WATER RESEARCH A Journal of the International Water Association
Editor-in-Chief Mark van Loosdrecht Delft University of Technology and KWR Watercycle Research The Netherlands E-mail:
[email protected] Editors Jean-Claude Block Université H. Poincaré, Nancy I France David Dixon University of Melbourne Australia Hiroaki Furumai University of Tokyo Japan Xiaodi Hao Beijing University of Civil Engineering and Architecture China Gregory Korshin University of Washington USA Anna Ledin Formas Sweden Eberhard Morgenroth ETH Zürich and Eawag, Swiss Federal Institute of Aquatic Science and Technology Switzerland Wolfgang Rauch University of Innsbruck Austria Maria Reis Universidade Nova de Lisboa/FCT Portugal Hang-Shik Shin Korea Advanced Institute of Science and Technology Korea Thomas Ternes Bundesanstalt für Gewässerkunde Germany Stefan Wuertz University of California, Davis USA
Associate Editors Andrew Baker University of New South Wales Australia Damien Batstone University of Queensland Australia
Guang-Hao Chen Hong Kong University of Science & Technology China Tom Curtis University of Newcastle upon Tyne UK Francis de los Reyes III North Carolina State University USA Ana Deletic Monash University USA Rob Eldridge Glen Waverley Australia Monica B. Emelko University of Waterloo Canada Rosina Girones University of Barcelona Spain Stephen Gray Victoria University Australia Kate Grudpan Chiang Mai University Thailand Edwin Herricks University of Illinois at Urbana-Champaign USA Peter Hillis United Utilities Plc UK Hong-Ying Hu Tsinghua University China Bruce Jefferson Cranfield University UK Ulf Jeppsson Lund University Sweden Sergey Kalyuzhnyi Moscow State University Russian Federation Jae-Hong Kim Georgia Institute of Technology USA Günter Langergraber University of Natural Resources and Applied Life Sciences, Vienna (BOKU) Austria
Yoshihiko Matsui Hokkaido University Japan Armand Maul Université de Metz France Max Maurer Eawag Switzerland How Yong Ng National University of Singapore Singapore Satoshi Okabe Hokkaido University Japan Jong Moon Park Pohang University of Science & Technology Korea
Shang-Lien Lo National Taiwan University Taiwan Dionisis Mantzavinos Technical University of Crete Greece
Anastasios Zouboulis Aristotle University of Thessaloniki Greece
Susan Richardson US Environmental Protection Agency USA Miguel Salgot University of Barcelona Spain David Sedlak University of California, Berkeley USA Jean-Philippe Steyer LBE-INRA France Masahiro Takahashi Hokkaido University Japan Nathalie Tufenkji McGill University Canada Kai Udert Eawag Switzerland Vayalam Venugopalan BARC Facilities India Eduardo von Sperling Federal University of Minas Gerais Brazil Julie Zilles University of Illinois at Urbana-Champaign USA
Editorial Office E-mail:
[email protected] Publication information: Water Research (ISSN 0043-1354). For 2011, volume 45 is scheduled for publication. Subscription prices are available upon request from the publisher or from the Elsevier Customer Service Department nearest you or from this journal’s website (http://www.elsevier.com/locate/watres). Further information is available on this journal and other Elsevier products through Elsevier’s website (http://www.elsevier.com). Subscriptions are accepted on a prepaid basis only and are entered on a calendar year basis. Issues are sent by standard mail (surface within Europe, air delivery outside Europe). Priority rates are available upon request. Claims for missing issues should be made within six months of the date of dispatch. Author Enquiries: For enquiries relating to the submission of articles (including electronic submission) please visit this journal’s homepage at http://www.elsevier.com/ locate/watres. Contact details for questions arising after acceptance of an article, especially those relating to proofs, will be provided by the publisher. You can track accepted articles at http://www.elsevier.com/trackarticle. You can also check our Author FAQs at http://www.elsevier.com/authorFAQ and/or contact Customer Support via http://support.elsevier.com. Orders, claims, and journal enquiries: Please contact the Elsevier Customer Service Department nearest you: St. Louis: Elsevier Customer Service Department, 3251 Riverport Lane, Maryland Heights, MO 63043, USA; phone: (877) 8397126 [toll free within the USA]; (+1) (314) 4478878 [outside the USA]; fax: (+1) (314) 4478077; e-mail:
[email protected] Oxford: Elsevier Customer Service Department, The Boulevard, Langford Lane, Kidlington OX5 1GB, UK; phone: (+44) (1865) 843434; fax: (+44) (1865) 843970; e-mail:
[email protected] Tokyo: Elsevier Customer Service Department, 4F Higashi-Azabu, 1-Chome Bldg, 1-9-15 Higashi-Azabu, Minato-ku, Tokyo 106-0044, Japan; phone: (+81) (3) 5561 5037; fax: (+81) (3) 5561 5047; e-mail:
[email protected] Singapore: Elsevier Customer Service Department, 3 Killiney Road, #08-01 Winsland House I, Singapore 239519; phone: (+65) 63490222; fax: (+65) 67331510; e-mail:
[email protected] Application for membership of International Water Association should be made to: Executive Director, IWA, Alliance House, 12 Caxton Street, London SW1H 0QS, U.K. (Tel.: +44 207 654 5500; Fax: +44 207 654 5555; e-mail:
[email protected]; website: http://www.IWAhq.org.uk). Registered Charity (England) No. 289269. Individual membership is available from £30 upwards. For details contact IWA.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 2 8 7 e4 3 1 0
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Retrospective of ecological approaches to excess sludge reduction Anwar Khursheed*, A.A. Kazmi Department of Civil Engineering, Indian Institute of Technology Roorkee, India
article info
abstract
Article history:
The problem of excess sludge handling produced during wastewater treatment is unde-
Received 18 June 2010
niable reality of grave concern with increasingly stringent legislations. The sludge
Received in revised form
synthesis yield being 0.4e0.6 kgVSS/kgCOD (0.57e0.8 kgCODcell/kgCOD), results in high
26 February 2011
power consumption on its digestion and therefore taken considerable attention to achieve
Accepted 22 May 2011
sustainable strategies.
Available online 31 May 2011
Solids reduction by physico-chemical methods results in buildup of chemicals. This may present risk to the environment and may require further treatment to remove the
Keywords:
chemicals of concern in future. Wastewater sludge reduction upto 100% by biological,
Low sludge production
sustainable, non-hazardous, and environment friendly methods has been successfully
MBR
tested at different levels. Therefore, above reasons were sufficient driving forces to confine
Oligochaeta
this review to non-chemically assisted processes. Similarly, the thermally assisted
Filamentous fungi
processes result in high carbon footprint and excluded from the scope of this review.
Maintenance metabolism
Enough has been reviewed on sludge reduction, as numbers of articles on the same subject
High oxygenation
with different angles have been reported, still the progress in the last few years is missing; hence, special emphasis is given herewith to highlight the efforts of the last five years. ª 2011 Elsevier Ltd. All rights reserved.
Contents 1.
2.
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.1. Carbonaceous oxidation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.2. Nitrification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.3. Denitrification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Sludge reduction efforts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1. Lysis-cryptic growth . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1.1. High oxygenation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2. Uncoupling metabolism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2.1. Initial substrate to biomass ratio (So/Xo) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2.2. Oxic-settling-anaerobic (OSA) process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3. Maintenance metabolism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3.1. Cost and energy considerations in MBR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
* Corresponding author. Tel.: þ91 9319077764; fax: þ91 1332 275568. E-mail addresses:
[email protected] (A. Khursheed),
[email protected] (A.A. Kazmi). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.05.018
4288 4288 4288 4289 4289 4291 4291 4292 4292 4293 4293 4297
4288
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 2 8 7 e4 3 1 0
2.4.
3. 4.
1.
Eco-manipulation (predation on bacteria) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.4.1. Two-stage system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.4.2. Oligochaeta . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.4.3. Filamentous fungi (FF) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Where; fsa ¼ fsa fm ¼ Net fraction of electron donor used in synthesis
Introduction
Organic matter and nutrient removal through biological oxidation is generally regarded as the most economically viable means of wastewater treatment. But, only those systems removing inorganic nutrients simultaneously and generating fewer volumes of excess sludge biomass can be regarded comprehensive, so as to give real protection to streams against eutrophication and environment against sludges. A very complex microbial ecosystem consisting of bacteria, protozoa, metazoa, viruses, worms, helminthes etc is involved in this sequential degradation of organics followed by nitrification, denitrification and phosphorous removal. Prerequisites to successful removal of all these substrates are efficient growth of biomass for organics, low COD, high dissolved oxygen (DO) and long sludge retention time (SRT) for nitrification, while reverse is true for denitrification in the form of sufficient COD in the absence of DO. This modus vivendi is itself a difficult preposition to optimize or even achieve (Metcalf and Eddy, 2003). In summary, the entire process is conversion of different components of wastewater by oxidation and reduction to end products, namely CO2, H2O, N2, new cells and fractions of intermediateries remaining due to inefficiencies (Eq (1.1)). microbes
v1 ðorganicsÞ þ v2 O2 þ v3 NH3 þ v4 PO3 4 ! v5 ðnew cellsÞ þ v6 CO2 þ v7 H2 O
(1.1)
Where, vi ¼ stoichiometric coefficient Although the outcome of pollution caused by the municipal or industrial effluents is environmental degradation, nevertheless its dimensions are different. Since, municipal wastewaters pose a universal problem; therefore, generalized expressions regarding whole process, which also gives stoichiometry of sludge growth are summarized below (Rittmann and McCarty, 2001; Metcalf and Eddy, 2003);
fsao ¼ Max fraction of electron donor used in synthesis ¼ 0.6 fm ¼ Fraction of electron donor used in maintenance feo ¼ Max fraction of electron donor used to provide energy Net fraction of electron donor used to provide energy ¼ fe fe þ fsa ¼ 1 Microbial biodegradable fraction, fd ¼ 0.8 b ¼ 0.15/d at 20 C, SRT ¼ qx, 1 þ 1 f d b qx 1 þ b qx ð1 þ 0:03 qxÞ f sa ¼ 0:6 ð1 þ 0:15 qxÞ 1 1 Synthesis; Ya ¼ f sa C5 H7 O2 N= C10 H19 O3 N 20 50
f sa ¼ f sa
C10 H19 O3 N þ 12:5 O2 ¼¼ 9CO2 þ 7H2 O þ NHþ 4 þ HCO3
Ya ¼ f sa
1.2.
1 f sa 1 1 O2 þ ¼¼ C10 H19 O3 N þ 9 5 f sa CO2 4 50 25 1 f sa 9 f sa H2 O þ 1 NHþ þ 1 4 2 20 25 1 f sa f sa C5 H7 O2 N HCO þ 3 þ 20 50 20
113 400 = ¼ 0:70625f sa gVSS=gUBOD 20 50
(1.5)
1 f sn 1 f sn f sn f sn 1 O2 þ NHþ HCO CO2 ¼¼ NO þ 4 þ 3 þ 3 4 20 5 8 20 8 1 f sn 9 3 1 f sn þ C5 H7 O2 N þ H2 O þ Hþ þ 2 20 20 8 4 (1.6) Where fractions of electrons and biodegradable fraction are in similar meaning; Max fraction of electron donor used in nitrifiers synthesis, fsno ¼ 0.14 b ¼ 0.11/d at 20 C 1 þ 1 f d b qx 1 þ b qx
f sn ¼ 0:14
(1.4)
Nitrification
o
Carbonaceous oxidation
(1.3)
COD ¼ UBOD of C10 H19 O3 N ¼ 12:5 2 16 ¼ 400
f sn ¼ f sn
1.1.
4298 4298 4300 4302 4302 4305 4305 4305
ð1 þ 0:022 qxÞ ð1 þ 0:11 qxÞ
(1.7)
f sn 1 f sn C5 H7 O2 N= þ NHþ 4 20 8 20 f sn 1 f sn 8:07 f sn 113= þ 14 ¼ ¼ g VSS=gN 20 2:5 þ f sn 8 20
Synthesis; Yn ¼
(1.2)
(1.8)
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 2 8 7 e4 3 1 0
Since, 14 g N is equivalent to 64 g O2, therefore
Yn ¼
1:765 f sn g VSS=g UBOD 2:5 þ f sn
1.3.
(1.9)
Denitrification
þ 1 f sd 1 1 NO C10 H19 O3 N þ 9 5f sd CO2 3 þ 1 f sd H ¼¼ 5 50 25 3 1 f sd 1 f sd 9 N2 þ 1þ H2 O þ 5 10 25 1 f sd f sd HCO C5 H7 O2 N þ 3 þ 20 50 20 1 f sd þ NHþ ð1:10Þ 4 50 20 Where; Max fraction of electron donor used in synthesis, fsdo ¼ 0.52. b ¼ 0.05/d, Microbial biodegradable fraction, fd ¼ 0.8 o
f sd ¼ f sd
f sd
1 þ 1 f d b qx 1 þ b qx
Synthesis; Yd ¼ f sd
Yd ¼ f sd
(1.11)
1 1 C5 H7 O2 N= C10 H19 O3 N 20 50
113 400 = ¼ 0:70625 f sd g VSS=gUBOD 20 50
(1.12)
The net synthesis in terms of NO3 (14 g N ¼ 64 g O2) would be; Yd ¼ f sd ¼
113 1 f sd 113x5 f sd g VSS=g NO = 3 ¼ 5 20 20x14 1 f sd
2:01 f sd gVSS=gNO 3 N 1 f sd
(1.13)
Therefore, the combined growth of the mixed culture would be, Y ¼ Ya þ Yn þ Yd ¼ 0:70625 f a þ 0:70625 f sd þ
1:765 f n 2:5 þ f n
ð1 þ 0:03 qxÞ ð1 þ 0:01 qxÞ ¼ 0:70625 0:6 þ 0:70625 0:52 ð1 þ 0:15 qxÞ ð1 þ 0:05 qxÞ ð1 þ 0:022qxÞ ð1 þ 0:022qxÞ 1 0:14 þ1:765 0:14 ð1 þ 0:11qxÞ ð1 þ 0:11qxÞ Y¼
0:424ð1 þ 0:03 qxÞ 0:367ð1 þ 0:01 qxÞ þ ð1 þ 0:15 qxÞ ð1 þ 0:05 qxÞ þ
explained above. For old or slow-growing cultures fs would be less than maximum value and correspondingly same would be the overall synthesis. However, there are other reactions also taking place simultaneously and the net sludge synthesis depends on numerous factors, mainly energy requirement for cell maintenance, decay of cells, endogenous respiration, grazing by predators and lysis due to suboptimal environmental conditions and toxicity. Bio-oxidation in activated sludge or similar process configurations is highly complex due to mixed cultures and their heterogeneity under variable conditions. Being substrate limiting, cell decay and endogenous respiration accounts for considerable oxygen consumption, this has been quantified as more than 50% in many practical cases (van Loosdrecht and Henze, 1999). Another intricacy is the re-dissolution of phosphates accumulated into the sludge due to in-situ excess sludge reduction and production of soluble microbial products (SMP) within it, which are basically biomasseassociated products. These products provide electron for heterotrophs and increase their mass (Laspidou and Rittmann, 2002).
2.
ð1 þ 0:01 qxÞ ¼ 0:52 ð1 þ 0:05 qxÞ
0:247ð1 þ 0:022 qxÞ 2:5ð1 þ 0:11 qxÞ 0:14ð1 þ 0:022 qxÞ
(1.14)
Based on above Eq. (1.14) the overall combined growth of the mixed culture would vary from 0.418 to 0.55 at 20 to 10 days SRT. The combined effect in terms of overall sludge growth of all heterotrophs and autotrophs is its dependence on fs value in each case independently and consequently on SRT as
4289
Sludge reduction efforts
“Sustainable sludge handling may be defined as a method that meets requirements of efficient recycling of resources, without supply of harmful substances to humans or the environment” (Commission of European Communities, 1998). Of the constituents removed by wastewater treatment, sludge is by far the largest in volume, therefore, it’s handling methods and disposal techniques are a matter of great concern, because, without a reliable disposal method for the sludge the actual concept of water protection cannot sustain. At the time of above legislative amendment, average dry weight per capita production of sewage sludge resulting from primary, secondary and tertiary treatment was 90 g per person per day in 1998 in Europe. At the time of the implementation of the above cited Urban Waste Water Treatment Directive (UWWTD) it was expected that by the year 2005, the sludge production would increase by 50%, i.e. 10 million tons annually. Therefore, by the year 2010 and 2050 the European Union targeted to reduce final waste disposal by 20% and 50% compared to the amount of sludge waste disposed in 2000 respectively (Lundin et al., 2004). Similarly the prohibitions of sludges in landfills or surface impoundments by virtue of 1984 amendment to the Resource Conservation and Recovery Act of 1976 (RCRA, 1984) in the form of Hazardous and Solid Waste Amendment (HSWA, 1984) by the US Congress was a clear intent to move sludge management towards more acceptable technologies. The legislative concern of European and American communities has been translated into a surge of scientific studies on sludge handling, which can best be realized from the fact that average number of published articles, which was around 83 per annum till the end of last century rise to 239 per annum up till now (Engineering Village, http://www. engineeringvillage2.com) as shown graphically below in Fig. 1: No such unified legislative concerns have been shown by the Asian countries, nonetheless the situation argues for a vigorous regulatory mechanism. But, economic realities often dictate that environmental concerns not be permitted to
4290
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 2 8 7 e4 3 1 0
Fig. 1 e Details of annual publication of articles on sludge reduction.
stand in the way of development and as a result the issue is inadequately addressed or not dealt with at all. The problem of handling of excess sludge produced during wastewater treatment is an undeniable reality of grave concern and gaining large proportion with increasingly stringent environmental legislations by the respective nations at different level. This resulted in a situation that sludge reduction has almost taken a central stage in overall wastewater scenario as its treatment and disposal accounts considerably (Horan, 1990). Consequently, there is a noticeable shifting of priorities from higher volumetric rate and effluent quality to less biomass production during wastewater treatment on the basis of overall mass balance of the inputs and outputs. At the outset, it is important to identify the various components of municipal excess sludge, which may be containing 0.75% from secondary clarifier to 10% solids from thickener and rest is water (Metcalf and Eddy, 2003). The major components of these solids in the form of residues of the original wastewater and synthesized biomass are around 60% organic compounds (non-toxic), organic and inorganic nitrogen, phosphorous, heavy metals (mainly Zn, Ni, Cd, Cu, Cr, Hg, Pb, As etc. in varying concentrations), trace organic matters such as pesticides, dioxins, phenols, polychlorinated biphenyls and polycyclic aromatic hydrocarbon etc., and microbes such as bacteria, viruses, pathogens, predators etc. Some of these are beneficial and reusable, while others are hazardous merely by presence or by virtue of their concentration. The sustainability of the sludge handling ultimately depends not only on volume and mass reduction only but also on recovery of usable and containment of adverse effects of hazardous components. The average sludge synthesis yield being 0.4e0.5 kg VSS/ kg COD consumed, which amounts to 0.57e0.8 kg COD cell/
kg COD consumed, obviously results in more than 57% of the power consumption on sludge digestion as rightly mentioned by Horan (1990) in his book and Chen et al. (2000). This could be attributed to the fact that most of the existing sludge reduction technologies are capital intensive and process-wise complex. As mentioned above, excess sludge reduction has taken considerable attention in order to achieve more sustainable strategies to cope up present as well as future requirements. It is not so that technologies are not available, the recent advent of sludge reduction technologies has validated the potential to significantly change the methods by which wastewater treatment biosolids are treated and handled across the globe, but still the real impetus is to explore economically viable one. Ideally, the solution should be in-pipe rather than subsequent treatment. In addition the real environmental friendly approach is to keep the process biological in place of physicochemical or mechanical. Solids reduction technologies can be categorized in three major categories according to their treatment methods viz.; physico-chemical, mechanical, and biological. The physicochemical and mechanical methods are fairly easily understood in how they might function, i.e. through the oxidation of organic material or the lysis of microbial material, thus making the overall mass more degradable, but the biological systems of sludge reduction, and the mechanisms behind them are much less understood. Since, the volumetric percentage of water derived from sewage treatment plants in receiving waters has been increasing; thereby, their quality is greatly influenced by the treated wastewater and most of the time resulting in bioaccumulation of chemicals. Research is in progress to carry out a meaningful ecological risk assessment of certain groups of chemicals of concern in wastewater treatment plant
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 2 8 7 e4 3 1 0
effluent which may present a risk to the environment. So that upgradation of treatment processes to remove the chemicals of concern could be implemented in future. The inclusion of disinfection byproducts may be due to anticipated effect of water recycling, where the potential exists for a gradual accumulation of disinfection byproducts in both the recycled water and wastewater. Chlorinated phenols constitute an important class of pollutants because of their wide use in sludge reduction as protonophores. Due to their strong toxicity and odour emission, persistence in environment, and suspected carcinogen and mutagen to the livings, chlorophenol have posed serious ecological problem (Armenante et al., 1999). Similarly, sonification and all the thermally assisted processes result in high carbon footprint and excluded from the scope of this review Therefore, lesser the assistance of chemicals, better for the environment, Hence, a detailed account of various engineering/technological approaches to achieve the goal of minimum sludge generation/reduction through bio-oxidation and without any chemical assistance is given in this paper based on classification as; (a) Biological lysis-cryptic growth (High oxygenation), (b) Biological uncoupling of metabolism (High So/Xo, OSA), (c) Maintenance Metabolism (MBR), and (d) Eco-manipulation (Predation). Since only in-place oxidation of excess sludges being natural process without any external and extraneous intervention is taken for discussion, therefore chemical uncoupling, has been left out. Moreover, enough has been reviewed on sludge reduction, still the progress in the last few years is missing, and hence, special emphasis is given herewith to highlight the efforts of the last five years.
2.1.
Lysis-cryptic growth
All cells have a plasma membrane, a protein-lipid bilayer that forms a barrier separating cell contents from the extracellular environment. Lysis (Greek, means to separate) refers to the death of a cell by breaking of the cellular membrane through different mechanisms that compromise its integrity. A solution containing the contents of lysed cells is called a “lysate”. Cell lysis is the first step in cell fractionation, which will release cell contents into the medium, thus providing an autochthonous substrate that contributes to the organic loading, microbial metabolism and a portion of the carbon is liberated as products of respiration and results in a reduced overall biomass production. The biomass grew on organic lysate is different from growth on original substrate, and therefore termed as cryptic (Greek word 徒ryptos means hidden) growth as it is not present from the beginning (Mason et al., 1986). It consists of lysis and biodegradation, where the former does not occur under normal conditions, however once lysed, it becomes easy for the living cells to biodegrade the lysed cells, therefore lysis is the rate-limiting step of lysis-cryptic growth, and an increase of the lysis efficiency can lead to an overall reduction of sludge production. The concept of cryptic growth was first introduced by Ryan (1959), followed by Mason’s (1986) PhD thesis on microbial death, lysis and cryptic growth. Mason and Hamer (1987) and Mason et al. (1986) on cryptic growth in Klebsiella
4291
pneumoniae in batch culture characterized sludge production yield of 0.33 g biomass/gCOD as against 0.56 g biomass/gCOD without cell lysis. The biodegradation of the cell wall is cited as the rate-limiting step and to increase it, several physicochemical and biological processes individually or in combination can be used in downstream processing. Several methods which have been applied so far for sludge disintegration summarized by Wei et al. (2003a) are herewith reported; (i) thermal treatment in the temperature range from 40 C to 180 C (Kepp et al., 1999; Barjenbruch et al., 1999), (ii) chemical treatment using acids or alkali (Tanaka et al., 1997,), (iii) mechanical disintegration using ultrasounds, mills, and homogenizers (Baier and Schmidheiny, 1997; Kopp et al., 1997; Camacho et al., 2002; Nolasco et al., 2002; Tiehm et al., 1997, 2001; Chu et al., 2001 and Onyeche et al., 2002), (iv) freezing and thawing (Chu et al., 1999), (v) biological hydrolysis with enzyme addition (Guellil et al., 2001), (vi) advanced oxidation processes such as wet air oxidation, using hydrogen peroxide and ozone (Weemaes et al., 2000a, 2000b; Shanableh, 2000 and Neyens et al., 2003c), and (vii) combination ways such as thermo-chemical treatment (Saiki et al., 1999; Neyens et al., 2003a, 2003b), combination of alkaline and ultrasonic treatment (Chiu et al., 1997). Ozonation has been successfully used on full scale (Yasui et al., 1996; Ahn et al., 2002; Egemen et al., 1999, 2001 and Deleris et al., 2002), Chlorination (Chen et al., 2001; Saby et al., 2002), integration of thermal/ultrasonic treatment and membrane (Canales et al., 1994; Chu et al., 2001), integration of alkaline and heat treatment (Rocher et al., 1999, 2001) are the other techniques in application for reducing sludge production, but on laboratory or pilot scale. The sludge reduction potential by ozonation, chlorination, thermo-chemical treatment, and high DO is reported as 100, 65, 60 and 25% respectively (Wei et al., 2003a). In abstract, the success of application of lysis-cryptic growth for sludge reduction depends on efficiency and economy of lysate production and till date no commercial success has been achieved except application of ozonation at full scale (Wei et al., 2003a).
2.1.1.
High oxygenation
Aerobic excess sludge digestion as a separate unit process is in application since ages. The three most commonly proven variations are; conventional, high-purity oxygen based and autothermal assisted digestion. However, excess sludge reduction by aerobic digestion on its place of production itself has been conflictingly reported in the literature from nil to 66% (Humenick and Ball, 1974; Roques et al., 1984; Sengewein, 1989; Metcalf and Eddy, 2003; Mudrack and Kunst, 1991; Bitton, 1994). McWhirter (1978) and Boon and Burgess (1974) reported 54% and 60% less yield with pure oxygen in comparison to air system, respectively. Abbasi et al. (2000) showed that a rise of the DO concentration from 2 to 6 mg/L led to about 25% sludge reduction at the sludge loading of 1.7 mg BOD5/mg MLSS d. The increase of the DO in the bulk liquid led to a deep diffusion of oxygen, which subsequently caused an enlargement of the aerobic volume inside the flocs, as a result, the hydrolyzed microorganisms in the floc matrix could be degraded and thus sludge
4292
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 2 8 7 e4 3 1 0
quantity was reduced, has been cited as a strong reason of increased sludge reduction. This has been substantiated by quantifying floc break-up due to mixing and oxygen and substrate profile inside the core of the sludge particle of 0.35 mm diameter size. This was accomplished by inducing nitrogen gas to keep the constant amount of diffused gas and consequently velocity gradient while increasing DO from 2 to 6 mg/L. The results showed 10% of the excess sludge reduced due to break-up of flocs as a result of the degree of mixing and remaining 12% was the result of elevated DO concentration in the mixed liquor (Abbasi et al., 2000). In conventional activated sludge (CAS) processes the oxygen transfer yields range from 0.6 to 4.2 kgO2/kW h depending on the methods of aeration, moreover, maintaining high DO in bulk liquid would further increase aeration cost very sharply, it is needless to mention that aeration itself costs more than 50% of the total energy consumption (Low and Chase, 1999a). Contrary to above, Wilen and Balmer (1999) concluded compact flocs at high and low effluent turbidity while working on effect of DO on structure and size of sludge flocs. However, Liu and Tay (2001) in their review rightly mentioned that the cell surface hydrophobicity, microbial activity and exopolymer production are linked to DO level in the reactor (Roques et al., 1984; Mishima and Nakamura, 1991; Palmegren et al., 1998; Pena et al., 2000). The role of DO as an uncoupler between anabolism and catabolism cannot be ruled out subject to further research. It is also mentioned that low DO favours growth of filamentous organism that cause sludge bulking (Richard et al., 1985; Nowak et al., 1986). On the contrary, activated sludge process operating on high DO can efficiently repress development of filamentous organisms in the aeration tank (McWhirter, 1978; Bitton, 1994; Abbassi et al., 2000). The choice of using air or pure oxygen has also been excercized subjective to degradability of excess activated sludge from a wastewater treatment plant at different temperatures (Zupancic and Ros, 2008). Considering the overall input in the digestion, there are advantages and disadvantages to oxygen aeration. Benefield and Randall (1980) reported number of advantages offered by pure oxygen process as compared to conventional air aeration process, such as ability to maintain a higher MLVSS concentration in the aeration tank; better sludge settling and thickening; lower net sludge production; higher oxygen transfer efficiency per horsepower and more stable operation. McWhirter (1978) observed that the growth yield in purified oxygenation activated sludge process can be reduced by 54% compared to conventional system, even at high sludge loading rates. Boon and Burgess (1974) reported 60% reduction in growth yield under similar SRT, utilizing non-purified air. Wunderlich et al. (1985) showed reduced sludge production from 0.38 to 0.28 mg VSS/mg COD removed as the SRT increased from 3.7 to 8.7 days in high-purity oxygen activated sludge system. These results indicated that the pure oxygen aeration process operated at a relatively longer SRT is more efficient in reduction of excess sludge production. Consequently, high oxygen process shows great industrial potential for minimization of excess sludge production as well as in improvement of system operation. However, economicefficiency and energy-balance calculations should be taken
in to consideration as important tool for performing the costbenefit analysis of a disintegration process.
2.2.
Uncoupling metabolism
The basic law of conservation of energy holds well in the biochemical pathways too, where the energy is transferred from electron donor to the microbes to perform different kind of activities. The reaction often starts with formation of electron carriers’ flavin adenine dinucleotide and nicotinamide adenine dinucleotide. These carriers during the oxidative phosphorylation give up electron and release proton. The proton during its discharge outside the cell membrane creates a charge imbalance and a pH gradient across the membrane. This is called proton motive force (PMF). The chemical energy stored in the proton gradient is used by the cells for adenosine triphosphate (ATP) formation from adenosine di-phosphate (ADP) apart from other products. The ATP thus formed during catabolism transfer it to cells for anabolism namely synthesis, maintenance and motility. The PMF is basically the driving force for the transfer of energy from catabolism to anabolism and is termed as energy coupling through rate limiting respiration (Rittmann and McCarty, 2001). The uncoupling is therefore short circuiting the PMF to restrict ATP formation, whilst simultaneously substrate oxidation (Stryer, 1988). This result in declined observed growth yield of biomass without reducing the removal rates of organic pollutants in biological wastewater treatment and may therefore provide a direct mechanism for reducing sludge production. The documented conditions of uncoupled metabolism are the presence of inhibitory compounds, heavy metals, abnormal temperatures, limitation of nutrients, excess energy source, and exposure of sludge to cyclic change in ATP content (Liu and Tay, 2001; Abbasi et al., 2000; Stouthamer, 1979 and Chudoba et al., 1992a). As mentioned in the beginning only last two are within the scope of unassisted uncoupling and hence would be discussed later in detail.
2.2.1.
Initial substrate to biomass ratio (So/Xo)
Chudoba et al. (1991) and Liu et al. (1998) observed decline in sludge growth at high initial substrate concentration to the initial biomass concentration (So/Xo as COD/biomass) in batch cultivation of mixed cultures as detailed in Table 1. The concept though modelled and verified experimentally (Liu, 2000) has yet to find a place in its engineering application in wastewater treatment plants, where the actual So/Xo ratios are 0.01e0.13 mg COD/mg MLSS for sewage (Chudoba et al., 1991c). Moreover, research on the process initiated by Chudoba et al. (1991) has not been globalized, and was only followed by Liu (1996, 2000) and Liu et al. (1998). Wei et al. (2003a) concluded that maintaining high food to microorganism ratio to achieve low sludge production may cause deterioration in effluent quality and would necessitates further treatment of wastewater (post-treatment) to meet the desirable levels of effluent organic matter. This would result in high capital and operation costs, and therefore makes it fit only for the biological treatment of high strength industrial wastewaters. No further progress in research was made since then.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 2 8 7 e4 3 1 0
4293
Table 1 e Summarized experiences with biological uncoupling metabolism. Reference(s)
Process
Highlights
Chudoba et al. (1991)
High So/Xo
Liu (1996), Liu et al., 1998, Ghigliazza et al. (1996)
High So/Xo
Liu, 2000
High So/Xo
Chudoba et al. (1992a,b) Chen et al. (2003)
OSA OSA
Saby et al., 2003 and Chen and Leung, 1999; Chen et al., 2003
OSA
Jun et al., 2004a
OSA
Energy dissipation (at So/Xo ratio 5 mg COD/mg MLSS) by leakage of ions through the cell membrane, which weakens the potential across it and thus uncouples oxidative phosphorylation. Substrate sufficiency in terms of high So/Xo cited a reason of lower yield owing to energy spilling due to uncoupling by excess substrate. The reduced sludge growth is quantified in terms of energy uncoupling coefficient (Eu) on a scale of 0e1.0. The kinetic model based on carbon partition between energy spilling-associated and growth-associated metabolisms at different So/Xo ratios provides mechanism of sludge reduction. Sludge yield were significantly lower by about 38e54% compared to CAS. Ruled out energy uncoupling, domination of slow growers, and SMPs as possible cause of sludge reduction and, instead low ORP is cited as the sole reason for the same. In a modified OSA process with a membrane module in the aeration tank, at a controlled ORP of 250 mV in the anaerobic tank, the excess sludge was reduced by 36%, and 58% compared to þ100 mV ORP or compared to CAS process. Increased soluble COD in the anaerobic tank, causing cryptic growth during aerobic phase caused low sludge yield in the overall OSA process. Established symbiosis between aerobes and methanogens by controlling DO from 0 to 0.5 mg/L in the aeration tank to reduce sludge production and higher removal efficiencies of total COD of 93%. The lower observed yield coefficients of 0.28 gVSS/ gCOD was achieved as against 0.41 and 0.37 in intermittent aeration and CAS reactors.
2.2.2.
Oxic-settling-anaerobic (OSA) process
Oxic-settling-anaerobic (OSA) process has well established the fact that anaerobic followed by aerobic in a cyclic order reduces the sludge through promotion of catabolism and demotion of anabolism by uncoupling the two reactions. Aerobic microorganisms capture energy in the form of ATP released from oxidation of organic content. The same microorganisms are unable to produce required energy when exposed to anaerobic conditions under severe food limiting condition and as a result consume their conserved ATP. Again on return to aerobic condition, augmentation of their depleted ATP reserves become first priority as against synthesis of new cell mass or anabolism. This in other words promotes catabolism and demotes anabolism by uncoupling the two reactions, thereby inducing sludge reduction (Chudoba et al. (1992a,b). Similarly, altering anaerobic and aerobic environment causes death of obligate aerobic and anaerobic microorganisms. Thus produced lysed microorganisms release the intracellular matters, which can be degraded by various extracellular enzymes. The first concept of anaerobic exposure to returned aerobic sludge was probably given by Westgarth et al. (1964). It consists of three tanks namely aeration tank followed by a settling tank and an anaerobic tank situated in the return sludge line to create an alternative oxic and anaerobic cycle, creating thereby a fasting/feasting condition to the exposed biomass, which ultimately caused reduction in excess sludge production. The OSA process has subsequently been investigated by various researchers such as Chudoba et al. (1991, 1992a,b); Ghiglizza et al. (1996); Copp and Dold (1998) and Chen et al. (2000, 2001a,b, 2003) (Table 1). Similar to explanation given by Chudoba et al. (1992a,b) regarding the process of sludge fasting/feasting the same was given by Chen and Liu (1999). Chen et al. (2003, 2001a,b) and
Saby et al. (2003) systematically investigated energy uncoupling, domination of slow growers, and SMPs as possible cause of sludge reduction and ruled out all of them, instead low ORP is cited as the sole reason for the same. Encouraged by the simultaneous nitrificationedenitrification at lower rates by controlling DO level below 0.5 mg/L in oxidation ditches (Trivedi and Heinen, 2000), Jun et al. (2004a,) established a similar symbiosis between aerobes and methanogens, a major group of archaea. The experiments were performed in an intermittent aeration reactor (I/A), an intermittent aeration reactor dosed with archaea solution once a day (I/A-arch) and a conventional activated system (CAS) (Table 1). However, validation of these theories is needed to substantiate the mechanism of sludge reduction in the OSA process and symbiosis between aerobes and methanogens. Citing ORP as the sole reason behind reduced sludge production by Chen et al. (2003) lack supporting mechanism. Nevertheless, it appears OSA process can successfully handle the problem of high excess sludge production, which is particularly important when handling high strength waste streams in order to maintain the economical feasibility.
2.3.
Maintenance metabolism
The energy obtained and captured in the form of ATP during biological oxidation is used by the microbial cells for their maintenance followed by synthesis. Therefore, long sludge age results in increased energy consumption for maintenance, which leaves less energy for cell synthesis. In other words, long SRT causes reduction in sludge loading rate or low food to microorganism (F/M) ratio reduces sludge production (van Loosdrecht and Henze, 1999). In case of higher concentration of reactor biomass and limited substrate, to the extent that it is just sufficient to cater the energy requirement for
4294
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 2 8 7 e4 3 1 0
maintenance, than in a dynamic situation no energy would remain available to sustain the growth of microorganisms anymore. The decreased observed yield consequent upon increased energy requirement induced by high concentration of NaCl in a culture of Saccharomyces cerevisiae due to more cell rupture was observed by Watson in 1970. Contemporarily, Lawrence and McCarty (1970) described that how observed yield (Yobs) decreased with increased SRT (qc) with reference to maximum theoretical or true yield Yobs ¼
Ym ð1 þ kd qcÞ
(2.1)
Pirt (1975) subsequently related observed yield (Ys or Yobs) and maximum or true yield (YG) to SOUR for maintenance requirement (qm), where; 1 q 1 ¼ mþ m Ys YG
(2.2)
Low and Chase (1999b) quantified the biomass production per unit volume based on their proposal that the mass balance on the utilization of the energy source is the sum of the substrate utilized by anabolism and the biomass for satisfying maintenance requirements. This justified the assumption that cells preferentially satisfy the energy requirements associated with maintenance functions and that additional cell synthesis occurs using the remaining substrate available. It was also assumed that with a constant supply of substrate and a situation where growth is substrate limited at a constant level, substrate uptake rate (rs) is also constant; 1 rs ¼ $rx qm X YG
(2.3)
Thus the biomass production per unit volume may be represented by; rx ¼ YG $rs qm X
(2.4)
Therefore, the constant supply of limiting substrate condition preferentially satisfies the energy requirements associated with maintenance functions and that additional cell synthesis occurs using the remaining substrate available, the substrate uptake rate (rs) is also constant. Thus, the rate of substrate uptake specifically for cell synthesis per unit volume (rsG) is given as; rsG ¼ ð1=YGÞ$rx ¼ rs þ qm X
(2.5)
It seems logical that with Pirt’s (1975) allocation of energy under such a condition of biomass sedimentation and recycle, when the biomass concentration within the reactor is divorced from biomass production, the meaningful determination of specific growth rate (m) is difficult. Hence, Low and Chase’s (1999b) model which excludes the specific growth rate, but incorporates the biomass concentration, would provide a more suitable description of a system with partial biomass recycle, as the biomass concentration is a function of the sludge return rate and therefore is an accessible control parameter. The model was validated using data presented by Bouillot et al. (1990) (Table 2). As it is impossible to increase the sludge concentration significantly in CAS processes by means of settling even assisted with coagulants, therefore, similar results were
obtained when biomass was concentrated by membrane separation in membrane bioreactor (MBR). Catering the need of comprehensive treatment, onsite sludge handling is difficult, but essentially required. Therefore, the core idea is to artificially restrain the normal sludge growth within an acceptable level through establishing a preferable long SRT system independent of HRT, in order to approach zero excess sludge discharge. Membrane bioreactor (MBR) is the application of concept of increased energy consumption in cell maintenance, leaving little for growth, ideally to attain no sludge from wastewater bio-oxidation. Yamamoto et al. (1989); Mu¨ller et al. (1995); Wagner and Rosenwinkel (2000); Visvanathan et al. (2000); Rosenberger et al. (2002) and Witzig et al. (2002) showed that the MBR caused little/zero sludge production, but expensive in terms of energy requirements. It has also been successfully applied in full-scale plants (Churchouse and Wildgoose, 1999). All the claims of excess sludge free performance of MBR depend on physical and rheological properties of the retained sludge, namely diluted sludge volume index, the capillary suction time, the specific resistance to filtration and the compressibility. The capillary suction time test involves the measurement of time to move a volume of filtrate over a specified distance as a result of the capillary suction pressure of dry filter paper. It provides information regarding the ease of separating the water portion from the solids portion of sludge (EPA, 1987). Specific resistance to filtration test was performed by recording the volume of filtrate versus time when applying a negative pressure (0.5 bar) on a sludge sample, and filtering it through standard filter paper (EPA, 1987). A bench scale MBR with complete solid retention upto 22e23 g TSS/L was investigated and compared for about one year with CAS. The sludge showed substantial similarity in terms of dewaterability. The application of Ostwald model (Blair et al., 1939) indicated less increase in apparent viscosity in proportion to increasing solid concentrations. Other parameters, such as the diluted SVI (Giokas et al., 2003) and the reduced hysteresis area appeared to be scarcely dependent on the MLSS concentration, although larger hysteresis areas were observed in the period of rapid biomass buildup, suggesting a possible link to the sludge growth (hysteresis area provides an indication of the reversibility of the effects of a shear stress, in other words, the energy provided to the system by increasing the shear rate is used partly to overcome the resistance opposed by viscosity and partly to break the links between particles flocs). Based on calculated Reynolds number (3000e600), energy consumption for mixing resulted in a limited increase of energy requirements by 25e30% at increased solid concentration from 3 to 30 g TSS/L. These results obtained by Pollice et al. (2007) on account of similarities in the basic properties of MBR and CAS sludges pointed out that MBR can be efficiently adopted not only for providing high-quality effluents, but also for reducing sludge handling and disposal costs with respect to typical activated sludge processes. The attempt of Pollice et al. (2007) is interesting in view of limited information available on sludge characteristics. An energy optimization may not be obtained by a better understanding of the fouling phenomena of the membranes. The essential requirement of maintenance of higher MLSS at
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 2 8 7 e4 3 1 0
4295
Table 2 e Summarized experiences with biological maintenance metabolism. Reference(s) Low and Chase (1999b)
Process
Highlights
MBR
The biomass production is the sum of the substrate utilized by anabolism and the biomass for satisfying maintenance requirements. Increased biomass from 1.7 to 10.3 g/L, reduced biomass by 44% at qm ¼ 0.023 0.005 g substrate/g biomass.h and YG ¼ 0.361 0.022 g biomass/g substrate. Bouillot et al. (1990) Coagulant assisted Acetate fed chemostat resulted in maintenance quotient qm being 0.027 g substrate/g biomass h and sludge thickening 5% decrease in the observed yield. 12% decrease in the observed yield was found when the F/M decreased from 0.22 to 0.11 g substrate/g biomass h as the biomass increased from 3 to 6 g/L in a coagulant assisted thickened recycled sludge. Chaze and Huyard MBR Reduced sludge production at SRT 100 days compared to 50 days. (1991), Cicek et al. (2001) MBR Similar results at long SRTs ranging from 2 to 30 days in MBR treating synthetic wastewater without causing any significant influence on COD removal and nitrification. Mu¨ller et al. (1995) MBR Reduced sludge to just 6% of the carbon fed to a cross flow MBR when the sludge concentration increased to 40e50 g/L. However, inorganic fraction in sludge increased to 23.5% from 21.6%. Rosenberger et al. MBR Low sludge production (0.002e0.032 kg/d) without sludge discharge for one year at 15e23 g SS/L and (1999; 2000) low F/M ratio of 0.07 kg COD/kg MLSS d. These studies mentioned maintenance metabolism as the main cause and ruled out role of protozoa/metazoa. Deterioration of sludge properties mainly; fragile flocs, viscous sludge, high SVI, poor settling, difficult dewatering. High SRT caused poor oxygenation, increased aeration cost and membrane fouling. Laera et al. (2005). MBR Near zero sludge yield as mmax/b approaching unity (¼ 1.5 0.6) at 41000 d SRT in 180 days with constant VSS of 16e18 g/L, at an OLR below 0.1 g COD/g VSS d and SOUR of 2e3 mg O2/g VSS h, with complete nitrification and good COD removal. Lobos et al. (2008) MBR and OSA Clubbing of uncoupling high F/M ratio (upto 12 g COD/g MLVSS d) and OSA in two immersed MBRs, one operating in sequential and another in continuous mode. The biomass production was half in the continuous MBR operation, because of substrate limitation. Sun et al. (2007) MBR High cell maintenance resulted low sludge yield and high decay rate of 0.115 g VSS/g COD and 0.024/ day, at higher sludge concentration of 14.5 g/L at prolonged SRT for 300 days. 99% removal efficiencies of both COD and TOC were attributed to the bio-fouling layer, which played as a prefilter to the membranes. MLVSS/MLSS ratio of 0.9 indicated that no accumulation of inorganic compounds occurred at high SRT due to production of larger than pore size hydrolysis or enzymatic solubilization compounds. This was inconsistent with other reported MBR systems (Huang and Qing, 2001; Han et al., 2005). Low SOUR at high SRT indicated lower energy requirement by the microorganisms due to maintenance metabolism. It was considered a controversial phenomenon owing to no reduction in the biological capability of the sludge despite changes in composition of mixed liquor caused by prolonged SRT. Heran et al. (2008) MBR Complete sludge retention is susceptible to induce high biomass concentration and consequently low F/M ratio, fed with easily biodegradable organic substrate. High efficiency, no accumulation of mineral solids and no decline of the membrane performance were observed. The yield reduction was 0.041. The use of ASM model provided a good simulation without integrating inert soluble COD produced by microorganism activity ( fps ¼ 1.1%). But the model overestimated the oxygen demands due to high lysis products, at high SRT. Wang et al. (2006) Nylon mesh Study was conducted on SBR equipped with a 100 mm pore size nylon mesh on a stainless steel frame Bioreactor and spacer, capable to increase MLSS upto 32 g/L at 10 days HRT resulting in 83.9% decline in excess sludge. The results indicated low SS, COD and colour. However, the decantation period was influenced by the extracellular polymers of the microbes in the reactors. Xing et al. (2006) Inclined-plate Another version of MBR (0.4 mm pore size, Mitsubishi Rayon SUR134 membranes) used to improve the MBR (iPMBR) anoxic condition by confining high MLSS sludge. The performance of a pilot iPMBR treating municipal wastewater was investigated at an HRT of 6 h for. The respective average removals of COD, ammonia nitrogen and turbidity were 92.1, 93 and 99.9% at 6 h HRT without any sludge discharge in 123 days. Xie et al. (2008) MBR with simple A submerged flat metal aerobic membrane bioreactor performed reasonably well by online back sponge cleaning flushing or simple sponge scrubbing and the system ran stably about 115 days at the permeate flux of 0.8e1 m3/m2$d without changing the membranes. The results showed that the mean COD and TN removal efficiencies were 96.69% and 32.12% under aerobic MBR mode, and 92.71% and 72.44% under A/O-MBR mode. A/O-MBR mode also resulted in perfect sludge reduction, but with serious irreversible membrane fouling. Lin et al. (2009) Gravel contact The GCOR provided an integrated aerobic, anaerobic and anoxic environment to form the oxidation reactor substrateebacteriaealgaeeprotozoaemetazoan food chain. The reduced excess biomass took place (GCOR) due to grazing by protozoa and metazoan on dead bacteria.
high SRT in the MBR increases fouling and energy demand and decreases sludge production. While less SRT decreases MLSS and associated energy consumption but increases sludge production, which further require energy for its treatment.
Therefore, MBR operation could be optimized between these two parameters on account of energy consumption and membrane fouling. Apart from suspended and colloidal fractions of the waste, autochthonous (i.e. microbial in origin)
4296
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 2 8 7 e4 3 1 0
organic solutes either EPS or SMP are the key foulants. EPS are extremely heterogeneous, encompasses carbohydrates, proteins, nucleic acids, (phospho) lipids and other polymeric compounds found in the intercellular space of microbial aggregates. The carbohydrate components of the SMP in some MBR membrane fouling studies have been correlated to promote fouling under certain conditions (Simon Judd, 2007). However, the methods used for assaying carbohydrates and proteins during EPS fractionation are not universally agreed, and consequently different studies have identified different components with the highest fouling propensity. Rosenberger et al. (2006) suggested that at lower SRTs it is the polysaccharide colloidal matter in the SMP that is primarily responsible for fouling. But, Drews (2007) concluded that fouling cannot be attributed to any one specific constituent of the mixed liquor for higher SRTs (>20 days). Drew’s review (2010) linked fouling complexity on many other parameters such as Calcium concentration, pH, temperature, which affect properties of SMP compounds which are relevant for fouling, like size, shape, charge, gelling potential and hydrophobicity. It is not only the supernatant SMP concentration that appears to be a generally useful indicator for fouling propensity, but spatial distribution of polysaccharides in the biocake on membranes has also to be taken into account. In order to universal application of MBR technology, a standardization of methods and an agreement on “good operating practice” is highly desirable. Use of modified biopolymers with a net cationic charge could also be a possible remedy to fouling. Membrane Performance Enhancer (Nalco’s patented Perma care Product) was successfully investigated while treating different wastewaters viz. Leachate, Food industry, Dairy industry/cheese industry and Paper mill (Wozniak, 2010). The advantages offered by performance enhancer are; reduction of the transmembrane pressure, improvement in the filtration rate, increasing the permeability, reduction of foam, reduction of colloidal EPS, improvement in the effluent quality, reduction in cleaning frequency, increase in the life of the membranes and no impact on the oxygen consumption. Consequently, it can save about 15e20% of the investment costs. This technology is not to clean the membranes, but it keeps the membranes clean during operation and increases the filterability of the cake layer, however, first a chemical cleaning must be carried out (Wozniak, 2010). The foregone discussion indicates that the main stumbling block in sludge thickening and reduction appears to be the membrane structure and their fouling. An excellent review covering almost all the aspects of membranes materials and their fouling, which is an important key to their application for the intended use of wastewater treatment including sludge reduction has been done by Meng et al. (2009). It was one of the top 25 hottest article of Water Research, theEarth and Planetary Sciences journal from April to June 2009. Membrane fouling and the high cost of membranes are main obstacles for wider application of MBRs. Over the past few years, considerable investigations have been performed to understand MBR fouling in detail and to develop high-flux or low-cost membranes. In this paper, recent advances of research on membrane fouling and membrane material in
MBRs were reviewed. From the viewpoint of fouling components, the fouling in MBRs is classified into three major categories namely; bio-fouling, organic fouling and inorganic fouling. The results obtained from recent investigations on bound EPS, SMP, filamentous bacteria and hydrodynamic conditions are updated. The article is itself a review and hence cannot be pre´cised further, a detailed reading is advised, and however, the areas suggested for future study on membrane fouling are reproduced below: (1) Studies on membrane fouling mechanisms should focus on identification and characterization of membrane foulants (i.e. chemical and biological components of foulants, bacteria community of the foulants). Cake formation, pore blocking, and (EPS)/SMP adsorption on/within the membranes are all important, but emphasis could be given to the interaction and interrelation between these mechanisms and sludge characteristics. (2) Development of procedures for the visualization and characterization of membrane fouling in MBRs. Direct monitoring and in-situ techniques will offer more useful information about the formation of membrane foulants. (3) Standardization of fouling characterization methods and an agreement on “good operating practice” is highly desirable. (4) Development of more effective and easy methods to control and minimize membrane fouling. Generally, removable fouling is controlled by creating shear stress on the membrane surface. Although air bubbles are used to promote shear stress and to enhance the membrane flux, they also have strong impact on biomass characteristics. Moreover, enforced aeration will need more energy. Research should be directed to optimization of the current coarse aeration methods for submerged membrane modules. Lastly, alternative filtration concepts to limit the deposition of foulants onto the membrane surface should also be developed. (5) Study of the fouling behaviour in full-scale MBR plants in order to reflect the real fouling behaviour. (6) Development of novel membrane modules for MBRs to reduce their capital costs and enhance their hydrodynamic conditions. (7) Modelling of mass transfer and membrane fouling by mathematical approaches such as computational fluid dynamics, Monte Carlo simulation, fractal theory, artificial neural network. In other words, a comprehensive investigation should be performed to understand, control and reduce membrane fouling, especially avoiding severe fouling; it is just like a systematic physical examination on a person to understand his/her health condition and to avoid the occurrence of illness, especially fatal diseases. In recent years, there are considerable investigations about the impacts of membrane materials, pore size, hydrophilicity/ hydrophobicity, etc., on membrane fouling; however, most of the recent investigations are focused on the application of low-cost filters to substitute the membranes, modification of membranes to enhance their hydrophilicity and use of dynamic membranes to improve the performance of membranes or low-cost filters. In the future, to our knowledge, the study on membrane materials in MBRs should still focus on development of anti-fouling membranes or
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 2 8 7 e4 3 1 0
modification of current membranes and enhancement of the performance of low-cost filters by modifying their surface properties (Meng et al., 2009). Since number of variables is high in every version of MBR application, therefore, a technique was needed to assess the performance on a common basis. Madoni’s (1994) Sludge Biotic Index (SBI) is the one, which is widely used specific biotic indexes to assess and predict the performance of the treatment system, particularly in view of the abundance and diversity of microfauna (Are´valo et al., 2009). In this study, the SBI index was applied to a pilot-scale membrane bioreactor equipped with polyvinylidenefluoride hollow fibre membranes (0.04 mm pore size). Two different sludge retention times (SRT) of 25 and 35 days were assayed, with a constant hydraulic retention time of 30 h. The experimental results showed a constant predominance of small flagellates, carnivorous ciliated protozoa and rotifers in assays with 35-day SRT, independently of effluent quality. However, continuous changes in micro-biota were observed, with a clear tendency for the SBI to increase over time as the sludge became more stable. Therefore, although it was not possible to establish an association between SBI and effluent quality for MBR activated sludge, the stabilization of sludge may be related to SBI and also in the identification of problems arising in the process, such as incorrect oxygenation and the presence of toxic substances. For this reason, there is a need for further experimentation and adaptation of this controlling index, in order to guarantee its successful application to MBR systems (Are´valo et al., 2009). The gravel contact oxidation reactor (GCOR) filled with crushed stone globular aggregates as carriers, has been demonstrated capable of reducing the excess sludge effectively through maintenance metabolism in some pilot and small scale engineering studies. In order to evaluate the variation and structure of the microbial community and their functions to excess sludge reduction in GCOR, a conventional activated sludge reactor was studied as a comparison (Lin et al., 2009). The best part of the article is a clear picture of phylogenetic diversity and population of microorganism grew in the porous carriers and on their biofilm in a GCOR, which was little known in-situ until then. Out of the 30 species grew on GCOR media the most abundant bacteria were those related to the b-Proteobacteria group followed by those related to g-Proteobacteria and then those related to phylum CFB. In the CAS the order was g-Proteobacteria, followed by beProteobacteria and then CFB. Shannon’s diversity index (H ) was as higher as 3.41 for diversity of bacteria extracted from carrier samples in GCOR than 2.71 for sludge sample in CAS. Species evenness (E, the relative abundance or proportion of individuals among the species) for the isolates from GCOR and CAS samples was 0.97 and 0.96, respectively. The total bacterial DNA concentration at normal operation on the carriers of GCOR was 8.98 105 mg/mL, about two times more than in CAS 4.67 105 mg/mL, which exhibited higher enrichment of bacteria. At the same time, the most representative eukarya were protozoa both the reactors, which were 15 no. per 20 mL in GCOR and 15 no. per 20 mL in CAS, next abundant group were attachment plants 10 no. per 20 mL in GCOR and 4 no.
4297
20 mL in CAS, respectively. Rotifers and copepoda belonging to metazoan were only present in GCOR (8 no. per 20 mL for rotifers and 8 no. 20 mL for copepoda). MLSS in the sediment tank of GCOR was only 4.5 mg/L. It was 25 times less than 115.4 mg/L in CAS at the time of two reactors’ normal operation. Microbial diversity (H) and their population difference (E) both in the carriers of GCOR and sludge of CAS indicated that diverse microbes and a large amount of biomass attached on the carriers were probably one of the main reason of excess sludge reduction in GCOR, where microbial community varied at different stages of the microbe incubation and normal wastewater treatment (Lin et al., 2009). It appears more genomics related studies are needed to focus on the phylogenetic diversity and abundance at different stages of bioreactors, and the microbial growth rate in association with substrate degradation, so as to more accurate exploration of microbial ecology fundamentals on the subject of excess sludge reduction in the bioreactors. The predominance of bProteobacteria as major constituents of the microbial community structure was also cited by Zubair et al. (2007).
2.3.1.
Cost and energy considerations in MBR
Information on energy consumption and consequently cost analysis of MBR plants is scarce in scientific literature. The sludge treatment cost minimization increases the aeration cost, therefore, optimum point could be identified between these two parameters. A methodology to obtain the most economical operational condition of membrane bioreactor was developed by Yoon et al. (2004), through which sludge production rate can be quantitatively estimated as functions of HRT and MLSS. When either target MLSS in bioreactor or HRT increases, sludge production rate decreases and aeration requirement increase. By summing the decreasing sludge treatment cost and increasing aeration cost, total variable operational cost is obtained. The energy costs in a Kubota flat sheet MBR pilot plant located in Southeast Spain were monitored at two fluxes, 19 and 25 L/m2.h for one year (Gil et al., 2010). Based on electricity price of 0.0806 V/kW, the prices of the treated water for the two fluxes were 0.49 and 0.39 V/m3 and the total consumptions were 6.06 and 4.88 kW h/m3 for above fluxes respectively. This energy consumption was on a much side as compared to reported value of 0.6e0.8 kW h/m3 of treated water. Coarse bubble aerator, followed by the mixer consumed about 50% and 25% of the total energy. But, the basis of energy data and cost evaluation has not been properly spelt out and statistically correlated in the paper. Fenu et al. (2010) validated a calibrated dynamic biological ASM model of a full scale MBR and CAS reactors in the form of two lanes at the treatment plant of Schilde. The overall energy consumption found to be 0.64 kW h/m3 of permeates in MBR lane in comparison to 0.3 kW h/m3 in CAS lane. The higher MLSS maintained in MBR implied significant mixing energy costs and reduced oxygen transfer, the smaller floc size did not reflect in a significant aeration energy saving. The impact of the filtration process on the overall energy consumption could be reduced if the coarse aeration flow would be better integrated in the biological process scheme of submerged MBRs. However this work indicated a minimal contribution of the coarse aeration flow to the biological oxygen requirements. Roels et al. (2010) identified
4298
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 2 8 7 e4 3 1 0
the following main energy consumers of the CAS system; 35e50% of the total energy accounts for coarse bubble aeration, 15e20% for influent flow, 10e15% for sludge recycle 5e15% for mixers. The overall filtration process, including the permeate extraction, the electrical heating of the cleaning in-place tank, the compressors for activation of the valves, accounted for more than 56% of energy consumption. The average energy cost of some municipal full scale MBRs ranges between 0.8 and 1.2 kW h per m3 (Mulder, 2009). The CAS in Schilde (Belgium) consumes 0.19 kW h per m3, whereas the MBR consumes 0.64 kW h per m3. The lower energy consumption in Schilde might be explained by an intensive cost reduction program started in 2006 (Garces et al., 2007). Judd (2007) prescribed two given below options for reducing energy demand in MBRs while reviewing fundamental facets of the MBR process; (i) use of ceramic membranes and (ii) anaerobic operation. Ceramic membranes are more fouling-resistant but are currently high in cost, which may expectedly be reduced with time owing to advances in fabrication techniques (Bishop, 2004). Submerged anaerobic MBRs may offer advantages over aerobic treatment because anaerobic operation demands least energy input in addition to energy generation via methane gas recovery. However, the advantages offered by this coupling of an anaerobic process with a membrane have not been quantified (Jefferson, 2007; Hu and Stuckey, 2006; Jeison, and van Lier, 2007). On account of cost of membranes, frequent replacement of membranes due to fouling, increased aeration owing to poor oxygen transfer, and taking the costebenefit scale into consideration, it is not always feasible to operate MBR with complete sludge retention or zero sludge bleeding in practice, and there must be a minimal rate at which excess sludge is wasted in order to keep an optimal range of sludge concentration in MBR. Moreover, the full-scale application documentation is only limited to one experience and still more studies on cost benefit analysis would justify MBR application.
2.4.
Eco-manipulation (predation on bacteria)
The principle of using microfauna to reduce excess sludge comes from the food chain. Materials and energy are dissipated (or lost) when they flow in the food chain, and hence the microfauna’s predation leads to sludge reduction. Most of the protozoa are aerobic organisms feed on organic sources including bacteria by the process called phagocytosis. Initially, their role was identified in effluent polishing by consuming dispersed bacteria by Curds and Cockburn (1970); Learner, 1979 and Gaudy and Gaudy (1981). Subsequent researchers have shown reduction in sludge production due to predation of bacteria by exploiting higher organisms such as protozoa and metazoa in the different variants of activated sludge process (Ratsak et al., 1996; Ratsak, 2001; Welander and Lee, 1994; Lee and Welander, 1994, 1996a,b; Rensink et al., 1996; Janssen et al., 1998; Ghyoot, 1998; Eikelboom, 1988). Rensink and Rulkens (1997) showed the grazing fauna in aerobic reactors as an Eltonian pyramid of numbers (Elton, 1935). The ecosystem in a biological wastewater treatment plant consists of bacteria, protozoa, metazoa, larvae of insects and arachnida (Curd and Hawkes, 1975), where bacteria are
the primary consumers, which themselves are consumed by protozoa and metazoa. It creates a food chain between each trophic level, linked by a predatoreprey relationship. During energy transfer from low to high trophic levels, energy is lost due to inefficient biomass conversion. Under optimal conditions the total loss of energy will be maximal and the total biomass production will thus be minimal Ratsak et al. (1996). The protozoa can be classified as ciliates (free swimming, crawling and sessile), flagellates, amoeba and heliozoa (Eikelboom, 2000). The metazoa consist normally of rotifers and nematode and occasionally Aeolosomatidae and Naididae. The predator population is normally dominated by protozoa in activated sludge system and metazoa in trickling filters. The presence of predators suppresses the growth of dispersed bacteria and results in favour of floc formation. As a result, major portions of the sludge remain unaffected by predation activity. This was overcome by making it a two-stage process by providing a short SRT completely mixed pretreatment tank to encourage the growth of dispersed bacteria in the absence of predators. Contrary to the conditions of the first stage, the second tank was conventional aeration tank with very high SRT (may be membrane assisted) to encourage growth of predators and consumption of earlier produced dispersed bacteria (Ratsak et al., 1994; Lee and Welander, 1996a; Ghyoot and Verstraete, 2000). In addition to the use of protozoa and metazoa as important indicators of process performance and efficiency of biological wastewater treatment processes, recently, many researchers have explored the potential of different kind of predators by manipulating the ecosystem as a biotechnological tool (Eikelboom, 2000; Eikelboom et al., 2001; Zhang, 2000; Luxmy et al., 2001; Lapinski and Tunnacliffe, 2003 and Wei et al., 2003b).
2.4.1.
Two-stage system
As mentioned above about the requirement of two stages for effective sludge reduction by predators, Ratsak et al. (1994) reported 12e43% of biomass reduction by employing the ciliated Tetrahymena pyriformis to graze on Pseudomonas fluorescens. Lee and Welander (1994, 1996a,b) tested the decreasing trend in sludge production, which was mediated by protozoa and metazoa. The high sludge yield of 0.17 g TSS/g COD removed in the acetic acid fed system was observed as the bacteria formed zoogloeal flocs, which protected them from grazing in the predator stage. Whereas, low sludge yield of 0.05 g TSS/g COD removed was observed in the system fed methanol as dispersed bacteria were obtained that were easily grazed by the protozoa and metazoa in the predator stage. Due to the mineralization activities of the predators, a significant release of nitrate (>7 mg N/L) and phosphate (>2.5 mg P/L) was observed in the effluent (Lee and Welander, 1996b). Similar study was performed on pulp and paper industry wastewater in a two-stage arrangement in both activated sludge and biofilm reactors were tested in second stage (Lee and Welander, 1996a). Results showed lower sludge yields of 0.01e0.23 g TSS/g COD removed as compared to 0.2e0.4 g TSS/ g COD removed in CAS processes treating the same wastewater. Growth of fast growing filaments, such as Sphaeroltilus natan and the formation of aggregates were encountered in the first stage, despite that their numbers were under control
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 2 8 7 e4 3 1 0
4299
Table 3 e Summarized experiences of sludge reduction by predation on bacteria. Reference(s)
Process
Lapinski and Tunnacliffe (2003)
Predation
Luxmy et al. (2001)
Predation
Rensink and Rulkens (1997) Predation and Ratsak et al. (1993)
Ratsak (1994)
Predation
Zhang (2000) and Eikelboom et al. (2001)
Predation
Wei et al. (2003b)
Predation
Elissen et al. (2006)
Predation
Hendrickx et al. (2009a)
Predation
Hendrickx et al. (2009b)
Predation
Liang et al. (2006a)
Predation
Highlights Bdelloid rotifers (‘‘leech-like wheel-bearers’’) are metazoans, which are ubiquitous in most fresh and wastewater habitats. A corona of cilia around the mouth opening draws suspended particles into mouth of rotifer. During grazing they consume several times their body weight per day. In addition they cause enhanced ‘‘bioflocculation’’ or a combination of both. The sludge reduction capacity of metazoa in an MBR and found insignificant sludge reduction at a metazoan population of 1000e2000 ind. per mL, instead greater abundances of metazoa were required, but metazoans were found to play an important role in fouling control of the membranes. In the activated sludge plants the ciliated protozoa were the dominant microfauna group in both the treatment systems, whereas metazoa, particularly Lecanidae rotifera were more abundant in the starch-enriched system. No significant differences in sludge production were found between the two systems. This was probably because metazoa levels in either system never exceeded more than 1500 ind/mL. Although the higher levels of metazoa in the starch-enriched system did not affect sludge settleability or production, they did have a great effect on the nitrogen removal efficiency (Puigagut et al., 2007). Predominance of oligochaetes worms especially Nais elinguis and Aeolosoma sp. and darkness preferring Tubificidae was observed. However, uncontrollability of Aeolosomatidae and Naididae i.e. their washout in effluent led to selection of Tubificidae for sludge reduction on different carriers. Two trickling filters one with lava slag and another with plastic media caused sludge reduction by 10e50% and 10e45% in the trickling filters compared with 10e15% and 10% without worms. The resulting sludge yield with and without Tubificidae was to 0.15 g MLSS/g COD and 0.40 g MLSS/g COD respectively. 25e50% sludge reduction by predominantly Nais elinguis followed by Pristina sp. and Aeolosoma hemprichicii worms was observed in one and a half year study. The number of worms varied both seasonally and among the aeration tanks. A major worm bloom resulted in a low SVI, lower energy consumption for oxygen supply and less sludge disposal. Sludge reduction was achieved using membranes to increase worm density. They found three species of worms namely; N. elinguis, Pristina sp. and Aeolosoma hemprichicii, but Aeolosoma sp. was predominant. It was contrary to the finding of Ratsak (1994) obtained in CAS. Wei et al. (2003b) found high worm density i.e. 2600e3800 Aeolosoma/mL mixed liquor in the MBR system, and resulted in a low sludge yield (0.10e0.15 kg SS/kg COD removed). It was found that worm growth in the CAS reactor was much better than in the MBR as the average worm density in the CAS reactor was 71 worms/mg VSS than 10 worms/mg VSS in the MBR. The prevalence of worms was 30 worms/mg VSS in CAS upto 172 days, however, worms did not produce naturally in MBR and their presence depend on sludge inoculation from the CAS. The impact of TSS, HRT, SRT, F/M; recycle ratio, temperature, pH and DO on the growth of worms in both the reactors was also correlated. Only sludge loading rate and SRT had no impact on the growth of worms in the MBR and CAS reactor, respectively. The separation of waste sludge and worm faeces was made possible with a new reactor concept in which L. variegatus was immobilized in a carrier material. This also eliminated the need to separate the worms from the sludge. The reactor concept consisted of a sludge compartment containing both waste sludge and worms. The open bottom of the sludge compartment was covered with a carrier material (polyamide mesh, 300 mm; with a surface area of 7.5 cm2) was used, through which the worms can protrude their tails. It was placed partially submerged in the SBR. L. variegatus respires and defecates via its tail, as a result; the worms kept their heads in the sludge compartment and protruded their tails into the aeration tank. Out of the total consumed waste sludge, 75% reduction in the amount of TSS was observed in addition to the natural sludge breakdown. Thus, carrier material acted as both a support material for the worms and a separation layer between the waste sludge and the worm faeces, which was beneficial to further processing. The effect of changes in DO concentration, ammonia concentration, temperature and light exposure on sludge reduction by L. variegatus in same reactor configuration used by Elissen et al. (2006) was studied to optimize the reactor. Sludge consumption rate was four times higher at DO above 8.1 mg/L, as against below 2.5 mg/L. Sludge reduction was 36 and 77% at these respective DO concentrations. Similar study was repeated by applying in continuous operation on increased mesh size to 350 mm over sludge. TSS reduction of 16e26% by the worms was achieved (22e30% VSS reduction). Mass balances showed that the worms contributed 41e71% towards total VSS reduction. The rest was caused by natural sludge breakdown. Very small Aeolosoma hemprichi worm (about 1e2 mm in size, doubling time about 1.4 days) grew at constant rate when the sludge concentration was higher than 300 mg VSS/L, higher growth rates of A. hemprichi were observed in higher protein containing and smaller particle size sludges. The sludge yield had negative correlation with the density of A. hemprichi regardless of differences in SRTs or F/M ratios. The relative sludge reduction by A. hemprichi was about 39e65% and in terms of per unit weight of A. hemprichi it was 0.53e6.32 mg VSS/mg A. hemprichi day. Stabilize the sludge settleability and total phosphorus removal but did not affect COD and NH4þ-N removals in the process. The CAS reactor consisted of three parts namely; aeration tank, transition region and settling tank, any effect of transition region on worms other than as a buffer for settling zone has not been spelt out in the paper. (continued on next page)
4300
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 2 8 7 e4 3 1 0
Table 3 (continued ) Reference(s)
Process
Song and Chen (2009)
Predation
Huang et al., (2007)
Predation
Wei et al. (2009a)
Predation
Highlights Different results were obtained while stabilizing the growth of A. hemprichi and minimizing the sludge production in batch and continuous experiments. A. hemprichi reached the maximum density when available VSS concentration was more than 3000 mg/L. No obvious difference was found between the initial specific growth rates (m) of A. hemprichi at various available VSS concentrations. Sludge reduction rate was correlated with both the growth rate and the density of A. hemprichi. The results indicated the sludge reduction rate was maximum at the density of 315 ind./mL. SRT more than 15 days did not affect the growth. Similar CAS reactor configuration which Liang et al. (2006a) used with A. hemprichi, was inoculated with oligochaete Tubifex tubifex. The results showed that the sludge reduction rate by T. tubifex (in terms of per unit dry mass of T. tubifex) was from 0.18 to 0.81 mg-VSS/mg-Tubifex.d. The sludge reduction capacity of the recycled sludge reactor was from 650 to 1080 mgSS/L.d. The optimum density of T. tubifex was 2500 mg/L and the optimum sludge recycled ratio was 1. The existence of T. tubifex did not affect COD and NH4þ-N removals in the process, but led to a slight decrease in TP removal, contrary to the findings of Liang et al. (2006a). SVI almost did not change when the T. tubifex density was lower than 3300 mg/L. A combination of a Tubificidae (known as Naididae) reactor with an integrated oxidation ditch with vertical circle (IODVC), was put to investigate excess sludge reduction (first stage) and returned sludge stabilization (second stage) as a new integrated system, dominantly containing Branchnria Sowerbyi worm after inoculation of Branchnria Sowerbyi and Limnodrilns sp. The maximal volume density of wet Tubificidae in vessels of the Tubificidae reactor was 17600 g/m3. The results showed that the excess sludge reduction rate was 46.4% in the first stage and the average sludge yield of the integrated system was 6.19 105 kg SS/kg COD in the second stage. Though the sludge returned to IODVC via the Tubificidae reactor, it had little impact on the effluent and the sludge quality.
in the second stage as compared to CAS. Different microfauna were observed under different set of conditions. Therefore, use of protozoa and metazoa for full-scale application for sludge reduction demands more understanding of microbial ecology of the system in question to ensure better operation and process control. Cech et al. (1994) also related a concomitantly poor phosphorous removal in a single stage laboratory scale reactor for a mixed population with a marked increase in predator numbers. The performance of CAS reactor and a submerged MBR as second stage was compared by Ghyoot (1998) in his thesis and later publication (Ghyoot and Verstraete, 2000). The sludge yield of the two-stage submerged MBR system was 20e30% lower than that of the two-stage CAS system under similar condition of SRT and F/M ratio owing increased grazing of predators in the two-stage MBR than those in the CAS reactor. However, the increased grazing on nitrifiers not only decreased the capacity of nitrification, but also resulted in high N and P concentration built-up in the effluent. MBR configuration also resulted in higher levels of soluble but poorly biodegradable COD. Subsequent research encounters with predators are shared in Table 3. It has been observed that there is a shift on research related to sludge reduction by predators towards oligochaeta rather than protozoa and metazoa mainly due to process uncertainties.
2.4.2.
Oligochaeta
Possessed with advantages such as low cost and no secondary pollution, worm (Oligochaete) technology, based on using microfauna’s predation to reduce excess sludge has recently begun to receive increasing attention by the researchers. It is a subclass in the biological phylum Annelida and includes various earthworms. Worms are the largest organisms found
in activated sludge and are already in application to regulate sludge. However, Wei et al. (2003b) mentioned that a practical application is still uncontrollable as there is no clear relationship between process conditions (e.g. retention times, temperature, sludge loading rates and shear forces) and worm growth. They state that one of the challenges is to maintain high densities of worms for a long time, particularly in fullscale applications. The main types of worms present in activated sludge system and trickling filters are Naididae, Aeolosomatidae and Tubificidae. Ghyoot and Verstraete (2000), Ratsak et al. (1996), Ratsak and Verkuijlen (2006), Rensink and Rulkens (1997), and Salvado et al. (1995) have performed research on oligochaeta to put forward advancements in handling and reduction of the excess sludge. Wei et al. (2003b) proved the effectiveness of using microfauna to reduce excess sludge, but the results were negatively correlated based only on sludge yield and density of the microfauna. As the sludge yield is easily affected by environmental and operational factors such as temperature, classes of bacteria, SRT and F/M ratio, therefore, any correlation excluding all these factors could be misleading on reduction scales obtained by comparing the correlation of sludge yield and the microfauna’s density alone. However, Luxmy et al. (2001) suggested that metazoa could not reduce sludge production in an aeration tank. Elissen et al. (2006) opined that microfauna could reduce sludge, but argued about which species of microfauna could reduce sludge best. These contradictory opinions were related to the lack of effective methods of detecting the rate of sludge reduction caused by microfauna. Therefore, it is important to accurately determine the rate of sludge reduction so as to compare and choose types of microfaunas as sludge predators. Table 3 gave a precise account of oligochaetes worms.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 2 8 7 e4 3 1 0
A dynamic energy-budget model was extended to describe the growth of Nais elinguis, a common oligochaete species frequently occurring in sewage treatment plants. N. elinguis reproduces asexually by dividing into an anterior and a posterior naidid. The daughter naidids initially have different growth rates due to the differences in energy reserves. The parameters of the model, such as maintenance rate coefficient and energy conductance, which represent physiological features of the organisms, facilitate insight into the processes underlying naidid growth. A numerical method was developed to predict asexual reproduction. The presented model was an important step towards a more comprehensive knowledge and modelling of organisms in activated sludge plant (Ratsak et al., 1993). On comparing the results obtained from sequencing batch reactor (SBR) with Lumbriculus variegatus worms immobilized on a carrier material and without worms, it was almost three times less sludge production than in the blank experiment. The worm faeces that are produced after sludge predation have an SVI that was approximately half of the initial waste sludge, capable to settle much faster than the initial waste sludge. Since, conditions beneficial to predator growth may not be optimal for bacterial growth and overall process, Lee and Welander (1996a) separated them into two-stages. The same principle was also applied with aquatic worms by Elissen et al. (2006), Table 3 describe the conclusions of studies reported by Hendrickx et al. (2009a), Liang et al. (2006a), Song and Xiaofei, (2009), Huang et al., (2007) and Wei et al. (2009a) The TSS reduction of 16e26% (Hendrickx et al., 2009b) was much lower than the 75% (Elissen et al., 2006) and 36% (Hendrickx et al., 2009a) with sludge from a municipal sewage treatment plant mentioned in Table 3. This effect may be attributed to the origin of the sludge and timings (same system, but during different months of the year) due to varying nutritional value of sludge for the worms. Recently, the results described above were summed up to give meaningful rector design parameters of a sequencing batch worm rectors using carrier materials with 300 and 350 mm mesh sizes (Hendrickx et al., 2010). The surface specific consumption rates were 45 and 58 g TSS/m2 d, respectively. The author mentioned 29% smaller reactor compared to using a 300 mm mesh, which is in fact 22% mainly due to 25% higher worm density of 1.1 kg ww/m2. To avoid substrate (sludge) limitation, a sludge load above 100 mg TSS/g ww d is recommended in the paper. Worm biomass growth rates were 0.026/ d for free worms and 0.009e0.01/d for immobilized worms on carrier material. Similarly, the decay rates under free and immobilized conditions were 0.018/d and 0.023/d, respectively. The SOUR of the worms required to supply and maintain desired oxygen level in the reactor is mentioned as 4.9 mg O2/g ww d, which amounts to 0.5e1.0 mg O2/mg VSS digested. Still the worm reactor size depends on the type of receiving sludges. Parts of nutrients are incorporated into the biomass during synthesis and then withdrawn with excess sludge. During sludge mineralization, increase of phosphorus, nitrogen, CO2 and even dissolved COD in effluent in the form of SMP is always an even possibility. The same is reported in the recent researches of sludge reduction induced by oligochaeta. Batch test and radioisotope 32P tracer test were carried out to further
4301
investigate nutrients release and phosphorus distribution among supernatant, sludge and worm during predation of oligochaeta on sludge (Wei et al., 2009b). The radioisotope 32P tracer was transferred step by step from synthetic wastewater to worms through a food chain (activated sludge cultivated by synthetic wastewater with exogenous Na2H32PO4, and predation of worms on activated sludge). Radioisotope activities for the radioisotope 32P tracer in supernatant, activated sludge and worms were then determined. Results showed that more nutrients release into supernatant occurred in the tests of worms with sterilized sludge than that of worms with activated sludge and release of nitrogen and phosphorus was few in the tests of worms with activated sludge. After 24 h 32P concentration of supernatant in the test of sludge with worms was 9% higher than that in the test of sludge without worms and 32P concentration of worm increased by 2.7%. Additionally, the release rate of phosphorus into supernatant caused by worm’s predation on activated sludge was 0.1211 mg TP/ worm (wet weight) h (Wei et al., 2009b). The optimum growth parameters of a number of oligochaeta have been cited by Inamori et al. (1983) and Kuniyasu et al. (1997), which are important for their comparative assessment. A comparative study was conducted to assess the potential of sludge reduction of Aeolosoma hemprichi, Daphnia magna, Tubifex tubifex and Physa acuta (Liang et al., 2006b). Since, Carbon accounts for more than 50% of all the ingredients, therefore, the rate of sludge reduction was correlated with the rate at which solid carbon form were changed into liquid and gas carbon form. The rates of sludge reduction using the four microfaunas were 0.8, 0.18, 0.54 and 0.1 mgsludge/mg-Microfauna/d, respectively, changing with the microfauna’s phylum or class and body size. The results were similar to those produced using the direct measuring method. The sludge reduction rate by T. Tubifex was not significantly affected by the difference in sludge, which came from different wastewater treatment plants in different cities. The microfauna in the class of Clitellata or subclass of Oligochaeta reduced the sludge more efficiently than those in the Crustacea or the Gastropoda classes. The sludge reduction rates of microfauna were related to their individual body size, even though they belong to the same phylum or class. For example, the sludge reduction rate of A. hemprichi was much higher than that of T. Tubifex because of their difference in individual weight. The smaller microfauna had higher sludge reduction rates due to their rapid metabolism. In addition, the amount of sludge reduced by microfauna was correlated to both the sludge reduction rate and the quantity of microfauna in the reactor. Though, the sludge reduction rate of A. hemprichi was higher than that of T. Tubifex, the quantity of T. Tubifex was more than that of A. hemprichi in the aeration tank (results of other experiments indicated 10e20 times higher quantity of T. Tubifex than that of A. hemprichi). Therefore, T. Tubifex reduced more sludge than A. hemprichi (Liang et al., 2006b). Biologically sludge reduction processes, highlight a weakness in the currently accepted approaches to modelling. Neither of these treatment processes can be accurately predicted with currently accepted simulators and models, such as International Water Association’s ASM Model (van Loosdrecht and Henze, 1999) and Anaerobic Digestion Model (ADM) (Batstone et al., 2002) series of models. To better
4302
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 2 8 7 e4 3 1 0
understand the mechanisms behind biological sludge reduction processes, a mathematical model similar to the ASM models (Henze et al., 2000) describing the interaction between nitrifiers, heterotrophs and predators in wastewater treatment has been developed by van Loosdrecht and coworkers (Moussa et al., 2005). The model includes six soluble compounds (DO, nitrogen gas, ammonia, nitrite, nitrate and COD) and five types of biomass (ammonia oxidizers, nitrite oxidizers, heterotrophs, predators and inert biomass) as particulate compounds. The kinetic expressions in the model are based on Monod equations. Two decay rates were considered in the model: aerobic decay occurred when the bacteria starved in the presence of oxygen, while anoxic decay occurred when the bacteria starved in the absence of oxygen and presence of NO3 (Leenen et al., 1997; Siegrist et al., 1999). Moreover, to assess the predator activity, a simple procedure was developed. The respiration under starvation condition with and without the presence of the predators was measured to determine the activity of predators. A shock load of NaCl was used to eliminate the predators. The impact of salt on the presence of the predators (protozoa, rotifers and nematodes) was followed in phase contrast mode using a light microscope. The minimum dose of NaCl and the contact time required for full elimination of the predators without affecting the activity of ammonia and nitrite oxidizers and of heterotrophs were 5 g NaCl as Cl/L and 1 h. The respiration activity of the samples treated with salt was estimated by measuring the OUR in a batch biological oxygen monitor. However, the use of salt as a selective inhibitor for predators under practical conditions demanded further investigations (Moussa et al., 2005). The results were then calibrated and validated on two laboratory scale SBRs operated at different SRT of 30 and 100 days for a period of 4 years. The fraction of active biomass (nitrifiers, denitrifiers and heterotrophs) predicted by the proposed model was only 33% and 14% at SRT of 30 and 100 days, respectively. The high fraction of inert biomass predicted by the model was in accordance with the microscopic investigations of biomass viability in both reactors. The results of this study presented the possibility for increasing the nitrification activity by suppressing the growth of predators in a nitrifying system or other systems in which slow-growing bacteria play an important role. The model results showed the need for careful optimization of systems operated at long SRT (such as membrane bioreactors), to avoid accumulation of high amounts of inert biomass and to avoid high operational costs without gaining any volumetric improvement. The model showed its capacity to elucidate the biological processes in activated sludge systems by including the effect of the predators. The practical application of the developed model and assessments of predators’ activity called for verification under full scale activated sludge plant operation (Moussa et al., 2005). The results of most of the studies on sludge reduction through worms are encouraging (Table 3), but still the control parameters of the system have not been established. Therefore, further research to correlate all the known parameters of wastewater treatment with the worms’ growth, density etc. in demonstration followed by full-scale needs to be undertaken.
2.4.3.
Filamentous fungi (FF)
Application of filamentous fungi is another possible area of eco-manipulation to preclude sludge growth. Fungi are saprophytic organisms and they obtain their nourishment from the degradation of dead organic matter. A great variety of fungi has been reported to be found in sludge, FF have been recognized for sludge treatment and possibly these strains can be utilized for simultaneous bioflocculation, solids and pathogens reduction and, removal and degradation of toxic compounds, a detailed account is given by More et al. (2010). The fungi treatment of sludge is less energy consuming. The oxygen supply needs of the microfungi are approximately one third of the oxygen requirement by a bacterial population. In addition, the microfungi will use all the forms of oxygen supply available in order to optimize the degradation of the organic matter. The fungi have advantages over bacteria because of fungi having capability to degrade more complex and variety of substrates. Certain fungi strains can also be chosen for their beneficial effects on the plants if treated sludge intended in the use as a fertilizer. Different fungi used for sludge treatment are; Phanerocheate chrysosporium (Alam et al., 2001), Mixed culture of Aspergillus niger and Penicillium corylophilum (Alam et al., 2003a, 2003b; Alam and Fakhru’l-Razi, 2003), Mixed culture of P. corylophilum WWZA1003 and Aspergillus niger SCahmA103 (Fakhru’l-Razi et al., 2002b), Mucor hiemalis (Fakhru’l-Razi and Molla, 2007), Aspergillus (Jamal et al., 2005), Aspergillus niger; Penicillium corylophilum (Mannan et al., 2005), Mixed cultures of Trichoderma harzianum with Phanerochaete chrysosporium 2094 and T. harzianum with Mucor Hiemalis (Molla et al., 2004), Penicillium expansum BS30 (Subramanian et al., 2006), P. expansum BS30 (Subramanian et al., 2008) However, the role of FF in sludge treatment is not well established and limited to shake flasks, which lack semblance to actual operating conditions. The all important selection of useful fungi and the effect of parameters such as dissolved oxygen, pH, temperature, agitation, suspended solid concentration, operating conditions and incubation time on fungal activity to achieve maximum solids degradation are yet to be established and therefore need further exploration. The specific studies on selection of fungal strains and influence of different concentration require attention. Therefore, extensive laboratory followed by pilot scale studies are needed for future applications of filamentous fungi as an eco-tool to sludge reduction.
3.
Discussion
Ecological imbalances are the fallout of economic and industrial development, the problem is particularly pronounced where the populations are large and increasing exponentially, for example the recent spur of development in Asia has its own environmental and social ramifications. The treatment and disposal of excess sludge is one amongst them. It is expected that by the year 2015, there will be more than 1000 operational municipal wastewater treatment plants (WWTPs) in China. In 2006, 39% of the Chinese population was served by municipal WWTPs, registering an increase by 18.2%, since the year 2000; while the wastewater treatment capacity has
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 2 8 7 e4 3 1 0
increased by 135%. In the year 2005, 297 cities in China had no WWTPs for treating wastewater. These facts indicate the potential growth of sewage treatment capability and sludge production along with economic and industrial growth (Idris et al., 2004; He et al., 2006; Jun et al., 2004b). Bhardwaj (2005) mentioned that about 26254 million litres per day (MLD) of wastewater were generated in the 921 towns in India (housing more than 70% of urban population) in 2004. It was only 18312 MLD in 644 cities in 1995. This reflects 30% growth of both urbanization and wastewater generation owing to industrial growth of the country. The municipal wastewater treatment capacity developed so far in India is about 7044 MLD e accounting for 27% of wastewater generation in urban areas (Bhardwaj, 2005; CPCB, 2004). These facts indicate the potential growth of sewage treatment capability and sludge production along with economic and industrial growth of these two economic power houses of Asia. It is further expected that the problem of sludge production will continue to increase with time due to increased social awareness and legislative concerns. As conventional or traditional methods of sludge reduction and disposal are under stringent control and almost phased out the immediate task is to find a novel cost effective and sustainable mechanism. Broadly speaking there are two approaches; online and offline, both with relative advantages and disadvantages. Varieties of possible techniques and their configurations have been and under investigation at different level, however, the research is in its infancy and therefore it is too early to come to a consensus. At best the relative merits and demerits can be precisely tabulated (Table 4). Cell fractionation is basically the rate limiting step in lysiscryptic growth and ozonation is the only successfully applied full scale process, however, enough scope still presents for the application of enzymatic hydrolysis, wet air oxidation and alkaline-ultrasound treatment. High rate oxygenation has limited application and efficiency, but it does not require any additional unit, however, process economy and energybalance needs further exploration. Non-chemically assisted uncoupling by virtue of high So/Xo and OSA process are promising efficient options. Their fullscale validation is still awaited. MBR is another technique which may have scope in future; however, further studies on membrane life, fouling and overall cost-benefit determination would further justify their large scale application. Sludge reduction by predation and oligochaeta is certainly environmental friendlier than others, but completely lack operational control from engineering point of view due to extreme uncertainties associated with them and keeping environmentalists away to apply on full scale. In summary, OSA and MBR are two such processes which could be short listed owing to their merits on account of friendliness to environment, defined mechanism of energy consumption in increased cell maintenance and flexibility in operation due to easy control. However, still the most important limitation in application of in-place sludge reduction is undefined support of process mathematics and subsequent modelling which is keeping us guessing and inviting for further research. The emphasis on in-place sludge reduction creates a unique or rather conflicting situation in simultaneous removal of organics, nitrogen, phosphorus and biomass as;
4303
Physiology of nitrifiers depends on high DO, while for denitrifiers it is low DO. High SRT required for organics and ammonia oxidation and sludge reduction, while low SRT for phosphorus uptake. Requirement of Carbon source for nitrification, while absence of any donor other than PHA/PHB in enhanced phosphate uptake. pH and alkalinity reduces during nitrification and denitrification, while high pH and alkalinity required for chemical precipitation of phosphate. The synthesized biomass normally contains 2e3% P by its dry weight, stoichiometric inclusion of it into biomass yields it as C5H7O2NP0.1 giving 2.67% P (i.e. 3.1 100/116g ¼ 2.67), in case of enhanced biological phosphate removal it could be as high as 2e5 times the above (5.4e13.4% of the cell mass). Biomass uptake is the only mechanism of phosphate removal, although chemical precipitation during denitrification in the form of Hydroxyapatite {Ca5(OH)(PO4)3} and Struvite (Mg NH4PO4) has also been reported (Quintana et al., 2005). Whereas, nitrogen has other route of removal through denitrification in addition to biomass uptake which is around 12% of the cell mass. The pinnacle of all efforts of nutrient removal is practically put-down the moment biological cell are lysed, hydrolyzed or digested causing reappearance of nitrogen in general and phosphorus in particular as mentioned above. Therefore, the phenomenon of in-place sludge reduction along with simultaneous removal of organics, nitrogen, and phosphorus depends on extent of reconciliation of conflicts as cited above. The phenomenon of sludges in Europe and USA is almost under strict surveillance through regulations, but the scenario is grim and uncertain in Asia, in view of social, political and legal infirmities and legislative inadequacies. There are three laws in Chinese environmental legislations that contain restrictions on sludge management, namely; Control standard for pollutants in sludge for agricultural use, GB4284-84 (promulgated in 1984 but never amended), Criteria for controlling the discharge of sewage and sludge from municipal wastewater treatment plant, CJ3025-93, and Criteria for controlling the discharge of pollutants from municipal wastewater treatment plant, GB18918-2002. The last one concludes that all sludge should be stabilized and shall meet the control degradation indexes pathogen concentrations after stabilization (Tottie, 2007). Solid Waste Pollution Prevention and Control Law enacted in 1995 establishes a broad national framework for the management of industrial, municipal and hazardous waste, aiming to safeguard human health by means of preventing and controlling solid waste pollution (Beyer, 2006). But, the reality is that the construction of WWTPs often is built only to achieve emission standards and not sludge quality standards. Similarly, the Municipal Solid Wastes (Management and Handling) Rules (2000), Government of India, do not provide for specific eco friendly sludge treatment. Realizing the dynamic economic development and the evolving political system, India’s environmental future is difficult to determine. The situation is so grim that due the absence of any suitable legislation and implementation of existing provisions even basic data on sludge generation and its expected growth is
4304
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 2 8 7 e4 3 1 0
Table 4 e Relative merits and demerits of Sludge reduction techniques. Process Pure oxygen
MBR
Merits High DO-activated sludge process can repress development of filamentous organisms. Ability to maintain a higher MLVSS concentration in the aeration tank. Better sludge settling and thickening. Higher oxygen transfer efficiency. More stable operation. Upto zero sludge growth Very high MLSS can be maintained Small footprint.
Flexibility of operation. Applied in full-scale
Demerits The efficacy of the process is not clear.
54 and 60%
Major References McWhirter (1978), Boon and Burges (1974)
The mechanism is not fully known.
High aeration cost.
Sludge settling and dewatering is more Upto 100% difficult Poor oxygenation: increased aeration cost. Membrane fouling, responsible for high cost Not feasible to operate with complete sludge retention, minimum wasting is desired. Energy requirements Cost of membranes. New materials still under investigation. Further research is needed to understand 38e54% the process and establish the optimum operational conditions and improve the process operation.
It is relatively easy to introduce the anaerobic zone to the conventional activated sludge process. Control of the growth of filamentous organisms. No physical or chemical forces are needed. The OSA process improves the COD removal and the settleability of activated sludge. Capable of handling high strength organic pollutants without serious sludge associated problems. PREDATION Predation on bacteria are used The worms growth is still uncontrollable, especially in the full scale application in large experimental operations today Kinetics is still undefined. Environmental friendly. High DO High degree of stabilization Inability to maintain a high MLVSS in the aeration tank Stable and Reliable operation The mechanism is not fully known. Easy to implement Inability to maintain a higher MLVSS concentration Can repress development of High energy consumption. filamentous organisms. No application of uncoupling Yet to find a place in its engineering So/Xo chemicals application Lower reactor volume Poor effluent quality. Wastewater post-treatment required. High capital and operation costs. Only applicable to high strength industrial wastewaters. OSA
Reduction in Growth
missing preventing even formulation of future blue print in this regard. But the scenario is different in Japan as it has been highly reliant on incineration for waste disposal (i.e. a total of 1320 plants in 2005), which is extremely high (i.e. 74%) compared to
12e75%
0e66%
Low and Chase (1999b), Churchouse and Wildgoose (1999), Rosenberger et al. (1999; 2002), Judd, 2007.
Chudoba et al. (1992a), Chen et al. (2001a,b, 2003)
Lee and Welander (1994, 1996a,b), Ghyoot and Verstraete (2000), Wei et al. (2003a, 2009a). Moussa et al. (2005). Abbassi et al. (2000) Ichinari et al. (2008)
Around 70% Chudoba et al. (1991), Ghigliazza et al. (1996), Liu (2000)
that in other Asian countries (Horio et al., 2009). 1997 saw the amendment to the 1991 Promotion of Resource Recycling and Reuse Law and a new Waste Disposal and Public Cleansing Law enacted in Japan as basic statutory regulations for waste disposal and recycling. This Law serves to promote recycling
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 2 8 7 e4 3 1 0
at the production, distribution, and consumption stages, effective use of resources, decreased waste generation, and environmental conservation (Tanaka, 1999). However, the element of eco friendly disposal of sludges is still missing as their treatment is largely confined to landfills and incineration. And still the amount of night soil, which can be collected from areas where sewage system has not been developed yet (i.e. 5% of all areas) was around 1.11million tons/year (PERI, 2007). However, the policy of enactment of law for the installation of onsite wastewater treatment in Japan called Johkasou paid rich dividends as it reduces sludge by 35% resulting in an observed growth yield of only 0.18. It is a combination of “Jouka” means purification and “Sou” means tub or tank (Ichinari et al., 2008). The success of Johkasou owe to the provision of a national subsidy and bond issuance to home owners, which increased from 100 million yen in 1987 to 21 billion yen in 2000 (JECES, 2004). Therefore, with determination some hard decisions have to be taken by the respective national governments and their regulatory authorities on stringent legislations, policies and maintenance of standards in the near future, lest the economic growth built assiduously will shortly return to same stupor as before.
4.
Summary
There is a noticeable shifting of priorities from higher volumetric rate and effluent quality to less biomass production during wastewater treatment on the basis of overall mass balance of the inputs and outputs. Particularly during the last decades there has been a major change in the ways sludges are disposed. Prior to 1998, land application as a fertilizer for agricultural reuse (37%), incineration (11%), simply land filling (40%) and use in some other areas such as forestry, land reclamation, seawaters disposal (12%) etc. were the common alternatives of municipal sludge disposal. Since 1998, onwards, European legislation prohibits the sea disposal of sewage sludge, in order to protect the marine environment and sludge deposits in landfills were also proposed to be phased out. This led to generation of significant scientific interest in the latest trends in the field of sludge management, i.e. agricultural reuse, biological oxidation, digestion, combustion, wet oxidation, pyrolysis, gasification and co-combustion of sewage sludge with other materials for further use as energy source. In addition to inherent minerals sewage sludge also contains nitrogen and phosphorous, resulting especially from nitrification and phosphorous accumulation phases in wastewater treatment process, which possesses fertilizer value. At the same time it may contain various other elements, which can be harmful when entering in human food chain, such as heavy metals, protonophores, disinfectants, pathogens and organic pollutants. An essential prerequisite of agriculture use is its storage; as sludge is being produced all year round whereas its application on land takes place once or twice a year; in addition, social unacceptability remains the grey area in the picture. Therefore, agricultural use is increasingly regarded as an insecure handling route. The other conventional route, such as land filling is eliminated due to stringent legislation and increased costs. Incineration on the
4305
other hand provides a large volume reduction of sewage sludge and results in improved thermal efficiency. However, the scrubbing costs of the product gases for air pollution control are usually very high in addition to high carbon footprint. Therefore, there are several driving forces for the search of alternatives for safe handling of the sewage sludges and reduction or minimization at the place of its production (inpipe) through biological lysis-cryptic growth, biological uncoupling of metabolism, maintenance metabolism and ecomanipulation. These sludge reduction technologies have validated the potential to significantly change the methods by which wastewater treatment biosolids are treated and handled across the globe, but still the real impetus is to explore more and more economically viable and environmentally friendlier alternatives. Some of the recent advent of sludge reduction technologies are just being introduced and thus are not yet well tested as combustion or incineration. To be precise, no consensus on sludge reduction approach has been evolved so far and at best the choice of technique depends on their relative merits and demerits. However, all the above summed issues indicate the importance of investigating thoroughly these novel trends in sewage sludge handling. The only conclusion drawn is that the total oxidation of synthesized biomass during wastewater treatment is indeed valid, albeit difficult to affectuated by current engineering practices evolved so far. It is also expected that this review would serve its purpose as dissemination of knowledge on this subject would take place.
Acknowledgements We are thankful to Editor Water Research for review, support and communication. We are grateful to the two anonymous reviewers, this article would not have taken shape without critical comments and time spared by them, their efforts deserve due acknowledgement. Thanks are due to Dr. Vinay K. Tyagi, M/S Abid A. Khan, Rubia, Muntajeer and Akanksha, whose support was our ultimate strength.
references
Abbassi, B., Dullstein, S., Rabiger, N., 2000. Minimization of excess sludge production by increase of oxygen concentration in activated sludge flocs: experimental and theoretical approach. Water Research 34 (1), 139e146. Ahn, K.H., Park, K.Y., Maeng, S.K., Hwang, J.H., Lee, J.W., Song, K. G., Choi, S., 2002. Ozonation of wastewater sludge for reduction and recycling. Water Science and Technology 46 (10), 71e77. Alam, M.Z., Fakhru’l-Razi, A., 2003. Enhanced settleability and dewaterability of fungal treated domestic wastewater sludge by liquid state bioconversion process. Water Research 37, 1118e1124. Alam, M.Z., Fakhru’l-Razi, A., Molla, A.H., 2003a. Biosolids accumulation and biodegradation of domestic wastewater treatment plant sludge by developed liquid state bioconversion process using batch fermentation. Water Research 37 (15), 3569e3578.
4306
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 2 8 7 e4 3 1 0
Alam, M.Z., Fakhru’l-Razi, A., Molla, A.H., 2003b. Optimization of liquid state. Bioconversion process for microbial treatment of domestic wastewater sludge. Journal of Environmental Engineering and Science 2, 299e306. Alam, M.Z., Fakhru’l-Razi, A., Molla, A.H., Roychoudhury, P.K., 2001. Treatment of wastewater sludge by liquid state bioconversion process. Journal of Environmental Science Health A 36 (7), 1237e1243. Are´valo, J., Moreno, B., Pe´rez, J., Go´mez, M.A., 2009. Applicability of the sludge biotic index (SBI) for MBR activated sludge control. Journal of Hazardous Materials 167], 784e789. Armenante, P.M., Kafkewitz, D., Lewandowski, G.A., Jou, C.J., 1999. Anaerobiceaerobic treatment of halogenated phenolic compounds. Water Research 33, 681e692. Baier, U., Schmidheiny, P., 1997. Enhanced anaerobic degradation of mechanically disintegrated sludge. Water Science and Technology 36 (11), 137e143. Barjenbruch, M., Hoffmann, H., Tra¨nckner, J., 2000. Minimizing of foaming in digesters by pre-treatment of the surplus sludge. Water Science and Technology 42 (9), 235e242. Batstone, D.J., Keller, J., Angelidaki, I., Kalyuzhnyi, S.V., Pavlostathis, S.G., Rozzi, A., Sanders, W.T.M., Siegrist, H., Vavilin, V.A., 2002. Anaerobic Digestion Model No. 1 (ADM1). IWA Publishing, London, UK, ISBN 1900222787. Benefield, L.D., Randall, C.W., 1980. Biological Process Design for Wastewater Treatment. Prentice-Hall, Inc., Englewood Cliffs, NJ. Beyer, S., 2006. Environmental law and policy in the People’s Republic of China. Chinese Journal of International Law 5 (1), 185e211. Bhardwaj, R.M. (2005) Status of Wastewater Generation and Treatment in India. IWG-Env Joint Work Session on Water Statistics, 20e22 June 2005, Vienna. Bishop, B. (2004) Use of ceramic membranes in airlift membrane bioreactors. 8th International Conference on Inorganic Membranes (ICIM8); 2004 July 18e22, 2004, Cincinnati, USA. Bitton, G., 2005. Wastewater Microbiology, third ed. Wiley-Liss, Hoboken, N.J. Boon, A.G., Burgess, D.R., 1974. Treatment of crude sewage in two high-rate activated sludge plants operated in series. Journal of the Water Pollution Control 74, 382. Bouillot, P., Canales, A., Pareilleux, A., Huyard, A., Goma, G., 1990. Membrane bioreactors for the evaluation of maintenance phenomena in wastewater treatment. Journal of Fermentation and Bioengineering 69 (3), 178e183. Camacho, P., Geaugey, V., Ginestet, P., Paul, E., 2002. Feasibility study of mechanically disintegrated sludge and recycle in the activated-sludge process. Water Science and Technology 46 (10), 97e104. Canales, A., Pareilleux, A., Rols, J., Goma, G., Huyard, A., 1994. Decreased sludge production strategy for domestic wastewater treatment. Water Science and Technology 30 (8), 97e106. Cech, J.S., Hartman, P., Macek, M., 1994. Bacteria and protozoa population dynamics in biological phosphate removal systems. Water Science and Technology 19 (7), 109e117. Chaize, S., Huyard, A., 1991. Membrane bioreactor on domestic wastewater treatment sludge production and modelling approach. Water Science and Technology 23, 1591e1600. Chen, G.H., Leung, D.H.W., 1999. Utilization of oxygen in a sanitary gravity sewer. Water Research 34 (15), 3813e3821. Chen, G.H., Liu, Y., 1999. Modeling of energy spilling in substratesufficient cultures. Journal of Environmental Engineering e ASCE 125 (6), 508e513. Chen, G.H., Mo, H.K., Liu, Y., 2002. Utilization of a metabolic uncoupler, 3,30 ,40 ,5 tetrachlorosalicylanilide (TCS) to reduce sludge growth in activated sludge culture. Water Research 36 (8), 2077e2083. Chen, G.H., Mo, H.K., Saby, S., Yip, W.K., Liu, Y., 2000. Minimization of activated sludge production by chemically stimulated energy spilling. Water Science and Technology 42 (12), 189e200.
