WATER RESEARCH A Journal of the International Water Association
Editor-in-Chief
Associate Editors
Mark van Loosdrecht Delft University of Technology and KWR Watercycle Research The Netherlands E-mail:
[email protected]
Andrew Baker University of New South Wales Australia
Editors Jean-Claude Block Université H. Poincaré, Nancy I France David Dixon University of Melbourne Australia Hiroaki Furumai University of Tokyo Japan
Damien Batstone University of Queensland Australia Guang-Hao Chen Hong Kong University of Science & Technology China Rob Eldridge Glen Waverley Australia Monica B. Emelko University of Waterloo Canada
Xiaodi Hao Beijing University of Civil Engineering and Architecture China
Tim Fletcher University of Melbourne Australia
Gregory Korshin University of Washington USA
Rosina Girones University of Barcelona Spain
Anna Ledin Formas Sweden Eberhard Morgenroth ETH Zürich and Eawag, Swiss Federal Institute of Aquatic Science and Technology Switzerland
Stephen Gray Victoria University Australia Kate Grudpan Chiang Mai University Thailand
Yoshihiko Matsui Hokkaido University Japan Armand Maul Université de Metz France Max Maurer Eawag Switzerland How Yong Ng National University of Singapore Singapore Jong Moon Park Pohang University of Science & Technology Korea Susan Richardson US Environmental Protection Agency USA Miguel Salgot University of Barcelona Spain David Sedlak University of California, Berkeley USA Masahiro Takahashi Hokkaido University Japan
Peter Hillis AECOM Australia
Nathalie Tufenkji McGill University Canada
Hong-Ying Hu Tsinghua University China
Vayalam Venugopalan BARC Facilities India
Jae-Hong Kim Georgia Institute of Technology USA
Eduardo von Sperling Federal University of Minas Gerais Brazil
Hang-Shik Shin Korea Advanced Institute of Science and Technology Korea
Günter Langergraber University of Natural Resources and Applied Life Sciences, Vienna (BOKU) Austria
Julie Zilles University of Illinois at Urbana-Champaign USA
Thomas Ternes Bundesanstalt für Gewässerkunde Germany
Shang-Lien Lo National Taiwan University Taiwan
Anastasios Zouboulis Aristotle University of Thessaloniki Greece
Stefan Wuertz University of California, Davis USA
Dionisis Mantzavinos Technical University of Crete Greece
Editorial Office
Wolfgang Rauch University of Innsbruck Austria Maria Reis Universidade Nova de Lisboa/FCT Portugal
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 ) 6 6 0 3 e6 6 1 4
Available online at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Waterborne transmission of protozoan parasites: Review of worldwide outbreaks e An update 2004e2010 Selma Baldursson, Panagiotis Karanis* University of Cologne, Faculty of Medicine, Center of Anatomy, Institute II, Medical and Molecular Parasitology, Joseph-Stelzmann-Strasse 35, D-50937 Cologne, Germany
article info
abstract
Article history:
The present update gives a comprehensive review of worldwide waterborne parasitic
Received 13 July 2011
protozoan outbreaks that occurred and were published globally between January 2004 and
Received in revised form
December 2010. At least one hundred and ninety-nine outbreaks of human diseases due to
11 October 2011
the waterborne transmission of parasitic protozoa occurred and were reported during the
Accepted 12 October 2011
time period from 2004 to 2010. 46.7% of the documented outbreaks occurred on the
Available online 20 October 2011
Australian continent, 30.6% in North America and 16.5% in Europe. Cryptosporidium spp. was the etiological agent in 60.3% (120) of the outbreaks, Giardia lamblia in 35.2% (70) and
Keywords:
other protozoa in 4.5% (9). Four outbreaks (2%) were caused by Toxoplasma gondii, three
Contamination
(1.5%) by Cyclospora cayetanensis. In two outbreaks (1%) Acanthamoeba spp. was identified as
Diarrhea
the causative agent. In one outbreak, G. lamblia (in 17.6% of stool samples) and Cryptospo-
Protozoan parasites
ridium parvum (in 2.7% of stool samples) as well as Entamoeba histolytica (in 9.4% of stool
Public health surveillance systems
samples) and Blastocystis hominis (in 8.1% of stool samples) were detected. In those coun-
Waterborne disease outbreak (WBDO)
tries that are likely affected most a lack of surveillance systems is noticeable. However,
Worldwide review
countries that established surveillance systems did not establish an international standardization of reporting systems. ª 2011 Elsevier Ltd. All rights reserved.
Contents 1. 2. 3. 4. 5.
1.
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Material and methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Introduction
Waterborne parasitic protozoan diseases have a worldwide distribution and are, in both developed and developing countries, reasons for epidemic and endemic human suffering
6603 6604 6604 6610 6612 6612
(Cotruva et al., 2004). They are one of the main reasons for 4 billion cases of diarrhea that causes annually 1.6 million deaths (www.who.int) and 62.5 million Disability Adjusted Life Years (DALYs) worldwide (Wright and Gundry, 2009). Diarrhea belongs to the five most common disease causes of
* Corresponding author. Tel.: þ49 221 478 5655; fax: þ49 221 478 3808. E-mail address:
[email protected] (P. Karanis). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.10.013
6604
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 0 3 e6 6 1 4
death (www.who.int) and is responsible for 21% of deaths of children younger than five years of age (Kosek et al., 2003). The most prevalent waterborne parasitic infections producing diarrhea are cryptosporidiosis and giardiasis, already the appearance of infections caused by Giardia lamblia account 2.8 10^8 cases yearly (Lane and Lloyd, 2002). Other parasitic protozoa with a waterborne transmission that cause human infections are Toxoplasma gondii, Entamoeba histolytica, Acanthamoeba spp., Cyclospora cayetanensis, Microsporidia, Isospora, Blastocystis hominis, Sarcocystis spp., Naegleria spp. and Balantidium coli. Since most of these protozoa use the faecal-oral way of transmission they can infect humans through sewage and contamination of land and rivers by animal or human feces (Lanata, 2003). Efficient sanitation and improved water supplies are the main safety measures against parasitic protozoan hazards. The United States created organizations like the Center for Disease Control and Prevention (CDC) and the US Environmental Protection Agency (USEPA) that enforced waterborne disease outbreak surveillance since 1971 (WBDOSS). In 1980 Sweden established a surveillance system (Stanwell-Smith et al., 2003) and in 1981 Japan started the National Epidemiological Surveillance of Infectious Diseases (NESID). The National Notifiable Diseases Surveillance System (NNDSS) in Australia was founded in 1990, the Health Protection Agency (HPA) of the United Kingdom in 2003, and furthermore, the Public Health Agency of Canada (PHAC) in 2004. Following the example of the USA European Countries created the European Center for Disease Control and Prevention (ECDC) in 2005. From most of these centers highly qualified information and documentations of waterborne parasitic protozoan outbreaks are approachable. In developing countries, governmental systems to register incidence and prevalence of protozoan infections or waterborne outbreaks are not established. Consequently there is a lack of documentation of waterborne parasitic protozoan outbreaks in developing countries. In 2007, the Journal of Water and Health published a review of worldwide waterborne outbreaks caused by parasitic protozoa (Karanis et al., 2007: Waterborne transmission of protozoan parasites: A worldwide review of outbreaks and lessons learnt) which spans a time period of almost hundred years from the beginning of the previous century. The aim of the present work is to update worldwide waterborne outbreaks of pathogenic protozoa during the time period of 2004e2011 and to figure out their worldwide distribution pattern.
2.
Material and methods
For the collection of data a variety of global literature sources was used. The research included articles from the databases MEDLINE/PubMed, MEDPILOT and Scopus as well as available electronic data from surveillance systems all over the world, like the Center for Disease Control and Prevention (CDC) and the European Center of Disease Prevention and Control (ECDC). The collection of data entailing this present review of waterborne parasitic protozoan outbreaks is based on a search of the medical literature databases MEDLINE/PubMed, MEDPILOT and Scopus, as well as on the use of electronic data from Morbidity and Mortality Weekly Report (MMWR by CDC), Euro
Surveillance (published by ECDC), Canada Communicable Disease Report (CCDR by PHAC), Communicable Disease Report (CDR by HPA) and CRYPTNET (www.mednetvet.org). In the named electronic databases the terms “outbreak (and) Cryptosporidium”, “outbreak (and) cryptosporidiosis”, “outbreak (and) Giardia”, “outbreak (and) giardiasis”, “outbreak (and) Cyclospora”, “outbreak (and) Blastocystis”, “outbreak (and) Entamoeba”, “outbreak (and) Acanthamoeba”, “outbreak (and) Amoebiasis”, “outbreak (and) Toxoplasma”, “outbreak (and) microsporidia”, “outbreak (and) Sarcocystis”, “outbreak (and) Naegleria”, “outbreak (and) Balantidium coli”, “outbreak (and) Isospora” were exerted and the listed articles critically reviewed.
3.
Results
During a time period of almost hundred years, between the previous century and 2004, a number of 325 waterborne protozoan parasitic outbreaks have been reported worldwide (Karanis et al., 2007), while in the considerable shorter time period of seven years, between 2004 and 2010, 199 reports of waterborne protozoan parasitic outbreaks were published. Between January 2004 and December 2010, one hundred and ninety-nine waterborne outbreaks of parasitic protozoan diseases occurring during this time period have been published worldwide and could be detected in the considered databases. The outbreaks are summarized in Tables 1e3. Table 1 documents worldwide waterborne outbreaks caused by Cryptosporidium spp., Table 2 shows worldwide waterborne outbreaks caused by G. lamblia and Table 3 summarizes the worldwide waterborne outbreaks caused by T. gondii, E. histolytica, Acanthamoeba spp., C. cayetanensis, Microsporidia, Isospora, Blastocystis spp., Sarcocystis spp., Naegleria spp. and Balantidium coli. Additional, Tables 1e3 point out the parameters of time (month and year), place (region and country), estimated cases and labor-confirmed cases in brackets if denoted in the original article. For each outbreak, the suspected cause and the key reference is annotated. In the reported outbreaks, Cryptosporidium spp. was the etiological agent in 60.3% (120) of the outbreaks, G. lamblia in 35.1% (70) and other protozoa in 4.5% (9). Four outbreaks (2%) were caused by T. gondii and three (1.5%) by C. cayetanensis. In two outbreaks (1%) Acanthamoeba was identified as causative agent. In one outbreak, four parasitic protozoa were implicated: G. lamblia (in 17.6% of stool samples) and Cryptosporidium parvum (in 2.7% of stool samples) as well as E. histolytica (in 9.4% of stool samples) and B. hominis (in 8.1% of stool samples) were detected. The outbreak occurred in Malaysia during April and May 2004 among Orang Asli (Aborigine) (Hakim et al., 2007). Summarizing the outbreaks, we count this outbreak as a single one, while it is mentioned in all three tables of each detected pathogen. From the Australian continent, 46.7% (93) of worldwide waterborne outbreaks caused by parasitic protozoa were reported. In New Zealand 80 outbreaks occurred (40.2%), in Australia 13 (6.5%). The waterborne outbreaks on the American continent amount to 33.1% (66) of worldwide waterborne outbreaks.
6605
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 0 3 e6 6 1 4
Table 1 e List of worldwide waterborne outbreaks caused by Cryptosporidium spp. Month/year Aug 2001
a
Location/country
Est. cases
Susp. cause
Nu¨rnberg/Germany
201 (15)
Potentially tap water during field exercise among military recruits Drinking water from a well in a factory G. lamblia (17.6%), Entamoeba (9.4%), Blastocystis hominis (8.1%), Cryptosporidium (2.7%)b/source of infection probably contaminated water or food Swimming pool Swimming pool
Jan 2004 ApreMay 2004
Ohio/USA Cameron Highlands/ Malaysia
82 Probably 79
Mar 2004 MayeJun 2004
4 (4) 7 (7)
Jun 2004
North West England Northern and Yorkshire/England Georgia/USA
14
Jul 2004
Ohio/USA
160
Jul 2004
Illinois/USA
37
Aug 2004
California/USA
336
AugeSep 2004
Nagano/Japan
41 (30)/74e288d
Aug 2004
Wisconsin/USA
6
Aug 2004 AugeSep 2004
Colorado/USA California/USA
6 >250
AugeNov 2004 Sep 2004 Oct 2004
133 8 10 (9)
2004
Bergen/Norway Illinois/USA Northern and Yorkshire/England New Zealand
Treated recreational water in a community pool Treated recreational water of a hotel pool Oocysts in backwash and sand from waterslides’ filter in a water park Additive during large giardiasis outbreak Treated recreational water in a hotel pool Swimming pool
8
Untreated water supply
2004 Mar 2005
Queensland/Australia South-east-Ireland
5 (5) 31
Public swimming pool Waterborne and person-to-person community outbreak
Apr 2005
West-Ireland
7
Waterborne
Jun 2005 Jun 2005
New York/USA Kentucky/USA
2307 53
Jun 2005 Jun 2005
Florida/USA Iowa/USA
47 24
Jun 2005
Kentucky/USA
9
Jul 2005
Kansas/USA
84
Jul 2005
Oregon/USA
20
JuleAug 2005
New York State/USA
>3000 (425)
JuleAug 2005 Aug 2005
Madison, Missouri/USA Ohio/USA
56 523
Aug 2005 Aug 2005 AugeSep 2005
New York/USA Louisiana/USA Copenhagen/Denmark
97 31 99 (13)
AugeDec 2005
London/England
84
Interactive fountain, state park Treated recreational water in community pools Treated recreational water in a hotel pool Treated recreational water in a community pool Treated recreational water in a community wading pool Treated recreational water in a water park pool Treated recreational water in a membership club pool Use of a recreational water interactive fountain Recreational pool water Treated recreational water in a community pool Treated recreational water in a camp pool Interactive fountain in a water park Carrots served in a basin with water at a canteen salad bar Swimming pools
Treated recreational water in a community pool Treated recreational water in a community wading pool Treated recreational water in a community pool, wading pool and interactive fountain Treated recreational water of a water park pool Swimming pool in a hotel
Key reference Brockmann et al., 2008 Liang et al., 2006 Hakim et al., 2007
Nichols et al., 2006 Nichols et al., 2006 Dziuban et al., 2006 Dziuban et al., 2006 Dziuban et al., 2006
Dziuban et al., 2006 Ichinohe et al., 2005, Yokoi et al., 2005, Takagi et al., 2008 Dziuban et al., 2006 Dziuban et al., 2006 Wheeler et al., 2006 Robertson et al., 2006a, b Dziuban et al., 2006 Nichols et al., 2006 Institute of Environmental Science and Research Ltd (ESR), New Zealand, 2004 Dale et al., 2010 Health Protection Surveillance Center (HPSC) 2006, Ireland Health Protection Surveillance Center (HPSC), Ireland Yoder et al., 2008 Yoder et al., 2008 Yoder et al., 2008 Yoder et al., 2008 Yoder et al., 2008 Yoder et al., 2008 Yoder et al., 2008 Schaffzin et al., 2006 Turabelidze et al., 2007 Yoder et al., 2008 Yoder et al., 2008 Yoder et al., 2008 Ethelberg et al., 2009 Nichols et al., 2006 (continued on next page)
6606
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 0 3 e6 6 1 4
Table 1 e (continued ) Month/year
Location/country
Est. cases
Sep 2005
Western Turkey
191 (15)
SepeNov 2005 SepeDec 2005
South East England North-west Wales
140 (140) 218
Oct 2005
New York/USA
22
Nov 2005
100e232
2005 2005 2005
Gwynedd and Anglesey, northwest Wales/UK Kentucky, Ohio/USA New South Wales/Australia New Zealand
2005 2005 2005
Victoria/Australia Victoria/Australia Victoria/Australia
9 (9) 6 (5) 20 (18)
Jan 2006 May 2006
England Florida/USA
14 (6) 55
May 2006
South West England
35
Jun 2006
Missouri/USA
116
Jun 2006
Wyoming/USA
29
Jun 2006
Pennsylvania/USA
13
JuneOct 2006
Wyoming/USA
34
JuneNov 2006
85
Jul 2006
Charleston region, South Carolina/USA Montana/USA
Jul 2006
Illinois/USA
65
Jul 2006
Louisiana/USA
29
Jul 2006
South-Ireland
28
Jul 2006
California/USA
16
Jul 2006
South Carolina/USA
12
Jul 2006
Missouri/USA
6
Jul 2006
South-Ireland
2
JuleAug 2006
Louisiana/USA
35
Aug 2006
Wisconsin/USA
22
Aug 2006
Georgia/USA
19
Aug 2006
Illinois/USA
18
>800 254 (254) 5 outbreaks, 17 cases
82
Susp. cause Public drinking water supply contaminated by sewage or animal waste following heavy rainfall Public water supply Drinking unboiled tap water, washing fruit or salad, fruit juice drinks, and other transmission ways Treated recreational water in a membership club pool Drinking water from several sources, incl. Lake Cwellyn Multiples modes of transmission Public swimming pool 63% untreated water supply, 22.2% contamination of water source, 7.4% contamination of reservoirs, 37% unknown factors (percentages relate to a total number of 27 waterborne outbreaksc) Public swimming pool Unknown swimming pool Suspected waterborne, public Swimming pool Swimming pool Treated recreational water, interactive fountain Faecally contaminated surface water, consumption of water from private well Treated recreational water, interactive fountain, water park Multiple community pools and untreated reservoir Treated recreational water, pool of a membership club Recreational water use at any public swimming pool and one local reservoir Recreational water venues and 13 day care centers Treated recreational water of community pools Treated recreational water in a water park of a day camp Treated recreational water, interactive fountain, water park pool Waterborne
Treated recreational water, interactive fountain Treated recreational water of a community pool Treated recreational water of a community pool Waterborne, private house
Recreational water use at commercial water park Treated recreational water of a community pool Treated recreational water in a community pool Treated recreational water of a water park
Key reference Aksoy et al., 2007
Nichols et al., 2006 Mason et al., 2010
Yoder et al., 2008 Carnicer-Pont et al., 2005; Nichols et al., 2006 Yoder et al., 2008 Dale et al., 2010 Institute of Environmental Science and Research Ltd (ESR), New Zealand, 2005
Dale et al., 2010 Dale et al., 2010 Dale et al., 2010 Davison, 2006 Yoder et al., 2008 Hoek et al., 2008 Yoder et al., 2008 Yoder et al., 2008 Yoder et al., 2008 Alden et al., 2007 Alden et al., 2007 Yoder et al., 2008 Yoder et al., 2008 Yoder et al., 2008 Health Protection Surveillance Center (HPSC) 2007, Ireland Yoder et al., 2008 Yoder et al., 2008 Yoder et al., 2008 Health Protection Surveillance Center (HPSC) 2007, Ireland Alden et al., 2007 Yoder et al., 2008 Yoder et al., 2008 Yoder et al., 2008
6607
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 0 3 e6 6 1 4
Table 1 e (continued ) Month/year
Location/country
Est. cases
Susp. cause
Aug 2006
Colorado/USA
15
Birthday party at indoor community swimming pool Day camp pool and community water park Treated recreational water of a community pool Treated recreational water of a water park Treated recreational water in a hotel pool Treated recreational water in school pools Drinking water from a well (in a church) Interactive water fountain in neighborhood Waterborne
Aug 2006
Illinois/USA
7
Aug 2006
Wisconsin/USA
4
Aug 2006
Illinois/USA
4
Aug 2006
Florida/USA
3
Sep 2006
Minnesota/USA
47
Sep 2006
Ohio/USA
10
Sep 2006
Florida/USA
49 (9)
Oct 2006
West-Ireland
6
Oct 2006
Georgia/USA
4
Nov 2006
South-east-Ireland
8
Treated recreational water in a community pool Waterborne community outbreak
Dec 2006
USA
7
Waterborne community outbreak
2006
Victoria/Australia
134 (134)
2006
Victoria/Australia
30 (11)
2006 2006
South Australia Victoria/Australia
19 (14) 18 (18)
2006
New Zealand
6 outbreaks, 15 cases
2006
Western Australia
14 (14)
2006
Victoria/Australia
10 (7)
Jan 2007
South-east-Ireland
7
Suspected waterborne, unknown swimming pool Suspected waterborne, school swimming pool Waterborne, unspecified source Suspected waterborne, public swimming pool 83.3% untreated water supply, 44.4% contamination of water source, 5.6% contamination of reservoirs, 5.6% unknown factors (percentages relate to a total number of 18 waterborne outbreaksc) Suspected waterborne, public swimming pool Suspected waterborne, public swimming pool Waterborne community outbreak
Feb 2007
182 (98)
Mar 2007
City and county of Galway/Ireland West-Ireland
304
Public water supply using Lough Corrib (a large lake) Waterborne community outbreak
Mar 2007
USA
186
Waterborne community outbreak
Mar 2007
Norway
25 (10)
Jun 2007
USA
2
Drinking water from hotel dispensers and tap water, consuming ice cubes in hotel bar, eating broccoli soup Waterborne, private house
JuneDec 2007 Jul 2007
Utah/USA South-Ireland
1902 2
Treated recreational water venues Waterborne, private house
Key reference Boehmer et al., 2009 Alden et al., 2007 Yoder et al., 2008 Yoder et al., 2008 Yoder et al., 2008 Yoder et al., 2008 Yoder et al., 2008 Eisenstein, 2008 Health Protection Surveillance Center (HPSC) 2008, Ireland Yoder et al., 2008 Health Protection Surveillance Center (HPSC) 2008, Ireland Departments of Public Health 2007 Dale et al., 2010 Dale et al., 2010 Dale et al., 2010 Dale et al., 2010 Institute of Environmental Science and Research Ltd (ESR), New Zealand, 2006
Dale et al., 2010 Dale et al., 2010 Health Protection Surveillance Center (HPSC) 2008, Ireland Pelly et al., 2007 Health Protection Surveillance Center (HPSC) 2008, Ireland Departments of Public Health 2007 Hajdu et al., 2008
Departments of Public Health 2007 Rolfs et al., 2008 Health Protection Surveillance Center (HPSC), Ireland (continued on next page)
6608
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 0 3 e6 6 1 4
Table 1 e (continued ) Month/year
Location/country
Est. cases
Susp. cause
Aug 2007 Aug 2007
Idaho/USA Stockholm/Sweden
50 23 (9)
OcteDec 2007 2007
Staffordshire/England Galway/Ireland
57 (39) 182
2007
New Zealand
5 outbreaks, 16 cases
2007
4 (2)
Jan 2008
Northern Territory/ Australia West-Ireland
Mar 2008
South-Ireland
2
Waterborne and person-to-person in a private house
JuneJul 2008
Northamptonshire/UK
33 (23)
Jul 2008
South-east-Ireland
2
Drinking water supplied from the Pitsford Reservoir Waterborne and animal transmission in a private house
2008
New Zealand
4 outbreaks, 17 cases
2009
New Zealand
5 outbreaks, 13 cases
Nov 2010
¨ stersund/Sweden O
10,000
3
Recreational water at splash parks Drinking water from a well at a camping site close to a field with sheep Swimming pool associated Heavy precipitation of historic proportions and the water source reaching the highest level on record 80% untreated water supply, 20% contamination of water source, 6.7% treatment process failure, 6.7% contamination of reservoirs, 6.7% post treatment contamination, 13.3% unknown factors (percentages relate to a total number of 15 waterborne outbreaksc) Suspected waterborne, public swimming pool Waterborne, private house
61.5% contamination of water source, 50% untreated water supply, 11.5% contamination of reservoirs, 15.4% unknown factors (percentages relate to a total number of 26 waterborne outbreaksc) 66.7% untreated drinking water supply, 20.8% contamination of reservoirs, 16.7% contamination of water sources, 20.8% unknown factors (percentages relate to a total number of 24 waterborne outbreaksc) Contaminated water supply
Key reference Jue and Schmalz, 2009 Persson et al., 2007
Coetzee et al., 2008 Pelly et al., 2007
Institute of Environmental Science and Research Ltd (ESR), New Zealand, 2007
Dale et al., 2010 Health Protection Surveillance Center (HPSC) 2009, Ireland Health Protection Surveillance Center (HPSC) 2009, Ireland Smith et al., 2010 Health Protection Surveillance Center (HPSC), Ireland Institute of Environmental Science and Research Ltd (ESR), New Zealand, 2008
Institute of Environmental Science and Research Ltd (ESR), New Zealand, 2009
Sveriges Radio 2010
( ): laboratory confirmed cases. a outbreak occurred before 2004, but was published at 2010 and is not included in the review of Karanis et al. (2007). b infectious rate of tested stool samples. c for some waterborne outbreaks more than one contributing factor was recorded. d variation between authors.
From the 61 outbreaks in North America (30.6%) 60 appeared in the United States (30.1%) and 1 in Canada (0.5%). The documented five outbreaks in South America (2.5%) occurred in Peru (two outbreaks, 1%), in Brazil (two outbreaks, 1%) and in French Guiana (one outbreak, 0.5%). Europe contributes 16.5% (33) of worldwide waterborne outbreaks. The distribution within the European countries is as follows: Ireland 6.5% (13) of worldwide waterborne outbreaks, the United Kingdom 5.5% (11), Norway 2% (4),
Sweden 1% (2) and at least 0.5% (1) in Finland, Denmark and Germany, respectively. In Asia 7 (3.5%) waterborne outbreaks of parasitic protozoan diseases were reported, 3 (1.5%) in Turkey and 1 (0.5%) in each of the following countries: Japan, China, India and Malaysia. The worldwide distribution of waterborne outbreaks caused by parasitic protozoa on the individual continents is presented in Fig. 1, while Fig. 2 shows the distribution across countries.
6609
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 0 3 e6 6 1 4
Table 2 e List of worldwide waterborne outbreaks caused by Giardia lamblia. Month/year
Location/country
Est. cases
Nov 2003eJul 2004 Mar 2004 AprileMay 2004
Trondheim/Norway Missouri/USA Cameron Highlands /Malaysia
12 9 (Probably 79)
MayeSep 2004 Jun 2004 AugeOct 2004
Ohio/USA Vermont/USA Bergen/Norway
3e1450 11 2500 (1300)
2004
New Zealand
11 outbreaks, 34 cases
Jul 2005
California/USA
3
Aug 2005 Oct 2005 2005
41 196 (7) 12 (3)
2005
California/USA Izmir/Turkey New South Wales /Australia New Zealand
May 2006 Sep 2006 2006
Colorado/USA Florida/USA New Zealand
6 49 (38) 7 outbreaks, 22 cases
JuleAug 2007
California/USA
50 (26)
Sep 2007 Nov 2007e Feb 2008 2007
New Hamshire/USA Nokia/Finland
31 ND
Northern Ontario /Canada
ND
2007
New Zealand
7 outbreaks, 32 cases
2008
New Zealand
14 outbreaks, 63 cases
2009
New Zealand
13 outbreaks, 44 cases
9 outbreaks, 34 cases
Susp. cause/parasite Waterborne, child day care center Untreated recreational water, lake G. lamblia (17.6%), Entamoeba (9.4%), Blastocystis hominis (8.1%), Cryptosporidium (2.7%)a/source of infection probably contaminated water or food Sewage contaminated groundwater Drinking water from a well in a camp Leaking sewage pipes and insufficient water treatment 8 outbreaks caused untreated water supplies, 1 by inadequate water source, 2 by multiple sources Water not intended to drink in a private residence Drinking water Faecal contamination in public water supply Suspected rainwater tank or bore water in a health spa resort 63% untreated water supply, 22.2% contamination of water source, 7.4% contamination of reservoirs, 37% unknown factors (percentages relate to a total number of 27 waterborne outbreaksb) River water in wilderness Interactive water fountain in neighborhood 83.3% untreated water supply, 44.4% contamination of water source, 5.6% contamination of reservoirs, 5.6% unknown factors (percentages relate to a total number of 18 waterborne outbreaksb) Unsterilized sand in slow sand water filtration system in a private recreational camp Community drinking water associated (tap water) Sewage contamination of drinking water distribution network Water source in a tree-planting camp
80% untreated water supply, 20% contamination of water source, 6.7% treatment process failure, 6.7% contamination of reservoirs, 6.7% post treatment contamination, 13.3% unknown factors (percentages relate to a total number of 15 waterborne outbreaksb) 61.5% contamination of water sources, 50% untreated water supply, 11.5% contamination of reservoirs, 15.4% unknown factors (percentages relate to a total number of 26 waterborne outbreaksb) 66.7% untreated drinking water supply, 20.8% contamination of reservoirs, 16.7% contamination of water sources, 20.8% unknown factors (percentages relate to a total number of 24 waterborne outbreaksb)
Key reference Wahl and Bevanger, 2007 Dziuban et al., 2006 Hakim et al., 2007
O’Reilly et al., 2007 Liang et al., 2006 Nyga˚rd et al., 2006, Robertson et al., 2006a, b; Strand et al., 2008 Institute of Environmental Science and Research Ltd (ESR), New Zealand, 2004 Yoder et al., 2008 Yoder et al., 2008 Tuncay et al., 2008 Dale et al., 2010 Institute of Environmental Science and Research Ltd (ESR), New Zealand, 2005 Yoder et al., 2008 Eisenstein, 2008 Institute of Environmental Science and Research Ltd (ESR), New Zealand, 2006 Karon et al., 2010 Daly et al., 2009 Rimhanen-Finne et al., 2010 C-EnterNet Annual Report 2007, Public Health Agency of Canada Institute of Environmental Science and Research Ltd (ESR), New Zealand, 2007
Institute of Environmental Science and Research Ltd (ESR), New Zealand, 2008
Institute of Environmental Science and Research Ltd (ESR), New Zealand, 2009
ND: no data. ( ): laboratory confirmed cases. a infectious rate of tested stool samples. b for some waterborne outbreaks more than one contributing factor was recorded.
In this review, 36.2% (72) of the documented worldwide outbreaks caused by protozoan pathogens had multiple ways of transmission, e.g. untreated water supplies, contamination of water sources, treatment process failures, contamination of
reservoirs and post treatment contamination. In 33.7% (67 outbreaks), the transfer through recreational water was detected as the source of infection, mainly due to the contamination with Cryptosporidium spp. in 32.7% (65) of
Mao et al., 2009 Khan et al., 2006 31 (23) 8 Jiangshan City/China Agronomica/Brazil MayeJun 2006 ND/between 2003 and 2005
112 (27) 138 37 (7) 191 (9) Lima/Peru 35 states and Puerto Rico Lima/Peru Izmir/Turkey Nov 2004 Jan 2005eMay 2007 Mar 2005 Sep 2005
249 (178) Coimbatore City/India Aug 2004eJul 2005
Fig. 1 e Distribution of worldwide waterborne outbreaks caused by parasitic protozoan between 2004 and 2010 by continent.
described outbreaks. In 20.6% (41) of the listed outbreaks, drinking water systems were contaminated with a protozoan pathogen. In 9.5% (19 outbreaks) of the reported outbreaks the way of transmission was not mentioned.
4.
( ): labor-confirmed cases. ND : no data. a outbreak occurred before 2004, but was published at 2010 and is not included in the review of Karanis et al. (2007). b infectious rate of tested stool samples.
Palanisamy et al., 2006, Balasundaram et al., 2010 Torres-Slimming et al., 2006 MMWR 2007 Mundaca et al., 2008 Aksoy et al., 2007
Vaudaux et al., 2010 Joslin et al., 2006 Demar et al., 2007 Hakim et al., 2007
T. gondii/contaminated cistern, municipal water supply Acanthamoeba (Keratitis)/probably waterborne T. gondii/unknown source G. lamblia (17.6%), Entamoeba (9.4%), B. hominis (8.1%), Cryptosporidium (2.7%)b/source of infection probably contaminated water or food T. gondii (ocular)/probably contaminated supplying water after heavy rainfall C. cayetanensis/way of transmission unclear Acanthamoeba (Keratitis)/probably waterborne C. cayetanensis/unknown way of transmission Cyclospora, co-infection with outbreak of cryptosporidiosis/ contaminated public drinking water supply E. histolytica T. gondii/non-treated water in common neighborhood 20 40 11 Probably 79 Santa Isabel do Ivai/Brazil Chicago-Gary-Kanosha area, Illinois/USA Patam/French Guiana Cameron Highlands/Malaysia OcteDec 2001 Jun 2003eNov 2005 Dec 2003eJan 2004 AprileMay 2004
Est. cases Location/country
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 0 3 e6 6 1 4
a
Month/year
Table 3 e List of worldwide waterborne outbreaks caused by other parasitic protozoa.
Parasite/susp. cause
Key reference
6610
Discussion
The gathering of worldwide waterborne parasitic protozoan outbreaks illustrates their global distribution pattern. The present survey shows a large number of outbreaks in a short time period, 199 worldwide outbreaks in 7 years, while the previous overview of Karanis et al. (2007) presented 325 reported outbreaks in 100 years. This significant difference in the number of reported outbreaks is caused by the substantial improvements of data reporting and the establishment of surveillance systems in developed countries. Both achievements of the developed countries are also the reasons for the distortion in the reflection of the global distribution pattern. The highest prevalence of parasitic protozoan infections is known to occur in developing countries due to their low hygiene standards. Thus, the highest rate of waterborne parasitic protozoan outbreaks should be estimated in these countries. Though, we found a higher rate of reported waterborne protozoan outbreaks in developed nations. These findings are due to the better technological and logistical possibilities of the developed countries. As the collection of data depends on the detection, investigation and reporting system of the outbreaks (Leclerc et al., 2002) many waterborne parasitic protozoan outbreaks stay unrecognized or unreported. In 1990, Craun estimated that
Fig. 2 e Distribution of worldwide waterborne outbreaks caused by parasitic protozoa between 2004 and 2010 by country.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 0 3 e6 6 1 4
only a low fractional amount of waterborne outbreaks affecting the United States are detected and reported, possibly as low as one-tenth (Craun, 1990). In 1995, Morris and Levin suggested that the annual incidence of waterborne infectious diseases in the United States could even be as high as 8 million cases of illness and around 1200 cases of death (Morris and Levin, 1995). Likewise, in the Annual Epidemiological Report on Communicable Diseases in Europe 2009, ECDC infectious diseases due to protozoan are described to be underreported, e.g. the data of cryptosporidiosis from several EU countries are lacking and several countries have not established a surveillance system for toxoplasmosis (www.ecdc.europa.eu). The National Institute of Hygiene in Poland (PZH) does not report human cases of cryptosporidiosis although they are registered and the annual number of human giardiasis is probably underestimated (Bajer, 2008). As water is the transmission route of the referred protozoa Giardia cysts and Cryptosporidium oocysts are widely distributed in aquatic ecosystems and could be detected in 81% and 87% of raw water samples of 66 surface water treatment plants in the United States 1991 (LeChevallier et al., 1991). Worldwide, G. lamblia is one of the most prevalently identified causative pathogen in waterborne disease outbreaks (WBDOs). Waterborne outbreaks of giardiasis are usually associated with ingestion of contaminated surface water or groundwater while the contamination of recreational water has also been noticed (Marshall et al., 1997). In developed countries G. lamblia and Cryptosporidium spp. are the most common waterborne pathogens. Slifko et al. (2000) described three reasons for this: domestic animals are often infected with cryptosporidiosis or giardiasis, the aquatic ecosystem is contaminated by the high frequency of environmental burden and lastly, Giardia cysts and Cryptosporidium oocysts are resistant against commonly used water disinfectants (Slifko et al., 2000). In the early 1900s waterborne diseases were accountable for 25% of deaths due to infectious diseases. Chlorination, the first used disinfection for public water supplies, lead to a dramatic decrease of waterborne diseases in the United States (Cutler and Miller, 2005). The WHO introduced treatment guidelines for the quality of drinking water which includes the treatment of water intended to drink with plain sedimentation, pre-filtration, slow sand filtration, coagulation, flocculation and sedimentation, rapid sand filtration, aeration and disinfection via boiling, UV radiation, chlorine and ozone. For water treatment plants multi-barrier systems are recommended (www.who.int). After the introduction of membrane filtration in North West England, the incidence of cryptosporidiosis decreased around 79% (Goh et al., 2005). The CDC in collaboration with the Pan American Health Organization (PAHO) generated a Safe Water System (SWS) which includes three elements: water treatment with sodium hypochlorite at the point-of-use (reduction of diarrheal disease incidence: 26e84%), storage of water in safe containers, as well as advanced hygiene and water handling practices. Since 1998 national, regional and local SWS projects have been implemented with NGO and government partners in over 30 countries. A systematic review of interventions to improve water quality for preventing diarrhea published 2007 by Clasen pointed out that interventions generally are effective (Clasen et al., 2007).
6611
USEPA and CDC established protection and prevention guidelines as well as national statistics of waterborne outbreaks caused by different agents in the United States. In 1994, USEPA published the Cryptosporidium Criteria Document (www.epa.gov) and thus declared Cryptosporidium as a primary drinking water contaminant. The attention on this protozoon raised and led to more investigation. Beside traditional microscopy and staining methods, enzyme immunoassays (EIAs), direct fluorescent-antibody (DFA) and the polymerase chain reaction (PCR) have become established as sensitive and specific detection methods (Morgan et al., 1996). As a result of the enhanced surveillance the USEPA announced the Interim Enhanced Surface Water Treatment Rule to minimize the level of Cryptosporidium spp. in finished water and estimated that the incidence of endemic illness caused by Cryptosporidium spp. will decrease by 463,000 cases each year (www.epa. gov). Due to the advertence and the surveillance system, the data reported for the United States are the most complete of waterborne diseases worldwide (Leclerc et al., 2002). The increasing quality of documentation led to a rise in reported outbreaks during recent years, especially in cryptosporidiosis and giardiasis (Craun et al., 2005). Nevertheless, the supreme incidence of waterborne parasitic protozoan infection is suggested to occur in developing countries. In Latin America, Asia and Africa around 600 million people live in unhealthy homes (Cotruva et al., 2004), 1.1 billion people lack access to improved water supplies and 2.6 billion people lack adequate sanitation (www.who.int). Hence high prevalence rates of waterborne infectious diseases can be expected in developing countries where water supply and waste disposal are deficient. Lanata (2003) stated that the majority of giardiasis outbreaks occur in Latin America, Africa and Asia, with about 5 105 new cases each year. However, gastrointestinal infections are under-diagnosed in developing countries and the prevalence is underestimated (Lanata, 2003). Current and Garcia (1991) recorded prevalence rates of Cryptosporidium in stool samples of patients with gastroenteritis of 1e4% in Europe and North America, while the prevalence rates in Africa, Asia, Australia and South America amount to between 3 and 20%. They also detected high rates of asymptomatic carriage of Cryptosporidium (10e30%) in developing countries compared to low rates of <1% in developed countries. Higher prevalence rates in developing countries can also be estimated for the other waterborne parasitic protozoan infections. However, most studies about prevalence of parasitic protozoan infections have been made in developed countries where health infrastructure and laboratory testing are more accessible than in developing countries (Mak, 2004). Yotthanooi and Choonpradub (2010) consider that data of diarrheal diseases incidence in Thailand are known to be underreported and the causal agents of diarrheal diseases are undiscovered. But although surveillance systems are established in developed countries, an international agreement of reporting structure is still missing. As Ho¨rmann et al. (2004) criticizes, the systems of surveillance and reporting are fundamentally different between various countries and a comparison of data is not always possible (Ho¨rmann et al., 2004). While the CDC registers each waterborne outbreak by agent, location and number of affected persons, European surveillance systems
6612
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 0 3 e6 6 1 4
are used to figure out national infection rates and incidences meanwhile neglecting the itemization of waterborne outbreaks. The index of human cryptosporidiosis and giardiasis is listed annually by Euro Surveillance but the sources of infection remain undiscovered. The number of human cases of giardiasis in the Ukraine was 30751 in 2007, 31638 in 2008 and even 32928 in 2009. However, these data doesn’t allow precise conclusions about their origin and no further information about their consistence is given. It remains unclear whether these data reflect incidence or prevalence or both and whether they also include outbreaks (www.eurosurveillance. org). In Japan, Taniguchi described the incidence rate of amebiasis with an annual average of 3.18 per 1,000,000 people in 1999e2008 as second highest nationwide (Taniguchi et al., 2008) but outbreaks and cases were not described. Furthermore, Craun et al. (2005) declared that “even in countries with surveillance systems, outbreak investigation activities have frequently been unable to identify sources of infection and etiologic agents”. This statement raises the question of the quality of surveillance. Even countries that already support a surveillance system should improve their methods of detection and disease diagnosis. The incorporation of molecular investigative tools including detection/diagnosis and genotyping and the introduction of new tools (OrtegaPierres et al., 2009; Karanis and Ongerth, 2009) will contribute to a better surveillance of waterborne parasitic outbreaks.
5.
Conclusions
The number of waterborne parasitic outbreaks is still increasing due to the better surveillance and reporting systems in several countries and continents. Since this review leans on documented waterborne parasitic protozoan outbreaks listed in worldwide databases and reported in single studies it can only give an overview of the detected and reported outbreaks. Quantity and intensity of the undiagnosed outbreaks stay uncovered. Data about those countries that are probably concerned most are lacking. The introduction of surveillance systems in these countries would be helpful to detect, and as a result to combat, parasitic protozoa with the final aim to improve the health of the population. Therefore it is necessary to develop reliable and accessible diagnostic tools for the detection of the causative organisms especially in these countries. Further investigation of emerging methods is required to provide more results concerning human infections caused by parasitic organisms. Additionally, an international standardized reporting system should be established in all affected countries. Standardized databases could lead to a closer and more successful collaboration in the battle against waterborne pathogenic protozoa.
references
Aksoy, U., Akisu, C., Sahin, S., Usluca, S., Yalcin, G., Kuralay, F., Oral, A., 2007. First reported waterborne outbreak of cryptosporidiosis with Cyclospora co-infection in Turkey. Euro Surveillance 12 (7), 3142.
Alden, N., Ghosh, T., Vogt, R., Andreasen, C., Buller-Fenton, S., Saathoff-Huber, L., Davis, J., Henry, S., Ratard, R., Roach, A., Drociuk, D., Meredith, J., Ball, R., Baker, L., Grandpre, J., Murphy, T., Van Houten, C., Beach, M., Bishop, H., DaSilva, A., Hill, V., Xiao, L., Boehmer, T., Brunkard, J., Christian, K., Luce, R., 2007. Cryptosporidiosis outbreaks associated with recreational water use e five states, 2006. Morbidity and Mortality Weekly Report 56, 729e732. Bajer, A., 2008. Cryptosporidium and Giardia spp. infections in humans, animals and the environment in Poland. Parasitology Research 104, 1e17. Balasundaram, M.B., Andavar, R., Palaniswamy, M., Venkatapathy, N., 2010. Outbreak of acquired ocular toxoplasmosis involving 248 patients. Archives of Ophthalmology 128, 28e32. Boehmer, T.K., Alden, N.B., Ghosh, T.S., Vogt, R.L., 2009. Cryptosporidiosis from a community swimming pool: outbreak investigation and follow-up study. Epidemiology and Infection 137, 1651e1654. Brockmann, S.O., Jakobi, V., Dreweck, C., Wagner-Wiening, C., Hagen, R.M., Kimmig, P., Petry, F., 2008. Serological and epidemiological analysis of an outbreak of gastroenteritis among Military recruits in Germany caused by Cryptosporidium parvum. Infection 36, 450e457. C-EnterNet Annual Report, 2007. Outbreak-Related Cases. Public Health Agency of Canada. Carnicer-Pont, D., Atenstaedt, R., Walker, M., Chalmers, R., Rees, A., Rowlands, K., Evans, M., Salmon, R., 2005. An outbreak of cryptosporidiosis in Wales, November 2005. Euro Surveillance 10 (49), 2854. Clasen, T., Schmidt, W., Rabie, T., Roberts, I., Cairncross, S., 2007. Interventions to improve water quality for preventing diarrhoea: systematic review and meta-analysis. BMJ 334 (7597), 782. Coetzee, N., Edeghere, O., Orendi, J., Chalmers, R., Morgan, L., 2008. A swimming pool-associated outbreak of cryptosporidiosis in Staffordshire, England, October to December 2007. Euro Surveillance 13 (45), 19028. Cotruva, J.A., Durfour, A., Rees, G., Bartram, J., Carr, R., Cliver, D.O., Craun, G.F., Fayer, R., Gannon, V.P.J., 2004. Waterborne zoonoses: identification, causes and control. World Health Organisation, IWA Publishing. Craun, G., 1990. Methods for the investigation and prevention of waterborne disease outbreaks. USEPA. Craun, G., Calderon, R., Craun, M., 2005. Outbreaks associated with recreational water in the United States. International Journal of Environmental Health Research 15, 243e262. Current, W., Garcia, L., 1991. Cryptosporidiosis. Clinical Microbiology Reviews 4, 325e358. Cutler, D., Miller, G., 2005. The role of public health improvements in health advances: the twentieth-century United States. Demography 42, 1e22. Dale, K., Kirk, M., Sinclair, M., Hall, R., Leder, K., 2010. Reported waterborne outbreaks of gastrointestinal disease in Australia are predominantly associated with recreational exposure. Australian and New Zealand Journal of Public Health 34, 527e530. Daly, E.R., Roy, S.J., Blaney, D.D., Manning, J.S., Hill, V.R., Xiao, L., Stull, J.W., 2009. Outbreak of giardiasis associated with a community drinking-water source. Epidemiology and Infection 138, 491e500. Davison, H., 2006. A swimming pool associated outbreak of cryptosporidiosis. Health Protection Agency North West. Quarterly Health Protection Bulletin, 2e4. Oct- Dec 2006. Demar, M., Ajzenberg, D., Maubon, D., Djossou, F., Panchoe, D., Punwasi, W., Valery, N., Peneau, C., Daigre, J.L., Aznar, C., Cottrelle, B., Terzan, L., Darde, M.L., Carme, B., 2007. Fatal outbreak of human toxoplasmosis along the maroni river:
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 0 3 e6 6 1 4
epidemiological, clinical, and parasitological aspects. Clinical Infectious Diseases 45, e88e95. Dziuban, E., Liang, J., Craun, G., Hill, V., Yu, P., Painter, J., Moore, M., Calderon, R., Roy, S., Beach, M., 2006. Surveillance for waterborne disease and outbreaks associated with recreational water d United States, 2003e2004. Morbidity and Mortality Weekly Report 55, 1e30. Eisenstein, L., 2008. Outbreak of giardiasis and cryptosporidiosis associated with a neighborhood interactive water fountain-Florida, 2006. Journal of Environmental Health 71, 18e22. Ethelberg, S., Lisby, M., Vestergaard, L.S., Enemark, H.L., Olsen, K.E.P., Stensvold, C.R., Nielsen, H.V., Porsbo, L.J., Plesner, A.M., MØLbak, K., 2009. A foodborne outbreak of Cryptosporidium hominis infection. Epidemiology and Infection 137, 348e356. Goh, S., Reacher, M., Casemore, D., Verlander, N., Charlett, A., Chalmers, R., Knowles, J., Pennington, A., Williams, J., Osborn, K., Richardstt, S., 2005. Sporadic cryptosporidiosis decline after membrane filtration of public water supplies, England, 1996-2002. Emerging Infectious Diseases 11, 251e259. Hajdu, A., Vold, L., Østmo, T.A., Helleve, A., Helgebostad, S.R., Krogh, T., Robertson, L., de Jong, B., Nyga˚rd, K., 2008. Investigation of Swedish cases reveals an outbreak of cryptosporidiosis at a Norwegian hotel with possible links to in-house water systems. BMC Infectious Disease 8, 152. Hakim, S., Gan, C., Malkit, K., Azian, M., Chong, C., Shaari, N., Zainuddin, W., Chin, C., Sara, Y., Lye, M., 2007. Parasitic infections among Orang Asli (Aborigine) in the Cameron Highlands, Malaysia. Southeast Asian Journal of Tropical Medicine and Public Health 38, 415e419. Health Protection Surveillance Center (HPSC), 2006. HPSC Annual Report 2005. Health Protection Surveillance Center (HPSC), 2007. Epidemiology of Cryptosporidiosis in Ireland 2006. Health Protection Surveillance Center (HPSC), 2008. HPSC Annual Report 2007. Health Protection Surveillance Center (HPSC), 2009. HPSC Annual Report 2008. Hoek, M.R., Oliver, I., Barlow, M., Heard, L., Paynter, S., Chalmers, R., 2008. Outbreak of Cryptosporidium parvum among children after a school excursion to an adventure farm, South West England. Journal of Water and Health 6, 333e338. Ho¨rmann, A., Korpela, H., Sutinen, J., Wedel, H., Ha¨nninen, M.L., 2004. Meta-analysis in assessment of the prevalence and annual incidence of Giardia spp. and Cryptosporidium spp. infections in humans in the Nordic countries. International Journal for Parasitology 34, 1337e1346. Ichinohe, S., Fukushima, T., Kishida, K., Sanbe, K., Saika, S., Ogura, M., 2005. Secondary transmission of cryptosporidiosis associated with swimming pool use. Japanese Journal of Infectious Diseases 58, 400e401. Institute of Environmental Science and Research Ltd (ESR), 2005. Annual Summary of Outbreaks in New Zealand 2004. Institute of Environmental Science and Research Ltd (ESR), 2006. Annual Summary of Outbreaks in New Zealand 2005. Institute of Environmental Science and Research Ltd (ESR), 2007. Annual Summary of Outbreaks in New Zealand 2006. Institute of Environmental Science and Research Ltd (ESR), 2008. Annual Summary of Outbreaks in New Zealand 2007. Institute of Environmental Science and Research Ltd (ESR), 2009. Annual Summary of Outbreaks in New Zealand 2008. Institute of Environmental Science and Research Ltd (ESR), 2010. Annual Summary of Outbreaks in New Zealand 2009. Joslin, C., Tu, E., McMahon, T., Passaro, D., Stayner, L., Sugar, J., 2006. Epidemiological characteristics of a Chicago-area Acanthamoeba Keratitis outbreak. American Journal of Ophthalmology 142, 212e217.
6613
Jue, R., Schmalz, T., 2009. Outbreak of cryptosporidiosis associated with a splash park e Idaho, 2007. Morbidity and Mortality Weekly Report 58, 615e618. Karanis, P., Ongerth, J., 2009. LAMPea powerful and flexible tool for monitoring microbial pathogens. Trends in Parasitology 25, 498e499. Karanis, P., Kourenti, C., Smith, H., 2007. Waterborne transmission of protozoan parasites: a worldwide review of outbreaks and lessons learnt. Journal of Water and Health 5, 1e38. Karon, A., Hanni, K., Mohle-Boetani, J., Beretti, R., Hill, V., Arrowood, M., Johnston, S., Xiao, L., Vugia, D., 2010. Giardiasis outbreak at a camp after installation of a slow-sand filtration water-treatment system. Epidemiology and Infection 29, 1e5. Khan, A., Jordan, C., Muccioli, C., 2006. Genetic Divergence of Toxoplasma gondii strains associated with ocular toxoplasmosis, Brazil. Emerging Infectious Diseases 12, 942e950. Kosek, M., Bern, C., Guerrant, R., 2003. The global burden of diarrhoeal disease, as estimated from studies published between 1992 and 2000. Bulletin of the World Health Organization 81, 197e204. Lanata, C., 2003. Studies of food hygiene and diarrhoeal disease. International Journal of Environmental Health Research 13 (Suppl. 1), S175eS183. Lane, S., Lloyd, D., 2002. Current trends in research into the waterborne parasite Giardia. Critical Reviews in Microbiology 28, 123e147. LeChevallier, M., Norton, C., Luo, R., 1991. Occurrence of Giardia and Cryptosporidium spp. in surface water supplies. Applied and Environmental Microbiology. 57, 2610e2616. Leclerc, H., Schwartzbrod, L., Dei-Cas, E., 2002. Microbial agents associated with waterborne diseases. Critical Reviews in Microbiology 28, 371e401. Liang, J., Dziuban, E., Craun, G., Hill, V., Moore, M., Gelting, R., Calderon, R., Beach, M., Roy, S., 2006. Surveillance for waterborne disease and outbreaks associated with drinking water and water not intended for drinking e United States, 2003e2004. Morbidity and Mortality Weekly Report 55, 31e58. Mak, J., 2004. Important zoonotic intestinal protozoan parasites in Asia. Tropical Biomedicine 21, 39e50. Mao, L., Zheng, J., Wang, X., Yao, L., Cheng, H., Zhu, Q., 2009. Epidemiological investigation on an outbreak of amoebic dysentery in Jiangshan City of Zhejiang Province. Zhongguo Ji Sheng Chong Xue Yu Ji Sheng Chong Bing Za Zhi 27, 182e183. Marshall, M., Naumovitz, D., Ortega, Y., Sterling, C., 1997. Waterborne protozoan pathogens. Clinical Microbiology Reviews 10, 67e85. Mason, B.W., Chalmers, R.M., Carnicer-Pont, D., Casemore, D.P., 2010. A Cryptosporidium hominis outbreak in North-West Wales associated with low oocyst counts in treated drinking water. Journal of Water and Health 8, 299e310. MMWR, 2007. Acanthamoeba Keratitis e multiple states, 2005e2007. Morbidity and Mortality Weekly Report 56, 532e534. Morgan, U.M., O’Brien, P.A., Thompson, A.R.C., 1996. The development of diagnostic PCR primers for Cryptosporidium using RAPD-PCR. Molecular and Biochemical Parasitology 77, 103e108. Morris, R.D., Levin, R., 1995. Estimating the incidence of waterborne infectious disease related to drinking water in the United States. In: Reichard, G., Zapponi, E. (Eds.), Assessing and Managing Health Risks from Drinking Water Contamination: Approaches and Applications. International Association of Hydrological Sciences, Wallingford (UK). Mundaca, C., Torresslimming, P., Araujocastillo, R., Moran, M., Bacon, D., Ortega, Y., Gilman, R., Blazes, D., 2008. Use of PCR to improve diagnostic yield in an outbreak of cyclosporiasis in
6614
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 0 3 e6 6 1 4
Lima, Peru. Transactions Royal Society Tropical Medical Hygiene 102, 712e717. Nichols, G., Chalmers, R., Lake, I., Sopwith, W., Regan, M., Hunter, P., Grenfell, P., Harrison, F., Lane, C., 2006. Cryptosporidiosis: a report on the surveillance and epidemiology of Cryptosporidium infection in England and Wales. In: Drinking Water Directorate Contract Number DWI 70/2/201, pp. 1e142. Nyga˚rd, K., Schimmer, B., Søbstad, Ø, Walde, A., Tveit, I., Langeland, N., Hausken, T., Aavitsland, P., 2006. A large community outbreak of waterborne giardiasis- delayed detection in a non-endemic urban area. BMC Public Health 6, 141. O’Reilly, C.E., Bowen, A.B., Perez, N.E., Sarisky, J.P., Shepherd, C.A., Miller, M.D., Hubbard, B.C., Herring, M., Buchanan, S.D., Fitzgerald, C.C., Hill, V., Arrowood, M.J., Xiao, L.X., Hoekstra, R. M., Mintz, E.D., Lynch, M.F., 2007. A waterborne outbreak of gastroenteritis with multiple etiologies among resort island visitors and residents: Ohio, 2004. Clinical Infectious Diseases 44, 506e512. Ortega-Pierres, G., Smith, H., Caccio`, S., Thompson, R.C.A., 2009. New tools provide further insights into Giardia and Cryptosporidium biology. Trends in Parasitology 25, 410e416. Palanisamy, M., Madhavan, B., Balasundaram, M.B., Andavar, R., Venkatapathy, N., 2006. Outbreak of ocular toxoplasmosis in Coimbatore, India. Indian Journal of Ophthalmology 54, 129e131. Pelly, H., Cormican, M., O’Donovan, D., Chalmers, R., Hanahoe, B., Cloughley, R., 2007. A large outbreak of cryptosporidiosis in western Ireland linked to public water supply: a preliminary report. European Surveillance 12 (18), 3187. Persson, K., Svenungsson, B., de Jong, B., 2007. An outbreak of cryptosporidiosis at a day-care centre in Sweden. European Surveillance 12 (8), E070823.3. Rimhanen-Finne, R., Ha¨nninen, M., Vuento, R., Laine, J., Jokiranta, T., Snellman, M., Pitka¨nen, T., Miettinen, I., Kuusi, M., 2010. Contaminated water caused the first outbreak of giardiasis in Finland, 2007: a descriptive study. Scandinavian Journal of Infectious Diseases 42, 613e619. Robertson, L., Forberg, T., Hermansen, L., Gjerde, B., Alvsva˚g, J., Langeland, N., 2006a. Cryptosporidium parvum infections in Bergen, Norway, during an extensive outbreak of waterborne giardiasis in autumn and winter 2004. Applied and Environmental Microbiology 72, 2218e2220. Robertson, L., Hermansen, L., Gjerde, B., Strand, E., Alvsva, J., Langeland, N., 2006b. Application of genotyping during an extensive outbreak of waterborne giardiasis in Bergen, Norway, during autumn and winter 2004. Applied and Environmental Microbiology 72, 2212e2217. Rolfs, R., Beach, M., Hlavsa, M., 2008. Community wide cryptosporidiosis outbreak e Utah, 2007. Morbidity and Mortality Weekly Report 57, 989e993. Schaffzin, J.K., Keithly, J., Johnson, G., 2006. Large outbreak of cryptosporidiosis associated with a recreational water sprayparkeNew York, 2005. In: Proceedings of the 55th Annual Conference of the Epidemic Intelligence Service. US Department of Health and Human Services, CDC, Atlanta. Slifko, T., Smith, H., Rose, J., 2000. Emerging parasite zoonoses associated with water and food. International Journal for Parasitology 30, 1379e1393. Smith, S., Elliot, A., Mallaghan, C., Modha, D., Hippisley-Cox, J., Large, S., Regan, M., Smith, G., 2010. Value of syndromic surveillance in monitoring a focal waterborne outbreak due to an unusual Cryptosporidium genotype in Northamptonshire, United Kingdom, June e July 2008. European Surveillance 15 (33), 19643.
Stanwell-Smith, R., Andersson, Y., Levy, D.A., 2003. National surveillance systems. In: Hunter, P.R., Waite, M., Ronchi, E. (Eds.), Drinking Water and Infectious Disease: Establishing the Links. CRC Press, IWA publishing, London, New York, Washington DC. Strand, E.A., Robertson, L.J., Hanevik, K., Alvsvag, J.O., Morch, K., Langeland, N., 2008. Sensitivity of a Giardia antigen test in persistent giardiasis following an extensive outbreak. Clinical Microbiology and Infection 14, 1069e1071. Sveriges Radio 2010, www.sverigesradio.se. Takagi, M., Toriumi, H., Endo, T., Yamamoto, N., Kuroki, T., 2008. An outbreak of cryptosporidiosis associated with swimming pools. Kansenshogaku Zasshi 82, 14e19. Taniguchi, K., Yoshida, M., Sunagawa, T., Tada, Y., Okabe, N., 2008. Imported infectious diseases and surveillance in Japan. Travel Medicine and Infectious Disease 6, 349e354. Torres-Slimming, P., Mundaca, C., Moran, M., Quispe, J., Colina, O., Bacon, D., Lescano, A., Gilman, R., Blazes, D., 2006. Outbreak of cyclosporiasis at a naval base in Lima, Peru. The American Society of Tropical Medicine and Hygiene 75, 546e548. Tuncay, S., Delibas, S., Inceboz, T., Over, L., Oral, A.M., Akisu¨, C., Aksoy, U., 2008. An outbreak of gastroenteritis associated with intestinal parasites. Tu¨rkiye Parazitoloji Dergisi 32, 249e252. Turabelidze, G., Lin, M., Weiser, T., Zhu, B.P., 2007. Communitywide outbreak of cryptosporidiosis in rural Missouri associated with attendance at child care centers. Archives of Pediatrics & Adolescent Medicine 161, 878e883. Vaudaux, J.D., Muccioli, C., James, E.R., Silveira, C., Magargal, S.L., Jung, C., Dubey, J.P., Jones, J.L., Doymaz, M. Z., Bruckner, D.A., Belfort, R., Holland, G.R., Grigg, M.E., 2010. Identification of an atypical strain of Toxoplasma gondii as the cause of a waterborne outbreak of toxoplasmosis in Santa Isabel do Ivai. Brazilian Journal of Infectious Diseases 202, 1226e1233. Wahl, E., Bevanger, L., 2007. An outbreak of giardiasis in a child day-care centre in Trondheim. Tidsskr Nor Laegeforen 127, 146e184. Wheeler, C., Vugia, D.J., Thomas, G., Beach, M.J., Carnes, S., Maier, T., Gorman, J., Xiao, L., Arrowood, M.J., Gilliss, D., Werner, S.B., 2006. Outbreak of cryptosporidiosis at a California watermark: employee and patron roles and the long road towards prevention. Epidemiology and Infection 135, 302e310. Wright, J., Gundry, S.W., 2009. Household characteristics associated with home water treatment: an analysis of the Egyptian demographic and health survey. Journal of Water and Health 7 (1), 21e29. www.who.int. Yoder, J., Hlavsa, M., Craun, G., Hill, V., Roberts, V., Yu, P., Hicks, L., Alexander, N., Calderon, R., Roy, S., Beach, M., 2008. Surveillance for waterborne disease and outbreaks associated with recreational water use and other aquatic facilityassociated health events e- United States, 2005-2006. Morbidity and Mortality Weekly Report 57 (SS09), 1e29. Yokoi, H., Tsuruta, M., Tanaka, T., Tsutake, M., Akiba, Y., Kimura, T., Tokita, Y., Akimoto, T., Mitsui, Y., Ogasawara, Y., Ikegami, H., 2005. Cryptosporidium outbreak in a sports center. Japanese Journal of Infectious Diseases 58, 331e332. Yotthanooi, S., Choonpradub, C., 2010. A statistical method for estimating under-reported incidence rates with application to child diarrhea in Thai provinces bordering Cambodia. Southeast Asian Journal of Tropical Medicine and Public Health 41, 203e214.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 1 5 e6 6 2 4
Available online at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Characterization of the settling process for wastewater from a combined sewer system P. Piro a,*, M. Carbone a, N. Penna a, J. Marsalek b a b
Department of Soil Conservation, University of Calabria, Italy National Water Research Institute, Environment Canada, Burlington, Ontario, Canada L7R 4A6
article info
abstract
Article history:
Among the methods used for determining the parameters necessary for design of waste-
Received 20 January 2011
water settling tanks, settling column tests are used most commonly, because of their
Received in revised form
simplicity and low costs. These tests partly mimic the actual settling processes and allow
24 June 2011
the evaluation of total suspended solids (TSS) removal by settling. Wastewater samples
Accepted 14 September 2011
collected from the Liguori Channel (LC) catchment in Cosenza (Italy) were subject to
Available online 2 October 2011
settling column tests, which yielded iso-removal curves for both dry and wet-weather flow conditions. Such curves were approximated well by the newly proposed power law func-
Keywords:
tion containing two empirical parameters, a and b, the first of which is the particle settling
Urban drainage
velocity and the second one is a flocculation factor accounting for deviations from discrete
Water pollutants
particle settling. This power law function was tested for both the LC catchment and
CSO (Combined Sewer Overflow)
literature data and yielded a very good fit, with correlation coefficient values (R2) ranging
CSO settleability
from 0.93 to 0.99. Finally, variations in the settling tank TSS removal efficiencies with
Settling columns
parameters a and b were also analyzed and provided insight for settling tank design. ª 2011 Elsevier Ltd. All rights reserved.
Wastewater treatment
1.
Introduction
There is a large body of evidence that wet-weather pollution, comprising uncontrolled or inadequately treated combined sewer overflows (CSOs) and discharges of urban stormwater, represents one of the major causes of long-term persistence of poor water quality in receiving waters (Sua`rez and Puertas, 2004). This impact is caused by a variety of pollutants present in urban wastewater and in the washoff of urban surfaces by rain, and in general, many such pollutants, including heavy metals, are associated with solid particles (Pettersson, 2002; Vaze and Chiew, 2004; Characklis et al., 2005; Vallet et al., 2010). Pollutant loads in washoff depend on such factors as catchment land use, population density, and traffic intensity (Butler and Davies, 2000), and their mitigation by common management measures is rather challenging.
Solid particles serving as carriers of pollutants are typically described as total suspended solids (TSS) and their concentrations are evaluated by standard laboratory tests (APHA et al., 2005). Among the TSS characteristics, the particle terminal settling velocity is the most important parameter for settling design and can be determined experimentally using a variety of experimental procedures and devices, which can be classified into two categories: (a) quiescent settling devices (i.e., various types of settling columns), with liquid at rest, and (b) dynamic settling devices, in which liquid may be flowing or is subject to mechanically generated turbulence (Marsalek et al., 2006). The removal of TSS is an essential step for reducing the pollution of receiving waters (Peavy et al., 1985) and can be accomplished by applying different treatment processes, including coagulation/flocculation, settling and filtration (Oke et al., 2006). Currently, settling tanks are one of
* Corresponding author. E-mail address:
[email protected] (P. Piro). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.09.034
6616
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 1 5 e6 6 2 4
the best practices for management of wet weather pollution, because they provide both volumetric control of polluted flows and their treatment by settling. The efficiency of settling tanks depends, to a great extent, on flow behaviour in the tank (Maus and Uhl, 2010) and on settling characteristics of TSS in the flow to be treated (Piro et al., 2011). Concerning the latter, the settleability of wetweather flows can be described by the settling velocity of TSS particles (Chebbo et al., 1998) and this parameter is used as an input in the majority of models simulating TSS settling in tanks. Thus, the hydraulic design of CSO settling tanks relies on empirical approaches providing input to computer modelling, as demonstrated, e.g., by He et al. (2006) and others, who conducted laboratory studies in small-scale physical models to advance the understanding of settling processes and their modelling by computational fluid dynamics (CFD) models. Several (CFD) models and a number of laboratory procedures (Klepiszewski et al., 2010) were used to evaluate the TSS settling velocity. CFD models simulate complex hydraulic conditions in settling tanks and help explain when and where sediments will settle. However, because of their complexity and computing time requirements, these models are not well suited for engineering practice (Vallet et al., 2010). At the same time, the essential design information can be obtained by laboratory experiments, such as settling column tests, which are relatively inexpensive. Assessments of settling column tests or of the relationship between experimental settling results and tank design criteria were reported by a number of researchers (Weber, 1972; Zanoni and Blomquist, 1975; Berthouex and Stevens, 1982; Eckenfelder, 1989; Oke et al., 2006; Piro et al., 2011). The main objective of this paper is to examine settling column data and propose a simple computational design methodology linking the treatment efficiency of CSO settling tanks to settling column test data. The proposed method can be applied quickly and produces results, which are equivalent to those obtained with the well-known conventional graphical methods for assessing the wastewater settleability.
2.
Methodology
2.1.
Study area
For experimental investigations, CSO samples were collected in the Liguori Channel (LC) catchment in Cosenza (Italy), which has an area of 414 ha and a population of 50,000 inhabitants. Forty-eight percent of the catchment area is densely urbanized with high imperviousness, but the rest (52%) is occupied by a pervious area, covered mainly by vegetation (Piro et al., 2011). A map of the catchment is shown in Fig. 1. Further information on catchment characteristics can be found elsewhere (Piro and Sole, 2001). The catchment is drained by a combined sewer system that conveys dry weather flows to the wastewater treatment plant (WWTP) at Montalto Uffugo. During significant rainfall events, the sewer flow exceeds the hydraulic capacity of the sewer system and the WWTP, and the excess flow is discharged directly into the
Fig. 1 e Experimental catchment of Liguori Channel (LC).
Crati River through overflow drop structures, without any treatment (Piro et al., 2011). The LC catchment was instrumented in the 1990s by installing a monitoring station, which was equipped with a rain gauge and an ultrasonic sensor measuring water depths in the outfall sewer. Such depths were then used to calculate the corresponding discharges. The station also contains a datalogger recording rain and water level data every minute. Furthermore, since 2004, sewer flow samples have been collected to characterize flow quality during both dry and wet weather conditions at this site (Piro, 2007).
2.2.
Methods of analysis
Settling is a natural method of removing suspended particles from wastewater. Inorganic solids are heavier than water and, therefore, gravity settling is the most common solid/liquid separation technique (Peavy et al., 1985). To determine the settling characteristics of a suspension and to measure the settling velocities of discrete particles in diluted suspensions, an indirect method was devised by Camp (1946), who had introduced for the first time the concept of the settling column test procedure, which was later described in detail in common handbooks of environmental engineering (e.g., Metcalf and Eddy, 2003). Generally, various types of settling columns are used, with the first and most common type being a stationary settling column with side withdrawal ports along the column length. The other types of columns which are also frequently used are variations of the original design and differ only in various methods that are used to pre-mix the sample before the column test begins (Pisano, 1996). In order to determine settling characteristics of solids in dry- and wet-weather sewer flows in the catchment studied, settling column tests were employed. The sampling campaign, on which this research is based, was carried out for two years, starting in Oct. 2007, and yielded data for 15 events. These events can be divided into two categories (Table 1) corresponding to dry-weather conditions (DW) and wet-weather conditions (WW).
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 1 5 e6 6 2 4
Table 1 e Division of the tests into dry and wet weather conditions. DW 24 October 2007 23 April 2008 14 May 2008 11 June 2008 9 July 2008 16 July 2008
WW 2 April 2008 13 January 2009 18 February 2009 11 March 2009 20 March 2009 21 April 2009 28 April 2009 22 September 2009 23 October 2009
Since a minimum column diameter of 127 mm is recommended to minimize wall effects (Eckenfelder, 1989), the column used in this study was a stationary settling column of 150 mm in diameter and 3 m in length, with 5 sampling ports equally spaced (Fig. 2). Theoretically, the depth of the column does not influence the analysis (Peavy et al., 1985), but the column depth was selected here as 3 m, which corresponds to a typical depth adopted in design of settling tanks (Piro et al., 2011). During the settling tests, samples are withdrawn from the settling column at certain time intervals and at several depths. In this study, 100-ml samples were withdrawn from each sampling port every 5 min for the total test duration of 40 min. This duration was established from the earlier
6617
findings developed for the settling tank design in the LC catchment, as a part of the research studies conducted by the Department of Soil Conservation (University of Calabria) in addressing the problem of CSO treatment. In such studies, a minimum settling tank volume of 10,000 m3 was established in 2005. After the average flow through the unit had been calculated for a critical event selected from the database of locally observed rain events (Piro, 2007), the detention time was fixed at 40 min (Piro et al., 2011). After finishing the settling tests, the collected samples were analyzed to determine TSS concentrations according to the APAT method (APAT and IRSA-CNR, 2003), which follows the protocol 2540D developed by APHA et al. (2005). These TSS concentrations are then used to compute the mass fraction removed at each depth and for each time, using the following expression: Eij ¼
Cij 100 1 C0
(1)
where Cij is the mass fraction, in percent, that is removed at the ith depth at the jth time interval, and C0 is the initial TSS concentration. The percent removals obtained from the test data are plotted at the appropriate depths and times, and the removal isolines are constructed by interpolating the plotted values (Fig. 3) (Zanoni and Blomquist, 1975). The curves thus drawn represent the limiting or maximum settling path for the indicated removal efficiency percentage (Eckenfelder, 1989). Further information on this traditional-graphical method can be found elsewhere (Piro et al., 2011). As shown in Fig. 3, when the concentrations of TSS remaining in suspension (expressed in percent of the initial concentration) are plotted versus time for several sampling depths, the plotted curves can be approximated by a power law function described by the following equation: h ¼ atb
(2)
where h is the depth, t is the residence time (also called detention or settling time), and a and b are two empirical parameters that will be determined for each curve of each set of column test data. In particular, parameter a represents the particle settling velocity, which can be demonstrated by differentiating Eq. (2) with respect to time (t): dh ¼ abtb1 dt
(3)
and assuming b ¼ 1 (which corresponds to discrete settling), Eq. (3) becomes: dh ¼a dt
Fig. 2 e Laboratory settling column (dimensions in metres).
(4)
which is the settling velocity of a discrete particle. Parameter b represents the flocculation factor; when b equals 1 the iso-percentage removal curves become straight lines, which is an indicator of discrete settling, while if b is greater than 1, flocculent settling occurs, yielding a curvilinear settling path. This new mathematical definition of the iso-removal curves using a simple function allows easy calculation of the removal efficiency of a settling unit. Such a method was
6618
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 1 5 e6 6 2 4
For the jth iso-removal curve of an individual event, the root mean square error (RMSE) for the detention times of any actual sample describing deviations from those of the average sample, tRj, can be calculated using the following equation:
tRj ¼
Fig. 3 e Iso-removal curves obtained from settling analysis.
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2 uP u N t i¼1 ti;j ti;j N
(5)
where j indicates the iso-removal curve, and N is the total number of the ith depths for which the average of the settling times for the jth iso-removal curve, ti;j , is calculated. Furthermore, for each event sample, a measure of the sample mean error was calculated as: PM Err ¼
applied to the experimental data from the LC catchment as well as to the literature data in order to test its more general applicability. Furthermore, for sites where a number of events were sampled and the corresponding settling data sets were determined, a characteristic (typical) event sample can be determined. Theoretically, using an average sample (i.e., a sample possessing average TSS removals determined for the full set of analyzed events) might seem more appropriate, but such a sample is hypothetical, and thus it is more realistic to use the characteristic (actual) sample instead. It is defined as the actual event sample of which characteristics best approximate the average of all the samples. In some way this approach is analogous to using actual (rather than synthesized) design storms in urban drainage (WEF and ASCE, 1992). The method adopted to determine the characteristic sample is based on the method used in hydrology to produce the flow duration curve of a water course (Da Deppo et al., 2004).
j¼1 tRj
M
(6)
where M is the number of iso-removal curves, and the sample with a minimum average RMSE ðErrÞ was adopted as the characteristic sample. The characteristic sample identified this way can be used in the design of settling facilities.
3.
Results and discussion
The results of the 15 settling column tests conducted for the LC catchment produced useful information for characterizing mathematically the settling of the LC wastewater. Figs. 4 and 5 show the TSS iso-removal efficiency curves obtained for DW and WW sample tests. When comparing the settling data in Figs. 4 and 5, for the same detention time of 40 min, it is possible to note greater efficiencies of settling in the dry weather than in wet-weather. This result can be explained by two factors promoting more intense flocculation in DW (wastewater) samples: a higher
Fig. 4 e Curves of constant percent TSS removal obtained for DW sample tests.
6619
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 1 5 e6 6 2 4
presence of fine-grained particles and greater organic contents, when compared to the properties of WW (CSO) samples (Piro et al., 2011). More efficient settling of DW (wastewater) samples was confirmed earlier by the analysis of PSD (particle size distribution) curves for DW and WW samples from the study site, indicating greater concentrations of finegrained particles in DW samples (Piro et al., 2010). The same study also confirmed greater organic contents in DW (wastewater) samples, because in WW (CSO) samples, wastewater is diluted by low organic content stormwater (Piro et al., 2011). The results confirm that the iso-removal curves can be approximated mathematically by the power function (Eq. (2)). The correlation coefficients, R2, were calculated for each curve of all data sets (see Table 2) and were generally higher than 0.95. Thus, the choice of a power function is justified and provides an excellent approximation of the experimental data. The residual values of the detention (settling) times were also calculated as the difference between the observed and fitted values for each data set. The average residual values of the detention time for individual iso-removal curves are given in Table 3 and indicate that the uncertainties caused by using the mathematical power function model rather than the traditional graphical procedure are relatively small, with the maximum residual being about 2.9 min.
Table 2 e Correlation coefficients (R2) for curves fitted to all settling column test data. Test
24 Oct 2007 23 Apr 2008 14 May 2008 11 Jun 2008 09 Jul 2008 16 Jul 2008 02 Apr 2008 13 Jan 2009 18 Feb 2009 11 Mar 2009 20 Mar 2009 21 Apr 2009 28 Apr 2009 22 Sep 2009 23 Oct 2009
Weather conditions DW DW DW DW DW DW WW WW WW WW WW WW WW WW WW
R2 10% 20% 30% 40% 50% 60% 0.99 0.97 0.99 0.99 e 0.99 0.99 e 0.98 0.98 0.96 0.99 e 0.99 e
0.99 0.96 0.99 0.99 0.99 0.99 0.97 0.99 0.99 0.99 0.90 0.99 0.99 0.98 0.99
0.99 0.99 0.99 0.99 0.99 0.96 0.95 0.97 0.98 0.99 0.91 0.99 0.99 0.99 0.99
0.99 0.99 0.99 0.99 0.99 0.94 0.96 0.99 0.93 0.99 0.99 0.99 0.98 0.99 0.99
0.99 0.99 0.99 0.99 0.99 0.94 e e e e e e e e e
0.99 e 0.99 0.99 0.99 0.99 e e e e e e e e e
Figs. 6 and 7 show variations in the TSS removal efficiency, E, with parameters a and b (from Eq. (2)), for both DW and WW conditions. As shown, E increased with the increasing values of the flocculation factor, but decreased with increasing
Fig. 5 e Curves of constant percent TSS removal obtained for WW sample tests.
6620
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 1 5 e6 6 2 4
Table 3 e Numerical values of the fitting residual (min) for each curve of each settling column test. Test
24 23 14 11 09 16 02 13 18 11 20 21 28 22 23
Weather conditions
Oct 2007 Apr 2008 May 2008 Jun 2008 Jul 2008 Jul 2008 Apr 2008 Jan 2009 Feb 2009 Mar 2009 Mar 2009 Apr 2009 Apr 2009 Sep 2009 Oct 2009
DW DW DW DW DW DW WW WW WW WW WW WW WW WW WW
Residual 10% 20% 30% 40% 50% 60% 0.00 1.27 0.00 0.00 e 0.11 0.01 e 0.95 1.04 0.11 0.49 0.00 0.00 e
0.15 1.66 0.66 0.00 0.06 0.00 0.77 0.42 0.22 0.29 1.34 0.00 0.81 0.85 0.47
0.26 1.40 0.91 0.00 0.10 0.80 1.99 1.06 0.77 0.58 1.87 0.00 1.93 0.13 1.32
0.23 0.00 0.12 1.19 0.00 1.85 0.72 0.48 1.24 0.47 0.61 0.00 0.46 0.06 0.12
1.83 0.00 0.00 2.91 0.79 2.77 e e e e e e e e e
0.00 e 0.00 0.08 0.00 1.84 e e e e e e e e e
(design) particle settling velocity. The latter tendency can be explained by the fact that in the settling tank design, the tank overflow rate V (V ¼ Q/A, where Q is the flow rate at which clarified water is produced and A is the surface area of the tank) is set equal to the settling velocity of the smallest particle to be captured, Vc (Vc ¼ Q/A), and therefore the larger the settling velocity, the smaller the surface area (and the tank) and the overall settling efficiency. Thus, higher removal efficiencies correspond to designs targeting capture of smaller particles settling at lower velocities. At the same time, flocculation of fine particles may increase, to some extent, their size and settling velocity. The results indicate that in the initial phase of the settling process the particles behave as discrete particles and later begin to show flocculation tendencies. Once flocs have reached the maximum sustainable size, beyond which they start to break up (Krishnappan and Marsalek, 2002), they are sufficiently distant from each other and start settling again as discrete particles (Piro et al., 2011). In order to characterize the settleability of wastewater (corresponding to the DW conditions) and CSOs
(corresponding to the WW conditions) at the LC experimental site, the characteristic samples were determined for both DW and WW conditions. Table 4 shows the RMSE of individual event detention times compared to those corresponding to the average sample, tRj, as determined by Eq. (5). Thus, according to the condition of a minimum average RMSE (Eq. (6)), it is possible to deduce that the sample of February 18, 2009 is the characteristic sample for WW conditions. For DW conditions, two events have rather similar minimum average RMSE, on 24 October 2007 and 16 July 2008; the event on 24 October 2007 was selected as yielding the characteristic sample, because its standard deviation of RMSE was smaller than that of the July, 16 2008 event. The applicability of the newly proposed power law function (Eq. (2)) used to approximate the iso-removal curves was also tested on the data found in five references (Weber, 1972; Zanoni and Blomquist, 1975; Berthouex and Stevens, 1982; Eckenfelder, 1989; Oke et al., 2006). Fig. 8 illustrates the fitted curves for each data set. The values of the correlation coefficients for each data set, including that from the LC site, are given in Table 5. For the 35 TSS removal curves listed, the goodness of fit was described by a mean value of correlation coefficient (R2) of 0.983 (0.015), and consequently, the proposed power law function (Eq. (2)) approximates well the settling column data not only for the LC catchment, but generally for wastewater samples from different source catchments characterized by various land use, population density and traffic intensity. Moreover, the trends in parameters a and b of the literature data are similar to those obtained for the LC catchment; the particle settling velocity decreases with the increasing removal efficiency, but the flocculation factor increases (Fig. 9). The design of treatment of CSOs requires the knowledge of settling characteristics; when CSOs are retained for clarification, larger sediments and settleable particles are mainly influenced by gravitational forces, while the finer suspended particles are also subject to coagulation and flocculation phenomena (Metcalf and Eddy, 2003). It was reported that CSOs generally contain flocculating particles and therefore the Stoke’s equation cannot be used to design clarifiers
Fig. 6 e Particle settling velocity (a) and flocculation factor (b) vs. TSS removal efficiency, for DW conditions.
6621
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 1 5 e6 6 2 4
Fig. 7 e Particle settling velocity (a) and flocculation factor (b) vs. TSS removal efficiency, for WW conditions.
(Marsalek et al., 2006). Thus, the criteria adopted for design of clarifiers have evolved in both practice and theory in order to account for many factors contributing to the flocculation process. Most frequently, the settling column tests (Peavy et al., 1985) were used to estimate the removal efficiency of TSS and the detention (settling) time of the flocculating wastewater, but in the traditional graphically-based application of settling results, this approach may be time consuming and, therefore, expensive for practical design purposes. The iso-removal curves allow determining the particle fractions that are completely removed from the water column, i.e. the particles with diameters d d0, for a given detention time t ¼ QH. Nevertheless the total removal efficiency will be greater because also some finer particles (with dimensions d < d0 and settling velocity v < v0) are partly removed. The results of the column tests can be used to assess the total removal efficiency (Etot) of sedimentation process (Metcalf and Eddy, 2003), as:
Etot ¼ EðqH Þ þ
X hi;iþ1 ðEi Eiþ1 Þ H i
(7)
where H is the column depth, E(QH) is the constant percent removal curve passing through point (QH, H ), Ei and Eiþ1 are the iso-removal efficiencies greater than E(QH) and hi,iþ1 is the depth of the middle point of the segment joining Ei and Eiþ1 curves at t ¼ QH. It is evident that this approach is time consuming for practical design purposes, especially when taking into account the high variability of the settling process during individual events. The knowledge of the analytical relationships for each iso-removal curve (i.e. hi ¼ ai tbi ), allows expressing Eq. (7) as: X
hi þ hiþ1 ðEi Eiþ1 Þ 2H 1 X b b ¼ EðQH Þ þ ðEi Eiþ1 Þ ai QHi þ aiþ1 QHiþ1 2H i
Etot ¼ EðQH Þ þ
i
(8)
Table 4 e (a) RMSEs (in minutes) for individual event samples and iso-removal curves in relation to the average sample, tRj, (b) average RMSE ðErrÞ, and (c) standard deviations (the characteristic tests are bolded). Test
Weather conditions
tRj 10%
20%
30%
40%
50%
60%
Err
Standard deviation
24 Oct 2007 23 Apr 2008 14 May 2008 11 Jun 2008 09 Jul 2008 16 Jul 2008
DW DW DW DW DW DW
2.8 6.9 0.8 2.4 e 1.7
1.8 7.9 3.6 4.8 6.6 2.2
2.9 9.1 7.7 7.4 9.0 2.7
3.1 5.7 8.8 5.4 6.9 1.1
1.4 4.8 8.0 7.5 3.5 0.9
3.1 e 5.5 5.6 4.6 6.0
2.5 6.9 5.7 5.5 6.1 2.4
0.75 1.69 3.08 1.85 2.12 1.87
02 Apr 2008 13 Jan 2009 18 Feb 2009 11 Mar 2009 20 Mar 2009 21 Apr 2009 28 Apr 2009 22 Sep 2009 23 Oct 2009
WW WW WW WW WW WW WW WW WW
0.8 e 1.0 2.5 5.1 5.2 e 3.2 e
3.6 2.7 1.4 6.4 2.5 4.4 3.2 2.6 3.4
6.7 2.6 2.1 7.5 2.7 3.6 2.1 3.1 7.1
9.4 4.2 2.3 4.7 2.4 5.3 1.2 5.9 3.6
e e e e e e e e e
e e e e e e e e e
5.1 3.2 1.7 5.3 3.2 4.6 2.2 3.7 4.7
3.72 0.92 0.59 2.15 1.29 0.78 1.03 1.50 2.05
6622
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 1 5 e6 6 2 4
Fig. 8 e Curves of constant percent removal obtained for published data (Weber, 1972; Zanoni and Blomquist, 1975; Berthouex and Stevens, 1982; Eckenfelder, 1989; Oke et al., 2006).
where the parameters ai and bi reflect the variability of the settling behaviour due to effluent characteristics. In practical terms, the approximation of the TSS isoremoval curves by a mathematical function serves two purposes: (a) It provides a unifying model for settling of various wastewaters (as shown by the applicability of this model to various types of wastewaters in Table 5), and (b) It facilitates a quick calculation of the total TSS removal using Eq. (8). Furthermore, the newly proposed model is physically based and can be used in characterizing settleability of the wastewater to be treated, which depends not only on the design particle size, density and detention time, but also on the characteristic ability of the wastewater to flocculate. The
feasibility of describing wastewater settling parameters as well-defined physical quantities (velocity of sedimentation and flocculation capacity) for specific events also opens the possibility of stochastic analysis for the events observed at a specific site. Ultimately, this may lead to a better definition of the wastewater characteristic parameters in a probabilistic manner, thus providing more information about the characteristics of the wastewater to be treated and comprehensive information for design and the proper sizing of treatment units. Towards this end, further experiments are in progress at the experimental site of the University of Calabria, with the overall goal of further evaluating the characteristic settleability values at this site.
6623
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 1 5 e6 6 2 4
Table 5 e Correlation coefficients, R2, describing the goodness of fit of the power law function (Eq. (2)) for literature and LC data. R2
Data source
Weber (1972) Zanoni and Blomquist (1975) Berthouex and Stevens (1982) Eckenfelder (1989) Oke et al. (2006) Piro et al. (2010) e DW Piro et al. (2010) e WW
10%
20%
30%
40%
50%
60%
70%
80%
90%
e e
0.99 0.99
0.99 0.98
0.99 0.99
0.99 0.95
0.99 0.99
0.99 e
e e
e e
e
e
e
e
0.99
0.96
0.99
0.99
0.99
e e 0.99 0.98
e 0.94 0.99 0.99
0.99 0.99 0.99 0.98
0.99 0.99 0.99 0.93
0.99 0.99 0.99 e
0.98 0.98 0.99 e
e e e e
e e e e
e e e e
Fig. 9 e Particle settling velocity (a) and flocculation factor (b) vs. TSS removal efficiency, for literature data.
4.
Conclusions
An experimental study was conducted in the context of a broader research project, which addresses the characteristics of WW and DW flows in combined sewers, with respect to the transport of pollutants and solids of specific particle size distributions. The aim of the study was to obtain empirical TSS settleability data and develop a simple methodology for design of sedimentation tanks. Towards this end, a novel method of analyzing settling column test data was presented in this paper. Using this method, it was possible to determine a mathematical function (power law) describing the isoremoval curves of TSS, for varying depths and detention times. This function is physically based, and both its fitted parameters, a and b, have physical meaning: a represents the particle settling velocity and b is a flocculation factor. The applicability of this equation was also tested on the literature data. For 35 TSS removal curves found in the literature, the goodness of fit was described by a mean value of correlation coefficient (R2) of 0.983 (0.015). Thus, the proposed function appears to be valid not only for the LC catchment data, but generally for wastewater samples from different catchments with different characteristics. The fitted equation can be used to describe TSS iso-removal curves and to estimate the removal efficiency of an ideal settling tank. Furthermore,
where a number of events were sampled and the corresponding settling data sets are available, a single representative design (characteristic) event can be identified as an actual event with least deviation from the average removal curve for the whole event set. Future research will focus on probabilistic description of the parameters a and b, and their use in analysis of settleability of CSOs generated by long rainfall series.
references
APAT, IRSA-CNR, 2003. Metodi analitici per le acque. Manuali e linee guida 29 (in Italian: standard methods for the examination of water). APHA, AWWA, WEF, 2005. In: Standard Methods for The Examination of Water and Wastewater, 21st ed. American Public Health Association, Washington, DC. Berthouex, P.M., Stevens, D.K., 1982. Computer analysis of settling test data. J. Environ. Eng. 108 (5), 1065e1069. Butler, D., Davies, J.W., 2000. In: Urban Drainage, first ed. E&FN SPON, London, pp. 489. Camp, T.R., 1946. Sedimentation and the design of settling tanks. Trans. ASCE 111, 895e958. Characklis, G.W., Dilts, M.J., Simmons III, O.D., Likirdopulos, C.A., Krometis, L.-A.H., Sobsey, M.D., 2005. Microbial partitioning to settleable particles in stormwater. Water Res. 39, 1773e1782.
6624
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 1 5 e6 6 2 4
Chebbo, G., Lucas-Aiguier, E., Bertrand-Krajewski, J.L., Gagne`, B., Hedges, P., 1998. Analysis of the methods for determining the settling characteristics of sewage and stormwater solids. Water Sci. Technol. 37 (1), 53e60. Da Deppo, L., Datei, C., Salandin, P., 2004. In: Sistemazione dei corsi d’acqua, fifth ed. Cortina, Padova, Italy, p. 815. Eckenfelder, W.W., 1989. In: Industrial Water Pollution Control, second ed. McGraw-Hill, New York, NY, pp. 400. He, C., Marsalek, J., Rochfort, Q., Krishnappan, B.G., 2006. Case study: refinement of hydraulic operation of a complex CSO storage/treatment facility by numerical and physical modeling. J. Hydraul. Eng., ASCE 312 (2), 131e139. Klepiszewski, K., Teufel, M., Seiffert, S., Henry, E., 2010. Measurement of flow velocity profiles in tank structures using the prototype device OCM Pro LR. In: Electronic Proceedings of the Seventh International Conference on Sustainable Techniques and Strategies in Urban Water Management, Lyon, France, June 27eJuly 1, Distributed by GRAIE, Lyon, France, 9 pp. Krishnappan, B.G., Marsalek, J., 2002. Modelling of flocculation and transport of cohesive sediment from an on-stream stormwater detention pond. Water Res. 36, 3849e3859. Marsalek, J., Krishnappan, B.G., Exall, K., Rochfort, Q., Stephens, R.P., 2006. An elutriation apparatus for assessing settleability of combined sewer overflows (CSOs). Water Sci. Technol. 54 (6, 7), 223e230. Maus, C., Uhl, M., 2010. Tracer studies for the modelling of sedimentation tanks. In: Electronic Proceedings of the Seventh International Conference on Sustainable Techniques and Strategies in Urban Water Management, Lyon, France, June 27eJuly, Distributed by GRAIE, Lyon, France, 10 pp. Metcalf and Eddy, 2003. In: Wastewater Engineering. Treatment and Reuse, fourth ed. McGraw-Hill, New York, NY, pp. 1819. Oke, I.A., Oladepo, K.T., Olajumoke, A.M., Ajayi, E.O., 2006. Settlement properties of solids in a domestic-institutional wastewater. J. Appl. Sci. Res. 2 (7), 385e390. Peavy, H.S., Rowe, D.R., Tchobanoglous, G., 1985. Environmental Engineering, International Edition. McGraw-Hill, New York, NY, pp. 699. Pettersson, T.J.R., Sep. 8e13, 2002. Characteristics of suspended particles in a small stormwater pond. In: Proceedings of the Ninth International Conference on Urban Drainage, Portland, OR. American Society of Civil Engineers, Reston, VA, Portland, OR, United States, pp. 1e12. Piro, P., Sole, A., 2001. Analisi di reti di drenaggio urbano mediante l’uso dei Gis. Applicazione al bacino urbano del
Liguori. (Urban drainage network analysis through the use of GIS. Application to the urban catchment Liguori). In: L’uso dei GIS per l’analisi e il controllo di problematiche territoriali (in Italian; The Use of GIS for Analysis and Control of Land Issues). Bios Publishers, Cosenza, Italy, pp. 143. Piro, P., 2007. Il bacino sperimentale urbano del Canale Liguori nella citta` di Cosenza. Osservazioni sperimentali qualiquantitative nel periodo 1995e2003. (An urban experimental catchment Canale Liguiori near the City of Cosenza. Experimental observations of water quantity and quality during the period of 1995e2003). Bios Publishers, Cosenza, Italy, pp. 71. Piro, P., Carbone, M., Garofalo, G., Sansalone, J., 2010. Size distribution of wet weather and dry weather particulate matter entrained in combined flows from an urbanizing sewer shed. Water Air Soil Pollut. 1e4 (206), 83e94. Piro, P., Carbone, M., Tomei, G., 2011. Assessing settleability of dry and wet weather flows in an urban area serviced by combined sewers. Water Air Soil Pollut. 214 (1), 107e117. Pisano, W.C., 1996. Summary: United States “sewer solids” settling characterization methods, results, uses and perspective. Water Sci. Technol. 33 (9), 109e115. Sua`rez, J., Puertas, J., 2004. Determination of COD, BOD, and suspended solids loads during combined sewer overflow (CSO) events in some combined catchments in Spain. Ecol. Eng. 24, 201e219. Vallet, B., Muschalla, D., Lessard, P., Vanrolleghem, P.A., 2010. A new dynamic stormwater basin model as a tool for management of urban runoff. In: Electronic Proceedings of the Seventh International Conference on Sustainable Techniques and Strategies in Urban Water Management, Lyon, France, June 27eJuly 1, Distributed by GRAIE, Lyon, France, 10 pp. Vaze, J., Chiew, F.H.S., 2004. Nutrient loads associated with different sediment sizes in urban stormwater and surface pollutants. J. Environ. Eng. 130, 391e396. Water Environment Federation (WEF) and American Society of Civil Engineers (ASCE), 1992. Design and Construction of Urban Stormwater Management Systems. ASCE, New York, NY. Weber, W.J., 1972. In: Physicochemical Processes for Water Quality Control, first ed. Wiley-Interscience, New York, NY, pp. 640. Zanoni, A.E., Blomquist, M.W., 1975. Column settling tests for flocculent suspensions. J. Environ. Eng. 101 (3), 309e318.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 2 5 e6 6 3 6
Available online at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Electron donor availability for microbial reductive processes following thermal treatment Kelly E. Fletcher a, Jed Costanza a,1, Kurt D. Pennell b, Frank E. Lo¨ffler c,d,e,* a
School of Civil and Environmental Engineering, Georgia Institute of Technology, 311 Ferst Drive, Atlanta, GA 30332-0512, USA Department of Civil and Environmental Engineering, Tufts University, Medford, MA 02155, USA c Department of Microbiology, University of Tennessee, M409 Walters Life Sciences, Knoxville, TN 37996-0845, USA d Department of Civil and Environmental Engineering, University of Tennessee, Knoxville, TN 37996, USA e Biosciences Division, Oak Ridge National Laboratory, Oak Ridge, TN 37831, USA b
article info
abstract
Article history:
Thermal treatment is capable of removing significant free-phase chlorinated solvent mass
Received 26 February 2011
while potentially enhancing bioremediation effectiveness by establishing temperature
Received in revised form
gradients in the perimeter of the source zone and by increasing electron donor availability.
16 June 2011
The objectives of this study were to determine the potential for enhanced reductive
Accepted 14 September 2011
dechlorination activity at the intermediate temperatures that establish in the perimeter of
Available online 2 October 2011
the heated source zone, and to evaluate the effect of electron donor competition on the performance of the microbial reductive dechlorination process. Microcosms, constructed
Keywords:
with tetrachloroethene- (PCE-) and trichloroethene- (TCE-) impacted soils from the Great
Thermal treatment
Lakes, IL, and Ft. Lewis, WA, sites were incubated at temperatures of 24, 35, 50, 70, and
Microbial reductive dechlorination
95 C for 4 months. Reductive dechlorination did not occur in microcosms incubated at
Biostimulation
temperatures above 24 C even though mesophilic PCE-to-cis-1,2-dichloroethene dechlor-
Bioaugmentation
inators were present in Ft. Lewis soil suggesting electron donor limitations. Five days after
Combined remedy approaches
cooling the microcosms to 24 C and bioaugmentation with the methanogenic, PCE-toethene-dechlorinating consortium OW, at least 85% of the initial PCE and TCE were dechlorinated, but dechlorination ceased prior to complete conversion to ethene. Subsequent biostimulation with hydrogen gas mitigated the dechlorination stall, and conversion to ethene resumed. The results of this study demonstrated that temperatures >35 C inhibit reductive dechlorination activity at the Great Lakes and Ft. Lewis sites, and that the majority of reducing equivalents released from the soil matrix during heat treatment are consumed in methanogenesis rather than reductive dechlorination. These observations suggest that bioaugmenting thermal treatment sites with cultures that do not contain methanogens may allow practitioners to realize enhanced dechlorination activity, a potential benefit of coupling thermal treatment with bioremediation. ª 2011 Elsevier Ltd. All rights reserved.
* Corresponding author. Department of Microbiology, University of Tennessee, M409 Walters Life Sciences, Knoxville, TN 37996-0845, USA. Tel.: þ1 865 974 4933; fax: þ1 865 974 4007. E-mail address:
[email protected] (F.E. Lo¨ffler). 1 Present address: USEPA/OCSPP/OPP, Antimicrobials Division, 2777 S. Crystal Dr., Arlington, VA 22202, USA. 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.09.033
6626
1.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 2 5 e6 6 3 6
Introduction
Tetrachloroethene (PCE) and trichloroethene (TCE) often exist in the subsurface as dense non-aqueous phase liquids (DNAPLs), which can serve as long-term sources of groundwater contamination (Ramsburg et al., 2004). In situ thermal treatment coupled with soil vapor extraction can rapidly remove significant amounts of chlorinated ethene mass (Heron et al., 1998, 2009; McGuire et al., 2006), but like other source zone remedies, thermal treatment is unlikely to remove 100% of the DNAPL mass (Davis, 1998; Sale and McWhorter, 2001; Christ et al., 2005). Anaerobic bioremediation has been employed successfully for the cleanup of chlorinated ethene plumes (Lo¨ffler and Edwards, 2006), and recent studies suggest that bioremediation may be an effective posttreatment polishing step following thermal treatment (Christ et al., 2005; Friis et al., 2006b, 2007). Anaerobic bioremediation of PCE and TCE occurs via stepwise reductive dechlorination reactions, i.e., PCE is transformed to TCE, TCE is transformed to dichloroethenes (mainly cis-1,2dichloroethene [cis-DCE]), DCEs are transformed to vinyl chloride (VC), and VC is transformed to non-toxic ethene. While numerous bacteria dechlorinate PCE and TCE to cisDCE, strictly hydrogenotrophic Dehalococcoides (Dhc) strains are the only known organisms capable of gaining energy from cisDCE to ethene reductive dechlorination (He et al., 2003). Dhc have stringent redox, temperature, and nutritional requirements (He et al., 2003; Friis et al., 2007; Amos et al., 2008). While previous studies have suggested that electrical resistance heating (ERH) may have little impact on redox conditions (Friis et al., 2005, 2006b), thermal treatment obviously affects subsurface temperature as source zones are routinely heated to temperatures of 100 C or greater (Heron et al., 1998). Optimal temperatures for complete microbial reductive dechlorination to ethene are 25e30 C, higher than those of most aquifers, but much lower than source zone temperatures during heat treatment (Friis et al., 2007; Costanza et al., 2009; Fletcher et al., 2010). Therefore, dechlorination activity is unlikely to occur during active thermal treatment of DNAPL source zones, but accelerated dechlorination and ethene formation may be observed during thermal treatment in the perimeter of the source zone where temperatures are in the range for Dhc activity. To effectively treat residual source zone contamination following thermal treatment, bioaugmentation with Dhc-containing PCE-toethene dechlorinating consortia may be necessary (Major et al., 2002; Lendvay et al., 2003; Lo¨ffler and Edwards, 2006; Hood et al., 2008; Scheutz et al., 2010). Previous studies reported the release of organic carbon from the subsurface matrix during thermal treatment and suggested that the increase in bioavailable electron donors could support microbial dechlorination (Friis et al., 2005, 2006a). However, the released organic carbon, suggested to be in the form of long-chain fatty acids (Friis et al., 2005), cannot directly serve as an electron donor for Dhc and must be fermented to produce H2. Furthermore, Dhc must compete with other organisms, such as methanogens, for available H2. Thermal treatment has been shown to reduce methanogenesis, even after bioaugmentation with the methanogenic
dechlorinating consortium KB-1, suggesting that dechlorinating Dhc may outcompete methanogens for H2 following thermal treatment (Friis et al., 2006a). Therefore, if thermal treatment both increases H2 flux and decreases the activity of competing H2 utilizers such as methanogens, dechlorination could be supported by indigenous electron donors, thus alleviating the need for biostimulation (i.e., the introduction of external substrates). To further explore the utility of combining thermal treatment and bioremediation, (i) the potential for dechlorination activity to occur in the perimeter of the source zone during thermal treatment and (ii) the impact of competing H2 utilizers (e.g., methanogens) on reductive dechlorination activity were evaluated experimentally. Microcosms, established with soils from two chlorinated ethene-contaminated sites, were incubated at 24, 35, 50, 70, and 95 C and, following cooling to 24 C, were bioaugmented with a methanogenic PCE-to-ethene-dechlorinating consortium. Reductive dechlorination and methanogenesis were monitored to determine if the activity of dechlorinators and/or their competitors correlated with previous incubation temperature.
2.
Materials and methods
2.1.
Site descriptions and soil preparation
Soil and groundwater samples impacted with TCE and cis-DCE (<0.01 mmol/mL of groundwater) were collected from the East Gate Disposal Yard (EGDY) in Fort Lewis, WA, from 28 to 36 feet below ground surface. Flushing with argon gas minimized exposure to oxygen and the analysis of geochemical parameters suggested reducing soil conditions prevailed (Costanza et al., 2009). The soil consisted of well-graded gravel in sand, silt, and clay with a total carbon content (TOC) of 0.69 g TOC/ kg. Additional details about the Ft. Lewis site are available (Friis et al., 2007). Ft. Lewis soil samples were collected prior to thermal treatment activities. Additional soil and groundwater samples impacted with PCE and less than 0.02 mmol/mL TCE, cis-DCE, and VC were collected from a former dry cleaner facility at the Naval Training Center in Great Lakes, IL, from between 8 and 10 feet below ground surface (Costanza et al., 2009, 2010). During sampling, the soil was exposed to air before it was broken into fragments with diameters no larger than 10 mm inside a disposable glove bag filled with ultra high purity argon. The solids were combined with an equal weight of site groundwater to create a slurry. Samples from both sites were collected prior to the commencement of thermal treatment activities.
2.2. Microcosm construction, incubation conditions, and sampling A total of 30 Ft. Lewis microcosms were constructed at room temperature (24 C) inside an anoxic chamber (Coy Laboratory Products, Ann Arbor, MI) that contained a 96% N2/4% H2 (vol/ vol) atmosphere. Serum bottles (70 mL, nominal capacity) were filled with 15 mL groundwater and soil was added to achieve a total volume of approximately 37 mL. A total of 20
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 2 5 e6 6 3 6
Great Lakes microcosms were constructed at room temperature inside an argon-filled glove bag. In 70 mL serum bottles, the soil/groundwater slurry was combined with 10e20 mL of sterile mineral salts medium (Amos et al., 2007) or site groundwater to yield a total volume of 40 mL. Mineral salts medium replaced groundwater when insufficient site groundwater was available. All microcosms were capped with sterile black butyl-rubber stoppers prior to removal from the anoxic tents. Preliminary experiments demonstrated that the headspace composition during microcosm setup did not change dechlorination profiles in the Ft. Lewis microcosms (unpublished data; Costanza et al., 2009). Immediately following microcosm construction, five Great Lakes and 15 Ft. Lewis microcosms were autoclaved at 121 C for 30 min to serve as killed controls. Triplicate Great Lakes and Ft. Lewis live microcosms and single Great Lakes and triplicate Ft. Lewis autoclaved controls were incubated statically in the dark in water baths adjusted to 24, 35, 50, 70, and 95 C. The temperature in microcosms incubated at 35 and 50 C was increased by 1 C per day whereas all other microcosms were immediately placed at the target temperatures. The 24e95 C temperature range was selected to include both optimal temperatures for the microbial reductive dechlorination process and temperatures that occur during thermal treatment. Because of the low chlorinated solvent concentrations in Ft. Lewis soil microcosms (less than 0.5 mmoles/microcosm) after 27 days of incubation, Ft. Lewis microcosms were amended with 5 mmoles of TCE dissolved in 2.25 mL filtersterilized, anoxic Ft. Lewis site groundwater. Great Lakes microcosms received no additional chlorinated solvent because PCE concentrations were at least 20 mmoles/microcosm. Following 4 months of incubation, the microcosms were cooled at a rate of 5 C per day to room temperature (24 C). An incubation period of 4 months was chosen to mimic site conditions where thermal remediation is generally employed for time periods of greater than weeks, but less than years (Heron et al., 2009). Microcosms were sampled periodically for chlorinated ethenes and gases including acetylene, CH4, CO2, and H2. Aqueous (1 mL) and gaseous (2 mL) samples were removed from microcosms using sterile plastic syringes with needles. Removed aqueous volumes were replaced with filter-sterilized groundwater or sterile medium, and removed gaseous volumes were replaced with sterile, oxygen-free N2 gas.
2.3. Transfer culture preparation, incubation conditions, and sampling After 138 days of incubation, 2 mL aqueous volumes were removed from each of the three Ft. Lewis microcosms previously incubated at 35 C and transferred to individual 160 mL serum bottles containing 100 mL mineral salts medium amended with vitamins (Wolin et al., 1963), 5 mM lactate, and 4 mmoles of TCE. The transfer cultures were initially incubated at 24 C and 1 mL aqueous samples were removed periodically for chlorinated ethene quantification. Once TCE was completely dechlorinated to cis-DCE during incubation at 24 C, transfer cultures were amended with another 4 mmoles of TCE and incubated at 35 C. A second batch of 35 C transfer
6627
cultures was generated by transferring culture suspension (3%, vol/vol) to the same mineral salts medium.
2.4.
