WATER RESEARCH A Journal of the International Water Association
Editor-in-Chief Mark van Loosdrecht Delft University of Technology Department of Biochemical Engineering Julianalaan 67 2628 BC Delft The Netherlands Tel: +31 15 27 81618 E-mail:
[email protected]
Editors J. Block Université H. Poincaré, Nancy I France David Dixon University of Melbourne Australia Hiroaki Furumai The University of Tokyo Japan Xiaodi Hao Beijing University of Civil Engineering and Architecture China Gregory Korshin University of Washington USA Anna Ledin Formas Sweden Eberhard Morgenroth Swiss Federal Institute of Aquatic Science and Technology (EAWAG) Switzerland W. Rauch University Innsbruck Austria Maria Reis Universidade Nova de Lisboa/FCT Portugal Hang-Shik Shin Korea Advanced Institute of Science and Technology Korea Thomas Ternes Bundesanstalt für Gewässerkunde Germany Stefan Wuertz Univ. of California, Davis USA
Associate Editors Andrew Baker University of New South Wales Australia
Damien Batstone The University of Queensland Australia G-H. Chen The Hong Kong University of Science & Technology Hong Kong China Tom Curtis Univ. of Newcastle upon Tyne UK Ana Deletic Monash University USA Francis de los Reyes III North Carolina State University USA Rob Eldridge The University of Melbourne Australia Rosina Girones University of Barcelona Spain Stephen Gray Victoria University Australia Kate Grudpan Chiang Mai University Thailand E.E. Herricks University of Illinois - Urbana USA Peter Hillis United Utilities Plc UK H-Y. Hu Tsinghua University China P.M. Huck University of Waterloo Canada Bruce Jefferson Cranfield University UK Ulf Jeppsson Lund University Sweden Sergey Kalyuzhnyi Moscow State University Russian Federation Jaehong Kim Georgia Institute of Technology USA Jes La Cour Jansen Lund Institute of Technology Sweden G. Langergraber BOKU/Univ. of Natural Res. and Applied Life Scs. Austria S-L. Lo National Taiwan University Taiwan Dionisis Mantzavinos Technical University of Crete Greece
Y. Matsui Hokkaido University Japan A. Maul Université Paul Verlaine-Metz France Max Maurer EAWAG Switzerland How Yong Ng National University of Singapore Singapore Satoshi Okabe Hokkaido University Japan S.L. Ong National University of Singapore Singapore Jong M. Park Pohang University of Science & Technology Korea Susan Richardson U.S. Environmental Protection Agency USA Miguel Salgot University of Barcelona Spain David Sedlak University of California, Berkeley USA Jean-Philippe Steyer LBE-INRA France M. Takahashi Hokkaido University Japan Kai Udert EAWAG Switzerland V.P. Venugopalan BARC Facilities India E. von Sperling Federal University of Minas Gerais Brazil Hanqing Yu University of Science & Technology of China China J. Zilles University of Illinois Urbana USA A.I. Zouboulis Aristotle University Greece
Editorial Office Mailing address Water Research Radarweg 29 1043 NX Amsterdam The Netherlands 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 ) 3 2 7 1 e3 2 7 8
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Occurrence of Mycobacterium avium subsp. paratuberculosis in raw water and water treatment operations for the production of potable water G. Aboagye a, M.T. Rowe a,b,* a b
Food Microbiology, The Queen’s University of Belfast, Belfast, Northern Ireland, United Kingdom Food Microbiology Branch, Agri-Food and Biosciences Institute, Newforge Lane, Belfast BT9 5PX, Northern Ireland, United Kingdom
article info
abstract
Article history:
Mycobacterium avium subsp. paratuberculosis (Map) causes Johne’s disease of cattle and is
Received 23 December 2010
implicated as a cause of Crohn’s disease in humans. The organism is excreted in animal
Received in revised form
faeces and can contaminate water catchment areas. This coupled with Map’s survival in
14 March 2011
the environment means that water destined for domestic use may be a source of exposure.
Accepted 15 March 2011
This work was designed to determine the occurrence of Map in Lough Neagh (the largest
Available online 21 March 2011
freshwater lake in the British Isles), used as a reservoir, and in two water treatment works (WTW1 and WTW2) which abstract from the lough and which have slow sand filtration
Keywords:
(SSF) and dissolved air flotation respectively as their principal treatment regimes. The
Mycobacterium avium subsp.
organism was not detected in lough water samples by culture (n ¼ 70) but 29% (20/70) were
paratuberculosis
positive by PCR. In the raw water to WTW1 and WTW2 no culture positives were detected
Water treatment
but 54% (13/24) and 58% (14/24) respectively were PCR positive. In WTW1 there were no culture positives at the SSF or final water but 31% (8/26) and 45% (9/20) respectively were PCR positive. In WTW2 similar results were obtained with 26% (6/23) and 48% (11/23) in the floccules and final water respectively. At WTW2 however one culture positive was detected in the final water. This latter finding is of concern. The inability to reach definitive conclusions indicates the need for further research, particularly in the detection methods for viable Map. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
Mycobacterium avium subspecies paratuberculosis (Map) is the known cause of Johne’s disease of ruminants, particularly affecting dairy cattle (Clarke, 1997). Although the disease is contracted in the early years of life of the animal overt clinical signs, such as emaciation and loss of milk yield are usually only manifested in later years (Fecteau and Whitlock, 2010). The animal disease is responsible for significant economic
losses prompting control programmes in many countries which have varying degrees of success (Bakker, 2010; Kennedy and Citer, 2010; Whitlock, 2010). In addition to the animal welfare and attendant agrieconomic issues Map has been implicated as a causal factor in a number of human conditions such as diabetes type 1 (Paccagnini et al., 2009; Rani et al., 2010) and in particular Crohn’s disease (Behr, 2010). This latter condition is incurable, causing abdominal pain and constipation which may
* Corresponding author. Food Microbiology Branch, Agri-Food and Biosciences Institute, Newforge Lane, Belfast BT9 5PX, Northern Ireland, United Kingdom. Tel.: þ44 (0)2890 255291; fax: þ44 (0)2890 255009. E-mail address:
[email protected] (M.T. Rowe). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.03.029
3272
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 2 7 1 e3 2 7 8
require surgical intervention (Dasari et al., 2010). Although the relationship between Map and Crohn’s disease is only putative it was deemed sufficiently strong for the UK Government to adopt the precautionary principle and advocate strategies to minimise the public’s exposure to the organism (Rubery, 2002). It has been shown that Map is excreted directly in the milk of infected animals and in high numbers in bovine faeces (Grant et al., 2002) that can contaminate the environment (Hermon-Taylor, 2009) and contribute to agricultural runoff which may be situated in water catchment areas (Pierce, 2009). Map has been detected in raw water sources by both culture and PCR in previous studies in the UK (Pickup et al., 2005; Whan et al., 2005a). However, the efficacy of water treatment processes in removing or killing the organism has not been determined to the best of the authors’ knowledge. The object of the work reported here was firstly to attempt both molecular and culture methods to detect Map in Lough Neagh over a seasonal cycle. Secondly, to test the incoming raw and outgoing final treated water from two water treatment works (designated WTW1 and WTW2) which abstract from the lough (Fig. 1). In addition, in the case of WTW1, the schmutzdecke or ‘dirty layer’ which is the biologically active site in a slow sand filter (SSF) and, in the case of WTW2, the surface floccules which contain the organic material as a result of dissolved air flotation (DAF) were also sampled.
2.
Materials and methods
2.1.
Lough Neagh and environs
It should be recognised that domestic grazing animals have access to the shoreline of Lough Neagh at many locations, there are comparatively few surrounding hills thus exposing it to the prevailing wind (south west with mean wind speed of 5 m s1) and it is comparatively shallow (maximum depth 31 m, average depth 10 m). These factors combined mean that although the two WTWs are not geographically proximal to each other (Fig. 1) they could be considered as having the same source water. This therefore allows, to some extent, a comparison between SSF, primary treatment process applied in WTW1 and DAF, primary treatment process employed in WTW2, for removal of or lethality for Map. Information on the Map infection status of herds contiguous to Lough Neagh was not available to the authors.
2.2.
Water treatment procedures at WTW1
The WTW1 (Fig. 2) abstracts raw water from the lough at a distance of 600 m from the shore at depths of 1.5 and 3 m. The raw water is pretreated with ozone (2.5 mg l1) or chlorine (concentration varied to have a minimum chlorine residual of
Fig. 1 e Geographical location of Lough Neagh showing the two water treatment sites.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 2 7 1 e3 2 7 8
WTW1
WTW2
Raw untreated water
Raw untreated water
Initial flocculation Initial ozonation
Initial chlorination Initial chlorination
First stage filtration (RGF)
Final flocculation
Final ozonation
Dissolved-air flotation (DAF)
Second stage filtration (SSF)
Intermediate chlorination
pH bal.*/orthophosphorylation
pH bal.*/orthophosphorylation
Final chlorination
Final chlorination
Final treated water
Final treated water
* bal. = balance
Fig. 2 e Flow diagram of water treatment at water treatment works 1 and 2 (WTW1 and 2).
0.2 mg l1 above the sand) if the ozone is not in service. Preferential use is made of ozone because of the high levels of particulate matter in the raw water which may result in unacceptable levels of trihalomethane (formed by reaction between chlorine and organic matter) which poses a health risk. The rapid gravity filters (RGF) sieve the water through sand grains for the inter-ozonation (final ozonation) stage which seeks to remove algae and total organic carbon from the pretreated water (Fig. 2). The next stage of treatment is water clarification in the slow sand filtration (SSF) system where particulate matter in the water is removed, including microorganisms, by the top dirty layer (schmutzdecke) which is mainly a biological process resulting in metabolism of organic compounds and sieving of particulate matter by accumulated microorganisms that form a meshwork. This is supported by sand grains underlain with large gravels that serve as an underdrain, conveying the clarified water to the final treatment stage where free chlorine at a concentration of 1.2e1.4 mg l1 is added. The pH, before the final chlorination stage, is adjusted to 7.0 by adding soda ash and also orthophosphate (1.0e1.5 mg l1) to prevent lead leaching into the final treated water before the water is led to holding reservoirs and finally distributed to the general public.
2.3.
Water treatment procedures at WTW2
The WTW2 (Fig. 2) is situated about 800 m from the lough and initiates water treatment by a flocculation procedure
3273
employing aluminium sulphate before dosing the pretreated water with chlorine at 2 mg l1 to inhibit algal growth in storage tanks. An initial flocculation stage, where gross particulate matter is removed from the water occurs prior to the main flocculation stage which occurs in the dissolved air flotation chamber where smaller particles including microorganisms are removed. Here, clarified water is saturated with pressurised air (generated by a pressurising pump at 172e483 kPa) at a maximum retention time of 3 min and, with a drop in pressure of the system, the air is released from the water and the air bubbles generated attach themselves to the particulate matter including colloidal particles and suspended solids which float to the surface. This flotation occurs as a result of the entrapped air within the particles becoming lower in specific gravity than that of water. The floccules generated are removed by a skimming procedure. A final dosing with chlorine is performed at a concentration between 1.5 and 2 mg l1 (Fig. 2) after dosing with soda ash (to adjust the pH to 7.0) and orthophosphate (1.0e1.5 mg l1) to prevent leaching of lead into the treated water before the final water is pumped to holding reservoirs for distribution to the general public.
2.4.
Collection of samples
Raw water samples were collected from the three sampling sites (Lough Neagh, WTW1 and WTW2) over a period of 12 months e 3-month intervals for Lough Neagh to coincide with the four seasons and once every month for WTW1 and WTW2. The sampling sites were; Lough Neagh and proximal to the two abstraction sites at a depth of between 1.5 and 12 m using a manual corer (Hth-Teknik, Lulea, Sweden) which produced cores of approximately 10 mm thickness of sediment and contiguous water column, incoming raw and final treated water from both WTWs (1 l each), schmutzdecke (20 g) down to a depth of approximately 1 cm from drained SSFs in the case of WTW1 and flotation floccules (20 g) in the case of WTW2 as shown in Table 1. Sampling the schmutzdecke at WTW1 was determined by which SSF had been drained at the time of sampling and was therefore outside the control of experimental staff. There were five occasions on which a SSF was sampled twice at different chronological times. At WTW2 there were two DAF tanks and choice at sampling time was dependent on which was operational at the time.
2.5.
Pre-treatment of samples prior to analysis
The water samples (Lough Neagh, raw and final treated waters from WTW1 and 2) were membrane filtered (0.2 mm pore size, 47 mm diameter) and the membranes transferred to 20 ml 0.75% w/v cetylpyridinium chloride (CPC) solution (SigmaeAldrich Company Ltd, Dorset, UK) and incubated at room temperature for 4 h. Subsequently the membranes were transferred to 10 ml phosphate buffered saline plus 0.05% Tween 20 (PBS-T20, pH 7.3) and the CPC filtrate re-filtered as before to capture any Map cells dislodged from the filters during the decontamination phase. This second membrane was transferred to the PBS-T20 solution containing the first membrane and both abraded with forceps before being shaken vigorously by hand for 2 min
3274
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 2 7 1 e3 2 7 8
Table 1 e IS900 and f57 PCR analyses of water and sediment samples obtained from Lough Neagh and two water treatment works. Sample location
Sample type
No. of samples taken
Number IS900 PCR positive/ total number of samples (%)
Number f57 Rt-PCR positive/ total number tested (%)
Lough Neagh
Water and sediment
70
20/70 (29%)
4/20 (20%)
WTW1
Raw water Schmutzdecke Final water
24 26 20 70
13/24 (54%) 8/26 (31%) 9/20 (45%) 30/70 (43%)
3/13 (23%) 3/8 (38%) 0/9 (0%) 6/30 (20%)
Raw water Floccules Final water
24 23 23 70
14/24 (58%) 6/23 (26%) 11/23 (48%) 31/70 (44%)
4/14 (29%) 3/6 (50%) 1/11 (9%) 8/31 (26%)
210
81/210 (39%)
18/81 (22%)
Total for WTW1 WTW2
Total for WTW2 Grand total
followed by vortex mixing for 2 min. The membranes were removed and the resulting suspension used for culture and PCR assay as described later in this paper. In respect of the samples from Lough Neagh these were treated as for the other water examples except that the filtering time was protracted, in some instances, because of the presence of excessive particulate matter. Twenty grams of the schmutzdecke and floccules were dispensed separately into 480 ml distilled water and stomached (Stomacher Model 400, Seward Ltd., Southdownview Way, Worthing, UK) at 260 rpm for 4 min and the resultant suspension subjected to double membrane filtration and CPC decontamination as described before.
2.6.
Culture protocols
One hundred microlitres of the PBS-T20 test cell suspensions were spread plated (British Standards Institution, 1984) onto Middlebrook 7H10 supplemented with VAN antibiotic cocktail (Sigma, Gillingham, Dorset, UK) and 2 mg l1 mycobactin J (Synbiotics Europe SAS, Lyon, France) and sealed with parafilm to avoid desiccation. One hundred microlitres of the suspension was also inoculated into triplicate tubes of M7H9 broth medium supplemented with PANTA antibiotic cocktail (Becton Dickinson and Company, Sparks, Maryland, USA) and mycobactin and triplicate Bactec 12B medium vials supplemented with PANTA and mycobactin. All samples were incubated at 37 C for at least 10 weeks and checked regularly for growth i.e. typical Map colonies on M7H10 plates, turbidity in M7H9 broths and positive growth index readings (30 units) for Bactec cultures. Middlebrook 7H10 was used in preference to Herrold’s Egg Yolk Medium, used previously for a similar study (Whan et al., 2005a), because it was found to give similar recovery rates, minimise the background microflora and allow better identification of Map colonies for further confirmation. In all cases if growth was detected an acid-fast stain and PCR assay were performed to confirm the presence of mycobacteria and Map respectively. In the case of the culture isolate obtained this was subjected to mycobactin dependency, f57 PCR assay and the IS900 PCR amplicon was sequenced and compared to Map using the NCBI BLAST GenBank database and also typed by mycobacterial interspersed repetitive unitvariable-number tandem-repeat (MIRU-VNTR). This method
is based on the polymorphism of tandemly repeated DNA sequences and has been used for genotyping several mycobacterial species (Supply et al., 2006). In this study, eight MIRU-VNTR loci were applied to differentiate Map strains (Thibault et al., 2007).
2.7.
Cell disruption for DNA release prior to PCR assay
To 1 ml of each test sample (PBS-T20 cell suspensions) in an Eppendorf tube, 10 ml of Map antibody beads (Matrix Pathatrix PM 50, Map test, Matrix Microscience, Newmarket, UK) were added and the solution agitated on a rotary mixer (LD-79, Labinco BV, The Netherlands) for 30 min at 10 m s1. The sample, after agitation, was placed in a magnetic particle concentrator (DYNAL MPC-S, Invitrogen Ltd, Paisley, UK) for 10 min and the resulting clear liquid aspirated and discarded. The pellet was re-suspended in 700 ml TEN lysis buffer (pH 8.0) containing 0.744 g l1 ethylenediaminetetraacetic acid (EDTA, sodium salt, Sigma), 23.376 g l1 sodium chloride (Fisher Scientific UK Ltd., Loughborough, UK), 1.576 g l1 TriseHCl (pH 8.0, Sigma) and 10 g l1 sodium dodecylsulphate (Amesham Biosciences, Uppsala, Sweden) supplemented with 25 mg ml1 proteinase K (Sigma) and filter sterilised (0.45 mm pore size). The suspension was incubated overnight at 37 C before being transferred to FastProtein Blue126 tubes (Qbiogene-ALEXIS Ltd, Carlsbad, California, USA) containing matrix B and subjected to ribolysation (Hybaid Ltd., Middlesex, UK) for 45 s at 6.5 m s1 before being placed on ice for 15 min to settle the foam generated. The DNA was extracted and purified as described below.
2.8.
Extraction and purification of DNA
Seven-hundred microlitres of phenol:chloroform:isoamylalcohol (25:24:1, pH 8.0, Sigma) was added to each FastProtein Blue tube and centrifuged at 7826 g for 10 min. The supernatant was aspirated into a micro-centrifuge tube containing 400 ml of isopropanol (99% v/v) and incubated at 20 C for 30 min to precipitate DNA. The centrifugation step was repeated followed by washing once with 500 ml of 70% v/v ethanol. Finally the ethanol was decanted off and the DNA pellet allowed to air dry for 30 min at room temperature before being re-suspended in 50 ml TriseEDTA (TE) buffer (pH 8.0) and stored at 20 C until required.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 2 7 1 e3 2 7 8
2.9.
IS900 amplification of extracted DNA
An Adiavet Paratb PCR kit (Adiagene, 38 Rue de Paris, 22000 Saint Brieuc, France) which targets the IS900 amplicon of Map was employed for this PCR assay. This kit has been optimised for the detection of Map from bovine faecal, milk and tissue matrices. In the work reported here the assay protocol was further optimised for the detection of Map DNA in PBS-T20 and sediment matrices (data not shown) and the sensitivity equates to 10 cells per ml due to the presence of between 10 and 14 copies of the IS900 sequence per cell. The sensitivity was calculated using a range of dilutions of Map ATCC 43015 in PBS-T20 and plating out as described previously in conjunction with PCR assay. Two microlitres of extracted DNA was subjected to PCR amplification in a 50 ml reaction according to the manufacturer’s instructions (Adiagene) using a thermal cycler (MBS Satellite 0.2G Thermocycler; Thermo Electron Corp., Milford, MA, USA). The reaction conditions were as follows; 1 cycle of 37 C for 30 s, 94 C for 5 min followed by 45 cycles of 94 C for 15 s, 62 C for 30 s, 72 C for 40 s followed by final extension of 1 cycle of 72 C for 10 min and held at 4 C before analysis.
2.10.
f57 real time PCR (Rt-PCR) of extracted DNA
Only if the sample tested positive for the IS900 PCR was it subjected to f57 PCR for confirmatory purposes. By inference therefore there can be no IS900 PCR negative and f57 positive samples. The method of Donaghy et al. (2010) was used with the final PCR assay mix being as follows; 25 ml Taqman universal master mix, 5 ml 10 exo ipc mix (Applied Biosystems, Warrington, UK), 1 ml 50 ipc DNA, 9.5 ml water, 1.5 ml forward f57 primer, 1.5 ml reverse f57 primer, 1.5 ml probe and 5 ml of template DNA. The cycling conditions for the f57 Rt-PCR were as follows; 1 cycle of 50 C for 2 min, 1 cycle of 95 C for 10 min, followed by 40 cycles of 95 C for 15 s and final extension of 1 cycle of 60 C for 1 min. The primers and probe were as follows: JM111F, Forward, 50 -CCG CGA TCC CAA AAG TTG-30 ; JM249R, Reverse, 50 -CTC GTA GCT GCC GAT TCA TG-30 ; JM165P, Probe, 50 -FAM-TCA CGG ACT AGA CCG GT-MGB-30 . The 50 end was labeled with 6-carboxyfluorescein (FAM; Applied Biosystems) and quenched with a minor groove bonder (MGB; Applied Biosystems). A Ct value 35 was considered a negative. Using this Ct value as a threshold and performing tandem Rt-PCR and plate counts the sensitivity was estimated to be 103 cfu ml1.
3.
Results and discussion
In the present study PCR positive results were found at all the sites sampled from Lough Neagh to final treated waters at the two WTWs (Table 1). The primary assay used was based on the IS900 insertion element, the specificity of which has been called into question (Kim et al., 2002; Taddei et al., 2008). However, the primer sequences used in the work reported here have been shown to give negative PCR assay results with Mycobacterium scrofulaceum (1 strain), Mycobacterium intracellulare (1 strain) and Mycobacterium SP2333 (Beatrice Blanchard, pers. comm.) which have been reported to give false
3275
positive responses with PCR assays based on the IS900 element. The IS900 element was chosen because it is present as multiple copies in the Map genome thereby making the assay sensitive enough to detect the low numbers of Map expected. The prepared DNA templates from samples which tested positive using the IS900 based primers were retested using primers directed at the f57 insertion element. This is reputed to be more specific (Mobius et al., 2008) but is present only as a single copy in the genome and therefore results in lower sensitivity (103 cells ml1) compared to IS900 (10 cells ml1). Bearing in mind that the f57 PCR assay was used as a confirmatory only on all IS900 positive PCR assays not all samples tested positive. Indeed only 18 out of 81 IS900 PCR positive samples (22%) tested positive using the f57 assay indicating that Map, if present, was likely to be only in low numbers or that the samples contained other mycobacteria, other than those mentioned previously, that possibly contained the IS900 insertion element. In Lough Neagh 20/70 (29%) of samples tested positive by PCR (Table 1) which is comparable to that observed by Pickup et al. (2005) who found an occurrence of 21/67 (31%) throughout lakes and water courses in the Lake District of England, UK that are used as a catchment area for the country. This number of IS900 PCR positives from Lough Neagh was lower than that from the raw water entering both WTWs (Table 1). This was unexpected since the Lough Neagh samples invariably contained sediment and Map is known to survive longer in sediment than in the water column (Whittington et al., 2005). There was no significant difference (P ¼ 1.0) in the raw untreated water entering the two WTWs i.e. 54 and 58% (Table 1) adding credence to the notion that the source water was essentially the same for both WTWs. Interestingly in a previous, albeit more limited study, of raw untreated waters entering WTWs from Lough Neagh Whan et al. (2005a) found 2/7 (28.8%) samples positive by IS900 PCR from a WTW distant from WTW1 and 2 and 0/26 (0%) from WTW2 itself. Although the IMSePCR methods used in the study of Whan et al. (2005a) and the work reported here were not directly comparable, the reported sensitivities were the same i.e. 10 Map cells ml1 (Whan et al., 2005b). To the authors’ knowledge there have been no significant changes in agricultural or other practices in the area surrounding WTW2 that would account for the difference in occurrence rates. In a study of the river Taff in Wales, UK Pickup et al. (2005) found that there was a significant association (1e5%) between rainfall and detection of Map. In the present study however no significant effect (binomial regression, P ¼ 0.544) of weather conditions (temperature, wind speed and rainfall) on occurrence of Map PCR positives was obtained from either the lough itself or the two WTWs. This disparity may be partly explained by the way the climate data was recorded in the two studies. In the case of Pickup et al. (2005) the mean rainfall values on days of PCR positive samples and on each of the preceding 7 days were compared with mean rainfall values on days of PCR negative samples. In the work reported here however the mean climatic parameters, compiled on a monthly basis, were used for comparison. These were obtained from a UK Met Office sited at Belfast International Airport situated approximately 4 and 22 km away from WTW1 and 2 respectively.
3276
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 2 7 1 e3 2 7 8
A positive result with PCR is indicative of presence but not necessarily viability of Map cells. The finding of Map in the schmutzdecke layer of the slow sand filter (Table 1) was not unexpected since the biological action of the schmutzdecke resides in the bacteriocidal action of environmental protozoa and there is evidence that Map can survive such ingestion (Hermon-Taylor and Bull, 2002; Whan et al., 2006). This also applies to ingestion by nematode larvae and other invertebrates (Fischer et al., 2001; Lloyd et al., 2001; Whittington et al., 2001) associated with the schmutzdecke layer. In addition mycobacteria, in general, can reside and survive in biofilms (Falkinham et al., 2001; Pickup et al., 2006; Rowe and Grant, 2006) and the schmutzdecke could be expected to provide a suitable stratum. The schmutzdecke from a number of drained SSFs were sampled during the course of the study, one per month, so the findings are not likely to be due an aberration with one individual SSF. In respect of WTW1 the comparable percentage of PCR positive samples of final water compared to the schmutzdecke may be due to lysis of intact cells and leaching of naked Map DNA into the final water. Certainly no culture positive samples of final water were detected in WTW1. However the sensitivity of the PCR method used is approximately 10 cfu ml1 (data not shown) which is likely to be greater than the culture method. Published information on culture methods, albeit using different matrices, is 102 cfu per 102 g of faeces (Jorgensen, 1982; Reddacliff et al., 2003) and 102e103 cfu ml1 for milk (Grant et al., 2003). It should be recognised however that decontamination was employed (0.75% CPC for 4 h) and no resuscitation was attempted. This would undoubtedly have reduced recovery because of the likely presence of injured or dormant cells. If however the results obtained reflect significant lysis of Map cells and subsequent leaching of naked DNA into the final water this would indicate an efficient treatment. In WTW2 the percentage of PCR positive floccule samples, as a result of the DAF process, were fewer than either the incoming raw water or final treated water (Table 1). It should be noted that the formation of floccules, in contrast to the schmutzdecke, is a purely physiochemical process with no biological bioremediation involved. It is known that the cell wall of Map is hydrophobic (Brennan and Nikaido, 1995). This, it would be expected, would induce an affinity between Map and the air bubbles and result in concentration of Map in the surface floccule layer but this was not detected in the system investigated. It is of concern that a viable Map culture was obtained in the final treated water from WTW2 even with the difficulties of obtaining viable isolates from environmental sources. The fact that the isolate was mycobactin dependent, positive with f57 PCR and the IS900 amplicon and when sequenced showed 100% homology with Map using the NCBI BLAST GenBank database provides substantive proof of its identity. The culture was subjected to MIRU-VNTR typing analysis as described by Thibault et al. (2007) and was assigned the INMV number of 2 indicating that it is one of the two predominant bovine strains in the UK as well as other European countries such as Netherlands, Finland and Spain (Stevenson et al., 2009). In summary Map was detected by PCR in Lough Neagh which is in agreement with previous work on the lough (Whan et al., 2005a). It was also found by PCR assay throughout both
WTWs, even in the final water which indicates the presence of Map but not necessarily that of intact viable cells. There was no clear difference between the corresponding results from both WTWs which, because of the topography of the lough, essentially abstracted the same source water. This means that if PCR positive tests are used as the index then the efficacy of both treatments was the same. What was most concerning was the culture positive sample in final treated water from WTW2 which shows that viable Map, albeit probably in very low numbers, is entering the water distribution system. It is worthy of note that Pickup et al. (2006) detected Map using PCR in 1 of 54 domestic water cisterns. The need to present the interpretation of the results obtained in this study in the form of a range of explanations rather than a definitive conclusion highlights the need for further research in this area and particularly improvements in the culture of Map from environmental sources.
4.
Conclusions
A limited survey (n ¼ 210) of Lough Neagh for M. avium subsp. paratuberculosis (Map) was performed along with two water treatment works (WTWs) that abstract from the lough using both PCR assay and culture. Map was detected by PCR in Lough Neagh and throughout the two WTWs, including the final treated water. One culture positive was confirmed and that was found in the final treated water. No difference was found in terms of Map removal between slow sand filtration (WTW1) and dissolved air floatation (WTW2). This work provides evidence that the public may be exposed to Map through water supplies.
Acknowledgements This work was supported by the following scholarships associated with Queen’s University of Belfast; Gibson, Harold Barbour and Mac Geough Bond. The authors would also wish to thank Northern Ireland Water for allowing access to their facilities, David Kilpatrick, Biometrics Branch, Agri-Food and Biosciences Institute (AFBI) for the statistical analyses and Dr. A. Gilmour, AFBI for helpful advice and guidance.
references
Bakker, D., 2010. Paratuberculosis control in Europe. In: Behr, M.A., Collins, D.M. (Eds.), Paratuberculosis. Organism, Control, Disease. CAB International, Oxfordshire, UK, pp. 306e318. Behr, M.A., 2010. The path to Crohn’s disease: is mucosal pathology a secondary event? Inflammatory Bowel Disease (5), 896e902. Brennan, P.J., Nikaido, H., 1995. The envelope of mycobacteria. Annual Review Biochemistry 64, 29e63.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 2 7 1 e3 2 7 8
British Standards Institution, 1984. Microbiological Examination for Dairy Purposes. Part 2: Methods of General Application for Enumeration of Microorganisms. Section 2.3: Enumeration of Microorganisms by Surface Plate Technique for Colony Count. BSI, London. Clarke, C.J., 1997. The pathology and pathogenesis of paratuberculosis in ruminants and other species. Journal of Comparative Pathology 116 (3), 217e261. Dasari, B.V., Maxwell, R., Gardiner, K.R., 2010. Assessment of complications following strictureplasty for small bowel Crohn’s disease. Irish Journal of Medical Science 179 (2), 201e205. Donaghy, J.A., Johnston, J., Rowe, M.T., 2010. Detection of Mycobacterium avium ssp. paratuberculosis in cheese, milk powder and milk using IS900 and f57-based qPCR assays. Journal of Applied Microbiology 110 (2), 479e489. Falkinham 3rd, J.O., Norton, C.D., LeChevallier, M.W., 2001. Factors influencing numbers of Mycobacterium avium, Mycobacterium intracellulare, and other Mycobacteria in drinking water distribution systems. Applied and Environmental Microbiology 67 (3), 1225e1231. Fecteau, M.-E., Whitlock, R.H., 2010. In: Behr, M.A., Collins, D.M. (Eds.), Paratuberculosis. CAB International, Wallingford, UK, pp. 144e156. Fischer, O., Matlova, L., Dvorska, P., Svastova, J., Bartl, I., Melicharek, I., et al., 2001. Diptera as vectors of mycobacterial infections in cattle and pigs. Medical and Veterinary Entomology 15 (2), 208e211. Grant, I.R., Ball, H.J., Rowe, M.T., 2002. Incidence of Mycobacterium paratuberculosis in bulk raw and commercially pasteurized cows’ milk from approved dairy processing establishments in the United Kingdom. Applied and Environmental Microbiology 68 (5), 2428e2435. Grant, I.R., Kirk, R.B., Hitchings, E., Rowe, M.T., 2003. Comparative evaluation of the MIGIT and BACTEC culture systems for the recovery of Mycobacterium avium subsp. paratuberculosis from milk. Journal of Applied Microbiology 196 (1), 196e201. Hermon-Taylor, J., 2009. Mycobacterium avium subspecies paratuberculosis, Crohn’s disease and the Doomsday scenario. Gut Pathogens 1 (1), 15. Hermon-Taylor, J., Bull, T., 2002. Crohn’s disease caused by Mycobacterium avium subspecies paratuberculosis: a public health tragedy whose resolution is long overdue. Journal of Medical Microbiology 51 (1), 3e6. Jorgensen, J.B., 1982. An improved medium for culture of Mycobacterium paratuberculosis from bovine faeces. Acta Veterinaria Scandinavica 23 (3), 325e335. Kennedy, D., Citer, L., 2010. Paratuberculosis control measures in Australia. In: Behr, M.A., Collins, D.M. (Eds.), Paratuberculosis. Organism, Disease, Control. CAB International, Oxfordshire, UK, pp. 330e343. Kim, S.G., Shin, S.J., Jacobson, R.H., Miller, L.J., Harpending, P.R., Stehman, S.M., et al., 2002. Development and application of quantitative polymerase chain reaction assay based on the ABI 7700 system (TaqMan) for detection and quantification of Mycobacterium avium subsp. paratuberculosis. Journal of Veterinary Diagnostic Investigation 14 (2), 126e131. Lloyd, J.B., Whittington, R.J., Fitzgibbon, C., Dobson, R., 2001. Presence of Mycobacterium avium subspecies paratuberculosis in suspensions of ovine trichostrongylid larvae produced in faecal cultures artificially contaminated with the bacterium. Veterinary Record 148 (9), 261e263. Mobius, P., Hotzel, H., Kohler, H., 2008. Comparison of 13 singleround and nested PCR assays targeting IS900, IS Mav2, f57 and locus 255 for detection of Mycobacterium avium subsp. paratuberculosis. Veterinary Microbiology 126 (4), 324e333. Paccagnini, D., Sieswerda, L., Rosu, V., Masala, S., Pacifico, A., Gazouli, M., et al., 2009. Linking chronic infection and autoimmune diseases: Mycobacterium avium subspecies
3277
paratuberculosis, SLC11A1 polymorphisms and type-1 diabetes mellitus. PLoS One 4 (9), e7109. Pickup, R.W., Rhodes, G., Arnott, S., Sidi-Boumedine, K., Bull, T.J., Weightman, A., et al., 2005. Mycobacterium avium subsp. paratuberculosis in the catchment area and water of the River Taff in South Wales, United Kingdom, and its potential relationship to clustering of Crohn’s disease cases in the city of Cardiff. Applied and Environmental Microbiology 71 (4), 2130e2139. Pickup, R.W., Rhodes, G., Bull, T.J., Arnott, S., Sidi-Boumedine, K., Hurley, M., Hermon-Taylor, J., 2006. Mycobacterium avium subsp. paratuberculosis in lake catchments, in river water abstracted for domestic use, and in effluent from domestic sewage treatment works: diverse opportunities for environmental cycling and human exposure. Applied and Environmental Microbiology 72 (6), 4067e4077. Pierce, E.S., 2009. Possible transmission of Mycobacterium avium subspecies paratuberculosis through potable water: lessons from an urban cluster of Crohn’s disease. Gut Pathogens 1 (1), 17. Rani, P.S., Sechi, L.A., Ahmed, N., 2010. Mycobacterium avium subsp. paratuberculosis as a trigger of type-1 diabetes: destination Sardinia, or beyond? Gut Pathogens 2 (1), 1. Reddacliff, L.A., Nicholls, P.J., Vadali, A., Whittington, R.J., 2003. Use of growth indices from radiometric culture for quantification of sheep strains of Mycobacterium avium subsp. paratuberculosis. Applied and Environmental Microbiology 69 (6), 3510e3516. Rowe, M.T., Grant, I.R., 2006. Mycobacterium avium ssp. paratuberculosis and its potential survival tactics. Letters in Applied Microbiology 42 (4), 305e311. Available from: Rubery, E., 2002. A Review of the Evidence for a Link Between Exposure to Mycobacterium paratuberculosis and Crohn’s Disease (CD) in Humans. A Report for the Food Standards Agency, January 2002, pp. 1e66 http://www.food.gov.uk/ multimedia/pdfs/mapcrohnreport.pdf (accessed 14.06.10). Stevenson, K., Alvarez, J., Bakker, D., Biet, F., de Juan, L., Denham, S., et al., 2009. Occurrence of Mycobacterium avium subspecies paratuberculosis across host species and European countries with evidence for transmission between wildlife and domestic ruminants. BMC Microbiology 9, 212. Supply, P., Allix, C., Lesjean, S., Cardoso-Oelemann, M., RuschGerdes, S., Willery, E., et al., 2006. Proposal for standardization of optimized mycobacterial interspersed repetitive unitvariable-number tandem repeat typing of Mycobacterium tuberculosis. Journal of Clinical Microbiology 44 (12), 4498e4510. Taddei, R., Barbieri, I., Pacciarini, M.L., Fallacara, F., Belletti, G.L., Arrigoni, N., 2008. Mycobacterium porcinum strains isolated from bovine bulk milk: implications for Mycobacterium avium subsp. paratuberculosis detection by PCR and culture. Veterinary Microbiology 130 (3e4), 338e347. Thibault, V.C., Grayon, M., Boschiroli, M.L., Hubbans, C., Overduin, P., Stevenson, K., Gutierrez, M.C., Supply, P., Biet, F., 2007. New variable-number tandem-repeat markers for typing Mycobacterium avium subsp. paratuberculosis and M. avium strains: comparison with IS900 and IS1245 restriction fragment length polymorphism typing. Journal of Clinical Microbiology 45 (8), 2404e2410. Whan, L., Ball, H.J., IGrant, I.R., Rowe, M.T., 2005a. Occurrence of Mycobacterium avium subsp. paratuberculosis in untreated water in Northern Ireland. Applied and Environmental Microbiology 71 (11), 7107e7112. Whan, L., Ball, H.J., Grant, I.R., Rowe, M.T., 2005b. Development of an IMSePCR assay for the detection of Mycobacterium avium ssp. paratuberculosis in water. Letters in Applied Microbiology 40 (4), 269e273. Whan, L., Grant, I.R., Rowe, M.T., 2006. Interaction between Mycobacterium avium subsp. paratuberculosis and environmental protozoa. BMC Microbiology 6, 63.
3278
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 2 7 1 e3 2 7 8
Whitlock, R.H., 2010. Paratuberculosis control measures in the USA. In: Behr, M.A., Collins, D.M. (Eds.), Paratuberculosis. Organism, Disease, Control. CAB International, Oxfordshire, UK, pp. 319e329. Whittington, R.J., Lloyd, J.B., Reddacliff, L.A., 2001. Recovery of Mycobacterium avium subsp. paratuberculosis from nematode
larvae cultured from the faeces of sheep with Johne’s disease. Veterinary Microbiology 81 (3), 273e279. Whittington, R.J., Marsh, I.B., Reddacliff, L.A., 2005. Survival of Mycobacterium avium subsp. paratuberculosis in dam water and sediment. Applied and Environmental Microbiology 71 (9), 5304e5308.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 2 7 9 e3 2 9 0
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Bacterial pathogens in Hawaiian coastal streamsdAssociations with fecal indicators, land cover, and water quality Emily J. Viau a, Kelly D. Goodwin b, Kevan M. Yamahara a, Blythe A. Layton a,1, Lauren M. Sassoubre a, Siobha´n L. Burns c, Hsin-I Tong c, Simon H.C. Wong a, Yuanan Lu c, Alexandria B. Boehm a,* a
Stanford University, Department of Civil & Environmental Engineering, 473 Via Ortega, Stanford, CA 94305, United States National Oceanic and Atmospheric Administration (NOAA), AOML, Miami, FL (stationed at SWFSC, San Diego, CA 92037), United States c University of Hawai’i at Manoa, Departments of Public Health Sciences and Microbiology, Honolulu, HI 96822, United States b
article info
abstract
Article history:
This work aimed to understand the distribution of five bacterial pathogens in O’ahu coastal
Received 7 January 2011
streams and relate their presence to microbial indicator concentrations, land cover of the
Received in revised form
surrounding watersheds, and physicalechemical measures of stream water quality. Twenty-
23 February 2011
two streams were sampled four times (in December and March, before sunrise and at high
Accepted 16 March 2011
noon) to capture seasonal and time of day variation. Salmonella, Campylobacter, Staphylococcus
Available online 12 April 2011
aureus, Vibrio vulnificus, and V. parahaemolyticus were widespread d12 of 22 O’ahu streams had all five pathogens. All stream waters also had detectable concentrations of four fecal indica-
Keywords:
tors and total vibrio with log mean standard deviation densities of 2.2 0.8 enterococci,
Salmonella
2.7 0.7 Escherichia coli, 1.1 0.7 Clostridium perfringens, 1.2 0.8 Fþ coliphages, and 3.6 0.7
Campylobacter
total vibrio per 100 ml. Bivariate associations between pathogens and indicators showed
Staphylococcus aureus
enterococci positively associated with the greatest number of bacterial pathogens. Higher
Vibrio
concentrations of enterococci and higher incidence of Campylobacter were found in stream
Fecal indicator
waters collected before sunrise, suggesting these organisms are sensitive to sunlight. Multi-
Tropical streams
variate regression models of microbes as a function of land cover and physicalechemical water quality showed positive associations between Salmonella and agricultural and forested land covers, and between S. aureus and urban and agricultural land covers; these results suggested that sources specific to those land covers may contribute these pathogens to streams. Further, significant associations between some microbial targets and physicalechemical stream water quality (i.e., temperature, nutrients, turbidity) suggested that organism persistence may be affected by stream characteristics. Results implicate streams as a source of pathogens to coastal waters. Future work is recommended to determine infectious risks of recreational waterborne illness related to O’ahu stream exposures and to mitigate these risks through control of land-based runoff sources. ª 2011 Elsevier Ltd. All rights reserved.
* Corresponding author. Tel.: þ1 650 724 9128; fax: þ1 650 723 7058. E-mail address:
[email protected] (A.B. Boehm). 1 Present address: Southern California Coastal Water Research Project, 3535 Harbor Blvd., Suite 110, Costa Mesa, CA 92626, United States. 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.03.033
3280
1.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 2 7 9 e3 2 9 0
Introduction
Each year exposures to marine waters contaminated by microbes cause an estimated 120 million gastrointestinal infections (GIs), 50 million acute respiratory infections (ARIs) (Shuval, 2003), and numerous skin infections (Yau et al., 2009). One source of microbial pollution to coastal waters is landbased runoff that discharges from rivers, streams and culverts to the nearshore. Runoff may contain pathogens from leaking sewage infrastructure, wild and domestic animal excreta, and other poorly understood environmental reservoirs such as soils and sands; in sum these are non-point sources of pollution (Boehm et al., 2009a). Several epidemiology studies have investigated health effects from recreational exposure to land-based runoff and the results are equivocal. Haile et al. (1999) found increased risks of GI and ARI for swimmers recreating near storm drains at a southern Californian marine beach and a correlation between risk and fecal indicator bacteria (FIB) concentrations. In contrast, Calderon et al. (1991) found no statistically significant association between swimmers’ illness risk and FIB in a freshwater pond contaminated by agricultural runoff. Dwight et al. (2004) found that Southern Californian surfers exposed to urban runoff had higher illness rates than Northern Californian surfers exposed to rural runoff. In these epidemiology studies, pathogen data were not readily available or limited and the exact source of fecal indicator organisms in the runoff was not known. Data on pathogens in runoff and insight into the factors that modulate their concentrations in runoff, would improve our ability to calculate and understand risks from exposure to terrestrial runoff. Most research on fecal pollution and risks from recreational swimming has been conducted in temperate climates. Both US and WHO standards for recreational water quality were promulgated using data collected in temperate zones (Boehm et al., 2009a). This is despite the fact that US tropical beaches receive more visitors than all temperate beaches combined (Leeworthy and Wiley, 2001). Furthermore, numerous studies find enterococci and Escherichia coli, the indicators used to assess water quality around the globe, in tropical soils and streams (Hardina and Fujioka, 1991; Hazen, 1988); there is concern that the presence of these organisms in tropical recreational waters may not indicate contamination or presence of pathogens. To improve waterborne pathogen surveillance in the tropics, researchers suggest monitoring alternative indicators, like Clostridium perfringens and Fþ coliphages (Fung et al., 2007; Luther and Fujioka, 2004). However, pathogen data are needed to corroborate whether or not an association exists between FIB, alternative indicators, and pathogens in the tropics. In the present study, we measured human bacterial skin and GI pathogens in tropical coastal streams discharging to marine waters, and tested their association with traditional and alternative indicator organisms, surrounding land cover, and physicalechemical water quality. Specifically, we documented the occurrence or concentrations of Salmonella, Campylobacter, Staphylococcus aureus, Vibrio parahaemolyticus and Vibrio vulnificus in 22 tropical streams of O’ahu, Hawai’i using a combination of culture-based and molecular methods.
One goal was to test pathogen associations with indicators including E. coli, enterococci, C. perfringens, Fþ coliphage, and total vibrio to determine which indicators have predictive power of bacterial pathogens. A second goal was to build multivariate statistical models to understand how land cover and physicalechemical stream parameters controlled pathogens and indicators. The modeling is premised on a conceptual model where bacterial concentrations in streams (1) increase due to microbial fluxes from the surrounding land; fluxes are affected by land cover, and (2) change in response to physicalechemical characteristics of the stream that affect organism persistence. The work presented here is unique in that it investigates the distribution of both GI and non-GI pathogens in a tropical climate, two research needs specifically mandated in the US BEACH Act of 2000.
2.
Materials and methods
2.1.
Sampling sites
Twenty-two streams were identified on O’ahu, Hawai’i for sampling (Table 1). Streams were selected because they discharge to coastal waters adjacent to popular swimming beaches. We also selected streams that drained watersheds with diverse land covers. In all cases, there were no known sewage point sources to the streams; all watersheds had ) was hio separate storm and sewage conveyances. One site (Ku a storm drain.
2.1.1.
Land cover
Land cover was determined for stream watersheds using ArcGIS (ESRI, Redlands, CA) and Hawai’i Land Cover 2001 data (NOAA 2001). The data consist of Landsat Enhanced Thematic Mapper data at 30 m resolution for 18 land coverage classes. These classes were aggregated into four broad categories: urban, agriculture, forested, and other (unclassified, unconsolidated shore, water, and bare land). Watershed boundary data were obtained from the Hawai’i Statewide GIS program (Hawaii, 2010). The fraction of each watershed that was urban, agricultural, and forested was calculated by normalizing the areas, respectively, by the total watershed area.
2.1.2.
Field sampling
Two water sampling campaigns were conducted at the 22 sites over five days in December 2009 (14e18 Dec 2009) and March 2010 (28 Mare3 Apr 2010). Daily precipitation during and prior to the field campaigns was obtained from a centralized rain gauge on O’ahu (Moanalua USGS site 212359157502601). It should be noted that rainfall is spatially variable over the island, so data from this gauge represent approximate rainfall in the studied watersheds. The streams have very limited USGS stream gauge coverage, so flow data were not available. During each campaign, the 22 sites were visited twice (once before the sun rose and once at high noon); in sum each stream was sampled four times. Twenty-liter water samples were collected in triple rinsed, 10% HCl-washed plastic containers. Water was sampled at an accessible location near the intersection of the stream and coastal ocean (Table 1).
3281
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 2 7 9 e3 2 9 0
Table 1 e O’ahu Stream Survey Site Description, Land Use. Hawai’i department of health water quality standards are provided where relevant. Stream values in exceedance of these standards are bolded. Location Stream
Watershed land cover
Ancillary Measurements
Latitude/ Associated FOR URB AG TEMP Salinity DO CHLa TURB NO2þNO3 DINd PO4 (mg/l) (mg/L)c (NTU)b,c (mg-N/L)c (mg-N/L)c (mg-P/L)c Longitude of beach ( C) sample collection
Ala Waia 21.288 N, 157.839 W Ala Moana Storm 21.271 N, 157.824 W hio hio Ku Ku Draina Wai’alae Kapakahi 21.270 N, 157.778 W Kahala Wai’alae Golf 21.273 N, 157.771 W Coursea 21.278 N, 157.750 W Wailupe Wailupea Moanalua 21.333 N, 157.894 W Ke’ehi lagoon Kalihi 21.332 N, 157.891 W Ke’ehi lagoon kua kua Ma Ma 21.530 N, 158.229 W Kaupuni 21.448 N, 158.193 W Poka’i Bay ’ili’ilia 21.429 N, 158.180 W Ma’ili Ma ’ilia 21.409 N, 158.177 W Ma’ili Ma na kuli na kuli Na Na 21.376 N, 158.140 W leakahana laekahana 21.673 N, 157.936 W Ma Ma Waimea Waimea 21.641 N, 158.063 W Hale’iwa Anahulu 21.594 N, 158.103 W Kaiaka Paukauila 21.580 N, 158.117 W Kaiaka Kiikii 21.579 N, 158.120 W nolo nalo 21.365 N, 157.709 W Waima Waima Kailua Ka’elepulu 21.398 N, 157.726 W Kailua Kawainui 21.426 N, 157.741 W Kahana Kahana 21.556 N, 157.869 W Punalu’u Punalu’u 21.579 N, 157.885 W All Stream Average Hawai’i DOH Water Quality Standardse
18% 79% 0% 18% 79% 0%
24.0 24.5
25.2 33.2
5.3 6.5
2.1 0.2
2.3 1.2
319 52
388 72
16 8.1
54% 45% 0% 67% 31% 0%
23.7 24.1
30.6 23.6
4.5 7.1
1.6 1.3
10 15
22 47
141 92
11 14
67% 79% 72% 97% 83% 78% 78% 85% 90% 98% 80% 38% 27% 81% 47% 72% 98% 95%
24.6 25.0 24.0 23.9 24.7 23.6 24.6 23.5 23.7 23.6 23.0 23.1 22.8 24.4 24.3 24.5 22.1 21.7 23.8 e
22.4 33.5 23.5 37.3 30.0 34.3 30.1 20.9 0.6 7.8 14.9 11.0 10.2 17.2 17.5 11.7 4.8 1.2 20.1 e
6.2 4.6 5.3 6.3 6.0 7.0 6.5 5.0 3.5 6.4 6.6 4.7 7.9 5.0 6.9 5.3 5.3 7.4 5.9 >5
1.1 4.0 2.6 7.9 1.1 0.7 0.6 6.1 0.5 1.4 0.5 3.6 7.4 1.8 3.3 2.2 0.3 0.3 1.5 e
7.9 7.7 5.1 4.4 4.7 1.8 3.3 6.7 5.7 3.5 1.8 5.4 6.7 3.9 1.8 2.8 1.9 1.8 3.9 <5
29 45 76 3.4 186 117 2128 5.9 5011 32 1310 1142 350 809 8.6 10 26 33 26 <70
83 133 159 4.5 218 146 2163 55 5143 87 1339 1250 358 922 34 22 52 49 41 <250f
12 7.1 15 1.4 25 2.8 4.2 7.2 7.6 4.0 88 39 4.8 33 2.3 14 5.6 9.7 9.4 <50f
31% 21% 28% 2% 13% 16% 16% 13% 3% 0% 3% 24% 29% 14% 52% 20% 0% 1%
0% 0% 0% 0% 2% 2% 2% 0% 6% 2% 17% 37% 42% 3% 0% 0% 0% 4%
, Ma ’ili/Ma ’ili’ili and Wai’alae Golf Course/Wailupe have shared watersheds. hio a Ala wai/Ku b NTU ¼ Nephelometric Turbidity Units. c Values represent geometric mean of n ¼ 4. d DIN ¼ dissolved organic nitrogen, which is the sum of NO2, NO3, and NH4þ. e Hawai’ian Department of Health (2001) water quality standard during rainy season for geometric mean. f Standards for total nitrogen and phosphorusdnote that stream values do not include organic nitrogen or phosphorus so are not directly comparable.
Field sampling controls were taken every other day by performing sampling procedures with distilled waterdfield blanks were assayed concurrently with water samples to confirm sterility of sampling procedures for all microbial assays. At the stream, salinity and temperature were measured using a sensor (YSI85, YSI Inc., Yellow Springs, OH), dissolved oxygen (mg/l) was measured with an optical probe (ProODO, YSI Inc.), and turbidity was measured using a benchtop turbidity meter (HF Scientific DRT-15CE, Fort Myers, FL).
2.2.
Laboratory analyses
Water samples were stored on ice during transport to the laboratory and processed within 6e8 h of collection in accordance with EPA and standard methods for water sampling (AWWA 2005, USEPA 2006).
2.2.1.
Fecal indicator and total vibrio enumeration
Samples were assayed for both traditional and alternative indicators, including E. coli, enterococci, C. perfringens, Fþ
coliphages, and total vibrio. In all cases, membrane filtration was used; water was filtered through 0.45 mm pore-size mixed cellulose ester filters (S-Pak Type HA, Millipore, Billerica, MA). Filtration blanks were run during each day of sampling for all assays and in all cases indicated no cross contamination. E. coli were enumerated in 1 ml, 10 ml, and 100 ml volumes with MI media (Difco Laboratories Inc, Detroit, MI), and incubated at 35 C for 24 h according to EPA method 1604 (lowest detectable concentration (LDC) ¼ 1 colony forming unit (CFU)/100 ml) (USEPA 2002). The same volumes were used to quantify enterococci by EPA Method 1600 with mEI agar (EMD Chemicals, Gibbstown, NJ) incubated at 41 C for 24 h (USEPA 2006) (LDC ¼ 1 CFU/100 ml). For C. perfringens, the Hawai’i Department of Health procedure was used for detection in 10 ml, 100 ml, and 500 ml water samples with mCP agar (see supporting materials (SM)) (LDC ¼ 0.2 CFU/ 100 ml). Total vibrio were assayed in 0.1 ml, 1 ml and 10 ml volumes with TCBS agar (EMD) following Hsieh et al. (2008). Filters on TCBS were incubated for 24 h at 35 C and yellow/ green colonies were counted as total vibrio (LDC ¼ 10 CFU/ 100 ml).
3282
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 2 7 9 e3 2 9 0
Fþ coliphages were enumerated following Boehm et al. (2009b). MgCl2 was added to a final concentration of 0.05 M in the water sample (Victoria et al. 2009) and 10 ml, 100 ml, and 500 ml were membrane filtered. Filters were placed on 0.3 ml of sterile 50% 1 PBS/50% glycerol and frozen at 80 C. Samples were assayed by a modified double agar layer method (USEPA 2001a) as described in the SM (LDC ¼ 0.2 plaque forming units (PFU)/100 ml).
2.2.2.
Culture-dependent pathogen detection
In each sample, the presence/absence of Salmonella, Campylobacter, and S. aureus was measured while Vibrio vulnificus and V. parahaemolyticus were quantified. The quantification of Salmonella, Campylobacter, and S. aureus was not possible due to finite time and staffing available during the field work. Presumptive isolates were confirmed with molecular methods using the respective primers and probes listed in Table 2donly pathogen isolates positive by both culture- and PCR-based confirmation methods are reported (quantification cycle < 35 were considered positive). Further details of pathogen assays are provided in the SM. Salmonella (SAL) presence/absence from 1 L water samples (LDC ¼ 1 most probable number (MPN)/L) was determined following Shellenbarger et al. (2008). Presumptive positives were confirmed using Salmonella-genus-specific PCR (Table 2). Campylobacter (CAMPY) presence/absence was determined for 1 L of water with a two-step enrichment procedure modified from Khan and Edge (2007) (LDC ¼ 1 MPN/L). Presumptive CAMPY isolates (two isolates for each positive sampling event) were first confirmed with Campylobacter genus-specific PCR followed by further typing of isolates as C. jejuni or C. coli using endpoint Taqman PCR (Table 2). S. aureus (STAPH) detection followed Goodwin and Pobuda (2009) (LDC ¼ 1.2 CFU/100 ml). Water volumes up to 80 ml were filtered and filters were placed on CHROMagar Staph aureus (BD, Franklin Lakes, New Jersey) at 37 C for 24 h. Putative STAPH were restreaked to isolation to confirm morphology followed by S. aureus-specific PCR (Table 2).
V. vulnificus (VVUL) and V. parahaemolyticus (VPARA) were quantified by filtering 0.1 ml, 1 ml, and 10 ml volumes and placing filters on CHROMagar Vibrio (BD). Plates were incubated at 40 C for 24 h (LDC ¼ 10 CFU/100 ml) (further details can be found in SM). Presumptive isolates were confirmed by endpoint Taqman PCR (Table 2).
2.2.3.
Nutrients and chlorophyll a
For nutrients analyses, 30 ml of water were filtered through 0.2 mm pore-size PES syringe filters (Millipore) and stored at 20 C. Phosphate, nitrate, nitrite, and ammonium were measured by standard methods with a nutrient autoanalyzer (Lachat QuikChem 8000, Loveland, CO). Lowest detectable concentrations were 0.5 mg/L PO3 4 -P/L, 2.8 mg NO3 -N/L, 1.4 mg þ NO2 -N/L, and 1.4 mg NH4 -N/L. To measure chlorophyll a, water was filtered in the dark through GF/F filters until pale green was seen (60e240 ml) and immediately frozen. Chlorophyll a was determined following EPA method 445 (USEPA 1997).
2.3.
Statistical analyses
Statistical analyses were carried out using Minitab 15.0 or SPSS 18.0. For continuous microorganism data, non-detects were substituted with lowest detectable concentrations (Harwood et al., 2005) and raw/transformed data (e.g., log10, square-root) were evaluated for normality with quantileequantile plots. Bivariate pairwise associations between bacterial pathogens were assessed using contingency tables, one-way analyses of variance (ANOVA), and Pearson’s correlation coefficient (rp). The same methods were used to investigate time of sampling effects. Bivariate associations between pathogens and microbial indicators were tested using generalized estimating equations (GEEs), with binomial logistic regression models for dichotomous dependent variables and linear regression models for continuous variables. Multivariate GEE models were developed for microorganisms as a function of land cover and physicalechemical water quality
Table 2 e PCR primers and probes used in the study. Target Group (gene) Salmonella (invA gene) Campylobacter (16S rRNA) Campylobacter jejuni (hipO) Campylobacter coli ( glyA) Staphylococcus aureus (clfA) Vibrio vulnificus (vvhA/B) Vibrio paraheomolyticus ( gyrB)
Primer and probe sequences (50 -30 )
Size (bp)
F-primer: GTGAAATTATCGCCACGTTCGGGCAA R-primer: TCATCGCACCGTCAAAGGAACC F-primer: CTAGAGTACAAACTAATAAGTCTC R-primer: ATTCTAAAACGCATCACTTCCTTG F-primer: TGCTAGTGAGGTTGCAAAAGAATT R-primer: TCATTTCGCAAAAAAATCCAAA Probe: FAM-ACGATGATTAAATTCACAATTTTTTTCGCCAAA-BHQ F-primer: CATATTGTAAAACCAAAGCTTATCGG R-primer: AGTCCAGCAATGTGTGCAATG Probe: FAM-TAAGCTCCAACTTCATCCGCAATCTCTCTAAATTT-BHQ F-primer: GCAAAATCCAGCACAACAGGAAACGA R-primer: CTTGATCTCCAGCCATAATTGGTGG F-primer: TGCCT(AG)GATGTTTATGGTGAGAAC R-primer: TCGACTGTGAGCGTTTTGTC Probe: FAM-TAGCCGAGT(AG)GCATCCGATCGTTGTT-BHQ F-primer: TGAAGGTTTGACTGCCGTTGT R-primer: TGGGTTTTCGACCAAGAACTCA Probe: FAM-TTCTCACCCATCGCCGATTCAACCGC-BHQ
284
(Malorny et al., 2003)
650e800
(Khan and Edge, 2007)
Reference
100
(LaGier et al., 2004)
133
(LaGier et al., 2004)
638
(Mason et al., 2001)
179
(Wetz et al., 2008)
148
(Cai et al., 2006)
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 2 7 9 e3 2 9 0
characteristics. GEEs were necessary because observations at streams represented repeated measurements and this technique accounts for data clustering amongst sites (Hardin and Hilbe, 2003; Walters et al., 2011). Statistics were deemed significant if P 0.05; some marginally significant associations (0.05 < P 0.1) were also discussed.