Chen, G.H., An, K.J., Saby, S., Brois, E., Djafer, M., 2003. Possible cause of excess sludge reduction in an oxic-settling anaerobic activated sludge process (OSA process). Water Research 37 (16), 3855e3866. Chen, G.H., Saby, S., Djafer, M., Mo, H.K., 2001a. New approaches to minimize excess sludge in activated sludge systems. Water Science and Technology 44 (10), 203e208. Chen, G.H., Yip, W.K., Mo, H.K., Liu, Y., 2001b. Effect of sludge fasting/feasting on growth of activated sludge cultures. Water Research 35 (4), 1029e1037. Chiu, Y.C., Chang, C.N., Lin, J.G., Huang, S.J., 1997. Alkaline and ultrasonic pretreatment of sludge before anaerobic digestion. Water Science and Technology 36 (11), 155e162. Chu, C.P., Chang, B.V., Liao, G.S., Jean, D.S., Lee, D.J., 2001. Observation on changes in ultrasonically treated wasteactivated sludge. Water Research 35 (4), 1038e1046. Chu, C.P., Feng, W.C., Chang, B.V., Chou, C.H., Lee, D.J., 1999. Reduction of microbial density level in wastewater activated sludge via freezing and thawing. Water Research 33 (16), 3532e3535. Chudoba, B., Morel, A., Capdeville, B., 1992b. The case of both energetic uncoupling and metabolic selection of microorganisms in the OSA activated sludge system. Environmental Technology 13, 761e770. Chudoba, P., Capdeville, B., Chudoba, J., 1992c. Explanation of biological meaning of the So/Xo ratio in batch cultivation. Water Science and Technology 26 (3e4), 743e751. Chudoba, P., Chevalier, J.J., Chang, J., Capdeville, B., 1991. Effect of anaerobic stabilization of activated sludge on its production under batch conditions at various So/Xo ratios. Water Science and Technology 23, 917e926. Chudoba, P., Chudoba, J., Capdeville, B., 1992a. The aspect of energetic uncoupling of microbial growth in the activated sludge process: OSA system. Water Science and Technology 26 (9e11), 2477e2480. Churchouse, S., Wildgoose, D., 1999. Membrane bioreactor from lab to full-scale application. In: The 2nd Symposium on Membrane Bioreactors for Wastewater Treatment. Cranfield University, UK. Cicek, N., Macomber, J., Davel, J., Suidan, M.T., Audic, J., Genestet, P., 2001. Effect of solids retention time on the performance and biological characteristics of a membrane bioreactor. Water Science and Technology 43 (11), 43e50. Commission of European Communities, 1998. Council Directive 91/271/EEC of 21 March 1991 Concerning Urban Wastewater Treatment amended by the 98/15/EC of 27 February 1998. Copp, J.B., Dold, P.L., 1998. Comparing sludge production under aerobic and anoxic conditions. Water Science and Technology 38, 285e294. CPCB, 2004. News Letter, Urbanisation & Wastewater Management in India. Ministry of Environment and Forests, Government of India. Curds, C.R., Hawkes, H.A., 1975. Ecological Aspects of Used-Water Treatment: The Organisms and their Ecology. Academic Press, New York. Curds, C.R., Cockburn, A., 1970. Protozoa in biological sewage treatment process-I: a survey of the protzoan fauna of British percolating filters and activated sludge plants. Water Research 4, 225e236. Deleris, S., Geaugey, V., Camacho, P., Debellefontaine, H., Paul, E., 2002. Minimization of sludge production in biological processes: an alternative solution for the problem of sludge disposal. Water Science and Technology 46 (10), 63e70. Drews, A., 2007. Does fouling in MBR depend on SMP?. In: Proceedings of the 4th IWA Membranes Conference, May 15e17, 2007. Cranfield University Press, Harrogate, UK. Drews, A., 2010. Membrane fouling in membrane bioreactors e characterisation, contradictions, cause and cures. Journal of Membrane Science 363, 1e28.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 2 8 7 e4 3 1 0
Egemen, E., Corpening, J., Nirmalakhandan, N., 2001. Evaluation of an ozonation system for reduced waste sludge generation. Water Science and Technology 44 (2e3), 445e452. Egemen, E., Corpening, J., Padilla, J., Brennan, R., Nirmalakhandan, N., 1999. Evaluation of ozonation and cryptic growth for biosolids management in wastewater treatment. Water Science and Technology 39 (10e11), 155e158. Eikelboom, D.H., 1988. Extra to epassingsmogelijkheden voor protozoa en metazoa bij de zuivering van afval water TNOrapport no. R88/286, The Netherlands. Eikelboom, D.H., 2000. Process Control of Activated Sludge Plants by Microscopic Investigation. IWA Publishing, UK. Eikelboom, D.H., Borger, A.R., van Houten, R.T., 2001. Reductiezuiveringsslib Door Borstelwormen. Fase 1. TNOrapport. Laboratoriumonderzoek, The Netherlands. Elissen, H.J.H., Hendrickx, T.L.G., Temmink, H., Buisman, C.J.N., 2006. A new reactor concept for sludge reduction using aquatic worms. Water Research 40, 3713e3718. Elton, C.S., 1935. Animal Ecology. Sidgwick and Jackson, London. engineeringvillage2.com, 2010. Web-based Information Service. http://www.engineeringvillage2.com. EPA, 1987. Design Manual-Dewatering Municipal Wastewater Sludge Washington, DC. Fakhru’l-Razi, A., Alam, M.Z., Idris, A., Abd-Aziz, S., Molla, A.H., 2002b. Domestic wastewater accumulation by liquid state bioconversion process for rapid composting. Journal of Environmental Science and Health 38 (8), 1533e1543. Fakhru’l-Razi., A., Molla, A.H., 2007. Enhancement of bioseparation and dewaterability of domestic wastewater sludge by fungal treated dewatered sludge. Journal of Hazardous Materials 147, 350e356. Fenu, A., Roels, J., Wambecq, T., De Gussem, K., Thoeye, C., De Gueldre, G., Van De Steene, B., 2010. Energy audit of a full scale MBR system. Desalination 262, 121e128. Garces, A., De Wilde, W., Thoeye, C., De Gueldre, G., 2007. Operational cost optimisation of MBR Schilde. In: Proceedings of the 4th IWA Membranes Conference, 15e17 May 2007. Harrogate Cranfield University Press, UK. Gaudy, A.F., Gaudy, E.T., 1981. Microbiology for Environmental Scientists and Engineers, First ed. McGraw Hill. Ghigliazza, R., Lodi, A., Converti, A., Nicolella, C., Rovatti, M., 1996. Influence of the ratio of the initial substrate concentration to biomass concentration on the performance of a sequencing batch reactor. Bioprocess and Biosystems Engineering 14 (3), 131e137. Ghyoot, W. (1998) Membrane-assisted bioreactor for anaerobic sludge digestion and low sludge production in aerobic wastewater treatment. PhD thesis, University of Gent, Belgium. Ghyoot, W., Verstraete, W., 2000. Reduced sludge production in a two-stage membrane-assisted bioreactor. Water Research 34 (1), 205e215. Gil, J.A., Tu´a, L., Rueda, A., Montan˜o, B., Rodrı´guez, M., Prats, D., 2010. Monitoring and analysis of the energy cost of an MBR. Desalination 250, 997e1001. Giokas, D.L., Daigger, G.T., von Sperling, M., Kim, Y., Paraskevas, P.A., 2003. Comparison and evaluation of empirical zone settling velocity parameters based on sludge volume index using a unified settling characteristics database. Water Research 37 (16), 3821e3836. Guellil, A., Boualam, M., Quiquampoix, H., Ginestet, P., Audic, J.M., Block, J.C., 2001. Hydrolysis of wastewater colloidal organic matter by extra-cellular enzymes extracted from activated sludge flocs. Water Science and Technology 43 (6), 33e40. Huang, X., Peng, L., Yi, Q., 2007. Excess sludge reduction induced by Tubifex tubifex in a recycled sludge reactor. Journal of Biotechnology 127, 443e451.
4307
Han, S.S., Bae, T.H., Jang, G.G., Tak, T.M., 2005. Influence of sludge retention time on membrane fouling and bioactivities in membrane bioreactor system. Process Biochem 40 (7), 2329e2400. Hazardous and Solid Waste Amendment Act, (1984) U.S. Environmental Protection Agency (EPA): P.L. 98e616, 98 Stat. 3221, Nov. 9, 1984. He, P.-J., Lu, F., Zhang, H., Shao, L.-M., Lee, D.-J., 2006. Sewage Sludge in China: Challenges toward a Sustainable Future. State Key Laboratory of Pollution Control and Resources Reuse. Tongji University, Shanghai and Department of Chemical Engineering, National Taiwan University. Hendrickx, T.L.G., Temmink, H., Elissena, H.J.H., Buisman, C.J.N., 2009a. The effect of operating conditions on aquatic worms eating waste sludge. Water Research 43, 943e950. Hendrickx, T.L.G., Temmink, H., Elissena, H.J.H., Buisman, C.J.N., 2009b. Aquatic worms eating waste sludge in a continuous system. Bioresource Technology 100, 4642e4648. Hendrickx, T.L.G., Temmink, H., Elissena, H.J.H., Buisman, C.J.N., 2010. Design parameters for sludge reduction in an aquatic worm reactor. Water Research 44 (3), 1017e1023. Henze, M., Gujer, W., Mino, M., van Loosdrecht, M.C.M., 2000. Activated Sludge Models ASM1, ASM2, ASM2d and ASM3. IWA Publishing, London, UK. Heran, M., Wisniewski, C., Orantes, J., Grasmick, A., 2008. Measurement of kinetic parameters in a submerged aerobic membrane bioreactor fed on acetate and operated without biomass discharge. Biochemical Engineering Journal 38, 70e77. Horan, N.J., 1990. Biological Wastewater Treatment Systems, Theory and Operation. John Wiley and Sons, Chichester, UK. Horio, M., Shigeto, S., Shiga, M., 2009. Evaluation of energy recovery and CO2 reduction potential in Japan through integrated waste and utility management. Waste Management 29 (7), 2195e2202. Hu, A.Y., Stuckey, D.C., 2006. Treatment of dilute wastewaters using a novel submerged anaerobic membrane bioreactor. Journal of Environmental Engineering 132, 190e198. Huang, X., Gui, P., Qing, Y., 2001. Effect of sludge retention time on microbial behaviour in a submerged membrane bioreactor. Process Biochemistry 36, 1001e1006. Humenick, M.J., Ball, J.E., 1974. Kinetics of activated sludge oxygenation. Journal of the Water Pollution Control Federation 46, 553e561. Ichinari, T., Ohtsubo, A., Ozawa, T., Hasegawa, K., Teduka, K., Oguchi, T., Kiso, Y., 2008. Wastewater treatment performance and sludge reduction properties of a household wastewater treatment system combined with an aerobic sludge digestion unit. Process Biochemistry 43, 722e728. Idris, A., Inanc, B., Hassan, M.N., 2004. Overview of waste disposal and landfills/dumps in Asian countries. Material Cycles and Waste Management 6, 104e110. Inamori, Y., Suzuki, R., Sudo, R., 1983. Mass culture of small aquatic oligochaeta. Research Report from the National Institute for Environmental Studies 47, 125e137 (in Japanese). Jamal, P., Alam, M.Z., Salleh, M.R.M., Akib, M.M., 2005. Sewage treatment plant sludge: a source of potential microorganism for citric acid production. American Journal of Applied Sciences 2 (8), 1236e1239. Janssen, P.M.J., Rulkens, W.H., Rensink, J.H., van der Poest, H.F., 1998. The potential for metazoa in biological wastewater treatment. Water Quality International, 25e27. JECES, 2004. Japanese Education Center for Environmental Sanitation. Jefferson, B. (2007) Low temperature municipal sewage treatment with anaerobic MBRs. The 6th International Membrane Science and Technology Conference: November 5e9, 2007; Sydney. Jeison, D., van Lier, J.B., 2007. Feasibility of thermophilic anaerobic submerged membrane bioreactors (AnSMBR) for
4308
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 2 8 7 e4 3 1 0
wastewater treatment. In: Proceedings of the 4th IWA Membranes Conference, May 15e17, 2007. Cranfield University Press, Harrogate, UK. Judd, S., 2007. The status of membrane bioreactor technology. Trends Biotechnology 26 (2), 109e116. Jun, H.B., Park, S.M., Park, N.B., Lee, S.H., 2004a. Nitrogen removal and sludge reduction in a symbiotic activated sludge system between anaerobic archaea and bacteria. Water Science and Technology 50 (6), 189e197. Jun, H.-J., Liu, D., Liang, P., 2004b. Sludge Treatment and Disposal. Qinghua Water Industry Technology Green Paper, First Series. Department of Qinghua University, China’s water network environment. Kepp, U., Machenbach, I., Weisz, N., Solheim, O.E., 1999. Enhanced stabilization of sewage sludge through thermal hydrolysis e three years of experience with a full scale plant. Water Science and Technology 42 (9), 89e96. Kopp, J., Muller, J., Dichtl, N., Schwedes, J., 1997. Anaerobic digestion and dewatering characteristics of mechanically disintegrated excess sludge. Water Science and Technology 36 (11), 129e136. Kuniyasu, K., Hayashi, N., Inamori, Y., Sudo, R., 1997. Effect of environmental factors on growth characteristics of oligochaeta. Japanese Journal of Water Treatment Biology 33, 207e214 (in Japanese). Laera, G., Pollice, A., Saturno, D., Giordano, C., Lopez, A., 2005. Zero net growth in a membrane bioreactor with complete sludge retention. Water Research 39, 5241e5249. Lapinski, J., Tunnacliffe, A., 2003. Reduction of suspended biomass in municipal wastewater using bdelloid rotifers. Water Research 37 (9), 2027e2034. Laspidou, C.S., Rittmann, B.E., 2002. A unified theory for extracellular polymeric substances, soluble microbial products, and active and inert biomass. Water Research 36, 2711e2720. Lawrence, A.W., McCarty, P.L., 1970. Unified basis for biological treatment design and operation. ASCE Journal of the Sanitary Engineering Division 96 (SA3), 757. Learner, M.A., 1979. The distribution and ecology of the Naididae (Oligochaeta) which inhabit the filter-beds of sewage-works in Britain. Water Research 13, 1291e1299. Lee, N.M., Welander, T., 1994. Influence of predator in nitrification in aerobic biofilm. Water Science and Technology 29 (7), 355e363. Lee, N.M., Welander, T., 1996a. Reducing sludge production in aerobic wastewater treatment through manipulation of the ecosystem. Water Research 30 (8), 1781e1790. Lee, N.M., Welander, T., 1996b. Use of protozoa and metazoa for decreasing sludge production in aerobic wastewater treatment. Biotechnology Letters 18 (4), 429e434. Leenen, E.J.T.M., Boogert, A.A., van Lammeren, A.A.M., Tramper, J., Wijffels, R.H., 1997. Dynamics of artificially immobilized Nitrosomonas europaea: effect of biomass decay. Biotechnology and Bioengineering 55, 630e641. Liang, P., Huang, X., Qian, Y., 2006a. Excess sludge reduction in activated sludge process through predation of Aeolosoma hemprichi. Biochemical Engg Journal 28, 117e122. Liang, P., Huang, X., Qian, Y., Wei, Y., Ding, G., 2006b. Determination and comparison of sludge reduction rates caused by microfaunas’ predation. Bioresource Technology 97, 854e861. Lin, S., Jin, Y., Fu, L., Quan, C., Yang, Y.S., 2009. Microbial community variation and functions to excess sludge reduction in a novel gravel contact oxidation reactor. Journal of Hazardous Materials 165, 1083e1090. Liu, Y., 1996. Bioenergetic interpretation on the So/Xo in substrate-sufficient batch culture. Water Research 30 (11), 2766e2770. Liu, Y., 2000. The So/Xo-dependent dissolved organic carbon distribution in substrate-sufficient batch culture of activated sludge. Water Research 34 (5), 1645e1651.
Liu, Y., Tay, J.H., 2001. Strategy for minimization of excess sludge production from the activated sludge process. Biotechnology Advances 19 (2), 97e107. Liu, Y., Chen, G.H., Paul, E., 1998. Effect of the So/Xo ratio on energy uncoupling in substrate-sufficient batch culture of activated sludge. Water Research 32 (10), 2833e2888. Lobos, J., Wisniewski, C., Heran, M., Grasmick, A., 2008. Sequencing versus continuous membrane bioreactors: effect of substrate to biomass ratio (F/M) on process performance. Journal of Membrane Science 317, 71e77. Low, E.W., Chase, H.A., 1999a. Reducing production of excess biomass during wastewater treatment. Water Research 33 (5), 1119e1132. Low, E.W., Chase, H.A., 1999b. The effect of maintenance energy requirements on biomass production during wastewater treatment. Water Research 33 (3), 847e853. Lundin, M., Olofsson, M., Pettersson, G., Zetterlund, H., 2004. Environmental and economic assessment of sewage sludge handling options. Resources, Conservation and Recycling 41, 255e278. Luxmy, B.S., Kubo, T., Yamamoto, K., 2001. Sludge reduction potential of metazoa in membrane bioreactors. Water Science and Technology 44 (10), 197e202. Madoni, P., 1994. A sludge biotic index (SBI) for the evaluation of the biological performance of activated sludge plants based on the microfauna analysis. Water Research 28, 67e75. Mannan, S., Fakhru’l-Razi, A., Alam, Z.Md, 2005. Use of fungi to improve bioconversion of activated sludge. Water Research 39, 2935e2943. Mason, C.A. (1986) Microbial death, lysis and Cryptic growth: fundamental and applied aspects. PhD Thesis, Swiss Fed. Inst. Tecnol. Zurich. Mason, C.A., Hamer, G., 1987. Cryptic growth in Klebsiella pneumonia. Applied Microbiology and Biotechnology 25, 577e584. Mason, C.A., Hamer, G., Bryers, J.D., 1986. The death and lysis of microorganism in environmental process. FEMS Microbiology Review 39, 373e401. McWhirter, J.R., 1978. Oxygen and Activated Sludge Process. The Use of High-Purity Oxygen in the Activated Sludge Process, vol. 1. CRC Press, Boca Raton, FL: USA. Meng, F., Chae, S., Drews, A., Kraume, M., Shin, H., Yang, F., 2009. Recent advances in membrane bioreactors (MBRs): membrane fouling and membrane material. Water Research 43 (6), 1489e1512. Metcalf and Eddy, Inc., 2003. Wastewater Engineering: Treatment, Disposal and Reuse, fourth rev. ed. Tata McGraw-Hill, New Delhi. Mishima, K., Nakamura, M., 1991. Self-immobilization of aerobic activated sludge: a pilot study of the aerobic upflow sludge blanket process in municipal sewage treatment. Water Science and Technology 23, 981e990. Molla, A.H., Fakhru’l-Razi, A., Alam, M.Z., 2004. Evaluation of solid-state bioconversion of domestic wastewater sludge as a promising environmental friendly disposal technique. Water Research 38 (19), 4143e4152. More, T.T., Yan, S., Tyagi, R.D., Surampalli, R.Y., 2010. Potential use of filamentous fungi for wastewater sludge treatment. Bioresource Technology 101, 7691e7700. Moussa, M.S., Hooijmans, C.M., Lubberding, H.J., Gijzen, H.J., van Loosdrecht, M.C.M., 2005. Modelling nitrification, heterotrophic growth and predation inactivated sludge. Water Research 39, 5080e5098. Mudrack, K., Kunst, S., 1991. Biologie der Abwasserreinigung (Biology of Wastewater Purification), third ed. Gustav Fischer Verlag, Stuttgart, Germany. Mulder, J.W. (2009) Operational experiences with the hybrid MBR Heenvliet, a smart way of retrofitting. Book of Proceedings of final MBR Network workshop, 31 Marche1 April, 2009, Berlin, Germany.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 2 8 7 e4 3 1 0
Muller, E.B., Stouthamer, A.H., van Verseveld, H.W., Eikelboom, D. H., 1995. Aerobic domestic waste water treatment in a pilot plant with complete sludge retention by cross-flow filtration. Water Research 29 (4), 1179e1189. Municipal Solid Wastes (Management and Handling) Rules, 2000. Ministry of Environment and Forests, Government of India. Neyens, E., Baeyens, J., Weemaes, M., De heyder., B., 2003a. Alkaline thermal sludge hydrolysis. Journal of Hazardous Materials 97 (1e3), 295e314. Neyens, E., Baeyens, J., Weemaes, M., De heyder, B., 2003b. Hot acid hydrolysis as a potential treatment of thickened sewage sludge. Journal of Hazardous Materials 98 (1e3), 275e293. Neyens, E., Baeyens, J., Weemaes, M., De heyder., B., 2003c. Pilot scale peroxidation (H2O2) of sewage sludge. Journal of Hazardous Materials 98 (1e3), 91e106. Nolasco, M.A., Campos, A.L.O., Springer, A.M., Pires, E.C., 2002. Use of lysis and recycle to control excess sludge production inactivated sludge treatment: bench scale study and effect of chlorinated organic compounds. Water Science and Technology 46 (10), 55e61. Nowak, G., Brown, G., Yee, A., 1986. Effects of feed pattern and dissolved oxygen on growth of filamentous bacteria. Journal of the Water Pollution Control Federation 58, 978e984. Onyeche, T.I., Schla¨fer, O., Bormann, H., Schro¨der, C., Sievers, M., 2002. Ultrasonic cell disruption of stabilized sludge with subsequent anaerobic digestion. Ultrasonics 40 (1e8), 31e35. Palmegren, R., Jorand, F., Nielsen, P.H., Block, J.C., 1998. Influence of oxygen limitation on the cell surface properties of bacteria from activated sludge. Water Science and Technology 37, 349e352. Pena, C., Trujillo-Roldan, M.A., Galindo, E., 2000. Influence of dissolved oxygen tension and agitation speed on alginate production and its molecular weight in cultures of Azotobacter vinelandii. Enzyme and Microbial Technology 27, 390e398. PEGASUS Research Institute (PERI), 2007. The Report of Questionnaire Survey on Waste Treatment Facilities in Japan (in Japanese). Pirt, S.J., 1975. Principles of Microbe and Cell Cultivation, first ed. Blackwell Scientific Publications, Oxford. Pollice, A., Giordano, C., Laera, G., Saturno, D., Mininni, G., 2007. Physical characteristics of the sludge in a complete retention membrane bioreactor. Water Research 41, 1832e1840. Puigagut, J., Salvado´, H., Tarrats, X., Garcı´a, J., 2007. Effects of particulate and soluble substrates on microfauna populations and treatment efficiency in activated sludge systems. Water Research 41, 3168e3176. Quintana, M., Sanchez, E., Colmenarejo, M.F., Barrera, J., ıa, G., Borja, R., 2005. Kinetics of phosphorus removal and Garc struvite formation by the utilization of by-product of magnesium oxide production. Chemical Engineering Journal 111, 45e52. Ratsak, C.H., Verkuijlen, J., 2006. Sludge reduction by predatory activity of aquatic oligochaetes in wastewater treatment plants: science or fiction? A review. Hydrobiologia 564, 197e211. Ratsak, C.H. (1994) Grazer induced sludge reduction in wastewater treatment. PhD thesis, Vrije University, The Netherlands. Ratsak, C.H., 2001. Effects of Nais elinguis on the performance of an activated sludge plant. Hydrobiologia 463, 217e222. Ratsak, C.H., Kooi, B.W., van Verseveld, H.W., 1994. Biomass reduction and mineralization increase due to the ciliate Tetrahymena pyriformis grazing on the bacterium Pseudomonas fluorescens. Water Science and Technology 29 (7), 119e128. Ratsak, C.H., Kooijman, S.A.L., Kooi, B.W., 1993. Modelling the growth of an oligochaete on activated sludge. Water Research 27 (5), 739e747. Ratsak, C.H., Maarsen, K.A., Kooijman, S.A.L., 1996. Effects of protozoa on carbon mineralization in activated sludge. Water Research 30 (1), 1e12.
4309
Rensink, J.H., Rulkens, W.H., 1997. Using metazoa to reduce sludge production. Water Science and Technology 36 (11), 171e179. Rensink, J.H., Corstanje, R., van der Pal, J.H. (1996) A new approach to sludge reduction by metazoa. 10th European Sewage and Reuse Symposium, IFAT 1996, Munchen. Resource Conservation and Recovery Act (RCRA of 1976), (1984) 42 USCx6901, 98e616, N. Richard, M., Hao, O., Jenkins, D., 1985. Growth kinetics of Sphaerotilus species and their significance in activated sludge bulking. Journal of the Water Pollution Control Federation 57, 68e81. Rittmann, B.E., McCarty, P.L., 2001. Environmental Biotechnology: Principles and Applications, first ed. McGraw-Hill, New York. Rocher, M., Goma, G., Begue, A.P., Louvel, L., Rols, J.L., 1999. Towards a reduction in excess sludge production inactivated sludge processes: biomass physicochemical treatment and biodegradation. Applied Microbiology and Biotechnology 51 (6), 883e890. Rocher, M., Roux, G., Goma, G., Begue, A.P., Louvel, L., Rols, J.L., 2001. Excess sludge reduction in activated sludge processes by integrating biomass alkaline heat treatment. Water Science and Technology 44 (2e3), 437e444. Roels, J., Wambecq, T., De Gussem, K., Fenu, A. (2010) LCA and nutrient removal. Neptune and Innowatech, End User Conference, Gent, Belgium. Roques, H., Capdeville, B., Seropian, J.C., Grigoropoulou, H., 1984. Oxygenation by hydrogen peroxide of the fixed biomass used in biological water treatment. Water Research 18, 103e110. Rosenberger, S., Kraume, M., Szewzyk, U., 1999. Sludge free management of membrane bioreactors. In: The 2nd Symposium on Membrane Bioreactor for Wastewater Treatment. The School of Water Sciences, Cranfield University, UK. Rosenberger, S., Kruger, U., Witzig, R., Manz, W., Szewzyk, U., Kraume, M., 2002. Performance of a bioreactor with submerged membranes for aerobic treatment of municipal wastewater. Water Research 36 (2), 413e420. Rosenberger, S., Laabs, C., Lesjean, B., Gnirss, R., Amy, G., Jekel, M. , Schrotter, J.C., 2006. Impact of colloidal and soluble organic material on membrane performance in membrane bioreactors for municipal wastewater treatment. Water Research 40, 710e720. Rosenberger, S., Witzig, R., Manz, W., Szewzyk, U., Kraume, M., 2000. Operation of different membrane bioreactors: experimental results and physiological state of the microorganisms. Water Science and Technology 41 (10e11), 269e277. Ryan, F.J., 1959. Bacterial mutation in a stationary phase and the question of cell turnover. Journal of General Microbiology 21, 530e549. Saby, S., Djafer, M., Chen, G.H., 2002. Feasibility of using a chlorination step to reduce excess sludge in activated sludge process. Water Research 36 (3), 656e666. Saby, S., Djafer, M., Chen, G.H., 2003. Effect of low ORP in anoxic sludge zone on excess sludge production in oxic-settling anoxic activated sludge process. Water Research 37 (1), 11e20. Saiki, Y., Imabayashi, S., Iwabuchi, C., Kitagawa, Y., Okumura, Y., Kawamura, H., 1999. Solubilization of excess activated sludge by self-digestion. Water Research 33 (8), 1864e1870. Salvado, H., Gracia, M.P., Amigo, J.M., 1995. Capability of ciliated protozoa as indicators of effluent quality in activated-sludge plants. Water Research 29, 1041e1050. Scott Blair, G.W., 1939. J. Phys. Chem. 43 (7), 853e864. Sengewein, H., 1989. Das Sauerstoffbelebungsverfahren, Abwasserreinigung mit reinem Sauerstoff. (The Activated Sludge Process on Oxygen, Waste Water Purification with Pure Oxygen). Academia Verlag Richarz, Sankt Augustin. Shanableh, A., 2000. Production of useful organic matter from sludge using hydrothermal treatment. Water Research 34 (3), 945e951.
4310
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 2 8 7 e4 3 1 0
Siegrist, H., Brunner, I., Koch, G., Phan, L.C., Le, V.C., 1999. Reduction of biomass decay rate under anoxic and anaerobic conditions. Water Sci. Technol 39, 129e137. Song, B., Xiaofei, C., 2009. Effect of Aeolosoma hemprichi on excess activated sludge reduction. Journal of Hazardous Materials 162, 300e304. Stouthamer, A.H., 1979. The search for correlation between theoretical and experimental growth yields. International Review of Biochemistry 21, 1e15. Stryer, L., 1988. Biochemistry, third ed. Freeman and Company, New York. Subramanian, B.S., Yan, S., Tyagi, R.D., Surampalli, R.Y. (2006) Isolation of extra cellular biopolymer producing microorganisms from wastewater sludge for sludge settling and dewatering. Proceedings of the 79th Annual Water Environment Federation Technical Exposition and Conference; Dallas, Texas Oct 21e25, 2006; Alexandria, Virginia, USA. Subramanian, B.S., Yan, S., Tyagi, R.D., Surampalli, R.Y., 2008. A new, pellet-forming fungal strain: its isolation, molecular identification, and performance for simultaneous sludgesolids reduction, flocculation, and dewatering. Water Environment Research 80 (9), 840e852. Sun, D.D., Khor, S.L., Hay, C.T., Leckie, J.O., 2007. Impact of prolonged sludge retention time on the performance of a submerged membrane bioreactor. Desalination 208, 101e112. Tanaka, M., 1999. Recent trends in recycling activities and waste management in Japan. Material Cycles and Waste Management 1, 10e16. Tanaka, S., Kobayashi, T., Kamiyama, K., Bildan, M.L., 1997. Effects of thermochemical pre-treatment on the anaerobic digestion of waste activated sludge. Water Science and Technology 36 (8), 209e215. Tiehm, A., Nickel, K., Neis, U., 1997. The use of ultrasound to accelerate the anaerobic digestion of sewage sludge. Water Science and Technology 36 (11), 121e128. Tiehm, A., Nickel, K., Zellhorn, M., Neis, U., 2001. Ultrasonic waste activated sludge disintegration for improving anaerobic stabilization. Water Research 35 (8), 2003e2009. Tottie, O. (2007) Evaluation of sludge management in Wuhan, China. Master Thesis, UPTEC W08 002, ISSN 1401e5765; Department of Microbiology, Swedish University of Agricultural Sciences, Uppsala University, Uppsala. Trivedi, H., Heinen, N., 2000. Simultaneous nitrification/ denitrification by monitoring NAOH Fluorescence in activated sludge. In: Proceedings of the Facility Operations II: Innovative Technology Forum; 73rd Annual Conference. WEF, Anaheim, CA. van Loosdrecht, M.C.M., Henze, M., 1999. Maintenance, endogenous respiration, lysis, decay and predation. Water Science and Technology 39 (1), 107e117. Visvanathan, C., Aim, R.B., Parameshwaran, K., 2000. Membrane separation bioreactors for wastewater treatment. Critical Reviews in Environmental Science and Technology 30 (1), 1e48. Wagner, J., Rosenwinkel, K.H., 2000. Sludge production in membrane bioreactors under different conditions. Water Science and Technology 41 (10e11), 251e258. Wang, W., Jung, Y., Kiso, Y., Yamada, T., Min, K., 2006. Excess sludge reduction performance of an aerobic SBR process equipped with a submerged mesh filter unit. Process Biochemistry 41, 745e751. Watson, T.G., 1970. Effects of sodium chloride on steady-state growth and metabolism of Saccharomyces cerevisiae. J. Genet. Microbiol. 64, 91e99. Weemaes, M., Grootaerd, H., Simoens, F., Huysmans, A., Verstraete, W., 2000a. Ozonation of sewage sludge prior to anaerobic digestion. Water Science and Technology 42 (9), 175e178.
Weemaes, M., Grootaerd, H., Simoens, F., Verstraetem, W., 2000b. Anaerobic digestion of ozonized biosolids. Water Research 34 (8), 2330e2336. Wei, Y., Van Houten, R.T., Borger, A.R., Eikelboom, D.H., Fan, Y., 2003b. Comparison performances of membrane bioreactor (MBR) and conventional activated sludge (CAS) processes on sludge reduction induced by Oligochaete. Environmental Science and Technology 37 (14), 3171e3180. Wei, Y., Van Houten, R.T., Borger, A.R., Eikelboom, D.H., Fan, Y., 2003a. Minimization of excess sludge production for biological wastewater treatment. Water Research 37, 4453e4467. Wei, Y., Wang, Y., Guo, X., Liu, J., 2009a. Sludge reduction potential of the activated sludge process by integrating an oligochaete reactor. Journal of Hazardous Materials 163, 87e91. Wei, Y., Zhu, H., Wang, Y., Li, J., Zhang, P., Hu, J., Liu, J., 2009b. Nutrients release and phosphorus distribution during oligochaetes predation on activated sludge. Biochemical Engineering Journal 43, 239e245. Welander, T, and Lee, N.M. (1994) Minimization of sludge production in aerobic treatment by use of predators. The Second International Symposium on Environmental Biotechnology, Brighton, UK. Westgarth, W.C., Sulzer, F.T., Okun, D.A. (1964) Anaerobiosis in the activated sludge process. Proceedings of the Second IAWPRC Conference. Tokyo, Japan. Wilen, B.M., Peter, B., 1999. The effect of dissolved oxygen concentration on the structure, size and size distribution of Activated sludge flocs. Water Research 33 (2), 391e400. Witzig, R., Manz, W., Rosenberger, S., Kruger, U., Kraume, M., Szewzyk, U., 2002. Microbiological aspects of a bioreactor with submerged membranes for aerobic treatment of municipal wastewater. Water Research 36 (2), 394e402. Wozniak, T., 2010. MBR design and operation using MPE-technology (Membrane Performance Enhancer). Desalination 250, 723e728. Wunderlich, R., Barry, J., Greenwood, D., Carry, C., 1985. Startup of a high-purity, oxygen-activated sludge system at the Los Angeles County Sanitation Districts’ Joint water pollution control plant. Journal of the Water Pollution Control Federation 57, 1012e1018. Xie, Y., Xu, C., Tong, Z., Kenji, F., Tsutomu, N., 2008. Operation Character of a Metal MBR in Treating the Domestic Sewage, vol. 29. Dongbei Daxue Xuebao/Journal of Northeastern University, Shengyang (Chinese)160e164. Xing, C.H., Yamamoto, K., Fukushi, K., 2006. Performance of an inclined-plate membrane bioreactorat zero excess sludge discharge. Journal of Membrane Science 275, 175e186. Yamamoto, K., Hiasa, M., Mahmood, T., Matsuo, T., 1989. Direct solideliquid separation using hollow fibre membrane in an activated sludge aeration tank. Water Science and Technology 21, 43e54. Yasui, H., Nakamura, K., Sakuma, S., Iwasaki, M., Sakai, Y., 1996. A full-scale operation of a novel activated sludge process without excess sludge production. Water Science and Technology 34 (3e4), 395e404. Yoon, S.-H., Kim, H.-S., Yeom, I.-T., 2004. The optimum operational condition of membrane bioreactor (MBR): cost estimation of aeration and sludge treatment. Water Research 38, 37e46. Zhang, S. (2000) Polluted water treatment by the combining processes of membrane separation and biodegradation. PhD thesis, Research Centre for Eco-Environmental Sciences, Chinese Academy of Sciences, China. Zubair, A., Chob, J., Lim, B., Song, K., Ahn, K., 2007. Effects of sludge retention time on membrane fouling and microbial community structure in a membrane bioreactor. Journal of Membrane Science 287, 211e218. Zupancic, G.D., Ros, M., 2008. Aerobic and two-stage anaerobiceaerobic sludge digestion with pure oxygen and air aeration. Bioresource Technology 99, 100e109.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 1 1 e4 3 4 0
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Bioassays as a tool for evaluating advanced oxidation processes in water and wastewater treatment Luigi Rizzo* Department of Civil Engineering, University of Salerno, via Ponte don Melillo 1, 84084 Fisciano (SA), Italy
article info
abstract
Article history:
Advanced oxidation processes (AOPs) have been widely used in water and wastewater
Received 5 April 2011
treatment for the removal of organic and inorganic contaminants as well as to improve
Received in revised form
biodegradability of industrial wastewater. Unfortunately, the partial oxidation of organic
24 May 2011
contaminants may result in the formation of intermediates more toxic than parent
Accepted 29 May 2011
compounds. In order to avoid this drawback, AOPs are expected to be carefully operated
Available online 12 June 2011
and monitored, and toxicity tests have been used to evaluate whether effluent detoxification takes place. In the present work, the effect of AOPs on the toxicity of aqueous
Keywords:
solutions of different classes of contaminants as well as actual aqueous matrices are
Antibiotics
critically reviewed. The dualism toxicityebiodegradability when AOPs are used as pre-
Biodegradability
treatment step to improve industrial wastewater biodegradability is also discussed. The
Drinking water
main conclusions/remarks include the followings: (i) bioassays are a really useful tool to
Dyes
evaluate the dangerousness of AOPs as well as to set up the proper operative conditions, (ii)
Emerging contaminants
target organisms for bioassays should be chosen according to the final use of the treated
Endocrine disruptors
water matrix, (iii) acute toxicity tests may be not suitable to evaluate toxicity in the
Industrial wastewater
presence of low/realistic concentrations of target contaminants, so studies on chronic
Oxidation intermediates
effects should be further developed, (iv) some toxicity tests may be not useful to evaluate
Pesticides
biodegradability potential, in this case more suitable tests should be applied (e.g., activated
Pharmaceuticals
sludge bioassays, respirometry).
Photocatalysis
ª 2011 Elsevier Ltd. All rights reserved.
Toxicity Urban wastewater Xenobiotics
Contents 1.
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.1. Bioassays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.1.1. Invertebrate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.1.2. Plants and algae . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.1.3. Microbial bioassays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.1.4. Fish bioassays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
* Tel.: þ39 089 969334; fax: þ39 089 969620. E-mail address:
[email protected]. 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.05.035
4312 4313 4313 4314 4314 4315
4312
2.
3.
4.
5.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 1 1 e4 3 4 0
1.2. Advanced oxidation processes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.3. Oxidation intermediates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Xenobiotics degradation by AOPs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1. Pharmaceuticals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2. Dyes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3. Pesticides . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Wastewater treatment by AOPs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1. Advanced treatment of urban wastewater . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2. Industrial wastewater treatment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2.1. Toxicity vs. biodegradability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2.2. Olive oil mill wastewater . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2.3. Textile wastewater . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2.4. Tannery wastewaters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2.5. Pesticides wastewaters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2.6. Pulp and paper mill wastewaters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Drinking water treatment by AOPs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1. NOM removal and DBPs control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2. Xenobiotics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.3. Microcystins . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusive remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Abbreviations AOC AOPs BOD5 BPA COD DBPs DOC EC EC50 EG FELST GAC ISO LC50 MBR MC
1.
assimilable organic carbon advanced oxidation processes biochemical oxygen demand after 5 days bisphenol A chemical oxygen demand disinfection by-products dissolved organic carbon emerging contaminants half maximal effective concentration energy gap fish early life stage toxicity test granular activated carbon international organization for standardization half maximal lethal concentration membrane biological reactor microcystins
Introduction
Advanced oxidation processes (AOPs) have been widely used in water and wastewater treatment for the removal of organic and inorganic contaminants as well as to improve biodegradability of industrial wastewater. In the last years different AOPs have been investigated in the removal of emerging contaminants from urban wastewater effluents (Baumgarten et al., 2007; Naddeo et al., 2009; Klamerth et al., 2010) and drinking water (Brose´us et al., 2009; Sanches et al., 2010). Unfortunately, the
4315 4316 4316 4316 4317 4317 4319 4319 4319 4319 4320 4323 4325 4325 4326 4329 4329 4330 4331 4334 4334 4334
NOAEL NOM OECD
no observed adverse effect level natural organic matter organization for economic cooperation and development OMW olive mill wastewater OUR oxygen uptake rate TOC total organic carbon TSS total suspended solids UF ultrafiltration membrane USEPA United States Environmental Protection Agency ultraviolet absorbance at xxx nm wavelength UVxxx UWWTP urban wastewater treatment plant YES yeast estrogen screen EDCs endocrine disrupting chemicals SUVA254 specific UV absorbance at 254 nm THMFP trihalomethanes formation potential
partial oxidation of organic contaminants may result in the formation of intermediates more toxic than parent compounds. In order to avoid this drawback, AOPs are expected to be carefully operated and monitored, and toxicity tests have been used to evaluate whether effluent detoxification takes place (Rizzo et al., 2009a; Klamerth et al., 2010). In the present work, the effect of AOPs on the toxicity of aqueous solutions of different classes of contaminants (such as dyes, pesticides and pharmaceuticals) as well as actual water (such as urban wastewater and drinking water) are critically reviewed.
4313
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 1 1 e4 3 4 0
AOPs have also been used as pre-treatment of industrial wastewater to improve biodegradability before the subsequent biological process (Arslan Alaton and Teksoy, 2007; Rizzo et al., 2008; Zapata et al., 2010). In this regard, sometime toxicity tests have been used to infer the behavior of treated wastewater in terms of biodegradability; but this approach may result in a misinterpretation of the effect of AOPs on the biodegradability of wastewater. Accordingly, some papers are also reviewed to elucidate the differences between toxicity and biodegradability when AOPs are used as pre-treatment step to improve industrial wastewater biodegradability. Moreover, the most used bioassays and AOPs are first introduced in this chapter. In particular it was believed important to provide the reader with some information about the main organisms used in the environmental applications as well as to summarize the corresponding standardized and most used methods.
1.1.
Bioassays
Bioassays rely on measuring the response of organisms exposed to contaminants relative to a control. They have been used to establish the toxicity levels of target contaminants and complex aqueous matrices (e.g., surface water, groundwater, wastewater) for aquatic organisms. The test organisms incorporated in these bioassays can be grouped in (Tothill and Turner (1996), Farre´ and Barcelo´ (2003)): microorganisms, plants and algae, invertebrates and fishes. Table 1 summarizes just (i) the most commonly used organisms to characterize toxicity of water and wastewater, (ii) the main organisms for each one group, (iii) the methods and (iv) some applications to water, wastewater and liquid waste (e.g., landfill leachate).
1.1.1.
Invertebrate
Invertebrates are widely used in the evaluation of toxic effects of pollutants in aqueous matrices. The most used organism to
Table 1 e Toxicity tests typically used in assessing risks to human health and aquatic life after water and wastewater treatment. Group Invertebrate
Plants and algae
Organism
Method
Daphnia magna
USEPA, 2002; ISO, 1996a
Sea urchin (Paracentrotus lividius and Sphaerechinus granularis) Brine shrimp (Artemia salina)
Pagano et al., 1982
Applications Disinfection of Hospital wastewater (Emmanuel et al., 2004) Drinking water treatment (Rizzo et al., 2005) Industrial wastewater treatment (Oral et al., 2007) Advanced treatment of urban wastewater (Rizzo et al., 2009a) Urban wastewater treatment (Hernando et al., 2005) Landfill leachate treatment (Marttinen et al., 2002) Industrial wastewater treatment (Meric¸ et al., 2005; Oral et al., 2007)
Persoone and Van Haecke, 1981; Migliore et al., 1997
Industrial wastewater treatment (Campos et al., 2002; Rodrigues et al., 2008; Pala´cio et al., 2009) Landfill leachate treatment (Silva et al., 2004)
Scenedesmus subspicatus Selenastrum capricornutum
ISO, 1989 ISO, 1989
Dunaliella tertiolecta Raphidocelis subcapitata Lettuce seeds (Lactuca sativa)
USEPA, 1988 Marttinen et al., 2002 OECD, 1984; USEPA, 1989
Industrial wastewater treatment (Tisler et al., 2004) Industrial wastewater treatment (Walsh et al., 1980; Tisler et al., 2004; Oral et al., 2007) Urban wastewater treatment (Hernando et al., 2005) Industrial wastewater treatment (Meric¸ et al., 2005) Landfill leachate treatment (Marttinen et al., 2002) Industrial wastewater treatment (Pala´cio et al., 2009)
Microrganisms Pseudomonas fluorescens strain P17 AOC method Drinking water quality and treatment and Spirillum sp. strain NOX (Van der Kooij et al., 1982; (Vrouwenvelder et al., 1998; Polanska et al., 2005; APHA/AWWA/WEF, 1998) Hammes et al., 2006; Lautenschlager et al., 2010) Wastewater reuse (Weinrich et al., 2010) Vibrio fischeri ISO, 1998 Disinfection of Hospital wastewater (Emmanuel et al., 2004) Industrial wastewater treatment (Tisler et al., 2004) Urban wastewater treatment (Hernando et al., 2005) Bacteria from activated sludge Respiration inhibition test Urban wastewater treatment (ISO, 2007; OECD, 2010) (Dalzell et al., 2002; Pagga et al., 2006) Industrial wastewater treatment (Burgess et al., 1999; Mert et al., 2010) Fishes
Zebrafish (Danio rerio) Rainbow trout (Oncorhynchus mykiss)
ISO, 1996b OECD, 1992
Industrial wastewater treatment (Tisler et al., 2004) Disinfected drinking water (Ferraris et al., 2005) Urban wastewater treatment plant effluent (Gagne´ et al., 2006) Advanced treatment of urban wastewater (Stalter et al., 2010)
4314
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 1 1 e4 3 4 0
characterize toxicity of water and wastewater treatment effluents is Daphnia magna. Acute lethality tests with Daphnia are well established and standardized (USEPA, 2002; ISO, 1996a). The organisms are exposed to target contaminants or aqueous matrices under controlled conditions and living (mobile) daphnias are counted after the required incubation period. Acute (24 and a 48 h) and chronic (21 days) toxicity tests with daphnids have also been reported (Tisler et al., 2004). The use of daphnids has many advantages for routine toxicity testing, such as high sensitivity to toxicants, short reproductive cycle and parthenogenetic reproduction (Tothill and Turner, 1996). Other tests have been developed such as those based on Artemia salina and sea urchin. A. salina serves as test organism in a wide range of toxicological assays and research; it was used in the screening of bioactive compounds in natural products, detection of cyanobacterial and algal toxicity in water, detection of anthropogenic chemicals in the environment and investigations into biochemical processes mediating acute toxic responses (Ruebhart et al., 2008). The Artemia bioassay is attractive for different reasons, including (i) the commercial availability of the cysts, (ii) Artemia can be maintained indefinitely in the laboratory in their cyst form and is easily induced to hatch, (iii) the assay is quick, simple, and performed at low cost, (iv) it requires small sample volume and can be performed with high sample throughput (microplates), and (v) it complies with animal ethics guidelines in many countries (Ruebhart et al., 2008). Sea urchin embryos have been either used to evaluate marine pollution or to test toxicity of specific pollutants and non-marine complex mixtures (Pagano et al., 2001; De Nicola et al., 2004). In addition, since tannery wastewater are typically characterized by high salinity, toxicity to sea urchin of tannery wastewater was also investigated (Meric¸ et al., 2005). Because of the high sensitivity of some invertebrates to high polluted aqueous matrices, such as industrial wastewater, these organisms may be not useful to characterize toxicity; in Section 3.2.1, the role of toxicity tests in industrial wastewater treatment by AOPs is better explained and discussed.
1.1.2.
Plants and algae
Bioassays based on plants are characterized by low maintenance cost and different assessment endpoints (e.g., germination rate, biomass weight, enzyme activity). For instance, Valerio et al. (2007) evaluated the sensitivity of lettuce in seed germination, root elongation, germination rate and root necrosis by exposing lettuce to different concentrations of soluble elements in soilewater solutions, as an alternative way to determine soil toxicity. Unfortunately, tests based on growth responses of plants require a long time, generally 4e6 days for root length measurements. Plant based bioassays have been used to evaluate the toxicity of organic and inorganic contaminants (Date et al., 2005; Wieczorek and Wieczorek, 2007; Di Salvatore et al., 2008), nanoparticles (Lin and Xing, 2007), contaminated soils (Robidoux et al., 2004; Zorrig et al., 2010), solid waste and sludges (Renoux et al., 2001). Due to their ubiquity and short life cycle, which make algae suitable for toxicological studies, toxicity tests based on these
organisms have also been developed (Joubert, 1980; Wong et al., 1995; Radix et al., 2000; Pehlivanoglu and Sedlak, 2004). At the end of the exposure time, the number of algae is assessed using an automatic particle counter and inhibition of algal growth is used as the indicator of toxicity. The main disadvantages of algal methods are difficulty in culturing and, sometimes, lack of reproducibility (Farre´ and Barcelo´, 2003).
1.1.3.
Microbial bioassays
A wide variety of microbial techniques has been developed and is used as toxicity screening procedures. These bioassays use different mechanisms based on (i) capacity of microorganisms to transform carbon, sulfur or nitrogen, (ii) enzymatic activity, (iii) growth, mortality or photosynthesis, (iv) glucose uptake activity, (v) oxygen consumption and (vi) luminescence output (Tothill and Turner, 1996). Among these assimilable organic carbon (AOC) test from one side and respiration inhibition test and luminescent microbial tests from the other find wide application in drinking water and wastewater toxicity characterization respectively. The AOC bioassay can be considered a measurement of the potential of a given water to support bacterial regrowth. Bacterial regrowth was found to be significantly limited for AOC values lower than a few decades of mg L1 (Van der Kooij, 1990; LeChevallier et al., 1991). In AOC test, water samples are inoculated with Pseudomonas fluorescens strain P17 and Spirillum sp. strain NOX up to achieve a given initial bacterial density. Inoculated samples are incubated and bacterial growth is periodically monitored by spread plating. The activated sludge respiration inhibition test has been established as an effective method for evaluating the toxicity of chemicals to activated sludge bacteria. This method was standardized since the late 80s by ISO and OECD, and today updated guidelines are available (OECD, 2010; ISO, 2007). Respiration inhibition test evaluates the effects of a substance on microorganisms from activated sludge by measuring their respiration rate in the presence of different concentrations of the test substance. The respiration rates of samples of activated sludge fed with synthetic sewage are measured in an enclosed cell containing an oxygen electrode after a contact time of 3 h (OECD, 2010). In order to investigate realistic exposure scenario, longer or shorter (e.g., in the presence of volatile substances or rapidly degraded abiotically via hydrolysis 30 min can be used) contact times could be appropriate. The sensitivity of activated sludge should be checked in parallel with a suitable reference substance. The test is typically used to determine the ECx (e.g. EC50) of the test substance and/or the NOAEL (No observed adverse effect level). Luminescent microorganisms have been used in the production of several toxicity test instruments. The most used microorganism is the marine bioluminescent bacterium, Vibrio fischeri. The test relies on the change in the bacterial luminescent when the microorganisms are exposed to toxic chemicals. The bioluminescence inhibition of the V. fischeri test has been standardized (ISO, 1998) and it is commercially available in different versions. The advantages of those toxicity tests include short time of analysis and simplicity of operation. The bacteria are provided by manufacturers in a lyophilized form and they can be stored for several months to be used “on demand”.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 1 1 e4 3 4 0
1.1.4.
Fish bioassays
The purpose of the acute toxicity test with fish species is to help in the assessment of possible risk to similar species in natural environments, which may be due to, for instance, the disposal of wastewater treatment plants effluents into aquatic environments. The rainbow trout (Oncorhynchus mykiss), and bluegill sunfish (Lepomis macrochirus), are preferred species to meet this requirement since they are sensitive indicator species and a large data base which characterizes the response to environmental contaminants is available, but other species as identified in USEPA guideline (USEPA, 1996) may be used. The goal of the test is to determine concentrationeresponse curves for fish mortality (LC50), for 1e4 days in a static, static-renewal, or flow-through system. Generally, fish bioassays show good sensitivities but they have some drawbacks such as standardization problems, time consuming and they need specialized equipment and operators with adequate skills (Farre´ and Barcelo´, 2003). In spite of in vitro screening methods have been preferred due to logistical, cost, time constraints and reliability, chronic in vivo tests (such as fish early life stage toxicity test with rainbow trout) have been also investigated to achieve a more comprehensive information about toxicity of oxidation intermediates formed upon ozonation of urban wastewater (Stalter et al., 2010).
1.2.
Advanced oxidation processes
AOPs can be broadly classified as oxidation methods which promote the formation of highly reactive species, such as hydroxyl radicals (the strongest oxidants (E0 ¼ 2.8 V) after fluorine), which allow the degradation of organic and inorganic pollutants. AOPs include a combination of oxidants (e.g., H2O2; ozone), UV radiation, catalysts (e.g., Fe2þ, TiO2) and ultrasounds. Among the most investigated AOPs in water and wastewater treatment there are the followings: heterogeneous and homogeneous photocatalysis based on near ultraviolet (UV) or solar visible irradiation, ozonation, Fenton reaction, ultrasound, electrochemical and wet-air oxidation. AOPs can be used in wastewater treatment in order to (Fig. 1): 1. improve the quality of urban wastewater treatment plant (UWWTP) effluent by removing residual xenobiotics in order to decrease final toxicity and make finished wastewater reusable (Fig. 1a); 2. disinfect biologically treated urban wastewater to be reused as alternative to conventional chemical disinfectants (such
4315
as chlorine, chlorine dioxide and ozone) which result in formation of toxic disinfection by-products (Fig. 1a); 3. increase industrial wastewater biodegradability before conventional biological process (Fig. 1b); 4. remove or convert in their nontoxic forms metals can be found in industrial wastewater. Industrial wastewater can be characterized by high chemical oxygen demand (COD) concentration. AOPs can be suitably used for the treatment of industrial wastewater with relatively small COD ( diuron > isoproturon > atrazine > alachlor, and the following order for respirometric data: pentachlorophenol > alachlor > atrazine > isoproturon ¼ diuron (the agreement was found only for pentachlorophenol). According to Gutie´rrez et al. (2002), the authors explained the disagreement considering the different nature of the organisms used. Zapata et al. (2009) investigated the effect of solar photo-Fenton treatment of commercial pesticides in water on toxicity (V. fischeri) and biodegradability (ZahneWellens test). In spite of biodegradability threshold was achieved as pesticides active ingredients were almost removed, toxicity was still over 50% inhibition (Fig. 5). Additionally, because of continuous changes in toxicity results along oxidation treatment, the authors inferred that V. fischeri toxicity test is not a suitable way to detect biodegradability threshold. Accordingly, Amat et al. (2009) found out that toxicity to V. fischeri was still high after solar photo-Fenton treatment of a mixture of four commercial pesticides, although the effluent might be compatible with biological process. They concluded that toxicity bioassays could be very useful as preliminary tests to find out the proper time for the application of long-time biodegradability tests. When Schrank and co-workers investigated different AOPs (UV, TiO2/UV, O3 and O3/UV) to improve biodegradability of tannery wastewater they found out that the removal of pollutants with aromatic structures did not increase biodegradability, in spite of toxicity to D. magna was found to be stabile or decreased (Schrank et al., 2004). Actually, already in 1995 Strotmann and co-workers suggested the use of both respiration inhibitory tests (such as heterotrophic respiration activity and nitrification activity tests) for monitoring the biological activity of wastewater treatment plant and toxicity tests (specifically luminescent bacteria based tests) for screening the toxicity of the effluent (Strotmann et al., 1995). They also investigated the effect of shock loading conditions by adding 2.3-dichlorophenol and 3.5-dichlorophenol and observed that nitrification and heterotrophic respiration activities as well as TOC degradation were significantly reduced. Moreover, an increased inhibition of luminescent bacteria in the effluent was observed during the shock loading experiments.
3.2.2.
Olive oil mill wastewater
Mediterranean countries (particularly Spain, Italy and Greece) are the main olive oil producers in the world with 2.5 106 t per year, the 95% of the annual world production (Brenes et al., 1999). The related amount of olive mill wastewaters (OMW) was estimated as high as 30 106 m3 per year (Mantzavinos and Kalogerakis, 2005). OMW are characterized by high organic and total suspended solids (TSS) loads as well as acidic pH (Beccari et al., 2002). The organic matter mainly consists of polysaccharides, sugars, phenols, polyalcohols, proteins, organic acids and oil (Cabrera et al., 1996). The high phenols concentrations results in phytotoxicity and toxicity to wastewater which make them difficult to be treated by biological processes (Borja et al., 1996; Gernjak et al., 2004). In order to solve this problem, several AOPs have been
Table 3 e Toxicity tests used in evaluating the degradation of dyes and their oxidation intermediates by AOPs. Target dye
Investigated AOPs
Toxicity tests V. fischeri
Reactive Black 5 (20e100 mg L1)
UV/TiO2, wet-air oxidation, electro-Fenton and UV/electro-Fenton.
P. putida
Reactive Black 5 (5e300 mg L1)
Two different ultrasound devices (high-frequency plate type (279 and 817 kHz), and a 20 kHz low-frequency probe system). ozonation, Fenton, UV/H2O2, and photo-Fenton
V. fischeri
Photo-Fenton (6 W Philips black-light fluorescent lamp, 1.38 109 Einstein s1 light intensity) Ultrasound (plate type piezoelectric transducer, 520 kHz, 100 W)
V. fischeri
Remazol Brilliant Blue R, Acid Black 1
Fenton oxidation and Fe0/air process
V. fischeri
Acid dyebath consisted of three different acid dyestuffs (C.I. Acid Yellow 242, C.I. Acid Red 360 and C.I. Acid Blue 264, 30 mg L1 each one) and two dye auxiliaries (a leveling agent (1500 mg L1) and an acid donor (500 mg L1))
Fenton
Activated sludge inhibition test
Reactive Red 120 (20e100 mg L1)
Disperse Red 354 (100 mg L1) Procion Red H-E7B (250 mg L1)
Reactive red 141, reactive black 5, basic brown 4, basic blue 3 (20e60 mg L1)
P. putida
V. fischeri
V. fischeri
Although DOC was almost 80% removed after 5 h of illumination, the toxicity of the solutions is slightly decreased. No bacterial growth inhibition was observed after 90 min treatment in all AOP investigated. The toxicity increased in the early 15 min of treatment; only electro-Fenton and UV/electro-Fenton (90 min treatment) decreased toxicity under initial value. No toxicity was detected at 20 mg L1 dye solution as well as no change in toxicity profile was observed, even after 6 h of ultrasound treatment (817 kHz, 100 W) The photo-Fenton process was found to result in the lowest toxicity after 15 min treatment. EC50 values were larger than the DOC content at every measurement time. RR141 and RBk5 were non-toxic at 20 and 30 mg/L, while BBl3 and BBr4 were toxic at both concentrations. Total toxicity removal was accomplished within shorter contact time than that necessary for total dye degradation. The toxicity of Fe0/air-treated solution was significantly lower than that of Fenton-treated solution; no toxicity was detected after treatment by the Fe0/air process. The inhibitory effect of acid dyebath toward sewage sludge can be completely eliminated after Fenton oxidation.
References Bizani et al., 2006 Kusvuran et al., 2004 Kusvuran et al., 2005 Vajnhandl and Le Marechal, 2007 Neamtu et al., 2004 Garcia-Montano et al., 2006 Tezcanli-Guyer and Ince, 2003
Chang et al., 2009
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 1 1 e4 3 4 0
UV/TiO2 (TiO2 Degussa P-25 and Hombikat UV-100, Philips HPK 125 W high-pressure mercury lamp). UV/TiO2, electro-Fenton, wet-air oxidation, and UV/electro-Fenton.
Cibacron Red FN-R and Cibacron Yellow FN2R (50 mg L1)
Comments
Arslan Alaton and Teksoy, 2007
4321
Target pesticide Pirimiphos-methyl (PMM)
Dichlorvos (10e50 mg L1)
Dimethoate (5e50 mg L1)
Prometryn (20 mg L1)
Pentachlorophenol (50 mg L1), chlorfenvinfos (50 mg L1), isoproturon (50 mg L1), diuron (42 mg L1), alachlor (50 mg L1) and atrazine (38 mg L1) Pentachlorophenol (50 mg L1), isoproturon (50 mg L1), diuron (42 mg L1), alachlor (50 mg L1) and atrazine (38 mg L1) A mixture of 5 pesticides (alachlor, atrazine, chlorfenvinphos, diuron, isoproturon) Methomyl (50 mg L1)
Methylparathion (10 mg L1) Methylparathion (10e20 mg L1)
Toxicity tests
Comments
References
TiO2 photocatalysis (simulation of solar radiation by 125 W high-pressure mercury lamp equipped with 335 nm cut-off filter transmitting wavelengths > 340 nm). Heterogeneous photocatalysis (125 W high-pressure mercury lamp) using TiO2 and ZnO as catalysts þ oxidants (H2O2 or K2S2O8). Heterogeneous photocatalysis (125 W high-pressure mercury lamp) using TiO2 and ZnO as catalysts þ oxidants (H2O2 or K2S2O8).
V. fischeri
The toxicity of PMM solution was found to increase in the early min than slowly decreased.
Herrmann et al., 1999
V. fischeri
Evgenidou et al., 2005a
Photo-assisted Fenton reaction (500 mL Pyrex reactor, 125 W high-pressure mercury lamp, 30e35 C temperature, 10e100 mg H2O2 L1, 0e6 mg Fe3þ L1) TiO2 photocatalysis (125 W high-pressure mercury lamp) þ oxidants (H2O2 or K2S2O8).
V. fischeri
Toxicity increased under all investigated conditions (6 h irradiation time, 0.1 g TiO2/L, 0.5 and 0.1 g ZnO/L) with and without oxidants. In the TiO2 system the toxicity was only slightly reduced after 6 h of treatment; only the addition of peroxide was able to achieve complete detoxification (120 min). Zinc dissolution and photodissolution increased toxicity in ZnO system. A complete and almost complete detoxification was achieved within 60 min of illumination for dimethoate and methylparathion respectively.
Photo-Fenton/ozone (PhFO) and TiO2photocatalysis/ozone (PhCO). 1.6 g h1 O3 dose and 6 W black-light lamp.
V. fischeri
O3 (1.75 g h1) photo-Fenton (6 W black-light)
V. fischeri
A decreased toxicity (increase of the EC50 value) for diuron, while even after 3 h treatment atrazine and alachlor solutions remain toxic.
Farre´ et al., 2007
Photo-Fenton (Hanau Suntest Simulator equipped with a xenon lamp with total radiant flux of 80 mW cm2). Solar driven photo-Fenton and TiO2 photocatalysis pilot plant
V. fischeri
A sharp decrease in toxicity was observed at the beginning of the photo-Fenton process, followed by a stable toxic level during all the rest of the photo-treatment. Although a clear decrease in water toxicity treated by both processes was observed, EC50 was not clearly beneath threshold after pesticide disappeared. Toxicity on V. fischeri and D. magna was reduced almost completely after 90 min photocatalytic treatment. Toxicity was significantly decreased (over 80%), faster for V. fischeri (30 min) than D. magna (90 min). The formation of malaoxon, isomalathion or trimethyl phosphate esters correlated well with the induced toxicity which was observed in photocatalysis of malathion and radotion. Photo-Fenton was found to decrease toxicity better than TiO2 photocatalysis.
Lapertot et al., 2007
TiO2 photocatalysis (six 20 W UV-A lamps, the light intensity of each one lamp was 2.4 mW cm2) Solar driven TiO2 photocatalysis
Malathion (14.7 mg L1), radotion (11.3 mg L1), malaoxon (18.0 mg L1) and isomalathion (15.7 mg L1)
TiO2 photocatalysis (six 20 W low-pressure mercury fluorescent lamps)
Dipyrone (50 mg L1)
Solar driven photo-Fenton and TiO2 photocatalysis pilot plant
V. fischeri
V. fischeri
V. fischeri, D. magna and S. capricornotum V. fischeri and D. magna V. fischeri and D. magna Inhibition of acetylcholinesterase
V. fischeri
After a slight increase in the toxicity a 50% reduction was observed after 6 h. In presence of H2O2 a 100% toxicity reduction was achieved after 2 h. For all investigated pesticides but alachlor, PhFO first increases toxicity than decreases (after 60e120 min treatment depending on specific pesticide).
Evgenidou et al., 2005b
Evgenidou et al., 2007b
Evgenidou et al., 2007a Farre´ et al., 2005
Fernandez-Alba et al., 2002 Kim et al., 2006 Zoh et al., 2006 Kralj et al., 2007
Pe´rez-Estrada et al., 2007
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 1 1 e4 3 4 0
Dimethoate and methylparathion (10 mg L1)
Investigated AOPs
4322
Table 4 e Toxicity tests used in evaluating the degradation of pesticides and their oxidation intermediates by AOPs.
Gonze et al., 1999 Toxicity to V. fischeri first decreased (2 h, 220 kW m3) due to PCP removal then increased (up to 10 h) because of oxidation intermediates and H2O2 formation. Similar behavior was observed for D. magna. D. magna and V. fischeri
Kim et al., 2007 TiO2 photocatalysis with UV-B radiation was the most effective method to remove PCP and decrease toxicity.
Sodium pentachlorophenate (PCP) (0.1 mM solutions)
Pentachlorophenol (10 mg L1)
D. magna and Bacillus subtilis sp.
Photo-Fenton (2 L reactor, three 6 W Philips black-light fluorescent lamps (I ¼ 5 106 Einstein s1), controlled temperature (25 C) Photolysis and photocatalytic indoor (six-columns continuous flow photoreactor equipped with 6 UV lamps) and outdoor solar (eight quartz tube modules) reactors Ultrasound (electrical generator: frequency 500 kHz, power 0e100 W). Temperature was maintained constant (20 C), pH in the range 6.8 7.5.
V. fischeri
Segura et al., 2008
V. fischeri
Phenoxy-acid pesticides 2,4-D (50 mg L1) and 3,6-dichloro-2methoxy-benzoic acid (110 mg L1) Imidacloprid (100 mg L1)
Ionizing radiation
Toxicity was found to increase at low irradiation doses because of the formation of more toxic oxidation intermediates, which can be decomposed at larger doses. Both acute toxicity to D. magna and genotoxicity to B. subtilis sp. remain detectable even after significant removal of the pesticide has been achieved.
Drzewicz et al., 2004
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 1 1 e4 3 4 0
4323
investigated either alone or in combination with biological processes (Beltra´n-Heredia et al., 2001; Amat et al., 2003; Rizzo et al., 2008) but just a few works investigated their effect on either toxicity and biodegradability (Table 6). The capacity of AOPs to decrease toxicity was found to depend on AOP investigated, OMW characteristics and dilution factor, as well as organisms used for toxicity tests. Although wet hydrogen peroxide oxidation implementing FeeBEA zeolites as heterogeneous catalysts, was found to decrease toxicity to V. fischeri at some extent (from 100 to 70%) (Najjar et al., 2009), other investigated AOPs, such as electrochemical oxidation, photo-Fenton and ozonation, did not significantly improved biodegradability neither reduced toxicity and phytotoxicity (Andreozzi et al., 2008; Chatzisymeon et al., 2009; Justino et al., 2009). In this regard it is point of discussion the proposed use of AOPs and toxicity tests. Chatzisymeon et al. (2009) investigated the effect of electrochemical oxidation on OMW biodegradability and used toxicity tests as support information. Justino et al. (2009) evaluated the effect of coupled photo-Fenton-biological processes to final toxicity. On the opposite, Andreozzi et al. (2008) investigated the effect of different AOPs to OMW toxicity and phytotoxicity in order to evaluate the risks related to OMW disposal on agricultural land as well as to verify whether fertirrigation could really represent a cost-effective solution to the OMW disposal problem. But, as previously discussed, the use of AOPs as sole treatment is expected to be really expensive because of the formation of organic oxidation intermediates which may be more refractory to oxidation treatment than parent compounds. Moreover, the oxidation intermediates may be more toxic than parent compounds, thus increasing final toxicity/phytotoxicity. Accordingly, toxicity and phytotoxicity tests should be used either as a support tool to evaluate biodegradability of industrial wastewater after AOP treatment or to evaluate final toxicity/phytotoxicity after coupled AOP-biological treatment.
3.2.3.
Textile wastewater
These wastewaters are typically characterized by: (i) strong color because of residual dyes, (ii) presence of recalcitrant compounds (such as dyes, surfactants and sizing agents), (iii) high salinity, (iv) high temperature and (v) highly variable pH. Table 7 summarizes data and information on the bioassays used in evaluating textile wastewater treatment by AOPs. Selcuk et al. (2006) investigated the effect of pre-ozonation treatment on COD fractions removal and detoxification of textile wastewater. The authors found out that, under optimum conditions (86e96% of color, 33e39% of soluble COD and 57e64% of total COD removals), toxicity to D. magna can be decreased at some extant in diluted samples, and they inferred that pre-ozonation process improved biodegradability too. Unfortunately, according to the previous discussion, the optimum operating conditions to decrease toxicity does not necessarily result in optimum conditions for biodegradability (Gutie´rrez et al., 2002; Farre´ et al., 2007). In this regard, in a recent study biodegradability and toxicity of pre-ozonated textile wastewater at pilot scale were investigated (Somensi et al., 2010). The authors found out that biodegradability slightly increased along ozonation time and a minimum acceptable biodegradability threshold (roughly 0.4 BOD5/COD)
4324
Table 5 e Bioassays used in evaluating final toxicity of UWWTP effluents after advanced treatment with AOPs. Wastewater characteristics Mixture of municipal and industrial wastewater pre-treated by MBR (COD 45 mg L1).
Target contaminants
Investigated AOPs 1
D. magna, Photobacterium phosphoreum, umu (genotoxicity) test
Solar driven photo-Fenton pilot plant (3 m2 total illuminated area, 35 L total volume, 22 L irradiated volume) 50 mg L1 of H2O2, 5 mg L1 of Fe2þ 3 pharmaceuticals: TiO2 photocatalysis, catalyst loading in amoxicillin (10 mg L1), the range of 0.2e0.8 g L1, 125 W blackcarbamazepine (5 mg L1), light fluorescent lamp, 120 min and diclofenac (2.5 mg L1). maximum irradiation time. 3 pharmaceuticals: Sonolysis, 20 kHz ultrasound amoxicillin, carbamazepine generator, power density 25e100 W L 1 , initial pH 3e11, and air sparging, and diclofenac in 200 mL of wastewater (WW) spiked at different initial various concentrations Ozonation (maximum applied ozone concentration 1 mg O3/mg DOC)
V. fischeri
Three different wastewater samples: effluent from secondary biological treatment (SBT) (COD 17 mg L1, pH 8.3), SBT þ ozonation (SBT-O3), SBTO3 þ sand filtration (SBT-O3-SF) (COD 15 mg L1, pH 8.3). Secondary treated wastewater samples (pH 7.8, HCO3 7.2 meq L1, TOC 17 mg L1) from a Motril (Granada, Spain).
Effluent from urban wastewater treatment plant (DOC varied from 14.2 to 19.7 mg L1)
Three nitroimidazoles: Metronidazole (MNZ), Dimetridazole (DMZ), Tinidazole, (TNZ), 65, 150 and 300 mM initial concentrations. 85 micropollutants
V. fischeri (toxicity) and Salmonella typhimurium (mutagenicity)
D. magna, P. subcapitata, L. sativum D. magna, P.subcapitata, L. sativum
FELST with rainbow trout (Oncorhynchus mykiss)
Comments Genotoxic potential could be reduced by O3 but no comment is provided about the effects of the investigated AOPs on toxicity tests. Ozonation may either increase or decrease the toxic potential of secondary effluents. In general, the toxicity and mutagenicity results were inconsistent. The high removal rate of EC probably resulted in a large amount of more toxic intermediates thus increasing toxicity. In general the photocatalytic treatment did not completely reduce toxicity under the investigated conditions. Sonication decreased the toxicity in single compounds and mixtures to some extent; however, this was not achieved at all the experimental conditions applied in this study. Ozonation resulted in a significant developmental retardation of tested organisms.
References Baumgarten et al., 2007
Petala et al., 2008
Klamerth et al., 2010
Rizzo et al., 2009a,b
Naddeo et al., 2009
Stalter et al., 2010
Gamma irradiation (high-level 60Co source) The initial activity 2.22 1014 Bq (6.000 Ci), 310 Gy h1 dose rate at 20 cm distance from the source.
V. fischeri
Although gamma irradiation removed Sa´nchez-Polo TOC (70% after 700 Gy of treatment) et al., 2009 MNZ degradation by-products were not significantly mineralized, and toxicity did not decrease.
Full scale reclamation plant using ozonation (2 mg L1 and 5 mg L1 of O3 in pre-oxidation and main oxidation step respectively) and activated carbon filtration
V. fischeri, estrogenic activity (E-SCREEN assay), arylhydrocarbon receptor response (CAFLUX assay), neurotoxicity (acetylcholinesterase inhibition assay), phytotoxicity (PSII inhibition I-PAM assay) and genotoxicity (umuC assay)
The effect of the investigated Reungoat processes varied from one bioassay to et al., 2010 another but their combination was almost totally responsible for the overall observed toxicity reduction.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 1 1 e4 3 4 0
Ciprofloxacin, enrofloxacin, O3 and O3/UV (40 mg L O3 dosage, 20 moxifloxacin and their and 40 min treatment), H2O2 and H2O2/ precursors at 50 mg L1 UV (6 mL L1 of 30% H2O2) each one. Ozonation (applied ozone 2.5e8.0 mg L1, 2e30 min treatment)
Effluent from secondary biological treatment, pH 7.8, 640 mg CaCO3 L1 alkalinity, 40.0 NTU turbidity, 0.17 cm1 of UV254, COD 60.0 mg L1, total nitrogen 7.0 mg L1. 15 EC at 100 mg L1 initial Effluent from secondary biological concentrations treatment, COD 60.0 mg L1, DOC 25 mg L1. Effluent from secondary biological treatment, pH 8.1, BOD5 10 mg L1, TOC 4.51 mg L1, UV280 0.110 cm1, TSS 11 mg L1 Effluent from secondary biological treatment, pH 7.5, BOD5 4 mg L1, COD 10.5 mg L1, TOC 4.4 mg L1, UV280 0.079 cm1, TSS 4 mg L1
Toxicity tests
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 1 1 e4 3 4 0
80 60
1
2
3
inhibition (%)
40 20 0 -100 -20
-50
0
50
100
150
200
250
300
t 30w (min)
-40 -60
Fig. 2 e Final toxicity to Vibrio fischeri in a real wastewater effluent spiked with 15 emerging contaminants at low concentrations, after advanced oxidation by photo-Fenton in a pilot-scale solar compound parabolic collector (revised from Klamerth et al., 2010, with permission from Elsevier).
was detected only after 240 min treatment. Although toxicity to V. fischeri was found to decrease, the authors did not explain if the results were achieved under same ozonation treatment conditions (particularly, ozonation time). In a previous work the effect of pre-oxidation treatment on biodegradability of simulated textile wastewater was evaluated in terms of inhibition of microbial growth of activated sludge (Ledakowicz and Gonera, 1999). The inhibitory effect of O3/UV and O3/UV/ H2O2 processes on microbial growth during subsequent biodegradation of textile wastewater accounted for only 10%.
3.2.4.
Tannery wastewaters
Wastewater from leather tanning industry includes high organic loads and priority pollutants such as sulphite, chromium, synthetic tannins, oils, resins, biocides, detergents (Jochimsen et al., 1997; Tisler et al., 2004). Because of their low
4325
biodegradability different AOPs have been investigated. Among these just a few works evaluated the effect of AOPs on final toxicity (Table 8). In particular, when UV, TiO2/UV, O3 and O3/UV were investigated, toxicity to D. magna was found to be stable or decreased, but no increase in biodegradability was observed (Schrank et al., 2004). The same research group also studied Fenton and H2O2/UV processes and did not find any significant effect on toxicity to D. magna and V. fischeri (Schrank et al., 2005). On the opposite when the effect of TiO2 photocatalysis to A. salina was investigated in lower organic loading tannery wastewater compared to Schrank et al.’s (2004) work, toxicity was found to increase in diluted and undiluted samples in spite of COD and BOD5 removals (Sauer et al., 2006), confirming that the organisms used for bioassay strongly effect toxicity final outcome. The authors generically related this result to the formation of toxic intermediates, but taking into account that they also detected an increase in ammonia concentration in parallel, probably related to the degradation of aromatic and aliphatic compounds containing nitrogen through the oxidative breakdown of the CeN bonds, the increased toxicity may just be due to higher ammonia concentration (Marttinen et al., 2002).
3.2.5.
Pesticides wastewaters
Because of their persistence, toxicity and no biodegradability, pesticides can bio-accumulate as well as result in serious health problem; some of them have been included in the list of persistent organic pollutants by EU members Countries (Directive 2000/60/EC, 2000). Pesticides contamination comes from point and non-point sources, such as field runoff, field drainage pipes, wastewater treatment plants effluents, sewer overflows and runoff from farmyards (Neumann et al., 2002). In the province of Almeria (southern Spain) pesticides containers are selectively collected to be recycled. The recycling process includes a washing step which results in the formation of highly pesticides polluted water (Malato et al.,
Fig. 3 e Toxicity to D. magna and P. subcapitata exposed to the mixture samples (M) during photocatalytic treatment using 0.4 g TiO2 LL1 (reprinted from Rizzo et al., 2009a, with permission from Elsevier).
4326
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 1 1 e4 3 4 0
pesticides mixture (Oller et al., 2007). Moreover, the high toxicity after pre-oxidation did not result in any toxic effect to biomass in the subsequent biological treatment. Subsequent studies confirmed the high efficiency of solar photo-Fenton process in the removal of pesticides mixtures. Amat et al. (2009) observed that initial active ingredients concentrations decreased under detection limits after 36 min (time normalized with respect to solar radiation) treatment but, as expected, mineralization was a more time-consuming process, due to both the formation of organic intermediates and the presence of solvents in the commercial pesticides investigated. When measurements based on activated sludge were performed, toxicity was found to decrease according to the removal of the active ingredients; on the opposite, toxicity to V. fischeri was still high after treatment. Table 9 summarizes additional information about some papers available in scientific literature where bioassays have been used in evaluating pesticides wastewater treatment by AOPs.
a
b
3.2.6.
Fig. 4 e Cell growth inhibition of P. subcapitata (%) (a) and germination index (%) of L. sativum (b) exposed to M-spiked wastewater (WW D M) samples before and after photocatalytic treatment at 0.8 g TiO2 LL1 loading (reprinted from Rizzo et al., 2009a, with permission from Elsevier).
2007). Homogeneous (photo-Fenton) and heterogeneous (TiO2) photocatalysis processes were investigated in parallel at pilot scale to evaluate the oxidation of a mixture of five pesticides commonly used in the province of Almeria, and photo-Fenton was found to be much more efficient than TiO2 photocatalysis process for the degradation and mineralization of the
100 S1
% Inhib. 30 min., without dilution % Inhib. 30 min., diluted 1:10
S2 S3
Samples for Zahn-Wellens test
150
S4
60
S5
Toxicity limit
40
S6
100
DOC (mg/L)
% Inhibition
200
DOC
80
50
20 Total elimination of active ingredients
0
0 0
50
100
150
200
t 30 W (min)
Fig. 5 e Vibrio fischeri inhibition after 30 min exposure to samples partially treated by photo-Fenton at 20 mg/L of Fe2D (reprinted from Zapata et al., 2009, with permission from Elsevier).
Pulp and paper mill wastewaters
Wastewater from pulp and paper mill is refractory to biological process because of high concentration of lignin, an irregular tri-dimensional hydrophobic and aromatic polymer causing brown color. Moreover, cellulose pulp bleaching step results in the formation of several chlorinated compounds via chlorination, and others toxic organic compounds, including lignin-derived refractory ones. Different AOPs have been investigated for the treatment of pulp and paper mill wastewater. Catalkaya and Kargi (2008) investigated the effect of UV, UV/H2O2, UV/TiO2 and UV/H2O2/TiO2 on TOC and toxicity removal from the effluent of biological wastewater treatment plant of a pulp and paper manufacturing. Among the investigated processes UV/TiO2 resulted in the highest TOC and toxicity removals. The effect of pH (3, 7, 11) was also investigated and the best performance in terms of toxicity removal (resazurin method) was detected at neutral pH (7). Unfortunately, it is not clear from the manuscript if pH was adjusted before toxicity tests, so it may be point of discussion if the toxicity effect at pH 3 and 11 is due to pH extreme conditions (pH not adjusted) or to the formation of higher toxicity oxidation intermediates. More than 250 chemicals have been identified in pulp and paper effluents among these resin acids, which are diterpenoid carboxylic acids with characteristic three-rings chemical structure, naturally occur in softwood cellulose raw material and are released into mill effluents by pulping processes (Ledakowicz et al., 2006). The concern related to the release of resin acids into the environment is due to their toxic effect to D. magna and rainbow trout (Salmo gairdneri) at quite low concentrations in water (24 h-LC50: 0.26e1.89 mg/L; 96 h-LC50 0.4e1.1 mg/L, respectively) (Peng and Roberts, 2000). Ledakowicz et al. (2006) investigated the effect of different O3 based AOPs (O3, O3 þ UV, O3 þ UV þ H2O2) on the removal of different resin acids mixtures as well as their effect on toxicity and biodegradability. O3 treatment was found to decrease toxicity to V. fischeri at the lower investigated doses, but a subsequent increase was observed when O3 dose was increased to completely remove resin acids. Moreover, based on final COD concentrations measured before and after treatment with O3 and O3 based AOPs of model wastewater,
Table 6 e Bioassays used in evaluating olive mill wastewater treatment by AOPs. Wastewater
Investigated AOPs 1
Toxicity/biodegradability tests
Pilot-scale plant combining electroFenton, anaerobic digestion and UF.
V. fischeri and L. sativum seeds.
Composite wastewater samples (pH 5.2, 115 g COD L1, 32 g SS/L, 5.6 g phenols L1) from an olive mill plant with a daily olive processing capacity of 30 tones in Bursa City, Turkey. Diluted and undiluted wastewater samples from a 3-phase process olive mill plant, North Portugal.
Fenton (FeSO4$7H2OeH2O2) and Fenton-like (FeCl3$6H2OeH2O2) processes; pH 3, 500 mg H2O2 L1, 500e3500 mg L1 of FeSO4 and FeCl3 doses.
Activated sludge inhibition tests (ISO 8192).
Photo-Fenton process before and after biological treatment by three species of fungi (P. sajor caju, T. versicolor and P. chrysosporium)
D. longispina.
Filtered wastewater sample (pH 4.4, 40 g COD mg L1, 0.6 g TSS L1, 3.5 g phenols L1) from three-phase olive oil mill company, in Chania, Crete, Greece.
Electrochemical oxidation over borondoped diamond (BDD) electrodes (70 cm2 area, distance between them 0.01 m).
V. fischeri and ZahneWellens biodegradability tests
Wastewater samples (pH 5.2, 64.4 g COD L1, 13.7 g BOD5 L1, 61.9 g TSS L1, 6.4 g phenols L1) from traditional discontinuous olive oil extraction plant in Sfax, southern Tunisia. Centrifuged wastewater sample (78.7 g COD mg L1, 4.4 g phenols mg L1) collected by a plant located in Bari (South Italy).
Wet hydrogen peroxide oxidation implementing FeeBEA zeolites as heterogeneous catalysts, differing in the amount of iron.
V. fischeri
Ozonation, solar photolysis, solar modified photo-Fenton, solar modified photo-Fentoneozonation.
Toxicity to P. subcapitata and phytotoxicity to seeds of R. sativus, C. sativus and L. sativa
References
Toxicity to V. fischeri decreased to 38% inhibition after anaerobic treatment and totally decreased after UF posttreatment. UF also increased germination index of L. sativum up to 124% compared to 12% of untreated OMW. Fenton and Fenton-like processes considerably removed inhibitory effect from untreated OMW.
Khoufi et al., 2009
Photo-Fenton process was found to preserve or even increase the effluent toxicity. Furthermore, when OMW was pre-treated by photo-Fenton, the treatment with fungi did not significantly decreased toxicity. BDD electrochemical oxidation did not improve biodegradability as assessed by the ZahneWellens test. All samples were found to be highly toxic to V. fischeri with EC50 values never exceeding 5%. Biodegradability increased and toxicity to V. fischeri decreased from 100 to 70%.
Justino et al., 2009
2 h ozonation treatment only moderately reduced phytotoxicity to seeds and the treated samples were still toxic to P. subcapitata. Both long-term aerated storage of OMW under sunlight and the combined process solar modified photo-Fentoneozonation did not significantly contribute to seed germination.
Mert et al., 2010
Chatzisymeon et al., 2009
Najjar et al., 2009
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 1 1 e4 3 4 0
Settled OMW (pH 5.2, 95 g COD L , 19 g BOD5 L1, 15 g TSS L1, 11.5 g polyphenols L1) from a discontinuous olive oil processing plant located in Sfax (southern Tunisia).
Remarks
Andreozzi et al., 2008
4327
Wastewater
Investigated AOPs
Toxicity/biodegradability tests
Remarks
References
Pre-treatment by O3, UV and UV/H2O2 had positive effect on subsequent biodegradation. Toxicity to V. fischeri was only tested on untreated single textile constituents. The inhibitory effect of acid dyebath toward sewage sludge can be completely eliminated after Fenton oxidation.
Ledakowicz et al., 2001
Simulated wastewater (pH 7.38, 2154 mg COD L1, 1050 mg BOD5 L1).
O3, UV, H2O2, O3/H2O2, UV/H2O2
Inhibition of microbial growth of activated sludge and toxicity tests to V. fischeri
Acid dyebath consisted of three different acid dyestuffs (C.I. Acid Yellow 242, C.I. Acid Red 360 and C.I. Acid Blue 264, 30 mg L1 each one) and two dye auxiliaries (a leveling agent (1500 mg L1) and an acid donor (500 mg L1)) Wastewater sample (330.5 mg DOC L1, pH 6) taken right after the dyebath and before entering the activated sludge treatment of textile industry located in Thessaloniki (Greece). Two raw textile wastewater samples (COD: 1600, 1560 mg L1; soluble COD: 950, 900 mg L1; BOD5, 150, 170 mg L1; TSS, 750, 250 mg L1; Color, 3300, 1000 PteCo; pH 8.8, 8.6) from the balancing tank of a textile industry located in Istanbul (Turkey) Raw (COD: 910 mg L1; filtered COD: 560 mg L1; BOD5: 150 mg L1; TSS: 150 mg L1; Color: 1570 PteCo unit; pH 10) and biologically treated (COD: 210 mg L1; filtered COD: 170 mg L1; BOD5: 20 mg L1; TSS: 90 mg L1; Color: 1450 PteCo unit; pH 8.5) textile wastewater samples from a textile finishing industry in Istanbul, Turkey. Raw wastewater (COD: 1505 mg L1; BOD5: 91 mg L1; Color: 0.754 as A455; pH 9.1) from the equalization storage tank of a textile mill in Blumenau (Brazil). Simulated textile wastewater diluted with 10% synthetic domestic sewage (pH: 7.38; COD: 2154 mg L1; BOD5: 1050 mg L1).
Fenton
Activated sludge inhibition test
TiO2 photocatalysis (1.5 L Pyrex reactor, diving Philips HPK 125 W highpressure mercury lamp jacked with a water-cooled Pyrex filter restricting the transmission of wavelengths below 290 nm). Ozonation (Pyrex glass reactor: 40 mm diameter, 1100 mm height; 18.5e24 mg L1 ozone dose; 0.566 m3 h1 air flowrate).
V. fischeri
Although a complete decolorization of the wastewater was achieved within 6 h, inhibition to V. fischeri only decreased to some extent (z35% for 6h treatment).
Bizani et al., 2006
D. magna
Ozone was effective for removing acute toxicity to D. magna from diluted textile wastewater.
Selcuk et al., 2006
Fenton (FeSO4 and H2O2 doses between 100 and 400 mg L1 and 600 to 1200 mg L1, respectively; pH 3.0 and temperature 40 C) and ozonation (1.4 g O3 L1 h1 applied dose to 1 L samples for 20 min without any pH adjustment (pH 10.0)).
D. magna
Both processes were effective in removing wastewater toxicity.
Meric¸ et al., 2005
Ozonation (20 g m3; flowrate 1 m3 h1)
BOD5/COD for biodegradability and V. fischeri for toxicity
Ozonation treatment reduced toxicity and improved biodegradability
Somensi et al., 2010
O3/UV, O3/UV/H2O2 (1.5 L stirred gaseliquid reactor; UV lamps: 150 W, l ¼ 254e578 nm, 15 W, l ¼ 254 nm)
Inhibition of microbial growth of activated sludge
The inhibitory effect of O3/UV and O3/ UV/H2O2 processes on microbial growth during subsequent biodegradation of textile wastewater accounts for only 10%.
Ledakowicz and Gonera, 1999
4328
Table 7 e Bioassays used in evaluating textile wastewater treatment by AOPs.
Arslan Alaton and Teksoy, 2007
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 1 1 e4 3 4 0
4329
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 1 1 e4 3 4 0
Table 8 e Bioassays used in evaluating tannery wastewater treatment by AOPs. Wastewater Effluent of a mechanical and chemical industrial wastewater treatment plant in Brazil (Bonato Couros SA) (COD: 2365 mg L1; BOD5:1010 mg L1; TOC: 820 mg L1; DOC: 720 mg L1). Coagulated/settled tannery wastewater (COD: 130 mg L1; BOD5:47 mg L1; TOC: 45 mg L1; DOC: 44 mg L1). Coagulated/settled wastewater (COD: 801 mg L1; BOD5: 349 mg L1; Chromium(VI) 0.1 mg L1; sulfate 1150 mg L1; pH adjusted at 2.5).
Investigated AOPs
Toxicity/ biodegradability tests
References
UV (15 W medium pressure mercury lamp), TiO2/UV (1 g TiO2 L1; 120 min), O3 and O3/UV (2.6 g O3 h1, 60 and 30 min respectively).
D. magna
In spite of toxicity to D. magna was found to stable or decreased, no increase in biodegradability was observed.
Schrank et al., 2004
Fenton (pH 3.5, 2 h, 50e240 mg L1 of FeSO4, 100e500 mg L1 of H2O2) and H2O2/UV (15 W medium pressure mercury lamp, different H2O2 and pHs) H2O2/UV, TiO2/H2O2/UV and TiO2/UV in a continuous operated annular reactor (15 W UV-lamp, 1 g TiO2 L1, 1 h).
D. magna and V. fischeri
No increase or only a slight decrease in toxicity was observed compared to the untreated wastewater
Schrank et al., 2005
A. salina
A. salina mortality increased after TiO2/UV treatment (1 h) for diluted and undiluted samples
Sauer et al., 2006
the authors concluded that the investigated processes do not improve biodegradation; unfortunately, since specific biodegradability tests were not performed, results cannot be considered conclusive. Yeber et al. (1999) investigated the effect of different AOPs (O3/UV, O3/UV/ZnO, O3/UV/TiO2, O2/ UV/ZnO, O2/UV/TiO2), on the biodegradability and toxicity of the effluent from bleaching process of Pinus radiata wood. Acute toxicity to V. fischeri was found to decrease between 30 and 60% after treatment, being the O2/UV/TiO2 system the most efficient process. Moreover, the investigated processes were found to increase biodegradability too in terms of BOD/ COD ratio and activated sludge dry weight. Their results found confirmation in the study from Rodrigues et al. (2008) which investigated the combined treatment of coagulationeflocculation followed by UV driven AOPs (UV/TiO2, UV/ H2O2 and UV/TiO2/H2O2). UV/TiO2/H2O2 treatment was found to decrease A. salina mortality and improve biodegradability in terms of COD/BOD ratio. Table 10 includes some additional information about the above discussed papers in relation to wastewater characteristics and AOPs operating conditions.
4.
Remarks
Drinking water treatment by AOPs
AOPs have been investigated over the past years in drinking water treatment to (i) decrease natural organic matter (NOM) typically found in waters, in order to prevent the formation of disinfection by-products (DBPs), (ii) disinfect water (iii) remove xenobiotics (particularly pharmaceuticals and endocrine disruptors compounds) and (iv) microcystins. Typically, these contaminants occur at low concentrations in natural waters and ecotoxicity tests may be not enough sensitive to characterize toxicity of both parent compounds and oxidation intermediates so, sometime, when pharmaceuticals and EDCs
have been investigated estrogenic activity was evaluated in parallel. In the following sub paragraphs the above mentioned topics are reviewed. Table 11 summarizes data and information (water characteristics, target contaminants, equipments and operating conditions for investigated AOPs, toxicity and/ or estrogenic activity tests implemented and related results) for the papers reviewed in the subsequent paragraphs.
4.1.
NOM removal and DBPs control
The formation of DBPs can be controlled by reducing NOM, which occurs in natural waters. NOM, particularly humic acids, can react with disinfectants (chlorine gas and hypochlorites) to form chlorination by-products (Nikolaou et al., 2007). Process efficiency can be characterized in terms of UV absorbance (typically measured at the wavelength of 254 nm, UV254), DOC, specific UV absorbance (UV254/DOC ratio, SUVA254) and trihalomethanes formation potential (THMFP) measurements. The effects of different treatment methods (ozonation, coagulation, ozonation followed by coagulation and TiO2 photocatalysis) on the removal of organic matter from surface waters taken from different regions of Istanbul (Turkey) and Salerno (Italy) were investigated in terms of UV254, SUVA254 and DBP formation potentials and toxicity to D. magna (Bekbolet et al., 2005). The related removal efficiencies as well as toxicity results were found to be site and treatment specific. The effect of pre-ozonation treatment on coagulation process was also investigated in terms of TOC and THMFP removals as well as toxicity to D. magna (Selcuk et al., 2007). The authors did not observe any doseeresponse relationship between the measured DBPs and toxicity to D. magna. Taking into account that more than 500 DBPs have been reported in scientific literature (Richardson, 1998), the toxicity may be due to one or more among the unmeasured DBPs. Moreover, DBPs
4330
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 1 1 e4 3 4 0
Table 9 e Bioassays used in evaluating pesticides wastewater treatment by AOPs. Wastewater A mixture of five pesticides (Methomyl, Dimethoate, Oxamyl, Cymoxanil and Pyrimethanil 50 mg L1 each one)
A mixture of four commercial pesticides including methyloxydemeton, methidathion, carbaryl, dimethoate as active ingredients (50 mg L1 each one). Wastewater (33700 2100 mg COD L1, 6100 700 mg BOD5 L1) from a pesticide factory in Hebei Province, North of China. A mixture of five pesticides (Methomyl, Dimethoate, Oxamyl, Iimidacloprid and Pyrimethanil, 40e100 mg DOC L1 each one).
Investigated AOPs Solar TiO2 photocatalytic experiments: 35-L compound parabolic collectors pilot plant, pH 6, 0.2 g TiO2 L1. Solar Photo-Fenton tests: 75 L pilot plant, pH adjusted to 2.7e2.9, 20 and 55 mg Fe2þ L1, 200e500 mg H2O2 L1 Solar Photo-Fenton tests: 40 L pilot plant, pH adjusted to 2.8, 20 mg Fe2þ L1, 200e500 mg H2O2 L1.
Toxicity/ biodegradability tests V. fischeri and biodegradability assay (ZahneWellens test)
V. fischeri and biodegradability assays (OUR inhibition test, ZahneWellens test, BOD5/ COD ratio)
Photo-Fenton was much more efficient than TiO2 photocatalysis for pesticide degradation and mineralization. The high toxicity after preoxidation did not result in any toxic effect to biomass in the subsequent biological treatment. Toxicity to V. fischeri was still high after the removal of pesticides active ingredients, although the effluent might be compatible with biological process.
References Oller et al., 2007
Amat et al., 2009
Fenton oxidation (optimum conditions: 97 mmol H2O2 L1 and 40 mmol Fe2þ L1 at initial pH 3).
BOD5/COD ratio
The biodegradability (BOD5/ COD) of the wastewater was enhanced from 0.18 to more than 0.47.
Chen et al., 2007
Solar Photo-Fenton tests: 50 L pilot plant (2.25 m2 irradiated surface, 22 L irradiated volume), pH adjusted to 2.7e2.9, 20 mg Fe2þ L1, 100 mg H2O2 L1
P. putida biodegradability assay
Biodegradability first increased with mineralization, then decreased at the end of the treatment because of the formation of less biodegradable by-products.
Ballesteros Martı´n et al., 2009
typically occur at low concentrations in drinking water (from ng L1 to mg L1), with no relevant acute toxic effect on the investigated organisms; therefore, more sensitive and/or reliable chronic bioassays should be developed to characterize possible toxic effects of AOPs in drinking water applications.
4.2.
Remarks
Xenobiotics
Conventional drinking water treatments such as coagulation, sedimentation, filtration and adsorption may not be effective for the removal of certain classes of xenobiotics (Westerhoff et al., 2005; Stackelberg et al., 2007). Moreover, chlorine, the most widely used oxidant/disinfectant in the treatment of ground and surface water, may not significantly improve the removal of these pollutants (Gibs et al., 2007). Although AOPs have been widely investigated in the removal of xenobiotic compounds (particularly pharmaceuticals and endocrine disruptors compounds) from aqueous solutions, only a few studies have been focused on the decontamination of natural waters (surface and ground waters) and AOPs effect on toxicity and estrogenic activity. While slurry heterogeneous photocatalytic reactors have been found to be effective in the removal of a wide range of pharmaceuticals and EDCs at bench-scale (Calza et al., 2006; Me´ndez-Arriaga et al., 2008; Rizzo et al., 2009a), a major limitation to scaling up this technology has been the difficulty in separating the TiO2 following treatment. Benotti et al. (2009)
investigated a patented TiO2 photocatalytic reactor membrane pilot system (Photo-Cat) for the removal of thirty-two pharmaceuticals and endocrine disrupting compounds from river water as well as for the effect on estrogenic activity. Higher than 70% removal was detected for twenty-nine investigated contaminants (the remaining three being removed at less than 50%), but no estrogenically active transformation products were detected after photocatalytic treatment. The degradation of EDCs by AOPs and the effect on estrogenic activity have also been investigated in the last years (Chen et al., 2006; Rosenfeldt et al., 2007). Among numerous EDCs detected into the environment, bisphenol A (BPA) has received a great deal of attention because of widespread use in the production of polycarbonates, epoxy resins, numerous plastic articles, and dental sealants (Staples et al., 1998) and effects on human health and environment (Segner et al., 2003; Kang et al., 2006). BPA has been detected into surface water in concentrations from ng to mg L1 (Belfroid et al., 2002; Jin et al., 2004). Since, conventional water treatment processes are not effective in the removal of most EDCs, Chen et al. (2006) investigated direct photolysis with low-pressure Hg UV lamps and UV/H2O2 processes for the degradation of BPA as well as changes in estrogenic activity. They found out that UV/ H2O2 significantly removed BPA and estrogenic activity in vitro and in vivo. Moreover, different sensitivities of the bioassays were observed because of removal rates of in vivo estrogenic activity were significantly lower than those observed in vitro.
4331
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 1 1 e4 3 4 0
Table 10 e Bioassay used in evaluating pulp and paper mill wastewater treatment by AOPs. Wastewater Wastewater from bleaching process of Pinus radiata wood of pulp mill plant (VIII Region, Chile) (Color: 4510 mg Pt L1; pH 10.8; COD: 1550 mg L1; BOD: 534 mg L1; Cl: 47 mmol L1). Effluent of biological wastewater treatment plant of a pulp and paper manufacturing (Izmir, Turkey) (pH 7.08; COD: 592 mg L1; BOD: 240 mg L1) Resins acids based solutions (BOD5: 200e1300 mg L1; COD: 300e1900 mg L1; initial toxicity to Vibrio fischeri 30e96%)
Effluent of cellulose and paper industry (pH: 9.8; COD: 1303 mg L1; BOD: 148 mg L1).
Investigated AOPs
Toxicity/ biodegradability tests
O3/UV, O3/UV/ZnO, O3/UV/TiO2, V. fischeri O2/UV/ZnO, O2/UV/TiO2 (125 W high-pressure Hg lamp; 2 g O3 h1)
Remarks
References
All AOPs significantly improved Yeber biodegradability (BOD/COD) et al., 1999 and decreased toxicity to some extent.
Catalkaya and Kargi, 2008
UV, UV/H2O2, UV/TiO2, UV/ H2O2/TiO2 (16 W low-pressure mercury vapor lamp
Washed activated sludge UV/TiO2 resulted in a higher (Resazurin reduction toxicity removal compared to method) UV/H2O2/TiO2.
O3, O3 þ UV, O3 þ UV þ H2O2 (0.03e2.9 mgO3/mgCOD, 15 W low-pressure lamp, 2 ml H2O2 L1 of solution, temperature: 20, 50 and 80 C, initial pH: 9e10, after ozonation: 7e8). Combined treatment of coagulationeflocculation followed by photocatalysis (UV/ TiO2, UV/H2O2 and UV/TiO2/ H2O2); 250 W low-pressure mercury lamp.
V. fischeri
O3 treatment decreased toxicity Ledakowicz et al., 2006 in the range of 0.3e0.5 mgO3/ mgCOD (in parallel, for 3 of 4 resins mixtures the removal was >90%), then, for higher O3 doses toxicity increased.
A. salina and biodegradability assay (COD/BOD)
A. salina mortality decreased after coagulation and was improved by photocatalysis treatment. Biodegradability improved from 0.14 to 0.5 after coagulation, to 0.63 and 0.71 after 2 and 4 h photocatalytic treatment respectively.
Furthermore, the UV/H2O2 process was found to be effective for reducing larval lethality in treated BPA solutions, suggesting that the degradation process did not result in the production of acutely toxic oxidation intermediates. The same investigation group studied the degradation and the corresponding effect on estrogenic activity of other two EDCs, 17-bestradiol (E2) and 17-a-ethinyl estradiol (EE2) following direct UV photolysis and UV/H2O2 treatments (Rosenfeldt et al., 2007). UV/H2O2 process effectively decreased (90%) estrogenic activity of E2 and EE2 at environmentally relevant concentrations (roughly 3 mg L1). Furthermore, no statistically significant difference between removal rates of E2 and EE2 and the subsequent reduction in estrogenic activity was observed, implying that the oxidation products of E2 and EE2 are not as estrogenic as the parent compounds. Although drinking water mainly comes from groundwater in Italy, the increasing chemical contamination in some area makes surface water an important alternative supply. Ozone based AOPs (UV/O3 and UV/O3/H2O2) were compared with resin and granular activated carbon (GAC) adsorption to evaluate their effectiveness in the removal of toxic and mutagenic organic micropollutants from Como Lake waters (Italy) (Guzzella et al., 2002). The results showed a decrease of the mutagenic and toxic activities after adsorption on GAC and resins, while AOPs were generally found out to increase these parameters. When a GAC adsorption step was performed in addition to the AOPs, no toxicity and mutagenic activity was observed.
4.3.
Rodrigues et al., 2008
Microcystins
Microcystins (MCs) are a group of monocyclic heptapeptide hepatotoxins produced by different cyanobacteria such as Microcystis, Anabaena, Nostoc and Oscillatoria which can be found in fresh waters (Pe´rez and Aga, 2005). MCs exhibit acute and chronic effects on humans and wildlife as well as they can result in serious damage to the liver (Gupta et al., 2003; Weng et al., 2007). MC-LR is usually regarded as one of the most acutely toxic cyanobacteria toxins and WHO recommends a guideline value of 1 mg L1 for drinking water (WHO, 2008). Since MCs are stable against physicochemical and biological factors (such as temperature, sunlight and enzymes) and conventional water treatment processes have been proven to be unreliable for the removal of these toxins (Jurczak et al., 2005), advanced treatment such as AOPs have been investigated (Lawton et al., 2003; Al Momani et al., 2008; Miao et al., 2010). Liu et al. (2002) found out that TiO2 photocatalysis can effectively destroy MC-LR in aqueous solutions, confirming previous studies (Lawton et al., 1999), but it can also significantly reduce protein phosphatase 1 (PP1) inhibition, which is potentially one of the most harmful effects to humans who may consume water contaminated by MC. In spite of a rapid disappearance of the MC-LR after photocatalysis treatment, the PP1 inhibition only slightly decreased after 20 min reaction time. However, the inhibition was found to rapidly decrease after 30 min treatment and only 20% residual inhibition was detected after 60 min. Moreover, when H2O2 was added to photocatalytic
Water characteristics
Natural ground (pH 7.5, HCO3 8.8 meq L1, TOC 10.3 mg/L) and surface (pH 8.3, HCO3 6.4 meq L1, TOC 11.9 mg L1) water samples were collected from a drinking water treatment plant (Motril, Spain). Microcystins aqueous solutions
Surface water (Tai Lake, China).
Investigated AOPs
Toxicity/Estrogenic activity tests
Comments
References
Disinfection by-products
Ozone (10.5 mg L1 min1 for 5 min) and TiO2 photocatalysis (125 W black-light fluorescent lamp)
D. magna
The immobilization test results were in accordance with the DBPs distribution rather than NOM removal efficiency.
