikova, N., Aro, E.-M., 2007. Cyanobacterial NDH-1 complexes: multiplicity in function and subunit composition. Physiol. Plant 131 (1), 22–32. Berenbaum, M.C., 1981. Criteria for analyzing interactions between biologically active agents. Adv. Cancer Res. 35, 269–335. Borgmann, U., Bennie, D.T., Ball, A.L., Palabrica, V., 2007. Effect of a mixture of seven pharmaceuticals on Hyalella azteca over multiple generations. Chemosphere 66 (7), 1278–1283.
water research 44 (2010) 427–438
Bovill, J.G., 1998. Analysis of drug interactions. Baillie`re’s Clin. Anaesthesiol 12 (2), 135–168. Brain, R.A., Johnson, D.J., Richards, S.M., Hanson, M.L., Sanderson, H., Lam, M.W., Young, C., Mabury, S.A., Sibley, P.K., Solomon, K.R., 2004. Microcosm evaluation of the effects of an eight pharmaceutical mixture to the aquatic macrophytes Lemna gibba and Myriophyllum sibiricum. Aquat. Toxicol 70 (1), 23–40. Castiglioni, S., Bagnati, R., Fanelli, R., Pomati, F., Calamari, D., Zuccato, E., 2006. Removal of pharmaceuticals in sewage treatment plants in Italy. Environ. Sci. Technol 40 (1), 357–363. Chen, C.-Y., Lu, C.-L., 2002. An analysis of the combined effects of organic toxicants. Sci. Total Environ 289 (1–3), 123–132. Chou, T.C., 1976. Derivation and properties of Michaelis–Menten type and Hill type equations for reference ligands. J. Theor. Biol. 59 (2), 253–276. Chou, T.C., 2006. Theoretical basis, experimental design, and computerized simulation of synergism and antagonism in drug combination studies. Pharmacol. Rev. 58 (3), 621–681. Chou, T.C., Martin, N., 2005. CompuSyn for Drug Combinations: PC Software and User’s Guide: A Computer Program for Quantification of Synergism and Antagonism in Drug Combinations and the Determination of IC50 and ED50 and LD50 Values. ComboSyn, Inc., Paramus, NJ. Chou, T.C., Talalay, P., 1984. Quantitative analysis of dose-effect relationships: the combined effects of multiple drugs or enzyme inhibitors. Adv. Enzyme Regul 22, 27–55. Christensen, A.M., Faaborg-Andersen, S., Ingerslev, F., Baun, A., 2007. Mixture and single-substance toxicity of selective serotonin reuptake inhibitors toward algae and crustaceans. Environ. Toxicol. Chem. 26 (1), 85–91. Cleuvers, M., 2003. Aquatic ecotoxicity of pharmaceuticals including the assessment of combination effects. Toxicol. Lett. 142 (3), 185–194. Cleuvers, M., 2004. Mixture toxicity of the anti-inflammatory drugs diclofenac, ibuprofen, naproxen, and acetylsalicylic acid. Ecotoxicol. Environ. Saf 59 (3), 309–315. Comerton, A.M., Andrews, R.C., Bagley, D.M., Yang, P., 2007. Membrane adsorption of endocrine disrupting compounds and pharmaceutically active compounds. J. Memb. Sci. 303 (1–2), 267–277. Crane, M., Watts, C., Boucard, T., 2006. Chronic aquatic environmental risks from exposure to human pharmaceuticals. Sci. Total Environ 367 (1), 23–41. Daughton, C.G., Ternes, T.A., 1999. Pharmaceuticals and personal care products in the environment: agents of subtle change? Environ. Health Perspect 107 (6), 907–938. English, N., Hughes, V., Wolf, C.R., 1994. Common pathways of cytochrome P450 gene regulation by peroxisome proliferators and barbiturates in Bacillus megaterium ATCC14581. J. Biol. Chem. 269 (43), 26836–26841. Fent, K., Escher, C., Caminada, D., 2006. Estrogenic activity of pharmaceuticals and pharmaceutical mixtures in a yeast reporter gene system. Reprod. Toxicol 22 (2), 175–185. Garbe, T.R., 2004. Co-induction of methyltransferase Rv0560c by naphthoquinones and fibric acids suggests attenuation of isoprenoid quinone action in Mycobacterium tuberculosis. Can. J. Microbiol. 50 (10), 771–778. Gros, M., Petrovic, M., Barcelo, D., 2007. Wastewater treatment plants as a pathway for aquatic contamination by pharmaceuticals in the Ebro river basin (northeast Spain). Environ. Toxicol. Chem. 26 (8), 1553–1562. Han, G.H., Hur, H.G., Kim, S.D., 2006. Ecotoxicological risk of pharmaceuticals from wastewater treatment plants in Korea: occurrence and toxicity to Daphnia magna. Environ. Toxicol. Chem. 25 (1), 265–271. Hernando, M.D., Petrovic, M., Ferna´ndez-Alba, A.R., Barcelo´, D., 2004. Analysis by liquid chromatography-electrospray
437
ionization tandem mass spectrometry and acute toxicity evaluation for beta-blockers and lipid-regulating agents in wastewater samples. J. Chromatogr. A 1046 (1–2), 133–140. Hernando, M.D., Agu¨era, A., Ferna´ndez-Alba, A.R., 2007. LC-MS analysis and environmental risk of lipid regulators. Anal. Bioanal. Chem. 387 (4), 1269–1285. Holoshitz, N., Alsheikh-Ali, A.A., Karas, R.H., 2008. Relative safety of gemfibrozil and fenofibrate in the absence of concomitant cerivastatin use. Am. J. Cardiol 101 (1), 95–97. ¨ stu¨n, B., 1999. Ince, N.H., Dirilgen, N., Apikyan, I.G., Tezcanli, G., U Assessment of toxic interactions of heavy metals in binary mixtures: a statistical approach. Arch. Environ. Contam. Toxicol 36, 365–372. International Organization for Standardization, 2007. Water quality - Determination of the inhibitory effect of water samples on the light emission of Vibrio fischeri (Luminescent bacteria test), ISO 11348-3 Revised version, Geneva, Switzerland. Jia, J., Zhu, F., Ma, X., Cao, Z.W., Li, Y.X., Chen, Y.Z., 2009. Mechanisms of drug combinations: interaction and network perspectives. Nat. Rev. Drug Discov 8 (2), 111–128. Junghans, M., Backhaus, T., Faust, M., Scholze, M., Grimme, L.H., 2006. Application and validation of approaches for the predictive hazard assessment of realistic pesticide mixtures. Aquat. Toxicol 76 (2), 93–110. Kasprzyk-Hordern, B., Dinsdale, R.M., Guwy, A.J., 2009. The removal of pharmaceuticals, personal care products, endocrine disruptors and illicit drugs during wastewater treatment and its impact on the quality of receiving waters. Water Res. 43 (2), 363–380. Lambropoulou, D.A., Hernando, M.D., Konstantinou, I.K., Thurman, E.M., Ferrer, I., Albanis, T.A., Fernandez-Alba, A.R., 2008. Identification of photocatalytic degradation products of bezafibrate in TiO2 aqueous suspensions by liquid and gas chromatography. J. Chromatogr. A 1183 (1–2), 38–48. Metcalfe, C.D., Miao, X.S., Koenig, B.G., Struger, J., 2003. Distribution of acidic and neutral drugs in surface waters near sewage treatment plants in the lower Great Lakes, Canada. Environ. Toxicol. Chem. 22 (12), 2881–2889. Miller, D.B., Spence, J.D., 1998. Clinical pharmacokinetics of fibric acid derivatives (fibrates). Clin. Pharmacokinet 34 (2), 155–162. Pomati, F., Orlandi, C., Clerici, M., Luciani, F., Zuccato, E., 2008. Effects and interactions in an environmentally relevant mixture of pharmaceuticals. Toxicol. Sci. 102 (1), 129–137. Quinn, B., Gagne, F., Blaise, C., 2009. Evaluation of the acute, chronic and teratogenic effects of a mixture of eleven pharmaceuticals on the cnidarian, Hydra attenuata. Sci. Total Environ 407 (3), 1072–1079. Rodea-Palomares, I., Gonza´lez-Garcı´a, C., Legane´s, F., Ferna´ndezPin˜as, F., 2009. Effect of pH, EDTA, and anions on heavy metal toxicity toward a bioluminescent cyanobacterial bioreporter. Arch. Environ. Contam. Toxicol. doi:10.1007/ s00244-008-9280-9. Rodrı´guez, A., Rosal, R., Perdigo´n, J.A., Mezcua, M., Agu¨era, A., Hernando, M.D., Leto´n, P., Ferna´ndez-Alba, A.R., Garcı´aCalvo, E., 2008. In: Barcelo´, D., Petrovic, M. (Eds.), Emerging Contaminants from Industrial and Municipal Waste. SpringerVerlag, Berlin, pp. 127–175. 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 Res. 42 (14), 3719–3728. Rosal, R., Rodea-Palomares, I., Boltes, K., Ferna´ndez-Pin˜as, F., Legane´s, F., Gonzalo, S., Petre, A., 2009. Ecotoxicity assessment of lipid regulators in water and biologically treated wastewater using three aquatic organism. Environ. Sci. Pollut. Res. doi:10.1007/s11356-009-0137-1.
438
water research 44 (2010) 427–438
Scatena, R., Bottoni, P., Botta, G., Martorana, G.E., Giardina, B., 2007. The role of mitochondria in pharmacotoxicology: a reevaluation of an old, newly emerging topic. Am. J. Physiol. Cell Physiol 293 (1), C12–C21. Staels, B., Dallongeville, J., Auwerx, J., Schoonjans, K., Leitersdorf, E., Fruchart, J.C., 1998. Mechanism of action of fibrates on lipid and lipoprotein metabolism. Circulation 98 (19), 2088–2093. Stumpf, M., Ternes, T.A., Wilken, R.D., Rodrigues, S.V., Baumann, W., 1999. Polar drug residues in sewage and natural waters in the state of Rio de Janeiro. Brazil. Sci. Total Environ 225 (1–2), 135–141.
Teuschler, L.K., 2007. Deciding which chemical mixtures risk assessment methods work best for what mixtures. Toxicol. Appl. Pharmacol 223 (2), 139–147. Vighi, M., Altenburger, R., Arrhenius, A., Backhaus, T., Bodeker, W., Blanck, H., Consolaro, F., Faust, M., Finizio, A., Froehner, K., Gramatica, P., Grimme, L.H., Gronvall, F., Hamer, V., Scholze, M., Walter, H., 2003. Water quality objectives for mixtures of toxic chemicals: problems and perspectives. Ecotoxicol. Environ. Saf 54 (2), 139–150.
water research 44 (2010) 439–448
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Toxication or detoxication? In vivo toxicity assessment of ozonation as advanced wastewater treatment with the rainbow trout Daniel Staltera,*, Axel Magdeburga, Mirco Weilb, Thomas Knackerb, Jo¨rg Oehlmanna a
Goethe University Frankfurt am Main, Biological Sciences Division, Department of Aquatic Ecotoxicology, Siesmayerstrasse 70, 60054 Frankfurt, Hessen, Germany b ECT Oekotoxikologie GmbH, Bo¨ttgerstrasse 2–4, 65439 Flo¨rsheim, Germany
article info
abstract
Article history:
Ozonation as advanced wastewater treatment method is an effective technique for
Received 23 March 2009
micropollutant removal. However, the application of this method carries the inherent
Received in revised form
danger to produce toxic oxidation byproducts. For an ecotoxicological assessment
16 July 2009
conventionally treated wastewater, wastewater after ozonation and ozonated wastewater
Accepted 18 July 2009
after sand filtration were evaluated in parallel at an operating treatment plant via the fish
Available online 25 July 2009
early life stage toxicity test (FELST) using rainbow trout (Oncorhynchus mykiss).
Keywords:
exposed to ozonated WW. This was accompanied by a significant decrease in body weight
Sewage treatment
and length compared to reference water, to the conventionally treated WW and to the
Pharmaceuticals
ozonated water after sand filtration. Hence sand filtration obviously prevents from adverse
Oxidation byproducts
ecotoxicological effects of ozonation.
The FELST revealed a considerable developmental retardation of test organisms
Vitellogenin
An additional test with yolk-sac larvae resulted in a significant reduction of vitellogenin
Emerging contaminants
levels in fish exposed to ozonated wastewater compared to fish reared in conventionally
Advanced oxidation process
treated wastewater. This demonstrates the effective removal of estrogenic activity by
Rainbow trout
ozonation.
Fish early life stage toxicity test
Adverse ozonation effects may have been a result of the conversion of chemicals into more toxic metabolites. However, sand filtration reduced toxication effects indicating that these oxidation byproducts are readily degradable or adsorbable. The results indicate that in any case ozonation should not be applied without subsequent post treatment appropriate for oxidation byproducts removal (e.g. sand filtration). ª 2009 Elsevier Ltd. All rights reserved.
1.
Introduction
Wastewater (WW) is one of the major sources of micropollutants introduced into the aquatic environment (Schwarzenbach et al., 2006). The large spectrum of pharmaceuticals and personal care products (PPCPs) occurring in
surface waters (Daughton and Ternes, 1999), particularly with regard to mixture toxicity, may pose a potential threat to aquatic wildlife. But also single substances may have the capability to endanger the ecosystem as for example diclofenac can lead to an impairment of the general health condition of rainbow trout at environmentally relevant concentrations
* Corresponding author. Tel.: þ49 69 7982 4882; fax: þ49 69 7982 4748. E-mail address:
[email protected] (D. Stalter). 0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2009.07.025
440
water research 44 (2010) 439–448
Nomenclature AOC C COD DOC EE2 ELISA EQS FELST FS GJIC MS222 NH4–N
assimilable organic carbon control chemical oxygen demand dissolved organic carbon 17a-ethinylestradiol enzyme linked immunosorbent assay environmental quality standard fish early life stage toxicity test wastewater after final sedimentation gap junctional intercellular communication tricaine methanesulfonate ammonium nitrogen
(Schwaiger et al., 2004; Triebskorn et al., 2004). 17a-ethinylestradiol (EE2) is considered to be the most potent estrogenic active compound for fish with the lowest observed effect concentration of 0.1 ng/L for vitellogenesis in rainbow trout (Purdom et al., 1994). Moreover it leads to a reduced egg fertilisation success and skewed sex ratio toward females of fat head minnows at a LOEC of 0.32 ng/L (Parrott and Blunt, 2005). Furthermore, industrial surfactants like alkylphenolic ethoxylates are likely to be responsible for feminization effects in rainbow trout (Jobling et al., 1996) and complex mixtures of estrogenic chemicals present in the environment have been shown to act additively (Brian et al., 2005). The concentrations of such compounds in WW often exceed effect concentrations (Ternes et al., 1999; Ying et al., 2002; Tixier et al., 2003) and thus even treated WW contains high amounts of estrogenic substances that impair the endocrine system of exposed fish (Larsson et al., 1999; Rodgers-Gray et al., 2001) resulting in feminization effects of males and consequently reduced fertility of field populations (Jobling et al., 2002). Besides, in the European Union the reduction of the contamination of surface waters with hazardous substances is defined by the Water Framework Directive requiring a ‘‘good status’’ for all coastal and inland waters until 2015 (European Commission, 2000). One tool is the implementation of environmental quality standards (EQSs) for mono substance pollutants exhibiting a significant risk to the aquatic ecosystem. The discharge of such compounds, classified as priority substances, is envisaged to be progressively reduced by 2025 or 5 years after inclusion in the list for priority substances, respectively. However, till now so called ‘‘emerging contaminants’’ are only marginally addressed and PPCPs are not included in this list. But the latter has to be reviewed at least every 4 years and provisions have been made to add several hazardous emerging contaminants. Amongst others the inclusion of diclofenac, EE2 and carbamazepine is discussed and additionally recommended by the European Parliament since 2007 (Jensen, 2007). Moreover EQSs meanwhile have been proposed for several PPCPs based on reliable ecotoxicity data (e.g. diclofenac: 0.1 mg/L, EE2: 0.03 ng/L, carbamazepine: 0.5 mg/L; Jahnel et al., 2006; Moltmann et al., 2007). Unfortunately, some PPCPs exceed the EQSs in surface waters of some urbanised
NO3–N O OECD OS P PAH PAC PPCPs rcf SPM VTG WW WWTP
nitrate nitrogen wastewater after ozonation Organisation for Economic Cooperation and Development wastewater after ozonation and sand filtration phosphate polycyclic aromatic hydrocarbons powdered activated carbon pharmaceuticals and personal care products relative centrifugal force suspended particulate matter vitellogenin wastewater wastewater treatment plant
regions. That is, inter alia, a result of the fact that conventional WW treatment plants are not designed for appropriate removal of PPCPs (Daughton and Ternes, 1999). Nevertheless in many countries surface waters serve as raw water resources for drinking water supply. Due to the public’s demand for safe drinking water and ecosystem health advanced WW treatment methods are required to allow for sufficient removal of micropollutants from sewage treatment effluents. Therefore, end of pipe techniques like activated carbon filtration and ozonation of WW might be crucial to achieve regulative requirements in a medium-term perspective. Huber et al. (2005) demonstrated that ozonation of secondary effluent is an effective tool for the removal of a wide range of pharmaceuticals among them diclofenac, carbamazepine and the estrogens estradiol, estrone and EE2, with degradation rates of more than 90% for ozone doses of 2 mg/L. Merely iodinated X-ray contrast media and a few acidic pharmaceuticals were oxidized only partially. Nakada et al. (2007) confirmed the effective elimination of 22 pharmaceuticals during ozonation in an operating treatment plant. A further advantage of the ozonation is the sanitizing property of the method (Tyrrell et al., 1995). The effective removal of micropollutants and pathogens indicates that ozonation is a suitable technique for advanced WW treatment to reduce the contamination of the aquatic environment. Actually a reduced toxicity of a mixture of six pharmaceuticals treated by ozonation towards algae and rotifer was demonstrated by Andreozzi et al. (2004). Reduced estrogenic activity of an EE2-containing WW after ozonation was shown by Huber et al. (2004) and ozone treatment of WW significantly decreased the toxic potency of urban pollutants to the freshwater mussel Elliptio complanata (Gagne et al., 2007). However, ozone treatment typically transforms chemical compounds but does not mineralize them entirely. Consequently, not only the concentrations of mother compounds but also the inclusion of transformation products and their mixtures should be assessed for toxicity, too. For a risk-benefit analysis an extensive ecotoxicological evaluation of ozonated WW is essential. In vitro bioassays covering different modes of toxic action (e.g. estrogenicity, acetylcholine esterase activity or non-specific baseline toxicity) and performed with enriched water samples are
water research 44 (2010) 439–448
2.2.
Table 1 – Wastewater quality summary (mg/L). SPM
COD
P
NH4–N
NO3–N
pH
0.1 0.04 0.27
5.4 10.1
8.3 8.12 8.4
Median 10% percentile 90% percentile
Final sedimentation 4.8 17 0.19 3.6 15 0.17 8.96 23.8 0.22
n
37
Median 10% percentile 90% percentile
Ozonation and 2 15 1.6 12.2 2.4 19.8
sand filtration 0.17 0.04 0.14 0.04 0.19 0.15
9.8 7.28 12.48
n
37
37
37
23
23
37
37
37
2
23
8.3 8.02 8.4 23
suitable tools for toxicity characterization of WW as shown by Escher et al. (2008). These methods were shown to be highly sensitive even for identification and characterization of low toxic water samples. Nevertheless, as typical screening tools they are not designed to replace chronic in vivo tests of whole effluent samples. To allow for a more comprehensive and integrative assessment of the potentially hazardous impact of WW ozonation we applied the fish early life stage toxicity test (FELST) with rainbow trout (Oncorhynchus mykiss).
2.
Materials and methods
Ozonated and conventionally treated WW were tested in parallel on site at the wastewater treatment plant (WWTP) Wu¨eri (Regensdorf, Switzerland) with the fish early life stage toxicity test (FELST) in a flow-through system.
2.1.
Characterization of the wastewater treatment plant
The municipal WWTP Wu¨eri operates experimentally with a full scale ozonation reactor after final sedimentation and with a sand filtration step after the ozone reactor. Table 1 shows the WW quality parameters. The water parameters after the final sedimentation and after the ozone reactor are on the same level and therefore exemplarily shown for final sedimentation water. Low ammonium and phosphate concentrations indicate that the treatment plant is already working well. The dissolved organic carbon (DOC) ranged from 5.4 to 5.9 mg/L and the pH was nearly constant in all test waters. The WWTP serves for a population equivalent of 25,000. The median discharge in the experimental period was 6190 m3/day (10th percentile: 4430 m3/day, 90th percentile: 10,500, n ¼ 109) and the applied ozone concentration was in a range between 0.4 and 1 mg O3/mg DOC.
441
Experimental setup
WW from three different sampling points of serial treatment steps was tested (Fig. 1): after the final sedimentation (FS), after the ozone reactor (O) and after additional sand filtration (OS). Test waters were passively transported through high-grade steel pipes to aerated high-grade steel reservoir tanks. The retention time in the pipes was adjusted to be at least 45 min to avoid ozone residuals reaching the exposure vessels. Indeed no ozone was detected by indigo blue method (Bader and Hoigne, 1981) in the reservoir tank during maximum required flow-through and during maximum applied ozone concentration (1 mg O3/mg DOC). From reservoir tanks test waters were transported through polytetrafluoroethylene tubes via a peristaltic pump (IPC24, Ismatec, Wertheim– Mondfeld, Germany) to the exposure vessels each equipped with a passive discharge device and tempered using a temperature-controlled water bath. Constant temperature conditions in the water bath were achieved with a flowthrough cooling system (Van der Heijden, Do¨rentrup, Germany). The flow-through rates in the exposure vessels ranged from 11 mL per minute up to 44 mL per minute (2–8 fold water exchange per day in the exposure vessels) depending on the fish size to match the loading rate criteria (OECD, 1992b). Reconstituted water according to OECD guideline 203 (1992a) was used as control water (C). The FELST was performed with the rainbow trout (Oncorhynchus mykiss) according to OECD guideline 210 (1992b) with a constant water temperature of 10 2 C as well as darkness for embryo development and 12 2 C with 12/12 h light/dark photoperiod post hatch. Sixty newly fertilized eggs per replicate were exposed to the test waters for 65 days in high-grade steel 10 L tanks. One test series was performed with unfiltered WW and a second with membrane filtered WW (pore size: 0.4 mm, Kubota Corp., Osaka, Japan) to minimize microbial impacts. Macromolecules and organic compounds were not retained. However, it has to be considered that membrane filtration additionally removes suspended particulate matter and consequently all particle bound pollutants. The filter membranes were placed in the reservoir vessels. A third test was performed with yolk-sac fry (5 days post hatch, 30 larvae per exposure vessel) and non-filtered test waters because we postulated a reduced sensitivity to pathogen contamination of the larvae compared to the egg stage. The test duration was 64 days. All tests were performed with undiluted WW to increase the probability to detect differences between the treatment groups. With the beginning of swim up (the swim up process marks a developmental transition from larval stage to juvenile fish stage and is characterized with the
Fig. 1 – Sampling points at the wastewater treatment plant. Abbreviations: FS, after final sedimentation; O, after the ozone reactor; OS, after sand filtration.
442
water research 44 (2010) 439–448
beginning of exogenous ingestion) the fish were fed four times per day (trout starter, 4% body weight per day). The tests were run with four (C) and three (FS, O, OS) replicates in the first FELST with unfiltered WW and two (C, OS) and three replicates (FS, O) per test water in the remaining two tests. The latter were performed in parallel and therefore it was not possible to use consistently three replicates per WW-group due to limited space capacities. In each test replicates were placed randomized in the water bath. Observations on egg coagulation, hatching, mortality, swim up, malformation and abnormal behaviour were recorded daily. Fish were humanely killed by MS222 (tricaine methanesulfonate, Sigma– Aldrich, St. Louis, USA) overdose. Individual fish were blotted dry, weight and body length were determined. Afterwards fish were frozen in liquid nitrogen and stored at 80 C until vitellogenin detection in whole body homogenates.
2.3.
Vitellogenin detection
Whole body homogenates of 11 fish (from the third test with yolk-sac fry) per replicate were prepared as described by Holbech et al. (2006) with slight modifications. Aliquots of 0.3 g frozen fish, excised between head and pectoral fin, were mixed with 10 fold of the body weight of homogenisation buffer (50 mM Tris–HCl pH 7.4; 1% protease inhibitor cocktail (P 8340, Sigma–Aldrich, St. Louis, USA)) and homogenated with a dispersing apparatus (T18 basic Ultra–Turrax, IKA, Staufen, Germany). The homogenate was centrifuged for 30 min at 20,000 rcf and the supernatant was used for vitellogenin analysis. Vitellogenin (VTG) was detected with a rainbow trout vitellogenin ELISA test kit (Biosense, Bergen, Norway) using a 1:20 dilution.
2.4.
Statistical analysis
Complete statistical analyses (Kruskal–Wallis with Dunn’s post test, Fisher’s exact test, non-linear regression with variable slope) were performed using GraphPad Prism version 5.0 for windows (GraphPad software, San Diego, CA, USA). All given error values indicate the standard error (SE) and in all figures error bars display the SE. Kruskal–Wallis with Dunn’s post test was chosen to test on significant differences because data were not normally distributed in all cases. For quantal data Fisher’s exact test was applied (mortality, swim up, coagulation). When toxicity endpoints could be analysed individually for each test animal (biomass, body length, vitellogenin), the collected data are presented and statistically evaluated on a per specimen basis.
3.
Results
3.1.
FELST with unfiltered wastewater
Unfiltered WW caused an increased coagulation rate of the exposed eggs in the FELST (Fig. 2). The eggs exposed to WW after final sedimentation (FS) were completely coagulated after 18 days with a 50% coagulation time of 12.1 days. The coagulation of the eggs exposed to WW after the ozone reactor (O) was considerably delayed compared to FS. However the
Fig. 2 – Cumulative coagulation of Oncorhynchus mykiss eggs (mean values ± SE) exposed to differently treated wastewaters. Abbreviations: C, control water; FS, final sedimentation; O, ozonation; OS, ozonation and sand filtration; SE, standard error. Time scale is logarithmized. After 40 days all treatment groups differ significantly from each other (Fisher’s exact test: p < 0.001; n [ 240 (C), 180 (FS, O, OS)).
coagulation after 40 days achieves still an average rate of 87.2 4.8% with a 50% coagulation time of 17.6 days. The lowest coagulation rate in the WW treatments occurred after sand filtration (OS; 64.4 4.0% after 40 days; 50% coagulation time: 25.8 days). After 10–15 days exposure, fungus mycelia (first appearing in the FS vessels) were observed in all WW exposure vessels. Mycelia were found on and between eggs as well as vorticellas on the eggs whereas the reference water (C) remained observably free from mycelia and vorticellas. Hatching success in the control group (53.4 5.2%) did not meet validity criteria (>66%; OECD guideline 210).
3.2.
FELST with membrane filtered wastewater
To exclude microbial impairment the second FELST was performed with membrane filtered WW and eggs were obtained from another fish hatchery. All test vessels remained free from microbial contamination throughout the test duration. The egg coagulation rates in the WW treatment groups of this experimental series were reduced compared to the first FELST with a maximum of 25% (O, OS) and only 20.1 1.1% after FS (Fig. 3A). Egg coagulation was significantly increased and hatching success significantly decreased in the WW treatment groups compared to the control ( p < 0.05, Fisher’s exact). The coagulation rates were slightly but not significantly increased in O and OS compared to FS (Fig. 3A). The hatching progress was slightly delayed in O compared to FS and OS but hatching success achieved at least 75% in all treatments (Fig. 3B). The control group met the validity criteria according to OECD guideline 210 (egg coagulation: 10.0 1.7%, hatching success: 90.0 1.7%).
water research 44 (2010) 439–448
443
Fig. 3 – Cumulative coagulation of eggs (A) and cumulative hatching of larvae (B) from Oncorhynchus mykiss (mean values ± SE) exposed to differently treated and membrane filtered wastewaters. Abbreviations: C, control water; FS, final sedimentation; O, ozonation; OS, ozonation and sand filtration; SE, standard error. Time scale is logarithmized. After 40 and 36 days, respectively all wastewater treatment groups are significantly different from the control (Fisher’s exact test: p < 0.01–0.05; n [ 120 (C, OS), 180 (FS, O)).
Fig. 4 shows the cumulative swim up of hatched fish beginning after 45 days of exposure. The swim up is considerably delayed in all WW treatment groups compared to the control. This effect is most notable in O, even if compared to
FS and OS. At the end of the experiment only 83.3 1.6% of the fish swam up in O while 100% swam up in C and FS and 97.6 2.4% in OS. Hereby the swim up success after 64 days is significantly decreased in O compared to the other treatment groups ( p < 0.01, Fisher’s exact). The 50% swim up time in the control is 47.4 1.0 days, which is only slightly increased in FS and OS (50.2 1.0; 49.3 1.0) but obviously increased in O (57.5 1.04). The biomass as well as the body length of fish is significantly decreased ( p < 0.001, Kruskal–Wallis with Dunn’s post test) in all WW treatments compared to the control (Fig. 5). Both endpoints in fish exposed to O are furthermore significantly decreased compared to the FS group ( p < 0.001) and significantly decreased compared to OS ( p < 0.05). Generally the mortality is comparatively low in all WW treatment groups (Fig. 6), as they are fulfilling the validity criteria for controls according to OECD guideline 210 (survival after hatch 70%). Nevertheless the mortality in the WW treatment groups is slightly increased compared to C. The highest and significantly increased mortality rate was detected in the ozonated water (24.0 4.0%, p < 0.05, Fisher’s exact).
3.3. Fig. 4 – Cumulative swim up of Oncorhynchus mykiss larvae (mean values ± SE) exposed to differently treated and membrane filtered wastewaters. Abbreviations: C, control water; FS, final sedimentation; O, ozonation; OS, ozonation and sand filtration; SE, standard error. Time scale is logarithmized. After 64 days the O group differs significantly from the other treatment groups (Fisher’s exact test: p < 0.001–0.01; n [ 95 (C), 116 (FS), 103 (O), 75 (OS)).
Fish test starting with yolk-sac fry & vitellogenesis
The fish test starting with the yolk-sac stage revealed no statistically significant differences in development between WW treatments and the control. The biomass exhibited no major deviations between exposure groups (<6% deviation from the control). Mortality was very low with the highest value in the control of 3.3 0.2% and a mortality of 0% in O (FS: 2.5 2.5%, OS: 1.7 1.7%). The vitellogenin content was significantly increased in specimens of the FS group (67.3 26.9 ng/mL; p < 0.05,
444
water research 44 (2010) 439–448
Fig. 5 – Biomass (A) and body length (B) in percentage relative to the control (0.28 ± 0.01 g and 34.1 ± 0.2 mm, respectively) of Oncorhynchus mykiss (mean values ± SE) exposed to differently treated and membrane filtered wastewaters. Abbreviations: C, control water; FS, final sedimentation; O, ozonation; OS, ozonation and sand filtration; SE, standard error. Significant differences to the control are indicated with white asterisks, between treatments with black asterisks (Kruskal–Wallis with Dunn’s post test: +, p < 0.05; +++, p < 0.001; n [ 73–114).
Kruskal–Wallis with Dunn’s post test) compared to fish maintained in reference water (11.0 6.6 ng/mL) whereas it was significantly decreased in the O and OS groups (4.6 1.9 and 4.8 1.9 ng/mL, respectively; p 0.01) compared to FS (Fig. 7).
4.
Discussion
4.1.
FELST with unfiltered wastewater
presumably are a result of microbial contamination as the reference water (C) remained free from mycelia and vorticellas whereas in the WW treatment groups fungi mycelia and vorticellas were found after 10–15 days exposure. However, the hatching success in the control (53.4 5.2%) did not comply with the validity criteria (>66%) according to OECD guideline (1992b) indicating that egg quality was not sufficient to run a valid test. The disinfectant effect of ozonation (Tyrrell et al., 1995) is likely to be responsible for the delayed egg
Unfiltered WW led to an increased coagulation rate of the exposed eggs in the FELST (Fig. 2). High coagulation rates
Fig. 6 – Mortality of Oncorhynchus mykiss post hatch (mean values ± SE) exposed to differently treated and membrane filtered wastewaters. Abbreviations: C, control water; FS, final sedimentation; O, ozonation; OS, ozonation and sand filtration; SE, standard error. Significant differences to control are indicated with white asterisks (Fisher’s exact test: +, p < 0.05; n [ 108 (C), 143 (FS), 135 (O), 90 (OS)).
Fig. 7 – Whole body vitellogenin concentration of Oncorhynchus mykiss (mean values ± SE) after 60 days exposure to differently treated wastewaters starting with the yolk-sac stage. Abbreviations: FS, final sedimentation; O, ozonation; OS, ozonation and sand filtration; C, control water; SE, standard error. Significant differences to control are indicated with white asterisks, between treatments with black asterisks (Kruskal–Wallis with Dunn’s post test: +, p < 0.05; ++, p < 0.01; n [ 22 (C, OS) – 33 (FS, O)).
water research 44 (2010) 439–448
coagulation in the O group compared to FS. Ozonation is also known to produce high amounts of assimilable organic carbons (AOC) which may have allowed a fast regrowth of microorganisms (Huang et al., 2005), resulting in still high coagulation rates in the O group. Sand filtration reduces the amount of suspended particulate matter (SPM) and of AOC (Wang and Summers, 1996). This may have reduced microbial development and resulting coagulation rates.
4.2.
FELST with membrane filtered wastewater
Exposure to membrane filtered WW revealed a distinct delay of the swim up process in the O group compared to FS and OS (Fig. 4) accompanied by a significant decrease in body weight and body length (Fig. 5). The reduced biomass and body length after ozonation is most likely a result of the delayed development because fish start exogenous ingestion at the onset of the swim up process. Consequently trout that swim up later may suffer from developmental disadvantages. Oxidative byproducts in ozonated WW may have impeded embryonic and/or larval development of fish. Possibly because sand filtration is able to remove ozonation metabolites (e.g. aldehydes, glyoxal, AOC; Wang and Summers, 1996) the effect in OS was reduced. However, no single compound could be clearly identified for the retardation effect. Possibly the sum of aldehydes (e.g. formaldehyde, glyoxal, acetaldehyde), carboxylic acids (e.g. formate), ketones and brominated organic compounds formed due to ozonation (Huang et al., 2005; Wert et al., 2007) caused these effects. Unfortunately, no chronic toxicity data for juvenile rainbow trout are available in literature for these compounds. Nevertheless, Wang and Summers (1996) were able to demonstrate that sand filtration efficiently removes aldehydes produced during ozone application, supporting this assumption. Besides, the formation of more complex organic metabolites evoked by ozonation may result in an increased toxicity compared to chemical precursors as it has already been documented for polycyclic aromatic hydrocarbons (PAHs). In the experiments of Luster-Teasley et al. (2002, 2005) chrysene and pyrene and byproducts as a result of ozonation were examined for their ability to disrupt gap junctional intercellular communication (GJIC), an indicator for tumor promoting properties. Among the transformation products, aldehydic compounds exhibited an increased toxicity compared to the precursor substance while their carboxylic structure analogue showed no GJIC disrupting activity. For the antiepileptic drug carbamazepine McDowell et al. (2005) identified three new oxidation products of unknown toxicity after ozone treatment. Furthermore, a possible increase in mutagenicity after ozonation of WW, as observed by Monarca et al. (2000) with the Ames test, verifies the potential of ozonation to produce toxic oxidation byproducts. In this context conversion of the widely used fungicide tolylfluanide into the carcinogen N-nitrosodimethylamine during ozonation was discovered by Schmidt and Brauch (2008). Based on these results it is likely that ozonation of WW leads to an increased number of unknown and potentially toxic metabolites depending on the composition of WW at the outset and the post treatment (e.g. sand filtration). Petala et al. (2006) demonstrated with the Vibrio fischeri bioluminescence test,
445
that toxicity originating from ozonated WW decreases with increasing storage time, indicating a rapid decomposition of toxic metabolites. In spite of the developmental retardation and the reduced biomass no significant differences in mortality rates were observed between the WW treatment groups (Fig. 6). Mortality was highest directly after ozonation (O, 24.0 4.0%). Considering the developmental retardation and the decreased biomass of fish exposed to ozonated water the increased mortality of O group specimens compared to FS and OS is noticeable and, although not significant, potentially a result of a reduced fitness of the fish in the ozonated water. It might be assumed that the retarded development is a consequence of an unspecific and general impairment of the fish’s health condition. This may increase their sensitivity towards environmental and anthropogenic stressors, thus leading to an increased mortality. Furthermore, under field conditions the delayed development possibly increases the risk for the fish to fall prey to predators since before swim up the larvae rest on the bottom, not capable to abscond effectively. Based on the assumption that the compounds causing adverse effects after ozonation are readily degradable it is possible that the detoxication after the ozone reactor will occur in the river as well. However, the discharge of ozonated WW without sand filtration would bear the risk to endanger fish populations in a considerable range of the receiving river, depending on WW load and flow velocity. Therefore, ozonation should always be followed by a sand filtration and further studies should focus on whether sand filtration is capable to remove oxidation byproducts sufficiently.
4.3.
Fish test starting with yolk-sac fry & vitellogenesis
The fish test with non-filtered WW starting with the yolk-sac stage resulted in negligible and insignificant deviations of development and biomass between treatment groups. Ozonation had no effect on these toxicity endpoints. These results indicate that yolk-sac fry 5 days post hatch are less sensitive to ozonation metabolites, compared to fish embryos and newly hatched fish, respectively. Nevertheless, the amount of suspended particulate matter in the exposure vessels and in the Teflon tubes was accompanied by an increased biofilm formation. This may have contributed to an increased detoxication of the ozonated WW due to (bio-) degradation of oxidation byproducts. The significant increase of vitellogenin content in fish exposed to FS compared to the reference water indicates an environmentally relevant contamination of the WW with estrogenic active compounds (Fig. 7). After ozonation the VTG content decreased even below the control level. The formation of antiestrogenic compounds is not likely because the VTG decrease compared to control level is not significant and phenols, as an important functional group interacting with the estrogen receptor (Nishihara et al., 2000), are particularly susceptible to ozone attack (von Gunten, 2003). These results confirm the high efficiency of ozonation to eliminate estrogenic contamination in WW as already shown by Huber et al. (2004) with an in vitro test system. Based on these results it can be assumed that ozonation is well suited to reduce the estrogenic burden of WW below environmental relevance.
446
water research 44 (2010) 439–448
Regarding ecosystem health it should be considered that estrogen and xenoestrogen concentrations detected in the environment have been shown to impair the sustainability of wild fish populations (Kidd et al., 2007), arguing for the establishment of an end of pipe technique for estrogen removal to protect fish populations in surface waters receiving high amounts of WW.
4.4.
Ozonation vs. alternatives
Currently there is a controversy regarding the appropriate advanced WW treatment techniques to reduce contamination of the aquatic ecosystem with micropollutants. Alternative processes other than ozonation also deliver promising results with regard to the removal of contaminants. Wastewater treatment with powdered activated carbon (PAC) achieves reduction rates of 75–90% for pharmaceuticals (including X-ray contrast media) with PAC dosages of 10–20 mg/L (Pu¨ttmann et al., 2008). The main advantage of PAC treatment is that a reduced chemical concentration is equivalent to removal of chemicals whereas ozonation leads basically to a transformation of the compounds. According to cost estimations PAC treatment is expected to be about 30% more expensive than ozonation (Joss et al., 2008). However, the main disadvantage of PAC treatment might be the need for the disposal of the used and contaminated carbon. Membrane filtration is suitable for micropollutants retention but as a result of considerably higher requirement for energy and technical equipment economically not competitive with ozonation or activated carbon treatment (Joss et al., 2008). Overall, end of pipe techniques are presumably a suitable solution to reduce toxicity of hazardous WW in a mediumterm perspective. Nevertheless in the long term source control strategies such as wastewater separation (e.g. urine separation), ecologically correct disposal of drugs by end users, reuse or recycling by the pharmaceutical industry or alternative medical treatments to drug therapies (Daughton, 2003; Joss et al., 2006) could offer environmentally friendly and affordable options.
5.
Conclusions
The sanitizing effect of ozonation was confirmed. Coagulation rates of eggs exposed to ozonated wastewater were on a lower level compared to those exposed to conventionally treated wastewater. Disinfection is presumably only efficient when linked to sand filtration because of rapid recovery of microorganisms due to increased assimilable organic carbon formation as a result of ozonation. Membrane filtered wastewater (for removal of microorganisms) reveals developmental retardation directly after ozonation. After sand filtration this adverse effect disappears. The reduced biomass and body length in fish exposed to ozonated wastewater is most probably a result of the formation of toxic oxidation byproducts. The mortality in the ozonated wastewater was significantly increased as a result of retarded development. Impairment of the fish’s health condition may increase the sensitivity towards environmental and anthropogenic stressors.
Developmental retardation might increase the risk for the fish to fall prey to predators because swim up stage is delayed. Ozonation should not be applied without appropriate barrier for oxidation byproducts. Effectiveness of sand filtration for removal of oxidation byproducts should be further evaluated. Reduction of vitellogenin content in fish exposed to ozonated wastewater on control level confirms the suitability of this technique to reduce estrogenic activity, possibly below environmental relevance.
Acknowledgements The authors should like to express their gratitude to the staff from WWTP Wu¨eri for their technical cooperation and Adriano Joss from the EAWAG for helpful suggestions and technical support. Ulrike Schulte-Oehlmann is acknowledged for critical comments and helpful suggestions on the manuscript. This study was part of the EU project Neptune (contract no 036845, SUSTDEV-2005-3.II.3.2) within the Energy, Global Change and Ecosystems Programme of the Sixth Framework (FP6-2005-Global-4) and co-funded by Bundesamt fu¨r Umwelt (BAFU), Bern (CH) within the Strategy MicroPoll Programme (contract no 05.0013.PJ/F471-0916).
references
Andreozzi, R., Campanella, L., Fraysse, B., Garric, J., Gonnella, A., Lo Giudice, R., Marotta, 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), 23–28. Bader, H., Hoigne, J., 1981. Determination of ozone in water by the indigo method. Water Research 15 (4), 449–456. Brian, J.V., Harris, C.A., Scholze, M., Backhaus, T., Booy, P., Lamoree, M., Pojana, G., Jonkers, N., Runnalls, T., Bonfa, A., Marcomini, A., Sumpter, J.P., 2005. Accurate prediction of the response of freshwater fish to a mixture of estrogenic chemicals. Environmental Health Perspectives 113 (6), 721–728. Daughton, C.G., 2003. Cradle-to-cradle stewardship of drugs for minimizing their environmental disposition while promoting human health. II. Drug disposal, waste reduction, and future directions. Environmental Health Perspectives 111 (5), 775–785. Daughton, C.G., Ternes, T.A., 1999. Pharmaceuticals and personal care products in the environment: agents of subtle change? Environmental Health Perspectives 107, 907–938. Escher, B.I., Bramaz, N., Quayle, P., Rutishauser, S., Vermeirssen, E.L.M., 2008. Monitoring of the ecotoxicological hazard potential by polar organic micropollutants in sewage treatment plants and surface waters using a mode-of-action based test battery. Journal of Environmental Monitoring 10 (5), 622–631. European Commission, 2000. 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. Official Journal of the European Communities L327. Retrieved from: http://eur-lex.europa.eu/
water research 44 (2010) 439–448
LexUriServ/LexUriServ.do?uri¼OJ: L:2000:327:0001:0072:EN: PDF [12.02.2009]. Gagne, F., Andre, C., Cejka, P., Gagnon, C., Blaise, C., 2007. Toxicological effects of primary-treated urban wastewaters, before and after ozone treatment, on freshwater mussels (Elliptio complanata). Comparative Biochemistry and Physiology C – Toxicology & Pharmacology 145 (4), 542–552. Holbech, H., Kinnberg, K., Petersen, G.I., Jackson, P., Hylland, K., Norrgren, L., Bjerregaard, P., 2006. Detection of endocrine disrupters: evaluation of a fish sexual development test (FSDT). Comparative Biochemistry and Physiology C – Toxicology & Pharmacology 144 (1), 57–66. Huang, W.J., Fang, G.C., Wang, C.C., 2005. The determination and fate of disinfection by-products from ozonation of polluted raw water. Science of the Total Environment 345 (1-3), 261–272. Huber, M.M., Ternes, T.A., von Gunten, U., 2004. Removal of estrogenic activity and formation of oxidation products during ozonation of 17a-ethinylestradiol. Environmental Science & Technology 38 (19), 5177–5186. Huber, M.M., Gobel, A., Joss, A., Hermann, N., Loffler, D., McArdell, C.S., Ried, A., Siegrist, H., Ternes, T.A., von Gunten, U., 2005. Oxidation of pharmaceuticals during ozonation of municipal wastewater effluents: a pilot study. Environmental Science & Technology 39 (11), 4290–4299. Jahnel, J., Neamtu, M., Schudoma, D., Frimmel, F.H., 2006. Scientific risk assessment of considered water relevant substances (Bestimmung von Umweltqualitaetsnormen fuer potenziell gewaesserrelevante Stoffe). Acta Hydrochimica et Hydrobiologica 34 (4), 389–397. Jensen, G.K., 2007. European Parliament increases number of chemicals earmarked for priority action by European Commission. The Lancet Oncology 8 (5), 382–383. Jobling, S., Sheahan, D., Osborne, J.A., Matthiessen, P., Sumpter, J.P., 1996. Inhibition of testicular growth in rainbow trout (Oncorhynchus mykiss) exposed to estrogenic alkylphenolic chemicals. Environmental Toxicology and Chemistry 15 (2), 194–202. Jobling, S., Coey, S., Whitmore, J.G., Kime, D.E., Van Look, K.J.W., McAllister, B.G., Beresford, N., Henshaw, A.C., Brighty, G., Tyler, C.R., Sumpter, J.P., 2002. Wild intersex roach (Rutilus rutilus) have reduced fertility. Biology of Reproduction 67 (2), 515–524. Joss, A., Klaschka, U., Knacker, T., Liebig, M., Lienert, J., Ternes, T.A., Wennmalm, A., 2006. Source control, source separation. In: Ternes, T.A., Joss, A. (Eds.), Human Pharmaceuticals, Hormones and Fragrances. The Challenge of Micropollutants in Urban Water Management. ISBN 1843390930, pp. 353–384. Joss, A., Siegrist, H., Ternes, T.A., 2008. Are we about to upgrade wastewater treatment for removing organic micropollutants? Water Science and Technology 57 (2), 251–255. 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), 8897–8901. Larsson, D.G.J., Adolfsson-Erici, M., Parkkonen, J., Pettersson, M., Berg, A.H., Olsson, P.E., Forlin, L., 1999. Ethinylestradiol – an undesired fish contraceptive? Aquatic Toxicology 45 (2-3), 91–97. Luster-Teasley, S.L., Yao, J.J., Herner, H.H., Trosko, J.E., Masten, S.J., 2002. Ozonation of chrysene: evaluation of byproduct mixtures and identification of toxic constituent. Environmental Science & Technology 36 (5), 869–876. Luster-Teasley, S.L., Ganey, P.E., DiOrio, M., Ward, J.S., Maleczka, R.E., Trosko, J.E., Masten, S.J., 2005. Effect of byproducts from the ozonation of pyrene: biphenyl-2,20 ,6,60 tetracarbaldehyde and biphenyl-2,20 ,6,60 -tetracarboxylic acid
447
on gap junction intercellular communication and neutrophil function. Environmental Toxicology and Chemistry 24 (3), 733–740. McDowell, D.C., Huber, M.M., Wagner, M., Von Gunten, U., Ternes, T.A., 2005. Ozonation of carbamazepine in drinking water: identification and kinetic study of major oxidation products. Environmental Science & Technology 39 (20), 8014–8022. Moltmann, J.F., Liebig, M., Knacker, T., Keller, M., Scheurer, M., Ternes, T., 2007. Relevance of Endocrine Disrupting Substances and Pharmaceuticals in Surface Waters (Gewa¨sserelevanz endokriner Stoffe und Arzneimittel). Final report. Federal Environment Agency. Funding code 20524205. Monarca, S., Feretti, D., Collivignarelli, C., Guzzella, L., Zerbini, I., Bertanza, G., Pedrazzani, R., 2000. The influence of different disinfectants on mutagenicity and toxicity of urban wastewater. Water Research 34 (17), 4261–4269. 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. Water Research 41 (19), 4373–4382. Nishihara, T., Nishikawa, J., Kanayama, T., Dakeyama, F., Saito, K., Imagawa, M., Takatori, S., Kitagawa, Y., Hori, S., Utsumi, H., 2000. Estrogenic activities of 517 chemicals by yeast two-hybrid assay. Journal of Health Science 46 (4), 282–298. OECD, 1992a. Fish Acute Toxicity Test. OECD Guidelines for the Testing of Chemicals. No. 203. Organisation for Economic Cooperation and Development, Paris. OECD, 1992b. Fish Early Life Stage Toxicity Test. OECD Guidelines for the Testing of Chemicals. No. 210. Organisation for Economic Cooperation and Development, Paris. Parrott, J.L., Blunt, B.R., 2005. Life-cycle exposure of fathead minnows (Pimephales promelas) to an ethinylestradiol concentration below 1 ng/L reduces egg fertilization success and demasculinizes males. Environmental Toxicology 20 (2), 131–141. Petala, M., Samaras, P., Zouboulis, A., Kungolos, A., Salkellaropoulos, G., 2006. Ecotoxicological properties of wastewater treated using tertiary methods. Environmental Toxicology 21 (4), 417–424. Pu¨ttmann, W., Keil, F., Oehlmann, J., Schulte-Oehlmann, U., 2008. Strategy to reduce pharmaceuticals in drinking water – technical approach (Wassertechnische Strategien zur Reduzierung der Trinkwasserbelastung durch Arzneimittelwirkstoffe). Umweltwissenschaften und Schadstoff-Forschung 20 (3), 209–226. Purdom, C.E., Hardiman, P.A., Bye, V.V.J., Eno, N.C., Tyler, C.R., Sumpter, J.P., 1994. Estrogenic effects of effluents from sewage treatment works. Chemical Ecology 8 (4), 275–285. Rodgers-Gray, T.P., Jobling, S., Kelly, C., Morris, S., Brighty, G., Waldock, M.J., Sumpter, J.P., Tyler, C.R., 2001. Exposure of juvenile roach (Rutilus rutilus) to treated sewage effluent induces dose-dependent and persistent disruption in gonadal duct development. Environmental Science & Technology 35 (3), 462–470. Schmidt, C.K., Brauch, H.J., 2008. N, N-dimethosulfamide as precursor for N-nitrosodimethylamine (NDMA) formation upon ozonation and its fate during drinking water treatment. Environmental Science & Technology 42 (17), 6340–6346. Schwaiger, J., Ferling, H., Mallow, U., Wintermayr, H., Negele, R.D., 2004. Toxic effects of the non-steroidal antiinflammatory drug diclofenac Part 1: histopathological alterations and bioaccumulation in rainbow trout. Aquatic Toxicology 68 (2), 141–150. Schwarzenbach, R.P., Escher, B.I., Fenner, K., Hofstetter, T.B., Johnson, C.A., von Gunten, U., Wehrli, B., 2006. The challenge
448
water research 44 (2010) 439–448
of micropollutants in aquatic systems. Science 313 (5790), 1072–1077. Ternes, T.A., Stumpf, M., Mu¨ller, J., Haberer, K., Wilken, R.D., Servos, M., 1999. Behavior and occurrence of estrogens in municipal sewage treatment plants – I. Investigations in Germany, Canada and Brazil. Science of the Total Environment 225 (1-2), 81–90. Tixier, C., Singer, H.P., Oellers, S., Mu¨ller, S.R., 2003. Occurrence and fate of carbamazepine, clofibric acid, diclofenac, ibuprofen, ketoprofen, and naproxen in surface waters. Environmental Science & Technology 37 (6), 1061–1068. Triebskorn, R., Casper, H., Heyd, A., Eikemper, R., Ko¨hler, H.R., Schwaiger, J., 2004. Toxic effects of the non-steroidal antiinflammatory drug diclofenac Part II. Cytological effects in liver, kidney, gills and intestine of rainbow trout (Oncorhynchus mykiss). Aquatic Toxicology 68 (2), 151–166.
Tyrrell, S.A., Rippey, S.R., Watkins, W.D., 1995. Inactivation of bacterial and viral indicators in secondary sewage effluents, using chlorine and ozone. Water Research 29 (11), 2483–2490. von Gunten, U., 2003. Ozonation of drinking water: Part I. Oxidation kinetics and product formation. Water Research 37 (7), 1443–1467. Wang, J.Z., Summers, R.S., 1996. Biodegradation behavior of ozonated natural organic matter in sand filters. Revue des Sciences de l’Eau 9 (1), 3–16. Wert, E.C., Rosario-Ortiz, F.L., Drury, D.D., Snyder, S.A., 2007. Formation of oxidation byproducts from ozonation of wastewater. Water Research 41 (7), 1481–1490. Ying, G.G., Williams, B., Kookana, R., 2002. Environmental fate of alkylphenols and alkylphenol ethoxylates – a review. Environment International 28 (3), 215–226.
water research 44 (2010) 449–460
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
The role of organic matter in the removal of emerging trace organic chemicals during managed aquifer recharge T. Rauch-Williams, C. Hoppe-Jones, J.E. Drewes* Advanced Water Technology Center (AQWATEC), Colorado School of Mines, Environmental Science and Engineering Division, Golden, CO 80401-1887, United States
article info
abstract
Article history:
This study explored the effect of different bulk organic carbon matrices on the fate of trace
Received 20 April 2009
organic chemicals (TOrC) during managed aquifer recharge (MAR). Infiltration through
Received in revised form
porous media was simulated in biologically active column experiments under aerobic and
11 August 2009
anoxic recharge conditions. Wastewater effluent derived organic carbon types, differing in
Accepted 20 August 2009
hydrophobicity and biodegradability (i. e., hydrophobic acids, hydrophilic carbon, organic
Available online 27 August 2009
colloids), were used as feed substrates in the column experiments. These carbon substrates while fed at the same concentration differed in their ability to support soil biomass growth
Keywords:
during porous media infiltration. Removal of degradable TOrC (with the exception of
Trace organic chemicals (TOrC)
diclofenac and propyphenazone) was equal or better under aerobic versus anoxic porous
Groundwater recharge
media infiltration conditions. During the initial phase of infiltration, the presence of
Effluent organic matter
biodegradable organic carbon (BDOC) enhanced the decay of degradable TOrC by
Managed aquifer recharge
promoting soil biomass growth, suggesting that BDOC served as a co-substrate in a co-
Riverbank filtration
metabolic transformation of these contaminants. However, unexpected high removal
Biotransformation
efficiencies were observed for all degradable TOrC in the presence of low BDOC concen-
Co-metabolism
trations under well adopted oligotrophic conditions. It is hypothesized that removal under
Primary substrate
these conditions is caused by a specialized microbial community growing on refractory carbon substrates such as hydrophobic acids. Findings of this study reveal that the concentration and character of bulk organic carbon present in effluents affect the degradation efficiency for TOrC during recharge operation. Specifically aerobic, oligotrophic microbiological soil environments present favorable conditions for the transformation of TOrC, including rather recalcitrant compounds such as chlorinated flame retardants. ª 2009 Elsevier Ltd. All rights reserved.
1.
Introduction
Managed aquifer recharge (MAR) systems, such as riverbank filtration (RBF) and soil aquifer treatment (SAT), are widely used natural processes for drinking water augmentation projects using source water that might be impaired by wastewater discharge. Previous studies have demonstrated that MAR systems are effective in dampening and reducing
the concentrations of dissolved organic carbon (DOC) as well as various trace organic contaminants (TOrC) that might be present in impaired source waters (Drewes and Fox, 1999; Brauch et al., 2000; Gru¨nheid et al., 2005). The presence of TOrC has become a key concern for drinking water augmentation projects during the past decade (Kolpin et al., 2002; Heberer, 2002; Focazio et al., 2008). Although adverse human health effects caused by these compounds at concentrations
* Corresponding author. E-mail address:
[email protected] (J.E. Drewes). 0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2009.08.027
450
water research 44 (2010) 449–460
commonly observed in impaired water resources are very unlikely (Schwab et al., 2005), minimizing exposure of wastewater derived contaminants in these projects is desired. Previous research on the fate of TOrC during MAR has primarily focused on collecting anecdotal and site specific information on their occurrence and removal (Drewes et al., 2002; Montgomery-Brown et al., 2003; Gru¨nheid et al., 2005; Dillon et al., 2008). Studies delineating the mechanisms and boundary conditions for the transformation of wastewater derived TOrC during MAR are lacking. Previous studies demonstrated that the type and bioavailability of effluent organic matter (EfOM) controls the extent of soil biomass growth in MAR systems (Rauch and Drewes, 2004; Rauch and Drewes, 2005). EfOM may consequently effect the degradation of TOrC by serving as a co-substrate in microbiologically facilitated transformations (Stratton et al., 1983). The diversity and expression of the soil microbial community also depends on composition and concentration of the organic carbon substrate controlling trophic cycles in the subsurface (Preuß and Nehrkorn, 1996; Szewzyk et al., 1998). The composition of EfOM (i.e. in terms of its bioavailability) is primarily determined by the degree of wastewater treatment employed (Drewes and Fox, 1999), which can vary widely from primary to conventional to advanced wastewater treatment. As a result, effluent qualities fed to MAR systems can vary in biodegradable dissolved organic carbon (BDOC) concentrations from less than 1 up to 15 mg/L or more. As a consequence, soil microbial communities growing on different levels of BDOC can differ widely in total biomass and diversity. The objectives of this research were to investigate the role of 1) abiotic vs. biotic conditions, 2) BDOC and 3) the type of organic carbon matrices on the removal of select TOrC, such as pharmaceutical residues, personal care products, and household chemicals, during MAR.
2.
Methodology
2.1.
Target organic contaminants
Compounds selected for this study represent small molecular weight organic chemicals (180 to 360 Dalton) that are hydrophilic at neutral pH regimes as indicated by an octanol/water partition coefficient at pH 7 (log DpH¼7 of less than 2.6). TOrC with these properties have a high potential to migrate into groundwater and are not expected to be adsorbed onto porous media. The molecular structures and physicochemical properties of the target compounds chosen for this study are presented in Table 1. These compounds cover a wide range of biodegradability as previously reported for soil/water systems. The anticonvulsants carbamazepine and primidone have been classified as recalcitrant during wastewater treatment and MAR in earlier studies (Heberer, 2002; Drewes et al., 2003; Clara et al., 2004). The chlorinated phosphate esters tris (1-chloroisopropyl)-phosphate (TCPP) and tris(2-chloroethyl)phosphate (TCEP) are two widely used flame retardants and are persistent in the aquatic environment (Heberer et al., 2001; Fries and Pu¨ttmann, 2003). During bank filtration in Germany, however, TCPP and TCEP exhibited a significant reduction, which was attributed to biotransformation in the aquifer
(Heberer et al., 2003). Amy and Drewes (2006) also reported removal of TCEP to concentrations below the detection limit after two years of subsurface travel in an MAR facility supporting that chlorinated flame retardants can be biotransformed under anoxic conditions. Propyphenazone is a poorly biodegradable analgesic that persists during RBF (Heberer et al., 2003) and SAT (Drewes et al., 2003). For diclofenac, a popular analgesic drug, low removal due to biodegradation or adsorption was reported (Buser et al., 1998; Mo¨hle et al., 1999) unless soils contain a high organic carbon content (Drillia et al., 2003). Several studies report a faster degradation of diclofenac under anoxic conditions as compared to aerobic conditions (Zwiener and Frimmel, 2003; Hua et al., 2003). Opposing results were reported by Schmidt et al. (2004) in that diclofenac was almost completely removed during aerobic bank filtration but recalcitrant during anaerobic recharge. Ibuprofen, ketoprofen, and naproxen are common analgesics that are well degradable during wastewater treatment (Buser et al., 1999; Zwiener and Frimmel, 2003; Carballa et al., 2004) and during soil infiltration (Sedlak and Pinkston, 2001; Drewes et al., 2003). Gemfibrozil is a commonly prescribed blood lipid regulator found in unconfined shallow aquifers impacted by wastewater infiltration in the low to moderate ng/L-concentration range (Heberer and Stan, 1997; Drewes and Shore, 2001; Heberer, 2002). Gemfibrozil was removed below the limit of detection within a few weeks during groundwater recharge using SAT (Drewes et al., 2003).
2.2.
Analytical methods
2.2.1.
GC/MS analysis
The following TOrC were analyzed by gas chromatography coupled with mass spectrometry (GC/MS) using a HP 6890 gas chromatograph and a HP 5973 quadrupole mass spectrometer from Agilent Technologies (Waldbronn, Germany) adopting a method published by Reddersen and Heberer (2003): gemfibrozil, primidone, diclofenac sodium salt, carbamazepine, ketoprofen, naproxen, phenacetine, tris (2-chloroethyl)phosphate (TCEP) (Sigma Aldrich Chemicals), tris(chloropropyl)phosphate (TCPP), and propyphenazone (Pfaltz & Bauer, Inc.). Stock solutions were prepared by dissolving the compounds in milli-Q water adjusted to a pH of 10, during sonification and in the dark. A volume of 500 mL of sample was collected and filtered (0.45 mm, Whatman) prior to solid phase extraction. 10,11-dihydrocarbamazepine (Sigma Aldrich Chemicals) and 2-(m-chlorophenoxy) propionic acid (Sigma Aldrich Chemicals) were used as surrogate standards. The limits of detection (LofD) and limit of quantification (LoQ) for the target compounds ranged from 5 to 10 ng/L and from 10 to 50 ng/L, respectively. Only parent target compounds were investigated in this study, metabolites and conjugates were not analyzed.
2.2.2.
HPLC analysis
A high performance liquid chromatography (HPLC) HP 1100 (Agilent Technologies) combined with a UV diode array detector (DAD) was used to quantify concentrations of primidone, phenacetine, carbamazepine, naproxen, diclofenac, and ibuprofen in the lower mg/L range during adsorption breakthrough tests. The samples were directly injected without filtration or other
451
water research 44 (2010) 449–460
Table 1 – Physical–chemical properties of target organic compounds. Compound
Carbamazepine (anticonvulsant)
Structure
pKa
log P (log DpH7)
Expected degradability
References
13.9
2.67
Persistent
Mersmann et al., 2002; Schmidt et al., 2004
Diclofenac sodium (analgesic/antiinflammatory)
4.0
4.06 (1.06)
Degradable (anoxic)
Buser et al., 1998; Mo¨hle et al., 1999; Drillia et al., 2003; Zwiener and Frimmel, 2003, Hua et al., 2003; Schmidt et al., 2004
Gemfibrozil (blood lipid regulator)
4.8
4.39 (2.19)
Relatively persistent
Heberer and Stan, 1997; Drewes and Shore, 2001; Heberer, 2002; Drewes et al., 2003
3.72 (1.12)
Degradable (better aerobic)
Buser et al., 1999; Zwiener and Frimmel, 2003; Carballa et al., 2004; Sedlak and Pinkston, 2001; Drewes et al., 2003
Degradable
Buser et al., 1999; Zwiener and Frimmel, 2003; Carballa et al., 2004; Sedlak and Pinkston, 2001; Drewes et al., 2003
Degradable
Buser et al., 1999; Zwiener and Frimmel, 2003; Carballa et al., 2004; Sedlak and Pinkston, 2001; Drewes et al., 2003
Relatively persistent
Buser et al., 1999; Zwiener and Frimmel, 2003; Carballa et al., 2004; Sedlak and Pinkston, 2001; Drewes et al., 2003
Ibuprofen (analgesic/antiinflammatory)
Ketoprofen (analgesic/antiinflammatory)
Naproxen (analgesic/antiinflammatory)
Phenacetine (antipyretic)
4.4
4.2
4.0
n/a
2.81 (0.01)
3.00 (0.00)
1.63 (1.63)
Primidone (anticonvulsant)
12.3
0.4 (0.4)
Persistent
Buser et al., 1999; Zwiener and Frimmel, 2003; Carballa et al., 2004; Sedlak and Pinkston, 2001; Drewes et al., 2003
Propyphenazone (analgesic)
n/a
1.74 (1.74)
Poor/ persistent
Heberer et al., 2003; Drewes et al., 2003
(continued on next page)
452
water research 44 (2010) 449–460
Table 1 (continued ) Compound
Structure
Tris(2chloroethyl)phosphate (TCEP) (flame retardant)
pKa
n/a
log P (log DpH7)
Expected degradability
0.48 (0.48)
Relatively persistent
References
Heberer et al., 2001; Heberer et al., 2003, Amy and Drewes, 2006; Fries and Pu¨ttmann, 2003
n/a not applicable. pKa and log P values calculated by software ACD. log DpH¼7 values calculated using equations proposed by Scherrer and Howard, (1977).
sample preparation. The mobile phase consisted of two solvents: 1) type 1 water (adjusted to pH 2.5 with o-phosphoric acid (HPLC grade, Fisher Scientific) and buffered with 25 mmol potassium phosphate monobasic (Fisher Scientific)), and 2) acetonitrile (HPLC grade, Mallinckrodt ChromAR HPLC) with the same concentration of o-phosphoric acid as solvent 1). Sample injection (100 mL) occurred in triplicates for each sample at a flow rate of 2 mL/min. A mobile phase gradient was applied during the run for solvent 2 (acetonitrile) of 30% at t ¼ 0 min., 80% at t ¼ 12 min., and 30% at t ¼ 15 min. with a post-run time of 7 min. for column cleaning (30% of solvent 2). The detection wavelengths (band width 10 nm) were set to 205 nm at t ¼ 0 min. for quantification of primidone, phenacetine, and carbamazepine, 230 nm at t ¼ 6.8 min. (naproxen), and 205 nm at 7.8 min. for detection of diclofenac and ibuprofen with a reference wavelength of 330 nm (band width 60 nm) for all compounds. Standards were run in the range of 5–500 mg/L. The detection limits for each compound were 5 mg/L.
2.2.3.
DOC/UV absorbance/nitrate/ammonium
A Sievers 800 Total Organic Carbon Analyzer (GE, Boulder) was used for DOC quantification after microfiltration (0.45 mm, Whatman) (Standard Method 5310C). UV absorbance (UVA) measurements were conducted at 254 nm using a Beckman Coulter DU 800 Spectrophotometer after 0.45 mm filtration (Standard method 5910B). The specific UV absorbance (SUVA) was calculated as the ratio of UVA and DOC. Ammonium and nitrate concentrations were measured using the Nessler and the chromotropic acid method (HACH) with a detection range of 0.02–2.0 mg/l and 0.2–30 mg/L, respectively.
2.2.4.
Biomass
Soil biomass was determined as total viable biomass (i.e. viable, not necessarily active bacteria) using phospholipid extraction (PLE) as described in Rauch and Drewes (2005). Analyses were conducted in triplicates from the top-soil (0–2 cm, infiltration zone) of the columns.
2.2.5.
Wastewater effluent
Secondary treated wastewater effluent served as the feed to column systems and as source for subsequent organic carbon fractionation. The secondary treated effluent was collected from a local wastewater treatment plant employing nitrification and partial denitrification. The average DOC concentration of the effluent was 8.74 1.44 mg/L.
2.2.6.
Organic carbon fractionation
Organic matter (less than 1 mm in size as defined for this study) of the secondary effluent sample was isolated into three organic fractions: colloidal organic carbon, hydrophobic acids (HPO-A), and hydrophilic carbon (HPI) following the procedure described in Rauch and Drewes (2005) with some modifications. The sample was concentrated by vacuum rotary evaporation at 45 C (concentration factor: 40). The concentrate was separated by dialysis (Spectra/Por, Spectrum, 6000– 8000 Dalton) at pH 4–5 into organic colloids and DOC. The permeate of the dialysis (containing DOC) was collected and further separated into HPI and HPO-A by XAD-8 fractionation according to Leenheer et al. (2000) using a capacity factor of k0 ¼ 4. A carbon mass balance was performed for each carbon fractionation based on UVA and DOC measurements for quality control. In average, the secondary treated effluent contained 16 percent colloidal organic carbon, 38 percent HPO-A, 37 percent HPI, and 8 percent hydrophobic neutrals (HPO-N) (Rauch and Drewes, 2004). Organic carbon isolates were diluted to 3 mg/L DOC using milli-Q water, adjusted for ion strength, macro nutrient (nitrogen, phosphate) and micro nutrient concentrations (Rauch and Drewes, 2004) and promptly utilized as column feed waters.
2.3.
Experimental set-up
2.3.1.
Anoxic column system
The columns (PC system) were operated to study TOrC removal during simulated MAR under anoxic conditions and up to 3–4 weeks of retention time in the subsurface. The anoxic columns consisted of four 1-m plexiglass columns (15 cm i.d.) in series filled with aquifer material (d50 ¼ 0.8 mm, foc ¼ 0.003%) (Fig. SI-1, Supplemental Information). The column system was operated in flow-through mode at a loading rate of 0.065 m/d under saturated, anoxic (denitrifying) flow conditions (Table 2). The hydraulic retention time of the system using four columns in series was previously determined as 25 days. The columns had been continuously fed with secondary or tertiary treated effluents for over 6 years and for 5 months with the secondary treated effluent quality employed in this study prior to spiking of TOrC. The column influent was regularly purged with nitrogen gas to keep dissolved oxygen concentrations below 0.5 mg/L. Samples were collected once to twice a week from column influent and the four column effluents, respectively, and analyzed for DOC, UV absorbance, pH, conductivity and TOrC concentrations.
453
water research 44 (2010) 449–460
Table 2 – Operational parameters and water quality changes of columns used in this study.a Label
PC C1 C2 C3 C4 Abiotic
Feed water matrix
Length (m)
Anoxic columns (EfOM) Hydrophobic acids (HPO-A) Hydrophilic carbon (HPI) Effluent organic matter (EfOM) Organic colloids (Org. Colloids) Abiotic column, Type I water
41m¼ 4m 0.3
Organic Carbon
Feed water (mg/L) 6.9
Effluent (mg/L)
BOC (mg/L)
Predominant Redox Condition
Total Viable Soil Biomassb (nmol PO3 4 /g d.w soil)
Anoxic
17.3 1.3
3.1 0.9
See Fig. 1 for DOC column profile 2.7 0.02 0.4 0.9
Aerobic
3.3 0.4
0.3
2.8 0.4
2.6 0.1
0.3 0.4
Aerobic
10.5 5.3
0.3
3.0 0.4
2.7 0.4
0.4 0.4
Aerobic
20.6 2.5
0.3
3.1 0.5
2.2 0.3
0.8 0.5
Aerobic
27.2 1.8
0.3
<0.1
<0.1
0
Abiotic
0.5 0.3
a All experiments conducted at feed water conductivity of 940–1500 mS/cm and pH 7.0–7.9. b Biomass samples were collected from the infiltration zone of all columns at 0–2 cm depth.
2.3.2.
Organic carbon fraction columns
Four columns (C1–C4) were operated to examine the effect of different organic carbon matrices on soil biomass growth and TOrC removal during aerobic conditions and short retention times (less than 1 day) in the subsurface. This system consisted of four parallel plexiglass columns (L ¼ 30 cm, i.d. ¼ 5 cm) filled with silica sand (d50 ¼ 0.65 mm, foc ¼ 0.004%). The columns were operated under saturated, aerobic flow conditions at a loading rate of 0.9 m/d representing a retention time of about 19 h in the porous media. The system had been acclimated for over two years to the infiltration of the four organic carbon fractions derived from secondary effluent (HPO-A, HPI, organic colloids, EfOM), respectively. Approximately six weeks prior to spiking TOrC to the column systems, the feed water concentrations of all four fractions was adjusted to 3 mg/L as DOC. Samples for DOC, UVA and pH were taken once to twice a week from the four column influents and effluents. Three consecutive TOrC spiking events were conducted. For each event, hydraulically corresponding samples of column influents and effluents were collected, respectively, after a complete breakthrough of the spike solution was observed using conductivity as a conservative tracer (typically after 24 h). After the 1st spiking event (day 0), the 2nd and 3rd spikes occurred on day 11 and 22, respectively. Set-up and operation conditions of the different columns employed in this study are summarized in Table 2. All small column systems received feed waters that were adjusted to a pH of 8.0 0.2 to minimize adsorption effects of acidic TOrC onto porous media.
2.3.3.
Adsorption test
An additional column was operated under abiotic conditions to assess the role of physical adsorption of selected TOrC onto porous media. The column set-up and operation was the same as described for the biologically active columns C1–C4. The adsorption column was filled with aquifer material from an RBF site (d50 ¼ 0.55 mm, bulk density ¼ 1.81 g/cm3, effective porosity ¼ 0.36, fOC ¼ 0.066%). Six TOrC were spiked into secondary effluent in a concentration range of 175–226 mg/L adjusted to a pH of 7.0 (feed water). To minimize biotransformation during this experiment, the column was kept
abiotic by adding sodium azide at a concentration of 2 mM. Sodium bromide was added as a conservative tracer and bromide was measured by ion chromatography (Dionex, Sunnyvale, CA) in the column effluent. The column was operated in single flow-through mode under saturated flow conditions at a hydraulic loading rate of 0.33 m/d. Effluent samples were collected using a fraction collector and analyzed by HPLC analysis. Retardation factors (Rd) were fitted to the step input breakthrough curves for each compound using the software program CXTFIT (Toride et al., 1995). Adsorption coefficients (Kd) were calculated according to the following relationship Rd ¼
rKd þ1 3
(1)
where Rd is the retardation factor, r is the sand bulk density, and 3 is the sand porosity.
3.
Results and discussion
3.1.
Adsorption behavior of target TOrC
In order to assess whether physical retardation would be a contributing factor in the removal of select TOrC in the column experiments, an abiotic column experiment was conducted to study sorption of the compounds onto the virgin porous media used in the columns. None of the six TOrC included in this experiment exhibited significant retardation during flow through porous media as compared to the conservative tracer bromide (indicated by Rd values very close to 1, Table 3). These findings reveal that sorption did not play an important role in the attenuation of these compounds onto the porous media used in the columns.
3.2.
Organic carbon removal
During the course of the study, the anoxic column system exhibited a steady removal of DOC (55%). The DOC removal observed in the column system followed a first-order kinetic
454
water research 44 (2010) 449–460
Table 3 – Retardation and sorption coefficients of select TOrC derived during abiotic column study. TOrC Naproxen Ibuprofen Phenacetine Diclofenac Primidone Carbamazepine
Rd
Kd (mL/g)
Mass recovery (%)
1.30 1.10 1.02 1.41 1.00 1.90
0.08 0.03 n.a. 0.01 <0.001 0.24
94.2 92.7 n.a. 93.6 102 94.0
degradation profile (Fig. 1). SUVA increased from 1.7 to 2.1 within 25 days of hydraulic residence time indicating a preferred removal of saturated organic compounds. Average DOC concentrations observed in the influent and effluent of column systems C1, C2, C3, and C4 fed with different organic carbon fractions are summarized in Table 2. Average influent DOC concentrations in all columns ranged from 2.8 to 3.1 mg/L. Columns C1–C4 exhibited average DOC removal by 14 (31) percent, 9 (15) percent, 12 (15) percent, and 25 (14) percent, respectively (Table 2). Standard deviations of BDOC concentrations in all columns were determined by propagation of error, and were for all fractions, with the exception of organic colloids, too high to compare statistically significant carbon removal among the different carbon feed fractions. In previous work, using the same column set-up but higher organic carbon influent concentrations (5 mg/L instead of 3 mg/L), the authors were able to show that the BDOC content in the organic carbon fractions increased relatively in the order of HPO-A, HPI, EfOM, and organic colloids (Rauch and Drewes, 2004).
3.3.
Prevailing redox conditions in the columns
Dissolved oxygen concentrations in the water fed to the anoxic column system were consistently below 0.3 mg/L. Denitrification resulted in a removal of 96 percent nitrate in the anoxic columns from an average influent concentration of 9.3 mg/L NO 3 -N. The pH increased slightly from 7.6 to 7.9 on average after a retention time of 25 days (Fig. 1). The columns C1–C4 were operated under fully aerobic conditions. Dissolved oxygen concentrations in the effluents of all four aerobic columns remained above 3 mg/L. Nitrification was observed in
Fig. 1 – DOC, SUVA, and soil biomass profile in anoxic columns (error bars [ Stdev). (The pH in the anoxic columns ranged between 7.6 and 7.9).
the columns (data not presented), but denitrification did not occur.
3.4.
Soil biomass
In the anoxic column system, soil biomass measured as PLE decreased rapidly with depth (Fig. 1). Due to little bioavailable organic carbon in the HPO-A feed water, column C1 sustained comparatively low amounts of soil microbial biomass (Table 2). However, in comparison to the virgin sand (prior to exposure to any aqueous organic carbon) biomass growth in presence of HPO-A was quantifiable even in such carbon poor (oligotrophic) conditions. Soil biomass in the infiltration zone (top 0–2 cm) of the HPI (C2), EfOM (C3) and organic colloid column (C4) increased in correlation with BDOC concentrations in the respective feed water matrices. This trend is not apparent from the BDOC data collected during this study, but is supported by previous work conducted when the same column systems were operated for a long period of time at slightly higher influent DOC concentrations of 5 mg/L (Rauch and Drewes, 2004). The variability of the DOC analysis in column influent and effluent samples likely resulted in a propagation of error for calculating the BDOC concentration that prevented showing the same trend for the lower feed DOC concentration used in this study.
3.5.
Removal of TOrC under anoxic conditions
The removal of TOrC under anoxic flow conditions was studied at environmentally relevant concentrations (100–900 ng/L) (Table 4). Compounds exhibiting concentrations of less than 100 ng/L in the secondary effluent (i.e. primidone, propyphenazone, diclofenac, and ketoprofen) were artificially amended prior to the experiments, in order to be able to derive meaningful removal efficiencies in the columns. TOrC influent and effluent concentrations for all column studies summarized in Table 4 are presented for synoptic samples taking into account the hydraulic residence time of the respective column systems. During the column studies, the concentrations of TOrC in the anoxic feed water remained stable for at least 4 weeks (with the exception of ibuprofen which exhibited a rapid loss after 8 days, data not shown). The observed removal of TOrC in the anoxic column system is presented in Table 4 and Fig. 2. Naproxen exhibited an almost linear removal during the first 2 m of infiltration (or 12.5 days residence time) and remained at a concentration of 10–25 ng/L during subsequent travel. Similarly, ketoprofen and diclofenac concentrations were reduced within 1–2 m of soil infiltration and subsequently remained relatively stable at concentrations below the limit of quantification for ketoprofen and in the range of 75–135 ng/L for diclofenac. Low and relatively stable TOrC concentrations in the 3rd and 4th anoxic column may indicate that concentration thresholds for certain TOrC exist below which further degradation is inhibited, as previously reported by others (Van der Meer et al., 1987; McCarty et al., 1981; Margesin and Schinner, 2001; Rittmann and McCarty, 2001). In contrast to bulk organic carbon removal (first-order kinetic), the TOrC attenuation (Fig. 2) seem to follow a linear removal with depth indicating a zero-order kinetic. While
Table 4 – Summary of TOrC concentrations observed in column studies. Anoxic Column TOrC (ng/L)
Carbamazepine Diclofenac sodium
Feed water 279–296 (n ¼ 3) 307–416 (n ¼ 2)
1-m Column effluent n.a. 75
2-m Column effluent n.a. 134
HPO-A (C1) 3-m Column effluent n.a. 110
4-m column effluent 318 80–120 (n ¼ 2)
444 141 (n ¼ 5)
323–325 (n ¼ 2)
<25
<25
<25
Ibuprofen
60
40
<10
<10
<10
Ketoprofen
707 111 (n ¼ 6)
570–592 (n ¼ 2)
<10
<10
<10
Naproxen
412 103 (n ¼ 5)
303–350 (n ¼ 2)
21
17–20 (n¼ 2)
17–20 (n ¼ 2)
Phenacetine
n.a.
Primidone Propyphenazone
562–573 (n ¼ 2) 642 107 (n ¼ 7)
TCEP
TCPP
323 68 (n ¼ 5)
805 172 (n ¼ 7)
629 520–630 (n ¼ 2)
352–377 (n ¼ 3)
955–1010 (n ¼ 3)
n.a. n.a.
n.a.
n.a.
n.a. 36–87 (n ¼ 2)
314
n.a.
764 42–112 (n ¼ 2)
378
942–1075 (n ¼ 2)
Column effluent
Column influent
n.a. 604 538 73 373 770 381 626 948 356 772 605 180 889 711 176 534 539 289
538 190 72 26 <25 62 10 84 <10 38 22 37 18 21 38 <50 <50 <50 133 809 556 1104 908 596 1080 1274 557
Column influent
n.a. n.a. 70 23 544 507 302 496 528 499 406 706 935 438 581 721 795 506 508
n.a. 148 809 542 856 966 572 995 1503 958
Column effluent
EfOM (C3)
n.a. n.a. <10 <10 431 307 59 6 33 <10 318 145 <25 232 36 21 n.a. <50 <50
n.a. n.a. 328 578
318 73 770 547 381 638 369 356 640 589 180 722 37 176 607 141 289
710 132
682 139
721
255
599 691
229 17
670 820
34 421
924 817
428 176
846 392
215 <50
222
605 848 556
863 572
n.a. 1087 958
588
605 n.a.
636 1120 596
n.a. <50 121
<50 n.a.
n.a. 868 404
Column effluent
n.a. 294 72 336 108 62 104 102 <10 500 244 37 4 44 38 <50 <50 <50
588 733 542
n.a.
755 973
Column influent
n.a. 698 341
336 651
Column effluent
Org. colloids (C4)
1057 n.a.
712 1720 557
water research 44 (2010) 449–460
Gemfibrozil
Column influent
HPI (C2)
949 n.a.
n.a.: not available.
455
456
water research 44 (2010) 449–460
indicate a linear removal profile for ketoprofen, naproxen, gemfibrozil, and ibuprofen, the removal profile might also have been affected by a change in soil microbial community composition with column depth caused by changing organic carbon substrate composition and concentrations along the flow path. Thus, definite conclusions about the kinetic order are not possible without additional experiments. Due to the low ambient concentrations of trace organic contaminants (occurring in the ng/L range) it is unlikely that energy from TOrC metabolism is sufficient for biomass maintenance and growth. It was hypothesized that at trace levels, TOrC were transformed by co-metabolism (co-utilization). In this context the term co-metabolism describes a transformation of a trace compound, which is by itself unable to support cell replication, and requires the presence of another transformable organic compound (primary carbon source) that allows microorganisms to obtain energy for their metabolism and growth (Brandt et al., 2003; Stratton et al., 1983; Clara et al., 2005). Because the primary substrate (biodegradable bulk organic carbon) was present in the column experiment in concentrations two orders of magnitude above common concentrations of TOrC in the environment, co-metabolism offers a possible explanation for the observed removal of the target TOrC, that were fed at trace concentrations. This hypothesis was further tested by examining the effect of organic carbon subgroups present in EfOM and differing in biodegradability on soil biomass growth and TOrC removal.
Fig. 2 – Removal of TOrC in anoxic column system.
first-order degradation kinetics are indicative of organic carbon degradation at concentrations below the half-velocity coefficient Ks, or under mass transport limitations that keep biodegradation rates below the biological capacity. Zero-order reaction kinetics are indicative of degradation below the biological degradation potential, where either substrate concentrations of trace contaminants are significantly higher than the affinity for the specific substrate, or the limiting substrate is delivered at a constant rate (Simoni et al., 2001; Alexander, 1999). It should be noted that, while the column results
Gemfibrozil
Ketoprofen
Naproxen
ng/L removed in column
900 800 700 600 500 400 300 200 100 0 3.3
10.4
27.2
HPO-A
HPI
Org. Colloids
nmol org-PO43- g-1
Fig. 3 – Relationship between TOrC removal, soil biomass, and organic carbon substrate in columns C1, C2, and C4 (1st spike event).
3.6. Effect of aqueous organic carbon matrices on TOrC removal In order to further study the effect of soil biomass and aqueous organic carbon matrices on the removal of TOrC, aerobic column experiments were conducted using different organic carbon bulk fractions (EfOM, HPO-A, HPI, and organic colloids) isolated from secondary treated effluent. The different organic bulk substrates were fed at similar DOC concentrations to different column systems and were able to support different amounts of viable soil biomass (Table 2). During three consecutive spiking tests, hydraulically corresponding influent and effluent samples were collected from each column (Fig. 3) with the exception of the column fed with organic colloids for which only two spiking tests were conducted. The results of the first spiking test for the HPI and organic colloid columns are presented in Fig. 3. Findings suggest that the more viable soil biomass was present on the column media, the more complete was the removal for gemfibrozil, ketoprofen, and naproxen. In other terms the concentration of biodegradable effluent derived organic carbon substrate limited the transformation of TOrC. (Phenacetine was removed below the quantification limit in all columns and removal could, thus, not be related to biomass activity in the columns.) Organic colloids (BDOC content 0.86 0.48 mg/L) appeared to have promoted the degradation of TOrC by serving as the primary substrate and establishing a relatively higher microbial population that was able to co-metabolize TOrC better as compared to HPI organic carbon. TOrC removal in the HPI column was possibly lower because HPI substrate (BDOC content 0.25 0.41 mg/L) supported less biomass
457
water research 44 (2010) 449–460
1st spike
2nd spike (day 11 after 1st spike)
3rd spike (day 22 after 1st spike) 925
100 90
544
506
700
508
490 495 499
561
243 550
80 removal %
70 60
206
50
200
40 30
87
113
20 10 0 Ketoprofen
Naproxen
Phenacetine
Gemfibrozil
Ibuprofen
Fig. 4 – Microbial adaptation to TOrC in HPI column (values on top of bars represent removal in ng/L).
growth to metabolize TOrCs. During longer exposure times to the contaminants the performance of the HPI column for TOrC removal improved. In the 2nd and 3rd spiking event, ketoprofen, phenacetine, gemfibrozil, and naproxen exhibited increasingly better removal exceeding 95 percent 22 days after the first spike (Fig. 4) suggesting that the microbial community had adapted to metabolizing TOrC. In accordance with the performance of the anoxic column system (Fig. 2), ibuprofen was equally efficient removed (>85%) in the presence of all organic carbon fractions and in all three spiking events (Table 4 and Fig. 4) under both aerobic and anoxic conditions. This confirms previously reported findings that ibuprofen is very well degradable under various field conditions (Sedlak and Pinkston, 2001; Drewes et al., 2003). In contrast to the anoxic column system, propyphenazone was not eliminated in any of the four short columns (Table 4). Consistent with previous studies, propyphenazone was found to require longer retention times (>1 week) to exhibit partial removal under MAR conditions (Drewes et al., 2003; Heberer et al., 2003). The anticonvulsant drugs were not eliminated in any of the columns investigated (Table 4). Based on these study results, effluentderived BDOC does not seem to be a suitable substrate to stimulate co-metabolic decay of the selected anticonvulsant drugs (i.e. primidone, carbamazepine) under environmentally relevant bulk and TOrC concentrations. Whereas TCPP was not removed in the anoxic columns, this flame retardant seems to be removed in the aerobic HPO-A, HPI, and EfOM column after a lag time of two to three weeks after starting to feed this compound to the column, possibly indicating biological acclimatization leading to some degree of attenuation of this compound (Table 4). However, results are inconsistent when comparing influent and effluent concentrations in the different columns, therefore further studies are recommended to solidify this observation. Given that the first column of the anoxic system was operated at a longer residence time (about 6 days) compared to the aerobic columns (less than 1 day), it is noticeable that most degradable TOrC (i.e. gemfibrozil, ibuprofen, ketoprofen, naproxen) were significantly better transformed under aerobic conditions as compared to anoxic recharge
conditions. Diclofenac exhibited consistent removal during anoxic conditions but not during aerobic recharge, which is consistent with findings reported by Zwiener and Frimmel (2003) and Hua et al. (2003).
3.7.
Removal of TOrC in oligotrophic environments
Although the HPO-fraction provided little soil biomass growth 1 (<5 nmol org-PO3 4 g ) in the infiltration zone (Table 2), column C1 exhibited-unexpectedly-a significant and instantaneous transformation of target TOrC after initial exposure during the first spiking event. Phenacetine, ibuprofen, gemfibrozil, naproxen, and ketoprofen were all consistently removed by more than 90 percent in all three spiking events (Fig. 3 and Table 4). In comparison to all other feed water matrices, the HPO-A column exhibited an equal (regarding ibuprofen) or faster removal for all degradable target TOrC. These findings suggest that the type of effluent derived bulk organic carbon effects both, the quantity of soil biomass growth and the composition of microbial community and that both parameters are essential in understanding TOrC transformation during MAR. It is postulated, that HPO-A (humic-like) carbon promoted the development of an oligotrophic soil microbial community well capable of degrading TOrC to low residual concentrations. Given that EfOM usually is a mixture of all three carbon fractions (HPO-A, HPI, and organic colloids), EfOM may have stimulated the growth of a more diverse biomass in column C3 as compared to C1, C2, or C4 that were only fed with a subfraction of the organic substrate. This may explain why TOrC removal was relatively better in the column fed with EfOM, than would be expected based on soil biomass present in C3 (i.e. for gemfibrozil and naproxen) (Table 4). Daughton and Ternes (1999) acknowledged the role of oligotrophic metabolism for the decay of TOrC in the environment. Accordingly, the biotransformation of bulk organic carbon is governed by enzyme-saturating substrate concentrations (copiotrophic metabolism) during soil infiltration. TOrC are present at low concentrations, at below enzyme-saturating levels, which may necessitate oligotrophic metabolism.
458
water research 44 (2010) 449–460
Because oligotrophic organisms are more likely to be prevalent in native environments characterized by low-carbon fluxes than in wastewater treatment plants, the degradation of TOrC may be more efficient during soil percolation than in wastewater treatment plants (Daughton and Ternes, 1999). It is anticipated that oligotrophic organisms reside in MAR systems in soil depths below the immediate infiltration zone (which in contrast is likely dominated by microbial organisms growing primarily on easily degradable EfOM). Further investigations directed at the interrelation between effluent derived organic carbon and the succession of biocommunities in soil based on trophic carbon sequences seem key to the understanding of mechanisms and limitations to TOrC removal in MAR systems.
Acknowledgments
4.
Appendix. Supplementary data
Conclusions
Hydrophilic, small-molecular size TOrC can survive conventional and advanced wastewater treatment processes and are one of the key concerns in drinking water augmentation projects using impaired source waters. More research is needed to better understand the trophodynamic role of organic carbon on the removal of TOrC in MAR applications. In this study we conducted column experiments to illuminate key factors for the metabolic removal of twelve TOrC during soil infiltration. Experiments were designed to resemble natural environments (flow through systems, long biological acclimatization times, complex aqueous carbon compositions, environmentally relevant bulk and TOrC concentrations) in order to reveal interactions between bulk organic carbon and TOrC removal that are relevant for full-scale MAR conditions. Removal efficiencies varied widely among the target compounds but were generally higher during aerobic soil infiltration with the exception of diclofenac that was faster degraded during anoxic recharge. Propyphenazone required a longer residence time for removal than was provided in the aerobic columns employed in this study. The biological reactions for all other degradable TOrC were fast at ambient concentrations (ng/L level) and completed within the first 2 m of porous media infiltration. Threshold concentrations below which further degradation was negligible were observed for acidic TOrC under anoxic conditions in the lower ng/L-range. Intensive research has been conducted on stimulating the degradation of certain pollutants by artificially adding carbon substrates (Horvath, 1973, Semprini, 1997). Findings of this study demonstrated that naturally available organic matter in aqueous environments can promote the degradation of certain TOrC by serving as a secondary carbon substrate for their removal. The composition of effluent-derived bulk organic carbon had a strong effect on the biological removal of the target TOrC and is based on several mechanisms. BDOC, prevalent in form of colloidal and hydrophilic carbon, stimulates soil biomass growth and induces secondary substrate utilization of TOrC. Soil biomass in aerobic systems was able to adapt to most acidic drugs showing an improved removal over time. Soil biomass growth occurred even in response to the infiltration of recalcitrant hydrophobic acids. This organic carbon substrate induced an oligotrophic microbial community that was more effective in removing TOrC than during
copiotrophic metabolism in the presence of higher concentrations of BDOC. Consistent with previous observations, the anticonvulsant drugs carbamazepine and primidone exhibited no removal under any conditions examined in this study.
Partial funding for this study was provided by the Gwangju Institute of Technology in Korea. We are thankful for analytical support during this study provided by Dr. Thomas Heberer at the Technical University in Berlin, Germany, Matt Oedekoven, and Stephan Wagner.
Supplementary data associated with this article can be found in the online version at doi:10.1016/j.watres.2009.08.027.
references
Amy, G., Drewes, J.E., 2006. Soil-aquifer treatment (SAT) as a natural and sustainable wastewater reclamation/reuse technology: fate of wastewater effluent organic matter (EfOM) and trace organic compounds. Environmental Monitoring and Assessment 129 (1–3), 19–26. Alexander, M., 1999. Biodegradation and Bioremediation. Academic Press, San Diego, California. Brandt, B.W., Van Leeuwen, I.M.M., Kooijman, S.A.L.M., 2003. A general model for multiple substrate biodegradation. Application to co-metabolism of structurally non-analogous compounds. Water Research 37, 4843–4854. Brauch, H.-J., Sacher, F., Denecke, E., Tacke, T., 2000. Effectiveness of bank filtration for the removal of polar organic trace contaminants. GWF Wasser Abwasser 141 (4) 2260–234 (in German). Buser, H.R., Poiger, T., Mu¨ller, M.D., 1998. Occurrence and fate of the pharmaceutical drug diclofenac in surface waters: rapid photodegradation in a lake. Environmental Science and Technology 32 (22), 3449–3456. Buser, H.R., Poiger, T., Mu¨ller, M.D., 1999. Occurrence and environmental behavior of the chiral pharmaceutical drug ibuprofen in surface waters and in wastewater. Environmental Science and Technology 33 (15), 2529–2535. Carballa, M., Omil, F., Lema, J.M., Lllompart, M., Jares, C.G., Rodriguez, I., Gomez, M., Ternes, T., 2004. Behaviour of pharmaceuticals, cosmetics and hormones in a sewage treatment plant. Water Research 38, 2918–2926. Clara, M., Strenn, B., Kreuzinger, N., 2004. Carbamazepine as a possible anthropogenic marker in the aquatic environment: investigations on the behaviour of carbamazepine in wastewater treatment and during groundwater infiltration. Water Research 38, 947–954. Clara, M., Kreuzinger, N., Strenn, B., Gans, O., Kroiss, H., 2005. The solids retention time – a suitable design parameter to evaluate the capacity of wastewater treatment plants to remove micropollutants. Water Research 39, 97–106. Daughton, C.G., Ternes, T.A., 1999. Pharmaceuticals and personal care products in the environment: agents of subtle change? Environmental Health Perspectives 107 (Suppl. 6) December 1999.
water research 44 (2010) 449–460
Dillon, P., Page, D., Vanderzalm, J., Pavelic, P., Toze, S., Bekele, E., Sidhu, J., Prommer, H., Higginson, S., Regel, R., RinckPfeiffer, S., Purdie, M., Pitman, C., Wintgens, T., 2008. A critical evaluation of combined engineered and aquifer treatment systems in water recycling. Water Science and Technology 57 (5), 753–762. Drewes, J.E., Fox, P., 1999. Fate of natural organic matter (NOM) during groundwater recharge using reclaimed water. Water Science and Technology 40 (9), 241–248. Drewes, J.E., Shore, L.S., 2001. Concerns about pharmaceuticals in water reuse, groundwater recharge, and animal waste. In: American Chemical Society (Ed.), Pharmaceuticals and personal care products in the environment. Symp. Ser, vol. 791, pp. 206–228. Drewes, J.E., Heberer, T., Reddersen, K., 2002. Fate of pharmaceuticals during indirect potable reuse. Water Science & Technology 46 (3), 73–80. Drewes, E.J., Heberer, T., Rauch, T., Reddersen, K., 2003. Fate of pharmaceuticals during groundwater recharge. Ground Water Monitoring and Remediation 23 (3), 64–72. Drillia, P., Stamatelatou, K., Lyberatos, G., 2003. Mobility of Pharmaceuticals in Soil, 8th Conference on Environmental Science and Technology, Lemnos, Greece. Focazio, M.J., Kolpin, D.W., Barnes, K.K., Furlong, E.T., Meyer, M.T., Zaugg, S.D., Barber, L.B., Thurman, M.E., 2008. A national reconnaissance for pharmaceuticals and other organic wastewater contaminants in the United States–II untreated drinking water sources. Science of the Total Environment 402 (2–3), 201–216. Fries, E., Pu¨ttmann, W., 2003. Monitoring of the three organophosphate esters TBP, TCEP and TBEP in river water and ground water (Oder, Germany). Journal of Environmental Monitoring 5 (2), 346–352. Gru¨nheid, S., Amy, G., Jekel, M., 2005. Removal of bulk dissolved organic carbon (DOC) and trace organic compounds by bank filtration and artificial recharge. Water Research 39, 3219–3228. Heberer, T., Stan, H.-J., 1997. Determination of clofibric acid and N-(phenylsulphonyl)-sarcosine in sewage, river and drinking water. International Journal of Environmental Analytical Chemistry 67, 113–124. Heberer, T., Verstraeten, I.M., Meyer, M.T., Mechlinski, A., Reddersen, K., 2001. Occurrence and fate of pharmaceuticals during bank filtration–preliminary results from investigations in Germany and the United States. Water Resources Update 20, 4–17. Heberer, T., 2002. Occurrence, fate and removal of pharmaceutical residues in the aquatic environment: a review of recent research data. Toxicology Letters 131, 5–17. Heberer, T., Mechlinski, A., Fanck, B., 2003. NASRI – Occurrence and Fate of Pharmaceuticals During Bank Filtration, Conference Wasser Berlin, KompetenzZentrum Wasser Berlin, pp. 41–49. Horvath, R.S., 1973. Enhancement of co-metabolism of chlorobenzoates by the co-substrate enrichment technique. Applied Microbiology 25 (6), 961–963. Hua, J., An, P., Winter, J., Gallert, C., 2003. Elimination of COD, microorganisms and pharmaceuticals from sewage by trickling through sandy soil below leaking sewers. Water Research 37, 4395–4404. 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, 1999–2000: a national reconnaissance. Environmental Science & Technology 36, 1202–1211. Leenheer, J.A., Croue, J.P., Benjamin, M., Korshin, G.V., Hwang, C.., Bruchet, A., Aiken, G.R., 2000. Comprehensive isolation of natural organic matter from water for spectral characterizations and reactivity testing. In: Barett, S.E., Krasner, S.W., Amy, G.L.
459
(Eds.), Natural Organic Matter and Disinfection By-Products, vol. 761. ACS Symposium Series, pp. 68–83. Margesin, R., Schinner, F., 2001. Bioremediation (natural attenuation and biostimulation) of diesel-oil-contaminated soil in an alpine glacier skiing area. Applied and Environmental Microbiology 67 (7), 2127–3133. McCarty, P.L., Reinhard, M., Rittmann, B.E., 1981. Trace organics in groundwater. Environmental Science and Technology 15 (1), 40–51. Mersmann, P., Scheytt, T., Heberer, T., 2002. Column experiments on the transport of pharmaceutically active compounds in the saturated zone. Acta Hydrochimica et Hydrobiologica 30 (5–6), 275–284 (in German). Mo¨hle, E., Kempter, C., Kern, A., Metzger, J.W., 1999. Examination of the degradation of drugs in municipal sewage plants using liquid chromatography-electrospray mass spectrometry. Acta Hydrochimica et Hydrobiologica 27 (6), 430–436 (in German). Montgomery-Brown, J., Drewes, J.E., Fox, P., Reinhard, M., 2003. Behavior of alkylphenol polyethoxlate metabolites during soil aquifer treatment. Water Research 37, 3672–3681. Preuß, G., Nehrkorn, A., 1996. Succession of microbial communities during bank filtration and artificial groundwater recharge Artificial Recharge of Groundwater. Proceedings of an International Symposium, A.-L. Kivima¨ki and T. Suokko, eds., Helsinki, Finland, June 3–5, 1996, Nordic Hydrological Programme, pp. 215–221. Rauch, T., Drewes, J.E., 2004. Assessing the removal potential of soil-aquifer treatment systems for bulk organic matter. Water Science & Technology 50 (2), 245–253. Rauch, T., Drewes, J.E., 2005. A novel approach for quantifying bulk organic carbon removal in groundwater recharge systems. Journal of Environmental Engineering 131 (6), 909–923. Reddersen, K., Heberer, T., 2003. Multi-compound methods for the detection of pharmaceutical residues in various waters applying solid phase extraction (SPE) and gas chromatography with mass spectrometric (GC-MS) detection. Journal of Separation Sciences 26, 1443–1450. Rittmann, B.E., McCarty, P.L., 2001. Environmental Biotechnology. Principles and Applications. McGraw-Hill, Boston. Scherrer, R.A., Howard, S.M., 1977. Use of distribution coefficients in quantitative structure-activity relationships. Journal of Medical Chemistry 20 (1), 53–58. Schmidt, C.K., Lange, F.T., Brauch, H.-J., 2004. Assessing the Impact of Different Redox Conditions and Residence Times on the Fate of Organic Micropollutants during Riverbank Filtration, Fourth International Conference on Pharmaceuticals and Endocrine Disrupting Chemicals in Water, October 13–15, Minneapolis, Minnesota. Schwab, B.W., Hayes, E.P., Fiori, J.M., Mastrocco, F.J., Roden, N.M., Cragin, D., Meyerhoff, R.D., D’aco, V.J., Anderson, P.D., 2005. Human pharmaceuticals in US surface waters: a human health risk assessment. Regulatory Toxicology and Pharmacology 42 (3), 296–312. Sedlak, D.L., Pinkston, K.E., 2001. Factors affecting the concentrations of pharmaceuticals released to the aquatic environment. Journal of Contemporary Water Research and Education 120, 56–64. September. Semprini, L., 1997. Strategies for the aerobic co-metabolism of chlorinated solvents. Current Opinion in Biotechnology 8 (3), 296–308. Simoni, S.F., Scha¨fer, A., Harms, H., Zehnder, A.J.B., 2001. Factors affecting mass transfer limited biodegradation in saturated porous media. Journal of Contaminant Hydrology 50, 99–120. Stratton, R.G., Namkung, E., Rittmann, B.E., 1983. Secondary utilization of trace organic by biofilms on porous media. Journal American Water Works Association 75, 463–469.
460
water research 44 (2010) 449–460
Szewzyk, U., Kalmbach, S., Manz, W., La˚ngmark, J., and Stenstro¨m, T.-A., 1998. Artificial groundwater as the future water supply of greater Stockholm IV: Phylogenetic diversity of the unsaturated zones of experimental sand filter columns. Artificial recharge of groundwater. Proceedings of the third international symposium on artificial recharge of groundwater–Tisar 98, J.H. Peters, ed., Amsterdam, Netherlands, 21–25 September 1998, Rotterdam, pp. 391–393. Toride, N., Leij, F.J., van Genuchten, M.T., 1995. The CXTFIT Code for Estimating Transport Parameters from Laboratory or Field
Tracer Experiments. Version 2.0. U.S. Salinity Laboratory Res. Rep. 137, U.S. Salinity, Riverside, CA. Van der Meer, J.R., Roelofsen, W., Schraa, G., Zehnder, A.J.B., 1987. Degradation of low concentrations of dichlorobezenes and 1,2,4thrichlorobenzene by Pseudomonas sp. strain P51 in nonsterile soil columns. FEMS Microbiology Letters 45 (6), 333–341. Zwiener, C., Frimmel, F.H., 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. Science of the Total Environment 309, 201–211.
water research 44 (2010) 461–476
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Toxicological relevance of emerging contaminants for drinking water quality Merijn Schriks a,*, Minne B. Heringa a, Margaretha M.E. van der Kooi a, Pim de Voogt a,b, Annemarie P. van Wezel a a
KWR Watercycle Research Institute, P.O. Box 1072, 3430 BB Nieuwegein, The Netherlands Institute for Biodiversity and Ecosystem Dynamics, University of Amsterdam, Nieuwe Achtergracht 166, 1018 WV, Amsterdam, The Netherlands
b
article info
abstract
Article history:
The detection of many new compounds in surface water, groundwater and drinking water
Received 17 May 2009
raises considerable public concern, especially when human health based guideline values
Received in revised form
are not available it is questioned if detected concentrations affect human health. In an
14 August 2009
attempt to address this question, we derived provisional drinking water guideline values
Accepted 21 August 2009
for a selection of 50 emerging contaminants relevant for drinking water and the water
Available online 26 August 2009
cycle. For only 10 contaminants, statutory guideline values were available. Provisional drinking water guideline values were based upon toxicological literature data. The
Keywords:
maximum concentration levels reported in surface waters, groundwater and/or drinking
Risk assessment
water were compared to the (provisional) guideline values of the contaminants thus
Drinking water
obtained, and expressed as Benchmark Quotient (BQ) values. We focused on occurrence
(Provisional) drinking water
data in the downstream parts of the Rhine and Meuse river basins. The results show that
guideline value Threshold of Toxicological Concern (TTC)
for the majority of compounds a substantial margin of safety exists between the maximum concentration in surface water, groundwater and/or drinking water and the (provisional) guideline value. The present assessment therefore supports the conclusion that the majority of the compounds evaluated pose individually no appreciable concern to human health. ª 2009 Elsevier Ltd. All rights reserved.
1.
Introduction
Due to anthropogenic activities, freshwater systems worldwide are confronted with thousands of compounds. In the European Union, for example, there are more than 100 000 registered chemicals (EINECS), of which 30 000–70 000 are in daily use. About 300 million tons of synthetic compounds annually used in industrial and consumer products, partially find their way to natural waters (Schwarzenbach et al., 2006). A major contribution to chemical contamination originates from wastewater discharges that impact surface water quality
with incompletely removed organic contaminants (Kolpin et al., 2004; Snyder et al., 2001). Additional contamination comes from diffuse agricultural activities, in which over 140 million tons of fertilizers and several million tons of pesticides are applied each year, and from atmospheric deposition. Such contamination can become an increasing problem for drinking water supplies, especially since the European REACH legislation may drive producers to develop newly designed less lipophilic/bioaccumulative chemicals that will be inherently more difficult to remove by traditional drinking water treatment techniques.
* Corresponding author. Tel.: þ31 30 6069564. E-mail address:
[email protected] (M. Schriks). 0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2009.08.023
462
water research 44 (2010) 461–476
Recently, Loos et al. (2009) presented an EU-wide monitoring study on 35 organic compounds in European river waters in concentrations up to 40 mg/L. In addition, we have shown the occurrence of emerging polar contaminants, such as benzotriazoles and metabolites of illicit drugs (e.g. benzoylecgonine, desalkylflurazepam and 9-carbonic acid-d9-THC) in groundwaters and surface waters in the Netherlands (Hogenboom et al., 2009; Van Leerdam et al., 2009). Many of these emerging contaminants raise considerable toxicological and public concern, especially when human health based guideline values are unavailable. At present, the World Health Organization (WHO) and the US Environmental Protection Agency (US EPA) have derived approximately 125 statutory guideline values for contaminants in drinking water (Cotruvo, 1988; US EPA, 2006; WHO, 2006). However, the potential health effects of many emerging contaminants present in the water cycle and the potential human health concern associated with direct water ingestion have not been evaluated and statutory standards are not available. Therefore, we proposed earlier to assess the potential human health concern of unknown non-genotoxic compounds (lacking structural alerts that raise concern for potential genotoxicity) by comparing the environmental or drinking water concentration to a TTC (Threshold of Toxicological Concern) derived target value (Mons et al., 2008). The TTC concept was developed in the context of food safety to obtain a first clue on risks of unregulated chemicals present at low levels. The TTC thus derived is based on the molecular structure of the chemical involved and its related mode of toxic action (Munro et al., 1996). Assuming a daily intake of 2 l/day of drinking water, and a maximum contribution of 10% from drinking water to the total exposure – both of which are standard assumptions for deriving drinking water-quality guidelines (WHO, 2006) – TTC based target values proposed for drinking water are 0.1 mg/L for non-genotoxic compounds and 0.01 mg/L for genotoxicants (Mons et al., 2008). This value is based on a TTC level of 1.5 mg/person/d (0.15 mg/person/d for substances containing structural alerts that raise concern for potential genotoxicity with an acceptable lifetime cancer risk of 106) for compounds in food (Kroes et al., 2004). Since the TTC based value is rather conservative, an over-estimation of the actual risk may be the result. For emerging contaminants, a more profound human health based assessment may therefore be very valuable. The objectives of the present study were twofold. The first objective was to collect existing drinking water guideline values for a selection of 50 emerging contaminants relevant for the water cycle. If existing guideline values were not available, provisional guideline values were derived with the aid of relevant toxicological literature data. The second aim was to compare the maximum concentration levels reported in surface water, groundwater and/or drinking water to the (provisional) guideline values of the contaminants thus obtained, and express this as a Benchmark Quotient (BQ) value (further abbreviated as ‘‘BQ value’’). The present study does not attempt to quantify mixture interactions, since for compounds with an unknown mode of action there is no accepted methodology for such an assessment.
2.
Materials and methods
The toxicological assessment of the compounds presented in this paper comprises of a tiered approach in five consecutive steps (Fig. 1). First, the compounds to be assessed were selected. Second, n-octanol–water partition coefficients (log Kow) were obtained and compounds with a log Kow > 3 were excluded from further assessment. This log Kow cut off value is applied as a default threshold; for compounds with a log Kow above 3 it is less likely that they pass drinking water treatment plants (Westerhoff et al., 2005). Third, if available, statutory drinking water guideline values were obtained from websites of competent authorities; else provisional guideline values were derived with the aid of toxicological data relevant for humans as reported in literature. Fourth, measured maximum surface water, groundwater and/or drinking water concentrations were obtained from various sources and compared to (provisional) guideline values. Finally, a BQ value was calculated from the maximum concentrations reported and the (provisional) drinking water guideline values obtained. These steps are described in more detail below.
2.1.
Selection of compounds for assessment
A priority list representing a broad range of chemical classes was formulated with more than 100 compounds of interest. The arguments for inclusion were (i) questions related to toxicity posed by Dutch drinking water companies, (ii) potential low removal efficiency during drinking water production, (iii) appearance in recent literature and (iv) occurrence in surface waters, groundwaters and drinking water as determined by ourselves and others in various screening studies.
2.2.
Collection of compound-specific data
2.2.1.
n-Octanol–water partition coefficients (log Kow)
All log Kow values were obtained with the aid of the estimation program KOWWIN (US EPA, v1.67). An exception was made for perfluoroctane sulfonate (PFOS) and perfluoroctanoic acid (PFOA), for which accurate log Kow values cannot be calculated with estimation software. For these compounds the log Kow values were obtained from a database (Krop and de Voogt, 2008).
2.2.2.
Toxicological data
As illustrated in Fig. 1 (step 3), the first step was to obtain existing statutory drinking water guideline values from e.g. the US EPA (URL1) and the WHO (URL2). If not available, the second step was to obtain an established (by an (inter)national organization) Tolerable Daily Intake (TDI), Acceptable Daily Intake (ADI) or Reference Dose (RfD) and subsequently a provisional drinking water guideline value was derived as further described in Section 2.3. If not available, in a third step toxicity data collection focused primarily on established (by an (inter)national organization) lowest/no observed (adverse) effect levels (LO/NO(A)ELs) and subsequently a TDI was calculated as further described in Section 2.3. Finally, in a fourth step, miscellaneous toxicological information was collected and a TDI was calculated accordingly. In the case of insufficient human relevant toxicological data the
water research 44 (2010) 461–476
463
Fig. 1 – Flow diagram of the assessment conducted in the present study. Abbreviations: log Kow, n-octanol–water partition coefficient; GLV, drinking water guideline value; ADI, acceptable daily intake; RfD, reference dose; TDI, tolerable daily intake; LO/NO(A)EL, lowest/no observed (adverse) effect level; numbers correspond to consecutive steps as described Section 2. Compound categories: see Section 2.2.2.
compound of interest was not further evaluated and removed from the list. To facilitate the interpretation for which compounds the toxicity database is strong and less strong, all compounds were categorized (Table 2); (A) representing compounds with a statutory drinking water guideline value, (B) representing compounds with an established TDI, ADI or RfD, (C) representing compounds for which the TDI was calculated with an established LO(A)EL or NO(A)EL and (D) representing compounds for which the TDI was calculated with miscellaneous toxicological information. TDIs, ADIs, RfDs and/or other chronic toxicity data were sourced from peer-reviewed scientific papers and from other sources such as ‘‘grey literature’’. In addition, a literature search was performed on the internet and/or toxicological relevant data were obtained from the US EPA IRIS database, the European Chemicals Bureau (ECB), the Organization for Economic Cooperation and Development (OECD), the Dutch National Institute for Health and the Environment (RIVM), the US Food and Drug Administration (US FDA), the Joint Meeting FAO/WHO Meetings on Pesticide Residues (JMPR), the Dutch Expert Committee for Occupational Standards (DECOS), the Dutch board for authorization of plant protection products and biocides (CTGB), the Scientific Committee on Occupational Exposure Limits (SCOEL), the US National Toxicology Program (NTP), the Joint FAO/WHO Committee on Food Additives (JECFA), the Food and Agriculture Organization (FAO), the Danish veterinary and food administration, the European Union (EU), the US National Research Council (NRC), the EFSA scientific panel on contaminants in the food chain (CONTAM) and the Hazardous Substance Data Bank (HSDB).
2.2.3.
Occurrence data
Collection of occurrence data focused primarily on maximum concentrations of compounds measured in the downstream parts of the Rhine and Meuse river basins during the past decade. If not available, maximum concentrations in other surface waters and/or groundwaters were sought. The primary source of occurrence data of compounds in surface waters were the annual reports of the Dutch Association of River Water Companies (RIWA) and the German Association of River Water Companies (ARW). Occurrence data of compounds in drinking water were obtained from the REWAB data set (restricted water-quality data from the Dutch water companies). If not available, alternative resources were searched such as ‘‘grey literature’’, unpublished data, or publicly available sources such as RIVM, the Dutch ministry of transport, public works and water management (Rijkswaterstaat) and the WHO. Additional data on the occurrence of compounds was obtained from peer-reviewed scientific papers and an internet based literature search.
2.3. Derivation of provisional drinking water guideline values A drinking water guideline value represents the concentration of a constituent that does not exceed tolerable risk to the health of the consumer over a lifetime (WHO, 2006). In some cases, an odour-threshold value may be much lower than the health based guideline value. To calculate a provisional health based guideline value, the general methodology was applied
464
water research 44 (2010) 461–476
Table 1 – List of compounds assessed in the present study. No.
Compound 2
CAS
log Kow
No.
Compound
1071-83-6 138261-41-3
4.0a 0.6a
Iohexol1
66108-95-0
3.1a
30
Iomeprol (iomeron)1
78649-41-9
1.4b
1.4b 2.3a 2.1a 2.0a 1.4a 2.5a 2.5a 1.5a
31 32 33 34 35 36 37 38
62883-00-5 73334-07-3 34123-59-6 1634-04-4 37350-58-6 3622-84-2 111991-09-4 62-75-9
2.4a 2.1a 2.9a 0.9a 1.9a 2.3b 1.2b 0.6a
1698-60-8 882-09-7
1.1a 2.6a
39 40
94-75-7
2.8a
41
84-66-2 134-62-3 109-89-7 111-96-6
2.4a 2.2a 0.6a 0.4a
42 43 44 45
Iopamidol1 Iopromide1 Isoproturon3 Methyl tert-butyl ether (MTBE)4 Metoprolol6 n-Butylbenzenesulphonamide2 Nicosulfuron3 n-Nitrosodimethylamine (NDMA)2 p,p0 -Sulfonyldiphenol2 Perfluoroctanesulfonate (PFOS) (potassium-salt)5 Perfluorooctanoic acid (PFOA)5 Phenazone6 Simazine3 Sulfamethoxazole6 Tolyltriazole2
67-43-6
4.2b
46
87674-68-8
2.2a
47
Dimethylamine (DMA)2
124-40-3
0.4a
48
24
Diuron3
330-54-1
2.7a
49
25
Ethyl tert-butyl ether (ETBE)4
637-92-3
1.9b
50
2 3
6 7 8 9 10 11 12 13
2,6-Dichlorobenzamide (BAM)3 4-Methylbenzenesulfonamide (p-toluenesulfonamide, 4-tolylsulfonamide)2 Acetylsalicylate (aspirin, acetyl salicylic acid)6 Alpha-amino-3-hydroxy5-methyl-4-isoxazole propionic acid (AMPA)3 Amidotrizoic acid (diatrizoic acid)1 Bentazone3 Benzene2 Benzothiazole2 Benzotriazole (1H-benzotriazole)2 Bis(2-chloroisopropyl)ether (BCIPE)2 Carbamazepine6 Carbendazim3
14 15
Chloridazon (pyrazon)3 Clofibric acid6
16
Dichlorophenoxyacetic acid (2,4-D)3
17 18 19 20
22
Diethyl phthalate2 Diethyl toluamide (DEET)3 Diethylamine (DEA)2 Diethylene glycol dimethyl ether (diglyme, bis(2-methoxy ethyl)ester))2 Diethylene triamine penta acetic acid (DTPA)2 Dimethenamid3
23
4 5
21
2008-58-4 70-55-3
0.8a 0.9b
27 28
Ethylenediamine tetra acetic acid (EDTA)2 Glyphosate3 Imidacloprid3
50-78-2
1.2a
29
77521-29-0
2.2b
117-96-4 25057-89-0 71-43-2 95-16-9 95-14-7 108-60-1 298-46-4 10605-21-7
Log Kow 3.9b
1,4-Dioxane
26
CAS 60-00-4
1
a
123-91-1
0.3
80-09-1 2795-39-3
1.7b 1.08c
335-67-1
2.8c
60-80-0 122-34-9 723-46-6 29385-43-1
0.4a 2.2a 0.9a NA
Trichloroethene2
79-01-6
2.4a
Triethylphosphate (ethylphosphate) (TEP)2 Triphenylphosphine oxide (TPPO)2 Tris(2-chloroethyl) phosphate (TCEP)2 Urotropine (methenamine, hexamine)2
78-40-0
0.8a
791-28-6
2.8a
115-96-8
1.4a
100-97-0
4.2b
NA not available. Chemical categories: 1Iodinated contrast media; 2Miscellaneous organic compounds; 3Miscellaneous pesticides; 4Oxygenated gasoline additives; 5Perfluorinated organic compounds; 6Pharmaceuticals. a log Kow values were derived from US EPA’s KOWWIN experimental database. b log Kow values were calculated with the aid of US EPA’s KOWWIN. c log Kow values were obtained from Krop and de Voogt (2008).
as described by Van Leeuwen (2000) and the WHO (2006). For compounds without a statutory drinking water guideline value, first the Tolerable Daily Intake (TDI) was determined. The point of departure (POD) for calculating the TDI was mostly a chronic LO(A)EL, NO(A)EL, benchmark dose level, maximum tolerated dose (MTD) or lowest effective safe dose. In case only inhalatory toxicity data could be found, a routeto-route extrapolation was carried out according to toxicological methods as described by Stokinger and Woodward (1958). An appropriate safety factor to extrapolate between species (inter-species differences), inter-individual differences (intraspecies differences), exposure route/duration and quality of the data was utilized as part of the TDI calculation (Van Leeuwen and Vermeire, 2007). Secondly, a drinking water equivalent level (DWEL) was calculated by multiplying the TDI
by a typical average body weight of 70 kg and division by a daily water consumption of 2 l. Finally, to account for the fraction of the TDI allocated to drinking water, the DWEL was multiplied by an allocation factor to give the provisional guideline value. In most cases, when there was insufficient exposure information to derive chemical-specific allocation factors, a default allocation factor of 10% was used.
2.4. Evaluation of water-quality data in the context of human health Of each compound the maximum concentration level reported in surface waters, groundwater and/or drinking water was compared to its (provisional) guideline value and was expressed as a BQ value (concentration in water divided by
Table 2 – Parameters used for derivation of (provisional) drinking water guideline values.
1,4-Dioxane
POD is an oral slope factor as derived by US EPA (1988a) of 0.011 per mg/kg bw/d from a 110-week study in rats (NCI, 1978). POD is a NOAEL of 4.5 mg/kg bw/d for decreased body weight in both sexes and increased liver weight in males as derived by the Danish EPA (2004) from a study in which dogs were fed BAM in the diet for 2 years (Wilson and Thorpe, 1971), with an uncertainty factor of 300 (100 for inter- and intraspecies variation and 3 for uncertainties in the dataset). POD is a NOAEL of 300 mg/kg bw/d for reproductive effects (decrease lactation index and litter weight at birth) as derived from a GLP compliant study in which rats were administered 4methylbenzenesulfonamide via oral gavage for 42 days (OECD/SIDS, 1994), with an uncertainty factor of 400 (100 for inter- and intraspecies variation and 4 for extrapolation to chronic exposure). POD is a LOEL of 10 mg/person from a human study as derived by the Dutch National Institute for Health and the Environment (RIVM) (Versteegh et al., 2007). POD is a NOAEL of 32 mg/kg bw/d as derived by the WHO (2005a) from a 26-month toxicity study in rats (study reference unknown). POD is the highest therapeutic dose of 50 mg/person/d (0.71 mg/kg bw/d) as derived by the Dutch National Institute for Health and the Environment (RIVM) (Versteegh et al., 2007). POD is a NOAEL of 9 mg/kg bw/d as derived by the WHO (1998) from a 2-year dietary toxicity study in rats (study reference unknown). POD is a risk estimate as derived by the WHO (2003a) from a 2-year gavage study in rats and mice (NTP, 1986). POD is a NOEL of 5.1 mg/kg bw/d as derived by the WHO/JECFA (2003) from a study in which rats were administered benzothiazole in the diet for 90 days (Morgareidge, 1971). Daily observations revealed no treatment related effects in histopathological/haematological parameters, body weight, food consumption and liver/kidney weights. An uncertainty factor of 200 was applied (100 for inter- and intraspecies variation and 2 for extrapolation to chronic exposure). POD is a LOAEL of 295 mg/kg bw/d for histological changes in the liver, decreased body weight gain and inflammation of the prostate/uterus as derived by the Dutch Expert Committee for Occupational Standards (DECOS, 2000) from a study in which rats were administered benzotriazole in the diet for 78 weeks (BIBRA Toxicology International, 1995), with an uncertainty factor of 1000 (100 for inter- and intraspecies variation and 10 for extrapolation of a LOAEL to a NOAEL). POD is a NOAEL of 35.8 mg/kg bw/d as derived by the US EPA (1989) from a 24-month chronic toxicity study in mice (Mitsumori et al., 1979). POD is a maximum tolerated dose (MTD) of 250 mg/kg bw/d as derived by Snyder et al. (2008) from a 2year study in rats showing evidence of carcinogenicity (Singh et al., 2005). POD is a NOAEL of 2.5 mg/kg bw/d as derived by the WHO/JECFA (1995) from a 2-year study in dogs (Sherman, 1972). POD is a NOAEL of 5.4 mg/kg bw/d for adverse histopathological/haematological changes, decreased food intake, lower body weight gain and higher organ weights (liver, kidney, thyroid gland) as derived from a study in which rats were orally administered chloridazon (method of administration unspecified) for 7 weeks (ECB, 2000b), with an uncertainty factor of 100 (inter- and intraspecies variation). POD is a LOEL of 1 mg/kg bw/d as derived by the Dutch National Institute for Health and the Environment (RIVM) (Versteegh et al., 2007) from an 8-week oral study in humans (Larsen et al., 1994).
2,6-Dichlorobenzamide (BAM)
4-Methylbenzenesulfonamide (p-toluenesulfonamide, 4-tolylsulfonamide) Acetylsalicylate (aspirin, acetyl salicylic acid) Alpha-amino-3-hydroxy-5-methyl4-isoxazole propionic acid (AMPA) Amidotrizoic acid (diatrizoic acid) Bentazone Benzene Benzothiazole
Benzotriazole (1H-benzotriazole)
Bis(chloroisopropyl)ether (BCIPE) Carbamazepine Carbendazim Chloridazon (pyrazon)
Clofibric acid
Categorya SUFb TDI, ADI or RfD (Provisional) (mg/kg bw/d) guideline value (mg/L) 30d
B
NA
NA
C
300
0.015
D
400
0.75
2600
B
20
0.007
25
A
100
0.3
900c
B
10
NA
250 000
A
100
0.1
300c
A
NA
NA
10c,d
C
200
0.026
90
C
1000
0.295
1000
B
1000
0.04
140
B
NA
0.00034
B
100
0.03
105
D
100
0.054
189
B
100
0.01
30
52.5
1
(continued on next page)
465
Point of departure (POD)
water research 44 (2010) 461–476
Compound
Compound
Dichlorophenoxyacetic acid (2,4-D) Diethyl phthalate Diethyl toluamide (DEET)
Diethylamine (DEA)
Diethylene triamine penta acetic acid (DTPA)
Dimethenamid Dimethylamine (DMA)
Diuron Ethyl tert-butyl ether (ETBE)
Ethylenediamine tetra acetic Acid (EDTA) Glyphosate Imidacloprid Iohexol
POD is a NOAEL of 1 mg/kg bw/d as derived by the WHO (2003b) from a 1-year study of toxicity in dogs and a 2-year study of toxicity and carcinogenicity in rats (study reference unknown). POD is a NOAEL of 750 mg/kg bw/d as derived by US EPA (1988b) from a 16-week toxicity study in rats (Brown et al., 1978). POD is a NOEL of 100 mg/kg/bw/d based on clinical signs, reduced haemoglobin/haematocrit levels and histological changes in liver, lymph nodes and uterus as derived by the California Environmental Protection Agency (California EPA, 2000) from a study in which beagle dogs were orally administered DEET (gelatin capsules) for 1 year (Goldenthal, 1994), with an uncertainty factor of 56 (14 for inter- and intraspecies variation and 4 for extrapolation to chronic exposure). POD is a LOAEL of 75 mg/m3 for reduced mean body weights and adverse histopathological effects (lesions of the nasal mucosa) as derived by the scientific committee on occupational exposure limits (SCOEL, 2002) from a study in which rats were exposed to DEA via the inhalatory route for 24 weeks (6.5 h/d, 5d/wk) (Lynch et al., 1986), with an uncertainty factor of 50 (5 for the absence of human data and a NOAEL and 10 for route-to-route extrapolation uncertainties). POD is a NOAEL 25 mg/kg bw/d for developmental effects (adversely affected implants per liter and decreased weight gain) as derived by the WHO (2002) from a study in which rabbits were administered diglyme via oral gavage for 13 days (NTP, 1987), with an uncertainty factor of 500 (100 for inter- and intraspecies variation and 5 for uncertainties in the dataset). POD is a NOAEL of 100 mg/kg bw/d for developmental effects (increased fetal deformations) as derived from a study according to OECD guideline 414 in which rats were administered DTPA (in its sodium form) via oral gavage during day 6–15 of pregnancy (ECB, 2000c), with an uncertainty factor of 1000 (100 for inter- and intraspecies variation and 10 for extrapolation to chronic exposure). POD is a NOAEL of 7 mg/kg bw/d as derived by the WHO/JMPR (2005) from a 24-month study in rats given diets containing racemic dimethenamid (study reference unknown). POD is a LOAEL of 19 mg/m3 for concentration-related lesions in the respiratory/olfactory mucosa as derived by the scientific committee on occupational exposure limits (SCOEL, 1991) from a study in which rats and mice were exposed to DMA via the inhalatory route for 2 years (6 h/d, 5d/wk) (CIIT, 1990), with an uncertainty factor of 50 (5 for the absence of human data and a NOAEL and 10 for routeto-route extrapolation uncertainties). POD is a NOEL of 0.625 mg/kg bw/d as derived by US EPA (1988c) from a 2-year feeding study in dogs (DuPont, 1964). POD is a NOAEL of 500 ppm (29.1 mg/kg bw/dh) for testes degeneration as derived from a study in which rats were exposed to ETBE via the inhalatory route for 13 weeks (Medinsky et al., 1999), with an uncertainty factor of 200 (100 for inter- and intraspecies variation and 2 for extrapolation to chronic exposure). POD is a NOAEL of 250 mg/kg bw/d (190 mg/kg bw/d as the free acid) as derived by the WHO/JECFA (1973) from a 2-year toxicity study in rats (study reference unknown). POD is a NOAEL of 32 mg/kg bw/d as derived by the WHO (2005a) from a 26-month study of toxicity in rats fed technical-grade glyphosate (study reference unknown). POD is a NOAEL of 5.7 mg/kg bw/d as derived by the WHO/JMPR (2001) from a 2-year study of toxicity and carcinogenicity in rats (study reference unknown). POD is a safe dose of 75 g/person/d (1.07 g/kg bw/d) as derived by the Dutch National Institute for Health and the Environment (RIVM) (Versteegh et al., 2007).
Categorya SUFb TDI, ADI or RfD (Provisional) (mg/kg bw/d) guideline value (mg/L) A
100
0.01
30c
B
1000
0.8
2800
C
56
1.8
6250
C
50
2.14f
750
C
500
0.05
175
D
1000
0.1
350
B
100
0.07
245
C
50
0.54g
190
B
300
0.002
7
D
200
0.15
525e
A
100
1.9
600b,c
A
100
0.3
900c
B
100
0.06
210
B
10
NA
375 000
water research 44 (2010) 461–476
Diethylene glycol dimethyl ether (diglyme, bis(2-methoxy ethyl)ester))
Point of departure (POD)
466
Table 2 (continued)
Iomeprol (iomeron)
Iopamidol Iopromide Isoproturon Methyl tert-butyl ether (MTBE)
Metoprolol n-Butylbenzenesulphonamide
p,p0 -Sulfonyldiphenol
Perfluoroctane sulfonate (PFOS) Perfluorooctanoic acid (PFOA)
Phenazone Simazine Sulfamethoxazole Tolyltriazole
D
2100
1.9
6700
B
10
NA
415 000
B
10
NA
250 000
A
1000
0.003
B
1000
0.3
B
100
0.014
50
D
600
0.083
292
B A
1000 NA
0.2 NA
D
600
0.017
B
200
0.00015
0.5
B
200
0.0015
5.3
B
100
0.036
125
A
1000
0.52
2c
B
200
0.13
440
D
600
0.25
875
9c 9400e
700 0.1c,d
60
467
(continued on next page)
water research 44 (2010) 461–476
Nicosulfuron n-Nitrosodimethylamine (NDMA)
POD is a NOEL of 2 g Iodine/kg bw/d (equal to 4 g iomeprol/kg bw/d) for adverse effects on liver and kidney (non-lipid cytoplasmic vacuolization of hepatocytes and renal tubular epithelium cells) as derived from a study in which dogs were intravenously exposed to iomeprol for 28 days (Morisetti et al., 1994), with an uncertainty factor of 2100 (35 for inter- and intraspecies, 10 for route-to-route extrapolation and 6 for extrapolation to chronic exposure). POD is a safe dose of 83 g/person/d (1.19 g/kg bw/d) as derived by the Dutch National Institute for Health and the Environment (RIVM) (Versteegh et al., 2007). POD is a safe dose of 50 g/person/d (0.71 g/kg bw/d) as derived by the Dutch National Institute for Health and the Environment (RIVM) (Versteegh et al., 2007). POD is a NOAEL of 3 mg/kg bw/d as derived by the WHO (2003c) from a 90-day study in dogs and a 2-year feeding study in rats (study reference unknown). POD is a NOAEL of 300 mg/kg bw/d as derived by the Dutch National Institute for Health and the Environment (RIVM) (Swartjes et al., 2004) from a 90-day oral toxicity study in rats (Robinson et al., 1990). POD is a LOEL of 100 mg/person/d (1.42 mg/kg bw/d) as derived by the Dutch National Institute for Health and the Environment (RIVM) (Versteegh et al., 2007). POD is a NOAEL of 50 mg/kg bw/d for adverse treatment related macro- or microscopic effects (liver enlargement, hepatocyte hypertrophy, thymic atrophy and lymphocytolysis) as derived from a study according to OECD guideline 407 (GLP compliant) in which rats were administered nbutylbenzenesulphonamide via oral gavage for 28 days (Proviron Fine Chemicals, 2003), with an uncertainty factor of 600 (100 for inter- and intraspecies differences, 6 for extrapolation to chronic exposure). POD is a NOAEL of 199 mg/kg bw/d as derived from a 2-year study in rats (URL3). POD is a tumor dose (TD05, dose level that causes 5% in increase in tumour incidence over the background) of 18 mg/kg bw/d as derived by the WHO (2008) from a detailed 2-year cancer dose– response study in rats (Peto et al., 1991a,b). POD is a NOEL of 10 mg/kg bw/d for histopathological effects (hyperplasia of the mucosal epithelium, hypertrophy of hepatocytes), decreased food consumption/body weight gain and increased liver weights as derived from a study according to OECD guideline 421 in which male and female rats were administered p,p0 -sulfonyldiphenol via oral gavage for respectively 45 days and from 14 days before mating to day 3 of lactation (Mitsubishi Chemical Safety Institute Ltd, date of study unknown), with an uncertainty factor of 600 (100 for inter- and intraspecies differences, 6 for extrapolation to chronic exposure). POD is a NOAEL of 0.03 mg/kg bw/d as derived by the Scientific Panel on Contaminants in the Food Chain (CONTAM) (EFSA, 2008) from a 182-day study in Cynomolgus monkeys (Seacat et al., 2002). POD is a benchmark dose level (BMDL10) of 0.3 mg/kg bw/d as derived by the Scientific Panel on Contaminants in the Food Chain (CONTAM) (EFSA, 2008) from a number of studies in mice and male rats (study references unknown). POD is a LOEL of 250 mg/person/day (3.57 mg/kg bw/d) as derived by the Dutch National Institute for Health and the Environment (RIVM) (Versteegh et al., 2007). POD is a NOAEL of 0.52 mg/kg bw/d as derived by the WHO (1996) from a 2-year combined chronic toxicity/oncogenicity study in rats (Ciba-Geigy, 1988; unpublished study submitted to WHO). POD is a NOAEL of 25 mg/kg/d as derived by Schwab et al. (2005) from a 60-week study in rats (Swarm et al., 1973). POD is a NOAEL of 150 mg/kg bw/d for observed mild apathy as derived from a study in which rats were administered tolyltriazole via oral gavage for 29 days (Benzotriazoles Coalition, 2001; ECB, 2000a), with an uncertainty factor of 600 (100 for inter- and intraspecies variation and 6 for extrapolation to chronic exposure).