Bioaugmentation and biostimulation
After the microcosms were cooled to 24 C, 10 mL of culture OW, a methanogenic, PCE-to-ethene-dechlorinating consortium, was added to both live and autoclaved microcosms. Consortium OW contains multiple Dhc strains along with Geobacter, Dehalobacter, and Sulfurospirillum populations implicated in chlorinated ethene reductive dechlorination (Daprato et al., 2007). Prior to inoculation, the OW consortium was removed from a draw-and-fill bioreactor (Daprato et al., 2007) and sparged with N2 gas for 30 min to remove chlorinated ethenes, ethene, and CH4. Inoculation (i.e., bioaugmentation) of Ft. Lewis and Great Lakes microcosms occurred on days 139 and 141 of incubation, respectively. The initial biostimulation event in Ft. Lewis and Great Lakes microcosms involved the amendment of microcosms with 5 mL (200 mmoles) of sterile H2 gas. Secondary biostimulation involved a one-time addition of 5 mL H2 gas to Ft. Lewis microcosms and involved the biweekly addition of 5 mL of H2 gas to Great Lakes microcosms. Initial and secondary biostimulation in Ft. Lewis microcosms occurred on days 153 and 258 of incubation or 14 and 119 days following bioaugmentation, respectively. In Great Lakes microcosms, initial and secondary biostimulation occurred on days 164 and 234 of incubation or 23 and 93 days following bioaugmentation, respectively.
2.5.
Analytical methods and calculations
Aqueous samples (1 mL) were collected for the quantification of chlorinated ethenes and ethene by gas chromatography (GC) following described procedures (Amos et al., 2007). Gaseous samples (2 mL) were collected for analysis of acetylene, CH4, CO2, and H2 using a Hewlett Packard 6890 GC equipped with a heated gas sampling valve, a 250 mL sample loop, and a 30 m length by 0.32 mm OD Carboxen-1010 column (Supelco, Bellefonte, PA) connected to a thermal conductivity detector. Detection limits were determined based on the lowest concentrations that were consistently quantifiable. To determine the rate of CO2 production, average CO2 concentrations measured after 7, 28, 62, 93, and 121 days were plotted versus incubation time. The data were fitted using linear regression analysis and, when concentrations were correlated with incubation time, the rate of CO2 production was determined from the slope of the linear regression line. The molar percent of a specific chlorinated ethene was determined according to the formula: MX =ðMPCE þ MTCE þ MDCE þ MVC þ Methene Þ ¼ MPX
(1)
Where Mx is the number of moles of the compound of interest (e.g., VC), MPCE is the number of moles of PCE, MTCE is the number of moles of TCE, MDCE is the number of moles of cisDCE, Mvc is the number or moles of vinyl chloride, Methene is the number of moles of ethene, and MPx is the molar percentage of the compound of interest. To determine how molar percentages change with time, the following formula was applied:
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 2 5 e6 6 3 6
MPx;tf MPx;ti ¼ DMPx
(2)
Where MPx,tf is the final molar percentage of the compound of interest, MPx,ti is the initial molar percentage of the compound of interest, and DMPx is the change in the molar percentage of the compound of interest over time, which is also represented as a percentage. The calculation of the moles of electrons consumed in reductive dechlorination of PCE-to-ethene assumed that each dechlorination step required two electrons (Lo¨ffler and Edwards, 2006) and therefore, the moles of electrons consumed in reductive dechlorination were calculated for the Great Lakes microcosms according to the formula: 2ðMTCE Þ þ 4 MDCES þ 6ðMVC Þ þ 8ðMethene Þ ¼ Mconsumed
(3)
Where Mconsumed is the number of moles of electrons consumed in reductive dechlorination of PCE to ethene. The moles of electrons used for reductive dechlorination of TCE to ethene in Ft. Lewis microcosms was calculated according to the formula: 2 MDCES þ 4ðMVC Þ þ 6ðMethene Þ ¼ Mconsumed
(4)
The calculation of the moles of electrons consumed in methanogenesis assumed that 4 moles of electrons are required per mole of CH4 formed (Ferry, 1999). Dechlorination extent in the Great Lakes microcosms was determined according to the formula: ð100Þ½1ðMTCE Þ þ 2ðMDCE Þ þ 3ðMVC Þ þ 4ðMethene Þ=½4ðMtot Þ ¼ DE
(5)
Where Mtot is the total number of moles of chlorinated ethenes and ethene, and DE is the dechlorination extent in percent (Friis et al., 2006b). In the Ft. Lewis microcosms, dechlorination extent was determined according to the formula: ð100Þ½1ðMDCE Þ þ 2ðMVC Þ þ 3ðMethene Þ=½3ðMtot Þ ¼ DE
(6)
Concentrations measured in microcosms were compared using two-tailed student’s t-tests and correlations were identified based on Pearson product moment correlation coefficients. In all cases, p-values below 0.05 were considered significant. In order to visualize values for multiple variables and correlations between variables in the same space, principal component analysis (PCA) was performed using the ViSta program (Young, 1996). PCA is a statistical approach used with multivariate data to reduce the number of dimensions that account for the observed variance to principal components (PCs). Prior to PCA, the data were normalized so that all arithmetic means had values of 0 and standard deviations of 1 using the ViSta program. The outputs of the PCA were the eigenvalue coordinates for specific microcosm treatments, the eigenvectors, which correspond to measured variables (e.g., the methane concentration, the time required for dechlorination), and the percentages of the total variation explained by each PC (e.g., PC 1 and PC 2). Eigenvalues and eigenvectors were plotted versus the PCs encompassing the highest percentages of the total variation (i.e., PC 1 and PC 2). Eigenvalue coordinates can be used to infer how similar the variables included in the PCA were between microcosms. The eigenvalue direction and length provide information as well.
For example, eigenvectors pointing in opposite directions indicate that variables have opposite profiles and may be negatively correlated. The longer the eigenvector, the more strongly the variable is related to the eigenvalue coordinates.
3.
Results
3.1.
Incubation at elevated temperatures
3.1.1.
Ft. Lewis microcosms
After 28 days of incubation, H2 concentrations in live Ft. Lewis microcosms incubated at 24, 35, and 50 C were significantly lower ( p < 0.05) compared to those measured in autoclaved microcosms (Fig. 1). In contrast, H2 concentrations in live Ft. Lewis microcosms incubated at 70 and 95 C were not significantly lower than those in autoclaved microcosms (Fig. 1). This pattern persisted throughout the 120 day incubation period. CO2 concentrations in all live microcosms, except those previously incubated at 95 C, were significantly lower ( p < 0.05) after 28 days of incubation than concentrations in autoclaved control microcosms incubated at the same temperatures. CO2 concentrations in live Ft. Lewis microcosms ranged from 1830 930 to 10,360 245 ppmv in microcosms incubated at 24 and 95 C, respectively, and were positively correlated with incubation temperature ( p < 0.05). Between day 7 and day 121, CO2 concentrations increased in all live microcosms and were positively correlated with incubation time ( p < 0.05) in microcosms incubated at 24, 35, 50, and 70 C. In fact, CO2 production rates decreased exponentially with increasing temperature (Fig. 2). No positive correlation was observed between CO2 concentrations and incubation time in the autoclaved microcosms. After 120 days of incubation, TCE concentrations in all live microcosms, except those incubated at 24 C, were not significantly lower than concentrations in autoclaved microcosms incubated at the same temperatures (Fig. S1). In live Ft. Lewis microcosms incubated at 24 C, TCE decreased to below the detection limit
H2 Concentration (ppmv)
6628
25000 20000
15000 10000 5000 0
* 24
35 50 70 95 Incubation Temperature (°C)
Fig. 1 e H2 concentrations in live (filled bars) and autoclaved (open bars) Ft. Lewis microcosms after 28 days of incubation. The asterisk indicates that H2 was below the detection limit in live microcosms incubated at 24 C. Error bars represent the standard error.
Rate of CO2 Production (ppmv/day)
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 2 5 e6 6 3 6
80
dechlorination products in all microcosms remained below 0.5 mmoles/microcosm.
60
3.2.
40
20
0
20
40
60
80
Incubation Temperature (°C) Fig. 2 e Exponential regression analysis of the rate of CO2 production in live Ft. Lewis microcosms during 7e121 days of incubation at temperatures of 24, 35, 50, and 70 C. The regression indicates that the rate of CO2 production is inversely exponentially correlated with incubation temperature. Error bars represent standard errors of CO2 production rates.
of 0.05 mmoles/microcosm after 120 days of incubation and cisDCE concentrations increased concomitantly; however, dechlorination beyond cis-DCE did not occur.
3.1.2.
6629
Great Lakes microcosms
In Great Lakes microcosms, H2 concentrations were below the detection limit in all microcosms after 58 days of incubation. After 28 days of incubation, CO2 concentrations were lower in all live microcosms incubated at 24, 35, 50, and 70 C (37,330 2,730, 44,560 6,870, 44,830 4,320, and 37,620 1080 ppmv, respectively) than in autoclaved microcosms incubated at the same temperatures (53,450, 73,450, 57,970, and 48,680 ppmv, respectively); however, CO2 concentrations in live microcosms incubated at 95 C (48,670 3360 ppmv) were similar to those in the autoclaved microcosm incubated at 95 C (49,500 ppmv). In live microcosms incubated at 24, 50, and 70 C, CO2 concentrations were positively correlated ( p < 0.05) with incubation time and increased to 60,680 5810, 64,020 9,480, and 75,060 1740 ppmv, respectively, after 123 days of incubation. CO2 concentrations were also positively correlated ( p < 0.05) with incubation time in autoclaved microcosms incubated at 50 and 70 C and increased to 97,460 and 70,990 ppmv, respectively, after 123 days of incubation. After 123 days of incubation, PCE persisted in all Great Lakes microcosms, but lower PCE concentrations were measured in the autoclaved microcosms incubated at 24, 35, 50, and 70 C compared to the live microcosms incubated at the same temperatures (Fig. S1). The PCE concentration in the autoclaved microcosm incubated at 95 C was within the range of concentrations measured in the live microcosms incubated at 95 C (11.0 and 11.4 1.2 mmoles/microcosm, respectively). TCE, cis-DCE, and VC concentrations were similar in live and autoclaved microcosms, and the concentrations of possible PCE
Effects of cooling
Ft. Lewis and Great Lakes microcosms were cooled to 24 C from elevated temperatures to determine if microbial activity would resume and in preparation for bioaugmentation. The concentration of H2 decreased significantly ( p < 0.05) during cooling in live Ft. Lewis microcosms previously incubated at 70 C whereas H2 concentrations changed insignificantly in autoclaved Ft. Lewis microcosms previously incubated at 70 C. With the exception of the Ft. Lewis microcosms incubated at 24 C, PCE and TCE persisted in all Great Lakes and Ft. Lewis microcosms, respectively, and no dechlorination products were formed. Higher PCE and TCE concentrations were present in microcosms previously incubated at lower temperatures and PCE and TCE concentrations were negatively correlated with previous incubation temperature ( p < 0.05).
3.3. Reductive dechlorination in Ft. Lewis transfer cultures Since reductive dechlorination occurred in Ft. Lewis microcosms incubated at 24 C, the lack of dechlorination activity in Ft. Lewis microcosms previously incubated at 35 C was investigated. In transfer cultures established from microcosms previously incubated at 35 C and cooled to 24 C, stoichiometric reductive dechlorination of TCE to cis-DCE occurred in 20 days during incubation at 24 C, indicating that dechlorinators survived the extended 35 C incubation period. After the transfer cultures were placed at 35 C and amended with additional TCE, TCE was completely dechlorinated to cisDCE within 2 days. When the culture suspension was transferred to fresh medium, TCE to cis-DCE reductive dechlorination activity again occurred during incubation at 35 C.
3.4. Reductive dechlorination following bioaugmentation Dechlorination of chlorinated ethenes to ethene did not occur in any of the Ft. Lewis or Great Lakes microcosms and all microcosms were bioaugmented with the PCE-to-ethenedechlorinating consortium OW. In Ft. Lewis microcosms, at least 95% of the TCE was dechlorinated to VC within 5 days of bioaugmentation (Fig. 3 and Figs. S3, S4). In Great Lakes microcosms, at least 85% of the PCE was dechlorinated to cisDCE and VC 3 days after bioaugmentation (Fig. 3 and Figs. S2eS4). The only Ft. Lewis microcosms demonstrating a significant increase ( p < 0.05) in the molar percentage of ethene 10e13 days following bioaugmentation were live microcosms previously incubated at 24 C. No significant increase in the molar percentages of VC and ethene occurred in any of the live Great Lakes microcosms between 11 and 23 days following bioaugmentation. In autoclaved Great Lakes microcosms, the molar percentage of VC increased by a maximum of 7.9% between day 11 and day 23 following bioaugmentation (Fig. 3). In all Ft. Lewis and Great Lakes
6630
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 2 5 e6 6 3 6
Fig. 3 e The change in the molar percentage of VC with time after bioaugmentation in Ft. Lewis and Great Lakes microcosms previously incubated at 24 (crosses), 35 (circles), 50 (triangles), 70 (diamonds), and 95 C (squares) in live (filled symbols) and autoclaved (open symbols) microcosms. Vertical dashed lines indicate the initial biostimulation event and the second biostimulation event (Ft. Lewis microcosms) or the beginning of biweekly secondary biostimulation (Great Lakes microcosms). Error bars have been omitted for clarity. Note that the time scales on the x-axes are shorter prior to the initial biostimulation event.
microcosms, H2 concentrations were below the detection limit 13 and 23 days after bioaugmentation, respectively. The electron consumption in reductive dechlorination was calculated to determine how previous incubation temperature affected electron donor utilization (Fig. 4). Over an order of magnitude more electrons were consumed in the reductive dechlorination process in Great Lakes microcosms than in Ft. Lewis microcosms. In microcosms from both sites, the moles of electrons consumed in reductive dechlorination following bioaugmentation (and prior to biostimulation) were negatively correlated with previous incubation temperature ( p < 0.05) (Fig. 4). Apparently, more reducing equivalents were directed toward the reductive dechlorination in microcosms that had been incubated at lower temperatures.
3.5. Reductive dechlorination following biostimulation with H2 Because dechlorination ceased prior to complete conversion of chlorinated ethenes to ethene, the microcosms were biostimulated with H2 14 days after bioaugmentation. The molar percentage of VC in Ft. Lewis microcosms decreased by 47.3 34.2% 17 days after biostimulation (i.e., 31 days following bioaugmentation) and, with the exception of live microcosms previously incubated at 50 and 70 C, 90% of VC was dechlorinated to ethene within 42 days of biostimulation (Table 1, Fig. 3). In Ft. Lewis live microcosms previously
incubated at 50 and 70 C, at least 70% of the initial TCE existed as VC 92 days after biostimulation in two of three and one of three microcosms, respectively. In all of these microcosms, H2 concentrations were below the detection limit. Following a second H2 amendment 105 days after the initial biostimulation, ethene production continued and 90% of the VC was dechlorinated within 29 days (Table 1, Fig. 3). In Great Lakes microcosms, the molar percentage of VC increased by 11.3 7.4% due to the conversion of cis-DCE to VC 2 days after biostimulation (25 days following bioaugmentation). In autoclaved and live Great Lakes microcosms previously incubated at 95 C, 100% of VC was converted to ethene 29 and 60 days after biostimulation, respectively. The molar percentage of VC did not decrease significantly from 60 to 68 days following biostimulation in live Great Lakes microcosms previously incubated at temperatures below 95 C. In autoclaved Great Lakes microcosms previously incubated at the same temperatures, the molar percentage of VC decreased by a maximum of 3% during the same period (Fig. 3). No H2 was detected in any of the Great Lakes microcosms 68 days after biostimulation. The molar percentage of VC decreased by 41.6 24.1% in response to 12 days of biweekly biostimulation. Within 27 days of the beginning of secondary biostimulation, VC was consumed to below the detection limit of 0.05 mmoles per vessel in all microcosms except the live microcosms previously incubated at 24 and 70 C (Table 1, Fig. 3). In live microcosms previously incubated at 24 and 70 C, the VC
6631
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 2 5 e6 6 3 6
Electrons (µmoles/microcosm)
16
Ft. Lewis - Live
16
12
12
8
8
4
4
0
0 24
300 250
Ft. Lewis Autoclaved
*
35
50
70
95
Great Lakes - Live
24 300
100
50
50
0
0 35
50
70
95
150
100
24
70
200
*
150
50
Great Lakes Autoclaved
250
200
35
95
Previous Incubation Temperature (C)
24
35
50
70
95
Previous Incubation Temperature (C)
Fig. 4 e Electron consumption in reductive dechlorination prior to bioaugmentation (black bars), following bioaugmentation (dark gray bars), following the initial biostimulation event (light gray bars), and following secondary biostimulation (white bars). Values are shown for Ft. Lewis and Great Lakes microcosms 13 and 23 days after bioaugmentation, 119 and 91 days following bioaugmentation, and after 90% dechlorination of VC and either complete dechlorination of VC or the completion of the experiment, respectively. Asterisks indicate that complete dechlorination did not occur during the incubation period. Error bars represent the standard error and negative error bars have been removed for clarity. No error bars are shown for autoclaved Great Lakes microcosms because only one microcosm was incubated at each temperature.
molar percentages were 0.5 0.1% and 73.6 37.4% 27 days after the beginning of secondary biostimulation. The times required for 90% of VC to be dechlorinated to ethene in Ft. Lewis microcosms and for 100% of VC to be dechlorinated to ethene in Great Lakes microcosms were not correlated with previous incubation temperatures.
3.6.
Competition for reducing equivalents
Prior to bioaugmentation, CH4 was below 600 ppmv in all microcosms. Within 23 days of bioaugmentation, average CH4 concentrations were at least 21,000 ppmv (Fig. 5). The ratio of the number of electrons consumed in methanogenesis versus reductive dechlorination prior to biostimulation was at least 1,300, indicating that the dechlorinators consumed less than 0.08% of the available reducing equivalents. CH4 concentrations in live Ft. Lewis microcosms previously incubated at 24 and 35 C were significantly higher ( p < 0.05) than in the autoclaved microcosms incubated at the same temperatures. In live Ft. Lewis microcosms previously incubated at temperatures of 50, 70, and 95 C, CH4 concentrations were not
significantly different from concentrations in autoclaved microcosms incubated at the same temperatures. To compare electron donor consumption in reductive dechlorination versus methanogenesis, the CH4 concentration, the number of electrons consumed in reductive dechlorination, and the dechlorination extent in Great Lakes microcosms following bioaugmentation and prior to biostimulation were compared using PCA (Fig. 6). The CH4 concentration and dechlorination extent eigenvectors point in a similar direction and CH4 concentration was significantly positively correlated ( p < 0.05) with dechlorination extent. The CH4 concentration eigenvector is perpendicular to the eigenvector for the moles of electrons consumed in reductive dechlorination, and CH4 concentrations were significantly negatively correlated ( p < 0.05) with the moles of electrons consumed in dechlorination. These results indicate that microcosms that produced more CH4 consumed fewer electrons in reductive dechlorination, but also exhibited higher dechlorination extents because lower PCE concentrations were present prior to bioaugmentation. The PC1 eigenvalues of microcosms previously incubated at lower temperatures were generally lower than those of microcosms that
6632
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 2 5 e6 6 3 6
Table 1 e The total numbers of days required for 90% of VC to be dechlorinated to ethene and for complete dechlorination of VC to ethene in Ft. Lewis and Great Lakes microcosms, respectively. The numbers of days following biostimulation are listed in parentheses and underlined values indicate that secondary biostimulation was required. Asterisks indicate that complete dechlorination of VC to ethene did not occur during the incubation period. Previous incubation temperature
Days for dechlorination to ethene Ft. Lewis microcosms Live
24 35 50 70 95
C C C C C
23 48 148 133 48
Autoclaved
(9) (34) (29) (14) (34)
48 38 56 20 20
experienced higher previous incubation temperatures. The direction of the eigenvectors suggests that microcosms previously incubated at lower temperatures consumed more electrons in the reductive dechlorination process and produced less CH4. In fact, the previous incubation temperature was significantly positively correlated ( p < 0.05) with CH4 concentration, indicating that microcosms previously incubated at lower temperatures generally produced less CH4 than microcosms incubated at higher temperatures. In Ft. Lewis microcosms, CH4 concentrations increased by over 30,000 ppmv following the initial H2 biostimulation event and increased by 16,070 9370 ppmv following the second biostimulation event (Fig. 5). At least 19,000 ppmv of CH4 were produced following the initial biostimulation event and prior
CH4 Concentration (ppmv)
300,000
Great lakes microcosms
Ft. Lewis - Live
Live
(34) (24) (42) (6) (6)
141* 105 120 141* 83
300,000 250,000
200,000
200,000
150,000
150,000
100,000
100,000
50,000
50,000
105 105 113 113 52
(12) (12) (20) (20) (29)
Ft. Lewis Autoclaved
0 24
400,000
(48) (12) (27) (48) (60)
to the beginning of secondary biostimulation events in all Great Lakes microcosms except for in those microcosms previously incubated at 50 C (Fig. 5). During the period following the initial biostimulation and prior to secondary biostimulation, no CH4 was produced in the live Great Lakes microcosms previously incubated at 50 C, and CH4 concentrations increased by less than 3000 ppmv in autoclaved microcosms previously incubated at 50 C (Fig. 5). In Great Lakes microcosms, CH4 concentrations increased by at least 74,000 ppmv during secondary biostimulation. These concentration increases were not correlated with previous incubation temperature. CH4 concentration, the time required to achieve 90% VC dechlorination, and the number of electrons consumed in
250,000
0
Autoclaved
35
50
70
95
Great Lakes - Live
24 400,000
300,000
300,000
200,000
200,000
100,000
100,000
35
50
70
95
Great Lakes Autoclaved
0
0 24
35
50
70
95
Previous Incubation Temperature ( C)
24
35
50
70
95
Previous Incubation Temperature ( C)
Fig. 5 e The concentration of CH4 prior to biostimulation (dark gray bars), following the initial biostimulation event (light gray bars), and following secondary biostimulation (white bars). Error bars represent the standard error and negative error bars have been removed for clarity. No error bars are shown for autoclaved Great Lakes microcosms because only one microcosm was incubated at each temperature.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 2 5 e6 6 3 6
6633
Fig. 6 e PCA ordination plots of Ft. Lewis microcosms after 90% VC dechlorination and of Great Lakes microcosms prior to biostimulation. Vectors represent the CH4 concentration, electrons consumed in reductive dechlorination, time required for 90% VC dechlorination (Ft. Lewis plot only) and dechlorination extent (Great Lakes plot only). The symbols represent microcosms incubated at 24 C (crosses), 35 C (circles), 50 C (triangles), 70 C (diamonds), and 95 C (squares) in live (filled symbols) and autoclaved (open symbols) microcosms. PC 1 and PC 2 are the PCs generated through PCA that encompass the greatest percentages of variation between the variables shown on each graph (i.e., in the Ft. Lewis graph, PC 1 encompasses the greatest percentage of variation in CH4 concentrations, electron consumption in reductive dechlorination, and the time required for dechlorination). The percentages of the total variation explained by each PC are shown in parentheses.
reductive dechlorination in Ft. Lewis microcosms following 90% VC dechlorination were compared using PCA (Fig. 6). None of the eigenvectors point in the same direction and, in fact, none of these variables were significantly positively correlated. Further, as all eigenvalues have similar lengths, each of these variables contributed approximately equally to eigenvalue variation. Similar to the PCA results obtained with the Great Lakes microcoms, the PC1 eigenvalues of microcosms previously incubated at lower temperatures were generally lower than those of microcosms previously incubated at higher temperatures. Based on the directions of eigenvectors, these findings indicate that microcosms previously incubated at lower temperatures produced less CH4 and required more electrons for 90% of VC to be dechlorinated to ethene. In fact, previous incubation temperature was positively correlated ( p < 0.05) with CH4 concentrations and negatively correlated ( p < 0.05) with the number of electrons consumed in reductive dechlorination (Fig. 6). Fewer electrons were required for 90% of VC to be dechlorinated to ethene in micorcosms previously incubated at elevated temperatures and TCE concentrations following cooling were negatively correlated with previous incubation temperature.
4.
Discussion
4.1. No degradation of chlorinated compounds during incubation at elevated temperatures PCE and TCE concentrations in autoclaved microcosms were either similar to or less than the concentrations in live microcosms (with the exception of the Ft. Lewis microcosms incubated at 24 C), indicating that biotic dechlorination did not occur in the Ft. Lewis microcosms incubated at
temperatures of 35 C and above or in any of the Great Lakes microcosms. Previous studies conducted with soils collected from the same sites demonstrated that the half-lives of PCE and TCE were approximately 7000 and 640 days during incubation at 95 C (Costanza and Pennell, 2007b; Costanza et al., 2009). Therefore, the differences in PCE and TCE concentrations are primarily due to loss (i.e., diffusion through or sorption to the rubber stopper) rather than degradation. This conclusion is consistent with a previous report, which demonstrated that the loss of PCE and TCE from vials sealed with polymer septa and incubated at 50 C was due to diffusion through and sorption to the septa (Costanza and Pennell, 2007a). The lack of microbial reductive dechlorination activity in Ft. Lewis microcosms previously heated in the laboratory to 100 C was also observed in an earlier study (Friis et al., 2007). Therefore, for both the Ft. Lewis and Great Lakes sites, transformation of PCE and TCE in the source zone is not likely to occur during thermal treatment.
4.2. Lack of dechlorination activity at non-inhibitory temperatures Transfer culture experiments demonstrated that microorganisms capable of TCE to cis-DCE dechlorination at 35 C were present in microcosms incubated at 35 C, yet no dechlorination occurred. Ft. Lewis soils and microcosms were handled under anoxic conditions, and previous studies have shown that redox conditions do not change due to heating (Friis et al., 2005, 2007). Therefore, unfavorable geochemical/ redox conditions cannot explain the lack of dechlorination activity. Increased temperatures affect the activity of both fermenters and consumers of fermentation products, including microbes that compete with the dechlorinators for electron donors (e.g., H2) (Wiegel, 1990; Rui et al., 2009; Noll
6634
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 2 5 e6 6 3 6
et al., 2010). Therefore, the lack of dechlorination activity in microcosms incubated at 35 C may have been due to inadequate electron donor availability. These findings indicate that dechlorination may not occur without electron donor amendment even when native bacteria capable of dechlorination are present and experience non-inhibitory temperatures. Thermal treatment has been shown to release bioavailable electron donors (Friis et al., 2005, 2006a); however, the findings of our study emphasize that careful evaluation of the electron donor availability (e.g., H2 concentration measurements) should be performed to ensure that the supply of reducing equivalents does not limit reductive dechlorination activity following thermal treatment.
4.3. Biostimulation was required to achieve complete dechlorination to ethene following bioaugmentation The production of CH4 coupled to the accumulation of cis-DCE and VC in all Ft. Lewis and Great Lakes microcosms following bioaugmentation demonstrates that the supply of electron donor was inadequate to achieve complete dechlorination to ethene due, at least in part, to consumption of H2 by competitors. Prior to the stall in dechlorination activity, on average one order of magnitude more electrons were consumed in reductive dechlorination in Great Lakes microcosms than in Ft. Lewis microcosms, but the major dechlorination products in all microcosms were cis-DCE and VC. These results suggest that cis-DCE and VC-dechlorinating Dhc were electron donor (i.e., H2) limited whereas the PCE and TCE dechlorinators were more competitive for electron donor(s). It is likely that methanogens outcompeted cis-DCE and VC-dechlorinating Dhc, but not PCE and TCE-dechlorinaters, because (i) Dhc are strict hydrogenotrophs whereas PCE/TCE dechlorinators are more versatile in terms of electron donor requirement, (ii) reductive dechlorination of cis-DCE and VC generally occurs at slower rates than PCE and TCE dechlorination, and (iii) VC dechlorination is inhibited by polychlorinated ethenes (Haston and McCarty, 1999; Maymo´-Gatell et al., 1999). After 90% of VC had been dechlorinated in Ft. Lewis microcosms, CH4 concentrations were positively correlated with previous incubation temperature. That is, more CH4 was produced in microcosms previously incubated at elevated temperatures, likely by methanogens introduced with the bioaugmentation inoculum. Similarly, following bioaugmentation in Great Lakes microcosms, more CH4 was generally produced in microcosms previously incubated at higher temperatures. Possible reasons for the higher concentrations of CH4 in microcosms previously incubated at higher temperatures include the elevated concentrations of CO2, which serves as an electron acceptor for methanogens (Ferry, 1999) and the release of fermentable carbon substrates as a source of H2, allowing fast-growing H2-consumers (e.g., hydrogenotrophic methanogens) to dominate (He et al., 2002).
4.4. Inhibition of reductive dechlorination following thermal treatment Dechlorination was incomplete in the live Great Lakes microcosms previously incubated at 70 C even after bioaugmentation and during secondary, biweekly biostimulation,
suggesting that inhibitory conditions prevailed. In fact, microcosms constructed with soils collected from both the Ft. Lewis and Great Lakes sites demonstrated limited dechlorination activity following cooling from 70 C. In a previous bioaugmentation and biostimulation microcosm study, dechlorination of TCE to VC and ethene occurred in unheated microcosms whereas TCE was only dechlorinated to cis-DCE in microcosms that had previously been heated to 100 C (Friis et al., 2006a). These observations suggest that exposing soils to elevated temperatures can impact dechlorination activity following cooling, possibly due to the release of compounds from the soil matrix that inhibit the microbial reductive dechlorination process. While limited dechlorination was observed in bioaugmented and biostimulated Ft. Lewis and Great Lakes microcosms previously incubated at 70 C, bioaugmentation and biostimulation supported complete reductive dechlorination to ethene in microcosms previously incubated at 95 C. These observations suggest that thermal treatment can affect the release and/or availability of compounds inhibitory to the reductive dechlorination process.
4.5.
Competition for reducing equivalents
Thermal treatment reduces microbial cell numbers and opens up ecological niches that will be reoccupied following cooling. In a natural setting, microbes surviving the heat treatment (e.g., spore-formers) and microbes transported with the groundwater flow from up-gradient locations will colonize available niches over time. The application of thermal treatment combined with bioaugmentation offers an opportunity to establish microbes with desirable activities, in this case bacteria capable of reductive dechlorination of chlorinated ethenes, in the thermally treated zones. Bioaugmentation can achieve a “founder effect” and establish a microbial community that efficiently ferments the available organic substrates (i.e., electron donors released during heating) to generate H2. The increased H2 flux supports chlorinated ethene-respiring bacteria, including Dhc, leading to the enhanced reductive dechlorination of residual contaminants. Since methanogens compete for the same electron donor (i.e., H2) and generally exhibit higher growth rates than chlorinated ethene-respiring bacteria, methanogens will dominate thermally treated sites following cooling (i.e., methanogens will reoccupy available niches more rapidly and consume H2). Therefore, consortia without methanogens may offer an advantage for implementing bioaugmentation at thermally treated sites and the application of methanogen-free, PCE-to-ethene-dechlorinating consortia should be explored at thermally treated sites.
5.
Conclusions
This study provides new information for the application of the microbial reductive dechlorination process during and following thermal treatment. The findings from this study are relevant for thermal treatment sites; however, the extrapolation of the findings to field sites should be done cautiously because only laboratory experiments were conducted with materials from two field sites. The major conclusions from this study include:
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 2 5 e6 6 3 6
Elevated temperatures above 35 C inhibit microbial reductive dechlorination of chlorinated ethenes. Even when temperatures are not inhibitory to dechlorinating populations, dechlorination may not occur due to electron donor limitations (i.e., the electron donors released during soil heating are insufficient or are being consumed in competing microbial processes). Thermal treatment may generate conditions favorable for competing microbial processes. Specifically, methanogenesis may consume a significant fraction of the reducing equivalents available from electron donor(s) released during thermal treatment, thus negatively impacting dechlorination extent. This observation suggests that bioaugmentation with non-methanogenic consortia can realize higher efficiency in terms of reducing equivalents directed toward the reductive dechlorination process.
Acknowledgments Support for this research was provided by the Strategic Environmental Research and Development Program (SERDP) under contract W912HQ-05-C-008 for Project ER-1419, “Investigation of Chemical Reactivity, Mass Recovery and Biological Activity During Thermal Treatment of DNAPL”. This work has not been subject to SERDP review and no official endorsement should be inferred.
Appendix. Supplementary data Supplementary data associated with this article can be found in online version at doi:10.1016/j.watres.2011.09.033.
references
Amos, B.K., Christ, J.A., Abriola, L.M., Pennell, K.D., Lo¨ffler, F.E., 2007. Experimental evaluation and mathematical modeling of microbially enhanced tetrachloroethene (PCE) dissolution. Environmental Science & Technology 41 (3), 963e970. Amos, B.K., Ritalahti, K.M., Cruz-Garcia, C., Padilla-Crespo, E., Lo¨ffler, F.E., 2008. Oxygen effect on Dehalococcoides viability and biomarker quantification. Environmental Science & Technology 42 (15), 5718e5726. Christ, J.A., Ramsburg, C.A., Abriola, L.M., Pennell, K.D., Lo¨ffler, F. E., 2005. Coupling aggressive mass removal with microbial reductive dechlorination for remediation of DNAPL source zones: a review and assessment. Environmental Health Perspectives 113 (4), 465e477. Costanza, J., Pennell, K.D., 2007a. Comparison of PCE and TCE disapperance in heated volatile organic analysis vials and flame-sealed ampules. Chemosphere 70 (11), 2060e2067. Costanza, J., Pennell, K.D., 2007b. Distribution and abiotic degradation of chlorinated solvents in heated field samples. Environmental Science & Technology 41 (5), 1729e1734. Costanza, J., Fletcher, K.E., Lo¨ffler, F.E., Pennell, K.D., 2009. Fate of TCE in heated Fort Lewis soil. Environmental Science & Technology 43 (3), 909e914.
6635
Costanza, J., Otan˜o, G., Callaghan, J., Pennell, K.D., 2010. PCE oxidation by sodium persulfate in the presence of solids. Environmental Science & Technology 44 (24), 9445e9450. Daprato, R.C., Lo¨ffler, F.E., Hughes, J.B., 2007. Comparative analysis of three tetrachloroethene to ethene halorespiring consortia suggests functional redundancy. Environmental Science & Technology 41 (7), 2261e2269. Davis, E., 1998. Steam Injection for Soil and Aquifer Remediation. EPA/540/S-97/505. Robert S. Kerr Environmental Research Laboratory, Ada, OH. Ferry, J.G., 1999. Enzymology of one-carbon metabolism in methanogenic pathways. FEMS Microbiology Reviews 23 (1), 13e38. Fletcher, K.E., Costanza, J., Cruz-Garcia, C., Ramaswamy, N.S., Pennell, K.D., Lo¨ffler, F.E., 2010. Effects of elevated temperature on Dehalococcoides dechlorination performance and DNA and RNA biomarker abundance. Environmental Science & Technology 45 (2), 712e718. Friis, A.K., Albrechtsen, H.J., Heron, G., Bjerg, P.L., 2005. Redox processes and release of organic matter after thermal treatment of a TCE-contaminated aquifer. Environmental Science & Technology 39 (15), 5787e5795. Friis, A.K., Albrechtsen, H.J., Cox, E., Bjerg, P.L., 2006a. The need for bioaugmentation after thermal treatment of a TCEcontaminated aquifer: laboratory experiments. Journal of Contaminant Hydrology 88 (3e4), 235e248. Friis, A.K., Heron, G., Albrechtsen, H.J., Udell, K.S., Bjerg, P.L., 2006b. Anaerobic dechlorination and redox activities after fullscale electrical resistance heating (ERH) of a TCEcontaminated aquifer. Journal of Contaminant Hydrology 88 (3e4), 219e234. Friis, A.K., Edwards, E.A., Albrechtsen, H.J., Udell, K.S., Duhamel, M., Bjerg, P.L., 2007. Dechlorination after thermal treatment of a TCE-contaminated aquifer: laboratory experiments. Chemosphere 67 (4), 816e825. Haston, Z.C., McCarty, P.L., 1999. Chlorinated ethene half-velocity coefficients (Ks) for reductive dehalogenation. Environmental Science & Technology 33 (2), 223e226. He, J.Z., Ritalahti, K.M., Yang, K.L., Koenigsberg, S.S., Lo¨ffler, F.E., 2003. Detoxification of vinyl chloride to ethene coupled to growth of an anaerobic bacterium. Nature 424 (6944), 62e65. He, J.Z., Sung, Y., Dollhopf, M.E., Fathepure, B.Z., Tiedje, J.M., Lo¨ffler, F.E., 2002. Acetate versus hydrogen as direct electron donors to stimulate the microbial reductive dechlorination process at chloroethene-contaminated sites. Environmental Science & Technology 36 (18), 3945e3952. Heron, G., Parker, K., Galligan, J., Holmes, T.C., 2009. Thermal treatment of eight CVOC sources zones to near nondetect concentrations. Ground Water Monitoring & Remediation 29 (3), 56e65. Heron, G., van Zutphen, M., Christensen, T.H., Enfield, C.G., 1998. Soil heating for enhanced remediation of chlorinated solvents: a laboratory study on resistive heating and vapor extraction in a silty, low-permeable soil contaminated with trichloroethylene. Environmental Science & Technology 32 (10), 1474e1481. Hood, E.D., Major, D.W., Quinn, J.W., Yoon, W.S., Gavaskar, A., Edwards, E.A., 2008. Demonstration of enhanced bioremediation in a TCE source area at Launch Complex 34, Cape Canaveral air force station. Ground Water Monitoring & Remediation 28 (2), 98e107. Lendvay, J.M., Lo¨ffler, F.E., Dollhopf, M., Aiello, M.R., Daniels, G., Fathepure, B.Z., Gebhard, M., Heine, R., Helton, R., Shi, J., Krajmalnik-Brown, R., Major, C.L., Barcelona, M.J., Petrovskis, E., Hickey, R., Tiedje, J.M., Adriaens, P., 2003. Bioreactive barriers: a comparison of bioaugmentation and biostimulation for chlorinated solvent remediation. Environmental Science & Technology 37 (7), 1422e1431.
6636
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 2 5 e6 6 3 6
Lo¨ffler, F.E., Edwards, E.A., 2006. Harnessing microbial activities for environmental cleanup. Current Opinion in Biotechnology 17 (3), 274e284. Major, D.W., McMaster, M.L., Cox, E.E., Edwards, E.A., Dworatzek, S.M., Hendrickson, E.R., Starr, M.G., Payne, J.A., Buonamici, L.W., 2002. Field demonstration of successful bioaugmentation to achieve dechlorination of tetrachloroethene to ethene. Environmental Science & Technology 36 (23), 5106e5116. Maymo´-Gatell, X., Anguish, T., Zinder, S.H., 1999. Reductive dechlorination of chlorinated ethenes and 1,2-dichloroethane by “Dehalococcoides ethenogenes” 195. Applied and Environmental Microbiology 65 (7), 3108e3113. McGuire, T.M., MDade, J.M., Newell, C.J., 2006. Performance of DNAPL source zone depletion technologies at 59 chlorinated solvent-impacted sites. Groundwater Monitoring & Remediation 26 (1), 73e84. Noll, M., Klose, M., Conrad, R., 2010. Effect of temperature change on the composition of the bacterial and archaeal community potentially involved in the turnover of acetate and propionate in methanogenic rice field soil. FEMS Microbiology Ecology 73 (2), 215e225. Ramsburg, C.A., Abriola, L.M., Pennell, K.D., Lo¨ffler, F.E., Gamache, M., Amos, B.K., Petrovskis, E.A., 2004. Stimulated
microbial reductive dechlorination following surfactant treatment at the Bachman Road site. Environmental Science & Technology 38 (22), 5902e5914. Rui, J.P., Peng, J.J., Lu, Y.H., 2009. Succession of bacterial populations during plant residue decomposition in rice field soil. Applied and Environmental Microbiology 75 (14), 4879e4886. Sale, T.C., McWhorter, D.B., 2001. Steady state mass transfer from single-component dense nonaqueous phase liquids in uniform flow fields. Water Resources Research 37 (2), 393e404. Scheutz, C., Broholm, M.M., Durnat, N.D., Weeth, E.B., Jørgensen, T.H., Dennis, P., Jacobsen, C.S., Cox, E.E., Chambon, J.C., Bjerg, P.L., 2010. Field evaluation of biological enhanced reductive dechlorination of chloroethenes in clayey till. Environmental Science & Technology 44 (13), 5134e5141. Wiegel, J., 1990. Temperature spans for growth e hypothesis and discussion. FEMS Microbiology Reviews 75 (2e3), 155e169. Wolin, E.A., Wolin, M.J., Wolfe, R.S., 1963. Formation of methane by bacterial extracts. Journal of Biological Chemistry 238 (8), 2882e2886. Young, F.W., 1996. ViSta: The Visual Statistics System. Thurstone Psychometric Laboratory Research Memorandum 94-1(c). University of North Carolina, Chapel Hill, North Carolina.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 3 7 e6 6 4 9
Available online at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Recycling algae to improve species control and harvest efficiency from a high rate algal pond J.B.K. Park a,*, R.J. Craggs a, A.N. Shilton b a b
National Institute of Water and Atmospheric Research Ltd (NIWA), Aquatic Pollution Group, P.O. Box 11-115, Hamilton, New Zealand Centre for Environmental Technology and Engineering, Massey University, Private Bag 11 222, Palmerston North, New Zealand
article info
abstract
Article history:
This paper investigates the influence of recycling gravity harvested algae on species
Received 2 February 2011
dominance and harvest efficiency in wastewater treatment High Rate Algal Ponds (HRAP).
Received in revised form
Two identical pilot-scale HRAPs were operated over one year either with (HRAPr) or without
14 July 2011
(HRAPc) harvested algal biomass recycling. Algae were harvested from the HRAP effluent in
Accepted 21 September 2011
algal settling cones (ASCs) and harvest efficiency was compared to settlability in Imhoff
Available online 29 September 2011
cones five times a week. A microscopic image analysis technique was developed to determine relative algal dominance based on biovolume and was conducted once a month.
Keywords:
Recycling of harvested algal biomass back to the HRAPr maintained the dominance of
Algae
a single readily settleable algal species (Pediastrum sp.) at >90% over one year (compared to
Biofuels
the control with only 53%). Increased dominance of Pediastrum sp. greatly improved the
High rate algal ponds
efficiency of algal harvest (annual average of >85% harvest for the HRAPr compared with
Wastewater treatment
w60% for the control). Imhoff cone experiments demonstrated that algal settleability was
Algal harvest
influenced by both the dominance of Pediastrum sp. and the species composition of
Algal species control
remaining algae. Algal biomass recycling increased the average size of Pediastrum sp.
Algal recycling
colonies by 13e30% by increasing mean cell residence time. These results indicate that recycling gravity harvested algae could be a simple and effective operational strategy to maintain the dominance of readily settleable algal species, and enhance algal harvest by gravity sedimentation. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
High rate algal ponds (HRAPs) provide cost-effective and efficient wastewater treatment with minimal energy consumption and have considerable potential to upgrade oxidation ponds (Craggs et al., 2003; Garcı´a et al., 2006; Heubeck et al., 2007; Park and Craggs, 2010). Furthermore, the algal biomass produced and harvested from these wastewater treatment systems could be converted through various pathways to biofuels, for example anaerobic digestion to biogas, transesterification of lipids to biodiesel, fermentation of
carbohydrate to bioethanol and high temperature conversion to bio-crude oil (Sukias and Craggs, 2011; Vasudevan and Fu, 2010; Craggs et al., 2011). Wastewater treatment and algal biofuel production both require rapid and cost-effective harvest of algal biomass from HRAP effluent, therefore, methods to improve algal harvest efficiency would be of great benefit (Benemann, 2003; Chen and Yeh, 2005; van Harmelen and Oonk, 2006; Brennan and Owende, 2010). However, algal cells are very difficult to remove due to their small size (<20 mm), similar density to water (1.08e1.13 g/mL) (Lavoie and de la Noue, 1986) and strong negative surface charge,
* Corresponding author. Tel.: þ64 7 856 1777; fax: þ64 7 56 0151. E-mail addresses:
[email protected] (J.B.K. Park),
[email protected] (R.J. Craggs),
[email protected] (A.N. Shilton). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.09.042
6638
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 3 7 e6 6 4 9
particularly during exponential growth (Moraine et al., 1979; Chen and Yeh, 2005). Gravity sedimentation is the most common and costeffective method of algal biomass removal in wastewater treatment because of the large volumes of wastewater treated and the low value of the algal biomass generated (Nurdogan and Oswald, 1996). However, the algal settling ponds which are typically used have relatively long retention times (1e2 d) and only remove 50e80% of the biomass (Nurdogan and Oswald, 1996; Brennan and Owende, 2010; Park and Craggs, 2010; Park et al., 2011). While various harvesting options including chemical and mechanical methods have been extensively investigated (Shen et al., 2009; Tampier, 2009; Brennan and Owende, 2010; Mata et al., 2010), most technologies greatly increase operational costs for algal production (Benemann, 2008a; Craggs et al., 2011). For example, chemical flocculation can be reliably used to remove small algae (<5 mm) from pond effluent by forming large (1e5 mm) sized algal flocs (Sharma et al., 2006). However, the process is highly sensitive to pH and the high flocculent dose required produces large amounts of sludge. Mechanical centrifugation could be used for the removal of algal biomass, but the high energy requirement of the system makes it only economically viable for secondary thickening of primary harvested algae (1e2% solids) up to 20e30% solids (Benemann, 2008b; Tampier, 2009). Algal species commonly found in wastewater treatment HRAPs include: Desmodesmus sp., Micractinium sp., Actinastrum sp., Pediastrum sp., Dictyosphaerium sp. and Coelastrum sp. These algae often form large settleable colonies (diameter: 50e200 mm), which enable cost-effective and simple algal biomass removal by gravity sedimentation (Lavoie and de la Noue, 1986; Garcı´a et al., 2000; Benemann, 2008b; Craggs et al., 2011; Park et al., 2011). Park and Craggs (2010) reported that algal colonies and bacterial biomass in wastewater treatment HRAP that have CO2 addition will aggregate to form large algal bioflocs (diameter: >500 mm), which settle rapidly in simple gravity settling cones with a retention time of 12 h or less. Therefore, operating wastewater treatment HRAP to promote both the growth of particular settleable algal species (particularly colonial species) and the aggregation of algalebacterial biomass could greatly enhance the efficiency of algal harvest from HRAP effluent. Recycling algal biomass harvested by gravity settling has previously been shown to increase the dominance of readily settleable algae in small-scale laboratory cultures (Benemann et al., 1977; Weissman and Benemann, 1979; Tillett, 1988). Another option to select for beneficial algal species is by adjustment of the HRAP hydraulic retention time to promote species based on their specific growth rate (Weissman and Benemann, 1979). While there are several examples of successful species control (e.g. Spirulina and Dunaliella) in outdoor commercial algal production HRAP, these algae grow under extreme conditions (e.g. high pH and salinity) that reduce the potential for contamination (Weissman and Benemann, 1979; Sheehan et al., 1998; Benemann, 2003; Benemann, 2008b). The feasibility of algal species control in wastewater treatment HRAP and the mechanisms involved are still poorly understood. This paper investigates the influence of recycling selectively harvested algae on species dominance and harvest
efficiency in a pilot-scale wastewater treatment HRAP over one year. Relative algal dominance was determined from algal cell or colony biovolume using microscopic image analysis and correlated with gravity harvest and settling efficiency. Several mechanisms that could contribute to improved algal harvest efficiency from wastewater treatment HRAP with biomass recycling are proposed.
2.
Materials and methods
2.1.
Experimental pilot-scale HRAP system
Experiments were conducted using two identical pilot-scale single-loop raceway HRAPs treating domestic wastewater at the Ruakura Research Centre, Hamilton, New Zealand (37 470 S, 175 190 E). Each HRAP had a surface area of 31.8 m2, a depth of 0.3 m and a total volume of 8 m3. The pond water was circulated around each raceway at a mean velocity of w0.15 m/s by a paddlewheel. A schematic diagram of a pilotscale HRAP is shown in Fig. 1. Further details of the HRAP construction and operation were previously described in Park and Craggs (2010).
2.2.
HRAP operation
The HRAPs received primary settled sewage (0.5e1 m3/d), which was diluted 1:1 with tap water (simulating recycling of treated effluent after complete algal biomass and nutrient removal), which gave hydraulic retention times (HRTs) of 8 and 4 d in winter and summer respectively. During the NZ spring (Sept 7eNov 22, 2009) and autumn (Mar 17eMay 25, 2010), the HRT of the HRAPs was maintained at 6 d with a total inflow of 1.3 m3/d (primary sewage diluted 1:1 with tap water). The primary sewage and tap water were temporarily stored in separate 1 m3 water tanks and were pumped in equal amounts into the HRAPs each hour using submersible pumps controlled by an electronic timer which was recalibrated at least twice a month. The maximum pH of the HRAPs was maintained below 8 through pH controlled CO2 addition to avoid free ammonia inhibition and to augment daytime carbon availability. The CO2 addition system consisted of pure CO2 (compressed in a gas cylinder), a gas regulator and flow meter (0e12 L/min range), a solenoid valve, and two gas diffusers placed on the pond bottom in turbulent zones (one just before the paddlewheel and the other before the downstream pond corner). Pond water pH was measured every 5 s with a pH probe and when the pH exceeded the pH 8 set point, the controller opened the solenoid valve and added CO2 into the ponds (2 L/ min) through the gas diffusers. When the pond water pH was reduced to 7.8 the controller closed the solenoid valve halting CO2 addition. The pH probes were calibrated 1e2 times a week with standard pH solutions (pH 7 and 10). More details were previously reported in Park and Craggs (2010). Pediastrum sp. was inoculated into both pilot-scale HRAPs using 500 L of pond water taken from three wastewater treatment mini-HRAPs (volume: 0.8 m3) that were operated next to the pilot-scale HRAPs. Pediastrum sp. naturally established in the mini-HRAPs in January 2009 and was dominant
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 3 7 e6 6 4 9
6639
Fig. 1 e Schematic diagram of the pilot-scale high rate algal pond (HRAP) followed by two algal settling cones (ASCs) in series.
on March 13th 2009 when the inoculum was added. The pilotscale HRAPs were completely drained and any solids which had accumulated on the pond bottom were removed. Once the inoculum had been added, each pilot-scale HRAP, was then filled with primary settled raw sewage over 8 d (flow rate: 1 m3/d). Both pilot-scale HRAPs were operated without algal biomass recycling until July 1st 2009.
2.2.1. Algal biomass removal and recycling harvested algal biomass Effluent from the HRAPs was taken from the pond bottom (upstream of the paddlewheel) and flowed by gravity into the first algal settling cone (ASC1, 250 L) at mid-depth. The effluent from the top of ASC1 then flowed into the second ASC (ASC2, 250 L) at mid-depth. The hydraulic retention time of each ASC varied seasonally (3 h in summer and 6 h in winter) depending on the hydraulic retention time in the HRAPs (4 d and 8 d respectively). Algal biomass collected at the bottom of each ASC and was removed each day using a peristaltic pump (Masterflex, ColeParmer, HV-07523-60). A 1 L of the volume of the readily settleable algal biomass that collected every 24 h in the ASC1 following HRAPr (shown in Fig. 1) was recycled back to the HRAPr each day. The HRAPr algal recycling rate was determined by multiplying the volume of algal biomass recycled per day (L/d) with the harvested algal biomass solids concentration (g/L) and then dividing by the total mass of algae biomass harvested from the pilot-scale HRAP on that day (Table 1). The second HRAP was operated without algal biomass recycling as a control (HRAPc) with all other operational parameters the same as HRAPr. The operational parameters of the two pilot-scale wastewater treatment HRAPs, one operated with harvested biomass recycling (HRAPr) and the other without (HRAPc) are also summarized in Table 1. Parameters include the design and weather compensated (actual) daily flow rate and hydraulic retention time (HRT), mean cell residence time (MCRT) and influent nutrient (NHþ 4 -N and PO4 -P) concentrations.
2.3. Measurement of algal harvest efficiency and Imhoff cone settling efficiency Samples of HRAP water and ASC effluents were taken at least five times a week for the measurement of total suspended solids (TSSs) according to standard methods (APHA, 2005). Large inorganic particles and invertebrate grazers (e.g. Moina)
were strained (1 mm mesh) from the pond water samples before analysis of TSS. Algal harvest efficiency was determined as the percentage of the algal biomass in the HRAP water (measured as TSS) that was removed in each of the ASCs. Algal settling efficiency was measured five times a week using 1 L Imhoff cones after 10, 30, and 60 min and 24 h under laboratory conditions. Water samples (50 mL) were taken using a syringe from the mid-depth of the Imhoff cone (w450 mL depth) and were used to measure TSS, which were then compared with the initial TSS to determine algal settling efficiency.
2.4.
Microscopic image analysis
2.4.1.
Processing of samples and equipment used
A Utermo¨hl chamber (25 mm diameter) was used to count pond water algal populations and for microscopic image analysis. A 1 mL sample of thoroughly mixed pond water was pipetted into the Utermo¨hl chamber, evenly distributed to cover the surface of the chamber and then settled for 30e60 min (depending on algal settling efficiency). The Utermo¨hl chamber was then examined using an inverted light microscope with 160 magnification (Leica model) equipped with a Leica microscopic camera (DFS 420c). If the cell/colony density in the HRAP water sample was too high (>300 mg TSS/ L) to allow adjacent algal cells or colonies to be distinguished, a 1 mL sample of diluted HRAP water was used.
2.4.2.
Taking microscopic images
Safi (2009) found that taking microscopic images of the whole settling chamber is neither necessary nor feasible, particularly when algal cell/colony numbers are high. If the algae have settled evenly across the base of the settling chamber (checked by scanning the whole settling chamber at low magnification), microscopic images can be taken of randomly selected fields of view (FOV). If the algae have not settled evenly, images may be taken of equally spaced FOV along transects that run perpendicular to any observed settling gradient (Safi, 2009). The number of microscopic images required for the measurement of cell/colony dimensions and counts varied with algal population density but were typically 10.
2.4.3.
Identification of algal species
The most abundant algal species in the microscopic images of the HRAP water were identified using an identification guide (Brook, 2002). Dominant invertebrate grazers (e.g. Moina) were
6640
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 3 7 e6 6 4 9
Table 1 e Operational parameters of wastewater treatment pilot-scale HRAP with and without recycling gravity harvested algae in terms of design/weather compensated inflow rate, influent nutrient concentrations (NH44D-N and PO43L-P), hydraulic retention time (HRT), algal mean cell residence time (MCRT) and algal recycling rate to the HRAPr. Parameters
Winter
Spring
Summer
Autumn
Winter
67 (July 1e Sept 6, 09) 1.0 1.1 0.2 30.3 4.0 5.5 1.1 8.0 7.7 1.2 8.9 1.4 7.7 1.2 105 32
76 (Sept 7e Nov 22, 09) 1.3 1.4 0.2 22.5 7.5 4.0 1.0 6.0 5.9 0.8 6.5 0.9 5.9 0.8 66 25
113 (Nov 23, 09e Mar 16, 10) 2.0 2.0 0.2 20.6 9.2 4.0 0.7 4.0 4.1 0.4 4.6 0.4 4.1 0.4 76 22
70 (Mar 17e May 25, 10) 1.3 1.3 0.2 39.5 6.5 5.9 1.0 6.0 6.1 0.5 7.2 0.6 6.1 0.5 170 89
34 (May 26e Jun 30, 10) 1.0 1.2 0.3 34.3 7.3 4.8 1.2 8.0 7.0 1.5 10.4 2.2 7.0 1.5 212 8
Design Actual
6 5.8 0.8 35 5.1
4.5 4.4 0.6 38 6.2
3 3.1 0.3 33 5.8
4.5 4.6 0.4 32 5.1
6 5.2 1.1 30 9.0
Design Actual
12 11.61.7
9 8.91.1
6 6.20.5
9 9.20.8
12 10.22.2
Experimental duration (d) Experimental period Inflow to HRAPs (m3/d) Influent nutrient concentrations (mg/L) HRAP HRT (d) MCRT (d)b Daily algal recycling rate to HRAPr (g recycled/kg production)c ASC1 HRT (h) Harvested biomass conc. (TS, g/L) ASC2 HRT (h)d
a b c d
Design Actuala NHþ 4 -N PO43-P Design Actuala HRAPr HRAPc
Weather compensated (daily precipitation and evaporation) daily inflow and hydraulic retention time (HRT). Calculated using Eq. (1) below. Volume of algal biomass recycled (L/d) harvested algal biomass concentration (g/L)/the total mass of algae biomass harvested (kg). Combined HRT of ASC1 and ASC2.
MCRT ¼
V1 X1 Qc X1 Qre Xh
ð1Þ
where MCRT ¼ algal mean cell residence time (d), V1: HRAP volume (m3), X1: HRAP algal biomass concentration (g/m3), Qc: compensated HRAP effluent flow rate (m3/d), Qre: algal biomass recycled per day (L/d), Xh: harvested algal biomass concentration (g/L).
also identified and counted; and the occurrence of fungal hyphae in the algal/bacterial floccs was confirmed using the Calcofluor White protocol suggested by Kagami et al. (2007) and Rasconi et al. (2009).
images depending on the algal biomass concentration. Therefore counts for each pond water sample had an accuracy of 10% with 95% confidence limits (Lund et al., 1958; Rott et al., 2007).
2.4.6. 2.4.4. Measurement of algal cell/colony dimensions and algal counts Only viable algal cells/colonies were included (clear or collapsed cells and cell fragments were ignored). Due to the variable and sometimes high number of cells in the colonies of the algae that often dominate HRAP (e.g. Pediastrum sp. can have 8, 16, 32, or 64 cells/colony and Desmodesmus sp. can have 2, 4, or 8 cells/colony), it was not practical to count every cell in a colony. Therefore, the number of colonies of each species was counted and the dimensions (length and width) of each colony were measured to determine the algal biovolume using microscopic image analysis software (Leica Application Suite, LAS version 3.1.0).
2.4.5. Total cell/colony number counts and measurement uncertainty Lund et al. (1958) recommended that for a FOV count, a minimum of 100 cells/colonies should be enumerated to ensure that the count is representative of the sample (20% accuracy with 95% confidence limits). In this study, 200e500 algal cells or colonies were counted from w10 microscopic
Calculation of algal biovolume
Algal biovolume is a more accurate measure of relative algal dominance (%) than cell counts because not all algal cells are the same size (Lyakh, 2007; Rott et al., 2007; Vadrucci et al., 2007). Hillebrand et al. (1999) and Vadrucci et al. (2007) developed geometrical equations to calculate the biovolume of algal species of different shapes from microscopically measured linear dimensions. The equations for the five most dominant algae in the HRAPs (Pediastrum sp., Desmodesmus sp., Micractinium sp., Dictyosphaerium sp. and unicellular algae (including Chlorella sp., flagellates and Thalassiosira sp.)) are shown in Table 2 along with the number of cells per colony. Microscopic image analysis was conducted each month and the data used to calculate algal biovolume, which was in turn used to determine relative algal dominance in the HRAP.
2.5.
Water quality monitoring
Weekly samples of HRAP influent (primary settled raw sewage) and effluent were taken and then analyzed using
Table 2 e Calculations of algal cell/colony biovolume using geometric measurement. Dominant Algae
Pediastrum sp.
Desmodesmus sp.
Micractinium sp.
Dictyosphaerium sp.
Unicellular algae
Photo
Disc/star-shaped, flat and single-layered
Flat, straight or slightly curved
Number of cells/colony
4/8/16/32/64-celled
2/4/8-celled
Calculation of single cell biovolume
Cuboidal, tetrahedral or polyhedral (spherical cells) >4-celled distinguishing into single cells
Spherical (Chlorella sp.), cube (Thalassiosira sp.)
>4-celled distinguishing into single cells
Single cell
p V ¼ $L3 6 (Colonial sphere) L: length (mm)
p V ¼ $d3 (Single 6 sphere) V ¼ d3 (Cube) L: length (mm)
6641
p p L2 p V ¼ $L$W$D V¼ $ $W 4 V ¼ $L3 4 6 4 6 (Elliptic disc) (Prolate spheroid) (Colonial sphere) L: length (mm) W: width (mm) D: depth (mm) L: length (mm) W: width (mm) L: length (mm)
Hollow, spherical
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 3 7 e6 6 4 9
Shape of the cell/colony
6642
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 3 7 e6 6 4 9
standard methods (APHA, 2005) for the following parameters: dissolved reactive phosphorus ammoniacal-N (NHþ 4 -N), (DRP). During the period when a unicellular diatom (Thalassiosira sp.) occurred in the ponds (6the30th March 2010) reactive Silica (as SiO2) was also measured (APHA, 2005) at weekly intervals.
3.
Results and discussion
3.1.
Algal species and algal biovolume
The types of algae found in the HRAPs over the one year experimental period are summarized in Table 3. These included 13 genera of green algae which are commonly found in eutrophic waters and four colonial algae including: two species of Pediastrum sp., Desmodesmus sp., Micractinium pusillum, and Dictyosphaerium sp., which typically dominate wastewater treatment HRAPs around the world (Benemann et al., 1978, 1983; Benemann, 1986; Garcı´a et al., 2000; Park and Craggs, 2010; Craggs et al., 2011). The two species of Pediastrum (Pediastrum boryanum and Pediastrum duplex) were easily distinguished since P. duplex has intercellular spaces and P. boryanum does not. The cell or colony biovolume of the five most abundant algae in the HRAPs (4 colonial species and unicellular algae including the diatom, Thalassiosira sp.) were calculated by image analysis of microscopic photographs of pond water samples. A positive correlation (r ¼ 0.881) was found between HRAP algal biovolume and biomass concentration (measured as TSS) (Fig. 2), indicating that biovolume is a particularly useful measure of algal biomass for wastewater treatment HRAPs. Particularly as the HRAPs selected for colonial algae (Micractinium sp., Dictyosphaerium sp., Desmodesmus sp. and Pediastrum sp.) with colonies of different shapes and varying numbers of cells. For example, the biovolume of a Pediastrum sp. colony can vary with both the number of cells (8, 16, 32 or 64 cells) and the size of the
-0.6
µ
Fig. 2 e The relationship between the algal biovolume and algal biomass concentration in the HRAPs (r [ 0.881).
cells within the colony depending on life-cycle stage and culture conditions.
3.2. Influence of harvested algal biomass recycling on relative algal dominance The relative dominance of the five most abundant algae in the HRAPs was determined at monthly intervals based on biovolume (Fig. 3). The colonial algae Pediastrum sp., Micractinium sp., and Dictyosphaerium sp. were each dominant in the control pond (HRAPc) for periods of two months or more (Fig. 3a). Changes in dominance between these algae occurred within a few weeks. For example, Pediastrum sp. was replaced by
Table 3 e Algae found in the wastewater treatment HRAPs over one year experimental period. Phyllum
Green algae (Chlorophyta)
Dominance
Dominant algae
Occasionally found algae
Diatoms (Bacillariophyceae)
e
HRAPc
HRAPr Genus
Species
Genus
Species
Pediastrum Pediastrum Desmodesmus Micractinium Dictyosphaerium Gonium Ankistrodesmus Monoraphidium Pandorina Radiococcus Kirchneriella Actinastrum Coelastrum Thalassiosira
boryanum duplex sp. pusillum sp. sp. falcatus sp. sp. sp. sp. hantzschii sp. sp.
Pediastrum Pediastrum Desmodesmus Micractinium Dictyosphaerium Gonium Ankistrodesmus Monoraphidium Dictyosphaerium Kirchneriella Actinastrum Coelastrum Chlamydomonas Thalassiosira
boryanum duplex sp. pusillum sp. sp. falcatus sp. sp. sp. hantzschii sp. sp. sp.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 3 7 e6 6 4 9
6643
Fig. 3 e Algal dominance based on calculated biovolume in the pilot-scale wastewater treatment HRAPs over a one year experimental period (from July 2009 to June 2010), a: control HRAPc without algal biomass recycling and; b: HRAPr with algal biomass recycling.
Micractinium sp. in October 2009, which was replaced by Pediastrum sp. in December 2009, which was replaced by the unicellular diatom Thalassiosira sp. in March 2010, which was then replaced by Dictyosphaerium sp. in April 2010. These shifts in algal dominance were probably caused by changes in environmental conditions (notably seasonal variation of solar radiation and pond water temperature which are known to affect species selection, succession and co-existence) and HRAP operational parameters such as hydraulic retention time (Benemann et al., 1977; Harris, 1978; Sommers, 1984; Oswald, 1988). Garcı´a et al. (2000) also reported similar changes in relation to environmental parameters by the dominant algae (including Dictyosphaerium sp., Chlorella sp., Micractinium sp., and Desmodesmus sp.) of a small-scale (0.5 m3) wastewater treatment HRAP without CO2 addition in Spain over a one year experimental period. As can be seen in Fig. 3b, recycling a portion of selectively harvested algal biomass increased the dominance of Pediastrum sp. (76e99% dominance) in the HRAPr compared to the control HRAPc (0e98% dominance) during the one year experimental period. Other colonial algae including Micractinium sp., Desmodesmus sp., Dictyosphaerium sp. and the unicellular diatom Thalassiosira sp. were temporarily present and co-occurred with Pediastrum sp. in the HRAPr but at much lower populations than in HRAPc. Maintaining dominance of a single algal species (Pediastrum sp., >90% dominance) over similarly sized algal species in wastewater treatment HRAP for over one year has not been previously reported in the literature. This suggests that recycling of settled algal biomass could provide a simple method to promote the dominance of readily settleable algal species such as Pediastrum sp. in HRAP.
3.3. Settling characteristics of the dominant algal species Recycling harvested algal biomass preferentially selected is algae that settle rapidly. Since all algae in the HRAP effluent are exposed to the same settling conditions (e.g. water viscosity (h) and temperature) in the algae settling cones, differences in settling efficiency between species, therefore, depend on their physiological state, and morphology (Smith, 1982; Alldredge and Gotschalk, 1989; Chen and Yeh, 2005; Choi et al., 2006). The morphology (i.e. size and irregularity) of the algal cell or colony influences settling velocity according to the frictional drag force exerted as it falls through the fluid under the pull of gravity (Smith, 1982; Padisa´k et al., 2003; Chen and Yeh, 2005; Choi et al., 2006). Therefore, cells or colonies with small surface areas settle faster than cells or colonies of the same density but with larger surface areas (Padisa´k et al., 2003; Choi et al., 2006). Average cell and/or colony biovolume, calculated nominal radius (rs) and approximate theoretical settling velocity (Vtheo) of the five dominant algae were determined using the one year experimental data and are summarized in Table 4. The nominal radius of Pediastrum sp. colonies (the most dominant algae in both HRAPs) ranged from 5.1 to 13.5 mm and depended on colony age and the number of cells per colony (8, 16, 32, or 64). The nominal radius of Pediastrum sp. colonies were 1.4e3.8 times larger than that of the unicellular algae (3.6 mm) and nearly 2 times larger than those of the other colonial algal species present (4.8e5.3 mm for 2e4 celled Desmodesmus sp., 5.5 mm for Micractinium sp., and 7.9 mm for Dictyosphaerium sp.). The theoretical settling velocity, Vtheo, of each alga was
6644
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 3 7 e6 6 4 9
Table 4 e Average algal cell or colony biovolume of the four dominant algal species/type over the one year experimental period, the number of cells, relative abundance with genera (%), nominal radius (ø, mm) of a sphere of equivalent biovolume to the algal cell or colony, and the approximate relative settling velocity (r2). Algal species
Pediastrum sp.
Desmodesmus sp. Micractinium sp. Dictyosphaerium sp. Unicellular algae
Relative Vtheob Nominal Cell numbers Relative Total cell Average biovolume Total biovolume 3 (mm s.d.) radius radius within abundance counts of cell/colony (rr)a (rs, mm s.d.) a colony within (counts/mL) (mm3 s.d.) genera (%) 8-celled 16-celled 32-celled 64-celled 2-celled 4-celled e e e
7.2 40.2 43.2 9.3 20.4 79.6 e e e
9.59Eþ06 3.37Eþ04 1.78Eþ04 1.10Eþ03 8.20Eþ02 2.87Eþ03 7.96Eþ03 5.45Eþ03 4.04Eþ04
561 308 1834 670 4115 2009 10,219 3217 458 265 617 292 679 446 2073 118 193 82
9.59Eþ06 5.33Eþ07 5.73Eþ07 1.23Eþ07 4.91Eþ05 1.92Eþ06 2.13Eþ06 1.20Eþ06 7.53Eþ06
5.1 4.2 7.6 5.4 9.9 7.8 13.5 9.2 4.8 4.0 5.3 4.1 5.5 4.7 7.9 3.0 3.6 2.7
1.4 2.1 2.8 3.8 1.3 1.5 1.5 2.2 1.0
2.0 4.5 7.7 14.1 1.8 2.2 2.3 4.9 1.0
a Radius relative to a unicellular algae. b Vtheo (theoretical relative settling velocity) is proportional to r2r according to Stoke’s law (Vs ¼ ð2=9Þðr2s gðrp rf Þ=hFr Þ), assuming all other parameters are same. where, g gravitational acceleration, (rp rf) excess density between particles and fluid, h viscosity of the medium, rs nominal radius of the sphere of equivalent biovolume to the algae, F form resistance (the effects of algal shape upon settling).
calculated from the biovolume, assuming form resistance (F) between algae was negligible (Table 4). The larger biovolume and nominal radius of Pediastrum sp. colonies suggest that they could have better settling characteristics (calculated theoretical settling velocity, Vtheo was 2e14 times greater than that of any other algae present in the HRAPs). Therefore, when the HRAP algal biomass was dominated by Pediastrum sp. it was easily harvested by gravity sedimentation.
3.4. Algal biomass harvest depending on dominant algal species The algal biomass concentrations (as TSS) in the HRAP, ASC1 and ASC2 effluents and the algal biomass harvest efficiency after the ASC2 are shown in Fig. 4. Algal biomass concentrations in the HRAP effluent varied from w50 to 420 g/m3 during the one year experimental period depending on seasonal algal
Fig. 4 e Algal biomass concentrations in the effluents from two pilot-scale high rate algal ponds and primary and secondary algal settling cones (ASCs); and calculated total removal efficiency measured over one year: a: HRAPc operated without harvested biomass recycling; b: HRAPr operated with harvested algal biomass recycling.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 3 7 e6 6 4 9
growth and invertebrate grazing. However, algal biomass concentrations of the ASC effluents and algal harvest efficiency were highly dependant on the algae that was dominant in the HRAP at that time. Increased dominance (annual average of >90%, Fig. 3) of Pediastrum sp. in HRAPr with algal biomass recycling greatly improved algal biomass harvest efficiency by gravity sedimentation (>75% and >85% algal harvest efficiency after ASC1 and ASC2 respectively). Low final effluent TSS concentrations (<25 g/m3) were consistently achieved for HRAPr during the one year experiment (Fig. 4b). In contrast, the control HRAPc had inconsistent and rather poor algal biomass harvest efficiency (48% and 64% after ASC1 and ASC2 respectively), mainly because less settleable algae were dominant in this pond (Fig. 3a). For example, increased dominance of Pediastrum sp. in the control HRAPc (80e98% dominance over 3 months from December to February 2010) greatly enhanced the algal biomass harvest efficiency to 90% (which was nearly comparable to that in the HRAPr which had >90% Pediastrum dominance). However, by June 2010 the Pediastrum dominance in the HRAPc had declined to less than 20%, as Pediastrum sp. was replaced by the poorly-setteable alga, Thalassiosira sp. (w85% dominance) by March 2010, and then Dictyosphaerium sp. (dominance increased from 40% to 85% by June 2010). These results demonstrate that maintaining readily settleable algae such as Pediastrum sp. as the dominant species could provide a way of promoting efficient algal harvest by gravity sedimentation. The temporary establishment of the unicellular diatom (Thalassiosira sp.) in both HRAPs (>82% dominance in the HRAPc and 32% dominance in HRAPr; Fig. 3) during a four week period from 6th to 30th March 2010 was an interesting observation, since it is very unusual for diatoms to grow at high concentration in wastewater treatment HRAP and has not been previously observed at our wastewater treatment HRAP research facility, or reported in the literature (Benemann et al., 1978, 1983; Benemann, 1986; Garcı´a et al., 2000; Craggs et al., 2011; Park and Craggs, 2010). The occurrence of Thalassiosira sp. may be explained by a temporary increase in the dissolved silica concentration (SiO2) of the HRAP influent wastewater, since silica is an essential element for diatom growth and the typical low concentration in wastewater limits the growth of diatoms. Analysis of the influent wastewater to the HRAPs for SiO2 during the period of Thalassiosira sp. occurrence showed a high dissolved silica level (w30 mg/L as SiO2), which declined to <5 mg/L by May 2010 when the diatom was no longer present in the HRAP. Thalassiosira sp. has very low settling efficiency and consequently its occurrence greatly reduced algal harvest efficiency of the ASCs for both HRAPs (HRAPc: 48% harvest efficiency; HRAPr: 64% harvest efficiency). Micractinium sp. and Dictyosphaerium sp. were dominant in the control HRAPc for periods of two months (Micractinium sp.: w70% dominance from October to November 2009; Dictyosphaerium sp.: 70e85% dominance from May to June 2010; Fig. 3a). These algae can form colonies of typically >100 spherical single cells with a nominal radius of 5.5e7.9 mm. When these two species were dominant in the ponds algal biomass harvest efficiency was particularly poor (63e77% for Micractinium sp. and only 16e38% for Dictyosphaerium sp, Fig. 4a). In contrast, during the period from December to
6645
February 2010 when Pediastrum sp. was dominant very high algal harvest efficiency (>85% after the ASC2 with 6 h HRT) was achieved. The lower settling efficiency of Micractinium sp. and Dictyosphaerium sp. compared with Pediastrum sp. might be due to lower density and higher drag resulting from the dispersed structure of the colonies (i.e. large spaces between groups of cells) compared with Pediastrum sp. colonies in which the cells are tightly packed together. Over the one year experimental period, nutrient removal 3 efficiency (86e98% NHþ 4 -N and 50e75% PO4 -P) of HRAPr (in which Pediastrum sp. was maintained at >90% dominance) 3 was similar to that (90e96% NHþ 4 -N and 52e68% PO4 -P) of HRAPc (which had a mixed population of algae). Further detailed wastewater treatment performance of the HRAP in terms of organic compounds (TSS/VSS and BOD5) and nutrient (N and P) removal and a nitrogen mass balance with CO2 addition have been described previously (Park and Craggs, 2010, 2011a, 2011b).
3.5. HRAP
Enhancing algal harvest in wastewater treatment
The Imhoff cone settling efficiency of algae in the HRAP effluent after 10, 30 and 60 min and 24 h was measured throughout the year and was found to be greatly improved when Pediastrum sp. was dominant (Fig. 5a). Average algal settling efficiencies of 75.3 13.7%, 86.0 9.1%, 93.6 2.8%, and 99.2 1.6% were achieved after 10, 30 and 60 min, and 24 h of settling respectively, when Pediastrum sp. was present at over 80% dominance in the HRAPs. However, as Pediastrum sp. dominance declined to less than 40%, algal settling efficiencies reduced to 19.8 8.6%, 25.6 10.2%, 35.2 10.1% and 76.0 10.9% for the respective settling periods. These algal settling efficiency test results confirmed that the dominance of Pediastrum sp. in the HRAP promoted algal settling efficiency particularly for settling periods of less than 1 h. Overall Imhoff cone settling efficiency was also influenced by settling characteristics of the algae that co-existed with Pediastrum sp. (Fig. 5a). For example, when Pediastrum sp. was present at 70% dominance and co-existed with other colonial algae such as Micractinium sp., and Desmodesmus sp. (shown in circle “i”) the 10 min settling efficiency was high (58%), however, when the co-existing algae were mainly poorly settleable unicellular algae such as Thalassiosira sp. (average cell size 3.6 2.7 mm) (shown in circle “ii”), the 10 min settling efficiency was only 38%. Analysis of the data on the population density of unicellular algae (counts/mL) including Thalassiosira sp. and algal settling efficiency during the one year experimental period indicates that algal settling efficiency at all settling times declined if the unicellular algal population was greater than 1 105 cells/mL (Fig. 5b). Therefore, controlling the population density of poorly settleable unicellular algae to below this level could be necessary to achieve efficient algal biomass harvest. The percentage solids (as % total solids) of the algal biomass collected at the bottom of an Imhoff cone after a 24 h settling period were also influenced by the settling characteristics of the dominant algal species and other biological factors (e.g. zooplankton grazing and fungal infection) in the HRAPs (Fig. 6). Pediastrum sp. dominant algal biomass (>90%
6646
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 3 7 e6 6 4 9
a
b
Fig. 5 e a. Relationship between: a. Pediastrum sp. dominance and algal biomass settling efficiency after 10, 30, and 60 min and 24 h settling in an Imhoff cone (Note: within circle i: 72% Pediastrum sp. D 16% Micractinium sp. D 12% Desmodesmus sp., and circle ii: 70% Pediastrum sp. D 30% Unicellular algae); and, b. unicellular algal counts (counts/mL) and algal biomass settling efficiency in an Imhoff cone.
dominance) was harvested as 2.5e3.0% solids after 24 h gravity sedimentation (December 2009 for HRAPc: Fig. 6a, NovembereDecember 2009 and MayeJune 2010 for HRAPr: Fig. 6b). However, when less settleable colonial (Micractinium sp., and Dictyosphaerium sp.) and poorly settleable unicellular algae (Thalassiosira sp.) were dominant in the control HRAPc, the % solids of harvested algae was only about 1.5e2.0% (Fig. 6a). Moreover, fungal infection in both HRAPs during the summer period of Pediastrum dominance (JanuaryeFebruary 2010) reduced the settled algal biomass solids concentrations from 2.5e3.0% to less than 2.0%, which may have been due to the lower density of algal/fungal floccs. Two species of Pediastrum (P. duplex and P. boryanum) were present in the HRAPs, of which, P. boryanum
a
(intercellular space absent) was the most prevalent species throughout the one year experimental period. In particular, during the three month summer period from December 2009 to February 2010, P. boryanum accounted for 80e98% of total algal biomass in both HRAPs. This period of P. boryanum prevalence coincided with very high algal harvest efficiency (80e95%) in the two ASC (HRT 6 h) (Fig. 4) of both HRAPs. Imhoff cone algal settling experiments in the laboratory conducted during the same period showed that algal biomass recycling in the HRAPr enhanced the algal settling efficiency by 8e15% (76.0 6.4%, 89.3 3.5%, and 97.2 1.9% removal after 10, 30, and 60 min respectively compared with 59.3 13.5%, 76.5 8.9%, and 88.2 9.9% for the control, P-value: <0.005, one-way ANOVA, Fig. 7). The
b
Fig. 6 e The relationship between the percentage solids of the algal biomass collected from the bottom of the Imhoff cone after 24 h and dominant algae in HRAPr (Fig. 5a) and the control HRAPc (Fig. 5b) (Note: fungal infection was observed in both HRAPs in summer from Jan to Feb 2010).
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 3 7 e6 6 4 9
6647
Fig. 7 e Imhoff cone algal settling efficiency after 10, 30, and 60 min and 24 h in laboratory measured from December 2009 to February 2010 when P. boryanum dominance was similar (>80%) in both HRAPs with and without harvested algal biomass recycling. (The data for each settling were compared to investigate the effect of algal biomass recycling using one-way ANOVA analysis.)
improved algal settling efficiency in the HRAPr may be explained by the presence of larger colonies of P. boryanum as recycling of harvested algae at an algal recycling rate of 66e212 g/kg/d (depending on season, Table 1) extended the mean cell residence time (MCRT) (i.e. the period for growth) in the pond by w0.5 d in summer and w3.4 d in winter.
Extending the MCRT increased the average size and biovolume of P. boryanum colonies in HRAPr by 13e30% and 50e80% respectively compared to those in the control HRAPc measured over the three month period when P. boryanum was prevalent in both HRAPs (December 2009 to February 2010) (Fig. 8).
a
b
Fig. 8 e Average size (a) and biovolume (b) of P. boryanum colonies with different numbers of cells (8e64) in HRAPr with biomass recycling and the control, HRAPc without biomass recycling during the period of P. boryanum dominance (>80%).
6648
4.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 3 7 e6 6 4 9
Conclusions
Recycling of harvested algal biomass to the HRAPr increased the dominance of the colonial alga, Pediastrum sp., which settles rapidly, so that >90% dominance was achieved for the majority of the one year study compared to the control HRAPc (53% dominance). Maintaining dominance of a single algal species over similarly sized algal species in wastewater treatment HRAP over one year has not been previously reported in the literature. Increased dominance of Pediastrum sp. greatly improved algal biomass harvest efficiency (90% compared to 60% in the control) from the HRAPr to consistently achieve low TSS concentrations (<25 g/m3) in the ASC effluent. Imhoff cone experiments measuring algal settling efficiency in the laboratory demonstrated that the dominance of Pediastrum sp. and the species composition of remaining algae both influenced algal biomass settling efficiency. Moreover, the settling characteristics of the dominant algal species in the HRAPs also influenced the % solids of 24 h settled algal biomass. Recycling of harvested algal biomass increased the average size and biovolume of Pediastrum sp. colonies in the HRAPr effluent by 13e30% and 50e80% respectively, possibly as a result of increasing the mean cell residence time. These results show that recycling of algal biomass is a simple and effective operational strategy to promote and maintain the dominance of readily settleable algae such as Pediastrum sp. to enhance algal biomass removal by gravity sedimentation.
references
Alldredge, A.L., Gotschalk, C.C., 1989. Direct observations of the mass flocculation of diatom blooms: characteristics, settling velocities and formation of diatom aggregates. Deep Sea Research Part A. Oceanographic Research Papers 36, 159e171. APHA, 2005. Standard Methods for the Examination of Water and Wastewater. American Public Health Association, Washington. Benemann, J.R., 1986. Microalgae Biotechnology: Products, Processes and Opportunities, vol. 1. OMEC International Inc. Benemann, J.R., 2003. Biofixation of CO2 and greenhouse gas abatement with michroalgae e Technology roadmap. Prepared for the U.S. Department of Energy: National Energy Technology Laboratory No: 7010000926. Benemann, J.R., 2008a. Open ponds and closed photobioreactors e comparative economics 5th Annual World Congress on Industrial Biotechnology & Bioprocessing, Chicago, April 30, 2008a. Benemann, J.R., 2008b. Opportunities and challenges in algae biofuels production, Algae World 2008, Singapore, November 17e18, 2008b, p. 15. Benemann, J.R., Goebel, R.P., Weissman, J.C., Augenstein, D.C., 1983. Production of liquid fuels and chemicals by microalgae. Proceedings of the March 1983 Principal Investigators Meeting, Aquatic Species Program Solar Energy Research Institute, Golden, Colorado, SERI/CP-231-1946, pp. 19e32. Benemann, J.R., Koopman, B.L., Weissman, J.C., Eisenberg, D.M., Oswald, W.J., 1977. Species control in large scale microalgae biomass production. Report to Univ. Calif. Berkeley SERL 77-5, SAN/740-77/1. Benemann, J.R., Koopman, B.L., Weissman, J.C., Eisenberg, D.M., Oswald, W.J., 1978. An integrated system for the conversion
of solar energy with sewage-grown microalgae. Report, Contract D(0e3)-34 U.S. Dept. of Energy, SAN-003-4-2, June 1978. Brennan, L., Owende, P., 2010. Biofuels from microalgae e a review of technologies for production, processing, and extractions of biofuels and co-products. Renewable and Sustainable Energy Reviews 14, 557e577. Brook, A.J., 2002. The Freshwater Algal Flora of The British Isles e An Identification Guide to Freshwater and Terrestrial Algae. Cambridge University Press, Cambridge. Chen, J.J., Yeh, H.H., 2005. The mechanisms of potassium permanganate on algae removal. Water Research 39, 4420e4428. Choi, S.K., Lee, J.Y., Kwon, D.Y., Cho, K.J., 2006. Settling characteristics of problem algae in the water treatment process. Water Science and Technology 53, 113e119. Craggs, R.J., Davies-Colley, R.J., Tanner, C.C., Sukias, J.P.S., 2003. Advanced ponds systems: performance with high rate ponds of different depths and areas. Water Science and Technology 48, 259e267. Craggs, R.J., Heubeck, S., Lundquist, T.J., Benemann, J.R., 2011. Algae biofuel from wastewater treatment high rate algal ponds. Water Science and Technology 63, 660e665. Garcı´a, J., Green, B.F., Lundquist, T., Mujeriego, R., Herna´ndezMarine´, M., Oswald, W.J., 2006. Long term diurnal variations in contaminant removal in high rate ponds treating urban wastewater. Bioresource Technology 97, 1709e1715. Garcı´a, J., Herna´ndez-Marine´, M., Mujeriego, R., 2000. Influence of phytoplankton composition on biomass removal from high-rate oxidation lagoons by means of sedimentation and spontaneous flocculation. Water Environment Research 72, 230e237. Harris, G.P., 1978. Photosynthesis, productivity and growth: the physiological ecology of phytoplankton. Ergebnisse der limnologie 10, 1e171. Heubeck, S., Craggs, R.J., Shilton, A., 2007. Influence of CO2 scrubbing from biogas on the treatment performance of a high rate algal pond. Water Science and Technology 55, 193. Hillebrand, H., Drselen, C.D., Kirschtel, D, Pollingher, U, Zohary, T, 1999. Biovolume calculation for pelagic and benthic microalgae. Journal of Phycology 35, 403e424. Kagami, M., de Bruin, A., Ibelings, B, Van Donk, E, 2007. Parasitic chytrids: their effects on phytoplankton communities and food-web dynamics. Hydrobiologia 578, 113e129. Lavoie, A., de la Noue, J., 1986. Harvesting of Scenedesmus obliquus in wastewaters: auto- or bioflocculation? Biotechnology and Bioengineering 30, 852e859. Lund, J.W.G., Kipling, C., Le Cren, E.D., 1958. The inverted microscope method of estimating algal numbers and the statistical basis of estimations by counting. Hydrobiologia 11, 143e170. Lyakh, A., 2007. A new method for accurate estimation of diatom biovolume and surface area. In: Kusber, W.H., Jahn, R. (Eds.), The 1st Central European Diatom Meeting Botanic Garden and Botanical Museum Berlin-Dahlem. Freie Universita¨t Berlin. Mata, T.M., Martins, A.A., Caetano, N.S., 2010. Microalgae for biodiesel production and other applications: a review. Renewable and Sustainable Energy Reviews 14, 217e232. Moraine, R., Shelef, G., Meydan, A., Levi, A., 1979. Algal single cell protein from wastewater treatment and renovation process. Biotechnology and Bioengineering 21, 1191e1207. Nurdogan, Y., Oswald, W.J., 1996. Tube settling rate of highrate pond algae. Water Science and Technology 33, 229e241. Oswald, W.J., 1988. Micro-algae and waste-water treatment. In: Borowitzka, M.A., Borowitzka, L.J. (Eds.), Microalgal Biotechnology. Cambridge University Press, New York, USA, pp. 305e328. Padisa´k, J., Soro´czki-Pinte´r, E´., Rezner, Z., 2003. Sinking properties of some phytoplankton shapes and the relation of form
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 3 7 e6 6 4 9
resistance to morphological diversity of plankton e an experimental study. Hydrobiologia 500, 243e257. Park, J.B.K., Craggs, R.J., 2010. Wastewater treatment and algal production in high rate algal ponds with carbon dioxide addition. Water Science and Technology 61, 633e639. Park, J.B.K., Craggs, R.J., 2011a. Algal production in wastewater treatment high rate algal ponds for potential biofuel use. Water Science and Technology 63, 2403e2410. Park, J.B.K., Craggs, R.J., 2011b. Nutrient removal and nitrogen balances in high rate algal ponds with carbon dioxide addition. Water Science and Technology 63, 1758e1764. Park, J.B.K., Craggs, R.J., Shilton, A.N., 2011. Wastewater treatment high rate algal ponds for biofuel production. Bioresource Technology 102, 35e42. Rasconi, S., Jobard, M., Jouve, L., Sime-Ngando, T., 2009. Use of calcofluor white for detection, identification, and quantification of phytoplanktonic fungal parasites. Applied and Environmental Microbiology 75, 2545e2553. Rott, E., Salmaso, N., Hoehn, E., 2007. Quality control of Utermo¨hl-based phytoplankton counting and biovolume estimatesdan easy task or a Gordian knot? Hydrobiologia 578, 141e146. Safi, K.A., 2009. Algal Services Methods: Standard Operating Procedure 2.0 Version 4.0., NIWA Algal Services Laboratory. NIWA Interal Precedure Manual p. 32. Sharma, B.R., Dhuldhoya, N.C., Merchant, U.C., 2006. Flocculantsdan ecofriendly approach. Journal of Polymers and the Environment 14, 196e202. Sheehan J., Dunahay T., Benemann J., Roessler P., 1998. A Look Back at the U.S. Department of Energy’s Aquatic Species Program- Biodiesel from Algae. National Renewable Energy Laboratory NREL/TP-580-24190. Shen, Y., Yuan, W., Pei, Z.J., Wu, Q., Mao, E., 2009. Microalgae mass production methods. Transactions of ASABE 52, 1275e1287.
6649
Smith, I.R., 1982. A simple theory of algal deposition. Freshwater Biology 12, 445e449. Sommers, U., 1984. The paradox of the plankton: fluctuation of phosphorus availability and natural diversity in flow-through cultures. Limnology and Oceanography 22, 633e656. Sukias, J.P.S., Craggs, R.J., 2011. Digestion of wastewater pond microalgae and inhibition from ammonium and alum. Water Science and Technology 63, 835e840. Tampier, M. Microalgae Technologies & Processes for Biofuels/ Bioenergy Production in British Columbia: Current Technology, Suitability & Barriers to Implementation. Prepared for The British Columbia Innovation Council. January 14, 2009. Tillett, D.M., 1988. Ph.D thesis: Lipid productivity and species competition in laboratory models of algae mass cultures. The School of Chemical Engineering. Georgia Institute of Technology, p. 316. Vadrucci, M.R., Cabrini, M., Basset, A., 2007. Biovolume determination of phytoplankton guilds in transitional water ecosystems of Mediterranean Ecoregion. Transitional Waters Bulletin 2, 83e102. van Harmelen, T., Oonk, H., 2006. Microalgae biofixation processes: applications and potential contributions to greenhouse gas mitigation options, TNO Built Environmental and Geosciences. Apeldoorn. Prepared for the International Network on Biofixation of CO2 and greenhouse gas abatement with Microalgae. Operated under the International Energy Agency Greenhouse Gas R&D Programme., The Netherlands (Order no. 36562). Vasudevan, P.T., Fu, B., 2010. Environmentally sustainable biofuels: advances in biodiesel research. Waste and Biomass Valorization, 1e17. Weissman, J.C., Benemann, J.R., 1979. Biomass recycling and species competition in continuous cultures. Biotechnology and Bioengineering 21, 627e648.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 5 0 e6 6 6 0
Available online at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Assessment of total uncertainty in cocaine and benzoylecgonine wastewater load measurements Christoph Mathieu a, Jo¨rg Rieckermann b, Jean-Daniel Berset c, Stefan Schu¨rch d, Rudolf Brenneisen a,* a
Dept. of Clinical Research, University of Berne, Berne, Switzerland Swiss Federal Institute of Aquatic Science and Technology (EAWAG), Du¨bendorf, Switzerland c Water and Soil Protection Laboratory (WSPL), Berne, Switzerland d Dept. of Chemistry and Biochemistry, University of Berne, Berne, Switzerland b
article info
abstract
Article history:
To check the effectiveness of campaigns preventing drug abuse or indicating local effects
Received 11 July 2011
of efforts against drug trafficking, it is beneficial to know consumed amounts of substances
Received in revised form
in a high spatial and temporal resolution. The analysis of drugs of abuse in wastewater
16 September 2011
(WW) has the potential to provide this information. In this study, the reliability of WW drug
Accepted 25 September 2011
consumption estimates is assessed and a novel method presented to calculate the total
Available online 19 October 2011
uncertainty in observed WW cocaine (COC) and benzoylecgonine (BE) loads. Specifically, uncertainties resulting from discharge measurements, chemical analysis and the applied
Keywords:
sampling scheme were addressed and three approaches presented. These consist of (i)
Sewage treatment plant
a generic model-based procedure to investigate the influence of the sampling scheme on
Wastewater
the uncertainty of observed or expected drug loads, (ii) a comparative analysis of two
Cocaine and benzoylecgonine loads
analytical methods (high performance liquid chromatographyetandem mass spectrometry
Analytical uncertainty
and gas chromatographyemass spectrometry), including an extended cross-validation by influent profiling over several days, and (iii) monitoring COC and BE concentrations in WW of the largest Swiss sewage treatment plants. In addition, the COC and BE loads observed in the sewage treatment plant of the city of Berne were used to back-calculate the COC consumption. The estimated mean daily consumed amount was 107 21 g of pure COC, corresponding to 321 g of street-grade COC. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
Drug abuse is a widespread problem of the modern community with significant socio-economic consequences, such as health treatment costs and higher incidence of criminality (EMCDDA, 2009). For drug policy makers it is therefore important to correctly identify trends, usage levels, hot spots, and prevalence of drug consumption to design prevention
campaigns or enforcement strategies, which is not trivial given current epidemiological data. Drug abuse is mostly estimated indirectly from population surveys, consumer interviews, medical records, crime statistics, drug production and seizure rates (EMCDDA, 2002). However, traditional methods are costly and the results are often obtained with considerable delay. Therefore, epidemiologists seek to explore novel data sources to improve addiction research or
* Corresponding author. Dep. Clinical Research, University of Berne, Murtenstr. 35, CH-3010 Berne, Switzerland. Tel.: þ41 31 632 8714; fax: þ41 31 632 3297. E-mail address:
[email protected] (R. Brenneisen). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.09.049
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 5 0 e6 6 6 0
drug early warning systems. Estimating drug consumption from analysis of drugs of abuse in wastewater (WW) has the potential for such improvement, because data could be collected in near-real time, non-intrusively from the majority of the population and without many other limitations of traditional surveys (EMCDDA, 2002). In 2001, Daughton (Daughton, 2001) hypothesized that WW can be regarded as a pooled urine sample of a large population and that mass flows of drugs of abuse in WW reflect the consumed amount. In 2004, for the first time a drug of abuse (3,4-methylenedioxymethamphetamine, MDMA, known as ecstasy) was detected in effluent water from a sewage treatment plant (STP) (Jones-Lepp et al., 2004). In 2005, Zuccato and coworkers (Zuccato et al., 2005) were the first suggesting a model to back-calculate usage figures from WW loads and coined the term “sewage epidemiology”. Consequently, this methodological approach has further been developed and implemented for other drugs of abuse (Castiglioni et al., 2008; Postigo et al., 2008; Van Nuijs, Castiglioni, et al., 2011; Van Nuijs et al., 2011). To further develop WW drug testing as a novel information source in addition to the classical epidemiological tools, Van Nuijs, Castiglioni, et al. (2011) formulated research needs to improve the validity of (i) measured “concentrations of drugs of abuse and/or metabolite(s) in influent wastewater” and (ii) “back-calculations from concentrations in wastewater to an amount of drugs of abuse (g/day)”, which range from errors in discharge measurements, adsorption to particulate matter (Metcalfe et al., 2010) and losses in the urban sewer system (Rutsch et al., 2008) to the variability of excretion rates (Cone et al., 2003) in a population of drug users. Interestingly, while errors in discharge measurements are often being considered in engineering analysis, there has been recent concern that the sampling scheme and frequency also introduce uncertainty in analytical studies. In a recent review, 87 peer-reviewed journal articles were analyzed regarding the fitness for purpose of the applied sampling procedures to monitor drugs of abuse as well as pharmaceuticals and personal care products (PPCPs) in sewers and STP influents (Ort et al., 2010a). One of the major findings was that sampling is mostly carried out according to existing tradition on the STP or standard laboratory protocols. Even for analysis of drugs of abuse the importance of shortterm pollutant variations on observed concentrations or loads is typically not addressed, although a high data quality is mandatory. As errors of up to 100% and more have been reported (Ort et al., 2010b), it remained unclear for the majority of reviewed studies whether observed variations can be attributed to “real” variations or sampling artefacts. In this study, we therefore take a first step towards assessing the reliability of WW drug consumption estimates and present a novel method to assess the total uncertainty in observed WW COC and BE loads based on measurements and stochastic dynamic load modelling. Specifically, we address uncertainties resulting from discharge measurements, chemical analysis and the applied sampling scheme, and present the following three innovative approaches: (i) A generic model-based procedure to investigate the influence of the sampling scheme on the uncertainty of observed or expected drug loads.
6651
(ii) A comparative analysis of two analytical methods, high performance liquid chromatographyetandem mass spectrometry (HPLCeMS/MS) and gas chromatographyemass spectrometry (GCeMS), including an extended cross-validation by influent profiling over several days. (iii) Monitoring COC and BE concentrations in WW of the largest Swiss STPs. For completeness, the COC and BE loads observed in the STP of Berne were used to back-calculate the COC consumption. However, we deliberately decided not to assess the uncertainty of back-calculated substance abuse figures, because, in our view, the current knowledge (e.g. on degradation) is incomplete.
2.
Methods
2.1. Framework to assess the total uncertainty of COC and BE loads The goal of our inter-disciplinary research project was to develop a methodology to assess the uncertainty resulting from discharge measurements, chemical analysis and sampling procedure. Our guiding principles were that the method should (i) deliver reproducible results, (ii) be simple to apply, (iii) be applicable at reasonable cost, (iv) support the experimental design as well as the a posteriori auditing of existing monitoring data and (v) explicitly account for the influence of the sampling strategy. To this aim, we performed an extensive review on literature regarding environmental sampling, with a focus on the monitoring of continuous WW streams, and identified three candidate procedures: the Theory of Sampling, the Guide to the Expression of Uncertainty in Measurement, and the Eurachem/CITAC procedure, with the latter two being very similar.
2.1.1.
Theory of sampling
The Theory of Sampling (TOS) (Guy, 1998; Petersen et al., 2005), presents a complete methodology for evaluating the total sampling error of process sampling from temporal sampling (integration error), homogeneity of material, etc. However, to assess the influence of the sampling strategy on the observed loads, the so-called “Point Selection Error”, it relies on empirical variographic methods and therefore on speciallytailored monitoring campaigns, has to be estimated. For our purpose, this procedure is not optimal because it only informs about sampling errors under the given conditions (e.g., at that specific morning or night where the sampling was performed). This can be a very strong assumption given the observed daily, weekly (see below) and yearly variability of WW substance loads. In addition, to produce meaningful results, it requires a large dataset at minute resolution. According to Esbensen et al. (2007) 60 or more samples should be available for each variographic analysis, which is prohibitively expensive in most cases.
2.1.2. Guide to the expression of uncertainty in measurement and Eurachem/CITAC guide The Guide to the Expression of Uncertainty (GUM) (JCGM 100, 2008) and the Eurachem guide (Eurachem/Citac, 2000; Ramsey
6652
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 5 0 e6 6 6 0
and Ellison, 2007), which are conceptually very similar, suggest the estimation of the overall uncertainty of a measurement by summation of individual uncertainty contributions, such as device characteristics, intra-day and intra-operator characteristics, etc. The individual influence factors are considered in an additive or multiplicative fashion, which results in a very generic procedure which can easily be extended. It is rather straight forward and thus meets most of our design principles presented above. Following the terminology and practical recommendations given in the Eurachem guide, the first step is to draw a causeeffect (“fishbone”) diagram which includes all relevant influence factors (Fig. 1). The corresponding mathematical model (Equation (1)) to compute analyte loads from the monitoring data is Z LDTR ¼
Q$CDTR dt$fs
(1)
2.1.4.
tc
where tc is the composition interval during which a composite sample is produced, Q the discharge, CDTR the measured concentration and fs a correction factor to account for uncertainty due to the applied sampling strategy (Eurachem/Citac, 2000). Then, the uncertainty of the observed drug loads (u(LDTR)) is assessed using a linear uncertainty propagation framework with independent components (Equation (2)): sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2 2ffi 2 u fs uðQÞ uðCDTR Þ þ þ uðLDTR Þ ¼ fs Q CDTR
(2)
The variable u(LDTR) describes the uncertainty in discharge (WW flow, Q), concentration of drug target residue (CDTR) and sampling uncertainty ( fs), respectively, which will be discussed in the following sections. While the assessment of uncertainty of discharge and concentration follows standard procedures and is straight forward, the assessment of the sampling error circumvents expensive variographic procedures by instead plugging in the results from a stochastic sewer load model.
2.1.3.
is no standard monitoring equipment, nor standard requirements or techniques of calibration. Therefore, we suggest to develop a meaningful measure of uncertainty together with the operators on the STP or at least consult the manufacturer’s information. Calibration studies would be desirable, especially to eliminate systematic deviations. Relative uncertainties u(Q)/Q should also be expressed as single standard deviation. Note the important assumption that the measurements are unbiased, which is not necessarily the case in practice (Thomann-Haller, 2002). Also, information provided by equipment manufacturers or the operators might rather reflect the uncertainty on an instantaneous value than the uncertainty on summary statistics such as the daily mean. Usually, we expect the uncertainty on the daily mean, which is derived from many hundreds of measurements, to be smaller.
Uncertainty of STP discharge data
WW influent measurements of an STP are crucial for the assessment of plant and process performance. However, there
Discharge (QDTR) Method
Sampling (fs) Sampling type
Water level
Sampling interval
Velocity
Load (LDTR) Calibration
Uncertainty of reported concentration values
The uncertainty of concentration measurements is well defined and intra- as well as inter-day precisions should be assessed. For our purpose, we suggest to define a relative uncertainty (u(CDTR)/CDTR), which should include all influence factors, such as purity of standards, extraction recovery, precision and accuracy of calibration and quantitation, etc. If in doubt, the adopted value should be expanded to avoid overconfidence in the analytical precision. Therefore, taking the single standard deviation is in line with a comparably safe assumption on the uncertainty of discharge measurements.
2.1.5.
Uncertainty due to the sampling scheme
As described above, an assessment based on variographic analysis using high-frequency monitoring data would be desirable, but to date is still prohibitively costly because online sensors are lacking. Therefore, we estimate the correction factor due to sampling uncertainty as suggested by Ort and Gujer (2005). Basically, they suggest to first replace the highfrequent observations needed for variographic analysis with synthetic loads from a physically-based stochastic model, which predicts time series of WW pollutant loads with a resolution of a few minutes or less. In the model, a population of users, distributed over the catchment of interest, is emitting WW pulses which are routed through the sewer network topology to the monitoring point using the wellknown analytical solution of the 1D advectionedispersion equation, which is suitable to model solute transport in gravity-driven sewer systems (Rieckermann et al., 2005). The dynamic load pattern at the monitoring point (LDTR) is then computed as the superposition of all the WW pulses during the simulation period (Equation (3)):
Method Sample prep.
Concentration (CDTR) Fig. 1 e Aggregated cause-effect (“fishbone”) diagram to assess the total uncertainty in observed benzoylecgonine (BE) loads in the STP influent due to uncertainty in discharge measurement, chemical analysis and sampling scheme. Factors in grey are examples for individual influence factors that do not necessarily apply to each STP.
LDTR ðtÞ ¼
NP X i¼1
m:pulsei
2 ! 1 S 1 pffiffiffiffiffiffiexp t Ti þ i n s2s;i ss;i 2p
(3)
with t ¼ simulation time, Np ¼ number of pulses discharged per hour, m.pulsei ¼ substance mass contained in the ith pulse, v ¼ mean velocity, Ti ¼ release time of the ith pulse, and si ¼ flow distance. The spread of substance pulse at monitoring station (ss,i) is calculated as follows (Equation (4)):
6653
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 5 0 e6 6 6 0
si 1 þ s2ini s2s;i ¼ 2Dx $ v v2
(4)
with s ¼ single standard deviation of a normal distribution, Dx ¼ longitudinal dispersion coefficient, sini ¼ spread of pulse at the house connection. As several parameters, such as m.pulsei, Ti, si, and sini are described by (empirical) probability distributions (Table 1), this results in a truly stochastic behaviour of the model (Fig. 2). Secondly, they estimate the sampling error by applying the original sampling procedure of the STP to the simulated loads. The relative sampling error S is then defined as the relative difference from the load of the obtained sample (ls) to the simulated reference load (lr) (Equation (5)). S¼
ls lr lr
Fig. 2 e Simulated benzoylecgonine (BE) time series for the STPs of Basel (upper black and grey profiles) and Lucerne (lower black and dark grey profiles).
(5)
2.2.
S: relative sampling error; ls: load of the obtained sample; lr: simulated reference load. To estimate the empirical distribution of fs, a Monte Carlo framework is suggested, where this procedure is repeated a large number of times with varying input factors of pulse masses, voiding times and sewer transport parameters (Table 2). Here, we suggest to use the mean of the obtained sampling errors for fs and the standard deviation of the obtained sample of S as a measure for u( fs). The original procedure has been developed on observed benzotriazole loads, which is an anticorrosive used for silver protection in dishwasher detergents (Ort et al., 2005). While it has been subsequently applied to a variety of other substances, such as gadolinium, our study is the first application to drug target residues (DTR), such as COC and BE. Here, we therefore adapt their procedure to drugs of abuse loads, which requires prior information on population, substance use, catchment characteristics, sewer topology and flow to predict realistic substance load patterns in our case study catchments. Details on this procedure, including important underlying assumptions and suggested simplifications are described together with the case studies in Section 3.
Analytical methods
HPLCeMS/MS and GCeMS were used to determine COC and BE in influent and effluent STP water samples. For experimental details concerning materials, calibration, and validation see Section 2 of Electronic supplementary material. Briefly, HPLCeMS/MS analysis was performed by direct injection reversed-phase liquid chromatography followed by electrospray ionization (ESI) tandem mass spectrometry (MS/ MS) detection using atmospheric pressure ionization (API) and a triple-quadrupole MS-MS system (Agilent 1200 HPLC and AB Sciex API 5000 triple-quadrupole mass spectrometer) (Berset et al., 2010). Quantitation was performed using labelled internal standards for each compound and applying the multiple reaction monitoring mode (MRM). For GCeMS analysis, after adding deuterated internal standards, 500 mL of an STP sample were filtered through a glass microfiber filter and then extracted on a manual SPE unit by using a mixed-mode cation-exchange column. The evaporated extract was then silylated and analyzed by GCeMS operated in the selected ion monitoring (SIM) mode. Peak assignment was achieved by retention times, the characteristic ions of COC and BE, and their ion ratios vs. those of control samples. Calibration was performed by using
Table 1 e Details on the sampling schemes and flow meters of the investigated STPs and the resulting sampling error. STP
Sampling
Sampling interval 3
dV[m ] Berne Basel Geneva Lucerne Zurich
vol. vol. vol. vol. vol.