3.
Results
Two sampling campaigns were designed to capture broad seasonal (December, March) and time of day (before sunriseAM, high noon-PM) differences in stream water quality during the Hawai’i rainy season (Nov 1eApr 30). Average rainfall during sample collection days was 0.1 cm/d in December and 0.3 cm/d in March, and was lower than the 1.5 cm/d overall average rainfall during the 2009e2010 rainy season. Thus, conditions were relatively dry. All data were evaluated for normality and transformed if necessary. Raw data were used for salinity, temperature (TEMP), and dissolved oxygen in mg/l (DO). Log10-transformations were employed for all microorganisms, turbidity (TURB), chlorophyll a (CHLa), and nutrients (PO3 4 , NO2 , NO3). Ammonium (NH4þ), forested (FOR), urban (URB), and agricultural (AG) land cover fractions were normalized by square-root transformations.
3.1. Land cover and physicalechemical properties of streams Land cover within the watersheds was largely tropical forest cover (FOR), but 18 of 22 streams had between 1% and 79% urban land cover (URB, Table 1). Agricultural land cover (AG) was present in 10 of 22 stream watersheds ranging from 2 to 42% of the total watershed area. URB and FOR land covers were strongly, negatively correlated (rp ¼ 0.918, P < 0.001), so URB is used hereafter to represent both variables. Mean stream salinity, TEMP, TURB, CHLa, and nutrient concentrations are reported in Table 1. Salinity ranged from 0.6 to 34.3, illustrating that streams spanned a range of marine influence. DO in 4 of 22 streams was below the state rainy season standard of 5 mg/l. Numerous streams exceeded water quality criteria for TURB and nutrients. Only TEMP and DO varied significantly between AM and PM samples (PM samples had higher TEMP by 2.8 C and DO by 1.5 mg/L, ANOVA, P 0.001). Other physico-chemical parameters showed no AM/PM fluctuations but varied significantly between streams (ANOVA, P ¼ 0.001). PO3 4 was the only parameter that differed significantly between DEC/MAR (ANOVA, P ¼ 0.011) with PO3 4 higher in MAR by 1.8 mg P/L relative to DEC. All collected water quality data are provided in Table S1. TEMP, salinity, and DO were significantly correlated (rp ¼ 0.32 for DO-TEMP and salinity-TEMP, P < 0.005), so TEMP was used to represent all three variables in multivariate models. Nutrient concentrations were positively correlated (0.40 < rp < 0.95, P < 0.001), so PO3 4 was used to represent nutrients. Additionally, TURB and CHLa were correlated (rp ¼ 0.35, P ¼ 0.001) so TURB was used to represent both parameters.
3.2.
3283
Pathogens
The presence of at least one pathogen, including Salmonella (SAL), Campylobacter, (CAMPY), and/or S. aureus (STAPH) was detected in 21 of the 22 streams (Fig. 1). V. vulnificus (VVUL) and V. parahaemolyticus (VPARA) were also widespread (Table S2). SAL was present in 15 of 22 streams (Fig. 1a) and was detected in two or more samples in 8/22 streams. There were no significant AM/PM or DEC/MAR differences in SAL occurrence (c2 test, P > 0.05). CAMPY were present in 18 of 22 streams (Fig. 1b). While no DEC/MAR differences were found in CAMPY (c2 test, P > 0.05), CAMPY was detected at a significantly higher frequency in AM compared to PM samples (P < 0.001). Confirmed CAMPY isolates from each of 25 samples were further typed to distinguish C. jejuni (CJEJ) and C. coli (CCOLI). In all but two samples, the confirmed CAMPY isolates were typed as the same species. In four of the 25 samples CCOLI were detected, in 7 CJEJ was detected, in 2 both CJEJ and an unknown CAMPY species were detected, and in the remaining 12 samples unknown CAMPY species were detected. In summary, 4 streams had more than one type of CAMPY isolated either in a single or multiple samples, while CAMPY isolated from 5 and 2 streams, respectively, were identified as CJEJ and CCOLI exclusively (Fig. 1b). STAPH was present in all but three streams and detected in 14/22 streams at least twice (Fig. 1c). STAPH was detected at a higher frequency in MAR than DEC (c2 test, P ¼ 0.001) with a maximum detection in 31% of the samples in DEC versus 68% in MAR; no differences were observed between AM/PM (P > 0.05). During DEC sampling, all samples were also tested for methicillin resistant S. aureus (MRSA) (methods in SM) but ’ili’ili, had a positive isolate so further only one stream, Ma analyses were not undertaken in MAR. Concentrations of VVUL and VPARA averaged 2.3 0.9 and 2.4 0.7 log CFU/100 ml, respectively. Reported densities include both pathogenic and non-pathogenic subspecies since PCRs did not target specific VVUL and VPARA virulence genes. No variations were noted between AM/PM or DEC/MAR samples but differences between streams were significant for both pathogens (Table S2, P < 0.05). Twelve of 22 streams were positive for SAL, STAPH, CAMPY, VVUL, and VPARA, but there were only a few significant associations between these organisms. SAL was detected at a higher frequency when STAPH was present (c2 test, P ¼ 0.012). VVUL was higher when SAL was present (ANOVA, P ¼ 0.012) and when CAMPY was present (ANOVA, P ¼ 0.017). VVUL concentrations were significantly correlated to VPARA (rp ¼ 0.42, P < 0.001).
3.3.
Indicators
E. coli (EC), Enterococcus (ENT), C. perfringens (CPERF), and Fþ coliphages (F þ PHAGE) as well as total vibrio (TOTVIB) were detected in all streams (Fig. 2). There were no significant differences in EC, CPERF, or TOTVIB between AM/PM or MAR/ DEC samples (ANOVA, P > 0.05). Seasonal fluctuations were significant for ENT (P ¼ 0.048) and F þ PHAGE (P ¼ 0.001), with MAR densities being lower than DEC by 0.25e0.5 log CFU/PFU per 100 ml. ENT concentrations also showed time of day
3284
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 2 7 9 e3 2 9 0
a
differences (P ¼ 0.001) with AM values averaging 0.5-log higher than PM values.
3.4.
b
Each pathogen tested (except STAPH and CCOLI) showed a positive, significant association (P 0.05) to at least one indicator based on GEE models (Table 3). SAL, CAMPY, and CJEJ were positively associated with ENT concentrationsdENT was 0.6e1 log higher when one of these pathogens was present compared to when it was absent (ANOVA, P < 0.05). ENT was marginally positively associated with STAPH (P < 0.1). CAMPY was marginally positively associated with CPERF e CPERF was 0.3 log unit higher when CAMPY was detected (ANOVA, P ¼ 0.09). SAL was also marginally associated with CPERF (P < 0.1). VVUL was positively associated with all indicators tested, while VPARA was positively associated with TOTVIB, CPERF, and EC.
3.5.
c
Pathogen associations with indicators
Multivariate models of microorganisms
Multivariate models of microorganisms were created using GEE models (Table 4, Fig. S1). The land cover variables URB, AG and physico-chemical characteristics TEMP, TURB, and PO3 4 were included as independent variables. SAL occurrence was marginally associated with AG, URB, TEMP, and PO3 4 (P < 0.10)eassociations were positive for all variables except URB. CAMPY occurrence was negatively associated with TEMP (P < 0.001) and marginally negatively associated with AG (P ¼ 0.1). STAPH was positively associated with both AG and URB (P < 0.05). No significant water quality or land use associations were found for VVUL while VPARA was marginally positively associated with URB and TURB (P < 0.1). For indicators, ENT, TOTVIB, and F þ PHAGE were significantly associated with more than one independent variable. ENT had a positive association with TURB and a negative association with TEMP (P < 0.05). While only of marginal significance (P < 0.1), ENT was also positively associated to both AG and PO3 4 . TOTVIB showed a significant negative relationship to AG and positive relationship to TEMP (P < 0.05). F þ PHAGE was negatively associated with AG (P < 0.05) and TEMP (P¼0.1), and positively associated with PO3 4 (P < 0.1). One stream parameter was significant for each of the other indicatorsdEC was positively related to PO3 4 (P < 0.05) and CPERF was positively associated with TURB (P < 0.05).
3.6. Comparison between ENT and CPERF as pollution indices Fig. 1 e Presence/absence of a) Salmonella, b) Campylobacter, and c) Staphylococcus aureus in 22 O’ahu coastal streams (shaded [ present, white [ absent) by time of day (AM/ PM) and season (DEC/MAR). Circles denote DEC with top half and MAR samples with bottom half. AM presence is on the left and PM presence is on the right. For CAMPY (1b), positive C. jejuni are lines, positive C. coli are dots, positive C. jejuni and other CAMPY are black, while other CAMPY are shaded. For STAPH (1c), positive MRSA in March is indicated by stripes while STAPH analyses not completed are black in that part of the circle.
Some researchers suggest that ENT found in pristine tropical soils interfere with their ability to predict pollution from fecal sources (Boehm et al., 2009a). Fujioka and colleagues suggest that CPERF may be a better indicator of fecal pollution than ENT for tropical waters and that CPERF concentrations can discern pollution sources based on a “Fung/Fujioka scale” of pollution (Fung et al., 2007). According to this scale, when CPERF is greater than 100 CFU/100 ml, sewage is the pollution source. When CPERF is between 10 and 100 CFU/100 ml, nonpoint pollution is the source. Finally, waters are considered
3285
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 2 7 9 e3 2 9 0
Fig. 2 e Log fecal indicator and total vibrio concentrations in 22 O’ahu coastal streams by time of day (AM/PM) and season (DEC/MAR)dindicators include E. coli (CFU), enterococci (CFU), C. perfringens (CFU), FD coliphages (PFU), and total vibrio (CFU). Box- and-whisker plots represent the median (inner box line), 25th and 75th percentiles (lower and upper outer box lines), whiskers cover 10th and 90th percentiles and data outliers are represented by open circles. Geometric means are indicated with a black circle (n [ 22).
uncontaminated when CPERF is present at less than 10 CFU/ 100 ml. We plotted CPERF versus ENT (Fig. 3) to gain insight on how the CPERF pollution scale compares to the USEPA ENT standard of 104 CFU/100 ml. Using the Fung/Fujioka scale, 53 of 88 samples indicated non-point source pollution while 6 samples indicated sewage pollution. Forty-four of these 59 “contaminated” samples (e.g. CPERF10 CFU/100 ml) also exceeded the ENT standard. Agreement between the Fung/ Fujioka scale for contamination and single-sample exceedance for ENT, as well as the pairwise correlation between ENT and CPERF (rp ¼ 0.50, P < 0.05) suggests that similar information can be obtained from ENT as CPERF in these tropical streams.
4.
Discussion
4.1.
Bacterial pathogens
Salmonella, Campylobacter, S. aureus, V. vulnificus, and V. parahaemolyticus, are implicated in recreational water and shellfish
outbreaks (USEPA 2009b). In this study, multiple isolations of these organisms in a given stream were frequent (Fig. 1, Table S2). STAPH and Vibrio spp. were most commonly detected (STAPH was present in 19/22 streams, while Vibrio spp. were detected in all streams) followed by CAMPY (18/22) and SAL (15/22); all pathogens were isolated from 12/22 streams. Salmonella, a leading cause of gastroenteritis in the US (USEPA 2009b), are frequently isolated from surface waters (Haley et al., 2009; Walters et al., 2011; Wilkes et al., 2009). We found SAL in the majority of O’ahu streams at concentrations greater than 1 MPN/1 L. SAL occurrence was positively associated with higher temperature (thus higher DO and salinity as well), nutrients, and AG; while negatively associated with URB (Table 4). Relationships between water quality and SAL may imply increased Salmonella persistence in warm, eutrophic, relatively saline water. Previously, Salmonella in microcosms showed decreased persistence in warm relative to cool waters and no real trends in persistence in variable saline waters (Evison, 1988; Wait and Sobsey, 2001), while Evison (1988) reported increased Salmonella persistence in waters
Table 3 e Bivariate pathogeneindicator relationships. GEE logistic regression (logit function) was used for dichotomous variables while GEE linear regression was applied for continuous variables. b is the parameter coefficient; P-value is provided. Microorganism
Enterococci b
Salmonella Campylobacter C. jejuni C. coli S. aureus V. vulnificus V. parahaemolyticus a P0.05. b 0.05 < P < 0.1.
1.09 1.12 2.23 0.10 0.49 0.43 0.16
E. coli
P-value a
0.012 <0.001a <0.001a 0.791 0.097b 0.001a 0.193
b 0.02 0.19 0.27 0.45 0.35 0.24 0.26
P-value 0.948 0.458 0.354 0.514 0.245 0.037a 0.004a
C. perfringens
Fþ coliphage
b
b
P-value
b
P-value
0.321 0.60 0.66 0.12 0.35 0.26 0.18
0.187 0.075b 0.124 0.761 0.119 0.019a 0.122
0.06 0.15 0.08 0.16 0.35 0.32 0.41
0.828 0.688 0.877 0.743 0.315 0.026a <0.001a
0.609 0.64 0.88 0.50 0.09 0.49 0.32
P-value b
0.095 0.030a 0.130 0.275 0.804 <0.001a 0.003a
Total Vibrio
3286
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 2 7 9 e3 2 9 0
Table 4 e Pathogen/indicator multivariate models with stream physico-chemical water quality parameters and watershed land use. Pathogen
GEE Model Parameter
Salmonella QICC ¼ 102.097a
Campylobacter QICC ¼ 106.064
S. aureus QICC ¼ 112.386
V. vulnificus QICC ¼ 78.53
V. parahaemolyticus QICC ¼ 50.306
Intercept AG URB TEMP TURB PO3 4 Intercept AG URB TEMP TURB PO3 4 Intercept AG URB TEMP TURB PO3 4 Intercept AG URB TEMP TURB PO3 4 Intercept AG URB TEMP TURB PO3 4
Coefficient 7.727 3.044 1.848 0.249 0.044 1.436 9.645 1.464 0.707 0.435 0.397 0.114 1.798 4.993 3.719 0.018 0.311 0.298 2.38 0.059 0.075 0.016 0.109 0.404 1.18 0.472 0.790 0.023 0.348 0.245
Indicator P-value b
0.024 0.084c 0.061c 0.076c 0.934 0.061c 0.001b 0.100c 0.506 <0.001b 0.572 0.843 0.576 0.046b 0.005b 0.903 0.704 0.644 0.051c 0.905 0.874 0.750 0.681 0.171 0.177 0.109 0.053c 0.543 0.084c 0.132
GEE Model Parameter
Enterococci QICC ¼ 52.416
E. coli QICC ¼ 52.573
C. perfringens QICC ¼ 48.224
Fþ coliphages QICC ¼ 58.575
Total vibrio QICC ¼ 48.826
Coefficient
P-value
5.427 0.473 0.188 0.163 0.543 0.338 3.105 0.314 0.087 0.040 0.109 0.450 0.376 0.217 0.246 0.006 0.904 0.169 2.498 0.692 0.132 0.066 0.078 0.322 1.620 0.951 0.243 0.072 0.376 0.035
<0.001b 0.086c 0.479 <0.001b 0.006b 0.087c 0.003b 0.236 0.739 0.394 0.566 0.004b 0.629 0.579 0.491 0.832 <0.001b 0.413 0.003b 0.007b 0.710 0.100c 0.742 0.075c 0.050b 0.014b 0.621 0.041b 0.116 0.810
Intercept AG URB TEMP TURB PO3 4 Intercept AG URB TEMP TURB PO3 4 Intercept AG URB TEMP TURB PO3 4 Intercept AG URB TEMP TURB PO3 4 Intercept AG URB TEMP TURB PO3 4
a Corrected Quasi likelihood under Independence Model criterioneinformation criteria are in small-is-better form. b P0.05. c 0.05 < P < 0.1.
with high nutrients. The positive AG and negative URB land cover associations suggest SAL sources related to agricultural activities (e.g. animal rearing, fertilizers) and within forested land covers (e.g. feral pigs, goats) may be important. This finding contrasts with Walters et al. (2011) who reported positive SAL associations with urban land cover in central California, but is consistent with frequent Salmonella isolations in studies of agricultural watersheds (Wilkes et al., 2009) and rural watersheds with known animal inputs (Haley et al., 2009). C. jejuni and C. coli are the most important human pathogens of the 17 Campylobacter species (USEPA 2009b), and results indicate they were prevalent in O’ahu streams, with C. jejuni in 8/18 streams and C. coli in 4/18 streams. CAMPY incidence was greater in AM compared to PM samples and was negatively associated with TEMP (thus also DO and salinity). Increased CAMPY occurrence when DO, TEMP, and salinity were lower is consistent with CAMPY being sensitive to oxygenated conditions and having increased persistence in low temperature, low salinity waters (Buswell et al., 1998). CAMPY detection in low salinity waters may also be explained by CAMPY-rich freshwater runoff being diluted by marine water. CAMPY was largely isolated in AM samples suggesting sensitivity to sunlight. Sinton et al. (2007) found CAMPY were
more sensitive to sunlight than SAL and EC. Finally, no correlations to land cover were found for CAMPY, but previous studies found that both human and animal sources contributed to surface water contamination (e.g. Vereen et al., 2007) so further work should determine sources of Campylobacter in O’ahu streams. STAPH was ubiquitous in O’ahu streams. Frequent isolation of S. aureus in coastal waters around O’ahu (Charoenca and Fujioka, 1993) and other recreational waters is common (Yau et al., 2009). Charoenca and Fujioka (1993) suggest that beach-goer shedding is the source of STAPH to O’ahu coastal waters. Our results suggest that STAPH may also emanate from coastal streams. Presence of STAPH was associated with URB and AG land covers suggesting that sources within those land covers contribute STAPH to streams. Future work is needed to pinpoint STAPH sources in streams and determine whether environmental reservoirs exist. S. aureus infections are a serious problem in Hawai’idthe state leads the US in annual deaths from STAPH and MRSA infections (HHIC 2007) Beach found S. aureus hio and an epidemiology study at Ku infections were four times more likely with marine water exposures (Charoenca and Fujioka, 1995). Vibrio vulnificus and V. parahaemolyticus are leading causes of shellfish-related illness (Iwamoto et al., 2010) and the two
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 2 7 9 e3 2 9 0
3287
could be higher in streams draining URB land covers. Future work will need to further investigate this association.
4.2.
Fig. 3 e C. perfringens versus enterococci (n [ 88). Black symbols indicate when enterococci were over 104 CFU/ 100ml, while white symbols indicate when under. Horizontal lines show 10 CFU and 100 CFU/100 ml of C. perfringens, which delineates non-point and sewage sources according to the Fung/Fujioka scale.
vibrio pathogens most frequently associated with US recreational water outbreaks (Dziuban et al., 2006)dthis study found that both organisms were widespread in O’ahu streams. It should be noted that PCRs used to confirm Vibrio species did not target virulence genes, so reported densities likely overrepresent pathogenic strains. V. vulnificus concentrations observed here were comparable to those measured in the Gulf of Mexico by Panicker et al. (2004) and during a tropical storm by Wetz et al. (2008) while V. parahaemolyticus levels were higher than those recorded by DePaola et al. (1990) in marine waters off the continental US. Various studies (e.g. Hsieh et al., 2008; Johnson et al., 2010) have shown that temperature and salinity were most important in modulating both V. vulnificus and V. parahaemolyticus concentrationsdoptimal environments for both pathogens include warmer temperatures (15e30 C) and mesosaline waters (5e25). No temperature or salinity associations were found here for either VVUL or VPARA. The lack of temperature correlation could be due to all streams being within optimal temperature ranges for vibrios. The lack of a linear salinity correlation could be due to the fact that mesosaline conditions are optimal (Johnson et al., 2010; Wetz et al., 2008). VPARA showed a marginal positive association with turbidity, consistent with Johnson et al. (2010) who found VPARA correlated to turbidity/chlorophyll a in the Gulf of Mexico. No land cover associations were found for VVUL, but VPARA were marginally positively related to URB land cover. Since vibrios are indigenous it is unlikely that this organism emanates from sources within URB land covers. Association does not prove causation, so the relationship between vibrios and URB land cover may arise due to an unmeasured variable which covaries with URB land cover. For example, perhaps vibrio growth or persistence is affected by DOC (not measured herein) which
Pathogeneindicator relationships
A goal of this study was to understand which indicators were best at predicting pathogens in tropical streams. Currently, the Hawai’i Department of Health (DOH) monitors both enterococci and C. perfringens in waters to dictate pollution advisories. Fþ coliphages are also suggested as a more conservative indicator of sewage pollution in the tropics due to increased persistence compared to ENT and EC (Fujioka and Yoneyama, 2002; Luther and Fujioka, 2004). No one indicator was associated with all pathogens, as has been reported previously in studies of temperate waters (Harwood et al., 2005; Wilkes et al., 2009). ENT correlated with SAL, CAMPY, CJEJ, and VVUL (and marginally with STAPH) while CPERF was associated with CAMPY, VVUL, and VPARA (and marginally with SAL) (Table 3). Strong relationships between ENT and CAMPY were reported previously in temperate stream waters (Wilkes et al., 2009) but the same study found EC was a better surrogate for SAL and parasites. In the current study, EC was only associated with VVUL. Fþ coliphages were also significantly associated with VVUL while only marginally correlated with CAMPY (P < 0.1). TOTVIB was a good indicator of both VVUL and VPARA. Overall, ENT was the best surrogate for bacterial pathogens in tropical stream waters, but CPERF was also a good predicator. ENT and CPERF concentrations covaried and the ‘FungeFujioka Scale’ of pollution was consistent with the current USEPA enterococci marine water standards for monitoring fecal pollution. Further quantification of bacterial and viral pathogens in streams could improve associations found here and may also show that Fþ coliphages are indicative of viral pathogens. Multivariate analyses of indicators highlight a number of possible reasons for pathogen/indicator correlations in these tropical streams (Table 4). For ENT and SAL, both had marginal relationships to AG land cover and nutrients suggesting that both can derive from agricultural land sources. ENT and CAMPY associations may be due to their negative correlations to the temperature/salinity/DO water quality parameters and increased ENT concentration/CAMPY occurrence in samples taken prior to sunrise; these findings suggest that both organisms are light sensitive and/or persist under similar water conditions. Finally, relationships between CPERF and VPARA could be due to their concurrent associations with turbidity/chlorophyll a.
4.3.
Limitations and future work
There are several limitations to this study. Associations between land cover and several microbial targets suggest that sources within specific land covers contribute microbes to streams. However, further work using host-specific source tracking markers is needed to fully understand pathogen sources in streams. Statistical associations between microbes and physicalechemical water quality variables were observed in this study. While these results are consistent with such variables affecting microbial persistence or fate in streams, laboratory experiments are needed to establish causal
3288
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 2 7 9 e3 2 9 0
relationships. Finally, it is important to note that this study was carried out during two field campaigns conducted under similar, relatively dry climatic conditions. Additional work should investigate the effects of rainfall and storm events on pathogen loadings in these coastal streams. In particular, the effect of stream discharge rate on microbial occurrence should be investigated. Such work may be best undertaken using temporally intensive sampling at a limited number of sites. The present study established widespread distribution of bacterial pathogens in streams discharging to coastal waters, insinuating the potential for recreational waterborne illness upon exposure to stream runoff. In order to determine potential risks, a quantitative microbial risk assessment (QMRA) must be performed; this requires knowledge of pathogen concentration. Measurements of concentration were only possible with vibrio pathogens in the present study. Future work should quantify pathogen concentrations when feasible. Additionally, etiologies of recreational waterborne illness are believed to be predominantly viral (USEPA 2009a), so future work should assess human viruses concentrations in these streams. Finally, tropical streams are known to harbor Leptospira interrogans, the etiological agent of leptospirosis. Hawai’i has the highest rates of leptospirosis in the US (USEPA 2009b), thus future investigations into the distribution of this pathogen are of interest.
5.
Conclusions
1. Salmonella, Campylobacter (including C. jejuni), S. aureus, V. vulnificus, and V. parahaemolyticus were widespread in O’ahu streams that discharge to coastal waters, indicating land-based activities and thus runoff as a source of pathogens. 2. Of the five indicators tested, enterococci were significantly associated with the most bacterial pathogens in O’ahu coastal streams. 3. Positive associations between specific land covers and Salmonella (AG, FOR), enterococci (AG), and S. aureus (AG, URB) suggest sources within these land covers contribute microbes to streams. Further microbial source tracking using, for example, molecular host-specific markers is needed to pinpoint specific microbial sources. 4. Significant associations between microbial targets and physicalechemical stream water quality (i.e., temperature, nutrients, turbidity) suggest that organism persistence is affected by stream water quality. Further work should investigate causal mechanisms. 5. Campylobacter incidence and enterococci concentrations were greater in samples collected before sunrise suggesting these organisms are sensitive to sunlight.
Acknowledgments The authors are indebted to the field sampling team in Hawai’i, including Alissa Rogers, Patie Boehm, Ari Mesa, Mike Cooke, Mark Mench, and Shimi Rii. Two anonymous reviewers
improved the manuscript. This work was funded by NSF Oceans and Human Health: Pacific Research Center for Marine Biomedicine Grant No. OCE/0910491.
Appendix. Supplementary material The supplementary data associated with this article can be found in the on-line version at doi:10.1016/j.watres.2011.03.033.
references
AWWA, 2005. Standard Methods for the Examination of Water and Wastewater. American Public Health Association, American Water Works Association, Water Environment Federation, Baltimore. Boehm, A., Ashbolt, N.J., Colford Jr., J.M., Dunbar, L.E., Fleming, L. E., Gold, M., Hansel, J., Hunter, P.R., Ichida, A.M., McGee, C., Soller, J.A., Weisberg, S.B., 2009a. A sea change ahead for recreational water quality criteria. Journal of Water and Health 7 (1), 9e20. Boehm, A.B., Yamahara, K.M., Love, D.C., Peterson, B.M., McNeill, K., Nelson, K.L., 2009b. Covariation and photoinactivation of traditional and novel indicator organisms and human viruses at a sewage-impacted marine beach. Environmental Science and Technology 43 (21), 8046e8052. Buswell, C.M., Herlihy, Y.M., Lawrence, L.M., McGuiggan, J.T.M., Marsh, P.D., Keevil, C.W., Leach, S.A., 1998. Extended survival and persistence of Campylobacter spp. in water and aquatic biofilms and their detection by immunofluorescent-antibody and -rRNA staining. Applied and Environmental Microbiology 64 (2), 733e741. Cai, T., Jiang, L., Yang, C., Huang, K., 2006. Application of real-time PCR for quantitative detection of Vibrio parahaemolyticus from seafood in eastern China. FEMS Immunology and Medical Microbiology 46 (2), 180e186. Calderon, R., Mood, E., Dufour, A., 1991. Health effects of swimmers and nonpoint sources of contaminated water. International Journal of Environmental Health Research 1 (1), 21e31. Charoenca, N., Fujioka, R.S., 1993. Assessment of Staphylococcus bacteria in Hawaii’s marine recreational waters. Water Science and Technology 27 (3e4), 283e289. Charoenca, N., Fujioka, R.S., 1995. Association of staphylococcal skin infections and swimming. Water Science and Technology 31 (5e6), 11e17. DePaola, A., Hopkins, L.H., Peeler, J.T., Wentz, B., McPhearson, R. M., 1990. Incidence of Vibrio parahaemolyticus in U.S. coastal waters and oysters. Applied and Environmental Microbiology 56 (8), 2299e2302. Dwight, R.H., Baker, D.B., Semenza, J.C., Olson, B.H., 2004. Health effects associated with recreational coastal water use: urban versus rural California. American Journal of Public Health 94 (4), 565e567. Dziuban, E.J., Liang, J.L., Craun, G.F., Hill, V., Yu, P.A., Painter, J., Moore, M.R., Calderon, R.L., Roy, S.L., Beach, M., 2006. Surveillance for water disease and outbreaks associated with recreational water e United States, 2003e2004. MMWR Surveillance Summary 55 (SS12), 1e24. Evison, L.M., 1988. Comparative studies on the survival of indicator organisms and pathogens in fresh and sea water. Water Science and Technology 20 (11e12), 309e315.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 2 7 9 e3 2 9 0
Fujioka, R.S., Yoneyama, B.S., 2002. Sunlight inactivation of human enteric viruses and fecal bacteria. Water Science and Technology 46 (11e12), 291e295. Fung, D.Y.C., Fujioka, R.S., Vijayavel, K., Sato, D., Bishop, D., 2007. Evaluation of Fung double tube test for Clostridium perfringens and Easyphage test for F-specific RNA coliphages as rapid screening tests for fecal contamination in recreational waters of Hawaii. Journal of Rapid Methods and Automation in Microbiology 15 (3), 217e229. Goodwin, K.D., Pobuda, M., 2009. Performance of CHROMagar Staph aureus and CHROMagar MRSA for detection of Staphylococcus aureus in seawater and beach sand e comparison of culture agglutination, and molecular analyses. Water Research 43 (19), 4802e4811. Haile, R.W., Witte, J.S., Gold, M., Cressey, R., McGee, C., Millikan, R.C., Glasser, A., Harawa, N., Ervin, C., Harmon, P., Harper, J., Dermand, J., Alamillo, J., Barrett, K., Nides, M., Wang, G., 1999. The health effects of swimming in ocean water contaminated by storm drain runoff. Epidemiology 10 (4), 355e363. Haley, B.J., Cole, D.J., Lipp, E.K., 2009. Distribution, diversity, and seasonality of waterborne salmonellae in a rural watershed. Applied and Environmental Microbiology 75 (5), 1248e1255. Hardin, J.W., Hilbe, J.M., 2003. Generalized Estimating Equations. Chapman and Hall/CRC, New York. Hardina, C.M., Fujioka, R.S., 1991. Soil: the environmental source of Escherichia coli and Enterococci in Hawai’i’s streams. Environmental Toxicology and Water Quality 6 (2), 185e195. Harwood, V.J., Levine, A.D., Scott, T.M., Chivukula, V., Lukasik, J., Farrah, S.R., Rose, J.B., 2005. Validity of the indicator organism paradigm for pathogen reduction in reclaimed water and public health protection. Applied and Environmental Microbiology 71 (6), 3163e3170. Hawaii Department of Health, 2001. Hawaii Administrative Rules: Chapter 11-54-05.2. “Inland Water Criteria”. Hawaii, G.I.S, 2010. Hawaii Statewide GIS Program. Office of Planning- Department of Business, Economic Development & Tourism e State of Hawaii, Honolulu, HI. Hazen, T.C., 1988. Fecal coliforms as indicators in tropical waters: a review. Toxicity Assessment 3 (5), 461e477. HHIC, 2007. Hawaii Health Information Corporation Inpatient Database. Health Care Cost and Utilization project. http://hhic. org/mrsa.asp. Hsieh, J.L., Fries, J.S., Noble, R.T., 2008. Dynamics and predictive modelling of Vibrio spp. in the Neuse River Estuary, North Carolina, USA. Environmental Microbiology 10 (1), 57e64. Iwamoto, M., Ayers, T., Mahon, B.E., Swerdlow, D.L., 2010. Epidemiology of seafood-associated infections in the United States. Clinical Microbiology Reviews 23 (2), 399e411. Johnson, C.N., Flowers, A.R., Noriea III, N.F., Zimmerman, A.M., Bowers, J.C., DePaola, A., Grimes, D.J., 2010. Relationships between environmental factors and pathogenic Vibrios in the Northern Gulf of Mexico. Applied and Environmental Microbiology 76 (21), 7076e7084. Khan, I.H.P., Edge, T.A., 2007. Development of a novel triplex PCR assay for the detection and differentiation of thermophilic species of Campylobacter using 16S-23S rDNA internal transcribed spacer (ITS) region. Journal of Applied Microbiology 103 (6), 2561e2569. LaGier, M.J., Joseph, L.A., Passaretti, T.V., Musser, K.A., Cirino, N.M., 2004. A real-time multiplexed PCR assay for rapid detection and differentiation of Campylobacter jejuni and Campylobacter coli. Molecular and Cellular Probes 18 (4), 275e282. Leeworthy, V.R., Wiley, P.C., 2001. In: National Ocean Service, S.P. O. (Ed.), Marine recreation participation and use, national
3289
survey on recreation and the environment. National Oceanic and Atmospheric Administration, Silver Spring, MD. Luther, K., Fujioka, R.S., 2004. Usefulness of monitoring tropical streams for male-specific RNA coliphages. Journal of Water and Health 2 (3), 171e181. Malorny, B., Hoorfar, J., Bunge, C., Helmuth, R., 2003. Multicenter validation of the analytical accuracy of Salmonella PCR: towards an international standard. Applied and Environmental Microbiology 69 (1), 290e296. Mason, W.J., Blevins, J.S., Beenken, K., Wibowo, N., Ojha, N., Smeltzer, M.S., 2001. Multiplex PCR protocol for the diagnosis of staphylococcal infection. Journal of Clinical Microbiology 39 (9), 3332e3338. NOAA, 2001. Hawaii Land Cover 2001 Data Set. National Oceanographic and Atmospheric Administration Coastal Services Center, Charleston, SC. Panicker, G., Myers, M.L., Bej, A.K., 2004. Rapid detection of Vibrio vulnificus in shellfish and Gulf of Mexico water by real-time PCR. Applied and Environmental Microbiology 70 (1), 498e507. Shellenbarger, G.G., Athearn, N.D., Takekawa, J.Y., Boehm, A., 2008. Fecal indicator bacteria and Salmonella in ponds managed as bird habitat, San Francisco Bay, CA. Water Research 42 (12), 2921e2930. Shuval, H., 2003. Estimating the global burden of thalassogenic diseases: human infectious diseases caused by wastewater pollution of the marine environment. Journal of Water and Health 1 (2), 53e64. Sinton, L., Hall, C., Braithwaite, R., 2007. Sunlight inactivation of Campylobacter jejuni and Salmonella enterica, compared with Escherichia coli, in seawater and river water. Journal of Water and Health 5 (3), 357e365. USEPA, 1997. In: Arar, E.J., Collins, G.B. (Eds.), Method 445.0: in vitro determination of chlorophyll a and pheophytin a in marine and freshwater algae by fluorescence. U. S. Environmental Protection Agency, Office of Research and Development. National Exposure Research Laboratory, Cincinnati, OH. USEPA, 2001a. Method 1601: male-specific (Fþ) and somatic coliphage in water by two-step enrichment procedure. U. S. Environmental Protection Agency: Office of Water, Washington D.C. USEPA, 2002. Method 1604: Total Coliforms and Escherichia coli in Water by Membrane Filtration Using a Simultaneous Detection Technique (MI Medium). EPA 821-R-02-024. Environmental Protection Agency Office of Water, Washington D.C. USEPA, 2006. Method 1600: Enterococci in Water by Membrane Filtration Using Membrane-Enterococcus Indoxyl-beta-DGlucoside Agar (MEI). EPA 821-R-06-009. Environmental Protection Agency: Office of Water, Washington D.C. USEPA, 2009a. Review of Published Studies to Characterize Relative Risks from Different Sources of Fecal Contamination in Recreational Water. EPA 822-R-09-001. Office of Water, United States Environmental Protection Agency, Washington D.C. USEPA, 2009b. Review of Zoonotic Pathogens in Ambient Waters. EPA 822-R-09-002. Office of Water, United States Environmental Protection Agency, Washington D.C. Vereen Jr., E., Lowrance, R., Cole, D.J., Lipp, E.K., 2007. Distribution and ecology of campylobacters in coastal plain streams (Georgia, United States of America). Applied and Environmental Microbiology 73 (5), 1395e1403. Victoria, M., Guimaraes, F., Fumian, T., Ferreira, F., Vierira, C., Leite, J.P., Miagostovich, M., 2009. Evaluation of an adsorptionelution method for detection of astrovirus and norovirus in environmental waters. Journal of Virological Methods 156 (1e2), 73e79.
3290
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 2 7 9 e3 2 9 0
Wait, D.A., Sobsey, M.D., 2001. Comparative survival of enteric viruses and bacteria in Atlantic Ocean seawater. Water Science and Technology 43 (12), 139e142. Walters, S.P., Thebo, A.L., Boehm, A.B., 2011. Impact of urbanization and agriculture on the occurrence of bacterial pathogens and stx genes in coastal waterbodies of central California. Water Research 45 (4), 1752e1762. Wetz, J.J., Blackwood, A.D., Fries, J.S., Williams, Z.F., Noble, R.T., 2008. Trends in total Vibrio spp. and Vibrio vulnificus concentrations in the eutrophic Neuse River Estuary, North Carolina, during storm events. Aquatic Microbial Ecology 53 (1), 141e149.
Wilkes, G., Edge, T.A., Gannon, V., Jokinen, C., Lyautey, E., Medeiros, D., Neumann, N., Ruecker, N., Topp, E., Lapen, D.R., 2009. Seasonal relationships among indicator bacteria, pathogenic bacteria, Cryptosporidiium oocysts, Giardia cysts, and hydrological indices for surface waters within an agricultural landscape. Water Research 43 (8), 2209e2223. Yau, V., Wade, T.J., de Wilde, C.K., Colford Jr., J.M., 2009. Skinrelated symptoms following exposure to recreational water: a systematic review and meta-analysis. Water Quality, Exposure and Health 1 (2), 79e103.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 2 9 1 e3 2 9 9
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Selective sludge removal in a segregated aerobic granular biomass system as a strategy to control PAOeGAO competition at high temperatures M.-K.H. Winkler a, J.P. Bassin a,b, R. Kleerebezem a, L.M.M. de Bruin d, T.P.H. van den Brand a,c, M.C.M. van Loosdrecht a,c,* a
Department of Biotechnology, Technische Universiteit Delft, Julianalaan 67, 2628 BC Delft, The Netherlands Programa de Engenharia Quı´mica, COPPE, Universidade Federal do Rio de Janeiro, Brazil c KWR, Groningenhaven 7, 3433 PE, Nieuwegein, The Netherlands d DHV, Laan 1914 35, 3818 EX Amersfoort, The Netherlands b
article info
abstract
Article history:
An aerobic granular sludge (AGS) reactor was run for 280 days to study the competition
Received 23 September 2010
between Phosphate and Glycogen Accumulating Organisms (PAOs and GAOs) at high
Received in revised form
temperatures. Numerous researches have proven that in suspended sludge systems PAOs are
4 March 2011
outcompeted by GAOs at higher temperatures. In the following study a reactor was operated at
Accepted 14 March 2011
30 C in which the P-removal efficiency declined from 79% to 32% after 69 days of operation
Available online 21 March 2011
when biomass removal for sludge retention time (SRT) control was established by effluent withdrawal. In a second attempt at 24 C, efficiency of P-removal remained on average at
Keywords:
71 5% for 76 days. Samples taken from different depths of the sludge bed analysed using
UASB
Fluorescent in situ hybridization (FISH) microscopy techniques revealed a distinctive microbial
Aerobic granular sludge
community structure: bottom granules contained considerably more Accumulibacter (PAOs)
SRT
compared to top granules that were dominated by Competibacter (GAOs). In a third phase the
PAO
SRT was controlled by discharging biomass exclusively from the top of the sludge bed. The
GAO
application of this method increased the P-removal efficiency up to 100% for 88 days at 30 C.
Segregation
Granules selected near the bottom of the sludge bed increased in volume, density and overall ash content; resulting in significantly higher settling velocities. With the removal of exclusively bottom biomass in phase four, P-removal efficiency decreased to 36% within 3 weeks. This study shows that biomass segregation in aerobic granular sludge systems offers an extra possibility to influence microbial competition in order to obtain a desired population. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
Aerobic granular sludge (AGS) reactors are based on the same principle as upflow anaerobic sludge blanket (UASB) reactors in which particles are freely suspended in an upward flow of air and liquid. Contrary to flocculent sludge processes, the biomass
in these reactors is not homogeneously mixed. Mixing with gas yields a higher concentration of biomass at the bottom of the reactor than the top. Granules do not have identical physical characteristics and therefore there is a segregation of granules. Granules with larger radius or higher specific density will develop more rapidly settling characteristics, are therefore
* Corresponding author. Department of Biotechnology, Technische Universiteit Delft, Julianalaan 67, 2628 BC Delft, The Netherlands. E-mail address:
[email protected] (M.C.M. van Loosdrecht). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.03.024
3292
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 2 9 1 e3 2 9 9
often at the bottom of the sludge bed. This segregation is viewed as a disadvantage in the particulate biofilm reactors because they lead to instabilities (Ro and Neethling, 1994; Safferman and Bishop, 1996; Trinet et al., 1991). These instabilities are attributed to a lower shear stress in the top of the sludge bed due to lower density (Gjaltema et al., 1997). The first attempts to numerically-model segregation of biomass in dependency of particle density and diameter caused by outgrowth were made by DiFelice et al., 1997 for fluidized bed reactors. Selection properties can be applied to prevent uncontrolled outgrowth of biomass by using batch wise feeding in granular sludge reactors (Beun et al., 2001). This generates a microbial population with a lower growth rate and hence smoother granules, which makes shear less important for a selection of well settling particles (de Kreuk and van Loosdrecht, 2004; Van Loosdrecht and Heijnen, 1993). Nicolella and colleagues wrote a review concerning the strength of particle-based biofilm reactors and their potential to develop compact and high rate nutrient removal processes. Nevertheless, for this particular technology segregation of granules is acknowledged to be a difficult process due to, for instance, clogging. However, researchers have neglected that segregation of differently sized granules may in fact offer different biological niches for bacteria (Nicolella et al., 2000). In UASB reactors as well as in AGS reactors segregation of microbial communities can occur as a function of depth within the different layers of the granules (Macleod et al., 1990; Tsuneda et al., 2003; Xavier et al., 2007). However, it has been neglected that segregation might also occur over the settled sludge bed due to differently sized granules. Circumstances like shear stress or substrate concentrations are different at certain depths. As a result, distinct biological niches can be generated within one reactor, by which one organism can be favoured over others. For instance, it might be possible to influence the SRT of certain organisms independent of other bacteria depending on the place of excess sludge withdrawal. The importance of controlled biomass removal in biofilm systems has already been experimentally and mathematically discussed earlier (Morgenroth and Wilderer, 1999). Previous research has demonstrated that PAOs were prevalent at 10 C regardless of the specific carbon source or pH (de Kreuk et al., 2005; Lopez-Vazquez et al., 2009b). At temperatures between 20 and 30 C GAOs are expected to dominate the culture, while at increasing temperatures common heterotrophs dominate the system (Erdal et al., 2003; Lopez-Vazquez et al., 2009a; Panswad et al., 2003; Whang et al., 2007). In our research we hypothesise that if PAOs and GAOs would be differentially distributed over the sludge bed, then selective removal of the GAO dominated section of the settled sludge bed would reduce the SRT for GAOs relative to PAOs. This would in turn make it feasible to obtain good biological phosphate removal at temperatures above 20 C.
2.
Material and methods
2.1.
Cycle operation and measurements
The granular sludge reactor operation was similar to that described by de Kreuk and van Loosdrecht (de Kreuk and van Loosdrecht, 2004). It consisted of a 60 min anaerobic feeding
period from bottom of the reactor in a plug flow regime followed by a 111 min period of aeration, 3 min settling, 5 min effluent withdrawal and a 1 min idle period. In the aeration period the dissolved oxygen (DO) concentration and pH were controlled at 20% air saturation and 7 0.2 pH units, respectively. Temperature was held constant at 30 C with a thermocycler and was protected against cooling with a cellular isolation placed around the reactor. The feed medium consisted of 3.1 mM NaCH3COO$3H2O (400 mg COD/L), 0.2 mM MgSO4$7H2O, 0.2 mM KCl, 2.1 mM NH4Cl (60 mg N/L), 0.2 mM K2HPO4 and 0.1 mM KH2PO4 (20 mg P/L). A ‘Vishniac and Santer’ solution was used to provide trace elements (Vishniac and Santer, 1957). A cycle measurement was conducted and samples were taken during the aerated mixing period to measure system performance of P- and COD- removal during one cycle of operation. Phosphate was analysed spectrophotometrically by the use of standard test kits (Hach-Lange). Calculations for P-removal efficiency are based on influenteeffluent basis and 100% efficiency was hence reached when no P was detectable in effluent.
2.2.
Long term reactor operation
The reactor operation can be divided into 4 phases. Firstly, the reactor was run for 69 days at 30 C and was inoculated with granules from an aerobic granular sludge pilot plant in Epe, The Netherlands, treating municipal wastewater and showing excellent N- and P- removal efficiencies. In the second phase, half of the granules were removed and replaced with granular sludge from a lab reactor that was operating at 20 C with an excellent P-removal efficiency. Temperature was decreased to 24 C, to favour PAOs, and the reactor was operated from day 69 until day 150 under the set conditions. During phase one and two the SRT was controlled by the sludge washed out with the effluent withdrawal. In the third phase which lasted from day 150 until day 240, sludge was manually withdrawn from the upper part of the sludge bed and the temperature was raised to 30 C. In the final and fourth phase, sludge was removed from the bottom to provoke the washout of PAOs. In the final two phases, the SRT was controlled at approximately 21 days by removing every third day 15% (phase 3 and 4a) or every sixth day 30% (phase 4b) of the settled sludge bed. The SRT was calculated taking the volatile suspended solids (VSS) from the reactor, effluent, and excess sludge into consideration. A sample was taken and transferred into a small measuring cylinder for determination of the VSS from the reactor and excess sludge. The volume of settled biomass in the small cylinder was used to calculate the ash and dry weight of the biomass. The VSS of settled biomass within the cylinder was then related back to the volume occupied by the settled sludge bed in the reactor as well as to the volume taken out as excess sludge by recording the height of settled sludge in the reactor and the height of excess sludge removed. The VSS was calculated on a mass basis by the following equation VSSr;ex ¼ DWset: r;ex ashset: r;ex ½gVSS. The SRT was calculated on the basis of the change in height of the sludge bed occurred due to growth (VSSr), the amount of excess sludge removed for manual SRT control (Qex,VSS) and by the sludge washed out with effluent withdrawal (Qeff,VSS). The calculation was conducted according to the following equation
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 2 9 1 e3 2 9 9
SRT ¼ Vr VSSr =Q eff;VSS þ Q ex;VSS ½day. Please refer Appendix 1 for the definitions and side calculations.
to
2.3. Density and size distribution measurements of top and bottom granules Bottom and top granules were sampled for measurements of particle size distribution, dry weight, ash content and granule density. Specific biomass density was measured with a pycnometer and size distribution measurements were conducted by the means of an image-analyser. Stokes law for laminar flow was used to calculate settling velocities by applying the equation SV ¼ g=18$rp rw =rw $d2p =nw for Reparticle 1 and compared to measured settling velocities which were recorded as the time that granules settled in a 30 cm volumetric cylinder. Definitions for Stokes law are given in Appendix 2. In phase four, the removed bottom biomass was sampled on a weekly basis to measure a change in settling behaviour and physical properties caused by selective sludge removal over time.
2.4.
Fluorescent in situ hybridization (FISH)
Bottom and top samples were taken over time for FISH analysis in order to assess microbiological properties. FISH was performed on crushed mixed (Fig. 3.1a and 1b), top (Fig. 3.2ae4a) and bottom granules (Fig. 3.2be4b) in order to determine GAO and PAO microbial populations. Different probes were tested to ensure a good representation of PAO and GAO population and the resulting probes and sequences are listed in Table 1. Crushing was accomplished on 10 ml granules by the means of a glass mortar (Glas-Col). From this suspension 500 ml were fixed in 4% paraformaldehyde and incubated for 120 min at room temperature. After fixation, samples were centrifuged for 2 min at 16,000 rpm, washed twice in 1x Phosphate buffer saline (PBS), and re-suspended in volume of 1:1 Ethanol/PBS buffer for storage at 20 C. For hybridization, the fixed samples were dried on a hybridization slide with 6 wells preventing mixing of probe in adjacent wells and dehydrated by incubating the microscope slides in 50%, 80% and 100% ethanol for 3 min in each solution. After dehydration, the hybridization solution (10 ml) and 25 ng of oligonucleotide probe tagged with
Fig. 1 e Typical concentration patterns of phosphate (A), nitrate (:) nitrite (X) and ammonium (-) during a cycle in an aerobic granular sludge reactor. COD (Җ) is completely consumed in the anaerobic period. Note that measurement during anaerobic phase is not possible due to a strict anaerobic plug flow operation without mixing.
3293
a fluorescent label (Fluos, Cy5 or Cy3) was added to each well, and the samples were incubated for 2 h in a humid chamber at 46 C. The hybridization buffer consisted of a mixture of 360 ml of 5 M NaCl, 40 ml of 1 M Tris (pH 8), 10 ml of a 10% (w/v) sodium dodecylsulfate buffer (SDS), 700 ml of formamide, and 900 ml of MilliQ water (Amann et al., 1990; Crocetti et al., 2002, 2000; Daims et al., 1999). After hybridization, the microscope slides were washed at 48 C for 15 min by immersing them into 50 ml of washing solution consisting of 800 ml of 5 M NaCl, 500 ml of 0.5 M EDTA, 1000 ml of 1 M Tris (pH 8), and 50 ml of 10% SDS (w/v). The samples were dried and prepared with 2 ml antifade fluorescent mounting oil and analysed with an epifluorescence microscope (Axioplan 2, Zeiss). Ratios between PAOs and GAOs were roughly estimated based on visual determination.
3.
Results
3.1.
Cycle operation
A cycle measurement was conducted in phase three to show a typical reactor performance during one cycle of operation (Fig. 1). A classical graph for N- and P- removal behaviour in aerobic granular sludge based systems is depicted as described by de Kreuk et al. (2005); nitrification occurs on the outer layers and denitrification and phosphate uptake in the core of the granules. One cycle lasted 3 h starting with 60 min of anaerobic feeding during which all incoming ammonium, phosphate and acetate were fed from the bottom of the reactor in a plug flow regime. All acetate was taken up during the anaerobic feeding period and phosphate was released due to the activity of PAOs. Samples for the cycle measurement were collected only during the aerobic mixing period since sampling is not possible during feeding due to the plug flow regime, which should not be disturbed.
3.2.
Long term operation
The reactor was operated in four phases over a time period of 280 days. The reactor initially established a P-removal efficiency of 79% when the reactor was run in phase one at 30 C. However, after 69 days of operation P-removal efficiency dropped to 32% (Fig. 2A phase 1). In order to enhance P-removal efficiencies in phase two, half of the sludge was discharged and the reactor was inoculated with new granular sludge from a lab-scale AGS reactor run at 20 C showing 100% P-removal efficiencies. Following this inoculation, the operational temperature was decreased to 24 C (phase 2) and the P-removal efficiency remained on average at 71 5% for 76 days (Fig. 2A). During the first two phases, the SRT was defined by washout of sludge during effluent withdrawal. In the second phase, FISH results illustrated that sludge samples taken from the top of the sludge bed consisted of more GAOs (Fig. 3.2a) whereas the bottom sludge contained more PAOs (Fig. 3.2b). In order to selectively remove GAOs (Competibacter) from the system and hence keep their SRT lower in respect to PAOs (Accumulibacter) a third phase was initiated in which biomass was withdrawn from the upper part of sludge bed. The amount of biomass withdrawn was established as such that an average SRT of 25 15 days was
3294
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 2 9 1 e3 2 9 9
Fig. 2 e (A) Phosphate effluent (C) and influent (A) concentration in mM PO4eP as well as SRT (:) in days. (B) shows the P release/COD uptake ratio (-) and the P-removal efficiency (▬) over time. Experimental setup was divided in four phases. In phase one the reactor was run at 30 C when removal efficiency dropped to 30%, a second phase started and the system was inoculated with new granular sludge with excellent P-removal capacity, after which P-removal efficiency remained 50%. In phase three, granules were manually withdrawn from the upper part of the sludge bed resulting in 100% P-removal efficiency. In phase four, granules were discarded from the bottom of sludge bed resulting in a collapse in removal efficiency.
achieved according to the SRT calculation as given in the material and methods section. Furthermore, the temperature was raised to 30 C to disfavour PAOs and in order to measure the effect of sludge control on the PAOeGAO competition. Within 3 weeks the P-removal efficiency improved to up to 100% and remained on average at 92 7% for an additional 67 days (Fig. 2A phase 3). In the fourth phase sludge was removed from the bottom to determine whether higher P-removal efficiencies from phase 3 are indeed due to selective removal of top sludge or solely an effect of lower SRT. Selective removal of the bottom PAO dominated sludge eventually resulted in a decrease in P-removal efficiency to 36% within 6 weeks (Fig. 2A phase 4). Phase four consisted of two sub phases: in phase 4a sludge was removed in the same manner as it was accomplished for removal of top sludge which was
based on removing approximately 15% of the settled sludge bed every three days, achieving on average an SRT of 21 7 days. During this phase removal efficiency remained on average at 83 8%. In phase 4b the selective removal was changed to see the effect of removing a bigger PAO fraction. Here the SRT was kept constant, however, instead of removing 15% of the settled sludge bed every three days, 30% was removed every six days. During this time the P-removal efficiency dropped from 85% to 36% stressing the importance of proper sludge control. The P release/COD uptake ratio is depicted in Fig. 2B. In a highly enriched PAO culture a P release/COD uptake ratio of about 0.5 P-mol/C-mol can be expected in contrast to a pure GAO culture in which this ratio would decline to zero (Brdjanovic et al., 1997; Smolders et al., 1994). Ratios of about
Fig. 3 e Hybridization with Cy3-red (GAO-Competibacter), Cy5-blue (Eub) and Fluos-green (PAO-Accumulibacter)-labelled probes. Epifluorescence photomicrographs are shown for top and bottom biomass at different time points during reactor operation. Picture 3.1a FISH image of mixed biomass at the end of phase 1, Picture 3.1b population used for inocculum to start up phase 2, Picture 3.2 segregation of PAOs and GAOs at a) top and b) bottom of the reactor at the end of phase 2 as well as same relation shown for phase 3 when top sludge was removed and phase 4 when bottom sludge was removed. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
3296
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 2 9 1 e3 2 9 9
Table 1 e Oligonucleotide probes, target microorganisms, and references used in this study. Probe PAO 462 PAO 651 PAO 846 GAO Q989 GAO Q431 EUB 338 EUB 338 II EUB 338 III
Sequence (from 50 to 30 )
Specificity
Reference
CCGTCATCTACWCAGGGTATTAAC CCCTCTGCCAAACTCCAG GTTAGCTACGGACTAAAAGG TTCCCCGGATGTCAAGGC TCCCCGCCTAAAGGGCTT GCTGCCTCCCGTAGGAGT GCAGCCACCCGTAGGTGT GCTGCCACCCGTAGGTGT
Most Accumulibacter Most Accumulibacter Most Accumulibacter Some Competibacter Some Competibacter Most bacteria Planctomycetales Verrucomicrobiales
(Crocetti et al., 2000) (Crocetti et al., 2000) (Crocetti et al., 2000) (Crocetti et al., 2000) (Crocetti et al., 2002) (Amann et al., 1990) (Daims et al., 1999) (Daims et al., 1999)
Probes PAOs were tagged with the fluorescent dye Fluos (green) GAOS with Cy3 (red) and Eub with Cy5 (blue). For analysis probes of one target group were mixed.
0.15 P-mol/C-mol were observed in phase 1 and 2 in which removal efficiencies were low and GAOs dominated the system, as indicated by FISH (Fig. 3). However, in phase three when SRT was controlled by selectively removing top granules the ratio gradually increased in correlation with removal efficiencies to 0.34 P-mol/C-mol. In phase four the GAOs became prevalent again and the ratio decreased in correlation with the decreasing P-removal efficiencies.
3.3.
FISH analysis of sludge
During the experiment, samples of the sludge were regularly subjected to analysis by FISH. Virtually all cells in the sludge where stained by either Accumulibacter (PAO) or Competibacter (GAO) probes, indicating that these formed the large majority of the microbial population in the sludge. Fig. 3 gives an overview of the most relevant samples. Firstly, the reactor was run at 30 C in which a mixed sludge sample, taken at the end of phase one (no distinction between bottom/top), revealed a higher dominance of GAOs (Fig. 3.1a). At the same time the removal efficiency was also low (32%, day 66). For phase two, half of the reactor sludge was discarded and inoculated with new granular sludge containing mainly PAOs (Fig. 3.1b) to ensure an equal starting point for competition of PAOs and GAOs. P-removal efficiency increased instantaneously after inoculation but declined over time. At day 140 bottom and top sludge were checked separately for their microbial community composition. FISH analysis of bottom and top sludge was conducted because stratification of biomass was visually observed in both phases. During the aerobic mixing period biomass density was higher at the bottom. Following the settling period, large, heavy granules remained closer to the bottom whereas smaller granules were concentrated at the top portion of the expanded sludge bed. FISH results revealed that the top sludge contained considerably more GAOs (Fig. 3.2a) whereas the bottom sludge were enriched by PAOs (Fig. 3.2b), overall indicating a vertical segregation of microorganisms over the sludge bed. Based on these observations a third phase was initiated in which top sludge was removed to favour PAOs over GAOs. P-removal efficiencies increased during this phase to 100% and FISH results of top sludge illustrated an increase in the PAO populations (Fig. 3.3a) and a dominance of PAOs in the bottom sludge (Fig. 3.3b). A fourth phase was conducted in order to show that the segregation of community composition over the sludge bed was indeed an effect of sludge control from a specific height of
the settled bed. During this phase, sludge was removed from the bottom while keeping the same SRT. P-removal efficiency dropped to 36% and the bottom and top microbial populations were dominated by GAOs (Fig. 3.4a and 4b).
3.4. Density and size distribution measurements of top and bottom granules The physical properties of top and bottom granules during phase three have been evaluated and are given in Table 2. During phase three, the sludge age was manually controlled by sludge removal from the top of the sludge blanket. There was a clear difference in ash percentage between top and bottom sludge. The higher ash content was also reflected in a higher density of the bottom granular sludge. The measured density of 1018 13 g/l for bottom granules versus 1004 4 g/l for top granules contributed to the higher settling rates of bottom granules. The diameter derived from the average surface area of bottom granules was also larger. The obtained physical parameters were used to estimate the settling velocity applying Stokes law as given in the material and methods section. There was an estimated 3e4 factor difference in settling velocities between top and bottom granules. Furthermore, calculations revealed that differences in settling velocities were influenced equally due to both changes in the radius and the density of the granules. The estimated velocities were similar to the measured velocities. Additionally, the settling properties of removed bottom sludge were measured during phase 4 (see Fig. 4). The settling rate of the bottom sludge decreased rapidly and continuously over time. Results revealed that in phase 4a, settling velocities of bottom sludge decreased from 80 m/h to 50 m/h when 15% of the settled
Table 2 e Physical properties of bottom and top granules during sludge control of top biomass (Phase 2). Parameter Settling velocity calculated m/h Settling velocity measured m/h Ash content % Density g/l Average diameter (mm)
Top 20 5 nm 15 0.1 1004 4 0.8 0.1
Bottom 80 66 34 1018 1.1
9 9 0.1 13 0.2
Measurements were conducted during phase 3 in which SRT was controlled by discarding top biomass. nm: not measured.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 2 9 1 e3 2 9 9
Fig. 4 e P-removal efficiency (A) and settling velocity in m/ h (-) of bottom granules in phase three (top sludge removal) and in phase four (bottom sludge removal). Settling velocities were calculated by Stokes law based on data obtained from density and size distribution measurements.
sludge bed was removed every third day. However, 30% of the settled sludge bed was removed every six days in phase 4b. As a consequence, P-removal efficiency dropped to 36% while settling velocities of bottom sludge declined to 12 m/h.
4.