Compound
Trichloroethene Triethylphosphate (ethylphosphate) (TEP)
Triphenylphosphine oxide (TPPO)
Tris(2-chloroethyl)phosphate (TCEP)
POD is a benchmark dose level (BMDL10) of 0.146 mg/kg bw/d as derived by the WHO (2005b)/Health Canada (2003) from a developmental toxicity study in rats (Dawson et al., 1993). POD is a NOEL of 335 mg/kg bw/d for fertility effects (effects on litter size) as derived from a study in which rats were administered TEP via the food for a unknown period (OECD/SIDS, 1998), with an uncertainty factor of 600 (100 for inter- and intraspecies variation and 6 for extrapolation to chronic exposure). POD is a NOAEL of 8 mg/kg bw/d for salivation, vomiting, diarrhea, histopathological (liver damage and skeletal muscle atrophy)/haematological (elevated GPT, GOT and alkaline phosphatase activities, reduced haemoglobin/haematocrit levels) parameters as derived from a study in which dogs were administered TPPO via the food for 3 months (ECB, 2000d), with an uncertainty factor of 1000 (100 for inter- and intraspecies variation and 10 for extrapolation to chronic exposure). POD is a NOAEL of 22 mg/kg bw/d for increased relative liver and kidney weights as derived from a study in which rats were administered TCEP via oral gavage for 16 weeks (NTP, 1991), with an uncertainty factor of 1000 (10 for inter- and intraspecies variation and 10 for extrapolation to chronic exposure and uncertainty in genotoxic potential). POD is a NOAEL of 15 mg/kg bw/d as derived by the WHO/JECFA (1974) from a teratogenicity study (exposure from the fourth to fifty-sixth day after mating) in dogs (Hurni and Ohder, 1973).
Categorya SUFb TDI, ADI or RfD (Provisional) (mg/kg bw/d) guideline value (mg/L) 20c
A
100
0.0015
D
600
0.56
1950
D
1000
0.008
28
D
1000
0.022
77
B
100
0.15
500
NA not available. a Categories: A) statutory drinking water guideline value available; B) established TDI, ADI or RfD available; C) TDI calculated with a established LO(A)EL or NO(A)EL; D) TDI calculated with miscellaneous toxicological information. b Uncertainty factors. c WHO drinking water guidelines (WHO, 2006). d Based on a specific cancer risk level of 105. e The odour-threshold value for drinking water preparation is 15 mg/L for MTBE and w1 mg/L for ETBE (Swartjes et al., 2004; Van Wezel et al., 2009). f Using the Stokinger–Woodward approach (Stokinger and Woodward, 1958), a TDI of 150 mg/person/d (2.14 mg/kg bw/d) can be calculated from the 8-h threshold limit value (15 mg/m3) assuming 100% oral/inhalatory absorption and a 8-h total workshift ventilation of 10 m3. g Using the Stokinger–Woodward approach (Stokinger and Woodward, 1958), a TDI of 40 mg/kg/person/d (0.54 mg/kg bw/d) can be calculated from the 8-hour threshold limit value (3.8 mg/m3) assuming 100% oral/inhalatory absorption and a 8-h total workshift ventilation of 10 m3. h A TDI of 0.15 mg/kg bw/day can be calculated from the NOAEL (500 ppm ¼ 29.1 mg/kg/bw/d) assuming 100% oral/inhalatory absorption, a specific lung retention of 25% and a minute ventilation volume for rat of 45 mL.
water research 44 (2010) 461–476
Urotropine (methenamine, hexamine)
Point of departure (POD)
468
Table 2 (continued)
water research 44 (2010) 461–476
guideline value) in order (i) to provide perspective on what the occurrence of emerging contaminants might signify to human health and (ii) to help prioritize further investigations. A BQ value of 1 represents a (drinking) water concentration equal to the (provisional) guideline value. Compounds with a BQ value of 1 in drinking water may be of potential human health concern if the water were to be consumed over a lifetime period. Compounds with a BQ value 0.1 in drinking water were identified as those that may warrant further investigation; this is consistent with various US State and Federal practices (Toccalino, 2007). For compounds found in surface waters and groundwater the BQ value threshold to carry out an additional assessment was set at an arbitrary value of 0.2, since these source waters are purified in drinking water treatment plants which provides extra safety. Compounds in surface waters/groundwater or drinking water with a BQ value of 0.2 or 0.1 respectively, are presumed to present no appreciable concern to human health.
3.
Results
3.1.
Selection of compounds
For only 50 compounds out of the original list, statutory guideline values or useful toxicity and occurrence data could be found. These compounds constitute the final list, which includes compounds from various groups such as iodinated contrast media, pharmaceuticals, oxygenated gasoline additives, perfluorinated organic compounds, miscellaneous organic compounds and pesticides (Table 1). Natural and synthetic steroid hormones such as 17b-estradiol, 17a-ethynylestradiol and estrone were not included in this assessment, as they are removed relatively easily in drinking water purification processes (Nghiem et al., 2004).
3.2.
(Provisional) drinking water guideline values
For 10 compounds WHO statutory drinking water guideline values were available and these compounds were classified as category A. For the remaining 40 compounds a provisional guideline value was established with the aid of toxicological data. An established TDI, ADI or RfD was available for 22 compounds (category B). In 7 cases when there was no TDI, ADI or RfD available, an established NO(A)EL or LO(A)EL was used to calculate a TDI and subsequently a provisional drinking water guideline value (category C). For the remaining 11 compounds, miscellaneous toxicological data was used to calculate a TDI and subsequently a provisional drinking water guideline value (category D). As tabulated in Table 2, (provisional) guideline values ranged from 0.0001 mg/L for NDMA to 415 mg/L for the iodinated contrast medium iopamidol. All iodinated contrast media had relatively high provisional guideline values, ranging from 6.7 mg/ L (iomeprol) to 415 mg/L (iopamidol). In the cases of MTBE and ETBE, the human health based guideline values were at least one order of magnitude higher than the corresponding odour-threshold based guideline values of respectively 15 mg/ L and w1 mg/L (Swartjes et al., 2004; Van Wezel et al., 2009). For two compounds (DMA and DEA) the provisional guideline
469
values were established by route-to-route extrapolation of inhalatory LOAELs to oral LOAELs. Since benzene was evaluated to be genotoxic/carcinogenic (IARC group I) and NDMA (IARC group 2A) and 1,4-dioxane (IARC group 2B) are respectively suspected non-genotoxic and genotoxic carcinogens, their corresponding (provisional) guideline values are provided as an upper bound lifetime cancer risk to an individual of 105, or the odds that one case of cancer would result for every 100 000 persons subjected to continuous exposure over a 70-year lifetime.
3.3. Concentration of compounds in surface waters, groundwaters and drinking water The maximum concentrations of compounds reported in surface waters and/or groundwaters are summarized in Table 3. Measured maximum surface water concentrations were available for 37 of the 50 compounds in the annual reports of RIWA and ARW. For two compounds (MTBE and clofibric acid) maximum concentrations in Dutch groundwater are reported. For the remaining compounds, the maximum concentration reported in surface waters was taken from other sources (see Section 2.2.3). The six compounds with the highest reported maximum concentrations in surface waters were EDTA (29 mg/L), DTPA (12.2 mg/L), p,p0 -sulfonyldiphenol (10 mg/L), urotropine (10 mg/ L), 1,4-dioxane (10 mg/L) and AMPA (5 mg/L), whereas in groundwater a relatively high concentration was found for MTBE (27.3 mg/L) showing the environmental relevance of this compound. The highest maximum concentration of iodinated contrast media in surface waters was reported for iomeprol (0.97 mg/L). Table 3 also summarizes the maximum concentrations of compounds reported in drinking water. Data on the occurrence of compounds in drinking water were relatively scarce, and limited to 35 compounds. For 18 compounds, drinking water concentrations were obtained from the Dutch REWAB database and for 17 compounds drinking water concentrations were taken from reports by others. Drinking water concentrations for the remaining compounds could not be found. The highest maximum concentration reported was for EDTA (13.6 mg/L), followed by DTPA (9 mg/L), metoprolol (2.1 mg/L) and BCIPE (1.9 mg/L).
3.4. Comparison of compound concentrations to (provisional) guideline values (BQ value) For all compounds found in surface waters, groundwaters and drinking water the calculated BQ value was <1 (Table 3). The three compounds exhibiting the highest BQ values (i.e. posing the highest potential human health concern) in surface water are 1,4-dioxane, carbamazepine and PFOS (Fig. 2A). For MBTE and ETBE BQ values of respectively 1.8 and 1.2 can be calculated, when comparing their maximum concentrations reported in surface water to the odourthreshold based guideline values of 15 mg/L and w1 mg/L (not shown in Fig. 2). These BQ values do not indicate a concern for human health per se, but rather indicate that the maximum environmental concentration reported exceeds the odour-threshold. This implies that for MTBE and ETBE
Compound
Surface waters and groundwaters Max conc (mg/L) (number of measurements, year) 10 (NA, 1997) 0.05 (40, 2002-2006) 0.06 (20, 2005)
Ref
Drinking water BQ valuea
Max conc (mg/L) (number of measurements, year)
Source
Ref
BQ valuea
SW, NL SW, NL SW, NL
(4) (11) (13)
0.3 0.001 0.00002
0.5 (NA) 0.23 (14, 2002) NA
UDW, NL FDW, USA
(4) (10)
0.02 0.004
0.065 (NA, 2007)
SW, NL
(15)
0.003
0.12 (12, 2007)
FDW, NL
(15)
0.005
5 (499, 2005)
SW, NL
(11)
0.006
1.1 (6, 2001)
FDW, NL
(10)
0.001
SW, GER SW, NL SW, NL SW, NL SW, NL GW, NL SW, NL SW, BE SW, BE BFGW, NL SW, NL SW, NL SW, NL SW, NL SW, NL
(2) (11) (11) (6) (11) (6) (11) (11) (11) (15) (11) (11) (11) (11) (11)
0.000003 0.0003 0.07 0.0003 0.0005 0.02 0.2 0.01 0.002 0.003 0.007 0.0003 0.00001 0.0004 0.02
0.25 (6, 2006) 0.28 (11, 2006) 0.96 (12, 2005) 0.01 (10, 2007) 0.2 (10, 2007) 1.9 (9, 1982-1984) 0.03 (2, 2007) NA NA 0.14 (2, 2007) 0.11 (5, 2002) NA 0.03 (1, 2005) NA 0.15 (13, 2007)
FDW, NL FDW, NL TW, NL FDW, NL FDW, NL FDW, NL FDW, NL
(10) (10) (10) (14) (14) (6) (15)
0.000001 0.0009 0.1 0.0001 0.0002 0.01 0.03
FDW, NL TW, NL
(15) (10)
0.005 0.004
TW, NL
(10)
0.000005
UDW, NL
(1)
0.0009
SW, NL SW, NL SW, NL SW, BE SW, GER SW, NL SW, NL SW, NL SW, NL SW, NL SW, NL SW, NL SW, NL GW, NL SW, NL SW, NL SW, NL SW, NL
(11) (9) (11) (11) (2) (11) (11) (11) (11) (11) (11) (11) (11) (7) (11) (6) (11) (11)
0.03 0.0005 0.002 0.1 0.002 (1.2b) 0.05 0.001 0.0003 0.000001 0.0001 0.000002 0.000002 0.03 0.003 (1.8b) 0.004 0.003 0.0002 0.07
9 (2, 2001) NA NA 0.08 (2, 2005) NA 13.6 (7, 2001) 0.46 (3, 2006) NA 0.06 (1, 2007) 0.01 (1, 2006) 0.1 (6, 2006) 0.04 (2, 2007) 0.02 (1, 2004) 1.25 (27, 2006) 2.1 (2, 2005) 0.05 (2, 2004) NA 0.002 (21, 2007)
FDW, NL
(10)
0.03
TW, NL
(10)
0.01
FDW, NL TW, NL
(10) (10)
0.02 0.0005
FDW, NL FDW, NL FDW, NL FDW, NL TW, NL FDW, NL FDW, NL TW, NL
(15) (10) (10) (15) (10) (10) (10) (13)
0.0000002 0.000001 0.0000002 0.0000002 0.002 0.0001 (0.08b) 0.04 0.0002
UDW, NL
(5)
0.02
0.63 0.1 0.74 0.03 0.54 2.9 0.227 1.5 0.3 0.091 0.2 0.9 0.06 0.29 3.64
(189, 2006) (126, 2007) (116, 2001) (3, 2008) (11, 2007) (15, 1984-1985) (263, 2003) (111, 2006) (68, 2002) (NA, 2007) (34, 2006) (8, 2005) (36, 2005) (38, 2007) (11, 2007)
12.2 (53, 2005) 0.12 (4, 2005) 0.34 (42, 2005) 0.68 (386, 2002) 1.2 (97, 2006) 29 (192, 2005) 1.2 (291, 2006) 0.06 (1, 2007) 0.5 (180, 2005) 0.97 (172, 2007) 0.714 (188, 2006) 0.56 (186, 2004) 0.31 (256, 2002) 27.3 (14, 2003-2005) 0.2 (114, 2006) 0.78 (300-400, 2004-2006) 0.17 (5, 2007) 0.0071 (38, 2006)
water research 44 (2010) 461–476
1,4-Dioxane 2,6-Dichlorobenzamide (BAM) 4-Methylbenzenesulfonamide (p-toluenesulfonamide, 4-tolylsulfonamide) Acetylsalicylate (aspirin, acetyl salicylic acid) Alpha-amino-3-hydroxy-5-methyl4-isoxazole propionic acid (AMPA) Amidotrizoic acid (diatrizoic acid) Bentazone Benzene Benzothiazole Benzotriazole (1H-benzotriazole) Bis(chloroisopropyl)ether (BCIPE) Carbamazepine Carbendazim Chloridazon (pyrazon) Clofibric acid Dichlorophenoxyacetic acid (2,4-D) Diethyl phthalate Diethyl toluamide (DEET) Diethylamine (DEA) Diethylene glycol dimethyl ether (diglyme, bis(2-methoxy ethyl)ester)) Diethylene triamine penta acetic acid (DTPA) Dimethenamid Dimethylamine (DMA) Diuron Ethyl tert-butyl ether (ETBE) Ethylenediamine tetra acetic Acid (EDTA) Glyphosate Imidacloprid Iohexol Iomeprol (iomeron) Iopamidol Iopromide Isoproturon Methyl tert-butyl ether (MTBE) Metoprolol n-Butylbenzenesulphonamide Nicosulfuron n-Nitrosodimethylamine (NDMA)
Source
470
Table 3 – Reported concentrations in surface waters, groundwaters and drinking water and comparison to (provisional) drinking water guideline values expressed as Benchmark Quotient (BQ) values.
Sources: bank-filtrated groundwater (BFGW); groundwater (GW); surface water (SW); finished drinking water (FDW); tap water (TW); unspecified drinking water (UDW). Locations: Belgium (BE); Germany (GER); the Netherlands (NL); United States of America (USA). References: (1) Anonymous data Dutch drinking water companies; (2) ARW database; URL4; (3) Brauch et al., 2000; (4) ECB, 2002; (5) Kleinneijenhuis and Puijker, 2008; (6) KWR internal data (KWR, 2009); (7) De Voogt et al., 2008; (8) Loos et al., 2009; (9) Mout et al., 2007; (10) REWAB database, 2009; (11) RIWA database; URL5 (12) Skutlarek et al., 2006; (13) van Beelen, E.S.E. (HWL, the Netherlands), personal communication; (14) Van Leerdam et al., 2009; (15) Versteegh et al., 2007. NA: Not available; an absence of measured concentrations above the detection limit. a BQ value: ratio of reported maximum concentration to the (provisional) guideline value (cf. Table 2). b Based on an odour-threshold value (cf. Table 2).
(6) FDW, NL
0.005
(10) FDW, NL
0.09
(12) (12) (15) (10) (15)
p,p0 -Sulfonyldiphenol Perfluoroctane sulfonate (PFOS) Perfluorooctanoic acid (PFOA) Phenazone Simazine Sulfamethoxazole Tolyltriazole Trichloroethene Triethylphosphate (ethylphosphate) (TEP) Triphenylphosphine oxide (TPPO) Tris(2-chloroethyl)phosphate (TCEP) Urotropine (methenamine, hexamine)
10 (300, 2005-2006) 0.11 (NA, 2008) 0.647 (NA, 2005/2006) 0.11 (26, 2005) 0.13 (124, 2006) 0.11 (170, 2005) 0.29 (11, 2007) 1.35 (206, 2006) 0.189 (25, 2007) 0.344 (88, 2003) 0.29 (8, 2006) 10 (NA, 1997-1998)
SW, BE SW, NL SW, GER SW, NL SW, BE SW, NL SW, NL SW, NL SW, NL SW, NL SW, NL SW, GER
(6) (8) (13) (12) (11) (11) (11) (11) (11) (11) (11) (3)
0.1 0.2 0.1 0.0009 0.07 0.0003 0.0003 0.07 0.0001 0.01 0.004 0.02
NA 0.02 (NA, 2005/2006) 0.52 (NA, 2005/2006) 0.03 (8, 2005) 0.06 (4, 2004) 0.03 (4, 2007) NA 1.75 (8, 2005) NA 0.13 (6, 2007) NA NA
TW, GER TW, GER FDW, NL FDW, NL FDW, NL
0.04 0.1 0.0002 0.03 0.00007
water research 44 (2010) 461–476
471
there is a concern for drinking water production, because consumers will not accept odourous drinking water. The BQ values of the remaining compounds calculated for surface waters and groundwater ranged between 0.1 (p,p0 -sulphonylphenol, diuron) and 0.000001 (iohexol). For all iodinated contrast media the concentration in surface waters was at least three orders of magnitude less than the (provisional) guideline value. For drinking water, the BQ values of these compounds were even lower (Fig. 2B). For two compounds (benzene and PFOA) found in drinking water, the BQ value was equal to 0.1 indicating that additional assessments such as establishing trends may be warranted. However, for 15 compounds occurrence data in drinking water were not available and therefore the human health concern associated with drinking water consumption due to presence of any of these compounds remains unknown.
4.
Discussion
Rapid new developments in analytical chemistry lead to the detection and quantification of many emerging contaminants in drinking water and its environmental sources (surface water and groundwater). Since toxicological information is often absent, such compounds are a growing concern for drinking water companies and their customers. The present study attempts to address potential human health concern associated with water containing emerging contaminants. The 50 compounds included in this study represent a broad range of chemical classes for which maximum concentrations in surface waters, groundwater and/or drinking water were obtained in the downstream parts of the Rhine and Meuse basins. The results as presented in Fig. 2 indicate that a substantial margin exists between the (provisional) guideline value and the maximum concentrations of most compounds reported in surface waters, groundwaters and/or drinking water. The compounds evaluated with a relatively high BQ value (i.e. a high potential human health concern) and a known carcinogenic action are 1,4-dioxane, benzene and NDMA. The (provisional) guideline values for 1.4-dioxane (30 mg/L), benzene (10 mg/L) and NDMA (0.1 mg/L) used in the present study are based on a specific cancer risk level of 105. However, when applying a specific risk level of 106, as is common practice in the Netherlands, the provisional guidelines value would be 3 mg/L, 1 mg/L and 0.01 mg/L, respectively. This would result in BQ values much higher than the arbitrary thresholds for surface waters and drinking water employed in the present study. This indicates that very low concentrations of these compounds in drinking water could lead to a potential carcinogenic effect, and we conclude that for these compounds it is important to monitor trends in their (environmental) occurrence. Furthermore, Fig. 2 illustrates that the (provisional) guideline values of the majority of non-genotoxic compounds are at least two orders of magnitude above the Threshold of Toxicological Concern (TTC)-based drinking water target value for non-genotoxic compounds (0.1 mg/L). In addition, the (provisional) guideline values (expressed as a specific risk level of 106) of the three compounds with known carcinogenic action (1,4-dioxane, benzene and NDMA) are equal (NDMA) or much higher (1,4-
472
water research 44 (2010) 461–476
Fig. 2 – Comparison of compound concentrations in (A) surface/groundwaters and (B) drinking water to (provisional) guideline values. Benchmark Quotient (BQ) thresholds are indicated with dashed lines. Threshold of Toxicological Concern (TTC) based target value for non-genotoxic compounds (0.1 mg/L) is indicated with a dotted line. Numbers correspond to compounds as tabulated in Table 1.
dioxane and benzene) than the TTC derived target value of 0.01 mg/L for genotoxic compounds. This illustrates that the TTC based drinking water target value may be a conservative value ideally suited for exposure based waiving of compounds for which there is no sufficient toxicological information, which can be followed up by a more data-intensive evaluation. Two perfluorinated organic compounds were evaluated in the present study. For PFOA in drinking water a BQ value equal to the arbitrary threshold of 0.1 was calculated, whereas for PFOS in surface waters a BQ value of 0.2 was calculated. These persistent compounds are becoming a global problem, and PFOA and PFOS have already been detected in the ng/L range in, e.g. European and Japanese tapwaters (Ericson et al., 2007; Loos et al., 2007; Norimitsu et al., 2004). Recently, Skutlarek et al. (2006) observed at sampling site Neheim (river Ruhr catchment, a tributary of the river Rhine, Germany)
concentrations of PFOA of 0.65 mg/L in Lake Moehne, and 0.53 mg/L in corresponding drinking water, respectively. The authors concluded that water treatment steps may not effectively eliminate perfluorinated compounds to a sufficient extent, although approximately 50% of the waterworks at the Ruhr river are equipped with activated carbon filters. Hence, more research should be devoted to the behavior of perfluorinated organic compounds in drinking water treatment processes. Several structurally related iodinated contrast media (iopamidol, iohexol, iomeprol and iopromide) were evaluated in the present research. However, their calculated BQ values are much lower than the BQ threshold above which further investigations would be warranted. Iopromide, for example, is a relatively non-toxic compound with a reported safe dose (intravenous) of 50 g/person/d (Versteegh et al., 2007). Despite
water research 44 (2010) 461–476
the absence of human health effects, these compounds may deserve further attention from an environmental impact point of view. Since environmental sublethal effects of iodinated contrast media to organisms are largely unknown, taken together with high persistence and environmental presence at relatively high concentrations, additional environmental assessments may be necessary. For compounds with a low (provisional) guideline value as identified in the present study (e.g. carbamazapine), additional environmental monitoring may be warranted to characterize concentrations and to establish trends in their occurrence. As shown by Walraven and Laane (2009), river flow rates may influence contaminant concentrations seasonally, thus resulting in substantially varying BQ values. For example, it can be observed that the riverine concentration of the fuel oxygenate MTBE is highly dependent on the flow of the river Meuse. Similar patterns may occur for other compounds, resulting in (temporarily) exceedance of the BQ threshold. The evaluation as presented here supports the conclusion that the majority of the selected compounds as found in surface waters, groundwater and drinking water do not pose an appreciable concern to human health. This finding of no adverse effect to human health from exposure to trace quantities of compounds (e.g. pharmaceuticals) in surface waters and/or drinking water is supported by other results reported in the literature. Kingsbury et al. (2008) recently evaluated the potential health effects of 148 organic compounds in source water and finished water. The authors showed that the annual mean concentration of all compounds detected in finished water was less than the established human health benchmarks. Furthermore, Snyder et al. (2008) arrived at the same findings after evaluating human health effects associated with potential drinking water exposure of a suite of 62 indicator pharmaceuticals and potential endocrine disrupting compounds. Despite the absence of any concern to human health, drinking water remains a major point of consumer concern and some residual uncertainties need further exploration. For example, drinking water guideline values are developed using toxicity information for single compounds. Hence, the longterm cumulative dose-additive or synergistic effects of low concentrations of contaminants co-occurring as mixtures on human health and potentially sensitive sub-populations remain currently unknown. Understanding and implementing of such information is important for the development of future (enforceable) guideline values. Finally, the relatively large data gap on occurrence of compounds in drinking water should compel further research and assessment, especially for those compounds with a low (provisional) guideline value.
5.
Major conclusions
For most compounds evaluated in the present assessment, a substantial margin exists between the (provisional) guideline value and the maximum concentrations in surface waters, groundwaters and/or drinking water. The TTC based drinking water target values (0.1 mg/L and 0.01 mg/L for non-carcinogenic compounds, respectively) as proposed earlier are the conservative values they are meant
473
to be. They are optimally suited to provide exposure based waiving. The concentrations in drinking water of compounds such as MTBE, ETBE, 1,4-dioxane, NDMA and benzene should be monitored closely, since their guideline values are easily exceeded. Alkylated perfluorinated compounds such as PFOA and PFOS are environmentally persistent compounds and their increasing occurrence in (the sources of) drinking water should be monitored closely. For compounds with a very low (provisional) guideline value (e.g. mutagenic and carcinogenic compounds) it is important to better establish trends in their environmental occurrence. From a toxicological point of view iodinated contrast media as present in drinking water, such as amidotrizoic acid iopamidol, iohexol and iopromide, are not a direct concern for human health. However, further environmental assessment may be necessary, especially since the sublethal (ecological) effects of these compounds are largely unknown. Better understanding of the potential mixture effects of emerging compounds present in drinking water is important for the development of future guideline values.
Acknowledgements The authors wish to acknowledge the significant contribution to this work by Leo Puijker (KWR) and Ruud Jansen. Thanks also to Dr. Corine Houtman (HWL, the Netherlands) for her valuable comments on the manuscript. The research described was funded by the Joint Research Programme of the Dutch Water utilities (BTO).
references
Benzotriazoles Coalition, 2001. Benzotriazoles category justification and testing rationale. Synthetic Organic Chemical Manufacturers Organization. http://www.epa.gov/oppt/ chemrtk/pubs/summaries/benzo/c13456.pdf (accessed June 2009). BIBRA Toxicology International, 1995. Toxicity Profile of Benzotriazole. BIBRA Toxicology International, Carshalton (Surrey), UK. Brauch, H.J., Sacher, F., Denecke, E., Tacke, T., 2000. Efficiency of river-bank filtration for the removal of polar organic compounds (Werksamkeit der Uferfiltration fuer die Enfernung von polaren organischen Spurenstoffen). WasserAbwasser 141 (4), 226–234. Brown, D., Butterworth, K.R., Gaunt, I.F., Grasso, P., Gangolli, S.D., 1978. Short-term oral toxicity study of diethyl phthalate in the rat. Food Cosmet. Toxicol. 16, 415–422. California EPA (California Environmental Protection Agency), 2000. N,N-diethyl-meta-toluamide (DEET) risk characterization document. Department of pesticide regulation. Document number RCD 00-01, Sacramento, California, USA. CIIT (Chemical Industry Institute of Toxicology), 1990. Twenty four month final report inhalation toxicity of dimethylamine in F344 rats and B6C3F1 mice. CIIT Archives, Docket #11957. Ciba-Geigy, 1988. Combined chronic toxicity/oncogenicity study in rats. Basle, Ciba-Geigy, 1988 (unpublished study submitted to WHO).
474
water research 44 (2010) 461–476
Cotruvo, J.A., 1988. Drinking water standards and risk assessment. Regul. Toxicol. Pharmacol. 8 (3), 288–299. Danish EPA (Danish Environmental Protection Agency), 2004. Evaluation of Health Hazards by Exposure to BAM (2,6Dichlorobenzamide) and Risk Characterisation of Drinking Water Exposure. Environmental Project Nr. 943 2004. Dawson, B.V., Johnson, P.D., Goldberg, S.J., Ulreich, J.B., 1993. Cardiac teratogenesis of halogenated hydrocarbon-contaminated drinking water. J. Am. Coll. Cardiol. 21 (6), 1466–1472. De Voogt, P., Vink, C., Puijker, L.M., 2008. MTBE and ETBE in Dutch groundwater (MTBE en ETBE in Nederlands grondwater), KWR 08.043, KWR, Nieuwegein, The Netherlands. DuPont de Nemours and Company, Inc., 1964. Report MRID No. 00017763, 00080899. DECOS (Dutch Expert Committee for Occupational Standards), 2000. 1,2,3-benzotriazole. Health-based recommended occupational exposure limit. No. 2000/14OSH, The Hague, The Netherlands. ECB (European Chemicals Bureau), 2000a. IUCLID dataset on benzotriazoles. ECB (European Chemicals Bureau), 2000b. IUCLID dataset on Chloridazon. ECB (European Chemicals Bureau), 2000c. IUCLID dataset on Diethylene triamine penta acetic acid. ECB (European Chemicals Bureau), 2000d. IUCLID dataset on triphenylphosphine oxide. ECB (European Chemicals Bureau), 2002. European Risk Assessment Report 1,4-dioxane. EFSA (European Food Safety Authority), 2008. Opinion of the scientific panel on contaminants in the food chain on perfluorooctane sulfonate (PFOS), perfluorooctanoic acid (PFOA) and their salts. EFSA J. 53, 1–131. Ericson, I., Nadal, M., van Bavel, B., Lindstro¨m, G., Domingo, J.L., 2007. Levels of perfluorochemicals in water samples from Catalonia, Spain: is drinking water a significant contribution to human exposure? Environ. Sci. Pollut. Res. 15, 614–619. Goldenthal, E.I. (International Research and Development Corp.), 1994. Evaluation of DEET in a one-year chronic oral toxicity study in dogs. DEET Joint Venture/Chemical Specialties Manufacturers Assoc. DPR vol. 50191-164, Rec. No. 131265. Health Canada, 2003. Benchmark Dose for TCE in Drinking Water. Health Canada, Healthy Environments and Consumer Safety Branch, Biostatistics Unit, April, Ottawa, Ontario. Hogenboom, A.C., Van Leerdam, J.A., De Voogt, P., 2009. Accurate mass screening and identification of emerging contaminants in environmental samples by liquid chromatography–hybrid linear ion trap Orbitrap mass spectrometry. J. Chromatogr. A 1216 (3), 510–519. Hurni, H., Ohder, H., 1973. Reproduction study with formaldehyde and hexamethylenetetramine in beagle dogs. Food Cosmet. Toxicol. 11 (3), 459–462. Kingsbury, J.A., Delzer, C.D., Hopple, J.A., 2008. Anthropogenic organic compounds in source water of nine community water systems that withdraw from streams, 2002–2005, USGS, Reston, Virginia, USA, Report 2008–5208. Kleinneijenhuis, A.J., Puijker, L.M., 2008. Monitoring program for DMSA, DMST, N,N-dimethylsulfamide and its ozone reaction product NDMA in untreated and ozone-treated water. TNO report 7686, Delft, The Netherlands. Kolpin, D.W., Skopec, M., Meyer, M.T., Furlong, E.T., Zaugg, S.D., 2004. Urban contribution of pharmaceuticals and other organic wastewater contaminants to streams during differing flow conditions. Sci. Total Environ. 328 (1–3), 119–130. Kroes, R., Renwick, A.G., Cheeseman, M., Kleiner, J., Mangelsdorf, I., Piersma, A., Schilter, B., Schlatter, J., Van Schothorst, F., Vos, J.G., Wurtzen, G., 2004. Structurebased thresholds of toxicological concern (TTC): guidance for
application to substances present at low levels in the diet. Food Chem. Toxicol. 42, 65–83. Krop, H. de Voogt, P., 2008. Physicochemical parameters and source markers of PFAS. Report IVAM ISO doc O/0818, Amsterdam, The Netherlands, pp. 1–147. KWR Watercycle Research Institute, 2009. KWR internal data, Nieuwegein, the Netherlands. Larsen, M.L., Illingworth, D.R., O’Malley, J.P., 1994. Comparative effects of gemfibrozil and clofibrate in type III hyperlipoproteinemia. Atherosclerosis 106, 235–240. Loos, R., Wollgast, J., Huber, T., Hanke, G., 2007. Polar herbicides, pharmaceutical products, PFOS, PFOA, and nonylpheniol and its carboxylates and ethoxylates in surface and tap waters around Lake Maggiore in Northern Italy. Anal. Bioanal. Chem. 387, 1469–1478. Loos, R., Gawlik, B.M., Locoro, G., Rimaviciute, E., Contini, S., Bidoglio, G., 2009. EU-wide survey of polar organic persistent pollutants in European river waters. Environ. Pollut. 157 (2), 561–568. Lynch, D.W., Moorman, W.J., Stober, P., Lewis, T.R., Iverson, W.O., 1986. Subchronic inhalation of diethylamine vapour in Fisher344 rats: organ toxicity. Fund. Appl. Toxicol. 6 (3), 559–565. Medinsky, M.A., Wolf, D.C., Cattley, R.C., Wong, B., Janszen, D.B., Farris, G.M., Wright, G.A., Bond, J.A., 1999. Effects of a thirteen-week inhalation exposure to ethyl tertiary butyl ether on Fischer-344 rats and CD-1 mice. Toxicol. Sci. 51, 108–118. Mitsubishi Chemical Safety Institute Ltd. (date of study unknown). 4,4’-sulfonyldiphenol. Summary of repeated dose and developmental studies. Kashima Laboratory, Ibaraki, Japan. Mitsumori, K., Usui, T., Takahashi, K., Shirasu, Y., 1979. Twentyfour month chronic toxicity studies of dichlorodiisopropyl ether in mice. J. Pestic. Sci. 4 (3), 323–335. Mons, M., Van der Hoek, J.P., Stoks, P., Van der Kooij, D., 2008. Vision on and guideline values for xenobiotic substances in drinking water (Visie op en streefwaarden voor milieuvreemde stoffen in drinkwater). H2O 4, 1–4 (in Dutch). Morgareidge, K., 1971. 90-Day feeding study with benzothiazole in rats. Unpublished report from Food and Drug Research Laboratories, Inc., Maspeth, New York, USA. Submitted to WHO by Flavor and Extract Manufacturers’ Association of the United States. Morisetti, A., Tirone, P., Luzzani, F., de Haen, C., 1994. Toxicological safety assessment of iomeprol, a new X-ray contrast agent. Eur. J. Radiol. 1 (18 Suppl), S21–31. Mout, L., Jansen, L., van der Pijll, S., Nierop, S., 2007. Implementation program dimethenamid-p. Clean sources, now and in the future (uitvoeringsprogramma dimethenamid-p. Schone bronnen, nu en in de toekomst: tweede reeks knelpunten (http://www. schonebronnen.nl/). The Hague, The Netherlands. Munro, I.C., Ford, R.A., Kennepohl, E., Sprenger, J.G., 1996. Correlation of structural class with no-observed effect levels: a proposal for establishing a threshold of concern. Food Chem. Toxicol 34, 829–867. NCI (National Cancer Institute), 1978. Bioassay of 1,4-dioxane for possible carcinogenicity, CAS No. 123-91-1. NCI Carcinogenesis Technical Report Series No. 80. DHEW Publication No. (NIH) PB-285-711, Bethesda, Maryland, USA. Nghiem, L.D., Schafer, A.I., Elimelech, M., 2004. Removal of natural hormones by nanofiltration membranes: measurement, modeling, and mechanisms. Environ. Sci. Technol. 38 (6), 1888–1896. Norimitsu, S., Kouji, H., Kayoko, I., Kazuaki, S., Takeo, Y., Akio, K., 2004. Perfluorooctanoate and perfluorooctane sulfonate concentrations in surface water in Japan. J. Occup. Health 46, 49–59. NTP (National Toxicology Program), 1986. Toxicology and carcinogenesis studies of benzene in F344/N rats and B6C3F1
water research 44 (2010) 461–476
mice (gavage studies). Research Triangle Park, NC, USA, US Department of Health and Human Services. (Technical report series No. 289). NTP (National Toxicology Program), 1987. Teratologic evaluation of diethylene glycol dimethyl ether (CAS No. 111-96-6) administered to New Zealand White rabbits on gestation days 6 through 19. Research Triangle Park, NC, USA, National Institute of Environmental Health Sciences, National Toxicology Program (NTP-87-108; PB 87–209532). NTP (National Toxicology Progam), 1991. Toxicology and carcinogenesis studies of tris(2-chloroethyl)phosphate in F344/N rats and B6C3F1 mice. Technical report series No. 391. NTP TR 391, Research Triangle Park, NC, USA. OECD/SIDS (Organization for Economic Development and Cooperation/Screening Information DataSet), 1994. pToluenesulfonamide. UNEP publications, Nairobi, Kenya. OECD/SIDS (Organization for Economic Development and Cooperation/Screening Information DataSet), 1998. Triethylphosphate. UNEP publications, Nairobi, Kenya. Peto, R., Gray, R., Brantom, P., Grasso, P., 1991a. Effects on 4080 rats of chronic ingestion of N-nitrosodiethylamine or Nnitrosodimethylamine: a detailed dose–response study. Cancer Res. 51, 6415–6451. Peto, R., Gray, R., Brantom, P., Grasso, P., 1991b. Dose and time relationships for tumor induction in the liver and esophagus of 4080 inbred rats by chronic ingestion of Nnitrosodiethylamine or N-nitrosodimethylamine. Cancer Res. 51, 6452–6469. Proviron Fine Chemicals, 2003. IUCLID dataset on Nbutylbenzenesulphonamide. REWAB Database, 2009. KWR Watercycle Research Institute, Nieuwegein, The Netherlands. Robinson, M., Bruner, R.H., Olson, G.R., 1990. Fourteen- and ninety-day oral toxicity studies of methyl-t-butyl ether in Sprague-Dawley rats. J. Am. Coll. Toxicol. 9, 525–540. Schwarzenbach, R.P., Escher, B.I., Fenner, K., Hofstetter, T.B., Johnson, C.A., Von Gunten, U., Wehrli, B., 2006. The challenge of micropollutants in aquatic systems. Science 313 (5790), 1072–1077. SCOEL (Scientific Committee on Exposure Limits), 1991. No 11b. Recommendations of the SCOEL for DMA. SCOEL (Scientific Committee on Exposure Limits), 2002. No 91. Recommendations of the SCOEL for DEA. Schwab, B.W., Hayes, E.P., Fiori, J.M., Mastrocco, F.J., Roden, N.M., Cragin, D., Meyerhoff, R.D., D’Aco, V.J., Anderson, P.D., 2005. Human pharmaceuticals in US surface waters: a human health risk assessment. Regul. Toxicol. Pharmacol. 42 (3), 296–312. Seacat, A.M., Thomford, P.J., Hansen, K.J., Olsen, G.W., Case, M.T., Butenhoff, J.L., 2002. Subchronic toxicity studies on perfluorooctanesulfonate potassium salt in cynomolgus monkeys. Toxicol. Sci. 68 (1), 249–264. Sherman, H., 1972. Long-term feeding studies in rats and dogs with 2-benzimadazole carbamic acid, methyl ester (INE-965) (50% and 70% MBC wettable powder formulations). Unpublished report from E.I. DuPont de Nemours and Co., Inc., Haskell Laboratory, Newark, Delaware, USA. Singh, G., Driever, P.H., Sander, J.W., 2005. Cancer risk in people with epilepsy: the role of antiepileptic drugs. Brain 128 (Pt 1), 7–17. Skutlarek, D., Exner, M., Farber, H., 2006. Perfluorinated surfactants in surface and drinking waters. Environ. Sci. Pollut. Res. 13 (5), 299–307. Snyder, S.A., Villeneuve, D.L., Snyder, E.M., Giesy, J.P., 2001. Identification and quantification of estrogen receptor agonists in wastewater effluents. Environ. Sci. Technol. 35 (18), 3620–3625. Snyder, S.A., Trenholm, R.A., Snyder, E.M., Bruce, G.M.P., Hemming, J.D.C., 2008. Toxicological Relevance of EDCs and
475
Pharmaceuticals in Drinking Water. Awwa Research Foundation, Denver, USA, p. 484. Stokinger, H.E., Woodward, R.L., 1958. Toxicologic methods for establishing drinking water standards. J. Am. Water Works Ass. 50, 515–529. Swarm, R.L., Roberts, G.K., Levy, A.C., Hines, L.R., 1973. Observations on the thyroid gland in rats following the administration of sulfamethoxazole and trimethoprim. Toxicol. Appl. Pharm. 24 (3), 351–363. Swartjes, F.A., Baars, A.J., Fleuren, R.H.L.J., Otte, P.F., 2004. Risk limits for MTBE (methyl tertiair-butyl ether) in soil, sediment, groundwater, surface water, drinking water and for drinking water production (risicogrenzen voor MTBE (Methyl tertiairButyl Ether) in bodem, sediment, grondwater, oppervlaktewater, drinkwater en voor drinkwaterbereiding) RIVM report 711701039/2004, Bilthoven, The Netherlands. Toccalino, P.L., 2007. Development and application of healthbased screening levels for use in water-quality assessments. US Geological Survey Scientific Investigations Report 20075106, 12 p, Reston, Virginia, USA. URL1: http://www.epa.gov/waterscience/criteria/drinking/ dwstandards.pdf, accessed June 2009. URL2: http://www.who.int/water_sanitation_health/dwq/ guidelines/en/, accessed June 2009. URL3: http://www.ctb.agro.nl/ctb_files/11995_01.html, accessed May 2009 (In Dutch). URL4: http://www.iawr.org/index.php, publications, accessed January 2009. URL5: http://www.riwa-rijn.org/riwa_en.html, publications, accessed January 2009. US EPA (United States Environmental Protection Agency) IRIS database (1988a). 1,4-dioxane. US EPA (United States Environmental Protection Agency) IRIS database, 1988b. Diethyl phthalate. US EPA (United States Environmental Protection Agency) IRIS database, 1988c. Diuron. US EPA (United States Environmental Protection Agency) IRIS database, 1989. Bis(chloroisopropyl)ether. US EPA (United States Environmental Protection Agency), 2006. National Primary Drinking Water Standards. http://www.epa. gov/safewater/consumer/pdf/mcl.pdf (accessed August 2009). Van Leerdam, J.A., Hogenboom, A.C., Kooi van der, M.E., de Voogt, P., 2009. Determination of polar 1-H-benzotriazoles and benzothialzoles in water by solid-phase extraction and liquid chromatography LTQ Orbitrap mass spectrometry. Int. J. Mass Spectrom. 282 (3), 99–107. Van Leeuwen, F.X., 2000. Safe drinking water: the toxicologist’s approach. Food Chem. Toxicol. 38 (1 Suppl), S51–S58. Van Leeuwen, C.J., Vermeire, T., 2007. Risk Assessment of Chemicals, second ed., Springer, ISBN 978-4020-6101-1. Van Wezel, A.P., Puijker, L., Vink, C., Versteegh, A., de Voogt, P., 2009. Odour and flavour thresholds of gasoline additives (MTBE, ETBE and TAME) and their occurrence in Dutch drinking water collection areas. Chemosphere 76, 672–676. Versteegh, J., Van Der Aa, N., Dijkman, E., 2007. Pharmaceuticals in drinking water and sources of drinking water - results of monitoring campaign 2005/2006 (geneesmiddelen in drinkwater en drinkwaterbronnen - resultaten van het meetprogramma 2005/2006), 53, RIVM, Bilthoven, The Netherlands. Number: R(T)-1-67. Walraven, N., Laane, R.W., 2009. Assessing the discharge of pharmaceuticals along the Dutch coast of the North sea. Rev. Environ. Contam. Toxicol. 199, 1–18. 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. Environ. Sci. Technol. 37 (17), 6649–6663. WHO (World Health Organization), 1996. Simazine in drinkingwater. Background document for development of WHO
476
water research 44 (2010) 461–476
Guidelines for drinking-water quality. WHO/SDE/WSH/03.04/ 42, Geneva, Switzerland. WHO (World Health Organization), 1998. Guidelines for drinking water quality. Second edition, volume 1 recommendations: addendum, ISBN 92 4 154514 3, Geneva, Switzerland. WHO (World Health Organization), 2002. Diethylene glycol dimethyl ether. Concise International Chemical Assessment Document 41, Geneva, Switzerland. WHO (World Health Organization), 2003a. Benzene in drinking water. WHO report WHO/SDE/WSH/03.04/24, Geneva, Switzerland. WHO (World Health Organization), 2003b. 2,4-D in drinking water. WHO report WHO/SDE/WSH/03.04/70, Geneva, Switzerland. WHO (World Health Organization), 2003c. Isoproturon in Drinking-water. Background document for development of WHO Guidelines for Drinking-water Quality WHO/SDE/WSH/ 03.04/37, Geneva, Switzerland. WHO (World Health Organization), 2005a. Glyphosate and AMPA in drinking-water. Background document for development of WHO Guidelines for Drinking-water Quality WHO/SDE/WSH/ 03.04/97, Geneva, Switzerland. WHO (World Health Organization), 2005b. Trichloroethene in Drinking-water, Background document for development of WHO Guidelines for Drinking-water Quality. WHO/SDE/WSH/ 05.08/22, Geneva, Switzerland. WHO (World Health Organization), 2006. Guidelines for drinking water quality, First Adendum to third ed., Recommendations, vol. 1, 2006, Geneva, Switzerland. WHO (World Health Organization), 2008. NNitrosodimethylamine in drinking-water. Background
document for preparation of WHO Guidelines for drinkingwater quality. Geneva, World Health Organization (WHO/HSE/ AMR/08.03/8), Geneva, Switzerland. WHO/JECFA (World Health Organization/joint FAO/WHO expert committee on food additives), 1973. Food Additives Series 32; Toxicological Evaluation of Certain Food additives and Contaminants. Prepared by the forty-first meeting of the joint FAO/WHO expert committee. WHO/JECFA (World Health Organization/joint FAO/WHO expert committee on food additives), 1974. Food Additives Series No. 5; Seventeenth Report of the Joint FAO/WHO Expert Committee on Food Additives, World Health Organization technical report series, 1974, No. 539; FAO Nutrition Meetings Report Series, 1974, No. 53. WHO/JECFA (World Health Organization/joint FAO/WHO expert committee on food additives), 1995. 892. Carbendazim (Pesticide residues in food: 1995 evaluations Part II Toxicological & environmental). WHO/JECFA (World Health Organization/joint FAO/WHO expert committee on food additives), 2003. WHO Food Additives Series No. 50; sulfur-containing heterocyclic compounds. WHO/JMPR (World Health Organization/Joint Meeting on Pesticide Residues), 2001. Toxicological evaluations imidacloprid. Pesticide residues in food. WHO/JMPR (World Health Organization/Joint Meeting on Pesticide Residues), 2005. Dimethenamid-p. pp. 189–239. Wilson, A.B., Thorpe, E., 1971. Toxicity studies on the ‘‘Prefix’’ residue 2,6-dichlorobenzamide: two year oral experiment with dogs. Tunstall Laboratory. Report number: Lab Project Number: T507531; Group research report: TLGR.0028.71, pp. 1–28.
water research 44 (2010) 477–492
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Monitoring the biological activity of micropollutants during advanced wastewater treatment with ozonation and activated carbon filtration M. Macova a, B.I. Escher a,*, J. Reungoat b, S. Carswell a,c, K. Lee Chue a, J. Keller b, J.F. Mueller a a
University of Queensland, National Research Centre for Environmental Toxicology (EnTox), 39 Kessels Road, Brisbane, QLD 4108, Australia University of Queensland, Advanced Water Management Centre (AWMC), Brisbane, QLD 4072, Australia c Queensland Health Forensic and Scientific Services, Organics Laboratory, 39 Kessels Road, Brisbane, QLD 4108, Australia b
article info
abstract
Article history:
A bioanalytical test battery was used to monitor the removal efficiency of organic micro-
Received 13 May 2009
pollutants during advanced wastewater treatment in the South Caboolture Water Reclamation
Received in revised form
Plant, Queensland, Australia. This plant treats effluent from a conventional sewage treatment
28 August 2009
plant for industrial water reuse. The aqueous samples were enriched using solid-phase
Accepted 7 September 2009
extraction to separate some organic micropollutants of interest from metals, nutrients and
Available online 16 September 2009
matrix components. The bioassays were chosen to provide information on groups of chemicals with a common mode of toxic action. Therefore they can be considered as sum indicators
Keywords:
to detect certain relevant groups of chemicals, not as the most ecologically or human health
Bioassays
relevant endpoints. The baseline toxicity was quantified with the bioluminescence inhibition
Baseline toxicity
test using the marine bacterium Vibrio fischeri. The specific modes of toxic action that were
Phytotoxicity
targeted with five additional bioassays included aspects of estrogenicity, dioxin-like activity,
Estrogenicity
genotoxicity, neurotoxicity, and phytotoxicity. While the accompanying publication discusses
Genotoxicity
the treatment steps in more detail by drawing from the results of chemical analysis as well as
Toxic equivalency concept
the bioanalytical results, here we focus on the applicability and limitations of using bioassays for the purpose of determining the treatment efficacy of advanced water treatment and for water quality assessment in general. Results are reported in toxic equivalent concentrations (TEQ), that is, the concentration of a reference compound required to elicit the same response as the unknown and unidentified mixture of micropollutants actually present. TEQ proved to be useful and easily communicable despite some limitations and uncertainties in their derivation based on the mixture toxicity theory. The results obtained were reproducible, robust and sensitive. The TEQ in the influent ranged in the same order of magnitude as typically seen in effluents of conventional sewage treatment plants. In the initial steps of the treatment chain, no significant degradation of micropollutants was observed, and the high levels of dissolved organic carbon probably affected the outcome of the bioassays. The steps of coagulation/flocculation/dissolved air flotation/sand filtration and ozonation decreased the effectbased micropollutant burden significantly. ª 2009 Elsevier Ltd. All rights reserved.
* Corresponding author. Tel.: þ61 (0) 7 3274 9180; fax: þ61 (0) 7 3274 9003. E-mail address:
[email protected] (B.I. Escher). 0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2009.09.025
478
1.
water research 44 (2010) 477–492
Introduction
Water quantity and water quality issues have led many countries to explore water reuse and water reclamation as alternatives to provide water for direct use in industrial applications as well as agricultural and gardening use, and for indirect use to supplement water reservoirs. Australia is amongst the countries most affected by uneven distribution of water resources and droughts. Therefore, various water treatment and desalination plants have been put into operation, among them the Water Reclamation Plant in South Caboolture, Queensland, Australia (van Leeuwen et al., 2003). Besides hygiene issues, organic micropollutants are of major concern because classical wastewater treatment with biological activated sludge only partially removes hazardous chemicals. Furthermore there is the potential that persistent and toxic micropollutants and transformation products could build up during the recycling process. Urban and agricultural wastewaters contain a high number of organic micropollutants, including pharmaceuticals and personal care products, hormones and industrial chemicals, biocides and pesticides (Schwarzenbach et al., 2006). A large number of studies have investigated the fate and removal of various individual and specific groups of micropollutants by treatment processes like biological activated sludge treatment (Ternes et al., 2004; Joss et al., 2005), ozonation and advanced oxidation processes (von Gunten, 2003; Huber et al., 2004; Huber et al., 2005; Lee et al., 2008) or activated carbon treatment (Sanchez-Polo et al., 2006; Snyder et al., 2007) but little is known how these treatment processes change the composition and biological activity of the micropollutants. Toxicity testing may provide complementary information to chemical analysis on the sum of micropollutants present in treated water. However, whole-effluent toxicity testing, which is a valuable tool for hazard assessment of industrial and municipal effluents (Chapman, 2000), is inappropriate to evaluate the fate of micropollutants during advanced water treatment because acute toxicity tests are not sensitive enough for purified water and long-term chronic tests are too expensive and time-consuming for routine monitoring. In addition, the physical treatment of wastewater alters the matrix and electrolyte composition and it becomes virtually impossible to differentiate the effect caused by micropollutants from that induced by the matrix, such as organic matter content, pH, ionic strength or nutrient content. Therefore, over the last years, we have developed a framework for bioanalytical quantification of organic micropollutants, which has three main cornerstones: Firstly, the aqueous samples are enriched using solid-phase extraction (SPE) to separate the organic micropollutants of interest from metals and matrix components. This SPE method was previously validated for recovery and extraction yield for selected chemicals including pesticides and pharmaceuticals (Bengtson Nash et al., 2005a; Escher et al., 2005; Escher et al., 2008a). Secondly, a battery of bioanalytical methods were selected that cover a non-specific bioassay, the bioluminescence inhibition of the marine bacterium Vibrio fischeri and
five bioassays for specific modes of toxic action that are indicative for groups of chemicals of particular relevance for human and environmental health, including aspects of estrogenicity, dioxin-like activity, genotoxicity, neurotoxicity, and phytotoxicity (Muller et al., 2007; Muller et al., 2008; Escher et al., 2008a; Escher et al., in press). Table 1 summarizes the key features of the bioassays and more details are given in the Supporting Information. Finally, we developed a coherent data evaluation method that is based on the toxic equivalency concept (Villeneuve et al., 2000; Escher et al., 2008b). Thus, the effects at various steps of the treatment chain can be easily compared and treatment efficiency can be compared between the different effect endpoints. Bioassays in combination with SPE have been successfully applied by several groups to evaluate various water treatment options (Escher et al., 2008a; Cao et al., 2009). In particular, the bioluminescence inhibition test with Vibrio fischeri and the Salmonella based test for genotoxicity have been widely used despite their recognized limitations (Johnson, 2005; Petala et al., 2006; Petala et al., 2008). The present work is more extensive covering several additional toxicity endpoints, allowing high-throughput monitoring applications while remaining cost-efficient. Here we present this framework for the first time to evaluate all steps of enhanced water treatment in the South Caboolture Water Reclamation Plant (van Leeuwen et al., 2003). This paper focuses on the advantages and limitations in the application of the bioanalytical test battery, and the interpretation and significance of the results. In the accompanying paper, the results obtained from the bioanalytical tools are compared with chemical analytical data for a variety of pharmaceuticals and pesticides and conclusions for advanced wastewater treatment are drawn (Reungoat et al., 2010).
2.
Materials and methods
2.1.
Samples and sites
Four consecutive sets of 24-hour composite samples were collected on 11-07-08, 22-07-08, 27-07-08 and 06-08-08 using refrigerated autosamplers at 10 sites of the South Caboolture Water Reclamation Plant (Table 2).
2.2.
Solid phase extraction
Immediately after sampling, concentrated HCl (36%) was added to a final concentration of 5 mM for preservation. It was demonstrated in earlier work that a pharmaceutical cocktail in a wastewater matrix had highest recoveries for HLB at pH 3 (Escher et al., 2005). Samples were extracted using 1 g OASIS HLB solid phase material in 20 mL cartridges (Waters, Australia) following centrifugation (only S1 sample) and filtration with a glass fibre filter. After conditioning the cartridges with 10 mL methanol and 20 mL of 5 mM HCl in MilliQ water, a known volume of sample was percolated under vacuum. The cartridges were dried overnight under vacuum and were eluted with 10 mL
479
water research 44 (2010) 477–492
Table 1 – Bioanalytical test battery (adapted from Muller (2008)). Assay
Measured by
Target mode of toxic action
Bioluminescence inhibition test
Reduction in luminescence of the naturally bioluminescent marine bacteria Vibrio fischeri
Effect on energy status through cytotoxicity and/or baseline toxicity
All chemicals
Acetylcholinesterase (AChE) Inhibition Assay
Product of the enzymatic reaction (colorimetric)
Organophosphates and carbamate insecticides
Imaging-PAM Assay
Increased Chl a fluorescence, inversely proportional to PS II photosynthetic yield Cellular proliferation in estrogenic dependent cell line
Inhibition of the enzyme, which hydrolyses the neurotransmitter acetylcholine PS II derived photosynthesis inhibition Estrogenicity
International Standard Organisation, 1998; Johnson, 2005; Farre´ et al., 2006; Escher et al., 2008a Ellman et al., 1961; Deutsche Norm, 1995; Hamers et al., 2000
Triazine and phenylurea herbicides
Schreiber et al., 2002; Schreiber et al., 2007
Estrogens, estrogenic industrial chemicals Polychlorinated dibenzodioxins/ furans (PCDD/F) and biphenyls (PCB) or polycyclic aromatic hydrocarbons (PAH) Chlorinated byproducts, aromatic amines, polycyclic aromatic hydrocarbons (PAH)
Soto et al., 1995; Ko¨rner et al., 1999
E-SCREEN
AhR-CAFLUX
Induction of green fluorescent protein under the control of Ah receptor; (fluorescence)
Dioxin-like activity; Aryl hydrocarbon (Ah) receptor activation
umuC assay
Induction of ß-galactosidase enzyme as an indicator of DNA damage; Product of the enzymatic reaction (colorimetric)
Genotoxicity; DNA damage (SOS-response)
methanol and 10 mL hexane:acetone (1:1). All eluates were evaporated to dryness and reconstituted with 0.5 mL of ethanol.
Chemicals that give response
dilution factorbioassay ¼
Literature reference
Nagy et al., 2002; Zhao and Denison, 2004
Oda et al., 1985; Reifferscheid et al., 1991; International Standard Organisation, 2000
volume of extract added to bioassay total volume of bioassay (2)
2.3.
Bioanalytical tools
The bioassays were performed as previously reported (Escher et al., 2005; Muller et al., 2007; Muller et al., 2008; Escher et al., 2008a; Escher et al., in press) using the methods cited in Table 1 and described in detail in the Supporting Information.
The final relative enrichment factor REF is the combination of the enrichment of the extraction and the dilution in the bioassay (Eq. 3) (Escher et al., 2006; Muller et al., 2007) and represents the enrichment or dilution of the original sample in each bioassay (Table 3). The REF is equivalent to concentration and is expressed in the units [Lwater sample/Lbioassay]. REF ¼ dilution factorbioassay enrichment factorSPE
2.4.
(3)
Relative enrichment factor of the samples REF
For each sample, the enrichment factor of the solid phase extraction was calculated using (Eq. 1) which represents the enrichment of the extract compared to the source water. enrichment factorSPE ¼
Vwater Vextract
(1)
Sample volume varied from 2.0 to 2.5 L. The final volume of each extract was 0.5 mL therefore the enrichment factorSPE was between 3500 and 5500 depending on the sample. An aliquot of the enriched sample extracts were then added to the microtiter plate of the respective bioassay and serially diluted by a test medium to obtain a concentrationeffect curve. A dilution factor of each bioassay was calculated using (Eq. 2). The dilution factors ranged from 0.05 to 0.007 for the highest dose depending on the bioassay.
2.5.
Evaluation of the concentration-effect curves
Due to the nature of the endpoints, the six bioanalytical tests of the battery could not be evaluated with a single data evaluation model (Table 3). In general, all concentration effect curves of the reference compounds and sample extracts followed a log-logistic function (Eq. 4), which was fitted using Prism 5.0 software (GraphPad, San Diego, CA, USA). Adjustable parameters are the minimum and maximum of the effect, min and max, the slope s, and the effect concentration inducing 50% of the maximum effect, EC50. If the concentration-effect curve dropped sharply at the range of higher doses indicating that the sample has a cytotoxic effect, the cytotoxic concentrations were excluded from the concentration-effect assessment.
480
water research 44 (2010) 477–492
Table 2 – Sample details. Samples were each taken after the described process. More details are given in Reungoat et al. (2010). TOC/DOC data represent the average of two samplings (22-07-08 and 06-08-08). Sample ID
Blank S1 S2 S3 S4 S5 S6 S7 C3 C4 C5
Site description
TOC (mg/L)
MilliQ water Influent (Effluent from WWTP) Denitrification Pre-Ozonation Coagulation/flocculation/DAFF Main ozonation Activated carbon filtration Effluent Biological sand filter fed with S4 water Biological activated carbon filter fed with S4 water Biological activated carbon filter fed with S5 water
DOC (mg/L)
avg
sd
avg
sd
n.a. 20 24 21 11 9.9 6.6 7.1 8.5 4.0 3.6
n.a. 4 n.a. 7 1 0.9 0.6 0.5 1 0.4 0.07
n.a. 17 21 21 11 10 6.6 6.8 8.8 4.2 4.1
n.a. 4 n.a. 4 0.5 0.9 0.9 0.7 0.8 0.04 0.08
avg - average; sd - standard deviation; n.a. - not applicable.
effect ¼ min þ
maxmin 1þ10sðlog EC50 log ðconcentration of reference compound or REF of sampleÞÞ (4)
Three bioassays, the inhibition of bioluminescence of Vibrio fischeri, inhibition of AChE and inhibition of PSII photosynthetic yield, all produced effects between 0% and 100%, therefore the maximum effect in (Eq. 4) could be set to 100% (Table 3). Two bioassays, the AhR-CAFLUX and the E-SCREEN, yield information on the induction of a receptor or proliferation of cells and there is no absolute maximum effect. Dependent on the mode of action, the maximum effect can be higher or lower than the maximum effect of the reference compounds. The extent of difference of maximally attainable effect yields some additional information on the mode of action and the type of compounds that induce the effect. Details of each assay are discussed below. The curve fit is then performed in such a way that the maximum effect in equation 4 is an additional fitting parameter. Finally the umuC assay exhibited a linear concentrationeffect curve. At low effect levels the log-logistic concentration effect curve (Eq. 4) is congruent with a linear concentrationeffect curve. Unlike the other assays, where effects were compared to a reference compound, genotoxicity is expressed as the concentration that induces the threshold of induction defined by the EN ISO guideline (International Standard Organisation, 2000). This way of data evaluation needs further refinement in the future because it is not directly comparable to the results in the other bioassays and the effect concentration cannot be translated into TEQ. The detection limit of the bioassays was defined as three times the standard deviation of the response using the lowest concentration of the standard that induced an effect significantly different from the control.
concentrations (Villeneuve et al., 2000; Escher et al., 2008b). The equivalent concentration represents the concentration of the reference compound required to produce the same effect as the mixture of different compounds in the sample. Equivalent concentrations were calculated from concentration-effect curves of the reference compound and the samples. Both reference compounds and samples generally followed a sigmoidal log–concentration-effect curve (Eq. 4), which is equivalent to a linear function with respect to nonlogarithmic concentration at small effect levels. Sample extracts are normally composed of a mixture of unknown substances at unknown concentrations, so the concentration unit of the sample in the concentration effect curve is based on the Relative Enrichment Factor (REF) in units of [Lwater sample/Lbioassay]. Toxic equivalent concentrations (TEQ) were calculated as the ratio of EC50 values of the reference compound to the EC50 of the sample (Eq. 5). TEQ ¼
EC50 ðreference compoundÞ EC50 ðsampleÞ g 3 2 L g bioassay 5 4Lwater sample ¼ Lwater sample
ð5Þ
Lbioassay
Since the EC50(reference compound) is expressed in concentration units such as [g/Lbioassay] and the EC50(sample) in concentration units [Lwater sample/Lbioassay], representing the enrichment or dilution (REF) of the sample required to elicit the 50% effect, the toxic equivalent concentration is expressed in [g/Lwater sample]. In principle, any effect level can be used to derive TEQ, provided that the concentration-effect curves are parallel i.e. have the same slope, which is in fact one of the prerequisites for the validity of the toxic equivalency concept (Villeneuve et al., 2000). Therefore, if the effect of the sample in the bioassay does not reach 50%, toxic equivalency can be estimated using EC20 of the sample and EC20 of the corresponding reference compound.
2.6. Estimating equivalent concentrations in the samples
2.7.
For an easy-to-follow reporting of the effect, in all but one bioassay, the effects were reported in terms of toxic equivalent
To determine background levels of contamination associated with the extraction process as well as assessing any effect of
QA/QC
481
water research 44 (2010) 477–492
Table 3 – Evaluation of the concentration effect curves and expression of the bioanalytical results. Assay
Concentration-effect assessment
REF range
Reference compound; EC50 of the reference Positive control compound
Result expression
Detection limit
Baseline Toxicity – Top: 100 Bioluminescence Bottom: 0 inhibition in Slope: 1 Vibrio fischeri Neurotoxicity – AChE Phytotoxicity – Max-I-PAM
0.4–38
Virtual baseline toxicant; Phenola
12 mg/L; 130 mg/L
Baseline toxicity equivalent concentrations (baseline-TEQ)
20% inhibition of bioluminescence
0.04–115
Parathion
120 mg/l
0.3 mg/L
1.5–258
Diuron
16 mg/L
Estrogenicity – E-SCREEN
Top: w Bottom: w Slope: 1
0.0006–55 17ß-Estradiol (E2)
2 ng/L
Ah-Receptor – AhR-CAFLUX
Top: at least 100 and then w Bottom: w Slope: 1 Linear regression
0.004–55
2,3,7,8Tetrachlorodibenzodioxin (TCDD)
8 ng/L
Parathion equivalent concentrations (PTEQ) Diuron equivalent concentrations (DEQ) Estradiol equivalent concentrations (EEQ); Relative proliferative effect (RPE) TCDD equivalent concentrations (TCDDEQ)
2.2–110
(-S9) 2-amino anthracenea (þS9) 4-nitroquinolineN-oxidea
n.a.
Genotoxicity – Umu
0.01 mg/L
0.02 ng/L
0.04 ng/L
REF that elicits genotoxic Maximum REF effect with threshold achieved in the Induction Ratio of assayb 1.5
a Only used as positive control, not as reference compound. b Quantification limit; w variable; n.a. - not applicable.
the solvent, 2.0 L of MilliQ water were extracted identically to the samples and analysed in all bioassays as a procedural blank. For quality control and assurance purposes two replicates of randomly selected samples (2 sites per each sampling; for details see Supporting information Table S1–S6) were collected and analysed in all bioassays on a different day than the first replicate to evaluate repeatability of the SPE and the bioassays. Each sample extract was tested in the bioassays in duplicates or triplicates per run, depending on the assay. All individual data sets are summarised in the Supporting information Table S1–S6. Since the variability among the four sampling events was not higher than the variability between the sample replicates collected at the same time, we report in the main manuscript the average sd of four sample extracts at four different consecutive sampling events (n ¼ 4, whith exception of S2, where n ¼ 3). Analysis of variance (ANOVA) was used to test the differences among the average TEQ of samples S1-C5, including the blank, in comparison to S1.
3.
Results and discussion
3.1. Baseline toxicity – bioluminescence inhibition in vibrio fischeri The EC50 of phenol was 130 37 mg/L over 8 plates in 3 different repetitions of the assay, confirming the reproducibility and repeatability of the assay. While in all other assays, the reference compound served two purposes, as quality control and to define the reference for the TEQ, in the bioluminescence inhibition test, phenol was only used for quality control.
Toxicity of the sample was expressed in baseline toxicity equivalents (baseline-TEQ) derived from a baseline toxicity QSAR (quantitative structure-activity relationship) (Fig. 1A) using a virtual compound with log Kow of 3 and molecular weight of 300 g/mol as a reference, which equates to an EC50 of 12 mg/L (Escher et al., 2008b). The rationale behind this choice is the fact that there is not a singular positive control, as every chemical exhibits baseline toxicity. For specifically acting compounds the baseline effect is marginal and cannot be resolved if the single chemical is tested. However, in a mixture with a large number of chemicals with a variety of specific modes of action, as is the case in a wastewater sample, the baseline toxicity might actually dominate the overall mixture effect (Warne and Hawker, 1995). Thus, baseline toxicity will provide an integrative measure of the combination of chemicals that act together in the mixture and each chemical’s contribution is essentially weighted by its hydrophobicity only (Escher and Schwarzenbach, 2002). Fig. 1A depicts the relationship between hydrophobicity and effect concentration on a log-log plot and how the EC50 of a virtual baseline toxicant is derived (note that the hydrophobicity scale is based on liposome-water partitioning not octanol-water partitioning), therefore the virtual baseline toxicant with log Kow of 3 is situated slightly higher than 3 on the log-hydrophobicity scale. Samples were tested for baseline toxicity at 8 different concentrations after serial dilution 1:2, with the REF ranging from 0.4 to 38. All previously tested baseline toxicants exhibited a common slope of 1 (Escher et al., 2008b), and therefore the slope of the samples was fixed to 1, too. Effect concentrations (EC50) of the samples used to calculate the toxic equivalent concentrations were expressed in REFs,
482
water research 44 (2010) 477–492
Fig. 1 – Concentration-effect curves of the reference compounds and selected samples tested in bioluminescence inhibition assay (A, B), AChE inhibition assay (C, D) and I-PAM (E, F) assay. Bottom and top of each curve was fixed to 0 and 100, respectively. Quality of the fit expressed as R2 was >0.90, except for S7-effluent sample tested in AChE assay, where the R2 was 0.74. (A) Baseline toxicity QSAR for the bioluminescence inhibition test with Vibrio fischeri and derivation of the EC50 of the virtual baseline toxicant that is used as reference compound in this assay (data and QSAR from Escher et al. (2008b)). Error bars indicate SD.
representing the enrichment or dilution of the sample required to elicit the 50% effect, and are summarised in Table 4. The EC50 of the blank was 59, which means that if MilliQ water were enriched 59 times, the blank extract would elicit 50% bioluminescence inhibition. This EC50 was extrapolated from experimental data, where the highest measured REF was 38, which caused a 25% effect. It is not reasonable to enrich the samples to a higher REF because the enrichment of impurities of solvents and materials or contamination during the enrichment procedure would mask the measurable effect. The influent, denitrification and ozonation samples all exhibited EC50 in the range of 4–5.6, i.e. the samples had to be
enriched by approximately five times to see an effect. The concentration window for a visible effect is relatively narrow. A sample that is not enriched (REF ¼ 1) or even diluted would not give any visible response. The necessary enrichment is not the only reason for using SPE for sample preparation: a majority of matrix compounds are removed by SPE, such as salts and particulates. All organisms are very sensitive to variable electrolyte concentration. In this assay too little salt would decrease the bioluminescence of the marine bacterium (Escher et al., 2008a), and salting up is difficult because the salt content of water samples is unknown. In contrast to salts and particulate matter, we assume that dissolved organic carbon
483
water research 44 (2010) 477–492
Table 4 – Sample enrichment required to produce a 50% effect in each assay expressed as the average ± sd of four 24 h sampling events. ID
S1 S2 S3 S4 S5 S6 S7 C3 C4
C5
Site description
Blank Influent (Effluent from WWTP) Denitrification Pre-ozonation Coagulation/flocculation/DAFF Main ozonation Activated carbon filtration Effluent Biological sand filter fed with S4 water Biological activated carbon filter fed with S4 water Biological activated carbon filter fed with S5 water
Bioluminescence inhibition
AChE
I-PAM
EC50 (REF)
EC50 (REF)
EC50 (REF)
avg
sd
59 5.6 4.4 4.0 9.8 14 28 29 11
0.81 1.6 0.94 0.98 3.7 3.2 7.8 11 2.6
36
14
BDL (<20%)
2381
1393
45
22
BDL (<20%)
2380
3349
(DOC), in particular the low molecular weight fraction, is partially retained by SPE. Baseline-TEQ increased slightly from influent (2.3 mg/L) to denitrification (2.8 mg/L) and after pre–ozonation (3.2 mg/L). However the stepwise increase was not statistically significant (ANOVA, p ¼ 0.05). Significant decrease by more than a factor of two in TEQ was observed after coagulation/flocculation/DAFF (1.4 mg/L). Baseline-TEQ after the main ozonation, activated carbon filtration and in the effluent ranged
avg
sd
avg
BDL (<20%) 58 19 53 12 42 12 80 14 92 17 BDL (<20%) 1093 930 129 13
sd
BDL (<10%) 429 272 133 70 146 93 1290 986 1187 758 2569 2116 1573 962 3205 4956
between 0.48 and 0.89 mg/L, levels not significantly different from the blank. Sand filtration as a polishing step directly after coagulation/flocculation/DAFF did not significantly alter the TEQ, while biological activated carbon filtration decreased the TEQs to the levels not significantly different from the blank (Table 5). Overall, DOC concentration correlated remarkably well with the baseline-TEQ (Fig. 2). The bioluminescence inhibition test cannot differentiate between micropollutants and organic
Table 5 – Summary of the bioanalytical results representing the average±sd of 4 samplings (expressed as equivalent concentrations except umuC assay where the data are expressed as REF of the sample required to elicit genotoxic effect). ID
Site description
Bioluminescence inhibition Baseline-TEQ (mg/L) avg
Blank S1 Influent (Effluent from WWTP) S2 Denitrification S3 Pre-ozonation S4 Coagulation/flocculation/ DAFF S5 Main ozonation S6 Activated carbon filtration S7 Effluent C3 Biological sand filter fed with S4 water C4 Biological activated carbon filter fed with S4 water C5 Biological activated carbon filter fed with S5 water
sd
AChE PTEQ (mg/L)
I-PAM
E-SCREEN
AhR-CAFLUX
umuC -S9
c
DEQ (mg/L) EEQ (ng/L) TCDDEQ (ng/L) 1/EC IR1.5 (1/REF)
avg sd avg
sd
avg
sd
avg
sd
avg
sd
0.21 2.3
0.01 0.7
<0.3 <0.01 3.2 1.0 0.12 0.1
<0.02 6.0 2.1
0.055 0.087b
0.01 0.03
<0.01 0.21
0.1
2.9 3.2a 1.4a
0.5 0.7 0.5
3.3 4.2 2.1
0.8 0.30 1.3 0.34 0.4 0.05
0.1 0.2 0.04
8.7 3.5 9.8a
0.10b 0.11b 0.11b
0.003 0.01 0.04
0.16 0.17 0.08a
0.04 0.06 0.06
0.91a,b 0.51a,b 0.52a,b 1.2a,b
0.2 0.2 0.2 0.3
1.9a 0.5 0.03 <0.3 0.02 0.36a 0.4 0.02 1.3a 0.1 0.08
0.03 0.02 0.01 0.06
0.43a 0.18 <0.02 <0.06 0.64a 0.53
0.091b 0.10b 0.10b 0.10b
0.02 0.06 0.01 0.01
0.02a <0.01 <0.01 0.075a
0.01
0.38a,b
0.1
<0.3
0.02
0.01
<0.02
0.070b
0.01
<0.01
0.34a,b
0.1
<0.3
0.04
0.03
<0.02
0.077b
0.02
<0.01
2.7 1.3 1.3
a Significantly different from influent, P < 0.05 tested against S1 with ANOVA, Bonferroni’s post test. b Not significantly different from the blank; ANOVA, Bonferroni’s post test. c TCDDEQ after acid silica gel clean-up.
0.08
484
water research 44 (2010) 477–492
Fig. 2 – Relationship between DOC concentration and baseline-TEQ. The black diamonds are the data from this study (from Tables 1 and 5). The empty squares refer to a previous study performed in Switzerland (all data refer to secondary effluent) (Escher et al., in press). Error bars indicate SD.
matter that is co-extracted with SPE. Assimilable organic carbon may be taken up by microorganisms and contribute to bioluminescence inhibition, while higher molecular-weight fractions of organic matter may bind micropollutants, potentially causing a decrease in their bioavailability and thus decrease the response. Note also that organic micropollutants contribute to DOC in DOC measurements. Therefore it is virtually impossible to differentiate between assimilable organic carbon and organic micropollutants. The TEQ of the influent of the reclamation plant, which is the effluent of a conventional wastewater treatment plant (WWTP), was higher than of the secondary effluent of a WWTP in Switzerland (Escher et al., 2008a; Escher et al., in press). It must also be noted that the DOC levels were significantly lower in the Swiss WWTP effluent than in the present study. In fact, the baseline-TEQ of the Swiss study (Escher et al., in press) fall on the lower left corner of Fig. 2, but they fit the correlation. This observation supports the conclusion that DOC contributed to the effects measured in this bioassay.
3.2. Neurotoxicity – acetylcholinesterase inhibition assay The concentration-effect curve of the reference compound, parathion, followed a log-logistic function with the slope 1 and the EC50 of 120 32 mg/L (Fig. 1C), using a commercially available acetylcholinesterase enzyme from Electrophorus electricus. The EC50 of parathion in our work was higher than data reported in literature: 27 mg/L (Escher et al., in press) and 26 mg/L (Escher et al., 2008a) with the use of bovine acetylcholinesterase. The difference in the sensitivity of the assay might be due to a different origin of acetylcholinesterase enzyme. The samples were tested at 12 different concentrations after serial dilution 1:2 with the REF range from 0.04 to 115. The best-fit slope of the samples’ concentration-effect curve was 1 (Fig. 1D) as it was for the reference compound. Effective concentrations required to elicit 50% inhibition are summarised in Table 4. Influent samples had to be enriched 52 times to elicit 50% inhibition, in contrast to effluent, which would
require an enrichment of 1100 times to elicit the same effect. As discussed above, we did not apply such high enrichment factors but had to extrapolate the EC50. The effect of the blank and samples treated with activated carbon was less than 20%, defined as the detection limit of the assay. Therefore the effective concentrations were not extrapolated from the concentration-effect curves. Results of the acetylcholinesterase inhibition assay were expressed as parathion equivalent concentrations (PTEQ) and are summarised in Table 5. PTEQ of the samples showed firstly a slight increase after pre-ozonation to 4.2 mg/L in comparison with the influent PTEQ of 3.1 mg/L, and then a significant decrease to 1.9 mg/L after the main ozonation. Activated carbon filtration further reduced PTEQ below the detection limit (<0.3 mg/L). The PTEQ in the effluent (0.36 mg/L) was only slightly above the detection limit of the assay. Sand filtration did not alter the PTEQ of the coagulation/flocculation/DAFF effluent, while treatment with the biological activated carbon filters reduced the PTEQ to below detection limit (<0.3 mg/L).
3.3.
Phytotoxicity – PS II inhibition I-PAM assay
The best-fit slope of the concentration-effect curve of the reference compound diuron was 1 and the EC50 was 16 6 mg/L (Fig. 1E). This EC50 represents an average of 12 concentrationeffect curves from 3 different runs indicating excellent repeatability of the assay. Diuron EC50 reported in this study was in the same order of magnitude as earlier work (6.7 mg/L) using Phaeodactylum tricornutum (Muller et al., 2007) and as the 2 hour endpoint in the combined algae test with Pseudokirchneriella subcapitata of 3 mg/L (Escher et al., 2008b; Escher et al., in press). The samples were enriched up to a REF of 258. The influent samples showed some cytotoxicity at these high enrichments so the phytotoxicity could not be evaluated. Despite the high enrichment of samples, the specific phytotoxic effects were still moderate, not exceeding 35% of the maximum effect of diuron. The best fit of the slope fitted to equation 5 would be higher than 1 (Fig. 1F, broken line) but given that only one data point forced this fit, the slope was fixed to the slope of the reference compound as usual (slope 1, Fig. 1F dotted line). We have previously observed that the yield of photosynthesis (Y)
Fig. 3 – Relationship between baseline-TEQ and PTEQ (triangles) or DEQ (diamonds) for the three samples influent, denitrification and pre-ozonation (S1 to S3). Error bars indicate SD.
water research 44 (2010) 477–492
485
Fig. 4 – Concentration-effect curves of selected samples and corresponding standards tested in E–SCREEN (A, B) and AhRCAFLUX (C, D) assay. The slope of the curves was fixed to 1. Bottom of the curves was an adjustable parameter in both assays. The top of the curve was adjustable in the E–SCREEN assay, while in AhR-CAFLUX, the top was set to the level of at least 100% of the TCDD maximal effect. Quality of the fit expressed as R2 ranged for selected samples and reference compounds from 0.96 to 0.99, except for Blank and S7-effluent samples tested in E-SCREEN assay, where the R2 was 0.3 and 0.4, respectively due to a very low response of both samples. Error bars indicate SD.
measurement is somewhat compromised, when several nonspecifically acting chemicals in the mixture mask the phytotoxic effect (Escher et al., 2008a). In fact, when single baseline toxicants (Escher et al., 2008b) were tested for the inhibition of photosynthesis yield they showed a response when measuring Y after 2 hours incubation, albeit at higher concentrations than for the corresponding growth inhibition endpoint, and the curve was steeper with a slope of 1.9. Since it is impossible to quantitatively differentiate between the contributions from phytotoxicity and cytotoxicity (unless cytotoxicity is so overwhelming that it makes the photosynthesis yield measurements impossible), we used the slope of the reference compound diuron to fit the samples even though the quality of fit was sub-optimal. The phytotoxic response of the samples after 2 hours of incubation expressed as diuron equivalent concentration DEQ is summarised in Table 5. Denitrification significantly increased the DEQ (0.29 mg/L) in comparison with the influent DEQ of 0.11 mg/L. The pre-ozonation step itself had no significant effect (0.34 mg/L). After coagulation/flocculation/DAFF, the DEQ decreased by 85% to 0.05 mg/L. DEQ decreased in small steps after the main ozonation and activated carbon filtration stages and in the effluent samples. No significant additional decrease in DEQ was observed after sand filter or biological activated carbon filters. All DEQ after coagulation/flocculation/DAFF are close to the detection limit and are based on
a single data point, therefore should be interpreted with some caution. The initial slight but not statistically significant increase in TEQ from sample S1 to S3 is unexpected at first sight. It is interesting to note, however, that this small increase was common for baseline toxicity and slightly less clear also for neurotoxicity (Figs. 2 and 3). The increase in baseline toxicity was explained by an increase of DOC, which is presumably partially co-extracted at low pH. The small increase in DEQ could also be caused by baseline toxicants interfering with the measurement of the photosynthesis yield. We have not observed such an effect in our previous work (Escher et al., 2008a; Escher et al., in press). However, we have never encountered DOC concentrations exceeding 10 mg/L in any of our previous work. It is unclear how much of the DOC is coextracted with the micropollutants by SPE. The effect of DOC needs to be investigated in future work but is beyond the scope of the present study. Possibly, extraction at a higher pH might reduce the fraction of co-extracted DOC due to the higher negative charge of DOC under these conditions. Previous work on comparing herbicides concentrations with DEQ often showed a relatively good correlation and a substantial fraction (> 50% of the response) could be explained by the measured herbicides (Bengtson Nash et al., 2005b; Bengtson Nash et al., 2006; Escher et al., 2006; Muller et al., 2008). There were some cases, such as samples collected
486
water research 44 (2010) 477–492
Table 6 – Sample enrichment required to produce a defined effect in each assay: 50% effect in E–SCREEN assay and 20% of maximal TCDD effect in AhR-CAFLUX assay. Results represent the average ± sd of 4 samplings. ID
Site description
E-SCREEN RPE
S1 S2 S3 S4 S5 S6 S7 C3 C4
C5
Blank Influent (Effluent from WWTP) Denitrification Pre-ozonation Coagulation/flocculation/DAFF Main ozonation Activated carbon filtration Effluent Biological sand filter fed with S4 water Biological activated carbon filter fed with S4 water Biological activated carbon filter fed with S5 water
Mode of Action
avg
sd
0.05 0.85 0.76 0.76 1.0 0.92 0.19 0.27 0.65
– 0.2 0.1 0.2 0.2 0.2 0.1 0.1 0.2
Not estrogenic Full agonist Partial agonist Partial agonist Full agonist Full agonist Not estrogenic Not estrogenic Partial agonist
0.12
0.06
0.16
0.1
EC50 (REF) avg
After Clean-up EC20 TCDD (REF)
sd
avg
sd
BDL (<20%) 0.45 0.1 0.32 0.1 0.79 0.2 0.27 0.06 6.7 2 BDL (<20%) BQL (<50%) 3.1 1
13 1.6 1.4 1.7 1.6 3.1 4.4 4.4 1.7
1 0.6 0.4 0.3 0.1 0.6 2 0.9 0.3
28 19 14 14 16 17 20 15 16
0.4 6 2 1 6 3 12 2 0.9
Not estrogenic
BDL (<20%)
6.1
1
22
3
Not estrogenic
BDL (<20%)
8.2
2
21
5
Estrogenic activity: E-SCREEN assay
In the E-SCREEN, bottom and top of the concentration-effect curve (Eq. 4), both adjustable parameters, refer to the minimum or maximum stimulatory response, i.e., proliferation of the estrogen-dependent cells in E–SCREEN. The slopes of the concentration-effect curves of the reference compound, 17ß-estradiol, and the samples were fixed to 1 (Figs. 4A and B). The maximum of the curve yields information on the mode of action, thus being able to distinguish among full agonist,
sd
Before Clean-up EC20 TCDD (REF) avg
from the Thames River, where the biological DEQ were much higher than the chemically analysed DEQ (Bengtson Nash et al., 2006). The authors suggested this difference was due to the presence of additional herbicides that were not analysed with the analytical chemical methodology, but given the new evidence it is also possible that additional cytotoxicity by nonspecifically acting micropollutants and DOC could have contributed to the discrepancy.
3.4.
AhR-CAFLUX
partial agonist and no estrogenic effect, a parameter called relative proliferative effect (RPE). The reference compound 17ß-estradiol had an EC50 of 2.4 ng/L, which is in good agreement with previous work of 2.2 ng/L (Ko¨rner et al., 1999). EC50 of the estradiol in the E–SCREEN assay was comparable with the EC50 in the ER– CALUX assay of 2.3 ng/L (Legler et al., 2002), indicating the similar sensitivity of both assays. The detection limit of the E– SCREEN assay was defined as 20% of the maximal estradiol effect. The detection limit for the whole method was calculated based on the highest relative enrichment of the samples, providing the sample was not cytotoxic, resulting in a detection limit of 0.02 ng/L. Samples collected in the influent and after denitrification were cytotoxic at the highest concentrations tested, therefore the two highest concentrations were excluded and the detection limit of the whole method for those samples was 0.2 ng/L. The assay is still more sensitive than chemical analysis of the estrogenic compounds. Nine concentrations of the sample extract serially diluted 1:4 were tested in the E–SCREEN assay, resulting in REF from 0.0006
Fig. 5 – Concentration-effect assessment of selected samples tested in umuC assay without (A) or with (B) metabolic activation. Samples follow a linear function with respect to non-logarithmic concentration. Quality of the fit expressed as R2 ranged for selected samples from 0.80 to 0.99, except for Blank and S7-effluent samples. Error bars indicate SD.
water research 44 (2010) 477–492
Table 7 – Sample enrichment required to produce a genotoxic effect [ Induction Ratio of 1.5 (IR1.5) in umuC assay without and with metabolic activation. Results represent the average ± sd of 4 samplings. ID
Site description
umuC -S9 EC
IR1.5
avg
S1 S2 S3 S4 S5 S6 S7 C3 C4
C5
Blank Influent (Effluent from WWTP) Denitrification Pre-ozonation Coagulation/flocculation/DAFF Main ozonation Activated carbon filtration Effluent Biological sand filter fed with S4 water Biological activated carbon filter fed with S4 water Biological activated carbon filter fed with S5 water
>78 5.2 6.9 6.3 15 78 >84 >69 24
umuC þ S9
(REF)
EC
sd
avg
2 2 2 7 20
13
IR1.5
>78 17 22 18 28 >87 >84 >73 34
>80
>80
>79
>79
(REF) sd
3 3 3 8
16
(1667 times dilution) to 55 (55 times enrichment of the original water sample). Sample REF required to produce a 50% effect (EC50) in the E–SCREEN assay ranged from 0.27 (equivalent to 4 times dilution of the original water sample) to 6.7 (equivalent to 6.7 times enrichment of the original water sample) (Table 3). The wide range of REF windows allows one to monitor a wide range of samples collected throughout the treatment chain. The estrogenicity of the samples was quantified with two parameters: the relative proliferative effect (RPE) and the estradiol equivalent concentration (EEQ). RPE of the samples, representing the ratio of the maximum proliferation induced by test samples compared to 17b-estradiol (also referred to as relative efficacy), are summarised in Table 6. The samples from the influent up to ozonation were fully or partially agonistic, while the remainder of the samples were not classified estrogenic and no EEQ could be derived for these samples. When a sample was classified as either a partial or full agonist, an estradiol equivalent concentration (EEQ) was calculated from the EC50. Influent to the reclamation plant (representing the effluent of the WWTP) showed full agonistic activity (RPE > 0.8) with the estradiol equivalent concentration EEQ of 6 ng/L (Table 5). This was comparable with the EEQs reported for WWTP effluent in the literature: <1–7.8 ng/L in Germany (E–SCREEN assay, (Ko¨rner et al., 2001)), <1 to >10 over a wide range of different WWTPs and 0.8–5.9 ng/L in one given WWTP at various samplings during three years in Switzerland (YES assay, (Rutishauser et al., 2004; Escher et al., in press)), <1–4.2 ng/L in Australia and New Zealand (estrogen receptor binding assay, (Leusch et al., 2006)), <1–16 ng/L in the Netherlands (ER–CALUX, (Murk et al., 2002)), 1–67 ng/L in Australia (E–SCREEN assay (Tan et al., 2007)). Samples collected after denitrification and pre-ozonation (S2 and S3, respectively) showed partial agonistic activity (RPE of 0.5–0.8), and the EEQs were not significantly different from the influent. Surprisingly, EEQ of the sample collected after coagulation/flocculation/DAFF treatment was significantly
487
higher (EEQ of 9.8 ng/L) than in influent and the sample again showed full agonistic activity. Full agonistic activity was also detected after the main ozonation (RPE > 0.8); however, the EEQ was significantly decreased to 0.43 ng/L. The remarkable removal efficiency of ozonation on estrogenicity is well established (Lee et al., 2008) and in a recent full-scale ozonation in a WWTP in Switzerland the ozonation step showed additional >95% removal efficiency (provided the influent of the ozonation reactor had EEQ > 2 ng/L) and the resulting effluent of the ozonation contained 0.6 ng/L EEQ (Escher et al., in press). Estrogenicity was further markedly altered after activated carbon filtration. Estrogenic effect of the sample treated with activated carbon (S6) was less than 20% of the estradiol effect (RPE < 0.2), which corresponds to the detection limit of the assay, therefore the EEQ was less than 0.02 ng/L. Estrogenicity (RPE) of the effluent sample was slightly increased, but did not reach 50% effect of estradiol, therefore the sample was classified as not estrogenic and the EEQ was not quantified. The estrogenic effect was below quantification limit. Taking into account the REF of the sample in this assay, the EEQ of the effluent was less than 0.06 ng/L. Besides the main treatment chain, additional treatment processes were assessed in this study: sand filtration of the coagulation/flocculation/DAFF effluent (C3) and biological activated carbon filtration of the coagulation/flocculation/ DAFF effluent (C4) and main ozonation effluent (C5). Sand filtration significantly decreased the EEQ of the coagulation/ flocculation/DAFF effluent from 9.8 ng/L to 0.63 ng/L. Biological activated carbon filtration decreased the estrogenic effect of both effluents to below detection limit (<0.02 ng/L). The same assay was also applied to detect estrogenicity in the raw sewage sample from Caboolture WWTP with high EEQ ranging from 68 to 91 ng/L in 3 different samples collected on different days (unpublished results). The levels were comparable with EEQ recently reported in the raw sewage in the Brisbane area using the same bioassay, with up to 74 ng/L EEQ (GWRC, 2008). The measured EEQ in the raw sewage was also similar to the levels reported in other studies, <4–185 ng/L in Australia and New Zealand (estrogen receptor binding assay, (Leusch et al., 2006)), up to 120 ng/L in the Netherlands (ER–CALUX, (Murk et al., 2002)) and 58–70 ng/L in Germany (E–SCREEN, (Ko¨rner et al., 2000)).
3.5.
Ah-receptor response: AhR-CAFLUX assay
In the AhR-CAFLUX assay, the bottom and top of the curve refer to the minimum and maximum induction of fluorescence under the control of AhR. Sample responses in AhRCAFLUX assay measured in relative fluorescence units (RFU) were converted to a percentage of maximal response of reference compound 2,3,7,8-tetrachlorodibenzo-p-dioxin (% TCDD max response). If the effect of the sample was higher than maximal TCDD effect, the top of the curve was adjustable. However, if the sample did not reach 100% TCDD effect, the top of the curve was fixed to 100. The slope of the concentration-effect curve of the reference compound TCDD and the samples was fixed to 1 (Figs. 4C and D). The EC50 of the reference compound TCDD ranged between 6.3 and 9.2 ng/L over 5 different runs, confirming the
488
water research 44 (2010) 477–492
reproducibility and repeatability of the assay and is in good agreement with the previous levels from the literature of 5.9 ng/L (Nagy et al., 2002) and 3.9 ng/L (Zhao and Denison, 2004). TCDD was chosen as the reference compound as it is one of the most potent agonists for this assay. However, high affinity ligands for the AhR also include dibenzofurans and biphenyls as well as a variety of polycyclic aromatic hydrocarbons, benzoflavones and other chemicals (Denison and Nagy, 2003). To assess the contribution of chemicals other than dioxins, furans and dioxin-like PCBs to the potency estimates, sulfuric acid silica gel clean-up was performed in this study, which removed most organic chemicals (e.g. PAHs) except persistent chemicals such as dioxins, furans and PCBs. Original sample extracts, as well as the extracts cleaned with sulfuric acid silica gel were tested with the AhR-CAFLUX assay at 5 different concentrations after serial dilution 1:10, with the REF ranging from 55 to 0.004. The response of many samples, especially after clean-up, did not reach 50% TCDD response. Therefore effective concentrations of the samples required to elicit 20% effect of TCDD (EC20 TCDD) were interpolated from concentration-effect curves (Table 6). The EC20 TCDD of the samples ranged from 1.4 to 8.2. The effective concentration of the same samples was significantly altered after clean up and ranged from 14 to 22, i.e. samples after acid silica gel clean up had to be enriched approximately 3–10 times more to elicit the same effect. The response in the AhR-CAFLUX assay was expressed as the 2,3,7,8-TCDD-equivalent concentration (TCDDEQ). TCDDEQ was significantly decreased by the main ozonation to 0.44 ng/L in comparison with the TCDDEQ of 0.83 ng/L in the influent. Sand filtration did not change the TCDDEQ of the coagulation/flocculation/DAFF effluent, while treatment with the biological activated carbon filters reduced the TCDD equivalent concentration to 0.33 ng/L a level not significantly different from the blank. The TCDDEQ of the samples after the sulfuric acid silica gel clean-up were reduced to levels not significantly different from the blank (Table 5). These results indicate that the observed effect in the AhR-CAFLUX assay prior to the sulfuric acid silica gel clean-up step may be attributed to chemicals other than dioxins, furans and PCBs.
3.6.
Genotoxicity-UmuC assay
The umuC genotoxicity assay can detect both cytotoxic and genotoxic effects. The extracts were tested both with (þS9) and without (S9) exogenous metabolic activation. Response on this assay is determined as an induction ratio (IR) (Fig. 5). An IR 1.5 is considered genotoxic, providing the sample is not cytotoxic (growth < 0.5). For samples that demonstrate significant genotoxic response, the effect concentrations ECIR 1.5 were derived from the linear concentration-effect curve as depicted in Fig. 5. Effect concentrations are in dimensionless units of REF and represent how many times the samples must be concentrated or diluted to elicit a threshold IR of 1.5 in the assay (Table 7). Results are expressed as 1/ECIR 1.5 therefore a higher number represents a higher genotoxic effect (Table 5). Unfortunately no TEQ could be derived for this endpoint but in theory this should be possible and in the future we will work on implementing the TEQ concept for this assay. This is somehow
problematic because of the frequently observed cytotoxicity of the samples, which can only be partially accounted for. Overall, the umuC test without metabolic activation gave higher responses than the test with metabolic activation indicating that some of the micropollutants that are responsible for genotoxicity are detoxified by the liver enzyme fraction S9. As in all other bioassays, the denitrification and preozonation did not affect the activity in the umuC test but coagulation/flocculation/DAFF decreased the genotoxicity without S9 by 58% and that with S9 by 35%. Subsequent main ozonation completely removed the effect for the test with prior S9 treatment and drastically reduced it in the absence of S9.
4.
Conclusion
A similar test battery has been previously used for various applications in wastewater treatment efficiency and water quality assessment. To apply the test battery to a treatment train with very subtle steps and for relatively pure water was a challenge. This challenge was met by the following measures: A comprehensive framework was developed that centres around the bioassays but links to sample preparation with SPE prior to testing and to the data evaluation scheme. Quality control was implemented and all effects were related to reference compounds, thereby correcting for slight day-to-day variability of the test results. The use of SPE allows a relatively non-specific extraction of water samples to concentrate a wide range of relevant pollutants to levels where they can exhibit a measurable response, while matrix components such as salts and metals are removed from the sample during SPE. The detection window of the bioassays is adjustable. For the application in the present study the concentration range of the sample in the bioassay ranged between 1700 times dilution and 250 times enrichment of the sample. The detection window depends on the constitution of the sample and the type of bioassay and is only limited by the effect of the blank at very high enrichment and nonspecific cytotoxicity of the sample for each of the assays. By dynamically changing the relative enrichment window of the sample throughout the study one can adapt the various bioassays to a wide variety of samples and results are robust and reproducible. Results are reported as toxic equivalent concentrations, a format that allows for comparison to results from chemical analysis. This is also a key feature of the bioanalytical toolsthey are no direct indicators for environmental health but they are indicators of the presence of mixtures of chemicals, accounting for the mixture toxicity of the unidentified chemicals in an environmental sample. To report bioanalytical results as toxic equivalent concentrations is better suited for hazard assessment and extrapolation than the commonly reported % effect of the water sample in a given bioassay because it is too difficult to extrapolate effects while it is possible to compare effect concentrations of chemicals across different level of biological organisation. So for example, if a sample elicits 0.5 mg/L DEQ one can compare this value with the environmental quality standard for
water research 44 (2010) 477–492
diuron to estimate what potential long-term effects could be expected from such an exposure. The study also highlighted some difficulties that remain and some lessons are to be learned. One of the weak points of the approach is the sample preparation with SPE. However, this is a common problem for chemical analysis and can only be circumvented if there are recovery standards for every single compound to be analysed. Such an approach is not possible for bioassays because spiked chemicals would add to the mixture toxicity. A shortcoming of the SPE method that was noticed here for the first time is the co-extraction of an unknown fraction of dissolved organic matter, an issue that requires further attention in the future. In addition, volatile compounds will evaporate during SPE but they would also cross contaminate in a 96 well plate, so it is a procedural advantage to have removed the volatile chemicals prior to toxicity testing. It must also be noted that oxidation by-products formed during ozonation are often more hydrophilic than their parent compound and might therefore not be extracted efficiently by SPE. In addition, one must be aware that mixture effects are assessed when applying bioanalytical tools to complex samples and there is always the problem that the subtle effect of newly formed by-products might be shielded by other micropollutants present in the mixture. Some of the oxidation byproducts stemming from ozonation are very reactive, for example aldehydes and other electrophiles formed from ß-blockers during ozonation (Benner and Ternes, 2009), others are very hydrophilic organics such as NDMA (N-nitrosodimethylamine) or inorganics such as bromate (Hollender et al., in press) none of these would be caught with the bioanalytical tools used here but require targeted chemical analysis for quantification. However, ozonation often only mildly oxidizes organic micropollutants and many resulting transformation products are likely to retain some of their toxic potential (Escher et al., 2008c) or lose their specific receptor-binding effect but retain non-specific baseline toxicity (Lee et al., 2008) and will thus contribute to the mixture toxicity in the bioanalytical tools applied in the present study. While the results are discussed in more detail and compared to chemical analysis in the accompanying paper (Reungoat et al., 2010), some general conclusions can be drawn: The influent of the water reclamation plant had similar levels of effects for the different endpoints as we have encountered in previous studies for wastewater after secondary treatment (Escher et al., 2008a; Escher et al., in press). The effluent of the water reclamation plant had strongly reduced effect levels compared to the influent, reducing the baseline-TEQ by 79%, the DEQ by 76%, the estrogenicity to below the detection limit of 0.02 ng/L EEQ, the PTEQ by 88% and the genotoxicity to below the detection limit. Only the AhR-CAFLUX assay did not yield any information on the removal efficiency of the treatment because the level of persistent inducers of the arylhydrocarbon receptor remained close to the detection limit. This was not too unexpected given that these are typically very hydrophobic compounds, which are usually adsorbed and not present in
489
the water column. In contrast, the non-persistent inducers of the AhR were clearly reduced by a factor of two during ozonation. The most efficient steps with respect to removal of the toxic responses in all selected assays proved to be the coagulation/flocculation/DAFF, the main ozonation and (biological) activated carbon filtration steps. This paper demonstrates the first comprehensive application of the bioanalytical tool framework on an advanced water treatment plant. One future goal would be to automate the assays to a degree that they can be used routinely by commercial laboratories or water companies themselves. At this stage, there remain some scientific questions to be resolved, including (1) the sample treatment and dosing, (2) extension of the battery and (3) data interpretation. (1) SPE and dosing by solvent spiking will always change the composition of an undefined mixture. A combination of passive sampling to a biomimetic polymer and passive dosing directly from the polymer into the bioassay might be a way to overcome this limitation. (2) A battery of bioassays will never be comprehensive because of the myriads of receptors and regulatory pathways in an organism. Nevertheless the battery would benefit from inclusion of additional endpoints, such as chemically induced immunosuppression or even other examples of mode-of-action categories already included, e.g. other endocrine endpoints. (3) At this stage we can say how much of a toxic equivalent concentrations is removed by a certain treatment step and we can give comparative results but we will never be able to deduce ecological or health consequences from that. This discussion must be lead independently. One solution would be to compare the TEQ of the mixture with the defined Environmental Water Quality Criteria (EQS) for single substances of the Water Framework Directive (European Commission, 2006) or guideline values for drinking water (e.g. NHMRC–NRMMC, 2004) but not for all reference compounds are such EQS and guideline values available.
Acknowledgements This work was co-funded by the Urban Water Security Research Alliance under the Enhanced Treatment Project, the CRC Water Quality and Treatment Project No. 2.0.2.4.1.1. Dissolved Organic Carbon Removal by Biological Treatment and by EnTox. The National Research Centre for Environmental Toxicology (EnTox) is a joint venture of the University of Queensland and Queensland Health Forensic and Scientific Services (QHFSS). The authors acknowledge the Moreton Bay Water for access to the South Caboolture Water Reclamation Plant; Ray McSweeny and Paul McDonnell (Moreton Bay Water) for their help during sampling and Chris Pipe-Martin (Ecowise) for providing information on the South Caboolture Water Reclamation Plant. The authors are grateful to Renee Muller (Gold Coast Water), Fred Leusch (Griffith University/EnTox), Karen Kennedy (EnTox) and Pam Quayle (EnTox) for constructive discussions and for help with the data analysis method, Marita Goodwin (EnTox) for running various assays and for helpful comments Christina Carswell (QHFSS) for extracting the
490
water research 44 (2010) 477–492
samples and Chris Paxman (EnTox) for technical assistance. We would like to thank also Michael Denison (University of California Davis, USA) for providing the H4G1.1c2 cells, Georg Reifferscheid (German Federal Institute of Hydrology, Germany) for providing the bacteria Salmonella typhimurium TA1535/pSK1002, and Ana Soto (Tufts University, USA) for providing the MCF7-BOS cells.
Appendix. Supporting information Supporting information related to this article can be found at doi:10.1016/j.watres.2009.09.025.
references
Bengtson Nash, S.M., Schreiber, U., Ralph, P.J., Mueller, J.F., 2005a. The combined SPE: ToxY-PAM phytotoxicity assay; application and appraisal of a novel biomonitoring tool for the aquatic environment. Biosensors and Bioelectronics 20 (7), 1443–1451. Bengtson Nash, S.M., McMahon, K., Eaglesham, G., Mueller, J.F., 2005b. Application of a novel phytotoxicity assay for the detection of herbicides in Hervey Bay and the Great Sandy Straits. Marine Pollution Bulletin 51, 351–360. Bengtson Nash, S.M., Goddard, J., Muller, J.F., 2006. Phytotoxicity of surface waters of the Thames and Brisbane river estuaries: a combined chemical analysis and bioassay approach for the comparison of two systems. Biosensors and Bioelectronics 21, 2086–2093. Benner, J., Ternes, T.A., 2009. Ozonation of propranolol: formation of oxidation products. Environmental Science & Technology 43, 5086–5093. Cao, N., Yang, M., Zhang, Y., Hu, J., Ike, M., Hirotsuji, J., Matsui, H., Inoue, D., Sei, K., 2009. Evaluation of wastewater reclamation technologies based on in vitro and in vivo bioassays. Ecotoxicology and Environmental Safety 407 (5), 1588–1597. Chapman, P.M., 2000. Whole effluent toxicity testing – usefulness, level of protection, and risk assessment. Environmental Toxicology and Chemistry 19 (1), 3–13. Denison, M.S., Nagy, S.R., 2003. Activation of the aryl hydrocarbon receptor by structurally diverse exogenous and endogenous chemicals. Annual Reviews in Pharmacology and Toxicology 43, 309–334. Deutsche Norm, 1995. In: DIN, N.W.N.i. (Ed.), Suborganismische Verfahren: Teil 1: Bestimmung von CholinesteraseHemmenden Organophosphat- und Carbamat-Pestiziden (Cholinesterase-Hemmtest) DIN 38415-1: 1995-02, Berlin, Germany. Ellman, G.L., Courtney, K.D., Andres, V., Featherstone, R.M., 1961. A new and rapid colorimetric determination of acetylcholinesterase activity. Biochemical Pharmacology 7, 88–95. Escher, B., Schwarzenbach, R.P., 2002. Mechanistic studies on baseline toxicity and uncoupling as a basis for modeling internal lethal concentrations in aquatic organisms. Aquatic Sciences 64, 20–35. Escher, B.I., Bramaz, N., Maurer, M., Richter, M., Sutter, D., von Ka¨nel, C., Zschokke, M., 2005. Screening test battery for pharmaceuticals in urine and wastewater. Environmental Toxicology and Chemistry 24 (3), 750–758. Escher, B.I., Quayle, P., Muller, R., Schreiber, U., Mueller, J., 2006. Passive sampling of herbicides combined with effect analysis in algae using a novel high-throughput phytotoxicity assay
(Maxi-Imaging- PAM). Journal Environmental Monitoring 8, 456–464. Escher, B.I., Bramaz, N., Quayle, P., Rutishauser, S., Vermeirssen, E., 2008a. Monitoring of the ecotoxicological hazard potential by polar organic micropollutants in sewage treatment plants and surface waters using a mode-of-action based test battery. Journal Environmental Monitoring 10, 622–631. Escher, B.I., Bramaz, N., Mueller, J.F., Quayle, P., Rutishauser, S., Vermeirssen, E., 2008b. Toxic equivalent concentrations (TEQs) for baseline toxicity and specific modes of action as a tool to improve evaluation of ecotoxicity tests on environmental samples. Journal Environmental Monitoring 10, 612–621. Escher, B.I., Baumgartner, R., Lienert, J., Fenner, K., 2008c. Predicting the ecotoxicological effects of transformation products. SpringerLink online. In: Boxall, A. (Ed.), Handbook of Environmental Chemistry – Degradation of Synthetic Chemicals in the Environment. Springer, Heidelberg, Germany. doi:10.1007/6998_2_015. http://www.springerlink. com/content/48r2t364q183q654/?p ¼ ec460152bfd 14906adc89378463e4997&pi ¼ 5 published Wednesday, April 30, 2008. Escher, B.I., Bramaz, N. and Ort, C., 2009. Monitoring the treatment efficiency of a full scale ozonation on a sewage treatment plant with a mode-of-action based test battery. Journal Environmental Monitoring, doi: 10.1039/ b907093a. European Commission, 2006. Proposal for a Directive of the European Parliament and the Council on Environmental Quality Standards in the Field of Water Policy and Amending Directive 2000/60/EC COM(2006) 398 Final. COM(2006) 398 Final. European Parliament, Committee on the Environment, Public Health and Food Safety, Brussels. Farre´, M., Martinez, E., Hernando, M., Fernandez-Alba, A., Fritz, J.I., Unruh, E., Mihail, O., Sakkas, V., Morbey, A., Albanis, T.A., Brito, F., Hansen, P.D., Barcelo, D., 2006. European ring exercise on water toxicity using different bioluminescence inhibition tests based on Vibrio fischeri, in support to the implementation of the water framework directive. Talanta, 323–333. GWRC, 2008. In: Leusch, F.D.L. (Ed.), Tools to Detect Estrogenic Activity in Environmental Waters. Global Water Research Coalition, London, UK, p. 84. Hamers, T., Molin, K.R.J., Koeman, J.H., Murk, A.J., 2000. A smallvolume bioassay for quantification of the esterase inhibiting potency of mixtures of organophosphates and carbamate insceticides in rainwater: development and optimization. Toxicological Sciences 58, 60–67. Hollender, J., Zimmermann, S.G., Koepke, S., Krauss, M., McArdell, C.S., Ort, C., Singer, H., von Gunten, U., Siegrist, H., 2009. Elimination of organic micropollutants in a municipal wastewater treatment plant upgraded with a full-scale postozonation followed by sand filtration. Environmental Science & Technology, Articles ASAP (Article), in press. doi: 10.1021/ es9014629. Huber, M.M., Ternes, T.A., von Gunten, U., 2004. Removal of estrogenic activity and formation of oxidation products during ozonation of 17 alpha-ethinylestradiol. Environmental Science & Technology 38, 5177–5186. Huber, M.M., Go¨bel, A., Joss, A., Hermann, N., Lo¨ffler, D., McArdell, C.S., Ried, A., Siegrist, H., Ternes, T.A., von Gunten, U., 2005. Oxidation of pharmaceuticals during ozonation of municipal wastewater effluents: a pilot study. Environmental Science & Technology 39, 4290–4299. International Standard Organisation 1998 Water Quality– determination of the inhibitory effect of water samples on the light emission of Vibrio fischeri (luminescent bacteria test), EN ISO 11348-3, Geneva, Switzerland.
water research 44 (2010) 477–492
International Standard Organisation 2000 Water quality – Determination of the genotoxicity of water and waste water using the umu-test, EN ISO 13829(2000) and 38415-3 (1996), Geneva, Switzerland. Johnson, B.T., 2005. Microtox acute toxicity test. In: Blaise, C., Ferard, J.-F. (Eds.), Small-Scale Freshwater Toxicity Investigations. Toxicity Test Methods, vol. 1. Springer. Joss, A., Keller, E., Alder, A.C., Gobel, A., McArdell, C.S., Ternes, T., Siegrist, H., 2005. Removal of pharmaceuticals and fragrances in biological wastewater treatment. Water Research 39 (14), 3139–3152. Ko¨rner, W., Hanf, V., Schuller, W., Kempter, C., Metzger, J., Hagenmaier, H., 1999. Development of a sensitive E-screen assay for quantitative analysis of estrogenic activity in municipal sewage plant effluents. Ecotoxicology and Environmental Safety 225 (1–2), 33–48. Ko¨rner, W., Bolz, U., Sussmuth, W., Hiller, G., Schuller, W., Hanf, V., Hagenmaier, H., 2000. Input/output balance of estrogenic active compounds in a major municipal sewage plant in Germany. Chemosphere 40 (9–11), 1131–1142. Ko¨rner, W., Spengler, P., Bolz, U., Schuller, W., Hanf, V., Metzger, J.W., 2001. Substances with estrogenic activity in effluents of sewage treatment plants in southwestern Germany. 2. Biological analysis. Environmental Toxicology and Chemistry 20 (10), 2142–2151. Lee, Y., Escher, B.I., von Gunten, U., 2008. Efficient removal of estrogenic activity during oxidative treatment of waters containing steroid estrogens. Environmental Science & Technology 42 (17), 6333–6339. Legler, J., Zeinstra, L.M., Schuitemaker, F., Lanser, P.H., Bogerd, J., Brouwer, A., Vethaak, A.D., De Voogt, P., Murk, A.J., Van der Burg, B., 2002. Comparison of in vivo and in vitro reporter gene assays for short-term screening of estrogenic activity. Environmental Science & Technology 36 (20), 4410–4415. Leusch, F.D., Chapman, H.F., van den Heuvel, M.R., Tan, B.L., Gooneratne, S.R., Tremblay, L.A., 2006. Bioassay-derived androgenic and estrogenic activity in municipal sewage in Australia and New Zealand. Ecotoxicology and Environmental Safety 65 (3), 403–411. Muller, R., Tang, J., Thier, R., Mueller, J., 2007. Combining passive sampling and toxicity testing for evaluation of mixtures of polar organic chemicals in sewage treatment plant effluents. Journal Environmental Monitoring 9, 104–109. Muller, R. 2008 Passive sampling and bioanalysis of organic chemicals in surface water. Ph.D. thesis. University of Queensland, St. Lucia, Qld, Australia, 201pp. Accessible at http://library.uq.edu.au/search~S7?/XRenee þ Muller&searchscope ¼ 7&SORT ¼ AZ/XRenee þ Muller&searchscope ¼ 7&SORT ¼ AZ&extended ¼ 0&SUBKEY ¼ Renee%20Muller/21%22C23%22C23%22CB/frameset&FF ¼ XRenee þ Muller&searchscope ¼ 27&SORT ¼ AZ&21%22C21%22C Muller, R., Schreiber, U., Escher, B.I., Quayle, P., Bengtson Nash, S., Mueller, J., 2008. Rapid exposure assessment of PSII herbicides in surface water using a novel chlorophyll a fluorescence imaging assay. Science Total Environment 401, 51–59. Murk, A.J., Legler, J., van Lipzig, M.M.H., Meerman, J.H.N., Belfroid, A.C., Spenkelink, A., van der Burg, B., Rijs, G.B.J., Vethaak, D., 2002. Detection of estrogenic potency in wastewater and surface water with three in vitro bioassays. Environmental Toxicology and Chemistry 21 (1), 16–23. Nagy, S., Sanborn, J., Hammock, B., Denison, M., 2002. Development of a green fluorescent protein-based cell bioassay for the rapid and inexpensive detection and characterization of Ah receptor agonists. Toxicological Sciences 65, 200–210. NHMRC–NRMMC, 2004. Australian Drinking Water Guidelines (ADWG). National Health and Medical Research Council
491
(NHMRC) in Collaboration With the Natural Resource Management Ministerial Council (NRMMC), Commonwealth of Australia. www.nhmrc.gov.au/publications/synopses/_ files/adwg_11_06.pdf. Oda, J., Nakamura, S.I., Oki, I., Kato, T., Shinagawa, H., 1985. Evaluation of the new system (umu-test) for the detection of environmental mutagens and carcinogens. Mutatation Research 147, 219–229. Petala, M., Samaras, P., Zouboulis, A., Kungolos, A., Salkellaropoulos, G., 2006. Ecotoxicological properties of wastewater treated using tertiary methods. Environmental Toxicology 21 (4), 417–424. Petala, M., Samaras, P., Zouboulis, A., Kungolos, A., Sakellaropoulos, G.P., 2008. Influence of ozonation on the in vitro mutagenic and toxic potential of secondary effluents. Water Research 42 (20), 4929–4940. Reifferscheid, G., Heil, J., Oda, Y., Zahn, R.K., 1991. A microplate version of the SOS/umu-test for rapid detection of genotoxins and genotoxic potentials of environmental samples. Mutation Research 253 (3), 215–222. Reungoat, J., Macova, M., Escher, B.I., Carswell, S., Mueller, J.F., Keller, J., 2010. Removal of micropollutants and reduction of biological adverse effects in a full scale reclamation plant using ozonation and activated carbon filtration. Water Research 44 (2), 625–637. Rutishauser, B.V., Pesonen, M., Escher, B.I., Ackermann, G.E., Aerni, H.-R., Suter, M.J.-F., Eggen, R.I.L., 2004. Comparative analysis of estrogenic activity in sewage treatment plant effluents involving three in vitro assays and chemical analysis of steroids. Environmental Toxicology and Chemistry 23, 857–864. Sanchez-Polo, M., Salhi, E., Rivera-Utrilla, J., von Gunten, U., 2006. Combination of ozone with activated carbon as an alternative to conventional advanced oxidation processes. Ozone-Science & Engineering 28 (4), 237–245. Schreiber, U., Mueller, J.F., Haugg, A., Gademann, R., 2002. New type of dual-channel PAM chlorophyll fluorometer for highly sensitive water toxicity biotests. Photosynthesis Research 74, 317–330. Schreiber, U., Quayle, P., Schmidt, S., Escher, B.I., Mueller, J., 2007. Methodology and evaluation of a highly sensitive algae toxicity test based on multiwell chlorophyll fluorescence imaging. Biosensors and Bioelectronics 22, 2554–2563. Schwarzenbach, R.P., Escher, B.I., Fenner, K., Hofstetter, T.B., Johnson, C.A., von Gunten, U., Wehrli, B., 2006. The challenge of micropollutants in aquatic systems. Science 313 (5790), 1072–1077. Snyder, S.A., Adham, S., Redding, A.M., Cannon, F.S., DeCarolis, J., Oppenheimer, J., Wert, E.C., Yoon, Y., 2007. Role of membranes and activated carbon in the removal of endocrine disruptors and pharmaceuticals. Desalination 202, 156–181. Soto, A.M., Sonnenschein, C., Chung, K.L., Fernandez, M.F., Olea, N. , Serrano, F.O., 1995. The E-SCREEN assay as a tool to identify estrogens: an update on estrogenic environmental pollutants. Environmental Health Perspectives 103 (Suppl. 7), 113–122. Tan, B.L.L., Hawker, D.W., Muller, J.F., Leusch, F.D.L., Tremblay, L.A., Chapman, H.F., 2007. Comprehensive study of endocrine disrupting compounds using grab and passive sampling at selected wastewater treatment plants in South East Queensland, Australia. Environment International 33 (5), 654–669. Ternes, T.A., Joss, A., Siegrist, H., 2004. Scrutinizing pharmaceuticals and personal care products in wastewater treatment. Environmental Science & Technology 38 (17), 393A–399A. van Leeuwen, J., Pipe-Martin, C., Lehmann, R.M., 2003. Water reclamation at South Caboolture, Queensland, Australia. Ozone: Science & Engineering 25, 107–120. Villeneuve, D.L., Blankenship, A.L., Giesy, J.P., 2000. Derivation and application of relative potency estimates based on in-vitro bioassays. Environmental Toxicology Chemistry 19, 2835–2843.
492
water research 44 (2010) 477–492
von Gunten, U., 2003. Ozonation of drinking water: part I. Oxidation kinetics and product formation. Water Research 37 (7), 1443–1467. Warne, M.S.J., Hawker, D.W., 1995. The number of components in a mixture determines whether synergistic and antagonistic or additive toxicity predominate – the funnel
hypothesis. Ecotoxicology and Environmental Safety 31 (1), 23–28. Zhao, B., Denison, M., 2004. Dvelopment and characterization of a green fluorescence protein-based rat cell bioassay system for detection of Ah receptor ligands. Organohalogen Compounds 66, 3332–3337.
water research 44 (2010) 493–504
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
An evaluation of a pilot-scale nonthermal plasma advanced oxidation process for trace organic compound degradation Daniel Gerrity*, Benjamin D. Stanford, Rebecca A. Trenholm, Shane A. Snyder Applied Research and Development Center, Southern Nevada Water Authority, River Mountain Water Treatment Facility, P.O. Box 99954, Las Vegas, NV 89193-9954, USA
article info
abstract
Article history:
This study evaluated a pilot-scale nonthermal plasma (NTP) advanced oxidation process
Received 22 May 2009
(AOP) for the degradation of trace organic compounds such as pharmaceuticals and
Received in revised form
potential endocrine disrupting compounds (EDCs). The degradation of seven indicator
3 September 2009
compounds was monitored in tertiary-treated wastewater and spiked surface water to
Accepted 10 September 2009
evaluate the effects of differing water qualities on process efficiency. The tests were also
Available online 17 September 2009
conducted in batch and single-pass modes to examine contaminant degradation rates and the remediation capabilities of the technology, respectively. Values for electrical energy per
Keywords:
order (EEO) of magnitude degradation ranged from <0.3 kWh/m3-log for easily degraded
Advanced oxidation process (AOP)
compounds (e.g., carbamazepine) in surface water to 14 kWh/m3-log for more recalcitrant
Nonthermal plasma (NTP)
compounds (e.g., meprobamate) in wastewater. Changes in the bulk organic matter based
Trace organic compound
on UV254 absorbance and excitation-emission matrices (EEM) were also monitored and correlated to contaminant degradation. These results indicate that NTP may be a viable
Pharmaceutical Endocrine
disrupting
compound
alternative to more common AOPs due to its comparable energy requirements for contaminant degradation and its ability to operate without any additional feed chemicals.
(EDC)
ª 2009 Elsevier Ltd. All rights reserved.
1.
Introduction
Wastewater-derived contaminants, including pharmaceuticals and personal care products (PPCPs), endocrine disrupting compounds (EDCs), and other trace organic compounds, have been a significant aspect of environmental research efforts for decades (Snyder et al., 2003). However, adverse impacts on aquatic ecosystems and increased public awareness of trace organic compounds in water supplies have stimulated recent interest in this issue (Snyder et al., 2003). Due to rapid population growth and increasingly stressed water supplies, particularly in semi-arid regions, many cities are turning to alternative sources of drinking water such as indirect potable reuse. In some systems, wastewater effluent is having a greater impact on reservoir water quality due to drought and overdraft (Benotti et al., 2009a). Over time, these trends may
exacerbate the effects of trace organic compounds and force utilities to address the issue by augmenting their treatment trains. Although scientists are conflicted on the direct human health effects of PPCPs and EDCs at trace levels, the demonstrated effects on aquatic ecosystems have prompted studies of occurrence in wastewater, receiving water, source water, and drinking water (Ternes, 1998; Kolpin et al., 2002; Kim et al., 2007; Benotti et al., 2009c). Uncertainty in health effects coupled with the ubiquity of these compounds has necessitated subsequent studies on the efficacy of various treatment processes for their removal or transformation. Some processes (e.g., coagulation/flocculation/sedimentation, UV irradiation at disinfection doses (z40 mJ/cm2), and microfiltration/ultrafiltration) are ineffective for the treatment of trace organic compounds, whereas others (e.g., chlorine,
* Corresponding author. Tel.: þ1 (702) 856 3666; fax: þ1 (702) 856 3647. E-mail address:
[email protected] (D. Gerrity). 0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2009.09.029
494
water research 44 (2010) 493–504
ozone, granular activated carbon (GAC), nanofiltration/reverse osmosis, and advanced oxidation) can achieve significant removal or transformation (Ternes et al., 2002; Westerhoff et al., 2005; Snyder et al., 2007). Advanced oxidation processes (AOPs) are often used to achieve further reductions in trace organics, target recalcitrant compounds, and comply with certain water reuse regulations (e.g., California Department of Public Health Title 22 requirements for NDMA and 1,4dioxane). The most common AOPs consist of UV/peroxide and ozone/peroxide, but emerging technologies such as nonthermal plasma (NTP) provide viable alternatives. This study focuses on a novel pilot-scale NTP reactor developed by Aquapure Technologies, Ltd. (Upper Galilee, Israel). The NTP uses high-voltage electrical pulses across fiber-like electrodes to form a corona discharge, which in turn generates UV light, ozone, and hydroxyl radicals (Locke et al., 2006). The light generated by the corona discharge spans wavelengths of approximately 250 nm to 1,000 nm with multiple peaks at wavelengths associated with radical formation (Locke et al., 2006). Although the light is lowintensity (Locke et al., 2006), it has been credited with a capacity for NDMA destruction (Even-Ezra et al., 2009), which is generally inefficient in ozone-based oxidation processes (Mitch et al., 2003). NTP is considered to be highly efficient because little energy is lost in heating the surrounding fluid, which allows the energy to be focused on the excitation of electrons (Pekarek, 2003). In contrast to the electron beam technology, which introduces electrons from an external source, the NTP used in this study employs a corona discharge that excites electrons in the ambient air directly above the target water matrix (Pekarek, 2003). Other NTP configurations are also available, including those that create a corona discharge in aerosols or within the target water matrix (Pekarek, 2003; Locke et al., 2006). The level of treatment can be adjusted by changing the frequency (e.g., from 500 to 1,000 Hz) or voltage (up to 40 kV) of the electrical pulses. The ionizing capability of the electrical pulses generates singlet oxygen atoms, which in turn form ozone and hydroxyl radicals. One of the most significant benefits of NTP is the fact that oxidants can be generated without the addition of costly chemicals or UV lamps, which require cleaning and are hindered by high turbidity and matrix absorbance. However, the NTP technology has limited commercial availability and has not been tested in full-scale drinking water and wastewater treatment applications so its long-term reliability, efficiency, and effectiveness are still unknown. In addition, the NTP technology is currently limited to low-flow applications due to its thin-film configuration. Although plasma-based technologies have been studied extensively for the degradation of volatile organic compounds (VOCs) (Masuda et al., 1995; Hwang and Jo, 2005; Ighigeanu et al., 2005; Thevenet et al., 2007; Kim et al., 2008), many of these studies have focused on air quality and/or bench-scale configurations. Due to limitations associated with conventional AOPsdprimarily the need for chemical additiond studies of novel AOPs for water and wastewater treatment applications are gaining popularity. A recent pilot-scale study utilizing the Aquapure technology evaluated the efficacy of NTP in degrading trichloroethylene (TCE), N-nitrosodimethylamine (NDMA), 1,4-dioxane, and methyl tert-butyl
ether (MTBE) in batch experiments and single-pass remediation (Even-Ezra et al., 2009). The current study uses the same pilot-scale system to evaluate the degradation of PPCPs and EDCs in water and wastewater. Despite the lack of PPCP/EDC data for plasma-based technologies, other AOPs, particularly UV/peroxide (Huber et al., 2003; Pereira et al., 2007; Snyder et al., 2007; Benotti et al., 2009b), ozone/peroxide (Acero and von Gunten, 2001; Snyder et al., 2007), and photocatalysis (Benotti et al., 2009b; Westerhoff et al., 2009), have been tested extensively with respect to trace organic degradation. In their review of NTP applications, Locke et al. (2006) discussed the need for comparisons of NTP with more common AOPs to determine whether there is a net benefit associated with employing NTP technologies. Among other research needs, Locke et al. (2006) specifically identified energy consumption and applicability to different waste matrices as useful points of comparison. To this end, the objective of this study is to expand the NTP knowledge base by evaluating its efficacy and energy efficiency in degrading seven PPCPs and EDCs in tertiary-treated wastewater and spiked surface water.
2.
Experimental section
2.1.
Selection of wastewater-derived contaminants
In order to narrow the scope of this research, a subset of the numerous compounds detected in previous occurrence studies was selected for evaluation. The indicator compounds were selected based on their magnitude and frequency of occurrence in water and wastewater (Snyder et al., 2007), varying physical/chemical characteristics and resulting susceptibility to treatment (Ternes et al., 2002; Westerhoff et al., 2005; Snyder et al., 2007), and ease of analytical methods (Trenholm et al., 2009). The seven compounds targeted in this study include meprobamate (anti-anxiety), dilantin (also known as phenytoin; anticonvulsant), primidone (anticonvulsant), carbamazepine (anticonvulsant), atenolol (betablocker), trimethoprim (antibiotic), and atrazine (herbicide and suspected EDC). Their structures and characteristics are described in Snyder et al. (2007). Although these compounds have generated considerable interest in the research, treatment, and regulatory arenas, only atrazine is currently regulated by the United States (U.S.) Environmental Protection Agency (EPA) at a maximum contaminant level (MCL) of 3 mg/L.
2.2.
Nonthermal plasma pilot unit
The pilot-scale unit depicted in Fig. 1 is an ‘‘electrode-to-plate’’ NTP prototype (Locke et al., 2006) developed by Aquapure Technologies, Ltd.; the pilot skid contains two reactors connected in series. Water flows in a thin film (z5 mm) along the stainless steel ground electrode (anode) where it is exposed to high-voltage electrical pulses. The electrical pulses from the generator have frequencies ranging from 500 to 1,000 Hz, maximum voltage of 8.0 kV, maximum current of 100 A, maximum energy of 1 J, and rise time of approximately 18 ns (Locke et al., 2006; Even-Ezra et al., 2009). Depending on the applied voltage and frequency, each reactor draws between 0.4 and 1.0 kW, and the external pumping and controls draw
water research 44 (2010) 493–504
495
Fig. 1 – Schematic of NTP pilot unit. The bottom figure illustrates a profile view of the electrodes within the reactor.
approximately 2.8 kW. The electrical pulses generate a corona discharge that stretches approximately 1 cm from the carbon fiber-like cathode to the water surface. The distance between the carbon electrode and water surface can be adjusted at multiple points along the reactor to level the system and optimize the process for a particular water matrix. The 1-cm distance was selected to achieve a consistent corona discharge and minimize sparking within the reactor, which would be destructive to the carbon fibers. Despite the harsh conditions experienced by the gas-phase electrode, no visible wear was detected on the carbon fibers over the duration of the experiments. With continuous, long-term operation of the NTP reactor, it is likely that the carbon fibers would eventually require replacement, but this life span is currently unknown. After traveling through the first reactor, the water is collected in a storage tank before being pumped into the ozone injector system or the second reactor. The ozone injector system combines a portion of the treated water from the first reactor with ozone-rich air (z2 g/m3) from the reactor headspace using a Venturi inductor. For the single-pass configuration, this mixture is collected in another storage tank for additional ozone contact time before being mixed and discharged with effluent from the first reactor. Excess air from the ozone contactor is then drawn out of the system by vacuum and quenched in an ozone destruct unit. The hydraulic residence times in the reactor, storage tank, and ozone contactor are approximately 0.5 min, 4–7.5 min, and 1.5 min, respectively. The hydraulic residence times in the reactor and storage tank depend on the process flow rate, whereas the residence time in the ozone contactor remains relatively constant because the injector flow rate (z40 L/min) is controlled by
a separate pump. The process is then repeated in the second reactor before being pumped out of the pilot skid. For the batch configuration, the effluent pump is replaced by a recirculation pump that allows the water to cycle continuously through a single reactor and ozone contactor. Additional details related to the NTP reactor used in this study are provided in Even-Ezra et al. (2009).
2.3.
Experimental design
The study was divided into three experiments: (1) a batch experiment with 150 L of tertiary-treated wastewater (i.e., grit removal, primary settling, activated sludge, secondary settling, and dual media filtration) at a recirculation rate of 8.0 L/min, (2) a single-pass experiment with tertiary-treated wastewater at a flow rate of 15.5 L/min, and (3) a single-pass experiment with spiked surface water at a flow rate of 11.4 L/min. For experiment (3), spiked surface water (Colorado River water from Lake Mead, NV) was pumped into the reactor from a 3,000-gallon polyethylene water tank (American Tank Company, Windsor, CA). The spiking solution was prepared by dissolving neat standards in water in order to avoid the introduction of radical scavenging solvents such as methanol. The spiking solution was added to the 3,000-gallon tank followed by recirculation for 24 h to facilitate homogenization. The two water types (wastewater and surface water) were selected to provide an assessment of NTP treatment efficiency for matrices with different organic loadings. The initial pH, total alkalinity, total organic carbon (TOC), UV254 absorbance, and trace organic compound concentrations for the three experiments are provided in Table 1. In all tables, ‘‘Batch’’ refers to the batch
496
water research 44 (2010) 493–504
4.8 kWh/m3 for the overall pilot skid. Samples were collected from the effluent line exiting the pilot skid. Residual oxidants in each sample were quenched with calcium thiosulfate. During the wastewater experiments, the dissolved ozone concentration in the NTP reactor effluent ranged from approximately 0.05 mg/L to 0.10 mg/L depending on the generator settings. The single-pass surface water experiment (3) incorporated nine different generator scenarios ranging from 0.6 to 2.6 kWh/m3 for the generators and 4.8 to 6.8 kWh/ m3 for the pilot skid. For the surface water, samples were taken from Reactor 1, Ozone Tank 1, Reactor 2, and the pilot skid effluent to evaluate the extent of contaminant degradation in each system component. During the surface water experiment, the dissolved ozone concentrations in the reactor effluent ranged from 0.25 to 0.50 mg/L depending on the generator settings.