Bekbolet et al., 2005
Disinfection by-products
Ozone (10.5 mg L1 min1 for 5 min)
D. magna
No doseeresponse relationship was observed between DBPs and toxicity to D. magna. Spiked bromide just resulted in higher levels of brominated DBPs but no significant change was observed in the immobilization of D. magna.
Selcuk et al., 2007
Three relevant metabolites (10 and 50 mg L1 initial concentrations) of the analgesic and antipyretic drug dipyrone, 4-methylaminoantipyrine (4-MAA), 4-formylaminoantipyrine (4-FAA) and 4-acetylaminoantipyrine (4-AAA) Three nitroimidazoles: Metronidazole (MNZ), Dimetridazole (DMZ), Tinidazole, (TNZ), 65, 150 and 300 mM initial concentrations.
Simulated solar irradiation system (Suntest) equipped with a 1100 W xenon arc lamp and special filters (wavelength < 290 nm).
D. magna
24 and 48 h inhibition increased to 27e55% and 60e85% respectively after photolysis treatment. Since investigated metabolites were totally removed, the toxicity to D. magna is due to the formation of toxic intermediates.
Go´mez et al., 2008
Gamma irradiation (high-level 60 Co source) The initial activity 2.22 1014 Bq (6.000 Ci), 310 Gy h 1 dose rate at 20 cm distance from the source.
V. fischeri
Toxicity during MNZ gamma irradiation increased with longer treatment time up to an irradiation dose of 200e300 Gy for both source waters, but decreased at higher doses.
Sa´nchez-Polo et al., 2009
TiO2 photocatalysis (480 W xenon lamp)
Protein phosphatase inhibition (PP1) assay
Liu et al., 2002
Ozone (1e6 mg O3 mg1 MC).
In vitro (protein phosphatase inhibition method) and in vivo (mouse bioassay method) tests.
The photocatalytic process significantly decreased PP1 inhibition. Ozonation treatment drastically decreased toxicity.
Microcystin-LR extracted from a bloom of Microcystis aeruginosa Two Microcystins (MCs) MC-LR and MC-RR extracted from the natural cyanobacteria of Microcystis Aeruginosa
Miao et al., 2010
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 1 1 e4 3 4 0
Surface water samples from Buyukcekmece and Omerli (Istanbul, Turkey) and Alento (Salerno, Italy) (pH 7.65, 7.18, 7.97; alk. 150, 70, 170 mg CaCO3 L1; bromide 274, 95, nd mg L1; chloride 98, 45, 14 mg L1, TOC 3.61, 3.05, 2.47 mg L1 respectively) Surface water samples from Buyukcekmece and Omerli (Istanbul, Turkey) and Carmine (Salerno, Italy) (pH 7.65, 7.18, 7.73; alk. 150, 70, 111 mg CaCO3 L1; bromide 274, 95, nd mg/L; chloride 98, 45, 17 mg L1, TOC 3.61, 3.05, 2.05 mg L1 respectively) Synthetic fresh water
Target contaminants
4332
Table 11 e Toxicity/estrogenic activity in drinking water after treatment with AOPs.
Organic micropollutants
UV/O3 and UV/O3/H2O2 were compared with GAC adsorption, amberlite and ion exchange resins
V. fischeri
Model water
bisphenol A
Photolysis with low-pressure Hg UV lamps and UV/H2O2
Two natural water samples (DOC 3.06 and 1.39 mg L1; alkalinity 2.63 and 0.67 mM; pH 8.0 and 7.98; UV254 0.04 and 0.11 cm1) taken from water treatment plant inlets before any treatment and filtered through a 0.45 mm filter. Surface water from Colorado River water, Lake Mead, NV, USA (2.6 mg TOC L1; alkalinity 137 mg L1; pH 8.0; UV254 0.036 cm1).
17-b-estradiol (E2) and 17-aethinyl estradiol (EE2) (10 nMe5 mM).
Direct UV photolysis and UV/ H2O2 advanced oxidation. Two irradiation systems: 1 kW medium pressure UV lamp and four 15-W low-pressure UV lamps.
Estrogenic activity was evaluated by in vitro yeast estrogen screen (YES) and in vivo vitellogenin (VTG) assays with Japanese medaka fish (Oryzias latipes). Estrogenic activity was evaluated by YES assay.
Thirty-two pharmaceuticals and endocrine disrupting compounds (16e1300 ng L1).
A patented photocatalytic reactor membrane pilot system (Phot-Cat), employing UV/TiO2 photocatalysis.
Estrogenic activity was evaluated by YES assay.
Opposite to GAC and resin treatment, AOP processes generally increased toxicity and mutagenic activity (Ames assay). The absence of mutagenic activity was detected only when a GAC step was performed in addition to the AOP process. UV/H2O2 significantly removed BPA and estrogenic activity in vitro and in vivo.
Guzzella et al., 2002
90% removal of estrogenic activity of E2 and EE2 (3 mg L1) was achieved with 5 mg H2O2 L1.
Rosenfeldt et al., 2007
Twenty-nine of the targeted compounds in addition to total estrogenic activity were greater than 70%. No estrogenically active transformation products were formed during treatment.
Benotti et al., 2009
Chen et al., 2006
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 1 1 e4 3 4 0
Surface water from Como lake, Italy (2e10 mg TOC L1).
4333
4334
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 1 1 e4 3 4 0
Table 12 e The role of AOPs in the treatment of actual aqueous matrices and the corresponding role of the bioassays. Aqueous matrix Fresh/drinking water
Urban wastewater
Industrial wastewater
Role of AOPs Metal oxidation To improve coagulation Disinfection Advanced treatment downstream of biological process to remove toxic/emerging contaminants before discharge or reuse Pre-treatment to biological process to improve biodegradability of wastewater Post-treatment to biological process to improve the removal of residual contaminants
system, the destruction of MC-LR took place much faster and toxicity dropped more quickly (a rapid reduction in PP1 inhibition within 5 min of photocatalytic treatment with a complete disappearance in 20 min was observed). Miao et al. (2010) investigated MC-LR and MC-RR degradation and detoxification by ozone. The toxicity of the MCs ozonation by-products was evaluated by assaying the protein phosphatase inhibition in vitro and acute toxicity in vivo. The toxicity evaluation did not show any adverse effects in vivo and in vitro of ozonation end-products, moreover MCs toxicity was completely removed.
5.
Role of bioassays
Conclusive remarks
In conclusion, bioassays, when properly used, showed to be a really useful tool to evaluate the dangerousness of AOPs as well as to set up the proper operative conditions. Table 12 summarizes the role of AOPs in the treatment of a given actual aqueous matrix and the corresponding role of the bioassays. The text organism should be chosen according to the final use of the treated water matrix; for instance, inhibition tests with D. magna may be suitable to evaluate the toxicity of wastewater treatment plant effluent before its disposal; plants bioassay may be suitable to characterize toxicity of wastewater treatment plant effluent before its agricultural reuse. Sometime, acute toxicity tests may not be the most suitable to evaluate the ecotoxicological hazard of micropollutants because of the low concentrations (Rizzo et al., 2005; Baumgarten et al., 2007), although a chronic effect can be expected (Crane et al., 2006). Accordingly studies on chronic effects should be further developed. Finally, some care should be taken in the characterization of the effect of AOPs on the biodegradability of industrial wastewater. Toxicity tests may be not suitable to achieve this aim, so they may be used just as screening test before to use more suitable biodegradability tests (e.g., activated sludge bioassays, respirometry).
Acknowledgment The author wishes to thank the University of Salerno for funding the project entitled “Rimozione di composti xenobiotici
To evaluate the toxicity of oxidation intermediates
Toxicity tests: to evaluate if toxicity decrease after AOPs treatment. Phytotoxicity tests: to evaluate if the effluent is suitable for agricultural reuse As support to biodegradability tests To evaluate the toxicity of oxidation intermediates
e detossificazione di acque reflue urbane destinate al riutilizzo mediante processi di ossidazione avanzata” (FARB, 2009).
references
Al Momani, F., Smith, D.W., El-Din, M.G., 2008. Degradation of cyanobacteria toxin by advanced oxidation processes. Journal of Hazardous Materials 150, 238e249. Amat, A.M., Arques, A., Beneyto, H., Garcı´a, A., Miranda, M.A., Seguı´, S., 2003. Ozonisation coupled with biological degradation for treatment of phenolic pollutants: a mechanistically based study. Chemosphere 53, 79e86. Amat, A.M., Arques, A., Miranda, M.A., Lo´pez, F., 2005. Use of ozone and/or UV in the treatment of effluents from board paper industry. Chemosphere 60, 1111e1117. Amat, A.M., Arques, A., Garcıa-Ripoll, A., Santos-Juanes, L., Vicente, R., Oller, I., Maldonado, M.I., Malato, S., 2009. A reliable monitoring of the biocompatibility of an effluent along an oxidative pre-treatment by sequential bioassays and chemical analyses. Water Research 43, 784e792. Andreozzi, R., Caprio, V., Insola, A., Marotta, R., 1999. Advanced oxidation processes (AOP) for water purification and recovery. Catalysis Today 53, 51e59. Andreozzi, R., Marotta, R., Pinto, G., Pollio, A., 2002. Carbamazepine in water: persistence in the environment, ozonation treatment and preliminary assessment on algal toxicity. Water Research 36, 2869e2877. Andreozzi, R., Caprio, V., Marotta, R., Vogna, D., 2003. Paracetamol oxidation from aqueous solutions by means of ozonation and H2O2/UV system. Water Research 37, 993e1004. Andreozzi, R., Campanella, L., Fraysse, B., Garric, J., Gonnella, A., Lo Giudice, R., Maritta, R., Pinto, G., Pollio, A., 2004. Effects of advanced oxidation processes (AOPs) on the toxicity of a mixture of pharmaceuticals. Water Science and Technology 50 (5), 23e28. Andreozzi, R., Canterino, M., Lo Giudice, R., Marotta, R., Pinto, G., Pollio, A., 2006. Lincomycin solar photodegradation, algal toxicity and removal from wastewaters by means of ozonation. Water Research 40, 630e638. Andreozzi, R., Canterino, M., Di Somma, I., Lo Giudice, R., Marotta, R., Pinto, G., Pollio, A., 2008. Effect of combined physico-chemical processes on the phytotoxicity of olive mill wastewaters. Water Research 42, 1684e1692. APHA/AWWA/WEF, 1998. Standard Methods for the Examination of Water and Wastewater, 20th ed. APHA/AWWA/WEF, Washington, DC, USA. Arslan Alaton, I., Teksoy, S., 2007. Acid dyebath effluent pretreatment using Fenton’s reagent: process optimization,
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 1 1 e4 3 4 0
reaction kinetics and effects on acute toxicity. Dyes and Pigments 73, 31e39. Auriol, M., Filali-Meknassi, Y., Tyagi, R.D., Adams, C.D., Surampalli, R.Y., 2006. Endocrine disrupting compounds removal from wastewater, a new challenge. Process Biochemistry 41, 525e539. ¨ tker, M., 2003. Treatment of pharmaceutical Balcıoglu, I.A., O wastewater containing antibiotics by O3 and O3/H2O2 processes. Chemosphere 50, 85e95. Ballesteros Martı´n, M.M., Sa´nchez Pe´rez, J.A., Garcı´a Sa´nchez, J.L., Casas Lo´pez, J.L., Malato Rodrı´guez, S., 2009. Effect of pesticide concentration on the degradation process by combined solar photo-Fenton and biological treatment. Water Research 43, 3838e3848. Baumgarten, S., Schro¨der, H.Fr., Charwath, C., Lange, M., Beier, S., Pinnekamp, J., 2007. Evaluation of advanced treatment technologies for the elimination of pharmaceutical compounds. Water Science and Technology 56 (5), 1e8. Beccari, M., Carucci, G., Lanz, A.M., Magone, M., Petrangeli Papini, M., 2002. Removal of molecular weight fractions of COD and phenolic compounds in an integrated treatment of olive oil mill effluents. Biodegradation 13, 401e410. Bekbolet, M., Uyguner, C.S., Selcuk, H., Rizzo, L., Nikolaou, A.D., Meric¸, S., Belgiorno, V., 2005. Application of oxidative removal of NOM to drinking water and formation of disinfection byproducts. Desalination 176, 155e166. Belfroid, A., van Velzen, M., van der Horst, B., Vethaak, D., 2002. Occurrence of bisphenol A in surface water and uptake in fish: evaluation of field measurements. Chemosphere 49, 97e103. ´ lvar, P.M., 2001. PH sequential Beltra´n, F.J., Garcı´a-Araya, J.F., A ozonation of domestic and wine-distillery wastewaters. Water Research 35, 929e936. Beltra´n, F.J., Aguinaco, A., Garcı´a-Araya, J.F., Oropesa, A., 2008. Ozone and photocatalytic processes to remove the antibiotic sulfamethoxazole from water. Water Research 42, 3799e3808. Beltra´n-Heredia, J., Torregrosa, J., Garcı´a, J., Dominguez, J.R., Tierno, J.C., 2001. Degradation of olive mill wastewater by the combination of Fenton’s reagent and ozonation processes with an aerobic biological treatment. Water Science and Technology 44 (5), 103e108. Benotti, M.J., Stanford, B.D., Wert, E.C., Snyder, S.A., 2009. Evaluation of a photocatalytic reactor membrane pilot system for the removal of pharmaceuticals and endocrine disrupting compounds from water. Water Research 43, 1513e1522. Bizani, E., Fytianos, K., Poulios, I., Tsiridis, V., 2006. Photocatalytic decolorization and degradation of dye solutions and wastewaters in the presence of titanium dioxide. Journal of Hazardous Materials 136, 85e94. Borja, R., Banks, C.J., Maestro-Duran, R., Alba, J., 1996. The effects of the most important phenolic constituents of olive mill wastewater on batch anaerobic methanogenesis. Environmental Technology 17, 167e174. Brenes, M., Garcia, A., Garcia, P., Rios, J.J., Garrido, A., 1999. Phenolic compounds in Spanish olive oils. Journal of Agricultural and Food Chemistry 47, 3535e3540. Brose´us, R., Vincent, S., Aboulfadl, K., Daneshvar, A., Sauve´, S., Barbeau, B., Pre´vost, M., 2009. Ozone oxidation of pharmaceuticals, endocrine disruptors and pesticides during drinking water treatment. Water Research 43, 4707e4717. Burgess, J.E., Quarmby, J., Stephenson, T., 1999. Micronutrient supplements to enhance biological wastewater treatment of phosphorus-limited industrial effluent. Process Safety and Environmental Protection 77, 199e204. Cabrera, F., Lopez, R., Martinez-Bordiu`, A., de Lome, E.D., Murillo, J. M., 1996. Land treatment of olive oil mill wastewater. International Biodeterioration and Biodegradation 38, 215e225. Calza, P., Sakkas, V.A., Medana, C., Baiocchi, C., Dimou, A., Pelizzetti, E., Albanis, T., 2006. Photocatalytic degradation
4335
study of diclofenac over aqueous TiO2 suspensions. Applied Catalysis B: Environmental 67, 197e205. Campos, J.C., Borges, R.M.H., Oliveira Filho, A.M., Nobrega, R., Sant’Anna Jr., G.L., 2002. Oilfield wastewater treatment by combined microfiltration and biological processes. Water Research 36, 95e104. Castiglioni, S., Bagnati, R., Fanelli, R., Pomati, F., Calamari, D., Zuccato, E., 2006. Removal of pharmaceuticals in sewage treatment plants in Italy. Environmental Science and Technology 40, 357e363. Catalkaya, E.C., Kargi, F., 2008. Advanced oxidation treatment of pulp mill effluent for TOC and toxicity removals. Journal of Environmental Management 87, 396e404. Chang, Shih-Hsien, Chuang, Shun-Hsing, Li, Heng-Ching, Liang, Hsiu-Hao, Huang, Lung-Chiu, 2009. Comparative study on the degradation of I.C. Remazol Brilliant Blue R and I.C. Acid Black 1 by Fenton oxidation and Fe0/air process and toxicity evaluation. Journal of Hazardous Materials 166, 1279e1288. Chatzisymeon, E., Xekoukoulotakis, N.P., Diamadopoulos, E., Katsaounis, A., Mantzavinos, D., 2009. Boron-doped diamond anodic treatment of olive mill wastewaters: statistical analysis, kinetic modeling and biodegradability. Water Research 43, 3999e4009. Chen, P.-J., Linden, K.G., Hinton, D.E., Kashiwada, S., Rosenfeldt, E.J., Kullman, S.W., 2006. Biological assessment of bisphenol A degradation in water following direct photolysis and UV advanced oxidation. Chemosphere 65, 1094e1102. Chen, S., Sun, D., Chung, J.-S., 2007. Treatment of pesticide wastewater by moving-bed biofilm reactor combined with Fenton-coagulation pretreatment. Journal of Hazardous Materials 144, 577e584. Coca, M., Pen˜a, M., Gonza´lez, G., 2005. Variables affecting efficiency of molasses fermentation wastewater ozonation. Chemosphere 60, 1408e1415. Coelho, A.D., Sans, C., Agu¨era, A., Go´mez, M.J., Esplugas, S., Dezotti, M., 2009. Effects of ozone pre-treatment on diclofenac: intermediates, biodegradability and toxicity assessment. Science of the Total Environment 407, 3572e3578. Comninellis, C., Kapalka, A., Malato, S., Parsons, S.A., Poulios, I., Mantzavinos, D., 2008. Advanced oxidation processes for water treatment: advances and trends for R&D. Journal of Chemical Technology and Biotechnology 83, 769e776. Crane, M., Watts, C., Boucard, T., 2006. Chronic aquatic environmental risks from exposure to human pharmaceuticals. Science of the Total Environment 367, 23e41. Dalrymple, O.K., Yeh, D.H., Trotz, M.A., 2007. Removing pharmaceuticals and endocrine disrupting compounds from wastewater by photocatalysis. Journal of Chemical Technology and Biotechnology 82, 121e134. Dalzell, D.J.B., Alte, S., Aspichueta, E., de la Sota, A., Etxebarria, J., Gutierrez, M., Hoffmann, C.C., Sales, D., Obst, U., Christofi, N., 2002. A comparison of five rapid direct toxicity assessment methods to determine toxicity of pollutants to activated sludge. Chemosphere 47, 535e545. Dantas, R.F., Canterino, M., Marotta, R., Sans, C., Esplugas, S., Andreozzi, R., 2007. Bezafibrate removal by means of ozonation: primary intermediates, kinetics, and toxicity assessment. Water Research 41, 2525e2532. Date, S., Terabayashi, S., Kobayashi, Y., Fujime, Y., 2005. Effects of chloramines concentration in nutrient solution and exposure time on plant growth in hydroponically cultured lettuce. Scientia Horticulturae 103, 257e265. De Nicola, E., Gallo, M., Iaccarino, M., Meric, S., Oral, R., Russo, T., Sorrentino, T., Tunay, O., Vuttariello, E., Warnau, M., Pagano, G., 2004. Hormetic vs. toxic effects of vegetable tannin in a multi-test study. Archives of Environmental Contamination and Toxicology 46, 336e344.
4336
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 1 1 e4 3 4 0
Directive 2000/60/EC of the Council and the European Parliament of 23 October 2000. OJ L 237, 22.12.2000. Di Salvatore, M., Carafa, A.M., Carratu`, G., 2008. Assessment of heavy metals phytotoxicity using seed germination and root elongation tests: a comparison of two growth substrates. Chemosphere 73, 1461e1464. Drzewicz, P., Na1ezcz-Jawecki, G., Gryz, M., Sawicki, J., Bojanowska-Czajka, A., G1uszewski, W., Kulisa, K., Wo1kowicz, S., Marek, T., 2004. Monitoring of toxicity during degradation of selected pesticides using ionizing radiation. Chemosphere 57, 135e145. Emmanuel, E., Keck, G., Blanchard, J.-M., Vermande, P., Perrodin, Y., 2004. Toxicological effects of disinfections using sodium hypochlorite on aquatic organisms and its contribution to AOX formation in hospital wastewater. Environment International 30, 891e900. Esplugas, S., Bila, D.M., Krause, L.G.T., Dezotti, M., 2007. Ozonation and advanced oxidation technologies to remove endocrine disrupting chemicals (EDCs) and pharmaceuticals and personal care products (PPCPs) in water effluents. Journal of Hazardous Materials 149, 631e642. Evgenidou, E., Fytianos, K., Poulios, I., 2005a. Semiconductorsensitized photodegradation of dichlorvos in water using TiO2 and ZnO as catalysts. Applied Catalysis B: Environmental 59, 81e89. Evgenidou, E., Fytianos, K., Poulios, I., 2005b. Photocatalytic oxidation of dimethoate in aqueous solutions. Journal of Photochemistry and Photobiology, A 175, 29e38. Evgenidou, E., Bizani, E., Christophoridis, C., Fytianos, K., 2007a. Heterogeneous photocatalytic degradation of prometryn in aqueous solutions under UVeVis irradiation. Chemosphere 68, 1877e1882. Evgenidou, E., Konstantinou, I., Fytianos, K., Poulios, I., 2007b. Oxidation of two organophosphorous insecticides by the photo-assisted Fenton reaction. Water Research 41, 2015e2027. Farre´, M., Barcelo´, D., 2003. Toxicity testing of wastewater and sewage sludge by biosensors, bioassays and chemical analysis. Trends Analytical Chemistry 22, 299e310. Farre´, M.J., Franch, M.I., Malato, S., Ayllo´n, J.A., Peral, J., Dome´nech, X., 2005. Degradation of some biorecalcitrant pesticides by homogeneous and heterogeneous photocatalytic ozonation. Chemosphere 58, 1127e1133. Farre´, M.J., Franch, M.I., Ayllo´n, J.A., Peral, J., Dome`nech, X., 2007. Biodegradability of treated aqueous solutions of biorecalcitrant pesticides by means of photocatalytic ozonation. Desalination 211, 22e33. Fernandez-Alba, A.R., Hernando, D., Aguera, A., Caceres, J., Malato, S., 2002. Toxicity assays: a way for evaluating AOPs efficiency. Water Research 36, 4255e4262. Ferraris, M., Chiesara, E., Radice, S., Giovara, A., Frigerio, S., Fumagalli, R., Marabini, L., 2005. Study of potential toxic effects on rainbow trout hepatocytes of surface water treated with chlorine or alternative disinfectants. Chemosphere 60, 65e73. Gagne´, F., Blaise, C., Andre, C., 2006. Occurrence of pharmaceutical products in a municipal effluent and toxicity to rainbow trout (Oncorhynchus mykiss) hepatocytes. Ecotoxicology and Environmental Safety 64, 329e336. Garcia-Montano, J., Torrades, F., Garcia-Hortal, J.A., Dome`nech, X. , Peral, J., 2006. Degradation of Procion Red H-E7B reactive dye by coupling a photo-Fenton system with a sequencing batch reactor. Journal of Hazardous Materials B134, 220e229. Gernjak, W., Maldonado, M.I., Malato, S., Caceres, J., Krutzler, T., Glaser, A., Bauer, R., 2004. Pilot-plant treatment of olive mill wastewater (OMW) by solar TiO2 photocatalysis and solar photo-Fenton. Solar Energy 77, 567e572. Gibs, J., Stackelberg, P.E., Furlong, E.T., Meyer, M., Zaugg, S.T., Lippincott, R.L., 2007. Persistence of pharmaceuticals and other
organic compounds in chlorinated drinking water as a function of time. Science of the Total Environment 373, 240e249. Gogate, P.R., 2002. Cavitation: an auxiliary technique in wastewater treatment schemes. Advances in Environmental Research 6, 335e358. Gogate, P.R., Pandit, A.B., 2004a. A review of imperative technologies for wastewater treatment I: oxidation technologies at ambient conditions. Advances in Environmental Research 8, 501e551. Gogate, P.R., Pandit, A.B., 2004b. A review of imperative technologies for wastewater treatment II: hybrid methods. Advances in Environmental Research 8, 553e597. Go´mez, M.J., Sirtori, C., Mezcua, M., Ferna´ndez-Alba, A.R., Agu¨era, A., 2008. Photodegradation study of three dipyrone metabolites in various water systems: identification and toxicity of their photodegradation products. Water Research 42, 2698e2706. Gonza´lez, O., Esplugas, M., Sans, C., Torres, A., Esplugas, S., 2009. Performance of a sequencing batch biofilm reactor for the treatment of pre-oxidized sulfamethoxazole solutions. Water Research 43, 2149e2158. Gonze, E., Fourel, L., Gonthier, Y., Boldo, P., Bernis, A., 1999. Wastewater pretreatment with ultrasonic irradiation to reduce toxicity. Chemical Engineering Journal 73, 93e100. Gupta, N., Pant, S.C., Vijayaraghavan, R., Rao, P.V.L., 2003. Comparative toxicity evaluation of cyanobacterial cyclic peptide toxin microcystin variants (LR, RR, YR) in mice. Toxicology 188, 285e296. Gutie´rrez, M., Etxebarria, J., de las Fuentes, L., 2002. Evaluation of wastewater toxicity: comparative study between Microtox and activated sludge oxygen uptake inhibition. Water Research 36, 919e924. Guzzella, L., Feretti, D., Monarca, S., 2002. Advanced oxidation and adsorption technologies for organic micropollutant removal from lake water used as drinking-water supply. Water Research 36, 4307e4318. Hammes, F., Salhi, E., Ko¨ster, O., Kaiser, H.-P., Egli, T., von Gunten, U., 2006. Mechanistic and kinetic evaluation of organic disinfection by-product and assimilable organic carbon (AOC) formation during the ozonation of drinking water. Water Research 40, 2275e2286. Hernando, M.D., Ferna´ndez-Alba, A.R., Tauler, R., Barcelo´, D., 2005. Toxicity assays applied to wastewater treatment. Talanta 65, 358e366. Herrmann, J.M., Guillard, C., Arguello, M., Agu¨era, A., Tejedor, A., Piedra, L., Ferna´ndez-Alba, A., 1999. Photocatalytic degradation of pesticide pirimiphos-methyl: determination of the reaction pathway and identification of intermediate products by various analytical methods. Catalysis Today 54, 353e367. ISO (International Organization for Standardization), 1989. Water QualitydFresh Water Algal Growth Inhibition Test with Scenedesmus subspicatus and Selenastrum capricornutum. ISO 8692, Geneva, Switzerland. ISO (International Organisation for Standardization), 1996a. Water Quality: Determination of the Inhibition of the Mobility of Daphnia magna Straus (Cladocera, Crustacea) e Acute Toxicity Test. ISO 6341, Geneva, Switzerland. ISO (International Organization for Standardization), 1996b. Water QualitydDetermination of the Acute Lethal Toxicity of Substances to a Freshwater Fish (Brachydanio rerio Hamilton Buchanan, Teleostei, Cyprinidae). Part 1: Static Method. ISO 7346-1, Geneva, Switzerland. ISO (International Organization for Standardization), 1998. Water QualitydDetermination of the Inhibitory Effect of Water Samples on the Light Emission of Vibrio fischeri (Luminescent Bacteria Test)dPart 2: Method Using Liquid-Dried Bacteria. ISO 11348-2, Geneva, Switzerland.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 1 1 e4 3 4 0
ISO (International Organization for Standardization), 2007. Water Quality-Test for Inhibition of Oxygen Consumption by Activated Sludge for Carbonaceous and Ammonium Oxidation. ISO 8192, Geneva, Switzerland. Jiao, S., Zheng, S., Yin, D., Wang, L., Chen, L., 2008. Aqueous photolysis of tetracycline and toxicity of photolytic products to luminescent bacteria. Chemosphere 73, 377e382. Jin, X., Jiang, G., Huang, G., Liu, J., Zhou, Q., 2004. Determination of 4-tert-octylphenol, 4-nonylphenol and bisphenol A in surface waters from the Haihe River in Tianjin by gas chromatographyemass spectrometry with selected ion monitoring. Chemosphere 56, 1113e1119. Jochimsen, J.C., Schenk, H., Jekel, M.R., Hegemann, W., 1997. Combined oxidative and biological treatment for separated streams of tannery wastewater. Water Science Technology 36, 209e216. Joubert, G., 1980. A bioassay application for quantitative toxicity measurements, using the green algae Selenastrum capricornutum. Water Research 14, 1759e1763. Jurczak, T., Tarczynska, M., Izydorczyk, K., Mankiewicz, J., Zalewski, M., Meriluoto, J., 2005. Elimination of microcystins by water treatment processesdexamples from Sulejow Reservoir, Poland. Water Research 39, 2394e2406. Justino, C.I., Duarte, K., Loureiro, F., Pereira, R., Antunes, S.C., Marques, S.M., Gonc¸alves, F., Rocha-Santos, T.A.P., Freitas, A.C., 2009. Toxicity and organic content characterization of olive oil mill wastewater undergoing a sequential treatment with fungi and photo-Fenton oxidation. Journal of Hazardous Materials 172, 1560e1572. Kang, J.-H., Kondo, F., Katayama, Y., 2006. Human exposure to bisphenol A. Toxicology 226, 79e89. Khoufi, S., Aloui, F., Sayadi, S., 2009. Pilot scale hybrid process for olive mill wastewater treatment and reuse. Chemical Engineering and Processing 48, 643e650. Kidak, R., Ince, N.H., 2007. Catalysis of advanced oxidation reactions by ultrasound: a case study with phenol. Journal of Hazardous Materials 146, 630e635. Kim, J.-K., Choi, K., Cho, I.-H., Son, H.-S., Zoh, K.-D., 2007. Application of a microbial toxicity assay for monitoring treatment effectiveness of pentachlorophenol in water using UV photolysis and TiO2 photocatalysis. Journal of Hazardous Materials 148, 281e286. Kim, T.-S., Kim, J.-K., Choi, K., Stenstrom, M.K., Zoh, K.-D., 2006. Degradation mechanism and the toxicity assessment in TiO2 photocatalysis and photolysis of parathion. Chemosphere 62, 926e933. Klamerth, N., Rizzo, L., Malato, S., Maldonado, M.I., Agu¨era, A., Ferna´ndez-Alba, A.R., 2010. Degradation of fifteen emerging contaminants at mg L1 initial concentrations by mild solar photo-Fenton in MWTP effluents. Water Research 44, 545e554. Klavarioti, M., Mantzavinos, D., Kassinos, D., 2009. Removal of residual pharmaceuticals from aqueous systems by advanced oxidation processes. Environment International 35, 402e417. Konstantinou, I.K., Albanis, T.A., 2003. Photocatalytic transformation of pesticides in aqueous titanium dioxide suspensions using artificial and solar light: intermediates and degradation pathways. Applied Catalysis B: Environmental 42, 319e335. Konstantinou, I.K., Albanis, T.A., 2004. TiO2-assisted photocatalytic degradation of azo dyes in aqueous solution: kinetic and mechanistic investigations: a review. Applied Catalysis B: Environmental 49, 1e14. Kralj, M.B., Cernigoj, U., Franko, M., Trebse, P., 2007. Comparison of photocatalysis and photolysis of malathion, isomalathion, malaoxon, and commercial malathiondproducts and toxicity studies. Water Research 41, 4504e4514. Kuch, H.M., Ballschmiter, K., 2001. Determination of endocrine disrupting phenolic compounds and estrogens in surface and
4337
drinking water by HRGCe(NCI)eMS in the pictogram per liter range. Environmental Science and Technology 35, 3201e3206. Kusic, H., Koprivanac, N., Horvat, S., Bakija, S., Loncaric Bozic, A., 2009. Modeling dye degradation kinetic using dark- and photo-Fenton type processes. Chemical Engineering Journal 155, 144e154. Kusvuran, E., Gulnaz, O., Irmak, S., Atanur, O.M., Yavuz, H.I., Erbatur, O., 2004. Comparison of several advanced oxidation processes for the decolorization of Reactive Red 120 azo dye in aqueous solution. Journal of Hazardous Materials B109, 85e93. Kusvuran, E., Irmak, S., Yavuz, H.I., Samil, A., Erbatur, O., 2005. Comparison of the treatment methods efficiency for decolorization and mineralization of Reactive Black 5 azo dye. Journal of Hazardous Materials B119, 109e116. Lapertot, M., Ebrahimi, S., Dazio, S., Rubinelli, A., Pulgarin, C., 2007. Photo-Fenton and biological integrated process for degradation of a mixture of pesticides. Journal of Photochemistry and Photobiology, A 186, 34e40. Lautenschlager, K., Boon, N., Wang, Y., Egli, T., Hammes, F., 2010. Overnight stagnation of drinking water in household taps induces microbial growth and changes in community composition. Water Research 44, 4868e4877. Lawton, L.A., Robertson, P.K.J., Cornish, B.J.P.A., Jaspars, M., 1999. Detoxification of microcystins (Cyanobacterial Hepatotoxins) using TiO2 photocatalytic oxidation. Environmental Science and Technology 33, 771e775. Lawton, L.A., Robertson, P.K.J., Cornish, B.J.P.A., Marr, I.L., Jaspars, M., 2003. Processes influencing surface interaction and photocatalytic destruction of microcystins on titanium dioxide photocatalysts. Journal of Catalysis 213, 109e113. LeChevallier, M.W., Schultz, W., Lee, R.G., 1991. Bacterial nutrients in drinking water. Applied Environmental Microbiology 57 (3), 857e862. Ledakowicz, S., Gonera, M., 1999. Optimisation of oxidants dose for combined chemical and biological treatment of textile wastewater. Water Research 33, 2511e2516. Ledakowicz, S., Solecka, M., Zylla, R., 2001. Biodegradation, decolourisation and detoxification of textile wastewater enhanced by advanced oxidation processes. Journal of Biotechnology 89, 175e184. Ledakowicz, S., Michniewicz, M., Jagiella, A., Stufka-Olczyk, J., Martynelis, M., 2006. Elimination of resin acids by advanced oxidation processes and their impact on subsequent biodegradation. Water Research 40, 3439e3446. Levec, J., Pintar, A., 2007. Catalytic wet-air oxidation processes: a review. Catalysis Today 124, 172e184. Lin, D., Xing, B., 2007. Phytotoxicity of nanoparticles: inhibition of seed germination and root growth. Environmental Pollution 150, 243e250. Liu, I., Lawton, L.A., Cornish, B., Robertson, P.K.J., 2002. Mechanistic and toxicity studies of the photocatalytic oxidation of microcystin-LR. Journal of Photochemistry and Photobiology A 148, 349e354. Malato, S., Blanco, J., Alarco´n, D.C., Maldonado, M.I., Ferna´ndezIba´n˜ez, P., Gernjak, W., 2007. Photocatalytic decontamination and disinfection of water with solar collectors. Catalysis Today 122, 137e149. Mantzavinos, D., Kalogerakis, N., 2005. Treatment of mill effluents Part I. organic matter degradation by chemical and biological processes e an overview. Environment International 31, 289e295. Mantzavinos, D., Psillakis, E., 2004. Enhancement of biodegradability of industrial wastewaters by chemical oxidation pre-treatment. Journal of Chemical Technology and Biotechnology 79, 431e454. Marttinen, K., Kettunen, R.H., Sormunen, K.M., Soimasuo, R.M., Rintala, J.A., 2002. Screening of physicalechemical methods for removal of organic material, nitrogen and toxicity from low strength landfill leachates. Chemosphere 46 (6), 851e858.
4338
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 1 1 e4 3 4 0
Matilainen, A., Sillanpa¨a¨, M., 2010. Removal of natural organic matter from drinking water by advanced oxidation processes. Chemosphere 80, 351e365. Me´ndez-Arriaga, F., Esplugas, S., Gime´nez, J., 2008. Photocatalytic degradation of non-steroidal anti-inflammatory drugs with TiO2 and simulated solar irradiation. Water Research 42, 585e594. Meric¸, S., De Nicola, E., Iaccarino, M., Gallo, M., Di Gennaro, A., Morrone, G., Warnau, M., Belgiorno, V., Pagano, G., 2005. Toxicity of leather tanning wastewater effluents in sea urchin early development and in marine microalgae. Chemosphere 61, 208e217. lu, K., 2010. Pre-treatment Mert, B.K., Yonar, T., Kilic¸, M.Y., Kestiog studies on olive oil mill effluent using physicochemical, Fenton and Fenton like oxidations processes. Journal of Hazardous Materials 174, 122e128. Miao, H.-F., Qin, F., Tao, G.-J., Tao, W.-Y., Ruan, W.-Q., 2010. Detoxification and degradation of microcystin-LR and -RR by ozonation. Chemosphere 79, 355e361. Migliore, L., Civitareale, C., Brambilla, G., Di Pelupis, G., 1997. Toxicity of several important agricultural antibiotics to Artemia. Water Research 31 (7), 1801e1806. Mishra, V.S., Mahajani, V.V., Joshi, J.B., 1995. Wet air oxidation. Industrial and Engineering Chemistry Research 34, 2e48. Muthukumar, M., Sargunamani, D., Selvakumar, N., Venkata Rao, J., 2004. Optimisation of ozone treatment for colour and COD removal of acid dye effluent using central composite design experiment. Dyes and Pigments 63, 127e134. Naddeo, V., Meric¸, S., Kassinos, D., Belgiorno, V., Guida, M., 2009. Fate of pharmaceuticals in contaminated urban wastewater effluent under ultrasonic irradiation. Water Research 43, 4019e4027. Najjar, W., Azabou, S., Sayadi, S., Ghorbel, A., 2009. Screening of FeeBEA catalysts for wet hydrogen peroxide oxidation of crude olive mill wastewater under mild conditions. Applied Catalysis B: Environmental 88, 299e304. Neamtu, M., Yediler, A., Siminiceanu, I., Macoveanu, M., Kettrup, A., 2004. Decolorization of disperse red 354 azo dye in water by several oxidation processesda comparative study. Dyes and Pigments 60, 61e68. Neumann, M., Schulz, R., Schafer, K., Muller, W., Mannheller, W., Liess, M., 2002. The significance of entry routes as point and non-point sources of pesticides in small streams. Water Research 36, 835e842. Nikolaou, A., Rizzo, L., Selcuk, H., 2007. Control of Disinfection ByProducts in Drinking Water Systems. Nova Science Publishers, Inc., 400 Oser Avenue, Suite 1600, Hauppauge, NY 11788, ISBN 88-7850-003-8. OECD (Organization for Economic Cooperation and Development), 1984. Terrestrial plants, growth test. In: Guidelines of the OECD for Testing Chemical Products. OECD Method 208, Paris. OECD (Organisation for Economic Cooperation and Development), 1992. Fish Early Life Stage Toxicity Test. OECD Guidelines for the Testing of Chemicals. OECD No. 210, Paris. OECD (Organization for Economic Cooperation and Development), 2010. Guidelines for the Testing of Chemicals: Activated Sludge, Respiration Inhibition Test (Carbon and Ammonium Oxidation). OECD No. 209, Paris. Oller, I., Malato, S., Sanchez-Perez, J.A., Maldonado, M.I., Gasso, R. , 2007. Detoxification of wastewater containing five common pesticides by solar AOPs e biological coupled system. Catalysis Today 129, 69e78. Oral, R., Meric¸, S., De Nicola, E., Petruzzelli, D., Della Rocca, C., Pagano, G., 2007. Multi-species toxicity evaluation of a chromium-based leather tannery wastewater. Desalination 211, 48e57. Pagano, G., Esposito, A., Giordano, G.G., 1982. Fertilization and larval development in sea urchins following exposure of
gametes and embryos to cadmium. Archives of Environmental Contamination and Toxicology 11, 47e55. Pagano, G., Iaccarino, M., De Biase, A., Meric, S., Trieff, N.M., 2001. Toxicity of the R6 fungicide mixture as related to the effects of its components, cymoxanil and copper oxychloride on sea urchin fertilization and development. Human and Experimental Toxicology 20, 404e411. Pagga, U., Bachner, J., Strotmann, U., 2006. Inhibition of nitrification in laboratory tests and model wastewater treatment plants. Chemosphere 65, 1e8. Pala´cio, S.M., Espinoza-Quin˜ones, F.R., Mo´denes, A.N., Oliveira, C.C. , Borba, F.H., Silva Jr., F.G., 2009. Toxicity assessment from electro-coagulation treated-textile dye wastewaters by bioassays. Journal of Hazardous Materials 172, 330e337. Pehlivanoglu, E., Sedlak, D.L., 2004. Bioavailability of wastewaterderived organic nitrogen to the alga Selenastrum capricornutum. Water Research 38, 3189e3196. Peng, G., Roberts, J.C., 2000. Solubility and toxicity of resin acids. Water Research 34, 2779e2785. Pe´rez-Estrada, Leonidas A., Malato, Sixto, Aguera, Ana, Fernandez-Alba, Amadeo R., 2007. Degradation of dipyrone and its main intermediates by solar AOPs e identification of intermediate products and toxicity assessment. Catalysis Today 129, 207e214. Pe´rez, S., Aga, D.S., 2005. Recent advances in the sample preparation, liquid chromatography tandem mass spectrometric analysis and environmental fate of microcystins in water. Trends in Analytical Chemistry 24, 658e670. Persoone, G., Van Haecke, P., 1981. Intercalibration exercise on a short-term standard toxicity test with Artemia naupli. Final Report. Contract CEE-ENV-396 B(N). Petala, M., SamarasZouboulis, P.A., Kungolos, A., Sakellaropoulos, G.P., 2008. Influence of ozonation on the in vitro mutagenic and toxic potential of secondary effluents. Water Research 42, 4929e4940. Pignatello, J.J., Oliveros, E., MacKay, A., 2006. Advanced oxidation processes for organic contaminant destruction based on the Fenton reaction and related chemistry. Critical Review in Environmental Science and Technology 36, 1e84. Polanska, M., Huysman, K., van Keer, C., 2005. Investigation of assimilable organic carbon (AOC) in flemish drinking water. Water Research 39, 2259e2266. Radix, P., Le´onard, M., Papantoniou, C., Roman, G., Saouter, E., Gallotti-Schmitt, S., Thie´baud, H., Vasseur, P., 2000. Comparison of four chronic toxicity tests using algae, bacteria, and invertebrates assessed with sixteen chemicals. Ecotoxicology and Environmental Safety 47, 186e194. , J., Petrovic , M., Barcelo´, D., 2009. Fate and distribution Radjenovic of pharmaceuticals in wastewater and sewage sludge of the conventional activated sludge (CAS) and advanced membrane bioreactor (MBR) treatment. Water Research 43, 831e841. Rajeshwar, K., Osugi, M.E., Chanmanee, W., Chenthamarakshan, C.R., Zanoni, M.V.B., Kajitvichyanukul, P., Krishnan-Ayer, R., 2008. Heterogeneous photocatalytic treatment of organic dyes in air and aqueous media. Journal of Photochemistry and Photobiology C: Photochemistry Reviews 9, 171e192. Renoux, A.Y., Tyagi, R.D., Samson, R., 2001. Assessment of toxicity reduction after metal removal in bioleached sewage sludge. Water Research 35, 1415e1424. Reungoat, J., Macova, M., Escher, B.I., Carswell, S., Mueller, J.F., Keller, J., 2010. Removal of micropollutants and reduction of biological activity in a full scale reclamation plant using ozonation and activated carbon filtration. Water Research 44, 625e637. Richardson, S.D., 1998. Encyclopedia of Environmental Analysis and Remediation. In: Meyers, R.A. (Ed.), Drinking water
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 1 1 e4 3 4 0
disinfection by-products, vol. 3. John Wiley & Sons, New York, pp. 1398e1421. Rizzo, L., Belgiorno, V., Gallo, M., Meric¸, S., 2005. Removal of THM precursors from a high-alkaline surface water by enhanced coagulation and behaviour of THMFP toxicity on D. magna. Desalination 176, 177e188. Rizzo, L., Lofrano, G., Grassi, M., Belgiorno, V., 2008. Pre-treatment of olive mill wastewater by chitosan coagulation and advanced oxidation processes. Separation and Purification Technology 63, 648e653. Rizzo, L., Meric, S., Guida, M., Kassinos, D., Belgiorno, V., 2009a. Heterogenous photocatalytic degradation kinetics and detoxification of an urban wastewater treatment plant effluent contaminated with pharmaceuticals. Water Research 43, 4070e4078. Rizzo, L., Meric, S., Kassinos, D., Guida, M., Russo, F., Belgiorno, V., 2009b. Degradation of diclofenac by TiO2 photocatalysis: UV absorbance kinetics and process evaluation through a set of toxicity bioassays. Water Research 43, 979e988. Rivas, J., Gimeno, O., Beltra´n, F., 2009. Wastewater recycling: application of ozone based treatments to secondary effluents. Chemosphere 74, 854e859. Robidoux, P.Y., Gong, P., Sarrazin, M., Bardai, G., Paquet, L., Hawari, J., Dubois, C., Sunahara, G.I., 2004. Toxicity assessment of contaminated soils from an antitank firing range. Ecotoxicology and Environmental Safety 58, 300e313. Rodrigues, A.C., Boroski, M., Shimada, N.S., Garcia, J.C., Nozaki, J., Hioka, N., 2008. Treatment of paper pulp and paper mill wastewater by coagulationeflocculation followed by heterogeneous photocatalysis. Journal of Photochemistry and Photobiology, A 194, 1e10. Rosal, R., Rodrı´guez, A., Perdigo´n-Melo´n, J.A., Mezcua, M., Hernando, M.D., Leto´n, P., Garcı´a-Calvo, E., Agu¨era, A., Ferna´ndez-Alba, A.R., 2008. Removal of pharmaceuticals and kinetics of mineralization by O3/H2O2 in a biotreated municipal wastewater. Water Research 42, 3719e3728. Rosenfeldt, E.J., Chen, P.J., Kullman, S., Linden, K.G., 2007. Destruction of estrogenic activity in water using UV advanced oxidation. Science of the Total Environment 377, 105e113. Ruebhart, D.R., Cock, I.E., Shaw, G.R., 2008. Brine shrimp bioassay: importance of correct taxonomic identification of Artemia (Anostraca) species. Environmental Toxicology 23, 555e560. Sacher, F., Lange, F.T., Brauch, H.-J., Blankenhorn, I., 2001. Pharmaceuticals in groundwaters: analytical methods and results of a monitoring program in Baden-Wu¨rttemberg, Germany. Journal of Chromatography A 938, 199e210. Sakkas, V.A., Calza, P., Medana, C., Villioti, A.E., Baiocchi, C., Pelizzetti, E., Albanis, T., 2007. Heterogeneous photocatalytic degradation of the pharmaceutical agent salbutamol in aqueous titanium dioxide suspensions. Applied Catalysis B: Environmental 77, 135e144. Sanches, S., Barreto Crespo, M.T., Pereira, V.J., 2010. Drinking water treatment of priority pesticides using low pressure UV photolysis and advanced oxidation processes. Water Research 44, 1809e1818. Sa´nchez-Polo, M., Lo´pez-Pen˜alver, J., Prados-Joya, G., FerroGarcı´a, M.A., Rivera-Utrilla, J., 2009. Gamma irradiation of pharmaceutical compounds, nitroimidazoles, as a new alternative for water treatment. Water Research 43, 4028e4036. Sarria, V., Parra, S., Adler, N., Pe´ringer, P., Benitez, N., Pulgarin, C., 2002. Recent developments in the coupling of photoassisted and aerobic biological processes for the treatment of biorecalcitrant compounds. Catalysis Today 76, 301e315. Sauer, T.K., Casaril, L., Oberziner, A.L.B., Jose´, H.J., Muniz Moreira, R.P., 2006. Advanced oxidation processes applied to tannery wastewater containing Direct Black 38delimination
4339
and degradation kinetics. Journal of Hazardous Materials B135, 274e279. Schrank, S.G., Jose´, H.J., Moreira, R.F.P.M., Schroder, H.Fr., 2004. Elucidation of the behavior of tannery wastewater under advanced oxidation conditions. Chemosphere 56, 411e423. Schrank, S.G., Jose´, H.J., Moreira, R.F.P.M., Schroder, H.Fr., 2005. Applicability of Fenton and H2O2/UV reactions in the treatment of tannery wastewaters. Chemosphere 60, 644e655. Segner, H., Caroll, K., Fenske, M., Janssen, C.R., Maack, G., Pascoe, D., Schafers, C., Vandenbergh, G.F., Watts, M., Wenzel, A., 2003. Identification of endocrine-disrupting effects in aquatic vertebrates and invertebrates: report from the European IDEA project. Ecotoxicology and Environmental Safety 54, 302e314. Segura, C., Zaror, C., Mansilla, H.D., Mondaca, M.A., 2008. Imidacloprid oxidation by photo-Fenton reaction. Journal of Hazardous Materials 150, 679e686. Selcuk, H., Eremektar, G., Meric, S., 2006. The effect of pre-ozone oxidation on acute toxicity and inert soluble COD fractions of a textile finishing industry wastewater. Journal of Hazardous Materials B137, 254e260. Selcuk, H., Rizzo, L., Nikolaou, A., Meric, S., Belgiorno, V., Bekbolet, M., 2007. DBPs formation and toxicity monitoring in different origin water treated by ozone and alum/PAC coagulation. Desalination 210, 31e43. Shemer, H., Kunukcu, Y.K., Linden, K.G., 2006. Degradation of the pharmaceutical metronidazole via UV, Fenton and photoFenton processes. Chemosphere 63, 269e276. Silva, A.C., Dezotti, M., Sant’Anna Jr., G.L., 2004. Treatment and detoxification of a sanitary landfill leachate. Chemosphere 55, 207e214. Somensi, C.A., Simionatto, E.L., Bertoli, S.L., Wisniewski Jr., A., Radetski, C.M., 2010. Use of ozone in a pilot-scale plant for textile wastewater pre-treatment: physico-chemical efficiency, degradation by-products identification and environmental toxicity of treated wastewater. Journal of Hazardous Materials 175, 235e240. Stackelberg, P.E., Gibs, J., Furlong, E.T., Meyer, M.T., Zaugg, S.D., Lippincott, R.L., 2007. Efficiency of conventional drinkingwater-treatment processes in removal of pharmaceuticals and other organic compounds. Science of the Total Environment 377, 255e272. Stalter, D., Magdeburg, A., Weil, M., Knacker, T., Oehlmann, J., 2010. Toxication or detoxication? In vivo toxicity assessment of ozonation as advanced wastewater treatment with the rainbow trout. Water Research 44, 439e448. Staples, C.A., Dorn, P.B., Klecka, G.M., O’Block, S.T., Harris, L.R., 1998. A review of the environmental fate, effects, and exposures of bisphenol A. Chemosphere 36, 2149e2173. Strotmann, U.J., Keinath, A., Hfittenhain, S.H., 1995. Biological test systems for monitoring the operation of wastewater treatment plants. Chemosphere 30, 327e338. Tezcanli-Guyer, G., Ince, N.H., 2003. Degradation and toxicity reduction of textile dyestuff by ultrasound. Ultrasonics Sonochemistry 10, 235e240. Tezcanli-Guyer, G., Ince, N.H., 2004. Individual and combined effects of ultrasound, ozone and UV irradiation: a case study with textile dyes. Ultrasonics 42, 603e609. Tisler, T., Zagorc-Koncan, J., Cotman, M., Drolc, A., 2004. Toxicity potential of disinfection agent in tannery wastewater. Water Research 38, 3503e3510. Tothill, I.E., Turner, A.P.F., 1996. Developments in bioassay methods for toxicity testing in water treatment. Trends in analytical Chemistry 75, 178e188. Trovo´, A.G., Nogueira, R.F.P., Agu¨era, A., Fernandez-Alba, A.R., Sirtori, C., Malato, S., 2009. Degradation of sulfamethoxazole
4340
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 1 1 e4 3 4 0
in water by solar photo-Fenton. Chemical and toxicological evaluation. Water Research 43, 3922e3931. USEPA, 1988. Methods for Toxicity Tests of Single Substances and Liquid Complex Wastes with Marine Unicellular Algae. Environmental Research Laboratory, Gulf Breeze, FL. EPA/600/04. USEPA, 1989. Lettuce seed germination (Lactuca sativa). In: Protocol for Short Term Toxicity Screening of Hazardous Waste Sites. Environmental Research Laboratory, Corvallis, OR EPA/600/3-88/029. USEPA, 1996. Ecological Effects Test Guidelines OPPTS 850.1075, Fish Acute Toxicity Test, Freshwater and Marine. Environmental Protection Agency, Prevention, Pesticides and Toxic Substances (7101). EPA 712-C-96-118. USEPA, 2002. Methods for Measuring the Acute Toxicity of Effluents and Receiving Waters to Freshwater and Marine Organisms, fifth ed. U.S. Environmental Protection Agency Office of Water (4303T), 1200 Pennsylvania Avenue, NW Washington, DC 20460. EPA-821-R-02-012. Vajnhandl, Simona, Le Marechal, Alenka Majcen, 2007. Case study of the sonochemical decolouration of textile azo dye Reactive Black 5. Journal of Hazardous Materials 141, 329e335. Valerio, M.E., Garcı´a, J.F., Peinado, F.M., 2007. Determination of phytotoxicity of soluble elements in soils, based on a bioassay with lettuce (Lactuca sativa L.). Science of the Total Environment 378, 63e66. Van der Kooij, D., Visser, A., Hijnen, W.A.M., 1982. Determination of easily assimilable organic carbon in drinking water. Journal of the American Water Works Association 74, 540e545. Van der Kooij, D., 1990. Assimilable organic carbon (AOC) in drinking water. In: McFeters, G.A. (Ed.), Drinking Water Microbiology, pp. 57e87. New York. Von Gunten, U., 2003. Ozonation of drinking water: Part I. Oxidation kinetics and product formation. Water Research 37, 1443e1467. Vrouwenvelder, H.S., van Paassen, J.A.M., Folmer, H.C., Hofman, J. A.M.H., Nederlof, M.M., van der Kooij, D., 1998. Biofouling of membranes for drinking water production. Desalination 118, 157e166. Walsh, G.E., Bahner, L.H., Horning, W.B., 1980. Toxicity of textile mill effluents to freshwater and estuarine algae, crustaceans and fishes. Environmental Pollution Series A, Ecological and Biological 21, 169e179. Weinrich, L.A., Jjemba, P.K., Giraldo, E., LeChevallier, M.W., 2010. Implications of organic carbon in the deterioration of water
quality in reclaimed water distribution systems. Water Research 44, 5367e5375. Weng, D., Lu, Y., Wei, Y.N., Liu, Y., Shen, P.P., 2007. The role of ROS in microcystin-LR-induced hepatocyte apoptosis and liver injury in mice. Toxicology 232, 15e23. Westerhoff, P., Yoon, Y., Snyder, S., Wert, E., 2005. Fate of endocrine-disruptor, pharmaceutical, and personal care product chemicals during simulated drinking water treatment processes. Environmental Science and Technology 39, 6649e6663. WHO, 2008. Guidelines for drinking-water quality: incorporating 1st and 2nd addenda, vol. 1, Recommendations, third ed. Wieczorek, J.K., Wieczorek, Z.J., 2007. Phytotoxicity and accumulation of anthracene applied to the foliage and sandy substrate in lettuce and radish plants. Ecotoxicology and Environmental Safety 66, 369e377. Wong, S.L., Wainwright, J.F., Piment, J., 1995. Quantification of total and metal toxicity in wastewater using algal bioassays. Aquatic Toxicology 31, 57e75. Yeber, M.C., Rodriguez, J., Freer, J., Baeza, J., Duran, N., Mansilla, H.D., 1999. Advanced oxidation of a pulp mill bleaching wastewater. Chemosphere 39, 1679e1688. Zapata, A., Oller, I., Rizzo, L., Hilgert, S., Maldonado, M.I., Sa´nchez-Pe´rez, J.A., Malato, S., 2010. Evaluation of operating parameters involved in solar photo-Fenton treatment of wastewater: interdependence of initial pollutant concentration, temperature and iron concentration. Applied Catalysis B: Environmental 97, 292e298. Zapata, A., Velegraki, T., Sa´nchez-Pe´rez, J.A., Mantzavinos, D., Maldonado, M.I., Malato, S., 2009. Solar photo-Fenton treatment of pesticides in water: effect of iron concentration on degradation and assessment of ecotoxicity and biodegradability. Applied Catalysis B: Environmental 88, 448e454. Zoh, K.-D., Kim, T.-S., Kim, J.-G., Choi, K., Yi, S.-M., 2006. Parathion degradation and toxicity reduction in solar photocatalysis and photolysis. Water Science and Technology 53 (3), 1e8. Zorrig, W., Rouached, A., Shahzada, Z., Abdelly, C., Davidiana, J.-C., Berthomieu, P., 2010. Identification of three relationships linking cadmium accumulation to cadmium tolerance and zinc and citrate accumulation in lettuce. Journal of Plant Physiology 167, 1239e1247.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 4 1 e4 3 5 4
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Review
Occurrence and control of nitrogenous disinfection by-products in drinking water e A review Tom Bond*, Jin Huang, Michael R. Templeton, Nigel Graham Department of Civil and Environmental Engineering, Imperial College London, London SW7 2AZ, United Kingdom
article info
abstract
Article history:
The presence of nitrogenous disinfection by-products (N-DBPs), including nitrosamines,
Received 10 November 2010
cyanogen halides, haloacetonitriles, haloacetamides and halonitromethanes, in drinking
Received in revised form
water is of concern due to their high genotoxicity and cytotoxicity compared with regu-
29 May 2011
lated DBPs. Occurrence of N-DBPs is likely to increase if water sources become impacted by
Accepted 30 May 2011
wastewater and algae. Moreover, a shift from chlorination to chloramination, an option for
Available online 7 June 2011
water providers wanting to reduce regulated DBPs such as trihalomethanes (THMs) and haloacetic acids (HAAs), can also increase certain N-DBPs. This paper provides a critical
Keywords:
review of the occurrence and control of N-DBPs. Data collated from surveys undertaken in
NDMA
the United States and Scotland were used to calculate that the sum of analysed haloni-
Nitrosamines
tromethanes represented 3e4% of the mass of THMs on a median basis; with Pearson
Haloacetonitriles
product moment correlation coefficients of 0.78 and 0.83 between formation of dihaloa-
Cyanogen halides
cetonitriles and that of THMs and HAAs respectively. The impact of water treatment
Halonitromethanes
processes on N-DBP formation is complex and variable. While coagulation and filtration are
Haloacetamides
of moderate efficacy for the removal of N-DBP precursors, such as amino acids and amines, biofiltration, if used prior to disinfection, is particularly successful at removing cyanogen halide precursors. Oxidation before final disinfection can increase halonitromethane formation and decrease N-nitrosodimethylamine, and chloramination is likely to increase cyanogen halides and NDMA relative to chlorination. ª 2011 Elsevier Ltd. All rights reserved.
Contents 1. 2.
Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Occurrence of N-DBPs in drinking water . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1. Interpreting N-DBP occurrence data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2. Haloacetonitriles (HANs) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3. Haloacetamides (HAcAms) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4342 4343 4343 4343 4344
* Corresponding author. Present address: Pollution Research Group, School of Chemical Engineering, University of KwaZulu-Natal, Durban 4041, South Africa. Tel.: þ27 31 260 3131; fax: þ27 31 260 3241. E-mail address:
[email protected] (T. Bond). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.05.034
4342
3. 4.
5.
1.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 4 1 e4 3 5 4
2.4. Cyanogen halides (CNX) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.5. Halonitromethanes (HNMs) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.6. Nitrosamines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.7. Other N-DBPs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Precursor sources and formation pathways . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Key factors in the formation of N-DBPs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1. Water quality parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1.1. The impact of pH . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1.2. The impact of bromide and iodide . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2. The impact of treatment and disinfection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2.1. The impact of treatment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2.2. The impact of pre-oxidation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2.3. The impact of chlorination and chloramination . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusions: minimising N-DBPs in water treatment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Background
Research into disinfection by-products (DBPs), an unintentional result of water treatment, stems from the mid-1970s, when the formation of trihalomethanes (THMs) was linked to reactions between chlorine and natural organic matter (NOM) in Dutch drinking water (Rook, 1974). By the turn of that decade the THMs were regulated in the USA at 100 mg L1 due to a cancer risk, and a second group of DBPs, the haloacetic acids (HAAs), had been identified in drinking water at comparable levels to THMs. Soon after, the haloacetonitriles (HANs), an important group of N-DBPs, were detected in chlorinated natural waters (Oliver, 1983; Trehy and Bieber, 1981). An enhanced risk of bladder cancer has been associated with exposure to DBPs (Villanueva et al., 2007), although the contribution of various DBPs to this association remains uncertain, given that no identified chlorination DBPs are believed to be plausible bladder carcinogens (Hrudey, 2009). Over 600 DBPs have been reported in drinking water or simulated laboratory disinfection tests, resulting from the use of chlorine and other disinfectants, notably chloramines, ozone and chlorine dioxide (Krasner et al., 2006). However, data for many nitrogenous DBPs (N-DBPs) remains relatively limited. In the United States (US) selected N-DBPs were analysed in water treatment plant (WTP) surveys undertaken in 1988e1989 (Krasner et al., 1989), 1997e98 (McGuire et al., 2002), 2000e2002 (Krasner et al., 2006; Weinberg et al., 2002) and 2006e2007 (Krasner et al., 2007; Mitch et al., 2009). The 2000e2002 study encompassed over 70 emerging DBPs and among the analysed N-DBPs were HANs, halonitromethanes (HNMs) and haloacetamides (HAcAms). There is less information available on the occurrence of N-DBPs in other countries, although relevant surveys have been carried out in Canada (Williams et al., 1995, 1997), Australia (Simpson and Hayes, 1998) and Scotland (Goslan et al., 2009). Another important N-DBP is N-nitrosodimethylamine (NDMA), formerly used in the production of rocket fuel and other industrial processes. Initially detected in Canadian drinking water in the 1980s (Jobb et al., 1994) NDMA has since been reported as a DBP produced from reactions
4345 4346 4346 4346 4348 4349 4349 4349 4350 4350 4350 4351 4351 4351 4352 4352
between monochloramine (NH2Cl) and dimethylamine (DMA) (Choi and Valentine, 2002). Several factors have seen a particular recent focus on NDBPs. Firstly, many N-DBPs are of greater perceived health risk than regulated DBP species. Comparison of data from in vitro mammalian cell tests demonstrated the HANs, HNMs and HAcAms are all far more cytotoxic and genotoxic than the non-nitrogenous THMs and HAAs, although the haloacetaldehydes also exhibit very high cytotoxicity and genotoxicity (Plewa and Wagner, 2009). Moreover, the nitrosamines may play a significant role in human carcinogenesis (Loeppky and Michejda, 1994) and the United States Office of Environmental Health Hazard Assessment (OEHHA) have issued a public health goal of 3 ng L1 for NDMA (OEHHA, 2006). At present, however, no N-DBPs are formally regulated by large governmental bodies anywhere in the world (Box 1). Secondly, water utilities are increasingly switching from chlorination to alternative disinfectants, particularly chloramines, in order to limit the formation of regulated THMs and HAAs (Seidel et al., 2005). Disinfectant type is a key factor in NDBP formation since, depending on the compound and reaction conditions, the nitrogen can derive either from the organic precursors, i.e. dissolved organic nitrogen (DON), or in the case of chloramination, from the disinfectant. Finally, the impact of human activity upon drinking water sources is increasingly being felt in the form of wastewater effluent and algal activity (Mitch et al., 2009). Since these are both enriched in DON, their presence is likely to lead to raised concentrations of many N-DBPs. Of the components of DON, amino acids are known to act as precursors of HANs, HAcAms and cyanogen halides (CNX) (Hirose et al., 1988; Ram, 1985; Reckhow et al., 2001; Trehy et al., 1986), while amine precursors of NDMA are believed to be largely anthropogenic in origin (Sacher et al., 2008; Schreiber and Mitch, 2006b), in contrast to the THMs and HAAs, where NOM, typically of terrestrial origin, is the main precursor pool. Hence, understanding and controlling the incidence of NDBPs is a contemporary challenge to the water industry. The objectives of this review are to highlight typical concentrations of identified N-DBPs drinking water, investigate
4343
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 4 1 e4 3 5 4
Box 1 Acronyms used. AOM CNBr CNCl CNMs CNX CNX DBPs BCAN DBAN DHAN DCAA DBAcAm DCAN DEA DHAN DHNMFP DHNMs DMA DON DPA EEM EfOM FP HAA9 HAAs HAcAms HAN4
Algal organic matter Cyanogen bromide Cyanogen chloride Chloronitromethanes Cyanogen halides CNX formation potential Disinfection by-products Bromochloroacetonitrile Dibromoacetonitrile Dihaloacetonitrile Dichloroacetic acid Dibromoacetamide Dichloroacetonitrile Diethylamine Dihaloacetonitrile Dihalonitromethane formation potential Dihalonitromethanes Dimethylamine Dissolved organic nitrogen Dipropylamine Excitation-emission matrix Effluent organic matter Formation potential Sum of nine surveyed HAAs Haloacetic acids Haloacetamides Sum of four HANs (DCAN, BCAN, DBAN and TCAN)
potential relationships between N-DBPs and other DBPs based on published data and to examine strategies for their mitigation. (Table 1).
2.
Occurrence of N-DBPs in drinking water
2.1.
Interpreting N-DBP occurrence data
A number of caveats should be kept in mind by those reviewing and comparing available data on N-DBPs. Firstly, important differences in disinfection practice between countries impact upon exposure of precursor material to disinfectants. Many US WTPs operate pre-chlorination or pre-chloramination (i.e. before or intermediate to other treatment) as well as post-disinfection, whereas in Europe post-chlorination/chloramination alone is typical. Furthermore, in the US 2000e2002 survey many selected WTPs had high bromide (median level 120 mg L1) and total organic carbon (median 5.8 mg L1) levels (Krasner et al., 2006) and were thus thought likely to generate relatively high DBP loads. A confounding factor when comparing relative effects of disinfectants on N-DBP formation is that it is often WTPs treating high bromide and/or organic carbon waters which switch from chlorination to chloramination in an attempt to lower formation of regulated DBPs. In the 2006e2007 US NDBP survey most WTP intake waters were impacted by upstream algal blooms and/or the discharge of treated wastewater (Mitch et al., 2009), where it can be expected that
HANs HNMs ICR MOR ND NDBA NDMA N-DBPs NDPA NEMA NMOR NOM NPOC NPYR NR OEHHA PYR TCAN TCNM THM4 THMs TMA TOC TOX UDMH US WTP WWTP
Haloacetonitriles Halonitromethanes Information collection rule Morpholine Not detected N-nitrosodibutylamine N-nitrosodimethylamine Nitrogenous disinfection by-products N-nitrosdipropylamine N-nitrosoethylmethylamine N-nitrosomorpholine Natural organic matter Non-purgable organic carbon N-nitrosopyrrolidine Not reported Office of Environmental Health Hazard Assessment Pyrrolidine Trichloroacetonitrile Trichloronitromethane (chloropicrin) Sum of four regulated THMs Trihalomethanes Trimethylamine Total organic carbon Total organic halogen Unsymmetrical 1,1-dimethylhydrazine United States Water treatment plant Wastewater treatment plant
N-DBP formation is above that found in more pristine water sources. Moreover, some N-DBPs have been identified in atypical waters: in the study where 2,3,5-tribromopyrrole was first detected (along with 3-bromopropanenitrile) bromide concentrations were 2 mg L1 (Richardson et al., 2003) and/or using sample concentration techniques followed by qualitative gas chromatography (GC) methods without commercially-available standards (Richardson et al., 1999, 2003). Meanwhile, disparate laboratory disinfection protocols are used for measuring DBPs formed from model compounds or isolates of NOM. To illustrate, in two studies testing the formation potential of NOM isolates the pH, chloramine dose and contact time were respectively 7.0, 45 mg of Cl2 per mg DOC (pre-formed monochloramine added) and 10 days (Lee et al., 2007) and 8.0, 3.0 mg Cl2 per mg DOC (ammonia added before chlorine at 1:3 weight ratio) and 3 days (Dotson et al., 2009).
2.2.
Haloacetonitriles (HANs)
Of the three major N-DBP groups captured by existing analytical methodologies e HANs, HAcAms and HNMs - in the 2000e2002 US survey, HANs occurred at the highest concentrations, with median and maximum levels of 3 and 14 mg L-1 respectively, and dichloroacetonitrile (DCAN) was the most prevalent species (Table 2) (Krasner et al., 2006; Weinberg et al., 2002). In the 2006e2007 US survey, median values for the sum of DCAN, bromochloroacetonitrile (BCAN), dibromoacetonitrile (DBAN) and trichloroacetonitrile (TCAN)
4344
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 4 1 e4 3 5 4
Table 1 e Important nitrogenous disinfection by-products (N-DBPs). Group/formula
Structure
Important species
R Haloacetonitriles (HANs) R3CCN
R
N R
Cyanogen halides (CNX) RCN
R NH2
R
R
N R
Halonitromethanes (HNMs) R3CNO2
N+
R
Dichloroacetamide (DCAcAm) (right), dibromoacetamide (DBAcAm), trichloroacetamide (TCAcAm)
Cyanogen chloride (CNCl) (right), cyanogen bromide (CNBr)
N Cl
N
Cl
O
Cl
NH2
H
Cl
O
N
O
N Cl
Trichloronitromethane (chloropicrin) (right), tribromonitromethane (bromopicrin), bromodichloronitromethane, dibromochloronitromethane
R
R
H
O-
R
Nitrosamines R2NNO
Cl
O
R Haloacetamides (HAcAms) R3CCONH2
Dichloroacetonitrile (DCAN) (right), bromochloroacetonitrile (BCAN), dibromoacetonitrile (DBAN), trichloroacetonitrile (TCAN) and tribromoacetonitrile (TBAN)
Structure
N-nitrosodimethylamine (NDMA) (right), N-nitrosopyrrolidine (NPYR), N-nitrosomorpholine (NMOR), N-nitrosodiethylamine (NDEA)
ON+
Cl
O
Cl
H3 C N
N
O
H3 C
R is typically Cl, Br, I, H or an alkyl group, though it can also be a larger aliphatic or aromatic group.
(collectively HAN4) were slightly higher at 4.0 mg L1 (Krasner et al., 2007), presumably a reflection of the mainly algal and wastewater-impacted waters chosen. In Australia, HAN4 levels up to 36 mg L1 have been observed, something most likely related to high organic content and bromide, which also resulted in THM levels up to 191 mg L1 (Simpson and Hayes, 1998). In contrast, HANs were lower in Scotland than the US, with median and maximum HAN4 concentrations of 1 mg L1 and 4 mg L1 respectively (Goslan et al., 2009). Median non-purgeable organic carbon (NPOC) and bromide levels were 3.6 mg L1 and 55 mg L1 respectively, in the Scottish waters, versus equivalent figures of 5.8 mg L1 and 120 mg L1 in the US survey (Goslan et al., 2009; Krasner et al., 2006). It has been proposed that the mass of HANs typically represents around 10% of the THMs (Krasner et al., 1989; Oliver, 1983). To investigate such rules the extensive DBP data from the US 2000e2002 survey was collated with that from the 2006e2007 US survey and Scotland and relevant ratios and correlations between the DBP groups calculated (Tables 3 and 4). At the 12 WTPs in the US 2000e2002 survey HAN4 as a proportion of the four regulated THMs (THM4) varied from 2% to 14%, with a median value of 8%, while median ratios for HAN4 and DHAN (HAN4 without TCAN) were respectively 2% and 7% in the Scottish and US 2006e2007 surveys (Table 3, Scottish ratios computed from median
values across the whole of the survey). Thus, the 10% value is an approximate guide to HAN formation. A good positive correlation (r ¼ 0.90) has previously been observed between HAN and THM formation (Krasner et al., 1989). For this review correlations between DHAN and THM4 was calculated as 0.78, with a correlation of 0.83 between HAN4 and the nine surveyed HAAs (HAA9) (Table 4, n ¼ 15). This indicates HAAs may be at least as good a predictor of HAN formation as THMs. As DCAN hydrolyses to 2,2-dichloroacetamide (DCAcAm) and consequently dichloroacetic acid (DCAA) in the presence of free chlorine or at alkaline pH (Reckhow et al., 2001) this is perhaps unsurprising. In the US 2000e2002 survey and Scotland median values of HAN4/DHAN in finished water accounted for 7% of HAA9 formation (Table 3).
2.3.
Haloacetamides (HAcAms)
HAcAms were reported for the first time during the 2000e2002 US survey, DCAcAm being the most prominent species, with a median concentration of 1.3 mg L1 (Table 2). The median and maximum concentrations of the sum of HAcAms were 1.4 mg L1 and 7.4 mg L1 respectively, though note that not all the possible brominated and chlorinated HAcAms were quantified. HAcAms were frequently identified in finished water from three sites where chlorine dioxide was applied prior to chlorine/chloramine. Krasner and co-workers noted
4345
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 4 1 e4 3 5 4
Table 2 e Summary of N-DBP occurrence in effluents of selected US WTPs (Krasner et al., 2006; Weinberg et al., 2002). Occurrence (mg L1)
Name
WTP of max occurrence
Median Maximum Haloacetonitriles Dichloroacetonitrile Bromochloroacetonitrile
Conditions of max occurrence TOC (mg L1)
Bromidea (mg L1)
pH
Disinfection
12.0 10 (opposite) 3.0e4.2 7.5e7.7 50e70 Cl2 - chloramines 3.0 2 (opposite), 1.6e3.0 7.9e8.7 120e400 Cl2 - chloramines 11 (opposite) and 12 (below) 2.9e4.2 7.4e7.5 160e210 ClO2eCl2 - chloramines Other HANs recorded: dibromoacetonitrile, trichloroacetonitrile, dibromochloroacetonitrile, chloroacetonitrile, bromoacetonitrile Sum of haloacetonitriles 3 14 10 (as above) Haloacetamides 2,2-dibromoacetamide 0.6 2.8 12 (opposite) 3.2e4.5 7.6e8.3 150e330 ClO2 - chloramines 2,2-dichloroacetamide 1.3 5.6 6 (opposite), 12 (as above) 3.5e4.5 5.8e7.0 39e80 ClO2eCl2 - chloramines Other HAcAms recorded: 2-bromoacetamide, 2,2,2-trichloroacetamide, 2-chloroacetamide Sum of haloacetamides 1.4 7.4 6 (as above) Halonitromethanes Chloropicrin 0.2 2.0 10 (as above) Bromodichloronitromethane 0.3 3.0 7 3.0e4.2 7.5e7.7 50e70 Cl2 - chloramines Dibromochloronitromethane ND 3.0 12 (as above) Bromopicrin ND 5.0 12 (as above) Other HNMs recorded: bromochloronitromethane, chloronitromethane, bromonitromethane, dichloronitromethane, dibromonitromethane Sum of halonitromethanes 1 10 12 (as above) 1 0.6
a ¼ Raw water bromide concentration.
that DCAcAm occurred at a similar level to DCAN (respective median values 1.3 mg L1 and 1.0 mg L1) and that HAcAm formation was w10% of HAAs, with dichloro representatives of the two groups found at higher levels than the trichloro species (Krasner et al., 2006).
2.4.
Cyanogen halides (CNX)
In the 1988e1989 US survey an association was noted between chloramination and CNCl formation, median values of CNCl in treatment works with free chlorine and chloramination were 0.4 mg L1 and 2.2 mg L1 respectively (Krasner et al., 1989). This finding has been re-confirmed by subsequent research (see Section 4.2.3). During the 2006e2007 NDBP survey CNX (i.e. CNCl plus CNBr) formation was generally
only observed at the plants with chloramination, in which the median and maximum formation was 2.6 mg L1 and 7.8 mg L1 respectively (Mitch et al., 2009). Nonetheless, CNX precursors were widely present, as shown by plant influent samples disinfected under laboratory conditions designed to maximise CNX (3 h pre-chlorination then chloramination for 21 h), which generated respective median and maximum levels of 12 and 34 mg L1 (Mitch et al., 2009). A low formation of CNX was found in plants where ozone was applied prior to biofiltration and chlorination/chloramination, suggesting biological treatment effectively removed formaldehyde and other CNX precursors resulting from ozonation (Krasner et al., 2007). In Australia CNCl has been recorded up to a level of 10 mg L1 from a WTP practicing monochloramination (Simpson and Hayes, 1998).
Table 3 e Ratios (mg/mg) between N-DBPs and other DBP groups in finished water samples from US and Scotland. Survey
US 2000e2002 Weinberg et al., 2002
US 2006e2007 a
Mitch et al., 2009
Ratios
Min
Median
Max
HAN4/THM4 DHAN/THM4 HAN4/HAA9 HAN4/DXAA TCNM/THM4 Sum of HNMs/THM4 Sum of HNMs/HAA9
0.02 0.02 0.02 0.02 0 0 0
0.08 0.08 0.07 0.13 0.00 0.03 0.03
0.14 0.14 0.12 0.2 0.01 0.23 0.19
Scotland b
Goslan et al., 2009 c
25th %ile
Median
90th %ile
0.07
0.07
0.16
Median 0.02 0.07
0 0
0.01 0.01
0.03 0.04
0.00
a ¼ Ratios calculated from mean values of 4e5 seasonal samples taken at each of 12 WTPs. Not reported taken as half the minimum reporting level; not detected taken as zero. Between four and nine HNMs quantified, depending on the sample (see Table 2). b ¼ Five HNMs quantified. c ¼ Ratios computed from median data across whole of survey. THM4 ¼ sum of four regulated THMs; HAA9 ¼ sum of nine surveyed HAAs. DHANs ¼ HAN4 e TCAN.
4346
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 4 1 e4 3 5 4
Table 4 e Correlationsa between DBP groups in USb and Scottishb DBP surveys.
HAA9 (n ¼ 16) DXAA (n ¼ 12) TXAA (n ¼ 12) HAN4 (n ¼ 16) DHAN (n ¼ 15) Sum of HNMs (n ¼ 15) TCNM (n ¼ 19)
THM4 (n ¼ 19)
HAA9
DXAA
TXAA
HAN4
DHANs
HNMs
0.81 0.76 0.80 0.30 0.78 0.02 0.09
0.84 0.89 0.71 0.83 0.14 0.54
0.42 0.59 0.61 0.56 0.44
0.83 0.79 0.21 0.62
0.99 0.06 0.42
0.18 0.67
0.30
a Correlations ¼ Pearson product moment correlation coefficients (r). b Bulk of data (n ¼ 12 for all analytes) from US 2000e2002 survey (mean values of 4e5 seasonal samples taken at each of 12 WTPs. Between four and nine HNMs quantified). Remainder (n ¼ 3) from US 2006e2007 survey (THM4, DHAN, sum of (five) HNMs and TCNM) and (n ¼ 4) from Scottish survey (THM4, HAA9 and TCNM). DHANs ¼ HAN4 e TCAN. TCNM ¼ trichloronitromethane.
2.5.
Halonitromethanes (HNMs)
For the sum of HNMs, median and maximum levels of 1 and 10 mg L1 were recorded during the 2000e2002 US survey. Bromopicrin and dibromochloronitromethane maxima were 5.0 and 3.0 mg L1, respectively (Table 2) at a site characterised by high bromide (150e330 mg L1), pre-oxidation with chlorine dioxide and post-chloramination. The sum of analysed HNMs represented respectively 3% and 1% of THM4 on a median basis in the US in 2000e2002 and 2006e2007(Table 3), although a maximum 23% of THM4 formation was recorded at one location where THM4 was relatively low (mean ¼ 8.5 mg L1). As intimated by this, HNM formation does not appear related to the THMs or HAAs, with no meaningful correlations calculated between the sum of HNMs and the two regulated groups (Table 4, n ¼ 15). The formation of chloropicrin (trichloronitromethane (TCNM)) in the 2006e2007 N-DBP study was higher, with median and maximum values of 0.5 and 7.6 mg/L, respectively (Krasner et al., 2007), again highlighting the importance of wastewater and algae as precursor sources. Ozonation before chlorination can dramatically enhance HNM formation (Hoigne and Bader, 1988) (see Section 4.2.2).
2.6.
Nitrosamines
NDMA is typically observed in the low ng L1 range in drinking water, although concentrations equal to or above 1000 ng L1 have been recorded in chloraminated raw water (Sacher et al., 2008) and chlorinated or chloraminated wastewater effluent (Krasner et al., 2009a; Mitch et al., 2003). Analysis of a Canadian drinking water supply in 1986 detected NDMA at concentrations between 5 and 115 ng L1 (Jobb et al., 1994), which stimulated a survey of 145 WTPs in Ontario, Canada. Finished water was under 5 ng L1 in the majority of cases. These and other studies highlighted associations between elevated NDMA occurrence and municipal and industrial wastewater input, chloramination, cationic polymers and ion exchange resins (Najm and Trussell, 2001). Various synthetic chemicals containing a DMA moiety have subsequently been identified as NDMA precursors. These include the pharmaceutical ranitidine (Sacher et al., 2008) and diuron, a herbicide (Chen and Young, 2008). Charrois et al. (2004) developed an ammonia positive chemical ionisation method which enabled detection of two
additional nitrosamines in drinking water: N-nitrosopyrrolidine (NPYR) and N-nitrosomorpholine (NMOR), at 2e4 ng L1 and 1 ng L-1 respectively, in addition to NDMA at 2e180 ng L1. Also evident were increased levels of NDMA in the distribution system (180 ng L1) relative to treated effluent (67 ng L1) of a plant using chloramination and UV disinfection. A recent study compared the formation of eight nitrosamines in finished water samples from six utilities using various treatments with raw water samples from the same sources chloraminated under laboratory conditions (Sacher et al., 2008) (Table 5). In contrast to the treated water samples, where NDMA peaked at 4.9 ng L1 and no other nitrosamines were reported, the laboratory disinfected samples generated NDMA up to 110 ng L1 and NPYR and Nnitrosoethylmethylamine (NEMA) at maxima of 7.6 and 3.4 ng L1, respectively, indicating precursors of these species were present in raw waters. Chloramination of 81 river and lake samples revealed a similar pattern: NDMA was always the dominant species, with median and maximum levels of 45 and 1000 ng L1 respectively, while other nitrosamines were periodically present, albeit always at least an order of magnitude lower than NDMA. NPYR and N-nitrosodiethylamine (NDEA) were the second and third most frequently recorded nitrosamines and reached respective peaks of 35 and 23 ng L1 (Table 5). Another nitrosamine, N-nitrosodibutylamine (NDBA), has been detected in one UK water distribution system at 6.4 ng L1 (Templeton and Chen, 2010).
2.7.
Other N-DBPs
Intermediates in the reactions schemes portrayed (Figs. 1e3) are either known or presumed to occur in drinking water. In particular, hydrazine is carcinogenic and has been detected at 0.5e2.6 ng L1 in chloraminated drinking water, though was not detected in chlorinated samples (Davis and Li, 2008). Other organic hydrazines presumably form during the unsymmetrical hydrazine pathway of nitrosamine formation (displayed for NDMA in Fig. 3) (Choi and Valentine, 2002). Organic chloramines form from chlorination or chloramination of DON, for example, amino acid chlorination (Fig. 1) and can lead to a w10% overestimation of disinfection capacity in chloraminated water systems (Lee and Westerhoff, 2009). There are additional N-DBPs for which very limited occurrence data exists. Benzeneacetonitrile, heptanenitrile and cyanoformaldehyde have been detected as ozone
4347
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 4 1 e4 3 5 4
Table 5 e Summary of nitrosamine levels in drinking water sources (Sacher et al., 2008) and of DBPs in wastewater (Krasner et al., 2009a and 2009b). Survey
Range of NDMA levels (ng/L)
Reference: Sacher et al., 2008 81 surface water samples (rivers and lakes) chloraminated under laboratory conditions (0.4 mM, 7 days). Eight nitrosamines. Original water samples from 6 water utilities. Samples from same 6 water utilities chloraminated under laboratory conditions (0.4 mM, 7 days). References: Krasner et al., 2009a and 2009b 23 US WWTPs at different seasons. Before chlorination/chloramination
Median (ng/L)
1e1000
45
= CN d d 2 Ko2 md > > ; ; m ¼ Ko2 ¼ CN nCN C ¼
(5)
where CN is DO concentration in the bulk water (see Fig. 2). A characteristic parameter, the Schmidt number defined by Eq. (6), appears in Eqs. (3) and (4). Sc ¼
n Dw
(6)
The Schmidt number for DO in water is a function of the temperature, e.g. Sc ¼ 1240 at 4 C (hypolimnion of deep lake or colder stream) and Sc ¼ 300 at 30 C (tropical and some shallower systems in summer). In this study, the Schmidt number is taken to be 500 at 20 C (Denny, 1993).
4384
3.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 8 1 e4 3 8 9
Solution
The non-dimensional basic equations (3) and (4) were expressed in the Crank-Nicolson finite-difference form with a spatial step (Dy*) of 0.01 and a time step (Dt*) of 0.01. Equations (3) and (4) are subject to two boundary conditions, respectively. The DO concentration at y ¼ N (far away from the sediment/water interface) is taken to be CN (the DO concentration in the bulk water) for Eq. (3). Although the continuous inflow is considered, the inflow rate is assumed to be much smaller than the size of the reservoir. For Eq. (4), the DO concentration at the bottom of the sediment (y ¼ -d) is take to be zero, i.e. C ¼ 0. Both equations (3) and (4) also require the DO concentration at the sediment/water interface (Cw). Its value is calculated explicitly. First, the DO concentration C1nþ1 next to the sediment/water interface in the water column is calculated as ¼ Cn1 þ Cnþ1 1
Dt 1 n $ C 2Cn1 þ Cnw Dy2 Sc 2
(7)
Fig. 3 e Variation of DO concentration profile with time for Sc [ 500, Ko2 [ 0.2 mg lL1, and m [ 100 mg lL1 dL1.
where superscript (n) denotes the time step. Second, the DO concentration C1nþ1 next to the sediment/water interface in the sediment for the next time step is calculated as n Cnþ1 1 ¼ C1 þ
1 Dt n m $Dt $ 2 C2 2Cn1 þ Cnw Cn1 n Ko2 þ C1 2Sc Dy
(8)
Using C1nþ1and C1nþ1, the DO concentration Cwnþ1 at the sediment/water interface is obtained as ¼ Cnþ1 w
Cnþ1 þ Cnþ1 1 1 2
(9)
The DO concentration profile as a function of time is determined by solving equations (3) and (4) implicitly. The computational domain went from y ¼ d ¼ 10 mm to y ¼ 100 cm.
4. Time variation of DO profile and DO flux at the sediment/water interface The time variation of the DO concentration profile near the sediment/water interface obtained by the model is shown in Fig. 3 for the Schmidt number Sc ¼ 500 and the maximum oxidation rate m ¼ 100 mg l1 d1. The normalized DO concentration (C*) was taken to be zero everywhere in the sediment and to be CN ¼ 10 mg l1 in the water column, initially. The DO profile varies substantially with time in the water column, but does not change significantly in the sediment. The normalized DO concentration at the sediment/ water interface (Cw/CN) is less than 0.1, and is close to zero regardless of biochemical activity characterized by the value of m. When flow is present over the sediment surface, the normalized DO concentration at the sediment/water interface (Cw/CN) and the oxygen penetration depth (ds) were reported to be 0.50e0.96, and 0.44e0.56 cm, respectively, for pffiffiffiffiffiffiffiffiffiffi m ¼ 100 mg l1 d1 depending on the shear velocity (U* ¼ s0 =r, where s0 is the boundary shear stress and r is the fluid density) (Higashino et al., 2004). Figs. 3 and 6 show that the DO
penetration depths are close to zero beside the maximum oxidation rate is small (m ¼ 10 mg l1 d1). This indicates that except highly organic sediment (m ¼ 10,000 mg l1 d1) organic matter can be oxidized sufficiently in an aerobic layer which is on the order of a few millimeter just below the sediment/ water interface, i.e. dissolved organic carbon is hardly transferred from the sediment to overlying water. The time variation of the DO flux at the sediment/water interface (SOD) obtained by the model is shown in Fig. S1 for m ¼ 100 mg l1 d1 and the detention time T ¼ 1 h. When the water body was replaced by a new one, the normalized DO concentration in the water column everywhere was taken to be unity again, i.e. C ¼ CN. The oxygen flux at the sediment/ water interface (J), i.e. SOD, can be calculated from the DO concentration profile by Fick’s law as
Fig. 4 e SODave for stagnant water versus detention time (T ) for Sc [ 500 and Ko2 [ 0.2 mg lL1.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 8 1 e4 3 8 9
4385
stagnant, the time-averaged SOD (SODave) over a period of the detention time (T) defined by Eq. (11) is introduced. SODave ¼
1 T
ZtþT JðtÞdt
(11)
t
Since the DO profile and corresponding DO flux at the sediment/water interface (SOD) vary with time when water is
Fig. 4 illustrates the SODave calculated by Eq. (11) for the Schmidt number Sc ¼ 500, and biochemical oxygen uptake rate m ¼ 10, 100, 1000, and 10,000 mg l1 d1. As shown in Eq. (1), the model assumes that there is no oxygen sink in the water column. Considering the short time periods (days to weeks) the model can be applicable when the sediment is not highly organic, e.g. m 1000 mg l1 d1 because an aerobic zone which is on the order of a few millimeters is present just below the sediment/water interface (see Fig. 6). Highly organic sediments in e.g. eutrophic lakes can be anoxic, and then, the mass transport of dissolved organic carbon from the sediment to overlying water may play an important role in determining the oxygen balance in the water column, and hence, DO concentration near the sediment/water interface. The detention time (T) is over a range from 1 h to 3 months. Since effects of benthic fluxes of dissolved organic carbon and ammonia from the sediment are not taken into account, the model may underestimate SOD for long time periods (T ¼ 2 and 3 months). The SODave is large when the detention time (T) is short, and diminishes as the detention time (T) becomes longer. The SODave decreases to 21, 8, and 4% of the SODave for T ¼ 1 h with increasing detention time, i.e. T ¼ 1 day, 1 week, and 1 month, respectively, for m ¼ 100 mg l1 d1. The SODave increases as the biochemical oxygen uptake rate (m) becomes larger. But dependence of the SODave on oxygen consumption rate (m) in the sediment is not as strong as the SOD with fluid flow (see Figs. 5 and 8 in Higashino et al., 2004). As can be expected, the water column DO next to the sediment/water interface gets lower with increasing detention time (T), and thus, the SODave becomes smaller depending on the microbial activity in the sediment described by m. For long time periods (more than several months), re-aeration,
Fig. 6 e Oxygen penetration depth (ds) versus detention time (T) for Sc [ 500 and Ko2 [ 0.2 mg lL1.
Fig. 7 e Dependence of SOD for stagnant water versus SOD with fluid flow (SODave/SODN) on detention time (T).
Fig. 5 e Boundary layer thickness (dw) versus detention time (T) for Sc [ 500 and Ko2 [ 0.2 mg lL1.
J ¼ Dw
vC vC ¼ D s vy y¼þ0 vy y¼0
(10)
The value of SOD gets smaller with time resulted from decreasing DO concentration gradient at the sediment/water interface (see Fig. 3). When the water body is replaced by a new one, the SOD increases discontinuously due to a large DO concentration gradient at the sediment/water interface. The change in the SOD during the detention time (T) shown in Fig. S1 repeats the same pattern every period of the detention time (T).
5.
Time-averaged SOD
4386
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 8 1 e4 3 8 9
longer. The DO penetration depth (ds) depends also on biochemical oxygen uptake. The DO penetration depth (ds) is large when the value of m is small, i.e. m ¼ 10 mg l1 d1, whereas decreases as the biochemical oxygen uptake rate (m) becomes larger. The DO penetration depth (ds) can close to zero for m > 1000 mg l1 d1. When flow is present over the sediment surface, the DO penetration depth was reported to be 0.44e0.56, and 0.10e0.18 cm for m ¼ 100 and 1000 mg l1 d1, respectively (Higashino et al., 2004). These are much larger than those for stagnant water.
6. Comparison of SOD for stagnant water to SOD with fluid flow
Fig. 8 e Dependence of SOD with fluid flow versus SOD for stagnant water on shear velocity.
photosynthesis, and the oxygen demand due to benthic flux of dissolved organic carbon (CBOD) and ammonia (NBOD) may have a significant effect on oxygen balance in the water column and on SOD. Since this paper considers relatively short time periods (days to weeks), these are not taken into account. Fig. 4 also shows the dependence of SOD on the detention time (T), and is useful for water quality management in manmade reservoirs when the detention time is short (less than 1 month). The diffusive boundary layer thickness was defined as the distance from the sediment/water interface (y ¼ 0) to the point where C/CN ¼ 0.99 when turbulence is present over the sediment surface (Higashino et al., 2004). Similarly, the boundary layer thickness above the sediment/water interface can be defined as the distance from the sediment/water interface (y ¼ 0) to the point where C/CN ¼ 0.99. For stagnant water the boundary layer starts to grow when the water body is replaced (t ¼ 0). The boundary layer thickens with time, and becomes thickest at t ¼ T (detention time). Simulated boundary layer thickness at t ¼ T is illustrated in Fig. 5. The boundary layer thickness (dw) depends on the detention time (T), i.e. the boundary gets thicker as the detention time (T) becomes longer, and is independent of the value of m. Fig. 5 also indicates how the oxygen depletion zone grows with increasing detention time (T) in lakes and reservoirs due to oxygen uptake by the sediment. When water is stagnant and respiration in the water column is ignored, SOD is a weak oxygen sink, and the oxygen depletion zone grows slowly. Flow above the sediment surface makes SOD a more significant oxygen sink, i.e. SOD increases as the velocity of water flowing above the sediment/water interface becomes larger (Higashino et al., 2004). Both flow over the sediment surface and respiration in the water column make the oxygen depletion zone grow faster. The DO penetration depth (ds) can be defined as the distance from the sediment/water interface to the point where C* ¼ 0.001, and is shown in Fig. 6. The DO penetration depth (ds) diminishes as the detention time (T) becomes
The effect of flow over the sediment surface on oxygen uptake by the sediment can be demonstrated by comparing SOD for stagnant water to SOD with fluid flow. SOD with fluid flow was previously analyzed using the diffusive boundary layer theory and zero-order kinetics, and was given as a function of the shear velocity (U*), and the zero-order rate constant m0 (Higashino et al., 2004) as vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 0 pffiffiffi u 2 12 u u 3 3 2 nSc 3 U A Ds m0 þ tðDs m0 Þ þ2m0 Ds CN @ 2p SOD ¼
pffiffiffi 2 3 3 nSc 3 U 2p
(12)
in which n(¼0.1) is a constant. Fig. S2 illustrates the analytical solution (Eq. (12)) of the SOD with fluid flow (SODflow) for the zero-order rate constant m0 ¼ 10, 100, 1000, and 10,000 mg l1 d1. The SODflow depends on the shear velocity (U*), i.e. the SODflow increases as the shear velocity (U*) increases, and approaches a constant value (SODN) given by Eq. (13) when the shear velocity (U*) is large enough (U*/N in Eq. (12)). SODN ¼
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2m0 Ds CN
(13)
The SODflow also depends on the biochemical oxygen uptake rate, i.e. the zero-order rate constant, in the sediment (m0). The value of SODN obtained by Eq. (13) for m0 ¼ 1000 mg l1 d1 is 10 times larger than that for m0 ¼ 10 mg l1 d1, which is different from the SODave for stagnant water (Fig. S2). SODN can be a good measure for oxygen uptake by the sediment when flow over the sediment surface is present. SODN is related to biochemical oxygen uptake measured by the value m0 in the sediment because when the shear velocity (U*) is large, oxygen transfer through the diffusive boundary layer is fast, and the DO flux at the sediment/water interface (SOD) is limited only by the oxygen consumption rate in the sediment (Higashino et al., 2004). The SODave for stagnant water in Fig. 4 is normalized by the SODN, and is shown in Fig. 7. When water is stagnant SOD is limited by oxygen transfer from the water column to the sediment surface. For weak biochemical activity in the sediment, i.e. m ¼ 10 mg l1 d1, SOD can be sediment-side controlled when the water body is replaced by a new one, and then, the SODave becomes larger than the SODN. This results from a high oxygen transfer rate induced by a large DO concentration gradient near
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 8 1 e4 3 8 9
the sediment/water interface (see Fig. 3). The DO concentration gradient becomes smaller with time, resulting in decreasing oxygen transfer rate, and finally, oxygen transfer in the water column becomes limiting when the detention time (T) is long. When biochemical activity in the sediment is strong, i.e. m > 100 mg l1 d1, SOD is completely controlled by oxygen transfer in the water column. The ratio SODave/SODN decreases as the detention time (T) becomes longer. The decrease in water column and sediment DO concentration causes reduction in SOD due to attenuation of the oxidation rate described by the Monod function in Eq. (2) with increasing detention time (T). Fig. S3 demonstrates that the oxygen attenuation factor (C/ (Ko2 þ C)) in Eq. (2) decreases as the detention time (T) becomes longer at the sediment/water interface for the biochemical oxygen uptake rate m ¼ 10, 100, 1000 and 10,000 mg l1 d1. When the detention time is longer than 10 days the ratio SODave/SODN decreases below 101 for m ¼ 100 and 1000 mg l1 d1, and 102 for m ¼ 10,000 mg l1 d1, respectively. When flow over the sediment surface is present, SOD can be much larger than that for stagnant water (SOD with fluid flow [SOD for stagnant). The SODflow is obtained by Eq. (12), and is normalized by the SODave for the detention time T ¼ 1 day (SODstagnant) as the reference value. Fig. S4 illustrates that the ratio SODflow/SODstagnant depends on the biochemical oxygen uptake rate in the sediment (m). The ratio SODflow/ SODstagnant is about 2, 7, 20 and 60 for m ¼ 10, 100, 1000, and 10,000 mg l1 d1, respectively, when the shear velocity (U*) is large enough. The dependence of the ratio SODflow/SODstagnant on the biochemical oxygen uptake rate (m) is attributed to the fact that SODstagnant is limited by oxygen transfer in the water column, whereas biochemical oxygen consumption controls SODflow when the shear velocity (U*) is large enough. This suggests that flow over the sediment surface hardly raises SOD much larger than that for stagnant water when the biochemical oxygen uptake rate (m) is small. The difference between SOD with fluid flow and SOD for stagnant water becomes significant as the biochemical oxygen uptake rate (m) becomes larger, i.e. SOD is 10e100 times larger when flow is present over highly organic sediments.
7.
SOD for hyporheic exchange
This paper considers impermeable sediments. When sediments consist of permeable materials, e.g. sand and gravel, “hyporheic exchange” can significantly raise the SOD. “Hyporheic exchange” occurs when water flows over the permeable sediment surface, and is due to interstitial flow driven by a spatially and temporally variable pressure distribution at the sediment/water interface. Interstitial flow can be driven by surface waves (Huettel and Webster, 2001; Qian et al., 2009), bedforms (e.g. Elliott and Brooks, 1997; Packman et al., 2004; Cardenas and Wilson, 2006), or near-bed coherent motions (Higashino et al., 2009). When the interstitial water motion is present due to pressure fluctuations at the sediment/water interface induced by near-bed coherent motions, the SOD for larger oxidation rate e.g. m ¼ 1000 and 2000 mg l1 d1 is almost 5 times that with no pore water flow (Higashino and Stefan, 2011).
8.
4387
Conclusions
This paper considers sedimentary oxygen demand (SOD) for stagnant water, and investigates how flow over the sediment surface raises SOD compared to stagnant water, depending on flow velocity and biochemical activity in the sediment mainly in manmade reservoirs (Fig. 1). A model was developed for stagnant water, and results were compared to those with fluid flow (Higashino et al., 2004). Specific conclusions are as follows: 1. The DO profile varies with time in the water column, but does not change significantly in the sediment. The normalized DO concentration at the sediment/water interface (Cw/CN) is less than 0.1, and is close to zero regardless of the biochemical oxygen uptake rate (m) in the sediment. The normalized DO concentration at the sediment/water interface (Cw/CN) with fluid flow is greater than 0.9 when biochemical oxygen uptake rate is small, i.e. m < 200 mg l1 d1 (Higashino et al., 2004). 2. Since the DO profile near the sediment/water interface, and corresponding DO flux at the sediment/water interface (SOD) varies with time when water is stagnant, the timeaveraged SOD (SODave) over a period of the detention time (T) was introduced (Eq. (11)). The value of SODave is large when the detention time (T) is short, and diminishes as the detention time becomes longer. The SODave decreases to 21, 8, and 4% of that for T ¼ 1 h with increasing detention time, i.e. T ¼ 1 day, 1 week, and 1 month, respectively. 3. The SODave increases as the biochemical oxygen uptake rate (m) becomes larger. But the dependence of SODave on oxygen consumption in the sediment is not as strong as SOD with fluid flow (see Figs. 5, and 8 in Higashino et al., 2004). 4. The boundary layer thickness, i.e. a distance from the sediment/water interface to the point where C/CN ¼ 0.99, gets thicker as the detention time (T) becomes longer, and is independent of the biochemical oxygen uptake rate (m) in the sediment. 5. The DO penetration depth, i.e. a distance from the sediment/water interface to the point where C/CN ¼ 0.001, diminishes with increasing detention time (T). The DO penetration depth decreases as the biochemical oxygen uptake rate (m) becomes larger. The DO penetration depth (ds) is much smaller than that with fluid flow (Higashino et al., 2004). 6. SOD is controlled by oxygen transfer through the diffusive boundary layer and/or by biochemical oxygen uptake in the sediment when flow is present over the sediment surface (Higashino et al., 2004). However, SOD is substantially limited by oxygen transfer in the water column when water is stagnant. 7. When flow over the sediment surface is present, SOD becomes larger than that for stagnant water depending on biochemical oxygen uptake rate (m) in the sediment, i.e. SOD with fluid flow (SODN, when the shear velocity is large) is about 2, 7, 20 and 60 times larger than that for stagnant water (SODave for T ¼ 1day) for m ¼ 10, 100, 1000, and 10,000 mg l1 d1, respectively. 8. Flow over the sediment surface hardly raises SOD much larger than that for stagnant water when biochemical
4388
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 8 1 e4 3 8 9
oxygen uptake rate (m) is small. The difference between SOD with fluid flow and SOD for stagnant water becomes significant as the biochemical oxygen uptake rate (m) becomes larger, i.e. SOD is 10e100 times larger when flow over the sediment surface is present for highly organic sediments.
Acknowledgments This work was supported by the Japan Society for the Promotion of Science (Young Researcher Overseas Visit Program, Number 21-5018), and by JSPS Grant-in-Aid for Scientific Research (22560522). Two anonymous reviewers provided helpful comments and suggestions on the manuscript. The author is grateful to these individuals and organizations for their support.