Discussion
In granular sludge systems segregation of biomass can easily occur due to slight variations in density and diameter of the particles. The opportunity to select for specific microbial groups at different heights within the column, allows for imposing additional selective pressure in granular sludge systems compared to traditional activated sludge systems. This is delineated by the ability of the granular sludge to maintain PAOs as the dominant group at 30 C by selectively removing top biomass dominated by GAOs and thus keeping their SRT low. Studies which were carried out at higher temperatures have shown that a very short cycle length or a low sludge age (3 days) can also lead to a stable P-removal efficiency at 30 C (Freitas et al., 2009; Whang and Park, 2006). Without sludge control, flocculent sludge systems have been shown to favour the enrichment of GAOs over PAOs (Lopez-Vazquez et al., 2009a). The results obtained in these experiments revealed that granules dominated by PAOs (Accumulibacter) were bigger, more dense and thus have the advantage to remain at the bottom of the reactor due to faster settling velocities. Since the reactor is fed in a plug flow regime from the bottom, it is evident that bottom granules have more substrate available, leading to a niche where PAOs were exposed to a higher percentage of the available substrate as compared to GAOs. Since sludge withdrawal is accomplished from the top of the sludge bed the SRT of the GAO population is effectively lower than for PAO dominated granules leading to a washout of GAOs over time. The main heterotrophic microbial population consisted of PAOs (Accumulibacter) or GAOs (Competibacter) since all acetate was always taken up before the aerobic period started leaving no more organic carbon available for normal heterotrophs. Moreover, the P release/COD uptake ratio positively correlated with
3297
removal efficiencies and data derived from FISH analysis. For example, when P-removal efficiencies and PAO content were high within the reactor, the P release/COD uptake ratios were additionally high. P-removal deteriorated in phase four when the required SRT was established by extracting bottom biomass instead of top biomass as performed in phase 3, demonstrating that the P-removal efficiencies were improved due to selective sludge removal and not only due to the lower SRT values. The difference in P-removal between phase 4a and b demonstrate that selective sludge control is strongly dependent on the amount of biomass extracted. The high values for the P release/COD uptake ratios in phase 4a suggest that although P-removal dropped, bottom biomass removal also stimulates new growth of PAOs. In phase 4b, when a larger proportion of bottom granules was removed, P-removal declined and at the same time also the settling velocities of bottom granules (due to a decrease in diameter and density), which minimizes segregation and herewith the advantage of bigger PAO dominated granules to always settle first to the bottom. Again, this is of advantage because bottom granules have more substrate available due to the plug flow feeding regime. This highlights the importance in obtaining knowledge concerning mechanisms leading to segregation of biomass, particularly how selective pressure of certain organisms over others can be influenced by removal of sludge from a specific depth within the reactor. In addition it is also significant to gain more knowledge about how selective sludge removal from the top or bottom of the sludge bed is influencing the SRT distribution of different granules in the reactor. For the renewal of bottom biomass and hence the growth of PAOs, sludge removal of bottom biomass could facilitate in avoiding the deterioration in P-removal over time. In order to better understand the effect of sludge removal it would be necessary in future studies to determine the P content of removed sludge and to make a proper P-mass balance. An explanation of how segregation occurs is that in PAO dominated granules the PO4 released per unit of acetate removed is higher than in GAO dominated granules, which is attributed to a lack of an active P-uptake/release metabolism in GAOs. Since the settling of biomass occurs after the aeration period PAO dominated granules have accumulated high amounts of poly-P, which will improve their settling properties in comparison to GAO dominated granules. The higher ash content of bottom PAO dominated granules might hence be due to higher poly-P content. Since chemical precipitation is strongly dependent on PO4 concentrations (Carlsson et al., 1997; Maurer et al., 1999) and only PAOs excrete phosphate, chemical precipitation in a PAO dominated granule might enhance this effect. This way the system has a self-enhancing effect on P-removal efficiencies since PAOs can remain in the system despite the high temperatures, which are known to be favourable for GAOs (Lopez-Vazquez et al., 2009b). Selective sludge removal at different heights in a granular sludge bed might offer a good opportunity to conduct microbial population engineering in AGS and UASB bed technology. Recent research has demonstrated the existence of segregation in other systems. For example, Volcke et al. (2010), illustrated that in a granular sludge nitritation/anammox system nitrite oxidising bacteria accumulate preferentially in smaller
3298
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 2 9 1 e3 2 9 9
granules, again allowing a method for controlling nitrite oxidising bacteria by selective sludge withdrawal. Additionally, selective sludge removal has been shown to enhance granulation processes (Li and Li, 2009). It is thus of interest to evaluate the activity of microorganisms in relation to the depth within the sludge bed and to design a sludge extraction protocol based on the specific population one wants to select for.
5.
Conclusions
In this work it was investigated whether segregation of biomass occurs along the sludge bed. PAOs were prevalent at the bottom, whereas GAOs dominated at the top of the sludge bed. By selective removal of GAO dominated sludge from the top of the sludge bed 100% P-removal efficiencies were achieved at 30 C. This study also shows that selective sludge withdrawal in granular sludge reactors can be used as an extra operational parameter to engineer the microbial population in the reactor.
Acknowledgements This study is partly funded by DHV and STOWA within the framework of the Dutch national Nereda research programme.
Appendix 1 Equations for calculating VSS of reactor and excess sludge as well as sludge retention time (SRT) during sludge control. SRT ¼
Vr VSSr ½day Q eff;VSS þ Q ex;VSS
Vset: r;ex ¼ pr2 hset: r;ex m3 DWset: ¼
DW Vset: r;ex ½g dry weight Vset: cyl
ashset: ¼
ash Vset: r;ex ½g ash Vset: cl
VSSr;ex ¼ DWset: r;ex ashset: r;ex ½gVSS Vr ¼ volume reactor [m3] VSSr,ex ¼ volatile suspended solids reactor or excess [gVSS] Qeff,VSS ¼ outflow VSS effluent [gVSSeff m3/day] Qex,VSS ¼ outflow VSS of bottom or top sludge control [gVSSex m3/day] hset. r,ex ¼ bed height settled sludge reactor and removed for excess [m] Vset. r,ex ¼ volume settled sludge bed in reactor or excess sludge [m3] Vset. cyl ¼ volume settled sludge bed in measuring cylinder [m3] SRT ¼ solid retention time [days]
Appendix 2 Equations for calculation of settling velocity by Stokes law 2
SV ¼
g rp rw dp $ for Reparticle 1 $ rw nw 18
SV ¼ sedimentation velocity of a single particle [m/s] dp ¼ particle diameter [m] rp ¼ density of particle [kg/m3] rw ¼ density of the fluid [kg/m3] g ¼ gravitational constant 9,81 [m/s2] nw ¼ kinematic viscosity water [m2/s] Re ¼ Reynolds number of a particle [e]
references
Amann, R.I., Krumholz, L., Stahl, D.A., 1990. Fluorescentoligonucleotide probing of whole cells for determinative, phylogenetic, and environmental-studies in microbiology. Journal of Bacteriology 172, 762e770. Beun, J.J., Heijnen, J.J., Van Loosdrecht, M.C.M., 2001. N-removal in a granular sludge sequencing batch airlift reactor. Biotechnology and Bioengineering 75, 82e92. Brdjanovic, D., van Loosdrecht, M.C.M., Hooijmans, C.M., Alaerts, G.J., Heijnen, J.J., 1997. Temperature effects on physiology of biological phosphorus removal. Journal of Environmental Engineering (New York) 123, 144e153. Carlsson, H., Aspegren, H., Lee, N., Hilmer, A., 1997. Calcium phosphate precipitation in biological phosphorus removal systems. Water Research 31, 1047e1055. Crocetti, G.R., Banfield, J.F., Keller, J., Bond, P.L., Blackall, L.L., 2002. Glycogen-accumulating organisms in laboratory-scale and full-scale wastewater treatment processes. Microbiology (Reading, United Kingdom) 148, 3353e3364. Crocetti, G.R., Hugenholtz, P., Bond, P.L., Schuler, A., Keller, J., Jenkins, D., Blackall, L.L., 2000. Identification of polyphosphate-accumulating organisms and design of 16S rRNAedirected probes for their detection and quantitation. Applied and Environmental Microbiology 66, 1175e1182. Daims, H., Bruhl, A., Amann, R., Schleifer, K.H., Wagner, M., 1999. The domain-specific probe EUB338 is insufficient for the detection of all bacteria: development and evaluation of a more comprehensive probe set. Systematic and Applied Microbiology 22, 434e444. 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, 761e769. de Kreuk, M.K., van Loosdrecht, M.C.M., 2004. Selection of slow growing organisms as a means for improving aerobic granular sludge stability. Water Science and Technology 49, 9e17. DiFelice, R., Nicolella, C., Rovatti, M., 1997. Mixing and segregation in water fluidised-bed bioreactors. Water Research 31, 2392e2396. Erdal, U.G., Erdal, Z.K., Randall, C.W., 2003. The competition between PAOs (phosphorus accumulating organisms) and GAOs (glycogen accumulating organisms) in EBPR (enhanced biological phosphorus removal) systems at different temperatures and the effects on system performance. Water Science and Technology 47, 1e8. Freitas, F., Temudo, M.F., Carvalho, G., Oehmen, A., Reis, M.A.M., 2009. Robustness of sludge enriched with short SBR cycles for biological nutrient removal. Bioresource Technology 100, 1969e1976.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 2 9 1 e3 2 9 9
Gjaltema, A., Vinke, J.L., vanLoosdrecht, M.C.M., Heijnen, J.J., 1997. Abrasion of suspended biofilm pellets in airlift reactors: importance of shape, structure, and particle concentrations. Biotechnology and Bioengineering 53, 88e99. Li, A.J., Li, X.Y., 2009. Selective sludge discharge as the determining factor in SBR aerobic granulation: numerical modelling and experimental verification. Water Research 43, 3387e3396. Lopez-Vazquez, C.M., Hooijmans, C.M., Brdjanovic, D., Gijzen, H.J., van Loosdrecht, M.C.M., 2009a. Temperature effects on glycogen accumulating organisms. Water Research 43, 2852e2864. Lopez-Vazquez, C.M., Oehmen, A., Hooijmans, C.M., Brdjanovic, D., Gijzen, H.J., Yuan, Z., van Loosdrecht, M.C.M., 2009b. Modeling the PAOeGAO competition: effects of carbon source, pH and temperature. Water Research 43, 450e462. Macleod, F.A., Guiot, S.R., Costerton, J.W., 1990. Layered structure of bacterial aggregates produced in an upflow anaerobic sludge bed and filter reactor. Applied and Environmental Microbiology 56, 1598e1607. Maurer, M., Abramovich, D., Siegrist, H., Gujer, W., 1999. Kinetics of biologically induced phosphorus precipitation in wastewater treatment. Water Research 33, 484e493. Morgenroth, E., Wilderer, P.A., 1999. Controlled biomass removal e the key parameter to achieve enhanced biological phosphorus removal in biofilm systems. Water Science and Technology 39, 33e40. Nicolella, C., van Loosdrecht, M.C.M., Heijnen, S.J., 2000. Particlebased biofilm reactor technology. Trends in Biotechnology 18, 312e320. Panswad, T., Doungchai, A., Anotai, J., 2003. Temperature effect on microbial community of enhanced biological phosphorus removal system. Water Research 37, 409e415. Ro, K.S., Neethling, J.B., 1994. Biological fluidized-beds containing widely different bioparticles. Journal of Environmental Engineering-Asce 120, 1416e1426.
3299
Safferman, S.I., Bishop, P.L., 1996. Aerobic fluidized bed reactor with internal media cleaning. Journal of Environmental Engineering-Asce 122, 284e291. Smolders, G.J.F., Vandermeij, J., Vanloosdrecht, M.C.M., Heijnen, J. J., 1994. Model of the anaerobic metabolism of the biological phosphorus removal process e stoichiometry and Ph influence. Biotechnology and Bioengineering 43, 461e470. Trinet, F., Heim, R., Amar, D., Chang, H.T., Rittmann, B.E., 1991. Study of biofilm and fluidisation of bioparticles in a threephase liquid fluidized bed reactor. Water Science and Technology 23, 1347e1354. Tsuneda, S., Nagano, T., Hoshino, T., Ejiri, Y., Noda, N., Hirata, A., 2003. Characterization of nitrifying granules produced in an aerobic upflow fluidized bed reactor. Water Research 37, 4965e4973. Van Loosdrecht, M.C.M., Heijnen, S.J., 1993. Biofilm bioreactors for waste-water treatment. Trends in Biotechnology 11, 117e121. Vishniac, W., Santer, M., 1957. Thiobacilli. Bacteriological Reviews 21, 195e213. Volcke, E.I.P., Picioreanu, C., De Baets, B., van Loosdrecht, M.C.M., 2010. How to deal with particle size distribution in granular sludge reactor models? In: Proceedings IWA/WEF Wastewater Treatment Modelling Seminar, WWTmod 2010, Que´bec, Canada, pp. 28e30. Whang, L.M., Filipe, C.D.M., Park, J.K., 2007. Model-based evaluation of competition between polyphosphate- and glycogen-accumulating organisms. Water Research 41, 1312e1324. Whang, L.M., Park, J.K., 2006. Competition between polyphosphateand glycogen-accumulating organisms in enhanced-biologicalphosphorus-removal systems: effect of temperature and sludge age. Water Environment Research 78, 4e11. Xavier, J.B., De Kreuk, M.K., Picioreanu, C., Van Loosdrecht, M.C. M., 2007. Multi-scale individual-based model of microbial and bioconversion dynamics in aerobic granular sludge. Environmental Science & Technology 41, 6410e6417.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 0 0 e3 3 0 8
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Effects of primary sludge particulate (PSP) entrapment on ultrasonic (20 kHz) disinfection of Escherichia coli Xiaofei Cui a, Jeffrey W. Talley a,b,*, Guojing Liu c, Steve L. Larson d a
Department of Civil and Environmental Engineering, Southern Methodist University, Suite 203, 3101 Dyer Street, Dallas, TX 75205, USA Environmental Technology Solutions, 75 W Baseline Road, Suite 32, Gilbert, AZ 85233, USA c 5710 Winterhaven Dr., Newark, DE 19702, USA d US Army Corps of Engineers, Engineer Research and Development Center, Environmental Laboratory, 3909 Halls Ferry Road, Vicksburg, MS 39180-6199, USA b
article info
abstract
Article history:
The role of primary sludge particulates (PSPs) in ultrasonic disinfection of Escherichia coli
Received 13 October 2010
(E. coli) was investigated. Entrapment of E. coli by PSP was directly observed through
Received in revised form
scanning electron microscope (SEM) after E. coli and PSP were incubated together in water
16 March 2011
for 24 h at 35 C. Entrapment coefficient was proposed for the first time to reflect the ability
Accepted 20 March 2011
of PSP to entrap E. coli and was estimated as 1.4 103 CFU/mg PSP under our experimental
Available online 31 March 2011
conditions. Ultrasonication (20 kHz) of different E. coli-PSPs solutions showed that the entrapped E. coli cells were protected by PSP from ultrasonication and the unentrapped
Keywords:
cells were not. However, the protection of entrapped E. coli cells gradually decreased as
Primary sludge particulates
ultrasonication proceeded, suggesting the ability of power ultrasonication to deprotect the
Ultrasonic disinfection
entrapped E. coli cells. SEM studies suggested a two-step mechanism for ultrasonic (20 kHz)
Entrapment of Escherichia coli
disinfection of entrapped E. coli: breakdown of the protective PSP refugia and disinfection of
Scanning electron microscopy
the exposed E. coli cells. This research will enable more informed decisions about disinfection of aqueous samples where porous PSP are present. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
Coming from household and toilet wastes, primary sludge (PS) is regularly removed from the bottom of primary sedimentation tanks (Cheng et al., 2009; Ji et al., 2010). Processed PS has been more and more applied to cropland, strip mines, public parks, and other areas in the United States and elsewhere (Gattie and Lewis, 2004). Fecal coliform density in PS is required by Class A sludge regulations on control of pathogens to be less than 1000 most probable number (MPN)/g PS (U.S. EPA, 1993). However, primary sludge discharged from most wastewater treatment plants does not meet this requirement (Han and Dague, 1997), and even commercially processed PS
has also been criticized for the illness caused by exposure to the pathogens in it (Gattie and Lewis, 2004). This situation requires effective disinfection of PS. Various organic substances originating from feces, vegetables, fruits, textiles, paper, etc. are present in PS and majority of them are particulates (Ji et al., 2010; Ubukata, 2007). These organic primary sludge particulates (PSPs) increase the difficulty of PS disinfection. Bacteria can be retained by suspended solids through non-specific Lifshitzevan der Waals forces and hydrogen and chemical bonding (Bayoudh et al., 2009). Pathogenic bacteria may also be entrapped by flocs of fine particles and therefore protected from biocides (Joyce et al., 2003; Narkis et al., 1995). Therefore,
* Corresponding author. Department of Civil and Environmental Engineering, Southern Methodist University, Suite 203, 3101 Dyer Street, Dallas, TX 75205, USA. Tel.: þ1 214 668 7925; fax: þ1 214 768 2164. E-mail address:
[email protected] (J.W. Talley). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.03.034
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 0 0 e3 3 0 8
a technique that can deprotect and disinfect pathogenic bacteria is desirable. Ultrasonication (20 kHze10 MHz) has been proposed to break apart agglomerates of suspended solids through mechanical effects (Blume and Neis, 2004; Tiehm et al., 1997). As ultrasound waves pass through liquid media, oscillating acoustic pressures create cavities. These cavities grow in size and ultimately implode (Fu et al., 2007). This process is instantaneous and thereby adiabatic, resulting in extremely high temperatures and pressures (5000 C and 1000 atm) in the cavities upon collapse (Findik et al., 2006). The collapse of cavities becomes milder as the ultrasonic frequency increases (Yasui, 2002). Therefore 20 kHz is the most used and ideal frequency to achieve the maximum mechanical effects including shock waves, liquid jets, and acoustic streaming (Dahlem et al., 1998; He et al., 2007). Mechanical effects have multiple functions to disintegrate PSP, including particleeparticle collision by shock wave, solid surface bombardment by liquid jet, etc (He et al., 2007). Ultrasonication can also disinfect pathogenic bacteria directly as the membrane of bacterial cells can be ruptured by shear forces resulting from bubble collapse (Kubo et al., 2005). PSP is concentrated during primary settlement and this makes the entrapment and protection of pathogenic bacteria more significant. Therefore, it is reasonable to deprotect and disinfect the entrapped pathogenic bacteria through ultrasonication before the primary settlement in the wastewater treatment plants. Ultrasonication has been extensively studied in the past concerning its effect on anaerobic digestion of sewage sludge (Tiehm et al., 1997; Xie et al., 2007), the reduction of secondary sludge (Foladori et al., 2007), the treatment of organic pollutants (Ning et al., 2005), and disinfection of bacteria directly (Drakopoulou et al., 2009; Madge and Jensen, 2002; Phull et al., 1997). There is an also significant amount of published work discussing the negative effects of suspended solids on disinfection through chlorination, ozone, and UV radiation (Boorman et al., 1999; Wang et al., 2006). However, little is known of the spatial relationship between Escherichia coli (E. coli) cells and PSP or the role PSP play in ultrasonic disinfection of E. coli, one of the most used indicator microorganisms of fecal contamination. The goals of this research were three-fold: (1) to characterize the entrapment of E. coli cells by PSP and to establish its correlation with PSP concentration; (2) to investigate the effects of PSP on ultrasonic disinfection of entrapped and unentrapped E. coli cells; (3) to clarify the mechanism for ultrasonic disinfection of entrapped E. coli cells.
2.
Experimental section
2.1.
Apparatus and materials
An ultrasonication system was used to generate ultrasound at 20 kHz, consisting of a MISONIX Sonicator S-4000 (QSONICA, LLC.), a titanium tip horn of 12.7 mm in diameter (QSONICA, LLC.), and a 500-ml jacketed beaker (Kimble Kontes LLC.). The horn was submerged 24 mm into the aqueous sample. Sample temperatures were kept between 20 and 27 C during ultrasonication at 20 kHz by circulating ice water through the beaker jacket. A Clinical Centrifuge (IEC) was used to prepare
3301
E. coli stock solution. An ultrasonics cleaner (VWR, B5500AMT, 42 kHz) was used to release the entrapped E. coli cells from PSP. An UV lamp (UVP, UVGL-15, 4 W, 254/365 nm) was used to kill the E. coli cells that were not entrapped by PSP. An Energy Dispersive X-ray Spectroscopy (using EDAX Genesis) coupled with LEO 1450 VPSE (variable pressure mode and 25 Pa of chamber pressure) was used to analyze the elemental composition of dewatered PSP. The probe current was 493 pA, high voltage was 15 kV, and magnification was 100. An XL 30 scanning electron microscope (Philips) equipped with a LaB6 source was employed to characterize the entrapment of E. coli cells by PSP and to study the mechanisms for ultrasonic (20 kHz) disinfection of entrapped E. coli cells. A description of materials used in this study is provided in the Supplementary material.
2.2.
Specific energy of ultrasonication
The parameter of specific energy (Es, kJ/L) was used in this work to reflect energy consumption during ultrasonication. It is calculated according to Equation (1) (Foladori et al., 2007): Es ¼ 103
Pdiss t v
(1)
Where Pdiss is the dissipated ultrasonic power into the samples (W) as determined by calorimetry (Mason et al., 1992), t is the time of ultrasonication (s), V is the volume of sample (L).
2.3.
Experimental procedures
The basic experimental procedures are summarized below in Fig. 1.
2.3.1. Preparation of E. coli stock solution and uniform dewatered PSP E. coli O157: H7 (ATCC 700728) was used in this study. Preparation of E. coli stock solution is described in the Supplementary material. The prepared stock solution had an E. coli concentration around 1.2 107 CFU/ml and was stored at 4 C. In this paper, Coliscan Easygel was used to detect and measure E. coli O157: H7. The Coliscan Easygel test kit consists of a bottle of Easygel (an agar replacement) and a pretreated petri dish. Special chromogenic materials are absorbed at the bottom of petri dishes, and can react with the enzyme of galactosidase to produce contrasting color. General coliform will produce galactosidase in lactose fermentation and therefore be detected by Coliscan Easygel (Micrology Laboratories, 2009). Since Coliscan Easygel is a culture-based technique, the E. coli cells that were disinfected by ultrasonication may become non-cultivable, and thus may not be detected by Coliscan Easygel. However, these non-cultivable damaged E. coli cells are not necessarily dead. Dewatered PSP were prepared by lyophilization of primary sludge collected from Central Wastewater Treatment Plant, Dallas, Texas. During lyophilization (also called freeze-drying), deterioration of the samples can be stopped and the primary structure can be protected (Ratti, 1999). The dewatered PSP was homogenized by hand. No E. coli was detected in PSP slurry by Coliscan Easygel before and after the PSP slurry was treated ultrasonically at 42 kHz based on our preliminary test.
3302
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 0 0 e3 3 0 8
Fig. 1 e Schematic diagram for the basic experimental procedures.
Inactivation of cells is possible during the freeze-drying process (Greaves, 1960). Energy Dispersive X-ray Spectroscopy analysis showed that the dewatered PSP contained 34.8% 1.48% of carbon, 10.8% 0.888% of nitrogen, 42.9% 1.42% of oxygen, 0.0795% 0.0705% of aluminum, 1.65% 0.133% of silicon, 1.51% 0.0742% of phosphorous, 1.17% 0.174% of sulfur, 3.03% 0.257% of calcium, and 3.26% 0.324% of iron.
2.3.2.
Entrapment of E. coli cells by PSP
In this paper, the E. coli cells entrapped by PSP refer to the E. coli cells that were associated with PSP, and made undetectable to Coliscan Easygel due to such entrapment. Dewatered PSP of 0.80 g was made into 50 ml slurry to which 9.0 ul of E. coli stock solution was added. The mixture was incubated by stirring at 100 RPM for 24 h at 35 C. Samples of 2 ml were collected before and after incubation, and were examined under SEM. Sample preparation for SEM study is described in the Supplementary material. Concentration of PSP ranges from 0.068 to 1.3 mg/ml in restaurant wastewater (Chen et al., 2000). In typical domestic sewage, the estimated PSP concentration is in the range from 0.1 to 0.35 mg/ml (McGhee, 1991). PSP concentration from 0.53 to 0.86 mg/ml was also reported in the sewage collected in Spain (Carballa et al., 2004). The samples studied in this work had PSP concentrations ranging from 0.17 to 1.3 mg/ml. Four E. coli-PSP solutions were prepared to study the correlation between the concentration of entrapped E. coli and that of PSP: dewatered PSP of 0.10, 0.33, 0.54, and 0.80 g were made into four 50 ml slurry samples and 9.0 ul of E. coli stock solution was added to each of them. The new E. coli-PSP
matrices were then incubated by stirring at 100 RPM for 24 h at 35 C. After incubation, the slurry samples were diluted into 600 ml solutions with distilled water. The resulting PSP dosages in the four samples were 17, 0.55, 0.90, and 1.3 mg/ml. They were radiated by UV light at wavelengths of 254 and 365 nm for 20 min to disinfect the unentrapped E. coli cells. Unlikely to penetrate into PSP (Qualls et al., 1983), UV radiation has no disinfection effects on the E. coli cells entrapped by the PSP. A volume of 150 ml from each of the 600 ml UV-radiated samples was treated in ultrasonics cleaner at 42 kHz for 14 min (Es ¼ 17 kJ/L) to release all the E. coli cells made undetectable to culture-based methods due to PSP entrapment. The correlation between the concentration of entrapped E. coli and that of PSP was established according to Equation (2) and the concentrations of entrapped E. coli concentrations (CE) used during correlation establishment were obtained according to Equation (3), CE ¼ B Css
(2)
where B is PSP entrapment coefficient (CFU/mg PSP), Css is PSP concentration (mg PSP/ml), and CE is concentration (CFU/ml) of E. coli cells made undetectable due to PSP entrapment before ultrasonication at 42 kHz. CE ¼ CU C0
(3)
where CE is the E. coli concentration (CFU/ml) measured by Coliscan Easygel after ultrasonication for 14 min (Es ¼ 17 kJ/L) at 42 kHz, and C0 is E. coli concentration (CFU/ml) measured by Coliscan Easygel before the ultrasonication at 42 kHz. In Coliscan Easygel measurement of E. coli, special chromogenic substrates pre-coated on the bottom of Petri plate
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 0 0 e3 3 0 8
interact with galactosidase, the enzyme specific to coliform, to produce pigment of contrasting color (Micrology Laboratories, 2009). Without contact with the chromogenic substrates due to PSP entrapment, the entrapped E. coli cells are undetectable by Coliscan Easygel. Therefore, CE can be used to represent the entrapped undetectable E. coli cells. Ultrasonics cleaner (42 kHz) was used to release the entrapped E. coli cells. An ultrasonic apparatus operating around 40 kHz was previously used for non-destructive, rapid, and reproducible removal of biofilm from standard materials (Oulahal-Lagsir et al., 2000). A preliminary test was conducted in our lab to evaluate the disinfection of ultrasonication at 42 kHz for 14 min (Pdiss ¼ 3 W and Es ¼ 17 kJ/L). E. coli stock was added to 150 ml DI water. The mixture was treated in ultrasonics cleaner (42 kHz) for 14 min (Es ¼ 17 kJ/L). E. coli concentrations were measured through Coliscan Easygel in triplicate before and after the treatment. After compared with the control sample, it was found that 14 min (Es ¼ 17 kJ/L) of ultrasonic treatment at 42 kHz did not change the E. coli concentration significantly, indicating that 14 min (Es ¼ 17 kJ/ L) of pretreatment in the ultrasonics cleaner (42 kHz) under our experimental conditions has no significant effect on E. coli disinfection. Another preliminary test was conducted to evaluate whether 14 min (17 kJ/L) of ultrasonication at 42 kHz was sufficient to release all entrapped E. coli. PSP of 0.8 g was made into 50 ml slurry and incubated by stirring at 100 RPM for 24 h at 35 C. After incubation the slurry was diluted into 600 ml solution with distilled water and was radiated by UV light at wavelengths of 254 nm and 365 nm for 20 min to disinfect the unentrapped E. coli cells. Then, 300 ml of the E. coli-PSP sample were treated in the ultrasonics cleaner for 28 min. E. coli concentrations were measured through Coliscan Easygel in triplicate initially and after 7, 14, 21, and 28 min of treatment in the ultrasonics cleaner. The results showed after 14 min of treatment, the E. coli concentration reached the maximum, indicating that 14 min (Es ¼ 17 kJ/L) of pretreatment in the ultrasonics cleaner (42 kHz) can release the maximum entrapped E. coli cells under our experimental conditions. Therefore, CE can be used to represent the maximum entrapped undetectable E. coli cells. A preliminary test was also conducted to evaluate whether E. coli may be entrapped by bacterial flocs. An incubated E. coliPSP solution was settled and the supernatant was analyzed. No increase of E. coli concentration (measured by Coliscan Easygel) was found after the supernatant was ultrasonicated (42 kHz) for 14 min (Es ¼ 17 kJ/L). This fact excludes the possibility of E. coli agglomeration into bacterial flocs, and therefore excludes the possibility that some E. coli cells might be undetectable by Coliscan Easygel due to being entrapped by bacterial flocs. Therefore, CE can be used to represent all the maximum E. coli cells that are undetectable due to PSP entrapment. E. coli concentrations before and after the ultrasonic (42 kHz) treatment were measured in triplicate by Coliscan Easygel.
2.3.3. Effects of PSP on ultrasonic (20 kHz) disinfection of entrapped and unentrapped E. coli cells Four E. coli-PSP solutions were prepared: one with 24 h of incubation at 35 C and three without. The incubated sample
3303
was prepared as follows: 0.80 g of dewatered PSP was made into 50 ml slurry to which 9.0 ul of E. coli stock solution was added. The mixture was incubated by stirring at 100 PRM for 24 h at 35 C. After incubation the slurry was diluted into 600 ml solution with distilled water and was radiated by UV light at wavelengths of 254 nm and 365 nm for 20 min to disinfect the unentrapped E. coli cells. For the three unincubated samples, 600 ml of distilled water and 50 ul of E. coli stock solution were mixed with dewatered PSP of 0.10, 0.33, and 0.80 g, respectively for 90 s at room temperature before experiments. A volume of 300 ml from each of the four samples was ultrasonicated at 20 kHz and 34 W (dissipated power, Pdiss) for 35 min in the ultrasonic disinfection system. Specific energy, Es, was calculated according to Equation (1). E. coli concentrations were measured in triplicate by Coliscan Easygel every 7 min during ultrasonication.
2.3.4. Mechanism for ultrasonic (20 kHz) disinfection of entrapped E. coli cells An incubated sample was prepared as follows: 0.80 g of dewatered PSP was made into 50 ml slurry to which 9.0 ul of E. coli stock solution was added. The mixture was incubated by stirring at 100 PRM for 24 h at 35 C. After incubation the slurry was diluted into 600 ml solution with distilled water and was radiated by UV light at wavelengths of 254 and 365 nm for 20 min to disinfect the unentrapped E. coli cells. A volume of 300 ml of the incubated sample was ultrasonicated at 20 kHz and 46 W (dissipated power, Pdiss) for 35 min in the ultrasonic disinfection system. Specific energy, Es, was calculated according to Equation (1). Samples of 2 ml were collected after 7, 14, and 35 min of ultrasonication, and were examined under SEM. Sample preparation for SEM imaging is described in the Supplementary material.
3.
Results and discussion
3.1.
Entrapment of E. coli cells by PSP
The morphology of PSP was characterized by SEM before and after incubation with E. coli cells in water for 24 h at 35 C (Fig. 2) to study the spatial relationship between PSP and E. coli. Fig. 2(a) showed that PSP had an irregular macroporous structure, and such chaotic morphology of PSP can be attributed to its various organic contents in forms of cellulose, microbial extracellular contents, and other complex carbohydrates, proteins, and lipids (Brar et al., 2008; Mahmoud et al., 2004; Zheng et al., 2009). Fig. 2(a) showed no sign of E. coli cells attaching to or entrapped by PSP before incubation. In preliminary test, no E. coli cells were detected by Coliscan Easygel in PSP before and after 42 kHz ultrasonic pretreatment either. Comparatively, Fig. 2(b) shows that after incubation, the rod-shaped bacteria not only attached to the PSP, but also were entrapped inside the macropores of PSP. It should be noted that even though the entrapped cells shown in Fig. 2(b) were not completely encapsulated, they demonstrated the ability of PSP to provide macropores to entrap E. coli cells, preventing them from contacting chromogenic coating in the pretreated petri dish used Coliscan Easygel.
3304
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 0 0 e3 3 0 8
Fig. 3 presents the relationship between CE and CSS in samples that were incubated for 24 h at 35 C. When CSS ranged from 0.17 to 1.3 mg/ml, CE positively correlated with CSS. This correlation was established through the Trendline function of Excel according to Equation (2) with an R2 of 0.94 and a root mean square error of 2.1 102 CFU/ml. The PSP entrapment coefficient, B, was estimated as 1.4 103 CFU/mg and represents the ability of PSP to entrap E. coli cells. A complimentary test was conducted to verify the entrapment coefficient: E. coli stock of 9.0 ul was incubated for 24 h at 35 C with 50 ml slurry that contained 0.80 g dewatered PSP. After incubation, the mixture was diluted into 600 ml from which 150 ml was collected and treated in ultrasonics cleaner at 42 kHz for 14 min (Es ¼ 17 kJ/L) to release all entrapped undetectable E. coli cells. The concentration of entrapped E. coli was calculated (Equation (3)) as 2.0 103 CFU/ml, which lies in the range of 1.4 CSS 2.1 102 CFU/ml (derived from the linear regression in Fig. 3). Therefore, the estimated value of 1.4 103 CFU/mg is valid as the entrapment coefficient under our experimental conditions. The correlation between CSS and CE provides an option for predicting the amount of entrapped undetectable pathogenic bacteria in samples where porous PSP is present. Further study is needed regarding the effects of PSP on ultrasonic (20 kHz) disinfection of both entrapped and unentrapped E. coli cells.
3.2. Effects of PSP on ultrasonic disinfection of entrapped and unentrapped E. coli cells
Fig. 2 e SEM images of PSP before (a) and after (b) incubation with E. coli cells at 35 C for 24 h. Pictures were taken at accelerating voltage of 15 kV. Samples were sputter coated with gold.
Our SEM images provide evidence for the entrapment of E. coli cells by PSP. These entrapped E. coli cells can be hard to detect by culture-based methods, and may survive the disinfection through traditional techniques such as chlorination, ozonation, and UV radiation, posing a potential threat to public health. Therefore, it is of great interests to develop a correlation between PSP concentration (CSS,) and the concentration (CE) of E. coli cells made undetectable to Coliscan Easygel due to PSP entrapment. The correlation was established according to Equation (2). The values of CE used during correlation establishment were obtained according to Equation (3) based on E. coli concentrations measured by Coliscan Easygel before and after 14 min (Es as 17 kJ/L) of ultrasonic (42 kHz) pretreatment. As described in the Experimental Section, CE represents all E. coli cells that were made undetectable to Coliscan Easygel by PSP entrapment under experimental conditions. Initial E. coli concentrations (C0, Equation (2)) were 1.1 102, 1.0 102, 3.1 102, and 1.0 103 CFU/ml for 0.17, 0.55, 0.90, and 1.3 mg PSP/ml, respectively.
Both incubated and unincubated samples were ultrasonicated (20 kHz) to study the effect of PSP on ultrasonic disinfection of entrapped and unentrapped E. coli cells. “Incubated” means that E. coli and PSP had been stirred together at 100 RPM for 24 h at 35 C before experiments, while “unincubated” means that the E. coli and PSP powder had just been mixed for 90 s at room temperature before experiments. N Two types of normalized E. coli concentration, CN t;d and Ct , are developed to evaluate ultrasonic disinfection (20 kHz) of
Fig. 3 e Values of CE in samples with CSS of 0.17, 0.55, 0.90, and 1.3 mg/ml. Standard error bars represent triplicate measurements and the figure shows that CE increased with CSS in a linear pattern.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 0 0 e3 3 0 8
3305
incubated and unincubated samples, and they are calculated as follows, CNt;d ¼
Ct C0 þ CE
(4)
CNt ¼
Ct þ CE;t C0 þ CE
(5)
Where CNt;d is normalized concentration of unentrapped E. coli (unitless), after ultrasonic (20 kHz) disinfection for time t, Ct is concentration of unentrapped E. coli (CFU/ml) measured by Coliscan Easygel after ultrasonic (20 kHz) disinfection for time t, C0 is the concentration of unentrapped E. coli (CFU/ml) measured by Coliscan Easygel before ultrasonic (20 kHz) disinfection, CE is the concentration of entrapped E. coli (CFU/ ml) measured before ultrasonic (20 kHz) disinfection, CNt is the normalized concentration of overall E. coli (unitless, including both unentrapped and entrapped E. coli) after ultrasonic (20 kHz) disinfection for time t, and CE,t is the concentration of entrapped E. coli (CFU/ml) measured after ultrasonication (20 kHz) for time t. For the incubated sample, the value of CE was estimated according to the formula of 1.4 103 CSS 2.1 102 (derived from the linear regression in Fig. 3). For the unincubated samples, our test showed CE and CE,t were both zero. ThereN fore, for unincubated samples, CN t;d is equal to Ct . The initial E. coli concentrations (combining both entrapped and unentrapped E. coli) of the samples were 2.6 103 2.2 102 CFU/ml for the incubated sample (1.3 mg PSP/ml), 1.2 103 1.7 102, 8.6 102 1.6 102, and 1.6 103 1.2 102 for the unincubated samples with PSP concentrations of 0.17, 0.55, and 1.3 mg/ml, respectively. F-test ( p ¼ 0.05) did not show significant differences in the CN t values among the three unincubated samples for the same Es value (Fig. 4). This indicates that PSP may not have significant effects on ultrasonic disinfection of unentrapped E. coli cells under our experimental conditions. The size of PSP was at the submillimeter level (Fig. 2), and is much smaller than the 20 kHz ultrasound wavelength which is 75 mm (Thangavadivel et al., 2009). This is called long wavelength regime in which scattering of ultrasound waves by PSP is negligible (Babick et al., 1998), and thereby unentrapped E. coli cells were not shielded by PSP from ultrasound waves. Solid particles were previously reported having both positive and negative effects on the performance of ultrasonication. On the one hand, particles act as impurities to reduce the cavitational threshold, resulting in more cavitational nuclei for a larger number of collapse events (Ince and Belen, 2001). On the other hand, particles can dampen ultrasonic waves through the mechanism of dissipation (Babick et al., 1998). Based on the results summarized in Fig. 4, it was concluded that during ultrasonic disinfection (20 kHz) of unentrapped E. coli cells under our experimental conditions both the positive and negative effects of PSP was probably neutralized by each other. The value of CN t;d in the incubated sample increased during the first 7 min of ultrasonic disinfection (20 kHz, from 0 to 48 kJ/L), and then decreased during the remaining 28 min of ultrasonic disinfection (20 kHz, from 48 to 240 kJ/L) (Fig. 4). Ultrasonication (20 kHz) had two functions toward the incubated sample: breakdown of the PSP around entrapped E. coli
N Fig. 4 e Profiles of CN t (for unincubated samples), and Ct;d (for the incubated sample) at different Es values during ultrasonic disinfection (20 kHz). Standard error bars represent triplicate measurements. The figure shows that entrapped E. coli cells were protected from ultrasonic disinfection and unentrapped E. coli were not. However, such protection gradually decreased as the ultrasonic disinfection proceeded. Es values of 48, 96, 144, 192 and 240 kJ/L correspond to 7, 14, 21, 28, and 35 min of ultrasonication (20 kHz) under our experimental conditions.
and disinfection of unentrapped E. coli (Blume and Neis, 2004; Drakopoulou et al., 2009; Madge and Jensen, 2002; Phull et al., 1997). Initial increase of CN t;d indicates the dominance of the former function (breakdown of the entrapping PSP) during the first 7 min of ultrasonic disinfection (20 kHz, from 0 to 48 kJ/L), and the following decrease of CN t;d indicates the dominance of the latter function (E. coli disinfection) during the remaining 28 min of ultrasonic disinfection (20 kHz, from 48 to 240 kJ/L). One-tailed t-test ( p ¼ 0.05) shows that at each point from Es ¼ 48 kJ/L to Es ¼ 192 kJ/L, the CN t;d value of the incubated sample (1.3 mg PSP/ml) was significantly higher than the CN t value of the unincubated sample with the same PSP dosage (1.3 mg PSP/L). It should be noted that for the incubated N sample, CN t is larger than or equal to Ct;d due to the possibilities that some E. coli cells might still be entrapped inside PSP even after certain time of ultrasonication at 20 kHz. Therefore, at each point from Es ¼ 48 kJ/L to Es ¼ 192 kJ/L, the CN t value of the incubated sample (1.3 mg PSP/ml) was significantly higher than the CN t value of the unincubated sample with the same PSP dosage (1.3 mg PSP/ml). The incubated sample had significant amounts of E. coli cells entrapped by PSP while the unincubated sample had none. Thus, the fact that the CN t value of the unincubated sample (1.3 mg PSP/L) was lower than the CN t value of the incubated sample (1.3 mg PSP/L) from Es ¼ 48 kJ/L to Es ¼ 192 kJ/L indicates that the performance of ultrasonic disinfection (20 kHz) in the incubated sample was reduced by the entrapment of E. coli by PSP. In other words, entrapped E. coli cells were protected from ultrasonic disinfection (20 kHz) by PSP. Protection of bacteria by suspended solids was also previously observed during disinfection by UV, ozone, and other disinfectants (Anderson et al., 1982; Qualls
3306
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 0 0 e3 3 0 8
et al., 1983). However, there do not appear to be any published studies about the protection of entrapped E. coli cells by PSP refugia from ultrasonic disinfection. At each Es value from 96 to 240 kJ/L, the difference between the CN t;d value of the incubated sample (1.3 mg PSP/ml) and the CN t value of the unincubated sample (1.3 mg PSP/ml) decreased from 0.22 (for Es ¼ 96 kJ/L), 0.20 (for Es ¼ 144 kJ/L), 0.10 (for Es ¼ 192 kJ/L), to 0.030 (for Es ¼ 240 kJ/L). This trend can be described by Inequality (6), CNtþDt;d ðincubatedÞ CNtþDt ðunincubatedÞ < CNt;d ðincubatedÞ CNt ðunincubatedÞ
(6)
With the ability to breakdown particle agglomerate (Blume and Neis, 2004), power ultrasonication at 20 kHz can reduce the concentration of entrapped E. coli cells, causing the value of ðCNt CNt;d Þ in the incubated sample to decrease with time. This trend can be described by Inequality (7), CNtþDt ðincubatedÞ CNtþDt;d ðincubatedÞ CNt ðincubatedÞ CNt;d ðincubatedÞ
(7)
Combining Inequality (6) to Inequality (7), yields Inequality (8) CNtþDt ðincubatedÞ CNtþDt ðunincubatedÞ < CNt ðincubatedÞ CNt ðunincubatedÞ
(8)
Inequalities (6)e(8) were applicable during ultrasonic disinfection (20 kHz) of the incubated sample and unincubated sample (1.3 mg PSP/ml for both samples) when Es increased from 96 to 240 kJ/L under our experimental conditions. Inequality (8) shows that the difference of CN t between the incubated and unincubated sample (1.3 mg PSP/L for both samples) decreased with time. This phenomenon indicates that as entrapped E. coli cells lost their protective PSP layer due
to ultrasonication (20 kHz), performance for ultrasonic disinfection of the incubated sample (1.3 mg PSP/ml) became more and more similar to that of the unincubated sample (1.3 mg PSP/ml). This feature agrees with the reported improvement of wastewater disinfection by power ultrasonication (Blume and Neis, 2004). Based on Fig. 4, after 240 kJ/L (Pdiss ¼ 34 W), the percentages of E. coli removal (from the perspective of all the active cultivable E. coli cells) were 96%, 98%, 94%, and 91% (or less) when PSP concentrations were 0.17 mg/ml (unincubated), 0.55 mg/ml (unincubated), 1.3 mg/ml (unincubated), and 1.3 mg/ml (incubated), respectively. Apparently, the dissipated power and the specific energy needed for such performance were much higher than those (Pdiss ¼ 3 W and Es ¼ 17 kJ/L) needed by ultrasonics cleaner to release the entrapped E. coli cells at 42 kHz.
3.3. Mechanism for ultrasonic disinfection of E. coli entrapped by PSP To better understand how ultrasonic disinfection of the entrapped E. coli occurs, the mechanism for ultrasonic disinfection (20 kHz) of the E. coli cells entrapped by PSP need to be clarified. The mechanisms were studied through SEM imaging of E. coli-PSP complexes when Es increased from 0, 64, 129, to 320 kJ/L during ultrasonication at 20 kHz. Before ultrasonication, E. coli cells were associated with and entrapped inside PSP (Fig. 5(a)). After 7 min of ultrasonication (20 kHz, 64 kJ/L), an E. coli-PSP complex has been severed and some E. coli cells have been exposed due to ultrasonic impact, but were still undamaged (Fig. 5(b)). Comparatively, an unentrapped E. coli cell was severely damaged after 7 min (20 kHz, 64 kJ/L) of ultrasonic disinfection (Fig. 5(c)). After 14 min of ultrasonic disinfection (20 kHz, 129 kJ/L), a bacteria-PSP complex has been more severed than
Fig. 5 e (a). PSP after (b) incubation with E. coli cells at 35 C for 24 h; (b). SEM images of an E. coli-PSP complexes after 7 min of ultrasonic disinfection (20 kHz and 64 kJ/L), showing a severed complex and exposing of some E. coli cells; (c). SEM image of a damaged E. coli cell that was not protected by PSP after 7 min of ultrasonic disinfection (20 kHz and 64 kJ/L); (d). SEM image of an E. coli-PSP complex after 14 min of ultrasonic disinfection (20 kHz and 129 kJ/L), showing a more severed complex and damaged E. coli cells; (e). SEM image of an E. coli-PSP complex after 35 min of ultrasonic disinfection (20 kHz and 320 kJ/L), showing a broken complex and few intact E. coli cells left.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 0 0 e3 3 0 8
the complex shown in Fig. 5(b), and while there were still some cells undamaged due to the protection by PSP, some cells have already been broken apart (Fig. 5(d)). After 35 min of ultrasonic disinfection (20 kHz, 320 kJ/L), an E. coli-PSP was shattered and few E. coli cells remained intact (Fig. 5(e)). Based on Fig. 5, ultrasonic disinfection of the entrapped E. coli cells occurs in two steps: (1) breakdown of the PSP refugia around the entrapped E. coli cells and (2) destruction of the exposed cells. As mentioned earlier, breakdown of protective PSP can occur through particleeparticle collision by shock waves, and pitting and erosion of PSP surface by liquid jets (He et al., 2007). This two-step mechanism explains why incubated samples were harder to disinfect than unincubated samples when PSP content was the same. However, once the protective PSP layer is peeled off the initially entrapped E. coli cells, complete disinfection of E. coli by ultrasonication may realize given sufficient energy (Fig. 5(e)). This phenomenon agrees with the conclusion from Fig. 4 that as the energy of ultrasonic disinfection increased, the performance for ultrasonic disinfection of incubated E. coli-PSP solution became more and more similar to that of the unincubated E. coli-PSP solution when both samples contained 1.3 mg PSP/ml.
3307
be economically feasible unless high-level disinfection is needed for high-concentration organic particulates samples (e.g. primary sludge).
Acknowledgments Thanks to Mr. Roy Beavers who has provided great advice on SEM imaging, Dr. Larry Ruben and Mr. Greg Hubbard from the Department of Biological Sciences of Southern Methodist University who have provided the lyophilizer and valuable assistance, Ms. Laurie Muller from Molecular and Cellular Imaging Facilities of University of Texas-Southwestern Medical Center for help in the SEM studies, and Dr. Andrew Quicksall, Mr. Gary Michael Nijak Jr. and Mr. Brian D. Fisher from the Department of Environmental and Civil Engineering of Southern Methodist University for their comments on the paper. We also thank sponsors of the Indiana 21st Century Project for their financial support, and Dallas, TX Central Wastewater Treatment Plant for supply of primary sludge.
Appendix. Supplementary material 4.
Conclusions
The spatial relationship between E. coli and PSP was studied after they were incubated together in water for 24 h at 35 C. Our study through SEM demonstrated a macroporous structure of PSP, and the entrapment of E. coli by PSP. Such entrapment can make E. coli cells undetectable to Coliscan Easygel, and correlation was established between the concentration of PSP and that of the E. coli cells made undetectable to Coliscan Easygel by PSP entrapment. A new parameter, entrapment coefficient, was proposed to reflect the ability of PSP to entrap E. coli cells, and it was estimated as 1.4 103 CFU/mg PSP under our experimental conditions when the initial E. coli concentrations (C0, Equation (2)) were 1.1 102, 1.0 102, 3.1 102, and 1.0 103 CFU/ml for 0.17, 0.55, 0.90, and 1.3 mg PSP/L, respectively. The effects of PSP on ultrasonic disinfection (20 kHz) of the entrapped or unentrapped E. coli cells have also been studied. Our results indicate that the entrapped E. coli cells were protected from ultrasonic disinfection while the unentrapped cells were not. Such protection gradually decreased as the energy of ultrasonic disinfection (20 kHz) increased. After 240 kJ/L, more than 90% of E. coli has been inactivated in the three unincubated samples with PSP dosage ranging from 0.17 to 1.3 mg/ml. The mechanism for ultrasonic disinfection (20 kHz) of the E. coli cells entrapped by PSP was studied through SEM. A twostep mechanism was tentatively concluded: breakdown of the entrapping PSP around the protected E. coli cells, and disinfection of the exposed cells. It has been found that ultrasonication (20 kHz) can overcome the entrapping PSP, and complete disinfection of entrapped E. coli cells can be expected given enough energy. PSP-E. coli complex has been completely shattered as evidenced by the SEM image in Fig. 5(e) when Es increased to 320 kJ/L under our experimental conditions. The energy level for this ultrasonic disinfection (20 kHz) may not
Supplementary data related to this article can be found online at doi:10.1016/j.watres.2011.03.034.
references
Anderson, A.C., Reimers, R.S., DeKenion, P., 1982. A brief review of the current status of alternatives to chlorine disinfection of water. Am. J. Public Health 72 (11), 1290e1293. Babick, F., Ripperger, F., Hinze, F., Stintz, M., 1998. Ultrasonic spectrometry for particle size analysis in dense submicron suspensions. Part. Part. Syst. Charact. 15 (5), 230e236. Bayoudh, B., Othmane, A., Mora, L., Ouada, H.B., 2009. Assessing bacterial adhesion using DLVO and XDLVO theories and the jet impingement technique. Colloids Surf., B 73 (1), 1e9. Blume, T., Neis, U., 2004. Improved wastewater disinfection by ultrasonic pre-treatment. Ultrason. Sonochem. 11 (5), 333e336. Boorman, G.A., Dellarco, V., Dunnick, J.K., Chapin, R.E., Hunter, S., Hauchman, F., Gardner, H., Cox, M., Sills, R.C., 1999. Drinking water disinfection byproducts: review and approach to toxicity evaluation. Environ. Health Perspect. 107 (Suppl. 1), 207e217. Brar, S.K., Surampalli, R.Y., Verma, M., Valro, J.R., Tyagi, R.D., 2008. Bacillus thuringiensis fermentation of primary and mixed sludge: rheology and process performance. J. Environ. Eng. 134 (8), 659e670. Carballa, M., Omil, F., Lema, J.M., Llompart, M., Garcı´a-Jares, C., Rodrı´guez, I., Go´mez, M., Ternes, T., 2004. Behavior of pharmaceuticals, cosmetics and hormones in a sewage treatment plant. Water Res. 38 (12), 2918e2926. Chen, X., Chen, G., Yue, P.L., 2000. Separation of pollutants from restaurant wastewater by electrocoagulation. Separat. Purif. Technol. 19 (1e2), 65e76. Cheng, H.A., Lucy, F.E., Graczyk, T.K., Broaders, M.A., Tamang, L., Connolly, M., 2009. Fate of Cryptosporidium parvum and Cryptosporidium hominis oocysts and Giardia duodenalis cysts during secondary wastewater treatments. Parasitol. Res. 105 (3), 689e696.
3308
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 0 0 e3 3 0 8
Dahlem, O., Demaiffe, V., Halloin, V., Reisse, J., 1998. Direct sonication system suitable for medium-scale sonochemical reactors. AIChE J. 44 (12), 2724e2730. Drakopoulou, S., Terzakis, S., Fountoulakis, M.S., Mantzavinos, D. , Manios, T., 2009. Ultrasound-induced disinfection of gramnegative and gram-positive bacteria in secondary treated municipal wastewater. Ultrason. Sonochem. 16 (5), 629e634. Findik, S., Gu¨ndu¨z, G., Gu¨ndu¨z, E., 2006. Direct sonication of acetic acid in aqueous solutions. Ultrason. Sonochem. 13 (3), 203e207. Foladori, P., Laura, B., Gianni, A., Giuliano, Z., 2007. Effects of sonication on bacteria viability in wastewater treatment plants evaluated by flow cytometryefecal indicators, wastewater and activated sludge. Water Res. 41 (1), 235e243. Fu, H., Suri, R.P.S., Chimchirian, R.F., Helmig, E., Constable, R., 2007. Ultrasound-induced destruction of low levels of estrogen hormones in aqueous solutions. Environ. Sci. Technol. 41 (16), 5869e5874. Gattie, D.K., Lewis, D.L., 2004. A high-level disinfection standard for land-applied sewage sludges (biosolids). Environ. Health Perspect. 112 (2), 126e131. Greaves, R.I.N., 1960. Preservation of living cells by freeze-drying. Ann. N. Y. Acad. Sci. 85, 723e728. Han, Y., Dague, R.R., 1997. Laboratory studies on the temperaturephased anaerobic digestion of domestic primary sludge. Water Environ. Res. 69 (6), 1139e1143. He, Z., Traina, S.J., Weavers, L.K., 2007. Sonolytic desorption of mercury from aluminum oxide: effects of pH, chloride, and organic matter. Environ. Sci. Technol. 41 (3), 779e784. Ince, N.I., Belen, R., 2001. Aqueous phase disinfection with power ultrasound: process kinetics and effect of solid catalysts. Environ. Sci. Technol. 35 (9), 1885e1888. Ji, Z., Chen, G., Chen, Y., 2010. Effects of waste activated sludge and surfactant addition on primary sludge hydrolysis and short-chain fatty acids accumulation. Bioresour. Technol. 101 (10), 3457e3462. Joyce, E., Phull, S.S., Lorimer, J.P., Mason, T.J., 2003. The development and evaluation of ultrasound for the treatment of bacterial suspensions. A study of frequency, power and sonication time on cultured Bacillus species. Ultrason. Sonochem. 10 (6), 315e318. Kubo, M., Onodera, R., Shibasaki-Kitakawa, N., Tsumoto, K., Yonemoto, T., 2005. Kinetics of ultrasonic disinfection of Escherichia coli in the presence of titanium dioxide particles. Biotechnol. Prog. 21 (3), 897e901. Madge, B.A., Jensen, J.N., 2002. Disinfection of wastewater using a 20-kHz ultrasound unit. Water Environ. Res. 74 (2), 159e169. Mahmoud, N., Zeeman, G., Gijzen, H., Lettinga, G., 2004. Anaerobic stabilisation and conversion of biopolymers in primary sludge e effect of temperature and sludge retention time. Water Res. 38 (4), 983e991. Mason, T.J., Lorimer, J.P., Bates, D.M., 1992. Quantifying sonochemistry: casting some light on a ‘black art’. Ultrasonics 30 (1), 40e42.
McGhee, T.J., 1991. Water Supply and Sewage, McGraw-Hill, New York. Micrology Laboratories, 2009. Coliscan Easygel. http://www. micrologylabs.com/Home/Our_Methods/Coliscan_Media/ Coliscan_Easygel (accessed 22.04.10.). Narkis, N., Armon, R., Offer, R., Orshansky, F., Friedland, E., 1995. Effect of suspended solids of wastewater disinfection efficiency by chlorine dioxide. Water Res. 29 (1), 227e236. Ning, P., Bart, H., Jiang, Y., Haan, A., Tien, C., 2005. Treatment of organic pollutants in coke plant wastewater by the method of ultrasonic irradiation, catalytic oxidation and activated sludge. Separat. Purif. Technol. 41 (2), 133e139. Oulahal-Lagsir, N., Martial-Gros, A., Bonneau, M., Blum, L.J., 2000. Ultrasonic methodology coupled to ATP bioluminescence for the non-invasive detection of fouling in food processing equipment: validation and application to a dairy factory. J. Appl. Microbiol. 89 (3), 433e441. Phull, S.S., Newman, A.P., Lorimer, J.P., Pollet, B., Mason, T.J., 1997. The development and evaluation of ultrasound in the biocidal treatment of water. Ultrason. Sonochem. 4 (2), 157e164. Qualls, R.G., Flynn, M.P., Johnson, J.D., 1983. The role of suspended particles in ultraviolet disinfection. J. Water Pollut. Control 55 (10), 1280e1285. Ratti, C., 1999. Hot air and freeze-drying of high-value foods: a review. J. Food Eng. 49 (4), 311e319. Thangavadivel, K., Megharaj, M., Smart, R.S.C., Lesniewski, P.J., Naidu, R., 2009. Application of high frequency ultrasound in the destruction of DDT in contaminated sand and water. J. Hazard. Mater. 168 (2e3), 1380e1386. Tiehm, A., Nickel, K., Neis, U., 1997. The use of ultrasound to accelerate the anaerobic digestion of sewage sludge. Water Sci. Technol. 36 (11), 121e128. Ubukata, Y., 2007. The role of particulate organic matter and acetic acid in the removal of phosphate in anaerobic/aerobic activated sludge processes. Eng. Life Sci. 7 (1), 61e66. United States Environmental Protection Agency, 1993. Standards for the use and disposal of sewage sludge. Fed. Regist. 58, 9398. Walker, A.S., and Yanko, W.A. Wang, J., Wang, L., Wang, B., Zhang, J., Zou, Q., 2006. Impact of suspended particles and enhancement techniques on ultraviolet disinfection of a secondary effluent. J. Ocean Univ. China 5 (4), 381e386. Xie, R., Xing, Y., Ghani, Y.A., Ooi, K., Ng, S., 2007. Full-scale demonstration of an ultrasonic disintegration technology in enhancing anaerobic digestion of mixed primary and thickened secondary sewage sludge. J. Environ. Eng. Sci. 6 (5), 533e541. Yasui, K., 2002. Influence of ultrasonic frequency on multibubble sonoluminescence. J. Acoust. Soc. Am. 112 (4), 1405e1413. Zheng, J., Kennedy, K.J., Eskicioglu, C., 2009. Effect of low temperature microwave pretreatment on characteristics and mesophilic digestion of primary sludge. Environ. Technol. 30 (4), 319e327.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 0 9 e3 3 1 7
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Hydroxyl radical involvement in the decomposition of hydrogen peroxide by ferrous and ferric-nitrilotriacetate complexes at neutral pH Yen Hai Dao, Joseph De Laat* Universite´ de Poitiers, Laboratoire de Chimie et Microbiologie de l’Eau (CNRS UMR 6008), Ecole Nationale Supe´rieure d’Inge´nieurs de Poitiers, 40, Avenue du Recteur Pineau, 86 022 Poitiers Cedex, France
article info
abstract
Article history:
The relative rates of degradation of three hydroxyl radical probe compounds (atrazine,
Received 12 December 2010
fenuron and parachlorobenzoic acid (pCBA)) by FeIII/H2O2 (pH ¼ 2.85), FeIIINTA/H2O2
Received in revised form
(neutral pH), FeII/O2, FeIINTA/O2, FeII/H2O2 and FeIINTA/H2O2 (neutral pH) have been
16 March 2011
investigated using the competitive kinetic method. Experiments were carried out in batch
Accepted 21 March 2011
and in semi-batch reactors, in the dark, at 25 C. The data showed that the three probe
Available online 29 March 2011
compounds could be degraded by all the systems studied, and in particular by FeIINTA/ H2O2 and FeIIINTA/H2O2 at neutral pH. The relative rate constants of degradation of the
Keywords:
three probe compounds obtained for all the systems tested were identical and equal to
Oxidation
1.45 0.03 and 0.47 0.02 for kAtrazine/kpCBA and kFenuron/kpCBA, respectively. These values
Fenton-like reactions
as well as the decrease of the rates of degradation of the probe compounds upon the
Competitive kinetics
addition of hydroxyl radical scavengers (tert-butanol, bicarbonate ions) suggest that the
Atrazine
degradation of atrazine, fenuron and pCBA by FeIINTA/O2, FeIINTA/H2O2 and FeIIINTA/H2O2
Fenuron
is initiated by hydroxyl radicals. ª 2011 Elsevier Ltd. All rights reserved.