Table 1 – Initial water quality conditions. Parameter pH Total Alkalinity TOC UV254 Abs. Meprobamate Dilantin Primidone Carbamazepine Atenolol Trimethoprim Atrazined
Batch (1)
WW (2)a,c
SSW (3)b,c
7.0 126 mg/L as CaCO3 5.9 mg-C/L 0.119 cm1 275 ng/L 119 ng/L 124 ng/L 176 ng/L 378 ng/L 36 ng/L N/A
7.0 126 mg/L as CaCO3 5.6 mg-C/L N/A 256 25e ng/L 165 10 ng/L 146 6 ng/L 219 9 ng/L 413 8 ng/L 39 2 ng/L N/A
8.0 137 mg/L as CaCO3 2.6 mg-C/L 0.036 cm1 933 15 ng/L 1,005 74 ng/L 1,375 96 ng/L 1,600 115 ng/L 1,018 57 ng/L 1,200 82 ng/L 1,118 159 ng/L
a WW (2) ¼ Single-pass wastewater experiment. b SSW (3) ¼ Single-pass spiked surface water experiment. c Four influent samples were taken throughout the experiment to calculate a representative ambient concentration and evaluate temporal variability in target compound concentrations. d Atrazine was spiked in the SSW experiment but was not detected in the wastewater. e Errors represent 1 standard deviation.
2.4.
Analytical methods
Ozone gas concentrations in the reactor headspace were determined by online measurements with an OMC20 ozone monitor and IN-2000-L2-LC ozone analyzer (IN USA, Norwood, MA). Dissolved ozone residuals were measured with the indigo trisulfonate method (Yates and Stenstrom, 2000). TOC and UV254 absorbance were analyzed according to standard methods 5310B and 5910, respectively. Excitation-emission matrices (EEM) were also developed for the single-pass wastewater samples using a QuantaMaster UV-Vis QM4 Steady State Spectrofluorometer (Photon Technology International, Inc., Birmingham, NJ). Trace organic compounds were extracted and analyzed using on-line solid phase extraction and liquid chromatography with tandem mass spectrometry (SPE-LC-MS/MS) according to Trenholm et al. (2009). All standards and reagents were of the highest purity commercially available. All
wastewater experiment, ‘‘WW’’ refers to the single-pass wastewater experiment, and ‘‘SSW’’ refers to the single-pass spiked surface water experiment. The generator settings for the NTP pilot differed for each of the experiments, as described in Table 2. For the batch experiment (1), a single reactor/generator was operated at a frequency of 500 Hz and a voltage of 8.0 kV. To assess the degradation rate, ten samples were collected at generator (AOP-specific) energy consumption values ranging from 0 to 7.3 kWh/m3. For the single-pass wastewater experiment (2), four different generator scenarios provided treatment levels ranging from 0.7 to 1.8 kWh/m3 for the generators and 3.8 to
Table 2 – Experimental test conditions. Experiment
Generator 1 Frequency (Hz)
Voltage (kV)
Generator 2 Energy (kWh/m3)
Frequency (Hz)
Voltage (kV)
Energy (kWh/m3)
Total generator energy
Total skid energy
(kWh/m3)
(kWh/m3)
(1)
Batcha
Off
Off
0.00
500
8.0
N/A
N/A
N/A
(2)
WWa – 1 WW – 2 WW – 3 WW – 4
Off 800 600 800
Off 8.0 8.0 8.0
0.00 1.04 0.87 1.04
500 Off 500 500
8.0 Off 8.0 8.0
0.71 0.00 0.71 0.71
0.71 1.04 1.58 1.75
3.75 4.08 4.62 4.79
(3)
SSWb – 1 SSW – 2 SSW – 3 SSW – 4 SSW – 5 SSW – 6 SSW – 7 SSW – 8 SSW – 9
Off Off 500 Off 1000 500 500 1000 1000
Off Off 8.0 Off 8.0 8.0 8.0 8.0 8.0
0.00 0.00 0.97 0.00 1.53 0.97 0.97 1.53 1.53
600 800 Off 1000 Off 600 1000 600 1000
5.9 5.9 Off 5.9 Off 5.9 5.9 5.9 5.9
0.63 0.85 0.00 1.07 0.00 0.63 1.07 0.63 1.07
0.63 0.85 0.97 1.07 1.53 1.60 2.04 2.16 2.60
4.78 5.01 5.12 5.23 5.68 5.75 6.19 6.31 6.75
a The batch and WWTP experiments were conducted with ambient concentrations of trace organics in tertiary-treated wastewater. b The SSW experiments were conducted with raw surface water (Lake Mead) spiked with a suite of trace organics.
497
water research 44 (2010) 493–504
3.
1.0
0.8
0.6
C/C0
pharmaceuticals were obtained from Sigma-Aldrich (St. Louis, MO, USA). Meprobamate-d3 and trimethoprim-d9 were obtained from Toronto Research Chemicals (Ontario, Canada). Phenytoin-d10 and atenolol-d7 were obtained from C/D/N Isotopes (Pointe-Claire, Canada). Carbamazepine-d10 and primidone-d5 were obtained from Cambridge Isotope Laboratories (Andover, MA, USA). All solvents were trace analysis grade from Burdick and Jackson (Muskegon, MI). Reagent water was obtained using a Milli-Q Ultrapure Water Purification System (Millipore, Bedford, MA, USA). All concentrated stocks were prepared in methanol and stored at 20 C, while mixed spiking solutions were prepared in reagent water and stored at 4 C. On-line SPE was performed using a Symbiosis Pharma (Spark Holland, Emmen, the Netherlands) automated SPE in XLC mode. Briefly, a 10-mL volume of sample was measured in a volumetric flask at which time isotopically-labeled standards were spiked at 100 ng/L. This provided sufficient sample volume for duplicates, matrix spikes, and dilutions, if necessary. A 1.5-mL aliquot of each sample was then transferred to a 2-mL autosampler vial, although only 1.0 mL was used for extractions. Extractions were performed using Waters Oasis HLB Prospekt cartridges (30 mm, 2.5 mg, 10 1 mm, 96 tray) (Milford, MA). Prior to sample loading, each cartridge was sequentially conditioned with 1 mL of dichloromethane (DCM), MTBE, methanol, and reagent water (Milli-Q). Samples were loaded onto the SPE cartridge at 1 mL/min after which the cartridge was washed with 1 mL of reagent water. After sample loading, the analytes were eluted from the SPE cartridge to the LC column with 200 mL methanol, using the LC peak focusing mode. A 5-mM ammonium acetate solution and methanol gradient was used for LC mobile phases with a flow rate of 800 mL/min. Analytes were separated using a 150 4.6 mm Luna C18(2) with a 5-mm particle size (Phenomenex, Torrance, CA). Method reporting limits (MRLs) were chosen at 3 to 5 times the method detection limit (MDL). MRLs were 10 ng/L for all compounds, except for atenolol, which was 25 ng/L. All analytes were quantified using isotope dilution. Stringent QA/QC protocols (i.e., matrix spikes, duplicate samples, field blanks, and laboratory blanks) were followed throughout the duration of the experiment. Using the on-line SPE method, the concentrations of duplicate samples rarely varied by greater than 5%.
0.4
0.2
0.0 0
1
2
3
4
5
6
7
8
Generator Energy Consumption (kWh/m3) Meprobamate
Dilantin
Primidone
Carbamazepine
Atenolol
Trimethoprim
Fig. 2 – Contaminant degradation profiles from the batch experiment (1) with ambient concentrations of trace organic compounds in tertiary-treated wastewater. This graph illustrates the relationship between contaminant degradation and generator (AOP-specific) energy consumption. The dashed lines represent samples that were below the MRL for those compounds.
The degradation of each detected contaminant was assumed to be pseudo first order based on a plot of ln(C/C0) versus generator energy consumption. Degradation rates, 95% confidence intervals, and R2 values are provided in Table 3. Due to the low initial trimethoprim concentration, the extent of degradation was limited, as indicated by the dashed line representing the MRL in Fig. 2. Despite the limited degradation, it is apparent that trimethoprim and carbamazepine are highly susceptible to NTP treatment. This result is consistent with previous studies assessing oxidation strategies for trace organic compounds. Westerhoff et al. (2005) reported nearly 100% degradation of trimethoprim and carbamazepine with standard chlorination and ozonation practices. Dilantin and atenolol degraded more slowly during the NTP treatment, which is also consistent with oxidation by chlorine or ozone (Westerhoff et al., 2005). Finally, meprobamate has been shown to be highly resistant to chlorination and ozonation in comparison to other trace organic compounds (Westerhoff et al., 2005). This conclusion is supported by the NTP results in that meprobamate required nearly double the energy to
Results and discussion
3.1. Experiment 1 – batch experiment with tertiarytreated wastewater Contaminant degradation profiles during the batch wastewater experiment are provided in Fig. 2. The initial ambient concentrations for the target contaminants ranged from 36 ng/L for trimethoprim to as high as 378 ng/L for atenolol (Table 1). Although included in the analytical method, atrazine was below the method reporting limit (MRL) for both wastewater experiments. As indicated in Table 2, the generator was operated at a frequency of 500 Hz, a voltage of 8.0 kV, and a recirculation rate of 8.0 L/min.
Table 3 – Pseudo first order rate constants for batch experiment (1) in tertiary-treated wastewater. Compound Meprobamate Dilantin Primidone Carbamazepine Atenolol Trimethoprim Atrazinea
k (m3/kWh)
95% CIb
R2
N
0.36 0.52 0.40 1.03 0.61 0.71 N/A
0.03 0.15 0.15 0.35 0.22 N/A N/A
0.98 0.92 0.88 0.93 0.88 0.92 N/A
9 5 5 4 5 2 N/A
a Atrazine concentrations were below MRL in all samples. b 95% CI ¼ 95% confidence interval.
498
water research 44 (2010) 493–504
achieve a specified level of degradation. Primidone was not quite as resistant as meprobamate, but its degradation was slower and more variable than the other compounds. With respect to bulk organic parameters, there was no significant change in TOC over the duration of the experiment. The consistent TOC level was expected considering that mineralization is highly energy intensive in hydroxyl radicaldominated processes (Gerrity et al., 2009), and high CT values are required in ozone-dominated processes (Rosal et al., 2009). The limited energy consumption, low dissolved ozone concentrations, and limited ozone contact time were likely insufficient to induce measurable organic mineralization in the NTP reactor. Although there was no reduction in TOC, UV254 absorbance was reduced by more than 65%; this change was correlated to the degradation of target compounds in Fig. 3 as a potential surrogate measure of process efficacy. Quantifying trace organic compounds requires specialized equipment and training, and the analytical methods are very expensive and time-consuming. For this reason, it is generally impractical for water and wastewater utilities to employ trace contaminant monitoring programs for their systems. Implementation of this type of surrogate framework (i.e., reduction in UV254 absorbance in lieu of trace contaminant degradation) would allow utilities to assess the performance of certain treatment processes for emerging contaminants. For oxidation processes, UV254 absorbance is an easily measured parameter that appears to show consistent correlations with the degradation of trace organic compounds (Wert et al., 2009). Fig. 3 provides support for this surrogate framework because it demonstrates strong correlations between the degradation of the target compounds and reduction in UV254 absorbance. The slopes of the regression lines indicate that relative contaminant degradation (i.e., percent degradation) is actually more rapiddmore than twice as fast for nearly all compoundsdthan the reduction in UV254 absorbance. The slopes also show a general agreement with the degradation profiles
in Fig. 2. Although trimethoprim was expected to degrade rapidly, its slope of 6.25 is likely an outlier and artifact of its small sample size. With the exception of trimethoprim, the magnitude of the slopes and the negative intercepts are similar to those reported in the Wert et al. (2009) ozonation study. The negative intercepts indicate that there are initial reductions in UV254 absorbance that precede degradation of the target compounds. This may be attributable to initial oxidant/radical scavenging by the bulk organic matter. Values for electrical energy per order (EEO) of magnitude degradation (Bolton et al., 1996), which is often measured in kWh/m3-log, are also provided in Table 4. Since the batch experiment involved a continuous treatment, the EEO values were calculated based on first order degradation rates and linear regression. In other words, a linear regression was generated for ln(C/C0) against generator energy consumption, and the EEO was calculated as ln(0.1) divided by the slope of the linear regression. The EEO values ranged from 2.2 kWh/m3 for carbamazepine to 6.4 kWh/m3 for meprobamate. It is important to note that the batch EEO values may be overestimates because the batch system requires approximately 10–15 min to achieve its maximum ozone concentration after starting the generator.
3.2. Experiment 2 – single-pass experiment with tertiary-treated wastewater Contaminant degradation profiles for the single-pass wastewater experiment are illustrated in Fig. 4. The initial ambient concentrations for the target contaminants were similar to the batch experiment and ranged from 39 ng/L for trimethoprim to as high as 413 ng/L for atenolol (Table 1). As indicated in Table 2, this experiment involved a single-reactor configuration and both reactors operating in series at a flow rate of 15.5 L/min. All scenarios for this experiment involved voltages of 8.0 kV. The reactor consumed 0.7 kWh/m3 with Generator 2 operating at a frequency of 500 Hz during the
% Degradation of Trace Organic Compounds
100%
80%
Linear Regression Models: 60%
Contaminant
40%
20%
Slope Intercept R2
N
Meprobamate
1.56
-14%
0.97
8
Dilantin
2.09
-22%
0.99
4
Primidone
2.08
-28%
0.98
4
Carbamazepine 2.75
-30%
1.00
3
Atenolol
2.68
-44%
0.99
4
Trimethoprim
6.25
-90%
1.00
2
0% 0%
20%
40%
60%
80%
100%
% Reduction in UV254 Absorbance
Fig. 3 – Correlation between the degradation of target compounds and UV254 absorbance during the batch experiment (1). This analysis excludes samples that were below the MRL and the final five carbamazepine samples (>3 kWh/m3; see Fig. 2) due to high variability near the MRL. The slopes were all significant at the 0.03 level or better, and intercepts were all significant at the 0.07 level or better.
499
water research 44 (2010) 493–504
Table 4 – AOP-specific EEO values (kWh/m3-log). Contaminant
(1)
(2)
(3)
Batcha
Min. WWb
Med. WWb
Max. WWb
Min. SSWb
Med. SSWb
Max. SSWb
6.4 4.4 5.8 2.2 3.8 3.2 N/A
4.6 2.7 2.8 1.3 2.5 1.5 N/A
10 3.1 4.3 1.8 2.9 2.2 N/A
14 3.5 4.8 2.1 3.3 3.0 N/A
2.1 1.1 1.1 <0.3 0.4 <0.3 2.2
3.5 2.0 2.2 <0.7 1.0 <0.7 3.7
5.3 3.1 3.3 <1.2 1.7 <1.2 6.3
Meprobamate Dilantin Primidone Carbamazepine Atenolol Trimethoprim Atrazine
UVc SSW
UV/H2O2c,d SSW
UV/TiO2c,e SSW
6.6 2.1 3.7 2.3 1.4 0.8 3.3
1.0 1.0 0.6 0.4 0.5 0.4 1.2
6.8 2.2 3.9 2.1 2.0 1.5 4.7
a Batch EEO values based on linear regression assuming first order kinetics. This analysis excludes samples that were below the MRL and the final five carbamazepine samples (>3 kWh/m3; see Fig. 2) due to high variability near the MRL. b WW and SSW were comprised of discrete sampling events so regression was not appropriate. c Adapted from Benotti et al. (2009b). The water matrix was identical to that of the SSW samples from this study (i.e., Lake Mead). d UV/H2O2 with 10 mg/L of H2O2. e UV/TiO2 (photocatalysis) with approximately 500 mg/L of titanium dioxide (TiO2).
WW-1 scenario, and the reactor consumed 1.0 kWh/m3 with Generator 1 operating at a frequency of 800 Hz during the WW-2 scenario. WW-3 and WW-4 involved both reactors operating in series at 600/500 Hz (1.6 kWh/m3) and 800/500 Hz (1.8 kWh/m3), respectively. The energy consumption values for the entire skid, which include energy related to the generators, pumps, and controls, are also provided for each scenario. The relative degradation rates were identical to those of the batch experiment. For example, carbamazepine and trimethoprim experienced the most rapid degradation, and primidone and meprobamate were the most recalcitrant compounds. At the highest level of energy consumption (1.8 kWh/m3), every compound except meprobamate experienced at least 70% degradation. The evaluation of
1.0
0.8
C/C0
0.6
0.4
0.2
0.0 0.0
0.5
1.0
1.5
2.0
Generator Energy Consumption (kWh/m3) Meprobamate
Dilantin
Primidone
Carbamazepine
Atenolol
Trimethoprim
Fig. 4 – Contaminant degradation profiles from the singlepass experiment (2) with tertiary-treated wastewater. The initial contaminant concentrations represent ambient conditions in the wastewater. This graph illustrates the relationship between contaminant degradation and generator (AOP-specific) energy consumption. The dashed line represents samples that were below the MRL for trimethoprim.
trimethoprim was hindered by the low initial ambient concentration in the wastewater. EEO values for the recalcitrant compounds (meprobamate and primidone) decreased by 64% and 42%, respectively, as the energy consumption increased from 0.7 to 1.8 kWh/m3. EEO values for dilantin and atenolol decreased by approximately 20% over the same range of energy consumption. However, the EEO values for the most susceptible compounds (carbamazepine and trimethoprim) actually increased by 38% and 100%, respectively, due to their rapid degradation at low levels of energy consumption. Most compounds were characterized by EEO values ranging from 1 to 5 kWh/m3-log, but meprobamate required 13–14 kWh/m3-log with one reactor in operation. In general, the single-pass wastewater experiment was more efficient than the batch experiment with respect to EEO values due to the presence of ozone at the beginning of each discrete scenario, which provides a more appropriate representation of actual treatment conditions. As with the batch experiment, TOC concentrations did not change during the various treatment scenarios, but the transformation of bulk organic matter was evaluated with excitation-emission matrices (Fig. 5). Based on the regions described in Chen et al. (2003), a portion of the organic matter was characterized as soluble microbial byproduct-like (region IV), though a much larger portion of the organic matter was characterized as humic acid-like (region V). It is apparent that the NTP treatment quickly transformed the natural organic matter in both of these regions as observed by the dramatic decrease in fluorescence intensity. Although the organic matter was transformed into compounds with less ability to fluoresce, the constant TOC values indicate that the treatment was insufficient to induce organic mineralization. This highlights the importance of transformation products in AOP applications. Although the target compounds are degraded, the lack of mineralization indicates the formation of byproducts related to the bulk organic matter and possibly the trace organic compounds, which may or may not be harmful to humans and/or the environment. Due to the complexity of the target water matrices, identifying byproducts associated with the target compounds was beyond the
500
water research 44 (2010) 493–504
Fig. 5 – Excitation-emission matrices as a function of generator energy consumption for the single-pass experiment (2) with tertiary-treated wastewater. The regions are described as follows: (I) Aromatic protein, (II) aromatic protein II, (III) fulvic acid-like, (IV) soluble microbial byproduct-like, and (V) humic acid-like (Chen et al., 2003).
scope of this study, but future studies should address this issue in greater detail.
3.3. Experiment 3 – single-pass experiment with spiked surface water Contaminant degradation profiles, including that of atrazine, for the single-pass spiked surface water experiment are illustrated in Fig. 6. The initial spiked concentrations were much higher than the ambient wastewater concentrations as all of the compounds were spiked at greater than 900 ng/L (Table 1). Similar to the single-pass wastewater experiment, this experiment involved a single-reactor configuration and both reactors operating in series at a flow rate of 11.4 L/min (Table 2). For this experiment, Generator 1 was operated at a voltage of 8.0 kV, and Generator 2 was operated at a voltage of 5.9 kV. The generator and overall skid energy consumption values ranged from 0.6 to 2.6 kWh/m3 and 4.8 to 6.8 kWh/m3, respectively. The relative degradation rates were similar to those of the wastewater experiments. For example, meprobamate was the most difficult to degrade, while trimethoprim and carbamazepine rapidly dropped below the MRL in all treated samples. The lower organic loading of the surface water (i.e., lower initial TOC concentration and UV254 absorbance), which
resulted in a higher residual ozone concentration (0.25 to 0.50 mg/L), likely contributed to the improved degradation. In contrast to the wastewater experiments, atenolol degradation was more similar to trimethoprim and carbamazepine, while primidone degradation was more similar to that of dilantin. As expected, the degradation of atrazine, which is considered a recalcitrant compound (Westerhoff et al., 2005), was almost identical to that of meprobamate. For all of the compounds with concentrations above the MRL, the degradation peaked at about 70% for the most recalcitrant compounds and 85% for the moderately recalcitrant compounds after reaching an energy consumption of 1.6 kWh/m3. At the highest energy level, the effluent concentrations of the residual compounds ranged from 140 ng/L for dilantin to 390 ng/L for atrazine. During Experiment 3, there was also greater fluctuation for several time points. Since the samples represent discrete, rather than continuous (i.e., batch), treatment levels, one would anticipate greater variability, but there was still a relatively consistent decrease in contaminant concentration over the range of energy values tested. The EEO values for all of the compounds were dramatically lower in the surface water experiment than the wastewater experiments, presumably due to the lower scavenging capacity of the surface water. Meprobamate and atrazine had the highest EEOs with maximum values of 5.3 and 6.3 kWh/
501
water research 44 (2010) 493–504
1
0.8
C/C0
0.6
0.4
0.2
0 0.0
0.5 Meprobamate
1.0 1.5 2.0 Generator Energy Consumption (kWh/m3) Dilantin
Primidone
Carbamazepine
Atenolol
2.5 Trimethoprim
3.0 Atrazine
Fig. 6 – Contaminant degradation profiles from the single-pass experiment (3) with spiked surface water. This graph illustrates the relationship between contaminant degradation and generator (AOP-specific) energy consumption. m3-log, respectively. The maximum values generally occurred during the experiments when one or both of the generators were operating at 1,000 Hz. Although operating both generators at 1,000 Hz achieved the highest level of degradation, increasing the pulse frequency generally yielded diminishing rates of return with respect to efficiency. Operating Generator 2 at a frequency of 600 Hz provided the most efficient treatment conditions for all of the target compounds. However, Generator 1 had been operated immediately prior to that scenario, so the residual ozone from Reactor 1 likely contributed to this increased efficiency. Although this overestimates the anticipated degree of degradation, it indicates that it may be beneficial to incorporate ozone contact tanks before and after each reactor to maximize degradation. Based on this
observation, the median values are likely the most appropriate EEO estimates since they always corresponded to both reactors operating simultaneously. For the compounds with reportable concentrations, the median EEO values ranged from 1.0 to 3.7 kWh/m3 for atenolol and atrazine, respectively. A subset of samples was also analyzed to assess the contribution of the system components (i.e., reactor vs. ozone contactor) to contaminant degradation. The results from one set of samples (SSW-8) are provided in Fig. 7; the other sample sets yielded similar trends and therefore are not shown. Based on these results, it appears that the ozone contactors provided the majority of the degradation for the more recalcitrant compounds, including meprobamate, dilantin, primidone,
1600
1400
Concentration (ng/L)
1200
1000
800
600
400
200
0 Influent
Reactor 1
Ozone 1
Reactor 2
Effluent
Sample Location Meprobamate
Dilantin
Primidone
Carbamazepine
Atenolol
Trimethoprim
Atrazine
Fig. 7 – Comparison of the pilot system components in degrading the target compounds during the single-pass experiment (3) with spiked surface water. This data refers to samples from SSW-8 with generator and skid energy consumption values of 2.2 and 6.3 kWh/m3, respectively. The small increase in concentrations from Ozone 1 to Reactor 2 is explained by mixing of Reactor 1 and Ozone 1 effluent prior to passage into Reactor 2 (see Fig. 1).
502
% Degradation of Trace Organic Compounds
water research 44 (2010) 493–504
100%
Linear Regression Models: 80%
60%
40%
Contaminant
Slope Intercept R2
N
Meprobamate
2.40
0
0.72
9
Dilantin
1.35
45%
0.84
9
Primidone
1.52
40%
0.81
9
Atrazine
2.27
0
0.74
9
20%
0% 0%
5%
10%
15%
20%
25%
30%
35%
% Reduction in UV254 Absorbance Meprobamate
Dilantin
Primidone
Atrazine
Fig. 8 – Correlation between the degradation of detected target compounds and UV254 absorbance during the single-pass experiment (3) with spiked surface water. The slopes were all significant at the 0.001 level, the intercepts for primidone and dilantin were significant at the 0.001 level, and the intercepts for meprobamate and atrazine were considered insignificant ( p > 0.09; regressions for meprobamate and atrazine were forced through the origin).
atenolol, and atrazine. However, carbamazepine and trimethoprim experienced rapid degradation in the reactor itself with the additional oxidation pathways (i.e., UV light, hydroxyl radicals, and ozone). The spiked surface water experiment was also evaluated based on general water quality parameters. For example, the initial bromide concentration in the surface water was approximately 62 mg/L, and additional analyses indicated that bromate increased only slightly (maximum of 2.4 mg/L) above the ambient surface water concentration of 1.7 mg/L. Similar to the wastewater experiments, there was no significant change in TOC over the duration of the experiment, but the reduction in UV254 absorbance was correlated to the degradation of the target contaminants in Fig. 8. In a manner similar to Fig. 3, Fig. 8 illustrates strong positive correlations between contaminant degradation and reduction in UV254 absorbance. Furthermore, the degradation of target contaminants was still more rapid than the reduction in UV254 absorbance. In contrast to the batch wastewater experiment and data provided in Wert et al. (2009), the regression lines for dilantin and primidone have positive intercepts, while the intercepts for meprobamate and atrazine were found to be insignificant ( p > 0.09). Wert et al. (2009) proposed that the correlations between contaminant degradation and UV254 absorbance were independent of wastewater quality, but it appears that the correlations differ when the matrices themselves are drastically different (i.e., wastewater vs. surface water). Specifically, the positive intercepts for dilantin and primidone in the surface water experiment indicate that these target compounds started to degrade (z40–45%) prior to any reduction in UV254 absorbance. In addition to contributing to their smaller slopes, this may indicate that there was less oxidant/radical scavenging and that the bulk organic matter was more recalcitrant in the surface water compared to the wastewater. It is also interesting to note that meprobamate has the smallest absolute intercept for the wastewater experiment and an insignificant
intercept for the surface water experiment, and atrazine also has an insignificant intercept for the surface water experiment. Therefore, the initial degradation of these highly resistant compounds is more similar to the initial degradation of the bulk organic matter. For the surface water experiments, this supports the theory that the aromatic organic matter in the surface water was more recalcitrant than that of the wastewater. Based on these observations, it appears that the UV254 surrogate framework is still valid, but there may be differences in the correlations for dissimilar water matrices. Table 4 provides EEO values for the three experiments in this study, and it also provides EEO values from Benotti et al. (2009b), which evaluated the efficacy of photolysis, UV/H2O2, and photocatalysis in degrading PPCPs and EDCs. Benotti et al. (2009b) provides an interesting point of comparison since the water matrix used for that study (spiked surface water from Lake Mead), target compounds, and analytical methods were identical. As indicated by the data, NTP was generally more efficient than photolysis and photocatalysis for each of the compounds, but was either comparable or slightly less efficient than UV/H2O2 at a peroxide dose of 10 mg/L. It is important to note that the NTP technology does not require UV lamps or costly peroxide feeds to generate the oxidative species. Therefore, NTP may be a viable alternative to the more common UV/H2O2 process despite the slightly higher energy requirements for some compounds. Although the NTP achieved comparable efficiencies to those of UV/H2O2 in these experiments, identification of the best alternative for a particular application also depends on capital costs and reliability. Since the NTP technology has not been fully commercialized, it is difficult to compare its capital costs and reliability with more established technologies. Future evaluations of large-scale NTP reactors, including long-term studies, will provide valuable information in determining how NTP compares to more established technologies on a life-cycle basis.
water research 44 (2010) 493–504
4.
Conclusions
With respect to treatment efficacy, the pilot-scale NTP system demonstrated rapid degradation of the target compounds. The relative degradation rates were consistent between the three phases of the study with carbamazepine and trimethoprim as the most susceptible compounds and meprobamate, primidone, and atrazine as the most recalcitrant compounds. Atenolol and dilantin were generally more resistant to oxidation than carbamazepine and trimethoprim. These similar relative degradation rates justify the use of indicator compounds in evaluations of PPCP and EDC treatment. The EEO values for the optimal scenarios were generally less than 5 kWh/m3-log for all of the target compounds, particularly in the spiked surface water with a lower organic loading. With process optimization (i.e., optimization of electrode height for each water matrix) and limited modifications to the system, including more effective use of ambient ozone, the EEO values could likely be reduced further. More common AOPs, including UV/H2O2 and ozone/H2O2, require significant chemical addition and residual H2O2 quenching, which represent a significant portion of their operational costs. The primary benefit of the NTP AOP is the ability to generate UV light, ozone, and hydroxyl radicals without chemical addition or the use of UV lamps. However, the true capital costs and reliability of large-scale NTP reactors for water and wastewater treatment are still relatively unclear. Also, since oxidation by molecular ozone may be the dominant mechanism for many compounds, future studies should provide direct comparisons of NTP with conventional ozone generators to compare system performance. These expanded studies are necessary before one can fully determine the applicability of the novel NTP technology for water and wastewater treatment. However, these preliminary results indicate that NTP may be a viable alternative to more common AOPs due to its chemical-free operation and comparable energy requirements for trace contaminant degradation.
Acknowledgments We would like to thank the SNWA Applied Research and Development Center staff, particularly Janie Zeigler, Shannon Ferguson, Christy Meza, and Elaine Go, for assisting with the various analyses. We would also like to thank Dvir Solnik, Itay Even-Ezra, Gal Bitan, and Shahar Nuriel of Aquapure Technologies for use of the nonthermal plasma pilot unit and for providing crucial training and guidance during the study.
references
Acero, J.L., von Gunten, U., 2001. Characterization of oxidation processes: ozonation and the AOP O3/H2O2. J. AWWA 93 (10), 90–100. Benotti, M.J., Stanford, B.D., Snyder, S.A., 2009a. Increased loading of pharmaceuticals and endocrine disrupting compounds to source drinking water due to climate change. J. Environ. Qual. Accepted.
503
Benotti, M.J., Stanford, B.D., Wert, E.C., Snyder, S.A., 2009b. Evaluation of a photocatalytic reactor membrane pilot system for the removal of pharmaceuticals and endocrine disrupting compounds from water. Water Res. 43 (6), 1513–1522. Benotti, M.J., Trenholm, R.A., Vanderford, B.J., Holaday, J.C., Stanford, B.D., Snyder, S.A., 2009c. Pharmaceuticals and endocrine disrupting compounds in U.S. drinking water. Environ. Sci. Technol. 43 (3), 597–603. Bolton, J.R., Bircher, K.G., Tumas, W., Tolman, C.A., 1996. Figuresof-merit for the technical development and application of advanced oxidation processes. J. Adv. Oxid. Technol. 1 (1), 13–17. Chen, W., Westerhoff, P., Leenheer, J.A., Booksh, K., 2003. Fluorescence excitation-emission matrix regional integration to quantify spectra for dissolved organic matter. Environ. Sci. Technol. 37 (24), 5701–5710. Even-Ezra, I., Mizrahi, A., Gerrity, D., Snyder, S., Salveson, A., Lahav, O., 2009. Application of a novel plasma-based advanced oxidation process for efficient and cost effective destruction of refractory organics in tertiary effluents and contaminated groundwater. Desalin. Water Treat. Accepted. Gerrity, D., Mayer, B., Ryu, H., Crittenden, J., Abbaszadegan, M., 2009. A comparison of pilot-scale photocatalysis and enhanced coagulation for disinfection byproduct mitigation. Water Res. 43, 1597–1610. Huber, M.M., Canonica, S., Park, G., von Gunten, U., 2003. Oxidation of pharmaceuticals during ozonation and advanced oxidation processes. Environ. Sci. Technol. 37, 1016–1024. Hwang, Y.H., Jo, Y.M., 2005. Decomposition of odorous gases in a pilot-scale nonthermal plasma reactor. J. KOSAE 21 (2), 57–65. Ighigeanu, D., Martin, D., Zissulescu, E., Macarie, R., Oproiu, C., Cirstea, E., Iovu, H., Calinescu, I., Iacob, N., 2005. SO2 and NOx removal by electron beam and electrical discharge induced non-thermal plasmas. Vacuum 77, 493–500. Kim, S.D., Cho, J., Kim, I.S., Vanderford, B.J., Snyder, S.A., 2007. Occurrence and removal of pharmaceuticals and endocrine disruptors in South Korean surface, drinking, and waste waters. Water Res. 41, 1013–1021. Kim, H., Han, J., Sakaguchi, Y., Minami, W., 2008. Simultaneous oxidation of NOx and SO2 by a new non-thermal plasma reactor enhanced by catalyst and additive. Plasma Sci. Technol. 10 (1), 53–56. 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, 1999–2000: A national reconnaissance. Environ. Sci. Technol. 36, 1202–1211. Locke, B.R., Sato, M., Sunka, P., Hoffmann, M.R., Chang, J.S., 2006. Electrohydraulic discharge and nonthermal plasma for water treatment. Ind. Eng. Chem. Res. 45, 882–905. Masuda, S., Hosokawa, S., Tu, X., Wang, Z., 1995. Novel plasma chemical technologiesdPPCP and SPCP for control of gaseous pollutants and air toxics. J. Electrostatics 34, 415–438. 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 (5), 389–404. Pekarek, S., 2003. Non-thermal plasma ozone generation. Acta Polytech. 43 (6), 47–51. Pereira, V.J., Linden, K.G., Weinberg, H.S., 2007. Evaluation of UV irradiation for photolytic and oxidative degradation of pharmaceutical compounds in water. Water Res. 41 (19), 4413–4423. Rosal, R., Rodriguez, A., Perdigon-Melon, J.A., Petre, A., GarciaCalvo, E., 2009. Oxidation of dissolved organic matter in the effluent of a sewage treatment plant using ozone combined with hydrogen peroxide (O3/H2O2). Chem. Eng. J. 149, 311–318.
504
water research 44 (2010) 493–504
Snyder, S.A., Wert, E.C., Lei, H., Westerhoff, P., Yoon, Y., 2007. Removal of EDCs and Pharmaceuticals in Drinking and Reuse Treatment Processes. Water Research Foundation, Denver, CO. Snyder, S.A., Westerhoff, P., Yoon, Y., Sedlak, D.L., 2003. Pharmaceuticals, personal care products, and endocrine disruptors in water: implications for the water industry. Environ. Eng. Sci. 20 (5), 449–469. Ternes, T.A., 1998. Occurrence of drugs in German sewage treatment plants and rivers. Water Res. 32 (11), 3245–3260. Ternes, T.A., Meisenheimer, M., McDowell, D., Sacher, F., Brauch, H., Haist-Gulde, B., Preuss, G., Wilme, U., ZuleiSeibert, N., 2002. Removal of pharmaceuticals during drinking water treatment. Environ. Sci. Technol. 36, 3855–3863. Thevenet, F., Guaitella, O., Puzenat, E., Herrmann, J.M., Rousseau, A., Guillard, C., 2007. Oxidation of acetylene by photocatalysis coupled with dielectric barrier discharge. Catal. Today 122, 186–194.
Trenholm, R.A., Vanderford, B.J., Snyder, S.A., 2009. On-line SPE LC-MS/MS analysis of pharmaceutical indicators in water: a green alternative to conventional methods. Talanta 79, 1425–1432. Wert, E.C., Rosario-Ortiz, F.L., Snyder, S.A., 2009. Using ultraviolet absorbance and color to assess pharmaceutical oxidation during ozonation of wastewater. Environ. Sci. Technol. 43, 4858–4863. 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. Environ. Sci. Technol. 39, 6649–6663. Westerhoff, P., Moon, H., Minakata, D., Crittenden, J., 2009. Oxidation of organics in retentates from reverse osmosis wastewater reuse facilities. Water Res. 43 (16), 3992–3998. Yates, R.S., Stenstrom, M.K., 2000. Gravimetric sampling procedure for aqueous ozone concentrations. Water Res. 34 (4), 1413–1416.
water research 44 (2010) 505–520
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Single-walled carbon nanotubes dispersed in aqueous media via non-covalent functionalization: Effect of dispersant on the stability, cytotoxicity, and epigenetic toxicity of nanotube suspensions Alla L. Alpatova a, Wenqian Shan a, Pavel Babica b, Brad L. Upham b,c, Adam R. Rogensues a, Susan J. Masten a, Edward Drown d,1, Amar K. Mohanty d,2, Evangelyn C. Alocilja e, Volodymyr V. Tarabara a,* a
Department of Civil and Environmental Engineering, Michigan State University, East Lansing, MI 48824, USA Center for Integrative Toxicology, Michigan State University, East Lansing, MI 48824, USA c Department of Pediatrics and Human Development, Michigan State University, East Lansing, MI 48824, USA d School of Packaging, Michigan State University, East Lansing, MI 48824, USA e Department of Biosystems and Agricultural Engineering, Michigan State University, East Lansing, MI 48824, USA b
article info
abstract
Article history:
As the range of applications for carbon nanotubes (CNTs) rapidly expands, understanding
Received 2 June 2009
the effect of CNTs on prokaryotic and eukaryotic cell systems has become an important
Received in revised form
research priority, especially in light of recent reports of the facile dispersion of CNTs in
14 September 2009
a variety of aqueous systems including natural water. In this study, single-walled carbon
Accepted 17 September 2009
nanotubes (SWCNTs) were dispersed in water using a range of natural (gum arabic,
Available online 30 September 2009
amylose, Suwannee River natural organic matter) and synthetic (polyvinyl pyrrolidone, Triton X-100) dispersing agents (dispersants) that attach to the CNT surface non-covalently
Keywords:
via different physiosorption mechanisms. The charge and the average effective hydrody-
Single-walled carbon nanotubes
namic diameter of suspended SWCNTs as well as the concentration of exfoliated SWCNTs
Dispersion
in the dispersion were found to remain relatively stable over a period of 4 weeks. The
Non-covalent functionalization
cytotoxicity of suspended SWCNTs was assessed as a function of dispersant type and
Cytotoxicity
exposure time (up to 48 h) using general viability bioassay with Escherichia coli and using
Epigenetic toxicity
neutral red dye uptake (NDU) bioassay with WB-F344 rat liver epithelia cells. In the E. coli viability bioassays, three types of growth media with different organic loadings and salt contents were evaluated. When the dispersant itself was non-toxic, no losses of E. coli and WB-F344 viability were observed. The cell viability was affected only by SWCNTs dispersed using Triton X-100, which was cytotoxic in SWCNT-free (control) solution. The epigenetic toxicity of dispersed CNTs was evaluated using gap junction intercellular communication (GJIC) bioassay applied to WB-F344 rat liver epithelial cells. With all SWCNT suspensions except those where SWCNTs were dispersed using Triton X-100 (wherein GJIC could not be measured because the sample was cytotoxic), no inhibition of GJIC in the presence of SWCNTs was observed. These results suggest a strong dependence of the toxicity of
* Corresponding author. Tel.: þ517 432 1755; fax: þ517 355 0250. E-mail address:
[email protected] (V.V. Tarabara). 1 Present address: Department of Chemical Engineering and Materials Science, Michigan State University, East Lansing, MI 48824, USA. 2 Present address: Department of Plant Agriculture, University of Guelph, 50 Stone Rd. E., Guelph, Ontario, N1 G 2W1 Canada. 0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2009.09.042
506
water research 44 (2010) 505–520
SWCNT suspensions on the toxicity of the dispersant and point to the potential of noncovalent functionalization with non-toxic dispersants as a method for the preparation of stable aqueous suspensions of biocompatible CNTs. ª 2009 Elsevier Ltd. All rights reserved.
1.
Introduction
The discovery (Iijima, 1991) and subsequent extensive characterization of carbon nanotubes (CNTs) have revealed a class of materials with extraordinary electrical, mechanical, and thermal properties (Tasis et al., 2006). The wider application of CNTs in electronic, optical, sensing, and biomedical fields has been impeded by the low solubility of as-produced CNTs in polar liquids and by the strong tendency of CNTs to aggregate due to hydrophobic–hydrophobic interactions (Lin et al., 2004). Dispersing CNTs and ensuring long-term stability of CNTs suspended in a liquid medium have proved especially challenging for aqueous systems.
2002; Star et al., 2002; Islam et al., 2003; Moore et al., 2003; McDonald et al., 2006; Grossiord et al., 2007). These treatments were applied both in the absence (Salzmann et al., 2007) and in the presence of solubilizing agents: NOM (Hyung et al., 2007), Triton X-100 (Islam et al., 2003), Triton X-405 (Chappell et al., 2009), PVP-1300 (Didenko et al., 2005), GA (Bandyopadhyaya et al., 2002), and starch (Star et al., 2002). A three-step approach to solubilizing single-walled carbon nanotubes (SWCNTs) with amylose was suggested by Kim et al. (Kim et al., 2004): dispersion of SWCNT in water by sonication followed by treatment with amylose in dimethylsulfoxide (DMSO)–H2O mixture, followed by sonication allowing for molecularly controlled encapsulation of CNTs.
1.1.
1.2.
Dispersing CNTs in aqueous media
Recent efforts on the development of efficient and facile methods of dispersing CNTs in aqueous media have been focused on the hydrophilization of CNT with molecules that bind to the CNT surface non-covalently (O’Connell et al., 2001; Bandyopadhyaya et al., 2002; Star et al., 2002; Islam et al., 2003; Moore et al., 2003; Didenko et al., 2005; Wang et al., 2005; Bonnet et al., 2007; Grossiord et al., 2007; Hyung et al., 2007; Liu et al., 2007). Such non-covalent functionalization has great promise as the modification-induced changes in the electronic and mechanical properties of CNTs are minimized (Yang et al., 2007). Various surfactants (Islam et al., 2003; Moore et al., 2003; Grossiord et al., 2007), synthetic polymers (e.g., polyvinyl pyrrolidone (O’Connell et al., 2001; Didenko et al., 2005), poly(ethylene glycol) (Vaisman et al., 2006), polyphosphazene (Park et al., 2006)), natural organic matter (NOM) (Hyung et al., 2007; Liu et al., 2007; Saleh et al., 2009), biomolecules (e.g., proteins (Karajanagi et al., 2006; Zong et al., 2007), aminoacids (Georgakilas et al., 2002), DNA (Enyashin et al., 2007)), and carbohydrates (e.g., cyclodextrines (Dodziuk et al., 2003), amylose (Bonnet et al., 2007), starch (Star et al., 2002), and GA (Bandyopadhyaya et al., 2002)) have been evaluated as dispersants for CNTs. Dispersion via non-covalent functionalization is based on the direct contact between a CNT and a dispersant molecule (Liu et al., 1998; Grossiord et al., 2007). Such modification of the CNT surface facilitates the disaggregation (i.e. debundling) of CNT bundles into smaller diameter bundles (Liu et al., 1998) or even individual CNTs (O’Connell et al., 2002; Hyung et al., 2007) and leads to the stabilization of suspended CNTs via steric or electrostatic repulsion mechanisms or both. (See Supporting Documentation (SD), Section S.1.1, for a brief review of mechanisms of non-covalent functionalization of CNTs.) The dispersion of CNTs in water has been enhanced by mixing (Didenko et al., 2005; Hyung et al., 2007), sonication (Bandyopadhyaya et al., 2002; Liu et al., 2007; Salzmann et al., 2007), or mixing followed by sonication (O’Connell et al., 2001,
Stability of CNT suspensions in water
Previous studies of the long-term changes in suspensions of dispersed CNTs have focused on monitoring changes in the concentration of suspended CNTs (Jiang and Gao, 2003; Tseng et al., 2006; Lee et al., 2007; Marsh et al., 2007). By measuring UV–vis absorption at certain wavelength: 253 nm (Jiang and Gao, 2003), 300 nm (Sinani et al., 2005), 500 nm (Bahr et al., 2001; Lee et al., 2007), 530 nm (Marsh et al., 2007), and 800 nm (Hyung et al., 2007) the change in the concentration of suspended CNTs with time was determined. Aqueous suspensions of non-functionalized CNTs are known to be unstable. There are considerable quantitative differences, however, in the reported stability data for nonfunctionalized CNTs. The concentration of unmodified multiwalled carbon nanotubes (MWCNTs) suspended in deionized water was reported to decline 86 % over 2 h in one study (Marsh et al., 2007) and only 50% over 500 h in another study (Jiang and Gao, 2003). The suspensions of unmodified SWCNTs in deionized water were found to completely precipitate after only 4 h (Tseng et al., 2006). It was found that non-covalent modification of CNT surface drastically improved the stability of CNT suspensions (Jiang and Gao, 2003; Tseng et al., 2006; Hyung et al., 2007; Lee et al., 2007; Marsh et al., 2007). The stability of the suspension of non-covalently functionalized CNTs was found to be a function of the type (Hyung et al., 2007; Lee et al., 2007) and concentration (Chappell et al., 2009) of the dispersant, CNT length (Marsh et al., 2007) and the presence of inorganic salts in the solution (Saleh et al., 2009). SWCNTs stabilized with oligothiophene-terminated poly(ethylene glycol) produced suspensions that were significantly more stable than CNTs dispersed in water with the aid of sodium dodecyl sulfate (SDS) and Pluronic F127 (Lee et al., 2007). CNT suspensions in model Suwannee River NOM (SRNOM) solutions and in Suwannee River water were found to be considerably more stable than suspensions of CNTs dispersed in 1% aqueous solution of SDS (Hyung et al., 2007). Chappell et al. showed
water research 44 (2010) 505–520
that the stability of MWCNTs dispersed using two types of humic acids, Triton X-405, Brij 35, and SDS was enhanced in dose-dependent manner (Chappell et al., 2009). In a study of the aggregation kinetics of MWCNTs in aquatic systems, Suwannee River humic acid was shown to significantly enhance stability of MWCNTs suspensions in the presence of monovalent and divalent salts (Saleh et al., 2009). While studying the stability of MWCNTs in water, Marsh and coworkers (Marsh et al., 2007) observed that wrapping of the annealed CNTs with 1% SDS or charge doping increased their stability in the suspension; the authors demonstrated that shorter CNTs form more stable dispersions than longer CNTs of the same diameter. In summary, there is growing evidence that by the appropriate choice of a dispersant that modifies CNT surface non-covalently, highly stable CNT suspensions can be produced. However, to date there have been no systematic investigations that correlated long-term changes in size and charge of CNTs dispersed via non-covalent functionalization with the dispersion stability.
1.3.
Toxicity of CNTs
1.3.1.
Toxicity of CNTs towards eukaryotic cells: cytotoxicity
Most nanomaterial toxicity studies have been performed with mammalian cells, in particular with lung and skin cell cultures reflecting the understanding that the most likely routes of an organism’s exposure to nanomaterials are respiratory and dermal contact. With the development of methods of CNT dispersion in aqueous media, the assessment of the toxicity of such dispersed CNTs becomes very important in view of their increased mobility and potential to enter water supplies. While the toxicity of CNTs was studied with respect to the type of the CNT (MWCNT versus SWCNT) (Ding et al., 2005; Jia et al., 2005; Kang et al., 2008a), surface functionalization (Sayes et al., 2006; Yu et al., 2007; Meng et al., 2009; Yun et al., 2009), CNT length (Muller et al., 2005; Kang et al., 2008b; Simon-Deckers et al., 2008), exposure dose (Cherukuri et al., 2006; Flahaut et al., 2006; Sayes et al., 2006; Kang et al., 2007; Pulskamp et al., 2007; Yang et al., 2008, 2009; Ye et al., 2009), degree of purification (Fiorito et al., 2006; Flahaut et al., 2006; Elias et al., 2007; Simon-Deckers et al., 2008), and degree of dispersion (Wick et al., 2007), only very limited information exists on the biocompatibility of CNTs as a function of surface coating. Several hypotheses have been put forth to explain the likely pathways of CNT toxicity: (i) oxidative stress induced by the formation of reactive oxygen species (ROS) generated at the surface of CNTs (Manna et al., 2005; Yang et al., 2008; Ye et al., 2009); (ii) the presence of residual catalyst, which is used during the CNT manufacturing process (Kang et al., 2008a); (iii) physical contact between a CNT and a cell (Kang et al., 2007); or (iv) a combination of these factors (Kang et al., 2008a). The dispersant used for the stabilization of CNTs in suspension was suggested as another possible cause of the observed nanotube toxicity (Sayes et al., 2006) (See SD, Section S.1.2 for a brief review of possible mechanisms of CNT toxicity). In most studies of the toxicity of non-covalently functionalized CNTs the dispersants were surfactants. In vivo assessment of the toxicity of SWCNTs modified with Pluronic F108 administered intravenously to rabbits showed the
507
absence of the acute toxicity (Cherukuri et al., 2006). Pluronic F-127-coated MWCNTs did not affect cell viability, apoptosis, and ROS formation in mouse and human neuroblastoma cells (Vittorio et al., 2009), and mouse cerebral cortex (Bardi et al., 2009). In contrast, MWCNTs dispersed using Pluronic F68 caused cell death, changes in cell size and complexity, ROS production, interleukin-8 (IL-8) gene expression and nuclear factor (NF)-jB activation (Ye et al., 2009). CNTs dispersed using Tween 80 were toxic to mesothelioma cells (Wick et al., 2007), led to an inflammation of murine allergic airway with augmented humoral immunity (Inoue et al., 2009), and induced inflammatory and fibrotic responses when intracheally administrated to rats (Muller et al., 2005). More ROS were observed upon exposure of human lung epithelial or primary bronchial epithelial cells to SWCNTs dispersed using dipalmitoylphosphatidylcholine, a major component of lung surfactant (Herzog et al., 2009), as compared to dipalmitoylphosphatidylcholine-free samples. Low toxicity was observed in in vivo experiments when SWCNTs were dispersed using Tween 80 and intravenously injected into mice (Yang et al., 2008). In another study, SWCNTs dispersed in 1% SDS aqueous solution showed no cytotoxicity with respect to the human alveolar epithelial cells (Wo¨rle-Knirsch et al., 2006). CNTs dispersed using Pluronic PF-127 solution did not affect viability, apoptosis and ROS generation in the human neuroblastoma cells after 3 days of incubation; however, cell proliferation decreased as incubation time increased to 2 weeks (Vittorio et al., 2009). In the only paper that mentioned the potential toxicity effect of the dispersant, the authors suggested that excess surfactant was responsible for the observed increase in toxicity; in this work, the controlled exposure of cells to 1% Pluronic F108 produced a 10% decrease in the cells viability (Sayes et al., 2006). Very little is known about the effect of non-covalent wrapping with dispersants other than surfactants on the biocompatibility of CNTs. In vitro cytotoxicity of CNTs wrapped with poly(methyl vinylketone) decorated with a-N-acetylgalactosamine (Chen et al., 2006), nano-1 peptide (Chin et al., 2007) and cholesterolend-capped poly(2-methacryloyloxyethyl phosphorylcholine) (Xu et al., 2008) were examined after their contact with human lung epithelial-like cells, normal skin fibroblasts and human umbilical vein endothelial cell line. No impact on cell growth and proliferation was demonstrated in all three cases. Cytotoxicity of GA-stabilized MWCNTs upon exposure to A549 cells was observed by LDH, XTT and MTT bioassays in (SimonDeckers et al., 2008). Two possible reasons were hypothesized to be responsible for this effect: (i) increased availability of GAstabilized MWCNTs and (ii) different intercellular accumulation pathway of GA-stabilized MWCNTs as compared to when carbon nanotubes were exposed in bundles.
1.3.2. Toxicity of CNTs towards eukaryotic cells: genotoxicity and epigenetic toxicity Most toxicological studies of CNTs focused on the evaluation of cytotoxicity (Sayes et al., 2006; Elias et al., 2007; Gutierrez et al., 2007; Kang et al., 2007; Wick et al., 2007; Kang et al., 2008a,b; Yang et al., 2008; Bardi et al., 2009; Herzog et al., 2009; Inoue et al., 2009; Kang et al., 2009; Vittorio et al., 2009; Yang et al., 2009). However, genotoxicity (Kang et al., 2008a; Di Sotto et al., 2009; Lindberg et al., 2009; Wirnitzer et al., 2009; Yang et al., 2009) and
508
water research 44 (2010) 505–520
epigenetic toxicity (see SD, Section S.1.3) (Upham et al., 1994); (Trosko et al., 1998) are other possible causes of cell damage or other adverse effects. The genotoxicity of SWCNTs dispersed using fetal bovine serum at (5–10) mg/mL concentration, with respect to primary mouse embryo fibroblasts was demonstrated using Comet assay (Yang et al., 2009). In another recent study, CNTs (dispersed in BEGM cell culture medium and subjected to ultrasonication) induced a dose-dependent increase in DNA damage as indicated by Comet assay and caused a significant increase in micronucleated cells (micronucleus assay) in human bronchial epithelial cells (Lindberg et al., 2009). Kang et al. observed high levels of stress-related gene products in Escherichia coli upon its exposure to CNTs, with the quantity and magnitude of expression being much higher in the presence of SWCNTs (Kang et al., 2008a); in this study CNTs were either deposited on a PVDF membrane surface or dispersed in saline solution. No mutagenic effect was observed in Salmonella microcosme test with baytubes (high purity MWCNTs agglomerates, sonicated 10 min in deionized water) (Wirnitzer et al., 2009) and in bacterial reverse mutation assay (Ames test) with Salmonella typhimurium TA 98 and TA 100 strains and with E. coli WP2uvrA strain exposed to MWCNTs dispersed in DMSO (Di Sotto et al., 2009). There have been no reports to date on the epigenetic toxicity of carbon nanomaterials.
1.3.3.
2007). Lyon et al. showed that in media with high salt content, the size of nC60 aggregates tends to increase as compared to that observed in low salt media (Lyon et al., 2005). In another study, MWCNTs were found to stimulate growth of unicellular protozoan Tetrahymena pyriformis in growth medium, which contained proteose peptone, yeast extract, and glucose; the increase in growth was attributed to the formation of peptone-MWCNTs conjugates, which were taken up by the microorganism (Zhu et al., 2006). While the aforementioned studies indicate that salt and organic composition of the medium, in which exposure studies are performed, may influence the interaction of CNTs with bacteria, there have been no studies that comparatively evaluated CNT toxicity in growth media with different organic loadings and salt contents.
1.5.
Objectives of this study
This study addressed some of the knowledge gaps identified above. Aqueous suspensions of SWCNTs functionalized by a range of non-covalently bound dispersants of natural (NOM, GA, amylose) and synthetic (PVP, Triton X-100) origin, were prepared and evaluated in terms of their physicochemical and toxicity properties. The study pursued the following objectives:
Toxicity of CNTs in prokaryotic cells
Most toxicity studies have focused on the effect of CNTs on mammalian cell lines. Only limited information exists on the cytotoxic effects of CNTs towards bacterial cells. Recently, antimicrobial activity of SWCNTs (Kang et al., 2007, 2008a, 2009) and MWCNTs (Kang et al., 2008a,b, 2009) either suspended in aqueous solution (Kang et al., 2007, 2008a) or deposited on the surface of a PVDF microfilter (Kang et al., 2007, 2008a,b, 2009) towards gram-negative bacteria (E. coli and Pseudomonas aeruginosa (Kang et al., 2007, 2008a,b, 2009) and gram-positive (Staphylococcus epidermidis and Bacillus subtilis (Kang et al., 2009)) bacteria was reported. It was suggested that membrane damage to the cells was caused by the direct physical contact between the CNT and cell (Kang et al., 2007) or by a combination of direct physical contact and oxidative stress (Kang et al., 2008a). No effect on the percentage of E. coli inactivation was observed upon exposure of SRNOM-stabilized SWCNTs as compared of SWCNTs dispersed in the absence of SRNOM (Kang et al., 2009). Significant antimicrobial activity of CNTs composite films against Staphylococcus aureus and Staphylococcus warneri was reported in a separated study (Narayan et al., 2005). In contrast to the findings of above studies, no inhibition of E. coli growth and proliferation was reported in a study where microchannel-structured MWCNTs scaffolds were immersed into the culture medium with the cells (Gutierrez et al., 2007).
1.4. Effect of growth media on the toxicity of nanoparticles The existing literature highlights the importance of the growth media in cytotoxicity testing and links the apparent cytotoxic effect of the nanomaterial to the salt and organic content of the culture and related physicochemical characteristics of the nanomaterial (Lyon et al., 2005; Tong et al.,
(i) To evaluate the long-term stability and its physicochemical determinants for non-covalently functionalized SWCNTs as a function of dispersant type. The evolution of the concentration, size, and charge of suspended SWCNTs was studied over the period of 28 days. (ii) To assess time-dependent cytotoxicity of non-covalently functionalized SWCNTs with respect to bacteria and mammalian cells as a function of dispersant type and growth media. The effect of SWCNTs on E. coli was studied by measuring cell viability after 3 h, 24 h, and 48 h of exposure in three types of growth media with different organic loadings and salt contents. The cytotoxicity of dispersed SWCNTs towards mammalian cells was assessed in NDU bioassay with rat liver epithelial cells after 30 min and 24 h of incubation. (iii) To assess time-dependent epigenetic toxicity of non-covalently functionalized SWCNTs as a function of dispersant type. In this study, we evaluated epigenetic toxicity of the SWCNTs to rat liver epithelial cells as a function of dispersant type in GJIC bioassay after 30 min and 24 h of incubation.
2.
Approach
The set of chemically diverse dispersants was chosen based on their demonstrated effectiveness in solubilizing CNTs (Bandyopadhyaya et al., 2002; O’Connell et al., 2002; Star et al., 2002; Islam et al., 2003; Moore et al., 2003; Hyung et al., 2007; Liu et al., 2007) and potential adverse effects (Burnette, 1960; Chourasia and Jain, 2004; Dayeh et al., 2004; Schmitt et al., 2008). NOM, GA and PVP have LD50 doses of (54.8–58.5) mg (intravenous administration in mice) (EMEA, 1999), 2,000 mg (oral administration in rats) (Schmitt et al., 2008), and 100,000 mg (oral administration in rats) (Burnette, 1960) per kg
water research 44 (2010) 505–520
of body weight, respectively. Amylose has been reported to be used in a colon-specific drug delivery due to its low toxicity and high biodegradability (Chourasia and Jain, 2004). Triton X-100 was reported to be toxic to protozoa, fish, and mammalian cells (Dayeh et al., 2004). Three batches of dispersed SWCNTs were prepared. The first batch was used to comprehensively evaluate the longterm stability of SWCNT suspensions over a period of 4 weeks in terms of concentration, effective hydrodynamic diameter, and z-potential of stabilized SWCNTs. The second batch was used to evaluate the viability of E. coli cells after their contact with dispersed SWCNTs by the quantification of colony forming units. To elucidate the effects of ionic and organic composition of the growth medium, the cytotoxicity of the SWCNTs suspended in three growth media of different organic and salt compositions was evaluated. First, in order to assess the acute cytotoxicity of dispersed SWCNTs, we conducted experiments in 0.1 M NaCl. Then, in order to investigate the effect of SWCNTs on the ability of E. coli form colonies over time, we employed three types of growth media: (i) LB medium with higher organic chemical and salt content, (ii) MD medium with low salt content and organic load, and (iii) 0.1 M NaCl solution. The third batch was used to evaluate cyto- and epigenetic toxicity of the dispersed SWCNTs against rat liver epithelial cells by the (i) NDU and (ii) GJIC bioassays. The NDU assesses cell viability by measuring the accumulation of neutral red dye in lysosomes, which depends on the cell’s capacity to maintain pH gradients through the maintaining membrane integrity and production of adenosine triphosphate. The amount of dye incorporated by the cell is quantified spectrophotometrically (Borenfreund and Puerner, 1985). The NDU bioassay has been used to assess cytotoxicity of CNTs (Flahaut et al., 2006). The GJIC bioassay (Borenfreund and Puerner, 1985; Weis et al., 1998; Herner et al., 2001; Satoh et al., 2003) utilizes the ability of epigenetic tumor promoters to alter level of GJIC (Yamasaki, 1990; Trosko et al., 1991). The degree of GJIC was quantified by measuring the distance (area) the fluorescent dye (1200 Da) travels (occupies) between the cells after a given time. Both NDU and GJIC bioassays were carried out with WB-F344 rat liver epithelial cells exposed to dispersed SWCNTs for different periods of time. The WB-F344 cell line was chosen because these normal diploid rat liver epithelial cells have already used in numerous studies of cytotoxic or epigenetic effects (e.g., Herner et al., 2001) thus allowing us to have a comparative basis.
3.
Materials and methods
3.1.
CNTs
SWCNTs (purity > 90%), produced by catalytic chemical vapor deposition, were obtained from Cheap Tubes, Inc (Brattleboro, VT) and used as received. The SWCNTs were used as obtained, allowing us to mimic what might occur in the environment. As indicated by the manufacturer, the SWCNTs had an inside diameter in the range of 0.8 nm to 1.6 nm, an outer diameter in the range of 1 nm to 2 nm and were 5 mm to 30 mm in length.
509
3.2. Non-covalent functionalization: dispersants and dispersion procedures 3.2.1.
Dispersants
Gum arabic (approx. 250 kDa; a complex mixture of saccharides and glycoproteins obtained from the acacia tree), PVP (approx. 29 kDa), Triton X-100 (approx. 625 Da; polyethylene glycol p-(1,1,3,3-tetramethylbutyl)-phenyl ether) and amylose (molecular weight not specified; a polymeric form of glucose and a constituent of potato starch) were purchased from Sigma–Aldrich (Milwaukee, WI). SRNOM reverse osmosis isolation was obtained from International Humic Substances Society (St Paul, MN). The molecular structure of dispersants used in this study is presented in Fig. 1.
3.2.2.
Dispersion procedure
Aqueous solutions of PVP, Triton X-100 were prepared by dissolving PVP (40 mg) and Triton X-100 (0.4 mL) in 40 mL of water and adjusting the pH of both solutions to 7 with 0.1 M HCl. The aqueous solution of GA was prepared by mixing 1 g of GA in 50 mL of water and adjusting the pH of the mixture to 7 with 0.1 M HCl; this mixture was left to settle for 24 h and then 40 mL of supernatant was collected for use in the stability or toxicity studies. To prepare NOM solutions, two flasks of 40 mL of water, each containing 8 mg of NOM, were stirred for 24 h. Each solution was then filtered through a 0.22 mm filter under vacuum. The pH of one solution was kept at its original value (approx. 3.5) while the pH in another solution was adjusted to 7 with 0.1 M NaOH. SWCNTs were added to the above solutions to result in a 1 mg (SWCNT)/mL loading and were sonicated in Aquasonic 50 T water bath (VWR Scientific Products Corp, West Chester, PA) for 20 min. SWCNTs were dispersed with amylose based on modified version of the three-step approach reported by Kim et al. (Kim et al., 2004). SWCNTs (40 mg) were sonicated in 25 mL of water (pH 7) at approximately 75 W for 5 min using Sonicator 3000 (Misonix, Inc., Farmingdale, NY) equipped with a microprobe. 100 mg of amylose in 6.28 mL of DMSO were prepared and added to the sonicated suspension of SWCNT in water so that the DMSO/water ratio was 20% by volume. The resulting mixture was sonicated for another 5 min in order to remove the excess amylose and DMSO. The suspension was sonicated, centrifuged and refilled with water four times (See SD, Section S.2.3). Following the sonication, SWCNTs suspensions were divided into 12 mL aliquots and each aliquot was transferred into 15 mL centrifuge tube. After 24 h standing time, the top 4 mL of each suspension was withdrawn in order to be used in the stability or toxicity studies. Overall, three batches of dispersed SWCNT suspensions were prepared and were used, correspondingly, for 1) the study of the long-term stability of SWCNTs suspensions, 2) the general viability bioassay with E. coli, and 3) NDU and GJIC bioassays.
3.3. Physicochemical characterization of SWCNT suspensions In order to quantitatively assess the long-term stability SWCNT suspensions, a combination of characterization methods was employed.
510
water research 44 (2010) 505–520
(a-1)
a
GALP
1
1 3
ARAF
3
3
ARAF
ARAF
GALP
O OH
OH
O OH
OH
1
1
3
CH2OH
3
ARAF
1 6
(a-2)
GALP
CH3
OH
1
1
3
GALP
3
1
6
GALP
1
3 3
1
3
RHAP
GALP
RHAP
6
GALP
1
1
1
GA
GA
4
4
1
1
ARAF
ARAF
OH
(a-3)
GALP 6 1
OH
OH
6
6
(a-4) HOOC
O
O
OH H2COH
HO HO
OH OH
OH
GALP = D-GALACTOPYRANOSE ARAF = L-ARABOFURANOSE GA = D-GLUCURONIC ACID RHAP = L-RHAMNOPYRANOSE
b
OH
c HO
O
O
O
N
OH
O
O COOH
H2N
OH
HO
n
HOOC
d
COOH
OH
e O
COH
OH
O
n
C8 H17
O
OH
O OH
n
Fig. 1 – Molecular structure of solubilizers: (a) Gum arabic (a-1) galactose, (a-2) rhamnose, (a-3) arabinose, (a-4) glucuonic acid; (b) Poly(vinyl pyrrolidone); (c) A building block of humic acids, which has a compositional similarity to SRNOM; Dots represent chiral centers; (d) Triton X-100; (e) Amylose.
3.3.1. UV–vis spectrophotometry. Quantification of SWCNTs concentration in suspension The absorbance of SWCNT suspensions over the (200–800) nm wavelength range was measured over a period of 4 weeks (Multi-spec 1501, Shimadzu, Kyoto, Japan). For each type of SWCNT suspension prepared, the absorbance of SWCNT-free solution of the corresponding dispersant was used as a baseline. SWCNT suspensions used in our toxicity studies likely contained both individual and bundled SWCNTs. In fact, DLS measurements indirectly confirmed (see Section 4.2) the presence of dispersed SWCNT bundles and direct TEM imaging (see SD, Section S.3.3, Fig. S5) also showed the presence of SWCNTs bundles (note that TEM results need to be interpreted with caution as evaporation-induced aggregation could have contributed to bundling during TEM sample preparation). Incomplete exfoliation is the most likely and most environmentally relevant scenario; unfortunately, it also
entails difficulties with quantifying the total concentration of suspended nanotubes as larger CNT bundles tend to separate from the suspension. In this study, we used UV–vis spectrophotometry to estimate the concentration of exfoliated SWCNTs in suspensions. It is known that only fully exfoliated SWCNTs absorb in the (200–1200) nm wavelength range. Bundled NTs do not absorb significantly in this range due to the tunnel coupling between metallic and semiconductive SWNTs (Lauret et al., 2004; Grossiord et al., 2007). Therefore, the UV absorbance by a SWCNT suspension can be used to selectively measure the concentration of individually dispersed (i.e. fully exfoliated or debundled) SWCNTs. To determine the concentration of suspended exfoliated SWCNTs, we first prepared a separate set of suspensions with ten-fold reduced SWCNT loading with respect to the CNT loading used in long-term stability and toxicity studies. Four suspensions (with PVP, NOM pH 3.5, NOM pH 7, and GA used
water research 44 (2010) 505–520
as dispersants) with 10-fold reduced SWCNT content (0.1 g/L) were prepared using the same procedures as described in Section 3.2 except that they were subjected to rigorous, prolonged sonication for 1 h at power of (70–80) W using horn sonicator (Sonicator 3000, Misonix, Inc., Farmingdale, NY). No settling of SWCNTs was observed over the short term following the application of this treatment indicating complete dispersion of SWCNTs. UV–vis absorbance spectra were recorded at different dilutions and a calibration curve for absorption at 500 nm (Bahr et al., 2001; Huang et al., 2002; Sinani et al., 2005; Lee et al., 2007; Salzmann et al., 2007) as a function of concentration was constructed, and coefficient of molecular extinction, 3, was determined. This coefficient was used to estimate the concentration of exfoliated SWCNTs in suspensions used in long-term stability and toxicity experiments. Note that by subjecting suspended SWCNTs to a very intense sonication treatment we aimed at maximizing the extent of exfoliation; however, it was not possible to ascertain that the exfoliation was complete. Thus coefficients of molecular extinction and exfoliated SWCNT concentrations determined using the recorded calibration curves were estimated values. SWCNT characteristics measured upon the preparation of CNT suspensions are given in Table S1 in SD.
3.3.2.
Size and charge of dispersed SWCNTs
The effective hydrodynamic diameter and z-potential of suspended SWCNTs were determined after 1, 4, 7, 14, 21, and 28 days of settling. The size and charge were measured by dynamic light scattering (DLS) and phase analysis light scattering techniques, respectively (ZetaPALS, BI_MAS Option, Brookhaven Instrument Corp., Holtsville, NY). The Smoluchowski equation was applied to convert the measured electrophoretic mobility of dispersed SWCNT to z-potential. Transmission electron microscopy (TEM) imaging was used as an auxiliary method to aid in the interpretation of dynamic light scattering data (see SD, Sections S.2.5 and S.3.4).
3.4.
Toxicity assessment: E. coli viability assay
3.4.1.
Media preparation
Luria–Bertani (LB) growth medium was prepared according to the standard procedure (Atlas, 1993). Minimal Davis medium with 90% reduced potassium phosphate concentration (MD medium) was prepared (Fortner et al., 2005). 0.1 M NaCl solution was prepared by dissolving 9 g of NaCl in 1 L of water (pH 7) and autoclaving it for 15 min at 1 bar and 121 C. For the toxicity assessments, each component of LB, MD and NaCl media was prepared in 4-fold higher concentration as compared to original protocol and the aliquot part of the corresponding medium was added to SWCNTs suspension (as further described in ‘‘Quantification of cell viability’’ Section). Luria–Bertani Petri plates were prepared according to the published method (Atlas, 1993).
3.4.2.
Preparation of E. coli culture
E. coli K12 stock was prepared in glycerol and stored at 80 C. Prior to use, the stock was defrosted, and 30 mL of LB medium were inoculated with 5 mL of the stock. After overnight growth at 37 C, 5 mL of this suspension was spread onto LB agar plate and cultured at 37 C. Once distinct colonies were formed, the
511
agar plate was transferred to the refrigerator and kept at 4 C for up to one month. E. coli suspensions to be used in SWCNT cytotoxicity studies were prepared by scraping one colony from the surface of a Petri plate by aseptic loop and immersing the loop into 10 mL of LB or MD media in a 50 mL centrifuge tube. Tubes were placed on a shaker in an incubator 37 C for 12 h. When 0.1 M NaCl was used as an exposure medium in colony forming units bioassay, E. coli were first grown in the LB medium, centrifuged for 5 min at 2250 rpm and washed with 0.1 M NaCl as follows: the supernatant was decanted and replaced with an equal volume of 0.1 M NaCl, vortexed and resuspended by centrifugation. The washing procedure was repeated twice, presuming that after this treatment most of the remaining organic constituents of LB medium were removed from the E. coli suspension.
3.4.3.
Quantification of cell viability
The SWCNTs suspension (1.425 mL), growth medium (475 mL) and 100 mL of the stock E. coli suspension were transferred into a 15 mL centrifuge tube and incubated under gentle shaking at 37 C. Samples were taken after 3, 24, and 48 h and a series of dilutions (104–106) was prepared for each sample. Five samples of 10 mL and 20 mL from each dilution were placed onto an agar plate and incubated at 37 C until distinct colonies developed. Colony forming units (CFU)/1 mL were calculated for each sample. Each experiment was run in triplicates with negative (bacterial suspension with the corresponding amount of ultrapure water) and vehicle (solution of the corresponding dispersant) controls, herein called vehicle control I and vehicle control II, respectively. The results are reported as a fraction of control (FOC) standard deviation (STD), calculated as the ratio of the average number of colonies grown after E. coli exposure to SWCNTs suspensions or vehicle control II to the average number of colonies grown in vehicle control I plates.
3.5. Toxicity assessment: neutral red dye uptake bioassay For the NDU bioassay we adapted a published procedure (Borenfreund and Puerner, 1985; Weis et al., 1998; Satoh et al., 2003). A solution (0.015 w/v) of neutral red dye (3-amino-7(dimethylamino)-2-methylphenazine hydrochloride) in D-medium was incubated at 37 C for 2 h and filtered through a 0.22 mm syringe filter (Millipore Corp., New Bedford, MA) to remove undissolved dye and ensure sterile conditions. Confluent WB-F344 cells (see SD, Section S.2.6 for details on the preparation of the cells) were exposed to 500 mL of each of SWCNTs suspension and incubated at 37 C for 30 min and 24 h under gentle shaking in a humidified atmosphere containing 5% CO2. After cells were exposed to SWCNTs, the exposure medium was removed by aspiration and the cells were washed with 1 mL of phosphate saline buffer (PBS). Following washing, 2 mL of the neutral red dye solution per plate was added and the cells were incubated for 1 h at 37 C in the humidified atmosphere containing 5% CO2. Upon incubation, the cells were rinsed three times with PBS, and 2 mL of aqueous solution containing 1% acetic acid and 50% ethanol was added to each plate to lyse the cells. 1.5 mL of the lysate was transported into 2 mL microcentrifuge tube and the optical density was
512
water research 44 (2010) 505–520
recorded at 540 nm using a Beckman DU 7400 diode array detector (Beckman Instruments, Inc., Schaumburg, IL). The background absorbance was measured at 690 nm and then subtracted from the original absorbance. Each experiment was conducted in triplicates. The neutral red dye uptake was reported as the FOC (absorption of neutral red in the chemically treated sample divided by the absorption of neutral red in the nontreated control I sample). FOC values of 1.0 indicates noncytotoxic response while FOC values <0.8 indicate that less dye was absorbed by cells and that chemical is cytotoxic at that dose (El-Fouly et al., 1987; Satoh et al., 2003).
3.6. Toxicity assessment: gap junction intercellular communication (GJIC) bioassay The principles of GJIC bioassay were originally described in (Borenfreund and Puerner, 1985) and further developed in (Weis et al., 1998; Herner et al., 2001; Satoh et al., 2003). SWCNTs suspensions (500 mL) were introduced to the cell culture plates with confluent WB-F344 cells and incubated for 30 min and 24 h at 37 C in humidified atmosphere containing 5% CO2 under gentle shaking. Then, cells were washed with PBS solution and 1 mL of 0.05% solution of Lucifer yellow fluorescent dye in PBS buffer was added to each plate. Three different scrapes were made on the bottom of a cell culture plate using a surgical steel scalpel blade. The dye was absorbed through the monolayer of confluent cells. The transfer of dye through gap junctions lasted 3 min, followed by a thorough rinse with PBS solution to remove extracellular dye. Cells were fixed with 0.5 mL of 4% formalin solution in PBS. The migration of the dye in the cells was observed at 200 magnification using a Nikon epifluorescence microscope equipped with a Nikon Cool Snap EZ CCD camera and the images were digitally acquired using a Nikon NIS-Elements F2.2 imaging system. The fluorescence area of the dye migration from the scrape line was quantified using Image J 1.40 g (a public domain Java image processing program, http:// rsbweb.nih.gov/ij/index.html). Each experiment was performed in duplicate with same controls as in the NDU bioassay. Data are reported as the FOC, which is the ratio of the area of dye spread in the chemically treated sample to the area of dye spread in the vehicle control I samples. A FOC value of 1.0 corresponds to full communication between cells, FOC values of (0.5–0.9) represent partial GJIC inhibition, (0.3–0.5) values correspond to significant inhibition of the GJIC and values <0.3 denote complete inhibition of the GJIC (Weis et al., 1998). Student’s t-tests at 95% confidence interval were run for all three bioassays to determine whether the difference between the results obtained in the control I plates and the results in the tested plates were significant. The concentrations of SWCNTs upon addition to suspension of cells in the general E. coli viability bioassay, and in NDU and GJIC bioassays are reported in Table 1.
4.
Results and discussion
The knowledge of the evolution of physicochemical properties (e.g., charge and size) of dispersed CNTs can provide insights into the basis for both CNT stability and possible toxicity.
4.1.
Concentration of dispersed SWCNTs in suspensions
First, SWCNTs were suspended directly in ultrapure water without a dispersant; this suspension was used as a control. After sonication for 20 min, the mixture became a lightly colored suspension that completely separated into a dark subnatant layer and a clear supernatant after only 1 h of settling. In contrast, 20 min of sonication of the SWCNT–water mixture in the presence of a dispersant produced relatively stable colored suspensions (see SD, Fig. S1) that precipitated gradually with the precipitation rate varying as a function of the type of dispersant. The amount of the precipitate that formed at the bottom of a centrifuge tube visibly increased with settling time for all types of SWCNT suspensions. Immediately after sonication, all SWCNT suspensions prepared using the different dispersants appeared black. After 4 weeks of settling, in suspensions of NOM-stabilized and amylose-stabilized SWCNTs the color of the aqueous solution of the dispersant was revealed. After 7 weeks of settling, the color change could be detected in all samples with the suspension of amylose-stabilized SWCNTs (see SD, Fig. S1) undergoing the most dramatic change to an almost transparent supernatant phase (Figure S1 (b)). The UV–vis absorption measurements were performed to determine the concentration of suspended exfoliated SWCNTs as a function of settling time. The calibration curves for the intensity of the absorbance at 500 nm as a function of SWCNT concentration (see Section 2.3) were recorded for all suspensions (see SD, Fig. S2) except for amylose-stabilized SWCNTs. Because of the observed loss of SWCNTs during centrifugation/washing steps for amylose-dispersed SWCNTs (see SD, Section S.2.3), the concentration of amylosedispersed SWCNTs could not be measured. The calibration curves were linear (R2 0.99). The extinction coefficients for suspensions of SWCNTs dispersed using GA-, PVP-, Triton X-100-, NOM (pH 3.5)-, and NOM (pH 7) were determined to be 1.81 102, 1.97 102, 2.51 102, 2.54 102, and 2.70 102 (L mg1 cm1), respectively. The concentration of exfoliated SWCNTs in suspension as a function of settling time is given in Fig. 2a. The dispersion and exfoliation of SWCNTs bundles in the presence of GA and PVP was attributed to the wrapping of a long polymeric chain around the nanotube core (Bandyopadhyaya et al., 2002; Didenko et al., 2005), whereas supramolecular encapsulation was hypothesized as the mechanism of CNT modification by helical amylose (Kim et al., 2004).
Table 1 – Concentrations of the SWCNTs upon addition to cells in the general viability bioassay with E. coli, NRDU bioassay, and GJIC bioassay. Solubilizer
Concentration of SWCNTs in general E. coli viability bioassay, mg/L
Concentration of SWCNTs in NRDU and GJIC bioassays, mg/L
NOM pH 3.5 NOM pH 7.0 PVP GA Triton X-100
68.27 40.12 77.38 142.91 107.19
11.22 11.93 23.42 34.97 26.43
513
water research 44 (2010) 505–520
b
200
Effective diameter (nm)
Concentration (mg/L)
a
160 120 80 40 0
1000 800 600 400 200 0
0
7
14
21
28
0
7
Time (d) 0
d
-15
-potential (mV)
-potential (mV)
c
14
21
28
21
28
Time (d)
-30
-45
0
-15
-30
-45
-60
0
7
14
21
28
Time (d)
0
7
14
Time (d)
Fig. 2 – The time-dependent characterization of (a) estimated concentration of the exfoliated SWCNTs in the dispersion, (b) effective diameter and (c) z-potential of dispersed SWCNTs, and (d) z-potential of dispersants alone (-B- for GA/SWCNTs, -,- for PVP/SWCNTs, -6- for NOM3.5/SWCNTs, -:- for NOM7/SWCNTs, --- for Triton X-100/SWCNTs, -C- for Amylose/ SWCNTs). Notes: (1) The z-potential measurement was not conducted for amylose solution, because amylose is not watersoluble under room temperature. (2) Reported are results of only those measurements that were statistically significant, i.e. when sufficiently high photon count rates were recorded in dynamic light scattering measurements. (3) The absorption by suspensions of amylose-stabilized SWCNTs was not measured because the nanotube content in these suspensions could not be precisely determined; this was due to the incomplete separation at the centrifugation step of the suspension preparation process.
Triton X-100 stabilizes nanotubes by the formation of hemimicelles that cover nanotube surface with benzene rings providing p–p stacking between the surfactant molecule and nanotube core (Islam et al., 2003). The exfoliation (debundling) of SWCNTs in the presence of SRNOM was suggested to occur through the interaction of the aromatic moieties of natural organic matter and nanotube surface (Hyung et al., 2007; Liu et al., 2007; Hyung and Kim, 2008). Despite the differences in the physiosorption mechanisms responsible for the stabilization of CNTs in aqueous suspensions, all dispersants were effective, albeit to somewhat different extents. For all SWCNT suspensions, the calculated values of the concentration of exfoliated SWCNTs correlated well to the visual observations described above. As seen from Fig. 2a, the dispersion efficiency in terms of the concentration of dispersed SWCNTs was a function of the type of the dispersant, with GA producing suspensions having highest concentration of SWCNTs. The data were consistent among the three prepared batches of SWCNTs (see SD, Table S1). Didenko et al. (Didenko et al., 2005) suggested that after covering one single carbon nanotube with a long polymeric molecule, the remaining strands would react with other uncovered or partially covered SWCNTs thus bundling several nanotubes together. We speculate that this mechanism where
one polymer molecule links two or more nanotubes may be responsible for the higher dispersing efficiency of GA (by far the largest molecule among the target dispersants) towards SWCNTs compared to other dispersants. This hypothesis is supported by the measurements of the effective hydrodynamic diameter (Fig. 2b), and the estimation of the length of suspended particles with the size and the length of GA/ SWCNTs being higher than those of PVP, Triton X-100 and SRNOM (both pHs). As the measured diameter of GA-stabilized SWCNTs aggregates (which could also be expected to be rather porous) was still relatively small, gravity settling did not lead to a significant removal of GA-stabilized SWCNTs from the dispersion. When comparing the solubilizing ability of SRNOM at different pH, one could see that SRNOM is a more efficient dispersant at pH 3.5 than at pH 7. This is consistent with the findings by Huyng et al. (Hyung and Kim, 2008) who reported that adsorption of SRNOM to MWCNTs increased when pH decreased due to more compact and coiled conformation of NOM at acidic pHs. When pH increases, carboxylic and phenolic groups of SRNOM deprotonate resulting in higher electrostatic repulsion between a CNT and a SRNOM molecule and, as a consequence, in a lower amount of organic matter adsorbed on the CNT surface. The better dispersion of
514
water research 44 (2010) 505–520
a Fraction of control (FOC)
SWCNTs at pH 3.5 could also be attributed, in part, to the steric hindrance imposed by SRNOM when a higher surface density of NOM on the surface of CNT bundles results in stronger repulsion between CNTs. This observation is supported by the long-term z-potential measurements (Fig. 2c), where the surface charge of stabilized SRNOM/SWCNTs at pH 3.5 became less negative up to day 7, and then stabilized with increasing settling time. At the same time the surface charge of SWCNTs dispersed in SRNOM solution at pH 7 gradually increased after day 5. Indeed, had the electrostatic repulsion been the sole mechanism of SWCNTs stabilization in SRNOM solutions, the surface charge on the SWCNTs would have more rapidly become less negative at pH 3.5 than at pH 7. In summary, in terms of the effectiveness of SWCNT stabilization, the dispersants were ranked as follows: GA > Triton X-100 > PVP > NOM (pH 3.5) > NOM (pH 7).
0.8 0.6 0.4 0.2
NOM 3.5 NOM 7
GA
Amylose
PVP
Triton
1.2
Charge of dispersed SWCNTs
For all suspensions, the z-potential of dispersed SWCNTs was negative and nearly constant over the entire duration of the experiment (Fig. 2c and SD, Table S1). Each z-potential
Fraction of control (FOC)
Hydrodynamic size of dispersed SWCNTs 1 0.8 0.6 0.4 0.2 0 NOM 3.5
c Fraction of control (FOC)
The hydrodynamic size of CNTs in a suspension is an important characteristic that affects the stability and, possibly, toxicity of dispersed CNTs. It should be noted that the effective hydrodynamic diameter of suspended SWCNTs measured using DLS can only be used as a very rough estimate. While the DLS data are interpreted with the assumption that the primary scatterers are spherical with an aspect ratio of 1, dispersed SWCNTs are long and tubular, with a very high aspect ratio (see SD, Fig. S3). To obtain a better approximation of the size distribution of dispersed SWCNT, the multimodal size distribution model was used. TEM imaging was employed to obtain auxiliary information on the size and morphology of stabilized SWCNT. Even though the measurement of effective hydrodynamic diameter cannot be relied on to compare the sizes of suspended SWCNTs in different dispersion media, these measurements can be used to compare how the average particle size for a given dispersant changes with settling time. The values of effective diameter as a function of time for different SWCNT suspensions are given in Fig. 2b. Generally, the effective size of SWCNTs dispersed using Triton X-100, GA and NOM did not change significantly with increasing settling time (Fig. 2b). In the case of amylosedispersed SWCNTs, a slight decrease in the effective size was observed due to settling of larger and unstable SWCNT aggregates, which left behind more uniformly sized smaller amylose-stabilized CNT clusters. The concentration of SWCNTs in the suspension was negatively correlated with the effective size of SWCNT. The notable apparent exception was the suspension of SWCNTs dispersed using GA. However, it should be noted that the aqueous suspension of GA contained GA colloids of the size that was 1) comparable to the size of dispersed SWCNTs and 2) decreased during the 4 weeks of settling (see SD, Fig S5). Thus, the effective size measurements for SWCNTs dispersed using GA should be interpreted with caution.
4.3.
1
0
b 4.2.
1.2
NOM 7
GA
Amylose
PVP
Triton
1.2 1 0.8 0.6 0.4 0.2 0 NOM 3.5 NOM 7
GA
Amylose PVP
Triton
Fig. 3 – Viability of E. coli in (a) 0.1 M NaCl, (b) LB medium, and (c) MD medium. Exposure time: (a) 3 h, (b) 48 h, (c) 48 h. Open, hatched and cross-hatched bars correspond to control I, control II and dispersed SWCNTs. In the case of amylose ultrapure water was used as control.
measurement for SWCNT suspensions was accompanied by a measurement of the z-potential of the dispersants in a SWCNT-free aqueous solution (Fig. 2d). After 1 d, only the GA solution scattered light sufficiently to enable a statistically meaningful z-potential measurement. After 4 d, solutions of NOM (pH 7), PVP, and Triton X-100 also scattered light
515
water research 44 (2010) 505–520
1.2
b Fraction of control (FOC)
Fraction of control (FOC)
a
1 0.8 0.6 0.4 0.2
1.2 1 0.8 0.6 0.4 0.2 0
0 NOM3.5 NOM7
GA
Amylose
PVP
Triton
NOM3.5
NOM7
GA Amylose
PVP Triton
Fig. 4 – Neutral red dye uptake by WB-F344 cells as a function of dispersant type (a – 30 min, b – 24 h of exposure). Open, hatched and cross-hatched bars correspond to control I, control II and dispersed SWCNTs. In the case of amylose control I and control II were ultrapure water.
sufficiently. After 7 d, all the dispersants except for NOM (pH 3.5) achieved acceptable count rates during z-potential measurements. The gradual change in the scattering ability of the dispersant solutions may correspond to the aggregation of the dispersant molecules. The dispersants were ranked in order from the most to the least negative z-potential of dispersed SWCNTs: NOM (pH 7) > NOM (pH 3.5) > GA z Triton X-100 > PVP z amylose. The highly negative charge of NOMstabilized SWCNTs was likely due to the charge of NOM. The low stability of amylose-dispersed SWCNTs could be attributed to the combination of low surface charge and steric repulsion between stabilized SWCNTs in the dispersion.
4.4. Assessment of cell toxicity in prokaryotic and eukaryotic systems 4.4.1.
General viability bioassay with E. coli (prokaryotic)
Immediately after E. coli cells were exposed to SWCNTs suspended in growth medium as well as after 3 h of exposure, no SWCNT aggregation was visually observed (see SD, Table S2) in experiments with all three types of the media – 0.1 M NaCl, MD (medium with lower salt organics content) and LB (medium with higher salt and organics content). After 24 h of incubation, limited precipitation was observed for all SWCNT
suspensions in LB and MD media. After 48 h of incubation more precipitation occurred in each type of media. No inhibition of E. coli colony forming ability was observed after 3 h of incubation with amylose-, NOM-, GA- and PVPstabilized SWCNTs in 0.1 M NaCl (Fig. 3a). The FOC values did not decline in any of these samples and no significant difference in the influence on CFU counts was found between dispersed SWCNTs and solutions of corresponding dispersants. When the bacteria suspension was brought in contact with Triton X-100-stabilized SWCNTs, approx. 25% loss of the cell viability was measured as compared to vehicle control I samples. However, there was no statistical difference between FOC of Triton-stabilized SWCNTs and the control (solution of Triton X-100 only). It remains unclear if the observed cytotoxicity of the these suspensions was due to i) SWCNTs stabilized by Triton X-100 or ii) the residual ‘‘free’’ (i.e. not associated with suspended SWCNTs) Triton X-100 potentially present in the solution or iii) the combined effect of both Triton X-100/SWCNT and dissolved Triton X-100. The complete separation of the Triton X-100/SWCNT and dissolved Triton X-100 could not be accomplished using centrifugation. Even for very long centrifugation times, the supernatant had grayish color indicating that some fraction of SWCNTs was not removed.
Fig. 5 – Representative phase contrast images of WB-F344 cells incubated with 500 mL of H2O (control I), 500 mL of Triton X-100 (control II) and 500 mL of Triton X-100-stabilized SWCNTs. Black dots observed in Control II and Triton X-100-solubilized SWCNTs samples correspond to dead WB-F344 cells. All images were taken at 2003 magnification. Scale bar is 50 mm.
516
water research 44 (2010) 505–520
Fig. 6 – Representative phase contrast (upper row) and UV epifluorescent (bottom row) images of WB-F344 cells incubated with 500 mL of H2O (control I), 500 mL of NOM pH 3.5 (control II) and 500 mL of NOM-stabilized SWCNTs (pH 4). All images were taken at 2003 magnification. Bright dots correspond to cells that absorb the dye; the absorption is indicative of cellular health. All images were taken at 2003 magnification. Scale bar is 50 mm.
1.2 1 0.8 0.6 0.4 0.2
plates. As the exposure time increased to 48 h, E. coli resumed its growth; with FOC of both Triton X-100 suspensions approaching values for vehicle control I sample (Fig. 3b). When dispersed SWCNTs were tested in MD medium, no losses in cell viability for amylose-, GA-, PVP- and NOM-stabilized SWCNTs were measured (Fig. 3c; also see SD, Fig. S7). In the case of Triton X-100 suspensions, the reduction in E. coli survival was observed after 3 h of exposure; comparable losses of the E. coli viability were also observed for both Triton X-100-stabilized SWCNTs and the Triton X-100 only solution. However, after 24 h of exposure, fewer E. coli colonies were
b
1.2
Fraction of control (FOC)
a Fraction of control (FOC)
The ability of E. coli to grow and to form colonies in the presence of amylose-, GA-, PVP-, and NOM-stabilized SWCNTs in LB medium mimicked both control samples regardless of the contact time (see SD, Fig. 3b and Fig. S6). On the contrary, 21 and 18 % mortality rates after 3 h of incubation were observed for E. coli in Triton X-100/SWCNTs suspension and Triton X-100 solution, respectively. After 24 h of contact, the number of colonies grown on the Petri plates decreased by 30 % when bacteria were in contact with Triton X-100-stabilized SWCNTs and by 27 % when bacteria were in Triton X-100 only solution (see SD, Fig. S6) as compared to vehicle control I
1 0.8 0.6 0.4 0.2 0
0 NOM 3.5
NOM 7
GA
Amylose
PVP
NOM 3.5
NOM 7
GA
Amylose
PVP
Fig. 7 – GJIC in WB-F344 cells as a function of dispersant type (a – 30 min, b – 24 h of exposure). Open, hatched and cross-hatched bars correspond to control I, control II and dispersed SWCNTs. In the case of amylose control I and control II were ultrapure water.
water research 44 (2010) 505–520
observed on Petri plates from suspensions of Triton X-100stabilized SWCNTs and Triton X-100 solution only (vehicle control II) (33 and 44 %, respectively) (see SD, Fig. S7). As the incubation time increased to 48 h, bacterial CFU counts in the presence of both Triton X-100 samples increased and approached that of vehicle control I samples (Fig. 3c). As was previously mentioned, the exposure of Triton X-100-stabilized SWCNTs to MD medium for 24 h resulted in formation of large flake-like aggregates, which settled out of the suspension. It is possible that this low salt medium induced specific interactions between bacterial cells and Triton X-100-coated SWCNTs, which were suppressed in the medium having a higher ionic strength. This led to a larger decrease in E. coli viability in these samples in comparison with the losses in Triton X-100 solution only. Damaged or dead cell attached to the SWCNTs aggregates and formed debris at the bottom of the testing tubes. The simultaneous effect of bacterial and nanoparticles settling out of the solution has been previously described (Tang et al., 2007) where the response of Shewanella oneidensis to C60–NH2 nanoparticles was studied. The fact that we observed inhibition of E. coli viability after 24 h of exposure in MD medium and did not observed this effect in LB medium highlights the importance of growth medium in cytotoxicity testing. Similarly, E. coli exposure to SWCNTs, stabilized with amylose, GA, PVP and NOM (both pH) did not affect CFU counts in any of the three media pointing to the importance of the dispersant in the assessments of biocompatibility of CNTs.
4.4.2.
NDU cytotoxicity bioassay on eukaryotic cells
The assessment of cytotoxicity was determined by the viable uptake of neutral red into F344-WB rat liver epithelial cells. Nonviable cells lack the ability to absorb this dye. Results (Fig. 4) indicated that SWCNTs dispersed using NOM (pH 3.5), NOM (pH 7), GA, amylose and PVP were not cytotoxic. However, Triton X-100 detergent induced a significant cytotoxic effect independent of SWCNTs. This latter observation can also be seen under phase contrast microscopy of cells treated with Triton X-100 for 30 min (Fig. 5). The morphology of the cells drastically changed from that of the normal control cells, in which the cells became clearer due to loss of most cytosolic contents, except for the cell nuclei. After a 24 h treatment with Triton X-100, the cells completely detached or solubilized from the culture plates (data not shown).
4.5. Assessment of epigenetic toxicity on eukaryotic cells: GJIC bioassay A representative fluorescent micrograph of the GJIC assay after treatment with (NOM þ SWCNTs) is shown in Fig. 6. The migration of the fluorescent dye through several cell layers is an indicator that the gap junction channels are open. After WB-F344 cells were incubated with dispersed SWCNTs suspensions for (1) 30 min and (2) 24 h, there was no significant affect on GJIC for all suspensions tested (Fig. 7). Due to Triton X-100-stabilized SWCNTs being cytotoxic, GJIC was not measured in cells exposed to this mixture. Also, when cells were examined under phase
517
contrast, no changes in size and shape of the individual cells were detected (Fig. 6). Hence, it can be concluded that under these experimental conditions WB-F344 cells exposed to dispersed SWCNTs retain normal intercellular communication irrespective of the applied treatment.
5.
Conclusions
The stability of aqueous suspensions of SWCNTs non-covalently functionalized by a range of natural (gum arabic, amylose, Suwannee River natural organic matter) and synthetic (polyvinyl pyrrolidone, Triton X-100) dispersants that bind to SWCNT surface via different physiosorption mechanisms was evaluated. Despite the differences, all SWCNT suspensions remained relatively stable over a period of 4 weeks as indicated by the evolution of the concentration of suspended exfoliated SWCNT and by the fact that both the average effective hydrodynamic diameter and z-potential of suspended SWCNTs were relatively constant. The epigenetic toxicity of dispersed SWCNTs was evaluated for the first time using a gap junctional intercellular communication assay with WB-F344 rat liver epithelial cells resulting in no inhibition by SWCNTs non-covalently functionalized using GA, PVP, amylose, and SRNOM. There were no cytotoxic effects of these SWCNTs suspensions on either the prokaryotic, bacterial (E. coli), or eukaryotic (WB-F344 rat liver epithelial) cell types. Only when the dispersant itself was toxic, were losses of cell viability observed. Bacterial CFU counts or neutral red uptake was affected only by SWCNTs dispersed using Triton X-100, which showed cytotoxicity in the SWCNT-free solution (vehicle control II). The results suggest a strong dependence of the toxicity of SWCNT suspensions on the toxicity of the solubilizing agent and point to the potential of non-covalent functionalization with non-toxic dispersants as a method for the preparation of aqueous suspensions of biocompatible CNTs.
Acknowledgements This work was supported by the National Science Foundation research grants CBET-0608320, CBET-056828 and OISE0530174, by the National Institute of Environmental Health Sciences grants R01 ES013268-01A2 and Grant-in-Aid for Science Research. We thank Michael Housinger for his assistance with the preparation of SWCNT suspensions. ARR acknowledges the support by Michigan AWWA Fellowship for Water Quality and Treatment Study. The contents of this work are solely the responsibility of the authors and do not necessarily represent the official views of the National Institute of Environmental Health Sciences.
Appendix. Supplementary data Supplementary data associated with this article can be found in the online version, at doi: doi:10.1016/j.watres.2009.09.042.
518
water research 44 (2010) 505–520
references
Atlas, R.M., 1993. Handbook of Microbiological Media. CRC Press, Boka Raton. Bahr, J.L., Mickelson, E.D., Bronikovski, M.J., Smalley, R.E., Tour, J.M., 2001. Dissolution of small diameter single-wall carbon nanotubes in organic solvents. Chemical Communications, 193–194. Bandyopadhyaya, R., Nativ-Roth, E., Regev, O., YerushalmiRozen, R., 2002. Stabilization of individual carbon nanotubes in aqueous solutions. Nano Letters 2 (1), 25–28. Bardi, G., Tognini, P., Ciofani, G., Raffa, V., Costa, M., Pizzorusso, T., 2009. Pluronic-coated carbon nanotubes do not induce degeneration of cortical neurons in vivo and in vitro. Nanomedicine: Nanotechnology, Biology, and Medicine 5, 96–104. Bonnet, P., Albertini, D., Bizot, H., Bernard, A., Chauvet, O., 2007. Amylose/SWNT composites: from solution to film – synthesis, characterization and properties. Composite Science and Technology 67, 817–821. Borenfreund, E., Puerner, J.A., 1985. Toxicology determined in vitro by morphological alterations and neutral red absorption. Toxicology Letters 24, 119–124. Burnette, L.W., 1960. A review of the physiological properties of PVP. Proceedings of the Scientific Section of the Toilet Goods Association 38, 1–4. Chappell, M.A., George, A.J., Dontsova, K.M., Porter, B.E., Price, C.L., Zhou, P., Morikawa, E., Kennedy, A.J., Steevens, J.A., 2009. Surfactive stabilization of multi-walled carbon nanotube dispersions with dissolved humic substances. Environmental Pollution 157, 1081–1087. Chen, X., Tam, U.C., Czlapinski, J.L., Lee, G.S., Rabuka, D., Zettl, A., Bertozzi, C.R., 2006. Interfacing carbon nanotubes with living cells. Journal of American Chemical Society 128, 6292–6293. Cherukuri, P., Gannon, C.J., Leeuw, T.K., Schmidt, H.K., Smalley, R.E., Curley, S.A., Weisman, R.B., 2006. Mammalian pharmacokinetics of carbon nanotubes using intrinsic nearinfrared fluorescence. Proceedings of the National Academy of Sciences 103, 18882–18886. Chin, S.F., Baughman, R.H., Dalton, A.B., Diekmann, G.R., Draper, R.K., Mikoryak, C., Musselman, I.H., Poenitzsch, V.Z., Xie, H., Pantano, P., 2007. Amphiphilic helical peptide enhances the uptake of single-walled carbon nanotubes by living cells. Experimental Biological Medicine 232 (9), 1236–1244. Chourasia, M.K., Jain, S.K., 2004. Polysaccharides for colon targeted drug delivery. Drug Delivery 11 (2), 129–148. Dayeh, V.R., Chow, S.L., Schirmer, K., Lynn, D.H., Bols, N.C., 2004. Evaluating the toxicity of Triton X-100 to protozoan, fish, and mammalian cells using fluorescent dyes as indicators of cell viability. Ecotoxicology and Environmental Safety 57 (3), 375–382. Di Sotto, A., Chiaretti, M., Carru, G.A., Bellucci, S., Mazzantia, G., 2009. Multi-walled carbon nanotubes: lack of mutagenic activity in the bacterial reverse mutation assay. Toxicology Letters 184, 192–197. Didenko, V.V., Moore, V.C., Baskin, D.S., Smalley, R.E., 2005. Visualization of individual single-walled carbon nanotubes by fluorescent polymer wrapping. Nano Letters 5 (8), 1563–1567. Ding, L.H., Stilwell, J., Zhang, T.T., Elboudware, J.O., Jiang, H.J., Selegue, J.P., Cooke, P.A., Gray, J.W., Chen, F.F., 2005. Molecular characterization of the cytotoxic mechanism of multiwall carbon nanotubes and nanoonions on human skin fibroblast. Nano Letters 5, 2448–2464. Dodziuk, H., Ejchart, A., Anczewski, W., Ueda, H., Krinichnaya, E., Dolgonosa, G., Kutner, W., 2003. Water solubilization,
determination of the number of different types of single-wall carbon nanotubes and their partial separation with respect to diameters by complexation with h-cyclodextrin. Chemical Communications, 986–987. El-Fouly, M.H., Trosko, J.E., Chang, C.C., 1987. Scrape-loading and dye transfer: a rapid and simple technique to study gap junctional intercellular communications. Experimental Cell Research 168, 422–430. Elias, A.L., Carrero-Sanchez, J.C., Terrones, H., Endo, M., Laclette, H.P., Mauricio, M., 2007. Viability studies of pure carbon- and nitrogen-doped nanotubes with Entamoeba Histolytica: from amoebicidal to biocompatible structures. Small 3 (10), 1723–1729. Enyashin, A.N., Gemming, S., Seifert, G., 2007. DNA-wrapped carbon nanotubes. Nanotechnology 18, 245702. Fiorito, S., Serafino, A., Andreola, F., Bernier, P., 2006. Effects of fullerenes and single-wall carbone nanotubes on murine and human macrophages. Carbon 44, 1100–1105. Flahaut, E., Durrieu, M.C., Remy-Zolghadri, M., Bareille, R., Baquey, C., 2006. Investigation of the cytotoxicity of CCVD carbon nanotubes towards human umbilical vein endothelial cells. Carbon 44 (6), 1093–1099. Fortner, J.D., Lyon, D.Y., Sayes, C.M., Boyd, A.M., Falkner, J.C., Hotze, E.M., Alemany, L.B., Tao, Y.J., Guo, W., Ausman, K.D., Colvin, V.L., Hughes, J.B., 2005. C60 in water: nanocrystal formation and microbial response. Environmental Science and Technology 39 (11), 4307–4316. Georgakilas, V., Tagmatarchis, N., Pantarotto, D., Bianco, A., Briand, J.-P., Prato, M., 2002. Amino acid functionalisation of water soluble carbon nanotubes. Chemical Communications, 3050–3051. Grossiord, N., Schhoo, P., Meuldijk, J., Koning, C.E., 2007. Determination of the surface coverage of exfoliated carbon nanotubes by surfactant molecules in aqueous solution. Langmuir 23, 3646–3653. Gutierrez, M.C., Garcia-Carvajal, Z.Y., Hortiguela, M.J., Yuste, L., Rojo, F., Ferrer, M.L., Monte, F., 2007. Biocompatible MWCNT scaffolds for immobilization and proliferation of E. coli. Journal of Materials Chemistry 17, 2992–2995. Herner, H.A., Trosko, J.E., Masten, S.J., 2001. The epigenetic toxicity of pyrene and related ozonation byproducts containing an aldehyde functional group. Environmental Science and Technology 35 (17), 3576–3583. Herzog, E., Byrne, H.J., Davoren, M., Casey, A., Duschl, A., Oostingh, G.J., 2009. Dispersion medium modulates oxidative stress response of human lung epithelial cells upon exposure to carbon nanomaterial samples. Toxicology and Applied Pharmacology 236, 276–281. Huang, W.J., Taylor, S., Fu, K.F., Lin, Y., Zhang, D.H., Hanks, T.W., Rao, A.M., Sun, Y.P., 2002. Attaching proteins to carbon nanotubes via diimide-activated amidation. Nano Letters 2 (4), 311–314. Humic acids and their sodium salts. EMEA-The European agency for the evaluation of medicinal products. Available via. http:// www.emea.europa.eu/pdfs/vet/mrls/055499en.pdf, 1999. Hyung, H., Fortner, J.D., Hughes, J.B., Kim, J.H., 2007. Natural organic matter stabilizes carbon nanotubes in the aqueous phase. Environmental Science and Technology 41 (1), 179–184. Hyung, H., Kim, J.-H., 2008. Natural organic matter (NOM) adsorption to multi-walled carbon nanotubes: effect of NOM characteristics and water quality parameters. Environmental Science and Technology 42 (12), 4416–4421. Iijima, S., 1991. Helical microtubules of graphitic carbon. Nature 354 (6348), 56–58. Inoue, K.-i., Koike, E., Yanagisawa, R., Hirano, S., Nishikawa, M., Takano, H., 2009. Effects of multi-walled carbon nanotubes on a murine allergic airway inflammation model. Toxicology and Applied Pharmacology 237 (3), 306–316.
water research 44 (2010) 505–520
Islam, M.F., Rojas, E., Bergey, D.M., Johnson, A.T., Yodh, A.G., 2003. High weight fraction surfactant solubilization of single-wall carbon nanotubes in water. Nano Letters 3 (2), 269–273. Jia, G., Wang, H., Yan, L., Wang, X., Pei, R., Yan, T., Zhao, Y., Guo, X., 2005. Cytotoxicity of carbon nanomaterials: singlewall nanotube, multi-wall nanotube, and fullerene. Environmental Science and Technology 39, 1378–1383. Jiang, J., Gao, L., 2003. Production of aqueous colloidal dispersions of carbon nanotubes. Journal of Colloid and Interface Science 260, 89–94. Kang, S., Herzberg, M., Rodrigues, D.F., Elimelech, M., 2008a. Antibacterial effects of carbon nanotubes: size does matter! Langmuir 24 (13), 6409–6413. Kang, S., Mauter, M.S., Elimelech, M., 2008b. Physicochemical determinants of multiwalled carbon nanotube bacterial cytotoxicity. Environmental Science and Technology 42 (19), 7528–7534. Kang, S., Mauter, M.S., Elimelech, M., 2009. Microbial cytotoxicity of carbon-based nanomaterials: implications for river water and wastewater effluent. Environmental Science and Technology 43 (7), 2648–2653. Kang, S., Pinault, M., Pfefferle, L.D., Elimelech, M., 2007. Singlewalled carbon nanotubes exhibit strong antimicrobial activity. Langmuir 23 (17), 8670–8673. Karajanagi, S.S., Yang, H.C., Asuri, P., Sellitto, E., Dordick, J.S., Kane, R.S., 2006. Protein-assisted solubilization of singlewalled carbon nanotubes. Langmuir 22 (4), 1392–1395. Kim, O.-K., Je, J., Baldwin, J.W., Kooi, S., Pehrsson, P.E., Buckley, L.J., 2004. Solubilization of single-wall carbon nanotubes by supramolecular encapsulation of helical amylose. Journal of American Chemical Society 125, 4426–4427. Lauret, J.-S., Voisin, C., Cassabois, G., Roussignol, P., Delalande, C., Capes, L.E., Valentin, E., Jost., O., 2004. Photocreated carrier dynamics in isolated carbon nanotubes. Semiconductor Science and Technology 19, S486–S488. Lee, J.U., Huh, J., Kim, K.H., Park, C., Jo, W.H., 2007. Aqueous suspension of carbon nanotubes via non-covalent functionalization with oligothiophene-terminated poly(ethylene glycol). Carbon 45, 1051–1057. Lin, Y., Taylor, S., Li, H.P., Fernando, K.A.S., Qu, L.W., Wang, W., Gu, L.R., Zhou, B., Sun, Y.P., 2004. Advances toward bioapplications of carbon nanotubes. Journal of Materials Chemistry 14 (4), 527–541. Lindberg, H.K., Falck, G.C.-M., Suhonen, S., Vippola, M., Vanhalab, E., Catala´na, J., Savolainena, K., Norppa, H., 2009. Genotoxicity of nanomaterials: DNA damage and micronuclei induced by carbon nanotubes and graphite nanofibres in human bronchial epithelial cells in vitro. Toxicology Letters 186, 166–173. Liu, J., Rinzler, A.G., Dai, H.J., Hafner, J.H., Bradley, R.K., Boul, P.J., Lu, A., Iverson, T., Shelimov, K., Huffman, C.B., RodriguezMacias, F., Shon, Y.S., Lee, T.R., Colbert, D.T., Smalley, R.E., 1998. Fullerene pipes. Science 280 (5367), 1253–1256. Liu, Y.Q., Gao, L., Zheng, S., 2007. Debundling of single-walled carbon nanotubes by using natural polyelectrolytes. Nanotechnology 18 (36), 365702. Lyon, D.Y., Fortner, J.D., Sayes, C.M., Colvin, V.L., Hughes, J.B., 2005. Bacterial cell association and antimicrobial activity of a C60 water suspension. Environmental Toxicology and Chemistry 24, 2757–2762. Manna, S.K., Sarkar, S., Barr, J., Wise, K., Barrera, E.V., Jejelowo, O., Rice-Ficht, A., Ramesh, G.T., 2005. Single-walled carbon nanotube induces oxidative stress and activates nuclear transcription factor-kB in human keratinocytes. Nano Letters 5, 1676–1684. Marsh, D.H., Rance, G.A., Zaka, M.H., Whitby, R.J., Khlobystov, A.N., 2007. Comparison of the stability of multiwalled carbon
519
nanotube dispersions in water. Physical Chemistry Chemical Physics 9 (40), 5490–5496. McDonald, T.J., Engtrakul, C., Jones, M., Rumbles, G., Heben, M.J., 2006. Kinetics of PL quenching during single-walled carbon nanotube rebundling and diameter-dependent surfactant interactions. Journal of Physical Chemistry B 110, 25339–25346. Meng, J., Yanga, M., Song, L., Kong, H., Wang, C.Y., Wang, R., Wang, C., Xie, S.S., Xua, H.Y., 2009. Concentration control of carbon nanotubes in aqueous solution and its influence on the growth behavior of fibroblasts. Colloids and Surfaces B: Biointerfaces 71, 148–153. Moore, V.C., Strano, M.S., Haroz, E.H., Hauge, R.H., Smalley, R.E., Schmidt, J., Talmon, Y., 2003. Individually suspended singlewalled carbon nanotubes in various surfactants. Nano Letters 3 (10), 1379–1382. Muller, J., Huaux, F., Moreau, N., Misson, P., Heilier, J.-F., Delos, M., Arras, M., Fonseca, A., Nagy, J.B., Lison, D., 2005. Respiratory toxicity of multi-wall carbon nanotubes. Toxicology and Applied Pharmacology 207 (31), 221–231. Narayan, R.J., Berry, C.J., Brigmon, R.L., 2005. Structural and biological properties of carbon nanotube composite films. Materials Science and Engineering B 123, 123–129. O’Connell, M.J., Bachilo, S.M., Huffman, C.B., Moore, V.C., Strano, M.S., Haroz, E.H., Rialon, K.L., Boul, P.J., Noon, W.H., Kittrell, C., Ma, J., Hauge, R.H., Weisman, R.B., Smalley, R.E., 2002. Band gap fluorescence from individual single-walled carbon nanotubes. Science 297, 593–596. O’Connell, M.J., Boul, P.J., Ericson, L.M., Huffman, C., Wang, Y., Haroz, E., 2001. Reversible water-solubilization of singlewalled carbon nanotubes by polymer wrapping. Chemical Physics Letters 342, 265–271. Park, H.J., Heo, H.Y., Lee, S.C., Park, M., Lee, S.-S., Kim, J., Chang, J.Y., 2006. Dispersion of single-walled carbon nanotubes in water with polyphosphazene polyelectrolyte. Journal of Inorganic and Organometallic Polymers and Materials 16 (4) 359–364. Pulskamp, K., Worle-Knirsch, J.M., Hennrich, F., Kern, K., Krug, H.F., 2007. Human lung epithelial cells show biphasic oxidative burst after single-walled carbon nanotube contact. Carbon 45, 2241– 2249. Saleh, N.D., Pfefferle, L.D., Elimelech, M., 2009. Aggregation kinetics of multiwalled carbon nanotubes in aquatic systems: measurements and environmental implications. Environmental Science and Technology 42 (21), 7963–7969. Salzmann, C.G., Chu, B.T.T., Tobias, G., Llewellyn, S.A., Green, M.L.H., 2007. Quantitative assessment of carbon nanotube dispersions by Raman spectroscopy. Carbon 45 (5), 907–912. Satoh, A.Y., Trosco, J., Masten, S.J., 2003. Epigenetic toxicity of hydroxylated biphenyls and hydroxylated polychlorinated biphenyls on normal rat liver epithelial cells. Environmental Science and Technology 37, 2727–2733. Sayes, C.M., Liang, F., Hudsona, J.L., Mendeza, J., Guob, W., Beach, J.M., Moore, V.C., Doyle, C.D., West, J.L., Billups, W.E., Ausmanb, K.D., Colvin, V.L., 2006. Functionalization density dependence of single-walled carbon nanotubes cytotoxicity in vitro. Toxicology Letters 161, 135–142. Schmitt, D., Tran, N., Riefler, S., Jacoby, J., Merkel, D., Marone, P., Naouli, N., 2008. Toxicologic evaluation of modified gum acacia: mutagenicity, acute and subchronic toxicity. Food and Chemical Toxicology 46, 1048–1054. Simon-Deckers, A., Gouget, B., Mayne-L’Hermiteb, M., HerlinBoimeb, N., Reynaudb, C., Carrie`rea, M., 2008. In vitro investigation of oxide nanoparticle and carbon nanotube toxicity and intracellular accumulation in A549 human pneumocytes. Toxicology 253, 137–146. Sinani, V.A., Gheith, M.K., Yaroslavov, A.A., Rakhnyanskaya, A.A., Sun, K., Mamedov, A.A., Wicksted, J.P., Kotov, N.A., 2005.
520
water research 44 (2010) 505–520
Aqueous dispersions of single-wall and multiwall carbon nanotubes with designed amphiphilic polycations. Journal of American Chemical Society 127 (10), 3463–3472. Star, A., Steuerman, D.W., Heath, J.R., Stoddart, J.F., 2002. Starched carbon nanotubes. Angewandte Chemie International Edition 41 (14), 2508–2512. Tang, Y.J., Ashcroft, J.M., Chen, D., Min, G., Kim, C.-H., Murkhejee, B., Larabell, C., Keasling, J.D., Chen, F.F., 2007. Charge-associated effects of fullerene derivatives on microbial structural integrity and central metabolism. Nano Letters 7 (3), 754–760. Tasis, D., Tagmatarchis, N., Bianco, A. and Prato, M., 2006. Chemistry of carbon nanotubes. 106 (3), 1105–1136. Tong, Z., Bischoff, M., Nies, L., Applegate, B., Turco, R., 2007. Impact of fullerene (C60) on a soil microbial community. Environmental Science and Technology 41, 2985–2991. Trosko, J.E., Chang, B.V., Madhucar, J.E., Klauning, J.E., 1991. Chemical, oncogene, and growth factor inhibition of gap junctional intercellular communication: an alternative hypothesis to carcinogenesis. Pathobiology 58, 265–278. Trosko, J.E., Chang, C.-C., Upham, B., Wilson, M., 1998. Epigenetic toxicology as toxicant-induced changes in intracellular signalling leading to altered gap junctional intercellular communication. Toxicology Letters 102–103, 71–78. Tseng, C.-H., Wang, C.-C., Chen, C.-Y., 2006. Modification of multi-walled carbon nanotubes by plasma treatment and further use as templates for growth of CdS nanocrystals. Nanotechnology 17, 5602–5612. Upham, B.L., Masten, S.J., Lochwood, B.R., Trosko, J.E., 1994. Nongenotoxic effects of polycyclic aromatic hydrocarbons and their ozonation by-products on the intercellular communication of rat liver epithelial cells. Fundamental and Applied Toxicology 23, 470–475. Vaisman, L., Marom, G., Wagner, H.D., 2006. Dispersions of surface-modified carbon nanotubes in water-soluble and water-insoluble polymers. Advanced Functional Materials 16 (3), 357–363. Vittorio, O., Raffa, V. and Cuschieri, A., Influence of purity and surface oxidation on cytotoxicity of multi-wall carbon nanotubes with human neuroblastoma cells. Nanomedicine: Nanotechnology, Biology, and Medicine, in press, doi:10.1016/j.nano.2009.02.006. Wang, Y., Iqbal, Z., Malhotra, S.V., 2005. Functionalization of carbon nanotubes with amines and enzymes. Chemical Physics Letters 402, 96–101. Weis, L.M., Rummel, A.M., Masten, S.J., Trosko, J.E., Upham, B.L., 1998. Bay or baylike regions of polycyclic aromatic hydrocarbons were potent inhibitors of gap junctional
intercellular communication. Environmental Health Perspectives 106 (1), 17–22. Wick, P., Manser, P., Limbach, L.K., Dettlaff-Weglikowskab, U., Krumeich, F., Roth, S., Stark, W.J., Bruinink, A., 2007. The degree and kind of agglomeration affect carbon nanotube cytotoxicity. Toxicology Letters 168, 121–131. Wirnitzer, U., Herbold, B., Voetz, M., Ragot, J., 2009. Studies on the in vitro genotoxicity of baytubes, agglomerates of engineered multi-walled carbon-nanotubes (MWCNT). Toxicology Letters 186, 160–165. Wo¨rle-Knirsch, J.M., Pulskamp, K., Krug, H.F., 2006. Oops they did it again! Carbon nanotubes hoax scientists in viability assays. Nano Letters 6 (6), 1261–1268. Xu, F.-M., Xu, J.-P., Ji, J., Shen, J.-C., 2008. A novel biomimetic polymer as amphiphilic surfactant for soluble and biocompatible carbon nanotubes (CNTs). Colloids and Surfaces B: Biointerfaces 67, 67–72. Yamasaki, Y., 1990. Gap junctional intercellular communication and carcinogenesis. Carcinogenesis 11, 1051–1058. Yang, H., Liu, C., Yang, D., Zhanga, H., Xia, Z., 2009. Comparative study of cytotoxicity, oxidative stress and genotoxicity induced by four typical nanomaterials: the role of particle size, shape and composition. Journal of Applied Toxicology 29, 69–78. Yang, S.-T., Wang, X., Jia, G., Guc, Y., Wang, T., Nie, H., Ge, C., Wang, H., Liu, Y., 2008. Long-term accumulation and low toxicity of single-walled carbon nanotubes in intravenously exposed mice. Toxicology Letters 181, 182–189. Yang, Z., Chen, X.H., Chen, C.S., 2007. Noncovalent-wrapped sidewall multi-walled carbon nanotubes functionalization with polyimide. Polymer Composites 28 (1), 36–41. Ye, S.-F., Wu, Y.-H., Hou, Z.-Q., Zhang, Q.-Q., 2009. ROS and NF-jB are involved in upregulation of IL-8 in A549 cells exposed to multi-walled carbon nanotubes. Biochemical and Biophysical Research Communications 379, 643–648. Yu, B.Z., Yang, J.S., Li, W.X., 2007. In vitro capability of multi-walled carbon nanotubes modified with gonadotrophin releasing hormone on killing cancer cells. Carbon 45 (10), 1921–1927. Yun, Y., Dong, Z., Tan, Z., Schulz, M.J., Shanov, V., 2009. Fibroblast cell behavior on chemically functionalized carbon nanomaterials. Materials Science and Engineering C 29, 719–725. Zhu, Y., Ran, T., Li, Y., Guo, J., Li, W., 2006. Dependence of the cytotoxicity of multi-walled carbon nanotubes on the culture medium. Nanotechnology 16, 4668–4674. Zong, S., Cao, Y., Jua, H., 2007. Direct electron transfer of hemoglobin immobilized in multiwalled carbon nanotubes enhanced grafted collagen matrix for electrocatalytic detection of hydrogen peroxide. Electroanalysis 19 (7–8), 841–846.
water research 44 (2010) 521–532
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Oxidation of atenolol, propranolol, carbamazepine and clofibric acid by a biological Fenton-like system mediated by the white-rot fungus Trametes versicolor Ernest Marco-Urrea a, Jelena Radjenovic´ b, Gloria Caminal c, Mira Petrovic´ b,d, Teresa Vicent a,*, Damia` Barcelo´ b,e a
Departament d’Enginyeria Quı´mica and Institut de Cie`ncia i Tecnologia Ambiental, Universitat Auto`noma de Barcelona (UAB), 08193 Bellaterra, Spain b Department of Environmental Chemistry, IDAEA-CSIC, c/Jordi Girona 18–26, 08034 Barcelona, Spain c Unitat de Biocata`lisis Aplicada associada al IQAC (CSIC-UAB), Escola Te`cnica Superior d’Enginyeria, UAB, 08193 Bellaterra, Spain d Institucio´ Catalana de Recerca i Estudis Avanc¸ats (ICREA), Barcelona, Spain e Institut Catala` de Recerca de l’Aigua (ICRA), Parc Cientı´fic i Tecnolo´gic de la Universitat de Girona, Pic de Peguera, 15, 17003 Girona, Spain
article info
abstract
Article history:
Biological advanced oxidation of the pharmaceuticals clofibric acid (CA), carbamazepine
Received 31 March 2009
(CBZP), atenolol (ATL) and propranolol (PPL) is reported for the first time. Extracellular
Received in revised form
oxidizing species were produced through a quinone redox cycling mechanism catalyzed by
16 September 2009
an intracellular quinone reductase and any of the ligninolytic enzymes of Trametes versi-
Accepted 21 September 2009
color after addition of the lignin-derived quinone 2,6-dimethoxy-1,4-benzoquinone (DBQ)
Available online 22 September 2009
and Fe3þ-oxalate in the medium. Time-course experiments with approximately 10 mg L1 of initial pharmaceutical concentration resulted in percent degradations above 80% after
Keywords:
6 h of incubation. Oxidation of pharmaceuticals was only observed under DBQ redox
Trametes versicolor
cycling conditions. A similar degradation pattern was observed when CBZP was added at
Pharmaceuticals
the environmentally relevant concentration of 50 mg L1. Depletion of DBQ due to the attack
Hydroxyl radical
of oxidizing agents was assumed to be the main limiting factor of pharmaceutical degra-
Carbamazepine
dation. The main degradation products, that resulted to be pharmaceutical hydroxylated
Beta-blockers
derivatives, were structurally elucidated. The detected 4- and 7-hydroxycarbamazepine
Clofibric acid
intermediates of CBZP degradation were not reported to date. Total disappearance of intermediates was observed in all the experiments at the end of the incubation period. ª 2009 Elsevier Ltd. All rights reserved.
1.
Introduction
The presence of pharmaceuticals and personal care products (PPCPs) and their metabolites in waters has become an emerging environmental issue. To date, most attention has been focused on identification, fate and distribution of PPCPs
in municipal wastewater treatment plants (WWTPs), which are commonly found at very low concentrations (low ppb levels) (Radjenovic et al., 2007, 2009, Reemtsma et al., 2006). To avoid their potential adverse health effects on both humans and environment, research efforts are underway to develop efficient techniques for their removal. Among the alternatives
* Corresponding author. Departament d’Enginyeria Quı´mica and Institut de Cie`ncia i Tecnologia Ambiental. Universitat Auto`noma de Barcelona (UAB). 08193 Bellaterra, Spain. Tel.: þ34 935812142; fax: þ34 935812013. E-mail address:
[email protected] (T. Vicent). 0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2009.09.049
522
water research 44 (2010) 521–532
to traditional water treatment processes, advanced oxidation processes (AOPs) have received great attention to degrade PPCPs (Klavarioti et al., 2009; Song et al., 2008). AOPs can be broadly described as aqueous phase oxidation methods based on the intemediacy of highly reactive species such as hydroxyl radicals (,OH). This radical acts as a non-selective, very strong oxidant agent with the ability to react with chemicals giving dehydrogenated or hydroxylated derivates, until achieving their total mineralization. To date, the production of ,OH radicals was based on chemical, photochemical, photocatalytical or electrochemical techniques, but little is known about the possibility to use biological systems to produce ,OH radicals for pollutant degradation (Klavarioti et al., 2009). This biological approach was proposed recently for several whiterot fungi (WRF) including Trametes versicolor and Pleurotus eringyi through a simple quinone redox cycling mechanism that is referred to as advanced biooxidation (Go´mez-Toribio et al., 2009a). WRF are basidiomycetes that are capable of extensive aerobic lignin depolymeration and mineralization due to the low substrate specificity and high reactivity of the enzymes they produce (principally laccase and peroxidases). Such combinations of enzymes with auxiliary oxidoreductive enzymes such as cytochrome P450 system are commonly involved in the degradation of several PPCPs as for example carbamazepine, clofibric acid, ibuprofen and several endocrine disrupting chemicals (Cabana et al., 2007; Marco-Urrea et al., 2009a). In this work, the anti-epileptic drug carbamazepine (CBZP), lipid regulator clofibric acid (CA), and b-blockers propranolol (PPL) and atenolol (ATL) were the selected pharmaceuticals due to their long-term use in Europe and North America and their subsequent occurrence in the aquatic environment (Alder et al., 2006). The metabolism of these pharmaceuticals in mammals has been well established. It is known that CBZP undergoes extensive hepatic metabolism by the cytochrome P450 system, with oxidation to 10,11-dihydro-10,11-epoxy carbamazepine, its further hydration to 11-dihydro-10,11epoxy carbamazepine and conjugation of the latter one with glucuronide being the main metabolic pathway of CBZP (Kerr et al., 1994). CA is the pharmacologically active derivative of clofibrate and several other fibrates, and it is metabolized to reactive acylating derivatives that have been shown to transacylate glutathione forming the corresponding glutathione conjugate (Grillo and Benet, 2001). The primary metabolic pathways of PPL are glucuronidation, side-chain oxidation and ring oxidation, with the main metabolites arising from Ndealkylation of the isopropanolamine side-chain, naphthalene ring hydroxylation, and side-chain O-glucuronidation (Luan et al., 2005). ATL is excreted eagerly via urine as an unchanged compound (90%) (Dollery, 1991), with a small percentage of atenolol-glucuronide (0.8–4.4%) and hydroxyatenolol (1.1– 4.4%, hydroxylation of the benzilic position). Limited removal of ATL and CA and no removal of CBZP have been frequently observed in municipal WWTPs and subsequently they deserve more attention due to the high risk of passing through later barriers in partly closed water cycles (Radjenovic et al., 2007; Reemtsma et al., 2006; Ternes, 1998). Joss et al. (2003) found that CBZP was not expected to produce acute toxic effects in the aquatic biota, but chronic and
synergistic effects with other chemicals could not be excluded. According to their results and regarding the present European legislation on the classification and labelling of chemicals (92/32/EEC), they classified CBZP as ‘‘R52/53 Harmful to aquatic organisms and may cause long-term adverse effects in the aquatic environment’’. On the other side, CA has been reported to be a non-hazardous, yet environmentally persistent compound (estimated persistence of 21 years in the environment) (Buser et al., 1998; Ferrari et al., 2003). Nevertheless, the possibility of endocrine disruption activity of CA through interference with cholesterol synthesis was reported (Pfluger and Dietrich, 2001), which is particularly important regarding its long-term effects. Although specific environmental effects of b-blockers are low, in a mixture with other bblockers the effect of concentration addition can occur, as shown in tests with Daphnia magna and phototoxicity assays with green algae (Escher et al., 2006). Considering these facts, removal techniques need to be developed to effectively degrade the selected pharmaceuticals. The objective of this study was to demonstrate for the first time the degradation of the aforementioned pharmaceuticals by biologically induced oxidizing species, produced by the WRF T. versicolor. The advanced biooxidation strategy consists of the incubation of fungi with a lignin-derived quinone (2,6,dimethoxy-1,4-benzoquione, DBQ) and chelated ferric ion (Fe3þ-oxalate). Under these conditions, fungi catalyze the conversion of the quinone into hydroquinone (DBQH2) by an intracellular quinone reductase, and subsequent oxidation of DBQH2 to semiquinone radicals (DBQ,) is performed in the extracellular medium by any of the lignin modifying enzymes of the WRF (laccase and peroxidases). Then, Fenton’s reagent is formed by DBQ, autoxidation catalyzed by Fe3þ, in which Fe2þ and superoxide anion radical (O, 2 ) are generated and Fe2þ(DBQ, þ Fe3þ-oxalate / DBQ þ Fe2þ-oxalate; , , 3þ oxalate þ O2 # Fe -oxalate þ O2 ), followed by O2 dis, þ , mutation (O, 2 þ HO2 þ H / O2 þ H2O2). Production of OH via quinone redox cycling described above was previously proposed by estimating production of 2-thiobarbituric acid reactive substances (TBARS) from 2-deoxyribose and hydroxylation of 4-hydroxybenzoic acid producing 3,4-dimethoxybenzoic acid, showing a high correlation between formation of Fenton’s reagent and TBARS production (Go´mez-Toribio et al., 2009a). The inhibitory effect of ,OH scavengers and catalase on TBARS production rate further strengthened ,OH generation by this mechanism (Go´mez-Toribio et al., 2009a; Guille´n et al., 2000). However, besides production of ,OH radicals, other oxidizing species such as ferryl ion might also be produced in typical Fenton reactions under certain conditions of pH and concentration of organic and inorganic ligands (Hug and Leupin, 2003).Therefore, more research is needed to ascertain the relative contribution of ,OH produced under quinone redox cycling conditions. This quinone redox cycle mechanism applied to WRF increases the range of environmental pollutants susceptible to be degraded by these microorganisms due to the high oxidation power and low substrate specificity of the oxidizing species generated in comparison with their ligninolytic enzymes (Marco-Urrea et al., 2009b). These oxidizing species are produced rapidly in the extracellular medium and almost total destruction of pollutants can be achieved during the first hours
water research 44 (2010) 521–532
of incubations without the need of adaptation of fungi to pollutants (Go´mez-Toribio et al., 2009a, b). Furthermore, degradation metabolites were elucidated, which were further oxidized either by the induced oxidizing agents or by the ligninolytic enzyme system of fungus. Studies concerning degradation of pharmaceuticals by fungi are very scarce, with 1- and 2-hydroxy ibuprofen, and 1,2-dihydroxy ibuprofen being the only fungal metabolites of pharmaceutical reported up to date (Marco-Urrea et al., 2009a). Nevertheless, the knowledge of these metabolites is necessary for safe application of fungi biocatalyst as bioremediation technology, as well as for the elucidation of the key steps of the degradation process.
2.
Materials and methods
2.1.
Chemicals
CA (CAS No. 882-09-7), CBZP (CAS No. 298-46-4), ATL (CAS No. 29122-68-7), PPL (CAS No. 3506-09-0) and DBQ (CAS No. 530-55-2) were obtained from Sigma–Aldrich Co. All other chemicals used were of analytical grade.
2.2.
Fungus and culture conditions
T. versicolor (ATCC#42530) was maintained by subculturing on 2% malt extract agar slants (pH 4.5) at room temperature. Subcultures were routinely made every 30 days. Pellets of T. versicolor were produced by inoculating 1 mL of a mycelial suspension, prepared as described previously (Marco-Urrea et al., 2008), in 1 L Erlenmeyer flask containing 250 mL of malt extract medium. This was shaken (135 rpm, r ¼ 25 mm) at 25 C for 5 days. Subsequent pellets formed by this process were transferred to another 1 L Erlenmeyer flask containing 250 mL of a defined medium described elsewhere (Marco-Urrea et al., 2008) and were also incubated for 2 days in shaking conditions.
2.3.
Degradation experiments
In time-course experiments, induction of oxidizing agents in T. versicolor via quinone redox cycling was routinely performed as follows. Mycelial pellets from each sample flask were collected by filtration and washed three times with MilliQ water. Appropriate amounts of 2-day old washed mycelium pellets were incubated in the reaction mixture (see figure legends). Reaction mixture contained 500 mM DBQ, 100– 300 mM Fe3þ-oxalate, and 100 mM Mn2þ in 25 mL 20 mM phosphate buffer, pH 5, based on previous optimization of ,OH production in quinone redox cycling (Go´mez-Toribio et al., 2009b). Twenty mL of a solution containing the corresponding pharmaceutical in acetonitrile was added into the flasks to give the desired final pharmaceutical concentration (approximately 10 mg L1). The flasks were incubated at 25 oC and 130 rpm and samples were taken at each point for analysis. The samples were filtered through a Millex-GV (Millipore) 0.22 mm filter and subsequently analyzed by HPLC. In order to investigate the degradation of pharmaceuticals present at environmentally relevant concentrations, CBZP was selected as a model compound and tests flasks were
523
amended with CBZP to a 50 mg L1 concentration. Erlenmeyer flasks containing 50 mL of the corresponding reaction mixture were sacrificed at each experiment time and were filtered through 0.45 mm glass fiber filter from Whatman. The target compound was extracted in one step by solid phase extraction with Oasis HLB cartridges (60 mg adsorbent, Waters, Barcelona, Spain) as is described elsewhere (Radjenovic et al., 2007). Briefly, the cartridges were preconditioned sequentially with 5 mL of methanol and 5 mL of deionized water at neutral pH. The cartridge was dried under vacuum and was eluted with two 2-mL portions of methanol and subsequently concentrated to dryness under a gentle nitrogen stream. The extracts were reconstituted with 0.5 mL 25:75 (v/v) acetonitrile-water. To obviate the possible influence of light on pharmaceuticals stability, all the experiments were carried out in the dark. Each experiment included control flasks (without Fe3þoxalate). All the results included in the text and shown in figures are the mean and standard deviation of duplicate experiments.