Prop. Prop. Prop. Prop. Prop.
500 1000 2100 ca. 200 ca. 600
Composition interval
Flow measuring principle
dt[min] 10 18 20 5 8
8:00 8:00 8:00 8:00 8:00
a.m.e8:00 a.m.e8:00 a.m.e8:00 a.m.e8:00 a.m.e8:00
a.m. a.m. a.m. a.m. a.m.
MID Venturi Ultrasonic MID Venturi
u(Q)
Np,BE
u(fs)
[%]
[pulses per day]
[%]
4992 9564 8144 3382 25,469
1.0 2.5 3.1 1.1 0.8
a
1 1.5b 2c 5d 5e10b
STP: sewage treatment plant; vol. Prop.: volume-proportional; dV: volumetric sampling interval; dt: corresponding temporal sampling interval based on measured flow data; MID: MagneticeInductive Flowmeter. a Volumetric calibration. b Expert opinion. c Manufacturer information. d Assumption; u(Q): uncertainty in discharge; Np,BE: number of wastewater pulses containing benzoylecgonine; u(fs): estimated uncertainty due to the applied sampling scheme.
6654
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 5 0 e6 6 6 0
Table 2 e Parameters for stochastic load model and estimation of sampling error by Monte Carlo simulation. Parameter n S Dx n Np,BE sini sampling interval composite interval n.MCS
Unit
Distribution
counts m m2s1 ms1 counts s min h counts
uniform uniform uniform e uniform e e e
Berne
Basel
Geneva
Lucerne
Zurich
139,050 [500; 10,000]
174,635 [500; 9000]
114,766 [500; 12,000]
273,360 [500; 15,000]
4992
9564
3382
25,469
10 24 5000
18 24 5000
242,764 [500; 12,000] [0.03; 0.3] [0.5; 1] 8144 [10; 300] 20 24 5000
5 24 5000
10 24 5000
n: population; S: sewer travel distance; Dx: longitudinal dispersion coefficient; v: velocity; Np,BE: number of wastewater pulses containing benzoylecgonine; sini: initial spread of wastewater pulse at discharge point; n.MCS: number of Monte Carlo Simulations for error propagation.
a spiked blank WW sample. Quantification was based on peak target ion ratios of non-deuterated to deuterated analytes abundance, and linear regression analysis (least squares model).
2.3.
Back-calculation of COC consumption
For the back-calculation of COC consumption we used the original model suggested by Zuccato et al. (2005) (see Equation (6)). They calculated COC by multiplying the BE concentration by the WW flow (Q) and a factor of 2.33. This factor takes into account the COC-to-BE molar mass ratio (1.05) and the midrange excretion percentage of 45% of a COC dose excreted as BE (Cone et al., 1998, 2003; Baselt, 2008; Postigo et al., 2008; Van Nuijs, Castiglioni, et al., 2011). Van Nuijs et al. (2011) recently applied a factor of 3 for the estimation of LCOC from BE concentrations assuming that only 35% of the COC dose is excreted as BE. Similar COC loads were calculated by them using formulae taking into account ecgonine methyl ester (EME) alone or together with BE. COCðg=dayÞ ¼ CBE ðg=LÞ FðL=dayÞ 2:33
3.2.
Modelling COC and BE loads in major Swiss STPs
(6)
3. Case studies: monitoring and modelling COC and BE loads in major Swiss STPs 3.1.
Further 24-h samples (08:00 a.m. to 08:00 a.m., Sunday and Wednesday) were obtained from each of the largest STPs in Switzerland, i.e. Zurich-Werdho¨lzli (Zurich; Table 3: QeS), Geneva-Aı¨re (Geneva, AeD), Basel-Pro Rheno (Basel, GeH), and Lucerne-Region (Lucerne, EeF). The samples were on-site acidified to pH 2 and stored at 4 C in the dark until sent by courier to the laboratory. For Berne and Zurich, the weekend samples (sample P and R) covered an important local 3-days music open-air festival (about 0.1 million participants, July 16e19, 2009) and the annual 1-day mass rave event “Street Parade” (about 0.5 million participants, taking place on August 8, 2009), respectively. Furthermore, the observations from the 14-days monitoring period enabled us to compare the performance of the HPLCeMS/MS and GCeMS methods, focussing on extraction, sensitivity, and general handling. In addition, we also collected effluent samples in the STP of Geneva (sample CeD) in parallel to the influent samples on two days to check the elimination efficiency of STPs.
Monitoring campaigns
For development and validation of the analytical procedures, 1-L WW samples were collected in June and July 2009 (Table 3: sample M) at the STP Berne-Neubru¨ck (Berne), the main WW treatment facility of the city of Berne. Volume-proportional 24-h composite samples were taken from 08:00 a.m. to 08:00 a.m. of the next day, in intervals of 500 m3 with the routine autosampler, which corresponds to an average temporal interval of 10 min (for details see Table 1). The pH of the samples was immediately set to 2 by adding 18% hydrochloric acid. The samples were stored light-protected at 4 C based on own stability experiments and according to Gheorghe et al. (2008). A blank WW sample (see Fig. S2, Electronic supplementary material) was collected from the STP of Da¨rligen, a small Swiss rural community of 400 inhabitants. End of August and beginning of September 2009, 24-h samples were taken from the STP Berne for a period of 14 days (Fig. 3).
As described above, we estimated the correction factor due to sampling uncertainty with a stochastic sewer load model. To simulate realistic load patterns, meaningful input variables and parameter values of the model had to be chosen. While we expected that there reasonable information about catchment characteristics and population is generally available, some stronger assumptions and results from recent research were necessary to specify those parameters which determine substance masses in excreted WW pulses and depend on drug use behaviour and pharmacokinetics.
3.2.1. Population figures, STP catchment characteristics and sewer transport parameters Population figures (n) (Table 2) for the age group 14e65 years were taken from regional statistical offices. For Basel and Geneva they were estimated from total population values applying a proportional scaling procedure. Sewer system topology and flow distances s were estimated based on analyses of the bounding polygon of the STP catchment and the position of the STP, using a Geographic Information System (Maurer and Merlyn, 2006). Sewer transport parameters, such as Dx, n and sini, were taken from Ort et al. (2005) and Rieckermann et al. (2005).
6655
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 5 0 e6 6 6 0
Table 3 e Concentrations (C, ng/L), and loads (L, g/day) of cocaine (COC) and benzoylecgonine (BE) in wastewater collected from major Swiss sewage treatment plants (STP) on different sampling dates and determined by GCeMS; Q: WW flow rate (L/day). Sample A B C D E F G H I K L M N O P Q R S
STP
Day, date
CCOC[ng/L]
LCOC [g/day]
CBE [ng/L]
LBE [g/day]
Q [m3/day]
Geneva Geneva Genevaa Genevaa Lucerne Lucerne Basel Basel Berne Bernec Bernec Berne Berne Berne Bernec Zurich Zurichh Zurich
Wed 2.9.09 Sun 6.9.09 Wed 2.9.09 Sun 6.9.09 Sun 30.8.09 Wed 2.9.09 Wed 26.8.09 Sun 30.8.09 Sun 12.7.09 Sat 18.7.09 Sun 19.7.09 JuneeJulyd JuneeJuly, SateSune JuneeJuly, MoneFrif SateSun, 18.e19.7.09 Wed 5.8.09g Sun 9.8.09 Sun 23.8.09
e 215 e e 16 e 25 11 18 28 17 16 18 14 22 e 216 134
269 1339 101 249 663 244 872 1299 1022 721 1132 655 944 522 927 e >2000i 1622
70 189 26 35 48 29 82 83 59 83 87 52 57 46 89 e 5657 219
260,247 141,236 260,247b 141,236b 71,885 120,263 94,300 63,700 57,800 114,800 77,000 79,226 60,167 88,023 95,900 180,000 195,000 135,000
a Effluent. b Qeffluent approx. corresponding to Qinfluent. c Three-days open-air music festival. d Mean (N ¼ 19). e Mean (N ¼ 6). f Mean (N ¼ 13). g Not available, not analyzed. h Day after mass rave event (“Street Parade”). i >ULOQ.
3.2.2.
Fig. 3 e Loads (L, g/day) of cocaine (COC) and benzoylecgonine (BE) in wastewater from the sewage treatment plant of Berne, collected on 14 consecutive days (AugusteSeptember 2009), and determined by high performance liquid chromatographyetandem mass spectrometry (HPLCeMS/MS). Weekends are marked by grey bars.
DTR masses in excreted WW pulses and numbers
The distribution of drug target residues of interest (DTR) mass in a WW pulse depends among other parameters on the intake route, the drug use behaviour, the voiding behaviour of each user, and the purity of the substance. To obtain uncertainty estimates that are on the safe side, we assumed that the data reported by Cone et al. (2003), represent a reasonable estimate of the variability of pulse masses of an entire population of drug users. The study, which investigated the excretion of COC and major metabolites following different routes of administration, provides detailed amounts of excreted masses and is to the best of our knowledge the most complete dataset available to date (for details see Section 1, Electronic supplementary material). We further assumed that a common COC dose consists of 100 mg/day, the maximum excretion rate is about 45% for BE (Cone et al., 2003; Baselt, 2008), and 5 toilet uses per day is an average value (Boedker et al., 1989). Based on these assumptions we computed the number of pulses (Np,BE) contained in the minimum values of the observed loads. We used BE for the computation of Np,BE, because the variability of the individual pulse masses is much larger (see Section 1, Electronic supplementary material) and consequently leads to larger sampling errors in comparison to COC. To simulate realistic load variations, we chose a model output temporal resolution of 1 min and a simulation time of 3 weeks. To estimate the distribution of fs we simulated 100 realizations of the influent load pattern for each STP with varying input variable of pulse masses, voiding times and
6656
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 5 0 e6 6 6 0
sewer transport parameters. Examples of output from the stochastic model are given for a 1-day window in Fig. 2.
3.3. Assessing the correction factor due to sampling uncertainty ( fs) From each of the 100 patterns fifty 24-h composite samples were computed based on the real-world sampling scheme of the respective STP, using varying starting times. Thus, for each observation, 5000 different realizations of S were computed. As described above, u( fs) was taken as the standard deviation of the empirical histogram.
4.
Results and discussion
4.1.
Analytical methods
For HPLCeMS/MS, correlation coefficients r2 were >0.999, showing excellent linearity for both analytes in the calibration range of 20e1000 ng/L. The lower (LLOQ) and upper limit of quantification (ULOQ) were 20 ng/L (signal-to-noise ratio 1300 and 510 for COC and BE) and 1000 ng/L, corresponding to the lowest and highest calibrator, respectively. Recoveries at the LLOQ were 105 and 117% and at 10 times the LLOQ were 95 and 97% for COC and BE, respectively. COC concentrations of 169 6 and 168 ng/L for water-based and standard addition calibration, respectively, show that comparable values were obtained using both quantification modes. With 453 12 and 453 ng/L, respectively, the BE concentrations were very similar, too. Therefore, stable-isotope-labelled internal standards efficiently compensate for WW matrix effects when analyzing COC and BE in STP samples. In conclusion, validation data clearly demonstrate the suitability of the direct injection HPLCeMS/MS approach for measuring COC and BE in WW. For GCeMS, correlation coefficients r2 0.999 for COC and BE indicate good linearity of the calibration graphs ranging
from 80 to 2000 ng/L. With 80 ng/L (signal-to-noise ratio, S/ N ¼ 3 and 6 for COC and BE, respectively), LLOQ was set at the lowest calibrator concentration. With intra- and inter-assay accuracies of 1 to 8 and 3 to 4% (deviation to target concentration, N ¼ 3) for COC and BE, respectively, the GCeMS method was precise. With intra- and inter-assay precisions of 3e12 and 1e11% (r.s.d., N ¼ 3) for COC and BE, respectively, the assay was also repeatable and reproducible. The overall recovery for COC was 88, 100 and 73% at the high (4 ng/mL), medium (2 ng/mL) and low (0.2 ng/mL) concentration level, respectively. For BE it was 34, 39 and 57%, respectively. When stored at 4 C and pH 2, the COC concentration dropped to 66%, whereas BE showed a loss of only 3% after 29 days. Apparently, the formation of BE by hydrolysis of COC is compensated by the degradation of BE when stored at pH 2 and 4 C in WW (our study) or in pond water (Gheorghe et al., 2008). Compared to HPLCeMS/MS the sensitivity of the GCeMS is lower, thus requiring larger sample volumes and efficient analyte enrichment by SPE. Even when performing an appropriate clean-up by SPE, in rare cases of very dirty WW samples the residual matrix still present in the derivatized extracts is slightly impairing the selectivity of GCeMS. Cleaner extracts may be obtained by repeated SPE washing steps, however resulting in lower analyte recoveries, or using 500 mg SPE cartridges. Due to the lower sensitivity of GCeMS, in some samples (Table 3: A, C, D, F, Q) COC and/or BE could only be quantified by HPLCeMS/MS. In some samples (Tables 3 and 4: D, E, F) considerably lower CBE were found by GCeMS compared to HPLCeMS/MS.
4.2.
Profiling of COC and BE in WW
A 14-days profiling of STP Berne WW was performed by HPLCeMS/MS and GCeMS from Saturday 22.8.2009 to Friday 4.9.2009. With 36 and 66 g/day the highest COC and BE loads (L) determined by HPLCeMS/MS were observed in Berne on Sunday 23.8.2009 and Saturday 29.8.2009, respectively (Fig. 3).
Table 4 e Concentrations (C, ng/L), and loads (L, g/day) of cocaine (COC) and benzoylecgonine (BE) in wastewater collected from major Swiss sewage treatment plants (STP) on different sampling dates and determined by HPLCeMS/MS; Q: WW flow rate (L/day). Sample A B C D E F I K L P Q R a b c d
STP
Day, date
CCOC[ng/L]
LCOC [g/day]
CBE [ng/L]
LBE [g/day]
Q [m3/day]
Geneva Geneva Genevaa Genevaa Lucerne Lucerne Berne Bernec Bernec Bernec Zurich Zurichd
Wed 2.9.09 Sun 6.9.09 Wed 2.9.09 Sun 6.9.09 Sun 30.8.09 Wed 2.9.09 Sun 12.7.09 Sat 18.7.09 Sun 19.7.09 SateSun, 18.e19.7.09 Wed 5.8.09 Sun 9.8.09
114 1928
30 272 e 15 21 15 19 37 23 30 73 195
280 1788 166 425 1040 350 954 694 853 774 802 2400
73 253 43 60 75 42 55 80 66 74 144 468
260,247 141,236 260,247b 141,236b 71,885 120,263 57,800 114,800 77,000 95,900 180,000 195,000
Effluent. Qeffluent approx. corresponding to Qinfluent. Three-days open-air music festival. Day after mass rave event (“Street Parade”).
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 5 0 e6 6 6 0
The remarkable fluctuation of the STP WW profiles is characterized by minimum levels during the first half of the week, e.g. MondayeTuesday, and maximum levels towards the end of the week (SaturdayeSunday). The loads for the 14-days sampling campaign in Berne were normally distributed (data not shown). There is a significant difference between loads on MondayeThursday and FridayeSunday ( p-value >0.01). This confirms our expectation and observations of others (Van Nuijs et al., 2008; Van Nuijs et al., 2011), that recreational use of COC is predominantly occurring on weekends, especially on Saturdays. Note that the low concentrations observed on 24.8., 25.8. as well as 1.e4.9.2009 are also due to dilution from precipitation, which in addition might have caused losses of WW due to combined WW overflows (daily precipitation > 4 mm). The results from the nation-wide monitoring campaigns at the STPs Geneva, Lucerne, Basel, Berne and Zurich are shown in Fig. 4 and summarized in Table 3 (for GCeMS) and Table 4 (for HPLCeMS/MS). With 272 and 468 g/day the highest COC and BE loads measured by HPLCeMS/MS were in Sunday influent samples from the STP Geneva and Zurich (sample B and R). It can be assumed that the extremely high BE level of the Zurich Sunday sample R is due to the “Street Parade”, which with about 0.5 million participants is one of the biggest annual European mass rave events taking place on that weekend. The average COC-to-BE ratios (mean s.d.) was 0.4 0.1 (HPLCeMS/MS). Factors for the unusually high ratio of 1.1 in the Geneva Sunday sample (B) could be (i) a reduced degradation of COC to BE in WW, (ii) disposal of COC to WW due to a police pursuit or (iii) seasonal and temperature variables (Van Nuijs et al., 2009). On Sunday 19.7.2009 (sample L), at the end of a 3-days event in Berne (open-air music festival, about 0.1 million participants), LCOC was surprisingly lower compared to the Sunday before (sample I), whereas LBE was slightly elevated. However, the COC load LCOC was almost 50% higher on this particular event than on a regular Sunday. The results of the elimination efficiency of the STP Geneva differ slightly, depending on the analytical method used (Tables 3 and 4: sample C and D). Overall, 0e6% of COC and 19e59% of BE were still present in the effluent. Therefore, although diluted, a significant amount of BE is reaching the
Fig. 4 e Loads (L, g/day) of cocaine (COC) and benzoylecgonine (BE) in wastewater collected from major Swiss sewage treatment plants (STP) on different sampling dates, determined by GCeMS and HPLCeMS/MS.
6657
surface water. This can be due to incomplete biological treatment, release of untreated WW during heavy rainfall and/or adhesion of BE to sludge particles. With 94e100% the removal efficiency for COC of the STP Geneva is very similar to that observed by Postigo et al. (2010), whereas for BE (41e81% vs. 88%) the elimination rate is significantly lower.
4.3. loads
Assessing the total uncertainty in observed DTR
4.3.1.
Discharge (Q)
In the present study, the different STPs operated flow meters based on electromagnetic induction, ultrasonic velocity and level measurements, and the venturi effect (see Table 1). For the STP Berne, reliable information on the uncertainty of the influent measurement (u(Q)/Q) is available, because it is checked experimentally by comparison to volumetric measurements in the primary clarifier basin on a routine basis. For the other STPs, we relied on the expert opinion of the operators or manufacturer information (see Table 1). From our experience, uncertainties of about 1e10% reflect the quality of very well maintained stationary flow monitoring stations and might very well be larger (Thomann-Haller, 2002). However, the real concern is a bias and not a random uncertainty component.
4.3.2.
Concentrations (C )
The validation of the analytical procedures showed intra- and inter-day precisions to be <20% for the lowest and <15% for medium and high concentrations. To avoid underestimation of the measurement uncertainty we choose u(CDTR)/ CDTR ¼ 0.20 for both COC and BE.
4.3.3.
Sampling uncertainty ( fs)
From our analysis we found that the expected sampling uncertainty for all investigated catchments and applied sampling schemes on each STP varies between 0.8 and about 3%, which is generally small (Table 1). This is mostly due to the relatively large estimated number of WW pulses containing drugs of abuse and the rather small variability of substance masses per pulse in comparison to other substance, such as gadolinium or benzotriazole. In Fig. 5A and B we present 2 histograms as examples of the empirical error distributions obtained from the Monte Carlo Simulations. With an average sampling interval of 20 min the largest uncertainty (about 3%) was computed for the catchment of Geneva (Fig. 5B), the smallest for STP Zurich. Fig. 5C depicts the individual error contributions for the Geneva data. Our results are consistent with Ort and Gujer (2005) in demonstrating that catchments with short sampling intervals on the STP and large number of discharged pulses in the catchment exhibit the smallest error. This is, because the expected short-time fluctuations are less pronounced for patterns which contain many similar WW pulses. In addition, we observe that the distributions are symmetric, which is another sign of less fluctuating pollutant concentrations.
4.3.4.
Total uncertainty
The total uncertainty from discharge, analytics, and sampling is estimated to be <20% for all observed samples (not shown).
6658
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 5 0 e6 6 6 0
A
B
C
24-h composite sample, dt: 20 min
24-h composite sample, dt: 18 min
1000
fs
200
Q
600
Frequency
400 200
Frequency
600
CBE
0
0
u(Load)
-5
0
5
10
Rel. error [%]
-10
-5
0
5
10
0
5000
10000
15000
u(y,xi) [mg/day]
Rel. error [%]
Fig. 5 e Assessment of uncertainty shown by the distribution of the relative sampling error from the STP Basel (A, s.d. [ 0.025) and Geneva (B, s.d. [ 0.031). By the total uncertainty of influent loads of STP Geneva (C) and individual error contributions it is clearly seen that the analytical uncertainty is the dominant factor.
As such, it is dominated by the analytical uncertainty (Fig. 5C). Flow measurement errors are in the same order of magnitude than sampling errors, but vary in importance for the individual STP and catchment.
4.4.
Swiss persons at the age of 15e30 years have at least once consumed COC, with 6.1% being men and 2.6% women (Addiction Info Switzerland, 2010). Taking into account that COC consumption is decreasing with age, this survey correlates reasonably with our estimation for Berne.
Back-calculation of COC consumption
As expected, with 132 and 95 g/day the estimated amounts of used COC (calculated according to Zuccato et al. (2005)) on weekends were in Berne higher compared to those on working days (Table 3: samples N and O; mean, N ¼ 6 and 13). For the period of JuneeJuly 2009 (sample M; mean, N ¼ 19) the mean daily COC consumption is estimated to be 107 21 g. During a 3-days open-air festival with about 0.1 million visitors the estimated weekend COC consumption was 198 g/day (sample P), which is 66 g/day higher than for other weekends in JuneeJuly 2009 (sample N). To calculate the approximate percentage of the population consuming COC, one has to make some assumptions. Firstly, it is estimated that, as an average, the consumed single dose of COC is 100 mg (United Nations Office on Drugs and Crime, 2004; Eve and Rave, 2011), corresponding to 10 doses per gram. Ten doses per gram multiplied with the grams of COC estimated to be consumed per day in Berne during JuneeJuly 2009, which is 107 21 g (sample M), result in an average of 1070 210 doses per day. In 2009, 92,290 of 130,290 inhabitants in the city of Berne were between 16 and 64 years old corresponding to 71% of the population. Seventy-one percent of the 196,711 people living in the STP Berne drainage area correspond to 139,665 people at the age of 16e64 years. This would mean that 0.76 0.15% of the 139,665 people in this particular age group is consuming 1 dose COC per day. However, this estimation is based on the assumption, that street COC is of 100% purity, i.e. not adulterated or diluted, which in reality is not the case. If we consider the median COC content to be 33% (Swiss Society of Legal Medicine, 2009), we estimate a daily COC consumption in Berne of 321 g, which is equivalent to 2.3% of the population living in the STP Berne drainage area consuming COC. According to a telephone survey, 4.4% of the interviewed
5.
Conclusions
In this study, we presented a novel method to assess the total uncertainty in observed WW COC and BE loads resulting from discharge measurements, chemical analysis and the applied sampling scheme and demonstrated its usefulness on data from the largest Swiss STPs. In summary we can draw the following conclusions: The monitoring of drugs or drug target residues in WW has the potential to provide supplementary information on short- and long-term, local, national, and international COC use, if the monitoring data are evaluated critically and accompanied by an estimate of uncertainty. This should take into account all relevant influence factors, such as discharge and chemical measurements as well as the applied sampling protocol. To reliably assess all influence factors, we suggest a linear uncertainty propagation framework which is based on the Eurachem guide and has been extended with a stochastic model-based assessment of sampling uncertainty. Thus, we avoid variographic analysis which requires an expensive set of preliminary data. Instead, the model-based assessment derives an empirical distribution of the sampling error from prior information on population, substance use and excretion, catchment characteristics, sewer topology and discharge, which are generally available. Results for the 5 largest Swiss STPs suggest that the total uncertainty in observed loads is smaller than 20% and that the analytical uncertainty is the dominating influence factor. We compute the sampling uncertainty in daily loads to 1e3%, which is about the same order of magnitude of
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 5 0 e6 6 6 0
stationary flow measurements. For the sampling uncertainty, we identified the number of excreted substance pulses and the sampling interval at the STP as important influence factors. These results could be generally valid for communities of more than 100,000 inhabitants and gravitydriven sewers, but most probably do not hold for small communities and systems with large WW pumps. The highest COC concentrations were found in Geneva influent water, resulting from a Sunday WW sample. The highest BE concentrations were measured in WW of the Zurich STP collected on a Sunday after a mass rave event. Back-calculations with a standard model suggest that, in Berne, about 2.3% of the people between the age of 16 and 64 years are estimated to consume 1 dose of COC per day. This is higher compared to data of The World Drug Report 2008, which estimated for Switzerland a 1.1% annual prevalence of COC consumption in the age group 15e64 years (United Nations Office on Drugs and Crime, 2008). For the first time, two analytical approaches using GCeMS and HPLCeMS/MS were compared on the same WW samples to monitor COC and BE over a longer time period. We found that, except for sensitivity and recovery (GCeMS < HPLCeMS/MS), the validation data showed a similar performance and that, at least for COC and BE quantification in WW, GCeMS can be recommended as an efficient and less expensive alternative. This was recently also stated by Gonzalez-Marino et al. (2010) using GCeMS/ MS. However, when available in a lab, HPLCeMS/MS should be the method of first choice for WW analysis. We deliberately did not tackle to assess the uncertainty of back-calculated user figures itself, because in our view today there is still a lack of understanding of important influence factors such as degradation and surface adsorption in the WW system, sewer leakage, short-term population fluctuations in the catchment and usage habits, which could introduce potential biases. Nevertheless, we believe that our method is useful to audit past and design future monitoring campaigns. Therefore we hope that our approach helps to ensure a high data quality and can serve as a first step towards providing better information on substance abuse in the future.
Acknowledgements Special thanks go to the following STP technicians for assisting in sampling, providing samples, and supporting the project: Peter Wyss and Tanja van der Heijden from the STP Berne, Mr. Caflisch and colleagues from the STP Da¨rligen, Mr. Wahl from the STP Geneva, Mr. Kopf from the STP Basel, Mr. Zumbach from the STP Lucerne, Mr. Langenegger and Mr. Pfund from the STP Zurich.
Appendix. Supplementary data Supplementary data related to this article can be found online at doi:10.1016/j.watres.2011.09.049.
6659
references
Addiction Info Switzerland, 2010. Use of Illegal Substances in Switzerland: Cocaine Lausanne, Switzerland. Baselt, R.C., 2008. Disposition of Toxic Drugs and Chemicals in Man. Biomedical Publications, Foster City, CA. Berset, J.D., Brenneisen, R., Mathieu, C., 2010. Analysis of llicit and illicit drugs in waste, surface and lake water samples using large volume direct injection high performance liquid chromatographyeelectrospray tandem mass spectrometry (HPLCeMS/MS). Chemosphere 81 (7), 859e866. Boedker, A., Lendorf, A., Nielsen, A.H., Glahn, B., 1989. Micturition pattern assessed by the frequency/volume chart in a healthy population of men and women. Neurourol. Urodyn. 8, 4212. 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 (4), 378e394. 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 (5), 556e560. 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 (7), 386e401. Daughton, C.G., 2001. In: Daughton, C.G., Jone-Lepp, T. (Eds.), Pharmaceuticals and Personal Care Products in the Environment: Scientific and Regulatory Issues. American Chemical Society, Washington, DC, pp. 348e364. European Monitoring Centre for Drugs and Drug Addiction (EMCDDA). EMCDDA Project CT.99.EP.08 B, 2002 EMCDDA, 2002. Handbook for Surveys on Drug Use Among the General Population. EMCDDA, Lisbon. Available at: http://www. emcdda.eu.int/?fuseaction¼public. AttachmentDownload&nNodeID¼1390. European Monitoring Centre for Drugs and Drug Addiction (EMCDDA) EMCDDA, 2009. The State of the Drug Problem in the European Union and Norway. Annual Report 2009. EMCDDA, Lisbon. Available at: http://www.emcdda.europa. eu/publications/annual-report/2009. Esbensen, K.H., Friis-Petersen, H.H., Petersen, L., Holm-Nielsen, J. B., Mortensen, P.P., 2007. Representative process sampling e in practice: variographic analysis and estimation of total sampling errors (TSE). Chemometrics Intell. Lab. Syst. 88, 41e59. Eurachem/Citac, 2000. Quantifying uncertainty in analytical measurement. Tech. Rep. Guide CG4, EU-RACHEM/CITEC EURACHEM/CITAC Guide. Eve & Rave, 2011. DrugseSubstance Informations. Eve & Rave, Switzerland. Available at: http://www.eve-rave.ch/drugs. Gheorghe, A., van Nuijs, A., Pecceu, B., Bervoets, L., Jorens, P.G., Blust, R., Neels, H., Covaci, A., 2008. Analysis of cocaine and its principal metabolites in waste and surface water using solidphase extraction and liquid chromatography-ion trap tandem mass spectrometry. Anal. Bioanal. Chem. 391 (4), 1309e1319. Gonzalez-Marino, I., Quintana, J.B., Rodriguez, I., 2010. Determination of drugs of abuse in water by solid-phase extraction, derivatisation and gas chromatography-ion traptandem mass spectrometry. J. Chromatogr. A. 1217 (11), 1748e1760. Guy, P., 1998. Sampling for Analytical Purposes. John Wiley & Sons, Chichester. JCGM 100, 2008. Evaluation of Measurement Data e Guide to the Expression of Uncertainty in Measurement. Available at: http://www.bipm.org/utils/common/documents/jcgm/JCGM_ 100_2008_E.pdf.
6660
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 5 0 e6 6 6 0
Jones-Lepp, T.L., Alvarez, D.A., Petty, J.D., Huckins, J.N., 2004. Polar organic chemical integrative sampling and liquid chromatography-electrospray/ion-trap mass spectrometry for assessing selected prescription and illicit drugs in treated sewage effluents. Arch. Environ. Contam. Toxicol. 47 (4), 427e439. Maurer, M., Merlyn, A., 2006. State, Cost and Investment Needs into Swiss Wastewater Infrastructure. (Zustand, Kosten und Investitionsbedarf der schweizersichen Abwasserentsorgung). in German, Du¨bendorf, Switzerland. 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 (10), 3179e3185. Ort, C., Gujer, W., 2005. Sampling for representative micropollutant loads in sewer systems. Water Sci. Technol. 54 (6e7), 169e176. Ort, C., Schaffner, C., Giger, W., Gujer, W., 2005. Modeling stochastic load variations in sewer systems. Water Sci. Technol. 52 (5), 113e122. Ort, C., Lawrence, M.G., Reungoat, J., Mueller, J.F., 2010a. Sampling for PPCPs in wastewater systems: comparison of different sampling modes and optimization strategies. Environ. Sci. Technol. 44 (16), 6289e6296. Ort, C., Lawrence, M.G., Rieckermann, J., Joss, A., 2010b. 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 (16), 6024e6035. Petersen, L., Minkkinen, P., Esbensen, K.H., 2005. Representative sampling for reliable data analysis: theory of sampling. Chemometrics Intell. Lab. Syst. 77, 261e277. Postigo, C., Lopez de Alda, M.J., Barcelo, D., 2008. Analysis of drugs of abuse and their human metabolites in water by LCeMS/MS: a non-intrusive tool for drug abuse estimation at the community level. Trends Anal. Chem. 27 (11), 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 (1), 75e84. Ramsey, M.H., Ellison, S.L.R., 2007. Eurachem/EUROLAB/CITAC/ Nordtest/AMC Guide: Measurement Uncertainty Arising from
Sampling: A Guide to Methods and Approaches. Eurachem. Available at: http://www.eurachem.org/guides/UfS_2007.pdf. Rieckermann, J., Neumann, M., Ort, C., Gujer, W., 2005. Dispersion coefficients of sewers from tracer experiments. Water Sci. Technol. 52, 123e133. Rutsch, M., Rieckermann, J., Cullmann, J., Ellis, J.B., Vollertsen, J., Krebs, P., 2008. Towards a better understanding of sewer exfiltration. Water Res. 42 (10e11), 2385e2394. Swiss Society of Legal Medicine, 2009. Personal Communication. Thomann-Haller, M., 2002. Datenkontrolle von Abwasserreinigungsanlagen mit Massenbilanzen, Experimenten und statischen Methoden, Dissertation Nr. 14824 ETH Zu¨rich. United Nations Office on Drugs and Crime, 2004. World Drug Report 2004. United Nations, Vienna. Available at: http://www. unodc.org/unodc/en/data-and-analysis/WDR-2004.html. United Nations Office on Drugs and Crime, 2008. World Drug Report 2008. United Nations, Vienna. Available at: http://www. unodc.org/unodc/en/data-and-analysis/WDR-2008.html. Van Nuijs, A.L., 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 (3), 612e621. Van Nuijs, A.L., Castiglioni, S., Tarcomnicu, I., Postigo, C., de Alda, M.L., Neels, H., Zuccato, E., Barcelo, D., Covaci, A., 2011. Illicit drug consumption estimations derived from wastewater analysis: a critical review. Sci. Total Environ. 409 (19), 3564e3577. Van Nuijs, A.L., Pecceu, B., Theunis, L., Dubois, N., Charlier, C., Jorens, P.G., Bervoets, L., Blust, R., Neels, H., Covaci, A., 2008. Cocaine and metabolites in waste and surface water across Belgium. Environ. Pollut. 157 (1), 123e129. Van Nuijs, A.L., Pecceu, B., Theunis, L., Dubois, N., Charlier, C., Jorens, P.G., Bervoets, L., Blust, R., Meulemans, H., Neels, H., Covaci, A., 2009. Can cocaine use be evaluated through analysis of wastewater? A nation-wide approach conducted in Belgium. Addiction 104 (5), 734e741. Zuccato, E., Chiabrando, C., Castiglioni, S., Calamari, D., Bagnati, R., Schiarea, S., Fanelli, R., 2005. Cocaine in surface waters: a new evidence-based tool to monitor community drug abuse. Environ. Health 4, 14.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 6 1 e6 6 6 7
Available online at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Simultaneous heterotrophic and sulfur-oxidizing autotrophic denitrification process for drinking water treatment: Control of sulfate production Erkan Sahinkaya a,*, Nesrin Dursun a, Adem Kilic a, Sevgi Demirel a, Sinan Uyanik a, Ozer Cinar b a b
Harran University, Environmental Engineering Department, Osmanbey Campus, 63000 Sanliurfa, Turkey Kahramanmaras Sutcu Imam University, Environmental Engineering Department, Kahramanmaras, Turkey
article info
abstract
Article history:
A long-term performance of a packed-bed bioreactor containing sulfur and limestone was
Received 12 July 2011
evaluated for the denitrification of drinking water. Autotrophic denitrification rate was
Received in revised form
limited by the slow dissolution rate of sulfur and limestone. Dissolution of limestone for
21 September 2011
alkalinity supplementation increased hardness due to release of Ca2þ. Sulfate production is
Accepted 28 September 2011
the main disadvantage of the sulfur autotrophic denitrification process. The effluent
Available online 19 October 2011
sulfate concentration was reduced to values below drinking water guidelines by stimulating the simultaneous heterotrophic and autotrophic denitrification with methanol
Keywords:
supplementation. Complete removal of 75 mg/L NO3eN with effluent sulfate concentration
Denitrification
of around 225 mg/L was achieved when methanol was supplemented at methanol/NO3eN
Sulfur-limestone autotrophic deni-
ratio of 1.67 (mg/mg), which was much lower than the theoretical value of 2.47 for
trification
heterotrophic denitrification. Batch studies showed that sulfur-based autotrophic NO2eN
Sulfate
reduction rate was around three times lower than the reduction rate of NO3eN, which led
Drinking water
to NO2eN accumulation at high loadings. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
In many countries, nitrate concentration in ground water has exceeded the maximum allowable limits for the drinking water. In the USA, between 10 and 25% of the ground waters used as a drinking water source has nitrate concentration above the maximum allowable concentration of 10 mg/L NO3eN (Liu et al., 2009). Some wells in Harran Plain, Sanliurfa, Turkey contain nitrate as high as 180 mg/L NO3eN and the average concentration for whole plain is 35 mg/L NO3eN (Yesilnacar et al., 2008). The most important sources of nitrate in ground waters are nitrogen containing fertilizers, and industrial and domestic wastewaters discharged without
being treated properly. Reverse osmosis, ion exchange, distillation, and electrodialysis are the physico/chemical methods used for the removal of nitrate from drinking waters. The main disadvantages of these processes are high operational cost, low selectivity, and the formation of secondary brine wastes after treatment. Moreover, these processes are expensive and not proper for in-situ applications. For these reasons, biological denitrification should be considered as an alternative process. Heterotrophic denitrifiers utilize simple organic compounds as a carbon and energy source. The main advantages of the process are high denitrification rate and treatment capacity. However, nitrite accumulates when organic is stoichiometrically insufficient and organic remains
* Corresponding author. Tel.: þ90 414 344 00 20; fax: þ90 414 344 00 31. E-mail address:
[email protected] (E. Sahinkaya). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.09.056
6662
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 6 1 e6 6 6 7
in the effluent when supplemented in excess amount (Liu et al., 2009 and Sierra-Alvarez et al., 2007). In the practice, the addition of exactly the right amount of organic is difficult to achieve (Oh et al., 2001). Organic supplementation is not required in autotrophic denitrification process, which leads to low biomass production, decreased risk of bacterial contamination and reduced operational cost. Elemental sulfur is an attractive source of energy for autotrophic denitrification of nitrate contaminated ground water (Moon et al., 2008 and Soares, 2002). Furthermore, elemental-sulfur is non-toxic, water insoluble, stable under normal conditions, and readily available (Soares, 2002). In this process, the elemental sulfur and nitrate act as an electron donor and an acceptor, respectively. Hence, when nitrate is reduced to nitrogen gas, sulfur is oxidized to sulfate (reaction (1)). þ 55S0 þ 50NO 3 þ 38H2 O þ 20CO2 þ 4NH4 /4C5 H7 O2 N
(1)
þ þ 55SO2 4 þ 25N2 þ 64H
Although sulfur-based autotrophic denitrification has several advantages, its main disadvantages are sulfate and acid formation (Park et al., 2002; Liu and Koenig, 2002; Moon et al., 2008). Lime stone is the most commonly used low-cost alkalinity source (Liu and Koenig, 2002). The main drawback of using limestone is the increased hardness of treated water due to Ca2þ dissolution. In sulfur-based autotrophic denitrification process, 7.54 mg sulfate is formed for per mg NO3eN removal (reaction (1)). Theoretically, around 33 mg/L NO3eN (or 150 mg/L NO 3 ) can be denitrified using sulfur-based autotrophic denitrification without exceeding sulfate limit value of 250 mg/L, set by US EPA (Oh et al., 2001), if treated water has not background sulfate. For the waters with higher nitrate concentrations, heterotrophic and autotrophic denitrification processes can be combined (mixotrophic process) to control the sulfate formation (Lee et al., 2001; Liu et al., 2009 and Oh et al., 2001). During heterotrophic denitrification, 3.57 g CaCO3 is produced for each gram NO3eN denitrified (reaction (2)) (Oh et al., 2001), which can be used by sulfur-oxidizing autotrophic denitrifiers (reaction (1)) in mixotrophic denitrification processes. Hence, mixotrophic process requires less alkalinity compared to autotrophic process. In an effort to reduce alkalinity requirement of autotrophic denitrification, mixotrophic denitrification of nitrified industrial effluents has been studied previously (Kim and Bae, 2000 and Lee et al., 2001). NO 3 þ 1:08CH3 OH þ 0:24H2 CO3 /0:056C5 H7 NO2 þ 0:47N2 þ 1:68H2 O þ HCO 3
(2)
Despite the high denitrification potential of sulfur-based mixotrophic processes, limited studies have been conducted for the denitrification of drinking water (Liu et al., 2009; Soares, 2002). The first study on combined heterotrophic and sulfur-based autotrophic process for drinking water denitrification was conducted by Liu et al. (2009). In this study, they used separate reactors for heterotrophic and autotrophic denitrification processes, which may increase the process cost. In this context, the present study aims at promoting simultaneous autotrophic and heterotrophic denitrification processes (mixotrophic) in one-reactor for drinking water treatment to decrease sulfate formation and alkalinity requirement of sulfur-oxidizing denitrification process.
2.
Materials and methods
2.1.
Bioreactor
A laboratory-scale glass column reactor with an empty bed volume of 350 mL was used. The column reactor was filled with sulfur (0.5e1 mm), lime-stone (0.5e1 mm) and activated carbon (1e1.5 mm) particles with equivalent volume of 117 mL. The activated carbon granules were used to improve the biofilm formation. The use of small sulfur and lime-stone particles was not to limit the denitrification rate as the dissolution of sulfur depends on surface area and decreasing the diameter of sulfur granules would increase the process performance. The reactor was covered with aluminum foil to prevent the growth of phototrophic bacteria. A denitrifying activated sludge obtained from the first anoxic tank of a 5stage Bardenpho process located in Harran University Campus (Sanliurfa, Turkey) was used as inoculum. The reactor was operated in batch mode for 3 days after inoculation, and then it was operated continuously in up-flow mode at 28e30 C in a temperature controlled room. In order to prevent biological activity, the feed container was kept refrigerated at 4 C. The freshly prepared feed solution was deoxygenated by passing through the N2 gas for 20 min. Then, the feed was kept under anaerobic conditions in collapsible feed containers. The reactor was fed with tap water supplemented with 50 mg/L K2HPO4 as a source of phosphorus, and different concentrations of KNO3 to obtain predetermined NO3eN concentrations. During the study of mixotrophic process, methanol was supplemented as an external organic carbon at varying concentrations (15e50 mg/L dissolved organic carbon (DOC)) (Table 1).
Table 1 e Operational conditions of the column reactors. Periods Days NO3eN (mg/L) HRT (h) Loading (g NO3eN/(L.d)) Methanol (mg/L)a
1
2
3
4
5
6
7
8
0e35 50 15 0.080 e
35e52 50 8.4 0.143 e
52e68 50 5.6 0.214 e
68e80 50 11.0 0.109 e
80e122 75 11.0 0.164 e
122e132 75 11.0 0.164 37.0 (14)
132e157 75 11.0 0.164 75 (28)
157e190 75 11.0 0.164 125 (47)
a Values in parenthesis shows the methanol concentrations as DOC.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 6 1 e6 6 6 7
2.2.
Experiments
2.2.1.
Continuous tests with column reactor
A laboratory-scale glass column reactor was operated at different NO3eN, hydraulic retention time (HRT), and autotrophic and mixotrophic conditions (Table 1) to evaluate their effects on denitrification performance, and sulfate production. HRT was calculated considering the empty bed volume. Each operating period shown in Table 1 was changed after reaching steady-state conditions, which was indicated less than 10% differences in at least three successive samplings. The reactor effluent was sampled at least three times a week for the measurement of NO3eN, NO2eN, sulfate, sulfide, DOC, pH, and alkalinity. The feed solution was sampled once a week for the determination of NO3eN, NO2eN, sulfate, DOC, pH, and alkalinity. Additionally, both the feed and the effluent of the reactor were sampled once a week for the measurement of Na, Ca, Mg, and Fe ions, and NH4eN. All measurements were at least double and the standard deviations were within 5% for NO3eN, NO2eN, sulfate and pH, and within 10% for DOC, alkalinity, and ion analysis. In order to promote simultaneous heterotrophic and autotrophic denitrification, the feed was supplemented with methanol after period 5. According to reaction (2), 2.47 g methanol (or 0.93 g methanol as DOC) is required to denitrify each g of NO3eN. Therefore, the theoretical fraction of NO3eN denitrified by heterotrophic bacteria under mixotrophic conditions were 20, 40, and 68%, respectively, for periods 6, 7 and 8 (Table 1). On day 139, 70 mL (20% of the bed) of the reactor content was replaced with fresh sulfur particles to improve the process performance. The daily gas production was measured using liquid displacement method and the gas production rate was compared with the theoretical value using the following equation (Moon et al., 2008). Theorethical N2 gas production rate ðml=dÞ ¼ removed NO3 N conc:ðmg=LÞ Flow rateðL=dÞ
2.2.2.
22:4ml Tempð KÞ 28mg 273:15ð KÞ (3)
Activity tests
The autotrophic denitrification activity was tested on day 140 in parallel 150 mL serum bottles at 30 C, filled with 100 mL medium supplemented with 50 mg/L NO3eN. The serum bottles were consisted of 3 g sulfur and lime-stone particles. After addition of sulfur and lime-stone, N2 gas was passed through the reactors for 5 min to ensure the removal of dissolved oxygen. Then, the serum bottles were inoculated with the 5 mL column bed. The initial biomass concentration in the serum bottles was 127 18 mg volatile suspended solids (VSS)/L. The serum bottles were sampled at regular intervals for the measurement of NO3eN, NO2eN, and sulfate. At the end of the study, biomass was analyzed for organic nitrogen for the calculation of biomass concentration.
2.2.3.
Kinetic tests
The autotrophic denitrification rates were tested in 150 mL parallel batch serum bottles at 30 C, similar to the activity
6663
tests. Again, 3 g sulfur and 3 g lime-stone particles were added to the batch bottles. Then, medium containing varying concentrations of NO3eN (30 mg/L, 80 mg/L, and 100 mg/L) was added to the batch bottles. Five mL suspension (127 18 mg VSS/L medium) of serum bottles used for activity tests (Section 2.2.2) were added as inoculum after passing N2 gas for 5 min. Elemental sulfur is insoluble and apolar mineral, which may make mass transfer an important rate-limiting factor (Sierra-Alvarez et al., 2007). In order to avoid the slow dissolution of sulfur become rate-limiting in batch assays, sulfur was added at much higher amount (3 g) than the theoretical requirement for complete reduction of NO3eN according to reaction (1). The contents in serum bottles were sampled and analyzed similar to the activity tests.
2.3.
Analytical methods
Samples were filtered using cellulose acetate syringe filters with pore size of 0.45 mm before the measurements of NO3eN, NO2eN, sulfate, DOC, and dissolved sulfide in the supernatant. NO3eN, NO2eN, and sulfate were measured using ion chromatography, Schimadzu, Prominence HIC-NS. Sulfide was analyzed spectrometrically using a Shimadzu UV-1601 Spectrophotometer following the method described by CordRuwish (1985). Alkalinity was measured according to Standard Methods (APHA AWWA WEF, 2005) in unfiltered samples titrated with 0.1 N HCI to a pH 4.5 end point. DOC was measured using TOC analyzer (Shimadzu, Japan). Organic nitrogen of biomass at the end of batch tests was measured according to Standard Methods (APHA AWWA WEF, 2005). Then, the biomass concentrations in the batch serum bottles were estimated assuming average cell formula C5H7O2N (Rittmann and McCarty, 2001). Cations were measured with an Inductively Coupled Plasma (ICP) combined with atomic emission. When the standard deviation was larger than the size of the plotting symbol, the standard deviation is shown by þ/ error bars.
3.
Results and discussion
3.1. Denitrification and sulfate production in the column reactor The laboratory-scale glass column reactor was operated under autotrophic (periods 1e5) and mixotrophic (periods 6e8) conditions for around 200 days. In the first three periods, the feed nitrate concentration was kept at 50 mg/L NO3eN and the effect of decreasing HRT from 15 h (period 1) to 5.6 (period 3) on the process performance was investigated (Table 1 and Fig. 1A). Denitrification was almost complete in the first two periods, but nitrite accumulated up to 10 mg/L NO2eN in the 3rd period with corresponding denitrification efficiency of around 80% (Fig. 1A). Nitrite accumulation is a reliable marker of over loading of sulfur-based denitrification process (SierraAlvarez et al., 2007). The loading rate was increased from 0.080 g NO3eN/(L.d) in period 1 to 0.214 g NO3eN/(L.d) in period 3. The maximum denitrification rate was around 0.20 g N/(L.d) observed in period 3. Similarly, Soares (2002) observed
6664
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 6 1 e6 6 6 7
Sulfate (mg/L)
NO3-N or NO2-N (mg/L)
PERIODS 1
80
2
3
4
5
6
7
8 Feed NO3-N
60
Effluent NO3-N
40
Effluent NO2-N
20
A
0 600 500 400 300 200 100 0 9
Feed Effluent Theoretical
B
pH
8 7
Feed Effluent
Alkalinity (mg/CaCO3)
6
C
300
Feed Effluent
250 200 150 100 50
D
0 0
50
100
150
200
Day Fig. 1 e Feed and effluent NO3eN, NO2eN, sulfate, alkalinity and pH variations. The theoretical sulfate concentration was calculated according to reaction (1) assuming that all NO3eN and NO2eN were autotrophically removed.
concentrations (Table 1). The fraction of nitrate denitrified by heterotrophs was determined indirectly from the production of sulfate (Oh et al., 2001). Addition of 37 mg/L methanol (14 mg/L as DOC) did not stimulated heterotrophic denitrification as the sulfate production was similar to the theoretical autotrophic sulfate production calculated according to the reaction (1) (Fig. 1B). Although methanol was almost completely oxidized (Fig. 2), it is surprising not to observe any heterotrophic denitrification activity, which should decrease the sulfate production due to the decreased fraction of nitrate denitrified by sulfur-oxidizing autotrophs. This may be interpreted as the presence of facultative chemolithoautotrophic denitrifiers, such as Thiobacillus versutus, Thiobacillus thyasiris, Thiosphaera pantotropha and Paracoccus denitrificans as they are not only able to grow autotrophically, by using reduced sulfur compounds as energy source, but are also capable of heterotrophic growth. Therefore, facultative chemolithoautotrophic denitrifiers can apparently adapt to different environments, such as autotrophic, heterotrophic or mixotrophic (Oh et al., 2001). In period 7, methanol concentration in the feed was increased to 28 mg/L DOC. The increase in methanol concentration neither started heterotrophic denitrification nor increased the process performance. On day 139, around 20% of the reactor content was replaced with fresh sulfur particles thinking that the sulfur dissolution limits the process performance, which resulted in accumulation of nitrate in the effluent. After that, the reactor performance increased sharply and denitrification efficiency approached 100% although hetereotophic denitrification did not start. In period 8, with the increase of methanol concentration to 47 mg/L DOC, sulfate concentration in the effluent decreased appreciably to below the 250 mg/L, which is the maximum level set by US EPA. Hence, the fraction of NO3eN denitrified by heterotrophs
PERIODS
A
1
60
3
4
5
6
7
8
Influent Effluent
40 30 20 10
B Gas Production (mL/day)
denitrification rate of up to 0.20 g N/(L.d) at a HRT of 1.0 h and NO3eN loading of 0.24 g NO3eN/(L.d). Lee et al. (2001) obtained much higher removal rate (around 5.0 g N/(L.d)) during simultaneous heterotrophic and sulfur-utilizing autotrophic denitrification for nitrified leachate containing 700e900 mg/L NO3eN. The limited dissolution of sulfur and/or lime-stone may limit the denitrification rate as will be discussed later. In period 4, in order to recover the process performance (i.e., the process without nitrite accumulation); HRT was increased back to 11.0 h. In period 5, the feed nitrate concentration was increased to 75 mg/L NO3eN. In this period, the effluent NO3eN and NO2eN increased over 15 mg/L and the approximate denitrification efficiency was 75% (Fig. 1A). Under autotrophic conditions (between periods 1 and 5), the measured sulfate concentrations are in good agreement with the theoretical sulfate production calculated according to reaction (1) (Fig. 1B). After period 5, the reactor was operated under mixotrophic conditions and heterotrophic denitrification was stimulated by the addition of methanol to the feed at various
DOC (mg/L)
50
2
0 120 Measured Theorethical
100 80 60 40 20 0 0
50
100 Time (day)
150
Fig. 2 e The variations of feed and effluent DOC concentrations (A), and produced and theoretical gas production rates (B).
200
6665
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 6 1 e6 6 6 7
þ 4S0 þ 4H2 O/3H2 S þ SO2 4 þ 2H
(4)
Due to the presence of nitrite in the middle of the reactor, sulfide was only detected at the effluent of the reactor. The produced sulfide may be oxidized in the reactor with nitrate as the electron acceptor. Similarly, Moon et al. (2008) reported that at lower part of the column Thiobacillus denitrificans, a sulfur-oxidizing bacterium, was dominant, whereas at the higher part of the column T. denitrificans was disappeared and Chlorobium limicola, a sulfide oxidizing bacterium, became dominant. This data shows that sulfate reduction may occur at the higher part of the column, where nitrate and nitrite were consumed.
3.2.
pH, alkalinity and hardness in the column reactor
The influent pH was around 8.0 throughout the study. The effluent pH (6.2e7.7) was lower than the influent pH until period 8 due to acid production of sulfur-utilizing autotrophic denitrifiers according to reaction (1). The optimum pH for sulfur-oxidizing denitrifiers is between 6 and 9 (Holt et al., 1994). According to reaction (1), 1 mol nitrate removal under autotrophic conditions produces 1.28 mol hydrogen ions, corresponding to 4.57 g CaCO3 consumption per gram of nitrate removal. In period 1, the effluent alkalinity was higher than the influent alkalinity due to rapid dissolution of limestone. In period 2, the effluent alkalinity decreased with increasing feed flow due to limited dissolution rate of limestone. With increasing feed flow and decreasing HRT to 5.6 h, the effluent alkalinity decreased appreciably as alkalinity supplied by lime-stone dissolution cannot balance the acid produced by autotrophic denitrifiers. Hence, the decrease of denitrification efficiency in the 3rd period may be due to decreased alkalinity in the reactor. Although the use of limestone seems to be an effective and economical way of alkalinity supplementation, its limited dissolution rate makes difficult to provide enough alkalinity at high nitrate ladings (Oh et al., 2001). In period 8, the effluent pH and alkalinity increased sharply reaching to the values higher than those of influent. According to reaction (2), 3.57 g alkalinity is produced per gram of NO3eN denitrified heterotrophically. Hence, the increase of pH and alkalinity in the 8th period was due to commencement of heterotrophic denitrification as discussed above. The required
PERIODS 8
2
3
4
5
6
7
8
NO3-N or NO2-N (mg/L)
1
6
Influent
4
2+
Ca (meq/L)
Effluent
2 0 0
40
80
120
160
Time (day) Fig. 3 e The variations of feed and effluent Ca2D concentrations.
200
NO3-N Sulfate NO2-N
50
1200 1000
40
800
30
600
20
400
10
200
0
0
10
20 30 Time (day)
40
50
Sulfate (mg/L)
was about 60% in period 8. According to reaction (2), the theoretical nitrate that can be removed by the addition of 47 mg/L DOC is around 50 mg/L NO3eN, corresponding to about 67% denitrification by heterotrophic denitrifies. Hence, the required amount of organic substances for heterotrophic denitrification is higher than the theoretical value, which may be due to higher cell yield than the prediction, and/or the use some portion of methanol as a carbon source by facultative chemolithoautotrophic denitrifiers as discussed before. Between days 141 and 162, samples from the middle of the reactor was withdrawn for the measurement of NO3eN and NO2eN (data not shown). The NO3eN and NO2eN concentrations averaged 2 2 mg/L and 13 3 mg/L, respectively. Results indicated that the conversion rate of NO3eN to NO2eN is faster than the rate of NO2eN conversion to N2, as also illustrated in batch tests, which will be discussed later. Fig. 2B illustrates that the collected and the theoretical N2 gas production rate are in close agreement as the N2 gas recovery throughout the study averaged 103%. Observing high oscillating results in collected N2 amount was due to the entrapment and sudden release of the produced gas in the column due to the use of small sulfur and lime-stone particles as the packing materials. Similarly, Moon et al. (2008) reported that a portion of the N2 gas produced during denitrification process was entrapped in the pores of the column support materials especially at low up-flow velocities. This entrapment of the gas in the column may decrease the denitrification rate due to mass transfer limitation. In the case of complete denitrification, sulfide (0.02e0.1 mg/L) was detected in the effluent of the column (data not shown) even in the absence of external-organic supplementation (periods 1e5). The reason of this observation was due to the activity of sulfate reducing bacteria at high sulfate concentrations. Biomass decay may supplement the organics required for the reduction of sulfate (Sahinkaya, 2009). Sulfur disproportionation according to reaction (4) may also be responsible for sulfide detection. Sulfur disproportionation develops under anaerobic conditions after depletion of all electron acceptors (NO 3 and NO2 ), which may indicate that the reactor was being operated at lower loadings than the maximum capacity (Luna-Velasco et al., 2010).
0
Fig. 4 e The variations of NO3eN, NO2eN and sulfate concentrations in activity test (Temperature: 30 C, initial nitrate concentration: 50 mg/L NO3eN, initial biomass concentration: 127 ± 18 mg VSS/L).
6666
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 6 1 e6 6 6 7
Activity and kinetic tests
During the batch assays, sulfate concentration increased as the NO2eN and NO3eN were consumed due to the activity of sulfur-oxidizing denitrifiers (Fig. 4). Nitrite was detected as an
NO3-N or NO2-N (mg/L)
30 20
500
100
400
80
300
60
200 10 0 70 60 50 40 30 20 10 0
100 0
C
1000 800 600 400
Sulfate (mg/L)
A
NO2-N or NO3-N (mg/L)
3.3.
200 0 0
10
20
30
Time (day)
40
NO3-N Sulfate NO2-N
B
600 500 400 300 200 100 0
40 20 0
D
400
120 100 80 60 40 20 0
Sulfate (mg/L)
intermediate during the conversion of the NO3eN and the maximum NO2eN concentration was reached when the NO3eN was completely depleted, similar to the results reported by Sierra-Alvarez et al. (2007). Similar to the column reactor, the NO2eN reduction is the limiting step in autotrophic denitrification activity tests. Although NO3eN was completely removed within 10 days (Fig. 4), NO2eN concentration did not change significantly for a further 10 days, then NO2eN was consumed rapidly. The linear portion of the nitrate or nitrite-time curve was used to calculate the removal rate. The specific NO3eN and NO2eN reduction rates were 65 mg NO3eN/(gVSS.d) and 21 mg NO2eN/g VSS.d, respectively. Therefore, NO3eN reduction was over three times faster than NO2eN reduction which caused NO2eN accumulation at high loadings in the column reactors. In the study of Sierra-Alvarez et al. (2007), NO3eN and NO2eN reduction rates increased with increasing initial NO3eN concentration and the maximum specific reduction rates were 77 mg NO3eN/ (gVSS.d) and 60.2 mg NO2eN/(gVSS.d), respectively. The reduction rates of NO3eN and NO2eN at various initial NO3eN concentrations were studied in batch serum bottles, inoculated with 5 mL mixed liquor of the serum bottles used in activity tests. The results of the kinetic tests were illustrated in Fig. 5. Lag phase, between 6 and 10 days, was observed in the tests depending on the initial nitrate concentration, which should be due to low initial biomass concentrations in the tests. Similar to the activity tests, the reduction of nitrate produced nitrite. When the initial NO3eN concentrations were 83 and 110 mg/L, NO2eN concentration reached plateau and its reduction did not commence for further 16 days. In the tests, the NO3eN reduction rate did not change significantly at varying initial concentrations, and it averaged 4.0 1.0 mg NO3eN/L.d. The NO2eN reduction rate for the initial NO3-N concentrations of 30 and 56 mg/L was similar, averaging 1.74 0.06 mg NO2eN/(L.d). The NO3eN reduction rate was around 2.3 times higher than NO2eN reduction rate, similar to
methanol for heterotrophic denitrification of each gram NO3eN was calculated as 2.72 g. Similarly, Liu et al. (2009) reported that when the methanol/NO3eN ratio was 2.47 the denitrification was not complete and methanol remained in the effluent when the ratio was 3.0. In addition to sulfate production and alkalinity consumption, increase of hardness due to lime-stone dissolution is another disadvantage of sulfur and limestone autotrophic denitrification (SLAD) process. Although Fe2þ, Mn2þ, Mg2þ and Naþ concentrations in the influent and effluent did not change, the average Ca2þ concentration in the effluent increased during autotrophic denitrification process due to dissolution of lime-stone (Fig. 3). The effluent Ca2þ increased up to 7.15 meq/L (or 358 mg/L CaCO3) after increasing feed NO3eN concentration from 50 mg/L to 75 mg/L, which was due to increased acid production according to reaction (1) and subsequent increase in the limestone dissolution. After methanol supplementation, effluent Ca2þ started to decrease, which should be due to decreased acid production as a consequence of heterotrophic denitrification activity and decreased limestone dissolution. In period 8, where heterotrophic denitrification was maximum and effluent sulfate concentration below 250 mg/L, effluent Ca2þ concentration decreased to around 2 meq/L (or 100 mg/L CaCO3) due to decreased limestone dissolution as a result of decreased acid generation under mixotrophic conditions. Hence, mixotrophic denitrification has four significant advantages over SLAD process: increased process efficiency, decreased effluent sulfate concentration, decreased alkalinity requirement and, lastly decreased effluent hardness.
300 200 100 0 0
10
20
30
40
Fig. 5 e The variations of NO3eN, NO2eN and sulfate concentrations in kinetic tests at different initial NO3eN concentrations (The serum bottles were inoculated with 5 mL suspension (127 ± 18 mg VSS/L medium) of serum bottles used for activity tests. Temperature was 30 C).
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 6 1 e6 6 6 7
the activity tests. The observation of similar reduction rates at varying initial NO3eN concentrations are in good agreement with the half-saturation constant (Ks) of 0.4 mg/L NO3eN for attached sulfur-oxidizing denitrifiers (Zeng and Zhang, 2005). The obtained results are in contradiction with those of SierraAlvarez et al. (2007) as they observed linear increase in the NO3eN reduction rate with increasing initial nitrate concentration. The reason for this increase in the study of SierraAlvarez et al. (2007) should be due to use of granular consortium (0.5e3 mm), which may result in diffusion limitation and higher apparent Ks values.
4.
Conclusions
Simultaneous sulfur-based autotrophic and heterotrophic denitrification can be achieved in one-reactor for drinking water treatment. Complete denitrification of 75 mg/L NO3eN with effluent sulfate concentration of around 225 mg/L was achieved when feed methanol/NO3eN ratio was 1.67 g/g, which was much lower than the heterotrophic theoretical value of 2.47 g/g. Stimulating mixotrophic denitrification also decreased Caþ2 release and alkalinity requirement of SLAD process due to the alkalinity production of heterotrophic process. Batch experiments showed that sulfur-based autotrophic NO2eN reduction rate was around three times lower than the NO3eN reduction rate.
Acknowledgments _ This study was funded by TUBITAK (Project No: 110Y256) and Harran University Research Fund (Project No: 1062).
references
APHA AWWA WEF, 2005. Standard Methods for the Examination of Water and Wastewater, first ed. American Public Health Association, American Water Works Association, Water Environmental Federation, Washington DC, USA. Cord-Ruwish, R., 1985. A quick method for the determination of dissolved and precipitated sulfides in cultures of sulfatereducing bacteria. J. Microbiol. Methods 4, 33e36.
6667
Holt, J.G., Krieg, N.R., Sneath, P.H.A., Staley, J.T., Williams, S.T., 1994. Bergey’s Manual of Determinative Bacteriology. Williams and Wilkins, Baltimore, MD. Kim, E.W., Bae, J.H., 2000. Alkalinity requirements and the possibility of simultaneous heterotrophic denitrification during sulfur-utilizing autotrophic denitrification. Water Sci. Technol. 42, 233e238. Lee, D.U., Lee, I.S., Chai, Y.D., Bae, J.H., 2001. Effects of external carbon source and empty bed contact time on simultaneous heterotrophic and sulfur-utilizing autotrophic denitrification. Process Biochem. 36, 1215e1224. Liu, L.H., Koenig, A., 2002. Use of limestone for pH control in autotrophic denitrification: batch experiments. Process Biochem. 37 (8), 885e893. Liu, H., Jiang, W., Wan, D., Qu, J., 2009. Study of a combined heterotrophic and sulfur autotrophic denitrification technology for removal of nitrate in water. J. Haz. Mat 169, 23e28. Luna-Velasco, A., Sierra-Alvarez, R., Castro, B., Field, J.A., 2010. Removal of nitrate and hexavalent uranium from groundwater by sequential treatment in bioreactors packed with elemental sulfur and zero-valent iron. Biotechnol. Bioeng. 107, 933e942. Moon, H.S., Shin, D.Y., Nam, K., Kim, J.Y., 2008. A long-term performance test on an autotrophic denitrification column for application as a permeable reactive barrier. Chemosphere 73, 723e728. Oh, S.E., Yoo, Y.B., Young, J.C., Kim, I.S., 2001. Effect of organics on sulfur-utilizing autotrophic denitrification under mixotrophic conditions. J. Biotechnol. 92, 1e8. Park, J.H., Shin, H.S., Lee, I.S., Bae, J.H., 2002. Denitrification of high NO3eN containing wastewater using elemental sulfur; nitrogen loading rate and N2O production. Environ. Technol. 23, 53e65. Rittmann, B.E., McCarty, P.L., 2001. Environmental Biotechnology: Principles and Applications. McGraw-Hill Book Co, New York. Sahinkaya, E., 2009. Microbial Sulfate reduction at low (8 C) temperature using waste sludge as a carbon and seed source. Int. Biodeter. Biodeg 63, 245e251. Sierra-Alvarez, R., Beristan-Cardoso, R., Salazar, M., Gomez, J., Razo-Flores, E., Field, J.A., 2007. Chemolithotrophic denitrification with elemental sulfur for groundwater treatment. Water Res. 41, 1253e1262. Soares, M.I.M., 2002. Denitrification of groundwater with elemental sulfur. Water Res. 36, 1392e1395. Yesilnacar, M.I., Sahinkaya, E., Naz, M., Ozkaya, B., 2008. Neural network prediction of nitrate in groundwater of Harran Plain, Turkey. Environ. Geol. 56, 19e25. Zeng, H., Zhang, T.C., 2005. Evaluation of kinetic parameters of a sulfurelimestone autotrophic denitrification biofilm process. Water Res. 39, 4941e4952.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 6 8 e6 6 7 6
Available online at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
On-farm treatment of dairy soiled water using aerobic woodchip filters Eimear M. Ruane a,b,*, Paul N.C. Murphy b, Mark G. Healy a, Padraig French b, Michael Rodgers a a
Civil Engineering, National University of Ireland, Galway, Ireland Livestock Systems Research Department, Animal and Grassland Research and Innovation Centre, Teagasc, Moorepark, Fermoy, Co. Cork, Ireland b
article info
abstract
Article history:
Dairy soiled water (DSW) is produced on dairy farms through the washing-down of milking
Received 25 April 2011
parlours and holding areas, and is generally applied to land. However, there is a risk of
Received in revised form
nutrient loss to surface and ground waters from land application. The aim of this study was
28 September 2011
to use aerobic woodchip filters to remove organic matter, suspended solids (SS) and
Accepted 30 September 2011
nutrients from DSW. This novel treatment method would allow the re-use of the final
Available online 20 October 2011
effluent from the woodchip filters to wash down yards, thereby reducing water usage and environmental risks associated with land spreading. Three replicate 100 m2 farm-scale
Keywords:
woodchip filters, each 1 m deep, were constructed and operated to treat DSW from 300
Dairy soiled water
cows over an 11-month study duration. The filters were loaded at a hydraulic loading rate
Woodchip
of 30 L m2 d1, applied in four doses through a network of pipes on the filter surface.
Filter
Average influent concentrations of chemical oxygen demand (COD), SS and total nitrogen
Wastewater filtration
(TN) of 5750 1441 mg L1, 602 303 mg L1 and 357 100 mg L1, respectively, were
Nitrogen removal
reduced by 66, 86 and 57% in the filters. Effluent nutrient concentrations remained rela-
Agricultural wastewater treatment
tively stable over the study period, indicating the effectiveness of the filter despite
Solidseliquid separation
increasing and/or fluctuating influent concentrations. Woodchip filters are a low cost, minimal maintenance treatment system, using a renewable resource that can be easily integrated into existing farm infrastructure. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
Dairy farming is a key sector in Irish agriculture and dairy products represent over a quarter of all Irish agri-food exports (Department of Agriculture, Food and Fisheries, 2010). Rising population levels, improved standards of living, and changing dietary patterns, particularly in Asia (Fuller and Beghin, 2004; OECD/FAO, 2009), have all contributed to increased demand for dairy food products. This increased demand has been, and will continue to be, met by more intensive agricultural practises (European Communities, 2008). The Farm Structure
Survey of 2007 (CSO, 2008) highlighted the trend towards a smaller number of dairy cow herds with increasing herd sizes. In 2007, there were a greater number of cow herds in the 50-99 head category compared with 1991 when the majority of cow herds fell within the 10-19 head category (CSO, 2008). Intensification on farms may lead to the production of greater volumes of wastewater, which will require effective management options. Agricultural activities are recognised as significant sources of nutrient inputs to European waters (EEA, 2002). These may contribute to a deterioration in water quality in the form of
* Corresponding author. Livestock Systems Research Department, Animal and Grassland Research and Innovation Centre, Teagasc, Moorepark, Fermoy, Co. Cork, Ireland. Tel.: þ353 87986523; fax: þ353 2542340. E-mail address:
[email protected] (E.M. Ruane). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.09.055
6669
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 6 8 e6 6 7 6
eutrophication (Carpenter et al., 1998), potential toxicity to aquatic species (Kadlec et al., 2005), and groundwater contamination (Knudsen et al., 2006). Legislation in the form of the EU Nitrates Directive (91/676/EEC; EEC, 1991) and the Water Framework Directive (WFD) (2000/60/EC; EC, 2000) has been introduced to address this issue. The aim of the Nitrates Directive is to enforce the protection of receiving water bodies against contamination by nitrate produced through agricultural activities. The WFD endeavours to protect and enhance the water quality of surface, ground and coastal waters, and to ensure that they achieve ‘good status’ by 2015 (Fenton et al., 2008). Agricultural pollution as a result of land spreading is classified as non-point source, or diffuse, meaning the focus of legislation has to be on farm and land management (FAO, 1996). Therefore, the farmer is more accountable for nutrient management (Longhurst et al., 2000). Dairy soiled water (DSW) is water from concreted areas, hard stand areas, and holding areas for livestock that has become contaminated by livestock faeces or urine, chemical fertilisers and parlour washings (SI No.610 of 2010; Martı´nezSuller et al., 2010). It contains high and variable levels of nutrients such as nitrogen (N) and phosphorus (P), as well as other constituents such as spilt milk and cleaning agents (Fenton et al., 2008). Its composition is inherently variable (Table 1) due to the different facilities and management practises that exist on farms, seasonal changes in weather, and management practices (Ryan, 1990; Mingoue et al., 2010). Dairy soiled water is legally defined in Ireland as having a five-day biochemical oxygen demand (BOD5) of less than 2500 mg L1 and less than 1% dry matter (DM) content (S.I. No.610 of 2010). Application of DSW to the land has long been the most common method of disposal employed by farmers (Fenton et al., 2008). However, when DSW is land applied at rates that exceed the nutrient requirements of the pasture, it can create a number of problems, the most significant threat being the loss of P and N in runoff (Silva et al., 1999; Regan et al., 2010) and subsurface leaching of N and, depending on the soil type, P (Knudsen et al., 2006). Other problems associated with the land application of wastes include odour, greenhouse gas (GHG) and ammonia (NH3) emissions (Bhandral et al., 2007), and the build-up of heavy metals in the soil (Wang et al., 2004). However, the European Communities (Good Agricultural Practice for the Protection of Waters) Regulations,
introduced in 2006 and amended in 2010 (S.I. No.610 of 2010), brought about the introduction of a number of restrictions with regard to land spreading of these wastes. Among the restrictions, it imposed a maximum application rate of 50,000 L ha1 in any 48-d period. In order to reduce costs and labour requirements, simple low-maintenance systems utilising natural processes are preferable for the treatment of waste streams on dairy farms. Constructed wetlands (CW) have been investigated for the treatment of agricultural wastewaters (Mantovi et al., 2003; Dunne et al., 2005; Wood et al., 2007). Sand filters (SF), noted for their simplicity, and low capital and operating costs, have been used to treat synthetic DSW at laboratory-scale (Campos et al., 2002; Healy et al., 2007). Constructed wetlands and SFs, however, require large areas of land as they have maximum respective organic loading rates (OLR) of approximately 5 g BOD5 m2 d1 and 22 g BOD5 m2 d1 (Healy et al., 2007). In Australia and New Zealand, waste stabilisation ponds are the most common method of treating DSW (Bolan et al., 2004). Though they are capable of successfully decreasing suspended solids (SS) and BOD5 concentrations to acceptable levels, they are not very successful at decreasing nutrient concentrations (Craggs et al., 2004). Woodchip filters may be effective in treating DSW. Woodchip is already in use on farms to provide outdoor standing areas for cattle during the winter months (Vinten et al., 2006; O’Driscoll et al., 2008). A study in Scotland (Vinten et al., 2006) found that filtration through these outdoor woodchip standing areas, known in Scotland as Corrals, resulted in a 5to 10-fold decrease in faecal indicator bacteria concentrations and dissolved organic carbon (DOC) when compared with fresh slurry. As a result of state schemes introduced in the 1980s to encourage afforestation, Ireland has a young forest stock and a large area of forests that have not yet been thinned (Teagasc Forestry Development Unit, 2007). Thinnings from these young forests may provide a steady supply of woodchips for use in wastewater filters. Such a treatment system may provide a more economical and sustainable alternative to current management practices. Studies have examined the potential of wood-based products to treat various types of contaminated water such as groundwater, high in nitrate, contaminated by septic systems (Robertson et al., 2000; Schipper and Vojvodic-Vukovic, 2001;
Table 1 e Chemical characteristics of dairy soiled water (DSW) for different studies. Reference
Location
BOD5
COD
TN
NH4eN
NO3eN
NO2eN
TP
PO4eP
SS
23 415
353 1a 250e600 1a 1645 6144
1
mg L Healy et al., 2007 Crumby et al., 1999 Sarkar et al., 2006 Longhurst et al., 2000 Schaafsma et al., 2000 Wood et al., 2007 Lansing and Martin, 2006 Mantovi et al., 2003 Di and Cameron, 2000 Martı´nez-Suller et al., 2010 a Unit %.
Ireland England India New Zealand USA UK USA Italy New Zealand Ireland
2208 6593 350e600 2178 2811 517 451 3084
2921 13,383 1500e3000
6690 1219
176 825
85 457
9
269 164 540
48 72 366 52 22 58 32
2 6
65 246 351
69 53 89
57 21
0
0.3
13 55 44
690 7400 12,000
6670
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 6 8 e6 6 7 6
Schipper et al., 2010a), aquaculture, other high-strength wastewaters (Healy et al., 2006; Saliling et al., 2007), and subsurface drainage water (Greenan et al., 2006). These studies focused on saturated woodchip filters and hypothesised that the carbon (C) contained in the woodchip acts as a C source for microbial respiration. Under anaerobic conditions in these filters, denitrification occurs. Buelna et al. (2008) developed a biofiltration system, BIOSORTM-Manure, consisting of a mixture of woodchips and peat moss, to treat high-strength pig manure. Despite a large variation in influent concentrations, the system, loaded at a hydraulic loading rate (HLR) of 12 m3 d1, maintained overall pollutant reductions of greater than 95, 97, 84 and 87% for BOD5, SS, total kjeldahl nitrogen (TKN) and total phosphorus (TP), respectively. The cationic exchange, adsorption and absorption capacity of the organic filter media contributed to the overall treatment of the influent across a wide variation of loads (Buelna et al., 2008). Ruane et al. (2011) investigated laboratory-scale woodchip filters to treat DSW and found SS, chemical oxygen demand (COD) and total nitrogen (TN) removals of >99%, >97% and >89%, respectively. Therefore, aerobic woodchip filtration appears to have the potential to treat DSW. An additional benefit of this system is that the filters act as a medium where liquidesolid separation occurs. This produces a liquid fraction that can be recycled on-farm and a solids fraction that can be composted, or used to produce bioenergy (Garcia et al., 2009). A large proportion of solids contained within the DSW are trapped within the woodchip matrix and a high proportion of the nutrients in DSW are associated with the solid fraction (Garcia et al., 2009; Ruane et al., in press). The aims of this paper were: (i) to assess the performance of woodchip filters, operated under normal farm conditions, to treat DSW (ii) to conduct an economic appraisal of the filters taking construction, recurring and operational costs into consideration, and (iii) to elucidate options for the treatment and/or re-use of final effluent from the filters. To address these aims, three replicate woodchip filters were constructed on a research farm at Teagasc, Moorepark Research Centre in South West Ireland. Each filter was capable of treating DSW generated by 100 cows. The filters were operational for eleven months and filter performance was tested by monitoring influent and effluent waters for nutrients, SS and COD.
2.
membrane, overlain by a felt cover to protect it from abrasion and tearing, was placed directly on top of the soil surface on which the units rested. The base of each pad was then filled with round washed stone (25.4e50.8 mm in size) to make a level surface up to ground level. Sitka Spruce (Picea sitchensis) thinnings, with the bark left on, were chipped onsite and placed directly on top of the stone layer. The size distribution of the woodchip filter media by weight, calculated as the percentage retained on each sieve, was: 28 mm: 9.11%; 20 mm: 2.74%; 14 mm: 28.58%; 10 mm: 29.45%; and on the base: 30.11%. The stone base extended out past the edge of the woodchip to allow for the movement of air underneath the base of the woodchip filter. This was to avoid the development of anaerobic conditions and the potential for denitrification. A wastewater distribution system, consisting of 38.1 mmdiameter plastic pipes placed on top of the woodchip, was constructed to ensure an even distribution of the effluent over the surface of the woodchip (Fig. 1). Distribution pipes were perforated by drilling 4 mm-diameter holes at 0.7 m-spacing on one side of the pipe. These holes were distributed evenly across the top of the filters with each exit hole delivering DSW
Materials and methods
Three replicate farm-scale filter pads were constructed at the Teagasc Animal and Grassland Research and Innovation Centre, Moorepark, Co. Cork, Ireland. The farm filters were operated for a study period of eleven months, from October 2009 (winter) to August 2010 (summer/autumn), inclusive. Each filter pad was constructed to the same specifications. The filter pads had a footprint of 12 m 12 m, a depth of 1 m, and a top surface area of 100 m2 (Fig. 1). The base of the filters was sloped at 1:10 towards a centre line which contained a 101.6 mm-diameter perforated pipe to collect effluent after it passed through the filter. The perforated collection pipe, running half the length of the base, was sloped 1:20 downwards towards a single deepest point (Fig. 1). All the effluent exited the base of the filter at this point. A 0.5 mm-deep plastic waterproof
Fig. 1 e Plan (a) and side view (b) of three farm-scale woodchip filters.
6671
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 6 8 e6 6 7 6
to an area of approximately 0.49 m2. The exit holes faced upwards to facilitate ease of cleaning, when necessary, and so that an even distribution of the effluent could be visually assessed by observing the spurts of water from each hole. Lateral pipes were closed off with a screw stop-end. These could be opened occasionally to allow access to the pipe to clear any build-up of solids that might restrict flow. The distribution system for each filter pad was connected to a separate submersible pump (Pedrollo, Tamworth UK) positioned in the final chamber of a 3-chamber DSW tank. A HLR of 30 L m2 d1 was applied to the filters. This was applied in equal volumes of 750 L, four times daily. Taking in to account head losses in the pipe, the number of bends in the pipe, and the flow curve for the pump, the time to deliver 750 L to each pad was adjusted accordingly to range from 582 s to 898 s. Effluent from all three filter pads was collected in a single tank and a submersible pump was used to pump the effluent to a lagoon on the farm. A 100-ml water sample, obtained from the pipe discharging into the collection tank, was taken from each pad separately for analysis twice weekly. Influent samples were taken, twice weekly, close to the location of the pumps delivering DSW to the filters. Samples were frozen immediately and tested within a period of 14 d. The following water quality parameters were measured: SS (filtered through 1.4 mm paper and dried overnight at 103e105 C); total COD (CODT) and filtered COD (CODF) (dichromate method); unfiltered TN (TN) and filtered TN (TNF) (persulfate method). After filtering through a 1.4 mm filter paper, the following parameters were analysed using a Konelab 20 nutrient analyser (Fisher Scientific, Wathan, Massachusetts): ammonium-N (NH4eN), nitrite N (NO2eN), total oxidised nitrogen (TON) and orthophosphate (PO4eP). NitrateN (NO3eN) was calculated by subtracting NO2eN from TON. Dissolved organic N (DON) was calculated by subtracting TON and NH4eN from TNF. Particulate N (PN) was calculated by subtracting TNF from TN. All tests were carried out in accordance with standard methods (APHA-AWWA-WEF, 1995). To assess the maximum amount of P the filter media was capable of adsorbing, a P adsorption isotherm test was carried out on the wood used in the woodchip filter. Solutions containing four known concentrations of PO4eP were made up: 21.51, 46.06, 61.4 and 92.13 mg PO4eP L1. Approximately 5 g of wood was added to a container and was mixed with 115 ml of each solution concentration (n ¼ 3). Each mixture was then shaken for 24 h using an end-over-end mixer. The solids were separated from the mixture using a centrifuge and tested for PO4eP. The data obtained was then modelled using a suitably fitting adsorption isotherm (Langmuir or Freundlich). The decrease in the concentration of nutrients and other water quality parameters was calculated as the influent concentration minus the effluent concentration, expressed as a percent of the influent concentration.
3.
Results and discussion
3.1.
Organic carbon and SS removal
Influent CODT concentrations averaged 5750 1441 mg L1 and the filters achieved a 66% decrease on the influent
concentration to produce an effluent that had a concentration of 1961 251 mg L1 (Table 2). Much of the influent CODT was associated with the particulate fraction, with CODF accounting for only 30% of CODT. While there was a 66% decrease in CODT, there was only a 43% decrease in CODF, indicating that the filters were less effective at decreasing soluble COD. Therefore, it was likely that physical filtration was the primary removal mechanism for CODT. The aerobic nature of the filters would suggest that oxidation of organic compounds also contributed to the decrease in concentrations of CODT and CODF. The woodchip filters achieved an average decrease of 86% in the concentration of SS, decreasing the concentration from an influent value of 602 303 mg L1 to 84 19 mg L1 (Table 2). From the start of operation, the filters achieved good decreases in the concentration of SS. A laboratory study by Ruane et al. (2011) found that the ability of woodchip filters to remove SS improved over time. In that study, the woodchip used had been de-barked and passed through a 10 mmdiameter sieve; therefore, the gradual build-up of SS in the pore space likely resulted in more immediate SS removal. The presence of bark and smaller woodchip particles in this study likely resulted in the immediate impact on SS concentrations.
3.2.
Nitrogen conversion
An average influent TN concentration of 357 100 mg L1 was decreased by 57% to give an effluent concentration of 153 24 mg L1 (Table 2). This compares favourably with another pilot-scale unit employing horizontal flow over a stack of plastic sheets, which achieved TN decreases in DSW of between 56 and 76% (Clifford et al., 2010). Particulate N accounted for 39% of TN and was decreased by 54%e 64 4 mg L1 in the effluent. The large decrease in PN was consistent with the hypothesis that physical filtration was a primary removal mechanism in the filters. The filters removed, on average, 58% of the influent TNF from 217 64 mg L1 giving an effluent concentration of
Table 2 e Mean chemical composition of influent and effluent dairy soiled water (DSW) treated in three woodchip filter pads over one year of operation. Influent
Effluent
Decrease %
1961 (251) 987 (133) 153(24) 64 (41) 74 (16) 64.80 (25) 37 (10) 4.69 (2) 22.46 (8) 27.15 (17) 91.64 (45) 24.70 (3) 84 (19) 7.8 (0.3)
66 43 57 54 58 68 72 182 74 87 56 31 86 3
1
mg L CODT CODF TN Particulate N TNF Dissolved Org N NH4eN NO2eN NO3eN Mineral N Org N PO4eP SS pH
5750 1744 357 140 217 202.15 134 1.66 12.88 14.54 207.43 36.01 602 7.6
(1441) (488) (100) (65) (64) (63) (45) (2) (10) (10) (77) (17) (303) (0.2)
6672
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 6 8 e6 6 7 6
74 16 mg L1. Dissolved organic N accounted for 31% of the influent TNF with the filters decreasing the DON concentration by 68% to 64.8 25 mg L1. The most likely mechanism for decreasing the concentration of DON is mineralisation to NH4eN. However, sorption onto the filter medium and biological uptake could also have contributed to the decrease of DON. The influent concentration of NH4eN was, on average, 134 45 mg L1 and decreased by 72% to 37 10 mg L1 (Table 2). The influent concentration fluctuated over the duration of the study (Fig. 2). The effluent concentrations reflected these fluctuations, which would suggest that the average rate of decrease of 72% was close to the maximum rate achievable by the filters (Fig. 2). Robertson et al. (2005) and Schipper et al. (2010b) found that once immobilization of N was complete, no substantial long-term removal of NH4eN by adsorption, anaerobic reduction of NO3 to NH4 (dissimilatory nitrate reduction to ammonia; DNRA), or microbial conversion of NO3 and NH4 to N2 gas via an intermediate NO2 (anaerobic ammonium oxidation; ANAMMOX), occurred in woodchip filters. Under aerobic conditions, nitrification is a likely mechanism for decreasing the concentration of NH4eN. This hypothesis is supported by the concurrent increase in NO3eN and decrease in NH4eN in the effluent (Fig. 2). There was a 74% increase in the concentration of NO3eN in the effluent from a concentration of 12.9 10 mg L1 to 22.5 8 mg L1 (Table 2). Some denitrification may also have occurred within the filter, leading to a loss of N in gaseous form as nitrogen gas (N2), N2O, or nitrogen oxide (NOx). A portion of the NH4eN may also have been volatilized. The pH of the effluent DSW was slightly alkaline (Table 2), which may have encouraged ammonia volatilization. However, further investigation into the emission of gases from the filter would be required to verify this.
3.3.
Phosphorus retention
An average influent concentration of 36 17 mg L1 was recorded for PO4eP. This decreased by 31% to an average effluent concentration of 24.7 3 mg L1 (Table 2). This is similar to the decrease of 35% achieved by Morgan and Martin (2008) in a study investigating DSW treatment using an ecological treatment system of aerobic and anaerobic reactors and subsurface wetlands. Using the Langmuir isotherm, the maximum mass of P adsorbed per mass of wood was calculated to be 1958 mg P kg1 woodchip (Fig. 3). Phosphorus adsorption rates for wood are not widely recorded. Comparing the P adsorption capacity of woodchip with the effectiveness of sand to adsorb P, woodchip demonstrated a greater P adsorption capacity. Healy et al. (2010) recorded a value of 85 mg P kg1 for sand. This would suggest that the woodchip could continue to adsorb P over a longer time period before all the potential P adsorption sites become exhausted. The relatively poor PO4eP removals measured (31%) suggest that the P adsorption sites on the woodchip were not fully utilized and that an additional P treatment capacity remained by the end of the study. This may have been a function of an insufficient average hydraulic retention capacity within the filter for the full adsorption of P.
Fig. 2 e The temperature of the wastewater exiting the filters ( C) and the influent and effluent concentration (mg LL1) of ammonium-N (NH4eN), nitrate-N (NO3eN) and unfiltered total nitrogen (TN). Closed diamond indicates influnet measurements. Open square indicates efflunet measurements. Fitted linear regression lines are also shown for influnet (solid line) and effluent measurements (hatched line). Standard deviations are shown for effluent concentrations.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 6 8 e6 6 7 6
6673
7000
Ce/(x/m) (mg L-1)
6000 5000 4000 3000 2000 1000 0 0
2
4
6
8
10
Ce (mg L-1)
Fig. 3 e Langmuir isotherm fitted for the woodchip media. Ce is theh concentration of P in solution at equilibrium (mg LL1), x/m is the mass of P adsorbed per unit mass of woodchip (g gL1) at Ce.
3.4. Impact of seasonal variations and influent concentrations on the data A comparison of the influent and effluent TN, SS, CODT, and PO4eP concentrations and seasonal variations in temperature are illustrated in Figs. 2 and 4. There was an increase in the influent concentration of all four parameters over the duration of the study period and, with the exception of CODT, this followed the same trend as seasonal variations in temperature. Martı´nez-Suller et al. (2010) had similar findings. The TN concentrations were lowest in the winter (NovembereMarch; Days 17e134) and highest in the summer (MayeAugust; Days 197e320) (Fig. 2). This occurred because the farm on which the study was carried out was operated on a seasonal production system and therefore only a small proportion of the herd were milked throughout the winter months. Effluent concentrations for all four parameters increased with the influent concentrations, albeit to a lesser degree for CODT, as indicated by the gradual slope of the fitted regression line for the CODT effluent data. In general, there was considerably less fluctuation in concentrations of the effluent compared to the influent. This would suggest that the woodchip filters are capable of producing a relatively consistent effluent concentration despite increasing and/or fluctuating influent concentrations. This is consistent with the findings of a laboratory study by Ruane et al. (2011) in which SS, CODT and TN concentration in the influent did not have a significant effect on the performance of woodchip filters.
3.5. Economic appraisal of woodchip filter construction and operation Presented in Table 3 are the estimated capital, operational and recurring costs associated with the construction and operation of an aerobic woodchip filter to treat DSW under Irish conditions. The figures presented are based on the three replicated farm-scale filters used in this study, and are presented for guidance purposes only. Calculations are presented for the costs associated with 1 m3 of woodchip, which would provide treatment for one cow on the basis that wash water
Fig. 4 e The influent and effluent concentration (mg LL1) of suspended solids (SS), chemical oxygen demand (COD) and ortho-phosphorus (PO4eP). Closed diamond indicates influnet measurements. Open square indicates efflunet measurements. Fitted linear regression lines are also shown for influnet (solid line) and effluent measurements (hatched line). Standard deviations are shown for effluent concentrations.
generated per cow is approximately 30 L d1 (Mingoue et al., 2010). Capital costs involved in the construction of farmscale filters include: use of a digger to dig out the filter base, a plastic liner to capture the effluent at the filter base, washed stone to make a level base for the woodchip; and pumps and pipes to deliver influent DSW and to collect the treated effluent at the base of the filter. The woodchips constitute the only recurring cost associated with the filters. Woodchip prices used in this paper are based on the cost of hiring a contractor to chip the wood onsite in June 2009. Costs associated with the delivery of woodchip to a farm may differ depending upon factors such as the
6674
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 6 8 e6 6 7 6
Table 3 e Estimated capital, recurring and operation costs associated with the construction and operating of an aerobic woodchip filter to treat dairy soiled water. No. cows
Q (L m2 d1)
Woodchipa (m3)
Costs V Capital
1
30
1
b
Total c
Recurring
Operational
25.48
0.72
33
59
a Including woodchip around the edges of the filter extending out 1 m and inclined at 45 . b Woodchip to be replaced when excessive ponding occurs on the surface of the filter. c Based on the average of three pumps (0.75 kW) at different distances and head losses used in this study operating for between 4.53 and 6.98 h per week for a year at EUR16 cent per unit of electricity (ESB, 2009).
distance of the farm from the woodchip supply base and moisture content of the woodchip. Moisture content can alter the weight of the woodchips and the price accordingly, if purchased on a per tonne basis. Woodchip would need replacing when ponding occurs on the surface of the filter, indicating that the pore space within the filter medium has reached capacity. Estimates suggest that this may occur after 2e3 yr of operation (Ruane et al., in press) and would depend on the concentration of SS in the DSW being applied to the filter. If the build-up of SS extends throughout the entire depth of the woodchip, then all the woodchip would need to be replaced. If SS build-up is restricted to the upper portion of the woodchip, then only this portion of the woodchip would need to be replaced. On-farm management practises should be considered prior to selection of the pump to deliver DSW to the filters and installation of the distribution system. Pump running costs depend upon: the water volumes generated, the head loss in the pipe delivering DSW to the filters, and distance from the holding tank to the woodchip filter. Ideally, the holding tank should consist of at least two compartments: the first compartment for the settlement of larger SS particles and the final compartment housing the pump to deliver DSW to the filter for treatment. The operational costs calculated in Table 3 are based on the average of three replicate woodchip filters, each a different distance from the holding tank (between 4 and 20 m) and with different associated head losses, using 0.75 kW pumps operated, four times daily, for between 582 and 898 s.
3.6.
Management options for woodchip effluent
Two management options may be employed to re-use the final effluent from the woodchip filters. Given the large volumes of fresh water used daily on farms to clean down the holding yard and milking parlour, the effluent could be recycled to wash down the holding yard. An alternative management option would be to apply the effluent to the land. The high concentration of plant available nutrients and low SS concentration would suggest it has potential to benefit plant growth and soil fertility without the traditional problems associated with the land spreading of fresh DSW. The low concentration of SS in the effluent means that, if land applied, the potential for surface sealing of the soil is decreased. The potential for runoff is lowered and the infiltration ability of effluent into the soil profile is increased. The lower concentration of solids reduces problems such as
clogging of pipes and aids the delivery of the effluent to distant fields for targeted irrigation via rotating arms (Petersen et al., 2007). The concentrations of NO3eN in the effluent are just above the maximum allowable concentration for discharge to a receiving water body of 50 mg NO3eN L1 (WHO, 2006). If the effluent from the woodchip filters was to be applied to the land, consideration would have to be given to the timing of application to avoid any potential leaching or runoff to nearby receiving water courses. If applied at a time when plant uptake is at it highest, this form of N would be very beneficial for plant growth. Ammonium-N is also easily utilised by plants (von Wire´n et al., 1997), and this form of N is not as susceptible to leaching due to its positive charge which attracts it to negatively charged soil and clay particles (Miller and Cramer, 2005). Organic N is not immediately plant available, but, in soil, it acts as a slow release fertiliser and mineralises to NH4eN, therefore becoming plant available (Zaman et al., 1999). It is not very mobile in soil, so application and timing rates would be determined based on the NO3eN concentration of the effluent from the woodchip filters. Further investigation into the other fractions of P present in the effluent from the woodchips would be required to determine the potential for long-term build-up of P in the soil matrix. If the effluent were to be reused as ‘flush down’ water in the holding yard of the milking parlour, the concentration of microbes in the effluent would have to be considered. This would determine the part of the farmyard on which this effluent is most suitable for use. Potable water is usually recommended for washing down the holding yard and milking parlour (ADF, 2008). A minimal maintenance and simple tertiary treatment system such as a sand filter may be used to polish the effluent. Using the treated effluent to wash down the holding yard would mean a reduction in the on-farm consumption of fresh water. The potential increase in concentration of NO3eN each time the water was cycled through the system, due to mineralisation and nitrification, would lead to a very nitrate-enriched effluent. As has already been outlined, this could be a very effective fertiliser, but care would also be needed with application rates and timing to minimise the risk of nitrate leaching. Solids from the DSW are trapped in the matrix of the woodchip filter. Spent filter chips could be composted or used in bio-energy production (Garcia et al., 2009). The woodchip provides long-term storage for the solids fraction and the working life of a woodchip filter is estimated to be around two to three years.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 6 8 e6 6 7 6
4.
Conclusions
The main conclusions from this study are: This farm-scale filter study confirmed the effectiveness of woodchip filters to treat DSW under normal operational conditions. Analysis of three farm-scale woodchip filters operating for a duration of 11 months shows that they were capable of decreasing the SS, COD, TN and PO4eP concentrations of fresh DSW by 86, 66, 57 and 31%, respectively. Physical filtration was the principal mechanism of decreasing influent nutrient concentrations in the filters. Mineralisation, nitrification and biological degradation were active processes within the filters. Sorption and biological uptake on the filter media also contributed to decreasing nutrient concentrations. Woodchip filters are capable of producing an effluent that is consistent in SS and nutrient concentration despite fluctuations in influent concentration. Effluent from the filters may be applied to the land. The woodchip filter decreases the influent SS, and the resulting effluent contains nutrients, such as NO3eN, NH4eN and PO4eP, that are readily plant available. The decrease in the concentration of SS in the effluent means that infiltration of DSW into the soil should be enhanced, delivering nutrients to the plant root system and decreasing potential for ammonia volatilisation. These characteristics of the effluent should improve the fertiliser value of nutrients in DSW.
Acknowledgements The authors are grateful to Teagasc for the award of a Walsh Fellowship to the first author and for financial support provided by the Research Stimulus Fund (Department of Agriculture, Fisheries and Food). The authors appreciate the technical help of J. Kennelly and D. Minogue (Teagasc) and E. Clifford and E. O’Reilly (NUI, Galway).
references
ADF, 2008. Assured Dairy Farm Standards & Guidelines for Assessment, 3.2:1 ed. Scotland, UK. APHA-AWWA-WEF, 1995. Standard Methods for the Examination of Water and Wastewater, nineteenth ed.. American Public Health Association, Washington, USA. Bhandral, R., Bolan, N.S., Saggar, S., Hedley, M.J., 2007. Nitrogen transformation and nitrous oxide emissions from various types of farm effluents. Nutrient Cycling in Agroecosystems 79 (2), 193e208. Bolan, N.S., Wong, L., Adriano, D.C., 2004. Nutrient removal from farm effluents. Bioresource Technology 94 (3), 251e260. Buelna, G., Dube´, R., Turgeon, N., 2008. Pig manure treatment by organic bed biofiltration. Desalination 231 (1e3), 297e304. Campos, L.C., Su, M.F.J., Graham, N.J.D., Smith, S.R., 2002. Biomass development in slow sand filters. Water Research 36, 4543e4551.
6675
Carpenter, S.R., Caraco, N.F., Correll, D.L., Howarth, R.W., Sharpley, A.N., Smith, V.H., 1998. Nonpoint pollution of surface waters with phosphorus and nitrogen. Ecological Applications 8, 559e568. Central Statistics Office (CSO), 2008. Farm Structure Survey 2007. Stationery Office, Dublin, Ireland. Clifford, E., Nielsen, M., Sørensen, K., Rodgers, M., 2010. Nitrogen dynamics and removal in a horizontal flow biofilm reactor for wastewater treatment. Water Research 44, 3819e3828. Craggs, R.J., Sukias, J.P., Tanner, C.T., Davies-Colley, R.J., 2004. Advanced pond system for dairy-farm effluent treatment. New Zealand Journal of Agricultural Research 47 (4), 449e460. Crumby, T.R., Brewer, A.J., Dimmock, S.J., 1999. Dirty water from dairy farms: I: biochemical characteristics. Bioresource Technology 67, 155e160. Department of Agriculture, Food and Fisheries, 2010. Annual Review and Outlook for Agriculture, Food and Fisheries 2008e2009. Department of Agriculture, Food and Fisheries, Dublin, p 83. Di, H.J., Cameron, K.C., 2000. Calculating nitrogen leaching losses and critical nitrogen application rates in dairy pasture systems using a semi-empirical model. New Zealand Journal of Agricultural Research 43 (1), 139e147. Dunne, E.J., Culleton, N., O’Donovan, G., Harrington, R., Daly, K., 2005. Phosphorus retention and sorption by constructed wetland soils in Southeast Ireland. Water Research 39 (18), 4355e4362. EC, 2000. Directive 2000/60/EC of the European Parliament and of the Council of 23 October 2000 establishing a framework for community action in the field of water policy. OJ L 327, 1e73. 22.12.2000. EEC, 1991. Council Directive (91/271/EEC) of 12th December 1991 concerning the protection of waters caused by nitrate from agriculture sources. Official Journal of the European Communities L375/1. Brussels, Belgium. ESB, 2009 http://www.esb.ie/esbcustomersupply/residential/ energy_efficiency/what_is_1kwh.jsp (accessed 31.10.10). European Communities, 2008. Eurostat Regional Yearbook 2008. Office for Official Publications of the European Communities, Luxembourg. European Environment Agency (EEA), 2002. Environment Signals 2002, Report No. 9. European Environment Agency, Copenhagen. FAO, 1996. Control of water pollution from agriculture. Series Title: FAO Irrigation and Drainage Papers 55 1996 W2598/E. Fenton, O., Healy, M.G., Schulte, R.P.O., 2008. A review of remediation and control systems for the treatment of agricultural waste water in Ireland to satisfy the requirements of the water framework directive. Biology and Environment 108 (2), 69e79. Fuller, F., Beghin, J.C., 2004. China’s growing market for dairy products. Iowa Ag Review 10 (3), 10e11. Garcia, M.C., Szogi, A.A., Vanotti, M.B., Chastain, J.P., Millner, P.D., 2009. Enhanced solid-liquid separation of dairy manure with natural flocculants. Bioresource Technology 100, 5417e5423. Greenan, C.M., Moorman, T.B., Kaspar, T.C., Parkin, T.B., Jaynes, D.B., 2006. Comparing carbon substrates for denitrification of subsurface drainage water. Journal of Environmental Quality 35, 824e829. Healy, M.G., Rodgers, M., Mulqueen, J., 2006. Denitrification of a nitrate-rich synthetic wastewater using various wood-based media materials. Journal of Environmental Science and Health - Part A Toxic/Hazardous Substances and Environmental Engineering 41 (5), 779e788. Healy, M.G., Rodgers, M., Mulqueen, J., 2007. Treatment of dairy wastewater using constructed wetlands and intermittent sand filters. Bioresource Technology 98 (12), 2268e2281.
6676
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 6 8 e6 6 7 6
Healy, M.G., Burke, P., Rodgers, M., 2010. The use of laboratory sand, soil and crushed-glass filter columns for polishing domestic-strength synthetic wastewater that has undergone secondary treatment. Journal of Environmental Science and Health Part A 45, 1635e1641. Kadlec, R.H., Tanner, C.C., Hally, V.M., Gibbs, M.M., 2005. Nitrogen spiraling in subsurface-flow constructed wetlands: implications for treatment response. Ecological Engineering 25 (4), 365e381. Knudsen, M.T.K., Kristensen, I.B.S., Berntsen, J., Petersen, B.M., Kristensen, E.S., 2006. Estimated N leaching losses for organic and conventional farming in Denmark. Journal of Agricultural Science 144 (2), 135e149. Lansing, S.L., Martin, J.F., 2006. Use of an ecological treatment system (ETS) for removal of nutrients from dairy wastewater. Ecological Engineering 28 (3 SPEC), 235e245. Longhurst, R.D., Roberts, A.H.C., O’Connor, M.B., 2000. Farm dairy effluent: a review of published data on chemical and physical characteristics in New Zealand. New Zealand Journal of Agricultural Research 43 (1), 7e14. Mantovi, P., Marmiroli, M., Maestri, E., Tagliavini, S., Piccinini, S., Marmiroli, N., 2003. Application of a horizontal subsurface flow constructed wetland on treatment of dairy parlor wastewater. Bioresource Technology 88 (2), 85e94. Martı´nez-Suller, L., Provolo, G., Carton, O.T., Brennan, D., Kirwan, L., Richards, K.G., 2010. The composition of dirty water on dairy farms in Ireland. Irish Journal of Agricultural and Food Research Volume 49 (1). Miller, A.J., Cramer, M.D., 2005. Root nitrogen acquisition and assimilation. Plant and Soil 274 (1-2), 1e36. Mingoue, D., Coughlan, F., Murphy, P., French, P., Bolger, T., 2010. Characterisation of Soiled Water on Irish Dairy Farms. Queen’s University, Belfast. BSAS/WPSA/Agricultural Research Forum Conference, 12e14 April. Morgan, J.A., Martin, J.F., 2008. Performance of an ecological treatment system at three strengths of dairy wastewater loading. Ecological Engineering 33 (3e4), 195e209. O’Driscoll, K., Boyle, L., French, P., Hanlon, A., 2008. The effect of out-wintering pad design on hoof health and locomotion score of dairy cows. Journal of Dairy Science 91, 544e553. OECD/FAO, 2009. Agricultural Outlook 2009e2018. Available at: http://www.agri-outlook.org/dataoecd/2/31/43040036.pdf Verified 30 Nov. 2010. Petersen, S.O., Sommer, S.G., Be´line, F., Burton, C., Dach, J., Dourmad, J.Y., Leip, A., Misselbrook, T., Nicholson, F., Poulsen, H.D., Provolo, G., Sørensen, P., Vinnera˚s, B., Weiske, A., Bernal, M.P., Bo¨hm, R., Juha´sz, C., Mihelic, R., 2007. Recycling of livestock manure in a whole-farm perspective. Livestock Science 112, 180e191. Regan, J.T., Rodgers, M., Healy, M.G., Kirwan, L., Fenton, O., 2010. Determining phosphorus and sediment release rates from five Irish tillage soils. Journal of Environmental Quality 39, 185e192. Robertson, W.D., Blowes, D.W., Ptacek, C.J., Cherry, J.A., 2000. Long-term performance of in situ reactive barriers for nitrate remediation. Ground Water 38, 689e695. Robertson, W.D., Ford, G.I., Lombardo, P.S., 2005. Wood-based filter for nitrate removal in septic systems. Transmission ASABE 48, 121e128.