Appendix. Supplementary data Supplementary data related to this article can be found online at doi:10.1016/j.watres.2011.04.051.
references
Barcelona, M.J., 1983. Sediment oxygen demand fractionation, kinetics and reduced chemical substances. Water Research 17, 1081e1093. Bear, J., 1972. Dynamics of fluids in porous media. American Elsevier Publishing Campany, New York. Boudreau, B.P., Joergensen, B.B. (Eds.), 2001. The Benthic Boundary Layer: Transport Processes and Biogeochemistry. Oxford University Press, UK. Boudreau, B.P., 1997. A one-dimensional model for bed boundary layer particle exchange. Journal of Marine Systems 11, 279e303. Cardenas, M.B., Wilson, J.L., 2006. The influence of ambient groundwater discharge on exchange zones induced by current-bedform interactions. Journal of Hydrology 331, 103e109. Chapra, S.C., 1997. Surface Water Quality Modeling. McGraw Hill. Dade, W.B., 1993. Near-bed turbulence and hydrodynamic control of diffusional mass transfer at the sea floor. Limnology and Oceanography 38 (1), 52e69. Denny, M.W., 1993. Air and Water. Princton University Press, Princton, N.J. Ditoro, D.M., 2001. Sediment Flux Modeling. Wiley, New York. Elliott, A.H., Brooks, N.H., 1997. Transfer of non-sorbing solutes to a streambed with bed forms: theory. Water Resources Research 33 (1), 123e136. Gantzer, C.J., Stefan, H.G., 2003. A model of microbial activity in lake sediments in response to periodic water column mixing. Water Research 37, 2833e2846. Hargrave, B.T., 1972. Aerobic decomposition of sediment and detritus as a function of particle surface area and organic content. Limnology and Oceanography 17, 583e596. Higashino, H.G., Stefan, M., 2011 . Dissolved oxygen demand at the sediment-water interface of a stream: near-bed turbulence and pore water flow effects. Journal of Environmental Engineering, ASCE 137 (7) pp. Higashino, M., Clark, J.J., Stefan, H.G., 2009. Porewater flow due to near-bed turbulence and associate solute transfer in a stream
or lake sediment bed. Water Resources Research 45 (12), W12414. doi:10.1029/2008WR007374. Higashino, M., Gantzer, C.J., Stefan, H.G., 2004. Unsteady diffusional mass transfer at the sediment/water interface: theory and significance for SOD measurements. Water Research 38, 1e12. Hondzo, M., Feyaerts, T., Donovan, R., O’Connor, B.L., 2005. Universal scaling of dissolved oxygen distribution at the sediment-water interface: a power law. Limnology and Oceanography 50 (5), 1667e1676. House, W.A., 2003. Factors influencing the extent and development of the oxic zone in river-bed sediment. Biogeochemistry 63, 317e333. Huettel, M., Webster, I.T., 2001. Porewater flow in permeable sediment. In: Boudreau, B.P., Joergensen, B.B. (Eds.), The Benthic Boundary Layer: Transport Processes and Biogeochemistry. Oxford University Press, UK, pp. 144e179. Iversen, N., Jørgensen, B.B., 1993. Diffusion coefficients of sulfate and methane in marine sediment. Geochimica et Cosmochimica Acta 57, 571e578. Josiam, R., Stefan, H.G., 1999. Effect of flow velocity on sediment oxygen demand: comparison of theory and experiments. Journal of the American Water Resources Association 35 (2), 433e439. Joergensen, B.B., Revsbech, N.P., 1985. Diffusive boundary layers and the oxygen uptake of sediment and detritus. Limnology and Oceanography 30 (1), 111e122. Joergensen, B.B., DesMarais, D.J., 1990. Diffusive boundary layer of sediments: oxygen microgradients over a microbial mat. Limnology and Oceanography 35 (6), 1343e1355. Mackenthun, A., Stefan, H.G., 1998. Effect of flow velocity on sediment oxygen demand: laboratory measurements. Journal of Environmental Engineering, ASCE 12 (3), 222e230. Nakamura, Y., Stefan, H.G., 1994. Effect of flow velocity on sediment oxygen demand: theory. Journal of Environmental Engineering, ASCE 120 (5), 996e1016. O’Connor, B.L., Hondzo, M., 2008. Dissolved oxygen transfer to sediments by sweep and eject motions in aquatic environment. Limnology and Oceanography 53 (2), 566e578. O’Connor, B.L., Hondzo, M., Harvey, J.W., 2008. Incorporating both physical and kinetic limitations in quantifying dissolved oxygen flux to aquatic sediments. Journal of Environmental Engineering, ASCE 135 (12), 1304e1314. Packman, A.I., Salehin, M., Zaramella, M., 2004. Hyporheic exchange with gravel beds: basic hydrodynamic interactions and bedform-induced advective flows. Journal of Hydraulic Engineering, ASCE 130 (7), 647e656. Pamatmat, M.M., 1971. Oxygen consumption by the seabed. 4. Shipboard and laboratory experiments. Limnology and Oceanography 16, 536e550. Qian, Q., Clark, J.J., Voller, V.R., Stefan, H.G., 2009. Depthdependent dispersion coefficient for modeling of vertical solute exchange in a lake bed under surface waves. Journal of Hydraulic Engineering, ASCE 135 (3), 187e197. doi:10.1061/ (ASCE)0733-9429. Rahm, L., Svensson, U., 1989. On the mass transfer properties of the benthic boundary layer with an application to oxygen fluxes. Netherlands Journal of Sea Research 24 (1), 27e35. Roy, H., Huttel, M., Jorgensen, B.B., 2004. Transition of oxygen concentration fluctuations through the diffusive boundary layer overlying aquatic sediments. Limnology and Oceanography 49, 686e692. Smith Jr., K.L., 1978. Benthic community respiration in the N.W. Atlantic Ocean: in situ measurements from 40 to 5200 m. Marine Biology 47, 337e347. Steinberger, N., Hondzo, M., 1999. Diffusional mass transfer at the sediment-water interface. Journal of Environmental Engineering, ASCE 125 (2), 192e200.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 8 1 e4 3 8 9
Nomenclature C: dissolved oxygen concentration (mg l1) CN: dissolved oxygen concentration in the bulk water (mg l1) Cw: dissolved oxygen concentration at the sediment/water interface (mg l1) C*: normalized dissolved oxygen concentration Dw: molecular diffusivity of dissolved oxygen in water (cm2 s1) Ds: effective diffusion coefficient for oxygen in the sediment (cm2 s1) Ko2: half-saturation constant for dissolved oxygen (mg l1) Sc: the Schmidt number SOD: sedimentary oxygen demand (g m2 d1) SODave: time-averaged SOD (g m2 d1) SODflow: SOD with fluid flow (g m2 d1)
4389
SODstagnant: SOD for stagnant water (g m2 d1) T: detention time of water in a lake or a reservoir (h) U*: bed shear velocity (cm s1) y: vertical coordinate (cm) d: sediment layer thickness (cm) dw: boundary layer thickness above the sediment/water interface (cm) ds: dissolved oxygen penetration depth in the sediment (cm) m: maximum oxidation rate (mg l1 d1) m0: zero-order rate constant (mg l1 d1) q: sediment tortuosity r: fluid (water) density (g cm3) s: bed shear stress (N cm2) 4: sediment porosity
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 9 0 e4 3 9 8
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Analysis of the bacterial community in a laboratory-scale nitrification reactor and a wastewater treatment plant by 454-pyrosequencing Lin Ye a, Ming-Fei Shao a, Tong Zhang a,*, Amy Hin Yan Tong b, Si Lok b a b
Environmental Biotechnology Laboratory, The University of Hong Kong, Hong Kong SAR, China Genome Research Center, The Li Ka Shing Faculty of Medicine, The University of Hong Kong, Hong Kong SAR, China
article info
abstract
Article history:
For full understanding of the microbial community in the wastewater treatment bioreac-
Received 25 January 2011
tors, one of the feasible and effective ways is to investigate the massive genetic informa-
Received in revised form
tion contained in the activated sludge. In this study, high-throughput pyrosequencing was
20 March 2011
applied to analyze the 16S rRNA gene of bacteria in a laboratory-scale nitrification reactor
Accepted 22 May 2011
and a full-scale wastewater treatment plant. In total, 27,458 and 26,906 effective sequence
Available online 31 May 2011
reads of the 16S rRNA gene were obtained from the Reactor and the wastewater treatment plant activated sludge samples respectively. The taxonomic complexities in the two
Keywords:
samples were compared at phylum and genus levels. According to the pyrosequencing
Activated sludge
results, even for a laboratory-scale reactor as simple as that in this study, a small size clone
Bacterial community
library is far from enough to reflect the whole profile of the bacterial community. In
Cloning
addition, it was found that the commonly used informatics tool “RDP classifier” may
454 High-throughput
drastically assign Nitrosomonas sequences into a wrong taxonomic unit resulting in
pyrosequencing
underestimation of ammonia-oxidizing bacteria in the bioreactors. In this paper the reasons for this mistakenly assignment were analyzed and correction methods were proposed. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
Nitrification is an important step for removing ammonium nitrogen from wastewater. The key role that bacterial communities play in wastewater treatment has been intensively studied in the past decades in laboratory-scale and fullscale bioreactors by the use of various molecular methods. PCR-DGGE (polymerase chain reaction - denaturing gradient gel electrophoresis) and T-RFLP (terminal restriction fragment length polymorphism) were used to investigate the diversity of the bacterial communities of activated sludge from different wastewater treatment plants (WWTPs) (Boon et al., 2002; Regan et al., 2002). More recently, analysis of clone
library (Matsumoto et al., 2009) as well as direct visualization of bacterial species by FISH (fluorescent in situ hybridization) (Hao et al., 2009) were employed to determine the community compositions. Although bacterial species involved in nitrification have also been characterized by the aforementioned approaches, the extraordinary diversity of microorganisms in the activated sludge exceeds the sensitivity and dynamic range of those molecular methods and precludes a complete characterization of the interplay of various components of the microbial community in nitrification. Pyrosequencing developed by Roche 454 Life Science (Branford, CT, USA) is a high-throughput analytical method that can generate huge amounts of DNA reads through
* Corresponding author. Fax: þ86 852 2559 5337. E-mail address:
[email protected] (T. Zhang). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.05.028
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 9 0 e4 3 9 8
a massively parallel sequencing-by-synthesis approach (Margulies et al., 2005). This technology have been used widely to analyze the microbial community in various environmental samples, such as marine water (Qian et al., 2011), soil (Roesch et al., 2007), human distal intestine (Claesson et al., 2009), wastewater treatment plant influent (McLellan et al., 2010) et al.. However, few studies have been conducted on activated sludge by this method. Kwon and colleagues (Kwon et al., 2010) investigated the microbial diversity in an integrated fixed-film activated sludge system. Their results showed the bacterial abundances were quite high, totally 3034 and 1451 operational taxonomic units (OTUs) were identified at the 3% cutoff for the suspended and attached samples, respectively. In the present study, we characterized and compared the bacterial communities in a laboratory-scale nitrification reactor with that from activated sludge of a wastewater treatment plant by 454-pyrosequencing. The diversity and abundance of the nitrifiers in these two samples were also investigated. Additionally, we found that RDP Classifier, a commonly used informatics tool for pyrosequencing data analysis, may have phylogenetically assigned Nitrosomonas sequences into a wrong order. The key nucleotides leading to the mistaken assignment have been identified and a method to overcome this problem was proposed based on the BLAST results.
4391
added to the influent to get the ammonia nitrogen concentration of 200 mg/L, In addition, 20 mg/L of KH2PO4 was added into the influent to provide sufficient phosphorus for the growth of microorganisms in the Reactor. Without adding organic matter, the total organic carbon (TOC) of the influent was as low as 0.64 0.05 mg/L. The hydraulic retention time (HRT) of the Reactor was 18.4 h. The Reactor was shielded with aluminum foil to avoid exposure to light. Prior to the present study, the Reactor was ran continuously for more than 500 days to investigate the ammonia-oxidizing bacteria (AOB) and nitrite-oxidizing bacteria (NOB) under different conditions (Jin et al., 2010; Ye and Zhang, 2010). Shatin WWTP is a full-scale wastewater treatment plant in Hong Kong. This WWTP treats saline sewage (salinity 1.2%) with a four-stage process (anoxiceaerobiceanoxiceaerobic) that may simultaneously remove organic compounds and nitrogen. The seed sludge of the Reactor described above was taken from Tank No.16 of the first aerobic stage. Activated sludge sample used to perform 454-pyrosequencing analysis was also taken from the same tank.
2.2.
Chemical analysis
2.
Materials and methods
Concentrations of ammonium, nitrite and nitrate were measured according to the Standard Methods (Eaton and Franson, 2005) by Nesslerizaion Method, Colorimetric Method and Ultraviolet Spectrophotometric Screening Method, respectively.
2.1.
Reactor operation and WWTP description
2.3.
In the present study, a fermentor (Sartorius Biostat A plus) (Goettingen, Germany) with a working volume of 2.6 L was configured for continuous operation to conduct nitrification studies. A pH value of 7.5 was held by automated addition of sodium bicarbonate. Dissolved Oxygen (DO) was maintained at 0.5 mg/L by stirring and aeration. The influent was made with deionized water (67%) and seawater (33%) to simulate the typical salinity of sewage found in Hong Kong. NH4Cl was
DNA extraction and PCR
Sludge samples of Day 165, 178, 190, and 201 (as indicated by the inverted triangle in Fig. 1) were taken from the Reactor for DNA extraction using FastDNA SPIN Kit for Soil (MP Biomedicals, Illkirch, France). For clone library construction, the above DNA mixture was pooled and amplified by PCR using primer set EUB8F (50 -AGAGTTTGATCMTGGCTCAG-30 ) (Heuer et al., 1997) and UNIV1392R (50 -ACGGGCGGTGTGTRC30 ) (Ferris et al., 1996). 30 ml PCR mixture contained 0.2 ml of Ex
Fig. 1 e Nitrogen concentration in the influent and effluent (The 4 inverted open triangles indicate the time points of sludge samples for cloning and pyrosequencing analysis).
4392
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 9 0 e4 3 9 8
Taq TM(TaKaRa, Dalian, China), 3 ml of 10 Ex Taq Buffer, 3 ml of dNTP mixture, 0.2 mM of each primer, and 20e50 ng of genomic DNA. The thermocycling steps were as follows: 95 C for 7 min, followed by 35 cycles at of 95 C for 1 min, 55 C for 1 min, 72 C for 1 min and a final extension step at 72 C for 10 min. For pyrosequencing, the above DNA mixture of the four samples was amplified with a set of primers targeting the hypervariable V4 region of the 16S rRNA gene (RDP’s Pyrosequencing Pipeline: http://pyro.cme.msu.edu/pyro/help.jsp). The forward primer is 50 -AYTGGGYDTAAAGNG-30 and the reverse primers are the mixture of four primers, i.e. 50 TACCRGGGTHTCTAATCC-30 , 50 -TACCAGAGTATCTAATTC-30 , 50 -CTACDSRGGTMTCTAATC-30 , and 50 -TACNVGGGTATCTAATCC-30 (Claesson et al., 2009). Barcodes that allow sample multiplexing during pyrosequencing were incorporated between the 454 adapter and the forward primers.
2.4.
Cloning
PCR products were purified using PCRquick-spinTM PCR Product Purification Kit (iNtRON Biotechnology, SangdaewonDong, Korea). The purified PCR products were cloned using the InsTAclone PCR Cloning Kit (Fermentas, Burlingtong, Ontario, Canada) following the instructions of the vendor. White colonies were selected for whole-cell PCR amplification with the M13F (50 -TGTAAAACGACGGCCAGT-30 ) and M13R (50 CAGGAAACAGCTATGAC-30 ) primer set. The PCR products were purified and sequenced on the ABI 3730xl capillary sequencer (Applied Biosystems, Foster City, CA, USA) using M13F or M13R primers. The clone library sequences in this study have been deposited in GenBank under accession numbers HM117160 to HM117171.
2.5.
qblast function in Biopython to run BLAST and search against “nr” database through the internet automatically for all the above-mentioned extracted sequences; 3) parse the BLAST results to check the top 10 hits whether there is a hit containing Nitrosomonas or Comamonas in the title; and 4) record the maximum identities between the query sequences and the subject sequences identified in last step. On the other hand, all sequences obtained from pyrosequencing in this study were compared with Greengenes 16S rRNA gene database (DeSantis et al., 2006) using NCBI’s BLASTN tool (Altschul et al., 1990) and the default parameters except for the maximum hit number of 100 (Claesson et al., 2009). Then the sequences were assigned to NCBI taxonomies with MEGAN (Huson et al., 2007) by using the Lowest Common Ancestor (LCA) algorithm and the default parameters, i.e. absolute cutoff: BLAST bitscore 35, and relative cutoff: 10% of the top hits.
3.
Results and discussion
3.1.
Reactor performance
Prior to this study, the Reactor was operated under a very low oxygen concentration condition (DO 0.15 mg/L). Accordingly, the bulk of the ammonium was partially oxidized to nitrite. In present study, the DO level was increased to 0.5 mg/L, the nitrite was gradually reduced and several molecular methods (DGGE, T-RFLP and Cloning) have been used to confirm that nitrite-oxidizing bacteria (Nitrospira) proliferated intensively in this period (Ye and Zhang, 2010). The operational condition and the performance of the Reactor were described in detail in our previous paper (Ye and Zhang, 2010).
High-throughput 454 pyrosequencing 3.2.
The composition of the PCR products of V4 region of 16S rRNA gene was determined by pyrosequencing using the Roche 454 FLX Titanium sequencer (Roche 454 Life Sciences, Branford, CT, USA). Samples in this study were individually barcoded to enable multiplex sequencing. The results are deposited into the NCBI short reads archive database (Accession Number: SRA026842.2).
2.6.
Sequence analysis and phylogenetic classification
Following pyrosequencing, Python scripts were written to: 1) remove sequences containing more than one ambiguous base (‘N’); 2) check the completeness of the barcodes and the adapter; 3) remove sequences shorter than 150 bps. The “RDP Align” tool in RDP’s Pyrosequencing Pipeline was used to align the effective sequences. A cluster file was generated with “RDP Complete Linkage Clustering” tool. From the cluster file, the rarefaction curve was generated using the “RDP Rarefaction” tool. Taxonomic classification of the sequences was performed using the RDP Classifier (Version 2.2) with a set confidence threshold of 50%. Python and Biopython (Cock et al., 2009) were used to create scripts to: 1) extract all sequences that were assigned into the order of Burkholderiales according to the result of RDP Classifier (the downloaded assignment detail text file); 2) use
Cloning results
Twelve OTUs were obtained with 3% nucleotide cutoff from 61 clones that were sequenced in 16S rRNA gene clone library. According to RDP Classifier, the 61 sequences examined by Sanger-dideoxy based sequencing can be assigned to 3 phyla. Based on both RDP Classifier and BLAST analysis (Table 1), OTU-2 and OTU-4 are Nitrosomonas and Nitrospira species, which accounted for 21.3% and 3.2% in total bacterial community, respectively, suggesting that these species represent the dominant AOB and NOB in the Reactor. It should be noted that according to the BLAST results, the most dominant OTU, OTU-1, is probably a heterotrophic species that is similar (max identity 99%) to an uncultured bacteria reported in the marine sediment.
3.3.
Taxonomic complexity of the bacterial community
Pyrosequencing of the Reactor sample and the WWTP sludge sample yielded 27,458 and 26,906 effective sequence tags respectively. The amount of sequences was comparable to those of other studies (Kwon et al., 2010; Lee et al., 2010; McLellan et al., 2010). RDP Classifier was firstly used to assign these sequence tags into different phylogenetic bacterial taxa. Fig. 2 and Supplementary Table S1 show the relative bacterial community abundances on the phylum level. Except
4393
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 9 0 e4 3 9 8
Table 1 e The affiliation and closest match of the bacterial OTUs. OTU 1 2 3 4 5 6 7 8 9 10 11 12
Clones
Percentage
Genus assignment based on RDP classifier [Probability]
Closest match from BLAST [Max Identity]
34 13 3 2 2 1 1 1 1 1 1 1
55.7% 21.3% 4.9% 3.2% 3.2% 1.6% 1.6% 1.6% 1.6% 1.6% 1.6% 1.6%
Coxiella [35%] Nitrosomonas [56%] Oleiphilus [10%] Nitrospira [100%] Loktanella [40%] Aminobacter [56%] Azoarcus [56%] Muricauda [100%] Adhaeribacter [31%] Roseivirga [74%] Roseivirga [45%] Marinicola [21%]
Uncultured bacterium [99%] Nitrosomonas sp. [99%] Uncultured geproteobacterium [99%] Uncultured Nitrospira sp. [98%] Uncultured bacterium [94%] Uncultured bacterium [94%] Denitromonas indolicum [97%] Muricauda sp. [98%] Uncultured bacterium [91%] Uncultured bacterium [98%] Uncultured bacterium [97%] Uncultured bacterium [91%]
for a small number of minor phyla accounting for no more than 0.6% of total community found only in WWTP sludge, the numbers of tabulated phylum present in the Reactor and the WWTP sludge were nearly identical. Proteobacteria, Firmicutes and Bacteroidetes were three phyla that were abundant in both samples. These three phyla were also ubiquitous in soil (Fierer et al., 2007). However, Actinobacteria and Chloroflexi were much less abundant in the Reactor than in the WWTP. Heterotrophic bacteria in these phyla may be depleted in the Reactor under such an oligotrophic environment. It was also observed that Nitrospira phylum in the Reactor was significantly more abundant than that in the WWTP sludge, suggesting that the elevated level of nitrite and oxygen present in the Reactor may have favored the propagation of these nitrite-oxidizing bacteria. Notably, although the organic matter in the influent of the nitrification reactor was very low, the heterotrophic bacteria were still dominant over the autotrophic bacteria. The carbon source of these heterotrophic bacteria were probably from soluble microbial products (SMP) in the Reactor (Barker and Stuckey, 1999; Rittmann and McCarty, 2001). 55%
454-Pyprosequencing provides at least three logs or more sensitivity over conventional Sanger-dideoxy based sequencing for assessing microbial diversity. Fig. 3 showed the diversity of bacteria in the Reactor was significantly reduced after 500 days’ operation compared with the seed sludge from WWTP. The diversity reduction from seeding sludge to lab-scale reactor was usually investigating by DGGE and cloning previously (Liu et al., 2002). Heterotrophic bacteria were greatly depleted in the Reactor. Especially, the phylum Actinobacteria, where most members are heterotrophs (Servin et al., 2008), were dramatically reduced (Fig. 2). While the absence of light under Reactor conditions avoids perturbing the growth of the nitrifiers (Sinha and Annachhatre, 2007), light dependent bacteria were depleted. A marked decrease of the phototropic Chloroflexi phylum (Holt et al., 1994) (Fig. 2) in the Reactor was observed compared with the seed sludge. In order to further compare the microbial communities of the two samples from the Reactor and WWTP, all-against-all comparison was conducted by using the MEGAN software. The sequences in each of the samples were normalized before doing the comparison. The tree created by MEGAN was shown in Fig. 4. The pie charts beside the leaves of the tree indicate the relative abundance of the genus in the two samples. From
50%
Reactor WWTP
45% 40%
3500
Percentage
35% 30%
3% Reactor 5% Reactor 3% WWTP 5% WWTP
3000
25% 2500
20%
OTUs
15% 10%
2000 1500
5% 1000
0% i us ia e M7 lex es ria tes etes pira e ria etes eria b ia f u te et o te T t t rm yd s ro ac Bac m ic r oid itro bac my c ba c ic r h e ha o m l b c r la e o co m s -T N ino c to Fi e o i ed Ch piro ct id t u Ch ot f n S Ba Ac erru c c Ac Pl a Pr ss i o V a c l o c in e Un D ria
Phylum name
Fig. 2 e Bacterial community compositions at phylum level revealed by pyrosequencing.
500 0 0
5000
10000
15000
20000
25000
30000
Number of sequences
Fig. 3 e Rarefaction curves of OTUs defined by 3% and 5% distances in Reactor and WWTP sludge samples.
4394
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 9 0 e4 3 9 8
Fig. 4 e Sequences from the Reactor and WWTP assigned into NCBI taxonomies with BLAST and MEGAN. (Pie charts indicate the relative abundance for each genus. The ratio of gray color area to dark color area in each pie represented the ratio of the relative abundance of the corresponding genus in WWTP to that in the Reactor.)
4395
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 9 0 e4 3 9 8
Fig. 4, it could be seen that at the genus level the microbial communities of the Reactor and the WWTP were quite different. There were some genera (such as Marinobacter, Pseudomonas, Aequorivita, Muricauda, etc.) appearing only in the Reactor. According to the previous reports (Bowman and Nichols, 2002; Gauthier et al., 1992; Yoon et al., 2005), these genera are usually halotolerant bacteria and exist in the marine environment. These bacteria may come from the seawater of the influent and could adapt themselves to the conditions of the Reactor. Many other genera, which were marked by gray color in Fig. 4, exist only in the WWTP. That indicates the diversity of the bacteria in the WWTP was much more complex than that in the Reactor. Also, there are some genera, including Phyllobacteriaceae, Rhodobacteraceae, Chromatiales, Comamonadaceae, Nitrospira, Rhodococcus, etc., exist both in the Reactor and the WWTP. According to the cluster files produced by the “RDP Complete Linkage Clustering” tool, there were 494 and 381 OTUs in Reactor sludge using two cutoffs levels of 3% and 5%, respectively. By contrast, WWTP sludge had 1986 and 1648 OTUs using the same cutoffs. According to the rarefaction curve (Fig. 3), the species complexity in the Reactor was sixfold less than that in WWTP. Judging from the numbers of OTUs obtained by pyrosequencing in this study, not to mention the WWTP, even for a laboratory reactor as simple as that in our study, a small scale clone library was not sufficient to reflect the whole profile of the bacterial community, especially for those minor populations. Furthermore, it can be deduced that the commonly used molecular methods (such as DGGE and T-RFLP) in environmental biotechnology may also have insufficient resolutions to characterize the microbial communities in the wastewater treatment bioreactors. High-throughput sequencing methods have the potential to be effective means for better understanding of the microorganisms in various environmental engineering facilities.
3.4.
Diversity and abundances of AOB and NOB
It was found that in the results of RDP Classifier (Table 2), the sequences that were assigned into Nitrosomonadales order were quite few. For the Reactor sludge sample, only 0.65% of the sequences were classified into this order, which was inconsistent with the limited results from Sangerdideoxy sequencing of the clone library. Such low abundance of AOB also conflicted with the performance (high ammonium removing rate) of the Reactor. A further check showed that there were large numbers of Nitrosomonadales
sequences that were wrongly assigned by the RDP Classifier into the order of Burkholderiales, a neighbor of Nitrosomonadales in Proteobacteria phylum. Most of these misassigned sequences were closely related (identity>97%) to Nitrosomonas species as shown by BLAST analysis. Following reclassification, 15.54% of the sequences from the reactor sludge sample were Nitrosomonas related, indicating that Nitrosomonas-like AOB were remarkably enriched in the reactor (Table 2). In both the Reactor and the WWTP sludge, the dominant AOB species was Nitrosomonas and the dominant NOB species was Nitrospira, which were affiliated to Nitrosomonadales and Nitrospirales order, respectively. This result was consistent with the previous reports of activated sludge from other researchers (Layton et al., 2005; Logemann et al., 1998). Except Nitrosomonas and Nitrospira, the other AOB and NOB genus were very rare in the reactor. Nitrosospira, another genus belonging to b-Proteobacteria AOB, accounted for only 0.6% and 0.056% in the reactor and the WWTP sludge samples, respectively. Only one sequence and seven sequences of Nitrosococcus were found in 27,458 and 26,906 sequences in the Reactor and WWTP sludge samples, respectively. For NOB, only 13 Nitrobacter sequences were found in the WWTP sludge and none was found in the reactor. The present results suggested that except Nitrosomonas and Nitrospira, all other species of bacterial nitrifiers play only a very small role in nitrification process in the wastewater treatment reactors.
3.5.
Mistakenly classified Nitrosomonas sequences
In this study, as aforementioned, RDP Classifier may wrongly assign sequences in Nitrosomonadales order into another order, consequently, leading to an underestimation of the abundance of AOB in the samples. Except for the samples in this study, we also investigated the pyrosequencing results of other 14 samples from WWTPs in another study, a lot of sequences were mistakenly assigned to Burkholderiales. Thirty sequences, including 10 from mistakenly assigned Nitrosomonas sequences, 10 from correctly assigned Nitrosomonas sequences, and 10 from Comamonas (a genus in Burkholderiales) sequences, were selected from the sequences obtained by the pyrosequencing, and used to draw a Neighbor-joining phylogenetic tree using Jukes-Cantor model (Fig. 5). It was found that the Nitrosomonas sequences and Comamonas sequences can be clearly classified into different groups, indicating that there are marked differences
Table 2 e Relative abundance of dominant AOB and NOB in the Reactor and WWTP sludge. Sample
Reactor WWTP
AOB/NOB
Nitrosomonadales Nitrospirales Nitrosomonadales Nitrospirales
RDP Classifier
Clone Library
BLAST-corrected RDP Classifier
MEGAN Assignment
Reads
%
Reads
%
Reads
%
Reads
%
179 1782 14 273
0.65% 6.49% 0.05% 1.01%
13 2 e e
21.3% 3.2% e e
4059 1782 14 273
14.78% 6.49% 0.05% 1.01%
4268 1817 13 273
15.54% 6.61% 0.05% 1.01%
4396
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 9 0 e4 3 9 8
Nitro_M-GPQYMAQ01D2WQ6 Nitro_M-GPQYMAQ01CPALL Nitro_M-GPQYMAQ01EE6QD Nitro_M -GPQYMAQ01DCDM8 Nitro_M-GPQYMAQ01BBFHN 40
I
Nitro_M-GPQYMAQ01B7TNH Nitro_M-GPQYMAQ01D96JO Nitro_M-GPQYMAQ01A2KAW
43
Nitro_M-GPQYMAQ01EIO8Z Nitro_M-GPQYMAQ01EI4S3
72
Nitro-GPQYMAQ01A9K6V Nitro-GPQYMAQ01E0QA6 Nitro-GPQYMAQ01C68KE Nitro-GPQYMAQ01DEH7U Nitro-GPQYMAQ01CCBG2
90
Nitro-GPQYMAQ01AHLH1
98
II
Nitro-GPQYMAQ01AYECJ
60
Nitro-GPQYMAQ01CXQ2B
86
67 Nitro-GPQYMAQ01BQZAW 31 75 82
Nitro-GPQYMAQ01BEAJB Comam-GPQYMAQ01EM28M Comam-GPQYMAQ01CGVBO
Comam-GPQYMAQ01BYP7R Comam-GPQYMAQ01EUM1H
100
Comam-GPQYMAQ01C0W9M 62
III
Comam-GPQYMAQ01ELIAF
33
Comam-GPQYMAQ01B2DQT
33
Comam-GPQYMAQ01DJK4R
25 Comam-GPQYMAQ01DPQIB
Comam-GPQYMAQ01AYTQ6
0.01
Fig. 5 e Neighbor-joining phylogenetic tree using Jukes-Cantor model of Nitrosomonas sequences and Comamonas sequences based on V4 region of 16S rRNA gene sequences (I - Mistakenly assigned Nitrosomonas sequences, II e Correctly assigned Nitrosomonas sequences, III e Comamonas sequences).
Fig. 6 e The key nucleotides that caused the mistakenly assigned Nitrosomonas sequences (I - Mistakenly assigned Nitrosomonas sequences, II e Correctly assigned Nitrosomonas sequences).
in the sequences of these two groups. It indicates the possible unrecognized deficiencies of the RDP Classifier. Further examination of these sequences reveals that the correct assignment of Nitrosomonas related sequences into the Nitrosomonas genus is dependent on a key dinucleotide position, as show in Fig. 6. If the ‘GC’ dincucleotide at position 106 of group I was changed to ‘AT’, as it is found in group II, the RDP Classifier would then assign group I correctly into Nitrosomonas. It is therefore advised that the phylogenetic results of Nitrosomonas species obtained using the RDP Classifier should be cross-validated by other independent tools such as Greengenes’ classification tool (DeSantis et al., 2006) and GAST (Huse et al., 2008). Accordingly, we have developed a batch BLAST method, which has been described in the materials and methods part, to confirm the suspicious sequences.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 9 0 e4 3 9 8
4.
Conclusions
The diversity of the bacterial community in the nitrification reactor was significantly reduced compared with seeding sludge from the WWTP after 500 days’ operation. While, the bacteria both in the reactor and the seeding sludge distributed almost over the same phyla. RDP classifier is a powerful tool for pyrosequencing data analysis but it appears to misassign Nitrosomonas sequences into wrong taxonomic rank. Some other tools, such as Blast and Greengenes classification tool, can be used to correct the results of the RDP classifier. We also developed a batch Blast method to confirm the suspicious sequences. According to the pyrosequencing results, for such a reactor, a small scale clone library is not enough to reflect the profile of the bacterial community. Although the influent of the nitrification reactor contained nearly no organic matter, the heterotrophic bacteria were still dominant and much more than autotrophic bacteria in the reactor.
Acknowledgments Dr. Ming-Fei Shao thanks HKU for the postdoctoral fellowship. Lin Ye thanks HKU for the postgraduate studentship. We also would like to thank Hong Kong General Research Fund (HKU7197-08E) for financial support of this study and W Chan and CK Wong for technical help in pyrosequencing.
Appendix. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.watres.2011.05.028.
references
Altschul, S., Gish, W., Miller, W., Myers, E., Lipman, D., 1990. Basic local alignment search tool. J. Mol. Biol. 215, 403e410. Barker, D.J., Stuckey, D.C., 1999. A review of soluble microbial products (SMP) in wastewater treatment systems. Water Res. 33, 3063e3082. Boon, N., Windt, W., Verstraete, W., Top, E., 2002. Evaluation of nested PCReDGGE (denaturing gradient gel electrophoresis) with group-specific 16S rRNA primers for the analysis of bacterial communities from different wastewater treatment plants. FEMS Microbiol. Ecol. 39, 101e112. Bowman, J., Nichols, D., 2002. Aequorivita gen. nov., a member of the family Flavobacteriaceae isolated from terrestrial and marine Antarctic habitats. Int. J. Syst. Evol. Microbiol. 52, 1533. Claesson, M., O’Sullivan, O., Wang, Q., Nikkila, J., Marchesi, J., Smidt, H., De Vos, W., Ross, R., O’Toole, P., 2009. Comparative analysis of pyrosequencing and a phylogenetic microarray for exploring microbial community structures in the human distal intestine. PloS One 4, e6669. Cock, P., Antao, T., Chang, J., Chapman, B., Cox, C., Dalke, A., Friedberg, I., Hamelryck, T., Kauff, F., Wilczynski, B., 2009. Biopython: freely available Python tools for computational molecular biology and bioinformatics. Bioinformatics 25, 1422.
4397
DeSantis, T., Hugenholtz, P., Larsen, N., Rojas, M., Brodie, E., Keller, K. , Huber, T., Dalevi, D., Hu, P., Andersen, G., 2006. Greengenes, a chimera-checked 16S rRNA gene database and workbench compatible with ARB. Appl. Environ. Microbiol. 72, 5069. Eaton, A., Franson, M., 2005. Standard methods for the examination of water & wastewater. Am Public Health Assn. Ferris, M., Muyzer, G., Ward, D., 1996. Denaturing gradient gel electrophoresis profiles of 16S rRNA-defined populations inhabiting a hot spring microbial mat community. Appl. Environ. Microbiol. 62, 340e346. Fierer, N., Bradford, M.A., Jackson, R.B., 2007. Toward an ecological classification of soil bacteria. Ecology 88, 1354e1364. Gauthier, M., Lafay, B., Christen, R., Fernandez, L., Acquaviva, M., Bonin, P., Bertrand, J., 1992. Marinobacter hydrocarbonoclasticus gen. nov., sp. nov., a new, extremely halotolerant, hydrocarbon-degrading marine bacterium. Int. J. Syst. Evol. Microbiol. 42, 568. Hao, X., Wang, Q., Zhang, X., Cao, Y., Mark Loosdrecht, C.M., 2009. Experimental evaluation of decrease in bacterial activity due to cell death and activity decay in activated sludge. Water Res. 43, 3604e3612. Holt, J., Krieg, N., Sneath, P., Staley, J., Williams, S., 1994. Bergey’s Manual of Systematic Bacteriology, nineth ed. Williams and Wilkins. Heuer, H., Krsek, M., Baker, P., Smalla, K., Wellington, E., 1997. Analysis of actinomycete communities by specific amplification of genes encoding 16S rRNA and gelelectrophoretic separation in denaturing gradients. Appl. Environ. Microbiol. 63, 3233e3241. Huse, S., Dethlefsen, L., Huber, J., Welch, D., Relman, D., Sogin, M., 2008. Exploring microbial diversity and taxonomy using SSU rRNA hypervariable tag sequencing. PLoS Genet. 4, e1000255. Huson, D., Auch, A., Qi, J., Schuster, S., 2007. MEGAN analysis of metagenomic data. Genome Res. 17, 377e386. Jin, T., Zhang, T., Yan, Q., 2010. Characterization and quantification of ammonia-oxidizing archaea (AOA) and bacteria (AOB) in a nitrogen-removing reactor using T-RFLP and qPCR. Appl. Microbiol. Biotechnol. 87, 1167e1176. Kwon, S., Kim, T., Yu, G., Jung, J., Park, H., 2010. Bacterial community composition and diversity of a full-scale integrated fixed-film activated sludge system as investigated by pyrosequencing. J. Microbiol. Biotechnol. 20, 1717e1723. Layton, A., Dionisi, H., Kuo, H., Robinson, K., Garrett, V., Meyers, A., Sayler, G., 2005. Emergence of competitive dominant ammonia-oxidizing bacterial populations in a fullscale industrial wastewater treatment plant. Appl. Environ. Microbiol. 71, 1105. Lee, T.K., Van Doan, T., Yoo, K., Choi, S., Kim, C., Park, J., 2010. Discovery of commonly existing anode biofilm microbes in two different wastewater treatment MFCs using FLX Titanium pyrosequencing. Appl. Microbiol. Biotechnol., 1e9. Liu, W.T., Chan, O.C., Fang, H.H.P., 2002. Microbial community dynamics during start-up of acidogenic anaerobic reactors. Water Res. 36, 3203e3210. Logemann, S., Schantl, J., Bijvank, S., Loosdrecht, M., Kuenen, J., Jetten, M., 1998. Molecular microbial diversity in a nitrifying reactor system without sludge retention. FEMS Microbiol. Ecol. 27, 239e249. Margulies, M., Egholm, M., Altman, W., Attiya, S., Bader, J., Bemben, L., Berka, J., Braverman, M., Chen, Y., Chen, Z., 2005. Genome sequencing in microfabricated high-density picolitre reactors. Nature 437, 376e380. Matsumoto, S., Katoku, M., Saeki, G., Terada, A., Aoi, Y., Tsuneda, S., Picioreanu, C., van Loosdrecht, M., 2009. Microbial community structure in autotrophic nitrifying granules characterized by experimental and simulation analyses. Environ. Microbiol. 12, 192e206.
4398
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 9 0 e4 3 9 8
McLellan, S., Huse, S., Mueller Spitz, S., Andreishcheva, E., Sogin, M., 2010. Diversity and population structure of sewage derived microorganisms in wastewater treatment plant influent. Environ. Microbiol. 12, 378e392. Qian, P., Wang, Y., Lee, O., Lau, S., Yang, J., Lafi, F., AlSuwailem, A., Wong, T., 2011. Vertical stratification of microbial communities in the Red Sea revealed by 16S rDNA pyrosequencing. ISME J. 5, 507e518. Regan, J.M., Harrington, G.W., Noguera, D.R., 2002. Ammonia-and nitrite-oxidizing bacterial communities in a pilot-scale chloraminated drinking water distribution system. Appl. Environ. Microbiol. 68, 73e81. Rittmann, B., McCarty, P., 2001. Environmental Biotechnology: Principles and Applications. McGraw-Hill, New York. Roesch, L., Fulthorpe, R., Riva, A., Casella, G., Hadwin, A., Kent, A., Daroub, S., Camargo, F., Farmerie, W., Triplett, E., 2007.
Pyrosequencing enumerates and contrasts soil microbial diversity. ISME J. 1, 283e290. Servin, J., Herbold, C., Skophammer, R., Lake, J., 2008. Evidence Excluding the Root of the tree of Life from the Actinobacteria. Mol. Biol. Evol. 25, 1e4. Sinha, B., Annachhatre, A., 2007. Partial nitrificationdoperational parameters and microorganisms involved. Rev. Environ. Sci. Biotechnol. 6, 285e313. Ye, L., Zhang, T., 2010. Estimation of nitrifier abundances in a partial nitrification reactor treating ammonium-rich saline wastewater using DGGE, T-RFLP and mathematical modeling. Appl. Microbiol. Biotechnol. 88 (6), 1403e1412. Yoon, J., Lee, M., Oh, T., Park, Y., 2005. Muricauda flavescens sp. nov. and Muricauda aquimarina sp. nov., isolated from a salt lake near Hwajinpo Beach of the East Sea in Korea, and emended description of the genus Muricauda. Int. J. Syst. Evol. Microbiol. 55, 1015.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 9 9 e4 4 0 8
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Characterization of bottom sediments in lakes using hydroacoustic methods and comparison with laboratory measurements Michael A. Anderson*, Porfirio Pacheco Department of Environmental Sciences, University of California, Riverside, CA 92521, USA
article info
abstract
Article history:
The acoustical properties of bottom sediments in two lakes were shown to be strongly
Received 29 December 2010
correlated with clay content, organic C and total N concentrations, and other important
Received in revised form
sediment properties. The fractal dimension of the bottom echo was more strongly corre-
28 April 2011
lated with sediment physical and chemical properties than energy-based measures. The
Accepted 22 May 2011
fractal dimension was also related to rates of PO4-P and NH4-N release from intact sedi-
Available online 31 May 2011
ment cores and sediment oxygen demand. Measurements made at 430-kHz were more sensitive to differences in sediment properties than 201- or 38-kHz. Hydroacoustic
Keywords:
measurements allow rapid assessment of properties important in lake restoration and
Hydroacoustic methods
water resource management. ª 2011 Elsevier Ltd. All rights reserved.
Sediment properties Fractal dimension
1.
Introduction
Bottom sediments are a critical component of lake and reservoir systems. Bottom sediments serve as habitat for benthic invertebrates (Peeters et al., 2004), influence macrophyte distribution (Duarte and Kalff, 1986), regulate nutrient recycling rates in lakes (Sondegaard et al., 2003), control concentrations of dissolved oxygen (DO), H2S and other constituents in bottom waters (Hatcher, 1986; Reese et al., 2008), accumulate contaminants such as metals, pesticides and other hydrophobic organic compounds (Baudo et al., 1989; Forstner and Wittmann, 1979; Karickhoff et al., 1979), and provide a record of past conditions in the lake, airshed, watershed and climate (Lehman, 1975; Kirby et al., 2007). Sediment properties can vary widely over space and time, however. Wave action tends to limit deposition of fine sediments and leave coarse-textured sediments in place in shallower waters, while fine organic sediments are often focused
into deeper regions of a lake or reservoir basin (Lehman, 1975; Hakanson and Jansson, 1983; Anderson et al., 2008). Organic sediments are generally zones of intense microbial activity, rapidly depleting DO concentrations, promoting denitrification, reduction of Fe and Mn oxyhydroxides and often resulting in sulfate reduction or methanogenesis (Hargrave, 1972; Reese et al., 2008). Reducing conditions often result in very high rates of release of phosphate and ammonium from bottom sediments (Holdren and Armstrong, 1980). In contrast, coarser textured sediments that are lower in organic carbon content support much slower rates of sediment oxygen demand and internal nutrient recycling (Hargrave, 1972). The capacity for carboxylic acid and other functional groups on organic matter to strongly bind metals can result in enrichment of organic sediments to very high concentrations of metals relative to the water column (Forstner and Wittmann, 1979). The potential for toxicity effects on benthic invertebrates and other organisms has led to development of sediment
* Corresponding author. Tel.: þ1 951 827 3757; fax: þ1 951 827 3993. E-mail address:
[email protected] (M.A. Anderson). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.05.029
4400
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 9 9 e4 4 0 8
quality criteria, and high contaminant concentrations can also lead to management and disposal issues. At the same time, sediments enriched in organic carbon exhibit very high partitioning coefficients for hydrophobic organic contaminants (Karickhoff et al., 1979), potentially leading to food-chain transfer and biomagnification (Lundgren et al., 2002). Understanding the properties and distribution of sediments in a lake or reservoir, then, can be critical to management and restoration. The distribution of sediments and their properties are typically determined using sediment grab samples or cores from a finite number of sites. The numbers of sites sampled are generally limited, since such sampling is time-consuming, and thus often fails to capture the detailed distribution and heterogeneities present in the lake (Downing and Rath, 1988). Models have been developed to understand sediment distribution in lake basins. For example, Hakanson (1982) developed the dynamic ratio model, where lake area and mean depth could be used to estimate the area of erosion and transport, and therefore also the area of deposition. Rowan et al. (1992) used wave theory to predict the depth that separates the erosional zone from transitional and depositional zones in lakes (the so-called mud energy boundary depth). Such models provide only coarse-scale predictions about depositional and erosional zones and do not make specific predictions about sediment properties, however (Anderson et al., 2008). Hydroacoustic measurements offer a way to rapidly collect information about sediment properties at a very large number of sites. Hydroacoustic methods involve emission of soundwaves at known energy and wavelength from a transducer; the soundwave is then propagated through the water column and reflected off of objects at a strength proportional to their size, shape and density and soundspeed contrasts with water. Bottom sediments represent the dominant source of acoustic backscatter in a lake, although aquatic plants (Zhu et al., 2007), fish (Kubecka and Duncan, 1998), zooplankton (Holliday and Pieper, 1980), larval insects (Knudsen et al., 2006) and gas bubbles (Ostrovsky et al., 2008) also scatter sound. Fisheries remain the most common application of hydroacoustic methods, although it is increasingly being used to identify different sediment types within nearshore ocean and estuarine studies (Tegowski, 2005; Freitas et al., 2005; Anderson et al., 2002). An early such application of hydroacoustic methods was conducted by Pouliquen and Lurton (1992), who found that cumulative energy curves of bottom echoes were related to sediment physical properties. Other acoustical properties have been used to distinguish sediments with different hardness, roughness and grain size, including the ratio of the first and second bottom echo intensities (Orlowski, 1984; Chivers et al., 1990) and the first and second parts of the first bottom echo (Sternlicht and de Moustier, 2003; BioSonics, 2008). The fractal dimension of the bottom echo has also been proposed (Tegowski and Lubniewski, 2000). These studies have, to this point, generally been conducted in nearshore and open ocean environments with relatively low frequency transducers (e.g., Freitas et al., 2005; Anderson et al., 2008). Characterization of bottom sediments in inland lakes and reservoirs using hydroacoustic methods has received much less attention.
The overall objective for this study was to evaluate the capability of hydroacoustic methods to reliably estimate bottom sediment properties in lakes. Specific objectives were to: (i) quantify the acoustical response of different types of sediment using 3 different frequencies; (ii) assess the capacity of different acoustical properties to quantify sediment characteristics; and (iii) test the extensibility of findings to other lakes.
2.
Materials and methods
2.1.
Hydroacoustic measurements
Hydroacoustic measurements were made using a BioSonics Inc. (Seattle, WA) DTX echosounder multiplexed with 430-kHz single beam, 201-kHz split beam and 38-kHz single beam transducers (Table 1). A JRC 212W real-time differential globalpositioning satellite (DGPS) receiver was mounted directly over the transducers and recorded differentially-corrected positions every 1-s. BioSonics VisualAcquisition 4.0 software on a Dell ATG laptop was used to set up transmit and receive information and acquire data during calibration and measurement. Data were acquired at 5 pings per sec at a pulse duration of 0.4 ms on each frequency (Table 1). Field calibration was conducted using tungsten-carbide spheres of known target strength. Echograms were analyzed using BioSonics VBT v.1.12 software with 30 log R time-varied gain (TVG) and depthnormalization to lake mean depth (Dommisse et al., 2005). The VBT software was used to extract 4 different attributes of the bottom echo envelope, i.e., the time-dependent amplitude of a received pulse (Horne, 2000). A soundwave emitted by the transducer propagates through the water at approximately 1500 m s1 and spreads radially depending upon the transducer half-beam angle (Table 1). The soundwave is then reflected off of the lake bottom and returned to the transmitter as the first bottom echo. The shape and energy of the reflected soundwave carries with it information about the bottom sediment. The increasing (first) part of the bottom echo (E10 ) contains information about hardness, while the decreasing (second) part of the bottom echo (E1) contains information about volume scattering and roughness (BioSonics, 2008; Ostrovsky and Tegowski, 2010). Sound reflected off of the bottom sediments is also reflected from the lake surface, and is subsequently reflected a second time off of the bottom sediments which can also be recorded by the transducer (the second bottom echo, E2). The ratio of the first and second parts of the first bottom echo (E10 /E1) can resolve different types of
Table 1 e Transducer configurations used in this study. Property Frequency (kHz) Beam angle ( ) Source level (dB/mPa) Receive sensitivity (dBC/mPa) Pulse length (ms) Pings per second (pps)
DTX-38
DTX-200
DTX-420
38 10.0 217.03 41.1 0.4 5
201 6.6 221.3 57.6 0.4 5
430 7.0 220.0 62.9 0.4 5
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 9 9 e4 4 0 8
bottom sediments and can serve as an indicator of hardness (Ostrovsky and Tegowski, 2010) The ratio of the second part of the first bottom echo and second bottom echo (E1/E2) has also been used to classify different types of sediments (Chivers et al., 1990). In addition to these energy-based measures, the shape and structure of the bottom echo also contains information about the sediments properties (Van Walree et al., 2005). VBT characterizes the shape of the bottom echo as a fractal dimension (FD) using the box-dimension approach (Van Walree et al., 2005; BioSonics, 2008). The FD of the bottom echo is thus a measure of the echo fluctuation and describes heterogeneity of the bottom (Tegowski, 2005).
2.2.
Study sites
Hydroacoustic surveys were conducted at Lake Elsinore and San Dieguito Reservoirs in southern California, USA. Lake Elsinore is the largest natural lake in southern California and is located in southwestern Riverside County about 110 km southeast of Los Angeles. The lake is nominally about 1200 ha in surface area, and has a mean depth near 4 m, although historical bathymetric data were derived from depth measurements made at about 50 locations across the lake. Lake Elsinore is highly eutrophic, with frequent algal blooms and periodic fish kills. A 270 km survey, on an orthogonal grid with 100 m spacing, was conducted over 6 days from June 27e30 and July 12e14, 2010. Sediment surface grab samples were collected from 28 sites on a staggered-start sampling grid matched to the hydroacoustic survey grid over several days in late MayeJune 2010 using an Ekman dredge (Fig. 1a, solid circles). An additional 6 sites were sampled from the lake on September 21, 2010 (Fig. 1a, open circles). Measurements were also made on San Dieguito Reservoir, a small (w20 ha) reservoir that serves as a source drinking
4401
water reservoir for the Santa Fe Irrigation District (SFID). Two 20 km surveys were conducted on April 24 and August 12, 2010. Sediment surface grab samples were collected from 5 sites on the lake during the summer sampling (Fig. 1b).
2.3.
Sediment analyses
Sediment grab samples collected with an Ekman dredge were homogenized and subsampled into individual 500-mL widemouth glass jars with Teflon-lined lid. Samples were stored on ice in a cooler and returned to the lab. Upon return to the lab, sediment was promptly rehomogenized and subsampled for sediment characterization and porewater analysis. Water content was determined on subsamples that were heated overnight at 105 C. This higher temperature, commonly used in soil analyses, was chosen over the lower temperatures (60e80 C) used by Hakanson and Jansson (1983) and others because of the high amount of hydrous clays and moderate organic matter content found in these sediments, although weight differences were minimal (often 0.50 statistically significant at p ¼ 0.01. Thus, a large number of statistically significant correlations were present (e.g., 11 out of 36 correlations were significant at 0.01 for the 430-kHz data) (Table 3). Of the acoustical properties, the fractal dimension (FD) and energy ratio of the first and second part of the first bottom echo (E10 /E1) were both statistically significantly correlated with a number of sediment properties, especially at the 430- and 201-kHz frequencies (Table 3). While statistical significance is important, r-values near 0.5 capture a relatively small fraction (25%) of the variance in the sediment properties; meaningful predictive capabilities require much higher rvalues. On that basis, one notes that generally very strong correlations exist between the fractal dimension of the first
Table 2 e Summary of sediment properties from Lake Elsinore (n [ 28). Property
The hydroacoustic survey on Lake Elsinore was first used to develop a more detailed bathymetric map for the lake at a surface elevation of 379.0 m above MSL, relative to an earlier one constructed at a lower surface elevation and from a limited number of measurements (Lawson and Anderson, 2007) (Fig. 1a). The region of maximum depth is located in the northeastern part of the lake, reaching about 9 m below one of the axial flow pump arrays installed on the lake
Depth Sand Silt Clay H2O content Organic C LOI CaCO3 Total N Total P
Units
Mean
Std dev
Min
Max
m % % % % % % % % mg kg1
4.6 30.8 38.8 30.5 58.5 2.63 8.31 10.44 0.35 788
1.9 35.9 19.6 23.4 25.1 1.93 5.69 6.90 0.20 475
1.33 0.0 8.4 3.3 15.6 0.04 0.36 0.11 0.01 106
7.5 88.0 72.4 65.4 83.1 4.99 16.04 18.17 0.59 1456
4403
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 9 9 e4 4 0 8
Table 3 e Correlation table showing correlation coefficient values between sediment properties and corresponding acoustical attributes (r-value >0.5 significant at 0.01). Sand 430-kHz FD E1/E2 E10 /E1 E1/FD 201-kHz FD E1/E2 E10 /E1 E1/FD 38-kHz FD E1/E2 E10 /E1 E1/FD
Silt
Clay
% H2O
Org C
LOI
CaCO3
Total N
Total P
0.83 0.35 0.51 0.21
0.42 0.34 0.40 0.18
0.92 0.24 0.45 0.17
0.86 0.35 0.56 0.18
0.91 0.26 0.48 0.14
0.91 0.32 0.51 0.17
0.84 0.34 0.55 0.17
0.85 0.26 0.41 0.13
0.13 0.06 0.09 0.10
0.76 0.16 0.65 0.65
0.34 0.22 0.54 0.41
0.89 0.06 0.54 0.65
0.81 0.16 0.66 0.71
0.87 0.14 0.56 0.72
0.87 0.16 0.61 0.76
0.79 0.11 0.60 0.71
0.82 0.15 0.40 0.67
0.08 0.18 0.16 0.06
0.89 0.39 0.14 0.72
0.69 0.19 0.10 0.52
0.79 0.44 0.14 0.68
0.89 0.48 0.22 0.74
0.85 0.45 0.09 0.78
0.88 0.46 0.14 0.80
0.88 0.59 0.25 0.71
0.75 0.44 0.02 0.70
0.17 0.26 0.15 0.03
bottom echo and a number of biogeochemically relevant sediment properties. For example, the FD at 430-kHz was strongly correlated with % organic C content (r ¼ 0.91), % clay content (r ¼ 0.92) and total N (r ¼ 0.85) (Table 3). The FD of the first bottom echo thus accounted for 83% of the variance in measured organic C content in the sediments of Lake Elsinore. Unlike the other sediment properties, the FD was negatively correlated with % sand content (r ¼ 0.83). Interestingly, total P concentration in sediments was not well-correlated with any acoustical property and, by extension, not correlated with other sediment properties as well (Table 3). The other, energy-based acoustical properties of bottom sediments, such as the energy ratio of the second part of the first bottom echo and second bottom echo (E1/E2), ratios of the first and second parts of the first bottom echo (E10 /E1) were less strongly correlated with physical and chemical characteristics than the FD or “shape” of the bottom echo (Table 3). The energy ratio of the first and second part of the first bottom echo (E10 /E1) at 430-kHz was nonetheless significantly correlated (at p ¼ 0.01) with % water content, loss-on-ignition and CaCO3 content, but provides much lower predictive power than FD. Correlation analysis yielded broadly similar results for most of the acoustical attributes at 201-kHz (Table 3). The FD was again the strongest correlate with sediment properties, with very strong but slightly lower r-values compared with 430-kHz. The ratio of E10 /E1 was somewhat more strongly correlated with % organic C and other properties at 201-kHz than 430-kHz (e.g., r-value of 0.56 vs. 0.48 for 201- and 430kHz, respectively, with organic C), but still only accounted for 31% of variance in organic C content. The ratio of E1/FD was more strongly correlated with most sediment properties at 201-kHz than 430-kHz (r-values up to 0.76 and accounting for up to 58% of variance), but also remained below values for FD (Table 3). Interestingly, r-values were often of the opposite sign at 38kHz when compared with the higher frequencies (Table 3). Fractal dimension was again the strongest correlate, but in this case, FD values generally declined with increasing clay content, organic C content (r-values of 0.79 to 0.85, respectively) and other properties. The exception was % sand
content that exhibited a positive r-value (opposite of that witnessed at 430- and 201-kHz) (Table 3). Frequency-dependent acoustic response has been used in plankton and fish studies to help resolve different types or size classes of scatterers (e.g., Jurvelius et al., 2008) and proposed to improve classification of bottom sediments (Anderson et al., 2008). Differences and ratios of E10 , E1, E2 and FD at 38-, 201- and 430-kHz respectively were not found to offer greater predictive power than single frequency measurements made at 430-kHz, however (data not shown). While r-values as provided in Table 3 provide a convenient summary of the strength of the relationship between acoustical and sediment physical and chemical properties, they can mask more subtle non-linear relationships. Plots of the FD of sediment bottom echo against physical properties indicate that strong linear relationships were, in fact, present for a number of properties, including clay content, organic C content, and total N content (Fig. 2aec) and loss-on-ignition (not shown), although a non-linear relationship between FD and water content was present (Fig. 2d). These latter two properties are often used as convenient surrogates for other more physically and chemically relevant properties (Hakanson and Jansson, 1983). Linear regressions of the FD of the bottom echo at 430-kHz with sediment properties from the initial 28 sites on Lake Elsinore (Fig. 2, solid circles) yielded the following equations: %Clay ¼ 528:5 FD 471:1
r2 ¼ 0:85
%Organic C ¼ 43:0 FD 38:2 %LOI ¼ 126:7 FD 111:9 %Total N ¼ 4:25 FD 3:68
r2 ¼ 0:83
r2 ¼ 0:83 r2 ¼ 0:72
(1) (2) (3) (4)
Included on these figures are sediment physicochemical properties and the FD of the bottom echo at 430-kHz from 6 additional sites from Lake Elsinore and the 5 sites sampled from San Dieguito Reservoir (Fig. 1). One sees that the additional 6 sites from Lake Elsinore (shown as open circles, Fig. 2) fell within the range of values found for the original 28 sites
4404
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 9 9 e4 4 0 8
a
100
b L. Elsinore Jul 2010 L. Elsinore Oct 2010 San Dieguito Res.
60
40
20
0 0.88
L. Elsinore Jul 2010 L. Elsinore Oct 2010 San Dieguito Res.
8
Organic C Content (%)
Clay Content (%)
80
10
6
4
2
0.92
0.96
1
1.04
0 0.88
1.08
0.92
Fractal Dimension
c
1.2
d L. Elsinore Jul 2010 L. Elsinore Oct 2010 San Dieguito Res.
0.8 0.6 0.4
1.04
1.08
100
60 L. Elsinore Jul 2010 L. Elsinore Oct 2010 San Dieguito Res.
40
20
0.2 0 0.88
1
80
Water Content (%)
Total N Content (%)
1
0.96
Fractal Dimension
0.92
0.96
1
1.04
1.08
Fractal Dimension
0 0.88
0.92
0.96
1
1.04
1.08
Fractal Dimension
Fig. 2 e Fractal dimension (FD) of bottom echo at 430-kHz vs. (a) clay content, (b) organic C content, (c) total N content and (d) % water content.
(shown at solid circles, Fig. 2). Thus sampling later in the summer did not substantively alter the relationship between FD and clay content, organic carbon content, total N or water content of the bottom sediments within this relatively narrow, transitional range of properties (Fig. 2). Also shown on these figures are the physicochemical and acoustical properties of sediments in San Dieguito Reservoir (Fig. 2). The five sites from the reservoir ranged in depth from 3.3 to 9.2 m, and varied strongly in clay, organic C, total N and water contents (Fig. 2). The highest of these components were found in the deepest water, consistent with the focusing of fine organic sediments there. It is noteworthy that these high concentrations exceeded by 50e100% the highest values found in Lake Elsinore, although the linear trends seen between clay, organic C and total N contents with FD in Lake Elsinore appear to hold for sediments collected from a very different lake (Fig. 2aec). The non-linear relationship between water content and FD also holds (Fig. 2d). The capability of FD to predict sediment properties can be quantitatively explored with regression equations fit to the original Lake Elsinore data using attributes measured at
additional sites from Lake Elsinore and at San Dieguito Reservoir. As illustrated in Fig. 2a, Eq. (1) yielded low predicted clay contents, on the order of 10e20%, in the additional Lake Elsinore samples collected in September in reasonable agreement with measured values (Fig. 2a, open circles). The regression model between FD and % clay content also reasonably predicted clay content in the sediments of San Dieguito Reservoir (Fig. 2a, shaded squares). The mean error in predicted clay content (predictedeobserved concentrations) was 1.2 7.4% (Table 3). The regression model for clay content was thus close to neutral bias, and accurate to within about 7% for these samples. The regression model developed from the original Lake Elsinore sites also reasonably predicted organic C contents of sediments collected from the lake in September, as well as those collected from San Dieguito Reservoir (Fig. 2b). The mean error estimate was biased slightly low, but generally within 1.0% of the measured values (Table 4). Similarly, %LOI and total N were reasonably well-described by the regression models, with mean errors of 0.14 1.97% and 0.02 0.16%,
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 9 9 e4 4 0 8
Table 4 e Summary of estimate errors (predictedeobserved) from regression equations when fit to San Dieguito Reservoir and additional Lake Elsinore samples (n [ 11). Estimate error
% Clay
% Organic C
Mean s.d. 1.2 7.4 0.21 1.00 Range 10.7e16.7 1.9e1.8
% LOI
% Total N
0.14 1.97 0.02 0.16 2.7e4.4 0.29e0.22
respectively (Table 4), although regression equation fitted to the initial Lake Elsinore samples was observed to underpredict total N in the organic-rich San Dieguito Reservoir sediments (Fig. 2c). The capability of predicting sediment properties from their hydroacoustic properties allows one to develop detailed spatial representations of sediment properties across a lake basin. For example, using the regression equation developed from the initial sampling of Lake Elsinore and validated with subsequent samples collected there as well as at San Dieguito Reservoir (Eq. (2)), the survey reveals that organic C concentrations in Lake Elsinore are highest in the central-northern part of the lake (Fig. 3a), coinciding with greatest depth there (Fig. 1a). Sediments with very low organic C contents (and low clay and high sand contents) were present in a shallow bay at the south end of the lake, consistent with wave-induced resuspension owing to the strong afternoon winds from the northwest. Distribution of organic C (Fig. 3a) thus broadly followed that of depth in this simple basin (Fig. 1a), although heterogeneities are clearly evident (e.g., note the 3e4% organic C sediment that intrudes into the south-central part of the lake basin where 4e5% organic C was more broadly found). The area of the lake basin with very low sediment organic C concentrations (4%) comprised a larger fraction of the lake bottom (28.9%). The finescale heterogeneities (on a scale less than the transect spacing) that are present in the figure are a result of the interpolation process (kriging) used to develop these contour plots.
4405
The distribution of organic C in San Dieguito Reservoir exhibited a weaker relationship with depth, with high organic C sediments present both in deep water near the dam, and also in the shallower water on the east side of the lake, while low organic C content sediments were present in the south part of the reservoir (Fig. 3b). San Dieguito Reservoir differs from Lake Elsinore in the presence of a well-developed littoral zone; in contrast, the seasonally and annually fluctuating surface elevation of Lake Elsinore (Lawson and Anderson, 2007) limits the development of an aquatic macrophyte community there. Thus, focusing of organic matter into the deeper part of San Dieguito Reservoir results in increased organic C concentrations there, although organic matter content of the sediments is also influenced by the presence of emergent and submerged aquatic macrophytes. Two additional factors are also thought to have influenced the distribution of organic C in San Dieguito Reservoir. First of all, filter backwash from the R.E. Badger Filtration plant is discharged to the shallow northeastern part of the lake that may have increased the total organic C content of sediments there. Secondly, prior to impoundment in 1918, Escondido Creek flowed through the site; as a result, native wetland soils would have also been present near and within some stretches of the creek. The region of maximum organic C content (reaching or exceeding 10% by weight) (Fig. 3b) was found in intermediate depth (about 6e7 m) near or within the creek channel (Fig. 1b). Since SOD and internal nutrient recycling are both correlated with the organic C content of sediments (Hargrave, 1972), it follows that hydroacoustic methods can also be used to infer the rates of these important processes in sediments. As found with clay content, organic C and other properties, the fractal dimension of the bottom echo was also linearly related to rates of release of PO4-P (r2 ¼ 0.98) and NH4N (r2 ¼ 0.96) from bottom sediments as well as SOD (r2 ¼ 0.94) (Fig. 4). Nonetheless, the FD values of bottom sediment represent the echo fluctuations due to heterogeneities in sediment physicochemical properties, and are thus indirectly related to rates of nutrient recycling or SOD, although depending upon one’s particular interests, the method allows
Fig. 3 e Distribution of organic C concentrations across: (a) Lake Elsinore and (b) San Dieguito Reservoir.
4406
b
16
NH4-N Flux (mg m-2 d-1)
PO4-P Flux (mg m-2 d-1)
12
8
4
0
-4 0.88
0.92
0.96
1
Fractal Dimension
1.04
c
160
120
80
40
0 0.88
1
0.8
SOD (g m-2 d-1)
a
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 9 9 e4 4 0 8
0.6
0.4
0.92
0.96
1
Fractal Dimension
1.04
0.2 0.88
0.92
0.96
1
1.04
Fractal Dimension
Fig. 4 e Fractal dimension (FD) at 430-kHz vs. (a) PO4-P flux, (b) NH4-N flux and (c) sediment oxygen demand (SOD) (Lake Elsinore).
estimation of physicochemical properties as well as (autocorrelated) biogeochemical processes.
4.
Discussion
The acoustical properties of bottom sediments from Lake Elsinore were significantly correlated with a number of physical and chemical properties, including textural information, water content, and organic C and total N contents. The fractal dimension of the first bottom echo was the acoustic property most strongly correlated with these sediment attributes (correlation coefficient values typically greater than 0.8) (Table 3). A survey from the North Sea that included mean grain size analysis on 50 grab samples also found that the fractal dimension of the bottom echo increased with smaller average particle size, and was more strongly correlated than other acoustical attributes (e.g., echo energy) at an echosounder frequency of 66 kHz (r ¼ 0.84) (Van Walree et al., 2005). A slightly weaker correlation was found at 150 kHz. The fractal dimension of the bottom echo was also found to effectively distinguish between different types of bottom sediments (silt, clay, fine-grained sand, and coarse sand þ gravel) in the southern Baltic Sea (Tegowski, 2005) and in Lake Kinneret in Israel (Ostrovsky and Tegowski, 2010). Coarse-textured sediment, often found in shallow regions of lakes, is frequently more homogeneous vertically than finer textured sediments with higher organic matter contents within depositional zones than can include varves, gas pockets, and other fine-scale features (Ostrovsky and Tegowski, 2010). That greater heterogeneity within the bottom sediments can alter the shape and increase the fractal dimension of the reflected soundwave. The slightly stronger correlations found at 430-kHz than either 201- or 38-kHz result from differences in penetration depths, and ability to resolve small spatial structures within the sediments (e.g., laminations, gas pockets). The lower
frequencies penetrate bottom sediments more effectively (Dunbar et al., 1999), but the longer wavelength reduces the ability to resolve fine-scale heterogeneities within the sediment. In a very simple way, assuming resolution is approximated by l/2 and the speed of sound is near 1500 m s1, the 430-kHz frequency would be able to resolve features on the scale of about 2 mm, while the 38-kHz frequency would only be able to resolve features 10 larger, or about 2 cm. In a direct comparison of acoustic sediment classes with ground-truth measurements using both 66-kHz and 150-kHz frequencies, Van Walree et al. (2005) also found the higher frequency was more effective at resolving sediment types in a North Sea survey. The negative correlations found for FD at 38-kHz for most of the measured sediment properties in this study (Table 3) are thought to result from greater variance in penetration depths in combination with the longer wavelength and lower resolution at this frequency. Results from this study demonstrate that the hydroacoustic characterization of bottom sediments, especially the fractal dimension of the bottom echo, can provide valuable insights into the properties and distribution of sediments in lakes and reservoirs. Beyond mean grain size and, in some cases, loss-on-ignition that has been evaluated in most coastal studies, the fractal dimension of the bottom echo can provide information about organic C and total N concentrations in bottom sediments, as well as estimates of rates of internal nutrient recycling and sediment oxygen demand. Findings here indicate that there is some general extensibility of results from measurements made on Lake Elsinore in JuneeJuly 2010 to those made on the lake somewhat later in the year, as well as to San Dieguito Reservoir. This suggests that extensive ground-truthing may not be necessary for similar lakes in this region if qualitative estimates of organic C and other physical and chemical properties are sufficient, although ground-truthing remains necessary to maximize the accuracy of the estimates. Ground-truthing is needed to understand rates of biogeochemical processes in sediments,
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 9 9 e4 4 0 8
however, owing to their sensitivity to water column properties, such as temperature and DO. Incorporation of the additional sample results from Lake Elsinore, as well as those from San Dieguito Reservoir, into the regression models provide somewhat improved correlations relative to the initial Lake Elsinore dataset (increasing r2 values by 0.04e0.11): %Clay ¼ 516:9 FD 460:4
r2 ¼ 0:89
%Organic C ¼ 46:6 FD 41:5 %LOI ¼ 128:5 FD 113:7 %Total N ¼ 5:15 FD 4:53
r2 ¼ 0:87
r2 ¼ 0:88 r2 ¼ 0:83
(5) (6) (7)
2. The fractal dimension of the bottom echo was most strongly correlated with sediment properties, with r2 values of 0.83e0.85 at 430-kHz. 3. The fractal dimension of the bottom echo was also linearly related with rates of PO4-P and NH4-N flux and sediment oxygen demand of bottom sediments. 4. Somewhat stronger correlations between sediment properties and fractal dimension of the bottom echo were found at 430-kHz relative to 201- and 38-kHz. 5. Hydroacoustic methods offer a rapid way to determine sediment physical and chemical properties and rates of important biogeochemical processes across a lake, and are thus a valuable tool for habitat assessment, water quality management and lake restoration.
(8)
Inclusion of additional sample results increased the r2 value for % total N more so than the other properties (r2 value increased from 0.72 to 0.83); the additional samples lowered the response at low FD, resulting in a higher slope than found from the initial sample set (Fig. 2c). Additional work would be needed to test the applicability of these regression equations to other lakes and to possible seasonal differences (e.g., due to sediment gas accumulation, production and water level fluctuations) (Ostrovsky and Tegowski, 2010), basin morphometries and water quality. Fine lateral scale heterogeneities are also thought to have influenced observations reported herein. That is, the Ekman dredge necessarily samples a small cross-sectional are of sediment (about 230 cm2). In contrast, the transducers acoustically sampled a much larger cross-sectional area that is a function of beam angle and sediment depth. For example, with a beam halfangle of 7 (Table 1), the 430-kHz transducer would acoustically sample a sediment area over 0.7 m2 (or 33 greater than the Ekman dredge) at the mean depth of the lake. This larger sampling area is advantageous in the context of sediment characterization within a lake or basin survey, but does introduce spatial scale differences during ground-truthing. As a result, sediment surface grab samples were collected from a relatively large number of sites during this study. Notwithstanding, the capacity to map sediment properties across a lake can provide critical information for use in habitat assessment, water quality management and lake restoration. For example, being able to map the distribution of PO4-P flux from bottom sediments based upon FD (Fig. 4a) allows targeted alum applications to the regions where most benefit can be attained. Similarly, understanding the distribution of SOD across a lake basin can aid in the design of hypolimnetic oxygenation and diffused aeration systems and placement of diffuser lines. The value of hydroacoustic mapping of sediments for benthic habitats has already been recognized in coastal, seafloor and mid-shelf systems (Hewitt et al., 2004; Freitas et al., 2006).
5.
4407
Conclusions
1. Sediment physical and chemical properties were significantly correlated with acoustical properties of bottom sediments.
Acknowledgments Thanks to the Lake Elsinore-San Jacinto Watersheds Authority (LESJWA) and the Santa Fe Irrigation District (SFID) for their support of this study. Special thanks to Pat Kilroy, City of Lake Elsinore, and Tim Bailey and Cor Shaffer, SFID, for their support and assistance, and to David Thomason and Ed Betty, University of California-Riverside, for assistance in the collection and analysis of the sediment samples.
references
Anderson, J.T., Gregory Gregory, R.S., Collins, W.T., 2002. Acoustic classification of marine habitats in coastal Newfoundland. ICES J. Mar. Sci. 59, 156e167. Anderson, J.T., Van Holliday, D., Kloser, R., Reid, D.G., Simard, Y., 2008. Acoustic seabed classification: current practice and future directions. ICES J. Mar. Sci. 65, 1004e1011. Anderson, M.A., Whiteaker, L., Wakefield, E., Amrhein, C., 2008. Properties and distribution of sediment in the Salton Sea, California: an assessment of predictive models. Hydrobiologia 604, 97e110. Anderson, M.A., 2004. Impacts of metal salt addition on the chemistry of Lake Elsinore, California: 2. Calcium salts. Lake Reservoir Manage. 20, 270e279. APHA, 1998. Standard Methods for the Examination of Water and Wastewater, twentieth ed. American Public Health Association, Washington, D.C. Aspila, K.I., Agemian, H., Chau, A.S.Y., 1976. A semi-automated method for determination of inorganic, organic and total phosphate in sediments. Analyst 101, 187e197. Baudo, R., Amantini, L., Bo, F., Cenci, R., Hannaert, P., Lattanzio, A., marengo, G., Muntau, H., 1989. Spatial-distribution patterns of metals in the surface sediments of Lake Orta (Italy). Sci. Total Environ. 87, 117e128. BioSonics, Inc., 2008. User Guide: Visual Bottom Typer 1.10. BioSonics, Inc., Seattle, WA, 113 pp. Chivers, R.C., Emerson, N., Burns, D.R., 1990. New acoustic processing for underway surveying. Hydrographic J. 56, 9e17. Dommisse, M., Urban, D., Finney, B., Hills, S., 2005. Potential depth biasing using BioSonics VBT seabed classification software. Mar. Technol. Soc. J. 39, 90e93. Downing, J.A., Rath, L.C., 1988. Spatial patchiness in the lacustrine sedimentary environment. Limnol. Oceanogr. 33, 447e458. Duarte, C.M., Kalff, J., 1986. Littoral slope as a predictor of the maximum biomass of submerged macrophyte communities. Limnol. Oceanogr. 31, 1072e1080.
4408
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 3 9 9 e4 4 0 8
Dunbar, J.A., Allen, P.M., Higley, P.D., 1999. Multifrequency acoustic profiling for water reservoir sedimentation studies. J.Sediment. Res. 69, 521e527. Forstner, U., Wittmann, G.T., 1979. Metal Pollution in the Aquatic Environment. Springer-Verlag, Berlin. Freitas, R., Sampaio, L., Rodrigues, A.M., Quintino, V., 2005. Seabottom classification across a shallow-water bar channel and near-shore shelf, using single-beam acoustics. Estuar. Coast. Shelf Sci. 65, 625e632. Freitas, R., Sampaio, L., Oliveira, J., Rodrigues, A.M., Quintino, V., 2006. Validation of soft bottom benthic habitats identified by single-beam acoustics. Mar. Pollut. Bull. 53, 72e79. Gee, G.W., Bauder, J.W., 1986. Particle-size analysis. In: Klute, A. (Ed.), Methods of Soil Analysis. Part 1Agronomy Monographs, second ed., vol. 9. ASA and SSSA, Madison, WI, pp. 383e411. Hakanson, L., Jansson, M., 1983. Principles of Lake Sedimentology. Springer-Verlag, New York, NY. Hakanson, L., 1982. Lake bottom dynamics and morphometry: the dynamic ratio. Water Resour. Res. 18, 1444e1450. Hargrave, B.T., 1972. Aerobic decomposition of sediment and detritus as a function of particle surface area and organic content. Limnol. Oceanogr. 17, 583e596. Hatcher, K.J., 1986. Sediment Oxygen Demand: Processes, Modeling & Measurement. Institute of Natural Resources, Univ. of Georgia, Athens, GA. Heiri, O., Lotter, A.F., Lemcke, G., 2001. Loss on ignition as a method for estimating organic and carbonate content in sediments: reproducibility and comparability of results. J. Paleolimnol. 25, 101e110. Hewitt, J.E., Thrush, S.F., Legendre, P., Funnel, G.A., Ellis, J., Morrison, M., 2004. Mapping of marine soft-sediment communities: integrated sampling for ecological interpretation. Ecol. Appl. 14, 1203e1216. Holdren, G.C., Armstrong, D.E., 1980. Factors affecting phosphorus release from intact sediment cores. Environ. Sci. Technol. 14, 79e87. Holliday, D.V., Pieper, R.E., 1980. Volume scattering strengths and zooplankton distributions at acoustic frequencies between 0.5 and 3 MHz. J. Acoust. Soc. Am. 67, 135e146. Horne, J.K., 2000. Acoustic approaches to remote species identification: a review. Fish. Oceanogr. 9, 356e371. Jurvelius, J., Knudsen, F.R., Balk, H., Marjomaki, T.J., Peltonen, H., Taskinen, J., Tuomaala, A., Viljanen, M., 2008. Echo-sounding can discriminate between fish and macroinvertebrates in fresh water. Freshwater Biol. 53, 912e923. Karickhoff, S.W., Brown, D.S., Scott, T.A., 1979. Sorption of hydrophobic pollutants on natural sediments. Water Res. 13, 241e248. Kirby, M.E., Lund, S.P., Anderson, M.A., Bird, B.W., 2007. Insolation forcing of Holocene climate change in Southern California: a sediment study from Lake Elsinore. J. Paleolimnol. 38, 395e417. Knudsen, F.R., Larsson, P., Jakobsen, P.J., 2006. Acoustic scattering from a larval insect (Chaoborus flavicans) at six echosounder frequencies: implication for acoustic estimates of fish abundance. Fish. Res. 79, 84e89. Kubecka, J., Duncan, A., 1998. Acoustic size vs. real size relationships for common species of riverine fish. Fish. Res. 35, 115e125. Lawson, R., Anderson, M.A., 2007. Stratification and mixing in Lake Elsinore, California: an assessment of axial flow pumps
for improving water quality in a shallow eutrophic lake. Water Res. 41, 4457e4467. Lehman, J.T., 1975. Reconstructing the rate of accumulation of lake sediment: the effect of sediment focusing. Quaternary Res. 5, 541e550. Loeppert, R.H., Suarez, D.L., 1996. Carbonate and gypsum. In: Sparks, D.L. (Ed.), Methods of Soil Analysis. Part 3Agronomy Monographs, third ed., vol. 9. ASA and SSSA, Madison, WI, pp. 437e474. Lundgren, K., Tysklind, M., Ishaq, R., Broman, D., Van Bavel, B., 2002. Polychlorinated naphthalene levels, distribution, and biomagnification in a benthic food chain in the Baltic Sea. Environ. Sci. Technol. 36, 5005e5013. Nelson, D.W., Sommers, L.E., 1982. Total carbon, organic carbon, and organic matter. In: Page, A.L., Miller, R.H., Keeney, D.R. (Eds.), Methods of Soil Analysis, Part 2Agronomy Monographs, second ed., vol. 9. ASA and SSSA, Madison, WI, pp. 539e580. Orlowski, A., 1984. Application of multiple echoes energy measurements for evaluation of sea bottom type. Oceanologia 19, 61e78. Ostrovsky, I., Tegowski, J., 2010. Hydroacoustic analysis of spatial and temporal variability of bottom sediment characteristics in Lake Kinneret in relation to water level fluctuation. Geo-Mar. Lett. 30, 261e269. Ostrovsky, I., McGinnis, D.F., Lapidus, L., Eckert, W., 2008. Quantifying gas ebullition with echosounder: the role of methane transport by bubbles in a medium-sized lake. Limnol. Oceanogr. Meth. 6, 105e118. Peeters, E.T.H.M., Gylstra, R., Vos, J.H., 2004. Benthic macroinvertebrate community structure in relation to food and environmental variables. Hydrobiologia 519, 103e115. Pouliquen, E., Lurton, X., 1992. Identification of the nature of the seabed using echosounders. J. Phys. IV (2), 941e944. Reese, B.K., Anderson, M.A., Amrhein, C., 2008. Hydrogen sulfide production and volatilization in a polymictic eutrophic saline lake, Salton Sea, California. Sci. Total Environ. 406, 205e218. Rowan, D.J., Kalff, J., Rasmussen, J.B., 1992. Estimating the mud deposition boundary depth in lakes from wave theory. Can. J.Fish. Aquat. Sci. 49, 2490e2497. Sondegaard, M., Jensen, J.P., Jeppesen, E., 2003. Role of sediment and internal loading in shallow lakes. Hydrobiologia 506e509, 135e145. Sternlicht, D.D., de Moustier, C.P., 2003. Time-dependent seafloor acoustic backscatter (10e100 kHz). J. Acoust. Soc. Am. 114, 2709e2725. Tegowski, J., Lubniewski, Z., 2000. The use of fractal properties of echo signal for acoustical classification of bottom sediments. Acoustica/Acta Acoust. 86, 276e282. Tegowski, J., 2005. Acoustical classification of the bottom sediments in the southern Baltic Sea. Quatern. Int. 130, 153e161. Van Walree, P.A., Tegowski, J., Laban, C., Simons, D.G., 2005. Acoustic seafloor discrimination with echo shape parameters: a comparison with ground truth. Cont. Shelf Res. 25, 2273e2293. Zhu, B., Fitzgerald, D.G., Hoskins, S.B., Rudstam, L.G., Mayer, C.M., Mills, E.L., 2007. Quantification of historical changes of submerged aquatic vegetation cover in two bays in Lake Ontario with three complementary methods. J. Great Lakes Res. 33, 122e135.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 0 9 e4 4 1 8
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Transport and deposition of CeO2 nanoparticles in water-saturated porous media Zhen Li a, Endalkachew Sahle-Demessie b, Ashraf Aly Hassan b, George A. Sorial a,* a
Environmental Engineering Program, School of Energy, Environmental, Biological, and Medical Engineering, University of Cincinnati, P.O. Box 210012, Cincinnati, OH 45221-0012, USA b U.S. Environmental Protection Agency, Office of Research and Development, NRMRL, 26 W. Martin Luther King Drive (MS 443), Cincinnati, OH 45268, USA
article info
abstract
Article history:
Ceria nanoparticles are used for fuel cell, metal polishing and automobile exhaust catalyst;
Received 16 March 2011
however, little is known about the impact of their release to the environment. The stability,
Received in revised form
transport and deposition of engineered CeO2 nanoparticles through water-saturated
16 May 2011
column packed with sand were studied by monitoring effluent CeO2 concentration. The
Accepted 23 May 2011
influence of solution chemistry such as ionic strength (1e10 mM) and pH (3e9) on the
Available online 31 May 2011
mobility and deposition of CeO2 nanoparticles was investigated by using a three-phase (deposition-rinse-reentrainment) procedure in packed bed columns. The results show
Keywords:
that water chemistry governs the transport and deposition of CeO2 nanoparticles. Trans-
Ceria
port is significantly hindered at acidic conditions (pH 3) and high ionic strengths (10 mM
Flow through porous media
and above), and the deposited CeO2 particles may not be re-entrained by increasing the pH
Modeling transport and deposition
or lowering the ionic strength of water. At neutral and alkaline conditions (pH6 and 9), and
Nanoparticles
lower ionic strengths (below 10 mM), partial breakthrough of CeO2 nanoparticles was observed and particles can be partially detached and re-entrained from porous media by changing the solution chemistry. A mathematical model was developed based on advection-dispersion-adsorption equations and it successfully predicts the transport, deposition and re-entrainment of CeO2 nanoparticles through a packed bed. There is strong agreement between the deposition rate coefficients calculated from experimental data and predicted by the model. The successful prediction for attachment and detachment of nanoparticles during the deposition and re-entrainment phases is unique addition in this study. This work can be applied to access the risk of CeO2 nanoparticles transport in contaminated ground water. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
Manufactured nanoparticles have been developed and used in a diverse range of products and industries in the past few decades due to their unique and novel physicochemical properties. While their applications can benefit medicine, textiles, electronics, agriculture, cosmetics, new
materials and environmental remediation; the potential impact of engineered nanomaterials on human health and aquatic animals is not fully understood once they are released to the environment (Aitken et al., 2006; Baun et al., 2008; Bystrzejewska-Piotrowskaa et al., 2009; Dunphy Guzma´n et al., 2006; Navarro et al., 2008; Nowack and Bucheli, 2007).
* Corresponding author. Tel.: þ1 513 556 2987; fax: þ1 513 556 4162. E-mail address:
[email protected] (G.A. Sorial). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.05.025
4410
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 0 9 e4 4 1 8
Several researchers have investigated the transport and retention of nanomaterials through water-saturated porous media in packed bed columns, which represents the behavior of nanomaterials in ground water environments or engineered granular filtration systems. Experimental and mathematical modeling studies of the transport and deposition of Fullerene (C60) nanoparticles have been conducted under varying flow conditions and electrolyte species in glass beads, quartz sands and Ottawa sand (Li et al., 2008). The mobility of Fullerene and metal oxides nanoparticles in aquatic system were evaluated and compared via column study, and the role of factors such as water velocity, electrolyte species and ionic strength was investigated (Lecoanet et al., 2004). In these studies, aqueous suspensions of nanoparticles were injected into columns packed with granular media and the behavior of the nanoparticles was commonly interpreted with clean-bed filtration theory, which describes the transport by mechanisms, interception on the particle by the media, sedimentation caused by gravity, and diffusion due to Brownian motion (Yao et al., 1971). CeO2 nanoparticles are used to make novel nanomaterials, and are widely applied for polishing materials (Kosynkin et al., 2000), automobile exhaust catalysts (Fu et al., 2001), fuel cell materials (Corma et al., 2004), and additives in glass and ceramic application (Livingston and Helvajian, 2005). However, CeO2 nanoparticles have significant chronic toxicity for algae (Thill et al., 2006) and large uptake of nanoscale cerium was found in the liver of zebrafish exposed via ingestion (Johnston et al., 2010), and human lung fibroblast cells fast absorb nanoCeO2 even at a low concentration (100 ppbe100 ppm) (Limbach et al., 2005). CeO2 nanoparticles produce significant oxidative stress in human lung cells, indicating lipid peroxidation and cell membrane damage (Lin et al., 2006). On the other hand, there is limited work on the stability, mobility, transport and deposition of CeO2 nanoparticles, which determines their potential exposure and bioavailability of environmentally released particles. Hence, understanding the transport and fate of CeO2 nanoparticles is of particular interest to fill this knowledge gap. In this study, we explored the transport and deposition of CeO2 nanoparticles in water-saturated sand columns, a process that is relevant to both ground water movement and the treatment of potable water by sand filtration methods. The objective of the present experimental and modeling study is to investigate the stability, transport and deposition of commercial manufactured CeO2 nanoparticles through watersaturated pre-cleaned sand columns, and to evaluate the effect of solution chemistry (ionic strength and pH) on the mobility of CeO2 nanoparticles.
2.
Theoretical consideration
For steady-state fluid flow, the transport and retention of CeO2 nanoparticles in a homogeneous porous medium can be described by the traditional 1-dimensional advectiondispersion-sorption/desorption kinetics (Eq. (1)). The kinetics of CeO2 attachment was expressed similar to the clean-bed filtration theory (Kuhnen et al., 2000; Li et al., 2008; Saiers et al., 1994) as a function of two coefficients: One is kmod, which is
the modeled rate of nanoparticle attachment along the depth, while the other Kdet estimates the detachment rates. Smax is the maximum retention capacity of CeO2 nanoparticles within the bed. Equations (1) and (2) given below are solved according to the initial and boundary conditions given in equation (3). vC v2 C vC r vS ¼ D 2 vp b vt vx vx qw vt
(1)
rb vS Smax S r kmod C b kdet S ¼ qw vt qw Smax
(2)
CðX; t ¼ 0Þ ¼ 0 Sðx; t ¼ 0Þ ¼ 0 B:C CðX ¼ 0; tÞ ¼ C0 vC ðX ¼ L; tÞ ¼ 0 vX
(3)
I:C
where, C is the concentration of CeO2 nanoparticles in solution, S is the concentration of CeO2 nanoparticles associated with the solid phase, t is time elapsed, x is the distance parallel to the flow, rb is the sand bulk density, D is the hydrodynamic dispersion coefficient, qwis the volumetric water content and vp is pore water velocity. Prior to the introduction of CeO2 nanoparticles, a nonreactive tracer test was conducted to assess water flow and hydrodynamic dispersion in the columns (data not shown). The hydrodynamic dispersion coefficient, D, used later in the model was calculated using the tracer study. The deposition and detachment rate coefficients can be estimated by fitting the transport equations (1) and (2) and initial and boundary conditions, equation (3), to the experimental breakthrough curve using a nonlinear least square technique. The mathematical model was solved by using Mathematica software (Wolfram, 1991). The deposition of nanoparticles in porous media is limited, by the frequency of collision between the colloids and the matrix surface, and by the collision efficiency. The frequency of collision is controlled by the characteristics of nanoparticles and the matrix, and hydrodynamics of the flow. The collision efficiency is the fraction of collisions resulting in the attachment and sorption of the nanoparticles on the matrix surface. In the analysis of experimental data, attachment efficiency, a, is defined as the ratio of the rate of particle deposition on a collector to the rate of collisions with that collector that has been used. For column studies, a can be expressed as a function of the relative effluent concentration C/C0, where C is the effluent concentration at time t, and C0 is the influent concentration, shown in Eq. (4) (Yao et al., 1971): 2dc a¼ lnðC=C0 Þ 3ð1 eÞh0 L
(4)
Where dc is the median diameter of the porous media, e is the bed porosity, L is the length of the bed, and h0 is the theoretical clean-bed single collector efficiency, which describes the particle transport to an individual collector before particle accumulation alters the collector geometry. Single collector efficiency can be calculated as the sum of individual contributions each transport mechanism, and the overall correlation equation (Tufenkji and Elimelech, 2003). The experimental particle deposition rate coefficient, kexp, which
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 0 9 e4 4 1 8
represents the rate of physical chemical filtration, is related to the single-collector efficiency, h0, when the transport of colloids in saturated porous media is within the advectiondispersion range. The expression for particle deposition rate is given as (Tufenkji and Elimelech, 2003): kexp ¼
3 ð1 eÞ Uah0 2 dc e
(5)
Where U is the Darcy velocity of flow.
3.
Experimental methods
3.1. Preparation and characterization of CeO2 nanoparticles High grade CeO2 nanoparticles were obtained from Alfa Aesar (Ward Hill, MA) in suspension form (18 wt% in water with dispersants and polishing additives) and were used as is. Stock solutions of CeO2 nanoparticles were prepared by diluting the original suspension. The electrolyte concentration and pH of stock solutions were adjusted to the desired levels by adding appropriate amount of standard solutions: NaCl solution (1 M), NaOH (1 M), and HCl (1 N). Both laser diffraction particle size analyzer (Beckman Coulter LS230) and dynamic light scattering (DLS) system (ZetaSizer Nano ZS, Malvern) were used to determine the size distribution of CeO2 nanoparticles. Laser Doppler velocimetry in conjunction with phase analysis light scattering (ZetaSizer Nano, Malven) was used to measure the zeta potential of CeO2 nanoparticles. For size and zeta potential measurements, CeO2 nanoparticles suspensions were prepared with various initial ionic strengths (1e100 mM) over a pH range of 2.83e10.23. Transmission electron microscopy (JEOL, JEM2010) images were obtained to confirm the primary size distribution of the particles. Suspended CeO2 nanoparticles were sonicated for 20 min before deposited onto a copper mesh grid and left to air-dry prior to TEM analysis. Powder XRD analysis was performed to determine the chemical composition and crystallographic structure of the nanoparticles. XRD data were collected with Cu Ka radiation (l ¼ 0.15059 nm) with a 0.02 2q step size at the speed of 2 min1. UVevis absorbance spectroscopy of CeO2 nanoparticles were measured by HP 8453 over wavelength range of 200e800 nm. Calibration was based on maximum absorbance of l ¼ 309 nm.
3.2.
Transport and deposition study
Glass columns (45 cm length 2.54 cm diameter) were packed with the 20 30 mesh size fraction (geometric mean diameter ¼ 0.717 mm) of industrial mineral silica sand that contains 98.2% SiO2 and trace amount of metal oxides (AGSCO Co., IL). To remove metal and organic impurities the packing sand was thoroughly cleaned by sequential washing (1 M HNO3) water rinsing, and oven-drying (55 C, 12 h).The columns were packed with pretreated sand using a wet packing method by adding 1 cm depth at a time, yielding a bed porosity of 0.34. Schematic diagram of experimental setup is shown in Supplementary Material (Fig. 1(A)). A
4411
three-phase procedure was applied to evaluate the role of solution chemistry on the transport, deposition and retention of CeO2 nanoparticles following the procedure reported earlier (McDowell-Boyer, 1992). In Phase I (Deposition Phase), six pore volumes of CeO2 nanoparticle suspension was introduced continuously to the column, followed by Phase II (Rinse Phase), during which the column was rinsed with four pore volumes of the particle free solution which has the same solution chemistry as the nanoparticle suspensions used in Phase I. The ionic strength and pH was kept the same during these two phases but varied for different tests. In Phase III (Re-entrainment Phase), MilliQ water was fed to the column to lower the ionic strength. The flow rate was kept constant at 20 mL/min for all three phases, providing NRe of 0.844. A sample of 3 mL was collected from the effluent stream every 1 min and was analyzed by UVevis spectrometer at the wavelength of l ¼ 309 1 nm. Calibration of UVevis spectrometer based on multi-concentration standard samples of CeO2 nanoparticles were used for quantitative analysis as shown in Fig. B-1 in supplementary material.
4.
Results
4.1.
Characterization of CeO2 nanoparticles
The particle size distribution of the CeO2 NPs in 1 mM NaCl solution was measured by both DLS and laser diffraction particle analyzer (Fig. 1(a)). DLS intensity based The average hydrodynamic diameter as determined by DLS analyzer was 152.7 nm, and the laser diffraction particle analyzer provided a mean diameter of 62.6 nm. DLS measurements also gave a wider size distribution. This difference can be accounted to differences in the techniques of the two methods. DLS measures the hydrodynamic diameter, which refers to how a particle diffuses in fluid, where a laser beam measures time-dependent fluctuations in scattered intensity caused by particles undergoing Brownian motion. The intensity fluctuations are auto correlated by particle size distribution according to Doppler Effect. Whereas, for laser diffraction particle analyzer, particle size distribution is obtained by measurements of low angle light scattering intensity as a function of the scattering angle, the wavelength and polarization of light measured based on applicable scattering models. Correlations for both measurement techniques are based on the assumption that the particles are spherical. For a monodispersed, spherical latex particle standard (Standard, L300, Nominal 300 nm Latex Particles, Beckman Coulter), the two techniques give identical size distribution. Hence the difference in size distribution of the two techniques (Fig. 1(a)) indicates that the CeO2 used in this study are non-spherical particles, which is confirmed by TEM image. A representative TEM image of CeO2 nanoparticles is shown in Fig. 1-b, which shows the variance in shape and size of the primary particles and provides evidence for aggregation. The images show that the 2dimensional projects of the primary particles range from ca.5 nm diameter to ca.60 nm diameter and appear to have triangle, pentagonal and hexagonal shapes similar to self-
4412
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 0 9 e4 4 1 8
Fig. 1 e Characterization of CeO2 nanoparticles. (a) Size distribution of CeO2 in MilliQ water, measured by DLS and Laser Diffraction pattern, with TEM image inserted, (b) XRD of CeO2 nanoparticles.
assembled CeO2 nanostructure reported in literature (Wang and Feng, 2003). XRD analysis verified the purity and crystallinity of CeO2 (Fig. 1(b)). The peaks of diffraction angle (28.6 , 33.1 , 47.6 , 56.5 , 76.9 , 79.2 , and 88.3 ) match the maximum diffraction angles and relative intensities of ceria from the Trace database (PDF number 75-0390) with primary particle size of 54 nm, confirming the CeO2 nanoparticles are ceria particles with nanocrystalline structure.
4.2.
Stability of CeO2 nanoparticles
The electrophorotic stability of CeO2 was studied by measuring the zeta potential across a range of pH (2.83e10.23) at ionic strength of 1 and 10 mM NaCl (Fig. 2-a). These data show decrease in the magnitude of the zeta potential with increasing ionic strength due to the compression of electrical double layers. As the pH increased from 2.8 to 10.23, zeta potential changed from 4.41 mV to 28.8 mV in 1 mM NaCl solution, and from 1.98 mV to 39.7 mV in 10 mM NaCl solution. Aggregation of CeO2 nanoparticles was studied by measuring the hydrodynamic size distribution after adding
Fig. 2 e (a) Zeta potential and electrophoretic mobility of CeO2 nanoparticles as a function of pH at two levels of ionic strength (1 mM and 10 mM) (b) Intensity-weighted size distribution of CeO2 in suspension under variable electrolyte concentration and pH.
electrolyte or changing pH followed by sonication for 15 min and samples were left undisturbed up to 24 h. At pH 6, the size distribution of CeO2 nanoparticles did not show significant change, indicating little aggregation occurred (Fig. 2-b). However, at pH 3, which approaches the point of zero charge (pHZPC), size distribution become wider with multiple peaks, indicating that aggregation and polydispersity occurred due to the formation of bi- and trimodal systems.
4.3.
Transport and deposition of CeO2 in porous media
4.3.1.
Screening study
A factorial screening experimental design was first conducted to study the influence of water chemistry (NaCl concentration of 1, 10 and 100 mM, and pH 3, 6, and 9). Effluent stream concentration of CeO2 nanoparticles were monitored and breakthrough curves are presented in Fig. 3. The normalized effluent nanoparticle concentration (C/C0) is shown as a function of cumulative volume normalized to pore volume. The breakthrough curves of CeO2 nanoparticle suspensions through porous media differ distinctively for different solution chemistry. The results are summarized into three cases. Case (1) Breakthrough during deposition phase, where sharp breakthrough curves were observed at one pore volume for two of the test conditions at pH ¼ 6, 1 mM NaCl, and pH 9, 10 mM NaCl. The normalized value of
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 0 9 e4 4 1 8
4413
breakthrough occurred during deposition phase, indicating all the nanoparticles were captured by the sand. However, sharp peaks were observed during the re-entrainment phase, when the ionic strength of the column was lowered by introducing MilliQ water and the pH was gradually brought back to 5.6 0.5 (the pH of MilliQ water). This indicates the detachment of weakly attached nanoparticles that were previously captured during the deposition phase by the porous media. Case (3) There were no breakthrough during deposition phase and no peaks were observed during reentrainment phase. For test runs at pH of 3, no CeO2 nanoparticles were detected in the effluent streams during deposition phase or re-entrainment phase, indicating stronger attachment of CeO2 to sand bed at lower that was not reversed at neutral pH or lowering the ionic strength.
4.3.2.
Fig. 3 e A profile of experimental breakthrough curves on an experimental design of 9 runs at three pH levels (pH 3, 6, and 9) and under ionic strength of a) 1 mM, b) 10 mM, and c) 100 mM. The lines represent model predictions.
C/C0 showed a step increase from the initial value of zero to a plateau of 0.9 0.5. During the rinse phase, the normalized effluent stream concentration decreased sharply to almost zero when particle free solution was fed to the column, indicating little or no deposition occurring at these two conditions. Case (2) No breakthrough was observed during deposition phase, but peaks were observed during reentrainment phase: For test runs operated at conditions of pH ¼ 6, 10 or 100 mM NaCl, pH ¼ 9, 10 or 100 mM NaCl, no
The role of ionic strength
The screening study has shown significant difference between 1 mM and 10 mM NaCl (Fig. 3a and 3b), while higher levels of ionic strength showed little effects (Fig. 3c). The results of rigorous test runs made at pH 6 with ionic strengths of 1, 2, 3, 5, and 10 mM for understanding the effects of ionic strength are presented in Fig. 4. During the deposition phase, fast and complete breakthrough curves were observed at ionic strength of 1 mM and 2 mM, where the normalized effluent concentration rapidly reached to 0.96 0.4. Whereas partial breakthroughs and gradual increase in C/C0 were observed with decreasing slope and plateau values as the ionic strength changed from 2, to 3 and 5 mM. Instead of fast, steep breakthrough as observed before, the normalized nanoparticle concentrations increased gradually to a plateau. The specific shape of the partial breakthrough can be explained by the blocking effect (Chen et al., 2002; Kuhnen et al., 2000; Song and Elimelech, 1994). As seen in Fig. 2, at pH 6, CeO2 NPs were negatively charged from 1 to 10 mM NaCl, and the size distribution of CeO2 NPs didn’t widen for 24 h duration, indicating that stable particleeparticle interaction predominates in these conditions. In this case, only a monolayer of deposited particles is formed on the sand surface and multi layer deposition could be neglected. Decrease in ceria deposition rate was reflected by the gradual increase in C/C0 with time, forming partially breakthrough. The maximum surface coverage was increased due to the increased screening of particle surface charge as could be seen in table with values of Smax of 0.1, 0.2, 1.2 mg/g corresponding to increased ionic strength of 3, 5 and 10 mM in runs 11, 12 and 6, respectively. During the rinse phase, C/C0 values decreased sharply to zero as particle free solution was fed to the column and during the re-entrainment phase a pulse increased with narrow, sharp peaks were observed. The peak height increased with increase in ionic strength of the influent nanoparticle suspensions. The attachment efficiency, a, and the deposition rate coefficient kexp were determined using equation (4) and 5 for each transport experiment. Mathematical expressions based on mass balance were developed to evaluate the mass fraction of nanoparticles recovered during the rinse phase, phase 2, and the re-entrainment phase, phase 3 (Franchi and O’Melia, 2003). The fractions of nano-CeO2 recovered during the two phases, FRC2 and FRE3 are defined as:
4414
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 0 9 e4 4 1 8
P Mass of NPs Recovered During Phase 2 FRC2 ¼ P Mass of NPs Deposited During Phase 1
ð5Þ
P
FRC3 ¼ P
Mass of NPs Recovered during Phase 3 Mass of NPs Deposited During Phase 1 Mass of NPs Recovered during Phase 2
Fig. 5-a shows attachment efficiency versus ionic strength at two distinct regions. At low ion concentrations (1e5 mM), where the attachment efficiency and deposition rate coefficient increased proportionally with increasing NaCl concentration, whereas at high ion concentrations above 10 mM deposition rate coefficient and the attachment efficiency were independent of NaCl concentration. Fig. 5-b presents logarithmic values of FRC2 and FRE3 as a function of log concentrations of NaCl. FRC2 values are in the range of 0.803w0.946 at ionic strength of 3 mM and lower, indicating almost complete detachment of nanoparticles during the rinse phase without changing the water chemistry. When the ionic strength increase to 5 mM and higher, there was a sharp drop in the FRC2 value from 0.9 to 0.105, indicating large retention of nanoparticles occurred in Phase 1. The values of FRE3 decreased as ionic strength increased. When the ionic strength was 1 mM and 2 mM, the corresponding values of FRE3 were0.694 and 0.390, respectively, indicating the nanoparticles deposited in phase 1 were subsequently reentrained. As ionic strength increased to 3 mM and above, FRE3 values dropped to below 0.033. Batch CeO2 nanoparticle suspensions left undisturbed showed little aggregation even after 24 h at pH 6, 1e100 mM NaCl. This suggests that, although particle aggregation and deposition kinetics are closely related, the hindered transport at higher ionic strength in this study is not caused by nanoparticle aggregation. Unlike other similar studies, such as fullerence C60 nanoparticles (Brant et al., 2005), where
(6)
deposition is related to particle aggregation induced by the screening effect of electrolyte concentration, these effects were less obvious with CeO2 nanoparticles. The increase of electrolyte concentration, at pH of 6, significantly increased nanoparticles attachment and deposition to the bed matrix, although there are limited aggregation of CeO2 nanoparticles. Derjaguin-Landau-Verwey-Overbeek (DLVO) theory (Derjaguin and and Landau, 1941; Verwey and Overbeek, 1948) was used to explain the particle-sand grain interaction energy at each ionic strength. The classical DLVO theory of colloidal stability describes the total interaction energy experienced by nanoparticles when approaching a collector surface as the sum of van der Waals (VDW) and electrical double layer (EDL) repulsion. Theoretical analysis was conducted with a spheareplate interaction to calculate the total interaction energy as a nanoparticle approaches sand surface (Derjaguin and Landau, 1941; Elimelech and O’Melia, 1990; Gregory, 1981; Healy and White, 1978; Hogg et al., 1966). The variation of the DLVO interaction energy with separation distance at different ionic strengths at pH 6 is shown an insert in Fig. 4. Transport behavior of CeO2 nanoparticles at various levels of ionic strength is in qualitative agreement with the DLVO theory (DLVO theory calculation is provided in the Supplimentary Material). At low ionic strength, diffusion layer surrounding nanoparticles and collector surfaces cause the screening effect of the salt is smaller than the electrostatic repulsion between particles and sand surface. For solutions with ionic strength of 1 and 2 mM, calculated values predict
Fig. 4 e The influence of ionic strength on the breakthrough curves at pH [ w6. The lines represent model prediction.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 0 9 e4 4 1 8
Fig. 5 e Effects of ionic strength on deposition and reentrainment of CeO2 nanoparticles, showing (a) attachment efficiency a, rate constant, and kexp, (b) mass fractions of recovered CeO2 during phase II (FRC2,) and Phase III (FRE3).
the presence of a substantial repulsive energy barrier to deposition ranging from 33 to 19 kT at 1 mM and 2 mM, respectively. The high-energy barrier suggested that only a small fraction of the particles can overcome the primary energy barrier and deposit onto sand surface. At conditions when complete breakthrough was observed, the attachment efficiency and deposition rate constants increased with the increase of ionic strength and strong electrostatic repulsion effectively hindered attachment of nanoparticles to the sand surface. Low attachment efficiency results smaller deposition rate and the deposition kinetic becomes reaction-limited. As the ionic strength increased, the diffuse layers of sand surface and nanoparticles are progressively compressed and consequently, electrostatic repulsion is reduced. At solution ionic strength of 3e5 mM, the interaction energy calculations indicate small energy barrier to deposition, also the secondary minimum is deeper and located at a closer distance of separation between sand surface, suggesting that mechanism can be a combination of disposition in both primary minimum and secondary minimum. This is supported by the observation that only a portion of the particles remaining in the porous media after Phase 2 attached to the sand surface. Partial breakthrough of the column studies suggests as CeO2 nanoparticles approach silica sand, they experience attractive
4415
force and deposition occurred at the shallow secondary minimum well, where the nanoparticles are continuously captured and released by the sand surface. The slope and plateau value of the breakthrough curves decrease with the increase of the ionic strength because the depth of decreasing secondary minimum. The deposition rate coefficient and the attachment efficiency increased when ionic strength increased from 3 to 5 mM, and became independent of ionic strength at 10 mM and above. The deposition rate coefficient is independent of the ionic strength at high ion concentrations duet to the reduction of electrostatic repulsion between nanoparticles and porous matrix surface. At these conditions, more collisions result in attachment, and the process was fast and the kinetic is limited by mass transfer. Analysis of effluent stream during Phase 2 indeed deposited in secondary energy-minimum. During Phase 3, sharp peaks of pulse release of nanoparticles were observed, which is vident of the deposition in secondary minimum. The introduction of MilliQ water to the column during Phase 3 eliminated the presence of secondary minimum, resulting rapid release of nanoparticles previously deposited at higher ionic strengths from sand surface. Fig. 5(b) shows the fraction of deposited particles that was eluted during Phase 3, given as FRE3, decreased with the increase of ionic strength. This indicates that as the ionic strength increased, the contribution of secondary minimum became smaller and more fraction of nanoparticles were deposited in the primary energy well. While the role of the calculated secondary minimum provided theoretically explanation for the deposition and reentraiment of CeO2, it needs to be noted that certain limitations of the DLVO calculation here. For instance, the zeta potential was used in the calculation as estimation for surface charge. Also sphere-to-plate interaction energy profiles are sensitive to changes in particle size. The height of the repulsive energy barrier and the depth of the secondary energy well both change with changing particle diameter. In this study the energy barrier height and the depth of secondary minimum calculation are based on the average diameter of the NPs. Due to the wide size distribution and polydispersitvity of the NPs, as mentioned before, the energy interaction profile can only represent the majority of the NPs in the suspension and a small fraction of NPs that are either very small or large cannot be represented here. Calculations of the theoretical predictions assume that the particles are spherical, while both the particles and collectors selected for this study are likely to be considerably more spherical than the majority of those encountered in realistic situations, neither the CeO2 NPs nor the sand grains are perfectly spherical. The angularity of the collector grain shape could also contribute to the removal of NPs by physical straining effect. These limitations could cause the deviation between the DLOV theoretical prediction and the actual deposition behaviors.