2.4.
Analytical procedures
2.4.1.
Analysis of pharmaceuticals and DBQ
Analysis of the pharmaceuticals and DBQ were performed using a Dionex 3000 Ultimate HPLC equipped with a UV detector at 230 nm. The column temperature was 30 C and a sample volume of 20 mL was injected from a Dionex autosampler. Chromatographic separation of CBZP, PPL, CA and DBQ was achieved on a GraceSmart RP 18 column (250 mm 4 mm, particle size 5 mm). The mobile phase consisted of 6.9 mmol L1 acetic acid adjusted to pH 4 (by NaOH) with 35% v/v acetonitrile. It was delivered isocratically at 1 mL min1 as was described elsewhere (Stafiej et al., 2007). ATL analysis was performed with a Ascentis C18 column (150 mm 4.6 mm, particle size 5 mm). Mobile phase of 0.01 M ammonium acetate (pH 7) and mobile phase B (acetonitrile) were delivered at flow rate of 1.2 mL min1 and used for gradient elution of ATL (t ¼ 0 min A ¼ 95%, t ¼ 20 min A ¼ 80%). The detection limit was calculated to be <0.125 mg L1.
2.4.2.
Identification of pharmaceutical metabolites
To identify the major metabolites of CBZP, CA, ATL and PPL, time-course experiments were carried out as described above. Accurate mass MS and MS/MS analyses of parent compounds and their microbial products were performed using a Waters/ Micromass QqToF-Micro system coupled to Waters ACQUITY UPLC system (Micromass, Manchester, UK). Samples from biodegradation experiments were analyzed on a Waters ACQUITY BEH C18 column (10 2.1 mm, 1.7 mm particle size). UPLC analysis was performed in the same way and with the same mobile phases on both instruments. After elution from the column, CBZP, ATL and PPL were analyzed in positive ion (PI) mode with mobile phases consisting of (A) 5 mM aqueous NH4Ac/acetic acid (pH 4.8); and (B) acetonitrile-methanol (2:1, v/v) at 350 mL min1. The elution of ATL and PPL started at 5% B for 1 min and then it was linearly increased to 80% of B in 10.5 min, further increased to 95% of B in the next 1 min, and then returned to initial conditions. Total run time, including the conditioning of the column to the initial conditions was 14 min. For CBZP after the first min at 5% of A the % of A was
524
water research 44 (2010) 521–532
Fig. 1 – Time-course of CBZP (a) and CA (b) degradation in T. versicolor cultures under quinone redox cycle conditions. Symbols: (B) CBZP and CA in the reaction mixture (25 mL 20 mM phosphate buffer containing 500 mM DBQ, 100–300 mM Fe3D-oxalate, 100 mM Mn2D, and 10 mg LL1 of the corresponding pharmaceutical); (C) CBZP and CA in the control treatment (non-containing DBQ and Fe3D-oxalate); (-) DBQ(H2) in the reaction mixture; (:) metabolite formation expressed as relative area (A/A0) where A is the corresponding metabolite and A0 is the parent drug in the control treatment at time zero. In the case of CBZP where two hydroxylated isomers were found, black and white triangles refer to P254A and B, respectively. Incubations were carried out with 1.5 ± 0.1 and 1.7 ± 0.2 mg dry weight mLL1 for CBZ and CA, respectively.
increased to 60% at 11 min, further increased to 90% of A in the next 3 min and held isocratically for 2 min. Together with returning to initial conditions, the total run time was 20 min. CA was analyzed in the negative ion (NI) mode with mobile phases consisting of (A) methanol; and (B) water. The elution started at 5% of A for the first min, which was raised to 70% in the next 7 min, to 90% in the next 2 min, and then returned to initial conditions. The total run time was 13 min. The injection volume of the sample was 10 mL. The mass spectrometry analysis on the QqToF instrument was performed in wide pass quadrupole mode, for MS experiments, with the ToF data being collected between m/z 50–800. The capillary and cone voltages were set to 3000 and 25–30 V, respectively. Data were collected in the centroid mode, with a scan accumulation time of 1 s. The instrument was operated at a resolution of 5000 (FWHM). The nebulisation gas was set to 500 L h1 at a temperature of 350 C,
the cone gas was set to 50 L h1, and the source temperature to 120 C. All analyses were acquired using an independent reference spray via the LockSpray interference to ensure accuracy and reproducibility. Sulfaguanidine (C7H10N4O2S) was used as the internal lock mass in both PI mode ([M þ H]þ ¼ m/z 215.0602) and NI mode ([M H] ¼ m/z 213.0446). The LockSpray frequency was set at 11 s. Fragmentation of precursor ions was done by applying collision energies in the range of 10–35 eV, using argon as a collision gas at a pressure of ~20 psi. Elemental compositions of the molecular ions and their fragments were determined and exact masses were calculated with the help of MassLynx V4.1 software incorporated in the instrument. Since the software calculation of the accurate mass of cation is performed by adding a hydrogen atom instead of proton, mass of one electron (i.e., 0.0005) was subtracted from the calculated mass.
water research 44 (2010) 521–532
525
Fig. 2 – Time-course of PPL (a) and ATL (b) degradation in T. versicolor cultures under quinone redox cycle conditions. Culture conditions were as described in the legend for Fig. 1. Symbols: (B) PPL and ATL in the reaction mixture; (C) PPL and ATL in the control treatment; (-) DBQ(H2) in the reaction mixture; (:) metabolite formation expressed as relative area (A/A0) where A is the corresponding metabolite and A0 is the parent drug in the control treatment at time zero. Incubations were carried out with 1.9 ± 0.1 mg dry weight mLL1 for both PPL and ATL.
2.4.3.
Mycelial dry weight
Mycelial dry weights were determined by vacuum filtering the cultures through reweighed glass filters (Whatman GF/C, Maidstone, England). The filters containing the mycelial mass were placed in glass dishes and dried at 100 C to constant weight.
3.
Results and discussions
3.1. Oxidation of CA, CBZP, PPL and ATL by T. versicolor under quinone redox cycling conditions The strategy proposed here to degrade the selected pharmaceuticals couples both AOP and biological strategies by inducing the production of oxidizing agents such as ,OH in the white-rot fungus T. versicolor. In the experiments conducted,
oxalate was used as chelant agent of Fe3þ instead of EDTA and Mn2þ was added into the incubation mixture, since the oxidizing power under these conditions was substantially improved as discussed previously (Go´mez-Toribio et al., 2009b). Figs. 1 and 2 show the time-courses degradation of the selected pharmaceuticals in test flasks amended with approximately 10 mg L1 of the target pharmaceuticals, when incubated under DBQ redox cycling conditions. Control treatments in the absence of DBQ and Fe3þ but containing fungus were performed to demonstrate both the positive involvement of oxidizing agents formed from the quinone redox cycling mechanism and to discard the role of fungal sorption in the removal of pharmaceuticals. Furthermore, control treatments permit us to rule out the intracellular and extracellular enzymatic system of T. versicolor in the degradation of the pharmaceuticals. Previously CA and CBZP were reported to be degraded by the cyt P450 system of T. versicolor grown in a chemically define medium (Marco-Urrea et al.,
526
water research 44 (2010) 521–532
experiment, adsorption and transformation of pharmaceutical by fungus cannot be clearly distinguished, since live cultures can also degrade adsorbed pollutants by intracellular mechanisms (Bla´nquez et al., 2004). In this case, adsorption was higher than that observed at 10 mg L1 probably due to the highest concentration of fungal mycelium in the experimental flasks (see figure legend). Similarly to that described for higher concentrations, the plateau in CBZP degradation under quinone redox cycling conditions was observed at the fifth hour, when DBQ(H2) was expected to be depleted under the tested conditions.
3.2. QqToF-MS2 fragmentation pathways of pharmaceuticals and their metabolites
Fig. 3 – Time-course of CBZP degradation added at low concentration of 50 mg LL1 in T. versicolor cultures under quinone redox cycle conditions. Symbols: (;) CBZP in the reaction mixture; (B) CBZP in the control treatment; (C) CBZP in uninoculated bottles. Incubations were carried out with 3.8 ± 0.1 mg dry weight mLL1. Values plotted are means ± standard deviations for triplicate cultures.
2009a) but as can be observed in Figs. 1 and 2 degradation did not occur in control treatments for the incubation period studied. Results in Figs. 1 and 2 show a steep decrease of the concentration of pharmaceuticals in the first hours of incubation when the fungus was incubated under quinone redox cycling conditions. Approximately 50% of CBZP, CA and ATL were degraded the first hour and a plateau was reached above 80% after the sixth hour of incubation. However, the incubation was continued to 24 h to investigate the further degradation of possible metabolites formed during the degradation process (see Section 3.3). The consumption of DBQ was also investigated in the same samples used for the determination of pharmaceutical degradation. This analysis included the determination of DBQ and its counterpart (DBQH2) levels and is shown in Figs. 1 and 2 as the sum of both: DBQ(H2). A high correlation between DBQ(H2) and pharmaceutical degradation was found, suggesting that DBQ(H2) consumption was due to the attack of the induced oxidizing agents. Furthermore, since quinone redox cycling and therefore the production of oxidizing species cannot be sustained without the presence of DBQ(H2), the plateau observed in the pharmaceutical degradation rate can be ascribed to DBQ(H2) depletion. In an effort to discuss the relevance of this strategy in regards to the normal concentrations of pharmaceuticals found in aqueous environments (that are frequently detected at levels from ng L1 to low mg L1), one experiment was carried out at 50 mg L1. For this purpose CBZP was chosen as model compound since it is found to be recalcitrant in sewer treatment facilities and in the environment (Zhang et al., 2008). As is shown in Fig. 3 the concentration of CBZP was decreased to approximately 10 mg L1, due to the mechanisms of fungal sorption and oxidation-induced degradation. In this
Table 1 summarizes the exact masses of molecular and fragment ions, together with recalculated mass errors and double bond equivalents (DBEs) given by the software (mass measurements accuracy threshold of 5 mg L1). Since all fragment ions had an even number of electrons (i.e., DBE was half integer number in all cases), all neutral losses had likewise even electron configurations. These data were obtained under optimized conditions of collision energy and cone voltage in ESI(þ) and ESI() MS2 experiments on the QqToF instrument. In ESI(þ) full-scan experiments performed at UPLC-QqToF instrument, ATL (molecular ion [M þ H]þ267) eluted at 3.4 min. Another peak was observed at 3.2 min, with the molecular ion [M þ H]þ283, denominated as metabolite P282. Considering that the mass of P282 was shifted 16 Da upwards relative to the parent compound, hydroxylation by ,OH radical attack was assumed. In Table 1 are presented calculated and measured masses of fragment ions of ATL and its metabolite P282, determined in ESI(þ)-MS2 experiments at the QqToF-MS instrument. The most abundant fragment ions in the MS2 spectrum of ATL were detected m/z 190 (loss of 77 Da, isopropylamine and water) and m/z 145 (further cleavage of CO and intermolecular cyclization and rearrangement) (see Fig. 4c). In the fragmentation pattern of the molecular ion [M þ H]þ283, characteristic losses of isopropyl group (42 Da) and ammonia (17 Da) afforded the fragment ions at m/z 241 and 224 (see Fig. 4d). The subsequent losses of water and CO from the fragment ion m/z 224 led to the formation of m/z 206 and 178 fragment ions. The fragment ion m/z 161 was formed through cleavage of isopropylamine, water, CO and ammonia and intramolecular cyclization, as suggested in the insert in Fig. 4d. The fragment ion m/z 116 was also present in the MS2 spectrum of P282, indicating that the hydroxylation had to occur at the p-hydroxyphenylacetamide fraction of ATL. From the comparison of the two fragmentation patterns, it was deduced that the –OH group was attached to the alkyl sidechain, at the C-atom next to the ether oxygen. In the case of PPL, MS2 fragmentation of the molecular ion [M þ H]þ260 rendered intense signals at m/z 183 and 157, corresponding to the subsequent cleavages of aminoisopropyl moiety and water, and C2H2, respectively (see Fig. 4a). The less intense fragment ions at m/z 242 and 218 were formed through cleavages of water and isopropylamine, respectively. The fungal metabolite of PPL eluted 0.6 min earlier, with the molecular ion [M þ H]þ276, and was denominated as P275. Besides the molecular ion with mass 16 Da greater than
Table 1 – Accurate mass measurement of product ions of carbamazepine (CBZP), clofibric acid (CA), propranolol (PPL), and atenolol (ATL) and their degradation products P254, P230, P275 and P282, respectively, as determined by UPLC-QqToF in ESI(D) MS2 mode for CBZP, PPL and ATL, and ESI(L) MS2 mode for CA. Comp.
CBZP
P254
CA
PPL
P275
ATL
P282
[M þ H]þ [M þ Na]þ [M þ H NH3]þ [M þ H CONH]þ [M þ H]þ [M þ Na]þ [M þ H NH3]þ [M þ H CONH]þ [M þ H CONH3]þ [M þ H CONH CO]þ [M H] [M H C4H6O2] [M H C6H4ClO] [M H] [M H C4H6O2] [M þ H]þ [M þ H H2O]þ [M þ H C(CH3)2]þ [M þ H H2O NH2CH(CH3)2]þ [M þ H H2O NH2CH(CH3)2 C2H2]þ [M þ H C10H8O]þ [M þ H]þ [M þ H H2O]þ [M þ H H2O O NH2CH(CH3)2]þ [M þ H H2O O NH2CH(CH3)2 C2H2]þ [M þ H C10H8O2]þ [M þ H]þ [M þ H C(CH3)2]þ [M þ H C(CH3)2 NH3]þ [M þ H C(CH3)2 H2O]þ [M þ H C(CH3)2 NH3 H2O CO NH3]þ [M þ H CHCONH2 H2O C(CH3)2 NH3]þ [M þ H C8H9NO2]þ [M þ H]þ [M þ H C(CH3)2]þ [M þ H C(CH3)2 NH3]þ [M þ H C(CH3)2 NH3 H2O]þ [M þ H C(CH3)2 NH3 H2O CO]þ [M þ H C(CH3)2 NH3 CO H2O NH3]þ [M þ H C8H9NO3]þ
Elemental formula
C15H13N2O C15H12N2ONa C15H10NO C14H12N C15H13N2O2 C15H12N2O2Na C15H10NO2 C14H12NO C14H10NO C13H12N C10H10ClO3 C6H4ClO C4H6O2 C10H10ClO4 C6H4ClO2 C16H22NO2 C16H20NO C13H16NO2 C13H11O C11H9O C6H14NO C16H22NO3 C16H20NO2 C13H11O C11H9O C6H14NO C14H23N2O3 C11H17N2O3 C11H14NO3 C11H12NO2 C10H9O C9H9O C6H14NO C14H23N2O4 C11H17N2O4 C11H14NO4 C11H12NO3 C10H12NO2 C10H9O2 C6H14NO
DBEa
Error
Exp.
Theor.
mDa
ppm
237.1019 259.0844 220.0761 194.0960 253.0976 275.0799 236.0710 210.0906 208.0762 182.0965 213.0327 126.9944 85.0294 229.0254 142.9909 260.1653 242.1547 218.1170 183.0814 157.0648 116.1066 276.1583 258.1492 183.0795 157.0664 116.1079 267.1696 225.1222 208.0975 190.0874 145.0650 133.0658 116.1070 283.1663 241.1186 224.0901 206.0807 178.0862 161.0617 116.1055
237.1022 259.0842 220.0757 194.0964 253.0972 275.0791 236.0706 210.0913 208.0762 182.0964 213.0324 126.9956 85.0290 229.0268 142.9905 260.1645 242.1539 218.1176 183.0804 157.0648 116.1070 276.1594 258.1489 183.0804 157.0648 116.1070 267.1704 225.1234 208.0969 190.0863 145.0648 133.0648 116.1071 283.1652 241.1183 224.0917 206.0812 178.0863 161.0597 116.1070
0.3 0.2 0.4 0.4 0.4 0.8 0.4 0.7 0 0.1 0.3 1.2 0.4 1.4 0.4 0.8 0.8 0.6 1.0 0 0.4 1.1 0.3 0.9 1.6 0.9 0.8 5.3 0.6 1.1 0.2 1.0 0.1 1.1 0.3 1.6 0.5 0.1 2.0 1.5
1.3 0.8 1.8 2.1 1.6 2.9 1.7 3.3 0 0.5 1.4 9.4 4.7 6.1 2.8 3.1 3.3 2.7 5.5 0 3.4 4.0 1.2 4.9 10.1 7.7 3.0 1.2 2.9 5.8 1.4 7.5 0.9 3.9 1.2 7.1 2.4 0.6 12.4 12.9
10.5 10.5 11.5 9.5 10.5 10.5 11.5 9.5 10.5 8.5 5.5 4.5 2.5 5.5 4.5 6.5 7.5 6.5 8.5 7.5 0.5 6.5 7.5 8.5 7.5 0.5 4.5 4.5 5.5 6.5 6.5 5.5 0.5 4.5 4.5 5.5 6.5 5.5 6.5 0.5
trb, min
8.93
7.24; 7.81
5.90
6.40 6.7
6.1
3.4
3.2
527
Only product ions with abundances higher than 10% are taken into account. a DBE, double bond equivalent. b tr-UPLC retention time.
Mass (m/z)
water research 44 (2010) 521–532
P230
Precursor ion/product ion
528
water research 44 (2010) 521–532
Fig. 4 – Spectra obtained in ESI(D)-MS/MS experiments at QqToF instrument (cone voltages 20–25 V, collision energies 15– 25 eV) for: (a) standard mixture of PPL in methanol/water (v/v, 25/75) at 10 mg LL1, (b) degradation product of PPL, P275, in the sample taken after 2 h, (c) standard mixture of ATL in methanol/water (v/v, 25/75) at 10 mg LL1, and (d) degradation product of ATL, P282, in the sample taken after 4 h.
[M þ H]þ260, fragment ion at m/z 258 was observed, equivalent to the m/z 242 in the MS2 spectrum of PPL (Fig. 4b). The rest of the observed fragment ions of the molecular ion [M þ H]þ276 were the same (i.e., m/z 116, 157 and 183), indicating that the naphthalene moiety was the site of ,OH radical attack. In full-scan experiments CBZP eluted at 8.94 min, whereas two new peaks emerged at 7.24 and 7.81 min. The identical fragmentation patterns of these two peaks and their molecular weight (MW) 16 Da higher than the parent compound (see Figs. 5c,d) suggested that the formed products were two hydroxylated isomers of CBZP, denominated as P254A and B. In Table 1 exact and measured masses were presented for a more intense M254B. The collision-induced-dissociation experiments with CBZP (molecular ion [M þ H]þ237) revealed the formation of only two fragment ions at m/z 220 and 194, as a consequence of the loss of ammonia and CONH group, respectively. Equivalent fragments were observed for P254A,B, at m/z 236 and 210. The strong signal observed at m/z 208 suggested that the –OH group was probably attached at the C4 and C7-atom for the two observed isomers, since the loss of 2 H-atoms could be favoured by intramolecular cyclization as presented in the insert of Fig. 5d. Further cleavage of CO in the
fragment ion m/z 208 afforded the signal at m/z 182. Nevertheless, based on the retention times only it could not be deduced which peak belongs to which product. Probably the formation of intramolecular hydrogen bond between the attached oxygen and hydrogen atoms of the amide group caused one of the isomers to elute ~0.6 min later than the other. Finally, ESI() full-scan screenings of samples from the experiments with CA (molecular ion [M þ H]þ213) revealed a new signal appearing 0.5 min after the CA peak, at 6.40 min. Similar to the other metabolites identified, its MW was deferring for 16 Da relative to the original drug, with the molecular ion [M þ H]þ229 (i.e., P230). Very poor MS2 fragmentation was observed for both parent compound and its fungal metabolite. The collision of [M þ H]þ213 rendered two fragment ions at m/z 126 and 85, representing the cleavage of the ether bond (see insert in Fig. 5a), whereas in the spectrum of [M þ H]þ229 only one fragment ion was detected at m/z 142, equivalent to the m/z 126 but with mass shifted upwards for 16 Da. Nevertheless, this was enough evidence to assume hydroxylation of the benzene ring, although the exact position of the attached –OH group could not be deduced.
water research 44 (2010) 521–532
529
Fig. 5 – Spectra obtained in MS/MS experiments at QqToF instrument (cone voltages 15–30 V, collision energies 15–25 eV) for: (a) standard mixture of CA in methanol/water (v/v, 25/75) at 10 mg LL1, (b) degradation product of CA, P230, in the sample taken after 2 h, (c) standard mixture of CBZP in methanol/water (v/v, 25/75) at 10 mg LL1, and (d) degradation product of CBZP, P254, in the sample taken after 2 h.
The degradation intermediates of each pharmaceutical detected in this section are found in Fig. 6.
3.3. Discussion of the results on the fungal metabolites of pharmaceuticals The formation of metabolites was depicted in Figs. 1 and 2 and was expressed as relative area (A/A0) measured by integration of LC-MS peaks of the corresponding metabolite (A) and the parent drug in the control treatments at time zero (A0), since due to the lack of authentic analytical standards for the newly identified products their quantitative determination was not possible. Examination of the profiles show qualitatively similar formation pattern of metabolites across the incubation time, obtaining a maximum value of A/A0 between the first and fourth hour and disappearing completely in all the cases after 24 h. The metabolites appeared to be formed at low concentration levels in comparison with the parent drug, with a maximum A/A0 values for each case in the range of 0.012 (PPL, Fig. 2a) to 0.27 (CA, Fig. 1b). The same compound that was identified in our experiments as the fungal metabolite of ATL, P282, was previously reported as one of the intermediate products of solar photo-
Fenton and TiO2 photocatalytic treatment of ATL in pilotscale compound parabolic collectors (Radjenovic et al., 2009). In the case of these photocatalytic treatments, P282 was observed together with its keto tautomer (P280), and was formed through ,OH-mediated reactions. Furthermore, Song et al. (2008) reported a product of ATL with MW 282, formed in the reaction of ATL with ,OH radicals. This product was identified by 60Co g-irradiation and LC-MS, although the ,OH attacks was assumed to occur at the benzene ring. In the same study hydroxylated derivative of PPL was also detected with MW 275 and –OH group attached at the naphthalene ring, which coincides with the structure of P275 identified in the present study. Hydroxylation of naphthalene moiety is phase I reaction in the human metabolism of PPL, in which 4hydroxy PPL is formed (Luan et al., 2005). Cytochrome P450 isozymes were found to be responsible for oxidative metabolic pathways of PPL in humans proceeding through naphthalene ring-hydroxylations at the 4-, 5-, and 7-positions and side-chain N-desisopropylation (Masubuchi et al., 1994). However, in our study no metabolites were found in fungal control treatments indicating that P275 was generated via induction of oxidizing agents instead of fungal cyt P450 mechanism.
530
water research 44 (2010) 521–532
Fig. 6 – Selected pharmaceuticals and proposed structures of their degradation products.
Sire´s et al. (2007) proposed the formation of a hydroxylated intermediate as the observed in our study after the hydroxylation of CA on its C2-position by electro-Fenton and photoelectro-Fenton. This proposed metabolite was not identified by pure standards in the mentioned study but it was assigned to a dehydrated species of 2-(4-chloro-2hydroxyphenoxy)-2-methylpropionic acid on the basis of its mass fragmentation spectra. On the other hand, by using an HPLC-DAD and FLD, Doll and Frimmel (2004) proposed TiO2 photocatalytic degradation pathway of CA, which did not include the hydroxylated product as the one observed in our study. During photocatalysis CA underwent substitution of –Cl by –OH group in the para-position, whereas products such as 2-(4-hydroxyphenoxy)-isobutyric acid and hydroquinone were formed. On the other side, cleavage of isobutyric acid from the side-chain afforded 4-chlorophenol and 4-chlorocatechol. Interestingly, in the abovementioned reports, substrates of laccase such as phenol and 4-chlorophenol have been described as typical metabolites of CA degradation by AOP and once produced they could therefore be mineralized by the ligninolytic enzymatic system of WRF. But perhaps even more interesting is the production of different
quinone metabolites (also described for CA degradation by , OH attack) that could theoretically be incorporated into the quinone redox cycling enhancing the degradation rates of the target pollutant. As far as CBZP is concerned, there are few studies regarding identification of products in ,OH-induced degradation pathway. Vogna et al. (2004) reported the formation of toxic acridine intermediates in UV/H2O2 treatment of CBZP, formed in ,OH and ,HO2-mediated reactions. Chiron et al. (2006) also found acridine as major photodegradation intermediate of CBZP after direct photolysis, but in the presence of Feþ3 they identified a hydroxycarbamazepine structure derived from a ,OH attack on the C6 aromatic ring. The formation of 2- and 3-hydroxylated derivatives of CBZP has been reported in the metabolism of CBZP by the cytochrome P450 in mammals (Mandrioli et al., 2001; Pearce et al., 2002) as well as in the two model fungi Cunninghamella elegans and Umbelopsis ramanniana used to study mammalian metabolism for many drugs (Kang et al., 2008). As far as we know, the two hydroxylated metabolites of CBZP in the 4- and 7-position (P254A and B) detected in this study are reported for the first time.
water research 44 (2010) 521–532
4.
Conclusions
Biological advanced oxidation of ATL, CA, CBZP and PPL was demonstrated for the first time by inducing extracellular oxidizing species in T. versicolor. Under quinone redox cycling conditions, pharmaceuticals degradation (added at 10 mg L1) reached a plateau after 6 h of incubation achieving degradation levels up to 80%. The plateau observed was due to DBQ depletion, which was shown to be concomitant to pharmaceutical degradation. The feasibility of this novel technology to degrade pharmaceuticals at low levels commonly found in the environment (50 mg L1) was also demonstrated. Hydroxylated intermediates of all pharmaceuticals were identified and they disappeared after the incubation period, suggesting the total mineralization of the target compound.
Acknowledgements This work was supported by the Spanish MICINN (project CTM2007-60971/TECNO) and MMAMRM (project 010/PC08/3-04). The Department of Chemical Engineering of the UAB is the Unit of Biochemical Engineering of the XRB de la Generalitat de Catalunya. E. Marco-Urrea, T. Vicent and G. Caminal are members of a Consolidated Research Group of Catalonia (2005SGR 00220). Jelena Radjenovic´ gratefully acknowledges the JAE Program (CSIC-European Social Funds).
references
Alder, A.C., Bruchet, A., Carballa, M., Clara, M., Joss, A., Lo¨ffler, D., McArdell, C.S., Miksch, K., Omil, F., Tuhkanen, T., Ternes, T.A., 2006. Consumption and occurrence. In: Ternes, T.A., Joss, A. (Eds.), Human Pharmaceuticals, Hormones and Fragrances. IWA Publishing, London, pp. 15–54. Bla´nquez, P., Casas, N., Font, X., Gabarrell, X., Sarra`, M., Caminal, G., Vicent, T., 2004. Mechanism of textile metal dye biotransformation by Trametes versicolor. Water Research 38 (8), 2166–2172. Buser, H.-R., Mu¨ller, M.D., Theobald, N., 1998. Occurrence of the pharmaceutical drug clofibric acid and the herbicide mecoprop in various Swiss Lakes and in the North Sea. Environmental Science and Technology 32 (1), 188–192. Cabana, H., Jones, J.P., Agathos, S.N., 2007. Elimination of endocrine disrupting chemicals using white-rot fungi and their lignin modifying enzymes: a review. Engineering in Life Sciences 7 (5), 429–456. Chiron, S., Minero, C., Vione, D., 2006. Photodegradation processes of the antiepileptic drug carbamazepine, relevant to estuarine waters. Environmental Science and Technology 40 (19), 5977–5983. Doll, T.E., Frimmel, F.H., 2004. Kinetic study of photocatalytic degradation of carbamazepine, clofibric acid, iomeprol and iopromide assisted by different TiO2 materials – determination of intermediates and reaction pathways. Water Research 38 (4), 955–964. Dollery, C., 1991. Therapeutic Drugs. Churchill Livingstone, Edinburgh. Escher, B.I., Bramaz, N., Richter, M., Lienert, J., 2006. Comparative ecotoxicological hazard assessment of beta-blockers and their
531
human metabolites using a mode-of-action-based test battery and a QSAR approach. Environmental Science and Technology 40 (23), 7402–7408. Ferrari, B., Paxe´us, N., Giudice, R.L., Pollio, A., Garric, J., 2003. Ecotoxicological impact of pharmaceuticals found in treated wastewaters: study of carbamazepine, clofibric acid, and diclofenac. Ecotoxicology and Environmental Safety 55 (3), 359–370. Go´mez-Toribio, V., Garcı´a-Martı´n, A.B., Martı´nez, M.J., Martı´nez, A.T., Guille´n, F., 2009a. Induction of extracellular hydroxyl radical production by white-rot fungi through quinone redox cycling. Applied and Environmental Microbiology 75 (12), 3944–3953. Go´mez-Toribio, V., Garcı´a-Martı´n, A.B., Martı´nez, M.J., Martı´nez, A.T., Guille´n, F., 2009b. Enhancing the production of hydroxyl radicals by Pleurotus eryngii via quinone redox cycling for pollutant degradation. Applied and Environmental Microbiology 75 (12), 3954–3962. Grillo, M.P., Benet, L.Z., 2001. Interaction of gglutamyltranspeptidase with clofibryl- S-acyl-glutathione in vitro and in vivo in rat. Chemical Research in Toxicology 14 (8), 1033–1040. Guille´n, F., Go´mez-Toribio, V., Martı´nez, M.J., Martı´nez, A.T., 2000. Production of hydroxyl radical by the synergistic action of fungal laccase and aryl alcohol oxidase. Archives of Biochemistry and Biophysics 383 (1), 142–147. Hug, S.J., Leupin, O., 2003. Iron-catalyzed oxidation of arsenic(III) by oxygen and by hydrogen peroxide: pH-dependent formation of oxidants in the Fenton reaction. Environmental Science and Technology 37 (12), 2734–2742. Joss, A., Repetto, G., Rios, J.C., Hazen, M.J., Molero, M.L., del Peso, A., Salguero, M., Fernandez-Freire, P., Perez-Martin, J.M., Camean, A., 2003. Ecotoxicological evaluation of carbamazepine using six different model systems with eighteen endpoints. Toxicology in Vitro 17 (5-6), 525–532. Kang, S.-I., Kang, S.-Y., Hur, H.-G., 2008. Identification of fungal metabolites of anticonvulsant drug carbamazepine. Applied Microbiology and Biotechnology 79 (4), 663–669. Kerr, B.M., Thummel, K.E., Wurden, C.J., Klein, S.M., Kroetz, D.L., Gonzalez, F.J., Levy, R.H., 1994. Human liver carbamazepine metabolism: role of CYP3A4 and CYP2C8 in 10,11-epoxide formation. Biochemical Pharmacology 47 (11), 1969–1979. Klavarioti, M., Mantzavinos, D., Casinos, D., 2009. Removal of residual pharmaceuticals from aqueous systems by advanced oxidation processes. Environment International 35 (2), 402–417. Luan, L.-J., Shao, Q., Ma, J.-Y., Zeng, S., 2005. Stereoselective urinary excretion of S-()- and R-(þ)-propranolol glucuronide following oral administration of RS-propranolol in Chinese Han subjects. World Journal of Gastroenterology 11 (12), 1822–1824. Mandrioli, R., Albani, F., Casamenti, G., Sabbioni, C., Raggi, M.A., 2001. Simultaneous high-performance liquid chromatography determination of carbamazepine and five of its metabolites in plasma of epileptic patients. Journal of Chromatography B 762 (2), 109–116. Marco-Urrea, E., Parella, T., Gabarrell, X., Caminal, G., Vicent, T., Reddy, C.A., 2008. Mechanistics of trichloroethylene mineralization by the white-rot fungus Trametes versicolor. Chemosphere 70 (3), 404–410. Marco-Urrea, E., Pe´rez-Trujillo, M., Vicent, T., Caminal, G., 2009a. Ability of white-rot fungi to remove selected pharmaceuticals and identification of degradation products of ibuprofen by Trametes versicolor. Chemosphere 74 (6), 765–772. Marco-Urrea, E., Aranda, E., Caminal, G., Guille´n, F., 2009b. Induction of hydroxyl radical production in Trametes versicolor to degrade recalcitrant chlorinated hydrocarbons. Bioresource Technology 100 (23), 5757–5762.
532
water research 44 (2010) 521–532
Masubuchi, Y., Hosokawa, S., Horie, T., Suzuki, T., Ohmori, S., Kitada, M., Narimatsu, S., 1994. Cytochrome P450 isozymes involved in propranolol metabolism in human liver microsomes. The role of CYP2D6 as ring-hydroxylase and CYP1A2 as N-desisopropylase. Drug Metabolism and Disposition 22 (6), 909–915. Pearce, R.E., Vakkalagadda, G.R., Steven Leeder, J., 2002. Pathways of carbamazepine bioactivation in vitro I. Characterization of human cytochromes P450 responsible for the formation of 2- and 3-hydroxylated metabolites. Drug Metabolism and Disposition 30 (11), 1170–1179. Pfluger, P., Dietrich, D.R., 2001. Effects on pharmaceuticals in the environmentdan overview and principle considerations. In: Ku¨mmerer, K. (Ed.), Pharmaceuticals in the EnvironmentdSources, Fate, Effects and Risks. Springer, Berlin, pp. 11–17. Radjenovic, J., Petrovic, M., Barcelo´, D., 2007. Analysis of pharmaceuticals in wastewater and removal using a membrane bioreactor. Analytical and Bioanalytical Chemistry 387 (4), 1365–1377. Radjenovic, J., Sirtori, C., Petrovic, M., Barcelo´, D., Sixto, M., 2009. Solar photocatalytic degradation of persistent pharmaceuticals at pilot-scale: Kinetics and characterization of major intermediate products, Applied catalysis B: Environmental 89 (1–2), 255–264.
Reemtsma, T., Weiss, S., Mueller, J., Petrovic, M., Gonza´lez, S., Barcelo´, D., Ventura, F., Knepper, T.P., 2006. Polar pollutants entry into the water cycle by municipal wastewater: a European perspective. Environmental Science and Technology 40 (17), 5451–5458. Sire´s, I., Arias, C., Cabot, P.Ll., Centellas, F., Garrido, J.A., Rodrı´guez, R.M., Brillas, E., 2007. Degradation of clofibric acid in acidic aqueous medium by electro-Fenton and photoelectro-Fenton. Chemosphere 66 (9), 1660–1669. Song, W., Cooper, W.J., Mezyk, S.P., Greaves, J., Peake, B.M., 2008. Free radical destruction of b-blockers in aqueous solution. Environmental Science and Technology 42 (4), 1256–1261. Stafiej, A., Pyrzynska, K., Regan, F., 2007. Determination of antiinflammatory drugs and estrogens in water by HPLC with UV detection. Journal of Separation Science 30 (7), 985–991. Ternes, T.A., 1998. Occurrence of drugs in German sewage treatment plants and rivers. Water Research 32 (11), 3245– 3260. Vogna, D., Marotta, R., Andreozzi, R., Napolitano, A., D’Ischia, M., 2004. Kinetic and chemical assessment of the UV/H2O2 treatment of antiepileptic drug carbamazepine. Chemosphere 54 (4), 497–505. Zhang, Y., Geissen, S.U., Gal, C., 2008. Carbamazepine and diclofenac: removal in wastewater treatment plants and occurrence in water bodies. Chemosphere 73 (8), 1151–1161.
water research 44 (2010) 533–544
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Loadings, trends, comparisons, and fate of achiral and chiral pharmaceuticals in wastewaters from urban tertiary and rural aerated lagoon treatments Sherri L. MacLeod a,1, Charles S. Wong a,b,* a
Department of Chemistry, University of Alberta, Edmonton, Alberta, T6G 2G2, Canada Environmental Studies Program and Department of Chemistry, Richardson College for the Environment, University of Winnipeg, Winnipeg, Manitoba, R3B 2E9, Canada b
article info
abstract
Article history:
A comparison of time-weighted average pharmaceutical concentrations, loadings and
Received 1 June 2009
enantiomer fractions (EFs) was made among treated wastewater from one rural aerated
Received in revised form
lagoon and from two urban tertiary wastewater treatment plants (WWTPs) in Alberta,
14 September 2009
Canada. Passive samplers were deployed directly in treated effluent for nearly continuous
Accepted 23 September 2009
monitoring of temporal trends between July 2007 and April 2008. In aerated lagoon effluent,
Available online 27 September 2009
concentrations of some drugs changed over time, with some higher concentrations in winter likely due to reduced attenuation from lower temperatures (e.g., less microbially
Keywords:
mediated biotransformation) and reduced photolysis from ice cover over lagoons; however,
Pharmaceuticals
concentrations of some drugs (e.g. antibiotics) may also be influenced by changing use
Passive samplers
patterns over the year. Winter loadings to receiving waters for the sum of all drugs were
Wastewater
700 and 400 g/day from the two urban plants, compared with 4 g/day from the rural plant.
Loadings
Per capita loadings were similar amongst all plants. This result indicates that measured
Trends
loadings, weighted by population served by WWTPs, are a good predictor of other effluent
Chiral drugs
concentrations, even among different treatment types. Temporal changes in chiral drug EFs were observed in the effluent of aerated lagoons, and some differences in EF were found among WWTPs. This result suggests that there may be some variation of microbial biotransformation of drugs in WWTPs among plants and treatment types, and that the latter may be a good predictor of EF for some, but not all drugs. ª 2009 Elsevier Ltd. All rights reserved.
1.
Introduction
Pharmaceutical occurrence, fate, and toxicity in aquatic ecosystems has been studied for more than 10 years (Daughton and Ternes, 1999; Halling-Sorensen et al., 1998; Kolpin
et al., 2002; Stamm et al., 2008). While drugs have been consistently detected in surface water, the effects of these complex mixtures of bioactive pollutants on non-target aquatic organisms are not fully understood (Fent et al., 2006). Accurate exposure scenarios are essential to assess toxicity
Abbreviations: CR, Capital Region; DL, method detection limit; EF, enantiomer fraction; GB, Gold Bar; LLB, Lac La Biche; PEC, predicted environmental concentration; POCIS, polar organic chemical integrative sampler; SD, standard deviation; TWA, time-weighted average; WWTP, wastewater treatment plant. * Corresponding author. Richardson College for the Environment, University of Winnipeg, Winnipeg, Manitoba, R3B 2E9, Canada. E-mail addresses:
[email protected] (S.L. MacLeod),
[email protected] (C.S. Wong). 1 Department of Chemistry, Universite´ de Montre´al, Montre´al, Quebec, H3C 3J7, Canada. 0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2009.09.056
534
water research 44 (2010) 533–544
and assign risk. Exposure, in turn, can be determined by either prediction or by measurement of wastewater or surface water concentrations. Prescription and sales information are not readily available, and may not be ideal for predicting environmental concentrations of some drugs (Coetsier et al., 2009) because of the uncertainty associated with use and metabolism (Letzel et al., 2009). In addition, removal of drugs from wastewater may range from 0 to 100% depending on the drug and the treatment process (Batt et al., 2007; Letzel et al., 2009, Lishman et al., 2006), even for the same compound such as diclofenac (Letzel et al., 2009), further complicating prediction of environmental concentrations. In Europe, where a tiered approach is taken for requirements of fate and effects studies, predicted environmental concentrations (PECs) are calculated as a worst-case scenario, without accounting for any metabolism, biodegradation, or retention by treatment (EMEA, 2006). For PEC values above 10 ng/L, higher tiered assessment is required (EMEA, 2006), and while in a French study, PECs were accurate for some drugs (carbamazepine, diclofenac, propranolol), local measured environmental concentrations of others were found to be much lower than those predicted, even very close (10 m) to effluent outfalls (Coetsier et al., 2009). Wastewater and surface water concentrations may be heavily affected by abiotic transformation processes such as photolysis, which depends on light intensity over the seasons (Vieno et al., 2005); as well as biotransformation, which would depend on microbial activity and on temperature. The latter is particularly significant for the numerous chiral drugs. As with other chiral pollutants (Wong, 2006), the enantiomers of pharmaceuticals are affected identically by abiotic processes, but may be affected differently by biologically mediated processes in wastewater treatment (Fono and Sedlak, 2005; MacLeod, 2009; MacLeod et al., 2007b; Nikolai et al., 2006). Drug enantiomers may also exhibit differential toxicity to aquatic life (Stanley et al., 2006). Thus, a robust means by which to predict environmental loadings and fate of drugs from wastewater effluents is valuable for exposure and risk assessment. Direct measurement of drug loadings from wastewater treatment plants (WWTPs) has proven accurate for prediction of drug concentrations in receiving waters (Letzel et al., 2009; MacLeod, 2009). Such measurements can be accomplished via repeated grab sampling (Sacher et al., 2008), which is tedious and time-consuming, or continuous flow-proportional effluent collection (Letzel et al., 2009). However, passive sampling devices such as the Polar Organic Chemical Integrative Sampler (POCIS) (Alvarez et al., 2004) can provide a simple means of obtaining time-weighted average (TWA) loadings of polar contaminants, such as drugs, over longer time periods with lower detection limits (Jones-Lepp et al., 2004; MacLeod et al., 2007a; Zhang et al., 2008). While a high-frequency, discrete sampling approach may capture the real-time heterogeneity of contaminant loadings, passive samplers have the advantage of providing integrated, continuous sample collection, and are thus useful as an alternative or complementary sampling approach for dissolved phase chemicals. Most research on drugs in the environment has, thus far, focused on measurement of drugs in the waste or receiving waters from larger (10,000 to millions) population centers (Carballa et al., 2007; Chen et al., 2006; Hua et al., 2006; Kimura et al., 2007; Lajeunesse et al., 2008; Letzel et al., 2009; Lissemore
et al., 2006; Loraine and Pettigrove, 2006; Managaki et al., 2007; Sacher et al., 2008; Vieno et al., 2005), with little attention paid to the drug output from the wastewater of smaller (<10,000) communities (Batt et al., 2007; Lishman et al., 2006; MacLeod et al., 2007a; Vasskog et al., 2008; Vieno et al., 2005). Thus, the risk of exposure to pharmaceuticals in receiving waters of smaller communities is unclear. In this study, we used POCIS to assess pharmaceutical fate via TWA loadings and temporal trends of achiral and chiral pharmaceuticals at three WWTPs (Fig. 1). The first plant serves Lac La Biche, a rural community of 4000 (Statistics Canada, 2009) in northeastern Alberta, Canada. As of 1999, 84% of Canadian inland municipalities use secondary or tertiary treatment (Environment Canada, 2009). Treatment effectively equivalent to secondary treatment is used in the Lac La Biche WWTP (LLB) through long retention (90 days) in waste stabilization ponds or aerated lagoons (Duigou, 2006). We compared winter drug output from the LLB WWTP with that from two urban WWTPs in central Alberta, Canada, with tertiary UV treatment: Edmonton Gold Bar (GB), serving 750,000 (City of Edmonton, 2009) and Capital Region (CR), serving 250,000 (Alberta Capital Region Wastewater Commission, 2009) in parts of Edmonton and surrounding
Fig. 1 – Map showing study sites in central and northeastern Alberta, Canada. GB, Gold Bar WWTP in Edmonton; CR, Capital Region WWTP in Fort Saskatchewan; LLB, Lac La Biche WWTP in Lac La Biche. Figure prepared by Charlene Nielsen, Geographic Information Systems, Department of Biology, University of Alberta.
water research 44 (2010) 533–544
communities. Treated effluent from LLB is discharged continuously to Field Lake which discharges into Lac La Biche (the lake) via Red Deer Brook, while GB and CR both discharge treated effluent to the North Saskatchewan River and we have previously measured drugs in both of these receiving waters (MacLeod, 2009; MacLeod et al., 2007a). To our knowledge, this is the first study to use POCIS to undertake long-term, nearly continuous, monitoring of drugs, particularly chiral drugs, in municipal effluents from a small population center.
2.
Materials and methods
2.1.
Materials
Table 1 – Pooled winter time-weighted average concentrations for drugs detected at all three WWTPs in each of the three deployment periods.
Atenolol Carbamazepine Celecoxib Citalopram Clarithromycin Codeine Diclofenac Erythromycin Gemfibrozil Metoprolol Naproxen Paroxetine Propranolol Sotalol Temazepam Triclosan Trimethoprim Sum of drugs (mg/L)
Gold Bar (ng/L) 72 200 17 23 220 450 1200 13 150 58 66 0.7 4.7 27 130 1.5 38 2.2
SD 270 130 8 5 140 380 680 10 59 47 58 0.9 3.0 17 71 2.3 39
Capital Region (ng/L) 120 290 28 67 420 920 2,000 19 190 65 530 1.4 14 66 160 53 65 5.0
SD 440 190 13 8 270 710 1300 15 72 52 510 1.8 8.5 22 76 69 70
Lac la Biche (ng/L) 36 63 22 5.7 120 990 460 3.1 72 29 280 0.2 1.4 16 120 6.9 10 2.2
SD 160 39 10 1.7 90 710 280 2.8 34 23 250 0.2 1.1 11 60 8.7 12
Pooled values calculated from n ¼ 6 for Gold Bar and Capital Region WWTPs, n ¼ 9 for Lac la Biche WWTP.
Passive sampling in WWTP effluent
Sampling of WWTP effluent took place between July 2007 and April 2008 (Table S1). During each sampling period at the CR and LLB WWTPs, two and three POCIS, respectively, were protected in perforated stainless steel cages secured with nylon rope, and deployed in final effluent. During each sampling period at GB WWTP, two caged samplers were deployed in a tank into which treated effluent was continuously pumped. Field blanks were brought to all sites at deployment and retrieval, at which time POCIS were individually put on ice until return to the laboratory, where they were stored at 20 C until chemical extraction.
2.3.
We used POCIS (Environmental Sampling Technologies, St. Joseph, MO) in the ‘pharmaceutical’ configuration: 200 mg Oasis HLB sorbent (Waters, Milford, MA) between two polyethersulfone membranes (45.8 cm2 total surface area) held together by stainless steel rings and attached to a spindle holder. This configuration was used as POCIS sampling rates are available for the target analytes (Tables 1 and S2), all commonly used drugs. Chemical analytes (>98% purity) and used as received (sources in Table S2). Solvents and reagents for chromatography (Fisher Scientific, Ottawa, Canada) were of HPLC grade; nanopure water (18 MU cm) was supplied by a Nanopure Ultrapure system (Barnsteam/Thermolyne, Dubuque, IA). Details are published elsewhere (MacLeod, 2009; MacLeod et al., 2007a, b).
Analyte
2.2.
535
Chemical extraction and instrumental analysis
Details on chemical extraction and instrumental analysis are provided elsewhere (MacLeod, 2009; MacLeod et al., 2007a, b) and in Supplemental Information. In brief, POCIS sorbent was washed into a glass column, and methanol used to elute target analytes. The extract was filtered, reduced to 5–10 mL via rotary evaporation, filtered through 0.22 mm Acrodisc polytetrafluoroethylene syringe filters (Pall Life Sciences, Ann Arbor, MI, USA), and nitrogen-evaporated to dryness. Internal standards (Table S2) were added, and extracts were reconstituted to 1 mL with methanol. Achiral chromatography for determining concentrations (MacLeod et al., 2007a) was performed with an Ultra C18 reversed-phase column (150 mm 4.6 mm internal diameter (i.d.) 5 mm particle size (dp), Restek, Bellefonte, PA, USA), while enantioselective chromatography for determining enantiomer compositions was performed with a Chirobiotic V column (250 mm 4.6 mm i.d. 5 mm dp, Advanced Separation Technologies, Whippany, NJ, USA) for all chiral analytes except temazepam, for which a Chiralpak AD-RH column (MacLeod, 2009) was used (150 mm 4.6 mm i.d. 5 mm dp, Daicel Chemical Industries, West Chester, PA). Analytes were detected by electrospray ionization-tandem mass spectrometry using an Applied Biosystems QTrap triple-quadruple instrument (Foster City, CA) in multiple reaction monitoring mode.
2.4. Quality assurance/quality control and data handling Detection limits (DLs, Table S2) were determined as the analyte concentration in 1 mL POCIS extract providing a signal-to-noise ratio of 3, and ranged from 0.001 to 1.6 ng/ POCIS. Drugs were not detected in any blanks. Calculated time-weighted average concentrations and loadings (hereafter called ‘‘concentrations’’ and ‘‘loadings’’, respectively) are presented as mean standard deviation (SD). Data analysis was carried out as previously described elsewhere (MacLeod, 2009; MacLeod et al., 2007a, b). Statistical comparisons between two groups were carried out by t-test, while three or more groups were compared by ANOVA, with either Tukey’s test for multiple comparisons or with Dunnett’s test for comparison with a control, as indicated. Differences noted are statistically significant (all p < 0.05). As expected, analytes for which the POCIS sampling rate was known with higher
536
water research 44 (2010) 533–544
precision (Table S2) were generally more conducive to statistical significance testing. For chiral drugs atenolol, citalopram, fluoxetine, metoprolol, nadolol, pindolol, propranolol, salbutamol, sotalol, and temazepam, enantiomer fractions (EFs) described enantiomer composition (Harner et al., 2000): EF ¼
E1 ðþÞ ¼ E1 þ E2 ð þ Þ þ ð Þ
(1)
where E1 and E2 are the first and second eluted enantiomers, respectively, when the enantiomer optical rotation was unknown (i.e., all analytes except atenolol, propranolol, and fluoxetine). Further information on data quality and handling is available in Supplementary Data.
3.
Results and discussion
3.1. Time-weighted average pharmaceutical concentrations in treated effluents 3.1.1.
Aerated lagoons: Lac la Biche
Most studies to date have relied on grab or short-term composite sampling to assess temporal trends in pharmaceutical concentrations in effluents (Batt et al., 2007; Chen et al, 2006; Hua et al., 2006; Loraine and Pettigrove 2006, Stu¨lten et al., 2008; Vieno et al., 2005). However, temporal trends assumed from grab sampling may not be representative of longer time periods. Because POCIS can provide longer and continuous temporal integration, it may provide more accurate assessment of temporal trends. However, passive samplers cannot provide reliable information on concentration fluctuations over time scales shorter than sample integrative periods, which can result in very different conclusions about temporal changes in concentration over time (Shaw and Mueller, 2009). Drug concentrations at LLB (Fig. 2) measured over six sampling periods of 35–51 days (Table S1) ranged from 0.07 ng/L (paroxetine, propranolol) to 1100 ng/L (codeine). Fenoprofen, fluoxetine, ketoprofen, omeprazole, roxithromycin, sildenafil, sulfadimethoxine, sulfamethazine, sulfapyridine, sulfisoxazole, tadalafil and vardenafil were not detected in LLB treated effluent above their respective DLs in any sampling period (Table S2). Concentrations of atenolol and propranolol were similar to effluent grab sampled from LLB in September 2005 (Nikolai et al., 2006); however, metoprolol was higher (200 ng/L). Diclofenac and naproxen concentrations at LLB were also similar to that in effluents of the activated sludge WWTP of Aura, Finland, which has a similar population (4000) to LLB (Vieno et al., 2005). Concentrations were also generally similar to those in an aerated lagoon WWTP in Louisiana, USA (Conkle et al., 2008) for drugs common to both studies: atenolol, carbamazepine, gemfibrozil, metoprolol, naproxen, and sotalol. Thus, our TWA concentrations are likely representative of those expected in small community WWTP effluents. There were statistically significant temporal concentration changes (Fig. 2) for celecoxib, citalopram, clarithromycin, codeine, diclofenac, erythromycin, gemfibrozil, naproxen, propranolol, sotalol, and temazepam. Temporal concentration trends were similar for clarithromycin, codeine, diclofenac,
erythromycin, naproxen and propranolol, with higher concentrations during winter sampling periods (December to April) than at other times (July to December, Fig. 2). For atenolol, carbamazepine, metoprolol, triclosan and trimethoprim, the effluent concentrations did not change significantly over the sampling periods (Fig. 2). For drugs with temporal differences in concentration, the changes in concentration ranged from 2 (propranolol) to 960 ng/L (codeine), whereas drugs without statistically significant temporal differences had smaller concentration changes over time, from 8 (triclosan) to 85 ng/L (atenolol). While the average daily volume of treated effluent discharged decreased over the six sampling periods (Table S1), this decrease was not statistically significant. Thus, TWA loadings are reflective of concentrations. Of the drugs studied, photodegradation likely plays a major role in the dissipation of diclofenac (Andreozzi et al., 2003), naproxen (Lin et al., 2006; Lin and Reinhard, 2005), and propranolol (Andreozzi et al., 2003) in aquatic systems. The higher effluent concentrations of those drugs in winter (Fig. 2) may be at least partially the result of decreased photolysis, due to ice cover on lagoons from November to March (Siebold, 2008) blocking sunlight from the water column (Vieno et al., 2005), and fewer hours of sunlight during those months (MacLeod, 2009; Vieno et al., 2005). It is impossible to ascertain the amount of photodegradation in the lagoon during ice-free periods with our data. While the LLB lagoon wastewater was not clear in color, it is likely that photodegradation of drugs took place at least at the lagoon surface during ice-free periods, and continuous aeration of lagoon waters for oxygenation purposes would bring water at depth to the surface to provide some photodegradation treatment. Lower winter temperatures also likely influenced elimination of pharmaceuticals, through lower reaction rate constants for abiotic transformation, and reduced microbial activity for biotransformation (Kim and Carlson, 2007). Effluent pH was 8 0.5 over the six sampling periods and effluent temperatures ( C) were 21 3, 16 2, 4 3, 1 0, 1.4 0.5, 4 2, chronologically for the six sampling periods (Siebold, 2008). The concentrations of those three drugs decreased again in March/April, as temperatures and number of daylight hours increased, and ice cover was reduced or absent on the lagoons (Fig. 2). Since naproxen can be readily biodegraded in sewage (Yu et al., 2006), decreased biodegradation may also contribute to the higher effluent concentrations for naproxen in winter. It is possible that changes in temperature may affect calculated TWA concentrations, as sampling rates used in this study were determined at room temperature (22–28 C, MacLeod et al., 2007a), and WWTP temperatures during colder months were considerably less. However, we cannot presently quantify the compound-dependent effect of such changes on our observed trends. Lower temperatures would result in lower analyte aqueous diffusion coefficients and hence lower sampling rates, estimated to be roughly 75% over a 20 C ambient temperature change (MacLeod et al., 2007a). When applied to water colder than under the calibration condition, sampling rates would then provide an underestimate of the calculated TWA concentrations, the effect of which would be compound-dependent. Such an effect may be pronounced in the case of drugs with higher
water research 44 (2010) 533–544
537
Fig. 2 – Time-weighted average concentrations (ng/L ± SD; ) and enantiomer fractions (EF) of chiral drugs (,) in LLB treated wastewater during six sampling periods from July 2007 to August 2008 (Table S1). Hash marks on x-axis indicate break in continuous sampling between September and October 2008. Significant concentration differences are indicated by lowercase letters while EF differences are indicated by upper-case letters. Plots without letters do not have concentrations or EFs, as appropriate, statistically different from one another over sampling periods. Non-racemic EFs are indicated by an asterisk, and EFs that were corrected for enantiomer-specific matrix effects are indicated by ^. RS, racemic standards. Error bars (SD) represent propagated error from replicate POCIS, analytical processing and instrumental uncertainty, and error associated with POCIS sampling rates.
concentrations in winter than in other times (e.g., antibiotics such as clarithromycin and erythromycin, Fig. 2). Temporal effects on concentrations were less for other analytes, and may either be insignificant (e.g., triclosan and gemfibrozil) or of little practical significance even if statistically significant, given the relatively narrow range (e.g., from a factor of two, to an order of magnitude) in concentrations over the seasons studied (Fig. 2).
Without measurements of influent concentration, we cannot discern whether temporal concentrations in effluent are more influenced by changes in removal efficiency, as suggested as likely by many studies (Hua et al., 2006; Kim and Carlson, 2007; Loraine and Pettigrove, 2006; Sacher et al., 2008; Vieno et al., 2005) or in use patterns over time, which may be the case for antibiotics. In a Swiss study (McArdell et al., 2003), higher macrolide concentrations were found in winter than in
538
water research 44 (2010) 533–544
other periods, consistent with our observations for the macrolides clarithromycin and erythromycin (Fig. 2). In the Swiss study, as with this study, raw wastewater concentrations were not measured so removal efficiency could not be assessed. However, the Swiss authors cited a personal communication which indicated that macrolide sales were higher in winter, indicating that increased use may contribute to higher environmental concentrations. The winter concentrations in LLB treated effluent, which were sampled mostly at the same time as those at GB and CR (see Section 3.1.3), changed over time for some but not all of our target analytes (Fig. 2). Concentration changes were less prevalent among the three winter sampling months, compared with other periods. Among the sampling periods between December and April (Tukey’s test), there were statistically significant changes in effluent concentration for only three drugs. The celecoxib concentration in December/ January was three times that in January/March and twice that in March/April with magnitudes of difference of 20 and 15 ng/L, respectively. Sotalol had December/January concentrations were eight times that in both January/March and March/April and magnitudes of 33 ng/L for both comparisons. While the concentration of citalopram in January/March was 50% higher than that in March/April, the magnitude of this difference was only 2 ng/L. The general lack of effluent concentration change over the winter indicates that there is no change in drug use patterns and/or in the ability of aerated lagoons to remove drugs effectively over this season. It is important to note that concentration changes in WWTPs serving small communities may well be sensitive to changes in use patterns of the population served. If a drug was completely excreted from the body in unchanged form and did not undergo any removal during wastewater treatment, then the addition of a single additional individual’s twice-aday, 100 mg active ingredient dose at LLB would result in an effluent concentration change of 100 ng/L, which would rival existing concentrations at LLB (Fig. 2)! In actuality, only a fraction of most drugs are excreted as parent compounds, and WWTP treatment provides some elimination. For example, 3% of carbamazepine is excreted unmetabolized from humans, with 30% removal based on a survey of Canadian WWTPs (Miao et al., 2005). Assuming these conditions hold for LLB, then a 2 ng/L change in effluent concentration, or a 3–10% increase (Fig. 2), would result from the same dosage. However, without specific usage information, excretion data, and WWTP-specific removal efficiencies, we cannot make more specific predictions of the sensitivity of drug concentrations in wastewater on usage in small communities.
3.1.2.
Tertiary treatment: Gold Bar and Capital Region
The concentration of detectable drugs (Table 1, Table S3) in GB effluent ranged from 0.03 (omeprazole) to 1350 ng/L (diclofenac) while that in CR ranged from 0.09 (roxithromycin) to 2520 ng/L (diclofenac). These concentrations are similar to grab sample measurements for our analytes in other Canadian WWTP effluents (Hua et al., 2006; Lishman et al., 2006; Miao et al., 2005), indicating that our passive sampler derived TWA concentrations were comparable with previous work. However, concentrations of chiral drugs in this study were generally lower by a factor of at least 10 compared with a grab
sample from GB in April 2007 (Macleod et al., 2007b), suggesting possible annual or other temporal changes in concentrations as discussed below. Since only two samplers were deployed in each of GB and CR during each sampling period (Table S1, Table S3), temporal trends could not be statistically assessed. However, few temporal trends were evident. In GB effluent, the temazepam concentration in March was double (range >100 ng/L) that in December/January, and in CR effluent, the naproxen concentration increased seven-fold over the winter (range >700 ng/L). Also in CR effluent, the roxithromycin concentration in December/January was nine times higher than that in both February/March and March (range 0.8 ng/L), whereas the opposite trend was seen for sulfamethazine (increased three times over time, range 3 ng/L). Fenoprofen, ketoprofen, sulfadimethoxine, sulfisoxazole, and tadalafil were not detected in either urban WWTP above their respective DLs. When the TWA concentrations for each drug were pooled for each WWTP (n ¼ 6 for each WWTP), statistical comparisons revealed no significant differences between urban effluent concentrations for any drug (Table 1). As with the aerated lagoon effluent, there was little evidence of changes in drug concentration over the monthly integrated sampling periods. Seasonality in removal efficiency and/or or drug use patterns could not be assessed for the tertiary plants as only winter data was collected. The concentrations of each drug were similar between the two tertiary plants (Table 1), likely a reflection of similar per capita drug and water use between the populations since treatment processes are nearly identical in the two plants (Letzel et al., 2009). We do not expect concentrations in these large-community WWTPs to be influenced significantly by use changes in a small number of individuals; using the same assumptions as in the previous section, the same individual’s carbamazepine dose change would result in a 0.02 ng/L increase in GB concentrations, an insignificant amount.
3.1.3. Comparison of concentrations in aerated lagoon and tertiary treated effluents in winter Drug concentrations in effluents depend on a number of factors. Removal efficiency during treatment is known to vary by drug and by WWTP (Batt et al., 2007; Letzel et al., 2009). Differences in sewage retention time (Batt et al., 2007; Go¨bel et al., 2007; Kimura et al., 2007), and/or the co-metabolic activity of microbes (Joss et al., 2006; Kim and Carlson, 2007; Kimura et al., 2007; Yu et al., 2006) can produce differences in the efficacy of wastewater treatment for removal of some drugs. For other chemicals, sorption capacity, rather than retention time or biodegradation, can be most important for removal during treatment (Carballa et al., 2007; Joss et al., 2006). Differences in sunlight attenuation while wastewater is in outdoor tanks (tertiary) and lagoons may also play a role for some drugs, as the turbidity of the wastewater may vary, thus varying the amount of sunlight attenuation and thus the amount available for photolysis. Drugs in GB and CR were exposed to sunlight for less than 24 h, and received at least 8 s of low power UV exposure (City of Edmonton, 2009) for disinfection of effluent before release. The drugs in LLB are exposed to sunlight for as long as the daylight hours of 90 days (Duigou, 2006), providing more time for photolysis.
water research 44 (2010) 533–544
Temperature can also play a role, as previously noted. In addition, both GB and CR use secondary activated sludge treatment. These tanks are warmed in the winter to maintain efficiency, especially in cold climates such as that in Alberta. Thus, the water temperatures in those tanks may be higher than that in the LLB lagoon pond under ice (w4 C). As a result, biological activity capable of biodegrading drugs may be more active in GB and CR compared with LLB. Notwithstanding these potential differences in incidental removal during treatment, the concentrations of several drugs detected at all three plants were generally comparable, with CR having the highest total concentration (Table 1). Most drugs (17 out of 24 detectable) were found in all three WWTPs in winter (Table 1). No differences in winter pooled TWA concentrations (three winter sampling periods with n ¼ 9 for LLB, and n ¼ 6 for GB and for CR) among the three effluents for atenolol, celecoxib, codeine, metoprolol, naproxen, paroxetine, temazepam, triclosan and trimethoprim. For eight other drugs, the concentrations in urban effluents were from 2 to 12 times higher than that in rural effluent, with concentration differences that ranged from 9 to 1600 ng/L (Table 1). For citalopram and gemfibrozil, the concentrations at GB were greater than those at LLB. For carbamazepine, citalopram, clarithromycin, diclofenac, erythromycin, gemfibrozil, propranolol, and sotalol, the concentrations in CR effluent were greater than those in LLB effluent. Between the two urban WWTPs, the concentrations of citalopram, propranolol and sotalol at CR were greater than those at GB. Several drugs (fluoxetine, paroxetine, roxithromycin, sildenafil, sulfamethazine, sulfapyridine, vardenafil), infrequently or not detected in the rural WWTP effluent, were consistently detected in the urban WWTP effluents (Tables S2–S3). These drugs were generally present at low concentrations (<4 ng/L) in urban WWTP effluents, with the notable exception of sulfapyridine (>100 ng/L). The discrepancy with sulfapyridine may be explained by differences in drug use among the communities, but is more likely caused by treatment differences. Sulfapyridine is generally administered via sulfasalazine, which is cleaved in the colon to release sulfapyridine (10–35%), while another 30–50% is released as sulfasalazine and a metabolite (Neumann, 1989), both of which may be cleaved in biological wastewater treatment to release sulfapyridine (Go¨bel et al., 2005). It is possible that such cleavage does not occur in the aerated lagoons, leading to a lower concentration of sulfapyridine in LLB effluent. For the other drugs, differences reflect differences between the communities in terms of water and/or drug use, and/or a difference in the fate of drugs in the WWTPs (Hua et al., 2006; Kim and Carlson, 2007; Loraine and Pettigrove, 2006; Sacher et al., 2008; Vieno et al., 2005).
3.2. Loadings of pharmaceuticals to surface waters from aerated lagoon and tertiary treated effluents Average daily loadings (Table S4) in winter were calculated for each plant by multiplying the average daily water discharge by the TWA concentration of each drug during each sampling period. Total winter loadings for GB, CR, and LLB were 700, 400 and 4 g/day, respectively. For the winter sampling periods, average daily loadings from GB were generally twice that from CR. Average daily loadings from LLB were on the order
539
of mg/day. This was at least an order of magnitude lower than loadings from GB and CR, as expected based on the populations served (Letzel et al., 2009). The above result suggests that population-weighted per capita TWA daily loadings would be similar among the three WWTPs. Indeed, this was the case for most drugs, for which there were no statistically significant differences among population-weighted average daily loadings (Fig. 3), e.g., atenolol, celecoxib, clarithromycin, codeine, gemfibrozil, metoprolol, naproxen, temazepam, triclosan, and trimethoprim. Urban per capita average daily loadings were greater than those from the rural WWTP in a few cases: CR was greater than LLB (carbamazepine, citalopram, diclofenac, propranolol, erythromycin, sotalol), GB was greater than LLB (citalopram), while the per capita loading from CR was greater than GB for three drugs (citalopram, propranolol, sotalol). These per capita loading differences, although statistically significant, may not be practically significant, as all were all less than an order of magnitude (two to eight times) and the absolute differences ranged from 2 (propranolol, CR > GB) to 390 mg/person per day (diclofenac, CR > LLB). All of these differences in loadings between WWTPs were also noted as differences in concentration. Loading differences were not found for clarithromycin and gemfibrozil, for which differences in concentration were found among WWTPs. Thus, we conclude that population is a good predictor of human-use drug release into receiving waters, at least for over-the-counter or long-term prescription drugs such as those in this study. Such predictions are likely to be more accurate when based on direct measurements, particularly from demographically similar regions (Stamm et al., 2008), and most accurate when also based on comparable wastewater treatment. In a ten-year study of drugs in the Rhine, concentrations of diclofenac and carbamazepine were found to be relatively constant and given consistency in use patterns, and although raw wastewater concentrations were not measured, the authors attributed any temporal variations to differences in WWTP removal efficiency (Sacher et al., 2008). Their conclusion was that any measures that had been implemented to reduce loadings were thus far ineffective. Because our estimates are based on continuous measurements of pharmaceutical concentrations (Letzel et al., 2009; Miao et al., 2005; Stamm et al., 2008), rather than instantaneous grab sampling, they are likely to be more robust as TWA concentrations from passive samplers are less susceptible to short-term fluctuations in concentration over periods shorter than the sampler deployment periods. Efforts have been undertaken to find a suitable mechanism for removal of drugs from the waste stream (Ikehata et al., 2008), and such efforts should continue, because as our results further highlight, even sophisticated WWTPs like GB and CR produce wastewater with drug concentrations that are comparable with water treated by aerated lagoons. As previously noted, loadings at WWTPs serving small communities are potentially more sensitive to small changes in use patterns than larger communities. The twice-a-day, 100 mg/dose from one individual at LLB would result in a loading increase of 50 mg/person per day at 100% excretion and no WWTP removal, or 1 mg/person per day (4% increase, Fig. 3) using the carbamazepine assumptions previously given. At GB, the equivalent increase in loadings would
540
water research 44 (2010) 533–544
Fig. 3 – Time-weighted average per capita daily loadings (:, mg/person per day) and enantiomer fractions (,, EF), both ± SD, of drugs at three WWTPs: GB, Gold Bar; CR, Capital Region (n [ 6), LLB, Lac La Biche. Differences in daily loadings are indicated by lower-case letters while EF Statistically significant differences are indicated by upper-case letters. Plots without letters do not have loadings or EFs, statistically different from one another over sampling periods. Asterisks indicate EFs that are significantly non-racemic compared with racemic standards. Error bars (SD) represent propagated error from replicate POCIS, analytical processing and instrumental uncertainty, and error associated with POCIS sampling rates.
water research 44 (2010) 533–544
generally be insignificant (Fig. 3), at 0.27 and 0.006 mg/person per day, respectively.
3.3. Enantiomer fractions of chiral drugs in treated effluents 3.3.1. Temporal trends in enantiomer fractions of chiral drugs in aerated lagoon effluent We measured EFs in LLB to discern whether temporal changes existed in enantiomer composition of chiral drugs released by WWTPs. Changes in EF may arise from differential interaction of chiral drug enantiomers with other chiral chemicals (e.g., enzymes). These may result from temporal differences in enantiomer-specific metabolism of patients (Mehvar and Brocks, 2001; Schmekel et al., 1999), and/or from enantiomerspecific, microbial enzyme mediated biotransformations during wastewater treatment, as previously noted for atenolol (MacLeod et al., 2007b; Nikolai et al., 2006), citalopram (MacLeod et al., 2007b) and propranolol (Fono and Sedlak, 2005; MacLeod et al., 2007b; Nikolai et al., 2006), but not metoprolol (Fono and Sedlak, 2005; MacLeod et al., 2007b; Nikolai et al., 2006), salbutamol (MacLeod et al., 2007b), or sotalol (MacLeod et al., 2007b). Because of enantiomer-specific metabolism by the target organism, chiral pharmaceuticals may not necessarily enter the environment with the same enantiomeric composition as the dose. This is in contrast to chiral persistent organic pollutants, which generally enter the environment as a racemate, or sometimes as a single enantiomer. Thus, changes in enantiomeric composition from the excreted EF, not just from a racemic value of 0.5, would be environmentally significant as such a change would indicate post-release biochemical weathering. However, given that few environmental measurements of drug EFs exist to date, our discussion here focuses on changes in EF from the racemate, the most likely enantiomeric composition of formulated chiral drugs not sold as single enantiomers (e.g., naproxen). With the exception of tempazepam, chiral drugs were generally non-racemic in LLB effluent (Fig. 2) compared with racemic standards (Dunnett’s test). Atenolol was enriched in ()-atenolol during some sampling periods, with changes over time, consistent with previous observations in 2005 at LLB (Nikolai et al., 2006). Citalopram was mainly enriched in the first-eluted enantiomer except during August/September. Metoprolol was enriched in the E2-enantiomer when nonracemic, which was not previously observed at LLB in individual grab samples (Nikolai et al., 2006). Salbutamol was always enriched in the second eluted enantiomer, except during January/March wherein it was E1-salbutamol that was enriched. Sotalol was always non-racemic, with no change in EF over time. Temazepam was racemic with one exception, wherein E2-temazepam was enriched. Propranolol EFs could not be measured (supplemental data). Temporal changes in EF were observed (Tukey’s test) for all drugs except sotalol in LLB effluent (Fig. 2, Table S5). Without direct measurement of the EF in influent and during the treatment process, we cannot conclusively determine whether EFs in effluent arose from metabolism prior to bodily excretion or from processes during treatment. The latter may be plausible, given changes in activity and composition of microbial consortia during biological sewage treatment from
541
temperature changes over the seasons. For drugs such as citalopram and salbutamol, changes in EF may also reflect differences in availability or use of single-enantiomer formulations (Maier et al., 2001). Because of the temporal changes in EF observed, our results show that an EF measured at one time point may not necessarily be characteristic of the enantiomer composition of chiral drugs discharged from a WWTP, and hence insufficient to characterize exposures in receiving waters.
3.3.2. Comparison of chiral drug EFs from aerated lagoons and tertiary treated wastewater Winter EFs from LLB and from the urban tertiary WWTPs were compared to gain insight on whether EFs could be predicted by treatment type, and may possibly be caused by similar human metabolism and release and/or microbial degradation during WWTP treatment. Due to low sample sizes, statistical tests and trend analyses were not carried out on the temporal EF data from the urban tertiary WWTPs. Instead, winter EFs were pooled for each drug from GB and CR and compared with EFstandard (Dunnett’s test), to each other (t-test, with Welch’s correction for unequal variances as needed), and to the pooled EFs from the three winter sampling periods at LLB (Tukey’s test). Standard deviations in pooled data reflect propagated variability from both temporal changes in EFsample and from replicates. The two tertiary WWTP effluents had similar EFs during winter for all chiral drugs except sotalol (Fig. 3), suggesting that both treatment plants were processing those drugs in the same stereoselective manner during microbial treatment. At all three WWTPs (Fig. 3), metoprolol EFs were racemic in effluents, while sotalol was enriched in the E2-enantiomer and differed in EF between the tertiary effluents. Racemic metoprolol was previously observed in both GB and LLB effluents (MacLeod, 2009; MacLeod et al., 2007b; Nikolai et al., 2006) and in the effluent of a WWTP in Dallas TX (Fono et al., 2006), with no change in EF from raw to treated wastewater (MacLeod et al., 2007b). After correction for enantiomer-specific matrix effects (MacLeod et al., 2007b), sotalol was previously found to be racemic in GB effluent (MacLeod et al., 2007b) via grab samples. This observation suggests that changes in sotalol EFs might have occurred over time in GB effluent; however, we cannot unequivocally conclude that this must be the case given our earlier caveats that a single EF cannot necessarily characterize the enantiomer composition. No statistically significant differences were observed for salbutamol EFs in the three effluents, despite enrichment of E2-salbutamol in both tertiary effluents. This observation is largely a result of the high variability in salbutamol EFs in LLB and CR. In turn, the variability may be a function of the drug’s input to WWTPs, given that salbutamol is available as an R-salbutamol-only formulation, while S-salbutamol is metabolized slower than its antipode (Schmekel et al., 1999). Citalopram (Fig. 3) was enriched in the E1-enantiomer in all effluents, with differences in EF between tertiary and aerated lagoon output. As with salbutamol, this difference could be influenced by differences in prescription patterns, given the availability of escitalopram, the S-enantiomer formulation. Citalopram was previously found with similar EFs (0.55–0.65) in GB effluent (MacLeod et al., 2007b). Temazepam was enriched in the E2-enantiomer at both GB and LLB, but racemic in CR,
542
water research 44 (2010) 533–544
wherein its EF was different from that in LLB, but not from that in GB. No information regarding temazepam EFs in treated effluent is available, except our previous results (MacLeod, 2009). For atenolol, all effluents had EFs that were different from one another, with GB containing racemic atenolol, CR containing more (þ)-atenolol and LLB containing more ()-atenolol. The EF of atenolol from healthy renal excretion is 0.476 (Mehvar and Brocks, 2001), and 90% of atenolol is excreted unchanged in urine (Bourne, 1981). Thus, a process other than healthy renal excretion, and potentially specific to microbial communities of WWTPs, was influencing atenolol EFs in effluent. Likewise, the differences in EFs observed among WWTP effluents is likely due, at least in part, to variations in enantioselective microbial biotransformation of drugs during wastewater treatment. As we have noted, it is not clear what the enantiomeric composition of drugs may be upon excretion from humans, but non-racemic release is likely. As such, we must delineate between environmental biotransformation and in changes in enantiomeric composition of the source material in assessing the environmental fate of drug enantiomers, as has been done for chiral pesticides (e.g., metolachlor) also released in nonracemic amounts (Kurt-Karakus et al., 2008).
4.
Conclusions
Although other studies have used POCIS to monitor drugs in treated wastewater, none have done so for more than 30 days (Jones-Lepp et al., 2004; Zhang et al., 2008). Herein, we monitored LLB effluent for 271 days and found that the concentration of drugs in effluent varied over seasonal time scales. Other studies have mainly focused on larger urban centers. By comparing drugs in both rural and urban wastewater effluents, we find that pharmaceutical loadings to aquatic ecosystems scale with population. However, we also found that the concentrations of drugs in effluent from WWTPs with biological activated sludge treatment and UV disinfection were comparable with concentrations in effluent from simpler treatment methods like aerated lagoons. Although absolute loadings were generally lower, drugs were still detectable in effluent from small communities. Thus, if pharmaceutical contamination of surface water presents a risk to aquatic life, that risk is also present near smaller centers, especially those which are landlocked and not discharging to a large surface water body to provide effluent dilution. Overall, temporal changes in EF were found for all chiral drugs except sotalol, with temporal changes in concentration for all chiral drugs except atenolol and metoprolol. The concentration of atenolol and metoprolol in LLB effluent did not change over time, but the EFs of these drugs did change. Conversely, although the concentration of sotalol in LLB effluent changed over time, the EF remained constant. Although temporal changes in concentration may not be occurring, the relative proportions of enantiomers being released to the environment may be changing over time. Also, for some drugs the EF may remain constant, but the concentration of the drug may change. Similar conclusions can be drawn from the comparison between EFs and per capita daily loadings among the three
WWTPs. Loading and EF differences were found among WWTPs for citalopram and sotalol, while for atenolol and temazepam, there were no differences in loading but there were changes in EF, and for metoprolol there were no differences in loading or EF among the three WWTPs. Simply put, although concentrations and loadings may be accurately predicted based on population-weighted measured effluent concentrations in other treatment plants, the EFs of chiral drugs released by one WWTP may not necessarily reflect that released by another, probably from site-specific biochemical activity. Thus, enantiomer-specific monitoring of chiral drugs may be useful, particularly for drugs with enantiomer-specific toxicity to aquatic life (Stanley et al., 2006).
Acknowledgements We thank D. Bleackley, V. Cooper, M. Ross, B. Asher and E. McClure (University of Alberta), R. Litwinow and D. Seehagel (GB WWTP), D. Cikaluk (CR WWTP) and G. Siebold (LLB WWTP) for sampling assistance, and H. Li (Trent University) for providing unpublished POCIS sampling rates for citalopram and sotalol. Funding was provided to C.S.W. from the Canada Research Chairs Program, Canada’s Natural Sciences and Engineering Research Council (NSERC), and the Society of Environmental Toxicology and Chemistry (SETAC) Early Career Award for Applied Ecological Research, cosponsored by the American Chemistry Council; and to S.L.M. as fellowships from NSERC, Alberta Ingenuity Fund, and the American Chemical Society Division of Analytical Chemistry, sponsored by DuPont.
Appendix. Supplementary data Information about sampling locations and times, analytical and data treatment procedures, concentrations, loadings, and EFs are available in the online supplementary data to Supplementary data associated with this article can be found in the online version at doi:10.1016/j.watres.2009.09.056.
references
Alberta Capital Region Wastewater Commission, 2009. http:// www.acrwc.ab.ca accessed February 25. Alvarez, D.A., Petty, J.D., Huckins, J.N., Jones-Lepp, T.L., Getting, D.T., Goddard, J.P., Manahan, S.E., 2004. Development of a passive, in situ, integrative sampler for hydrophilic organic contaminants in aquatic environments. Environ. Toxicol. Chem. 23, 1640–1648. Andreozzi, R., Raffaele, M., Nicklas, P., 2003. Pharmaceuticals in STP effluents and their solar photodegradation in the aquatic environment. Chemosphere 50, 1319–1330. Batt, A.L., Kim, S., Aga, D.S., 2007. Comparison of the occurrence of antibiotics in four full-scale wastewater treatment plants with varying designs and operations. Chemosphere 68, 428–435. Bourne, G.R., 1981. The metabolism of b-adrenoreceptor blocking drugs. In: Bridges, J.W., Chasseaud, L.F. (Eds.), Progress in Drug Metabolism 6. John Wiley & Sons Ltd, pp. 77–110.
water research 44 (2010) 533–544
Carballa, M., Omil, F., Ternes, T., Lema, J.M., 2007. Fate of pharmaceutical and personal care products (PPCPs) during anaerobic digestion of sewage sludge. Water Res. 41, 2139–2150. Chen, M., Ohman, K., Metcalfe, C., Ikonomou, M.G., Amatya, P.L., Wilson, J., 2006. Pharmaceuticals and endocrine disruptors in wastewater treatment effluents and in the water supply system of Calgary, Alberta, Canada. Water Qual. Res. J. Can. 41, 351–364. City of Edmonton, 2009. http://www.edmonton.ca. accessed February 25. Coetsier, C.M., Spinelli, S., Lin, L., Roig, B., Touraud, E., 2009. Discharge of pharmaceutical products (PPs) through a conventional biological sewage treatment plant: MECs vs PECs? Environ. Int 35, 787–792. Conkle, J.L., White, J.R., Metcalfe, C.D., 2008. Reduction of pharmaceutically active compounds by a lagoon wetland wastewater treatment system in Southeast Louisiana. Chemosphere 73, 1741–1748. Daughton, C.G., Ternes, T.A., 1999. Pharmaceuticals and personal care products in the environment: agents of subtle change? Environ. Health. Perspect 107, 907–938. Duigou, E., 2006. Town of Lac La Biche, Personal communication. EMEA, 2006. Note for guidance on environmental risk assessment of medicinal products for human use. The European Agency for the Evaluation of Medicinal Products (EMEA) CMPC/SWP/4447/00. Environment Canada, 2009. http://www.ec.gc.ca/soer-ree/ English/Indicator_series/new_issues.cfm?issue_id¼7&tech_ id¼28#bio_pic. accessed February 27. Fent, K., Weston, A.A., Caminada, D., 2006. Ecotoxicology of human pharmaceuticals. Aquat. Toxicol 76, 122–159. Fono, L.J., Sedlak, D.L., 2005. Use of the chiral pharmaceutical propranolol to identify sewage discharges into surface waters. Environ. Sci. Technol 39, 9244–9252. Fono, L.J., Kolodziej, E.P., Sedlak, D.L., 2006. Attenuation of wastewater-derived contaminants in a effluent-dominated river. Environ. Sci. Technol 40, 7257–7262. Go¨bel, A., McArdell, C.S., Joss, A., Siegrist, H., Giger, W., 2007. Fate of sulfonamides, macrolides, and trimethoprim in different wastewater treatment methodologies. Sci. Total Environ 372, 361–371. 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. Environ. Sci. Technol 39, 3981–3989. Halling-SBrensen, B., Nielsen, S.N., Lanzky, P.F., Ingerslev, F., Holten-Lu¨tzhBft, H.C., JBrgensen, S.E., 1998. Occurrence, fate and effects of pharmaceutical substances in the environment – a review. Chemosphere 36, 357–393. Harner, T., Wiberg, K., Norstrom, R., 2000. Enantiomer fractions are preferred to enantiomer ratios for describing chiral signatures in environmental analysis. Environ. Sci. Technol 34, 218–220. Hua, W.Y., Bennett, E.R., Miao, X.-S., Metcalfe, C.D., Letcher, R.J., 2006. Seasonality effects on pharmaceuticals and s-triazine herbicides in wastewater effluent and surface water from the Canadian side of the Upper Detroit River. Environ. Toxicol. Chem. 25, 2356–2365. Ikehata, K., El-Din, M.G., Snyder, S.A., 2008. Ozonation and advanced oxidation treatment of emerging organic pollutants in water and wastewater. Ozone Sci. Eng 30, 21–26. Jones-Lepp, T.L., Alvarez, D.A., Petty, J.D., Huckins, J.N., 2004. Polar organic chemical integrative sampling and liquid chromatography-electrospray/ion-trap mass spectrometry for assessing selected prescription and illicit drugs in treated sewage effluents. Arch. Environ. Contam. Toxicol 47, 427–439. Joss, A., Zabcynski, 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
543
wastewater treatment: proposing a classification scheme. Water Res. 40, 1686–1696. Kim, S.-C., Carlson, K., 2007. Temporal and spatial trends in the occurrence of human and veterinary antibiotics in aqueous and river sediment matrices. Environ. Sci. Technol 41, 50–57. Kimura, K., Hara, H., Watanabe, Y., 2007. Elimination of selected acidic pharmaceuticals from municipal wastewater by an activated sludge system and membrane bioreactors. Environ. Sci. Technol 41, 3708–3714. 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, 1999–2000: a national reconnaissance. Environ. Sci. Technol 36, 1202–1211. Kurt-Karakus, P.B., Bidleman, T.F., Muir, D.C.G., Cagampan, S.J., Struger, J., Sverko, E., Small, J.M., Jantunen, L.M., 2008. Chiral current-use herbicides in Ontario streams. Environ. Sci. Technol 42, 8452–8458. Lajeunesse, A., Gagnon, C., Sauve´, S., 2008. Determination of basic antidepressants and their N-desmethyl metabolites in raw sewage and wastewater using solid-phase extraction and liquid chromatography-tandem mass spectrometry. Anal. Chem. 80, 5325–5333. Letzel, M., Metzner, G., Letzel, T., 2009. Exposure assessment of the pharmaceutical diclofenac based on long-term measurements of the aquatic input. Environ. Int 35, 363–368. Lin, A.Y.-C., Reinhard, M., 2005. Photodegradation of common environmental pharmaceuticals and estrogens in river water. Environ. Toxicol. Chem. 24, 1303–1309. Lin, A.Y.-C., Plumlee, M.H., Reinhard, M., 2006. Natural attenuation of pharmaceuticals and alkylphenol polyethoxylate metabolites during river transport: photochemical and biological transformation. Environ. Toxicol. Chem. 25, 1458–1464. Lishman, L., Smyth, S.A., Sarafin, K., Kleywegt, S., Toito, J., Peart, T., Lee, B., Servos, M., Beland, M., Seto, P., 2006. Occurrence and reductions of pharmaceuticals and personal care products and estrogens by municipal wastewater treatment. Sci. Total Environ 367, 544–588. Lissemore, L., Hao, C., Yang, P., Sibley, P.K., Mabury, S., Solomon, K.R., 2006. An exposure assessment for selected pharmaceuticals within a watershed in Southern Ontario. Chemosphere 64, 717–729. Loraine, G.A., Pettigrove, M.E., 2006. Seasonal variations in concentrations of pharmaceuticals and personal care products in drinking water and reclaimed wastewater in Southern California. Environ. Sci. Technol 40, 687–695. MacLeod, S.L., 2009. Pharmaceutical occurrence and fate in wastewater and receiving surface waters in two Alberta watersheds. PhD dissertation. University of Alberta, Edmonton, Alberta. MacLeod, S.L., McClure, E.L., Wong, C.S., 2007a. Laboratory calibration and field deployment of the Polar Organic Chemical Integrative Sampler for pharmaceuticals and personal care products in wastewater and surface water. Environ. Toxicol. Chem. 26, 2517–2529. MacLeod, S.L., Sudhir, P., Wong, C.S., 2007b. Stereoisomer analysis of wastewater-derived beta-blockers, selective serotonin re-uptake inhibitors, and salbutamol by highperformance liquid chromatography-tandem mass spectrometry. J. Chromatogr. A 1170, 23–33. Maier, N.M., Franco, P., Lindner, W., 2001. Separation of enantiomers: needs, challenges, perspectives. J. Chromatogr. A 906, 3–33. Managaki, S., Murata, A., Takada, H., Tuyen, B.C., Chiem, N.H., 2007. Distribution of macrolides, sulfonamides, and trimethoprim in tropical waters: ubiquitous occurrence of
544
water research 44 (2010) 533–544
veterinary antibiotics in the Mekong Delta. Environ. Sci. Technol 41, 8004–8010. McArdell, C.S., Molnar, E., Suter, M.J.-F., Giger, W., 2003. Occurrence and fate of macrolide antibiotics in wastewater treatment plants and in the Glatt Valley Watershed. Switzerland. Environ. Sci. Technol 37, 5479–5486. Mehvar, R., Brocks, D.R., 2001. Stereospecific pharmacokinetics and pharmacodynamics of beta-adrenergic blockers in humans. J. Pharm. Pharm. Sci. 4, 185–200. Miao, X.-S., Yang, J.-J., Metcalfe, C.D., 2005. Carbamazepine and its metabolites in wastewater and in biosolids in a municipal wastewater treatment plant. Environ. Sci. Technol 39, 7469–7475. Neumann, J., 1989. Untersuchungen zur Bioverfu¨gbarkeit und Pharmakokinetik von Sulfasalazin und seinen Metaboliten. PhD thesis. Free University of Berlin. Nikolai, L.N., McClure, E.L., MacLeod, S.L., Wong, C.S., 2006. Stereoisomer quantification of the b-blocker drugs atenolol, metoprolol, and propranolol in wastewaters by chiral highperformance liquid chromatography-tandem mass spectrometry. J. Chromatogr. A 1131, 103–109. Sacher, F., Ehmann, M., Gabriel, S., Graf, C., Brauch, H.-J., 2008. Pharmaceutical residues in the river Rhine-results of a onedecade monitoring programme. J. Environ. Monit. 10, 664–670. Schmekel, B., Rydberg, I., Norlander, B., Sjo¨swa¨rd, K.N., Ahlner, J., Anderson, R.G.G., 1999. Stereoselective pharmacokinetics of S-salbutamol after administration of the racemate in healthy volunteers. Eur. Respir. J 13, 1230–1235. Shaw, M., Mueller, J.F., 2009. Time integrative passive sampling: how well do Chemcatchers integrate fluctuating pollutant concentrations? Environ. Sci. Technol 43, 1443–1448. Siebold, G., 2008. Lac La Biche County, Personal communication.
Stamm, C., Alder, A.C., Fenner, K., Hollender, J., Kraus, M., McArdell, C.S., Ort, C., Schneider, M.K., 2008. Spatial and temporal patterns of pharmaceuticals in the aquatic environment. Geog. Compass 2/3, 920–955. Stanley, J.K., Ramirez, A.J., Mottaleb, M., Chambliss, C.K., Brooks, B.W., 2006. Enantiospecific toxicity of the b-blocker propranolol to Daphnia magna and Pimephales promelas. Environ. Toxicol. Chem. 25, 1780–1786. Statistics Canada, 2009. http://www12.statcan.ca. accessed February 25. Stu¨lten, D., Zu¨hlke, S., Lamsho¨ft, M., Spiteller, M., 2008. Occurrence of diclofenac and selected metabolites in sewage effluents. Sci. Total Environ 405, 310–316. Vieno, N.M., Tuhkanen, T., Kronberg, L., 2005. Seasonal variation in the occurrence of pharmaceuticals in effluents from a sewage treatment plant and in the recipient water. Environ. Sci. Technol 39, 8220–8226. Vasskog, T., Anderssen, T., Pedersen-Bjergaard, S., Kallenborn, R., Jensen, E., 2008. Occurrence of selective serotonin reuptake inhibitors in sewage and receiving waters at Spitsbergen and in Norway. J. Chromatogr. A 1185, 194–205. Wong, C.S., 2006. Environmental fate processes and biochemical transformations of chiral emerging organic pollutants. Anal. Bioanal. Chem. 386, 544–558. 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 Mgmt 86, 72–80. Zhang, Z., Hibberd, A., Zhou, J.L., 2008. Analysis of emerging contaminants in sewage effluent and river water: comparison between spot and passive sampling. Anal. Chim. Acta 607, 37–44.
water research 44 (2010) 545–554
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Degradation of fifteen emerging contaminants at mg LL1 initial concentrations by mild solar photo-Fenton in MWTP effluents N. Klamerth a,b, L. Rizzo c, S. Malato a,*, Manuel I. Maldonado a, A. Agu¨era b, A.R. Ferna´ndez-Alba b a
Plataforma Solar de Almerı´a-CIEMAT. Carretera Sene´s km 4, 04200 Tabernas (Almerı´a), Spain Pesticide Residue Research Group, University of Almerı´a, 04120 Almerı´a, Spain c Department of Civil Engineering, University of Salerno, via Ponte don Melillo, 84084 Fisciano (SA), Italy b
article info
abstract
Article history:
The degradation of 15 emerging contaminants (ECs) at low concentrations in simulated and
Received 9 March 2009
real effluent of municipal wastewater treatment plant with photo-Fenton at unchanged pH
Received in revised form
and Fe ¼ 5 mg L1 in a pilot-scale solar CPC reactor was studied. The degradation of those
23 September 2009
15 compounds (Acetaminophen, Antipyrine, Atrazine, Caffeine, Carbamazepine, Diclofe-
Accepted 28 September 2009
nac, Flumequine, Hydroxybiphenyl, Ibuprofen, Isoproturon, Ketorolac, Ofloxacin, Proges-
Available online 4 October 2009
terone, Sulfamethoxazole and Triclosan), each with an initial concentration of 100 mg L1, was found to depend on the presence of CO2 3 and HCO3 (hydroxyl radicals scavengers) and
Keywords:
on the type of water (simulated water, simulated effluent wastewater and real effluent
Emerging contaminants
wastewater), but is relatively independent of pH, the type of acid used for release of
Pharmaceuticals treatment
hydroxyl radicals scavengers and the initial H2O2 concentration used. Toxicity tests with
Photo-Fenton
Vibrio fisheri showed that degradation of the compounds in real effluent wastewater led to
Solar photocatalysis
toxicity increase. ª 2009 Elsevier Ltd. All rights reserved.
1.
Introduction
Due to their growing use, pharmaceuticals, like the antiinflammatory ibuprofen, the antibiotic flumequine and the antiepileptic carbamazepine, endocrine disruptors like bisphenol A and atrazine, personal care products like oxybenzone and parabens (PHBA), synthetic musks and fragrances like musk xylene and galaxolide, pesticides like isoproturon and endosulphan and illicit drugs like THC aznd cocaine, to name just a few, and other xenobiotic substances, are found in increasing quantities in wastewater, surface water, and even in drinking water (Kasprzyk-Hordern et al., 2009; Kim et al., 2007; Mitch et al., 2003; Esplugas et al., 2007; Ternes, 1998). Since the use of these substances cannot be controlled or eliminated as they are ever present in our daily
lives, their release into the environment has to be optimized and restricted, as they pose risks to the environment, public health and aquatic systems and they are responsible for building up microbiological resistance, feminisation of higher organisms and ecotoxicological issues (Laville et al., 2004). Particularly relevant examples of such emerging contaminants (ECs), such as those mentioned above, which are ubiquitously present in influents and effluents of MWTPs in the high ng L1 to low mg L1 range, do not need to be persistent to be hazardous, because they are introduced continuously into the environment (Fono et al., 2006; Jackson and Sutton, 2008; Nakada et al., 2008; Petrovic et al., 2003). Conventional MWTPs, typically based on biological processes, are capable of removing some substances, but nonbiodegradable compounds may escape the treatment and be
* Corresponding author. Tel.: þ34 950387940; fax: þ34 950365015. E-mail address:
[email protected] (S. Malato). 0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2009.09.059
546
water research 44 (2010) 545–554
released into the environment (Go¨bel et al., 2007; Carballa et al., 2004; Ternes et al., 2007). ECs have been found in the MWTP effluents at mean concentrations ranging from 0.1 to 20 mg L1 (Castiglioni et al., 2006; Martı´nez Bueno et al., 2007; Radjenovic´ et al., 2007; Richardson, 2007; Zhao et al., 2009). Concern about the growing problem of the continuously rising concentrations of these compounds must be emphasized, and therefore, the application of more thorough wastewater treatment protocols, including the use of new and improved technologies, is a necessary task. Reusable water should be free of these persistent, toxic, endocrine-disrupting or non-biodegradable substances, (Radjenovic´ et al., 2007; Teske and Arnold, 2008), and therefore, an effective tertiary treatment is required to remove these substances completely. Among the advanced technologies that may be used to remove these pollutants, (Gogate and Pandit, 2004; Saritha et al., 2007; Huber et al., 2005) advanced oxidation processes (AOPs), through the generation of hydroxyl radicals which are able to mineralise most organic molecules yielding CO2 and inorganic ions as final products, are a particularly attractive option. (Farre´ et al., 2005; Gebhardt and Schro¨der, 2007; Gu¨ltekin and Ince, 2007; Ning et al., 2007; Rosenfeldt and Linden, 2004; Rosenfeldt et al., 2007; Ternes et al., 2003) The generation of the OH radicals can be achieved: electrochemically (Can˜izares et al., 1999; Pelegrini et al., 2001; Zhou et al., 2005), sonochemically (Mantzavinos et al., 2004; Papadaki et al., 2004; Lesko et al., 2006), photochemically (Esplugas et al., 2005; Bali et al., 2003; Bremner et al., 2006), and by homogeneous or heterogenous catalysis (Zepp et al., 1992; Martinez et al., 2005) in acid or basic media (Glaze et al., 1987; Hislop and Bolton, 1999; Neyens and Baeyens, 2003). Most of the AOPs make use of a combination of either oxidants and irradiation (O3/H2O2/ UV), or a catalyst and irradiation (Fe2þ/H2O2; UV/TiO2). The drawbacks which make them economically disadvantageous depend on the specific AOP: (i) High electricity demand (e.g. ozone and UV-based AOPs), (ii) the relatively large amounts of oxidants and/or catalysts (e.g. ozone, hydrogen peroxide and iron-based AOPs), and (iii) the pH operating conditions (e.g. Fenton and photo-Fenton). This is why, although AOPs are well known for their capacity for oxidising and mineralising almost any organic contaminant, commercial applications are still scarce. Processes like photo-Fenton may be applied to commercial applications by using solar energy as a light source, optimizing the pH range and the amounts of oxidant/ catalyst required. AOP efficiency in the removal of ECs has typically been studied in demineralised water and bench scale, at initial concentrations in the milligram-to-gram range, which is not realistic compared to the concentrations detected in real water and wastewater (Farre´ et al., 2005; Lapertot et al., 2007; Kassinos et al., 2009; Malato et al., 2007). This work focused on solar photo-Fenton degradation of the ECs typically found in the effluents of MWTPs, leaving the treated wastewater suitable for reuse. Moreover, to make the process of interest for practical applications high iron concentrations (mM range), excessive amounts of H2O2 and a pH under 3 must be avoided (Pignatello et al., 2006). A new approach aimed at finding a very mild photo-Fenton treatment (low iron concentration and H2O2 dose at neutral pH), has been proposed (Moncayo-
Lasso et al., 2008; Klamerth et al., 2009). In this paper, a pilotscale solar photo-Fenton treatment was run at a pH between 6 and 7, with starting concentrations of 5 mg L1 Fe, and 50 mg L1 H2O2. Synthetic water (SW), simulated effluent wastewater (SE) and real effluent wastewater (RE) were tested in this study to which a mixture of 15 ECs, consisting of pharmaceuticals, pesticides and personal care products, selected from a list of 80 compounds found in MWTP effluents in previous studies (Martı´nez Bueno et al., 2007), were added at low concentrations (100 mg L1 each).
2.
Material and methods
2.1.
Reagents
All reagents used for chromatographic analyses, acetonitrile, methanol, and ultrapure (MilliQ) water, were HPLC grade. Analytical standards for chromatography analyses were purchased from Sigma-Aldrich. Table 1 and Scheme 1 list the 15 compounds (pharmaceuticals, pesticides and personal care products) used. Photo-Fenton experiments were performed using iron sulphate (FeSO4$7H2O), reagent-grade hydrogen peroxide (30% w/v), sulphuric acid and hydrochloric acid for carbonate stripping, all provided by Panreac. The filters used were syringe driven 0.2 mm Millex nylon membrane filters from Millipore. The reagents used in the preparation of the synthetic water (NaHCO3, CaSO4$2H2O, MgSO4, and KCl), and simulated effluent wastewater (Peptone, Meat extract, Urea, K2HPO4, CaCl2$2H2O, MgSO4$7H2O and NaCl) were provided by Panreac.
2.2.
Polluted waters
Taking into consideration that typical EC concentrations in the effluent are in the 0.1–20.0 mg L1 range, it was decided to work at 100 mg L1, which is a compromise between (i) a high enough concentration to characterise kinetics using conventional
Table 1 – LOD, LOQ and absorption l of the selected compounds. Name Acetaminophen Antipyrine Atrazine Caffeine Carbamazepine Diclofenac Flumequine Hydroxybiphenyl Ibuprofen Isoproturon Ketorolac Ofloxacin Progesterone Sulfamethoxazole Triclosan
LOD (mg L1)
LOQ (mg L1)
Absorption l [nm]
1.4 2.1 0.6 0.7 0.8 2.7 2.1 1.6 2.5 1.5 2.1 1.3 2.0 1.4 5.0
2.8 4.2 1.2 1.5 1.6 5.5 4.3 3.2 5.0 3.0 4.2 2.6 4.0 2.7 10.0
245 205 223 205 211 277 248 243 222 205 321 295 248 267 280
547
water research 44 (2010) 545–554
Name
Structure
Name Ibuprofen
Acetaminophen HO
analgesic
Structure
NH
nonsteroidal
/
O
antipyretic
anti-
OH
inflammatory
O
O
Isoproturon
Antipyrine
H N N
analgesic
phenylurea
N
O
herbicide
H N
N
O
H N
N
Ketorolac
Atrazine
N N
N
OH
anti-
herbicide
inflammatory
Cl
O N
O
Caffeine
F
Ofloxacin
N
gramnegative
O
N
N
O
stimulant
O
N
N
antibiotic
N
O
OH
O
Progesterone
Carbamazepine N
steroid
anticonvulsant
hormone
NH2
O
O O
Diclofenac Cl
O
Sulfamethoxaz OH
ole
H2N
S
H N
anti-
H N N
O
O
bacteriostatic
inflammatory
antibiotic
Cl O
Flumequine
F
COOH
Cl
Triclosan
OH O
broadspectrum
anti-bacterial/ N
antibiotic
Hydroxybiphenyl
fungal agent
Cl
OH
biocide
Scheme 1 – Name and structure of the 15 selected compounds.
Cl
548
water research 44 (2010) 545–554
1.0
Acetaminophen Caffeine Ofloxacine Antipyrine Sulfamethoxazole
illumination
0.5 Fenton
Carbamazepine Flumequine Ketorolac Atrazine Isoproturon
0.5
0.0 1.0
Hydroxybenzone Diclofenac Ibuprofen Progesterone Triclosan
0.5
0.0 -40
-20
0
20
tEXP, min
40
60
80
100
t30W, min
Fig. 1 – Degradation of the 15 ECs (0.1 mg LL1 each) by photo-Fenton with 5 mg LL1 Fe without pH adjustment in SW acidified for release of carbonates.
2.3.
analytical techniques, and (ii) low enough to simulate real conditions. Synthetic water (SW), simulated effluent wastewater (SE) and real effluent wastewater (RE) to which a mixture of the 15 ECs previously listed had been added at low concentrations (100 mg L1) were tested. The experimental protocol was designed to study solar photo-Fenton with a relatively uncomplicated aqueous matrix (SW) first, before gradually increasing complexity by studying SE and finally RE, in order to acquire information on process operation. Demineralised water used for preparing SW and SE in the pilot plant was supplied by the Plataforma Solar de Almerı´a
1.0
Solar photo-Fenton pilot plant
Photo-Fenton experiments were performed at the Plataforma Solar de Almerı´a in a pilot compound parabolic collector (CPC) solar plant designed for solar photocatalytic applications. This reactor is composed of two 11dL modules with twelve Pyrex glass tubes (30 mm O.D.) mounted on a fixed platform tilted 37 (local latitude). The water flows (20 L min1) directly from one module to the other and finally to a 10dL reservoir. The piping and valves (3 L) between the reactor and the tank are black HDPE, which is highly resistant to chemicals, weatherproof and opaque, preventing any photochemical effect from outside the collectors. The total illuminated area is 3 m2, the total volume (two modules þ reservoir tank þ piping and valves) is 35 L (VT) and the irradiated volume is 22 L (Vi). Solar ultraviolet radiation (UV) was measured by a global UV radiometer (KIPP&ZONEN, model CUV 3) mounted on a platform tilted 37
Acetaminophen Caffeine Ofloxacine Antipyrine Sulfamethoxazole
illumination
0.5
C/C0
0.0 1.0
Carbamazepine Flumequine Ketorolac Atrazine Isoproturon
0.5
0.0 1.0
Fenton 120
consumed H2O2
0.5
Hydroxybenzone 90 Diclofenac 60 Ibuprofen Progesterone 30 Triclosan
0.0
0 0
tEXP, min
60
120
180
240
300
consumed H2O2, mg L-1
C/C0
0.0 1.0
(PSA) distillation plant (conductivity < 10 mS cm1, Cl ¼ 0.7– 1 1 0.8 mg L1, NO 3 ¼ 0.5 mg L , organic carbon < 0.5 mg L ). The preparation of standard moderately-hard freshwater (SW) was carried out taking into account the characteristics of ground water in the province of Almerı´a (southern Spain). SW was prepared by mixing 96 mg L1 of NaHCO3, 60 mg L1 of CaSO4$2H2O, 60 mg L1 of MgSO4, and 4 mg L1 of KCl (Standard Methods, 1998). The simulated effluent wastewater (SE) consists of SW and Peptone (32 mg L1), Meat extract (22 mg L1), Urea (6 mg L1), K2HPO4 (28 mg L1), CaCl2$2H2O (4 mg L1), MgSO4$7H2O (2 mg L1) and NaCl (7 mg L1) which derives to an initial DOC (dissolved organic carbon) of 25 mg L1 (OECD, 1999). RE was taken downstream of the Almerı´a MWTP secondary biological treatment and used as received within the next 2 days. Initial COD and DOC were at 60 mg L1 and 25 mg L1, respectively.
t30W, min
Fig. 2 – Degradation of the 15 ECs (0.1 mg LL1 each) by photo-Fenton (5 mg LL1 Fe, no pH adjustment) and hydrogen peroxide consumption in SE acidified for carbonate release.
549
water research 44 (2010) 545–554
Table 2 – Final concentration of ECs in two different experiments with SE after similar irradiation times, with [Fe]0 [ 5 mg LL1: 300 min with [H2O2]0 [ 50 mg LL1 and addition of 50 mg LL1 when the initial dose was almost consumed and so on; 336 min with [H2O2]0 [ 5 mg LL1 and addition of H2O2 every 30–60 min until the end of the experiment. Fenton degradation (%) in the dark and first order kinetic constants (k) for photo-Fenton are also included. [H2O2]0 ¼ 50 mg L1
[H2O2]0 ¼ 5 mg L1
t30W
C/C0
Fenton deg. X [%]
k [min1]
t30W
C/C0
Fenton deg. X [%]
k [min1]
Acetaminophen Caffeine Ofloxacin Antipyrine Sulfamethoxazole Carbamazepine Flumequine Ketorolac Atrazine Isoproturon Hydroxybiphenyl Diclofenac Ibuprofen Progesterone Triclosan
261 300 56 300 300 300 56 191 300 300 300 111 261 300 300
16.1 21.7 63.2 10.5 10.2 18.0 39.3 21.8 6.1 12.4 15.6 35.2 11.3 18.3 14.9
0.016 0.008 0.063 0.008 0.012 0.014 0.078 0.022 0.004 0.016 0.012 0.032 0.011 0.011 0.005
248 336 114 336 336 336 114 248 336 336 336 133 294 336 336
< LOD 0.09
14.1 1.2 56.7 3.8 4.5 6.6 30.0 11.6 0.0 6.9 6.2 20.1 11.3 8.9 15.6
0.019 0.007 0.034 0.007 0.010 0.013 0.036 0.017 0.004 0.011 0.010 0.028 0.010 0.011 0.005
Conditions TOC [mg L1] Used H2O2 [mg L1]
Initial 37.8
End 31.1 110
Initial 39.9
(the same as the CPCs). The temperature inside the reactor was continuously recorded by a temperature probe (Crioterm PT-100 3H) inserted in the piping. A plant diagram has been published elsewhere (Kositzi et al., 2004). With Eq. (1), combination of the data from several days’ experiments and their comparison with other photocatalytic experiments is possible,
lasted as long as there was H2O2 to be consumed (usually 3–4 h), in the case of experiments which lasted longer than that, the experiment was covered at the end of the day and the next day, after recirculation for 15 min, a sample was taken, peroxide was added (50 mg L1) and the covers again removed.
2.5. t30W;n
UV Vi ¼ t30W;n1 þ Dtn ; Dtn ¼ tn tn1 ; t0 ¼ 0ðn ¼ 1Þ 30 VT
(1)
where tn is the experimental time for each sample, UV is the average solar ultraviolet radiation (l < 400 nm) measured between tn1 and tn, and t30W is a ‘‘normalized illumination time’’. In this case, time refers to a constant solar UV power of 30 Wm2 (typical solar UV power on a perfectly sunny day around noon).
2.4.
Experimental setup
The mixture of the 15 compounds (100 mg L1 each) dissolved in Methanol at 2.5 g L1 (mother solution) was added directly (1.4 mL) into the pilot plant and well homogenized by turbulent recirculation for 15 min. Methanol was used to dissolve the contaminants (TOC from methanol was 12 mg L1). The pH in SW was 7.6, in SE 7.8 and in RE 8.0, which favoured bicarbonate ions, so at the beginning of the process, when the collectors and HCO were still covered, enough acid to remove CO2 3 3, known OH radical scavengers, was added. Recirculation time for this process was between 15 and 60 min. When the total inorganic carbon (TIC) concentration was below the desired level (10 mg L1), 50 mg L1 of peroxide was added and homogenised by recirculation for 15 min. Finally, Fe2þ as FeSO4$7H20 was added (5 mg L1). After 15 min of recirculation, in which the Fenton reaction started, the collectors were uncovered and photo-Fenton began. Hydrogen peroxide and iron were measured in every sample taken. The experiments normally
End 38.7 52
Analytical measurements
Experiments were performed with 100 mg L1 of each of 15 different compounds in synthetic water (SW), simulated effluent wastewater (SE) and real effluent wastewater (RE). Therefore, no study of degradation intermediates or mineralisation was done, as under these conditions, it is really almost impossible to monitor degradation intermediates. Wastewater background DOC was in the range of tens of mg L1, and all 15 compounds together at concentrations of 100 mg L1 produce no more than 1 mg L1 of DOC. Therefore, results of DOC degradation with EC mineralisation would be meaningless. Dissolved Organic Carbon (DOC) and total inorganic carbon (TIC) were measured by direct injection of samples filtered with 0.2 mm syringe-driven filters into a Shimadzu – 5050A TOC analyser. Colorimetric determination of total iron concentration was used with 1,10-phenantroline according to ISO 6332. Hydrogen peroxide was analyzed by a fast, simple spectrophotometric method using ammonium metavanadate, which allows the H2O2 concentration to be determined immediately based on a red-orange peroxovanadium cation formed during the reaction of H2O2 with metavanadate, with a maximum absorption at 450 nm. The peroxide concentrations are calculated from absorption measurements according to the method proposed by Nogueira et al. (2005). The concentration profile of each compound during degradation was determined by UPLC-UV (series 1200, Agilent technologies, Palo Alto, CA) using a method specifically developed for this application (Martı´nez-Bueno et al., 2007).
550
water research 44 (2010) 545–554
Table 3 – Final EC concentration in RE with [Fe]0 [ 5 mg LL1, [H2O2]0 [ 50 mg LL1 after an irradiation time of t30W [ 276 min. Fenton degradation (%) in the dark and first order kinetic constant (k) for photo-Fenton are also included. t30W
C/C0
Fenton deg. X [%]
k [min1]
Acetaminophen Caffeine Ofloxacin Antipyrine Sulfamethoxazole Carbamazepine Flumequine Ketorolac Atrazine Isoproturon Hydroxybiphenyl Diclofenac Ibuprofen Progesterone Triclosan
109 158 102 276 198 158 102 43 276 122 109 64 109 158 276
16.1 36.4 11.4 21.3 20.5 19.7 23.1 33.3 1.4 15.3 14.7 38.0 10.1 23.7 15.0
0.030 0.020 0.042 0.015 0.025 0.026 0.040 0.107 0.009 0.032 0.033 0.066 0.030 0.024 0.005
Conditions TOC [mg L1] Used H2O2 [mg L1]
Initial 55
the wavelength of maximum light absorption of each compound with the response limits of detection (LODs) and limits of quantification (LOQs) as shown in Table 1.
2.6.
Toxicity test with Vibrio fisheri
Photo-Fenton experiments with SE and RE were repeated for toxicity test sampling. During the experiments, 0.5 mL of H2O2 (which corresponds to the original H2O2 concentration of 5 mg L1) were added every 30 min. The time had been previously estimated to be long enough to make H2O2 residual totally disappear; so, after 30 min a sample was taken for toxicity testing and then more H2O2 was added. The pH of the samples was adjusted to 7. Toxicity tests based on the inhibition of the luminescence emitted by the marine bacteria V. fisheri, were carried out using a Biofix Lumi-10. The reagent is a freeze-dried preparation of a specially selected strain of the marine bacterium V. fisheri (formerly known as Photobacterium phosphoreum, NRRL number B-11177). The drop in light emission is measured after 30-min contact periods.
End 34.5 45
Analytes were separated using a reversed-phase C-18 analytical column (XDB-C-18 1.8 mm, 4.6 50 mm) using acetonitrile (mobile phase A) and ultrapure water (25 mM formic acid, mobile phase B) at a flow rate of 1 mL min1. A linear gradient progressed from 10% A (original conditions) to 82% A in 12 min. Re-equilibration time was 3 min. A 25 mL sample was filtered through a 0.2dmm syringe-driven filter, and the filter was washed with 3 mL acetonitrile in order to remove any adsorbed compounds. The two solutions were mixed and a 100dmL aliquot was injected into the UPLC. The UV signal was recorded at different wavelengths according to
3.
Results and discussion
3.1.
Photo-Fenton tests using SW
Previous experiments have confirmed that low iron concentrations without pH adjustment can lead to extremely slow degradation kinetics because of the presence of carbonate and HCO species (CO2 3 3 ) which compete with organic contaminants for hydroxyl radicals (Klamerth et al., 2009). Therefore, SW was acidified (41 mg L1 HCl) to remove carbonates prior to photo-Fenton. 15 min recirculation in the pilot plant was necessary to remove carbonates (final TIC ¼ 5 mg L1) with pH between 5.2 and 6.0. Degradation of pollutants (Fig. 1) started in the dark (Fenton), 15 min before uncovering the
100
2
1
3
inhibition, %
75 50 25 0
SE RE
-25 -50 1.0
0
10
20
30
40
Antrazine Diclofenac Carbamazepine Ketorolac
0.5
C/C0
50 SE
0.0 1.0
RE Atrazine Diclofenac Carbamazepine Ketorolac
0.5 0.0 0
10
20
30
40
50
dose H2O2, mg L-1 Fig. 3 – Toxicity tests and correlation of toxicity with concentrations of 4 selected compounds.
water research 44 (2010) 545–554
reactor (tExP ¼ 15 min). Removal of all compounds but atrazine (18%) was in the range between 30% (caffeine) and 60% (ibuprofen). Fenton degradation was very quick after adding just 5 mg L1 of Fe2þ but it did not proceed further until the reactor was uncovered, as Fe3þ reduction to Fe2þ was inefficient without illumination. All the pollutants but triclosan (87%) and atrazine (77%) were efficiently degraded (removal over 94%) under photoFenton conditions, at t30W ¼ 90 min. Moreover, some pollutants (ofloxacin and diclofenac) were degraded to below the LOD as early as t30W ¼ 32 min. The total amount of H2O2 consumed was 37 mg L1 and the iron concentration decreased due to precipitation from 5 mg L1 to 3.1 mg L1, probably due to colloidal aggregation, while the pH varied from an original 7.6–5.3 (after adding the FeSO4$7H2O) to a final pH ¼ 3.8 at t30W ¼ 90 min. Similar results were found by adding H2SO4 (56 mg L1) instead of HCl. H2SO4 has certain advantages (handling, volatility, etc.) over HCl, and Cl ions may act as radical scavengers. Experiments did not show any significant difference between using HCl or H2SO4. Indeed, the total concentration of chloride and sulphate measured in SW was 43 mg L1 Cl when HCl was added and 145 mg L1 SO2 4 when H2SO4 was added respectively. It is well known that inorganic species (chloride, sulphate, phosphates, etc) are usually detrimental to the photo-Fenton reaction rate (Pignatello et al., 2006) due to complexation of these ions with Fe2þ or Fe3þ, or to the scavenging of hydroxyl radicals and formation of less reactive inorganic ions (De Laat et al., 2004), but it has also been observed that they significantly reduce the photoFenton reaction rate only if they occur at concentrations over 500 mg L1, usually very far from the concentration found in natural water (Bacardit et al., 2007; Zapata et al., 2009).
3.2.
Photo-Fenton tests using SE
The following experiment dealt with the photo-Fenton treatment of SE, which was characterised by original TOC and COD of 37 mg L1 and 80 mg L1, respectively. Two different approaches to the H2O2 dose were studied: (i) 50 mg L1 were added at the beginning and again when the initial dose was almost consumed and so on; (ii) 5 mg L1 were added every 30–60 min (also depending on consumption) until the end of the experiment. In this case, 56 mg L1 H2SO4 acid were added. A recirculation time of 15 min was necessary for carbonates to be released (TIC < 5 mg L1). As can be seen in Fig. 2, degradation is much slower with both Fenton (t < 0) and photoFenton (t30W > 0) than in the experiment with SW shown in Fig. 1. The residual concentration of the contaminants at the end of the experiments (t30W ¼ 300 min and 336 min, respectively) can be seen in Table 2. The pH varied during the experiment with high H2O2 concentrations from the original 7.6–4.9 at the end, while the iron concentration varied from 5 mg L1 to 3.2 mg L1. During the experiment with low H2O2 concentrations the pH varied from 8.1 to 4.9, while iron concentration went from 4.8 to 3.1 mg L1. The EC degradation behaviour did not change significantly, no matter which hydrogen peroxide dose approach was used, as can be seen in Table 2, although the overall amount of peroxide consumed was much lower (110 mg in 50-mg-L1 additions compared to 52 mg with the 5-mg-L1 additions). It should be mentioned, although not relevant to the purpose of the experiments,
551
because the main purpose was to degrade ECs, that TOC mineralisation was rather low in both cases (around 25%).
3.3.
Photo – Fenton treatment of RE
The initial DOC, TIC and COD were 36 mg L1, 106 mg L1 and 60 mg L1, respectively. In this case, 406 mg L1 H2SO4 were added to reach < 20 mg L1 TIC and recirculation time in the pilot plant necessary to remove carbonate species was 30 min. According to previous tests, photo-Fenton treatment at a TIC over 20 mg L1 leads to very slow degradation. Interestingly, although the degradation rate of all ECs was faster than in the previous experiments in SE, as observed in Table 3, DOC degradation was similar to SE (around 20%). This can be explained by the presence of humic acids in RE which produce solvated electrons and hydroxyl radicals upon irradiation (Fukushima and Tatsumi, 2001; Prosen and Zupancˇicˇ-Kralj, 2005; Auger et al., 2002).
3.4. The effect of photo-Fenton treatment on V. fisheri toxicity In Fig. 3, the results of V. fisheri toxicity tests with SE and RE after photo-Fenton experiments are compared. Three different steps in toxicity behaviour (marked 1, 2 and 3 in Fig. 3) can be described. Before photo-Fenton starts, SE toxicity (20–30% inhibition) may be due to the mixture of pollutants as well as to the formation of intermediates during Fenton in the dark, particularly from the fast oxidation of some of the ECs. On the contrary, simultaneous activation (negative inhibition) is also observable in the RE experiment. This completely different behaviour may be explained by the RE – nutrients and salt contents, which in someway favoured the growth of V. fisheri until the toxic effect of pollutants and their oxidation intermediates was neutralized. In both SE and RE, when photo-Fenton starts up, from an of H2O2 dose ¼ 0 mg L1 up to H2O2 dose ¼ 50 mg L1, the degradation of parent pollutants begins forming more toxic intermediates, which drastically increase the inhibition rate in SE (Fig. 3, Step 2). The previously favourable conditions of RE probably can no longer efficiently balance the increasingly toxic effect, so inhibition start to gradually increase in RE too. In the last step (3) inhibition with RE is quite constant, probably because toxic organic intermediates take longer to mineralise; on the other hand, toxicity in SE almost totally disappeared. If we compare these results with residual concentration of the four parent ECs selected for comparison (selected for their representativeness of the behaviour of the 15 different compounds) it may be observed that photo-Fenton treatment was more efficient in the RE experiment than in the SE. So in terms of toxicity, the higher removal rate in RE experiment may result in a correspondingly larger amount of more toxic intermediates than in the SE experiment or the generation of other intermediates from RE organics. Therefore, toxicity assessment be should taken into account as well as EC degradation during AOP wastewater treatment.
4.
Conclusions - The experiments showed that ECs at low concentrations (mg L1 range) can be successfully degraded to negligible
552
water research 44 (2010) 545–554
concentrations with solar photo-Fenton at low iron concentrations (5 mg L1) and low initial H2O2 (50 mg L1) concentrations without adjusting the pH. - One limiting factor for the degradation rate is the presence of CO2 3 and HCO3 which are very efficient OH radical scavengers and which have to be removed prior to the photoFenton reaction. This can be done either with HCl or H2SO4, as the range of the final concentration of Cl or SO2 4 does not influence the reaction. Therefore, either one may be used as economic/operating reasons recommend, but not for any kinetic effect. It should be taken into account that around 400 mg L1 of H2SO4 (0.05 V kg1) were necessary for CO2 stripping in RE, which means around 0.025 V m3. However, these costs should be evaluated on a case by case basis as it strongly depends on the particular RE. - Changing the operating conditions by lowering H2O2 concentration (5 mg L1) and adjusted by properly controlled dosing could lead to comparable degradation of the compounds with lower peroxide consumption. - Degradation is faster in RE due to humic acids producing solvated electrons and hydroxyl radicals upon irradiation, and these contribute to the radicals produced by photoFenton. Toxicity from the formation of oxidation intermediates should be taken into account in the treatment of effluents from wastewater treatment plants by AOPs, because they could increase final toxicity.
Acknowledgements Funding for this work was provided by the Spanish Ministry of Science and Innovation under the Consolider-Ingenio 2010 programme (Project CSD2006–00044 TRAGUA; http://www. consolider-tragua.com) and by the Andalusia Regional Government (Project no. P06-TEP-02329). Nick Klamerth would to thank the University of Almerı´a and CIEMAT for his Ph. D. research grant. Luigi Rizzo wishes to thank the University of Salerno (Italy) and the Province of Salerno for the research grant which allowed him to work in the above mentioned project at the Plataforma Solar de Almerı´a. The authors also wish to thank Mrs. Deborah Fuldauer for English language correction.
references
Auger, J.P., Richard, C., Trubetskaya, O., Trubetskoj, O., Le´ve`que, J., Andreux, F., 2002. Photoinductive efficiency of soil extracted humic and fulvic acids. Chemosphere 49, 259–262. Bacardit, J., Sto¨tzner, J., Chamarro, E., Esplugas, S., 2007. Effects of salinity on the photo-Fenton process. Industrial and Engineering Chemistry Research 46, 7615–7619. Bali, U., Catalkaya, E.C., Sengul, F., 2003. Photochemical degradation and mineralisation of phenol: a comparative study. Journal of Environmental Science and Health – Part A Toxic/Hazardous Substances and Environmental Engineering 38, 2259–2275. Bremner, D.H., Burgess, A.E., Houllemare, D., Namkung, K., 2006. Phenol degradation using hydroxyl radicals generated from zero-valent iron and hydrogen peroxide. Applied Catalysis B: Environmental 63, 15–19.
Can˜izares, P., Domı´nguez, J.A., Rodrigo, M.A., Villasen˜or, J., Rodrı´guez, F., 1999. Effect of the current intensity in the electrochemical oxidation of aqueous phenol wastes at an activated carbon and steel anode. Industrial and Engineering Chemistry Research 38, 3779–3785. Carballa, M., Omil, F., Lema, J.M., Llompart, M., Garcia-Jares, C., Rodriguez, I., Gomez, M., Ternes, T., 2004. Behaviour of pharmaceuticals, cosmetics and hormones in a sewage treatment plant. Water Research 38, 2918–2926. 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, 357–363. De Laat, J., Truong Le, G., Legube, B., 2004. A comparative study of the effects of chloride, sulfate and nitrate ions on the rates of decomposition of H2O2 and organic compounds by Fe(II)/H2O2 and Fe(III)/H2O2. Chemosphere 55, 715–723. Esplugas, S., Rodrı´guez, M., Malato, S., Pulgarin, C., Contreras, S., Curco´, D., Gime´nez, J., 2005. Optimizing the solar photo Fenton process in the treatment of contaminated water. Determination of intrinsic kinetic constants for scale-up. Solar Energy 79, 360–368. Esplugas, S., Bila, D., Krause, L., Dezotti, M., 2007. Ozonation and advanced oxidation technologies to remove endocrine disrupting chemicals (EDCs) and pharmaceuticals and personal care products (PPCPs) in water effluents. A review. Journal of Hazardous Materials 149, 631–642. Farre´, M.J., Franch, M.I., Malato, S., Ayllon, J., Peral, J., Dome´nech, X., 2005. Degradation of some biorecalcitrant pesticides by homogeneous and heterogeneous photocatalytic ozonation. Chemosphere 58, 1127–1133. Fono, L.J., Kolodziej, E.P., Sedlak, D.L., 2006. Attenuation of wastewater-derived contaminants in an effluent dominated river. Environmental Science and Technology 40, 7257–7262. Fukushima, M., Tatsumi, K., 2001. Degradation pathways of pentachlorophenol by photo-Fenton systems in the presence of iron(III), humic acids and hydrogen peroxide. Environmental Science and Technology 35, 1771–1778. Gebhardt, W., Schro¨der, H., 2007. Liquid chromatography(tandem) mass spectrometry for the follow-up of the elimination of persistent pharmaceuticals during wastewater treatment applying biological wastewater treatment and advanced oxidation. Journal of Chromatography A 1160, 34–43. Glaze, W., Kang, W., Chapin, H., 1987. Chemistry of water treatment processes involving ozone, hydrogen peroxide and ultraviolet radiation. Ozone Science & Engineering 9, 335–352. Go¨bel, A., McArdell, C., Joss, A., Siegrist, H., Giger, W., 2007. Fate of sulfonamides, macrolides and trimethoprim in different wastewater treatment technologies. Science of the Total Environment 372, 361–371. Gogate, P.R., Pandit, A.B., 2004. A review of imperative technologies for wastewater treatment. I: oxidation technologies at ambient conditions. Advances in Environmental Research 8, 501–551. Gu¨ltekin, I., Ince, N.H., 2007. Synthetic endocrine disruptors in the environment and water remediation by advanced oxidation processes. Journal of Environmental Management 85, 816–832. Hislop, K.A., Bolton, J.R., 1999. The photochemical generation of OH radicals in the UV–Vis/ferrioxalate/H2O2 system. Environmental Science and Technology 33, 3119–3126. Huber, M., Coronen, S., Ternes, T., von Gunten, U., 2005. Oxidation of pharmaceuticals during water treatment with chlorine dioxide. Water Research 39, 3607–3617. Jackson, J., Sutton, R., 2008. Sources of endocrine disrupting chemicals in urban wastewater, Oakland, CA. Science of the Total Environment 405, 153–160.
water research 44 (2010) 545–554
Kassinos, D., Varanva, N., Michael, C., Piera, P., 2009. Homogenous oxidation of aqueous solutions of atrazine and fenitrothion through dark and photo-Fenton reactions. Chemosphere 74, 866–872. Kasprzyk-Hordern, B., Dinsdale, R.M., Guwy, A.J., 2009. The removal of pharmaceuticals, personal care products, endocrine disruptors and illicit drugs during wastewater treatment and its impact on the quality of receiving waters. Water Research 43, 363–380. Kim, S.D., Cho, J., Kim, I.S., Vanderford, B.J., Snyder, S.A., 2007. Occurrence and removal of pharmaceuticals and endocrine disruptors in South Korean surface, drinking, and waste waters. Water Research 41, 1013–1021. Klamerth, N., Miranda, N., Malato, S., Agu¨era, A., Ferna´ndezAlba, A.R., 2009. Degradation of emerging contaminants at low concentrations in MWTPs effluents with mild solar photoFenton and TiO2. Catalysis Today 144, 124–130. Kositzi, M., Poulios, I., Malato, S., Ca´ceres, J., Campos, A., 2004. Solar photocatalytic treatment of synthetic municipal wastewater. Water Research 38, 1147–1154. 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: Chemistry 186, 34–40. Laville, N., Ait-Aissa, S., Gomez, E., Casellas, C., Porcher, J.M., 2004. Effects of human pharmaceuticals on cytotoxicity, EROD activity and ROS production in fish hepatocytes. Toxicology 196, 41–55. Lesko, T., Colussi, A., Hoffmann, M., 2006. Sonochemical decomposition of phenol: evidence for a synergistic effect of ozone and ultrasound for the elimination of total organic carbon from water. Environmental Science and Technology 40, 6818–6823. 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, 137–149. Mantzavinos, D., Vassilakis, C., Pantidou, A., Psillakis, E., Kalogerakis, N., 2004. Sonolysis of natural phenolic compounds in aqueous solutions: degradation pathways and biodegradability. Water Research 38, 3110–3118. Martinez, F., Calleja, G., Melero, J.A., Molina, R., 2005. Heterogeneous photo-Fenton degradation of phenolic aqueous solutions over iron containing SBA-15 catalyst. Applied Catalysis B: Environmental 60, 181–190. Martı´nez Bueno, M.J., Agu¨era, A., Go´mez, M.J., Hernando, M.D., Garcı´a-Reyes, J.F., Ferna´ndez-Alba, A.R., 2007. Application of liquid chromatography/quadrupole-linear ion trap mass spectrometry and time-of-flight mass spectrometry to the determination of pharmaceuticals and related contaminants in wastewater. Analytical Chemistry 79, 9372–9384. Mitch, W.A., Sharp, J.O., Trussel, R.R., Valentine, R.L., AlvarezCohen, L., Seldak, D.L., 2003. N-Nitrosodimethylamine (NDMA) as a drinking water contaminant: A review. Environmental Engineering and Science 20, 389–404. Moncayo-Lasso, A., Pulgarin, C., Benı´tez, N., 2008. Degradation of DBPs’ precursors in river water before and after slow sand filtration by photo-Fenton process at pH 5 in a solar CPC reactor. Water Research 42, 4125–4132. Nakada, N., Kiri, K., Shinohara, H., Harada, A., Kuroda, K., Takizawa, S., Takada, H., 2008. Evaluation of pharmaceuticals and personal care products as water-soluble molecular markers of sewage. Environmental Science and Technology 42, 6347–6353. Ning, B., Graham, N., Zhang, Y., Nakonechny, M., El-Din, M.G., 2007. Degradation of endocrine disrupting chemicals by ozone/AOPs. Ozone Science & Engineering 29, 153–176.
553
Neyens, E., Baeyens, J., 2003. A review of classic fenton’s peroxidation as an advanced oxidation technique. Journal of Hazardous Materials 98, 33–50. Nogueira, R.F.P., Oliveira, M.C., Paterlini, W.C., 2005. Simple and fast spectrophotometric determination of H2O2 in photoFenton reactions using metavanadate. Talanta 66, 86–91. OECD Guidelines for testing of Chemicals, Simulation TestAerobic Sewage Treatment 303A, 1999. Papadaki, M., Emery, R.J., Abu-Hassan, M.A., Diaz-Bustos, A., Metcalfe, I.S., Mantzavinos, D., 2004. Sonocatalytic oxidation processes for the removal of contaminants containing aromatic rings from aqueous effluents. Separation and Purification Technology 34, 35–42. Pelegrini, R.T., Freire, R.S., Duran, N., Bertazzoli, R., 2001. Photoassisted electrochemical degradation of organic pollutants on a DSA type oxide electrode: process test for a phenol synthetic solution and its application for the E1 bleach kraft mill effluent. Environmental Science and Technology 35, 2849–2853. Petrovic, M., Gonzalez, S., Barcelo´, D., 2003. Analysis and removal of emerging contaminants in wastewater and drinking water. Trends in Analytical Chemistry 22, 685. Pignatello, E., Oliveros, A., McKay, 2006. Advanced oxidation processes for organic contaminant destruction based on the Fenton reaction and related chemistry. Critical Reviews in Environmental Science and Technology 36, 1–84. Prosen, H., Zupancˇicˇ-Kralj, L., 2005. Evaluation of photolysis and hydrolysis of atrazine and its first degradation product in the presence of humic acids. Environmental Pollution 133, 517–529. Radjenovic´, J., Petrovic´, M., Barceloc´, D., Petrovic´, M., 2007. Advanced mass spectrometric methods applied to the study of fate and removal of pharmaceuticals in wastewater treatment. Trends in Analytical Chemistry 26, 1132–1144. Richardson, S., 2007. Water analysis: emerging contaminants and current issues. Reviews in Analytical Chemistry 79, 4295–4324. 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 and Technology 38, 5476–5483. 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, 105–113. Saritha, P., Aparna, C., Himabindu, V., Anjaneyulu, Y., 2007. Comparison of various advanced oxidation processes for the degradation of 4-chloro-2-nitrophenol. Journal of Hazardous Materials 149, 609–614. United Book Press Inc.. In: Clesceri, L.S., Greenberg, A.E., Eaton, A. D. (Eds.), Standard Methods for the Examination of Water and Wastewater, 20th ed. American Public Health Association, American Water Works Association, Water Environment Federation, Baltimore, Maryland. Ternes, T., 1998. Occurrence of drugs in German sewage treatment plants and rivers. Water Research 32, 3245–3260. Ternes, T., Struber, J., Herrmann, N., McDowell, D., Ried, A., Kampmann, M., Teiser, B., 2003. Ozonation: a tool for removal of pharmaceuticals contrast media and musk fragrances from wastewater. Water Research 37, 1976–1982. Ternes, T., Bonerz, M., Herrmann, N., Teiser, B., Andersen, R., 2007. Irrigation of treated wastewaters in Braunschweig, Germany: an option to remove pharmaceuticals and musk fragrances. Chemosphere 66, 894–904. Teske, S.S., Arnold, R.G., 2008. Removal of natural and xenoestrogens during conventional wastewater treatment. Reviews in Environmental Science and Biotechnology 7, 107–124. Zapata, A., Oller, I., Bizani, E., Sa´nchez-Pe´rez, J.A., Maldonado, I., Malato, S., 2009. Evaluation of operational parameters involved in solar photo-Fenton degradation of a commercial pesticide mixture. Catalysis Today 144, 94–99.