Ruane, E.M., Murphy, P.N.C., Clifford, E., O’Reilly, E., French, P., Rodgers, M., 2011. Treatment of dairy soiled water using a woodchip filter. Journal of Environmental Management. doi: 10.1016/j.jenvman.2011.09.007. Ryan, M., 1990. Properties of different grades of soiled water and strategies for safe disposal. Environmental Impact of Landspreading of wastes. In: Proceedings of the Symposium Held at Johnstown Castle Centre for Soils and Environmental Research and Development, Co, pp. 43e58. Wexford, Ireland, May 30e31, 1990. Saliling, W.J.B., Westerman, P.W., Losordo, T.M., 2007. Wood chips and wheat straw as alternative biofilter media for denitrification reactors treating aquaculture and other wastewaters with high nitrate concentrations. Aqua Engineering 37, 222e233. Sarkar, B., Chakrabarti, P.P., Vijaykumar, A., Kale, V., 2006. Wastewater treatment in dairy industries e possibility of reuse. Desalination 195, 141e152. Schaafsma, J.A., Baldwin, A.H., Streb, C.A., 2000. An evaluation of a constructed wetland to treat wastewater from a dairy farm in Maryland, USA. Ecological Engineering 14, 199e206. Schipper, L.A., Vojvodic-Vukovic, M., 2001. Five years of nitrate removal, denitrification and carbon dynamics in a denitrification wall. Water Research 35, 3473e3477. Schipper, L.A., Robertson, W.D., Gold, A.J., Jaynes, D.B., Camerone, S.C., 2010a. Denitrifying bioreactorsean approach for reducing nitrate loads to receiving waters. Ecological Engineering 30, 1532e1543. Schipper, L.A., Cameron, S.C., Warneke, S., 2010b. Nitrate removal from three different effluents using large-scale denitrification beds. Ecological Engineering 36, 1552e1557. Silva, R.G., Cameron, K.C., Di, H.J., Hendry, T., 1999. A lysimeter study of the impact of cow urine, dairy shed effluent, and nitrogen fertiliser on nitrate leaching. Australian Journal of Soil Research 37, 357e369. Statutory Instrument 610, 2010. European Communities (Good Agricultural Practice for Protection of Waters) Regulations 2010. http://www.irishstatutebook.ie/2010/en/si/0610.html. Teagasc Forestry Development Unit, 2007. A Road Map for the Farm Forestry Sector to 2015. Vinten, A.J.A., Donnelly, S., Ball, B.C., Crawford, C.E., Richie, R.M., Parker, J.P., 2006. A field trial to evaluate the pollution potential to ground and surface waters from woodchip corrals for over-wintering livestock outdoors. Soil Use and Management 22, 82e94. von Wire´n, N., Gazzarrini, S., Frommer, W.B., 1997. Regulation of mineral nitrogen uptake in plants. Plant and Soil 196 (2), 191e199. Wang, H., Magesan, G.N., Bolan, N.S., 2004. An overview of the environmental effects of land application of farm effluents. New Zealand Journal of Agricultural Research 47, 389e403. WHO, 2006. Guidelines for Drinking-Water Quality, Recommendations, vol 1:3. WHO, Geneva, Switzerland. Wood, J., Fernandez, G., Barker, A., Gregory, J., Cumby, T., 2007. Efficiency of reed beds in treating dairy wastewater. Biosystems Engineering 98, 455e469. Zaman, M., Di, H.J., Cameron, K.C., 1999. A field study of gross rates of N mineralization and nitrification and their relationships to microbial biomass and enzyme activities in soils treated with dairy effluent and ammonium fertilizer. Soil Use and Management 15 (3), 188e194.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 7 7 e6 6 8 7
Available online at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Frequency of use controls chemical leaching from drinking-water containers subject to disinfection Syam S. Andra a,b, Konstantinos C. Makris a,*, James P. Shine b a
Water and Health Laboratory, Cyprus International Institute for Environmental and Public Health in association with the Harvard School of Public Health, Cyprus University of Technology, Irenes 95, Limassol 3041, Cyprus b Exposure, Epidemiology and Risk Program, Department of Environmental Health, Harvard School of Public Health, Boston, MA, USA
article info
abstract
Article history:
Microbial-, and chemical-based burden of disease associated with lack of access to safe
Received 29 May 2011
water continues to primarily impact developing countries. Cost-effective health risk-
Received in revised form
mitigating measures, such as of solar disinfection applied to microbial-contaminated
24 August 2011
water stored in plastic bottles have been increasingly tested in developing countries
Accepted 1 October 2011
adversely impacted by epidemic water-borne diseases. Public health concerns associated
Available online 12 October 2011
with chemical leaching from water packaging materials led us to investigate the magnitude and variability of antimony (Sb) and bromine (Br) leaching from reused plastic
Keywords:
containers (polyethylene terephthalate, PET; and polycarbonate, PC) subject to UV and/or
Antimony
temperature-driven disinfection. The overall objective of this study was to determine the
Bottled water
main and interactive effects of temperature, UV exposure duration, and frequency of bottle
Bottle reuse
reuse on the extent of leaching of Sb and Br from plastic bottles into water. Regardless of
Bromine
UV exposure duration, frequency of reuse (up to 27 times) was the major factor that
Polyethylene terephthalate
linearly increased Sb leaching from PET bottles at all temperatures tested (13e47 C).
Polycarbonate
Leached Sb concentrations (w360 ng L1) from the highly reused (27 times) PET bottles
Polybrominated biphenyl ethers
(minimal Sb leaching from PC bottles, <15 ng L1) did not pose a serious risk to human health according to current daily Sb acceptable intake estimates. Leached Br concentrations from both PET and PC containers (up to w15 mg L1) did not pose a consumer health risk either, however, no acceptable daily dose estimates exist for oral ingestion of organobrominated, or other plasticizers/additives compounds if they were to be found in bottled water at much lower concentrations. Additional research on potential leaching of organic chemicals from water packaging materials is deemed necessary under relevant environmental conditions. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
Bottled water market has shown a 25% global increase in average consumption per capita between 2004 and 2009 (Rodwan, 2010). Increases in global population and urbanization along with climate change effects on water supply and
availability have been charged with fluctuating consumer preference toward bottled water in both developed and developing countries. For example, Mexico, a developing country leads the list with the highest consumption of bottled water per capita, worldwide (Rodwan, 2010). Lack of access to safe water, sanitation and hygiene primarily impacts
* Corresponding author. Tel.: þ357 25002398; fax: þ357 25002676. E-mail address:
[email protected] (K.C. Makris). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.10.001
6678
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 7 7 e6 6 8 7
developing countries via water-borne bacterial and other emerging infectious diseases. Abundant, although often carrying a heavy microbial load, drinking-water in developing countries is a primary cause of elevated prevalence’s of gastrointestinal illnesses (Ezzati et al., 2002). Solar disinfection of water packaged in plastic bottles (SODIS) is an increasingly adopted intervention measure aimed at reducing the public health risk from oral ingestion of contaminated water (Sommer et al., 1997; du Preez et al., 2010; Graf et al., 2010). The SODIS method is widely used around the world, enjoying use in 33 developing countries (Meierhofer and Landolt, 2009). In brief, SODIS-treated water is produced by exposure of drinking-water containers to solar UV radiation that inactivates microorganisms (SANDEC, 2002). While the primary working principle of SODIS is the germicidal effect of solar UVA radiation, the associated elevated temperatures can also induce pasteurization effects. The efficacy of the SODIS process, therefore, is driven by both solar-, and temperaturebased effects. Consumer concerns about drinking water quality often stem from episodic events of chemical leaching from watercontact materials (bottled water) (Sax, 2010), but they are often considered unfounded (IBWA, 2011). Examples of chemical leaching from bottled water are: antimony (Shotyk et al., 2006; Shotyk and Krachler, 2007; Westerhoff et al., 2008; Krachler and Shotyk, 2009; Keresztes et al., 2009; Andra et al., 2012), bisphenol A (Casajuana and Lacorte, 2003; Cao and Corriveau, 2008; Le et al., 2008), phthalates (Biscardi et al., 2003; Casajuana and Lacorte, 2003; Criado et al., 2005; Bosnir et al., 2007; Schmid et al., 2008; Leivadara et al., 2008), adipates (Schmid et al., 2008), and 4-nonylphenol (Amiridou and Voutsa, 2011). Little, if any, attention has been paid upon possible contaminant migration during solar disinfection of plastic bottles containing microbially-contaminated water. While the solar disinfection method mentions “aging reduces UV transmittance and hence use fresh bottles” (SANDEC, 2002), it is unlikely that individuals in the developing countries follow this recommendation, because it is not economically viable or convenient to purchase new bottle after each use, leading to frequent reuse of bottles. Studies conducted using PET containers under recommended SODIS conditions claim no potential human threat, despite release of DEHA, DEHP (Schmid et al., 2008) and aliphatic aldehydes (Wegelin et al., 2001). However, though statistically nonsignificant, slightly elevated levels of DEHA and DEHP in water were observed from reused (unknown frequency of reuse) SODIS containers compared to the fresh ones (Schmid et al., 2008). The effect of the frequency of bottle reuse on the magnitude of chemical leaching as well as on the disinfection effectiveness is currently unknown. However, aged and reused plastic bottle containers (19 L) by bottled water distributing companies may be subject to relatively harsh cleaning procedures, i.e., reuse of water containers for as much as 50 times after bottle washing with sodium hydroxide and hydrogen peroxide at 75 C (personal communication, Xristodoulou Bottled Water Company, Limassol, Cyprus). In India, these bottles were used at least for 100 cycles with mild soap and sodium hypochlorite wash between cycles, followed by drying with hot air (personal communication, Ram Das, Sri Krishna Bottling Company, Delhi, India). Similarly, up to 80
times of consecutive reuse of PC bottled water containers after sun exposure for 1 h, autoclaving and chemical disinfection has been demonstrated in Bangladesh (personal communication, Shwapon Biswas, ICDDRB, Dhaka, Bangladesh). Several studies evaluated antimony leaching behavior from water-contact materials as a function of either temperature, or storage time, or container material, or UV exposure (Table SI-1 with references). However, none of these studies took bottle reuse into consideration, except for Schmid et al. (2008) who compared leaching of DEHA and DEHP between fresh and reused (unknown frequency) containers exposed to solar water disinfection (SODIS) procedures. We decided to focus on antimony (Sb), so far the only inorganic contaminant leaching from plastics for which induction of neoplasias, or endocrine disrupting effects have been reported (Choe et al., 2003; WHO, 2003), and bromine (Br) representing brominated compounds being part of water and food contact materials that has received little attention. Polybrominated diphenyl ethers (PBDE), in particular decabromodiphenyl oxide (BDE209), are used as flame retardants in the preparation of both PET and PC plastics (Albemarle, 2011), while tetrabromobisphenol A (TBBPA) is used in epoxy and PC resins for the same purposes (Talsness et al., 2009). The presence of PBDEs (BDE-153, 183, 196, 197, 203, 206, 207, and 209 congener) in plastics, such as high density polyethylene (HDPE), polystyrene (PS), polypropylene (PS) has been reported by Mingwu et al. (2010), and tetrabromodiphenyl ether in PET by CDS Analytical, Inc. (2010). Other brominated compounds used as plastics additives in the making of PET and PC materials are pentabromobenzyl acrylate, pentabromotoluene, ethylene bis-tetrabromo phthalimide, pentabromobenzyl acrylate in PET, and ethylene bis-tetrabromo phthalimide, 1, 2-bis(2, 4, 6tribromophenoxy)ethane in PC material (Covaci et al., 2011). Using a central composite experimental design (CCD) (Mason et al., 1989), this study concomitantly investigated the effects of frequency of bottle reuse, temperature, and UV exposure on the extent of leaching of Sb and Br from UVdisinfected (SODIS) plastic containers (PET and PC). The generated knowledge of this study may be extremely useful to developing countries widely using solar disinfection as a means of improving access to safe water, sanitation and hygiene. The specific aims of this study were to (i) determine the main and interactive effects of temperature, UV exposure, and frequency of bottle reuse on soluble Sb and Br levels in bottled water, and (ii) estimate human daily Sb and Br intakes associated with leaching from water packaging materials under different usage conditions (SODIS), including a worstcase scenario.
2.
Materials and methods
2.1.
Solar disinfection simulated experimental setup
Recommendations for solar disinfection of microbialcontaminated water include filling colorless and transparent 2 L plastic bottles made of PET, not exceeding 10 cm water depth when laid horizontally for solar UV exposure either 1 h at >50 C, or 6 h if up to 50% cloudy day, or for 2 consecutive days if >50% clouds (SANDEC, 2002). Following SODIS
6679
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 7 7 e6 6 8 7
recommendations, commercially available PET bottled water containers of a single brand with 2 L volume, 30 cm height and 10 cm diameter dimensions were procured from a supermarket in Boston, MA (USA) and emptied prior to experimentation. Due to extensive use of 19 L PC bottles around the world, being disinfected via a combination of chemical and high temperature treatments, PC bottles were also included in the leaching experiments. Laboratory PC bottles (2 L), 20 cm height and 15 cm diameter with similar dimensions to those of PET bottles were emptied, washed thoroughly, air dried and used in our experiments. Bottles were cleaned after each use (emptying all contents) using 0.5 mL of liquid soap and tap water, followed by thorough rinsing with deionized water, and air drying. Each bottle was filled with 2 L ultra-pure water (18.2 MU-cm) from a Barnstead nanopure water purification system (Thermo Scientific, NJ, USA) as a migration stimulant. Solar exposure was simulated via the use of a UV-A lamp (365 nm, UVL-28 EL series, UVP Company, Upland, CA, USA); the 320e400 nm wavelength of UV-A light possesses germicidal properties (Wegelin et al., 2001). The UV-A lamp intensity was about 1500 mW cm2 at a 5 cm distance from the lamp, which was placed within the incubator (personal communication, technical support staff, UVP Company, CA, USA). In a temperature-controlled biological incubator (Blue-M, model #B2730Q, Blue Island, IL, USA), bottles representing the n-th time of reuse from a specific experimental run of the CCD design were subject to the following procedure repeated for n times: deionized-water filled bottle was exposed to UV-A radiation for 1 h at 50 C, being equivalent to one SODIS cycle (SANDEC, 2002); after each cycle, aliquots were collected and analyzed for soluble Br and Sb concentrations. After subjecting bottles to the respective frequency of reuse, they
were exposed to simulated SODIS conditions in the incubator according to Table 1 factorial combinations of the CCD. Filled bottled waters were placed horizontally (on long axis) in the incubator (no shaking) and the following range of SODISsimulated conditions was applied: i) temperatures in the range of 13 C to 47 C; ii) UV-A exposure in the range of 0 h to 16 h; and iii) frequency of bottle reuse in the range of 0 to 27 times (Table 1).
2.2.
Central composite design
Conventional full factorial designs have limitations when the factor number is greater than two since the experimental runs appear to be very large (N ¼ 3k) making it difficult in practice. The central composite design (CCD) presents fewer experimental points and it has been widely used in the fields of science and engineering (Mason et al., 1989). Examples of CCD use in related fields are studies measuring chlorobenzenes in water (Vidal et al., 2007), arsenic bioavailability in contaminated soils (Makris et al., 2008), polybrominated diphenyl ethers in sediment (Yusa et al., 2006), aluminum in fruit juices and soft drinks (Jalbani et al., 2006), toxic metals in mussel tissues etc. (Farfal et al., 2004), polychlorinated biphenyls in human serum (Lopez et al., 2007), just to name a few. Influence of temperature, UV exposure duration, and frequency of reuse on leaching of Sb and Br into bottled water was evaluated with the CCD using Design-Expert software, version 6.0 (Stat-Ease, Inc., Minneapolis, MN, USA). Five levels of each variable, ranging from 13 to 47 C, 0 to 16 h, and 0 to 27 times for the temperature, UV exposure duration, and frequency of reuse, respectively, were studied using the CCD (Table 1). Number of experiments used by the CCD was determined by
Table 1 e Factorial combinations and measured values of leached Sb and Br in bottled water under the experimental conditions of the central composite design. #
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Factors
Response
Temperature ( C)
UV-A exposure (h)
Bottle reuse (#)
PET_Sb (ng L1)
PC_Sb (ng L1)
PET_Br (mg L1)
PC_Br (mg L1)
20 40 20 30 30 20 13 40 40 20 30 30 30 30 30 30 30 40 47 30
0 0 12 6 0 0 6 12 0 12 6 6 6 6 6 6 16 12 6 6
0 0 0 0 10 20 10 0 20 20 10 10 10 10 10 10 10 20 10 27
1.06 2.7 5.9 6.0 6.3 5.8 5.8 5.0 5.5 5.1 8.7 9.2 8.1 7.8 8.9 8.2 11 9.3 10.3 13.8
1.2 6.8 8.2 8.8 13.0 7.5 6.6 8.4 7.8 9.5 12.2 12.2 11.3 12.2 11.6 13.1 10.3 10.8 10.0 14.8
a Limit of detection for Sb and Br were 0.5 ng L1 and 0.1 mg L1, respectively.
6680
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 7 7 e6 6 8 7
the equation n ¼ 2k-p þ 2k þ c, where k is the number of study variables, p the fractionalization element set at 0 in case of a full design as in this study, and c is the number of central points (Barker, 1985). This equation with k ¼ 3, p ¼ 0, and c ¼ 6, yielded 20 experimental runs to assess the main and interaction effects between the three variable combinations (Table 1), which is clearly a big advantage over using conventional factorial design that requires hundreds units of experimental runs to study the same. Responses from each experimental run were recorded as total soluble Sb and Br concentrations in bottled water contained in either PET or PC bottles (Table 1). Data was fitted to all available models e linear, interaction, quadratic, and cubic; to identify the best fit for each response. Statistical significance of each response model was assessed with analysis of variance (ANOVA) and least squares analyses followed by Fisher’s statistical test (F-test). Prediction equations for the responses were constructed based on the significant (95% confidence interval) response surface models. Correlation (Pearson’s) graphs were constructed for measured Sb and Br concentrations that leached from PET and PC containers.
2.3.
Analysis of soluble Sb and Br in bottled water
Water samples collected from each experimental run in the CCD were analyzed for total soluble Sb and Br concentrations using inductively coupled plasma mass spectrometer (ICP-MS) (PE Sciex Elan-DRC-II axial field technology, PerkinElmer, USA). Details of the ICP-MS parameters used during Sb and Br analyses can be found in the Supporting Information (Table SI-2). Bottled water samples were acidified with 1% HNO3 and allowed to equilibrate at room temperature for 2 h (Krachler and Shotyk, 2009). Dilutions were performed using ultra-pure water (18.2 MU-cm) from a Barnstead nanopure water purification system (Thermo Scientific, NJ, USA). Both nitric acid and deionized water used for sample preparation and analysis were tested for Sb and Br, being below detection limits (Table SI-3). Limit of detection for Sb and Br were 0.5 ng L1 and 0.1 mg L1, respectively (Table SI-3). Given the differences in Sb and Br concentrations, each sample was analyzed twice using two different calibration ranges: 1 to 1000 ng L1 for Sb made from stock Sb calibration standard solution (PerkinElmer, USA) and 0.5 to 100 mg L1 for Br made from stock potassium bromide standard solution (Sigma Aldrich, USA) (Table SI-3). Quantification of Sb and Br was based on monitoring the analyte isotope and corresponding internal standard, which were 121Sb/175Lu and 79Br/128Te, respectively. Quality assurance (QA) and quality control (QC) practices were strictly observed. In specific, a minimum of 25% QC runs were included for all analyses (Table SI-3). In addition to internal QCs, standard reference materials (SRMs) for Sb and Br were analyzed as external QCs. The SRM 1643e, Trace Elements in Water (National Institute of Standards and Technology, Gaithersburg, MD, USA), was used for Sb analysis, while the IAEA A-13, animal blood SRM (International Atomic Energy Agency, Vienna, Austria), was used for Br (Table SI-3). IAEA A-13 was the closest Br available reference material to water matrix. Recoveries from spiked samples and SRMs for both Sb and Br were always >90% (Table SI-3).
3.
Results and discussion
3.1. Simultaneous leaching of antimony and bromine into bottled water The magnitude of Sb and Br leaching was a complex function of plastic type and experimental factorial combinations as laid out by the CCD. Antimony leaching increased from PET bottles under certain factorial combinations in the CCD setup, reaching as high as 360 ng L1, while minimal Sb leaching occurred for PC bottles (up to 17 ng L1) (Table 1). Bromine leaching increased from either PET or PC bottles under certain factorial combinations in the CCD setup, reaching concentrations up to 15 mg L1 (Table 1). Certain factors, such as temperature and storage time facilitated leaching of Sb from plastic containers into water (Westerhoff et al., 2008; Keresztes et al., 2009). Leaching of both Sb and Br was observed for 31 bottled water samples collected from Boston, MA, USA markets after storage for 60 days (23 C temperature, no-shaking and 12 h/12 h light/dark) (Andra et al., 2012). It was, thus, warranted to study the factors influencing the magnitude and uncertainty of Sb and Br leaching from PET and PC water packaging materials. The CCD setup allowed for the selection of the most significant (at the 95% confidence interval) model explaining the observed variability in measured Sb and Br concentrations subject to concomitant influence of the three tested factors (temperature, UV exposure length, and frequency of bottle reuse). Linear models were chosen for both Sb and Br leaching from PET and PC types of containers, except for the case of Br leaching from PC where a quadratic model was chosen, being the only significant model at the 95% confidence level. Based on ANOVA tables, frequency of bottle reuse was the most significant ( p < 0.001) factor influencing Sb and Br leaching from PET and PC plastic containers into water (Table SI-4). Main effects of the three tested factors influencing Sb and Br concentration leaching from PET and PC followed the order of frequency of bottle reuse > UV exposure > temperature (Table SI-4). Interaction terms of the three aforementioned factors were not significant at the 95% confidence level with the exception of Sb leaching from PC bottles, where interaction of frequency of bottles reuse and UV exposure was significant ( p < 0.05) (Table SI-4). Frequency of bottle reuse was the major factor that linearly increased Sb leaching from PET bottles at all temperatures tested (13 C to 47 C) regardless of UV exposure duration (Fig. 1). In the case of PC bottles, Sb leaching was significantly greater for UV-exposed bottles that were never used before (just purchased container), but not for reused PC bottles (Fig. 1). Upon frequent PC bottle reuse, Sb leaching was unaffected by changes in UV exposure duration at any temperature (Fig. 1). However, the magnitude of leached Sb concentrations in PC bottles was minimal, overall (<15 ng L1), indicative of the sporadic use of Sb as a flame retardant in PC polymeric applications (Fig. 1) (Albemarle, 2011). Leaching behavior of Br from PET and PC bottles followed a similar to Sb trend (Fig. 2). Bromine leaching from PET bottles linearly increased with frequency of reuse, irrespective of temperature increase in the range of 13 C to 47 C, or UV exposure duration (Fig. 2). In the case of PC bottles, a quadratic
6681
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 7 7 e6 6 8 7
Temperature (30C)
Temperature (13C) UV exposure (h)
150
0
300
150
0
-150
-150 0
7
14
20
27
0
7
20
27
10
0
-10 20
7
27
14
20
27
Bottle reuse (#)
Temperature (47C) UV exposure (h)
30
20
10
0
20
10
0
-10
-10 14
0
PC_Sb (ng L-1)
PC_Sb (ng L-1)
PC_Sb (ng L-1)
14
UV exposure (h)
30
20
Bottle reuse (#)
0
Temperature (30C)
UV exposure (h)
7
150
Bottle reuse (#)
Temperature (13C)
0
300
-150
Bottle reuse (#)
30
UV exposure (h)
450
PET_Sb (ng L-1)
300
Temperature (47C)
UV exposure (h)
450
PET_Sb (ng L-1)
PET_Sb (ng L-1)
450
0
7
14
20
Bottle reuse (#)
27
0
7
14
20
27
Bottle reuse (#)
Fig. 1 e Frequency of bottle reuse and UV exposure duration linear interaction plots for Sb leaching from PET and PC bottles at three different temperatures. Circles, triangles and rectangles stand for measured CCD design points, model predicted 12 h UV exposure points, and no UV exposure points, respectively. Black and red lines represent UV exposure (h) at their low (no UV) and high (16 h UV exposure) levels from the CCD. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
model explained the significant effect of frequency of reuse on leaching of Br concentrations at all tested temperatures (Fig. 2); no interaction was significant ( p > 0.05). Aggregating all CCD data obtained for both Sb and Br leaching from PET and PC bottles resulted in a significant ( p < 0.001) linear correlation between leached Sb and Br concentrations obtained under all factorial combinations of CCD (Fig. 3). Because of the significant correlation, it was suggested that a common mechanism was responsible for the concurrent leaching of Sb and Br, but elucidation of possible leaching mechanisms was not the focus of this study (Fig. 3). Data from a market-representative bottled water survey conducted in Boston, USA showed a highly significant ( p < 0.001) correlation between log-transformed soluble Br and Sb concentrations indicating similar leaching mechanisms from aggregated data obtained from non-carbonated, carbonated, and enriched (vitamin, fruit aroma, etc.) bottled water classes (Andra et al., 2012). The final pH of all bottles used in our CCD study did not change ( p > 0.05) for all factorial combinations, remaining approximately equal to 6.5. Chemical migration from water packaging materials has been thoroughly studied, showing Sb leaching from PET bottles as a function of temperature (Westerhoff et al., 2008;
Keresztes et al., 2009) and solar UV exposure duration (Westerhoff et al., 2008). In this study, PET or PC bottle reuse resulted in gradual Sb leaching without reaching pseudoequilibrium even after 27 times of reuse (Fig. 4). Water Sb concentrations in a PET bottle reused up to 27 times were 250 ng L1, being lower than the maximum contaminant level of Sb in drinking-water (5 mg L1, EU Council Directive 98/83/ EC, 1998). According to CCD model, conditions inducing minimum leaching of Sb from bottled water were considered those of 20 C temperature, 0 h UV (no UV) exposure, and new (never reused) bottles (Table 2). Concentrations of Sb in PET and PC bottled water under these conditions were below limit of quantification (0.5 ng L1), since bottles were initially filled with deionized water (Table 2). Leaching of Sb and Br was predicted (based on the linear model) for the SODIS conditions (6 h at 30 C) to be significantly higher for a reused (20), when compared with that of a never used (0) PET or PC bottle (Table 2). In effect, three times higher Sb concentrations were predicted for a reused, versus a brand new PET or PC bottle subject to SODIS treatment (Table 2). Five times higher Sb concentrations leached from PET or PC reused (27) bottles subject to a worst-case treatment (12 h UV, 47 C) when compared with Sb leaching from new bottles subject to SODIS (Table 2).
6682
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 7 7 e6 6 8 7
Temperature (30C)
Temperature (13C) UV exposure (h)
5
0
-5
10
5
0
7
14
20
27
0
20
27
0
27
14
20
27
Temperature (47C) UV exposure (h)
30
20
10
0
-10
-10 20
7
Bottle reuse (#)
PC_Br (µg L-1)
PC_Br (µg L-1)
10
14
0
UV exposure (h)
30
20
7
0
Temperature (30C)
UV exposure (h)
0
5
Bottle reuse (#)
Temperature (13C)
PC_Br (µg L-1)
14
7
Bottle reuse (#)
30
10
-5
-5 0
UV exposure (h)
15
PET_Br (µg L-1)
10
Temperature (47C)
UV exposure (h)
15
PET_Br (µg L-1)
PET_Br (µg L-1)
15
20
10
0
-10 0
7
Bottle reuse (#)
14
Reuse (#)
20
27
0
7
14
20
27
Reuse (#)
Fig. 2 e Frequency of bottle reuse and UV exposure duration interaction plots for Br leaching from PET (linear model) and PC (quadratic model) bottles at three different temperatures. Circles, triangles and rectangles stand for experimentally obtained CCD design points, model predicted 12 h UV exposure points, and no UV exposure points, respectively. Black and red lines represent UV exposure (h) at their low (no UV) and high (16 h UV exposure) levels from the CCD. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
Antimony concentrations in PC bottles were remarkably lower than those leaching from PET, since Sb is not used in the polymerization process of PC, but could be perhaps added in minor quantities as a flame retardant (Albemarle, 2011). The highest concentrations of leached Br were observed for PET bottles subject to SODIS treatment. No effect of the tested factors used by SODIS or the worst-case scenario influenced leaching of Br from PC bottles, remaining unaffected at w7e8 mg L1 (Table 2). A similar, but not identical parameter of chemical leaching to frequency of bottle reuse is storage time, since storage time refers to storing a bottle with the same liquid for a long time. Storage time has been shown to affect chemical leaching from plastic bottles, primarily PET (Shotyk et al., 2006; Shotyk and Krachler, 2007; Andra et al., 2012). Storage of PET bottles for three months at room temperature resulted in considerable Sb leaching, reaching average concentrations of 630 ng L1 from 360 ng L1 at time zero analysis (Shotyk et al., 2006). Keresztes et al. (2009) reported similar storage effect on Sb leaching from non-carbonated PET bottles from 100 ng L1 (after 10 days) increased to 950 ng L1 after 950 days at room temperature. Increasing both storage time and temperature resulted in considerable increase in leached Sb concentrations
from PET bottles (Keresztes et al., 2009). In specific, storage at 60 C for 72 h resulted in Sb leaching of up to 2000 ng L1 (Keresztes et al., 2009). Interestingly, a specific PET brand used by Westerhoff et al. (2008) leached >2400 ng Sb L1 at 45 C under UV exposure for 6 h, much higher than the Sb concentrations encountered by our PET bottles. Differences in the magnitude of Sb leaching could be ascribed to differences in purity of plastic material (recycled or not) and/or differences in the bottled water class, i.e., non-carbonated, or carbonated, or enriched with vitamin or color (Andra et al., 2012). Enriched bottled water along with carbonated bottled PET bottles leached significantly higher Sb concentrations than the typical non-carbonated bottled water (Andra et al., 2012). Frequency of reuse has never been tested before with respect to its influence on chemical leaching from drinkingwater plastic containers. Our study showed that frequency of reuse exerted greater influence on Sb and Br leaching from PET and PC containers when compared with that attributed to UV exposure duration and temperature. This finding comes in contrast with earlier work on plasticizer leaching from PET bottles subject to SODIS procedure that were reused, but they did not report significant increase in plasticizer leaching from
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 7 7 e6 6 8 7
6683
Fig. 3 e Relationship between measured Sb and Br concentrations in bottled water from PET and PC bottles under the study conditions. ** and *** represents probability of Pearson’s correlation coefficient less than 0.01 and 0.001, respectively.
reused containers; the authors did not mention how many times containers were reused (Schmid et al., 2008). In certain countries e for example India, the big 19 L PC water containers are reused for about 6 to 12 months until they are deformed beyond visual acceptance and/or break beyond repair. During this period these bottles will undergo up to 100 cycles of refilling by bottling companies (personal communication, Ram Das, Sri Krishna Bottling Company, Delhi, India). Figure SI-1 depicts changes in physical appearance and condition of these bottled water containers that hints to an accelerated constituent leaching from plastic walls into contained water. Our new findings come handy with the recently proposed public health goal (PHG) of 0.7 mg Sb L1 in drinking-water by California’s Office for Environmental Health and Hazard Assessment (OEHHA, 2009). The proposed PHG underlines current attempts in revisiting Sb critical human health effects
in conjunction with increasing human exposure to Sbcontaining food and water packaging materials. In our study, water consumption of 3.08 L day1 person1 and an average body mass of 70 kg were used in calculating daily and cumulative Sb and Br intakes, per OEHHA suggestions (OEHHA, 2009). The cumulative average daily Sb intake due to leaching from PET and PC bottles that were reused for at least 27 times were 100 and 6.3 ng kg1 day1, respectively, (Fig. 4). Minimal Sb leaching from PC bottles occurred as a function of bottle reuse, possibly due to the magnitude of concentrations being close to the limit of quantification (1 ng L1). In specific, Sb leaching increased up to 13 cycles of PC bottle reuse (w8 ng L1) and remained almost unchanged thereafter, while Sb leaching from PET continued to increase even after 27 cycles of bottle reuse (w230 ng L1) (Fig. 4). Current regulatorybased tolerable daily intake (TDI) estimate for Sb is 6 mg kg1 body weight (WHO, 2003), which is much higher than our
6684
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 7 7 e6 6 8 7
Fig. 4 e Antimony leaching and cumulative daily intake values as a function of bottle reuse. Circle data points refer to Sb cumulative daily intake (second y-axis). Each reuse number (x-axis) indicates one cycle of deionized-water filled bottle exposure to UV-A radiation for 1 h at 50 C, being equivalent to one SODIS cycle (SANDEC, 2002).
estimate of Sb daily intake under SODIS conditions (Fig. 4). Migrated Sb concentrations from PET and PC bottles did not pose a serious risk to human health per the existing MCL value of 6 mg Sb L1 in drinking water that corresponds to an acceptable daily dose of 1.1 mg kg1 day1 (OEHHA, 1997). This suggests that under our experimental conditions, leached Sb into bottled water despite repeated bottle reuse may not be a major human health consideration based on current MCL value of Sb for drinking-water. However, Sb MCL is currently under revision by regulatory agencies based on new findings, such as, recent in-vitro studies call for increased proliferation of breast cancer cells in the presence of metalloestrogens, like Sb (Choe et al., 2003). In addition, in-vitro rapid tests evaluating biologic activity of bottled water call for increased genotoxicity (Ceretti et al., 2010; Ubomba-Jaswa et al., 2010) and estrogenicity (Pinto and Reali, 2009; Wagner and Oehlmann, 2009, 2010). Contrasting results on the magnitude of estrogenic activity in bottled water (Wagner and Oehlmann, 2009, 2010; Pinto and Reali, 2009; Stanford et al., 2010) may be ascribed to differences in study design, PET material, raw water composition prior to bottling, or test sensitivity (Sax, 2010). Similarly, the cumulative average daily Br intake due to leaching from PET and PC bottles that were reused for at least 27 times were 4.3 and 7.8 mg kg1 day1, respectively (Fig. 5). Bromine leaching increased up to 22 cycles of both PET and PC bottle reuse (w7 and w10 ng L1, respectively) and remained almost unchanged thereafter (Fig. 5). Total soluble Br concentrations leaching to bottled water were not considered toxic, since Br is not considered a primary water contaminant. Toxicity associated with Br presence in drinking water containers is most likely ascribed to organo-brominated compounds such as PBDE or TBBPA, used as flame retardants in polymeric applications of PET and PC (Albemarle, 2011; CDS Analytical, 2010; Talsness et al., 2009). Assuming all of the migrated Br in the form of bromide, our measured Br intake estimates were below the acceptable daily Br intake value of 0.4 mg kg1 day1 (EMEA, 1997). No toxicological-based health
Table 2 e Predicted leaching responses of Sb and Br into bottled water using the central composite design under various scenarios relevant to SODIS treatment. Antimony and Br leaching prediction models were based on the central composite design. Limit of detection for Sb and Br were 0.5 ng LL1 and 0.1 mg LL1, respectively. A [ UV Exposure (h); B [ Temperature ( C); C [ Bottle Reuse (#). Criteria
Conditions
Response
UV Exposure (h)
Temperature ( C)
Bottle reuse (#)
PET_Sb (ng L1)
PC_Sb (ng L1)
PET_Br (mg L1)
PC_Br (mg L1)
0 6 6 12
20 30 30 47
0 0 20 27
3 5 9 14
2 7 8 8
Minimum leaching SODIS simulation-1 SODIS simulation-2 Worst-case scenario
R2
Adjusted R2
-1
1. PET_Sb (ng L ) 2. PC_Sb (ng L-1)
0.62 0.88
0.55 0.83
0.36 0.51
3. PET_Br (mg L-1) 4. PC_Br (mg L-1)
0.51 0.82
0.42 0.66
0.12 0.32
Analyte
Prediction R2
Prediction equation PET_Sb ¼ 83.52 þ 7.99 * A þ 3.57 * B þ 7.24 * C PC_Sb ¼ 5.86 þ 1.04 * A þ 0.16 * B þ 0.21 * C e 0.01 * A * B 0.03 * A * C þ 0.01* B * C PET_Br ¼ 0.99 þ 0.25 * A þ 0.09 * B þ 0.20 * C PC_Br ¼ 15.7 þ 1.13 * A þ 1.26 * B þ 0.58 * C e 0.05 * A* A e 0.02 * B * B e 0.01* C * C e 0.01 * A * B e 0.01* A * C e 0.01* B * C
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 7 7 e6 6 8 7
Fig. 5 e Bromine leaching and cumulative daily Br intake values as a function of bottle reuse. Circle data points refer to cumulative daily Br intake values (second y-axis). Each reuse number (x-axis) indicates one cycle of deionizedwater filled bottle exposure to UV-A radiation for 1 h at 50 C, being equivalent to one SODIS cycle (SANDEC, 2002).
risk criteria exist for organo-brominated compounds, if the plastics leachants into bottled water were to include organicbrominated compounds that present potential health concerns at even lower levels. Our most recent work has qualitatively confirmed the presence of BDE-209 in bottled water, but additional work is warranted to quantify leached concentrations of BDE-209 (Andra et al., 2012).
4.
6685
health risk based upon existing MCL value (5 or 6 mg L1). Antimony leaching from PET seems to be the most concerning observation, while the mechanism(s) and health implications behind leaching of soluble Br (associated with organobromine compounds, such as, PBDE) are yet to be determined. Our cost-effective experimental approach in studying main and interactive effects of three major tested factors influencing chemical leaching enabled us classifying the studied factors according to their contribution to chemical leaching from frequency of reuse > UV exposure duration > temperature. This study illustrated the importance of a so far unaccounted factor, i.e., frequency of bottle reuse in everyday activities that could enhance substantial leaching of plastic constituents into packaged water. Bottle reuse represents an everyday scenario for either household owner applying SODIS to reused PET bottles, or those consuming drinking water from reused PC large 19 L bottles supplied by bottling water companies. SODIS is a method typically used by developing countries, while drinking-water consumption from reused 19 L PC containers steadily gains popularity among both developed and developing countries. Our study illustrated the importance of a so far underestimated factor of deteriorating bottled water quality, e.g., that of frequency of bottle reuse, which in certain countries may reach as high as 100 times (personal communication, Ram Das, Sri Krishna Bottling Company, Delhi, India). Leaching of plastic constituents other than Sb and Br remains to be investigated for reused water containers. Cost-benefit analyses may soon be needed to tackle effectiveness of recycling plastic bottles in light of our findings calling for increased health risk associated with reused bottles. It is currently unknown whether frequent recycling of water packaging materials (smaller frequency of bottle reuse) as advocated by this study’s findings is supported by sustainability metrics of energy and economic aspects. Additional research is warranted on evaluating the effect of the three major tested conditions encountered during solar disinfection on other chemicals associated with the polymeric structure of the drinking water containers. It is warranted that further studies need to focus on organobrominated compounds instead of total soluble Br as a better indicator of neurotoxic and highly potent PBDE leaching into bottled water. Such chemicals under investigation are plasticizer compounds like phthalates, BPA, alkyl phenols and PBDE, i.e., decabromodiphenyl ether congener 209 (Andra et al., 2012) that are currently under investigation in our laboratory.
Conclusions
Solar disinfection of microbial-contaminated water in plastic containers at the household level represents a cost-effective intervention measure in reducing total social cost associated with mortality and morbidity burden of water-borne disease in developing countries. Concerns related to chemical leaching from plastic containers subject to solar disinfection were not validated, based on our Sb and Br leaching data. Antimony leaching in water was primarily enhanced with extending frequency of reuse, but not to levels posing serious public
Funding Partial support for this work was provided by the HarvardCyprus program.
Competing interests None declared.
6686
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 7 7 e6 6 8 7
Acknowledgments \We would like to thank Mr. Nicola Lupoli, Drs. Innocent Jayawardene and Chitra Amarasiriwardena at the Trace Metals Laboratory, Harvard School of Public Health for providing assistance and guidance with the ICP-MS.
Appendix. Supplementary data Supplementary data related to this article can be found online at doi:10.1016/j.watres.2011.10.001.
references
Albemarle, 2011. Brominated Flame Retardants. Albemarle’s Fire Safety Solutions Product Selector. Available at: http://www. albemarle.com/Products-and-Markets/Polymer-Solutions/ Fire-Safety-373Solutions/Brominated-Flame-Retardants-183. html (accessed 24.08.11.). Amiridou, D., Voutsa, D., 2011. Alkylphenols and phthalates in bottled waters. Journal of Hazardous Materials 185 (1), 281e288. Andra, S.S., Makris, K.C., Shine, J.P., Lu, C.S., 2012. Co-leaching of brominated compounds and antimony from bottled water. Environment International 38 (1), 45e53. Barker, T.B., 1985. Quality by Experimental Design. Marcel Dekker, New York. Biscardi, D., Monarca, S., De Fusco, R., Senatore, F., Poli, P., Buschini, A., Rossi, C., Zani, C., 2003. Evaluation of the migration of mutagens/carcinogens from PET bottles into mineral water by Tradescantia/micronuclei test, Comet assay on leukocytes and GC/MS. The Science of Total Environment 302 (1e3), 101e108. Bosnir, J., Puntaric, D., Galic, A., Skes, I., Dijanic, T., Klaric, M., Grgic, M., Curkovic, M., Smit, Z., 2007. Migration of phthalates from plastic containers into soft drinks and mineral water. Food Technology and Biotechnology 45 (1), 91e95. Cao, X.L., Corriveau, J., 2008. Migration of bisphenol A from polycarbonate baby and water bottles into water under severe conditions. Journal of Agricultural and Food Chemistry 56 (15), 6378e6381. Casajuana, N., Lacorte, S., 2003. Presence and release of phthalic esters and other endocrine disrupting compounds in drinking water. Chromatographia 57 (9e10), 649e655. CDS Analytical, Inc., 2010. CDSolutions: applications information using advanced sample handling technology. Brominated Flame Retardant in PET Available at: http://www. cdsanalytical.com/appnotes/104.pdf (accessed 24.08.11.). Ceretti, E., Zani, C., Zerbini, I., Guzzella, L., Scaglia, M., Berna, V., Donato, F., Monarca, S., Feretti, D., 2010. Comparative assessment of genotoxicity of mineral water packed in polyethylene terephthalate (PET) and glass bottles. Water Research 44 (5), 1462e1470. Choe, S.Y., Kim, S.J., Kim, H.G., Lee, J.H., Choi, Y., Lee, H., Kim, Y., 2003. Evaluation of estrogenicity of major heavy metals. Science of the Total Environment 312 (1), 15e21. Covaci, A., Harrad, S., Mohamed Abdallah, M.A.E., Ali, N., Law, R.J., Herzke, D., de Wit, C.A., 2011. Novel brominated flame retardants: a review of their analysis, environmental fate and behaviour. Environment International 37 (2), 532e556. Criado, M.V., Pinto, V.E.F., Badessari, A., Cabral, D., 2005. Conditions that regulate the growth of moulds inoculated into bottled mineral water. International Journal of Food Microbiology 99 (3), 343e349.
EMEA, 1997. Bromide, Sodium Salt. Committee for Veterinary Products. EMEA/MRL/182/97:1e3. European Agency for the Evaluation of Medicinal Products, London. Available at:. http:// www.ema.europa.eu/docs/en_GB/document_library/ Maximum_Residue_Limits_-413_Report/2009/11/ WC500011113.pdf (accessed 24.08.11.). EU Council Directive 98/83/EC, 1998. Available at: http://eurLex. europa.eu/LexUriServ/416LexUriServ.do?uri=CELEX: 31998L0083:EN:HTML (accessed 23.08.11.). Ezzati, M., Lopez, A.D., Rodgers, A., Vander Hoorn, S., Murray, C.J. L., 2002. Selected major risk factors and global and regional burden of disease. Lancet 360 (9343), 1347e1360. Farfal, C.P., Pin˜eiro, A.M., Barrera, A.B., Barrera, P.B., Cancino, H.P., Henrı´quez, L.G., 2004. Use of enzymatic hydrolysis for the multi-element determination in mussel soft tissue by inductively coupled plasma-atomic emission spectrometry. Talanta 64 (3), 671e681. Graf, J., Togouet, S.Z., Kemka, N., Niyitegeka, D., Meierhofer, R., Pieboji, J.G., 2010. Health gains from solar water disinfection (SODIS): evaluation of a water quality intervention in Yaounde, Cameroon. Journal of Water and Health 8 (4), 779e796. International Bottled Water Association (IBWA), 2011. Bottled Water Facts: IBWA Corrects EWG’S Faulty Report. Bottled Water Reporter, February/March 2011 issue. Available at: http://www.bottledwater.org/news/bottled-water-labelsscorecard-environmental-working-group-fda429epa-tapwater-spring-water-purified-water-food (accessed 24.08.11.). Jalbani, N., Kazi, T.G., Arain, B.M., Jamali, M.K., Afridi, H.I., Sarfraz, R.A., 2006. Application of factorial design in optimization of ultrasonic-assisted extraction of aluminum in juices and soft drinks. Talanta 70 (2), 307e314. Keresztes, S., Tatar, E., Mihucz, V.G., Virag, I., Majdik, C., Zaray, G., 2009. Leaching of antimony from polyethylene terephthalate (PET) bottles into mineral water. Science of the Total Environment 407 (16), 4731e4735. Krachler, M., Shotyk, W., 2009. Trace and ultratrace metals in bottled waters: survey of sources worldwide and comparison with refillable metal bottles. Science of the Total Environment 407 (3), 1089e1096. Le, H.H., Carlson, E.M., Chua, J.P., Belcher, S.M., 2008. Bisphenol A is released from polycarbonate drinking bottles and mimics the neurotoxic actions of estrogen in developing cerebellar neurons. Toxicology Letters 176 (2), 149e156. Leivadara, S.V., Nikolaou, A.D., Lekkas, T.D., 2008. Determination of organic compounds in bottled waters. Food Chemistry 108 (1), 277e286. Lo´pez, R., Gon˜i, F., Etxandia, A., Milla´n, E., 2007. Determination of organochlorine pesticides and polychlorinated biphenyls in human serum using headspace solid-phase microextraction and gas chromatography-electron capture detection. Journal of Chromatography B 846 (1e2), 298e305. Makris, K.C., Quazi, S., Nagar, R., Sarkar, D., Datta, R., Sylvia, V.L., 2008. In vitro model improves the prediction of soil arsenic bioavailability: worst-case scenario. Environmental Science and Technology 42 (16), 6278e6284. Mason, R.L., Gunst, R.F., Hess, J.L., 1989. Statistical Design and Analysis of Experiments with Applications to Engineering and Science. John Wiley, New York. Meierhofer, R., Landolt, G., 2009. Factors supporting the sustained use of solar water disinfection e Experiences from a global promotion and dissemination programme. Desalination 248 (1e3), 144e151. Mingwu, S., Chao, W., Yongjuan, J., Xinhua, D., Xiang, F., 2010. Determination of selected polybrominated diphenyl ethers and polybrominated biphenyl in polymers by ultrasonicassisted extraction and high-performance liquid chromatography-inductively coupled plasma mass spectrometry. Analytical Chemistry 82 (12), 5154e5159.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 7 7 e6 6 8 7
OEHHA, 1997. Public Health Goal for Antimony in Drinking Water. Office of Environmental Health Hazard Assessment. California Environmental Protection Agency, Oakland/Sacramento, CA. Available at: http://oehha.ca.gov/water/phg/pdf/anti3_c.pdf (accessed 24.08.11). OEHHA, 2009. Public Health Goal for Antimony in Drinking Water (Draft). Office of Environmental Health Hazard Assessment. California Environmental Protection Agency, Oakland/ Sacramento, CA. Available at: http://oehha.ca.gov/water/phg/ pdf/PHGAntimonyD072309.pdf (accessed 23.08.11.). Pinto, B., Reali, D., 2009. Screening of estrogenLike activity of mineral water stored in PET bottles. International Journal of Hygiene and Environmental Health 212 (2), 228e232. du Preez, M., McGuigan, K.G., Conroy, R.M., 2010. Solar disinfection of drinking water in the prevention of dysentery in South African children aged under 5 years: the role of participant motivation. Environmental Science and Technology 44 (22), 8744e8749. Rodwan Jr., J.G., 2010. Bottled Water 2009. International Bottled Water Association. Available at: http://www.bottledwater.org/ files/2009BWstats.pdf (accessed 23.08.11.). SANDEC, 2002. Solar Water Disinfection: A Guide for the Application of SODIS. SANDEC Report No 06/02. Available at: http://www.sodis.ch/methode/anwendung/ ausbildungsmaterial/dokumente_material/manual_e.pdf (accessed 23.08.11.). Sax, L., 2010. Polyethylene terephthalate may yield endocrine disruptors. Environmental Health Perspectives 118 (4), 445e448. Schmid, P., Kohler, M., Meierhofer, R., Luzi, S., Wegelin, M., 2008. Does the reuse of PET bottles during solar water disinfection pose a health risk due to the migration of plasticisers and other chemicals into the water? Water Research 42 (20), 5054e5060. Shotyk, W., Krachler, M., 2007. Contamination of bottled waters with antimony leaching from polyethylene terephthalate (PET) increases upon storage. Environmental Science and Technology 41 (5), 1560e1563. Shotyk, W., Krachler, M., Chen, B., 2006. Contamination of Canadian and European bottled waters with antimony from PET containers. Journal of Environmental Monitoring 8 (2), 288e292. Sommer, B., Marino, A., Solarte, Y., Salas, M.L., Dierolf, C., Valiente, C., Mora, D., Rechsteiner, R., Setter, P., Wirojanagud, W., Ajarmeh, H., AlHassan, A., Wegelin, M., 1997. SODIS e An emerging water treatment process. Journal of Water Supply Research and Technology e Aqua 46 (3), 127e137.
6687
Stanford, B.D., Snyder, S.A., Trenholm, R.A., Holady, J.C., Vanderford, B.J., 2010. Estrogenic activity of US drinking waters: a relative exposure comparison. Journal of American Water Works Association 102 (11), 55e65. Talsness, C.E., Andrade, A.J.M., Kuriyama, S.N., Taylor, J.A., vom Saal, F.S., 2009. Components of plastic: experimental studies in animals and relevance for human health. Philosophical Transactions of the Royal Society of London. Series B, Biological Sciences 364 (1526), 2079e2096. Ubomba-Jaswa, E., Fernandez-Ibanez, P., McGuigan, K.G., 2010. A preliminary Ames fluctuation assay assessment of the genotoxicity of drinking water that has been solar disinfected in polyethylene terephthalate (PET) bottles. Journal of Water and Health 8 (4), 712e719. Vidal, L., Psillakis, E., Domini, C.E., Grane´, N., Marken, F., Canals, A., 2007. An ionic liquid as a solvent for headspace single drop microextraction of chlorobenzenes from water samples. Analytica Chimica Acta 584 (1), 189e195. Wagner, M., Oehlmann, J., 2009. Endocrine disruptors in bottled mineral water: total estrogenic burden and migration from plastic bottles. Environmental Science and Pollution Research 16 (3), 278e286. Wagner, M., Oehlmann, J., 2010. Endocrine disruptors in bottled mineral water: estrogenic activity in the E-Screen. Journal of Steroid Biochemistry and Molecular Biology. doi:10.1016/j. jsbmb.2010.10.007. Wegelin, M., Canonica, S., Alder, A.C., Marazuela, D., Suter, M.J.F., Bucheli, T.D., Haefliger, O.P., Zenobi, R., McGuigan, K.G., Kelly, M.T., Ibrahim, P., Larroque, M., 2001. Does sunlight change the material and content of polyethylene terephthalate (PET) bottles? Journal of Water Supply Research and Technology e Aqua 50 (3), 125e133. Westerhoff, P., Prapaipong, P., Shock, E., Hillaireau, A., 2008. Antimony leaching from polyethylene terephthalate (PET) plastic used for bottled drinking water. Water Research 42 (3), 551e556. World Health Organization, 2003. Antimony in Drinking-water. Background Document for Development of WHO GUIDELINES for Drinking-water Quality. Geneva, Switzerland. Available at: http://www.who.int/water_sanitation_health/dwq/chemicals/ antimony.pdf (accessed 24.08.11.). Yusa`, V., Pardo, O., Pastor, A., Guardia, M., 2006. Optimization of a microwave-assisted extraction large-volume injection and gas chromatography-ion trap mass spectrometry procedure for the determination of polybrominated diphenyl ethers, polybrominated biphenyls and polychlorinated naphthalenes in sediments. Analytica Chimica Acta 557 (1e2), 304e313.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 8 8 e6 6 9 8
Available online at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Sulfur transformations in pilot-scale constructed wetland treating high sulfate-containing contaminated groundwater: A stable isotope assessment Shubiao Wu a, Christina Jeschke b, Renjie Dong c,*, Heidrun Paschke d, Peter Kuschk e, Kay Kno¨ller b a
Key Laboratory of Agricultural Engineering in Structure and Environment of Ministry of Agricultural, College of Water Conservancy & Civil Engineering, China Agricultural University, 100083 Beijing, PR China b Department of Catchment Hydrology, Helmholtz Centre for Environmental Research - UFZ, Theodor-Lieser-Strasse 4, Halle D-06120, Germany c College of Engineering, China Agricultural University, 100083 Beijing, PR China d Department of Groundwater Remediation, Helmholtz Centre for Environmental Research - UFZ, Permoserstrasse 15, Leipzig D-04318, Germany e Department of Environmental Biotechnology, Helmholtz Centre for Environmental Research - UFZ, Permoserstrasse 15, Leipzig D-04318, Germany
article info
abstract
Article history:
Current understanding of the dynamics of sulfur compounds inside constructed wetlands is
Received 17 June 2011
still insufficient to allow a full description of processes involved in sulfur cycling. Experiments
Received in revised form
in a pilot-scale horizontal subsurface flow constructed wetland treating high sulfate-
30 September 2011
containing contaminated groundwater were carried out. Application of stable isotope
Accepted 7 October 2011
approach combined with hydro-chemical investigations was performed to evaluate the sulfur
Available online 19 October 2011
transformations. In general, under inflow concentration of about 283 mg/L sulfate sulfur, sulfate removal was found to be about 21% with a specific removal rate of 1.75 g/m2$d. The
Keywords:
presence of sulfide and elemental sulfur in pore water about 17.3 mg/L and 8.5 mg/L, respec-
Constructed wetland
tively, indicated simultaneously bacterial sulfate reduction and re-oxidation. 70% of the
Bacterial sulfate reduction
removed sulfate was calculated to be immobilized inside the wetland bed. The significant
Sulfide re-oxidation
enrichment of 34S and 18O in dissolved sulfate (d34S up to 16&, compared to average of 5.9& in
Stable isotopes
the inflow, and d18O up to 13&, compared to average of 6.9& in the inflow) was observed clearly correlated to the decrease of sulfate loads along the flow path through experimental wetland bed. This enrichment also demonstrated the occurrence of bacterial sulfate reduction as well as demonstrated by the presence of sulfide in the pore water. Moreover, the integral approach shows that bacterial sulfate reduction is not the sole process controlling the isotopic composition of dissolved sulfate in the pore water. The calculated apparent enrichment factor (3 ¼ 22&) for sulfur isotopes from the d34S vs. sulfate mass loss was significantly smaller than required to produce the observed difference in d34S between sulfate and sulfide. It indicated some potential processes superimposing bacterial sulfate reduction, such as direct reoxidation of sulfide to sulfate by oxygen released from plant roots and/or bacterial disproportionation of elemental sulfur. Furthermore, 41% of residual sulfate was calculated to be
* Corresponding author. Tel.: þ86 10 62737852; fax: þ86 10 62737885. E-mail address:
[email protected] (S. Wu). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.10.008
6689
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 8 8 e6 6 9 8
from sulfide re-oxidation, which demonstrated that the application of stable isotope approach combined with the common hydro-chemical investigations is not only necessary for a general qualitative evaluation of sulfur transformations in constructed wetlands, but also leads to a quantitative description of intermediate processes. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
Since the 1980s, constructed wetlands as an alternative ecological technology for wastewater treatment have been increasingly used for remediation of various contaminated waters (Bulc, 2006; Braeckevelt et al., 2008; Garcia et al., 2010). However, despite the successful application, the knowledge of the removal processes inside the systems is still insufficient, because of the variability of the redox states in the near-root zones and the complex interactions of different microbial transformations (Faulwetter et al., 2009). In the wetland bed, the spatial and temporal micro-scale gradients of oxygen concentrations and redox states established close to root surfaces enable the development of microbial biofilms of functionally different microorganisms. Those microorganisms can simultaneously mediate processes such as nitrification, denitrification, mineralization of organic carbon, methanogenesis, sulfate reduction, and sulfide oxidation on a small spatial scale (Lee et al., 1999; Holmer and Storkholm, 2001). As sulfate is a common constituent of wastewaters, different processes of sulfur cycling, depending on the availability of organic carbon and/or oxygen, are accordingly prevalent in constructed wetlands (Sturman et al., 2008; Wiessner et al., 2010). The wetland system can influence the sulfur cycling by e.g. releasing organic carbon compounds and/or oxygen from the plant roots to enhance the sulfate reduction or re-oxidation of the reduced sulfur compounds. Moreover, the processes of sulfur transformations, such as sulfate reduction can also influence the conditions for the biochemical processes, changing the pH and redox conditions (Leon et al., 2002; Geurts et al., 2009). In addition, the nitrogen removal and other removal processes under sulfur cycling can be influenced as well due to the toxicity of the reduced sulfur compounds like H2S, as well as competition for oxygen (Aesoy et al., 1998; Stein et al., 2007). Previous studies of sulfur transformations focus mainly on the removal of metals by precipitation with sulfide in constructed wetlands, particularly in treating acid mine drainage (Webb et al., 1998; Machemer et al., 1993; Woulds and Ngwenya, 2004), and some focus on negative effects caused by sulfide toxicity on wetland plants and microorganisms (Aesoy et al., 1998; Wiessner et al., 2010; Armstrong et al., 1996). However, the knowledge of evaluation on the dynamics of sulfur cycling in the constructed wetlands, particularly the quantitative description of intermediate turnover is still limited (Choi et al., 2006; Faulwetter et al., 2009). Moreover, the quantification of microbial sulfur transformations by concentration pattern of reactant (SO2 4 ) consumption or product formation (S2) in aquifers may be obscured by concurrent abiotic transformations, e.g. dilution due to the rainfall, concentration resulted from high
evapotranspiration, matrix effects, and mineral precipitation such as precipitation of gypsum and sulfide with metals (Anderson and Lovley, 2000; Schroth et al., 2001; Kno¨ller et al., 2006; Rahman et al., 2008). Furthermore, the coexisting of biotic processes, such as bacterial dissimilatory sulfate reduction and sulfide re-oxidation as well as disproportionation of different reduced sulfur compounds under specific micro gradients, also makes the understanding of different turnover processes of sulfur species inside the constructed wetlands be insufficient (Finster et al., 1998; Liesack et al., 2000). As a tool to discern microbial activity from abiotic transformations and assess pathways and rates of sulfur transformations, stable isotope analyses have been found increasingly applied in recent years (Schroth et al., 2001; Kno¨ller et al., 2004, 2008). Microbial sulfate reduction usually results in significant isotope enrichment of 34S in residual sulfate coupled to a depletion of 34S in produced sulfide (Rees, 1973; Fry et al., 1988; Bottrell et al., 1995). Sulfur isotope fractionation appears to be a valuable indicator for microbial sulfate reduction in complex environments. Thus, sulfur isotope fractionation in groundwater was previously observed in forest hydrological studies (Robertson and Schiff, 1994; Alewell and Giesemann, 1996) as well as in contaminated aquifers (Kno¨ller et al., 2006, 2008). Presently, little is know about sulfur isotope fractionation in constructed wetlands with complex coexisting reductive and oxidative conditions. In this study, evaluation of application of stable isotope investigation combined with common hydro-chemical examination in pilot-scale constructed wetland treating high sulfate-containing contaminated groundwater was conducted for: 1) the identification of microbial sulfate reduction; 2) recognition of further bacterial sulfur transformations superimposing sulfate reduction, such as disproportionation of reduced sulfur compounds and re-oxidation of sulfide; 3) the impact of wetland plants on sulfur cycling which was facilitated by the oxygen and organic carbon compounds released from plant roots; 4) quantitative assessment of reoxidation of sulfide to sulfate.
2.
Materials and methods
2.1.
Site description and experimental setup
The experimental pilot-scale constructed wetland was built at the SAFIRA research site in Bitterfeld, Germany in 2003 (Braeckevelt et al., 2008). The local groundwater used for loading this system during this experiment contained monochlorobenzene (MCB) as the main organic compound with concentration of 6e12 mg/L and sulfate (710e920 mg/L) as the
6690
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 8 8 e6 6 9 8
main inorganic compound. The constructed wetland was designed to operate in a horizontal subsurface flow mode with dimension of 6 m length 1 m width and was filled to an average depth of 0.5 m with autochthonous quaternary aquifer material consisting predominantly of about 25% mica sand and about 67% gravel (porosity 0.35). The wetland bed was planted with common reed (Phragmites australis). The loading contaminated groundwater from a well installed in 22 m depth was pumped up and supplied continuously at a flow rate of 5.0 L/h to the wetland bed. The water level was maintained at approximate 0.1 m below the surface of the wetland. The samples took from 4 m (distance from inlet) were defined as effluent and the theoretical hydraulic retention time in the wetland bed was about 6 days. The outflow volumes were measured by a flow meter in order to determine water loss and allow the calculation of load removal. The system had been run with the same groundwater for 7 years before the current investigation was carried out from March to November in 2010.
2.2.
Sampling
Pore water samples for the evaluation of hydro-chemical parameters were taken from the wetlands at 0.5, 1, 2, 3 and 4 m distance from the inflow at 0.3, 0.4 and 0.5 m depth of the bed. Inflow samples were taken from the feeding pipe. The 18 2 isotope fractionation samples for d34SeSO2 4 , d OeSO4 were added 5% Zn-Acetate solution to removal the interference of sulfide. The samples for d34S in sulfide were only taken from 0.5 m depth of the bed and precipitated with 5% Zn-Acetate solution. Water sampling at each depth was carried out using stainless steel lancets (3.5 mm inner diameter) and peristaltic pumps at a rate of 78 ml/min. Water samples were stored without headspace at 4 C until analysis.
2.3.
Physical-chemical analysis
The redox potential was determined in the field using a SenTix ORP electrode (WTW, Weilheim, Germany) and the temperature was measured with a temperature sensor (PT 1000, PreSens, Regensburg, Germany). Sample filtration was carried out using a 5 mm-syringe filter (Ministart NML, Sartorius) for particles removal before the quantitative ions analysis. The pH value was measured by a SenTix41 electrode with pH 537 Microprocessor (WTW, Weilheim, Germany). For the conservation of Fe(II) samples, hydrochloric acid was added and after derivatisation with ferrocin, photometric measurement was carried out at 562 nm using a Cadas 100 photometer (Hach Lange, Dusseldorf, Germany) (Lovley and Phillips, 1986). Sulfate concentrations were determined by ion chromatography (DX 500) with an IonPacAG11 (4 250 mm) column (Dionex Corporation, Sunnyvale, USA) and conductivity detection (CD 20). Sulfide concentrations were measured by photometric method using Test kit LCW053 from HACH LANGE, Germany. Elemental sulfur in the pore water was estimated according to Rechmeier et al. (Rethmeier et al., 1997), by extracting samples with chloroform and subsequent detection by HPLC (Beckman, USA) using a Li-Chrospher 100, RP 18 column (5 mm, Merck, Germany) equipped with a UV-detector at 263 nm.
2.4.
Isotope analysis
Sulfur isotope analysis was conducted according to Kno¨ller and Schubert (2010). The precipitated ZnS was removed by filtration (0.45 mm). After adding concentrated hydrochloric acid in the laboratory, the hydrogen sulfide was stripped with N2 gas and then trapped as ZnS in Zn-acetate solution. The precipitated ZnS was subsequently converted to Ag2S by addition of 0.1 M AgNO3 solution. Dissolved sulfate was recovered by precipitation as BaSO4 at 70 C after the pH of the solution was adjusted to 3.0 and BaCl2 solution was added. Sulfur isotopic compositions were measured after conversion of BaSO4 (or Ag2S) to SO2 using an elemental analyzer (continuous flow flash combustion technique) coupled with an isotope ratio mass spectrometer (delta S, ThermoFinnigan, Bremen, Germany). Sulfur isotope measurements were performed with an analytical error of better than 0.3& and results are reported in delta notation (d34S) as part per thousand (&) deviations relative to the Vienna Can˜on Diablo Troilite (VCDT) standard (according to general Eq. (1)) dð&Þ ¼
Rsample Rstandard
Rstandard 1000
(1) 34
32
where R is the ratio of the heavy to light isotopes (e.g. S/ S or 18O/16O). Oxygen isotope analysis with barium sulfate samples was carried out by high temperature pyrolysis at 1450 C in a TC/EA connected to a delta plus XL mass spectrometer (ThermoFinnigan, Bremen, Germany) with an analytical error of better than 0.5&. According to Eq. (1), results of oxygen isotope measurements are expressed in delta notation (d18O) as part per thousand (&) deviations relative to Vienna Standard Mean Ocean Water (VSMOW). For normalizing the d34S data, the IAEA-distributed reference materials NBS 127 (BaSO4) and IAEA-S1 (Ag2S) were used. The assigned values were þ20.3& (VCDT) for NBS 127 and 0.3& (VCDT) for IAEA-S1. The normalization of oxygen isotope data of sulfate was carried out using the reference material NBS 127 with an assigned d18O value of þ8.7& (VSMOW).
2.5.
Calculation
Water loss generally occurs in constructed wetlands via evaporation from the filter surface and transpiration by plants, which is combined to be called evapotranspiration. The area specific water loss (ΔV) during a defined period is calculated by measuring the influent and effluent streams as well as rainfall, DV ¼ fðVin þ Prain A Vout Þ=Vin g 100
(2)
where ΔV is the water loss by evapotranspiration in %, Vin and Vout are the influent and effluent volumes in L/d, and Prain is the precipitation, which was measured by a weather station near the SAFIRA site and its amount was related to specific area in L/m2d. For the evaluation of treatment performance of constructed wetlands, the consideration of water loss and load calculation is necessary. The following assumptions were made: 1) Evapotranspiration along the wetland flow path increases in a linear way
6691
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 8 8 e6 6 9 8
2) An simplified ideal plug flow though the soil filter exists. Therefore, the load of the contaminants in the defined sampling points can be expressed as Eq. (3) and the residual fraction of the contaminants through the bed (e.g. 4 m sampling point) was defined as Eq. (4). Load4m ¼ ðVin þ Prain A4m DV4m Vin Þ C4m
(3)
Residual fraction ð%Þ ¼ fðVin þ Prain A4m DV4m Vin Þ C4m g=Vin Cin
(4)
where Cin is the influent concentration in mg/L, C4m is the concentration at 4 m sampling point in mg/L, and A4m is the area from the inflow to the sampling point ¼ 4 m2.
3.
Results
Concentrations of sulfate, sulfide and elemental sulfur in the pore water of experimental constructed wetland along flow path from inlet to outlet are presented in Table 1. The sulfide and elemental sulfur depending on the distance from inlet to outlet gradually increased to 18.5 mg/L and 8.2 mg/L, respectively. However, the concentrations of sulfate-sulfur varied from 266 mg/L to 293 mg/L through the wetland and did not show any decreasing tendency. The clearly production of sulfide and elemental sulfur was not proved by the decrease of sulfate concentrations along the flow path, which was attributed to the concentrating effect of considerable water losses occurring via evapotranspiration. As shown in Fig. 1, the monthly average water loss calculated from inflow and outflow streams varied from 55% to 1%, correlating to the changes of temperature from 25 C to 8 C during sampling months. The loss of water from the wetland was mainly (about 92%) conducted by the plants transpiration.
Concerning the considerable water losses from wetland, the concentrations can not really reflect the microbial sulfur transformations here. Thus, according to Eqs. (2) and (3) the daily loads along the flow path from inlet to outlet were calculated and used to interpret the sulfur dynamics through the wetland. As shown in Fig. 2, with gradual increase of sulfide and elemental sulfur in the pore water through the wetland, sulfate-sulfur decreased correspondingly from 34 g/ d to 27 g/d, indicating the calculation of load with consideration of water loss fits the case study in the field. The reduction of sulfate was generally achieved 21% with a specific area removal rate of 1.75 g/m2$d. The compositions of isotopic sulfur and oxygen of sulfate and sulfide along the flow path from inlet to outlet were shown in Table 2 and Fig. 3. In general, d34S of sulfate and sulfide and d18O of sulfate increased from inlet with values of þ5.7&, 33.2& and þ6.7& to outlet with values of þ16&, 25& and þ13&, respectively. Significant enrichment of heavy isotopes in this experiment was observed. The pH measured in all sampling points was within the range of 6.5e6.9. Redox potential gradually decreased from about 50 mV in the inflow to 120 mV in the outflow (4 m sampling point). Ferrous iron from the pore water through the whole bed was constant around 0.5 mg/L.
4.
Discussion
Sulfide is a product of bacterial dissimilatory sulfate reduction (BSR) by using organic compounds as electron donors. Elemental sulfur is a product of sulfide oxidation, which may be performed by abiotic oxidation and/or biological oxidation by using different electron acceptors, such as oxygen, nitrite and nitrate (Buisman et al., 1990; Mahmood et al., 2009; Zheng, 2007; Zheng and Cai, 2007). In this study, the coexisting of sulfide and elemental sulfur in the pore water of experimental
Table 1 e Sulfur species concentrations of the soil pore water of experimental pilot-scale constructed wetland along flow path from inlet to outlet (in mg/L, n [ 15). Flow path (m)
0 (inflow) 0.5
1
2
3
4 (outflow)
SO2 4 S
Depth (m)
0.3 0.4 0.5 0.3 0.4 0.5 0.3 0.4 0.5 0.3 0.4 0.5 0.3 0.4 0.5
S2
S0
Mean
STDEV
Mean
STDEV
Mean
STDEV
283.1 281.5 282.0 282.7 280.5 284.6 292.7 261.2 270.5 269.2 288.4 285.2 263.9 266.0 257.9 288.1
12.1 12.1 10.4 12.4 11.7 11.1 15.3 19.8 16.9 18.5 15.8 18.3 18.5 17.2 24.1 14.8
B.D.L 2.2 1.1 1.1 3.2 2.9 1.8 14.2 10.3 8.4 15.5 15.8 18.5 14.1 14.6 17.9
B.D.L 0.9 0.6 0.8 2.8 2.8 1.7 5.1 2.5 3.6 7.1 3.6 4.9 4.6 3.1 6.2
B.D.L 1.3 1.3 0.4 1.6 0.8 0.4 7.2 4.3 4.1 6.7 6.9 8.2 6.1 6.5 6.8
B.D.L 0.8 1.0 0.3 1.4 0.6 0.3 1.7 1.4 1.7 2.6 3.8 3.7 2.7 5.1 2.4
B.D.L means the estimated concentration below detection limit. The detection limit for sulfide and elemental sulfur is 0.1 mg/L.
6692
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 8 8 e6 6 9 8
90
35
80
30
Temperature (°C)
60
20
50
15
40 30
10
W ater loss (%)
70 25
20 5 0 07/2010
10
08/2010
09/2010
10/2010
11/2010
0 12/2010
Date (month / year) Daily average temperature Monthly average temperature Monthly water loss
into the rhizosphere, the spatial and temporal micro-scale gradients of oxygen concentrations and redox states are consequently established close to root surfaces (Colmer, 2003; Bezbaruah and Zhang, 2004). The coexisting reductive and oxidative conditions in the root-near zones could enable microbial processes such as mineralization of organic carbon, sulfate reduction and sulfide re-oxidation realized on a small spatial scale simultaneously (Holmer and Storkholm, 2001). In concert with decreasing sulfate-sulfur load from inlet to outlet presented in Fig. 3, the significant enriched 34S of sulfate values along flow path (Fig. 4) are strong evidence for the occurrence of BSR. Generally, under closed system conditions, 34 the relationship between SO2 4 concentrations and d S values is expressed by a Rayleigh Eq. (5) where 3 is the respective enrichment factor for sulfur and f stands for the fraction of 2 residual sulfate expressed as CeSO2 4 /C0eSO4 . d34 S ¼ d34 SSO42initial þ εlnðfÞ
Fig. 1 e The dynamics of temperature and water loss in the experimental constructed wetland. The down arrows show the isotope sampling campaigns. Box plot shows median (central thick lines), 25% and 75% quartile ranges around the median, upper and lower edge (hinge) of the box. The ends of the vertical lines (whiskers) indicate the minimum and maximum data values.
Since considerable water loss via plant transpiration occurred in this study, the concentrations of sulfate measured in the pore water can really not reflect the BSR. Thus, the 2 parameter of f in Eq. (5) expressed as CeSO2 4 /C0eSO4 can also not fit here. Accordingly, the load of sulfate calculated as a result of combined concentration and water loss was used in this study and the modified f in Eq. (5) was expressed as L2 SO2 4 /L0-SO4 . Besides, the application of Eq. (5) on field data often only yields an apparent enrichment factor 3 , because closed system conditions are rarely achieved under aquifer conditions (Kno¨ller et al., 2006). For the investigation in this
40
4
35
3
S (g/d)
30
2-
SO42--S (g/d)
wetland (Table 1 and Fig. 2) indicates the simultaneous reduction and re-oxidation of sulfur compounds. Regarding the oxygen and organic compounds released from the roots
(5)
2
25
1
20 2.0
0 0
1
2
3
4
Distance from inlet (m)
S0 (g/d)
1.5
1.0
0.5
0.0 0
1
2
3
4
Distance from inlet (m)
Fig. 2 e The load dynamics of sulfate-sulfur, sulfide and elemental sulfur along flow path in the experimental constructed wetland from inflow to outflow. The meaning of box plot was described in Fig. 2.
6693
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 8 8 e6 6 9 8
Table 2 e Isotope fractionation data for investigated sampling points of experimental pilot-scale constructed wetland along flow path from inlet to outlet (in &). Flow path (m) 0 (inflow) 0.5
1
2
3
4 (outflow)
Depth (m)
Sampling campaign in August SeSO2 4
34
d
d
5.8 6.7 6.0 5.9 6.5 6.2 6.3 12.0 9.8 9.3 13.0 13.6 13.2 16.6 16.3 14.3
0.3 0.4 0.5 0.3 0.4 0.5 0.3 0.4 0.5 0.3 0.4 0.5 0.3 0.4 0.5
OeSO2 4
18
7.0 7.9 7.7 7.9 8.0 7.6 7.3 11.2 9.6 8.3 11.4 10.2 11.4 12.8 12.4 11.2
33.2
28.7
28.4
24.9
13
‰ (VSMOW )
16
Delta- O-SO4
10 8
18
34
2-
2-
‰ (VCDT)
14
12
6 4 -24
d18OeSO2 4
6.0 7.0 6.2 5.7 6.4 6.0 6.3 8.0 7.6 7.1 9.4 9.9 10.2 11.8 12.4 13.0
6.7 7.8 7.4 7.0 7.6 7.7 7.4 9.7 10.1 9.0 12.0 11.4 9.6 12.1 11.6 10.3
d34S-HS
31.0
32.0
28.3
Besides the enrichment of heavy sulfur isotope in the residual sulfate during BSR, the fractionation of oxygen isotopes in the residual sulfate molecule was also carried out. Even though the breaking of the SeO bondage will result in a kinetic isotopic effect as observed for sulfur, the kinetic isotopic effect of oxygen is masked by an isotope exchange between oxygen in sulfate and oxygen in ambient water in natural environments (Kno¨ller and Schubert, 2010). The enrichment of heavy oxygen
18
14
d SeSO2 4 34
d S-HS
study, the apparent isotopic enrichment factor (3 ) for 34S was 22&, which obtained by fitting the logarithmic Eq. (5) to the measured data (Fig. 4). The convincing logarithmic relationship shown in Fig. 4 as well as the high value of correlation coefficient (R2) suggest that BSR could be the main process determining the distribution and isotope composition of sulfate at the site, as compared to the processes like adsorption and/or desorption.
Delta- S-SO4
Sampling campaign in October
34
12 11 10 9 8 7 6 0
1
2
3
4
-28
34
Delta- S-S
2-
‰ (VCDT)
Distance from inlet (m) -26
-30
-32
-34 0
1
2
3
4
Distance from inlet (m)
Fig. 3 e The dynamics of isotopic compositions of d34S and d18O in sulfate and d34S in sulfide of pore water along flow path in the experimental constructed wetland from inflow to outflow.
6694
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 8 8 e6 6 9 8
8 15 6
y= -21.9 * ln(x) + 5.7 R2=0.86
10
4
2
5
0 1.0
0.9
0.8
0.7
0.6
0.5
Fraction residual sulfate (L/L0) 18
O-sulfate S-sulfate Logarithmic fitting of 34S-sulfate 34
Fig. 4 e The correlation of isotopic compositions of d34S and d18O in sulfate and fraction residual sulfate which expressed as fraction of the residual sulfate load to the initial inflow sulfate load.
in residual sulfate via isotope exchange usually was reached up to a certain equilibrium value. This equilibrium takes place within the sulfate reducing bacteria via intermediate intracellular sulfur compounds formed during BSR that are subject to re-oxidation (Brunner et al., 2005; Kno¨ller et al., 2006; Mangalo et al., 2007; Turchyn et al., 2010). As shown in Figs. 3 and 4, the significant enrichment of oxygen in the residual sulfate may again imply that the reduction of sulfate in constructed wetlands is exclusively due to BSR. As a product of BSR, sulfide is generally present in the pore water of the constructed wetlands. However, if reactive metals (e.g. iron) are also present in the aquifer, a considerable part of the sulfide may immediately be precipitated as metal sulfide and then immobilized in the wetland matrix. The isotopic evolution of the sulfide pool is depending on the isotopic composition of the precursor sulfate, enrichment factor (3 ) and the fraction of sulfide which is removed from the dissolved pool by precipitation with metals (Kno¨ller and Schubert, 2010). In closed systems, the estimation of isotopic composition of sulfide pool could be obtained by two different derivations according to Rayleigh Eq. (5). If the sulfide is precipitated under conditions with sufficient metals, the approximate d34S values of the immediate produced sulfide (d34Ssulfide-Instanous-t) can be calculated from Eq. (6). If in case of accumulation of sulfide in the pore water reservoir without any precipitation with metals, the isotopic composition of sulfide at a certain stage of BSR can be obtained according to Eq. (7). d34 SsulfideInstanoust ¼ d34 Ssulfatet þ ε
(6)
d34 Ssulfidereservoirt ¼ d34 Ssulfateinitial f$ε$lnf=ð1 fÞ
(7)
12
y= x
10
0
10
8
2-
20
Delta-18O-SO42- ‰ (VSMOW )
2-
Delta-34S-SO4 ‰ (VCDT)
12
The correlation between assumed bacterial sulfate consumption and the sum of existing reduced sulfur compounds such as sulfide and elemental sulfur are shown in Fig. 5. The sum of estimated sulfide and elemental sulfur in the pore water from bacterial sulfate consumption was lower than the expected product reservoir accumulation. In general, the measured reduced sulfur only takes 30% of the decreased sulfate in this experimental wetland bed and 70% was immobilized in the wetland matrix such as precipitation of sulpide with metals and elemental sulfur. In addition, as well as immobilization in the matrix, formation of dissolved organic sulfur compounds and plant uptake could also account for some of the lost sulfur as shown by the few studies that have examined these in peatlands (Steinmann and Shotyk, 1997; Bottrell et al., 2010; Bartlett et al., 2009). Unfortunately, the data on sulfur species in the substrate of wetland was not available in this study and should be further investigated. Considering the presence of sulfide measured in the pore water (Table 1 and Fig. 2) and considerable immobilization of sulfur compounds (Fig. 5), the d34S values of dissolved sulfide should be plotted within a theoretical range defined by Eqs. (6) and (7). As shown in Fig. 6, this range is enclosed by the two curves modeled from Eqs. (6) and (7). However, surprisingly, no sulfide samples were plotted within the expected isotopic range and all samples were quite below this range in this study. This finding indicated that the enrichment factor for sulfur isotopes calculated from the d34S vs. sulfate mass loss is significantly smaller than required to produce the observed large difference in d34S between sulfate and sulfide. Also, this
Sum of dissolved S and S (g/d)
14
25
6
70%
4
y= 0.3 x (R2= 0.53)
2
30% 0 0
2
4
6
8
10
12
Sulfate reduction (g/d)
Fig. 5 e Correlation between assumed bacterial sulfate reduction and sum of estimated sulfide and elemental sulfur in the pore water of constructed wetland. The straight dashed line illustrates the theoretical correlation for a complete accumulation of reduced sulfur reservoir. The straight line stands for the regression of measured sum of sulfide and elemental sulfur in the pore water. Regarding the slopes of the two lines, 30% of the reduced sulfate was expressed as dissolved sulfide and elemental sulfur, indicating 70% of the reduced sulfate was deposited in the wetland matrix such as precipitation of sulfide with metals and elemental sulfur.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 8 8 e6 6 9 8
suggested one or more sulfur transformation processes superimposing on BSR, such as bacterial disproportionation of elemental sulfur (Bo¨ttcher et al., 2001; Holmkvist et al., 2011) and direct oxidation of sulfide to sulfate by elemental oxygen which was introduced by plant roots. In constructed wetlands, the release of oxygen from plant roots into rhizosphere was well reported (Armstrong et al., 1990). The re-oxidation of sulfide to elemental sulfur using oxygen as electron acceptor is accordingly quiet reasonable here. Besides, nitrite and nitrate as a product of ammonium oxidation can also easily stimulate sulfide oxidation to elemental sulfur in wetlands (Krishnakumar and Manilal, 1999; Londry and Suflita, 1999). The process associated with sulfide re-oxidation to elemental sulfur only provides a minor isotope effect and the isotopic composition of sulfide pool would not change significantly (Bo¨ttcher et al., 1990; Balci et al., 2007). However, if elemental sulfur undergoes microbial disproportionation into sulfate and sulfide (Eq. (8)), considerable sulfur isotope fractionation and significant isotopic difference between sulfate and sulfide can be easily reached. þ 4H2 O þ 4S0 /3H2 S þ SO2 4 þ 2H
(8)
The microbial disproportionation of elemental sulfur produces more enriched sulfate and more depleted sulfide. 20 15
y= -21.9 * ln(x) + 5.7 R2=0.86
34
2-
Delta- S-SO4 ‰ (VCDT)
10 5 0 -5 -10 -15 -20 -25 -30 -35 1.0
0.9
0.8
0.7
0.6
0.5
Fraction residual sulfate (L/L0) Sulfate Sulfide Calculated sulfide instantaneous Calculated sulfide product reservoir Fitting sulfate
Fig. 6 e Relationship between the fraction of the residual sulfate and the measured sulfur isotopic composition of d34S in dissolved sulfate and sulfide. The curves represent the calculated isotopic evolution of the sulfate pool (Eq. (5)), of the instantaneously produced sulfide (Eq. (6)), and of the sulfide pool in case of an accumulation of a product reservoir (Eq. (7)) during progressing bacterial sulfate reduction under closed system conditions. For the calculation, the field based apparent isotopic enrichment factor of L21.9& was used.
6695
When the new produced sulfate mixed with the precursor sulfate, the isotopic pool of 34S will be slightly enriched. In like manner, the new sulfide pool is isotopically lighter than the precursor pool (Bo¨ttcher et al., 2001). Consequently, the isotopic difference between sulfate and sulfide will increase, as compared to the difference as a result of BSR only. Moreover, if the newly formed sulfide is oxidized again to S0, its disproportionation will further increase the isotopic difference. In this case with periodical re-oxidation of sulfide and disproportionation of elemental sulfur, the isotopic difference between coexisting sulfate and sulfide would be significantly enlarged. So far, no direct evidence for the occurrence of bacterial sulfur disproportionation has been provided in constructed wetlands. A stable isotope fractionation during formation of sulfate via this disproportionation process was experimentally investigated in a pure culture. It was found that the dissolved sulfate was enriched in 18O by 17.4& (Bo¨ttcher et al., 2001). Compared to this, the smaller enrichment of 18O in dissolved sulfate (6&) in this study demonstrated the potential occurrence of disproportionation of reduced sulfur compounds in CWs. In addition to the microbial disproportionation, direct reoxidation of sulfide to sulfate by elemental oxygen released from wetland plant roots, may also be a possible explanation for the unexpected enlarged isotopic differences between the coexisting sulfate and sulfide. Considering the hypothesis of significant direct re-oxidation of sulfide to sulfate, the enrichment factor (3 ) obtained by fitting the logarithmic analysis of measured sulfate isotope data can be underestimated to reflect the real enrichment factor. The isotopic composition of sulfate pool can be depleted when new formed sulfate from reoxidation of sulfide with a more depleted sulfate isotopic composition mixed with the precursor sulfate. Compared to the sulfur isotope fractionation during BSR only, the sulfide pool does not significantly change its isotopic composition during re-oxidation of sulfide to sulfate. Consequently, if considerable re-oxidation of sulfide to sulfate occurred, the enrichment factor obtained by fitting the logarithmic analysis of measured sulfate isotope data should not be the real evaluation of BSR. According to Eq. (6) under the assumption that the produced sulfide from sulfate reduction was immediately precipitated in the matrix, the real enrichment factor obtained for pure BSR process can be presented as isotopic difference between coexisting sulfate and sulfide. Based on this case, the real isotope curves from process of BSR only in this study was modeled using the average isotopic difference (38.9&) between sulfate and sulfide as the enrichment factor (Fig. 7). In this way, the sulfide samples as shown in Fig. 7 were perfectly enclosed within the range which calculated from the modeled sulfide isotope fractionation curve. If one considers the hypothesis of re-oxidation as valid in this study, the fraction of newly produced sulfate from sulfide re-oxidation can be modeled according to Eq. (9). d34 Smixed ¼ d34 Sprecursor ð1 XÞ þ d34 Sproduced X
(9)
in which the X stands for the fraction of newly produced sulfate from oxidation of sulfide. Using liner regression analysis of the modeled real sulfate isotope (as precursor pool) and fitting sulfate by measured values (as mixed pool), the fraction
6696
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 8 8 e6 6 9 8
of newly produced sulfate from oxidation of sulfide (X) was calculated as 41%. This fraction indicates that there was 41% of the measured sulfate resulted from sulfide re-oxidation. The mixture of 41% new produced sulfate with precursor sulfate made the enrichment factor of the whole sulfate pool decreased from 38.9& to 21.9&. Moreover, the reaction of sulfide oxidation to sulfate drived by oxidizing reactant like oxygen can be expressed as Eq. (10). Considering the fraction of 41% new produced sulfate, the oxygen equivalent consumption was calculated up to be 4.1 g/ m2d by using the minimum mean sulfate load of 27 g/d in 4 m from the inlet of the wetland with corresponding area of 4 m2 in this study. S2 þ 2O2 /SO2 4
(10)
Regarding the fine sand as the matrix used in the wetland and the horizontal saturated flow model, the oxygen diffusion from atmosphere into the wetland could be neglected in this
30
y= -38.9 * ln(x) + 5.7
25
Delta-34S-SO42- ‰ (VCDT)
20 15
y= -21.9 * ln(x) + 5.7 R2=0.86
10 5 0 -5 -10 -15
study. The oxygen supply here was mainly attributed to the plants. The oxygen flux from the roots of P. australis into their surroundings reported by Lawson (1985) and Armstrong et al. (1990) was up to be 4.3 g/m2$d and 5e12 g/m2$d, respectively. The calculated plant oxygen transfer capacity of 4.1 g/m2d using isotopic technology here was well agreed. This agreement strongly indicates the important role of plants in constructed wetlands and also indicates the potential application of isotope fractionation in constructed wetlands for deep understanding of the internal complex sulfur transformation processes. Besides the role of plants in constructed wetlands for the release of oxygen, the release of organic carbon as providing extra electron donors is as well important to influence the various microbial processes. The inflow sulfate load of the experimental wetland is about 34 g/d and outflow (4 m) sulfate is reduced to be approximate 27 g/d, yielding a net difference of 7 g/d. Regarding the 41% (11.6 g/d) of the outflow sulfate coming from the re-oxidation of sulfide, the net flux of bacterial sulfate reduction could be calculated up to be 18.6 g/ d with a specific area reduction rate of 4.65 g/m2d. The concentration of MCB as the main organic compound in the inflow water is around 0.3 g/m2d in this study. According to the reaction of MCB as electron donor coupled to the reduction of sulfate as electron acceptor expressed as Eq. (11) (Colberg, 1990), the reduction of 4.65 g/m2d sulfate needs 1.56 g/m2d consumption of MCB. The net difference of 1.26 gMCB/m2d from inflow supply and theoretical consumption strongly gives the indication of extra electron donors involved in the sulfate reduction process, underlining the role of organic compounds release from plant roots deriving sulfur cycling in constructed wetlands.
-20 -25
þ 2C6 H5 Cl þ 7SO2 4 þ 5H /12CO2 þ 7HS þ 2Cl þ 4H2 O
-30 -35 -40 1.0
0.9
0.8
0.7
0.6
0.5
Fraction residual sulfate (L/L0) Sulfide Sulfate Initial sulfate Fitting sulfate Calculated sulfide instantaneous Calculated sulfide product reservoir
Fig. 7 e Relationship between the fraction of the residual sulfate and the measured sulfur isotopic composition of d34S in dissolved sulfate and sulfide. The curves represent the calculated isotopic evolution of the initial sulfate pool which processing by BSR only, of the measured sulfate pool (Eq. (5)), of the instantaneously produced sulfide (Eq. (6)), and of the sulfide pool in case of an accumulation of a product reservoir (Eq. (7)) during progressing bacterial sulfate reduction under closed system conditions. For the calculation, the isotopic enrichment factor for the initial sulfate pool was modified to L38.9& according to the average isotopic difference between coexisting dissolved sulfate and sulfide.
(11)
In general, the effect of vegetation in constructed wetland was well reported (Brix, 1997; Chazarenc et al., 2009) and the role of plants improving the performance by releasing organic carbon compounds and/or oxygen from roots is also proved (Merbach et al., 1999; Picek et al., 2007). However, due to quite a biodegradation availability of the released organic carbons, the fast consumption by various microbes sitting around the roots surface makes the quantification of the released organic compounds into the wetland bed become extreme difficult. The consumption of some pollutants only under irreversible reaction using organic compounds as electron donor can be used to calculate the theoretical amount of organic carbon released from roots. However, as sulfate is undergoing microbial reversible reactions including sulfate reduction and re-oxidation of reduced sulfur compounds, the estimation of organic carbon consumption here is extremely difficult. In this study, a progressive step was made by using the approach of stable isotope combined with the common hydro-chemical parameters, and the capacity of organic carbon release from roots was calculated to 1.26 g/m2d MCB equivalent. But this value was derived by only considering the process of bacterial sulfate reduction. If considering the consumption of organic compounds by denitrification, methanogenesis and microbial respiration, the capacity of organic matter release from roots should be larger.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 8 8 e6 6 9 8
5.
Conclusion
The significant enrichment of heavy isotopes of sulfur and oxygen in dissolved sulfate was observed to be clearly correlated to the decrease of sulfate loads along the flow path through an experimental horizontal subsurface flow wetland bed. This strongly indicates the occurrence of bacterial dissimilatory sulfate reduction. - Lack of sulfur isotope mass balance between sulfate removed and sulfide produced implies that other processes are superimposed on bacterial sulfate reduction. These include re-oxidation of sulfide to sulfate by oxygen and bacterial disproportionation of elemental sulfur in constructed wetlands. - The application of the stable isotope approach combined with common hydro-chemical investigations enables a general qualitative evaluation of sulfur transformations in constructed wetlands, but also leads to a quantitative description of intermediate processes. -
Acknowledgement This work was supported by a grant of China Scholarship Council (CSC) and by the Helmholtz Centre for Environmental Research e UFZ within the scope of the SAFIRA II Research Program (Revitalization of Contaminated Land and Groundwater at Megasites, subproject ‘‘Compartment Transfer e CoTra”). We are grateful to Martina Neuber and Sandra Zuecker-Gerstner of the stable isotope laboratory Halle/Salle for conducting isotope analyses of the samples. Thanks are also addressed to the A. Al-Dahoodi and M. Schro¨te for their valuable assistance in the laboratory and field. Furthermore, we would like to thank Simon Botrell for his valuable comments that considerably helped improve the paper.
references
Aesoy, A., Odegaard, H., Bentzen, G., 1998. The effect of sulphide and organic matter on the nitrification activity in a biofilm process. Water Science and Technology 37 (1), 115e122. Alewell, C., Giesemann, A., 1996. Sulfate reduction in a forested catchment as indicated by d34S values of sulfate in soil solutions and runoff. Isotopes in Environmental and Health Studies 32, 203e210. Anderson, R.T., Lovley, D.R., 2000. Anaerobic bioremediation of benzene under sulfate-reducing conditions in a petroleumcontaminated aquifer. Environmental Science & Technology 34, 2261e2266. Armstrong, J., Armstrong, W., Beckett, P.M., 1990. Measurement and modeling of oxygen release from roots of Phragmites australis. In: Cooper, P.F., Findlater, B.C. (Eds.), Constructed Wetlands in Water Pollution Control. Pergamon Press, Oxford, United Kingdom, pp. 41e52. Armstrong, J., Afreen-Zobayed, F., Armstrong, W., 1996. Phragmites die-back: sulphide- and acetic acid-induced bud and root death, lignifications, and blockages within aeration and vascular systems. New Phytologist 134, 601e614.
6697
Balci, N., Shanks III, W.C., Mayer, B., Mandernack, K.W., 2007. Oxygen and sulfur isotope systematics of sulfate produced by bacterial and abiotic oxidation of pyrite. Geochimica et Cosmochimica Acta 71, 3796e3811. Bartlett, R., Bottrell, S.H., Coulson, J.P., Lee, J., Forbes, L., 2009. 34S tracer study of pollutant sulfate behaviour in a lowland peatland. Biogeochemistry 95, 261e275. Bezbaruah, A.N., Zhang, T.C., 2004. pH, redox, and oxygen microprofiles in rhizosphere of bulrush (Scirpus validus) in a constructed wetland treating municipal wastewater. Biotechnology and Bioengineering 88, 60e70. Bo¨ttcher, J., Strebel, O., Voerkelius, S., Schmidt, H.L., 1990. Using isotope fractionation of nitrate nitrogen and nitrate oxygen for evaluation of microbial denitrification in a sandy aquifer. Journal of Hydrology 114, 413e424. Bo¨ttcher, M.E., Thamdrup, B., Vennemann, T.W., 2001. Oxygen and sulfur isotope fractionation during anaerobic bacterial disproportionation of elemental sulfur. Geochimica et Cosmochimica Acta 65, 1601e1609. Bottrell, S.H., Hayes, P.J., Bannon, M., Williams, G.M., 1995. Bacterial sulfate reduction and pyrite formation in a polluted sand aquifer. Geomicrobiology Journal 13, 75e90. Bottrell, S.H., Hatfield, D., Bartlett, R., Spence, M.J., Bartle, K.D., Mortimer, R.J.G., 2010. Concentrations, sulfur isotope compositions and origin of organosulfur compounds in pore waters of a highly polluted raised peatland. Organic Geochemistry 41, 55e62. Braeckevelt, M., Mirschel, G., Wiessner, A., Rueckert, M., Reiche, N., Vogt, C., Schultz, A., Paschke, H., Kuschk, P., Kaestner, M., 2008. Treatment of chlorobenzenecontaminated groundwater in a pilot-scale constructed wetland. Ecological Engineering 33, 45e53. Brix, H., 1997. Do macrophytes play a role in constructed treatment wetlands? Water Science and Technology 35 (5), 11e17. Brunner, B., Bernasconi, S.M., Kleikemper, J., Schroth, M.H., 2005. A model for oxygen and sulfur isotope fractionation in sulfate during bacterial sulfate reduction processes. Geochimica et Cosmochimica Acta 69, 4773e4785. Bulc, T.G., 2006. Long term performance of a constructed wetland for landfill leachate treatment. Ecological Engineering 26, 365e374. Buisman, C., Uspeert, P., Janssen, A., Lettinga, G., 1990. Kinetics of chemical and biological sulphide oxidation in aqueous solutions. Water Research 24, 667e671. Chazarenc, F., Gagnon, V., Comeau, Y., Brisson, J., 2009. Effect of plant and artificial aeration on solids accumulation and biological activities in constructed wetlands. Ecological Engineering 35, 1005e1010. Choi, J.H., Park, S.S., Jaffe´, P.R., 2006. The effect of emergent macrophytes on the dynamics of sulfur species and trace metals in wetland sediments. Environmental Pollution 140, 286e293. Colberg, P.J.S., 1990. Role of sulfate in microbial transformations of environmental contaminants: chlorinated aromatic compounds. Geomicrobiology Journal 8, 147e165. Colmer, T., 2003. Long distance transport of gases in plants: a perspective on internal aeration and radial oxygen loss from roots. Plant, Cell & Environment 26, 17e36. Faulwetter, J.L., Gagnon, V., Sundberg, C., Chazarenc, F., Burr, M. D., Brisson, J., Camper, A.K., Stein, O.R., 2009. Microbial processes influencing performance of treatment wetlands: a review. Ecological Engineering 35, 987e1004. Finster, K., Liesack, W., Thamdrup, B., 1998. Elemental sulfur and thiosulfate disproportionation by Desulfocapsa sulfoexigens sp. nov., a new anaerobic bacterium isolated from marine surface sediment. Applied and Environmental Microbiology 64, 119. Fry, B., Gest, H., Hayes, J., 1988. 34S/32S fractionation in sulfur cycles catalyzed by anaerobic bacteria. Applied and Environmental Microbiology 54, 250.
6698
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 8 8 e6 6 9 8
Garcia, J., Rousseau, D.P.L., Morato, J., Lesage, E., Matamoros, V., Bayona, J.M., 2010. Contaminant removal processes in subsurface-flow constructed wetlands: a review. Critical Reviews in Environmental Science and Technology 40, 561e661. Geurts, J.J.M., Sarneel, J.M., Willers, B.J.C., Roelofs, J.G.M., Verhoeven, J.T.A., Lamers, L.P.M., 2009. Interacting effects of sulphate pollution, sulphide toxicity and eutrophication on vegetation development in fens: a mesocosm experiment. Environmental Pollution 157, 2072e2081. Holmer, M., Storkholm, P., 2001. Sulphate reduction and sulphur cycling in lake sediments: a review. Freshwater Biology 46, 431e451. Holmkvist, L., Ferdelman, T.G., Jørgensen, B.B., 2011. A cryptic sulfur cycle driven by iron in the methane zone of marine sediment (Aarhus Bay, Denmark). Geochimica et Cosmochimica Acta. Kno¨ller, K., Fauville, A., Mayer, B., Strauch, G., Friese, K., Veizer, J., 2004. Sulfur cycling in an acid mining lake and its vicinity in Lusatia, Germany. Chemical Geology 204, 303e323. Kno¨ller, K., Schubert, M., 2010. Interaction of dissolved and sedimentary sulfur compounds in contaminated aquifers. Chemical Geology, 284e293. Kno¨ller, K., Vogt, C., Feisthauer, S., Weise, S., Weiss, H., Richnow, H., 2008. Sulfur cycling and biodegradation in contaminated aquifers: insights from stable isotope investigations. Environmental Science & Technology 42, 7807e7812. Kno¨ller, K., Vogt, C., Richnow, H.-H., Weise, S.M., 2006. Sulfur and oxygen isotope fractionation during benzene, toluene, ethyl benzene, and xylene degradation by sulfate-reducing bacteria. Environmental Science & Technology 40, 3879e3885. Krishnakumar, B., Manilal, V., 1999. Bacterial oxidation of sulphide under denitrifying conditions. Biotechnology Letters 21, 437e440. Lawson, G., 1985. Cultivating Reeds (Phragmites Australis) for Root Zone Treatment of Sewage. Contract Report ITE Project 965. Water Research Centre, Cumbria, UK. Lee, R., Kraus, D., Doeller, J., 1999. Oxidation of sulfide by Spartina alterniflora roots. Limnology and Oceanography 44, 1155e1159. Leon, P.M.L., Falla, S.-J., Samborska, E.M., Dulken, I.A.R.v., Hengstum, G.v., Roelofs, J.G.M., 2002. Factors controlling the extent of eutrophication and toxicity in sulfate-polluted freshwater wetlands. Limnology and Oceanography 47, 585e593. Liesack, W., Schnell, S., Revsbech, N.P., 2000. Microbiology of flooded rice paddies. FEMS Microbiology Reviews 24, 625e645. Londry, K., Suflita, J., 1999. Use of nitrate to control sulfide generation by sulfate-reducing bacteria associated with oily waste. Journal of Industrial Microbiology and Biotechnology 22, 582e589. Lovley, D., Phillips, E., 1986. Organic matter mineralization with reduction of ferric iron in anaerobic sediments. Applied and Environmental Microbiology 51, 683. Machemer, S.D., Reynolds, J.S., Laudon, L.S., Wildeman, T.R., 1993. Balance of S in a constructed wetland built to treat acid mine drainage, Idaho Springs, Colorado, USA. Applied Geochemistry 8, 587e603. Mahmood, Q., Zheng, P., Hayat, Y., Ren-cun, J., Azim, M.R., Jilani, G., Islam, E., Ahmed, M., 2009. Effect of nitrite to sulphide ratios on the performance of anoxic sulphide oxidizing reactor. Arabian Journal for Science and Engineering 34, 45e54. Mangalo, M., Meckenstock, R.U., Stichler, W., Einsiedl, F., 2007. Stable isotope fractionation during bacterial sulfate reduction
is controlled by reoxidation of intermediates. Geochimica et Cosmochimica Acta 71, 4161e4171. Merbach, W., Mirus, E., Knof, G., Remus, R., Ruppel, S., Russow, R., Gransee, A., Schulze, J., 1999. Release of carbon and nitrogen compounds by plant roots and their possible ecological importance. Journal of Plant Nutrition and Soil Science 162, 373e383. Picek, T., Cizkova, H., Dusek, J., 2007. Greenhouse gas emissions from a constructed wetland-Plants as important sources of carbon. Ecological Engineering 31, 98e106. Rahman, K., Wiessner, A., Kuschk, P., Mattusch, J., Ka¨stner, M., Mu¨ller, R., 2008. Dynamics of arsenic species in laboratory-scale horizontal subsurface-flow constructed wetlands treating an artificial wastewater. Engineering in Life Sciences 8, 603e611. Rees, C., 1973. A steady-state model for sulphur isotope fractionation in bacterial reduction processes. Geochimica et Cosmochimica Acta 37, 1141e1162. Rethmeier, J., Rabenstein, A., Langer, M., Fischer, U., 1997. Detection of traces of oxidized and reduced sulfur compounds in small samples by combination of different highperformance liquid chromatography methods. Journal of Chromatography A 760, 295e302. Robertson, W., Schiff, S., 1994. Fractionation of sulphur isotopes during biogenic sulphate reduction below a sandy forested recharge area in south-central Canada. Journal of Hydrology 158, 123e134. Schroth, M.H., Kleikemper, J., Bolliger, C., Bernasconi, S.M., Zeyer, J., 2001. In situ assessment of microbial sulfate reduction in a petroleum-contaminated aquifer using pushpull tests and stable sulfur isotope analyses. Journal of Contaminant Hydrology 51, 179e195. Stein, O.R., Borden-Stewart, D.J., Hook, P.B., Jones, W.L., 2007. Seasonal influence on sulfate reduction and zinc sequestration in subsurface treatment wetlands. Water Research 41, 3440e3448. Steinmann, P., Shotyk, W., 1997. Chemical composition, pH and redox state of sulphur and iron in complete vertical profiles from two Sphagnum peat bogs, Jura mountains, Switzerland. Geochimica et Cosmochimica Acta 61, 1143e1163. Sturman, P., Stein, O., Vymazal, J., Kro¨pfelova´, L., 2008. Sulfur cycling in constructed wetlands. In: Vymazal, J. (Ed.), Wastewater Treatment, Plant Dynamics and Management in Constructed and Natural Wetlands. Springer, pp. 329e344. Turchyn, A.V., Bruchert, V., Lyons, T.W., Engel, G.S., Balci, N., Schrag, D.P., Brunner, B., 2010. Kinetic oxygen isotope effects during dissimilatory sulfate reduction: a combined theoretical and experimental approach. Geochimica et Cosmochimica Acta 74, 2011e2024. Webb, J., McGinness, S., Lappin-Scott, H., 1998. Metal removal by sulphate-reducing bacteria from natural and constructed wetlands. Journal of Applied Microbiology 84, 240e248. Wiessner, A., Rahman, K.Z., Kuschk, P., Ka¨stner, M., Jechorek, M., 2010. Dynamics of sulphur compounds in horizontal subsurface flow laboratory-scale constructed wetlands treating artificial sewage. Water Research 44, 6175e6185. Woulds, C., Ngwenya, B.T., 2004. Geochemical processes governing the performance of a constructed wetland treating acid mine drainage, Central Scotland. Applied Geochemistry 19, 1773e1783. Zheng, P., 2007. Anoxic sulfide biooxidation using nitrite as electron acceptor. Journal of Hazardous Materials 147, 249e256. Zheng, P., Cai, J., 2007. Comparison of anoxic sulfide biooxidation using nitrate/nitrite as electron acceptor. Environmental Progress 26, 169e177.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 9 9 e6 7 0 8
Available online at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Nitrogen removal assessment through nitrification rates and media biofilm accumulation in an IFAS process demonstration study Pusker Regmi a,*, Wes Thomas b, Gary Schafran a, Charles Bott c, Bob Rutherford c, David Waltrip c a
Old Dominion University, Civil and Environmental Engineering, Rm 135 Kaufman Hall, Hampton Blvd, Norfolk, VA 23529, USA Virginia Polytechnic Institute and State University, USA c Hampton Roads Sanitation District, USA b
article info
abstract
Article history:
An IFAS demonstration study was conducted at the 76,000 m3/day (20MGD) James River
Received 5 April 2011
Wastewater Treatment Plant (JRTP) located in Newport News, Virginia by converting one
Received in revised form
fully-aerobic conventional aeration basin with dedicated secondary clarification to
23 September 2011
a 7041 m3/day (8404 m3/day max month) IFAS train in a modified Ludzack-Ettinger (MLE)
Accepted 11 October 2011
configuration. During the study, biomass concentrations on the biofilm carriers were
Available online 19 October 2011
monitored (weekly) as well as nitrogen species concentrations in the IFAS reactor to quantify the nitrogen transformations occurring within the demonstration tank. In
Keywords:
a related effort, nitrification kinetics for ammonia and nitrite oxidizing bacteria were
Biofilm carriers
monitored on a weekly basis for IFAS media alone, IFAS process mixed liquor without
IFAS
media, and IFAS mixed liquor and media together in an effort to identify the location of
MLE
nitrification activity (i.e. on the media or in the suspended culture) in the IFAS process.