4.3.3. model
Predicting transport of nano-CeO2 with mathematical
The mathematical model for the transport and retention of CeO2 nanoparticles was optimized to estimate the breakthrough under corresponding experimental conditions. The values of kmod and Smax were determined by solving Eq. (1) and Eq. (2) and the necessary initial and boundary conditions,
4416
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 0 9 e4 4 1 8
Table 1 e Attachment and detachment coefficients at different operating conditions.
Run Run Run Run Run Run Run Run Run Run Run Run
1 2 3 4 5 6 7 8 9 10 11 12
Ionic strength
pH
Initial CeO2 Concentration mg/L
Smax (mg/g)
kmod (1/Hr)
kexp (1/Hr)
kdet (1/Hr)
10 mM 1 mM 100 mM 1 mM 10 mM 10 mM 1 mM 100 mM 100 mM 2 mM 3 mM 5 mM
9 6 9 9 3 6 3 6 3 6 6 6
10 10 50 50 50 50 10 50 50 50 50 50
0.1 1.0 1.2 1.0 1.0 1.2 0.9 1.1 1.0 1.0 0.1 0.2
24.5 0.3 34.9 0.1 27.0 25.1 20.2 25.6 27 0.5 2.3 8.4
26.1 0.70 30.7 0.1 30.7 23.37 22.5 27.6 31.4 0.77 1.91 11.62
0.7 0.0 1.2 0.1 0.0 0.6 0.0 0.8 0.0 0.1 0.1 0.1
Eq. (3), for the Deposition Phase, while assuming kdet is negligible. The effect of kdet was then considered for the whole column run while optimizing the value of kdet in the Reentrainment Phase. The initial conditions presented in Eq.(3) apply only for the first phase. For the second and third phases, the initial conditions used are the final distributions calculated for both CeO2 concentrations in solution (C ) and associated to the solid phase (S ) from the earlier phase. The model successfully captured the sharp increase and decline of the CeO2 concentration in the first two phases. There was a strong agreement of the deposition rate coefficients calculated from transport experimental data, kexp, and predicted by the model, kmod,(see Fig. 5-a), confirming successful simulation of the deposition process. The dispersion term in Equation (1) accounts for broadening the elution curve. For conditions where breakthrough of the nano-CeO2 occurred this term was insignificant (Chen et al., 2006). The volumetric water content was determined independently from physical measurements in the column. The results are plotted together with the experimental data in Figs. 3 and 4. Model predictions of effluent concentration fitted well with experimental data for Phases 1 and 2, however, the model parameter optimization failed to converge and did not fully capture the sharp increases and drop of effluent curve during Phase 3. Although, previous studies have shown steady-state condition of the first deposition phase (Liu et al., 2009) both first phases (Hydutsky et al., 2007; Li et al., 2008) or a pulse injection in the column (Hanna et al., 2010), the modeling effort for attachment and detachment of nanoparticles with changes in the fluid chemistry during the deposition and reentrainment phase is a unique contribution of this study. The mathematical modeling efforts have the advantage in predicting the amount and distribution of CeO2 held within the columns during all the phases. Particle deposition continues to grow within the column during the test run of pH 6, and ionic strength of 3 mM. As shown in Table 1, the maximum attachment concentration occurred at the column inlet and the amount increased during the deposition phase. The attachment distribution continued to grow linearly along the length of the column with the highest deposition observed near the inlet. Deposition within the column redistributed suspended particles to sand surface during Phase 2. The same behavior was observed in run 12 (pH 6, ionic strength 5 mM),
where higher deposition rates were observed. The maximum attachment at the inlet calculated for this run was 0.085 mg/g maintaining a ratio of 10% of the maximum deposition Smax (0.2 mg/g) similar to the previous run. For the runs without breakthrough the deposition was also maximum near the inlet. The deposited mass of nanoparticles dropped rapidly by an order of magnitude at half the length of the column. For instance, in run 9 the maximum attachment calculated was 0.35 mg/g which is 35% of the total Smax. On the other hand, with full breakthrough, i.e. run 2, the deposition was almost negligible. Experimental effort was made to validate the distribution of deposited nanoparticles in the column by taking portions of the bed along the length of the column, washing and sonicating in MilliQ water to detach deposited nano-CeO2. Unfortunately, desorbed silica interfered strongly with the analysis. A selective and quantitative technique is needed that can give 3-D distribution of deposited particles.
5.
Conclusion
Accidental or deliberate introduction of CeO2 nanoparticles into subsurface environments may lead to contamination of drinking water supplies and can act as colloidal carriers for sorbed contaminants. This study highlights the implication of CeO2 nanoparticle and provides important insights to confirm the mobility of CeO2 nanoparticle under typical ground water movements. It is not straightforward to precisely predict the behavior of CeO2 nanoparticle upon their environmental release because of the compliance of various environmental parameters. However, the values of attachment efficiency, deposition rate coefficients, fraction recovered/re-entrained at varying ionic strength and the modeling results provided in this study can be used to further estimate the relative mobility and evaluate the potential exposure and risk of CeO2 nanoparticles. Results of column studies clearly showed that the water chemistry governs the transport, deposition and reentrainment of nanoparticles. The increase of ionic strength decreased the mobility of CeO2 nanoparticles due to the compression of electrostatic double layer repulsion, which is in general agreement with DLVO theory prediction. The three phase flow method allowed better understanding of the
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 0 9 e4 4 1 8
influence of ionic strength in determining the condition whether the capture-release dynamics is reaction-limited or transport limited. A mathematical model was developed and successfully simulates both complete and partial breakthrough and re-entrainment of the CeO2 breakthrough curves. The model is also capable of predicting the distribution of nanoparticle deposition within the porous media.
Acknowledgments The authors are grateful for the financial support from U.S. Environmental Protection Agency under contract No. PRC108-1170.
Appendix. Supplementary data Supplementary data related to this article can be found online at doi:10.1016/j.watres.2011.05.025.
references
Aitken, R.J., Chaudhry, M.Q., Boxall, A., Hull, M., 2006. Manufacture and use of nanomaterials: current status in the UK and global trends. Occupational Medicine 56 (5), 300e306. Baun, A., Hartmann, N., Grieger, K., Kusk, K., 2008. Ecotoxicity of engineered nanoparticles to aquatic invertebrates: a brief review and recommendations for future toxicity testing. Ecotoxicology 17 (5), 387e395. Brant, J., Lecoanet, H., Wiesner, M.R., 2005. Aggregation and deposition characteristics of fullerene nanoparticles in aqueous systems. Journal of Nanoparticle Research 7 (4), 545e553. Bystrzejewska-Piotrowskaa, G., Golimowskib, J., Urbana, P.L., 2009. Nanoparticles: their potential toxicity, waste and environmental management. Waste Management 29, 2587e2595. Chen, J.Y., Klemic, J.F., Elimelech, M., 2002. Micropatterning microscopic charge heterogeneity on flat surfaces for studying the interaction between colloidal particles and heterogeneously charged surfaces. Nano Letters 2 (4), 393e396. Chen, K.L., Mylon, S.E., Elimelech, M., 2006. Aggregation kinetics of alginate-coated hematite nanoparticles in monovalent and divalent electrolytes. Environmental Science and Technology 40 (5), 1516e1523. Corma, A., Atienzar, P., Garcia, H., Chane-Ching, J.-Y., 2004. Hierarchically mesostructured doped CeO2 with potential for solar-cell use. Nature Materials 3 (6), 394e397. Derjaguin, B.V., Landau, L., 1941. Acta Physicochimca URSS. Dunphy Guzma´n, K.A., Taylor, M.R., Banfield, J.F., 2006. Environmental risks of nanotechnology: National nanotechnology initiative funding, 20002004. Environmental Science and Technology 40 (5), 1401e1407. Elimelech, M., O’Melia, C.R., 1990. Effect of particle size on collision efficiency in the deposition of Brownian particles with electrostatic energy barriers. Langmuir 6 (6), 1153e1163. Franchi, A., O’Melia, C.R., 2003. Effects of natural organic matter and solution chemistry on the deposition and reentrainment of colloids in porous media. Environmental Science and Technology 37 (6), 1122e1129.
4417
Fu, Q., Weber, A., Flytzani-Stephanopoulos, M., 2001. nanostructured AueCeO2 catalysts for low-temperature wateregas shift. Catalysis Letters 77 (1), 87e95. Gregory, J., 1981. Approximate expressions for retarded Van der waals interaction. Journal of Colloid and Interface Science 83 (1), 138e145. Hanna, K., Rusch, B., Lassabatere, L., Hofmann, A., Humbert, B., 2010. Reactive transport of gentisic acid in a hematite-coated sand column: experimental study and modeling. Geochimica et Cosmochimica Acta 74 (12), 3351e3366. Healy, T.W., White, L.R., 1978. Ionizable surface group models of aqueous interfaces. Advances in Colloid and Interface Science 9 (4), 303e345. Hogg, R., Healy, T.W., Fuerstenau, D.W., 1966. Mutual coagulation of colloidal dispersions. Transactions of the Faraday Society 62, 1638e1651. Hydutsky, B.W., Mack, E.J., Beckerman, B.B., Skluzacek, J.M., Mallouk, T.E., 2007. Optimization of nano- and microiron transport through sand columns using polyelectrolyte mixtures. Environmental Science and Technology 41 (18), 6418e6424. Johnston, B.D., Scown, T.M., Moger, J., Cumberland, S.A., Baalousha, M., Linge, K., van Aerle, R., Jarvis, K., Lead, J.R., Tyler, C.R., 2010. Bioavailability of nanoscale metal oxides TiO2, CeO2, and ZnO to Fish. Environmental Science and Technology 44 (3), 1144e1151. Kosynkin, V.D., Arzgatkina, A.A., Ivanov, E.N., Chtoutsa, M.G., Grabko, A.I., Kardapolov, A.V., Sysina, N.A., 2000. The study of process production of polishing powder based on cerium dioxide. Journal of Alloys and Compounds 303-304, 421e425. Kuhnen, F., Barmettler, K., Bhattacharjee, S., Elimelech, M., Kretzschmar, R., 2000. Transport of iron oxide colloids in packed quartz sand media: monolayer and multilayer deposition. Journal of Colloid and Interface Science 231 (1), 32e41. Lecoanet, H.F., Bottero, J., Wiesner, M.R., 2004. Laboratory assessment of the mobility of nanomaterials in porous media. Environmental Science and Technology 38 (19), 5164e5169. Li, Y., Wang, Y., Pennell, K.D., Abriola, L.M., 2008. Investigation of the transport and deposition of Fullerene (C60) nanoparticles in quartz sands under varying flow conditions. Environmental Science and Technology 42 (19), 7174e7180. Limbach, L.K., Li, Y., Grass, R.N., Brunner, T.J., Hintermann, M.A., Muller, M., Gunther, D., Stark, W.J., 2005. Oxide nanoparticle uptake in human lung fibroblasts: effects of particle size, agglomeration, and diffusion at low concentrations. Environmental Science and Technology 39 (23), 9370e9376. Lin, W., Huang, Y., Zhou, X., Ma, Y., 2006. Toxicity of cerium oxide nanoparticles in human lung cancer cells. International Journal of Toxicology 25 (6), 451e457. Liu, X., Wazne, M., Christodoulatos, C., Jasinkiewicz, K.L., 2009. Aggregation and deposition behavior of boron nanoparticles in porous media. Journal of Colloid and Interface Science 330 (1), 90e96. Livingston, F.E., Helvajian, H., 2005. Variable UV laser exposure processing of photosensitive glass-ceramics: maskless microto meso-scale structure fabrication. Applied Physics A: Materials Science and Processing 81 (8), 1569e1581. McDowell-Boyer, L.M., 1992. Chemical mobilization of micron-sized particles in saturated porous media under steady flow conditions. Environmental Science and Technology 26 (3), 586e593. Navarro, E., Baun, A., Behra, R., Hartmann, N., Filser, J., Miao, A.-J., Quigg, A., Santschi, P., Sigg, L., 2008. Environmental behavior and ecotoxicity of engineered nanoparticles to algae, plants, and fungi. Ecotoxicology 17 (5), 372e386. Nowack, B., Bucheli, T.D., 2007. Occurrence, behavior and effects of nanoparticles in the environment. Environmental Pollution 150 (1), 5e22. Saiers, J.E., Hornberger, G.M., Liang, L., 1994. First- and secondorder kinetics approaches for modeling the transport of
4418
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 0 9 e4 4 1 8
colloidal particles in porous media. Water Resource Research 30 (9), 2499e2506. Song, L., Elimelech, M., 1994. Transient deposition of colloidal particles in heterogeneous porous media. Journal of Colloid and Interface Science 167 (2), 301e313. Thill, A., Zeyons, O., Spalla, O., Chauvat, F., Rose, J., Auffan, M., Flank, A.M., 2006. Cytotoxicity of CeO2 nanoparticles for Escherichia coli. Physico-chemical insight of the cytotoxicity mechanism. Environmental Science and Technology 40 (19), 6151e6156. Tufenkji, N., Elimelech, M., 2003. Correlation equation for predicting single-collector efficiency in physicochemical
filtration in saturated porous media. Environmental Science and Technology 38 (2), 529e536. Verwey, E., Overbeek, J., 1948. Theory of the Stability of Lyophobic Colloids. Elsevier, Amsterdam. Wang, Z.L., Feng, X., 2003. Polyhedral shapes of CeO2 nanoparticles. The Journal of Physical Chemistry B 107 (49), 13563e13566. Wolfram, S., 1991. Mathematica: A System for Doing Mathematics by Computer, second ed. Addison Wesley Longman Publishing Co., Inc, Redwood City, CA, USA. Yao, K., Habibian, M.T., O’Melia, C.R., 1971. Water and waste water filtration. Concepts and applications. Environmental Science and Technology 5 (11), 1105e1112.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 1 9 e4 4 2 7
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Kinetics of model high molecular weight organic compounds biodegradation in soil aquifer treatment Peter Fox, Roshan Makam* Department of Civil and Environmental Engineering, Arizona State University, PO Box 5306, Tempe, AZ 85287-5306, USA
article info
abstract
Article history:
Soil Aquifer Treatment (SAT) is a process where treated wastewater is purified during
Received 6 December 2010
transport through unsaturated and saturated zones. Easily biodegradable compounds are
Received in revised form
rapidly removed in the unsaturated zone and the residual organic carbon is comprised of
14 May 2011
primarily high molecular weight compounds. This research focuses on flow in the satu-
Accepted 22 May 2011
rated zone where flow conditions are predictable and high molecular weight compounds
Available online 15 June 2011
are degraded. Flow through the saturated zone was investigated with 4 reactors packed with 2 different particle sizes and operated at 4 different flow rates. The objective was to
Keywords:
evaluate the kinetics of transformation for high molecular weight organics during SAT.
Soil aquifer treatment
Dextran was used as a model compound to eliminate the complexity associated with
High molecular weight organics
studying a mixture of high molecular weight organics. The hydrolysis products of dextran
Hydrolysis
are easily degradable sugars. Batch experiments with media taken from the reactors were
Dextran
used to determine the distribution of microbial activity in the reactors. Zero-order kinetics
Kinetics
were observed for the removal of dextran in batch experiments which is consistent with
Saturated
hydrolysis of high molecular weight organics where extracellular enzymes limit the substrate utilization rate. Biomass and microbial activity measurements demonstrated that the biomass was independent of position in the reactors. A Monod based substrate/ biomass growth kinetic model predicted the performance of dextran removal in the reactors. The rate limiting step appears to be hydrolysis and the overall rate was not affected by surface area even though greater biomass accumulation occurred as the surface area decreased. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
Water Reuse has become important in arid areas throughout the world including regions such as the southwestern United States and the Mideast. Soil aquifer treatment (SAT) is a technology where natural systems have been employed to treat wastewater for indirect potable reuse (Bouwer and Rice, 1984; Bouwer, 1985; Amy et al., 1993; Drewes and Jekel, 1998; Wilson et al., 1995, Wild and Reinhard, 1999). SAT involves
the water quality benefits derived during percolation through vadose zone sediments and subsequent ground water transport. The organic matter in treated wastewater is a complex mixture of simple carbohydrates, amino acids, alcohols, volatile fatty acids mixed with polymers and heteropolymers including proteins (1/3 of COD), polysaccharides (1/5 of COD) and lipids (1/3 of COD) along with Natural Organic Matter (NOM) and Soluble Microbial Products (SMPs) (Raunkjaer et al., 1994). In the unsaturated zone, easily biodegradable low
* Corresponding author. Biotechnology Department, PES Institute of Technology, 82 East End ’B’ Main Road, Jayanagar 9th Block, Bangalore, Karnataka 560069, India. Tel.: þ91 80 26633721. E-mail addresses:
[email protected] (P. Fox),
[email protected] (R. Makam). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.05.023
4420
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 1 9 e4 4 2 7
molecular weight organics are biodegraded while residual high molecular weight organics are biodegraded over longer time periods in the saturated zone (Fox et al., 2001; Amy et al., 2006). The removal of dissolved organic carbon may be modeled as a first order reaction where the organic components are divided into fractions with different biodegradation rates. Dissolved organic carbon (DOC) is often used as a surrogate to monitor the removal of organic compounds without consideration for the characteristics of individual organic compounds. During saturated flow, the majority of organic carbon transformations are associated with high molecular weight organics. The high molecular weight fraction cannot be assimilated directly by microorganisms (Levine et al., 1985). Bacteria assimilate high molecular weight compounds via hydrolysis by extracellular enzymes. These extracellular enzymes are either bound to the cell surface (ecto-enzymes) (Chrost, 1991) or released (exoenzymes) (Vetter and Deming, 1999; Wetzel, 1991) into the medium so as to hydrolyze high molecular weight compounds. Cadoret et al. (2002) has demonstrated the utilization of low and high molecular weight substrates as a consequence of extracellular enzymes in whole and dispersed activated sludges (e.g.: azocasein MW ¼ 26,000, amulose azure 32,000 < MW < 86,000). Slow hydrolysis of organic matter by exoenzymes has been demonstrated by Eliosov and Argaman (1995) and Guellil et al. (2001). When substrate utilization kinetics for high molecular weight DOC are dominated by hydrolysis of complex substrates, zero-order kinetics with respect to substrate concentration may be observed. This research focuses on understanding the rate limiting step for the biodegradation of DOC in the saturated zone during SAT. These transformations are vital for the sustainability of SAT and they support microbial activity for long sub-surface travels times allowing for potential co-metabolic reactions of trace anthropogenic compounds (Nalinakumari et al., 2010; Wild and Reinhard, 1999). The objective of this research is to evaluate the kinetics of biodegradation of a model high molecular weight compound during flow through a saturated media. A model compound was chosen to avoid the complexities associated with a mixture of many compounds and provide insight into the mechanisms of removal for a high molecular weight compound. To accomplish this, two different particle size sands were used to provide two different surface areas for microbial attachment and the flow rates were varied. The experimental design maintained aerobic conditions to eliminate complexities from varying redox conditions.
2.
Methods
2.1.
Dextran as a model compound
Dextran was chosen as the model compound because it is a high molecular weight compound (MW ¼ 10,000), mimics polysaccharide soluble microbial products and is readily biodegradable. The hydrolysis products of dextran are sugars that easily biodegradable. Analysis of carbohydrates can be used for monitoring dextran and its hydrolysis products. When combined with dissolved organic carbon measurements, a mass balance on dextran can be done.
2.2.
Reactors
The experimental apparatus was designed to simulate saturated flow in a sand aquifer. The experimental apparatus consisted of 4 cylindrical reactors packed with with two different particle sizes of sand. The reactors were 0.915 m tall with a 0.076 m inner diameter and were constructed of Plexiglass. Two different clean silica sand sieve sizes were used as packing material (Border Products, Arizona). One size was US standard sieve 16 x 30 with a geometric mean diameter of 0.6 mm and the second size was US standard sieve 40 x 60 with a geometric mean diameter of 0.353 mm. The sands were washed with de-ionized water to remove any residual fines and dried before packing. The sands were packed in the reactors to achieve an average dry packing density of 1.4 g/ cm3. Reactors 1 and 4 were packed with 0.6 mm silica sand and Reactors 2 and 3 were packed with 0.353 mm silica sand. The reactors were operated in an upflow mode. The 4 reactors were seeded by feeding filtered nitrified/ denitrified effluent from the Mesa Northwest Reclamation Plant, Arizona for a period of 50 days. After seeding, the reactors were operated with synthetic feed containing dextran (average M.W. ¼ 10,000) (Sigma, St. Louis, Missouri, USA) as the substrate. The synthetic feed had a nominal concentration of 6.8 mg DOC/L added to dechlorinated drinking water. The background DOC of the dechlorinated tap water used to formulate the influent feed was found to be at 1.26 þ 0.4 mg DOC/L. During previous studies using the same tap water in soil columns, less than 0.2 mg/L of the DOC was removed over a 30 day retention time (Nalinakumari et al., 2010). During this study, the retention times were much lower and therefore the natural organic matter in the tap water should not significantly influence the results. Based on the stoichiometry, the tap water contained sufficient nutrients and there was no need add supplemental nutrients. Weekly monitoring of turbidity, UVA254 and dissolved organic carbon was done. The flow rates chosen for simulating sub-surface transport provided a Reynolds Number less than 1 and the Peclet Number ranged from 0.2 to 6. Ground water recharge sites have saturated flow conditions with Reynolds Numbers less than 1 and for most aquifer materials the Peclet Numbers range from 0.2 to 36. The reactors were operated in two phases with different flow rates to evaluate a range of conditions that occur during saturated flow in SAT systems. During Phase 1, Reactors 1 and 2 were operated at 0.5 L/day and Reactors 3 and 4 were operated at 4 L/day. All 4 reactors were operated a minimum of 150 days under saturated aerobic conditions to achieve steady state. Once Phase I was completed, Phase II was initiated. During Phase II, the flow rate to Reactors 1 and 2 was increased 4 fold to 2 L/day and the flow rate to Reactors 3 and 4 was decreased by a factor of 4 to 1 L/day. All 4 reactors were run for at least 50 days under saturated aerobic conditions during Phase II. Table 1 shows the operating conditions for the four reactors during Phases 1 and 2. In this experimental design, the empty bed contact in Reactors 1 and 2 was the always identical and the empty bed contact time in Reactors 3 and 4 was always the same. Table 1 also shows the mean effluent concentrations under these conditions along with the student’s t-test results for comparing the reactors.
4421
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 1 9 e4 4 2 7
Table 1 e Effluent DOC Concentrations during Phase I and Phase II. Reactor 3
Particle Size (m 10 ) 3
3
1
2
3
4
0.600
0.353
0.353
0.600
Phase I
Flow rate (m /day 10 ) Empty Bed Contact Time (days) Reynolds Number (103) Peclet Number Mean Effluent Concentration (mg/L)
0.500 8.340 0.762 0.762 1.980
0.500 8.340 0.448 0.448 1.740
4.000 1.040 3.584 3.584 2.060
4.000 1.040 6.091 6.091 2.070
Phase II
Flow rate (m3/day 103) Empty Bed Contact Time (days) Reynolds Number (103) Peclet Number Mean Effluent Concentration (mg/L)
2.000 2.080 3.046 3.046 1.950
2.000 2.080 1.792 1.792 1.860
1.000 4.170 0.896 0.896 1.940
1.000 4.170 1.523 1.523 1.930
Student’s t-test
p-value (based on particle size) p-value (based on phase flow rate)
0.183 0.320
0.500 0.172
0.500 0.172
0.183 0.320
2.3.
Batch kinetic tests
A modified biodegradable dissolved organic carbon (BDOC) reactor method was used to measure the kinetics of dextran biodegradation using native sand originally acclimated to tertiary effluent. The native sand used was from the Agua Fria River Basin (passed through a 2 mm screen) at a planned recharge site in Arizona, USA. In addition batch tests were completed with the clean silica sands used in the four reactors. The BDOC sand reactors were 500 ml Erlenmeyer Flasks containing 100 g of biologically active sand in each of the reactors. The biologically active sand was seeded initially with the same Mesa Tertiary Effluent used to seed the reactors. The BDOC reactors were seeded with 300 ml aliquots of effluent for three sequential batch tests where each test was 5 days. The reactors were then acclimated with 300 ml of dextran at a target concentration of 5 mg-C/L. After each 5 day reaction period, the solution was decanted and the acclimated sand in the reactors was washed with 100 ml of solution containing 0.15 M Sodium Chloride with 1 mM Magnesium Chloride. The washing procedure was repeated 3 times before new feed was added to initiate a new batch experiment. The reactors were then incubated with 300 ml of dextran at a nominal concentration of 9.66 mg-C/L to simulate the synthetic feed used in the column studies. The initial concentration in the batch tests was 40% higher than the concentration used in the column studies to provide kinetic information over a larger concentration range. Initial samples were taken to measure the concentration of DOC at the beginning of each experiment after mixing with the sand and biomass. The DOC and UVA254 were monitored every day for a period of 5 days consistent with other BDOC tests (Cha et al., 2004). Also, 5 g of sand was removed daily from the batch reactors and the sand was analyzed for biomass using organic nitrogen and carbohydrate analysis. The results were normalized to the quantity of sand and biomass after each sampling event to account for the removal of the sand and biomass. The data obtained from DOC measurements were used for analyzing the kinetics of dextran utilization for each reactor. The data from organic nitrogen analysis was then used to calculate a yield coefficient. By normalizing the substrate utilization rate data to the biomass content, a specific substrate utilization ratio was
calculated. This value was then compared to values determined for the reactors.
2.4.
Biomass characrerization in the columns
Immediately after Phase II was completed, the sand from the column reactors was extruded and divided into 11 to 12 sections to provide a profile of the sand as a function of reactor length. Each section was approximately 0.076 m in thickness and the sand was analyzed for biomass composition by determining the organic nitrogen, carbohydrate and volatile suspended solid attached to the sand.
2.5.
Analytical methods
2.5.1.
UVA254
The ultraviolet absorbance (UVA) at 254 nm was routinely measured using a Model 8452 A Hewlett Packard Diode Array Spectrophotometer. A 1 cm pathlength was used with a quartz cuvette.
2.5.2.
Dissolved organic carbon
The dissolved organic carbon (DOC) was routinely monitored using a Shimadzu TOC 5000 A Total Organic Carbon Analyzer in accordance with Method 5310 in Standard Methods for Examination of Water and Wastewater (Andrew et al., 2007). The minimum detection level was 0.5 mg/L.
2.5.3.
Biomass extraction and Quantification
The extraction of attached biomass from column reactor media or BDOC reactor media was done using the following procedure. Approximately 5 g of wet media was weighed and transferred into a 10 ml volumetric flask and 3 ml of 20% (w/v) of tricholoroacetic acid (TCA) was added. The flasks were placed in a VWR Scientific Aqua Sonic Model 150 T ultrasonic cleaner and sonicated for 10 min. After sonication, the solution was decanted from the flask and transferred to a second 10 ml volumetric flask. The sand remaining in the flask was rinsed once with 10 ml of water and the rinse water was transferred to the second volumetric flask. The sonication and transfer steps were repeated two more times to complete the separation of attached biomass. The extracts were analyzed
4422
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 1 9 e4 4 2 7
for organic nitrogen and carbohydrate to characterize the biomass. The sand used in each analysis was then dried and weighed.
2.5.4.
Organic nitrogen analysis
The extracted biomass was quantified for organic nitrogen using a Hach Total Nitrogen kit with the Persulfate Digestion Method in accordance with Method 4500 in Standard Methods for Examination of Water and Wastewater (Andrew et al., 2007) and a Hach DR/4000 Spectrophotometer.
2.5.5.
Carbohydrate analysis
The extracted biomass was also quantified for carbohydrate content. A 1 ml sample was mixed with 1 ml of 5% (w/v) of aqueous phenol solution. Then 5 ml of concentrated sulfuric acid was added (Herbert et al., 1971). A calibration standard with 0e100 mg glucose was used. The carbohydrate was measured at 488 nm using a Model 8452 A Hewlett Packard Diode Array Spectrophotometer. A 1 cm pathlength was used with a quartz cuvette.
2.5.6.
Indirect dextran measurements
Dextran was measured indirectly using both DOC and carbohydrate analyses. The carbohydrate analysis measured both simple and complex carbohydrates and can provided a direct measurement of dextran before biodegradation. There was less than þ3% difference between dextran measurements using carbohydrate analysis as compared to dextran measurements using DOC analysis. The correlation between DOC and carbohydrates was found to be valid throughout the studies even though the carbohydrate analysis will measure sugars in biomass such as riboses. This correlation also demonstrated that measuring carbohydrate on biomass extracted from the sand media was primarily due to biomass and the influence of dextran was negligible.
2.5.7.
Data analysis
And the specific rate of substrate utilization is q ¼ qmax
S Ks þ S
(4)
Where qmax is the maximum specific substrate utilization rate (g-C/g-biomass-t) Assuming hydrolysis is the rate limiting step, Ks is very small since there are not many extracellular enzymes and the concentration of extracellular enzymes is constant (Guellil et al., 2001). The hydrolysis products of dextran are sugars that should be rapidly degraded at a rate faster than the hydrolysis rate. The specific growth rate (Eq. (1)) and specific rate of substrate utilization (Eq. (4)) becomes (5) m ¼ mmax q ¼ qmax ¼ k
(6)
where k ¼ zero-order substrate utilization rate constant (g-C/ g-biomass-day) Considering a fixed-bed and assumingplug flow, steady state conditions with zero-order kinetics with respect to substrate utilization and first order kinetics with respect to biomass growth, we get using the following definitions in the model: L ¼ Length of reactor (m), U velocity in direction of flow (m/day), S ¼ Substrate concentration (g-C/m3), X Biomass concentration (g-biomass/m3), Y Yield Coefficient (g-C/ g-biomass), k zero-order substrate utilization rate constant (g-C/g-biomass-day), b decay coefficient (day1), z axial direction (m), Xf Biofilm density (g/m3), Lf Biofilm thickness (m), q ¼ residence time (day), a specific surface area (m2/m3). Where the biomass concentration may be expressed in terms of the biofilm density and thickness as X ¼ Xf Lf a
(7)
Biofilm kinetics may be analyzed using a classic biofilm model (Fig. 1) where external mass transfer resistance and diffusion through the biofilm are considered. However, if
2.5.7.1. Monod substrate utilization/biomass growth kinetic model. A model was developed for substrate utilization and
Y
biomass growth in the column reactors studied. Monod in 1942 gave an empirical model (Eq. (1)) for microbial growth kinetics introducing the concept of a growth limiting substrate.
dX dS
YX=S dS ¼ YX=S q m¼ X dt
Lf
Bulk Liquid
Diffusion Layer
(1)
where m ¼ specific growth rate (day1), mmax ¼ maximum specific growth rate (day1), S ¼ substrate concentration (g-C/m3), Ks ¼ substrate saturation constant i.e., substrate concentration at half mmax (g-C/m3) In addition to this, Monod also related the Yield Coefficient Yx/s (Eq. (2)) to the specific biomass growth rate and the specific rate of substrate utilization q (Eq. (3)) YX=S ¼
Z
Biofilm
S Ks þ S
Xf
Sand Particle
m ¼ mmax
X
S
L
(2)
(3)
Fig. 1 e Biofilm Model e Flow is In the Z-Direction while Diffusion is in the X-Direction. As Dextran Approaches the Biofilm Extracellular Enzymes Can Hydrolyze the Dextran.
4423
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 1 9 e4 4 2 7
hydrolysis by extracellular enzymes is the rate limiting step, diffusion into the biofilm is only important for the products of hydrolysis. For dextran, the hydrolysis products are rapidly degraded and diffusion into the biofilm was assumed to not be rate limiting. A Steady State Substrate Mass Balance may be written assuming external mass transfer to the biofilm is negligible.
0¼V
dS ¼ dt
! dS U kXf Lf a V dz
DS=Dt ¼ ks whereDS ¼ mass substrate utilized over the time Dt
(13)
The yield of biomass per mass of substrate utilized substrate (YX/S) is obtained using equation 14 where DXX is the biomass growth over the time Dt. YX=S ¼ DXX=DS
(8)
Steady State Biomass Mass Balance 0 ¼ aV
The rate of substrate (S) utilization is obtained using Eq. (13).
dXf Lf ¼ ðYk bÞXf Lf aV dt
(9)
(14)
where YX/S yield of biomass on substrate (mg-X/mg-C) The specific substrate utilization rate (kbatch) for the batch experiment is obtained using Eq. (15). kbatch ¼ ðks =XXÞ ð300ml=1000ml=LÞ
(15)
Recognizing that the substrate utilization rate-k Xf is the rate of substrate utilization, then Eq. (9) reduces to Eq. (10) where J is the flux into the biofilm in gC/m2ed.
where kbatch specific substrate utilization rate (mg-C/mgbiomass-day) The specific growth rate is then obtained according to Eq. (16).
JY ¼ bXf Lf
m ¼ YX=S kbatch
(10)
which is the classic result for a steady state biofilm where growth equals decay. Solving Eq. (8) for X where X XfLfa yields X¼
ðS0 SÞU Lk
(11)
Since hydrolysis limits substrate utilization to zero-order kinetics, the model predicts biomass content becomes independent of position.
2.6.
(16) 1
where m specific growth rate (day ) For each particle size of soil, the kinetic parameters are tabulated in Table 2. The specific substrate utilization rate constants for column experiments (kcolumn) are then obtained according to Eq. 17 and listed in Table 3. kcolumn ¼ kbatch particle surface area 5=residence time (17) The predicted biomass from the model is then obtained using Eq. (18) and is listed in Table 3.
Batch experiments data analysis
Predicted Biomass ¼ X V Batch experiments were analyzed for substrate utilization and microbial growth kinetics. Analysis was done to determine the zero-order substrate utilization rate (ks). The specific substrate utilization rate (kbatch) was determined by normalizing to biomass attached to the sand which was measured in terms of organic nitrogen per gram of sand. The quantity of attached biomass (XX) in mg was calculated using organic nitrogen data assuming an empirical cell composition of C5H7O2N according to Eq. (12). In Eq. (12), Org-N is the measured organic nitrogen in mg-N/g-sand. XX ¼ ðOrg N=g sandÞ 100 g sand 113 mg C5 H7 O2 N=14 mgN
(12)
where XX ¼ amount of biomass formed (mg-X)
(18)
where V volume of reactor (L), X concentration of biomass according to Eq. (11) (mg-X/m3)
3.
Results and discussion
3.1.
Column/Reactor experiments with dextran
The influent and effluent DOC concentrations were monitored as a function of time for the four reactors and the average concentrations were used to evaluate the reactors for the two phases of operation. Effluent DOC was used as a surrogate to understand the removal of dextran in the soil columns since these values correlated with carbohydrate analysis as
Table 2 e Kinetic parameters from Batch Kinetic Experiments. Reactor
1 2 3 4
Particle Size (m 103)
Substrate rate constant ks (mg-C/L/day)
Final Biomass-N per gram of sand DN (mg-N/g-sand)
Amount of Biomass formed X (mg-X)
Amount of Substrate utilized S (mg-C)
Yield of Biomass on Substrate YX/S (mg-X/mg-C)
Specific substrate utilization rate kbatch (mg-C/mg-X/day)
Specific Growth Rate m (day1)
0.600 0.353 0.353 0.600
0.4426 0.8856 0.8856 0.4426
0.023 0.041 0.041 0.023
18.564 33.093 33.093 18.564
0.664 1.328 1.328 0.664
27.958 24.920 24.920 27.958
0.0072 0.0080 0.0080 0.0072
0.2 0.2 0.2 0.2
4424
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 1 9 e4 4 2 7
Table 3 e Biomass according to substrate utilization/microbial growth model and experimentation Reactor
1 2 3 4
Experimental Experimental Predicted Predicted Particle Flux Flow rate Particle Biofilm Size (m 103) (m3/day 103) surface area (mg-C/m2/day) Thickness (mm) (m2 106) 0.600 0.353 0.353 0.600
0.500 0.500 4.000 4.000
1.133 0.392 0.392 1.133
1.550 0.910 7.290 12.390
mentioned above. Phase I was run for a period of 150 days such that the DOC concentration became constant with time reflecting the development of an acclimated population and pseudo-steady state conditions. The reactors were operated under saturated and aerobic conditions. The feed dissolved oxygen (DO) was maintained near equilibrium with atmospheric oxygen with an approximate concentration of 6 mg/L. Aerobic conditions in the reactors were maintained as 0.7 mg of oxygen was needed per mg of dextran according to stoichiometry (Eq. (19)) and the average feed dextran concentration was 6.7 mg/L resulting in an effluent dissolved oxygen concentration greater than 2 mg/L. þ 0:042C6 H10 O5 þ 0:02HCO 3 þ 0:02NH4 þ 0:15O2 /0:02C5 H7 O2 N
þ 0:25CO2 þ 0:188H2 O (19) Routine monitoring of DO, Turbidity, UV254 and DOC was performed on the reactors. The mean feed concentration which includes the background DOC of tap water was 6.7 þ 0.5 mg DOC/L and the average effluent concentrations ranged from 1.70 to 2.07 mg DOC/L for Phase 1 (Table 1). During Phase 2, the effluent concentrations in all four reactors were very similar and ranged only from 1.86 to 1.95 mg DOC/L (Table 1). Also, listed in Table 1 are the Reynolds Number and Peclet Number corresponding to the mean effluent concentrations during Phase 1 and Phase 2. A Student’s t-test was performed on the mean effluent concentrations in the reactors based on particle size and flow rate for both the Phases
Fig. 2 e BDOC Kinetic Experiment with different sands.
0.004 0.001 0.010 0.028
Column Predicted Experiment Rate Biomass Biomass Constant (Eq (18)) (mg-X) kcolumn (mg-X) (mg-C/mg-X/L/day) 0.0176 0.0068 0.0541 0.1406
508.48 1363.38 1287.20 501.12
393.110 1185.680 1196.140 498.060
and was found to have no statistical difference between the various mean effluent concentrations.
3.1.1.
Batch kinetic experiments with dextran as substrate
Batch experiments with the modified BDOC reactors were completed using Agua Fria sand and two clean silica sands sieved to the same geometric diameter as used in the column reactors which was 0.353 mm and 0.6 mm. Prior to running the kinetic experiments the BDOC reactors were acclimated with Mesa Tertiary Effluent as described above. Acclimatization was determined complete when the final DOC concentration after each test was the same as the previous test. After acclimation, the kinetic experiments were carried out with a nominal initial dextran concentration 40% higher than the influent to the reactors. The results are summarized in Fig. 2. Fig. 2 shows that the substrate concentration decreases linearly as a function of time which is consistent with zeroorder kinetics for the different types of sand studied. i.e., Agua Fria sand, 0.353 mm silica sand and 0.6 mm silica sand. The UVA254 in the effluent increased from zero to 0.4 cm1 for 0.353 mm silica sand and from zero to 0.18 cm1 for 0.6 mm silica sand during the batch tests. The increase in UVA254 is consistent with the presence of aromatic compounds as the microorganisms were producing or desorbing soluble microbial products. The larger increase with the 0.353 mm silica sand is consistent with the larger surface area and biomass content. The increase in UVA254 was not enough to significantly impact the DOC concentration. The organic nitrogen data used to evaluate the growth of biomass
Fig. 3 e Organic Nitrogen data during the kinetic experiment with different sands.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 1 9 e4 4 2 7
4425
Biomass (C5H7O2N) (mg-X/g-sand)
Reactor Biomass 0.40 Reactor Reactor Reactor Reactor
0.35 0.30
3 4 2 1
0.25 0.20 0.15 0.10 0.05 0.00
0
2
4 6 8 Reactor Section (Bottom-Top)
10
12
Fig. 4 e Reactor Biomass versus Reactor Sections.
on the sand is shown in Fig. 3. The gradual increase with reaction time for the different types of sand is consistent with microbial growth. From the kinetic experiments with the BDOC reactors, the rate of DOC biodegradation kinetics was observed to be zero-order with respect to substrate concentration. The rate for the smaller particle size was found to be approximately twice that of the larger particle size. The smaller particle size had approximately three times the surface area of the larger particle size. Assuming the microbial population was primarily attached, the surface area appeared to directly influence the removal rate. The BDOC reactor experiments support a zero-order relationship for substrate utilization which might be expected when hydrolysis of high molecular weight compounds is the rate limiting step. The specific substrate utilization rate, yield of biomass on substrate and the specific growth rate were calculated according to the Batch Experiments Data Analysis and are listed in Table 2. The rate constants for the column experiments were then obtained in accordance with Eq. (17).
3.1.2.
Biomass data from columns
Once, the column experiments were completed the sand in the reactors were cut into 11 or 12 sections and were analyzed for biomass using Org-N, carbohydrate and volatile suspended solids. Fig. 4 shows the biomass profiles for the 4 reactors based on the Org-N measurements. The Organic-N profiles,
Fig. 5 e Activity Kinetics for Reactor 1.
Fig. 6 e Activity Kinetics for Reactor 2.
which represent a measure of the protein content of the cells was relatively constant along the length of the reactors. The organic-N could also represent extracellular polymers, nucleic acids and the organic-N does not distinguish between active cells and cellular debris. We see from Fig. 4 that the biomass growth was similar for reactors 1 and 4 and for reactors 2 and 3. This indicates that the biomass growth was independent of the applied substrate flux and was primarily affected by the surface area and hence the particle size. Table 3 lists the quantity of biomass according to the biomass growth model (Eq. (18)) and from the experimental investigation. We see from Table 3 that the biomass is proportional to the surface area in the reactors. The biomass increases by a factor of 3 for a 3-fold increase in individual particle surface area when comparing the reactors with 0.6 mm sand to reactors with 0.353 mm sand. Table 3 also shows that the biofilm thickness is less than one micron. Therefore, diffusion in the biofilm can be assumed and this verifies the assumption made in model development. A fully-penetrated biofilm with the traditional Monod model for substrate utilization/biomass growth agrees with experimental results.
3.1.3.
Activity kinetics
The sand samples extruded as sections from the reactors were analyzed for microbial activity using the modified BDOC
Fig. 7 e Activity Kinetics for Reactor 3.
4426
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 1 9 e4 4 2 7
Fig. 8 e Activity Kinetics for Reactor 4.
kinetic test. Sand extruded from every 4th section of each of the reactors was tested to determine the removal of dextran over a period of 5 days. The activity test data for sand samples taken from reactors 1 through 4 is presented in Figs. 5e8, respectively. The kinetics were found to be zero-order with respect to substrate removal for samples from all the reactors. For each section of the reactors, the rate was independent of the distance from the influent to the effluent. Hence, the overall rate for a reactor could be calculated based on the results for the average of sections from a reactor. The individual rate for reactor 1 was found to be equal to 0.1734 mg-C/L-day, for reactor 2 it was 0.2849 mg-C/L-day, for reactor 3 it was 0.3989 mg-C/L-day and for reactor 4 it was 0.3257 mg-C/L-day. The overall rates for the column reactors were calculated using the ratio of total mass of sand in the columns divided by the mass used for batch testing. These values were found to be equal to 10.404 mg-C/L-day for reactor 1, 17.094 mg-C/L/day for reactor 2, 23.934 mg-C/L/day for reactor 3 and 19.542 mg-C/ L/day for reactor 4. Reactors 1 and 2 always had the same hydraulic retention time and Reactors 3 and 4 always had the same hydraulic retention time. We see that the reactors with smaller particle size had higher rates than that of reactors with larger particle diameter for the same flow rates (i.e. Reactor 2 Reactor 1 and Reactor 3 Reactor 4). The smaller difference between Reactors 3 and 4 could be from an accumulation of biomass during Phase 1 when the loading rate was 8 times the loading rate in Phase II. Similar data was observed when the BDOC reactors were used to evaluate kinetics using a single dextran feed concentration. This is interesting since the BDOC reactors were acclimated in a batch system while the sand from the reactors was acclimated in a column, yet the effects of surface area were similar.
4.
Summary
Dextran (average MW 10,000 Da) was observed to biodegrade with zero-order substrate utilization kinetics during batch kinetic experiments. This is consistent with the expected kinetics when hydrolysis is the rate limiting step. During continuous flow column experiments, the biomass distribution did not vary significantly based on activity measurements
and organic nitrogen analyses. Four column experiments were completed with two different particle sizes (hence surface area contact) and 4 different flow rates. The removal of substrate was independent of flow rate and particle size. However, the surface area had a positive relationship with biomass accumulation. The ratio of organic-N (hence biomass) was a factor of 3 higher for 0.353 mm particle size when compared with 0.6 mm particle size in one set of paired columns (Columns 1 and 2) and a factor of 2.4 higher for 0.353 mm particle size when compared with 0.6 mm particle size in the second set of paired columns (Columns 3 and 4). The surface area for the columns with a particle size of 0.353 mm was 3 times greater than the columns with 0.6 mm particle size. The experimental results demonstrated that biological removal of a biodegradable high molecular weight compound was robust during flow over a porous media. The Monod based substrate/growth kinetic model does predict the removal of a single type of biodegradable high molecular weight and the biofilm thickness was insufficient to cause diffusion limitations. The rate limiting step appears to be hydrolysis. Hydrolysis may be the rate limiting step for mixtures of high molecular weight compounds that are present in actual systems. The mixtures contain compounds with different biodegradabilities and their hydrolysis products might not be easily biodegradable resulting in higher order kinetics.
references
Amy, G., Wilson, L.G., Conroy, A., Chahbandour, J., Ahai, W., Siddiqui, M., 1993. Fate of chlorination byproducts and nitrogen species during effluent recharge and soil aquifer treatment. Water and Environmental Research 65, 726e734. Amy, G.L., Drewes, J., Westerhoff, P., 2006. Organic matter in soilaquifer treatment systems. Journal of Environmental Engineering 132 (11), 1447e1458. Andrew, D.E., Rice, E.W., Baird, R.B., 2007. In: Standard Methods for the Examination of Water and Wastewater, first ed. American Public Health Association. Bouwer, H., 1985. Renovation of Wastewater with Rapidinfiltration Land Treatment Systems. In: Asano, T. (Ed.), Artificial Recharge of Groundwater. Butterworth, Boston, pp. 249e282. Bouwer, H., Rice, R.C., 1984. Renovation of wastewater at the 23rd avenue rapid infiltration project. Journal of Water Pollution Control Federation 36, 76e83. Cadoret, Aurore, Conrad, Arnaud, Block, Jean-Claude, 2002. Availability of low and high molecular weight substrates to extracellular enzymes in whole and dispersed activated sludges. Enzyme and Microbial Technology 31, 179e186. Cha, W., Choi, H.C., Fox, P., 2004. Abiotic and biotic removal mechanisms for organic carbon during soil aquifer treatment. Water Environment Research 76, 756e804. Chrost, R.J., 1991. Environmental control of the synthesis and activity of aquatic microbial ecto-enzymes. In: Chrost, R.J. (Ed.), Microbial Enzymes in Aquatic Environments. Springer, New York, pp. 29e59. Drewes, J.E., Jekel, M., 1998. Behavior of DOC and AOX using advanced treated wastewater for groundwater recharge. Water Research 32, 3125e3133. Eliosov, B., Argaman, Y., 1995. Hydrolysis of particulate organics in activated sludge systems. Water Research 29, 155e163.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 1 9 e4 4 2 7
Fox, P., Narayanaswamy, K., Genz, A., Drewes, J.E., 2001. Water quality transformations during soil aquifer treatment at the nesa Northwest water reclamation plant USA. Water Science and Technology 43 (10), 343e350. Guellil, A., Boualam, M., Quiquampoix, H., Ginsestet, P., Audic, J.M., Block, J.C., 2001. Hydrolysis of wastewater colloidal organic matter by extracellular enzymes extracted from activated sludge flocs. Water Science and Technology 43 (6), 33e40. Herbert, D., Phipps, P.J., Strange, R.E., 1971. Carbohydrate analysis. vol. 5B. In: Norris, J.R., Ribbons, D.W. (Eds.), Methods in Microbiology. Academic Press, New York, pp. 265e301. Levine, A.D., Tchobanoglous, G., Asano, T., 1985. Characterization of the size distribution of contaminants in wastewater: treatment and reuse implications. Journal of Water Pollution Control Federation 57, 805e816. Nalinakumari, B., Cha, W., Fox, P., 2010. Effects of primary substrate concentration on N-nitrosodimethylamine (NDMA) during simulated aquifer recharge. ASCE Journal of Environmental Engineering 136 (4), 373e380.
4427
Raunkjaer K, Hvitved-Jacobsen T., Nielsen P.H. 1994. Measurement of Pools of Protein, Carbohydrate and Lipid in Domestic Wastewater. 28(2), 251e262. Vetter, Y.A., Deming, J.W., 1999. Growth rates of marine bacterial isolates on particulate organic substrates solubilized by freely released extracellular enzymes. Microbial Ecology 37, 86e94. Wetzel, R.G., 1991. Extracellular enzymatic interactions: storage, redistribution and inter specific communication. In: Chrost, R.J. (Ed.), Microbial Enzymes in Aquatic Environments. Springer, New York, pp. 6e28. Wild, D., Reinhard, M., 1999. Biodegradation residual of 4-Octylphenoxyacetic acid in laboratory columns under groundwater recharge conditions. Environ. Sci. Technol. 33 (24), 4422e4426. Wilson, L.G., Amy, G.L., Gerba, C.P., Gordon, H., Johnson, B., Miller, J., 1995. Water quality changes during soil aquifer treatment of tertiary effluent. Water and Environmental Research 67, 371e376.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 2 8 e4 4 3 6
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Accuracy and precision of Legionella isolation by US laboratories in the ELITE program pilot study Claressa E. Lucas*, Thomas H. Taylor Jr., Barry S. Fields Division of Bacterial Diseases, Centers for Disease Control and Prevention, 1600 Clifton Rd NE MS G03, Atlanta, GA 30333, USA
article info
abstract
Article history:
A pilot study for the Environmental Legionella Isolation Techniques Evaluation (ELITE)
Received 9 December 2010
Program, a proficiency testing scheme for US laboratories that culture Legionella from
Received in revised form
environmental samples, was conducted September 1, 2008 through March 31, 2009.
28 March 2011
Participants (n ¼ 20) processed panels consisting of six sample types: pure and mixed
Accepted 26 May 2011
positive, pure and mixed negative, pure and mixed variable. The majority (93%) of all
Available online 7 June 2011
samples (n ¼ 286) were correctly characterized, with 88.5% of samples positive for Legionella and 100% of negative samples identified correctly. Variable samples were incorrectly
Keywords:
identified as negative in 36.9% of reports. For all samples reported positive (n ¼ 128),
Proficiency testing
participants underestimated the cfu/ml by a mean of 1.25 logs with standard deviation of
Environmental sampling
0.78 logs, standard error of 0.07 logs, and a range of 3.57 logs compared to the CDC re-test
Bacteria enumeration
value. Centering results around the interlaboratory mean yielded a standard deviation of
Legionella monitoring
0.65 logs, standard error of 0.06 logs, and a range of 3.22 logs. Sampling protocol, treatment regimen, culture procedure, and laboratory experience did not significantly affect the accuracy or precision of reported concentrations. Qualitative and quantitative results from the ELITE pilot study were similar to reports from a corresponding proficiency testing scheme available in the European Union, indicating these results are probably valid for most environmental laboratories worldwide. The large enumeration error observed suggests that the need for remediation of a water system should not be determined solely by the concentration of Legionella observed in a sample since that value is likely to underestimate the true level of contamination. Published by Elsevier Ltd.
1.
Introduction
Legionellaceae are ubiquitous in moist environments and a frequent contaminant of building water systems (Fields et al., 2002). Inhalation of aerosolized water contaminated with legionellae by susceptible individuals results in legionellosis that may present as the acute pneumonia, Legionnaires’ disease, or the less severe Pontiac Fever. Legionella pneumophila is the most common etiological agent, however all species of legionellae are presumed to be capable of causing disease (Alli et al., 2003; Palusinska-Szysz and Cendrowska-Pinkosz, 2009). * Corresponding author. Tel.: þ1 404 639 3564; fax: þ1 866 638 0199. E-mail address:
[email protected] (C.E. Lucas). 0043-1354/$ e see front matter Published by Elsevier Ltd. doi:10.1016/j.watres.2011.05.030
Legionellosis cannot be spread person-to-person and is only acquired from environmental sources. The widespread presence of legionellae precludes their removal from the environment so that disease prevention methods instead focus on reducing transmission of bacteria to susceptible hosts (Fields et al., 2002; Sehulster and Chinn, 2003; Freije, 2004; Tablan et al., 2004; Fields and Moore, 2006). Monitoring levels of legionellae in building water systems by routine environmental sampling has been employed by some as a means of controlling transmission, though the relationship between the presence of legionellae and
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 2 8 e4 4 3 6
incidence of disease remains unclear (O’Neill and Humphreys, 2005; Den Boer et al., 2007). Environmental sampling for Legionella spp. as an approach for primary control can provide useful information to institutions that house persons at high risk for disease, such as chronic care and transplant facilities (Anonymous, 1997; Butler et al., 1997; Yu, 1997; Fiore et al., 1999; Fields and Moore, 2006). However, routine culture in the absence of documented cases of legionellosis is an area of considerable controversy, mostly because there is no generally accepted protocol for the choice of sampling sites, frequency of sampling events, or endpoints for remediation (Den Boer et al., 2007; Stout et al., 2007; Ditommaso et al., 2010). Recommendations that suggest the use of routine sampling for primary control acknowledge that the acceptable limits of contamination, either measured in cfu/ml or as a percentage of positive sites sampled, are based on limited data (Force, 1997; Sehulster and Chinn, 2003; Stout et al., 2007). Another confounding factor that is not frequently mentioned in the discussion on routine sampling is the considerable variability in recovery of legionellae from repeated sampling of sites or even seeded tap water suggesting that interlaboratory enumeration error may be high (Boulanger and Edelstein, 1995; Bentham, 2000; Napoli et al., 2009). The Centers for Disease Control and Prevention (CDC) guidelines for the reduction of legionellosis advise that there is no acceptable level of legionellae contamination and that if Legionella spp. are detected a plan to prevent transmission to susceptible individuals should be employed (1997). Domestic and international organizations have published procedures for the recovery of Legionella bacteria from environmental samples but the application of these techniques can still require considerable expertise from the laboratorian (Anonymous, 1998, 2000, 2004, 2005a,b, 2008). Numerous variables contribute to effective recovery including properly taking and transporting samples, the application of sample pre-treatments intended to increase the representation of viable legionellae, and the use of suitable selective media (Reeves et al., 1981; Tesh and Miller, 1981; Shahamat et al., 1991; Lee et al., 1993; Ta et al., 1995; Leoni and Legnani, 2001; Wiedenmann et al., 2001; Bartie et al., 2003; Luck et al., 2004). Characterization of isolates entails the ability to distinguish Legionella colony morphology from autochthonous microbiota and serological methods for confirmation of identity, both of which determinations are subjective and sometimes difficult to interpret without considerable prior experience by the laboratorian (Benson and Fields, 1998; Helbig et al., 2007; Wagner et al., 2007). PCR-based methods of isolate identification offer greater sensitivity and specificity but are often cost prohibitive for laboratories with a low volume of tests for legionellae and do not yet provide enough discrimination between strains for epidemiological surveillance (Thurmer et al., 2009; Tronel and Hartemann, 2009). In the European Union (EU) laboratories that culture environmental samples for Legionella spp. are required to participate in a proficiency testing (PT) scheme to ensure baseline quality standards throughout the industry (2005). The EU scheme, in operation since June 2004, requires participants to process PT products as would be performed for potable water and report results to the program administrators for scoring
4429
according to ISO standards (1997; 1997). The EU scheme has been successful in assessing the status of the industry and publishes quarterly reports of their trends and performance distributions over time. A new PT scheme, the Environmental Legionella Isolation Techniques Evaluation (ELITE) program was created by CDC to capture similar information for US laboratories. A Pilot Program was performed September 1, 2008eMarch 31, 2009 to determine industry limits of detection and ascertain the accuracy of legionellae enumeration by participating laboratories. The results indicate that US laboratories are generally capable of a qualitative assessment of environmental samples for the presence of legionellae but that quantitation displays significant inter- and intralaboratory variability.
2.
Methods and materials
2.1. Sample creation, panel composition, quality control, and re-tests Legionellae and heterotrophic bacteria were grown from freezer stocks of less than three passages on BCYE media (BBL agar base plus 10 g/L L-cysteine) for 3e5 days at 35 C with 2.5% (v/v) CO2. Samples distributed for PT were removed from plates by suspension in sterile de-ionized water. Bacterial suspensions were diluted in 10% (v/v) AYE broth prior to lyophilization. The formula for full strength AYE broth is as follows: 5 g bovine serum albumin fraction V, 10 g ACES, 10 g yeast extract, 0.4 g L-cysteine, 0.25 g iron pyrophosphate. Bring volume to 1 L with sterile de-ionized water, adjust pH to 6.9 with 1 N potassium hydroxide, and filter to sterilize. The AYE broth helped to stabilize lyophilized cells during storage and formed a consistent pellet regardless of sample composition. Samples were made in batches monthly with 20e25 aliquots per lot. A representative sample from each lot was examined for quality control (QC) one week after lyophilization by plating serial dilutions from the vial reconstituted in 1 ml sterile de-ionized water on BCYE in triplicate at each serial dilution. Plates were incubated 5 days at 35 C with 2.5% (v/v) CO2 prior to being read. The total bacterial count and Legionella-specific count per plate were determined. At least one representative Legionella-like colony per plate was confirmed as a cysteine auxotroph. PT panels consisted of 6 samples, one of each sample type: pure positive, mixed positive, pure negative, mixed negative, pure variable, and mixed variable (see Table 1). Positive samples were either Legionella in pure culture or mixed with a low ratio of heterotrophs. Negative samples contained no viable legionellae, though they might contain heat killed legionellae. Variable samples consisted of either low levels of Legionella in pure culture or a mixture of organisms with a high ratio of heterotrophs to legionellae. Laboratories were scored on a pass/fail basis, required to correctly identify both positive and both negative samples for a passing score. Variable samples did not count toward the score but were included in each panel to assess lower limits of detection. Panels identical to those sent to participants were shipped back to the program administrators for re-tests. Re-test samples were first reconstituted in the vial with 1 ml of sterile,
Sample ID no
A21-08052203 A48-08071517 A65-08091607 A67-08091619 A62-08091601 A20-08052201 A28-08052204 A6508091615 A48-08071518 A6708091617 A62-08091609 A50-08081904 A71-08111906 A38-08061703 A25-08052218 A84-08120201 A88-08120215 A50-08081908 A28-08052207 A89-08120213 A14-08041609 A39-08061716 A86-08120217 A31-08061717 A89-08120211 A62-08091607 A32-08061706
Negative e media Variable e mixed Variable e low Positive e pure Negative e mixed Positive e mixed Negative e media Negative e mixed Variable e mixed Positive e mixed Positive e pure Variable e pure Negative e pure nonLegionella Negative e mixed Variable e mixed Variable e pure Positive e mixed Positive e pure Negative e media Negative e mixed Variable e pure Variable e mixed Positive e mixed Positive e pure Negative e media Negative e mixed Variable e pure Positive e mixed Positive e pure Variable e mixed Negative e media Negative e mixed Variable e pure Positive e mixed Variable e mixed Positive e pure Negative e mixed Variable e mixed Variable e pure Positive e mixed
Panel
Sample composition
QC total cfu/ml SD
QC Legionella cfu/ml SD
Re-testa total cfu/ml SD
Re-test Legionella cfu/ml
Re-test protocol
I I I I I, A I, A II II II II II II, A III
Blank media L. pneumophila Sg1 þ HTs L. pneumophila Sg3 L. pneumophila Sg8 HTs L. bozemani þ HTs Blank media HTs L. pneumophila Sg1 þ HTs L. pneumophila Sg3 þ HTs L. pneumophila Sg8 L. cherrii HTs
0 5 0.3 Pure Pure 1370 374 4400 712 0 703 9 41 583 26 Pure Pure 3330 471
0 0.6 0.4 53 13 211 14 0 170 11 0 0 0.7 4 48 4 254 17 74 0
0 7 0.3 Pure Pure 861 5 3000 0 300 41 300 Pure Pure ND
0 1 52 139 0 170 0 0 1 35 344 2 0
D, BCYE D, GPCV D, BCYE SD, BCYE SD, BCYE SD, PCV D, BCYE SD, BCYE D, GPCV SD, A, PCV SD, BCYE D, BCYE SD, BCYE
III III III III III, A IV IV IV IV IV IV V V V V V V, A VI VI VI VI VI VI VII VII VII VII
HTs L. pneumophila Sg1 þ HTs L. pneumophila Sg1 L. pneumophila Sg8 þ HTs L. pneumophila Blank media HTs L. pneumophila Sg8 L. pneumophila Sg1 þ HTs L. pneumophila Sg1 þ HTs L. pneumophila Sg8 Blank e broth only HTs L. rubrilucens L. dumoffii HTs L. feelei L. pneumophila Sg1 þ HTs Blank e broth only HTs L. pneumophila Sg10 L. rubrilucens þ HTs L. rubrilucens þ HTs L. dumoffii HTs L. rubrilucens þ HTs L. pneumophila Sg1 L. pneumophila Sg3 þ HTs
83,000 3740 7870 309 Pure 4730 624 Pure 0 15, 200 2380 Pure 883 58 18, 7000 1250 Pure 0 15,600 1130 Pure 39,000 5720 Pure 600 36 0 13,600 8740 Pure 16,200 9420 70,700 7040 Pure 7870 141 205 23 Pure 11,700 5350
0 150 39 593 170 2000 630 7600 1280 0 0 61 7 257 38 4000 1410 3300 804 0 0 10 0.4 1670 125 27,300 5910 33 5 0 0 183 24 1290 173 2400 57 74,200 17,000 0 35 4 59 17 1480 398
ND ND Pure ND Pure 0 ND Pure ND ND Pure ND ND Pure ND Pure ND 0 ND Pure ND ND Pure ND ND Pure ND
0 13 120 311 3000 0 0 16 193 412 3000 0 0 3 1000 16500 12 0 0 102 540 2000 34300 0 17 60 200
SD, BCYE A, SD, PCV SD, BCYE A, SD, GPCV SD, BCYE D, BCYE SD, BCYE D, BCYE A, SD, BCYE A, SD, PCV SD, BCYE D, BCYE SD, BCYE D, BCYE A, SD, GPCV SD, BCYE A, SD, BCYE D, BCYE SD, BCYE SD, BCYE A, SD, PCV A, SD, BCYE SD, BCYE SD, BCYE SD, BCYE D, BCYE A, SD, GPCV
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 2 8 e4 4 3 6
A50-08081901 A49-08071519 A32-08061701 A12-08041602 A31-08061718 A25-08052219 A50-08081905 A30-08061719 A49-08071520 A32-08061707 A12-08041604 A32-08061702 A31-08061704
Status
4430
Table 1 e PT sample characteristics: All PT samples used during the Pilot Program (Panel column: Panels number IeVIII and Accelerated) are listed here with sample composition, QC, and re-test results. Samples were categorized into sample types according to QC results. Re-test treatments specific to each sample are abbreviated: D [ direct plating, SD [ serial dilutions, A [ 15 min acid.
4431
VIII VIII VIII VIII VIII Negative e mixed Variable e mixed Variable e pure Positive e mixed Positive e pure A71-08111907 A89-08120212 A65-08091608 A32-08061708 A62-08091604
VIII Negative e media A90-09010509
a Heterotrophic plate counts were enumerated by re-test only for the first round (Panels I and II). Heterotrophic cfu/ml were not determined (ND) by re-test for panels III through VIII.
SD, BCYE A, GPCV SD, BCYE A, SD, BCYE SD, BCYE 0 5 45 200 835 7670 858 1130 17 Pure 4430 377 Pure
0 23 12 300 183 37 1330 557
ND ND Pure ND Pure
D, BCYE 0 0 0
0
1410 0 VII VII, A A62-08091603 A90-09010505
Positive e pure Negative e media
L. pneumophila Sg1 L. pneumophila Sg1 [heat killed] L. pneumophila Sg10 [heat killed] HTs L. rubrilucens þ HTs L. pneumophila Sg1 L. pneumophila Sg3 þ HTs L. pneumophila Sg1
Pure 0
3100 804 0
Pure 0
SD, BCYE D, BCYE
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 2 8 e4 4 3 6
de-ionized water then diluted 1:1000 in a bottle of sterile, deionized water to yield a 1 L Test Sample. All Test Samples were filtered. Each Test Sample was then processed according to an individual protocol determined by sample composition, which was designed for maximum recovery and enumeration of legionellae (Table 1). Three replicates of each dilution were plated but only the highest count was used as the re-test value for data analysis. All re-test values were within one standard deviation of the mean of the re-test replicate plate counts (data not shown). Test Samples were subjected to combinations of pre-treatment with acid (18 parts 0.2 M KCl to 1 part 0.2 M HCl), serial dilutions prior to plating, and/or plating on selective media PCV (BCYE plus 13.22 mg/L polymixin B, 80 mg/L cyclohexamide, and 5 mg/L vancomycin) or GPCV (PCV plus 2 g/L glycine). QC results, re-test results, and re-test treatment protocols are listed for all samples in Table 1.
2.2.
Pilot participants
Because current industry capabilities were unknown at the inception of the ELITE Program, a pilot study was conducted September 1, 2008 through March 31, 2009 to generate baseline values. The Pilot Program was limited to ten participating laboratories. A total of 20 laboratories enrolled in the program prior to the August 31, 2008 deadline for inclusion in the Pilot Program. Laboratories that could not be accommodated in the Pilot Program due to space limitations were enrolled as Accelerated Members. Commercial laboratories comprised the majority of enrollees but the Pilot Program was designed to include at least one of each type of laboratory as listed in Table 2. CDC reference laboratory personnel separate from those administering the ELITE Program participated in the pilot. To avoid confusion in the text, CDC reference laboratory Pilot Program participants are designated the “federal” laboratory in the text and figures, while administrators’ results are labeled “CDC”. For data analysis purposes, State, County, and Hospital laboratories were grouped into the category “Local Public Health.” Participants were asked to process the panels according to their standard in-house protocol and return results as they would be given to clients. Participants were directed to rehydrate the lyophilized pellets in 1 ml sterile water then dilute the suspension 1:1000 in sterile water to yield a test sample of the volume dictated by the participant’s in-house protocol. Supplement 1 contains the complete Sample Handling Instructions available to all participants. Pilot Members processed four panels shipped September 3, 2008, November 3,
Table 2 e Pilot and accelerated program demographics: the number and type of each participant in the pilot or accelerated program.
Commercial Federal State County Hospital Total
Pilot
Accelerated
Total
6 1 1 1 1 10
6 0 3 0 1 10
12 1 4 1 2 20
4432
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 2 8 e4 4 3 6
2008, December 29, 2008, and March 2, 2009. Accelerated Members processed a single panel consisting of samples already tested by Pilot Members that was shipped March 20, 2009. All laboratories enrolled during this time were given two months to process a panel and report results.
2.3.
Data collection and analysis
Combined results from Pilot and Accelerated Members were captured from the CDC SQL 2005 database June 12, 2009. Reported laboratory concentrations were tabulated and summarized. Differences were computed and tabulated between the reported concentrations and (a) the CDCvalidated concentration (re-test values) and (b) the intralaboratory mean among all laboratories reporting a value. In addition, tabulations of un-quantified identifications were made. Quantitative results were analyzed directly and through log transformations. Reporting laboratories were categorized, respectively, by: type of laboratory (state, local, federal, hospital, commercial); annual volume of Legionella tests performed; how long (years) the laboratory has tested for Legionella; number of full-time and Legionella-dedicated employees; and by protocol-related factors such as incubation time and temperature, water volume, use of CO2, and standard protocol reference (ASM, ISO, or CDC). State, county, and hospital laboratories were grouped into the category “Local Public Health” for further analyses. For each categorization of laboratories or procedures, Chi-square tests were computed for dichotomous outcomes such as finding Legionella or not or being within a specified error, such as one log, of the respective (re-test or interlaboratory mean) reference concentration. In addition, for continuous measures, t-tests were performed comparing the errors of mutually-exclusive groups of laboratories, such laboratories which used less than 750 ml of water versus those who used 750 ml or more.
3.
Results
3.1.
PT sample qualitative identification
Concordance between expected positive or negative results, as determined by QC, and results returned by Pilot or Accelerated Members was high regardless of laboratory type (Table 3). The majority of positive samples (88.5%) were correctly identified. No false positives from a negative sample were reported by any member. In contrast, 36.9% of all variable samples tested (n ¼ 103) were incorrectly identified (Table 4). The federal laboratory was most often correct, followed by commercial laboratories, and then local public health laboratories. Correct identification of positive samples depended on the Legionella concentrations in the samples. Samples with less than 10 cfu/ml, as determined by re-test, were identified as negative in 93.1% of reports while samples with 10 cfu/ml or more were reported positive in 85.3% of reports. These results were independent of whether the sample was pure or mixed, positive or variable indicating that 10 cfu/ml is at or near the lower limit of detection. In addition to whether a sample was positive or negative for Legionella growth, participating
Table 3 e Concordance of positive and negative sample reports by laboratory type: displays the agreement between expected results and participants’ reported results for all positive or negative samples, both mixed and pure. Variable samples’ reported results are not included in this table. Definitions of column headings are: True Negative [ expected negative and reported negative, False Negative [ expected positive but reported negative, True Positive [ expected positive and reported positive.
Commercial Federal Local public health Total results (n)
True negative
False negative
True positive
Total results (n)
56 (100%) 8 (100%) 32 (100%)
3 (6.3%) 0 (0%) 7 (21.9%)
45 (93.8%) 7 (100%) 25 (78.1%)
104 15 64
96 (100%)
10 (11.5%)
77 (88.5%)
183
laboratories had the option to report species and serogroup of legionellae recovered. All laboratories that provided optional results correctly identified the species and serogroup of recovered Legionella.
3.2.
Accuracy and precision of Legionella quantitation
Pilot and Accelerated Members were encouraged to also provide results for their observed concentration of legionellae in samples (n ¼ 128). The expected concentration for each sample was determined by re-test (Table 1). The log error for each report is displayed graphically in Fig. 1. The majority (99.5%) of reported results underestimated the expected concentration by an average of 1.25 logs with a standard deviation of 0.78 logs, standard error of 0.07 logs, and range of 3.57 (3.30 to 0.27) logs. Centering the results on the interlaboratory mean forced an overall net difference of zero logs (Fig. 2). The standard deviation was 0.62 logs, standard error was 0.06 logs and the range was 3.22 (1.89 to 1.33) logs. The standard deviation from the interlaboratory mean of the results from laboratories that entered at least three concentrations for evaluation (n ¼ 11) ranged from 0.53 to 0.96 and standard error from 0.14 to 0.58. No sampling protocol, treatment regimen, incubation procedure, or organizational structure type analyzed produced significantly greater accuracy when compared to re-test values or precision when compared to the interlaboratory mean concentration (Table 5).
Table 4 e Accuracy of variable sample reports by laboratory type: Displays the reported results for all variable samples, both mixed and pure. No positive or negative sample results are included in this table.
Commercial Federal Local public health Total results (n)
Reported positive
Reported negative
Total results (n)
39 (62.9%) 7 (77.8%) 19 (59.4%)
23 (37.1%) 2 (22.2%) 13 (40.6%)
62 9 32
65 (63.1%)
38 (36.9%)
103
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 2 8 e4 4 3 6
4433
Fig. 1 e Accuracy of quantitative results relative to re-test values for all samples reported positive: Displays the log error of the difference between re-test values and reported concentrations for all samples reported positive. Each marker represents a sample reported positive by a participant (n [ 190). Markers are coded by laboratory type: commercial (green circle), federal (red square), local public health (blue triangle). Enumeration error [Error(Log(CFU))] was calculated by taking the difference between the log value of the expected concentration in cfu/ml and the log value of the concentration reported by a participant in cfu/ml. Participant responses are plotted as enumeration error ( y-axis) vs. expected concentration derived from re-test values (x-axis). Samples reported positive without a concentration value (n [ 62) are indicated by markers connected with the blue line ( y [ Lx) and are not included in calculations measuring accuracy. The green line depicts the mean enumeration error across the range of reported concentrations (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
4.
Discussion
The results from the ELITE Pilot Program indicate that most participating US laboratories are capable of qualitatively
identifying Legionella spp. from a water sample and that the lower limit of detection is approximately 10 cfu/ml. According to 2009 EU Health Protection Agency external quality assessment for Legionella isolation from water samples reports, the accuracy of the European PT participants (Shah et al., 2009a,
Fig. 2 e Precision of quantitative results relative to interlaboratory means for all samples reported positive: Displays the log error of the difference between interlaboratory mean values and reported concentrations for all samples reported positive. Each marker represents a sample reported positive by a participant (n [ 190). Markers are color coded by laboratory type: commercial (green circle), federal (red square), local public health (blue triangle). Enumeration error [Error(Log(CFU))] was calculated by taking the difference between the log value of the interlaboratory mean cfu/ml and the log value of the concentration reported by a participant in cfu/ml. Participant responses are plotted as enumeration error ( y-axis) vs. expected interlaboratory mean (x-axis). Samples reported positive without a concentration value (n [ 62) are indicated by markers connected with the blue line (y [ -x) and were not included in calculations to determine the interlaboratory mean (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
4434
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 2 8 e4 4 3 6
Table 5 e Statistical analysis of accuracy as a function of procedural variables: The enumeration errors associated with treatment, protocol, and experience categories were compared with t-tests. The p-values are given when the baseline is calculated from either re-test values or the interlaboratory mean. Measurement
Incubation temperature Incubation with CO2 Water volume Incubation time Sample composition Sampling protocol Years of testing Number of dedicated employees Number of samples processed annually
Categories
p Value compared to re-test
p Value compared to interlaboratory mean
35.0 C or >35.0 C 10 years 2 or >2 employees 200 or >200
0.297 0.537 0.454 0.318 0.065 0.221 0.672 0.993 0.919
0.311 0.332 0.476 0.562 >0.999 0.216 0.159 0.960 0.565
2009b, 2009c, 2009d) averaged 93% concordance (range 78e98%). US laboratories in this study averaged 86% concordance with positive samples, well within the range of EU performance. On the other hand, with regard to precision, US laboratories demonstrated more extreme variability in enumerating bacteria in the PT samples both within and between participants. US labs underestimated the concentration of Legionella by an average of 1.25 (0.78) logs when compared to re-test values. Further, US labs exhibited 0.62 logs variance around the interlaboratory mean. In contrast, EU laboratory results from 2009 clustered very close to the interlaboratory mean and FEPTU median (a reference standard comparable to re-test values), usually within a 0.5 log range of either score, with a mean standard deviation of 0.44 logs (Shah et al., 2009a, 2009b, 2009c, 2009d). Thus, EU and US laboratories demonstrate similar accuracy in identifying positive samples but US laboratories appear to be less precise than EU laboratories in enumerating viable bacteria. The apparent greater precision of enumeration by EU laboratories compared to US laboratories is partially due to the bacterial concentrations used in the two sources of PT samples. EU sample concentrations spanned 0.2e110 cfu/ml while US PT samples contained between 1 and 34,300 cfu/ml (as determined by re-test), allowing the possibility for a larger range of enumeration error within US samples. However, a greater contribution to the disparity in variance between EU and US reported results was the different methodologies used to generate reference standards for comparison. US re-test values were calculated from the highest count recovered from a representative sample after treatment with an individualized protocol that took sample concentration and composition into account. Plating serial dilutions from the PT samples in triplicate, a feature of the majority of individualized re-test protocols (Table 1) but no referenced standard protocol (Anonymous, 1998, 2004, 2005a, 2005b, 2008, 2010), resulted in plates that had reduced numbers of obscuring heterotrophs and increased physical distance between colonies, making Legionella cfu easier to determine. In contrast, EU FEPT medians were generated from the results of 10 samples treated according to a single standard protocol without regard to sample concentration or composition, making the results inherently less variable. Both reference standards are valuable
in assessing participant results but address different laboratory capabilities. Re-tests return the most accurate count of viable bacteria within a sample, independent of protocol bias while the EU FEPT median and ELITE interlaboratory mean values address the precision with which a sample can be enumerated using a (set of) standard protocol(s). Thus, comparison to the re-test value indicates how close a laboratory can come to a ‘true’ answer while FEPT medians and interlaboratory means illustrate how reproducible the results are between and within a laboratory. Laboratories that submitted more than three concentration reports were statistically indistinguishable from each other in enumeration error and also demonstrated large intralaboratory variance (0.53e0.96). Two PT samples were quantified with significantly different log error from re-test values compared to the rest. These observations can be explained by sample composition. A62-08091603 (0.71 vs. 1.27, p ¼ 0.002, n ¼ 5) was a pure sample with a concentration within the optimum range for allowing distinct colony growth on media from a single 10-fold dilution and so could be measured more precisely than other samples. In contrast, sample A3208061706 (2.19 vs. 1.21, p ¼ 0.006, n ¼ 5) was heavily mixed with heterotrophs that had a spreading colony morphology, obscuring legionellae growth, resulting in a sizeable difference between QC and re-test results (Table 1), and a substantial interlaboratory enumeration error. Enumeration of the other 46 samples was statistically indistinguishable. No sampling protocol, treatment, incubation, or experience level analyzed in this study affected the accuracy or precision of enumeration (Table 5) at the 0.05 significance level. Taken together, these results indicate that intralaboratory and interlaboratory variance in precision and accuracy were similar in degree and magnitude for all pilot study participants. Two obvious caveats to these results are the small number of participating laboratories and their method of selection. ELITE Program participation is voluntary rather than mandated by legal statutes, as in the EU, and there was little promotion for the creation and implementation of the program. Most Pilot and Accelerated Members discovered the program through scientific meetings and/or word of mouth. Pilot Members were also likely to have a previous relationship with CDC, multiple years experience with Legionella isolation,
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 2 8 e4 4 3 6
and/or a large number of well-trained personnel dedicated to Legionella testing. Thus, it is possible that Pilot Members do not accurately reflect the capabilities of all US laboratories that culture environmental samples but are a subset of those whose primary focus is Legionella.
5.
Conclusions
Overall, Legionella PT qualitative and quantitative results were similar between US and EU laboratories in 2009 despite differences in sample composition, delivery, and reference standard determination. The observed variability in enumeration by both US and EU laboratories is probably due to the inherent inconsistency in assessing a sample by culture techniques. Given these data no protocol can be recommended to yield more accurate or precise results than any other. Responses from ELITE Program participants will continue to be monitored and analyzed to determine if more data can illuminate practices that contribute to increased accuracy and precision. However, the current findings have several implications for the use of routine sampling as a primary method of legionellosis control. Agreement between EU and US PT schemes suggest these results are applicable worldwide to environmental sampling laboratories. Since a sample qualitatively identified as positive could represent a 3 log cfu/ml range of viable legionellae it would be in the best interests of public health to consider any detectable level a hazard. Therefore, primary legionellosis prevention should consider the risk posed by an individual water system, assessing the likelihood of transmission and population affected, to determine if remediation is required rather than relying on a contamination cutoff level to take action.
Acknowledgments The authors would like to thank the Safe Water Program of the Division of Environmental Hazards and Health Effects, NCEH for funding the ELITE Program implementation and website administration October 1, 2007 through September 30, 2009. The ELITE Program website URL is https://wwwn.cdc.gov/ elite/Public/EliteHome.aspx.
Appendix. Supplementary data Supplementary data associated with this article can be found in the online version, at doi:10.1016/j.watres.2011.05.030.
references
Alli, O.A., Zink, S., von Lackum, N.K., Abu-Kwaik, Y., 2003. Comparative assessment of virulence traits in Legionella spp. Microbiology 149, 631e641. Anonymous, 1997. Guidelines for prevention of nosocomial pneumonia. MMWR Recomm Rep 46, 1e79. Centers for Disease Control and Prevention.
4435
Anonymous, 1998. Water Quality e Detection and Enumeration of Legionella Edited by I.S. Organization. Rule no. 11731:1998 TC no. 147/SC4. Anonymous, 2000. Minimizing the Risk of Legionellosis Associated with Building Water Systems. American Society of Heating, Refrigerating, and Air-Conditioning Engineers, Inc, Atlanta, GA. Document No. 12-2000:1-6. Anonymous, 2004. In: I.S. Organization (Ed.), Water Quality e Detection and Enumeration of Legionella e Part 2: Direct Membrane Filtration Method for Waters with Low Bacterial Counts Rule no. ISO 11731-2:2004. Anonymous, 2005a. European Guidelines for Control and Prevention of Travel Associated Legionnaires’ Disease. European Parliament and the Council. January 2005. 2119/ 98/EC. Anonymous, 2005b. In: Fields, B.S. (Ed.), Procedures for the Recovery of Legionella from the Environment. US Department of Health and Human Services, Public Health Service, Centers for Disease Control and Prevention, Atlanta, GA, pp. 1e13. http://www.cdc.gov/legionella/files/LegionellaProcedures-508. pdf. Recommendation January 2005. Anonymous, 2008. Standard Guide for The Inspection of Water Systems for Legionella and The Investigation of Possible Outbreaks of Legionellosis (Legionnaires’ Disease or Pontiac Fever) ASTM Rule no. D5952-08, 2008. Anonymous, 2010. Legionella Isolation Scheme. European Union Health Protection Agency, London, UK. www.hpa.org.uk/web/ HPAwebFile/HPAweb_C/1254510409973. Bartie, C., Venter, S.N., Nel, L.H., 2003. Identification methods for Legionella from environmental samples. Water Research 37, 1362e1370. Benson, R.F., Fields, B.S., 1998. Classification of the genus Legionella. Seminars in Respiratory Infections 13, 90e99. Bentham, R.H., 2000. Routine sampling and the control of Legionella spp. in cooling tower water systems. Current Microbiology 41, 271e275. Boulanger, C., Edelstein, P., 1995. Precision and accuracy of recovery of Legionella pneumophila from seeded tap water by filtration and centrifugation. Applied and Environmental Microbiology 61, 1805e1809. Butler, J., Fields, B.S., Breiman, R.F., 1997. Prevention and Control of Legionellosis. Infectious Disease and Clinical Practice 6, 458e464. Den Boer, J.W., Verhoef, L., Bencini, M.A., Bruin, J.P., Jansen, R., Yzerman, E.P., 2007. Outbreak detection and secondary prevention of Legionnaires’ disease: a national approach. International Journal of Hygiene and Environmental Health 210, 1e7. Ditommaso, S., Giacomuzzi, M., Gentile, M., Moiraghi, A.R., Zotti, C.M., 2010. Effective environmental sampling strategies for monitoring Legionella spp contamination in hot water systems. American Journal of Infection Control 38 (5), 341e343. Fields, B.S., Benson, R.F., Besser, R.E., 2002. Legionella and Legionnaires’ disease: 25 years of investigation. Clinical Microbiology Reviews 15, 506e526. Fields, B.S., Moore, M.R., 2006. Control of legionellae in the environment: a guide to the US guidelines. ASHRAE Transactions 112, 691e699. Fiore, A.E., Butler, J.C., Emori, T.G., Gaynes, R.P., 1999. A survey of methods used to detect nosocomial legionellosis among participants in the National Nosocomial Infections Surveillance System. Infection Control and Hospital Epidemiology 20, 412e416. Force, A.C.H.D.L.T., 1997. Approaches to Prevention and Control of Legionella Infection in Allegheny County Health Care Facilities. Allegheny County Health Department, Pittsburgh, PA, pp. 1e15.
4436
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 2 8 e4 4 3 6
Freije, M.R., 2004. The word on water: new CDC guidelines recommend a proactive approach to Legionella. Health Facilities Management 17, 33e34. 36. Helbig, J.H., Benson, R.F., Pelaz, C., Jacobs, E., Luck, P.C., 2007. Identification and serotyping of atypical Legionella pneumophila strains isolated from human and environmental sources. Journal of Applied Microbiology 102, 100e105. Lee, T.C., Vickers, R.M., Yu, V.L., Wagener, M.M., 1993. Growth of 28 Legionella species on selective culture media: a comparative study. Journal of Clinical Microbiology 31, 2764e2768. Leoni, E., Legnani, P.P., 2001. Comparison of selective procedures for isolation and enumeration of Legionella species from hot water systems. Journal of Applied Microbiology 90, 27e33. Luck, P.C., Igel, L., Helbig, J.H., Kuhlisch, E., Jatzwauk, L., 2004. Comparison of commercially available media for the recovery of Legionella species. International Journal of Hygiene and Environmental Health 207, 589e593. Napoli, C., Iatta, R., Fasano, F., Marsico, T., Montagna, M.T., 2009. Variable bacterial load of Legionella spp. in a hospital water system. The Science of The Total Environment 408, 242e244. O’Neill, E., Humphreys, H., 2005. Surveillance of hospital water and primary prevention of nosocomial legionellosis: what is the evidence? The Journal of Hospital Infection 59, 273e279. Palusinska-Szysz, M., Cendrowska-Pinkosz, M., 2009. Pathogenicity of the family Legionellaceae. Archivum Immunologiae et Therapiae Experimentalis 57, 279e290. Reeves, M.W., Pine, L., Hutner, S.H., George, J.R., Harrell, W.K., 1981. Metal requirements of Legionella pneumophila. Journal of Clinical Microbiology 13, 688e695. Sehulster, L., Chinn, R.Y., 2003. Guidelines for environmental infection control in health-care facilities. Recommendations of CDC and the Healthcare Infection Control Practices Advisory Committee (HICPAC). MMWR Recommendations and Reports 52, 1e42. Shah, H., Lang, N.L., Russell, J.E., 2009a. External quality assessment for: Legionella isolation from water samples. In: Russell, J.E. (Ed.), Summary of Results, G64. Center for Infections/Food and Environmental Proficiency Testing Unit, London, UK, pp. 1e15. http://www.hpa.org.uk/web/ HPAwebFile/HPAweb_C/1194947324524. Shah, H., Lang, N.L., Russell, J.E., 2009b. External quality assessment for: Legionella isolation from water samples. In: Russell, J.E. (Ed.), Summary of Results, G65. Center for Infections/Food and Environmental Proficiency Testing Unit, London, UK, pp. 1e15. http://www.hpa.org.uk/web/ HPAwebFile/HPAweb_C/1245581545311. Shah, H., Lang, N.L., Russell, J.E., 2009c. External quality assessment for: Legionella isolation from water samples. In: Russell, J.E. (Ed.), Summary of Results, G66. Center for Infections/Food and Environmental Proficiency Testing Unit,
London, UK, pp. 1e15. http://www.hpa.org.uk/web/ HPAwebFile/HPAweb_C/1249454476861. Shah, H., Lang, N.L., Russell, J.E., 2009d. External quality assessment for: Legionella isolation from water samples. In: Russell, J.E. (Ed.), Summary of Results, G67. Center for Infections/Food and Envinronmental Proficiency Testing Unit, London, UK, pp. 1e15. http://www.hpa.org.uk/web/ HPAwebFile/HPAweb_C/1259151931733. Shahamat, M., Paszko-Kolva, C., Keiser, J., Colwell, R.R., 1991. Sequential culturing method improves recovery of Legionella spp. from contaminated environmental samples. Zentralbl Bakteriol 275, 312e319. Stout, J.E., Muder, R.R., Mietzner, S., 2007. & other authors Role of environmental surveillance in determining the risk of hospital-acquired legionellosis: a national surveillance study with clinical correlations. Infection Control and Hospital Epidemiology 28, 818e824. Ta, A.C., Stout, J.E., Yu, V.L., Wagener, M.M., 1995. Comparison of culture methods for monitoring Legionella species in hospital potable water systems and recommendations for standardization of such methods. Journal of Clinical Microbiology 33, 2118e2123. Tablan, O.C., Anderson, L.J., Besser, R., Bridges, C., Hajjeh, R., 2004. Guidelines for preventing health-careeassociated pneumonia, 2003: recommendations of CDC and the Healthcare Infection Control Practices Advisory Committee. MMWR Recommendations and Reports 53, 1e36. Tesh, M.J., Miller, R.D., 1981. Amino acid requirements for Legionella pneumophila growth. Journal of Clinical Microbiology 13, 865e869. Thurmer, A., Helbig, J.H., Jacobs, E., Luck, P.C., 2009. PCR-based ’serotyping’ of Legionella pneumophila. Journal of Medical Microbiology 58, 588e595. Tronel, H., Hartemann, P., 2009. Overview of diagnostic and detection methods for legionellosis and Legionella spp. Letters in Applied Microbiology 48, 653e656. Wagner, C., Kronert, C., Luck, P.C., Jacobs, E., Cianciotto, N.P., Helbig, J.H., 2007. Random mutagenesis of Legionella pneumophila reveals genes associated with lipopolysaccharide synthesis and recognition by typing monoclonal antibodies. Journal of Applied Microbiology 103, 1975e1982. Wiedenmann, A., Langhammer, W., Botzenhart, K., 2001. A case report of false negative Legionella test results in a chlorinated public hot water distribution system due to the lack of sodium thiosulfate in sampling bottles. International Journal of Hygiene and Environmental Health 204, 245e249. Yu, V.L., 1997. Prevention and control of Legionella: an idea whose time has come. Infectious Disease and Clinical Practice 6, 420e421.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 3 7 e4 4 4 8
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Refining the estimation of illicit drug consumptions from wastewater analysis: Co-analysis of prescription pharmaceuticals and uncertainty assessment Foon Yin Lai a, Christoph Ort b,c,*, Coral Gartner d, Steve Carter e, Jeremy Prichard f, Paul Kirkbride g, Raimondo Bruno h, Wayne Hall d, Geoff Eaglesham a,e, Jochen F. Mueller a a
The University of Queensland, The National Research Centre for Environmental Toxicology (Entox), 39 Kessels Road, Coopers Plains, QLD 4108, Australia b The University of Queensland, Advanced Water Management Centre (AWMC), St. Lucia, QLD 4072, Australia c Eawag, Swiss Federal Institute of Aquatic Science and Technology, CH 8600 Du¨bendorf, Switzerland d The University of Queensland, UQ Centre for Clinical Research, Royal Brisbane and Women’s Hospital, Herston, QLD 4029, Australia e Queensland Health Forensic Scientific Services (QHFSS), Queensland Government, 39 Kessels Road, Coopers Plains, QLD 4108, Australia f Law Faculty, University of Tasmania, Private Bag 89, Hobart TAS 7001, Australia g Australian Federal Police, Forensic and Data Centers, GPO Box 401, Canberra, ACT 2601, Australia h School of Psychology, University of Tasmania, Private Bag 30, Hobart TAS 7001, Australia
article info
abstract
Article history:
Wastewater analysis is a promising monitoring tool to estimate illicit drug consumption at
Received 21 February 2011
the community level. The advantage of this technique over traditional surveys and other
Received in revised form
surveillance methods has been emphasized in recent studies. However, there are meth-
21 May 2011
odological challenges that can affect reliability. The objectives of this study were to
Accepted 28 May 2011
systematically reduce and assess uncertainties associated with sampling (through a strin-
Available online 14 June 2011
gent optimization of the sampling method) and the back calculation of per capita drug consumption (through a refined estimation of the number of people actively contributing
Keywords:
to the wastewater in a given period). We applied continuous flow-proportional sampling to
Normalization
ensure the collection of representative raw wastewater samples. Residues of illicit drugs,
Estimated population
opioids, prescription pharmaceuticals and one artificial sweetener were analyzed by liquid
Error propagation
chromatography coupled with tandem mass spectrometry. A parameter estimating the
LC-MS/MS
number of people actively contributing to wastewater over a given period was calculated
Australia
from the measured loads of prescription pharmaceuticals, their annual consumption and relative excretion data. For the calculation of substance loads in sewage, uncertainties were propagated considering five individual components: sampling, chemical analysis, flow measurements, excretion rates and the number of people contributing to the wastewater. The daily consumption per 1000 inhabitants was estimated to be almost 1000 mg for cannabis and several hundred mg for cocaine, methamphetamine and ecstasy. With the best sampling practice and current chemical analysis, we calculated the remaining uncertainty to be in the range of 20e30% (relative standard deviation, RSD) for the estimation of consumed drug masses in the catchment; RSDs for the per capita consumption were lower (14e24%), as one of the biggest uncertainty components (i.e. error in flow measurements) cancels out in the proposed method for the estimation of the number of
¨ berlandstrasse 133, P.O. Box 611, CH 8600 Du¨bendorf, Switzerland. Tel.: þ41 (0) 58 765 * Corresponding author. Present address: Eawag, U 52 77; fax: þ41 (0) 58 765 53 89. E-mail address:
[email protected] (C. Ort). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.05.042
4438
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 3 7 e4 4 4 8
people contributing to the daily wastewater volume. In this study, we provide methodological improvements that substantially enhance the reliability of the estimation method e a prerequisite for the application of this technique to meaningfully assess changes in drug consumption and the success of drug intervention strategies in future studies. ª 2011 Elsevier Ltd. All rights reserved.
List of abbreviations A
annual consumption of prescription pharmaceutical C concentration of chemical in raw wastewater E average excretion fraction of a given drug residue ENCP estimated number of contributing people (i.e. the number of people that actively contributed to the daily wastewater volume which was sampled) EDDP 2-ethylidene-1,5-dimethyl-3,3diphenylpyrrolidine F total flow in a day load of chemical in sewer Lsewer per capita consumption of a given illicit drug in MENCP the catchment normalized with ENCP Mcatchment consumed mass of chemical in the catchment MDA 3,4-methylenedioxyamphetamine MDEA 3,4-methylenedioxyethamphetamine MDMA 3,4-methylenedioxymethamphetamine P population
1.
Introduction
The use of illicit drugs in Australia and many other developed countries adversely affects population health, social order and the economy (Collins and Lapsley, 2008; Degenhardt et al., 2004). It is difficult to estimate the prevalence of an illegal, highly stigmatized and clandestine activity. To date, the extent and trends of illicit drug use at the national level have been monitored through indirect methods, such as socioepidemiological studies, mortality data from drug-related causes (e.g. opioid overdoses) and monitoring of border seizures and seizures from domestic traffickers and users (Shand et al., 2003; UNODC, 2010). In addition to these methods, Daughton (2001) proposed estimating illicit drug consumption through the analysis of drug residues in raw wastewater. Zuccato et al. (2005) applied this method by quantifying excreted parent drugs and their key metabolite(s) in wastewater samples. Their data provided the first direct estimates for illicit drug use (Zuccato et al., 2005, 2008). They also reported predictable temporal variations, with the highest loads of certain drugs occurring on weekends. A number of research teams have since applied this approach in cities in Europe, North America and Australia (Boleda et al., 2009; Bones et al., 2007; Chiaia et al., 2008; Huerta-Fontela et al., 2007, 2008; Irvine et al., in press; Karolak et al., 2010; Metcalfe et al., 2010; Postigo et al., 2010; Terzic et al., 2010; van Nuijs et al., 2009a,b,c; Zuccato et al., 2008). A potential key application is to estimate and assess changes in illicit drug
R ratio of molar mass of parent drug to its metabolite RSD relative standard deviation STP sewage treatment plant THC Δ9-tetrahydrocannabinol THCeCOOH 11-nor-9-carboxy-D9-tetrahydrocannabinol Total population in Australia in 2009 (i.e. 22 TPAUS million) uncertainty due to chemical analysis UC uncertainty of the estimated number of UENCP contributing people uncertainty of the fraction of metabolite to parent UE compound uncertainty of flow meter UF uncertainty of chemical loads in sewer ULsewer UMcatchment uncertainty of illicit drug loads consumed in the catchment uncertainty of illicit drug loads normalized to 1000 UMENCP people (ENCP) sampling uncertainty US
consumption in a given population. This approach provides valuable data for health and law enforcement agencies regarding, for example, the impact of strategies that attempt to reduce the supply of, or the demand for, illicit drugs. In these contexts, it is essential to recognize and minimize uncertainties related to the parameters used for the calculation of illicit drug consumption based on wastewater analysis. Many of the previous wastewater studies provided sophisticated analytical methods, explained the back calculation (i.e. estimation of drug consumption per capita) and demonstrated the applicability and potential of these methods to estimate illicit drug consumption in a given population. However, there are numerous uncertainties affecting these estimates. To date, uncertainties related to sample collection and the number of people contributing to the wastewater have been insufficiently reported or not reported at all. Recently, Ort et al. (2010c) noted that these studies typically have not presented evidence that the analyzed samples were collected in an appropriate manner; proper sampling is fundamental for the accurate calculation of chemical loads (Ort et al., 2010b). Depending on the characteristics of the sewer system and the exact sampling location, a flow-proportional, high-frequency sampling method is required, since common sampling modes and frequencies can result in systematic and random artifacts ranging from “insignificant” to “100% or more” (Ort et al., 2010c). This becomes even more important if samples are collected farther upstream in small sub-catchments or in the effluent of
4439
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 3 7 e4 4 4 8
individual institutions to evaluate intervention strategies (Prichard et al., 2010). Furthermore, the back calculation requires reliable information on the size of the population that has contributed to the sampled wastewater. Census data or design capacities of sewage treatment plants (STPs) have typically been used for these calculations. However, in reality this may not reflect the actual number of people in the contributing population, and may be variable due to daily commuting patterns, seasonal variability and/or special events. Zuccato et al. (2008) recognized the limitation of assuming one fixed population size and suggested that the accuracy of the back calculation could be improved by measuring human biomarkers in wastewater to estimate the number of people in a catchment. One candidate marker is creatinine (Chiaia et al., 2008). Recently van Nuijs et al. (2011) proposed using the average of loads of nitrogen, phosphorus and chemical and biological oxygen demand to estimate a day-specific number of inhabitants in the catchment. However, the loads of these conventional water quality parameters may be significantly and systematically influenced by other sources such as industrial discharges. To our knowledge, no practical correction methods using indicators that are specific to humans have been used to date to estimate the number of people who have contributed to sampled wastewater. In this study, we aimed to enhance the reliability of the back calculation method for illicit drug consumption through: (a) the collection of representative wastewater samples using a continuous flow-proportional sampling technique at the influent of a municipal sewage treatment plant; (b) the evaluation of daily variations in the number of people that actively contributed to a given sample by conducting analysis of prescription pharmaceuticals in wastewater; and (c) the assessment of the remaining uncertainty of the estimates.
2.
Material and methods
2.1.
Sample collection
Sampling was conducted during a dry weather period from 20th November to 1st December in 2009 at the influent of a municipal STP located in South-East Queensland. The STP serves a growing population of approximately 300,000 to 350,000 (Regional Council data). Sewage volumes, from 8:00 AM to 8:00 AM the following day, ranged from 52.6 to 55.5 ML day1. Daily 24-h composite samples from the raw influent were collected before the primary clarifier, refrigerated at 4 C during collection and kept on ice during transportation. The online signal of the flow sensor in the inlet of the STP was used to control the speed of a peristaltic sampling pump: a continuous flow-proportional side stream (a few mL min1) ensured representative samples (Ort et al., 2010b). Samples were preserved at pH 2 with 2M hydrochloric acid on site and stored at 20 C until analysis in amber glass bottles which were prerinsed with dichloromethane and methanol.
2.2.
Chemicals of interest
We measured residues of ten illicit drugs (five parent compounds and five primary metabolites) and four psychoactive
opioids that are most frequently used in Australia according to the National Drug Strategy Household Survey 2007 (AIHW, 2008). These included cocaine, benzoylecgonine, ecgonine methyl ester, amphetamine, methamphetamine, 3,4-methylenedioxymethamphetamine (MDMA), 3,4-methylenedioxyethamphetamine (MDEA), 3,4-methylenedioxyamphetamine (MDA), Δ9-tetrahydrocannabinol (THC), 11-nor-9-carboxy-D9tetrahydrocannabinol (THC-COOH), methadone, 2-ethylidene1,5-dimethyl-3,3-diphenylpyrrolidine (EDDP), codeine and morphine. Furthermore, we included and analyzed five prescription pharmaceuticals, i.e. atenolol, carbamazepine, gabapentin, hydrochlorothiazide and venlafaxine, to assess their suitability to refine the estimation of the number of people that contributed to wastewater in our catchment. Additionally, we also analyzed acesulfame, an artificial sweetener contained in beverages and food that are widely consumed. Acesulfame is almost completely excreted as parent compound, persistent in wastewater and can be measured accurately.
2.3.