Parachlorobenzoic acid Tert-butanol Bicarbonate ion
1.
Introduction
Advanced Oxidation Processes (AOPs) can be successfully used in the field of wastewater treatment to reduce the chemical oxygen demand and the toxicity of industrial wastewaters, to convert toxic and biorecalcitrant contaminants into biodegradable by-products, to remove colour or to obtain a complete mineralization of organic pollutants (Pera-Titus et al., 2004). Among the AOPs, the Fenton and the Fenton-like oxidation processes (FeII/H2O2, FeIII/H2O2) have been applied to many industrial wastewaters such as chemical, petrochemical, pharmaceutical and textile effluents (Pignatello et al., 2006; Bautista
et al., 2008). These oxidation processes are also popular methods for the remediation of contaminated soils and groundwater (Bergendahl et al., 2003; Watts and Teel, 2005). The main benefits of the Fenton and Fenton-like reactions are the use of environmentally friendly and low cost reagents. However, these oxidation processes have also some limitations. The FeII/H2O2 and FeIII/H2O2 systems must be operated at low pH values (pH z 3) to prevent the precipitation of ferric oxyhydroxides and to produce hydroxyl radicals. In addition, the efficiency of the Fenton reaction can be markedly decreased in the presence of high concentrations of inorganic salts such as chloride and sulfate ions because of the formation of inactive
* Corresponding author. Tel.: þ33 5 49 45 39 21; fax: þ33 5 49 45 37 68. E-mail address:
[email protected] (J. De Laat). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.03.043
3310
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 0 9 e3 3 1 7
iron(III)-complexes (De Laat and Le, 2005, 2006). To overcome these drawbacks, the use of organic and inorganic ironchelating agents has been investigated by several authors in order to enhance the efficiency of the homogenous Fenton-like oxidation processes at neutral pH (Sun and Pignatello, 1992, 1993; Li et al., 2007; Lee and Sedlak, 2009; Rastogi et al., 2009) or of heterogenous systems involving ion bearing minerals (Xue et al., 2009) or granular zero-valent iron (Keenan and Sedlak, 2008; Lee et al., 2008). In their pioneering works, Sun and Pignatello (1992) have tested several classes of iron(III)-chelates for the decomposition of H2O2 and of 2,4-dichlorophenoxyacetic at pH 6. Of the 50 compounds tested, nitrilotriacetic acid (NTA) was one of the most active iron(III)-chelate. The iron-NTA/H2O2 system was also found to be effective for the degradation of tetrachloroethene in contaminated soils (Howsawkeng et al., 2001; Ndjou’ou et al., 2006). Kim and Kong (2001) showed that the FeIIINTA/H2O2 system degraded more efficiently 1-hexanol and carbon tetrachloride at pH 9 than at pH 3 and that the degradations of 1-hexanol and carbon tetrachloride are initiated by hydroxyl radical (HO) and by superoxide anion radical (HO2/O2), respectively. The nature of the reactive oxidant species (free and bound HO radicals, high-valent-oxoiron species) generated by the reaction of H2O2 with free and complexed Fe(II) and Fe(III) species in aqueous solution remains a controversial issue (Walling et al., 1975; Yamazaki and Piette, 1991; Bossmann et al., 1998; Pignatello et al., 2006). In the absence of chelating agents, it is generally accepted that the reaction of H2O2 with Fe2þ at low pH yields hydroxyl radicals whereas other oxidants such as the ferryl ion may be produced at circumneutral pH (Gallard et al., 1998; Rivas et al., 2001; Hug and Leupin, 2003; Keenan and Sedlak, 2008; Katsoyiannis et al., 2008). In all these studies, proofs of the formation of one or another oxidizing species are based on EPR spin-trapping techniques, distribution of oxidation by-products, effects of added hydroxyl radical scavengers on reaction rates or on kinetic modeling studies. In this work, the competitive kinetic method has been applied to the degradation of three hydroxyl radical probes (atrazine, fenuron and parachlorobenzoic acid) in order to highlight the formation of hydroxyl radicals by the FeIINTA/ H2O2 and FeIIINTA/H2O2 systems at neutral pH. The degradation of probe compounds by the FeII/O2 and FeIINTA/O2 systems has also been investigated because dissolved oxygen readily oxidizes ferrous ion (King et al., 1995; Rose and Waite, 2002) and FeIINTA (Harris and Aisen, 1973; Welch et al., 2002) at neutral pH. The effects of the addition of HO radical scavengers (tert-butanol or bicarbonate ions) on the degradation rates of probe compounds and of H2O2 will also be examined. However, the effects of experimental parameters on the rate of decomposition of H2O2 by FeIINTA and FeIIINTA will be presented in a next paper and will not be discussed here.
2.
Materials and methods
2.1.
Reagents and preparation of solutions
All chemicals used in this work were of reagent grade or higher and were used without further purification. All aqueous solutions were prepared with Milli-Q water.
The stock solution of pCBA (100 mM) was prepared at pH 10.5. The stock solution of fenuron was prepared with a precise concentration (0.5 mM). The stock solution of atrazine was prepared by dissolving the required amount of atrazine in water. After a mixing time of about 48 h in the darkness, the solution was filtered through a 0.45 mm filter and the concentration of dissolved atrazine in the filtrate was determined by HPLC. Stock solutions of Fe(III) (10 mM) were prepared by dissolving the required amount of iron(III) perchlorate (Fe(ClO4)3, 9 H2O) in HClO4 0.1 M. Since iron(III) perchlorate has a great tendency to hydrate, its degree of hydration was checked regularly by spectrophotometric titration using 1,10-orthophenanthroline. Solutions of FeIIINTA were prepared by mixing appropriate volumes of stock solutions of ferric perchlorate (10 mM) and of sodium nitrilotriacetate (5e7 mM). The pH was then adjusted with NaOH 1 M. All FeIIINTA solutions were freshly prepared each time before use. Aqueous solutions of FeII and FeIINTA at neutral pH are extremely sensitive to dissolved oxygen and must be prepared in oxygen-free water and handled under a protective nitrogen atmosphere. For the FeII/O2 and FeII/H2O2 oxidation experiments, a stock solution of Fe(II) (50.0 0.5 mM) was prepared by dissolving the appropriate weight of iron(II) perchlorate hydrate (Fe(ClO4)2, 7 H2O) in oxygen-free pure water. For the stock solution of FeIINTA (50.0 0.5 mM; FeII:NTA molar ratio ¼ 1:3), 1.435 g of NTA powder (7.5 mmol) was dissolved into 48.0 0.5 mL of water. After dissolution, the NTA solution was deoxygenated by nitrogen bubbling for at least 20 min. Then, 0.98 g (2.5 mM) of iron(II) perchlorate hydrate (Fe(ClO4)2, 7 H2O) was added to the solution and a pale green solution was obtained. A nitrogen flux was then kept over the solution during the experiments.
2.2.
Experimental conditions
All the experiments were carried with initial concentrations of atrazine, fenuron and pCBA of 5 mM and were conducted in the dark and at 25.0 0.5 C. All the oxidation experiments conducted with FeIINTA or FeIIINTA complexes have been performed by using [NTA]0:[FeII]0 and [NTA]0:[FeIII]0 molar ratios of 3:1. Under these conditions, FeII and FeIII were present only as FeIINTA and FeIIINTA complexes at the beginning of the reaction (Motekaitis and Martell, 1994). It should also be noted that uncomplexed NTA is therefore always present in the solutions and shall compete with atrazine, fenuron and pCBA for the reaction with reactive species generated in solution. Oxidation experiments with the FeIII/H2O2 (pH ¼ 2.85) and the III Fe NTA/H2O2 (7 < pH < 9) were carried out using a batch reactor. The batch reactor consisted of a 1.3-L cylindrical double-wall jacketed reactor in order to circulate thermostated water with an external circulating pump connected to a thermostated water bath (25.0 0.5 C) (Fig. S1 in Supplementary material). The reactor was covered by a black plastic film to protect the aqueous solution from ambient light. The reactor was filled with 1.25 L of solution containing FeIII or FeIIINTA and the three organic solutes. The solution was mixed using a magnetic stirrer at nearly 800 rpm during all the course of the reaction. Initial pH was adjusted to the desired value using HClO4 or NaOH. During the course of the reaction, the pH was kept constant using a pH
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 0 9 e3 3 1 7
transmitter (OPM 223, Endress þ Hauser) and a peristaltic pump (Gilson Minipuls 3) for the injection of NaOH 1 M (liquid flow rate : 10e15 mL/h). The reaction was initiated by adding H2O2 into the reactor. At regular time intervals, samples were withdrawn from the reactor and immediately analyzed for H2O2 (1 or 2 mL samples) or immediately quenched with 1 mL of methanol (for 1 mL sample) for the analyses of HO probe compounds. Because of the very fast reactions between O2 or H2O2 with free FeII or FeIINTA at neutral pH, concentrationetime profiles for HO probe compounds could not be obtained using a batch reactor for the FeII/(O2 or H2O2) and the FeIINTA/(O2 or H2O2) systems. A series of experiments were carried out using a semi-batch reactor with a continuous introduction of iron(II) perchlorate or of iron(II)-NTA. The reactor was equipped with a pH electrode connected to a pH transmitter (OPM 223, Endress þ Hauser) and a WTW CellOx 325 dissolved oxygen sensor connected to a calibrated oxygen meter (WTW Oxi 340i-A/SET). A volume of 1.25 L of a solution containing the three HO probe compounds (pH ¼ 7.0) was introduced into the reactor. For the FeII/O2 and the FeIINTA/O2, the solution was oxygenated by bubbling air at a flow rate of nearly 1 L/min. When the equilibrium concentration of dissolved oxygen in water was reached (z8.3 mg/L), the gas bubbling was stopped and all the valves were closed to avoid external oxygen contact during the reaction. The stock solution of FeII(ClO4)2 or of FeIINTA (50 mM) was then injected at time t ¼ 0, at a constant flow rate (z25 or z50 mL/h) with a Gilson Minipuls 3 peristaltic pump. During the course of the reaction, the pH was kept constant at pH 7.0 0.1 using the pH transmitter and a peristaltic pump for the injection of NaOH (0.1e1 M) or of HClO4 (0.1e1 M) (liquid flow rate 10e15 mL/h). For the FeII/H2O2 and the FeIIINTA/H2O2 oxidation experiments, the solution containing the probe compounds (5 mM each) and H2O2 (0.5 mM) was either deoxygenated by bubbling nitrogen gas ([O2] in water < 0.1 mg O2/L) or aerated with air ([O2] z 8.3 mg O2/L) during at least 30 min before injecting the stock solution of FeII(ClO4)2 or FeII/NTA (50 mM) at a constant flow rate (z25 or 50 mL/h). During the course of the reaction, the pH and the concentrations of dissolved oxygen were noted every 30 or 60 s. Samples were taken from the reactor at various reaction times and immediately analyzed without filtration for H2O2, Fe(II) and for total Fe concentrations. Samples were also withdrawn from the reactor and immediately quenched with methanol (1 mL methanol for 1 mL sample), filtered through 0.45 mm pore size membrane filters and then analyzed by HPLC for the determination of the concentrations of HO probes. Batch experiments have also been carried out with the FeII/ H2O2 and FeIINTA/H2O2 systems. For these experiments, 100 mL of a stock solution of FeII(ClO4)2 (50 mM) or FeIINTA (50 mM, FeII:NTA molar ratio ¼ 1:3) were introduced into a series of flasks containing 100 mL of solution of probe compounds (5 mM each) and H2O2 at concentrations ranging from 0 to 1 mM ([H2O2]0 ¼ 0, 10, 25, 50, 80, 100, 150, 200, 500 and 1000 mM). The initial pH before the introduction of FeII or FeIINTA was equal to 7.0. The addition of the FeII or FeIINTA solution was done under vigorous mixing. After a reaction time of 3 h (residual H2O2 ¼ 0 mM, and complete oxidation of FeII into FeIII), the pH of the solutions was determined and the solutions quenched with methanol (100 mL/mL of sample) to stop the degradation of the probe compounds.
2.3.
3311
Analytical methods
The concentration of hydrogen peroxide in stock solutions of H2O2 was determined by iodometric titration. Concentrations of hydrogen peroxide in solutions containing Fe(II) or Fe(III) were determined spectrophotometrically using the TiCl4 method (Eisenberg, 1943) and a molar absorption coefficient of 724 M1 cm1 for the titanium peroxocomplex. The concentrations of Fe(II) or of Fe(III) (after reduction of Fe(III) by hydroxylamine hydrochloride) were measured by the o-phenanthroline colorimetric method and by using a molar extinction coefficient of 1.105 104 M1 cm1 at 510 nm for the Fe(II)ephenanthroline complex (Tamura et al., 1974). pH measurements were made with a Meter Lab PHM 240 pH meter calibrated with standard buffers. Dissolved oxygen was analyzed using a calibrated oxygen meter (WTW Oxi 340i-A/SET). Atrazine, fenuron and pCBA were analyzed by HPLC using a Waters HPLC system equipped with a Waters 717 plus auto sampler, a Waters 996 photodiode array detector, a Waters Millennium software, a Waters Spherisorb C8 column (5 mm, 4 250 mm). A mobile phase consisting of 50% methanol and 50% H2O (pH 2, acidification with CF3COOH) was used at a flow rate of 0.8 mL/min. Absorbance was measured continuously in the range of 200e400 nm. Fenuron and pCBA, peaks were normally quantified at 240 nm and atrazine at 220 nm.
2.4.
Competitive kinetic expression
Competitive kinetic experiments using three probe compounds for HO radicals have been conducted to ascertain the formation of hydroxyl radicals. By assuming that the degradation of the model compounds are only due to an attack by the hydroxyl radical and by using pCBA as the reference compound, the relative degradation rates of atrazine and fenuron should be described by the following competitive kinetic expressions: ½Atrazinet kAtrazine ½pCBAt ¼ ln ln ½Atrazine0 kpCBA ½pCBA0
(1)
½Fenuront kFenuron ½pCBAt ln ¼ ln ½Fenuron0 kpCBA ½pCBA0
(2)
where kAtrazine, kFenuron and kpCBA are the absolute rate constants for the reaction of HO radicals with atrazine, fenuron and pCBA respectively. Previous competitive kinetic studies carried out with the Fenton reaction (Fe2þ/H2O2, pH ¼ 3) showed that kFenuron/kpCBA ¼ 1.47 0.04 (Acero et al., 2002). A value of 2.92 0.04 has been determined for kFenuron/katrazine from H2O2/UV experiments (254 nm, pH 7.2) (Mazellier et al., 2007). From these values, a value of 0.50 0.03 can be calculated for kAtrazine/kpCBA. By using a value of 5.0 109 M1 s1 for the reaction rate constant of HO radicals with parachlorobenzoate anion (Elovitz and von Gunten, 1999) and by assuming that the rate constants for the reaction of HO radicals with the acid and the basic forms of pCBA (pKa of pCBA z 4) are identical (kpCBA ¼ 5.0 109 M1 s1), the absolute second-order rate constants kAtrazine and kFenuron should be equal to 2.4 109 M1 s1 and 7.0 109 M1 s1, respectively. If the plots of ln([A]t/[A]0) against ln([B]t/[B]0) (A ¼ atrazine or fenuron, B ¼ pCBA) obtained from different experimental conditions or from various oxidation processes yield
3312
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 0 9 e3 3 1 7
superimposable straight lines with a slope equal to the expected kA/kB value, one can therefore say that the degradation of the model compounds can be attributed to the hydroxyl radical.
a
1 III
Fe /H2 O 2 ; pH 2.85 Tert -Butanol : 50 mM (Exp.A4)
Results
For the concentrations of reactants used in the present work, control experiments showed that, in the dark, the HO probe compounds were not degraded by H2O2 alone, by FeIII alone (acidic pH), by FeIIINTA alone (acidic and neutral pH), by FeII or by FeIINTA alone in deoxygenated solutions at neutral pH (data not shown). In addition, coagulation experiments carried out with iron(III) perchlorate (0e1 mM) and pH showed that the maximum losses of HO probe compounds by adsorption on the iron(III) hydroxide surface were less than 3% and 5% for a total doses of FeIII equal to 0.25 mM and 1 mM, respectively. Degradation of probe compounds was only observed in the presence of H2O2 for solutions containing iron species (FeII, FeIII, FeIINTA, FeIIINTA) or in the presence dissolved oxygen for solutions containing ferrous species (FeII and FeIINTA).
3.1. Oxidation by FeIII/H2O2 (pH 2.85) and FeIIINTA/ H2O2 (neutral pH)
0.6
Atrazine pCBA III
0.4
Fenuron
Fe /H2 O 2 ; pH 2.85 Tert -Butanol : 0 mM (Exp. A1-A3)
0.2
0
5000
10000
15000
Time (s)
b
1 III
Fe /H2 O2 ; pH 2.85 Tert -Butanol : 0 mM (Exp. A1-A3)
0.8
- ln (Ct / C 0 )
3.
Ct / C0
0.8
Fenuron y = 1.481x
0.6 0.4
Atrazine
0.2
y = 0.462x
Figs. 1 and 2 present data obtained for the degradation of the three HO radical probes by FeIII/H2O2 at pH 2.85 and by FeIIINTA/H2O2 at neutral pH. The experimental conditions and the relative rate constants kAtrazine/kpCBA and kFenuron/kpCBA obtained for all experiments carried with these two oxidation processes have been reported in Table 1. For each experiment, detailed data can be found in the Supplementary material. Identical initial concentrations of FeIII (0.2 mM) and H2O2 (0.5 mM) were used in order to compare the rates of degradation of the organic compounds by FeIII/H2O2 and FeIIINTA/H2O2. As shown the data in Fig. 1b, competitive oxidation experiments carried out with the classical Fenton-like reaction (FeIII/ H2O2, pH ¼ 2.85, Experiments A1eA3 in Table 1) yielded relative rate constant values (kAtrazine/kpCBA ¼ 0.47 0.01 and kFenuron/ kpCBA ¼ 1.48 0.01; mean value of three experiments) which are consistent with the expected values for a HO radical-initiated mechanism of degradation of organic compounds by the Fenton-like reaction. In addition, the degradations of the three organic solutes were totally inhibited in the presence of 50 mM of tert-butanol (Fig. 1a). If HO radicals are generated in solutions, they can react with the probe compounds with secondorder rate constants (ki) roughly equal to 5 109 M1 s1, H2O2 (ki z 3 107 M1 s1), tert-butanol (ki z 6 108 M1 s1, Buxton et al., 1988), free NTA (ki ¼ 6.1 107, 5.5 108 and 4.2 109 M1 s1 at pH 2, 6 and 10, respectively (Sahul and Sharma, 1987)) and with FeIINTA (minor reaction, ki unknown). A comparison of the values of the scavenging term ki Ci (ki : rate constant for the reaction of HO with a solute i, and Ci concentration of the solute i) calculated for tert-butanol (6 108 5 102 ¼ 3 107 s1), a probe compound (z5 109 5 106 z 2.5 104 s1), H2O2 (z3 107 5 104 z 1.5 104 s1) and free NTA (<5 108 2 103 ¼ < 1 106 s1 at pH < 3) indicate that more than 97% of HO radicals would be scavenged by tert-butanol under our conditions. All these data confirm that the hydroxyl radical is the main
0
0
0.2
0.4
0.6
0.8
- ln([pCBA] t / [pCBA]0 ) Fig. 1 e Oxidation of probe compounds by FeIII/H2O2 at pH 2.85 in the absence and in the presence of tert-butanol. (a) Normalized concentrationetime profiles of atrazine, fenuron and pCBA in the absence and in the presence of 50 mM tert-butanol; (b) Plots of the competitive kinetic equation for the determination of the relative rate constants (Experiments A1eA4 in Table 1).
oxidant responsible for the degradation of organic compounds by the FeIII/H2O2 at acidic pH. Fig. 2 presents typical data obtained for the FeIIINTA/H2O2 system in the pH range 7e9 and with a molar FeIII:NTA ratio of 1:3 ([FeIII]0 ¼ 0.2 mM, [NTA]0 ¼ 0.6 mM). At neutral pH, the FeIII/ H2O2 is not efficient for the decomposition of H2O2 and for the oxidation of organic compounds because of the formation of non-soluble and inactive iron(III) hydroxides. As shown the data in Fig. 2, the addition of NTA leads to the formation of soluble FeIIINTA complexes which can decompose H2O2 (Fig. 2a) and degrade the model organic compounds (Fig. 2b) at neutral pH. The decomposition rate of H2O2 by FeIIINTA was faster at pH 8 than at pH 7 or 9, in agreement with previous data obtained by Tachiev et al. (2000) for the study of the effects of pH on the initial rates of decomposition of H2O2 by several iron(III)-complexes with aminopolycarboxylic acids including NTA. For an iron(III) complex such as FeIIINTA, Tachiev et al. (2000) showed that the decomposition rate of H2O2 is strongly dependent on the stability constants for the formation of the various FeIIINTA complexes and on the reactivity of the various FeIIINTA complexes toward H2O2.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 0 9 e3 3 1 7
a
1
[H 2 O 2 ]t / [H2O 2 ]0
pH = 7.0 0.8
pH = 8.0 pH = 9.0
0.6 0.4 0.2 0 0
1000
2000
3000
4000
5000
Time (s)
b
1
Tert -Butanol: 10 mM ; pH 8.3
[C] t / [C] 0 (Fenuron)
-
HCO3 : 50 mM ; pH8.3
0.8 -
HCO 3 : 25 mM ; pH8.3
0.6
-
pH 7.0
0.4
HCO 3 : 10 mM ; pH 8.3
pH 9.0
0.2 pH 8.0
0
0
1000
2000
3000
Time (s)
- ln ([C] t / [C] 0 )
c
III
Fe -NTA/H2O 2 7 < pH < 9 NaHCO3 : 0 - 50 mM
2 1.6
Fenuron y = 1.433 x
1.2 0.8
Atrazine y = 0.460 x
0.4 0
0
0.5
1
1.5
2
2.5
-ln([pCBA]t / [pCBA] 0 ) Fig. 2 e Oxidation of probe compounds by FeIIINTA/H2O2 at neutral pH. (a) Normalized concentrationetime profiles of H2O2 at pH 7, 8 and 9; (b) Normalized concentrationetime profiles of fenuron in the absence and in the presence of tertbutanol (10 mM) or of bicarbonate ion (10e50 mM); (c) Plots of the competitive kinetic equation for the determination of the relative rate constants (Experimental conditions are given in Table 1, Experiments B1eB11 and C1eC4).
The degradation rates of model organic compounds were also faster at pH 8 than at pH 7 or 9 and decreased in the presence of tert-butanol (10 mM) or of bicarbonate ions (10e50 mM) (Fig. 2b). Under the conditions used, the inhibition of the degradation rates of the HO probes was greater in the
3313
presence of tert-butanol than of bicarbonate ion, in agreement with the values of the scavenging terms for tert-butanol (ki Ci ¼ 6.0 106 s1) and for bicarbonate ions (ki Ci ¼ 1.2 105, 3.0 105 and 6.0 105 s1 for Ci ¼ 10, 25 and 50 mM and pH ¼ 8.3). The apparent second-order rate constant for the reaction of HO radicals with bicarbonate ion (ki) at pH 8.3 is equal to 1.2 107 M1 s1 and has been calculated by using pKa values of 6.35 and 10.33 for H2CO3 and second-order rate constants of 0, 8.6 106 and 3.9 108 M1 s1 for the reaction of HO radicals with H2CO3, HCO3 and CO32, respectively (Buxton et al., 1988). The plots of ln([Atrazine]0/[Atrazine]t) and of ln([Fenuron]0/ [Fenuron]t) against ln([pCBA]0/[pCBA]t) yield straight lines and the slopes of the straight lines were not affected by the pH of the solutions and by the concentration of bicarbonate (Fig. 2c). The mean values of the ratios kAtrazine/kpCBA and kFenuron/kpCBA determined from the 15 experimental values obtained with the FeIIINTA/H2O2 system were equal to 0.46 (s.d. 0.02) and 1.44 (s.d. 0.03), respectively (Experiments B1eB11, C1eC4, Table 1). These values of relative rate constants as well as the inhibiting effect of bicarbonate ion demonstrate that the degradation of model organic compounds by the FeIIINTA/H2O2 system can be attributed to the hydroxyl radical. It should also be noted that the identical relative rate constants obtained for the degradation of the HO probes by FeIII/H2O2 at pH 2.85 and by FeIIINTA/ H2O2 at pH 7 indicate that the absolute rate constants of HO radicals with the acid and the basic forms of pCBA are quite identical confirming the assumption made above.
3.2. pH
Oxidation by FeII/O2 and by FeIINTA/O2 at neutral
In the classical Fenton-like reaction (Fe2þ/H2O2, acidic pH), the reaction of H2O2 with Fe2þ represents the unique source for the generation of HO radicals and the reduction reaction of iron(III) into iron(II) represents the rate-limiting step for the overall rate of decomposition of H2O2 (De Laat and Gallard, 1999). At neutral pH, iron(II) species (FeOHþ and Fe(OH)2) are readily oxidized by H2O2 and can also be rapidly oxidized by dissolved oxygen at pH >7 (King et al., 1995; Santana-Casiano et al., 2005). It has also been found that the presence of NTA significantly enhances the rate of oxidation of iron(II) by oxygen (Kurimura et al., 1968) because of the formation of more reactive FeIINTA complexes. As dissolved oxygen oxidizes ferrous ion and FeIINTA at neutral pH, a series of experiments was carried out without adding H2O2 to check if the reactions of dissolved oxygen with free FeII species or with FeIINTA lead to the formation of intermediates that can degrade the three model compounds used in the present work. These experiments were carried out in a semibatch reactor with a continuous introduction of FeII(ClO4)2 (50 mM) or FeIINTA (50 mM; FeII:NTA molar ratio ¼ 1:3) at a flow rate of about 25 or 50 ml L1 h1. For each experiment, the feed doses of FeII or of FeIINTA (0.72e2.3 mM L1 h1, Table 2) were calculated from the flow rate delivered by the peristaltic pump and the exact concentration of Fe(II) in the stock solutions of Fe(II). The feed doses were confirmed by the slope of the straight lines obtained for the concentrationetime profiles of total iron in the reactor (Fig. 3a). Data in Fig. 3a show a decrease of the concentration of dissolved oxygen with increasing reaction time and a nearly
3314
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 0 9 e3 3 1 7
Table 1 e Relative rate constants of degradation of model organic solutes obtained for the FeIII/H2O2 system (experiments A1eA4) and the FeIIINTA/H2O2 system in the absence (experiments B1eB9) and in the presence of sodium hydrogenocarbonate (experiments C1eC4). (Batch reactor; [Fe(III)]0 [ 0.206 mM; [NTA]0 [ 0.6 mM; [H2O2]0 [ 0.5 mM; [Atrazine]0 [ [Fenuron]0 [ [pCBA]0 [ 5 mM; 25 C). Exp. A1 A2 A3 A4
Process tested III
Fe /H2O2 FeIII/H2O2 FeIII/H2O2 FeIII/H2O2 III
pH 2.85 2.85 2.85 2.85
0.05 0.05 0.05 0.05
B1 B2 B3 B4 B5 B6 B7 B8 B9 B10 B11
Fe NTA/H2O2 FeIIINTA/H2O2 FeIIINTA/H2O2 FeIIINTA/H2O2 FeIIINTA/H2O2 FeIIINTA/H2O2 FeIIINTA/H2O2 FeIIINTA/H2O2 FeIIINTA/H2O2 FeIIINTA/H2O2 FeIIINTA/H2O2
7.0 7.0 7.0 7.0 8.0 8.0 8.0 8.3 8.3 9.0 9.0
0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1
C1 C2 C3 C4 C5
FeIIINTA/H2O2 FeIIINTA/H2O2 FeIIINTA/H2O2 FeIIINTA/H2O2 FeIIINTA/H2O2
8.3 8.3 8.3 8.3 8.3
0.1 0.1 0.1 0.1 0.1
[NaHCO3]0 (mM)
Tert-Butanol (mM)
kAtrazine/kpCBA
kFenuron/kpCBA
50
0.471 0.473 0.441 e
1.482 1.476 1.486 e
0.482 0.494 0.458 0.453 0.436 0.451 0.444 0.420 0.441 0.464 0.463
1.437 1.486 1.428 1.404 1.453 1.440 1.439 1.443 1.389 1.403 1.430
0.466 0.488 0.456 0.480 e
1.412 1.417 1.468 1.448 e
0.460 0.020
1.441 0.030
5 10 25 50 10
Mean value Standard deviation
complete conversion of ferrous species into ferric species (z95% for FeII; z97% for FeIINTA at a reaction time of 20 min). The consumptions of dissolved oxygen by free FeII and FeIINTA were equal to 0.243 0.005 mol of O2/mole of FeII or FeIINTA (Table 2), in agreement with the expected overall stoichiometry for the oxidation of ferrous species by O2. The rates of consumption of dissolved oxygen by FeII and FeIINTA were identical because the introduction doses of FeII and FeIINTA were identical and represented the rate-limiting step in the overall rate of consumption of dissolved oxygen at pH 7.0(Fig. 3a). The data also demonstrate that atrazine, fenuron and pCBA could be degraded by FeII and FeIINTA at neutral pH in the presence of dissolved oxygen and that the degradation rates were faster with FeIINTA than with FeII (Fig. 3b). In addition, the
values of the relative rate constants determined from the competitive kinetic expression (kAtrazine/kpCBA ¼ 0.46 0.01 and kFenuron/kpCBA ¼ 1.45 0.02, Fig. 3c, Table 2) showed that the degradation of the organic solutes can be attributed to HO radicals and provide evidence for the formation of HO radicals during the oxidation of FeII and FeIINTA by dissolved oxygen. These results are consistent with other data showing the involvement of hydroxyl radicals in the degradation of organic compounds during the oxidation of iron(II) (Burns et al., 2010) or of FeIINTA (Kachur et al., 1998; Keenan and Sedlak, 2008) by dissolved oxygen at circumneutral pH. In the FeII/O2 and FeIINTA/O2 systems, HO radicals can be formed via the Fenton reaction because H2O2 can be produced from the two-step reduction of dissolved oxygen into HO2/O2 and then into H2O2.
Table 2 e Dissolved oxygen consumption (D[O2]/D[FeII]) and relative rate constants of degradation of probe compounds obtained with the FeII/O2 and FeIINTA/O2 systems (Semi-batch reactor, [O2]0 [ 8.3 ± 0.1 mg O2 LL1; [Atrazine]0 [ [Fenuron]0 [ [pCBA]0 [ 5 mM; Feed dose of FeII or of FeIINTA (FeII:NTA molar ratio [ 1:3) [ 0.72e2.33 mM LL1 hL1; pH [ 7.0 ± 0.1; 25 C). Exp.
Process tested II
D1 D2 D3
Fe /O2 FeII/O2 FeII/O2
E1 E2 E3 E4 E5
FeIINTA FeIINTA FeIINTA FeIINTA FeIINTA
Mean value Standard deviation
(1:3)/O2 (1:3)/O2 (1:3)/O2 (1:3)/O2 (1:3)/O2
FeII introduced (mM/L1 h1)
DO2/D FeII (mol/mol)
kFenuron/kpCBA
kAtrazine/kpCBA
2.295 2.255 1.874
0.243 0.246 0.245
1.427 1.436 1.445
0.466 0.459 0.447
2.350 2.316 2.331 0.727 0.872
0.244 0.241 0.245 0.239 0.24
1.469 1.476 1.472 1.445 1.453
0.478 0.468 0.462 0.458 0.467
0.243 0.003
1.453 0.018
0.463 0.009
3315
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 0 9 e3 3 1 7
0.3
a
1
II
Open symbols : Fe /O2 Dark symbols Fe NTA/O2
0.8 Total iron
0.2
0.6
Dissolved oxygen
0.4 0.1
0.2 Ferrous iron
0
0
0
300
600
900
1200
0,2 Total iron
H2O2
0,3 0,2
0,1
0,1
Ferrous iron
0
200
400 600 Time (s)
800
0 1000
1
1 0.8
0.8
II
II
Fe /O2
Atrazine pCBA Fenuron
0.6
[C] t / [C] 0
[C] t / [C] 0
0,4
0
b
0,3
II
Open symbols : Fe /H2O2 (Exp. F1) II Dark symbols : Fe NTA/H2O2 (Exp. G4)
0,5
1500
Time (s)
b
0,6
[Iron species] (mM)
[Iron species] (mM)
[Dissolved oxygen] (mM)
II
[Hydrogen peroxide] (mM)
a
0.4
Fe /H2O2 (Exp. F1)
0.6
Atrazine pCBA Fenuron
0.4 II
II
0.2
Fe -NTA/H2O2 (Exp. G4)
0.2
Fe NTA/O 2
0
0
0
300
600
900
1200
1500
1800
0
200
400
600
800
1000
Time (s)
Time (s)
c
Fenuron
II
Fe NTA / O2 FeII-NTA/O2 II Fe NTA / O2 FeII-NTA/O2 II Fe / O2 FeII/O2 II FeII/O2 Fe / O2
2.50 2.00
- ln ([C]t / [C]0
Fig. 4 e Oxidation of probe compounds by FeII/H2O2 and FeIINTA/H2O2 at neutral pH. (a) Concentrationetime profiles of H2O2 and of iron species (unfiltered samples), (b) Normalized concentrationetime profiles of probe compounds, (Experimental conditions are given in Table 3, Experiments F1 and G4).
3.00
1.50
y = 1.453x
Atrazine
1.00 0.50
y = 0.463x
0.00
0
0.5
1
1.5
2
- ln([pCBA] t / [pCBA]0 ) Fig. 3 e Oxidation of probe compounds by FeII/O2 and FeII/ NTA/O2 at pH [ 7.0. (a) Concentrationetime profiles of O2 and of iron species(unfiltered samples); (b) Normalized concentrationetime profiles of probe compounds; (c) Plots of the competitive kinetic equation for the determination of the relative rate constants (Experimental conditions are given in Table 2).
3.3. Oxidation by FeII/H2O2 and by FeIINTA/H2O2 at neutral pH Oxidation experiments with the FeII/H2O2 ([O2]0 < 0.1 mg/L) and FeIINTA/H2O2 ([O2]0 < 0.1 and 8.5 mg/L) systems were also carried
out in the semi-batch reactor with a continuous introduction of a solution of FeII(ClO4)2 (45 mM) or of FeIINTA (45 mM, FeII:NTA molar ratio ¼ 1:3) into the reactor (experiments F1 and G1eG4 in Table 3 and Fig. 4). The initial concentration of H2O2 was 0.5 mM and the pH was regulated at pH ¼ 7.0 0.1. Contrary to the experiments carried out with FeII/O2, FeIINTA/O2 and FeII/H2O2 systems, ferrous ion was never detected in the experiments conducted with FeIINTA/H2O2 (Fig. 4a) because of the very fast reaction of H2O2 with FeIINTA at neutral pH. The data showed that the hydrogen peroxide consumptions were nearly equal to 1.3 and 1.8 0.1 mol H2O2/mol FeII for the FeII/H2O2 and the FeIINTA/H2O2 systems, respectively. These data will be discussed in a next paper. The data in Fig. 4b also showed that the degradation rates of HO probes were faster with FeIINTA/H2O2 than with FeII/H2O2, probably because the reaction rates by the FeII/H2O2 are likely to be limited by the precipitation of iron(III) hydroxides (De Laat and Gallard, 1999). Similar conclusions could be drawn from data obtained from batch experiments (0.5 mM; [NTA]0 ¼ 0 or 1 mM), initial (pH ¼ 7.0) and with initial concentrations of H2O2 ranging from 0 to 1 mM (Table S4). The plots of competitive kinetic expression to data obtained from batch and semi-batch experiments yielded straight lines with
3316
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 0 9 e3 3 1 7
Table 3 e Hydrogen peroxide consumption (D[H2O2]/D[FeII]) and relative rate constants of degradation of probe compounds obtained with the FeII/H2O2 and FeIINTA/H2O2 systems in the absence and in the presence of dissolved oxygen. (Semi-batch reactor, [H2O2]0 [ 0.5 mM; [Atrazine]0 [ [Fenuron]0 [ [pCBA]0 [ 5 mM; Feed dose of FeII or of FeIINTA (FeII:NTA molar ratio [ 1:3) [ 1.0e1.2 mM LL1 hL1, pH [ 7.0 ± 0.1; 25 C). Exp. F1 G1 G2 G3 G4
Process tested FeII/H2O2 FeIINTA (1:3)/H2O2 FeIINTA (1:3)/H2O2 FeIINTA (1:3)/H2O2 FeIINTA (1:3)/H2O2
FeII introduced (mM L1 h1) [O2]0 (mg L1) DH2O2/DFeII (mol mol1) kFenuron/kpCBA kAtrazine/kpCBA 1.038 1.193 1.152 1.135 1.033
<0.1 <0.1 <0.1 8.5 8.5
slopes equal to 0.48 0.02 and 1.46 0.02 for kAtrazine/kpCBA and kFenuron/kpCBA, respectively (Table 3). These relative rate constant values as well as the inhibition of the degradation of the three probes molecules in the presence of tert-butanol (50 mM) demonstrate the formation of HO radicals during the catalytic decomposition of FeIINTA by H2O2.
4.
Conclusions
The data obtained in the present work showed that the catalyzed decomposition of oxygen and/or hydrogen peroxide by iron(II) and iron(III)-nitrilotriacetate complexes (FeIINTA/O2, FeIINTA/H2O2 and FeIIINTA/H2O2 systems) can degrade atrazine, fenuron and parachlorobenzoic acid at neutral pH whereas the classical Fenton and Fenton-like oxidation processes (FeII/H2O2 and FeIII/H2O2) were not efficient at pH 7. The competitive kinetic study and the observed decrease of the reaction rates in the presence of hydroxyl radical scavengers (tert-butanol and bicarbonate ions) clearly demonstrate that the hydroxyl radical is the unique oxidant involved in the initial step of the degradation of the three model compounds at neutral pH. These data do not exclude the formation of ferryl ion but they demonstrate that ferryl ion is not the primary oxidant of the model compounds studied in the present work. The data also showed that the rates of decomposition of H2O2 and of the model compounds observed with the FeIIINTA/ H2O2 system were at a maximum at pH z 8. The study of the effects of pH as well as of other parameters on the rates of decomposition of H2O2 by FeIIINTA complexes is currently investigated in our lab and will be presented in a next paper.
Appendix. Supplementary material Supplementary data related to this article can be found online at doi:10.1016/j.watres.2011.03.043.
references
Acero, J.L., Benitez, F.J., Gonzalez, M., Benitez, R., 2002. Kinetics of fenuron decomposition by single-chemical oxidants and combined systems. Ind. Eng. Chem. Res. 41, 4225e4232. Bautista, P., Mohedano, A.F., Casas, J.A., Zazo, J.A., Rodriguez, J.J., 2008. An overview of the application of Fenton oxidation to
1.315 1.832 1.763 1.789 1.955
1.48 1.443 1.410 1.435 1.456
0.501 0.485 0.479 0.481 0.471
industrial wastewaters treatment. J. Chem. Technol. Biot. 83, 1323e1338. Bergendahl, J., Hubbard, S., Grasso, D., 2003. Pilot-scale Fenton’s oxidation of organic contaminants in groundwater using autochthonous iron. J. Hazard. Mater. 99, 43e56. Bossmann, S.H., Oliveros, E., Go¨b, S., Siegwart, S., Dahlen, E.P., Payawan, A.L., Straub, M., Wo¨rner, M., Braun, A.M., 1998. A new evidence against hydroxyl radicals as reactive intermediates in the thermal and photochemically enhanced Fenton reactions. J. Phys. Chem. A 102, 5542e5550. Burns, J.M., Craig, P.S., Shaw, T.J., Ferry, J.L., 2010. Multivariate examination of Fe(II)/Fe(III) cycling and consequent hydroxyl radical generation. Environ. Sci. Technol. 44, 7226e7231. Buxton, G.U., 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. J. Phys. Chem. Ref. Data 17 (2), 513e886. De Laat, J., Gallard, H., 1999. Catalytic decomposition of hydrogen peroxide by Fe(II) in homogenous aqueous solution: mechanism and kinetic modelling. Environ. Sci. Technol. 33, 2726e2732. De Laat, J., Le, T.G., 2005. Kinetics and modelling of the Fentonlike reaction in the presence of sulfate in acidic aqueous solutions. Environ. Sci. Technol. 39, 1811e1818. De Laat, J., Le, T.G., 2006. Effects of chloride ions on the iron(III)catalyzed decomposition of hydrogen peroxide and the efficiency of the Fenton-like oxidation process. Appl. Catal. B. Environ. 66, 137e146. Eisenberg, G.M., 1943. Colorimetric determination of hydrogen peroxide. Ind. Eng. Chem. 15, 327e328. Elovitz, M., von Gunten, U., 1999. Hydroxyl radical/ozone ratios during the ozonation processes. I. The Rct concept. Ozone Sci. Eng. 21, 239e260. Gallard, H., De Laat, J., Legube, B., 1998. Influence du pH sur la vitesse d’oxydation de compose´s organiques par FeII/H2O2. Me´canismes re´actionnels et mode´lisation. New J. Chem., 263e268. Harris, D.C., Aisen, P., 1973. Facilitation of Fe(II) autooxidation by Fe(III) complexing agents. Biochim. Biophys. Acta 329, 156e158. Howsawkeng, J., Watts, R.J., Washington, D.L., Teel, A.L., Hess, T.F., Crawford, R.L., 2001. Evidence for simultaneous abiotic-biotic oxidations in a microbial-Fenton’s system. Environ. Sci. Technol. 35, 2961e2966. Hug, S.J., Leupin, O., 2003. Iron-catalyzed oxidation of arsenic(III) by oxygen and by hydrogen peroxide: pH-dependent formation of oxidants in the Fenton reaction. Environ. Sci. Technol. 37, 2734e2742. Kachur, A.V., Tuttle, S.W., Biaglow, J.E., 1998. Autooxidation of ferrous ion complexes: a method for the generation of hydroxyl radicals. Radiat. Res. 150, 475e482. Katsoyiannis, I.A., Ruettimann, T., Hug, S.J., 2008. pH dependence of Fenton reagent generation and As(III) oxidation and removal by corrosion of zero valent iron in aerated water. Environ. Sci. Technol. 42, 7424e7430.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 0 9 e3 3 1 7
Keenan, C.R., Sedlak, D.L., 2008. Ligand-enhanced reactive oxidant generation by nanoparticulate zero-valent iron and oxygen. Environ. Sci. Technol. 42, 6936e6941. Kim, Y.-S., Kong, S.-H., 2001. Treatment of 1-hexanol and carbon tetrachloride in modified Fenton’s reagent using iron(III)-NTA complex. J. Korean Ind. Eng. Chem. 12, 787e792. King, D.W., Lounsbury, H.A., Millero, F.J., 1995. Rates and mechanism of Fe(II) oxidation at nanomolar total iron concentrations. Environ. Sci. Technol. 29, 818e824. Kurimura, Y., Ochiai, R., Matsuura, N., 1968. Oxygen oxidation of ferrous ions induced by chelation. Bull. Chem. Soc. Jpn. 41, 2234e2239. Lee, C., Keenan, C.R., Sedlak, D.L., 2008. Polyoxometalate oxidation of organic compounds by nanoparticulate zerovalent iron and ferrous ion in the presence of oxygen. Environ. Sci. Technol. 42, 4921e4926. Lee, C., Sedlak, D.L., 2009. A novel homogeneous Fenton-like system with Fe(III)-phosphotungstate for oxidation of organic compounds at neutral pH values. J. Mol. Catal. A: Chem. 311, 1e6. Li, Y.C., Bachas, L.G., Bhattacharyya, D., 2007. Selected chloroorganic detoxifications by polychelate (poly(acrylic acid)) and citrate-based Fenton reaction at neutral pH environment. Ind. Eng. Chem. Res. 46, 7984e7992. Mazellier, P., Busset, C., Delmont, A., De Laat, J., 2007. A comparison of fenuron degradation by hydroxyl and carbonate radicals in aqueous solution. Water Res. 41, 4585e4594. Motekaitis, R.J., Martell, A.E., 1994. The iron(III) and iron(II) complexes of nitrilotriacetic acid. J. Coord. Chem. 31, 67e78. Ndjou’ou, A.-C., Bou-Nasr, J., Cassidy, D., 2006. Effect of Fenton reaction dose on coexisting chemical and microbial oxidation in soil. Environ. Sci. Technol. 40, 2778e2783. Pera-Titus, M., Garcıa-Molina, V., Ban˜os, Miguel A., Gime´nez, J., Esplugas, S., 2004. Degradation of chlorophenols by means of advanced oxidation processes: a general review. Appl. Catal. B. Environ. 47, 219e256. Pignatello, J., Oliveros, E., MacKay, A., 2006. Advanced oxidation process for organic contaminant destruction based on the Fenton reaction and related chemistry. Crit. Rev. Environ. Sci. Technol. 36, 1e84. Rastogi, A., Al-Abed, S.R., Dionysiou, D.D., 2009. Effect of inorganic, synthetic and naturally chelating agents of Fe(II) mediated advanced oxidation of chlorophenols. Water Res. 43, 684e694.
3317
Rivas, F.J., Beltran, F.J., Frades, J., Buxeda, P., 2001. Oxidation of p-hydroxybenzoic acid by Fenton’s reagent. Water Res. 35, 387e396. Rose, A.L., Waite, T.D., 2002. Kinetic model for Fe(II) oxidation in seawater in the absence and presence of natural organic matter. Environ. Sci. Technol. 364, 433e444. Santana-Casiano, J.M., Gonza´lez-Da´vila, M., Millero, F.J., 2005. Oxidation of nanomolar levels of Fe(II) with oxygen in natural waters. Environ. Sci. Technol. 39, 2073e2079. Sahul, K., Sharma, B.K., 1987. Gamma radiolysis of nitrilotriacetic acid (NTA) in aqueous solutions. J. Radioanal. Nucl. Ch. 109, 321e327. Sun, Y., Pignatello, J.J., 1992. Chemical treatment of pesticide wastes. Evaluation of Fe(III) chelates for catalytic hydrogen peroxide oxidation of 2,4-D at circumneutral pH. J. Agric. Food Chem. 40, 322e327. Sun, Y., Pignatello, J.J., 1993. Activation of hydrogen peroxide by iron(III) chelates for abiotic degradation of herbicides and insecticides in water. J. Agric. Food Chem. 41 (2), 308e312. Tachiev, G., Roth, J.A., Bowers, A.R., 2000. Kinetics of hydrogen peroxide decomposition with complexed and “free” iron catalysts. Int. J. Chem. Kinet. 32, 24e35. Tamura, H., Goto, K., Yotsuyanagi, T., Nagayama, M., 1974. Spectrophotometric determination of iron(II) with 1. 10-phenanthroline in the presence of large amounts of iron(III). Talanta 21, 314e318. Walling, C., Partch, R.E., Weil, T., 1975. Kinetics of the decomposition of hydrogen peroxide catalyzed by ferric ethylenediaminetetraacetate complex. Proc. Nat. Acad. Sci. USA 72 (1), 140e142. Watts, R.J., Teel, A.L., 2005. Chemistry of modified Fenton’s reagent (catalyzed H2O2 propagations - CHP) for in situ soil and groundwater remediation. J. Environ. Eng. 131, 612e622. Welch, K.D., Davis, T.Z., Aust, S.D., 2002. Iron autoxidation and free radical generation: effects of buffers, ligands and chelators. Arch. Biochem. Biophys. 397, 360e369. Xue, X., Hanna, K., Despas, C., Wu, F., Deng, N., 2009. Effect of chelating agent on the oxidation of PCP in the magnetite/H2O2 system at neutral pH. J. Mol. Catal. A: Chem. 311, 29e35. Yamazaki, I., Piette, L., 1991. EPR spin-trapping study on the oxidizing species formed in the reaction of the ferrous ion with hydrogen peroxide. J. Am. Chem. Soc. 113, 7588e7593.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 1 8 e3 3 3 0
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Selective removal of phosphorus from wastewater combined with its recovery as a solid-phase fertilizer Sukalyan Sengupta a,*, Arka Pandit b,1 a
Department of Civil & Environmental Engineering, University of Massachusetts Dartmouth, 285 Old Westport Road, North Dartmouth, MA 02747, USA b Department of Civil & Environmental Engineering, Georgia Institute of Technology, Atlanta, GA 30332, USA
article info
abstract
Article history:
Influx of Phosphorus (P) into freshwater ecosystems is the primary cause of eutrophication
Received 26 October 2010
which has many undesirable effects. Therefore, P discharge limits for effluents from
Received in revised form
WWTPs is becoming increasingly common, and may be as low as 10 mg/L as P. While
20 March 2011
precipitation, filtration, membrane processes, Enhanced Biological Phosphorus Removal
Accepted 23 March 2011
(EBPR) and Physico-chemical (adsorption based) methods have been successfully used to
Available online 31 March 2011
effect P removal, only adsorption has the potential to recover the P as a usable fertilizer. This benefit will gain importance with time since P is a non-renewable resource and is
Keywords:
mined from P-rich rocks. This article provides details of a process where a polymeric anion
Phosphate
exchanger is impregnated with iron oxide nanoparticles to effectuate selective P removal
Ion-exchange
from wastewater and its recovery as a solid-phase fertilizer. Three such hybrid materials
Eutrophication
were studied: HAIX, DOW-HFO, & DOW-HFO-Cu. Each of these materials combines the
Hydrated iron oxide
durability, robustness, and ease-of-use of a polymeric ion-exchanger resin with the high
Selective phosphate removal
sorption affinity of Hydrated Ferric Oxide (HFO) toward phosphate. Laboratory experiments
Ligand
demonstrate that each of the three materials studies can selectively remove phosphate
Nanoparticle
from the background of competing anions and phosphorus can be recovered as a solid-
Lewis acid-base
phase fertilizer upon efficient regeneration of the exchanger and addition of a calcium or magnesium salt in equimolar (Ca/P or Mg/P) ratio. Also, there is no leaching of Fe or Cu from any of these hybrid exchangers. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
Eutrophication of freshwater sources is a widespread global pollution issue. Surveys reveal that around 50% of all the lakes and reservoirs in all the continents barring Africa and Oceania are eutrophic. In the US, an EPA (USEPA, 1997) study found that 36% of the lakes are eutrophic. In more focused studies, the problem has been found to be even more intensive. For example, over 1000 waterbodies in Idaho, Oregon, and Washington are identified as being impaired due to excessive
Phosphorus (P) loading and are included on State Clean Water Act lists for water quality problems (USEPA, 2007). Eutrophication affects the survival of many aquatic species e the most critical being commercial fish, affects the safety of drinking water supplies, affects the aesthetics of recreational areas, and the ability to navigate through rivers and lakes. It is commonly agreed upon that P is the rate-limiting nutrient in freshwater ecosystems, i.e. the level of P in freshwater bodies governs the growth of aquatic organisms. Many streams and lakes in the US have very little capacity to assimilate P loading
* Corresponding author. Tel.: þ1 508 999 8470; fax: þ1 508 999 8964. E-mail address:
[email protected] (S. Sengupta). 1 At the time that this research was conducted, graduate student in the Department of Civil & Environmental Engineering, University of Massachusetts Dartmouth, Massachusetts. 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.03.044
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 1 8 e3 3 3 0
during the "critical" warm and dry summer period without significant water quality degradation. Large diurnal swings in pH and dissolved oxygen may occur as excessive amounts of nutrients are metabolized by aquatic plants and algae. The range of these swings is often measured to exceed the state water quality criteria established to protect fish and other aquatic organisms in their various life stages. Therefore, the amount of phosphorus currently entering these waters exceeds the seasonal loading capacity and must be reduced if these water quality problems are to be resolved. The primary source of P in the environment is anthropogenic, present as both point sources (synthetic detergent in domestic wastewater) and non-point sources (agricultural run-off). The approximate contributions of P from major sources to domestic wastewater is estimated as 0.6 kg P/cap/yr from human wastes, 0.3 kg P/cap/yr from laundry detergents with no restrictions on P content, and 0.1 kg P/cap/yr from household detergents and other cleaners (Sedlak, 1991). Without significant commercial or industrial loads, the influent concentration of total phosphorus in WWTP effluent may range from 6 mg/L to 8 mg/L as P. There has been a concerted effort to reduce the amount of P in laundry detergents, and now there is growing momentum to eliminate P from dishwashing detergents. But even if these efforts are totally successful, the P concentration in WWTP effluent would reduce only to 4e5 mg/L P (USGS, 1999). Unfortunately, this is much higher than the natural capacity of rivers and lakes to assimilate P. Studies (Sharpley et al., 1994, 2003; Heathwaite and Sharpley, 1999; Seviour et al., 2003) indicate that lake water concentrations of P above 0.02 ppm generally accelerate eutrophication. Therefore, federal and state regulators have been applying strict P effluent limits. For instance, the Everglades Forever Act mandates a total P concentration of 10 ppb (Florida Everglades Forever Act, 1994); additionally, a recent plan establishes a systemic P reduction in effluent WWTP streams emptying into the Spokane river, with an ultimate mandate of 10 ppb (Hansen, 2006). While the 10 ppb limit is the strictest and rare, many WWTPs have been forced to retrofit or modify existing processes to meet limits such as 0.07 mg/L (Durham Advanced Wastewater Treatment Facility, Tigard, OR, and Rock Creek AWTF in Hillsboro, OR), or 0.1 mg/L (effluent Total P limit for all major POTWs discharging to Assabet river in MA). And there are many other WWTPs that are currently regulated to meet discharge standard of 500 ppb but have been notified that the limits will be reduced to 100 ppb. Thus, WWTPs have to prepare themselves for stricter P discharge limits in the range of 10e100 ppb. Another important aspect of the phosphorus issue is the potential for its decrease in worldwide supplies. Phosphorus is a non-renewable source, and is primarily available from phosphate rock, high-quality deposits of which are controlled by just five countries. Some predictions suggest (Herring and Fantel, 1993) that if the current trend of phosphate rock mining is continued the stock is predicted to dwindle in next fifty years. Thus, it is imperative that phosphorus recycle and reuse become an integral part of a WWTP process. Techniques for phosphorus removal from wastewater are in general biological and physicalechemical. It is readily recognized that traditional biological nutrient removal (BNR) and precipitationesorption processes are unable to reduce
3319
phosphate concentrations below 100 mg/L as P (Jenkins et al., 1971; Jenkins and Hermanowicz, 1991; Cooper et al., 1993; Kuba et al., 1993). About 95% of the P in municipal wastewater treatment plants (WWTPs) is as phosphate species (including organic phosphate, polyphosphate and orthophosphate; Snoeyink and Jenkins, 1980). Phosphoric acid (H3PO4) is a triprotic acid with the 3 pKa values being 2.0, 7.0, & 12.0. At near-neutral pH that is typical of WWTP processes, the species of orthophosphate phosphorus that predominate are HPO42 & H2PO4. Both species are strong ligands and oxides of polyvalent metals, namely, Fe(III), Ti(IV) and Zr(IV) exhibit strong ligand sorption (of HPO42 & H2PO4) through formation of innersphere complexes (Dzombak and Morel, 1990; Stumm and Morgan, 1995; Suzuki et al., 2000; Dutta et al., 2004; Gao et al., 1995; Yuchi et al., 2003). Of them, hydrated iron(III) oxide [FeOOH] or HFO is benign, inexpensive and stable over a wide pH range. Many previous studies have confirmed that HFO or a-Goethite has high sorption affinity for phosphates or a similar oxy-anion such as arsenite/arsenate(Huang and Vane, 1989; Golterman, 1995; Hiemstra and van Riemsdjik, 1999; Katsoyiannis and Zouboulis, 2002; Seida and Nakano, 2002; Onyango et al., 2003; Hristovski et al., 2008; Mezenner and Bensmaili, 2009). Granular activated alumina, zirconium oxide and iron oxide are a few well-studied inorganic sorbents in this regard (Seida and Nakano, 2002; Tanada et al., 2003; Kang et al., 2003; Zeng et al., 2004; Genz et al., 2004; Chubar et al., 2005). However, these metal oxide particles lack the mechanical strength and attrition resistance properties needed for prolonged operation in fixed-bed units; consequently the fines formed lead to increased head loss in the system (Blaney et al., 2007). The metal oxides referred to above are not also amenable to efficient regeneration and are hence recommended for single-use applications only (Genz et al., 2004).
1.1.
Polymeric anion exchangers
Polymeric materials offer durability and mechanical strength but commercially available polymeric anion exchangers lack specific selectivity or affinity for H2PO4 or HPO42, as compared to sulfate, an anion that is ubiquitous in WWTP effluent. A new genre of polymeric resins called Polymeric Ligand Exchangers (PLEs) overcomes this limitation by taking a free-form resin such as bisepicolylamine (shown in Fig. 1a) and loading it with a salt of a heavy-metal such as Cu2þ (Fig. 1b). The immobilized Cu2þ forms strong coordination bond with HPO42 or H2PO4 and this provides selectivity of orthophosphate anion over competing anions such as Cl, SO42, HCO3, and NO3 (Zhao and SenGupta, 1998). Another approach is to impregnate Hydrated Ferric Oxide (HFO; FeOOH) nanoparticles within a strong-base anionexchange polymeric resin (Fig. 1c). This material has displayed significantly increased sorption capacity (Cumbal et al., 2003; DeMarco et al., 2003; Cumbal and SenGupta, 2005; Blaney et al., 2007). The increased sorption capacity results from the Donnan co-ion exclusion effect exerted by the fixed positive charge. Donnan effect has been explained in many articles (Glueckauf and Watts, 1962; Seidel et al., 2001; Cumbal and SenGupta, 2005) and thus has not been discussed in this
3320
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 1 8 e3 3 3 0
Fig. 1 e a: Bisepicolylamine functional group of the parent free-form exchanger of DOW-HFO. b: Cu2D - loaded bis-picolylamine functional group of DOW-HFO-Cu. c: Quaternary ammonium functional group of the parent anion exchanger of HAIX.
article. However, based on the above approaches of making a HPO42/H2PO4 selective polymeric exchanger, the next logical step would be to impregnate HFO in a PLE. Since in this case the selective uptake of P would be due to the cumulative effects of bonding of HPO42 or H2PO4 with Cu2þ and HFO, it is imperative to quantify the effect of each. This article presents results of a comparative study of three phosphate-selective polymeric exchangers: HAIX, which consists of impregnated HFO nanoparticles in a strong-base anion-exchange polymer; DOW-HFO, which is a PLE loaded with HFO nanoparticles; and DOW-HFO-Cu, a PLE loaded with Cu2þ and then impregnated with HFO nanoparticles. The primary objectives of this study are: To compare the performance (bed volumes at breakthrough of P) of the three above-listed exchangers for a typical postsecondary WWTP effluent; To compare the regeneration performance of the three above-listed exchangers after exhaustion with a typical post-secondary WWTP effluent; To postulate an equilibrium and kinetic model of phosphate uptake for one phosphate-selective exchanger, viz, HAIX.