554
water research 44 (2010) 545–554
Zepp, R.G., Faust, B.C., Hoigne´, J., 1992. Hydroxyl radical formation in aqueous reactions pH(3–8) of iron(II) with hydrogen peroxide: the photo-fenton reaction. Environmental Science and Technology 26, 313–319. Zhao, J.-L., Ying, G.-G., Wang, L., Yang, J.-F., Yang, X.-B., Yang, L.-H., Li, X., 2009. Determination of phenolic endocrine disrupting chemicals and acidic pharmaceuticals in surface water of the
Pearl River in South China by GC-MS. Science of the Total Environment 407, 962–974. Zhou, M., Dai, Q., Lei, L., Ma, C., Wang, D., 2005. Long life modified lead dioxide anode for organic wastewater treatment: electrochemical characteristics and degradation mechanism. Environmental Science and Technology 39, 363–370.
water research 44 (2010) 555–566
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Oxidative transformation of micropollutants during municipal wastewater treatment: Comparison of kinetic aspects of selective (chlorine, chlorine dioxide, ferrateVI, and ozone) and non-selective oxidants (hydroxyl radical) Yunho Lee a, Urs von Gunten a,b,* a b
Eawag, Swiss Federal Institute of Aquatic Science and Technology, Ueberlandstrasse 133, P.O. Box 611, 8600 Duebendorf, Switzerland Institute of Biogeochemistry and Pollutant Dynamics, ETH Zurich, 8092 Zurich, Switzerland
article info
abstract
Article history:
Chemical oxidation processes have been widely applied to water treatment and may serve
Received 1 October 2009
as a tool to minimize the release of micropollutants (e.g. pharmaceuticals and endocrine
Received in revised form
disruptors) from municipal wastewater effluents into the aquatic environment. The
13 November 2009
potential of several oxidants for the transformation of selected micropollutants such as
Accepted 23 November 2009
atenolol, carbamazepine, 17a-ethinylestradiol (EE2), ibuprofen, and sulfamethoxazole was
Available online 27 November 2009
assessed and compared. The oxidants include chlorine, chlorine dioxide, ferrateVI, and ozone as selective oxidants versus hydroxyl radicals as non-selective oxidant. Second-
Keywords:
order rate constants (k) for the reaction of each oxidant show that the selective oxidants
Oxidation processes
react only with some electron-rich organic moieties (ERMs), such as phenols, anilines,
Ozone
olefins, and deprotonated-amines. In contrast, hydroxyl radicals show a nearly diffusion-
Chlorine
controlled reactivity with almost all organic moieties (k 109 M1 s1). Due to a competi-
Chlorine dioxide
tion for oxidants between a target micropollutant and wastewater matrix (i.e. effluent
FerrateVI
organic matter, EfOM), a higher reaction rate with a target micropollutant does not
Hydroxyl radical
necessarily translate into more efficient transformation. For example, transformation efficiencies of EE2, a phenolic micropollutant, in a selected wastewater effluent at pH 8 varied only within a factor of 7 among the selective oxidants, even though the corresponding k for the reaction of each selective oxidant with EE2 varied over four orders of magnitude. In addition, for the selective oxidants, the competition disappears rapidly after the ERMs present in EfOM are consumed. In contrast, for hydroxyl radicals, the competition remains practically the same during the entire oxidation. Therefore, for a given oxidant dose, the selective oxidants were more efficient than hydroxyl radicals for transforming ERMs-containing micropollutants, while hydroxyl radicals are capable of transforming micropollutants even without ERMs. Besides EfOM, ammonia, nitrite, and bromide were found to affect the micropollutant transformation efficiency during chlorine or ozone treatment. ª 2009 Elsevier Ltd. All rights reserved.
* Corresponding author at: Eawag, Swiss Federal Institute of Aquatic Science and Technology, Ueberlandstrasse 133, P.O. Box 611, 8600 Duebendorf, Switzerland. Tel.: þ41 44 823 5270; fax: þ41 44 823 5028. E-mail address:
[email protected] (U. von Gunten). 0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2009.11.045
556
1.
water research 44 (2010) 555–566
Introduction
Effluents of municipal wastewater treatment plants (WWTPs) have been identified as a major source of micropollutants, such as hormones, pharmaceuticals, and personal care products (Ternes and Joss, 2006). To minimize a discharge of micropollutants from WWTPs to the receiving waters and thus to prevent adverse ecological effects, there have been discussions on upgrading WWTPs with additional treatment steps such as activated carbon and ozonation (Nowotny et al., 2007; Hollender et al., 2009). This is also desirable for the quality of drinking water if the wastewater-impacted receiving waters are used as a drinking water source via indirect potable reuse. The focus of the current study is on the chemical oxidation. Chemical oxidants, such as chlorine, chlorine dioxide, ozone, and hydroxyl radicals, have been widely applied for disinfection of drinking waters and wastewaters and often for the transformation/elimination of undesired micropollutants from drinking water (Letterman, 1999). Chlorine is worldwide one of the most commonly used disinfectant for drinking water and wastewater treatment (Letterman, 1999). Chlorine dioxide has been used as an alternative disinfectant for chlorine, often to minimize the formation of chlorine-based disinfection by-products (e.g. trihalomethane) (Gates, 1998; Letterman, 1999). The application of ozone in drinking water treatment is also widespread and the main reasons for its use are disinfection and oxidation (e.g. micropollutants transformation) or a combination of both (Hoigne´, 1998; von Gunten, 2003a). Even though chlorine, chlorine dioxide, and ozone have been widely used for disinfection (Letterman, 1999), their potentials for transforming micropollutants in wastewaters have received attention only in recent years (Ternes and Joss, 2006). In addition, ferrateVI is an emerging water treatment chemical due to its dual functions as an oxidant and a subsequent coagulant/precipitant as ferric hydroxide (Lee et al., 2009). Finally, the hydroxyl radical is a primary oxidant in advanced oxidation processes (AOPs) such as UV/H2O2 and O3/H2O2 and their application are considered for transforming various micropollutants in drinking waters as well as wastewaters for water reuse (Letterman, 1999; Asano et al., 2007). The hydroxyl radical is quite different from other oxidants with respect to its high reactivity and thus low selectivity (Buxton et al., 1988). In this paper, chlorine, chlorine dioxide, ferrateVI and ozone will be referred as selective oxidants as opposed to non-selective hydroxyl radicals. The oxidants mentioned above can achieve a transformation of various micropollutants during wastewater treatment. The transformation efficiency mainly depends on: (i) the reactivity of the oxidant to target micropollutants; and (ii) towards matrix components present in water (e.g. dissolved organic matter, DOM) which determines the stability of the oxidant. In general, the reactions of the oxidants follow second-order kinetics, i.e. the reaction rate is proportional to the concentrations of both reactants. The corresponding second-order reaction rate constants (k) range from 1 to 1010 M1 s1 in aqueous solution (Buxton et al., 1988; Deborde and von Gunten, 2008; Lee et al., 2009; Neta et al., 1988; von
Gunten, 2003a). Currently, k values are available for a wide range of organic compounds including many micropollutants (Fig. 1 and Table 1 as examples). For ozone reactions more than 500 k values have been measured (Neta et al., 1988 von Gunten, 2003a), and a few thousand k values are available for hydroxyl radical reactions (Buxton et al., 1988). There are also a few hundred k values available for chlorine, chlorine dioxide, and ferrateVI (Deborde and von Gunten, 2008; Lee et al., 2009; Neta et al., 1988). These second order rate constants show that the selective oxidants react preferentially with electron-rich organic moieties (ERMs), such as activated aromatic compounds (i.e. phenol, aniline, and polycyclic aromatics), organosulfur compounds, and deprotonated amines. Additionally, ozone and ferrateVI react with olefins. Contrary to the selective oxidants, hydroxyl radicals react with almost all organic moieties (e.g. aliphatic CeH bond) with nearly diffusion-controlled rates (i.e. k 109 M1 s1). Therefore, micropollutants containing the ERMs mentioned above can be potentially transformed by the selective oxidants. On the contrary, hydroxyl radicals can be effective for transforming any type of micropollutant. However, a high reactivity of the oxidants to the ERMs is also responsible for a rapid oxidant consumption by the ERMs included in the matrix components, such as DOM. In addition to DOM, some inorganic species such as ammonia, nitrite, sulfide, iron(II), and manganese(II) can be potential consumers of the selective oxidants. In the case of hydroxyl radicals, their rapid consumption by almost all types of matrix components can lead to a significantly lower transformation efficiency of micropollutants than the selective oxidants. Therefore, the kinetic information (i.e. k values) of the oxidants with the ERMs and some other moieties is crucial to assess the transformation efficiency of structurallydiverse micropollutants by a given oxidant in a real wastewater matrix. The aim of this study was to compare each of the selective oxidants and non-selective hydroxyl radicals with respect to their efficiency for transforming micropollutants during the treatment of the wastewater effluents. This comparison is important to provide criteria for choosing an appropriate oxidation process for a specific class of micropollutants based on the chemical structure (i.e. presence of ERM). For this purpose, the following issues were addressed: The k of each oxidant with selected organic model compounds and micropollutants (containing various ERMs) available in literature were summarized and compared. The oxidant consumption kinetics in a wastewater effluent were determined. The oxidative transformations of selected micropollutants containing various ERMs were conducted in a wastewater effluent for varying oxidant doses (0–150 mM). The transformation efficiency was then compared for the oxidants based on the required oxidant dose to achieve a certain degree of a micropollutant transformation. The effect of ammonia, nitrite, and bromide was investigated in respect to the transformation efficiency of selected micropollutants during the treatment of a wastewater effluent by the selected oxidants and hydroxyl radicals.
557
water research 44 (2010) 555–566
a
phenol 1010
c
aniline
HO .
butenol (olefin)
HO .
HO .
O3
108
k, M-1 s-1
b
106
O3 ClO2
ClO2
104
O3
HOCl HFeO4–
102
HOCl
HFeO4–
HFeO4–
100
d
glycine (1o amine) 1010
e
trimethylamine (3o amine)
f
HO .
HO .
HO .
108
k, M-1 s-1
dimethylamine (2o amine)
106
HOCl
104
O3
102
O3
O3
ClO2
HOCl
HFeO4–
HOCl
ClO2
HFeO4–
100 5
6
7
8
9
10
5
6
pH
7
8
9
10
pH
5
6
7
8
9
10
pH
Fig. 1 – pH dependent second-order rate constants (k) for the reaction of the oxidants, chlorine (HOCl), chlorine dioxide (ClO2), ferrateVI (HFeOL 4 ), hydroxyl radicals (HO ), and ozone (O3) with (a) phenol, (b) aniline, (c) butenol, (d) glycine, (e) dimethylamine, and (f) trimethylamine. k values for chlorine: Deborde and von Gunten, (2008); for chlorine dioxide and ozone: Neta et al., (1988); for ferrateVI: Lee et al. (2005a, 2008, 2009); and for hydroxyl radicals: Buxton et al., (1988). Speciesspecific second-order rate constants and the pKa values of the compounds used to generate the data are provided in Supplementary Information, SI-excel.
2.
Materials and methods
2.1.
Reagents and wastewater
Atenolol (ATL), carbamazepine (CBZ), 17a-ethinylestradiol (EE2), ibuprofen (IBP), and sulfamethoxazole (SMX) were obtained from Sigma–Aldrich with a purity higher than 98%. Stock solutions of these pharmaceuticals were prepared with Milli-Q purified water (Barenstad). Other chemicals and solvents were purchased from various commercial suppliers and used as received. A 10-L grab sample of secondary wastewater effluent (RDWW) was obtained from the Regensdorf wastewater treatment plant near Zurich, Switzerland. The sample was transported to the laboratory within several hours of sampling and vacuum-filtered with a 0.45 mm cellulose nitrate membrane prior to storage at 4 C. RDWW has a DOC of 5.0 mg C L1, 1.2 mM ammonia (17 mg N L1), 6 mM nitrite (84 mg N L1), 5.3 mM carbonate alkalinity, 0.35 mM (28 mg L1) of Br and a pH of w8. The concentrations of selected micropollutants (ATL, CBZ, EE2, IBP, and SMX) originally present in RDWW
were negligible relative to the concentrations spiked for oxidation experiments, which were 0.2–1 mM for each substrate.
2.2.
Oxidants
Stock solutions of chlorine (5–20 mM) were prepared by diluting a commercial solution of sodium hypochlorite (10% active chlorine, Riedel-deHaen, Germany). Chlorine dioxide (w5 mM) was produced by mixing potassium peroxodisulfate (K2S2O8) with sodium chlorite (NaClO2) according to a method described by Gates (1998). Ozone was produced with a Fischer 502 ozone generator and its stock solutions (w1.5 mM) were produced by sparging ozone-containing oxygen through MilliQ water that was cooled in an ice bath (Bader and Hoigne´, 1981). Potassium ferrateVI of high purity (w90%) was prepared by the method of Thompson et al. (1951). Stock solutions of ferrateVI (2 mM) were freshly prepared by dissolving solid samples of potassium ferrateVI (K2FeO4) in Milli-Q water and subsequent filtration (pH z 9.2). Hydroxyl radicals were in situ generated by photolysis of H2O2 (0.2–5 mM). Irradiations
558
water research 44 (2010) 555–566
Table 1 – Second-order rate constants for the reaction of oxidants with selected pharmaceuticals and inorganic species (ammonia, nitrite, and bromide).a Compounds
pKa
Oxidant
10.4
HO O3 ClO2 HFeO 4 HOCl
17a-Ethinylestradiol (EE2), synthetic steroid estrogen 5.7
HO O3 ClO2 HFeO 4 HOCl
Sulfamethoxazole (SMX), sulfonamide antibiotic
k (species of compound), M1 s1
k at pH 8, M1 s1
Ref.
9.8 109 1.8 105 (neutral), 3.7 109 (anion) 4.6 108 (anion) 9.4 102 (neutral), 5.4 105 (anion) 3.5 105 (anion)
9.8 109 1.5 107
Huber et al. (2003) Deborde et al. (2005)
1.8 106 4.2 102
Huber et al. (2005) Lee et al. (2005a)
3.4 102
Deborde et al. (2004)
5.5 109 4.7 104 (neutral), 5.7 105 (anion) 7.9 103 (anion) 9.2 103b 1.1 103 (neutral), 2.4 103 (anion)
5.5 109 5.7 105
Huber et al. (2003) Dodd et al. (2006)
7.9 103 95 5.7 102
Huber et al. (2005) Lee et al. (2009) Dodd and Huang (2004)
–
HO O3 ClO2 HFeO 4 HOCl
8.8 109 3.0 105 <0.1c 1.1 102 <0.1c
8.8 109 3.0 105 <0.1 16 <0.1
Huber et al. (2003) Huber et al. (2003) This study Lee et al. (2009) This study (for CBZ-HOCl)
9.6
HO O3
8.0 109 1.6 104
Benner et al. (2008) Benner et al. (2008)
ClO2 HFeO 4 HOCl
8.0 109 6.3 105 (neutral) w50 (neutral)c w2 103 (neutral)c w6 106 (neutral)c
w1 w7 w3.5 104
This study This study This study
4.9
HO O3 ClO2 HFeO 4 HOCl
7.4 109 9.6 (anion) <0.1 (anion)c <0.1 (anion)c <0.1 (anion)c
7.4 109 9.6 <0.1 <0.1 <0.1
Huber et al. (2003) Huber et al. (2003) This study This study This study
9.3
HO O3 ClO2 HFeO 4
4.3 106 1 <0.1 w0.4
Buxton et al. (1988) Neta et al. (1988) Neta et al. (1988) Sharma et al. (1998)
HOCl
9 107 20 (neutral) <0.1 (neutral) w1.5 (cation), w20 (neutral)d 3 106 (neutral)
3.4 104
Deborde and von Gunten (2008)
HO O3 ClO2 HFeO 4
1 1010 (anion) 3.7 105 (anion) 1.1 102 (anion) 1.2 102 (anion)d
1 1010 3.7 105 1.1 102 16
HOCl
2.3 1012 [Hþ]e
w6 103
Buxton et al. (1988) Neta et al. (1988) Neta et al. (1988) Lee and von Gunten (in preparation) Johnson and Margerum (1991)
Carbamazepine (CBZ), antiepileptic drug
Atenolol (ATL), b-blocker
Ibuprofen (IBP), antiphlogistic NHþ 4 /NH3 ammonia
HNO2/NO 2 nitric acid/nitrite
3.1
559
water research 44 (2010) 555–566
Table 1 (continued) Compounds
Br bromide
pKa
Oxidant
k (species of compound), M1 s1
k at pH 8, M1 s1
–
HO O3 ClO2 HFeO 4
w1.1 109 2.6 102 <0.1 <0.1
w1.1 109 2.6 102 <0.1 <0.1
HOCl
1.6 103
3.7 102
Ref.
von Gunten (2003b) Liu et al. (2001) Neta et al. (1988) Lee and von Gunten (in preparation) Kumar and Margerum (1987)
a pH-dependent k are provided in Supplementary Information, SI-excel. b k represents the second-order rate constant for the reaction between the fully protonated species of the Fe(VI)-SMX intermediate and SMX, for details refer Ref. Lee et al. (2009). c Estimated from the k with the corresponding ERM (Fig. 1) and the transformations in RDWW (Fig. 3). d The species specific rate constants were re-calculated in this study from the original k values of Sharma et al. (1998), see also Supplementary Information, SI-excel. e For details see Ref. Johnson and Margerum (1991), see also Supplementary Information, SI-excel.
were performed in quartz tubes using a merry-go-round photoreactor (Hans Mangels GmbH, Bornheim-Roisdor, Germany), equipped with a low pressure mercury lamp (Heraues Noblelight model TNN 15/32, nominal power 15 W) emitting essentially monochromatic light at l ¼ 254 nm. In the case of SMX, a medium pressure mercury lamp (Heraues Noblelight model TQ 718, nominal power 500 W) with a 0.25 M sodium nitrate filter solution producing light at l > 320 nm was used to minimize the direct photo-transformation of SMX (less than 10%). Fluence and fluence rate were determined by chemical actinometry at low optical density using 5 mM aqueous atrazine as an actinometer described by Meunier et al. (2006). Stock solutions of each oxidant were standardized spectrophotometrically based on their molar absorption coefficient: 3 ¼ 350 M1 cm1 at 292 nm for OCl (Johnson and Margerum, 1991), 3 ¼ 1230 M1 cm1 at 359 nm for ClO2 (Gates, 1998), 3 ¼ 3000 M1 cm1 at 260 nm for O3 (Liu et al., 2001), 3 ¼ 1150 M1 cm1 at 510 nm for FeO2 4 (Lee et al., 2005b), and 3 ¼ 40 M1 cm1 at 240 nm for H2O2 (Bader et al., 1988).
each oxidant (<1 mL) were added under vigorous stirring to a series of identical 20 mL reaction solutions. After 1 h, 1 mL of the reaction solution was sampled and quenched with 50 mL of a thiosulfate solution (100 mM) and analyzed for residual micropollutant concentrations by using a HPLC/UV system. For ferrateVI experiments at doses higher than 40 mM, the treated samples were filtered through a 0.45 mm PTFE filter (BGB Analytik AG, Switzerland) to remove iron-precipitates. Losses of micropollutants during the filtration were negligible. In the case of hydroxyl radicals, which were generated in situ in RDWW, a series of reaction solutions (20 mL) spiked with H2O2 (0.2–0.6 mM for the low pressure lamp and 2 mM for the medium pressure lamp) and a micropollutant (or parachlorobenzoic acid, pCBA as hydroxyl radical probe compound) was irradiated for different time periods ranging from 0.5 to 20 min for the low pressure lamp and from 2.5 to 60 min for the medium pressure lamp, respectively. At proper time intervals, 1 mL of the reaction solution was sampled for the analysis of residual H2O2 and micropollutants (or pCBA). Each experiment was conducted at least in duplicate.
2.3. Oxidant consumption kinetics and transformation of selected micropollutants in wastewater Oxidant consumption experiments for the selective oxidants were performed in a secondary wastewater effluent (RDWW) at pH 8 (20 mM borate buffer) in 50 mL reaction volume. Forty to 45 mM of each oxidant was applied to RDWW and 0.2–1 mL of the reaction solution was sampled for the measurements of residual oxidants over a 1 h reaction time. Transformation of the selected micropollutants was tested in RDWW spiked with each micropollutant and treated by each oxidant at various oxidant doses (0–150 mM) at 24 1 C and pH 8. The micropollutants, ATL (amine), CBZ (olefin), EE2 (phenol), and SMX (aniline), were selected as representative substances with different ERMs that belong to different usage classes and have an environmental relevance (Ternes and Joss, 2006). IBP was chosen as a model micropollutant without the ERMs. Each micropollutant was spiked separately at concentrations of 0.2–1 mM into 20 mL of the wastewater buffered at pH 8. For the selective oxidants, stock solutions of
2.4. Quantification of hydroxyl radical doses in wastewater To calculate the formation and consumption (or scavenging) rate of hydroxyl radicals in a wastewater effluent, the method developed by Zepp and co-workers (Zepp et al., 1987) was used. The method is based on a decrease of a trace hydroxyl radical probe compound in a system of interest that contains a known concentration of external hydroxyl radical scavenger. In this study, 0.5 mM of para-chlorobenzoate (pCBA, kOH þ pCBA ¼ 5.2 109 M1 s1 (Buxton et al., 1988)) and 125– 1000 mM of tertiary-butanol (tBuOH, k OH þ tBuOH ¼ 6.0 108 M1 s1 (Buxton et al., 1988)) were employed as the hydroxyl radical probe and scavenger, respectively. Applications of this method provided hydroxyl radical formation rates as a function of initial hydrogen peroxide concentrations during a UV254 nm/H2O2 treatment of RDWW. This relationship was used to quantify hydroxyl radical doses at a given UV254 nm
560
water research 44 (2010) 555–566
Oxidant concentrations in wastewater effluents were measured by several colorimetric methods. Chlorine and chlorine dioxide were determined by the ABTS method developed by Pinkernell et al. (2000); ozone by the indigo method (Bader and Hoigne´, 1981); ferrateVI by the ABTS method described by Lee et al. (2005b); H2O2 by the DPD/ peroxidase method (Bader et al., 1988). Micropollutant (or pCBA) concentrations at sub to low mM range were determined with an Agilent 1100 series HPLC system equipped with a Nucleosil 100-5 C18 column (125 4 or 250 4 mm) and a UV (diode array) and/or fluorescence detector. Separation was achieved by different gradient programs with the aqueous phase consisting either of 10 mM phosphoric acid or 10 mM citric acid (pH 5) and the solvent phase either of acetonitrile or methanol. The injection volume was 100 mL and the flow rate was set to 1 mL min1. The repeatability was determined to be 4% while limit of quantifications (LOQ) were inferred from the lowest standard concentrations used. The limits of quantification were in all cases sufficient to determine a transformation of micropollutants of more than 96% (1.4 logtransformation).
In the pH range 5–10, hydroxyl radicals react with phenol, aniline, and butenol with nearly diffusion-controlled rates. However, the k values for the reaction with three amine compounds are one or two orders of magnitude lower (107– 108 M1 s1 at pH 6–8). This is due to the relative low reactivity of hydroxyl radicals with the protonated-amine moieties. Nevertheless, the k of hydroxyl radicals with the amine compounds with longer hydrocarbon-chains can be higher than 109 M1 s1. This is due to the high reactivity of hydroxyl radical with CeH bonds (k [ 108 M1 s1) (Buxton et al., 1988). In addition, hydroxyl radicals react with benzenes (i.e. non-activated aromatics) with k > 109 M1 s1 (Buxton et al., 1988). Ozone reacts fast with all ERMs shown in Fig. 1: the k values decrease in the order of aniline phenol > butenol dimethylamine z trimethylamine > glycine in the pH range 6–8. Chlorine dioxide shows a high reactivity only with phenol, aniline, and trimethylamine. Chlorine is reactive to all ERMs except butenol. The k values decrease in the order of glycine > dimethylamine aniline > phenol z trimethylamine in the pH range 6–8. Finally, ferrateVI reacts with all ERMs except trimethylamine. The k values decrease in the order of aniline > phenol z glycine > butenol dimethylamine in the pH range 6–8. For the selective oxidants, ozone shows the widest range of k values (i.e. w7 orders of magnitude) while ferrateVI shows the narrowest range of k values (i.e. w4 orders of magnitude) toward the selected ERMs in the pH range 6–8.
3.
3.2. Consumption kinetics of the selective oxidants in a wastewater effluent
irradiation time. For details of the kinetic method, see Supplementary Information, SI-word.
2.5.
Analytical methods
Results and discussion
3.1. Rate constants of oxidants with selected organic compounds Fig. 1 summarizes the pH-dependent k for the reaction of the five oxidants (chlorine, chlorine dioxide, ferrateVI, ozone, and hydroxyl radical) with the selected ERM-containing model compounds, such as phenol and aniline (activated aromatic compounds), butenol (olefin), glycine (primary (1 ) amine), dimethylamine (secondary (2 ) amine), and trimethylamine (tertiary (3 ) amine). If the k value is lower than 1 M1 s1, it is not shown in Fig. 1. The species-specific k values which were used to generate Fig. 1 are provided in Supplementary Information, SIexcel. For more details on the kinetics of each oxidant, refer the corresponding references provided in SI-excel. The k values for phenol and aniline, as two aromatic compounds, decrease in the order of hydroxyl radical > ozone > chlorine dioxide > chlorine z ferrateVI (Figs. 1a,b). The k values for the other compounds show a slightly different reactivity order. Chlorine and chlorine dioxide are not reactive to butenol, therefore the reactivity trend for butenol is hydroxyl radical > ozone > ferrateVI (Fig. 1c). For three amine compounds, chlorine shows a particular high reactivity whereas chlorine dioxide has a low reactivity. For glycine as 1 amine, the k values decrease in the order of hydroxyl radical > chlorine > ozone > ferrateVI > chlorine dioxide (Fig. 1d). For dimethylamine as 2 amine, the reactivity order is hydroxyl radical > ozone z chlorine > ferrateVI > chlorine dioxide (Fig. 1e). Finally, for trimethylamine as 3 amine, the reactivity order is hydroxyl radical > ozone > chlorine dioxide chlorine > ferrateVI (Fig. 1f).
Fig. 2 shows the consumption kinetics of the selective oxidants (ozone, ferrateVI, chlorine, and chlorine dioxide) in a secondary wastewater effluent from Regensdorf, Switzerland (RDWW) at pH 8 for an oxidant dose of 40–45 mM. The selected RDWW has a low concentration of inorganic nitrogen species ([ammonia] ¼ 1.2 mM and [nitrite] ¼ 6 mM), therefore the consumption of each oxidant is mainly caused by EfOM ([DOC] ¼ 5 mg C L1). Except ferrateVI, all oxidants showed fast consumptions within 2 min after the oxidant addition. The stability of the oxidants follows the order of ferrateVI z chlorine dioxide chlorine [ ozone. Ozone was depleted in less than 2 min (Fig. 2a, inset), whereas the other oxidants were slowly consumed within 60 min. The observed low stability of ozone is consistent with its high reactivity to all ERMs shown in Fig. 1. Chlorine and chlorine dioxide showed a similar decay pattern: a rapid decrease within 2 min after the oxidant addition (termed ‘initial phase’), followed by a slow decrease over 60 min of reaction time (termed ‘second phase’). Based on the k values shown in Fig. 1, the chlorine consumption in the initial phase is expected to be mainly caused by aniline- and 1 & 2 aminemoieties while in the second phase phenolic- and 3 aminemoieties become important. In the case of chlorine dioxide, the consumption in the initial phase might be mainly caused by phenolic- and aniline-moieties while in the second phase by 2 & 3 amine-moieties. Contrary to chlorine and chlorine dioxide, ferrateVI decreased smoothly along the reaction time, which is consistent with its narrow distribution of k values with the ERMs shown in Fig. 1. The achieved oxidant exposures for the conditions shown in Fig. 2 were 4.5 102 M1
561
water research 44 (2010) 555–566
a
b
50 40
ferrateVI
50
Concentration, µM
Concentration, µM
ozone 30 20 10
40 30 20 10 0 0.0
0.5
1.0
1.5
2.0
2.5
3.0
Time, min
0 0
c
10
20
30
40
50
60
50
10
20
30
d
chlorine Concentration, µM
0
40
50
60
chlorine dioxide
40 30 20 10 0 0
10
20
30
40
50
60
0
Time, min
10
20
30
40
50
60
Time, min
Fig. 2 – Consumption kinetics of the selective oxidants, (a) ozone, (b) ferrateVI, (c) chlorine, and (d) chlorine dioxide, in a secondary wastewater effluent (RDWW) at pH 8. Symbols represent measured data and lines connect each data point to show the trend.
s1 for ferrateVI and chlorine dioxide, 3.6 102 M1 s1 for chlorine, and 6.0 104 M1 s1 for ozone.
micropollutants, and will be discussed in the following sections.
3.3.1. 3.3. Transformation of selected micropollutants in a wastewater effluent: comparison between selective oxidants and non-selective hydroxyl radicals A comparison was made for the transformation efficiency of selected micropollutants in a secondary wastewater effluent (RDWW) by the selective oxidants (chlorine, chlorine dioxide, ferrateVI, and ozone) and the non-selective hydroxyl radicals. The selected micropollutants include compounds with ERMs, such as 17a-ethinylestradiol (EE2, phenol), sulfamethoxazole (SMX, aniline), carbamazepine (CBZ, olefin), and atenolol (ATL, 2 amine), and ibuprofen (IBP, no ERM). For the selective oxidants, the reaction time of 1 h was given to simulate realistic treatment conditions. Ozone was completely consumed after 1 h whereas for chlorine, chlorine dioxide, and ferrateVI, more than 10% of the oxidant concentration remained after 1 h if the oxidant doses were higher than 40 mM (Fig. 2). For hydroxyl radicals, the oxidant doses were quantified by the kinetic method described in the Supplementary Information, SI-word. Fig. 3 shows the logarithm of the relative residual concentration of each selected micropollutant during treatment of RDWW at pH 8 as a function of the oxidant doses. The transformation efficiency of micropollutants varied significantly depending on the types of the selective oxidants and
EE2 and SMX
For these two micropollutants containing activated aromatic moieties, all four selective oxidants were more efficient than hydroxyl radicals in terms of the oxidant doses required to achieve more than one-log (90%) transformation (Figs. 3a,b). For the selective oxidants, especially in the case of chlorine and ferrateVI, there was a lag-phase at lower oxidant doses where the micropollutant concentration decreased little with increasing oxidant doses. The low transformation efficiency in the lag-phase is attributed to the high competition for the selective oxidants between EE2/SMX and the ERMs of EfOM. With an increase of the selective oxidant doses higher than the initial consumption by ERMs, the residual concentration of EE2 and SMX started to decrease significantly. This can be understood by the depletion of the most reactive ERMs of EfOM and the corresponding smaller competition for EE2 and SMX toward the selective oxidants. For hydroxyl radicals, on the contrary, the log-relative residual concentration of EE2 and SMX was linearly proportional to the applied hydroxyl radical dose. This indicates that the magnitude of the competition for hydroxyl radicals between micropollutants and wastewater effluent matrix remained constant during the entire oxidation. Therefore, it can be concluded that the initial oxidation products of EfOM by hydroxyl radicals have similar k values for their reaction with hydroxyl radicals compared to
562
water research 44 (2010) 555–566
a
b
17a-ethinylestradiol (EE2)
0
c
Sulfamethoxazole (SMX)
0
Carbamazepine (CBZ)
0
log C/C0
ClO2 HOCl HO.
HO.
O3
-1
-1
HFeO4-
ClO2
-
HFeO4
HOCl
-2 0
20
60
80
100
HOCl
0
60
30
e
Atenolol (ATL)
0
90
120
150
0
30
60
90
120
150
Oxidant dose, µM
f
Ibuprofen (IBP)
0
HFeO4-
-2
Oxidant dose, µM
Oxidant dose, µM
d
O3
ClO2 O3
-2
40
HO.
-1
p-chlorobenzoate (pCBA)
0
log C/C0
ClO2 HOCl HFeO4HO.
-1
O3 + t-BuOH
ClO2 HOCl HFeO4HO.
-1
HO.
O3
-1
O3 -2 0
30
60
90
Oxidant dose, µM
120
150
-2 0
30
60
90
Oxidant dose, µM
120
150
-2 0
30
60
90
120
150
Oxidant dose, µM
Fig. 3 – Logarithm of the residual concentrations (log(c/c0)) of selected micropollutants as a function of oxidant doses in a secondary wastewater effluent (RDWW) at pH 8: (a) EE2, (b) SMX, (c) CBZ, (d) ATL, and (e) IBP. Symbols represent measured data and lines connect each data point to show the trend. The lines for hydroxyl radicals represent the linear regression of data. For the selective oxidants, the reaction time of 1 h was given to simulate realistic treatment conditions. Hydroxyl radicals were in situ produced by UV254 nm/H2O2 photolysis and their doses were quantified by the kinetic method described in the Supplementary Information, SI-word. For (e) IBP, ‘O3 D tBuOH’ indicates that the ozonation was conducted in presence of 60 mM tBuOH to exclude the contribution of hydroxyl radicals to the IBP transformation.
the parent EfOM. This leads to a constant consumption rate of hydroxyl radicals. Hence, selective oxidants can be much more efficient than hydroxyl radicals especially when a higher degree of transformation is required (e.g. >90%) for EE2 or SMX. The characteristics of transformations of EE2 and SMX by hydroxyl radicals (i.e. a linear log-transformations vs. hydroxyl radical doses) were also observed for CBZ, ATL, and IBP (Figs. 3c–e, respectively). For EE2 (Fig. 3a), the transformation efficiency to achieve an one-log transformation for the oxidants was in the order of chlorine dioxide (3 mM) > ozone z ferrateVI (10 mM) > chlorine (20 mM) > hydroxyl radical (50 mM) where the value in the parenthesis shows the required oxidant dose. The very high efficiency of chlorine dioxide for EE2 transformation can be understood by its high reactivity to phenolic moieties (Fig. 1a), but relative low reactivity to amine moieties (Figs. 1d–f). Since phenolic and amine moieties are expected to be the two main ERMs present in EfOM, chlorine dioxide can be selective for the transformation of the micropollutants containing phenolic moieties in presence of EfOM. In contrast, the relatively low efficiency of chlorine for EE2 transformation can be understood by its high reactivity to amine moieties (Figs. 1d–f), but low reactivity to phenolic moieties (Fig. 1a). The one-log transformation efficiency in the case of SMX (Fig. 3b) was in the order of chlorine dioxide (15 mM) z ozone (45 mM) z chlorine (48 mM) > hydroxyl (20 mM) > ferrateVI
radical (80 mM). The relative lower transformation of SMX compared to EE2 at the same oxidant dose (Figs. 3a vs. b) are consistent with the lower k values of SMX than EE2 for the reaction with the oxidants at pH 8 except the case of chlorine (Table 1). The sulfonyl-moiety deactivates the aniline-moiety of SMX toward oxidation as an electron-withdrawing group. Therefore, the k values for the reaction of SMX with the selective oxidants are lower than those for aniline (SI-excel). In contrast, the cyclic ring-moiety activates the phenolicmoiety of EE2 toward oxidation as an electron-donating group. Therefore, the k values for the reaction of EE2 with the selective oxidants are higher than those for phenol (SI-excel). The smaller transformation of SMX than EE2 at the same chlorine doses (Figs. 3a vs. b) is not consistent with the 1.7-fold higher k for the reaction of chlorine with SMX at pH 8 (Table 1). Dodd et al. found that the reaction of chlorine with SMX produced the N-chlorinated SMX (i.e. chlorine-adduct to the aniline-moiety of SMX) as an initial product (Dodd and Huang, 2004). The k values reported in Table 1 correspond to the rate constant of this initial reaction step. The N-chlorinated SMX can be reduced back to the parent, SMX, upon its reaction with reductants such as thiosulfate, which was employed as a quenching reagent in this study. Therefore, the actual extent of SMX transformation during chlorination could be smaller than the predictions based on the k values corresponding to the initial chlorine addition to SMX. Finally, even though the k
water research 44 (2010) 555–566
values of ozone or chlorine dioxide for its reaction with EE2 and SMX are several orders of magnitude higher than those of chlorine or ferrateVI (Table 1), the transformation efficiency in RDWW among the selective oxidants differed only within a factor of 20. This is caused by their competitive consumption by micropollutants and EfOM.
3.3.2.
CBZ
The one-log transformation efficiency in the case of CBZ (Fig. 3c) was in the order of ozone (30 mM) > ferrateVI (43 mM) > hydroxyl radical (100 mM) [ chlorine z chlorine dioxide. Chlorine and chlorine dioxide achieved little transformations of CBZ (<5%) even at an oxidant dose of 150 mM, which is consistent with the very low reactivity of these two oxidants towards the olefin-moieties (Fig. 1c). In addition, the relative lower transformations of CBZ compared to EE2 or SMX at the same oxidant dose of ozone or ferrateVI (Figs. 3a,b vs. c) are consistent with the lower k values of CBZ for the reaction with these oxidants (Table 1).
3.3.3.
ATL
In the case of ATL transformation (Fig. 3d), only ozone was more efficient than hydroxyl radicals. The required oxidant dose for one-log transformation was 40 and 85 mM for ozone and hydroxyl radicals, respectively. FerrateVI achieved 0.6-log (60%) transformation at its highest dose of 150 mM. Chlorine and chlorine dioxide showed little transformations (<16%) even at an oxidant dose of 150 mM. The poor transformation by chlorine is surprising because chlorine has a high reactivity toward amine-moieties (Figs. 1d–f). This is because, similar to the case of SMX chlorination, the intermediate formed from the fast reaction of chlorine with ATL (N-chlorinated ATL) is stable toward further degradations and can be transformed back to ATL by thiosulfate. According to Pinkston and Sedlak, the N-chloro ATL was degraded less than 10% even after 12 h of exposure to chlorine at pH 8 (Pinkston and Sedlak, 2004). Therefore, the transformation of ATL (and possibly other 2 amine-containing micropollutants) can be poor if a dechlorination by sulfite is applied for the chlorinated-wastewater effluents.
3.3.4.
IBP
Due to a very low reactivity of the selective oxidants to IBP (k < 10 M1 s1, Table 1) which contains no ERMs, hydroxyl radicals were most efficient in this case: more than one-log transformation at the oxidant doses higher than 75 mM (Fig. 3e). In addition, ozone achieved one-log transformation of IBP at its dose of w120 mM. Little transformations (<13%) of IBP were observed during ozonation in the presence of tBuOH as a hydroxyl radical scavenger (Fig. 3e). This indicates that the transformation of IBP during ozonation was primarily caused by hydroxyl radicals which are produced from the ozone decomposition (von Gunten, 2003a). The transformation of other ERM-containing micropollutants (i.e. EE2, SMX, CBZ, and ATL) are expected to be dominated by ozone rather than hydroxyl radicals especially at higher ozone doses. This is based on the observation that the variation of the logarithm of the relative micropollutant transformation as a function of ozone dose is similar to that of the other selective oxidants (Fig. 3).
3.4.
563
Efficiency of oxidants in real water matrices
Based on the kinetic information in Fig. 1 and the experimental results in Figs. 2 and 3, the efficiency of each oxidant to transform various micropollutants in real wastewaters can be predicted. Ozone can significantly transform micropollutants containing any ERMs shown in Fig. 1. However, its concurrent high reactivity to the ERMs present in EfOM will decrease the efficiency of micropollutant transformation. Chlorine dioxide can significantly transform phenolic- and aniline-containing micropollutants and moderately 3 amine-containing micropollutants. The consumption of chlorine dioxide by EfOMs should be smaller than ozone because only phenolic-, aniline-, and 3 amine-moieties of EfOM consume chlorine dioxide rapidly. Chlorine can transform micropollutants containing all ERMs except olefin-moieties. However, chlorine is not efficient to transform micropollutants containing 1 & 2 amine-moieties if dechlorination by sulfite is applied afterwards. Because chlorine is rapidly consumed by all ERMs except olefin-moieties present in EfOM, its efficiency can be moderate or low to transform micropollutants containing phenolic and aniline moieties. FerrateVI can transform micropollutants containing any ERMs shown in Fig. 1 except 3 amine-moieties. FerrateVI is more efficient to transform phenol-containing than olefin- and amine-containing compounds. Finally, hydroxyl radicals can transform almost any micropollutant based on their high second-order rate constants. However, their rapid consumption by any wastewater matrix component is an important factor, decreasing their efficiency to transform micropollutants considerably.
3.5. Assessing wastewater matrix effects on micropollutants transformation 3.5.1.
EfOM
Consumption of oxidants by EfOM is an important factor determining the transformation efficiency of micropollutants. In the case of the selective oxidants, the ERMs present in EfOM are the actual oxidant consumers. Currently, there is limited information available on the type and concentration of ERMs present in EfOM. Therefore, the consumption of the selective oxidants should be experimentally measured at different oxidant doses in a given water (e.g. Fig. 2). In the case of hydroxyl radicals, the carbon concentration of EfOM (i.e. DOC) have been known as a parameter representing the consumption rate of hydroxyl radicals. Based on previous studies and this study, the k values for the reaction of hydroxyl radicals with dissolved organic matter (including EfOMs) are narrowly distributed and centered at 2.5 104 L mg1 s1 (Hoigne´, 1998; Westerhoff et al., 2007; Zepp et al., 1987). In addition, the consumption rate of hydroxyl radicals by dissolved organic matter did not change along the course of the oxidation. Therefore, the consumption rate of hydroxyl radicals can be readily calculated from a carbon concentration of a given water and the k value of 2.5 104 L mg1 s1. However, a recent study reported that the k values for the reaction of hydroxyl radicals with non-isolated EfOMs are three to five times higher than the previous average value and vary significantly depending on the source water (Rosario-Ortiz et al., 2008), which encourages more studies on this subject.
564
3.5.2.
water research 44 (2010) 555–566
and Margerum, 1991). Reactions of these intermediates with EE2 might be responsible for the incomplete inhibition of EE2 transformation by chlorine at the high nitrite concentration. For ozone, the EE2 transformation decreased gradually from 80 to 20% with an increase of the nitrite concentration from 6 to 100 mM ([NO 2 ]0/[O3]0 ¼ from 0.8 to 12.5). Nitrite had little effect on the EE2 transformation in the case of chlorine dioxide and ferrateVI, which is consistent with the relatively low k values of these two oxidants with nitrite (Table 1). For hydroxyl radical, nitrite has a slightly positive effect on EE2 transformation even though it reacts rapidly with nitrite (OH þ NO 2 / OH þ NO2 , 10 1 1 k ¼ 10 M s , Table 1). In RDWW spiked with nitrite at the concentration of 100 mM, w80% of the produced hydroxyl radicals is scavenged by nitrite during the initial phase of the UV/H2O2 treatment. This is based on the relative reaction rate of hydroxyl radicals with nitrite vs. the remaining matrix components (i.e. EfOM, H2O2, and carbonate). Therefore, the slightly enhanced transformation of EE2 in presence of 100 mM nitrite might be explained by the reaction of EE2 with some intermediates from the reaction of hydroxyl radical with nitrite such as NO2 (i.e. EE2 þ NO2 / EE2oxid þ NO 2 ). The k value for the reaction of NO2 with EE2 is estimated to be w105 M1 s1 at pH 8 from the k value for the reaction of NO2 with 4-methylphenol (Neta et al., 1988).
Ammonia and nitrite
In addition to EfOMs, inorganic nitrogen compounds, such as ammonia and nitrite, commonly present in poorly-nitrified/ denitrified wastewater effluents can rapidly consume some oxidants and affect the transformation efficiency of micropollutants. The effect of ammonia and nitrite was investigated using EE2 as an example. From preliminary experiments, oxidant doses to achieve w80% transformation of EE2 in RDWW were determined. As a next step, the RDWW spiked with additional ammonia or nitrite (up to 100 mM) was treated with the pre-determined oxidant dose. In the case of the selective oxidants, the residual EE2 concentration was determined after 1 h of reaction time. Fig. 4a shows that in the case of chlorine, ammonia significantly decreased the degree of transformation of EE2. At an ammonia concentration higher than 20 mM ([NHþ 4 ]0/ [HOCl]0 1), the EE2 transformation was less than 10%. This is because chlorine reacts rapidly with ammonia, forming chloramines (Deborde and von Gunten, 2008). When the [NHþ 4 ]0/[HOCl]0 1, monochloramine (NH2Cl) is the predominant chloramine species which has a low reactivity to EE2 (k ¼ 0.24 M1 s1) (Lee et al., 2008). For the other selective oxidants and hydroxyl radicals, the EE2 transformation was not affected by the presence of ammonia within the tested concentration. This is consistent with the low reactivity of these oxidants with ammonia (Table 1). Fig. 4b shows that nitrite significantly reduced the transformation of EE2 in the case of chlorine and ozone, which is consistent with the high k values of these two oxidants with nitrite (Table 1). For chlorine, the EE2 transformation decreased from 80 to 40% with an increase of the nitrite concentration from 6 to 30 mM ([NO 2 ]0/[HOCl]0 ¼ from 0.3 to 1.5) and remained rather constant at w40% with a further increase of the nitrite concentration even up to 100 mM ([NO 2 ]0/[HOCl]0 ¼ 5). The oxidation of nitrite by chlorine to nitrate proceeds through some reactive intermediates such as NO2Cl and NOþ 2 (Johnson
3.5.3.
Bromide
Bromide is present in waters and wastewaters at concentrations of 10 to several hundred mg L1 (von Gunten, 2003b). Since the oxidation of bromide typically produces bromine, which is highly reactive to phenol- and amine-moieties, the presence of bromide during oxidative wastewater treatment can affect the transformation efficiency of micropollutants. Among the selective oxidants, only chlorine and ozone react with bromide (Table 1), generating bromine. Since bromine is about three orders of magnitude more reactive toward phenols than chlorine (Lee and von Gunten, 2009), transformation of −
[NO2 ]0/[O3]0 0
a
[NH4+]0/[HOCl]0 0
1
2
% transformation
100
3
4
5
b
2
4
6
8
10
−
[NO2 ]0/[HOCl]0 0
1
2
3
4
100
O3 ClO2 Fe(VI) HO .
12
5
HO . ClO2 Fe(VI)
80
80
60
60
40
40
HOCl
20
O3
HOCl
20
0
0 0
20
40
60 +
[NH4 ]0, µM
80
100
0
20
40
60
80
100
−
[NO2 ]0, µM
L Fig. 4 – Effect of (a) ammonia (NHD 4 ) and (b) nitrite (NO2 ) on the transformations of EE2 during treatment of a secondary wastewater effluent (RDWW) by different oxidants at pH 8. Preliminary experiments were conducted to determine the oxidant dose for each oxidant to achieve a 80% transformation of EE2 in RDWW without additionally spiked ammonia and nitrite. They were 20 mM for chlorine, 3 mM for chlorine dioxide, 8 mM for ozone, 8 mM for ferrateVI, and 37 mM for hydroxyl radicals. Symbols represent measured data and lines connect each data point to show the trend.
water research 44 (2010) 555–566
phenolic micropollutants can be significantly enhanced during chlorination of bromide-containing waters. A recent study showed that bromine produced from bromide was mainly responsible for 17a-ethinylestradiol transformation during chlorination of a wastewater effluent (Lee and von Gunten, 2009). In the case of ozone, most phenolic- and amine-moieties are already oxidized by ozone before the significant formation of bromine from bromide. Hence, the presence of bromide affects little the transformation efficiency of micropollutants during ozonation. However, an enhanced transformation of 1 amine-containing micropollutants is expected due to the relatively low reactivity of ozone versus high reactivity of bromine to 1 amines. For example, the k values at pH 7 for the reaction with glycine are 1.6 102 and 4.7 105 M1 s1 for ozone and bromine, respectively (SI-excel). In addition, like to the enhanced ammonia oxidation by ozone in the presence of bromide as a catalyst (Haag et al., 1984), N-brominated 1 amines can be oxidized by ozone with a faster rate than the parent 1 amines and the corresponding reaction regenerates bromide. Finally, bromide has only a small effect on the transformation efficiency of micropollutants during advanced oxidation processes. This is due to (i) the low reactivity of hydroxyl radicals with bromide (OH þ Br / OH þ Br, k ¼ 1.1 109 M1 s1, Table 1) which results in the formation of HOBr and in the case of H2O2-based AOPs (ii) the fast reaction of HOBr with H2O2 leading to bromide (von Gunten, 2003b). Formation of potentially carcinogenic bromate during ozonation of bromide-containing waters is one of drawbacks of ozonation. In recent two studies, only low concentration of bromate formation (i.e. <10 mg L1) were observed during ozonation of wastewater effluents at typical ozone doses (i.e. <1 mg L1 of ozone per mg L1 of DOC) (Wert et al., 2007; Hollender et al., 2009). The low bromate formation can be explained by relative low bromide levels (16–31 and 210–280 mg L1, respectively) or high dissolved organic matter concentrations (4.2–6.0 and 7.1–7.2 mg C L1, respectively) of the investigated wastewaters.
4.
Conclusions
In this study, the potential of various oxidants for their transformation of micropollutants were assessed and compared during treatment of wastewater effluents. The oxidants include chlorine, chlorine dioxide, ferrateVI, and ozone as the selective oxidants and hydroxyl radicals as the non-selective oxidant.
565
Therefore, competition for oxidants between target micropollutants and EfOM determines the transformation efficiency. The competition depends on the relative reaction rate of a given oxidant with ERMs present in a target micropollutant and the EfOM. Accordingly, a higher rate constant of an oxidant with a target micropollutant does not necessarily translate into more efficient transformation. For the selective oxidants, the competition disappears rapidly after the ERMs present in EfOM are consumed. In contrast, for hydroxyl radicals, the competition remains practically the same during the entire oxidation process. Therefore, the efficiency of hydroxyl radicals is much lower than that of the selective oxidants for transforming micropollutants containing ERMs. In addition, the difference in transformation efficiency becomes larger if higher extents of transformations of micropollutants should be achieved. Ammonia and nitrite can significantly decrease transformation efficiency of micropollutants (i.e. phenolic- and aniline containing) during chlorination. Nitrite can also decrease transformation efficiency during ozonation. Therefore, in poorly nitrified or -denitrified wastewaters, transformation of micropollutants can be low during a treatment with chlorine or ozone. In contrast, bromide can significantly increase transformation efficiency of phenolicmicropollutants by forming bromine during chlorination. In addition, an enhanced transformation of 1 amine-containing micropollutants is expected during ozonation of bromide-containing waters.
Acknowledgements This study was funded by the Swiss Federal Offices for the Environment (Contract No. 07.0142.PJ/I232-2755). The authors would like to thank E. Salhi and H. Laubscher for laboratory assistance and S. Canonica for support with the photochemical experiments.
Appendix A. Supplemental material Supplementary information for this manuscript can be downloaded at doi:10.1016/j.watres.2009.11.045.
references The selective oxidants react only with some electron-rich organic moieties (ERMs), such as phenols, anilines, olefins, and amines, with the exception of the following reactions: chlorine and chlorine dioxide with olefins, chlorine dioxide with 1 and 2 amines, and ferrateVI with 3 amines show a negligible reactivity (k < 1 M1 s1). In contrast, hydroxyl radicals show a very high reactivity (k 108 M1 s1) with almost all organic moieties, even including CeH bonds. Therefore, hydroxyl radicals can transform any type of micropollutant with a similar efficiency (e.g. within a factor of 4 for those having k 2 109 M1 s1). Effluent organic matter (EfOM) as a major wastewater matrix component contains ERMs and thus consumes the oxidants.
Asano, T., Burton, F.L., Leverenz, H.L., Tsuchihashi, R., Tchobanoglous, G., 2007. Water Reuse: Issues, Technologies, and Applications, first ed. McGraw-Hill, New York. Bader, H., Hoigne´, J., 1981. Determination of ozone in water by the indigo method. Water Res. 15, 449–456. Bader, H., Sturzenegger, V., Hoigne´, J., 1988. Photometric method for the determination of low concentrations of hydrogen peroxide by the peroxidase catalyzed oxidation of N, Ndiethyl- p-phenylenediamine (DPD). Water Res. 22, 1109–1115. Benner, J., Salhi, E., Ternes, T., von Gunten, U., 2008. Ozonation of reverse osmosis concentrate: kinetics and efficiency of beta blocker oxidation. Water Res. 41, 2481–2490.
566
water research 44 (2010) 555–566
Buxton, G.V., Greenstock, C.L., Helman, W.P., Ross, W.P., 1988. Critical review of rate constants for reactions of hydrated electrons, hydrogen atoms and hydroxyl radicals (OH/O) in aqueous solution. J. Phys. Chem. Ref. Data 17, 513–886. Deborde, M., von Gunten, U., 2008. Reactions of chlorine with inorganic and organic compounds during water treatmentdkinetics and mechanisms: a critical review. Water Res. 42, 13–51. Deborde, M., Rabouan, S., Gallard, H., Legube, B., 2004. Aqueous chlorination kinetics of some endocrine disruptors. Environ. Sci. Technol 38, 5577–5583. Deborde, M., Rabouan, S., Duguet, J.P., Legube, B., 2005. Kinetics of aqueous ozone-induced oxidation of some endocrine disruptors. Environ. Sci. Technol 39, 6086–6092. Dodd, M.C., Huang, C.H., 2004. Transformation of the antibacterial agent sulfamethoxazole in reactions with chlorine: kinetics, mechanisms, and pathways. Environ. Sci. Technol 38, 5607–5615. 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. Environ. Sci. Technol 40, 1969–1977. Gates, D., 1998. The Chlorine Dioxide Handbook. American Water Works Association, Denver. Haag, W.R., Hoigne´, J., Bader, H., 1984. Improved ammonia oxidation by ozone in the presence of bromide ion during water treatment. Water Res. 18, 1125–1128. Hoigne´, J., 1998. Chemistry of aqueous ozone and transformation of pollutants by ozonation and advanced oxidation processes. In: Hrubec, J. (Ed.), The Handbook of Environmental Chemistry. Springer-Verlag, Berlin Heidelberg, pp. 83–141. Hollender, J., Zimmermann, S.G., Koepke, S., Krauss, M., McArdell, C.S., Ort, C., Singer, H., von Gunten, U., Siegrist, H., 2009. Elimination of organic micropollutants in a municipal wastewater treatment plant upgraded with a full-scale postozonation followed by sand filtration. Environ. Sci. Technol 43, 7862–7869. Huber, M.M., Canonica, S., Park, G.Y., von Gunten, U., 2003. Oxidation of pharmaceuticals during ozonation and advanced oxidation processes. Environ. Sci. Technol 37, 1016–1024. Huber, M.M., Korhonen, S., Ternes, T.A., von Gunten, U., 2005. Oxidation of pharmaceuticals during water treatment with chlorine dioxide. Water Res. 39, 3607–3617. Johnson, D.W., Margerum, D.W., 1991. Non-metal redox kinetics: a reexamination of the mechanism of the reaction between hypochlorite and nitrite ions. Inorg. Chem. 30, 4845–4851. Kumar, K., Margerum, D.W., 1987. Kinetics and mechanisms of general-acid-assisted oxidation of bromide by hypochlorite and hypochlorous acid. Inorg. Chem. 26, 2706–2711. Lee, Y., Yoon, J., von Gunten, U., 2005a. Kinetics of the oxidation of phenols and phenolic endocrine disruptors during water treatment with ferrate (Fe(VI)). Environ. Sci. Technol 39, 8978–8984. Lee, Y., Yoon, J., von Gunten, U., 2005b. Spectrophotometric determination of ferrate (Fe(VI)) in water by ABTS. Water Res. 39, 1946–1953. Lee, Y., Escher, B.I., von Gunten, U., 2008. Efficient removal of estrogenic activity during oxidative treatment of waters containing steroid estrogens. Environ. Sci. Technol 42, 6333–6339.
Lee, Y., von Gunten, U., 2009. Transformation of 17aethinylestradiol during water chlorination: effects of bromide on kinetics, products, and transformation pathways. Environ. Sci. Technol 43, 480–487. Lee, Y., von Gunten, U. Kinetics of ferrateVI reactions with inorganic compounds in water. in preparation. Lee, Y., Zimmermann, S.G., Kieu, A.T., von Gunten, U., 2009. Ferrate (Fe(VI)) application for municipal wastewater treatment: a novel process. Environ. Sci. Technol 43, 3831–3838. Letterman, R.D., 1999. Water Quality and Treatment, fifth ed. McGraw-Hill, New York. Liu, Q., Schurter, L.M., Muller, C.E., Aloisio, S., Francisco, J.S., Margerum, D.W., 2001. Kinetics and mechanisms of aqueous ozone reactions with bromide, sulfite, hydrogen sulfite, iodide, and nitrite ions. Inorg. Chem. 40, 4436–4442. Meunier, L., Canonica, S., von Gunten, U., 2006. Implications of sequential use of UV and ozone for drinking water quality. Water Res. 40, 1864–1876. Neta, P., Huie, E., Ross, A.B., 1988. Rate constants for reactions of inorganic radicals in aqueous solution. J. Phys. Chem. Ref. Data 17, 1027–1264. Nowotny, N., Epp, B., von Sonntag, C., Fahlenkamp, H., 2007. Quantification and modelling of the elimination behavior of ecologically problematic wastewater micropollutants by adsorption on powdered and granulated activated carbon. Environ. Sci. Technol 41, 2050–2055. Pinkernell, U., Nowack, B., Gallard, H., von Gunten, U., 2000. Methods for the photometric determination of reactive bromine and chlorine species with ABTS. Water Res. 34, 4343–4350. Pinkston, K.E., Sedlak, D.L., 2004. Transformation of aromatic ether- and amine-containing pharmaceuticals during chlorine disinfection. Environ. Sci. Technol 38, 4019–4025. Rosario-Ortiz, F.L., Mezyk, S.P., Doud, D.F.R., Snyder, S.A., 2008. Quantitative correlation of absolute hydroxyl radical rate constants with non-isolated effluent organic matter bulk properties in water. Environ. Sci. Technol 42, 5924–5930. Sharma, V.K., Bloom, J.T., Joshi, V.N., 1998. Oxidation of ammonia by ferrate(VI). J. Environ. Sci. Health A33, 635–650. Ternes, T.A., Joss, A., 2006. Human Pharmaceuticals, Hormones and Fragrances. The Challenge of Micropollutants in Urban Water Management. IWA Publishing, London, New York. Thompson, G.W., Ockerman, G.W., Schreyer, J.M., 1951. Preparation and purification of potassium ferrate. VI. J. Am. Chem. Soc. 73, 1379–1381. von Gunten, U., 2003a. Ozonation of drinking water: Part I. Oxidation kinetics and product formation. Water Res. 37, 1443–1467. von Gunten, U., 2003b. Ozonation of drinking water: Part II. Disinfection and by-products formation in presence of bromide, iodide or chlorine. Water Res. 37, 1469–1487. Wert, E.C., Rosario-Ortiz, F.L., Drury, D.D., Snyder, S.A., 2007. Formation of oxidation byproducts from ozonation of wastewater. Water Res. 41, 1481–1490. 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. Environ. Sci. Technol 41, 4640–4646. Zepp, R.G., Hoigne´, J., Bader, H., 1987. Nitrate-induced photooxidation of trace organic chemicals in water. Environ. Sci. Technol 21, 443–450.
water research 44 (2010) 567–577
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
A review of phytoestrogens: Their occurrence and fate in the environment Ze-hua Liua,b,*, Yoshinori Kanjoa, Satoshi Mizutania a
Department of Urban Engineering, Graduate School of Engineering, Osaka City University, 3-3-138 Sugimoto, Sumiyoshi-ku, Osaka 558-8585, Japan b Department of Chemistry and Chemical Engineering, Shanghai University of Engineering Science, 333 Longteng Road, Songjiang, Shanghai 201620, PR China
article info
abstract
Article history:
Phytoestrogens are plant compounds with estrogenic activities. Many edible plants, some
Received 11 December 2008
of which are common in the human diet, are rich in phytoestrogens. Almost all phytoes-
Received in revised form
trogens eaten daily by people were reported partly recovered in urine or feces, which can
12 March 2009
be regarded as one of the main sources of their occurrence in municipal wastewaters. As
Accepted 16 March 2009
they may act as one part of the endocrine disrupting compounds (EDCs) in water systems,
Available online 24 March 2009
some phytoestrogens have been monitored and detected in wastewater and other various environments. It is very difficult to monitor numerous unknown EDCs in complex waste-
Keywords:
water samples, and it is helpful if some estimation of target EDCs can be done before
Phytoestrogens
monitoring. With this in mind, this review will: (1) summarize estrogenic activities or
Endocrine disrupting compounds
estrogenic potencies of phytoestrogens by different bioassays; (2) summarize daily urinary
Wastewater
excretion rates of phytoestrogens by humans, and compare their urinary excretion rates to
Urinary excretion rate
that of estrone, which suggests that most phytoestrogens may occur in municipal waste-
Monitoring
waters; (3) collect and summarize published data on the occurrence and fate of phytoestrogens in various environments. ª 2009 Elsevier Ltd. All rights reserved.
1.
Introduction
Endocrine disrupting compounds (EDCs) are chemicals with the potential to elicit negative effects on the endocrine systems of humans and wildlife. They include a broad class of chemicals such as natural estrogens/androgens, synthetic estrogens/androgens, industrial chemicals as well as phytoestrogens (Liu et al., 2009). As EDCs exist in extremely low concentrations (mg/L or ng/L) in the environment, one of the key points for the research is sensitive and accurate measurement of EDCs. Nowadays, two categories of methods are available; chemical analyses such as LC–MS/MS, GC–MS as well as HPLC and bioassays (in vitro and in vivo). Chemical
analyses can measure precise concentrations of known target EDCs in environmental samples, while bioassays provide direct information of the estrogenic activity of complex mixtures of EDCs that are likely to occur in water irrespective of the causative compounds. For large-scale monitoring research, bioassays (in vitro) seem most appropriate due to their low cost and high throughput capability. However, for the complexity of environmental samples (usually wastewater samples), values measured by different bioassays on the same samples differed greatly, which have lead to questions over the validity and effectiveness of bioassays for environmental samples. To find widely acceptable standard bioassays for environmental samples, the chemically derived
* Corresponding author: Department of Urban Engineering, Graduate School of Engineering, Osaka City University, 3-3-138 Sugimoto, Sumiyoshi-ku, Osaka 558-8585, Japan. Tel./fax: þ81 6 6605 3048. E-mail address:
[email protected] (Z.-h. Liu). 0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2009.03.025
568
water research 44 (2010) 567–577
values are often compared to those of bioassays, which accelerates monitoring a multitude of endocrine-like chemicals in environmental samples (Furuichi et al., 2004; Nakada et al., 2004; Tan, 2006; Nelson et al., 2007; Salste et al., 2007). In Fernandez et al. (2007), as many as 30 EDCs have been monitored in 9 wastewater treatment plants in Canada and the results of chemically derived estrogen equivalents (EEQ) were also compared with those measured by recombinant yeast assay (RYA). The monitoring of numerous EDCs in wastewater is difficult, and would be preferable if some guidelines were applicable to wastewater in which EDCs may exist in high values, while other EDCs may not. Phytoestrogens are defined as any plant compounds structurally and/or functionally similar to ovarian and placental estrogens and their active metabolites (Whitten and Patisaul, 2001). Phytoestrogens with rather high concentrations widely exist in numerous plants such as soybeans, fruits, cabbages and so on (Liggin et al., 2000; US Department of Agriculture, USDA, 2003). Some of these are widelyconsumed foods, especially in Asia, where high amounts of soy are consumed in direct or indirect ways (Adlercreutz et al., 1991; Nagata et al., 2007). Although most research on phytoestrogens has suggested that they yield positive effects on human health (Setchell and Cassidy, 1999; Lissin and Cook, 2000; Kris-Etherton et al., 2002), some research denoted that most phytoestrogens also acted as endocrine disruptors, which could cause intersex in aquatic organisms (Kiparissis et al., 2003; Tzchori et al., 2004). In addition, genotoxicity of phytoestrogens was reviewed by Stopper et al. (2005), who pointed out some possible adverse effects of phytoestrogens on humans. Due to the possible negative effects mentioned above, much attention has been paid to the occurrence and fate of phytoestrogens by environmental researchers in municipal/ industrial wastewater or surface waters, in which they were detected with high frequency (Kiparissis et al., 2001; Lagana et al., 2004; Bacaloni et al., 2005; Erbs et al., 2007). Liu et al. (in press) detected 15 EDCs at two municipal wastewater treatment plants, in which the concentrations of daidzein in influents were from 3900 to 12,000 ng/L and its chemically derived estrogenic equivalents (EEQ) were 0.9–5.3% of those measured by the ER-binding assay (bisphenol A (BPA) and nonylphenol (NP) totally contributed 0.7–4.1% in the same samples), indicating a relatively high proportion in wastewater which was not negligible. Like natural estrogens or androgens, the main source of phytoestrogens in municipal wastewaters can also be attributed to daily excretions of urine or feces by humans. Therefore, a thorough understanding of phytoestrogens from daily excretions by humans is helpful for estimating their possible concentrations in municipal wastewater. This estimation has been applied to natural estrogens (Johnson et al., 2000; Johnson and Williams, 2004), and it is also helpful for suggesting possible priority phytoestrogens when monitoring for numerous EDCs in wastewater samples is carried out. Although there is much data on the excretion rates of phytoestrogens from human urine, few conclusions have been drawn for the purpose above. In addition, in terms of convenience to the comparison of chemical analyses with bioassays, relative binding affinities (RBA) or estrogenic/ androgenic potencies are also important.
The objectives of this review are: (1) to summarize reported estrogenic/androgenic activities of phytoestrogens by different bioassays; (2) to summarize the daily urinary excretion rates of phytoestrogens by humans; and (3) to summarize reported data on the occurrence and fate of phytoestrogens in the various environments.
2. Estrogenic/androgenic activities of phytoestrogens Phytoestrogens are a large family with two major classes, flavonoids and lignans. The number of flavonoids is greater than the number of lignans. Flavonoids also possess higher relative estrogenic activities than those in lignans (Whitten and Patisaul, 2001). One interest for researchers of environmental engineering relating to EDCs is the estrogenic/androgenic activities of phytoestrogens. There is much available data on their estrogenic/androgenic activities by different bioassays. To give an overview of their estrogenic/androgenic activities, some published data are outlined in Table 1, in which the estrogen standard substance is 17b-estradiol (E2) and the androgen standard substance is testosterone (Te) or methyltrienolone (R1881). From Table 1, it is easily seen that almost all phytoestrogens are far weaker estrogenic substances, except in the transient gene expression assay (TGEA), in which the estrogenic potencies of some phytoestrogens are at the same level, or even stronger than that of E2. Compared to the estrogenic activities of industrial chemicals BPA or NP (Blair et al., 2000; Gutendorf and Westendorf, 2001; Liu et al., 2009), most phytoestrogens are in the same order of magnitude, which denote that the estrogenic activities derived from phytoestrogens can not be neglected when they remain in high concentrations in water samples. In addition, some phytoestrogens also possess androgenic activities (Table 1). With the RBA or estrogenic potency in Table 1, we can calculate the contribution ratio of phytoestrogens (Ratio) to values measured by bioassays using Eq. (1) and the contribution ratio of phytoestrogens (Ratioc) to the chemically derived EEQ using Eq. (2): Ratio ¼
EEQc ¼ EEQb
P
RBAi ðEPi Þ Ci EEQb
(1)
where EEQc and EEQb are the chemically calculated EEQ and the measured EEQ by bioassay, respectively; Ci represents the concentration of an individual EDC in wastewater sample or the urinary excretion rate of an individual phytoestrogen from human and RBAi(EP)i is its corresponding estrogenic activity. Ratioc ¼
EEQi RBAi ðEPi Þ Ci ¼P EEQc RBAi ðEPi Þ Ci
(2)
where EEQi is the chemically derived EEQ of an individual phytoestrogen in urine or an individual EDC in wastewater. To monitor phytoestrogens in complex wastewater or other water samples, it is very important to estimate their concentrations possibly existing in environmental samples. For phytoestrogens in municipal wastewater, one of the main sources can be regarded as human urine or feces. Therefore, their daily urinary excretion rates were concluded as shown below.
Table 1 – Estrogenic/androgenic activities of phytoestrogens by different bioassays*. Class
Subclass
Phytoestrogen
CAS
RBA (ER) a;b
hER(a)
Flavonoids
Isoflavone
Isoflavanones Isoflavans Coumestans Flavones
Flavanones Chalcone Phytosterols Lignans –
– Mycoestrogens
hER(b)
a
hAR(a)
Estrogenic potency (EP)
rAR(a)
c
a
RGA (ER(a))
YESd (ER(a))
TGEAb ER(a)
ER(b)
Daidzein Dihydrodaidzein Formononetin Genistein Biochanin A Glycitein ODMA Equol Coumestrol Flavone Apigenin 7-hydroxyflavone Chrysin Tangeretin Kaemperide Kaempferol Quercetin Naringenin Phloretin b-Sitosterol
486-66-8 17238-05-0 485-72-3 446-72-0 491-80-5 40957-83-3 21255-69-6 531-95-3 479-13-0 525-82-6 520-36-5 6665-86-7 480-40-0 481-53-8 491-54-3 520-18-3 117-39-5 480-41-1 60-82-2 83-46-5
1.8e-3; 1e-3 –; – 2.65e-4; <1e-4 1.2e-3; 4e-2 1.26e-5; <1e-4 2.8e-4i –; – 8.6e-3j 2.64e-3; 0.2 N.B; <1e-4 1.79e-4; 3e-3 N.B; – N.B; <1e-4 –; – N.B; – 2.9e-4; 1e-3 –; 1e-4 5.33e-5; 1e-4 1.54e-4; 2e-3 8.75e-4g
5e-3 – <1e-4 0.87 <1e-4 – – 0.02j 1.4 <1e-4 0.06 – <1e-4 – – 0.03 4e-4 1.1e-4 7e-3 1.6e-2g
N.B – – N.B – – – – – – 1.56e-4 – – – – – – – – –
– – – 3.6e-5 – – – 4.1e-5 N.B 4.0e-5 – N.B – – – – – – – N.B
6.67e-5 – 2.94e-5 4.17e-4 9.09e-6 – – 6.1e-4f 5e-4 – 1.75e-5 2.27e-6 5.88e-6 – 1.09e-6 8.33e-6 – 6.67e-6 2.04e-6 –
3e-6 – – 3e-3 <3e-6 <3e-7 – 3e-5 1e-3 <3e-6 3e-5 – – – – <3e-6 <3e-6 1e-5 <3e-6 <3e-6
0.97 – 0.06 1.98 0.36 – – – 1.02 0.02 0.5 – 0.01 – – 0.35 0.03 0.36 0.49 –
0.8 – 0.02 1.82 0.53 – – – 0.98 0.02 0.49 – 0.02 – – 0.53 0.02 0.45 0.10 –
Enterolactone Enterodiol Zearalenone Zearalanone a-Zearalanol b-Zearalanol a-Zearalenol b-Zearalenol
78473-71-9 80226-00-2 17924-92-4 5975-78-0 26538-44-3 42422-68-4 36455-72-8 71030-11-0
3.2e-5; 6.5e-4e –; – 2.83e-2; 7e-2 3.22e-2; – –; – 0.176; – 0.523; 0.48k 1.68e-2; 0.13k
– 0.05 – – – – –
– – 7.63e-5 – – 1.22e-3 – 6.13e-4
– – – – – – – –
– – 1.56e-2 3.12e-2 – 1.16e-2 1 4.35e-4
– – 7.5e-2e 1.5e-2h 3.3e-2h 2.6e-3h 5.5e-2h 2.6e-3h
– – 0.91 – – – – –
– – 0.27 – – – – –
569
* RBA, relative binding affinity by estrogen receptor (ER)/androgen receptor (AR) ligand competitive binding assay; hER, human estrogen receptor; hAR, human androgen receptor; rAR, rat androgen receptor; RGA, reporter gene assay; TGEA, transient gene expression assay; –, not available; N.B., not binding. a National Institute of Technology and Evaluation, Japan (NITE,2008). b Kuiper et al., 1998. c Fang et al., 2003. d Nishihara et al., 2000. e Mueller et al., 2004. f Markiewicz et al., 1993. g Gutendorf and Westendorf, 2001. h Bovee et al., 2004. i Song et al., 1999. j Kostelac et al., 2003. k Matthews et al., 2000.
water research 44 (2010) 567–577
Flavonols
RBA (AR) b
570
water research 44 (2010) 567–577
Table 2 – Daily excretion rates of phytoestrogens from human urine. Phytoestrogens
Genistein
Gender condition (number)
Men (267) Women (250) Women (19) Women (268) Women (18)
Men (49) Women (49) Men (2) Women (3) Women (18) Daidzein
Men (267) Women (250) Women (19) Women (18)
Men (49) Women (49) Men (2) Women (3) Women (18) Women (41)
Diet
Daily excretion rates (average)
Mean EEQ
mg/d
mgE2/dc
Normala Normal Soy protein diets Normal Normal 7.1 mg total isoflavones/d 65 mg total isoflavones/d 132 mg isoflavones/d Normal Normal Normal Normal (2 vegetarians) Normal
n.d.–7579.4 (132.4) –b (2172.7) 1810.6–2837.5 (2357.8) 31.5–1251.3 (219.8) 183.5–280.8 (227) 362.4–492.9 (422.7) 1521.1–2096.2 (1785.7) 3045.8–4144.6 (3553.1) – (78.4) – (45.9) 368.6–587.2 (477.9) 482.6–3653.3 (1749.9) 10.8–281.0 (48.6)
0.159 2.607 2.829 0.264 0.272 0.507 2.143 4.264 0.094 0.055 0.573 2.100 0.058
Normal Normal Soy protein diets Normal 7.1 mg total isoflavones/d 65 mg total isoflavones/d 132 mg isoflavones/d Normal Normal Normal Normal (2 vegetarians) Normal Normal
n.d.–4309.8 (227.0) – (4390.7) 2517.0–5008.5 (3686.4) 200.8–323.4 (254.7) 303.8–523.5 (398.9) 1281.6–2255.8 (1700.3) 2722.6–4692.5 (3574.3) – (211.0) – (94.1) 1074.2–1113.6 (1093.9) 1240.4–5865.3 (3624.1) 28.0–2333.9 (180.5) 56.1–322.9 (124.2)
0.409 7.903 6.636 0.458 0.718 3.061 6.434 0.380 0.169 1.969 6.523 0.325 0.224
Reference
Low et al., 2005 Dai et al., 2002 d Lampe et al., 2001 Tonkelaar et al., 2001d Xu et al., 2000
Lampe et al., 1999 Adlercreutz et al., 1995 e Adlercreutz et al., 1995 Lampe et al., 1994 Low et al., 2005 Dai et al., 2002 Lampe et al., 2001 Xu et al., 2000
Lampe et al., 1999 Adlercreutz et al., 1995 Adlercreutz et al., 1995 Lampe et al., 1994 Adlercreutz et al., 1986
Dihydrodaidzein
Women (250) Women (18)
Normal Normal 7.1 mg total isoflavones/d 65 mg total isoflavones/d 132 mg isoflavones/d
– (1340.2) 151.2–222.7 (183.5) 221.7–339.5 (274.4) 853.1–1334.1 (1066.8) 1528.8–2341.4 (1891.9)
– – – – –
Dai et al., 2002 Xu et al., 2000
ODMA
Men (267) Women (250) Women (19) women (18)
Normal Normal Soy protein diets Normal 7.1 mg total isoflavones/d 65 mg total isoflavones/d 132 mg isoflavones/d Normal Normal (2 vegetarians) Normal
n.d.–381.9 (48.1) – (1299.1) 206.6–1859.5 (707.0) 142.6–214.4 (174.8) 206.4–297.8 (247.9) 1045.2–1344.5 (1265.5) 2236.9–3227.8 (2687.3) 22.8–335.2 (179.0) 171.0–627.1 (379.6) 6.5–27.4 (12.7)
– – – – – – – – – –
Low et al., 2005 Dai et al., 2002 Lampe et al., 2001 Xu et al., 2000
Normal Soy protein diet Normal 7.1 mg total isoflavones/d 65 mg total isoflavones/d 132 mg isoflavones/d Normal Normal Normal Normal (2 vegetarians) Normal Normal
n.d.–1280.5 (18.2) n.d.–581.4 (271.0) 45.1–64.4 (53.8) 61.3–120.4 (85.8) 126.9–249.5 (178.1) 189.5–372.4 (265.8) – (21.8) – (17.0) 26.6–117.5 (72.1) 16.9–60.8 (37.7) 14.5–43.6 (26.6) 15.5–24.7 (20.4)
0.157 2.331 0.447 0.738 1.532 2.286 0.187 0.146 0.620 0.324 0.229 0.175
Men (2) Women (3) Women (41) Equol
Men (267) Women (19) Women (18)
Men (49) Women (49) Men (2) Women (3) Women (18) Women (41)
Adlercreutz et al., 1995 Adlercreutz et al., 1995 Adlercreutz et al., 1986 Low et al., 2005 Lampe et al., 2001 Xu et al., 2000
Lampe et al., 1999 Adlercreutz et al., 1995 Adlercreutz et al., 1995 Lampe et al., 1994 Adlercreutz et al., 1986
571
water research 44 (2010) 567–577
Table 2 (continued) Phytoestrogens
Gender condition (number)
Diet
Daily excretion rates (average)
Mean EEQ
mg/d
mgE2/dc
Reference
Glycitein
Men (267) Women (250) Women (18)
Normal Normal Normal 7.1 mg total isoflavones/d 65 mg total isoflavones/d 132 mg isoflavones/d
n.d.–556.2 (51.3) – (690.8) 73.9–94.1 (83.3) 109.4–141.3 (120.8) 149.0–292.8 (209.9) 222.3–436.9 (311.8)
0.014 0.193 0.023 0.034 0.059 0.087
Low et al., 2005 Dai et al., 2002 Xu et al., 2000
Coumestrol
Women (18)
Normal 7.1 mg total isoflavones/d 65 mg total isoflavones/d 132 mg isoflavones/d
8.3–9.5 (8.9) 7.9–9.2 (8.6) 8.4–9.7 (9.0) 8.8–10.2 (9.5)
0.023 0.023 0.024 0.025
Xu et al., 2000
Enterolactone
Men (267) Women (250) Women (19) Women (268) Women (18)
n.d.–13218.5 (1020.4) – (1873.5) 477.3–2386.7 (1555.1) 19.0–6157.9 (1276.2) 364.3–417.1 (389.6) 347–489.6 (412.3) 599.6–858.3 (717.2) 960.6–1352.0 (1139.6) – (683.2) – (584.7) 426.9–1592.8 (1009.9) 1305.2–2488.7 (1851.2) 611.6–1244.0 (914.3)
0.033 0.060 0.050 0.041 0.012 0.013 0.023 0.036 0.022 0.019 0.032 0.059 0.029
Low et al., 2005 Dai et al., 2002 Lampe et al., 2001 Tonkelaar et al., 2001 Xu et al., 2000
Men (49) Women (49) Men (2) Women (3) Women (41)
Normal Normal Soy protein diets Normal Normal 7.1 mg total isoflavones/d 65 mg total isoflavones/d 132 mg isoflavones/d Normal Normal Normal Normal (2 vegetarians) Normal
Adlercreutz et al., 1995 Adlercreutz et al., 1995 Adlercreutz et al., 1986
Naringenin
Women (250) Men and women (77)
Normal Normal
– (1217.0) – (653.8)
0.065 0.035
Dai et al., 2002 Nielsen et al., 2002
Quercetin
Men and women (77)
Normal
– (23.4)
–
Nielsen et al., 2002
Phloretin
Men and women (77)
Normal
– (84.8)
0.013
Nielsen et al., 2002
Kaempferol
Men and women (77)
Normal
– (51.3)
0.015
Nielsen et al., 2002
Lampe et al., 1999
a Normal diet. b Not available. c Calculated as EEQi in Eq. (2), values of RBA based on ER-binding assay (NITE, 2008). d Transferred from nmol/mg creatinine or mmol/mol creatinine to mg/d, and creatinine excretion rate of 1 g/d for women was chosen (Jackson, 1966). e Calculated from their free phytoestrogens and their conjugates.
3. Occurrence and fate of phytoestrogens in the environment 3.1.
Phytoestrogens in human urine
Phytoestrogens in human urine are from the diet or dietary metabolites, and their concentrations in urine have been found dose dependent on their consumption (Karr et al., 1997; Zhang et al., 1999; Xu et al., 2000; Peeters et al., 2003). For disease related items, there has been much research on the daily excretion rates of urinary phytoestrogens, in which the excretion rates varied greatly with race and dietary customs (Horn-Ross et al., 1997; Seow et al., 1998). To give an overview, published data on urinary excretion rates are outlined in Table 2. In order to compare the estrogenic potencies of phytoestrogens from human urine, the authors
converted the urinary excretion rate of the individual phytoestrogens into a corresponding value, calculated as EEQi in the numerators of Eq. (2); we label it ‘‘excretion rate of EEQ’’. As is shown in Table 2, out of 12 phytoestrogens, genistein and daidzein are the two most important phytoestrogens evaluated by both urinary excretion rate and excretion rate of EEQ; followed by equol, for which its urinary excretion rates by normal diets are low but with a relatively high corresponding excretion rate of EEQ. Enterolactone and naringenin are the phytoestrogens with high urinary excretion rates, but very low estrogenic potencies (low RBA, Table 1). Therefore, the corresponding excretion rates of EEQ by enterolactone and naringenin seem relatively low. Data of urinary excretion rates of some phytoestrogens, especially for quercetin, phloretin as well as kaempferol, is very limited. For better comparison, it would be helpful if more data were available.
572
water research 44 (2010) 567–577
In addition, biochanin A and formononetin have been found existing in high concentrations in red clovers of Canada, and have also been detected in many fruits, beans and vegetables (Mazur et al., 1996; Sivesind and Seguin, 2005; Kuhnle et al., 2007, 2008). Few data on urinary excretion rates are available. As a reference, one published paper showed the average fecal excretion rate was about 1.5 mg/d for biochannin A and 11.9 mg/d for formononetin (Jenner et al., 2005). A phytosterol b-sitosterol with estrogenic activity (Table 1) was found rich in a wide range of fruits, vegetables, nuts, and many kinds of seed oils (Grosso et al., 2000; Beveridge et al., 2002, 2005; Iwatsuki et al., 2003; Holser et al., 2004; Phillips et al., 2005; Jimenez-Escrig et al., 2006; Zhang et al., 2006). Although there is no available data on the urinary excretion rate of b-sitosterol, very high human fecal excretion rates of 52–322 mg/d and a high human fecal concentration of 121 mg/g in dry feces (average) were reported (Eneroth et al., 1964; Leeming et al., 1996), which suggested that b-sitosterol might also exist in wastewater with a very high concentration. From the view of statistics, a rough estimation on the arithmetic average of urinary excretion rates was done on the data of normal diets in Table 2, which was also applied on human urinary excretion rates of natural estrogens and androgens (Liu et al., submitted for publication). If the loss of phytoestrogens during the course of flow to wastewater treatment plants is low, their concentrations in wastewater can be compared to those of natural estrogens. Although it is well known that 17b-estradiol (E2) is the strongest natural estrogen, estrone (E1) is the one which was detected frequently with high concentrations in effluent of WWTPs (Baronti et al., 2000; Joss et al., 2004; Johnson et al., 2007; Liu et al., 2009). Therefore, the average urinary excretion rates of phytoestrogens were compared to that of E1 as shown in Fig. 1, and the urinary excretion rate of E1 was modified from Liu et al. (submitted for publication), for which the urinary excretion rates of two pregnant women out of 177 persons (11.3&) were also included. From Fig. 1 it is known that the urinary excretion rates of phytoestrogens are 1–86 times that of E1, except for coumestrol (0.45 times); Among phytoestrogens in Table 2, genistein and daidzein, frequently detected in wastewaters, are 35 and 86 times that of E1,
urinary excretion rates(µg/d)
10000
1000
100
10
G
en
E1 ist e D in D ai ih d yd ze ro da in id ze in O D M A G ly ci te in E Co qu um ol En est te rol rla c N tone ar in g Q erin ue rc e Ph tin lo re tin K ae m pe ro l
1
Fig. 1 – Comparison of urinary excretion rates between phytoestrogens and E1.
respectively. According to published results, the average concentrations of genistein and daidzein in influent of one WWTP in Italy were about 3 and 9 times that of the detected E1 (Lagana et al., 2004), while the concentrations of daidzein in 12 influent samples of two municipal WWTPs in Japan showed about 8–46 times (22 times on average) that of the detected E1 (Liu et al., in press). Compared to detected concentrations for genistein and daidzein, their estimated concentrations in Fig. 1 are about 10 times and 2–10 times higher. The large differences may result from the different loss rates before they reach WWTPs, different excretion ratios of urine and feces for phytoestrogens and E1, as well as regional dietary customs. Although coumestrol is the leastpresent urinary excretion phytoestrogen in Fig 1, it was also reported existing in municipal WWTPs (Bacaloni et al., 2005). Based on this fact, the other 11 phytoestrogens in Fig. 1 are very likely to be present in municipal wastewater. Based on EEQ, phytoestrogens such as daidzein, genistein, equol and glycitein should not be neglected, for which the excretion rates of EEQ are about 35%, 10%, 2% and 1% that of human urinary E1. As enterlactone and naringerin are much weaker estrogens than daidzein (Table 1), although their urinary excretion rates are on the same level as daidzein, the corresponding excretion rates of EEQ are just about 0.5% and 0.7% that of human urinary E1. However, the urinary excretion rates of enterlactone in some regions may be much higher than that in Fig. 1 (Lampe et al., 1994; Wang, 2002; Milder et al., 2005; Valentin-Blasini et al., 2005; Penalvo et al., 2008), which may yield a higher contribution ratio in wastewater. In addition, the urinary excretion rates of dihydrodaidzein and ODMA are almost the same as that of daidzein, but the data on their estrogenic activities is not available; therefore, their corresponding excretion rates of EEQ can not be applied here.
3.2. Phytoestrogens in wastewaters and the other water bodies As predicted above that most phytoestrogens may exist in wastewater, some phytoestrogens have been proven to not only exist in wastewater but also in surface waters (Kawanishi et al., 2004; Lagana et al., 2004; Bacaloni et al., 2005; Erbs et al., 2007). Mycoestrogens (Table 1) are zearalenone and its derivatives. They are macrocyclic toxins produced by several Fusarium species, colonizing maize, sorghum, wheat, barley oats and other cereal grains, which were also reported occurring in wastewaters and surface waters (Lagana et al., 2004). Published data on the occurrence and fate of phytoestrogens in the course of WWTPs are included in Table 3. As stated in Section 1, to give a whole profile of EDCs in the wastewater of WWTPs, a combination of chemical analyses and bioassays has been used, and estrogenic contribution ratios of detected EDCs to those of chemical derived EEQ have also been compared. As shown in Table 3, it is known that some phytoestrogens have been detected in both influent and effluent, and their removal efficiencies varied greatly as shown for natural estrogens and androgens (Liu et al., 2009). Data in Table 3 also suggested that the estrogenic contribution ratios of phytoestrogens could not be ignored in wastewaters when present in relatively high concentrations. For example,
573
water research 44 (2010) 567–577
Table 3 – Occurrence and fate of phytoestrogens in wastewater. Phytoestrogens Daidzein
Daidzin Genistein
Genistin Glycitein Formononetin Coumestrol Biochanin-A Equol Zearlenone a-Zearalanol b-Zearalanol a-Zearalenol b-Sitosterol
Enterolactone
Country
WWTP type (n)
Methods
LOD
Influent
Effluent
Average removal (%)
Reference
Germany Italy Italy Japan Italy Canada Italy Italy Italy Italy Italy Italy Italy Italy Japan Italy Italy Italy Germany Switzerland Canada
– LC–MS/MS LC–MS/MS LC–MS/MS LC–MS/MS LC–MS/MS LC–MS/MS LC–MS/MS LC–MS/MS LC–MS/MS LC–MS/MS LC–MS/MS LC–MS/MS LC–MS/MS LC–MS/MS LC–MS/MS LC–MS/MS LC–MS/MS – GC–MS GC–HRMS
10 2.5–3.5 3.5 5–60b 9–15 – 1.3–4.3 5–6 9–13 6–8 5–8 2–5 0.9–5.4 5–9 – 0.6–1.7 5.4–8.3 3.7–8.4 10 125 12–39
– 75–120 97–1685 3900–12000 18–74 13100 195–384 160–954 11–62 6–25 n.d.–10 9–23 8–18 14–76 9.4e5–1.1e6 3–18 n.d.–10 n.d.–8 – – 7303–35033
n.d.–10 7–16 5–81 5.7–23 n.d.–22 10100 15–83 7–22 n.d.–19 n.d.–16 n.d. n.d.–19 3–5 n.d.–21 170–410 3–10 n.d.–7 n.d.–5 20 or 30 400–1300 350–42110
– 88 88 or 98 99.8 69 or 100 23 81 97 66 or 100 62 or 88 – 66 or 100 73 67 or 90 99.99 53 0 33 – – –
Pawlowski et al., 2003 Lagana et al., 2004 Bacaloni et al., 2005 Liu et al., in press Bacaloni et al., 2005 Kiparissis et al., 2001 Lagana et al., 2004 Bacaloni et al., 2005 Bacaloni et al., 2005 Bacaloni et al., 2005 Bacaloni et al., 2005 Bacaloni et al., 2005 Lagana et al., 2004 Bacaloni et al., 2005 Furuichi et al., 2006 Lagana et al., 2004 Lagana et al., 2004 Lagana et al., 2004 Pawlowski et al., 2003 Baig et al., 2008 Fernandez et al., 2007
USA Germany Germany Germany France
AS (2) AS (1) AS (2) AS (2) AS (2) – (1) As (1) AS (2) AS (2) AS (2) AS (2) AS (2) As (1) AS (2) – (1) AS (1) AS (1) AS (1) AS (2) – (1) AS (3); TF/SC (1); L (1) – (2) – (2) AS (2) – (7) AS (2)
GC–MS GC–MS – GC–MS/MS GC–MS
– – 50 – –
32000a 14000or 16000 – 15900 4194–49920
– – 60 or 90 1900 1100–50165
– – – 88 –
Australian Finland
– (1) AS (1)
LC–MS LC–MS/MS
1 –
600 1223
– 659
– 46
Cain et al., 2008 Radke, 2005 Pawlowski et al., 2003 Heberer, 2002 Quemeneur and Marty, 1994 Kang et al., 2006 Smeds et al., 2007
AS, activated sludge; TF/SC, trickling filter/solid contact; L, lagoon; –, not available; n.d., not detected; LOD, limit of detection. a Maximum concentration. b Limit of quantification.
the authors calculated that six phytoestrogens in influents of a municipal WWTP contributed over 2.4% of total chemically calculated EEQ including E1, E2, E3 and EE2 (RBA values of EDCs are chosen from NITE, 2008), and increased to over 4.5% for the corresponding effluents (Lagana et al., 2004); In 12 influent samples of two municipal WWTPs, the contribution ratios of daidzein accounted for 3.0–14.8% of total chemically calculated EEQ (Liu et al., in press). In the influents of a swine wastewater treatment plant, the contribution ratios of phytoestrogen equol accounted for as high as about 64% of total chemically derived EEQ (Furuichi et al., 2006). In addition, for the extremely high concentrations of b-sitosterol in wastewaters (Table 3), its estrogenic contribution ratio in wastewater seems relatively high and it also can not be neglected. Similar to natural estrogens and androgens, about 97– 100% of phytoestrogens excreted from human urine are phytoestrogen conjugates, in which 62–97% of phytoestrogens are glucuronide conjugates (Adlercreutz et al., 1995). Although there is no information on the occurrence and fate of phytoestrogen conjugates in WWTPs, the glycoside conjugates of daidzein and genistein (daidzin and genistin), which have structures similar to those of their glucuronide conjugates, have been proven hydrolyzable to daidzein and
genistein by the glucosidase of human intestinal bacteria (Hur et al., 2000). The detected daidzin and genistin in some effluents of WWTPs suggested their incomplete hydrolysis during the whole activated sludge process (Table 3). In addition, based on the fact that natural estrogen conjugates occurred both the influent and effluent of WWTPs (D’Ascenzo et al., 2003; Isobe et al., 2003; Komori et al., 2004; Reddy et al., 2005; Nakada et al., 2006), phytoestrogen conjugates may also be present in influent and effluent of WWTPs with high frequency. Published data on the occurrence of phytoestrogens in other various environments are collected in Table 4. From Table 4, it is known that most phytoestrogens including mycoestrogens were found in rivers, and in some occasions the concentrations of phytoestrogens in drainage water were much higher than those in river water, which indicated that the drainage water was also a source of phytoestrogens in river waters. In Table 4 b-sitosterol showed very high concentrations in rivers and it was also detected in the atmosphere. However, it has not been detected in drinking water (Quemeneur and Marty, 1994; Heberer, 2002; Martins et al., 2007; Stackelberg et al., 2007). In addition, enterlactone was seldom monitored in wastewater as a micro-pollutant,
574
water research 44 (2010) 567–577
Table 4 – Phytoestrogens in the other various environments. Phytoestrogens Daidzein
Daidzin Genistein
Genistin Glycitein Formononetin
Coumestrol Biochanin-A
Equol Zearlenone a-Zearalanol a-Zearalenol b-Zearalanol b-Sitosterol
Enterolactone
River water (ng/L)
Drainage water (ng/L)
River sediments (ng/g)
Sea water (ng/L)
Drinking water
Air (pg/m3)
2–3 [1]; 2–4 [2]; 42900 [4]; <3.0–526 [6] n.d.b [2] 4–7 [1]; 3–5 [2]; 143400 [4]; <2.6–507.1 [6] n.d. [2]; <400 [4] n.d.–6 [2] n.d. [2]; dc-21 [3] 1 [16] 1–2 [2] 1–2 [1]; 1–3 [2]; d.–12 [3]; <8.4–60.2 [6] d.–22 [3] 2–5 [1] n.d.–3 [1] <10 [19] n.d.–3 [1] <1400–6000 [5]; 700–3000d [11] <930 [12] 5 [16]
5–30 [3]
–a
<0.43 [10]
–
–
– d.–14 [3]
– –
– <0.61 [10]
– –
– –
–-
–
–
–
–
– 44–157 [3]
– –
– –
– –
– –
n.d. [3] 7–22 [3]
– –
– –
– –
– –
34–121 [3] <30 [9] n.d. [9] – n.d. [9] 2000 [13]
– – – – – <2400–6300 [5]; 16.1–204.3 [7] 54–36142 [15] –
– – – – – 63.4–758.7e [7] 194–3156e [18]
– – – – – n.d. [11]
– – – – – 321–1765 [8] 1588 [17]
26 [14]
13 [14]
–
–
[1] Lagana et al., 2004; [2] Bacaloni et al., 2005; [3] Erbs et al., 2007; [4] Kawanishi et al., 2004; [5] Cain et al., 2008; [6] Ribeiro et al., in press; [7] Martins et al., 2007; [8] Radke, 2005; [9] Hartmann et al., 2007; [10] Beck et al., 2005; [11] Stackelberg et al., 2004; [12] Stackelberg et al., 2007; [13] Devane et al., 2006; [14] Smeds et al., 2007; [15] Chou and Liu, 2004; [16] Kang et al., 2006; [17] Beck and Radke, 2006; [18] Fattore et al., 1996; [19] Pawlowski et al., 2003. a Not available. b Not detected. c Detected but under the limit of quantification. d Values are drawn from figure. e ng/g sediments.
whereas the recent investigation denoted that it not only existed in wastewater, but also in river water, lake water, sea water and drinking water (Table 4).
WWTPs, will grow. On some occasions, chemical advanced oxidation should also be adopted to degrade phytoestrogens when purer water is required.
4.
references
Conclusions
Urinary excretion rates of phytoestrogens from humans and their occurrence and fate in the environment were summarized here. It was concluded that most phytoestrogens may be present in municipal wastewaters as can be estimated from their urinary excretion rates. Although phytoestrogens are very weak estrogens when compared to natural estrogens, their possible high concentrations in wastewater may lead them to contribute a non-negligible part of EEQ. Compared to natural estrogens or industrial chemicals such as BPA, NP and Octylphenol (OP), the research of phytoestrogens in wastewaters is just beginning. However, with the progress of analytic methods, knowledge of phytoestrogens in wastewater, such as removal mechanisms of phytoestrogens at
Adlercreutz, H., Fotsis, T., Bannwart, C., Wahala, K., Makela, T., Brunow, G., Hase, T., 1986. Determination of urinary ligans and phytoestrogen metabolites, potential antiestrogens and anticarcinogens, in urine of women on various habitual diets. J. Steroid Biochem 25, 791–797. Adlercreutz, H., Honjo, H., Higashi, A., Fotsis, T., Hamalainen, E., Hasegawa, T., Okada, H., 1991. Urinary excretion of ligans and isoflavonoid phytoestrogens in Japanese men and women consuming a traditional Japanese diet. Am. J. Clin. Nutr 54, 1093–1100. Adlercreutz, H., van der Wildt, J., Kinzel, J., Attalla, H., Wahala, K., Makela, T., Hase, T., Fotsis, T., 1995. Lignan and isoflavonoid conjugates in human urine. J. Steroid Biochem. Mol. Biol. 52, 97–103.
water research 44 (2010) 567–577
Bacaloni, A., Cavaliere, C., Faberi, A., Foglia, P., Samperi, R., Lagana, A., 2005. Determination of isoflavones and coumestrol in river water and domestic wastewater sewage treatment plants. Analytica. Chimica. Acta 531, 229–237. Baig, S., Hansmann, G., Paolini, B., 2008. Ozone oxidation of estrogenic active substances in wastewater and drinking water. Wat. Sci. Technol 58, 451–458. Baronti, C., Curini, R., D’Ascenzo, G., Corcia, A.D., Gentili, A., Samperi, R., 2000. Monitoring natural and synthetic estrogens at activated sludge sewage treatment plants and in a receiving river water. Environ. Sci. Technol 34, 5059–5066. Beck, M., Radke, M., 2006. Determination of sterols, estrogens and inorganic ions in waste water and size-segregated aerosol particles emitted from waste water treatment. Chemosphere 64, 1134–1140. Beck, I.C., Bruhn, R., Gandrass, J., Ruck, W., 2005. Liquid chromatography-tandem mass spectrometry analysis of estrogenic compounds in coastal surface water of the Baltic Sea. J. Chromatogr. A 1090, 98–106. Beveridge, T.H.J., Li, T.S.C., Drover, J.C.G., 2002. Phytosterol content in American ginseng seed oil. J. Agric. Food Chem. 50, 744–750. Beveridge, T.H.J., Girard, B., Kopp, T., Drover, J.C.G., 2005. Yield and composition of grape seed oils extracted by supercritical carbon dioxide and petroleum ether: varietal effects. J. Agric. Food Chem. 53, 1799–1804. Blair, R.M., Fang, H., Branham, W.S., Hass, B.S., Dial, S.L., Moland, C.L., Tong, W., 2000. The estrogen receptor relative binding affinities of 188 natural and xenochemicals: structural diversity of ligands. Toxicol. Sci. 54, 138–153. Bovee, T.F.H., Helsdingen, R.J.R., Rietjens, I.M.C.M., Keijer, J., Hoogenboom, R.L.A.P., 2004. Rapid yeast estrogen bioassays stably expressing human estrogen receptor a and b, and green fluorescent protein: a comparison of different compounds with both receptor types. J. Steroid Biochem. Mol. Biol. 91, 99–109. Cain, T.G., Kolpin, D.W., Vargo, J.D., Wichman, M.D., 2008. Occurrence of antibiotics, pharmaceuticals and sterols at select surface and wastewater sites in Iowa. http://info.ngwa. org/gwol/pdf/042379985.pdf. 2008.11.24. Chou, C.C., Liu, Y.P., 2004. Determination of fecal sterols in the sediments of different wastewater outputs by GC-MS. Intern. J. Environ. Anal. Chem. 84, 379–388. D’Ascenzo, G., Corcia, A.D., Mancini, A.G.R., Mastropasqua, R., Nazzari, M., Samperi, R., 2003. Fate of natural estrogen conjugates in municipal sewage transport and treatment facilities. Sci. Total Environ 302, 199–209. Dai, Q., Franke, A.A., Jin, F., Shu, X.O., Hebert, J.R., Custer, L.J., Cheng, J.R., Gao, Y.T., Zheng, W., 2002. Urinary excretion of phytoestrogens and risk of breast cancer among Chinese women in Shanghai. Cancer Epidem. Biomarkers 11, 815–821. Devane, M., Saunders, D., Gilpin, B., 2006. Faecal sterols and fluorescene whiteners as indicators of the source of faecal contamination. Chem. New Zealand 70, 74–77. Eneroth, P., Hellstrom, K., Ryhage, R., 1964. Identification and quantification of neutral fecal steroids by gas-liquid chromatography and mass spectrometry: studies of human excretion during two dietary regimens. J. Lipid Res. 5, 245–262. Erbs, M., Hoerger, C.C., Hartmann, N., Bucheli, T.D., 2007. Quantification of six phytoestrogens at the nanogram per liter level in aqueous environmental samples using 13C3-labeled internal standards. J. Agric. Food Chem. 55, 8339–8345. Fang, H., Tong, W., Branham, W.S., Moland, C.L., Dial, S.L.D., Hong, H., Xie, Q., Perkins, R., Owens, W., Sheehan, D.M., 2003. Study of 202 natural, synthetic, and environmental chemicals for binding to the androgen receptor. Chem. Res. Toxicol 16, 1338–1358. Fattore, E., Benfenati, E., Marelli, R., Cools, E., Fanelli, R., 1996. Sterols in sediment samples from Venice lagoon. Italy. Chemosphere 33, 2383–2393.
575
Fernandez, M.P., Ikonomou, M.G., Buchanan, I., 2007. An assessment of estrogenic organic contaminants in Canadian wastewaters. Sci. Total Environ 373, 250–269. Furuichi, T., Kannan, K., Giesy, J.P., Masunaga, S., 2004. Contribution of known endocrine disrupting substances to the estrogenic activity in Tama River water samples from Japan using instrumental analysis and in vitro reporter gene assay. Wat. Res. 38, 4491–4501. Furuichi, T., Kannan, K., Suzuki, K., Tanaka, S., Giesy, J.P., Masunaga, S., 2006. Occurrence of estrogenic compounds in and removal by a swine farm waste treatment plant. Environ. Sci. Technol 40, 7896–7902. Grosso, N.R., Nepote, V., Guzman, C.A., 2000. Chemical composition of some wild peanut species (Arachis L.) seeds. J. Agric. Food Chem. 48, 806–809. Gutendorf, B., Westendorf, J., 2001. Comparison of an array of in vitro assays for the assessment of the estrogenic potential of natural and synthetic estrogens, phytoestrogens and xenoestrogens. Toxicology 166, 79–89. Hartmann, N., Erbs, M., Wettstein, F.E., Schwarzenbach, R.P., Bucheli, T.D., 2007. Quantification of estrogenic mycotoxins at the ng/L level in aqueous environmental samples using deuterated internal standards. J. Chromatogr. A 1138, 132–140. Heberer, T., 2002. Tracking persistent pharmaceutical residues from municipal sewage to drinking water. J. Hydrol 266, 175–189. Holser, R.A., Bost, G., Boven, M.V., 2004. Phytosterol composition of hybrid Hibiscus seed oils. J. Agric. Food Chem. 52, 2546–2548. Horn-Ross, P.L., Barnes, S., Kirk, M., Coward, L., Parsonnet, J., Hiatt, R.A., 1997. Urinary phytoestrogen levels in young women from a multiethnic population. Cancer Epidem. Biomar 6, 339–345. Hur, H.G., Lay Jr., J.O., Beger, R.D., Freeman, J.P., Rafii, F., 2000. Isolation of human intestinal bacteria metabolizing the natural isoflavone glycosides daidzin and genistin. Arch. Microbiol. 174, 422–428. Isobe, T., Shiraishi, H., Yasuda, M., Shinoda, A., Suzuki, H., Morita, M., 2003. Determination of estrogens and their conjugates in water using solid-phase extraction followed by liquid chromatography-tandem mass spectrometry. J. Chromatogr. A 984, 195–202. Iwatsuki, K., Akihisa, T., Tokuda, H., Ukiya, M., Higashihara, H., Mukainaka, T., Izuka, M., Hayashi, Y., Kimura, Y., Nishino, H., 2003. Sterol ferulates, sterols, and 5-alk(en)yl resorcinols from wheat, rye and corn bran oils and their inhibitory effects on Epstein-Barr virus activation. J. Agric. Food Chem. 51, 6683– 6688. Jackson, S., 1966. Creatinine in urine as an index of urinary excretion rate. Health Phys. 12, 843–850. Jenner, A.M., Rafter, J., Halliwell, B., 2005. Human fecal water content of phenolics: the extent of colonic exposure to aromatic compounds. Free Radic. Biol. Med 38, 763–772. Jimenez-Escrig, A., Santos-Hidalgo, A.B., Saura-Calixto, F., 2006. Common sources and estimated intake of plant sterols in the Spanish diet. J. Agric. Food Chem 54, 3462–3471. Johnson, A.C., Williams, R.J., 2004. A model to estimate influent and effluent concentrations of estradiol. estrone, and ethinylestradiol at sewage treatment works. Environ. Sci. Technol 38, 3649–3658. Johnson, A.C., Belfroid, A., Di Corcia, A., 2000. Estimating steroid oestrogen inputs into activated sludge treatment works and observations on their removal from the effluent. Sci. Total Environ 256, 163–173. Johnson, A.C., Williams, R.J., Simpson, P., Kanda, R., 2007. What difference might sewage treatment performance make to endocrine disruption in rivers. Environ. Pollut 147, 194–202. Joss, A., Andersen, H., Ternes, T., 2004. Removal of estrogens in municipal wastewater treatment under aerobic and anaerobic
576
water research 44 (2010) 567–577
conditions: consequences for plant optimization. Environ. Sci. Technol 38, 3047–3055. Kang, J.G., Price, W.E., Hick, L.A., 2006. Simultaneous determination of isoflavones and lignans at trace levels in natural waters and wastewater samples using liquid chromatography/electrospray ionization ion trap mass spectrometry. Rapid Commun. Mass Spectrom 20, 2411–2418. Karr, S.C., Lampe, J.W., Hutchins, A.M., Slavin, J.L., 1997. Urinary isoflavonoid excretion in humans is dose dependent at low to moderate levels of soy-protein comsumption. Am. J. Clin. Nutr 66, 46–51. Kawanishi, M., Takamura-Enya, T., Ermawati, R., Shimohara, C., Sakamoto, M., Matsukawa, K., Matsuda, T., Murahashi, T., Matsushi, T., Wakabayashi, K., Watanabe, T., Tashiro, Y., Yagi, T., 2004. Detection of genistein as an estrogenic contaminant of river water in Osaka. Environ. Sci. Technol 38, 6424–6429. Kiparissis, Y., Hughes, R., Metcalfe, C., Ternes, T., 2001. Identification of the isoflavonoid genistein in bleached kraft mill effluent. Environ. Sci. Technol 35, 2423–2427. Kiparissis, Y., Balch, G.C., Metcalfe, T.L., Metcalfe, C.D., 2003. Effects of the isoflavones genistein and equol on the gonadal development of Japanese medaka (Oryzias latipes). Environ. Health Persp 111, 1158–1163. Komori, K., Tanaka, H., Okayasu, Y., Yasojima, M., Sato, C., 2004. Analysis and occurrence of estrogen in wastewater in Japan. Wat. Sci. Technol 50, 93–100. Kostelac, D., Rechkemmer, G., Briviba, K., 2003. Phytoestrogens modulate binding response of estrogen receptor a and b to the estrogen response element. J. Agric. Food Chem. 51, 7632– 7635. Kris-Etherton, P.M., Hecker, K.D., Bonanome, A., Coval, S.M., Binkoski, A.E., Hilpert, K.F., Griel, A.E., Etherton, T.D., 2002. Bioactive compounds in foods: their role in the prevention of cardiovascular disease and cancer. Am.. J. Med 113 (Suppl), 71–88. Kuhnle, G.G.C., Dell’Aquila, C., Kussmaul, M., Bingham, S.A., 2007. Extraction and quantification of phytoestrogens in foods using automated solid-phase extraction and LC/MS/MS. Anal. Chem. 79, 9234–9239. Kuhnle, G.G.C., Dell’aquila, C., Aspinall, S.M., Runswick, S.A., Mulligan, A.A., Bingham, S.A., 2008. Phytoestrogen content of beverages, nuts, seeds, and oils. J. Agric. Food Chem. 56, 7311– 7315. Kuiper, G., Lemmen, J.G., Carlsson, B., Corton, J.C., Safe, S.H., van der Saag, P.T., van der Burg, B., Gustafsson, J.A., 1998. Interaction of estrogenic chemicals and phytoestrogens with estrogen receptor b. Endocrinology 139, 4252–4263. Lagana, A., Bacaloni, A., De Leva, I., Faberi, A., Fago, G., Marino, A., 2004. Analytical methodologies for determining the occurrence of endocrine disrupting chemicals in sewage treatment plants and natural waters. Analytica. Chimica. Acta 501, 79–88. Lampe, J.W., Wartini, M.C., Kurzer, M.S., Adlercreutz, H., Slavin, J. L., 1994. Urinary lignan and isoflavonoid excretion in premenopausal women consuming flaxseed powder. J. Clin. Nutr 60, 122–128. Lampe, J.W., Gustafson, D.R., Hutchins, A.M., Martini, M.L., Li, S., Wahala, K., Grandits, G.A., Potter, J.D., Slavin, J.L., 1999. Urinary isoflavonoid and lignan excretion on a western diet: relation to soy, vegetable, and fruit intake. Cancer Epidem. Biomarkers 8, 699–707. Lampe, J.W., Skor, H.E., Li, S., Wahala, K., Howald, W.N., Chen, C., 2001. Wheat bran and soy protein feeding do not alter urinary excretion of the isoflavan equol in premenopausal women. J. Nutr 131, 740–744. Leeming, R., Ball, A., Ashbolt, N., Nichols, P., 1996. Using faecal sterols from humans and animals to distinguish faecal pollution in receiving waters. Wat. Res. 30, 2893–2900.
Liggin, J., Bluck, L.J.C., Runswick, S., Atkinson, C., Coward, W.A., Bingham, S.A., 2000. Daidzein and genistein content of fruits and nuts. J. Nutr. Biochem 11, 326–331. Lissin, L.W., Cook, J.P., 2000. Phytoestrogens and cardiovascular health. J. Am. Coll. Cardiol 35, 1403–1410. Liu, Z.H., Kanjo, Y., Mizutani, S., 2009. Removal mechanisms for endocrine disrupting compounds (EDcs) in wastewater treatment-physical means, biodegradation, and chemical advanced oxidation: A review. Sci. Total Environ 407, 731–748. Liu, Z.H., Kanjo, Y., Mizutani, S. Excretions of natural estrogens and androgens from humans, and their occurrence and fate in the environments. Sci. Total Environ, submitted for publication. Liu, Z.H., Mamoru, I., Kanjo, Y., Yamamoto, A. Profile and removal of endocrine disrupting chemicals (EDCs) in two municipal wastewater treatment plants by using an ER (AR) competitive ligand binding assay and chemical analyses. J. Environ. Sci. in press. doi:10.1016/S1001-0742(08)62356-6. Low, Y.L., Taylor, J.I., Grace, P.B., Dowsett, M., Folkerd, E., Doody, D., Dunning, A.M., Scollen, S., Mulligan, A.A., Welch, A. A., Luben, R.N., Khaw, K.T., Day, N.E., Wareham, N.J., Bingham, S.A., 2005. Polymorphisms in the CYP19 gene may affect the positive correlations between serum and urine phytoestrogen metabolites and plasma androgen concentrations in men. J. Nutr 135, 2680–2686. Markiewicz, L., Garey, J., Adlercreutz, H., Gurpide, E., 1993. In vitro bioassays of non-steroidal phytoestrogens. J. Steroid Biochem. Mol. Biol. 45, 399–405. Martins, C., Fillmann, G., Montone, R.C., 2007. Natural and anthropogenic sterols inputs in surface sediments of Patos Lagoon. Brazil. J. Braz. Chem. Soc. 18, 106–115. Matthews, J., Celius, T., Halgren, R., Zacharewski, T., 2000. Differential estrogen receptor binding of estrogenic substances: a species comparison. J. Steroid Biochem. Mol. Biol. 74, 223–234. Mazur, W., Fotsis, T., Wahala, K., Ojalas, S., Salakka, A., Adlercreutz, H., 1996. Isotope dilution gas chromatographicmass spectrometric method for the determination of isoflavonoids, coumestrol, and lignans in food samples. Anal. Biochem 233, 169–180. Milder, I.E.J., Feskens, E.J.M., Arts, I.C.W., de Mesquita, H.B.B., Hollman, P.C.H., Kromhout, D., 2005. Intake of the plant lignans secoisolariciresinol, matairesinol, lericiresinol, and pinoresinol in Dutch men and women. J. Nutr 135, 1202–1207. Mueller, S., Simon, S., Chae, K., Metzler, M., Korach, K., 2004. Phytoestrogens and their human metabolites show distinct agonistic and antagonistic properties on estrogen receptora(ERa) and ER(b) in human cells. Toxicol. Sci. 80, 14–25. Nagata, Y., Miyanaga, N., Okumura, K., Goto, K., Naito, S., Fujimoto, K., Hirao, Y., Takahashi, A., Tsukamoto, T., Akaza, H., 2007. Dietary isoflavones may protect against prostate cancer in Japanese men. J. Nutr 137, 1974–1979. Nakada, N., Nyunoya, H., Nakamura, M., Hara, A., Iguchi, T., Takada, H., 2004. Identification of estrogenic compounds in wastewater effluent. Environ. Toxicol. Chem. 23, 2807–2815. Nakada, N., Higashitani, T., Miyajima, K., Komori, K., Suzuki, Y., 2006. Development of method for identification of major substances inducing estrogenic activity contained in sewage and river waters. J. Environ. Chem. 16, 389–401 (in Japanese with English abstract). Nelson, J., Bishay, F., van Roodselaar, A., Ikonomou, M., Law, F.C. P., 2007. The use of in vitro bioassays to quantify endocrine disrupting chemicals in municipal wastewater treatment plants effluents. Sci. Total Environ 374, 80–90. Nielsen, S.E., Freese, R., Kleemola, P., Mutanen, M., 2002. Flavonoids in human urine as biomarkers for intake of fruits and vegetables. Cancer Epidem. Biomarkers 11, 459–466.
water research 44 (2010) 567–577
Nishihara, T., Nishikawa, J.I., Kanayama, T., Dakeyama, F., Saito, K., Imagawa, M., Takatori, S., Kitagawa, Y., Hori, S., Utsumi, S., 2000. Estrogenic activities of 517 chemicals by yeast two-hybrid assay. J. Health Sci. 46, 282–298. NITE, 2008. http://www.safe.nite.go.jp/english/index.html. 2008.11. Pawlowski, S., Ternes, T., Bonerz, M., Kluczka, T., van der Burg, B., Nau, H., Erdinger, L., Braunbeck, T., 2003. Combined in situ and in vitro assessment of the estrogenic activity of sewage and surface water samples. Toxicol. Sci. 75, 57–65. Peeters, P.H.M., Keinan-Boker, L., van der Schouw, Y.T., Grobbee, D.E., 2003. Phytoestrogens and breast cancer risk. Review of the epidemiological evidence. Breast Cancer Res. Treat 77, 177–183. Penalvo, J.L., Adlercreutz, H., Uehara, M., Ristimaki, A., Watanabe, S., 2008. Lignan content of selected foods from Japan. J. Agric. Food Chem. 56, 401–407. Phillips, K.M., Ruggio, D.M., Ashraf-Khorassan, M., 2005. Phytosterol composition of nuts and seeds commonly consumed in the United States. J. Agric. Food Chem. 53, 9436– 9445. Quemeneur, M., Marty, Y., 1994. Fatty acids and sterols in domestic wastewaters. Wat. Res. 28, 1217–1226. Radke, M., 2005. Sterols and anionic surfactants in urban aerosol: emissions from wastewater treatment plants in relation to background concentrations. Environ. Sci. Technol 39, 4391– 4397. Reddy, S., Iden, C.R., Brownawell, B., 2005. Analysis of steroid conjugates in sewage influent and effluent by liquid chromatography-tandem mass spectrometry. Anal. Chem. 77, 7032–7038. Salste, L., Leskinen, P., Virta, M., Kronberg, L., 2007. Determination of estrogens and estrogenic activity in wastewater effluent by chemical analysis and bioluminescent yeast assay. Sci. Total Environ 378, 343–351. Seow, A., Shi, C.Y., Franke, A.A., Hankin, J.H., Lee, H.P., Yu, M.C., 1998. Isoflavonoid levels in spot urine are associated with frequency dietary soy intake in a population-based sample of middle-aged and older Chinese in Singapore. Cancer Epidem. Biomar 7, 135–140. Setchell, K.D.R., Cassidy, A., 1999. Dietary isoflavones: biological effects and relevance to human health. J. Nutr 129 (Suppl), 758–767. Sivesind, E., Seguin, P., 2005. Effects of the environment, cultivar, maturity, and preservation method on red clover isoflavone concentration. J. Agric. Food Chem. 53, 6397–6402. Smeds, A.I., Willfor, S.M., Pietarinen, S.P., Peltonen-Sainio, P., Reunanen, M.H.T., 2007. Occurrence of ‘‘mammalian’’ lignans in plant and water sources. Planta 226, 639–646.
577
Stackelberg, P.E., Furlong, E.T., Meyer, M.T., Zaugg, S.D., Hendersen, A.K., Reissman, D.B., 2004. Persistence of pharmaceutical compounds and other organic wastewater contaminants in a conventional drinking-water-treatment plant. Sci. Total Environ 329, 99–113. 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. Sci. Total Environ 377, 255–272. Song, T.T., Hendrich, S., Murphy, P.A., 1999. Estrogenic activity of glycitein, a soy isoflavone. J. Agric. Food Chem. 47, 1607–1610. Stopper, H., Schmitt, E., Kobras, K., 2005. Genotoxicity of phytoestrogens. Mutat. Res. 574, 139–155. Tan, B.L.L., 2006. Chemical and biological analyses of selected endocrine disruptors in wastewater treatment plants in South East Queensland, Australia. Doctorate dissertation. Griffith University, Australia. Tonkelaar, I.D., Keinan-Boker, L., Veer, P.V., Arts, C.J.M., Adlercreutz, H., Thijssen, J.H.H., Peeters, H.M., 2001. Urinary phytoestrogens and postmenopausal breast cancer risk. Cancer Epidem. Biomarkers 10, 223–228. Tzchori, I., Degani, G., Elisha, R., Eliyahu, R., Hurvitz, A., Vaya, J., Moav, B., 2004. The influence of phytoestrogens and oestradiol-17b on growth and sex determination in the European eel (Anguilla anguilla). Aquat. Res. 35, 1213–1219. USDA (US Department of Agriculture), 2003. USDA database for the flavonoid content of selected foods. http: www.nal.usda. gov/fnic/foodcomp. Valentin-Blasini, L., Sadowski, M.A., Walden, D., Caltabiano, L., Needham, L.L., Barr, D.B., 2005. Urinary phytoestrogen concentrations in the U.S. population (1999–2000). J. Expo. Anal. Environ. Epidemiol 15, 509–523. Wang, L.Q., 2002. Mammalian phytoestrogens: enterodiol and enterolactone. J. Chromatogr. B 777, 289–309. Whitten, P.L., Patisaul, H.B., 2001. Cross-species and interassay comparisons of phytoestrogen action. Environ. Health Persp 109 (Suppl), 5–20. Xu, X., Duncan, A.M., Wangen, K.E., Kurzer, M.S., 2000. Soy consumption alters endogenous estrogen metabolism in postmenopausal women. Cancer Epidem. Biomarkers 9, 781–786. Zhang, Y., Wang, G.J., Song, T.T., Murphy, P.A., Hendrich, S., 1999. Urinary disposition of the soybean isoflavones daidzein, genistein and glycitein differs among humans with moderate fecal isoflavone degradation activity. J. Nutr 129, 957–962. Zhang, X., Cambrai, A., Miesch, M., Roussi, S., Raul, F., AoudeWerner, D., Marchioni, E., 2006. Separation of D5-and D7phytoesterols by adsorption chromatography liquid chromatography for quantitative analysis of phytosterols in foods. J. Agric. Food Chem. 54, 1196–1202.
water research 44 (2010) 578–588
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Occurrence of emerging pollutants in urban wastewater and their removal through biological treatment followed by ozonation Roberto Rosala,*, Antonio Rodrı´gueza, Jose´ Antonio Perdigo´n-Melo´na, Alice Petrea, Eloy Garcı´a-Calvoa, Marı´a Jose´ Go´mezb, Ana Agu¨erab, Amadeo R. Ferna´ndez-Albab a
Department of Chemical Engineering, University of Alcala´, 28771 Alcala´ de Henares, Spain Department of Analytical Chemistry, University of Almerı´a, 04010 Almerı´a, Spain
b
article info
abstract
Article history:
This work reports a systematic survey of over seventy individual pollutants in a Sewage
Received 24 February 2009
Treatment Plant (STP) receiving urban wastewater. The compounds include mainly phar-
Received in revised form
maceuticals and personal care products, as well as some metabolites. The quantification in
26 June 2009
the ng/L range was performed by Liquid Chromatography–QTRAP–Mass Spectrometry and
Accepted 3 July 2009
Gas Chromatography coupled to Mass Spectrometry. The results showed that para-
Available online 9 July 2009
xanthine, caffeine and acetaminophen were the main individual pollutants usually found
Keywords:
detected in the ppb level. A group of compounds including the beta-blockers atenolol,
Emerging pollutants
metoprolol and propanolol; the lipid regulators bezafibrate and fenofibric acid; the anti-
Pharmaceuticals
biotics erythromycin, sulfamethoxazole and trimethoprim, the antiinflammatories diclo-
Personal care products
fenac, indomethacin, ketoprofen and mefenamic acid, the antiepileptic carbamazepine
Wastewater treatment
and the antiacid omeprazole exhibited removal efficiencies below 20% in the STP treat-
in concentrations over 20 ppb. N-formyl-4-amino-antipiryne and galaxolide were also
Ozonation
ment. Ozonation with doses lower than 90 mM allowed the removal of many individual pollutants including some of those more refractory to biological treatment. A kinetic model allowed the determination of second order kinetic constants for the ozonation of bezafibrate, cotinine, diuron and metronidazole. The results show that the hydroxyl radical reaction was the major pathway for the oxidative transformation of these compounds. ª 2009 Elsevier Ltd. All rights reserved.
1.
Introduction
The presence of a wide variety of pharmaceutical and personal care products (PPCP) in water and wastewater has been frequently reported after the findings of Ternes (1998) and Daughton and Ternes (1999). These compounds are a source of concern because they are used and released in large quantities and their physical and chemical properties
contribute to their widespread distribution into the environment. The presence of small concentration of PPCP has been associated to chronic toxicity, endocrine disruption and the development of pathogen resistance. The consequences are particularly worrying in aquatic organisms as they are subjected to multigenerational exposure (Halling-Sørensen et al., 1998). The presence of micropollutants also endangers the reuse of treated wastewater, a generally proposed solution to
* Corresponding author. Department of Chemical Engineering, Facultad de Quamica, University of Alcala´, 28771 Alcala´ de Henares, Madrid, Spain. Tel.: þ34 91 885 5099; fax: þ34 91 885 5088. E-mail address:
[email protected] (R. Rosal). 0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2009.07.004
water research 44 (2010) 578–588
kL
Notation Ci CHO$ CO3 CO3 DO3 Dow fOH Ha
concentration of a given organic compound (M) concentration of hydroxyl radical (M) concentration of dissolved ozone (M) equilibrium concentration of dissolved ozone (M) diffusivity of ozone in water, m2 s1 apparent octanol–water partition coefficient (dimensionless) fraction of a given compound degraded by hydroxyl radicals Hatta number, defined in Eq. (5) (dimensionless)
achieve a sustainable water cycle management (Mun˜oz et al., 2009). PPCP represent a rising part of the trace organic micropollutants found in urban and domestic wastewaters that reach sewage treatment plants (STP), either metabolised or not (Castiglioni et al., 2006). Many of these substances escape to conventional activated sludge wastewater treatments allowing them to reach surface water streams and distribute in the environment (Tauxe-Wuersch et al., 2005). The need for treatment technologies that can provide safe treated effluents led to the proposals for upgrading STP and to implement new competing technologies for biological degradation of organic matter like membrane bioreactor (Radjenovic et al., 2009). In addition to these strategies, effective tertiary treatment technologies are also needed in order to ensure a safe use for reclaimed wastewater. The available technologies include oxidation processes alone or combined with nanofiltration or reverse osmosis (Ernst and Jekel, 1999). Many oxidation processes have been described for the removal of organic compounds in wastewater. Ozone-based and Advanced Oxidation Processes (AOP) using hydrogen peroxide or radiation have been repeatedly proposed for this task (Gogate and Pandit, 2004a; Ikehata et al., 2006). Proper combinations of AOP can also be considered in order to treat the more refractory pollutants. Fenton and Fenton-based systems, heterogeneous photocatalysis, and ultraviolet or ozone-based oxidation processes have been described (Gogate and Pandit, 2004b; Comninellis et al., 2008). The choice of the most suitable technology or combination lies on the quality required for the reclaimed water. Ozone has been largely used as oxidant in drinking water treatment and repeatedly proposed to remove organics in wastewater treatment (Raknes, 2005; Beltra´n, 2004). The ozone molecule can react with many organic compounds, particularly those unsaturated or containing aromatic rings or heteroatoms also being able to decompose in water to form hydroxyl radicals. In a previous work (Rosal et al., 2008a) we studied the ozonation of wastewater from the secondary clarifier of urban and domestic STP by using ozone (pH w 8) and ozone–hydrogen peroxide (O3/H2O2). The presence of hydrogen peroxide improved the mineralization of dissolved carbon from 15% to over 90% after one hour, from which most part took place during the first five minutes. The disappearance of a selected group of over thirty pharmaceuticals revealed removal efficiencies were over >99% for most compounds after
kO3 kHO$ kR kLa Kow TOD
579
liquid-phase individual mass transfer coefficient, m s–1 second order ozone-based rate constant for a direct ozonation reaction (M1 s1) second order rate constant for a reaction with hydroxyl radical (M1 s1) second order rate constant for the ozonation of a given compound (M1 s1) volumetric mass transfer coefficient (s1) octanol–water partition coefficient for neutral species (dimensionless) dose of ozone transferred to the liquid (M)
five minutes on stream, with slightly better results in the absence of hydrogen peroxide. This result is consistent with the fact that many PPCP directly react with ozone with large second order kinetic constants (Huber et al., 2003). The objective of this research was to verify the occurrence and fate of 84 pollutants of different classes, mainly PPCP traced during wastewater treatment in a conventional urban STP. These include: pharmaceuticals (analgesics, antidepressants, antiinflammatories, antibiotics, antiepileptics, betablockers and lipid regulators among others), personal care products (sunscreen agents, synthetic musks), stimulants (caffeine, nicotine) and some metabolites (clofibric acid, cotinine, several metabolites of dipyrone). The effectiveness of the STP process for the removal of these compounds has been assessed in a monitoring program undertaken in Alcala´ de Henares (Madrid) during a one-year period with samples taken before and after a biological activated sludge with nutrient removal process. The research was also trying to identify the impact of ozone exposure on individual pollutants encountered in the secondary effluent. The doses of ozone required for a given degree of removal of the most representative individual pollutants has been assessed. For certain pollutants, the determination of second order kinetic constants for the ozonation reaction in a real wastewater matrix was also performed.
2.
Material and methods
2.1.
Materials and plant description
Wastewater samples were taken every month over a one-year period from the input and output of the secondary clarifier of a STP located in Alcala´ de Henares (Madrid). This plant treats a mixture of domestic and industrial wastewater from some facilities located near the city with a nominal capacity of 3000 m3/h of raw wastewater. The Council Directive 91/271/ EEC concerning urban wastewater treatment established strict limits for the wastewater discharged to sensitive areas particularly by plants serving an equivalent population of 10 000 or more. In the period prior to the sampling campaign, the need to adapt STP to the new conditions forced, the implementation of a biological nutrient removal process aimed to the elimination of phosphorous and nitrogen. The
580
water research 44 (2010) 578–588
biological treatment worked under a traditional A2O multistage configuration with nitrification–denitrification and enhanced phosphorus removal by phosphorus-accumulating microorganisms. The treatment takes place in three zones: anaerobic, anoxic and oxic. The nitrate produced in the oxic zone is recycled, with mixed liquor to the anoxic zone where denitrification takes place. The return sludge from the settler is recycled to the anaerobic zone where the influent and sludge are mixed under anaerobic conditions. All samples were immediately processed or stored in a refrigerator (<4 C) inside glass bottles. The main characterization parameters of wastewater before and after the biological treatment of the STP are shown in Table 1.
2.2.
Liquid chromatography–QTRAP–Mass spectrometry
Previous to LC–QTRAP–MS analysis, wastewater samples were preconcentrated by automated solid phase extraction (SPE) using Oasis HLB cartridges (Waters, 200 mg, 6cc). The operational procedure, which mainly includes a preconditioning step with MeOH and deionized water (pH ¼ 8); sample loading (200 mL of effluent wastewater and 400 mL of the treated samples at pH ¼ 8) and final elution with 2 4 mL of MeOH, has been described in detail elsewhere (Bueno et al., 2007). The final extracts were evaporated and reconstituted with 1 mL of MeOH:H2O, 10:90 (v/v), filtered, and diluted 1:1 with MeOH:H2O (10:90) before the analysis. Sample analysis was performed in a HPLC series 1100 (Agilent Technologies) coupled to a 3200 QTRAP MS/MS system (Applied Biosystems) using a turbo ionspray source in positive and negative modes. Separation was performed in a C-18 analytical column (ZORBAX SB, 250 mm; 3.0 mm I.D.; 5 mm). For the analysis in positive mode, the compounds were separated using acetonitrile (A) and water with 0.1% formic acid (B) at a flow rate of 0.2 mL/min. A linear gradient progressed from 10% A to 100% A in 40 min, and was maintained at 100% A for 10 min. In negative mode mobile phase composition was acetonitrile (A) and water (B) at a flow
Table 1 – Wastewater characterization parameters for influents and effluents from Alcala´ STP. Standard deviations in parenthesis. Parameters
Influent
Effluent
pH TSS (mg/L) Turbidity (NTU) Conductivity (mS/cm) COD (mg/L) BOD5 (mg/L) TOC (mg/L) Total-P (mg/L) N–NO3 (mg/L) N–NH4 (mg/L) Sulphate (mg/L) Chloride (mg/L) Sodium (mg/L) Potassium (mg/L) Magnesium (mg/L) Calcium (mg/L) Alkalinity (mg/L CaCO3)
7.54 (0.24) 67.8 (40.9) 65 (14) 703 (139) 269 (50) 42 (17) – 4.8 (0.3) 0.50 (0.47) 14.7 (5.6) 100 (36) 74.2 (24.0) 60.2 (13.0) 23.6 (15.1) 17.8 (3.7) 45.3 (8.0)
7.63 7.5 5.3 589 59 7.8 6.9 0.82 3.7 8.5 107 78.8 60.3 11.3 17.0 37.5 472
–
(0.17) (9.3) (4.4) (131) (24) (3.7) (1.4) (1.34) (2.3) (1.8) (30) (21.4) (10.6) (1.5) (2.4) (3.8) (39)
rate of 0.3 mL/min. LC gradient started with 30% A, was linearly increased to 100% A in 7 min, and was maintained at 100% A for 8 min. The volume of injection was of 20 mL in both modes.