Nitrogen
Biomass quantity on the media was generally observed to be inversely related to
Nutrient removal
temperature except during a period when an auxiliary carbon source contaminated with fungi was introduced. Both ammonia oxidizing and nitrite oxidizing bacterial activity were elevated on the carriers compared to the suspended culture (AOBmedia: 4.97 mgNOx/gMLSS/ hr; AOBsuspended: 1.72 mgNOx/gMLSS/hr; NOBmedia: 7.55 mgNOx/gMLSS/hr; NOBsuspended: 0.82 mgNOx/gMLSS/hr) during all periods of the study. In-basin nitrification rates calculated based on nitrogen profiling efforts averaged 0.90 mgNOx/m2/day which was in good agreement with the average of 0.89 mgNOx/m2/day for IFAS mixed liquor and media from batch testing. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
Biofilm technologies are increasingly being implemented in wastewater treatment due to their advantages with respect to smaller reactor sizes, ease of operation, less demanding solids
separation requirements, and the increased specialization of attached biomass (Ødegaard, 2006). One of these technologies is the integrated fixed-film activated sludge (IFAS) process which is a hybrid process relying on microorganisms both in suspended culture and attached to free-floating media to
* Corresponding author. Tel.: þ1 757 255 8465. E-mail addresses:
[email protected],
[email protected] (P. Regmi). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.10.009
6700
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 9 9 e6 7 0 8
achieve soluble BOD and nitrogen removal. Because IFAS media are retained within an aeration basin, process decoupling of the growth rate of nitrifying organisms from the SRT of the suspended mixed liquor phase occurs (Onnis-Hayden et al., 2011). This situation is particularly attractive for processes targeting nitrogen removal since slow-growing nitrifying organisms can be retained in the process where otherwise they would be lost under operational conditions where SRT is shorter than needed for suspended culture populations to proliferate. If this understanding is applicable it means in colder months, nitrification is more dependent on attached growth nitrification activity than on suspended growth and this activity will determine the successful operation of an IFAS treatment process for nitrogen removal during cold conditions. The fact that in an IFAS process nitrification takes place predominantly in the attached biofilm whereas removal of BOD is carried out by the suspended growth makes it possible to keep SRT of the activated sludge shorter than when nitrification occurs solely within a suspended culture. In attached growth systems such as a moving-bed bioreactor (MBBR), the surface area for biofilm growth can be increased by designing carriers with a higher specific surface area or by adding more biofilm carriers to the same reactor volume (Ødegaard et al., 2000). This allows future expansion of treatment capacity without the need for construction of new reactors. However, mixing problems restrict the volumetric fill fraction for carriers to exceed approximately70%, which imposes limitation on increasing the surface area in a given reactor volume (Ødegaard, 2006). Generally principles of microbial ecology apply to biofilms, however, biofilm communities also have distinct features that affect their microbiology and community dynamics (Wuertz et al., 2004). The nitrification rates in integrated biofilm systems are also less affected by changes in temperature (Christensson and Welander, 2004), which makes them favorable in cold climates as a nitrogen removal process. Higher dissolved oxygen concentrations at lower temperature can also compensate for the low growth rate of nitrifiers as there will be deeper penetration of oxygen within the biofilm promoting the growth of active nitrifiers. Hubbell and McDowell (2003) reported significant ammonia reduction at Donner Summit Public Utility District at influent wastewater temperatures as low as 8 C. Stable nitrification has been demonstrated by IFAS processes at even lower mixed liquor concentrations than those where it can be achieved by the conventional activated sludge process (Ross et al., 2004). To date, there are more than 300 wastewater treatment plants operating or under construction based on IFAS-like biofilm processes worldwide with most of them in Europe (Ødegaard, 2006). Several studies have reported successful installation and operation of IFAS (Sriwiriyarat and Randall, 2005) but these plants are located in more northern temperate zones than the JRTP. In North America there are only two full-scale IFAS plants using free-floating plastic carriers. One of these is the Broomfield Wastewater Treatment Plant (BWTP) in Denver, Colorado which started operation in November 2002 as an IFAS process (Rutt et al., 2006; Onnis-Hayden et al., 2007) and has shown capabilities of N
and P removal (Rutt et al., 2006). In Ontario, Canada, the Lakeview Wastewater Treatment Plant has been operating as a full-scale IFAS demonstration project (Ross et al., 2004; Maas et al., 2008; Stricker et al., 2009) since August 2003. Because IFAS is a relatively new technology, limited experience exists in the design of the process particularly retrofitting it into existing facilities. To better understand the process operation and to guide the design for a full-scale plant conversion, the study described here was conducted as a fullscale demonstration study using one of nine aeration tanks and two dedicated secondary clarifiers. The goals of the demonstration study were: 1) To demonstrate the capability of the IFAS treatment process to achieve an annual average settled effluent total nitrogen concentration of 12 mg N/L under design load and flow conditions; 2) To understand the temporal and spatial changes in biomass attached to the biofilm carriers over time and their effect on performance of the IFAS process; 3) Investigate nitrogen conversion across the IFAS process train; 4) Identify the location of nitrogen conversion activity and understand the influence of warm and cold weather conditions on this activity in both the carriers and in the mixed liquor.
2.
Material and methods
2.1.
Demonstration train
To conduct a full-scale demonstration study, one of nine aeration tanks was modified and two secondary clarifiers were isolated and dedicated to the IFAS demonstration process train. The IFAS demonstration reactor consisted of anoxic and aerobic zones similar to a modified LudzackEttinger (MLE) configuration, with the anoxic zone consisting of 35% of total tank volume and the aerobic zone occupying 65% (Fig. 1). The IFAS train consisted of five compartments labeled R1 through R5, with each compartment separated by a baffle wall. Direct-drive submersible mechanical mixers in each of the anoxic chambers provided mixing. The plastic carrier media (AnoxKaldnes K3) was contained within chamber R4 while chamber R5 (no media) was also aerated by a small aeration grid to provide further ammonia reduction. The narrow tank configuration (length to width ratio of 4:1) resulted in high approach velocities and led to more media accumulation at the downstream side of R4 particularly under high recirculation rates of nitrified mixed liquor. The demonstration tank included two internal mixed liquor recycle (IMLR) pumps. Of these two, one is installed at the upstream end of R4 and the other one installed in R5. The purpose of the IMLR pump installed at the upstream end of R4
Fig. 1 e IFAS reactor schematic and average flows during the study.
6701
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 9 9 e6 7 0 8
in monitoring influent and effluent of IFAS train for the above mentioned parameters are presented in Table 3.Operational parameters, flow rate, SRT, and dissolved oxygen were also closely monitored and recorded.
Table 1 e Periods of time for studies carried out during the IFAS demonstration. Study effort
Period
Influent/effluent monitoring Biomass measurements Nitrification rate study Nutrient profiling
January 2008 to June 2009a February 2008 to February 2009 January 2008 to June 2009 June 2008 to January 2009
2.4.
a The influent/effluent monitoring was carried out from November 2007 to June 2009, however, data before January 2008 is not presented in this paper. During this time IFAS train was receiving as much as 14000 m3/d while it was designed to handle 8330 m3/ d (Rutherford, 2010).
(IMLR1) was to limit the high approach velocities caused by the recycle from the second IMLR pump installed in R5 (IMLR2) and ensure more even distribution of media across the R4 chamber. The plastic carrier media used in this demonstration project was AnoxKaldnes K3 media made of virgin polyethylene with a density (0.95 g/cm3) similar to water measuring 12 mm deep and 25 mm in diameter and having specific surface area of 500 m2/m3. The carriers are designed to provide a large protected surface for the microorganisms and good mass transfer. The biofilm carriers were installed in the aerobic reactor and kept suspended and in continuous movement by a coarse bubble aeration system. The carriers were kept within the aerobic reactor by a sieve arrangement at the reactor outlet. At the beginning of the study, 465 m3of this media was installed in the aerobic reactor with the volume corresponding to a fill fraction of 50% and providing the aerobic reactor with an effective media surface area of 250 m2/ m3 of aeration tank volume.
2.2.
Study periods
Several activities or studies were conducted throughout the IFAS demonstration beginning at different dates and lasting for different periods of time. These efforts are summarized in Table 1. Some of the key parameters under which the demonstration tank was operated during these study periods are presented in Table 2. Temporal variation in influent flow, temperature and influent TKN are also presented for the period of biomass monitoring (Fig. 2).
2.3.
Influent/effluent monitoring
Throughout the IFAS demonstration, nitrogen species (TKN, NHþ 4 N, NO3 N and NO2 N), phosphorus, BOD, COD were monitored in the influent (primary clarifier effluent) and effluent to the IFAS reactor. The methods and procedures used
Biomass concentration measurements
Media samples were collected from three different locations within the aerobic bioreactor to identify whether spatial variation of biomass on the media occurred. Biomass measurements were conducted on an approximately weekly basis from January 2008 to January 2009. Total biofilm solids on the media were measured by drying twenty carrier samples collected from each of three locations (60 total) at 105 C for 2 h. The dried samples were weighed and the biomass removed by placing the carriers in a 2 N H2SO4 solution, shaking vigorously using a vortex mixer for 3 min, and then soaking the carriers in the same container and solution for more than 2 h. The carriers were then rinsed several times using de-ionized, distilled water and then dried for more than 2 h at 105 C. High-pressure air was then applied to each media individually to ensure that no dry biofilm remained. The difference in initial and final weight was used to calculate the biomass on the carriers. This method to quantify the biomass on the media was similar to that used by Maas et al. (2008).
2.5.
Attached and suspended biomass calculations
The total suspended biomass in the IFAS zone was calculated by multiplying the MLSS (mg/L) concentration and reactor volume; reactor volume occupied by the mixed liquor was 8.74 106 L. The total attached biomass was calculated based on the biomass density (g/m2) and total surface area of the media in the IFAS zone. The total surface is the product of specific surface area (500 m2/m3) and reactor volume (465 m3) occupied by the media. Multiplying total surface area with biomass density gave the mass of the attached biomass in the IFAS zone.
2.6.
Nitrification rate measurements
Nitrification rates were measured on an approximately weekly basis from January 2008 to February 2009. Both AOB and NOB activity were measured in separate bench-scale experiments. In these experiments nitrification rates for the IFAS mixed liquor alone, IFAS media alone and IFAS media and mixed liquor combined were measured. The bench scale reactors used were 9 L (7 L active volume) rectangular polycarbonate containers with provisions for DO and pH probes access. A sample port located in the middle of the active volume was used to collect samples during different
Table 2 e Summary of selected demonstration tank influent wastewater characteristics and operational conditions (Period: January 2008 to June 2009).
Average Max Min
Flow, m3/d
TKN, mg/L
Temperature, C
SRT, days
COD/TKN
MLSS, mg/L
6743 10542 2468
40.3 69.2 25.5
21.3 29.0 14.0
4.8 13.6 1.9
7.0 10.1 2.9
2514 6980 1180
6702
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 9 9 e6 7 0 8
Additional details of the analytical procedures can be found in Thomas et al. (2009).
2.7. Half-saturation coefficient evaluation (low NHþ 4 N and NO2eN)
Fig. 2 e Influent flow, temperature and influent TKN during influent/effluent monitoring period.
experiments. Mixing was provided by a gang stirrer (Phipps and Bird) to ensure equal mixing to the four reactors. The dissolved oxygen was maintained roughly between 3 and 5 mg/L using pure oxygen. The temperature of the four reactors was maintained similar to that of the IFAS tank by placing them in a temperature controlled water bath and pH was maintained between 7.2 and 7.3 by automated addition of a base (concentrated Na2CO3). Dissolved oxygen, pH and temperature values were recorded directly to a laptop computer. For AOB experiments, reactors with samples were spiked with 40e50 mg/L NHþ 4 N and sampled continuously for 2 h at 25 min intervals. In a similar manner reactors with freshly aerated samples were spiked with 40e50 mg/L of NO 2 N for NOB experiments. To avoid nutrient limitation, reactors were also spiked with 5 mg/L of phosphorus. All collected samples were analyzed for NHþ 4 N, NO2 N, and NO3 N. For field þ analysis, NH4 N was analyzed using HACH colorimetric method TNT plus 832. Colorimetric methods were performed per HACH protocols with evaluation of matrix interference and other issues as needed. NO 3 N was analyzed using HACH colorimetric method TNT plus 836. Nitrite analysis was conducted in the field as soon as possible after sample filtration. The standard colorimetric method was performed using HACH Nitriver 3 reagent and 10 mL samples. HACH method 8507 is a low range method, so all samples were diluted to concentrations between 0.002 and 0.3 mg/L NO2eN and analyzed per standard protocols.
Variable D.O. concentrations of 2, 4, 6, and 8 mg/L were maintained in the controlled reactors and allowed to mix and aerate with pure O2 until each reactor reached the desired D.O. level. The reactors were spiked to 25 mg/L of either ammonia or nitrite; in addition to that phosphate is added to ensure that the biomass was not nutrient limited. Samples were taken at desired intervals to capture the linear rate of nitrification as well as nitrification rates in low substrate concentrations. The recorded results from these tests were analyzed to calculate half-saturation coefficient using four different D.O. concentrations. These tests were conducted three times for AOB and NOB activity on the media at 175 RPM and 100 RPM and as twice for AOB and NOB activity on the media at 100 RPM. In addition to testing the IFAS media, the same experiments were conducted one for AOB and NOB activity in the mixed liquor at 175 RPM.
2.8.
Nutrient profiling
To better understand the fate of the various forms of nitrogen and the total nitrogen removal occurring across the IFAS train, samples were collected at seven locations from the inlet to outlet of the demonstration tank. Three samples were collected from the anoxic basin (referred to as AN1 through AN3 and corresponding to R1, R2, and R3 chambers) and four samples across the aerated basin (OX1-OX4). This effort was carried out on a weekly basis over a seven-month period (June 2008 to January 2009). The profiling conducted over an extended period allowed an evaluation of both spatial and temporal trends in performance across the IFAS basin. To avoid loss/conversion of nitrogen due to biological activities samples were filtered through a 0.45 mm filter immediately after collection and on site and transported on ice back to the laboratory within 45 min. These samples were then analyzed for total dissolved nitrogen (TDN) using a TOC analyzer (Shimadzu TOC-Vcsn, Kyoto, Japan) connected in series with a TNM-1 total
Table 3 e Methods and procedures used for influent/effluent monitoring of IFAS train. Parameter Total and Volatile Suspended Solids (TSS, VSS) Total Kjeldahl Nitrogen (TKN) Total Phosphorous (TP) Ammonia (NH3eN) Nitrate/nitrite (NO2, NO3) Nitrite (NO2) Nitrate (NO3) by calculation Biochemical Oxygen Demand (BOD) Chemical Oxygen Demand (COD)
Method title
Reference method
TSS e Total suspended solids dried at 103e105 C TVSS e Fixed and volatile solids ignited at 550 C Determination of Total Kjeldahl Nitrogen by flow injection analysis colorimetry (Block Digestion) Determination of Total Phosphorous by flow injection analysis colorimetry (Acid Persulfate Digestion) Determination of ammonia by flow injection analysis colorimetry Determination of nitrate/nitrite by flow injection analysis colorimetry
SM 20th 2540D SM 18th 2540E EPA 351.2 10-107-06-2-I EPA 365.1 Lachat 10-115-01-1-E EPA 350.1 Lachat 10-107-06-1-C EPA 353.2 Lachat 10-107-04-1-C/A
Biochemical Oxygen Demand (BOD) 5-day BOD test Chemical Oxygen Demand, Reactor Digestion Method
SM 18TH 5210B Hach 8000
6703
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 9 9 e6 7 0 8
nitrogen module, nitrate and nitrite by ion chromatography (Dionex Model 4500 IC, Sunnyvale, CA, USA), and ammonia using the indophenol blue method (Method 4500-Ammonia, Standard Methods, 1998). The soluble org-N fraction was calculated as the difference of TDN and the sum of the inorganic nitrogen species (NH4eN, NO3eN and NO2eN).
3.
Results and discussion
During the period between February 2008 and June 2009 the average TN concentration in the IFAS effluent was 9.95 mg/L corresponding to 75.1% TN removal (Rutherford, 2010). The average chemical oxygen demand (COD) influent to the IFAS process train was 267 mg/L and the COD/TKN ratios remained low during the period averaging 7.0. The temperature of wastewater in the IFAS tank ranged from 14 C to 29 C and averaged 23 C. The mixed liquor suspended solids concentrations were on average 2786 mg/L and varied between a low of 1180 mg/L and high of 6980 mg/L corresponding to changes in operational SRT varying between 1.9 and 13.6 days. Controlling of SRT by wasting sludge from a return line proved difficult because of wide variations in return sludge concentrations (Rutherford, 2010). The problem was obviated by wasting directly from the IFAS tank.
3.1.
Biomass coverage of K3 media
The total solids attached to the IFAS media was monitored from February 2008 to January 2009. During this study period biofilm total solids on the K3 media ranged from a high of 12.6 g/m2 to a low of 2.4 g/m2, with an average of 6.0 g/m2. The difference in the total biofilm solids among the three sites sampled within the IFAS tank was minimal (Fig. 3), and there was also no visible difference observed among the IFAS media samples at the three sites in terms of color and physical appearance. This observation suggests that mixing in the IFAS tank caused the media to be well mixed along the axis of flow as well as laterally. In the first month of measurement (February 2008) the biofilm solids concentration was 7.0 g/m2 and increased through March 2008 at which point it was 12 g/m2. This increase may be related to acclimatization of the biofilm, which had not reached steady state conditions prior to the first sampling. The temperature of wastewater in the IFAS aerobic bioreactor was approximately 17 C for the month of February and most of March increasing to 20 C by mid-April. Beginning in mid-April the biofilm mass on the IFAS media began decreasing and by the end of the month it was less than 3 g/m2. Examining the full year of biofilm measurements, a seasonal pattern can be observed (biofilm mass inversely related to temperature, Fig. 3). In suspended growth systems the role temperature and its impact have been well understood, however, in attached growth systems temperature effects are not fully known (Wheaton et al., 1994; Okey and Albertson, 1989). The difficulty in assessing temperature effects on biofilm based systems stems from the fact that temperature has a deep impact on biological kinetics as well as substrate diffusion and transport (Zhu and Chen, 2002). In biofilm systems gradients of electron acceptors such as
Fig. 3 e Biofilm solids concentrations at three sites (influent end, middle, effluent end) along the aerobic bioreactor in the IFAS process train from February 2008 to January 2009.
dissolved oxygen and nitrate can potentially cause stratification of microorganisms exhibiting vastly different metabolisms (Okabe et al., 1996). Similar relationships between temperature and biomass on the carriers have been observed by Rutt et al. (2006) and Maas et al. (2008). Bjonberg et al. (2010) also reported an almost inverse relationship between biomass and operating temperature at The Moorhead, Minnesota Wastewater Treatment Facility (WWTF) operating a moving-bed biofilm reactor (MBBR). Boltz and Daigger (2010) have attributed this relationship to macronutrient and other substrates not being exhausted immediately under cold water conditions resulting in biochemical processes occurring deeper into the biofilm resulting in thicker and heavier biomass coverage. In the case of nitrifying biofilm on IFAS media at higher temperature the depth of NH3 diffusion is reduced as biochemical processes occur faster at the interface. As a response to cold weather the MBBR/IFAS biofilm mass increases as a consequence of slower biochemical transformations (Boltz and Daigger, 2010). The biomass on the media accounted for on average 38% of the total biomass within the IFAS zone, which is a significant amount considering that biofilm carriers were present at 50% fill with 88% void spaces (Table 4). During the winter months the biomass on the media was comparable to the suspended biomass while the quantity was far less during the summer months (Fig. 4). In a completely different situation the biomass on the media increased during the month of June 2008 (Fig. 3). This increase occurred due to the introduction to the IFAS train of an external organic substrate (to improve denitrification), which after introduction to the system was found to be high in yeast. Rapid microbial growth on the media was microscopically confirmed to be due to yeast
Table 4 e Amount Suspended and Attached Biomass and Attached Fraction (Period: February 2008 to February 2009).
Average Max Min
Suspended biomass, kg
Attached biomass, kg
Attached biomass/ total biomass
2280 6102 1250
1383 2919 558
38% 68% 10%
6704
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 9 9 e6 7 0 8
during the yeast episode. Therefore, it appears that at higher biomass concentrations recycle NO 3 concentrations will be decreased and the total nitrogen removal will decrease. The episode highlights the importance of successful monitoring of the biomass on the media to resolve situations such as this one that was encountered during the study period. The sixweek-long yeast problem was resolved by over-aerating the IFAS tank and raising the pH to encourage more nitrification (Rutherford, 2010).
3.2.
Fig. 4 e Temporal changes in biomass on suspended and attached biomass phase in the IFAS zone.
proliferation causing process upset in terms of nitrogen removal most notably observed in the effluent ammonia concentration (Fig. 5). During the period when the media was at its thickest coverage, the media migrated and accumulated at the effluent end of the aerobic basin. This condition likely inhibited water movement through the media compared to when the media was more evenly distributed throughout the aerobic basin and highly mixed. The high approach velocities of the wastewater coming through the anoxic to the aerobic zone are experienced due to the low width to length ratio of the IFAS bioreactor and coupled with the operation of the IMLR 2 pump, which pulled the media toward the downstream end of the aerobic basin. By design and to counteract this effect, the IMLR 1 pump was operated at the upstream end of the aerobic basin. However, during the period of high biomass coverage on the media the IMLR 1 pump was not able to keep the media from migrating toward the downstream end of the aeration basin. Consequently, the IMLR 2 pumping rate was decreased and the flow rate of the IMLR 1 pump was increased. The compounding effect of operating the IMLR 1 pump at much higher flow rate was that in the upstream end of the aeration basin (OX1) the NO 3 concentration was significantly lower than at the downstream end (OX4). Considering that the NO 3 recycle is the backbone for successful nitrogen removal by the MLE process, the low NO 3 concentration in the recycle had a negative effect on total nitrogen removal performance
Fig. 5 e Temporal variation of primary effluent TKN, IFAS Effluent TN and IFAS effluent ammonia during yeast event.
Nitrogen removal
Nitrogen removal across the IFAS tank was evaluated from profiling data collected during the study and significant removal of nitrogen was observed from February 2008 to June 2009 (Fig. 5). The effluent total dissolved nitrogen averaged around 12.8 mg/L (profiling period) which slightly exceeded the target of 12 mg/L. On average the percent nitrogen removal achieved by the IFAS demonstration tank was sixty-seven, and ranged between 39 and 89 percent. The effluent ammonia concentration remained low, averaging 1.6 mg/L, signifying extensive nitrification during the nutrient profiling period. There were several days with effluent ammonia below detection indicating complete nitrification. As observed with nitrifying/denitrifying treatment processes with internal mixed liquor recycle similar to the IFAS demonstration process, the dominant species of nitrogen in the effluent was NO 3 . With nitrification being relatively stable in the IFAS tank during the study period it was the extent of denitrification that controlled the effluent total nitrogen concentration. The nitrogen species at seven sampling sites across the demonstration tank illustrate transformations that occurred (Fig. 6). The total nitrogen concentration did not change dramatically over the length of the demonstration tank, but the nitrogen species distribution did. The effluent ammonia concentration remained low, on average 1.4 mg/L, while NO 3 concentration averaged 8.4 mg/L signifying extensive nitrification. Nitrite concentrations generally remained low averaging 0.2 mg/L across all sites illustrating that there was no nitrite accumulation in the demonstration tank. This observation suggests a stable population of AOB and NOB were present causing extensive conversion of ammonia to nitrate. The soluble organic nitrogen concentration also remained low in the effluent. On average there was 0.6 mg N/L of soluble organic nitrogen leaving the aeration basin to the secondary clarifier. The fact that there was an average of 2 mg/L of soluble organic-N entering the aerobic basin and only 0.6 mg/L leaving illustrates that ammonification or direct uptake of organic nitrogen was occurring across the IFAS reactor. The measurement of nitrogen species across the IFAS process train allowed assessment of the denitrifcation that occurred in the anoxic zone. On average about 54% of NO 3 was denitrified. Denitrification was limited and there was always some effluent NO 3 escaping the anoxic basin to the aerobic basin (Fig. 6). The rate of denitrification in a pre-denitrification system may be governed by the NO 3 concentration, the bioavailable organic matter concentration and/or by dissolved oxygen concentration in a denitrifying anoxic reactor. At NO 3 concentrations of approximately 3 mg/L and above, the denitrification rate will be entirely controlled by the quantity
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 9 9 e6 7 0 8
6705
Fig. 6 e Average concentrations of nitrogen species and dissolved oxygen from June 2008 to January 2009 at the seven sampling sites.
and type of biodegradable organic matter (Rusten et al., 1995). The COD/TKN ratio during the study period was on average 7.0 which is lower than the desired COD/TKN ratio (>9) for higher nitrogen removal efficiencies (Grady et al., 1999).
3.3.
Nitrification kinetics and in-basin nitrification rates
The results of nitrification rate testing (AOB and NOB activity) during the study period for IFAS mixed liquor, the IFAS media, and the media and mixed liquor combined is presented in Fig. 7. As noted previously, this effort was conducted to determine where nitrification activity was taking place in the process (i.e. suspended or attached) and how the operational and seasonal conditions influenced nitrogen transformations. The average AOB activity on the IFAS media during the study was 4.97 mgNOx/gMLSS/hr which was much higher than the average of 1.72 mgNOx/gMLSS/hr for the IFAS mixed liquor. The NOB activity on the media averaged 7.55 mgNOx/gMLSS/ hr much higher than the IFAS mixed liquor, which was only 0.82 mgNOx/gMLSS/hr on average. In an IFAS study OnnisHayden et al. (2011) reported 75% nitrification activity residing on the carried media rather than in the mixed liquor which is similar to the distribution of activity observed in this study. The AOB activity in the IFAS suspended culture was present throughout the study period, but the activity was substantially lower in suspension compared to the media during colder temperatures (Fig. 7). The fluctuations in AOB activity were more gradual in IFAS media as opposed to mixed liquor where it drastically increased with warmer temperatures. The fact that there was presence of AOB activity year round in the IFAS mixed liquor, even in low SRT in winter, suggests that sloughing of biofilm biomass was seeding the mixed liquor continuously. Unlike AOB activity, the NOB activity resided attached to the media even in warmer temperature (Fig. 7). In the full-scale plant (conventional activated sludge (CAS): average flow of 96,700 m3/d, MLSS of
2483 mg/L, SRT of 3.7 days and temperature of 21.4 C for the month November) profiling revealed that the NOB activity in November 2008 was much lower than AOB activity leading to nitrite accumulation (CAS NO 2 N ¼ 11.2 mg N/L measured on 11/18/2008). In contrast, there was no nitrite accumulation in the IFAS zone throughout the study (Fig. 6) as nitrite
Fig. 7 e Nitrification rates for AOB and NOB experiments.
6706
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 9 9 e6 7 0 8
produced by the AOB activity was immediately converted to nitrate by NOB activity. The half-saturation constants for microorganisms on the biofilm carriers were modeled using Monod kinetics using the results of the nitrification kinetic studies (Thomas, 2009). The calculated KS values in the IFAS process were higher than accepted KS (0.7 mg N/L for AOB and 0.05 mg N/L for NOB according to Envirosim (2008)) values for mixed liquor in activated sludge systems as expected. Modeled NOB effective KS values for the carriers were much higher (2168%) than KS of a typical suspended growth culture. This suggests nitrite accessibility within the biofilm is greatly limited by substrate diffusion. Compared to ammonia and oxygen, nitrite has a lower diffusivity, which explains the high effective NOB halfsaturation coefficient in the attached growth systems. It also supports the hypothesis that NOB reside deeper within the biofilm. Unlike the NOB KS value, the AOB KS value was only 33% higher than what is accepted for suspended cultures of microorganisms. This indicates ammonia was more available to biofilm and supports the hypothesis that AOB populations inhabited the surface of the biofilm where substrate diffusion has less effect. It also explains the lower NOB activity in the IFAS mixed liquor that may be attributed by the NOB population growing deeper into the biomass and detaching less than AOB population. The lower NOB population in the mixed liquor may also depend on the fact that NOB have a lower maximum specific growth rate than AOB at temperatures higher than 15 C. The difference between nitrite and ammonia concentration should be also considered. Nitrite concentrations were everywhere lower, so k-strategist rather than rstrategist NOB could be favored. In a molecular analysis using florescent in-situ hybridization (FISH) at the Broomfield WWTP Nitrospira spp. was observed to be the most abundant NOB (Onnis-Hayden et al., 2007). Schramm et al. (1999) proposed that Nitrospira spp. might be k-strategist capable of reaching higher population densities at weaker nitrite environments. This is again suggesting that NOB require a higher SRT and explains why they are more prevalent in attached form. Close monitoring of the nitrogen species across the IFAS process train allowed the calculation of nitrification rates for each sampling date over the period of the nutrient profiling effort. The calculated nitrification rates allow an indirect evaluation of nitrifying activity. In-basin nitrification rates
Fig. 9 e Nitrification rates measured along the IFAS tank.
were calculated based on the NO X produced rather than ammonia removed and represents the difference in NO X measured at sites AN3 and OX4. The in-basin nitrification rate, which is based on a mass balance of dissolved nitrogen species across the aerobic basin, includes ammonia removal through assimilation by attached and suspended microorganisms. The nitrification rate averaged 0.90 gNOxeN/m2/ d (Fig. 8); samples less than 1 mg N/L ammonia were excluded from the average rate calculation since low ammonia concentrations achieved before the end of the tank would artificially lower the nitrification rate due to the availability of ammonia. This value was in good agreement with the average nitrification rate of 0.89 gNOxeN/m2/d for IFAS media and mixed liquor observed from the kinetic testing phase. OnnisHayden et al. (2007) reported a nitrification rate of 1.12 gNOxeN/m2/d for the Broomfield IFAS demonstration that was similar to the IFAS demonstration at JRTP. The carrier biomass was measured to be constant along the length of IFAS tank; however, profiling of the IFAS tank clearly suggested concentration gradients of nitrogen species (ammonia, nitrite and nitrate) over the length of the tank. This prompted assessment of IFAS media nitrification rates over the length of IFAS tank. The results indicate that the effect of location was not significant on the carrier nitrification potential (Fig. 9). This high activity on the IFAS carriers throughout of tank volume can be advantageous over a two reactor design where lower nitrification rates are observed in the downstream reactor. The slightly higher nitrification rates at OX3 and OX4 might be attributed to higher dissolved oxygen in the bench scale test reactors (Fig. 9).
4.
Fig. 8 e Biofilm solids and in basin nitrification rates for the IFAS train during the study period.
Conclusion
A CAS train at JRTP was converted from carbonaceous removal to nitrification/denitrification within the existing footprint by dividing the tank into anoxic and aerobic compartments and installing the free-floating plastic carriers in the aerobic stage of the reactor. Carrier biomass was monitored throughout the study, which helped assess biofilm conditions. Biofilm density was constant in three sampling locations, which indicate that biomass growth was homogeneous throughout the IFAS tank. Similarly, uniform
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 9 9 e6 7 0 8
nitrification activity was observed independent of location in the IFAS tank. The biomass growth on the media exhibited a seasonal pattern where biomass coverage was highest during winter months and lowest during summer excluding the yeast event. The decrease of biomass occurred while the SRT of mixed liquor remained fairly stable thus the trend appears independent of suspended culture conditions. The observed increase in suspended biomass (Fig. 4) was due to the rapid loss of biomass from the carrier media toward the end of the yeast event. As temperature cooled in the basin in the late autumn and early winter, biomass content on the carriers steadily increased from about 3 g MLSS/m2 to about 7 g MLSS/m2. This increase in biomass should correlate directly with increased nitrifier abundance on the media. Another possible reason for increased biomass on the carriers could be a reduction in mixing input to the tank. At low temperatures, less aeration was required to maintain dissolved oxygen in the tank, so the decreased aeration intensity likely also resulted in decreased agitation of the media and thus less sloughing of biomass. From June 2008 to January 2009 the demonstration tank was monitored for nitrogen species at seven sampling sites along the length of the IFAS process train. The results of profiling have shown that under different loading and temperature conditions the IFAS process train was able to achieve significant removal of total nitrogen. The effluent TDN level averaged 12.8 mg/L during the nutrient profiling period while effluent ammonia averaged 1.6 mg/L during the same time. Nitrate was the most dominant nitrogen species at the effluent averaging 9.8 mg/L. Soluble organic nitrogen in the treated effluent was consistently less than 0.6 mg/L across the range of load and temperature conditions. During the study period aerobic nitrification was consistently over 87 percent whereas anoxic denitrification resulted in 51 percent conversion of nitrate. There was high nitrate concentration (average 4.0 mg/L) in the anoxic effluent which indicates that denitrification was incomplete. The fact that the COD/TKN ratio of influent wastewater remained low averaging around 7.0 and significant dissolved oxygen was recycled from the IFAS zone by two IMLR pumps indicating that denitrification was COD limited. Consequently, one way to improve the total nitrogen removal efficiencies is to improve the anoxic denitrification through augmentation of available biodegradable organic matter. The activities of both ammonia oxidizing and nitrite oxidizing were elevated on the carriers compared to the suspended culture (AOBmedia: 4.97 mgNOx/gMLSS/hr; AOBsuspended: 1.72 mgNOx/gMLSS/hr; NOBmedia: 7.55 mgNOx/gMLSS/hr; NOBsuspended: 0.82 mgNOx/gMLSS/hr) during all periods of the study. In-basin nitrification rates calculated based on nitrogen profiling efforts averaged 0.90 mgNOx/m2/day which was in good agreement with the average of 0.89 mgNOx/m2/day for IFAS mixed liquor and media from batch testing.
Acknowledgment The authors would like to thank Mr. Jim McQuarrie, CH2MHILL (at the time) and Mr. Rick Baumler, Hampton Roads Sanitation
6707
District for their invaluable input. Funding for this study was provided by the Hampton Roads Sanitation District.
references
Bjonberg, C., Lin, W., Zimmerman, R., 2010. Kinetic evaluation and model simulation of temperature impact on biofilm growth and nitrification in a full-scale MBBR system. Proceedings of the Water Environment Federation, 4146e4171. Boltz, J.P., Daigger, G.T., 2010. Uncertainty in bulk-liquid hydrodynamics and biofilm dynamics creates uncertainties in biofilm reactor design. Water Science and Technology 61 (2), 307e316. Christensson, M., Welander, T., 2004. Treatment of Municipal Wastewater in a Hybrid Process Using a New Suspended Carrier with Large Surface Area, pp. 207e214. IWA Publishing. Envirosim, 2008. BioWin Version 3.0 Default Parameters. Grady, C.P.L., Daigger, G.T., Lim, H.C., 1999. Biological Wastewater Treatment. Marcel Dekker, New York. Hubbell, S.B., McDowell, C., 2003. Cold temperature BNR using integrated fixed film activated sludge (IFAS) hybrid technology. Proceedings of the Water Environment Federation, 162e172. Maas, C.L.A., Parker, W.J., Legge, R.L., 2008. Oxygen uptake rate tests to evaluate integrated fixed film activated sludge processes. Water Environment Research 80 (12), 2276e2283. Ødegaard, H., Gisvold, B., Strickland, J., 2000. The influence of carrier size and shape in the moving bed biofilm process. Water Science and Technology 41 (4e5), 383e391. Ødegaard, H., 2006. Innovations in wastewater treatment: the moving bed biofilm process. Water Science and Technology 53 (9), 17e33. Okabe, S., Hiratia, K., Ozawa, Y., Watanabe, Y., 1996. Spatial microbial distributions of nitrifiers and heterotrophs in mixed-population biofilms. Water Science and Technology 50, 24e35. Okey, R., Albertson, O., 1989. Evidence for oxygen-limiting conditions during tertiery fixed film nitrification. Journal Water Pollution Control Federation 61, 510e519. Onnis-Hayden, A., Dair, D., Johnson, C., Schramm, A., Gu, A.Z., 2007. Kinetcis and nitrifying populations in nitrogen removal processes at a full-scale integrated fixed-film activated sludge (IFAS) plant. Proceedings of the Water Environment Federation, 3099e3119. Onnis-Hayden, A., Majed, N., Schramm, A., Gu, A.Z., 2011. Process optimization by decoupled control of key microbial populations: distribution of activity and abundance of polyphosphate-accumulating organisms and nitrifying populations in a full-scale IFAS-EBPR plant. Water Research 45 (13), 3845e3854. Ross, D., Fernandes, W., Briggs, T., Kim, N., Booth, G., Neely, D., Welp, J., 2004. Integrated fixed film activated sludge (IFAS) at the lakeview WWTP the real implementation issues. Proceedings of the Water Environment Federation, 286e298. Rutherford, B., 2010. IFAS enables Virginia plant to meet permit demands. Water Environment & Technology, 64e69. May 2010. Rusten, B., Hem, L.J., Degaard, H., 1995. Nitrogen removal from dilute wastewater in cold climate using moving-bed biofilm reactors. Water Environment Research 67, 65e74. Rutt, K., Seda, J., Johnson, C.H., 2006. Two year case study of integrated fixed film activated sludge (IFAS) at broomfield, CO WWTP. Proceedings of the Water Environment Federation, 225e239. Schramm, A., de Beer, D., van den Heuvel, J.C., Ottengraf, S., Amann, R., 1999. Microscale distribution of populations and
6708
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 6 9 9 e6 7 0 8
activities of nitrosospira and Nitrospira spp. along a macroscale gradient in a nitrifying bioreactor: quantification by in situ hybridization and the use of microsensors. Applied and Environmental Microbiology 65 (8), 3690. Sriwiriyarat, T., Randall, C.W., 2005. Performance of IFAS wastewater treatment processes for biological phosphorus removal. Water Research 39 (16), 3873e3884. Stricker, A.E., Barrie, A., Maas, C.L.A., Fernandes, W., Lishman, L., 2009. Comparison of performance and operation of side-byside integrated fixed-film and conventional activated sludge processes at demonstration scale. Water Environment Research 81 (3), 219e232. Thomas, W.A., Bott, C.B., Regmi, P., McQuarrie, J., Rutherford, B., Baulmer, R., Waltrip, D., 2009. Evaluation of nitrification
kinetics for a 2.0 MGD IFAS process demonstration. Proceedings of the Water Environment Federation. Thomas, W.A. 2009 Evaluation of Nitrification Kinetics for a 2. 0 MGD IFAS Process Demonstration; Master of Science thesis; Environmental Engineering Department; Virginia Polytechnic Institute and State University: Blacksburg, VA. Wheaton, F., Hochheimer, J., Kaiser, G., Krones, M., 1994. Nitrification Principles: Aquaculture and Reuse Systems. Elsevier, Amsterdam. Wuertz, S., Okabe, S., Hausner, M., 2004. Microbial communities and their interactions in biofilm systems: an overview. Water Science and Technology 49 (11e12), 327e336. Zhu, S., Chen, S., 2002. The impact of temperature on nitrification rate in fixed film biofilters. Aquacultural Engineering 26, 221e237.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 0 9 e6 7 1 7
Available online at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Reconnaissance of selected PPCP compounds in Costa Rican surface waters Alison L. Spongberg a,*, Jason D. Witter a, Jenaro Acun˜a b, Jose´ Vargas b, Manuel Murillo b, Gerardo Uman˜a b, Eddy Go´mez b, Greivin Perez c a
Department of Environmental Sciences, University of Toledo, 2801 W. Bancroft St, Toledo, OH 43606, United States Centro de Investigacio´n en Ciencias del Mar y Limnologı´a (CIMAR), Universidad de Costa Rica, San Jose´, Costa Rica c Centro de Investigacio´n en Contaminacion Ambiental (CICA), Universidad de Costa Rica, San Jose´, Costa Rica b
article info
abstract
Article history:
Eighty-six water samples were collected in early 2009 from Costa Rican surface water and
Received 27 January 2011
coastal locations for the analysis of 34 pharmaceutical and personal care product
Received in revised form
compounds (PPCPs). Sampling sites included areas receiving treated and untreated
5 October 2011
wastewaters, and urban and rural runoff. PPCPs were analyzed using a combination of
Accepted 8 October 2011
solid phase extraction and liquid chromatography tandem mass spectrometry. The five
Available online 17 October 2011
most frequently detected compounds were doxycycline (77%), sulfadimethoxine (43%), salicylic acid (41%), triclosan (34%) and caffeine (29%). Caffeine had the maximum
Keywords:
concentration of 1.1 mg L1, possibly due to coffee bean production facilities upstream.
Costa Rica
Other compounds found in high concentrations include: doxycycline (74 mg L1), ibuprofen
Pharmaceutical and personal care
(37 mg L1), gemfibrozil (17 mg L1), acetominophen (13 mg L1) and ketoprofen (10 mg L1).
products
The wastewater effluent collected from an oxidation pond had similar detection and
Liquid
chromatography
mass spectrometry
tandem
concentrations of compounds compared to other studies reported in the literature. Waters receiving runoff from a nearby hospital showed higher concentrations than other areas for many PPCPs. Both caffeine and carbamazepine were found in low frequency compared to other studies, likely due to enhanced degradation and low usage, respectively. Overall concentrations of PPCPs in surface waters of Costa Rica are inline with currently reported occurrence data from around the world, with the exception of doxycycline. Published by Elsevier Ltd.
1.
Introduction
The active ingredients in pharmaceutical and personal care products (PPCPs) and veterinary antibiotics have increasingly been detected in a wide variety of environmental matrices. These include surface and groundwaters, wastewater treatment end products (effluents, reclaimed waters and sludges), soils and biota in the United States and Europe (Kolpin et al., 2002; Loos et al., 2009; Martı`nez-Carballo et al., 2007;
Ramirez et al., 2009). The widespread therapeutic and preventative use in both human and animal populations of products containing these active ingredients and their incomplete elimination in both the body and conventional wastewater treatment has resulted in their environmental introduction. Major pathways into the aquatic environment for these compounds include runoff from areas where both animal and human waste is not confined and treated (i.e landfills, manure piles, land application of sewage sludges),
* Corresponding author. Tel.: þ1 419 530 4091; fax: þ1 419 530 4421. E-mail addresses:
[email protected],
[email protected] (J.D. Witter). 0043-1354/$ e see front matter Published by Elsevier Ltd. doi:10.1016/j.watres.2011.10.004
(A.L.
Spongberg),
[email protected]
6710
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 0 9 e6 7 1 7
and the direct discharge of untreated wastewater and treated wastewaters (effluent) into surface waters. Once released into the environment many of these compounds persist and can possibly be transported to locations far from the source (e.g. Walters et al., 2010; Wu et al., 2009). Research has mostly focused on the occurrence of PPCPs in temperate environments, with only limited studies being conducted in the warm and humid sub-tropical and tropical climates. Managaki et al. (2007) analyzed the occurrence of twelve veterinary antibiotics in waters from the Mekong Delta, Viet Nam and the urban Tamagawa River in Japan. Increased detections and higher concentrations were found for the Japanese sampling compared to the Mekong Delta for most analyzed compounds, although some PPCPs were common in both locations. Locatelli et al. (2010) surveyed Brazilian surface waters for eight antibiotic compounds and found that usage and sample collection during dry versus wet season determined the distribution and occurrence patterns. Martins et al. (2008) analyzed Brazilian hospital effluent for the presence of ciprofloxacin, a fluoroquinolone antimicrobial agent. The high concentrations detected caused the authors to conclude that the risk associated with the use and emission of pharmaceuticals into the environments of developing countries might be higher than in developed countries. Siemens et al. (2008) reported detection of several acidic and basic pharmaceutical compounds surviving the treatment process in reclaimed wastewaters used as an irrigation source in Mexico City. The inefficient or lack of treatment of wastewaters in developing countries can lead to increased introduction into the environment of these “emerging” contaminants. Costa Rica is a moderately developed country with major urban and vast rural areas. Undeveloped land, rural agricultural areas and highly contaminated sites where identified during previous studies on PCB and pesticide residues (Spongberg and Davis, 1999; Spongberg, 2004). Wastewater treatment in Costa Rica can range from modern regional or city level treatment plants, to areas with primary treatment only, to discharge of untreated waste into local waterways by runoff or pipe. No previous study has been conducted in Costa Rica to assess the occurrence of PPCPs in surface waters. The goal of this study was to detail the occurrence of PPCPs in Costa Rican surface waters, a tropical country, with relation to previous sampling sites used for PCB and pesticide determination, wastewater sources and potentially contaminated runoff. Thirty-four PPCPs ranging in therapeutic class and usage were chosen for analysis. Compounds were selected based on occurrence data reported in other similar large-scale studies, established analytical methodology in our laboratory and suspected usage in Costa Rica.
2.
Materials and methods
2.1.
Nationwide sampling
Eighty-six sampling locations, along with one wastewater effluent sample, were selected from a wide range of localities within Costa Rica and are presented in Fig. 1 with detailed descriptions of the sites listed in Table 1 and S-1 (supplementary material). Site locations were based on
access, previous survey for other compounds, correlation with other studies, or location near a possible source of contamination (wastewater treatment plant, city waste discharge, hospital discharge, urban runoff etc.). Sampling points were recorded using a GPSMAP 76 global positioning satellite receiver (Garmin, Olathe, KS, USA).
2.2.
Sample collection
One liter of surface water from both salt and freshwater environments was collected in high-density polyethylene (HDPE) sampling bottles (Fisher Scientific, Pittsburg, PA, USA). Prior to sampling bottles were washed with dilute hydrochloric acid and methanol. At each sampling site bottles used were rinsed and shaken twice with a full volume of surface water. When necessary bulk samples were collected using a 5liter bucket, then transferred to the 1-liter HDPE containers. All samples were stored on ice until being processed at the Centro de Investigacion en Contaminacion Ambiental (CICA), located at the University of Costa Rica (typically less than 48 h). Once in the laboratory samples were filtered through 47 mm, 0.7 micron glass fiber filters (Fisher Scientific, Pittsburg, PA, USA) using a vacuum apparatus, containers were then rinsed with a 50% (v:v) methanol in water solution, combined and subsequently extracted using solid phase extraction.
2.3.
Chemicals and reagents
All pharmaceutical standards (purity, 90%w99%) were purchased from SigmaeAldrich (St. Louis, MO), except clarithromycin (purity, 98%), obtained from Abbott (Chicago, IL). Instrumental internal standards 13C3-Caffeine (purity, 99%), josamycin (purity, 98%), and 2-(3-chlorophenoxy) propionic acid (purity, 99%) were also obtained from SigmaeAldrich (St. Louis, MO) and simatone was obtained from AccuStandard (New Haven, CT). All other chemicals and solvents were American Chemical Society certified or HPLC grade and supplied by Fisher Chemicals (Fair Lawn, NJ). Deionized water (18.3 MU) was provided by a Barnstead NANOpureÒ Infinity Water System (Dubuque, IA).
2.4.
Solid phase extraction (SPE)
SPE was conducted according to the method reported in Wu et al. (2008). For all samples an aliquot of 350 mL was transferred to a glass container. Twenty-eight milligrams of Na2-EDTA was added and allowed to dissolve while mixing. Sample pH was then adjusted to 5 using H2SO4 and/or 5% (v:v) NH4OH in water. For SPE, Phenomenex Strata X polymeric cartridges, 6 mL 200 mg packing (Torrance, CA, USA), were conditioned three times with 2 mL methanol then three times with 2 mL deionized water containing 1% (w/v) Na2-EDTA in water. Each 350 mL sample aliquot was loaded into the SPE cartridge at a rate of 10 mL min1 using large volume sampling tubes connected to a 24 port SPE vacuum manifold (Phenomenex, Torrance, CA, USA). After loading, cartridges were washed with 2 mL of 5% (v/v) methanol in water and dried under vacuum for 2 min. The analytes were then eluted twice with 3 mL methanol without the use of
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 0 9 e6 7 1 7
6711
Fig. 1 e Overall map showing sampling locations, surface hydrography and points of interest in Costa Rica, Central America (10N, 85W).
vacuum. The eluate was collected in a glass conical vial and evaporated using nitrogen and a water bath at 40 C to around 200 mL using a Turbo Vap LV (Caliper Life Sciences, Hopkinton, MA, USA). Samples were spiked with 100 ng instrumental internal standards, reconstituted to 0.45 ml using 50% (v/v) methanol in water, vortexed and transferred to 2 ml amber glass vials. SPE extracts were stored at 20 C until instrumental analysis.
2.5. Liquid chromatography mass spectrometry analysis (LC-MS/MS) The LC-MS/MS system consists of a ProStarÒ 210 solvent delivery module with a ProStar 430 autosampler and a 1200L triple-stage quadrupole mass spectrometer with a dual offaxis electrospray ionization interface (Varian Inc., Walnut Creek, CA). Analytes were separated using a Supelco DiscoveryÒ HS C18 column (150 4.6 mm, 3 mm). The column was maintained at 25 C using a ThermaSphere TS-430 Column Chiller/Heater (Phenomenex, Torrance, CA, USA). Mobile phase A was 0.1% (v/v) formic acid in water, mobile phase B was 100% acetonitrile and the total flow rate was 0.3 mL min1. The gradient started with 5% B, held for 2 min, ramped to 100% in 18 min, held 10 min, dropped to 5% B in 2 min, and equilibrated for 8 min. The precursor ions and two most abundant transition ions are provided in Table 2.
Detailed MS/MS parameters and method development procedure is presented elsewhere (Wu et al., 2008).
2.6.
Quantification and method validation
Instrument control, peak detection and integration were accomplished using Varian MS Workstation (Version 6.8). Data acquisition was performed under multiple reaction monitoring (MRM) mode. Identification of the target analytes was based on the presence of two MRM transitions and match of retention time with the reference standard. The ratio of two MRM transitions was used for confirmation. The most abundant transition was selected for quantification. Instrumental internal standards (Wu et al., 2008) were added prior to LCMS/MS analysis, but after SPE, to compensate for instrumental and ionization variation (matrix effect). Ratios of the analyte peak areas to appropriate internal standard peak area were used to construct calibration curves and for sample extract quantification. Powdered standards were dissolved in 450 mL of methanol at five concentration levels (10e500 mg L1), including 100 ng internal standards, to create calibration curves for external quantification. All calibration curves were linear (r2 > 0.98) between 10 and 500 mg L1. Method blanks and reagent water used in the extractions were also run for quality assurance, and data were adjusted for any carryover and background accordingly. A quality control sample (50 mg L1 calibration standard) was run every six injections, and the
6712
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 0 9 e6 7 1 7
Table 1 e Detailed sampling location information for Costa Rica study. CRP Type Influence 1 2 3 7 8
Fresh Fresh Fresh Tap Salt
Rural Rural Rural Urban Urban
9
Salt
Urban
10 11 12
Salt Salt Salt
Urban Urban Urban
13 14
Salt Salt
Urban Urban
15 16 17
Fresh Salt Fresh
Rural Rural Rural
18 19 20 21 22 23 24 25 26
Fresh Fresh Fresh Fresh Fresh Fresh Fresh Salt Salt
27 28 29 30 31 32 33 34
Salt Salt Salt Salt Salt Salt Salt Salt
Urban Effluent Rural Urban Rural Rural Urban Urban Open water Urban Urban Urban Urban Rural Rural Rural Urban
35
Salt
Urban
36 37
Salt Salt
38 39 40
Description Rı´o Para´ Tributary to Rı´o Virilla Rı´o San Miguel UCR Campus Puntarenas Estuary, exit, near the Ferry Puntarenas beach, near cruise ship Puntarenas dock Puntarenas, 1 km east of dock Puntarenas, Fertica channel, dry dock Puntarenas, Fertica channel Puntarenas, north side of peninsula Rı´o Barranca, end of low tide Puntarenas, Port Caldera Puntarenas, Port Caldera, Mata de Limo´n San Isidro, upstream of wastewater San Isidro, downtream of wastewater Dominical Beach Corte´s, Rı´o Balsar, upstr Osa hospital Rı´o Te´rraba, Palmar Norte Rı´o Esquinas, from Road Golfito, Rı´o Can˜azas Golfo Dulce, Punta Gallardo Golfo Dulce, Punta Gallardo
CRP Type Influence Urban Effluent Urban Urban Rural
Liberia, near oxidation ponds Liberia, WWT oxidation pond Liberia, Rı´o Liberia, upstream of WWT Rı´o Liberia, near hospital Can˜as, Tilapias El Sol, irrigation canal
53 Fresh
Rural
Can˜as, Rı´o Corobicı´
54 Fresh 55 Fresh 56 Salt
Rural Rural Rural
Can˜as, irrigation channel Arenal, Bahı´a San Luis Parque Nacional Manuel Antonio, beach
57 Fresh 58 Fresh
Urban Rural
Quepos, downstream of hospital Quepos, upstream of hospital
59 Fresh 60 Fresh 61 Fresh
Urban Urban Rural
Quepos, downstream of city, low tide Jaco Rı´o Ta´rcoles
62 63 64 65 66 67 68 69 70
Fresh Fresh Fresh Salt Fresh Fresh Fresh Fresh Fresh
Urban Rural Rural Rural Rural Rural Rural Rural Rural
Puntarenas, Talmana Estuary, residential Rı´o Bebedero, falling tide Rı´o Tempisque Golfo de Nicoya, ECMAR dock, high tide Rı´o Lagarto Santa Marta, Te´rraba-Sierpe wetland El Caite, Te´rraba-Sierpe wetland Rı´o Te´rraba, Samu, Te´rraba-Sierpe wetland Isla Loros, Te´rraba-Sierpe wetland
71 72 73 74 75 76 77 78
Fresh Salt Salt Salt Salt Fresh Fresh Fresh
Rural Urban Urban Urban Urban Urban Urban Urban
Golfo Dulce, Golfito, Isla Pelı´cano
79 Fresh
Urban
Urban Urban
Golfo Dulce, Golfito, Purruja Estuary Golfo Dulce, Golfito
80 Fresh 81 Fresh
Urban Urban
Salt Salt Salt
Urban Urban Urban
Golfo Dulce, Golfito, near cemetary Golfo Dulce, Golfito, municipal pier Golfo Dulce, Golfito, hospital
82 Fresh 83 Fresh 84 Fresh
Rural Urban Urban
41 42
Fresh Fresh
Urban Urban
Golfo Dulce, Golfito, urban drainage Golfo Dulce, Rı´o Coto-Colorado, near ferry
85 Fresh 86 Fresh
Urban Urban
43 44
Fresh Fresh
Urban Urban
Golfo Dulce, Golfito, ditch drainage for hospital 87 Fresh Rı´o Corredores, Neily, downstream of hospital 88 Fresh
Urban Urban
45 46
Fresh Fresh
Urban Urban
89 Fresh 90 Fresh
Urban Urban
47
Fresh
Rural
Rı´o Corredores, Neily Rı´o Java, San Vito, near animal feed factory Rı´o Te´rraba
91 Fresh
Urban
Rı´o Sierpe, Te´rraba-Sierpe wetland Limo´n, Cieneguita Limo´n, Vargas Park Limo´n, near hospital, accumulation of waste Limo´n, Moı´m pier Stream running through UCR campus Rı´o Torres, Barrio Tournon River at end of airport runway, Juan Santamarı´a Rı´o San Joaquı´n de Flores, near medical clinic Rı´o Pirro Rı´o Bermu´dez, btwn San Pablo-Santo Domingo Ciudad Quesada, upstream of city Ciudad Quesada, upstream of city, residential Ciudad Quesada, upstream, before confluence Ciudad Quesada, Rı´o Platanar, in city Ciudad Quesada, Rı´o Platanar, hospital and residential drainage Cartago, Rı´o Purires Cartago, Rı´o Reventado, downstream of slum homes Cartago, Quebrada Creek,center of city Cartago, Rı´o Agua Caliente, la Ciudad de los Nin˜os (Hervidero) Cartago, Rı´o Toyogres, drains san Rafeal de Oreamuno
Golfo Golfo Golfo Golfo Golfo Golfo Golfo Golfo
Dulce, Dulce, Dulce, Dulce, Dulce, Dulce, Dulce, Dulce,
Rı´o Tigre, coral reef Puerto Jime´nez Puerto Jime´nez near Hotel Cocodrilo Rı´o Coto-Colorado Rı´o Coto-Colorado, inland Coto-Colorado mouth Golfito, end of low tide
response factor was found to vary less than <5% RSD (relative standard deviation) for all compounds. Sample extracts with compound concentrations outside of the linear range (previously given) were diluted with methanol and rerun. Final
48 49 50 51 52
Fresh Waste Fresh Fresh Fresh
Description
concentrations presented here have been adjusted for any dilution and concentration factors utilized during analysis. SPE extraction recoveries using fresh and saltwater were generated by fortifying 350 mL of each matrix in triplicate with
6713
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 0 9 e6 7 1 7
175 mL of a 200 mg L1 standard mix solution, yielding a final spiked concentration of 100 ng L1, within reported environmental ranges for surface waters for these analytes. Conductivity, total dissolved solids and salinity for the fresh and saltwater used for extraction recovery calculations were 0.78 mS m1, 0.26 g L1 and 0.35 g L1, and 32 mS m1, 28.58 g L1 and 19.57 g L1, respectively. Recoveries were typically above 70%, and variation was within 10% between fresh and saline matrices, with some exceptions. Limit of quantitation (LOQ) was calculated using the lowest detectable concentration on the instrument (IDL) with a signal to noise ration of 3 multiplied by the concentration factor divided by the sample volume. LOQs varied considerably between the first and second block of samples collected and analyzed (block 1; 1/9/2009 to 1/29/2009; block 2; 2/13/2009 to 2/16/2009). This difference was due to instrumental sensitivity problems during analysis of block 2 samples; therefore, two LOQ’s are given in Table 2.
3.
Results and discussion
3.1.
General trends
Surface waters collected from coastal and interior areas were analyzed for 16 classes of PPCPs, and Fig. 2 presents the percent of total measured concentration for each class. The predominant PPCP class detected in surface water samples was dominated by the central nervous system (CNS) stimulant (caffeine). This is not surprising since caffeine has been shown to be ubiquitous in waters affected by domestic wastewater pollution (Buerge et al., 2003), and is commonly found as an ingredient in food products. Other classes of note, in order of contribution, included tetracycline antibiotics, non-steroidal anti-inflammatory drugs (NSAIDs), the lipid regulator gemfibrozil (tradename Lopid) and the analgesic acetominophen. These trends are not surprising since all of
Table 2 e Classification, MS/MS ions, method extraction recovery (mean ± sd) and limits of quantitation (LOQ) for the analytes of interest in this study. Classificationa
Compound
MS Precursor Transition ions
LOQc [ng L1]
Recovery [%]
Quant Confirm Fresh water Salt water Fresh water Salt water Analgesic, antipyretic Anticonvulsant, antidepressant Antimicrobial Antohypertensive Bacteristatic CNS Stimulant Caffeine metabolite Fluoroquinolone
Acetominophen Carbamazepine Triclosan Diltiazem Trimethoprim Caffeine Paraxanthine Ciprofloxacin Norfloxacin Ofloxacin Histamine H2-Receptor antagonist Cimetidine Ranitidine Lincosomide Clindamycin Lincomycin Lipid regulator, metabolite Gemfibrozil Clofibric acid Macrolide Clarithromycin Erythromycinb Roxithromycin Tylosin NSAIDs Diclofenac Ibuprofen Indomethacin Ketoprofen Sulfanomide Sulfadimethoxine Sulfamethazine Sulfamethoxazole Sulfathiazole Tetracyclines Doxycycline Oxytetracycline Tetracycline b-lactam Oxacillin Penicillin G Skin care product ingredient Salicylic acid
152.2 237.1 286.8/289.0 415.1 291.1 192.2 181.2 332.1 320.1 362.1 253.1 315.1 425.2 407.1 249.2 213.1 748.5 716.4 837.6 916.4 293.8 204.9 358.0 255.1 311.1 279.1 254.1 256.1 445.1 461.2 445.1 402.1 335.1 136.8
110.0 194.0 35.1 178.0 230.0 138.0 124.0 314.0 302.0 318.0 159.0 176.0 126.0 126.0 121.1 126.8 158.0 158.0 158.0 174.0 249.8 161.0 139.0 209.0 156.0 186.0 156.0 156.0 428.0 426.0 410.0 143.9 127.9 92.8
93.0 192.0 149.9 230.0 110.0 96.0 230.9 230.9 344.0 117.0 130.0 359.1 85.0 590.3 558.0 679.3 156.0 159.0 174.0 105.0 108.0 124.0 92.0 92.0 320.8 443.1 153.9 186.0 160.0
97 115 79 82 85 85 78 55 51 64 78 54 87 102 91 89 79 99 90 70 113 98 95 75 82 74 88 103 79 109 79 77 65 60
17 4 6 3 8 9 8 8 8 2 13 3 2 11 16 2 10 8 13 5 3 3 2 8 5 3 4 6 14 11 8 8 6 12
a CNS: central nervous system; NSAIDs: non-steroidal anti-inflammatory drugs. b Determined as Erythromycin-H2O. c Two LOQ’s given based on change in instrument sensitivity for samples processed after 2/10/2009.
43 95 47 83 64 110 88 103 100 70 112 79 97 67 77 110 90 77 77 81 103 100 82 90 76 84 63 71 75 101 75 70 75 110
6 6 25 13 4 34 10 5 10 3 14 9 21 8 5 4 10 14 10 42 7 5 7 5 5 4 4 9 10 9 6 3 13 20
7, 37 1, 3 10, 91 1, 1 7, 10 18, 498 8, 12 21, 251 20, 498 22, 90 6, 9 5, 9 3, 12 1, 2 41, 275 11, 18 5, 10 8, 55 116, 157 974, 4909 12, 15 5, 984 7, 9 6, 20 1, 6 3, 7 11, 14 4, 7 18, 1607 <1, 156 44,4133 63, 6750 84, 100 11, 19
15, 17 1, 3 6152 1, 1 5,13 24, 386 7, 14 31, 133 38, 255 14, 82 6, 8 6, 7 3, 11 1, 2 35, 325 14, 15 5 10 11, 42 100, 183 844, 5666 14, 14 5, 969 6, 10 7, 17 1, 6 4, 6 8, 19 5, 6 19, 1693 <1, 144 2, 4353 70, 7425 86, 97 6, 47
6714
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 0 9 e6 7 1 7
Percent of Total Measured Concentration [%] 0.00
0.01
0.10
1.00
10.00
100.00
Analgesic and antipyretic (1) Anticonvulsant (1) Antihypertensive (1) CNS stimulant (2) H2-receptor antagonist (3) Lipid regulator (2) NSAID (4) Skin care (1) Antimicrobial (1) Bacteriostatic (1) -lactam (2) Lincosomide (2) Macrolide (4) Quinolone (3) Sulfanomide (4) Tetracyclines (3)
Fig. 2 e The percent of total measured concentration by drug class. Value in parenthesis indicates number of compounds in class.
these are widely available and relatively cheap, with possible heavy use in Central American countries. Table 3 provides the summary statistics for PPCP compounds analyzed in surface waters in this investigation. Complete data for each location can be found in Table S-2 in the supplementary material. Of the 34 compounds chosen for analysis, seven were not detected in any of the environmental water samples above the LOQ; clofibric acid, diltiazem, erythromycin, penicillin G, ranitidine, roxithromycin, and tylosin. The five compounds detected with greatest frequency were doxycycline (77%), sulfadimethoxine (43%), salicylic acid (41%), triclosan (34%) and caffeine (29%). The six compounds found in highest concentrations include: caffeine (1.12 mg L1), doxycycline (74 mg L1), ibuprofen (37 mg L1), gemfibrozil (17 mg L1), acetominophen (13 mg L1) and ketoprofen (10 mg L1). Detection frequency in general for all other compounds was low (<30%), and median values were below the LOQ for all but one compound (doxycycline). This presents the heterogeneous nature of PPCP contamination in Costa Rica, linked to untreated wastewaters from both urban and rural sources. This trend is similar to what has been found in other studies around the globe.
3.2.
Caffeine
Caffeine is often the compound reported with the highest frequency in similar studies and has previously been used as an indicator of anthropogenic contamination (Seiler et al., 1999; Buerge et al., 2003). In South Korea caffeine was detected in eight out of eight surface water samples and six out of seven effluent samples with a mean concentration of 105 ng L1 (Kim et al., 2007). In Costa Rican surface water samples caffeine was detected in only 29% of samples, likely due to rapid biodegradation. The presence of paraxanthine (20%), a metabolite of caffeine supports this idea. The highest concentration detected in any sample was for caffeine, with 1.1 mg L1, obtained from the sediment-laden waters of Rio Java downstream from San Vito (CRP-46). Upstream from this sampling location, coffee beans are washed frequently during processing, likely the source of the high concentration. Caffeine was found in locations with both rural and urban influences, albeit with low frequency and typically low concentrations. The 75th percentile concentration for caffeine
was 21 ng L1, well below the mean caffeine concentrations reported previously in Brazil (Ferreira, 2005), but similar in magnitude to those reported in surface water in Europe (Loos et al., 2009) and the United States (Kolpin et al., 2002). The concentrations reported here are higher than those typically found in wastewater effluent samples collected in the United States (e.g. Spongberg and Witter, 2008) or Europe (Buerge et al., 2003). Regular influxes of caffeine from point sources present in the drainage areas of rivers sampled, including coffee manufacturing facilities and rural wastewater, are likely, though the degradation properties are largely unknown in this tropical environment.
3.3.
Carbamazepine
Carbamazepine is a prescription medication used to control certain types of seizures, mental illnesses and depression and has been shown to persist in the aquatic and terrestrial environment (i.e. Tixier et al., 2003). Along with caffeine it has been proposed as an anthropogenic waste marker (Clara et al., 2004) and is typically detected in a majority of aquatic samples. Carbamazapine was detected in seven out of eight surface waters in South Korea with a mean concentration of 25 ng L1 (Kim et al., 2007). In Costa Rican surface waters carbamazepine was detected in only 10% of samples, with a maximum concentration of 82 ng L1(site 59, downstream of a large tourist area). Concentrations at other sampling sites were typically close to the LOQ (1e3 ng L1), indicating low usage for this compound in Costa Rica. This could be due to the availability and cost of the drug in this country. Prescription records and pricing were not readily available.
3.4.
Antibiotics
Antibiotic residues from varying classes have been reported in surface waters around the globe. In this study several antibiotic compounds were detected with frequencies greater than 10%: doxycycline (77%), sulfadimethoxine (43%), norfloxacin (28%), tetracycline (22%), ciprofloxacin (15%) and sulfamethazine (12%). Doxycycline had a maximum concentration of 74 mg L1. Neither Hirsch et al. (1999) in Europe or Focazio et al. (2008) in the United States detected any doxycycline in aquatic samples in sewage treatment plant discharge or surface waters. Tetracycline antibiotics have been shown to dissipate rapidly in the aqueous environment, and the relatively high occurrence and concentration found here likely indicates high usage in Costa Rica compared to previously reported study areas. The broad-spectrum antibiotic ciprofloxacin was detected with a maximum concentration of 740 ng L1 Martins et al. (2008) provides a detailed table of comparison values for ciprofloxacin, with concentrations in surface waters as high as 6.3 mg L1, much higher then observed during our sampling campaign. Sulfamethoxazole was only detected above LOQ in 4% of samples, unlike similar studies in Taiwan (96%, maximum concentration of 5.8 mg L1, Lin et al., 2008) and in Vietnamese and Japanese waters (Managaki et al., 2007). The input and persistence of antibiotics and antimicrobials, such as doxycycline, into the environment is of major concern due to the possible development of bacterial resistance to these compounds (e.g. Khachatourians, 1998). Further
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 0 9 e6 7 1 7
Table 3 e Summary statistics for PPCP compounds detected in Costa Rican surface water samples (n [ 86). Compound
% Frequency detected
Maximum [ng L-1]
Median [ng L-1]
27 29 10 8 15 3 6 8 77 26 19 20 27 5 28 3 2 3 20 41 43 12 3 5 22 34 9
13,216 1,121,446 82 63 740 63 8 266 73,722 17,036 36,788 2323 9808 11 1744 335 7571 428 592 274 20 1626 56 39 93 263 122
<15 <24 <1 <6 <31 <5 <3 <14 74 <41 <5 <7 <7 <1 <38 <22 <70 <1 <8 <11 <1 <4 <11 <5 <44 <11 <7
Acetominophen Caffeine Carbamazepine Cimetidine Ciprofloxacin Clarithromycin Clindamycin Diclofenac Doxycycline Gemfibrozil Ibuprofen Indomethacin Ketoprofen Lincomycin Norfloxacin Ofloxacin Oxacillin Oxytetracycline Paraxanthine Salicylic acid Sulfadimethoxine Sulfamethazine Sulfamethoxazole Sulfathiazole Tetracycline Triclosan Trimethoprim
Compounds not detected above LOQ: Clofibric acid, Tylosin, Ranitidine, Diltiazem, Roxithromycin and Penicillin G.
study is needed to identify if input is larger or if any conditions prevalent in Costa Rica and tropical systems slow the environmental persistence of antibiotics.
3.5.
Geographic trends
The site with the highest total concentration of all compounds was CRP-46, on the Rio Java. However, this sample was overwhelmed by the high amount of caffeine and has only four other compounds detected above LOQ, norfloxacin paraxanthine (a metabolite of caffeine), salicylic acid and triclosan (an antimicrobial ingredient in many personal care products). Two sites had the highest compound frequencies, both 59%: CRP-43, a ditch located downstream from a regional hospital in Golfito (on Golfo Dulce) collecting runoff from an effluent pipe; and CRP-49, sampled directly at the effluent pipe of a wastewater treatment oxidation pond in the city of Liberia. Compounds with high concentrations in both of these wastewater dominated samples include caffeine, gemfibrozil, doxycycline, paraxanthine, oxacillin, diclofenac, triclosan, and ibuprofen. Four compounds (acetominophen, oxacillin, sulfamethazine and triclosan) were not present in the oxidation pond effluent but were present in the hospital effluent dominated ditch. In general the concentrations in the oxidation pond effluent were 1e2 orders of magnitude higher then in the hospital effluent and the rest of the surface water samples with detections. The
6715
range of compounds and concentrations for the oxidation pond effluent are similar with those typically detected in other WWTP effluents (Spongberg and Witter, 2008; Stumpf et al., 1999). CRP-59, downstream from the tourist destination city of Quepos, sampled at low tide, also had high frequency (56%) and similar concentrations to the Liberia effluent sample. The sample was collected at a large national park (Parque´ National Manuel Antonio), indicating the possibility of wastewaters from the surrounding hotels and area contaminating the park environment. CRP 61, taken on the Rio Tarcoles, also had a high frequency of compound detection (44%). The high contamination found at this site poses a risk to the present ecosystem, known as crocodile habitat, because of possible bioaccumulation of some chemicals in the food chain This was also one of the only sites (along with CRP-22 and 23) to have detection for oxytetracycline. Site CRP 89, located in the Moline ravine, was located downstream from the city of Cartago and represented the main drainage for the city. This sample was full of suspended material and had a bad odor at the time of collection. Analysis identified 14 of the 34 compounds above LOQ in this sample. The creek passes by new developments and the local hospital. All samples (CRP 8791) from the Cartago area had detection frequencies >20% and high concentrations for doxycycline, gemfibrozil and ketoprofen. The Golfito area was found to be one of the most polluted sites in Costa Rica with respect to PCBs and hydrocarbons (Spongberg, 2004; Spongberg and Davis, 1999). However, most of the sites around Golfito were among the cleanest with respect to PPCPs (w10% frequency), with the exception of the freshwater ditch leaving the hospital mentioned previously (CRP-43). Sites receiving drainage from the city, local cemetery and local hotel were comparable to samples from the middle of Golfo Dulce (CRP-26) used as a clean reference site. Other locations sampled within Golfo Dulce (CRP 25s-33) had fairly low frequency (3e12%), although doxycycline, sulfadimethoxine and triclosan were present in almost all samples. Other sites with clean samples (<6% frequency and low concentrations) include the waters from the irrigation canal at Can˜as (CRP-52 and 53), the mouth of the Barranca River in Puntarenas at the onset of rising tide (CRP-15) and the low tide samples taken along the downstream parts of the Rio Coto-Colorado (CRP-32, 33). The Terraba-Sierpe wetland along the Sierpe River (CRP 71) also had low frequency of detection, although the concentration of ketoprofen was very high (10 ug L1), similar to another site sampled in the wetland.
3.6.
Comparison of fresh and salt water locations
This study sampled both fresh and saltwater locations. In general, the freshwater samples had higher concentrations and a greater number of detected compounds than the saltwater samples. This reflects both the dilution process into larger saltwater areas, and the selection of many fresh water sample sites based on proximity to potential pollution sources (urban runoff, point sources etc). The average concentration of the 34 PPCPs analyzed in the saltwater samples was 727 ng L1. However, the average for the Pacific samples was
6716
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 0 9 e6 7 1 7
only 75 ng L1. Several of the Pacific sites were chosen based on their remote locations and have been used in other studies as reference sites. However, even sites chosen close to shore near potential sources (e.g. Golfito samples) still had low values. The Caribbean sites had an average concentration of 4.5 mg L1. All of the Caribbean samples were taken near Limo´n near sites suspected of having a point sources, including near the local hospital (CRP-74). The data seem to reflect the presence of these point sources.
4.
Conclusions
Surface water samples and one wastewater effluent sample were collected from saltwater and freshwater environments throughout Costa Rica for the analysis of 34 PPCP compounds in early 2009. Sampling locations were chosen based on previous contaminant studies and proximity to possible sources of PPCP compounds. Although a large number of sites had low PPCP occurrence and concentration, data indicate the potential for abundant contaminant levels in areas were urban and agricultural wastewaters are treated inadequately. Compound concentrations at sites near hospitals and other heavily used waters were higher than those reported in similar studies in other countries, indicating a need to treat these wastewaters. The over the counter painkiller acetominophen, along with tetracycline antibiotics, gemfibrozil and ketoprofen were commonly associated with areas receiving both urban and rural runoff. The high occurrence and concentrations found for doxycycline show the need to further understand both usage and environmental fate of antibiotic compounds in tropical environments. This study has shown that Costa Rica faces the same wastewater treatment and management challenges posed to the rest of the world when dealing with these new emerging contaminants of concern.
Acknowledgments This work was partially supported by the United States Department of Agriculture (USDA) Cooperative State Research, Education and Extension Service (CSREES) program under grant numbers 2008-38898-19239 and 2003-3889402032. The authors would like to thank Davis our driver and Eleazar Ruiz Campos for logistical support, el Centro de Investigacio´n en Ciencias del Mar y Limnologı´a de la Universidad de Costa Rica (CIMAR) for coordinating the project, el Centro de Investigacion en Contaminacion Ambiental (CICA) for laboratory facilities and preliminary analyses, Chenxi Wu for guidance on analytical analysis, and an anonymous reviewer for helpful comments.
Appendix. Supplementary data Supplementary data related to this article can be found online at doi:10.1016/j.watres.2011.10.004.
references
Buerge, I., Poiger, T., Mu¨ller, M., Buser, H., 2003. Caffeine, and anthropogenic marker for wastewater contamination of surface waters. Environmental Science and Technology 37 (4), 691e700. Clara, M., Strenn, B., Kreuzinger, N., 2004. Carbamazepine as a possible anthropogenic marker in the aquatic environment: investigations on the behaviour of carbamazepine in wastewater treatment and during groundwater infiltration. Water Research 38, 947e954. Ferreira, A.P., 2005. Caffeine as an environmental indicator for assessing urban aquatic ecosystems. Cadernos de Sau´de Pu´blica 21, 1884e1892. Focazio, M.J., Kolpin, D.W., Barnes, K.K., Furlong, E.T., Meyer, M.T. , Zaugg, S.D., Barber, L.B., Thurman, M.E., 2008. A national reconnaissance for pharmaceuticals and other organic wastewater contaminants in the United States e II) untreated drinking water sources. Science of the Total Environment 402, 201e216. Hirsch, R., Ternes, T., Haberer, K., Kratz, K.-L., 1999. Occurrence of antibiotics in the aquatic environment. The Science of the Total Environment 225, 109e118. Khachatourians, G., 1998. Agricultural use of antibiotics and the evolution and transfer of antibiotic-resistant bacteria. Canadian Medical Association Journal 159, 1129e1136. Kim, S.D., Cho, J., Kim, I.S., Vanderford, B.J., Snyder, S.A., 2007. Occurrence and removal of pharmaceuticals and endocrine disruptors in South Korean surface, drinking and wastewater. Water Research 41, 1013e1021. 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. Environmental Science and Technology 36, 1202e1211. Lin, A.Y., Yu, T.H., Lin, C.F., 2008. Pharmaceutical contamination in residential, industrial, and agricultural waste streams: risk to aqueous environments in Taiwan. Chemosphere 74, 131e141. Locatelli, M., Sodre´, F., Jardim, W., 2010. Determination of antibiotics in Brazilian surface, waters using liquid chromatographyeelectrospray tandem mass spectrometry. Archives of Environmental Contamination and Toxicology, 1e9. Loos, R., Gawlik, B.M., Locoro, G., Rimaviciute, E., Contini, S., Bidoglio, G., 2009. EU-wide survey of polar organic persistent pollutants in European river waters. Environmental Pollution 157, 561e568. Managaki, S., Murata, A., Takada, H., Tuyen, B.C., Chiem, N.H., 2007. Distribution of macrolides, sulfonamides, and trimethoprim in tropical water: ubiquitous occurrence of veterinary antibiotics in the mekong delta. Environmental Science and Technology 41, 8004e8010. Martı`nez-Carballo, E., Gonz$lez-Barreiro, C., Scharf, S., Gans, O., 2007. Environmental monitoring study of selected veterinary antibiotics in animal manure and soils in Austria. Environmental Pollution 148, 570e579. Martins, A.F., Vasconcelos, T.G., Henriques, D.M., Frank, C., da, S., Ko¨nig, A., Ku¨mmerer, L., 2008. Concentration of ciprofloxacin in Brazilian hospital effluent and preliminary risk assessment: a case study. CLEAN 36, 264e269. Ramirez, A.J., Brain, R.A., Usenko, S., Mottaleb, M.A., O’Donnell, J. G., Stahl, L.L., Wathen, J.B., Snyder, B.D., Pitt, J.L., PerezHurtado, P., Dobbins, L.L., Brooks, B.W., Chambliss, C.K., 2009. Occurrence of pharmaceuticals and personal care products in fish: results of a national pilot study in the United States. Environmental Toxicology and Chemistry 28, 2587e2597.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 0 9 e6 7 1 7
Seiler, R.L., Zaugg, S.D., Thomas, J.M., Howcroft, D.L., 1999. Caffeine and pharmaceuticals as indicators of wastewater contamination in wells. Groundwater 37, 405e410. Siemens, J., Huschek, G., Siebe, C., Kaupenjohann, M., 2008. Concentrations and mobility of human pharmaceuticals in the world’s largest wastewater irrigation system, Mexico CityMezquital Valley. Water Research 42, 2124e2134. Spongberg, A.L., 2004. An initial investigation of PCB contamination in surface sediments in the coastal waters of Costa Rica. Journal of Tropical Biology 52 (Suppl. 2), 1e9. Spongberg, A.L., Davis, P., 1999. Preliminary investigation of pesticide contamination in Golfo Dulce, Costa Rica. Journal of Tropical Biology 46 (Suppl. 5), 111e124. Spongberg, A.L., Witter, J.D., 2008. Pharmaceutical compounds in wastewater treatment plant influent, effluent and sludge from northwest Ohio. The Science of the Total Environment 397, 148e157. Stumpf, M., Ternes, T.A., Wilken, R.-D., Silvana Vianna, R., Baumann, W., 1999. Polar drug residues in sewage and natural
6717
waters in the state of Rio de Janeiro, Brazil. The Science of the Total Environment 225, 135e141. Tixier, C., Singer, H.P., Oellers, S., Mu¨ller, S.R., 2003. Occurrence and fate of carbamazepine, clofibric acid, diclofenac, ibuprofen, ketoprofen and naproxen in surface waters. Environmental Science and Technology 37, 1061e1068. Walters, E., McClellan, K., Halden, R.U., 2010. Occurrence and loss over three years of 72 pharmaceuticals and personal care products from biosolids-soil mixtures in outdoor mesocosms. Water Research 44, 6011e6020. Wu, C., Spongberg, A.L., Witter, J.D., 2008. Use of solid phase extraction and liquid chromatograph-tandem mass spectrometry for simultaneous determination of various pharmaceuticals in surface water. The International Journal of Environmental Analytical Chemistry 88, 1033e1048. Wu, C., Witter, J.D., Spongberg, A.L., Czajkowski, K.P., 2009. Occurrence of selected pharmaceuticals in an agricultural landscape, Western Lake Erie Basin. Water Research 43, 3407e3416.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 1 8 e6 7 2 4
Available online at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Breakage and re-growth of flocs: Effect of additional doses of coagulant species Wenzheng Yu a,b, John Gregory a,*, Luiza C. Campos a a
Department of Civil, Environmental and Geomatic Engineering, University College London, Gower Street, London WC1E 6BT, UK State Key Laboratory of Urban Water Resource and Environment (SKLUWRE), School of Municipal & Environmental Engineering, Harbin Institute of Technology, No 73 Huanghe Road, Nangang District, Harbin 150090, China b
article info
abstract
Article history:
Several polyaluminum chloride (PACl) coagulants were prepared, with different OH/Al
Received 30 March 2011
ratios (B values), and characterized by Ferron assay. These were used in studies of floc
Received in revised form
formation, breakage and re-growth with kaolin suspensions under controlled shear
11 October 2011
conditions, using a continuous optical monitoring method. Particular attention was paid to
Accepted 12 October 2011
the effect of small additional coagulant dosages, added during the floc breakage period, on
Available online 20 October 2011
the re-growth of broken flocs. The results showed that the re-growth ability was greatly dependent on the nature of the PACl species added as second coagulant. The re-growth
Keywords:
ability of broken flocs was greatest when the second coagulant was PACl0 (i.e. AlCl3, with
Floc breakage
B ¼ 0) and least with PACl25 (B ¼ 2.5). In the latter case there was no effect on floc re-growth,
Floc re-growth
irrespective of the initial coagulant used. PACls with intermediate B values gave some
Kaolin
improvement in floc re-growth, but less than that with PACl0. Additional dosage of PACl0
Alum
gave re-grown flocs about the same size or even larger than those before breakage. The re-
PACl
growth of broken flocs is significantly correlated with the species Ala (monomeric) and Alb (polymeric), as determined by Ferron assay. The amorphous hydroxide precipitate formed from PACl0, (mainly Ala) can greatly improve the adhesion between broken flocs and give complete re-growth. However, for PACl25, mostly composed of Alb, the nature of the precipitate is different and there is no effect on floc re-growth. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
For hydrolyzing coagulants based on Al and Fe, charge neutralization and sweep coagulation are recognized as the important mechanisms (Duan and Gregory, 2003). However, the nature of the floc growth process and the mode of attachment between floc particles are still not fully understood. In particular, hydroxide flocs often show irreversible breakage, so that broken flocs do not fully re-grow (Yukselen and Gregory, 2004b). The reasons for this behavior are not yet clear. The nature of the floc surface is important, but there
is comparatively little information on this subject. The role of the zeta potential of flocs, as measured by electrophoretic mobility (EM), in floc growth is also a matter of debate (Xiao et al., 2008). As mentioned above, it is well known that there may be a significant irreversibility of floc breakage. Yukselen and Gregory (2004a, b) found that only limited re-growth of kaolin flocs occurred when the coagulant was alum or PACl. Li et al. (2007) also showed that broken kaolin flocs were difficult to re-grow when alum was used. Solomentseva et al. (2007) showed that the formation, breakage and re-growth of flocs
* Corresponding author. Tel.: þ44 20 7689 7818; fax: þ44 20 7380 0986. E-mail address:
[email protected] (J. Gregory). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.10.016
6719
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 1 8 e6 7 2 4
could be repeated several times, but after each breakage cycle the re-grown flocs were smaller. The irreversibility of floc breakage at high shear has been found not only for kaolin suspensions, but also NOM (Jarvis et al., 2005; Wang et al., 2009). The flocs formed in natural water with alum and PACl showed the same phenomenon (McCurdy et al., 2004). Physical properties such as floc size, compaction, and strength have been investigated in recent years (Jarvis et al., 2005, 2006; Wang et al., 2009), as these properties may influence the regrowth ability of broken flocs. It may be that these physical properties of flocs do influence the re-growth ability, but, according to our work, the surface properties of the flocs may change as a result of the breakage process, which restricts the re-growth of broken flocs. Ehrl et al. (2008) considered that, after the initial phase of self-accelerating aggregate growth, the kinetics gradually slows down as breakage sets in and the flocs eventually reach a steady-state size. This steady state predominantly depends on the size and concentration as well as surface and bulk properties of the primary particles, composition of the liquid phase, type of coagulant and average value and distribution of the shear rate in the vessel. According to Serra et al. (2008), at higher shear rates (G > 30 s1), the dominant effect of breakup of latex flocs was shown through reduced maximum floc sizes with increasing shear rates. Xiao et al. (2008) considered that enmeshment of particles by voluminous flocs was independent of the EM of particles, but solution pH and Al (III) concentration were found to strongly affect coagulation efficiency. The relative importance of the properties of primary particles or the precipitate of coagulant in determining coagulation efficiency should be further investigated. Previous studies suggested that when hydrolyzing coagulants are used, there is distinct irreversibility of the floc breakup process so that floc re-growth is limited. However, there is still no adequate model to explain these findings (Duan and Gregory, 2003). This paper focuses on surface properties of flocs, and investigates the effect of additional dosage of different species of PACl on floc re-growth ability. The results should lead to a better understanding of factors influencing floc re-growth after breakage.
2.
Materials and methods
2.1.
Preparation of PACl
The various PACl samples were prepared using an alkali titration method at room temperature (Wang et al., 2004). A calculated amount of 0.2 mol/L AlCl3 solution was transferred into a 300 mL jar. Under rapid stirring, this solution was titrated slowly (0.1 mL/min) with 0.5 mol/L NaOH using a syringe pump (74900 series, ColeeParmer Instrument company, Herts, UK). The amount of NaOH added varied with the target OH/Al ratio (B value), i.e. 0, 1.0, 1.5, 2.0, and 2.5, denoted, respectively as PACl0, PACl10, PACl15, PACl20, and PACl25. The final concentration of aluminum was 0.1 mol/L in all cases. The reagents used in this study were all of analytical grade. The samples, after aging for one week, were analyzed using the Ferron method as described below. The results of speciation and pH measurement are shown in Table 1 (Section 3.1).
2.2.
Ferron method
The Ferron reagent used by some previous researchers exhibited some unstable features. A continuous change in the absorption spectrum of the Ferron reagent with aging was observed, and it was found to be relatively stable only after 10 days (Duffy and Vanloon, 1994). Therefore, it has been suggested that the mixed Ferron solution may be used after 10 days and the experiments should preferably be completed within one day. The main procedure used was as follows (Wang et al., 2004): Reagent A (0.2% Ferron): under rapid stirring, 2 g Ferron (Sigma Chemical Co., MO, USA) was dissolved in 1000 mL deionized water, then, filtered and stored in a 1 L bottle. Reagent B (20% w/v NaAc): 200 g NaAc was dissolved in 1 L of deionized water. Reagent C (1:9 v/v HCl): 100 mL HCl (37%) was mixed with 900 mL of deionized water. Prior to the speciation experiments, the mixed Ferron solution was obtained by mixing reagents A, B, and C in the proportions 2.5:2:1. For each test 2.2 mL of the mixed reagent was transferred into 25-mL graduated glass tube and 10 mL of deionized water was added. Then 1 mL of the PACl sample was quickly added to give a final volume of 13.2 mL. After mixing for 15 s, a portion of the reacting sample was transferred to a 1-cm glass cuvette. The timed absorbance measurements (at 366 nm), using a Camspec M350 UVeVisible spectrophotometer, were carried out from 30 s and recorded for a further 2 h. The absorbance after the first minute was assumed to be due to Ala (mononuclear Al species) and the absorbance developed from 1 min to 2 h was due to Alb (reactive, polynuclear Al). The remaining fraction, Alc (inert or colloidal Al species) was calculated as AlT minus (Ala þ Alb), where AlT is the known total Al concentration.
2.3.
Suspension and jar test
Kaolin clay (Imerys, St Austell, Cornwall, UK) was used as the model suspension. 200 g of kaolin was dispersed in 500 mL of deionized (DI) water in a high-speed blender. To obtain full dispersion it was necessary to raise the pH of the suspension to about 7.5, which was achieved by adding 5 mL of 0.1 M NaOH. After blending at 4000 rpm for 10 min, the clay suspension was diluted to 1 L with DI water and allowed to stand overnight in a measuring cylinder. The top 800 mL was decanted and its solids content was determined gravimetrically and found to be 133 g/L. For the flocculation tests, the stock suspension was diluted to give a final clay concentration of 50 mg/L.
Table 1 e Species distributions and pH values of PACl samples. Sample PACl0 PACl10 PACl15 PACl20 PACl25
B
Ala (%)
Alb (%)
Alc (%)
pH
0.0 1.0 1.5 2.0 2.5
94.5 52.5 45.8 10.5 3.4
5.5 46.6 53.4 82.8 92.0
0 0.8 0.8 6.7 4.6
2.95 3.68 3.80 4.00 5.19
6720
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 1 8 e6 7 2 4
2.4.
Electrophoretic mobility
Electrophoretic mobility (EM) was measured on the kaolin suspension before coagulant addition and for samples after coagulant dosing and 1 min of rapid mixing, by a Particle Electrophoresis Apparatus Mk 2 (Rank Brothers Ltd, Cambridge, UK). In addition, the EM of flocs after 1 min of breakage at high stirring speed was measured. The delay time of the measurement was about 40 s. The average EM value for a sample was determined from 20 measurements.
3.
Results and discussion
3.1.
The characteristics of PACl samples
The Al species distribution in the PACl samples was analyzed by the Ferron method as described in Section 2.2. The kinetic differences of the reactions between aluminum and Ferron reagent are shown in Fig. 1. From these data, the proportions of Ala, Alb and Alc species can be calculated, as outlined earlier. Table 1 shows the speciation distributions and pH
values of PACl samples. It is clear that for PACl0 (i.e. AlCl3) nearly all of Al (around 95%) exists in monomeric form (Ala). For the other samples, the proportion of Ala decreases with increasing B values and there is a corresponding increase in the proportion of the polymeric species (Alb). It is known (Wu et al., 2007) that most of Al in the Alb fraction is in the form of the tridecameric polycation Al13O4(OH)7þ 24 (‘Al13’). PACl10 and PACl15 contain around half Alb, whereas for PACl20 and PACl25, nearly all (>80%) of the Al exists as Alb. For this reason, PACl0, PACl15 and PACl25 were chosen for the coagulation experiments.
3.2.
Effect of different species and dosage of PACls on EM
The effect of different species of PACl, at different concentrations, on the EM of flocs is shown in Fig. 2. Prior to dosage, the EM of kaolin particles was negative (around 3.6 mm s1 V1 cm1). The addition of low dosages of PACl0 caused a rapid increase of EM and the charge became slightly positive at dosages of around 0.2 mM Al or higher. For the other PACl samples, very low dosages caused a steep rise in EM and charge reversal occurred at about 0.05 mM Al or less. At higher dosages of these coagulants the EM reached a plateau value around þ1 mm s1 V1 cm1. From Fig. 2, it can be seen that the EM values for different PACls at the same Al dosage were in the order: PACl25 > PACl15 > PACl0. The dosages of PACl25, PACl15, PACl0 required to give EM ¼ 0 (the isoelectric point) were 0.02, 0.04 and 0.1 mM, respectively, showing that increasing the B (OH/Al) value increased the charge neutralizing ability of the PACl samples.
3.3.
Re-growth ability of broken flocs
Fig. 3 shows the results of experiments in which kaolin suspensions were coagulated at 50 rpm with either PACl0 or PACl25, both at a dosage of 0.1 mM Al, followed by floc breakage at an increased stirring speed, as described in 2.3. In some cases additional coagulant (0.03 mM Al) was added during the breakage period. The initial coagulation with PACl0 (Fig. 3a) gave higher FI values (and hence larger flocs) than with PACl25 (Fig. 3b), very likely because the latter coagulant at 0.25
0.20
0.15
UV 366
The test solutions were prepared using DI water with addition of 5 mM NaHCO3. For the flocculation tests, the stock suspension was diluted to give a clay concentration of 50 mg/ L. The pH of the solution was adjusted to 7 by prior addition of a predetermined amount of 0.1 M HCl. All reagents used were of analytical grade. The experiments were conducted at a temperature of 251 C. A known amount of PACl was added into the test suspension (800 mL) and at the same time the stirring of a semiautomatic jar test device (Flocculator 2000, Kemira, Sweden) was started. The stirring speed was set at 200 rpm (G z 184 s1) for 1 min for rapid mixing and 50 rpm (G z 23 s1) for 10 min to allow flocs to grow. The speed was then increased to 200 rpm for 1 min to break the flocs and then back to 50 rpm for 10 min for flocs to re-grow. In some cases, additional doses of different PACl species were added half way through the breakage period (i.e. at 30 s after increasing the stirring speed). The values of effective shear rate, G, quoted are derived from the work of Chavez and Jimenez (2000), who also used a Flocculator 2000. Experiments on the kinetics of formation, breakage and subsequent re-growth of flocs were performed using the ‘turbidity fluctuation’ technique, as used in the Photometric Dispersion Analyzer (PDA-2000, Rank Brothers, UK). The experimental procedure was similar to that of Yukselen and Gregory (2004a). In this method, the average intensity of light transmitted through the flowing suspension (dc value) and the root mean square (rms) value of the fluctuating component are measured. The ratio (rms/dc) is often termed the Flocculation Index (FI) and it provides a sensitive measure of particle aggregation (Gregory and Nelson, 1986). It significantly increases as aggregation occurs, and decreases when aggregates are broken. In this work, after the FI value reached an initial steady value, coagulant was added into the suspension and the FI value was recorded by a PC data acquisition system (Pico ADC-11, Pico Technology, UK) at 1 s intervals.
B C D E F
0.10
0.05
0.00
0
1000
2000
3000
4000 5000 Time (s)
6000
7000
8000
Fig. 1 e Variation of UV366 with reaction time by different species of PACls.
6721
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 1 8 e6 7 2 4
Fig. 2 e Electrophoretic mobility vs. dosage for different species of PACl at pH 7.
a dosage of 0.1 mM gave charge reversal and a positive zeta potential (Fig. 2). With no additional coagulant, the re-growth of flocs after breakage was limited in both cases, with the FI values returning to only about 60% of their values before
breakage. This type of behavior has been observed several times previously (e.g. Yukselen and Gregory, 2004b). The effect of the additional coagulant was greatly dependent on its composition. For PACl0, the additional dosage gave a marked improvement in floc re-growth. When PACl0 was the initial coagulant (Fig. 3a), the further dosage gave re-growth to about the same FI value as that before breakage. However, when PACl25 was the second coagulant, no improvement in floc regrowth was found. For PACl15, there was significant improvement in floc re-growth. Both PACl0 and PACl15 contain significant amounts of monomeric Ala, whereas PACl25 contains very little (Table 1). It is thus very likely that Ala plays a large part in the floc re-growth process. This is confirmed by the results in Fig. 3b, where the initial coagulant was PACl25. When the additional coagulant was also PACl25, no improvement in floc re-growth was found. However, with an addition of PACl0 during breakage, a large improvement in re-growth occurred, giving an FI value for the re-grown flocs significantly higher than that for flocs before breakage. Again, PACl15 gave intermediate results. It is possible that the observed effects might be explained by the changes in the surface properties of flocs as a result of breakage. Electrophoretic mobility results for flocs immediately after breakage, under the same conditions as for Fig. 3, are shown in Table 2. The EM values after floc breakage for 0.1 mM PACl0 and PACl25 alone are not significantly different from the values just after coagulant dosing (Fig. 2). Additional dosages of PACl modify the EM of broken flocs, making the values more positive when the initial coagulant is PACl0. For PACl25 as the first coagulant, additional coagulant gives less positive EM values. From the data in Table 2 it is difficult to find convincing evidence that charge effects can explain the differences in floc re-growth shown in Fig. 3. For instance, PACl0 alone shows only limited floc re-growth, even though the EM values before and after breakage are about the same. An additional low dosage of the same coagulant during breakage gives much improved floc re-growth (Fig. 3a) and yet the EM value of the broken flocs is only slightly less negative (changed from 0.21 to 0.17). When the second coagulant is PACl25, the EM value of the broken flocs is þ0.51, but the regrowth is about the same as for PACl0 alone, where the EM value is 0.21. For PACl25 alone, the EM value is þ1.25 and this probably accounts for the smaller initial FI value (Fig. 3b) than with PACl0 alone (Fig. 3a). Additional coagulant dosing gives reduced positive EM values and these appear to correlate with the increased floc re-growth in Fig. 3b. Thus, with PACl25 as the second coagulant the EM is only slightly reduced (from þ1.25 to þ1.19), whereas the other two coagulants give less positive EM values and better floc re-growth. However,
Table 2 e EM (mm sL1 VL1 cm) of flocs with and without additional coagulant dosage. Conditions as for Fig. 3. Coagulant
Fig. 3 e The effect of different additional coagulant species on floc growth, breakage and re-growth at pH 7 for 0.1 mM of (a) PACl0 and (b) PACl25 added initially.
0.1 mM PACl0 0.1 mM PACl25
Additional coagulant None
0.03 mM PACl0
0.03 mM PACl15
0.03 mM PACl25
0.21 1.25
0.13 0.55
0.25 0.71
0.51 1.19
6722
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 1 8 e6 7 2 4
comparing the cases PACl0 þ PACl25 and PACl25 þ PACl0 in Table 2, the final EM values are almost the same (þ0.51 and þ0.55) but the re-growth of flocs is very different. It appears that from the above results that EM values are not directly relevant to floc re-growth after breakage. In order to investigate this point further, tests were carried out with lower doses of PACl25, to give EM values close to zero.
3.4. Re-growth of PACl25 flocs close to charge neutralization The dosage of PACl25 needed to give charge neutralization is about 0.02 mM Al (Fig. 2). Therefore, kaolin suspensions were coagulated with 0.02 mM PACl25 or lower, to investigate floc formation and re-growth under conditions where the EM was close to zero. Results are shown in Fig. 4, from three trials with the following coagulants: 0.02 mM PACl25 alone 0.01 mM PACl25 initially and 0.01 mM PACl25 added during floc breakage 0.02 mM PACl25 initially and 0.03 mM PACl0 added during floc breakage The first two of these have the same final coagulant dosage after floc breakage, and show very little difference. The initial rate of rise of FI is slightly higher for the higher initial dosage, but the plateau FI values are almost the same. Note that the plateau FI value for this coagulant is significantly higher than that in Fig. 3, where a higher dosage was used and the flocs had a fairly high positive EM value. Breakage and re-growth are also closely similar for the first two cases. An additional dosage of PACl25 cannot improve the re-growth ability, even though the EM of flocs is very close to zero. However, for the third case, where an additional dosage of PACl0 is introduced during breakage, there is a much improved re-growth of flocs, giving a final FI value slightly higher than that before breakage. The most important point about the results in Fig. 4 is that the EM values for flocs in all three cases are very close to zero. As before, it is the addition of PACl0, containing mostly
Fig. 4 e The effect of PACl25 and PACl0 on the reversibility of floc breakage at pH 7, with PACl25 as the first coagulant.
Ala, that gives good floc re-growth and these results confirm that EM is not a factor.
3.5.
Mechanisms
The fact that re-growth of broken flocs may be restricted, even when the EM is close to zero, is difficult to understand in terms of basic colloid stability theory. The floc fragments should be able to come into contact with little or no repulsion. In sheared suspensions the collision efficiency depends on particle interactions, particle size and the effective shear rate (Gregory, 1982). Since the FI value (and hence size) for the broken flocs was the same in all cases (Fig. 4) and the shear conditions were identical, the explanation for the different regrowth potential must involve different interactions between floc fragments. We need to consider possible changes in floc surface properties not directly related to surface charge. All of the reported experiments were conducted at pH 7, where Al solubility is close to its minimum value. Although PACls have slightly higher solubility than simple Al salts (Pernitsky and Edzwald, 2003), at the coagulant concentrations used here (at least 20 mM Al) the Al solubility should always be exceeded and an amorphous hydroxide precipitate (AHP) would be expected. Hence, it is likely that all of the coagulation reported here would be under ‘sweep floc’ conditions, where precipitates attach to and enmesh the kaolin particles. Under these conditions EM values seem to have little influence on coagulation performance (Xiao et al., 2008). The irreversible nature of floc breakage found with hydrolyzing coagulants (Yukselen and Gregory, 2004b) is very likely associated with the nature of hydroxide precipitates, since flocs formed with other types of coagulant can show almost complete re-growth after breakage. It is likely that, after breakage of hydroxide flocs, certain areas of the surface of the floc fragments are ‘inactive’, so that they do not easily form attachments with other broken flocs. It may be that the ‘inactive’ areas are those that formed contacts in flocs before breakage. Only those areas that had not formed attachments previously would remain ‘active’ and able to form attachments during floc re-growth. This concept is shown schematically in Fig. 5. In this way, the collision efficiency of broken flocs would be reduced, since fewer collisions would result in attachment. With further floc growth, breakage and re-growth cycles, the proportion of ‘active’ surface would become progressively less and the collision efficiency would be further reduced. Each cycle would give smaller re-grown flocs than the previous, as has been observed experimentally (Solomentseva et al., 2007; Yukselen and Gregory, 2002). It is still not clear why floc breakage leads to ‘inactive’ surface sites, but it seems, from the present study and our previous work (Yu et al., 2010a, b) that small doses of additional coagulant can effectively ‘re-activate’ these sites so that floc breakage can be fully reversed. However, additional coagulant added at the beginning of a long breakage period (several minutes) did not improve the re-growth of Al-humic flocs (Yu et al., 2010a). This suggests that freshly formed hydroxide precipitate is able to re-coat broken flocs, so that inactive surface sites are re-activated, but this effect is short-lived and a prolonged period of high shear causes further inactivation.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 1 8 e6 7 2 4
Fig. 5 e Schematic picture showing floc breakage and the formation of ‘inactive’ surface sites. The areas of contact in the floc before breakage are shaded black and it is possible that these regions are not able to form attachments during floc re-growth and are hence ‘inactive’. The ‘active’ areas are those that were not in direct contact in the original floc and these would still be able to form attachments during floc re-growth.
An important finding of the present work is that different PACl species give very different floc re-growth when added as the second coagulant. The results in Fig. 3 show that the effect of additional coagulant on floc re-growth decreases in the order PACl0 > PACl15 > PACl25, with the last giving no improvement. It has been shown above (3.4) that surface charge effects are unlikely to provide an explanation of these effects. The major difference between the coagulants is the relative proportion of Ala and Alb species. In PACl25, most of the Al (92%) exists as Alb, which is thought to be mainly Al13. It has been found (Van Benschoten and Edzwald, 1990) that precipitates formed from Al13 solutions are very different from the amorphous hydroxide precipitates typically formed from solutions of aluminum salts. In particular, the Al13 structure is largely retained in the precipitate. From the results in Figs. 3 and 4, it appears that fresh amorphous hydroxide precipitate formed from PACl0 is highly effective in promoting the regrowth of broken flocs, whereas the Al13 precipitate is ineffective. More detailed studies are needed to elucidate the difference in behavior between these coagulants.
4.
Conclusions
1. The PACls used can be divided into three species: PACl0 (nearly all monomeric Ala), PACl10 and PACl15 (composed of roughly 50-50 Ala and polymeric Alb), and PACl20 and PACl25 (nearly all Alb). 2. The re-growth ability of broken flocs could be improved by a small additional coagulant dosage, but this effect was greatly dependent on the PACl type. The effect on floc regrowth was in the following order: PACl0>PACl15 > PACl25, which was significantly correlated with the species distribution of Ala and Alb. 3. The electrophoretic mobility of flocs was not a significant factor in their re-growth ability. 4. As a result of floc breakage at high shear, some areas of the floc surface may become ‘inactive’ and unable to form
6723
attachments with other flocs, thus reducing the collision efficiency. Amorphous hydroxide precipitate formed by additional coagulant dosage can coat the floc surface and increase collision efficiency. Precipitates formed from Alb species are not effective. 5. There are potentially significant practical implications of this work. Where floc breakage is a problem, as a result of exposure to high shear, then a small second addition of coagulant may facilitate floc re-growth when more quiescent conditions are restored. It appears that pre-hydrolyzed coagulants of high passivity would be less effective in promoting floc re-growth.
references
Chavez, A., Jimenez, B., 2000. Particle size distribution (psd) obtained in effluents from an advanced primary treatment process using different coagulants. In: Hahn, H.H., Hoffmann, E., Odegaard, H. (Eds.), Chemical Water and Wastewater Treatment VI. Springer, Berlin, pp. 257e268. Duan, J.M., Gregory, J., 2003. Coagulation by hydrolysing metal salts. Advances in Colloid and Interface Science 100, 475e502. Duffy, S.J., Vanloon, G.W., 1994. Characterization of amorphous aluminum hydroxide by the ferron method. Environmental Science & Technology 28, 1950e1956. Ehrl, L., Soos, M., Morbidelli, M., 2008. Dependence of aggregate strength, structure, and light scattering properties on primary particle size under turbulent conditions in stirred tank. Langmuir 24, 3070e3081. Gregory, J., 1982. Particle interactions in flowing suspensions. Advances in Colloid and Interface Science 17, 149e160. Gregory, J., Nelson, D.W., 1986. Monitoring of aggregates in flowing suspensions. Colloids and Surfaces 18, 175e188. Jarvis, P., Jefferson, B., Parsons, S.A., 2005. Breakage, regrowth, and fractal mature of natural organic matter flocs. Environmental Science & Technology 39, 2307e2314. Jarvis, P., Jefferson, B., Parsons, S.A., 2006. Floc structural characteristics using conventional coagulation for a high doc, low alkalinity surface water source. Water Research 40, 2727e2737. Li, T., Zhu, Z., Wang, D.S., Yao, C.H., Tang, H.X., 2007. The strength and fractal dimension characteristics of alum-kaolin flocs. International Journal of Mineral Processing- 82, 23e29. McCurdy, K., Carlson, K., Gregory, D., 2004. Floc morphology and cyclic shearing recovery: Comparison of alum and polyaluminum chloride coagulants. Water Research 38, 486e494. Pernitsky, D.J., Edzwald, J.K., 2003. Solubility of polyaluminium coagulants. Journal of Water Supply: Research and Technology-AQUA 52, 395e406. Serra, T., Colomer, J., Logan, B.E., 2008. Efficiency of different shear devices on flocculation. Water Research 42, 1113e1121. Solomentseva, I., Barany, S., Gregory, J., 2007. The effect of mixing on stability and break-up of aggregates formed from aluminum sulfate hydrolysis products. Colloids and Surfaces A 298, 34e41. Van Benschoten, J.E., Edzwald, J.K., 1990. Chemical aspects of coagulation using aluminum salts .1. Hydrolytic reactions of alum and polyaluminum chloride. Water Research 24, 1519e1526. Wang, D.S., Sun, W., Xu, Y., Tang, H.X., Gregory, J., 2004. Speciation stability of inorganic polymer flocculant-pacl. Colloids and Surfaces A 243, 1e10. Wang, Y., Gao, B.Y., Xu, X.M., Xu, W.Y., Xu, G.Y., 2009. Characterization of floc size, strength and structure in various
6724
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 1 8 e6 7 2 4
aluminum coagulants treatment. Journal of Colloid and Interface Science 332, 354e359. Wu, X.H., Ge, X.P., Wang, D.S., Tang, H.X., 2007. Distinct coagulation mechanism and model between alum and high al-13-pacl. Colloids and Surfaces A 305, 89e96. Xiao, F., Zhang, X.R., Lee, C., 2008. Is electrophoretic mobility determination meaningful for aluminum(iii) coagulation of kaolinite suspension? Journal of Colloid and Interface Science 327, 348e353. Yu, W.Z., Gregory, J., Campos, L., 2010a. Breakage and regrowth of Al-humic flocs - effect of additional coagulant dosage. Environmental Science and Technology 44, 6371e6376.