Chemical analysis
We adopted and validated our analytical method to analyze illicit drugs and opioids from those reported in the literature (e.g., Boleda et al., 2007; Castiglioni et al., 2006, 2008). Briefly, acidified and filtered samples (200 mL) were spiked with the deuterated analogs of targeted chemicals (10e100 ng) and then loaded onto Oasis MCX cartridges (6 cc, 150 mg, 30 mm) preconditioned with methanol (6 mL), Milli-Q water (4 mL) and HCl-acidified Milli-Q water at pH 2 (4 mL). Subsequently, cartridges were dried by centrifugation at 3000 rpm for 3 min. The target analytes were eluted into two different fractions with methanol and then 2% ammonia hydroxide in methanol. Both fractions were concentrated using a gentle stream of high purity nitrogen gas. Final extracts of the first and second fractions were reconstituted with methanol and an aqueous solution of 5% acetonitrile and 0.1% formic acid, respectively, and then measured by high-performance liquid chromatography coupled to a triple quadrupole tandem mass spectrometer. For prescription pharmaceuticals and acesulfame, filtered samples (400 mL) were spiked with deuterated standards (8e40 ng) and then directly analyzed (for details see Tables S1 and S2, in the Supplementary Information). The mean recovery of the reported illicit drug and opioid residues in Milli-Q water and wastewater samples were in the range of 89e106% (inter-day RSD 4e19%) and 83e113% (inter-day RSD 2e19%), respectively. The inter-day precision of the quantification of prescription pharmaceuticals and acesulfame was 3e20% (for details see Table S2).
2.4.
Back calculation
The common approach to estimate the daily consumption of a drug normalized per 1000 inhabitants in a given catchment by means of wastewater analysis can be found e.g. in Zuccato et al. (2008): 0
1 mg Ri Ci $F$ day Ei B C Daily drug consumptioni @ A ¼ P 1000 people
4440
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 3 7 e4 4 4 8
where Ci is the concentration of a given drug residue i (parent drug or metabolite) measured in raw wastewater samples, F is the total flow during the sampling period (typically 24 h), P is the number of people in the catchment, Ri is the ratio of molar mass of parent drug to its metabolite and Ei is the average excretion rate of a drug residue i (see Table 1). Here we present a set of equations (see Table 2) to consequently assess the uncertainties associated with each step in a transparent way, from the drug load in the sewer (Lsewer), over the back calculation to the consumed mass in the catchment (Mcatchment) to the consumed mass normalized per 1000 people (MENCP). The uncertainties (U) affecting the accuracy of Lsewer are US (sampling), UC (chemical analysis) and UF (flow measurement). Additionally, for Mcatchment, UE (excretion rate) is considered; UB (biodegradation in sewer) was not assessed in our study. The magnitude of uncertainty was determined either experimentally in the laboratory (UC), adopted from the literature (UE), estimated based on a model (US) or specified by STP operators (UF). While the values in our equation may not be independent (e.g. higher flow due to rain may imply lower concentrations of illicit drugs), the tools and methods to quantify each value are independent. Hence, we can apply Gaussian error propagation to calculate the total uncertainty (for multiplications and divisions of independent uncertainties the squared RSDs can be summed). The uncertainty of ENCP is calculated based on the four individual uncertainty components introduced above.
2.4.1.
Load in sewer (Lsewer)
Three factors of uncertainty influence the calculation of Lsewer (Eq. 1.2): US and UC, both affecting the accuracy of the average concentration (C), and UF. In theory, with the sampling mode we applied, US would be zero. However, we conservatively assign US a value of 5% to account for unforeseen or unknown uncertainties; as we will see later, this value will not contribute substantially to the total uncertainty. The values for UC were determined based on the inter-day precision (RSD) in the wastewater matrix which is 2e19% for the illicit drug and
Table 1 e Pharmacokinetic data of targeted illicit drug residues. Parent drug Cocaine Methamphetamine MDMA THC
Illicit drug residue Cocaine Benzoylecgonine Methamphetamine Amphetamine MDMA THCeCOOH
Ei (range) (%) b,c,e,i
7.5 (0.08e15) 35 (14e55)b,c,e,i 39 (2e76)d,g 5.5 (2e76)d,g 15 (6e25)a,j 0.6d,h(20e35)f,k
Ri 1.0 1.1 1.0 1.1 1.0 0.9
a Abraham et al., 2009. b Ambre et al., 1984. c Ambre et al., 1988. d Baselt, 2008. e Cone et al., 2003. f Grotenhermen, 2003. g Postigo et al., 2008. h Zuccato et al., 2008. i From 26 subjects over different administration routes (intravenous, intranasal and smoke). j From 32 subjects in oral dose. k Acid metabolites in urine.
opioid residues and 3e20% for the prescription pharmaceuticals and acesulfame (for details see Table S2). Usually, when asking the STP staff for a value for UF, only precision (factory settings) can be obtained. These random errors are typically very small and cancel each other out over the course of a day when integrating the flow to calculate the daily wastewater volume. However, systematic errors are difficult to recognize and quantify (e.g. offset from installation or deterioration over time). Through dialogue with the staff of the STP sampled in this study, a conservative estimate of 20% was made, which seems reasonable in view of other studies (e.g. Thomann, 2008).
2.4.2.
Consumed mass in the catchment (Mcatchment)
Two additional uncertainty factors are involved in the back calculation to estimate Mcatchment based on Lsewer: UE and UB. The effect of UB was not assessed but is discussed in section 3.4. The variable (uncertain) excretion rate (UE) of a user is compoundspecific and was estimated from literature (Table 1). Values for excretion rates are reported sparsely in the literature and sometimes only ranges instead of individual values are specified; it is difficult to meaningfully derive a distribution and average from a small number of values or a minimum to maximum range. To avoid excluding values outside the reported min to max range, we assumed a normal distribution instead of a uniform distribution for this study and calculated the average as (min þ max)/2. The standard deviation of the normal distribution was calculated as ((maxmin)/2)/O(3). The estimated standard deviation was then used to calculate the relative standard deviation (i.e. UE) for the error propagation. Taking the excretion of cocaine as an example, the reported range is 0.08e15% (Table 1). The average is 7.5% [(0.08 þ 15)/2] and the standard deviation is 4% [((15e0.08)/2)/ O(3)]. For the error propagation UE needs to be a RSD, which in this example is about 57% [4/7.5]. It has to be noted that this value for UE is for one single user. The combined excretion rate of many users will tend toward the mean and the uncertainty UE for n users decreases by O(n) (see Eq. 2.2). The consumed mass of cocaine was back calculated from both cocaine and benzoylecgonine measurements, the mass for methamphetamine was back calculated from both methamphetamine and amphetamine measurements, the mass for MDMA was back calculated only from MDMA measurements and the mass for THC was back calculated only from THC-COOH measurements (Table 1)
2.4.3.
Estimated number of contributing persons (ENCP)
According to Eq. 3.1, the number of people (ENCP) that are present in the catchment and actively contributed to the sampled wastewater can be derived from measured mass loads of prescription pharmaceuticals in the sewer relative to the expected consumption (national audit data, PBS 2009) after correcting for excretion rates (see Table 3). This idea holds true if changes in levels of prescription pharmaceuticals that need to be consumed regularly in consistent amounts are assumed to reflect changes in the overall number of people in the catchment. A similar approach has been proposed in a recent review to reduce errors and thus advance the utility of the back calculation (Daughton, 2011). If no local or regional consumption data are available for the catchment and prescription drug of interest one must assume a homogenous
4441
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 3 7 e4 4 4 8
Table 2 e Equations used to back calculate consumed drug masses and assess effects of uncertainties. Eq.
Formula
Eq.
Lsewer;i ¼ Ci $F
1.1
Uncertainty qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ULsewer;i ¼ U2Si þ U2Ci þ U2F vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi !2 u u UE UMcatchment;i ¼ tU2Si þ U2Ci þ U2F þ pffiffiffiffii ffi þ ½U2B;i ni vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi !2 u u UEj UENCPj ¼ tU2Si þ U2Ci þ U2F þ pffiffiffiffiffi þ ½U2B;j nj vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ffi !2 !2 u u U U E E j t 2 2 2 2 i UMENCPi;j ¼ U2Si þ U2Ci þ pffiffiffiffiffi þ ½UB;i þ USi þ UCj þ pffiffiffiffiffi þ ½UB;j nj ni
1.2 Ci $F$REii
2.2
R
3.2
2.1
Mcatchment;i ¼
3.1
ENCPj ¼ Cj $F$Ejj $TPAAUS j 365
4.1 MENCPi;j ¼
Ci $
Ri Ei
4.2
R
Cj $Ejj $TPAAUS j 365
Eq. 1.1e1.2: Ci is the measured concentration of a given drug residue i (parent drug or metabolite) measured in raw wastewater samples; F is the total flow during the sampling period (typically 24 h); Lsewer: chemical load in sewer; USi is the uncertainty of sampling; UCi is the uncertainty of chemical analysis; UF is the uncertainty of flow measurement; ULsewer is the uncertainty of Lsewer. To the consumed mass of the parent drug; Eq. 2.1e2.2: Ri is the ratio of molar mass of parent drug to its metabolite; Ei is the average excretion rate of a drug residue i; UEi is the uncertainty of excretion rate; ni is the number of users in the catchment; USi is the uncertainty of chemical biodegradation; Mcatchment;i is the consumed mass of the parent drug; UMcatchment;i is the uncertainty of calculating the consumed mass of the parent drug; Eq. 3.1e3.2: Cj is the measured concentration of the prescription pharmaceutical j (parent drug) in raw wastewater samples; TPAUS is the total population in Australia in 2009 (22 million) (ABS, 2009); Aj is the annual consumption of the prescription pharmaceutical; UENCPj is the uncertainty of calculating ENCP; Eq. 4.1e4.2: MENCPi;j is per capita consumption of the illicit drug in the catchment normalized with ENCP; UMENCPi;j is the uncertainty of MENCPi;j .
distribution of the annual national consumption. It is then necessary to demonstrate that the demographics in the catchment do not significantly differ from the national average. As shown in Eq. 3.2, UENCP is equal to that of Mcatchment (Eq. 2.2) for prescription pharmaceuticals.
2.4.4.
Per capita consumption in the catchment (MENCP)
The consumed mass of a drug in the catchment (Mcatchment) is normalized with ENCP to calculate the consumption expressed per 1000 people (MENCP). The remaining uncertainty is calculated according to Eq. 4.2. The values for the individual uncertainty components are reported in Table 4. It should be noted that the uncertainties due to systematic flow measurement errors cancel each other out: for example, an erroneously high flow reading would result in an overestimation of Mcatchment of any illicit drug, but also for any load of prescription pharmaceuticals, resulting in an equivalent overestimation of ENCP.
2.5.
Statistical analysis
The relationships among the measured loads of illicit drug and opioid residues, prescription pharmaceuticals and acesulfame in the samples were examined using non-parametric Spearman’s rank correlation because the dataset did not follow a normal distribution (KolmogoroveSmirnov test). The statistical analysis was carried out in XLSTAT e Pro 7.1 software. The results are shown in Table S3 (Supplementary Information).
3.
Results and discussion
3.1.
Estimated drug loads in sewer (Lsewer)
Table 5 summarizes the measured daily loads of illicit drug and opioid residues, prescription pharmaceuticals and acesulfame in the sewer over 12 monitoring days (see also Figs. S1aec,
Table 3 e Total annual consumption and pharmacokinetic data of five prescription pharmaceuticals and one artificial sweetener. Chemicals Atenolol (beta-blocker) Gabapentin (anti-convulsant) Hydrochlorothiazide (cardiac agent, diuretic/kidney treatment, etc) Methadone (addiction treatment, etc) Venlafexine (anti-depressant) Acesulfame (artificial sweetener) a Data from Australia PBS (2009). b average excretion fraction as parent drug in urine. c Australian Medicines Handbook (2010). d Baselt (2008). e Buerge et al. (2009). f Howell et al., 1993 g Lienert et al. (2007). h eMIMS Version 5.01.0100, 2010.
Total consumptiona (kg/year)
Eb (range) (%)
Dosec,h (mg)
6906 6718 3748 107 10709 Not available
37 (33e40)g 78.5 (76e81)d 82 (68e95)g 27.5 (5e50) d 4.7 (SD ¼ 3.1)f 100.5 (100e101)e
25, 50, 100 100, 300 12.5, 25, 200 10, 60, 80 37.5, 300 Not available
4442
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 3 7 e4 4 4 8
Table 4 e Evaluated uncertainty values (RSD) related to sampling, chemical analysis, flow measurement and excretion rate. Illicit drugs and prescription pharmaceuticals
Cocaine CocaineBenzoylecgonine Methamphetamine MethamphetamineAmphetamine MDMA THC Methadone Atenolol Gabapentin Venlafaxine Hydrochlorothiazide Acesulfame
US
5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0
UC
1.6 3.5 2.7 6.0 6.8 4.0 4.2 12 5.3 20 8.9 3.3
UF
20 20 20 20 20 20 20 20 20 20 20 20
ULsewer
21 21 21 21 22 21 21 24 21 29 22 21
UE (n¼1)
57 34 55 16 35 16 47 5.5 1.8 62 10 0.3
UMcatchment
UMcatchmentðnÞ
if n ¼ 1
if n ¼ 100
n
61 40 59 27 41 26 52 24 21 68 24 21
21 21 22 22 22 21 22 24 21 29 23 21
880 670 2960 3330 460 2350 126 6370 1850 9760 1450 nd
21 21 21 21 22 21 21 24* 21* 29* 22* 21*
UMENCP
14 14 14 15 16 14 15 18 15 24 17 nd
CocaineBenzoylecgonine and MethamphetamineAmphetamine: back calculation using the metabolites benzoylecgonine for cocaine and amphetamine for methamphetamine; n: estimated average number of users per day in the catchment when assuming that one person takes one dose per day (average dose: cocaine 100 mg, methamphetamine 20 mg, MDMA 100 mg, THC 125 mg, methadone 50 mg, atenolol 50 mg, gabapentin 200 mg, venlafaxine 170 mg, hydrochlorothiazide 80 mg); nd: not determined. * ¼ UENCP.
Supplementary Information). Of the ten illicit drugs and four opioid residues targeted in the analyses, codeine was found at the highest daily loads (up to 185 g day1). Cocaine, benzoylecgonine, methamphetamine, MDMA and morphine were also measured at relatively high daily loads (3.5 g day1
(minimum for MDMA) to 75 g day1 (maximum for morphine)). Loads of amphetamine, MDA, THCeCOOH, methadone and EDDP were approximately one order of magnitude lower than the other illicit drug and opioid residues (0.65 g day1 (minimum for MDA) to 5.1 g day1 (maximum for EDDP)).
Table 5 e Results in range (median) of the illicit drug and psychoactive opioid residue, prescription pharmaceutical and an artificial sweetener in raw wastewater over the 12 monitoring days. Chemicals
Measured daily loads in sewer (Lsewer) [g/day]
Illicit drug and psychoactive opioid residue Cocaine 3.96e11.9 (5.83) Benzoylecgonine 10.9e45.1 (20.3) Ecgonine methyl ester nd Amphetamine 2.40e4.66 (3.18) Methamphetamine 17.1e34.9 (21.9) MDMA 3.52e13.6 (6.73) MDA 0.653e4.61 (1.78) MDEA nd THC nd THC-COOH 1.23e2.70 (1.90) Morphine 39.2e75.1 (46.9) Codeine 116e185 (136) EDDP 2.54e5.13 (3.49) Prescription pharmaceutical Atenolol 73.6e199 (109) Gabapentin 213e412 (291) Hydrochlorothiazide 72.5e160 (83.3) Methadone 1.35e2.36 (1.64) Venlafaxine 65.1e131 (71) Artificial sweetener Acesulfame 1360e3180 (1760)
Parent drug consumption (census based) [mg/day/1000 peoplea]
Parent drug consumption (ENCP based) [mg/day/1000 peopleb]
176e531 (259) 109e450 (222)d nd 161e312 (213)e 146298 (187) 76.6e297 (147)f nd nd nd 623e1370 (964)g nd nd 37.6e75.9 (51.6) 663e1790 (986) 905e1750 (1240) 295e649 (338) 16.4e28.6 (19.9) 4340e8750 (4720) 4520e10600 (5870)
nd: not determined. a P in Eq. (1) is a constant value estimated from council’s census data with P ¼ 300,000 persons. b P ¼ ENCP evaluated from atenolol. c calculated from its annual consumption of 22 million Australians. Predicted parent illicit drug consumption. d cocaine. e methamphetamine. f MDMA. g THC.
129e614 (202) 98.8e521 (160)d nd 123e360 (176)e 110e345 (145) 44.7e343 (116)f nd nd nd 364e1410 (908)g nd nd 34.8e60.8 (44.0) 860c 837c 467c 13.0c 1330c Not available
4443
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 3 7 e4 4 4 8
The results clearly reflect different patterns among illicit drug and opioids, prescription pharmaceuticals and acesulfame residues in the wastewater over the 12 monitoring days (Figs. S1aec, Supplementary Information). Overall, we found a positive correlation among the loads of different prescription pharmaceuticals, opioids and acesulfame over the sampling period. However, no significant correlation between illicit drug residue loads and prescription pharmaceutical loads or acesulfame loads could be identified (Table S3, Supplementary Information). This finding implies that a higher number of people in the catchment does not necessarily relate to a higher level of illicit drug consumption. While this is not relevant when assessing Lsewer, it must be taken into consideration when determining per capita drug consumption; i.e. applying an ENCP instead of using one number for the population over the monitoring period. Ecgonine methyl ester, MDEA and THC were not quantifiable in the samples. In laboratory tests we found that ecgonine methyl ester was only efficiently recovered from the acidified Milli-Q water (Table S2) but not from the wastewater samples along the whole analytical method, which suggests that ecgonine methyl ester may not be stable in the acidified wastewater matrix. The concentrations of THC-COOH in the samples were found to be between the limits of detection and quantification over the sampling period. Prescription pharmaceuticals were measured in greater daily loads than all illicit drug residues in the samples (Table 5): atenolol (73.6e199 g day1), gabapentin (213e412 g day1), hydrochlorothiazide (72.5e160 g day1) and venlafaxine (65.1e131 g day1). The measured loads of methadone were considerably lower than those of the other prescription pharmaceuticals. Carbamazepine was consistently below the limit of quantification (1 mg L1) and therefore not reported here. Acesulfame, an artificial sweetener, was found in the greatest mass loads among all the analyzed chemicals, at about 9.3 2.5 mg person1 day1 which is comparable to the data from nine STPs in Switzerland (10 3.4 mg person1 day1) (Buerge et al., 2009).
3.2.
Estimated number of contributing persons (ENCP)
3.2.1.
Selection of a suitable marker
The selection of prescription pharmaceuticals as suitable markers to calculate ENCP was based on the following criteria:
1) wide and common usage; 2) known consumption data; 3) frequently detected at relatively high concentrations in wastewater samples in Australia (Ort et al., 2010a); 4) assumed not to significantly biodegrade in sewers; and 5) readily measurable without sample extraction. Gabapentin, hydrochlorothiazide, methadone and venlafaxine were found to be inappropriate because at least one of the above criteria was not met: (a) a relatively small number of patients consume these prescription pharmaceuticals (methadone and gabapentin); (b) the annual PBS consumption data did not capture a substantial proportion of some of the drug use (i.e. in the case of methadone when dispensed through the Methadone Program in hospitals and drug treatment facilities); (c) inaccurate prescription and/or excretion data (i.e. in the application of venlafaxine for which the ENCP resulted in a substantial overestimation of the population); or (d) non-representative or non-homogenous usage (i.e. the database suggests that the use of many of the hydrochlorothiazide analogs is much lower in Queensland than in the rest of Australia (PBS 2009)). This is the reason why mean values for ENCP based on different prescription pharmaceuticals vary significantly (Table 6). Atenolol was found to fulfill the criteria best since (a) the prevalence of atenolol use is relatively high (based on daily consumption and daily doses about 1e3% of the Australian population use atenolol (Table. 6), which needs to be consumed on a regular basis by patients); (b) it is expected to be relatively persistent and mobile in sewers since it is not effectively removed in wastewater treatment plants (Castiglioni et al., 2006; Mie`ge et al., 2009; Onesios et al., 2009); and (c) the per capita atenolol consumption in Queensland is representative for Australia. Twenty percent of the total Australian population lived in Queensland (ABS, 2009) and about 22% of the total atenolol usage in Australia is in Queensland (PBS Item Reports, 2009; Medicare Australia Statistics). Furthermore, the age demographic for atenolol usage (greater than 65 years of age, ABS, 2006), in Australia’s and Queensland’s populations is about 13% and 12% respectively, which is comparable to that in our study catchment of 14% in 2009 (ABS, 2009). This implies that this age group in our study area is representative for Queensland and Australia and so is the average atenolol consumption.
Table 6 e Daily estimated number of contributing people (ENCP) for 12 days according to different prescription pharmaceuticals. Prescription pharmaceuticals Atenolol Gabapentin Hydrochlorothiazide Methadone Venlafaxine
Rangea of ENCPs (1000 inhabitants)
Mean of ENCP (1000 inhabitants)
CVb (total observed variation, n ¼ 12)
Average percentage of users [% of total population]c
240e620 340e620 190e410 340e640 930e1950
370 440 250 470 1200
29% 19% 31% 20% 27%
3, 2, 1 0.8, 0.3 3.7, 1.9, 0.2 0.02, 0.02, 0.13 3.6, 0.4
Estimated population: 300,000e350,000 (based on census data). a minimumemaximum, for details see Table S4. b Coefficient of variation. As a normalized standard deviation, it covers approximately the range/accuracy, one would/could expect based on the complete uncertainty analysis of the whole method if the population was assumed to be constant. However, it is unknown how much the number of people actually varied (inter- and intra-day) due to commuting or any special events during sampling. Therefore, individual values of the ENCP can plausibly be significantly larger or smaller on a daily basis c references for doses are reported in Table 3.
4444
3.2.2.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 3 7 e4 4 4 8
ENCP based on atenolol loads
The average consumed mass of atenolol in the catchment over 12 days was 299 g day1 and results in an average ENCP of approx. 370,000 with an uncertainty of 24% (UENCP). This is in line with the estimate from the Regional Council (growing population of 300,000e350,000). The total observed variability of the ENCP based on measured atenolol loads over the 12 monitoring days was 29% (Table 6). If the methodological uncertainty (UMcatchment ¼ UENCP for atenolol ¼ 24%, Table 4) is subtracted from the total observed variation, a value of 16% [ ¼ O(292e242)] remains, which can be attributed to ‘real’ variation of the number of people in the catchment. This value seems to be reasonable in view of the fact that the sampling period coincided with the end of the school term where considerable movement of people and subsequent variation from the resident population could be expected (i.e. recreational visitors and seasonal tourists, festivals or events). If the total observed variation was smaller than the complete uncertainty estimate of the method for an individual value, none of the observed variation could be attributed to ‘real’ variation.
3.3.
Estimated drug consumption per 1000 people
Table 5 and Fig. 1 show the estimated drug consumption per 1000 people based on both census data and ENCP over the 12-day monitoring period. The rank order of estimated drug consumption was: THC > methamphetamine z cocaine > MDMA. Consumption of cocaine, methamphetamine and MDMA generally increases during weekends and declines during weekdays. This pattern is more distinct when calculated with ENCP (Fig. 1B) instead of a single value based on census data (Fig. 1A) since the latter cannot account for day-to-day variation of the people in the catchment.
3.3.1.
Cocaine
The daily consumption of cocaine estimated from cocaine loads was slightly higher than the estimate based on its primary metabolite benzoylecgonine (average difference about 15%). Benzoylecgonine is expected to result in a more reliable estimation of cocaine consumption because of its higher persistence in wastewater and because it originates only from human excretion (van Nuijs et al., in press). The high market value of cocaine and the strong correlation with the benzoylecgonine pattern suggest that dumping of high levels of cocaine into the wastewater is unlikely. The observed difference may indicate that the fraction of excreted benzoylecgonine and/or cocaine used in our study was slightly elevated and/or declined respectively to some extent. This may be related to the route of administration (Cone et al., 1998). For example, users in Australia typically snort or inject cocaine (Stafford and Burns, 2010), whereas there is more ‘crack’ (smoked) cocaine use in North America (DASIS, 2007).
3.3.2.
Amphetamine and methamphetamine
We recognize that both amphetamine and methamphetamine consumption results in the excretion of amphetamine (Baselt, 2008). Police seizure data indicate that the use of amphetamine is negligible in Australia (ACCR, 2010). Hence, we assumed that the measured trace amounts of amphetamine
in the samples arise from methamphetamine consumption. This is substantiated by the fact that methamphetamine loads are consistent when back calculated from either amphetamine or methamphetamine.
3.3.3.
THC
THC showed much less daily variation over the period. We also found that its metabolite THCeCOOH did not correlate with other illicit drug and opioid residues, opioids or even prescription pharmaceuticals and acesulfame (Table S3). The lack of inter-day variability in THC in the wastewater results could be more related to the long excretion half-life of THC and its metabolites (range 0.8e9.8 days, average 3 days) (Baselt, 2008) than the actual weekly usage pattern of THC.
3.3.4.
Comparison with other studies
To our knowledge this study provides the first published data on the consumption of illicit drugs estimated from wastewater analysis in Queensland, Australia. Irvine et al. (in press) have recently measured three illicit drug residues in metropolitan and rural areas in the State of South Australia. While MDMA and methamphetamine loads in our catchment were similar, benzoylecgonine loads were about five to six times higher. In general the observed weekly pattern of illicit drug use was similar to the ones described in other studies in Europe and North America (e.g., Huerta-Fontela et al., 2008; Karolak et al., 2010; Terzic et al., 2010; van Nuijs et al., 2009b; Zuccato et al., 2008). However, the absolute average levels of individual illicit drug loads per 1000 people vary (e.g. lower benzoylecgonine).
3.4.
Uncertainty assessment
To evaluate the results it is important to communicate the expected uncertainty associated with the estimations (Ort et al., 2010a). As can be seen from Table 4 the uncertainty due to unknown systematic errors in flow measurements (UF) dominate the uncertainty estimates for the chemical loads in sewers in our study. This uncertainty can only be reduced with a conscientious calibration of flow meters. The uncertainty due to a wide range of reported excretion rates for individual users may suggest that the back calculation for consumed masses in the catchment is subject to even higher uncertainty. This uncertainty decreases with the number of users. In our catchment we estimate more than 100 users for each illicit drug (Table 4) which implies a reduction by at least a factor of 10 [O(100)] for the uncertainty of the ‘combined’ excretion rate by many users. However, systematic errors are not compensated by a large number of users. The limitation of the excretion data available in the literature have been recently reviewed and highlighted (van Nuijs et al., in press). For example, the published clinical data originate from urinary analysis of a very limited number of young or healthy men and adults (Ambre et al., 1984, 1988; Baselt, 2008; Cone et al., 2003). Furthermore, poly-consumption of drugs (i.e. drugedrug interactions) and different administration routes such as intravenous, intranasal, smoked and/or the combination with alcohol can result in a higher or lower excretion rate which is also not taken into account. To date comprehensive data on biodegradation of illicit drugs in sewers is missing. It must be assumed that
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 3 7 e4 4 4 8
4445
Fig. 1 e Daily consumption of illicit drugs per 1000 people over a 12-day monitoring period. A: using census data (300,000 persons) to normalize for population size (error bars in Fig. 1A indicate total uncertainty for the estimated per capita consumption (UMcatchment)) and B: using ENCP (estimated number of contributing people based on atenolol loads (see text for more details). The error bars in Fig. 1B indicate the total uncertainty for the estimated per capita consumption (UMENCP, see Table 4 and Eq. 4.2). Gray shadows indicate weekend days. COC: cocaine; COCBE: cocainebenzoylecgonine; MA: methamphetamineamphetamine; MAAM: methamphetamineamphetamine. The subscripts in cocainebenzoylecgonine and methamphetamineamphetamine imply that the consumed load of cocaine was back calculated from the metabolites benzoylecgonine and methamphetamine from its metabolite amphetamine. biodegradation leads to a systematic underestimation of Mcatchment. The effects of temperature, pH, hydraulic residence time, biofilm, total suspended solids, etc. have not been assessed yet and therefore we cannot assign a justified value
for UB at this stage. However, we assume that the effect of biodegradation is more or less constant within a given sewer system and over a short sampling period (i.e. days) and that inter-day variability is negligible. This may not hold true when
4446
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 3 7 e4 4 4 8
data among different locations or within a location over a longer time span (i.e. year, seasonal effects) is compared. The uncertainty of the estimated number of people contributing to the wastewater is in our case the uncertainty of the estimated mass of consumed atenolol (UENCP ¼ 24%). Since census data only accounts for the permanent residential population and not for tourists or commuters, a meaningful, independent direct comparison is difficult. However, the census data (300,000e350,000 people) is within the 95% confidence interval of observed ENCP (370,000) and observed coefficient of variation. Another systematic underestimation might be caused by exfiltration: through leaky sewers causing a part of the wastewater collected in the catchment to never arrive at the inlet of the STP under investigation (Rieckermann et al., 2007).
3.5.
Potential strengths, limitations and alternatives
The advantage of estimating the number of people in the catchment based on a parameter in the collected wastewater samples is two-fold: 1) it can account for day-to-day variation; and 2) systematic errors due to flow measurement uncertainties can be avoided with this normalization approach (Table 2, Eq. 4.1, 4.2). This also holds true qualitatively for effects due to wastewater losses (exfiltration through leaky sewers) and potentially even to biodegradation. With atenolol we suggest a compound for which sales data are known and which is regularly taken by a relatively large fraction in the Australian population (1 in 30e100). To our knowledge this is the first paper objectively assessing methodological uncertainty vs. observed variation: over the 12-day period all observed variation in illicit drug use was larger than the methodological uncertainty (see Table S5, Supplementary Information). We recognize two aspects as limitations of this approach: 1) degree of patient compliance; and 2) stability of compounds. Furthermore, sampling errors must not only be minimized for the determination of illicit drug loads, but also for any substance used to calculate the ENCP to keep the overall uncertainties low. In view of these strengths and limitations we suggest using multiple wastewater quality parameters (beyond other prescription pharmaceuticals), which might be adequate in other countries, to calculate the ENCP: acesulfame is one of them. Unfortunately, the consumption data were not available. However, we still could use it as a normalization factor to assess day-to-day variation, but not from month to month or between different locations, since consumption could be different and change over these time frames. The advantage of using traditional water quality parameters such as biological or chemical oxygen demand (BOD, COD), total nitrogen (Ntot) or phosphorus (P) is that they are often measured routinely by the STP operators. In addition to the limitations outlined above it has to be noted that human excretion is not the only source for BOD/COD/Ntot and P. A study in Switzerland showed an average industrial contribution in the range of 10e40% (depending on the parameter and assumed per capita excretion, unpublished data). Besides knowing the type and level of industry in the catchment one would also have to know their discharge pattern. If it is not accounted
for, it is expected to result in an underestimation of the per capita consumption of illicit drugs during the week (because increased industrial pollutant loads would suggest a higher number of people in the catchment) and a relative overestimation of the weekend consumption if industry is not active during the weekend. NH4eN, which is more exclusively from human excretion might be the parameter which is least affected by industry.
4.
Conclusions
Estimating the number of people contributing to a sample: Among five prescription pharmaceuticals atenolol was found to be the most appropriate candidate to estimate the number of people contributing to the sampled wastewater in the studied catchment. Acesulfame concentration could be used to estimate population and may be a better basis for calculations given its demographic relevance to illicit drug consumption. Unfortunately national consumption figures for this artificial sweetener were not available. We suggest the development of a method based on a combination of prescription pharmaceutical loads, acesulfame loads and conventional parameters for the more precise and reliable estimation of the number of people contributing to a specific wastewater sample. Estimating illicit drug use: We estimated the daily consumption per 1000 persons using both census data and the estimated number of contributing people in the wastewater derived from atenolol loads, respectively: 623e1370 mg and 364e1410 mg (cannabis (THC)); 176e531 mg and 129e614 mg (cocaine predicted based on cocaine), 109e450 mg and 98.8e521 mg (cocaine predicted based on benzoylecgonine); 146e298 mg and 110e345 mg (methamphetamine predicted by methamphetamine), 161e312 mg and 123e360 mg (methamphetamine predicted by amphetamine); 76.6e297 mg and 44.7e343 mg (ecstasy (MDMA)) in the studied catchment. Higher consumption was found at weekends, which was similar to other studies. For both cocaine and methylamphetamine estimations of the daily consumption based on either the load of the parent drug itself or its metabolites were in good agreement. Uncertainty assessment: Our study describes an approach to estimate the number of people contributing to the wastewater stream and how to quantify the methodological uncertainty. The uncertainty components include sampling, chemical analysis, flow measurement, excretion fraction and estimated number of people that contributed to the wastewater. With current best practice we determined a remaining uncertainty applicable to an estimate of consumed chemical mass to be about 20e30% at the study catchment while that of per capita drug consumption to be about 14e24% (using our method to estimate the number of contributing people). In other words, any difference or change in drug loads smaller than this uncertainty cannot be considered to be significant. This is important for all kinds of studies (i.e. screening, policy, intelligence, operational purposes) and particularly to evaluate the success of intervention strategies aiming at a reduction of drug consumption in a given community or institution (e.g. prison, school).
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 3 7 e4 4 4 8
Acknowledgments The authors sincerely thank Mr. Jake O’Brien (Entox) and Mr. Jack Thompson (Entox) for assisting with sampling and SouthEast Queensland Council for sampling opportunities. We also acknowledge the assistance of Dr. Shalona Anuj-Outten (QHFSS) for instrumental training. Vanna Mabbott and Chris Raymond from The Drug Utilisation Sub-Committee of the Pharmaceutical Benefits Advisory Committee, Department of Health and Ageing, Commonwealth of Australia (providing Australian drug-use statistics). Financial assistance for this research was from the QHFSS/Entox Collaborative Research Funds and the Australian Future Forensics Innovation Network as supported by the Department of Employment, Economic Development and Innovation, through the National and International Research Alliance Program. Wayne Hall is funded by an NHMRC Australia Fellowship. Entox is a joint venture of The University of Queensland and QHFSS. We also thank Dr. Margaret Murphy (City University of Hong Kong) and Dr. Jo¨rg Rieckermann (Eawag) for valuable input and discussion.
Appendix. Supplementary material Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.watres.2011.05.042.
references
Abraham, T.T., Barnes, A.J., Lowe, R.H., Kolbrich Spargo, E.A., Milman, G., Pirnay, S.O., Gorelick, D.A., Goodwin, R.S., Huestis, M.A., 2009. Urinary MDMA, MDA, HMMA, and HMA excretion following controlled MDMA administration to humans. J. Anal. Toxicol. 33, 439e446. ABS (Australian Bureau of Statistics), 2006. 4821.0.55.001 e Cardiovascular Disease in Australia: a Snapshot, 2004e05. ABS, Canberra. Available. http://www.abs.gov.au/ausstats/
[email protected]/mf/4821.0.55.001. ABS (Australian Bureau of Statistics), 2009. 3201.0 e Population by Age and Sex, Australian States and Territories, Jun 2009. ABS, Canberra. Available. http://www.abs.gov.au/Ausstats/abs@. nsf/mf/3201.0. ACCR (Australian Crime Commission Report), 2010. Illicit Drug Data Report 2008e09. Available. http://www.crimecommissi on.gov.au/publications/iddr/index.htm. AIHW (Australian Institute of Health and Welfare), 2008. 2007 National Drug Strategy Household Survey: Detailed Findings. Drug Statistics Series No. 22. Cat. No. PHE 107. AIHW, Canberra. Available. http://www.aihw.gov.au/publications/ index.cfm/title/10674. Ambre, J., Fischman, M., Ruo, T., 1984. Urinary excretion of ecgonine methyl ester, a major metabolite of cocaine in humans. J. Anal. Toxicol. 8, 23e25. Ambre, J., Ruo, T., Nelson, J., Belknap, S., 1988. Urinary excretion of cocaine, benzoylecgonine and ecgonine methyl ester in humans. J. Anal. Toxicol. 12, 301e306. AMH (Australian Medicines Handbook), 2010. The Australasian Society of Clinical and Experimental Pharmacologists and Toxicologists, the Pharmaceutical Society of Australia and the
4447
Royal Australian College of General Practitioners. July 2010 Edition. Available 1998. http://amh.hcn.net.au/view.php. Baselt, R.C., 2008. Disposition of Toxic Drugs and Chemicals in Man, eighth ed., Biomedical Publications, Foster City, CA. Boleda, M.R., Galceran, M.T., Ventura, F., 2009. Monitoring of opiates, cannabinoids and their metabolites in wastewater, surface water and finished water in Catalonia, Spain. Water Res. 43, 1126e1136. Boleda, M.R., Galceran, M.T., Ventura, F., 2007. Trace determination of cannabinoids and opiates in wastewater and surface waters by ultra-performance liquid chromatographytandem mass spectrometry. J. Chromatogr. A 1175, 38e48. Bones, J., Thomas, K.V., Paull, B., 2007. Using environmental analytical data to estimate levels of community consumption of illicit drugs and abused pharmaceuticals. J. Environ. Monit. 9, 701e707. Buerge, I.J., Buser, H.R., Kahle, M., Mu¨ller, M.D., Poiger, T., 2009. Ubiquitous occurrence of the artificial sweetener acesulfame in the aquatic environment: an ideal chemical marker of domestic wastewater in groundwater. Environ. Sci. Technol. 43, 4381e4385. Castiglioni, S., Zuccato, E., Chiabrando, C., Fanelli, R., Bagnati, R., 2008. Mass spectrometric analysis of illicit drugs in wastewater and surface water. Mass Spectrom. Rev. 27, 378e394. Castiglioni, S., Zuccato, E., Crisci, E., Chiabrando, C., Fanelli, R., Bagnati, R., 2006. Identification and measurement of illicit drugs and their metabolites in urban wastewater by liquid chromatographyetandem mass spectrometry. Anal. Chem. 78, 8421e8429. Chiaia, A.C., Banta-Green, C., Field, J., 2008. Eliminating solid phase extraction with large volume injection LC/MSMS: analysis of illicit and legal drugs and human urine indicators in US wastewaters. Environ. Sci. Technol. 42, 8841e8848. Collins, D.J., Lapsley, H.M., 2008. The Costs of Tobacco, Alcohol and Illicit Drug Abuse to Australian Society in 2004/05, ISBN 1-74186437-2 Available. http://www.health.gov.au/internet/drugstrate gy/publishing.nsf/Content/34F55AF632F67B70CA2573F60 005D42B/$File/mono64.pdf Online. Cone, E.J., Sampson-Cone, A.H., Darwin, W.D., Huestis, M.A., Oyler, J.M., 2003. Urine testing for cocaine abuse: metabolic and excretion patterns following different routes of administration and methods for detection of false-negative results. J. Anal. Toxicol. 27, 386e401. Cone, E.J., Tsadik, A., Oyler, J., Darwin, W.D., 1998. Cocaine metabolism and urinary excretion after different routes of administration. Ther. Drug Monit. 20, 556e560. Daughton, C.G., 2001. Illicit drugs in municipal sewage: proposed new non-intrusive tool to heighten public awareness of societal use of illicit/abused drugs and their potential for ecological consequences. In: Daughton, C.G., Jones-Lepp, T.L. (Eds.), Pharmaceuticals and Personal Care Products in the Environment. Scientific and Regulatory Issues. American Chemical Society, Washington, DC, pp. 348e364. Symposium Series 791. Daughton, C.G., 2011. Illicit drugs: contaminants in the environment and utility in forensic epidemiology. Rev. Environ. Contam. Toxicol. 210, 59e110. Degenhardt, L., Hall, W., Lynskey, M., Warner-Smith, M., 2004. Illicit drug use. In: Ezzati, M., Lopez, A.D., Rodgers, A., Murray, R. (Eds.)Comparative Quantification of Health Risks: Global and Regional Burden of Disease Attributable to Selected Major Risk Factors, second ed., chapter 13. World Health Organization, Geneva, pp. 1109e1176. DASIS (Drug and Alcohol Services Information System), 2007. The DASIS Report: Cocaine Route of Administration Trends: 1995e 2005. Substance Abuse and Mental Health Services Administration, Office of Applied Studies, Rockville, MD. Available. http://www.oas.samhsa.gov/2k7/crackTX/crackTX.pdf.
4448
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 3 7 e4 4 4 8
Grotenhermen, F., 2003. Pharmacokinetics and pharmacodynamics of cannabinoids. Clin. Pharmacokinet. 42, 327e360. Howell, S.R., Husbands, G.E.M., Scatina, J.A., Sisenwine, S.F., 1993. Metabolic disposition of 14C-venlafaxine in mouse, dog, rhesus money and man. Xenobiotica 23, 349e359. Huerta-Fontela, M., Galceran, M.T., Ventura, F., 2007. Ultraperformance liquid chromatography-tandem mass spectrometry analysis of stimulatory drugs of abuse in wastewater and surface waters. Anal. Chem. 79, 3821e3829. Huerta-Fontela, M., Galceran, M.T., Martin-Alonso, J., Ventura, F., 2008. Occurrence of psychoactive stimulatory drugs in wastewaters in north-eastern Spain. Sci. Total Environ. 397, 31e40. Irvine, R.J, Kostakis, C, Felgate, P.D, Jaehne, E.J, Chen, C, White, J. M. Population drug use in Australia: a wastewater analysis. Forensic Sci. Int. in press, doi:10.1016/j.forsciint.2011.01.037. Karolak, S., Nefau, T., Bailly, E., Solgadi, A., Levi, Y., 2010. Estimation of illicit consumption by wastewater analysis in Paris area. Forensic Sci. Int. 200, 153e160. Lienert, J., Gudel, K., Escher, B.I., 2007. Screening method for ecotoxicological hazard assessment of 42 pharmaceuticals considering human metabolism and excretory routes. Environ. Sci. Technol. 41, 4471e4478. Metcalfe, C., Tindale, K., Li, H., Rodayan, A., Yargeau, V., 2010. Illicit drugs in Canadian municipal wastewater and estimates of community drug use. Environ. Pollut. 158, 3179e3185. Mie`ge, C., Choubert, J.M., Ribeiro, L., Euse`be, M., Coquery, M., 2009. Fate of pharmaceuticals and personal care products in wastewater treatment plants e conception of a database and first results. Environ. Pollut. 157, 1721e1726. Onesios, K.M., Yu, J.T., Bouwer, E.J., 2009. Biodegradation and removal of pharmaceuticals and personal care products in treatment systems: a review. Biodegradation 20, 441e466. Ort, C., Lawrence, M.G., Reungoat, J., Eaglesham, G., Carter, S., Keller, J., 2010a. Determining the fraction of pharmaceutical residues in wastewater originating from a hospital. Water Res. 44, 605e615. Ort, C., Lawrence, M.G., Reungoat, J., Mueller, J.F., 2010b. Sampling for PPCPs in wastewater systems: comparison of different sampling modes and optimization strategies. Environ. Sci. Technol. 44, 6289e6296. Ort, C., Lawrence, M.G., Rieckermann, J., Joss, A., 2010c. Sampling for pharmaceuticals and personal care products (PPCPs) and illicit drugs in wastewater systems: are your conclusions valid? A critical review. Environ. Sci. Technol. 44, 6024e6035. Prichard, J., Ort, C., Bruno, R., Gartner, C., Kirkbride, P., Hall, W., Lai, F.Y., Carter, S., Mueller, J., Salinas, A., 2010. Developing a method for site-specific wastewater analysis: implications for prisons and other agencies with an interest in illicit drug use. J. Law Inf. Sci. 20, 15e27. Postigo, C., Lopez de Alda, M.J., Barcelo, D., 2008. Analysis of drugs of abuse and their human metabolites in water by LC-MS/MS: a non-intrusive tool for drug abuse estimation at the community level. Trends Anal. Chem., 1053e1069.
Postigo, C., Lopez de Alda, M.J., Barcelo, D., 2010. Drugs of abuse and their metabolites in the Ebro river basin: occurrence in sewage and surface water, sewage treatment plants removal efficiency, and collective drug usage estimation. Environ. Int. 36, 75e84. Rieckermann, J., Bare s, V., Kracht, O., Braun, D., Gujer, W., 2007. Estimating sewer leakage from continuous tracer experiments. Water Res. 41 (9), 1960e1972. Shand, F., Topp, L., Darke, S., Makkai, T., Griffiths, P., 2003. The monitoring of drug trends in Australia. Drug Alcohol Rev. 22, 63e74. Stafford, J., Burns, L., 2010. Australian Drug Trends 2009. Findings from the Illicit Drug Reporting System (IDRS). Australian Drug Trend Series No. 37. National Drug and Alcohol Research Centre, University of New South Wales, Sydney. Available. http://www.med.unsw.edu.au/NDARCWeb.nsf/page/National %25%20Reports. Terzic, S., Senta, I., Ahel, M., 2010. Illicit drugs in wastewater of the city of Zagreb (Croatia) e estimation of drug abuse in a transition country. Environ. Pollut. 158, 2686e2693. Thomann, M., 2008. Quality evaluation methods for wastewater treatment plant data. Water Sci. Technol. 57 (10), 1601e1609. UNODC (United Nations Office on Drugs and Crime), 2010. World Drug Report. Methodology. Available. http://www.unodc.org/ unodc/en/data-and-analysis/WDR-2010.html. van Nuijs, A.L.N., Pecceu, B., Theunis, L., Dubois, N., Charlier, C., Jorens, P.G., Bervoets, L., Blust, R., Neels, H., Covaci, A., 2009a. Cocaine and metabolites in waste and surface water across Belgium. Environ. Pollut. 157, 123e129. van Nuijs, A.L.N., Pecceu, B., Theunis, L., Dubois, N., Charlier, C., Jorens, P.G., Bervoets, L., Blust, R., Neels, H., Covaci, A., 2009b. Spatial and temporal variations in the occurence of cocaine and benzoylecgonine in waste- and surface water from Belgium and removal during wastewater treatment. Water Res. 43, 1341e1349. van Nuijs, A.L.N., Pecceu, B., Theunis, L., Dubois, N., Charlier, C., Jorens, P.G., Bervoets, L., Blust, R., Meulemans, H., Neels, H., Covaci, A., 2009c. Can cocaine use be evaluated through analysis of wastewater? A nation-wide approach conducted in Belgium. Addiction 104, 734e741. van Nuijs, A.L.N., Castiglioni, S., Tarcomnicu, I., Postigo, C., de Alda, M.L., Neels, H., Zuccato, E., Barcelo, D., Covaci, A. Illicit drug consumption estimations derived from wastewater analysis: a critical review. Sci. Total Environ., in press. van Nuijs, A.L.N., Mougel, J.F., Tarcomnicu, I., Bervoets, L., Blust, R., Jorens, P.G., Neels, H., Covaci, A., 2011. Sewage epidemiology e a real-time approach to estimate the consumption of illicit drugs in Brussels, Belgium. Environ. Int. 37, 612e621. Zuccato, E., Chiabrando, C., Castiglioni, S., Bagnati, R., Fanelli, R., 2008. Estimating community drug abuse by wastewater analysis. Environ. Health Persp. 116, 1027e1032. Zuccato, E., Chiabrando, C., Castiglioni, S., Calamari, D., Bagnati, R., Schiarea, S., et al., 2005. Cocaine in surface waters: a new evidence-based tool to monitor community drug abuse. Environ. Health 4, 14e20.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 4 9 e4 4 5 8
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Fulvic acid mediated photolysis of ibuprofen in water Laura E. Jacobs a,1, Ryan L. Fimmen b, Yu-Ping Chin b,*, Heath E. Mash c, Linda K. Weavers d a
Environmental Science Graduate Program, The Ohio State University, Columbus, OH 43210, USA School of Earth Sciences, The Ohio State University, 125 South Oval Mall, 275 Mendenhall Laboratory, Columbus, OH 43210, USA c United States Environmental Protection Agency, 26 W. Martin Luther King Drive, Cincinnati, OH 45268, USA d Department of Civil and Environmental Engineering and Geodetic Science, The Ohio State University, 470 Hitchcock, 2070 Neil Avenue, Columbus, OH 43210, USA b
article info
abstract
Article history:
Photolysis of the non-steroidal anti-inflammatory drug ibuprofen was studied by exposure
Received 22 February 2011
to a solar simulator in solutions of fulvic acid (FA) isolated from Pony Lake, Antarctica;
Received in revised form
Suwannee River, GA, USA; and Old Woman Creek, OH, USA. At an initial concentration of
22 May 2011
10 mM, ibuprofen degrades by direct photolysis, but the presence of FA significantly increases
Accepted 28 May 2011
reaction rates. These reactions proceeded up to 6 faster in FA solutions at lower ibuprofen
Available online 7 June 2011
concentrations (0.1 mM), but the rates are highly dependent upon DOM composition. Incomplete quenching of the reaction in the presence of isopropanol suggests that the
Keywords:
hydroxyl radical is only partially responsible for ibuprofen’s photodegradation in FA solu-
Pharmaceuticals
tions, and other reactive transients likely play an important role. Liquid chromatography-
Ibuprofen
quadrupole time-of-flight mass spectrometry and NMR spectroscopy reveal the formation of
Photolysis
multiple photoproducts, with three byproducts identified as 1-(4-isobutylphenyl)ethanol,
Dissolved organic matter
isobutylacetophenone, and a phenol derivative. Pony Lake FA significantly increases the
Photodecarboxylation
production of the major byproduct relative to yields produced by direct photolysis and the
Fulvic acids
other two FA. Thus, the photolytic fate of ibuprofen in sunlit waters is affected by its initial concentration and the source of dissolved organic matter present. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
Pharmaceuticals and personal care products (PPCPs) have been detected in natural waters (e.g., wetlands, lakes, and rivers) throughout the world and are widely recognized as emerging environmental contaminants (Ternes, 1998; Kolpin et al., 2002; Daughton and Ternes, 1999; Nikolaou et al., 2007; plus many others). While many PPCPs (but not all) can be efficiently removed in wastewater treatment plants (from 90%) these compounds can enter waterways from treated sewage effluent, septic systems and combined sewer overflows (Winker et al., 2008; Lindqvist et al., 2005; Nakada et al., 2006,
2007). Most of these compounds exist at low levels (sub part per billion) and exhibit relatively low acute toxicity (EC50 > 1 mg/L), but little is known about synergistic effects i.e., mixtures and long-term chronic exposure of these substances to aquatic organisms (Fent et al., 2006). Moreover, the toxicological impacts of their unidentified derivatives have not been well studied. In 1999/2000 a United States Geological Survey study detected the pharmaceutical ibuprofen in w10% of 139 streams sampled across the United States at concentrations as high as 5 nM (Kolpin et al., 2002). Ibuprofen is a non-steroidal anti-inflammatory drug (NSAID) and is one of the most widely used over-the-counter medications. It is an alkylbenzene with a carboxylic acid
* Corresponding author. Tel.: þ1 614 292 6953; fax: þ1 614 292 7688. E-mail address:
[email protected] (Y.-P. Chin). 1 Present address: The National Academies, 500 5th St. NW, Washington DC 20001, USA. 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.05.041
4450
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 4 9 e4 4 5 8
functional group, a pKa of 4.8, and a log n-octanol/water partition coefficient (Kow) of 2.48 at pH 7 (Scheytt et al., 2005). Ibuprofen is chiral and its pharmacological activity is attributable to the S-isomer, but is administered as a racemic mixture (Hutt and Caldwell, 1983). Over 70% of ibuprofen is metabolized in the body and excreted in urine. Hydroxylated and carboxylated compounds are the predominant metabolized forms (Hutt and Caldwell, 1983). Despite this loss mechanism, levels of ibuprofen as high as 0.015 mM have been detected in wastewater influent (Buser et al., 1999). NSAIDs, as well as many other classes of pharmaceuticals, are known to degrade by both photochemical and biological processes (Boreen et al., 2003; Packer et al., 2003). With respect to ibuprofen, it is rather efficiently removed in wastewater treatment plants (Winker et al., 2008; Lindqvist et al., 2005; Nakada et al., 2006, 2007), but the rates of ibuprofen degradation vary depending upon the microbial consortia and processes involved e.g., biofilms, membranes, etc. Nonetheless our interpretation of the literature reveals that the biodegradation kinetics of ibuprofen is similar to or slower than photochemical processes (Zwiener and Frimmel, 2003; Quintana et al., 2005; Xu et al., 2009; Yu et al., 2006). Packer et al. (2003) showed that ibuprofen is susceptible to degradation by both direct and indirect photolytic pathways, whereby dissolved organic matter (DOM) was demonstrated to be an important photosensitizer. DOM and in particular the fulvic acid (FA) fraction can generate reactants such as reactive oxygen species (ROS) and triplet DOM (3DOM) (Gerecke et al., 2001; Miller and Chin, 2002; Vaughan and Blough, 1998), which can subsequently transform ibuprofen. Fulvic acids represent the chromophoric portion of DOM and as a result are the most photochemically active fraction (Aguer et al., 1997; Brown et al., 2004). Packer et al. (2003) also showed that the indirect photolysis of ibuprofen occurs in part through reaction with hydroxyl radicals (OH) generated from excited-state DOM. The OH mediated pathway, however, does not explain all the observed reactivity and an additional reactive transient of unknown origin also contributes to the photodegradation of ibuprofen in natural waters (Packer et al., 2003). Several studies have identified a number of PPCP derivatives resulting from photochemical reactions in aqueous media and proposed chemical mechanisms that form these derivatives (Edhlund et al., 2006; Boreen et al., 2005; Pe´rezEstrada et al., 2005). Indeed, light-induced photodecarboxylation of NSAIDs and the derivatives formed from this pathway have been studied in both non-aqueous (Castell et al., 1987; Budac and Wan, 1992; Bosca´ et al., 1994, 2001; Monti et al., 1997) and aqueous systems (Pe´rez-Estrada et al., 2005). This process is a mechanism whereby a benzylic radical is formed followed by the subsequent ejection of CO2 and a solvated electron. This electron can then be scavenged by triplet molecular oxygen (O2), generating superoxide (,O 2) leading to a series of oxygenated byproducts (Bosca´ et al., 1994). Castell et al. (1987) proposed the same mechanism for the decomposition of ibuprofen when photolyzed by a medium pressure Hg lamp in methanol. This irradiation gave rise to several byproducts via light-induced excitation of ibuprofen, which cleaves the CeC bond a to the carboxyl group generating a benzylic radical intermediate followed by
hydrogen abstraction, dimerization, methanol addition, or reaction with dioxygen to generate various byproducts. This study examines the mechanism(s) responsible for the photochemical degradation of ibuprofen in the aquatic environment and the effect of DOM (specifically aquatic fulvic acids) on this process. We studied the direct and indirect photolysis of ibuprofen in the presence of fulvic acids that represent the continuum of DOM derived from terrestrial and microbial processes. We chose to use fulvic acids rather than whole water DOM because the manner in which they were processed (by XAD chromatography from a single sampling event) rendered them temporally stable i.e., experiments can be repeated using the same batch of freeze-dried FA. In addition we examined how the initial concentration of ibuprofen influences indirect photolysis in the presence of DOM. Finally, we examine the chemical mechanism of ibuprofen degradation through the identification of its photoderivatives and an analysis of byproduct production in the presence of various fulvic acids.
2.
Methods
2.1.
Chemicals and fulvic acids
Milli-Q (18 MU-cm, Millipore) water was used for all aqueous solutions. All solvents (acetonitrile, hexane, methanol, isopropanol and ether) and chemicals (hydrochloric acid, isobutylacetophenone, etc.) were reagent-grade or higher. CDCl3 (99.8% -D) and ibuprofen (racemic mixture) with a purity of 99% were obtained from Acros Organics. Suwannee River fulvic acid (SRFA) was obtained from the International Humic Substances Society and Pony Lake fulvic acid (PLFA) provided by Dr. Diane McKnight. Additionally, we isolated a fulvic acid from Old Woman Creek, a National Estuarine Research Reserve wetland in Huron, OH, using the XAD-8 protocol (Leenheer, 1981; Thurman and Malcolm, 1981), which we refer to in this paper as OWCFA.
2.2.
Photolytic reactions
Ibuprofen stock solutions were prepared in acetonitrile. Experimental solutions of 10 mM or 0.1 mM were prepared by addition of an appropriate volume of stock, complete evaporation of the acetonitrile, and re-constitution into the desired aqueous matrix. The pH of the solutions was adjusted to 7 (0.1) with HCl and/or NaOH. Some solutions were prepared anoxically by either sparging the sample with argon for 1 min/mL and transferring it to photolysis tubes within a glove box (95% N2/5% H2) or subjecting solutions to 3 freeze-pump-thaw cycles in a Schlenk line followed by cannula transfer into photolysis tubes. After this treatment, oxygen levels in photolysis solutions were below the limit of detection as determined by both a dissolved oxygen probe and chemical indicators. To investigate the role of DOM on the indirect photolysis of ibuprofen, each solution was spiked with the same amount of the desired fulvic acid (by weight) and an aliquot was used for total organic carbon analysis (Shimadzu TOC-5000). Airtight quartz tubes (1 cm path length capped with Teflon-lined quartz lids) were used for the photolysis experiments.
4451
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 4 9 e4 4 5 8
Photolysis of ibuprofen was conducted using a solar simulator (Atlas Suntest CPSþ) fitted with a 500 W Xenon arc lamp at a temperature of 25 C and a duration equivalent to at least one half-life. Selected photolysis experiments were evaluated over three half-lives in order to determine the appropriate rate model for degradation kinetics. For experiments where the photolytic rates were very slow e.g., direct photolysis and indirect experiments in the presence of scavengers we did not exceed 50 h irradiation time. Chemical actinometry ( p-nitroanisole/pyridine) (Dulin and Mill, 1982) was conducted in both the solar simulator and under natural sunlight in Columbus, OH (40 40 41" N, 83 40 24" W) in June at noon. Actinometry showed no significant change in the light intensity of our solar simulator over the course of experiments and the difference in p-nitroanisole rate constants between our light source and sunlight under the conditions described above was a factor of 3.7 i.e., our light source was 3.7 times more intense than June sunlight at noon in the absence of clouds in Columbus, OH. Dark controls were run concurrently in foil-covered tubes. To ascertain the role of OH in our experiments, we used isopropanol (20 mM) as a hydroxyl radical scavenger (Packer et al., 2003). Temperature, pH, and radiometer readings were monitored during each experiment, with all three parameters remaining consistent. Finally, the procedure of Zhou and Mopper (1990) was used to quantify hydroxyl radical production in sample solutions, and is further described in the Supplemental Information (SI) file.
2.3.
Analysis of ibuprofen and its photoderivatives
Ibuprofen and its photolysis byproducts were analyzed by UVeVisible spectrophotometry (Varian Cary 13) and by highpressure liquid chromatography (HPLC) using a UVeVisible dual wavelength detector at l ¼ 223 nm and l ¼ 264 nm respectively (Waters Corp. 2487, Breeze 3.3 software). Aqueous
samples were directly injected into the HPLC (50 mL) and analytes separated using a Restek C-18 reverse-phase chromatography column. Each sample was eluted isocratically at 1 ml min1 using a 60% acetonitrile and 40% water mobile phase (v/v) buffered with phosphate to pH 3. Photolysis byproducts from experiments ([Ibuprofeno] ¼ 10 mM) were further analyzed using proton nuclear magnetic resonance (1H NMR) spectroscopy in both 1-D and 2-D mode, liquid chromatography-quadrupole time-of-flight mass spectrometry (LC-Q-TOFMS), and gas chromatography mass spectrometry (GCeMS). A detailed description of the methods employed is described in the SI.
3.
Results and discussion
3.1.
Photolysis experiments
Ibuprofen (10 mM initial concentration) degrades very slowly by direct photolysis, with a pseudo first order rate constant of 0.0025 h1 (Table 1). We observed no statistical difference between ibuprofen as a racemic mixture or its “S” isomer (Table 1). This translates to a half-life greater than 200 h, and is nearly identical to the half-life reported by Lin and Reinhard (2005) who used the same light source and faster than in natural sunlight (600 h) as reported by Yamamoto et al. (2009). Ibuprofen degrades approximately 5 times faster in the presence of each fulvic acid than by direct photolysis (Table 1: Fig. 1). We attribute this slow direct photolytic rate to its low absorbance at wavelengths present in sunlight (>290 nm: Fig. 2), and the observed increase in its reaction kinetics in the presence of DOM can be attributed to indirect pathways. Surprisingly, the photolytic rate constants in the presence of each fulvic acid are statistically identical even though they are derived from different precursor materials i.e., higher plants vs. bacteria and phytoplankton. Small differences do occur if one normalizes the rate constants to DOC whereby the
Table 1 e 10 mM and 0.1 mM racemic and S-(D) (when indicated) ibuprofen degradation rate constants (kobs hL1) in the presence of simulated sunlight, OWCFA (Old Woman Creek Fulvic Acid), SRFA (Suwannee River Fulvic Acid), and PLFA (Pony Lake Fulvic Acid) at indicated dissolved organic carbon levels (DOC), in the absence of molecular oxygen, and in the presence of isopropanol. (L) Denotes not applicable. [Racemic Ibuprofen]0 ¼ 10 mM
Direct Direct (S-(þ)-Ibuprofen) OWCFA SRFA PLFA SRFA low O2 OWCFA low O2 [Racemic Ibuprofen]0 ¼ 0.1 mM
OWCFA SRFA PLFA SRFA low O2 OWCFA low O2 PLFA low O2
[DOC] mgC/L
kobs (h1)
Half-lives (hours)
Fulvic Acid þ 20 mM Isopropanol kobs (h1)
e e
0.0025 0.001 0.0031 0.001 0.015 0.001 0.016 0.001 0.014 0.001 0.017 0.002 0.019 0.001
277 224 46 43 50 41 36
e e 0.009 0.001 0.007 0.001 0.016 0.001 e e
5.56 7.20 5.45 7.20 5.56
0.18 0.20 0.19 0.20 0.18
[DOC] mgC/L
5.56 7.20 5.45 7.20 5.56 5.45
0.18 0.20 0.19 0.20 0.18 0.19
kobs (h1) 0.019 0.029 0.079 0.015 0.022 0.056
0.001 0.004 0.013 0.002 0.006 0.009
Fulvic Acid þ 20 mM Isopropanol kobs (h1) 36 24 9 46 32 12
0.011 0.002 0.020 0.004 0.013 0.006 e e e
4452
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 4 9 e4 4 5 8
Fig. 1 e Pseudo first order (ln(Concentration/Initial Concentration)) degradation of Ibuprofen (10 mM) direct photolysis degradation and in the presence of Old Woman Creek Fulvic Acid (OWCFA) with and without isopropanol.
transformation of ibuprofen by PLFA w OWCFA > SRFA. Nonetheless we believe that this observation is coincidental as experiments using chemical scavengers reveal significantly different photolytic pathways that appear to be dependent upon DOM composition. DOM can produce a plethora of reactive species, which varies with the fulvic acid composition (White et al., 2003; Hoigne´ et al., 1989; Cooper et al., 1989; Canonica et al., 1995; Canonica and Hoigne´, 1995; Canonica and Freiburghaus, 2001; Halladja et al., 2007, plus many others). Our choice of fulvic acids represents a wide diversity of compositional differences ranging from SRFA (derived from predominantly higher plant detritus) to PLFA (a microbially derived fulvic acid isolated from a hyper-eutrophic lake in Antarctica). Finally, OWCFA’s composition is comprised of organic matter derived from a combination of microbial and terrestrial precursors. The differences in chemical composition of all three FA are well documented in the literature
Fig. 2 e UVeVisible absorption spectrum of a 10 mM solution of ibuprofen.
(Thurman and Malcolm, 1981; Brown et al., 2004; Cawley et al., 2009) and are not presented in detail in this paper. As stated previously, we believe that the actual transformation pathways differed depending upon the fulvic acid used even though the degree of photosensitization appears to be composition independent. By using scavengers or altering experimental conditions we were able to elucidate specific pathways. To probe the role of hydroxyl radicals formed from photo-irradiated fulvic acids, the known OH scavenger, isopropanol, was added to photolysis experiments. While isopropanol is not the most ideal OH scavenger (due to its ability to react with other ROS) these data provide us with a rough estimate of the hydroxyl radical’s role in ibuprofen’s photofate. We observed significant decreases in the degradation rate of ibuprofen for OWCFA (kobs ¼ 0.009 0.001 h1; 46% reduction: see Fig. 1) and SRFA (k ¼ 0.007 0.001 h1; 64% reduction), corroborating observations by Packer et al. (2003) who used the same scavenger. Surprisingly, in PLFA, the addition of isopropanol does not change the rate of ibuprofen degradation (Fig. 3). Measurements of the steady state hydroxyl radical concentration ([OH]ss) corroborate our observations because photolyzed PLFA solutions (8.80 1017 M) yielded OH levels that are a factor of three lower than those measured for OWCFA (2.42 1016 M). Thus, other pathways are more important than OH for PLFA. Details for the determination of [OH]ss are found in SI. The role of ROS produced by irradiated organic matter was studied by photolyzing anoxic ibuprofen solutions in the presence of each fulvic acid. While the rate of ibuprofen degradation in the presence of SRFA remains statistically the same as oxic solutions, its kinetics was enhanced in OWCFA solutions (Table 1). We speculate that in the anoxic OWCFA system ibuprofen may also react via 3DOM oxidation in addition to OH pathways. Ibuprofen has an electron donating branched alkyl group that renders it susceptible to oxidation by 3DOM (Canonica and Freiburghaus, 2001). In the absence of dioxygen more 3DOM would be made available to react with ibuprofen thereby increasing its degradation rate.
Fig. 3 e Photoinduced pseudo first order (ln(Concentration/ Initial Concentration)) degradation of Ibuprofen (10 mM) in Milli-Q (Direct) photolysis and in the presence of Pony Lake Fulvic Acid (PLFA) with and without 20 mM isopropanol.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 4 9 e4 4 5 8
Unfortunately, the PLFA results were inconclusive (due to possible air contamination in the solutions and our limited amounts prevented us from repeating the experiment for this study) and are not included in this paper. Environmental levels of ibuprofen in surface waters are significantly smaller than the concentrations used in this and other studies. Because of ibuprofen’s electron donating alkyl group we investigated whether its indirect photolysis kinetics are influenced by its initial concentration. Canonica and Freiburghaus (2001) observed concentration effects for other electron rich probe molecules (alkyl and methoxy phenols) whereby the reaction kinetics of these compounds increases significantly (w 2 to 3) at low concentrations (0.1 mM vs. 5 mM) when irradiated in the presence of fulvic acids. We conducted similar experiments for all three fulvic acids at the lowest detectable concentration for ibuprofen that can be measured by our HPLC (Co ¼ 0.1 mM with an average detection limit of w0.02 mM). For all three fulvic acids the degradation rate constant increased compared to a Co of 10 mM concentration (Fig. 4). The increase in reaction kinetics varied among the fulvic acids where respective PLFA and SRFA rate constants increased by a factor of 5 and 2, while ibuprofen photodegradation in the presence of OWCFA exhibited the smallest enhancement (Table 1, Fig. 4). Canonica and Freiburghaus (2001) were able to show that electron rich aromatic compounds reacted with DOMgenerated photo-oxidants of various lifetimes. At high concentrations degradation is dominated by short-lived species such as 3DOM while at low concentrations reaction kinetics reflect the combined effect of short-lived (2 ms) species and long-lived species (>>2 ms, but variable by several orders of magnitude depending upon the species), which may include peroxyl, oxyl, or phenoxyl radicals, and excited states of DOM chromophoric constituents. Our results for SRFA and PLFA are consistent with those reported by others (Canonica and Hoigne´, 1995; Richard and Canonica, 2005; Canonica and Freiburghaus, 2001; Halladja et al., 2007). The increase in
Fig. 4 e 10 mM and 0.1 mM ibuprofen pseudo first order degradation rate constants (kobs hL1) in the presence of Pony Lake Fulvic Acid (PLFA), Suwannee River Fulvic Aci (SRFA), and Old Woman Creek Fulvic Acid (OWCFA).
4453
degradation rates at low ibuprofen concentrations (Fig. 4) reflects the influence of long-lived and short-lived reactive transients generated by the fulvic acids on ibuprofen photodegradation. As observed by us and others (Canonica and Freiburghaus, 2001; Canonica and Laubscher, 2008; Cawley et al., 2009) fulvic acids derived from algal/microbial sources e.g., PLFA appears to be more reactive by these pathways relative to DOM that includes significant amounts of organic matter derived from higher plants. Our results show that the degree to which these long-lived photo-transients are formed is dictated by the composition of the fulvic acid given that the effect is most pronounced for PLFA and is smallest in the presence of OWCFA. To further probe reactive transients at low concentrations, anoxic ibuprofen photolytic experiments were performed in the presence of each fulvic acid. In contrast to the experiments conducted at the higher ibuprofen concentration, degradation rates in the anoxic PLFA and SRFA solutions decreased (Table 1 and Fig. 5). This result is not unexpected given that fulvic acid- generated long-lived transients such as peroxyl and oxyl radicals require dioxygen to form (Richard and Canonica, 2005). Conversely, low initial concentration ibuprofen kinetics remained unchanged in the presence of OWCFA under suboxic conditions (Fig. 5). This fulvic acid was also the least sensitive to changes in the initial concentration of ibuprofen, which suggests that it is either producing longlived radicals that can form in the absence of dioxygen e.g., phenoxyl radicals (Mvula et al., 2001) and/or is an exceptionally good scavenger of these oxidants (Canonica and Laubscher, 2008). Thus, whatever long-lived transients are responsible for the increased degradation of ibuprofen at low concentrations, their effectiveness appears to be highly dependent upon the composition of the fulvic acid. The addition of isopropanol decreased ibuprofen’s degradation rates at low initial concentrations in the presence of all the fulvic acids. Similar observations made at the higher initial ibuprofen concentration (10 mM) were observed for SRFA and OWCFA, but not PLFA where isopropanol had no
Fig. 5 e 0.1 mM ibuprofen pseudo first order degradation rate constants (kobs hL1) in the presence of Pony Lake Fulvic Acid (PLFA), Suwannee River Fulvic Acid (SRFA), and Old Woman Creek Fulvic Acid (OWCFA) under oxic and suboxic systems and with 20 mM isopropanol (Iso).
4454
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 4 9 e4 4 5 8
effect. Thus, it appears that isopropanol reacts with some of the unknown long-lived radicals capable of reacting with ibuprofen at these lower initial concentrations. This study demonstrates that DOM composition plays an important role in the type and amount of reactive transients formed when irradiated. Work conducted by us (Cawley et al., 2009; Guerard et al., 2009) showed that compounds that were transformed by different pathways were highly sensitive to the photosensitizer composition. While ibuprofen reacts with both OH and 3DOM it appears that the former pathway is more important for terrestrially-derived DOM, while the later is the dominant mechanism for DOM derived from microbial (algal and bacterial) precursors. This corroborates recent work by Wenk et al. (2011) who showed that terrestrially-derived DOM is effective at inhibiting the oxidation of organic compounds by the 3DOM pathway, but not by the hydroxyl radical.
3.2.
Photolysis byproduct formation
Multiple byproducts were formed and accumulated in irradiated ibuprofen solutions in the absence and presence of SRFA, OWCFA, and PLFA. In particular one byproduct dominates in the HPLC chromatograms, while another shows up as a redshifted peak in our UVeVis spectrophotometry scans (264 nm absorbance vs. 223 nm for ibuprofen) (Fig. 6a and b). Further analysis of these photo-derivatives using 1H-NMR spectroscopy (Figs S.1 and S.2 in the SI: Note that all figures in SI will have an “S” prefix) and LC-QTOF-MS (Fig. 7a) of unreacted ibuprofen and the photolyzed solution revealed multiple compounds, consistent with our HPLC data. LC-QTOF-MS in positive ion mode identified an unknown compound with a molecular weight of 176.2 that could be two possible derivatives in the photolysis solution: isobutylacetophenone (IBAP), which is an intermediate in the synthesis of ibuprofen (Fig. 7b and c) and/or 1-(4isobutylphenyl) ethanol (4IBPE). Our GCeMS data corroborates these results (Table S.1). The identification of IBAP agrees with the results of Castell et al. (1987), who identified this compound as a byproduct of ibuprofen irradiated in methanol (w8% IBAP is formed at an [Ibuprofen]o ¼ 16 mM). A UVeVis spectrum of an IBAP standard also confirms its existence as its absorbance maximum occurs at the same absorbance (264 nm) of one of the photo-derivatives (Fig. 6 and Fig S.3), and the HPLC retention time for IBAP is also identical to one of our byproduct peaks. Finally parts of our 1H NMR spectrum of the IBAP standard are similar to the photolyzed solution (Figs S.2 and S.4) further confirming this structural assignment (note the quartet at 2.5 ppm and doublet at 7.8 ppm in both spectra). Indeed, isobutylacetophenone has been detected at concentrations of 0.00023 mM in waters in urban areas, citing ibuprofen degradation as its environmental precursor (Zorita et al., 2007). We were able to isolated one of the byproduct peaks (Fig. 6b; peak retention time w7 min) from the HPLC waste line effluent and analyzed it using both 1-D and 2-D correlation spectroscopy (COSY) 1H NMR. Surprisingly, the HPLC isolated peak revealed 1-D spectra that are quite complex and lack the methyl group quartet at 2.5 ppm and doublet at 7.8 ppm (Fig S.5) and is thus not IBAP. The 1-D NMR spectra’s complexity
Fig. 6 e a. UVeVis scan of photolyzed ibuprofen solution at various time points (hours) demonstrating byproduct accumulation at 264 nm; b. HPLC chromatogram (l [ 264 nm) of solution (Co [ 10 mM) photolyzed for 48 h (ibuprofen retention time w5 min; “major” photolysis byproduct retention time w7 min, additional byproduct retention time w8 min).
leads us to speculate that multiple compounds are co-eluting from the HPLC column at 7 min. 2-D 1H NMR (COSY) of the HPLC isolated fraction reveals structural connectivity between the quartet at 4.1 ppm and a doublet at 1.2 ppm (Fig S.6). This quartet at 4.1 ppm, when compared to the ibuprofen scan (Fig S.1), is shifted upfield from 3.7 ppm after photolysis, but is still well resolved. This shift, combined with the observed COSY connectivity, indicates the presence of a more electronegative functional group at the carboxylic acid group of ibuprofen, which has presumably been altered due to photolysis. This was not observed in the 2-D 1H NMR spectrum for IBAP (Fig S.7). Taken together, the NMR data reveal the structure of the major byproduct observed in our HPLC chromatograms as 4IBPE plus other co-eluting compounds and confirms our mass spectrometry results. Our analysis of the byproducts revealed other substances in addition to IBAP and 4IBPE. We observed aromatic doublets in the 1-D 1H NMR data that shifted from the left of the solvent peak (CHCl3 w 7.4 ppm) prior to irradiation to the right after
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 4 9 e4 4 5 8
4455
Fig. 7 e Liquid Chromatography Electrospray Ionization Quadrupole Time of Flight Mass Spectrometer (LC-ESI-QTOF-MS) of 10 mM ibuprofen solution photolyzed 48 h and proposed byproduct structures.
photolysis (Fig S.5). This shift suggests the attachment of a more electronegative moiety to the benzene ring or more specifically, hydroxylation of the aromatic spin system. The NMR spectra taken together with the GCeMS and LC-QTOFMS data suggest a phenol derivative of the parent compound. Photodecarboxylation pathways of NSAIDs in nonaqueous systems have been extensively probed (Castell et al., 1987; Budac and Wan, 1992; Bosca´ et al., 1994, 2001; Monti et al., 1997). Castell et al. (1987) demonstrated that IBAP and 4IBPE are the result of reactions involving dioxygen. These investigators also report equal percent yield of IBAP and 4IBPE, upon irradiation of ibuprofen in methanol in contrast to our systems, which demonstrated that the 4IBPE derivative is the dominant photoproduct. Because Castell et al. (1987) conducted their experiments in methanol, where the solubility of oxygen is significantly higher than in water the yields
of IBAP is presumably higher. Our identification of both these byproducts strongly suggests that photodecarboxylation of ibuprofen can also occur in aqueous systems (Scheme 1). Finally, we have also shown that hydroxylation of the benzene ring takes place (Fig S.5). Fulvic acid composition can also affect photo-induced byproduct formation. At [Ibuprofen]o ¼ 10 mM in the presence of SRFA and OWCFA, HPLC peak area units of the major byproduct do not significantly change when compared to direct photolysis (Fig S.8). In the presence of PLFA, however, the byproduct peak at retention time w7 min (see Fig. 6b) increases relative to direct photolysis and in the presence of the other FA (Fig S.8: filled triangles). We have no explanation as to why PLFA is able to increase the production of this byproduct. Formation of the byproduct peak (7 min retention time) in the presence of fulvic acids at the low initial ibuprofen
4456
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 4 9 e4 4 5 8
Scheme 1 e Proposed reactions: Ibuprofen photodecarboxylation (1) followed by oxygen addition to carbon centered radical and subsequent rearrangement resulting in the formation of isobutylacetophenone (2) and the hydroxylation of carbon centered radical to form 1-(4-isobutylphenyl)ethanol.
concentrations (0.1 mM) does not show the same trend (Fig S.9). Indeed, no correlation is apparent between byproduct peak area generation and the presence of the fulvic acids. Moreover, the linear increase in photo-derivative abundance demonstrated at 10 mM (Fig S.8) is not replicated at the lower concentration (Fig S.9). Our low concentration experiments demonstrate that reaction of ibuprofen with both long- and short-lived reactive transients at this concentration most probably results in the formation of yet-to-be identified derivatives.
4.
Conclusions
The presence of dissolved organic matter significantly enhances the photolytic fate of ibuprofen in sunlit natural waters. Even though our results show that the composition of DOM does not appear to influence the rate of ibuprofen photolysis at high (10 mM) initial concentrations, controlled scavenging and anoxic studies reveal that it undergoes phototransformation by different mechanisms. Further the type of DOM influences the relative abundance of byproducts formed. We were able to identify three important byproducts of ibuprofen photolysis; isobutylacetophenone (a precursor in the synthesis of ibuprofen), 1-(4-isobutylphenyl)ethanol, and a phenol derivative. We observed a significant enhancement in photolytic reaction rates in the presence of DOM at lower initial ibuprofen concentrations. Further DOM composition plays
a more important role at these lower concentrations whereby the microbially derived Pony Lake fulvic acid proved to be the most reactive toward ibuprofen. We attribute these differences in reaction rates to the ability of the DOM phase to form “long-lived” reactive photo-transients of unknown origin when irradiated. Thus, at environmentally relevant concentrations the photolytic transformation of ibuprofen is significantly enhanced by DOM, which may in part explain its extremely low concentrations in sunlit surface waters.
Acknowledgments We thank the members of the Chin research group for helping us isolate the Old Woman Creek fulvic acid and especially Collin Ward for helping LEJ with the photolysis experiments. We also thank Silvio Canonica for his helpful discussions with LEJ as well as comments provided by two anonymous reviewers. This work was partially supported by NOAA/NERR Fellowship awarded to LEJ and by a grant from the National Science Foundation CBET 0504434.
Appendix. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.watres.2011.05.041.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 4 9 e4 4 5 8
references
Aguer, J., Richard, C., Andreux, F., 1997. Comparison of the photoinductive properties of commercial, synthetic and soilextracted humic substances. J. Photochem. Photobiol. A: Chem. 103 (1e2), 163e168. Boreen, A.L., Arnold, W.A., McNeill, K., 2003. Photodegradation of pharmaceuticals in the aquatic environment: a review. Aquat. Sci. 65 (4), 320e341. Boreen, A.L., Arnold, W.A., McNeill, K., 2005. Triplet-sensitized photodegradation of sulfa drugs containing six-membered heterocyclic groups: identification of an SO2 extrusion photoproduct. Environ. Sci. Technol. 39 (10), 3630e3638. Bosca´, F., Marı´n, M.L., Miranda, M.A., 2001. Photoreactivity of the non-steroidal anti-inflammatory 2-arylpropionic acids with photosensitizing side effects. Photochem. Photobiol. 74 (5), 637e655. Bosca´, F., Miranda, M.A., Carganico, G., Mauleo´n, D., 1994. Photochemical and photobiological properties of ketoprofen associated with the benzophenone chromophore. Photochem. Photobiol. 60 (2), 96e101. Brown, A., McKnight, D.M., Chin, Y.P., Roberts, E.C., Uhle, M., 2004. Chemical characterization of dissolved organic matter in Pony Lake, a saline coastal pond in Antarctica. Mar Chem. 89 (1e4), 327e337. Budac, D., Wan, P., 1992. Photodecarboxylation: mechanism and synthetic utility. J. Photochem. Photobiol. A: Chem. 67 (2), 135e166. Buser, H., Poiger, T., Muller, M.D., 1999. Occurrence and environmental behavior of the chiral pharmaceutical drug ibuprofen in surface waters and in wastewater. Environ. Sci. Technol. 33 (15), 2529e2535. Canonica, S., Freiburghaus, M., 2001. Electron-rich phenols for probing the photochemical reactivity of freshwaters. Environ. Sci. Technol. 35 (4), 690e695. Canonica, S., Hoigne´, J., 1995. Enhanced oxidation of methoxy phenols at micromolar concentration photosensitized by dissolved organic matter. Chemosphere 30 (12), 2365e2374. Canonica, S., Jans, U., Stemmler, K., Hoigne´, J., 1995. Transformation kinetics of phenols in water: photosensitization by dissolved natural organic material and aromatic ketones. Environ. Sci. Technol. 29 (7), 1822e1831. Canonica, S., Laubscher, H., 2008. Inhibitory effect of dissolved organic matter on triplet-induced oxidation of aquatic contaminants. Photochem. Photobiol. Sci. 7 (5), 547e551. Castell, J.V., Gomez, M.J., Miranda, M.A., Morera, I.M., 1987. Photolytic degradation of ibuprofen. Toxicity of the isolated photoproducts on fibroblasts and erythrocytes. Photochem. Photobiol. 46 (6), 991e996. Cawley, K.M., Hakala, J.A., Chin, Y.P., 2009. Evaluating the triplet state photoreactivity of dissolved organic matter isolated by chromatography and ultrafiltration using an alkylphenol probe molecule. Limnol. Ocean. Meth. 7, 391e398. Cooper, W.J., Zika, R.G., Petasne, R.G., Fischer, A.M., 1989. Sunlight induced photochemistry of humic substances in natural waters: major reactive species. ACS Symp. Ser. 219, 333e362. Daughton, C.G., Ternes, T.A., 1999. Pharmaceutical and personal care products in the environment: agents of subtle change? Environ. Health Persp 107 (6), 907e938. Dulin, D., Mill, T., 1982. Development and evaluation of sunlight actinometers. Environ. Sci. Technol. 16 (11), 815e820. Edhlund, B.L., Arnold, W.A., McNeill, K., 2006. Aquatic photochemistry of nitrofuran antibiotics. Environ. Sci. Technol. 40 (17), 5422e5427. Fent, K., Weston, A.A., Caminada, D., 2006. Ecotoxicology of human pharmaceuticals. Aquat. Tox. 76 (2), 122e159.
4457
Gerecke, A.C., Canonica, S., Muller, S.R., Scharer, M., Schwarzenbach, R.P., 2001. Quantification of dissolved organic matter (DOM) mediated phototransformation of phenylurea herbicides in lakes. Environ. Sci. Technol. 35 (19), 3915e3923. Guerard, J.J., Miller, P.L., Trouts, T.D., Chin, Y., 2009. The role of fulvic acid composition in the photosensitized degradation of aquatic contaminants. Aquat. Sci. 71 (2), 160e169. Halladja, S., Ter Halle, A., Auger, J.P., Boukamh, A., Richard, C., 2007. Inhibition of humic substances mediated photooxygenation of furfuryl alcohol by 2,4,6-trimethylphenol evidence for reactivity of the phenol with humic triplet excited states. Environ. Sci. Technol. 41 (17), 6066e6073. Hoigne´, J., Faust, B.C., Haag, W.R., Scully, F.E., Zepp, R.G., 1989. Aquatic humic substances and sinks of photochemically produced transients reactants. ACS Symp. Ser. 219, 363e381. Hutt, A.J., Caldwell, J., 1983. The metabolic chiral inversion of 2arylpropionic acids- a novel route with pharmacological consequences. J. Pharm. Pharmacol. 35 (11), 693e704. Kolpin, D.W., Furlong, E.T., Meyer, M.T., Thurman, E.M., Zaugg, S. D., Barber, L.B., Buxton, H.T., 2002. Pharmaceuticals, hormones, and other organic, wastewater contaminants in U. S. streams, 1999e2000: a national reconnaissance. Environ. Sci. Technol. 36 (6), 1202e1211. Leenheer, J.A., 1981. Comprehensive approach to preparative isolation and fractionation of dissolved organic carbon from natural waters and wastewaters. Environ. Sci. Technol. 15 (5), 578e587. Lin, A.Y., Reinhard, M., 2005. Photodegradation of common environmental pharmaceuticals and estrogens in river water. Environ. Toxicol. Chem. 24 (6), 1303e1309. Lindqvist, N., Tuhkanen, T., Kronberg, L., 2005. Occurrence of acidic pharmaceuticals in raw and treated sewages and in receiving waters. Wat. Res. 39 (11), 2219e2228. Miller, P.L., Chin, Y.P., 2002. Photoinduced degradation of carbaryl in a wetland surface water. J. Agric. Food Chem. 50 (23), 6758e6765. Monti, S., Sortino, S., De Guidi, G., Marconi, G., 1997. Photochemistry of 2-(3-benzoylphenol)proponic acid (ketoprofen) part 1: a picosecond and nanosecond time resolved study in aqueous solution. J. Chem. Soc. Faraday Trans. 93 (13), 2269e2275. Mvula, E., Schuchman, M.N., Sonntag, C., 2001. Reactions of phenol-OH-adduct radicals. Phenoxyl radical formation by water elimination vs. oxidation by dioxygen. J. Chem. Soc. Perkin Trans. 2, 264e268. Nakada, N., Tanishima, T., Shinohara, H., Kiri, K., Takada, H., 2006. Pharmaceutical chemicals and endocrine disrupters in municipal wastewater in Tokyo and their removal during activated sludge treatment. Wat. Res. 40 (17), 3297e3303. Nakada, N., Shinohara, H., Murata, A., Kiri, K., Managaki, S., Sato, N., Takada, H., 2007. Removal of selected pharmaceuticals and personal care products (PPCPs) and endocrine-disrupting chemicals (EDCs) during sand filtration and ozonation at a municipal sewage treatment plant. Wat. Res. 41 (19), 4373e4382. Nikolaou, A., Meric, S., Fatta, D., 2007. Occurrence patterns of pharmaceuticals in water and wastewater environments. Anal. Bioanal. Chem. 387 (4), 1225e1234. Packer, J.L., Werner, J.J., Latch, D.E., McNeill, K., Arnold, W.A., 2003. Photochemical fate of pharmaceuticals in the environment: naproxen, diclofenac, clofibric acid, and ibuprofen. Aquat. Sci. 65 (4), 342e351. Pe´rez-Estrada, L.A., Malato, S., Gernjak, W., Agu¨era, A., Thurman, E.M., Ferna´ndez-Alba, A.R., 2005. Photo-Fenton degradation of diclofenac: identification of main intermediates and degradation pathway. Environ. Sci. Technol. 39 (21), 8300e8306. Quintana, J.B., Weiss, S., Reemtsma, T., 2005. Pathways and metabolites of microbial degradation of selected acidic
4458
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 4 9 e4 4 5 8
pharmaceutical and their occurrence in municipal wastewater treated by a membrane bioreactor. Wat. Res. 39 (12), 2654e2664. Richard, C., Canonica, S., 2005. Aquatic phototransformation of organic contaminants induced by coloured dissolved natural organic matter. In: The Handbook of Environmental Chemistry, vol. 2, Part M, pp. 299e323. Scheytt, T., Mersmann, P., Lindsta¨dt, R., Heberer, T., 2005. 1-octanol/water partition coefficients of 5 pharmaceuticals from human medical care: carbamazepine, clofibric acid, diclofenac, ibuprofen, and propyphenazone. Water Air Soil Poll. 165 (1e4), 3e11. Ternes, T.A., 1998. Occurrence of drugs in German sewage treatment plants and rivers. Wat. Res. 32 (11), 3245e3260. Thurman, E.M., Malcolm, R.L., 1981. Preparative isolation of aquatic humic substances. Environ. Sci. Technol. 15 (4), 463e466. Vaughan, P.P., Blough, N.V., 1998. Photochemical formation of hydroxyl radical by constituents of natural waters. Environ. Sci. Technol. 32 (19), 2947e2953. Wenk, J., von Gunten, U., Canonica, S., 2011. Effect of dissolved organic matter on the transformation of contaminants induced by excited triplet states and the hydroxyl radical. Environ. Sci. Technol. 45 (4), 1334e1340. White, E.M., Vaughan, P.P., Zepp, R.G., 2003. Role of the photoFenton reaction in the production of hydroxyl radicals and photobleaching of colored dissolved organic matter in a coastal river of the southeastern United States. Aquat. Sci. 65 (4), 402e414.
Winker, M., Faika, D., Gulyas, H., Otterpohl, R., 2008. A comparison of pharmaceutical concentrations in raw municipal wastewater and yellow water. Sci. Total Env. 399 (1e3), 96e104. Xu, J., Wu, L., Chang, A.C., 2009. Degradation and adsorption of selected pharmaceuticals and personal care products (PPCPs) in agricultural soils. Chemosphere 77 (10), 1299e1305. Yamamoto, H., Nakamura, Y., Moriguchi, S., Nakamura, Y., Honda, Y., Tamura, I., Hirata, Y., Hayashi, A., Sekizawa, J., 2009. Persistence and partitioning of eight selected pharmaceuticals in the aquatic environment: laboratory photolysis, biodegradation, and sorption experiments. Wat. Res. 43 (2), 351e362. Yu, J.T., Bouwer, E.J., Coelhan, M., 2006. Occurrence and biodegradability studies of selected pharmaceuticals and personal care products in sewage effluent. Agric. Water Manag. 86 (1e2), 72e80. Zhou, X., Mopper, K., 1990. Determination of photochemically produced hydroxyl radicals in seawater and freshwater. Mar. Chem. 30, 71e88. Zorita, S., Barri, T., Mathiasson, L., 2007. A novel hollow-fiber microporous membrane liquid-liquid extraction for determination of free 4-isobutylacetophenone concentration at ultra trace level in environmental aqueous samples. J. Chromatogr. 1157 (1e2), 30e37. Zwiener, C., Frimmel, F., 2003. Short-term tests with a pilot sewage plant and biofilm reactors for the biological degradation of the pharmaceutical compounds clofibric acid, ibuprofen, and diclofenac. Sci. Total Environ. 309 (1e3), 201e211.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 5 9 e4 4 6 9
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Intrinsic biodegradation potential of aromatic hydrocarbons in an alluvial aquifer e Potentials and limits of signature metabolite analysis and two stable isotope-based techniques Barbara Morasch a,*, Daniel Hunkeler a, Jakob Zopfi b, Brice Temime c, Patrick Ho¨hener c a
Center for Hydrogeology, University of Neuchaˆtel, Rue Emile Argand 11, 2009 Neuchaˆtel, Switzerland Laboratory of Microbiology, University of Neuchaˆtel, Rue Emile Argand 11, 2009 Neuchaˆtel, Switzerland c Laboratoire Chimie Provence, Universite´ de Provence, CNRS, Place Victor Hugo, F-13331 Marseille Cedex 3, France b
article info
abstract
Article history:
Three independent techniques were used to assess the biodegradation of monoaromatic
Received 21 February 2011
hydrocarbons and low-molecular weight polyaromatic hydrocarbons in the alluvial aquifer
Received in revised form
at the site of a former cokery (Fle´malle, Belgium).
12 May 2011
Firstly, a stable carbon isotope-based field method allowed quantifying biodegradation
Accepted 28 May 2011
of monoaromatic compounds in situ and confirmed the degradation of naphthalene. No
Available online 14 June 2011
evidence could be deduced from stable isotope shifts for the intrinsic biodegradation of larger molecules such as methylnaphthalenes or acenaphthene. Secondly, using signature
Keywords:
metabolite analysis, various intermediates of the anaerobic degradation of (poly-) aromatic
Groundwater contamination
and heterocyclic compounds were identified. The discovery of a novel metabolite of ace-
Natural attenuation
naphthene in groundwater samples permitted deeper insights into the anaerobic biode-
(Poly-) aromatic hydrocarbons
gradation of almost persistent environmental contaminants. A third method, microcosm
Signature metabolites
incubations with
Stable isotopes
techniques one and two by providing quantitative information on contaminant biodegra-
Biodegradation rates
dation independent of molecule size and sorption properties. Thanks to stable isotope
13
C-labeled compounds under in situ-like conditions, complemented
labels, the sensitivity of this method was much higher compared to classical microcosm studies. The
13
constants for
13
C-microcosm approach allowed the determination of first-order rate C-labeled benzene, naphthalene, or acenaphthene even in cases when
degradation activities were only small. The plausibility of the third method was checked by comparing 13C-microcosm-derived rates to field-derived rates of the first approach. Further advantage of the use of
13
C-labels in microcosms is that novel metabolites can be linked
more easily to specific mother compounds even in complex systems. This was achieved using alluvial sediments where
13
C-acenaphthyl methylsuccinate was identified as trans-
formation product of the anaerobic degradation of acenaphthene. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
Biodegradation is claimed to be the key process leading to decontamination of many abandoned industrial sites impaired
with coal- and tar oil-derived compounds. In practice, it is often difficult to judge whether degradation is taking place or not. Under environmental conditions, particularly in the absence of oxygen, mono- and polyaromatic contaminants are
* Corresponding author. Present address: Environmental Mineralogy and Chemistry, Center for Applied Geoscience (ZAG), University of Tuebingen, Sigwartstrasse 10, 72076 Tuebingen, Germany. Tel.: þ49 7071 2973135. E-mail address:
[email protected] (B. Morasch). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.05.040
4460
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 5 9 e4 4 6 9
biodegraded at low rates or are supposedly persisting (Zamfirescu and Grathwohl, 2001; Foght, 2008). Consequently, solid information is needed on the in situ biodegradation of coal- and tar oil-derived pollutants. Due to low solubility in water and tendencies to sorb, turnover of (poly-) aromatic compounds (PAHs) may often be limited by mass transfer and not by microbial activity (Bosma et al., 1997). It is controversial whether biodegradation of aromatic hydrocarbons can be distinguished at all from partitioning when only a small decrease in concentration is measured (Foght, 2008). Even though it is considered as an issue of increasing importance, no universal technique exists to measure biodegradation at contaminated sites. In this study, three independent approaches were used in combination to assess the intrinsic biodegradation of aromatic and heterocyclic environmental contaminants. Particular attention was paid to useful approaches for the quantification of in situ biodegradation of compounds that show small degradation activities or have the tendency to sorb. The potentials and limits of the three techniques are compared and discussed.
1.1. Compound-specific stable isotope analysis of individual pollutants in groundwater samples of contaminated sites This technique has proven appropriate for the direct assessment of biodegradation at those field sites where a contaminant plume has established. Along the centerline of a contaminant plume, in situ biodegradation is resolved over distance in the stable isotope shifts of individual groundwater pollutants. Stable isotope fractionation during degradation of monoaromatic hydrocarbons, naphthalenes, alkanes, chlorinated solvents, and gasoline additives was studied under various redox conditions (Hunkeler and Morasch, 2010). Larger molecules of high environmental relevance (e.g. three and more ring PAHs) have been investigated much less extensively and no enrichment of heavier isotopes above the analytical error was reported (Mazeas et al., 2002). Furthermore, compound-specific stable isotope analysis (CSIA) was used in field studies to calculate the percentage of in situ biodegradation and first-order rate constants (Richnow et al., 2003b; Batlle-Aguilar et al., 2009).
1.2. Screening for signature metabolites indicating contaminant biodegradation in groundwater samples Signature metabolites are highly specific reaction intermediates produced only during biodegradation of target contaminants. They need to be excluded as contaminants themselves in order to be indicative of on-site remediation (Phelps et al., 2002). In previous studies, various signature metabolites were either identified in batch culture experiments or were extracted from groundwater samples directly. Recognized molecules include intermediates of the anaerobic degradation of alkylated aromatic compounds, e.g. benzylsuccinates that are formed in addition reactions with fumarate (Elshahed et al., 2001; Beller, 2002). Apart from methylated benzenes and naphthalenes, methylated heterocyclic compounds are anaerobically degraded also via initial fumarate addition (Annweiler et al., 2000; Safinofski et al., 2006).
1.3. Incubations of field material in the lab under in situ-like conditions adding isotopically labeled contaminants The intrinsic biodegradation potential of individual compounds can be assessed by incubating field material in the laboratory under in situ-like conditions. Even in complex systems, biodegradation can be specifically tracked when isotopically labeled compounds are supplied as markers. Isotope labels circumvent difficulties related to classical microcosm experiments with field material that potentially contains organic background contaminants. Briefly, the method is based on the recovery of isotope labels in CO2 produced from mineralization of isotope-labeled substrates. Complete oxidation of naphthalene under anoxic conditions e.g. was shown by adding 14C-labeled naphthalene to aquifer material and to marine harbor sediments (Chapelle et al., 1996; Coates et al., 1996; Langenhoff et al., 1996). Recently, naphthalene was also used in 13C-labeled form to confirm biodegradation in sedimentegroundwater microcosms (Morasch et al., 2007). All techniques mentioned above can provide evidence for natural attenuation of contaminants according to the three lines of evidence established by the National Research Council (2000). By definition, conditions are met if (I) the loss of contaminant is documented at the field scale, (II) the presence of degrading microorganisms is confirmed by means of incubation experiments using field material, and (III) the direct evidence for microbial activity in situ can be provided. Techniques one (CSIA) and two (signature metabolite analysis) meet criterion (III). They have already been applied in combination, e.g. at a former gasworks site with a longstanding contamination with mono- and polyaromatic hydrocarbons and in a controlled release experiment of benzene, toluene, and o-xylene at a US Air Force base (Griebler et al., 2004; Beller et al., 2008). However, the combination of CSIA and signature metabolite analysis has its limitations. CSIA has the potential to quantify the biodegradation of smaller contaminants, while its sensitivity decreases with increasing molecular size of the compounds of interest. Since frequently the biodegradation rates also get lower with molecular size the method is not applicable for larger contaminants. Signature metabolite analysis can e in a qualitative way e provide information on the biodegradation of these larger compounds. Hence, degradation kinetics of larger contaminant molecules remains unrevealed. In this study we use a third technique (microcosms with 13C-labeled compounds) to overcome these limitations and to provide additional quantitative information on in situ biodegradation independent of molecular size. Results of all three techniques are compared and their potentials and limits evaluated.
2.
Materials and methods
2.1.
Field site
The study site was a former coke and gas factory that was dismantled in 1984. The property of 400 m 250 m was
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 5 9 e4 4 6 9
situated in an industrial environment at 25 m distance from the Meuse River upstream of the city of Lie`ge, Belgium (Fig. 1; geographical location: þ50 360 19.7600 , þ5 290 13.5200 , www. google.com/maps). The topmost 4 m below the surface consisted of backfill material followed by silt, sand, and clay deposits of about 2 m thickness, and 8 m of fine gravel above the carboniferous shale bedrock. The water table was located in the alluvial gravel layer at 5.5e7 m depth. Groundwater flow was in eastern direction. Previous field characterization evidenced a hydrological gradient of 0.3% and a saturated hydraulic conductivity that varied between 105 and 103 m/s (Batlle-Aguilar et al., 2009). Severe contamination with heavy metals, cyanides, mineral oils, as well as with mono- and polyaromatic hydrocarbons reached down to 11 m depth. The contamination was of unknown horizontal extent according to various measurements by the Walloon Environmental Protection Agency. In the north-western part of the site (Fig. 1), Eh values of the groundwater were at 300 mV and nitrate was almost depleted. Strictly reducing conditions prevailed up to 100 m in southeastern direction toward the Meuse River before the Eh rose to þ100 mV and nitrate concentrations of up to 15 mg L1 were observed. Sulfate concentrations in groundwater were between 500 and 2100 mg/L, hence sulfate was assumed to be the major electron acceptor in the degradation of organic contaminants after O2 had been depleted (For a redox zonation map, see Batlle-Aguilar et al., 2009).
2.2.
Sampling
Sampling campaigns took place in March 2005 and July 2006 where 17 and 23 groundwater wells were sampled, respectively (Table S1). Water was pumped with submersible pumps at a rate of 1e5 L/min. Groundwater table, temperature,
4461
conductivity, pH, and dissolved O2 were recorded using specific field probes (WTW; Weilheim, Germany). Subsequently, water was sampled to determine alkalinity, Mn2þ, Fe2þ, HS-, and methane. Additional samples were taken for the quantification of aromatic contaminants, CSIA, and the identification of signature metabolites. Water samples were conserved immediately on site by adding 0.1% (vol/vol) of NaOH (5 M). Upgradient from the contamination, groundwater wells E6p and F4 served as references (Fig. 1). In 2005, freshly drilled core material from location U13 adjacent to the source zone was sampled every 0.5 m for microcosm experiments.
2.3.
Microcosm set-up and sampling
Aquifer material was collected during drillings, filled into brown-glass bottles of 120 mL with Teflon-sealed caps, cooled immediately, and stored at 5 C until microcosms were prepared. Sediments of core U13 were pooled to four different depth layers (5e6, 7e8, 9, and 12e13 m). Working under an atmosphere of N2 in a glove bag (SigmaeAldrich), 53 microcosms were set up in culture bottles of 50 mL volume. Every microcosm consisted of 25 g of sediment and 15 mL of O2-free groundwater that had been sterile-filtered and diluted fivefold with autoclaved nanopure water, to achieve nitrate concentrations that better resembled groundwater concentrations of the strictly reducing zone. The diluted groundwater contained 6 mg/L of nitrate and 140 mg/L of sulfate. Benzene, naphthalene, or acenaphthene labeled with 13C at six positions (99% purity, Cambridge Isotope Laboratories) was dissolved in the anoxic groundwater that was added to the culture bottles (Morasch et al., 2007). The headspace either was air for sediments from the unsaturated zone or N2 for microcosms with sediments from the saturated zone. Microcosms were closed with non-absorptive Viton rubber stoppers. Controls were
Fig. 1 e Site map of the former cokery of Fle´malle, Belgium, located in the direct vicinity of the river Meuse. Locations of groundwater (GW) samples are shown as circles, the sediment sampling location U13 is depicted as square. The arrow roughly describes the groundwater flow direction. The map has been modified from Batlle-Aguilar et al. (2009).
4462
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 5 9 e4 4 6 9
prepared without addition of substrate or microcosms were autoclaved twice at days 1 and 4 of the experiment. Incubations took place in the dark at 16 C.
2.4.
Analysis of organic contaminants
Concentrations of benzene, toluene, ethylbenzene, and mxylene (BTEX), and of low-molecular weight PAHs were analyzed using a gas chromatograph (Varian 3800) with a CP8410 autoinjector for solid phase microextraction (SPME). The aromatic compounds were extracted from the headspace of 2 mL vials filled with 0.5 mL of groundwater and 0.3 g of NaCl using polydimethylsiloxane fibers (100 mm film thickness, Supelco). Extraction and analysis were performed as described previously (Morasch et al., 2007).
2.5.
Stable carbon isotope analysis
Groundwater for 13C/12C isotope analysis was sampled in volumes of 1e2 L. CSIA of aromatic hydrocarbons was performed using a Trace GC coupled to a Delta Plus XP isotope ratio mass spectrometer (IRMS) via a GC combustion III interface (Thermo Finnigan). Complete protocols of sample preparation and analysis are provided in Supplementary material. For stable carbon isotope analysis of CO2, 2 mL headspace samples were taken from microcosms through the Viton stoppers with a gas tight syringe working under an atmosphere of N2. Sampling intervals were between three and seven days at the beginning of the incubation and three-monthly toward the end of the experiment (Table S2). Concentration measurements of CO2 and stable carbon isotope measurements were performed connecting a headspace autosampler (Tekmar Dohrmann 7000) to the GC-IRMS using a previously described protocol (Morasch et al., 2007). CO2 concentrations were quantified based on five-point calibration curves with an average correlation of r ¼ 0.98. Stable isotope ratios were determined relative to an external CO2 reference gas and reported as d [&] deviation to the VPDB standard d13 C ½& ¼
Rsample 1 1000 Rstd
(1)
where Rsample and Rstd are the carbon stable isotope ratios of the sample and of the standard, respectively.
2.6.