2.3.
Gas chromatography/Mass spectrometry (GC–MS)
For GC–MS analysis, sample aliquots of 500 mL (pH ¼ 3) were preconcentrated by liquid–liquid extraction (LLE) using n-hexane as extraction solvent (Go´mez et al., 2009). The organic phase was evaporated at a final volume of 3 mL. A higher preconcentration factor was applied to the treated samples, for which an aliquot of 1 mL of the final extract was evaporated to 0.4 mL. GC–MS analyses were run on an Agilent 7890 series gas chromatograph (Agilent Technologies, Santa Clara, CA) interfaced to an Agilent 5975 mass-selective detector. Analytes were separated in an Agilent HP-5MSi capillary column (5% biphenyl/95% dimethylsiloxane), 15 m 0.25 mm i.d., 0.25 mm film thickness. The inlet operating conditions were as follows: injection volume 10 mL; the temperature programme went from 79 C (0.25 min) to 300 C (2 min) at 710 C min1. The oven temperature programme was 70 C (1 min), to 150 C at 50 C min1, then to 200 C at 6 C min1 and finally to 280 C at 16 C min1; it was kept at this temperature for 5 min. Electron impact (EI) mass spectra in full-scan mode were obtained at 70 eV; the monitoring was from m/z 50 to 400. The ion source and quadrupole analyzer temperatures were fixed at 230 C and 150 C, respectively.
2.4.
Other analyses
The concentration of dissolved ozone was continuously measured using an amperometric Rosemount 499A OZ analyser whose signal was sent to a Rosemount 1055 SoluComp II Dual Input. The analyzer was calibrated against the standard Indigo Colorimetric Method (SM 4500-O3B). The pH of the reaction mixture was determined by means of a CRISON electrode connected to a Eutech alpha-pH100 feed-back control system whose final control element was a LC10AS Shimadzu pump delivering a solution of sodium hydroxide. The signals corresponding to the concentration of dissolved ozone, pH and temperature were recorded using an Agilent 34970 Data Acquisition Unit connected to a computer. Based on the dynamic response of the three measuring devices, the sampling period was set as 5 s. The determination of Total Organic Carbon (TOC) was performed using a Shimadzu TOC-VCSH analyzer equipped with ASI-V autosampler. Inorganic anions were determined by means of a Dionex DX120 Ion Chromatograph with conductivity detector and an IonPac AS9-HC 4 250 mm analytical column þ ASRS-Ultra suppressor using 9.0 mM Na2CO3 with a flow of 1.0 mL/min as eluent. Inorganic cations were determined by an IonPac CS512A 4 250 mm cation exchange analytical column with Dionex CSRS Ultra II suppressor and 20 mM metasulphonic acid as eluent. Total suspended solids were determined by the American Public Health Association (APHA) Method 2540 D, ‘‘Total Suspended Solids Dried at 103–105 C’’. The determination of COD and BOD followed APHA Methods 5220C and 5210B respectively. The Standard Method SM 4500-P E was used for the determination of ortho-phosphate. Nitrates were determined according to ISO 7890/1 and ammonia nitrogen by means of SM 4500-NH3 D. The total alkalinity was determined by titration.
water research 44 (2010) 578–588
2.5.
Ozonation
The ozonation runs were carried out in a 5-L glass jacketed reactor operating in semi-batch mode. A temperature of 25 C, chosen to be close to average ambient temperature, was kept using a Huber Polystat cc2 and monitored throughout the runs by means of a Pt100 Resistance Temperature Detector (RTD). Ozone was produced by a corona discharge ozonator (Ozomatic, 119 SWO100) fed by an AirSep AS-12 PSA oxygen generation unit. The gas containing about 9.7 g/Nm3 ozone was bubbled by means of a porous glass disk with a gas flow of 0.36 Nm3/h. The reaction vessel was agitated with a Teflon four-blade impeller at 1000 rpm. The mass transfer coefficient was determined in transient runs with pure water with a value of kLa ¼ 0.010 0.005 s1. Additional details on the experimental set-up and procedure are given elsewhere (Rosal et al., 2008a,b). Throughout the runs, certain samples were withdrawn for analysis at prescribed intervals. Residual ozone was removed by bubbling nitrogen in order to prevent oxidation reactions to continue. For the desorption conditions used, residual ozone fell down below 10% in less than 20 s. During the ozonation reactions a moderate increase of pH took place that can be attributed to the stripping of CO2 from solution. In conditions that favour the generation of hydroxyl radicals, this effect is not observed and pH tend to decrease during ozonation due to the accumulation of carboxylic acids. The choice of a pH higher than that of the raw wastewater not only favours hydroxyl radical mediated reactions, but allowed to keep an almost constant pH value of 8.5 0.1 by using the feed-back control procedure described above.
3.
Results and discussion
3.1.
Occurrence
The pollutants analyzed were mainly PPCP and their metabolites together with some agrochemicals. A list of the 72 anthropogenic emerging pollutants detected in at least one wastewater sample from the influent to the biological treatment is detailed in the Appendix together with the analytical method applied, the molecular formula and the octanol–water partition coefficient, when available. The limit of quantification for the biological effluent (LOQ), for most compounds in the tens of ng/L, is also shown in the Appendix that also lists the compounds checked but not detected with their limits of detection (LOD). Several pesticides like chlorfenvinphos and the herbicide isoproturon were never detected, as expected considering the urban origin of wastewater. Atrazine, however, was found in all samples with an average concentration at the inlet of the biological treatment of 109 ng/L. Diuron was encountered in two samples and simazine in three with maximum concentrations of 196 and 32 ng/L, respectively. Some drugs like paroxetine, the antibiotic cefotaxime and the antiinflammatory fenoprofen were not detected in any of the analyzed samples. A similar negative result was reported for paroxetine by Terzic´ et al. (2008), whereas the occurrence of fenoprofen was reported in concentrations as high as 0.759 mg/L in Canadian wastewater facilities (Metcalfe et al., 2004).
581
The detailed data for the concentrations of the most significant individual pollutants are shown in Table 2. It includes maximum and minimum values of those compounds encountered over their quantification limit in at least 4 samples in the influent of the biological treatment. The compounds excluded for not complying with this criterion comprise, among other, carbamazepine-10,11-epoxide that was detected in 2 samples with a maximum concentration of 63 ng/L, sotalol, detected three times in the influent with concentrations in the 25–30 ng/L range or celestolide with a maximum influent concentration of 30 ng/L. In addition, fenofibrate was encountered in one sample with 101 ng/L, whereas its human metabolite, fenofibric acid has been detected in all samples at concentrations as high as 117 ng/L. Table 2 also excludes some compounds not systematically analyzed during the sampling campaign due to improvements in the analytical procedure. Certain compounds such as terbutaline or simazine have been encountered in several samples, but these data have not been considered statistically significant because they were found at concentrations so close to the LOQ that removal efficiency in STP could not be properly assessed and, therefore, they were not included in Table 2. Among them, diazepam was detected in half of the samples with a maximum concentration of 8 ng/L in the influent and 5 ng/L in the effluent, the average in both cases being near 3 ng/L (that equals LOQ). Also, mepivicaine was found in 7 samples with an average concentration of 8 ng/L in the influent and 7 ng/L in the effluent and maximum concentration in the influent of 14 ng/L. Paraxanthine, caffeine and acetaminophen were the individual pollutants usually found in higher concentration, with averages near 20 ppb in the influent. N-formyl-4-amino-antipiryne (4-FAA) and galaxolide also exhibited averages in the mg/L level. The fact that caffeine is the dominant micropollutant in STP is not new. Caffeine has been detected in many surface streams and STP effluents in concentration as high as 230 mg/L (Ternes et al., 2001; Heberer et al., 2002). Recently, Wilcox et al. (2009) also report caffeine, paraxanthine and acetaminophen as the most frequently detected compounds, in the influent of conventional septic systems. Mun˜oz et al. (2009) found several substances at mg/L level, with the highest concentrations corresponding to caffeine, acetaminophen, atenolol, and paraxanthine, that exceeded 40 mg/L each. In this work, atenolol was part of the group of compounds found with averages over the ppb limit that included ciprofloxacin, hydrochlorthiazide, ibuprofen, N-acetyl-4-amino-antipiryne (4-AAA), naproxen, nicotine, and ofloxacin. The occurrence of dipyrone (metamizol) residues in STP effluents has been less frequently assessed, but Feldmann et al. (2008) have recently reported concentrations up to 7 mg L1 in influents and effluents of STP in Berlin. These dipyrone residues have been attributed to effluents originated in hospitals more than to private households. Nicotine was found in concentrations as high as 12 mg/L in the influent and 148 ng/L in the effluent. Cotinine, its major urinary metabolite, was detected in the biological effluent with a concentration of 100 ng/L. These values agree with those published by Buerge et al. (2008) who found concentrations of cotinine of approximately 1–10 mg/L in untreated wastewater, and 0.01–0.6 mg/L in the treated effluent, corresponding to elimination efficiencies of over 90%.
582
water research 44 (2010) 578–588
Table 2 – Concentrations of pollutants in the influent and effluent of the studied STP calculated for compounds detected over LOQ in at least four influent samples along the monitoring programme. Averages and removal efficiencies have been calculated excluding concentrations below LOQ. Compound
4-aminoantipyrine (4-AA) 4-methylaminoantipyrine (4-MAA) Acetaminophen Antipyrine Atenolol Bezafibrate Benzophenone-3 Caffeine Carbamazepine Ciprofloxacin Clofibric acid Codeine Diclofenac Diuron Erythromycin Fenofibric acid Fluoxethine Furosemide Galaxolide Gemfibrozil Hydrochlorothiazide Ibuprofen Indomethacine Ketoprofen Ketorolac Mefenamic acid Metoprolol Metronidazole N-acetyl-4-amino-antipiryne (4-AAA) Naproxen N-formyl-4-amino-antipiryne (4-FAA) Nicotine Ofloxacin Omeprazole Paraxanthine Propanolol Ranitidine Sulfamethoxazole Tonalide Triclosan Trimethoprim
pKaa
4.3 4.3 9.4 1.4 9.6 3.3 7.6 10.4 13.9 8.9 3.2 8.2 4.2 NA 8.9 2.9 10.1 3.9 – 4.7 7.9 4.9 4.5 4.5 3.5 4.2 9.6 2.4 4.6 4.2 5.0 8.0 7.9 7.1 8.5 9.4 1.9 5.7 – 7.8 6.8
Influent (ng/L)
Effluent (ng/L)
Maximum
Minimum
Average
Maximum
Minimum
Average
3325 1894 37 458 72 2432 361 904 65 625 173 13 625 127 2087 561 196 2310 117 1827 1051 24 971 17 055 10 018 4113 113 801 2793 220 52 165 22 200 5228 71 000 11 671 5286 2134 98 500 61 1466 530 1932 2417 197
262 314 1571
1517 880 23 202 40 1197 141 393 22 849 129 5524 26 521 232 109 346 79 585 413 10 022 3525 2514 2687 42 441 407 141 20 90 8333 2363 17 579 4368 2275 365 26 722 36 524 279 952 860 104
2253 1098
127 34
676 291
Removal efficiency (%) 55.4 66.9 100 32.8 14.4 9.1 78.2 94.9 9.5 57.0 54.2 69.3 5.0 61.5 4.3 1.3 61.9 59.8 87.8 76.0 53.2 95.0 11.1 11.2 43.9 1.8 6.5 38.7 46.1 60.9 68.2 98.7 64.1 8.5 96.9 1.0 31.2 17.3 84.7 74.5 5.1
a Mun˜oz et al. (2008).
Concerning the less polar compounds, the UV filters 3-(4-methylbenzylidene) camphor, octocrylene and ethylhexyl methoxycinnamate were detected with maximum concentrations in the effluent of the STP of 55 ng/L, 114 ng/L and 234 ng/L, respectively. Other works reported higher mean for the secondary effluent or a STP of 70 ng/L and <10 ng/L for 3-(4-methylbenzylidene) camphor and octocrylene, respectively (Kupper et al., 2006). Celestolide, a synthetic musk was detected in low concentrations with a maximum of 30 ng/L in the influent. Several other compounds, not checked in the influent, were detected in the treated wastewater. It is the case of famotidine with concentration over 1 mg/L or the antiacid lansoprazole, 337 ng/L. Minor constituents include
loratadine, 29 ng/L, musk xylene, 52 ng/L or the antibiotic norfloxacin, with 38 ng/L.
3.2.
Removal in STP
As indicated before, Table 2 shows the pKa values for those compounds for which data were considered sufficient and statistically significant. The table also shows the maximum and minimum values for concentration together with the average for influent and effluent to the biological treatment sequence. The removal efficiency for every individual compound was determined from average concentrations calculated excluding samples whose concentrations fell
583
water research 44 (2010) 578–588
2008). The removal efficiency obtained in this work for the compounds indicated in Table 2 has been related to Dow and the results represented in Fig. 1. For a significant group of compounds, ranging from ketorolac (44%) to galaxolide (88%), a clear relationship is observed between removal efficiency and Dow. A group of five compounds were almost completely removed in the STP even they present relatively low Dow values. They are the metabolite of caffeine paraxanthine, caffeine itself, acetaminophen (paracetamol), ibuprofen and nicotine, compounds otherwise usually found in high concentrations in raw urban wastewater (Table 2). A similar result was reported by Mun˜oz et al. (2008) who reported concentrations of caffeine, acetaminophen, atenolol, and paraxanthine that exceeded 40 mg/L each but a substantial removal in an activated sludge STP (90%, >99%, 43%, and 67%, respectively). Joss et al. (2005) reported removal efficiencies for ibuprofen beyond its quantification limit (>90%) and also coincident are data corresponding to naproxen (50–80%). This work also identified 14 compounds with removal efficiencies fell below 20%, but exhibiting intermediate Dow values (2 < Dow < 2.5). They have been marked by the lower square in Fig. 1 and are the beta-blockers atenolol, metoprolol and propanolol; the lipid regulator bezafibrate and the metabolite of fenofibrate fenofibric acid; the antibiotics erythromycin, sulfamethoxazole and trimethoprim; the antiinflammatories diclofenac, indomethacin, ketoprofen and mefenamic acid; the antiepileptic carbamazepine and the antiacid omeprazole. Joss et al. (2005) reported no significant removal of sulfamethoxazole and carbamazepine and partial elimination of diclofenac, results generally coincident with this work. For musk fragrances, galaxolide and tonalide, Joss
below LOQ. It is well known that during wastewater treatment in STP, PPCP as well as their metabolites partition into the solid phase or remain dissolved depending on their hydrophobicity. The hydrophobicity of a neutral compound can be expressed as its octanol–water partition coefficient, Kow, but in the case of compounds that can exist in ionized form, the acid–base equilibrium must also be taken into account. The pH-dependent or apparent octanol–water distribution coefficient, Dow, that considers both the dissociation constant of acidic of basic solutes, pKa, and the current pH of wastewater can be derived from the Herderson–Hasselbalch equations (Scheytt et al., 2005). For acidic compounds, like indomethacin or naproxen, that are dissociated at the pH of wastewater, the equation yields: Dow ¼
Kow 1 þ 10pHpKa
(1)
In the case of basic drugs, the apparent partition coefficient can be expressed using the pKa for the corresponding conjugate acid: Dow ¼
Kow 1 þ 10pKa pH
(2)
Codeine or fluoxethine are among compounds that are substantially dissociated at ambient pH, whose average was 7.61 0.39, with boundaries representing the 95% confidence interval. For neutral substances, Dow ¼ Kow. There is experimental evidence that the removal of organic pollutants in STP is largely controlled by sorption process with solid–water distribution coefficients being a function of the octanol–water distribution coefficient Dow (Carballa et al., 100
1
90
3
4 5
2
26 25
80
23
Removal efficiency (%)
22
24
17
70 60
8 9
7
50
11 10 12 16
18 19 14 21 15
6
40 30 20
28 27
10 0 -8.00
29
-6.00
-4.00
-2.00
32 31 30
35
39
34 36
0.00
37
2.00
40
4.00
6.00
8.00
log Dow Fig. 1 – Removal efficiency during conventional activated sludge treatment: (1) paraxanthine, (2) caffeine, (3) acetaminophen,(4) nicotine, (5) ibuprofen,(6) ketorolac,(7) clofibric acid,(8) furosemide, (9) ciprofloxacin,(10) fluoxethine, (11) ofloxacin,(12) naproxen,(13) hydrochlorothiazide, (14) 4-aminoantipyrine, (15) metronidazole, (16) N-acetyl-4-aminoantipiryne, (17) codeine, (18) N-formyl-4-amino-antipiryne, (19) 4-methylaminoantipyrine, (20) ranitidine, (21) antipyrine, (22) gemfibrozil, (23) benzophenone-3,(24) triclosan,(25) tonalide, (26) galaxolide, (27) atenolol, (28) sulfamethoxazole, (29) fenofibric acid, (30) metoprolol, (31) bezafibrate, (32) ketoprofen, (33) trimethoprim, (34) diclofenac, (35) indomethacine, (36) propanolol, (37) mefenamic acid, (38) omeprazole, (39) carbamazepine, (40) erythromycin.
584
water research 44 (2010) 578–588
et al. (2005) reported, however, removal efficiencies somewhat lower (>50%) than those found in this work (w85%). Carballa et al. (2004) also reported overall removal efficiencies of the ranging between 70% and 90% for fragrances in coincidence with this work and generally higher for antiinflammatories (40–60%) and sulfamethoxazole. In a further work, Carballa et al. (2007) indicated different removal efficiencies for several PPCP during the anaerobic digestion of sewage sludge with high efficiencies for naproxen or sulfamethoxazole. In general, a high variability can be expected as a consequence of different pollutant concentrations, changes in the processes of the STP as well as operational conditions (Liu et al., 2008).
3.3.
Removal by ozonation
The ozonation of a dissolved organic compound may take place by direct reaction with ozone molecule and also through the action of secondary oxidants produced from ozone in aqueous medium. Among them, the strongest oxidant is hydroxyl radical, generally associated with the oxidation of the more refractory substances. A mass balance to a given compound in solution yields the following expression: dCi ¼ kO3 Ci CO3 þ kHO$ Ci CHO$ dt
(3)
Elovitz and von Gunten (1999) introduced a parameter Rct defined as the relationship between the concentrations of ozone and hydroxyl radical at any moment. Based on the observation that Rct can be considered constant at least though certain periods of the ozonation runs, the concentration of the two main oxidants involved in ozonation reactions can be related so that Eq. (3) can be solved without experimental determination of CHO$ : Z Z t Z t t Ci CO3 dt þ kHO$ CHO$ dt ¼ kO3 þ Rct kHO$ CO3 dt ln ¼ kO3 Cio 0 0 0 Z t ¼ kR CO3 dt
2003) mass transfer is likely to limit the ozonation rate during the first reaction minutes. Once ozone appears in solution, the following inequality holds: X kL a CO3 CO3 > kO3 Ci;o CO3
(6)
i
Therefore an upper limit can be obtained for Ha as follows:
Ha <
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi C kL a COO3 1 DO3 3
(7)
kL
where DO3 is the diffusivity of ozone in water (1.77 109 m2 s1). The equilibrium concentration of ozone CO3 was calculated from Henry’s law using the correlation of Rischbieter et al. (2000) from the concentration in the gas phase that was 9.4 g/Nm3. The value of the mass transfer coefficient, kL ¼ 5.5 105 m s1, was calculated by using the correlation of Calderbank and Moo-Young (1961). At the beginning of the run, there was no ozone in solution and Ha is supposed to reach high values. After about three minutes, the concentration of ozone that increased during runs approaching equilibrium was over 2 103 mM. This, considering that CO3 ¼ 0:05 mM, ensures Ha < 0.3 and, therefore, the kinetic regime was slow and for the sample taken at 4 min and those taken thereafter. Second order kinetic constants could be obtained as indicated before for bezafibrate, cotinine, diuron, ketoprofen and metronidazole. The logarithmic concentration decay is represented against the integral ozone dose in Fig. 2 in which, the experimental points represent data from samples taken at 4, 6, 10 and 15 min, the first being considered the initial concentration, Cio. The comparison with literature data shows that the transformation of ozone-resistant micropollutants takes place primarily via indirect radical oxidation inhibited by the wastewater matrix. The values reported in the literature for the second order direct and indirect rate constants are: bezafibrate, kO3 ¼ 590 M1 s1 and kHO$ ¼ 7:40 109 M1 s1 in Huber et al. (2005); ketoprofen, kO3 ¼ 0:4 M1 s1 and kHO$ ¼ 8:40 109 M1 s1 in Real et al. (2009), diuron,
0
(4)
Ha ¼
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi kO3 Ci;o DO3 kL
(5)
In the case of wastewater treatment, ozone reacts with many compounds in a complex parallel reaction system so P that kO3 Ci;o can be substituted by kO3 Ci;o . As raw wastewater i contains a number of compounds whose second order direct reaction constants with ozone are very large (Huber et al.,
6
4
5
3
ln(Cio/Ci)
4 3
2
ln(Cio/Ci)
The kinetic parameter kR would behave as second order kinetic constant provided Rct is constant throughout the sampling period. As shown below, this work shows that Rct is constant during ozonation at least for the samples taken after ozone appeared in solution. The preceding findings also lie on the assumption of slow kinetic regime. The kinetics of a heterogeneous gas–liquid semicontinuous process is determined by the relative rates of absorption and chemical reaction and is characterized by Hatta number. It represents the maximum rate of chemical reaction relative to the maximum rate of mass transfer, yielding, for a second order reaction, the following expression:
2
1 1
0
0
0.2
0.4
0.6
0.8
1
0
Fig. 2 – Logarithmic decay of the concentration of diuron (B), metronidazole (C), ketoprofen (D) bezafibrate (,) and cotinine (-) as a function of the integral ozone exposure. (Two y-axes have been used for clarity.)
water research 44 (2010) 578–588
kO3 ¼ 16:5 M1 s1 and kHO$ ¼ 4:6 109 M1 s1 in Benitez et al. (2007) and kO3 ¼ 14:7 M1 s1 and kHO$ ¼ 6:6 109 M1 s1 in De Laat et al. (1996) ; metronidazole, kO3 ¼ < 350 M1 s1 in Sa´nchez-Polo et al. (2008) and kHO$ ¼ 1:98 109 M1 s1 in Johnson and Mehrvar (2008). No data have been published for cotinine. The ozonation of bezafibrate was also studied by Dantas et al. (2007) who reported a direct kinetic constant kO3 ¼ 4:24 103 0:66 103 M1 s1 for pH 7 and kR ¼ 1.0 104 1.07 103 M1 s1 for pH 8. The reported direct rate constant for ketoprofen is particularly low and allows to calculate Rct by dividing the value of kR obtained in this work by kHO$ reported by Real et al. (2009). The value found, Rct ¼ 3.62 107, is very close to those that could be derived using Eq. (4) for bezafibrate and diuron. The partial contribution of the direct ozone and radical pathways can be calculated for a given organic compound as follows: fOH ¼
kHO$ CHO$ Ci kHO$ Rct ¼ kHO$ CHO$ Ci þ kO3 Ci kHO$ Rct þ kO3
(8)
Using literature constants, the fraction degraded by hydroxyl radicals were 0.819 for bezafibrate, 0.996 for diuron, 1.000 for ketoprofen and >0.887 for metronidazole. The contribution of direct ozone reaction is obviously low because the five compounds studied are relatively refractory to ozonation. By means of Eq. (8) and taking into account that kR ¼ kO3 þ kHO$ $Rct , the second order rate constants, kR, can be derived from literature data with the following results: metronidazole 1.07 103 M1 s1, bezafibrate 3.27 103 M1 s1 and diuron 2.41 103 M1 s1 (Benitez et al., 2007). The two last are in good agreement with experimental values. In the case of metronidazole, the low value obtained for Rct suggests that kHO could have been underestimated. The efficiency of ozonation for the removal of the main micropollutants whose concentration in biologically treated wastewater was >10 ng/L is indicated in Table 3. The table shows the evolution of the concentration for samples taken during ozonation up to a reaction time of 15 min and the amount of ozone required for a given degree of removal. Only concentrations above LOQ are shown. These LOQ apply for ozonated samples and are lower than those reported in the Appendix for untreated samples. They could be reached, because of the higher preconcentration factor applied to the ozonated samples (see experimental section) and the reduction in the matrix effects observed in these cleaner extracts. The amount of ozone transferred to the liquid at a certain reaction time, TOD, was determined, from the integration of the ozone absorption rate equation: 0 TODðtÞ ¼ kL a@CO3 t
Z 0
t
1 CO3 dtA
(9)
For the experimental conditions used in this work CO3 was always less then 10% of CO3 and, therefore, TOD(t) was essentially linear with time. The last right column of Table 3 shows the dose of ozone required for the complete removal (no detection) of a given compound or to achieve certain removal efficiency in cases where ozonation was unable to get complete oxidation in less than 15 min corresponding to a dose of ozone of 0.34 mmol/L of wastewater. It is to be noted that a change in the pH of wastewater modifies the ozone
585
doses required for a given effect. A pH increase causes a higher hydroxyl radical exposure increasing Rct and leading, for a similar integral radical exposure, to lower ozone doses. Also, it is important to point out that a change of pH may affect the direct ozonation rates of dissociating compounds. In our case, raising pH from that of raw wastewater to 8.5, at which the ozonation took place, could affect the ozonation rate of some proton-accepting compounds whose pKa w 8, particularly benzophenone-3, hydrochlorthiazide, nicotine, ofloxacin and triclosan. A group of 15 compounds rapidly disappear during the first 120 s on stream, with ozone doses <50 mM. These include codeine, diclofenac, indomethacin or naproxen, whose disappearance during the first minutes of ozonation has already been reported in a former work whose target was the mineralization of dissolved organics (Rosal et al., 2008a). Antipyrine, erythromycin, ketorolac, norfloxacin, propanolol, ranitidine, and trimethoprim became completely removed for doses <90 mM. Carbamazepine, ciprofloxacin, citalopram hydrobromide, hydrochlorothiazide, metoprolol, omeprazole, venlafaxine and the two metabolites of metamizol, (4-AAA and 4-FAA), required <130 mM of ozone to disappear. Higher ozone doses area necessary to remove atenolol, lansoprazole, loratadine, primidone, sulfamethoxazole and diuron. Another group of four personal care products were completely refractory to ozone, not being removed at all after 15 min of reaction. These were the UV filters 3-(4-methylbenzylidene) camphor and ethylkexyl methoxycinnamate together with the sunscreen agent benzophenone-3 and the aromatic nitro musk xylene. Finally, certain compounds were removed in different extents using an ozone dose of 340 mM. From ketoprofen, metronidazole or ofloxacin, with removal efficiencies over 95% to the musk ketone and the UV filter octocrylene, with 38% and 20% removal respectively. It is interesting to point out than some compounds like diclofenac, indomethacin or the beta-blockers atenolol, metoprolol and propanolol, which are poorly removed in the activated sludge conventional treatment, exhibit large ozonation rates that allow their removal from wastewater using moderate ozone doses. It is interesting to note that during the first part of the reaction there is no ozone in solution and therefore the concepts of homogeneous kinetics cannot be applied. During this period, the reaction must take place at the interface and the rate of reaction depends on the surface concentration, that is related to the polarity of the compound. This may explain why diclofenac and sulfamethoxazole, with similar direct ozonation constants (Huber et al., 2005) behave in a rather different way, the later being detected during a much longer period. The apparent octanol–water partition coefficient (see Appendix and Table 2) is, at pH 8.5, tow orders of magnitude larger diclofenac (0.21) than for sulfamethoxazole (1.91). A certain insight into the efficiency of the ozonation process can be reached by considering the global removal of micropollutants as a function of ozone doses. Our data showed that the removal of micropollutants reached 86% for a dose of ozone of 90 mM. For higher doses, the efficiency of removal considerably decreased with very limited gains over 130 mM. In fact, an additional ozone dose of 200 mM results in less than 1% removal of the compounds studied in this work.
586
water research 44 (2010) 578–588
Table 3 – Removal of pollutants contained in wastewater during ozonation. The ozone doses are those required to reach concentrations below the limit of quantification (LOQa) in treated samples. LOQa
0
2
4
6
10
15
Ozone doses for remotion
3-(4-methylbenzylidene) camphor 4-aminoantipyrine 4-methylaminoantipyrine (4-MAA) Antipyrine Atenolol Azithromycin Benzophenone-3 Bezafibrate
39 19 2 8 3 12 33 4
55 58 389 30 911 235 123 115
50 – – 16 655 – 89 72
65 – – – 265 – 100 67
39 – – – 24 – 102 37
72 – – – – – 119 15
54 – – – – – 119 4
Carbamazepine Carbamazepine epoxide Ciprofloxacin Citalopram hydrobromide Clarithromycin Codeine Cotinine
1 9 5 2 5 5 12
106 32 522 60 39 378 100
17 23 334 31 – – 61
2 19 28 4 – – 54
– 13 – – – – 48
– – – – – – 38
– – – – – – 28
1 1
433 100
– 60
– 46
– 25
– 6
– 1
10 15 14 2 82 23 1 1 2 2
72 234 1045 17 840 1486 332 707 37 162
16 274 – – – 749 50 461 – 156
– 322 – – – 552 18 199 – 102
– 231 – – – 178 19 – – 68
– 214 – – – 196 15 – – 18
– 204 – – – 173 15 – – 3
2 9 3 1 2 3 3
533 337 12 29 59 27 113
165 162 – 18 – 17 73
– 84 – 7 – 5 85
– 32 – 2 – – 56
– – – – – – 14
– – – – – – 3
Musk xylene Musk ketone N-acetyl-4-aminoantipyrine (4-AAA) Naproxen N-formyl-4-aminoanttipyrine (4-FAA) Nicotine
3 36 50 12 17 4
89 123 8605 109 1776 81
92 125 2419 – 471 12
95 140 101 – 21 10
89 95 – – – 13
98 105 – – – 10
91 76 – – – 14
Norfloxacin Octocrylene Ofloxacin
8 16 3
38 114 3594
56 115 276
– 113 18
– 81 11
– 95 9
– 91 10
Omeprazole Primidone Propanolol Propyphenazone Ranitidine Sulfamethoxazole Sulfapyridine Tonalide Triclosan
3 5 2 2 2 8 12 19 52
1015 80 32 23 111 95 50 188 246
231 86 7 – 3 39 – 131 55
7 65 – – – 19 – 130 72
– 40 – – – 15 – 53 79
– – – – – – – 67 70
– – – – – – – 53 53
2 6
73 179
7 127
– 21
– –
– –
– –
Not removed <50 mM <50 mM <90 mM <220 mM <50 mM Not removed Still detected at 340 mM <130 mM <220 mM <130 mM <130 mM <50 mM <50 mM 28% remained for 340 mM <50 mM Still detected at 340 mM <90 mM Not removed <50 mM <50 mM <50 mM 83% for 340 mM 95% for 340 mM <130 mM <50 mM Still detected at 340 mM <90 mM <220 mM <50 mM <220 mM <50 mM <130 mM Still detected at 340 mM Not removed 38% for 340 mM <130 mM <50 mM <130 mM Still detected at 340 mM <90 mM 20% for 340 mM Still detected at 340 mM <130 mM <220 mM <50 mM <90 mM <90 mM <220 mM <50 mM 72% for 340 mM Still detected at 340 mM <90 mM <130 mM
Ozonation time (min)
Diclofenac Diuron Erythromycin Ethylkexyl methoxycinnamate Famotidine Fluoxethine Furosemide Galaxolide Gemfibrozil Hydrochlorothiazide Indomethacine Ketoprofen Ketorolac Lansoprazole Lincomycin Loratadine Mefenamic acid Metoprolol Metronidazole
Trimethoprim Venlafaxine a LOQ calculated in ozonated samples.
kR (M1 s1)
3260 780
680 29
3890 200
3040 770
3100 780
water research 44 (2010) 578–588
4.
Conclusions
This research showed the regular presence of over seventy anthropogenic individual pollutants, some of which are encountered in relatively high amounts. In raw sewage, 25 compounds were detected in the mg/L range, 15 of which exceeded this level in yearly averages. Acetaminophen (paracetamol) and caffeine were persistently detected over 1 ppb in untreated wastewater. Galaxolide was also encountered in high concentration, but its occurrence showed a significant variability. Two metabolites, paraxanthine from caffeine and 4-FAA from metamizol were also found in high amounts in almost all samples. Other metabolites detected were 4-AA and 4-MAA from dipyrone, fenofibric acid from fenofibrate and 4-AAA also from metamizol. This findings stress the need for exploring not only the pattern PPCP, but also their metabolic or photodegradation intermediates. The efficiency of removal of PPCP in STP was roughly dependent on its hydrophobicity expressed as apparent octanol–water distribution coefficient, Dow, a parameter that takes into account the octanol–water partition coefficient, Kow, as well as the dissociation constant of acidic or basic compounds. For most compounds, the removal efficiency during biological treatment increased with hydrophobicity as expected considering the higher sorption of non-polar compounds on sludge. Certain compounds showed important deviations with a group of relatively polar substances formed by paraxanthine, caffeine, acetaminophen, ibuprofen and nicotine that were almost completely removed during biological treatment. Another group formed by 14 compounds exhibited removal efficiencies below 20%, even their octanol– water distribution coefficient was not particularly low. They are all important pharmaceuticals prescribed and delivered to sewage in high amounts and include the beta-blockers atenolol, metoprolol and propanolol; the lipid regulator bezafibrate, fenofibric acid; the antibiotics erythromycin, sulfamethoxazole and trimethoprim; the antiinflammatories diclofenac, indomethacin, ketoprofen and mefenamic acid; the antiepileptic carbamazepine and the antiacid omeprazole. An ozonation treatment yielded high removal efficiencies of most individual pollutants detected in treated wastewater. The kinetic analysis of the part of the run that takes place in the slow kinetic regime, allowed the determination of second order kinetic constants for the ozonation of bezafibrate, cotinine, diuron, ketoprofen and metronidazole in its wastewater matrix. This work shows that Rct, a concept developed for drinking water, can be applied for the ozonation of wastewater as it was constant at least for the samples taken after ozone appeared in solution. The ratio of the concentrations of ozone and hydroxyl radical, Rct, during this period was 3.62 107 and the fraction degraded by hydroxyl radicals was in the 0.819–1.000 range for the aforementioned compounds. Even when most compounds disappear for doses lower than 340 mM, and most for less than 90 mM, some pollutants, essentially personal care products were not significantly removed during ozonation. These were the UV filters 3-(4-methylbenzylidene) camphor and Ethylkexyl methoxycinnamate, the sunscreen agent Benzophenone-3 and the aromatic nitro musk xylene. On the other hand, certain polar
587
compounds like diclofenac, indomethacin or the betablockers atenolol, metoprolol and propanolol, which are poorly removed in the activated sludge conventional treatment, exhibit large ozonation rates and can be removed from treated wastewater using moderate ozone doses.
Acknowledgements The authors wish to express their gratitude to the Ministry of Education of Spain (Contract CONSOLIDER-INGENIO 2010 CSD2006-00044), the Direccio´n General de Universidades e Investigacio´n de la Comunidad de Madrid, Research network 0505/AMB-0395.
Appendix. Supplementary information Supplementary data associated with this article can be found in the doi: 10.1016/j.watres.2009.07.004.
references
Beltra´n, F.J., 2004. Ozone Reaction Kinetics for Water and Wastewater Systems. CRC, Boca Raton, pp. 113–149. Benitez, F.J., Real, F.J., Acero, J.L., Garcia, C., 2007. Kinetics of the transformation of phenyl-urea herbicides during ozonation of natural waters: rate constants. Water Res. 41, 4073–4084. Bueno, M.J.M., Agu¨era, A., Go´mez, M.J., Hernando, M.D., Garcı´a, J.F., Ferna´ndez-Alba, A.R., 2007. Application of liquid chromatography/quadrupole-linear ion trap mass spectrometry and time-of-flight mass spectrometry to the determination of pharmaceuticals and related contaminants in wastewater. Anal. Chem. 79, 9372–9384. Buerge, I.J., Kahle, M., Buser, H.R., Mu¨ller, M.D., Poiger, T., 2008. Nicotine derivatives in wastewater and surface waters: application as chemical markers for domestic wastewater. Environ. Sci. Technol. 42, 6354–6360. Calderbank, P.H., Moo-Young, M.B., 1961. The continuous phase heat and mass transfer properties of dispersions. Chem. Eng. Sci. 16, 39–54. Carballa, M., Fink, G., Omil, F., Lema, J.M., Ternes, T., 2008. Determination of the solid–water distribution coefficient (Kd) for pharmaceuticals, estrogens and musk fragrances in digested sludge. Water Res. 42, 287–295. Carballa, M., Omil, F., Lema, J.M., Llompart, M., Garcı´a-Jares, C., Rodrı´guez, I., Go´mez, M., Ternes, T., 2004. Behavior of pharmaceuticals, cosmetics and hormones in a sewage treatment plant. Water Res. 38, 2918–2926. Carballa, M., Omil, F., Ternes, T., Lema, J.M., 2007. Fate of pharmaceutical and personal care products (PPCPs) during anaerobic digestion of sewage sludge. Water Res. 41, 2139–2150. Castiglioni, S., Bagnati, R., Fanelli, R., Pomati, F., Calamari, D., Zuccato, E., 2006. Removal of pharmaceuticals in sewage treatment plants in Italy. Environ. Sci. Technol. 40, 357–363. 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. J. Chem. Technol. Biotechnol. 83, 769–776.
588
water research 44 (2010) 578–588
Dantas, R.F., Centerino, M., Marotta, R., Sans, C., Esplugas, S., Andreozzi, A., 2007. Bezafibrate removal by means of ozonation: primary intermediates, kinetics, and toxicity assessment. Water Res. 41, 2525–2532. Daughton, C.G., Ternes, T.A., 1999. Pharmaceuticals and personal care products in the environment: agents of subtle change? Environ. Health Perspect. 107, 907–938. De Laat, J., Maouala-Makata, P., Dore, M., 1996. Rate constants for reactions of ozone and hydroxyl radicals with several phenylureas and acetamides. Environ. Technol. 17, 707–716. Elovitz, M.S., von Gunten, U., 1999. Hydroxyl radical/ozone ratios during ozonation processes I. The Rct concept. Ozone Sci. Eng. 21, 239–260. Ernst, M., Jekel, M., 1999. Advanced treatment combination for groundwater recharge of municipal wastewater by nanofiltration and ozonation. Water Sci. Technol. 40, 277–284. Feldmann, D.F., Zuehlke, S., Heberer, T., 2008. Occurrence, fate and assessment of polar metamizole (dipyrone) residues in hospital and municipal wastewater. Chemosphere 71, 1754–1764. Gogate, P.R., Pandit, A.B., 2004a. A review of imperative technologies for wastewater treatment I: oxidation technologies at ambient conditions. Adv. Environ. Res. 8, 501–551. Gogate, P.R., Pandit, A.B., 2004b. A review of imperative technologies for wastewater treatment II: hybrid methods. Adv. Environ. Res. 8, 553–597. Go´mez, M.J., Go´mez-Ramos, M.M., Agu¨era, A., Mezcua, M., Herrera, S., Ferna´ndez-Alba, A.R., 2009. A new gas chromatography/mass spectrometry method for the simultaneous analysis of target and non-target organic contaminants in waters. J. Chromatogr. A, 1216, 4071–4082. Halling-Sørensen, B., Nielsen, B.N., Lanzky, P.F., Ingerslev, F., Lu¨tzhoft, H.C.H., Jorgensen, S.E., 1998. Occurrence, fate and effects of pharmaceutical substances in the environment, a review. Chemosphere 36, 357–394. Heberer, T., Reddersen, K., Mechlinski, A., 2002. From municipal sewage to drinking water: fate and removal of pharmaceutical residues in the aquatic environment in urban areas. Water Sci. Technol. 46, 81–88. Huber, M.M., Canonica, S., Park, G.Y., Gunten, U., 2003. Oxidation of pharmaceuticals during ozonation and advanced oxidation processes. Environ. Sci. Technol. 37, 1016–1024. Huber, M.M., Go¨bel, A., Joss, A., Hermann, N., Lo¨ffler, D., Mcardell, C.S., Ried, A., Siegrist, H., Ternes, T.A., von Gunten, U., 2005. Oxidation of pharmaceuticals during ozonation of municipal wastewater effluents: a pilot study. Environ. Sci. Technol. 39, 4290–4299. Ikehata, K., Naghashkar, N.J., El-Din, M.G., 2006. Degradation of aqueous pharmaceuticals by ozonation and advanced ozonation processes: a review. Ozone Sci. Eng. 28, 353–414. Johnson, M.B., Mehrvar, M., 2008. Aqueous metronidazole degradation by UV/H2O2 process in single- and multi-lamp tubular photoreactors: kinetics and reactor design. Ind. Eng. Chem. Res. 47, 6525–6537. 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, 3139–3152. Kupper, T., Plagellat, C., Bra¨ndli, R.C., de Alencastro, L.F., Grandjean, D., Tarradellas, J., 2006. Fate and removal of polycyclic musks, UV filters and biocides during wastewater treatment. Water Res. 40, 2603–2612. Liu, Z., Kanjo, Y., Mizutani, S., 2008. Removal mechanisms for endocrine disrupting compounds (EDCs) in wastewater treatment d physical means, biodegradation, and chemical advanced oxidation: a review. Sci. Total Environ.. doi:10.1016/ j.scitotenv.2008.08.039.
Metcalfe, C., Miao, X.S., Hua, W., Letcher, R., Servos, M., 2004. Pharmaceuticals in the Canadian environment. In: Ku¨mmerer, K. (Ed.), Pharmaceuticals in the Environment, Sources, Fate Effects and Risks. Springer, pp. 67–90. Mun˜oz, I., Go´mez, M.J., Molina-Dı´az, A., Huijbregts, M.A.J., Ferna´ndez-Alba, A.R., Garcı´a-Calvo, E., 2008. Ranking potential impacts of priority and emerging pollutants in urban wastewater through life cycle impact assessment. Chemosphere 74, 37–44. Mun˜oz, I., Rodrı´guez, A., Rosal, R., Ferna´ndez-Alba, A.R., 2009. Life cycle assessment of urban wastewater reuse with ozonation as tertiary treatment. A focus on toxicity-related impacts. Sci. Total Environ. 407, 1245–1256. Radjenovic, J., Petrovic, M., Barcelo´, D., 2009. Fate and distribution of pharmaceuticals in wastewater and sewage sludge of the conventional activated sludge (CAS) and advanced membrane bioreactor (MBR) treatment. Water Res. 43, 831–841. Raknes, K.L., 2005. Ozone in Drinking Water Treatment: Process Design, Operation, and Optimization. American Water Works Association, Denver. Real, F.J., Benitez, F.J., Acero, J.L., Sagasti, J.J.P., Casas, F., 2009. Kinetics of the chemical oxidation of the pharmaceuticals primidone, ketoprofen, and diatrizoate in ultrapure and natural waters. Ind. Eng. Chem. Res. 48, 3380–3388. Rischbieter, E., Stein, H., Shumpe, A., 2000. Ozone solubilities in water and aqueous salt solutions. J. Chem. Eng. Data 45, 338–340. 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., 2008a. Removal of pharmaceuticals and kinetics of mineralization by O3/H2O2 in a biotreated municipal wastewater. Water Res. 42, 3719–3728. Rosal, R., Rodrı´guez, A., Perdigo´n-Melo´n, J.A., Petre, A., Garcı´aCalvo, E., 2008b. Oxidation of dissolved organic matter in the effluent of a sewage treatment plant by ozone combined with hydrogen peroxide (O3/H2O2). Chem. Eng. J. 149, 311–318. Sa´nchez-Polo, M., Rivera-Utrilla, J., Prados-Joya, G., FerroGarcıa, M.A., Bautista-Toledo, I., 2008. Removal of pharmaceutical compounds, nitroimidazoles, from waters by using the ozone/carbon system. Water Res. 42, 4163–4171. Scheytt, T., Mersmann, P., Lindsta¨dt, R., Heberer, T., 2005. 1-Octanol/water partition coefficients of 5 pharmaceuticals from human medical care: carbamezepine, clofibric acid, diclofenac, ibuprofen and propyphenazone. Water Air Soil Pollut. 165, 3–11. Tauxe-Wuersch, A., De Alencastro, L.F., Grandjean, D., Tarradellas, J., 2005. Occurrence of several acidic drugs in sewage treatment plants in Switzerland and risk assessment. Water Res. 39, 1761–1772. Ternes, T.A., 1998. Occurrence of drugs in German sewage treatment plants and rivers. Water Res. 32, 3245–3260. Ternes, T., Bonerz, M., Schmidt, T., 2001. Determination of neutral pharmaceuticals in wastewater and rivers by liquid chromatography–electrospray tandem mass spectrometry. J. Chromatogr. A 938, 175–185. Terzic´, S., Senta, I., Ahel, M., Gros, M., Petrovic´, M., Barcelo, D., Mu¨ller, J., Knepper, T., Martı´, I., Ventura, F., Jovancˇic´, P., Jabucˇar, D., 2008. Occurrence and fate of emerging wastewater contaminants in Western Balkan Region. Sci. Total Environ. 399, 66–77. Wilcox, J.D., Bahr, J.M., Hedman, C.J., Hemming, J.D.C., Barman, M.A.E., Bradbury, K.R., 2009. Removal of organic wastewater contaminants in septic systems using advanced treatment technologies. J. Environ. Qual. 38, 149–156.
water research 44 (2010) 589–595
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Degradation of the emerging contaminant ibuprofen in water by photo-Fenton Fabiola Me´ndez-Arriaga*, Santiago Esplugas, Jaime Gime´nez Departamento de Ingenierı´a Quı´mica, Facultad de Quı´mica, Universidad de Barcelona, C/ Martı´ i Franque`s 1, 08028 Barcelona, Spain
article info
abstract
Article history:
In this study the degradation of the worldwide Non-Steroidal Anti-Inflammatory Drug
Received 18 April 2009
(NSAID) ibuprofen (IBP) by photo-Fenton reaction by use of solar artificial irradiation was
Received in revised form
carried out. Non-photocatalytic experiments (complex formation, photolysis and UV/Vis-
3 July 2009
H2O2 oxidation) were executed to evaluate the isolated effects and additional differentiated
Accepted 8 July 2009
degradation pathways of IBP. The solar photolysis cleavage of H2O2 generates hydroxyl-
Available online 15 July 2009
ated-IBP byproducts without mineralization. Fenton reaction, however promotes hydroxylation with a 10% contamination in form of a mineralization. In contrast photo-Fenton in
Keywords:
addition promotes the decarboxylation of IBP and its total depletion is observed. In absence
AOP
of H2O2 a decrease of IBP was observed in the Fe(II)/UV–Vis process due to the complex
Photo-Fenton
formation between iron and the IBP-carboxylic moiety. The degradation pathway can be
Non-steroidal anti-inflammatory
described as an interconnected and successive principal decarboxylation and hydroxyl-
drug ibuprofen
ation steps. TOC depletion of 40% was observed in photo-Fenton degradation. The iron-IBP binding was the key-point of the decarboxylation pathway. Both decarboxylation and hydroxylation mechanisms, as individual or parallel process are responsible for IBP removal in Fenton and photo-Fenton systems. An increase in the biodegradability of the final effluent after photo-Fenton treatment was observed. Final BOD5 of 25 mg L1 was reached in contrast to the initial BOD5 shown by the untreated IBP solution (BOD5 < 1 mg L1). The increase in the biodegradability of the photo-Fenton degradation byproducts opens the possibility for a complete remediation with a final post-biological treatment. ª 2009 Elsevier Ltd. All rights reserved.
1.
Introduction
The recent public interest regarding the presence of pharmaceutical pollutants in water has raised important concerns due to the unknown environmental impact and possible damages to the botany and fauna present in aquatic systems (Halling-Sorensen et al., 1998; Ternes et al., 2004). One of the most consumed medications corresponds to the classification of the NonSteroidal Anti-Inflammatory Drugs (NSAIDs) with more than 70 million annual prescriptions in the world (Takagi et al., 2006). Moreover, several recent reports have confirmed the presence of
the NSAID ibuprofen (IBP), in effluents of wastewater treatment plants (WTPs) (Alonso et al., 2006; Go´mez et al., 2008). Concentrations of IBP in the environment are reported between 10 ng L1 and 169 mg L1 (Santos et al., 2007). A significant removal rate of this drug is reached by biological oxidation, in some cases more than 70% (Carballa et al., 2006). In spite of, its apparently high biodegradation rate, the ecological risk remains still high, due to the main byproducts generated during the biological oxidation. These byproducts, hydroxyl-IBP and carboxy-IBP, have shown quite similar toxicological consequences in aquatic environment (Roberts and Thomas, 2006).
* Corresponding author. Tel.: þ34 93 403 9789; fax: þ34 93 402 1291. E-mail address:
[email protected] (F. Me´ndez-Arriaga). 0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2009.07.009
590
water research 44 (2010) 589–595
On the other hand, recent research has shown promising results in the removal of pharmaceutical pollutants through the application of Advanced Oxidation Processes (AOPs). The AOPs are oxidative processes applied for treatment of contaminants in water, soils and air, based on the presence and reactivity of the hydroxyl radicals (OH) which are generated in atmospheric or subcritical conditions of temperature and pressure with or without catalyst and/or reactive energy (electrochemical, UV–Vis or ultrasounds) (Me´ndez-Arriaga et al., 2009). The final common byproducts after AOPs have been reported to be easily degraded by biological oxidation (Torres et al., 2003). Although AOPs use different reacting systems (homogeneous or heterogeneous phases, irradiated or dark conditions, etc.), they are mainly characterized by the production of hydroxyl radicals with consecutive unselective attack on the organic material. Several AOPs have been employed to remove NSAIDs (as diclofenac, naproxen, IBP, ketoprofen, etc.) but special focus was placed on the heterogeneous photocatalyst with TiO2, photo-Fenton solar process, ozonation process, electrochemistry (Sire´s Sadornil Ignacio, 2006) and more recently on ultrasound energy (Hartmann et al., 2008; Me´ndez-Arriaga et al., 2008b). Photo-Fenton reaction is one of the most widely used AOPs due to its environment friendly application of solar energy. The extremely complex chemistry of the oxidation system based on the reaction between iron and hydrogen peroxide, the called Fenton reaction, has become of great interest for environmental applications. Well documented information, in a critical compilation about the basis of Fenton chemistry as well as its applications to wastewater treatment, can be founded in Venkatadri and Peters (1993), Bigda (1995), Chen and Pignatello (1999), Tarr (2003), Burkitt (2003), Neyens and Baeyens (2003), SafarzadehAmiri et al. (1996) and more recently in a widely recommended critical review, written by Pignatello et al. (2006). Photo-Fenton, an attractive alternative for degradation of emerging contaminants, is based on the redox reaction between Fe(II) and H2O2 with the concomitant generation of OH and consecutive catalytic role of iron (Fe(II)/Fe(III)) and it is improved under UV–Vis irradiation. Even the Fenton and photo-Fenton reactions have been applied for wide types of organic contaminants, their use for the degradation of pharmaceutical compounds has been applied to a smaller extent than other AOPs like photocatalysis or O3. The Fenton and photo-Fenton processes have been employed in the treatment of wastewater containing pharmaceutical pollutants such as azathriopine, asparginase, DCF, penicillin, a-methyl-phenylglycine, metronidazole, lincomicyn metabolites, sulfamethazine, sulfamethoxazole, amoxicillin, bezafibrate, paracetamol, biguanides, guanidine, triamines, pharmaceutical industry and hospital wastewaters among others (Klaviaroti et al., 2008; Gonza´lez et al., 2007; Pe´rez-Moya et al., 2007; Pe´rez-Estrada et al., 2005). Degradation of IBP by photo-Fenton reaction has not been reported yet. Therefore, the goals of this work were to evaluate the photo-Fenton ability to degrade IBP in an aqueous solution with a simulated solar irradiation source as well as to assess the biodegradability of the identified byproducts in the effluent after the treatment.
2. Reagents, equipments and analytical procedures IBP (Sigma–Aldrich), FeSO4$7H2O (Panreac) and H2O2 (30% w/v) (Panreac) were used without pre-treatment. The experimental device has been already described in earlier communications (Me´ndez-Arriaga et al., 2008a). Initial IBP aqueous solution (0.87 mM, pH 6.25 0.25) was pumped from the feeding stirred tank to the Duran-photoreactor (0.08 L) placed inside the Solarbox (Co.Fo.Me.Gra, Italy). The Xe lamp (Phillips OP 1 kW, 6.9 mEinsteins s1 in 290–400 nm wavelength range), placed top-inside of the Solarbox, was switched on, and samples at different irradiation times were withdrawn. In all experiments, fresh H2O2 and Fe(II)-acid dilutions were daily prepared. For photo-Fenton experiments a peristaltic pump was used to dose solutions of H2O2 and H2SO4 in the entrance of the photoreactor. Several concentrations of H2O2 were chosen, from 0 to 0.32 mM (0.04, 0.08, 0.16 and 0.32 mM), with a dose rate of 0.032 mM min1. pH adjust was carried out with H2SO4 diluted solution reaching a final pH 3. IBP concentration was monitored by HPLC by using an RP-C18 Trace Extrasil OD52-5 Micromet 25 0.46 Teknockroma column in isocratic mode with acetonitrile (99.8% Panreac) and acetic acid (0.25 M) in 75–25 proportion. Aliquots were also used to measure the residual concentration of H2O2 by spectrophotometric method (Nogueira et al., 2005) with a Perkin Elmer UV/VIS Lambda 20. Total Organic Carbon (TOC) measurements were obtained by a Shimadzu TOC-V CNS auto sampler instrument. BOD5 determinations were obtained according to Standard Methods (5120) by respirometric single measuring system OxiTop procedure. A Liquid Chromatography–Electrospray Ionization Jasco AS-2050 plus IS mass spectrometer (LC/ESI-MS, TOF Mariner) was employed to identify several byproducts.
3.
Results and discussion
3.1. Photolysis, H2O2 and UV–Vis/H2O2 photolysis effects Non-photocatalytic assays were undertaken to evaluate their isolated influence on the degradation of IBP. For the photolysis experiment, the IBP solution (initial concentration of 0.87 mM) was irradiated throughout the photoreactor during 2 h. On the other hand, several ratios of IBP-H2O2 (ranging from 1:0.001 to 1:10) were mixed in 30 mL vials at free pH conditions (pH 6.25 0.25), constant controlled temperature (30 C) and continuous dark stirring during 24 h. Furthermore, H2O2 photolysis process was carried out irradiating a solution of 0.87 mM of IBP with 0.32 mM of H2O2 during 2 h. Photolysis let IBP and TOC concentrations unchanged. The negligible degradation attained by photolysis was not unexpected, due to the low IBP molar absorption coefficient above 300 nm. Similar results were observed for all ratios IBP-H2O2 after 24 h darkness. Any difference in the original concentration of IBP was observed and TOC conversion showed an insignificant change, thus addition of peroxide in the dark gave a quite stable IBP solution. In contrast, UV–Vis/H2O2 process showed an evident decrease in the concentration of
591
water research 44 (2010) 589–595
IBP. When solution was irradiated in presence of H2O2, almost 40% of degradation took place after 2 h. The degradation of IBP can be attributed to the photochemical cleavage of hydrogen peroxide to yield hydroxyl radicals by light absorption (Tuhkanen, 2004):
1.00
IBP/ IBPo
0.75 0.50 0.25
H2O2 þ hv / 2OH
(1) 0.00
3.2.
Fenton dark and photo-Fenton reaction
For Fenton dark experiments, several Fe(II) concentrations (0.15 mM, 0.29 mM; 0.60 mM and 1.2 mM) were selected. Iron acid solutions were added to 1 L IBP solution (0.87 mM, pH 6.25 0.25) under stirring and controlled temperature (30 C). H2O2 (0.32 mM) was immediately added in the tank and samples were periodically withdrawn. On the other hand, photo-Fenton experiments were carried out with similar conditions in 1.5 L IBP solution as described in Section 2. Iron acid solution (1.2 mM) was added in the continuously flowed solution. Several concentrations of H2O2 were chosen, from 0 to 0.32 mM, with a dose rate of 0.032 mM min1. pH adjust was carried out with H2SO4 diluted solution reaching a final pH 3. Fig. 1 depicts the IBP conversion after 2 h of Fenton dark reaction. As higher Fe(II) concentration, higher IBP degradation is reached. IBP depletion by Fenton involves the chemical reaction with H2O2 as free OH generator (Haber and Weiss, 1934):
Fe(II) þ H2O2 / Fe(III) þ OH þ OH
Fe(III) þ H2O2 / Fe(II) þ O2H þ Hþ
(2)
(3)
Iron acts as catalyst by the simultaneous species Fe(II)/ Fe(III) present during the reaction meanwhile H2O2 is continually consumed forming OH. The increase of Fe(II) promotes IBP decrease, in the maximum case ca. of 60% corresponding to 10% of TOC reduction. Early report (Kennedy et al., 1990) suggests that for ratios equal or higher to 0.13 of iron/IBP all
0
30
60
90
120
Time (min) Fig. 1 – Effect of Fe(II) concentration on removal of IBP (0.87 mM) by Fenton dark reaction.(B) 0.15 mM Fe(II),(:)0.29 mM Fe(II); (,)0.60 mM Fe(II) and (A)1.2 mM Fe(II). In all cases 0.32 mM of H2O2.
the coordination sites of the metal are occupied by IBP and the OH production is hampered. However, in this work under the ratios tested – between 0.12 and 0.33 – the catalyst has shown to be reactive with H2O2 promoting the degradation of IBP. On the other hand, in photo-Fenton experiments more than 80% removal of IBP and almost 40% decrease of the initial TOC were achieved by using 0.32 mM H2O2 and 1.2 mM Fe(II). These values are four folds higher than the observed under dark conditions. Fig. 2 shows the IBP/IBPo for photo-Fenton process together with photolysis, H2O2/UV–Vis, Fenton and Fe(II)/UV–Vis process, and Fig. 3 shows the concomitant TOC removal for all the above described processes. Fig. 2 shows an evident change in the kinetic degradation of IBP by means of photo-Fenton reaction in comparison with the dark Fenton series above described. The increase in the reactivity of the photo-Fenton reaction is attributed to the simultaneous OH generation from the H2O2/UV–Vis and photo-Fenton systems. The photo-assisted reaction (photoFenton) gives typically faster rates of degradation and mineralization than the thermal (‘‘dark’’) reaction. In photoFenton reaction the production of OH is significantly increased under illuminated conditions due to the stimulating photoreduction of Fe(III) to Fe(II):
1.00
Photolysis
0.75
IBP/ IBPo
In the configuration of our system (Xe lamp and Duranglass photoreactor) there is an overlap between the emitted light and the irradiation yet available to be absorbed for H2O2 approximately from 290 to 350 nm. Thus, the elimination of IBP has the unique contribution of the hydroxyl radical attack produced by H2O2 photolysis but not due to their separate elements (irradiation or H2O2). An important remark is the fact that TOC did not decrease. Thus, the composition of the remained organic solution after 2 h of UV–Vis/H2O2 process could be addressed either to the hydroxylated byproducts of IBP with weak acidic properties according to the pH evolution observed from 6.5 to 5.5. The hydroxylation process of IBP is a consequence of its scavenger ability in favor of OH. Several indications of the OH scavenger property in NSAID have been widely reported (Hamburger and McCay, 1990; Bilodeau et al., 1995; Aruoma and Halliwell, 1988; Halliwell et al., 1995; Patrico` et al., 1999).
H2O2/UV-Vis
0.50
Fenton Fe(II)/UV-Vis
0.25
0.00 0
30
60
Photo-Fenton (1.2mM Fe(II)) 0.04mM H2O2 0.08mM H2O2 0.16mM H2O2 0.32mM H2O2
90
120
Irradiation Time (min) Fig. 2 – Removal of IBP (0.87 mM) by (-) Photolysis; (,) H2O2/UV–Vis (0.32 mM H2O2); (A) Fenton (1.2 mM Fe(II), 0.32 mM H2O2); (>)Fe(II)/UV–Vis (1.2 mM Fe(II)); PhotoFenton in all cases 1.2 mM Fe(II) and (:) 0.04 mM H2O2; (6) 0.08 mM H2O2; (C) 0.16 mM H2O2; (B) 0.32 mM H2O2.
592
water research 44 (2010) 589–595
of volatile compounds (see also Section 3.5). Thus IBP oxidation and TOC depletion depend on experimental conditions and ratios between reactants, in a rather complex way due to the large number of reactions involved.
50 40 30 20
3.3. Iron effect: IBP as organic ligand of iron and its photolysis
10 0
Fig. 3 – TOC removal for H2O2/UV–Vis (0.32 mM H2O2); Fenton (1.2 mM Fe(II), 0.32 mM H2O2); Fe(II)/UV–Vis (1.2 mM Fe(II)); Photo-Fenton with 1.2 mM Fe(II) and 0.04 mM H2O2, 0.08 mM H2O2, 0.16 mM H2O2, 0.32 mM H2O2. In all cases 0.87 mM initial IBP concentration.
Fe(III) þ H2O þ hv / Fe(II) þ Hþ þ OH
(4)
Ferrous ion is continuously recycled by irradiation and therefore it is not depleted during the course of the oxidation, as stated by Zepp et al. (1992). Additionally Pignatello et al. (1999) have shown the formation of iron complexes, as (FeOH)2þ – predominant in acidic conditions – which have adsorption bands in the UV–Vis range (between 180 and 410 nm) (Sire´s Sadornil Ignacio, 2006), and act as additional hydroxyl radical source in photo-Fenton reaction.
Fe(II) þ H2O2 / Fe(OH)2þ þ OH
(5)
Degradation of IBP was directly proportional to the amount of H2O2 employed. However, H2O2 was not in all cases full consumed. For example, almost 60% of 0.32 mM of H2O2 remained unconsumed. The same proportion is observed by using 0.16 mM however a clear decrease tendency was observed between 90 and 120 min. In contrast 0.08 and 0.04 mM of H2O2 were almost full consumed after 120 min. Fenton reaction shows a rapid degradation phase resulting from a burst of $OH by Eq. (2). However when H2O2 is in large excess the extent of this burst phase will depend on the Fe/ contaminate molar ratio because the ratio determines the OH/contaminant ratio in the burst phase (Pignatello et al., 2006). Moreover due to the pH was adjusted into the first 30 min, an early parallel formation of Fe(III) could promote in strong manner the Fenton-like reaction (Eq. (3)) rather than Fenton reaction. Since reaction in Eq. (3) is so much slower than reaction in Eq. (2), the parallel formation of Fe(III) also results in a slower initial rate of IBP and similar slow consume of H2O2. In the dark series more than 90% of the H2O2 remained after 2 h of reaction. TOC depletion was enhanced under irradiated conditions in the photo-Fenton reaction. As noted in Fig. 3 the maximum TOC removal is observed around 40%. The elimination of organic carbon is also due to the formation
An additional important query in Fenton process is the evaluation of possible complex formation between the organic compound and the metallic catalyst. Normally in Fenton applications the iron-complex formation with organic ligands is employed in order to widen the pH range, to avoid precipitation of Fe(III)-oxyhydroxide and/or to increase the ironphotolysis process (Andreozzi et al., 2006). However, undesirable effects can be also observed if the active coordination sites in the metal are not able to generate hydroxyl radicals in the Fenton reaction. The experiment carried out with 1.2 mM Fe(II) without H2O2, under irradiated condition (> doted in Fig. 2), gave an unexpected 60% of IBP removal as shown in Fig. 2. The removal of IBP using Fe(II)/UV–Vis configuration (without H2O2) was in deep studied by additional experiments varying the iron concentration under irradiated and free pH 5 conditions. Fig. 4 shows the IBP evolution, for 0.012, 1.2 and 5 mM FeSO4.7H2O concentrations, during 2 h of irradiation. Elimination of IBP is directly proportional to the amount of the irradiated-iron employed. IBP binds iron and under UV–Vis irradiated conditions this iron–IBP complex is photoactive. At pH 5, a large fraction of iron is in the form of Fe(III). Coordination of IBP with Fe(II) is probably negligible and would not be photolabile in the wavelength range used. Therefore, the reaction was due to the Fe(III)-complex photolysis. Iron forms complexes with organic compounds, especially those acting as polydentate ligands like Fe(III)–carboxylate or Fe(III)–polycarboxylate complexes (Pignatello et al., 2006). In the case of NSAID – which are characterized for their carboxylic acid moiety – some authors have shown their capacity to bind iron to form coordination complex. For instance, Kennedy et al. (1990) suggest that the complex formation between IBP and iron can be reached through the carboxylic moiety. Under our experimental conditions the possible production of decarboxylated byproducts and acidic
1.00 0.75
IBP (mM)
% TOC removal
Photo-Fenton 1.2mM Fe(II)
0.50 0.25 0.00 0
30
60
90
120
Time (min) Fig. 4 – Effect of Fe(III)/IBP coordinated complex photolysis on IBP degradation during 120 min of UV–Vis irradiation. Initial Fe(II) acid ion added (C) 0.012 mM; (:) 1.2 mM; (-) 5 mM and 0.87 mM IBP initial concentration.
593
water research 44 (2010) 589–595
conditions, the catalytic reaction is less favorable and low TOC removal is observed. In Table 1 all defined stages of photoFenton process are described together with their simultaneous contributions of isolated and combined reagents. The degradation of IBP by iron irradiated process seems to be the principal way as initial mechanism of degradation even if H2O2 is not present (0.7 and 20% of IBP conversion and TOC removal respectively). In photo-Fenton process, the OH yield improves and simultaneous scavenges the isobutyl moiety of IBP. Thus the carboxylic moiety of IBP binds the iron while the isobutyl moiety scavenges the OH radical generated.
Table 1 – Isolated and combined effects of Fenton and photo-Fenton reagents on the IBP conversion and TOC removal observed at 2 h. Process
AþB/C
Fe(III) IBP þ hv H2O2 þ IBP H2O2 þ IBP þ hv Fe(II) þ H2O2 þ IBP
Decarboxylated byproducts Stable solution without reaction Hydoxylated bycompounds Fe(III)-IBP coordinated complex þ hydroxylated byproducts Fe(II) þ H2O2 þ hv Decarboxylated þ hydroxylated þ IBP þ biodegradable compounds
XIBP
TOC removal
0.70 0.00 0.30 0.50
20% 0% 2% 10%
1.00
40%
3.5. aliphatic compounds are related with the change of the pH during the reaction from initial 5 to final 3.5. Higher iron–IBP complex amount, higher depletion of IBP through the carboxylic moiety, in agreement with reports of oxidation in iron-complex with organic functional group RCOO (Pignatello et al., 2006). The ligand-to-metal-charge-transfer reduction of the metal center promotes decarboxylation of IBP, at least partially, and further degradation of decarboxylated bycompounds is also driven under whole photo-Fenton system.
Byproducts from photo-Fenton process
IBP byproducts generated by photo-Fenton treatment (1.2 mM Fe(II), 0.32 mM H2O2, 2 h of irradiation) were identified by an LC/ESI-TOF-MS, into the m/z range of 65–1000 in negative ionization mode. Fig. 5 depicts the proposed reaction pathway with the corresponding intermediates identified. The remaining TOC after treatment of IBP (m/z ¼ 206, TR 28) consisted mostly of two principal types of byproducts: decarboxylated and hydroxylated metabolites of IBP. Hydroxy-IBP byproducts identified were the three structures 1 (m/z ¼ 222, TR 16), 2 (m/ z ¼ 222, TR 23) and 3 (m/z ¼ 222, TR 18). The dihydroxylated compound 10 (m/z ¼ 238, TR 19) was founded at low intensity. Compound 1 suffers either the cleavage of C1–C2 bond of the isobutyl moiety, forming 4 (m/z ¼ 164, TR 13) and isopropanol or ketone acetone; or the decarboxylation forming 5
3.4. Comparison between photo-Fenton processes and its isolated or combined reagents The photo-Fenton degradation of IBP is strongly marked for the hydroxyl attack and decarboxylation process. In dark
CH3 OH
CH3 O
H3C
IBP CH3
CH3 OH
CH3 HO
1
O CH3
OH H3C
O
H3C
O
HO
5
CH3
O
H 3C
OH CH3
O
CH3
H3C
6
OH
3
CH3
CH3
CH3
CH3
CH3
CH3
4
CH3
2
HO
OH
CH3
CH3 O
CH3
H3C
H3C
OH
O
10
7
OH
HO
O CH3 H3C
8
CH3 CH3 H3C
OH
9
Fig. 5 – Byproducts identified and possible degradation pathway by photo-Fenton reaction (1.2 mM Fe(II), 0.32 mM H2O2, 2 h of irradiation, 0.87 mM IBP initial concentration). ID Name, Molecular Mass [Fragments m/z] IBP Ibuprofen, 206 [205L; 161L]. 1 2-HydroxyIBP, 222 [221L; 177L]. 2 1-HydroxyIBP-, 222 [221L; 177L]. 3 2-hydroxy-2-[4-(2methylpropyl)phenyl]propanoic acid, 222 [221L; 191L]. 4 (2RS)-2-(4-Methylphenyl)propanoic acid, 164 [163L; 133L]. 5 1-ethyl-4-(2-hydroxy)isobutylbenzene, 178 [177L, 149L]. 6 1-ethyl-4-(1-Hydroxy)isobutylbenzene, 176 [175L; 133L]. 7 1-[4-(2-Methylpropyl)phenyl]ethanone, 176 [175L; 133L]. 8 4-(1-hydroxy-2-methylpropyl)acetophenone, 192 [191L; 177L]. 9 2-Methyl-1-phenylpropane, 134[133L; 119L]. 10 2-hydroxy-2-[4-(2methylpropyl)phenyl]peroxic acid, 238[237L; 221L].
594
water research 44 (2010) 589–595
(m/z ¼ 178, TR 15.5) and formic acid. In the case of 2, the evidence of its decarboxylation could be represented by the byproduct 6 (m/z ¼ 178, TR 15.5). However undifferentiated retention times by MS analysis with its analogous compound 5 were obtained. In the case of 3, the degradation process follows two possible pathways by either decarboxylation and deprotonation to reach 7 (m/z ¼ 176, TR 21) and formic acid, and by dihydroxylation to obtain the byproduct 10. However byproduct 10 can be also formed by a non-free radical pathway involving a nucleophilic exchange of OH of the carboxylic acid with H2O2. 7 can be again hydroxylated in position 1 of the isobutyl to reach 8 (m/z ¼ 191, TR 25) and also can be decarboxylated to form 9 (m/z ¼ 133, TR 15) and acetaldehyde. An approximation of the main byproducts remained after photo-Fenton treatment could be assigned to compounds 7 and 9 with 23 and 36% of the total intensities. Thus the degradation of IBP by photo-Fenton goes through the simultaneous cleavage of the carboxyl acid moiety to produce formic acid or acetaldehyde and scavenger of hydroxyl radicals. As described before the formation of high soluble bycompounds isopropanol or formic acid and acetone promotes stable dissolution. The presence of those soluble compounds after 2 h of reaction improves the biodegradability in the effluent treated. Initial BOD5 for 0.87 mM of untreated-IBP was <1 ppm, however after photo-Fenton treatment was 25 mg L1.
4.
Conclusions
Photolysis experiment let the IBP (0.87 mM initial concentration) and TOC concentration unchanged. Similar results were observed for all ratios IBP–H2O2 after 24 h darkness. In contrast, UV–Vis/H2O2 process (0.32 mM H2O2) showed an evident decrease in the concentration of IBP (0.87 mM initial concentration). When solution was irradiated in presence of 0.32 mM of H2O2, almost 40% of degradation occurred after 2 h. In Fenton series, degradation of IBP (0.87 mM initial concentration) ranged from 25 to 60% by use of 0.15–1.2 mM Fe(II) respectively in presence of 0.32 mM of H2O2. In contrast, in photo-Fenton series, the degradation of IBP (0.87 mM initial concentration) was direct proportional to the amount of hydrogen peroxide used between 80 and 100% for 0.04 mM and 0.32 mM of H2O2 respectively in presence of 1.2 mM of Fe(II). In regard to the mineralization, photo-Fenton reached 40% of TOC removal (0.87 mM initial IBP concentration; 0.32 mM of H2O2 and 1.2 mM of Fe(II)) whereas in dark-Fenton conditions it was 10% (0.87 mM initial IBP concentration; 0.32 mM of H2O2 and 1.2 mM of Fe(II)). Strong binding ability between iron and IBP was observed and an important photolability of the Fe(III)– IBP coordinated complex promoted in strong manner the decarboxylation of IBP. Higher depletion of IBP through the carboxylic moiety was observed as higher Fe(III)–IBP coordinated complex presence. Complex formation and OH scavenger, as individual process or both at the same time, are responsible of the IBP and TOC depletion in photo-Fenton degradation. The main byproducts observed consisted mostly of the decarboxylated and the hydroxylated byproducts of IBP. Stable biodegradable final effluent showed a BOD5 increase of
25 mg L1. Thus a biological post-treatment can be suggested in order to diminish the environmental impact of IBP in surface waters.
Acknowledgments Authors wish to thank the financial support from the Barcelona University, the Spanish Ministry of Education and Science (project CTQ2005-00446/PPQ) to NOVEDAR-CONSOLIDER 2007 project as well as to the Serveis Cientifico-Tecnics of Barcelona University. Also the comments of two anonym referees to enhance the discussion of the paper are great acknowledged.
references
Alonso, E., Villar, P., Santos, A., Aparicio, I., 2006. Fractionation of heavy metals in sludge from anaerobic wastewater stabilization ponds in southern Spain. Waste Manage. 26, 1270–1276. Andreozzi, R., Canterino, M., Marotta, R., 2006. Fe(III) homogeneous photocatalysis for the removal of 1,2-dichlorobenzene in aqueous solution by means UV lamp and solar light. Water Res. 40, 3785–3792. Aruoma, O., Halliwell, B., 1988. The iron-binding and hydroxyl radical scavenging action of anti-inflammatory drugs. Xenobiotica 18, 459–470. Bigda, R.J., 1995. Consider Fenton’s chemistry for waste water treatment. Chem. Eng. Prog. 12, 62–66. Bilodeau, J.F., Wang, M.Y., Chung, F.L., Castonguay, A., 1995. Effects of nonsteroidal antiinflammatory drugs on oxidative pathways in A/J mice. Free Radic. Biol. Med. 18, 47–54. Burkitt, M.J., 2003. Chemical, biological and medical controversies surrounding the Fenton reaction. Prog. React. Kinet. Mech. 28, 75–103. Carballa, M., Omil, F., Alder, A.C., Lema, J.M., 2006. Comparison between the conventional anaerobic digestion of sewage sludge and its combination with a chemical or thermal pretreatment concerning the removal of pharmaceuticals and personal care products. Water Sci. Technol. 53, 109–117. Chen, R., Pignatello, J.J., 1999. Structure–activity study of electronshuttle catalysis by quinones in the oxidation of aromatic compounds by the Fenton reaction. J. Adv. Oxid. Technol. 4, 447–453. Gonza´lez, O., Sans, C., Esplugas, S., 2007. Sulfamethoxazole abatement by photo-Fenton: Toxicity, inhibition and biodegradability assessment of intermediates. Journal of Hazardous Materials 146 (3), 459–464. Go´mez, M.J., Martı´nez Bueno, M.J., Lacorte, S., Ferna´ndez-Alba, A.R., Agu¨era, A., 2008. Pilot survey monitoring pharmaceuticals and related compounds in a sewage treatment plant located on the Mediterranean coast. Chemosphere 66, 993–1002. Haber, F., Weiss, J., 1934. The catalytic decomposition of hydrogen peroxide by iron salts. Proc. Roy. Soc. A. 134, 332–351. Halliwell, B., Murcia, M.A., Chirico, S., Aruoma, O.I., 1995. Freeradicals and antioxidants in food and in-vivo - what they do and how they work. Crit. Rev. Food Sci. 35, 7–20. Hamburger, S., McCay, P.B., 1990. Spin trapping of ibuprofen radicals - evidence that ibuprofen is a hydroxyl radical scavenger. Free Radic. Res. Commun. 9, 337–342. Halling-Sorensen, B., Nors Nielsen, S., Lanzky, P.F., Ingerslev, F., Holten Lu¨tzhoft, H.C., Jorgensen, S.E., 1998. Occurrence, fate
water research 44 (2010) 589–595
and effects of pharmaceutical substances in the environment a review. Chemosphere 36, 357–393. Hartmann, J., Bartels, P., Mau, U., Witter, M., Tumpling, W.V., Hofmann, J., Nietzschmann, E., 2008. Degradation of the drug diclofenac in water by sonolysis in presence of catalysts. Chemosphere 70, 453–461. Kennedy, T.P., Rao, N.V., Noah, W., Michael, J.R., Jafri, M.H., Gurtner, G.H., Hoidal, J.R., 1990. Ibuprofen prevents oxidant lung injury and in vitro lipid-peroxidation by chelating iron. J. Clin. Invest. 86, 1565–1573. Klavarioti, M., Mantzavinos, D., Kassinos, D., 2008. Removal of residual pharmaceuticals from aqueous systems by advanced oxydation processes. Review article. Environment International 35 (2), pp. 402–417. Me´ndez-Arriaga, F., Esplugas, S., Gimenez, J., 2008a. Photocatalytic degradation of non-steroidal antiinflammatory drugs with TiO2 and simulated solar irradiation. Water Res. 42, 585–594. Me´ndez-Arriaga, F., Torres Palma, R., Pe´trier, C., Esplugas, S., Gimenez, J., Pulgarin, C., 2008b. Ultrasonic treatment of water contaminated with ibuprofen. Water Res. 42, 4243–4248. Me´ndez-Arriaga F., March 2009. Advanced Oxidation Processes (Photocatalysis, Photo-Fenton and Sonolysis) for Removal of Pharmaceutical Pollutants in Water. Doctoral thesis, Barcelona University. Neyens, E., Baeyens, J., 2003. A review of the classic Fenton’s peroxidation as an advanced oxidation technique. J. Hazard. Mater. 98, 33–50. Nogueira, R.F.P., Oliveira, M.C., Paterlini, W.C., 2005. Simple and fast spectrophotometric determination of H2O2 in photoFenton reactions using metavanadate. Talanta 66, 86–91. Patrico`, D., Pasin, M., Barry, O.P., Ghiselli, A., Sabatino, G.A., Iuliano, L., FiztGerald, G., Violi, F., 1999. Iron-dependent human platelet activation and hydroxyl radical formation: involvement of protein kinase C. Circulation 99, 3118–3124. Pe´rez-Estrada, L., Malato, S., Gernjak, W., Agu¨era, A., Thurman, M., Ferrer, I., Fernandez-Alba, A., 2005. Photo Fenton degradation of diclofenac: identification of main intermediates and degradation pathway. Environ. Sci. Technol. 39 (21), 8300–8306. Pe´rez-Moya, M., Graells, M., del Valle, L., Centelles, E., Mansilla, H. , 2007. Fenton and photo-Fenton degradation of 2-chlorophenol: Multivariate analysis and toxicity monitoring. Catalysis Today, 124, 3-4, 30 2007, 163–171. Pignatello, J.J., Liu, D., Huston, P., 1999. Evidence for an additional oxidant in the photoassisted Fenton reaction. Environ. Sci. Technol. 33, 1832–1839.
595
Pignatello, J.J., Oliveros, E., MacKay, A., 2006. Advanced oxidation processes for organic contaminant destruction based on the Fenton reaction and related chemistry. Crit. Rev. Environ. Sci. Technol. 36, 1–84. Roberts, P.H., Thomas, K.V., 2006. The occurrence of selected pharmaceuticals in wastewater effluent and surface waters of the lower Tyne catchment. Sci. Total Environ. 356, 143–153. Safarzadeh-Amiri, A., Bolton, J.R., Cater, S.R., 1996. The use of iron in advanced oxidation technologies. J. Adv. Oxid. Technol. 1, 18–26. Santos, J.L., Aparicio, I., Alonso, E., 2007. Occurrence and risk assessment of pharmaceutically active compounds in wastewater treatment plants. A case study: Seville city (Spain). Environ. Int. 33, 596–601. Sire´s Sadornil Ignacio, 2006. Electrochemical Advanced Oxidation Processes for the Removal of the Drugs Paracetamol, Clofibric Acid and Chlorophene from Waters. Doctoral thesis, Universidad de Barcelona. Takagi, T., Ramachandran, C., Bermejo, M., Yamashita, S., Yu, L.X., Amidon, G.L., 2006. A provisional biopharmaceutical classification of the top 200 oral drug products in the United States, Great Britain, Spain, and Japan. Mol. Pharmacol. 3, 631–643. Tarr, M.A., 2003. Chemical degradation methods for wastes and pollutants. Chemical Degradation Methods for Wastes and Pollutants. In: Environmental Science and Pollution Control Series, vol. 26 165–200. Ternes, T.A., Herrmann, N., Bonerz, M., Knacker, T., Siegrist, H., Joss, A., 2004. A rapid method to measure the solid-water distribution coefficient (K–d) for pharmaceuticals and musk fragrances in sewage sludge. Water Res. 38, 4075–4084. Torres, R., Sarria, V., Torres, W., Peringer, P., Pulgarin, C., 2003. Electrochemical treatment of industrial wastewater containing 5-amino-6-methyl-2-benzimidazolone: toward and electrochemical–biological coupling. Water Res. 37, 7–13. Tuhkanen, T., 2004. UV/H2O2 processes. In: Parsons, S. (Ed.), Advanced Oxidation Processes for Water and Wastewater Treatment. IWA, London UK, pp. 86–109. Venkatadri, R., Peters, R., 1993. Chemical oxidation technologies: ultraviolet light/hydrogen peroxide, Fenton’s reagent and titanium dioxide-assisted photocatalysis, hazard. Waste Hazard. Mater. 10, 107–149. Zepp, R.G., Faust, B.C., Hoigne, J., 1992. Hydroxyl radical formation in aqueous reactions (pH 3–8) of iron(II) with hydrogen-peroxide - the photo-Fenton reaction. Environ. Sci. Technol. 26, 313–319.
water research 44 (2010) 596–604
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Polar pollutants in municipal wastewater and the water cycle: Occurrence and removal of benzotriazoles Thorsten Reemtsmaa,*, Ulf Mieheb, Uwe Duennbierc, Martin Jekelb a
Federal Institute for Risk Assessment, Chemicals Safety, Thielallee 88–92, 14195 Berlin, Germany Technical University of Berlin, Department of Water Quality Control, Sekr KF 4, Strasse des 17 Juni 135, 10623 Berlin, Germany c Berliner Wasserbetriebe, Laboratory, Motardstrasse 35, 13629 Berlin, Germany b
article info
abstract
Article history:
1H-benzo-1,2,3-triazole (BTri) and its methylated analogues (tolyltriazole, TTri) are corro-
Received 3 April 2009
sion inhibitors used in many industrial applications, but also in households in dishwashing
Received in revised form
agents and in deicing fluids at airports and elsewhere. BTri and one of the TTri-isomers
9 July 2009
(4-TTri) are typical examples of polar and poorly degradable trace pollutants. Benzotriazole
Accepted 11 July 2009
elimination in four wastewater treatment plants (WWTP) in Berlin ranged from 20 to 70%
Available online 18 July 2009
for 5-TTRi over 30 to 55% for BTri to insignificant for 4-TTri. WWTP effluent concentrations
Keywords:
4-TTri proved to be omnipresent in surface waters of the rivers Rhine and Elbe with
were in the range of 7–18 mg/L of BTri, 1–5 mg/L of 4-TTri and 0.8–1.2 mg/L of 5-TTri. BTri and Municipal wastewater
concentrations increasing from <0.05 mg/L to around 0.5 mg/L of BTri and 0.2–0.5 mg/L of
Wastewater
4-TTri over 600–700 km. Bank filtration is an important process to generate raw water for
Treatment
drinking water production from surface waters. Even after residence times of several
Benzotriazole
months BTri and 4-TTri were determined in concentrations of a few hundred ng/L in bank
Bank filtration
filtration water. Isotherm data from batch experiments indicate that activated carbon
Ozonation
filtration should be suitable to avoid intrusion of TTri into drinking water in partially closed
Activated carbon
water cycles. For BTri, however, sorption to activated carbon appears to be too weak and
Groundwater
ozonation may be mandatory to remove it from raw waters.
Corrosion inhibitor
ª 2009 Elsevier Ltd. All rights reserved.
Rivers Rhine Elbe
1.
Introduction
Benzotriazoles, in which a 1,2,3-triazole ring is condensed to a benzene ring, are a class of high production volume chemicals (HPVC) used in a wide variety of applications. The basic benzotriazole components are 1H-benzotriazole (BTri, also abbreviated as BTZ, BTA or BTAH) and the 1H-methylbenzotriazoles (tolyltriazoles; TTri, available as technical
mixture of the 4- and 5-isomer, also abbreviated as MeBTri or MeBTA). BTri and TTri are compounds of high polarity (log Kow 1.23 for BTri and 1.89 for TTri; Hart et al., 2004) and can, thus, be expected to be quite mobile in the aquatic environment. Their mobility is further enhanced by the fact that 1H-benzotriazoles, contrary to many other aromatic amines, are NHacidic heterocycles (pKa 8.2–8.5; Hart et al., 2004). Thus,
* Corresponding author. Tel.: þ49 30 8412 3607. E-mail addresses:
[email protected],
[email protected] (T. Reemtsma). 0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2009.07.016
water research 44 (2010) 596–604
benzotriazoles may dissociate and occur as anions at slightly basic pH or in contact with metal cations. At neutral pH in aqueous solution benzotriazoles are expected to be non-ionic rather than cationic. Owing to their NH acidity, benzotriazoles show metal complexing properties. BTri and TTri prevent corrosion of metals and steels, especially of copper and brass (Hart et al., 2004), by forming a thin complexing film on the metal surface (Chadwick and Hashemi, 1978). For this reason BTri or TTri are used in the metal finishing industry, in semiconductor industry (Hollingsworth et al., 2005), in milk processing (Verheyen et al., 2009), and added to various fluids that come in contact with metals, such as aircraft deicing and antiicing fluids (Gruden et al., 2001), cooling liquids, brake fluids, metalworking fluids, etc. BTri is also used for silver protection in dishwashing machines (Ort et al., 2005). The annual production of benzotriazoles was reported to be in the range of 9000 tons/year worldwide (Reemtsma et al., 2006). Benzotriazole can also be employed in the developing process in the photographic industry (Bergthaller, 2002) and is a biocide included in Annex I of Council Directive 98/8/EC (European Commission, 2007). Structurally more complex hydroxiphenyl-benzotriazoles are found in personal care products, where they act as stabilizers for UVA filter components (De Orsi et al., 2006). In the European Union some benzotriazole derivatives are included in Annex VII, part I of Council Directive 76/768/EEC as registered cosmetic ingredients (European Commission, 2000). Structurally related and partly chlorinated benzotriazole derivatives are used as stabilizers in plastic material such as polyethylene terephthalate (PET) (Bentayeb et al., 2007). First reports on the occurrence of benzotriazoles in the environment were related to their use in open systems as aircraft deicing fluids at airports (Cancilla et al., 1998, 2003). A few years later it was recognized that BTri and TTri are much more widely distributed in surface waters (Reemtsma et al., 2006; Voutsa et al., 2006; Giger et al., 2006) in 0.1–1 mg/L concentrations and are regularly discharged with municipal wastewater even after state-of-the art biological treatment in mg/L concentrations (Weiss and Reemtsma, 2005; Voutsa et al., 2006; Weiss et al., 2006). Only one of these studies differentiated between the two TTri-isomers (Weiss et al., 2006) by liquid-chromatography tandem mass spectrometry (LC-MS/MS) (Weiss and Reemtsma, 2005). With this method it could be shown that the two isomers behave quite differently in the environment. While 5-TTri was mineralized within 15 days in an aerobic biodegradation test with activated sludge, the 4-TTri isomer remained stable (Weiss et al., 2006). The much higher stability of 4-TTri as compared to 5-TTri has also been observed in wastewater treatment and in surface waters (Weiss and Reemtsma, 2005; Weiss et al., 2006). Unlike the N-substituted 1,2,4-triazoles, which are used as herbicides, the benzo-1,2,3-triazoles have only limited biological activity. EC50 values for the acute toxicity to Vibrio fisheri were in the range of 7–21 mg/L and NOAEC levels for invertebrates and vertebrates in the range of 10–100 mg/L (Pillard et al., 2001). BTri has been shown to exhibit antiestrogenic activity in vitro in a yeast assay but not in vivo to fish (Harris et al., 2007).
597
Despite these favourable ecotoxicity data the widespread occurrence of benzotriazoles in the environment should be a reason to improve our knowledge on the environmental behaviour of these compounds. Inter alia, the fate of benzotriazoles in surface waters is not well studied, especially for the TTri-isomers. It could also be worthwhile to become aware of those processes that are suitable for the removal of benzotriazoles from wastewater as well as from raw waters used for drinking water production.
2.
Materials and methods
2.1.
Wastewater treatment plants
Samples are taken from the four largest wastewater treatment plants (WWTPs) in Berlin with treated water flows between 40,000 and 200,000 m3/day. All plants are equipped with primary sedimentation, conventional activated sludge (CAS) treatment with nutrient removal (N/P removal) and secondary clarification. The plants differ in the P removal process: WWTP-R and W use biological P-elimination (Bio-P), WWTP-S with Bio-P and chemical precipitation and WWTP-M with chemical precipitation only. Hydraulic retention time (HRT) in the plants ranged from 11 to 24 h for the biological stage and the sludge retention time (SRT) from 8 to 25 days. The influent to WWTP-R had an organic load of 300–900 mg/L chemical oxygen demand (COD), while it ranged from 700 to 1100 mg/L COD for the other three plants. Samples of the influent (raw sewage) and the effluent of the secondary clarifier were taken as 24 h mixed samples between June and December 2006. Sampling was performed only at dry weather flow, and the sampling of effluent was corrected for hydraulic retention time in each of the plants.
2.1.1.
Flocculation-rapid filtration (F-RF)
A polishing of the effluent of WWTP-R was investigated by rapid filtration on two experimental dual-media filters (0.8 m sand/1 m anthracite layer) that treated 100–180 L/h. Particle sizes were 0.7–1.25 mm for sand and 1.4–2.5 mm for anthracite in Filter 1 and 1.1–1.6 mm for sand and 2.5–4.0 mm for anthracite in Filter 2. Filter velocities ranged from 6 to 10 m/h. Filter backwash was performed every 1–2 days. Before filtration the effluent was treated by flocculation with either Al3þ (1 mg/L) or Fe3þ (2 mg/L). During some phases 10–20 mg/L of powdered activated carbon (PAC; Norit Super SAE) were added during flocculation (flocculation, PAC, rapid filtration: F-PAC-RF).
2.2.
Field sites and field samples
2.2.1.
Rivers Rhine and Elbe
Single grab samples were taken with a bucket in January 2007 at sites used for sampling for the environmental specimen bank of the Federal Environment Agency (UBA).
2.2.2.
River Havel
Samples of the river Havel were taken at the surface water treatment plants Spandau (upstream, n ¼ 5) and Beelitzhof (downstream; n ¼ 8) from July to November 2006.
598
2.2.3.
water research 44 (2010) 596–604
Bank filtration
Bank filtration was studied at two sites in Berlin (Germany), at Lake Tegel and Lake Wannsee/Havel. Details of the bank filtration, the hydrology of the sites and the transects installed have been published before (Gruenheid et al., 2005; Massmann et al., 2008). Samples were withdrawn from several observation wells and from the production well at a distance of about 100 m from the lakeshore. Samples, taken in April 2008 from the Tegel site and in May 2008, from both sites, were analyzed for benzotriazoles.
2.2.4.
Surface water treatment plants
Two surface water treatment plants (Tegel and Beelitzhof) were investigated. These plants are designed to reduce the total phosphorus load of the surface water (median of 265 mg/L for Tegel and 129 mg/L for Beelitzhof) to values between 5 and 30 mg/L by coagulation/flocculation, sedimentation and dualmedia filtration. The influent of the Tegel plant consists of 25–80% effluent of WWTP-S, diluted with surface water. The Tegel plant treats 130,000–300,000 m3/day, the final filters have a height of 1.8 m and an average filter velocity of 6.6 m/h. The Beelitzhof plant treats surface water of Lake Wannsee (part of the River Havel system) downstream of the City of Berlin. A volume of 12,000 m3/day is treated in this plant and filtration occurs on filters of 1.6 m height and at a filter velocity of 5 m/h.
2.3.
Lab experiment
2.3.1.
PAC sorption studies
Batch experiments for isotherm calculation were performed in either ultrapure water or effluent of WWTP-R with start concentrations of 100 mg/L of BTri or TTri (technical mix consisting of 40% 4-TTri and 60% 5-TTri (Weiss and Reemtsma, 2005). Between 1 and 25 mg/L of PAC (Norit Super SAE) were added from suspensions (0.1–1 g/L) stirred at 600 rpm. The bottles were continuously mixed on a horizontal shaker (120 rpm) for 24 h. All experiments were performed in duplicate. Stability of the compounds was recorded in a control experiment without PAC addition. Dissolved concentrations of BTri and TTri (sum of both isomers) were determined after membrane filtration (0.45 mm cellulose-nitrate membrane filters) by LC-MS/MS analysis. A Freundlich model was fitted to the data.
2.4.
Analysis of benzotriazoles
All field samples were stored in glass bottles, transported to the lab in ice chests, filtered over 0.45 mm membrane filters and stored in a refrigerator at 4 C pending analysis, usually within the next 24 h. Benzotriazoles were analyzed by LC-MS/MS with an isocratic elution by methanolic eluents on a diphenyl column (Pursuit Diphenyl, 3 mm, 150 2 mm) that allowed separation of the two TTri-isomers. Details have been published before (Weiss and Reemtsma, 2005). Samples of surface water and WTTP influents and effluents were analyzed by direct injection (20 mL for surface water and WWTP effluent; 10 mL for WWTP influent) into the LC-MS/MS system, while groundwater samples were analyzed after enrichment of a volume of 50 mL by solid phase extraction (SPE) on a Oasis HLB 200 mg cartridge. With this procedure an LOQ of 0.15 mg/L (surface water) and 0.25 mg/L (wastewater) was obtained for direct injection and 0.05 mg/L after SPE (groundwater) for each of the three analytes.
3.
Results and discussion
3.1.
Wastewater treatment plants
Four WWTP in Berlin, treating municipal wastewater as a mix of household and industrial wastewater were investigated for the occurrence and removal of benzotriazoles over a 6 month period (Fig. 1, Table S1). Mean influent concentration for the four plants ranged from 17 to 44 mg/L for BTri and from 1.1 to 4.9 mg/L for each of the TTri-isomers. BTri and TTri concentrations were both higher in WWTP-M as compared to the other three plants. This was not due to a generally more concentrated wastewater in this plant, because the BTri/DOC ratio of 0.6& was also twice as high as the value for the other three plants (0.2–0.3&). The source of this elevated input of benzotriazoles to this WWTP is unknown. These influent concentrations are comparable to those reported in previous studies: mean concentrations of 6.6 mg/L BTri and 0.7 mg/L for TTri were found in a monitoring of eight WWTP in Western Europe (Reemtsma et al., 2006). In a study carried out in Switzerland on 10 WWTP even higher influent concentrations of 15–73 mg/L for BTri and 1.1–5.6 mg/L for
Fig. 1 – Concentrations of (a) BTri, (b) 5-TTri and (c) 4-TTri in the influent and effluent of the four largest WWTP in Berlin (R (n [ 5), S (n [ 8), W (n [ 9): 05/2006–04/2007; M (n [ 8): 10/2005–04/2007).
water research 44 (2010) 596–604
TTri-isomers were recorded (Voutsa et al., 2006). In a previous study on WWTP-R in Berlin 12.0 mg/L for BTri and 1.3 mg/L for 5-TTri and 2.1 mg/L for 4-TTri were reported as the 2-year average (Weiss et al., 2006), which is at the lower end of concentrations determined in this study. Benzotriazoles in municipal wastewater may originate from household due to their use as corrosion inhibitors in dishwashing agents. This could be proven for a residential area by modeling the loads in its sewer system (Ort et al., 2005). When wastewater of a larger urban area is collected in one sewer system, industrial effluents, e.g. from metal finishing or semiconductor industry, may also be of relevance. Milligram per litre concentrations of TTri have only recently been reported in wastewater from milk processing (Verheyen et al., 2009). In mixed sewer systems runoff from streets (traffic) may also contribute. Airports, which are a well studied point source of benzotriazole emission (Cancilla et al., 1998, 2003), may also contribute to the load in municipal wastewater if their surface runoff is collected and directed to the municipal sewer system (Weiss et al., 2006). Of the three benzotriazoles, none was eliminated completely during wastewater treatment. Generally relative removal was best for 5-TTri, but highly variable between the plants; it ranged from 19 to 69%. BTri showed moderate removal (29–58%), whereas for 4-TTri only one plant showed a significant but still limited removal (34%) (Fig. 1, Table S1). Only moderate removal of BTri and TTri (sum of both isomers) has been reported before in a monitoring of eight WWTP in Western Europe (Reemtsma et al., 2006) and for a variety of WWTP in Switzerland (Voutsa et al., 2006). This different removal of the two TTri-isomers observed in this study is in agreement with lab experiments, which showed that 5-TTri can be mineralized by activated sludge while 4-TTri is persistent (Weiss et al., 2006). These differences are attributed to the position of the methyl substituent at the ring system. Previous studies reported that the aerobic biodegradability of positional isomers of bicyclic aromatic compounds depends upon the substitution pattern and that isomers with the ortho-substituent (as in 4-TTri) were not biodegradable, while the isomer with the meta-substituent (as in 5-TTri) were degraded (Weiss and Reemtsma, 2005). The different removal behaviour of the three benzotriazoles can be highlighted by calculating the concentration ratios of either BTri or 5-TTri to the most persistent 4-TTri. Both these ratios tend to decrease during biological treatment from 6–7 to 4–5 for BTri/4-TTri and from 0.7–0.9 to 0.3–0.8 for 5-TTri/4-TTri (Table S1). These ratios are especially useful to differentiate between dilution (decrease in concentrations but no change in the concentration ratios) and biodegradation (decrease in concentrations and in concentration ratios). This underlines that an appropriate analytical method that separates the isomers is essential to adequately study the environmental fate of TTri (Weiss and Reemtsma, 2005). Owing to the incomplete removal BTri median values of 7–18 mg/L were found for the effluents of the four WWTP. Single values up to 38 mg/L were recorded in the effluent of WWTP-M. In a Swiss study BTri effluent concentrations ranging from 10 to 100 mg/L have been found (Voutsa et al., 2006). These are remarkably high concentrations for single compounds in WWTP effluents. In a study of eight European
599
WWTP, in which 28 trace pollutants have been monitored, BTri ranked second in effluent concentration; it was outcompeted only by ethylenediamine tetraacetic acid (EDTA) with 100 mg/L (Reemtsma et al., 2006). As shown previously the elimination of 5-TTri and BTri in biological wastewater treatment can be improved to about 60% by using membrane bioreactors. But even with this upgraded biological treatment system the 4-TTri concentration was not reduced (Weiss et al., 2006). As a measure to prioritize trace pollutants in WWTP effluents with regard to their putative concentration in further compartments of the water cycle a so-called water cycle spreading index (WCSI) has been proposed (Reemtsma et al., 2006). The WCSI is the ratio of the effluent concentration and the degree of removal in a WWTP. In previous studies the order of the three benzotriazoles in terms of increasing WCSI was 5-TTri (16 mg/L), BTri (22 mg/L) and 4-TTri (46 mg/L) (Reemtsma et al., 2006; Weiss et al., 2006). From the influent and effluent data of these four WWTP the order would be 5-TTri (1–13 mg/L), 4-TTri (4–27 mg/L, 74 mg/L) and BTri (10–54 mg/L). Accordingly, BTri would be expected to be the benzotriazole occurring in highest concentration in surface waters and also in later compartment of a water cycle, closely followed by 4-TTri. From its WCSI (lower effluent concentration and better removal) 5-TTri would be the least critical components.
3.2.
Polishing of WWTP effluents
If the biological processes in WWTP are not suited to remove polar pollutants from the wastewater, physico-chemical posttreatment may be used to avoid discharge of these pollutants into surface waters and their spreading along partially closed water cycles. For benzotriazoles, ozonation has been shown to be a suitable post-treatment process. An ozone dose of 1 mg/mg DOC in wastewater was sufficient for complete removal of BTri and TTri (Weiss et al., 2006). No difference in the reactivity of the three benzotriazoles towards ozone could be determined.
3.2.1. Flocculation, sand filtration and powdered activated carbon In the present study we investigated whether flocculation followed by rapid sand filtration, with and without addition of powdered activated carbon (PAC) can also be used for such a polishing step. A flocculation and rapid filtration (F-RF) unit (100–180 L/h) was operated for more than 2 years on site of the WWTP-R with different flocculants (Al3þ 1 mg/L, Fe3þ 2 mg/L) and different filter velocities (6–7.5 m/h). Based on experiences in the deironing step of water treatment plants it was hypothesized that the biomass establishing in the subsequent (bio-) filters may remove some of the trace pollutants. The flocculation with either Al3þ or Fe3þ had no effect on the concentration of any of the benzotriazoles. However, 5-TTri showed an average removal of 50% in the subsequent rapid filtration (RF) step (Table S1). During some periods the removal was almost complete (Fig. 2). Thus, biodegradation of 5-TTri on the filter column must have been comparatively fast, as the residence time in the filter was about 20 min, only. These data
600
water research 44 (2010) 596–604
Table 1 – Concentration of benzotriazoles in single surface water samples of the rivers Rhine and Elbe (01/ 2006) and mean concentration in river Havel (07/2006–12/ 2006). kma
Fig. 2 – Mean concentration of BTri, 4-TTri and 5-TTri in the feed water and the effluent of a rapid filtration column (filled with 0.8 m sand and 1 m anthracite) after flocculation with 1–2 mg/L Fe3D (F-RF) with and without addition of PAC (10–20 mg/L) The error bars indicate standard deviation of three samplings at independent days.
confirm the much better biodegradability of 5-TTri as compared to BTri and 4-TTri. The whole process was much more effective when PAC was added during flocculation (F-PAC-RF) (Fig. 2, Table S1). With the addition of 20 mg/L of PAC BTri could be reduced by 81% to filter effluent concentrations around 3 mg/L and 5-TTri as well as 4-TTri were removed to levels below the LOQ of 0.25 mg/L. A PAC concentration of 10 mg/L proved insufficient in effluent polishing (Fig. 2). In batch experiments with WWTP effluent, however, 10 mg/L of PAC were able to reduce the BTri and TTri concentration by 40 and 80%, respectively, in 24 h. This indicates that the contact time in the rapid filtration process was too short to make full use of the sorption capacity of the PAC.
3.3.
Occurrence in surface water
3.3.1.
Rhine and Elbe
The two major rivers of Germany and also important European rivers, Rhine (catchment area around 200,000 km2) and Elbe (catchment area around 150,000 km2), were analyzed for their benzotriazole concentrations. BTri was quantified (LOQ ¼ 0.05 mg/L) in all but one sample with concentrations ranging from 0.09 to 0.68 mg/L and 4-TTri in six of the nine samples (0.05– 0.46 mg/L; Table 1). Clear trends are seen along both rivers, such as the increasing concentration of BTri (from 0.13 to 0.35 mg/L over the 700 km in the Rhine and from <0.05 to 0.48 mg/L over 640 km in the Elbe River) and of 4-TTri (<0.05–0.17 mg/L in the Rhine, 0.1–0.45 mg/L in the Elbe River). Finally a relative increase in 4-TTri concentration as compared to BTri is visible in both rivers (Table 1). 5-TTri could be quantified in only two samples downstream of the city of Hamburg in the Elbe river. These increasing concentrations indicate that the continuing emission of BTri and 4-TTri into the rivers Rhine and Elbe exceeds the self-purification potential of the rivers and the dilution by increasing water flow. This phenomenon
BTri (mg/L)
5-TTri (mg/L)
4-TTri (mg/L)
BTri/4-TTri
Rhine Weil Iffezheim Koblenz Bimmen
171 333 591 865
0.13 0.09 0.39 0.35
<0.05 <0.05 <0.05 <0.05
<0.05 <0.05 <0.05 0.17
9.23 2.08
Elbe Prossen Zehren Barby Cumlosen Blankenese
13 89 293 470 638
<0.05 0.07 0.48 0.68 0.48
<0.05 <0.05 <0.05 0.09 0.13
0.10 <0.05 0.18 0.46 0.45
2.69 1.46 1.06
7
<0.05
<0.05
<0.05
13
1.57
0.34
2.14
Havel Upstream of Berlin (n ¼ 5) Downstream of Berlin (n ¼ 8)
0.73
a Rhine: distances (in km) according to an international agreement starting at Lake Constance; Elbe: distances (in km) from the Czech/ German boarder; Havel: distance (in km) from Spandau harbour.
is worse for 4-TTri, for which the WWTP data already indicated highest persistence. Therefore, also a relative increase in 4-TTri as compared to BTri, expressed as a decreasing BTri/ 4-TTri ratio, is observable (Table 1). For the biodegradable 5-TTri this ‘‘enrichment’’ is much weaker and visible only for two downstream samples. High concentrations at station Blankenese may be a local phenomenon as this site is located downstream of the city of Hamburg. The concentrations found in Rhine and Elbe match well with previous monitoring data for Western Europe (0.31 mg/L of BTri and 0.18 mg/L for TTri; Reemtsma et al., 2006) and for a number of Swiss rivers (0.16–2.68 mg/L of BTri and 0.04– 0.32 mg/L sum of TTri; Giger et al., 2006). In a recent monitoring study covering grab samples of 100 rivers in 27 European countries BTri and TTri ranked 2nd and 3rd in the median concentration of the 34 analytes covered in that campaign (Loos et al., 2009). These previous studies did not differentiate between the two TTri-isomers. Based on the data presented here it is likely, that most of the TTri concentration reported in previous studies was due to the more stable 4-TTri isomer. According to the high concentration of benzotriazoles in WWTP discharges it is reasonable to assume that these are the major contributor to benzotriazole concentrations in surface waters (Giger et al., 2006). Surface runoff from airports can be another point source with remarkably high benzotriazole concentrations in winter time (Cancilla et al., 2003; Giger et al., 2006; Corsi et al., 2003), either via direct discharge into surface water or through WWTPs (Weiss et al., 2006). According to their uses also in automotive cooling liquids, brake fluids, etc., occurrence of benzotriazoles in runoff from impervious streets can also be anticipated. Street runoff may either be collected in combined sewer systems and would enhance the benzotriazole load of WWTP influents at wet weather flow, or
water research 44 (2010) 596–604
discharged directly into trenches and creeks. However, the concentration of benzotriazoles in such surface runoffs has not been analyzed, yet.
3.3.2.
River Havel
Also, water of the river Havel that passes the City of Berlin from the north to the south was analyzed for a period of 5 months, because it is an important component of the partially closed water cycle of the city (Gruenheid et al., 2005). While the upstream samples did not show measurable concentrations of the triazoles, the downstream samples contained an average 1.6 mg/L of BTri, 0.34 mg/L of 5-TTri and 2.1 mg/L of 4-TTri (Table 1). Here 4-TTri is the most prominent benzotriazole, which is consistent with our view on the relative stability of these three compounds. These surface water concentrations in river Havel downstream of the city of Berlin are about a factor of 5–10 higher than those found in Rhine and Elbe (Table 1). This is due to the high portion of WWTP effluent in the rivers of Berlin with their low water flow. Increased surface water concentration of BTri with increasing wastewater portion has been found also for different Swiss rivers (Giger et al., 2006). The widespread occurrence of BTri and 4-TTri in surface waters may also have consequences for the quality of raw waters used for drinking water production. Along all three rivers, Rhine, Elbe and Havel, water is abstracted via bank filtration for drinking water production.
3.4.
Treatment of surface water
3.4.1.
Flocculation and rapid filtration
At several locations in Berlin (Fig. S1) surface water is treated by flocculation with Fe3þ, sedimentation and rapid filtration to remove dissolved phosphorous and to avoid eutrophication in critical areas. Because removal of some trace pollutants, such as para-toluenesulfonamide, has been recorded in these surface water treatment plants (Richter et al., 2008) we investigated whether benzotriazoles could also be removed. However, no significant removal was obtained for any of the three compounds in each plant. For BTri and 4-TTri, these results agreed to those obtained for effluent polishing by F-RF (see above; Table S1). For 5-TTri, a 50% removal had been obtained in the RF step of the effluent, presumably by biodegradation. The lack of removal in the corresponding step with surface water, at the same contact time of 20 min, may indicate, that either microorganisms from the activated sludge may be required for this degradation or that the concentration of 5-TTri in the surface water (0.3–0.5 mg/L) was too low to induce this biodegradation process.
3.4.2.
(River) bank filtration
River bank filtration is an important source of raw water for drinking water production and the passage of the littoral and the underground is an important process to improve and ensure water quality. This is true on a global scale, for Germany (16% of bank filtration; Hiscock and Grischek, 2002) but especially for the City of Berlin that gains more than 50% of the drinking water for its 3.5 Mio inhabitants from bank filtrate (Gruenheid et al., 2005).
601
The concentrations of the three benzotriazoles were determined along two bank filtration transects towards the production wells at two different sites (Lake Tegel and Lake Wannsee) in Berlin (Fig. S1) At both sites the distance of the production well from the lakeshore is about 100 m, corresponding to travel times of the water in the range of 4–5 months (Gruenheid et al., 2005; Massmann et al., 2008). Surface water in Lake Tegel consists of 15–30% effluent of WWTP-S (Gruenheid et al., 2005). The water abstracted in the production well consists of approx. 75% lake water and 25% land-sided groundwater (inland aquifer), corresponding to a volumetric contribution of previous wastewater of 11–22% (Gruenheid et al., 2005). Redox conditions are not stable but fluctuate between anoxic and aerobic depending on water temperature and the mode of operation of production wells (Gruenheid et al., 2005). At the second site the hydrological situation is largely similar. For two reasons the concentration data (Fig. 3) are a bit scattered. (i) Not all observation wells sampled at the same depth and, thus, redox conditions are not identical. (ii) All samples were taken within 2 days while the water needs months to travel along the transects from the lakeshores to the production wells. Thus, not the same body of water was sampled along one transect. Long-term investigations at these two sites have shown that seasonal fluctuations can be recorded in the groundwater body (Massmann et al., 2008). Despite these difficulties both transects show an ongoing continuous decrease in BTri concentration from the lakeshore (distance 0) to the respective production well (Figs. 3a,d). At Lake Tegel the lakewater concentration of around 2 mg/L is diminished to 0.2 mg/L in the production well, which corresponds to a 90% concentration decrease due to dilution and removal (Fig. 3a). If one considers the dilution by land-sided groundwater with no BTri contamination, the removal of BTri during this underground passage would be 86%. At the bank filtration site at Lake Wannsee the BTri concentration was reduced by 75% to 0.3 mg/L in the production well (Fig. 3d). The concentration gradient of BTri at the Lake Tegel transect is quite parallel to the DOC gradient (Fig. 3a), which may suggest a cometabolic degradation of BTri. Removal of the least stable 5-TTri from surface water was obviously much faster and, likely, complete (Figs. 3c,f). This is well visible for the Wannsee site, where the first observation well was placed right in the colmation zone at the lakeshore. Here the 5-TTri concentration was already diminished to 0.1 mg/L (Fig. 3f), while the concentrations of 4-TTri was only slightly reduced as compared to the lake water (Fig. 3e). However, also 4-TTri was removed significantly along both transects, but the concentration gradient was less steep (Figs. 3b,e). At Lake Wannsee, where the 4-TTri concentration in surface water tends to be higher than in Lake Tegel the concentration in the production well in 100 m distance remained at 0.13 mg/L (Fig. 3e). It is unclear by which mechanism 4-TTri was removed. For sandy aquifer material a retardation factor of 1.1 has been experimentally determined (Jia et al., 2007). As sandy material is predominant also at these bank filtration sites, a removal by sorption appears unlikely. Rather a very slow microbial degradation may occur. Slow biodegradation of trace concentrations of organic contaminants taking place in
602
water research 44 (2010) 596–604
Fig. 3 – Concentration of (a, d) BTri, (b, e) 4-TTri and (c, f) 5-TTri in bank filtration transects at two sites in Berlin. (left) at Lake Tegel (04/2008 and 05/2008); DOC concentration also shown in (a); (right) at Lake Wannsee (04/2008). PW: production well. Shaded area below LOQ.
time periods of months are, however, difficult to verify experimentally. Whatever process may be responsible, bank filtration was the most effective natural process for removal of benzotriazoles from waters. This may also be true for other poorly degradable polar substances. However, one has to consider the duration of this treatment process of several months. The concentration profiles of the transects also show that these long residence times in the subsurface are essential to remove poorly degradable polar pollutants such as BTri and 4-TTri, while for 5-TTri much shorter residence times would have been sufficient. These findings consistently show that neither BTri nor 4-TTri are completely eliminated during bank filtration, even after a travel time of 4 months (120 days).
3.5.
Drinking water treatment
Benzotriazoles are widespread in surface waters and ‘‘natural’’ processes such as biological wastewater treatment,
biologically active filters and bank filtration are not sufficient to remove them completely. Thus, BTri and 4-TTri may occur in raw waters used for drinking water production and technical treatment processes should be available to remove such polar pollutants during drinking water treatment. While the suitability of ozonation for the removal of BTri and TTri with moderate dosages of ozone has been shown before (Weiss et al., 2006; Legube et al., 1987), it was investigated here whether activated carbon treatment could also be used to remove trace concentrations of benzotriazoles. Batch experiments with deionized water were performed to record isotherms for BTri and TTri (as the sum of both isomers; Fig. 4). The Freundlich parameters KF ¼ 84 (mg/g)/ (mg/L)n for BTri and 187 (mg/g)/(mg/L)n for TTri and n ¼ 0.44 for BTri and 0.52 for TTri indicate a moderate tendency to adsorb onto activated carbon. It is stronger for the methylated benzotriazoles (TTri) than for the non-substituted analogue (BTri). Based on these isotherm data one can assume that trace concentrations of TTri that may occur in raw waters are well
water research 44 (2010) 596–604
603
cycles with indirect potable reuse can be avoided by these technical treatment steps. Data presented here clearly show an increasing resistance to biodegradation from 5-TTri over BTri to 4-TTri. Thus, the widespread contamination of the water cycle with benzotriazoles could be avoided to a large extent if solely 5-TTri would be used instead of BTri and of the technical mixture of 5-TTri and 4-TTri.
Acknowledgements Fig. 4 – Isotherms for the adsorption of BTri and TTri (sum of both isomers) onto PAC (Norit Super SAE) from deionized water with a start concentration C0 [ 100 mg/L (mean of two experiments).
removable by GAC filtrations, as it has been found for the pharmaceutical diclofenac with a comparable KF values of 140 (mg/g)/(mg/L)n in a previous study (Ternes et al., 2002). For BTri with its lower KF value, however, the suitability of GAC filtration is questionable. Rather breakthrough in GAC filtration may occur, as found for clofibric acid with a comparable KF value of 71 (mg/g)/(mg/L)n (Ternes et al., 2002). A final evaluation of the suitability of GAC for the removal of the triazoles would have to be performed site-specific, as the sorption tendency and the sorption kinetics in a certain water may be influenced by its matrix constituents and its pH.
4.
Conclusion
Benzotriazoles are typical examples of polar and poorly degradable pollutants and they behave characteristically in partially closed water cycles. Biological treatment is insufficient for their complete removal from water, as shown for wastewater treatment, surface waters, bank filtration and biologically active filters. Owing to its wide use and biological resistance BTri appears to be one of the most prominent single trace pollutants in WWTP effluents (mg/L concentrations), and it is omnipresent also in surface waters. Concentrations of 4-TTri tend to increase along the large European river systems Rhine and Elbe, indicating the continuous discharge of this compound into the rivers and its high stability also in surface waters. Bank filtration with residence times of several months in the underground proved the most effective natural process for removing these polar and poorly degradable trace pollutants. But neither BTri nor 4-TTri were removed completely and both compounds may occur at reduced concentration in raw waters used for drinking water production. Fortunately, these aromatic compounds of high electron density can be removed by ozonation and, at least, TTri also by activated carbon filtration, either in wastewater effluent polishing or in drinking water treatment. Thus, an intrusion of BTri and TTri from wastewater into drinking water in partially closed water
Funding of this research by Berliner Wasserbetriebe (project ‘‘Barrieren’’) is gratefully acknowledged. We are grateful to Jutta Jakobs for assistance in the laboratory and to Ina Engel for performing the activated carbon study. The river water samples were received from the Department of Hydrogeology of the Free University of Berlin.
references
Bentayeb, K., Batlle, R., Romero, J., Nerin, C., 2007. UPLC-MS as a powerful technique for screening the nonvolatile contaminants in recycled PET. Anal. Bioanal. Chem. 388, 1031–1038. Bergthaller, P., 2002. Couplers in color photography – chemistry and function – Part 3. Imaging Sci. J. 50, 233–276. Cancilla, D.A., Martinez, J., Van Aggelen, G.C., 1998. Detection of aircraft deicing/antiicing fluids in perched water monitoring well at an international airport. Environ. Sci. Technol. 32, 3834–3835. Cancilla, D.A., Baird, J.C., Rosa, R., 2003. Detection of aircraft deicing additives in groundwater and soil samples from Fairchild Air Force base, a small to moderate user of deicing fluids. Bull. Environ. Contam. Toxicol. 70, 868–875. Chadwick, D., Hashemi, T., 1978. Adsorbed corrosion inhibitors studied by electron spectroscopy: benzotriazoles on copper and copper alloys. Corros. Sci. 18, 39–51. Corsi, S.R., Zitomer, D.H., Field, J.A., Cancilla, D.A., 2003. Nonylphenol ethoxylates and other additives in aircraft deicers, antiicers and waters receiving airport runoff. Environ. Sci. Technol. 37, 4031–4037. De Orsi, D., Gianni, G., Gogliardi, L., Porra, R., Berri, S., Bolasco, A., Carpani, I., Tonelli, D., 2006. Simple extraction and HPLC determination of UV-A and UV-B filters in sunscreen products. Chromatographia 64, 509–515. European Commission, 2000. Twenty-Fourth Commission Directive 2000/6/EC Adapting to Technical Progress Annexes II, III, VI and VII to Council Directive 76/768/EEC on the Approximation of the Laws of the Member States Relating to Cosmetic Products. European Commission, 2007. Commission Regulation (EC) No. 1451/2007 of Directive 98/8/EC Concerning the Placing of Biocidal Products on the Market. Giger, W., Schaffner, C., Kohler, H.-P.E., 2006. Benzotriazole and tolyltriazole as aquatic contaminants. 1. Input and occurrence in rivers and lakes. Environ. Sci. Technol. 40, 7186–7192. Gruden, C.L., Dow, S.M., Hernandez, M.T., 2001. Fate and toxicity of aircraft deicing fluids additives through anaerobic digestion. Water Environ. Res. 73, 72–79. Gruenheid, S., Amy, G., Jekel, M., 2005. Removal of bulk dissolved organic carbon (DOC) and trace organic compounds by bank filtration and artificial recharge. Water Res. 39, 3219–3228.
604
water research 44 (2010) 596–604
Harris, C.A., Routledge, E.J., Schaffner, C., Brian, J.V., Giger, W., Sumpter, J.P., 2007. Benzotriazole is antiestrogenic in vitro but not in vivo. Environ. Toxicol. Chem. 26, 2367–2372. Hart, D.S., Davis, L.C., Erickson, L.E., Callender, T.M., 2004. Sorption and partitioning of benzotriazole compounds. Microchem. J. 77, 9–17. Hiscock, K.M., Grischek, T., 2002. Attenuation of groundwater pollution by bank filtration. J. Hydrol. 266, 139–144. Hollingsworth, J., Sierra-Alvarez, R., Zhou, M., Ogden, K.L., Field, J.A., 2005. Anaerobic biodegradability and methanogenic toxicity of key constituents in copper chemical mechanical planarization effluent of the semiconductor industry. Chemosphere 59, 1219–1228. Jia, Y., Breedveld, G.D., Aagaard, P., 2007. Column studies on transport of deicing additive benzotriazole in a sandy aquifer and a zerovalent iron barrier. Chemosphere 69, 1409–1418. Legube, B., Guyon, S., Dore, M., 1987. Ozonation of aqueous solutions of nitrogen heterocyclic compounds: benzotriazoles, atrazine and amitrole. Ozone Sci. Eng. 9, 233–246. Loos, R., Gawlik, B.M., Locoro, G., Rimaviciute, E., Contini, S., Bidoglio, G., 2009. EU-wide survey of polar organic persistent pollutants in European river waters. Environ. Pollut. 157, 561–568. Massmann, G., Sueltenfuss, J., Duennbier, U., Knappe, A., Taute, T., Pekdeger, A., 2008. Investigation of groundwater residence times during bank filtration in Berlin: a multi-tracer approach. Hydrol. Process 22, 788–801. Ort, C., Schaffner, C., Giger, W., Gujer, W., 2005. Modeling stochastic load variations in a sewer system. Water Sci. Technol. 52, 113–122.
Pillard, D.A., Cornell, J.S., Dufresne, D.L., Hernandez, M.T., 2001. Toxicity of benzotriazole and benzotriazole derivatives to three aquatic species. Water Res. 35, 557–560. Reemtsma, T., Weiss, S., Mueller, J., Petrovic, M., Gonzalez, S., Barcelo, D., Ventura, F., Knepper, T.P., 2006. Polar pollutant entry into the water cycle by municipal wastewater: a European perspective. Environ. Sci. Technol. 40, 5451–5458. Richter, D., Massmann, G., Duennbier, U., 2008. Identification and significance of sulphonamides (p-TSA, o-TSA, BSA) in an urban water cycle (Berlin, Germany). Water Res. 42, 1368–1378. Ternes, T.A., 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, 3855–3863. Verheyen, V., Cruickshank, A., Wild, K., Heaven, M.W., McGee, R., Watkins, M., Nash, D., 2009. Soluble, semivolatile phenol and nitrogen compounds in milk-processing wastewaters. J. Dairy Sci. 92, 3484–3493. Voutsa, D., Hartmann, P., Schaffner, C., Giger, W., 2006. Benzotriazoles, alkylphenols and bisphenol A in municipal wastewaters and in the Glatt River, Switzerland. Environ. Sci. Pollut. Res. 13, 333–341. Weiss, S., Reemtsma, T., 2005. Determination of benzotriazoles corrosion inhibitors from aqueous environmental samples by liquid chromatography-electrospray ionization-tandem mass spectrometry. Anal. Chem. 77, 7415–7420. Weiss, S., Jakobs, J., Reemtsma, T., 2006. Discharge of three benzotriazole corrosion inhibitors with municipal wastewater and improvements by membrane bioreactor treatment and ozonation. Environ. Sci. Technol. 40, 7193–7199.
water research 44 (2010) 605–615
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Determining the fraction of pharmaceutical residues in wastewater originating from a hospital Christoph Ort a,*, Michael G. Lawrence a, Julien Reungoat a, Geoff Eaglesham b, Steve Carter b, Jurg Keller a a b
The University of Queensland, Advanced Water Management Centre (AWMC), Brisbane, QLD 4072, Australia Queensland Health Forensic and Scientific Services, Organics Laboratory, QLD 4108, Australia
article info
abstract
Article history:
Pharmaceutical residues in water are frequently analysed and discussed in connection with
Received 14 May 2009
sewage treatment, ecotoxicity and, natural and drinking water quality. Among different
Received in revised form
localities hospitals are suspected, or implied, to be a major and highly variable source of
21 July 2009
pharmaceuticals that substantially contribute to the total wastewater load. In this study, the
Accepted 1 August 2009
contribution of pharmaceuticals from a hospital to a sewage treatment plant (STP) serving
Available online 5 August 2009
around 45,000 inhabitants was evaluated. Approximately 200 hospital beds result in a hospital bed density of 4.4 beds per 1000 inhabitants, which is a typical value for developed world
Keywords:
countries. Prior to sampling, a sound systems analysis was performed, and a sophisticated
Experimental design
continuous flow-proportional sampling regime was applied. Hence, overall experimental
Audit data
uncertainty was reduced to a minimum, and measurements provide clear evidence that, for 28
Quantification
of 59 investigated substances, over 85% of the pharmaceutical residue loads do not originate
Prediction
from the hospital when applying a conservative error estimation. Only for 2 substances,
Loads
trimethoprim (18%) and roxithromycin (56%), was the maximum observed contribution of the
Consumption
hospital >15%. On average, the contribution of the hospital for the compounds detected in both, hospital effluent and sewage treatment plant influent was small and fairly constant. Five compounds were only detected in hospital wastewater, and 24 neither in the hospital wastewater nor in the total wastewater at the influent of the STP. For these compounds no experimental contribution could be calculated. For the compounds where audit data for both the national consumption and the specific hospital under investigation were available, a prediction of the fraction of pharmaceuticals originating from the hospital was performed. Three quarters of the compounds, classified with the existing audit data, were in the same ‘‘hospital contribution category’’ as determined by measurements. For most of the other compounds, plausible reasons could be identified to explain the observed deviations. ª 2009 Elsevier Ltd. All rights reserved.
1.
Introduction
1.1.
Brief overview
Hospital wastewater (HWW) is normally discharged directly, without pre-treatment, to sewers. Despite mostly being only
a small fraction of the total wastewater volume in the influent of a sewage treatment plant (STP), HWW has gained increasing scientific and public attention in the last decade. This is, in part due to the observation and expectation that HWW is a source for undesirable constituents, such as (multi-)antibiotic-resistant bacteria (Baquero et al., 2008; Kummerer, 2004). In other
* Corresponding author. Tel.: þ61 7 3365 4730; fax: þ61 7 3365 4726. E-mail address:
[email protected] (C. Ort). 0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2009.08.002
606
water research 44 (2010) 605–615
publications, the emission from hospitals was estimated for antibiotics, anaesthetics, disinfectants, heavy metals, AOX (Adsorbable Organic Halogens), iodised X-ray contrast media and cytostatic agents (e.g. Kummerer, 2001). The latter were also investigated in detail by Lenz et al. (2007). Furthermore, a number of toxicity assays were performed (Boillot et al., 2008; Ferk et al., 2009; Hartmann et al., 1998). As a result, it has been suggested in some studies that pre-treatment of HWW prior to discharge into the sewers provides a reasonable solution (Gautam et al., 2007; Lenz et al., 2007; Pauwels and Verstraete, 2006). However, this view is not unanimously supported. The separate treatment of HWW to reduce the development of resistant bacteria was questioned (Kummerer, 2009): the substantial amount of antibiotics used outside of hospitals (in Germany more than 75%) seems to be a plausible reason that resistant bacteria are also abundant in wastewater not receiving any HWW. Additionally, Boillot et al. (2008) found quantitatively far fewer microorganisms in the effluents of hospitals than in urban wastewaters which is consistent with other studies. With regard to pharmaceuticals, Lenz et al. (2007) report that 1) for some pharmaceuticals merely a small fraction of the amounts administered in the hospital were actually found in its effluent (i.e. 0.1–0.2% for doxorubicin, 0.5– 4.5% for 5-fluorouracil and 27–34% for total platinum) and 2) a complete onsite wastewater treatment process is needed to significantly remove targeted pharmaceuticals. This includes full physical and biological treatment steps, not only advanced processes. Capturing all sources within a hospital (wards, laboratories) may be further complicated by the fact that different facilities discharge through different pipes to the common sewer. This particularly holds true for large existing hospital complexes. Therefore, local circumstances need to be considered and the contribution of an individual hospital needs to be assessed in relation to the total load in a STP catchment. To our knowledge, only a few publications explicitly quantify pharmaceutical residues (subsequently referred to as ‘pharmaceuticals’) excreted within hospitals compared to the total pharmaceutical load in the corresponding STP influents (Feldmann et al., 2008; Heberer and Feldmann, 2005; Thomas et al., 2007). However, these studies are limited to a small number of pharmaceuticals, or make an assumption on the water flow instead of measuring the wastewater flow onsite to determine actual loads. In view of the local situation in South East Queensland where it is proposed to recycle wastewater for indirect potable reuse, it is sensible to consider whether pre-treatment of HWW will provide a significant benefit. From two previous research papers relevant for the region of interest also dealing with pharmaceuticals the contribution of hospitals cannot be derived (Khan and Ongerth, 2004; Watkinson et al., 2009). Therefore, the goal of our study is to determine accurately the contribution of a hospital to the total pharmaceutical load found at the inlet of the corresponding STP by means of measurements. Additionally, this experimentally data obtained from a limited time period is then compared with readily available audit data. It shall be assessed whether the contribution of a hospital can be predicted reliably without any additional administrative effort, i.e. without extra surveys on the hospital wards for day-specific consumptions. If
measurements matched with the prediction, the same kind (comprehensiveness and quality) of information can be used at other locations to make a prediction, a priori without laborious measurements. The focus of this research is on dissolved pollutants which cannot be eliminated in conventional wastewater treatment. Pollutants showing poor to moderate biological removal need to be transformed by chemical reactions (e.g. oxidation) or separated by physical processes (e.g. adsorption onto activated carbon).
1.2.
Systems analysis
The prediction and experimental quantification of pharmaceutical mass fluxes in the wastewater of a specific STP catchment are laborious. A sound understanding of the whole system is required prior to setting up a predictive model, and performing a confirmative sampling campaign. This particularly holds true when attempting to attribute different fractions to a multitude of individual sources, for example if there are several hospitals and multiple smaller health care facilities in a catchment. Due to the lack of generally accessible consumption data at sufficiently high spatial and temporal resolution, models often provide only a prediction of an average load. Additionally, the latter is prone to uncertainty due to varying transformations of pharmaceuticals during human metabolism. While it would be ideal to have a list of all health care facilities with size, services provided and precise pharmaceutical consumption, just obtaining generally available consumption data is a tedious task in itself. The ‘‘institutional resolution’’ is often not sufficient without additional administrative effort, i.e. temporary surveys of the wards in the hospital(s) under investigation (Feldmann et al., 2008; Kummerer, 2001). Furthermore, the (average) household pharmaceutical consumption needs to be estimated from national or state-wide sales and/or prescription data if regional data is not available. Moreover, collecting representative samples requires a thorough knowledge of the sewer layout and awareness of potentially highly variable concentrations and loads in the course of a day. Clearly, accurate chemical analysis of a nonrepresentative sample is not adequate to characterise a real full-scale system.
1.3.
Sampling issues
Accurately quantifying pharmaceutical loads in hospital effluents or sewers close to any source (sub-catchments, households or industry) is a demanding undertaking. It requires a substantial experimental effort and is still prone to uncertainties. The latter are extremely hard to quantify if sampling is carried out with conventional (unsophisticated) devices, i.e. auto-samplers operated in a discrete sampling mode with (too) long time intervals, or grab samples. Rarely are fluctuations of concentrations and loads assessed in separate experiments at high temporal resolution prior to the ‘‘real’’ measuring campaigns. These pre-experiments are very expensive and may not provide the data to answer the actual research question. However, if the applied sampling protocol does not result in the collection of a representative sample, then the care taken in the following
water research 44 (2010) 605–615
processes of transport, storage, preparation and chemical analyses with a sophisticated method cannot make up for this deficiency (de Gruijter et al., 2006). Subsequent (even sophisticated) statistical analyses of non-representative samples are unreliable and the resulting conclusions will therefore be of limited value. In some cases, the large variation observed in previous studies may not be ‘‘true natural variation’’ but instead, may simply be an artefact caused by inadequate sampling (Ort et al., in preparation). Therefore, strong emphasis has been put on obtaining representative samples for this study. In Ort and Gujer (2006) a method was presented to estimate the required sampling frequency in order to not exceed a certain sampling error. In gravity sewers this results in fairly short time intervals if the substance of interest is contained in a small number of ‘‘wastewater pulses’’ per day (e.g. toilet flushes containing a specific excreted pharmaceutically active compound). Sampling frequencies that are too low result in large sampling uncertainties, especially in the case of only a few patients per day (Weissbrodt et al., 2009). The often claimed problem of ‘‘limited storage capacity in an auto-sampler’’ can be easily solved by replacing the glass bottles more than once per day. This may be more laborious, but it is a much better solution than using a time-proportional sampling mode, which does not take samples weighted according to the flow in the sewer. In contrast, physical boundary conditions such as deep sewers resulting in long dead times for purging the sampling hose or limited access to pressurised sewers are more difficult to overcome.