2.
Materials and experimental methods
2.1.
Materials
DOWEX M4195 (Fig. 1a) was obtained from SigmaeAldrich Co. 10.0 g of the dried virgin resin was added to 1.0 L of 1% solution of FeCl3$6H2O (2.0 g/L as Fe) at pH z 2.0 (to ensure that Fe(OH)3 does not precipitate) and the mixture was vigorously stirred for an hour. The pH of the solution was then stepwise increased to z8 uniformly paced over 3 h, by drop wise addition of 1.0 N sodium hydroxide [NaOH] solution to form the FeOOH precipitate. The mixing speed was reduced for the next 24 h to optimize the contact of the resin beads with freshly formed HFO nanoparticles. After 24 h, the resultant solution was repeatedly rinsed with deionized water to remove excess precipitate. The resin was then oven dried at 45 2 C for 24 h and stored in moisture free glass vials. The average HFO impregnation was found to vary between 45 and 60 mg as Fe/g of resin. This resin would be referred to as DOWHFO henceforth for typographical convenience. Another batch of DOW-HFO was loaded with Cu2þ based on the method presented by Zhao and SenGupta (1996). The average HFO impregnation ranged from 40 mg Fe/g to 50 mg Fe/g of dry original resin, and Cu(II) loading was z60 mg/g of dry original resin. This material would henceforth be called DOW-HFO-Cu. The development of HAIX has been detailed in many publications (Cumbal and SenGupta, 2005; Puttamraju and SenGupta, 2006; Blaney et al., 2007). For this study HAIX (brand LayneRT) was obtained from Solmetex Co. in Massachusetts (www.solmetex.com); however, no commercial endorsement is implied. The salient properties of the different resins used for this study are listed in Table 1.
2.2.
Batch equilibrium tests
Batch equilibrium tests were carried out in batch reactors (1.0 L or 2.0 L glass beakers) by adding known mass of the exchanger, ranging from 0.5 g to 1.0 g, into a known volume of a synthetic solution. Na2HPO4 was used as the P source. P concentration ranged from 5.0 to 75.0 (mg/L as P). The tests were conducted at pH of 5.0 or 8.0 to pH 8.0 to assess the effect of pH on the uptake of orthophosphate phosphorus. To evaluate equilibrium P removal capacity of the three resins (HAIX, DOW-HFO and DOW-HFO-Cu), tests were carried out at room temperature (20 2 C). The reactors were agitated with magnetic stirrer at 400 rpm for 48 h to ensure proper mixing and attainment of equilibrium. After equilibrium was attained, the exchanger was filtered out, rinsed with deionized water and stirred with a known volume (25 mL) of regenerant solution containing 2.5% NaCl and 2.0% NaOH.
2.3.
Batch kinetic tests
Kinetic tests were carried out in batch reactors (1.0 L or 2.0 L glass beakers) by adding a predetermined mass of the exchanger ranging from 0.5 g to 1.0 g into 1.0 or 2.0 L of solution at pH 8.0 containing different initial concentrations of orthophosphate phosphorus ranging from 10.0 mg to 45.0 mg as P/L. To ensure the attainment of uniform soluteesorbent
3321
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 1 8 e3 3 3 0
Table 1 e Salient Properties of Polymeric Ion Exchangers Used. Characteristics Structure Appearance Functional group Iron content Bulk density Particle size Ion-exchange capacity
HAIX
DOW-HFO
DOW-HFO-Cu
Macroporous PolystyreneDivinylbenzene Brown spherical beads Quaternary ammonium (see Fig. 1a) 75e90 mg as Fe/g resin 790e840 g/L 300e1200 mm 1 meq/mL
Macroporous PolystyreneDivinylbenzene Tan to dark brown opaque beads Bisepicolylamine (see Fig. 1b) 45e60 mg as Fe/g resin 673 g/L 297e841 mm None
Macroporous PolystyreneDivinylbenzene Tan to dark brown opaque beads Bisepicolylamine (see Fig. 1c) 40e50 mg as Fe/g resin 673 g/L 297e841 mm 1 meq/mL
contact and to eliminate the diffusion resistance in the liquid film (i.e., the Nernst film), the solution was agitated on a magnetic stirrer at 400 rpm. Thus, intra-particle diffusion was the rate-limiting step under the experimental conditions, and the intra-particle diffusivity ðDÞ was determined. The diameter of each exchanger particle was obtained from the parent resin supplier’s brochure. The average particle size for HAIX was 750 mm (0.75 mm); for DOW-HFO and DOW-HFO-Cu it was 570 mm (0.57 mm). Samples (6 ml each) were collected at every 30 min interval for 6.0 h and then again after 20.0, 24.0 and 48.0 h. Uptake of orthophosphate phosphorus species or other solutes by the sorbents at different times were determined from mass balance calculations on the solution samples retrieved from the batch reactor.
2.4.
Fixed-bed column runs
Fixed-bed column runs were carried out in Adjusta Chrom #11 (Ace Glass Inc., Vineland NJ) glass columns 300 mm long with 10 mm inner diameter. The influent was pumped into the column in a downflow direction using synchronous pumps, FMI Lab Pump, Model QSY (Fluid Metering Inc., Syosset, NY). Effluent sample from the column was collected by a Spectra/ Chrom CF-1 Fraction Collector (Spectrum Chromatography). The flow was adjusted to maintain constant empty bed contact time (EBCT) for all the column runs at 3.0 min. An online pH meter was also connected in the system to monitor the effluent pH. Regeneration of the bed was performed similarly by passing the regenerant in a downflow direction and the spent regenerant samples were also collected in a similar fashion. The regenerated bed was rinsed with deionized water till the effluent pH was <8.5. The regenerated bed was then used for another exhaustion run.
2.5. Regeneration, recycling of spent regenerant and recovery of phosphate Based on earlier studies (Pandit, 2010), a 2.5% NaCl þ 2.0% NaOH solution was chosen as the regenerant for all the regeneration runs. The sorbent media was regenerated with z20 bed volumes (BVs) of the regenerant solution. Following analysis of orthophosphate phosphorus in the spent regenerant samples, Ca(NO3)2$4H2O or a mixture of MgSO4$7H2O and NH4Cl was added to the spent regenerant to precipitate Ca3(PO4)2 or Mg(NH4)PO4 (struvite), the latter being a slow-release fertilizer. The precipitate was allowed to settle for a couple of hours and the supernatant was filtered through 0.45 mm filter and analyzed for different solutes. The precipitate was mildly
rinsed with deionized water, dried in an oven at 85 C, and elemental analysis using Energy Dispersive X-Ray (EDS) was performed on it using an Oxford Instruments, Inc. (High Wycombe, United Kingdom) INCA Energy EDX system in conjunction with a Scanning Electron Microscope (JSM 5610 from JEOL, Inc. Peabody, MA) at an accelerating voltage of 15 kV. Following the precipitation of orthophosphate phosphorus from the spent regenerant, 1.5% NaOH was added to it to compensate for the loss in OH ions in the regenerant due to regeneration of the media and possible precipitation and this mixture was again used as a regenerant for the next cycle.
2.6.
Analytical procedures
All the chemicals were analyzed as per the Standard Methods of Water and Wastewater Treatment (APHA et al., 1998). All anions such as orthophosphate phosphorus, Cl, NO3, & SO42 were analyzed using a DIONEX Ion Chromatograph (Model e ICS 900) coupled with an AS40 autosampler. Dissolved organic carbon (DOC) was measured using a total organic carbon (TOC) analyzer (Teledyne Tekmar Phoenix 8000 UVPersulfate TOC Analyzer). The iron content in the resins was determined using a HACH spectrophotometer (Model e DV 4000 UV/Vis) by the FerroVer method. Before analysis the samples were digested as per the Standard Methods of Water and Wastewater Treatment (APHA et al., 1998).
3.
Results
3.1.
Fixed-bed column run: exhaustion
Fig. 2 shows the P breakthrough curve for the three anion exchangers studied in this study when the influent synthetic solution was identical. The SO42 and P concentration being 130.0 & 4.25 mg/L respectively, this is a case where P is a trace species. P breakthrough is the earliest for DOW-HFO. Since this is a free-form resin, it has no exchange capacity. Therefore, the only mechanism of P removal is the Lewis Acid Base (LAB) innersphere complex formed between HPO42 (since the pH is 8) and FeOH2þ. The P breakthrough curve for HAIX closely follows that of DOW-HFO. In the case of HAIX, P removal can occur by ion exchange (of the preloaded Cl with HPO42) and LAB between HPO42 and FeOH2þ. The following affinity sequence is generally obtained for strong-base anion exchange resins (the parent resin of HAIX) and has been demonstrated in previous works (Zhao and SenGupta, 1996; Blaney et al., 2007): SO42 > PO43 > NO3 > Br > NO2 > Cl.
3322
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 1 8 e3 3 3 0
Fig. 2 e Comparative breakthrough profile of P with different sorbents.
Thus, for the case of anion exchange where SO42 is present at a much higher concentration than P, it provides stronger Coulombic Interaction (CI) and almost completely prevails over P for the preloaded Cl sites. P removal is almost completely due to LAB interaction between HPO42 and FeOH2þ. The role of SO42 vis-a`-vis P being a trace species is discussed in the following section, but it will suffice to state here that competition by SO42 does not impair the performance of HAIX. For the third exchanger, DOW-HFO-Cu, the
capacity of the sorbent media is enhanced and the breakpoint of P is observed much later than (about 200 BVs) HAIX or DOWHFO. In this particular scenario, P was removed by LAB interaction with both the dispersed HFO nanoparticles and the Cu2þ ion immobilized on the surface. For the Cu2þ sites on the exchanger, LAB þ CI for HPO42 is higher than CI alone for SO42; therefore, SO42 competition is avoided and this exchanger provides the highest P uptake. These experimental data validate that if two different ligand exchangers can
Fig. 3 e Comparison of breakthrough profiles of P for HAIX for different background concentration of SO42- (P is present as trace species compared with SO42-).
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 1 8 e3 3 3 0
3323
Fig. 4 e Comparative breakthrough profile of P for HAIX at low and high background concentration of SO42-.
successfully be hosted on a polymeric backbone, increased sorption capacity is attained with little or no compromise on the individual capacity of the ligand exchangers. The role played by SO42 concentration in the case of P removal by HAIX was further evaluated. Fig. 3 provides the P breakthrough curve for a HAIX column that was fed an influent that varied only in SO42 concentration (161.6 and 245.9 mg/L). Since the P: SO42 is 40.4 or 61.5, P can be considered as a trace species in both the cases. Fig. 3 shows that the P breakthrough behavior remained the same for both
the cases and it can be inferred that as long as P is a trace species compared to SO42 (that is ubiquitous in wastewater effluents), P uptake capacity of HAIX is not affected by SO42 concentration. The typical SO42 and P concentration in the secondary effluent of a municipal wastewater treatment plant being in this range, little or no competition is expected from the increase in concentration of SO42. However, when the fixed-bed column was operated with a much lower background concentration of SO42, at P: SO42 ¼ 1.0: 5.0, there was significant improvement in the
Fig. 5 e CEffluent/CInfluent graph of P and SO42- for HAIX at low background concentration of SO42-.
3324
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 1 8 e3 3 3 0
Fig. 6 e a: Typical regeneration profile of P for HAIX. b: Typical regeneration profile of P for DOW-HFO. c: Typical regeneration profile of P for DOW-HFO-Cu.
3325
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 1 8 e3 3 3 0
performance of the column. The breakthrough BV for P almost doubled, even though there was some competition from the SO42, which is evident from its later breakpoint in the effluent graph and chromatographic elution of P. Figs. 4 and 5 illustrate the P breakthrough behaviors obtained from the HAIX- loaded fixed-bed column runs for high and low background concentration of SO42. The sudden drop in P concentration at w700 BV as observed in Fig. 4 was due to an interruption of flow, and the nature of the curve is further elucidated in Section 4.4. In all the fixed-bed column runs performed, the pump used to drive the inflow through the fixed-bed needed to be turned off (once or more, depending on the length of the run) for a brief period to prevent overheating. This observation was consistent in all cases. However, this interruption test had no effect on the ultimate removal capacity. The removal of P is attributed to Coulombic Interaction (CI) þ Lewis Acid-base (LAB) Interaction, with the latter being the major player. Whereas the removal of SO42 by HAIX is purely due to CI and LAB interaction has no role to play. Thus, when the background concentration of SO42 is very high, the ion exchange sites get exhausted rapidly with SO42 and its breakthrough occurs almost instantaneously. P being present as a trace species does not pose a competition to the SO42 for CI removal, but gets exchanged due to LAB interaction with HFO till the exhaustion of its (HFO) capacity. On the other hand, when the background concentration of SO42 is low, both P and SO42 compete for the ion exchange sites, in addition to P removal by LAB interaction. Since SO42 is more preferred by the CI exchange mechanism, P undergoes a brief chromatographic elution before the breakpoint of SO42, where SO42 replaces P from the exchange sites and gets removed from the solution phase. However, P removal due to LAB interaction is not compromised at any stage. Hence, it can be inferred that, P removal by HAIX is attributed to both CI and LAB exchange mechanism at low background concentration of SO42, while at high background concentration of SO42 it is entirely attributable to LAB interaction and is independent of SO42 concentration.
3.2. Fixed-bed column run: regeneration, recovery, and recycling Excellent regeneration of the exchanger was observed in all cases. Fig. 6aec shows the regeneration profile for the three exchangers studied for a regenerant of 2.5% NaCl þ 2% NaOH. Recovery of more than 90.0% of the sorbed P was attained within 10 BVs in all cases, even after 10 cycles of exhaustion and regeneration run. Desorption of the SO42 was also remarkable and >90.0 5.00% of the sorbed SO42 was desorbed within 15 BVs in all cases. The mass of SO42 in the exchanger in increasing order of magnitude was DOWHFO < DOW-HFO-Cu < HAIX, as may be expected from the earlier discussion. Also, there was no compromise of the performance of the regenerated resin with that of the virgin resin. A decrease of only about 1.5% in the breakpoint of P was observed after 10 cycles of exhaustion-regeneration of the same resin bed. After addition of Ca(NO3)2$4H2O or a mixture of MgSO4$7H2O and NH4Cl to precipitate the P in the regenerant solution as Ca3(PO4)2 or Mg(NH4)PO4 (struvite) respectively, >90% of the orthophosphate phosphorus present was
recovered as a high-purity fertilizer. Table 2 shows the comparative elemental distribution of the recovered fertilizer and the Sigma standard. Fig. 7a & b shows the EDS peaks for the appropriate elements. It can be inferred that the fertilizers obtained from the spent regenerant do not contain any impurity except for z3.6 wt.% of NaCl in the Ca3(PO4)2. The benefits of this approach compared to conventional P precipitation by adding a Ca or Al salt are discussed in the next section. The phosphate-free spent regenerant was then recycled after necessary adjustments as elaborated in Section 2.5, and the regeneration performance was observed to be close to that of the virgin regenerant solution (Fig. 8). The slightly high retention of phosphate by the resin in case of the recycled regenerant may be attributed to the presence of trace concentration of phosphate in the recycled regenerant as 100% recovery was not achieved. However, the regeneration efficiency was greater than 90% even after 3e4 recycling cycles.
4.
Discussion
4.1.
Phosphorus recovery
Chemical precipitation of phosphorus in wastewater by addition of aluminum or ferric salt is a conventional process for removal of phosphorus but many studies (Takacs et al., 2006; Szabo et al., 2008; Banu et al., 2008) have shown the critical role of pH and alkalininity, and the process may need addition of an external buffer to raise the pH/alkalinity before attempting precipitation of Al(or Fe) salt. Moreover, it is clear (Debarbadillo et al., 2010; Esvelt et al., 2010) that to achieve effluent P in the range of <50 mg/L, the ratio of moles of Al(or Fe) required/mole of P removed is >>1, and may reach 200. The technique employed here used 1 mol of Ca (or Mg)/mole of P removed. The main features of the chemical addition step are: (1) the regenerant solution already is highly alkaline (pH > 12), and (2) the regenerant solution volume is much lower (1/20 of the volume of wastewater treated). Moreover, after filtering the Ca(or Mg) salt from the regenerant, the regenerant solution is reused for the next cycle, as explained in Section 2.5. Thus, although the process described in this
Table 2 e EDS Analysis: Elemental Composition. Fertilizer from Regenerant Solution Element
Weight %
Atomic %
Standard (Sigma) Weight %
Atomic %
I e Calcium Phosphate Ca3(PO4)2 Oxygen 49.35 68.96 53.57 72.87 Sodium 1.82 1.71 e e Phosphorus 17.74 12.36 19.60 13.18 Chlorine 3.11 1.89 e e Calcium 27.99 15.07 26.82 13.94 II e Magnesium Ammonium Phosphate Mg(NH4)PO4 e Struvite Oxygen 66.19 75.56 56.67 65.92 Phosphorus 16.99 10.02 22.12 13.29 Magnesium 13.60 10.21 13.12 10.04 Nitrogen 3.23 4.21 8.10 10.76
3326
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 1 8 e3 3 3 0
Fig. 7 e a: EDS analysis of solid-phase fertilizer obtained from regenerant by adding Ca(NO3)2.4H2O. See Table 2 for elemental weight % of constituents. b: EDS analysis of Ca3(PO4)2 standard from Sigma Chemicals. See Table 2 for elemental weight % of constituents.
article uses chemicals, the quantity needed is much lower than alternative chemical precipitation processes, and the solution volume on which the chemicals are applied is only about 5% of the wastewater treated.
4.2.
Sorption isotherm and ultimate sorption capacity
A Freundlich isotherm fit of the experimental data yielded the following equation: qA ¼ 4:754 C0:2975 A
(1)
where qA is in mg P/g exchanger and CA is in mg/L as P. Since the equilibrium tests were carried out at pH ¼ 8.0, the ligands formed with HFO nanoparticles were more probably a combination of both monodentate and bidentate ligands (though the percentage of monodentate ligands is expected to be small). However, previous studies (Golterman, 1995) have hypothesized that for every molecule of H2PO4 or HPO42 adsorbed, two FeOOH sites become inactive. From the equilibrium calculations, the following equation was proposed which is identical to Freundlich isotherm (Golterman, 1995):
Pads ¼ A ðo PÞ0:333
(2)
where, Pads ¼ mass of phosphate adsorbed onto FeOOH (mg/g); A ¼ adsorption constant; o-P ¼ concentration of orthophosphate species in equilibrium (mg/L). It is noteworthy that the Freundlich isotherm fit resulted in 1/n ¼ 0.2975, which is very close to Eq. (2). This substantiates that the ligand formation mechanism in this study was analogous to the hypothesis proposed by Golterman (1995). A probable stoichiometric equation scheme for sorption of H2PO4/HPO42 onto HFO can be proposed as:
2 þ 2 2FeOHþ 2 Cl þ HPO4 #2FeOH2 HPO4 þ 2Cl
(3)
þ þ 2FeOHþ 2 Cl þH2 PO4 #FeOH2 HPO4 þjFeOH2 Cl jinactive
þCl
(4)
2 2FeOH þ HPO2 4 #2FeOH HPO4
(5)
2FeOH þ H2 PO 4 #FeOH H2 PO4 þ jFeOHjinactive
(6)
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 1 8 e3 3 3 0
3327
Fig. 8 e Comparison of the regeneration efficiency of the fresh and recycled regenerant for HAIX column.
The hypothesis that two molecules of HFO are inactivated per molecule of phosphate (be it monodentate or bidentate) as expressed in the above reactions also holds good in determination of the ultimate sorption capacity of the sorbent. The maximum sorption capacity as observed experimentally was 23.0 mg P/g sorbent at both pH of 5.0 and 9.0 with H2PO4 and HPO42 being the predominant species respectively. According to the stoichiometric equations proposed above, for 80 mg Fe/g sorbent the maximum capacity would have been about 21.5 mg P/g sorbent, which is in excellent agreement with the experimental observation. On the other hand, had the H2PO4 formed a monodentate ligand inactivating only one site, the theoretical capacity should have been z43 mg P/g sorbent, which was not observed at any pH.
4.3. Quantifying the Relative contribution of different mechanisms in phosphate sorption Conventional strong-base anion exchange resins have poor P selectivity compared to SO42. Liberti et al. (1981) report the aP/S for different SBA resins; the maximum value is 0.25. The separation factor or a can be defined as ‘the preference of the ion exchanger for one of the two counterions’ (Helfferich, 1995). It is determined as the quotient of the ratios of the two counterions in the ion-exchanger and the solution and is expressed as aA/B to denote the preference of A over B. If the ratio is >1, Then A is the preferred ion, while if the ratio is <1, B is the preferred ion. As opposed to the aP/ S values for conventional SBA resins reported by Liberti et al. (1981), the aP/S for HAIX was computed to be 46.0 This wide difference can only be attributed to the hypothesis that HAIX binds to P through electrostatic and Lewis acid-base interaction whereas conventional SBA interaction with P is
only electrostatic. Since the experiments to determine aP/S in HAIX were performed at a pH z 8, the electrostatic interaction between HPO42 (the predominant P species) and SO42 is almost equal since both are divalent anions. Though calculations are not included due to page limitation, we have estimated the Lewis acid-base interaction to be of the order of 3 kcal/mol. This result is in excellent agreement with previous research by Miltenburg and Golterman (1998).
4.4.
Kinetic modeling of sorption
Fixed-bed sorption mechanism is often governed by the kinetics of the sorption process. Generally, the process is controlled by the mass transfer (intra-particle and/or film diffusion mass transfer) rate. From the results of interruption test it was inferred that intra-particle diffusion was the rate limiting step in the sorbent’s sorption of phosphate. This kinetic behavior has been validated for selective sorption of trace solutes in absence of chemical reactions by many researchers. Similar interruption data and interpretation in favor of such an inference for selective ion exchangers and sorbents are available in open literature (Helfferich, 1995; Gjerde and Fritz, 1978). Fig. 9 shows the breakthrough profile emphasizing the interruption test. During one of the fixed-bed column run experiment with HAIX, the flow of the influent was intentionally stopped after 392 bed volumes with the effluent phosphate concentration of 3.117 mg P/L, for approximately 3.0 h. After that, the original flow condition was restored and the effluent phosphate concentration reached 3.25 mg P/L at 416 bed volume, i.e., within 24 bed volumes (as shown in Fig. 9).
3328
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 1 8 e3 3 3 0
Fig. 9 e Breakthrough profile of phosphate showing interruption test.
4.4.1.
Determination of intra-particle diffusivity
Results of batch kinetic study for phosphate-chloride binary system were studied to determine the intra-particle diffusivity, D. Restricting the discussion to cases where the diffusion is radial, the equation for a constant diffusion co-efficient can be expressed as: 2 vC v C vC þ ¼D vt vr2 vr
(7)
where, D ¼ Intra-particle diffusivity; C ¼ concentration; r ¼ radial coordinate (distance from the bead center); and t ¼ time. On substituting, u ¼ Cr, where u is the mass flux entering the spherical bead, we get, vu v u ¼D 2 vt vr 2
Fig. 10 e Intra-particle diffusion model plot with experimental data.
Assuming, phosphate to be a trace species under experimental conditions, for a well-stirred solution of limited volume the exact analytical solution under suitable initial and boundary conditions can be written as (Crank, 1975): Dq2n t 2 X N 6aða þ 1Þe a Mt ¼ 1 2 n¼0 MN 9 þ 9a þ qn a2
(9)
where, Mt ¼ sorbent phase concentration of phosphate at time t ¼ V(CP,0 CP,t); MN ¼ sorbent phase concentration of phosphate at equilibrium ¼ V(CP,0 CP,N); a ¼ radius of the bead; and, the qns are the non-zero roots of 3qn tan qn ¼ (10) 3 þ aq2n The parameter a is expressed in terms of the final fractional uptake of the solute by the sphere by the relation,
(8) MN 1 ¼ VC0 1 þ a
(11)
where, V ¼ batch volume, and C0 ¼ initial aqueous phase concentration of phosphate. The batch kinetic tests were carried out for three different concentrations of phosphate, namely 13.01 mg P/L, 22.58 mg P/L and 44.22 mg P/L. The average value of a obtained was 1.661. The values of qns were interpolated from the values provided in open literature (Crank, 1975). The first six values of qn were only considered as it converges within that range. The other terms being known, the intra-particle diffusivity was obtained by fitting the batch kinetic data to Eq. (9) as shown in Fig. 10. The mean effective intra-particle diffusivity was found to be 5.278 1010 cm2/sec (with standard deviation ¼ 0.011 1010 cm2/sec). Excellent fit was observed for all three initial concentrations of phosphate. The solid-line in the figure shows the best-fit line. The observed values were a magnitude
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 1 8 e3 3 3 0
3329
higher than phosphate exchange in OH-type strong basic anion exchanger (Galinada and Yoshida, 2004)
Acknowledgments
5.
This project received partial funding through the National Science Foundation Partnerships for Innovation Program (IIP-0650163).
Conclusions
Selective removal of trace phosphate from ubiquitous background presence of other competing anions present in wastewater is a major challenge faced by the industry. Chemical precipitation is limited by the production of excessive sludge and biological removal processes are restricted by the inability to remove trace concentration. Also, both these processes suffer from inadequate recovery of the removed phosphate, which is a non-renewable resource. The overall goal of this study was to estimate the removal capacity of phosphate by three iron-oxide impregnated polymers, HAIX, DOW-HFO, and DOW-HFO-Cu, at different background concentration of competing anions and to assess recovery of the phosphate as a solid-phase fertilizer. Laboratory experiments were carried out to investigate the key features of the fixed-bed sorption process, which is capable of removing trace concentrations of orthophosphate phosphorus, to recover the P and to reuse the spent regenerant to minimize the production of any solid or liquid stream. Theoretical analysis and experimental evidence were also provided to demonstrate the underlying principles of sorption and kinetics of the process. The major conclusions that can be drawn from this study are synopsized as follows: 1. HAIX, DOW-HFO, & DOW-HFO-Cu showed high selectivity towards orthophosphate phosphorus when compared to competing anions, especially sulfate. 2. Lewis acid-base interaction (i.e. formation of coordination bond between the anionic ligand and the central metal atom forming inner-sphere complexes) accompanied by the electrostatic attraction (i.e. ion-pair formation) is the core mechanism leading to the high sorption affinity for HAIX and DOW-HFO-Cu, whereas for DOW-HFO Lewis acid-base interaction is the only mechanism of orthophosphate phosphorus sorption. 3. All three materials studied, HAIX, DOW-HFO, and DOWHFO-Cu, are amenable to efficient regeneration. Single step regeneration with 2.5% sodium chloride and 2.0% sodium hydroxide consistently recovered more than 95.0% of sorbed orthophosphate phosphorus within 10 bed volumes. Only minor capacity drop (about 1.5%) was observed after 10 cycles of exhaustion-regeneration. 4. The spent regenerant may be reused after supplementing for the hydroxide lost in regeneration. The regeneration efficiency was unaltered compared to virgin regenerant. 5. Orthophosphate phosphorus can be recovered from the spent regenerant as calcium phosphate or magnesium ammonium phosphate (struvite) upon addition of calcium nitrate or a combination of ammonium chloride and magnesium sulfate, respectively. 6. No significant bleeding of the iron (from HAIX) or copper (from DOW-HFO-Cu) was found in the regenerant. 7. Intra-particle diffusion is the rate limiting step for orthophosphate phosphorus sorption by HAIX.
references
APHAAWWAWEF, 1998. Standard Methods for the Examination of Water and Wastewater, twentieth ed.. American Public Health Association, American Water Works Association, Water Environment Federation, Baltimore, MD. Banu, R.J., Do, K.U., Yeom, I.T., 2008. Phosphorus removal in low alkalinity secondary effluent using alum. International Journal of Environment Science and Technology 5 (1), 93e98. Blaney, L.M., Cinar, S., SenGupta, A.K., 2007. Hybrid anion exchanger for trace phosphate removal from water and wastewater. Water Research 41 (7), 1603e1613. Chubar, N.I., Kanibolotskyy, V.A., Strelko, V.V., Gallios, G.G., Samanidou, V.F., Shaposhnikova, T.O., 2005. Adsorption of phosphate ions on novel inorganic ion exchangers. Colloids and Surfaces A: Physicochemical and Engineering Aspects 255 (1e3), 55e63. Cooper, P., Dee, T., Yang, G., 1993. Nutrient removal-methods of meeting the EC urban wastewater directive. In: Paper Presented at the Fourth Annual Conference on Industrial Wastewater Treatment, Esher, Surrey, March 10, 1993. Crank, J., 1975. The Mathematics of Diffusion, second ed. Oxford University Press, Oxford. Cumbal, L., SenGupta, A.K., 2005. Arsenic removal using polymersupported hydrated iron (III) oxide nanoparticles: role of donnan membrane effect. Environmental Science and Technology 39 (17), 6508e6515. Cumbal, L., Greenleaf, J., Leun, D., SenGupta, A.K., 2003. Polymer supported inorganic nanoparticles: characterization and environmental applications. Reactive and Functional Polymers 54 (1e3), 167e180. Debarbadillo, C., Shellswell, G., Cyr, W., Edwards, B., Waite, R., Sabherwal, B., Mullan, J., Mitchell, R., 2010. Development of full-scale sizing criteria from tertiary pilot testing results to achieve ultra-low phosphorus limits at innisfil, Ontario. In: Proceedings of the 83rd Annual Water Environment Federation Technical Exhibition and Conference, New Orleans, Louisiana, October, pp. 2e6. DeMarco, M.J., SenGupta, A.K., Greenleaf, J.E., 2003. Arsenic removal using a polymeric/inorganic hybrid sorbent. Water Research 37 (1), 164e176. Dutta, P.K., Ray, A.K., Sharma, V.K., Millero, F.J., 2004. Adsorption of arsenate and arsenite on titanium dioxide suspensions. Journal of Colloid and Interface Science 278 (2), 270e275. Dzombak, D.A., Morel, F.M., 1990. Surface Complexation Modeling: Hydrous Ferric Oxide. John Wiley & Sons, Inc, Hoboken, NJ. Esvelt, L., Esvelt, M., Walker, B., Hendron, L., 2010. Pilot studies for reducing RPWRF effluent TP for discharge to the Spokane river. In: Proceedings of the 83rd Annual Water Environment Federation Technical Exhibition and Conference, New Orleans, Louisiana, October, pp. 2e6. Florida Everglades Forever Act, 1994. Florida State Legislature (Tallahassee, Florida). Galinada, W.A., Yoshida, H., 2004. Intraparticle diffusion of phosphates in OH-type strongly basic ion exchanger. AIChE Journal 50 (11), 2806e2815.
3330
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 1 8 e3 3 3 0
Gao, Y.-M., Sengupta, A.K., Simpson, D., 1995. A new hybrid inorganic sorbent for heavy metals removal. Water Research 29 (9), 2195e2205. Genz, A., Kornmu¨ller, A., Jekel, M., 2004. Advanced phosphorus removal from membrane filtrates by adsorption on activated aluminium oxide and granulated ferric hydroxide. Water Research 38 (16), 3523e3530. Gjerde, D.T., Fritz, J.S., 1978. Ion Chromatography. A. Hu¨thig, Heidelberg/New York, Germany/USA. Glueckauf, E., Watts, R.E., 1962. The Donnan law and its application to ion exchange polymers. Proceedings of the Royal Society of London, Series A: Mathematical and Physical Sciences 268 (1334), 339e349. Golterman, H., 1995. Theoretical aspects of the adsorption of ortho-phosphate onto iron-hydroxide. Hydrobiologia 315, 59e68. Hansen, B., 2006. Long-term plan seeks to reduce phosphorus in Spokane river. Civil Engineering 76 (10), 24e25. Heathwaite, L., Sharpley, A., 1999. Evaluating measures to control the impact of agricultural phosphorus on water quality. Water Science and Technology 39, 149e155. Helfferich, F.G., 1995. Ion Exchange. Dover Publications Inc, Mineola, NY. Herring, J.R., Fantel, R.J., 1993. Phosphate rock demand into the next century: impact on world food supply. Natural Resources Research, 226e246. Hiemstra, T., van Riemsdjik, W., 1999. Surface structural ion adsorption modeling of competitive binding of oxyanions by metal (hydr)oxides. Journal of Colloid and Interface Science 210 (1), 182e193. Hristovski, K., Westerhoff, P., Mo¨ller, T., Sylvester, P., Condit, W., Mash, H., 2008. Simultaneous removal of perchlorate and arsenate by ion-exchange media modified with nanostructured iron (hydr)oxide. Journal of Hazardous Materials 152 (1), 397e406. Huang, C.P., Vane, L.M., 1989. Enhancing As5þ removal by a Fe2þtreated activated carbon. Journal of Water Pollution Control Federation 61, 1596e1603. Jenkins, O., Fergusson, J.F., Menar, A.B., 1971. Chemical processes for phosphate removal. Water Research 5 (7), 369e387. Jenkins, D., Hermanowicz, S.W., 1991. Principles of chemical phosphorus removal. In: Sedlak, R.I. (Ed.), Phosphorus and Nitrogen Removal from Municipal Wastewater: Principles and Practice, second ed. Lewis Publishers, Chelsea, MI. Kang, S.K., Choo, K.-H., Lim, K.-H., 2003. Use of iron oxide particles as adsorbents to enhance phosphorus removal from secondary wastewater effluent. Separation Science and Technology 38, 3853e3874. Katsoyiannis, I.A., Zouboulis, A.I., 2002. Removal of arsenic from contaminated water sources by sorption onto iron-oxidecoated polymeric materials. Water Research 36 (20), 5141e5155. Kuba, T., Smolders, G.J.F., van Loosdrecht, M.C.M., Heijnen, J.J., 1993. Biological phosphorus removal from wastewater by anaerobic-anoxic sequencing batch reactor. Water Science and Technology 27 (5e6), 241e252. Liberti, L., Boari, G., Petruzzelli, D., Passino, R., 1981. Nutrient removal and recovery from wastewater by ion exchange. Water Research 15 (3), 337e342. Mezenner, N.Y., Bensmaili, A., 2009. Kinetics and thermodynamic study of phosphate adsorption on iron hydroxide-eggshell waste. Chemical Engineering Journal 147 (2e3), 87e96. Miltenburg, J., Golterman, H., 1998. The energy of the adsorption of o-phosphate onto ferric hydroxide. Hydrobiologia 36, 93e97. Onyango, M.S., Matsuda, H., Ogada, T., 2003. Sorption kinetics of arsenic onto iron-conditioned zeolite. Journal of Chemical Engineering of Japan 36 (4), 477e485.
Pandit, A., 2010. Selective removal and recovery of phosphate from wastewater, University of Massachusetts Dartmouth, MS thesis. Puttamraju, P., SenGupta, A.K., 2006. Evidence of tunable on-off sorption behaviors of metal oxide nanoparticles: role of ion exchanger support. Industrial and Engineering Chemistry Research 45 (22), 7737e7742. Sedlak, R. (Ed.), 1991. Phosphorus and Nitrogen Removal from Municipal Wastewater: Principles and Practice. Lewis Publishers. Seida, Y., Nakano, Y., 2002. Removal of phosphate by layered double hydroxides containing iron. Water Research 36 (5), 1306e1312. Seidel, A., Waypa, J.J., Elimelech, M., 2001. Role of charge (Donnan) exclusion in removal of arsenic from water by a negatively charged porous nanofiltration membrane. Environmental Engineering Science 18 (2), 105e113. Seviour, R.J., Mino, T., Onuki, M., 2003. The microbiology of biological phosphorus removal in activated sludge systems. FEMS Microbiology Reviews 27, 99e127. Sharpley, A.N., Chapra, S.C., Wedepohl, R., Sims, J.T., Daniel, T.C., Reddy, K.R., 1994. Managing agricultural phosphorus for protection of surface waters: issues and options. Journal of Environmental Quality 23, 437e451. Sharpley, A.N., Daniel, T., Sims, T., Lemunyon, J., Stevens, R., Parry, R., 2003. Agricultural Phosphorus and Eutrophication, second ed. United States Department of Agriculture, Agricultural Research Service. Snoeyink, V.L., Jenkins, D., 1980. Water Chemistry. John Wiley & Sons, New York. Stumm, W., Morgan, J.J., 1995. Aquatic Chemistry: Chemical Equilibria and Rates in Natural Waters, third ed. John Wiley & Sons, New York. Suzuki, T.M., Bomani, J.O., Matsunaga, H., Yokoyama, Y., 2000. Preparation of porous resin loaded with crystalline hydrous zirconium oxide and its application to the removal of arsenic. Reactive and Functional Polymers 43 (1e2), 165e172. Szabo, A., Takacs, I., Murthy, S., Daigger, G.T., Liksco, I., Smith, S., 2008. Significance of design and operational variables in chemical phosphorus removal. Water Environment Research 80 (5), 407e416. Takacs, I., Murthy, S., Smith, S., McGrath, M., 2006. Chemical phosphorus removal to extremely low levels: experience of two plants in the Washington, DC area. Water Science and Technology 53 (12), 21e28. Tanada, S., Kabayama, M., Kawasaki, N., Sakiyama, T., Nakamura, T., Araki, M., 2003. Removal of phosphate by aluminum oxide hydroxide. Journal of Colloid and Interface Science 257 (1), 135e140. USEPA., 1997. Chesapeake Bay Nutrient Reduction Program and Future Directions. USEPA Chesapeake Bay Program, Anapolis, MD. USEPA., 2007. Advanced wastewater treatment to achieve low concentration of phosphorus. EPA 910-R-07e002. USGS, 1999. Phosphorus in a Ground-Water Contaminant Plume Discharging to Ashumet Pond, Cape Cod, Massachusetts. USGS, Northborough, MA. Yuchi, A., Ogiso, A., Muranaka, S., Niwa, T., 2003. Preconcentration of phosphate and arsenate at sub ng ml1 level with a chelating polymer-gel loaded with zirconium (IV). Analytica Chimica Acta 494 (1e2), 81e86. Zeng, L., Li, X., Liu, J., 2004. Adsorptive removal of phosphate from aqueous solutions using iron oxide tailings. Water Research 38 (5), 1318e1326. Zhao, D., SenGupta, A.K., 1996. Selective removal and recovery of phosphate in a novel fixed-bed process. Water Science and Technology 33 (10e11), 139e147. Zhao, D., SenGupta, A.K., 1998. Ultimate removal of phosphate from wastewater using a new class of polymeric ion exchangers. Water Research 32 (5), 1613e1625.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 3 1 e3 3 4 0
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Examining the link between terrestrial and aquatic phosphorus speciation in a subtropical catchment: The role of selective erosion and transport of fine sediments during storm events Jason G. Kerr a,*, Michele A. Burford a, Jon M. Olley a, Stuart E. Bunn a, James Udy b a b
Australian Rivers Institute, Griffith University, Nathan, QLD, Australia Seqwater, 240 Margaret Street, Brisbane, QLD, Australia
article info
abstract
Article history:
This study examined the link between terrestrial and aquatic phosphorus (P) speciation in
Received 6 January 2011
the soils and sediments of a subtropical catchment. Specifically, the study aimed to
Received in revised form
identify the relative importance of P speciation in source soils, erosion and transport
22 March 2011
processes upstream, and aquatic transformation processes as determinants of P speciation
Accepted 23 March 2011
in lake sediments (Lake Wivenhoe). Using a sequential extraction technique, NH4Cl
Available online 31 March 2011
extractable P (NH4Cl-P; exchangeable P), bicarbonateedithionite extractable P (BD-P; reductant soluble P), NaOH extractable P (NaOH-rP; Al/Fe oxide P), HCl extractable P (HCl-P;
Keywords:
apatite-P), and residual-P (Res-P; organic and residual inorganic P) fractions were compared
Phosphorus speciation
in different soil/sediment compartments of the upper Brisbane River (UBR) catchment,
Sediment
Queensland, Australia. Multidimensional scaling identified two distinct groups of samples,
Soil
one consisting of lake sediments and suspended sediments, and another consisting of
Selective erosion
riverbed sediments and soils. The riverbed sediments and soils had significantly higher HCl-P and lower NaOH-rP and Res-P relative to the lake and suspended sediments (P < 0.05). Analysis of the enrichment factors (EFs) of soils and riverbed sediments showed that fine grained particles (<63 mm) were enriched in all but the HCl-P fraction. This indicated that as finer particles are eroded from the soil surface and transported downstream there is a preferential export of non-apatite P (NaOH-rP, NaOH-nrP, BD-P and Res-P). Therefore, due to the preferential erosion and transport of fine sediments, the lake sediments contained a higher proportion of more labile forms of inorganic-P relative to the broader soil/sediment system. Our results suggest that a greater focus on the effect of selective erosion and transport on sediment P speciation in lakes and reservoirs is needed to better target management strategies aimed at reducing P availability, particularly in P-limited water bodies impacted by soil erosion. ª 2011 Elsevier Ltd. All rights reserved.
* Corresponding author. Present address: Department of Geography, Trent University, 1600 West Bank Drive, Peterborough, Ontario K9J 7B8, Canada. Tel.: þ1 705 748 1011x7692; fax: þ1 705 748 1205. E-mail addresses:
[email protected] (J.G. Kerr),
[email protected] (M.A. Burford),
[email protected] (J.M. Olley),
[email protected] (S.E. Bunn),
[email protected] (J. Udy). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.03.048
3332
1.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 3 1 e3 3 4 0
Introduction
P is a major contributor to the eutrophication of freshwater ecosystems. While there has been a substantial amount of research on P in aquatic and terrestrial environments, there has been little integration of the two disciplines (Grimm et al., 2003). This is important because the effective management of P in lakes and reservoirs requires a whole of catchment approach. Our ability to predict the potential effectiveness of various mitigation strategies is currently constrained by a lack of knowledge about the link between terrestrial and aquatic P. In particular, while much is known about the total amount of P moving from the terrestrial to the aquatic parts of the catchment, there is little known about the potential availability of this P, particularly in the particulate fraction (Sharpley and Tunney, 2000). Furthermore, even less is known about the principle drivers of soil/sediment P speciation across a gradient from the source (soils) to the site of impact (e.g. lake sediment). Sediment is an important source of P in many lakes and reservoirs (Wetzel, 2001). As a result, reductions in P loadings to lakes have not always resulted in substantial improvements in trophic status due to internal P loading from the sediments (Golterman, 2001). The extent to which P is released from sediment will depend on the different forms of P present and the physiochemical properties at the sedimentewater interface (Pettersson and Istvanovics, 1988; Penn et al., 1995; Ruban et al., 1999). Therefore, there has been a substantial amount of research investigating the release of various forms of P from lake sediment (Bostro¨m et al., 1988; Istva´novics, ^ek et al., 2005; Lake 1994; Kleeberg and Dudel, 1997; Kopa´c et al., 2007). These studies have provided useful information on the potential for various P fractions to be released from sediments over the short to mid-term under varying conditions (e.g. lowered redox and changes in pH). However, the question of what drives P speciation in lake sediments has not been sufficiently addressed. In particular, there is little known about the importance of terrestrial P as a driver of P speciation in lake sediments. Broadly speaking, the speciation of P in lake sediments will be determined by the composition of the source material, and any physical, chemical or biological processes which might alter this composition prior to, or following, deposition within the lake. The underlying bedrock geology, soils and land use of the catchment can have a strong influence on the P content of suspended sediments (Walling et al., 2001; Ballantine et al., 2008). Terrestrial soils provide the parent material from which sediments are derived and therefore the speciation of P in lake sediments may be largely governed by the P speciation of the catchment soils. However, the selective erosion and transport of soils with different grain sizes in overland and stream flow may alter the composition of suspended sediments relative to the bulk source material. Selective erosion is a process whereby finer soils are transported more frequently and farther in overland and stream flows relative to coarser soils (Quinton et al., 2001; McDowell and Sharpley, 2003). In addition, physical winnowing of sediments within lakes can result in further sorting of sediment based on grain size and density (SanClements et al., 2009). If various P fractions are distributed unevenly between different grain sizes, selective erosion and
transportation processes could substantially alter the P speciation of suspended sediments relative to the source soils. Stone and English (1993) examined P speciation in suspended sediments of varying grain size in two Lake Erie, USA/ Canada tributaries and found that exchangeable P, Fe- and Al-bound P, and organic P increased with decreasing grain size. Conversely, the apatite inorganic-P concentration of the suspended sediments decreased with decreasing grain size (Stone and English, 1993). A study of suspended sediments from a hardwater stream in Switzerland found that the smallest sized particles (median diameter 6 mm) were enriched in organic P and exchangeable P (Pacini and Gatcher, 1999). However, unlike the results from Stone and English (1993) the finest suspended sediments were relatively poor in Al- and Fe-bound forms of P. While these studies indicate that different forms of P are not evenly distributed between different grain sizes, the results are contradictory in terms of which species were enriched. Furthermore, a study of streams in Washington, USA, found that the median particle size of suspended sediments did not correlate with the P content of suspended sediment or the proportion of particulate P that was bioavailable (Ellison and Brett, 2006). Therefore, while there is potential for selective erosion to play an important role in determining the P speciation of receiving environments, it is unclear if this overrides other factors. Biological and chemical transformation processes either within the receiving waters or within the upstream rivers and streams may also be a key determinant of P speciation in lake sediments. Within aquatic ecosystems, processes such as sorption, reduction/oxidation, precipitation, and mineralization result in the transformation of P between inorganic and organic forms (Rydin, 2000; Pardo et al., 2003; Ellison and Brett, 2006). As soils are deposited and transported in streams and rivers these processes may alter the balance of P between different species (Froelich, 1988; House and Denison, 2002; Jarvie et al., 2005). Furthermore, sediments deposited in lakes and reservoirs can undergo additional physical, chemical, and biological processing (Søndergaard et al., 1996; ^ek et al., 2000; Rydin, 2000). While this provides Kopa´c evidence of P transformation in lake sediments following deposition, it remains unclear whether these processes lead to a major shift in P speciation in lake sediments relative to the soils from which they were derived. The speciation of P in sediments has been widely studied using various sequential extraction techniques which produce a number of operationally defined P fractions (Chang and Jackson, 1957; Hieltjes and Lijklema, 1980; Psenner et al., 1988; Ruttenberg, 1992; Ruban et al., 2001; Pardo et al., 2004). Although sequential extraction techniques have limitations in terms of their capacity to identify specific P species (Psenner and Puckso, 1988; Ruttenberg, 1992; Ruban et al., 1999), they have provided useful information on the potential for various P fractions to be released from sediments over the short to mid-term. While sequential extraction techniques have been employed to understand within-lake cycling of P, they have rarely been extended to the catchment scale to investigate the link between terrestrial and aquatic P. The aim of this research was to apply sequential extraction techniques on soils, suspended sediments, river sediments and lake sediments to determine the relative importance of parent
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 3 1 e3 3 4 0
material, erosion and transport processes, and internal P transformations in determining the P speciation of sediments in a P limited subtropical reservoir.
2.
Methods
2.1.
The study site e upper Brisbane River
The study was conducted in the upper Brisbane River (UBR) catchment in southeast Queensland, Australia, approximately 80 km northwest of the city of Brisbane. The hydrology of the upper Brisbane River catchment is characterized by a high degree of inter- and intra-annual variation in flow, and as such, P loads are predominately associated with infrequent but intense storm events (Kerr, 2009). Land use in the catchment is primarily associated with cattle grazing (z70%) with minimal urban or intensive agricultural uses. The catchment has particular ecological, social and economic importance because it flows into southeast Queensland’s largest potable water storage, Lake Wivenhoe. Lake Wivenhoe stratifies in the summer months between October and March and previous studies suggested that algal growth in the reservoir is P limited (Burford et al., 2007). Douglas et al. (2007) used a combination of NdeSr isotopic analysis and a Bayesian mixing model to quantitatively estimate the contribution of catchment sources to sediment in Lake Wivenhoe. They found that 36% of the catchment area, comprising three major rock formations (Neara Volcanics, Esk Formation and Maronghi Creek Beds), provided the majority of sediments to Lake Wivenhoe (Douglas et al., 2007).
2.2.
Study design
P fractionation was contrasted between different compartments of the catchment soil/sediment system. These
3333
compartments comprised: streambank soils and surface soils collected from source areas of the UBR catchment; suspended sediment collected from the upper and mid-lower reaches of the UBR during flow events; bed sediments of the UBR; and bed sediments in Lake Wivenhoe (Fig. 1). Statistical comparisons between these compartments were used to identify changes in P speciation as soils are transported into the river system, processed in stream and eventually deposited and further processed in the receiving environment. The effect of parent material as a determinant of P speciation in Lake Wivenhoe sediment was assessed by comparing the P fractionation of Lake Wivenhoe sediments with the P fractionation of bulk soils collected from areas identified by Douglas et al. (2007) as being a major source of sediment to the reservoir. To identify the importance of selective erosion in altering the speciation of soils, P fractionation was compared between suspended sediments collected during event flows and the bulk source material. In addition, an enrichment factor (EF) was calculated for all soils and riverbed sediments to determine if P fractions were distributed unevenly between fine (<63 mm) and bulk samples. To investigate the importance of transformation processes the P fractionation of river and lake sediments was compared with suspended sediments and soils. In addition P fractionation in sediments was compared across a spatial gradient from the upper reaches of the UBR to the dam wall at Lake Wivenhoe.
2.3.
Sampling
Surface soils, streambank soils, and bed sediments were collected in December 2007 from sites in the upper Brisbane River. Samples of bed sediment, surface soil and streambank soil were taken from sites ranging from the upper to lower reaches of the upper Brisbane River catchment (Fig. 1). The sampling sites covered regions identified by Douglas et al. (2007) as being the major sources of sediment and P to
Fig. 1 e Location of study sites in the upper Brisbane River catchment.
3334
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 3 1 e3 3 4 0
Wivenhoe dam. River sediments were collected from the riverbed using a PVC corer and the top 2 cm of sediment was taken. Soil samples were also collected from the top 2 cm of upland surrounding slopes (surface soils) and exposed streambanks (streambank soils). Surface soils, as defined in our study, represent upland soils susceptible to erosion via overland flow. They may include subsurface soils exposed by previous erosion events. At each sampling point, three replicate samples of soil or sediment were collected and homogenized on site to form a composite sample. A total of five composite samples were taken for each soil/sediment type at each site. Suspended sediments were collected during two major flow events in the upper Brisbane River. Flows at the gauging stations at UBRG (2020 ML d1) and UBR (Linville) (1367 ML d1) were above the 95th percentile of historical flows (1990e2004) (1275 ML d1 and 537 ML d1, respectively). Events of this magnitude account for the majority of P inputs into Lake Wivenhoe (Kerr, 2009). Samples were collected at UBRG and UBRA on the 5th of January 2008. Samples were collected on the 15th of January 2008 from sites in the upper catchment at UBR12, UBR10 and UBR8 (Fig. 1). Suspended sediment samples were obtained by collecting bulk water samples into 25 L containers from mid stream during storm events. Suspended sediments were separated in the laboratory using a continuous flow centrifuge. The sediment collected from each site was homogenized and split into sub-samples (n ¼ 5) for the sequential extraction. Sediment samples were also collected at 4 sites within Lake Wivenhoe in February 2008. Sites ranged from the inlet (WIV4) to the dam wall (WIV1) (Fig. 1) with water depth increasing from the inlet to the dam wall (WIV4 ¼ 5 m, WIV3 ¼ 9 m, WIV2 ¼ 18 m, and WIV1 ¼ 23 m). Five replicate samples of surface sediment were taken from each site using a grab sampler. All sediment samples were stored in zip lock bags placed inside larger zip lock bags filled with sediment to prevent oxidation of bed sediments during collection and storage (Baldwin, 1996).
2.4.
Analytical
Prior to the sequential extraction analysis, bulk sediment and soil samples were first sieved through a 63 mm sieve to obtain fine (<63 mm) and bulk (0e2 mm) size fractions for subsequent analysis. All samples were sieved wet/field moist, and all subsequent sequential extractions were performed on wet/field moist sediment and soil samples. Samples were analyzed wet because previous studies of UBR sediments found that drying had a substantial effect on P fractionation (Kerr et al., 2010). Sediment P fractionation was determined based on the sequential extraction procedure of Psenner et al. (1988) which produces five operationally defined P fractions (Table 1). 0.5 g Dry weight equivalent of sediment/ soil was weighed into 50 ml centrifuge tubes and 25 ml of 1.0 M NH4Cl (pH 7) was added. The mixture was shaken on an orbital shaker for 2 h and the supernatant separated by centrifugation at 4000 g and filtered through 0.45 mm membrane filters (Millipore, USA) before analysis of exchangeable P (Ex-P). This procedure was repeated on the sediment residue before 25 ml of 0.11 M NaHCO3eNa2S2O4 was added and the mixture was shaken for 30 min at 40 C. The supernatant was separated by centrifugation at 4000 g
Table 1 e Sequential extraction procedure based on Psenner et al. (1988) showing extraction reagents and their targeted P forms (adapted from Hupfer et al., 1995). Extractant/fraction NH4Cl Bicarbonateedithionite NaOH HCl TP e (Ex-P þ BD-P þ NaOH-rP þ HCl-P)
Targeted P form Loosely sorbed P and porewater P Redox sensitive P, mainly bound to Fe-hydroxides and Mn compounds Metallic oxide-bound P (mainly Al- and Fe-bound) exchangeable against OH-ions Apatite- and CaCO3-bound P Organic P and mineral P resistant to previous extractions
and filtered for analysis of reductant soluble P (bicarbonatee dithionite P; BD-P). The residue was then extracted with 1.0 M NaOH for 16 h at 25 C before centrifugation and filtration of supernatant before determination of Al/Fe oxide P (NaOH-rP). The residue was then shaken for 24 h in 25 ml of 0.5 M HCl at 25 C followed by centrifugation and filtration before analysis of apatite-P (HCl-P). Residual-P (organic P and mineral P resistant to previous extractions) was calculated as the differences between TP and the sum of the four fractions. Total-P (TP) was determined by calcination at 450 C for 3 h and subsequent extraction (16 h) with 3.5 M HCl (Pardo et al., 2004). All sediment extracts were measured for reactive P following neutralization or oxidation of samples where appropriate. Reactive P was determined using the colorimetric ascorbic acid method (APHA, 1998).
2.5.
Statistical
Statistical analyses were performed using the Statistical Package for the Social Sciences (SPSS) version 14.0.1. One way ANOVA was used to test for differences between soil/sediment types for each P fraction. Where there was significant heterogeneity of variance between treatments (i.e. soil types and sites) logarithmic transformation of data sets was performed prior to ANOVA analysis. Where homogeneity of variance could not be achieved through logarithmic transformation non-parametric Kruskal Wallace and Manne Whitney U tests were performed to identify statistically significant differences between treatments. An enrichment factor (EF) was calculated from the concentration of each P fraction in the fine sample divided by the concentration in the bulk sample. A value of 1 indicates that there is no enrichment or depletion in the fine versus bulk particle size samples. To determine if enrichment or depletion of each P fraction was significant, a one sample t-test was employed to determine if EF was significantly different than 1. Multidimensional scaling (ALSCAL) was employed to group samples (site * soil/sediment type) based on P fractionation across the entire catchment. This was done to test whether samples were more closely related based on soil/sediment type, catchment location or a combination of the two. The percent TP of each fraction was used as the basic sample unit for the analysis. Similarity between groups was defined by the Euclidian distance with similar samples having a lower Euclidian distance.
3335
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 3 1 e3 3 4 0
Fig. 2 e The mean (±1 SD) enrichment factor (EF) for each P fraction in riverbed sediments, streambank soils and surface soils (n [ 25, * indicates that EF is significantly different than 1 at a [ 0.05).
3.
Results
3.1. Enrichment factors of potential sources of sediment to Lake Wivenhoe Analysis of the enrichment factors (EFs) of P in surface soils, streambank soils and riverbed sediments showed that TP was higher in fine soils and sediments (<63 mm) relative to the bulk sample (0e2 mm). The EF of TP in these soils and sediments was significantly higher than 1 (P < 0.05). With the exception of HCl-P, the mean EF of each P fraction was also significantly higher than 1 (P < 0.05) for both soil types and also for riverbed sediments. For HCl-P, the EF was significantly lower than 1 in the riverbed sediments and the streambank soils. In the surface soils the EF of HCl-P was not significantly different from 1 (Fig. 2).
3.2. A comparison of P fractionation in different soil and sediment types of the UBR catchment There were distinct differences in P fractionation between the different soil and sediment types (Table 2). The mean concentrations of HCl-P in streambank soils, surface soils and riverbed sediments were significantly higher than the suspended and lake sediments (P < 0.05). Mean HCl-P in soils and riverbed sediments was approximately 10 times higher than in suspended sediments and approximately 3 times higher than in lake sediments. Unlike HCl-P, mean NaOH-rP concentration was significantly higher (P < 0.05) in suspended sediments relative to the other soil/sediment types (Table 1). The mean concentration of NaOH-rP in suspended sediments was more than double that in the soils and riverbed sediments. Mean NaOH-rP in lake sediments was also relatively
high. The mean NaOH-rP concentration in lake sediments was significantly higher than soils and riverbed sediments (P < 0.05). Res-P exhibited a similar pattern to NaOH-rP with higher mean Res-P in the suspended sediments and lake sediments relative to soils and riverbed sediments (P < 0.05). For BD-P the pattern was similar with significantly higher mean BD-P in lake and suspended sediments relative to soils. However, unlike NaOH-rP and Res-P, the mean BD-P concentration of suspended sediments was not significantly higher than riverbed or lake sediments (P < 0.05). Mean BD-P in lake sediments was actually higher than suspended sediments although the difference was not significant (P > 0.05). However, on a percentage basis, the BD-P fraction in lake and river sediments (z20%) was significantly higher (P < 0.05) than suspended sediments and soils (13e16%). For Ex-P, the mean concentration was significantly higher in surface soils and streambank soils relative to suspended and lake sediments (P < 0.05). In the riverbed sediments, Ex-P was below the limit of detection (1 mg kg1 dry wt.).