Yu, W.Z., Gregory, J., Campos, L., 2010b. The effect of additional coagulant on the re-growth of alum-kaolin flocs. Separation and Purification Technology 74, 305e309. Yukselen, M.A., Gregory, J., 2002. Breakage and re-formation of alum flocs. Environmental Engineering Science 19, 229e236. Yukselen, M.A., Gregory, J., 2004a. The effect of rapid mixing on the break-up and re-formation of flocs. Journal of Chemical Technology & Biotechnology 79, 782e788. Yukselen, M.A., Gregory, J., 2004b. The reversibility of floc breakage. International Journal of Mineral Processing 73, 251e259.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 2 5 e6 7 3 6
Available online at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Modelling the photochemical fate of ibuprofen in surface waters Davide Vione a,b,*, Pratap Reddy Maddigapu a, Elisa De Laurentiis a, Marco Minella a, Marco Pazzi a, Valter Maurino a, Claudio Minero a, Sofia Kouras c,d, Claire Richard c,d a
Dipartimento di Chimica Analitica, Universita` di Torino, Via P. Giuria 5, 10125 Torino, Italy1 Centro Interdipartimentale NatRisk, Universita` di Torino, Via Leonardo da Vinci 44, 10095 Grugliasco (TO), Italy2 c Clermont Universite´, Universite´ Blaise Pascal, Laboratoire de Photochimie Mole´culaire et Macromole´culaire, BP 10448, F-63000 Clermont-Ferrand, France d CNRS, UMR 6505, Laboratoire de Photochimie Mole´culaire et Macromole´culaire, BP80026, F-63171 Aubie`re Cedex, France b
article info
abstract
Article history:
We show that the main photochemical processes involved in the phototransformation of
Received 29 June 2011
anionic ibuprofen (IBP) in surface waters are the reaction with OH, the direct photolysis
Received in revised form
and possibly the reaction with the triplet states of chromophoric dissolved organic matter
11 October 2011
(3CDOM*). These conclusions were derived by use of a model of surface water photo-
Accepted 12 October 2011
chemistry, which adopted measured parameters of photochemical reactivity as input data.
Available online 20 October 2011
The relevant parameters are the polychromatic UVB photolysis quantum yield (FIBP ¼ 0.33 0.05, m s), the reaction rate constant with OH (kIBP,OH ¼ (1.0 0.3), 1010 M1 s1), the 1O2 rate constant (kIBP; 1 O2 ¼ (6.0 0.6),104 M1 s1), while the reaction
Keywords:
with CO 3 can be neglected. We adopted anthraquinone-2-sulphonate (AQ2S) and ribo-
Pharmaceuticals
flavin (Ri) as CDOM proxies and the reaction rate constants with the respective triplet
PCCPs Non-steroidal
anti-inflammatory
states were kIBP,3AQ2S* ¼ (9.7 0.2),109 M1 s1 and kIBP,3Ri* ¼ 4.5,107 M1 s1. The reaction
drug
with 3CDOM* can be an important IBP sink if its rate constant is comparable to that of
Surface water photochemistry
3
Sensitised phototransformation
chemical pathways mainly lead to the transformation (oxidation and/or shortening) of the
Hydroxyl radical
propanoic lateral chain of IBP, which appears to be significantly more reactive than the
Singlet oxygen
isobutyl one. Interestingly, none of the detected intermediates was produced by substi-
AQ2S*, while it is unimportant if the rate constant is similar to the 3Ri* one. The photo-
tution on the aromatic ring. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
Ibuprofen (2-(4-(2-methylpropyl)phenyl)propanoic acid, hereafter IBP) is a non-steroidal anti-inflammatory drug that is widely used nowadays as the active principle of many “over
the counter” pharmaceutical products. Excretion by humans after partial metabolism and incorrect drug disposal are two important pathways of IBP to sewage waters. For instance, it has been the sixth most abundant pharmaceutical detected in the influent flow of wastewater treatment plants (WWTPs) in
* Corresponding author. Dipartimento di Chimica Analitica, Universita` di Torino, Via P. Giuria 5, 10125 Torino, Italy. Tel.: þ39 011 6705296; fax: þ39 011 6707615. E-mail address:
[email protected] (D. Vione). URL: http://chimica.campusnet.unito.it/do/docenti.pl/Show?_id¼dvione 1 http://www.environmentalchemistry.unito.it. 2 http://www.natrisk.org. 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.10.014
6726
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 2 5 e6 7 3 6
Italy (Castiglioni et al., 2006). Although it is susceptible to biodegradation (Tiehm et al., 2011; de Graaf et al., 2011), IBP is only partially removed by WWTPs (Oulton et al., 2010; Morasch et al., 2010) and the treatment efficiency undergoes a considerable decrease during the winter months (Castiglioni et al., 2006; Santos et al., 2009), when the IBP environmental inputs tend to increase considerably (Daneshvar et al., 2010). The combination of widespread use and incomplete WWTP elimination results into an extensive occurrence of IBP in surface waters. Indeed, this compound has been detected in water systems all over the word (Zuccato et al., 2005; Zhao et al., 2009, 2010; Yu and Chu, 2009; Pailler et al., 2009; Moldovan et al., 2009; Fernandez et al., 2010; CamachoMunoz et al., 2010; Loos et al., 2010; Teijon et al., 2010; Helenkar et al., 2010; Waiser et al., 2011; Wang et al., 2010, 2011; Lewandowski et al., 2011). IBP has been found to accumulate in fatty fish tissues and muscles (Zhang et al., 2010) and to cause disturbance to both amphipoda (De Lange et al., 2009) and cnidaria (Quinn et al., 2008). IBP effects can be increased when it is present in mixture with other pharmaceuticals. Negative effects at environmentally significant levels have been observed for both cnidaria (Quinn et al., 2009) and human embryonic cells (Pomati et al., 2006). The latter finding is potentially of very high concern when considering the occurrence of IBP in drinking water, at very variable levels depending on the location (Kleywegt et al., 2011; Valcarcel et al., 2011). Biodegradation and phototransformation are potentially important transformation processes for IBP in surface waters (Lin and Reinhard, 2005). Phototransformation might involve direct photolysis and indirect reaction with transient species 1 3 (OH, CO 3 , O2, CDOM*) generated by CDOM, nitrite and nitrate under irradiation (Hoigne´, 1990; Peuravuori and Pihlaja, 2009). As far as the direct photolysis is concerned, IBP is a poor absorber of sunlight (Lin and Reinhard, 2005; Peuravuori and Pihlaja, 2009) but its highest-wavelength absorption band (maximum around 265 nm) is characterised by a quite efficient photolysis (Yuan et al., 2009). Poor sunlight absorption and elevated photolysis quantum yield are contrasting issues as far as the environmental importance of direct photolysis is concerned. A similar case holds for the herbicide 2-methyl-4-chlorophenoxy acetic acid (MCPA), and we have recently found that MCPA direct photolysis can be a significant process but it is highly influenced by the environmental variables (Vione et al., 2010). The available literature data do not allow a clear assessment of the importance of the direct photolysis as a potential transformation pathway for IBP in surface waters (Packer et al., 2003; Lin and Reinhard, 2005; Matamoros et al., 2009; Peuravuori and Pihlaja, 2009). We have recently developed a model approach to describe the transformation kinetics of dissolved compounds as a function of water chemical composition, column depth and photochemical reactivity (i.e. direct photolysis quantum yield 1 3 and reaction rate constants with OH, CO 3 , O2 and CDOM*) (Maddigapu et al., 2011). Such an approach was able to effectively predict field data in the case of 2,4-dichloro-6nitrophenol in the Rhoˆne delta lagoons (Chiron et al., 2007). The goal of the present paper is to assess the role of direct and indirect photochemistry in the degradation of IBP in surface waters. To this purpose, we combine a kinetic study of IBP
transformation by direct photolysis and reaction with OH, 1 3 CO 3 , O2 and CDOM*, with a photochemistry model. The latter enables the prediction of IBP transformation kinetics as a function of photoreactivity data and of key environmental variables. IBP phototransformation intermediates, formed in the first step of the environmentally most significant processes were also identified. IBP has pKa ¼ 4.4 (Martell et al., 1997) and its anionic form would prevail under most conditions that are relevant to surface waters. In this work, irradiation was carried out at both pH 2 and 8 where the neutral and the anionic IBP forms prevail, respectively.
2.
Experimental section
2.1.
Reagents and materials
Ibuprofen (IBP, purity grade 98%), anthraquinone-2-sulphonic acid, sodium salt (AQ2S, 97%), furfuryl alcohol (98%), NaNO3 (>99%), NaHCO3 (98%), anhydrous Na2SO4 (99%), NaCl (99.5%), HClO4 (70%) and H3PO4 (85%) were purchased from Aldrich, NaOH (99%), methanol and 2-propanol (both LiChrosolv gradient grade) and dichloromethane (GC Suprasolv) from VWR Int., Rose Bengal (RB) from Alfa Aesar, riboflavin (>98%) from Sigma.
2.2.
Irradiation experiments
Solutions to be irradiated (5 mL) were placed inside Pyrex glass cells (4.0 cm diameter, 2.3 cm height, 295 nm cut-off wavelength) and magnetically stirred during irradiation. Irradiation of IBP þ nitrate to study reactions with OH and CO 3 , and irradiation of IBP alone to study the direct photolysis were carried out under a Philips TL 01 UVeVis lamp, with emission maximum at 313 nm and 3.0 0.2 W m2 UV irradiance in the 300e400 nm range, measured with a power meter by CO.FO.ME.GRA. (Milan, Italy) equipped with a UV-sensitive probe. The incident photon flux in solution was actinometrically determined using the ferrioxalate method (Kuhn et al., 2004). By knowing, as a function of the wavelength, the fraction of 2þ radiation absorbed by FeðC2 O4 Þ3 3 , the quantum yield of Fe photoproduction and the shape of the lamp spectrum (vide infra), it is possible to use the measured formation rate of Fe2þ to fix the value of the incident spectral photon flux density p (l). The photon flux of the UVeVis lamp between 300 and R 500 nm was Po ¼ p ðlÞdl ¼ 2.0$105 E L1 s1. The transl
formation of IBP photosensitised by AQ2S was studied under a Philips TL K05 UVA lamp, with emission maximum at 365 nm, 28 W m2 UV irradiance (300e400 nm), and 2.1,105 E L1 s1 incident photon flux in solution between 300 and 500 nm. The photodegradation of IBP sensitised by Rose Bengal (RB) via 1O2 was studied under a Philips TL D 18W/16 yellow lamp, with emission maximum at 545 nm and 11 W m2 irradiance in the visible, measured with the CO.FO.ME.GRA. power meter equipped with a probe sensitive to visible radiations. The mixtures IBP-riboflavin were irradiated at 365 nm in a cuvette in parallel beam, using a Xenon lamp (1600 W)
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 2 5 e6 7 3 6
equipped with a Bausch and Lomb monochromator. The incident photon flux in solution, measured by ferrioxalate actinometry, was 1.2 106 E L1 s1. The choice of the lamps had the purpose of exciting each photosensitiser as selectively as possible. The direct photolysis of IBP was studied under a lamp having maximum emission in the UVB region, upon consideration of the IBP absorption spectrum. The emission spectra of the lamps were taken with an Ocean Optics SD 2000 CCD spectrophotometer and normalised to the actinometry results, also taking into account the absorbance of the Pyrex glass walls of the irradiation cells. Note that, irrespective of the presence of the Pyrex glass, no lamp emitted radiation below 300 nm. The absorption spectra of the relevant compounds were taken with a Varian Cary 100 Scan UVeVis spectrophotometer. The various emission and absorption spectra are reported in Fig. 1.
2.3.
Monitoring of IBP transformation
After the scheduled irradiation time, the cells were withdrawn from the lamp and the irradiated solutions were analysed by high-performance liquid chromatography (HPLC-UV) to monitor the time evolution of IBP. The adopted Merck-Hitachi instrument was equipped with an autosampler AS2000A (100 mL sample volume), pumps L-6200 and L-6000 for highpressure gradients, a reverse-phase column Merck LiChrocart RP-C18 packed with LiChrospher 100 RP-18
6727
(125 mm 4.6 mm 5 mm), and a UVeVis detector L-4200 (detection wavelength 279 nm). It was adopted an isocratic elution with a 80:20 mixture of CH3OH:aqueous H3PO4 (pH 2.8) at a flow rate of 1.0 mL min1. The retention time of IBP was 3.3 min, the column dead time 0.90 min. The time evolution of furfuryl alcohol to quantify the formation rate of 1O2 under the yellow lamp was also monitored by HPLC-UV, as reported in Minella et al. (2011).
2.4.
Identification of IBP transformation intermediates
Intermediate identification was carried out with gas chromatography coupled with mass spectrometry. To this purpose, aqueous solutions after irradiation were extracted with 3 mL dichloromethane, dehumidified with anhydrous Na2SO4 and evaporated to dryness. Each sample was reconstructed with 100 mL dichloromethane. The solution was transferred into a vial and injected into a capillary gas chromatograph (Agilent 6890) coupled with a mass spectrometer (Agilent 5973 inert). The injection system used was a Gerstel CIS4 PTV. Initial injection temperature was 40 C, programmed at 5 C/s; final temperature was 320 C, held for 9 min. The injection volume was 2 ml in the splitless mode. The capillary column used was a HP-5MS, 30 m 0.25 mm 0.25 mm film thickness. Initial column temperature was 40 C and was increased by 15 C/ min to 300 C. The carrier gas was ultrapure He (1.0 mL/min; SIAD, Bergamo, Italy). The ionization source worked in the
Fig. 1 e (a) Absorption spectrum of ibuprofen (IBP) at pH 2 and 8. Incident spectral photon flux density p (l) of the adopted TL 01 lamp (emission maximum at 313 nm). (b) Absorption spectrum of Rose Bengal (RB). Incident spectral photon flux density of the yellow lamp (TL D 18W/16 Yellow). (c) Absorption spectrum of antraquinone-2-sulphonate (AQ2S). Incident spectral photon flux density of the UVA lamp (TL K 05).
6728
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 2 5 e6 7 3 6
electronic impact (EI) mode and the mass spectrometer worked in the Scan mode from 44 to 450 Th. Identification of spectra was performed by using the Wiley 7n library (Agilent Part No. G1035B).
2.5.
Kinetic data treatment
Reaction rates were determined by fitting the time evolution data of IBP with pseudo-first order equations of the form Ct ¼ Co exp (k t), where Ct is the concentration of IBP at the irradiation time t, Co its initial concentration and k the pseudofirst order degradation rate constant. The initial degradation rate is RDCNP ¼ k Co. The reported errors on the rates (s) were derived by curve fitting and depend on the scattering of the experimental data around the fitting curve. The reproducibility of repeated runs was around 10e15%. The data plots, the fits and the numerical integrations to determine the absorbed photon fluxes were all carried out with the Fig. P software package (BIOSOFT, Cambridge, UK).
3.
Results and discussion
3.1.
Kinetics of IBP photochemical transformation
3.1.1.
Direct photolysis
IBP (initial concentration 20 mM) was irradiated under the TL 01 lamp (emission maximum at 313 nm, see Fig. 1) at pH 2 and 8. The time evolution of IBP under the adopted experimental set-up is reported in Figure S1-SM (SM ¼ Supplementary Material). Note that significant IBP degradation was obtained at both pH values and that the adopted lamp did not emit radiation below 300 nm. Such results are in agreement with Matamoros et al. (2009). The transformation rate was RateIBP ¼ (8.84 1.07),1011 M s1 at pH 2 and (4.28 0.65),1011 M s1 at pH 8. The photon flux absorbed by IBP can be expressed as R PIBP p ðlÞ½1 10εIBP ðlÞb½IBP dl, where p (l) is the incident a ¼
are expected to take place in the system under consideration are the following (Mack and Bolton, 1999; Buxton et al., 1988): þ NO 3 þ hn þ H / OH þ NO2
½F1 ¼ 0:01
2 Propanol þ OH/Products
k2 ¼ 1:9:109 M1 s1
IBP þ OH/Products ½k3
(1) (2) (3)
Upon application of the steady-state approximation to [OH], one gets the following expression for the initial transformation rate of IBP in the presence of 2-propanol: RateIBP ¼
R:OH $k3 $½IBP k3 $½IBP þ k2 $½2 Propanol
(4)
Fig. 2 reports RateIBP as a function of the concentration of 2propanol, upon irradiation of 10 mM NaNO3 þ 20 mM IBP at pH 2 (open squares) and at pH 8 (solid stars). For each pH value it is also reported the fit curve of Equation (4) to the experimental data and the two 95% confidence bands (dotted for pH 2, dashed for pH 8). It is evident that in both cases the alcohol inhibited the transformation of IBP, which is consistent with the scavenging of OH in reaction (2). Under the adopted conditions (irradiation time up to 4 h) the rate of IBP direct photolysis could be neglected. The experimental rate data were fitted with Equation (4), fixing the values of k2 and [IBP] and letting ROH and k3 vary. From the fit we obtained k3 ¼ (1.5 0.1), 1010 M1 s1 at pH 2 and (1.0 0.3),1010 M1 s1 at pH 8. These values are very near previous results by Parij et al. (1995). Packer et al. (2003) have obtained a somewhat different value, k3 ¼ (6.5 0.2),109 M1 s1 at pH 3.5 upon adoption of the Fenton reaction as OH source. Considering that the Fenton process yields other oxidising species (e.g. ferryl) in addition to OH (Bossmann et al., 2004; Prousek, 2007), which are usually
l
spectral photon flux density of the lamp, εIBP(l) the molar absorption coefficient of IBP at pH 2 (neutral form) or pH 8 (anionic one, see Fig. 1), b the optical path length in solution (0.4 cm), and [IBP] the initial IBP concentration (20 mM). One 11 E L1 s1 at pH 2 and 1.30,1010 E L1 s1 gets PIBP a ¼ 8.73,10 at pH 8. From these data it is possible to obtain the polychromatic photolysis quantum yield of IBP, FIBP, in the UVB region where the spectra of the lamp and IBP overlap. It is 1 ¼ 1.01 0.12 at pH 2 and 0.33 0.05 at pH FIBP ¼ RateIBP (PIBP a ) 8 (m s). The latter datum is not too far from the quantum yield value of 0.2 obtained by Yuan et al. (2009) upon UVC irradiation of ibuprofen at pH 7. This high quantum yield of photolysis is consistent with the observed photodecarboxylation reactions (vide infra), which are reported to be very efficient photolysis pathways (Budac and Wan, 1992).
3.2.
Reaction with OH
The reaction rate constant between IBP and OH was determined upon competition kinetics with 2-propanol, using nitrate photolysis as the OH source. The main reactions that
Fig. 2 e Initial transformation rates of 20 mM IBP upon irradiation of 10 mM NaNO3, as a function of the concentration of added 2-propanol, at pH 2 (open squares, dottes curves) and 8 (solid stars, dashed curves). The pH values were adjusted with HClO4 and NaOH, respectively. Note that there are three curves for each data set. The central curve is the fit one, the lateral curves are the 95% confidence bands. nM [ 10L9 M. Irradiation was carried out under the TL 01 lamp.
6729
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 2 5 e6 7 3 6
less reactive than OH itself, it cannot be excluded that use of the Fenton reaction could lead to a slight underestimation of the OH rate constant.
3.2.1.
Reaction with CO 3
A semi-quantitative assessment of the reactivity of can be carried out by studying the a substrate with CO 3 effect of bicarbonate on the transformation photoinduced by nitrate (Vione et al., 2009a). Fig. 3 reports the initial transformation rate of 20 mM IBP upon irradiation of 10 mM NaNO3, with the addition of variable NaHCO3 concentrations. In the presence of bicarbonate up to 10 mM, the solution pH gradually increased from 6.5 to 8.5. The trend of the IBP rate with irradiated nitrate is also shown upon addition of a phosphate buffer (NaH2PO4 þ Na2HPO4). The total concentration of phosphate and the ratio NaH2PO4/Na2HPO4 were chosen to have the same concentration values as for NaHCO3, and the same pH within 0.1 units. Finally, the transformation rate of IBP in the presence of NaHCO3 without nitrate is also reported. From Fig. 3 it can be derived that: (i) the direct photolysis of IBP (irradiation without nitrate) was negligible under the adopted experimental conditions; (ii) bicarbonate inhibited the transformation of IBP upon nitrate irradiation, while phosphate had practically no effect. Nitrate photolysis yields OH (reaction 1), and the addition of NaHCO3 induces the production of CO 3 by reaction between the hydroxyl radical 2 and HCO 3 /CO3 (Buxton et al., 1988). There would be compe 2 tition between IBP and HCO 3 /CO3 for reaction with OH, but CO3 (although it is less reactive than OH) could also contribute to the transformation of IBP. Vione et al. (2009a) have found that the formation rate of OH þ CO 3 by irradiation of nitrate and bicarbonate is higher than the rate of OH formation with nitrate alone. A likely explanation is that OH and NO2, photogenerated upon nitrate photolysis, undergo
recombination to nitrate and Hþ when still in the solvent cage and before diffusing to the solution bulk. An excess bicarbonate could react with cage OH and inhibit recombination, thereby enhancing the production of reactive species upon nitrate photolysis (reactions (5e8), Bouillon and Miller, 2005; Nissenson et al., 2010). þ NO 3 þ hn þ H / OH þ NO2 cage
(5)
½ OH þ NO2 cage / OH þ NO2
(6)
þ ½ OH þ NO2 cage /NO 3 þH
(7)
½ OH þ NO2 cage þHCO 3 /CO3 þ H2 O þ NO2
(8)
Because it induces the formation of a higher amount of a less reactive species, the addition of bicarbonate to irradiated nitrate enhances the transformation of the compounds and OH that show a ratio of the rate constants with CO 3 higher than 0.01 (Vione et al., 2009a). The same compounds could also undergo significant transformation by CO 3 in the environment (Vione et al., 2009b). In contrast, inhibition of transformation by bicarbonate is observed for molecules that would also undergo insignificant reaction with CO 3 in surface waters, such as 4-nitrophenol. However, considering that the addition of bicarbonate would also alter the solution pH, with implications for nitrate photolysis and possibly for substrate reactivity, enhancement or inhibition by bicarbonate are defined by comparison with phosphate buffers at the same pH values (Vione et al., 2009a). In the case of IBP, the inhibition of transformation by bicarbonate compared to phosphate suggests that the reaction between IBP and CO 3 in surface waters would not be important (the relevant second-order rate constant is expected to be significantly below 108 M1 s1).
3.2.2.
Reaction with 1O2
Fig. 4 reports the initial transformation rate of IBP, as a function of its initial concentration, upon irradiation of 10 mM Rose Bengal (RB) adopted as a source of 1O2 (reaction (9); Miller, 2005). Irradiation was carried out only at pH 8, because at pH 2 RB is protonated and unable to absorb visible radiation to a significant extent. The reaction (10) between IBP and 1O2 would be in competition with the thermal deactivation of singlet oxygen (reaction (11); Wilkinson and Brummer, 1981; Rodgers and Snowden, 1982): RB þ hn þ O2 /RB þ 1 O2 IBP þ 1 O2 /Products Fig. 3 e Initial transformation rates upon irradiation of (,) 20 mM IBP and 10 mM NaNO3, as a function of the concentration of NaHCO3; (D) 20 mM IBP and 10 mM NaNO3, as a function of the concentration of added phosphate buffer (same concentration as NaHCO3 and same pH, within 0.1 units); (>) 20 mM IBP without nitrate, as a function of NaHCO3 concentration. nM [ 10L9 M. Irradiation was carried out under the TL 01 lamp.
1
O2 /O2 þ heat
½k10
k11 ¼ 2:5$105 s1
(9) (10) (11)
Upon application of the steady-state approximation to 1O2 one gets the following expression for the initial transformation rate of IBP (RateIBP): RateIBP ¼
R1O2 $k10 $½IBP
k11 þ k10 $½IBP
(12)
6730
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 2 5 e6 7 3 6
Fig. 4 e Initial transformation rates of IBP upon irradiation of 10 mM Rose Bengal (RB) under the yellow lamp (Philips TL D 18W/16), as a function of the IBP concentration. The solution pH was 8, adjusted with NaOH. The lamp and RB spectra are reported in Fig. 1b. pM [ 10L12 M. The regression line is dashed, the 95% confidence bands are dotted.
Fig. 5 e Initial transformation rates of IBP upon UVA irradiation of 1 mM AQ2S, as a function of the concentration of IBP at pH 2 and 8. The pH values were adjusted with HClO4 and NaOH, respectively. The regression lines are dashed, the 95% confidence bands are dotted.
where R1O2 is the formation rate of 1O2 by 10 mM RB under the adopted irradiation device. The measurement of R1O2 was carried out upon irradiation at pH 8 of 10 mM RB þ 0.1 mM furfuryl alcohol (FFA), which reacts with 1O2 with a rate constant kFFA ¼ 1.2,108 M1 s1 (Haag et al., 1984). The initial transformation rate of FFA under the adopted conditions was RateFFA ¼ (1.26 0.06),107 M s1. Photogenerated 1O2 could undergo deactivation or reaction with FFA, and upon application of the steady-state approximation to [1O2] one gets:
adopted concentration value (10 mM). Therefore, it is possible to simply determine the photon flux absorbed by AQ2S as R PAQ2S ¼ p ðlÞ½1 10εAQ2S ðlÞ b ½AQ2S dl ¼ 2.01,106 E L1 s1 a
R1 O2 ¼ RateFFA $
k11 þ kFFA $½FFA kFFA $½FFA
(13)
From Equation (13) one gets R1O2 ¼ (2.75 0.13),106 M s1. The linear trend of RateIBP vs. [IBP] in Fig. 4 suggests that, in equation (12), k10 [IBP] « k11. If this approximation holds, one gets RateIBP ¼ R1O2 k10 (k11)1 [IBP]. From Fig. 4 one obtains RateIBP ¼ (6.60 0.33),107 [IBP], and from the known values of R1O2 and k11 one gets k10 ¼ (6.00 0.58),104 M1 s1 as the reaction rate constant between IBP and 1O2. This result confirms that the hypothesis k10 [IBP] « k11 was justified.
3.2.3.
Reaction with irradiated AQ2S
The choice of AQ2S is motivated by the fact that this compound has a reactive triplet state but it does not yield either OH or 1O2 under irradiation, thus avoiding formation of potentially interfering transients (Loeff et al., 1983; Maurino et al., 2008; Maddigapu et al., 2010c). Fig. 5 reports the initial transformation rate of IBP as a function of its initial concentration, upon UVA irradiation of 0.1 mM AQ2S at pH 2 and pH 8. From the linear trend of the plot one gets RateIBP ¼ (4.66 0.17),104 [IBP] at pH 2 and (3.19 0.09),104 [IBP] at pH 8. AQ2S would be by far the main radiation absorber at both pH values. From the absorption spectra (Fig. 1), one gets that at 330 nm the absorbance of 0.1 mM AQ2S would be 104 times or more higher than that of IBP at the highest
l
(p (l) is the spectral incident photon flux density of the UVA lamp and b ¼ 0.4 cm). The polychromatic quantum yield of IBP phototransformation by AQ2S would thus be )1 ¼ (2.32 0.08),102 [IBP] at pH 2 and FIBP,AQ2S ¼ RateIBP (PAQ2S a (1.59 0.04),102 [IBP] at pH 8. The reactive triplet state 3AQ2S* accounts for the degradation processes that take place with AQ2S under irradiation. It has a formation quantum yield F3AQ2S* ¼ 0.18 and a deactivation rate constant k3AQ2S* ¼ 1.1,107 s1 (Loeff et al., 1983). , The formation rate of 3AQ2S* would be R3AQ2S* ¼ F3AQ2S* PAQ2S a and its deactivation would be in competition with the reaction with IBP (rate constant k3AQ2S*,IBP). Therefore, the transformation rate of IBP by irradiated AQ2S could be expressed as follows: $ RateIBP ¼ F3 AQ2S $PAQ2S a
k3 AQ2S ; IBP $½IBP k3 AQ2S þ k3 AQ2S ; IBP $½IBP
(14)
Under the hypothesis that k3AQ2S*,IBP [IBP] « k3AQ2S*, one gets k3AQ2S*,IBP (k3AQ2S*)1 [IBP], which is RateIBP ¼ F3AQ2S* PAQ2S a compatible with the linear trend reported in Fig. 5. It is also )1 ¼ F3AQ2S* k3AQ2S*,IBP (k3AQ2S*)1 [IBP]. FIBP,AQ2S ¼ RateIBP (PAQ2S a By comparison with the experimental data, FIBP,AQ2S ¼ (2.32 0.08),102 [IBP] at pH 2 and (1.59 0.04),102 [IBP] at pH 8, one gets k3AQ2S*,IBP ¼ (1.42 0.05),1010 and (9.70 0.24),109 M1 s1 at pH 2 and 8, respectively. This finding confirms that the hypothesis k3AQ2S*,IBP [IBP] « k3AQ2S* was reasonable.
3.2.4.
Reaction with irradiated riboflavin
Riboflavin is another well-known sensitiser. It produces 1O2 under irradiation and its excited triplet state (3Ri*) also reacts
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 2 5 e6 7 3 6
through electron transfer with electron-donor molecules (Barbieri et al., 2008). The solutions containing riboflavin (5.4 mM) and IBP (10 mM) were buffered at pH 8 and irradiated in aerated solution for 1 h, to limit riboflavin loss to 10%. The rate of IBP consumption was quite small: 5.5,1011 M s1. Considering that the absorbance of riboflavin at 365 nm is 0.053, the incident photon flux is 1.2,106 E L1 s1 and the quantum yield of 3Ri* formation is 0.375 in neutral water (Islam et al., 2003), the rate of 3Ri* production in our system would be 5.1,108 M s1. In another set of experiments, we observed that 3Ri* is efficiently quenched by oxygen with a rate constant of 1.06,109 M1 s1. The pseudo first-order rate constant of 3Ri* deactivation (k3Ri*) in neutral deoxygenated aqueous solution is 1.5,105 s1 (Kouras et al., unpublished results), which raises to 4.2,105 s1 in aerated systems because of the reaction between 3Ri* and O2. When considering both the reaction between 3Ri* and IBP and the production of 1O2 by 3Ri*, followed by reaction of 1O2 with IBP, it is possible to write the rate of IBP transformation as follows:
RateIBP ¼5:1$108 $
k3 Ri ; IBP $½IBP 2:7$105 k10 $½IBP þ 5 4:2,10 þk3 Ri ; IBP $½IBP 4:2$105 k11 þk10 $½IBP
(15) The first term in the bracket corresponds to the oxidation of IBP by 3Ri* and the second to the oxidation of IBP by 1O2. Using k10 ¼ (6.00 0.58),104 M1 s1 and k11 ¼ 2.5,105 s1, the rate of IBP consumption through 1O2 is equal to (7.9 0.8),1014 M s1 and is thus negligible. This means that in our experimental conditions IBP mainly disappears by direct reaction with 3Ri*. The rate constant of the reaction k3Ri*,IBP is estimated to be (4.5 0.4),107 M1 s1. This value is two orders of magnitude lower than that measured in the case of AQ2S at the same pH value.
3.3. Modelling the phototransformation kinetics of IBP in surface waters The availability of the direct photolysis quantum yield and of the reaction rate constants of the anionic IBP form with OH, 1 O2 and 3AQ2S*/3Ri* (proxies for 3CDOM*) allow the calculation of the first-order rate constant of IBP transformation (kIBP) as
6731
a function of water chemical composition and column depth. The model approach to link substrate reactivity and environmental features is described in detail elsewhere (Albinet et al., 2010a; Maddigapu et al., 2010a, 2011; Vione et al., 2010; Hatipoglu et al., 2010). Figure S2-SM reports the sunlight spectrum adopted for the calculations, which corresponds to a 22 W m2 irradiance in the UV (Frank and Klo¨pffer, 1988). A major issue for the assessment of outdoor reactivity is that sunlight irradiance is not constant. The time unit adopted in the cited model is the summer sunny day (SSD), which corresponds to a fair-weather 15 July at 45 N latitude. The incident UV energy in an SSD is equivalent to 10 h continuous irradiation at 22 W m2 UV irradiance (Maddigapu et al., 2010b). When applying the model, the first-order rate constant kIBP of IBP phototransformation is given in units of SSD1, while the halflife times sIBP ¼ ln 2 (kIBP)1 have units of SSD. Inclusion in the model of the IBP kinetic data obtained here suggests that direct photolysis and reactions with OH and possibly 3CDOM* would be the main IBP transformation pathways in surface waters. Under the hypothesis that k3CDOM*,IBP w k3AQ2S*,IBP ¼ 9.7,109 M1 s1, with low dissolved organic carbon (DOC, expressed as NPOC, Non-Purgeable Organic Carbon) the order of importance of the three relevant pathways would be OH > photolysis > 3CDOM*. At high NPOC it would be 3CDOM* > photolysis > OH. Such a scenario could be compatible with previous results of IBP photodegradation in surface water samples under irradiation, which suggest that reaction with OH would not be the only relevant process (Packer et al., 2003). More recently, significant reactivity has been found between IBP and fulvic acids under irradiation (Jacobs et al., 2011). On the other hand, taking k3CDOM*,IBP w k3Ri*,IBP ¼ 4.5,107 M1 s1, the reaction with 3 CDOM* would be minor even at high NPOC. Fig. 6 reports the model results for kIBP as a function of nitrite and NPOC, with fixed nitrate (51 mM), bicarbonate (2.1 mM) and carbonate (26 mM) and with a water column depth of 1 m. In Fig. 6a it is hypothesised that IBP reacts with 3 CDOM* with k3CDOM*,IBP ¼ 9.7,109 M1 s1, following the results obtained with AQ2S. In Fig. 6b it is hypothesised k3CDOM*,IBP ¼ 4.5,107 M1 s1, in analogy with the riboflavin data. As far as Fig. 6a is concerned, at low nitrite kIBP has
Fig. 6 e (a) First-order transformation rate constant of IBP, as a function of nitrite concentration and the NPOC values. Other conditions: 51 mM nitrate, 2.1 mM bicarbonate, 26 mM carbonate. Here it is hypothesised kIBP,3CDOM* [ 9.7,109 ML1 sL1. (b) First-order transformation rate constant of IBP, under the hypothesis that kIBP,3CDOM* [ 4.5,107 ML1 sL1. Other conditions are same as before.
6732
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 2 5 e6 7 3 6
a minimum for 1 mg C L1 NPOC. In the system described by Fig. 6a nitrite would be the main OH source (with the exclusion of the highest NPOC values, where CDOM would prevail as source), DOM the main OH sink, and 3CDOM* would obviously be produced by organic matter alone. At low NPOC, the kIBP decrease with NPOC is due to OH scavenging by DOM. The increase of kIBP after the minimum would be accounted for by the role of 3CDOM* in the transformation of IBP. If the reaction between IBP and 3CDOM* is negligible, a continuous decrease of kIBP with increasing NPOC is expected as shown in Fig. 6b. In the latter case, DOM would scavenge OH and its chromophoric moieties would compete with IBP for radiation absorption, thereby inhibiting the IBP direct photolysis. The data of Fig. 6a foresee an IBP half-life time sIBP ¼ ln 2 (kIBP)1 of less than one week, while those of Fig. 6b suggest that the half-life time could vary from a few days to a couple of months. Also note that the adopted column depth of 1 m is quite favourable to photochemistry. An assessment of the importance of the reaction between IBP and 3CDOM* can be obtained by application of our model to the IBP degradation kinetics reported by Packer et al. (2003) upon irradiation of Mississippi river water. Water data were 63 mM nitrate, 8.9 mg C L1 NPOC, bicarbonate and carbonate not reported (but they are minor OH sinks and the reaction between IBP and CO 3 is negligible) (Packer et al., 2003). The solutions have been placed in quartz bottles with diameter d ¼ 4.0 cm, which would yield a geometrical slab-equivalent optical path length b ¼ 2 d/p ¼ 2.5 cm (Albinet et al., 2010b). Note however that, for such small b values where the innerfilter effect of the solution is expected to be small, the IBP phototransformation kinetics would depend very little on the path length. The adopted lamp was approximately five times more intense than sunlight. With this apparatus, Packer et al. (2003) obtained a first-order rate constant of 0.45 day1, which would become 0.09 day1 under normal sunlight. With the input data reported above, our model yielded kIBP ¼ 0.24 SSD1 with k3CDOM*,IBP ¼ 4.5,107 M1 s1 and 0.73 SSD1 with k3CDOM*,IBP ¼ 9.7,109 M1 s1. Under the hypothesis that 1 SSD ¼ 1 day of Packer et al., it is quite likely that CDOM in Mississippi river water is poorly reactive toward IBP. Moreover, our model foresees that direct photolysis would be the main transformation reaction in the system, with OH only contributing to 0.02 SSD1. Note that the model adopts the sunlight spectrum reported in Figure S2-SM, which might be significantly different from that of the lamp used by Packer et al. (medium-pressure Hg-vapour lamp filtered by borosilicate glass), in particular at the wavelengths around 300 nm that are the most significant for the direct photolysis of IBP. It is also possible to compare the IBP elimination rate constant in late summer months, derived from the field data of Tixier et al. (2003) in the epilimnion of the Swiss Lake Greifensee (5 m depth) with the model results, to get insight into the potential importance of photochemistry in the fate of IBP in surface waters. Tixier et al. (2003) obtained a field rate constant of 0.022 day1, which would include all possible processes such as outflow, sedimentation, OH-induced photodegradation and possibly biodegradation. Note that the known processes accounted for approximately one half of the observed transformation. With the Lake Greinfensee data (0.1 mM nitrate, 3.5 mg C L1 NPOC, 2 mM bicarbonate, 10 mM carbonate)
(Canonica et al., 2005) and d ¼ 5 m, one obtains from our model kIBP w 0.030 SSD1 for the sum of OH reaction and direct photolysis, of which 0.024 SSD1 would be accounted for by direct photolysis alone. Tixier et al. (2003) report that unfavourable meteorology would approximately halve the importance of photochemistry in the period under study, while our model hypothesises constant good weather. By halving kIBP and its two components (OH and photolysis) in our model, one gets a direct photolysis contribution of 0.012 SSD1. Under the hypothesis that 1 SSD ¼ 1 day, one gets a reasonable match between model results concerning direct photolysis (0.012 SSD1) and the IBP degradation kinetics in Lake Greifensee that could not be accounted for by the known processes (0.006e0.011 day1). Such a comparison suggests that direct photolysis might be a significant pathway for IBP transformation in the lake epilimnion during the summer months.
3.4. Identification of IBP phototransformation intermediates To identify the intermediates of the potentially most important photochemical transformation pathways of IBP in surface waters (direct photolysis, OH, 3AQ2S*), a significantly higher initial IBP concentration (1 mM) was used than in the kinetic studies. It was adopted pH 8, adjusted with NaOH. Other conditions were as follows: 1 M NaNO3 for OH, 1 mM AQ2S for triplet state reactivity. The adopted irradiation times were up to 6 h for OH, up to 3 h for 3AQ2S*, and up to 3 days for the direct photolysis. Table 1 reports the IBP transformation intermediates identified by GCeMS following the different pathways. Experimental mass spectra and the comparisons with the spectra libraries are reported as SM. The direct photolysis of IBP yielded III and IV as detected phototransformation intermediates. Interestingly, the same compounds have been detected upon 254-nm irradiation of IBP (Szabo´ et al., 2011), which is reasonable because both UVC irradiation and our conditions (UVB) would excite the highestwavelength absorption band of IBP. Compound III was also detected in the presence of OH, and IV was detected in all the three pathways under study (OH, direct photolysis and 3 AQ2S*). Considering that all the three systems were irradiated, direct photolysis could be operational at some extent also in the presence of nitrate and AQ2S. However, III has been identified upon IBP transformation in photo-Fenton systems (Me´ndez-Arriaga et al., 2010), where it probably originated by OH substitution at the carbon in alpha to the carboxylic group (probably via H abstraction followed by OH addition) followed by decarboxylation/oxidation. Moreover, we detected IV in the system IBP þ AQ2S that was irradiated under the UVA lamp, where the direct photolysis of IBP should be low to negligible. The same compound has been detected by Jacobs et al. (2011) in the presence of irradiated fulvic acids, under conditions where the direct photolysis of IBP was negligible. Compound I was identified with IBP þ 3AQ2S* and derives by transformation of the lateral chain containing the carboxylic group. Transformation of the COOH-containing (propanoic) chain was also observed for III and IV (already discussed). Compound II, only detected in the presence of 3 AQ2S*, would be produced upon shortening of both lateral chains of IBP. Interestingly, this is the only detected
6733
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 2 5 e6 7 3 6
Table 1 e IBP transformation intermediates detected under the different studied conditions. Near each formula it is reported the GC retention time (tR) and the reaction pathway that generated the relevant intermediate. DP [ direct photolysis, n/a [ not applicable, (*) [ tentative identification. The match between the experimental mass spectra and the library ones is reported in the Supplementary Material. Identification implied an overall mass spectrometric match >85% and the match of the most abundant ions and of the molecular one. Compound
Acronym
tR, min
Formation pathway
IBP
15.04
n/a
I
11.65
3
AQ2S*
II
12.59
3
AQ2S*
III
13.08
IV
13.40
V (*)
14.72
3
OH, DP
OH, DP, 3AQ2S*
AQ2S*
intermediate that underwent transformation of the isobutyl chain, which is thus expected to be much less reactive compared to the propanoic one. The identification of V (only formed with 3AQ2S*) was only tentative, because of the lack of the relevant mass spectrum in the adopted library. Identification was mainly based on the mass of the molecular ion that yielded the likely chemical formula C13H18O, namely an oxygen atom less than IBP. If this is the case, V would arise from IBP þ 3AQ2S* following a reductive pathway, which might possibly involve the reduced radical species AQ2S and/or AQ2S-H, or superoxide/hydroperoxide. The latter may be formed via the following reactions (Maurino et al., 2008, 2011; Maddigapu et al., 2010c) (ISC ¼ Inter-System Crossing, SeH ¼ substrate): ISC
AQ2S þ hn!3 AQ2S
(16)
3
AQ2S /AQ2S or products
(17)
3
AQ2S þ AQ2S/AQ2S þ AQ2Sþ
(18)
3
AQ2S þ S H/AQ2S H þ S
(19)
Scheme 1 e A summary of the processes leading to the detected intermediates under the studied irradiation conditions. Note that identification of V is only tentative, as is the assignment of III and IV as IBP transformation intermediates upon reaction with OH and/or 3AQ2S* (see dashed arrows). DP [ direct photolysis; (*) [ tentative identification.
6734
3
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 2 5 e6 7 3 6
AQ2S þ S H/AQ2S þ S Hþ
(20)
Appendix. Supplementary data
AQ2S þ O2 /AQ2S þ O 2
(21)
Supplementary data related to this article can be found online at doi:10.1016/j.watres.2011.10.014.
AQ2S H þ O2 /AQ2S þ HO2
(22)
Scheme 1 summarises the transformation pathways leading to the detected IBP intermediates. The dashed arrows connecting IBP with III and IV via 3AQ2S* and OH account for the theoretical possibility for them to be produced by IBP direct photolysis when nitrate and AQ2S are irradiated. Interestingly, the transformation intermediates detected in the present study do not match those identified upon IBP degradation with irradiated TiO2 (Me´ndez-Arriaga et al., 2008), which suggest that IBP in surface waters and upon TiO2 photocatalysis would follow quite different reaction pathways.
4.
Conclusions
Photochemistry can be an important transformation process of IBP in surface waters. The prevailing pathways involved are the reaction with OH, the direct photolysis and possibly, depending on IBP reactivity, transformation induced by 3CDOM*. Anionic IBP, which prevails at the pH values of surface waters has a polychromatic UVB photolysis quantum yield FIBP ¼ 0.33 0.05, a reaction rate constant with OH kIBP,OH ¼ (1.0 0.3),1010 M1 s1, a 1O2 rate constant kIBP; 1O2 ¼ (6.00 0.58),104 M1 s1, a 3AQ2S* rate constant kIBP,3AQ2S* ¼ (9.70 0.24),109 M1 s1, and a 3Ri* rate constant kIBP,3Ri* ¼ (4.5 0.4),107 M1 s1. Moreover, IBP does not react with CO 3 to a significant extent. The results obtained with AQ2S and riboflavin as model systems for CDOM are quite contrasting, showing that the reactivity of IBP with oxidant triplets is greatly influenced by the triplet nature. We compared our model predictions with literature data of irradiation of Mississippi water samples and with a field study carried out in Lake Greinfensee, Switzerland. In both cases it is suggested a poor reactivity of 3CDOM* toward IBP and an important role of the direct photolysis. IBP phototransformation mainly takes place via lateral chain shortening, and the propanoic chain is much more reactive than the isobutyl one. The IBP aldehyde, which would arise upon IBP reduction, was also tentatively identified in the presence of AQ2S under irradiation.
Acknowledgements Financial support by PNRA e Progetto Antartide is gratefully acknowledged. The work of PRM in Torino was supported by a Marie Curie International Incoming Fellowship (IIF), under the FP7-PEOPLE programme (contract n PIIF-GA-2008-219350, project PHOTONIT). The PhD grant of EDL was funded by Progetto Lagrange - Fondazione CRT, Torino, Italy.
references
Albinet, A., Minero, C., Vione, D., 2010a. UVA irradiation induces direct phototransformation of 2,4-dinitrophenol in surface water samples. Chemosphere 80 (7), 759e763. Albinet, A., Minero, C., Vione, D., 2010b. Phototransformation processes of 2,4-dinitrophenol, relevant to atmospheric water droplets. Chemosphere 80 (7), 753e758. Barbieri, Y., Massad, W.A., Dı´az, D.J., Sanz, J., Amat-Guerri, F., Garcı´a, N.A., 2008. Photodegradation of bisphenol A and related compounds under natural-like conditions in the presence of riboflavin: kinetics, mechanism and photoproducts. Chemosphere 73 (4), 564e571. Bossmann, S.H., Oliveros, E., Kantor, M., Niebler, S., Bonfill, A., Shahin, N., Worner, M., Braun, A.M., 2004. New insights into the mechanisms of the thermal Fenton reactions occurring using different iron(II)-complexes. Water Science and Technology 49 (4), 75e80. Bouillon, R.C., Miller, W.L., 2005. Photodegradation of dimethyl sulfide (DMS) in natural waters: Laboratory assessment of the nitrate-photolysis-induced DMS oxidation. Environmental Science and Technology 39 (24), 9471e9477. Budac, D., Wan, P., 1992. Photodecarboxylation-Mechanism and synthetic utility. Journal of Photochemistry and Photobiology A: Chemistry 67 (2), 135e166. Buxton, G.V., Greenstock, C.L., Helman, W.P., Ross, A.B., 1988. Critical review of rate constants for reactions of hydrated electrons, hydrogen atoms and hydroxyl radicals (OH/O) in aqueous solution. Journal of Physical and Chemical Reference Data 17 (3), 1027e1284. Camacho-Munoz, M.D., Sanots, J.L., Aparicio, I., Alonso, E., 2010. Presence of pharmaceutically active compounds in Donana Park (Spain) main watersheds. Journal of Hazardous Materials 177 (1e3), 1159e1162. Canonica, S., Kohn, T., Mac, M., Real, F.J., Wirz, J., Von Gunten, U., 2005. Photosensitizer method to determine rate constants for the reaction of carbonate radical with organic compounds. Environmental Science & Technology 39 (23), 9182e9188. Castiglioni, S., Bagnati, R., Fanelli, R., Pomati, F., Calamari, D., Zuccato, E., 2006. Removal of pharmaceuticals in sewage treatment plants in Italy. Environmental Science and Technology 40 (1), 357e363. Chiron, S., Minero, C., Vione, D., 2007. Occurrence of 2,4dichlorophenol and of 2,4-dichloro-6-nitrophenol in the Rhoˆne river delta (Southern France). Environmental Science and Technology 41 (9), 3127e3133. Daneshvar, A., Svanfelt, J., Kronberg, I., Weyhenmeyer, G.A., 2010. Winter accumulation of acidic pharmaceuticals in a Swedish river. Environmental Science and Pollution Research 17 (4), 908e916. de Graaf, M.S., Vieno, N.M., Kujawa-Roeleveld, K., Zeeman, G., Temmink, H., Buisman, C.J.N., 2011. Fate of hormones and pharmaceuticals during combined anaerobic treatment and nitrogen removal by partial nitritation-anammox in vacuum collected black water. Water Research 45 (1), 375e383. De Lange, H.J., Peeters, E.T.H.M., Lurling, M., 2009. Changes in ventilation and locomotion of Gammarus pulex (Crustacea, Amphipoda) in response to low concentrations of
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 2 5 e6 7 3 6
pharmaceuticals. Human and Ecological Risk Assessment 15 (1), 111e120. Fernandez, C., Gonzales-Doncel, M., Pro, J., Carbonell, G., Tarazona, J.V., 2010. Occurrence of pharmaceutically active compounds in surface waters of the Henares-Jarama-Tajo river system (Madrid, Spain) and a potential risk characterization. Science of the Total Environment 408 (3), 543e551. Frank, R., Klo¨pffer, W., 1988. Spectral solar photon irradiance in Central Europe and the adjacent North Sea. Chemosphere 17 (5), 985e994. Hatipoglu, A., Vione, D., Yalcin, Y., Minero, C., Cinar, Z., 2010. Photo-oxidative degradation of toluene in aqueous media by hydroxyl radicals. Journal of Photochemistry and Photobiology A: Chemistry 215 (1), 59e68. Haag, W.R., Hoigne´, J., Gassman, E., Braun, A.M., 1984. Singlet oxygen in surface waters - part I: furfuryl alcohol as a trapping agent. Chemosphere 13 (5e6), 631e640. Helenkar, A., Sebok, A., Zaray, G., Molnar-Pearl, I., VasanitsZsigrai, A., 2010. The role of the acquisition methods in the analysis of the non-steroidal anti-inflammatory drugs in Danube River by gas chromatography e mass spectrometry. Talanta 82 (2), 600e607. Hoigne´, J., 1990. Formulation and calibration of environmental reaction kinetics: Oxidations by aqueous photooxidants as an example. In: Stumm, W. (Ed.), Aquatic Chemical Kinetics. Wiley, NY, pp. 43e70. Islam, S.D.M., Penzkofer, A., Hegemann, P., 2003. Quantum yield of triplet formation of riboflavin in aqueous solution and of flavin mononucleotide bound to the LOV1 domain of Phot1 from Chlamydomonas reinhardtii. Chemical Physics 291 (1), 97e114. Jacobs, L.E., Fimmen, R.L., Chin, Y.-P., Mash, H.E., Weavers, L.K., 2011. Fulvic acid mediated photolysis of ibuprofen in water. Water Research 45 (15), 4449e4458. Kleywegt, S., Pileggi, V., Yang, P., Hao, C.Y., Zhao, X.M., Rocks, C., Thach, S., Cheung, P., Whitehead, B., 2011. Pharmaceuticals, hormones and bisphenol A in untreated source and finished drinking water in Ontario, Canada and treatment efficiency. Science of the Total Environment 409 (8), 1481e1488. Kouras, S., Amine-Khodja, A., Richard, C. Unpublished results. Kuhn, H.J., Braslavsky, S.E., Schmidt, R., 2004. Chemical actinometry. Pure and Applied Chemistry 76 (12), 2105e2146. Lewandowski, J., Putschew, A., Schwesig, D., Neumann, C., Radke, M., 2011. Fate of organic micropollutants in the hyporheic zone of a eutrophic lowland stream: results of a preliminary field study. Science of the Total Environment 409 (10), 1824e1835. Lin, A.Y.C., Reinhard, M., 2005. Photodegradation of common environmental pharmaceuticals and estrogens in river water. Environmental Toxicology and Chemistry 24 (6), 1303e1309. Loeff, I., Treinin, A., Linschitz, H., 1983. Photochemistry of 9,10anthraquinone-2-sulfonate in solution. 1. Intermediates and mechanism. Journal of Physical Chemistry 87 (14), 2536e2544. Loos, R., Locoro, G., Contini, S., 2010. Occurrence of polar organic contaminants in the dissolved water phase of the Danube River and its major tributaries using SPE-LC-MS2 analysis. Water Research 44 (7), 2325e2335. Mack, J., Bolton, J.R., 1999. Photochemistry of nitrite and nitrate in aqueous solution: a review. Journal of Photochemistry and Photobiology A: Chemistry 128 (1e3), 1e13. Maddigapu, P.R., Minero, C., Maurino, V., Vione, D., Brigante, M., Mailhot, G., 2010a. Enhancement by anthraquinone-2sulphonate of the photonitration of phenol by nitrite: Implication for the photoproduction of nitrogen dioxide by coloured dissolved organic matter in surface waters. Chemosphere 81 (11), 1401e1406. Maddigapu, P.R., Vione, D., Ravizzoli, B., Minero, C., Maurino, V., Comoretto, L., Chiron, S., 2010b. Laboratory and field evidence
6735
of the photonitration of 4-chlorophenol to 2-nitro-4chlorophenol and of the associated bicarbonate effect. Environmental Science and Pollution Research 17 (5), 1063e1069. Maddigapu, P.R., Bedini, A., Minero, C., Maurino, V., Vione, D., Brigante, M., Mailhot, G., Sarakha, M., 2010c. The pHdependent photochemistry of anthraquinone-2-sulfonate. Photochemical and Photobiological Sciences 9 (3), 323e330. Maddigapu, P.R., Minella, M., Vione, D., Maurino, V., Minero, C., 2011. Modeling phototransformation reactions in surface water bodies: 2,4-Dichloro-6-nitrophenol as a case study. Environmental Science and Technology 45 (1), 209e214. Martell, A.E., Smith, R.M., Motekaitis, R.J., 1997. Critically Selected Stability Constants of Metal Complexes Database, Version 4.0. Matamoros, V., Duhec, A., Albaiges, J., Bayona, J.M., 2009. Photodegradation of carbamazepine, ibuprofen, ketoprofen and 17-alpha-ethinylestradiol in fresh and seawater. Water, Air and Soil Pollution 196 (1e4), 161e168. Maurino, V., Borghesi, D., Vione, D., Minero, C., 2008. Transformation of phenolic compounds upon UVA irradiation of anthraquinone-2-sulfonate. Photochemical and Photobiological Sciences 7 (3), 321e327. Maurino, V., Bedini, A., Borghesi, D., Vione, D., Minero, C., 2011. Phenol transformation photosensitised by quinoid compounds. Physical Chemistry Chemical Physics 10 (4), 601e609. 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. Me´ndez-Arriaga, F., Esplugas, S., Gime´nez, J., 2010. Degradation of the emerging contaminant ibuprofen in water by photofenton. Water Research 44, 589e595. Miller, J.S., 2005. Rose-Bengal sensitized photooxidation of 2chlorophenol in water using solar simulated light. Water Research 39 (2e3), 412e422. Minella, M., Romeo, F., Vione, D., Maurino, V., Minero, C., 2011. Low to negligible photoactivity of lake-water matter in the size range from 0.1 to 5 mm. Chemosphere 83 (11), 1480e1485. Moldovan, Z., Chira, R., Alder, A.C., 2009. Environmental exposure of pharmaceuticals and musk fragrances in the Somes River before and after upgrading the municipal wastewater treatment plant Cluj-Napoca, Romania. Environmental Science and Pollution Research 16 (Suppl. I), 46e54. Morasch, B., Bonvin, F., Reiser, H., Grandjean, D., de Alencastro, L. F., Perazzolo, C., Chevre, N., Kohn, T., 2010. Occurrence and fate of micropollutants in the Vidy Bay of Lake Geneva, Switzerland. Part II: Micropollutant removal between wastewater and raw drinking water. Environmental Toxicology and Chemistry 29 (8), 1658e1668. Nissenson, P., Dabdub, D., Das, R., Maurino, V., Minero, C., Vione, D., 2010. Evidence of the water-cage effect on the 2þ photolysis of NO 3 and FeOH . Implications of this effect and of H2O2 surface accumulation on photochemistry at the airwater interface of atmospheric droplets. Atmospheric Environment 44, 4859e4866. Oulton, R.L., Kohn, T., Cwiertny, D.M., 2010. Pharmaceuticals and personal care products in effluent matrices: a survey of transformation and removal during wastewater treatment and implications for wastewater management. Journal of Environmental Monitoring 12 (11), 1956e1978. 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. Aquatic Sciences 65 (4), 342e351. Pailler, J.-Y., Krein, A., Pfister, L., Hoffmann, L., Guignard, C., 2009. Solid phase extraction coupled to liquid chromatographytandem mass spectrometry analysis of sulfonamides,
6736
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 2 5 e6 7 3 6
tetracyclines, analgesics and hormones in surface water and wastewater in Luxenbourg. Science of the Total Environment 407 (16), 4736e4743. Parij, N., Nagy, A.M., Neve, J., 1995. Linear and non linear competition plots in the deoxyribose assay for determination of rate constants for reaction of non steroidal antiinflammatory drugs with hydroxyl radicals. Free Radical Research 23 (6), 571e579. Peuravuori, J., Pihlaja, K., 2009. Phototransformation of selected pharmaceuticals under low-energy UVA-vis and powerful UVB-UVA irradiations in aqueous solutions e The role of natural dissolved organic chromophoric material. Analytical and Bioanalytical Chemistry 394 (6), 1621e1636. Pomati, F., Castiglioni, S., Zuccato, E., Fanelli, R., Vigetti, D., Rossetti, C., Calamari, D., 2006. Effects of a complex mixture of therapeutic drugs at environmental levels on human embryonic cells. Environmental Science and Technology 40 (7), 2442e2447. Prousek, J., 2007. Fenton chemistry in biology and medicine. Pure and Applied Chemistry 79 (12), 2325e2338. Quinn, B., Gagne, F., Blaise, C., 2008. The effects of pharmaceuticals on the regeneration of the cnidarian, Hydra attenuata. Science of the Total Environment 402 (1), 62e69. Quinn, B., Gagne, F., Blaise, C., 2009. Evaluation of the acute, chromic and teratogenic effects of a mixture of eleven pharmaceuticals on the cnidarian, Hydra attenuata. Science of the Total Environment 407 (3), 1072e1079. Rodgers, M.A.J., Snowden, P.T., 1982. Lifetime of 1O2 in liquid water as determined by time-resolved infrared luminescence measurements. Journal of the American Chemical Society 104 (20), 5541e5543. Santos, J.L., Aparicio, I., Callejo´n, L., Alonso, E., 2009. Occurrence of pharmaceutically active compounds during 1-year period in wastewaters from four wastewater treatment plants in Seville (Spain). Journal of Hazardous Materials 164 (2e3), 1509e1516. Szabo´, R.K., Megyeri, Cs., Ille´s, L., Gajda-Schrantz, K., Mazellier, P., Dombi, A., 2011. Phototransformation of ibuprofen and ketoprofen in aqueous solutions. Chemosphere 84, 1658e1663. Teijon, G., Candela, L., Tamoh, K., Molina-Diaz, A., FernandezAlba, A.R., 2010. Occurrence of emerging contaminants, priority substances (2008 105 CE) and heavy metals in treated wastewater and groundwater at Depurbaix facility (Barcelona, Spain). Science of the Total Environment 408 (17), 3584e3595. Tiehm, A., Schmidt, N., Stieber, M., Sacher, F., Wolf, L., Hoetzl, H., 2011. Biodegradation of pharmaceutical compounds and their occurrence in the Jordan Valley. Water Resources Management 25 (4), 1195e1203. Tixier, C., Singer, H.P., Oellers, S., Mu¨ller, S.R., 2003. Occurrence and fate of carbamazepine, clofibric acid, diclofenac, ibuprofen, ketoprofen, and naproxen in surface waters. Environmental Science and Technology 37 (6), 1061e1068. Valcarcel, Y., Alonso, S.G., Rodriguez-Gil, J.L., Maroto, R.R., Gil, A., Catala, M., 2011. Analysis of the presence of cardiovascular and analgesic/anti-inflammatory/antipyretic pharmaceuticals in river- and drinking-water of the Madrid Region in Spain. Chemosphere 82 (7), 1062e1071. Vione, D., Khanra, S., Cucu Man, S., Maddigapu, P.R., Das, R., Arsene, C., Olariu, R.I., Maurino, V., Minero, C., 2009a.
Inhibition vs. enhancement of the nitrate-induced phototransformation of organic substrates by the OH scavengers bicarbonate and carbonate. Water Research 43 (18), 4718e4728. Vione, D., Maurino, V., Minero, C., Carlotti, M.E., Chiron, S., Barbati, S., 2009b. Modelling the occurrence and reactivity of the carbonate radical in surface freshwater. Comptes Rendus Chimie 12 (8), 865e871. Vione, D., Khanra, S., Das, R., Minero, C., Maurino, V., Brigante, M. , Mailhot, G., 2010. Effect of dissolved organic compounds on the photodegradation of the herbicide MCPA in aqueous solution. Water Research 44 (20), 6053e6062. Waiser, M.J., Humphries, D., Tumber, V., Holm, J., 2011. Effluentdominated streams. Part 2: presence and possible effects of pharmaceuticals and personal case products in wascana Creek, Saskatchewan, Canada. Environmental Toxicology and Chemistry 30 (2), 508e519. Wang, C.A., Shi, H.L., Adams, C.D., Gamagedara, S., Stayton, I., Timmons, T., Ma, Y.F., 2011. Investigation of pharmaceuticals in Missouri natural and drinking water using high performance liquid chromatography e tandem mass spectrometry. Water Research 45 (4), 1818e1828. Wang, L., Ying, G.G., Zhao, J.L., Yang, X.B., Chen, F., Tao, R., Liu, S., Zhou, L.J., 2010. Occurrence and risk assessment of acidic pharmaceuticals in the yellow River, Hai River and Liao River of north China. Science of the Total Environment 408 (16), 3139e3147. Wilkinson, F., Brummer, J., 1981. Rate constants for the decay and reactions of the lowest electronically excited singlet-state of molecular oxygen in solution. Journal of Physical and Chemical Reference Data 10 (4), 809e1000. Yu, C.P., Chu, K.H., 2009. Occurrence of pharmaceuticals and personal care products along the West Prong little Pigeon River in east Tennessee, USA. Chemosphere 75 (10), 1281e1286. Yuan, F., Hu, C., Hu, X., Qu, J., Yang, M., 2009. Degradation of selected pharmaceuticals in aqueous solution with UV and UV/H2O2. Water Research 43 (6), 1766e1774. Zhang, X., Oakes, K.D., Ciu, S.F., Bragg, L., Servos, M.R., Pawliszyn, J., 2010. Tissue-specific in vivo bioconcentration of pharmaceuticals in rainbow trout (Oncorhynchus mykiss) using space-resolved solid-phase microextraction. Environmental Science and Technology 44 (9), 3417e3422. Zhao, J.L., Ying, G.G., Wang, L., Yang, J.F., Yang, X.B., Yang, L.H., Li, X., 2009. Determination of phenolic endocrine disrupting chemicals and acidic pharmaceuticals in surface waters of the Pearl Rivers in South China by gas chromatography-negative chemical ionization-mass spectrometry. Science of the Total Environment 407 (2), 962e974. Zhao, J.L., Ying, G.G., Liu, Y.S., Chen, F., Yang, J.F., Wang, L., Yang, X.B., Stauber, J.L., Warne, M.S., 2010. Occurrence and a screening-level risk assessment of human pharmaceuticals in the Pearl River system, South China. Environmental Toxicology and Chemistry 29 (6), 1377e1384. Zuccato, E., Castiglioni, S., Fanelli, R., 2005. Identification of the pharmaceuticals for human use contaminating the Italian aquatic environment. Journal of Hazardous Materials 122 (3), 205e209.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 3 7 e6 7 4 4
Available online at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Rejection of micropollutants by clean and fouled forward osmosis membrane Rodrigo Valladares Linares*, Victor Yangali-Quintanilla, Zhenyu Li, Gary Amy King Abdullah University of Science and Technology, KAUST, Water Desalination and Reuse Center, Al-Jazri Bldg (4) Of. 4255 ws07, Thuwal 23955-6900, Saudi Arabia
article info
abstract
Article history:
As forward osmosis (FO) gains attention as an efficient technology to improve wastewater
Received 30 July 2011
reclamation processes, it is fundamental to determine the influence of fouling in the
Received in revised form
rejection of emerging contaminants (micropollutants). This study focuses on the rejection
15 October 2011
of 13 selected micropollutants, spiked in a secondary wastewater effluent, by a FO
Accepted 15 October 2011
membrane, using Red Sea water as draw solution (DS), differentiating the effects on the
Available online 25 October 2011
rejection caused by a clean and fouled membrane. The resulting effluent was then desalinated at low pressure with a reverse osmosis (RO) membrane, to produce a high quality
Keywords:
permeate and determine the rejection with a coupled forward osmosis e low pressure
Forward osmosis
reverse osmosis (FOeLPRO) system. When considering only FO with a clean membrane, the
Fouling
rejection of the hydrophilic neutral compounds was between 48.6% and 84.7%, for the
Micropollutants
hydrophobic neutrals the rejection ranged from 40.0% to 87.5%, and for the ionic
Reverse osmosis
compounds the rejections were between 92.9% and 96.5%. With a fouled membrane, the rejections were between 44.6% and 95.2%, 48.7%e91.5% and 96.9%e98.6%, respectively. These results suggest that, except for the hydrophilic neutral compounds, the rejection of the micropollutants is increased by the presence of a fouling layer, possibly due to the higher hydrophilicity of the FO fouled membrane compared to the clean one, the increased adsorption capacity of hydrophilic compounds and reduced mass transport capacity, membrane swelling, and the higher negative charge of the membrane surface, related to the foulants composition, mainly NOM acids (carboxylic radicals) and polysaccharides or polysaccharide-like substances. However, when coupled with RO, the rejections in both cases increased above 96%. The coupled FOeLPRO system was an effective double barrier against the selected micropollutants. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
Organic micropollutants, or emerging organic contaminants, are substances that include, but are not limited to, pharmaceutically active compounds (PhACs), endocrine disrupting compounds (EDCs), disinfection by-products and pesticides, that are present in the environment and, due to the increasing
concentration detected in recent studies, are arising concern among researchers and regulatory agencies, because most of them are not yet regulated and their impacts on human life are not quite known (Snyder et al., 2003). There is an urging need to understand their partitioning, accumulation and removal from water, soil, air and biota; unfortunately, there is not a clear idea of the exact risks of
* Corresponding author. Tel.: þ966 2 808 4997. E-mail address:
[email protected] (R. Valladares Linares). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.10.037
6738
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 3 7 e6 7 4 4
chronic exposure to a mixture of organic micropollutants, including quantities ingested through drinking water, leaving a big question yet to be answered (Yangali-Quintanilla et al., 2010b). There are current technologies that can remove these organic micropollutants from wastewaters, but generally they are energy intensive processes, such as nanofiltration (NF) and reverse osmosis (RO) (Kimura et al., 2003; Bellona et al., 2004). Conventional wastewater treatment plants will not remove completely these chemicals, which will partition mainly into the sludge produced and the water effluent, generating a threat for the discharge site and the downstream areas (Ku¨mmerer et al., 2000; Snyder et al., 2003; Ternes et al., 2004). Findings indicate that fouling significantly impacts on membrane performance and varies the removal pattern of neutral trace organic contaminants (Agenson and Urase, 2007). Xu et al. found that membrane fouling facilitated the transport of hydrophobic and hydrophilic organic contaminants through a cellulose triacetate (CTA) RO membrane (Xu et al., 2006). For trace organic contaminants (hormones), colloidal fouling in an RO membrane (CTA) caused a significant increase in the permeate concentration of these substances (Ng and Elimelech, 2004). Three mechanisms have been identified to govern the influence of fouling in the rejection capacity of a NF membrane: modification of the membrane charge surface, pore restriction, and cake enhanced concentration polarization (Nghiem and Hawkes, 2007). Fouling can change the surface characteristics of the membrane, either to improve or degrade the rejection capability and flux of it. There are studies that can prove both situations for NF and RO membranes (Bellona et al., 2004; Xu et al., 2006; Nghiem and Coleman, 2008), but there is no literature on the precise effects of organic contaminants and their rejection by FO membranes. Forward osmosis (FO) coupled with RO membranes have been proposed in the last years to be an efficient double barrier system to reject most of emerging contaminants (Cornelissen et al., 2008; Cath et al., 2011). Cath et al. studied the removal of micropollutants by a FOeRO system using a membrane biological reactor (MBR) and the effluent from an activated sludge tank as feed waters for the spiral wound FO membrane, and synthetic seawater as draw solution (DS). They measured the concentration of the micropollutants present in the MBR and the effluent (no micropollutants were spiked in the water), varying from 2 to 400 ng/L (in some cases the compounds were below the limit of quantification). The rejections were mostly high with the FO membrane, except for Bisphenol A. When the hybrid process was considered, very high percentages of rejection (>99%) were achieved (Cath et al., 2011). This study focuses on the study of the difference in the rejection of a cocktail of 13 micropollutants that are known to occur in aquatic environments (Heberer, 2002) (5 hydrophilic nonionic, 3 hydrophobic nonionic and 4 hydrophilic ionic micropollutants) by a clean FO membrane (inorganic solution with ionic-strength used as feed) and a fouled FO membrane (secondary wastewater effluent (SWWE) used as feed). Significantly higher-than-normal concentrations of the micropollutants are used in the experiment to mainly account for membrane adsorption after steady-state saturation (Yangali-Quintanilla et al., 2009).
2.
Materials and methods
2.1.
FO membrane
Hydration Technology Innovations (HTI, Albany, OR) FO membranes were used for this study. The original coupons, measuring 400 600 , were cut into flat-sheet membranes to fit the custom-made plate and frame cell with an area of 202 cm2 for each side made out of poly (methyl methacrylate), forming a channel where the DS recirculates.
2.2.
RO membrane
The RO membrane used is an aromatic polyamide membrane produced for desalination of brackish water by Dow-Filmtec (Midland, MI, USA) under the name BW30. The RO system consists of a positive displacement pump (HydraCell, MN, USA), a crossflow filtration cell accommodating a 139 cm2 membrane (SEPA CF II, Sterlitech, Kent, WA, USA), needle valves, pressure gauges, a proportional pressure relief valve and stainless steel tubing (Swagelok BV, Netherlands).
2.3.
Source waters
For the feed water of the FO process, a SWWE was used; the effluent was collected from the Al-Ruwais wastewater treatment plant in Jeddah, Saudi Arabia, after sedimentation of the effluent of the activated sludge aeration tanks (secondary treatment). The SWWE has a pH of 7.2 and a conductivity of 3300 mS/cm. The biological oxygen demand (BOD5) of the wastewater effluent was 20 mg/L, and the DOC was 5 mg/L. The effluent received no pretreatment, except that it was kept at 4 C to prevent changes in the composition. Seawater from the Red Sea was taken from the feeding line of the desalination plant at KAUST, Thuwal, Saudi Arabia, and was used as DS for the FO process. The water has a pH of 7.8, the dissolved organic carbon (DOC) was determined to be 1 mg/L, the total dissolved solids (TDS) are 40.5 g/L, and the conductivity was 57500 mS/cm. The raw seawater was pretreated with a 0.45 mm filter. Table S1 (supporting information) shows the list of chemicals and the concentration used to create an inorganic synthetic solution (SS) to simulate the ionic-strength of a wastewater effluent, avoiding the presence of suspended solids and fouling agents. It was calculated based on the results of the analysis done to the wastewater effluent from Jeddah following the Standard Methods (Eaton et al., 2005) for common inorganic elements such as ammonium, nitrate, nitrite, total nitrogen and phosphate. Both feed waters and the DS were kept at 20 0.5 C during the FO experiments with a chilling/heating device. For the RO experiment to be compared to FOeLPRO, the RO feed water was prepared diluting prefiltered seawater with DI water, reaching the conductivity of the diluted draw solution (DDS) after the FO process. For the FOeLPRO, the same DDS coming from the FO process with SWWE as feed water is used in the LPRO system.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 3 7 e6 7 4 4
2.4.
Experimental setup and procedure
6739
The FO membrane cell was connected, via a gear pump (Coleparmer), to the DS container. A balance (TE6101, Sartorius AG, Germany) was used as flow recorder when connected to a computer; based on the difference in weight and the effective membrane area, the flux can be calculated. The feed water tank contained 18 L of SWWE and the DS tank held 500 ml of pretreated seawater. The recirculation flow was set at 50 mL/min. The recording of the weight data was activated immediately after the membrane was submerged in the feed tank; at the same time the osmotic dilution process started, because the tubing system and the cell were prefilled with DS before being in contact with the feed water. A stirring device was used to create turbulence in the feed water and achieve proper mixing. A schematic is presented in Fig. 1. The dilution experiment was performed for 24 h; the DS increased its volume due to continuous osmosis between the feed water and the DS recirculating in the cells. The dilution cycle was repeated for 5 days by replacing the fresh DS at the beginning of each new day. The same procedure was followed with a new membrane and the SS as feed water to test the capability of the clean membrane to reject the micropollutants (no fouling agents present in the feed water). The operating parameters for the RO system were set as follows: transmembrane pressure of 15 bars, permeate flow of 1.6 mL/min, concentrate flow of 80 mL/min and a RO flux of 7 L/m2-h. The recovery of the process was stabilized to 2%.
All of the compounds, except for the 17a-ethynilestradiol, were prepared in a stock solution of 10 mg/L with Milli-Q water. This organic micropollutant was diluted in 1% wt. ethanol solution, and then diluted into a concentration of 10 mg/L with Milli-Q water. The organic micropollutants were spiked into the SS, SWWE and RO feed from the stock solution prepared with a concentration of 10 mg/L. The target concentration in the FO and RO feed of each micropollutant independently was 10 mg/ L. Samples of the FO and RO feed water were taken before the experiment started, and control samples were taken before the addition of the micropollutants. For the comparison of rejections between the SWWE and the SS, the 2nd and 3rd day samples of DDS were mixed to form a composite. Afterward, rejection over time for the SWWE was tested between 2nd and 3rd day and 4th and 5th day composites. A sample of blank DI water was taken as control. The micropollutants in water samples were analyzed by Technologiezentrum Wasser, (TZW, Karlsruhe, Germany). The information about the procedures for analyses of micropollutants in water samples were referenced in a previous study (Yangali-Quintanilla et al., 2010a). The uncertainty of measurement was 20%, not determined for each compound individually sampled, but determined during method validation, for all of the compounds listed in Table 1. Limits of quantification for each compound can be found in the Table S2; limits of detection usually were three-fold lower.
2.5.
2.6.
Micropollutants stock preparation and analyses
All the micropollutants were purchased from Sigma Aldrich (Munich, Germany). The list of micropollutants used to prepare the stock solutions is presented in Table 1. Compounds were classified according to their speciation in water as hydrophobic when the logarithm of the distribution coefficient (log D), which refers to the ratio of the sum of the concentrations of all forms of the compound in each of the two immiscible phases forming a mixture, generally water and octanol, is higher than 2.6, and hydrophilic when log D 2.6. Ionic compounds shown in the table are negatively charged at pH 7, which was determined by using ADME/Tox Web Software. Physicochemical properties were calculated using Molecular Modeling Pro.
FO membrane characterization
An Anton Paar Zeta Potential Analyzer (Austria) was used to determine the zeta potential (ZP) of the FO membrane. It uses a clamping cell where two pieces of membrane are used to create a channel of 25 mm of length and 5 mm width, with the active layers facing each other, and then the charge of the membrane in mVolts is measured when an electrolyte flows through it. For this analysis, two electrolytes were used, 10 mM KCl and the SWWE previously used as feed water. The ZP is measured in the pH range in which the membrane can operate (4e8), so the proper injection of acid (0.1 M HCl) or base (0.1 M NaOH) is added in the titration process. The results presented in this study were calculated with the HelmholtzeSmoluchowski equation.
Fig. 1 e FOeRO system layout. TC [ temperature controller, GP [ gear pump, CP [ conductivity probe, FR [ flow recorder, PDP [ positive displacement pump.
6740
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 3 7 e6 7 4 4
Table 1 e Properties of the micropollutants spiked to the feed water. Name
ID
1,4-dioxane Acetaminophen Metronidazole Phenazone Caffeine Bisphenol A Carbamazepine 17a-ethynilestradiol Ibuprofen Naproxen Fenoprofen Gemfibrozil Ketoprofen
DIX ACT MTR PHZ CFN BPA CBM EE2 IBF NPX FNP GFB KTP
MW (g/mol) log Da (pH 7) Molec. length (nm)b Width (nm)b,c Depth (nm)b Equiv. width (nm)b 88 151 171 188 194 228 236 296 206 230 242 250 254
0.17 0.23 0.27 0.54 0.45 3.86 2.58 3.98 3.97 0.34 0.38 2.3 0.13
0.71 1.14 0.93 1.17 0.98 1.25 1.20 1.48 1.39 1.37 1.16 1.58 1.16
0.66 0.68 0.9 0.78 0.87 0.83 0.92 0.87 0.73 0.78 0.93 0.94 0.92
0.52 0.41 0.48 0.56 0.56 0.75 0.58 0.84 0.55 0.75 0.74 0.65 0.74
0.59 0.53 0.66 0.66 0.70 0.79 0.73 0.85 0.64 0.76 0.83 0.78 0.83
a ADME/Tox Web Software. b Molecular Modeling Pro. c equivalent width ¼(width depth)0.5.
The contact angles of clean and fouled FO membranes were measured with a goniometer CAM200 (KSV, Finland) by using the sessile drop method. The fouled membrane samples were dried for 24 h at room temperature (20 C) before measurements. Confocal laser scanning microscopy (CLSM) (LSM710 uplight confocal microscope, Zeiss, Germany) was used to compare the surface of the FO membrane when different feed waters were used in the FO process.
3.
Results and discussion
3.1.
Zeta potential and contact angle
The ZP of the new membrane (virgin membrane) and the fouled membrane (after being used in the 5-day experiment with SWWE as feed water) was determined to account for the membrane surface charge. Fig. 2 shows the graph for the ZP in mVolts for each value of pH, using different electrolytes. The new membrane has a negative charge for the experiments carried out in this study, because the pH of the SWWE is 7.6, compared to the isoelectric point of the new membrane, which is below pH 3 (using the SWWE as electrolyte). Table S6 (supporting information) shows the results for the ZP and the
standard deviation of the calculated voltage for each electrolyte used in the new membrane and for the fouled membrane with SWWE as electrolyte. The fouling layer after the FO process increased the negative charge of the membrane. When the membrane was in contact with the wastewater effluent, the ZP was negative, which means that the ionic micropollutants were rejected due to electrostatic repulsion. This rejection was higher when the membrane was fouled; the charge became more negative due to the presence of negatively charged radicals in the foulants, mainly coming from the NOM acids (carboxylic radicals) and polysaccharides, as described in the literature (Cho et al., 1998; Fan et al., 2001; Shim et al., 2002). There was also an increase in the hydrophilicity of the fouled membrane, based on the measurements of the contact angle of the FO membrane in both conditions, being 58.8 0.3 and 49 3 for the clean and fouled membrane, respectively. The foulants increased the adsorption capacity of the membrane for hydrophilic compounds, and thus, the rejection of hydrophobic neutral micropollutants was higher when the membrane was fouled.
3.2.
Rejection of micropollutants by FO
Table S4 and Table S5 in the supporting information show the results for the initial concentration of micropollutants in the spiked SWWE/SS, the final concentration of contaminants in the DDS, rejection percentage and deviation errors. Similarly, the results for the RO process, and the hybrid FOeLPRO process are shown in Table S6. Figure S1 shows the flux decline curves when SS and SWWE are used as feed. No flux decline was reported for the former, but a 20% flux decline was observed for the latter. Rejections of the FO and RO membranes were calculated with Equation (1). Rejection ¼ Fig. 2 e Zeta Potential of the FO membrane for varying pH values and electrolytes.
1
C C0
100
(1)
For FO rejection when the feed water (SWWE or SS) was spiked with the micropollutants, C0 is the concentration of the
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 3 7 e6 7 4 4
Hydrophilic neutral
Hydrophobic neutral
Hydrophilic ionic
100
Rejection (%)
80 60 40
20 0 (MW)
DIX (88)
ACT MTR PHZ CFN BPA CBM EE2 IBF NPX FNP GFB KTP (151) (171) (188) (194) (228) (236) (296) (206) (230) (242) (250) (254) FO clean
FO fouled
Fig. 3 e Rejection of micropollutants by a clean and fouled FO membrane.
feed water, and C is the concentration of the DDS. For the RO rejection, the C0 is the concentration of the RO feed water, and the C is the concentration of the RO permeate. For the FOeLPRO process, the initial FO feed water concentration is C0 and C is the concentration of the RO permeate. Fig. 3 shows the comparison for the rejection of the 13 micropollutants by a clean FO membrane and a fouled FO membrane for the composite samples of the 2nd and 3rd day of experiments. Fig. 4 illustrates and summarizes the rejection mechanisms of the FO membrane. It can be noticed that for the hydrophilic neutral compounds the rejection with the SS as feed water (clean membrane) is higher than the rejection when SWWE is used as feed (fouled membrane), ranging from 2% difference for Dioxane to 6% difference for Phenazone, except for Caffeine. Two hypothesis might explain these results: i) the increase in surface charge might potentially
6741
result in a higher molecular weight cut-off (MWCO) due to membrane swelling, as described by Xu et al. (2006) for NF and RO membranes, reducing the rejection of the hydrophilic neutral compounds, and ii) higher hydrophilicity generated by the fouling layer on the membrane may allow a higher amount of micropollutants to partition through the membrane, and eventually, decrease the rejection (shown in Fig. 4a and b). This hypothesis is more significant than the first one, because it is based independently from the type of membrane used and focuses on the fouling layer instead, which is certainly occurring. In the case of Caffeine, the rejection for the fouled membrane increases 15% compared to the clean membrane; this can be explained by the high hydrophilicity of the compound (log D e 0.45), being absorbed by the cake layer formed in the membrane (mainly hydrophilic) and thus, preventing the compound from partitioning through the membrane into the DS, forming a double-barrierlike system. Along with this hypothesis, the MWCO of the membrane (z200 Da) is really close to the MW of this compound, and the foulants can be blocking the pores of the membrane, increasing the rejection of Caffeine with the fouled membrane. For the hydrophobic neutrals, the increase in rejections goes from 8.7% for Bisphenol A, to a 9.8% for Carbamazepine. In the case of the ionic contaminants, the increase goes from 0.4% for Naproxen, to 6% for Ibuprofen when the membrane is fouled. The increment in the rejection of hydrophobic neutral compounds is due to the higher hydrophilicity of the membrane when the fouling cake layer is present; this phenomenon is also associated with an increased adsorption capacity (for hydrophilic compounds) and reduced mass transport capacity (diffusion and partitioning) of hydrophobic compounds through the fouled membrane (Fig. 4c and d). For
Fig. 4 e Rejection mechanisms of different types of micropollutants by a clean and fouled FO membrane. a) and b) For the hydrophilic neutral substances, the fouled membrane is able to reject less efficiently the contaminants due to membrane swelling. c) and d) For the hydrophobic compounds, rejection by the fouled membrane is increased due to reduced mass transport capacity of hydrophobic compounds. e) and f) For the hydrophilic ionic compounds, the fouling layer increases the negative charge of the membrane, increasing the rejection of these compounds.
6742
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 3 7 e6 7 4 4
Fig. 5 e CLSM images of a FO membrane. a) Membrane used with SS as feed; b) membrane used with SWWE as feed. For the FO membrane used with the inorganic solution (SS), no fouling can be seen on the surface of the membrane; however, when the membrane is used with SWWE as feed, foulants accumulate on the surface.
process were lower. Nevertheless, when the coupled process was considered, the results of this work and that of Cath et al. were similar, achieving very high percentages of rejection (>99%). Fig. 7 shows the rejection of micropollutants for a composite sample of the 2nd and 3rd day DDS of the FO process with SWWE as feed, and the rejection percentage for a composite sample of the 4th and 5th day DDS with the same feed water. Fig. 7a shows the hydrophilic neutral pollutants; note that for all of them, excluding Acetaminophen, the rejection decreases over time due to the formation of the fouling layer, which increases the hydrophilicity of the surface and promotes partitioning of these contaminants into the DS side. For Fig. 7b, rejection of hydrophobic neutrals decreases over time, even for Carbamazepine, which can have the same decreasing tendency once the experimental error is considered. The cake layer formed in the surface of the membrane increases the effect of external concentration polarization (ECP), and even though the rejection of the 4th and 5th day composite is still higher than the rejection of the clean membrane (SS as feed water), ECP plays an important role in the reduction of rejection over time for the hydrophobic neutral compounds. Fig. 7c shows an increase in the rejection for hydrophilic ionic micropollutants over time,
Hydrophilic neutral
Hydrophobic neutral
Hydrophilic ionic
100
80 Rejection (%)
the ionic compounds, the negative charge of the membrane is greater when fouled (described in Section 3.1), increasing eventually the electrostatic repulsion (Donnan exclusion) between the negative charge of the membrane surface (Fig. 4e and f) as mentioned by Verliefde et al. (2008) and Xu et al. (2006) and the negative charge of the compound at pH 7.3. Nevertheless, the rejection is still low for hydrophobic neutral compounds, being Bisphenol A again the lowest with 48.7% rejection with the fouling layer. Fig. 5 shows images of the active layer of a FO membrane in different feed waters. Fig. 5a corresponds to a CLSM image of the FO membrane used with SS as feed water, where the support mesh can be clearly distinguished and no fouling is present on the surface of the membrane. Nevertheless, when SWWE is used as feed for the FO process, the foulants start to accumulate on the surface of the membrane, forming a cake layer clearly identifiable in the CLSM image on Fig. 5b, covering almost completely the support mesh. Consequently, rejection hypothesis based on the fouling layer formation are encouraged in this study. Fig. 6 compares the rejection percentages for the RO process and the coupled FOeRO process. For the RO process alone, the rejection of most compounds is higher than 99%, except for the two smallest micropollutants, Dioxane (67.9%) and Acetaminophen (89.4%), which have a MW lower or equal than the MWCO of the RO membrane, which can be estimated as 150Da. When the RO processes is coupled to the FO, the results for rejection can go as high as 89.1% for Dioxane, 96.3% for Acetaminophen, 98.9% for Metronidazole and for the rest of the compounds the rejection was higher than 99%. These results can be compared with the results presented by Cath et al. (2011) described in Section 1. In the case of the experiments mentioned in the present study, the concentrations of the spiked MPs were in the range of 1e10 mg/L, at least 2 orders of magnitude higher than those used by Cath et al. and there was no crossflow velocity on the feed side, which means a higher concentration polarization effect, leading to the accumulation of micropollutants in the membrane surface; thereby, the rejections achieved from the single FO
60 40 20 0
(MW)
DIX (88)
ACT MTR PHZ CFN BPA CBM EE2 IBF NPX FNP GFB KTP (151) (171) (188) (194) (228) (236) (296) (206) (230) (242) (250) (254)
FO-LPRO
LPRO
Fig. 6 e Rejection of micropollutants by an RO membrane and a coupled FOeLPRO system.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 3 7 e6 7 4 4
a 100.00 Rejection (%)
80.00 DIX (88) 60.00
ACT (151) MTR (171)
40.00
PHZ (188) CFN (194)
20.00 0.00 Day 2-3
Day 4-5
b 100.00 Rejection (%)
80.00 60.00
6743
caused by membrane swelling which results in a higher MWCO, and the increased adsorption capacity for hydrophilic compounds. In real conditions of water reuse applications, using SWWE as feed and seawater as DS, FO membranes were able to reject most of the micropollutants; rejections were moderate for hydrophilic neutral compounds (44e95%), moderate for hydrophobic neutral contaminants (48e92%), and high for the hydrophilic ionic micropollutants (96e99%). FO coupled with LPRO was effective rejecting low molecular weight hydrophilic neutral micropollutants, with rejections that went beyond 89.1%. For the rest of the compounds, rejections were greater than 99%. Therefore, the coupled FOeLPRO system proves to be an effective double barrier against the 13 micropollutants tested.
BPA (228) CBM (236)
40.00
EE2 (296)
Acknowledgments
20.00 0.00 Day 2-3
Day 4-5
c 100.00 Rejection (%)
99.00 98.00
IBF (206)
97.00
NPX (230)
96.00
FNP (242)
The authors appreciate the help of Hydration Technology Innovations and Dow-Filmtec for providing the FO and RO membranes, respectively. The help and assistance of the WDRC group is greatly acknowledged. The authors thank Dr. Frank Sacher from TZW (Technologiezentrum Wasser, Karlsruhe), who, along with his team, analyzed the concentration of micropollutants in the water samples for all of the experiments realized in this study.
GFB (250)
95.00
KTP (254)
Appendix. Supplementary material
94.00 93.00 Day 2-3
Day 4-5
Fig. 7 e Rejection of micropollutants on the 2nd and 3rd day and rejection on the 4th and 5th day of the FO process. a) Hydrophilic neutrals, b) hydrophobic neutrals, and c) hydrophilic ionic.
which was expected due to the formation of the fouling layer in the membrane and the change in the surface charge over time.
4.
Conclusions
Hydrophilic ionic compounds were rejected more effectively when the FO membrane was fouled due to the higher negative charge of the fouled membrane, due to the presence of negatively charged radicals in the foulants. Hydrophobic neutral compounds had higher rejections with the fouled membrane due to the higher hydrophilicity induced by the fouling layer, and the increase in adsorption capacity of hydrophilic compounds reducing passage of hydrophobic compounds. Rejection of hydrophilic neutral compounds decreased an average of 5% with the fouled membrane,
Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.watres.2011.10.037.
references
Agenson, K.O., Urase, T., 2007. Change in membrane performance due to organic fouling in nanofiltration (NF)/reverse osmosis (RO) applications. Separation and Purification Technology 55, 147e156. Bellona, C., Drewes, J.E., Xu, P., Amy, G., 2004. Factors affecting the rejection of organic solutes during NF/RO treatmentea literature review. Water Research 38, 2795e2809. Cath, T., Hancock, N., Xu, P., Heil, D., 2011. A comprehensive study of micropollutants rejection by forward osmosis and hybrid FOeRO. In: 2012 Membrane Technology Conference American Water Works Association (Glendale, Arizona). Cho, J., Amy, G., Pellegrino, J., Yoon, Y., 1998. Characterization of clean and natural organic matter (NOM) fouled NF and UF membranes, and foulants characterization. Desalination 118, 101e108. Cornelissen, E.R., Harmsen, D., de Korte, K.F., Ruiken, C.J., Qin, J.-J., Oo, H., Wessels, L.P., 2008. Membrane fouling and process performance of forward osmosis membranes on activated sludge. Journal of Membrane Science 319, 158e168. Eaton, A.D., Clesceri, L.S., Rice, E.W., Greenberg, A.E., 2005. Standard Methods for the Examination of Water and Wastewater, 21st. edition (New York).
6744
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 3 7 e6 7 4 4
Fan, L., Harris, J.L., Roddick, F.A., Booker, N.A., 2001. Influence of the characteristics of natural organic matter on the fouling of microfiltration membranes. Water Research 35, 4455e4463. Heberer, T., 2002. Occurrence, fate, and removal of pharmaceutical residues in the aquatic environment: a review of recent research data. Toxicology Letters 131, 5e17. Kimura, K., Amy, G., Drewes, J.E., Heberer, T., Kim, T.-U., Watanabe, Y., 2003. Rejection of organic micropollutants (disinfection by-products, endocrine disrupting compounds, and pharmaceutically active compounds) by NF/RO membranes. Journal of Membrane Science 227, 113e121. Ku¨mmerer, K., Al-Ahmad, A., Mersch-Sundermann, V., 2000. Biodegradability of some antibiotics, elimination of the genotoxicity and affection of wastewater bacteria in a simple test. Chemosphere 40, 701e710. Ng, H.Y., Elimelech, M., 2004. Influence of colloidal fouling on rejection of trace organic contaminants by reverse osmosis. Journal of Membrane Science 244, 215e226. Nghiem, L.D., Coleman, P.J., 2008. NF/RO filtration of the hydrophobic ionogenic compound triclosan: transport mechanisms and the influence of membrane fouling. Separation and Purification Technology 62, 709e716. Nghiem, L.D., Hawkes, S., 2007. Effects of membrane fouling on the nanofiltration of pharmaceutically active compounds (PhACs): mechanisms and role of membrane pore size. Separation and Purification Technology 57, 176e184. Shim, Y., Lee, H.-J., Lee, S., Moon, S.-H., Cho, J., 2002. Effects of natural organic matter and ionic species on membrane surface charge. Environmental Science & Technology 36, 3864e3871.
Snyder, S.A., Westerhoff, P., Yoon, Y., Sedlak, D.L., 2003. Pharmaceuticals, personal care products, and endocrine disruptors in water: Implications for the water industry. Environmental Engineering Science 20, 21. Ternes, T.A., Joss, A., Siegrist, H., 2004. Scrutinizing pharmaceuticals and personal care products in wastewater treatment. Environmental Science & Technology 38, 392ae399a. Verliefde, A.R.D., Cornelissen, E.R., Heijman, S.G.J., Verberk, J.Q.J. C., Amy, G.L., Van der Bruggen, B., van Dijk, J.C., 2008. The role of electrostatic interactions on the rejection of organic solutes in aqueous solutions with nanofiltration. Journal of Membrane Science 322, 52e66. Xu, P., Drewes, J.E., Kim, T.-U., Bellona, C., Amy, G., 2006. Effect of membrane fouling on transport of organic contaminants in NF/RO membrane applications. Journal of Membrane Science 279, 165e175. Yangali-Quintanilla, V., Maeng, S.K., Fujioka, T., Kennedy, M., Amy, G., 2010a. Proposing nanofiltration as acceptable barrier for organic contaminants in water reuse. Journal of Membrane Science 362, 334e345. Yangali-Quintanilla, V., Sadmani, A., McConville, M., Kennedy, M., Amy, G., 2009. Rejection of pharmaceutically active compounds and endocrine disrupting compounds by clean and fouled nanofiltration membranes. Water Research 43, 2349e2362. Yangali-Quintanilla, V., Sadmani, A., McConville, M., Kennedy, M., Amy, G., 2010b. A QSAR model for predicting rejection of emerging contaminants (pharmaceuticals, endocrine disruptors) by nanofiltration membranes. Water Research 44, 373e384.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 4 5 e6 7 5 2
Available online at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
2-Fluorophenol degradation by aerobic granular sludge in a sequencing batch reactor Anouk F. Duque a, Vaˆnia S. Bessa a, Maria F. Carvalho a, Merle K. de Kreuk b,c, Mark C.M. van Loosdrecht b, Paula M.L. Castro a,* a
CBQF/Escola Superior de Biotecnologia, Universidade Cato´lica Portuguesa, Rua Dr. Anto´nio Bernardino de Almeida, 4200-072 Porto, Portugal b Department of Biotechnology, Delft University of Technology, Julianalaan 67, 2628BC Delft, The Netherlands c Waterboard Hollandse Delta, Handelsweg 100, 2988DC Ridderkerk, The Netherlands
article info
abstract
Article history:
Aerobic granular sludge is extremely promising for the treatment of effluents containing
Received 6 June 2011
toxic compounds, and it can economically compete with conventional activated sludge
Received in revised form
systems. A laboratory scale granular sequencing batch reactor (SBR) was established and
22 September 2011
operated during 444 days for the treatment of an aqueous stream containing a toxic
Accepted 15 October 2011
compound, 2-fluorophenol (2-FP), in successive phases. Initially during ca. 3 months, the
Available online 25 October 2011
SBR was intermittently fed with 0.22 mM of 2-FP added to an acetate containing medium. No biodegradation of the target compound was observed. Bioaugmentation with
Keywords:
a specialized bacterial strain able to degrade 2-FP was subsequently performed. The reactor
Aerobic granular sludge (GS)
was thereafter continuously fed with 0.22 and 0.44 mM of 2-FP and with 5.9 mM of acetate
2-Fluorophenol (2-FP)
(used as co-substrate), for 15 months. Full degradation of the compound was reached with
Bioaugmentation
a stoichiometric fluoride release. The 2-FP degrading strain was successfully retained by
Sequencing batch reactor (SBR)
aerobic granules, as shown through the recovering of the strain from the granular sludge at
Wastewater treatment
the end of the experiment. Overall, the granular SBR has shown to be robust, exhibiting a high performance after bioaugmentation with the 2-FP degrading strain. This study corroborates the fact that bioaugmentation is often needed in cases where biodegradation of highly recalcitrant compounds is targeted. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
The use of organofluorine compounds as aerosol propellants, surfactants, agrochemicals, adhesives, refrigerants, fire retardants, pharmaceuticals, among others, has increased during the last century (Key et al., 1997). They may have significant biological effects as enzyme inhibitors, modifiers of cellecell communication, disrupting membrane transport and processes for energy generation. Biodegradation of fluoroaromatic compounds,
among them fluorophenols, has been scarcely investigated. They are usually biodegraded via (halo)catechols (Haggblom, 1992; Murphy et al., 2009). The conversion of these compounds is often the rate limiting step and therefore they easily accumulate in reactors (Carvalho et al., 2006a; Fava et al., 1995). Micropollutants need to be removed efficiently from wastewater to protect receiving waterbodies from their ecotoxicological effects. The removal of (halo)aromatics from wastewaters using biological technologies has been shown before (Buitron et al., 2005;
* Corresponding author. Tel.: þ351 22 5580059; fax: þ351 22 5090351. E-mail addresses:
[email protected] (A.F. Duque),
[email protected] (V.S. Bessa),
[email protected] (M.F. Carvalho),
[email protected] (M.K. de Kreuk),
[email protected] (M.C.M. van Loosdrecht),
[email protected],
[email protected] (P.M.L. Castro). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.10.033
6746
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 4 5 e6 7 5 2
Carvalho et al., 2006b; Osuna et al., 2008). As these pollutants occur in wastewaters discontinuously and at low concentrations, maintenance of a good population of (halo)aromatics degraders in bioreactors is highly desirable. Bioreactor systems with high biomass retention are extremely promising for the treatment of wastewaters containing toxic compounds. The granular sludge SBR is such a reactor type. The granular sludge SBR is a relative novel design, which has especially gained interest after the recent observation that under aerobic conditions biomass can be grown in granules similar to anaerobic granular sludge reactors (Beun et al., 1999, 2000, 2002; Morgenroth et al., 1997). Aerobic granular sludge presents several advantages over activated sludge, such as excellent settling properties, high biomass retention and biosorption, ability to deal with high organic loading rates and to perform simultaneously diverse biological processes, such as COD, N and P removal (De Bruin et al., 2004; De Kreuk et al., 2005; Xu et al., 2004). Aerobic granular sludge can economically compete with the conventional activated sludge systems. Recently, Carucci et al. (2010) compared a granular sludge sequencing batch reactor (GSBR) with a conventional sequencing batch reactor and a membrane bioreactor for the treatment of 4-chlorophenol. They reported the GSBR as the most suitable technology when the critical parameters are low land requirement, system simplicity/flexibility and short start up times. Furthermore, it has been reported in literature that aerobic granular sludge can successfully degrade and deal with the presence of phenol and chlorinated compounds, suggesting that the granular structure protects the microorganisms against toxicity (Carucci et al., 2008, 2009, 2010; Jiang et al., 2002; Tay et al., 2005a, b; Wang et al., 2007). Nevertheless, if unacclimated sludge is not able to readily degrade a toxic or a poorly degradable compound, bioaugmentation with specialized strains can be an option, a strategy that is lately gaining more interest (Quan et al., 2003; Rittmann and Whiteman, 1994; Yu and Mohn, 2001). There are two main problems associated with the augmentation of a bioreactor: (i) as toxic and/or recalcitrant (micro)pollutants may appear intermittently and/ or may be present at very low concentrations, loss of degrading capacity by the specialized strain may occur, due to the absence of the selective pressure (ii) the presence of protozoa can affect the success of the bioaugmentation, as the added culture can be grazed by these microorganisms. Nevertheless, Quan et al. (2003) has shown the success of bioaugmenting a flocculated sludge SBR with an immobilized mixed culture able to biodegrade 2,4-dichlorophenol. The main aim of this study was to investigate the robustness and performance of a laboratory scale SBR with aerobic granular sludge toward shock loadings of 2-FP, using acetate as the growth substrate, before and after bioaugmentation with a specialized strain able to degrade this compound.
2.
Material and methods
2.1.
SBR set up and operation
A 2.5 L SBR with 110 cm height and an internal diameter of 6.5 cm was established. The experimental set up is
schematically shown in Fig. 1. The system was operated in cycles using an automatic timer (Siemens Logo! 230RC) to start and stop pumps for influent, aeration (4 L min1; superficial air velocity of 84.8 m h1) and effluent withdrawal. The operating conditions tested in the SBR are described in Table 1. Dissolved oxygen (DO) and pH were measured online. DO was measured as percentage of the oxygen saturation concentration. The oxygen saturation level was monitored, but not controlled during the cycle. The pH was maintained at 7.0 0.8 by dosing 1 M NaOH or 1 M HCl. The reactor was operated in successive cycles of 3 h (during phases I and II), consisting of 60 min influent feeding (which was introduced in the bottom of the reactor), 112 min aeration, 3 min settling and 5 min effluent withdrawal. During phases III and IV, the aeration period was increased to 652 min in the 12 h cycle and then decreased to 412 min in the 8 h cycle (phase V) and, afterward, to 172 min in the 4 h cycle (phases VI and VII), so that the cycle length would not represent a limitation for all bioconversion processes (Table 1). In each cycle, about 40% of the reactor liquid was withdrawn. The settling time was chosen such that only particles with a settling velocity larger than 6 m h1 were effectively retained in the reactor. The reactor was operated at a sludge retention time (SRT) of 30 days. Aerobic granular sludge (500 ml wet granules) was collected from a pilot plant treating sewage, in the Netherlands. This biological phosphate removing sludge was used as inoculum for the start up of the reactor.
2.1.1.
Bioaugmentation with Rhodococcus sp. strain FP1
For bioaugmentation of the SBR, a bacterial strain able to degrade 2-FP, named as Rhodococcus sp. strain FP1, previously isolated in our laboratories, was used (HM210775) (unpublished). FP1 pure cultures were grown in sealed flasks containing a mineral salts liquid medium (Caldeira et al., 1999) and 2-FP at a concentration of 50 mg L1. The cultures were incubated on an orbital shaker (100 rpm) at 25 C. The optical density at 600 nm (OD600) was followed to monitor growth. The reactor was inoculated with 1.25 L of an FP1 pure culture with an OD600 of 0.8.
2.2.
Media
The composition of the SBR influent media was as described by De Kreuk et al. (2005). During shock loadings, 2.38 mM (phases II, III and VII) and 4.76 mM (phases IVeVI) of 2-FP was added to the influent medium. From each media, 89 ml per cycle were dosed together with 772 ml of tap water.
2.3.
Analysis
The DO concentration in the reactor was measured with a DOsensor (InPro 6820, Mettler-Toledo) and the pH was monitored using a pH-electrode (InPro 3030, Mettler-Toledo). Chemical oxygen demand (COD) was determined according to Standard Method 5220 (APHA, 1998). The concentration of fluoride ions in supernatants was measured with an ion-selective combination electrode (model CH-8902, Mettler-Toledo GmbH, Urdorf, Switzerland), which was calibrated with NaF (0.01e5 mM) in mineral salts medium
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 4 5 e6 7 5 2
6747
Fig. 1 e Schematic representation of the SBR.
(SBR influent medium without the carbon sources). Biomass was previously removed from culture samples by centrifugation at 7000 rpm for 10 min. The ionic strength of the standards and of the samples was adjusted with a buffer solution, named total ionic strength adjustment solution (TISAB). The composition of the TISAB solution was NaCl 1 M, CH3COOH 0.25 M, NaCH3COO 0.75 M and sodium citrate 0.002 M. 2-FP was analyzed by high performance liquid chromatography (HPLC), on a System Gold 126 (Beckman Coulter, Fullerton, USA) with a LiChroCART 25-4 LiChrospher 100 RP18 reversed-phase column, 5 mm particle size (Merck, Darmstadt, Germany). The samples were filtered through a 0.45 mm filter prior to HPLC analysis. The mobile phase consisted of 50% (v/v) acetonitrile and water and was used with a flow rate of 0.8 ml min1. The run time was 10 min (elution time about 5.1 min) and the volume injected was 20 ml. The compound was detected at 220 nm using a diode array detector.
2.4. Analysis of the presence of Rhodococcus sp. strain FP1 in the SBR 2.4.1.
Bacterial isolation and DNA extraction
In order to obtain a representative sample of the population present in the SBR, a sample of aerobic granular sludge was taken from the reactor during the aeration phase. The granules were then crushed using a sterile potter and pestle. Serial dilutions in saline solution (0.85% w/v NaCl) were made and 0.1 ml of each dilution was spread onto nutrient agar (NA) (LABM, UK). Plates were incubated at 25 C for 3 days. Based on size, morphology and pigmentation, different bacterial colonies were isolated from NA plates using the streak-plate procedure. Genomic DNA from each isolate was extracted by picking a colony with a sterile loop, suspending the cells in 200 mL sterile ultrapure water and incubating the suspension for 15 min at 95 C. Samples were then kept in ice for 7 min and vortexed. Subsequently, samples were centrifuged at
6748
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 4 5 e6 7 5 2
Table 1 e Operating conditions tested in the SBR. Phase
I IIc
Length of operation (days)
Days of operation
0e99 100e209
Cycle time (h)
Acetate
2-FP
3 3
5.9 5.9
0 0.22
12 12 8 4 4
5.9 5.9 5.9 5.9 5.9
0.22 0.44 0.44 0.44 0.22
99 109
Bioaugmentation with Rhodococcus sp. strain FP1 III 210e222 12 IV 223e229 6 V 230e266 36 VI 267e400 133 VII 401e444 43
Inlet carbon sources concentrations (mM)
HRTa (h)
OLRb (kg m3 d1) Acetate
2-FP
7.9 7.9
1.06 1.06
e 0.075
31.6 31.6 21.1 10.5 10.5
0.26 0.26 0.40 0.79 0.79
0.019 0.037 0.056 0.112 0.056
a HRT e hydraulic residence time. b OLR e organic loading rate. c Organic shock loadings with 2-FP applied 1 cycle/2 days.
14,000 rpm for 5 min and the supernatant was transferred to a new sterile microtube. DNA was stored at 20 C.
2.4.2.
DNA sequencing analysis
Isolates were subsequently identified by 16S rRNA sequencing analysis. The amplification was carried out with the universal primers f27 and r1492 (Lane, 1991) under standard polymerase chain reaction (PCR) conditions (Rainey et al., 1996), with Taq polymerase from Promega (Madison, WI). The amplified fragments were sequenced by Macrogen Inc. (Seoul, Republic of Korea). To determine the phylogenetic affiliation, similarity analysis was performed using the BLAST program (Altschul et al., 1997).
2.5. Fluoride adsorption tests performed on granular sludge In order to estimate the amount of fluoride adsorbed to the aerobic granules, 20 ml of aerobic granular sludge was taken from the bioreactor at the end of a cycle fed with 2-FP and was added to 500 ml flasks containing 180 ml of SBR medium without 2-FP. The flasks were incubated in an orbital shaker at 25 C, 150 rpm. Samples for fluoride analysis were taken periodically during 1 month.