2.
Material and methods
2.1. Sewage treatment plant and catchment characteristics A total of approximately 45,000 inhabitants in two geographically separated sub-catchments, Morayfield and Caboolture, are connected to the South Caboolture STP (subsequently only referred to as STP) which is operated with two sequencing batch reactors (SBRs). It treats a daily dry weather flow of approximately 10,000 m3. During long dry periods with high level water restrictions, this value can drop to 7500 m3 day1. Morayfield is drained by gravity sewers and contributes two thirds of the total wastewater. It is only pumped once, at the STP itself. Caboolture makes up for one third of the total influent and is a largely pressurised sewer system with numerous pumping stations. At specific times of the day the flow is diverted at the influent of the STP and stored in two large buffer tanks (800 m3 each) before being pumped to the SBRs. This combination of sewers and the complex influent layout of the STP results in very high hydraulic fluctuations (see Fig. 1). Hours with almost zero flow contrast with hours around 250–300 L s1 and in between, the flow varies rapidly and significantly. During wet weather the relative flow variations are less significant due to higher base flow.
supporting information). A small private hospital providing mainly day surgery (only around 10 beds) and a small dental surgery also drain into the same sewer. The wastewater from the private hospital cannot be accessed separately. Other small health care facilities within this sewer catchment make consultations to out-patients, and therefore, the wastewater from these facilities are not expected to significantly add to the pharmaceutical load of the STP. The hospital bed density for the whole STP catchment is 4.4 beds per 1000 inhabitants. All HWW is collected in a sewage pumping station (SPS CT-51, subsequently referred to as SPS) before being pumped to the primary rising main. There is no residential wastewater contributing to this SPS and the hydraulic residence time in the main sewer to the STP is approximately 30 min to 1 h (hydraulic calculations provided by the Regional Council for the decisive time in the morning when samples at the SPS and the STP needed to be coordinated). The average daily volume during dry periods pumped at the SPS is approximately 75 m3 which is 1% of the total wastewater volume discharged to the STP. The occupancy of hospital beds in Caboolture during the sampling period was close to 100% which is representative for the year to date average. Unfortunately no comprehensive database exists with regard to other health care facilities in the catchment of the STP. Hence, an internet search was performed. Four aged care facilities with a total capacity of 443 beds were found (297 high care and 146 low care) with an unknown occupancy rate. Furthermore, a total of 14 addresses for doctors plus 12 dentists were found. If mass fluxes at the influent of the STP were significantly higher than expected from average national consumption and hospital usage, further investigations of these facilities would be warranted.
2.3.
Hospital characteristics
Caboolture Public Hospital has 190 beds and offers all services of a modern regional hospital (listed in Table SI 1, see
Sampling
Continuous flow-proportional sampling modes were applied in this study to minimise sampling error. Continuously diverting a small flow-proportional side stream is conceptually the best solution to obtain representative samples for dissolved compounds. However, low velocities in the side stream prevent proper sampling of solids and long-term operation may lead to biofilm growth. Due to the limited time of sampling biofilm growth is not considered problematic in this instance. Sampling over consecutive days was preferred to the alternative option of collecting samples on single days distributed over a longer period. This drastically reduces the effect of unknown system behaviour: missing a ‘‘decisive’’ HWW packet at the STP is then limited to the first hour of the first day and the last hour of the last day. All other water packets are captured, although they might be attributed to the STP sample a day later. However, this would merely lead to higher variability of the hospital’s contribution and not to a non-quantifiable effect.
2.3.1. 2.2.
607
Sampling protocol for Caboolture Hospital (SPS CT-51)
The HWW is not easily accessible before it enters the SPS. Furthermore it would have been very difficult to set up an accurate flow meter to measure flow in a small open channel with intermittent, partially very low flows and to use this data
608
water research 44 (2010) 605–615
500 Dry weather flow (cv = 0.72) Wet weather flow (cv = 0.44)
Flow [L s-1]
400
300
200
100
0 8am 10am 12pm 2pm
4pm
6pm
8pm 10pm 12am 2am
4am
6am
8am
Time [h] Fig. 1 – Two examples for typical flow patterns at the influent of the sewage treatment plant; cv [ coefficient of variation (standard deviation/mean).
operates at about 2500 L min1 and the sampling side stream was adjusted with the tap to approximately 1 L min1, resulting in a sampling volume of about 10 L per pump cycle. In comparison, the dead volume of the tap installation including hose was 0.5 L (ca. 5% of the sampling volume). The three samples were collected in separate glass bottles, and analysed separately. The concentrations of the individual samples were multiplied with the flow for the corresponding pump cycle, and summed to obtain a 24-h load. Rough diurnal variations could also be determined with this sampling procedure, but they are not relevant for the system and time scales under investigation, and hence they are not further discussed in this paper.
to control the speed of the sampling pump. Instead, plumbers from the Regional Council fitted a tap in the rising main of the SPS (see Fig. 2). The tap is upstream of the non-return valve before the HWW enters the primary rising main leading to the STP. Electricians from the Regional Council installed an actuator after the tap which only opens when the pump of the SPS empties the wet well. Water runs without a sampling pump due to the pressure in the rising main. Under normal operating conditions there are about 24 pumping cycles per day, triggered automatically based on to the water level in the SPS. While it was found that the flow during one cycle is fairly constant, it can vary significantly among cycles due to variable hydraulic conditions in the primary rising main. Therefore, a manual operating mode was adopted, disabling the auto level control. This allowed for using the full storage capacity of the wet well. Starting at 7 AM it was emptied again at 12 PM, 6 PM and 7 AM the following day which required personnel to be present three times per day (confined space). The pump
2.3.2.
Sampling protocol at the sewage treatment plant
To sample for the same ‘‘water packets’’ as at the SPS, sampling started at 7:45 AM in the influent of the STP. The storage tanks start filling at 8 AM and are emptied completely during night
dry pit
control unit
to STP
hose
sample bottle
primary rising main
actuator
pump
tap
non-return valve
wet well SPS CT-51
efflue
nt Ca
boolt
ure H
ospit
al
Fig. 2 – Schematic drawing of the sampling point at the sewage pumping station (SPS) CT-51 (not to scale): All hospital wastewater is discharged to the wet well of the SPS and intermittently pumped to the primary rising main leading to the sewage treatment plant (STP). Upstream of the non-return valve a stand pipe with a tap and an actuator was fitted. This allows for taking flow-proportional samples during individual pump cycles.
water research 44 (2010) 605–615
time, and in the early morning hours. This guarantees that wastewater is not stored and dragged on over different 24-h sampling periods. Flow rates in the influent are routinely measured at high temporal resolution. A wire connected to an analogue digital converter provides a 4–20 mA signal from the PLC (programmable logic controller) linear to the flow in the sewer to control the speed of the sampling pump. The peristaltic pump (Watson Marlow 520UN, programmable interface, water proof casing, equipped with a 520R2 pump head and 3.2 mm tube bore) was tested in the lab to ensure its linear behaviour over the full speed range under similar physical boundary conditions (suction height approximately 2 m, pressure height negligible). The pump speed was set to 0 rpm (revolutions per minute) for 0 L s1 in the sewer (pumping 0 mL min1) and to 34 rpm for 1000 L s1 (pumping 69.4 mL min1). The finest increment of the pump is 0.1 rpm equivalent to 2.9 L s1 wastewater flow in the influent of the STP. With this setup approximately 15 L of wastewater were collected in a 20 L glass bottle which was located in a refrigerated container. Two field blanks were collected: to this end 0.5 L of MilliQ water was used to rinse the sampling tube and subsequently 0.5 L MilliQ water was pumped through the tube to be analysed in the laboratory. No substances were detected above the limit of quantification.
2.4.
Chemical analyses
After collection, the continuously refrigerated samples were transported to the laboratory where they were filtered the same day and preserved before analysis. All samples were analysed for 59 substances by Queensland Health Forensic and Scientific Services (QHFSS). A detailed description of the method consisting of solid phase extraction followed by concentration prior to quantification by LC–MS/MS (liquid chromatography coupled with tandem mass spectrometry) is given in the supplementary information SI 2, accompanied with an alphabetical list of all compounds (Tables SI 2.2 and SI 2.3). As the method does not allow for correction of absolute analytical extraction recoveries in raw wastewater samples, we report relative loads. In order to compare hospital effluent samples with samples from the influent of the STP, it is necessary to assume that matrix effects between these sample types are similar. Any systematic error in recovery is therefore cancelled out when calculating ratios of loads, i.e. contribution of HWW to the total influent of the STP.
2.5.
Uncertainty assessment
Flows in completely filled pressurised pipes can be measured more accurately than flows in open water channels (gravity flow). For this study a maximum error of 10% was assumed, which equals to 6% (¼10/30.5) as single standard deviation of a normal distribution. For chemical analysis a random uncertainty (reproducibility) of 20% for all compounds was chosen (see Tables SI 2.1–2.3). The two errors are independent, and Gaussian error propagation results in an overall uncertainty estimate for calculated loads of 21% (¼[62 þ 202]0.5). The flow-proportional continuous sampling procedure covers all fluctuations in the wastewater over time. Since it is a reasonable assumption that dissolved compounds are completely mixed over the whole pipe cross section in the
609
influent works, no additional errors need to be taken into account due to sampling.
2.6.
Pharmaceutical audit data
2.6.1.
National consumption
An extract from the DUSC database (Drug Utilisation SubCommittee) for the year 2008 is listed for the compounds investigated in this study (see supporting information, Table SI 3). It comprises the sum of subsidised drugs (subsidised under the Pharmaceutical Benefits Scheme (PBS) and the Repatriation Pharmaceutical Benefits Scheme (RPBS) and processed by Medicare Australia) and non-subsidised drugs (under PBS copayment and private prescriptions). The amounts of non-subsidised drugs were estimated from continuous data on all prescriptions dispensed from a validated sample of 370 community based pharmacies. The available data do not include drugs dispensed to public hospital in-patients, pharmacy over-the-counter drugs (i.e. non-prescription) and drugs supplied by supermarkets.
2.6.2. Amounts administered to in-patients in Caboolture Public Hospital No specific survey was carried out during the sampling period on the wards. Routinely stored audit data for a current 12-month period (2007–2008) was made available by the pharmacy of the Caboolture Public Hospital. For each pharmaceutical, a specific database query was performed to derive the amounts exclusively used for hospitalised in-patients; pharmaceuticals given to out-patients (in consultations and pharmacy) were not considered, since they will be taken and excreted at home. The total annual hospital load was determined after summing the contributions of all medications containing the pharmaceutically active compound of interest. It has to be noted that the amounts derived from this database are amounts supplied by the pharmacy to the individual wards and not the amounts effectively administered. However, it is generally not the hospital’s policy to discard drugs to the (solid or liquid) waste system, both from a financial and environmental point of view. Nevertheless, some unused drugs for in-patients may be collected on the wards and returned to the pharmacy for reuse or proper disposal. Hence, these drugs do not contribute to the load in the HWW. However, in discussion with relevant hospital staff these amounts are considered to be very limited and are not assessed within this study.
3.
Results and discussion
3.1.
Evaluation of wastewater volumes
The four consecutive weekdays, mid-February 2009 when sampling took place, were during a wet period, with flows 1.5– 2 times higher than normal dry weather flow (i.e. surface runoff in catchments and infiltration to sewage pumping stations). In Table 1 the flows at the two sampling locations over the corresponding 24-h periods are summarised. During the sampling period, the hospital contributed less than 1% of the total wastewater flow to the STP.
610
water research 44 (2010) 605–615
Table 1 – Wastewater volumes over 24 h at the SPS CT-51 (hospital wastewater) and the influent to the STP. Influent STP [m3]
7 AM–7 AM of the following day [m3]
Fraction of influent STP [%]
14,064 16,921 19,059 14,347
109 129 138 127
0.8 0.8 0.7 0.9
Evaluation of relative pharmaceutical loads
The concentrations in the hospital effluent were on average 2 times higher than in the STP influent. The STP influent loads show only little day-to-day variation (cv ¼ 0.06, cv ¼ coefficient of variation ¼ standard deviation/mean). Day-to-day variation is smaller than the estimated overall uncertainty. The loads in the hospital effluent varied more (cv ¼ 0.27). On average the hospital contributed only 1.8% to the total atenolol load in the influent of the STP. For a conservative error estimation, a maximum contribution of the hospital was calculated by dividing the upper uncertainty value of the hospital effluent by the lower uncertainty value of the STP influent for each day (see Fig. 3). Over all four days, the highest maximum contribution for atenolol was 3.5%. Over all four days, the smallest minimum contribution for atenolol was 0.9% (analogue procedure as in ). The prediction for an average contribution of the hospital based on audit data is 0.6% (see more details in Section 3.4).
fraction of STP influent
fraction of STP influent
day 3
day 2
day 4 7% % day 4 16
0.00
7-
STP influent
% day 3
0.0
0.05 % day 2
0.2
13
0.00
0.4
6-
% day 4
day 3 8%
10 4-
day 2 7%
3-
day 1 7%
3-
3-
STP influent
day 4
day 3
0.0
0.02 day 2
0.2
15
0.4
0.10 0.6
6-
0.6
1.20 0.15
0.8
day 4
0.04
day 3
0.8
Hospital effluent
1.0
day 2
0.06
Trimethoprim 1.2
day 1
Hospital effluent
1.20
Loads (normalised to max. STP influent load)
D
1.0
0.00
fraction of STP influent
Paracetamol 1.2
day 1
Loads (normalised to max. STP influent load)
C
0.01
3-
STP influent
3%
0.0
1-
0.2
1-
day 4
0.02
0.4
% day 1
day 3
2%
0.00
0.03
0.6
1-
day 2
4% 2-
day 1 3%
1-
1-
STP influent
2%
day 4
day 3
0.0
day 2
0.2
0.01
0.8
18
0.4
0.04
day 1
0.02 0.6
1.20
1.0
2%
0.8
Hospital effluent
8-
1.0
0.03
Gabapentin 1.2
day 4
Hospital effluent
1.20
Loads (normalised to max. STP influent load)
B
Atenolol 1.2
day 1
Loads (normalised to max. STP influent load)
A
2%
Concentration values for atenolol in the influent of the STP were, on average, 10 times higher than the limit of quantification (LOQ).
1-
To obtain relative pharmaceutical loads, measured concentrations were multiplied with the corresponding 24-h flow at each sampling location and normalised by the highest STP influent load. Four examples representing four different groups of pharmaceuticals are charted in Fig. 3. Absolute concentration values are not reported because they are difficult to compare among different studies; they highly depend on the sewer system (separate or combined) and on the hydraulic conditions (dry or wet weather flow). The key figures chosen for statistical evaluation are presented in Table 2, and discussed subsequently in detail for one example (atenolol, a beta-blocker, see also Fig. 3A). The numbers in black circles ( ) refer to the corresponding column in Table 2:
day 3
16/2/09 17/2/09 18/2/09 19/2/09
day 2
3.2.
1 2 3 4
7:45 AM – 7:45 AM of the following day
day 1
Day Day Day Day
Hospital wastewater (flow at SPS)
fraction of STP influent
Fig. 3 – Measured, relative pharmaceutical loads over 24-h periods in the influent of the STP and effluent of the hospital for four consecutive weekdays. Error bars include uncertainty of flow measurements (±6%) and chemical analysis (±20%), resulting in an overall uncertainty of ±21% (single standard deviation). Note the different scales for the y-axis of STP influent and hospital effluent.
611
water research 44 (2010) 605–615
Table 2 – Classification of substances according to the contribution of the hospital to the total load in the influent of the STP (see Section 3.3 for more explanations of key figures marked with black circles ). LOQs for all compounds are between 0.1 and 2 mg LL1. Classification
Substance
Therapeutic group f
CSTP LOQ
CHospital CSTP
Coefficient of variation for loads
Contribution of hospital wastewater [% of total STP influent] Measured
Influent STP Max 5%
Min
Mean
Max
average predicted with audit datad
0.34 0.41 0.33 0.65 0.23 0.27
0.9 – 1.4 0.0 0.0 – 0.5 0.1 – 1.1 1.4 0.8 0.4 – 0.2 0.7 0.9
1.8 3c 2.6 0.4 0.4 4c 1.5 0.2 1c 2.0 2.1 2.3 1.2 3c 0.8 1.6 2.0
3.5 – 4.4 1.3 1.2 – 3.7 0.3 – 3.7 3.2 4.4 2.8 – 2.2 3.1 5.0
0.6 0.9 – 1.9 5.9 1.6 6.6 – 1.8 0.5 – 0.3 1.5 5.0 6.7 4.3 2
Hospital
Atenolol Atorvastatin Caffeine Carbamazepine Cephalexina Citalopram Codeine DEET Diclofenac Hydrochlorthiazide Iopromideb Naproxen Oxazepam Oxycodone Sulphamethoxazole Temazepam Venlafaxine
BB HL – AC AB AD AG IR AI DI XC AI AL AG AB SE AD
10.7
2.0 – 3.0 0.6 0.9 – 1.8 0.2 – 2.7 3.9 3.0 1.2 – 1.1 1.9 2.6
0.06 – 0.07 0.17 0.21 – 0.08 0.22 – 0.24 0.20 0.11 0.08 – 0.05 0.06 0.16
5 % < max < 15 %
Erythromycin Furosemide Gabapentin Gemfibrozil Ibuprofen Metoprolol Paracetamol Ranitidine Salicylic acid Tramadol Triclosan
AB DI AC HL AI BB AG HB m AG BI
7.7 13.3 56.5 3.9 70.6 3.5 1293.6 3.2 60.1 11.0
2.8 6.9 3.2 5.3 6.3 4.6 6.8 6.2 4.9 3.4 –
0.41 0.20 0.13 0.26 0.08 0.19 0.11 0.12 0.26 0.18 –
0.28 0.17 0.49 0.85 0.18 0.29 0.10 0.38 0.18 0.26 0.25
0.8 2.6 1.0 0.7 2.7 2.0 2.8 1.3 1.8 1.2 –
2.6 5.8 2.3 4.1 4.6 4.1 5.1 4.9 4.9 2.5 6c
5.5 13.7 6.8 10.0 8.5 7.0 9.8 11.0 10.8 6.0 –
4.3 5.9 4.6 0.4 49 2.3 10 5.7 11 6.7 –
max > 15%
Roxithromycin Trimethoprim
AB AB
1.4 3.4
28.4 13.3
0.22 0.11
0.20 0.15
11.68 5.7
25.66 10.14
56.0 18.3
19 14
All values at STP < LOQ
10 Ciprofloxacin AB
All values < LOQe
0.27 0.47 0.17 0.65 1.33 0.19 0.47 0.18 0.53 0.30 –
– Not available. a Only detected twice in hospital (influent STP four times). b Only detected once in hospital (influent STP twice). c Calculated with average loads measured in the influent of the same STP (three non-consecutive days in 2008 during very low dry weather flows). d Audit data for hospital 2007–2008, audit data for national consumption 2008 (DUSC database). e Numbers in brackets are the fraction of the hospital based on audit data (same as for column for the other compounds). f AB ¼ antibiotic, AC ¼ anticonvulsant, AD ¼ antidepressant, AG ¼ analgesic, AI ¼ anti-inflammatory, AL ¼ anxiolytic, BB ¼ beta-blocker, BI ¼ biocide, DI ¼ diuretic, HB ¼ histamine blocker, HL ¼ hypolipidemic agent, IR ¼ insect repellent, m ¼ metabolite, SE ¼ sedative, XC ¼ X-ray contrast media. g More than 97% of national consumption used in agriculture (Watkinson et al., 2009). h When assuming CSTP ¼ LOQ.
612
water research 44 (2010) 605–615
Table 3 – Comparison with other hospital wastewater studies. Number of hospitals
Number of beds Investigated substances per 1000 inhabitants (% in influent of the corresponding STP originating from hospitals)
2
4.4
1
3.6
More than 5
12.1
More than 5
12.1
diclofenac (1.4)a ibuprofen (0.7) a metoprolol (1.5) a paracetamol (12) a tetracycline (0.5) a trimethoprim (14) a 5 X-ray contrast media (50) b cytostatics (max 7.5) b carbamazepine (15) c diclofenac (10) c metamizol (50) c
Location of study
Oslo, Norway (Thomas et al., 2007)
Winterthur, Switzerland (Weissbrodt et al., 2009)
Berlin, Germany (Heberer and Feldmann, 2005) Berlin, Germany (Feldmann et al., 2008)
a Concentrations measured over 12 weeks, loads estimated with water consumption, sum of the two major hospitals (an unknown number of other smaller hospitals/health care facilities are located in the catchment). b Influent STP was not measured in this study, percentage refers to loads of pharmaceuticals quantified in the hospital’s effluent compared to day-specific administered amounts. c Only measured in the effluent of one hospital and then extrapolated for the whole catchment based on audit data of the other hospitals.
Classification of all substances according to maximum observed contribution from the hospital ( ). The consistent results for atenolol are reflected across most of the 30 detected substances. Representatives of other pharmaceutical groups show also fairly constant loads over the four-day period: gabapentin (an anticonvulsant), paracetamol (an analgesic) and trimethoprim (an antibiotic, see Fig. 3B–D). From the 59 substances, 5 were detected only in the HWW but not in the influent of the STP and 24 substances were not detected above the LOQ in any of the samples. The 30 substances detected at both locations were classified for the hospital’s contribution to the total influent of the STP. To this end, the maximum observed contribution including uncertainty as a conservative estimate was used (see description before in ). The hospital’s contribution for 17 substances was at all times ‘‘smaller than 5%’’, 11 additional substances fall in
Four out of the 5 substances only detected in the HWW were just above the LOQ. With the 100 fold dilution in the influent of the STP the LOQ would have to be at least three orders of magnitude lower to reliably quantify the hospital’s (high) contribution. When assuming that the concentrations in the influent of the STP were equivalent to the corresponding LOQ, only a one-sided estimation with regard to the hospital’s contributions from >5% up to >50% can be made. However, in some cases this deviates from the prediction based on audit data (see chapter 3.3).
3.3.
Comparison with audit data
If the consumption of pharmaceuticals in a STP catchment can be estimated from existing national sales or prescription data, and audit data for the hospital are available, the contribution of the hospital can be calculated with the following equation
ConsCab:Hosp: $excretion ratio ConsCab:Pop: $excretion ratio þ ConsCab:Hosp: $excretion ratio measured loadðhospitalÞ$recovery$accuracy ConsAUS y $45; 000 with ConsCab:Pop: ¼ 20; 000; 000 measured loadðSTP catchmentÞ$recovery$accuracy contributionðhospitalÞ ¼
the category ‘‘smaller than 15%’’ and only 2 substances were ‘‘above 15%’’ (trimethoprim and roxithromycin with a worst case estimate of 18% and 56% respectively). For most substances quantified in both STP influent and hospital effluent, the variations of the loads in the HWW were on average 2.4 times higher than in the influent to the STP. The small number of hospital patients compared to the potentially large number of individuals taking these pharmaceuticals at home is a valid explanation for this observed difference in variation.
ð1Þ
where Cons is the consumption, Cab stands for Caboolture, Pop. for population, AUS for Australia and Hosp. for hospital. It becomes evident that the transformation due to human metabolism (excretion ratio) cancels out of the equation when assumed to be similar for patients in the hospital and for people at home. The consumption of pharmaceuticals in the STP catchment is estimated by calculating an average per capita consumption from the national consumption data multiplied with the number of inhabitants in the catchment. The consumption of in-patients in the hospital is added to the
water research 44 (2010) 605–615
domestic consumption to obtain an estimate for the total STP influent load (see also Table SI 3). The prediction for 27 compounds where both national and hospital audit data were available, in some cases deviated significantly from the experimentally determined values. However, only 8 substances would have been classified differently based on audit data when applying strict boundaries for the classification which does not change the overall picture substantially. Possible reasons for three examples are briefly discussed: 1) The overestimation in the case of ibuprofen may be reasonably explained by the fact that the national consumption is likely to be substantially underestimated because ibuprofen can also be obtained over the counter and in supermarkets without prescription. 2) A patient who regularly takes histamine blockers (at home) is likely to take them with him if he is being hospitalised (for any treatment not related or interfering with histamine blockers). This is one of the cases where patients may bring their own medication to the hospital and is also assumed to be valid for beta-blockers and diuretics. 3) In some countries trimethoprim is often applied together with sulphamethoxazole (combination item) and hence would be expected in a similar ratio. In Australia, the consumption pattern is different: 70% of trimethoprim is sold as single item (general public) and in the hospital under investigation even 90% are administered as individual compound. In other cases the explanation may be sought in a higher or lower than average number of patients being treated during the sampling period in the hospital. However, if the number of treated patients shall be estimated from measurements, the excretion ratio and absolute recoveries for chemical analyses need to be taken in to account (see Eq. (1)). This makes it difficult to compare measured influent loads from a STP with audit data from an individual health care facility to reliably calculate the health care facility’s contribution to the total influent of the STP.
3.4.
Comparison with other studies
The Caboolture catchment, with 4.4 beds per 1000 inhabitants, is comparable with two other studies (3.6 and 4.4 beds per 1000 inhabitants, see Table 3). Without audit data for the hospitals and general public, the load estimations based on measured concentrations and an estimate for wastewater based on average water consumption in the study by Thomas et al. (2007) make a direct comparison difficult. However, higher contributions were also found for paracetamol and trimethoprim. In the study by Weissbrodt et al. (2009) the loads at the influent of the STP were not measured. The percentage determined in this study is the amounts measured in the sewer divided by the amounts administered on the corresponding days. The compounds investigated in the Swiss study are iodinated X-ray contrast media and cytostatics, both compounds almost exclusively administered in hospitals. Only 50% of the X-ray contrast media and a maximum of 7.5% of the cytostatics were quantified in the hospital’s effluent, implying that the remaining part is most likely ‘‘carried home’’ by patients and excreted in household toilets. In the studies by Heberer and Feldmann (2005) and Feldmann et al. (2008) the hospital bed density is significantly higher (12.1 beds per 1000 inhabitants) with a sub-catchment bed density of 24. Pharmaceutical loads
613
were measured in the influent of the STP and in selected hospital effluents. With day-specific hospital consumption data the contribution of the other hospitals was estimated, resulting in a total hospital contribution of 15% (carbamazepine), 10% (diclofenac) and 50% (metamizole, not measured in our study). Although the results seem to be in good agreement with our study, the limited number of compounds, the various approaches used, and the different catchment characteristics preclude a comprehensive comparison.
3.5. Hospital wastewater treatment and catchments in South East Queensland Over 800 pharmaceuticals, disinfectants and other substances are recorded in the DUSC and the hospital database. Whilst the 59 substances analysed for in this study presents one of the more comprehensive studies of the relative contribution of a hospital to total load in wastewater, we do not claim that these results can be extrapolated for each of these 800 substances, at all hospitals, or medical research activities in general. As is often the case, the selection of these 59 substances was based upon the availability of a validated analytical method. Despite this ‘‘limitation’’, even if there were substances that originate almost exclusively from hospital wastewater, or if measures were taken to prevent pharmaceutical residues entering hospital wastewater (source control, separate collection of urine and faeces (Heinzmann et al., 2008)) or if hospital wastewater was treated on site, over 85% of the total load for the majority of the pharmaceuticals investigated in this study would still reach to the STP because they are excreted by the public at home in their households. Even for very specific compounds, almost exclusively administered in hospitals, the trends in many health care systems are moving towards shorter hospitalisations or even treatment of out-patients (particularly diagnostics). Two examples are the iodinated X-ray media and cytostatics: although administered in high amounts in hospitals, they cannot be recovered to 100% and hence solely attributed to hospital effluent (Weissbrodt et al., 2009). Relevance to other catchments in South East Queensland (SEQ): The three catchments of main interest within SEQ, the ones with advanced water treatment plants for providing purified recycled water to the region (for planned indirect potable reuse scheme), have approximately 8 hospital beds per 1000 inhabitants (Luggage Point, eleven hospitals), 0.4 (Gibson Island, one hospital) and 1.7 (Bundamba, five hospitals). While the hospital in Caboolture (this study) contributes 4.2 beds per 1000 inhabitants (total in the catchment 4.4), the biggest individual hospital in the catchment of Luggage Point accounts for only 1.5. A desktop exercise analysing audit data from the sum of all hospitals in these catchments is proposed to evaluate if further steps are required. This includes the planning of future sampling campaigns and the potential benefit of treating some hospitals’ wastewater at the source.
4.
Conclusions
Measurements: For several, widely applied pharmaceuticals, an individual hospital seems to be a small additional point
614
water research 44 (2010) 605–615
source in the catchment of a sewage treatment plant. In this study a hospital with 4.4 hospital beds per 1000 inhabitants contributed less than 15% to the total load in the influent of the sewage treatment plant for 28 substances, detected in both hospital effluent and STP influent, which is in good agreement with estimates from other studies. Considering a conservative worst case uncertainty estimation, the hospital contribution only exceeded 15% for two substances, roxithromycin (max. 56%) and trimethoprim (max. 18%). Audit data: The contribution of the hospital calculated with audit data and the chosen classification reveals good agreement with actual measurements for three quarters of the substances. National audit data to calculate the consumption by the general public in a catchment and hospital data for in-patients appear to be good predictors. This approach can be used with some confidence for substances where no analytical method exists to experimentally determine concentrations and loads or where the LOQ is not low enough. This needs to be tested for other countries (dependant upon the comprehensiveness and quality of national and hospital audit data). Sampling in general: Sampling campaigns in hospital wastewater are prone to high uncertainty due to a highly dynamic system (flow and concentrations). All effort should be undertaken to understand the system (behaviour) prior to setting up a sound sampling protocol to ensure that representative samples can be obtained. Other catchments in South East Queensland: The preliminary analysis based on hospital bed densities suggests focusing on the catchment of the STP at Luggage Point (approximately 8 hospital beds per 1000 inhabitants). However it has to be noted that this hospital bed density consists of 3 major public hospitals and a series of private hospitals. Since measurements will be very expensive to assess all hospitals’ contributions. A detailed desktop analysis of all audit data is planned to identify if there are major sources and if measurements at selected locations may be appropriate. Hospital wastewater treatment: If, for whatever motivation, hospital wastewater shall be treated separately onsite, it must be noted, that for many substances no major overall reduction can be achieved since many pharmaceuticals are taken on a regular basis at home. With the current trend to shorter hospitalisations and treatments (diagnostics) of outpatients, this also holds true for compounds mainly administered in hospitals.
Acknowledgments We would like to acknowledge the following institutions and individuals who contributed to this study: the South East Queensland Urban Water Security Research Alliance (financial support); David Fillmore, Rick Jones and Wayne Batchler from Moreton Bay Water (assistance in setting up proper sampling points and support during the campaign); Benjamin Tan and Mary Hodge from Queensland Health
Forensic and Scientific Services (logistics and processing samples); John Doonan, Greg Jackson and Daniel Field from Queensland Health (making contacts and providing hospital data); Maxine Robinson and Chris Raymond from The Drug Utilisation Sub-Committee of the Pharmaceutical Benefits Advisory Committee, Commonwealth Department of Health and Aged Care (providing Australian drug-use statistics); Christa McArdell from Eawag (valuable discussion and review of manuscript); and the Swiss National Science Foundation (Grant PBEZP2-122958 awarded to the first author).
Appendix. Supplementary information Information about the supplementary material can be found at doi:10.1016/j.watres.2009.08.002.
references
Baquero, F., Martı´nez, J.-L., Canto´n, R., 2008. Antibiotics and antibiotic resistance in water environments. Current Opinion in Biotechnology 19, 260–265. Boillot, C., Bazin, C., Tissot-Guerraz, F., Droguet, J., Perraud, M., Cetre, J.C., Trepo, D., Perrodin, Y., 2008. Daily physicochemical, microbiological and ecotoxicological fluctuations of a hospital effluent according to technical and care activities. Science of the Total Environment 403 (1–3), 113–129. de Gruijter, J., Brus, D., Bierkens, M., Knotters, M., 2006. Sampling for Natural Resource Monitoring. Springer. Feldmann, D.F., Zuehlke, S., Heberer, T., 2008. Occurrence, fate and assessment of polar metamizole (dipyrone) residues in hospital and municipal wastewater. Chemosphere 71 (9), 1754–1764. Ferk, F., Misik, M., Grummt, T., Majer, B., Fuerhacker, M., Buchmann, C., Vital, M., Uhl, M., Lenz, K., Grillitsch, B., Parzefall, W., Nersesyan, A., Knasmuller, S., 2009. Genotoxic effects of wastewater from an oncological ward. Mutation Research – Genetic Toxicology and Environmental Mutagenesis 672 (2), 69–75. Gautam, A.K., Kumar, S., Sabumon, P.C., 2007. Preliminary study of physico-chemical treatment options for hospital wastewater. Journal of Environmental Management 83 (3), 298–306. Hartmann, A., Alder, A.C., Koller, T., Widmer, R.M., 1998. Identification of fluoroquinolone antibiotics as the main source of umuC genotoxicity in native hospital wastewater. Environmental Toxicology and Chemistry 17 (3), 377–382. Heberer, T., Feldmann, D., 2005. Contribution of effluents from hospitals and private households to the total loads of diclofenac and carbamazepine in municipal sewage effluents – modeling versus measurements. Journal of Hazardous Materials 122 (3), 211–218. Heinzmann, B., Schwarz, R.-J., Schuster, P., Pineau, C., 2008. Decentralized collection of iodinated X-ray contrast media in hospitals – results of the feasibility study and the practice test phase. Water Science and Technology 57 (2). Khan, S.J., Ongerth, J.E., 2004. Modelling of pharmaceutical residues in Australian sewage by quantities of use and fugacity calculations. Chemosphere 54 (3), 355–367.
water research 44 (2010) 605–615
Kummerer, K., 2001. Drugs in the environment: emission of drugs, diagnostic aids and disinfectants into wastewater by hospitals in relation to other sources – a review. Chemosphere 45 (6–7), 957–969. Kummerer, K., 2004. Resistance in the environment. Journal of Antimicrobial Chemotherapy 54 (2), 311–320. Kummerer, K., 2009. Antibiotics in the aquatic environment – a review – part II. Chemosphere 75, 435–441. Lenz, K., Mahnik, S.N., Weissenbacher, N., Mader, R.M., Krenn, P., Hann, S., Koellensperger, G., Uhl, M., Knasmuller, S., Ferk, F., Bursch, W., Fuerhacker, M., 2007. Monitoring, removal and risk assessment of cytostatic drugs in hospital wastewater. Water Science and Technology 56 (12), 141–149. Ort, C., Gujer, W., 2006. Sampling for representative micropollutant loads in sewer systems. Water Science and Technology 54 (6–7), 169–176. Ort, C., Lawrence, M.G., Joss, A. Sampling for pharmaceuticals and personal care products to assess representative
615
environmental loads: sewers and wastewater treatment plants, in preparation. Pauwels, B., Verstraete, W., 2006. The treatment of hospital wastewater: an appraisal. Journal of Water and Health 4 (4), 405–416. Thomas, K.V., Dye, C., Schlabach, M., Langford, K.H., 2007. Source to sink tracking of selected human pharmaceuticals from two Oslo city hospitals and a wastewater treatment works. Journal of Environmental Monitoring 9 (12), 1410– 1418. Watkinson, A.J., Murby, E.J., Kolpin, D.W., Costanzo, S.D., 2009. The occurrence of antibiotics in an urban watershed: from wastewater to drinking water. Science of the Total Environment 407, 2711–2723. Weissbrodt, D., Kovalova, L., Ort, C., Pazhepurackel, V., Moser, R., Hollender, J., Siegrist, H., McArdell, C., 2009. Mass flows of Xray contrast media and cytostatics in hospital wastewater. Environmental Science & Technology 43 (13), 4810–4817.
water research 44 (2010) 616–624
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
A three-compartment model for micropollutants sorption in sludge: Methodological approach and insights Maialen Barret, Dominique Patureau, Eric Latrille, He´le`ne Carre`re* INRA, UR050, Laboratoire de Biotechnologie de l’Environnement, Avenue des Etangs, 11100 Narbonne, France
article info
abstract
Article history:
In sludge resulting from wastewater treatment, organic micropollutants sorb to particles
Received 11 June 2009
and to dissolved/colloidal matter (DCM). Both interactions may influence their physical and
Received in revised form
biological fate throughout the wastewater treatment processes. To our knowledge, sludge
17 August 2009
has never been considered as a three-compartment matrix, in which micropollutants
Accepted 20 August 2009
coexist in three states: freely dissolved, sorbed-to-particles and sorbed-to-DCM. A meth-
Available online 27 August 2009
odology is proposed to concomitantly determine equilibrium constants of sorption to
Keywords:
as model compounds for the experiments. The logarithm of estimated equilibrium
Activated sludge
constants ranged from 3.1 to 4.3 and their usual correlation to PAH hydrophobicity was
Bioavailability
verified. Moreover, PAH affinities for particles and for DCM could be compared. Affinity for
Isotherm
particles was found to be stronger, probably due to their physical and chemical charac-
Sorption model
teristics. This work provided a useful tool to assess the freely dissolved, sorbed-to-particles
Xenobiotic
and sorbed-to-DCM concentrations of contaminants, which are necessary to accurately
particles (Kpart) and to DCM (KDCM). Polycyclic Aromatic Hydrocarbons (PAHs) were chosen
predict their fate. Besides, guidelines to investigate the link between sorption and the fundamental concept of bioavailability were proposed. ª 2009 Elsevier Ltd. All rights reserved.
1.
Introduction
Organic micropollutants such as Polycyclic Aromatic Hydrocarbons (PAHs) present very low water solubility and are highly hydrophobic. Due to these properties, they tend to sorb to organic matter. In the context of wastewater treatment, sorption phenomena firstly impact on physical fate of these compounds, determining the fluxes distribution throughout the physical separation steps (eg. settling, centrifugation). But these physical mechanisms only determine micropollutants dispersion, whereas biological mechanisms are mainly responsible for their removal. Biological mechanisms may also be influenced by sorption, throughout bioavailability limitation towards biodegradation. Bioavailability might be the main limiting factor of PAHs biodegradation in sludge.
Micropollutants bioavailability has been widely investigated in soils and many tools were subsequently designed to quantify it (Reid et al., 2000; Tang et al., 2002; van Straalen et al., 2005; Oleszczuk, 2007; Hickman et al., 2008; Oleszczuk, 2008). However, these studies referred to different definitions of bioavailability. Moreover, bioavailability and bioaccessibility were sometimes confused. Semple et al. (2004, 2007) gathered most of the encountered concepts into two definitions: a bioavailable compound is ‘‘freely available to cross an organism’s cellular membrane from the medium the organism inhabits at a given time’’; a bioaccessible compound is ‘‘available to cross an organism’s cellular membrane from the environment, if the organism has access to the chemical’’. These definitions imply that a bioaccessible compound can become bioavailable after some time.
* Corresponding author. Tel.: þ33 468 425 168; fax: þ33 468 425 160. E-mail address:
[email protected] (H. Carre`re). 0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2009.08.029
water research 44 (2010) 616–624
Nomenclature
Caqu
DCM DM
KDCM
PAH Cfree CDCM Cpart
Dissolved and Colloidal Matter Dry Matter of the sludge, which includes particles and DCM Polycyclic Aromatic Hydrocarbon Concentration of freely dissolved PAHs (mg/mL) Concentration of sorbed-to-DCM PAHs (mg/gDCM) Concentration of sorbed-to-particles PAHs (mg/gPART)
On the contrary, micropollutants bioavailability in sludge has been little studied. Related concepts and mechanisms are less precisely described than in soil. Some of them may be similar between soil and sludge, although matrix characteristics and micropollutants/matrix contact time completely differ. From a general point of view, micropollutants are usually assumed to be bioavailable when they are located in the aqueous phase and bioaccessible when they are initially sorbed-to-particles and can be transferred to the aqueous phase during the process (Byrns, 2001; Artola-Garicano et al., 2003; Langford et al., 2005; Urase and Kikuta, 2005; Dionisi et al., 2006). Otherwise, the assumption that a fraction of sorbed-to-particles compounds would be directly bioavailable can also be encountered (Fountoulakis et al., 2006). Anyway, even if the nature of the link between sorption phenomena and bioavailability was never demonstrated and is still subject to controversy when studying sludge, the existence of a link is generally admitted. Thus, to study sorption phenomena appears to be the first step towards bioavailability investigation. In published studies dealing with micropollutants sorption in sludge, sludge have been considered as a two-compartment system, corresponding to two states for micropollutants: sorbed-to-particles and aqueous (Byrns, 2001; Artola-Garicano et al., 2003; Ternes et al., 2004; Dionisi et al., 2006; Carballa et al., 2008). However, sorption phenomena also occur inside aqueous phase. Indeed, micropollutants have been shown to interact with dissolved and colloidal matter (DCM) of various origins (Chiou et al., 1986; Perminova et al., 2001) including sludge from wastewater treatment plant (Holbrook et al., 2004). The sorption to DCM is likely to influence micropollutants bioavailability in sludge as well as sorption to particles (Mackay and Fraser, 2000; Vinken et al., 2004). Thus, the freely dissolved and sorbed-to-DCM micropollutants states should be differentiated. Sludge should then be considered as a threecompartment matrix, with three states of micropollutants: freely dissolved, sorbed-to-DCM and sorbed-to-particles. To our knowledge, such a model has been used to study sorption phenomena in aquatic environmental systems such as sediments (Schrap and Opperhuizen, 1992; Lee and Kuo, 1999; Amiri et al., 2005) but it has never been applied to sludge. Micropollutants sorption equilibria in environmental samples including sludge are often modelled by Freundlich isotherms with a Freundlich coefficient close to 1, which is equivalent to a linear isotherm (Hung et al., 2004; Ivashechkin et al., 2004; Arias-Estevez et al., 2007). As a consequence, the equilibrium of sorption to particles is usually assumed to fit linear equations (Ternes et al., 2004; Carballa et al., 2006; Dionisi et al., 2006; Katsoyiannis et al., 2006) as well as the
Kpart Kglobal
617
Concentration of apparently dissolved micropollutant (mg/mL) Equilibrium constant of PAHs sorption to DCM (mL/gDCM) Equilibrium constant of PAHs sorption to particles (mL/gPART) Coefficient of PAHs partition between aqueous phase and particles (mL/gPART)
equilibrium of sorption to DCM (Yamamoto et al., 2003; Zhou et al., 2007). The objective of this study was to propose and to validate a methodological approach to study micropollutants sorption equilibria in sludge, considering sludge as a three-compartment matrix. Even though this investigation focused on PAHs, the methodology and its implications may be transposed to a wider range of micropollutants.
2.
Materials and methods
2.1.
Sludge source
All experiments were performed using activated sludge from a urban wastewater treatment plant. The influent was entirely domestic. The plant consisted in preliminary treatments, primary settling, aerated tank, secondary settling and thermophilic anaerobic digestion of sewage sludge. Hydraulic retention time in the aerated tank was 0.36 day. After collection, sludge was divided into several samples and stored at 20 C in order to guaranty that all following experiments were performed with exactly the same matrix. DCM fraction represented the dissolved and colloidal matters which were not differentiated, defined as the centrifugation supernatant (12 000 g, 20 min, 35 C) filtered at 1.2 mm (Whatman GF/C glass microfiber filter). As some particles were unwillingly suspended back after centrifugation because of little pellet cohesiveness, the filtration step was performed to remove these particles. The DCM mass retained by the filter was negligible, as the DCM fraction beyond the size of 1.2 mm was insignificant (data not shown). On total sludge and on DCM fraction, the dry matter (DM) was measured by weighing the samples after heating at 105 C during 24 h and the organic matter after heating at 550 C during 2 h. The defrosted activated sludge contained 65.4 1.2 gDM/L, constituted of 7 1% of DCM and 93 2% of particles. Organic matter respectively accounted for 72 and 80% of particles and DCM dry matter.
2.2.
Chemicals
All solvents were purchased from J.T.Baker. Mixtures are indicated in volume percentage. 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 and indeno(1,2,3,c,d)pyrene powders were obtained from Dr Ehrenstorfer GmbH. Each PAH was dissolved in
618
water research 44 (2010) 616–624
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. This mixture was used in equilibrium experiments, as competitive phenomena are often assumed not to occur at low concentration (Krauss and Wilcke, 2005; Dionisi et al., 2006; Zhou et al., 2007). 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.
Micropollutant
Particle
Fig. 1 – Representation of the three-compartment model of a micropollutant in sludge.
KDCM ¼
CDCM Cfree
(2)
where Kpart is the equilibrium constant of micropollutant sorption to particles (mL/gPART) and KDCM is the equilibrium constant of sorption to DCM (mL/gDCM). Cfree is very difficult to obtain experimentally whereas Caqu (concentration of apparently dissolved micropollutant, sum of the freely dissolved and sorbed-to-DCM states, mg/mL) and Cpart are easily measurable. From Caqu and Cpart measurement, Kglobal (mL/gPART) can be estimated: Kglobal ¼
Cpart Kpart Cfree Kpart ¼ ¼ Caqu KDCM Cfree ½DCM þ Cfree KDCM ½DCM þ 1
(3)
where [DCM] represents the DCM concentration (g/mL). Kglobal is not a thermodynamical equilibrium constant but a partition coefficient, which is system-dependant. Equation (3) was previously used in several studies dealing with the effect of DCM on micropollutants sorption in environmental aquatic systems (Schrap and Opperhuizen, 1992; Amiri et al., 2005).
2.5.
Equilibration time
The kinetics of PAHs sorption was monitored in order to determine the equilibration time. For this, eight sludge samples were firstly mixed at 100 rpm at 35 C in centrifugation flasks in order to stabilize the matrix. Then, each sample was spiked at 5 mgPAH/gDM to start with the sorption kinetics. The aqueous PAHs concentration was measured after 1, 3, 4, 6, 8, 10, 14 and 24 h of incubation (extractions were carried out in duplicate).
2.6.
and
Dissolved and colloidal matter (DCM)
The three-compartment model
A micropollutant present in sludge can be located in one among three physical compartments (Fig. 1): the freely dissolved one (concentration Cfree, mg/mL), the sorbed-to-DCM one (concentration CDCM, mg/gDCM) and the sorbed-to-particles one (concentration Cpart, mg/gPART). At equilibrium, the three-compartment system can be described by the two following equations: Kpart ¼
K DCM
PAHs quantification in sludge
400 mL of sludge were centrifuged during 20 min at 12 000 g, at 35 C. The supernatant was filtered at 1.2 mm with glass microfiber filters (GF/C Whatman) to obtain the aqueous phase containing the free and sorbed-to-DCM micropollutants, to which 1% of formaldehyde at 37% was added for conservation. The particulate phase (pellet) was frozen at 20 C and freeze-dried. Extraction from aqueous phase was performed using SPE cartridges with 1 g C18 sorbent from Macherey-Nagel. Cartridges were washed and conditioned with 2 5 mL of methanol:toluene:acetonitrile (33:33:33), 2 5 mL of methanol and 2 5 mL of ultra pure water. 100 mL of sample were percolated through the extraction columns at a flow rate of 2 mL/min. After the sample percolated, the sorbent was dried in a stream of atmospheric air during 30 min. Adsorbates were eluted with 2 5 mL of methanol:toluene (50:50). Extracts were dried under nitrogen flow and dissolved into 1 mL of acetonitrile. Each sample was extracted in duplicate and corrected by the recovery performance determined on the same aqueous sample (same DCM concentration) spiked at approximately the same PAHs concentration. Extraction from solid phase was carried out as described by Trably et al. (2004). PAHs quantification in extracts from aqueous and solid phases was performed according to the method optimised by the same authors (Trably et al., 2004). The sum of PAHs concentration in aqueous and in solid phase was compared to the spiking concentration to ensure that no significant losses occurred during the experiment.
2.4.
Kpart
Cpart Cfree
(1)
Linearity verification
At constant DCM concentration, Kglobal is constant. As a consequence, the linearity of Cpart ¼ f(Caqu) is a linearity indicator of equilibria (1) and (2). Five samples at 0.7 gDCM/L and 4.9 gPART/L were therefore directly put in centrifugation flasks and equilibrated at 35 C on a mixing table (100 rpm). After 30 min of equilibration, they were spiked with the spiking mix at different concentrations ranging from 1 to 5 mgHAP/gDM and incubated again at 35 C for 4 h. The samples were then centrifuged at 35 C and both Cpart and Caqu were measured according to quantification section.
619
water research 44 (2010) 616–624
Kpart and KDCM can be extracted from equation (3) by assessing Kglobal with various DCM concentrations for each PAH. Variations of DCM concentrations from 0.6 to 4.5 gDCM/L were obtained by diluting sludge with its own supernatant previously separated and with water in different proportions. To ensure that this dilution method did not modify pH and that no particulate matter was released into DCM, the pH and the chemical oxygen demand (COD) were measured in 10 samples diluted by 1.2 (3.7 gDCM/L) to 200 (0.02 gDCM/L) fold factor. The COD was chosen here because it is a fast and little volume consuming method. The same procedure previously detailed was applied to equilibrate, spike, incubate and quantify Cpart and Caqu in the samples from 0.6 to 4.5 gDCM/L. Hence, Kglobal dependence to DCM concentration was measured. A non linear regression algorithm of Levenberg-Marquardt type (Marquardt, 1963) was used to minimize the sum of square errors and to estimate the two parameters of the Kglobal model: Kpart and KDCM. In some cases, the algorithm did not converge, due to the measurement noise and to high KDCM[DCM] values compared to 1. Indeed, when the ‘‘þ1’’ term becomes negligible in the denominator of equation (3), this leads to Equation (4): Kpart a Kglobal z ¼ KDCM ½DCM ½DCM
(4)
An infinity of couples (Kpart;KDCM) can be solution of this equation, defined by the ratio a¼Kpart/KDCM. a could be precisely determined by the algorithm. Afterwards, introduction of a into equation (3) allowed us to estimate Kpart for each Kglobal measurement and its corresponding [DCM]: Kpart ¼
Kglobal 1 Kglobal ½DCM a
(5)
The average of calculated Kpart values and the corresponding KDCM were used with the Levenberg-Marquardt algorithm in order to evaluate standard deviations and correlation coefficient. This procedure was applied to dibenzo(a,h)anthracene, benzo(g,h,i)perylene and indeno(1,2,3,c,d)pyrene.
3.
Results and discussion
3.1.
Hypotheses verification
From the initial DCM concentration in sludge (12 700 mgO2/L), the DCM concentration in diluted samples (from 1.2 to 200 fold factor) could be predicted, assuming that no matter was transferred from particles into aqueous phase. These predicted concentrations were compared to measured values. The ratio between them was very close to 1 (Fig. 2). Only for dilution factors higher than 20, a little fraction of particulate matter (15%) was found to be released, since the ratio reached 1.15. The highest dilution factor applied in sorption experiments was 7.5, which ensures that no particulate matter release occurred during these experiments. As a consequence, the DCM concentrations estimated thanks to the dilution factor were reliable. Moreover the pH was not modify by the dilution
1,3
6,4
1,2
6,2
1,1
6
pH
Determination of sorption equilibrium constants Measured COD (mgO2/L) / theoretical COD (mgO2/L)
2.7.
1
5,8
0,9
5,6
0,8 1
10
100
5,4 1000
Dilution factor Fig. 2 – Ratio between measured and theoretical aqueous COD (mg/L) (black circles) and pH (grey triangle) as a function of sludge dilution factor. The dotted frame indicates the range of dilution factors used for sorption experiments.
(Fig. 2), implying that the physicochemical properties of DCM can be assumed to be unchanged. Besides, the absence of particulate matter release in diluted samples indicates that sludge DCM does not behave like sorbable/desorbable matter. Indeed, matter is likely to be either dissolved/colloidal or particulate according to intrinsic parameters. This implies that PAHs sorption to particles via the sorption of DCM on which PAHs are sorbed does not occur, which validates the ‘‘three-compartment model / two equilibria’’ description (Fig. 1). The kinetics of fluorene and benzo(a)pyrene sorption in sludge are presented in Fig. 3. The sorption equilibrium state for this two PAHs was achieved within 1 h shaking, and it was the same for all other PAHs (data not shown). This value of 1 h for equilibrium achievement is consistent with literature (Dionisi et al., 2006) and suggests that during wastewater treatment processes, bioavailability is not limited by dynamic phenomena, as contact between xenobiotics and sludge lasts for several hours or days. Furthermore, a 4 h incubation time was selected for equilibrium experiments. The linear regressions performed to extract Kglobal from Cpart ¼ f(Caqu) produced regression coefficients R2 ranging from 0.85 to 0.95 (Table 1). This is consistent with the hypothesis of linearity assumed in the model. Moreover, Kglobal was plotted as a function of DCM concentration (Fig. 4) for each PAH. The obtained models fitted very well the experimental data with R2 ranging from 0.68 to 0.95, which again reinforced the model.
3.2.
Kglobal dependence to DCM concentration
According to the considered PAH, the measured partition coefficients varied of a ten fold factor in the set of tested conditions (Fig. 4). Indeed, a little shift of DCM concentration was shown to imply an important variation of Kglobal, above all at low concentrations. Moreover, when sludge is not considered as a three-compartment system and measured Kglobal is equated to an equilibrium constant, this so-called constant can be largely underestimated. For example, measured Kglobal
620
water research 44 (2010) 616–624
Fluorene
80 60 40 20 0
0
10
Benzo(a)pyrene
100
Aqueous fraction (%)
Aqueous fraction (%)
100
20
80 60 40 20 0
30
0
10
t (h)
20
30
t (h)
Fig. 3 – Percentage of initially spiked PAHs that remain in aqueous phase as a function of equilibration time. Since PAHs are spiked in liquid form, at time 0, the aqueous fraction is assumed to be 100%.
for benzo(b)fluoranthene at 0.6 gDCM/L was about 2000 mL/ gPART whereas the real equilibrium constant Kpart was estimated at 6920 mL/gPART (Fig. 4). In this case, assimilating Kglobal to the equilibrium constant would lead to an underestimation of 71%. Overall, this underestimation varied from 57 to 94% within the PAHs family and the more hydrophobic the PAH, the higher the error. Such an underestimation prevents from accurately predicting micropollutants sorption in sludge. All these results highlighted the necessity to consider the third phase to precisely assess the micropollutants fate. The obtained partition coefficients could not be strictly compared to literature data. Indeed, the studies published until now about PAHs sorption in sludge are very few, and authors neither take into account the three compartments nor measure DCM concentration. Moreover, our results underlined the strong Kglobal dependence to this concentration, which obliges to compare Kglobal values at equivalent DCM concentrations.
3.3.
Comparison between Kpart and KDCM
The logarithm of thermodynamical equilibrium constants Kpart and KDCM extracted from Kglobal dependence to DCM concentration are presented in Table 1. Standard deviation
was higher for highly hydrophobic compounds than for weakly hydrophobic ones, due to mathematical difficulties mentioned in the paragraph 2.7. The first way to reduce standard deviations would be to multiply measurements in order to decrease experimental uncertainty due to (i) sludge heterogeneity and (ii) quantification procedure. The second way would be to optimize measurements distribution on the DCM concentration scale. However, this option is limited for practical reasons. Indeed, the upper DCM concentration limit comes from the DCM concentration in raw sludge that cannot be passed over. Besides, to minimize DCM concentration, no supernatant volume should be added and a small sludge volume should be diluted with water. Nevertheless, particles concentration would concomitantly decrease and particles quantity needed for PAH quantification would constitute the lower DCM concentration limit. Nonetheless, standard deviation values for most PAHs were acceptable in comparison with the literature data on adsorption topic, since standard deviations, although very seldom estimated, are in this range (Kordel et al., 1997; Dionisi et al., 2006; Carballa et al., 2008). Again, Kpart comparison with previously published data was hardly feasible. However, at low particulate matter concentration (1 g/L), which indicates a probable very low DCM concentration, Dionisi et al. (2006) measured a Kglobal of
Table 1 – log Kow, partition coefficient Kglobal and coefficient R2 obtained from linear regression of Cpart [ f(Caqu) at 0.7 gDCM/L, and equilibrium constants extracted from Kglobal dependence to DCM concentration in the range 0.6–4.5 gDCM/L. Compound
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
log Kow
4.18 4.46 4.5 4.9 4.88 5.63 5.63 6.04 6.21 6.06 6.86 6.78 6.58
Partition coefficient
Equilibrium constants 2
Kglobal (mL/ gpart)
R
log Kpart
log KDCM
638 2142 1493 2357 3196 2712 1833 2381 1857 566 1701 1542 1035
0.89 0.92 0.92 0.85 0.95 0.92 0.92 0.89 0.90 0.89 0.90 0.89 0.87
3.21 0.23 3.56 0.24 3.45 0.17 4.01 0.44 3.71 0.37 3.78 0.21 3.86 0.40 3.84 0.38 4.27 0.76 4.01 0.55 4.34 1.57 4.11 1.10 4.03 0.83
3.24 0.37 3.35 0.36 3.11 0.29 3.46 0.61 3.20 0.62 3.33 0.32 3.47 0.56 3.49 0.52 4.02 0.85 3.83 0.65 4.12 1.71 4.00 1.20 3.81 0.97
621
water research 44 (2010) 616–624
Fluorene
Kglobal (mL/gPART)
1200
800
Kglobal (mL/gPART)
1
3
4
5
Fluoranthene
8000
2000
2000
1000 0
1
2
3
4
5
Chrysene
3000
0 0 5000
Kpart = 7 251 KDCM = 2 940 R2 = 0.854
3
4
5
Pyrene
0
1
2
3
4
5
Benzo(a)anthracene Kpart = 7 941 KDCM = 3 337 R2 = 0.876
4000 3000
2000
1000
1000 0
1
2
3
4
5
Kpart = 10 283 KDCM = 6 745 R2 = 0.940
3000
0 0 8000
5000 4000
1000 1
2
3
4
5
Benzo(b)fluoranthene
1
2
3
4
5
Dibenzo(a,h)anthracene
5000
3
4
5
0 0
3
4
5
Kpart = 18 604 KDCM = 10 429 R2 = 0.950
0
0
1
2
4
5
α = 1.278
4000
Kpart = 22 076 KDCM = 13 202 R2 = 0.863
3
Benzo(g,h,i)perylene
5000
α = 1.672
Kpart = 12 770 KDCM = 9 992 R2 = 0.895
3000 2000
2000
2
2
2000
4000
1
1
Benzo(k)fluoranthene
6000
2000
0
0
4000
6000
1000
0 8000
Kpart = 6 920 KDCM = 3 073 R2 = 0.877
3000
2000
0
0
5000
Kpart = 5 089 KDCM = 1 594 R2 = 0.680
4000
Benzo(a)pyrene Kglobal (mL/gPART)
2
2000
4000
0
1
3000
4000
5000
Kglobal (mL/gPART)
0 0
1000
4000
Kpart = 10 225 KDCM = 2 894 R2 = 0.823
6000
0
Kglobal (mL/gPART)
2
Kpart = 3 636 KDCM= 2 117 R2 = 0.833
2000
1000
0
Anthracene
3000
Kpart = 3 659 KDCM = 2 244 R2 = 0.902
2000
400 0
Phenanthrene
3000
Kpart = 1 606 KDCM = 1 717 R2 = 0.857
1000 1
2
3
DCM (g/L)
Indeno(1,2,3,c,d)pyrene
4
5
0
0
1
2
3
4
5
DCM (g/L)
α = 1.639
4000
Kpart = 10 599 KDCM = 6 465 R2 = 0.823
3000 2000 1000 0 0
1
2
3
4
5
DCM (g/L)
Fig. 4 – Measured (mark) and modelled (line) Kglobal (mL/gPART) as a function of DCM concentration (g/L). The a coefficient (for dibenzo(a,h)anthracene, benzo(g,h,i)perylene and indeno(1,2,3,c,d)pyrene), Kpart, KDCM and regression coefficients estimated thanks to the algorithm are presented on each graph. The corresponding logarithmic values and standard deviation for Kpart and KDCM are presented in Table 1.
9700 5000 mL/gPART for pyrene in an activated sludge. This value might be compared to our Kpart, as Kpart corresponds to Kglobal limit when DCM concentration tends to zero. We calculated a Kpart in the range 2200–12 000 mL/gPART for pyrene (Table 1), which is in good agreement with Dionisi et al.’s (2006) values. KDCM is easier to compare with literature data than Kpart, since several studies were performed in absence of particles. As an illustration, Holbrook et al. (2004) studied pyrene adsorption onto sludge DCM. Depending on sample origin and size
fractionation, the authors measured some KDCM values ranging from below 1000–80 000 mL/gOC (g of organic carbon). Since organic carbon usually represents approximately 25% of the total matter in sludge (Dignac et al., 2000), these values are widespread from below 250 to 20 000 mL/gDCM. This range is in accordance with our estimation of 380–6607 mL/gDCM (Table 1). The common and general correlation between the logarithm of equilibrium constants and hydrophobicity of PAHs (Karickhoff et al., 1979; Poerschmann and Kopinke, 2001) was observed for both Kpart and KDCM (Fig. 5). This demonstrates
622
water research 44 (2010) 616–624
35
7
30
y = 0.308x + 2.13
5
25 4 20 3
%
log Kpart, log KDCM
6
y = 0.348x + 1.61
15
2
10
1 4
4.5
5
5.5
6
6.5
7
log Kow
Fig. 5 – log Kpart (black rounds) and log KDCM (grey rounds) as a function of PAHs log Kow.
5 0 0
2
4
6
8
DCM (g/L)
that PAHs sorption to particles and to DCM may be driven by hydrophobic interactions. Moreover, the linear regressions revealed very similar slopes (0.31 and 0.35) but different origins (2.13 and 1.61), resulting in different values (Fig. 5). Indeed, the PAHs affinity for particles was found to be stronger than for DCM. This finding can be related to the compartments characteristics since equilibrium constants reflect the affinity of PAHs for the corresponding compartment. In this work, the compartments were differentiated according to their physical criteria, as defined in the 2.1. Section: the density criteria imposed by centrifugation, followed by the size criteria of filtration. Other physical properties such as hydrophobicity, surface charge and specific surface area differentiate particles and DCM. In addition of physical properties, Bougrier et al. (2008) demonstrated that biochemical composition (chemical oxygen demand, carbohydrates and proteins content) is different in particles and DCM. These properties are influenced by sludge history in terms of substrate (wastewater), involved microbiological population, and process operating parameters. For example, high sludge retention time in activated sludge process reduces particles specific surface area (Liss et al., 2002) and leads to more hydrophobic particles (Liao et al., 2001) with higher lipid content (Liss et al., 2002). Quantifying the influence of physical and biochemical sludge characteristics on the sorption equilibrium constants of PAHs would be very interesting.
3.4.
Assessment of PAHs compartment distribution
Kpart and KDCM determination and dependence to hydrophobicity provided the parameters needed to simulate PAHs distribution between the three sludge compartments. Aqueous and freely dissolved percentages were simulated in two theoretical and extreme cases: for a weakly hydrophobic PAH and for a strongly hydrophobic one (Fig. 6). The simulation demonstrated that freely dissolved fraction does not exceed 0.4% of the total concentration for a very hydrophobic compound and 2.5% for a weakly hydrophobic one. This fraction is slightly influenced by DCM concentration in both cases. On the contrary, aqueous fraction of both PAHs appeared to strongly depend on DCM concentration. This result is of
Fig. 6 – Percentage of free (plots) and aqueous (linked plots) concentrations modelled for a theoretical weakly hydrophobic PAH (log Kow [ 4.2, square) and for a theoretical very hydrophobic PAH (log Kow [ 6.8, triangle) as a function of dissolved and colloidal matter concentration, particulate concentration being fixed at 10 g/L. For this simulation, Kpart and KDCM of both PAHs were estimated thanks to the relationship presented in Fig. 5. The reversion point is located at 6.01 gDCM/L, which corresponds to 25.5% of aqueous PAHs. Over this value, aqueous fraction is slightly higher for the very hydrophobic PAH than for the weakly hydrophobic one.
great concern since aqueous fraction is usually equated to bioavailable fraction, as previously mentioned. It means that low DCM concentrations favour higher percentage in aqueous phase for weakly hydrophobic PAHs than for very hydrophobic ones, whereas reasonable high concentrations discriminate less PAHs. The extrapolation of the model to extremely concentrated conditions ([DCM] > 6 g/L) shows that the aqueous concentration of a very hydrophobic PAH could equal and even exceed a weakly hydrophobic one (Fig. 6). In the usual bioavailability hypothesis ‘‘aqueous is bioavailable’’, this would mean that a very hydrophobic PAH could be more bioavailable than a weakly hydrophobic one. This case is really atypical, since micropollutants are usually assumed to be less bioavailable when they are more hydrophobic (Mackay and Fraser, 2000). It is an important insight derived from the three-compartment approach. The reversion point would be located at [DCM] ¼ 6.0 g/L for the studied activated sludge, as particulate matter concentration was fixed at 10 g/L for the simulation. However, it has to be reminded that the model was established for DCM concentrations up to 4.5 g/L. Moreover, DCM at the reversion point would represent 37.5% of the total matter, which is much higher than common values (Bougrier et al., 2008). As a conclusion, the latter case is little probable in real systems, but it could be interesting to create it in laboratory investigations. Apart from the descriptive issue previously discussed, such simulations constitute a new tool for micropollutants physical fate assessment. They will allow to predict the fate of PAHs throughout the separation steps encountered in wastewater
water research 44 (2010) 616–624
treatment plants. In the example of a settler, the flux of sorbedto-particles PAHs in solid effluent (for agricultural disposal) can be quantified as well as the flux of sorbed-to-DCM and dissolved PAHs carried by clarified water (discharged into surface water). Indeed, Katsoyiannis and Samara (2007) observed an influence of DCM concentration on micropollutants fate, without quantifying this effect. The threecompartment approach could provide the missing data to quantitatively take this interaction between micropollutants and DCM into account. As a result, the methodology presented here constitutes a real improvement in the assessment of micropollutants fate throughout treatment plants and of their environmental discharges.
3.5.
Towards bioavailability
Quantifying PAHs distribution in sludge samples has an important outcome concerning its consequences on bioavailability topic. Indeed, it may be useful to verify the usual hypothesis of identity between aqueous and bioavailable fractions (Artola-Garicano et al., 2003; Dionisi et al., 2006). This assumption may be verified by measuring biodegradation at different aqueous concentrations. Initial biodegradation rate in batch experiments could be an appropriate indicator for biodegradation. The assumption would be confirmed if biodegradation rate correlates with aqueous concentration. Guidelines for such an experiment are provided by the present work: it suggests that different aqueous concentrations of one PAH might be achieved thanks to different DCM concentrations. On the other hand, investigation on the link between freely dissolved and bioavailable fractions of one PAH appears more difficult, as freely dissolved concentration varied little in the presented conditions. It means that this link might be investigated through either the comparison between different compounds in one matrix or the comparison of one compound in various matrixes for which this compound has different affinities. These guidelines are of great importance for the modeling of micropollutants biological fate in sludge, since until now, models have been built on hypotheses which have never been clearly demonstrated.
4.
Conclusions
The three-compartment methodology proposed to investigate micropollutants sorption in sludge was validated. The results underlined the fact that a reliable tool for micropollutants behaviour prediction in sludge should integrate a threecompartment model since DCM concentration influences micropollutants distribution in a significant way. Indeed, the usual biphasic approach was shown to lead to an important underestimation of equilibrium constants. Despite the lack of published data about PAH sorption in sludge, the results could be confronted to a few studies. The confrontation showed the consistency of equilibrium constant values and of dependency to PAH hydrophobicity. Besides, PAH affinities for particles and for DCM could be compared. Affinity for particles was found to be stronger. This
623
may be due to the differences between particles and DCM in term of physical and chemical characteristics. Further investigation is necessary to include the influence of both sludge and micropollutant characteristics in the three-compartment model. This research, being presently undergone by the authors, will enable to predict PAHs behaviour in any sludge, as a function of its characteristics. The future predictive model would allow to estimate the PAH distribution in any sludge in order (i) to determine fluxes throughout physical separation steps and (ii) to confront PAHs distribution between compartments to their biodegradation, which might help to elucidate bioavailability issue.
Acknowledgements This study was funded by the Institut National de Recherche Agronomique and the Ministe`re de l’Enseignement Supe´rieur et de la Recherche du Gouvernement Franc¸ais, France, and the authors gratefully acknowledge them.
references
Amiri, F., Bornick, H., Worch, E., 2005. Sorption of phenols onto sandy aquifer material: the effect of dissolved organic matter (DOM). Water Research 39, 933–941. Arias-Estevez, M., Fernandez-Gandara, D., Garcia-Falcon, M.S., Garcia-Rio, L., Mejuto, J.C., Simal-Gandara, J., 2007. Sorption of PAHs to colloid dispersions of humic substances in water. Bulletin of Environmental Contamination and Toxicology 79, 251–254. 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 and Technology 37, 116–122. Bougrier, C., Delgene`s, J.P., Carre`re, H., 2008. Effects of thermal treatments on five different waste activated sludge samples solubilisation, physical properties and anaerobic digestion. Chemical Engineering Journal 139, 236–244. Byrns, G., 2001. The fate of xenobiotic organic compounds in wastewater treatment plants. Water Research 35, 2523–2533. Carballa, M., Fink, G., Omil, F., Lema, J.M., Ternes, T., 2008. Determination of the solid-water distribution coefficient (K-d) for pharmaceuticals, estrogens and musk fragrances in digested sludge. Water Research 42, 287–295. Carballa, M., Omil, F., Alder, A.C., Lema, J.M., 2006. Comparison between the conventional anaerobic digestion of sewage sludge and its combination with a chemical or thermal pretreatment concerning the removal of pharmaceuticals and personal care products. Water Science and Technology 53, 109–117. Chiou, C.T., Malcolm, R.L., Brinton, T.I., Kile, D.E., 1986. Water solubility enhancement of some organic pollutants and pesticides by dissolved humic and fulvic acids. Environmental Science and Technology 20, 502–508. 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, 4185–4194. Dionisi, D., Bertin, L., Bornoroni, L., Capodicasa, S., Papini, M.P., Fava, F., 2006. Removal of organic xenobiotics in activated sludges under aerobic conditions and anaerobic digestion of
624
water research 44 (2010) 616–624
the adsorbed species. Journal of Chemical Technology and Biotechnology 81, 1496–1505. 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 and Technology 54, 119–128. Hickman, Z.A., Swindell, A.L., Allan, I.J., Rhodes, A.H., Hare, R., Semple, K.T., Reid, B.J., 2008. Assessing biodegradation potential of PAHs in complex multi-contaminant matrices. Environmental Pollution 156, 1041–1045. Holbrook, R.D., Love, N.G., Novak, J.T., 2004. Investigation of sorption behavior between pyrene and colloidal organic carbon from activated sludge processes. Environmental Science and Technology 38, 4987–4994. Hung, N.V., Tateda, M., Ike, M., Fujita, M., Tsunoi, S., Tanaka, M., 2004. Sorption of biodegradation end products of nonylphenol polyethoxylates onto activated sludge. Journal of Environmental Sciences-China 16, 564–569. Ivashechkin, P., Corvini, P.F.X., Dohmann, M., 2004. Behaviour of endocrine disrupting chemicals during the treatment of municipal sewage sludge. Water Science and Technology 50, 133–140. Karickhoff, S.W., Brown, D.S., Scott, T.A., 1979. Sorption of hydrophobic pollutants on natural sediments. Water Research 13, 241–248. Katsoyiannis, A., Samara, C., 2007. The fate of dissolved organic carbon (DOC) in the wastewater treatment process and its importance in the removal of wastewater contaminants. Environmental Science and Pollution Research 14, 284–292. Katsoyiannis, A., Zouboulis, A., Samara, C., 2006. Persistent organic pollutants (POPs) in the conventional activated sludge treatment process: model predictions against experimental values. Chemosphere 65, 1634–1641. Kordel, W., Hennecke, D., Franke, C., 1997. Determination of the adsorption-coefficients of organic substances on sewage sludges. Chemosphere 35, 107–119. Krauss, M., Wilcke, W., 2005. Persistent organic pollutants in soil density fractions: distribution and sorption strength. Chemosphere 59, 1507–1515. Langford, K.H., Scrimshaw, M.D., Birkett, J.W., Lester, J.N., 2005. The partitioning of alkylphenolic surfactants and polybrominated diphenyl ether flame retardants in activated sludge batch tests. Chemosphere 61, 1221–1230. Lee, C., Kuo, L., 1999. Quantification of the dissolved organic matter effect on the sorption of hydrophobic organic pollutant: application of an overall mechanistic sorption model. Chemosphere 38, 807–821. Liao, B.Q., Allen, D.G., Droppo, I.G., Leppard, G.G., Liss, S.N., 2001. Surface properties of sludge and their role in bioflocculation and settleability. Water Research 35, 339–350. Liss, S.N., Liao, B.Q., Droppo, I.G., Allen, D.G., Leppard, G.G., 2002. Effect of solids retention time on floc structure. Water Science and Technology 46, 431–438. Mackay, D., Fraser, A., 2000. Bioaccumulation of persistent organic chemicals: mechanisms and models. Environmental Pollution 110, 375–391. Marquardt, D., 1963. An algorithm for least-squares estimation of nonlinear parameters. SIAM Journal of Applied Mathematics 11, 431–441. Oleszczuk, P., 2007. Investigation of potentially bioavailable and sequestrated forms of polycyclic aromatic
hydrocarbons during sewage sludge composting. Chemosphere 70, 288–297. Oleszczuk, P., 2008. Tenax-TA extraction as predictor for free available content of polycyclic aromatic hydrocarbons (PAHs) in composted sewage sludges. Journal of Environmental Monitoring 10, 883–888. Perminova, I.V., Grechishcheva, N.Y., Kovalevskii, D.V., Kudryavtsev, A.V., Petrosyan, V.S., Matorin, D.N., 2001. Quantification and prediction of the detoxifying properties of humic substances related to their chemical binding to polycyclic aromatic hydrocarbons. Environmental Science and Technology 35, 486–490. Poerschmann, J., Kopinke, F.D., 2001. Sorption of very hydrophobic organic compounds (VHOCs) on dissolved humic organic matter (DOM). 2. Measurement of sorption and application of a Flory–Huggins concept to interpret the data. Environmental Science and Technology 35, 1142–1148. Reid, B.J., Stokes, J.D., Jones, K.C., Semple, K.T., 2000. Nonexhaustive cyclodextrin-based extraction technique for the evaluation of PAH bioavailability. Environmental Science and Technology 34, 3174–3179. Schrap, S.M., Opperhuizen, A., 1992. On the contradictions between experimental sorption data and the sorption partitioning model. Chemosphere 24, 1259–1282. 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 and Technology 38, 228A–231A. Semple, K.T., Doick, K.J., Wick, L.Y., Harms, H., 2007. Microbial interactions with organic contaminants in soil: definitions, processes and measurement. Environmental Pollution 150, 166–176. Tang, J.X., Liste, H.H., Alexander, M., 2002. Chemical assays of availability to earthworms of polycyclic aromatic hydrocarbons in soil. Chemosphere 48, 35–42. Ternes, T.A., Herrmann, N., Bonerz, M., Knacker, T., Siegrist, H., Joss, A., 2004. A rapid method to measure the solid-water distribution coefficient (K-d) for pharmaceuticals and musk fragrances in sewage sludge. Water Research 38, 4075–4084. 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 in sludge samples. International Journal of Environmental Analytical Chemistry 84, 995–1008. Urase, T., Kikuta, T., 2005. Separate estimation of adsorption and degradation of pharmaceutical substances and estrogens in the activated sludge process. Water Research 39, 1289–1300. van Straalen, N.M., Donker, M.H., Vijver, M.G., van Gestel, C.A.M., 2005. Bioavailability of contaminants estimated from uptake rates into soil invertebrates. Environmental Pollution 136, 409–417. Vinken, R., Hollrigl-Rosta, A., Schmidt, B., Schaffer, A., Corvini, P.F.X., 2004. Bioavailability of a nonylphenol in dependence on the association to dissolved humic substances. Water Science and Technology 50, 277–283. Yamamoto, H., Liljestrand, H.M., Shimizu, Y., Morita, M., 2003. Effects of physical–chemical characteristics on the sorption of selected endocrine disruptors by dissolved organic matter surrogates. Environmental Science and Technology 37, 2646–2657. Zhou, J.L., Liu, R., Wilding, A., Hibberd, A., 2007. Sorption of selected endocrine disrupting chemicals to different aquatic colloids. Environmental Science and Technology 41, 206–213.
water research 44 (2010) 625–637
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Removal of micropollutants and reduction of biological activity in a full scale reclamation plant using ozonation and activated carbon filtration J. Reungoat a,*, M. Macova b, B.I. Escher b, S. Carswell b,c, J.F. Mueller b, J. Keller a a
The University of Queensland, Advanced Water Management Centre (AWMC), Qld 4072, Australia The University of Queensland, National Research Centre for Environmental Toxicology (EnTox), QLD 4108, Australia c Queensland Health Forensic and Scientific Services, Organics Laboratory, QLD 4108, Australia b
article info
abstract
Article history:
Pharmaceutical compounds are found in secondary treated effluents up to mg L-1 levels and
Received 12 May 2009
therefore discharged into surface waters. Since the long term effects of these compounds
Received in revised form
on the environment and human health are, to date, largely unknown, implementation of
14 September 2009
advanced treatment of wastewaters is envisaged to reduce their discharge. This is of
Accepted 21 September 2009
particular relevance where surface waters are used as drinking water sources and when
Available online 1 October 2009
considering indirect potable reuse. This study aimed at assessing the removal of organic micropollutants and the concurrent reduction of their biological activity in a full scale
Keywords:
reclamation plant treating secondary effluent. The treatment consists of 6 stages: deni-
Water reuse
trification, pre-ozonation, coagulation/flocculation/dissolved air flotation and filtration
Oxidation
(DAFF), main ozonation, activated carbon filtration and final ozonation for disinfection. For
Adsorption
that purpose, representative 24-hour composite samples were collected after each stage.
Pesticides
The occurrence of 85 compounds was monitored by LC/MS-MS. A battery of 6 bioassays
Pharmaceuticals
was also used as a complementary tool to evaluate non-specific toxicity and 5 specific toxic
Bioassays
modes of action. Results show that, among the 54 micropollutants quantified in the influent water, 50 were removed to below their limit of quantification representing more than 90% of concentration reduction. Biological activity was reduced, depending on the specific response that was assessed, from a minimum of 62% (AhR response) to more than 99% (estrogenicity). The key processes responsible for the plant’s performances were the coagulation/flocculation/DAFF, main ozonation and activated carbon filtration. The effect of these 3 processes varied from one compound or bioassay to another but their combination was almost totally responsible for the overall observed reduction. Bioassays yielded complementary information, e.g. estrogenic compounds were not detected in the secondary effluent by chemical analysis, but the samples had an estrogenic effect. The main ozonation formed oxidation by-products of the organic micropollutants but decreased the level of non-specific toxicity and other specific toxic modes of action, demonstrating that the mixture of oxidation by-products was less potent than the mixture of the parent compounds for the considered effects. ª 2009 Elsevier Ltd. All rights reserved.
* Corresponding author. Tel.: þ61 734466251; fax: þ61 733654726. E-mail address:
[email protected] (J. Reungoat). 0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2009.09.048
626
1.
water research 44 (2010) 625–637
Introduction
In the last decade, numerous studies around the world have demonstrated the presence of pharmaceutical compounds in domestic wastewaters. These compounds are removed to different degrees by the commonly used biological activated sludge processes. While some compounds (e.g. ibuprofen, paracetamol) are effectively removed, others (e.g. carbamazepine, diclofenac) are barely affected by the biological treatment (Onesios et al., 2009). As a result, some pharmaceutical compounds are released into the environment, contaminating surface waters. This situation is of concern as these compounds have been designed to affect biological systems and long term exposure effects on the biological elements in the environment are still largely unknown. Human health is also at risk when treated wastewater is discharged to water bodies that are used as drinking water sources or considered for indirect potable reuse. Therefore, advanced treatment of wastewaters has to be envisaged to reduce the load of these compounds in discharged effluents. Several technologies have proven to be effective in removing pharmaceutical compounds from water: activated carbon adsorption (Nowotny et al., 2007; Snyder et al., 2007; Ternes et al., 2002; Westerhoff et al., 2005; Yu et al., 2008), ozonation and advanced oxidation processes (Esplugas et al., 2007; Huber et al., 2003; Huber et al., 2005; Kim et al., 2008; Nakada et al., 2007; Ternes et al., 2003; Zwiener and Frimmel, 2000) and membrane filtration (Kimura et al., 2004; Snyder et al., 2007; Yoon et al., 2007). Activated carbon adsorption and ozonation are considered to be economically feasible for tertiary treatment of wastewaters by Joss et al. (2008). Nevertheless, none of these processes can remove all the compounds of concern and to date, no extensive study has been carried out on the efficiency of the sequential combination of these two processes for the removal of micropollutants from secondary treated wastewater. Chemical analysis of micropollutants in water has gone through major developments in recent years and today’s analytical techniques can measure some compounds at concentrations as low as a few ng L1. Chemical analysis is powerful to measure the concentration of targeted compounds but does not cover the whole range of chemicals that might be present in the effluent of a wastewater treatment plant. Analytical techniques usually focus on pharmaceutically active parent compounds whereas a variable fraction is excreted from the human body as metabolites. Moreover, biological treatment of wastewater may form yet unidentified compounds through biodegradation reactions; e.g. 17b-estradiol is degraded to estrone (Ternes et al., 1999). Subsequent chemical treatments such as ozonation can also form by-products. Chemical analysis of a limited suite of compounds does not allow assessment of the potential biological adverse effects of the wastewater as the cumulative effect of the mixture of chemicals that may be present cannot be easily integrated. Bioassays have been utilised as complementary monitoring tools for the assessment of possible biological effects of chemicals that are present in a particular water sample (Escher et al., 2008b; Muller et al., 2007). These bioanalytical tools are designed to quantify non-specific
toxicity or particular toxic modes of action (e.g. estrogenicity, genotoxicity, phytotoxicity) induced by a sample on a biological organism or a biological process. To date, bioassays have not been widely used in the evaluation of water treatment processes and have rarely been accompanied with chemical analysis (Macova et al., 2010). The present study assessed the removal of selected micropollutants (pharmaceuticals and pesticides) and the decrease of biological activity along a full scale water reclamation plant using ozonation and activated carbon adsorption to treat a secondary effluent. This aimed at identifying the key steps in the treatment process as well as the key operating parameters. Results of chemical analysis and bioassays were compared to determine whether the use of both techniques was redundant or allowed a deeper understanding of the performances of individual treatment processes. The performance and relevance of the treatment train is further discussed in the context of indirect potable reuse.
2.
Materials and methods
2.1.
South Caboolture Water Reclamation Plant
The South Caboolture Water Reclamation Plant was designed to reduce riverine pollution from the 40,000 population equivalent wastewater treatment plant and to provide recycled water to industry and community consumers. Whilst the plant provides water for non-potable applications, it has been designed to meet drinking water standards. The treatment process detailed in Fig. 1 incorporates biological denitrification, pre-ozonation, coagulation/flocculation/dissolved air flotation-sand filtration (DAFF), main ozonation, biological activated carbon filtration and final ozonation for disinfection. The activated carbon was renewed in March 2008 after 9 years of service, its adsorption capacity was assumed not to be exhausted at the time sampling took place, 4 months later. van Leeuwen et al. (2003) published more details on the process and its performances.
2.2.
Sample collection
Four sets of samples were collected over winter 2008 under dry weather conditions including three during week days and one during the weekend (11-07-08, 22-07-08, 27-07-08 and 0608-08). Water temperature across the plant was 22 2 C and pH was 7.0 0.5. Samples were collected at 7 sampling points along the treatment train, labelled S1 to S7 on Fig. 1, in order to evaluate the performance of individual treatment steps. As the flow rate in the reclamation plant is constant (8 megalitres per day), representative samples were collected as time proportional 24-h composites. For S1 and S7, refrigerated auto-samplers collected 200 mL every 15 minutes; for the remaining sampling locations, a continuous flow pump collected samples at a flow rate of 7 mL min-1. At each point, samples were collected into a glass bottle pre-washed with MilliQ water and HPLC grade acetone. The samples were protected from light and refrigerated during collection. For micropollutant analysis, 1 L was transferred into solvent
water research 44 (2010) 625–637
627
Fig. 1 – South Caboolture Water Reclamation Plant process (adapted from van Leeuwen et al. (2003)). S1 to S7: sampling locations. HRT: hydraulic retention time. SRT: sludge retention time.
washed amber glass bottles and preserved with sodium thiosulfate. For the bioassays, 2 L were transferred into solvent washed amber glass bottles and hydrochloric acid (36%) was added to a final concentration of 5 mM for preservation. For dissolved organic carbon (DOC) measurements, 100 mL were filtered through a 0.45 mm PTFE membrane and collected in plastic (HDPE) bottles rinsed with the sample beforehand. All samples were transported on ice and stored frozen or at 4 C.
2.3.
Chemical analyses
2.3.1.
Dissolved organic carbon
The dissolved organic carbon (DOC) was measured as nonpurgable organic carbon (NPOC) with an Analytik Jena multi N/C 3100 instrument at the Advanced Water Management Centre (AWMC). For each sample, 2d3 replicates were measured, giving a relative standard deviation of less than 3%.
2.3.2.
basis of quantity of usage of the particular compounds, their potential toxicity and their resistance to degradation. Their limit of quantification (LOQ) was 0.01 mg L-1 in most cases. Concentrations were calculated using an internal calibration method. Results were not corrected for recovery efficiency of the extraction method. Recoveries of single compounds are assumed to be consistent along the treatment as the matrix is similar therefore; calculation of the removal efficiency is not affected. Indeed, Gros et al. (2009) determined the recoveries of 73 pharmaceutical compounds by SPE in raw wastewater and treated effluent and their results show that the recovery for a given compound generally vary by less than 20% from one matrix to the other despite the fact that these matrices are very different. The standard deviation of the method is 20% (Table SI 1 in the supporting information SI 1). When the reported outlet concentration was below the LOQ of the compound, removal efficiency was calculated as a minimum value using the LOQ as outlet concentration.
Micropollutants
Micropollutant analysis was carried out by Queensland Health Forensic and Scientific Services (QHFSS, accredited by the National Association of Testing Authorities, Australia, and ISO 9001 certified). The method consisted of solid phase extraction (SPE), concentration and quantification by liquid chromatography coupled with tandem mass spectrometry (LC/MSdMS). This method, described in details in the supporting information SI 1, allowed the quantification of 85 compounds (listed in Table SI 3 in the supporting information SI 2) selected on the
2.3.3.
Bioassays
Six bioassays, discussed in details in Macova et al. (2010), were applied to the samples collected in the study. Baseline toxicity was determined with the non-specific Vibrio fischeri bioluminescence inhibition test. Five additional specific toxic modes of action were also evaluated: estrogenic activity (E-SCREEN assay), arylhydrocarbon receptor response (CAFLUX assay), neurotoxicity (acetylcholinesterase inhibition assay), phytotoxicity (PSII inhibition I-PAM assay) and genotoxicity (umuC
628
water research 44 (2010) 625–637
assay). Water samples were extracted by SPE using Oasis HLB cartridges. Full dose response curves were determined for a serial dilution of the extract for each bioassay. Results were expressed as toxic equivalent concentrations (TEQ) except for the umuC assay. The TEQ represents the concentration of a given reference compound that would be required to produce the same effect as the mixture of the various compounds in the sample. When the outlet TEQ was below the LOQ of the bioassay, removal efficiency was calculated as a minimum value using the LOQ as outlet TEQ. In the umuC assay, the response is determined as an induction ration (IR), an IR 1.5 is considered genotoxic. For genotoxic samples, ECIR1.5 corresponds to how many times the sample must be concentrated or diluted to elicit an IR of 1.5. Results are expressed as 1/ECIR1.5 therefore a higher number represents a higher genotoxic effect.
3. 3.1.
Results and discussion Micropollutants
In the reclamation plant’s influent, 54 of the 85 targeted compounds had a median concentration above their LOQ confirming that conventional activated sludge treatment does not completely remove these micropollutants from wastewater (Fig. 2). The concentrations ranged from 0.01 to 2.10 mg L-1 with the exception of gabapentin, which was consistently found at higher concentrations ranging from 5.60–6.50 mg L-1 (data provided in the supporting information in Table SI 3). The factor between the minimum and the maximum concentrations measured for each individual compound was generally close to or lower than 2, with a maximum of 3.6 observed for iopromide. No clear pattern
Fig. 2 – Number of compounds quantified and dissolved organic carbon (DOC) after indicated stage along the treatment train. Bars represent the number of compounds (left y-axis) with a median concentration above the LOQ (4 samples). Dots represent DOC on two different sampling days (right y-axis). Denit: denitrification; Pre-O3: preozonation; DAFF: dissolved air flotation-sand filtration; Main O3: main ozonation; AC: activated carbon; Fin O3: final ozonation.
could be distinguished between the different sampling days. The increase or decrease of single compound concentrations from one day to another appeared to be random, even when comparing the sample collected during the weekend to samples collected during week days. Twenty-five compounds had an influent median concentration above 0.10 mg L–1, their fate is further detailed hereafter. The removal efficiency of these compounds was determined in each treatment step except when the concentration before treatment was lower than ten times the LOQ and below LOQ after treatment. This criterion was used to allow the determination of removals up to 90% in any case. The DOC was measured for two sets of samples (22-07-08 and 06-08-08), and varied from 14.2 to 19.7 mg L-1 in the influent water. The following paragraphs discuss the performance of each individual stage of the treatment by comparing concentrations of the selected compounds in the inlet and outlet water for the considered process.
3.1.1.
Denitrification
The number of compounds above LOQ was unchanged after denitrification and the concentrations of the 25 selected compounds decreased by less than 20% (with the exception of atenolol, 38%). To our knowledge, the removal of micropollutants by denitrifying bacteria has not been specifically studied to date. In the present case, methanol is added to the water prior to the denitrification reactor and the bacteria use it as the primary electron donor which may prevent or reduce the degradation of micropollutants. The observed increase in DOC after the denitrification was probably due to residual methanol.
3.1.2.
Pre-ozonation
The number of compounds quantified was stable through the pre-ozonation process (Fig. 2) and the concentration of the 25 selected compounds decreased by less than 30%. The ozone dose applied in the pre-ozonation step is approximately 2 mg L-1 and the DOC was approximately 20 mg L-1 corresponding to an ozone dose of 0.1 mgO3 mg-1 DOC. In such conditions, ozone is rapidly consumed by the organic matter (Buffle et al., 2006a,b) therefore micropollutants oxidation is limited. Ozonation has been proved to be very effective for oxidising various micropollutants in secondary treated wastewaters (Hollender et al., 2009; Huber et al., 2005; Snyder et al., 2006; Ternes et al., 2003; Wert et al., 2009) but with higher ozone doses of at least 0.25 to 0.50 mgO3 mg-1 DOC. Snyder et al. (2006) also investigated a full scale wastewater treatment plant using pre-ozonation with a low ozone dose (not specified) prior to biological activated carbon filtration, and similarly, observed removal of micropollutants inferior to 30%.
3.1.3.
Coagulation/flocculation/DAFF
In the coagulation/flocculation/DAFF process, the negatively charged colloids are neutralised by the addition of a chemical agent (aluminium sulphate) which causes them to aggregate in floccs (flocculation) which are subsequently removed by dissolved air flotation and sand filtration. The main purpose of this stage is to decrease the DOC, which was achieved effectively as demonstrated by the 40–50% DOC reduction (Fig. 2). This reduction of the DOC is important to achieve maximal
water research 44 (2010) 625–637
performance of the main ozonation process as discussed above. After the DAFF, 53 compounds were quantified and the concentrations of the 25 selected compounds decreased by less than 20% except for atenolol (42%), caffeine (29%), gabapentin (44%), gemfibrozil (32%) and roxithromycin (37%). It is also interesting to note that paracetamol concentration increased by 44% in the process, which could not be explained. No literature reporting investigations of the removal of micropollutants from treated wastewaters by coagulationflocculation could be identified. However, several studies on lab-scale and full scale drinking water treatment systems have reported removals lower than 50% for several pharmaceuticals and pesticides by such solids removal processes (Adams et al., 2002; Ternes et al., 2002; Thuy et al., 2008; Vieno et al., 2006; Westerhoff et al., 2005). Suarez et al. (2009) also showed similar limited removal in hospital wastewaters. Nevertheless, Suarez et al. (2009) and Westerhoff et al. (2005), showed that removals of up to 80% can be achieved for highly hydrophobic compounds (i.e logKow > 6) suggesting that the micropollutants removal during coagulation-flocculation occurs via hydrophobic interactions with neutral particles. Most of the pharmaceuticals and pesticides measured in this study are hydrophilic or of moderate hydrophobicity with logKow < 4 (Table SI 3 in supporting information SI 2); therefore little removal can be expected in the coagulation/flocculation/ DAFF stage. Among the exceptions observed, the removal of gemfibrozil can be explained by its more hydrophobic nature (logKow ¼ 4.77, EPI SUITE 4.0) but other compounds are very hydrophilic and no explanation for their removal could be advanced.
3.1.4.
Main ozonation
The main ozonation stage (approximately 0.5 mgO3 mg-1 DOC) decreased the concentration of 26 compounds below LOQ (Fig. 2). Among the 25 selected compounds, 9 showed
629
a reduction of more than 90% and 13 others were reduced by more than 70% while iopromide and gabapentin were reduced by 55% (Fig. 3). Removal of naproxen was not determined because its concentration was lower than ten times its LOQ before treatment and below LOQ after. The higher removal efficiency obtained by the main ozonation stage compared with the pre-ozonation is due to the higher ozone dosage relative to the DOC content. These results are in agreement with previous findings (Buffle et al., 2006b; Hollender et al., 2009; Huber et al., 2005; Snyder et al., 2006; Ternes et al., 2003; Wert et al., 2009). The reaction of organic compounds with molecular ozone is selective and only certain groups of compounds react rapidly, e.g. aliphatic molecules with double bonds, deprotonated amines and aromatics with an activating group. Most of the pharmaceuticals and pesticides possess one of these moieties and would be expected to react with molecular ozone with high second order rate constants (kO3). The results show that the removal of micropollutants increased with increasing value of kO3 but even compounds weakly reactive with ozone (i.e. iopromide and MCPA; Table 1) were oxidised. Oxidation in ozonation processes can also occur via the attack of hydroxyl radicals formed by ozone decomposition in water. Hydroxyl radicals are highly reactive with almost any organic molecule; the second order rate constants of oxidation by hydroxyl radicals are typically in the 109 M-1 s-1 range (Table 1). Therefore, the observed removals of iopromide and MCPA suggest a substantial exposure to hydroxyl radicals. Indeed, Buffle et al. (2006a,b) showed that the ozonation of wastewaters leads to high exposure to hydroxyl radicals; in the first instants of the reaction, the hydroxyl radical production is equivalent to, or greater than what can be obtained with advanced oxidation processes. To our knowledge, gabapentin oxidation by ozone has not been reported in the literature and no reaction rate data could be found. Gabapentin possesses both carboxylic and amine
Fig. 3 – Median removal of selected compounds (median of influent concentration > 0.10 mg L-1) by the main ozonation stage and the combination of the main ozonation and the activated carbon (AC) filtration stages. Error bars represent minimum and maximum removal, no error bar means that the compound was below LOQ after treatment; therefore removal was calculated as a minimum using the LOQ. CS4: concentration before main ozonation; CS5: concentration after main ozonation; CS6: concentration after activated carbon filtration.
630
water research 44 (2010) 625–637
Table 1 – Removal of selected pharmaceuticals in the main ozonation stage and their first order reaction rate constants with ozone (kO3) and hydroxyl radicals (kHO ). The temperature (T ) and pH given are the experimental conditions used for the determination of the reaction rates.
Compound
Removal
kO3 (M1 s1)
pH
T ( C)
kHO (109.M1 s1)
Iopromide MCPA Atenolol
55% 77% 77%
<0.08 47.7 (1.7 0.4) 103
7 2 7
20 20 20–22
Roxithromycin
84%
7 104 6.3 104 (2.0 0.6) 103
7 7 7
20 20 20–22
3.0 104–1.0 106 2.7 105 ~2.5 106 5.5 105 ~106 (1.84 0.15) 104 4.11 106 k ~3 105 (7.81 1.31) 104
2 7 7 7 7 6 7 7 25
20 20 20 20 20 25 25 20
Metoprolol
86%
Ranitidine Trimethoprim Sulphamethoxazole
>91% >93% >93%
Diclofenac
>94%
Paracetamol Carbamazepine
>96% >98%
pH
T ( C)
Ref.
3.3 0.6 6.6 8.0 0.5 7.05 0.27
7 9 7 7
25 20 20–22 room
a
5.4 0.3 7.3 0.2 8.39 0.06
7 7 7
25 20–22 room
6.5 0.2 5.5 0.7
7
25
7
25
b c d a e c d f
a e a
7.5 1.5 2.2 0.017 8.8 1.2
e
7 5.5 7
25 25 25
g h a i
2.05 0.14
5
room
j
a Huber et al. (2003). b Benitez et al. (2004). c Benner et al. (2008). d Song et al. (2008). e Dodd et al. (2006). f Rivas et al. (2009). g Vogna et al. (2004b). h Andreozzi et al. (2003). i Andreozzi et al. (2002). j Vogna et al. (2004a). k Calculated from Andreozzi et al (2003).
groups, their pKa are 3.89 and 9.56 respectively (Zour et al., 1992). Therefore gabapentin is present as a zwitterion under usual wastewater pH conditions. Protonated amines and deprotonated carboxylic acids are typically not highly reactive with ozone (Hoigne´ and Bader, 1983) and gabapentin oxidation can be expected to be controlled by hydroxyl radical reactions. As a primary amine, gabapentin has an expected second order rate constant for the reaction with ozone of about 50 M-1 s-1 which is significantly higher than for iopromide. However, the similar behaviour of the two substances shows that their decrease is controlled by hydroxyl radical reactions. The DOC decreased by less than 10% in the process, showing that ozonation does not lead to substantial mineralization of dissolved organic matter.
3.1.5.
Activated carbon filtration
Activated carbon filtration further removed the micropollutants and only 2 compounds were quantified above their LOQ downstream: roxithromycin (0.01 mg L-1) and gabapentin (0.70 mg L-1). The removal efficiency of the compounds having a median concentration of at least ten times their LOQ prior to activated carbon filtration was calculated: oxazepam, tramadol and venlafaxine were removed by more than 90% and gabapentin was removed by 53%. Adsorption onto activated carbon also removed 20–30% of the DOC showing a less effective adsorption of organic matter than of micropollutants. Previous studies (Ormad et al., 2008; Snyder et al., 2007; Ternes et al., 2002; Westerhoff et al., 2005) demonstrated
that powdered and granular activated carbon can efficiently remove micropollutants from natural water sources used for drinking water. To date, the removal of micropollutants from wastewaters by activated carbon has not been extensively studied; Snyder et al. (2007) found that a water reuse facility using granular activated carbon without regular replacement/ regeneration provided little removal. In the present work, the activated carbon had been replaced about 4 months before the collection of the samples so it is assumed that its adsorption capacity was not yet exhausted. Adsorption onto activated carbon is driven mainly by hydrophobic interactions and Westerhoff et al. (2005) observed that the trend in the removal efficiency of micropollutants by activated carbon can be predicted from logKow values. Gabapentin is zwitterionic at neutral pH as shown above and much more hydrophilic (logKow ¼ -1.37) compared to oxazepam, tramadol and venlafaxine (logKow ¼ 2.32, 3.01 and 3.28 respectively). This explains the difference observed in the removal efficiencies of these compounds. It can be expected that, with time, the adsorption capacity of the activated carbon filter will decrease and eventually be exhausted while biological activity will develop in the filter and contribute to biodegradation of the by-products formed by ozonation (Simpson, 2008).
3.1.6.
Final ozonation
In the final effluent, after final ozonation, 4 compounds had a median concentration above the LOQ: gabapentin (0.45 mg L-1), roxithromycin (0.01 mg L-1), DEET (0.03 mg L-1) and caffeine
water research 44 (2010) 625–637
(0.02 mg L-1). The median concentrations of caffeine and DEET are very close to the LOQ and are below the LOQ prior to the final ozonation stage; therefore one can assume that their quantification is due to analytical variations rather than an actual concentration variation. It is difficult to evaluate the efficiency of the final ozonation in the removal of micropollutants because they all had concentrations equal or below the LOQ before the treatment except gabapentin. The median removal of gabapentin in the final ozonation was only 20%, which is low compared to the main ozonation, and is due to the lower ozone dose employed (approximately 0.3 mgO3 mg-1 DOC). The DOC was not removed in this treatment step.
3.1.7.
Overall treatment
The full treatment decreased the concentration of 50 of the 54 compounds quantified in the influent water to levels below LOQ (Fig. 2). Concomitantly, DOC was also reduced by 55–60% in the effluent compared to the influent water. Overall, among the 25 selected compounds, 11 were removed by more than 95% and 11 by more than 89%. The median removal of gabapentin was 86% and the removals of naproxen and iopromide were not calculated because their concentration was lower than 10 times their LOQ in the influent and below their LOQ in the effluent. The 4 remaining compounds were gabapentin (0.45 mg L-1), roxithromycin (0.01 mg L-1), DEET (0.03 mg L-1) and caffeine (0.02 mg L-1). The first 3 stages of the treatment train (i.e. denitrification, pre-ozonation and coagulation/flocculation/DAFF) did not decrease the number of micropollutants quantified in the water (Fig. 2). After these 3 stages, the concentrations of the 25 compounds that had an influent median concentration of at least 0.10 mg L-1 were generally still greater than 50% compared to the influent concentration (Fig. 4). The main ozonation and
631
the activated carbon adsorption played a key role in the treatment bringing the concentration of 26 and 25 compounds below LOQ respectively (Fig. 2). The main ozonation generally decreased the micropollutants to less than 20% of their influent concentration and activated carbon filtration further removed the compounds to levels below LOQ except for gabapentin and roxythromycin (Fig. 4). The combined effects of the main ozonation and the activated carbon filtration processes decreased the concentration of 10 of the 25 selected micropollutants by more than 95% and by more than 89% for 12 of the 15 remaining compounds compared to their concentration prior to the main ozonation (Fig. 3). Gabapentin concentration was reduced by 79%. These results show that ozonation followed by activated carbon filtration is a very effective combination of processes to remove micropollutants from secondary treated wastewater. The key steps in the removal of the DOC were the DAFF and the activated carbon adsorption. Although the DAFF reduced the concentration of micropollutants by less than 30%, it also played a key role indirectly by reducing the DOC which enhanced the performances of the main ozonation.
3.2.
Bioanalytical results
The influent biological activity was higher than the blank (MilliQ water) in all the bioassays (Table 2). The effect of the treatment train on each bioassay is discussed individually hereafter.
3.2.1.
Baseline toxicity
The Vibrio fischeri bioluminescence inhibition test is a nonspecific bacterial toxicity test widely recognised in the field of ecotoxicology as the standard assay for acute cytotoxicity. The
Fig. 4 – Median relative concentrations of selected compounds (median of influent concentration > 0.10 g L-1) after the dissolved air flotation-sand filtration (DAFF, S4), the main ozonation (S5) and the activated carbon (S6) stages (error bars represent maximum and minimum values). C is the concentration after the specified treatment step and the reference concentration, C0, is the concentration in the influent water of the reclamation plant.
632
water research 44 (2010) 625–637
Table 2 – Maximum, median and minimum biological activity in the influent and effluent of the reclamation plant and overall maximum, median and minimum decrease observed. Bioassay
Baseline toxicity Estrogenicity AhR response Genotoxicity Neurotoxicity Phytotoxicity
Result expression
a
Blank
1
Baseline toxicity EqC (TEQ, mg L ) Estradiol EqC (EEQ, ng L1) TCDD EqC (TCDDEQ, ng L1) 1/ECIR1.5 Parathion EqC (PTEQ, mg L1) Diuron EqC (DEQ, mg L1)
0.21 <0.02 0.08 <0.01 <0.3 <0.01
Influent
Effluent
Decrease (%)
Max
Med
Min
Max
Med
Min
Max
Med
Min
2.9 7.6 0.98 0.32 3.9 0.22
2.1 5.8 0.82 0.19 3.1 0.10
2.0 5.7 0.59 0.13 2.8 0.05
0.72
0.52 <0.06 0.31 <0.01 <0.3 0.03
0.31
84
78
0.26
69
<0.01
>91
62 >93 >90 75
67 >99 46 82 57 50
0.36 0.04 1.2 0.05
a EqC ¼ equivalent concentration.
assay reflects the general ‘‘energy status’’ of the bacteria and can indicate the toxic potency of a broad spectrum of compounds with different modes of action. Denitrification and pre-ozonation did have a slight stimulatory effect on the baseline toxicity whereas the following treatment steps were able to substantially decrease baseline toxicity (Fig. 5). The TEQ was reduced to 62, 37 and 21% of the influent water level after the coagulation/flocculation/DAFF, the main ozonation and the activated carbon filtration respectively. As is discussed in more details in Macova et al. (2010), an almost linear correlation exists between DOC level and TEQ. The slight increase in TEQ after denitrification cannot be related to the residual methanol added before this process, which is still partially present after denitrification as is discussed above, but is likely to be related to some non-volatile organic chemicals. The 52% decrease of TEQ in the DAFF stage is accompanied by a 40–50% reduction in DOC, whereas micropollutants concentrations were generally reduced by less than 30% in the meantime. Although the SPE that is performed prior to toxicity testing should be able to remove a substantial fraction of the DOC and will remove all of the residual methanol, some DOC, most likely smaller breakdown products that have similar
physicochemical properties and similar molecular weight, may still be present. Consequently, the baseline toxicity is, at least partially, induced by compounds having a fate differing from the fate of measured micropollutants in the treatment process and that chemical analysis alone cannot effectively assess the effect of the treatment processes on water quality. The main ozonation reduced the TEQ by 31% whereas the concentration decay observed for most of the micropollutants was higher than 70%. The DOC was not affected. It is known that some organic compounds are poorly reactive with ozone and the results of the micropollutant analysis showed that some compounds were only partially degraded in the main ozonation step (i.e iopromide and gabapentin). Moreover, ozonation does not typically lead to complete mineralization but to the formation of by-products. The oxidation products of ozonation are in general more polar and more hydrophilic molecules than the parent compounds but the modification is not drastic. Therefore the oxidation products of ozonation will still have a considerable effect in a non-specific assay like the bioluminescence inhibition test with Vibrio fischeri, where the toxicity is generally directly related to the hydrophobicity of the mixture components (Escher et al., 2008a).
Fig. 5 – Relative response of the bioassays and relative dissolved organic carbon (DOC) after indicated stage along the treatment train compared to the influent (DOC0 [ influent DOC). Bars are the median of 4 values for bioassays and error bar represent maximum and minimum values. Dots are the average of 2 values for DOC and error bars represent maximum and minimum values. Denit: denitrification; Pre-O3: pre-ozonation; DAFF: dissolved air flotation-sand filtration; Main O3: main ozonation; AC: activated carbon; Fin O3: final ozonation. Baseline toxicity: Vibrio fischeri bioluminescence inhibition test; estrogenicity: E-SCREEN assay; arylhydrocarbon (AhR) receptor response: CAFLUX assay; genotoxicity: umuC assay; neurotoxicity: acetylcholinesterase inhibition assay; phytotoxicity: PSII inhibition I-PAM assay.
water research 44 (2010) 625–637
Activated carbon filtration reduced the baseline toxicity by 50% and the DOC by 30 to 35%. Activated carbon effectively adsorbs the more hydrophobic compounds, which is again consistent with the general trend discussed above that the more hydrophobic compounds have a higher toxic activity than the more hydrophilic ones. Based on this fact, identification of the compounds exhibiting a high toxic activity could start with the identification of the more hydrophobic compounds. The final ozonation did not further reduce the baseline toxicity compared to after activated carbon filtration. The effluent TEQ was approximately 80% lower than the influent TEQ (Fig. 5) and only 2.5 times higher than the blank (Table 2). This, indicates that the residual toxicity is of no concern, unless the residual organic chemicals and organic matter inducing this effect were of very specific potency. This latter question was tested with a series of specific endpoints that respond to environmentally relevant modes of toxic action.
3.2.2.
Estrogenic activity
The E-SCREEN assay specifically responds to natural hormones and other compounds that can mimic the activity of the female sex hormone estradiol. The estrogenic activity of the samples is expressed as an estradiol equivalent concentration (EEQ). The median influent EEQ was 5.8 ng L-1; higher than levels previously reported in South East Queensland. Most of the effluents from 12 activated sludge wastewater treatment plants tested by Leusch et al. (2006) had EEQs below 4 ng L-1 and sometimes below 1 ng L-1. Hormones and endocrine disrupting compounds (EDCs) were not investigated in this study but an earlier sampling campaign of the secondary treated wastewater showed that the concentrations of measured estrogenic compounds (17b-estradiol, 17a-ethynylestradiol, estrone, estriol, bisphenol A, nonylphenol) were all below the LOQ of 1 ng L-1 (Table SI 4 in supporting information SI 3). This demonstrates the relevance of using bioassays as complementary tools to chemical analysis for the assessment of water quality and process performances. Denitrification did not affect the estrogenicity (Fig. 5). Preozonation with an ozone dose of approximately 0.10 mgO3 mg-1 DOC reduced the EEQ by 34% compared to the influent. This is higher than the removal previously observed by Snyder et al. (2006) who measured the EEQ reduction induced by various ozone doses in treated wastewater with a DOC of 6.38 mg L-1. They found that an ozone dose of 2.1 mg L-1 (0.33 mgO3 mg1 DOC) only removed 18% of the EEQ but, with ozone doses of 3.6 mg L-1 (0.56 mgO3 mg1 DOC) and above, 90% or more removal could be achieved. In a recent study on a full scale ozonation in a Swiss sewage treatment plant (STP), the dose-dependency of removal of micropollutants yielded similar results (Escher et al., 2009). While most endpoints showed a clear dose-dependency of reduction of effects, the reduction of estrogenicity was already large at low ozone doses and depended more on the EEQ than on the ozone dose. When estrogenicity was already below a certain level, which was very close to the detection limit, the quantification of further reduction became difficult and prone to large uncertainty. For the remaining samples, ozone doses of 1.6– 5.3 mg L-1 in the presence of 4.2–6.0 mg L-1 DOC lead to more
633
than 90% reduction of estrogenicity. This is consistent with laboratory experiments that demonstrated that almost all first-generation transformation products of estrogenic chemicals had severely decreased estrogenic potency (Lee et al., 2008). Thus ozonation can be considered as a fairly selective oxidation, where even low doses selectively target one of the most environmentally relevant modes of toxic action, namely estrogenicity. After the coagulation/flocculation/DAFF stage the EEQ increased drastically by a median factor of 3.3 compared to the level prior to treatment. At this treatment step, the concentration of DOC is greatly reduced (by 40–50%), and there is a likelihood that the estrogenic chemicals that were bound to DOC were released during this treatment step. We have previously observed with another estrogenicity assay that DOC appears to reduce the bioavailability of estrogens (B. Escher, unpublished results). Estrogenic chemicals are typically relatively hydrophobic and bind well to DOC (Neale et al., 2008). In general DOC is not bioavailable in bioassays (the discussion on the small breakdown products above is an exception of this general paradigm) and micropollutants sorbed to DOC would not be bioavailable either. A large fraction of the matrix and also the DOC is supposed to be removed by SPE but, given the color of the extracts, it is possible that a substantial fraction of larger DOC is co-extracted. In addition, for the E-SCREEN test, it was demonstrated that the presence of serum proteins modulates the free and bioavailable concentration of estrogenic chemicals (Heringa et al., 2004). This effect was also hydrophobicity dependent and was much more pronounced for the more hydrophobic octylphenol than for the less hydrophobic estradiol. Protein binding is generally less important than binding to DOC or lipids, therefore while the effect on bioavailability was not very large for estradiol in the study of Heringa et al. (2004); it might well be relevant under the conditions of the present study. This hypothesis needs to be evaluated in the future by exploring the correlation between size distribution of naturally occurring DOC and effect on bioavailability, estrogenicity and toxicity. The main ozonation reduced the EEQ by a median value of 92 and 95% compared to the level of the reclamation plant’s influent and to the level before treatment respectively whereas DOC was not affected. It can be concluded that the mixture of by-products formed by the oxidation of the estrogenic compounds by ozone and hydroxyl radicals have a much lower estrogenic activity than the mixture of parent compounds, which is consistent with expectations as discussed above and in Lee et al. (2008). Activated carbon filtration was able to efficiently adsorb residual estrogenic compounds and further reduced the EEQ by another 95% to levels below the detection limit of 0.02 ng L-1 and the final effluent concentration was below the quantification limit of 0.06 ng L-1. The overall treatment efficiency for the removal of estrogenic activity was greater than 99%. This is in good agreement with observations on a full scale ozonation in a Swiss STP (Escher et al., 2009). As discussed above the analytically determined concentrations of (xeno)estrogens were below the quantification limit, therefore for this endpoint the very sensitive bioassay poses a great advantage despite the observed limitations due to matrix effects.
634
3.2.3.
water research 44 (2010) 625–637
Ah-receptor response
The CAFLUX assay targets dioxins and dioxin-like compounds such as polychlorinated biphenyls (PCBs) but can also respond to other chemicals such as polycyclic aromatic hydrocarbons (PAHs) (Macova et al., 2010). The results of the test are expressed as 2,3,7,8-tetrachlorodibenzo-p-dioxin equivalent concentration (TCDDEQ). The median TCDDEQ of the influent water was 0.82 ng L-1 and there was no significant variation along the first three steps of the treatment process; i.e denitrification, pre-ozonation and coagulation/flocculation/DAFF (Fig. 5). The main ozonation removed about 50% of the TCDDEQ but subsequent activated carbon filtration and final ozonation did not show further important removal and the median TCDDEQ of the final effluent was approximately 3.9 times higher than the blank (Table 2). Two sets of samples were submitted to a sulphuric acid silica gel clean up procedure that aims at removing organic chemicals except those that are not oxidised such as polychlorinated dibenzodioxins, -furans and PCBs. The samples were then tested again with the CAFLUX assay to evaluate the contribution of these very persistent chemicals (i.e dioxins, furans and dioxin-like PCBs). Results showed that after clean up the TCDDEQ was not significantly different from the blank (values ranged from 0.09 to 0.11 ng L-1). This shows that the effect induced by the samples without sulphuric acid silica gel clean up is not due to the presence of dioxins, furans or dioxin-like PCBs but was caused by other chemicals. Since none of these groups of chemicals were quantified by chemical analysis in this study, no comparison between chemical and biological analysis is possible.
3.2.4.
Genotoxicity
The umuC assay responds specifically to genotoxic compounds that cause DNA damages. To detect genotoxic effects caused by metabolites, the test is also performed in presence of a rat liver extract that can transform indirect genotoxicants to metabolites that are DNA damaging compounds. The median influent 1/ECIR1.5 were 0.19 and 0.060 in the absence and presence of the rat liver extract respectively, showing that the sample was less genotoxic after metabolisation. This is what one would commonly expect, an exception would be PAHs that are activated by metabolism. Denitrification and pre-ozonation did not have a substantial influence on genotoxicity (Fig. 5). The coagulation/flocculation/DAFF stage decreased 1/ECIR1.5 by 59% compared to the influent. The main ozonation drastically reduced the genotoxicity, 1/ECIR1.5 was reduced by 80 and 93% compared to the DAFF effluent and to the influent of the plant respectively. After activated carbon filtration as well as in the final effluent, 1/ECIR1.5 was below the LOQ of the bioassay (Table 2). In every case, the genotoxicity of the metabolised sample was lower than the non-metabolised sample, indicating that the type of chemicals inducing the genotoxic effect did not change over the treatment.
3.2.5.
reduce the PTEQ whereas DAFF decreased it by 31% compared to influent (Fig. 5). Unlike the other bioassays, the effect of the main ozonation on PTEQ was not significant but activated carbon filtration reduced it drastically to level below the quantification limit of the bioassay (0.30 mg L-1) which represents more than an 80% and 90% decrease compared to the main ozonation effluent and the plant influent water respectively. This observation is consistent with theoretical expectation, as it is known that compounds like diazinon and chlorpyrifos, which often constitute a large fraction of the acetylcholinesterase inhibitors, are not well oxidised by ozone. In contrast, these compounds are fairly hydrophobic (logKow ¼ 3.96 and 4.66 respectively), therefore sorption to activated carbon can be expected. A similar removal pattern has been observed for acetylcholinesterase inhibitors in the above-mentioned Swiss STP: none of the single removal steps (biological treatment, ozonation, sand filtration) had a high removal efficiency but all steps taken together produced a satisfactory overall removal (Escher et al., 2009).
3.2.6.
Phytotoxicity
The I-PAM assay is sensitive to herbicides that directly inhibit photosynthesis; the results are reported as a diuron equivalent concentration (DEQ). The DEQ of the influent water ranged from 0.05 to 0.22 mg L-1 with a median value of 0.10 mg L-1 (Table 2). Diuron concentrations were measured by chemical analysis; it was reported in every sample of the influent water from 0.02 to 0.04 mg L-1, suggesting that its contribution to the effect observed was limited. Among the other herbicides quantified, only simazine is also a photosystem II inhibitor with a relative potency of 0.15 (Muller et al., 2008). Simazine concentrations in the influent ranged from 0.05 to 0.19 mg L-1. These two compounds considered together accounted for 17–93% of the measured DEQ demonstrating the interest of this bioassay to take into account non-measured compounds. The DEQ increased by factors of 2.2 and 3.5 after denitrification and preozonation respectively but variation from one day to another was large therefore it is difficult to draw a conclusion (Fig. 5). This increase was accompanied with a slight increase in baseline toxicity and could therefore be caused by baseline toxicants interfering with the measurement of the photosynthesis yield (Macova et al., 2010). The coagulation/filtration/DAFF stage reduced DEQ by 67% and 88% compared to the plant’s influent water and to the pre-ozonated water respectively. After the main ozonation the DEQ was not further significantly reduced; Diuron’s concentrations were equal to or below the LOQ of 0.01 mg L-1 and simazine was removed by approximately 50%, their contribution accounting for 16–38% of the observed DEQ. Subsequent activated carbon filtration and final ozonation did not further affect the DEQ but diuron and simazine were removed below their LOQ. The overall treatment achieved 75% median decrease of DEQ, the effluent median DEQ was 0.03 mg L-1 (Table 2).
Neurotoxicity
Neurotoxicity is measured by the inhibition of the enzyme acetylcholinesterase (AChE). Organophosphate and carbamate pesticides specifically bind to this enzyme and the results are expressed as parathion equivalent concentration (PTEQ). The median PTEQ in the secondary treated wastewater was 3.1 mg L-1; denitrification and pre-ozonation did not
3.2.7.
Overall treatment
The biological activity of the effluent was lower compared to the influent and close or equal to the blank level showing that the treatment process could effectively decrease the biological adverse effects observed with the bioassays; from 62% for the AhR response to more than 99% for estrogenicity (Table 2). The
water research 44 (2010) 625–637
key treatment steps responsible for the decrease of biological activity are the DAFF stage, the main ozonation and the activated carbon adsorption (Fig. 5).
3.3.
Indirect potable reuse considerations
Micropollutants reported concentrations were compared to the guidelines values for indirect potable reuse given in the Australian Guidelines for Water Recycling: Augmentation of Drinking Water Supplies (Table SI 3 in supporting information SI 2). The reported concentrations of the measured compounds were found to be below the guideline values in the influent water of the reclamation plant before any treatment. After removal by the advanced treatment process, concentrations were several orders of magnitude below the guideline values. For information purpose, median equivalent concentrations obtained with the bioassays were compared to the corresponding reference compound’s guideline value when available. Note however, that the effect caused by a mixture cannot be compared directly to a guideline value of a single compound. Moreover, the bioassays used here are acute tests and no conclusions can be drawn about chronic effects. Nevertheless such a comparison gives an impression on the expected hazard of the mixture but must be communicated with caution to a lay audience. For estrogenicity, neurotoxicity and phytotoxicity the reference compounds were estradiol, parathion and diuron and the guidelines values were 175 ng L-1, 10 mg L-1 and 30 mg L-1 respectively. Similarly to individual compounds concentrations, the bioassays equivalent concentrations were already below the guidelines values in the water entering the reclamation plant. Final effluent median equivalent concentrations were also several orders of magnitude below the corresponding guideline values, i.e. more than 2900, 33 and 428 fold for estrogenicity, neurotoxicity and phytotoxicity respectively. For the parameters considered, the water quality complies with the requirements of the Australian Guidelines for Water Recycling: Augmentation of Drinking Water Supplies. This suggests that such a treatment train could be considered as an alternative to the combination of microfiltration and reverse osmosis for indirect potable reuse schemes. It has the advantage of not producing a waste stream and would be certainly less energy intensive. Nevertheless, before this process can be recommended for indirect potable reuse, additional consideration needs to be given to the overall risk management strategies of the treatment train. Moreover, the removal of pathogens such as viruses and bacteria has to be assessed as well as the potential to form disinfection by-products due to the remaining DOC levels.
4.
Conclusions
The assessment of a tertiary treatment train regarding the removal of micropollutants and decrease of biological activity leads to the following conclusions: 1) The treatment train reduced the concentration of the 54 micropollutants quantified in the secondary treated wastewater as well as the biological activity observed in
2)
3)
4)
5)
635
bioassays. Overall concentration reductions were typically higher than 90% and most of the compounds were removed to levels lower than 0.01 mg L1. The observed effects decreased by 62% to more than 90% depending on the assay and levels in the final effluent were close to or equivalent to the blank’s. The effect of individual processes varied from one compound and bioassay to another but the combination of the coagulation/flocculation/DAFF stage, the main ozonation and the activated carbon filtration was responsible for the major part of the observed reduction. The use of a battery of bioassays as complementary tools to chemical analysis yielded valuable additional information on the water quality and the process efficiency. The results showed the limitations of chemical analysis to assess potential biological adverse effects and the ability of bioassays to take into account the presence of non-measured compounds, formed transformation products and/or mixture effects. Concurrent measurement of DOC, micropollutants and biological activity showed that the DOC level can influence the effect observed in the bioassays. Baseline toxicity was shown to correlate well with the DOC levels. It is supposed that non-targeted organic matter is co-extracted with the targeted organic micropollutants in the SPE purification step. More understanding of the nature and quantity of these co-extracted compounds is needed. Ozone dose relative to DOC content is a key parameter for the ozonation performance. A low ozone dose of 0.1 mgO3 mg-1 DOC did not affect the concentration of micropollutants or biological activity whereas a dose of 0.5 mgO3 mg-1 DOC was able to remove most of the micropollutants by more than 70% and substantially decrease the biological activity. Degradation products are a major concern in ozonation processes. Here, the results from the battery of bioassays show that the main ozonation leads to lower baseline and specific toxic effects. It can be concluded that the mixture of degradation products formed have an overall less harmful potential than the mixture of parent compounds.
Acknowledgements This work was co-funded by the Urban Water Security Research Alliance under the Enhanced Treatment Project, the CRC Water Quality and Treatment Project No. 2.0.2.4.1.1 – Dissolved Organic Carbon Removal by Biological Treatment and by EnTox. The National Research Centre for Environmental Toxicology (EnTox) is a joint venture of the University of Queensland and Queensland Health Forensic and Scientific Services. The authors acknowledge the following institution and individuals who contributed to this study: Moreton Bay Water for access to the South Caboolture Water Reclamation Plant; Ray McSweeny and Paul McDonnell (Moretom Bay Water) for their help during sampling; Chris Pipe-Martin (Ecowise) for providing information on the South Caboolture Water Reclamation Plant; Geoff Eaglesham, Steve Carter and Mary Hodge (Queensland Health Forensic and Scientific Services) for the analysis of micropollutants and discussions on the analytical method; Wolfgang Gernjak, Michael G. Lawrence, Christoph Ort and Maria Jose Farre Olalla
636
water research 44 (2010) 625–637
(Advanced Water Management Centre) for fruitful discussions and proof reading of the manuscript.
Appendix. Supplementary data Supplementary data associated with this article can be found in the online version, at doi:10.1016/j.watres.2009.09.048.
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–ASCE 128 (3), 253–260. 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 (5), 993–1004. 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 (11), 2869–2877. Benitez, F.J., Acero, J.L., Real, F.J., Roman, S., 2004. Oxidation of MCPA and 2,4-D by UV radiation, ozone, and the combinations UV/H2O2 and O-3/H2O2. Journal of Environmental Science and Health Part B-Pesticides Food Contaminants and Agricultural Wastes 39 (3), 393–409. Benner, J., Salhi, E., Ternes, T., von Gunten, U., 2008. Ozonation of reverse osmosis concentrate: Kinetics and efficiency of beta blocker oxidation. Water Research 42 (12), 3003–3012. Buffle, M.O., Schumacher, J., Meylan, S., Jekel, M., von Gunten, U., 2006a. Ozonation and advanced oxidation of wastewater: effect of O-3 dose, pH, DOM and HO center dot-scavengers on ozone decomposition and HO center dot generation. Ozone: Science and Engineering 28 (4), 247–259. Buffle, M.O., Schumacher, J., Salhi, E., Jekel, M., von Gunten, U., 2006b. Measurement of the initial phase of ozone decomposition in water and wastewater by means of a continuous quench-flow system: application to disinfection and pharmaceutical oxidation. Water Research 40 (9), 1884–1894. 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), 1969–1977. Escher, B.I., Bramaz, N., Mueller, J.F., Quayle, P., Rutishauser, S., Vermeirssen, E.L.M., 2008a. Toxic equivalent concentrations (TEQs) for baseline toxicity and specific modes of action as a tool to improve interpretation of ecotoxicity testing of environmental samples. Journal of Environmental Monitoring 10 (5), 612–621. Escher, B.I., Bramaz, N., Ort, C., 2009. Monitoring the treatment efficiency of a full scale ozonation on a sewage treatment plant with a mode-of-action based test battery. Journal of Environmental Monitoring 11 (10), 1836–1846. Escher, B.I., Bramaz, N., Quayle, P., Rutishauser, S., Vermeirssen, E.L.M., 2008b. Monitoring of the ecotoxicological hazard potential by polar organic micropollutants in sewage treatment plants and surface waters using a mode-of-action based test battery. Journal of Environmental Monitoring 10 (5), 622–631. 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 (3), 631–642. Gros, M., Petrovic, M., Barcelo, D., 2009. Tracing pharmaceutical residues of different therapeutic classes in environmental waters by using liquid chromatography/quadrupole-linear ion trap mass spectrometry and automated library searching. Analytical Chemistry 81 (3), 898–912. Heringa, M.B., Schreurs, R., Busser, F., Van Der Saag, P.T., Van Der Burg, B., Hermens, J.L.M., 2004. Toward more useful in vitro toxicity data with measured free concentrations. Environmental Science and Technology 38 (23), 6263–6270. Hoigne´, J., Bader, H., 1983. Rate constants of reactions of ozone with organic and inorganic compounds in water–II: dissociating organic compounds. Water Research 17 (2), 185–194. Hollender, J., Zimmermann, S.G., Koepke, S., Krauss, M., McArdell, C.S., Ort, C., Singer, H., von Gunten, U., Siegrist, H., 2009. Elimination of organic micropollutants in a municipal wastewater treatment plant upgraded with a full-scale postOzonation followed by sand filtration. Environmental Science and Technology 43 (20), 7862–7869. Huber, M.M., Canonica, S., Park, G.Y., von Gunten, U., 2003. Oxidation of pharmaceuticals during ozonation and advanced oxidation processes. Environmental Science and Technology 37 (5), 1016–1024. Huber, M.M., Gobel, A., Joss, A., Hermann, N., Loffler, D., McArdell, C.S., Ried, A., Siegrist, H., Ternes, T.A., von Gunten, U., 2005. Oxidation of pharmaceuticals during ozonation of municipal wastewater effluents: a pilot study. Environmental Science and Technology 39 (11), 4290–4299. Joss, A., Siegrist, H., Ternes, T.A., 2008. Are we about to upgrade wastewater treatment for removing organic micropollutants? Environmental Science and Technology 57 (2), 251–255. Kim, I.H., Tanaka, H., Iwasaki, T., Takubo, T., Morioka, T., Kato, Y., 2008. Classification of the degradability of 30 pharmaceuticals in water with ozone, UV and H2O2. Water Science and Technology 57 (2), 195–200. Kimura, K., Toshima, S., Amy, G., Watanabe, Y., 2004. Rejection of neutral endocrine disrupting compounds (EDCs) and pharmaceutical active compounds (PhACs) by RO membranes. Journal of Membrane Science 245 (1–2), 71–78. Lee, Y., Escher, B.I., Von Gunten, U., 2008. Efficient removal of estrogenic activity during oxidative treatment of waters containing steroid estrogens. Environmental Science and Technology 42 (17), 6333–6339. Leusch, F.D.L., Chapman, H.F., van den Heuvel, M.R., Tan, B.L.L., Gooneratne, S.R., Tremblay, L.A., 2006. Bioassay-derived androgenic and estrogenic activity in municipal sewage in Australia and New Zealand. Ecotoxicology and Environmental Safety 65 (3), 403–411. Macova, M., Escher, B.I., Reungoat, J., Carswell, S., Lee Chue, K., Keller, J., Mueller, J.F., 2010. Monitoring the biological activity of micropollutants during enhanced wastewater treatment with ozonation and activated carbon filtration. Water Research 44 (2), 477–492. Muller, R., Schreiber, U., Escher, B.I., Quayle, P., Nash, S.M.B., Mueller, J.F., 2008. Rapid exposure assessment of PSII herbicides in surface water using a novel chlorophyll a fluorescence imaging assay. Science of the Total Environment 401 (1–3), 51–59. Muller, R., Tang, J.Y.M., Thierb, R., Mueller, J.F., 2007. Combining passive sampling and toxicity testing for evaluation of mixtures of polar organic chemicals in sewage treatment plant effluent. Journal of Environmental Monitoring 9 (1), 104–109. 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
water research 44 (2010) 625–637
endocrine-disrupting chemicals (EDCs) during sand filtration and ozonation at a municipal sewage treatment plant. Water Research 41 (19), 4373–4382. Neale, P.A., Escher, B.I., Schafer, A.I., 2008. Quantification of solute–solute interactions using negligible-depletion solidphase microextraction: measuring the affinity of estradiol to bulk organic matter. Environmental Science and Technology 42 (8), 2886–2892. Nowotny, N., Epp, B., vonSonntag, C., Fahlenkamp, H., 2007. Quantification and modeling of the elimination behavior of ecologically problematic wastewater micropollutants by adsorption on powdered and granulated activated carbon. Environmental Science and Technology 41 (6), 2050–2055. 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 (4), 441–466. Ormad, M.P., Miguel, N., Claver, A., Matesanz, J.M., Ovelleiro, J.L., 2008. Pesticides removal in the process of drinking water production. Chemosphere 71 (1), 97–106. Rivas, J., Gimeno, O., Encinas, A., Beltra´n, F., 2009. Ozonation of the pharmaceutical compound ranitidine: Reactivity and kinetic aspects. Chemosphere 76 (5), 651–656. Simpson, D.R., 2008. Biofilm processes in biologically active carbon water purification. Water Research 42 (12), 2839–2848. Snyder, S.A., Adham, S., Redding, A.M., Cannon, F.S., DeCarolis, J., Oppenheimer, J., Wert, E.C., Yoon, Y., 2007. Role of membranes and activated carbon in the removal of endocrine disruptors and pharmaceuticals. Desalination 202 (1–3), 156–181. Snyder, S.A., Wert, E.C., Rexing, D.J., Zegers, R.E., Drury, D.D., 2006. Ozone Oxidation of Endocrine Disruptors and Pharmaceuticals in Surface Water and Wastewater. Ozone: Science and Engineering 28 (6), 445–460. Song, W.H., Cooper, W.J., Mezyk, S.P., Greaves, J., Peake, B.M., 2008. Free radical destruction of beta-blockers in aqueous solution. Environmental Science & Technology 42 (4), 1256–1261. Suarez, S., Lerna, J.M., Omil, F., 2009. Pre-treatment of hospital wastewater by coagulation-flocculation and flotation. Bioresource Technology 100 (7), 2138–2146. Ternes, T.A., Kreckel, P., Mueller, J., 1999. Behaviour and occurrence of estrogens in municipal sewage treatment plants – II. Aerobic batch experiments with activated sludge. Science of the Total Environment 225 (1–2), 91–99. Ternes, T.A., Meisenheimer, M., McDowell, D., Sacher, F., Brauch, H.J., Gulde, B.H., Preuss, G., Wilme, U., Seibert, N.Z., 2002. Removal of
637
pharmaceuticals during drinking water treatment. Environmental Science and Technology 36 (17), 3855–3863. Ternes, T.A., Stuber, J., Herrmann, N., McDowell, D., Ried, A., Kampmann, M., Teiser, B., 2003. Ozonation: a tool for removal of pharmaceuticals, contrast media and musk fragrances from wastewater? Water Research 37 (8), 1976–1982. Thuy, P.T., Moons, K., van Dijk, J.C., Anh, N.V., Van der Bruggen, B., 2008. To what extent are pesticides removed from surface water during coagulation-flocculation? Water and Environment Journal 22 (3), 217–223. van Leeuwen, J., Pipe-Martin, C., Lehmann, R.M., 2003. Water reclamation at South Caboolture, Queensland, Australia. Ozone: Science and Engineering 25 (2), 107–120. Vieno, N., Tuhkanen, T., Kronberg, L., 2006. Removal of pharmaceuticals in drinking water treatment: effect of chemical coagulation. Environmental Technology 27 (2), 183–192. Vogna, D., Marotta, R., Andreozzi, R., Napolitano, A., d’Ischia, M., 2004a. Kinetic and chemical assessment of the UV/H2O2 treatment of antiepileptic drug carbamazepine. Chemosphere 54 (4), 497–505. Vogna, D., Marotta, R., Napolitano, A., Andreozzi, R., d’Ischia, M., 2004b. Advanced oxidation of the pharmaceutical drug diclofenac with UV/H2O2 and ozone. Water Research 38 (2), 414–422. Wert, E.C., Rosario-Ortiz, F.L., Snyder, S.A., 2009. Effect of ozone exposure on the oxidation of trace organic contaminants in wastewater. Water Research 43 (4), 1005–1014. 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 (17), 6649–6663. Yoon, Y., Westerhoff, P., Snyder, S.A., Wert, E.C., Yoon, J., 2007. Removal of endocrine disrupting compounds and pharmaceuticals by nanofiltration and ultrafiltration membranes. Desalination 202 (1–3), 16–23. Yu, Z.R., Peldszus, S., Huck, P.M., 2008. Adsorption characteristics of selected pharmaceuticals and an endocrine disrupting compound – naproxen, carbamazepine and nonylphenol – on activated carbon. Water Research 42 (12), 2873–2882. Zour, E., Lodhi, S.A., Nesbitt, R.U., Silbering, S.B., Chaturvedi, P.R., 1992. Stability studies of gabapentin in aqueous-solutions. Pharmaceutical Research 9 (5), 595–600. Zwiener, C., Frimmel, F.H., 2000. Oxidative treatment of pharmaceuticals in water. Water Research 34 (6), 1881–1885.
water research 44 (2010) 638–648
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Removal of natural hormone estrone from secondary effluents using nanofiltration and reverse osmosis Xue Jin 1, Jiangyong Hu*, Say Leong Ong Division of Environmental Science and Engineering, National University of Singapore, 10 Kent Ridge Crescent, 119260, Singapore
article info
abstract
Article history:
The rejection of steroid hormone estrone by nanofiltration (NF) and reverse osmosis (RO)
Received 19 December 2008
membranes in treated sewage effluent was investigated. Four NF/RO membranes with
Received in revised form
different materials and interfacial characteristics were utilized. To better understand
1 June 2009
hormone removal mechanisms in treated effluent, effluent organic matters (EfOM) were
Accepted 23 September 2009
fractionated using column chromatographic method with resins XAD-8, AG MP-50 and IRA-
Available online 13 October 2009
96. The results indicate that the presence of EfOM in feed solution could enhance estrone rejection significantly. Hydrophobic acid (HpoA) organic fraction made a crucial contribu-
Keywords:
tion to this ‘‘enhancement effect’’. Hydrophobic base (HpoB) could also improve estrone
Estrone
rejection while hydrophobic neutral (HpoN) and hydrophilic acid (HpiA) with low aroma-
Nanofiltration
ticity had little effects. The increment in estrone rejection was predominantly attributed to
Reverse osmosis
the binding of estrone by EfOM in feed solutions, which led to an increase in molecular
Rejection
weight and appearance of negative charge (for the HpoA case) and thus an increased level
Treated effluent
of estrone rejection. However, the improvement of estrone rejection by HpoA decreased
Binding
with increasing calcium ion concentration. The important conclusion of this study is, first, hydrophobic acid macromolecules are recommended to be added into feed water to improve the rejection of trace hormone during NF/RO membrane process, and, second, removal of calcium ions via pretreatment and application of membrane with more negative charge at its interface can greatly intensify this ‘‘enhancement effect’’. ª 2009 Elsevier Ltd. All rights reserved.