Metabolite extraction, analysis, and identification
Putative degradation intermediates of aromatic or heterocyclic contaminants were extracted with dichloromethane from 1 L groundwater samples acidified to pH 1e2 with HCl (37%). The extraction was repeated and dichloromethane fractions were pooled. The complete protocol is provided in Supplementary material. After the completion of degradation experiments, potential metabolites were extracted from the water phase and the sediment of microcosms. For metabolite extraction from the sediment fraction, 20 mL of acetone was added into the microcosm bottles and placed into an ultrasonic bath for 10 min. The acetone phase was decanted from the microcosm bottle and the procedure was repeated with 20 mL of
dichloromethane. Then, extracts from the water phase and the sediment fraction were combined. The complete extraction protocol may be found in Supplementary material. Analysis by GCeMS was performed using a Trace GC coupled to a Polaris Q Ion Trap Mass spectrometer (Finnigan). Identity of substances was confirmed by co-elution with reference compounds and by comparison of mass spectra with published data.
2.7.
Calculations
In situ biodegradation of BTEX was calculated using the approach of Richnow et al. (2003b): B½% ¼
ct 100 1 c0
(2)
where B is the percentage of biodegradation of the substrate; c0 and ct are substrate concentrations at the source and at a downgradient monitoring point. Independently from concentration measurements, ct was obtained from c0 in combination with the stable isotope ratios R0 at the source and Rt at a downgradient monitoring point, and substrate-specific stable isotope enrichment factors (3) derived from laboratory studies:
Rt ct ¼ c0 R0
1000 3
(3)
First-order biodegradation rates l were calculated from field isotope data according to l ¼ Dd13 C=ð3 tc Þ
(4)
13
where Dd C is the shift in the carbon isotope ratio between the source and a downgradient monitoring point, and tc is the travel time of the contaminant (Hunkeler et al., 2002; Blum et al., 2009). Travel times of the contaminants were estimated based on an intermediate groundwater flow velocity of 0.29 m d1 (Batlle-Aguilar et al., 2009) and compound-specific retardation factors 1f FR ¼ 1 þ r Kd f
(5)
that were obtained assuming a sediment density of r ¼ 2500 kg/m3 and a mobile porosity of f ¼ 0.2. Solidewater distribution coefficients Kd ¼ fOC KOC were calculated assuming a fraction of organic carbon in the sediments of fOC ¼ 0.001. Organic carbon-normalized distribution coefficients KOC were predicted from compound class-specific log KOC log KOW relationships (Schwarzenbach et al., 2003) taking octanolewater coefficients KOW from the Physical Properties Database (SRC Inc., 2009). Half-life times t1/2 and half-concentration distances x1/2 were defined as follows: t1=2 ¼
lnð2Þ l
(6)
x1=2 ¼
v t1=2 FR
(7)
where v is the intermediate groundwater flow velocity in the aquifer.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 5 9 e4 4 6 9
2.8.
3.2. Qualitative assessment of in situ biodegradation using CSIA
Inorganic carbon mass balance
To assess the intrinsic biodegradation potential of the 13Clabeled contaminants, an inorganic carbon mass balance was applied to sedimentegroundwater microcosms (a detailed description is provided in the Supplementary material). The initial amount of inorganic carbon M0 was approximated from the sum of CO2 (g) as determined by GC-IRMS at the beginning of the microcosm experiment, H2CO3 (aq), and the concentration of HCO3 (aq) of 2 mM determined by alkalinity titration for the groundwater added to microcosm bottles at the beginning of the experiment (see Supplementary material for determination of alkalinity). CO2 dissolved in the water phase was calculated from CO2 concentrations in the gas phase using Henry’s law according to ½H2 CO3 ðaqÞ ¼ ½CO2 ðgÞ Kh with a Henry coefficient of Kh ¼ 0.77 (SRC Inc., 2009). The amount of inorganic carbon produced upon the biodegradation of 13C substrates at time t, is depicted MS. Compared to MS, the isotope signature of 13 C-CO2 originating from the degradation of undefined background carbon sources (MBG) was considered equal to the isotope signature M0 at the beginning of the experiment. The biodegradation rate was calculated using a first-order type equation: l$t MSð0Þ MS ¼ MSð0Þ e FR
(8)
with MS(0) being the amount of 13C [mmol] added to the bottle at t ¼ 0 d, the first-order rate constant l, and the reciprocal value of the retardation factor 1/FR designating the fraction of contaminant present in the water phase of the microcosms (Schwarzenbach et al., 2003). For hydrophobic compounds such as BTEX and PAH, the correction by FR is necessary to account for sorption to the sediment matrix since only the dissolved compound fraction is readily available for biodegradation. 13C-CO2-based biodegradation rates were validated previously by comparing the changes in the 13C-aniline concentrations in the water phase to the concomitant evolution of 13C-CO2 in the gas phase (Morasch et al., 2011).
3.
4463
Results and discussion
3.1. Abundance of aromatic hydrocarbons in groundwater Based on a systematic assessment of contaminant concentrations in the aquifer of the Fle´malle site, one major source zone was identified in the north-western part around piezometers D2bis, D1p, and D3p (Fig. 1). In that zone, groundwater concentrations of benzene, toluene, and m-xylene were in the mg/L range; ethylbenzene was in the mg/L range (Table S3). Also the highest concentrations of naphthalene (up to 25 mg/L) were detected in D2bis, D1p, and D3p and decreased along the groundwater flow path toward the Meuse River in eastern direction (Fig. 2a). Concentrations of low-molecular weight PAHs were elevated in piezometer 14 located 39 m downgradient of well D2bis. At well 15, 133 m east of the source zone, there was an additional point of increased contamination with the three-ring compounds acenaphthene and fluorene (Table S3).
13
C/12C isotope signatures of residual groundwater contaminants were measured using CSIA in 2005 and 2006. Improved protocols (Purge&Trap for BTEX, SPME for PAHs) resulted in a higher number of isotope signatures that could be determined in 2006 (Table S4). Most negative d-values of BTEX were found in the major source zone (D2bis, D3p, D1p). Then, 13C became progressively enriched in the residual BTEX along the groundwater flow path in the methanogenic to sulfate-reducing aquifer section between the source and well U13. Less reducing groundwater sampled further toward the river showed lower 13C/12C ratios in the residual contaminants. Representative for all aromatic hydrocarbons detected at the Fle´malle site, stable carbon isotope ratios at the Fle´malle site are displayed for benzene, naphthalene, and acenaphthene (Fig. 2bed). According to CSIA, no naphthalene degradation took place within the first 90 m of the contaminant plume. However, 13 C-enriched naphthalene (with a signature of 19.7&) in the groundwater of well 15, suggested biodegradation beyond the strictly reducing zone in direction toward the river. No conclusive stable isotope shift was obtained for acenaphthene. For comparison, at a former gas manufacturing plant in Southern Germany d13C shifts of 3.3 and 3.6& provided evidence for intrinsic anaerobic biodegradation of benzene and naphthalene (Griebler et al., 2004). At that site, acenaphthene formed long contaminant plumes with almost constant concentrations and insignificant stable carbon isotope shifts over a distance of more than 135 m (Zamfirescu and Grathwohl, 2001; Steinbach et al., 2004). This lack of evidence for in situ biodegradation of acenaphthene is in agreement with our findings (Fig. 2d).
3.3.
Quantitative assessment of in situ biodegradation
The BTEX plume originating from the source zone around piezometer D2b was delineated using the piezometric lines (as established by Batlle-Aguilar et al., 2009) in combination with 13 C ratios of the residual BTEX of the campaign 2006 (Fig. S1). Due to the continuous increase in the d13C values of residual benzene, toluene, ethylbenzene, and m-xylene with distance in the strongly reducing part of the aquifer, piezometers D2bis, D1p, U4, 11, and U13 were attributed to the same contaminant plume. Significantly more negative d-values in wells U6 and 12 (partially even below the isotope signatures at the source D2bis) suggested local secondary sources of contaminants with BTEX concentrations that were orders of magnitude below the concentration in the major source zone (well D1p). Wells U6 and 12 were consequently excluded from the quantitative assessment of in situ biodegradation. Mixing of contaminants originating from secondary sources with the main contaminant plume that is more enriched in 13C, would lead to slight underestimations of the in situ degradation. In the following moderately reducing to oxic aquifer section 90e160 m from the major source zone, no further 13C-enrichment in residual BTEX was observed. Occasionally, the residual monoaromatic contaminants showed more negative d13C ratios (Fig. S1). Well 15 was not considered for quantitative evaluation, since it was surrounded by a less reducing to oxic zone (Batlle-Aguilar et al., 2009).
4464
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 5 9 e4 4 6 9
Fig. 2 e aed: Concentrations of benzene, naphthalene, and acenaphthene and corresponding stable carbon isotope ratios determined in groundwater of the Fle´malle site in 2006. Benzene (outer circle), naphthalene (middle circle), and acenaphthene (inner circle) were chosen as representatives of mono-, di-, and tri-ring aromatic compounds. b was modified from Batlle-Aguilar et al. (2009).
100 Biodegradation [%]
The percentage of biodegradation was based on the 13Cenrichment in residual groundwater contaminants with distance from the source (Eqs. (2) and (3)). The results confirmed that a considerable part of benzene, toluene, and m-xylene was already biodegraded within the strictly reducing zone (Fig. 3). According to CSIA, the percentage of biodegradation was highest for m-xylene (>90% relative to D2bis); 80% of the initial benzene and toluene, and 40% of the initial ethylbenzene were removed by biodegradation. Due to decreasing d-values in the residual BTEX at 90e160 m distance from the source zone, it was not possible to quantify the biodegradation in this aquifer section using the stable isotopebased approach (Fig. 3). This is comparable to another anoxic aquifer, where 99% of the decrease in toluene and o-xylene could be attributed to intrinsic biodegradation based on the observed stable isotope shifts (Richnow et al., 2003a). Apart from the percentage of biodegradation, the kinetics of the anaerobic in situ biodegradation was determined. Firstorder rate constants (l) were calculated based on the continuous 13C-enrichment downgradient the source zone (Eq. (4)). For this, distances were converted into travel times using an intermediate groundwater flow velocity of 0.26 m/d. For the single stretches from D2bis to the individual piezometers
80
U13
60
8
7
U6
40
11 D1p
20 0
Reducing to oxic
Strongly reducing
0 25 (D2bis)
P5
U4 15
50
75
100
125
150
175
Distance from source [m]
Fig. 3 e Percentage of biodegradation of benzene (A), toluene (C), ethylbenzene (:), and m-xylene (-) along the groundwater flow path. Filled symbols indicate strictly reducingd, open symbols mildly reducing to oxic groundwater samples. Large symbols represent samples, which were incorporated in the first-order rate calculations (see text for details).
4465
240
178 3.9 10 R2 ¼ 0.96
3
2.9 103
187 3.7 103
770 0.9 10 R2 ¼ 0.80
3
1.4 103
491
4.6 727 1.0 103
161 187
4.3 10 R2 ¼ 0.99
3
3.7 10 R2 ¼ 0.74
3
175 4.0 103 178 3.9 103
165 4.2 103 2.6
1.3
337 2.1 103 0.0 1.1 371 1.9 103 0.0 1.1 185 3.7 103 0.0 1.0
152 389 117 208 4.6 103 1.8 103 5.9 103 3.3 103
Linear regression
t1/2 [d] l [/d] l [/d]
t1/2 [d]
Dd [&]
l [/d]
t1/2 [d]
Dd [&]
l [/d]
t1/2 [d]
Dd [&]
m-Xylene Ethylbenzene
Dd [&]
0.0 1.3 0.9 3.9 3.8 Average
Acenaphthyl methylsuccinate was not commercially available as reference compound; therefore, the potential first
D2bis D1p U4 11 U13
3.5. Tentative identification of acenaphthene methylsuccinate
Toluene
Screening campaigns for metabolites in groundwater samples were performed in 2005 and 2006. In the first year, a series of carboxylic acid and alcohol derivatives of aromatic compounds were detected in groundwater samples of wells D2bis and C3bis, however, no signature metabolites in the strict sense could be identified. Groundwater of more than 94 m distance from the source mostly was free of contamination with aromatic hydrocarbons (wells 1, 2, 7, 8, P4, and P6), and no potential metabolites of BTEX or PAHs were detected. In order to improve the sensitivity, larger volumes of groundwater were extracted in the second campaign. Methylsuccinyl-adducts of five different aromatic hydrocarbons and carboxylic acids of twelve different aromatic and heterocyclic compounds were identified in form of their methyl esters in groundwater samples of ten different wells (Table S5). The fumarate adduct of toluene was identified by comparison with an authentic benzylsuccinate standard and had a GC-retention time of 28.0 min. Methyl esters of the fumarate adducts of xylene, methyl- and dimethylnaphthalene, were tentatively identified based on published reference spectra without distinguishing between the different isomers (Table S5). Fumarate adducts of BTEX were only detected in samples of well D1p (source zone), whereas methylsuccinates of naphthalene and methylnaphthalene were extracted from several wells in the strictly reduced zone of the aquifer. Putative acenaphthyl methylsuccinate was present in groundwater from D2bis, D1p, and well 14 (for the identification of this compound, see paragraph below). Carboxylic acids that were potentially related to the degradation of BTEX, biphenyl, or PAHs occurred in groundwater of the source zone and close by, as well as in piezometers 7 and 15, both in the reducing section of the aquifer in 117 and 133 m distance from the source, respectively. Carboxylic acids of the heterocyclic compounds benzothiophene, benzofuran, indane, and indene were equally present in the source zone and in the strictly reducing groundwater sampled further downgradient (Table S5).
Benzene
3.4. Screening for signature metabolites of contaminant biodegradation in groundwater
Well
downgradient of the source, anaerobic in situ biodegradation rate constants were in the range of 1.4 103/d for ethylbenzene and 4.0 103/d for toluene (Table 1). Assessing firstorder biodegradation between D2bis and U13 via linear regression resulted in comparable rate constants. Recently, the stable isotope-based model was applied to determine first-order rate constants for o-xylene and naphthalene degradation at a former wood preservation plant (Blum et al., 2009). Under strictly reducing conditions, biodegradation of o-xylene proceeded at a rate of 2 103/d which was in the same range as the rate constants for monoaromatic compounds at the Fle´malle site. At the other site, it was possible to determine first-order rate constants of naphthalene of 4 103/d and 3 103/d for the anoxic and oxic sections of the plume, respectively (Blum et al., 2009).
Table 1 e First-order rate constants l and corresponding half-life times t1/2 of anaerobic in situ biodegradation of monoaromatic groundwater contaminants at the site of the former cokery of Fle´malle. Values were calculated between isotope signatures in groundwater of D2bis (source) and individual piezometers located along the groundwater flow path in the strongly reducing part of the aquifer. An intermediate groundwater flow velocity of 0.26 m/d was assumed (Batlle-Aguilar et al., 2009). Calculations were based on average isotope enrichment factors taken from the literature (Hunkeler and Morasch, 2010). For comparison, rate constants from linear regression analysis retrieved from the stable isotope-based first-order biodegradation model are shown.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 5 9 e4 4 6 9
4466
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 5 9 e4 4 6 9
Fig. 4 e a and b: Mass spectra of dimethyl esters of putative acenaphthyl methylsuccinate (A), and naphthyl methylsuccinate (B) tentatively identified by comparison with a reference spectrum from Annweiler et al. (2000).
fragmentation pattern of acenaphthyl methylsuccinate was compared to commercially available acenaphthene succinate. An absolute difference of mass fragments in 14 m/z was observed between the two respective dimethyl esters (Fig. 4a, Fig. S2a). Their GC-retention times were 41.7 and 40.5 min. Thirdly, several microcosms that had been incubated with 13 C-acenaphthene were solvent-extracted at the end of the incubation period in search of the metabolite additionally bearing six 13C-atoms. Operating the GCeMS in single ion mode for higher sensitivity, putative 13C6-acenaphthyl methylsuccinyl dimethyl ester was detected at 42.2 min (Fig. S2b). Although, several studies reported on the anaerobic degradation of acenaphthene in microcosm experiments, the degradation pathway has remained unknown (Mihelcic and Luthy, 1988; Rothermich et al., 2002; Chang et al., 2003; Yuan and Chang, 2007). The metabolite that was tentatively identified in this study suggested the introduction of a methyl group in the acenaphthene skeleton and a subsequent fumarate addition in analogy to the anaerobic degradation of naphthalene (Safinofski and Meckenstock, 2006; Foght, 2008).
115 133
165 127 152
141
178
193
207
195 221
153
20
213 239
40
286
60
181
80
167
226 252
100
312
Relative abundance [%]
metabolite of methyl acenaphthene was tentatively identified based on three independent lines of evidence. Firstly, mass spectra of putative acenaphthyl methylsuccinate were compared to those of naphthyl methylsuccinate; both converted to their dimethyl esters (Fig. 4a and b). Due to the additional ethylene bridge of the acenaphthene skeleton, both molecules had an absolute difference of 26 m/z in their respective mass peaks (312 and 286 m/z). This shift of 26 m/z was reflected in all major peaks of both MS-fractionation patterns. Comparing the relative abundances of the major fragments of naphthyl methylsuccinate- and putative acenaphthyl methylsuccinate dimethyl ester, a correlation coefficient of R ¼ 0.879 was obtained (Fig. 5). Secondly, the MS-
0
Eliminated fragment [m/z] Fig. 5 e Comparison of mass fragmentation patterns of naphthyl methylsuccinate (gray bars) and putative acenaphthyl methylsuccinate (black bars). On the x-axis displayed is the size of the eliminated fragment (m/z), in case of deviating fragment size, the first value refers to naphthyl methylsuccinate and the second to acenaphthyl methylsuccinate. Values on top of the bars are major mass fragments of the original GCeMS spectra.
3.6. Microcosms under in situ-like conditions with labeled substrates
13
C-
The intrinsic biodegradation potential was studied in sedimentegroundwater microcosms that were spiked with benzene, naphthalene, or acenaphthene in 13C-labeled form and incubated under in situ-like conditions. Sediments originated from four different depth layers of drilling location U13 in the strictly reducing part of the aquifer (Fig. 1). The content in d13C-CO2 in the headspace of all alive and dead controls stayed constant over the whole incubation experiment (Table S2). The aerobic degradation of benzene and naphthalene started within the first day, and the aerobic degradation of acenaphthene within less than one week after microcosm set-
4467
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 5 9 e4 4 6 9
Table 2 e Intrinsic first-order rate constants l derived from sedimentegroundwater microcosms at the site of the former cokery of Fle´malle. Drilling material from different depth layers of location U13 was used. The table shows mean values of duplicate incubations, except for 13C-benzene at 5 m, where one microcosm did not grow. n.d.: not determined. Depth [m]
5 7 9 12
Conditions
Oxic Anoxic Anoxic Anoxic
13
13
C-Benzene
13
C-Naphthalene
C-Acenaphthene
l [/d]
t1/2 [d]
l [/d]
t1/2 [d]
l [/d]
t1/2 [d]
2.1 102 1.2 103 1.6 103 0.9 103
33 581 422 771
1.5 103 0.3 104 0.7 104 1.1 104
457 28,250 9412 6104
8.7 103 3.4 104 n.d. n.d.
80 2056 n.d. n.d.
up. Under anoxic conditions, the mineralization of 13Cbenzene started within five days and reached a plateau after 48 days (158 days for the less permeable sediment from 7 to 8 m depth). Comparably, the anaerobic mineralization of naphthalene started within the first week of the incubation period but proceeded much slower than under oxic conditions. In sediments from two of the three examined anoxic depth layers, acenaphthene biodegradation started after a lag phase of more than 100 days. In anoxic sediment from 12 to 13 m depth, no intrinsic biodegradation potential for acenaphthene was detected over the whole duration of the microcosm experiment of 327 days (Table S2). Compared to Morasch et al. (2007), we applied an advanced, quantitative approach where first-order rate constants (l) of intrinsic contaminant degradation were determined based on an inorganic carbon mass balance (Eq. (8)). For sediments of the
unsaturated zone incubated in presence of air, l values were 2.1 102/d, 1.5 103/d, and 8.7 103/d for benzene, naphthalene, and acenaphthene (Table 2). These rates were equivalent to half-life times (t1/2) of 0.1, 1.3, and 0.2 years, respectively (Eq. (6)). In anoxic microcosms that contained sediments from the saturated zone, the 15e30 times lower rate constants corresponded to mean half-life times of 1.6, 40, and 5.6 years. Comparison of microcosm-derived first-order rate constants of benzene with the field-derived rates of the first approach revealed rates that were on average three times lower.
3.7.
Implications for the field site
Based on stable isotope generated first-order rate constants (Eq. (7)), predicted half-concentration distances (x1/2) of BTEX under anoxic conditions were between 15 and 32 m which is
Table 3 e Comparison of the three different approaches applied in this study summarizing their potentials and limits proving the in situ biodegradation of benzene, naphthalene, and acenaphthene. The quantitative and qualitative assessment of the three approaches is divided by a slash; D designates appropriate and L designates inappropriate approaches, ± stands for limited applicability. Stable isotopes (field)
13
C-microcosms
Signature metabolites
þ/þ
/b
þ/þ
þ/a
/þ
þ/þ
/
/c
þ/d
Restrictions
Fast Multi compound For smaller molecules
Slow, long term Single compound Time-consuming
Needs
Experimental 3 value
Prospects
Determination of reaction mechanisms from isotope effect
Fast Multi compound For compounds with signature metabolites Knowledge of degradation pathway Identification of new metabolites
Assessment
Gray fields mark the most suitable options. a Method successful in another study. b Metabolites are ambiguous. c New metabolites postulated based on this method. d Duration of experiment in parts too short for rate determination. e Stable isotope probing e for a review, see Madsen, 2006.
13C substrates Detection of metabolites Microcosms can be used for SIPe
4468
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 5 9 e4 4 6 9
in agreement with the flow and transport model of benzene (Batlle-Aguilar et al., 2009). In contrast, predictions based on the half-life times of benzene, naphthalene, and acenaphthene in anoxic microcosms resulted in x1/2 of 100, 412, and 16 m, respectively. The groundwater flow path from the contaminant source D2bis to the river bank is approximately 220 m long and the first 100 m are under anoxic conditions. For the following oxic plume interval, shorter half-concentration distances of 6, 13, and 1 m for benzene, naphthalene, and acenaphthene were predicted based on microcosmderived rate constants that were corrected for contaminant retardation in groundwater. Event-based infiltration of surface water supplies additional O2 in this part of the aquifer (Batlle-Aguilar et al., 2009). Predictions on the basis of stable isotopes matched the observations from the field: neither BTEX nor PAHs were ever detected in piezometers close to the river bank (Table S3).
4. Synthesis e complementarity of approaches In a recent review on the assessment of in situ biodegradation, Bombach et al. (2010) recommended using several approaches in combination according to the local conditions. In the present study, we employed three approaches in order to gain qualitative and quantitative information on the in situ biodegradation of BTEX and PAHs (Table 3). What distinguishes our combination from many others is its applicability to a wider variety of contaminants e independent of molecule size and hydrophobicity. CSIA of groundwater pollutants, the first technique that we applied, is useful for collecting data on the anaerobic in situ biodegradation of several BTEX compounds at once. Using CSIA, biodegradation rates of BTEX can be determined and the fate of naphthalene can be assessed in a qualitative way. However, substituted naphthalenes and larger PAHs cannot be examined. In practice, the CSIA-based field approach is barely applicable to study the intrinsic biodegradation potential of polyaromatic compounds because the bulk isotope effect is below the detection limit of the method (Elsner, 2010). Signature metabolite analysis, the second approach, bears the potential to identify new degradation intermediates and pathways, as presented in this study for the anaerobic degradation of acenaphthene and elsewhere for heterocyclic compounds (Safinofski et al., 2006). Nevertheless, its biggest potential lies in a reliable detection of the anaerobic biodegradation of (methylated) aromatic- and aliphatic hydrocarbons as well as heterocyclic compounds. The signature metabolite approach thus provides additional qualitative insights into the biodegradation of larger compounds where CSIA is not applicable. Microcosms with 13C-labeled substrates, the third approach, allow the quantitative assessment of biodegradation for any 13C-labeled compound of interest (Table 3). Their substrate specificity combined with very sensitive detection, makes 13C-microcosms a particularly interesting option for compounds that sorb, are rather recalcitrant, or cannot be studied by CSIA. Moreover, novel 13C-labeled metabolites of
the specific substrate may be extracted from the microcosms and provide new insights into degradation pathways. Even though 13C-microcosms may also be used as stand-alone technique, combination with CSIA and signature metabolite analysis in the field overcomes the limitation of substrate specificity and allows conclusions on a wider spectrum of contaminants.
Acknowledgements This work was funded by the EU integrated project Aquaterra. We thank M. Aragno for providing lab space and J. BatlleAguilar and S. Brouye`re for coordinating work on site. F. Chatelain is acknowledged for technical assistance and D. Grandjean for assistance at the GCeMS. We thank three anonymous reviewers for their valuable comments and suggestions.
Appendix. Supplementary data Supplementary data associated with this article can be found in the online version, at 10.1016/j.watres.2011.05.040.
references
Annweiler, E., Materna, A., Safinofski, M., Kappler, A., Richnow, H.H., Michaelis, W., Meckenstock, R.U., 2000. Anaerobic degradation of 2-methylnaphthalene by a sulfate-reducing enrichment culture. Applied and Environmental Microbiology 66, 5329e5333. Batlle-Aguilar, J., Brouye`re, S., Dassargues, A., Morasch, B., Hunkeler, D., Ho¨hener, P., Diels, L., Vanbroekhoven, K., Seuntjens, P., Halen, H., 2009. Benzene dispersion and natural attenuation in an alluvial aquifer with strong interactions with surface water. Journal of Hydrology 369, 305e317. Beller, H.R., 2002. Analysis of benzylsuccinates in groundwater by liquid chromatography/tandem mass spectrometry and its use for monitoring in situ BTEX biodegradation. Environmental Science and Technology 36, 2724e2728. Beller, H.R., Kane, S.R., Legler, T.C., McKelvie, J.R., Sherwood Lollar, B., Pearson, F., Balser, L., MacKay, D.M., 2008. Comparative assessments of benzene, toluene, and xylene natural attenuation by quantitative polymerase chain reaction analysis of a catabolic gene, signature metabolites, and compound-specific isotope analysis. Environmental Science and Technology 42, 6065e6072. Blum, P., Hunkeler, D., Weede, M., Beyer, C., Grathwohl, P., Morasch, B., 2009. Quantification of biodegradation for oxylene and naphthalene using first order decay models, MichaeliseMenten kinetics and stable carbon isotopes. Journal of Contaminant Hydrology 105, 118e130. Bombach, P., Richnow, H.H., Kastner, M., Fischer, A., 2010. Current approaches for the assessment of in situ biodegradation. Applied Microbiology and Biotechnology 86, 839e852. Bosma, T.N.P., Middeldorp, P.J.M., Schraa, G., Zehnder, A.J.B., 1997. Mass transfer limitation of biotransformation: quantifying bioavailability. Environmental Science and Technology 31, 248e252. Chang, B.V., Chang, S.W., Yuan, S.Y., 2003. Anaerobic degradation of polycyclic aromatic hydrocarbons in sludge. Advances in Environmental Research 7, 623e628.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 5 9 e4 4 6 9
Chapelle, F.H., Bradley, P.M., Lovley, D.R., Vroblesky, D.A., 1996. Measuring rates of biodegradation in a contaminated aquifer using field and laboratory methods. Ground Water 34, 691e698. Coates, J.D., Anderson, R.T., Lovley, D.R., 1996. Oxidation of polycyclic aromatic hydrocarbons under sulfate-reducing conditions. Applied and Environmental Microbiology 62, 1099e1101. Elshahed, M.S., Gieg, L.M., McInerney, M.J., Suflita, J.M., 2001. Signature metabolites attesting to the in situ attenuation of alkylbenzenes in anaerobic environments. Environmental Science and Technology 35, 682e689. Elsner, M., 2010. Stable isotope fractionation to investigate natural transformation mechanisms of organic contaminants: principles, prospects and limitations. Journal of Environmental Monitoring 12, 2005e2031. Foght, J., 2008. Anaerobic biodegradation of aromatic hydrocarbons: Pathways and prospects. Journal of Molecular Microbiology and Biotechnology 15, 93e120. Griebler, C., Safinofski, M., Vieth, A., Richnow, H.H., Meckenstock, R.U., 2004. Combined application of stable carbon isotope analysis and specific metabolites determination for assessing in situ degradation of aromatic hydrocarbons in a tar oil-contaminated aquifer. Environmental Science and Technology 38, 617e631. Hunkeler, D., Ho¨hener, P., Zeyer, J., 2002. Engineered and subsequent intrinsic in situ bioremediation of a diesel fuel contaminated aquifer. Journal of Contaminant Hydrology 59, 231e245. Hunkeler, D., Morasch, B., 2010. Isotope fractionation during transformation processes. In: Aelion, C.M., Ho¨hener, P., Hunkeler, D., Aravena, R. (Eds.), Environmental Isotopes in Biodegradation and Bioremediation. Taylor & Francis Group, Boca Raton, pp. 79e125. Langenhoff, A.A.M., Zehnder, A.J.B., Schraa, G., 1996. Behaviour of toluene, benzene and naphthalene under anaerobic conditions in sediment columns. Biodegradation 7, 267e274. Madsen, E.L., 2006. The use of stable isotope probing techniques in bioreactor and field studies on bioremediation. Current Opinion in Biotechnology 17, 92e97. Mazeas, L., Budzinski, H., Raymond, N., 2002. Absence of stable carbon isotope fractionation of saturated and polycyclic aromatic hydrocarbons during aerobic bacterial biodegradation. Organic Geochemistry 33, 1259e1272. Mihelcic, J.R., Luthy, R.G., 1988. Microbial degradation of acenaphthene and naphthalene under denitrification conditions in soilewater systems. Applied and Environmental Microbiology 54, 1188e1198. Morasch, B., Ho¨hener, P., Hunkeler, D., 2007. Evidence for in situ degradation of mono- and polyaromatic hydrocarbons in alluvial sediments based on microcosm experiments with 13 C-labeled contaminants. Environmental Pollution 148, 739e748.
4469
Morasch, B., Ho¨hener, P., Hunkeler, D., 2011. Determination of in situ biodegradation rates using 13C-labeled aniline. In: Schirmer., M., Hoehn, E., Vogt, T. (Eds.), GQ10: Groundwater Quality Management in a Rapidly Changing World. IAHS, Zu¨rich, Switzerland, pp. 287e290. National Research Council (NRC), 2000. Natural Attenuation for Groundwater Remediation. In: Remediation, C.o.I., Board, W.S. a.T., Management, B.o.R.W., Commission on Geosciences, E., Resources (Eds.). National Academy Press, Washington, D.C. Phelps, C.D., Battistelli, J., Young, L.Y., 2002. Metabolic biomarkers for monitoring anaerobic naphthalene biodegradation in situ. Environmental Microbiology 4, 532e537. Richnow, H.H., Annweiler, E., Michaelis, W., Meckenstock, R.U., 2003a. Microbial in situ degradation of aromatic hydrocarbons in a contaminated aquifer monitored by carbon isotope fractionation. Journal of Contaminant Hydrology 65, 101e120. Richnow, H.H., Meckenstock, R.U., Ask, L., Baun, A., Ledin, A., Christensen, T.H., 2003b. In situ biodegradation determined by carbon isotope fractionation of aromatic hydrocarbons in an anaerobic landfill leachate plume (Vejen, Denmark). Journal of Contaminant Hydrology, 59e72. Rothermich, M.M., Hayes, L.A., Lovley, D.R., 2002. Anaerobic, sulfate-dependent degradation of polycyclic aromatic hydrocarbons in petroleum-contaminated harbor sediment. Environmental Science and Technology 36, 4811e4817. Safinofski, M., Griebler, C., Meckenstock, R.U., 2006. Anaerobic cometabolic transformation of polycyclic and heterocyclic aromatic hydrocarbons e evidence from laboratory and field studies. Environmental Science and Technology 40, 4165e4173. Safinofski, M., Meckenstock, R.U., 2006. Methylation is the initial reaction in anaerobic naphthalene degradation by a sulfatereducing enrichment culture. Environmental Microbiology 8, 347e352. Schwarzenbach, R.P., Gschwend, P.M., Imboden, D.M., 2003. Environmental Organic Chemistry. John Wiley & Sons, New York. SRC Inc., 2009. SRC Physical Properties Database. Available from: http://www.syrres.com. Steinbach, A., Seifert, R., Annweiler, E., Michaelis, W., 2004. Hydrogen and carbon isotope fractionation during anaerobic biodegradation of aromatic hydrocarbons: a field study. Environmental Science and Technology 38, 609e616. Yuan, S.Y., Chang, B.V., 2007. Anaerobic degradation of five polycyclic aromatic hydrocarbons from river sediment in Taiwan. Journal of Environmental Science and Health Part B e Pesticides Food Contaminants and Agricultural Wastes 42, 63e69. Zamfirescu, D., Grathwohl, P., 2001. Occurrence and attenuation of specific organic compounds in the groundwater plume at a former gasworks site. Journal of Contaminant Hydrology 53, 407e427.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 7 0 e4 4 8 2
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Determination of sorption of seventy-five pharmaceuticals in sewage sludge Maritha Ho¨rsing a,1,*, Anna Ledin a,1, Roman Grabic b,2, Jerker Fick b, Mats Tysklind b, Jes la Cour Jansen c, Henrik R. Andersen a a
Department of Environmental Engineering, Technical University of Denmark, B113, DK-2800 Kgs Lyngby, Denmark Department of Chemistry, Umea˚ University, SE-901 87 Umea˚, Sweden c Water and Environmental Engineering at Department of Chemical Engineering, Lund University, P.O. Box 124, SE-221 00 Lund, Sweden b
article info
abstract
Article history:
Sorption of 75 active pharmaceutical ingredients (APIs) to three different types of sludge
Received 14 December 2010
(primary sludge, secondary sludge with short and long sludge age respectively) were
Received in revised form
investigated. To obtain the sorption isotherms batch studies with the APIs mixture were
26 May 2011
performed in four nominal concentrations to water containing 1 g of sludge. The range of
Accepted 28 May 2011
APIs concentrations was between ng L1 to mg L1 which are found in the wastewater
Available online 14 June 2011
effluents. Isotherms were obtained for approximately 45 of the APIs, providing distribution coefficients for linear (Kd), Freundlich (Kf) and Langmuir (KL) isotherms. Kd, Kf and KL ranging
Keywords:
between 7.1 104 and 3.8 107, 1.1 102 and 6.1 104 and 9.2 103 and 1.1 L kg1,
Distribution coefficients (Kd)
respectively. The obtained coefficients were applied to estimate the fraction of APIs in the
Pharmaceuticals
water phase (see Abstract Graphic). For 37 of the 75 APIs, the predicted presence in the
Sorption isotherms
liquid phase was estimated to >80%. 24 APIs were estimated to be present in the liquid
Sludge
phase between 20 and 80%, and 14 APIs were found to have 3 103 are considered to be volatile (Ternes and Joss, 2008). The aim was to get four API equilibriums with water concentration (Cw; g L1) in the range of
Table 1 e The experimental set up, including API concentrations, pH sludge concentration and the number of bottles per blank/zero sludge/sludge. From each bottle triplicate solid phase extractions (SPE) were made. API conc. (mgL1) Blank for each sludge No sludge Primary sludge Avedøre Secondary sludge short sludge age Sjo¨lunda Secondary sludge long sludge age Klagshamn
0 0.08 0.08 0.08 10 10 10 0.08 10
0.4 0.4 0.4
2 2 2
10 10 10
0.4
2
10
pH
Sludge conc. (gL1)
No. of bottles
7 7 7 7
1 0 1 1 10 50 1
3 4 4 4 1 1 1 4 1
6 7 8
4472
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 7 0 e4 4 8 2
Table 2 e Sorption isotherms obtained during sorption to primary sludge. P is the significance that the linear model has a better fit than another model tested, the model tested was Freundlich and Langmuir. Linear model 1
Kd (L kg ) Afluzosin Amitryptiline Atenolol Atracurium Azelastine Biperiden Bupropion Chloprothixene Citalopram Clomipramine Clonazepam Clotrimazol Cyproheptadine Desloratidine Dicycloverin Donepezil Duloxetine Etonogestrel Ezetimibe Fexofenadine Fluoxetine Flutamide Glibenclamide Glimepiride Haloperidol Hydroxyzine Irbesartan Ketoconazole Loperamide Maprotiline Megesterol Mianserin Nefazodone Oxazepam Paroxetine Pizotifen Progesterone Repaglinide Risperidone Sertraline Sulfamethoxazol Telmisartan Tramadol Trimethoprim Verapamil
3
1.8 10 4.1 103 4.6 102 3.5 102 6.4 103 8.2 102 85 3.8 104 5.4 102 1.7 104 5.7 102 3.2 104 1.1 104 3.7 103 1.4 103 3.6 103 1.3 104 No fit 2.3 103 2.7 103 1.0 104 1.5 103 3.6 103 2.1 103 1.0 104 1.2 103 7.0 102 9.7 103 1.4 104 6.7 103 No fit 3.0 103 1.4 104 7.9 102 1.4 104 4.7 103 7.5 102 1.7 102 1.9 103 3.5 104 3.2 102 1.3 103 1.1 102 3.9 102 1.8 103
Freundlich model
2
R (%)
1
Kf (L g )
n
2
R (%)
Langmuir model P (%)
99 98 88 100 82 88 98 98 76 99 96 91 98 99 92 96 77
2.7 3.1 4.0 0.24 46 1.3 4.8 103 8.3 0.97 7.6 0.10 19.0 2.2 1.9 1.2 10.4 30.8
1.1 0.96 1.3 0.96 1.4 1.1 0.76 0.77 1.1 0.88 0.84 0.92 0.81 0.92 0.98 1.2 1.2
99 98 91 100 86 88 99 99 76 99 96 92 91 100 94 96 78
42 70 17 40 11 83 0.99 0.98 83 8.5 33 59 30 17 90 21 53
97 82 99 88 77 92 76 98 92 88 98 99
4.9 102 1.1 105 2.7 0.13 No fit 1.9 103 60.9 3.2 1.5 102 1.6 0.60 1.7
0.68 0.38 0.83 0.77
99 94 99 90
1.1 0.15 3.6 27
0.53 1.4 1.1 0.70 0.79 0.69 0.83
97 80 99 93 90 99 100
0.8 21 14 19 30 0.04 0.05
81 98 90 96 100 98 94 99 97 77 100 94 98 99
12.5 2.7 0.80 5.6 No fit 0.18 1.4 3.1 4.1 0.91 0.86 4.9 104 0.43 1.1
1.2 0.79 1.0 0.87
83 99 90 97
36 4.0 99 28
0.86 1.3 1.1 0.72 1.1 0.95 0.63 1.0 0.94
99 95 99 99 77 100 96 98 99
14 19 36 0.4 79 51 3.3 91 26
90% of the starting concentration (C0; g L1) and the LOQ. The water solubility should be larger than the starting concentration; Cw C0; LOQ < Cw. Sludge concentration in the experiments was 1 g L1, with exception for one sludge where it was increased to 10 and 50 g L1 (Table 1). The 1 g L1 sludge density was about 5- to 10-fold more than the realistic sludge production that can occur in a WWTP which ensures that APIs that adsorb significantly to sludge in a real WWTP also would be removed significantly from the water phase in the batch experiment which was the basis for determining a Kd value, while low sorbing APIs Kds will not be determined.
1
smax (L g ) 5.5 No fit 5.7 No fit 2.0 2.4 No fit No fit 1.0 No fit No fit No fit No fit No fit 6.2 1.9 1.6 No fit No fit No fit No fit No fit No fit No fit 1.5 1.9 No fit No fit No fit No fit No fit 1.6 No fit 2.1 No fit No fit No fit 7.6 3.2 No fit 8.6 No fit No fit 3.7 No fit
R2 (%)
KL
P (%)
5
99
31
103
1.4 104
90
18
104 104
5.8 104 4.6 105
89 89
3.5 46
104
6.7 105
104 104 104
2.7 105 2.6 104 1.1 103
94 97 79
66 7.3 37
104 104
1.2 103 8.3 105
81 99
12 16
104
2.9 104
85
17
104
4.3 105
90
71
103 104
2.8 105 7.0 105
94 99
53 17
103
5.4 105
78
58
104
1.1 105
98
73
10
4
3.7 10
7.7
59
A stock solution including the 75 API of 0.1 g L1 was prepared in MeOH from which the four MeOH stock solutions of 0.4, 2.0, 10.0 and 50.0 mg L1 were prepared, respectively. These stock solutions suited the design (Table 1) and were theoretically determined to fit the criteria for the present study. Bo¨hm and During (2010) showed that there was no significant difference between determination of the distribution coefficient KDOC for single compounds or mixtures. Artificial media was used in the study for the water phase (Berg and Nyholm, 1996), modified as described by Andersen et al. (2005). The artificial sewage was a phosphate-buffered mineral media containing Ca2þ, Kþ, Mg2þ, Naþ, Cl and SO42.
4473
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 7 0 e4 4 8 2
Table 3 e Sorption isotherms obtained during sorption to secondary sludge long sludge age P is the significance that the linear model has a better fit than another model tested, the model tested was Freundlich and Langmuir. Linear model 1
Kd (L kg ) Afluzosin Alprazolam Amitryptiline Atenolol Atracurium Azelastine Biperiden Bisoprolol Bupropion Chloprothixene Citalopram Clomipramine Clotrimazol Cyproheptadine Desloratidine Dicycloverin Donepezil Duloxetine Eprosartan Estradiol Etonogestrel Ezetimibe Fexofenadine Fluoxetine Flutamide Glibenclamide Glimepiride Haloperidol Hydroxyzine Irbesartan Ibuprofen Ketoconazole Levonorgestrel Loperamide Maprotiline Medroxyprogesterone Megesterol Mianserin Nefazodone Orphenadrine Oxazepam Paroxetine Pizotifen Progesterone Repaglinide Risperidone Sertraline Sotalol Sulfamethoxazol Telmisartan Tramadol Trimethoprim Venlafaxine Verapamil
3
10 102 103 103 102 103 102 102 102 104 102 103 104 103 103 103 102 103 71 No fit No fit 3.0 103 3.6 102 6.0 103 7.5 102 1.3 103 9.6 102 2.9 103 7.2 102 9.4 102 3.6 102 8.5 103 2.6 102 5.5 103 4.5 103 1.2 102 5.9 102 9.1 102 8.3 103 6.4 102 1.1 103 8.3 103 3.1 103 1.1 103 2.1 102 6.5 102 1.7 104 3.6 102 3.7 102 No fit 1.9 102 4.2 102 1.0 102 4.0 102 1.2 7.4 2.8 1.6 4.7 2.0 7.5 1.1 1.4 2.0 2.1 6.7 3.4 3.6 2.9 1.7 9.7 2.9
2
R (%)
Freundlich model 1
Kf (L kg )
100 94 99 94 100 99 98 4.4 99 98 94 100 96 100 100 99 99 98 93
2.2 1.6 102 1.9 22 0.37 3.2 0.13 46 0.35 9.4 0.90 3.4 4.5 3.7 2.1 0.50 4.6 2.8 0.43
96 95 99 90 93 99 98 98 94 91 91 61 97 99 70 82 99 96 99 87 97 100 87 17 98 92 99 96
3.0 0.2 1.3 1.5 0.2 0.2 3.5 0.54 5.3 1.6 0.70 18 0.53 0.65 5.2 6.7 1.4 0.96 0.12 11 0.63 1.9 8.6 10 2.8 1.3 0.56 5.2
99 99 85 84
102
104
104 102
103
9.0 102 0.18 0.77 20
2
n
R (%)
Langmuir model P (%)
1
smax(L kg ) 4
4.0 4.6 40 0.0 40 31 7.7 0.14 4.2 14 25 1.4 2.2 93 40 13 0.0 98 38
3.7 No No 7.8 No 5.8 No 9.9 5.3 No 3.8 No No No No No 1.0 No 1.8
10 fit fit 103 fit 104 fit 102 103 fit 103 fit fit fit fit fit 104 fit 103
0.63 0.93 0.83 0.50 0.82 0.83 1.0 0.97 0.54 0.73 0.73 1.9 103 0.77 0.79 1.8 1.4 1.1 0.76 0.84 1.4 0.73 0.94 1.3 1.8 1.2 0.71 1.1 0.68
100 96 100 99 100 99 98 68 100 99 95 100 98 100 100 99 100 98 94 0.0 0.0 99 96 100 96 99 99 98 98 97 94 95 81 98 100 81 85 99 98 98 91 98 100 90 98 99 97 99 98
0.0 74 0.80 0.13 8.2 11 78 68 1.2 38 0.8 1.4 0.83 0.05 0.08 16 45 1.6 2.6 9.2 0.07 26 17 0.0 1.1 1.3 35 5.6
No No No No No No 1.1 No No No No 1.9 No No 9.5 5.3 2.6 No No 6.5 No No 7.4 2.7 8.9 No 1.3 No
fit fit fit fit fit fit 105 fit fit fit fit 103 fit fit 102 103 104 fit fit 103 fit fit 103 103 103 fit 104 fit
0.92 0.91 1.3 1.8
99 99 86 96
39 16 32 0.0
No No 2.1 3.3
fit fit 103 103
1.1 0.69 0.95 1.5 0.97 1.1 0.84 3.0 1.1 0.89 1.2 0.91 0.73 1.0 0.96 0.88 1.2 1.0 1.2
Sludge was added to 1 L borosilicate glass bottles. In order to inhibit microbial growth oxygen was removed by purging with N2(g) for 1 min and Na2SO3 was added to a final concentration of 50 mg L1 to each bottle. The bottles were left on stirring in the dark at þ4 C in order for the sludge to
KL
R (%)
P (%)
5
100
1.3
4.9 104
99
0.0
3.9 105
99
1.3 103 3.5 105
86 100
8.1 105
95
1.5 104
100
6.6 105
94
32
2.8 105
98
76
6.6 104
87
4.4 104 2.4 104 4.0 105
85 87 99
3.8 7.1 29
3.6 104
92
5.2
2.9 104 2.8 104 1.2 104
92 99 99
5.4 0.0 0.15
3.3 105
99
14
7.9 105 4.3 104
0.0 98
25 0.0
4.0 10
22 0.0 1.4 20
0.0
0.23
rehydrate. After 12 h, 200 mL of the API stock solutions of 0.4, 2.0, 10.0 or 50.0 mg L1 were added to the bottles using a Hamilton syringe giving the final concentrations presented in Table 1. The bottles were left on stirring in the dark at þ4 C for 12 h. Experiments were performed at three pH values
4474
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 7 0 e4 4 8 2
Table 4 e Sorption isotherms obtained during sorption to secondary sludge short sludge age. P is the significance that the linear model has a better fit than another model tested, the model tested was Freundlich and Langmuir. Linear model 1
Kd (L kg ) Alprazolam Amitryptiline Atenolol Atracurium Azelastine Biperiden Bisoprolol Bupropion Chloprothixene Clomipramine Clotrimazol Cyproheptadine Desloratidine Dicycloverin Diltiazem Duloxetine Estradiol Etonogestrel Ezetimibe Fexofenadine Fluoxetine Flutamide Glibenclamide Glimepiride Haloperidol Hydroxyzine Irbesartan Ibuprofen Ketoconazole Levonorgestrel Loperamide Maprotiline Medroxyprogesterone Megesterol Mianserin Nefazodone Orphenadrine Oxazepam Paroxetine Pizotifen Progesterone Repaglinide Risperidone Sertraline Sotalol Sulfamethoxazol Telmisartan Trimethoprim Verapamil
2
4.3 10 2.8 103 1.9 103 6.1 102 1.4 103 8.4 102 94 2.0 102 No fit 7.3 103 No fit 5.3 103 3.2 103 1.7 103 4.4 102 3.2 103 2.3 102 2.4 102 8.5 103 6.7 102 5.7 103 1.2 103 2.3 103 2.6 103 1.7 103 6.0 102 No fit 2.0 102 No fit 2.0 102 1.1 104 3.9 103 2.5 102 8.3 102 5.2 102 8.9 103 5.4 102 1.6 103 8.6 103 3.1 103 1.1 103 5.1 102 3.3 102 No fit 7.4 102 2.8 102 No fit 2.8 102 6.3 102
2
Freundlich model 1
R (%)
Kf (L kg )
80 96 100 98 96 96 64 97
No fit 4.2 102 No fit 2.9 102 1.7 102 0.13 7.5 1.9 102
88 97 96 97 100 80 72 73 87 94 91 93 97 94 98 95
Langmuir model 1
smax(L kg )
n
R (%)
P (%)
0.66
98
0.27
0.73 0.65 0.82 2.0 0.79
99 99 96 79.6 98
6.0 103
0.49
97
0.0
0.23 0.10 0.11 0.65 1.7 0.16 0.11 0.72 2.5 1.4 1.6 0.16 9.1 0.90 0.27
0.70 0.69 0.75 1.0 0.37 0.97 0.92 0.98 0.60 0.47 0.49 0.75 0.70 1.0 0.91
99 99 98 100 99 72 73 87 98 99 99 98 94 96 95
0.0 0.22 3.1 35 0.0 93 87 91 0.54 0.0 0.0 1.2 2.0 75 56
No fit No fit No fit 5.2 104 No fit 1.6 104 No fit 9.4 104 No fit No fit No fit No fit No fit No fit No fit
81
62 61 0.88 3.1 65 66 22 0.0 25 40 0.0 0.0 20 28 45
105
103 103 104 102
0.1 0.0 26 2.0 6.2
80
1.1
1.3
86 91 91 96 86 94 92 98 97 82 98 96 98 98
0.40 0.16 9.3 0.46 0.30 6.6 3.0 0.17 3.1 5.6 0.20 0.21 1.3 0.15
1.1 0.62 0.66 1.1 0.89 0.80 0.54 0.88 1.1 0.37 0.74 0.83 1.1 0.92
83 96 95 96 86 95 99 98 97 98 1005 96 98 98
99 95
0.22 No fit
0.87
99
6.3
98 96
1.3 102 0.90
0.75 1.0
100 96
0.15 75
102
102 102
10
5
(Table 1). The pH 7.0 was chosen based on a typical pH in WWTPs and pH 6.0 and 8.0 were based on low and high values from WWTPs. Control batches without APIs in the sludge and standard APIs solutions were prepared (Table 1).
2.3.
2
Extraction and chromatography
Triplicate extractions were made from each 1 L borosilicate glass bottle. After 12 h of stirring, the samples were allowed to
KL
R (%)
P (%)
3.4 104
81
1.2
9.2 106
100
27
1.6 105
72
85
1.0 104
87
65
2.4 103
1.3 104
82
56
1.1 104 No fit No fit 9.7 103 No fit No fit No fit No fit 2.3 104 No fit No fit No fit 2.5 104 No fit
1.9 105
83
74
3.0 105
96
59
8.7 105
98
33
2.4 105
98
44
2.1 105
96
63
No fit
No fit No fit 1.1 103 No fit No fit
No fit No fit No fit 3.3 104
stand for 30 min in order to let the sludge settle. In order to remove particles, the liquid phase was decanted and filtered through a glass microfiber filters (GC/F; VWR Denmark). In the filtered samples (100 g) a surrogate standard mixture was added followed by solid phase extraction using OASIS HLB (6cc, Waters, Sweden). A detailed description of sample preparation and analyses employing the LC-MS/MS methodology reported in Grabic et al., (unpublished data) and Fick et al. (2009) may be found as Supplementary data S1.
4475
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 7 0 e4 4 8 2
2.4.
Data analysis
The measured concentrations of APIs in the water phase (C0) where no sludge was added and when sludge was added (Cw) and SS were used to calculate Cs as follows in each experimental replicate:
Removed fraction ðFR Þ ¼
Cs ¼ ðC0 Cw Þ=SS
(1)
The sorption isotherms define the equilibrium between the concentration of a chemical in aqueous and solid phases (Schwarzenbach et al., 2003). With batch sorption experiments including multiple concentrations, sorption isotherms may be constructed, from which the solidewater distribution coefficients can be determined. Three equations are used here to describe the sorption isotherms; linear (Eq. (2)), Freundlich (Eq. (3)) and Langmuir (Eq. (4)). Cs is the concentration sorbed to the sludge (g kg1) and Kd is the linear sorption constant. Kf is the Freundlich coefficient and n is the Freundlich exponent. s represents the total number of surface sites per mass of sorbent and KL is the Langmuir coefficient. Linear Cs ¼ Kd Cw Freundlich Cs ¼ Kf Cw
(3)
smax KL Cw 1 þ KL Cw
(6)
The sludge production from treatment of municipal sewage can be considered reasonably constant irrespective of the methods of treatment. Based on Henze et al. (2002) the typical amounts of sludge removed from a WWTP can be calculated for primary and secondary sludge. For a WWPT with 2 h of settling time, the removal of primary sludge was estimated to 210 g m3 of treated wastewater. In order to estimate the removal of secondary sludge, the yield coefficient for a low load treatment plant was employed giving a removal of 110 g m3 for the secondary sludge. The corresponding value for a high load treatment plant would be 165 g m3.
Estimation of the sorbed fraction
The estimation of the sorbed fraction of each API was made by employing the obtained Kd-values. In those cases where both Freundlich and Langmuir isotherms were found to have a better fit than the linear, the one with the best significance was chosen as the best fit. For the APIs where the Freundlich or Langmuir isotherm gave the best fit (see Tables 2e4), Kdvalues were calculated for a water concentration of 1 mg L1. The fraction of the APIs at equilibrium for a given sludge concentration (SS; kg L1) was calculated using the Kd-values (L kg1) according to Eq. (5). Sorbed fraction ðFS Þ ¼ 1 Fw ¼
3.
Results and discussion
3.1.
Sludge properties
(4)
The linear isotherm is the simplest case where the affinity of the API remains constant over the concentration interval. The Langmuir isotherm may have the best fit in cases where the sorbent becomes saturated at higher concentrations of API. Freundlich, is commonly employed to describe experimentally obtained sorption data (Schwarzenbach et al., 2003). The software GraphPad Prism 5 for Windows (GraphPad Software, Inc.) was used for data evaluation, using a 95% confidence interval for the best-fit sorption isotherms. The two hypotheses tested were whether the linear isotherm was a better fit than the Freundlich isotherm, and whether the linear was a better fit compared to Langmuir isotherm. Furthermore, in order to qualify as the best fit the R2-value for the curve should be >0.7, otherwise no fit was made.
2.5.
RESS Kd 1 þ RESS Kd
(2) 1 n
Langmuir Cs ¼
which would not be lost either by degradation or stripping, but that will be removed at equilibrium can be calculated as shown in Eq. (6). Thus, the fraction of APIs in the water phase can be calculated after removal of sludge and the APIs sorbed to the sludge removed.
CS SS Kd ¼ CW þ SS CS 1 þ SS Kd
(5)
Furthermore, if the mass of the sludge removed from the WWTP per volume of treated sewage (RESS; kg L1) is known, the fraction of the total APIs load into the activate sludge tank
An ocular inspection of the freeze-dried sludge showed that the primary sludge may be described as wadding, whereas the two types of secondary sludge had an appearance as instant coffee. The ocular differences between the different types of sludge may be due to their origin. Primary sludge is mainly settable particles of wastewater including faeces, toilet paper and particles of food, while secondary sludge consists of bacterial biomass and biopolymers created by bacteria. It is likely that the mainly plant/wood derived primary sludge has different densities of functional groups and aromatic rings compared to the bacteria derived secondary sludge.
3.2.
Sorption isotherms
Due to analytical limitations and the experimental conditions it was possible to determine sorption isotherms for 4452 APIs out of the 75 APIs (Table S2). The linearly obtained Kd values ranged from 85 to 38 400, 199 to 11 340 and 71 to 34 050 L kg1 for primary sludge, secondary sludge with short sludge age and secondary sludge with long sludge age, respectively. Examples of the obtained isotherms are shown in Fig. 1. Tables 2e4 presents for 1 g L1 sludge the obtained distribution coefficients, Freundlich coefficient, Freundlich exponent, smax and Langmuir coefficient in the cases where the isotherms fitted these isotherm descriptions. The sorption isotherm with the best fit was predominantly linear followed by the Freundlich isotherm within the studied concentration range (0.08e10 mg L1). Tables 2e4 exhibit the order of significance for each hypothesis tested. The isotherms coefficients did not change significantly even at the higher sludge densities (Table S3). The average difference between the Kd’s obtained from 1 g L1 sludge
4476
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 7 0 e4 4 8 2
Fig. 1 e Example of the obtained sorption isotherms. From the top Pizotifen linear isotherm, second Maprotiline Freundlich isotherm and at the bottom Bisoprolol Langmuir isotherm.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 7 0 e4 4 8 2
compared to the Kd obtained, when 10 and 50 g L1 sludge was 4%.
3.3.
Sorption results and literature comparison
It was not possible to obtain sorption isotherms under the chosen experimental conditions for all APIs included. For Amiodiarone, Bromocriptine, Chlorpromazine, Clemastine, Dihydroergotamin, Fluphenazine, Levomepromazine, Meclozine, Miconazole, Perphenazine, Prometazine, Roxithromycine and Tamoxifen, the possible reason was too strong sorption to the glass surface of the bottle and/or the water surface. Thirteen of the APIs exhibited so low sorption on the sludge that a sorption isotherm was impossible to determine. The low affinity to sludge obtained for Buprenorphine, Cilazapril, Carbamazepine, Codeine, Diclofenac, Estrone, Flecainide, Fluconazole, Metoprolol, Naloxone and Rosuvastatine, Tramadol and Venlafaxine in all three sludges demonstrated the limit of the design in the opposite direction. The experimental design of the present study and the LOQ in the present matrixes for the individual APIs measured, provides the limits of Kd. The highest Kd that could be obtained within the concentration range 0.08e10 mg L1 and given the median LOQ, 8 ng L1 was 1.2 106 L kg1. However, if the LOQ was low (1 ng L1) or high (170 ng L1), the highest Kd could be expected within the range from 5 104 to 1 107 L kg1. Furthermore, assuming that at least 10% was required to be sorbed in order to determine a Kd value the lowest Kd value obtained within the present study was 100 L kg1. Some of the APIs were on the border between water soluble and low sorption. Bisoprolol, Clonazepam, Diltiazem, Eprosartan, Estradiol, Ibuprofen, Levonorgestrel, Medroxyprogesterone, Orphenadrine and Sotalol were the APIs which specifically exhibited low sorption for the primary sludge and no sorption isotherm could be obtained, which it were for the secondary sludges (Tables 3 and 4). Possible explanation for this behaviour could be the different surface properties of the sludges due to different origin. It should be noticed that Eprosartan exhibited low sorption in the secondary sludge long sludge age (Table 3) and that the sorption in the secondary sludge short sludge age was extremely low (i.e. could not be obtained). Diltiazem, an API for which sorption was obtained in the secondary sludge short sludge age but not in the secondary sludge long sludge age or the primary sludge, it could be speculated that some moieties in the sludge enhance the sorption. Alfuzosin, Citalopram, and Donepezil, the APIs were sorption was obtained in the primary sludge (Table 2) and the a secondary sludge long sludge age (Table 3) but not in the secondary sludge short sludge age this behavior give rise to speculations like there was something within the sludge which prevented sorption to occur or enhanced sorption based on unknown similarities and differences within the sludges, however no clear explanation could be found. However, the difference between the secondary sludge was not larger than 6 percentage points with the median at 4 percentage points. Greater difference (ca 20%) was found between aerobic and anaerobic sludge for EE2 (Zeng et al., 2009). Several of the Kd values reported in the literature would in comparison with the present study be below or around the lowest border for Kd values that could be obtained. Examples
4477
of such APIs are Codeine (Wick et al., 2009) and Estrone, (Andersen et al., 2005; Carballa et al., 2008). Further, compared to the present study, low Kd values were reported by Ternes et al. (2004) with Carbamazepine in primary and secondary sludge urban canals > lakes > the HR, which indicates that the municipal WWTPs were the main point sources of PFCs into surface waters. Finally, we found that the composition profiles PFCs in surface waters (this study) was similar to those in tap water (Mak et al., 2009), but not consistent with those in adult blood samples (Liu et al., 2009) from Shenyang, which suggests a poor removal efficiency of PFCs between surface water and tap water. The total daily intake calculation of PFOS by adults from Shenyang showed that the contribution of drinking water to human exposure of PFOS was minor.
4489
Acknowledgments This paper was supported by Ministry of Science and Technology (No. 2009DFA92390) of China and Nature Science Foundation (No. 20877043).
Appendix. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.watres.2011.05.036.
references
Ahrens, L., Barber, J.L., Xie, Z., Ebinghaus, R., 2009. Longitudinal and latitudinal distribution of perfluoroalkyl compounds in the surface water of the Atlantic Ocean. Environ. Sci. Technol. 43, 3122e3127. Ahrens, L., Xie, Z., Ebinghaus, R., 2010. Distribution of perfluoroalkyl compounds in seawater from Northern Europe, Atlantic Ocean, and Southern Ocean. Chemosphere 78, 1011e1016. Armitage, J., Cousins, I.T., Buck, R.C., Prevedouros, K., Russell, M.H., Macleod, M., Korzeniowski, S.H., 2006. Modeling globalscale fate and transport of perfluorooctanoate emitted from direct sources. Environ. Sci. Technol. 40, 6969e6975. Bao, J., Jin, Y., Liu, W., Ran, X., Zhang, Z., 2009. Perfluorinated compounds in sediments from the Daliao River system of northeast China. Chemosphere 77, 652e657. Becker, A.M., Gerstmann, S., Frank, H., 2008. Perfluorooctane surfactants in waste waters, the major source of river pollution. Chemosphere 72, 115e121. Boulanger, B., Vargo, J., Schnoor, J.L., Hornbuckle, K.C., 2004. Detection of perfluorooctane surfactants in Great Lakes water. Environ. Sci. Technol. 38, 4064e4070. Boulanger, B., Peck, A.M., Schnoor, J.L., Hornbuckle, K.C., 2005a. Mass budget of perfluorooctane surfactants in Lake Ontario. Environ. Sci. Technol. 39 1920-1920. Boulanger, B., Vargo, J.D., Schnoor, J.L., Hornbuckle, K.C., 2005b. Evaluation of perfluorooctane surfactants in a wastewater treatment system and in a commercial surface protection product. Environ. Sci. Technol. 39, 5524e5530. Busch, J., Ahrens, L., Xie, Z., Sturm, R., Ebinghaus, R., 2010. Polyfluoroalkyl compounds in the East Greenland Arctic Ocean. J. Environ. Monitor 12, 1242e1246. Fenton, S.E., Reiner, J.L., Nakayama, S.F., Delinsky, A.D., Stanko, J. P., Hines, E.P., White, S.S., Lindstrom, A.B., Strynar, M.J., PetropoulouS-S, E., 2009. Analysis of PFOA in dosed CD-1 mice. Part 2: Disposition of PFOA in tissues and fluids from pregnant and lactating mice and their pups. Reprod. Toxicol. 27, 365e372. Fromme, H., Schlummer, M., Mller, A., Gruber, L., Wolz, G., Ungewiss, J., Bhmer, S., Dekant, W., Mayer, R., Liebl, B., 2007. Exposure of an adult population to perfluorinated substances using duplicate diet portions and biomonitoring data. Environ. Sci. Technol. 41, 7928e7933. Giesy, J.P., Mabury, S.A., Martin, J.W., Kannan, K., Jones, P.D., Newsted, J.L., Coady, K., 2006. Perfluorinated Compounds in the Great Lakes. In: The Handbook of Environmental Chemistry, vol. 5, pp. 391e438. Guo, R., Sim, W.-J., Lee, E.-S., Lee, J.-H., Oh, J.-E., 2010. Evaluation of the fate of perfluoroalkyl compounds in wastewater treatment plants. Water Res. 44, 3476e3486. Huset, C.A., Chiaia, A.C., Barofsky, D.F., Jonkers, N., Kohler, H.P.E., Ort, C., Giger, W., Field, J.A., 2008. Occurrence and mass flows
4490
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 8 3 e4 4 9 0
of fluorochemicals in the Glatt Valley watershed, Switzerland. Environ. Sci. Technol. 42, 6369e6377. Ju, X., Jin, Y., Sasaki, K., Saito, H., 2008. Perfluorinated surfactants in surface, subsurface water and microlayer from Daliao coastal waters in China. Environ. Sci. Technol 42, 3538e3542. Kissa, E., 2001. Fluorinated Surfactants and Repellents, second ed. Marcel Dekker, Inc., New York. Lau, C., Anitole, K., Hodes, C., Lai, D., Pfahles-Hutchens, A., Seed, J., 2007. Perfluoroalkyl acids: a review of monitoring and toxicological findings. Toxicol. Sci. 99, 366e394. Li, F.S., Sun, H.S., He, N., Hao, Z.N., Zhao, L.J., Zhang, T., Sun, T.H., 2011. Perfluorinated compounds in Haihe River and Dagu Drainage Canal in Tianjin, China. Chemsphere 84, 265e271. Li, F., Zhang, C.J., Qu, Y., Chen, J., Chen, L., Liu, Y., Zhou, Q., 2010. Quantitative characterization of short- and long-chain perfluorinated acids in solid matrices in Shanghai. China. Sci. Total Environ. 408, 617e623. Liu, J., Li, J., Luan, Y., Zhao, Y., Wu, Y., 2009. Geographical distribution of perfluorinated compounds in human blood from Liaoning Province, China. Environ. Sci. Technol. 43, 4044e4048. Loganathan, B., Sajwan, K., Sinclair, E., Senthilkumar, K., Kannan, K., 2007. Perfluoroalkyl sulfonates and perfluorocarboxylates in two wastewater treatment facilities in Kentucky and Georgia. Water Res. 41, 4611e4620. Loos, R., Locoro, G., Comero, S., Contini, S., Schwesig, D., Werres, F., Balsaa, P., Gans, O., Weiss, S., Blaha, L., 2010a. Pan-European survey on the occurrence of selected polar organic persistent pollutants in ground water. Water Res. 44, 4115e4126. Loos, R., Locoro, G., Contini, S., 2010b. Occurrence of polar organic contaminants in the dissolved water phase of the Danube River and its major tributaries using SPE-LC-MS2 analysis. Water Res. 44, 2325e2335. Mak, Y.L., Taniyasu, S., Yeung, L.W.Y., Lu, G., Jin, L., Yang, Y., Lam, P.K.S., Kannan, K., Yamashita, N., 2009. Perfluorinated compounds in tap water from china and several other countries. Environ. Sci. Technol. 43, 4824e4829. Mclachlan, M.S., Holmstrom, K.E., Reth, M., Berger, U., 2007. Riverine discharge of perfluorinated carboxylates from the European continent. Environ. Sci. Technol. 41, 7260e7265. Nakayama, S., Strynar, M.J., Helfant, L., Egeghy, P., Ye, X., Lindstrom, A.B., 2007. Perfluorinated compounds in the Cape Fear Drainage Basin in North Carolina. Environ. Sci. Technol. 41, 5271e5276. Naile, J.E., Khim, J.S., Wang, T., Chen, C., Luo, W., Kwon, B., Park, J., Koh, C., Jones, P.D., Lu, Y., Giesy, J.P., 2010. Perfluorinated compounds in water, sediment, soil and biota from estuarine and coastal areas of Korea. Environ. Pollut. 158, 1237e1244. OECD, 2002. Hazard Assessment of Perfluorooctane Sulfonate (PFOS) and Its Salts. Organisation for Economic Co-operation and Development. Prevedouros, K., Cousins, I.T., Buck, R.C., Korzeniowski, S.H., 2006. Sources, fate and transport of perfluorocarboxylates. Environ. Sci. Technol. 40, 32e44. Pan, G., You, C., 2010. Sedimentewater distribution of perfluorooctane sulfonate (PFOS) in Yangtze River Estuary. Environ. Pollut. 158, 1363e1367. Quinones, O., Snyder, S.A., 2009. Occurrence of perfluoroalkyl carboxylates and sulfonates in drinking water utilities and related waters from the United States. Environ. Sci. Technol. 43, 9089e9095. Sinclair, E., Mayack, D.T., Roblee, K., Yamashita, N., Kannan, K., 2006. Occurrence of perfluoroalkyl surfactants in water, fish, and birds from New York State. Arch. Environ. Contam. Toxicol. 50, 398e410. Skutlarek, D., Exner, M., Fa¨rber, H., 2006. Perfluorinated surfactants in surface and drinking waters. Environ. Sci. Pollut. Res. 13, 299e307.
So, M.K., Taniyasu, S., Yamashita, N., Giesy, J.P., Zheng, J., Fang, Z., Im, S.H., Lam, P.K.S., 2004. Perfluorinated compounds in coastal waters of Hong Kong, South China, and Korea. Environ. Sci. Technol 38, 4056e4063. So, M.K., Miyake, Y., Yeung, W.Y., Ho, Y.M., Taniyasu, S., Rostkowski, P., Yamashita, N., Zhou, B.S., Shi, X.J., Wang, J.X., Giesy, J.P., Yu, H., Lam, P.K.S., 2007. Perfluorinated compounds in the Pearl River and Yangtze River in China. Chemosphere 68, 2085e2095. Stockholm Convention, 2009. http://chm.pops.int/Convention/ Pressrelease/OP4Geneva9May2009/tabid/542/language/en-US/ Default.aspx (accessed 10.1.2009). Sun, H.W., Gerecke, A.C., Giger, W., Alder, A.C., 2011. Long-chain perfluorinated chemicals in digested sewage sludges in Switzerland. Environ. Pollut 159, 654e662. Takagi, S., Adachi, F., Miyano, K., Koizumi, Y., Tanaka, H., Mimura, M., Watanabe, I., Tanabe, S., Kannan, K., 2008. Perfluorooctanesulfonate and perfluorooctanoate in raw and treated tap water from Osaka, Japan. Chemosphere 72, 1409e1412. Taniyasu, S., Kannan, K., Soc, M.K., 2005. Analysis of fluorotelomer alcohols, fluorotelomer acids, and short- and long-chain perfluorinated acids in water and biota. J. Chromatogr. A 1093, 89e97. Thompson, J., Lorber, M., Toms, L.L., Kato, K., Calafat, A.M., Mueller, J.F., 2010. Use of simple pharmacokinetic modeling to characterize exposure of Australians to perfluorooctanoic acid and perfluorooctane sulfonic acid. Environ. Int. 36, 390e397. Thompson, J., Eaglesham, G., Reungoat, J., Poussade, Y., Bartkow, M., Lawrence, M., Mueller, J.F., 2011. Removal of PFOS, PFOA and other perfluoroalkyl acids at water reclamation plants in South East Queensland Australia. Chemosphere 82, 9e17. Toms, L.-M.L., Calafat, A., Kato, K., Thompson, J., Harden, F., Hobson, P., Sjodin, A., Mueller, J.F., 2009. Polyfluoroalkyl chemicals (PFCs) in human blood serum from children and adults in Australia. Environ. Sci. Technol. 43, 4194e4199. US EPA, 2010. Long-chain perfluorinated chemicals (PFCs)Action Plan summary. http://www.epa.gov/opptintr/ existingchemicals/pubs/actionplans/pfcs.html. Wang, Y., Fu, J., Wang, T., Liang, Y., Pan, Y., Cai, Y., Jiang, G., 2010. Distribution of perfluorooctane sulfonate and other perfluorochemicals in the ambient environment around a manufacturing facility in China. Environ. Sci. Technol. 44, 8062e8067. Wei, S., Chen, L.Q., Taniyasu, S., So, M.K., Murphy, M.B., Yamashita, N., Yeung, L.W.Y., Lam, P.K.S., Wei, S., Chen, L.Q., 2008. Distribution of perfluorinated compounds in surface seawaters between Asia and Antarctica. Mar. Pollut. Bull. 54, 1813e1838. Yamashita, N., Kannan, K., Taniyasu, S., Horii, Y., Petrick, G., Gamo, T., 2005. A global survey of perfluorinated acids in oceans. Mar. Pollut. Bull. 51, 658e668. Yeung, L.W.Y., So, M.K., Jiang, G., Taniyasu, S., Yamashita, N., Song, M., Wu, Y., Li, J., Giesy, J.P., Guruge, K.S., Lam, P.K.S., 2006. Perfluorooctanesulfonate and related fluorochemicals in human blood samples from China. Environ. Sci. Technol. 40, 715e720. Zhang, T., Sun, H.W., Wu, Q., Zhang, X.Z., Yun, S.H., Kannan, K., 2010a. Perfluorochemicals in meat, eggs and indoor dust in China: assessment of sources and pathways of human exposure to perfluorochemicals. Environ. Sci. Technol. 44, 3572e3759. Zhang, T., Wu, Q., Sun, H.W., Zhang, X.Z., Yun, S.H., Kannan, K., 2010b. Perfluorinated compounds in whole blood samples from infants, children, and adults in China. Environ. Sci. Technol. 44, 4341e4347.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 9 1 e4 5 0 0
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Effect of temperature shocks on membrane fouling in membrane bioreactors Paula van den Brink a,b,*, On-Anong Satpradit c, Andre´ van Bentem d, Arie Zwijnenburg b, Hardy Temmink b,e, Mark van Loosdrecht a a
Department of Biotechnology, Delft University of Technology, Julianalaan 67, 2628 BC Delft, The Netherlands Wetsus, Centre for Sustainable Water Technology, PO Box 113, 8900 CC Leeuwarden, The Netherlands c van Hall Larenstein, University of Applied Sciences, PO Box 1528, 8901 BV Leeuwarden, The Netherlands d DHV Water, PO Box 1132, 3800 BC Amersfoort, The Netherlands e Department of Environmental Technology, Wageningen University, PO Box 8129, 6700 EV Wageningen, The Netherlands b
article info
abstract
Article history:
Temperature is known to influence the biological performance of conventional activated
Received 13 March 2011
sludge systems. In membrane bioreactors (MBRs), temperature not only affects the
Received in revised form
bioconversion process but is also shown to have an effect on the membrane performance.
16 May 2011
Four phenomena are generally reported to explain the higher resistance for membrane
Accepted 31 May 2011
filtration found at lower temperatures: (1) increased mixed liquor viscosity, reducing the
Available online 12 June 2011
shear stress generated by coarse bubbles, (2) intensified deflocculation, reducing biomass floc size and releasing EPS into the mixed liquor, (3) lower backtransport velocity and (4)
Keywords:
reduced biodegradation of COD. Although the higher resistance at low temperatures has
Membrane fouling
been reported in several papers, the relation with supernatant composition has not been
Temperature
investigated before. In this paper, the composition of the soluble fraction of the mixed
Membrane bioreactor
liquor is related to membrane performance after exposing the sludge to temperature
Particle size
shocks. Flux step experiments were performed in an experimental system at 7, 15, and 25
Flux step method
Celsius with sludge that was continuously recirculated from a pilot-scale MBR. After cor-
Polysaccharides
recting the permeate viscosity for temperature, higher membrane fouling rates were
Fouling mechanisms
obtained for the lower temperature in combination with low fouling reversibility. The soluble fraction of the MBR mixed liquor was analysed for polysaccharides, proteins and submicron particle size distribution. At low temperature, a high polysaccharide concentration was found in the experimental system as compared to the MBR pilot. Upon decreasing the temperature of the mixed liquor, a shift was found in particle size towards smaller particles. These results show that the release of polysaccharides and/or submicron particles from sludge flocs could explain the increased membrane fouling at low temperatures. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
Temperature is known to influence the biological performance of conventional activated sludge systems (Farrell and
Rose, 1967; McClintock et al., 1993; Metcalf and Eddy, 2004). Both the rate of treatment and the microbial composition are affected by temperature (Chiemchaisri and Yamamoto, 1994; Wile´n et al., 2000a, 2000b). Temperature also influences the
* Corresponding author. Department of Biotechnology, Delft University of Technology, Julianalaan 67, 2628 BC Delft, The Netherlands. E-mail address:
[email protected] (P. van den Brink). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.05.046
4492
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 9 1 e4 5 0 0
structure of sludge flocs, although conflicting results have been reported. Experiments in which the sludge settleability was studied at different temperatures, showed an increased tendency for sludge bulking at lower temperatures (5 C) (Henry and Salenieks, 1980). Long-term monitoring of a fullscale municipal MBR revealed that settleability was positively influenced by higher temperatures (Lyko et al., 2008). In other studies, higher temperatures showed the opposite effect. An increase of sludge volume index (SVI) was found at higher temperatures (Su¨ru¨cu¨ and C ¸ etin, 1990). Poor settling was also observed at elevated temperatures (35 C as compared to 15 C) for SBR reactors running at low SRTs (Krishna and van Loosdrecht, 1999). In MBR’s, temperature has a similar effect on the biological processes as in conventional activated sludge systems. However, because temperature also has an impact on sludge morphology, the membrane filtration process can be affected. Seasonal variations in membrane fouling, with increased membrane fouling at lower temperatures, have been mentioned for several full scale systems (Rosenberger et al., 2006; Lyko et al., 2008; Miyoshi et al., 2009; Moreau, 2010; Wang et al., 2010b). Temperature has an impact on membrane filtration because it determines permeate viscosity. However, even after correcting for this change in permeate viscosity, increased membrane fouling has been quantified before for a small temperature shift (17e18 C to 13e14 C) (Jiang et al., 2005). Four phenomena, induced by low temperatures, were mentioned that could explain this effect: (1) increased mixed liquor viscosity, reducing the stress generated by coarse bubbles; (2) more severe deflocculation, reducing biomass floc size and releasing extracellular polymeric substances (EPS) into the mixed liquor that may cause membrane fouling; (3) lower particle backtransport velocity because Brownian diffusion is linearly related to temperature; (4) reduced biodegradation of COD, resulting in higher concentrations of soluble and particulate COD in the reactor (Tian et al., 1994; Lishman et al., 2000), including potential foulants. Apart from these hypotheses, other explanations or phenomena could be of importance as well. Changes in the cake layer thickness and/or porosity were possible explanations for the increased membrane fouling found at low temperatures for a laboratory scale MBR treating domestic wastewater, in which the temperature was lowered every two weeks in steps of 5 C from 25 C to 5 C (Chiemchaisri and Yamamoto, 1994). Temperature also has an effect on bubble size: for pure water, bubble coalescence and consequently bubble size was increased as the liquid temperature was raised from 10 C to 30 C (Ribeiro jr. and Mewes, 2006). Coarse bubbles are generally accepted to be more effective in fouling mitigation than smaller bubbles (Li et al., 1997; Judd, 2005; Zhang et al., 2009). Therefore, low temperature might induce membrane fouling. Although more severe membrane fouling at low temperatures has been reported in several papers, the relation with supernatant composition has not been investigated before. In the experiments presented in this paper, the composition of the
supernatant with respect to organic carbon concentrations and submicron particle size distribution and its relation to membrane fouling was investigated in short term temperature experiments. The focus on submicron particle size was chosen, because these particles are more likely to cause irreversible fouling (Geilvoet, 2010). By changing the sludge temperature, while keeping the influent composition constant, the individual temperature experiments could be more properly compared than data from full scale MBRs run at different temperatures. Operational data from a full scale MBR were used to inventorise long term temperature effects on membrane fouling. A better understanding of both short and long term temperature effects on membrane fouling in MBRs can help to develop more effective membrane fouling control strategies.
2.
Materials and methods
2.1.
Experimental set-up
A pilot-scale MBR (Fig. 1) with a working volume of 85 L and fed with municipal wastewater was used as a source of activated sludge for the experiments. The wastewater was screened (5 mm) before entering the biological reactors. The COD of the wastewater was 350 mg L1 on average, but fluctuated between 150 and 600 mg L1. The overall hydraulic retention time (HRT) of the MBR was 6 h. The dissolved oxygen concentration in the aerobic tank was measured on-line and controlled at 1.5 mg L1 with a fine-bubble diffuser. The membrane tank was equipped with 5 PE flat sheet membranes with a nominal pore size of 0.3 mm and a surface area of 0.063 m2 each (Kubota, Japan). The channel width between the membranes was 7 mm. Coarse bubble aerators placed below the membranes provided a specific aeration demand (SAD) of 1.4 m3 m2 h1 (equivalent to 0.1 cm s1) to scour the membrane surface and in this manner reduce fouling. With a peristaltic pump, permeate was continuously extracted at a flux of 47.9 L m2 h1. The pilot reactor was operated at an HRT of 6.3 h and SRT of 25 days. The average mixed liquor suspended solid (MLSS) concentration was around 4.0 g L1 in the aerobic tank of the MBR. This low MLSS concentration was due to the low strength of the municipal wastewater. The MBR pilot had good COD removal and full nitrification. To measure filterability, sludge from the aerobic reactors of the MBR was continuously recirculated over a vessel with a volume of 5 L and a retention time of 1 h (Fig. 1). The vessel contained two homemade submerged flat-sheet PVDF membranes, with a nominal pore size of 0.1 mm and a surface area of 0.014 m2. The temperature in the aerated tank of the pilot was continuously monitored. For temperature control, a cooling spiral connected to a water bath (Thermo Haake DC30-K10, US) was placed in the vessel. The filterability set-up (also called “experimental system”) was equipped with a coarse bubble aerator to scour the membrane surface at a flow rate of about 28 m3 m2 h1 (equivalent to 2.1 cm s1). This SAD was very high compared to practice to compensate for the broad channel width between the membrane sheets. Also, the air flow was divided by a relatively small surface area of the membranes used in these tests, resulting in a high
4493
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 9 1 e4 5 0 0
Fig. 1 e Pilot-scale MBR and filterability set-up (Remy et al., 2010).
specific air flow. Although the applied membranes and hydrodynamics are not the same in the pilot reactor and the filterability set-up, the individual experiments performed in the filterability set-up can be properly compared. The transmembrane pressure (TMP) was monitored on-line (Endress þ Hauser, Cerabar M PMC 41), both in the pilot MBR and in the experimental system.
2.2.
Experimental procedure
Membrane fouling can be quantified using different methods. In the past few years, efforts have been made to quantify membrane fouling, and more specifically the critical flux, in a reproducible way by developing the so-called “flux step method” (Le-Clech et al., 2003). The critical flux concept was originally developed by Field et al. (1995) and described as such: “the critical flux hypothesis for microfiltration is that on start-up there exists a flux below which a decline of flux with time does not occur; above it, fouling is observed”. The improved flux step method gives, apart from the critical flux value, additional information on reversibility and irreversibility of the membrane fouling (van der Marel et al., 2009). Flux step experiments were performed at three different temperatures in the filterability set-up: 7, 15 (reference temperature) and 25 C. These temperatures were selected because they are representative for operating conditions in winter and summer times: values from 8.4 to 26.8 C have been reported in an extensive report on filterability in pilot and full-scale MBRs all over Europe (Moreau, 2010). In order to properly compare the lower and higher temperatures with the reference temperature, experiments were performed in the following sequence: 15 e 7 e 15 e 25 e 15 C, using the same membrane for all 5 experiments. The filtrations were performed in duplicate, i.e. two membrane sheets were filtering the same mixed liquor but had separate TMP measurements. The complete sequence of filtrations was carried out 3 times to minimise effects of varying wastewater composition. After reaching the required temperature (taking 1e3 h) in the filtration set-up, flux step experiments lasting 5 h were
performed. Samples of the mixed liquor in both the filtration set-up and the aerated tank of the pilot were taken after the first flux step cycle, i.e. one filtration and one relaxation step. The filtrations were performed according to the flux step method as described by van der Marel et al. (2009), applying 5 membrane fluxes for JH: 20, 40, 60, 80, and 100 L m2 h1. Each relaxation step of 15 min was followed by a filtration step of 45 min. Before starting the experiment, a clean water filtration was performed for 30 min to determine the clean water permeability of the membranes. After each experiment, the membranes were mechanically cleaned by thorough rinsing with a nozzle to remove all accumulated fouling. The total filtration resistance was calculated from the TMP, membrane flux and permeate viscosity, according to the following expression: Rt ¼ TMP=ðhJÞ
(1)
where Rt is total resistance (m1), h the permeate viscosity (Pa s) and J the membrane flux (m3 m2 s1). A temperature correction was performed on permeate viscosity, according to the following relation (Roorda and van der Graaf, 2001) h ¼ 0:497ðT þ 42:5Þ1:5
(2)
with temperature T in degrees Celsius. This relation agrees with the values as reported in the CRC Handbook of Chemistry and Physics and first published by Sengers and Watson (1986). Total fouling rate, i.e. the increase of the total filtration resistance in time, was calculated at each flux step from the slope of the resistance in time (R(t)) after 30 min of filtration. Reversibility upon relaxation for 15 min at a membrane flux of 5 L m2 h1 was expressed as a percentage of the total resistance increase for one membrane flux step: % Reversibility ¼ Rhigh n Rlow n
Rhigh n Rlow n1 100
(3)
Rhigh,n represents the resistance at the end of one filtration step, Rlow,n represents the average resistance at the
4494
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 4 9 1 e4 5 0 0
corresponding relaxation step and Rlow,n1 represents the average resistance at the previous relaxation step. For the schematic representation of this calculation, the reader is referred to van den Brink et al. (2009).
2.3.
Sampling and fractionation
As mentioned above, samples of the mixed liquors of the aerobic tank of the MBR pilot and of the experimental system were taken after the first flux step cycle. Mixed liquor supernatant was obtained with a centrifuge (Sigma, 2e16) at 3260 g for 10 min. After centrifugation, the supernatant was subsequently paper filtered (Whatman Black Ribbon 589/1, 12e25 mm) and membrane filtered (Cronus PTFE syringe filter, nominal pore size of 0.45 mm). The membrane filtrate, which is referred to as the soluble fraction of the supernatant, was used for all analyses.
2.4.
Analyses
Concentrations of polysaccharides and proteins were measured according to the methods of Dubois et al. (1956) and Biorad based on Bradford (1976), respectively using glucose and bovine serum albumin (BSA) as standards. Apart from chemical techniques, fouling potential can also be assessed by measuring the particle size distribution of the mixed liquor. Particle size measurements are often used in membrane fouling studies (Wisniewski and Grasmick, 1998; Wang et al., 2008, 2010a; Zhang et al., 2010; Remy et al., 2010). However, usually particle sizes of 1 mm and higher are measured, while this is not the interesting range for a direct relation to membrane fouling. Particle size measurements in the micrometer range could be interesting to, for example, study flocculation and deflocculation. For assessing the fouling potential of a solution, rather the submicron range should be considered. Particle size distribution of the soluble fractions (4log10 MS2 for the duration of the
Point-of-use technologies
experiment (287 days), while BSF with steel wool removed >4log10 MS2 for the first 160
Physical and chemical processes
days. Plug flow for the BSF was shown to depend on uniformity between the iron oxide
Virus
material and sand media grains. The results suggest that the duration of effective virus removal by iron-amended biosand filtration depends on source water conditions and the quantity and composition of iron material added. Overall, this study provides evidence that iron-amended BSFs may advance the field of point-of-use technologies and bring relief to millions of people suffering from waterborne diseases. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
An estimated 884 million people e13% of the world population e lack access to safe drinking water (UNICEF/WHO, 2008). As a result, millions of people die each year from water-related diseases (WHO, 2008). While it is not possible to quantify the proportion of deaths directly due to unsafe drinking water and not attributed to other fecaleoral transmission routes (Curtis et al., 2000), access to clean drinking water and proper sanitation can provide substantial improvements in health (Logsdon et al., 2002; Nelson and Murray, 2008). Point-of-use (POU) water treatments, which allow the purification of water at the point of consumption rather than at a centralized location, allow water quality to be improved at the household scale (Sobsey, 2002).
Already widespread in their usage, as of 2007, 19 million people are estimated to use POU water treatment, in addition to the 350 million people who boil their water (Clasen et al., 2007). Studies indicate that the improvement of water quality through the use of POU technologies results in 30e40% reductions in diarrheal disease (Clasen et al., 2007; Esrey et al., 1985, 1991; Fewtrell et al., 2005). One of the most promising and widespread POU technologies is the biosand filter (BSF), a household-scale, intermittently operated slow sand filter, in which the upper layers of sand media remain saturated in between operations to allow the formation of a biologically active layer (Sobsey et al., 2008). The BSF consists of a plastic or concrete hollow chamber that tapers slightly toward the bottom (CAWST, 2010). A drainage gravel layer is laid at the bottom of the chamber,
* Corresponding author. Tel.: þ1 217 244 5965; fax: þ1 217 333 6968. E-mail address:
[email protected] (T.H. Nguyen). 1 Present address: Department of Civil and Environmental Engineering, Stanford University, United States. 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.05.045
4502
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 0 1 e4 5 1 0
covered by a separating gravel layer and a filtration sand layer. Approximately 5 cm above the filtration sand layer sits a diffuser. An outlet tube collects water from bottom of the chamber, passes the water vertically, and discharges the water at the outlet located at a height between the diffuser and the top of the filtration sand layer. During 24-h cycle of filter usage, water is poured into the inlet reservoir. As a result, the hydraulic head pushes the water downwards through the sand filtration layer and into the drainage gravel layer, where it is collected by the outlet tube and ultimately discharged. As the water level drops in the inlet reservoir, the flow rate decreases. Flow ceases when the standing water within the inlet reservoir is at a height equal to the height of the outlet. The top portion of the filtration sand layer collects the suspended solids, organic constituents, and microorganisms of the source water. Since the filtration sand remains saturated during and in between operation, a biological zone, wherein the sand grains are covered in a biofilm, develops in the top 10e20 cm of the filtration sand layer. The biofilm is credited with the enhanced removal of suspended solids and pathogens through mechanical trapping, adsorption, predation, and natural death. Development of the biofilm results in greater removal of suspended solids and pathogens, but its development also reduces flow rates (Elliott, 2010; Elliott et al., 2008; Weber-Shirk and Dick, 1997). Dr. David Manz developed the first BSFs in the 1990s at the University of Calgary as a way of improving water quality for low-income families in rural areas with restricted access to safe drinking water (Manz et al., 1993). Since then, BSFs have been chosen by hundreds of humanitarian groups as the best method for improving water quality in developing countries and, as of 2009, it is estimated that over 200,000 BSFs have been implemented in over 70 countries (CAWST, 2010). Surveys reveal its wide acceptance by users due to the improved appearance, smell, and taste of the treated water (Ngai et al., 2007). Considering the criteria of water quantity produced, water quality, ease of use, and ease of access, BSFs have been identified as the point-of-use technology having the most potential to deliver sustainable potable water treatment to the developing world (Sobsey et al., 2008). Both laboratory and field studies have documented improved microbiological water quality through the use of the BSF. BSFs remove greater than 99.9% of Giardia cysts and Cryptosporidium oocysts (Palmateer et al., 1999). Bacterial concentrations are reduced 70e99.99%, depending on biofilm development and time of sampling (Baumgartner et al., 2007; Buzunis, 1995; Elliott et al., 2008; Stauber, 2006). The improved water quality has been attributed to at least 20% reductions in frequency of diarrheal illness in studies conducted in the Dominican Republic (Stauber, 2006) and Kenya (Tiwari et al., 2009). However, both field and laboratory researches have identified a critical shortcoming: BSFs are not highly effective in removing viruses (Elliott et al., 2008). Viruses cause approximately 40% of diarrheal illnesses in developing countries (Ramani and Kang, 2009), with rotavirus being the leading cause of childhood diarrhea hospitalizations worldwide (Parashar et al., 2006). In natural water conditions of pH 6e8, sand and most viruses are negatively charged, causing a net repulsion and reducing virus removal efficacy by sand filtration (Jin et al., 2000). Thus, water contaminated with
pathogenic viruses is not yet potable after passing through a BSF, and a form of virus removal is required to treat the effluent (CAWST, 2010). The addition of zerovalent iron to the sand media results in filters that more effectively remove viruses from water (You et al., 2005) as corrosion on the iron surface generates a positively charged oxide layer (Lukasik et al., 1999) to which the viruses may electrostatically adsorb (Ryan et al., 1999). The primary objectives of this study were to determine: (1) the efficacy of virus removal during the daily operation of the iron-amended BSF; (2) the efficacy of virus removal using different iron oxide sources; (3) the duration for which ironamended biosand filtration effectively removed viruses. This study is unique to other studies for two reasons. First, unlike previous iron-amended BSFs, the iron source was added to the top half of the sand media rather than in the diffuser basin to increase the contact time between viruses and iron oxides. Second, to our knowledge, this is the longest BSF study conducted examining virus removal in both iron-amended and unmodified BSFs. Both small-scale columns and householdscale BSFs were tested using bacteriophage MS2. Rotavirus was used for select tests with small-scale columns, due to difficulty in propagating the virus.
2.
Methods
2.1.
Virus selection and assay
Bacteriophage male specific type 2 (MS2) was selected as a model virus because of its structural resemblance to many human enteric viruses and its ease of use. MS2 was replicated and purified as described previously (Gutierrez et al., 2009, 2010; Kitis et al., 2003; Page et al., 2009; Sirikanchana et al., 2008) with the following modifications. Briefly, Escherichia coli (ATCC 15597) grown in tryptic soy broth solution was inoculated with MS2 and incubated, followed by the separation of MS2 via centrifugation. After cell lysis and virus release, debris was removed via microfiltration through 0.2mm and 0.05-mm low-protein-binding polycarbonate tracketched membranes (Whatman Nucleopore, USA). Virus was concentrated on a 100-kDa membrane surface (Koch Membranes, USA) in a Millipore ultra-microfiltration unit (Whatman Nucleopore, USA). The virus accumulated on the membrane surface was washed extensively with sterilized 1 mM NaCl solution to remove nutrients, microbial products and dissolved organic matter. The final MS2 stock was stored at 4 C at a concentration of 1011 PFU/mL. MS2 was enumerated by the double agar layer procedure USEPA Method 1602. Briefly, plaques formed due to the inoculation of E. coli with MS2 at 37 C for 16 h, and plates with between 20 and 200 plaques were used for calculating the concentration of MS2. Any plates containing more than 200 plaques were quantified from a higher dilution plate. Select experiments were also performed using rotavirus (RV) to verify MS2 results. Group A porcine rotavirus OSU strain was obtained from the American Type Culture Collection (catalog # VR892). Rotavirus was propagated in embryonic African green monkey kidney cells (MA-104 cells) and extracted from culture as described by Rolsma et al. (1994).
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 0 1 e4 5 1 0
The purification protocol was the same as for MS2, except with an additional filtration step in which a 0.05-mm membrane was used. Rotavirus was cleaned and stored in 1 mM sodium chloride (NaCl) plus 0.1 mM calcium chloride (CaCl2) during the 100 kDa ultrafiltration to prevent the dissociation of the outer capsid proteins (Ruiz et al., 1996). The final RV stock was stored at 4 C at a concentration of >105 focus forming unit (FFU) per mL. Rotavirus infectivity assays were carried out following the procedures described by Rolsma et al. (1998).
2.2.
Solution chemistry
Newmark groundwater (NGW) was used as the background solution. Collected from a natural aquifer underneath the Newmark Civil Engineering Laboratory (205 N. Matthews, Urbana, Illinois, 61801), NGW is greensand-filtered for manganese and iron removal. NGW has been characterized and used in previous studies (Li et al., 2003). Content of total organic carbon (TOC) was measured using a Phoenix 8000 TOC Analyzer (Dohrmann, USA) in previous studies and found to be 2.35 mg/L (Gutierrez et al., 2009). Turbidity of the groundwater and the effluent were 0.25 NTU and 0.70 NTU, respectively, as measured by a Hach Turbidity Meter 43900. Newmark groundwater was chosen over available surface waters for the following reasons: (1) Nearby river water would be impacted by the seasonal runoff from agricultural fields in the surrounding and upstream area, resulting in shifting water chemistry throughout the course of the study. (2) Stagnant water bodies would freeze over in the wintertime, causing inconsistent water collection over the duration of the experiment, which spans multiple months. (3) Newmark groundwater has already been documented and provides a consistent water quality that better suites the longevity of the project. Pasteurized primary wastewater effluent from a local wastewater treatment plant was used to establish a biofilm in the MS2 experiments, as previously suggested by Elliott et al. (2008). Treated wastewater effluent from the same treatment plant was used to conduct rotavirus experiments with high levels (20 mg/L TOC) of natural organic matter (NOM). This wastewater treatment plant uses conventional activated sludge treatment.
2.3.
Column experiments
2.3.1.
Continuous flow test for MS2
Small-scale column tests were conducted to compare results with previous research examining MS2 removal through sand and sand/iron oxide columns (You et al., 2005), and to determine what effect, if any, orientation of iron particles in the sand column has on virus removal. Results were used to determine how household-scale filters would be incorporated with iron as an addition to the sand media. In addition, smallscale columns were used to examine the removal of rotavirus in the proposed design. Due to the difficulty of propagating the virus and the maximum concentrations necessary for use in the BSF, experiments on the household-scale filter with rotavirus were not possible. The glass columns had the following dimensions: 3.8 cm tapered ends, 8.9 cm main body length,
4503
2.5 cm inner diameter, and a total volume between 146.7 cm3 and 161.5 cm3 (Fig. 1). The control column was dry packed with 224.1 g clean quartz sand. The iron column was packed with a 10% iron by volume mixture with sand (30.4 g iron; 182.7 g sand) and a layer of sand only (42 g) at the influent. Zerovalent iron particles (ETI8/50) used in the column studies were obtained from Peerless Metal Powders & Abrasive (Detroit, MI). The iron was used without pretreatment. Sand used in the column studies was an industrial quartz obtained from Unimin Corporation (Le Sueur, MN) described as 5020 (20% retained on 50 mesh or coarser) and 0.15 mm effective size in filtration. The sand was washed with deionized water until the supernatant was no longer cloudy. Each column was flushed using a peristaltic pump for 10 pore volumes (PVs) of NGW at a flow rate of 1.36 mL/min to establish a steady-state flow. Held vertically, the bottom end of the column was used as the influent entrance (Fig. 1). Plug flow condition was verified by conducting tracer tests using 3 pore volumes of a 0.1 M NaCl solution. Each column was then flushed with another 10 pore volumes of NGW. A solution containing w109 PFU/mL of MS2 was introduced and samples were taken in 1.5 mL centrifuge tubes until breakthrough was well-established. Samples were taken from the control column every 5 min for 7 pore volumes, while samples were taken from the iron column every 15 min for 20 pore volumes. The effluent concentrations of MS2 were determined by plaque assay.
2.3.2.
Non-continuous flow test for MS2
To simulate the normal operation of a BSF, 4 additional columns (with the same dimensions and packing procedure as described in Section 2.3.1) were charged daily with 1 pore volume NGW containing w108 PFU/mL of MS2. Primary effluent (PE) obtained from the local wastewater treatment plant, was also added to the solution (2.5% PE) to stimulate biofilm growth at the influent sand surface, located at the
Fig. 1 e (Left) Breakthrough curves of NaCl tracer from columns packed with sand and zerovalent iron. (Right) Diagram of glass columns used for all small-scale experiments. For non-continuous tests, the column was inverted after each daily charge and the influent exposed to air to allow biofilm growth. The sketch for the column is not made to scale.
4504
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 0 1 e4 5 1 0
bottom of the column during charges. The columns were stored between charges in an inverted position, so that the influent entrance was located on top and exposed to air (Fig. 1). Three of the columns were packed with different orientations of iron particles (10% iron by total volume): an even mixture throughout the sand, a middle layer, and a layer at the influent. One column was packed with only sand.
2.3.3.
Non-continuous flow test for rotavirus
Two different columns (sand only and 10% iron by total volume) were then flushed with aquifer water according to the procedures described in Section 2.3.2. Columns were charged with NGW and 2.5% PE for 6 weeks to establish a biofilm. Two sets of experiments were conducted. For the first set, the influent water was Newmark groundwater with 2.35 mg/L TOC. About 30 days after the first set of experiments, the second set of experiments was conducted, for which the influent water was effluent from a local wastewater treatment plant. The treated wastewater effluent was not disinfected and had a TOC of 20 mg/L. For each set of experiments, influent water was seeded with rotavirus (w104 FFU/mL or w105 FFU/mL) and each column was charged with 1 pore volume of solution. After 24 h, effluent samples were collected and determined by rotavirus infectivity tests.
2.4.
Biosand filter experiments
2.4.1.
Filter and media preparation
Two 60 L capacity plastic BSFs were obtained from HydrAid (Grand Rapids, MI). The filters were packed according to the four layer system established by Manz (2007). Each filter contained 5 cm of under drain gravel (6.25e12.5 mm), 5 cm of medium sized support gravel (3.125e6.25 mm) to separate the drainage layer from the filtration sand, and 40 cm of filter media (effective size 0.4 mm) with a 5 cm top layer of fine sand (effective size 0.15 mm). Sand was manually sieved using the appropriate meshes. The effective sand size was similar to that used previously in research and practice (CAWST, 2010; Elliott et al., 2008). One filter was amended with iron by adding 5.54 kg of a mild steel nail (40 mm length, 2 mm diameter, “bright” finish) mixed evenly throughout the top 20 cm of the filter media, excluding the layer of fine sand. The initial maximum flow rates for the unmodified plastic BSF and iron-amended plastic BSF following the first 20 L charges were 0.64 L/min (0.70 m/h) and 0.93 L/min (1.01 m/h), respectively. Every 24 h each filter was charged with 20 L of a solution containing w108 PFU/mL MS2 and 2.5% PE to establish biofilm growth. PE was not added after day 20, when biofilm activity was well-established. Three concrete BSFs were constructed using a steel mold built to specifications (McCarroll, 2009) provided by the Centre for Affordable Water and Sanitation Technology (CAWST) and were representative of BSFs in use worldwide (CAWST, 2010, 2011). Each concrete filter was packed using the same specifications provided for the plastic filter. One filter was amended with iron by adding 5.54 kg of zerovalent iron particles (ETI8/ 50, same as used in column experiments) mixed evenly throughout the top 20 cm of the filter media, excluding the layer of fine sand. Another filter was packed with extra fine
steel wool (Red Devil, Inc. #0000), which was mixed evenly throughout the top 20 cm of filter media. Due to the steel wool’s low weight and large volume, only 0.26 kg of steel wool was used. The initial maximum flow rates for the unmodified concrete BSF, iron-particle-amended concrete BSF, and steelwool-amended BSF following the first 20 L charges were 0.64 mL/min (0.70 m/h), 0.59 mL/min (0.64 m/h), 0.50 mL/min (0.54 m/h), respectively. Every 24 h each filter was charged with 20 L of a solution containing w107 PFU/mL MS2 and 2.5% PE to establish biofilm growth. PE was not added after day 20, when biofilm activity was well-established. X-ray diffraction (XRD) experiments were performed on rusted iron particle samples to determine the mineral composition of the rust. A Rigaku D/Max-b diffractometer with a copper X-ray source controlled by MDI’s DataScan was used. The following parameters were used: 45 kV and 20 mA, the scanning angle range (2q) was 15e80 , and the scanning rate was 0.6 /min with a step increment of 0.05.
2.4.2.
Water chemistry
Influent and effluent samples were collected in 15 mL tubes for analysis of pH, dissolved oxygen (DO) content, alkalinity, NO3, NH4þ, Cl. Samples for MS2 were taken in 1.5 mL sterilized centrifuge tubes and held in a 4 C refrigerator until analysis within 24 h. Trace metal concentrations were determined using inductively coupled plasma mass spectrometry (ICPeMS) with an ELAN Dynamic Reaction Cell instrument (PerkinElmer, Norwalk, USA). Samples were diluted to a total dissolved solid concentration of 0.25%, and the light wavelength intensity from excited atom species was used to determine analyte concentrations. Turbidity of the groundwater and the final effluent was 0.25 and 0.7 NTU as measured by a Hach Turbidity Meter 43900.
3.
Results
3.1. Tracer tests for column experiments with noncontinuous flow Four NaCl breakthrough curves through sand and iron columns with varying orientations of zerovalent iron are plotted in Fig. 1. The three zerovalent iron orientations (10% iron by volume) were an even mixture throughout the sand (‘mixed’), a middle layer (‘band’), and a layer at the influent (‘top’). Independent estimates of the pore volumes (sand only/top placement: 50.0 0.1 mL, mixed/band: 54.9 0.1 mL) were found by measuring the water volume necessary to fill each column completely. All columns had a resulting porosity of 34%. The NaCl tracer response suggests uniform plug flow with a sharp incline in conductivity after one pore volume and a sharp decline after a negative step input was introduced. The Morrill Dispersion Index (MDI) was calculated for all four tracer tests using the method defined by Tchobanoglous et al. (2003) to confirm plug flow. The MDI for each test was approximately 1.4. The USEPA classifies flow with an MDI of less than 2.0 as effective plug flow, while an ideal plug flow reactor has an MDI of 1.0 (USEPA, 1986).