3.
Results and discussion
3.1.
SBR performance
A SBR was operated in order to assess its performance when treating a synthetic wastewater containing 2-FP. Several operating scenarios, divided in 7 different phases, were tested (Table 1).
3.1.1.
Before bioaugmentation
The reactor was inoculated with aerobic granular sludge obtained from a pilot SBR operating in the Netherlands and was operated without oxygen control, which led to dissolved oxygen concentrations during the aeration phase close to saturation. At the SBR start-up the granules were smooth, with a regular shape and dark (Fig. 2a). After 2e3 months of
operation (phase I) the granules became denser and with irregular shape (Fig. 2b and c) and after 6 months (phase II) they were larger and with “cauliflower” shape (Fig. 2d). At the end of the experiment (after 1.5 year, phase VII), the granules became very dense and small, looking like sand (Fig. 2e). Tay et al. (2005b) reported that the compact structure of the granules protects them against toxic compounds, minimizing sludge wash out, which could well explain the evolution observed here for the granule morphology. The overall performance of the SBR after 2-FP feeding is shown in Fig. 3. During phase I (99 days), acetate was fed as the sole carbon and energy source with the objective of achieving stable granules. In phase II, the SBR was exposed to intermittent organic feeding with 0.22 mM of 2-FP (1 cycle/2 days) with an HRT of 7.9 h, in order to evaluate the capacity of the endogenous granule population to biodegrade 2-FP. By operating the SBR at a volume exchange ratio of 40%, the 2-FP concentration inside the bioreactor was diluted to a concentration of 0.09 mM. These intermittent 2-FP organic shocks were applied along a period of 109 days. As it can be observed from Fig. 3, 2-FP was not degraded during phase II, with no fluoride release observed and with 2-FP being detected at the outlet, indicating that there was no acclimatization of the biomass to the toxic compound. It has been described in the literature that acetate-fed aerobic granules can easily degrade other recalcitrant compounds (Tay et al., 2005a). This was clearly not observed in this work for 2-FP. Nancharaiah et al. (2008) studied the bioaugmentation of aerobic microbial granules with Pseudomonas putida carrying the TOL plasmid and observed that granules grown in acetate may need long adaptation periods before they can metabolize recalcitrant compounds. Furthermore, they suggested that bioaugmentation could be a rapid and efficient way to provide aerobic granules with the adequate catabolic genes, thus reducing the start-up time. Consequently, in the present study, the SBR was bioaugmented with a bacterial strain able to degrade 2-FP, indicated as Rhodococcus sp. strain FP1, previously isolated in our laboratories (unpublished). As it is known from the literature that the start-up time decreases with the increase in inoculum size (Quan et al., 2003), a 50% augmentation with a pure culture of Rhodococcus sp. strain FP1 cells was performed.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 4 5 e6 7 5 2
6749
Fig. 2 e Morphology of the granules at the start-up of the SBR and during reactor operation. Granules used as inoculum for the start-up of the SBR (a); granules appearance after: 2 months operation (b); 3 months operation (c); 6 months operation (d) and 18 months operation (e). The size of the bar is 0.5 mm.
3.1.2.
After bioaugmentation
Bioaugmentation was performed using strain FP1 which is able to completely mineralize 2-FP with stoichiometric release of fluoride ion in suspension cultures supplied with 0.44 mM of the compound (unpublished). This strain is able to degrade 2-FP up to concentrations of 4 mM. When the bioaugmentation was carried out, the SBR cycle time was changed to 12 h and the settling time was increased to 20 min (particles with a settling velocity larger than 0.9 m h1 are retained in the reactor). The settling time was then gradually decreased back to 3 min (particles with a settling velocity larger than 6 m h1 are retained in the reactor) during the first
3 months after augmentation. This strategy was chosen to prevent loss of Rhodococcus sp. strain FP1, as its settling velocity was much lower than that of the aerobic granular sludge present in the reactor, and also to promote its attachment to the granules. There are two main advantages in the attachment and integration of the strain into the granules, which are avoidance of wash out of the strain from the system, resultant from the retention of the microorganism within the granules, and providing higher protection from protozoan grazing (Nancharaiah et al., 2008). In phase III the SBR was continuously fed with 0.22 mM of 2-FP and 5.9 mM of the co-substrate, acetate. The 2-FP
Fig. 3 e Biodegradation of 2-FP in the SBR before and after bioaugmentation with Rhodococcus sp. strain FP1. 2-FP inlet concentration measured after the 60 min feeding (3), 2-FP outlet concentration (-), 2-FP degraded based on fluoride release (B) and organic load based on acetate COD (L) are indicated.
6750
e 4.13
e 4.27
e 0.028
Bioaugmentation with Rhodococcus sp. strain FP1 III 1.53 0.08 IV 2.11 0.05 V 8.40 1.15 VI 17.90 0.18 VII 1.31 0
1.64 1.80 6.54 18.07 2.14
Total from phase III to VII 31.25
30.19
1.46
a 2-FP degraded based on fluoride release.
concentration was then increased to 0.44 mM in phases IV to VI. The results showed that 2-FP was completely degraded, suggesting that bioaugmentation was successfully achieved (Fig. 3, Table 2). During 2-FP degradation no intermediate metabolites, such as catechols and/or fluorocatechols, were detected in the samples. In some sampling days, such as in days 237 (phase V), 246 (phase V) and 336 (phase VI), the fluoride release obtained in the outlet of the SBR was higher than expected. This was most probably due to the adsorption of fluoride to the granules structure, which was then released to the effluent. The adsorption of fluoride to the aerobic granular sludge was studied in parallel with the SBR experiment and the results showed that granular sludge adsorb ca. 0.6 mmol F/Lgranules (data not shown). In some effluent samples, a small accumulation of 2-FP was observed (ca. 0.02 mM of 2-FP). However, in the beginning of phase V, 2-FP presence in the effluent of the SBR was more significant, which was most probably due to the decrease of the SBR cycle time from 12 h to 8 h. This leads to a higher loading rate of 2-FP. The 2-FP biodegradation profile observed in phase IV showed that this compound was degraded after 8 h and, based on this result, in phase V the SBR cycle time was reduced to 8 h. Subsequently, it was observed that, during phase V, the 2FP was completely consumed within 4 h (Fig. 4a), leading again to a change of the cycle time to 4 h (Fig. 4b). The main difference between phases VI and VII is the 2-FP concentration fed to the SBR (0.44 mM and 0.22 mM, respectively) (Table 1). The 2-FP conversion profile observed in phases VI and VII shows, respectively, that 0.44 mM of 2-FP are consumed within a 4 h period, with stoichiometric fluoride release after 4 h, and that 0.22 mM of 2-FP is consumed in just 2 h, with stoichiometric fluoride release after 3 h, suggesting that in phase VII the cycle time could have been further improved to a 3 h cycle. The SBR bedvolume was measured along the different phases of bioreactor operation (Fig. 5). The results showed that immediately after bioaugmentation (phase III), the bedvolume of the SBR increased, suggesting that Rhodococcus sp. strain FP1 was attaching to the granules. From phase IV, a step decrease in the bedvolume was observed, being more significant between phases VI and VII. This was due to the total length of reactor operation, which was 444 days, after which
b
0.12
0.12
0.06
0.06
0.00
0.00
0.18
0.18
-
2-FP degradeda
0.18
--- Phase VII 0.12
0.12
0.06
0.06
0.00
0
1
2
3
4
5
6
7
8
-
2-FP in the outlet
0.18
F release (mM)
2-FP fed
2-FP concentration (mM)
I II
2-FP mass balance (mmol)
2-FP concentration (mM)
Phase
a
0.00
Cycle time (h)
Fig. 4 e 2-FP biodegradation profiles during phase V (a) and phases VI and VII (b). 2-FP concentration (A) and fluoride release (,) are indicated.
the granules became very small, like sand, but still very dense and with extraordinary settling properties (Fig. 2e).
3.2. Analysis of the presence of Rhodococcus sp. strain FP1 in the SBR In order to follow the success of bioaugmentation, here defined as the capacity of the SBR to retain the degrading strain, aerobic granular sludge was crushed and plated in NA, close to the end of the experiment. After 3 days of incubation, different bacteria were isolated and the isolates were characterized through 16S rRNA gene analysis. According to Blast 35
Bedvolume (cm)
Table 2 e Summary of the performance of the SBR for 2-FP degradation.
F release (mM)
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 4 5 e6 7 5 2
30 25
222 d
229 d
209 d
266 d 400 d
20 15
444 d
10 5 0
I & II
III
IV
V
VI
VII
Phase Fig. 5 e Profile of the bedvolume during SBR operation. The last day of reactor operation corresponding to each phase is indicated at the top of the bars. Values are means ± standard error of the mean (SEM).
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 4 5 e6 7 5 2
results Rhodococcus sp. strain FP1 was recovered from the reactor. This strongly indicates the success of the bioaugmentation, suggesting that granular sludge is capable to incorporate and retain specialized degraders. Furthermore, this indicates that strain FP1 probably had a key role on 2-FP degradation. Previous studies have shown unsuccessful flocs bioaugmentation with labeled strains, indicating grazing by protozoa and cell wash out as the main reasons (Eberl et al., 1997; Van Veen et al., 1997), which did not happen in our study. Containment of bioaugmented strains within GAC biofilm reactors has been proven successful (Carvalho et al., 2006b; Emanuelsson et al., 2008), but in granule technology there is no physical immobilization material and, thus, we could speculate that containment of the degrading strains would have been more difficult. This further corroborates the robustness of the granular technology for application in the treatment of wastewaters.
4.
Conclusions
This study showed that it is possible to biologically remove toxic compounds, like 2-FP used in this study, from wastewaters using granular sludge SBRs. The main conclusions drawn from this work are: The aerobic granular sludge used to inoculate the SBR was not able to degrade the 2-FP fed to the reactor; Bioaugmentation with strain FP1 was successfully achieved, as complete biodegradation of 2-FP was reached; Strain FP1 was successfully recovered from the aerobic granules after 444 days of SBR operation, suggesting that granular sludge can integrate specialized degraders; Granular SBRs are very promising for the treatment of wastewaters containing toxic compounds as granulation may help retaining inoculated specialized degrading strains; This study clearly reinforces the need for bioaugmentation in cases where biodegradation of highly recalcitrant compounds is targeted.
Acknowledgments Anouk F. Duque and Maria F. Carvalho thank the research grants from Fundac¸a˜o para a Cieˆncia e Tecnologia (FCT), Portugal (Ref. SFRH/BD/30771/2006 and SFRH/BPD/44670/2008, respectively) and Fundo Social Europeu (FSE) (Programa Operacional Potencial Humano (POPH), Quadro de Refereˆncia Estrate´gico Nacional (QREN)). This work was supported by the FCT Project e PTDC/BIO/67306/2006.
references
Altschul, S.F., Madden, T.L., Schaffer, A.A., Zhang, J., Zhang, Z., Miller, W., Lipman, D.J., 1997. Gapped BLAST and PSI-BLAST: a new generation of protein database search programs. Nucleic Acids Research 25, 3389e3402.
6751
APHA, 1998. Standard Methods: For the Examination of Water and Wastewater, twentieth ed. American Public Health Association/American Water Works Association/Water Environment Federation, Washington, DC, USA. Beun, J.J., Hendriks, A., van Loosdrecht, M.C.M., Morgenroth, E., Wilderer, P.A., Heijnen, J.J., 1999. Aerobic granulation in a sequencing batch reactor. Water Research 33 (10), 2283e2290. Beun, J.J., van Loosdrecht, M.C.M., Heijnen, J.J., 2000. Aerobic granulation. Water Science and Technology 41 (4e5), 41e48. Beun, J.J., van Loosdrecht, M.C.M., Heijnen, J.J., 2002. Aerobic granulation in a sequencing batch airlift reactor. Water Research 36 (3), 702e712. Buitron, G., Schoeb, M.-E., Moreno-Andrade, I., Moreno, J., 2005. Evaluation of two control strategies for a sequence batch reactor degrading high concentration peaks of 4chlorophenol. Water Research 39, 1015e1024. Caldeira, M., Heald, S.C., Carvalho, M.F., Vasconcelos, I., Bull, A.T., Castro, P.M.L., 1999. 4-Chlorophenol degradation by a bacterial consortium: development of a granular activated carbon biofilm reactor. Applied Microbiology and Biotechnology 52, 722e729. Carucci, A., Milia, S., De Gioannis, G., Piredda, M., 2008. Acetate-fed aerobic granular sludge for the biodegradation of chlorinated phenols. Water Science and Technology 58 (2), 309e315. Carucci, A., Milia, S., De Gioannis, G., Piredda, M., 2009. Acetatefed aerobic granular sludge for the degradation of 4chlorophenol. Journal of Hazardous Materials 166, 483e490. Carucci, A., Milia, S., Cappai, G., Muntoni, A., 2010. A direct comparison amongst different technologies (aerobic granular sludge, SBR and MBR) for the treatment of wastewater contaminated by 4-chlorophenol. Journal of Hazardous Materials 177, 1119e1125. Carvalho, M.F., Ferreira, M.I.M., Moreira, I.S., Castro, P.M.L., Janssen, D.B., 2006a. Degradation of fluorobenzene by Rhizobiales strain F11 via ortho cleavage of 4-fluorocatechol and catechol. Applied Environmental Microbiology 72, 7413e7417. Carvalho, M.F., Ferreira Jorge, R., Pacheco, C.C., De Marco, P., Henriques, I.S., Correia, A., Castro, P.M.L., 2006b. Long-term performance and microbial dynamics of an up-flow fixed bed reactor established for the biodegradation of fluorobenzene. Applied Microbiology and Biotechnology 71, 555e562. De Bruin, L.M.M., de Kreuk, M.K., van der Roest, H.F.R., Uijterlinde, C., van Loosdrecht, M.C.M., 2004. Aerobic granular sludge technology: an alternative to activated sludge? Water Science and Technology 49 (11e12), 1e7. De Kreuk, M.K., Heijnen, J.J., van Loosdrecht, M.C.M., 2005. Simultaneous COD, nitrogen, and phosphate removal by aerobic granular sludge. Biotechnology and Bioengineering 90 (6), 761e769. Eberl, L., Schulz, R., Ammendola, A., Geisenberger, O., Erhart, R., Sternberg, C., Molin, S., Amann, R., 1997. Use of green fluorescent protein as a marker for ecological studies of activated sludge communities. FEMS Microbiology Letters 149, 77e83. Emanuelsson, M.A.E., Osuna, M.B., Sipma, J., Castro, P.M.L., 2008. Treatment of halogenated organic compounds and monitoring of microbial dynamics in up-flow fixed bed reactors under sequentially alternating pollutant scenarios. Biotechnology and Bioengineering 99, 800e810. Fava, F., Armenante, P., Kafkewitz, D., 1995. Aerobic degradation and dechlorination of 2-chlorophenol, 3-chlorophenol and 4chlorophenol by a Pseudomonas pickettii strain. Letters in Applied Microbiology 21, 307e312. Haggblom, M.M., 1992. Microbial breakdown of halogenated aromatic pesticides and related compounds. Microbiology Reviews 103, 29e72. Jiang, H.L., Tay, J.H., Tay, S.T.L., 2002. Aggregation of immobilized activated sludge cells into aerobically grown microbial granules for the aerobic biodegradation of phenol. Letters in Applied Microbiology 35, 439e445.
6752
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 4 5 e6 7 5 2
Key, B., Howell, R., Criddle, C., 1997. Fluorinated organics in the biosphere. Environmental Science and Technology 31, 2445e2454. Lane, D.J., 1991. 16S/23S rRNA sequencing. In: Stackebrandt, E., Goodfellow, M. (Eds.), Nucleic Acid Techniques in Bacterial Systematics. John Wiley and Sons Ltd, West Sussex, England. Morgenroth, E., Sherden, T., van Loosdrecht, M.C.M., Heijnen, J.J., Wilderer, P.A., 1997. Aerobic granular sludge in a sequencing batch reactor. Water Research 31 (12), 3191e3194. Murphy, C.D., Clark, B.R., Amadio, J., 2009. Metabolism of fluoroorganic compounds in microorganisms: impacts for the environment and the production of fine chemicals. Applied Microbiology and Biotechnology 84, 617e629. Nancharaiah, Y.V., Joshi, H.M., Hausner, M., Venugopalan, V.P., 2008. Bioaugmentation of aerobic microbial granules with Pseudomonas putida carrying TOL plasmid. Chemosphere 71, 30e35. Osuna, M.B., Sipma, J., Emanuelsson, M.A.E., Carvalho, M.F., Castro, P.M.L., 2008. Biodegradation of 2-fluorobenzoate and dichloromethane under simultaneous and sequential alternating pollutant feeding. Water Research 42, 3857e3869. Quan, X., Shi, H., Wang, J., Qian, Y., 2003. Biodegradation of 2,4dichlorophenol in sequencing batch reactors augmented with immobilized mixed culture. Chemosphere 50, 1069e1074. Rainey, F.A., Ward-Rainey, N., Kroppenstedt, R.M., Stackebrandt, E., 1996. The genus Nocardiopsis represents a phylogenetically coherent taxon and a distinct
actinomycete lineage: proposal of Nocardiopsaceae fam. nov. International Journal of Systematics Bacteriology 46, 1088e1092. Rittmann, B.E., Whiteman, R., 1994. Bioaugmentation: a coming age. Water Quality International 1, 12e16. Tay, S.T.L., Moy, B.Y.P., Jiang, H.L., Tay, J.H., 2005a. Rapid cultivation of stable phenol degrading granules using acetate fed granules as microbial seed. Journal of Biotechnology 115, 387e395. Tay, S.T.L., Moy, B.Y.-P., Maszenan, A.M., Tay, J.-H., 2005b. Comparing activated sludge and aerobic granules as microbial inocula for phenol biodegradation. Applied Microbiology and Biotechnology 67, 708e713. Van Veen, J.A., van Overbeek, L.S., van Elsas, J.D., 1997. Fate and activity of microorganisms introduced into soil. Microbiology and Molecular Biology Reviews 61, 121e135. Wang, S.G., Liu, X.W., Zhang, H.Y., Gong, W.X., Sun, X.F., Gao, B.Y., 2007. Aerobic granulation for 2,4-dichlorophenol biodegradation in a sequencing batch reactor. Chemosphere 69, 769e775. Xu, H., Tay, J.H., Foo, S.K., Yang, S.F., Liu, Y., 2004. Removal of dissolved copper(II) and zinc(II) by aerobic granular sludge. Water Science and Technology 50, 155e160. Yu, Z., Mohn, W.W., 2001. Bioaugmentation with resin-aciddegrading bacteria enhances resin acid removal in sequencing batch reactors treating pulp mill effluents. Water Research 35 (4), 883e890.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 5 3 e6 7 6 2
Available online at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Establishment of a real-time PCR method for quantification of geosmin-producing Streptomyces spp. in recirculating aquaculture systems Marc Auffret a,*, Alexandre Pilote b, E´milie Proulx b, Daniel Proulx b, Grant Vandenberg b, Richard Villemur a a b
INRS-Institut Armand-Frappier, 531 Boulevard des Prairies, Laval, Que´bec H7V 1B7, Canada De´partement des sciences animales, Universite´ Laval, Que´bec, QC G1V 0A6, Canada
article info
abstract
Article history:
Geosmin and 2-methylisoborneol (MIB) have been associated with off-flavour problems in
Received 16 May 2011
fish and seafood products, generating a strong negative impact for aquaculture industries.
Received in revised form
Although most of the producers of geosmin and MIB have been identified as Streptomyces
26 September 2011
species or cyanobacteria, Streptomyces spp. are thought to be responsible for the synthesis
Accepted 16 October 2011
of these compounds in indoor recirculating aquaculture systems (RAS). The detection of
Available online 24 October 2011
genes involved in the synthesis of geosmin and MIB can be a relevant indicator of the beginning of off-flavour events in RAS. Here, we report a real-time polymerase chain
Keywords:
reaction (qPCR) protocol targeting geoA sequences that encode a germacradienol synthase
Off-flavour
involved in geosmin synthesis. New geoA-related sequences were retrieved from eleven
Geosmin
geosmin-producing Actinomycete strains, among them two Streptomyces strains isolated
2-Methylisoborneol
from two RAS. Combined with geoA-related sequences available in gene databases, we
geoA
designed primers and standards suitable for qPCR assays targeting mainly Streptomyces
Real-time PCR
geoA. Using our qPCR protocol, we succeeded in measuring the level of geoA copies in sand
Recirculating aquaculture systems
filter and biofilters in two RAS. This study is the first to apply qPCR assays to detect and quantify the geosmin synthesis gene ( geoA) in RAS. Quantification of geoA in RAS could permit the monitoring of the level of geosmin producers prior to the occurrence of geosmin production. This information will be most valuable for fish producers to manage further development of off-flavour events. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
Geosmin and 2-methylisoborneol (MIB) are the most common causes of odour problems in drinking water obtained from surface water (Suffet et al., 1996), and have also been associated with off-flavour problems in fish and seafood products (Smith et al., 2008). Although these two compounds have not been associated with any serious health effects, the unpleasant taste
and odour from their presence in water (Smith et al., 2002) and fish (Howgate, 2004) are perceived as unsafe by consumers. Moreover, these compounds are characterized by an exceptionally low detection threshold by human senses, in the order of 2e10 ng/L (Cook et al., 2001). Finally, off-flavours have a strong negative impact for water and aquaculture industries by increasing production costs by as much as U.S. $47 million in 1999 (Hanson, 2003). Off-flavour testing or the consequences
* Corresponding author. Tel.: þ1 450 687 5010; fax: þ1 450 686 5501. E-mail address:
[email protected] (M. Auffret). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.10.020
6754
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 5 3 e6 7 6 2
generated by delays of production are the prevalent causes of these economical lost (Hanson, 2003). Although a wide variety of microorganisms have been shown to produce these secondary metabolites (Dickschat et al., 2005), most of the producers of geosmin and MIB have been identified as Streptomyces species or cyanobacteria (Izaguirre and Taylor, 1998; Ludwig et al., 2007). Streptomyces spp. are thought to be responsible for the synthesis of these compounds in indoor recirculating aquaculture systems (RAS) (Guttman and van Rijn, 2008; Schrader and Summerfelt, 2010). Therefore, early detection of off-flavour-metabolite producing Streptomyces spp. can be very useful in preventing the occurrence of such molecules in RAS by appropriate operating actions. Detection techniques for actinomycetes such as plate count (Zaitlin et al., 2003) and fluorescence in situ hybridization with catalyzed reporter deposition (CARD-FISH), detect both off-flavour and non-off-flavour producers and showed poor correlation with geosmin and/or MIB concentrations (Klausen et al., 2005). The detection and quantification of genes involved in the synthesis of these compounds can be a relevant indicator of the beginning of off-flavour events in RAS. Geosmin is produced by the conversion of farnesyl diphosphate (FPP) to germacradienol then geosmin (Jiang et al., 2007), by a sesquiterpene synthase encoded by homologous genes found in Streptomyces coelicolor A3(2) (cyc2), Streptomyces avermitilis ( geoA) or Streptomyces peucetius (Spterp13) (Cane and Watt, 2003; Cane et al., 2006; Ghimire et al., 2008). Nucleic sequences related to terpene synthases-encoding genes were detected in cyanobacteria (Ludwig et al., 2007; Giglio et al., 2008) but the correlation between their presence and the production of geosmin was only demonstrated by Giglio et al. (2008). A putative terpene synthase gene was found in the genome of Myxococcus xanthus (accession number YP_634376). Genes encoding functional monoterpene cyclase (as well named 2-MIB synthase) involved in the synthesis of 2-
MIB were identified in the genome of several Streptomyces strains (Komatsu et al., 2008; Wang and Cane, 2008). Wang et al. (2011) suggested that the synthesis of MIB is similar in cyanobacteria and actinomycetes but cyanobacteria represent a unique lineage possibly explained by multiple rearrangements of the genes in this operon (Giglio et al., 2011). However, such genes were not found in the genome of other MIB producers such as myxobacteria. In RAS, detection of geosmin and MIB is performed by gas chromatographyemass spectrometry (GCeMS). Fish are known to concentrate these compounds in their flesh, therefore when these molecules are detected in fish flesh, this usually means that the off-flavour producers are well established in RAS. We hypothesized that by targeting the geosmin and/or the MIB synthesis genes in RAS by real-time PCR (qPCR), we can measure the level of off-flavour-metabolite producers in RAS before the occurrence of off-flavour in fish. PCR assays were used before to detect these genes (Ludwig et al., 2007; Giglio et al., 2008), but the primers were not specifically designed to quantitate geosmin synthesis gene by qPCR. In this paper, we report new geoA and tpc sequences from Streptomyces strains, and the development of a qPCR protocol for the detection and the quantification of geoA in RAS.
2.
Materials and methods
2.1.
Recirculating aquaculture system and sampling
The indoor RAS units were located in the Laboratoire Re´gional des Sciences Aquatiques (LARSA, ULaval, Qc, Canada). Four different units schematically presented in Fig. 1 were studied for the detection of geosmin and/or MIB producers. Fresh water contained in the pumping tank was transferred to the sand filter. This compartment serves to collect suspended
Fish tank(s)
Water
Pumping basin
(2)
NH4+ + O2
(CH2O)n + nO2
Sand filter
Water Cooler
Biofilter
(1)
Pump
NO3nCO2 + H2O
Fig. 1 e Schematic presentation of the RAS. Solid arrows indicate the direction of the water flow. The chemical reactions of nitrification (1) and mineralization of the organic matter (2) in the biofilter compartment are indicated. Not in scale.
6755
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 5 3 e6 7 6 2
solids (faeces, uneaten feed and bacterial flocs). Then the water passed through a water cooler before going to the fish tanks, and to the trickling biofilter (height 3 m). The biofilter was composed of polystyrene as biofilm attachment carriers, and was inoculated with a nitrification/denitrification commercial bacterial consortium (Bacta-Pur N3000, IETAquarecherche Lte´e, North Hatley QC, Canada). When the optimum criteria for the water quality and fish growth were achieved, fish were added in the fish tanks. The recirculation rate was more than 90%. New fresh water was added at 20 L h1 with a retention time of 3.5 days. Samples were taken from four different RAS units and the different criteria applied in these RAS units are described in Table 1. Water samples from the fish tank of units 1 and 3 were collected and preserved at 4 C. Two hundred mL of water from the sand filter (unit 1) was collected in a 500 mL sterilized bottle and preserved at 4 C before filtration. Biofilter media (unit 2) were collected, washed with 5 mL of phosphate buffer (20 mM, pH 7.2) and vortexed twice at maximum speed for 10 s to detach the biomass. Fifty mL from the sediment trap at the bottom of the fish tank in the unit 4 was collected. This trap was used to collect the faecal waste produced by fish.
2.2.
Bacterial strains
The strains iafA, iafB, iafE, iafF, iafG, iafH and iafI affiliated to the genus Streptomyces were from a bacterial collection at INRSInstitut Armand Frappier (IAF), and cultured on actinomycete isolation agar (Aia medium; Becton Dickinson (BD), Franklin lakes, NJ, USA) supplemented with glycerol for 2e3 days, at 30 C. Affiliation was determined by PCR amplification of the 16S ribosomal RNA (rRNA) gene sequences with the 27f-YM/ 1492r primers (Table 2) and the sequencing of resulted amplicons (accession number JN652249eJN652257;
JN652249;JN652250;JN652251;JN652252;JN652253;JN652254;JN6 52255;JN652256;JN652257). S. avermitilis ATCC 31267 and Saccharopolyspora erythraea ATCC 11635 were from American Type Culture Collection (ATCC). They were also cultured on Aia medium. The Aia medium supplemented with glycerol was inoculated with 100 mL of water samples from unit 1 or unit 3, and incubated for 2e3 days, at 30 C. The plates were inspected for the presence of actinomycete isolates. Suspected isolates characterized by the presence of aerial mycelium and spore mass colour were transferred to fresh Aia plates. Ten isolates from each unit were randomly chosen and their 16S rRNA gene sequence determined. Respective of each unit, these sequences were identical, and all affiliated to the genus Streptomyces (accession numbers JN652249 and JN652250). Representative isolates (strain AMU11 from unit 1 and strain AMU14, from unit 3) were further purified by restreaking at least 3 times.
2.3.
Geosmin and MIB measurements
Organic matter and the faecal waste produced by the unit 4 fish tank did not exhibit the characteristic off-flavour odour. This sample was not tested by GCeMS due to the exceptionally low detection threshold by human sensory analyses (2e10 ng/L; Cook et al., 2001). Although off-flavours could have been present below this level, we considered unit 4 as a negative control for qPCR assays. Water samples (25 mL) collected from unit 2 biofilter and the unit 1 sand filter were placed in glass vials (40 mL, 8DRM W/cap 24-400, Agilent Technologies, Santa Clara, CA, USA). Vials were capped with foil-lined caps and septum. Bacterial strains were cultured in 5 mL M14 medium (Page´ et al., 1996) at 30 C for 48 h before their transfer directly in glass vials.
Table 1 e Characterization of the recirculating aquaculture systems (RAS). RAS unit
1
2
3 Yes
4
Operated in a recirculation mode Fish species
No
No
No fish
No fish
Biomass of fish (kg/m3) Commercial Dieta
0
0
NA
NA
Corey optimum
Biofilter status Oxygen level (% of saturation) Water volume in the unit (L) Water temperaturec ( C) Off-flavoursd
Inactive 90e100
Activeb 90e100
Active 90e100
400
400
800
3200
15 Yes
15 Yes
7e15 Yes
8 No
Brook trout Salvelinus fontinalis 25
Yes Rainbow trout Oncorhynchus mykiss 33 Corey optimum or Martin Mills Classic or Martin Mills High Energy or Skretting Orient Active 95e100
NA: not applicable. a More than one commercial diet was used in different fish tanks belonging to the same RAS unit. b Ammonium salt (NH4Cl, 150 g /week) was added to maintain the denitrification/nitrification microbial activity in the biofilter. c The temperature of water was controlled with a water cooler prior to the introduction of fish. d Detected by human senses.
6756
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 5 3 e6 7 6 2
Table 2 e Oligonucleotides. Expected length (bp)
Annealing temp. ( C)
Reference
AGAGTTTGATYMTGGCTCAG AAGGAGGTGATCCAGCCGCA
w1500
55
Franck et al., 2008 Edwards et al., 1989
CycFW CycRW
TGGTAYGTITGGGTITTYTTYTTYGAYGAYCAYTT CATRTGCCAYTCRTGICCICCISWYTGCCARTCYTG
730
52
Ludwig et al., 2007
geoA
250F 971R
TTCTTCGACGAYCACTTCC CCCTYGTTCATGTARCGGC
743
60
Giglio et al., 2008
tpc
AMmib-F AMmib-R
TGGACGACTGCTACTGCGAG AAGGCGTGCTGTAGTTCGTTGTG
592
58
This paper
AMgeo-F AMgeo-R
GAGTACATCGAGATGCGCCGCAA GAGAAGAGGTCGTTGCGCAGGTG
167
66
This paper
Target gene
Primers
16S rRNA
27f-YM 1492r
geoA
QPCR primers geoAa
Sequence (5’e3’)
a AMgeo-F/R primers correspond to nucleotides 532e554 and 676e698 of the geoA gene of S. avermitilis MA4680 (locus tag SAV_2163), respectively. See Figs. S1 and S3 in supplemental data for sequence alignments for designing the geoA and tpc primers.
Twenty mL of demineralized water, 6 g of sodium chloride and a magnetic stir bar were then added just prior to analysis. Water samples and bacterial cultures vials were placed at 4 C until analysis. Geosmin and MIB concentrations were measured by using the SPME protocol based on Lloyd et al. (1998) and modified as follows. Vials were incubated 20 min in a water bath at 40 C, then a 65 mm PDMS-DVB fibre (Supelco, Bellefonte, PA, USA) were injected through the septum of the vial. After 35 min with agitation, the fibres were taken and incubated for 3 min at 250 C into the splitless operated injector of a HP5890 Series II Plus GC with a HP5972 MS. The GC was operated with a DB-17ms column (Agilent Technologies). Helium was used as the carrier gas at constant flow rate of 2 mL/min. Oven temperature was held at 60 C for 2 min from injection, increased to 132 C at 4 C/min and held at 132 C for 1 min. The temperature increased again to 250 C at 20 C/min and held at this maximum temperature for 5 min. Standard solutions (Supelco) were used to determine the molecular ion base peaks and the retention time. These values were monitored at m/z 95, 135 and 168 for 2-MIB and at m/z 112, 126 and 182 for geosmin. The retention time was 11.7 min for MIB and 19.7 min for geosmin.
2.4.
DNA extraction
Bacterial cultures were centrifuged 3 min at 3000 rpm (1509 g). The cell pellets were then dispersed in 500 mL TEN (50 mM TriseHCl pH 8.0; 100 mM ethylenediaminetetraacetic acid (EDTA) pH 8.0, 150 mM NaCl) and transferred in 2 mL tubes containing 250 mg glass beads (0.25e0.50 mm). The cells were disrupted twice for 20 s at speed 5.0 with a FastPrep homogenizer (MP Biomedicals {Qbiogene}, Solon, OH, USA) and put on ice. The homogenate was centrifuged for 15 min at 13,000 rpm (16,060 g). The supernatant was then extracted once with phenolechloroformeisoamyl alcohol (25:24:1) and once with chloroformeisoamyl alcohol (49:1). DNA was precipitated overnight at 20 C by adding ammonium acetate (0.1 M final concentration), 5 mL of a diluted solution (1:50) of glycogen at 5 mg/mL (Ambion, Austin, TX, USA) and two
volumes of ethanol. After centrifugation at 13,000 rpm (16,060 g) for 15 min, the DNA pellet was washed once with 70% ethanol then with 95% ethanol and dissolved in 50 mL TE (10 mM TriseHCl, 0.1 mM EDTA, pH 8). The water sample from sand filter (unit 1) was filtered on 0.22 mm 47-mm MF mixed cellulose ester filters (Millipore, Billerica, MA, USA). The filter was cut in small pieces under sterilized conditions and put in 2 mL tubes containing 250 mg glass beads to prepare the DNA extraction as described previously. The biomass collected from the biofilter (unit 2) was centrifuged at 3000 rpm (1509 g) for 15 min and dispersed in TEN. DNA was extracted as described above. The 50 mL samples from the sediment trap (unit 4) were centrifuged for 20 min at 8000 rpm (10,732 g), and dispersed in TEN. DNA was extracted as described above.
2.5. PCR amplification, cloning and phylogenetic analysis ClustalW multiple alignment analyses (Thompson et al., 1994) were performed with all of the homologous tpc gene sequences available in GenBank (Fig. S3). The AMmibF/ AMmibR primers designed for tpc (Table 2) were tested on all samples to obtain the amplicon with the expected length (ca 600 bp; from positions 710e1301 in Streptomyces lasaliensis, accession number AB547324). PCR were performed in a 50 mL volume with 1.5 mM MgCl2, 20 mg bovine serum albumin, 200 mM deoxynucleotide triphosphate, 25 pmol of each primer (Table 2), total DNA (10e50 ng), and 2.5 U of Taq DNA polymerase with Thermopol buffer (New England Biolabs, Ipswich, MA, USA). The addition of 1 mL of dimethyl sulfoxide 100% (Finnzymes, Vantaa, Finland) per PCR pre-mix was necessary to amplify the geoA and tpc sequences. Amplifications were performed at 96 C for 5 min, then 30 cycles at 95 C for 30 s, 30 s at the specific annealing temperature (Table 2), 70 C for 60 s, and a final extension period of 7 min at 72 C. The PCR products were concentrated by precipitation with ammonium acetate (1 M final concentration) and ethanol
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 5 3 e6 7 6 2
(two volumes), washed with 70% and 95% ethanol, and dissolved in 10 mL of water. The geoA and tpc products were cloned into the T-vector pGEM-T Easy (Promega, Madison, WI, USA) and transformed in the Escherichia coli DH5a host strain (Invitrogen, Burlington, ON, Canada), according to the manufacturers’ specifications. Representative clones were sent for sequencing at the “Centre d’innovation Ge´nome Que´bec, Universite´ McGill” (Montre´al, QC, Canada). PCRamplified 16S rRNA gene sequences were sent directly to the sequencing service. Raw sequence data were assembled in BioEdit v7.05 (Ibis Bioscience, Carlsbad, CA). The sequences were manually aligned by comparing forward and reverse sequences. Amino acid sequences were aligned using ClustalW with both closely related representatives from NCBI BLASTP as well as novel complete or partial sequences obtained from BLASTP. Phylogenetic relationships were constructed with evolutionary distances (Poisson-correction distances) and the neighbourjoining method using the MacVector 9.0 software package (Accelrys, San Diego, CA). The bootstrap analyses for the phylogenetic trees were calculated by running 1000 replicates for the neighbour-joining data.
2.6.
Design of primers and standards for qPCR analysis
From our strain collection or from sequences provided by gene databases, 21 nucleic acid sequences related to geosmin synthase, mostly from Streptomyces species, were aligned (Fig. S1). Consensus sequences were based on Streptomyces sequences containing two degenerate positions. To improve the qPCR efficiency, a set of non-degenerate primers was designed by using the base most frequently found in these positions in our alignment (Powell et al., 2006). qPCR primers (AMgeo-F/R) were designed to amplify a specific fragment of geoA at the positions 532e698 in S. avermitilis MA4680 (strain ATCC 31267, GenBank accession number BA000030, gene tag SAV_2163) covering one metal (Mg2þ)-binding motif of germacradienol/ geosmin synthase (Ludwig et al., 2007). Four geoA sequences from strains AMU14, iafH and iafI, and S. avermitilis ATCC 31267 were used as standards for qPCR assays. Plasmid DNA from a representative clone of these geoA sequences was produced using the method of Serghini et al. (1989). geoA sequences were PCR amplified with the sp6/T7 primers that flanked the cloned geoA sequence in the plasmid vector. The resulted amplicons were purified by agarose gel electrophoresis with the Wizard SV Gel and PCR Clean-Up System (Promega) according to the manufacturers’ specifications. The concentration of each PCR fragment was measured by fluorescence using the Quant-iT PicoGreen dsDNA Reagent and Kits (Invitrogen) using Lambda DNA as standard. The number of gene copies per mL (N ) was calculated as: 1 1 1 N molecules=mL ¼ C g=mL Y bp 660 g mole bp 6:022 molecules=mole where C is the concentration of the PCR fragment and Y the PCR fragment length in base pair (bp). These amplicons were used individually in a range from 4 104 to 4 108 gene copies/reaction, or were mixed (ratio 1:1:1:1) and used in a range from 4 102 to 4 108 gene copies/reaction. The
6757
amplification efficiencies for these standards were calculated using the formula E ¼ [10(1/Slope) 1]. The slope was determined from plotting the log of gene copy number by the threshold cycle number Ct values.
2.7.
qPCR protocol
qPCR were performed in triplicate using a Rotor-Gene Q instrument (Qiagen, Venlo, Netherlands). Each qPCR reaction was performed in 20 mL with the Perfecta SYBR Green Fast Mix (QuantaBiosciences, Gaithersburg, MD, USA), 250 nM of each primer (AMgeo-F/R), and the template DNA. Ten ng of total DNA extracted from RAS units were used. The amplification conditions were as follows: preheating at 95 C for 10 min and then 40 cycles of 95 C for 15 s, 66 C for 60 s and 72 C for 15 s. Melting curves were performed to confirm purity of the amplified product.
2.8. Determination of the genome concentration in liquid cultures Streptomyces sp. strains AMU14 and iafH, and S. erythraea ATCC 11635 were grown in TSB medium at 30 C to obtain an OD600nm of 1.5 after 48 h. Cultures of strains AMU14 and iafH were diluted at 1:10 and 1:100 in sterile phosphate buffer (20 mM, pH 7.2). One mL of undiluted and diluted cultures was centrifuged at 13,000 rpm (16,060 g) for 1 min. DNA extraction was then performed on the pellets. Quantification of total double strand DNA was performed in triplicate by using the Quant-iT PicoGreen kit on 1 mL DNA sample. qPCR was performed in triplicate with 1 mL DNA sample.
2.9. Assessment of qPCR inhibition by environmental biomass Streptomyces sp. strains AMU14 and iafH were cultured and diluted as described in the previous section. One mL of undiluted and diluted cultures was centrifuged at 13,000 rpm (16,060 g) for 1 min, and dispersed in 1 mL of the water sample taken from the sediment trap of unit 4, which contained high level of organic matter These were then centrifuged at 13,000 rpm (16,060 g) for 1 min, before DNA extraction was performed on the pellets. qPCR was performed in triplicate with 1 mL DNA sample.
3.
Results
3.1.
New putative geoA and tpc gene sequences
Only few geoA gene sequences were available in gene databases at the beginning of this study. As we wanted to design qPCR primers to target as many as possible Streptomyces species, we used the PCR primers targeting geoA in cyanobacteria (CycFW/CycRW and 250F/971R, Table 2; Giglio et al., 2008; Ludwig et al., 2007) to detect this gene in seven Streptomyces strains from our strain collections, two Streptomyces strains isolated from RAS units (AMU11 and AMU14; see M&M section), in S. avermitilis ATCC 31267 and in the Actinobacteria S. erythraea ATCC 11635. All these strains produced geosmin in our
6758
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 5 3 e6 7 6 2
3.2.
culture conditions as determined by GCeMS analysis (Table 3). In all these cases, we detected a PCR product of the expected size (ca. 750 bp). Cloning and sequencing of these PCR products revealed in all cases putative geoA gene sequences with identity in the deduced amino acid sequences ranging from 62.3% to 100%. Phylogenic analysis showed that the Streptomyces geoA deduced amino acid sequences can be divided in four clusters based on the “geosmin synthase” (Fig. 2): Streptomyces sp. strains iafA, iafF, iafG and iafI grouped with the putative germacradienol/geosmin synthase (group A), which might include the geoA found in Kitasatospora setae. S. avermitilis strain MA4680 (corresponding to the strain ATCC 31267) formed a group with S. peucetius and Streptomyces sciabies (group B) encoding a germacradienol/geosmin synthase. Streptomyces sp. strains iafE and iafH grouped with a putative cyclase (group C) affiliated to the closest GeoA found in S. coelicolor strain A3(2). Finally, Streptomyces sp. strains AMU11, AMU14 and iafB grouped with the putative germacradienol synthase (group D) found in Streptomyces griseus and Streptomyces flavogriseus. We also detected geoA-related sequences in two RAS units (sand filter from unit 1 and biofilter from unit 2) that were producing geosmin at the time of the sampling (Table 3). The geoA product revealed in unit 2 was most closely related to geoA found in M. xanthus DK1622 (81% identity), while the one identified in unit 1 grouped with the geoA found mostly in Streptomyces spp. (group D; Fig. 2). We designed PCR primers targeting tpc based on the analysis of bacterial genome databases by Komatsu et al. (2008). Putative tpc sequences were detected only in Streptomyces sp. strains AMU11, AMU14, iafB and iafH (Table 3). However, MIB was only detected in strains AMU11 and AMU14 in the culture conditions used. tpc sequences from strains iafB and AMU11 are highly identical to S. coelicolor A3(2) tpc, and strains iafH and AMU14 to S. griseus tpc (Fig. 3). Due to the lack of sequence diversity in tpc sequences, we decided to focus on the geoA sequences in the development of a qPCR protocol.
Development of qPCR assays for geoA
The AMgeo primers (Table 2) were designed for qPCR protocol based of the geoA sequences available in gene databases and in this study. Four geoA sequences from strains AMU14 (group D), iafH (group C) and iafI (group A), and S. avermitilis ATCC 31267 (group B) were used as standards for qPCR assays. Each amplicon was representative of their corresponding group (AeD) allowing the coverage a large diversity of geosmin synthesis genes. The amplification efficiencies for each amplicon ranged from 0.9 to 1.1 with a linear response for the standard curve (r2) over 0.99. To mimic environmental samples where more than one geoA sequences from Streptomyces species can occur, qPCR efficiency assays were performed with a mix of the four amplicons to verify whether different geoA gene sequences could influence this efficiency. The amplification efficiency was similar to the ones performed with the individual amplicon with a linear range from 4 102 to 4 108 geoA copies. The mix of the four amplicons was then used as standards in the following experiments.
3.3. Quantification of geoA in pure cultures and in geosmin-producing RAS units The geoA concentration (gene copies/mL) in cultures of Streptomyces strains AMU14 and iafH was determined by qPCR, and these values were compared to the theoretical genome copy number as determined by measuring the DNA concentration. We assume that 0.01 pg of genomic DNA represent approximately one genome equivalent to a single bacterial cell, and that one geoA gene copy is present in the genome of these strains (Bentley et al., 2002; Ohnishi et al., 2008; Omura et al., 2001). The quantitative methods were performed on DNA extracted from 48-h undiluted cultures, but also on 10- and 100-time dilutions of these cultures. For strains AMU14, and
Table 3 e Characterization of the bacterial strains. Bacterial strains (accession number) Strain AMU11 (JN652249) Strain AMU14 (JN652250) Strain iafA (JN652251) Strain iafB (JN652252) Strain iafE (JN652253) Strain iafF (JN652254) Strain iafG (JN652255) Strain iafH (JN652256) Strain iafI (JN652257) Streptomyces avermitilis ATCC 31267 Saccharopolyspora erythraea ATCC 11635 Sand filter unit 1 Biofilter unit 2 Water unit 4
GCeMS analysis Geosmin
MIB
PCR geoA
tpc
Affiliation by 16S rRNA gene sequence
% identity
þ þ þ þ þ þ þ þ þ þ
þ þ
þ þ þ þ þ þ þ þ þ þ
þ þ þ þ
S. anulatus strain HBUM174206 S. flavogriseus ATCC 33331 S. champavatii isolate XSD-106 S. flavogriseus ATCC 33331 S. griseorubens strain 70063 S. lividans S. flavogriseus ATCC 33331 S. tricolor strain HBUM175093 S. champavatii strain 173753 NA
100% 100% 99% 99% 98% 99% 99% 99% 99% NA
þ
þ
NA
NA
þ þ ND
ND
þ þ
NA NA NA
NA NA NA
The CycFW/CycRW and 250F/971R primers were used to detect geoA, and the AmibF/R primers for tpc. ND: Not done. NA: Not applicable. Geosmin and MIB in units 1 and 2 were detected in the fish flesh.
6759
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 5 3 e6 7 6 2
Actinosynnema mirum DSM 43827 (ACU34935) Sa. erythraea ATCC 11635 (YP_001107098) Frankia sp. EAN1pec (YP_001509819) 89
Myxococcus xanthus DK 1622 (YP_634376)
97
Biofilter U2 (JN652244) Kitasotospora setae NBRC 14216 (BAJ30389)
91
S. champavatii strain iafA (JN652236)
100 98 97
S. lividans strain iafF (JN652239) Streptomyces sp. strain iafG (JN652240)
Group A
S. champavatii strain iafI (JN652242) S. peucetius ATCC 27952 (ABY50951)
75
S. scabies 87.22 (YP_003487693)
99
Group B
S. avermitilis ATCC 31267 (NP_823339) S. griseorubens strain iafE (JN652238)
93 100 92
S. coelicolor A3(2) (Q9X839.3)
Group C
S. tricolor strain iafH (JN652241) 100
S. flavogriseus ATCC 33331 (ADW07414) S. flavogriseus strain iafB (JN652237)
73
S. griseus NBRC 13350 (BAG23668)
97 100
Group D
S. anulatus AMU11 (JN652234) S. flavogriseus AMU14 (JN652235) Sand filter U1 (JN652243)
Fig. 2 e Phylogenetic neighbour-joining tree of GeoA. geoA amino acid sequences from GenBank and derived from our strain collection (see Table 3) were aligned. The tree is based on protein distance matrix analysis and the neighbour-joining method with Poisson correction within MacVector version 9.0 software. Bootstrap values greater than 50% derived from 1000 replicates are reported at the nodes. The bar represents 5% sequence divergence. See Fig. S2 for the amino acid alignment.
Sa. erythraea ATCC 11635 (YP_001105919) Stackebrandtia nassauensis DSM 44728 (YP_003510780) Micromonospora sp. (ZP_04609212) Nocardiopsis dassonvillei (ZP_04336204) Catenulispora acidiphila (YP_003115314)
88 100
Oscillatoria limosa LBD 305b (ADU79150) Planktothricoides raciborskii CHAB3331 (AEK21538) Pseudanabaena limnetica (ADU79148)
78
Pseudanabaena sp. dqh15 (AEK21531) Pseudanabaena sp. NIVA-CYA (ADU79149) S. flavogriseus ATCC 33331 (ADW07061)
75 96
S. griseus NBRC 13350 (BAG18098)
S. flavogriseus AMU14 (JN652246) 99 S. tricolor strain iafH (JN652248) Kitasatospora setae KM-6054 (BAJ32779) S. bingchenggensis BCW-1 (ADI12075)
100
S. coelicolor A3(2) (Q9F1Y6.1) 99 99 81
Streptomyces anulatus AMU11 (JN652245) S. flavogriseus strain iafB (JN652247)
S. ambofaciens ATCC 23877 (CAJ89344) S. lasaliensis ATCC 31180 (BAI77523) S. scabies 87.22 (YP_003493696)
Fig. 3 e Phylogenetic neighbour-joining tree of Tpc. tpc amino acid sequences from GenBank and derived from our strain collection (see Table 3) were aligned. See Fig. 2 legend for detail. The bar represents 20% sequence divergence.
6760
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 5 3 e6 7 6 2
iafH both quantitative methods showed similar results in all dilutions (Table 4). As one of the main objectives was to monitor geoA in RAS by qPCR, we assessed the impact of the biomass from a RAS sample on the quantification of geoA gene in Streptomyces strains AMU14 and iafH cultures. Water sample taken from the sediment trap of unit 4 (geosmin not detected) was added to the AMU14 and iafH cultures prior to DNA extraction. With strain iafH, the RAS biomass had no significant effect on the determination of geoA concentration by qPCR (Table 4). With strain AMU14, the geoA concentration was underestimated by a factor 2 (Table 4). The geoA concentration was determined in S. erythraea ATCC 11635 cultures, and compared with the genome concentration. The geoA gene sequence in S. erythraea ATCC 11635 is 71.8e74.8% identical in nucleic acid sequence to the 4 standard geoA sequences. The qPCR method underestimated the concentration of geoA by a factor 2, which is a factor similar than the results obtained with the AMU14 and iafH pure cultures. Finally, the geoA concentrations were measured in DNA samples retrieved from two RAS units that showed production of geosmin. In the sand filter unit 1, the geoA concentration was 1.63 107 copies/mL (SD: 0.94 107). It was 7 times less abundant in the biofilter of the unit 2 at 2.31 106 copies/mL (SD: 0.19 106).
4.
Table 4 e Quantification of geoA by qPCR and by DNA concentration. Dilution factor
1:1
1:10
1:100
Genome or geoA copy number/mL 109 (SD)
108 (SD)
107 (SD)
4.91 (0.11) 3.84 (0.08)
2.82 (0.20) 3.73 (0.06)
3.12 (0.77) 2.96 (0.14)
Strain AMU14 þ RAS 1.68 (0.13) geoAb
1.62 (0.07)
1.45 (0.08)
Strain iafH Genome geoAa
4.48 (1.37) 6.92 (0.015)
6.46 (1.08) 6.21 (0.24)
6.80 (1.33) 5.57 (0.40)
Strain iafH þ RAS geoAb
6.12 (0.23)
8.36 (1.00)
5.51 (0.05)
Strain AMU14 Genome geoAa
Saccharopolyspora erythraea ATCC 11635 Genome 1.08 (0.04) ND 0.51 (0.03) ND geoAa
ND ND
SD: standard deviation of triplicate of two independent assays. ND: not done. a qPCR on genomic DNA of strain AMU14, strain iafH or Saccharopolyspora erythraea. b qPCR were performed on DNA extracted from strain AMU14 or strain iafH cultures and mixed with a water sample taken from the unit 4 sediment trap.
Discussion
In this study, we report new putative geoA and tpc sequences, encoding germacradienol/geosmin synthase and 2methylisoborneol synthase, respectively, from different Streptomyces strains. Presence of geoA-related sequences correlated with the detection of geosmin in culture of these strains, suggesting that these sequences encode for GeoA. Combined with the known sequences, we derived a consensus alignment to derive qPCR primers targeting geoA. To improve the amplification efficiency criteria during qPCR assays, nondegenerate primers were derived, which resulted in primers prioritizing Streptomyces geoA. This strategy could have impaired our assays of detecting non-Streptomyces geoA in the RAS, as variability exists between geoA sequences found in different bacterial genus and cyanobacteria (Guttman and van Rijn, 2008; Ludwig et al., 2007). However, using these primers, the detection and quantification of geoA in S. erythraea were possible, and the qPCR product retrieved from unit 2 was related to Myxococcus geoA. These results suggest that our qPCR primers can cover a broad range of geoA sequences. Among the 11 actinomycete strains, putative tpc were detected in four of them, but only 2 strains were producing MIB in the culture conditions used. These two, strains AMU11 and AMU14, were the ones retrieved from the RAS units. However, MIB was not detected in unit 1 and the detection of this compound in unit 3 was not realized. Some environmental or chemical factors can affect the production of MIB by these strains such as the temperature (Blevins et al., 1995). This study is the first to apply qPCR assays to detect and quantify the geosmin synthesis gene ( geoA) in RAS. The possibility of having more than one geoA sequences in RAS
could have had an impact on its quantification. Using a mixture of four standards comprising PCR products from different geoA sequences, we showed that the amplification efficiency of the primers was within the prescribed values (0.9 < E < 1.1) with linearity (r2 0.99) ranging from 102 to 108 copies/reaction. These criteria were similar to those found in qPCR studies on functional genes involved in denitrifying process (Geets et al., 2007), hydrocarbon degradation (Powell et al., 2006) or the detection of aerobic arsenite oxidizers (Que´me´neur et al., 2010). To our knowledge, this study is the first to use a mix of homologous genes as standards with one set of primers. By comparing the qPCR method with another quantification method, such as the copy number of genomes, we showed that these two methods generated very similar values, in the same order of magnitude. Furthermore, we verified that the presence of PCR inhibitory compounds like humic acids could affect the qPCR efficiency (Schriewer et al., 2011). The presence of environmental biomass such as a water sample from the sediment trap of a RAS having a high organic load did not have a negative impact on the geoA quantification in strain iafH cultures. This values was however twice lower in strain AMU14 cultures. All these results suggest that our qPCR protocol was reliable for the quantification of different geoA sequences, and variations could occur but in the same order of magnitude. Detection of geosmin producers is normally based on chemical analysis (SPMEeGCeMS) to confirm the presence of this compound in the fish flesh (Hurlburt et al., 2009). However, chemical analysis could not be used to prevent the occurrence of off-flavours, because the detection of geosmin
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 5 3 e6 7 6 2
in fish flesh means that geosmin producers are already well established in the RAS. Culture methods with specific media for actinomycetes can also be used, but it can be cumbersome and not reliable, especially for actinomycetes (Rintala and Nevalainen, 2006). Using our qPCR method, we were able to detect and quantify geoA in the units 1 and 2 where geosmin was detected. No PCR amplification was observed in the nonproducing geosmin unit 4.
5.
Conclusions
New geoA-related gene sequences were retrieved from geosmin-producing actinomycete strains. Combined with geoA-related sequences available in gene databases, primers and standards suitable for qPCR assays targeting different Streptomyces geoA were designed. The qPCR protocol we developed succeeded in the quantification of geoA gene copies in two geosmin-producing RAS. The advantage of using qPCR is that sampling can be done throughout the RAS. Geosmin is known to be a secondary metabolite produced during the development of aerial hyphae on agar plate or just before the stationary phase of growth in liquid medium (Bibb, 2005). Therefore, quantification of geoA early during Streptomyces development could permit the detection of geosmin producers prior to the occurrence of geosmin production in RAS. This information will be most valuable for fish producers to manage further development of such bacteria. This qPCR protocol can also be used as a tool to assist in the development of adequate operating procedures for RAS to avoid the development of off-flavour events.
Acknowledgements This work was supported by a Strategic Projects grant from the Natural Sciences and Engineering Research Council of Canada, the Re´seau Aquaculture Que´bec and the Socie´te´ de recherche et de de´veloppement en aquaculture continentale inc. We are very grateful to Franc¸ois Shareck for providing the streptomycetes iaf strains and advices on Streptomyces culture conditions. We thank Jean-Christophe Therrien and Karla Vazquez for technical supports.
Appendix. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.watres.2011.10.020.
references
Bentley, S.D., Chater, K.F., Cerden˜o-Ta´rraga, A.M., Challis, G.L., Thomson, N.R., James, K.D., Harris, D.E., Quail, M.A., Kieser, H., Harper, D., Bateman, A., Brown, S., Chandra, G., Chen, C.W., Collins, M., Cronin, A., Fraser, A., Goble, A., Hidalgo, J., Hornsby, T., Howarth, S., Huang, C.H., Kieser, T., Larke, L.,
6761
Murphy, L., Oliver, K., O’Neil, S., Rabbinowitsch, E., Rajandream, M.A., Rutherford, K., Rutter, S., Seeger, K., Saunders, D., Sharp, S., Squares, R., Squares, S., Taylor, K., Warren, T., Wietzorrek, A., Woodward, J., Barrell, B.G., Parkhill, J., Hopwood, D.A., 2002. Complete genome sequence of the model actinomycete Streptomyces coelicolor A3(2). Nature 417, 141e147. Bibb, M.J., 2005. Regulation of secondary metabolism in streptomycetes. Current Opinion in Microbiology 8, 208e215. Blevins, W.T., Schrader, K.K., Saadoun, L., 1995. Comparative physiology of geosmin production by Streptomyces Halstedii and Anabaena sp. Water Science and Technology 31 (11), 127e133. Cane, D.E., Watt, R.M., 2003. Expression and mechanistic analysis of a germacradienol synthase from Streptomyces coelicolor implicated in geosmin biosynthesis. Proceedings of the National Academy of Sciences of the United States of America 100, 1547e1551. Cane, D.E., He, X., Kobayashi, S., Omura, S., Ikeda, H., 2006. Geosmin biosynthesis in Streptomyces avermitilis, molecular cloning, expression, and mechanistic study of the germacradienol/geosmin synthase. Journal of Antibiotics 59 (8), 471e479. Cook, D., Newcombe, G., Sztajnbok, P., 2001. The application of powdered activated carbon for 2-MIB and geosmin removal: predicting PAC doses in four raw waters. Water Research 35, 1325e1333. Dickschat, J.S., Bode, H.B., Mahmud, T., Muller, R., Schulz, S., 2005. A novel type of geosmin biosynthesis in myxobacteria. Journal of Organic Chemistry 70, 5174e5182. Edwards, U., Rogall, T., Blo¨cker, H., Emde, M., Bo¨ttger, E.C., 1989. Isolation and direct complete nucleotide determination of entire genes. Characterization of a gene coding for 16S ribosomal RNA. Nucleic Acids Research 17 (19), 7843e7853. Franck, J.A., Reich, C.I., Sharma, S., Weisbaum, J.S., Wilson, B.A., Olsen, G.J., 2008. Critical evaluation of two primers commonly used for amplification of bacterial 16S rRNA genes. Applied and Environmental Microbiology 74 (8), 2461e2470. Geets, J., de Cooman, M., Wittebolle, L., Heylen, K., Vanparys, B., De Vos, P., Verstraete, W., Boon, N., 2007. Real-time PCR assay for the simultaneous quantification of nitrifying and denitrifying bacteria in activated sludge. Applied Microbiology and Biotechnology 75, 211e221. Ghimire, G.P., Oh, T.J., Lee, H.C., Kim, B.G., Sohng, J.K., 2008. Cloning and functional characterization of the germacradienol synthase (spterp13) from Streptomyces peucetius ATCC 27952. Journal of Microbiology and Biotechnology 18, 1216e1220. Giglio, S., Jiang, J., Saint, C.P., Cane, D.E., Monis, P.T., 2008. Isolation and characterization of the gene associated with geosmin production in cyanobacteria. Environmental Science & Technology 42 (21), 8027e8232. Giglio, S., Chou, W.K.W., Ikeda, H., Cane, D.E., Monis, P.T., 2011. Biosynthesis of 2-methylisoborneol in Cyanobacteria. Environmental Science & Technology 45 (3), 992e998. Guttman, L., van Rijn, J., 2008. Identification of conditions underlying production of geosmin and 2-methylisoborneol in a recirculating system. Aquaculture 279, 85e91. Hanson, T.R., 2003. Economic Impact of Off-flavor to the U.S. Catfish Industry. American Chemical Society (Chapter 2), pp. 13e29. Howgate, P., 2004. Tainting of farmed fish by geosmin and 2-methyl-iso-borneol: a review of sensory aspects and of uptake/depuration. Aquaculture 234, 155e181. Hurlburt, B., Lloyd, S.W., Grimm, C.C., 2009. Comparison of analytical technique for detection of geosmin and 2-methylisoborneol in aqueous samples. Journal of Chromatographic Science 47 (8), 670e673.
6762
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 5 3 e6 7 6 2
Izaguirre, G., Taylor, W.D., 1998. A Pseudanabaena species from Castaic lake California, that produces 2-Methylisoborneol. Water Research 32 (5), 1673e1677. Jiang, J., He, X., Cane, D.E., 2007. Biosynthesis of the earthy odorant geosmin by a bifunctional Streptomyces coelicolor enzyme. Nature Chemical Biology 3, 711e715. Klausen, C., Nicolaisen, M.H., Strobel, B.W., Warnecke, F., Nielsen, J.L., Jørgensen, N.O.G., 2005. Abundance of actinobacteria and production of geosmin and 2methylisoborneol in Danish streams and fish ponds. Applied and Environmental Microbiology 52, 265e278. Komatsu, M., Tsuda, M., Omura, S., Oikawa, H., Ikeda, H., 2008. Identification and functional analysis of genes controlling biosynthesis of 2-methylisoborneol. Proceedings of the National Academy of Sciences of the United States of America 105, 7422e7427. Lloyd, S.W., Lea, J.M., Zimba, P.V., Grimm, C.C., 1998. Rapid analysis of geosmin and 2-methylisoborneol in water using solid-phase micro-extraction procedures. Water Research 32 (7), 2140e2146. Ludwig, F., Medger, A., Bornick, H., Opitz, M., Lang, K., Gottfert, M., Roske, I., 2007. Identification and expression analyses of putative sesquiterpene synthase genes in Phormidium sp. and prevalence of geoA-like genes in a drinking water reservoir. Applied and Environmental Microbiology 73, 6988e6993. Ohnishi, Y., Ishikawa, J., Hara, H., Suzuki, H., Ikenoya, M., Ikeda, H., Yamashita, A., Hattori, M., Horinouchi, S., 2008. Genome sequence of the streptomycin-producing microorganism Streptomyces griseus IFO 13350. Journal of Bacteriology 190, 4050e4060. Omura, S., Ikeda, H., Ishikawa, J., Hanamoto, A., Takahaski, C., Shinose, M., Takahaski, Y., Horikawa, H., Nakazawa, H., Osonoe, T., Kikuchi, H., Shiba, T., Sakaki, Y., Hattori, M., 2001. Genome sequence of an industrial microorganism Streptomyces avermitilis: deducing the ability of producing secondary metabolites. Proceedings of the National Academy of Sciences of the United States of America 98, 12215e12220. Page´, N., Kluepfel, D., Shareck, F., Morosoli, R., 1996. Effect of signal peptide alteration and replacement on export of xylanase A in Streptomyces lividans. Applied and Environmental Microbiology 62, 109e114. Powell, S.M., Ferguson, S.H., Bowman, J.P., Snape, I., 2006. Using real-time PCR to assess changes in the hydrocarbon-degrading microbial community in Antarctic soil during bioremediation. Microbial Ecology 52, 523e532. Que´me´neur, M., Ce´bron, A., Billard, P., Battaglia-Brunet, F., Garrido, F., Leyval, C., Joulian, C., 2010. Population structure
and abundance of arsenite-oxidizing bacteria along an arsenic pollution gradient in waters of the Upper Isle river basin, France. Applied and Environmental Microbiology 76 (13), 4566e4570. Rintala, H., Nevalainen, A., 2006. Quantitative measurement of streptomycetes using real-time PCR. Journal of Environmental Monitoring 8, 745e749. Schrader, K.K., Summerfelt, S.T., 2010. Distribution of offflavor compounds and isolation of geosmin-producing bacteria in a series of water recirculating systems for rainbow trout culture. North American Journal of Aquaculture 72, 1e9. Schriewer, A., Wehlmann, A., Wuertz, S., 2011. Improving qPCR efficiency in environmental samples by selective removal of humic acids with DAX-8. Journal of Microbiological Methods 85 (1), 16e21. Serghini, M.A., Ritzenthaler, C., Pinck, L., 1989. A rapid and efficient ‘miniprep’ for isolation of plasmid DNA. Nucleic Acids Research 17, 3604. Smith, V.H., Sieber-Denlinger, J., deNoyelles, F., Campbell, S., Pan, S., Randtke, S.J., Blain, G.T., Strasser, V.A., 2002. Managing taste and odor problems in a eutrophic drinking water reservoir. Lake and Reservoir Management 18, 319e323. Smith, J.L., Boyer, G.L., Zimba, P.V., 2008. A review of cyanobacterial odorous and bioactive metabolites: impacts and management alternatives in aquaculture. Aquaculture 280, 5e20. Suffet, I.H., Corado, A., Chou, D., Butterwoth, S., McGuire, M.J., 1996. AWWA survey of taste and odor. Journal of American Water Works Association 88, 168e180. Thompson, J.D., Higgins, D.G., Gibson, T.J., 1994. CLUSTAL W: improving the sensitivity of progressive multiple sequence alignment through sequence weighting position specific gap penalties and weight matrix choice. Nucleic Acids Research 22, 4673e4680. Wang, C.M., Cane, D.E., 2008. Biochemistry and molecular genetics of the biosynthesis of the earthy odorant methylisoborneol in Streptomyces coelicolor. Journal of the American Chemical Society 130 (28), 8908e8909. Wang, Z., Xu, Y., Shao, J., Wang, J., Renhui, L., 2011. Genes associated with 2-methylisoborneol biosynthesis in cyanobacteria: isolation, characterization, and expression in response to light. PloS ONE 6 (4), e18665. Zaitlin, B., Watson, S.B., Dixon, J., Steel, D., 2003. Actinomycetes in the Elbow River basin, Alberta, Canada. Water Quality Research Journal of Canada 38, 115e125.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 6 3 e6 7 6 9
Available online at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Relationships between sand and water quality at recreational beaches Matthew C. Phillips a,b, Helena M. Solo-Gabriele a,b,*, Alan M. Piggot a,d, James S. Klaus a,c, Yifan Zhang a,b a
University of Miami, NSF NIEHS Oceans and Human Health Center, Miami, FL 33149, USA Department of Civil, Arch. and Environmental Engineering, University of Miami, Coral Gables, FL 33146, USA c Department of Geological Sciences, University of Miami, Coral Gables, FL 33146, USA d Division of Marine Geology and Geophysics, Rosenstiel School of Marine and Atmospheric Sciences, University of Miami, Miami, FL 33149, USA b
article info
abstract
Article history:
Enterococci are used to assess the risk of negative human health impacts from recreational
Received 30 June 2011
waters. Studies have shown sustained populations of enterococci within sediments of
Received in revised form
beaches but comprehensive surveys of multiple tidal zones on beaches in a regional area
12 October 2011
and their relationship to beach management decisions are limited. We sampled three tidal
Accepted 17 October 2011
zones on eight South Florida beaches in Miami-Dade and Broward counties and found that
Available online 25 October 2011
enterococci were ubiquitous within South Florida beach sands although their levels varied greatly both among the beaches and between the supratidal, intertidal and subtidal zones.
Keywords:
The supratidal sands consistently had significantly higher ( p < 0.003) levels of enterococci
Enterococci
(average 40 CFU/g dry sand) than the other two zones. Levels of enterococci within the
Beach sand
subtidal sand correlated with the average level of enterococci in the water (CFU/100mL) for
Marine water
the season during which samples were collected (rs ¼ 0.73). The average sand enterococci content over all the zones on each beach correlated with the average water enterococci levels of the year prior to sand samplings (rs ¼ 0.64) as well as the average water enterococci levels for the month after sand samplings (rs ¼ 0.54). Results indicate a connection between levels of enterococci in beach water and sands throughout South Florida’s beaches and suggest that the sands are one of the predominant reservoirs of enterococci impacting beach water quality. As a result, beaches with lower levels of enterococci in the sand had fewer exceedences relative to beaches with higher levels of sand enterococci. More research should focus on evaluating beach sand quality as a means to predict and regulate marine recreational water quality. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
Enterococci are recommended by the Environmental Protection Agency (EPA) for use in assessing the health risk of
recreational waters (U.S. EPA, 1986). While enterococci are not commonly pathogenic, they have been shown in past epidemiological studies to correlate with adverse human health effects. These negative health effects are commonly
* Corresponding author. University of Miami, Department of Civil, Arch. and Environmental Engineering, P.O. Box 248294, Coral Gables, Florida 33124-0630, USA. Tel.: þ1 305 284 2908; fax: þ1 305 284 3492. E-mail address:
[email protected] (H.M. Solo-Gabriele). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.10.028
6764
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 6 3 e6 7 6 9
associated with pathogens found in sewage and thus enterococci are used to protect human health in waters thought to be impacted by fecal pollution (Cabelli et al., 1982; Dufour, 1984; Wade et al., 2003). In recent years however, researchers have discovered that beach sediments can sustain populations of enterococci and are potential non-sewage sources of these indicator bacteria in recreational waters (Shibata et al., 2004; Whitman and Nevers, 2003; Badgley et al., 2010; Wright et al., 2011; Abdelzaher et al., 2010). Even without a point source of sewage, high enterococci levels in beach water may still represent an increased risk for bathers at these beaches (Fleisher et al., 2010; Sinigalliano et al., 2010) and enterococci levels in sand have been shown to correlate with higher levels of human pathogens in beach sand (Shah et al., 2011). Due to this risk to human health, it is important to assess the beach sediment’s affect on the overall quality of the beach and associated recreational water. Regulatory agencies, such as the State Departments of Health (DOH), use water quality measurements to assess the human health risk of recreational swimming at local beaches. In Miami-Dade and Broward counties, Florida, the DOH collects weekly water samples from waist deep water at each recreational beach and analyzes these samples for enterococci in the water. Beach advisories are posted to alert swimmers of increased health risks when the level of bacteria in the water for two consecutive samples are measured above a maximum of 104 colony forming units
(CFU) per 100 mL of water or if the geometric mean for the past 5 water samples is above 35 CFUs/100 mL (U.S. EPA, 1986). The present study set out to compare results from regulatory water quality monitoring with the enterococci levels within the sediments of 8 South Florida Beaches and to evaluate if sediment levels of enterococci have a potential for predictive regulatory measures. Two water quality measures were evaluated; one that utilized the percent exceedances of the regulatory maximum value and another that utilized the actual measured levels of enterococci in water. In the process of comparing the results for the 8 beaches, the study also established the ubiquity of enterococci within the beach sands of South Florida’s beaches and characterized the spatial distribution of enterococci across the supratidal, intertidal and subtidal zones.
2.
Methods
2.1.
Site descriptions and sampling
Sediment samples were collected from 8 South Florida Beaches (Fig. 1; Supplemental Table 1). Beaches were chosen based on where the DOH conducts their weekly sampling and to represent a wide range of both beach conditions and DOH recorded levels of enterococci present in the water. Seven of the eight
Fig. 1 e Beach sampling locations and bacterial levels in beach sand on September 11, 2010 and February 12, 2011. Background map from Bing Maps.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 6 3 e6 7 6 9
beaches chosen for this study are located in Miami-Dade County and one (Hallandale Beach) is located in Broward County. Two of the eight are located in Biscayne Bay (Hobie Cat Beach and Matheson Hammock) thereby experiencing weaker wave conditions. Matheson Hammock beach is located within an artificial lagoon with limited inflow of bay water providing for particularly calm water conditions. The remaining six beaches face the Atlantic Ocean with larger wave exposure and more dilution and flushing of its shoreline. The primary sources of indicator bacteria to these beaches are diffuse (i.e., nonpoint sources). All beaches experience significant human traffic during warmer weather with the highest usage occurring during weekends and holidays (Wang et al., 2010). Beach usage declines during the winter months and increases again in April. Gulls and other seabirds are present at all beaches especially during the fall and winter months. Hobie Cat Beach is the only beach where dogs are allowed. Of the beaches sampled, Cape Florida, Matheson Hammock and Crandon Park are run by the state park system. The other beaches are public and maintained by local jurisdictions. The three beaches run by the state parks are gated and require visitors to pay for entrance into the park and thus these beaches tend to be better maintained. Although sewage discharge pipes are located offshore north of Hallandale Beach and east of Hobie Cat Beach, extensive studies have shown that these discharges do not impact the local beaches (U.S. EPA, 2003). Sand sampling locations were located along transects consistent with DOH water sampling locations. Samples were collected at low tide during two different sampling events, the morning of September 11th, 2010 and the morning of February 12th, 2011. Three composite samples (one from each of the following zones: supratidal, intertidal, and subtidal) were collected at each beach. Supratidal samples were located at 0.25 m above the wrack (mean high tide) line, the intertidal samples were located halfway between the wrack line and water line and the subtidal samples were located below the water line in 30 cm deep water. Each composite sample consisted of a series of 31 sediment cores (2.54 cm, 4 cm deep) that were collected every 20 cm along a 6-meter transect parallel to the shoreline. Upon collection the 31 cores were placed into a sterile tin and covered for transport. In an effort to document physicalechemical parameters (pH, salinity, turbidity) for each beach site, a water sample was also collected upstream of the subtidal transect using a sterile Whirl-Pak bag. Water temperature was documented using a multi parameter probe (YSI 650MDS, YSI Incorporated, Yellow Springs, Ohio).
2.2.
Laboratory analysis of sediment samples
Laboratory analysis was conducted within 4 h of sample collection. Each tin containing the composite sand sample was mixed thoroughly using sterile spoons for 3 min prior to analysis. Triplicate aliquots (20 g) of each homogenized sand sample were analyzed gravimetrically (dried for 24 h at 110 C) to obtain the moisture content. Grain size analysis was conducted utilizing the wet sieve method (Alekseeva and Sval’nov, 2005). Data from the grain size analysis was used to extrapolate total surface area of the sediment samples and D50 (diameter at which 50% of sand grains by weight are finer). See Supplemental text for details.
6765
In order to extract microbes from the sand grains, approximately 10 g of sand was aseptically transferred into a sterile bottle containing 100 mL of sterile phosphate buffered saline solution and shaken vigorously for 2 min (Shibata et al., 2004; Boehm et al., 2009). After allowing the sediment to settle (2 min), two volumes of the eluent (25ml and 3ml) were filtered (Pall, GN-6 Grid) and placed on mEI agar as per Method 1600 (U.S. EPA, 2006).
2.3.
Enterococci water monitoring data
Enterococci beach water monitoring data was provided by the Florida Department of Health as part of the Florida Healthy Beaches program (http://esetappsDOH.DOH.state.fl.us/ irm00beachwater/default.aspx) which implements weekly monitoring for regulatory purposes. Only the DOH’s routine weekly samples were used in calculating water enterococci levels. Daily supplementary samples taken after exceedances were excluded to avoid skewing the averages with multiple high values. The regulatory water quality data was then averaged over set time periods (yearly, seasonally, monthly, and based on rainfall) to minimize the impacts of spatial and temporal variability inherent in single water samples (Boehm et al., 2002). Yearly averaging corresponded to the year prior to each of the two sand sampling events at all beaches except Hobie Cat Beach. Hobie Cat Beach was closed for sand renourishment between October 12th, 2009 and May 31st, 2010. During this time sand on the beach was replaced with exogenous sand, and so only beach water monitoring samples taken after the beach reopened were included when evaluating the data from Hobie Cat Beach. For the seasonal averaging, the September sampling data was paired with DOH water enterococci measurements taken during wet season in south Florida (May, 2010eOctober, 2010) and the February data was paired with DOH water samples taken during the dry season (November, 2010eApril, 2011). Monitoring data for one month after each sand sampling event was also used to analyze the possible predictive capacity of the beach sand. Averaging based upon rainfall corresponded to averaging water quality values through an amount of time needed to acquire a certain amount of rainfall.
2.4.
Data analysis
All statistical analysis was performed using Microsoft Excel, XL STAT (Addinsoft USA, New York, NY). Averages from 78 DOH water samples (September 14, 2009 to March 29, 2011) and 6 sand samples per beach (September 11, 2010 and February 12, 2011) were compared and correlations were acquired using Spearman’s Rank Order Test (rs). Comparisons between sand tidal zones and sampling days were made using a single variable ANOVA. An alpha of 0.05 was used for all tests.
3.
Results and discussion
3.1.
Variation between sand tidal zones
Enterococci were found at every beach sampled during this study (Fig. 2) which is consistent with other studies that have
6766
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 6 3 e6 7 6 9
Fig. 2 e Levels of enterococci in all beach zones and the 1 year prior average of enterococci in the water. Levels of enterococci within the water varied greatly over the year prior to sampling at all beaches. Error bars represent one standard deviation.
quantified fecal indicator bacteria in beach sands (Alm et al., 2003; Whitman et al., 2006; Halliday and Gast, 2011), Moreover, enterococci levels were consistently higher in the supratidal sand compared to the intertidal sand and subtidal sand (Fig. 3). This was particularly evident at sites that were characterized by higher levels of enterococci. The higher level in the supratidal sand was significant when considering results from both sampling days ( p < 0.02) but was more pronounced for the September sampling day ( p < 0.05) as compared to the February sampling day ( p ¼ 0.08). Although higher levels of enterococci were observed in the supratidal sand these levels did not correlate with any grain size parameters of the sands including surface area (28.0e83.2 cm2/g), the diameter at which 50% of the sand grains are finer (0.32e0.89 mm) and uniformity coefficient, a relative measure of similarity in the sizes of sand grains (1.69e5.45) (1 is perfectly uniform, the higher the number the less uniform the sand grains). Thus differences in grain size characteristics among the different beaches did not appear to impact enterococci levels. A negative correlation was observed between moisture content and log normalized enterococci for all beach sands (rs ¼ 0.52, p < 0.0005) which is consistent with past studies of Hobie Cat Beach (Shah et al., 2011). This trend was not noted in sands within the same tidal zones. The elevated levels of enterococci within the supratidal zone may be due to the combined effects of proximity to on-shore non-point sources of enterococci such as humans, birds, leachate from garbage cans, run off after storms, lack of tidal washing and regrowth (Elmir et al., 2007; Bonilla et al., 2007; Abdelzaher et al., 2011; Hartz et al., 2008).
Differences in these factors could account for the variability seen between the 8 beaches sampled during this study. Since the supratidal sands are not often exposed to the water, it is unlikely that the enterococci could have been deposited there from the ocean. The supratidal sand is also not exposed to the same wave energy and tidal forces experienced by the sands
Fig. 3 e Average enterococci levels of south Florida beach sand. Results indicate that supratidal sand is consistently and significantly (P < 0.01) higher than both the intertidal and subtidal sand while there was no significant difference between intertidal and subtidal sands. Results showed very high levels of variability between the 8 beaches. The 52.3 CFU/g value was a statistical outlier of the intertidal samples during the February sampling. Error bars represent one standard deviation.
6767
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 6 3 e6 7 6 9
Table 1 e Summary of correlations between DOH water quality data and levels of enterococci within the sand. Correlated with
rs
P
Average sand ENT across all tidal zones ENT in supratidal sand ENT in subtidal sand Average sand ENT across all tidal zones ENT in supratidal sand ENT in subtidal sand Average sand ENT across all tidal zones ENT in supratidal sand ENT in subtidal sand
0.5 0.48 0.56 0.73 0.73 0.58 0.38 0.27 0.83
0.05 0.06 0.03 0.05 0.05 0.15 0.36 0.54 0.02
DOH water quality measurements Percentage of prior year exceedences (Both)
Percentage of prior year exceedences (Feb)
Percentage of prior year exceedences (Sept)
in the intertidal and subtidal zones. Thus intertidal and subtidal sands would have lower levels of enterococci because the energy from the waves and tides is capable of removing the on-shore sources of enterococci that deposit in the sand (Yamahara et al., 2007; Ge et al., 2010).
3.2. Correlations with DOH water quality measurements Both water quality measures evaluated (average levels over a pre-described period of time and percent exceedences), showed significant correlations with sand enterococci levels. The prior year average DOH water enterococci levels correlated with the average level of sand enterococci across all tidal zones (rs ¼ 0.64, p ¼ 0.008). The average DOH water enterococci levels over the prior year also positively correlated with levels of enterococci in supratidal sand (rs ¼ 0.6, p ¼ 0.015) (Figure S2). The levels of enterococci in the subtidal sand also correlated (rs ¼ 0.69, p ¼ 0.004) with the average levels of DOH water enterococci for the season during which the samples were collected. The percent of DOH samples from the prior year that exceeded 104 CFU/100 mL positively correlated with the average level of sand enterococci across all tidal zones for all sampling days (rs ¼ 0.5, p ¼ 0.05). This was primarily driven by the significant correlation of percent exceedance to levels of enterococci in the subtidal sand (rs ¼ 0.56, p ¼ 0.03) and the non-significant correlation (rs ¼ 0.48, p ¼ 0.06) to levels in the supratidal sand. This trend between the percent exceedance and sand enterococci levels was noted for the February sampling (rs ¼ 0.73, p ¼ 0.05) but not for the September sampling (rs ¼ 0.38, p ¼ 0.36). The lack of correlation in the September data set was primarily due to the supratidal sand not correlating with percent exceedance (rs ¼ 0.27, p ¼ 0.54), although the subtidal sand did show a correlation (rs ¼ 0.83, p ¼ 0.02). For the February data set, both subtidal and supratidal sand levels showed relatively high correlations (rs > 0.5) (although subtidal was not significant) (supratidal: rs ¼ 0.73, p ¼ 0.05; subtidal: rs ¼ 0.58, p ¼ 0.15). All correlations are summarized in Table 1 and raw data is available in Table S-1, located in the supplemental text. The heavy rains during and prior to September could possibly be responsible for the lack of correlation seen in the supratidal sands because of more frequent sand washing. Also the heavier beach traffic could have raised levels in the supratidal sand (Elmir et al., 2007) relatively quickly, in comparison to the levels observed in the
water during the prior year. There were no trends seen with enterococci levels in intertidal sands, possibly because of wave action and tidal movements making this tidal zone more variable and inconsistent with ambient enterococci levels in the water (Yamahara et al., 2007; Abdelzaher et al., 2010). At beaches where the predominant source of enterococci is thought to be the sand; the potential exists to use levels of enterococci in the sand as guidance for management decisions such as where and how often to sample. In this study, water enterococci for the two weeks following the sand samplings positively correlated (rs ¼ 0.63, p ¼ 0.01) with enterococci levels in the subtidal sand. The water enterococci for the month following the sand samplings also positively correlated (rs ¼ 0.54, p ¼ 0.04) with the average enterococci over all the tidal zones on the beach. Correlations could also be seen by setting averaging times equal to the amount of time needed to acquire a certain amount of rainfall. For example, the average enterococci in the water, for a time period after sand sampling corresponding to 1.25 cm of rain at the beach, correlated (rs ¼ 0.56, p ¼ 0.05) with the levels of bacteria in the subtidal sand for all sampling days. This amount was also significant for the February sampling (rs ¼ 0.85, p ¼ 0.02) and was strengthened if the time series was lengthened until the beach received 2.54 cm of rainfall (rs ¼ 0.96, p ¼ 0.003). While these correlations are not enough to encourage beach closures based solely on sand sampling, they suggest that sand sampling could be used to allocate monitoring resources to focus on non-point source beaches with higher likelihoods of exceedences. Instead of sampling all beaches at an equal frequency (example: once per week), beaches with low levels of sand enterococci could be sampled less frequently (example: once per month) as opposed to beaches with high levels of enterococci within the sand (example: twice per week). Further studies should be conducted to confirm results and evaluate the geographic extent of these potential trends.
4.
Conclusions
Enterococci were found at all 8 of the South Florida beaches sampled. They were also consistently found at higher levels in the supratidal sand than in other zones on the beach. This suggests a constant, on-shore source of enterococci throughout South Florida (Zhu et al., 2011). Percent exceedance and average levels of enterococci in the water can be looked at as relative measures of the overall
6768
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 6 3 e6 7 6 9
health of the recreational water. This would mean that the overall health of recreational waters in south Florida was correlated with the levels of enterococci present in the beach sands. One of most important implications from this study was that the average quality of the beach waters and the quality of the beach sands seem to be related. Beaches with low levels of enterococci in the sand generally have lower levels in the water and thus will have fewer closures than beaches with higher levels of enterococci in the beach sands. Since beach sands represent a more time-averaged representation of enterococci inputs to the beach environment, they might be more likely to reflect human health risks for specific beaches. Utilizing beach sand composite samples could help account for the inherent variability in enterococci levels of single water samples and help avoid needless beach closures because of this variability. Furthermore, sand samples could be used to focus beach monitoring efforts towards high-risk beaches with a higher likelihood of exceedences.
5.
Recommendations
The correlations found in this study should be further validated by evaluating the mechanisms by which enterococci in the beach sands are transported into the water column (Phillips et al., 2011). Both the horizontal shear forces of waves and the vertical pumping action of tides should be examined for a better understanding of how enterococci enter into recreational waters. Further studies should be conducted to evaluate the causes of elevated enterococci levels at certain beaches and between tidal zones. The different sources of enterococci should be examined (number of human visitors, types and number of animals, practices for waste collection and disposal) and the characteristics of the different beach sediments should also be documented (moisture content, grain size, levels of biofilm, composition of sand grains and volatile organics). This data should be used to establish possible correlations between causative parameters to levels of enterococci. Controlled laboratory experiments with microcosms could validate these correlations. Further sampling should also be conducted over longer periods of time to further corroborate the trends noticed in this study. Expanding to more beaches in other areas of the country would also assist in determining if the levels of enterococci in the sediment actually can be used as a predictive tool to guide beach management decisions.
Acknowledgments We would like to thank those who participated in the laboratory analyses including Amber Enns, Laura Vogel, Amir Abdelzaher, David Hernandez, Malia Carpio, Sara Johnson, Rafael Hernandez, Gabriela Toledo, Kieran Swart, Scott Hawley, Noha Abdel-Mottaleb and Nick Bill. Funding for this project was received through the NSF REU program and through the National Science Foundation (NSF) and the
National Institute of Environmental Health Sciences (NIEHS) Oceans and Human Health Center at the University of Miami Rosenstiel School [NSF 0CE0432368/0911373] and [NIEHS P50 ES12736].
Appendix. Supplementary material Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.watres.2011.10.028.
references
Abdelzaher, A.M., Wright, M.E., Ortega, C., Hasan, A.R., Shibata, T., Solo-Gabriele, H.M., Kish, J., Withum, K., He, G., Elmir, S.M., Bonilla, J.A., Bonilla, T.D., Palmer, C.J., Scott, T.M., Lukasik, J., Harwood, V.J., McQuaig, S., Sinigalliano, C.D., Gidley, M.L., Wanless, D., Plano, L.R.W., Garza, A.C., Zhu, X., Stewart, J.R., Dickerson, J.W., Yampara-Iquise, H., Carson, C., Fleisher, J.M., Fleming, L.E., 2011. Daily measures of microbes and human health at a non-point source marine beach. Journal of Water and Health 9 (3), 443e457. doi:10.2166/wh.2011.146. Abdelzaher, A.M., Wright, M.E., Ortega, C., Solo-Gabriele, H.M., Miller, G., Elmir, S., Newman, X., Shih, P., Bonilla, J.A., Bonilla, T.D., Palmer, C.J., Scott, T., Lukasik, J., Harwood, V.J., McQuaig, S., Sinigalliano, C., Gidley, M., Plano, L.W.R., Zhu, X., Wang, J.D., Fleming, L.E., 2010. Presence of pathogens and indicator microbes at a non-point source subtropical recreational marine beach. Applied and Environmental Microbiology 76 (3), 724e732. doi:10.1128/AEM.02127-09. Alekseeva, T.N., Sval’nov, V.N., 2005. The refined wet sieving method for the analysis of fine-graded sediments. Lithology and Mineral Resources 40 (6), 564e576. Alm, E.W., Burke, J., Spain, A., 2003. Fecal indicator bacteria are abundant in wet sand at freshwater beaches. Water Research 37 (16), 3978e3982. Badgley, B.D., Nayak, B.S., Harwood, V.J., 2010. The importance of sediment and submerged aquatic vegetation as potential habitats for persistent strains of enterococci in a subtropical watershed. Water Research 44 (20), 5857e5866. doi:10.1016/j. watres.2010.07.005. Boehm, A.B., Grant, S.B., Kim, J.H., Mowbray, S.L., McGee, C.D., Clark, C.D., Foley, D.M., Wellman, D.E., 2002. Decadal and shorter period variability of surf-zone water quality at Huntington beach, California. Environmental Science & Technology 36 (18), 3885e3892. doi:10.1021/es020524u. Boehm, A.B., Griffith, J., McGee, C., Edge, T.A., Solo-Gabriele, H.M., Whitman, R., Cao, Y., Getrich, M., Jay, J.A., Ferguson, D., Goodwin, K.D., Lee, C., Madison, M., Weisberg, S.B., 2009. Faecal indicator bacteria enumeration in beach sand: a comparison study of extraction methods in medium to coarse sands. Journal of Applied Microbiology 107, 1740e1750. doi:10.1111/j.1365-2672.2009.04440.x. Bonilla, T.D., Nowosielski, K., Cuvelier, M., Hartza, A., Greenb, M., Esiobub, N., McCorquodalea, D.S., Fleisher, J.M., Rogerson, A., 2007. Prevalence and distribution of fecal indicator organisms in South Florida beach sand and preliminary assessment of health effects associated with beach sand exposure. Marine Pollution Bulletin 54 (9), 1472e1482. doi:10.1016/j.marpolbul. 2007.04.016. Cabelli, V.J., Dufour, A.P., McCabe, L.J., Levin, M.A., 1982. Swimming associated gastroenteritis and water quality. American Journal of Epidemiology 115 (4), 606e616. Dufour, A.P., 1984. Health Effects Criteria for Fresh Recreational Waters. EPA-600/1-84-004. Office of Research and
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 6 3 e6 7 6 9
Development, US Environmental Protection Agency, Cincinnati, OH. Elmir, S.M., Wright, M.E., Abdelzaher, A., Solo-Gabriele, H.M., Fleming, L.E., Miller, G., Rybolowik, M., Shih, M.P., Pillai, S.P., Cooper, J.A., Quaye, E.A., 2007. Quantitative evaluation of bacteria released by bathers in a marine water. Water Research 41 (1), 3e10. doi:10.1016/j.watres.2006.10.005. Fleisher, J.M., Fleming, L.E., Solo-Gabriele, H.M., Jonathan, K.K., Sinigalliano, C.D., Plano, L.R.W., Elmir, S.M., Wang, J.D., Withum, K., Shibata, T., Gidley, M.L., Abdelzaher, A.M., He, G., Ortega, C., Zhu, X., Wright, M., Hollenbeck, J., Backer, L.C., 2010. The BEACHES study: health effects and exposures from non-point source microbial contaminants in subtropical recreational marine waters. International Journal of Epidemiology 39 (5), 1291e1298. doi:10.1093/ije/dyq084. Ge, Z., Nevers, M.B., Schwab, D.J., Whitman, R.L., 2010. Coastal loading and transport of Escherichia coli at an embayed beach in lake Michigan. Environmental Science & Technology 44 (17), 6731e6737. Halliday, E., Gast, R.J., 2011. Bacteria in beach sands: an emerging challenge in protecting coastal water quality and bather health. Environmental Science & Technology 45 (2), 370e379. 2011 Jan 15. Hartz, A., Cuvelier, M., Nowosielski, K., Bonilla, T.D., Green, M., Esiobu, N., McCorquodale, D.S., Rogerson, A., 2008. Survival potential of Escherichia coli and Enterococci in subtropical beach sand: implications for water quality managers. Journal of Environmental Quality 37, 898e905. Phillips, M.C., Solo-Gabriele, H.M., Reniers, A.J.H.M., Wang, J.D., Kiger, R.T., Abdel-Mottaleb, N., 2011. Pore water transport of enterococci out of beach sediments. Marine Pollution Bulletin 62, 2293e2298. Shah, A.H., Abdelzaher, A.M., Phillips, M., Hernandez, R., SoloGabriele, H.M., Kish, J., Scorzetti, G., Fell, J.W., Diaz, M.R., Scott, T.M., Lukasik, J., Harwood, V.J., McQuaig, S., Sinigalliano, C.D., Gidley, M.L., Wanless, D., Ager, A., Lui, J., Stewart, J.R., Plano, L.R.W., Fleming, L.E., 2011. Indicator microbes correlate with pathogenic bacteria, yeasts, and helminthes in sand at a subtropical recreational beach site. Journal of Applied Microbiology 110, 1571e1583. Shibata, T., Solo-Gabriele, H.M., Fleming, L., Elmir, S., 2004. Monitoring marine recreational water quality using multiple microbial indicators in an urban tropical environment. Water Research 38, 3119e3131. doi:10.1016/j.watres.2004.04.044. Sinigalliano, C.D., Fleisher, J.M., Gidley, M.L., Solo-Gabriele, H.M., Shibata, T., Plano, L., Elmir, S.M., Wang, J.D., Wanless, D.,
6769
Bartowiak, J., Boiteau, R., Withum, K., Abdelzaher, A., He, G., Ortega, C., Zhu, X., Wright, M., Kish, J., Hollenbeck, J., Backer, L.C., Fleming, L.E., 2010. Traditional and molecular analyses for fecal indicator bacteria in non-point source subtropical recreational marine waters. Water Research 44 (13), 3763e3772. doi:10.1016/j.watres.2010.04.026. U.S. Environmental Protection Agency, 1986. Ambient Water Quality Criteria for Bacteria. EPA A440/5-84-002. U.S. EPA, Washington, DC. U.S. Environmental Protection Agency, 2003. Relative Risk Assessment of Management Options for Treated Wastewater in South Florida. EPA 816-R-03e010. U.S. EPA, Washington D.C. U.S. Environmental Protection Agency, 2006. Enterococci in Water by Membrane Filtration Using Membrane-Enterococcus Indoxyl-b-D-Glucoside Agar (MEI). EPA-821-R-06e009. U.S. EPA, Washington, DC. Wade, T.J., Pai, N., Eisenberg, J.N., Colford, J.M.J., 2003. Do U.S. environmental protection agency water quality guidelines for recreational waters prevent gastrointestinal illness? A systematic review and meta-analysis. Environmental Health Perspectives 111, 1102e1109. Wang, J.D., Solo-Gabriele, H.M., Abdelzaher, A.M., Fleming, L.E., 2010. Estimation of enterococci input from bathers and animals on a recreational beach using camera images. Marine Pollution Bulletin 60, 1270e1278. Whitman, R.L., Nevers, M.B., 2003. Foreshore sand as a source of Escherichia coli in nearshore water of a Lake Michigan beach. Applied and Environmental Microbiology 69 (9), 5555e5562. doi:10.1128/AEM.69.9.5555-5562.2003. Whitman, R.L., Nevers, M.B., Byappanahalli, M.N., 2006. Examination of the watershed-wide distribution of Escherichia coli along Southern Lake Michigan: an integrated approach. Applied and Environmental Microbiology 72 (11), 7301e7310. Wright, M.E., Solo-Gabriele, H.M., Abdelzaher, A.M., Elmir, S., Fleming, L.E., 2011. The inter-tidal zone is the geographic location of elevated concentrations of enterococci. Water Science and Technology 63 (3), 542e549. Yamahara, K.M., Layton, B.A., Santoro, A.E., Boehm, A.B., 2007. Beach sands along the California coast are diffuse sources of fecal bacteria to coastal waters. Environmental Science & Technology 41, 4515e4521. doi:10.1021/es062822n. Zhu, X., Wang, J.D., Solo-Gabriele, H.M., Fleming, L.E., Elmir, S., 2011. A microbial water quality model for recreational marine beaches. Water Research 45, 2985e2995.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 7 0 e6 7 8 0
Available online at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Evaluation of the occurrence and biodegradation of parabens and halogenated by-products in wastewater by accuratemass liquid chromatography-quadrupole-time-of-flight-mass spectrometry (LC-QTOF-MS) Iria Gonza´lez-Marin˜o, Jose´ Benito Quintana*, Isaac Rodrı´guez, Rafael Cela Department of Analytical Chemistry, Nutrition and Food Sciences, IIAA - Institute for Food Analysis and Research, University of Santiago de Compostela, 15782 Santiago de Compostela, Spain
article info
abstract
Article history:
An assessment of the sewage occurrence and biodegradability of seven parabens and three
Received 4 July 2011
halogenated derivatives of methyl paraben (MeP) is presented. Several wastewater samples
Received in revised form
were collected at three different wastewater treatment plants (WWTPs) during April and
30 September 2011
May 2010, concentrated by solid-phase extraction (SPE) and analysed by liquid
Accepted 17 October 2011
chromatography-electrospray-quadrupole-time-of-flight mass spectrometry (LC-QTOF-
Available online 25 October 2011
MS). The performance of the QTOF system proved to be comparable to triple-quadrupole instruments in terms of quantitative capabilities, with good linearity (R2 > 0.99 in the
Keywords:
5e500 ng mL1 range), repeatability (RSD < 5.6%) and LODs (0.3e4.0 ng L1 after SPE). MeP and
Parabens
n-propyl paraben (n-PrP) were the most frequently detected and the most abundant analytes
Halogenated by-products
in raw wastewater (0.3e10 mg L1), in accordance with the data displayed in the bibliography
Biodegradation
and reflecting their wider use in cosmetic formulations. Samples were also evaluated in
Water samples
search for potential halogenated by-products of parabens, formed as a result of their reac-
Accurate-mass mass spectrometry
tion with residual chlorine contained in tap water. Monochloro- and dichloro-methyl par-
Wastewater
aben (ClMeP and Cl2MeP) were found and quantified in raw wastewater at levels between 0.01 and 0.1 mg L1. Halogenated derivatives of n-PrP could not be quantified due to the lack of standards; nevertheless, the monochlorinated species (ClPrP) was identified in several samples from its accurate precursor and product ions mass/charge ratios (m/z). Removal efficiencies of parabens and MeP chlorinated by-products in WWTPs exceeded 90%, with the lowest percentages corresponding to the latter species. This trend was confirmed by an activated sludge biodegradation batch test, where non-halogenated parabens had half-lives lower than 4 days, whereas halogenated derivatives of MeP turned out to be more persistent, with up to 10 days of half-life in the case of dihalogenated derivatives. A further stability test performed with raw wastewater also showed that parabens degrade rapidly in real sewage, with half-lives lower than 10 h for n-butyl-paraben, while dihalogenated species again turned out to be more stable, with half-lives longer than a week. ª 2011 Elsevier Ltd. All rights reserved.
* Corresponding author. Tel.: þ34 881814290; fax: þ34 981595012. E-mail address:
[email protected] (J.B. Quintana). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.10.027
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 7 0 e6 7 8 0
1.
Introduction
Parabens, esters of 4-hydroxybenzoic acid, are extensively employed as preservatives not only in a wide range of personal care products (PCPs) such as tooth pastes, deodorants, beauty creams, bath gels and shampoos, but also in canned foods, beverages and pharmaceuticals (Meyer et al., 2007; Lundov et al., 2009). This extensive use has awakened the concern about their potential long-term effects on human health and, in fact, recent studies have suggested a possible relationship between them and breast cancer, presumably favoured by prolonged dermal expositions to paraben-containing deodorants (Darbre et al., 2004). Even though this hypothesis has not been fully proved and additional studies are needed to confirm their carcinogenicity, a new generation of paraben-free PCPs has emerged in the market recently. As in the case of many other personal care chemicals, these preservatives are continuously released in urban wastewater at relatively high levels (Lee et al., 2005; Canosa et al., 2006b; Gonza´lez-Marin˜o et al., 2009; Villaverde-de-Sa´a et al., 2010) and, despite being considerably removed during conventional sewage treatments (Lee et al., 2005; Oppenheimer et al., 2007; Jonkers et al., 2009; KasprzykHordern et al., 2009), they have been still detected in river water samples at low ng L1 (Benijts et al., 2004; KasprzykHordern et al., 2008; Gonza´lez-Marin˜o et al., 2009; Villaverdede-Sa´a et al., 2010). The main concern once they reach the environment is that they have proved to show oestrogenic activity (Routledge et al., 1998; Golden et al., 2005; Soni et al., 2005), relatively weak compared to that of 17b-oestradiol but not negligible, as they occur at much higher concentrations than the latter compound. Besides this, they can easily react with free chlorine when mixed with chlorinated tap water (Canosa et al., 2006a), yielding mostly mono and dichlorinated/brominated derivatives that have already been detected in raw wastewater. Although halogenation masks the apparent oestrogenic activity of the parent compounds (Terasaki et al., 2009a), the resulting chlorinated by-products show higher acute toxicity responses in the Daphnia magna test (Terasaki et al., 2009b), a fact that should be taken into account in case they reach the aquatic medium. However, the occurrence of such derivatives in the environment has not been investigated yet and, to the best of our knowledge, the biodegradability of parabens and their by-products during wastewater treatments still remains unknown. Therefore, the aim of this work was the evaluation of the occurrence and biodegradability of parabens and their chlorination by-products in raw and treated wastewater. To this end, samples were preconcentrated by solid-phase extraction (SPE) followed by liquid chromatography-quadrupole-time-offlight mass spectrometry (LC-QTOF-MS) determination. The use of LC-QTOF-MS has emerged in the last years as a valuable technique for the identification of by-products from emerging pollutants (Kosjek and Heath, 2008) due to the unique combination of high selectivity and structural information derived from accurate-mass MS and MS/MS spectra, as well as novel software implementations, which allow the comparison with empirical formulae databases (Go´mez et al., 2010). Thus,
6771
the potential of a modern LC-QTOF-MS was evaluated in terms of both qualitative and quantitative capabilities.
2.
Materials and methods
2.1.
Chemicals
Fig. 1 shows the structure of the analytes included in this study. Methyl (MeP), ethyl (EtP), n-propyl (n-PrP), n-butyl (nBuP) and benzyl (BzP) esters of 4-hydroxybenzoic acid were purchased from Aldrich (Milwaukee, WI, USA); iso-propyl paraben (i-PrP) and iso-butyl paraben (i-BuP) were from TCI Europe (Zwijndrecht, Belgium). Halogenated derivatives of MeP: 3-chloro-, 3,5-dichloro- and 3,5-dibromo-methyl paraben (ClMeP, Cl2MeP and Br2MeP) were obtained from ABCR GmbH&Co (Karlsruhe, Germany). As internal standards (ISs), methyl 4-hydroxybenzoate-2,3,5,6-d4 (MeP-d4) and n-propyl 4hydroxybenzoate-2,3,5,6-d4 (n-PrP-d4) were from CDN Isotopes (Quebec, Canada). Stock solutions of each compound (1000 mg mL1) and mixtures of all of them or their deuterated analogues (10 mg mL1) were prepared in methanol and stored at 20 C until use. Calibration standards with increasing concentrations of the analytes and 100 ng mL1 of ISs were prepared in methanol:water (1:1). HPLC grade methanol and ammonium acetate, employed as mobile phase additive, were supplied by Merck (Darmstadt, Germany). Salts used in the preparation of the assay medium during the activated sludge biodegradation study (ammonium chloride, potassium dihydrogen phosphate anhydrous, dipotassium monohydrogen phosphate anhydrous and disodium monohydrogen phosphate dihydrate) were also purchased from Merck. Magnesium sulphate heptahydrated was from Aldrich (Milwaukee, WI, USA), and calcium chloride and iron (III) chloride hexahydrated were supplied by Riedel de Hae¨n (Seelze, Germany).
2.2.
Samples and sample extraction
Raw and treated sewage samples were collected in different days during April and May 2010, at three different urban WWTPs: codes A (3 days study), B (3 days study) and C (5 days study). These plants receive the discharges from small and medium size cities (18,000, 15,000 and 125,000 inhabitants, respectively) located in the same metropolitan area of the northwest of Spain. All of them comprise a primary and a secondary (activated sludge) treatment. Samples were collected in amber glass bottles previously rinsed with methanol and ultrapure water and extracted (SPE) in less than 6 h after sampling. Prior to extraction, particulate matter was removed using a combination of glass fibre prefilters and 0.45 mm nitrocellulose filters (Millipore, Bedford, USA). Extraction was performed as detailed elsewhere (Gonza´lezMarin˜o et al., 2009). In brief, Oasis HLB cartridges (60 mg, 3 mL), obtained from Waters (Milford, MA, USA), were sequentially conditioned with 3 mL of methanol and 3 mL of ultrapure water. Subsequently, 200 mL filtered samples, spiked with isotopically labelled standards and, in the case of recovery studies, also with analytes, were passed through them at
6772
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 7 0 e6 7 8 0
ClMeP
O O
R
HO
Cl2MeP
O
Cl
O
CH3
HO
MeP
CH3
EtP
CH2CH3
i-PrP
CH(CH3)2
n-PrP
CH2CH2CH3
i-BuP
CH2CH(CH3)2
n-BuP
CH2CH2CH2CH3
BzP
CH2Ph
O
Cl
O
CH3
HO Cl
Br2MeP Br
O
ClPrP O
HO
CH3
Cl
O O
CH2CH2CH3
HO
Br
Fig. 1 e Structures of parabens and halogenated derivatives included in the study.
a flow rate of approximately 10 mL min1. The sorbent was dried under vacuum for 30 min and the analytes were eluted with 4 mL of methanol. Extracts were concentrated down to ca. 0.5 mL with a gentle stream of nitrogen (99.999%) in a Turbovap II concentrator (Zymark, Hopkinton, MA, USA), diluted to a final volume of 1 mL with ultrapure water and injected (10 mL) into the LC-MS system.
2.3.