3.3. Spatial patterns of P fractionation in soils and sediments of the UBR catchment There were some clear spatial patterns in terms of P fractionation in the soils and sediments of the UBR catchment (Fig. 3). In particular, several P fractions exhibited a substantial decrease in concentration from the upper catchment through to the mid to lower catchment. The concentration of HCl-P in surface soils was highest in the upper to mid catchment (UBR17eUBR3), while in the streambank soils HCl-P was higher in UBR17 relative to the other sites. A similar pattern was found for the riverbed sediments with highest concentrations in the upper to mid reaches and a steady decline with distance downstream and into Lake Wivenhoe. For NaOH-rP, the concentration was also highest in the upper catchment for the surface soils. Unlike HCl-P, this also coincided with higher NaOH-rP in suspended sediments collected from the upper catchment sites. However, for streambank soils and riverbank sediments there was no clear spatial trend between the upper and lower catchment sites. The only other noticeable trend for NaOH-rP was in lake sediments where the concentration of NaOH-rP increased from WIV4 to WIV1. In the streambank and surface soils Res-P was particularly high in the upper catchment at UBR17. Similarly, Res-P in the suspended sediments was highest in the upper catchment at UBR12 and UBR10. In the riverbed sediments, Res-P also decreased from the upper catchment sites (UBR17 and UBR12) but then
Table 2 e Mean P concentration (±1 Std Deviation) of each P fraction for each soil and sediment type (n [ 25) in the upper Brisbane River catchment (
BD-P 129 93 90 140 167
(38) (16) (25) (38) (60)
NaOH-rP 117 152 159 364 261
(23) (36) (32) (78) (101)
HCl-P 202 203 234 17 69
(71) (54) (40) (5) (28)
Res-P 209 191 238 392 319
(52) (76) (142) (64) (47)
TP 658 662 748 909 818
(103) (83) (199) (131) (162)
3336
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 3 1 e3 3 4 0
Fig. 3 e Spatial variation in mean sediment P concentration (mg kgL1 dry wt.) of each P fractionation ( y axis left) and total P ( y axis right) in soils and sediments of the upper Brisbane River catchment.
increased from UBR3 to WIV3. In lake sediments, BD-P concentration increased from the upper Wivenhoe site (WIV 4) to the dam wall (WIV 1) but there were no other clear spatial patterns in BD-P for the other soil and sediment types. Similarly, there were no clear spatial patterns for Ex-P in soils or sediments.
3.4. Analysis of similarities in P speciation between soil and sediment samples in the UBR catchment Multidimensional scaling of the data set was used to determine if samples were grouped predominately by soil/sediment type or by spatial factors. The analysis produced a low stress value (0.0484) indicating the distances derived from the model give a close representation of dissimilarity between samples. The greatest level of dissimilarity was along the horizontal axis (dimension 1) and this was primarily due to differences between soil/sediment types (Fig. 4). There were
two distant groups of samples; one consisting of the suspended sediment and lake sediment samples and another consisting of the surface soils, streambank soils and riverbed sediments. Correlation analysis of values along the horizontal axis of the MDS plot versus the percent contribution of each P fraction found that variation along dimension 1 was significantly correlated (P < 0.05) with % HCl-P (r ¼ 0.99), % NaOH-rP (r ¼ 0.78) and % Res-P (r ¼ 0.73). Dissimilarities along the vertical axis (dimension 2) appear to be due to spatial differences within soil/sediment groups (Fig. 4). There was greater spatial similarity for suspended sediments, and to a lesser extent lake sediments, relative to riverbed sediments and soils. Values for each sample along the vertical axis (dimension 2) were significantly correlated (P < 0.05) with Res-P (r ¼ 0.68) and NaOH-rP (r ¼ 0.58). With the exception of the sediment and streambank soils at UBR3, there were few examples of close spatial links between sediments and soils.
Fig. 4 e Plot showing results of multidimensional scaling analysis. Distances between samples are Euclidian distances and represent the degree of dissimilarity between samples based on P fractionation (% TP of each fraction).
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 3 1 e3 3 4 0
4.
Discussion
4.1. The effect of selective erosion and transport processes on sediment P speciation in Lake Wivenhoe The differences in P fractionation between the major soil/ sediment components of the UBR catchment can be attributed to an uneven distribution of non-apatite (BD-P, NaOH-rP and Res-P) and apatite-P (HCl-P) fractions between different grain sizes in catchment soils, and the preferential export of fine sediment particles in overland and stream flow. Evidence for this includes the close grouping of lake sediments and suspended sediments from the MDS analysis; substantially lower EF values for HCl-P (apatite-P) compared with non-apatite P in the catchment soils; and the significantly higher HCl-P in soils and riverbed sediments relative to suspended sediments and deposited lake sediments. Furthermore, the trend of decreasing HCl-P in lake sediments from the inlet to the dam wall indicates further sorting of fine particles within Lake Wivenhoe as water depth increases from the inlet (WIV 4) to the dam wall (WIV 1). Therefore, our results suggest that selective erosion and transport of fine particles across the catchment are a major driver of sediment P speciation in Lake Wivenhoe, while the P speciation of the bulk soils and sediments upstream play a much smaller role. SanClements et al. (2009) suggested that the selective erosion and transportation of fine particles were an important determinant of lake P speciation in a forested catchment in Maine, USA. In their study, this produced an increase of HCl-P in lake sediments (112 mg kg1) relative to soils (33 mg kg1) and stream sediments (89 mg kg1) upstream. HCl-P in the UBR catchment (soils and riverbed sediments) was considerably higher than these values. However, HCl-P in Lake Wivenhoe was considerably lower than in the lake sediments from Maine. This further demonstrates that the composition of the bulk soil matrix is a poor predictor of sediment P speciation in the receiving environment. The comparison of our results with those of SanClements et al. (2009) also shows that selective erosion processes may alter the speciation of sediment P in different ways in different systems. The differences between the UBR and Maine catchments in terms of HCl-P probably reflect a greater degree of weathering of fine soils in the UBR catchment leading to the formation of secondary inorganic species associated with Al/Fe or organic P (Zehetner et al., 2008).
4.2. The effect of within stream/lake processes on P speciation in Lake Wivenhoe While selective erosion and transportation appear to be the dominant drivers of sediment P speciation in Lake Wivenhoe, there is some evidence that transformation processes may be important in determining the proportion of BD-P and Ex-P. The significantly lower Ex-P in lake sediments relative to soils could not be explained by sediment transport processes. The EF values of Ex-P in surface soils and streambank soils were greater than 1. This was not surprising as Ex-P is located on the soil surface and as such finer particles with a higher surface area should have a higher Ex-P content. However, unlike the NaOH-rP fraction, this did not translate to elevated
3337
Ex-P in lake sediments. Importantly, unlike the NaOH-rP fraction, Ex-P in suspended sediments was significantly lower than in the soils. This suggests that during rain events the Ex-P associated with fine particles in soils is desorbed prior to the soil entering the river system as suspended sediment. Soils in the UBR catchment have poor sorption capacity and high equilibrium P concentrations indicating that when these soils come into contact with overland flow (during storm events) there will be a net movement of PO42 from soils to the overland flow (Kerr et al., 2011). This would explain why the Ex-P concentration of suspended sediments was substantially lower than in the catchment soils, and why there was no enrichment of Ex-P in the riverbed or lake sediments. Although the EF value of BD-P in soils indicated that selective erosion played a role in the higher BD-P of lake sediments, other results suggest that some of this increase may be due to within stream/lake transformations. The higher % BD-P in riverbed and lake sediments relative to suspended sediments suggests an accumulation of P in Fe-bound forms in the post deposition period. Furthermore, there was a significantly higher BD-P in riverbed sediments relative to soils. If the higher BD-P in riverbed sediments was due solely to selective erosion, it would be expected that the NaOH-rP and Res-P fractions would also be elevated in riverbed sediments. Unlike catchment soils, riverbed sediments in the UBR have a high P sorption capacity and are likely to function as a sink for water column PO42 (Kerr et al., 2011). Importantly, the high P sorption capacity of riverbed sediments relative to catchment soils is associated with higher amorphous Fe in sediments relative to soils. Furthermore, P sorption in the UBR catchment is related to amorphous Fe but not amorphous Al. This suggests that PO42 sorption onto amorphous Fe surfaces is an important retention mechanism in riverbed sediments of the UBR (Kerr et al., 2011). This should lead to an accumulation of Fe-bound P in riverbed sediments and this could explain why the proportion of TP as BD-P was higher in riverbed sediments relative to suspended sediments and soils.
4.3. Sediment P delivery from streambank soil versus surface soil The distribution of P species between fine and bulk soil particles showed similar patterns for the surface and streambank soils. Therefore, any selective erosion of fine particles from either source should have a similar affect on the speciation of suspended sediment (i.e. enriched NaOH-rP and depleted HCl-P). However, due to the close proximity of streambank soils to stream flow, there are some important considerations in terms of P supply to the river system. Firstly, while both streambank soils and surface soils had relatively high Ex-P, for streambank soils there is a much higher level of connectivity between sites of potential desorption and stream flow. As such, the Ex-P of streambank soils is likely to provide a more immediate and potentially sustained source of PO43 to the water column, relative to surface soils where Ex-P pools are frequently disconnected from stream flow. Secondly, the substantially shorter transport pathway, together with the potential for bank failure in streambank soils should result in a higher proportion of coarse sediment inputs relative to upland surface soils (Prosser et al., 2001; Bartley et al., 2007).
3338
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 3 1 e3 3 4 0
Therefore, based on the distribution of P species between fine and coarse soils in the UBR catchment, it is reasonable to suggest that streambank soils will contribute a higher proportion of total P as HCl-P relative to surface soils.
4.4. Implications of selective erosion and transport processes for P cycling in Lake Wivenhoe The selective erosion and transport of non-apatite P fractions (BD-P and NaOH-rP) at the expense of apatite-P (HCl-P) have important implications for P cycling in Lake Wivenhoe. The solubility of apatite-P is pH dependent (Bostro¨m et al., 1988). Although apatite is soluble in acidic waters, the pH levels in Lake Wivenhoe (mean pH ¼ 7.7e8.2; Burford et al., 2007; Orr et al., 2010) are high enough that any release of P via dissolution of apatite is unlikely (Harouiya et al., 2007). The BD-P fraction is redox sensitive and therefore P release from this fraction is expected to occur during periods of hypolimnetic anoxia (Lake et al., 2007). However, compared to NaOH-rP, the BD-P fraction comprised a relatively small proportion of TP. Assuming NaOH-rP is associated primarily with Al(OH)3, the P associated with this fraction is unaffected by redox conditions and is considered to be relatively stable at circumneutral pH ^ek et al., 2000, 2005). However, at low or high pH the (Kopa´c solubility of Al(OH)3 increases resulting in an increased potential for P release (Bostro¨m et al., 1988; Lewandowski et al., 2003; Xiangcan et al., 2006). Based on the historic pH levels of Lake Wivenhoe (7.7e8.2), it is therefore likely that the NaOH-rP fraction is an important source of PO42 to the water column. Therefore, in terms of the importance of selective erosion and transport processes to P cycling in Lake Wivenhoe, these processes are important because they result in the export of more labile forms of P (BD-P and NaOH-rP) at the expense of a more recalcitrant form (HCl-P). The proportion of non-apatite inorganic-P relative to apatite-P in Lake Wivenhoe was generally at the higher end of values reported from studies of surface sediments in European and North American lakes (Søndergaard et al., 1996; ^ek et al., 2005; Lake Rydin, 2000; Kaiserli et al., 2002; Kopa´c et al., 2007). This is despite the fact that the soil and sediment upstream contained high levels of apatite-P (HCl-P). The relatively high non-apatite P levels in Lake Wivenhoe compared to lakes in the Northern Hemisphere is likely due to: (i) the age and degree of weathering of Australian soils relative to the Northern Hemisphere (Australian soils are older and more weathered) and (ii) the selective erosion and transport of these highly weathered clays across the soil/sediment system. These two factors could explain the relatively high proportion of non-apatite P in Lake Wivenhoe compared to lakes in the Northern Hemisphere.
4.5. Implications of selective erosion and transport processes for catchment P management Our results have implications for how P is managed in the UBR catchment and also in other catchments where soil erosion is a major source of P to downstream ecosystems. For example, vegetative buffer strips (VBS) are increasingly used to minimize the transport of diffuse pollutants (Stutter et al., 2009). While these systems trap and retain sediment, they do not trap
particles of varying grain size equally (Syversen and Borch, 2005). Our results suggest that this could alter the balance between labile and recalcitrant forms of particulate P transported to streams. If more labile forms of P are associated with the finer soil fraction (e.g. UBR catchment), VBS may not substantially reduce the transport of more labile forms of P. Alternatively, if more recalcitrant forms of P are associated with the finer soil fraction (e.g. SanClements et al., 2009), then VBS may be far more effective in terms of retaining the most labile forms of particulate P. Therefore a simple estimate of total P retention may fail to account for key differences in the retention of different forms of P. This in turn has important implications for P cycling in downstream ecosystems and therefore should be given greater consideration when assessing the potential benefits of implementing VBS as a P mitigation measure.
5.
Conclusion
Our results provide important information on the link between terrestrial and aquatic P in a subtropical catchment impacted by grazing. In systems such as the UBR, where widespread grazing has resulted in substantial soil erosion, understanding the processes operating on soils as they move from the terrestrial to the aquatic components of the catchment is critical if we are to effectively mitigate the impact of P downstream. Our results suggest that the selective erosion and transportation of fine soil particles substantially altered the P speciation of lake sediments relative to the soils and river sediments upstream. In addition, there was some evidence that adsorption/desorption processes might be an important driver of change in the Ex-P and BD-P fractions from soil to lake sediment. These processes represent important links between P speciation in the terrestrial and aquatic components of the catchment. Importantly, they have implications for P availability in Lake Wivenhoe and also in other systems where soil erosion is a major source of P, and P fractions are distributed unevenly between different grain sizes. It is important that these factors are considered when assessing the potential benefits of various management options aimed at reducing the impact of soil P inputs to lakes and reservoirs. The benefit of reducing the transport of P from terrestrial to aquatic components of the catchment cannot be properly assessed based on total P alone. Our study provides important information in this regard by identifying the key drivers of P speciation across the entire soil/sediment system.
Acknowledgments We thank the Australian Research Council, the Southeast Queensland Healthy Waterways Partnership and Seqwater for their financial support. We would also like to thank Dominic Valdez and Stephen Faggotter for their assistance with field sampling.
references
APHA, 1998. Standard Methods for the Examination of Water and Wastewater. American Public Health Association, American
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 3 1 e3 3 4 0
Water Works Association, Water Environment Federation, Washington. Baldwin, D.S., 1996. Effects of exposure to air and subsequent drying on the phosphate sorption characteristics of sediments from a eutrophic reservoir. Limnology and Oceanography 41 (8), 1725e1732. Ballantine, D.J., Walling, D.E., Collins, A.L., Leeks, G.J.L., 2008. The phosphorus content of fluvial suspended sediment in three lowland groundwater-dominated catchments. Journal of Hydrology 357, 140e151. Bartley, R., Hawdon, A., Post, D.A., Roth, C.H., 2007. A sediment budget for a grazed semi-arid catchment in the Burdekin basin, Australia. Geomorphology 87, 302e321. Bostro¨m, B., Andersen, J.M., Fleischer, S., Jansson, M., 1988. Exchange of phosphorus across the sediment-water interface. Hydrobiologia 170, 229e244. Burford, M.A., Johnson, S.A., Cook, A.J., Packer, T.V., Taylor, B.M., Townsley, E.R., 2007. Correlations between watershed and reservoir characteristics, and algal blooms in subtropical reservoirs. Water Research 41, 4105e4114. Chang, S.C., Jackson, M.L., 1957. Fractionation of soil phosphorus. Soil Science 84, 133e144. Douglas, G., Palmer, M., Caitcheon, G., Orr, P., 2007. Identification of sediment sources to Lake Wivenhoe, south-east Queensland, Australia. Marine and Freshwater Research 58, 793e810. Ellison, M.E., Brett, M.T., 2006. Particulate phosphorus bioavailability as a function of stream flow and land cover. Water Research 40, 1258e1268. Froelich, P.N., 1988. Kinetic control of dissolved phosphate in natural rivers and estuaries: a primer on the phosphate buffer mechanism. Limnology and Oceanography 33, 649e668. Golterman, H.L., 2001. Fractionation and bioavailability of phosphates in Lacustrine sediments: a review. Limnetica 20 (1), 15e29. Grimm, N.B., Gergel, S.E., McDowell, W.H., Boyer, E.W., Dent, C.L., Groffman, P., Hart, S.C., Harvey, J., Johnston, C., Mayorga, E., McClain, M.E., Pinay, G., 2003. Merging aquatic and terrestrial perspectives of nutrient biogeochemistry. Oecologia 442, 485e501. Harouiya, N., Chaı¨rat, C., Ko¨hler, S.J., Gout, R., Oelkers, E.H., 2007. The dissolution kinetics and apparent solubility of natural apatite in closed reactors at temperatures from 5 to 50 C and pH from 1 to 6. Chemical Geology 244, 554e568. Hieltjes, A.H.M., Lijklema, L., 1980. Fractionation of inorganic phosphorus in calcareous sediments. Journal of Environmental Quality 8, 130e132. House, W.A., Denison, F.H., 2002. Exchange of inorganic phosphate between river waters and bed-sediments. Environmental Science and Technology 36, 4295e4301. Hupfer, M., Ga¨chter, R., Giovanoli, R., 1995. Transformation of phosphorus species in settling seston and during early sediment diagenesis. Aquatic Sciences 57 (4), 305e324. Istva´novics, V., 1994. Fractional composition, adsorption and release of sediment phosphorus in the Kis-Balaton reservoir. Water Research 28, 717e726. Jarvie, H.P., Ju¨rgens, M.D., Williams, R.J., Neal, C., Davies, J.J.L., Barret, C., White, J., 2005. Role of river bed sediments as sources and sinks of phosphorus across two major eutrophic UK river basins: the Hampshire Avon and Hertfordshire Wye. Journal of Hydrology 304, 51e74. Kaiserli, A., Voutsa, D., Samara, C., 2002. Phosphorus fractionation in lake sediments e Lakes Volvi and Koronia, N. Greece. Chemosphere 46, 1147e1155. Kerr, J.G., 2009. The Transformation of Catchment Phosphorus in Dry Subtropical River Systems. Dissertation, Griffith University. Kerr, J.G., Burford, M., Olley, J., Udy, J., 2010. The effects of drying on phosphorus sorption and speciation in subtropical river sediments. Marine and Freshwater Research 61, 928e935.
3339
Kerr, J.G., Burford, M., Olley, J., Udy, J., 2011. Phosphorus sorption in soils and sediments: implications for phosphate supply to a subtropical river in southeast Queensland, Australia. Biogeochemistry 102, 73e85. Kleeberg, A., Dudel, G.E., 1997. Changes in extent of phosphorus release in a shallow lake (Lake Groer Muggelsee; Germany, Berlin) due to climatic factors and load. Marine Geology 139, 61e75. ^ek, J., Borovec, J., Hejzlar, J., Ulrich, K.-U., Norton, S.A., Kopa´c Amirbahman, A., 2005. Aluminium control of phosphorus sorption by Lake sediments. Environmental Science and Technology 39, 8784e8789. ^ek, J., Hejzlar, J., Borovec, J., Porcal, P., Kotorova, I., 2000. Kopa´c Phosphorus inactivation by aluminium in the water column and sediments: lowering of in-lake phosphorus availability in an acidified watershed-lake ecosystem. Limnology and Oceanography 45, 212e225. Lake, B.A., Coolidge, K.M., Norton, S.A., Amirbahman, A., 2007. Factors contributing to the internal loading of phosphorus from anoxic sediments in six Maine, USA, lakes. Science of the Total Environment 373, 534e541. Lewandowski, J., Schauser, I., Hupfer, M., 2003. Long term effects of phosphorus precipitations with alum in hypereutrophic Lake Su¨sser See (Germany). Water Research 37, 3194e3204. McDowell, R.W., Sharpley, A.N., 2003. Uptake and release of phosphorus from overland flow in a stream environment. Journal of Environmental Quality 32, 937e948. Orr, P.T., Rasmussen, J.P., Burford, M.A., Eaglesham, G.K., Lennox, S.M., 2010. Evaluation of quantitative real-time PCR to characterise spatial and temporal variations in cyanobacteria, Cylindrospermopsis raciborskii (Woloszynska) Seenaya et Subba Raju and cylindrospermopsin concentrations in three subtropical Australian reservoirs. Harful Algae 9, 243e254. Pacini, N., Gatcher, R., 1999. Speciation of riverine particulate phosphorus during rain events. Biogeochemistry 47, 87e109. Pardo, P., Lo´pez-Sa´nchez, J.F., Rauret, G., 2003. Relationships between phosphorus fractionation and major components in sediments using the SMT harmonised extraction procedure. Analytical and Bioanalytical Chemistry 376, 248e254. Pardo, P., Rauret, G., Lo´pez-Sa´nchez, J.F., 2004. Shortened screening method for phosphorus fractionation in sediments: a complementary approach to the standards, measurements and testing harmonised protocol. Analytica Chimica Acta 508, 201e206. Penn, M.R., Auer, M.T., Van Orman, E.L., Korienek, J.J., 1995. Phosphorus diagenesis in Lake Sediments: investigations using fractionation techniques. Marine and Freshwater Research 46, 89e99. Pettersson, K., Istvanovics, V., 1988. Sediment phosphorus in Lake Balaton e forms and mobility. Archiv fiir Hydrobiologie, Beihefre, Ergebnisse der Limnologie 30, 25e41. Prosser, I.P., Rutherfurd, I.D., Olley, J.M., Young, W.J., Wallbrink, P.J., Moran, C.J., 2001. Large-scale patterns of erosion and sediment transport in river networks, with examples from Australia. Marine and Freshwater Research 52, 81e99. Psenner, R., Puckso, R., 1988. Phosphorus fractionation: advantages and limits of the method for the study of sediment P origins and interactions. Archiv fiir Hydrobiologie, Beihefre, Ergebnisse der Limnologie 30, 43e59. Psenner, R., Bostrom, B., Dinka, M., Pettersson, K., Puckso, R., Sager, M., 1988. Fractionation of phosphorus in suspended matter and sediment. Archiv fiir Hydrobiologie, Beihefre, Ergebnisse der Limnologie 30, 98e103. Quinton, J.N., Catt, J.A., Hess, T.M., 2001. The selective removal of phosphorus from soil: is event size important? Journal of Environmental Quality 30, 538e545. Ruban, V., Sanchez, J.F., Rauret, G., Muntau, H., Quevauviller, P., 1999. Selection and evaluation of sequential extraction
3340
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 3 1 e3 3 4 0
procedures for the determination of phosphorus forms in lake sediment. Journal of Environmental Monitoring 1, 51e56. Ruban, V., Lopez-Sanchez, J.F., Pardo, P., Rauret, G., Muntau, H., Quevauviller, P., 2001. Development of a harmonised phosphorus extraction procedure and certification of a sediment reference material. Journal of Environmental Monitoring 3, 121e125. Ruttenberg, K.C., 1992. Development of a sequential extraction method for different forms of phosphorus in marine sediments. Limnology and Oceanography 37, 1460e1482. Rydin, E., 2000. Potentially mobile phosphorus in Lake Erken sediment. Water Research 34, 2037e2042. Søndergaard, M., Windolf, J., Jeppesen, E., 1996. Phosphorus fractions and profiles in the sediment of shallow Danish lakes as related to phosphorus load, sediment composition and lake chemistry. Water Research 30, 992e1102. SanClements, M.D., Fernandez, I.J., Norton, S.A., 2009. Soil and sediment phosphorus fractions in a forested watershed at Acadia National Park, ME, USA. Forest Ecology and Management 258, 2318e2325. Sharpley, A., Tunney, H., 2000. Phosphorus research strategies to meet agricultural and environmental challenges of the 21st century. Journal of Environmental Quality 29, 176e181.
Stone, M., English, M.C., 1993. Geochemical composition, phosphorus speciation and mass transport of fine-grained sediment in two Lake Erie tributaries. Hydrobiologia 253, 17e29. Stutter, M.I., Langan, S.J., Lumsdon, D.G., 2009. Vegetated buffer strips can lead to increased release of phosphorus to waters: a biogeochemical assessment of the mechanisms. Environmental Science and Technology 43, 1858e1863. Syversen, N., Borch, H., 2005. Retention of soil particle fractions and phosphorus in cold-climate buffer zones. Ecological Engineering 25, 382e394. Walling, D.E., Russell, M.A., Webb, B.W., 2001. Controls on the nutrient content of suspended sediment transported by British rivers. Science of the Total Environment 266, 113e123. Wetzel, R.G., 2001. Limnology: Lake and River Ecosystems. Academic Press, California, USA. Xiangcan, J., Wang, S., Pang, Y., Chang Wu, F., 2006. Phosphorus fractions and the effect of pH on the phosphorus release of the sediments from different trophic areas in Taihu Lake, China. Environmental Pollution 139, 288e295. Zehetner, F., Lair, G.J., Maringer, F.-J., Gerzabek, M.H., Hein, T., 2008. From sediment to soil: floodplain phosphorus transformations at the Danube River. Biogeochemistry 88, 117e126.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 4 1 e3 3 5 0
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Photodegradation of estrone enhanced by dissolved organic matter under simulated sunlight Emilie Caupos a, Patrick Mazellier a,b,*, Jean-Philippe Croue a,c a
Universite´ de Poitiers, CNRS e UMR 6008, Laboratoire de Chimie et Microbiologie de l’Eau, ENSIP, Poitiers, France Universite´ de Bordeaux, CNRS, UB1, UB4, UMR 5805, EPOC-Laboratoire de Physico et Toxico Chimie de l’Environnement, France c King Abdullah University of Science and Technology (KAUST), Water Desalination and Reuse Center, Saudi Arabia b
article info
abstract
Article history:
In the present work the degradation of estrone (E1) a natural estrogenic hormone has been
Received 3 October 2010
studied under simulated solar irradiation. The photodegradation of E1 has been investigated
Received in revised form
in the absence and in the presence of 7.7e8.9 mg L1 of dissolved organic carbon (DOC),
2 February 2011
under solar light simulation with irradiance approximating that of the sun. DOC extracts
Accepted 23 March 2011
from different origins have been used. Half-lives ranging between 3.9 h and 7.9 h were
Available online 31 March 2011
observed. Results indicated that E1 was photodegraded even in the absence of DOC. The presence of DOC was found to enhance the degradation of E1. Experiments performed with
Keywords:
the addition of reactive species scavengers (azide ions and 2-propanol) have shown that
Photodegradation
these two species play a significant role in the photodegradation. Some experiments have
Estrone
been performed with a DOC previously submitted to solar irradiation. Changes in optical and
Dissolved organic carbon - DOC
physico-chemical properties of DOC strongly affect its photoinductive properties, and hence
Photoproducts
its efficiency on E1 degradation. A part of the study consisted in the investigation of
Sunlight
photoproducts structures. Five photoproducts were shown by chromatographic analysis: one arising from direct photolysis and the four others from DOC photoinduced degradation. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
The growing interest related to the presence and analysis of Endocrine Disrupting Compounds (EDCs) is justified because their impact on ecosystem and human health is not yet completely understood. Numerous studies have shown the occurrence of these emerging pollutants in the environment, especially estrogenic compounds. Surface water is the first media receiving these estrogenic compounds mainly arising from WasteWater Treatment Plant (WWTPs) discharges (Kolpin et al., 2002). Only minor reduction is generally observed through WWTPs which have not been designed to remove these emerging pollutants. The major estrogenic compounds detected in urban wastewater effluents are two
natural estrogens (estrone and 17b-estradiol) and a synthetic one (17a-ethinylestradiol) (Desbrow et al., 1998). As some hormones are also present in WWTPs as sulfate-conjugates, it has been shown that very common bacteria such as Escherichia coli were able to deconjugate these hormones, giving back the hormone (Panter et al., 1999). E1 is the most frequently detected estrogen with the highest concentrations in WWTPs effluents (Ternes et al., 1999). In surface water concentrations of estrogens have been measured in the range of ng L1 but high concentrations up to mg L1 have also been determined (Kolpin et al., 2002; Hohenblum et al., 2004). However, even with the lowest concentration, the impact on the sexuality and reproduction of vertebrates and invertebrates has been demonstrated (Kang et al., 2002; Brion et al., 2004).
* Corresponding author. Universite´ de Bordeaux, CNRS, UB1, UB4, UMR 5805, EPOC-Laboratoire de Physico et Toxico Chimie de l’Environnement, France. Tel.: þ33 (0) 553352429. E-mail address:
[email protected] (P. Mazellier). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.03.047
3342
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 4 1 e3 3 5 0
Once in surface water, chemical processes like sunlight photolysis can lead to the degradation of organic pollutants. Some studies have already shown the efficiency of UV photolysis to transform these hormones (Liu and Liu, 2004; Rosenfeld and Linden, 2004; Mazellier et al., 2008) but only few data discuss their photodegradation under natural sunlight in the presence of Dissolved Organic Carbon (DOC). DOC is a ubiquitous component of natural waters. The presence of DOC can significantly impact organic pollutants photodegradation (Simmons and Zepp, 1986). These macromolecules coming from natural degradation of fauna and flora present different chemical and optical properties according to their origins (Martin-Mousset et al., 1997; Sierra et al., 2005). DOC contains chromophoric molecules sensitive to UV sunlight (Wetzel et al., 1995; Osburn et al., 2001). Hence, DOC controls the penetration of light in the water column and may alter the kinetics of pollutants direct photodegradation, acting as an inner filter. Moreover, the absorption of light by DOC generates reactive species such as singlet oxygen, hydroxyl and peroxyl radicals, solvated electron and triplet excited state that are able to promote pollutant degradation (Zepp et al., 1985, 1987; Canonica et al., 1995; Richard et al., 2004). Phototransformation of DOC is also observed as a consequence of light absorption leading to significant modifications of its structural and optical properties (Osburn et al., 2001; Brinkmann et al., 2003; Carvalho et al., 2008). Recent studies have underlined the natural photodegradation of E1 and E2 in surface waters under simulated sunlight (Lin and Reinhard, 2005; Leech et al., 2009; Chowdhury et al., 2010). In this work we have studied the photoinductive degradation of estrone (E1) by dissolved organic carbon (DOC) from surface water. Estrone has been chosen as a probe compound since it is the most abundant estrogen found in aquatic systems. Simulated sunlight photolysis experiments have been performed with different DOC extracts and the photodegradation efficiency was measured. Other experiments were performed with DOC extracts previously photodegraded, the phototransformation of E1 was then compared with photoinduced properties of the unreacted DOC. The formation of photoproducts was also investigated.
2.
Materials and methods
2.1.
Chemicals & solutions
Estrone (purity >99%, E1) was purchased from SigmaeAldrich Corporation, methanol and isopropanol from Carlo Erba (HPLC quality 99.9%), sodium azide (purity 99%, NaN3) from Acros Organics. Purified water (PW) was obtained from a Millipore purification system (Millipore Milli RX75/Synergy 185). Buffer solutions at pH 4.0 (phthalate), 7.0 (phosphate) and 10.0 (borate) were purchased from Acros Organics. As E1 is weakly soluble in water, aqueous solutions of E1 were prepared as follows: small amounts of E1 were magnetically stirred in purified water for two days; the undissolved E1 was removed by filtration through a 0.45 mm Millipore HVLP membrane (single use). The aqueous solution obtained was quantified according to calibration curves obtained using methanolewater E1 standard solutions. The solubility of E1 in water was found to be 4.0 0.4 mM under our experimental conditions at room temperature (20 2 C). A
dilute aqueous solution of E1 was prepared weekly; the pH was adjusted at 7.3 0.1 by addition of NaOH.
2.2.
Origin of DOC
Fulvic acids (Pinail FA, Suwannee FA and South Platte FA) have previously been extracted from surface waters according to the isolation procedure described by Thurman and Malcolm (1981) with minor modifications (Croue´, 2004). Suwannee FA and South Platte FA fractions were purchased from the International Humic Substances Society (IHSS). Pinail FA was isolated from the water of a strongly colored pond located in a forest near Poitiers (France). Suwannee FA originated from the Suwannee river (Georgia e U.S.A.) and South Platte FA from the corresponding river in Rocky Mountains with water sampling carried out upstream from Denver (Colorado e U.S.A.). This DOC was characterized by its high hydrophilic character.
2.3.
Analytical methods
Dissolved Organic Carbon was measured with a Shimadzu TOC-VCSH analyzer. UVeVisible spectra were obtained with a 5 cm quartz cell using a double beam Safas Monaco spectrophotometer between 200 and 600 nm. Fluorescence spectra were measured in a 1 cm quartz cell with a FluoroMax-2 spectrofluorometer (Jobin Yvon) at constant temperature of 20 C. Excitation-emission matrix (EEM) was generated by scanning excitation wavelength from 250 to 450 nm with 5 nm steps, and detecting the fluorescence between 300 and 600 nm with 5 nm steps. The excitation spectra of all DOC solutions showed that the emission of light was maximum for an excitation wavelength of 315 nm. High-pressure size exclusion chromatography (HPSEC) was used to characterize the molecular weight distribution of DOC solutions. Elution was carried out on a gel column (AIT, Reprosil 200 SEC, length: 300 mm, ID: 8 mm; particle size: 5 mm) using a binary pump (Model 1525, Waters) and a UV detector (Model 2487, Waters). The eluent was an aqueous solution of 0.01 M sodium acetate adjusted to pH 7.0, the flow rate was 1 mL min1 and injection volume 500 mL. The UV detector was set at 254 nm. E1 concentration was measured through high performance liquid chromatography with a UV detector (HPLC-UV). The equipment is a Waters system, which included a W600 pump, a W717 autosampler and a W486 UV detector (l ¼ 220 nm). Separation was performed with a Kromasil C18 250 4.6 mm ˚ , particule size: 5 mm) and column (endcapping, pore size: 100 A a water:methanol 30:70 v:v mobile phase at 1 mL min1. E1 stock solution (100 mM) and standard solutions (0.1e1 mM) were prepared in a mixture of methanol:water 3:2 v:v. External calibration was used. Gas and liquid chromatography coupled with mass spectrometry (GC-MS and LC-MSn) were used to identify E1 photoproducts. GC-MS analyses were performed with an Agilent 6890/5975 equipped with a CombiPAL injector. Solid-phase extraction was performed on irradiated samples (50 mL) with OASIS HLB 3 mL (60 mg) cartridges as described by Mazellier et al. (2008). GC-MS injection parameters were: pulsed splitless mode, injection temperature of 290 C, pulse pressure of 30 psi during 10 min then constant pressure of 8.56 psi. Initial oven temperature was 50 C during 10 min, then temperature
3343
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 4 1 e3 3 5 0
increased to 250 C with a rate of 25 C min1, then to 290 C (10 C min1) for 10 min and finally held at 300 C during 10 min (10 C min1). Separation was performed with a Agilent HP-1MS column 30 m; 0.25 m ID; 0.25 mm. The LC-DAD-MSn apparatus Thermo Surveyor chromatographic system includes two detectors: a Thermo Surveyor diode array detector and a Thermo DECA XP Plus ion trap mass spectrometer. Separation was carried on a Kromasil C18 ˚ , particle size: 5 mm) 250 3.2 mm (endcapping, pore size: 100 A with a water:methanol gradient elution (v:v 55:45 to 20:80 in 82 min) at 0.3 mL min1. Chemical ionization was performed in atmospheric pressure chemical ionization mode (APCI). The parameters were: capillary temperature of 200 C, vaporizer temperature 450 C, gas flow 7.4 L min1, auxiliary gas flow 9.9 L min1, corona discharge at 5 mA with a voltage 4.5 kV and capillary voltage 31 V. Mass range detection was 50e600 uma (to detect the formation of dimers). MSn experiments have been performed in order to identify photoproducts. MSn experiments were performed as follows: MS2 for m/z ¼ 271 with collision energy of 35%, Q activation of 0.25 and activation time of 30 ms, MS3 on m/z ¼ 253 with collision energy: 30%, Q activation of 0.25 and activation time of 30 ms.
2.4.
Photodegradation experiments
DOC photochemical modification and all photodegradation experiments were performed with a Suntest CPS þ solar simulator from Atlas Material Testing Solutions (Moussy Le Neuf e France). The system was equipped with a 1500 W xenon lamp filtered by a UV filter from Atlas (ref 56052371), delivering a light emission spectrum similar to that of the sun. Irradiations were performed with a light power of 250 W m2. Pyrex vials containing 4 mL of E1 with or without DOC were used for light exposition, with an irradiation time up to 8 h. A vial was removed at selected irradiation time. Control vials covered with aluminum foils have been introduced for each experiment since an increase of temperature has been observed under illumination (up to 33 C). No change in E1 concentration has been observed within 8 h of irradiation in aluminum-covered vials meaning that the degradation phenomenon described later was only due to light exposition. For the pre-photolyzed DOC, Pinail FA extract was used at a concentration of DOC of 20 mg L1. Samples were placed into 25 mL glass tubes and exposed in the Suntest during a week. Control tubes covered with aluminum foil were also prepared. Irradiations dose was set at 250 W m2 during a week and samples were taken every day. DOC, UVeVisible and fluorescence spectra, and high performance size exclusion chromatography (HPSEC-UV) analyses were performed at selected
reaction times. E1 was added in a pre-photolyzed DOC solution and samples were further irradiated under simulated sunlight for 4 h.
2.5.
Rate constants measurements
The degradation of E1 in different experimental conditions obeys apparent first order kinetics. The first order rate constant was calculated from the slope of the semi-logarithmic plot of E1 concentration as a function of the irradiation time. Each experimental condition was repeated at least two times; most of them three times. Standard deviations with 95% confidence interval were calculated.
3.
Results and discussion
3.1.
Characterization of DOC
A preliminary study of DOC optical properties was performed with DOC isolates (Pinail PPHPOA, South Platte SPHPOA and Suwannee SWHPOA). Table 1 presents the chemical and optical properties of these 3 extracts. UV absorbance is due to the presence of chromophoric compounds (compounds with aromatic structures) and fluorescence emission properties relate to the presence of fluorophores. The absorbances were measured at 300 and 315 nm. The wavelength 300 nm roughly corresponds to the beginning of the emission of the suntest xenon lamp and, as already mentioned, 315 nm corresponds to the optimum excitation wavelength with respect to maximum fluorescence emission (350 and 600 nm with an excitation at 315 nm). For similar DOC content, Pinail FA and Suwannee FA exert higher UV315 absorbance (higher specific UV absorbance at 315 nm i.e. SUVA315) as compared to the South Platte FA. The first two isolates are more enriched in aromatic and/or chromophoric groups corresponding to a stronger hydrophobic character. Pinail and Suwannee FA also generate strong fluorescence emission signal. As the fluorescence intensity depends on the rate of light absorption, the direct comparison of fluorescence intensity of DOC isolates is difficult. It is necessary to correct the fluorescence intensity measured with the light absorbed at the excitation wavelength used as reference. I f ¼ ff I a ff I0 ¼
;
Ia ¼ I0 1 10A
;
ff ¼
If 1 10A
If I0 1 10A
;
where If, Ia and I0 being the fluorescence intensity, the intensity of light absorbed and the intensity of incident light,
Table 1 e Absorption and fluorescence properties of DOC isolate.
Pinail PPHPOA Suwannee SWHPOA South Platte SPHPOA Irradiated Pinail
DOC (mg C L1)
A315 (cm1)
SUVA315 (L mg C1 cm1)
If (315 nm) x106
If x107 1 10A
8.4 8.7 8.9 7.7
0.185 0.174 0.098 0.102
0.022 0.020 0.011 0.013
4.85 5.85 4.20 2.52
1.40 1.77 2.08 1.20
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 4 1 e3 3 5 0
3.2.
E1 photodegradation in the presence of DOC
Photodegradation of estrone has been performed in the presence of the three different DOC extracts (Pinail FA, Suwannee FA and South Platte FA). All solutions contained approximately 10 mg C L1 of DOC and the pH was adjusted to pH 7.3. The kinetics of E1 photodegradation in the presence of different FA is presented in Fig. 1a. For comparison, the kinetics of E1 photodegradation in purified water (in the absence of DOC isolate) is also plotted. The intensity of the lamp was always set at 250 W m2. Even in the absence of DOC, the photodegradation of E1 is quite rapid; 50% of E1 was degraded after 8 h of irradiation (no degradation in the vial surrounded with aluminum foils). This means that direct phototransformation of estrone will occur in sunlit surface waters. This result had already been described by Mazellier et al. (2008) for 17b-estradiol and 17aethinylestradiol, but with an irradiation device that was not completely representative of sun emission. In the presence of DOC and whatever its origin, an enhancement of E1 photodegradation was observed (Fig. 1a). DOC sample from South Platte (SPHPOA) was shown to be the most efficient to photoinduce the degradation of E1. The percentages of E1 photodegradation in 8 h of irradiation were obtained as follows: SPHPOA (76%) > PPHPOA (66%) > SWHPOA (56%). This result is in agreement with the optical properties of the three DOC samples; the higher the fluorescence efficiency, the stronger the photoinductive properties. A similar trend was established by Chasson (2003) for the photoisomerisation/ photodegradation of sorbate in the presence of different DOC isolates, and by Richard et al. (2004), drawing a link between the fluorescence properties of different fractions of HAs and their photoinductive activity for the phototransformation of TMP. As previously mentioned, the kinetics of E1 obeys an apparent first order kinetic law. The first order rate constants obtained from these curves ðk meas: Þ are gathered in Table 2. In the presence of DOC isolates, the direct phototransformation of E1 cannot be neglected upon sunlight irradiation, even if the rate of light absorption by E1 is very low (with a molar absorption coefficient evaluated to be 60 M1 cm1 at 315 nm). Therefore the apparent first order constant k meas: corresponds to two degradation phenomena: the direct phototransformation and the transformation induced by the presence of DOC. The direct phototransformation occurs with
a
1
0,8
[E1]/[E1]0
respectively, ff the quantum yield of fluorescence and A the absorbance of the solution at the excitation wavelength. If The so-calculated values are directly proportional 1 10A to fluorescence quantum yields and can be compared. Results showed that South Platte FA presented the highest fluorescence efficiency followed by Suwannee FA and Pinail FA. This finding can be explained by a lower inter/intramolecular fluorescence quenching effect of the South Platte FA associated with a more aliphatic structure. It is worth noting that the rank is the opposite to that obtained by comparison of corrected UV absorbance SUVA315 which gave Pinail FA > Suwannee FA > South Platte FA. This inverse relationship for fluorescence and UV absorbance should be confirmed by studying additional FA.
0,6 Purified water PPHPOA
0,4
SWHPOA SPHPOA
0,2 0
2
4
6
8
Irradiation time (h)
b
1
0,8
[E1]/[E1]0
3344
0,6
Purified water PPHPOA PPHPOA + NaN3
0,4
PPHPOA + propanol PPHPOA + propanol + NaN3
0,2 0
2
4
6
8
Irradiation time (h) Fig. 1 e Kinetics of E1 photodegradation (a) in the absence and in the presence of the 3 DOC isolates (b) in the presence of scavengers (azide ions and 2-propanol).
an apparent first order kinetic constant of 0.09 h1 in the absence of DOC. It is therefore possible to calculate the value of E1 direct photolysis first order rate constant in the presence of DOC, if DOC only acted as an inner filter (i.e. without any photosensitizing properties). The initial fraction of light absorbed by E1 alone and in the presence of different DOC extracts have been calculated at 315 nm according to the following equation: Ia Ia AE1 ¼ I0 E1=m I0 m Am Ia represents the fraction of light absorbed by E1 I0 E1=m Ia represents the fracin the presence of DOC at 315 nm, I0 m tion of light absorbed by the mixture E1/DOC (it mainly corresponds to the light absorbed by DOC) at 315 nm, AE1 represents the absorbance of E1 (constant as E1 concentration was constant), Am represents the absorbance of the mixture, both at 315 nm. The apparent first order rate constant of E1 direct phototransformation is proportional to the fraction of light absorbed where
3345
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 4 1 e3 3 5 0
Table 2 e Apparent first order rate constants, and half-lives in different experimental conditions (different DOC isolates or in the presence of scavengers). DOC solution (mg C L1)
k
Purified water
0.090 0.006
0.090
e
7.9
PPHPOA (8.4) SWHPOA (8.7) SPHPOA (8.9)
0.122 0.007 0.100 0.006 0.178 0.004
0.070 0.071 0.077
43% 29% 57%
5.3 6.8 3.9
PPHPOA (8.4) þ N3 PPHPOA (8.4) þ 2-propanol PPHPOA(8.4) þ 2-propanol þ N3
0.100 0.007 0.100 0.008 0.070 0.008
0.070 0.070 0.070
Irradiated PPHPOA (7.7)
0.100 0.008
0.086
14%
7.0
meas:
(h1)
k
calc:
(h1)
Contribution of DOC to degradation
s½ (h)
Standard deviations have been determined by using 95% confidence interval. kE1 ¼ 0.090 0.006 corresponds to the measured first order rate of E1 direct photolysis in pure aqueous solution, k meas: corresponds to the measured first order rate constant, k calc: corresponds to the calculated direct photolysis first order rate constant by taking into account the filter effect of DOC.
by E1 in the mixture. In the presence of DOC, the light emitted by the lamp is mainly absorbed by DOC. Therefore, according to the following equation, the values of k calc: can be calculated: Ia I0 E1=m k calc: ¼ kE1 Ia I0 E1 Ia with being the fraction of light absorbed by E1 in pure I0 E1 solution, kE1 the first order apparent rate constant of E1 phototransformation in the absence of DOC. k calc: stands for the calculated first order rate constant of E1 degradation in the presence of DOC acting only as a inner filter. The so-calculated k calc: values are gathered in Table 2. They can be compared to the measured k meas: . This approach provides more quantitative information on how important the presence of DOC accelerates the phototransformation of E1. With South Platte FA, about 60% of the overall degradation of E1 is due to photosensitized reactions. Similar results but with lower participation were obtained for Pinail FA and Suwannee FA (about 50% for Pinail and about 30% for Suwannee). The enhancement of contaminant degradation through DOC inductive photolysis has already been described by numerous authors (e.g. Simmons and Zepp, 1986). Aguer and coworkers demonstrated that few commercial, synthetic or natural humic substances can present different photoinductive properties leading to the enhancement of organic compounds phototransformation (Aguer et al., 1997). The absorption of solar light by DOC lead to the formation of singlet and triplet excited states from which further generation of highly reactive species: singlet oxygen, solvated electrons or hydroxyl and peroxyl radicals (Canonica et al., 1995; Aguer and Richard, 1996; Aguer et al., 1999). These reactive species are key intermediates for the further degradation of pollutants. Several species may be involved in the phototransformation of E1 in the presence of DOC. Experiments have been performed with the addition of 2 mM of azide ions (as a singlet oxygen scavenger) or 0.26 M of 2-propanol (as a hydroxyl radicals scavenger) or both scavengers at the same time. The results obtained in the presence of Pinail extract are presented in Fig. 1b as an example. It can be seen that E1
photodegradation is less important in the presence of scavengers. This is confirmed by the calculation of first order rate constants (Table 2). However, the degradation remains always slightly more important than the direct photolysis itself and even more when we consider the inner filter effects of DOC (0.122 h1 without scavengers, 0.100 h1 in the presence of a scavenger to be compared to 0.070 h1 for the corrected photolysis rate constant - Table 2). The experiments performed with both scavengers led to a slower degradation than the direct photolysis alone. After correction for the absorption of light by DOC, the apparent first order of E1 photolysis was evaluated to be 0.07 h1 which corresponded to the first order rate constant determined in the presence of both scavengers. The main reactive species responsible for the degradation of E1 in the presence of DOC are hydroxyl radicals and singlet oxygen. However, the participation of other reactive species, not investigated here, cannot be ruled out since scavengers are not completely selective: excited triplet state, solvated electrons or peroxyl radicals as mentioned by many authors (Zepp et al., 1985, 1987; Canonica et al., 1995). When reactive species are formed from DOC irradiation, a competition exists between the reaction with organic compounds and DOC itself (see for example Zhang et al., 2007). In a recent paper, Leech et al. (2009) studied the simulated sunlight photodegradation of 17b-estradiol (E2) in the presence of DOC from Suwannee River. A first order rate constant of 0.05 h1 was obtained for a starting concentration of 1.0 mM in E2 and an irradiance dose of 650 W m2. In our case, E1 photodegradation is more efficient since the rate constant is 0.09 h1 at an irradiance dose of 250 W m2. The difference between E1 and E2 is the carbon in position 17 (alcohol function for E2 and aldehyde function for E1). Leech et al. observed a quite efficient photodegradation of E2 in the presence of DOC, increasing when the concentration of DOC increased. They have also shown the involvement of reactive species by using 2-propanol as a scavenger.
3.3. Impact of solar light exposition on DOC structure and efficiency of E1 photodegradation The ability of DOC isolates to promote pollutants photodegradation varies with their structural characteristics. The
3346
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 4 1 e3 3 5 0
photoinductive transformation of E1 was investigated with a DOC extract that was previously subjected to long term (up to 64 h) solar light exposition in order to modify its chemical structure and optical properties. Aqueous solutions of Pinail
a
0.4 t=0 t=1d t=3d
Absorbance
0.3
t=5d t=7d 0.2
0.1
0 200
300
400
500
600
Wavelength (nm)
b
6.E+06 t=0
Fluorescence emission
t=1d t=3d t=5d
4.E+06
t=7d
2.E+06
0.E+00 300
400
500
600
700
FA were exposed to simulated sunlight (suntest) for an entire week. In order to characterize structural changes, aliquots were removed at selected reaction times. The samples were characterized by UVeVisible and fluorescence spectroscopy to observe the alteration of chromophores and fluorophores. Sunlight exposure led to a progressive decrease with time of UVeVisible absorbance (Fig. 2a) and fluorescence emission (Fig. 2b) as previously described in literature (Carvalho et al., 2008). After one day of irradiation, a decrease of about 25% was obtained for the absorbance value (300 and 315 nm) whereas the decrease was 40% for the fluorescence emission. The phototransformation of DOC under simulated sunlight irradiation leads to the degradation of aromatic moieties or to the modification of the structures resulting in the alteration of the fluorescence properties. On the contrary, the reduction in carbon content of the irradiated solution is minor; the DOC decreased from 8.3 to 7.7 mgC L1 after 5 days of simulated sunlight irradiation. The evolution of the HPSEC-UV chromatograms of the irradiated Pinail FA aqueous solutions is represented in Fig. 2c. During the photolysis, the peaks corresponding to high molecular weight compounds (HMW, retention time about 8 min) decreased. Simultaneously, an increase of lower molecular weight (LMW) compounds was observed (between 12 and 14 min). Sunlight exposition of DOC led to the degradation of HMW to LMW compounds as already mentioned in literature (Osburn et al., 2001; Brinkmann et al., 2003; Carvalho et al., 2008). The simultaneous investigation of HPSEC chromatograms, UVeVisible and fluorescence spectra at different reaction times brings to the conclusion that photodegradation of hydrophobic molecules with a high molecular weight like aromatic compounds leads to the formation of smaller molecules more hydrophilic and with lower abilities to absorb light and emit fluorescence. These DOC samples were tested as photosensitizers for E1 degradation. E1 was added to a 64 h-irradiated DOC solution and then subjected to a maximum of 4 h of irradiation time. HPLC-UV analyses were carried out on the irradiated sample.
Emission wavelength (nm) 0.1 t=0
0.075
1
t=1d
300 nm
t=3d
315 nm
t=5d
0.05
0.025
Normalized Intensity
Absorbance Units (a.u.)
c
0.8
Exc. 315 nm Em. 460 nm
0.6
0.4
0 5
7
9
11
13
15
Retention time (min) Fig. 2 e Evolution of UVeVisible spectra (a), fluorescence emission spectra with lexc. [ 315 nm (b) and UV-HPSEC chromatograms (c) during Pinail FA photodegradation upon a week.
0.2 0
2
4
6
Irradiation time (d) Fig. 3 e Evolution of UV absorbance (at 300 and 315 nm) and fluorescence emission (lexc. [ 315 nm, lem. [ 460 nm) during Pinail FA photodegradation for a week.
3347
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 4 1 e3 3 5 0
part of photoinduced reaction has been evaluated to 14% with pre-irradiated Pinail FA whereas it was about 46% with initial Pinail FA (Table 2) corresponding to a decrease by a factor of 3. So, the major degradation of the aromatic moieties (41% of absorbance reduction at 315 nm; 48% of fluorescence efficiency, Fig. 3) strongly diminished the initial photoinductive properties of the Pinail FA toward E1 degradation as demonstrated by the values of first order rate constants.
1
[E1]/[E1]0
0.875
0.75
3.4.
Purified water
0.625
Pinail
The identification of photoproducts and the elucidation of their chemical structures are major challenges scientists are facing. This is particularly true for emerging pollutants that are weakly soluble such as hormones. In this work, aqueous solutions of E1 (with or without DOC) were photolyzed at different irradiation times. These samples were analyzed by LC-MS and GC-MS after an extraction-concentration step performed by SPE as described by Mazellier et al. (2008). In the absence of DOC, a unique photoproduct (P1) was observed at a retention time higher than E1, by direct HPLC analyses (without any extraction-concentration step). After SPE extraction, the UVeVisible and mass spectra concerning this photoproduct have been obtained. The UVeVisible spectrum of P1 is identical to that of E1 and GC-MS spectra of these two compounds present the same major fragments (Table 3). LCMSn experiments have been performed in positive APCI for E1 and P1. MS, MS2 and MS3 spectra were identical (Table 3). This set of results strongly supported the assumption that P1 was an isomer of E1. As the UV spectrum of P1 was identical to that of E1, this certainly means that the aromatic moiety remained unchanged and that structural changes were on the steroid moiety. Therefore, according to the structure of E1 and the similar characteristics of UV and mass spectra, the isomeric structure of P1 is quite difficult to put forward and the
Pre-photolysed 0.5 0
1
2
3
Identification of E1 photoproducts
4
Irradiation time (h) Fig. 4 e E1 photodegradation in the presence of prephotolyzed Pinail FA (64 h).
As previously mentioned, Pinail FA optical properties have changed with a high decrease of UV absorbance and fluorescence emission (Fig. 3), causing the decrease of its photoinductive efficiency as it is shown in Fig. 4. E1 photodegradation in the presence of the pre-photolyzed DOC is almost identical to what was obtained in purified water. Table 1 shows a strong decrease of the optical properties of pre-irradiated Pinail FA (t ¼ 64 h). The modified DOC lost 41% of SUVA315 and 48% of its maximum fluorescence emission. It also shows that the fluorescence efficiency (fluorescence quantum yield) value only decreases by 15% indicating a quite low relative degradation of Pinail FA fluorophores. However, pre-irradiated Pinail FA is less efficient to photoinduce E1 degradation. The
Table 3 e Major mass peaks and fragments observed in GC/MS and LC/MSn analyses of an aqueous solution of E1 irradiated for 8 h. For LC-MSn the ions in italics corresponds to the selected and fragmented ion. Compound
GC
Major fragments in MS mode or daughters ions in MSn mode
Structure
LC
E1
MS
MS
MS2
MS3
270.1 (100) 185.1 (36) 172.1 (27) 146.1 (32)
271.5 (100) 253.5 (8)
253.1 (100) 197.3 (16) 157.3 (13)
253.2 (86) 197.2 (100) 157.2 (71)
O
HO
P1
270.1 (100) 185.1 (34) 172.1 (26) 145.9 (17)
271.5 (100) 253.6 (30)
253.0 (100)
253.2 (100) 197.2 (22) 157.2 (67)
OH
HO
3348
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 4 1 e3 3 5 0
the presence of different DOC isolates. The kinetics is presented in Fig. 5. The area corresponding to the chromatographic peak of P1 has been plotted as a function of the percentage of E1 degradation to take into account the difference of E1 phototransformation rate. It can be seen that P1 was formed in weaker quantity when DOC was added. If P1 is an isomer of E1, it is certainly formed by direct photolysis and as previously mentioned the part of direct photolysis decreased when DOC was added. Therefore the amount of P1 also decreased. As an example, we have compared the formation of P1 in the presence and in the absence of any DOC. The apparent first order rate constants of E1 degradation (Table 2) have shown a factor of 2 between the experiments performed without DOC and those performed with South Platte FA, with about 40% of direct photolysis. The formation of P1 in the presence of South Platte FA was almost lower by a factor of 2 by comparison with that obtained in purified water, for an identical E1 conversion percentage. This is strongly in favor of our hypothesis of a photoisomer structure. In order to better observe other degradation products mainly formed with DOC, photodegradation experiments were performed with E1 solutions at higher concentrations with the presence of DOC at about 10 mg L1. A gradient elution permitted the separation of 5 major peaks, beside that of E1 and P1 at 43 and 47 min, respectively. Photoproducts P2, P3, P4, P5 and P6 are detected at retention times of 7, 15, 22, 27 and 37 min as shown in Fig. 6. They are very weak upon irradiation in purified water. Their formation was mainly linked to the presence of DOC. P2 is characterized by
0.4 P1 Purified water P1 Pinail 0.3
P1 Suwannee P1 South Platte
0.2
0.1
0 0
20
40
60
80
E1 degradation (%) Fig. 5 e Photoproduct P1 formation vs E1 degradation in purified water and in the presence of the three DOC extracts.
assumption that P1 corresponds to a photo enol structure should be further investigated (see Table 3). As previously stated, this photoproduct is the unique degradation product observed in direct analyses by HPLC-UV with a starting E1 concentration of 0.8 mM. Therefore, the formation and depletion of this product has been measured in
42.78
100
m/z 269
90
E1
P1 46.95
Relative Abundance
80 70 60 50 40 30 20 10 15.02 14.91
0 100
m/z 285
34.06
P3
90
Relative Abundance
80 70 60
21.81
P4
50
P5
40 30
P6
27.28
20
26.65 36.68
10 15.71 18.22
7.0 3
0 5
10
15
20
23.58
2 9.2 8
25
30
35 .43 35
43.24 40
45
50
Time (min)
Fig. 6 e LC-MS selected ions chromatograms obtained in negative APCI mode (trE1 [ 42.78 min) after 24 h of irradiation of E1 in the presence of DOC (10 mg LL1).
3349
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 4 1 e3 3 5 0
Table 4 e Chemical structures proposed from LC-MS analyses for E1 photoproducts after 24 h suntest irradiation in the presence of DOC. Photoproduct
Negative molecular ion
Positive molecular ion
Structure
P3, P4, P5, P6
O
O HO
HO
HO OH
285
287/269
O
O
HO
HO
HO OH
a molecular ion at 258 uma in negative APCI and at 227 uma in positive one. It may correspond to a mixture of non separated degradation products (supported by the shape of the UV spectrum without any maximum). It has not been possible to propose any structure corresponding to this peak. The chromatographic peak P4 certainly corresponded to at least 2 products as it can be seen in Fig. 6. Peaks P3, P4, P5 and P6 presented mainly a molecular ion at 285 uma in negative APCI and 269 in positive APCI. 269 uma was scarcely observed certainly due to a deshydratation occurring in the ionization source. These masses are strongly in agreement with hydroxylation processes. Hydroxylation can take place onto the aromatic cycle (in ortho position of the phenolic group i.e. position 2 and 4) or on alicyclic ring (position 6 and 9). Structures proposed for P3, P4, P5 and P6 are shown in Table 4. Compounds as 2-OH E1 and 4-OH E1 (P5 and P6) are known to be metabolites of E1 (Hsu et al., 2007). Monohydroxylations on aromatic and alicyclic groups have been observed during UV and polychromatic photolysis of 17b-estradiol E2 (Mazellier et al., 2008). MSn fragmentations do not bring additional informations since only water elimination was observed (difference of 18 uma). This remains a strong limitation of the technique with this kind of compound.
4.