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 0 1 e4 5 1 0
3.2.
Column experiments
3.2.1.
Short term removal using continuous flow (MS2)
The log10 reductions obtained by continuous flow through clean quartz sand with no iron particles and sand mixed with iron (10% iron by volume) are shown in Fig. 2. For the sand with no iron particles, the breakthrough concentration of MS2 was achieved after one pore volume, with approximately half of the MS2 being retained (49.9%). MS2 concentrations in the effluent of the column of sand mixed with iron particles were under the limit of quantification (20 PFU/plate of undiluted sample) until 6 pore volumes continuous flow had passed, with the breakthrough concentration around 5log10 removal (99.999%). In addition, overnight (i.e. 18-h) batch experiments using 15 mL tubes containing 10 mL of solution and 5 mL of iron particles or rusted iron nails revealed complete removal of 4 106 PFU/mL of MS2. Thus, MS2 removal was due to adsorption onto iron oxides. Sorption of MS2 in the iron column most likely occurred due to electrostatic interactions between positively charged iron oxides formed during iron corrosion (e.g. hematite and magnetite) and negatively charged virion particles. A number of studies have also reported the adsorption and inactivation of viruses by iron oxides (Gutierrez et al., 2009; Moore et al., 1981; Rao et al., 1968; Sagripanti et al., 1993). You et al. found complete breakthrough of MS2 by sand columns in a similar study (You et al., 2005). Although 49.9% removal is higher than observed in previous studies, overall removal by the quartz sand was limited. Differences in sand composition may have led to the increased reduction. The quartz sand in this study was cleaned and dried, but it was not treated with a citrate solution to remove existing metal ions and oxides. Trace levels of metal oxides may have been present, leading to heightened MS2 removal. You et al. used a solution containing sodium citrate and citric acid to decrease iron levels in the sand columns from 32.5 mg iron/kg sand to below the detection limit (0.02 mg iron/kg sand) before testing (You et al., 2005). The higher reduction may also be due to the pH of the Newmark groundwater (pH 6.2), which is lower than that of artificial groundwater (AGW; pH 7.5) used in other experiments. The isoelectric point (IEP), the pH at which the surface charge of the virus is neutral or zero, is 3.5e3.9 for MS2
Fig. 2 e Log10 reduction of bacteriophage MS2 in groundwater through clean quartz sand and iron particles (10% by volume) mixed evenly with quartz sand.
4505
(Gutierrez et al., 2009; Overby et al., 1966; Penrod et al., 1995; Zerda and Gerba, 1984). As the pH of the aqueous solution decreases and approaches the IEP, the net negative charge of the MS2 decreases, resulting in the reduction of electrostatic repulsive interactions between the negatively charged virion particles and the negatively charged sand particles (Logsdon et al., 2002). A lower pH in the source water would result in less repulsion and, therefore, greater virus removal by filtration.
3.2.2. Long term removal of MS2 in columns with biofilm growth The long term (72 day) log10 removal for four glass columns simulating a daily, 1 pore volume charged BSF is shown in Fig. 3. While the sand column averaged only 1log10 (90%) removal, all three iron columns had greater than 5log10 removal for the duration of the experiment. Reduction of MS2 by the iron columns ranged from 5log10 to complete removal (>8log10). In comparison, the virus reduction by the sand column ranged from no removal to 3log10 (99.9%). After about 1 week, biofilm growth could be seen in all 4 columns, but removal rates did not increase as the biofilm developed. Biofilm formation occurred only at the influent entrance, which was tapered down significantly from the maximum diameter of the column. Since biofilm formed only on a relatively small area, it made minimal contribution to virus removal. In contrast, Elliot et al. found that the reduction of bacteriophages MS2 and PRD-1 in BSFs increased from 0 to 1.3log10 as the biofilm developed (Elliott et al., 2008). Previous studies have also shown that bacteriophage adsorption (MS2) is reduced in the presence of natural organic matter (NOM) and phosphates that compete for and fill iron oxide adsorption sites (Chiew et al., 2009). Influent and effluent total organic carbon levels for all four columns were measured as an indicator of NOM. The influent concentration of TOC for each column was approximately 3.1 mg/L. Although each iron column retained more TOC (0.5e0.3 mg/L) than the sand column (0.1 mg/L), virus removal was not adversely affected. The extended contact time (24 h) most likely contributed to the increased removal, combating any negative effects caused by the introduction of NOM. It was
Fig. 3 e Log10 reductions of MS2 by three iron columns (different orientations) and one sand-only column.
4506
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 0 1 e4 5 1 0
expected that the iron column with a layer of iron particles at the influent entrance (the “top” column) would experience reduced adsorption because the absence of sand media and established biofilm above it would lead to increased exposure to NOM. However, as discussed previously, relatively small biofilm development occurred. Therefore, no significant differences in removal rates between columns of different iron particle orientations were observed.
3.2.3.
Removal of rotavirus in columns with biofilm growth
For the first set of rotavirus experiments (described in Section 2.3.3), the iron column was charged on two separate days with w104 FFU/mL of rotavirus and obtained removal below the detection limit on each daily charge. The second set of experiments run on the same iron and sand columns used treated wastewater effluent with 10 times higher TOC to evaluate the effect of NOM on virus removal. With the treated wastewater effluent, we obtained 5.2log10 removal of rotavirus by the iron column and 1.1log10 removal by the sand-only column. Note that even with higher TOC, the iron column still allowed substantial rotavirus removal. The rotavirus results are consistent with the MS2 experiments performed in these column studies and rotavirus experiments performed by Gutierrez et al. (2009). In a study using flow-through columns packed with hematite coated glass fibers, it was found that 3 104 FFU of rotavirus was removed per gram of hematite nanoparticles. Furthermore, Gutierrez et al. (2009) found that only 2% of adsorbed rotavirus remained infectious after attachment. Previous research suggests that electrostatic interactions between viruses and iron oxides are so strong that they can cause viruses to disintegrate (Gutierrez et al., 2009; Ryan et al., 2002). With an IEP of 4.5 (Gutierrez et al., 2009), the highly negative potential of rotavirus in source water leads to favorable conditions for viruses to adsorb and be rendered non-infectious.
3.3.
Biosand filter experiments
3.3.1.
Head-loss development over time
As the BSF accumulates bacteria and organic particulate matter to develop a biofilm, the head-loss of the system increases, and users will notice the effluent flow rate decreasing over time as the BSF is used repeatedly. The buildup of organic matter and development of the biological layers (known as “ripening”) results in a more effective filter (Elliott et al., 2008; Stauber, 2006). The flow rates for the two plastic filters are plotted in Fig. 4. To promote biofilm growth, 2.5% primary effluent (PE) was added to both filters until day 20. The BSF without iron nails behaved as expected: as time progressed and the filter ripened, the flow rate decreased and removal efficacy increased. The iron nail filter, however, behaved differently. The flow rate did not decrease as expected, indicating a problem in establishing uniform flow across the entire cross-sectional area of the filter. Although the filter with iron nails distributed evenly throughout the sand established uniform flow early in the experiment, independent flow paths eventually developed. Water bypassed the iron nails at an elevated flow rate, despite the continued growth of biofilm.
Fig. 4 e Flow rates for sand BSF (plastic), iron nail BSF (plastic), steel wool BSF (concrete), and iron particle BSF (concrete).
Fig. 4 also shows the flow rate of the concrete filter packed with iron particles compared to the flow rate of the plastic BSF packed with iron nails. Flow rate problems were effectively eliminated by using an iron material closer in size to the filtration sand media grain size, and the flow rate for the iron particle BSF behaved as desired. As organic matter accumulated, the filter ripened and the flow rate decreased. The steel wool filter displayed similar results.
3.3.2.
Water characteristics
Analytical parameters including pH, DO, alkalinity, and NO3 remained consistent throughout each BSF experiment. Influent and effluent pH was between 7.1 and 8.2 with an average alkalinity of 22.5 mg/L as CaCO3. DO was reduced in the filters but always remained above 1.01 mg/L. Iron oxides formed during iron corrosion were filtered out by the sand media, and iron was not present in the effluent (ICPeMS detection limit of 0.0059 mg/L). Leaching of other trace metals from the iron sources was not seen, and heavy metals such as chromium, lead, cadmium, and arsenic were not present in the effluent.
3.3.3.
MS2 removal
Results for the removal of MS2 through the sand/iron nail BSF are shown in Fig. 5. The iron nail filter started with a 6.5log10 (99.99997%) removal, but adsorption of virus particles quickly declined as flow paths short-circuited around the iron nails. Subsequent filter charges did not achieve plug flow, causing
Fig. 5 e MS2 reduction in the BSF containing iron nails.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 0 1 e4 5 1 0
reduced contact time with iron particles. After 10 days, removal for the iron nail BSF was less than 4log10 (99.99%), and successive samples were between 3 and 4log10 (99.9e99.99%). These results indicate that the iron nails used are not appropriate for use in the BSF; smaller iron material must be used to obtain a more evenly graded media that will promote uniform flow and not detract from the original (without iron) efficacy of the BSF. The flow paths created also highlight the importance of contact time, a point further illustrated in Fig. 6. Water that has been allowed to sit overnight in the filter (24 h) achieves additional removal of w1log10 (90%). The concrete sand filter achieved between 4log10 removal and no removal, with an average of about 2log10 (99%) removal (Fig. 7) for the first 150 days. In comparison, the two filters packed with iron (particles and steel wool) achieved between 7log10 (99.99999%) and >5log10 removal, with an average of 6log10 (99.9999%) reduction of virion particles for the first 170 days of filter use. The iron particles and steel wool resulted in filters with a more even media grade than in the filters containing nails. This promoted uniform, steady flow throughout the entire cross-sectional area of the filter, allowing sufficient contact time between each filter charge and the iron media. The corresponding increase in MS2 removal shows that ironamended BSFs can effectively remove select viruses, with removal rates that exceed U.S. Environmental Protection Agency (USEPA) drinking standards (4log10 removal) (USEPA, 1991). Iron samples from the iron particle BSF were taken for XRD analysis after the study ended. Iron particle composition after corrosion was 48% iron, 40% magnetite, 9% hematite, and 3% other. With 48% of the particles still present in the form of iron after 376 days of operation, further iron oxide generation was possible. However, removal in the steel wool filter decreased significantly after day 170, dropping to an average of 2log10 removal. XRD analysis revealed that the steel wool present in the steel wool BSF was completely oxidized. This indicates that virus sorption to the steel wool stopped once its adsorption sites were exhausted, and remaining removal of viruses was due to the sand media. The reduction in removal capacity to levels below those observed in the sand-only filter may indicate that the addition of steel wool to the sand media adversely affected the traditional filter’s removal efficacy. The sand-only BSF saw a steady increase in removal from 2log10 to >4log10. Previous research has shown that biofilm ripening and media aging contribute significantly to the MS2
Fig. 6 e Comparison of MS2 reduction at two different pause periods for the BSF containing iron nails.
4507
Fig. 7 e Long term MS2 reduction by three different concrete filters: sand, steel wool, and iron particles.
removal capacity of the traditional sand BSF (Elliott et al., 2008), but long term studies (>2 months) have not been previously conducted. However, it is well-known that filter ripening plays a crucial role in particle removal during granular filtration (Kim et al., 2008; Kretzschmar et al., 1997; Tobiason and Omelia, 1988). The results from this study suggest that traditional BSFs may experience significant virus removal (>2log10) once filters have been in use for several months. If this is the case, filters would only need to be amended with an appropriate iron source, preferably zerovalent iron particles, for the initial ripening period.
4.
Discussion
Previous testing in small-scale columns has shown that MS2 has minimal sorption to sand media. In particular, studies have demonstrated lower adsorption of MS2 when compared to bacteriophages ( phiX-174) and human pathogenic viruses (rotavirus, echovirus-12, and poliovirus). Data from smallscale saturated flow studies showed that MS2 had no sorption compared to phiX-174 with about 80% removal (Jin et al., 1997). Higher removal of phiX-174 was attributed to a higher isoelectric point (IEP ¼ 6.6) than that of MS2 (IEP ¼ 3.5e3.9). Virus sorption is largely governed by electrostatic interactions and van der Waal’s forces. Because quartz sand is negatively charged in source water (IEP ¼ w2.2), and MS2 is more negatively charged than phiX-174, MS2 experiences higher repulsion at pH ¼ 7. Bales et al. (1991, 1993) also demonstrated that phiX-174 and poliovirus exhibit similar removal through sand columns due to their shared IEP of 6.6. Additionally, Bales et al. (1993) showed that poliovirus sorbed to silica beads much more readily than MS2. Small-scale studies with sand columns have also shown higher adsorption of rotavirus (IEP ¼ 4.5) than MS2 (Gutierrez et al., 2009). It is also important to note that hydrophobic interactions may be important for bacteriophage adsorption to hydrophobic surface due to the presence of hydrophobic areas on the surface of bacteriophage such as MS2 and PRD-1 (Bales et al., 1991). In large-scale studies using BSFs similar to those used in this study, Elliott et al. (2008) saw greater removal of echovirus-12 (>2log) than bacteriophages MS2 and PRD-1
4508
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 0 1 e4 5 1 0
(w1log). These differences were also attributed to echovirus12 having a higher IEP (echoviruses have IEP of 5e6.4) than MS2 and PRD-1, which have similar IEPs of 3.5e3.9 and 4.2, respectively (Michen and Graule, 2010). When positively charged iron oxides are added throughout the sand media, it is expected that all viruses that are negatively charged at pH ¼ 7 will be able to readily adsorb to iron oxide surfaces. At the same time, MS2 experiences high repulsion from the sand particles due to its low isoelectric point, and based on past studies, is a more difficult virus, compared to echovirus type 12, to remove through conventional sand filtration (Elliott et al., 2008). Having a low IEP would make a virus relatively negative at pHs near neutral. This charge leads to less removal in a sand-only BSF, due to the repulsion of negatively charged sand particles, and greater removal in an iron-amended BSF, due to stronger electrostatic attraction to positively charged iron oxides. While MS2 bacteriophage can be used as a surrogate for rotavirus because these two virus have similar interfacial properties (Brady-Estevez et al., 2010; Gutierrez et al., 2009, 2010; Mylon et al., 2010), other enteric viruses may require different surrogates. When a virus attaches to particles in water, the overall size of the aggregate is greater than that of a monodispersed virus. As a result, virus-particle aggregates would be easier to remove by filtration (Kretzschmar et al., 1999; Petosa et al., 2010). Experiments presented here were conducted as a worst case scenario for monodispersed viruses. In addition, a recent publication showed that norovirus, which has a similar size as the studied virus MS2, associated with 0.45e180 mm particles, and attachment and settling was not an effective removal mechanism for viruses in waste stabilization ponds (Da Silva et al., 2008). Similarly, recent study with bacteriophage MS2 and rotavirus has found that steric repulsion prevents viruses from aggregating and adsorbing to silica and organic matter surfaces (Gutierrez et al., 2010; Mylon et al., 2010). It is expected that biofilm development is highly dependent on source water characteristics, including concentration and identity of native microorganisms that attach to the sand, nutrients from which the microbes derive an energy source, NOM that could shelter or cover the microbes, DO levels, and even water temperature. In our household-scale studies, MS2 removal in the sand-only BSF increased with biofilm development in a trend that matches previous studies (Elliott et al., 2008), despite the utilization of different water sources. While investigating biofilms developed from different water sources was not a focus during this phase of the research, the currently on-going research involves collecting data from BSFs installed and operating at different locations in Guatemala.
5.
Conclusions
The following conclusions resulting from this study indicate a potential advancement in household water treatment technologies by amending the BSF with iron materials: Both MS2 and rotavirus were treated to USEPA standards for virus removal, greater than 99.99% removal, through the
adsorption to positively charged iron oxides in small-scale studies. Untreated iron material distributed uniformly in sand media of a BSF will oxidize and effectively remove >4log10 MS2 and rotavirus from a natural source water. The duration of effective virus removal by iron-amended biosand filtration depends on both source water conditions and quantity and composition of iron material added, all of which need to be researched further. After 200 days in operation, the unmodified BSF underwent significant ripening and was able to provide more than 4log10 of MS2 virus removal. Further research is needed to determine the effects of competitive adsorption with other water constituents on the efficacy of virus removal by iron-amended biosand filtration. For relatively little cost (approximately 4 USD per filter based on initial cost estimates), BSFs can be amended with local iron materials, thereby providing a substantially improved barrier against waterborne viruses and, hopefully, bringing relief to millions of lives in the process.
Acknowledgments This work was partially supported by the Center of Advanced Materials for the Purification of Water with Systems (WaterCAMPWS), a Science and Technology Center under the National Science Foundation (NSF) Award No. CTS-0120978, under the United States Environmental Protection Agency (USEPA) People, Prosperity, and the Planet (P3) Phase 1 (SU834296) and Phase 2 (SU834754) grants, the NSF Career grant (0954501) to THN, and the University of Illinois College of Engineering International Programs in Engineering (IPENG). Leo Gutierrez and Ofelia Romero were acknowledged for preparing MS2 and rotavirus stocks and assays. BSFs were run daily by IB and AS with the assistance of Kevin Swanson and other members of the Engineers Without Borders (EWB) University of Illinois at Urbana-Champaign Chapter.
references
Bales, R.C., Hinkle, S.R., Kroeger, T.W., Stocking, K., Gerba, C. P., 1991. Bacteriophage adsorption during transport through porous-media e chemical perturbations and reversibility. Environmental Science & Technology 25 (12), 2088e2095. Bales, R.C., Li, S.M., Maguire, K.M., Yahya, M.T., Gerba, C.P., 1993. MS-2 and poliovirus transport in porous-media e hydrophobic effects and chemical perturbations. Water Resources Research 29 (4), 957e963. Baumgartner, J., Murcott, S., Ezzati, M., 2007. Reconsidering ‘appropriate technology’: the effects of operating conditions on the bacterial removal performance of two household drinking-water filter systems. Environmental Research Letters 2 (2), 6. Brady-Estevez, A.S., Nguyen, T.H., Gutierrez, L., Elimelech, M., 2010. Impact of solution chemistry on viral removal by a single-walled carbon nanotube filter. Water Research 44 (13), 3773e3780.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 0 1 e4 5 1 0
Buzunis, B.J., 1995. Intermittently Operated Slow Sand Filtration: A New Water Treatment Process. University of Calgary, Calgary. CAWST, 2010. Biosand Filter Manual: Design, Construction, Installation, Operation, and Maintenance. Centre for Affordable Water and Sanitation Technology, Calgary, Alberta, Canada, p. 59. CAWST, 2011. About Us. http://www.cawst.org/en/about-us. Chiew, H., Sampson, M.L., Huch, S., Ken, S., Bostick, B.C., 2009. Effect of groundwater iron and phosphate on the efficacy of arsenic removal by iron-amended bios and filters. Environmental Science & Technology 43 (16), 6295e6300. Clasen, T., Schmidt, W.P., Rabie, T., Roberts, I., Cairncross, S., 2007. Interventions to improve water quality for preventing diarrhoea: systematic review and meta-analysis. British Medical Journal 334, 7597e7782. Curtis, V., Cairncross, S., Yonli, R., 2000. Review: domestic hygiene and diarrhoea e pinpointing the problem. Tropical Medicine & International Health 5 (1), 22e32. Da Silva, A.K., Le Guyader, F.S., Le Saux, J.C., Pommepuy, M., Montgomery, M.A., Elimelech, M., 2008. Norovirus removal and particle association in a waste stabilization pond. Environmental Science & Technology 42 (24), 9151e9157. Elliott, M.A., 2010. Mechanisms of Microbial Reduction and Implications for Design and Operation of the Biosand Water Filter PhD thesis. University of North Carolina at Chapel Hill, Chapel Hill. Elliott, M.A., Stauber, C.E., Koksal, F., DiGiano, F.A., Sobsey, M.D., 2008. Reductions of E. coli, echovirus type 12 and bacteriophages in an intermittently operated household-scale slow sand filter. Water Research 42 (10e11), 2662e2670. Esrey, S.A., Feachem, R.G., Hughes, J.M., 1985. Interventions for the control of diarrhoeal diseases among young children: improving water supplies and excreta disposal facilities. Bulletin of the World Health Organization 63 (4), 757e772. Esrey, S.A., Potash, J.B., Roberts, L., Shiff, C., 1991. Effects of improved water supply and sanitation on ascariasis, diarrhoea, dracunculiasis, hookworm infection, schistosomiasis, and trachoma. Bulletin of the World Health Organization 69 (5), 609e621. Fewtrell, L., Kaufmann, R.B., Kay, D., Enanoria, W., Haller, L., Colford Jr., J.M., 2005. Water, sanitation, and hygiene interventions to reduce diarrhoea in less developed countries: a systematic review and meta-analysis. Lancet Infectious Diseases 5 (1), 42e52. Gutierrez, L., Li, X., Wang, J.W., Nangmenyi, G., Economy, J., Kuhlenschmidt, T.B., Kuhlenschmidt, M.S., Nguyen, T.H., 2009. Adsorption of rotavirus and bacteriophage MS2 using glass fiber coated with hematite nanoparticles. Water Research 43 (20), 5198e5208. Gutierrez, L., Mylon, S.E., Nash, B., Nguyen, T.H., 2010. Deposition and aggregation kinetics of rotavirus in divalent cation solutions. Environmental Science & Technology 44 (12), 4552e4557. Jin, Y., Yates, M.V., Thompson, S.S., Jury, W.A., 1997. Sorption of viruses during flow through saturated sand columns. Environmental Science & Technology, 548e555. Jin, Y., Chu, Y., Li, Y., 2000. Virus removal and transport in saturated and unsaturated sand columns. Journal of Contaminant Hydrology 43 (2), 111e128. Kim, J., Nason, J.A., Lawler, D.F., 2008. Influence of surface charge distributions and particle size distributions on particle attachment in granular media filtration. Environmental Science & Technology 42 (7), 2557e2562. Kitis, M., Lozier, J.C., Kim, J.H., Mi, B.X., Marinas, B.J., 2003. Microbial removal and integrity monitoring of RO and NF membranes. Journal of American Water Works Association 95 (12), 105e119. Kretzschmar, R., Barmettler, K., Grolimund, D., Yan, Y.D., Borkovec, M., Sticher, H., 1997. Experimental determination of
4509
colloid deposition rates and collision efficiencies in natural porous media. Water Resources Research 33 (5), 1129e1137. Kretzschmar, R., Borkovec, M., Grolimund, D., Elimelech, M., 1999. Advances in Agronomy, vol. 66. Academic Press Inc, San Diego, pp. 121e193. Li, Q., Snoeyink, V., Marinas, B., Campos, C., 2003. Elucidating competitive adsorption mechanisms of atrazine and NOM using model compounds. Water Research 37, 773e784. Logsdon, G.S., Kohne, R., Abel, S., LaBonde, S., 2002. Slow sand filtration for small water systems. Journal of Environmental Engineering & Science 1 (5), 339e348. Lukasik, J., Cheng, Y.F., Lu, F.H., Tamplin, M., Farrah, S.R., 1999. Removal of microorganisms from water by columns containing sand coated with ferric and aluminum hydroxides. Water Research 33 (3), 769e777. Manz, D.D.H., 2007. Preparation of Media for the BioSand Water Filter: Four Layer System. Manz, D.H., Buzunis, B., Morales, C., 1993. Nicaragua Household Water Supply and Testing Project. University of Calgary, Calgary, p. 10. McCarroll, R. (Ed.), 2009. The All Natural Biosand Filter Manual: Design, Construction, and Installation. The Centre for Affordable Water and Sanitation Technologies, Calgary, Alberta, Canada. Michen, B., Graule, T., 2010. Isoelectric points of viruses. Journal of Applied Microbiology 109 (2), 388e397. Moore, E.S., Tylor, D.H., Sturman, L.S., Reddy, M.M., 1981. Poliovirus adsorption by 34 minerals and soils. Applied and Environmental Microbiology 42, 963e975. Mylon, S.E., Rinciog, C.I., Schmidt, N., Gutierrez, L., Wong, G.C.L., Nguyen, T.H., 2010. Influence of salts and natural organic matter on the stability of bacteriophage MS2. Langmuir 26 (2), 1035e1042. Nelson, K.L., Murray, A., 2008. Sanitation for unserved populations: technologies, implementation challenges, and opportunities. Annual Review of Environment and Resources 33, 119e151. Ngai, T.K.K., Shrestha, R.R., Dangol, B., Maharjan, M., Murcott, S.E., 2007. Design for sustainable development e household drinking water filter for arsenic and pathogen treatment in Nepal. Journal of Environmental Science and Health e Part A Toxic/Hazardous Substances and Environmental Engineering 42 (12), 1879e1888. Overby, L.R., Barlow, G.H., Doi, R.H., Jacob, M., Spiegel, S., 1966. Comparison of two serologically distinct ribonucleic acid bacteriophages. Journal of Bacteriology 91 (1), 442e448. Page, M.A., Shisler, J.L., Marinas, B.J., 2009. Kinetics of adenovirus type 2 inactivation with free chlorine. Water Research 43 (11), 2916e2926. Palmateer, G., Manz, D., Jurkovic, A., McInnis, R., Unger, S., Kwan, K.K., Dutka, B.J., 1999. Toxicant and parasite challenge of Manz intermittent slow sand filter. Environmental Toxicology 14 (2), 217e225. Parashar, U.D., Gibson, C.J., Bresee, J.S., Glass, R.I., 2006. Rotavirus and severe childhood diarrhea. Emerging Infectious Diseases 12 (2), 304e306. Penrod, S.L., Olson, T.M., Grant, S.B., 1995. Whole particle microelectrophoresis for small viruses. Journal of Colloid and Interface Science 173 (2), 521e523. Petosa, A.R., Jaisi, D.P., Quevedo, I.R., Elimelech, M., Tufenkji, N., 2010. Aggregation and deposition of engineered nanomaterials in aquatic environments: role of physicochemical interactions. Environmental Science & Technology 44 (17), 6532e6549. Ramani, S., Kang, G., 2009. Viruses causing childhood diarrhoea in the developing world. Current Opinion in Infectious Diseases 22 (5), 477e482.
4510
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 0 1 e4 5 1 0
Rao, V.C., Sullivan, R., Read, R.B., Clarke, N.A., 1968. A simple method for concentrating and detecting viruses in water. Journal of American Water Works Association 60, 1288e1294. Rolsma, M.D., Gelberg, H.B., Kuhlenschmidt, M.S., 1994. Assay for evaluation of rotavirusecell interactions e identification of an enterocyte ganglioside fraction that mediates group-A porcine rotavirus recognition. Journal of Virology 68 (1), 258e268. Rolsma, M.D., Kuhlenschmidt, T.B., Gelberg, H.B., Kuhlenschmidt, M.S., 1998. Structure and function of a ganglioside receptor for porcine rotavirus. Journal of Virology 72 (11), 9079e9091. Ruiz, M., Charpilienne, A., Liprandi, F., Gajardo, R., Michelangeli, F., Cohen, J., 1996. The concentration of Ca21 that solubilizes outer capsid proteins from rotavirus particles is dependent on the strain. Journal of Virology 70 (8), 4877e4883. Ryan, J.N., Elimelech, M., Ard, R.A., Harvey, R.W., Johnson, P.R., 1999. Bacteriophage PRD1 and silica colloid transport and recovery in an iron oxide-coated sand aquifer. Environmental Science & Technology 33 (1), 63e73. Ryan, J.R.H., Metge, D., Elimelech, M., Pieper, A., 2002. Field and laboratory investigations of inactivation of viruses (PRD1 and MS2) attached to iron oxide-coated quartz sand. Environmental Science & Technology 36, 2403e2413. Sagripanti, J.L., Routson, L.B., Lytle, C.D., 1993. Virus inactivation by copper or iron ions alone and in the presence of peroxide. Applied and Environmental Microbiology 59, 4374e4376. Sirikanchana, K., Shisler, J.L., Marinas, B.J., 2008. Effect of exposure to UV-C irradiation and monochloramine on adenovirus serotype 2 early protein expression and DNA replication. Applied and Environmental Microbiology 74 (12), 3774e3782. Sobsey, M.D., 2002. Managing Water in the Home: Accelerated Health Gains from Improved Water Supply, Geneva. Sobsey, M.D., Stauber, C.E., Casanova, L.M., Brown, J.M., Elliott, M. A., 2008. Point of use household drinking water filtration: a practical, effective solution for providing sustained access to
safe drinking water in the developing world. Environmental Science & Technology 42 (12), 4261e4267. Stauber, C.E., 2006. Characterisation of the biosand filter for E. coli reductions from household drinking water under controlled laboratory and field use conditions. Water Science and Technology 54 (3), 1e7. Tchobanoglous, G., Burton, F.L., Stensel, H.D., 2003. Wastewater Engineering: Treatment and Reuse. McGraw-Hill Higher Education, New York. Tiwari, S.S.K., Schmidt, W.P., Darby, J., Kariuki, Z.G., Jenkins, M. W., 2009. Intermittent slow sand filtration for preventing diarrhoea among children in Kenyan households using unimproved water sources: randomized controlled trial. Tropical Medicine and International Health 14 (11), 1374e1382. Tobiason, J.E., Omelia, C.R., 1988. Physicochemical aspects of particle removal in depth filtration. Journal of American Water Works Association 80 (12), 54e64. UNICEF/WHO, 2008. Progress on Drinking Water and Sanitation: Special Focus on Sanitation. USEPA, 1986. Design Manual, Municipal Wastewater Disinfection, Cincinnati, OH. USEPA, 1991. Guidance Manual for Compliance with the Filtration and Disinfection Requirements for Public Water Systems Using Surface Water Sources. Prepared for the Office of Drinking Water, Science and Technology Branch, Criteria and Standards Division, Washington, DC. Weber-Shirk, M.L., Dick, R.D., 1997. Biological mechanisms in slow sand filters. Journal of American Water Works Association 89 (2), 72e83. WHO, 2008. Safer Water, Better Health: Costs, Benefits, and Sustainability of Interventions to Protect and Promote Health. You, Y., Han, J., Chiu, P.C., Jin, Y., 2005. Removal and inactivation of waterborne viruses using zerovalent iron. Environmental Science & Technology 39 (23), 9263. Zerda, K.S., Gerba, C.P., 1984. Agarose isoelectrofocusing of intact virions. Journal of Virological Methods 9 (1), 1e6.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 1 1 e4 5 2 1
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
A new dynamic model for bioavailability and cometabolism of micropollutants during anaerobic digestion Liliana Delgadillo-Mirquez*, Laurent Lardon, Jean-Philippe Steyer, Dominique Patureau INRA, UR050, Laboratoire de Biotechnologie de l’Environnement, Avenue des Etangs, Narbonne F-11100, France
article info
abstract
Article history:
Organic micropollutants (OMPs) are present in wastewater and sludge. Their possible
Received 8 February 2011
impact to the environment contributes to their increasing scientific and social interest.
Received in revised form
Anaerobic digestion has been shown as a potential biological process for removal of these
25 May 2011
compounds. An accurate description of OMP distribution in the environmental system can
Accepted 31 May 2011
be used to better understand which compartment is used for degradation and to improve
Available online 12 June 2011
their depletion in conventional wastewater treatment technologies. In this work, we proposed a dynamical model with a four-compartment distribution to describe the Poly-
Keywords:
cyclic Aromatic Hydrocarbons (PAHs) fate during anaerobic digestion. The model is cali-
Biodegradation
brated and validated using experimental data obtained from two continuous reactors fed
Methanogenic conditions
with primary and secondary sludge operated under mesophilic conditions. A non-linear
PAHs
least square method was used to optimize the model parameters. The resulted model is
Sorption
in accordance with the experimental data. The PAH biodegradation rate is well modeled
Xenobiotic
when considering the aqueous fraction (including free and sorbed to dissolved/colloidal matter PAHs) as the bioavailable compartment. It was also demonstrated in the simulations that the PAHs biodegradation is linked to a mechanism of cometabolism. The model proposed is potentially useful to better understand the micropollutant distribution, predict the fate of PAHs under anaerobic condition and help to optimize the operation process for their depletion. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
Organic micropollutants (OMPs) have become an important environmental topic in recent years due to the risk they pose on aquatic environment and on human health e.g. endocrine disrupting effects (Press-Kristensen et al., 2007; Couillard et al., 2008) and to the development of highly accurate analytical methodologies with lower detection limits (Trably et al., 2004). OMPs are frequently detected in different environmental compartments (rivers, lakes, groundwaters,
sediments, wastewaters, drinking waters) at low concentration (ng to mg/L and mg to mg/kg dry matter). In wastewater treatment plants (WWTPs), OMPs are partially removed by abiotic and biotic processes, including volatilization, stripping, sorption to sludge and biological and/or chemical transformation (Alder et al., 1997; Byrns, 2001; Lindblom et al., 2009). However, the conventional treatment technologies have not been specifically designed for removing OMPs but they can reduce OMPs concentrations as well as their potential environmental impact. Furthermore, these removals are
* Corresponding author. E-mail addresses:
[email protected] (L. Delgadillo-Mirquez),
[email protected] (L. Lardon),
[email protected] (J.-P. Steyer),
[email protected] (D. Patureau). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.05.047
4512
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 1 1 e4 5 2 1
dependant of the OMPs physico-chemical properties, the sludge characteristics and the WWTPs operational conditions (Clara et al., 2005; Dionisi et al., 2006; Joss et al., 2006). Several mathematical models described the fate and the distribution of OMPs between the aqueous and the solid phase (Byrns, 2001; Dionisi et al., 2006; Joss et al., 2004, 2006; Lindblom et al., 2009; Plo´sz et al., 2010). In general, these models used the solidewater partition coefficient to describe equilibrium condition and assumed that the aqueous phase is available for microbial biodegradation activity and the solid phase is bioaccessible and can be transferred to aqueous phase during the process (Artola-Garicano et al., 2003). Nevertheless, Barret et al. (2010b) have demonstrated that the sorption phenomena also occur onto the aqueous phase containing dissolved and colloidal matter. In this study, the sludge has been considered as a three-compartment system with two equilibrium constants. The presence of this third compartment can thus influence the distribution and the pollutants bioavailability. In fact, the distribution in three compartments can help to find the real bioavailable fraction of OMPs. Furthermore, sorption onto the solid phase and dissolved/colloidal matter of sludge is a very fast mechanism in comparison with biological anaerobic kinetics (Chang et al., 2003; Dionisi et al., 2006; Barret et al., 2010d). The bioavailability is influenced by a variety of factors including (i) sorption-desorption processes that could be ratelimiting for biodegradation, (ii) irreversibility or sequestration phenomena due to physical and/or chemical interactions and (iii) presence of other compounds that might compete for sorption sites. Moreover, the biodegradation of multiple substrates can also take place the cometabolism process i.e. a compound of interest does not function as a growth substrate (Criddle, 1993). Previous publications have shown the cometabolism as a mechanism approach in the transformation of some recalcitrant contaminants (Chang et al., 1993, 2003; Criddle, 1993; Tiehm and Fritzsche, 1995; Yuan et al., 2001; Clara et al., 2005; Plo´sz et al., 2010; Barret et al., 2010c). This study aimed to propose, analyze and validate a dynamic model for the Polycyclic Aromatic Hydrocarbons (PAHs) fate under anaerobic condition considering sludge as a four-compartment system. To this end, two hypotheses were evaluated. The first hypothesis consists in modeling the PAHs biodegradation kinetics with a cometabolism kinetics and to compare it with a Monod-type kinetics. The second hypothesis tests which one of the compartments is really available for degradation: the free dissolved one, the aqueous one or the sum of all compartments. This approach should improve the prediction of PAHs distribution, bioavailability and biodegradation.
2.
Material and methods
2.1.
Sludge source
All experiments were performed using activated sludge from an urban wastewater treatment plant. The primary sludge sample (PS) was collected at the outlet of a primary settling tank of a domestic wastewater treatment plant treating 33,000 PE (Person Equivalent). The secondary sludge sample (SS) came from the biological aerobic unit of another domestic plant treating 250,000 PE with a very low hydraulic retention time (0.4 day). Prior to their direct use, PS and SS were stored at 20 C. All these samples were finally diluted with deionized water to reach 24 5 gCOD/L and spiked at 5 mg/gDM for each PAH except for indeno(1,2,3,c,d)pyrene (1 mg/gDM). Table 1 shows the main characteristics of these primary and secondary sludge.
2.2.
Micropollutants
Polycyclic Aromatic Hydrocarbons (PAHs) were selected as model micropolluants (Table 2). All solvents were purchased from J.T.Baker. Mixtures are indicated in volume percentage. PAH powders were obtained from Dr Ehrenstorfer GmbH. Each PAH was dissolved in dichloromethane at 1 g/L. The spiking mix was prepared from these individual concentrated solutions, adding 5 mL of each, evaporating solvent under gentle nitrogen flow and dissolving in 50 mL of acetonitrile. Final concentrations were 100 mg/L for each PAH. The 10 mg/L standard solution of PAHs in acetonitrile was provided by Dr Ehrenstorfer GmbH. Dilution factors from 10 to 1000 were applied to obtain 6 calibration levels. Standards were stored at 20 C.
2.3.
Experimental set-up
Two continuous reactors have been operated at a constant organic load of 1.2 0.2 gCOD/L.d and a hydraulic retention time of 20 days. Temperature was regulated at 35 C using hot water circulation in the external jacket. The reactors were fed with primary (PS) and secondary sludge (SS). The feed was stored at 4 C. Six times a day, it was pumped into the reactor just after pumping out the digested sludge, collected in tanks at 4 C. For the start-up, reactors were filled with methanogenic sludge coming from an anaerobic mesophilic reactor adapted to PAHs-polluted sludge, and directly fed at the normal operating conditions. The pH and the biogas volumetric production were monitored on line. Seven days
Table 1 e Primary (PS) and secondary sludge (SS) characteristics (average value and standard deviation from 5 measurements performed at steady state). Sludge
PS SS
COD
DM
DCM
Proteins
Carbohydrates
Lipids
VFA
gO2/L
gDM/L
% of total DM
geqBSA/gDM
geqGlu/gDM
gPEEM/gDM
gVFA/gDM
28 4 23 2
22 2 19 1
51 44
0.27 0.03 0.25 0.02
0.29 0.09 0.30 0.10
0.13 0.05 0.10 0.04
0.03 0.02 0.04 0.04
Source: Barret et al., 2010d. Chemical oxygen demand (COD), dry matter (DM), dissolved/colloidal matter (DCM) and volatile fatty acids (VFAs).
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 1 1 e4 5 2 1
Table 2 e Physicochemical characteristics of the PAHs. Kp (mL/gCODLpart) and KDCM(mL/gCODLDCM), equilibrium constants of sorption determined from a threecompartment methodology (Barret et al., 2010b) for PS and SS. PAH
M (g/mol)
Fluorene Phenanthrene Anthracene Fluoranthene Pyrene Benzo(a)anthracene Chrysene Benzo(b)fluoranthene Benzo(k)fluoranthene Benzo(a)pyrene Dibenzo(a,h)anthracene Benzo(g,h,i)perylene Indeno(1,2,3,c,d)pyrene
166 178 178 202 202 228 228 252 252 252 278 276 276
PS
SS
log Kp
log KDCM
log Kp
log KDCM
0.098 0.398 0.398 0.398 0.698 0.798 0.898 0.898 0.998 0.898 1.198 1.098 0.898
0.481 0.681 0.681 0.781 0.981 1.281 1.281 1.381 1.281 1.281 1.481 1.481 1.681
0.283 0.683 0.583 0.603 0.883 0.983 1.083 1.083 1.183 1.083 1.383 1.283 1.083
0.902 1.102 1.102 1.202 1.302 1.302 1.502 1.602 1.802 1.702 1.902 1.902 1.702
composite samples were taken once a week from the feed tank, the outlet tank and the gaseous phase.
2.4.
Analytical methods
Inlet and outlet composite samples were analyzed for their chemical oxygen demand (COD) in both soluble and particulate fraction, dry matter (DM), organic matter (OM), proteins, carbohydrates, organic carbon in particles (POC) and in dissolved/colloidal matter (DCOC) and volatile fatty acids (VFAs), according to Barret et al. (2010a). The percentage of methane (CH4) and carbon dioxide (CO2) in the biogas were measured using a gas chromatograph (Shimadzu GC-8A), with argon as the carrier gas and equipped with a thermal conductivity detector. PAHs were extracted from the ORBO cartridge using a Soxhlet setup, operated during 16 h at 60 C, with 200 mL of hexane/acetone (50:50 v:v). Extraction from inlet and outlet sludge samples were performed according to Trably et al. (2004). Equilibrium constants (Table 2) of sorption to particles (Kp) and to dissolved/colloidal matter (KDCM) were determined according to the experimental methodology proposed by Barret et al. (2010b) for the thirteen PAHs and the two sludge (PS and SS).
2.5.
4513
distributions of these compounds influence their bioavailability and it may be the main limiting factor of OMP biodegradation in sludge. OMPs are assumed to be able to sorb onto either particles (P) or dissolved/colloidal matter (DCM) (Barret et al., 2010b). In the model, OMPs is thus assumed to be distributed between four physical compartments (Fig. 1): the free dissolved (Cf, mg/L), the gas (Cg, mg/L), the sorbed to DCM (cDCM, mg/gCOD-DCM) and the sorbed to particles (cp, mg/gCOD-p) compartments. At equilibrium, the four-compartment system can be described by the three following equations: Kp ¼
cp Cf
KDCM ¼
(1) cDCM Cf
(2)
and KH ¼
Cg Cf
(3)
where Kp is the equilibrium constant of OMPs sorption to particle (L/gCOD-p), KDCM is the equilibrium constant of OMPs sorption to DCM (L/gCOD-DCM) and Henry constant (KH, dimensionless) describes the equilibrium between gas phase and free dissolved concentration of OMPs. The total mass of OMP can be expressed as total liquid concentration (Ct,liq, mg/L) and gas concentration (Cg, mg/L) of OMP: Ct;liq ¼ Cf þ cp Sp þ cDCM SS
(4)
Cg ¼ KH Cf
(5)
where Sp is the particulate substrate (particulate concentration, gCOD-p/L) and SS is the soluble substrate (dissolved and colloidal concentration, gCOD-DCM/L). Thus, based on the experimental measurement of total liquid concentration and on Equations (1), (2) and (4), the concentrations in the different compartments and initial condition can be estimated from Equations (6)e(8) for each
Simulation software
Simulations presented in this work have been developed in MatLab-Simulink. Optimization toolbox for solving nonlinear least square problems has been used to estimate the model parameters.
3.
Model description and assumptions
3.1.
The four-compartment model
The model describes the physical exchanges of PAHs between compartments in the reactor. Hence, the physical
Fig. 1 e Representation of the four-compartment model of an OMP.
4514
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 1 1 e4 5 2 1
Table 3 e Matrix representation the fate of OMP in the four-compartment model.
a
pollutant. Indeed, the Table 2 shows the equilibrium constants Kp and KDCM calculated for the thirteen PAHs and the two sludge (PS and SS), according to previously developed methodology (Barret et al., 2010b). Cf ¼
Ct;liq 1 þ Kp Sp þ KDCM SS
(6)
Ct;liq Kp cp ¼ 1 þ Kp Sp þ KDCM SS cDCM ¼
(7)
Ct;liq KDCM 1 þ Kp Sp þ KDCM SS
(8)
Kinetics of sorption and desorption of a pollutant between the free dissolved compartment and the particle one and between the free dissolved compartment and the DCM one are described by Equations (9) and (10): part rsorp=desorp
¼ k1
cp
cp ¼ k1 Kp Cf cp
(9)
(10)
where cp and cDCM are the OMPs particle and DCM concentrations at equilibrium, respectively. k1 and k2 are the first-order kinetic constants of sorption to particle and DCM, respectively.
3.2.
khyd
Sp /SS m
SS /X þ CH4 þ CO2 Table 3 represents the mathematical model with seven components and nine processes. The metabolism of substrate growth is incorporated into the three first processes. Hydrolysis is described with a first order kinetics to represent the enzymatic degradation of particulate substrate in soluble substrate. Beside, decay is assumed in the transformation of activated biomass into particulate substrate. Biomass (X, gCOD/ L) growth is linked to soluble substrate uptake and modeled with Monod-type kinetics: m ¼ mmax
rDCM sorp=desorp ¼ k2 cDCM cDCM ¼ k2 KDCM Cf cDCM
gCOD/L) and then soluble substrate (SS, gCOD/L) biodegradation to biogas.
Biodegradation
A two-steps model has been used to describe the anaerobic digestion of sludge: first hydrolysis to particulate matter (Sp,
SS KS þ SS
(11)
where mmax (1/d) is the maximum bacterial growth rate and KS (gCOD/L) is the half-saturation constant associated with the soluble substrate SS. OMP biodegradation may be considered as a classical metabolism with a specific OMP-degrader (Fig. 2b) and modeled with a Monod-type kinetics (Equation (11): mmax,OMP, KS,OMP). Nevertheless, the removal of pollutant present only in trace levels (ng/L or mg/L) could not result in any significant biomass growth (Clara et al., 2005). Thus, we assumed that cometabolism is the main OMP biodegradation mechanism (Fig. 2a). Criddle (1993) proposed a cometabolism model between a growing substrate and a non-growing substrate in simple
4515
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 1 1 e4 5 2 1
b
a
Fig. 2 e Scheme illustrating (a) cometabolism and (b) classic metabolism of OMP.
ecosystem. This equation is based on the assumption that the cometabolic degradation rate is enhanced by the generation of reductants caused by the degradation of growth substrate (SS) and, in its absence, the cometabolic transformation is linked to endogenous decay. Moreover, the cometabolic model includes competitive inhibition between growth and nongrowth substrate and negative effect of the toxic products. However, we did not include those two last terms in this work because of the low concentration of OMPs in the system and a large number of kinetic parameters can complicate the modeling effort under current conditions. In our case, PAHs are the non-growing substrate and Criddle’s equation has to be modified. In particular, the bioavailability limitation can be accounted for by replacing the total concentration by the bioavailable one (Cbioav):
1 tf
(12)
where Tc is the OMPs transformation capacity (mgOMP/gCODSs ) standing for cometabolic interaction between the soluble substrate metabolism and the OMPs metabolism, kc is the maximum specific rate of OMPs biodegradation in absence of primary substrate (mgOMP/gCOD-X.d) and KSC is the half saturation constant of OMPs in the Monod formalism (mgOMP/L). The kc and KSC parameters are representative of the response of the OMP metabolic route to the OMP bioavailable fraction including transporters and enzymes affinity for their substrate. m is the growth rate (1/d) and Y is the growth yield (gCOD-X/gCODSs ). Furthermore, the four-compartment distribution may help us to find which compartment is the real bioavailable fraction (Cbioav) to be biodegraded. Beside, the four-compartment model can be modified. By this way, the OMPs biodegraded fraction can be assumed to be the free dissolved fraction (Cf e only process 7) or the aqueous fraction (Cf and CDCM e processes 7 and 8), or the sum of all fractions (Cf, CDCM and Cp e processes 7, 8 and 9) as proposed by Fountoulakis et al. (2006).
3.3.
sq ¼
Sensitivity analysis
The model has a cascade structure, which means that the variables X, SS and Sp are not influenced by the other variables and, then, by the parameters associated with the other state variables. This cascade structure is an advantage to find the parameter set. First of all, we can estimate the parameters of biomass and substrates (mmax, KS, Y, b and khyd), and then, the parameters linked with the biodegradation of each OMP (Tc, kc and KSC).
Ztf 0
zqþDq zq dt zq
(13)
where tf is the test duration, zq is the variable z associated with base value of parameter q, and zqþDq is the variable z when the parameter q is changed an amount Dq. The sensitivity coefficient is presented in Fig. 3. Growth yield, Y, is the most sensitive parameter. A strong influence of parameters linked to soluble substrate uptake (Y, mmax, KS) can be noted. This is in agreement with cometabolism concept, where the micropollutant fate is associated to growth substrate degradation. Half saturation constant of OMP KSC and the specific biodegradation rate of OMP kc show a relative sensibility. These parameters are representative of the OMP metabolic route and depend on the type of microbial
Y kc
μmax Parameter
rbio
m C bioav X ¼ Tc þ kc KSC þ Cbioav Y
A sensitivity analysis for OMP total concentration was conducted to identify the most sensitive parameters in the four-compartment model with free dissolved compartment (Cf) as the available fraction. In reference to a given set of parameter values, initial condition and characteristics of pollutants and reactors, eleven parameters were changed over 10%, 20%, 30% and 50% of their based values. In steady state, nine simulations were run at each of these values to generate nine concentrations profiles of compartments (Cf, Cp, CDCM, Cg) for each parameter. A sensitivity coefficient, sq, of the variable z to the parameter q, defined by Equation (13) (Bernard et al., 2001; Myint et al., 2007), was calculated to quantify the average spread for each parameter.
KS KSC khyd b Tc KLa k1 k2 0
0,1
0,2
0,3
Sensitivity coefficient, σq Fig. 3 e Sensitivity coefficient of OMP concentration for the model parameters.
4516
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 1 1 e4 5 2 1
Model calibration
The model simulations were compared with experimental data of anaerobic digestion of PAHs for two mesophilic continuous reactors fed with primary (PS) and secondary sludge (SS). The influent and effluent macroscopic performances of reactors and biogas production are represented in Table 4. Volatile fatty acids (VFAs) did not accumulate and the methane content in biogas was about 70% in the two mesophilic reactors. The overall removals in COD, dry matter and organic matter were higher than 60%. The variation of PAHs removal between reactors was slightly different while between PAHs in one reactor was high. The reported overall PAHs removal rates were from 32 to 74% and from 38 to 73% for PS and SS reactor, respectively. In the model, the set of parameters were estimated using a non-linear least square method between simulated values
Table 4 e Anaerobic performances of PS and SS reactors. Average and standard deviation calculated from 5 measurements performed at steady state. Parameter COD input (g/L) COD output (g/L) % COD removal DM input (g/L) DM output (g/L) % DM removal VFAs input (gCOD/L) VFAs output (gCOD/L) % VFAs removal % CH4 L CH4/d
PS 28.1 13.4 52.4 22.4 12.1 46.0 0.70 0.11 86.2 0.67 0.80
3.6 1.7 7.5 1.7 2.0 8.4 0.31 0.12 12.9 0.23 0.21
SS 23.0 1.9 9.1 2.6 60.4 6.5 19.5 1.4 10.2 1.8 52.9 6.9 0.90 0.47 0.20 0.24 89.4 9.6 0.68 0.25 0.61 0.10
Parameter
Meaning
Unit
Value PS
1
mmax KS
Maximum growth rate d Half saturation of growth gCODSs/L substrate Growth yield gCODX/gCOD-Ss First-order endogenous d1 decay First-order kinetic of d1 hydrolysis
Y b khyd
SS
0.62 0.63 3.25 5.10 0.75 0.75 0.05 0.05 0.07 0.13
and measurements. We take here advantage of the cascade structure of the model with the identification of a first set of parameters. Table 5 summarizes the values of the parameters linked to biomass and substrates for PS and SS. The KS values suggest that the metabolism and the implied microbial population could be different in PS and SS biodegradation. Fig. 4 shows the behavior of soluble and particulate substrate in PS and SS reactors. The simulations closely followed the dynamic evolution of the soluble substrate. Note the difference of the particulate substrate concentration between PS and SS, as well as, the fast decline of the particulate substrate in SS reactor. It could explain the dissimilarity of the hydrolysis coefficients found for both digesters. However, the particulate substrate in PS reactor is not well predicted by the model. This may be due to the first order kinetics used in the hydrolysis step which may be different between a non-stabilized sludge (PS) by comparison to a stabilized one (SS). Moreover, it is well known that the hydrolysis is the rate-limiting step in the anaerobic digestion for particulate substrate and the first order kinetics may be inaccurate to describe the hydrolysis of certain complex substrates (Vavilin et al., 2008). The set of parameters estimated in this section were used in the model calibration for evaluation of cometabolism and bioavailability hypotheses.
4.1.
Cometabolism evaluation
The physical and chemical characteristics of the OMP, as well as environmental factors, may influence their biodegradability. There are numerous references reporting that 4
20
3
15
Sp (gCOD/L)
4.
Table 5 e Estimated values of the biomass and substrates parameters.
SS (gCOD/L)
consortium. Moreover, hydrolysis step (khyd) has little influence on the compartments concentration despite it is the ratelimiting step in the anaerobic digestion. As pointed out by the results, the least sensitive parameters are Tc, KLa, k1 and k2, and therefore they will be less precisely estimated. In fact, the volatilization of PAHs is negligible, and then the OMP gas concentrations are small, as a result KLa does not influence the OMP concentration. On the other hand, hydrophobic character of PAHs makes easy their sorption onto sludge. Therefore, the first order kinetic constants (k1, k2) have high values (Dionisi et al., 2006). Beside, Kordel et al. (1997), Dionisi et al. (2006) and Barret et al. (2010b) have demonstrated that the sorption equilibrium state for PAHs was achieved after 1 or 2 h shaking. This sorption mechanism is faster compared to the biodegradation of these compounds under anaerobic condition (Chang et al., 2003). As a consequence for hydrophobic pollutant, this little influence of the first-order kinetic constants of sorption to particle and DCM (k1, k2) on the model showed that the sorption kinetics are not the rate-limiting steps and that the equilibrium state can be sufficient to represent the sorption phenomenon in the OMPs removal. Indeed, the processes 5 and 6 can be replaced by their equilibrium in the model in the case of hydrophobic compounds.
2 1 0
0
20
40
60
10 5 0
0
20
40
60
Time (d) Fig. 4 e Behavior of soluble (SS) and particulate (Sp) substrate for reactors PS (gray) and SS (black). Circles (C): experimental data and solid line: model.
4517
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 1 1 e4 5 2 1
4.2.
recalcitrant compounds may be transformed in the presence of another compound used as carbon and energy source, i.e. cometabolism (Chang et al., 1993, 2003; Criddle, 1993; Tiehm and Fritzsche, 1995; Yuan et al., 2001; Clara et al., 2005; Plo´sz et al., 2010; Barret et al., 2010c). It is usually assumed that the cometabolism may occur relatively slower than metabolism of growth substrate. Consequently, the general scheme adopted for the cometabolism of an OMP is shown on Fig. 2a. Total biomass consortium synthesizes enzymes for soluble substrate uptake and OMP degradation. Likewise, we propose the free dissolved compartment of OMP as bioavailable fraction. The behavior of the model with cometabolism has been compared to a classic metabolism of OMP (Fig. 2b). In this case, a specific degrader uptake OMP (XOMP) with Monod-type kinetics. Besides, non-linear least square optimization method has been used to estimate PAH parameters in the cometabolism kinetics (Tc, kc and KSC) and classic kinetics (i.e. Monod kinetics for an OMP: mmax,OMP, KS,OMP, YOMP) for each reactor (PS and SS) and thirteen PAHs. Fig. 5 displays the simulated (with both models) and experimental results for the fluoranthene in PS and SS reactors. Similar behaviors have been obtained for the other PAHs (data not shown). The model simulations with cometabolism closely follow the dynamic evolution of the total fluoranthene concentration and its compartments. In contrast, the model with a classic metabolism resulted in an overestimation of the experimental data. The residual values evaluated are 50.1 and 0.82 for metabolism and cometabolism, respectively, in Cf compartment for fluoranthene and PS. Moreover, cometabolic route of our results can be reinforced by following facts: (i) Barret et al. (2010c) reported a strong correlation between PAH and dry matter removal rates, it agrees with the results of Trably et al. (2003) and (ii) under anaerobic condition, Chang et al. (2003) and Trably et al. (2003) shown no growth with PAH as source of carbon.
PS 200
100
100
50
0
0
20
40
60
0
0
20
40
60
Various definitions of bioavailability are used across many disciplines (Semple et al., 2004). In this paper, a bioavailable compound is the chemical fraction that can be freely transformed by a microorganism. From a general point of view, a sorbed micropollutant is not available for microbial degradation; while its biodegradation occurs predominantly in the bulk aqueous phase (Byrns, 2001; Artola-Garicano et al., 2003; Urase and Kikuta, 2005; Dionisi et al., 2006; Plo´sz et al., 2010; Barret et al., 2010c). However, few studies have concluded that at least some microorganisms are capable of degrading compounds directly from the sorbed phase (Haws et al., 2006; Fountoulakis et al., 2006). In order to find the real OMP bioavailable fraction, the fourcompartment model was modified into the biotic process matrix for Cf, CDCM and Cp compartments (Table 3 and processes 7, 8 and 9). To this end, we have tested three hypotheses in the model (Table 3). Hypothesis 1 (processes 1e7): the bioavailable fraction was assumed to only be the free dissolved compartment, given that it is possible to separate the free dissolved fraction and sorbed to DCM of the aqueous phase in the model. Hypothesis 2 (processes 1e8): aqueous fraction is available for the microbial degradation activity. This in concordance with the widespread assumption that the aqueous fraction of OMP corresponds to their bioavailable compartment (Chang et al., 2003; Artola-Garicano et al., 2003; Dionisi et al., 2006; Barret et al., 2010c). Indeed, two mechanisms were proposed for the degradation of micropollutant sorbed to DCM: (i) low molecular weight DCM might be able to cross the microbial membrane in form of micropollutant-DCM complex, because some molecules up to a few kDa were shown to cross bacterial membrane (Nikaido, 2003) and (ii) recently it had been demonstrated that particles might transport the sorbed micropollutant directly or to vicinity of
Free-dissolved Compartment ( g/L)
Effluent HAP ( g/L)
Influent HAP ( g/L)
Bioavailability evaluation
Sorbed to DCM Compartment ( g/L)
Sorbed to particle Compartment ( g/L)
3
30
60
2
20
40
1
10
20
0
0
20
40
60
0
0
20
40
60
0
0
20
40
60
0
20
40
60
SS 200
100
100
50
0
0
20
40
60
0
0
20
40
60
3
30
60
2
20
40
1
10
20
0
0
20
40
60
0
0
20
40
60
0
Time (d) Fig. 5 e Fluoranthene behavior in PS and SS reactors: (gray line) influent concentration of PAH, (black circles) experimental data, (white circles) values estimated from equilibrium constants, (black line) model with cometabolism and (dashed line) model with classic metabolism.
4518
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 1 1 e4 5 2 1
PS 150
Free-dissolved Compartment ( g/L)
Effluent HAP ( g/L)
Influent HAP ( g/L)
Sorbed to DCM Compartment ( g/L)
80
1
20
40
0.5
10
60
100
40
50 0
Sorbed to particle Compartment (μg/L)
20
0
20
40
60
0
0
20
40
60
0
0
20
40
60
0
0
20
40
60
0
0
20
40
60
0
20
40
60
SS 150
80
1
20
40
0.5
10
60
100
40
50 0
20
0
20
40
60
0
0
20
40
60
0
0
20
40
60
0
0
20
40
60
0
Time (d) Fig. 6 e Chrysene behavior in PS and SS reactors: (gray line) influent concentration of PAH, (black circles) experimental data, (white circles) values estimated from equilibrium constants, (black line) hypothesis 1, (dark-gray line) hypothesis 2 and (light-gray line) hypothesis 3.
Rate: rSs (gCOD/gCOD.d), rOMP ( g/gCOD.d)
the cell surface (Smith et al., 2009), prior to the diffusion of free micropollutant throughout cell membrane. Finally, the hypothesis 3 (processes 1e9): All compartments are bioavailable as proposed by Fountoulakis et al. (2006). This is probably a mechanism of transport of sorbed micropollutant to vicinity of microorganism. Non-linear least square optimization method has been used to estimate PAH parameters (Tc, kc and KSC) for each case, thirteen PAHs and two sludge. The residuals value taken up to quantify the best fitting do not present a high variation between cases, for example the residual values are 0.11, 0.10 and 0.12 for free dissolved, aqueous phase and all compartments as available fraction, respectively, for chrysene in Cf compartment and PS. Fig. 6 shows the comparison of model predictions (three hypotheses) for chrysene in PS and SS reactors. Similar behaviors have been obtained for the other PAHs (data not shown). Such results could suggest that the PAH degradation occurs at the same time into free, aqueous and solid fraction, i.e sorbed fractions into particle and dissolved colloidal matter could be bioavailable to degraders. However, the half
saturation constants (KSC) of PAHs estimated in the first case (the free dissolved fraction is the bioavailable one) are ten times and hundred times lower than for the two other hypotheses (aqueous fraction and the sum of all compartments, respectively) as shown on Fig. 7. Consequently, in the hypothesis 1, the degradation rate of OMP is faster than that of soluble substrate as shown on Fig. 7a. This disagrees with the cometabolism studies demonstrating that the cometabolism is relatively slow by comparison to the metabolism of growth substrate (Chang et al., 1993; Haws et al., 2006). In addition, half saturation value estimated in the hypothesis 3 is higher than the OMP concentration in the free compartment (KSC>>Cf). This implies that the free compartment degradation is negligible compared to the particle compartment one and suggests high affinity for the OMP sorbed to particle. This case is really atypical, since it is generally considered that the sorbed chemicals are unavailable for microorganisms unless desorption occurs first. Indeed, Feng et al. (2000) have demonstrated that some bacteria can degrade sorbed chemical but it is not more important than the OMP aqueous phase
b
KSC = 0.7
a
KSC = 8.8
c
KSC = 63
Concentration: Ss (gCOD/L), OMP (μg/L) Fig. 7 e Comparison between substrate degradation rate (solid line) and OMP degradation rate (dashed line) for three hypotheses. (a) hypothesis 1: Cbioav [ Cf, (b) hypothesis 2: Cbioav [ Cf, CDCM and (c) hypothesis 3: Cbioav [ Cf, CDCM, Cp.
4519
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 1 1 e4 5 2 1
0.16
0.008
kc/KSC (L/gCOD.d)
Tc/KSC (L/gCOD)
0.010
R2 = 0.50
0.006 0.004 2 R = 0.67
0.002 0
150
200
250
0.12
0.08
2 R = 0.49
0.04
0
300
R2 = 0.55
150
M (g/mol)
200
250
300
M (g/mol)
Fig. 8 e Tc/KSC and kc/KSC as a function of molecular masses of PAHs. PS (gray) and SS (black).
degradation. In contrast, hypothesis 2 presents a cometabolism slower than metabolism of soluble substrate (Fig. 7b) and the affinity for free compartment and DCM compartment are comparable. Finally, based on the three-compartment model, Barret et al. (2010c) reported a strong correlation between PAHs aqueous fraction degradation and the dry matter removal and shown that the PAH biodegradation depended on a combination of bioavailability and cometabolism. Thus, the widespread assumption that the aqueous fraction of PAHs corresponds to their bioavailable compartment (Chang et al., 2003; Artola-Garicano et al., 2003; Dionisi et al., 2006; Barret et al., 2010c) was validated by our results. It is worth noting that PAH volumetric gas fraction does not exceed 0.05% of micropollutant total concentration for both low and higher molecular weight PAHs (data not shown). Moreover, hydrophobic character of PAH proves strong PAH affinity (higher to 98%) for both particle and DCM while PAH free concentration is hardly detectable (1.5%). However, PAH sorption to particle in PS reactor (85 5%) was higher than SS (65 5%) one and PAH affinity for DCM was 13 5% and 33 5% for PS and SS, respectively. As a consequence in this study, higher biological removal of individual PAH was observed in SS reactor in contrast with PS one.
4.3.
Kinetic parameters
The results suggest that the three cometabolism parameters (Tc, kc and KSC) could explain the different biodegradation rates between PAHs and between bioreactors. This is valid under the assumption that the aqueous fraction (sum of free and sorbed to DCM compartments) is the bioavailable compartment. Fig. 8 shows the variation of Tc/KSC and kc/KSC as function of molecular weight of PAHs. The transformation capacity values Tc did not present differences between PAHs and reactors (value close to 0.05 mgOMP/gCODSs ). As a result, the Tc value could correlate with a molecular structure family. Indeed, this term links PAHs degradation to soluble substrate utilization, so it might play a role in the different fates of PAHs in the reactor fed with PS and SS. However, previous PS and SS characterizations presented similar composition and reported slight differences in proteins and lipids content (Table 1). As a result, Tc can present trifling variation between substrates (PS and SS). Half saturation constant KSC is likely to vary as a function of PAH molecular weight. KSC values indeed increase when PAHs
molecular weight increases. This is in accordance with the idea that high molecular weight PAHs are less efficiently removed (Chang et al., 2003). Moreover, specific biodegradation rate, kc was shown to vary between reactors in a similar range (PS: 0.70e0.85 and SS: 0.60e0.90 mgOMP/gCOD-X.d). Therefore, half saturation constant KSC and kc probably depend on microbial consortium. In this study, it was shown that different consortia exhibit different KSC and kc. This microbial effect could account for biodegradation differences reported when bioaugmentation strategy has been developed (Trably et al., 2003).
5.
Conclusion
A four-compartment model of the fate of thirteen PAHs during anaerobic digestion of contaminated sludge was developed and confronted with experimental data. The model includes abiotic and biotic processes: volatilization, sorption, biodegradation as metabolism or cometabolism. Furthermore, in the case of hydrophobic pollutants, the sorption process can be represented by the equilibrium state. The model helps in elucidating which fraction of the PAHs distribution at equilibrium state is the real bioavailable compartment. Thus, the simulation carried out in this study validated the accepted assumption that the aqueous phase is bioavailable. Indeed, biodegradation affinity for OMP free dissolved and OMP sorbed to DCM are comparable. Furthermore, PAH biodegradation rate was coupled to cometabolism. The PAH removal was linked to soluble substrate uptake in the anaerobic digestion of sludge. The modified cometabolism model predicted well the relation between bioavailability and cometabolism of OMP. As a result of the numerical simulation, the three cometabolism parameters (Tc, kc and KSC) were shown to be molecule-dependant. These estimated parameter values could explain the different biodegradation rates between PAHs and between reactors. Nevertheless, the applied methodology for the parameters identification may converge toward several values but this study can be considered as a starting point, given that the parameter values comparison with previously published data was hardly feasible. A limitation in this model is that it does not include the OMPs inhibition and toxic effect, which could be considered in future work. However, the model proposed is potentially useful to better understand the micropollutant distribution, predict the
4520
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 1 1 e4 5 2 1
fate of PAHs under anaerobic condition and help to optimize the operation process for their removal.
Acknowledgments We would like to thank the Ibague University and COLCIENCIAS (Administrative Department of Science, Technology and Innovation) e Colombia for their financial support.
references
Alder, A.C., Siegrist, H., Fent, K., Egli, T., Molnar, E., Poiger, T., Schaffner, C., Giger, W., 1997. The fate of organic pollutants in wastewater and sludge treatment: significant processes and impact of compound properties. CHIMIA International Journal of Chemistry 51 (12), 922e928. Artola-Garicano, E., Borkent, I., Damen, K., Jager, T., Vaes, W.H.J., 2003. Sorption kinetics and microbial biodegradation activity of hydrophobic chemicals in sewage sludge: model and measurements based on free concentrations. Environmental Science & Technology 37 (1), 116e122. Barret, M., Carrere, H., Latrille, E., Wisniewski, C., Patureau, D., 2010a. Micropollutant and sludge characterization for modeling sorption equilibria. Environmental Science & Technology 44 (3), 1100e1106. Barret, M., Patureau, D., Latrille, E., Carre`re, H., 2010b. A threecompartment model for micropollutants sorption in sludge: methodological approach and insights. Water Research 44 (2), 616e624. Barret, M., Carre`re, H., Delgadillo, L., Patureau, D., 2010c. PAH fate during anaerobic digestion of contaminated sludge: do bioavailability and/or cometabolism limit their biodegradation? Water Research 44 (13), 3797e3806. Barret, M., Cea Barcia, G., Guillon, A., Carrere, H., Patureau, D., 2010d. Influence of feed characteristics on the removal of micropollutants during the anaerobic digestion of contaminated sludge. Journal of Hazardous Materials 181 (1e3), 241e247. Bernard, O., Hadj-Sadok, Z., Dochain, D., Genovesi, A., Steyer, J.P., 2001. Dynamical model development and parameter identification for an anaerobic wastewater treatment process. Biotechnology and Bioengineering 75 (4), 424e438. Byrns, G., 2001. The fate of xenobiotic organic compounds in wastewater treatment plants. Water Research 35 (10), 2523e2533. Chang, M.K., Voice, T.C., Criddle, C.S., 1993. Kinetics of competitive inhibition and cometabolism in the biodegradation of Benzene, Toluene, and pXylene by two Pseudomonas isolates. Biotechnology and Bioengineering 41 (11), 1057e1065. Chang, B.V., Chang, S.W., Yuan, S.Y., 2003. Anaerobic degradation of polycyclic aromatic hydrocarbons in sludge. Advances in Environmental Research 7 (3), 623e628. Clara, M., Kreuzingera, N., Strenna, B., Gansb, O., Kroissa, H., 2005. The solids retention time-a suitable design parameter to evaluate the capacity of wastewater treatment plants to remove micropollutants. Water Research 39 (1), 97e106. Couillard, C.M., Courtenay, S.C., Macdonald, R.W., 2008. Chemical-environment interactions affecting the risk of impacts on aquatic organisms: a review with a Canadian perspective - interactions affecting vulnerability. Environmental Reviews 16, 19e44. Criddle, C.S., 1993. The kinetics of cometabolism. Biotechnology and Bioengineering 41 (11), 1048e1056.
Dionisi, D., Bertin, L., Bornoroni, L., Capodicasa, S., Petrangeli Papini, M., Fava, F., 2006. Removal of organic xenobiotics in activated sludges under aerobic conditions and anaerobic digestion of the adsorbed species. Journal of Chemical Technology & Biotechnology 81 (9), 1496e1505. Feng, Y., Park, J.H., Voice, T., Boyd, S., 2000. Bioavailability of soilsorbed Biphenyl to Bacteria. Environmental Science & Technology 34 (10), 1977e1984. Fountoulakis, M.S., Stamatelatou, K., Batstone, D.J., Lyberatos, G., 2006. Simulation of DEHP biodegradation and sorption during the anaerobic digestion of secondary sludge. Water Science & Technology 54 (4), 119e128. Haws, N.W., Ball, W.P., Bouwer, E.J., 2006. Modeling and interpreting bioavailability of organic contaminant mixtures in subsurface environments. Journal of Contaminant Hydrology 82 (3e4), 255e292. Joss, A., Zabczynski, S., Go¨bel, A., Hoffmann, B., Lo¨ffler, D., McArdell, C.S., Ternes, T.A., Thomsen, A., Siegrist, H., 2006. Biological degradation of pharmaceuticals in municipal wastewater treatment: proposing and classification scheme. Water Research 40 (8), 1686e1696. Joss, A., Andersen, H., Ternes, T., Richle, P.R., Siegrist, H., 2004. Removal of estrogens in municipal wastewater treatment under aerobic and anaerobic condition: consequences for plant optimization. Environmental Science & Technology 38 (11), 3047e3055. Kordel, W., Hennecke, D., Franke, C., 1997. Determination of the adsorption-coefficients of organic substances on sewage sludges. Chemosphere 35 (1e2), 107e119. Lindblom, E., Press-Kristensen, K., Vanrolleghem, P.A., Mikkelsen, P.S., Henze, M., 2009. Dynamic experiments with high bisphenol-A concentrations modelled with an ASM model extended to include a separate XOC degradation microorganism. Water Research 43 (13), 3169e3176. Myint, M., Nirmalakhandan, N., Speece, R.E., 2007. Anaerobic fermentation of cattle manure: modeling of hydrolysis and acidogenesis. Water Research 41 (2), 323e332. Nikaido, H., 2003. Molecular basis of bacterial outer membrane permeability revisited. Microbiology and Molecular Biology Reviews 67 (4), 593e656. Plo´sz, B.G., Leknes, H., Thomas, K.V., 2010. Impacts of competitive inhibition, parent compound formation and partitioning behavior on the removal of antibiotics in municipal wastewater treatment. Environmental Science & Technology 44 (2), 734e742. Press-Kristensen, K., Ledin, A., Schmidt, J.E., Henze, M., 2007. Identifying model pollutants to investigate biodegradation of hazardous XOCs in WWTPs. Science of the Total Environment 373 (1), 122e130. Semple, K.T., Doick, K.J., Jones, K.C., Burauel, P., Craven, A., Harms, H., 2004. Defining bioavailability and bioaccessibility of contaminated soil and sediment is complicated. Environmental Science & Technology 38 (12), 228e231. Smith, K.E.C., Thullner, M., Wick, L.Y., Harms, H., 2009. Sorption to humic acids enhances polycyclic aromatic hydrocarbon biodegradation. Environmental Science & Technology 43 (19), 7205e7211. Tiehm, A., Fritzsche, C., 1995. Utilization of solubilized and crystalline mixtures of polycyclic aromatic hydrocarbons by a mycobacterium sp. Applied Microbiology and Biotechnology 42 (6), 964e968. Trably, E., Patureau, D., Delgene`s, J.P., 2003. Enhancement of polycyclic aromatic hydrocarbons (PAH) removal during anaerobic treatment of urban sludge. Water Science & Technology 48 (4), 53e60. Trably, E., Delgenes, N., Patureau, D., Delgenes, J.P., 2004. Statistical tools for the optimization of a highly reproducible method for the analysis of polycyclic aromatic hydrocarbons
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 1 1 e4 5 2 1
in sludge samples. International Journal of Environmental Analytical Chemistry 84, 995e1008. Urase, T., Kikuta, T., 2005. Separate estimation of adsorption and degradation of pharmaceutical substances and estrogens in the activated sludge process. Water Research 39 (7), 1289e1300.
4521
Vavilin, V.A., Fernandez, B., Palatsi, J., Flotats, X., 2008. Hydrolysis kinetics in anaerobic degradation of particulate organic material: an overview. Waste Management 28 (6), 939e951. Yuan, S.Y., Chang, J.S., Yen, J.H., Chang, B.V., 2001. Biodegradation of phenanthrene in river sediment. Chemosphere 43 (3), 273e278.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 2 2 e4 5 3 0
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Micellar enhanced ultrafiltration process for the treatment of olive mill wastewater Abdelilah El-Abbassi a, Mohamed Khayet b, Abdellatif Hafidi a,* a b
Laboratory of Food Sciences, Faculty of Sciences-Semlalia, Bd. My Abdellah, PB: 2390, 40090 Marrakech, Morocco Department of Applied Physics I, Faculty of Physics, University Complutense of Madrid, Av. Complutense s/n, 28040 Madrid, Spain
article info
abstract
Article history:
Olive mill wastewater (OMW) is an important environmental pollution problem, especially
Received 30 March 2011
in the Mediterranean, which is the main olive oil production region worldwide. Environ-
Received in revised form
mental impact of OMW is related to its high organic load and particularly to the phytotoxic
24 May 2011
and antibacterial action of its phenolic content. In fact, polyphenols are known as powerful
Accepted 31 May 2011
antioxidants with interesting nutritional and pharmaceutical properties. In the present
Available online 14 June 2011
work, the efficiency of OMW Micellar Enhanced Ultrafiltration (MEUF) treatment for removal and concentration of polyphenols was investigated, using an anionic surfactant
Keywords:
(Sodium Dodecyl Sulfate salt, SDS) and a hydrophobic poly(vinyldene fluoride) (PVDF)
Olive mill wastewater
membrane. The effects of the process experimental conditions on the permeate flux were
Polyphenols
investigated, and the secondary membrane resistance created by SDS molecules was
Micellar enhanced ultrafiltration
evaluated. The initial fluxes of OMW processing by MEUF using SDS were 25.7 and
Treatment
44.5 l/m2 h under transmembrane pressures of 3.5 and 4.5 bar, respectively. The rejection
Secondary resistance
rate of polyphenols without using any surfactant ranged from 5 to 28%, whereas, it reached
Rejection rate
74% when SDS was used under optimum pH (pH 2). The MEUF provides a slightly colored permeate (about 88% less dark), which requires clearly less chemical oxygen demand (COD) for its oxidation (4.33% of the initial COD). These results showed that MEUF process can efficiently be applied to the treatment of OMW and for the concentration and recovery of polyphenols. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
Olive oil extraction processes produce large quantities of olive mill wastewater (OMW), which exhibit phytotoxic properties, mainly due to natural phenolic compounds. In Morocco about 5.8 105 m3 of OMW are produced yearly (Achkari-Begdouri and Klimm, 2004). The traditional oil extraction system produces about 0.5 m3 of liquid waste per ton of olives. The three-phase system produces 1.1e1.5 times the weight of milled olives (Paraskeva et al., 2007). Besides, OMW chemical composition depends on the cultivar, climate, soil nature, olives ripeness and oil extraction process (Parinos et al., 2007).
To protect the environment and the crops from possible damage, a specific law prohibiting direct discharge of OMW without pre-treatment has been recently adopted in Morocco. Similar situations can be found in other Mediterranean countries, where olive oil is produced. Under mechanical processing, only about 1% of the total polyphenols present in olives can be found in oil. Most part of the olive polyphenols remain in OMW and also in solid wastes (Niaounakis and Halvadakis, 2004). Several studies demonstrate the negative impact of a direct discharge of OMW on soil properties (Rinaldi et al., 2003). The antibacterial and phytotoxic properties of polyphenols were
* Corresponding author. Tel.: þ212(0)661412030; fax: þ212(0)524437412. E-mail address:
[email protected] (A. Hafidi). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.05.044
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 2 2 e4 5 3 0
List of symbols Cf Cp Cg COD J Lp OD ΔP R Rmem
polyphenols concentration in the feed (g/l) polyphenols concentration in the permeate (g/l) gelation concentration (M) Chemical Oxygen Demand (g of O2/l) permeate flux (l/m2 h) hydraulic membrane permeability (l/m2 h Pa) optical density () transmembrane pressure (bar) rejection rate of phenolic compounds (%) intrinsic membrane resistance (m1)
confirmed as well as the negative effects of the acidic pH and high salinity, brought by traditional salting practices of olives prior to oil extraction, on soil (Sierra et al., 2001). However, when diluted, OMW are used for fertilizing, long term effects on micro-organisms, fertility and soil properties are still not completely understood. Numerous researches have tried to develop efficient technologies for OMW treatment. These technologies are mainly based on biological degradation (anaerobic and aerobic) (Ammary, 2005) or advanced oxidation process, such as ozonation, Fenton’s reagent, electrochemical oxidation, etc., (Marques et al., 1997), as well as on various combinations of them (Khoufi et al., 2006). It should be noted that the efficiency of the process, the complexity and the costs involved may vary significantly. High costs are often the main reason for not adopting these OMW treatment methods. Almost all the above cited treatments aim the destruction of the organic matter present in OMW although it contains high valuable compounds such as phenolic compounds. It is worth quoting that wastewaters from olive oil mills, can be interesting biological sources of high added value compounds, such as hydroxytyrosol or other antioxidant phenolic compounds. In fact, the phenolic compounds are known as important natural antioxidants with nutritional and pharmaceutical properties (Tuck and Hayball, 2002). Several medicinal and pharmaceutical scientists have demonstrated that olive phenols, especially hydroxytyrosol (3,4-di-hydroxyphenyl-ethanol), were effective in preventing and curing some important diseases (Visioli et al., 1995). Nevertheless, hydroxytyrosol is not commercially available in large quantities like other food additives. Several methods have been proposed for the production of hydroxytyrosol by means of chemical (Visioli et al., 1998) or enzymatic synthesis (Tuck et al., 2000). Such protocols are usually slow and expensive, resulting in few numbers of commercially available products containing pure hydroxytyrosol. Various methods have been proposed to recover phenols from olive mill wastewater such as solvent extraction, resin chromatography, solideliquid or liquideliquid extraction, and supercritical fluid extraction. Unfortunately, almost all tested processes are expensive and/or inappropriate (Ferna´ndezBolan˜os et al., 2002). Membrane technology has been proposed as a promising tool for oil wastewaters treatment. Paraskeva et al. (2007) tested different combinations of three membrane processes (ultrafiltration (UF), nanofiltration (NF),
Rs Rcp Rf Rtot s t vp vr hw
4523
membrane secondary resistance (m1) resistance related to the concentration polarization (m1) resistance related to the fouling phenomenon (m1) total resistance of the membrane (m1) effective membrane area (cm2) operating time (min) permeate volume (ml) retentate volume (ml) viscosity (Pa s)
and/or reverse osmosis (RO)) for olive mill wastewater fractionation. The combination of centrifugation and UF allow a chemical oxygen demand (COD) reduction of about 90% (Turano et al., 2002). The treatment consists in a preliminary centrifugation step, in which the suspended solids are removed in a selective separation phase by UF of the centrifuge supernatant. Stoller and Bravi (2010) studied the effect of particle size distributions in OMW on the critical flux and membrane fouling when different pre-treatment processes are applied prior to membrane processing. The reported results showed that optimal pre-treatment process, which allows the highest critical flux, should reduce significantly the solute concentration and increase the particle size over the membrane pore sizes. A treatment of olive mill wastewater (OMW) by combining an UF technique and an advanced oxidation process using UV/ H2O2 was applied to remove a large part of total solids and organic carbon (Drouiche et al., 2004). A combined application of RO and adsorption processes was also applied to the treatment of OMW (Canepa et al., 1988), and allowed a COD reduction of about 99%. Nevertheless, the main limitation of these proposed treatments is the high involved costs. Recently, membrane distillation (MD) has been efficiently applied for OMW treatment and concentration for polyphenols recovery (El-Abbassi et al., 2009). However, the limitation of such process was the obtained low permeate fluxes compared to pressure-driven membrane technologies. The efficiency of micellar enhanced ultrafiltration (MEUF) to recover phenolic derivatives from model mixtures have been evaluated and have shown a high rejection of such compounds (Purkait et al., 2005). So far, MEUF has been used to separate different organic and inorganic compounds, using various surfactants (Purkait et al., 2006). However, research studies using MEUF for the treatment of multi-solute systems such as OMW are very scarce within membrane literature. The concentration of OMW may reduce significantly the costs of the process used for polyphenols recovery. The natural phenolic compounds can be used as food additives; antimicrobials, pesticides; cosmetics and pharmaceutical compounds (Visioli et al., 1999). MEUF is a separation process in which surfactants are added to a waste stream to promote the removal of smaller molecules. Beyond a certain concentration level called the Critical Micellar Concentration (CMC), the surfactant
4524
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 2 2 e4 5 3 0
molecules will aggregate into structures known as micelles and will then solubilize the organic compound (OC) to form large OC-surfactant structures. The solution may be then subjected to UF using an appropriate membrane to concentrate the waste and produce a high quality permeate. The aim of the present work is to study the efficiency of MEUF when treating OMW and concentrating its phenolic compounds. The effects of some operating parameters such as pH and SDS concentration on the permeate flux and the rejection rate as well as on the membrane fouling were investigated.
2.
Materials and methods
2.1.
Olive oil mill wastewater samples
residue was dissolved in 10 ml of pure methanol. This final extract was called hereafter methanolic extract.
2.3.1.2. Determination of total phenolic content. The total phenolic content was determined following the Folin-Ciocalteau spectrophotometeric method according to Singleton et al. (1999) and using Tyrosol as a standard. The methanolic extract (0.2 ml) was diluted with distilled water (6.8 ml). Folin-Ciocalteau reagent (0.5 ml) was added and the contents of flask were mixed thoroughly. After 3 min, 1 ml of a sodium carbonate anhydrous solution (20 wt%) was added, and then the mixture was allowed to stand for 1 h in the dark. The optical density of the bluecolored samples was measured at 765 nm. The total phenolic content was determined as tyrosol equivalents (TYE) and values are expressed as mg of tyrosol/l of OMW. 2.3.2.
Membrane preparation
2.3.
Procedures
PVDF flat sheet membranes were prepared by phase inversion technique, using PEG as an additive and DMAc as a solvent. First, 80 ml of DMAc was mixed with 3 g of PEG. The mixture was allowed to stand for at least 12 h at ambient temperature. After complete dissolution of PEG in DMAc, 17 g of the PVDF polymer was added to the mixture and heated at 45 C during 24 h. The formed polymer solution was degassed at a room temperature before casting the polymer solution into a glass plate. The cast film together with the glass plate, were immersed in distilled water for coagulation. The contact of the cast film with distilled water induces a diffusional mass exchange, solvent-out and water-in the nascent membrane. During gelation, it was observed that the membrane peeled out from the glass plate spontaneously. After 12 h in distilled water (coagulant bath), the formed membrane was subjected to solvent exchange method. First, the membrane was immersed in 50% aqueous methanol solution for 6 h. Subsequently, the membrane was dried at ambient conditions for at least 24 h.
2.3.1.
Physico-chemical characterization of OMW samples
2.4.
OMW samples were obtained from semi-modern units based on press-based oil extraction process located in region of Marrakech (Morocco). The samples were freshly used for the physico-chemical analysis and UF experiments.
2.2.
Chemicals
The anionic surfactant SDS (Sodium Dodecyl Sulfate salt), the poly(vinyldene fluoride) (PVDF, MW ¼ 180 kD), the additive poly(ethylene glycol) (PEG, 1 kD), the solvent N,N-Dimethyl acetamide (DMAc), methanol, ethyl acetate, Folin-Ciocalteu reagent, Tyrosol and sodium bicarbonate were procured from SigmaeAldrich Co. Hydrochloric acid and n-hexane were purchased from Rieden-deHae¨n Co. All chemicals were of analytical grade. Tyrosol was used as standard for the determination of total phenolic content. All chemical solutions were prepared using distilled water.
The electrical conductivity (EC) and pH of the OMW samples were measured directly using WTW Multilab (P5 Germany). The humidity and dry extract were determined before and after drying the sample overnight at 105 C. The chemical oxygen demand (COD) was measured using closed refluxcolorimetric method. The total organic carbon (TOC) was determined by the Anne method described by Aubert (1978). Proteins and sugars were measured using Bradford reagent (Bradford, 1976). Total phosphorus was measured colorimetrically as a molybdovanadate phosphoric acid (APHA, 1981). The total phenolic content of both the feed and permeate samples was determined colorimetrically by the FolinCiocalteau reagent (Folin and Ciocalteau, 1927), after the liquideliquid extraction detailed below:
2.3.1.1. Liquideliquid extraction of polyphenols from OMW. About 10 ml of each sample (feed or permeate) were acidified to pH 2 with HCl (6 N) and washed twice with 10 ml of nhexane to eliminate traces of lipids. The water phase was extracted three times with 10 ml of ethyl acetate by centrifugation at 3000 g for 5 min. The three organic phases were recuperated and evaporated under reduced pressure and the
MEUF experiments
UF experiments were carried out at room temperature in a stirred UF cell (AMICON 8200, Millipore USA) with a 200 ml volume. The effective area (s) of the membrane is 28.7 cm2. A schematic diagram of experimental apparatus is shown in Fig. 1. Before each run, pure water was employed to determine the permeability of the membrane. Then, the surfactant solutions were prepared by solubilizing sodium dodecyl sulfate in distilled water to reach concentrations 10 times the critical micellar concentration (CMC). The CMC of the used surfactant (SDS) in distilled water is 9.7 mM (Rosen, 2004). The feed solutions were mixed adequately for at least 5 h before UF tests. Stirring was set at 500 rpm and the transmembrane pressure was regulated with a manometer. To evaluate the membrane separation efficiency for removal of natural phenolic compounds from the feed solution, the following equation is used:
Rð%Þ ¼
Cp 100 1 Cf
(1)
4525
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 2 2 e4 5 3 0
COD removal ð% ¼
CODpermeate 100 1 CODfeed
(3)
3.
Results and discussion
3.1.
Physico-chemical characterization of OMW
The physico-chemical properties of the used OMW are shown in Table 1. This effluent is relatively dense, meanly acidic and charged in salt with a high organic load that reaches values as high as 156 g/l COD. It contains large amounts of dry matter up to about 90 g/l. Amongst other organic constituents, OMW contain high concentrations of phenolic compounds up to about 4 g/l, which exceeds widely the environment legislation limit, set at 0.5 ppm for phenol (Pinto et al., 2005).
3.2.
Fig. 1 e Schematic diagram of the stirred ultrafiltration cell and experimental set-up used for the MEUF: (1) Permeate outlet, (2) Membrane disc, (3) Feed tank, (4) Stirrer, (5) Blowoff valve, (6) Nitrogen inlet pressure, (7) Valve, (8) Manometer, (9) Nitrogen cylinder, (10) Magnetic stirrer and (11) graduated cylinder.
where R is the rejection rate, Cp is the permeate concentration and Cf is the feed concentration, of the phenolic compounds, respectively.
2.4.1.
It was observed that UF permeate volumes (v) of distilled water show a straight linear relationship with time (t) (v ¼ at þ b). Thus, the permeate flux could be directly obtained as: J¼
dv a ¼ sdt s
2.4.1.1. Decolorization. The color of OMW was monitored by measuring the absorbency at different wavelengths. Measurements at 395 nm and 465 nm after 100 times dilution against distilled water were carried out (Jaouani et al., 2003). The removal of color was estimated using the following equation. ODpermeate 100 % of decolorization ¼ 1 ODfeed
(2)
where OD is the optical density.
2.4.1.2. Chemical oxygen demand (COD). COD of OMW feed, permeate and retentate was determined using the dichromate method. An appropriate amount of sample was introduced in a previously prepared digestion solution containing potassium dichromate, sulfuric acid and mercuric sulfate (LaPara et al., 2000). The mixture was then incubated for 120 min at 150 C in a COD reactor (Thermoreactor CR3000). COD concentration was measured colorimetrically using UVeVisible spectrophotometer. The process efficiency in reducing COD was expressed as follows:
(4)
where s is the membrane area. Fig. 2 shows the obtained water flux (Jw) for different transmembrane pressures (ΔP). A linear relationship was obtained between Jw and ΔP with a high correlation factor (0.9993). The hydraulic membrane permeability (Lp) was determined according to the following equation (Cassano et al., 2008):
Pollution removal efficiency of the MEUF technique
To determine the efficiency of the MEUF technique in reducing the recalcitrant pollutants, the color and the chemical oxygen demand (COD) were considered.
Ultrafiltration of distilled water
Lp ¼
Jw DP
(5)
Moreover the membrane resistance (Rmem) was estimated as follows (Cassano et al., 2008): Rmem ¼
1 hw LP
(6)
Table 1 e Physico-chemical characteristics of OMW sample. Parameters Humidity (%) pH Conductivity (mS/cm) Dry extract (g/l) Oil (g/l) Total Organic Carbon (g/l) COD (g of O2/l) Total Phosphorus (g/l) Total polyphenols (g of TYE/l) Sugar (g/l) Proteins (g/l)
Average value Standard deviation 90 5.3 24 90 7 25 156 0.6 4.1
15 0.3 8 36 1 7 27 0.2 0.6
4.3 0.3 1.8 0.4
4526
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 2 2 e4 5 3 0
Fig. 2 e Effect of transmembrane pressure on permeate fluxes of distilled water ultrafiltration.
Fig. 3 e Permeate volume versus UF operation time for the surfactant solution (10 CMC of SDS) tested under different transmembrane pressures.
where hw is the viscosity of the permeate. The water viscosity is equal to 0.00083 Pa s (at 28 C and under the atmospheric pressure). The obtained values are given in Table 2. The hydraulic permeability (Lp) of the membrane is found to be 11.2 104 0.8 l/m2 h Pa, and the membrane resistance (Rmem) is 3.87 0.3 1012 m1. It is worth quoting that the membrane permeability and the membrane resistance decrease slightly with the increase of the transmembrane pressure.
3.3.
Ultrafiltration of surfactant solutions
As it was stated earlier, a surfactant solution, 10 CMC of SDS in distilled water (pH ¼ 2) was tested by UF. The results are shown in Fig. 3 for different transmembrane pressures. It can be seen that the UF permeate volume does not follow a straight line as a function of time. Therefore the results were fitted to a second order polynomial equation (v ¼ at2 þ bt þ c) and the permeate flux was calculated as follows: JðtÞ ¼
dv 2at þ b ¼ sdt s
(7)
As it was expected, it was found that the permeate flux increased with the enhancement of the transmembrane pressure. However, the transmembrane pressure effect on the permeate flux was found to be less significant compared to distilled water flux (Fig. 2). Moreover, a decrease of the
Table 2 e Permeate flux (Jw), membrane permeability (Lp) and membrane resistance (Rmem) under different transmembrane pressures. Lp (104 m)
permeate flux was observed during processing time. This reduction of flux may depend on four phenomena: pore diameter reduction due to the solute sorption on the pore wall, formation of a dense solute layer with low permeability directly on the membrane surface, formation of a compressible filter cake and presence of a concentration polarization layer. This fouling phenomenon is known to be the main limitation of the pressure-driven membrane separation processes, and different procedures were adopted to limit this membrane fouling (feed solution pre-treatment, membrane cleaning, membrane surface modification, etc.). By changing sensibly the particle size distributions of the suspended solids, flocculation pre-treatment was reported to reduce membrane fouling effects and the membrane’s performances demonstrated to be strictly dependant on the coagulant type used (Stoller, 2009). Effective mixing is efficient to reduce the influence of the concentration polarization. Therefore, the observed reduction of flux could be attributed to the formation of a gel, as was already observed by many other researchers (Urbanski et al., 2002). A linear relationship of permeate flux versus the logarithm of the surfactant concentration (by CMC) in the retentate could be expected. Since only monomeric molecules of SDS can pass through the membrane, and consequently the surfactant concentration in the MEUF permeate will not exceed 1 CMC (Talens-Alesson, 2007). The surfactant concentration in the retentate (CR) could be approximated from the mass balance. nsp þ nsr ¼ nso
(8)
Rmem (1012 m1)
ΔP (bar)
Jw (l/m2.h)
2 2.5 3 3.5 4
200.32 271.94 340.31 409.17 488.20
10 10.9 11.3 11.7 12.2
4.31 3.97 3.81 3.69 3.54
Mean value
11.2 0.8
3.87 0.3
where nsP is the quantity of the surfactant in the permeate ðnsP ¼ vp $CMCÞ, nsR is the quantity of the surfactant in the retentate ðnsR ¼ ðvo vp Þ$CR Þ and nsO is the initial quantity of the surfactant in the feed solution ðnso ¼ vo $10 CMCÞ, where vo and vp refer to the feed initial volume and the permeate volume, respectively. Therefore, the concentration of the surfactant in the retentate (CR) is expressed as follows:
4527
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 2 2 e4 5 3 0
10vo vp CMC CR ¼ vo vp
(9)
If ð10vo vp Þ=ðvo vp Þ is replaced by a, Eq. (9) is written as: CR ¼ a$CMC
(10)
where a is a dimensionless coefficient and the subindexes S, P and R denote the surfactant, the permeate and the retentate, respectively. Fig. 4 shows the permeate flux as a function of the logarithm of the coefficient a for different transmembrane pressures. Linear relationships were observed for each transmembrane pressure. The permeate fluxes will be zero at more or less similar values. The corresponding concentration in the retentate is the so called the gelation concentration Cg which can be deduced at J ¼ 0 as (Cg ¼ a$CMC). The obtained values of Cg are depicted in Table 3. The average SDS gelation concentration was found to be 28 CMC (0.27 M). This value is small compared to the values reported in literature, 0.87 M (Urbanski et al., 2002) and 0.57 M (Christian and Scamehorn, 1989). This difference can be justified by the various experimental conditions affecting the concentration of the surfactant at the membrane surface (e.g. membrane characteristics, temperature, pH, ionic strength of the solution). Moreover, PVDF has ferroelectric properties and it is electrostatically charged. Consequently, some interactions between the PVDF membrane and charged solutes may occur favoring a rapid formation of a gel. The secondary resistance (Rs) caused by the SDS layer formation (gelation) could be calculated as follow: RS ¼
DP Rmem hw JðtÞ
(11)
When considering the resistance-in-series model (Kim et al., 2002), the total resistance can be written as: Rtot ¼ Rmem þ Rcp þ Rf ¼ Rmem þ Rs
(12)
where the subscripts tot, mem, cp and f are the total, membrane, concentration polarization and fouling resistances. Rs is the secondary resistance. The most significant parameter affecting permeate flux decline is fouling resistance (Rf), which contributes to a great extent to the secondary
Fig. 4 e Permeate flux versus Ln(a) for distilled water ultrafiltration under different transmembrane pressures.
Table 3 e Gelation concentration of SDS for the feed solution prepared with a surfactant solution of pH 2 and an SDS concentration of 10 CMC. ΔP (bar) 2 3.5 4
Ln(a) (at J ¼ 0)
a
Cg (M)
3.29 3.33 3.35
26.91 27.89 28.37
0.26 0.27 0.28
Mean value
27.73 0.74
0.27 0.01
resistance. Nevertheless, Rf can be reduced by appropriate methods such as cross-flow ultrafiltration. The membrane fouling can be divided into reversible and irreversible fouling depending on the attachment strength of solutes to the membrane surface. The secondary resistance caused by SDS molecules was calculated at t0 using Eq. (11). The permeate flux measured under a transmembrane pressure of 4 bar was used. The SDS molecules were found to contribute to about 91% of the total resistance (Rtot ¼ 41.67 1012 m1 vs Rs ¼ 38.13 1012 m1).
3.4.
Micellar enhanced ultrafiltration of OMW
3.4.1.
pH effect
As many natural phenolic compounds show weak acid properties, the pH of the solutions can affect their solubilization into SDS micelles. Therefore, first the optimal pH under which SDS shows high efficiency for polyphenols solubilization was determined. The variation of polyphenols rejection by MEUF of OMW is shown in Fig. 5 for different pH values. The pH of the feed was adjusted by mean of HCl and NaOH (2 N) to the desired values and the initial surfactant concentration was maintained the same in all experiments (10 CMC). The results show that polyphenols rejection (R) decreases with the increase of the pH and R is the highest (>70%) when the pH value was adjusted to 2. For pH values higher than 8, R decreases with the increase of the pH to reach values under 30%. A reduction of the pH considerably lowers the SDS Critical Micellar Concentration from 0.9 mM at pH 6 to 0.34 mM at pH 2.9 (Paulenovfi et al., 1998) and from 4.6 mM at pH 7 to
Fig. 5 e Polyphenols retention of OMW by MEUF applying different pH values (Transmembrane pressure: 4 bar, SDS concentration: 10 CMC).
4528
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 4 5 2 2 e4 5 3 0
1.26 mM at pH 2 (Macisek and Danihlik, 1998). The increase of pH may cause deprotonation of phenols and as a result all types of interactions with the polar head of the surfactant will be reduced. These two considerations may explain the high rejection of the phenolic compounds at low pH values. The pH value 2 was considered as an optimal value and used in all the subsequent experiments.
3.4.2.
Effect of SDS concentration
The variation of polyphenols rejection (R) at different surfactant concentrations (0; 5 and 10 CMC) is presented in Fig. 6. In the case of OMW treated without surfactant (0 CMC of SDS), it was observed that the rejection is small ( 1500 mJ cm2 while iohexol required a UV fluence of 720 mJ cm2. In a pilot plant treating pre-treated lake water, pharmaceutical transformation ranged from 67 to 98% with a medium pressure UV fluence of 540 mJ cm2 and a H2O2 dose of 6 mg L1 (Kruithof et al., 2007). This research was conducted to provide a more detailed understanding of factors controlling BAC transformation rates in UV photolysis and UV/H2O2 processes. The BACs tested in this study were the antimicrobial compounds sulfamethoxazole (SMX), sulfamethazine (SMZ), sulfadiazine (SDZ), and trimethoprim (TMP), the EDC bisphenol A (BPA), and the analgesic diclofenac (DCL). Specific objectives were to (1) determine the effects of solution pH on photolysis and photooxidation rates of BACs for which the degree of ionization can change considerably over the pH range typically encountered in water treatment, and (2) describe both experimentally and mathematically the influence of lake water (LW) and wastewater treatment plant effluent (WWTPE) matrices on BAC transformation rates.
2.
Materials and methods
2.1.
Experimental approach
Batch photolysis and UV/H2O2 oxidation experiments were carried out in a bench scale quasi-collimated beam (QCB) apparatus (Bolton and Linden, 2003). The purpose of the QCB apparatus is to ensure that UV rays reaching the sample are perpendicular to the water surface, which permits the accurate measurement of UV irradiance at the water surface and thus the accurate determination of the UV fluence (dose) delivered to the sample. The QCB was equipped with four low-pressure (LP) UV lamps, and a UV radiometer (UVX Radiometer, Upland, CA, USA) was used to measure the UV irradiance at the surface of the sample. An iodide/iodate actinometer was used to calibrate the radiometer readings (Rahn et al., 2006). The delivered UV fluence to the sample was calculated with the method described by Bolton and Linden (2003). Photolysis and UV/H2O2 oxidation experiments were conducted at initial BAC concentrations of 4 (1) mM, and the parent compound transformation was monitored as a function of UV dose (mJ cm2). SMX, SMZ, SDZ, TMP, BPA, and DCL served as target compounds in this study, and their characteristics are shown in Table 1. The effects of the following factors on BAC photooxidation rates were evaluated: (1) pH, (2) H2O2 concentration, and (3) background water matrix composition (hydroxyl radical scavenging due to background organic matter, alkalinity and other inorganic constituents). To determine the pH-dependent quantum yield of SMX, SMZ, SDZ, and TMP, photolysis experiments were conducted in
ultrapure water (UPW) buffered at pH values at which either the neutral or ionic form of the antimicrobial compounds was dominant (neutral form of sulfonamides at pH 3.6, anionic form of SMX at pH 7.85, anionic forms of SMZ and SDZ at pH 9.7, neutral form of TMP at pH 9.7, and cationic form of TMP at pH 3.6). For the UV/H2O2 degradation of SMX, SMZ, SDZ and TMP in UPW, only experiments at pH 3.6 and 7.85 were conducted to evaluate pH effects on oxidation rates. At pH 9.7, the reaction between the hydroxyl radical (OH) and the BAC would be affected by elevated carbonate concentrations due to enhanced dissolution of atmospheric CO2. Solution pH effects were not evaluated for BPA and DCL, because their pKa values (9.78 and 4.15, respectively) are distant from typical water treatment pH values; at the tested pH of 7.85, the neutral form of BPA and the anionic form of DCL was dominant. To quantify BAC oxidation rates in the UV/H2O2 process, QCB experiments were conducted with H2O2 concentrations of 2, 6, and 10 mg L1. Experiments in the presence of background organic matter were conducted at pH 7.85 in LW collected from Lake Wheeler (Raleigh, NC) and in WWTPE (Cary, NC). Both LW and WWTPE were filtered through a 0.45-mm nylon membrane (Magna-R, MSI, Westboro, MA) prior to use. Characteristics of LW and WWTPE are shown in Table 2. Photolysis and oxidation rate data were evaluated using two approaches. First, a fluence-based pseudo-first order reaction rate approach was used to evaluate the effects of pH, H2O2 concentration, and background water matrix on BAC transformation rates. Furthermore, the quantum yield for each BAC was determined from the fluence-based pseudofirst order photolysis rate. Second, the second order rate constant (kOH) describing BAC oxidation by OH in ultrapure water at pH 7.85 was obtained through competition kinetics (Huber et al., 2003; Pereira et al., 2007). In this study, pchlorobenzoic acid ( p-CBA) was selected as the reference compound because it is not measurably degraded by direct photolysis. Benitez et al. (2004) reported a quantum yield of 0.0030 mole Einstein1 for p-CBA at pH 7 and a wavelength of 254 nm. The second order rate constant describing the oxidation of p-CBA by OH has a value of 5 109 M1 s1 in the pH range of 6e9.4 (Buxton et al., 1988). In addition, kOH of the target compound was corrected by the percentage of the pseudo-first order reaction rate that accounted for the direct
Table 2 e Representative water quality parameters for the lake water (LW) and wastewater treatment plant effluent (WWTPE) used in this study.
DOC (mg L1) A254 (cm1) Alkalinity (mg L1as CaCO3) Nitrate (mg L1) Nitrite (mg L1) Sulfate (mg L1) Chloride (mg L1) Bromide (mg L1)
LWa
WWTPEa
5.1 0.130 24.4 1.9 BPA z TMP. For a H2O2 dose of 10 mg L1, the required UV dose to achieve 90% SMX and DCL transformation was