1.
Introduction
Recently, the use of treated municipal wastewater for groundwater recharge and indirect drinking water reuse has become promising worldwide. Nanofiltration (NF) and reverse osmosis (RO) membranes are widely used in this kind of environmental application (Scha¨fer et al., 2005). However, the presence of endocrine disrupting chemicals (EDCs) in the aquatic environment has raised great consumer concern due to their potential health risk (Tim, 1997; Adams, 1998; Croley
et al., 2000; Baker, 2001; Iguchi et al., 2001). Amongst many kinds of EDCs, the impact of steroid hormones such as estrone, estradiol and ethinylestradiol is prominent because their potency can be several thousand times higher than that of others (Desbrow et al., 1998; Ternes et al., 1999; Johnson and Sumpter, 2001). It is noteworthy to emphasize that concentrations of some steroid hormones in secondary effluent are often high enough to harm wildlife although it is within ng/L range (Baronti et al., 2000; Williams et al., 2003; Soliman et al., 2004). In the light of the magnitude of this problem, NF/RO
* Corresponding author. Tel.: þ65 65164540; fax: þ65 67744202. E-mail addresses:
[email protected] (X. Jin),
[email protected] (J. Hu),
[email protected] (S.L. Ong). 1 Current address: Civil & Environmental Engineering Department, University of California, 5732 Boelter Hall, Los Angeles, CA 900951593, USA 0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2009.09.057
water research 44 (2010) 638–648
membranes and different post-treatment technologies are essential to remove EDCs from treated effluent before water reuse and ensure the safety of drinking water. Removal of EDCs using NF/RO membranes in single-organic solution (the experiments were conducted with target EDCs spiked in synthetic model solutions in the absence of other organics) has been reported in several recent studies (Kimura et al., 2003, 2004; Scha¨fer et al., 2003; Nghiem et al., 2004a, b). Many factors may influence the interaction between EDCs and membrane and thus their rejection including physicochemical properties of EDCs, such as molecular weight (MW), molecular size, charge (as a function of dissociation constant), hydrophobicity and polarity; membrane properties, such as material, surface charge (zeta potential), hydrophobicity, molecular weight cut-off (MWCO), porosity and desalting degree; and solution chemistry, such as pH, ionic strength and presence of divalent cations (Bellona et al., 2004, Scha¨fer et al., 2005). Another important factor influencing the rejection efficiency is initial concentration of target EDCs in feed solution. Kimura et al. (2003) reported that experiments conducted at a ng/l concentration range resulted in a lower rejection efficiency as compared to experiments conducted at a mg/l range. This observation substantiates that experiments conducted within the ‘‘realistic’’ concentration range are necessary to provide an accurate value of rejection efficiency. It should, however, be pointed out that the results achieved from a single-organic solution cannot be extrapolated to the mixtures typically encountered in a real aquatic environment. In previous works, we have investigated the influence of other dissolved organic matter (DOM) in feed solution on the removal of steroid hormone estrone during NF processes (Jin et al., 2007). The results indicated that estrone rejection depended on the type of DOM co-present. Neutral and highly hydrophilic dextran had little effect on estrone rejection. The presence of humic acid without phenolic groups but great aromaticity improved estrone rejection slightly. In contrast, estrone rejection was obviously increased by the presence of hydrophobic acid fraction (HpoA) derived from secondary effluent. Therefore, it is expected that the rejection mechanism of EDCs in real water matrix, such as secondary effluent, is very complicated. Organic matter in secondary effluent is referred to as effluent organic matter (EfOM) and can be categorized into fractions based on the difference in hydrophobicity and charge properties using a column chromatographic method (Leenheer, 1981; Thurman and Malcolm, 1981; Imai et al., 2002; Hu et al., 2003). Adsorption of some fractions in EfOM onto membranes would be expected to alter the membrane surface properties, which subsequently results in changes in the rejection of EDCs (Ng and Elimelech, 2004; Yoon et al., 2004; Xu et al., 2006; Steinle-Darling et al., 2007). Alternatively, EDCs can bind to some fractions in EfOM and be retained together, which can also affect their rejection characteristics (Agbekodo et al., 1996; Devitt et al., 1998; Majewska-Nowak et al., 2002; Nghiem et al., 2004a, b; Hu et al., 2007; Jin et al., 2007). Both interactions mentioned above (EfOM-membrane and EfOMEDCs) are markedly influenced by the physicochemical characteristics of EfOM. Therefore, different fractions of EfOM would demonstrate diverse impacts on EDCs rejection by membranes. To our knowledge, however, very little information is available on the rejection of trace EDCs by NF/RO
639
membranes with the presence of different fractions of EfOM. In order to better understand the removal mechanism of EDCs in secondary effluent by NF/RO membranes and optimize membrane performance, it is important to clarify this issue. It is known that divalent cations (especially calcium) affect the characteristics of EfOM, membrane charges and thus the two above-mentioned interactions (EfOM–membrane and EfOM–EDCs). For example, Devitt et al. (1998) reported that atrazine–NOM association decreased in the presence of calcium. Therefore, calcium ion is predicted to affect the rejection of EDCs with the presence of other EfOM. However, to date, the intricate relationship between physicochemical characteristics of EfOM, calcium ion concentration and membrane performance on EDC rejection remains poorly understood. In order to gain insight into the removal mechanism of EDCs in treated effluent, more research effort is needed to better understand the above relationship. In this study, we investigated the removal of steroid hormone estrone from secondary effluent using a bench-scale cross-flow NF/RO system. We examined the effect of various organic fractions isolated from secondary effluent and membrane characteristics on the removal of estrone by this process. After determining the key organic fraction which makes a crucial contribution to the enhanced estrone rejection by EfOM, the influence of calcium ion concentration on estrone rejection with the presence of the key fraction was investigated.
2.
Materials and methods
2.1.
NF/RO membranes and characterization
Four NF/RO membranes, denoted as DL, CK, AK and CG (Sepa membranes, OSMONICS, USA), were used. DL and AK are polyamide (PA) composite membranes. CK and CG are made of cellulose acetate (CA) polymer. New membranes were soaked in ultrapure water for 24 h with the water being replaced regularly prior to use. Membrane properties were characterized as described elsewhere (Jin et al., 2007) and are shown in Table 1. Briefly, contact angle was determined with ultrapure water at pH 5.8 0.2 using a goniometer (Rame´-hart contact angle goniometer, Imaging System, Mountain Lakes, NJ, USA). Zeta potential of the membrane surface was measured in the electrolyte background solution (1 mM NaHCO3 and 8 mM NaCl, pH 7) using an Electro Kinetic Analyzer (EKA) from Anton Paar (Austria). To characterize the membrane pore size, MWCO was estimated by filtering the electrolyte background solution containing polyethylene glycol (PEG) with MW of 200, 400, 600 and 1000 Da, purchased from Fluka (Switzerland). Filtration tests were performed using the same cross-flow membrane test unit as used in the estrone filtration tests with clean membrane. A PEG concentration of approximately 6 ppm as dissolved organic carbon (DOC) was used for each measurement.
2.2.
Membrane filtration unit
A laboratory-scale cross-flow membrane filtration unit was used (Fig. S1). Membrane coupon (13 cm by 13 cm) was housed in a stainless steel membrane cell. Feed solution was fed from
640
water research 44 (2010) 638–648
Table 1 – Membrane properties. Membrane DL CK AK CG
Membrane type
Contact angle ( )
Zeta potentiala
MWCO (Da)b
Polyamide NF Cellulose acetate NF Polyamide RO Cellulose acetate RO
30.7 2.7 54.2 1.4 41.7 1.3 63.2 1.7
16.3 2.7 6.0 0.3 19.9 0.2 10.6 0.6
490 560 <200 300
Typical flux/pressure (m/s per Pa)c 1:46 105 =6:89 105 1:32 105 =1:52 106 1:23 105 =7:93 105 1:41 105 =2:90 106
a Zeta potential on membrane surface was measured in the electrolyte background solution (1 mM NaHCO3 and 8 mM NaCl, pH 7). b MWCO was estimated by filtering the electrolyte background solution with a certain molecular weight range of poly-ethylene glycol (PEG). A PEG concentration of approximately 6 ppm as DOC was used for each measurement. c Information was provided by the manufacturer.
a reservoir by a pump, capable of providing a maximum pressure of 4137 kPa and a maximum flow of 20 l/min. The temperature inside the feed reservoir was maintained constant (22 0.1 C) by a recirculating chiller. The pressure (1379 kPa with DL, CK and AK, and 2758 kPa with CG) inside the membrane cell was set with a back-pressure regulator and a bypass valve. Each experiment was conducted in two steps: (1) the membrane was compacted with ultrapure water until the water flux remained constant; (2) the reservoir was emptied and filled with test solution. The solution was then filtered at constant transmembrane pressure. All experiments were conducted for 24 h. During filtration, both permeate and concentrate were recirculated back to the feed tank. At specified intervals, permeate and feed samples were collected for analysis. Rejection (R) is defined as R ¼ 100 ð1 Cp =Cf Þ, where Cp and Cf are the permeate and feed concentrations, respectively. The effect of EfOM on estrone rejection was determined based on the rejection measured at the end of filtration where stable rejection was achieved.
2.3.
EfOM fractionation method
Secondary effluent was collected from a local municipal wastewater treatment plant. The water was first filtered through a microfiltration (MF) unit (0.05 mm, Desal/Osmonics, USA) to simulate pretreatment ahead of NF/RO processes in the real plant, which is commonly used to remove suspended solids and colloids. The total organic carbon (TOC) and total dissolved solid (TDS) of the water samples ranged from 9.2 to 10.8 ppm and 455 to 618 ppm, respectively. Other characteristics of the MF treated secondary effluent are summarized in Table S1. The filtrates were subsequently fractionated into six subcomponents: HpoA, hydrophobic bases (HpoB), hydrophobic neutrals (HpoN), hydrophilic acids (HpiA), hydrophilic bases (HpiB) and hydrophilic neutrals (HpiN) using a method developed by Leenheer (1981). Supelite XAD-8 resin (SUPELCO, USA.), AG-MP-50 cation exchange resin (BioRad, USA) and Amberlite IRA-96 anion exchange resin (Rohm and Haas, France) were the resins used. Details on the instruments and fractionation methods have been published previously (Hu et al., 2003). The DOC fractionation results for the secondary effluent are presented in Table 2. HpoA predominated in secondary effluent and accounted for 36.6–40.3% of the total amount of dissolved organic matter. These results do not agree with an earlier study by Imai et al. (2002) who found different results on the fractions with HpiA as the most
abundant fraction for the samples taken from Japan. Thus, DOM composition in secondary effluent varies with the origin of wastewater and the type of treatment process used. In this study, the experimental results obtained in MF treated secondary effluent should be analyzed based on its specific DOM-fraction distribution as reported here.
2.4.
Chemical and solution chemistry
Pure standard of estrone (E1) and estrone-2,4,16,16-d4 (E1-d4) were purchased as powders from Sigma–Aldrich (Singapore) and are reported to be of 97% purity or higher. The latter compound was used as an internal standard. The structure and characteristics of estrone are described elsewhere (Scha¨fer et al., 2003; Nghiem et al., 2004a). MW of the estrone molecule is 270 Da. Estrone, with pKa above 10, is predominantly nonionic at the neutral pH investigated in this study. Its octanol–water partitioning coefficient (logKow) is 3.43. As for its structure, the estrone molecule possesses one aromatic ring. Moreover, its phenolic hydroxyl functional groups can facilitate the formation of hydrogen bonding between estrone and other organics. There were three types of feed solution used in this study: (1) estrone in electrolyte background solution (1 mM NaHCO3 and 8 mM NaCl, pH 7); (2) estrone in MF treated secondary effluent; (3) estrone in the presence of specific organic fraction. To prepare the 3rd type of solution, isolated organic fraction was added to ultrapure water to achieve the desired TOC concentration (1 ppm or its original value in secondary effluent) followed by addition of 1 mM NaHCO3. After that, pH was adjusted to 7 and the appropriate amount of NaCl was added to achieve the same ionic strength as that in the 1st type of solution. Estrone was finally spiked into the solution.
Table 2 – DOC fractionation results. Fraction HpoA HpoB HpoN HpiA HpiB HpiN
DOC (ppm)
Percentage of total DOC (%)
3.4–4.2 0.5–0.6 1.2–1.3 0.5–0.9 0.3–0.5 2.4–3.2
36.6–40.3 5.2–6.5 10.7–14.0 5.7–9.5 3.3–5.1 25.4–35.0
water research 44 (2010) 638–648
In all feed waters, initial estrone concentration was maintained at 100 ng/L. Solution with different calcium ion concentrations (0, 0.3 and 0.6 mM) were accomplished by adding CaCl2 stock solution, while the total ionic strength was kept constant by adjusting NaCl concentration.
2.5.
Analytical methods
2.5.1.
Estrone detection
Estrone was extracted from aqueous sample by OASIS HLB cartridge (500 mg) from Waters, Singapore. Before the analytes were extracted, 80 ng/l of E1-d4 was added to the sample water to increase accuracy of the analytical data. The cartridge was subsequently conditioned by 6 ml of diethylether, 5 ml of methanol and 10 ml of ultrapure water. The sample water (400 ml) was loaded through the cartridge at flow rate of 3 ml/min with the aid of a vacuum pump. After sample loading, the cartridge was eluted with 10 ml of a methanol–diethylether solution (10:90, v/v). The eluant was collected in a brown glass vial and was allowed to dry under a gentle stream of nitrogen. After drying, the residuals were re-dissolved in 0.4 ml of methanol and then analyzed by liquid chromatography–tandem mass spectrometry (MDS Sciex API 2000 tandem triple-quadropole mass spectrometer) from Applied Biosystems, USA. Analytes were chromatographed on a 2.1 150 mm column filled with 3.5 mm C18 reversed phase packing (Agilent, Singapore). The multiple reaction monitoring (MRM) mode was chosen for quantification. After observing collision-induced dissociation (CID) spectra obtained by full-scan production experiments, the following MRM pairs were chosen: E1: 269.2/145.0; 269.2/143.0; E1-d4: 273.0/146.9. Estrone recovery was not matrix-dependent due to using internal standard. The mean recovery and relative standard deviation (RSD) was 99 7.4% (n ¼ 16). The limit of detection was 0.5 ng/l, calculated using a signal-to-noise ratio of 3. Details with respect to basic information on this method are provided separately (Jin, 2007).
2.5.2.
Results and discussion
3.1. Rejection of estrone by NF/RO membranes in electrolyte background solution Filtration of estrone in electrolyte background solution (1 mM NaHCO3 and 8 mM NaCl, pH 7) provides the baseline from which the effect of other organic matter on estrone rejection can be found. The open symbols in Fig. 1 present the observed rejections of estrone in electrolyte background solution by four kinds of NF/RO membranes. It appeared that estrone rejection was initially higher than 90% and decreased dramatically and then stabilized at later filtration stage. The excellent removal performances of all membranes at the initial filtration stage can be attributed to their adsorption capabilities and steric hindrance. However, the adsorption effect can only contribute to the short-term removal of estrone. As the feed solution is continuously filtered through the membrane, more and more available sites on the membrane are occupied by adsorbed estrone. When the partition of estrone between feed solution and membrane reaches equilibrium, there is no further net adsorption effect taking place and thus the contribution from adsorption would be negligible. Under this condition, size exclusion would become the overriding removal mechanism at the later filtration stage. Consistent with the MWCO values of the membranes, the ultimate rejection of estrone followed the order: AK > CG > DL > CK. AK membrane with MWCO value below the MW of estrone can remove estrone efficiently. However, for the other three kinds of membranes with MWCO values larger than the MW of estrone, ultimate rejection was extremely low. To avoid overestimation of estrone rejection, in this paper, all of the determined rejections were based on 23 h of filtration when stable rejection was achieved.
3.2. Rejection of estrone by NF/RO membranes in secondary effluent matrix
TOC and UV254 analysis
In all cases, the concentrations of EfOM were detected by a TOC analyzer (Shimadzu TOC-Vcsh, Japan). TOC and DOC are used interchangeably in this study because secondary effluent was filtered through 0.05 mm membrane before using. The specific ultraviolet absorption of selected organic fraction at 254 nm (SUVA254) was calculated by dividing the absorbance at 254 nm measured using the Shimadzu UV-1700 UV-Visible Spectrophotometer (Japan) by the DOC level of the solution.
2.5.3.
3.
641
Molecular Weight Measurement
High performance size exclusion chromatography (HPSEC) was performed to determine both weight- and number-averaged molecular weights. Instrumentation consisted of a SHIMADZU (Japan) HPLC instrument LC-10AT VP with a Polymer Laboratories aquagel-OH column and a SHIMADZU SPD-M10A diode array detector operating at 260 cm. The HPSEC system was calibrated using polystyrene sulfonates standards of known average MW between 210 and 13000, and acetone. Mobile phase was composed of ultrapure water buffered with phosphate to pH of 6.8.
Estrone rejection in MF treated secondary effluent, compared with that in electrolyte background solution, using four kinds of NF/RO membranes was monitored and the results are presented in Fig. 1, whereas the corresponding DOC rejection in secondary effluent is presented in Fig. 2. Estrone removal efficiencies in MF treated secondary effluent matrix for all NF/RO membranes used were consistently higher than the results obtained in electrolyte background solution. The improvements in ultimate estrone rejection were 6.5%, 32.5%, 21.6% and 11.6% for AK, DL, CK and CG membranes, respectively. The results obtained here are consistent with an earlier study by Comerton et al. (2008), who investigated the influence of water matrix on EDC rejection by RO and NF membranes. They reported a significant increase in EDCs rejection by tight NF membranes in membrane bioreactor effluent and natural waters when compared to the ultrapure water. The improvement in estrone rejection could be due to the following reasons. Firstly, permeate flux decline was observed in MF treated secondary effluent matrix during separation processes (in electrolyte background solution, reduction in permeate flux was not observed for any membrane tested,
642
water research 44 (2010) 638–648
b
100
Electrolyte background Secondary effluent
80 60 40 20
Estrone Rejection (%)
Estrone Rejection (%)
a
100
Electrolyte background Secondary effluent
80 60 40 20 0
0
0
5
10
15
20
0
25
5
10
15
20
25
Filtration Time (h)
Filtration Time (h)
d
c
100
80 60
Electrolyte background Secondary effluent
40 20 0
Estrone Rejection (%)
Estrone Rejection (%)
100
60 40 20 0
0
5
10
15
20
25
Electrolyte background Secondary effluent
80
0
5
10
15
20
25
Filtration Time (h)
Filtration Time (h)
Fig. 1 – Estrone rejection in electrolyte background solution and MF treated secondary effluent by (a) DL, (b) CK, (c) AK and (d) CG.
which can be attributed to the very low concentration of estrone in the feed solution). After 23 h of filtration, the flux decreased by about 13.0%, 8.9%, 33.8% and 17.5% for DL, CK, AK and CG membrane, respectively. This suggests that the membranes were fouled by EfOM. The entrapment of EfOM onto the membrane active layer restricted the passage of contaminants, resulting in an increase in the observed rejection of both DOC (Fig. 2) and estrone at the ng/l level (Fig. 1). Similar increases in trace organic rejection as a result of organic fouling have been reported in previous studies (Agenson and Urase, 2007; Bellona and Drewes, 2007; Nghiem and Hawkes, 2007). However, simple EfOM fouling may not be the main reason, because the sequence
100
DOC Rejection (%)
95 90 85 80 75
DL CK AK CG
70 65 60
0
5
10 15 Filtration Time (h)
20
25
Fig. 2 – DOC rejection in MF filtered secondary effluent by four kinds of NF/RO membranes.
of the improvement in estrone rejection (DL > CK > CG > AK) was mismatched with the sequence based on permeate flux decline in secondary effluent matrix (AK > CG > DL > CK). Therefore the enhanced estrone rejection can be mainly attributed to the other reason, namely that, secondly, estrone molecules may bind to some fractions of EfOM in bulk solution and be retained together, resulting in a reduction in the passage of estrone molecules through membrane. The remainder of this paper is focused on proving this hypothesis.
3.3. Rejection of estrone in the presence of specific organic fractions at same TOC level In order to optimize membrane performance, it is essential to identify which organic fraction in secondary effluent would make a crucial contribution to the ‘‘enhancement effect’’ on estrone removal. Amongst the six fractions derived from EfOM, HpiB was present in insignificant quantities and thus was expected to have a negligible influence on estrone removal. HpiN, as the last fraction derived from the resin fractionation system, inevitably contains some impurities of other fractions (Hu et al., 2003). Moreover, as shown in our previous work (Jin et al., 2007), dextran representing hydrophilic neutral organics had little effect on estrone rejection. Therefore, in this study, the research was focused on the influences of HpoA, HpoB, HpoN and HpiA on estrone rejection by NF/RO membranes. As shown earlier, DL and CK membranes demonstrated a greater improvement in estrone rejection in MF treated secondary effluent compared to AK and CG membranes. Investigations were firstly conducted to compare the capability of each isolated fraction to improve estrone removal
water research 44 (2010) 638–648
a 100
Electrolyte background
Estrone Rejection (%)
With HpoA With HpoB 80
With HpoN With HpiA
60
40
20
0
0
5
10
15
20
25
Filtration Time (h)
b 100
Estrone Rejection (%)
Electrolyte background With HpoA
80
With HpoB With HpoN 60
With HpiA
40
20
0
0
5
10
15
20
25
Filtration Time (h) Fig. 3 – Observed rejection of estrone with the presence of different organic fractions at same DOC level (1 ppm) by (a) DL and (b) CK.
using DL and CK membranes. To avoid the influence of DOC values, the initial concentration of each organic fraction was adjusted to a DOC value of approximately 1 ppm, which was close to the original concentrations of HpoB, HpoN and HpiA present in secondary effluent (Table 2). Estrone rejection in the presence of each fraction over a 24 h filtration period is presented in Fig. 3. It is evident that estrone rejection was significantly improved by the presence of HpoA, rising from 15.3% to 41.9% with DL membrane and from 8.2% to 15.3% with CK membrane. The presence of HpoB also resulted in an improvement in estrone rejection of as high as 3.6% and 4.1% with DL and CK, respectively. In contrast, HpoN and HpiA did not exhibit any obvious increase in estrone rejection for any of the membranes tested, although HpoN and HpiA were removed more efficiently than HpoB (Fig. 4). The differences in the ‘‘enhancement effect’’ between HpoA and HpiA indicate that hydrophobic organic fractions have a greater contribution to the enhanced estrone rejection in secondary effluent than hydrophilic organic fractions. In all experiments, no obvious permeate flux decline (<7%, data not shown) was noticed, probably due to insufficient
643
DOM accumulation onto membrane. Therefore, under the experimental conditions in this part, the increment in ultimate estrone rejection by membranes would most probably be attributed to the fact that binding of estrone molecules to bulk organic matter in feed solution would retard the passage of estrone molecules through membrane. Based on this understanding, it could be inferred that HpoA and HpoB would bind estrone molecules in feed solutions more favorably than HpoN and HpiA would since the presence of either HpoA or HpoB could lead to improvement in estrone removal while the presence of HpoN and HpiA could not. According to SUVA254 values (Table 3), HpoA and HpoB fractions contained more aromatic constituents than HpoN and HpiA fractions. Greater aromaticity presents greater binding capability with estrone for HpoA and HpoB, suggesting the important role of DOM’s aromaticity in the binding mechanism between estrone and bulk organic matter. This is consistent with the observation of Yamamoto et al. (2003), who reported a positive correlation between aromaticity of DOM and their binding with selected EDCs. Furthermore, in our previous work (Jin et al., 2007), it was indicated that hydrogen bonding between phenolic groups of estrone and HpoA played the dominating role in the binding between them in the water phase where estrone concentration is extremely low (at ng/l level). Organic compounds such as aromatic amine are categorized in the HpoB fraction (Imai et al., 2002). The interaction between estrone and the HpoB fraction in feed solution may be attributed to the hydrogen bonding between phenolic groups in estrone molecules and aromatic amine functional groups (such as the anilinic group) in the HpoB fraction because phenolic groups in estrone molecules are the most interactive sites due to the resonance structure of aromatic groups (Nghiem et al., 2004a). The possible hydrogen bonding formations are shown in Fig. 5. To our knowledge, few works on the binding of trace EDCs to DOM were documented. Therefore, further investigation at the molecular level is needed to verify the above hypothesis.
3.4. Rejection of estrone in the presence of hydrophobic organic fractions at their original TOC levels in SE In addition to the individual characteristics of bulk organic matters, their feed concentrations may also affect the rejection of target solute by membranes. In this section, experiments were performed to compare estrone rejection in the presence of isolated hydrophobic organic fractions at their original DOC concentrations present in treated effluent. Two NF membranes (DL and CK) and one RO membrane (CG) were used in this instance. Fig. 6 shows the rejection of estrone in the presence of each hydrophobic organic fraction over a 24 h filtration period. The corresponding DOC rejection attained by each membrane is presented in Fig. 7. The addition of HpoN did not show any obvious influence on estrone removal. As discussed above, this is due to the little binding between estrone and HpoN in aquatic matrix. The presence of HpoB remarkably enhanced estrone rejection by DL and CG membranes. However, in the case of CK membrane, no significant ‘‘enhancement effect’’ of HpoB was observed. As explained above, the improved estrone rejection was attributed to the fact that one portion of estrone
644
water research 44 (2010) 638–648
a 100 90
DOC Rejection (%)
80 70 60 50 40 HpoA HpoB HpoN HpiA
30 20 10 0
0
5
10
15
20
25
Filtration Time (h) Fig. 5 – Schematic of possible hydrogen bonding between phenolic groups of estrone molecule and HpoA and between phenolic group in estrone molecule and aromatic amine group in HpoB (A–E stand for any possible functional groups within HpoB fraction).
b 100 90
DOC Rejection (%)
80 70 60 50 40 HpoA HpoB HpoN HpiA
30 20 10 0 0
5
10
15
20
25
Filtration Time (h) Fig. 4 – Observed rejection of bulk organics (as DOC) with the presence of different organic fractions at same DOC level (1 ppm) by (a) DL and (b) CK.
molecules can bind to bulk DOM in feed matrix and subsequently be retained together. In this case, the ‘‘enhancement effect’’ on estrone removal was controlled not only by the binding between estrone and DOM in feed solution, but also by the rejection efficiency of the DOM with which estrone molecules are associated. Therefore, the weak effect of HpoB on estrone rejection is due to the inefficient removal of HpoB by CK membrane (Fig. 7b). It is apparent that the HpoA fraction, for all membranes used, had a significantly greater ‘‘enhancement effect’’ than the other two hydrophobic fractions on ultimate estrone removal. After 23 h of filtration, the presence of HpoA increased estrone rejection by 29.8%, 13.3% and 20.5% for DL, CK and CG membranes, respectively. This could be attributed to a combination of the following reasons. Firstly, the structural characteristics of HpoA (phenolic group)
allowed estrone molecules to bind with HpoA in feed solution through hydrogen bonding, thus estrone could be removed together with HpoA. Secondly, HpoA with the highest concentration (Table 2) in feed solution would have the highest chance to associate with estrone molecules. Thirdly, HpoA with the largest MW of more than 4000 Da and a negative charge could be removed effectively (Fig. 7). The formation of estrone-HpoA complex in bulk solution led to a significant increase in MW (from 270 Da to more than 4000 Da) and the appearance of negative charge and thus an increased level of estrone rejection by negative charged membranes based on the mechanisms of size exclusion and charge repulsion. Therefore, it could be concluded that the HpoA fraction was the greatest contributor to the ‘‘enhancement effect’’ on estrone rejection in treated effluent. It is worthwhile to note that the ‘‘enhancement effect’’ of HpoA followed the order of DL > CG > CK. This sequence agrees with the sequence based on the negative charge of the membrane surface (Table 1). According to the above discussion, the binding of estrone with HpoA in feed solution was critically important for the ‘‘enhancement effect’’. Less negative charge of membrane surface would result in more HpoA adsorption onto membrane due to the weaker electrostatic repulsion between HpoA and membrane. Consequently, less HpoA molecules would be available in feed solution for binding with estrone, resulting in a weaker improvement in estrone rejection. The results suggest that a more negatively-charged
Table 3 – Characteristics of specific organic fractions derived from secondary effluent. Parameter MW (Da) r SUVA254 (m1 L/mg C)
HpoA
HpoB
HpoN
HpiA
4192–4442 1.29–1.43 2.58–4.25
762–1751 1.01–1.84 2.06–3.81
1406–2658 1.01–1.27 1.19–1.57
3628–3825 1.18–1.32 0.75–0.81
645
water research 44 (2010) 638–648
a
a 80
100 90
Electrolyte background With HpoA With HpoB With HpoN
DOC Rejection (%)
Estrone Rejection (%)
100
60 40 20
80 70 60 50 40 HpoA HpoB HpoN
30 20 10
0
0
5
10
15
20
0
25
0
5
Filtration Time (h)
b
b Electrolyte background With HpoA With HpoB With HpoN
80 60 40 20 0
70 60 50 40 30 HpoA HpoB HpoN
10 0
5
10
15
20
0
25
Electrolyte background With HpoA With HpoB With HpoN
80
0
5
10
15
60 40 20
c
5
10
15
25
100 90 80 70 60 50 40 30
HpoA HpoB HpoN
20 10
0
20
Filtration Time (h)
DOC Rejection (%)
100
25
80
20
c Estrone Rejection (%)
20
100
Filtration Time (h)
0
15
90
DOC Rejection (%)
Estrone Rejection (%)
100
10
Filtration Time (h)
20
25
Filtration Time (h)
0 0
5
10
15
20
25
Filtration Time (h) Fig. 6 – Observed rejection of estrone with the presence of different hydrophobic organic fractions (at their original concentration in secondary effluent) by (a) DL, (b) CK and (c) CG.
membrane would benefit the interaction between HpoA and estrone and thus estrone rejection.
3.5. Effect of calcium ion concentration on estrone rejection in HpoA-containing solution Based on the discussion above, the estrone-binding capacity of HpoA is the major contributor to its significant ‘‘enhancement effect’’ on estrone rejection by NF/RO membranes. It is known that divalent cations, such as calcium, can influence the binding of trace contaminants by humic substances
Fig. 7 – Observed rejection of bulk organics (as DOC) with the presence of different hydrophobic organic fractions (at their original concentration in SE) by (a) DL, (b) CK and (c) CG.
(Schlautman and Morgan, 1993). Therefore, calcium ion concentration in feed solution is predicted to affect the estrone rejection in HpoA-containing solution. Fig. 8 presents the influence of HpoA on estrone rejection by DL membrane at different calcium concentrations. It is evident that while the presence of HpoA could increase the rejection of estrone, a higher calcium ion concentration tended to reverse this effect. In particular, the addition of 0.3 mM calcium ion in feed solution resulted in the ‘‘enhancement
646
water research 44 (2010) 638–648
Estrone Rejection (%)
a
100 w/o HpoA with HpoA
80 60 40 20 0
0
5
10
15
20
25
Filtration Time (h)
Estrone Rejection (%)
b
100 w/o HpoA with HpoA
80 60 40 20 0
0
5
10
15
20
25
Filtration Time (h)
Estrone Rejection (%)
c 100
w/o HpoA with HpoA
80
from other published investigations (Hong and Elimelech, 1997; Tang et al., 2007), in which a more serious permeate flux decline with an increased calcium concentration was reported. This phenomenon was possibly due to the relatively low calcium ion concentration added in this study. Such low calcium ion concentration might not be enough to overcome the electrostatic repulsive energy barrier between HpoA and membrane and thus cannot result in a severe permeate flux decline. In this case, the diminished effect of HpoA on estrone rejection by membrane with increasing calcium concentration can most probably be attributed to the lower estrone–HpoA complex formation. Calcium ions can interact specifically with acidic functional groups (predominantly carboxylic groups) of humic substances (Hong and Elimelech, 1997). Such interactions reduce the charge repulsion between negative functional groups on HpoA molecules due to charge shielding and neutralization. As a result, the structure of HpoA tends to change from a stretched and linear configuration to a coiled and compact configuration with increased calcium concentration (Clark and Lucas, 1998). This would result in the interactive sites within HpoA molecules becoming less easily reachable by estrone molecules by restricting access to interior sorption sites. In this case, the increase in calcium concentration would decrease the ability of HpoA to bind estrone molecules in feed water. This observation is similar to the results of Schlautman and Morgan (1993) who reported that the presence of calcium at neutral pH decreased the ability of humic acid to bind perylene. They attributed the finding to the fact that the compression of the HA structure decreased the size of voids in humic acid into which target solute could partition. Since the presence of calcium ions was not conducive to the binding of estrone by HpoA, the effect of increasing calcium concentration, even at relatively low levels, was to push estrone rejection towards the ‘‘free’’ estrone limit of rejection obtained in the absence of HpoA.
60
4. 40 0
5
10
15
20
25
Filtration Time (h) Fig. 8 – Influence of HpoA on estrone rejection by DL membrane at different calcium ion concentration (a) 0 mM Ca2D, (b) 0.3 mM Ca2D and (c) 0.6 mM Ca2D.
effect’’ of HpoA on ultimate estrone rejection by DL membrane decreasing to 18.0% as compared to the improvement of 29.8% with the absence of calcium ion. When calcium ion concentration was increased to 0.6 mM, HpoA did not exhibit observable improvement in estrone rejection. In addition, negligible differences in permeate flux decline and DOC rejection were observed with increasing calcium concentration (data are not shown, but are available on request). This observation is unexpected and also differs
Conclusions
The results reported in this paper demonstrate that NF and loose RO membranes cannot remove trace steroid hormone efficiently in single-organic solution. However, the removal efficiency can be enhanced significantly in the presence of EfOM in feed solution. The HpoA fraction played a paramount role in the ‘‘enhancement effect’’. This enhancement is mainly attributed to the binding of estrone molecules to HpoA in feed solutions and being removed with the negatively charged HpoA macromolecules based on the mechanisms of size exclusion and charge repulsion. However, the presence of calcium ion tended to diminish this effect. Application of membrane with more negative-charge at its interface can greatly intensify this ‘‘enhancement effect’’.
Supplementary data Supplementary data associated with this article can be found in online version at doi:10.1016/j.watres.2009.09.057.
water research 44 (2010) 638–648
references
Adams, N.R., 1998. Clover phyto-oestrogens in sheep in Western Australia. Pure and Applied Chemistry 70 (9), 1855–1862. Agbekodo, K.M., Legube, B., Dard, S., 1996. Atrazine and simazine removal mechanisms by nanofiltration: influence of natural organic matter concentration. Water Research 30 (11), 2535–2542. Agenson, K.O., Urase, T., 2007. Change in membrane performance due to organic fouling in nanofiltration (NF)/reverse osmosis (RO) applications. Separation and Purification Technology 55 (2), 147–156. Baker, V.A., 2001. Endocrine disrupters – testing strategies to assess human hazard. Toxicology in Vitro 15 (4-5), 413–419. Baronti, C., Curini, R., D’Ascenzo, G., Di Corcia, A., Gentili, A., Samperi, R., 2000. Monitoring natural and synthetic estrogens at activated sludge sewage treatment plants and in a receiving river water. Environmental Science & Technology 34 (24), 5059–5066. Bellona, C., Drewes, J.E., 2007. Viability of a low-pressure nanofilter in treating recycled water for water reuse applications: a pilotscale study. Water Research 41 (17), 3948–3958. Bellona, C., Drewes, J.E., Xu, P., Amy, G., 2004. Factors affecting the rejection of organic solutes during NF/RO treatment – a literature review. Water Research 38 (12), 2795–2809. Clark, M.M., Lucas, P., 1998. Diffusion and partition of humic acid in a porous ultrafiltration membrane. Journal of Membrane Science 143 (1–2), 13–25. Comerton, M.A., Andrews, C.R., Bagley, M.D., Hao, C., 2008. The rejection of endocrine disrupting and pharmaceutically active compounds by NF and RO membranes as a function of compounds and water matrix properties. Journal of Membrane Science 313 (1-2), 323–335. Croley, T.R., Hughes, R.J., Koenig, B.G., Metcalfe, C.D., March, R.E., 2000. Mass spectrometry applied to the analysis of estrogens in the environment. Rapid Communications in Mass Spectrometry 14 (13), 1087–1093. Desbrow, C., Routledge, E.J., Brighty, G.C., Sumpter, J.P., Waldock, M., 1998. Identification of estrogenic chemicals in STW effluent. 1. Chemical fractionation and in vitro biological screening. Environmental Science & Technology 32 (11), 1549–1558. Devitt, E.C., Ducellier, F., Cote, P., Wiesner, M.R., 1998. Effects of natural organic matter and the raw water matrix on the rejection of atrazine by pressure-driven membranes. Water Research 32 (9), 2563–2568. Hong, S., Elimelech, M., 1997. Chemical and physical aspects of natural organic matter (NOM) fouling of nanofiltration membranes. Journal of Membrane Science 132 (2), 159–181. Hu, J.Y., Ong, S.L., Shan, J.H., Kang, J.B., Ng, W.J., 2003. Treatability of organic fractions derived from secondary effluent by reverse osmosis membrane. Water Research 37 (19), 4801–4809. Hu, J.Y., Jin, X., Ong, S.L., 2007. Rejection of estrone by nanofiltration: influence of solution chemistry. Journal of Membrane Science 302 (1-2), 188–196. Iguchi, T., Watanabe, H., Katsu, Y., 2001. Developmental effects of estrogenic agents on mice, fish, and frogs: a mini-review. Hormones and Behavior 40 (2), 248–251. Imai, A., Fukushima, T., Matsushige, K., Kim, Y.H., Choi, K., 2002. Characterization of dissolved organic matter in effluents from wastewater treatment plants. Water Research 36 (4), 859–870. Jin, X., 2007. Rejection of steroid hormone estrone by NF/RO membranes. PhD thesis. Department of Civil Engineering, National University of Singapore, Singapore. Jin, X., Hu, J.Y., Ong, S.L., 2007. Influence of dissolved organic matter on estrone removal by NF membranes and the role of their structures. Water Research 41 (14), 3077–3088.
647
Johnson, A.C., Sumpter, J.P., 2001. Removal of endocrinedisrupting chemicals in activated sludge treatment works. Environmental Science & Technology 35 (24), 4697–4703. Kimura, K., Amy, G., Drewes, J.E., Heberer, T., Kim, T.U., Watanabe, Y., 2003. Rejection of organic micropollutants (disinfection by-products, endocrine disrupting compounds, and pharmaceutically active compounds) by NF/RO membranes. Journal of Membrane Science 227 (1–2), 113–121. Kimura, K., Toshima, S., Amy, G., Watanabe, Y., 2004. Rejection of neutral endocrine disrupting compounds (EDCs) and pharmaceutical active compounds (PhACs) by RO membranes. Journal of Membrane Science 245 (1–2), 71–78. Leenheer, J.A., 1981. Comprehensive approach to preparative isolation and fractionation of dissolved organic carbon from natural waters and wastewaters. Environmental Science & Technology 15 (5), 578–587. Majewska-Nowak, K., Kabsch-Korbutowicz, M., Dod, M., Winnicki, T., 2002. The influence of organic carbon concentration on atrazine removal by UF membranes. Desalination 147 (1–3), 117–122. Ng, H.Y., Elimelech, M., 2004. Influence of colloidal fouling on rejection of trace organic contaminants by reverse osmosis. Journal of Membrane Science 244 (1–2), 215–226. Nghiem, L.D., Hawkes, S., 2007. Effects of membrane fouling on the nanofiltration of pharmaceutically active compounds (PhACs): Mechanisms and role of membrane pore size. Separation and Purification Technology 57 (1), 176–184. Nghiem, L.D., Manis, A., Soldenhoff, K., Scha¨fer, A.I., 2004a. Estrogenic hormone removal from wastewater using NF/RO membranes. Journal of Membrane Science 242 (1–2), 37–45. Nghiem, L.D., Scha¨fer, A.I., Elimelech, M., 2004b. Removal of natural hormones by nanofiltration membranes: measurement, modeling, and mechanisms. Environmental Science & Technology 38 (6), 1888–1896. Scha¨fer, A.I., Fane, A.G., Waite, T.D., 2005. Nanofiltration: Principles & Applications. Elsevier Advanced Technology, Oxford, New York. Scha¨fer, A.I., Nghiem, L.D., Waite, T.D., 2003. Removal of the natural hormone estrone from aqueous solutions using nanofiltration and reverse osmosis. Environmental Science & Technology 37 (1), 182–188. Schlautman, M.A., Morgan, J.J., 1993. Effects of aqueous chemistry on the binding of polycyclic aromatic hydrocarbons by dissolved humic materials. Environmental Science & Technology 27 (5), 961–969. Soliman, M.A., Pedersen, J.A., Suffet, I.H., 2004. Rapid gas chromatography–mass spectrometry screening method for human pharmaceuticals, hormones, antioxidants and plasticizers in water. Journal of Chromatography A 1029 (1-2), 223–237. Steinle-Darling, E., Zedda, M., Plumlee, H.M., Ridgway, F.H., Reinhard, M., 2007. Evaluating the impacts of membrane type, coating, fouling, chemical properties and water chemistry on reverse osmosis rejection of seven nitrosoalklyamines, including NDMA. Water Research 41 (17), 3959–3967. Tang, Y.C., Kwon, Y.N., Leckie, O.J., 2007. Fouling of reverse osmosis and nanofiltration membranes by humic acid – effects of solution composition and hydrodynamic conditions. Journal of Membrane Science 290 (1–2), 86–94. Ternes, T.A., Stumpf, M., Mueller, J., Haberer, K., Wilken, R.D., Servos, M., 1999. Behavior and occurrence of estrogens in municipal sewage treatment plants - I. Investigations in Germany, Canada and Brazil. The Science of the Total Environment 225 (1–2), 81–90.
648
water research 44 (2010) 638–648
Thurman, E.M., Malcolm, R.L., 1981. Preparative isolation of aquatic humic substances. Environmental Science & Technology 15 (4), 463–466. Tim, Z., 1997. In vitro bioassays for assessing estrogenic substances. Environmental Science & Technology 31 (3), 613–623. Williams, R.J., Johnson, A.C., Smith, J.J., Kanda, R., 2003. Steroid estrogens profiles along river stretches arising from sewage treatment works discharges. Environmental Science & Technology 37 (9), 1744–1750. Xu, P., Drewes, J.E., Kim, T.U., Bellona, C., Amy, G., 2006. Effect of membrane fouling on transport of organic contaminants in
NF/RO membrane applications. Journal of Membrane Science 279 (1–2), 165–175. Yamamoto, H., Liljestrand, H.M., Shimizu, Y., Morita, M., 2003. Effects of physical-chemical characteristics on the sorption of selected endocrine disruptors by dissolved organic matter surrogates. Environmental Science and Technology 37 (12), 2646–2657. Yoon, Y., Westerhoff, P., Yoon, J., Snyder, S.A., 2004. Removal of 17 beta estradiol and fluoranthene by nanofiltration and ultrafiltration. Journal of Environmental Engineering 130 (12), 1460–1467.
water research 44 (2010) 649–657
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Screening of antimycotics in Swedish sewage treatment plants – Waters and sludge Richard H. Lindberg*, Jerker Fick, Mats Tysklind Department of Chemistry, Umea˚ University, SE-901 87 Umea˚, Sweden
article info
abstract
Article history:
Concentrations of six pharmaceutical antimycotics were determined in the sewage water,
Received 25 March 2009
final effluent and sludge of five Swedish sewage treatment plants (STPs) by solid phase
Received in revised form
extraction, liquid/solid extraction, and liquid chromatography–electrospray tandem mass
19 October 2009
spectrometry. The antimycotics were quantified by internal standard calibration. The
Accepted 23 October 2009
results were used to estimate national flows that were compared to predictions based on
Available online 31 October 2009
sales figures. Fluconazole was the only one of the six investigated antimycotics that was detected (at concentrations ranging from 90 to 140 ng L1) in both raw sewage water and
Keywords:
final effluent. Negligible amounts of this substance were removed from the aqueous phase,
Antimycotics
and its levels were below the limit of quantification in all of the analyzed sludge samples.
Sewage water
In contrast, clotrimazole, ketoconazole and econazole were present in all of the sludge
Sludge
samples, at concentrations ranging between 200 and 1000 mg kg1, dry weight. There were
Mass flow
close correlations between the national measured and predicted antimycotic mass flows. Antimycotic fate analysis, based on sales figures, indicated that 53% of the total amount of fluconazole sold appeared in the final effluents of the STPs, while 1, 155, 35, 209 and 41% of the terbinafine, clotrimazole, ketoconazole, econazole and miconazole sold appeared in the digested dewatered sludge. ª 2009 Elsevier Ltd. All rights reserved.
1.
Introduction
There are approximately 40 antimycotics, i.e. pharmaceuticals that can be used to treat fungal infections in humans, many of which are imidazoles, triazoles or allylamines. Antimycotics are usually administered topically, although a few, such as ketoconazole and fluconazole, are also used perorally. All antimycotics currently used are general cytochrome P450 inhibitors that reportedly affect CYP families 1–3. For instance, the imidazoles and triazoles strongly inhibit CYP3A4 and 2C9, while the allylamines are known to affect CYP2D6 (Sweetman, 2007). This has potentially significant implications for their use as pharmaceuticals since the P450s are responsible for the initial metabolism of ca. 75% of xenobiotics in humans, hence
they could have undesirable effects on patients’ ability to eliminate other substances (Sweetman, 2007; Guengerich, 2008). More specifically, the imidazole and triazole antimycotics inhibit the essential activity of CYP51 (also known as ERG11 in fungi) in ergosterol synthesis in fungal pathogens, thereby increasing the permeability of their membranes, inhibiting their growth and/or killing them (Graybill and Drutz, 1980; Yoshida et al., 2000; Lepesheva and Waterman, 2007). The antimycotic activity of allylamine antimycotics is also based on interference with ergosterol biosynthesis, but by inhibiting squalene epoxidase (Ryder, 1992). Gunnarson et al. have recently shown that six species commonly used in aquatic ecotoxicology tests possess CYP51 orthologs that have more than 70% homology to the drug
* Corresponding author. Tel.: þ46 90 786 5464; fax: þ46 90 12 7655. E-mail address:
[email protected] (R.H. Lindberg). 0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2009.10.034
650
water research 44 (2010) 649–657
targets (Gunnarsson et al., 2008). Further, imidazole and triazole antimycotics have been shown to inhibit CYP19, which participates in the conversion of androgens to estrogens (Tro¨sken et al., 2004; Gyllenhammar et al., 2009), with serious potential consequences for sex differentiation in exposed vertebrates, including testicular development in female salmon (Piferrer et al., 1994). Gyllenhammar et al. (2009) have also shown that exposure to the imidazole clotrimazole can have complex effects on aromatase activity in the gonads and brain of Xenopus tropicalis larvae; decreasing brain aromatase activity during their gonadal differentiation and increasing gonadal aromatase activity during metamorphosis. Concentrations of antimycotics have been analyzed (generally by solid phase extraction, SPE, followed by liquid (or gas) chromatography mass spectrometry, LC(GC)–MS/MS) in various aquatic environments and sewage flows (Kolpin et al., 2004; Thomas and Hilton, 2004; Roberts and Bersuder, 2006; Roberts and Thomas, 2006; Peschka et al., 2007; Kahle et al., 2008; Van De Steene and Lambert, 2008). Clotrimazole levels in the Main and Rhine rivers in Germany, and both the raw sewage water and final effluent from a German sewage treatment plant (STP), have been reported to be in the low ng L1 range (Peschka et al., 2007). Comparable concentrations of clotrimazole have also been reported in two UK studies, which assessed sewage water from the Howton STP and samples collected along the Tyne estuary (Thomas and Hilton, 2004; Roberts and Thomas, 2006). Somewhat higher levels of clotrimazole and fluconazole, between 10 and 110 ng L1, have been found in Swiss STP waters, along with fluconazole at levels close to its limit of quantification (LOQ), 1 ng L1, in Swiss lakes (Kahle et al., 2008). In addition, quantifiable levels of miconazole have been found in both surface and sewage waters in the UK (Roberts and Bersuder, 2006), although it was not found at levels exceeding its reporting limit (17.5 ng L1) in an analysis of the contribution of wastewater contaminants to US streams (Kolpin et al., 2004). However, no published studies to our knowledge have analyzed the concentrations of antimycotics in digested sewage sludge. The six substances examined here were chosen to represent a range, in terms of applications and physicochemical properties, of antimycotics commonly used in Sweden. Information on these six substances and the internal standard sulconazole (which is not used in this country) is presented in Table 1. Little is known about the fate and ecotoxic potential of these substances, thus the aim of the presented study was to increase information on their occurrence, levels and fate within STPs. In addition, the collected data were cross-validated against predictive equivalents. To our knowledge, this is the first published study to present national screening data on antimycotics in both sewage water and sludge.
(Deventer, the Netherlands), acetonitrile (HPLC-grade) from Fischer chemicals (Zurich, Switzerland), and sulfuric acid from Merck (Darmstadt, Germany). Water was purified (to 18.2 MU cm resistivity) using a MAXIMA HPLC ultra pure water system (ELGA, High Wycombe Bucks, England). Stock and working (including calibration standards) solutions of individual antimycotics were prepared in methanol and diluted in water, respectively.
2.2. Sample sites, collection and pre-extraction procedure Raw sewage water, final effluent and sludge samples were taken from five Swedish STPs during October and November 2007. The STPs surveyed were Stockholm (Henriksdal), Gothenburg (Ryaverken), Umea˚, Alingsa˚s (Nolhaga), and Bollebygd, which were selected to represent a wide range of STPs in terms of size and catchment area. Sewage water from the Umea˚ county hospital (UCH) was also collected during the same time period. In addition, raw sewage water, raw sewage water particles, raw sludge and digested dewatered sludge were collected from the Umea˚ STP in April 2008 to assess the fate of antimycotics during the treatment process. The locations of the STPs and data on their waste flows and catchment areas are provided in Fig. 1. All of the STPs treat sewage water mechanically, chemically and biologically. Nitrogen is also actively removed at the Henriksdal, Ryaverken and Nolhaga STPs, and the final solid product of all the STPs is anaerobically digested sludge, except at Bollebygd, where only clarifiers are used. Flow-proportional samples of raw sewage water and final effluents were collected from each of the five STPs in each case over a full single day, while grab samples of the UCH sewage water and digested dewatered sludge were collected. After collection, the sewage water (2 L in brown glass bottles) and sludge samples (500 g in brown glass beakers) were immediately stored at 18 C prior to extraction. Grab sampling was also used in the antimycotic fate study at the Umea˚ STP, in which raw sewage water (1 þ 10 L), final effluent (2 L), primary sludge (2 L) and 500 g of digested dewatered sludge samples were collected between 8:30 and 9:00 on each of two consecutive days in April 2008 (except for the digested dewatered sludge, which was only collected on the first of the sampling days due to technical problems). The raw sewage water (1 L) and final effluent samples for SPE were filtered (through 0.45 mm filters) immediately after collection, while the remaining raw sewage water (10 L) and primary sludge samples were centrifuged at 4800 rpm for 20 min, to concentrate the particles sufficiently for analysis (Lindberg et al., 2006), then air-dried for 24 h. All of the samples were then stored at 18 C awaiting extraction.
2.
Experimental
2.3.
2.1.
Chemicals
The aqueous samples were thawed, filtered through 0.45 mm MF membrane filters (Millipore, Sundbyberg, Sweden), acidified to pH 3 with sulfuric acid and 500 ng of the internal standard (IS) sulconazole was added to 500 mL portions of the resulting samples, which were then subjected to SPE using Evolute ABN (6 mL, 200 mg) solid sorbent columns (Biotage,
All of the antimycotics were bought from Sigma Aldrich (Stockholm, Sweden) except for terbinafine, which was obtained from LGC Standards (Bora˚s, Sweden). Formic acid and methanol (HPLC-grade) were purchased from JT Baker
Sample preparation – sewage waters
651
water research 44 (2010) 649–657
Table 1 – Information on the antimycotics investigated in this study. Application
EASb Log Kowc PECd UDe BCFf Usea 1 (kg year ) (ng L1)
N Terbinafine (allylamine)
Topical
43
g
5.81
6
m
5370
Topical
11
g
6.26
2
m
13,183
Topical/peroral 399/22
–/0%
4.45 (4.35)
60
r
447
Peroral
121
80% u 0.25
20
r
3
Topical
67
g
5.61
10
r
4169
Topical
134
g
6.25
20
r
12,882
N N Clotrimazole (imidazole)
Cl N N Ketoconazole (imidazole)
O O N
N
O
O
Cl
Cl
H N N
N
Fluconazole (triazole)
N
N
N
HO
F
F
N N Econazole (imidazole)
Cl
O Cl
Cl N N
Cl Miconazole (imidazole)
O Cl
Cl
Cl N N
IS sulconazole (imidazole)
Cl Not used in Sweden
S Cl
6.46
Cl
a Estimated from Apoteket AB’s Swedish sales during 2007. b Excretion of active substance through urine and/or feces (LIF, 2008). c Log Kow values estimated by KowWin, EPI suite (EPI Suite) (experimental values in parentheses). d Predicted environmental concentrations in surface waters, calculated according to EMEA guidelines Phase 1, except that Swedish sales values were used rather than the supposed penetration factor (Fpen), in combination with the daily defined dose (European Medicines Agency, 2006). e Ultimate degradation (m ¼ months, r ¼ recalcitrant) estimated by BioWin, EPI suite (EPI Suite). f Estimated bioconcentration factor as estimated by BcfWin, EPI suite (EPI Suite). g No excretion due to topical application.
652
water research 44 (2010) 649–657
Fig. 1 – Mass flows of the measured antimycotics (g dayL1) in the raw sewage water, final effluent and digested dewatered sludge of five Swedish sewage treatment plants.
Uppsala, Sweden) that had been conditioned with 6 mL of methanol and 6 mL water adjusted to pH 3 with sulfuric acid. Each sample was applied at a flow rate of 5 mL min1, cleaned using 5 mL of aqueous methanol (5%), and eluted by two washes of 3 mL methanol with 0.5% formic acid. The eluates were evaporated almost to dryness and finally reconstituted with 0.7 mL aqueous acetonitrile (10%).
2.4.
Sample preparation – sludge
The sludge samples were thawed, then 2.0 g portions from each sample were placed in separate glass centrifuge tubes. The IS was added, 500 ng, and the samples were subjected to liquid/solid extraction step by adding 10.0 mL of methanol to each tube and ultrasonically agitating them for 10 min. They were subsequently centrifuged at 4800 rpm for 10 min, and the supernatants were placed in separate 30 mL glass vials. The residues were then re-extracted in the same manner and the two supernatants were combined. The combined extracts (approximately 19 mL) were made up to 200 mL with purified
water, acidified to pH 3, and finally subjected to SPE using the protocol described above for the aqueous samples. The dry weights of the samples were determined by weighing portions after drying them at 105 C for 24 h.
2.5.
Liquid chromatography–mass spectrometry
Sample extracts and calibration solutions (10 mL) were injected using an AS 3000 autoinjector (Thermo Finnigan, San Jose, CA, US) into a Hydrosphere C18 guard column (10 4 mm i.d., 5 mm particle size; YMC Inc., Wilmington, NC, US) followed by an analytical column of the same type (150 4.6 mm i.d., 5 mm particle size). The antimycotics were then eluted by a linear gradient of 90:10 to 70:30 A:B (v/v) over 17 min followed by a further 4-min linear gradient to 30:70 (v/v), where A and B are water and acetonitrile, respectively (both acidified with 0.1% formic acid), at a 0.8 mL min1 flow rate, generated using a P4000 HPLC pump (Spectra system, Thermo Finnigan) at a constant temperature of 25 C. The eluting antimycotics were detected by an LCQ Duo ion trap mass spectrometer (Thermo Finnigan) with an
653
water research 44 (2010) 649–657
electrospray ion source operating in positive ion mode. The source voltage was maintained at a constant 6.0 kV and the heated capillary temperature was set to 250 C. The MS/MS parameters and the collision energies were optimized semiautomatically and manually, respectively.
2.6.
Extraction yield assessment
Absolute extraction yields (including matrix suppression of precursor/product ions during instrumental analysis) were investigated by fortifying 500 ng of each analyte and the IS to raw sewage water (500 mL, n ¼ 4) and digested sludge samples (2.0 g, n ¼ 3). The SPE of the water samples began immediately but the sludge samples were left for an hour prior to the liquid/ solid extraction. The extraction yield was evaluated by a comparison of the analyte and IS peak areas of the samples to the corresponding peak areas in a calibration solution. The same experimental set up was used in order to assess corrected extraction yields with the exception that the IS was added post extraction. The corrected extraction yields were based on peak area ratios (analyte peak area/IS peak area) of the samples to the corresponding peak areas ratios in a calibration solution.
2.7.
Quantification
The antimycotics were identified by retention time, precursor and product ion m/z. The most abundant transitions were monitored and the IS calibration method (peak area ratios of analytes and IS) was used for all substances except fluconazole (external calibration without recovery correction). The ten-point calibration curve, based on 500 ng of sulconazole and ranging from 1 ng mL1 to 1000 ng mL1 of the analytes, was prepared in aqueous acetonitrile (10%).
2.8.
supplied by staff of the STPs. The predicted mass flows (g day1) were based on total sales of the active substances to both institutional and non-institutional care during October and November 2007 (Svensson, 2008), the number of people connected to each STP and the total population of Sweden (Statistics Sweden).
Mass flow calculations
Mass flows (g day1) of each antimycotic substance at each STP were calculated using the obtained concentrations and the aqueous and solid flows during the sample collection periods, and the daily production of digested dewatered sludge (dw) obtained from monthly production figures
3.
Results and discussion
3.1.
Quantification
At least two transitions were monitored for each of the antimycotics (see Table 2), except ketoconazole, for which only one detectable product ion was formed, high CEs only resulted in additional low m/z product ions with weak signal intensities. An approach that could have been used to improve the level of confirmation is the use of the other Cl-isotope of ketoconazole, unfortunately it was not considered in this work. In addition, although two product ions were recorded for fluconazole, signals from one of them were often severely suppressed by matrix components, so it was not used for quantification. In general, matrix suppression of the analyte and IS precursor/product ion signals was immense and in this context it should be noted that the Evolute ABN sorbent column used in the SPE process was designed specifically for quantifying analytes in dirty extracts, since it reduces interference by matrix components such as macromolecules, but the sewage water and sludge extracts were brownish in color, indicating that it failed to remove substantial proportions of low molecular weight substances from these samples. It should also be noted that IS calibration was not used to quantify fluconazole, since its extraction yields were much higher than those of the other antimycotics (including the IS), and it correlated most poorly with the IS in terms of retention time and Log Kow. In addition, the IS was not natively present in the samples, thus an underestimation of the antimycotics was avoided. Terconazole was also considered as an IS, however, traces of it were found in the sewage waters and sludge from the southern part of Sweden, most likely due to its use as a pesticide in the agricultural areas, and it was therefore not used.
Table 2 – Monitored ions, retention times, method detection limits and extraction yields of the antimycotics in aqueous and solid matrices. MS/MS transitions
CEa (AU) tr (min)
MLOQsb
Extraction yields (%)
Waters (ng L1) Sludge (mg kg1) Waters Solids Terbinafine Clotrimazole Ketoconazole Fluconazole Econazole Miconazole Sulconazole (IS)
292.8 > 141.4, 205.4, 170.3 277.2 > 165.5, 242.5 532.9 > 490.7 307.3 > 238.3 383.0 > 125.3, 193.1 417.7 > 161.4, 229.4 399.3 > 330.8, 183.2
31 35 33 23 34 38 22
17.3 17.2 14.9 13.3 18.6 19.8 19.1
5 5 20 50 80 100 –
3 3 10 40 40 50
27 9 23 5 95 104 12 10 6 53 13 10
40 7 34 8 48 15 64 6 53 16 41 12 43 10
IS corrected extraction yields (%) Waters 133 48 117 31 104 21 – 115 21 69 28 –
Solids 109 26 79 6 108 9 – 115 10 95 17 –
a Collision energy in LC–MS/MS (arbitrary units). b Method limits of quantification based on the second calibration curve point in terms of mass per liter (liquid samples) and dry weight (sludge samples).
654
water research 44 (2010) 649–657
3.2. Concentrations and mass flows of the antimycotics in sewage water and sludge The results from the Umea˚ STP antimycotic fate and the national screening study are presented in Table 3. The concentrations of the antimycotics in both the aqueous and solid matrices were generally similar to, or slightly lower than, previously reported levels of antimycotics (Kahle et al., 2008) and other pharmaceuticals, such as antibiotics (Golet et al., 2003; Carballa et al., 2004; Go¨bel et al., 2005a,b; Lindberg et al., 2005, 2006). However, the only substances found at higher levels than their respective LOQs in the aqueous samples were fluconazole (in both the Umea˚ STP and hospital samples) and ketoconazole (in the hospital samples). The levels found in the final effluent were in close correlation with the raw sewage water (just below LOQ in raw sewage water of Stockholm but the chromatographic peak was identified and above three times the signal to noise ratio) hence elimination from the aqueous phase is minimal. Fluconazole was not present in any of the sludge samples thus it is the only one of the monitored antimycotics that will be directly released into recipient waters of the STPs. Kahle et al. came to the same conclusion regarding the fate of fluconazole in their study made at Swiss STPs (Kahle et al., 2008). Three of the other antimycotics (clotrimazole, ketoconazole and econazole) were detected in sludge samples collected from all of the studied STPs. However, miconazole was not detected in any samples
from the Alingsa˚s or Bollebygd STPs, and terbinafine was not detected in samples from the Umea˚ and Alingsa˚s STPs. According to EMEA guidelines (see Table 1), predicted environmental surface water concentrations can only be considered acceptable for fluconazole, and at best misleading for the other substances, since sludge is not included in the equation. It should be noted here that the distribution of the antimycotics between the aqueous and solid phases seems to be consistent with their respective Log Kow values. However, models based on Log Kow are not suitable to predict the environmental fate of pharmaceuticals (Tolls, 2001), and substances with similar Log Kow values have been shown to sorb to sludge to widely differing degrees (Lindberg et al., 2005, 2006). Antimycotic mass flows (g day1) were established to: (a) investigate the absolute distribution of these substances between the aqueous and solid phases, and (b) cross-validate them with mass flows based on sales statistics. The recorded mass flows from the STP screening study are presented together with the predicted values in Fig. 1. The recorded antimycotic mass flows were similar (within an order of magnitude) to their respective predicted equivalents at each STP with two exceptions: a) the measured mass flows in Bollebygd STP were ca. 100-fold lower than predicted, possibly because it is the smallest of the investigated STPs, and that Bollebygd is a small town lacking both a hospital and large industries; and b) the much lower
Table 3 – Sewage water and sludge antimycotic concentrations. National screening studya,b STP Um RSW, ng L Terbinafine Clotrimazole Ketoconazole Fluconazole Econazole Miconazole
d d d 90 d d
STP St
STP Al
STP Go
STP Bo
d d d 140 d d
RSW, ng L1 d d d d d d
d d d d d d
d d d d d d
d d d d d d
d 30 910 d 590 160
d d 30 570 d d
d d d 120 d d FE, ng L1
d d d 100 d d
d d d d d d
d d d d d d
d d d d d d
DDS, mg kg1 (dw) Terbinafine Clotrimazole Ketoconazole Fluconazole Econazole Miconazole
UCH
1
FE, ng L1 Terbinafine Clotrimazole Ketoconazole Fluconazole Econazole Miconazole
Fate in Umea˚ STP studya,c
4 120 910 d 1000 970
d 120 480 d 660 d
7 110 670 d 680 190
30 50 280 d 210 d
d d d 100 d d RSP, mg kg1 (dw)
RS, mg kg1 (dw)
DDS, mg kg1 (dw)
d 10 980 d 470 d
40 310 1300 d 240 d
d 60 1800 d 550 370
a RSW ¼ raw sewage water, FE ¼ Final effluent, DDS ¼ digested dewatered sludge, RSP ¼ raw sewage particles, and RS ¼ raw sludge. b STP Um ¼ Umea˚; STP St ¼ Stockholm, Henriksdal; STP Al ¼ Alingsa˚s; STP Go ¼ Gothenburg, Ryaverken; STP Bo ¼ Bollebygd; UCH ¼ Umea˚ county hospital (Norrlands Universitetssjukhus). c Mean concentrations from two days of sampling. d Below LOQ.
water research 44 (2010) 649–657
observed mass flow of terbinafine, which suggests that this topically used substance (of the ones included) is only to a small extent present in the sewage waters post application. In Umea˚, the use of ketoconazole and fluconazole in the institutional care accounts for ca. 23% and 56% of the total amounts sold, respectively. However, while the mass flow of fluconazole in the UCH sewage water represented 31% of the total mass flow in the STP raw sewage water, a substantially smaller proportion of the ketoconazole of UCH (1%) was found in the solid flows of the STP (digested dewatered sludge). In addition, mean mass flow values normalized to the number of people connected to each STP indicated that 53% of the purchased fluconazole eventually appeared in the final effluents, while 1, 35, 41, 155 and 209% of terbinafine, ketoconazole, miconazole, clotrimazole and econazole, respectively, appeared in digested dewatered sludge. In addition, the relative distribution between the antimycotics based on sales statistics (total sales nonand institutional care) in comparison to the observed results in dewatered digested sludge, with the exception of final effluents for fluconazole, can be seen in Fig. 2. Although the comparison would be more accurate by using raw sewage water and raw sewage particles and sales statistics of each catchment area of the STPs, the posttreatment findings at the STPs show similar distribution of the antimycotics as to the predicted equivalent.
655
Fig. 2 – The relative distribution of antimycotics based on sales records compared to observed results. Abbreviations: STP Um [ Umea˚; STP St [ Stockholm, Henriksdal; STP Al [ Alingsa˚s; STP Go [ Gothenburg, Ryaverken; STP Bo [ Bollebygd; STP mean [ mean value of the distribution of the antimycotics.
3.3.
Fate analysis within Umea˚ sewage treatment plant
The mass flow results from the fate analysis in Umea˚ STP are given in Fig. 3. Again, fluconazole was not detected in any of the solid matrices, and its mass flows in the raw sewage water
Fig. 3 – Mass flows of antimycotics (g dayL1) within the Umea˚ sewage treatment plant.
656
water research 44 (2010) 649–657
and the final effluent were very similar (removal efficiency ca. 13%). Thus, these results corroborate the indications from the screening study that fluconazole is at best marginally affected by current sewage water treatments. In contrast, clotrimazole, ketoconazole and econazole were all present in the solid flow of the raw sewage water, and they were also, together with terbinafine, detected in the raw sludge. The results for clotrimazole are consistent with data presented by Kahle et al. (2008), who found extractable amounts of this substance in the suspended solids of raw sewage water of Swiss STPs. Clotrimazole, ketoconazole, econazole and terbinafine were not strongly affected by sludge digestion either, and all of them (apart from terbinafine) were found at levels exceeding their respective LOQs in the resultant sludge. Traces of miconazole were also observed post-digestion. Our results show that fluconazole is released at significant levels into the aquatic environment, and although this substance is reportedly the least effective CYP19 aromatase inhibitor of the azole antimycotics (Tro¨sken et al., 2004), it is a moderate Cyp3A4 and a potent Cyp2C9 inhibitor, and thus may have significant direct ecotoxicological effects. The possibility that it may also have adverse indirect environmental effects cannot be excluded, since (for example) ketoconazole has been shown to increase the sensitivity of rainbow trout to ethynylestradiol (Hasselberg et al., 2008). However, sludge is the major route for antimycotics into the environment, and thus their exposure levels and potential negative effects will depend on the fate of the sludge. Sludge is commonly used as a fertilizing agent (e.g. for golf courses and forests) or as landfill material, so we recommend that future environmental risk assessments of this group of pharmaceuticals, not only focus on the aqueous environment, but also consider their potential impact on terrestrial organisms.
4.
Conclusions
The levels of fluconazole in the raw sewage water were similar to those of the final effluent and in addition it was not observed in digested dewatered sludge. Terbinafine, clotrimazole, ketoconazole, econazole and miconazole were only present in sludge samples, in turn, not strongly affected by sludge digestion. Observed antimycotic mass flows were in general similar to their respective predicted equivalents at each STP except for terbinafine and for the smallest STP included in this study. Future environmental risk assessments of antimycotics should consider both the aqueous and the terrestrial environment.
Acknowledgements We gratefully acknowledge financial support from the Swedish Research Council for Environment, Agricultural Sciences and Spatial Planning (FORMAS). We also thank the personnel at the sewage treatment plants for their assistance.
references
Carballa, M., Omil, F., Lema, J.M., Llompart, M., Garcia-Jares, C., Rodriguez, I., Gomez, M., Ternes, T., 2004. Behaviour of pharmaceuticals, cosmetics and hormones in a sewage treatment plant. Water Res. 38, 2918–2926. EPI Suite was Developed by the EPA’s Office of Pollution Prevention Toxics and Syracuse Research Corporation (SRC), 2000. U.S. Environmental Protection Agency. European Medicines Agency, 2006. Pre-authorisation Evaluation of Medicines for Human Use. http://www.emea.europa.int/ pdfs/human/swp/444700en.pdf (accessed 11.12.2008). Go¨bel, A., Thomsen, A., McArdell, C.S., Joss, A., Giger, W., 2005a. Occurrence and sorption behaviour of sulfonamides, macrolides, and trimethoprim in activated sludge treatment. Environ. Sci. Technol. 39, 3981–3989. Go¨bel, A., McArdell, C.S., Suter, M.J.-F., Giger, W., 2005b. Trace determination of macrolide and sulfonamide antimicrobials, a human sulfonamide metabolite, and trimethoprim in wastewater using liquid chromatography coupled to electrospray tandem mass spectrometry. Anal. Chem. 76, 4756–4764. Golet, E.M., Xifra, I., Siegrist, H., Alder, A.C., Giger, W., 2003. Environmental exposure assessment of fluoroquinolone antibacterial agents from sewage to soil. Environ. Sci. Technol. 37, 3243–3249. Graybill, J.R., Drutz, D.J., 1980. Ketoconazole – a major innovation for treatment of fungal disease. Ann. Int. Med. 93, 921–923. Guengerich, P.F., 2008. Cytochrome P450 and chemical toxicology. Chem. Res. Toxicol. 21, 70–83. Gunnarsson, L., Jauhiainen, A., Kristiansson, E., Nerman, O., Larsson, D.G.J., 2008. Evolutionary conservation of human drug targets in organisms used for environmental risk assessments. Environ. Sci. Technol. 42, 5807–5813. Gyllenhammar, I., Eriksson, H., So¨derqvist, A., Lindberg, R.H., Fick, J., Berg, C., 2009. Clotrimazole exposure modulates aromatase activity in gonads and brain during gonadal differentiation in Xenopus tropicalis frogs. Aquat. Toxicol. 91, 102–109. Hasselberg, L., Westerberg, S., Wassmur, B., Celander, M.C., 2008. Ketoconazole, an antifungal imidazole, increases the sensitivity of rainbow trout to 17a-ethynylestradiol exposure. Aquat. Toxicol. 86, 256–264. Kahle, M., Buerge, I.J., Hauser, A., Mu¨ller, M.D., Poiger, T., 2008. Azole fungicides: occurrence and fate in wastewaters and surface waters. Environ. Sci. Technol. 42, 7193–7200. Kolpin, D.W., Skopec, M., Meyer, M.T., Furlong, E.T., Zaugg, S.D., 2004. Urban contribution of pharmaceuticals and other organic wastewater contaminants to streams during differing flow conditions. Sci. Total Environ. 328, 119–130. Lepesheva, G.I., Waterman, M.R., 2007. Sterol 14 alpha-demethylase cytochrome P450 (CYP51), a P450 in all biological kingdoms. Biochim. Biophys. Acta-General Subjects 1770, 467–477. LIF, 2008. La¨kemedelsindustrifo¨reningen (the Swedish association of the pharmaceutical industry). FASS. Elanders, Kungsbacka. Lindberg, R.H., Wennberg, P., Johansson, M.I., Tysklind, M., Andersson, B.A.V., 2005. Screening of human antibiotic substances and determination of weekly mass flows in five sewage treatment plants in Sweden. Environ. Sci. Technol. 39, 3421–3429. Lindberg, R.H., Olofsson, U., Rendahl, P., Johansson, M.I., Tysklind, M., Andersson, B.A.V., 2006. Behavior of fluoroquinolones and trimethoprim during mechanical, chemical and active sludge treatment of sewage water and digestion of sludge. Environ. Sci. Technol. 40, 1042–1048.
water research 44 (2010) 649–657
Peschka, M., Roberts, P.H., Knepper, T.P., 2007. Analysis, fate studies and monitoring of the antifungal agent clotrimazole in the aquatic environment. Anal. Bioanal. Chem. 389, 959–968. Piferrer, F., Zanuy, S., Carillo, M., Solar, I.I., Devlin, R.H., Donaldson, E.M., 1994. Brief treatment with an aromatase inhibitor during sex differentiation causes chromosomally female salmon to develop as normal, functional males. J. Exp. Zool. 270, 255–262. Roberts, P.H., Bersuder, P., 2006. Analysis of OSPAR priority pharmaceuticals using high-performance liquid chromatography–electrospray ionisation tandem mass spectrometry. J. Chromatogr. A 1134, 143–150. Roberts, P.H., Thomas, K.V., 2006. The occurrence of selected pharmaceuticals in wastewater effluent and surface waters of the lower Tyne catchment. Sci. Total Environ. 356, 143–153. Ryder, N.S., 1992. Terbinafine: mode of action and properties of the squalene epoxidase inhibition. Br. J. Dermatol. 126 (Suppl. 39), 2–7. Statistics Sweden. http://www.scb.se/templates/ tableOrChart____78315.asp (accessed 02.10.08). Svensson, Elin, 2008. Apoteket AB, Sweden. Personal communication.
657
Sweetman, S.C. (Ed.), 2007. Martindale: The Complete Drug Reference. Pharmaceutical Press, London Electronic version, Edition 070214. Thomas, K.V., Hilton, M.J., 2004. The occurrence of selected human pharmaceutical compounds in UK estuaries. Mar. Pollut. Bull. 49, 436–444. Tolls, J., 2001. Sorption of veterinary pharmaceuticals in soils: a review. Environ. Sci. Technol. 35, 3397–3406. Tro¨sken, E.R., Scholz, K., Lutz, R.W., Volkel, W., Zarn, J.A., Lutz, W. K., 2004. Comparative assessment of the inhibition of recombinant human CYP19 (aromatase) by azoles used in agriculture and as drugs for humans. Endocr. Res. 30, 387–394. Van De Steene, J.C., Lambert, W.E., 2008. Validation of a solidphase extraction and liquid chromatography–electrospray tandem mass spectrometric method for the determination of nine basic pharmaceuticals in wastewater and surface water samples. J. Chromatogr. A 1182, 153–160. Yoshida, Y., Aoyama, Y., Noshiro, M., Gotoh, O., 2000. Sterol 14demethylase P450 (CYP51) provides a breakthrough for the discussion on the evolution of cytochrome P450 gene superfamily. Biochem. Biophys. Res. Commun. 273 (3), 799–804.
water research 44 (2010) 658–668
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Pharmaceuticals and personal care products in archived U.S. biosolids from the 2001 EPA national sewage sludge survey Kristin McClellan, Rolf U. Halden* Center for Environmental Biotechnology, The Biodesign Institute at Arizona State University, 1001 S. McAllister Avenue, Tempe, AZ 85287-5701, USA
article info
abstract
Article history:
In response to the U.S. National Academies’ call for a better assessment of chemical
Received 1 July 2009
pollutants contained in the approximately 7 million dry tons of digested municipal sludge
Received in revised form
produced annually in the United States, the mean concentration of 72 pharmaceuticals and
15 December 2009
personal care products (PPCP) were determined in 110 biosolids samples collected by the
Accepted 21 December 2009
U.S. Environmental Protection Agency (EPA) in its 2001 National Sewage Sludge Survey.
Available online 4 January 2010
Composite samples of archived biosolids, collected at 94 U.S. wastewater treatment plants from 32 states and the District of Columbia, were analyzed by liquid chromatography
Keywords:
tandem mass spectrometry using EPA Method 1694. Thirty-eight (54%) of the 72 analytes
Municipal sludge
were detected in at least one composite sample at concentrations ranging from 0.002 to
Organic wastewater contaminants
48 mg kg1 dry weight. Triclocarban and triclosan were the most abundant analytes with
Risk assessment
mean concentrations of 36 8 and 12.6 3.8 mg kg1 (n ¼ 5), respectively, accounting for
Land application
65% of the total PPCP mass found. The loading to U.S. soils from nationwide biosolids recycling was estimated at 210–250 metric tons per year for the sum of the 72 PPCPs investigated. The results of this nationwide reconnaissance of PPCPs in archived U.S. biosolids mirror in contaminant occurrences, frequencies and concentrations, those reported by the U.S. EPA for samples collected in 2006/2007. This demonstrates that PPCP releases in U.S. biosolids have been ongoing for many years and the most abundant PPCPs appear to show limited fluctuations in mass over time when assessed on a nationwide basis. The here demonstrated use of five mega composite samples holds promise for conducting cost-effective, routine monitoring on a regional and national basis. ª 2009 Elsevier Ltd. All rights reserved.
1.
Introduction
Pharmaceuticals and personal care products (PPCPs) are common contaminants of the environment, and have been detected in surface water (Kolpin et al., 2002; Cahill et al., 2004; Moldovan, 2006; Roberts and Thomas, 2006; Tamtam et al., 2008), groundwater (Heberer et al., 2000; Lindsey et al.,
2001; Fick et al., 2009), and drinking water (Stackelberg et al., 2004; Loraine and Pettigrove, 2006; Loos et al., 2007; Focazio et al., 2008), as well as in agricultural soils subject to land application of digested municipal sludge (Kinney et al., 2008; Kupper et al., 2004; Wu et al., 2009), also known as biosolids. Wastewater treatment plants were identified as one possible source for surface water contamination. Over-the-counter
Abbreviations: DL, Detection limit; EPA, Environmental protection agency; NEBRA, North east biosolids and residuals association; NSSS, National sewage sludge survey; PPCP, Pharmaceuticals and personal care products; RPD, Relative percent difference; TNSSS, Targeted national sewage sludge survey. * Corresponding author. Tel.: þ1 480 727 0893; fax: þ1 480 727 0889. E-mail addresses:
[email protected] (K. McClellan),
[email protected] (R.U. Halden). 0043-1354/$ – see front matter ª 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2009.12.032
water research 44 (2010) 658–668
and prescription drugs enter the wastewater via excretion of urine and feces containing parental drugs and their conjugates as well as other metabolites, or from disposal of unwanted or expired medications (Halling-Sorensen et al., 1998; Fent et al., 2006). Similarly, chemical constituents of personal care products may be directly disposed of into domestic wastewater. Removal of PPCPs during municipal wastewater treatment is rarely complete, thereby creating a pathway for entry of these compounds into aquatic environments via wastewater reclamation (Halling-Sorensen et al., 1998; Ternes, 1998; Daughton and Ternes, 1999; Hirsch et al., 1999) and into terrestrial environments via land application of biosolids (Ternes et al., 2004a). Of the more than 7 million tons of sewage sludge produced in the United States in 2004, about 50% was applied to land as fertilizer or soil amendment, and 45% was disposed of in landfills or as landfill cover (NEBRA, 2007). Terrestrial environments can offer effective biological, physical, and chemical attenuation mechanisms for manmade pollutants. However, they also can act as a source term for chemical migration into surface and groundwater from biosolids runoff and leachate. Pharmaceutical compounds are designed to be biologically active and therefore may have effects on non-target organisms even at trace concentrations extant in terrestrial and aquatic environments. While acute toxic effects of pharmaceuticals on non-target organisms have been investigated for some compounds, chronic toxicity and potential subtle environmental effects are only scarcely known (Fent et al., 2006). Also insufficiently investigated is the effect of mixtures of pharmaceuticals on aquatic organisms, although biochemical interactions of drugs in humans are well known. Of additional concern is the possible uptake of contaminants into food crops grown on agricultural fields that were fertilized with biosolids (Kumar et al., 2005; Dolliver et al., 2007). Currently, no regulation exists in the United States for PPCPs contained in biosolids, and a need for more information on the occurrence of and risk from these compounds has been noted by the National Research Council of the National Academies of the United States (National Research Council, 2002). In the past, analytical methods were limited, especially for trace analyses of complex environmental samples. In 2007, the release of U.S. EPA method 1694 (USEPA, 2007a) for the analysis of PPCPs in various matrices afforded the opportunity to analyze biosolids samples using a standardized protocol. The U.S. EPA has performed national sewage sludge surveys (NSSS) in 1989, 2001, and 2007. The survey conducted in 2001 served to evaluate the potential need for regulations of trace levels of dioxins and polychlorinated biphenyls (USEPA, 2007b). After the 2001 survey was completed, unused samples were released to a nationwide repository of biosolids samples now maintained at the Biodesign Institute at Arizona State University. This investigation evaluated occurrences and concentrations of PPCPs in biosolids from the year 2001 to enable risk assessments and to establish a national baseline for evaluating temporal trends of PPCPs in U.S. biosolids. The analysis of composite samples by EPA Method 1694 was employed to determine average concentrations of 72 PPCPs in archived biosolids collected by the EPA, as a representative sample of the more than 16,000 treatment plants located in the contiguous United States.
2.
Materials and methods
2.1.
Sampling procedure
659
Biosolids samples with solid contents between 1% and 30% were obtained from 94 wastewater treatment plants in 32 states and the District of Columbia for the 2001 National Sewage Sludge Survey (USEPA, 2007b). They were selected by the U.S. EPA to obtain a representative estimate of the occurrence of chemical contaminants in sewage sludge that is disposed of primarily by land application. Information on the exact sampling locations is available in Table S1 of the supplementary material (SM). This survey aimed to estimate levels of dioxins, dibenzofurans, and coplanar polychlorinated biphenyls in biosolids. The sampling was conducted by the U.S. EPA between February and March 2001, and samples were collected according to sampling procedures developed by the U.S. EPA (USEPA, 2001). Samples were only taken from fully processed sewage sludges intended for disposal. Eighty-nine of the 94 WWTPs had one single system for sludge treatment, therefore one sample was collected. Five facilities had two systems for treating their sludges, therefore two samples were taken from each of these plants. In addition, duplicate samples were collected from 15% of facilities (14 samples) for precision analysis. This amounted to 113 samples overall. After completion of the 2001 NSSS, the samples were acquired by the Halden laboratory for further studies. For the 7-year period between acquisition and analysis in the spring of 2008, samples were stored at 20 C.
2.2.
Composite sample preparation
From the 113 biosolids samples acquired from the EPA, three were excluded from analysis because the sample containers were broken or compromised; the remaining 110 samples were randomly grouped into five groups. Composite samples were prepared by weighing out approximately 1 g of dry weight from each sample and pooling it to obtain 5 composites each containing solids from between 21 and 24 individual samples. A duplicate of composite sample #3 was prepared to serve as a blind duplicate.
2.3.
Sample analysis
The samples were analyzed by AXYS Analytical Services (2045 Mills Road West, Sydney, British Columbia, Canada V8L 3S8) according to EPA method 1694 (USEPA, 2007a). For the purpose of compound detection, the 72 analytes were divided into four groups. All analytes were separated by liquid chromatography and detected by tandem mass spectrometry. For compounds with a respective labeled analog, the concentration was determined using the isotope dilution technique. The corresponding concentrations are deemed to be of high quality. For compounds where a labeled analog was not available, the concentration was determined using an external calibration. Quantitative data for analytes not determined by the isotope dilution method are judged to be less robust. More detailed information on the analysis method and accuracy and precision criteria for acceptance of analytical data is available in
660
water research 44 (2010) 658–668
supplementary material (SM). Further information on study limitations can be found in the discussion section.
2.4.
Quality assurance
2.6.
To ensure system and laboratory performance, several tests were performed before sample analysis. Calibration accuracy was verified using a calibration standard solution with labeled and native analytes. Retention times of native and labeled compounds had to be within 15 seconds of the respective retention time established during the previous calibration. In addition, ongoing precision and recovery were ensured. Lab blanks were analyzed before each sample analysis. A duplicate sample analysis was performed by the lab for each batch between 7 and 20 samples. In addition to these standard procedures, a blind duplicate was included in the sample set to evaluate analysis precision. Precision is expressed as relative percent difference (RPD) between each pair of measured concentrations. It was calculated using the following equation, jCsample Cduplicate j 100 RPD½% ¼ Csample þ Cduplicate 2
(1)
where C sample and C duplicate are, respectively, the concentration detected in the original sample and in its duplicate sample, and RPD is the relative percent difference.
2.5.
Modeling of soil and porewater concentration
For assessment of their potential environmental impact, the concentrations of PPCPs after mixing with agricultural soil were calculated. The mixing ratio with soil was assumed based on the EPA-recommended rate of biosolids application of up to 4.5 dry kg per m2 and by further assuming incorporation into soil to a depth of 10 cm (USEPA, 1994). The soil bulk density was assumed to equal 1.3 g cm3 and the biosolids bulk density 1.6 g cm3. Though soil moisture content is known to vary widely depending on soil properties, for the purpose of this calculation, a soil moisture content of 22% (v/v) was used, as reported by others for agricultural soil (De Lannoy et al., 2006). A typical soil organic carbon fraction was assumed (Causarano et al., 2008), as was a biosolids organic carbon fraction of 0.4 (USEPA, 2007b). The concentration in soil and porewater at equilibrium was calculated based on the organic carbon fraction of the soil-biosolids mixture and the compound specific organic-carbon distribution coefficient (KOC) of each analyte (Chalew and Halden, 2008). Calculations took into consideration the soil moisture content, as well as the volume of biosolids added to the soil, and were conducted for an environmentally relevant pH range of 7–9. The concentrations of PPCPs in porewater of soil/biosolid mixtures were calculated using the equations below,
Cporewater
Csoil ¼
mbiosolids Cbiosolids msoilþbiosolids ¼ fporewater rporewater þ KOC fOC msoilþbiosolids
fporewater rporewater mbiosolids Cbiosolids Cporewater msoilþbiosolids msoilþbiosolids
and fOC the dimensionless fractions of, respectively, porewater and organic carbon in the soil/biosolids mixture.
(2)
(3)
where m is the dry mass in kg m3 of the solid matrix, C the concentration in mg kg1, r the density in kg m3, and fporewater
Drug usage and ecotoxicity data
Information of prescription drug consumption was obtained from Internet sources (www.drugtopics.com). Exotoxicity data were taken from EPA’s Ecotox database (www.epa.gov/ ecotox).
2.7.
Modeling of annual loading to agricultural soil
The annual loading of PPCPs contained in biosolids was based on a production of 5.6–7 million dry tons of sewage sludge in the U.S., of which 50–60% is applied to land (National Research Council, 2002; Jones-Lepp and Stevens, 2007; NEBRA, 2007).
3.
Results
3.1.
Data quality assurance
Lab blanks showed no detections above the detection limit for any of the analytes except for ciprofloxacin [61 mg kg1 dry weight (dw); detection limit (DL) 32 mg kg1 dw] and erythromycin-H2O (2.6 mg kg1 dw; DL 1.5 mg kg1 dw)]. However, concentrations detected in biosolids were 100- and 40-times greater than the background level detected in lab blanks. Therefore, measured concentrations for both analytes were accepted. Recovery for all analytes typically was good with an average of 112% but some notable outliers were observed leading to a range of 12–493% (See Table 1 and Table S2, SM). A total of 10 analytes exceeded the methods’ lower and upper control limits. For 5 analytes (anhydrotetracycline, azithromycin, cimetidine, 4-epianhydrochlortetracycline, and 4-epianhydrotetracycline) recovery rates were below the method’s lower control limit of 40%. The concentrations reported for these analytes in biosolids may represent underestimates. The recovery rates of 5 analytes (clinafloxacin, enrofloxacin, lomefloxacin, ofloxacin, and sarafloxacin) were above the method’s upper control limits. The respective concentrations reported for these analytes may represent overestimations. All of the above mentioned analytes were quantified using labeled surrogate standards. In the case of clinafloxacin, enrofloxacin, lomefloxacin, ofloxacin, and sarafloxacin, 13C3-15N-ciprofloxacin was used as an isotope-labeled surrogate standard. The recoveries were determined from spiked quality control samples, where recovery of 13C3-15N-ciprofloxacin was below the method’s lower control limit. This led to extremely high recoveries of the above mentioned analytes. In the biosolids samples, recovery of 13C3-15N-ciprofloxacin was always within the method’s control limits. However, the lack of adequate recovery of 13C3-15N-ciprofloxacin in quality control samples suggests that concentrations reported in biosolids for clinafloxacin, enrofloxacin, lomefloxacin, ofloxacin, and sarafloxacin should be interpreted with caution. The duplicate analysis revealed 18% relative percent difference (RPD) for all analytes. The blind duplicate analysis of a subset of 10 analytes revealed 28% RPD. Both RPD values
Table 1 – Analytical results and summary statistics for pharmaceuticals and personal care products detected in U.S. biosolids collected in 2001 and reported on a dry weight basis. Substance name
CAS RN
Detection limit [mg kg1]
Frequency Standard measured deviation [%] of detection limit [mg kg1] n¼5
Maximum Mean detected concentration concentration 1 [mg kg ] [mg kg1]
Standard deviation of mean concentration [mg kg1] n ¼ 5
Recovery [%]
Isotope dilution quantification
Use
Anhydrotetracyclineg Azithromycin g
4496-85-9 83905-01-5
74.1 5.6
7.4 1.6
100 100
880 1220
392 838
249 224
51 12
Antibiotic Antibiotic
Caffeine
58-08-2
59.0
16.4
100
643
248
200
78
Stimulant
Carbamazepine
298-46-4
5.6
1.6
100
238
163
56.4
139
Anticonvulsant
Chlortetracycline
57-62-5
22.9
6.2
60
23.4
16.9
159
Antibiotic
Cimetidine g
51481-61-9
6.4
2.0
100
43.5 893 10800
504
Ciprofloxacin
85721-33-1
20.9
6.4
100
Clarithromycin Codeine
81103-11-9 76-57-3
5.6 11.2
1.6 3.2
100 20
94.6 29.7
66.2 n.d.
Cotinine
486-56-6
9.9
3.3
100
38.6
Diltiazem Diphenhydramine Doxycycline Enrofloxacinh
42399-41-7 58-73-1 564-25-0 93106-60-6
1.3 2.3 23.9 19.5
0.1 0.6 6.3 6.5
100 100 100 60
4-Epianhydrotetracycline g 4-Epichlortetracycline 4-Epitetracycline
4465-65-0
77.1
20.2
100
14297-93-9
56.3
16.0
40
23313-80-6
74.4
18.8
100
25.5
114 114
Antibiotic Analgesic
28.1
8.3
103
109 1740 1780 28.6
45.2 1166 966 n.d.
34.2 516 436
110 101 85 190
399
261
71.7
n.d. 2376
517
0.05 a (Bantle et al., 1994)
830–1000
100 b(Jos et al., 2003) 36 c(Brain et al., 2004)
550–680 80–100 1700–2100
5d(HallingSorensen, 2001)
23,000–28,000 220–270
Nicotine
100–120
metabolite Antianginal Antihistamine Antibiotic Antibiotic
39
Antibiotic
161
Antibiotic
82
Antibiotic
Projected annual land application [kg yr1]f
1300–1600 2800–3500
Antacid Antibiotic
3040
2348
41 98
93.0
6858
208
Lowest effect concentration for aquatic biota [mg L1]
150–190 3900–4800 3200–4000 49 d(Robinson et al., 2005) 900–1100
8000–9800
Erythromycin-H2O
114-07-8
1.6
0.2
100
183
81.5
52.3
97
Antibiotic
22 700 c (Williams et al., 1992)
Fluoxetine
54910-89-3
8.2
1.0
100
258
171
46.6
89
Antidepressant
36h, c(Flaherty and Dodson, 2005)
580–710
Gemfibrozil
25812-30-0
12.2
12.3
100
159
152
13.2
107
Antihyperlipidemic
30,040 c(Zurita et al., 2007)
510–630
Ibuprofen
15687-27-1
80
359
246
109
Anti-inflammatory
Isochlortetracycline Lomefloxacinh
514-53-4 98079-51-7
70 388
Antibiotic Antibiotic
Metformin
657-24-9
Antidiabetic
1000–1300
Miconazole Minocycline
22916-47-8 10118-90-8
Antifungal Antibiotic
2600–3200 6300–7800
122 22.5 13.2
123
36.0 16.1
121
6.4 2.1
60 40
n.d. n.d.
119
32.5
80
456
305
152
116
5.8 1880
1.3 524
100 80
1100 2630
777 1884
266 939
86 54
270–340
830–1000
106 e(Robinson et al., 2005)
(continued on next page)
Table 1 (continued) Substance name
Naproxen
CAS RN
22204-53-1
Detection limit [mg kg1]
24.3
Frequency Standard measured deviation [%] of detection limit [mg kg1] n¼5 24.6
100
Maximum Mean detected concentration concentration [mg kg1] [mg kg1]
273
119
Norfloxacin
70458-96-7
63.5
6.3
100
418
289
Ofloxacinh
82419-36-1
7.7
4.2
100
8140
5446
Oxytetracyclin
79-57-2
Ranitidine
66357-35-5
22.8
5.8
100
6.5
1.7
100
114 30.1
1941
136
Antibiotic
493
Antibiotic
21.0
8.0
51
Antacid
Projected annual land application [kg yr1]f
400–500 b
18 000 (Yang et al., 2008) 21 d(Robinson et al., 2005) 50 d(Hanson et al., 2006)
970–1200 18000–23,000 300–360 70–90
9 e(Brain et al., 2008)
98
Antibiotic
691
101 124
Antibiotic Antibiotic
131
103
Fungicide
36060
8049
96
Disinfectant
12640
3816
105
Disinfectant
0.12 b(Wilson et al., 2003)
42,000–52,000
Antibiotic
16 000 b(Luetzhoft et al., 1999)
90–110
20
3.3
Sulfanilamide Tetracycline
63-74-1 60-54-8
56.2 57.5
16.0 15.5
40 100
87.3 2790
n.d. 1914
Thiabendazole
148-79-8
5.9
1.2
100
370
110
Triclocarban
101-20-2
183
67.0
100
48100
Triclosan
3380-34-5
487
100
19700
Trimethoprim
738-70-5
60.5
74.0
Lowest effect concentration for aquatic biota [mg L1]
Antiinflammatory
Antibiotic
0.7
60
106
Use
90
2.9
3.8
79
Isotope dilution quantification
22.2
723-46-6
10.6
Recovery [%]
87.5
Sulfamethoxazole
493
Standard deviation of mean concentration [mg kg1] n ¼ 5
n.d.
26.0
21.5
97
47 e(Brain et al., 2004)
6400–7900
310 c(EPA, 2000)
370–460
0.101 c (EPA/OTS, 1992)
120,000–150,000
a Amphibium. b Green algae. c Crustacean. d Cyanobacteria. e Macrophyte. f Projected deposition rates of PPCPs on land based on 5.6–6.9 million dry tons of annual sewage sludge production and a land application rate of 60% (National Research Council, 2002; USEPA, 2003; USGS, 2006; Jones-Lepp and Stevens, 2007). g Concentration may be underestimated due to low recovery in quality control samples. h Concentration may be overestimated due to high recovery in quality control samples.
663
water research 44 (2010) 658–668
were in control with respect to the target RPD of 30% or less. The RPD value improved to 11% for 9 analytes, when excluding results for metformin, whose lack of detection in the blind duplicate unfavorably increased the summary statistic for measurement precision. A non-blinded duplicate analysis, performed by the contract laboratory for these 10 analytes, showed an RPD value of 11%.
3.2.
not detected in any of the composite biosolids samples. The mean total concentration of all targeted PPCPs combined in the five composite samples was 74.4 mg kg1 dw of sewage sludge 21.4 mg kg1 standard deviation (n ¼ 5). The two most abundant contaminants were the disinfectants triclocarban (48% of total detected PPCP mass) and triclosan (17%) (Fig. 1). Their mean concentrations were 36 8 and 12.6 3.8 mg kg1 dw (n ¼ 5), respectively (Table 1). The second most abundant class of PPCPs found was antibiotics. In order of decreasing concentration, ciprofloxacin, ofloxacin, 4epitetracycline, tetracycline, minocycline, doxycycline and azithromycin were found at concentrations between 6.8 2.3 and 0.8 0.2 mg kg1 dw (n ¼ 5) (Table 1). The combined mass of all antibiotics constituted about 29% of the total mass of PPCPs per sample. In addition to the 3 tetracyclines reported above, three tetracycline antibiotics were found for which peer-reviewed occurrence data in sewage sludge thus far were lacking. Anhydrotetracycline, 4-epianhydrotetracycline, and chlortetracycline (in order of decreasing concentration) together accounted for 6.9 2.5 mg kg1 dw (n ¼ 5) (Table 1). Also not previously reported in biosolids were two prescription drugs, metformin and ranitidine. They together accounted for 0.4% of the total mass of PPCPs found in sewage sludge composites.
Study representativeness and sample integrity
The prolonged storage of samples between sampling event and analysis may have allowed for the chemical degradation of labile analytes to occur. Therefore, any results of this study are conservative with respect to the detection frequency and concentration of compounds found. In other words, due to pooling of a large number of samples, analytes occurring infrequently and at low concentrations may have been diluted out to below the detection limit. A comparison of the presented data with the EPA data (USEPA, 2009), reveals that the mean concentrations of all analytes show no statistically significant difference within the 95th percentile confidence interval. Therefore, the prolonged storage did not impair the detection of multiple analytes at elevated concentrations in archived samples.
3.3.
Occurrence of PPCPs in biosolids
All composites tested positive for at least 26 analytes. Of the 72 PPCPs targeted, 38 (54%) were detected at concentrations ranging from the low parts-per-billion (ppb) to the parts-permillion (ppm) range. These 38 PPCPs include 8 that have not previously been reported in biosolids in the peer-reviewed literature but that also were observed in the U.S. EPA’s TNSSS published online (USEPA, 2009). The remaining 34 PPCPs were
4.
Discussion
4.1.
Study limitations
For this study, a relatively large number of individual samples were combined to form 5 composite pools or mega composites. This approach served to reduce the number of samples to be analyzed in order to obtain a defensible estimate of mean
10 000
ppm 1000
ppb 100
sulfamethoxazole *
codeine *
lomefloxacin ‡*
enrofloxacin ‡*
ranitidine
isochlortetracycline *
chlortetracycline *
cotinine
trimethoprim *
4-epichlortetracycline
diltiazem
sulfanilamide *
clarithromycin
oxytetracycline
erythromycin-H2O
naproxen
thiabendazole
gemfibrozil
fluoxetine
carbamazepine
caffeine
ibuprofen *
4-epianhydrotetracycline ‡
metformin *
norfloxacin
cimetidine ‡
anhydrotetracycline ‡
miconazole
doxycycline
azithromycin ‡
diphenhydramine
tetracycline
minocycline *
ofloxacin ‡
triclosan
ciprofloxacin
1
4-epitetracycline
10
triclocarban
Concentration in biosolids [µg kg-1 dry weight]
100 000
Fig. 1 – Rank order of mean concentrations for 38 PPCPs detected in composites of a total of 110 U S biosolids samples from 94 treatment plants in 32 states and the District of Columbia. Newly detected compounds are shown in darker hue. Error bars depict ± one standard deviation (n [ 5). Some concentrations represent estimates only (z) and some analytes were detected inconsistently (*).
664
water research 44 (2010) 658–668
analyte concentrations across the various treatment facilities represented. While being efficient and economical for the intended purpose, this approach was not well suited to capture the full spectrum of concentrations of individual PPCPs as a function of plant type, treatment processes employed, populations served, as well as of geographical locations and climate zones represented. As a comparison with the EPA TNSSS data reveals (USEPA, 2009), the detection frequency of less abundant analytes was significantly reduced in composite samples compared to individual sample analysis. Therefore, analytes that were not detected will represent a conservative estimate, and may still occur at detectable concentrations in individual samples from specific plants. While the mega composite approach cannot serve to determine variability between the large numbers of WWTPs studied, it was found to be suitable for identifying major contaminants of concern as well as their average concentration in a large sample set.
4.2.
Sanitizing agents
Triclocarban, which in previous U.S. studies had been found in concentrations ranging from 5.97 to 51 mg kg1 dry weight (dw) (Heidler et al., 2006; Chu and Metcalfe, 2007; Sapkota et al., 2007), was detected in every composite sample assayed. Also found in significant amounts was triclosan, which has been observed in a number of U.S. studies of sewage sludge with reported concentrations ranging from 0.53 to 30 mg kg1 dw (Chu and Metcalfe, 2007; Heidler and Halden, 2007; Kinney et al., 2008; McAvoy et al., 2002; USEPA, 2003). The EPA TNSSS (USEPA, 2009) found triclocarban and triclosan at concentrations of up to 441 and 133 mg kg1 dw, respectively, with mean concentrations at 38.7 59.7 and 12 18 mg kg1 dw (n ¼ 74), respectively. The relatively high mean concentrations of triclocarban and triclosan reported here and by the U.S. EPA (USEPA, 2009) in U.S. biosolids are in line with the intense usage of these antimicrobials and their high octanol/water partitioning coefficient (log KOW) of 4.9 and 4.8, respectively (both at neutral pH), which indicates significant potential of both compounds for sorption to biosolids (Halden and Paull, 2005). In addition, triclosan was found to be persistent in biosolids after aeration (Ying et al., 2007). Both triclosan and triclocarban concentrations found in the present study fall within the range of concentrations previously reported. Information on previously reported concentrations of PPCPs in biosolids is available in Table S2, SM.
4.3.
Antibiotics
The most abundant antibiotic was ciprofloxacin (6.8 2.3 mg kg1 dw; n ¼ 5), which is among the 30 most prescribed drugs in the United States, according to Internet sources. Ciprofloxacin, a metabolite of enrofloxacin, is polar and therefore prone to electrostatic interactions with the negatively charged surfaces of microbes that are found in high concentrations especially in secondary sludge (Ternes et al., 2004b). Ciprofloxacin has been detected in sewage sludge by several studies conducted in Sweden and Switzerland. Detected concentrations ranged from 6 105–11 mg kg1 dw (Golet et al., 2002; Lindberg et al., 2005, 2006, 2007). Therefore, the mean concentration found in the present study (6.8 2.3 mg kg1
dw; n ¼ 5) falls in the mid range of concentrations reported from outside of the U.S. The EPA TNSSS also identified ciprofloxacin as an abundant microcontaminant in sewage sludge, which was detected in every sample analyzed, yielding a mean concentration of 8.7 8.5 mg kg1 dw. At a mean concentration of 5.4 1.9 mg kg1 dw, ofloxacin was the fourth most abundant contaminant found in biosolids. It is fairly hydrophilic (log KOW of 0.2) and does not appear on the list of the top 200 prescription drugs. Its detection at elevated levels in every composite sample came as a surprise. However, ofloxacin had been found in concentrations of up to 2 mg kg1 dw in biosolids from Sweden (Lindberg et al., 2005). It can be speculated that the carboxyl group contained in ofloxacin may form an ion complex with exchangeable cations associated with negatively charged surfaces. Among the 10 most abundant PPCPs of the present study were three tetracycline antibiotics, 4-epitetracycline, tetracycline and minocycline, which had not been reported in biosolids before in the peer-reviewed literature. In addition, doxycycline was found as the ninth most abundant PPCP at a mean concentration of 1 0.4 mg kg1 dw. It had not been detected in biosolids from the U.S. before and only once in a Swedish study that found concentrations similar to those reported here (Lindberg et al., 2005). All tetracycline antibiotics are fairly hydrophilic (log KOW of 1.33 for tetracycline/4-epitetracycline, 0.42 for minocycline, and 1.36 for doxycycline). Despite their hydrophilic character, the mean concentrations reported here are around 1–2 mg kg1 dw. Tetracycline antibiotics are known to precipitate with ions of magnesium, calcium and ferric iron, and therefore accumulate in the solid fraction during wastewater treatment. Tetracycline, doxycycline and minocycline also rank among the 200 prescription drugs most widely used in the U.S. Other tetracycline antibiotics analyzed in this study were only found at mean concentrations below 0.4 mg kg1 dw, but the concentrations of anhydrotetracycline, 4-epianhydrotetracycline and 4-epianhydrochlortetracycline were possibly underestimated due to low recoveries. The sum of tetracycline antibiotics found in sewage sludge constitute about 8 1.3 mg kg1 dw, which is similar to the findings of the EPA TNSSS that found about 5 mg kg1 dw (USEPA, 2009). Azithromycin was found at 0.8 0.2 mg kg1 dw. It ranked as the 6th most frequently prescribed drug in 2007 and is also fairly hydrophobic. Due to both these properties and the fact that biodegradation of azithromycin was found to be insignificant (Ericson, 2007), one would expect high concentrations in biosolids. Yet, there are 7 drugs (not counting triclocarban and triclosan) that were found at higher concentrations. However, a low recovery of azithromycin of only 12% may indicate that actual azithromycin concentrations in biosolids are much higher than the detected concentration. Azithromycin has been detected in previous studies at concentrations of up to 6.5 mg kg1 dw in the U.S. (Jones-Lepp and Stevens, 2007) and at up to 0.16 mg kg1 dw in sludge from Germany and Switzerland (Gobel et al., 2005a, 2005b).
4.4.
Bioavailability and soil/porewater equilibria
To explore the importance of bioavailability of PPCPs sequestered in biosolids, the concentrations of individual PPCPs that are anticipated to occur in the solid and liquid
665
water research 44 (2010) 658–668
and the two sanitizing agents triclosan and triclocarban (Fig. 2).
4.5.
Risk assessment data gaps
These results suggest that aside from the 7 notable exceptions discussed above, the majority of the PPCPs detected in biosolids in this study likely exert no acute effects on aquatic organisms, assuming that biosolids are applied as regulated by the EPA (USEPA, 1994) and that the migration of solids from agricultural land to surface water via soil erosion and runoff can be completely prevented. While the former assumption is plausible, the latter may not always apply, as soil erosion is a common phenomenon. However, chronic toxicity as well as effects from mixtures of PPCPs on non-target organisms cannot be assessed due to lack of appropriate toxicity data. It has been speculated that the presence of sub-therapeutic concentrations of antibiotics may adversely affect soil microbial community structures as well as induce spreading of resistance among bacterial pathogens. In addition to influencing microbial populations, it has been shown that some
1000 100 10 1 0.1
100 000
100 000
10 000
10 000
1000
1000
100
100
10
10 1
1
0.1
0.1
*
* *
‡
*
* * ‡* *
‡*
*
0.0001
* ‡
0.0001
‡ ‡
0.01 0.001 ‡ *
0.01 0.001
Effect concentration [µg L-1]
4-epianhydrotetracycline ‡ 4-epichlortetracycline * 4-epitetracycline anhydrotetracycliine ‡ azithromycin ‡ caffeine carbamazepine chlortetracycline * cimetidine ‡ ciprofloxacin clarithromycin codeine * cotinine diltiazem diphenhydramine doxycycline enrofloxacin ‡ * erythromycin-H2O fluoxetine gemfibrozil ibuprofen * isochlortetracycline * lomefloxacin ‡ * metformin * miconazole minocycline * naproxen norfloxacin ofloxacin ‡ oxytetracycline ranitidine sulfamethoxazole * sulfanilamide * tetracycline thiabendazole triclocarban triclosan trimethoprim *
0.01
4-epianhydrotetracycline 4-epichlortetracycline 4-epitetracycline anhydrotetracycliine azithromycin caffeine carbamazepine chlortetracycline cimetidine ciprofloxacin clarithromycin codeine cotinine diltiazem diphenhydramine doxycycline enrofloxacin erythromycin-H2O fluoxetine gemfibrozil ibuprofen isochlortetracycline lomefloxacin metformin miconazole minocycline naproxen norfloxacin ofloxacin oxytetracycline ranitidine sulfamethoxazole sulfanilamide tetracycline thiabendazole triclocarban triclosan trimethoprim
Conc. in porewater at pH 7-9 [µg L-1]
Conc. in dry weight soil at pH 7-9 [µg kg-1]
phases upon mixing of land applied biosolids into soil were calculated (Fig. 2). In fully equilibrated soil-biosolids mixtures (assuming an EPA-recommended mixing ratio of approximate 25:1), concentrations of PPCPs on soil particles are expected to fall into the ppb range for most analytes, except for triclocarban, which was projected to be present at around 1 ppm(w/w) dw. Since most of the PPCPs found in biosolids are fairly hydrophobic, their calculated concentrations in porewater at equilibrium typically were quite low (Fig. 2 B) (Kinney et al., 2008). A comparison of these estimated dissolved PPCP levels in porewater with the lowest effect concentrations for aquatic organisms (red circles in Fig. 2 B) suggests that the leaching of dissolved PPCPs into surface waters probably does not present an important mechanism for exposure of aquatic biota for the majority of analytes detected. Concentrations calculated for soil porewater typically were several orders of magnitude below the lowest effect concentration reported for aquatic organisms. Notable exceptions were 6 analytes (Fig. 2) that are expected to yield potentially problematic concentrations in porewater after land application and partitioning of biosolidsderived PPCPs. These include the antibiotics ciprofloxacin, ofloxacin and tetracycline, as well as the stimulant caffeine,
Fig. 2 – Predicted equilibrium concentrations of PPCPs associated with particulates of soil-biosolids mixtures (top) and dissolved in porewater (bottom) after land application of biosolids on agricultural soil. Data depict the environmentally relevant range between pH 7 and 9. Circles represent the lowest ecological effect concentrations contained in the EPA Ecotox database. Some concentrations were calculated based on estimates only (z) and some analytes were detected inconsistently (*).
666
water research 44 (2010) 658–668
antibiotics, specifically drugs belonging to the tetracyclines, fluoroquinolones and sulfonamides, may be taken up by crop plants (Migliore et al., 2003; Kumar et al., 2005; Dolliver et al., 2007). This presents a potential exposure pathway to humans through ingestion of contaminated food and may result in the promotion of resistant bacteria in humans (Shoemaker et al., 2001). Furthermore, information is lacking to determine risks to the health of agricultural soils and soil-dwelling organisms. Few studies have examined the half-lives in soils of PPCPs sequestered in biosolids. The effect of ppm levels of sanitizing agents on soil microbial communities has rarely been investigated to date (Liu et al., 2009). Similarly, the half-life of PPCPs in biosolids-amended soils will require additional research to inform risk assessment analyses. Passage through municipal digesters and chemical aging may reduce the bioavailability of PPCPs and with it the risk of chemical uptake of and exposure to soil-dwelling organisms. However, a reduced bioavailability also may imply a prolonged half-life of these compounds in the environment, with possible delayed release in sensitive compartments. Furthermore, studies are lacking on the potential of biosolids-derived antimicrobials and antibiotics to exert selective pressure for the enrichment of drug-resistant microorganisms, a scenario demonstrated in vitro (Braoudaki and Hilton, 2004). Also unavailable are threshold concentrations for toxic effects in terrestrial organisms of many PPCPs, including some that have been detected in biosolids at ppm levels. The EPA’s Ecotox database (www.epa.gov/ecotox) provides only some toxicity values in aquatic organisms for the PPCPs investigated here. Bioaccumulation and biomagnification are other aspects that will need to be considered for risk assessment purposes. Although PPCPs are not typically thought of as representing persistent hydrophobic pollutants, some may be subject to bioaccumulation and possibly biomagnification thereafter in both terrestrial and aquatic environments. Bioaccumulation was demonstrated for triclosan and triclocarban in lab and field studies that examined uptake of these compounds from soil, sediment and water (Coogan and La Point, 2008; Higgins et al., 2009; Kinney et al., 2008).
5.
Conclusions
Overall, this study reemphasizes the significance of biosolids recycling as a mechanism for the release of PPCPs into the environment. Based on the mean concentrations of all analytes detected, it is estimated that the total loading to U.S. soils from nationwide biosolids recycling is on the order of 210–250 metric tons per year for the 72 PPCPs investigated here. It is concluded that, despite large variations found by the U.S. EPA between different treatment plants, mean concentrations of PPCPs in U.S. biosolids on a nationwide basis have remained fairly constant between 2001 and 2007 and possibly longer. Good agreement between this and the U.S. EPA study further suggests that the here demonstrated use of mega composite samples represents a cost-effective approach for collecting regional and nationwide information on average concentrations of contaminants in biosolids.
Acknowledgements We thank Rick Stevens and Harry B. McCarty from the U.S. EPA for providing samples from the 2001 National Sewage Sludge Survey. We also thank Randhir Deo and Jochen Heidler for their valuable input. This work was supported in part by the National Institute of Environmental Health Sciences by NIEHS grant 1R01ES015445 and by the Johns Hopkins University Center for a Livable Future.
Appendix. Supplementary material Additional details regarding experimental procedures and results can be found in the Supplementary Material accompanying this article. This information is available free of charge via the Internet at www.iwaponline.com. Supplementary data associated with this article can be found in the online version at doi:10.1016/j.watres.2009.12. 032.
references
Bantle, J.A., Burton, D.T., Dawson, D.A., Dumont, J.N., Finch, R.A., Fort, D.J., Linder, G., Rayburn, J.R., Buchwalter, D., Gaudethull, A.M., Maurice, M.A., Turley, S.D., 1994. Fetax interlaboratory validation study - Phase II testing. Environmental Toxicology and Chemistry 13 (10), 1629–1637. Brain, R.A., Johnson, D.J., Richards, S.M., Sanderson, H., Sibley, P.K., Solomon, K.R., 2004. Effects of 25 pharmaceutical compounds to Lemna gibba using a seven-day static-renewal test. Environmental Toxicology and Chemistry 23 (2), 371–382. Brain, R.A., Ramirez, A.J., Fulton, B.A., Chambliss, C.K., Brooks, B. W., 2008. Herbicidal Effects of Sulfamethoxazole in Lemna gibba: Using p-Aminobenzoic Acid As a Biomarker of Effect. Environmental Science & Technology 42 (23), 8965–8970. Braoudaki, M., Hilton, A.C., 2004. Low level of cross-resistance between triclosan and antibiotics in Escherichia coli K-12 and E. coli O55 compared to E. coli O157. Fems Microbiology Letters 235, 305–309. Cahill, J.D., Furlong, E.T., Burkhardt, M.R., Kolpin, D., Anderson, L.G., 2004. Determination of pharmaceutical compounds in surfaceand ground-water samples by solid-phase extraction and highperformance liquid chromatography-electrospray ionization mass spectrometry. Journal of Chromatography A 1041, 171–180. Causarano, H.J., Franzluebbers, A.J., Shaw, J.N., Reeves, D.W., Raper, R.L., Wood, C.W., 2008. Soil organic carbon fractions and aggregation in the Southern Piedmont and Coastal Plain. Soil Science Society of America Journal 72, 221–230. Chalew, T., Halden, R., 2008. Environmental exposure of aquatic and terrestrial biota to triclosan and triclocarban. Journal of the American Water Resources Association 45 (1), 4–13. Chu, S.G., Metcalfe, C.D., 2007. Simultaneous determination of triclocarban and triclosan in municipal biosolids by liquid chromatography tandem mass spectrometry. Journal of Chromatography A 1164, 212–218. Coogan, M.A., La Point, T.W., 2008. Snail bioaccumulation of triclocarban, triclosan, and methyltriclosan in a North Texas, USA, stream affected by wastewater treatment plant runoff. Environmental Toxicology and Chemistry 27, 1788–1793.
water research 44 (2010) 658–668
Daughton, C.G., Ternes, T.A., 1999. Pharmaceuticals and personal care products in the environment: agents of subtle change? Environmental Health Perspectives 107, 907–938. De Lannoy, G.J.M., Verhoest, N.E.C., Houser, P.R., Gish, T.J., Van Meirvenne, M., 2006. Spatial and temporal characteristics of soil moisture in an intensively monitored agricultural field (OPE3). Journal of Hydrology 331, 719–730. Dolliver, H., Kumar, K., Gupta, S., 2007. Sulfamethazine uptake by plants from manure-amended soil. Journal of Environmental Quality 36, 1224–1230. Ericson, J.F., 2007. An evaluation of the OECD 308 water/sediment systems for investigating the biodegradation of pharmaceuticals. Environmental Science & Technology 41, 5803–5811. Fent, K., Weston, A.A., Caminada, D., 2006. Ecotoxicology of human pharmaceuticals. Aquatic Toxicology 76, 122–159. Fick, J., Soderstrom, H., Lindberg, R.H., Phan, C., Tysklind, M., Larsson, D.G.J., 2009. Contamination of surface, ground, and drinking water from pharmaceutical production. Environmental Toxicology and Chemistry 28, 2522–2527. Focazio, M.J., Kolpin, D.W., Barnes, K.K., Furlong, E.T., Meyer, M.T., Zaugg, S.D., Barber, L.B., Thurman, M.E., 2008. A national reconnaissance for pharmaceuticals and other organic wastewater contaminants in the United States – II untreated drinking water sources. Science of the Total Environment 402, 201–216. Gobel, A., Thomsen, A., McArdell, C.S., Alder, A.C., Giger, W., Theiss, N., Loffler, D., Ternes, T.A., 2005a. Extraction and determination of sulfonamides, macrolides, and trimethoprim in sewage sludge. Journal of Chromatography A 1085, 179–189. Gobel, A., Thomsen, A., McArdell, C.S., Joss, A., Giger, W., 2005b. Occurrence and sorption behavior of sulfonamides, macrolides, and trimethoprim in activated sludge treatment. Environmental Science & Technology 39, 3981–3989. Golet, E.M., Strehler, A., Alder, A.C., Giger, W., 2002. Determination of fluoroquinolone antibacterial agents in sewage sludge and sludge-treated soil using accelerated solvent extraction followed by solid-phase extraction. Analytical Chemistry 74, 5455–5462. Halden, R.U., Paull, D.H., 2005. Co-occurrence of triclocarban and triclosan in US water resources. Environmental Science & Technology 39, 1420–1426. Halling-Sorensen, B., Nielsen, S.N., Lanzky, P.F., Ingerslev, F., Lutzhoft, H.C.H., Jorgensen, S.E., 1998. Occurrence, fate and effects of pharmaceutical substances in the environment a review. Chemosphere 36, 357–394. Halling-Sorensen, B., 2001. Inhibition of aerobic growth and nitrification of bacteria in sewage sludge by antibacterial agents. Archives of Environmental Contamination and Toxicology 40 (4), 451–460. Hanson, M.L., Knapp, C.W., Graham, D.W., 2006. Field assessment of oxytetracycline exposure to the freshwater macrophytes Egeria densa Planch and Ceratophyllum demersum L. Environmental Pollution 141 (3), 434–442. Heberer, T., Fuhrmann, B., Schmidt-Baumler, K., Tsipi, D., Koutsouba, V., Hiskia, A., 2000. Occurrence of pharmaceutical residues in sewage, river, ground-, and drinking water in Greece and Germany. Abstracts of Papers of the American Chemical Society 219 U623. Heidler, J., Halden, R.U., 2007. Mass balance assessment of triclosan removal during conventional sewage treatment. Chemosphere 66, 362–369. Heidler, J., Sapkota, A., Halden, R.U., 2006. Partitioning, persistence, and accumulation in digested sludge of the topical antiseptic triclocarban during wastewater treatment. Environmental Science & Technology 40, 3634–3639.
667
Higgins, C.P., Paesani, Z.J., Chalew, T.E.A., Halden, R.U., 2009. Bioaccumulation of triclocarban in Lumbriculus variegates. Environmental Toxicology and Chemistry 65, 141–148. Hirsch, R., Ternes, T., Haberer, K., Kratz, K.L., 1999. Occurrence of antibiotics in the aquatic environment. Science of the Total Environment 225, 109–118. Jones-Lepp, T.L., Stevens, R., 2007. Pharmaceuticals and personal care products in biosolids/sewage sludge: the interface between analytical chemistry and regulation. Analytical and Bioanalytical Chemistry 387, 1173–1183. Jos, A., Repetto, G., Rios, J.C., Hazen, N., Molero, M.L., del Peso, A., Salguero, M., Fernandez-Freire, P., Perez-Martin, J.M., Camen, A., 2003. Ecotoxicological evaluation of carbamazepine using six different model systems with eighteen endpoints. Toxicology in Vitro 17 (5–6), 525–532. Kinney, C.A., Furlong, E.T., Kolpin, D.W., Burkhardt, M.R., Zaugg, S.D., Werner, S.L., Bossio, J.P., Benotti, M.J., 2008. Bioaccumulation of pharmaceuticals and other anthropogenic waste indicators in earthworms from agricultural soil amended with biosolid or swine manure. Environmental Science & Technology 42, 1863–1870. 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 US streams, 1999–2000: a national reconnaissance. Environmental Science & Technology 36, 1202–1211. Kumar, K., Gupta, S.C., Baidoo, S.K., Chander, Y., Rosen, C.J., 2005. Antibiotic uptake by plants from soil fertilized with animal manure. Journal of Environmental Quality 34, 2082–2085. Kupper, T., Berset, J.D., Etter-Holzer, R., Furrer, R., Tarradellas, J., 2004. Concentraions and specific loads of polycyclic musks in sewage sludge originating from a monitoring network in Switzerland. Chemosphere 54, 1111–1120. Lindberg, R.H., Wennberg, P., Johansson, M.I., Tysklind, M., Andersson, B.A.V., 2005. Screening of human antibiotic substances and determination of weekly mass flows in five sewage treatment plants in Sweden. Environmental Science & Technology 39, 3421–3429. Lindberg, R.H., Olofsson, U., Rendahl, P., Johansson, M.I., Tysklind, M. , Andersson, B.A.V., 2006. Behavior of fluoroquinolones and trimethoprim during mechanical, chemical, and active sludge treatment of sewage water and digestion of sludge. Environmental Science & Technology 40, 1042–1048. Lindberg, R.H., Bjorklund, K., Rendahl, P., Johansson, M.I., Tysklind, M., Andersson, B.A.V., 2007. Environmental risk assessment of antibiotics in the Swedish environment with emphasis on sewage treatment plants. Water Research 41, 613–619. Lindsey, M.E., Meyer, M., Thurman, E.M., 2001. Analysis of trace levels of sulfonamide and tetracycline antimicrobials, in groundwater and surface water using solid-phase extraction and liquid chromatography/mass spectrometry. Analytical Chemistry 73, 4640–4646. Liu, F., Ying, G.-G., Yang, L.-H., Zhou, Q.-X., 2009. Terrestrial ecotoxicological effects of the antimicrobial agent triclosan. Ecotoxicology and Environmental Safety 72, 86–92. Loos, R., Wollgast, J., Huber, T., Hanke, G., 2007. Polar herbicides, pharmaceutical products, perfluorooctanesulfonate (PFOS), perfluorooctanoate (PFOA), and nonylphenol and its carboxylates and ethoxylates in surface and tap waters around Lake Maggiore in Northern Italy. Analytical and Bioanalytical Chemistry 387, 1469–1478. Loraine, G.A., Pettigrove, M.E., 2006. Seasonal variations in concentrations of pharmaceuticals and personal care products in drinking water and reclaimed wastewater in Southern California. Environmental Science & Technology 40, 687–695. Luetzhoft, H.C.H., Halling-Sorensen, B., Jorgensen, S.E., 1999. Algal toxicity of antibacterial agents applied in Danish fish
668
water research 44 (2010) 658–668
farming. Archives of Environmental Contamination and Toxicology 36 (1), 1–6. McAvoy, D.C., Schatowitz, B., Jacob, M., Hauk, A., Eckhoff, W.S., 2002. Measurement of triclosan in wastewater treatment systems. Environmental Toxicology and Chemistry 21, 1323–1329. Migliore, L., Cozzolino, S., Fiori, M., 2003. Phytotoxicity to and uptake of enrofloxacin in crop plants. Chemosphere 52, 1233–1244. Moldovan, Z., 2006. Occurrences of pharmaceutical and personal care products as micropollutants in rivers from Romania. Chemosphere 64, 1808–1817. National Research Council, 2002. Biosolids Applied to Land: Advancing Standards and Practices Washington, D.C. North East Biosolids and Residuals Association, 2007. A National Biosolids Regulation, Quality, End Use & Disposal Survey Tamworth, NH. Roberts, P.H., Thomas, K.V., 2006. The occurrence of selected pharmaceuticals in wastewater effluent and surface waters of the lower Tyne catchment. Science of the Total Environment 356, 143–153. Robinson, A.A., Belden, J.B., Lydy, M.J., 2005. Toxicity of fluoroquinolone antibiotics to aquatic organisms. Environmental Toxicology and Chemistry 24 (2), 423–430. Sapkota, A., Heidler, J., Halden, R.U., 2007. Detection of triclocarban and two co-contaminating chlorocarbanilides in US aquatic environments using isotope dilution liquid chromatography tandem mass spectrometry. Environmental Research 103, 21–29. Shoemaker, N.B., Vlamakis, H., Hayes, K., Salyers, A.A., 2001. Evidence for extensive resistance gene transfer among Bacteroides spp. and among Bacteroides and other genera in the human colon. Applied and Environmental Microbiology 67, 561–568. Stackelberg, P.E., Furlong, E.T., Meyer, M.T., Zaugg, S.D., Henderson, A.K., Reissman, D.B., 2004. Persistence of pharmaceutical compounds and other organic wastewater contaminants in a conventional drinking-watertreatment plant. Science of the Total Environment 329, 99–113. Tamtam, F., Mercier, F., Le Bot, B., Eurin, J., Dinh, Q.T., Clement, M., Chevreuil, M., 2008. Occurrence and fate of antibiotics in the Seine river in various hydrological conditions. Science of the Total Environment 393, 84–95. Ternes, T.A., 1998. Occurrence of drugs in German sewage treatment plants and rivers. Water Research 32, 3245–3260. Ternes, T.A., Herrmann, N., Bonerz, M., Knacker, T., Siegrist, H., Joss, A., 2004a. A rapid method to measure the solid-water
distribution coefficient (K-d) for pharmaceuticals and musk fragrances in sewage sludge. Water Research 38, 4075–4084. Ternes, T.A., Joss, A., Siegrist, H., 2004b. Scrutinizing pharmaceuticals and personal care products in wastewater treatment. Environmental Science & Technology 38, 392A–399A. USEPA, 1994. Biosolids Recycling: Beneficial Technologies for a Better Environment. Office of Water. U.S. Environmental Protection Agency, Washington, D.C. EPA 832-R-94–009. USEPA, 2001. Sampling Procedures for the 2001 National Sewage Sludge Survey. Office of Science and Technology, Washington, D.C. USEPA, 2003. Workshop on Emerging Pollutants. Region/ORD Workshop on Emerging Pollutants. U.S. EPA, Chicago, IL. USEPA, 2007a. 2001 National Sewage Sludge Survey Report Washington, D.C. EPA-822-R-07-006. USEPA, 2007b. Method 1694: Pharmaceuticals and Personal Care Products in Water, Soil, Sediment, and Biosolids by HPLC/MS/ MS Washington, D.C. EPA-821-R-08–002. USEPA, 2009. Targeted National Sewage Sludge Survey Washington, D.C. EPA-822-R-08–014. Williams, R.R., Bell, T.A., Lightner, D.V., 1992. Shrimp antimicrobial testing: II. Toxicity testing and safety determination for twelve antimicrobials with penaeid shrimp larvae. Journal of Aquatic Animal Health 4 (4), 262–270. Wilson, B.A., Smith, V.H., Denoyelles, F., Larive, C.K., 2003. Effects of three pharmaceutical and personal care products on natural freshwater algal assemblages. Environmental Science & Technology 37 (9), 1713–1719. Wu, C.X., Witter, J.D., Spongberg, A.L., Czajkowski, K.P., 2009. Occurrence of selected pharmaceuticals in an agricultural landscape, western Lake Erie basin. Water Research 43, 3407–3416. Yang, L.H., Ying, G.G., Su, H.C., Stauber, J.L., Adams, M.S., Binet, M.T., 2008. Growth-inhibiting effects of 12 antibacterial agents and their mixtures on the freshwater microalga Pseudokirchneriella subcapitata. Environmental Toxicology and Chemistry 27 (5), 1201–1208. Ying, G.G., Yu, X.Y., Kookana, R.S., 2007. Biological degradation of triclocarban and triclosan in a soil under aerobic and anaerobic conditions and comparison with environmental fate modelling. Environmental Pollution 150, 300–305. Zurita, J.L., Repetto, G., Jos, A., Salguero, M., Lopez-Artiguez, M., Camean, A.M., 2007. Toxicological effects of the lipid regulator gemfibrozil in four aquatic systems. Aquatic Toxicology 81 (1), 106–115.