Conclusion
This study demonstrated that estrone (E1) can be degraded by sunlight. A half-life of 8 h has been observed in purified water and kinetic constants have been calculated. The ability of dissolved organic carbon (DOC) to photoinduce the degradation of E1 has been established for three DOC’s from different origins. In the presence of DOC, E1 photodegradation was more efficient. The addition of scavengers has permitted to better understand the participation of reactive species as singlet oxygen and hydroxyl radicals, even if the participation of other mechanisms like hydrogen atom transfer or the
formation of DOC excited triplet states directly reacting with E1 cannot be ruled out. The ability of DOC for photoinduced degradation varies according to its origin and its optical and physico-chemical properties. Transformation and degradation of DOC by direct photolysis lead to the cleavage of aromatic structures of DOC and of high molecular size structures. Results have shown that the photodegradation involves the decrease of DOC aromaticity and the change of its optical and physico-chemical properties. This transformation was observed by UV and fluorescence spectrometry and chromatographic analyses. During this process, DOC UV absorbance and its fluorescence emission decrease. The pre-photodegradation leads to the loss of DOC photoinductive efficiency compared to unmodified DOC. So, optical and physico-chemical properties of DOC are important parameters to take into account in natural photochemistry. The formation of E1 photoproducts has also been studied. Four products have been observed, corresponding to a hydroxylation of estrone. Beside these products, one other compound appeared with the same mass than E1. This product is certainly an isomer of estrone (maybe the enol derivative). Our study demonstrate that estrone natural photodegradation in aquatic media is possible, the intensity of suntest xenon lamp (250 W m2) being closed to that of sunlight. The presence of DOC can lead to a faster degradation of E1. This phenomenon should be of major significance in photic zone of aquatic media. Furthermore, our study brings information about natural photodegradation of E1 in the environment but it should be prolonged by taking into account natural and seasonal variations of sunlight and the influence of water column on the penetration of light.
Acknowledgment Emilie Caupos thanks the Region Poitou-Charentes and CNRS for providing her the grant of her PhD. The authors thank the Region Poitou-Charentes for the financial support.
3350
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 4 1 e3 3 5 0
references
Ague, J.P., Richard, C., 1996. Reactive species produced on irradiation at 365 nm of aqueous solutions of humic acids. J. Photochem. Photobiol. A Chem. 93, 193e198. Aguer, J.P., Richard, C., Andreux, F., 1997. Comparison of the photoinductive properties of commercial, synthetic and soilextracted humic substances. J. Photochem. Photobiol. A Chem. 103, 163e168. Aguer, J.P., Richard, C., Andreux, F., 1999. Effects of light on humic substances: production of reactive species. Analusis 27, 387e390. Brinkmann, T., Sartorius, D., Frimmel, F.H., 2003. Photobleaching of humic rich dissolved organic matter. Aquat. Sci. 65, 415e424. Brion, F., Tyler, C.R., Palazzi, X., Laillet, B., Porcher, J.M., Garric, J., Flammarion, P., 2004. Impacts of 17b-estradiol, including environmentally relevant concentrations, on reproduction after exposure during embryo-larval-, juvenile- and adult-life stages in zebrafish (Danio rerio). Aquat. Toxicol. 68, 193e217. Canonica, S., Jans, U., Stemmler, K., Hoigne´, J., 1995. Transformations kinetics of phenols in water: photosensitization by dissolved organic material and aromatic ketones. Environ. Sci. Technol. 29, 1822e1831. Carvalho, S.I.M., Otero, M., Duarte, A.C., Santos, E.B.H., 2008. Effects of solar radiation on the fluorescence properties and molecular weight of fulvic acids from pulp mill effluents. Chemosphere 71, 1539e1546. Chasson, G., 2003. Direct photolysis and photolysis induced by natural aquatic organic matter of the diene function. [Photolyse directe et photolyse induite par la matie`re organique naturelle aquatique de la fonction die`ne]. PhD Thesis, Universite´ de Poitiers. Chowdhury, R.R., Charpentier, P., Madhumita, B.R., 2010. Photodegradation of estrone in solar irradiation. Ind. Eng. Chem. Res. 49, 6923e6930. Croue´, J.-P., 2004. Isolation of humic and non-humic fractions: structural characterization. Environ. Monit. Assess. 92, 193e207. Desbrow, C., Routledge, E.J., Brighty, G.C., Sumpter, J.P., Waldock, M., 1998. Identification of estrogenic chemicals in STW effluent. 1. Chemical fractionation and in vitro biological screening. Environ. Sci. Technol. 32, 1549e1558. Hohenblum, P., Gans, O., Moche, W., Scharf, S., Lorbeer, G., 2004. Monitoring of selected estrogenic hormones and industrial chemicals in groundwaters and surface waters in Austria. Sci. Total Environ. 333, 185e193. Hsu, J.F., Chang, Y.C., Chen, T.H., Lin, L.C., Liao, P.C., 2007. Evaluation of electrospray ionization and atmospheric pressure chemical ionization for simultaneous detection of estrone and its metabolites using high-performance liquid chromatography/tandem mass spectrometry. J. Chromatogr. B 860, 49e56. Kang, I.J., Yokota, H., Oshima, Y., Tsuruda, Y., Yamaguchi, T., Maeda, M., Imada, N., Tadokoro, H., Honjo, T., 2002. Effect of 17b-estradiol on the reproduction of Japanese medaka (Orysias latipes). Chemosphere 47, 71e80. Kolpin, D., Furlonge, E.D., Meyer, M.T., Thurman, E.M., Zaugg, S., Barber, L.B., Buxton, H.T., 2002. Pharmaceuticals, hormones, and others organic wastewater contaminants in U.S. streams, 1999-2000: a national reconnaissance. Environ. Sci. Technol. 36, 1202e1211.
Leech, D.M., Snyder, M.T., Wetzel, R.G., 2009. Natural organic matter and sunlight accelerate the degradation of 17 b-estradiol in water. Sci. Total Environ. 407, 2087e2092. Liu, B., Liu, X., 2004. Direct photolysis of estrogens in aqueous solutions. Sci. Total Environ. 320, 269e274. Lin, A.Y.-C., Reinhard, M., 2005. Photodegradation of common environmental pharmaceuticals and estrogens in river water. Environmental Toxicology and Chemistry 24 (6), 1303e1309. Martin-Mousset, B., Croue´, J.P., Lefe`bvre, E., Legube, B., 1997. Distribution and characterization of dissolved organic matter of surface waters. [Distribution et caracte´risation de la matie`re organique dissoute d’eaux naturelles de surface]. Water Res. 31, 541e553. Mazellier, P., Me´ite´, L., De Laat, J., 2008. Photodegradation of steroid hormones 17b-estradiol (E2) and 17a-ethinylestradiol (EE2) in dilute aqueous solution. Chemosphere 73, 1216e1223. Osburn, C.L., Morris, D.P., Thorn, K.A., Moeller, R.E., 2001. Chemical and optical changes in freshwater dissolved organic matter exposed to solar radiation. Biogeochemistry 54, 251e278. Panter, G.H., Thompson, R.S., Beresford, N., Sumpter, J.P., 1999. Transformation of a non-oestrogenic steroid metabolite to an oestrogenically active substance by minimal bacterial activity. Chemosphere 38, 3579e3596. Richard, C., Trubetskaya, O., Trubetskoj, O., Reznikova, O., Afanas’Eva, G., Aguer, J.P., Guyot, G., 2004. Key role of low molecular size fraction of soil humic acids for fluorescence and photoinductive activity. Environ. Sci. Technol. 38, 2052e2057. Rosenfeld, E.J., Linden, K.G., 2004. Degradation of endocrine disrupting chemicals bisphenol A, ethinylestradiol, and estradiol during UV photolysis and advanced oxydation processes. Environ. Sci. Technol. 38, 5476e5483. Sierra, M.M.D., Giovanela, M., Parlanti, E., Soriano-Sierra, E.J., 2005. Fluorescence fingerprint of fulvic and humic acids from varied origins as viewed by single-scan and excitation/ emission matrix techniques. Chemosphere 58, 715e733. Simmons, M.S., Zepp, R.G., 1986. Influence of humic substances on photolysis of nitroaromatic compounds in aqueous systems. Water Res. 20, 899e904. Ternes, T.A., Stumpf, M., Mueller, J., Haberer, K., Wilken, R.-D., Servos, M., 1999. Behavior and occurrence of estrogens in municipal sewage treatment plants. I. Investigations in Germany, Canada and Brazil. Sci. Total Environ. 225, 81e90. Thurman, E.M., Malcolm, R.L., 1981. Preparative isolation of aquatic humic substances. Environ. Sci. Technol. 15, 463e466. Wetzel, R.G., Hatcher, P.G., Bianchi, T.S., 1995. Natural photolysis by ultraviolet irradiance of recalcitrant dissolved organic matter to simple substrates for rapid bacterial metabolism. Limnol. Oceanogr. 40, 1369e1380. Zepp, R.G., Schlotzhauer, P.F., Sink, R.M., 1985. Photosensitized transformations involving electronic energy transfer in natural waters: role of humic substances. Environ. Sci. Technol. 19, 74e81. Zepp, R.G., Braun, A.M., Hoigne´, J., Leenheer, J.A., 1987. Photoproduction of hydrated electrons from natural organic solutes in aquatic environments. Environ. Sci. Technol. 21, 485e490. Zhang, Y., Zhou, J.L., Ning, B., 2007. Photodegradation of estrone and 17b-estradiol in water. Water Res. 41, 19e26.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 5 1 e3 3 5 8
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Municipal wastewater treatment and biomass accumulation with a wastewater-born and settleable algal-bacterial culture Yanyan Su a,b,*, Artur Mennerich b, Brigitte Urban a a
Faculty of Environmental Sciences and Engineering, Institute of Ecology and Environmental Chemistry, Leuphana University of Lueneburg, Lueneburg 21335, Germany b Campus Suderburg, Ostfalia University of Applied Sciences, Suderburg 29556, Germany
article info
abstract
Article history:
A wastewater-born and settleable algal-bacterial culture, cultivated in a stirred tank pho-
Received 17 October 2010
tobioreactor under lab conditions, was used to remove the carbon and nutrients in
Received in revised form
municipal wastewater and accumulate biomass simultaneously. The algal-bacterial
20 March 2011
culture showed good settleable property and could totally settle down over 20 min,
Accepted 23 March 2011
resulting in a reduction of total suspended solids from an initial 1.84 to 0.016 g/l. The
Available online 31 March 2011
average removal efficiencies of chemical oxygen demand, total kjeldahl nitrogen and
Keywords:
the average biomass productivity was 10.9 1.1 g/m2$d. Accumulation into biomass,
Algal-bacterial culture
identified as the main nitrogen and phosphorus removal mechanism, accounted for
Nutrient removal
44.9 0.4% and 61.6 0.5% of total inlet nitrogen and phosphorus, respectively. Micro-
Biomass accumulation
scopic analysis showed the main algae species in the bioreactor were filamentous blue-
N and P accumulation
green algae. Furthermore, denaturing gradient gel electrophoresis and 16S rDNA gene
16S rDNA gene
sequencing revealed that the main bacteria present in the photobioreactor were consortia
phosphate were 98.2 1.3%, 88.3 1.6% and 64.8 1.0% within 8 days, respectively, while
with sequences similar to those of Flavobacteria, Gammaproteobacteria, Bacteroidia and Betaproteobacteria. This study explores a better understanding of an algae-bacteria system and offers new information on further usage of biomass accumulated during treatment. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
The concept of algal-bacterial culture as an engineered system in domestic and industrial wastewater treatment has experienced increased momentum over the past few years (Bordel et al., 2009; de-Bashan et al., 2002; Garcia et al., 2000; Gutzeit et al., 2005; Medina and Neis, 2007; Munoz et al., 2005). It is especially favorable in regions with year-round high solar radiation and temperature as the removal is an entirely natural process. When illuminated, algae produce oxygen that can be used by aerobic bacteria to biodegrade pollutants
whilst, in return, they consume the carbon dioxide released from bacterial respiration (Oswald, 1988), which provides a cheaper and safer alternative to mechanical aeration and contributes to CO2 mitigation (Guieysse et al., 2002; Munoz and Guieysse, 2006). Another advantage of this technology is that more nitrogen and phosphorus could accumulate into the algal and bacterial biomass during the removal process. When using a bacterial system to treat acetonitrile (1 g/l), only 26% NeNHþ 4 was assimilated into biomass. Under the same conditions using algal-bacterial culture, 53% NeNHþ 4 was assimilated into algae
* Corresponding author. Faculty of Environmental Sciences and Engineering, Institute of Ecology and Environmental Chemistry, Leuphana University of Lueneburg, Lueneburg 21335, Germany. Tel.: þ49 582698861650. E-mail address:
[email protected] (Y. Su). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.03.046
3352
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 5 1 e3 3 5 8
and bacteria biomass (Munoz, 2005). Oswald also inferred that, under optimal operating conditions (e.g., sufficient light intensity and a proper bioreactor configuration), almost all the available ammonia nitrogen appeared in the form of algal cell material (Oswald and Gotass, 1957). All the above gives algalbacterial biomass, accumulated during wastewater treatment, the potential to be used as a fertilizer in agriculture (Benemann et al., 1977; Mulbry et al., 2005). The major limitation of the exploitation of this technology is the requirement for cost-effective biomass harvesting techniques. A technical separation unit consisting of filtration or centrifugation has to be applied (Mohn, 1988), but this will raise the operation cost. Adding chemicals such as Ca2þ or slaked lime will result in secondary pollutants (Imase et al., 2008; Nurdogan and Oswald, 1995). Using an immobilization system is another possible solution (Mallick, 2002; MorenoGarrido, 2008), but all media are costly and inefficient over a long operation time. Therefore, a more effective biomass harvesting strategy such as a settleable algae-bacteria system is required. The identification and biometry of the dominant algal species in the algal-bacterial culture were well-studied in previous works (Godos et al., 2009; Garcia et al., 2000; Oswald, 2003), but little information was available about the bacterial community involved in this symbiotic system. Investigation of the microbial composition and their functionalities could provide some insights into the biological catalytic and symbiotic mechanisms. Moreover, the purification capacity could potentially be improved by addressing microbial constraints. In this study, for the first time, a settleable algal-bacterial culture was cultivated from domestic wastewater, and its treatment efficiencies, biomass generation, N and P accumulation processes and microbial diversity were investigated.
2.
Material and methods
2.1.
Settleable algal-bacterial culture enrichment
was used as bacterial inoculum and nutrient supply. The settleable algal-bacterial culture was cultivated under laboratory conditions at around 19 C. The stirred tank photobioreactor (for culture enrichment) was made of transparent PVC 40 cm in depth and 29 cm in diameter. The total volume of the medium in the reactor was 14 l (approx. 25 cm in depth). Constant mixing was maintained using a magnetic stirring bar (100 rpm) to avoid algae sedimentation. Two compact fluorescent lamps (Sylvania, F20W/860/E27) were used to irradiate the tank with about 360 mE s1 m2 (measured at the top of the liquid surface) for a period of 12 h per day. In order to cultivate the settleable algal-bacterial culture, the mixing procedure was stopped every 23 h for 1 h and the floating biomass was discarded with a screen (0.5 mm). 600 ml pretreated wastewater was exchanged after sedimentation every 3 days to maintain a nutrient supply. After one month of cultivation, dark green and pea green microalgae were visible and distributed evenly in the reactor.
2.2.
The same laboratory-scale reactor system as in the cultivation process was used for the batch mode. The pretreated wastewater was used as feed for the reactor in the following experiments, unless otherwise stated. The characterization of the pretreated wastewater used in the different batches is shown in Table 1. Before starting the batch experiment, the algal-bacterial biomass was allowed to settle by stopping the stirrer for half an hour. At the end of each 8-days cycle, 12.5 l of suspension was removed and replaced by fresh wastewater as above. The irradiation device was the same as that for the cultivation process. The photoperiod was a 12 h light-12 h dark cycle. 150 ml samples for further analysis (see below) were collected near the midway of the reactor with a pipe every day, 4 h after starting the irradiation period.
2.3.
The algae inoculum was obtained from the second clarifier wall of the Suderburg municipal wastewater treatment plant (County of Uelzen, Lower Saxony, Germany). The collected algae solution (exposure to bacteria was unavoidable) was firstly settled down for 1 h, and then 30 g (wet weight) of settled solid was used as algae inoculum for algal-bacterial culture enrichment. The wastewater collected from the second clarifier at the same site was used as medium, and 600 ml pretreated wastewater collected from the same plant (after preliminary screening, grit removal and primary sedimentation process)
Experimental operation
Analytical procedures
The temperature and dissolved oxygen (DO) were measured near the midway of the reactor by using a microprocessor oximeter (Oxi 320/SET. WTW, Germany) coupled with an O2 sensor (CellOx 325, WTW, Germany). pH was determined using a Crison pH electrode (pH 197-S). Chemical Oxygen Demand (COD), Total Kjeldahl Nitrogen (TKN) and Total Suspended Solid (TSS) were analyzed according to DIN 38409-H 41(44), DIN EN 25663H11 and DIN ISO 11465 (DEV., 2002). NHþ 4 , total phosphorus and dissolved phosphorus ðPO3 4 Þ were determined according to DIN 38406-E5-1 and DIN EN ISO 6878-D11 (DEV., 2002) using a UV/Vis Spectrometer (Perkin Elmer, Lambda 40, USA). NO 3 and NO2
Table 1 e Characterization of wastewater. Parameter Chemical oxygen demand (COD) Total organic carbon (TOC) TKN Ammonium Total phosphorus
Unit
First batch
mg O2/l mg C/l mg N/l mg N/l mg P/l
132.7 3.0 49.8 1.2 25.7 0.3 14.6 1.0 4.9 0.1
Second batch 103.0 36.9 25.3 18.4 3.9
5.0 1.5 0.2 0.8 0.1
Third batch
Fourth batch
190.9 3.0 63.6 1.0 23.1 0.6 17.0 1.2 4.7 0.1
140.8 4.0 52.1 0.8 35.4 0.3 18.9 1.0 3.8 0.1
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 5 1 e3 3 5 8
were determined using an Ion Chromatograph (Dionex DX-100, USA) according to DIN EN ISO 10304-1 (DEV., 2002). Total Organic Carbon (TOC) and Inorganic Carbon (IC) were determined using a TOC analyzer (Elementar liqui TOC Ⅱ, Germany) according to DIN EN 1484-H3 (DEV., 2002). Before analysis of the above parameters in liquid, samples were membrane filtered (0.45 mm). To measure nitrogen or phosphorus in biomass, the samples were first divided into two identical parts. One part was homogenized, while another part was filtered to remove the solids. Nitrogen or phosphorus in biomass was calculated as the total nitrogen or phosphorus difference between the homogenized samples and the filtered samples. All the experiments were performed in duplicate. An optical microscope (OLYMPUS CHT, Japan) was used for morphological characterisation of microalgae.
2.4.
Community analysis
Bacteria of the algae-bacteria system were collected at the end of each batch test by centrifugation at 10,000 rev min1 for 10 min at 4 C. Each sample was washed twice with phosphate buffer (pH 7.0). Genomic DNA was isolated using the QIAamp DNA Stool Mini Kit (QIAGEN, 51504) according to the manufacturer’s instructions. The polymerase chain reaction (PCR), denaturing gradient gel electrophoresis (DGGE) and 16S rDNA analysis were done as described previously (Su et al., 2009; Schauer et al., 2000). Dominant bands were sequenced (Macrogen, the Netherlands). Sequences were subjected to Basic Local Alignment Search Tool (BLAST) and Ribosomal Database Project (RDP) analysis (Zhang et al., 2009). Phylogeny was determined with the RDP classifier and Sequmatch. The sequences used here have been deposited in the GenBank under the accession numbers HQ327478eHQ327485.
as nearly all the algal-bacterial biomass settled to the bottom of glass cylinder within 20 min, resulting in a reduction of TSS from an initial 1.84 to 0.016 g/l. The corresponding sludge settling ratio (SV %) was 12%, which also implied its good settleability (Sekine et al., 1984). Compared with an uncultivated algal-bacterial culture, the good settleability of the system might be due to the special cultivation strategy used in this study. The alternate mixing and non-mixing operation in the cultivation period promoted the selection of settleable algae and bacteria, which provided an effective way to harvest algalbacterial biomass. The harvest technology of this settleable system has three advantages over other algal-bacterial harvesting technologies (Mohn, 1988; Imase et al., 2008; Nurdogan and Oswald, 1995). First, the operating cost was greatly reduced, as no extra energy (or equipment) was required. Second, secondary pollutants were eliminated during operation and biomass harvest, as no extra chemicals were added. Third, the settleability could be guaranteed during long-term operation. The sedimentation was the characteristic of the cultivated algal-bacterial culture and not dependent on any immobilization medium which was inefficient over long time operation (Mallick, 2002; Moreno-Garrido, 2008). A microscopic photograph of the developed algal-bacterial flocs is shown in Fig. 2. From Fig. 2A, it can be seen that the main species in the bioreactor were filamentous blue-green algae. Fig. 2B shows the wastewater bacteria attached to the algae filaments, which forms the cooperative system. Obviously, binding mechanisms supporting the bio-flocculation process led to the formation of settleable biomass. There may be some factors responsible for the settling process, such as the algae cell surface properties, extracellular polymeric substances (EPS) and the content of cations, which will influence the formation and the stability of the settleable algal-bacterial biomass (Gutzeit et al., 2005).
3.
Results and discussion
3.2. Carbon and nutrients removal in municipal wastewater with algal-bacterial culture
3.1.
Settleability of the algal-bacterial culture
3.2.1.
The settleability of the cultivated algal-bacterial culture was investigated and is shown in Fig. 1. It showed good settleability,
3353
Temperature, dissolved oxygen and pH
Changes in temperature, pH and dissolved oxygen during operation were monitored to determine their effects on the treatment process. As shown in Fig. 3A, the culture temperature
Fig. 1 e Settleability of algal-bacterial biomass. (A) Initial completely mixed sample. (B) After 5 min of sedimentation. (C) After 20 min of sedimentation.
3354
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 5 1 e3 3 5 8
Fig. 2 e Microscopic photographs of algal-bacterial flocs. (A) Light microscope 3 100. (B) Light microscope 3 400. was around 12 C (the same as outdoor temperature) at the beginning of each batch test. It might be due to the fact that the wastewater was added into the bioreactor immediately after collection from the municipal wastewater plant. After that, the temperature increased to room temperature and remained stable until the end of each batch test. Similarly, at the beginning of each batch test, the dissolved oxygen concentration (DO) was the same as that of the pretreated wastewater. When starting the batch run, the DO values dropped significantly to around zero, which indicated
7
25 second batch
first batch
6
third batch
fourth batch 20
5 15
4 3
10
DO T
2
Temperature(°C)
5 1 0
0 0
4
8
12
16 20 Time(day)
24
28
32
3.2.2.
B 14
70
200
12 second batch
first batch
third batch
COD TOC
fourth batch
10
third batch
fourth batch
60
160 COD(mgO 2/l)
pH
Elimination of organic carbon
As shown in Fig. 4, the COD decreased with time and was lower than 3 mg/l at the end of each batch test. The COD and TOC removal efficiencies were around 98% and 75.2% for the four batches, respectively (Fig. 4). Both algae and bacteria were able to use organic carbon through mixotrophic or heterotrophic metabolism (Abeliovich and Weisman, 1978). It was obvious that carbon sources were eliminated significantly
8 6 4
first batch
50
second batch
120
40 30
80
TOC (mg/l)
Dissolved Oxygen(DO, mg/l)
A
that after initial consumption of the DO in the wastewater, the O2 released from the algal photosynthesis was almost consumed by processes such as heterotrophic carbon oxidation and nitrification. After three days, the oxygen concentration increased gradually to around 2 mg/l and continued to increase until it was around 5.5 mg/l (70% saturation) at the end of each batch run (Fig. 3A). No significant variation in culture pH during the four batches was detected in the system. Only a slight pH decrease occurred due to the intensive nitrification over the first five days. After that, pH increased gradually to 8.4 (see Fig. 3B). There are several factors which may influence the culture pH, such as micro-algal growth (pH increase as a result of CO2 uptake), NHþ 4 nitrification (pH decrease due to the release of Hþ) and the excretion of acidic or basic metabolites from organic matter biodegradation (Gonzalez et al., 2008b).
20
2
40 10
0
0
4
8
12
16 20 Time(day)
24
28
32
Fig. 3 e Changes in temperature, dissolved oxygen and pH in the algal-bacterial system. (A) Temperature and dissolved oxygen. (B) pH. Arrows indicate the start of a new batch test.
0
0 0
4
8
12
16 20 Time(day)
24
28
32
Fig. 4 e The concentration of COD and TOC in the algalbacterial system over four batch runs. Arrows indicate the start of a new batch test.
3355
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 5 1 e3 3 5 8
A
36 NH4+
NO3-
TKN
+
TKN (mg/l); N-NH 4 (mg/l) N-NO3 (mg/l)
32 first batch
28
second batch
fourth batch
third batch
24 20 16 12 8 4 0
B
0
4
8
12
16 Time(day)
first batch
5
20
24
28
32
third batch fourth batch
second batch 3P-PO4 (mg/l)
4 3 2 1 0
0
4
8
12
16 20 Time (day)
24
28
32
L 3L Fig. 5 e The concentration of TKN, NHD 4 , NO3 and PO4 L 3L over four batch runs. (A) TKN, NHD and NO . (B) PO 3 4 . 4 Arrows indicate the start of a new batch test.
3.2.4.
within the first four days (Fig. 4). This result was in agreement with that of previous studies (Gutzeit et al., 2005). The COD and TOC removal were slow after the fourth day. There were two possible reasons for this. First, the carbon source left in the system was small after the fourth day. Second, the remaining carbon may be some colloidal, slowly biodegradable material. Usually, carbon was the limiting factor when algae were cultured in sewage. However, algae may also obtain CO2 from the air in an open system, when the amount of carbon (inorganic and organic) in wastewater is inadequate for algal photosynthesis (Gonzalez et al., 2008a; Oswald, 1988).
3.2.3.
100%, while the TKN removal efficiency were slightly different, ranging from 76.6% to 97.8% (Fig. 5A). The TKN removal efficiency in the last two batches was slightly lower compared with that in the first two batches. Obviously, there was still some organic nitrogen (ON, the difference of TKN and NeNHþ 4 ) in the effluent. The ON might be made of a small amount of inseparable organic matter produced during algae growth and wastewater treatment processes Oswald and Gotass, 1957. Additionally, NeNO3 was always detected in the effluent due to incipient nitrification (around 5.2 mg/l). Nitrite was never detected at the end of each batch run. The nitrogen balance was also investigated for a better understanding of nitrogen removal mechanisms. Based on total nitrogen removal and biomass concentration (discussed later), nitrogen assimilation into biomass accounted for approx. 52.9 0.3%, 43.1 0.4%, 43.0 0.5% and 40.7 0.4% of the total inlet nitrogen in the four batches, respectively (Table 2). The contribution of ammonia volatilization to total nitrogen removal in this system could be negligible due to the low NHþ 4 concentrations and relatively low pH (<8.5). Conversion into nitrate only accounted for 20.0 0.2%, 18.3 0.3%, 19.0 0.1% and 17.4 0.1% of total inlet nitrogen in the four batches, respectively (Table 2). The remaining missing nitrogen might be due to denitrification processes, which could occur at DO below 2 mg/l (Godos et al., 2009).
Elimination of nitrogen
The influent nitrogen was mainly in the form of NeNHþ 4 (70e90%), followed by total organic nitrogen (10e30%). Fig. 5A shows the TKN, NeNHþ 4 removal and NeO3 generation process. þ NeNH4 removal efficiencies for all the four batches were nearly
Elimination of phosphate
The time course of PePO3 4 is shown in Fig. 5B. The removal of phosphate was a much slower process compared to that observed for nitrogen, but the same general pattern was apparent (Fig. 5B). The removal efficiencies of phosphate ranging from 54.5% to 72.6% were observed. The relatively slower and lower PePO3 4 removal efficiencies, compared with , were probably due to the fact that nitrogen was the NeNHþ 4 limiting nutrient, not phosphate, in this system. Previous studies have shown that the optimal ratio for maximum nitrogen and phosphorus uptake by algal-bacterial culture is N:P ¼ 30:1 (Chevalier and de la Noue, 1985). However, the ratio of nitrogen to phosphorus was lower than 3 in this study. Similarly, the balance of phosphorus was also investigated. Phosphorus accumulation into the biomass was still the main mechanism in this system, which accounted for 62.6 0.5%, 66.6 0.4%, 50.1 0.6% and 67.1 0.5% of the total inlet phosphorus for the four batches, respectively (Table 2). Phosphorus can be eliminated through both biotic phosphorous assimilation into the biomass and abiotic phosphorous precipitation (Godos et al., 2009). Nurdogan and Oswald (1995) reported on abiotic P removal which took place mainly in the
Table 2 e Nitrogen and phosphorus balance over the four batch tests. Batch Inlet TN Outlet TN Inlet nitrogen (mg N/l) (mg N/l) oxidized in NO 3 (%) 1 2 3 4
25.8 25.5 23.3 35.5
1.3 1.0 1.8 1.3
6.8 5.4 10.1 11.7
1.5 0.9 1.0 0.7
20.0 18.3 19.0 17.4
0.2 0.3 0.1 0.1
Inlet nitrogen accumulated in biomass (%) 52.9 43.1 43.0 40.7
0.3 0.4 0.5 0.4
Inlet Phosphorus (mg P/l)
Outlet Phosphorus (mg P/l)
4.9 0.1 3.9 0.1 4.7 0.1 3.8 0.1
1.7 0.1 1.1 0.1 2.1 0.1 1.2 0.1
Inlet phosphorus accumulated in biomass (%) 62.6 66.6 50.1 67.1
0.5 0.4 0.6 0.5
3356
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 5 1 e3 3 5 8
form of orthophosphate precipitation at high pH (9e11). In our system, the pH below 9 was not sufficient to promote this removal mechanism.
(e.g., illumination conditions and the irradiated water surface area of the reactor) (Chevalier and de la Noue, 1985).
3.3.2. 3.3. Biomass generation and nutrients accumulation processes 3.3.1.
Biomass generation during the operation
As shown in Fig. 6A, the TSS increased during operation, from 0.6 g/l at the beginning to 1.84 g/l at the end. The mean biomass generation rate was 10.9 1.1 g/m2 d. These values were lower than those reported by previous studies, in which maximum biomass productivities of 27.7 g/m2 d were observed with 10 times diluted swine manure (2417 481 mg COD/l; 214 53 mg NHþ 4 /l) during June and August in high rate algae ponds (HRAPs) in Valladolid, Spain (average irradiations 7062 81 Wh/m2 d) (de Godos et al., 2009). The lower availability of carbon and nitrogen in municipal wastewater compared with diluted swine manure, together with the lower irradiation, may be possible reasons for this. Although there was continuous algal and bacterial growth in the system during the treatment process, the removal efficiencies of both carbon and nutrients and their general patterns were quite similar for the four batches. One possible reason for this might be that the further increased algae concentration has less effect on the uptake rate, since the nutrient uptake by algae was also determined by other factors
A
Nutrient accumulation processes
The accumulation of nitrogen and phosphorus in biomass during the treatment process is shown in Fig. 6B. The nitrogen concentration in biomass increased from 27.8 to 32.6 mg/g, at a rate of nearly 0.15 mg/g d. Meanwhile, the phosphorus concentration in the biomass increased from 0.72 to 4.74 mg/g, at a rate of 0.12 mg/g d. The most rapid accumulation period of nitrogen and phosphorus occurred in the first batch. After that, the accumulation rates were much lower and the amount of nitrogen and phosphorus in biomass was stable. It has been reported that the biomass generated from a swine manure removal process, ranging from 22.9 to 76.6 mg N/g biomass and from 4.3 to 25.2 mg P/g biomass, had fertilizer value to promote plant growth (Mulbry et al., 2005; Mulbry et al., 2006). The concentrations of nitrogen and phosphorus in biomass in this study were comparable to those of the swine manure treatment system. However, it is worth noting that the nutrient concentration in municipal wastewater is much lower than that of swine manure, so it leaves room to improve the accumulation of nitrogen and phosphorus in the settleable algal-bacterial culture.
1.9 1.7 third batch
TSS(g/l)
1.5 second batch
1.3
fourth batch
first batch
1.1 0.9 0.7
B
0
4
8
12 16 Time(day)
24
28
32
34
8 fourth batch
33 Nitrogen in biomass. (mg/g)
20
second batch
32
7
third batch
6
first batch 31
5
30 4 29 3
28
2
27
N in biomass P in biomass
26 25
phosphorus in biomass. (mg/g)
0.5
0
4
8
12
16 20 Time(day)
24
28
1 32
0
Fig. 6 e Changes in total suspended solids, nitrogen and phosphorus accumulation in biomass over four batch runs. (A) Total suspended solids. (B) Nitrogen and phosphorus concentration in biomass. Arrows indicate the start of a new batch test.
Fig. 7 e DGGE bands of bacteria communities. W: wastewater; B1 to B4: batch 1 to 4; 1 to 11: the name of each band.
3357
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 5 1 e3 3 5 8
Table 3 e DGGE 16S rDNA band identifications. Band
Run
Genbank accession no.
Closest relatives (%Sequence similarityd)
Classc
HQ327478 HQ327479 HQ327480 HQ327481 HQ327482 HQ327483 HQ327484 HQ327485
Uncultured Flavobacteria bacterium clone ATB-LH-7295 (96%) Uncultured bacterium isolates DGGE gel band a3 (81%) Uncultured bacterium clone sls1360 (97%) Uncultured bacterium clone LL141-8H16 (92%) Dysgonomonas sp. enrichment culture clone YFZ1 (100%) Uncultured bacterium 6week9 (94%) Uncultured Bacteroidetes bacterium clone 298 (83%) Uncultured bacterium clone RW6944 (86%)
Flavobacteria Gammaproteobacteria Flavobacteria Bacteroidia Bacteroidia Bacteroidia Bacteroidia Betaproteobacteria
Wa B1 B2 B3 B4 b
1 2 3 4 5 6 7 8 a b c d
Inoculum. Existence under the condition. The phylotypes were assigned to phyla based on Ribosomal Database Project II (RDP II) taxonomy classifications. Percent values represent similarities between the associated DGGE band sequence and the closest match sequence from GenBank.
3.4.
Community analysis
The DGGE profiles of the bacteria community sampled from the reactor at the end of each batch run as shown in Fig. 4 are summarized in Fig. 7. It was clear that the bacterial populations changed with time and became stable after operation through three batches (Fig. 7). At the same time, bacteria related to nutrient removal might have been enriched and stable, as indicated by the high removal efficiency observed along the successive tests, as shown in Fig. 4 and Fig. 5. Some bacteria in the acclimatized bacteria consortium were not present in the inoculum, suggesting that some new communities were enriched after operation. The above results also indicate that the bacteria community in algal-bacterial culture required time to acclimate to this commensalism system. In order to provide greater insight into microbial ecology and diversity, eight predominant species extracted from DGGE bands were sequenced. Based on the 16S rDNA gene library results (Table 3), the acclimatized bacteria consortium was predominated by Bacteroidia (50% of clones), followed by Flavobacteria (25% of clones), Betaproteobacteria (12.5% of clones) and Gammaproteobacteria (12.5% of clones). It has been observed that Flavobacteria and Bacteroidetes phylotypes were present in high numbers in ammonia-oxidizing processes (Ducey et al., 2010; Nakano et al., 2008; Zang et al., 2008). Bafana et al. (2007) also observed that a Gammaproteobacteria (54%) phylotype was dominant in the acclimatized sludge of wastewater treatment plants. It was noticed that the band 1, 3 and 5 were stronger than other bands during operation (Fig. 7), thus their respective microorganisms might play a very important role in wastewater nutrient removal.
4.
Conclusion
A settleable algal-bacterial culture, cultivated from wastewater, was successfully used to treat municipal wastewater in a stirred tank photobioreactor. The algal-bacterial culture showed good settleability, while the total suspended solid could be reduced to 0.016 g/l within 20 min sedimentation. The average removal efficiencies of COD, TKN and phosphate
were 98.2 1.3%, 88.3 1.6% and 64.8 1.0% within 8 days, respectively, while the average biomass productivity was 10.9 1.1 g/m2$d. Biomass uptake was the main mechanism for nutrient removal. The main algae species in the bioreactor were filamentous blue-green algae, while the main bacteria present in the photobioreactor were a consortium with sequences similar to Flavobacteria, Gammaproteobacteria, Bacteroidia and Betaproteobacteria. This study provides new insights into rapidly settleable algal-bacterial culture enrichment strategies and supplements the information on microbial ecology and diversity in algal-bacterial culture.
Acknowledgments The authors thank Yifeng Zhang (Technical University of Denmark, Demark) for help and advice on the molecular biotechnology work, Sabine Henschel and Kerstin Lammersfor advice on chemical water analysis work and Ingeborg Joost and Pamela Holweg (Ostfalia University of Applied Sciences, Germany) for their kind assistance. This work was supported by Deutscher Akademischer Austausch Dienst (DAAD) and the LiWa project.
references
Abeliovich, A., Weisman, D., 1978. Role of heterotrophic nutrition in growth of the alga scenedesmus obliquus in high-rate oxidation ponds. Applied and Environmental Microbiology 35 (1), 32e37. Bafana, A., Devi, S.S., Krishnamurthi, K., Chakrabarti, T., 2007. Kinetics of decolourisation and biotransformation of direct black 38 by C. hominis and P. stutzeri. Applied Microbiology and Biotechnology 74 (5), 1145e1152. Benemann, J.R., Weissman, B.L., Koopman, B.L., Oswald, W.J., 1977. Energy production by microbial photosynthesis. Nature 268 (5615), 19e23. Bordel, S., Guieysse, B., Munoz, R., 2009. A mechanistic model for the reclamation of industrial wastewaters using algalbacterial photobioreactors. Environmental Science and Technology 43 (9), 3200e3207.
3358
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 5 1 e3 3 5 8
Chevalier, P., de la Noue, J., 1985. Efficiency of immobilized hyperconcentrated algae for ammonium and orthophosphate removal from wastewater. Biotechnology Letters 7 (6), 395e400. de-Bashan, L.E., Moreno, M., Hernandez, J.P., Bashan, Y., 2002. Removal of ammonium and phosphorus ions from synthetic wastewater by the microalgae Chlorella vulgaris coimmobilized in alginate beads with the microalgae growthpromoting bacterium Azospirillum brasilense. Water Research 36 (12), 2941e2948. de Godos, I., Gonzalez, C., Becares, E., Garcia-Encina, P.A., Munoz, R., 2009. Simultaneous nutrients and carbon removal during pretreated swine slurry degradation in a tubular biofilm photobioreactor. Applied Microbiology and Biotechnology 82 (1), 187e194. DEV., 2002. German Standard Methods for the Examination of Water, Wastewater and Sludge. WILEY-VCH, Weinheim, and Beuth, Berlin, Germany. Ducey, T.F., Vanotti, M.B., Shriner, A.D., Szogi, A.A., Ellison, A.Q., 2010. Characterization of a microbial community capable of nitrification at cold temperature. Bioresource Technology 101 (2), 491e500. Garcia, J., Muheriego, R., Hernandez-Marine, M., 2000. High rate algal pond operating strategies for urban wastewater nitrogen removal. Journal of Applied Phycology 12, 331e339. Godos, I., Blanco, S., Garcia-Encina, P.A., Becares, E., Munoz, R., 2009. Long-term operation of high rate algal ponds for the bioremediation of piggery wastewaters at high loading rates. Bioresource Technology 100 (19), 4332e4339. Gonzalez, C., Marciniak, J., Villaverde, S., Garcia-Encina, P.A., Munoz, R., 2008a. Microalgae-based processes for the biodegradation of pretreated piggery wastewaters. Applied Microbiology and Biotechnology 80 (5), 891e898. Gonzalez, C., Marciniak, J., Villaverde, S., Leon, C., Garcia, P.A., Munoz, R., 2008b. Efficient nutrient removal from swine manure in a tubular biofilm photo-bioreactor using algaebacteria consortia. Water Science and Technology 58 (1), 95e102. Guieysse, B., Borde, X., Munoz, R., Hatti-Kaul, R., NugierChauvin, C., 2002. Influence of the initial composition of algal bacterial microcosms on the degradation of salicylate in fed batch culture. Biotechnology Letters 24, 531e538. Gutzeit, G., Lorch, D., Weber, A., Engels, M., Neis, U., 2005. Bioflocculent algal-bacterial biomass improves low-cost wastewater treatment. Water Science and Technology 52 (12), 9e18. Imase, M., Watanabe, K., Aoyagi, H., Tanaka, H., 2008. Construction of an artificial symbiotic community using a Chlorella-symbiont association as a model. FEMS Microbiology Ecology 63 (3), 273e282. Mallick, N., 2002. Biotechnological potential of immobilized algae for wastewater N, P and metal removal: a review. Biometals 15 (4), 377e390. Medina, M., Neis, U., 2007. Symbiotic algal bacterial wastewater treatment: effect of food to microorganism ratio and hydraulic retention time on the process performance. Water Science and Technology 55 (11), 165e171.
Mohn, F.H., 1988. In: Borowitzka, M.A., Borowitzka, L.J. (Eds.), Harvesting of Micro-Algal Biomass. Cambridge, pp. 395e414. Moreno-Garrido, I., 2008. Microalgae immobilization: current techniques and uses. Bioresource Technology 99 (10), 3949e3964. Mulbry, W., Westhead, E.K., Pizarro, C., Sikora, L., 2005. Recycling of manure nutrients: use of algal biomass from dairy manure treatment as a slow release fertilizer. Bioresource Technology 96 (4), 451e458. Mulbry, W., Kondrad, S., Pizarro, C., 2006. Biofertilizers from algal treatment of dairy and swine manure effluents: characterization of algal biomass as a slow release fertilizer. Journal of Vegetable Science 12 (4), 107e125. Munoz, R. (2005) Algal-bacterial photobioreactors for the degradation of toxic organic pollutants. Phd thesis. Lund University. Munoz, R., Guieysse, B., 2006. Algal-bacterial processes for the treatment of hazardous contaminants: a review. Water Resource 40 (15), 2799e2815. Munoz, R., Jacinto, M., Guieysse, B., Mattiasson, B., 2005. Combined carbon and nitrogen removal from acetonitrile using algal-bacterial bioreactors. Applied Microbiology and Biotechnology 67 (5), 699e707. Nakano, M., Shimizu, Y., Okumura, H., Sugahara, I., Maeda, H., 2008. Construction of a consortium comprising ammoniaoxidizing bacteria and denitrifying bacteria isolated from marine sediment. Biocontrol Science 13 (3), 73e89. Nurdogan, Y., Oswald, W.J., 1995. Enhanced nutrient removal in high-rate ponds. Water Science and Technology 31 (12), 33e43. Oswald, W.J., 1988. In: Borowitzka, M.B.L. (Ed.), Micro-Algae and Waste-water Treatment. Cambridge, pp. 305e328. Oswald, W.J., 2003. My sixty years in applied algology. Journal of Applied Phycology 15, 99e106. Oswald, W.J., Gotass, H.B., 1957. Photosynthesis in Sewage treatment. American Society of Civil Engineers Transactions 122, 73e105. Schauer, M., Massana, R., Pedros-Alio, C., 2000. Spatial differences in bacterioplankton composition along the Catalan coast (NW Mediterranean) assessed by molecular fingerprinting. FEMS Microbiology Ecology 33 (1), 51e59. Sekine, T., Tsugura, H., Urushibara, S., Furuya, N., Fujimoto, E., Matsui, S., 1984. Evaluation of settleability of activated sludge using a sludge settling analyzer. Water Research 23 (3), 361e367. Su, Y., Zhang, Y., Wang, J., Zhou, J., Lu, X., Lu, H., 2009. Enhanced bio-decolorization of azo dyes by co-immobilized quinonereducing consortium and anthraquinone. Bioresource Technology 100 (12), 2982e2987. Zang, K., Kurisu, F., Kasuga, I., Furumai, H., Yagi, O., 2008. Analysis of the phylogenetic diversity of estrone-degrading bacteria in activated sewage sludge using microautoradiography-fluorescence in situ hybridization. Systematic and Applied Microbiology 31 (3), 206e214. Zhang, Y., Min, B., Huang, L., Angelidaki, I., 2009. Generation of electricity and analysis of microbial communities in wheat straw biomass-powered microbial fuel cells. Applied Environmental Microbiology 75 (11), 3389e3395.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 5 9 e3 3 6 8
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Adsorption of cellular peptides of Microcystis aeruginosa and two herbicides onto activated carbon: Effect of surface charge and interactions Petra Hnatukova, Ivana Kopecka, Martin Pivokonsky* Institute of Hydrodynamics, Academy of Sciences of the Czech Republic, Pod Patankou 5, 166 12 Prague 6, Czech Republic
article info
abstract
Article history:
In this research, the adsorption of two herbicides, alachlor (ALA) and terbuthylazine (TBA), on
Received 8 December 2010
granular activated carbon (GAC) in the presence of well-characterized peptide fraction of
Received in revised form
cellular organic matter (COM) produced by cyanobacterium Microcystis aeruginosa was studied.
25 March 2011
Two commercially available GACs were characterized using nitrogen gas adsorption and
Accepted 26 March 2011
surface charge titrations. The COM peptides of molecular weight (MW) < 10 kDa were isolated
Available online 6 April 2011
and characterized using MW fractionation technique and high-performance size exclusion chromatography (HPSEC). The effect of surface charge on the adsorption of COM peptides was
Keywords:
studied by means of equilibrium adsorption experiments at pH 5 and pH 8.5. Electrostatic
Cellular organic matter
interactions and hydrogen bonding proved to be important mechanisms of COM peptides
Granular activated carbon
adsorption. The adsorption of ALA and TBA on granular activated carbon preloaded with COM
Molecular weight distribution
peptides was influenced by solution pH. The reduction in adsorption was significantly greater
Surface charge
at pH 5 compared to pH 8.5, which corresponded to the increased adsorption of COM peptides
Cyanobacterial peptides
at pH 5. The majority of the competition between COM peptides and both herbicides was attributed to low molecular weight COM peptides with MW of 700, 900, 1300 and 1700 Da. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
The use of pesticides represents a risk for water quality in agricultural areas. Many water treatment plants deal with elevated concentrations of pesticides in raw water, which have to be reduced below the established limit of the total amount of pesticides of 0.5 mg L1 and 0.1 mg L1 for any single pesticide in the European Union. Alachlor or 2-chloro-2,6-diethyl-Nmethoxymethyl acetanilide has been one of the most commonly employed pre-emergence herbicide for crops. Terbuthylazine or 2-(tert-butylamino)-4-chloro-6-(ethylamino)-s-triazine belongs to triazine group of selective herbicides and it has been recently extensively used as a substitute for more toxic atrazine. The removal of pesticides by conventional water treatment processes such as coagulation/flocculation/sedimentation or
filtration is not effective (Miltner et al., 1989; Ormad et al., 2008). Meanwhile, the application of granular activated carbon (GAC) adsorbers is a very efficient technique for the removal of the pesticides from the water. Investigations on the application of GAC adsorbers for treatment of river water reported an alachlor removal efficiency of 75% (Badriyha et al., 2003). In the case of terbuthylazine, the lower efficiencies of about 60% were reported (Ormad et al., 2008). Triazines, owing to the presence of eNHe groups, are more hydrophilic compounds and their adsorption onto activated carbon can be obstructed (McCreary and Snoeyink, 1980). Although the adsorption capacity of GAC adsorbers is sufficient under ideal conditions, in real water treatment systems the presence of natural organic matter (NOM), exemplified by humic substances, proteins and polysaccharides,
* Corresponding author. Tel.: þ42 0233109068; fax: þ42 0233324361. E-mail address:
[email protected] (M. Pivokonsky). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.03.051
3360
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 5 9 e3 3 6 8
could significantly reduce the efficiency of activated carbon process (Kilduff and Karanfil, 2002; Li et al., 2003a). NOM adsorbs on GAC and changes the surface properties to such extent that the trace organic compounds uptake and rate of adsorption is greatly reduced (Newcombe and Drikas, 1997). In fixed-bed GAC columns, the mass transfer zone of NOM components move down more rapidly than that of trace organic compounds due to slower adsorption kinetics of NOM. Consequently, the NOM preloads the carbon ahead of the trace organic pollutant. This results in reduced adsorption of trace organic compounds, due to direct competition for adsorption sites or pore blockage by NOM (Newcombe et al., 2002b; Li et al., 2003b). Direct site competition only happens in pores that are accessible to both NOM and the trace organic compound. Therefore, NOM components of low molecular weight (MW) compete in micropores, where the adsorption of trace organic compounds occurs. NOM molecules of higher MW adsorb in mesopores and do not compete for the same adsorption sites. In case there is not enough surface area available in mesopores, larger NOM molecules tend to cause pore blockage (Pelekani and Snoeyink, 1999). Several factors influence the impact of preloading on activated carbon performance. One of the most important factors which determine the competitive adsorption is the pore size distribution of activated carbon relative to the molecular weight of adsorbates (trace organic compounds and NOM) (Quinlivan et al., 2005). This factor plays a crucial role particularly when the adsorption of organic molecules is governed by non-specific dispersive interactions (van der Waals interactions, hydrophobic interactions, hydrogen bonds) (Moreno-Castilla, 2004; Newcombe, 2006). Other important factors are chemical composition of NOM, surface functional group composition and surface charge of adsorbent (Newcombe and Drikas, 1997; Kilduff and Karanfil, 2002). Solution pH controls the (de)protonation of surface functional groups, which determines the surface charge of activated carbon pores and consequently the electrostatic interactions between the adsorbent and NOM molecules (Bjelopavlic et al., 1999). Ionic strength is the other key factor that controls the electrostatic interactions. Thus, these interactions, either attractive or repulsive, can be reduced by increasing the ionic strength of the solution. This is due to a shielding effect of the surface charge produced by the added salt. At certain conditions, the increase in ionic strength can improve adsorption by reduction of the intra- and intermolecular repulsions (Moreno-Castilla, 2004; Campinas and Rosa, 2006). Until now, most attention has been given to the influence of NOM presented by humic and fulvic acids on the adsorption of trace organic pollutants (Newcombe et al., 2002b; Li et al., 2003b; Matsui et al., 2003; Quinlivan et al., 2005). A major problem for surface water purification is seasonal development of large amounts of phytoplankton, which is accompanied by a considerably increased concentration of algal organic matter (AOM) (Pivokonsky et al., 2006; Campinas and Rosa, 2010; Henderson et al., 2010). However, a review of literature revealed that no study has addressed the effect of AOM preloading on the adsorption of trace organic compounds. The AOM is the result of metabolic activities of cyanobacteria and algae that produce extracellular organic matter
(EOM) and, during decay, cellular organic matter (COM) (Takaara et al., 2007; Fang et al., 2010). The AOM is known to comprise proteins, peptides, polysaccharides, oligosaccharides, lipids, nucleic acids, amino acids and other organic acids (Hoyer et al., 1985; Leenheer and Croue, 2003) of which proteins and polysaccharides comprise the majority (Myklestad, 1995; Henderson et al., 2010). Consequently, the AOM composition can be characterized as protein and nonprotein organic matter (Pivokonsky et al., 2006). In addition, under certain conditions of growth, cyanobacteria produce toxins as secondary metabolites which pose a significant health risk. The hepatotoxic and tumor-promoter microcystins are among the most commonly occurring cyanotoxins in surface water reservoirs used for water supply (Campinas and Rosa, 2006). The chemical purification of water by means of coagulation/flocculation is very sensitive to the outbreak of these substances which can cause severe problems (Bernhardt et al., 1985; Takaara et al., 2007). A previous study showed that COM from the decay of cyanobacterium M. aeruginosa were relatively difficult to remove by coagulation/flocculation processes. It was found that COM proteins of a higher molecular weight were removed more efficiently than COM proteins of lower molecular weight (Pivokonsky et al., 2009). Consequently, the residual fraction of COM proteins can affect adsorption of trace organic compounds such as pesticides in granular activated carbon adsorbers. In addition, water treatment plants usually have to deal with increased pesticides concentrations in the summer period, which also brings increased AOM concentrations during algal blooms in eutrophicated water. In the present study, the competitive effect of the COM protein fraction on the adsorption of two herbicides (alachlor and terbuthylazine) was investigated at equilibrium. The protein fraction was studied in order to investigate the effect of carbon surface charge together with the charge of COM peptides on electrostatic interactions taking part in adsorption and consequent competitive effect on the adsorption of herbicides. Two chosen granular activated carbons were commercially available, designed for pesticides removal in the water treatment process. The equilibrium adsorption of both herbicides on carbons preloaded with the COM peptides of MW <10 kDa was studied. Carbons and the COM were extensively characterized in order to describe the effect of COM peptides preloading. The objectives of the study were to: (i) elucidate the adsorption mechanism of COM peptides, (ii) determine the influence of pH value on the competitive adsorption of herbicides and COM peptides, (iii) identify the most likely competitive mechanism and components in this system.
2.
Materials and methods
2.1.
Adsorbents
The adsorbents were two commercial granular activated carbons, Filtrasorb 400 (Chemviron Carbon, USA) and Norit 1240 (Norit Americas Inc., USA). Both adsorbents were
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 5 9 e3 3 6 8
bituminous coal-based carbons activated by steam. The sieve fraction between 0.8 and 1 mm was employed. As in Sotelo’s et al. (2002) study, a full-scale diameter of GAC was used instead of a fine graded carbon, because only equilibrium adsorption experiments were performed, not kinetic experiments. Prior to experiments, the activated carbons were treated by extensive washing in ultrapure water, dried at 110 C for 24 h and stored in a desiccator.
2.1.1.
GAC characterization
Textural properties (specific surface area, micropore volume, total pore volume and mesopore surface area) of activated carbons were evaluated from 77 K nitrogen physical adsorption isotherms obtained with the volumetric instrument ASAP2020 (Micromeritics, USA). Before analysis, the samples were dried at 105 C and 0.1 Pa for 24 h. The specific surface area (SBET) was evaluated by BET method (Brunauer et al., 1938), micropore volume, (Vmicro) and mesopore surface area (Smeso) by the t-plot method with LeclouxePirard master isotherm (Lecloux and Pirard, 1979) and pore size distribution by the advanced BJH method (Barret et al., 1951).
2.1.2.
ultrapure water without the assistance of organic solvent. Undissolved herbicide particles were removed by filtering the solution through a 0.22 mm membrane filter (Millipore, USA). Gas chromatography with electron capture detector (Agilent Technologies 6890N, USA) and a DB-XLB capillary column (30 m 0.32 mm i.d. 0.17 mm film thickness) (Agilent Technologies, USA) was used for the analysis of ALA and TBA. Target herbicides were extracted from water phase by SPE columns with C18 sorbent (Agilent SAMPLI Q C18, Agilent Technologies, USA) and eluted by acetone prior to analysis.
2.3.
M. aeruginosa cultivation
The cyanobacterium M. aeruginosa, Kutz. (Zapomelova, 2006/2) was used in this study. Inoculum of this strain was obtained through the kind generosity of the Department of Culture Collection of Algal Laboratory, Institute of Botany, AS CR, Czech Republic. The strain of M. aeruginosa was cultured according to the methodology described in the literature (Pivokonsky et al., 2006, 2009). The culture of M. aeruginosa was harvested on the 16th day of cultivation during steady-state growth.
Surface charge determination
Surface charge determinations were undertaken by surface titrations using an Orion 960 Autotitrator (Thermo Scientific, USA). The carbon was titrated with 0.1 M HCl to pH 3, after the addition of 0.1 M NaOH. The titration was conducted at two electrolyte concentrations, 0.01 M and 0.3 M NaCl. A blank titration was also performed. Relative surface charge was determined from the difference between the surface titration curves and the blank curves. Relative surface charge was then plotted against pH. The pH at which the curves of two electrolyte concentrations crossed was the pH at which the absolute charge on the surface was zero (pHpzc). This method was described in detail in the literature (Bjelopavlic et al., 1999).
2.2.
3361
Adsorbates
Alachlor (ALA) of 99,2% purity and terbuthylazine (TBA) of 98,8% purity were supplied by SigmaeAldrich (USA). These herbicides present two common surface water contaminants. The structural formulas of ALA and TBA are shown in Fig. 1. ALA contains benzene ring, while TBA consists of triazine heterocycle and two amino groups. They have similar molecular weight (ALA 269.8 Da, TBA 229.7 Da) and both are relatively hydrophobic (aqueous solubility at 20 C: ALA 148 mg L1, TBA 8.5 mg L1) (Badriyha et al., 2003; Bruzzoniti et al., 2006). The stock solutions of concentration 10 mg L1 were prepared by adding a weighed amount of herbicides to
2.4.
COM preparation
The COM samples were prepared by disruption of the microorganisms’ cells, which were separated from the growth medium by a 0.22 mm membrane filter (Millipore, USA). The separated cells were stirred with ultrapure water (200 mL) and disrupted using an ultrasonic homogenizer (HD 3200, 20 kHz, 60W) for 20 min. The efficiency of cells destruction was confirmed by an optical microscope (Optech B4T, UK). The residual solids were removed by a 0.22 mm membrane filter, and the filtrates were concentrated tenfold in a rotary evaporator (Laborota 4000 HB/G1, Germany) at 30 C. The concentrated COM was stored at 18 C.
2.5.
COM characterization
2.5.1.
DOC analysis
Dissolved organic carbon (DOC) was analyzed using a Shimadzu TOC-VCPH analyzer (Shimadzu Corporation, Japan). All samples were first filtered through a 0.22 mm membrane filter (Millipore, USA). All measurements were conducted in triplicate and errors were less than 2%.
2.5.2.
Determination and isolation of COM protein portion
The COM was characterized in terms of the amount of protein (DOCP) and non-protein (carbohydrates) (DOCNP) organic matter. Proteins were isolated from the COM using (NH4)2SO4 as a protein precipitant. The protein precipitate was then separated from the dissolved organic matter by filtration through a 0.22 mm membrane filter (Millipore, USA), and DOCNP was analyzed in the filtrate. The protein portion DOCP was calculated as follows: DOCP ¼ DOCT DOCNP
Fig. 1 e Molecular structures of alachlor (left) and terbuthylazine (right).
(1)
where DOCP is the amount of protein DOC, DOCT the total DOC of the COM, and DOCNP the amount of non-protein (carbohydrate) DOC. The protein precipitations were carried out in
3362
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 5 9 e3 3 6 8
triplicate and errors in measurement of DOCP were less than 5%. The protein precipitate was then dissolved in ultrapure water for the purpose of adsorption experiments.
2.5.3.
Molecular weight fractionation
COM protein fraction was characterized in terms of molecular weight (MW) distribution. Centrifugation (4000 rpm, T ¼ 40 min) was used to drive the COM protein fraction through Amicon Ultra-15 centrifugal filters of 100, 50, 30, 10 and 3 kDa NMWL (Millipore, USA). The MW distribution was expressed as DOC portion of each MW fraction. Each MW distribution was repeated in triplicate. The low MW fraction of COM peptides <10 kDa was collected as an experimental material for adsorption tests. The MW fractionation by high-performance size exclusion chromatography (HPSEC) was performed using Agilent Bio ˚ and 300A ˚ (7.8 300 mm, 5 mm). The separation SEC-5 100A ˚ and range applied was 100e1 250 000 Da using Bio SEC-5 100A ˚ columns in series. The HPLC system (Agilent Bio SEC-5 300A 1100 series, Agilent Technologies, USA) was coupled with a diode array detector (DAD) operated at 280 nm. Prior to HPSEC, the maximum absorption wavelength (lmax ¼ 280 nm) of COM peptides <10 kDa was detected in UVeVIS absorbance spectra using a UVeVIS 8452A spectrophotometer (Agilent Technologies, USA). Therefore, the wavelength 280 nm was used especially for the COM peptides detection. The HPSEC mobile phase used for the MW fractionation was 0.05 M phosphate buffer (pH 7.0). The flow rate was 1 mL min1 at the temperature of 23 C and the sample injection volume was 60 mL. Data analysis was performed using Agilent Technologies Chemstation software (Agilent Technologies, USA). The system was calibrated using the following SEC standards (SigmaeAldrich, USA): d-biotin (224 Da), cyanocobalamin (1.35 kDa), aprotinin (6.5 kDa), cytochrome c (12.4 kDa), carbonic anhydrase (29 kDa), albumin (66 kDa), alcohol dehydrogenase (150 kD), apoferritin (443 kDa) and thyroglobulin (670 kDa). A semi-log calibration curve was used to calculate the MW (R2 ¼ 0.96). BioRad gel filtration standards of chicken ovalbumin (44 kDa) and bovine gamma globulin (158 kDa) were used as control samples. Standard error was 0.58 kDa for chicken ovalbumin and 0.89 kDa for gamma globulin. Reproducibility of the MW fractionation of COM peptide samples was very good, with MW deviations of less than 3% from repeated measurements.