LC-QTOF-MS
LC separations were carried out on a 100 2.1 mm Halo C18 (2.7 mm) porous shell column, consisting of a 1.7 mm inert core coated with a 0.5 mm C18 layer (Advanced Materials Technology, Nes-Ziona, Israel). The column was protected with a 4 2 mm C18 guard cartridge provided by Phenomenex (Torrance, CA, USA) and thermostatted at 45 C. A dual eluent system of water (A) and methanol (B), both containing 5 mM of ammonium acetate and adjusted to pH 4.5 with acetic acid, was used. The flow rate was maintained at 0.2 mL min1 and the gradient was as follows: 0 min (5% B), 10 min (40% B), 15 min (55% B), 20 min (55% B), 28 min (100% B), 30 min (100% B), 32 min (5% B) and 42 min (5% B). Analyses were performed by LC-QTOF-MS using an Agilent 1200 Series liquid chromatograph (comprising a membrane degasser, a binary high-pressure gradient pump, a thermostatted LC column compartment and an autosampler) interfaced to a quadrupole-time-of-flight mass spectrometer (Agilent 6520 accurate-mass) equipped with a Dual electrospray ion source. Nitrogen, used as nebulising and drying gas, was provided by a nitrogen generator (Erre Due srl, Livorno, Italy). High purity nitrogen (99.9995%, Carburos Meta´licos, A Corun˜a, Spain) was used as collision gas for MS/MS measurements. The voltage of the ESI needle was set at 4 kV in the negative ionisation mode. The gas temperature of the source was 350 C, the drying gas flow 11 L min1 and the nebulising gas pressure 45 psig. In the collision cell, nitrogen was kept at 18 mTorr. Analytes were quantified in single-MS mode from the accurate-mass extracted chromatograms (10 ppm mass window). Moreover, MS/MS spectra were simultaneously recorded for confirmation purposes (2 spectra per second, time window of 1.5 min centred in the retention time of each
analyte). The fragmentor voltage was set at 160 V and the collision energy at 20 V. The instrument was operated in the 2 GHz (extended-dynamic range) mode and tuned at the beginning of each analysis series (ca. every 1e2 days) with a tuning solution containing different m/z values in the 100e1700 m/z range, according to the manufacturer instruction (Agilent Technologies). During each chromatographic run, the mass-axis was constantly recalibrated. To this end, the second sprayer was continuously infused with a reference solution provided by Agilent Technologies, for which in the negative mode (ESI-), the reference masses were 68.995758, 112.985587 and 980.016375 m/z (FWHM resolution: ca. 4700 at m/z 113 and ca. 11000 at m/z 980). The m/z values of the MS quantification ions, also used as precursors for MS/MS, and confirmation fragment ions, as well as acquisition times and internal standards used for each analyte, are compiled in Table 1. Instrument control, data acquisition and evaluation were performed with the Mass Hunter software (Agilent Technologies). A database containing the empirical formulae of all the mono and dihalogenated possible parabens, with the halogen being Cl or Br, was created in order to identify those byproducts for which standards were not available.
2.4. Aerobic biodegradability and stability in raw wastewater The biological degradation of parabens was evaluated through two different series of assays. First of all, a batch of aerobic biodegradation tests was carried out based on the ISO 7827 international standard (ISO, 1994). Thus, 1.2 L glass bottles were filled with 1 L of ultrapure water containing a phosphate buffer, a pool of inorganic salts (ISO, 1994) and 5 mg L1 of one of the considered parabens (MeP, EtP, n-PrP, i-BuP, n-BuP, ClMeP, Cl2MeP or Br2MeP). Then, 20 mg of activated sludge from WWTP-C were added to each solution as inoculum. Control tests, without and with poisoned (HgCl2) sludge, and with aniline as a control substrate, were also run in parallel in order to check for abiotic degradation, sorption processes and sludge activity. All solutions were kept in the dark at 20 2 C under continuous stirring. Samples (ca. 40 mL) were taken every few days,
6773
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 7 0 e6 7 8 0
immediately filtered through 0.45 mm membrane filters (cellulose acetate; Sartorius, Goettingen, Germany) and stored frozen until being analysed by LC-QTOF-MS. A second series of assays was performed in order to evaluate the stability of parabens in real wastewater. A raw wastewater sample, characterized by a pH of 7.1, a total suspended solids content of 100 mg L1 and a chemical oxygen demand of 240 mg L1, collected from WWTP-C was used for this purpose. Non-filtered aliquots (10 mL) were poured in 16 mL amber vials, spiked with one of the considered parabens (50 ng mL1) and the closed vials were kept in the dark at room temperature (20 2 C). Control poisoned (HgCl2) tests were carried out in parallel. Fractions of ca. 1 mL were taken at different times, from a few hours up to 6 days, passed through 0.20 mm membrane filters (cellulose acetate; Sartorius, Goettingen, Germany) and stored frozen until analysis. In this case, 50 mL were injected in the LC-MS system, so that the achieved LODs stayed below 0.2 ng mL1 for all species. Thus, it was possible to follow their degradation up to a percentage higher than 99%. All degradation data were fitted to a logistic model with the software Graphpad Prism 5. This type of mathematic model, very common in enzymatic reactions, can account for the initial lag phase and the subsequent first-order degradation. A simplified model (Gimsing et al., 2006) was used: C K ¼ C0 1 þ c$ert where K, c and r are the fitting parameters. K represents the inhibiting factor and r the rate constant. In the case that K and c become very large, then the logistic function will approach a first-order exponential decay. Thus, the half-lives (t1/2) can be calculated as follows (Gimsing et al., 2006): t1=2 ¼ ln
K 0:5 1 $ c$0:5 r
Or, in general, the time when C/C0 ¼ x, tx, is calculated from the general expression: Kx 1 tx ¼ ln $ c$x r
3.
Results and discussion
3.1.
Method performance
The SPE-LC-QTOF method was based on a previously published work (Gonza´lez-Marin˜o et al., 2009). However, since the instrument was changed from a triple-quadrupole to a QTOF system and several halogenated by-products were included among the target analytes, SPE and LC-MS procedures required slight adaptations and were therefore revalidated. The information on the performance of the method is compiled in Table 2. In the first instance, the original composition of the LC eluents, which were originally buffered with 5 mM ammonium acetate (pH ca. 7) (Gonza´lez-Marin˜o et al., 2009), needed to be acidified to pH 4.5. Otherwise, the halogenated parabens showed broad peaks and were poorly retained into the reversed-phase C18 column, due to their more acidic character (calculated pKa ¼ 5.3e6.9) versus the original parabens (calculated pKa values in the 8e8.5 range) (SciFinder, 2011). Then, the performance of the LC-QTOF-MS method was tested in terms of repeatability and, particularly, LODs and linearity, since a short dynamic range has been reported as one of the main drawbacks of (Q)TOF instruments in quantitative analysis (Ferrer et al., 2005; Ferna´ndez-Alba and Garcı´a-Reyes, 2008). However, new implementations in modern instruments can overcome this problem. In the case of the system used in this work, the Analogue-to-DigitalConverter (ADC) can either work in the 4 GHz mode, which grants a higher mass resolution, and in the 2 GHz mode (resolution ca. half of 4 GHz), where data are acquired at two gain levels, expanding the linear range. Thus, this latter mode was employed, and the obtained linearity was satisfactory from 5 to 500 ng mL1 (IS concentration of 100 ng mL1) with R2 values varying from 0.9986 to 0.9996 (Table 2). Instrumental precision studies were carried out at two different levels (20 and 200 ng mL1) by seven injections of the same standard over a 24 h period; obtained relative standard deviations (RSDs) ranged between 1.6 and 5.4% in the first case and between 2.2 and 5.6% in the second one. Instrumental LODs of the LC-MS method were defined for
Table 1 e Accurate m/z ratios of the [MLH]L precursor ions, confirmation fragment ions and internal standards used in each case. Compound
tR
IS
(min) MeP EtP ClMeP Cl2MeP Br2MeP i-PrP n-PrP i-BuP n-BuP BzP ClPrP
15.9 19.3 19.8 20.3 21.1 21.6 22.2 25.6 26.1 26.3 26.8
MeP-d4 n-PrP-d4 n-PrP-d4 n-PrP-d4 n-PrP-d4 n-PrP-d4 n-PrP-d4 n-PrP-d4 n-PrP-d4 n-PrP-d4 n-PrP-d4
[MH]
Confirmation fragment ions
(m/z)
(m/z)
(m/z)
151.0401 165.0557 185.0011 218.9621 308.8591 179.0714 179.0714 193.0870 193.0870 227.0714 213.0324
136.0166 136.0166 169.9776 159.9488 249.8458 136.0166 136.0166 136.0166 136.0166 136.0166 170.9854
92.0268 92.0268 125.9878 131.9539 78.9189 92.0268 92.0268 92.0268 92.0268 92.0268 125.9878
6774
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 7 0 e6 7 8 0
Table 2 e Performance parameters of the method. Compound
LC-TOF-MS 2a
R
1
b
LOD (ng L )
%RSD
Mass accuracy relative error (ppm)
20 ng mL1 200 ng mL1 MeP EtP ClMeP Cl2MeP Br2MeP i-PrP n-PrP i-BuP n-BuP BzP a b c d
0.9996 0.9990 0.9995 0.9986 0.9986 0.9992 0.9991 0.9987 0.9993 0.9994
0.8 0.3 0.05 0.2 0.09 0.3 0.3 0.4 0.4 0.4
3.9 2.0 3.2 5.4 3.1 3.2 2.1 1.6 2.1 3.4
SPE-LC-TOF-MS
3.4 2.9 4.0 5.6 5.6 3.0 4.7 4.8 2.2 2.9
10 ng mL1
100 ng mL1
2.7 2.1 6.5 2.1 1.5 2.0 1.4 1.6 1.6 1.8
1.7 0.9 1.4 1.4 1.0 0.8 0.6 0.8 0.5 1.1
c
LOD (ng L1) %Rd SD
4.0 1.4 0.3 0.9 0.5 1.6 1.3 2.2 2.2 1.9
99 104 99 119 125 92 103 108 103 102
4 4 2 2 5 2 5 8 7 8
Calibration range 5e500 ng mL1 (IS 100 ng mL1). n ¼ 7 replicates. Mean values from two two replicates. Percentages of recovery for 200 mL ultrapure water samples spiked with 1.25 ng mL1 of each analyte and 0.5 ng mL1 of each IS.
a peak-to-peak signal-to-noise (S/N) ratio of 3 (mean of three replicates); measuring the noise in the ca. 2 min time region before and after the retention time of each analyte. Achieved LODs ranged from 0.05 ng mL1 (ClMeP) to 0.80 ng mL1 (MeP), considering an injection volume of 10 mL. Thus, the performance of the LC-QTOF system in the single-MS mode turned out to be comparable with that of triple-quadrupole instruments (Gonza´lez-Marin˜o et al., 2009). Furthermore, valuable qualitative information was provided and accurate-mass MS/ MS spectra could be recorded simultaneously for the target analytes, minimising the risk of false positives. A further tested parameter was the relative error of mass assignations to [MH] ions, which was maintained below 3 ppm, except ClMeP (6.5 ppm), for a concentration level of 10 ng mL1, and below 2 ppm for all compounds at 100 ng mL1 concentrations (Table 2). Finally, recoveries of the SPE protocol were also reevaluated using 200 mL aliquots of ultrapure water spiked with 1.25 ng mL1 of analytes and 0.5 ng mL1 of ISs; obtained values ranged from 92.0% (i-PrP) to 125.4% (Br2MeP).
3.2.
Screening of halogenated parabens
As it has been proved by Canosa et al. (Canosa et al., 2006a), halogenated derivatives of parabens can be formed by reaction of the parent bactericides with residual chlorine in tap water during showering and bathing. Thereafter, these byproducts might enter the aquatic environment through sewage water. Occurrence of such by-products was screened by LC-QTOF-MS, as pure standards were only available for some of the MeP derivatives. First, a database containing all the potential mono and dichlorinated/brominated derivatives of target parabens was created with the Personal Compound Database Library software, included within the Mass Hunter package. For each possible halogenated paraben, this database comprises its name and empirical formula; other information such as structure, CAS No. and Chemspider No. (directly linking to
PubChem and Chemspider public internet databases with more information on each particular chemical) can also be added. Then, the Mass Hunter software provides a “Find by Formula” function that automatically generates the accurate m/z values of the ionised compounds according to the ESI-MS polarity (negative in this case) and the parameters considered, i.e. deprotonation, formation of acetate adducts, etc. In this case, neither adducts nor dimers/trimers were expected to occur, so they were not contemplated. Subsequently, the software searches for peaks with these accurate-masses (within an m/z window of 5 ppm), accurate-mass extracted chromatograms are generated and their peak spectra compared with the calculated one in terms of three parameters: mass accuracy, isotopic match and spacing between the different molecular ions observed in each cluster of signals, determined by the natural abundance of elemental isotopes in each ion. These three parameters are combined into an overall score, where the mass accuracy contributes a 48%, the isotopic distribution a 28% and the spacing between ions a 24%. Hence, an overall score of 100 would represent a perfect match. More details on the Mass Hunter algorithm have been previously reported by Go´mez et al. elsewhere (Go´mez et al., 2010). Other (Q)TOF manufacturers use similar algorithms (Herna´ndez et al., 2011). Applying this methodology, ClMeP and Cl2MeP were found in all the influents and in a few effluents and ClPrP could be also detected in all the influents. Indeed, in the case of the chlorinated derivatives of MeP, this screening approach was unnecessary as standards were commercially available and unequivocal identification and quantification could be performed. Nevertheless, Cl2MeP illustrates a good example of QTOF potential due to the finding of two possible peaks with the same empirical formula and different retention times in a sample of WWTP-B. As shown in Fig. 2a, the MS spectrum of the first eluting compound matched very well (overall score 99.56) with the theoretical spectrum of the deprotonated Cl2MeP (ion formula: C8H5Cl2O3). This fact indicates that this species has the same empirical formula than Cl2MeP, but, obviously, this is not enough for a positive identification, as
6775
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 7 0 e6 7 8 0
a
Unknown compound extracted chromatogram: 218.9621 + 220.9593
MATCH WITH THE LIBRARY
x104 x103
1
7
0.9
218.9620 (M-H)-
6
0.8
5
0.7
220.9591 (M-H)-
4
0.6
3
0.5
2
0.4
219.9657 (M-H)-
1
0.3
0
0.2 0.1
Unknown compound MS spectrum
222.9571
217 218 219 220 221 222 223 224 225 m/z
Formula (M)
C8 H6 Cl2 O3
Ion Formula
C8 H5 Cl2 O3
Score
99.56
Calc m/z
218.9621
Diff (ppm)
0.45
Diff (mDa)
0.1
Mass Match
99.96
Abund Match
98.66
Spacing Match
99.84
0 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 Time (min)
b
INTERPRETATION
x104 1.6
159.9480
Cl2MeP MS/MS spectrum
1.2 0.8 218.9611
0.4 0 x102
131.9536 174.9714
4
Unknown compound MS/MS spectrum
3 2 1
138.9943
0 110 120 130 140 150 160 170 180 190 200 210 220 230 240 m/z
m/z
159.9480
131.9536
Formula
C6 H2 Cl2 O
C5 H2 Cl2
Diff (ppm)
5.2
2.31
Diff (mDa)
0.83
0.31
Loss Mass
59.0133
87.0082
Loss Formula
C2 H3 O2
C3 H3 O3
m/z
174.9714
138.9943
Formula
C7 H5 Cl2 O
C7H4ClO
Diff (ppm)
4.9
9.26
Diff (mDa)
0.86
1.29
Loss Mass
43.9897
74.9668
Loss Formula
CO2
CHClO2
Fig. 2 e Distinction between Cl2MeP and an unknown compound with the same empirical formula and same precursor m/z: (a) extracted accurate-mass MS chromatogram and MS spectrum of the unknown compound and empirical formula match; (b) MS/MS spectrum of Cl2MeP facing MS/MS spectrum of the unknown species.
a search in e.g. SciFinder Scholar database (SciFinder, 2011) revealed 169 known chemicals whose formula is C8H6Cl2O3. Yet, when both compounds were submitted to reinjection and CID fragmentation, their high resolution MS/MS spectra turned out to be completely different, proving that the first eluting compound was actually a different one and not Cl2MeP itself. In fact, its fragmentation was dominated by a decarboxylation and a further, or simultaneous, loss of HCl, indicating that it is a carboxylic acid (Fig. 2b). On the other hand, Cl2MeP fragmented through the typical parabens pattern, i.e. loss of the carboxylic group together, or sequentially, with the side chain (nominal m/z 160). On the other hand, the monochlorinated derivative of PrP, ClPrP, could not be quantified in any of the samples since no standard was available in the laboratory. However, it was identified in all of the raw sewage samples at low concentrations from its accurate-mass MS and MS/MS spectra (Fig. 3a and b, respectively); the latter one showing the typical paraben fragmentation pattern, i.e. loss of side chain from the
ester (nominal m/z 171) and production of the chlorophenolate anion by elimination of the carboxylic ester group (nominal m/z 126), confirming the identity of ClPrP. Obviously, once a by-product is identified for the first time it needs to be re-injected in order to obtain the MS/MS spectrum. However, when the retention time is known, its precursor ion can be introduced in the targeted MS/MS list of the method in order to automatically obtain the MS/MS confirmatory spectrum; therefore, a single injection suffices for a confident identification in the remaining set of samples. Hence, the findings from the screening study corroborated the data described by Canosa et al., who also detected the presence of the dichlorinated forms of methyl and propyl paraben in raw wastewater by GCeMS with an iontrap instrument after analytes’ derivatisation (Canosa et al., 2006a). The non-detection of brominated derivatives can be attributed to the low levels of bromide in the geographical area of investigation. Also, it is logical to detect only MeP and n-PrP derivatives, as these two precursor
6776
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 7 0 e6 7 8 0
a
ClPrP MS chromatogram:
MATCH WITH THE LIBRARY
213.0324 + 215.0297
x10 3
x103
213.0320
2
3
1.6
2.5
1.2
2
0.8
1.5
0.4
1
0
ClPrP MS spectrum
(M-H)-
211.1332
215.0290 217.0025
211 212 213 214 215 216 217 218 219 220
m/z
0.5
Formula (M)
C10 H11 Cl O3
Ion Formula
C10 H10 Cl O3
Score
94.03
Calc m/z
213.0324
Diff (ppm)
1.71
Diff (mDa)
0.37
Mass Match
99.51
Abund Match
97.81
Spacing Match
78.56
0 25
b
25.5
26
26.5
27
27.5
28
28.5
29
29.5
30
30.5 Time (min)
ClPrP MS/MS chromatogram: 213.0000 -> 125.9865
x10
2
x10 2 2
2
125.9868
1.6
1.2
1.2
0.8 0.4
0.8
spectrum 170.9845
213.0296
111.0811
0
110 120 130 140 150 160 170 180 190 200 210
m/z
0.4 0
ClPrP MS/MS
1.6
25
25.5
26
26.5
27
27.5
28
28.5
29
29.5
30
30.5 Time (min)
Fig. 3 e Identification of ClPrP from its accurate-mass precursor and product ion mass spectra: (a) extracted MS chromatogram and MS spectrum and empirical formula match; (b) MS/MS chromatogram and spectrum and empirical formula match.
parabens are the ones found at higher concentrations in wastewater (see Section 3.3).
3.3. Occurrence of target parabens and halogenated byproducts in urban wastewaters After the screening procedure, samples were also investigated in order to quantitate those compounds with commercially available standards. Table 3 compiles the overall concentrations found in all the WWTPs influents and effluents, the frequency of detection and the removal efficiency. Although these data were obtained from grab samples, they arise from 11 pairs of samples from three different WWTPs, so they provide a reliable estimation of the occurrence of the target analytes. In fact, due to the instability of parabens in influent (raw) wastewater, 24-h composite samples could even lead to a severe underestimation of their concentrations (see Section 3.4). For calculation of average and median values, samples below the LOD were treated as if their concentrations were half of the LOD. MeP was the most frequently detected compound, in 100% of the samples, and also the most abundant (average concentrations of 4200 ng L1 and 25 ng L1 in raw and treated wastewater, respectively). This fact is a reflection of its ubiquitous presence in cosmetic formulations. Following this
trend, n-PrP and EtP, the next two most frequently used parabens, also occurred in all the influents (average values of 1400 and 880 ng L1, respectively) and in some of the analysed effluents (maximum concentration of 21 ng L1). n-BuP and iBuP were also present in all the raw wastewater samples, although at lower concentrations (average values: 140 and 57 ng L1, respectively) but they were not detected in any of the treated wastewater samples. Finally, i-PrP was only found in some of the influents of plants B and C at marginal levels (lower than 6 ng L1) and BzP was not detected in any sample. These concentration ranges are in good agreement with the literature, showing the prevalence of MeP followed by either EtP or n-PrP and then n-BuP, whereas detection of BzP has only been reported at very low levels (Lee et al., 2005; Jonkers et al., 2009; Kasprzyk-Hordern et al., 2009). It is also interesting to notice that the n-BuP/i-BuP influent concentration ratio was quite constant (2.4 0.3, average standard deviation) and very similar to the value found in two different grab samples in the only published data concerning the levels of both compounds in raw wastewater (1.9 0.1 and 2.0 0.1, respectively) (Gonza´lez-Marin˜o et al., 2009). This may simply be due to the fact that they are produced as a technical mix of both isomers. Regarding halogenated parabens, Br2MeP was not detected in any sample, whereas ClMeP and Cl2MeP were found in all
6777
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 7 0 e6 7 8 0
Table 3 e Concentrations (ng LL1), percentage of samples above the LOD in both raw and treated wastewater and mean removal values considering the three WWTPs (n [ 11 samples); BzP and Br2MeP were not detected in any sample; n.d.: not detected (
Influent Effluent Influent Effluent Influent Effluent Influent Effluent Influent Effluent Influent Effluent Influent Effluent Influent Effluent
4200 25 880 4.0 40 1.2 46 2.6 2.1 n.d. 1400 4.8 57 n.d. 140 n.d.
Median 2500 19 760 2.5 39 n.d. 40 n.d. n.d. n.d. 1400 n.d. 65 n.d. 130 n.d.
SD 3200 14 520 3.3 18 2.0 29 4.3 1.8 n.d. 670 6.4 28 n.d. 68 n.d.
the influents at similar levels (average values of 40 and 46 ng L1, respectively) and could still be found in some of the effluents at lower concentrations (up to 12 ng L1). To our knowledge, these are the first quantitative data on the occurrence of halogenated parabens in wastewater. Taking into account the highest oestrogenic values published for these compounds (Terasaki et al., 2009a), the sum of average paraben concentrations in influents and effluents would be equivalent to ca. 1.6 ng L1 and 0.01 ng L1 of 17boestradiol, respectively. Therefore, they do not seem to represent a hazard in terms of estrogenicity, as concentrations of natural and synthetic oestrogens are much higher, ca. 10e100 ng L1 of 17b-oestradiol equivalents in raw and treated wastewater (Ying et al., 2002). However, more investigations are required in the case of halogenated derivatives, which are more toxic than their precursors (Terasaki et al., 2009b) and may appear at higher concentrations in other areas where higher chlorine doses are applied and/or tap water contains significant amounts of bromide. Finally, an estimate of the removal efficiency was calculated from the average concentration values measured in influents and effluents. As it can be seen in Table 3, removal percentages were higher than 90% in all cases. Yet, it is noteworthy that they were higher than 9% for all nonhalogenated parabens that could be detected in some of the effluents. The removal of i-PrP was not calculated as its influent concentration was already marginal. Also, the average removals of the butylated parabens could only be assured to be higher than 96%, as they were not detected in effluents and the calculation of their removal is limited by their LODs. Regarding the two chlorinated by-products, their removals were slightly lower than that of their precursor compound, MeP, ranging from 94% (Cl2MeP) to 97% (ClMeP). Although this does not represent a statistically relevant difference, it may suggest a slightly higher persistence of the halogenated derivatives compared to their parent compounds, which was studied in detail within a lab biodegradation test (Section 3.4). Removal values of non-halogenated parabens are in good
Min 290 6.1 250 n.d. 12 n.d. 8.0 n.d. n.d. n.d. 520 n.d. 13 n.d. 39 n.d.
Max 10000 50 1600 9.8 61 6.9 90 12 5.6 n.d. 2800 21 110 n.d. 270 n.d.
% samples > LOD 100 100 100 73 100 18 100 45 45 0 100 36 100 0 100 0
Mean removal (%) 99.4 99.5 96.9 94.3 e 99.7 >96.2 >98.4
agreement with previous findings for some of these chemicals in Europe and America (Lee et al., 2005; Oppenheimer et al., 2007; Jonkers et al., 2009; Kasprzyk-Hordern et al., 2009), and also recent investigations in grey water showed a good removal (>90%) on lab-scale bioreactors (Herna´ndez Leal et al., 2010).
3.4. Biodegradability and wastewater stability evaluation Taking into account the previous results (Section 3.3), only the biodegradation of those parabens which occurred in raw wastewater at significant levels was investigated. Furthermore, it was also assayed for the three halogenated derivatives of MeP, for which there were commercially available standards: ClMeP, Cl2MeP and Br2MeP. Although the latter byproduct was not detected in any of the analysed samples, it may occur in other locations with higher concentrations of bromide in natural waters (Canosa et al., 2006a). The assay was carried out following the ISO 7827 standard in terms of sludge inoculum preparation and tested concentrations, as detailed in Section 2.4. No concentration changes were observed for any compound neither in the control nor in the inhibited media in the course of the whole study, proving that neither adsorption nor degradation due to abiotic processes occurred and that any loss in the test solutions had to be attributed to biological routes (data not shown). Biodegradation fitted profiles and experimental data points are displayed in Fig. 4. Table 4 compiles the fitting model parameters, as well as estimated half-lives (t1/2) and time required to reach a degradation level of 99% for each compound, i.e. C/C0 ¼ 0.01 (t0.01). The obtained R2 values, higher than 0.97, demonstrate that the logistic model fitted quite well to the experimental data. This parameter could not be calculated in the case of MeP and EtP, since their degradation was too fast to obtain enough data points. As it can be seen, all considered non-halogenated parabens were readily biodegraded (Fig. 4a), presenting half-lives lower
6778
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 7 0 e6 7 8 0
a
Cl2MeP
nPrP
Br2MeP
0.75
C/C0
iBuP nBuP
0.50 0.25 0.00
ClMeP
1.00
EtP
0.75
C/C0
b
MeP
1.00
0.50 0.25
0
2
4
0.00
6
time (days)
0
5
10
15
20
time (days)
c
d MeP
1.00
1.00
EtP nPrP
0.75
iBuP
C/C0
C/C0
0.75
nBuP
0.50 0.25 0.00
0.50
ClMeP Cl2MeP
0.25
0
25
50
75
100
125
150
time (h)
0.00
Br2MeP
0
25
50
75
100
125
150
time (h)
Fig. 4 e Biodegradation profiles in the activated sludge batch test (a) and (b); and in the raw sewage batch test (c) and (d).
than 3 days and reaching a 99% degradation level in less than 5 days (Table 4). However, their persistence was observed to slightly be increased with the length of the hydrocarbonated chain. Hence, propylated and butylated parabens required from 3.7 to 4.5 days to reach a 99% of degradation, while MeP and EtP required only 2.1 days. On the other hand, halogenated derivatives of MeP showed slower biodegradation kinetics than their parent compound (Fig. 4b), with half-lives of 3.3 days for the monochlorinated species, 8.6 days for the dichlorinated and 9.7 days for di-brominated one (Table 4). This fact corroborates the findings on the WWTP removal, where efficiency slightly decreases as chlorination degree increases, in agreement with the relative stabilities of mono and dihalogenated paraben by-products in presence of free chlorine (Canosa et al., 2006a). No peaks of possible transformation products were detected by the LC-QTOF-MS system, indicating that the tested compounds were completely metabolised and incorporated into the sludge biomass or, at least, that the transformation products were not easily ionised by ESI. A second series of assays was performed with raw wastewater spiked with parabens at the 50 ng mL1 level in order to test the stability of these compounds in a real influent for a week (Fig. 4c and d). A poisoned control sample was run in parallel, without any degradation being observed. As presented in Table 4, again degradation kinetics fitted quite well the logistic model. Half-lives of native parabens ranged between 9.6 and 35 h, but, in this case, the species with longer hydrocarbonated chain underwent a faster degradation, in contrast with the activated sludge batch test. This may account for the different bacteriological and enzymatic composition of
sewage as compared to activated sludge. On the other hand, stability of methyl paraben derivatives showed the same trend than the previous study, increasing with the number of halogen substituents, as t1/2 of ClMeP was 28 h, whereas for Cl2MeP and Br2MeP it was longer than 1 week (Table 4). Another important datum shown in Table 4 for this study is the C/C0-12 h. This value represents the concentration of compound remaining in the sample after 12 h, which is the average time that a real sample would stand if the commonly used 24 h composite sampling was performed. Hence, it can be appreciated that, after 12 h, more than 50% of the butylated parabens would have been degraded, while for the rest of native parabens degradation would range from 17 to 34%. Degradation of halogenated parabens during sampling would be also an issue for ClMeP (16% loss) but not for the dihalogenated species (less than 3% of degradation). In view of these results, 24-h composite samples would clearly lead to underestimated concentrations, particularly for the long-length chain parabens. Therefore, it was decided to perform grab sampling through this work, with a total of 11 samples being used to estimate average concentrations and removal percentages. Indeed, this would result in a random error, but it is expected to be much less important than the bias introduced by composite sampling for these personal care compounds. In fact, Ort et al. have recently evaluated the error derived from grab sampling versus different composite sampling approaches (Ort et al., 2010). They observed that using just 4 grab samples to derive mean concentrations resulted in less than 30% error for the pharmaceuticals used in a higher extent, as compared to the best sampling system provided that analytes are stable (continuous flow-proportional composite
6779
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 7 0 e6 7 8 0
Table 4 e Parameters from the logistic kinetic fitting, half-lives (t1/2) and time required for 99% degradation (C/C0 [ 0.01; t0.01) from the activated sludge and real wastewater batch tests. In the last case, C/C0 after 12 h is also presented (C/C0e12 h). Activated sludge batch test
MeP EtP n-PrP i-BuP n-BuP ClMeP Cl2MeP Br2MeP
K
c
r (day1)
R2
t1/2 (days)
t0.01 (days)
1.00 1.00 0.98 1.00 0.96 0.99 0.93 0.90
1.4E11 3.0E11 8.5E04 3.4E08 8.1E06 5.0E05 3.3E03 8.8E04
13.79 13.43 2.60 5.97 4.02 3.04 0.65 0.70
* * 0.9978 1.0000 0.9976 0.9999 0.9748 0.9855
1.8 1.8 2.7 2.9 2.9 3.3 8.6 9.7
2.1 2.1 4.5 3.7 4.0 4.8 15.9 16.4
Real wastewater test K MeP EtP n-PrP i-BuP n-BuP ClMeP Cl2MeP Br2MeP
1.09 1.35 6.3E þ 11 9.3E þ 11 3.01 1.04 1.12 1.95
c
r (h1)
R2
t1/2 (h)
C/C0e12 h
6.6E02 3.5E01 6.4E þ 11 9.4E þ 11 1.9 7.4E02 1.3E01 9.4E01
0.082 0.057 0.034 0.058 0.103 0.095 0.009 0.003
0.9974 0.9976 0.9947 0.9972 0.9896 0.9978 0.9753 0.9855
35.2 27.5 20.3 11.8 9.6 28.2 237.1 449.5
0.93 0.79 0.66 0.49 0.41 0.84 0.97 0.99
*R2 values not calculated owing to a very high degradation rate.
sampling). In our case, with 11 samples being taken, this error is expected to be reduced to less than 20%, owing also to the wide usage level of parabens; as compared to 24-h composite samples that would lead up to 60% biased results.
4.
Innovation (Ramo´n y Cajal research program). IGM acknowledges the Spanish Ministry of Education (Ministerio de Educacio´n) for her FPU grant. Finally, we are indebt to Aquagest and Espina & Delfı´n, water supply/quality control companies, for kindly providing access to wastewater samples.
Conclusions
It has been proved that the new generation of QTOF instruments provides adequate limits of detection and good linearity, comparable to triple-quadrupole instruments for quantitative purposes. Moreover, their accurate-mass determinations, in the full scan and MS/MS acquisition modes permit the screening of transformation by-products without pure standards, which could otherwise not be detected with triple-quadrupole instruments. This is the case of ClPrP, which could not be quantified due to the lack of standard, but was detected in all influents. Parabens and the halogenated ClMeP and Cl2MeP were quantified in all raw wastewater samples, where MeP and n-PrP were the prevalent analytes. Their removal in the WWTPs was however high in all cases (>94%). Laboratory degradation tests, both with activated sludge and raw wastewater, demonstrated that the dihalogenated derivatives of MeP had significantly higher half-lives than MeP itself.
Acknowledgements This research was funded by the Spanish Ministry of Science and Innovation (Ministerio de Ciencia e Innovacio´n) and FEDER funds; projects CTQ2010-18927 and CTQ2009-08377. JBQ extends his gratitude to the Spanish Ministry of Science and
references
Benijts, T., Lambert, W., De Leenheer, A., 2004. Analysis of multiple endocrine disruptors in environmental waters via wide-spectrum solid-phase extraction and dual-polarity ionization LC-ion trap-MS/MS. Analytical Chemistry 76, 704e711. Canosa, P., Rodrı´guez, I., Rubı´, E., Negreira, N., Cela, R., 2006a. Formation of halogenated by-products of parabens in chlorinated water. Analytica Chimica Acta 575, 106e113. Canosa, P., Rodrı´guez, I., Rubı´, E., Bollaı´n, M.H., Cela, R., 2006b. Optimisation of a solid-phase microextraction method for the determination of parabens in water samples at the low ng per litre level. Journal of Chromatography A 1124, 3e10. Darbre, P.D., Aljarrah, A., Miller, W.R., Coldham, N.G., Sauer, M.J., Pope, G.S., 2004. Concentrations of Parabens in human breast tumours. Journal of Applied Toxicology 24, 5e13. Ferna´ndez-Alba, A.R., Garcı´a-Reyes, J.F., 2008. Large-scale multiresidue methods for pesticides and their degradation products in food by advanced LC-MS. Trends in Analytical Chemistry 27, 973e990. Ferrer, I., Garcı´a-Reyes, J.F., Ferna´ndez-Alba, A., 2005. Identification and quantitation of pesticides in vegetables by liquid chromatography time-of-flight mass spectrometry. Trends in Analytical Chemistry 24, 671e682. Gimsing, A.L., Sørensen, J.C., Tovgaard, L., Jørgensen, A.M.F., Hansen, H.C.B., 2006. Degradation kinetics of glucosinolates in soil. Environmental Toxicology and Chemistry 25, 2038e2044.
6780
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 7 0 e6 7 8 0
Golden, R., Gandy, J., Vollmer, G., 2005. A review of the endocrine activity of parabens and implications for potential risks to human health. Critical Reviews in Toxicology 35, 435e458. Go´mez, M.J., Go´mez-Ramos, M.M., Malato, O., Mezcua, M., Fe´rnandez-Alba, A.R., 2010. Rapid automated screening, identification and quantification of organic microcontaminants and their main transformation products in wastewater and river waters using liquid chromatographyquadrupole-time-of-flight mass spectrometry with an accurate-mass database. Journal of Chromatography A 1217, 7038e7054. Gonza´lez-Marin˜o, I., Quintana, J.B., Rodrı´guez, I., Cela, R., 2009. Simultaneous determination of parabens, triclosan and triclocarban in water by liquid chromatography/electrospray ionisation tandem mass spectrometry. Rapid Communications in Mass Spectrometry 23, 1756e1766. Herna´ndez Leal, L., Vieno, N., Temmink, H., Zeeman, G., Buisman, C.J.N., 2010. Occurrence of Xenobiotics in gray water and removal in three biological treatment systems. Environmental Science and Technology 44, 6835e6842. Herna´ndez, F., Bijlsma, L., Sancho, J.V., Dı´az, R., Iba´n˜ez, M., 2011. Rapid wide-scope screening of drugs of abuse, prescription drugs with potential for abuse and their metabolites in influent and effluent urban wastewater by ultrahigh pressure liquid chromatography-quadrupole-timeof-flight-mass spectrometry. Analytica Chimica Acta 684, 96e106. ISO, International Standards Organization, 1994. ISO-7827: Evaluation in an aqueous medium of the “ultimate” aerobic biodegradability of organic compounds e Method by analysis of dissolved organic carbon (DOC). Jonkers, N., Kohler, H.-P.E., Dammsha¨user, A., Giger, W., 2009. Mass flows of endocrine disruptors in the Glatt River during varying weather conditions. Environmental Pollution 157, 714e723. Kasprzyk-Hordern, B., Dinsdale, R.M., Guwy, A.J., 2008. The effect of signal suppression and mobile phase composition on the simultaneous analysis of multiple classes of acidic/neutral pharmaceuticals and personal care products in surface water by solid-phase extraction and ultra performance liquid chromatography-negative electrospray tandem mass spectrometry. Talanta 74, 1299e1312. Kasprzyk-Hordern, B., Dinsdale, R.M., Guwy, A.J., 2009. The removal of pharmaceuticals, personal care products, endocrine disruptors and illicit drugs during wastewater treatment and its impact on the quality of receiving waters. Water Research 43, 363e380.
Kosjek, T., Heath, E., 2008. Applications of mass spectrometry to identifying pharmaceutical transformation products in water treatment. Trends in Analytical Chemistry 27, 807e820. Lee, H.B., Peart, T.E., Svoboda, M.L., 2005. Determination of endocrine-disrupting phenols, acidic pharmaceuticals, and personal-care products in sewage by solid-phase extraction and gas chromatography-mass spectrometry. Journal of Chromatography A 1094, 122e129. Lundov, M.D., Moesby, L., Zachariae, C., Johansen, J.D., 2009. Contamination versus preservation of cosmetics: a review on legislation, usage, infections, and contact allergy. Contact Dermatitis 60, 70e78. Meyer, B.K., Ni, A., Hu, B., Shi, L., 2007. Antimicrobial preservative use in parenteral products: past and present. Journal of Pharmaceutical Sciences 96, 3155e3167. Oppenheimer, J., Stephenson, R., Burbano, A., 2007. Characterizing the passage of personal care products through wastewater treatment processes. Water Environment Research 79, 2564e2577. Ort, C., Lawrence, M.G., Reungoat, J., Mueller, J.F., 2010. Sampling for PPCPs in wastewater systems: comparison of different sampling modes and optimization strategies. Environmental Science and Technology 44, 6289e6296. Routledge, E.J., Parker, J., Odum, J., Ashby, J., Sumpter, J.P., 1998. Some alkyl hydroxy benzoate preservatives (parabens) are estrogenic. Toxicology and Applied Pharmacology 153, 12e19. SciFinder Scholar Database: http://www.cas.org/products/sfacad/ index.html (accessed March 2011). Soni, M.G., Carabin, I.G., Burdock, G.A., 2005. Safety assessment of esters of p-hydroxybenzoic acid (parabens). Food and Chemical Toxicology 43, 985e1015. Terasaki, M., Kamata, R., Shiraishi, F., Makino, M., 2009a. Evaluation of the estrogenic activity of parabens and their chlorinated derivatives by using yeast two-hybrid assay and the enzyme-linked immunosorbent assay. Environmental Toxicology and Chemistry 28, 204e208. Terasaki, M., Makino, M., Tatarazako, N., 2009b. Acute toxicity of parabens and their chlorinated by-products with Daphnia magna and Vibrio fischeri bioassays. Journal of Applied Toxicology 29, 242e247. Villaverde-de-Sa´a, E., Gonza´lez-Marin˜o, I., Quintana, J.B., Rodil, R., Rodrı´guez, I., Cela, R., 2010. In-sample acetylation-non-porous membrane-assisted liquid-liquid extraction for the determination of parabens and triclosan in water samples. Analytical and Bioanalytical Chemistry 397, 2559e2568. Ying, G.G., Kookana, R.S., Ru, Y.J., 2002. Occurrence and fate of hormone steroids in the environment. Environmental International 28, 545e551.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 8 1 e6 7 8 8
Available online at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Transformation of arsenic in offshore sediment under the impact of anaerobic microbial activities Liying Xu, Zhixi Zhao, Shaofeng Wang, Rongrong Pan, Yongfeng Jia* Key Laboratory of Pollution Ecology and Environmental Engineering, Institute of Applied Ecology, Chinese Academy of Sciences, Shenyang 110016, China
article info
abstract
Article history:
Sediment bound arsenic usually undergoes phase transformation processes when it is
Received 12 August 2011
transported and buried in deeper settings. This work investigated anaerobic microbial
Received in revised form
mediated speciation change of the arsenic in offshore sediment and monitored the
8 October 2011
transformation process of oxyhydroxide associated arsenate to sulfide associated forms.
Accepted 19 October 2011
The fate of arsenic and possible pathways of transformation were discussed based on
Available online 28 October 2011
quantitative analysis of aqueous and solid arsenic and iron, and qualitative characterization using X-ray absorption near edge spectroscopy (XANES). Arsenic was released and
Keywords:
reduced upon development of anoxic conditions but was resequestered by authigenic
Arsenic
minerals later. Most of the arsenic in the sediment was converted to orpiment-like
Speciation
material. Sulfide may have played double roles in arsenic redistribution process, i.e.
Sediment
promoting arsenic release from host oxyhydroxides in early stage and removal of arsenite
Microorganisms
from solution in the form of arsenic sulfide in later stage. The findings have implications
Sulfide
about the pathways of arsenic transformation when arsenate is transported and buried below redox boundaries in offshore sediment. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
Most of the arsenic entering into aquatic environment either from discharging of wastewater or leaching of arsenic-bearing solid wastes accumulates in sediment. Under the impact of physicochemical and/or microbial factors, arsenic would undergo consecutive and/or parallel biogeochemical processes such as adsorption/desorption, precipitation or coprecipitation, oxidation/reduction, etc. (Smedley and Kinniburgh, 2002). These processes transform arsenic speciation from one form to another, hence leading to its sequestration or mobilization due to varying stability of different arsenic speciation in aquatic environment (Burton et al., 2011). Arsenic in aquatic and sedimentary environments may exist in one or more of the following oxidation states, þ5, þ3,
1 depending on redox conditions (Oremland and Stolz, 2003). As(V) is the predominant arsenic form in oxidizing environments, whereas As(III) and As(-I) predominates in reducing environments. As(V) is usually associated with various oxide or clay minerals in the form of adsorbed and coprecipitated arsenate, while As(III) may be present as arsenic sulfide in addition to adsorbed and coprecipitated forms. Arsenic may also be present as As(-I) minerals (e.g. arsenopyrite) in sulfidic environments. When redox condition changes from oxic-suboxic to anoxic-sulfidic settings, microbial reduction will be the dominating factor controlling arsenic speciation. The most important oxidation state conversions are microbial mediated reduction of As(V) to As(III) (or even further down to As(-I)) and Fe(III) to Fe(II). When the adsorbed and/or coprecipitated As(V)
* Corresponding author. Tel./fax: þ86 2483970503. E-mail address:
[email protected] (Y. Jia). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.10.041
6782
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 8 1 e6 7 8 8
is reduced to As(III), it may be mobilized due to weaker affinity of minerals for As(III) (Manning and Goldberg, 1997; Manning et al., 1998; Goldberg, 2002). Microbial reduction of Fe(III) to Fe(II) may result in dissolution of iron oxyhydroxides, hence leading to the release of associated As(V) or As(III) into aqueous phase (Cummings et al., 1999). If the media contains abundant sulfate (as in the case of seawater), the behavior of arsenic in sediment becomes more complicated (Kirk et al., 2010; Burton et al., 2011). When the environment changes from oxic-suboxic to anoxic-sulfidic condition, sulfate is reduced to sulfide, leading to the formation of FeS, pyrite, or arsenic sulfide precipitate (O’Day et al., 2004). The authigenic iron sulfide may act as scavenger for arsenic. Arsenopyrite may also be formed in case of As(III) being further reduced to As(-I) (Bostick and Fendorf, 2003). There have been extensive studies on microbial reduction and mobilization of iron oxyhydroxides adsorbed arsenic and some works on the behavior of arsenic in sediment upon anaerobic microbial activities. Arsenic is mainly associated with iron oxyhydroxides in subsurface sediment of Meghna riverbank in Bangladesh (Datta et al., 2009), while in shallow aquifer sediment infiltrated with seawater of San Francisco Bay, realgar is dominant arsenic species (O’Day et al., 2004). In intertidal sediment, arsenic is associated mainly with iron oxides in shallow sediment and with sulfide minerals in deeper sediment (Masuda et al., 2005). It is important to shed more light on the behavior and the fate of arsenic in offshore sediment of seawater systems, especially the pathways of arsenic speciation conversion from oxyhydroxide-associated to sulfide-associated forms. The objective of the present work is to study the transformation process and the fate of arsenic in shallow offshore sediment under the effects of anaerobic microorganisms.
2.
Experimental section
2.1.
Sediment collection
Surface sediment that contains w735 mg kg1 of As (equivalent to 9.8 mmol kg1, determined after digestion in aqua regia) was collected from Jinzhou Bay in Northeastern China. This bay is heavily polluted by a variety of toxic heavy metals due to the operation of a zinc smelter. The sampling site is close to the outlet stream of the wastewater from the smelter. The sediment sample was collected using a gravity sediment corer. The top 10-cm sediment was retrieved and stored in PE bottles while purging with N2. The sediment sample was transported to lab right away after collection. The sediment sample was stored at 20 C in a freezer before use.
2.2.
Bacterial inocula
The sediment was mixed with artificial seawater media (ASM) with lactate as carbon source at liquid/solid ratio of 100/1 (120 ml of seawater, 1.2 g of sediment) and incubated at 30 C for 24 h for the enrichment of sediment microorganisms. The ASM was made with the following constituents (g L1): MgCl2$6H2O, 0.2; KH2PO4, 0.5; NaCl, 23; Na2SO4, 3; NH4Cl, 1; CaCl2$2H2O, 0.1; yeast extract, 0.25; sodium lactate, 6; adjusted
to pH 7.0 using HCl. The mixture was incubated in the dark at 30 C. After 24 h enrichment, 0.4 mL of the initial slurry was transferred to another fresh 10 mL anoxic ASM, and incubated for another 24 h. After two times of such transfers and incubations, a stable enrichment culture was obtained. All experiments were performed under strict anoxic conditions using the Hungate technique. All transfers were made using needles and syringes.
2.3.
Incubation of the sediment
The sediment was sterilized by autoclaving and mixed with artificial seawater media (ASM) at solid/liquid ratio of 1/100 in 50-mL serum bottles. The concentration of total As in the systems was 98 mM. The incubations were inoculated with 1% (v/v) of inoculums in each culture serum bottle. The abiotic controls were autoclaved at 121 C for 15 min after inoculation of microcosms. Autoclaving treatment of the sediment may potentially cause mineralogical changes of the minerals, especially conversion of amorphous ferric oxyhydroxide to more stable forms. According to a previous report, no recrystallization was observed using XRD when ferrihydrite was subjected to short time (e.g. 15 min) autoclaving treatment (Zhang et al., 2008). The biotic trials and the abiotic controls were then incubated in the dark at 30 C in a rotary shaker. All trials were conducted in triplicate under strict anoxic conditions. At different time intervals, the serum bottles were removed from the incubator and transferred to an anaerobic (ultra pure N2) glove chamber. The slurries were filtered through 0.22-mm filters and the solids were extracted separately with 1 M phosphate (pH 5) for 24 h and 1 M HCl for 2 h. The filtrates and the extractants were analyzed for As and Fe concentrations. All concentration data presented in this work are the averages of triplicate incubations.
2.4.
Analysis of arsenic, iron and sulfide concentration
The concentrations of total arsenic and As(III) were analyzed according to previously reported procedures (Maity et al., 2004; Zhang et al., 2008; Zhao et al., 2011). An atomic fluorescence spectrometer coupled with a hydride generator (HGAFS) was used and the detection limit for As was 0.01 mg L1. The samples were pretreated with an ascorbic acid/thiourea reducing agent (5 g of thiourea and 5 g of ascorbic acid in 100 mL of H2O) to reduce all arsenic to As(III) prior to hydride generation. Borohydride solution was prepared by dissolving 2 g of KBH4 in 100 mL of 0.5% (w/v) NaOH solution and was used as reducing agent for AsH3 generation. 5% HCl solution was used as carrier solution. To selectively determine the concentration of As(III) in solution, the sample was not pretreated with ascorbic acid/ thiourea reducing agent, but mixed with a pH 4.5 sodium citrate buffer (0.4 M). The sodium citrate buffer was used as carrier solution instead of HCl solution. Total Fe concentration was determined using a flame atomic absorption spectrophotometer (AA240, Varian) with detection limit of 0.05 mg/L. The concentration of Fe(II) was determined by light absorbance measurement at 510 nm after
6783
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 8 1 e6 7 8 8
2.5.
XANES spectroscopy
All samples for XANES measurements were handled in an anaerobic (ultrapure N2) glove chamber, freeze-dried, and stored in N2 atmosphere. The samples for As XANES measurements were sealed and pressed mechanically between two layers of Kapton tape under the protection of N2. The samples were transported to synchrotron facilities in N2 atmosphere. Arsenic K-edge XANES spectra were collected on beamline U7C (XAFS station) at National Synchrotron Radiation Laboratory (NSRL) of China. The NSRL storage ring was operated at 0.8 GeV and the ring currents of 300 mA. The monochromator was a fixed exit double crystal monochromator with Si(111) crystals. Data was acquired in transmission mode. Sulfur K-edge XANES spectra were collected on the mid-energy X-ray station at Beijing Synchrotron Radiation Facilities with a storage ring of 2.2 GeV and 100 mA. The measurements were carried out at ambient temperature under high vacuum (106108 mbar). Sodium arsenate, sodium arsenite, realgar, orpiment and arsenopyrite were chosen as reference compounds for As XANES measurements while ferric sulfate, ferrous sulfate, elemental sulfur, realgar, orpiment, arsenopyrite, pyrite and amorphous FeS were used as references for S XANES characterization. Amorphous FeS was prepared according to the method reported in literature (Wolthers et al., 2005). Realgar, orpiment, arsenopyrite and pyrite were sampled from the mineral specimens collected in our lab. Small pieces of mineral specimens were crushed before use. All other chemicals were purchased from Sinopharm Chemical Reagent Co. These reagents were of analytical grade and used without further purification.
3.
Results
3.1.
Dissolved As
Sediment bound As was destabilized upon anaerobic incubation. As started to release from the sediment to aqueous phase within hours of incubation (Fig. 1). Dissolved As accumulated rapidly in solution and the distribution of As(III) and As(V) varied with incubation time. The concentration of total dissolved As reached a maximum of w37 mM in 24 h, which accounted for w38% of As in the sediment. In comparison, the abiotic control released only 16 mM As, accounting for w16% of total As in the sediment, probably from desorption of loosely bound species by ASM. Total dissolved As remained nearly constant at the maximum for 3 days, followed by a fast decrease thereafter, indicating the released As being resequestered by the solid phase. As(V) reduction occurred within several hours of incubation experiment. As(V) concentration increased in the first day of incubation, after that As(V) diminished quickly from
50 As concentration in solution (µM)
complexing with 1,10-phenanthroline using a Shimadzu UV2550 spectrophotometer (Zhao et al., 2011). The concentration of aqueous sulfide was analyzed using methylene blue method (Eaton et al., 1995), on a Shimadzu UV-2550 spectrophotometer.
total As As(III) As(III)-CK As(V) As(V)-CK
40
30
20
10
0
0
50
100
150
200
250
Time (h)
Fig. 1 e Change of dissolved As concentration with incubation time (CK: abiotic control).
aqueous phase, while at the same time dissolved As(III) concentration increased rapidly and peaked at day 4. As(III) dominated dissolved As species thereafter but was quickly removed from solution. The concentration of As(III) fell quickly from w33 mM at the peak to w15 mM after 8 days.
3.2.
Phosphate extracted As
Phosphate extraction is commonly used to determine strongly adsorbed As in soils and sediment (Keon et al., 2001). Adsorbed As species accounted for w25% of total As in starting material (w9.8 mmol kg1) and w80% of which was present as As(V). The concentration of adsorbed As(V) in the incubated solids decreased sharply from around 2.2 to 0.5 mmol kg1 after 4 days of incubation. Desorbed As was not accumulated in solution since the concentration of total dissolved As changed little during this period (constant at the plateau of w37 mM, see Fig. 1). The concentration of adsorbed As(III) in solid throughout the incubation course remained nearly unchanged at a level similar to the negative control (Fig. 2). Hence, As(III) removal from solution after day 4 (Fig. 1) was due to precipitation or coprecipitation with authigenic minerals rather than readsorption onto the solid phases.
3.3.
HCl extracted As
HCl-extraction is employed to mainly target the As coprecipitated with very amorphous Fe oxyhydroxides, AVS and carbonate etc. (Keon et al., 2001). HCl-liberated H2S may react with As(III) to form arsenic sulfide precipitate (Wilkin and Ford, 2002). To minimize this effect, HgCl2 was introduced into the extraction system according to recent literature (Huang and Kretzschmar, 2010). In the abiotic controls, HClextracted As was w3.0 mmol kg1 accounting for 30% of total As in the sediment (9.8 mmol kg1) (Fig. 2), while the remaining 70% of As occurred as non HCl-dissolvable phase. Reductive dissolution of crystalline Fe oxyhydroxides may release associated As. Anaerobic incubation of the sediment remarkably increased HCl-extracted As(V) in the initial stage,
6784
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 8 1 e6 7 8 8
300
Fe concentration in solution (µM)
A
a
250 200 150 100
Fe(II) Fe(II)-CK
50 0 0
50
100
B
3.4.
Dissolved and HCl extracted Fe
Reduction and subsequent dissolution of Fe(III) oxyhydroxides occurred during the incubation process (Fig. 3). Dissolved Fe(II) concentration increased notably and reached a plateau of w220 mM after w5 days of incubation. HCl-extracted Fe(III) diminished rapidly while there was an apparent Fe(II) accumulation in solid phase concurrently. Although most HCl-
250
300
240
180 total Fe Fe(II) Fe(III)
120
60
0
reaching a maximum of w6 mmol kg1 on day 4. This suggests that the released As(V) has been significantly incorporated into newly formed HCl-dissolvable minerals. The sum of HClextracted and aqueous forms was 98e108% of total As in the sediment from day 2 to day 4. This is indicative that the As associated with crystalline Fe oxyhydroxides or other minerals, which are refractory to acid attack, was completely activated. HCl-extracted As(V) decreased gradually after day 4. It may be reduced to As(III) and incorporated into non-HCl leachable phases, e.g. pyrite or As sulfide. HCl-extracted As(III) was nearly constant throughout the incubation course. Hence, it is suggested that the removed As(III) from solution after 4 days (Fig. 1) was precipitated as non-HCl leachable phases instead of coprecipitation with amorphous Fe oxyhydroxides and AVS.
200
b
-1
Fe concentration in solid (mmol kg )
300
Fig. 2 e Change of the concentration of (A) phosphateextracted As and (B) HCl-extracted As in solid with incubation time (CK: abiotic control).
150 Time (h)
0
50
100
150
200
250
300
Time (h)
Fig. 3 e Change of the concentration of dissolved Fe in (a) aqueous phase and (b) HCl-extracted solid (CK: abiotic control).
extractable Fe(III) was reduced to Fe(II), only small fraction (w10%) of Fe(II) was released into solution. It is also interesting to note that HCl-extracted total Fe also increased during the anaerobic incubation process. This suggests that non HClextractable Fe(III) oxyhydroxides was reduced and transformed to Fe(II)-containing minerals such as authigenic amorphous FeS and green rust etc.
3.5.
As and S oxidation states
The change of As and S oxidation state in solid phase during the incubation process was analyzed using As and S K-edge XANES spectroscopy. Arsenic XANES spectra of the sediment samples are compared with those of reference compounds in Fig. 4. The absorption peaks of the reference compounds are easily distinguishable from one another. The peak positions are in following order: arsenate > arsenite > orpiment > realgar > arsenopyrite, with the difference of w3.0, w1.5, w0.8 and w0.5 eV respectively. This is in good agreement with previous reports (O’Day et al., 2004; Beauchemin and Kwong, 2006). Arsenic XANES spectrum of the as-received sediment shows a well resolved absorption peak located at the position
6785
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 8 1 e6 7 8 8
AR
H
Incu 60d
J
Incu 20d
I H
G
As2S3
Intensity
E
As2S3 AsS
D
Na3AsO3 Na3AsO4
B A
11860
S
E
FeAsS
C
FeSO4
F
AsS
D
Fe2(SO4)3
G
AR F
Incu 20d
FeAsS
C
FeS2
B
FeS
A 11870
11880
11890 2460
2470
2480
2490
Photon Energy (eV)
Photon energy (eV)
Fig. 4 e Left: Arsenic XANES spectra: (A) arsenate; (B) arsenite; (C) arsenopyrite; (D) realgar; (E) orpiment; (F) as-received sediment; (G) the sediment incubated at 30 C for 20 days; (H) the sediment incubated at 30 C for 60 days. Right: Sulfur XANES spectra: (A) disordered mackinawite; (B) pyrite; (C) arsenopyrite; (D) realgar; (E) orpiment; (F) elemental sulfur; (G) ferrous sulfate; (H) ferric sulfate; (I) the sediment incubated at 30 C for 20 days; (J) as-received sediment.
similar to that of sodium arsenate, suggesting that arsenate is the principal As species in the fresh sediment. Anaerobic incubation apparently changed As speciation in solid phase as indicated by absorption peak shift. After 20 days of incubation, a new band emerges at the position similar to that of orpiment, indicating the development of As2S3 phase. No clearly visible evidence can be observed for the presence of arsenite in the solid as major phase. This is supported by phosphate and HCl-extraction data (Fig. 2) which demonstrate that arsenite is not significantly sequestered as adsorbed or coprecipitated forms in the process of incubation. After two months of incubation, As(V) band almost disappears and the XANES spectrum is dominated by a well resolved absorption peak at orpiment position. This suggests that As2S3 phase has become the principal As species in the solid phase. FeAsS is less likely to be a part of As phases in the incubated sediment due to the big difference in absorption peak positions. Detection of As sulfide minerals using XRD was unsuccessful probably due to being minor phases and nanoparticulates in nature. Arsenic sulfide was not detected using XRD in previous work either (O’Day et al., 2004). Sulfur K-edge XANES spectra of sediment are compared with those of reference compounds in Fig. 4. The S in the asreceived sediment is present predominantly as sulfate with minor amount of elemental S-like phase. After 20 days of incubation, the intensity of sulfate band is significantly reduced while that of elemental S-like phase is markedly increased. A strong shoulder emerges at lower energy position
similar to amorphous FeS. No visible evidence is observed from S XANES spectra for the presence As sulfides. The absorption bands of the minor phases may have been obscured by the strong peaks of abundant FeS and elemental S.
4.
Discussion
4.1.
Mobilization of As
The mobility of As in aqueous environment is largely controlled by the redox conditions because As speciation and the ways of its sequestrations vary significantly when the environment shifts from oxic to anoxic conditions. In oxicsuboxic environment, dissolved As is sequestered by soil/ sediment mainly via adsorption on and/or incorporation into iron oxyhydroxides with different binding strength for arsenate and arsenite, whereas in anoxic-sulfidic environment, dissolved As may be sequestered by Fe sulfide minerals or via formation of As sulfide and arsenopyrite depending on the activities of various constituents and reaction kinetics (O’Day et al., 2004). Whether As is mobilized when the sediment is subjected to anaerobic incubation depends its speciation. Arsenic will not be released if it is mainly associated with sulfide minerals (Polizzotto et al., 2006). In present work, As was rapidly released from the sediment when anoxic condition was developed within 24 h as a result of anaerobic incubation. This
6786
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 8 1 e6 7 8 8
suggests that As in the as-received sediment is less likely associated with sulfide minerals but probably with Fe oxyhydroxides minerals. Arsenic is present mainly as As(V) although shallow sediment is very likely under suboxic conditions. Surface adsorbed and very amorphous Fe oxyhydroxides coprecipitated As accounts for only w30% of total As in the sediment. The remaining large fraction of As can only be released under reducing conditions. In Red River sediment where suboxic conditions prevails, very little As was leached by HCl. Most of As was liberated by ascorbic reduction of host Fe oxides (Postma et al., 2010). Microbial mediated Fe(III) and As(V) reduction may lead to As dissociation from the host minerals and release into aqueous phase (Smedley and Kinniburgh, 2002). This is a proposed common mechanism for elevated concentration of As in groundwater. Reduction of Fe(III) may result in either dissolution of host Fe oxyhydroxides or conversion to low surface area oxide minerals hence releasing their associated As. Reduction of As(V) to As(III) may also cause its release since As(III) is well known for its weaker affinity to various minerals in environment compared to As(V). For the surface-bound As(V), its reduction may proceed before Fe(III) reduction, whereas the As(V) incorporated in the structure of Fe oxyhydroxides is released first as a result of reductive dissolution of Fe followed by aqueous reduction to As(III) (Postma et al., 2010). Rapid increase of aqueous As(V) within 24 h is indicative of the liberation of arsenate from the structure of Fe oxyhydroxides, and/or desorption. However, reductive dissolution of host Fe oxyhydroxides alone cannot explain the significant release of arsenate because Fe(II) was not appreciably released to the solution within 24 h. Even after 10 days, only w10% of reduced Fe (HCl-extracted) was solubilized while most Fe(II) was still retained in solid phase. Reduction of Fe(III) may transform Fe oxyhydroxides to magnetite and/or green rust due to re-incorporation of the produced Fe(II) and most of their associated As can still be uptaken by the newly formed Fe oxide minerals (Saalfield and Bostick, 2009; Coker et al., 2006; Islam et al., 2005; Pedersen et al., 2006). In case there is excess sulfate and a sulfidic environment is developed, the produced Fe(II) will react with sulfide to precipitate amorphous FeS. Compared to Fe oxyhydroxides (magnetite, ferrihydrite, etc.), FeS sorbs much less As although it is considered as an important scavenger for trace elements in anoxic environments (Wolthers et al., 2007; Kocar et al., 2010). Recent studies suggested the triggering effect of sulfide on recrystallization of poorly ordered ferric (hydr)oxides to more ordered forms of (hydr)oxides (Kocar et al., 2010; Burton et al., 2011), hence leading to reduced surface area and sorption capacity. Sulfide can also act as a strong reductant and participate in Fe(III) reduction, hence accelerating Fe oxyhydroxides dissolution and subsequent As release. Additionally, it was reported that the produced sulfide could displace sorbed arsenite from the surface of Fe oxyhydroxides (Kocar et al., 2010). Accumulation of sulfide in aqueous phase may lead to the formation of soluble thioarsenic which also have mobilization effect on arsenic (Burton et al., 2011). Thereby, sulfate reduction and subsequent FeS precipitation may have significantly contributed to As release from the sediment. Enhanced As (mainly As(V)) release from HFO-As(V) was also reported when microbial sulfate reduction was stimulated
followed by FeS precipitation (Kocar et al., 2010). Arsenic redistribution process and the underlined mechanism for static systems may be different from those under flow-through conditions. A recent study using advective-flow columns which simulated subsurface ferrihydrite-rich environments with advective groundwater flow showed that microbial sulfidogenesis triggered significant As mobilization (Burton et al., 2011). The authors suggested that Fe(II) catalyzed transformation of ferrihydrite to goethite, rather than mackinawite formation, played a key role in As mobilization.
4.2.
As sequestration by Fe sulfide
Arsenic was resequestered into solid phase in two consecutive steps, both involving authigenic sulfide minerals formation, i.e. coprecipitation with FeS followed by precipitation of As sulfide. In the first 3 days, As(V) was released in large quantities from solid phase to solution. In addition to partial reduction to As(III) (Fig. 1), As(V) was significantly incorporated into newly formed HCl-dissolvable minerals (Fig. 2). This coprecipitated amount increased from w20% of total As in the beginning to w60% at day 3e4. The HCl-dissolvable authigenic minerals formed in the anaerobic incubation process is mainly FeS and other AVS. FeS is the first authigenic mineral to precipitate during sulfidization of As-containing Fe oxyhydroxides and is often called as nanoparticulate or disordered mackinawite, amorphous FeS (Wolthers et al., 2007). S K-shell XANES results in present study suggest the formation of amorphous FeS after anaerobic incubation of the sediment. Mackinawite is well known for its capability of sequestrating As via adsorption or coprecipitation in sulfidic environments (Postma et al., 2010; Farquhar et al., 2002; Wolthers et al., 2005). Arsenic species coprecipitated into the authigenic amorphous FeS is dominated by As(V) (Fig. 2). Previous laboratory tests also suggested disordered mackinawite having much stronger affinity and sorption capacity for As(V) compared to As(III) (Wolthers et al., 2007, 2005). Thus, As(III) may show higher mobility than As(V) in sulfidic sediment provided its concentration not exceeding As sulfide solubility. In lake sediment, As(III) is difficult to be sorbed by FeS and As removal does not occur at the depth where arsenite dominate As speciation (Couture et al., 2010). In comparison, As(V) can be easily sorbed by poorly crystalline mackinawite and this is proposed as a plausible mechanism of As removal in sulfidic waters (Couture et al., 2010).
4.3.
As sequestration by As sulfide
Removal of As(III) from solution occurred after w4 days of incubation. Aqueous As(III) was not readsorbed to the solid or coprecipitated with newly formed AVS in appreciable quantity. It was removed from solution mainly via formation As sulfide precipitate. Precipitation of As sulfide usually occurs in sulfidic waters provided that As(III) and S(-II) concentrations exceed the solubility of As sulfide. Although As(III) was accumulated in solution, formation of As sulfide did not occur in the early stage (before day 4) due to unsaturated state with respect to As sulfide. Disordered mackinawite is easier to precipitate than As sulfide. In an Fe(II)e As(III)eS(-II) system, disordered mackinawite is the initial
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 8 1 e6 7 8 8
precipitate to form (Kirk et al., 2010). In addition to precipitating Fe(II), S(-II) is also involved in Fe(III) reduction. Thus, aqueous S(-II) is continuously consumed in the reactions of FeS precipitation and Fe(III) reduction. This will limit rapid accumulation of S(-II) in solution, hence making S(-II) concentration unable to rise quickly to the level of As sulfide saturation until these reactions reaching equilibrium (Kocar et al., 2010). Fe(II) concentrations in both aqueous phase and HCl-extracted phase (mainly AVS) gradually reached a plateau after 4 days of incubation (Fig. 3). This is indicative of FeS precipitation reaction reaching equilibrium, thus allowing rapid accumulation of S(-II) ions in solution. Coinciding with this observation, dissolved As(III) concentration started to decrease at similar time, suggesting occurrence of As sulfide precipitation. The saturation state of the system with respect to arsenic sulfide at different incubation time was estimated using Geochemist’s Workbench (Kirk et al., 2010; Burton et al., 2011) (see Fig. 5). Arsenic sulfide was not saturated in the first three days of incubation. With the accumulation of As(III) and S(-II) in aqueous phase, the system was approaching saturation state of amorphous As2S3 on the fourth day, indicating the beginning of As2S3(am) precipitation. In present work, it is proposed from XANES evidence that As2S3 is the major As sulfide phase precipitated. It was also reported in previous work that precipitation of orpiment-like mineral was the major mechanism for As sequestration in an estuarine sediment (Bostick et al., 2004). In comparison, realgar was identified as the dominant As species in the aquifer sediment infiltrated with marine water of San Francisco Bay (O’Day et al., 2004). It is predicted based on a thermodynamic model that realgar is the first As sulfide species to precipitate in high-Fe/low-S reducing environments whereas in Fe-poor/S-rich environments orpiment will precipitate instead (O’Day et al., 2004). In a very recent study, orpiment
was suggested to be much less soluble than realgar and its precipitation should proceed initially (Kirk et al., 2010).
4.4.
-
H2AsS3
log a H2 S(aq)
-3
As2 S3 (am)
-4
(1) Arsenic in the sediment was released and reduced to arsenite upon the impact of anaerobic microbial activities. Ferric iron was also reduced to ferrous and released to aqueous phase but was precipitated in the form of FeS as a result of sulfate reduction. (2) The released arsenic was resequestered in later stage probably in the form of As2S3(am) after sulfide was accumulated to the saturation state of arsenic sulfide.
-5 -6
Realgar(alpha)
-7
H3 AsO3 (aq)
-8 -9 -10 -8
Conclusions
This work investigated the behavior and fate of arsenic in shallow offshore sediment of seawater systems under the effects of anaerobic microorganisms. Possible mechanism involved and the pathways of arsenic transformation were also discussed. The major findings are:
-1 -2
Environmental relevance
The present work documented a transformation process of Fe oxyhydroxides associated As to sulfide associated As upon environment shifting from oxic-suboxic to anoxic-sulfidic condition in a sulfur-rich system, e.g. offshore sediment. This simulates the microbial mediated conversion process when As-bearing minerals are transported and buried in deeper sediment. Arsenic is mainly present as As(V) and associated with Fe oxyhydroxides in shallow sediment. When it is buried in deeper settings, As(V) will be dissociated from host Fe oxyhydroxides and partly reduced to As(III). Arsenic is then incorporated into secondary hosts, the authigenic minerals including disordered mackinawite and As sulfide and finally transformed to orpiment-like materials. The present study also sheds more light on the pathways of As redistribution process for the field observations: In shallow intertidal sediment, As is associated mainly with Fe oxyhydroxides, while in deeper settings As is principally sequestered by sulfide minerals (Masuda et al., 2005). Arsenic is reduced and released to water when shallow Amazon sediment are buried, and then resequestered from porewater (Sullivan and Aller, 1996), indicating having undergone a similar speciation conversion process as discussed above. The As associated with Fe oxyhydroxides in reducing sulfide freshwater sediment was also reported to be assimilated into diagenetic sulfide minerals (Moore et al., 1988). However, in sulfur-poor environment, arsenic may undergoes different process and transforms to different species, i.e. not mainly associating with sulfide minerals.
5. 0
6787
Claudetite -7
-6
-5
-4
-3
-2
-1
0
log a H 3 AsO 3 (aq) Fig. 5 e Calculated arsenic speciation diagram of As(III)eS(II) system at pH 7 using Geochemist’s Workbench. The dots show As(III) speciation at different incubation time (from left: 0, 24, 96, 168 h).
Acknowledgements The financial supports by NSFC (40803032, 40925011, 41073086) are acknowledged. The authors also thank Drs. Shiqiang Wei, Bo He, Zhi Xie and Danhong Zhang at NSRL and Drs. Yidong Zhao and Chenyan Ma at BSRL for their assistances with XANES measurements.
6788
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 8 1 e6 7 8 8
references
O’Day, P.A., Vlassopoulos, D., Root, R., Rivera, N., 2004. The influence of sulfur and iron on dissolved arsenic concentrations in the shallow subsurface under changing redox conditions. Proceedings of the National Academy of Sciences of the United States of America 101 (38), 13,703e13,708. Beauchemin, S., Kwong, Y.T.J., 2006. Impact of redox conditions on arsenic mobilization from tailings in a wetland with neutral drainage. Environmental Science & Technology 40 (20), 6297e6303. Bostick, B.C., Fendorf, S., 2003. Arsenite sorption on troilite (FeS) and pyrite (FeS2). Geochimica et Cosmochimica Acta 67 (5), 909e921. Bostick, B., Chen, C., Fendorf, S., 2004. Arsenite retention mechanisms within estuarine sediment of Pescadero. CA. Environmental Science & Technology 38 (12), 3299e3304. Burton, E.D., Johnston, S.G., Bush, R.T., 2011. Microbial sulfidogenesis in ferrihydrite-rich environments: effects on iron mineralogy and arsenic mobility. Geochimica et Cosmochimica Acta 75 (11), 3072e3087. Coker, V.S., Gault, A.G., Pearce, C.I., Van Der Laan, G., Telling, N.D., Charnock, J.M., Polya, D.A., Lloyd, J.R., 2006. XAS and XMCD evidence for species-dependent partitioning of arsenic during microbial reduction of ferrihydrite to magnetite. Environmental Science & Technology 40 (24), 7745e7750. Couture, R.-M., Gobeil, C., Tessier, A., 2010. Arsenic, iron and sulfur co-diagenesis in lake sediment. Geochimica et Cosmochimica Acta 74 (4), 1238e1255. Cummings, D.E., Caccavo Jr., F., Fendorf, S., Rosenzweig, R.F., 1999. Arsenic mobilization by the dissimilatory Fe(III)reducing bacterium Shewanella alga BrY. Environmental Science & Technology 33 (5), 723e729. Datta, S., Mailloux, B., Jung, H.-B.M., Hoque, A., Stute, M., Ahmed, K.M., Zheng, Y., 2009. Redox trapping of arsenic during groundwater discharge in sediment from the Meghna riverbank in Bangladesh. Proceedings of the National Academy of Sciences of the United States of America 106 (40), 16930e16935. Eaton, A.D., Clesceri, L.S., Greenberg, A.E., 1995. Standard Methods for The Examination of Water and Wastewater. American Public Health Association, American Water Works Association, and Water Environmental Federation. Farquhar, M.L., Charnock, J.M., Livens, F.R., Vaughan, D.J., 2002. Mechanisms of arsenic uptake from aqueous solution by interaction with goethite, lepidocrocite, mackinawite, and pyrite: an X-ray absorption spectroscopy study. Environmental Science & Technology 36 (8), 1757e1762. Goldberg, S., 2002. Competitive adsorption of arsenate and arsenite on oxides and clay minerals. Soil Science Society of America Journal 66 (2), 413e421. Huang, J.-H., Kretzschmar, R., 2010. Sequential extraction method for speciation of arsenate and arsenite in mineral soils. Analytical Chemistry 82 (13), 5534e5540. Islam, F.S., Pederick, R.L., Gault, A.G., Adams, L.K., Polya, D.A., Charnock, J.M., Lloyd, J.R., 2005. Interactions between the Fe(III)-reducing bacterium Geobacter Sulfurreducens and arsenate, and capture of the metalloid by biogenic Fe(II). Applied and Environmental Microbiology 71 (12), 8642e8648. Keon, N.E., Swartz, C.H., Brabander, D.J., Harvey, C., Hemond, H.F., 2001. Validation of an arsenic sequential extraction method for evaluating mobility in sediment. Environmental Science & Technology 35 (13), 2778e2784. Kirk, M.F., Roden, E.E., Crossey, L.J., Brealey, A.J., Spilde, M.N., 2010. Experimental analysis of arsenic precipitation during microbial sulfate and iron reduction in model aquifer
sediment reactors. Geochimica et Cosmochimica Acta 74 (9), 2538e2555. Kocar, B.D., Borch, T., Fendorf, S., 2010. Arsenic repartitioning during biogenic sulfidization and transformation of ferrihydrite. Geochimica et Cosmochimica Acta 74 (3), 980e994. Maity, S., Chakravarty, S., Thakur, P., Gupta, K.K., Bhattacharjee, S., Roy, B.C., 2004. Evaluation and standardisation of a simple HG-AAS method for rapid speciation of As(III) and As(V) in some contaminated groundwater samples of West Bengal, India. Chemosphere 54 (8), 1199e1206. Manning, B.A., Goldberg, S., 1997. Arsenic(III) and arsenic(V) adsorption on three California soils. Soil Science 162 (12), 886e895. Manning, B., Fendorf, S., Goldberg, S., 1998. Surface structures and stability of arsenic(III) on goethite: spectroscopic evidence for inner-sphere complexes. Environmental Science & Technology 32 (16), 2383e2388. Masuda, H., Yamatani, Y., Okai, M., 2005. Transformation of arsenic compounds in modern intertidal sediment of Iriomote Island, Japan. Journal of Geochemical Exploration 87 (2), 73e81. Moore, J.N., Flcklin, W.H., Johns, C., 1988. Partitioning of arsenic and metals in reducing sulfidic sediment. Environmental Science & Technology 22 (4), 432e437. Oremland, R.S., Stolz, J.F., 2003. The ecology of arsenic. Science 300 (5621), 939e944. Pedersen, H.D., Postma, D., Jakobsen, R., 2006. Release of arsenic associated with the reduction and transformation of iron oxides. Geochimica et Cosmochimica Acta 70 (16), 4116e4129. Polizzotto, M.L., Harvey, C.F., Li, G., Badruzzman, B., Ali, A., Newville, M., Sutton, S., Fendorf, S., 2006. Solid-phases and desorption processes of arsenic within Bangladesh sediment. Chemical Geology 228 (1e3), 97e111. Postma, D., Jessen, S., Hue, N.T.M., Duc, M.T., Koch, C.B., Viet, P.H., Nhan, P.Q., Larsen, F., 2010. Mobilization of arsenic and iron from Red River floodplain sediment, Vietnam. Geochimica et Cosmochimica Acta 74 (12), 3367e3381. Saalfield, S.L., Bostick, B.C., 2009. Changes in iron, sulfur, and arsenic speciation associated with bacterial sulfate reduction in ferrihydrite-rich systems. Environmental Science & Technology 43 (23), 8787e8793. Smedley, P.L., Kinniburgh, D.G., 2002. A review of the source, behaviour and distribution of arsenic in natural waters. Applied Geochemistry 17 (5), 517e568. Sullivan, K.A., Aller, R.C., 1996. Diagenetic cycling of arsenic in Amazon shelf sediment. Geochimica et Cosmochimica Acta 60 (9), 1465e1477. Wilkin, R.T., Ford, R.G., 2002. Use of hydrochloric acid for determining solid-phase arsenic partitioning in sulfidic sediments. Environmental Science & Technology 36 (22), 4921e4927. Wolthers, M., Charlet, L., Van Der Weijden, C.H., Van Der Linde, P. R., Rickard, D., 2005. Arsenic mobility in the ambient sulfidic environment: sorption of arsenic(V) and arsenic(III) onto disordered mackinawite. Geochimica et Cosmochimica Acta 69 (15), 3483e3492. Wolthers, M., Butler, I.B., Rickard, D., 2007. Influence of arsenic on iron sulfide transformations. Chemical Geology 236 (3, 4), 217e227. Zhang, X., Jia, Y., Wang, X., Xu, L., 2008. Phylogenetic analysis and arsenate reduction effect of the arsenic-reducing bacteria enriched from contaminated soils at an abandoned smelter site. Journal of Environmental Sciences 20 (12), 1501e1507. Zhao, Z., Jia, Y., Xu, L., Zhao, S., 2011. Adsorption and heterogeneous oxidation of As(III) on ferrihydrite. Water Research. doi:10.1016/j.watres.2011.09.051.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 8 9 e6 7 9 7
Available online at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Monte Carlo simulation of pore blocking phenomena in cross-flow microfiltration Yingbo Chen a,*, Xiaoyu Hu b, Hern Kim c a
School of Materials Science and Engineering, Tianjin Polytechnic University, Tianjin 300387, China State Key Laboratory of Hollow Fiber Membrane Materials and Process, Tianjin Polytechnic University, Tianjin 300160, China c Department of Environmental Engineering and Energy, Myongji University, Yongin, Kyonggi-do 449-728, Republic of Korea b
article info
abstract
Article history:
An off-lattice Monte Carlo method was developed to examine the pore blocking
Received 18 April 2011
phenomena in cross-flow microfiltration. Membranes were generated by randomly
Received in revised form
distributed pores with a given pore size distribution on a two-dimensional plane. The
21 September 2011
permeability of particles through the membrane pores was calculated, and the effects of
Accepted 17 October 2011
crucial factors on the reduction of permeability were discussed. Reasonable fouling rules
Available online 24 October 2011
for simulations were examined and selected. At the initial stage the flux decreases sharply as the filtration time increases and then a steady state is reached. The results fit the pore
Keywords:
blocking mechanism well. The simulation model developed in this study can be effectively
Monte Carlo method
used for analyzing a number of factors influencing physical fouling of membranes. ª 2011 Elsevier Ltd. All rights reserved.
Off-lattice Permeability Pore blocking Fouling
1.
Introduction
Membrane filtration, such as microfiltration and ultrafiltration, is widely used in water purification (Meier-Haack and Booker, 2003), wastewater treatment (Chang et al., 2002), food processing (Babu et al., 2006) and medicine separation lu et al., 2005). The permeate flux in membrane (Bayramog filtration decreases with the processing time. This phenomenon is directly related to membrane fouling, which is a main limitation of membrane filtration. Membrane fouling can be categorized into two types: internal fouling which is caused by pore wall adsorption and blocking, and external fouling which is the result of formation of cake layer on the membrane surface. During membrane filtration, the fate of a particle deposited on the membrane depends on the membrane property and morphology, particle size and position, and
hydraulic conditions. A particle may pass through the membrane, be adsorbed on the pore wall, stick to the membrane surface, or be taken away by hydraulic force. Fouling mechanisms of membrane has been studied by many researchers. Many models have been developed to simulate the fouling problem during membrane filtration. The recent progress in understanding particle fouling of filtration membrane was reviewed (Aimar, 2003), where some of simulation works on membrane fouling were listed. The blocking mechanisms was thoroughly analyzed (Marchese et al., 2003) in relation to the performance of the filter media used, with particular reference to the ‘standard blocking’ process. As the results, all the blocking mechanisms were revised and reformulated in a common frame of powerlaw non-Newtonian fluids. A 2-D Voronoi tessellation model
* Corresponding author. Tel.: þ86 22 8395 5078; fax: þ86 22 8395 5055. E-mail addresses:
[email protected],
[email protected] (Y. Chen). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.10.023
6790
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 8 9 e6 7 9 7
(Baralla et al., 2001) was developed to simulate the porous membrane structure. Size blocking rules were used to observe the fouling of the pore blocking. A two-dimensional stochastic model (Wessling, 2001) was used to simulate small particles sticking to the membrane pore region and membrane surface, where decrease in flux and increase in membrane resistance were observed. Some three-dimensional simulations of particle deposition were used to illustrate the pore blocking phenomena by either equal-size particle or equal-size pore models (Kawakatsu et al., 1995; Yoon et al., 1999). Also, a three-dimensional model (Frey and Schmitz, 2000; Tung and Chuang, 2002) was used to find how the first layer of particle deposit was built at the membrane surface by determining the capture probability density at the membrane surface and especially close to a pore of the porous surface. A fixed matching size criterion (Seminario et al., 2002) was used to define the pore blocking in cross-flow microfiltration. Recently, a combined model including pore blockage model, pore constriction model, and cake filtration model was used to describe flux decline behavior and assess the importance of each fouling mechanism (Duclos-Oresello et al., 2006). The mechanisms of pore blocking and cake formation were probed by considering particle and membrane pore size distribution as well as the adhesion between particle and the membrane surface (Broeckmam et al., 2006). However, a clear understanding of the pore blocking is not available. Especially why and how the particles stick and block membranes are not well explained from the previous simulation models. In the present work, a Monte Carlo method was used to simulate particle capture and pore blocking of membrane pores in cross-flow microfiltration to better understand the mechanism of pore blockage and permeability reduction. A reasonable rule was established in terms of particle size, membrane pore size and filtration conditions to define pore blocking. A number of influencing factors from particles, membrane pores, and filtration conditions were examined on the normalized permeability reduction. The simulation results were explained and compared to the existing filtration theories and models. Compared with the above mentioned models which usually use fixed pore or particle sizes, the model in this paper, we use varied pore size and particle size. The size range and distribution can be set according to the reality. (e.g., in this paper, we select the particle size with range of 0.1e10 mm in lognormal size distribution, where contains almost all particle sizes properly existing in aqueous solution with suitable distribution.). Moreover, this model is especially suitable for the simulation the pore blocking during the early stage of cross-flow microfiltration.
Fig. 1 e Solute particle flow through the membrane system.
taken out from the system directly by the hydrodynamic force. However, the rests adhere to the surface and block the pore entrance. In this model, the interaction between the particles and membrane surface is assumed to be negligible in the condition of low concentration and adequate flowing rate (that is to say we only consider the interaction between the particles and the membrane pores). This assumption allows us to investigate the pore blocking phenomenon (particles sticking to the pore entrance and decreasing the effective pore size) without the influence triggered by cake layer formation. While the particle encounters the surface pore, it is retained on the pore and thus leads to the decrease of the pore size if the particle size is comparable to the pore size.
2.2.
Rules of particles trajectory
The details of Monte Carlo simulation of the fouling phenomena in cross-flow microfiltration are as follows: A particle is randomly chosen with a certain size and deposited to the membrane surface. The particle deposited randomly on the membrane surface touches either the membrane surface or the open cross-section of a pore. The rules for the particle movement are describes as follows: (i) If a particle hits the membrane surface, it will not affect the pores and their permeability. It is taken away by the cross-flow; (ii) If a particle hits the cross section of a pore, its fate depends on the following matching size rule. The ratio of the particle size to pore size is defined as b¼
Rpart Rpore
(1)
where Rpart is the radius of particle and Rpore is the radius of pore.
2.
Simulation methods
Case I: particle size < pore size; i. e., b < 1.
2.1.
Flow mode of the simulation
The particle can pass through the pore if there is no interaction between the particle and the pore. However, in real processing, some particles could be retained because of the energy barrier that exists between the particle and the pore. Such an energy barrier can be considered as the transition probability (Duclos-Orsello et al., 2004)
The membrane was modeled as a two-dimensional microstructure with holes, which represent the surface pores randomly distributed with a certain size distribution. As shown in Fig. 1, the feed solution flows cross the membrane surface. While some small suspended particles pass through the membrane pores unhindered, some large particles are
g1 ¼ a1 bexpðF=kTÞ
(2)
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 8 9 e6 7 9 7
6791
where a1 is a fitted parameter, F energy barrier, k boltzmann’s constant, and T absolute temperature. So if g1 < 1 the particle can pass through the pore. So g1 ¼ a1 bexpðF=kTÞh1
(3)
Rearranging Eq. (3) b<
1 a1 expðF=kTÞ
(4)
When Eq. (4) is satisfied, the particle can pass through the membrane pore, and the pore size and the permeability are not changed. On the contrary, if g1 > 1; b>
1 a1 expðF=kTÞ
(5)
is satisfied, the particle adheres to the pore surface and the pore size decreased due to blocking. Then it is assumed that the pore shape keeps circular and pore size reduces in proportional to square root of difference of the areas of the pore and of the particle. So R¼
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi R2pore R2part
(6)
Case II: particle size > pore size; i. e., b > 1 The particle covers the pore segment completely if it adheres to the pore entrance. But, as we know, if the particle size is much larger than 1 mm, the role of gravity can not be ignored and hydrodynamic effects are expected to dominate over diffusive effects. So the larger particles can be taken away by the hydrodynamic force. We define g2 ¼ a2 bexpðF=kTÞ
(7)
where a2 is a fitted parameter. So g2 > 1, that is g2 ¼ a2 bexpðF=kTÞi1
(8)
Rearranging Eq. (8) we got: b>
1 a2 expðF=kTÞ
(9)
Thus when Eq. (9) is satisfied, the particle is taken away by the hydrodynamic force, and while Eq. (10) b<
1 a2 expðF=kTÞ
(10)
is satisfied, the particle covers the pore completely and the permeability of that pore becomes zero. The fates of the particles are schemed in Fig. 2. In this model, the fitted parameters a1 and a2 are chosen to satisfy the Eqs. (3)e(5), and Eqs. (8)e(10). For b < 1/[a1exp(F/ kT)], the right hand side term depends on the energy barrier, temperature, and fitted parameter a1. According to DerjaguinLandau-Verwey-Overbeek (DLVO) theory (Oliveira, 1997), the energy of the particulate materials is mainly determined by the adhesion including van der Waals attraction and electrostatic double-layer repulsion (calculation of F and the discussion of effects of F on membrane fouling were published elsewhere (Chen and Kim, 2008)). a1 is a parameter which is influenced by the processing conditions (mainly
Fig. 2 e Four fates of the particles.
depends on the pressure difference across the membrane or perpendicular flow rate). A particle is much easier to adhere to the pore surface when the energy barrier is low. So the adhesion is detrimental to membrane fouling. The particle with high attraction to the pore surface fouls the membrane and blocks the pore easily. On the contrary, with increasing the transmembrane pressure, the hydrodynamic force perpendicular to the membrane surface increases, thus the probability of capturing a particle on the pore decreases (the perpendicular hydrodynamic force tends to take away the particles). So the parameter a1 and energy barrier F competitively affect the adsorption of a particle. They determine b value which decides whether a particle adsorbs on the pore or passes through the pore. For b > 1/[a2exp(F/kT)], the right hand side term mainly depends on the energy barrier, temperature, and fitted parameter a2. The effect of energy barrier was described above. a2 is a fitted parameter which is influenced by the processing conditions (mainly depends on the cross-flow rate). With increasing the cross-flow rate, the probability of capturing a particle on the pore decreases. So the parameter a2 and energy barrier F give a pair of competitive effects on the sticking of a particle. They determine b value which decides whether a particle sticks on the pore surface or is taken away by the hydraulic force from the pore.
2.3.
Analysis of the simulation
If all the particles of a certain concentration (i.e. a fix number of particles) are all deposited, one Monte Carlo step (MCS) is executed. The permeability of the membrane is calculated after each MCS and a new iteration begins until the time is reached at a maximum set. The normalized permeability is calculated as the area of the pore cross-section which allows the particles permeate through at certain time divided by the initial area of the pore cross section. The permeate flux is calculated as the number of particles passing trough the membrane divided by the total number of distributed particles. For each simulation and all the results are averaged for five configurations of the membranes.
6792
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 8 9 e6 7 9 7
The hydraulic resistance is described by Poiseuille flow as (Mortensen et al., 2005): Rhyd h
hL S2
(11)
where h is the dynamic viscosity of the liquid, L the channel length, and S is the cross-sectional area. Integrating all the pores on the membrane at time t the total membrane hydraulic resistance is obtained as hL Rhyd ðtÞ ¼ PNðtÞ 2 i¼1 SðtÞi
size ranges from mg =d3g to mg d3g , where 99% of the theoretical pore size population is included (Lu and Tsai, 2002). Typical pore size distributions are shown in Fig. 3(a) and (b). N pores were generated according to the given distribution, where N is such a value that controls the porosity of the membrane surface which is defined as the sum of the pore cross-section area divided by the total surface area of the membrane. The area of 200 200 is used as the membrane surface. The pores within the membrane were generated without overlapping
(12)
where S(t)i is the cross-sectional area of the ith pore at time t and N is the total number of pores on the membrane at time t. For the microfiltration of constant transmembrane, the permeate flux can be described by Darcy’s law as: J0 ¼
DP mRm
(13)
where J0 is the permeation, DP the transmembrane pressure, m the absolute viscosity of the liquid, and Rm is the resistance of the clean membrane. The initial permeate J0 mainly depends on Rm since there is no pore blocking initially. During the microfiltration, pore blocking causes the increasing of Rp and the permeate can be given, by modifying Darcy’s law, as: DP J¼ m Rm þ Rp
(14)
According to the physical blocking mechanisms of the pore, the pore fouling can be subdivided as standard blocking (particle deposition on the pore) and complete blocking (pore clogging). The complete pore blocking model can be described by (Lim and Bai, 2003): J ¼ J0 exp Kp t
(15)
Or lnJ ¼ Kp t þ lnJ0
(16)
where Kp is the system constant related to pore blocking resistance.
2.4.
Pore and particle size distribution
In this study two different distributions for surface pore generation were used: normal and lognormal. For the normal distribution, the density function of distribution is given as # " 1 ðr mÞ2 f ðrÞ ¼ pffiffiffiffiffiffiexp 2d2 d 2p
(17)
where m is the mean radius of the pores and d is the standard deviation. The pore size ranges from 0.1 to m þ 3d. For lognormal distribution, the density function of distribution is given as 2 # " lnr lnmg 1 pffiffiffiffiffiffiexp f ðrÞ ¼ 2 rlndg 2p 2 lndg
(18)
where mg is the geometric mean radius of the pores and dg is the geometric standard deviation of the distribution. The pore
Fig. 3 e Membrane pore size distributions: (a) normal distribution with mean radius 1.5 and standard deviation 1.72, (b) lognormal distribution with mean radius 1.53 and standard deviation 0.89, and (c) Particle size distribution with mean radius 0.72 and standard deviation 0.51.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 8 9 e6 7 9 7
and periodic boundary condition was used to extend the simulation domains. The particles are also generated with lognormal size distribution ranging from 0.1 to 10 mm (See Fig. 3(c)). The number of particles is relative to the feed concentration. When all the particles are deposited, one Monte Carlo Step is finished and a new iteration begins. The Monte Carlo steps are relative to the filtration time. The overall flowchart of Monte Carlo simulation procedures is shown in Fig. 4.
3.
Results and discussion
3.1.
Influence of particle concentration
The number of particles released in per MCS is strongly related to the concentration of the particles in the feed. Different concentrations of 2500, 5000 and 10,000 were used to observe the influence on the normalized permeability of the processing. As shown in Fig. 5 all curves of the permeability as a function of the filtration time can be separated into two parts. Within 100 MCS, the permeability decreases sharply. After that the filtration process reaches a pseudo steady-state stage with a long term decreasing of the permeability. The permeability tends to a stable value although it has a lightly
6793
decrease. Within the 100 MCS showing pore blocking of the membrane surface, the higher the concentration the faster the pore was clogged. It can be seen from the figure that the slopes of the curves increased with increasing the particle concentration. It was also shown that the permeability decreases with increasing the particle concentration. As the particle concentration increases, the rate of convective particle transport toward the membrane surface increases and hence, the overall rate of colloid deposition onto the membrane increases. Consequently, the amount of deposited colloid increases, resulting in higher resistance and lower permeability (Zhu and Elimelech, 1997).
3.2.
Influence of pore size distribution
In order to examine the effect of pore size distribution on the normalized permeability of the membrane filtration, two membrane surface pore models with different pore size distributions were constructed with the same mean pore radius. The lognormal and normal pore size distributions are described as f(r, m ¼ 1.53, d ¼ 0.89) and f(r, m ¼ 1.5, d ¼ 0.4667) and the size in the range of (mg =d3g , mg d3g ) and (0.1, m þ 3d), respectively. Under this setting 99% of the particle population is included (both of the integrals are more than 99.7%). Fig. 6 illustrates the influence of pore size distributions on the
Fig. 4 e Flow chart of the simulation program.
6794
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 8 9 e6 7 9 7
Fig. 5 e Influence of particle concentration on normalized permeability as a function of time.
permeability. The membrane with a lognormal size distribution holds a higher normalized permeability than the membrane with a normal distribution during all filtration processing period. The normal distribution with wider deviation from the average size induces faster and more drastic blocking of the membrane than lognormal one. From the view point of membrane preparation, it is important to regulate the morphology and pore size distribution of the membrane thus to control the membrane fouling.
3.3.
Influence of filtration conditions
The capture probability of a particle is influenced by filtration conditions such as membraneeparticle interaction, transmembrane pressure, feeding rate, concentration and viscosity of the solution, temperature and so forth. While b < 1, the size of particle is less than that of the membrane pore it approaching to. Whether the particle passes through the pore or is adsorbed on the pore mainly depends on the interaction energy barrier between the particles and the pore, the transmembrane pressure, and the absolute temperature. The g1 value with the ranges of 0.5e1, 0.7e1, and 0.9e1 (for g1 < 1, the values set as 0.5, 0.7 and 0.9) were chosen for the particle capture at the pore, and the permeability reduction was examined and shown in Fig. 7(a). The narrower the range of g1
Fig. 6 e Influence of membrane pore size distribution on normalized permeability as a function of time.
value the higher the permeability. It can be explained that, for the narrower range of g1 value which corresponds to a lower membraneeparticle interaction and a higher transmembrane pressure, the particle holds a lower probability to adsorb on the pore. When b > 1, the size of particle is bigger than that of the corresponding membrane pore. The particle may stick to the pore or flow away by the bulk solution, which mainly depends on the interaction energy barrier between the particle and the pore, the filtration rate, and the absolute temperature. The g2 value with the ranges of 1e1.2 and 1e2 (for g2 > 1, the values set as 1.2 and 2.0) were used to study the sticking of particles and the results are presented in Fig. 9. As shown in Fig. 7(b), for the g2 value of 1e2.0, the number of particles sticking to the pore surface is much more than that for the g2 value of 1e1.2. The increasing of particles sticking onto the pore can be explained that for higher g2 value, the membraneeparticle interaction is high while the cross-flow velocity is low, which increases the concentration polarization and thus more particles stick to the membrane pores. The results show good agreement with those reported by Niina Laitinen et al. (Laitinen et al., 2001) who also observed that the flux increases with enhancing the filtration pressure and the fouling diminished with increasing the cross-flow velocity.
3.4.
Influence of surface porosity of membrane
Surface porosity of membranes is an important factor influencing the performance of the filtration. Increasing the surface porosity can advance the flux of the membrane. However, the probability of particles adsorbed and sticking to the pore increases as increasing the surface porosity of the
Fig. 7 e Influence of (a) g1 and (b) g2 on normalized permeability as a function of time.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 8 9 e6 7 9 7
6795
membrane. The results were shown from Fig. 8. In Fig. 8(a) the numbers of particles passing through the membrane both increase for membrane surface porosity of 0.1 and 0.2 as time goes by. But the membrane with the porosity of 0.2 has more particles transported than the membrane with the porosity of 0.1, which illustrates that the flux strongly depends on the surface porosity of the membrane. The increasing rate (from the slopes as shown in Fig. 8(a)) for both membranes decreases as time goes by. It is explained that, as the time goes by, the particles block the membrane and thus induce the reduction of the particles passing through the membrane. Fig. 8(b) and (c) show the number of particles adsorbed on the pore and sticking to the pore, respectively. Both of the numbers of adsorbed and sticking particles increase sharply within 100 MCS for both membranes with a porosity of 0.1 and 0.2. And after that a slightly increasing or stable value reached for the sever fouling of the membranes.
3.5.
Analysis of the simulation
As the processing time increasing, the membrane hydraulic resistance increases due to the formation of pore blocking. As time goes by, some pores are clogged completely or blocked partially inducing the decreasing of the total pore crosssectional area and increasing of the hydraulic resistance. As shown in Fig. 9, before processing, the membrane resistance is 1 1012 L/(m.s). At the beginning of the filtration, the resistance increases sharply to 1 1017 L/(m.s), and then slowly increases to 1 1018 L/(m.s). After 800 MCS, the resistance has some jumps. It can be considered from two aspects. First, after
Fig. 8 e Influence of membrane surface porosity on cumulative number of particles (a) passing through the membrane, (b) adsorbed in the pore and (c) sticking on the pore as a function of time.
Fig. 9 e Hydraulic resistance (log Rhyd) as a function of time.
Fig. 10 e ln J as a function of time for membrane with (a) different b value and (b).
6796
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 8 9 e6 7 9 7
long time filtration, almost all of the pores are clogged or blocked, and the resistance becomes sensitive to the change of cross-sectional area. Secondly, in this model the cake formation was not considered. It is reasonable at the initial stage but not after long time filtration. The relationship between ln J and time obtained from the results of the simulation is a linear as shown in Fig. 10, which well satisfy Eq. (16). It is noted that the results fit the complete pore blocking model well in the initial stage. The values of Kp (slope of the curves in Fig. 10(a)) increased with increasing b value, which demonstrates that, with increasing b value, the complete blocking is easier to occur. It agrees with rule of the model and the results. Changing the initial membrane porosity has little obvious influence on the Kp value (see the slopes of the curves in Fig. 10(b)). It is also demonstrated that the flux changes but not the fouling mechanism with varied membrane porosity.
4.
Conclusions
Pore blocking is the dominant fouling mechanism in the early stage of microfiltration. The decline of permeability caused by pore blocking was simulated in the present model using a reasonable size matching criterion. Membrane with a high porosity and narrow distribution (lognormal distribution) is more effective to decrease the pore blocking and thus increase the permeability. Particle concentration and size distribution influence on the number of particles passing through, adsorbed, and sticking on the membrane pore can be predicted using the model. b value which is related to a number of parameters, including a1, a2, 4, and T, is successfully used to predict the degree of pore blocking at the initial stage. With increasing b value, the mechanism of pore blocking tends to complete pore blocking and the pore blocking mechanism does not change with varying the porosity of membranes. Very good agreements were found between the model and existing theories.
Acknowledgment This research was supported by National Natural Science Foundation of China (50903063) and Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Education, Science and Technology (2010-0010353).
references
Aimar, P., 2003. Recent Progress in Understanding Particle Fouling of Filtration Membranes. Proc. IMSTEC‘03 conference, 10e14 November. University of New South Wales, Sydney, Australia, ISBN 0 7334 2089 3. Babu, B.R., Rastogi, N.K., Raghavarao, K.S.M.S., 2006. Mass transfer in osmotic membrane distillation of phycocyanin colorant and sweet-lime juice. J. Membr. Sci. 272, 58e69. Baralla, G., Mattea, M., Gekas, V., 2001. A computer-aided model to simulate membrane fouling processes. Sep. Purif. Technol. 22-23, 489e498. lu, G., Yalc¸ın, E., Arıca, M.Y., 2005. Characterization of Bayramog polyethylenimine grafted and Cibacron Blue F3GA
immobilized poly(hydroxyethylmethacrylate-coglycydylmethacrylate) membranes and application to bilirubin removal from human serum. Colloid Surf. A 264, 195e202. Broeckmam, A., Busch, J., Wintgens, T., Marquardt, W., 2006. Modeling of pore blocking and cake layer formation in membrane filtration for wastewater treatment. J. Membr. Sci. 189, 97e109. Chang, I.-S., Clech, P.L., Jefferson, B., Judd, S., 2002. Membrane fouling in membrane bioreactors for wastewater treatment. J. Environ. Eng. 128, 1018e1029. Chen, Y., Kim, H., 2008. Monte Carlo Simulation of pore blocking and cake formation by interfacial interactions during membrane filtration. Desalination 233, 258e266. Duclos-Orsello, C., Kelly, W.P., Grant, D.C., Zahka, J., Thom, V., 2004. Neutral adsorptive capture of particles by membranes: network modeling near the membrane isoelectric point. J. Membr. Sci. 237, 167e180. Duclos-Oresello, C., Li, W., Ho, C.C., 2006. A three mechanism model to describe fouling of microfiltration membranes. J. Membr. Sci. 280, 856e866. Frey, J.M., Schmitz, P., 2000. Particle transport and capture at the membrane surface in cross-flow microfiltration. Chem. Eng. Sci. 55, 4053e4065. Kawakatsu, T., Nakajima, M., Nakao, S.-I., 1995. Threedimensional simulation of random packing and pore blocking phenomena during microfiltration. Desalination 101, 203e209. Laitinen, N., Michaud, D., Piquet, C., 2001. Effect of filtration conditions and backflushing on ceramic membrane ultrafiltration of board industry wastewater. Sep. Purif. Technol. 24, 319e328. Lim, A.L., Bai, R., 2003. Membrane fouling and cleaning in microfiltration of activated sludge wastewater. J. Membr. Sci. 216, 279e290. Lu, S.-Y., Tsai, C.-M., 2002. Membrane microstructure resulting from deposition of polydisperse particles. J. Membr. Sci. 177, 55e71. Marchese, J., Ponce, M., Ochoa, N.A., Pra´danos, P., Palacio, L., Herna´ndez, A., 2003. Fouling behaviour of polyethersulfone UF membranes made with different PVP. J. Membr. Sci. 211, 1e11. Meier-Haack, J., Booker, N.A., Carroll, T., 2003. A permeabilitycontrolled microfiltration membrane for reduced fouling in drinking water treatment. Water Res. 37, 585e588. Mortensen, N.A., Okkels, F., Bruus, H., 2005. Reexamination of Hagen-Poiseuille flow: shape dependence of the hydraulic resistance in microchannels. Phys. Rev. E 71 (057301), 1e4. Oliveira, R., 1997. Understanding adhesion: a means for preventing fouling. Exper. Therm. Fluid Sci. 14, 316e322. Seminario, L., Rozas, R., Bo´rquez, R., Toledo, P.G., 2002. Pore blocking and permeability reduction in cross-flow microfiltration. J. Membr. Sci. 209, 121e142. Tung, K.-L., Chuang, C.-J., 2002. Effect of pore morphology on fluid flow and particle deposition on a track-etched polycarbonate membrane. Desalination 146, 129e134. Wessling, M., 2001. Two-dimensional stochastic modeling of membrane fouling. Sep. Purif. Technol. 24, 375e387. Yoon, S.-H., Lee, C.-H., Kim, K.-J., 1999. Three-dimensional simulation of the deposition of multi-dispersed charged particles and prediction of resulting flux during cross-flow microfiltration. J. Membr. Sci. 161, 7e20. Zhu, X., Elimelech, M., 1997. Colloidal fouling of reverse osmosis membranes: measurement and fouling mechanism. Environ. Sci. Technol. 31, 3654e3662.
Glossary J: [m3/m2s] permeability J0: [m3/m2s] initial permeability k: [kgm2s2K1] Boltzmann’s constant (1.381 1023) Kp: [s1] system constant related to pore blocking resistance L: [m] length of the pore channel
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 6 7 8 9 e6 7 9 7
MCS: [-] Monte Carlo Step N: [-] number of pore in the membrane Np: [-] number of particle R: [m] radius of pore after fouling Rhyd: [m1s1] hydraulic resistance Rm: [m1] membrane resistance Rp: [m1] blocking resistance Rpart: [m] radius of particle Rpore: [m] radius of pore S: [m2] cross-sectional area of pore T: [K] absolute temperature t: [s] filtration time
6797
a1, a2: [-] fitted parameter b: [m/m] ratio of particle size to pore size g1, g2: [-] revision of b d: [-] standard deviation of normal distribution dg: [-] geometric standard deviation of lognormal distribution DP: [kgm1s2] transmembrane pressure ε: [m2/m2] porosity of membrane h: [m2s1] dynamic viscosity m: [-] mean of normal distribution mg: [-] geometric mean of lognormal distribution ms: [gm1s1] viscosity of the solution F: [kgm2s2] energy barrier
water research 45 (2011) 6798
Available online at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Corrigendum
Corrigendum to “Fate of N-nitrosodimethylamine, trihalomethane and haloacetic acid precursors in tertiary treatment including biofiltration” [Water Research 45 (2011) 5695e5704] Maria Jose´ Farre´*, Julien Reungoat, Francois Xavier Argaud, Maxime Rattier, Ju¨rg Keller, Wolfgang Gernjak The University of Queensland, Advanced Water Management Centre (AWMC), Level 4 Gehrmann Bldg, Research Road, Brisbane Qld 4072, Australia
On page 5699 right column, second line, the last sentence of Section 3.1.1 should read ‘below’ instead of ‘beyond’. Therefore the sentence should read as follows: “NDMA and NMOR were detected below LOQ across the plant (i.e., 5 ng/L for NDMA and 10 ng/L for NMOR).”
DOI of original article: 10.1016/j.watres.2011.08.033. * Corresponding author. Tel.: þ61 7 33463233; fax: þ61 7 33654726. E-mail address:
[email protected] (M.J. Farre´). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.10.044