2.5.4.
Determination of protein isoelectric point
The isoeletric point (pI ) of isolated COM peptides <10 kDa was determined by isoelectric focusing (IEF) and was carried out with a Multiphor II electrophoresis system (Pharmacia, Sweden). The IEF gel (7.5%) was prepared using ampholines of pI 2.5e5.0 and 3.5e10.0 (Pharmacia, Sweden). A standard calibration curve with broad-pI protein calibration kit, pI 3.0e10.0 (Pharmacia, Sweden), was used to determine the isoelectric points. Gels were stained with silver staining kit (Bio-Rad Silver Stain, USA) and activity-stained with 0.005 M guaiacol.
<10 kDa to determine the effect of COM peptides preloading on adsorption capacity of target herbicides on GAC. Synthetic water was prepared by herbicide stock solution and ultrapure water with alkalinity adjusted to 1.45 mmol L1 by 0.125 M NaHCO3. Different doses of GAC were agitated in 2 L borosilicate jars. In the case of the experiment with fresh GAC, the flasks were agitated on a magnetic stirrer for 7 days to reach adsorption equilibrium. Blanks were included to determine the initial target herbicide concentration. During COM peptides preloading experiments, different doses of fresh GAC were preloaded with COM peptides and agitated for 7 days. Initial DOC concentration of experimental solution was 6 mg L1. Furthermore, 50 mg L1 of sodium azide was added to both solutions with fresh and preloaded GAC to eliminate biological activity and decomposition and to provide similar ionic strength of 12.8 mM. The experiments with preloaded GAC were undertaken at acidic and alkaline pH (pH 5.0 and 8.5) adjusted by 0.1 M NaOH and HCl. DOC samples were taken after preloading to determine the surface concentrations of COM peptides. Herbicide stock solution was then spiked into the experimental solution to obtain the initial concentration of 100 mg L1 and the jars were agitated for the next 7 days to reach adsorption equilibrium. After equilibrium was reached, the GAC particles were separated by filtering the water samples through a 0.22 mm membrane filter (Millipore, USA) and the liquid-phase target herbicide concentration and DOC were measured. Afterward, the rest of the filtrates were concentrated using vacuum rotary evaporator (Laborota 4002 control, Heidolph, Germany) in order to obtain DOC concentration 100 mg L1 for HPSEC determination. The Freundlich model was used to describe the adsorption equilibrium: q ¼ Kf Ce1=n
where q (mg mg1) and Ce (mg L1) represent equilibrium surface and solution concentrations, respectively, and Kf and 1/n are Freundlich parameters.
3.
Results and discussion
3.1.
GAC characterization
Table 1 provides information on the textural properties of the two carbons investigated. Both carbons had relatively similar surface properties, as expected for the same starting material (coal) and the steam activation method. Nevertheless, the microporosity (Vmicro/Vtotal) of the carbon F400 was higher
Table 1 e Textural properties of granular activated carbons.
2
2.6.
Equilibrium adsorption experiments
Batch adsorption tests with ALA and TBA were conducted using fresh GAC and GAC preloaded with COM peptides of MW
(2)
1
SBET (m g ) Smeso (m2 g1) Vtotal (cm3 g1) Vmicro (cm3 g1) Vmicro/Vtotal (%)
N1240
F400
1110 536 0.70 0.29 41
1025 416 0.59 0.30 50
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 5 9 e3 3 6 8
than that of the N1240. It was 50% for the F400 and 41% for the N1240. Consequently, in terms of mesopores, the carbon N1240 had a higher mesopore surface, Smeso 536 m2 g1, compared to Smeso of F400, which was 416 m2 g1. This characteristic is important with regard to the pore blockage phenomenon. It was previously reported that large NOM molecules tend to cause pore blockage when there is not enough surface area available in mesopores (Pelekani and Snoeyink, 1999; Li et al., 2003a). Net surface charge was determined by potentiometric titration, the results are shown in Fig. 2. The two investigated carbons displayed both positive and negative charge, depending on the pH. The carbon F400 had a pHpzc of about 6.7, while N1240 was more acidic carbon with pHpzc of about 5.5. Therefore, both carbons had a positive net charge at applied acidic conditions of pH 5 and negative net charge at basic pH 8.5 during adsorption experiments in this study. Negative surface charge of activated carbon is attributed to surface acidic oxygen groups, e.g. carboxyl, phenol, lactone and lactol (Newcombe, 2006). Whereas, positive surface charge is usually assigned to surface oxygen complexes of basic character like pyrones or chromenes, or to the existence of electron-rich regions within the graphene layers acting as Lewis basic centers, which accept protons from the aqueous solution and form H30þ- p complexes (Moreno-Castilla, 2004). However, it is known that the titration method of surface charge determination measures net surface charge only; it is not possible from these results to determine the actual concentration of positive and negative sites.
3.2.
COM characterization
The protein portion, determined as DOCP, was measured at approximately 60.2% of DOCT in the COM, and the non-protein portion organic matter made up the balance of 39.8%. The COM proteins were characterized in terms of MW distribution expressed as DOC portion of ultrafiltration fractions. Percentage distribution of the MW fractions of <3; 3-10; 10-30; 30-50; 50-100 and >100 kDa was 19.4; 1.0; 5.8; 9.9; 18.6 and 45.6%, respectively. The COM peptides fraction of MW <10 kDa, which was consequently used in equilibrium adsorption experiments, comprised 20.4 3.7% of the whole
Fig. 2 e Surface charge versus pH of activated carbons (pHpzc [ pH of point of zero charge).
3363
COM protein portion. This fraction was chosen to represent the residual peptide fraction, which is not supposed to be aggregated and separated during chemical water treatment as was previously reported (Pivokonsky et al., 2009). Another study of Ebie et al. (2001) reported that the residual NOM after alum coagulation of a surface water displayed a range of molecular weights of approximately 1e5 kDa. Fig. 3 shows HPSEC chromatogram with apparent molecular weights of the isolated COM peptides <10 kDa. Peptides of MW of approximately 700, 900, 1300, 1700, 1900, 2300, 2700, 5300 and 6300 Da were identified as main components of this COM protein fraction. These peptides may include hepatotoxic microcystins. Since HPSEC is a size exclusion method, the molecular weight values should not be considered to be absolute values. The isoeletric points of isolated COM peptides <10 kDa were determined by isoelectric focusing (IEF). The measured values of peptides pI were 5.25; 5.45; 5.80; 6.10; 6.25; 7.15; 7.85; 7.95 and 8.05. The number of peptides pI identified by IEF corresponded to the number of peptides separated by HPSEC.
3.3.
Adsorption of COM peptides
3.3.1.
DOC removal
Adsorption experiments were conducted at two different GACs at pH 5 and pH 8.5. DOC concentrations were measured after a 7 day preloading. Initial DOC concentration of COM peptides was 6 mg L1. Fig. 4 shows the dissolved organic carbon removed as a function of carbon dose for the COM peptides <10 kDa. Adsorption isotherms were not used as the way of display for COM peptides adsorption because the experimental conditions resulted in a very narrow range of DOC solution and surface concentrations for the isotherms. DOC removal was approximately twice as high at pH 5 than at pH 8.5. Moreover, at pH 5, higher adsorption of COM peptides was seen on carbon N1240 compared to F400. It can be attributed to higher available pore volume of carbon N1240, see Table 1. The pore size distribution as well as the character and the MW distribution of the NOM has been found to be a major influence on the adsorption by several authors (Ebie et al., 2001; Kilduff and Karanfil, 2002; Li et al., 2003a). Some
Fig. 3 e HPSEC chromatogram displaying apparent molecular weights of COM peptides <10 kDa.
3364
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 5 9 e3 3 6 8
Fig. 4 e DOC removed as a function of carbon dose.
studies have also shown that the surface chemistry, in particular surface charge, of the activated carbon has a strong influence on the adsorption. Then, solution pH controls the charge and extent of (de)protonation of the adsorbate as well as the net charge of activated carbon pore surfaces (Knappe et al., 1998; Moreno-Castilla, 2004). In our study, at pH 5, the differences in total pore volume between studied carbons became apparent. This is in agreement with studies of Bjelopavlic et al. (1999) and Newcombe et al. (2002a) who reported consistent dependence of the adsorption of NOM on the pore volume of different carbons at acidic pH. The result of IEF indicated that, at pH 5, the positive charge of COM peptides predominated. On the contrary, at pH 8.5, the COM peptides were negatively charged. It was due to their amphoteric character, as the range of determined peptides pI was 5.25e8.05. Similarly, at pH 5, the net charge of both GACs was positive and it was negative at pH 8.5, according to determined pHpzc 5.5 for carbon N1240 and 6.7 for F400. The schema in Fig. 5 illustrates fundamental interactions at experimental conditions. Hydrophobic interaction is a major driving force of
adsorption, but in case of the adsorption of proteins and peptides, the electrostatic interactions and hydrogen bonding are very important, so the adsorption is controlled by the surface charge (Burns et al., 1996; Yoon et al., 1999). During adsorption experiments at pH 5, the pH was lower than determined pI values of COM peptides. Increased adsorption onto both carbons at pH 5 (Fig. 4) can be explained by formation of hydrogen bonds between protonated functional groups of COM peptides and protonated surface groups of carbon. On the other hand, at pH 8.5, COM peptides were negatively charged due to their pI values (5.25e8.05). Therefore, the repulsive electrostatic interactions between deprotonated carboxyl groups of COM peptides and deprotonated surface acidic oxygen groups of the carbons reduced adsorption significantly. A similar concept of protein-surface interactions was described in several literature (Yoon et al., 1999; Zhou et al., 2007; Katiyar et al., 2010). This effect became more apparent in case of the carbon N1240. The extent of electrostatic and hydrogen bonding contribution to the adsorption is governed by the densities of surface acidic oxygen groups, such as carboxyl groups. The surface of the carbon N1240 is expected to contain more acidic oxygen groups than F400 due to its lower pHpzc (Fig. 2), consequently, the effect of the repulsive electrostatic interactions can be more pronounced. In this study, adsorption experiments were conducted only at low ionic strength (IS ¼ 12.8 mM). However, a shielding effect of increased ionic strength diminishing electrostatic repulsions between carbon surface and adsorbate as well as lateral repulsions between NOM molecules was described in literature (Bjelopavlic et al., 1999; Campinas and Rosa, 2006). Since COM peptides carried similar charges in relation to GACs surface, positive at pH 5 and negative at pH 8.5, and all adsorption experiments were performed at high surface concentrations (22e300 mg g1 F400, 19e395 mg g1 N1240), an enhanced adsorption by shielding effect would be expected by increase in ionic strength.
3.3.2.
Change in MW distribution
Chromatograms on Fig. 6 compare MWs of remaining COM peptides after adsorption by different doses of F400 and N1240
Fig. 5 e Fundamental interactions involved in COM peptides adsorption at experimental conditions.
3365
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 5 9 e3 3 6 8
Fig. 6 e Molecular weight distributions of remaining COM peptides after adsorption on F400 and N1240 at different GAC doses at pH 5 and pH 8.5.
at pH 5 and pH 8.5. The results of MW distribution of remaining COM peptides in solution after adsorption onto F400 and N1240 were similar for both studied carbons. This was presumably caused by the fact that the textural properties of both carbons, in terms of microporosity and mesopore surface, were not significantly different to affect MW distribution. COM peptides of MW 700, 900, 1300 and 1700 Da were adsorbed proportionally to carbon doses. Even with high carbon doses, a negligible amount of COM peptides with MW higher than 2700 Da was adsorbed. With 40 mg L1 of F400, almost all COM peptides with MW of 700, 900 and 1700 Da were removed at pH 8.5. Similarly, with 100 mg L1 of F400, a significant amount of COM peptides with MW of 1700, 1300, 900 and 700 Da was removed at pH 5. These results demonstrated that among COM peptides <10 kDa, the low MW peptides were preferentially adsorbed. This observation is similar to conclusions of authors who studied adsorption of different NOM fractions (Ebie et al., 2001; Li et al., 2003a). However, at pH 8.5, the COM peptides of the lowest MW (<1700 Da) were adsorbed more quantitatively compared to pH 5 (Fig. 6). With regard to the fact that the overall adsorption of COM peptides was higher at pH 5 (Fig. 4), this could be caused by the pore blockage phenomenon. At pH 5, the higher adsorption of high MW COM peptides in mesopores could block micropore openings and restrict their accessibility for low MW COM peptides.
3.4.
Adsorption of herbicides
Isotherm tests were conducted with ALA and TBA in COM-free water using both GACs N1240 and F400. The initial concentrations of both herbicides used were 100 mg L1. The adsorption capacity of the carbon N1240 was higher for both herbicides which can be related to the total pore volume. This corresponded to the Freundlich adsorption parameters for the 7 day single-solute herbicide isotherms which are summarized in Table 2. Moreover, the obtained Kf parameters for both carbons indicated that the adsorption capacity for ALA was up to tenfold higher in comparison with TBA during adsorption experiments. With regard to the similar molecular weight of ALA and TBA, this can be attributed to the different chemical
Table 2 e Freundlich parameters for ALA and TBA adsorption onto fresh GACs. Adsorbate
GAC
Kf (mg/mg)(L/mg)1/n
1/n
ALA
N1240 F400
29.4 12.6
0.94 0.93
TBA
N1240 F400
4.9 2.7
0.29 0.36
3366
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 5 9 e3 3 6 8
Fig. 7 e Adsorption isotherms of ALA and TBA onto two GACs in the absence and presence of competing COM peptides at pH 5 and pH 8.5.
structure of these herbicides. Attractive interactions between delocalized p electrons of aromatic rings, contained in ALA structure, and p electrons of polyaromatic basal planes of the carbon have been reported by several authors (Newcombe and Drikas, 1997; Moreno-Castilla, 2004). On the other hand, TBA contains two amino groups, which can obstruct adsorption onto activated carbon due to their hydrophilic character (McCreary and Snoeyink, 1980). Equilibrium adsorption experiments with ALA and TBA were also performed onto carbons preloaded with COM peptides <10 kDa at pH 5 and pH 8.5. The impact of COM peptides preloading on TBA and ALA adsorption at pH 5 and pH 8.5 is shown by the isotherms plotted in Fig. 7. The corresponding Freundlich parameters are not shown since the Freundlich model did not fit the adsorption isotherms obtained for carbons preloaded by COM peptides (particularly at pH 5). ALA and TBA uptake by preloaded carbons was lower than uptake by fresh carbons. The reduction in adsorption was significantly greater for ALA than for TBA. The difference between the adsorption of ALA and TBA onto preloaded carbons is negligible compared to the difference in adsorption capacity of both herbicides onto fresh carbons. Moreover, the reduction in adsorption was significantly greater at pH 5 compared to pH 8.5, which corresponded to the increased adsorption of COM peptides at pH 5. Probable explanation of greater impact of COM peptides preloading on ALA adsorption is that COM peptides adsorbed to the oxygen functional groups by hydrogen bonding became secondary adsorption centers, which retained other COM peptides. This reduced the accessibility of the aromatic ring of ALA to the hydrophobic parts of the carbon surface. Since, the major adsorption driving forces of ALA are pep dispersion interactions between the aromatic ring of ALA and the aromatic structure of the graphene layers. According to HPSEC chromatograms visualized in Fig. 6, the COM peptides of 700, 900, 1300 and 1700 Da were identified as the most relevant components in competition for adsorption sites with ALA and TBA. The mechanism of direct site competition of low molecular weight NOM components was reported in several studies with different MW fractions of NOM (Pelekani and Snoeyink, 1999; Ebie et al., 2001). On the contrary, COM peptides with MW of 2300 and 6300 Da which were not quantitatively adsorbed proportionally to carbon
doses are supposed to cause pore blockage effect, since both carbons in this study represented relatively microporous carbons. This is consistent with the study of Newcombe et al. (2002a, 2002b), who reported that the higher molecular weight compounds restricted access to the adsorption sites for smaller NOM compounds in microporous carbons, while access was not restricted in mesoporous carbons.
4.
Conclusions
This study investigated the competitive adsorption on granular activated carbons between two herbicides (ALA and TBA) and COM peptide fraction <10 kDa, which represented the hardly separable COM protein fraction during chemical water treatment. Moreover, the effect of surface charge on the adsorption of COM peptides was studied at pH 5 and pH 8.5. The extent of electrostatic and hydrogen bonding contribution to the adsorption was also investigated. Electrostatic interactions and hydrogen bonding proved to be important mechanisms of COM peptides adsorption. The solution pH controlled the charge of peptides, as well as the net charge of activated carbon pore surfaces. The extent of electrostatic repulsion could be predicted by peptides characterization by means of pI and by the determination of net surface charge of used carbon. At acidic pH, the formation of hydrogen bonds between protonated acidic oxygen groups of peptides and carbon surface contributes to the adsorption. The adsorption of TBA and ALA onto activated carbon was influenced by their different chemical structures. The impact of COM peptides preloading on herbicides adsorption was significantly greater for ALA in comparison with TBA. As expected, the reduction in adsorption was significantly greater at pH 5 compared to pH 8.5, which corresponded to the increased adsorption of COM peptides at pH 5. Adsorbed COM peptides were supposed to become secondary adsorption centers which could consequently prevent the migration of herbicides to a large portion of carbon surface. The low molecular weight COM peptides were adsorbed to a greater extent than the larger peptides, due to size exclusion effects. The majority of the competition between COM peptides and both herbicides can be attributed to COM peptides of MW from
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 5 9 e3 3 6 8
700 to 1700 Da, where the hepatotoxic microcystins may be included. It is generally acknowledged that the pore size distribution plays a major role in the adsorption of NOM onto activated carbon. Nevertheless, the adsorption of charged COM peptides showed the influence of electrostatic effects related to the charge of the adsorbate and of the carbon surface. The selection of the carbon with broad pore size distribution can lead to a higher removal of micropollutants coexistent with COM in drinking water sources. Moreover, the knowledge of surface charge dependence on solution pH is fundamental for the selection of the proper activated carbon.
Acknowledgments The research project has been funded by the Grant Agency of AS CR under the project No. IAA200600902, by the Czech Science Foundation under the project No. P105/10/P515 and Institutional Research Plan No. AVOZ20600510. The authors acknowledge the financial assistance on this project.
references
Badriyha, B.N., Ravindran, V., Den, W., Pirbazari, M., 2003. Bioadsorber efficiency, design, and performance forecasting for alachlor removal. Water Research 37, 4051e4072. Barret, E.P., Joyner, L.G., Halenda, P.P., 1951. The determination of pore volume and area distributions in porous substances. I. Computations from nitrogen isotherms. Journal of American Chemical Society 73, 373e380. Bernhardt, H., Hoyer, O., Shell, H., Lu¨sse, B., 1985. Reaction mechanisms involved in the influence of algogenic organic matter on flocculation. Journal for Water and Wastewater Research (Zeitschrift fur Wasser und Abwasser Forschung) 18 (1), 18e30. Bjelopavlic, M., Newcombe, G., Hayes, R., 1999. Adsorption of NOM onto activated carbon: effect of surface charge, ionic strength, and pore volume distribution. Journal of Colloid and Interface Science 210, 271e280. Brunauer, S., Emmett, P.H., Teller, E., 1938. Adsorption of gases in multimolecular layers. Journal of American Chemical Society 60, 309e319. Bruzzoniti, M.C., Sarzanini, C., Costantino, G., Fungi, M., 2006. Determination of herbicides by solid phase extraction gas chromatography-mass spectrometry in drinking waters. Analytica Chimica Acta 578, 241e249. Burns, N.L., Holmberg, K., Brink, C., 1996. Influence of surface charge on protein adsorption at an amphoteric surface: effects of varying acid to base ratio. Journal of Colloid and Interface Science 178, 116e122. Campinas, M., Rosa, M.J., 2006. The ionic strength effect on microcystin and natural organic matter surrogate adsorption onto PAC. Journal of Colloid and Interface Science 299, 520e529. Campinas, M., Rosa, M.J., 2010. Assessing PAC contribution to the NOM fouling control in PAC/UF systems. Water Research 44, 1636e1644. Ebie, K., Li, F., Azuma, Y., Yuasa, A., Hagishita, T., 2001. Pore distribution effect of activated carbon in adsorbing organic micropollutants from natural water. Water Research 35 (1), 167e179.
3367
Fang, J., Yang, X., Ma, J., Shang, Ch, Zhao, Q., 2010. Characterization of algal organic matter and formation of DBPs from chlor(am)ination. Water Research 44, 5897e5906. Henderson, R.K., Parsons, S.A., Jefferson, B., 2010. The impact of differing cell and algogenic organic matter (AOM) characteristics on the coagulation and flotation of algae. Water Research 44, 3617e3624. Hoyer, O., Lu¨sse, B., Bernhardt, H., 1985. Isolation and characterization of extracellular organic matter (EOM) from algae. Journal of Water and Wastewater Research (Zeitschrift fur Wasser und Abwasser Forschung) 18 (2), 76e90. Katiyar, A., Thiel, S.W., Guliants, V.V., Pinto, N.G., 2010. Investigation of the mechanism of protein adsorption on ordered mesoporous silica using flow microcalorimetry. Journal of Chromatography A 1217, 1583e1588. Kilduff, J.E., Karanfil, T., 2002. Trichlorethylene adsorption by activated carbon preloaded with humic substances: effects of solution chemistry. Water Research 36, 1685e1698. Knappe, D.R.U., Matsui, Y., Snoeyink, V.L., Roche, P., Parados, M.J., Bourbigot, M.M., 1998. Predicting the capacity of powdered activated carbon for trace organic compounds in natural waters. Environmental Science and Technology 32 (11), 1694e1698. Lecloux, A., Pirard, J.P., 1979. The importance of standard isotherms in the analysis of adsorption isotherms for determining the porous texture of solids. Journal of Colloid and Interface Science 70, 265e281. Leenheer, J.A., Croue, J.P., 2003. Characterizing aquatic dissolved organic matter. Environmental Science and Technology 37 (1), 18Ae26A. Li, Q., Snoeyink, V.L., Marin˜as, B.J., Campos, C., 2003a. Elucidating competitive adsorption mechanisms of atrazine and NOM using model compounds. Water Research 37, 773e784. Li, Q., Snoeyink, V.L., Marin˜as, B.J., Campos, C., 2003b. Pore blockage effect of NOM on atrazine adsorption kinetics of PAC: the roles of PAC pore size distribution and NOM molecular weight. Water Research 37, 4863e4872. Matsui, Y., Fukuda, Y., Inoue, T., Matsushita, T., 2003. Effect of natural organic matter on powdered activated carbon adsorption of trace contaminants: characteristics and mechanism of competitive adsorption. Water Research 37, 4413e4424. McCreary, J.J., Snoeyink, V.L., 1980. Characterization and activated carbon adsorption of several humic substances. Water Research 14 (2), 151e160. Miltner, R.J., Baker, D.B., Speth, T.F., Fronk, C.A., 1989. Treatment of seasonal pesticides in surface waters. Journal American Water Works Association 81 (1), 43e52. Moreno-Castilla, C., 2004. Adsorption of organic molecules from aqueous solutions on carbon materials. Carbon 42, 83e94. Myklestad, S.M., 1995. Release of extracellular products by phytoplankton with special emphasis on polysaccharides. The Science of the Total Environment 165, 155e164. Newcombe, G., 2006. Removal of natural organic material and algal metabolites using activated carbon. In: Newcombe, G., Dixon, D. (Eds.), Interface Science in Drinking Water Treatment, Theory and Applications. Elsevier Ltd., Amsterdam, The Netherlands, pp. 133e153. Newcombe, G., Drikas, M., 1997. Adsorption of NOM onto activated carbon: electrostatic and non-electrostatic effects. Carbon 35 (9), 1239e1250. Newcombe, G., Morisson, J., Hepplewhite, C., 2002a. Simultaneous adsorption of MIB and NOM onto activated carbon. I. Characterisation of the system and NOM adsorption. Carbon 40, 2135e2146. Newcombe, G., Morrison, J., Hepplewhite, C., Knappe, D.R.U., 2002b. Simultaneous adsorption of MIB and NOM onto activated carbon. II. Competitive effects. Carbon 40, 2147e2156.
3368
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 5 9 e3 3 6 8
Ormad, M.P., Miguel, N., Claver, A., Matesanz, J.M., Ovelleiro, J.L., 2008. Pesticides removal in the process of drinking water production. Chemosphere 71, 97e106. Pelekani, C., Snoeyink, V.L., 1999. Competitive adsorption in natural water: role of activated carbon pore size. Water Research 33 (5), 1209e1219. Pivokonsky, M., Kloucek, O., Pivokonska, L., 2006. Evaluation of the production, composition and aluminum and iron complexation of algogenic organic matter. Water Research 40, 3045e3052. Pivokonsky, M., Polasek, P., Pivokonska, L., Tomaskova, H., 2009. Optimized reaction conditions for removal of cellular organic matter of Microcystis aeruginosa during the destabilization and aggregation process using ferric sulphate in water purification. Water Environment Research 81 (5), 514e522. Quinlivan, P.A., Li, L., Knappe, D.R.U., 2005. Effects of activated carbon characteristics on the simultaneous adsorption of
aqueous organic micropollutants and natural organic matter. Water Research 39, 1663e1673. Sotelo, J.L., Ovejero, G., Delgado, J.A., Martı´nez, I., 2002. Comparison of adsorption equilibrium and kinetics of four chlorinated organics from water onto GAC. Water Research 36, 599e608. Takaara, T., Sano, D., Konno, H., Omura, T., 2007. Cellular proteins of Microcystis aeruginosa inhibiting coagulation with polyaluminum chloride. Water Research 41, 1653e1658. Yoon, J.-Y., Kim, J.-H., Kim, W.-S., 1999. The relationship of interaction forces in the protein adsorption onto polymeric microspheres. Colloids and Surfaces A 153, 413e419. Zhou, H., Wu, T., Dong, X., Wang, Q., Shen, J., 2007. Adsorption mechanism of BMP-7 on hydroxyapatite (001) surfaces. Biochemical and Biophysical Research Communications 361, 91e96.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 6 9 e3 3 7 7
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Occurrence and formation potential of N-nitrosodimethylamine in ground water and river water in Tokyo Nguyen Van Huy a, Michio Murakami b,*, Hiroshi Sakai a, Kumiko Oguma a, Koji Kosaka c, Mari Asami c, Satoshi Takizawa a a
Department of Urban Engineering, Graduate School of Engineering, The University of Tokyo, 7-3-1 Hongo, Bunkyo, Tokyo 113-8656, Japan “Wisdom of Water” (Suntory), Corporate Sponsored Research Program, Organization for Interdisciplinary Research Projects, The University of Tokyo, 7-3-1 Hongo, Bunkyo, Tokyo 113-8656, Japan c Department of Water Supply Engineering, National Institute of Public Health, 2-3-6 Minami, Wako, Saitama 351-0197, Japan b
article info
abstract
Article history:
N-nitrosodimethylamine (NDMA), a disinfection byproduct of water and wastewater
Received 11 November 2010
treatment processes, is a potent carcinogen. We investigated its occurrence and the
Received in revised form
potential for its formation by chlorination ðNDMA FPCl2 Þ and by chloramination ðNDMA
28 January 2011
FPNH2 Cl Þ in ground water and river water in Tokyo. To characterize NDMA precursors, we
Accepted 27 March 2011
revealed their molecular weight distributions in ground water and river water. We
Available online 5 April 2011
collected 23 ground water and 18 river water samples and analyzed NDMA by liquid
Keywords:
water samples with free chlorine for 24 h at pH 7.0 while residual free chlorine was kept at
NDMA formation potential
1.0e2.0 mgCl2/L. NDMA FPNH2 Cl was evaluated by dosing water samples with mono-
chromatography-tandem mass spectrometry. NDMA FPCl2 was evaluated by chlorinating
NDMA precursors
chloramine at 140 mgCl2/L for 10 days at pH 6.8. NDMA precursors and dissolved organic
Ground water
carbon (DOC) were fractionated by filtration through 30-, 3-, and 0.5 kDa membranes.
Chlorination
NDMA concentrations were <0.5e5.2 ng/L (median: 0.9 ng/L) in ground water and
Chloramination
<0.5e3.4 ng/L (2.2 ng/L) in river water. NDMA concentrations in ground water were slightly
Disinfection byproducts
lower than or comparable to those in river water. Concentrations of NDMA FPCl2 were not much higher than concentrations of NDMA except in samples containing high concentrations of NH3 and NDMA precursors. The increased NDMA was possibly caused by reactions between NDMA precursors and monochloramine unintentionally formed by the reaction between free chlorine and NH3 in the samples. NDMA precursors ranged from 4 to 84 ng-NDMA eq./L in ground water and from 11 to 185 ng-NDMA eq./L in river water. Those in ground water were significantly lower than those in river water, suggesting that NDMA precursors were biodegraded, adsorbed, or volatilized during infiltration. The molecular weight of NDMA precursors in river water was dominant in the <0.5 kDa fraction, followed by 0.5e3 kDa. However, their distribution was inconsistent in ground water: one was dominant in the <0.5 kDa fraction, and the other in 0.5e3 kDa. Molecular weight distributions of NDMA precursors were very different from those of DOC. This is the first study to reveal the widespread occurrence and characterization of NDMA precursors in ground water. ª 2011 Elsevier Ltd. All rights reserved.
* Corresponding author. Tel.: þ81 3 5841 6263; fax: þ81 3 5841 8529. E-mail address:
[email protected] (M. Murakami). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.03.053
3370
1.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 6 9 e3 3 7 7
Introduction
N-nitrosodimethylamine (NDMA) is a highly water-soluble N-nitrosamine (WHO, 2008). NDMA had been used as an intermediate in the production of rocket fuel, an inhibitor of nitrification in soil, a plasticizer in the manufacture of rubber and polymers, a solvent in the fiber and plastic industry, an antioxidant, a softener of copolymers, and an additive to lubricants (WHO, 2002). Recently, NDMA was found to be a disinfection byproduct of chlorination, chloramination (Najm and Trussell, 2001; Choi et al., 2002; Mitch and Sedlak, 2002; Chen and Young, 2008; Zhou et al., 2009a), and ozonation (Andrzejewski and Nawrocki, 2007; Andrzejewski et al., 2008; Oya et al., 2008). Its occurrence in drinking water has been investigated throughout Canada, the USA (Charrois et al., 2007), and Japan (Asami et al., 2009). A survey of 20 municipal drinking water systems in Alberta, Canada, showed concentrations of <5e100 ng/L (Charrois et al., 2007). A national survey of Japanese drinking water treatment plants revealed concentrations in finished water of <1e10 ng/L (Asami et al., 2009). NDMA in waters has been causing concern because of its risk to health. NDMA has been classified as a probable human carcinogen (B2) by the Integrated Risk Information System (IRIS) of the United States Environmental Protection Agency (USEPA, 1987). WHO (2008) has set the guideline value for NDMA in drinking water at 100 ng/L. Health Canada proposed a maximum acceptable concentration of NDMA in drinking water of 40 ng/L (Health Canada, 2010). In Japan, the Ministry of Health, Labor and Welfare added NDMA to items for further study in the setting of drinking water quality standards and adopted the WHO’s guideline value as the target in April 2010. NDMA likely reaches aquifers owing to its high polarity (log octanol/water partition coefficient ¼ 0.57) (Singer et al., 1977) and moderate biodegradation rate. Zhou et al. (2009b) estimated that 90% of NDMA by mass recharged from surface water to ground water was biodegraded over 7 years in Los Angeles, USA. In the USA, the detection of NDMA in ground water is commonly attributed to its use and release in association with asymmetrical dimethylhydrazine, a rocket fuel component, at aerospace facilities, or to the infiltration of effluent from municipal wastewater treatment plants (Fleming et al., 1996; Gunnison et al., 2000; Zhou et al., 2009b). The occurrence of NDMA in ground water in Japan has not been investigated. The use of ground water by some sectors such as hospitals, hotels, and small factories in Tokyo has increased recently. However, some aquifers in Tokyo are heavily polluted by high concentrations of NH3 and organic matter (Kuroda et al., 2007, 2008). Nakada et al. (2008) also revealed that ground water in Tokyo is contaminated by pharmaceuticals and personal care products, probably owing to leakage from decrepit sewer pipes, the past practice of sewage disposal underground, and infiltration by contaminated river water, and estimated that the average composition of ground water is w1% sewage across Tokyo. Since NDMA can be formed from precursors such as dimethylamine and natural organic matter by chloramination or chlorination in the presence of high concentrations of NH3 (Choi and Valentine, 2002; Mitch and Sedlak, 2002; Gerecke and Sedlak, 2003; Mitch et al., 2003; Chen and
Valentine, 2007), the formation of NDMA after disinfection of these waters is a matter of concern. It is now required to investigate NDMA and its potential for formation in ground water to avoid detrimental impacts to health. Although NDMA precursors, normally estimated by monochloramine reaction during 10 days (Mitch et al., 2003), have been measured in surface water (Gerecke and Sedlak, 2003; Schreiber and Mitch, 2006) and wastewater (Mitch and Sedlak, 2004; Sedlak et al., 2005; Pehlivanoglu-Mantas and Sedlak, 2006b), there are no studies of the occurrence of NDMA precursors in ground water over a wide area. Our research had three aims. First, we investigated the extent of the occurrence of NDMA in ground water in Tokyo, comparing the results from river water. Second, we evaluated the potential for NDMA formation by chlorination ðNDMA FPCl2 Þ and by chloramination ðNDMA FPNH2 Cl Þ and investigated the factors influencing it. NDMA FPCl2 was analyzed to mimic a practical chlorination process in Japan. NDMA FPNH2 Cl was analyzed to estimate total NDMA precursors. Third, we revealed the molecular weight distributions of NDMA precursors in ground water and river water. Mitch and Sedlak (2004) investigated their molecular weight distributions in municipal wastewater using a series of ultrafiltration membranes with cutoffs of 30, 10, and 3 kDa and showed a dominant size of <3 kDa. So we used membranes with cutoffs of 30, 3, and 0.5 kDa. To our knowledge, this is the first study to reveal the widespread occurrence and characterization of NDMA precursors in ground water.
2.
Materials and methods
2.1.
Ground water and river water sampling
We collected 23 samples from springs and from private and public wells in Tokyo from October 2009 to May 2010. During the daytime we also collected 18 samples from the surface of 5 rivers at 15 locations in Tokyodthe Iruma River (R.I), the Ara River (R.A1e4), the Edo River (R.E1e3), the Tama River (R.T1e4), and the Tsurumi River (R.TS1e3) (Fig. 1). Sampling dates, aquifer type, and basic water quality parameters are shown in Supplementary Tables S1 and S2. The samples were filtered through pre-baked GF/F glass fiber filters (pore size 0.7 mm, Whatman) and divided between 2 glass bottles for NDMA and NDMA-FP measurements. Sodium thiosulfate, a quenching agent, was added into the bottle for NDMA measurements. All samples were stored in the dark at 4 C before analysis.
2.2.
Chemical analysis
2.2.1.
Chemicals
NDMA was purchased from Supelco. NDMA-d6 (98%) was purchased from Cambridge Isotope Laboratories. HPLC-grade distilled water, formic acid, and acetonitrile were purchased from Wako Pure Chemical. Methanol and dichloromethane (DCM) of pesticide residue and PCB analysis grade were purchased from Kishida. Special grade sodium bicarbonate, sodium thiosulfate, and sulfuric acid, 1st grade monopotassium phosphate, and sodium hypochlorite and sodium
3371
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 6 9 e3 3 7 7
Fig. 1 e Sampling locations.
hydroxide were purchased from Kishida. Special grade ammonium chloride was purchased from Wako. Distilled water was passed through a tC18 cartridge (Waters) to remove trace levels of NDMA precursors and was then used to prepare reagent solutions for NDMA-FP measurements. Monochloramine (NH2Cl) solutions were prepared fresh daily following Mitch and Sedlak (2002). The monochloramine concentration was confirmed by the indophenol method using a colorimeter (HACH).
2.2.2. MS
NDMA analysis by solid-phase extraction and LC-MS/
NDMA in the filtrate was concentrated by a factor of w2500 by solid-phase extraction. To 500 mL of sample, 1 g of sodium bicarbonate and 5 ng of NDMA-d6 were added. The samples were passed through an EPA 521 method cartridge (Resprep) preconditioned with 10 mL DCM, 10 mL methanol, and 20 mL distilled water. The flow rate was 5 mL/min. The cartridge was then dried under a gentle stream of nitrogen gas. NDMA was
6
12
5
10
4
8
2
6 4
0
0
Unconfined aquifer
Confined aquifer
River water
Fig. 2 e NDMA concentrations in ground water and river water.
GW1807 GW1808 GW1904 GW0402 GW0907 GW1204 GW1207 GW1401 GW1402 GW1406 GW1801 GW1806 GW2102 GW2103 GW0303 GW0307 GW0308 GW0404 GW0502 GW0905 GW1206 GW1213 GW1810
2
R.I R.A1 R.A2 R.A3 R.A4 R.E1 R.E2 R.E3 R.T1 R.T2 R.T3 R.T4 R.TS1 R.TS2 R.TS3
1
Spring
NDMA-FP Cl2
Spring
Unconfined aquifer
Confined aquifer
R.I R.A1 R.A2 R.A3 R.A4 R.E1 R.E2 R.E3 R.T1 R.T2 R.T3 R.T4 R.TS1 R.TS2 R.TS3
NDMA (ng/L)
3
GW1807 GW1808 GW1904 GW0402 GW0907 GW1204 GW1207 GW1401 GW1402 GW1406 GW1801 GW1806 GW2102 GW2103 GW0303 GW0307 GW0308 GW0404 GW0502 GW0905 GW1206 GW1213 GW1810
NDMA (ng/L)
NDMA
River water
Fig. 3 e NDMA concentrations between before and after chlorination.
3372
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 6 9 e3 3 7 7
b
6
R.A2
5
Increased NDMA* (ng/L)
Increased NDMA* (ng/L)
a
GW0502 4
GW0303
3 2 1 0
Ground water
-1
River water
-2
6 5 4 3 2 1 0 Ground water -1
River water
-2 0
5
10
15
NH 3-N (mg/L)
0
50
100
150
200
250
NDMA precursors (ng-NDMA eq./L)
* Increase = NDMA-FPCl2 – NDMA. Fig. 4 e Relationship between increased NDMA due to chlorination and (a) NH3 and (b) NDMA precursors. *Increase [ NDMA FPCl2 e NDMA.
eluted with 10 mL DCM. The eluate was concentrated to 200 mL under nitrogen gas. NDMA was separated in an Acquity UPLC system (Waters) with a BEH C18 column (Waters). The mobile phase was composed of 0.1% formic acid aqueous solution (eluent A) and 100% acetonitrile (eluent B). The flow rate was 0.2 mL/min at all stages, and the sample injection volume was 30 mL. NDMA was detected with an Acquity TQD tandem mass spectrometer (Waters) operated in electrospray/chemical ionization positive-ion mode. Multiple reaction monitoring transitions were m/z 74.9e43.1 (quantification) and m/z 74.9e57.9 (confirmation) for NDMA and m/z 81.0e46.0 for NDMA-d6. NDMA concentrations were label-recovery corrected. The reproducibility and recovery rate were confirmed by spiking samples with 5 ng-NDMA standard in 500 mL. The rate of recovery from distilled water (n ¼ 5) was 116% and the
NDMA precursors (ng-NDMA eq./L)
200
150
relative standard deviation (RSD) was 4%. The rate of recovery from ground water (n ¼ 3) was 83%. The rates of recovery of NDMA-d6 were 78% from ground water and 72% from river water. The limit of quantification (LOQ) was 0.5 ng/L. An operational blank was run with every batch, and NDMA was normally less than LOQ.
2.2.3. Potential for NDMA-formation by chlorination ðNDMA FPCl2 Þ NDMA FPCl2 was analyzed by following the method for the investigation of disinfection byproducts (Japan Water Works Association, 2001). Briefly, 570 mL water was buffered with 30 mL 0.2 M monopotassium phosphate at pH 7.0 0.2, chlorinated by free chlorine, and then incubated at 20 C in the dark for 24 h. The residual free chlorine was kept at 1e2 mgCl2/L. Chlorine was analyzed by the DPD method using a colorimeter. The reactions were halted by the addition of sodium thiosulfate solution, and NDMA was measured. The reproducibility was confirmed by using ground water (n ¼ 4; RSD ¼ 10%). NDMA FPCl2 in the operational blank ranged from <0.5 to 0.9 ng/L.
2.2.4. Potential for NDMA-formation by chloramination ðNDMA FPNH2 Cl Þ
100
50
Spring
Unconfined aquifer
Confined aquifer
R.I R.A1 R.A2 R.A3 R.A4 R.E1 R.E2 R.E3 R.T1 R.T2 R.T3 R.T4 R.TS1 R.TS2 R.TS3
GW1807 GW1808 GW1904 GW0402 GW0907 GW1204 GW1207 GW1401 GW1402 GW1406 GW1801 GW1806 GW2102 GW2103 GW0303 GW0307 GW0308 GW0404 GW0502 GW0905 GW1206 GW1213 GW1810
0
River water
Fig. 5 e NDMA precursor concentrations in ground water and river water.
NDMA FPNH2 Cl was analyzed according to Mitch et al. (2003). Briefly, 510 mL water was mixed with 30 mL 0.2 M monopotassium phosphate, dosed with 60 mL 20 mM (1400 mgCl2/L) monochloramine solution at pH 6.8 0.2, and then incubated at 20 C in the dark for 10 days. The residual total chlorine was analyzed by the DPD method using a colorimeter. The reactions were halted by the addition of sodium thiosulfate, and NDMA was measured. The reproducibility was confirmed by using river water (n ¼ 4; RSD ¼ 4%). NDMA FPNH2 Cl in the operational blank using distilled water passed through a tC18 cartridge was 5.4 0.7 ng/L (arithmetic mean standard error; n ¼ 7), whereas those in the operational blank using Milli-Q water and distilled water were
3373
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 6 9 e3 3 7 7
a
b
200
100
150
NDMA precursors (ng-NDMA eq./L)
Ground water River water
100
50
0
60 40 20
TIN (mg/L)
50
500 1000 1500
20
40
15
30
10
20
5
0
0
0
10
NDMA precursors (ng-NDMA eq./L)
Ground water 80
Crotamiton (ng/L)
Fig. 6 e Relationship between NDMA precursors and (a) TIN and (b) crotamiton (Kuroda, 2010). (a) Ground water: r2 [ 0.02, P > 0.1; river water: r2 [ 0.81, P < 0.01. (b) Ground water: r2 [ 0.003, P > 0.1.
13.9 2.6 ng/L (n ¼ 2) and 8.6 1.5 ng/L (n ¼ 2), respectively. Use of the tC18 cartridge reduced NDMA FPNH2 Cl in the operational blank, indicating that the cartridge removes trace levels of NDMA precursors in distilled water. NDMA FPNH2 Cl in the operational blank was not subtracted from that in ground water and river water, because a proportion of the NDMA precursors may have come from the distilled water, even after it had been passed through the cartridge. We expected total NDMA precursors in the water samples to convert to NDMA during the 10 days of chloramination (Mitch et al., 2003; Mitch and Sedlak, 2004). Therefore, we considered the increased concentration to represent total NDMA precursors, and defined the precursors as NDMA FPNH2 Cl minus initial NDMA. The concentration was expressed as ng-NDMA equivalent/L (ng-NDMA eq./L). Under these conditions, monochloramine decays mainly by disproportionation and other autodecomposition reactions (Valentine and Jafvert, 1988; Vikesland et al., 1998). The
residual total chlorine concentrations were similar among all samples except two, and were approximately 9 mgCl2/L, which was comparable to a previous result (Mitch and Sedlak, 2004). NDMA precursor concentrations were possibly underestimated in samples GW0307 and R.A4, which showed 0.1 mg/L of residual total chlorine.
2.2.5.
Other chemical analyses
After the samples were filtered through CE membrane filters (pore size 0.45 mm, Advantec), NH3 was analyzed by colorimetry using a salicylate method or indophenol blue absorptiometry. After the samples were filtered through PTFE membrane filters (0.45 mm, ADVANTEC), dissolved organic carbon (DOC), NO2, NO3, and UV absorbance were analyzed. DOC was analyzed with a total organic carbon analyzer (TOCV, Shimadzu). NO2 and NO3 were analyzed by ion chromatography (761 Compact IC, Metrohm). The sum of NH3eN, NO2eN, and NO3eN was regarded as total inorganic N (TIN). UV absorbance at 254, 260, and 272 nm (UV254, UV260, UV272) was analyzed by spectrophotometer (U-2010, Hitachi).
200 Ground water
2.3.
NDMA precursors (ng-NDMA eq./L)
River water 150
100
50
0 0
2
4
6
8
10
DOC (mg/L) Fig. 7 e Relationship between NDMA precursors and DOC. Ground water: r2 [ 0.16, P [ 0.06; river water: r2 [ 0.47, P < 0.01.
Fractionation of NDMA precursors
The NDMA precursors in two ground water samples (GW0905 and GW0907) and two river water samples (R.TS3 and R.T3) were fractionated by molecular weight by the filtration-inseries method (Lohwacharin et al., 2009). The samples were sequentially fractionated through 30-kDa regenerated cellulose (PLTK), 3 kDa regenerated cellulose (PLBC), and 0.5 kDa cellulose acetate (YC05) membranes (Millipore) in a 2000 mL Millipore Amicon stirred cell unit. Before use, all membranes were immersed in distilled water for 24 h to remove the wetting agent. Distilled water passed through a tC18 cartridge was then flushed through the membranes for 30 min to obtain pure water permeability in a quasi-steady state. Dead-end filtration was operated at a constant trans-membrane pressure. Constant transmembrane pressures of 100, 207, and 401 kPa for the 30-, 3-, and 0.5 kDa membranes, respectively, were maintained during fractionation at a constant stirring rate of 100 rpm.
3374
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 6 9 e3 3 7 7
Fig. 8 e Molecular weight distributions of DOC and NDMA precursors. (a) DOC concentration in each molecular fraction. (b) Molecular weight distribution of DOC. (c) NDMA precursors in each molecular fraction. (d) Molecular weight distribution of NDMA precursors.
After fractionation, water samples were subjected to NDMA FPNH2 Cl , DOC, and UV absorbance measurements.
3.
Results and discussion
3.1.
Occurrence of NDMA in Tokyo
In ground water, NDMA was detected in 20 out of 23 samples, at <0.5e5.2 ng/L, with a median of 0.9 ng/L (Fig. 2). NDMA probably decreased during infiltration owing to biodegradation (Zhou et al., 2009b), gaseous diffusion, or volatilization (Arienzo et al., 2006). In river water, NDMA was detected at 13 out of 15 locations (16/18 samples) at <0.5e3.4 ng/L (median: 2.2 ng/L). Because the river water samples were collected from the surface during the daytime, direct photolysis of NDMA might have occurred in these samples (Plumlee and Reinhard, 2007). These results are comparable to those of Asami et al.
(2009), who reported a maximum NDMA concentration in surface water in Japan of 4.3 ng/L. NDMA concentrations in ground water were slightly lower than or comparable to those in river water. The highest concentrations were 5.2 ng/L in ground water and 3.4 ng/L in river water, or <10% of the WHO guideline for NDMA in drinking water. No strong relationships were found between NDMA and other water quality parameters (Table S3) in ground water and river water (Tables S4, S5).
3.2.
NDMA FPCl2
NDMA FPCl2 concentrations in ground water ranged from <0.5 to 10.8 ng/L, with a median of 1.8 ng/L (Fig. 3). Those in river water ranged from 0.7 to 7.8 ng/L (median: 2.3 ng/L). Concentrations were not much increased (<4.0 ng/L) in most samples, but increased from 0.6 to 5.2 ng/L in GW0303, from 5.2 to 10.8 ng/L in GW0502, and from 3.3 to 7.8 ng/L in R.A2.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 6 9 e3 3 7 7
Samples containing high concentrations of NH3 and NDMA precursors showed large increases in NDMA by chlorination (>4.0 ng/L) (Fig. 4). In bench-scale chlorine disinfection experiments, peak NDMA production occurred near the theoretical monochloramine maximum in the sub-breakpoint region of the disinfection curve (Charrois and Hrudey, 2007). The increased NDMA in our study was possibly caused by reactions between NDMA precursors and monochloramine that was unintentionally formed by the reaction between free chlorine and NH3 in the samples.
3.3.
Occurrence of NDMA precursors
Total NDMA precursors in ground water ranged from 4 to 84 ng-NDMA eq./L, with a median of 10 ng-NDMA eq./L (Fig. 5). Total precursors in river water ranged from 11 to 185 ngNDMA eq./L (median: 51 ng-NDMA eq./L). Concentrations in ground water were significantly lower than those in river water (t-test, P < 0.01). The difference indicates that the NDMA precursors were biodegraded, adsorbed, or volatized during infiltration. Although NDMA precursors are degraded only slightly in wastewater-impacted river water (Schreiber and Mitch, 2006; Pehlivanoglu-Mantas and Sedlak, 2006a), they disappeared during the relatively long process of infiltration into the ground water. To identify the sources of NDMA precursors in the water, we investigated the relationships between the precursors and TIN (Fig. 6a). There was a significant strong relationship in river water (r2 ¼ 0.81, P < 0.01). This relationship suggests that NDMA precursors in river water are derived from wastewater treatment plants, because most N in urban surface waters in Japan comes from wastewater treatment plants (Toyoda et al., 2009; Ohte et al., 2010). In contrast, there was no significant relationship in ground water (r2 ¼ 0.02, P > 0.1). Since N in ground water in Tokyo comes from a wide variety of sources, such as natural soils, fertilizers, and sewage (Kuroda et al., 2007), we analyzed crotamiton, a conservative marker of domestic sewage (Nakada et al., 2008). The crotamiton was analyzed in ground water collected from the same well in a different year (Kuroda, 2010). There was no significant relationship (Fig. 6b), suggesting that the major source of NDMA was not leakage of domestic sewage, or that there is a large difference in infiltration behavior between NDMA precursors and crotamiton. There were weak but significant relationships between NDMA precursors and DOC in ground water (r2 ¼ 0.16, P ¼ 0.06) and river water (r2 ¼ 0.47, P < 0.01) (Fig. 7), but no relationships with UV absorbance or specific UV absorbance (SUVA) (Fig. S1). The arithmetic mean standard error of the NDMA precursor-to-DOC ratio was 20 5 ng-NDMA eq./mg in ground water and 39 5 ng-NDMA eq./mg in river water. Again, this result indicates that NDMA precursors in ground water were less abundant than those in river water. After reacting water samples with monochloramine at 70 mg/L Cl2 for 7 days at pH 7.2 0.2, Gerecke and Sedlak (2003) found 1.1 ng-NDMA eq./ mg in ground water (n ¼ 1) and 3.5 0.7 ng-NDMA eq./mg in surface waters (n ¼ 7) in the USA. The NDMA precursor-toDOC ratios in our study were approximately one order of magnitude higher than those of Gerecke and Sedlak (2003), although the measurement of NDMA precursors differed
3375
between the two studies. Our results indicate that ground water and river water in Tokyo are heavily contaminated by NDMA precursors, possibly owing to urban activities.
3.4.
Molecular size distributions of NDMA precursors
The distributions of DOC were inconsistent between the two ground water samples tested (Fig. 8a and b, Table S6). DOC in GW0907 was distributed substantially in all four fractions: highest in the <0.5 kDa fraction (45%), followed by 3e30 kDa (27%), 0.5e3 kDa (21%), and >30 kDa (7%). But DOC in GW0905 was dominant in the >30 kDa (58%) and <0.5 kDa (38%) fractions, and very small in the 3e30 kDa (3%) and 0.5e3 kDa (1%) fractions. On the other hand, DOC was similarly distributed in the two river water samples tested (Fig. 8a and b, Table S6): dominant in the 3e30 kDa fraction (35%e40%), followed by 0.5e3 kDa (28%e32%), <0.5 kDa (18%e29%), and >30 kDa (7%e10%). The distributions of NDMA precursors (Fig. 8c and d, Table S7) were very different from those of DOC (above) and UV absorbance (Fig. S2, Tables S8eS10). There were also no clear relationships between NDMA precursors and SUVA (Fig. S3). The molecular weight distributions of NDMA precursors were inconsistent between the two ground water samples tested (Fig. 8c and d): NDMA precursors in GW0905 were dominant in the <0.5-kDa fraction (73%), whereas those in GW0907 were dominant in the 0.5e3 kDa fraction (53%). NDMA precursors in both river water samples were dominant in <0.5 kDa fraction (49%e57%), followed by 0.5e3 kDa (24%e43%). In general, NDMA precursors were dominantly (>70%) distributed in the <3 kDa fraction in both ground water and river water samples. This result is consistent with the same finding in municipal wastewater (Mitch and Sedlak, 2004).
4.
Conclusions
(1) NDMA concentrations in ground water ranged from <0.5 to 5.2 ng/L and were slightly lower than or comparable to those in river water. (2) NDMA was not greatly increased (<4.0 ng/L) by chlorination, except in two ground water and one river water samples. NDMA was increased greatly in samples containing high concentrations of NH3 and NDMA precursors. (3) Concentrations of NDMA precursors ranged from 4 to 84 ng-NDMA eq./L in ground water and from 11 to 185 ngNDMA eq./L in river water. There were weak but significant relationships between NDMA precursors and DOC in both sources, and the NDMA precursor-to-DOC ratios were 20 5 ng-NDMA eq./mg in ground water and 39 5 ngNDMA eq./mg in river water. NDMA precursors in ground water were less abundant than those in river water, indicating that the NDMA precursors were biodegraded, adsorbed, or volatized during infiltration. (4) The molecular weight of NDMA precursors in river water was dominant in the <0.5 kDa fraction, followed by 0.5e3 kDa. However, their distribution was inconsistent in ground water: one was dominant in the <0.5 kDa fraction, and the other in 0.5e3 kDa. Molecular weight distributions of NDMA precursors were different from those of DOC.
3376
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 6 9 e3 3 7 7
Acknowledgments This research was partially supported by JSPS Grants-in-Aid for Scientific Research (22760406, 21860018, and 19360237) and a CREST project grant for ‘Development of Well-balanced Urban Water Use System Adapted for Climate Change’ from the Japan Science and Technology Agency.
Appendix. Supplementary data Supplementary data related to this article can be found online at doi:10.1016/j.watres.2011.03.053.
references
Andrzejewski, P., Kasprzyk-Hordern, B., Nawrocki, J., 2008. N-nitrosodimethylamine (NDMA) formation during ozonation of dimethylamine-containing waters. Water Res. 42 (4e5), 863e870. Andrzejewski, P., Nawrocki, J., 2007. N-nitrosodimethylamine formation during treatment with strong oxidants of dimethylamine containing water. Water Sci. Technol. 56 (12), 125e131. Arienzo, M., Gan, J., Ernst, F., Qin, S., Bondarenko, S., Sedlak, D.L., 2006. Loss pathways of N -nitrosodimethylamine (NDMA) in turfgrass soils. J. Environ. Qual. 35 (1), 285e292. Asami, M., Oya, M., Kosaka, K., 2009. A nationwide survey of NDMA in raw and drinking water in Japan. Sci. Total Environ. 407 (11), 3540e3545. Charrois, J.W.A., Boyd, J.M., Froese, K.L., Hrudey, S.E., 2007. Occurrence of N-nitrosamines in Alberta public drinkingwater distribution systems. J. Environ. Eng. Sci. 6 (1), 103e114. Charrois, J.W.A., Hrudey, S.E., 2007. Breakpoint chlorination and free-chlorine contact time: Implications for drinking water Nnitrosodimethylamine concentrations. Water Res. 41 (3), 674e682. Chen, Z., Valentine, R.L., 2007. Formation of Nnitrosodimethylamine (NDMA) from humic substances in natural water. Environ. Sci. Technol. 41 (17), 6059e6065. Chen, W.H., Young, T.M., 2008. NDMA formation during chlorination and chloramination of aqueous diuron solutions. Environ. Sci. Technol. 42 (4), 1072e1077. Choi, J., Duirk, S.E., Valentine, R.L., 2002. Mechanistic studies of N-nitrosodimethylamine (NDMA) formation in chlorinated drinking water. J. Environ. Monit. 4 (2), 249e252. Choi, J.H., Valentine, R.L., 2002. Formation of Nnitrosodimethylamine (NDMA) from reaction of monochloramine: a new disinfection by-product. Water Res. 36 (4), 817e824. Fleming, E.C., Pennington, J.C., Wachob, B.G., Howe, R.A., Hill, D. O., 1996. Removal of N-nitrosodimethylamine from waters using physical-chemical techniques. J. Hazard. Mater. 51 (1e3), 151e164. Gerecke, A.C., Sedlak, D.L., 2003. Precursors of Nnitrosodimethylamine in natural waters. Environ. Sci. Technol. 37 (7), 1331e1336. Gunnison, D., Zappi, M.E., Teeter, C., Pennington, J.C., Bajpai, R., 2000. Attenuation mechanisms of N-nitrosodimethylamine at an operating intercept and treat groundwater remediation system. J. Hazard. Mater. 73 (2), 179e197. Health Canada, 2010. N-nitrosodimethylamine (NDMA) in Drinking Water Document for public comment.
Japan Water Works Association, 2001. Standard Methods for the Examination of Water (in Japanese). Kuroda, K., 2010. Ph.D Thesis. The University of Tokyo (in Japanese). Kuroda, K., Fukushi, T., Takizawa, S., Aichi, M., Hayashi, T., Tokunaga, T., 2007. Source estimation of nitrogen contamination in groundwaters in Tokyo metropolitan area. Environ. Eng. Res. 44, 31e38 (in Japanese). Kuroda, K., Fukushi, T., Takizawa, S., Murakami, M., Takada, H., Nakada, N., Aichi, M., Hayashi, T., Tokunaga, T., 2008. Groundwater Qual.: securing Groundwater Qual. Urban Industr. Environ.. Sources of and Factors Influencing Groundwater Contamination in the Tokyo Metropolitan Area, vol. 324. IAHS Publ.. pp16e23. Lohwacharin, J., Oguma, K., Takizawa, S., 2009. Ultrafiltration of natural organic matter and black carbon: factors influencing aggregation and membrane fouling. Water Res. 43 (12), 3076e3085. Mitch, W.A., Gerecke, A.C., Sedlak, D.L., 2003. A Nnitrosodimethylamine (NDMA) precursor analysis for chlorination of water and wastewater. Water Res. 37 (15), 3733e3741. Mitch, W.A., Sedlak, D.L., 2002. Formation of Nnitrosodimethylamine (NDMA) from dimethylamine during chlorination. Environ. Sci. Technol. 36 (4), 588e595. Mitch, W.A., Sedlak, D.L., 2004. Characterization and fate of N-nitrosodimethylamine precursors in municipal wastewater treatment plants. Environ. Sci. Technol. 38 (5), 1445e1454. Najm, I., Trussell, R.R., 2001. NDMA formation in water and wastewater. J. Am. Water Works Assoc. 93 (2), 92e99. Nakada, N., Kiri, K., Shinohara, H., Harada, A., Kuroda, K., Takizawa, S., Takada, H., 2008. Evaluation of pharmaceuticals and personal care products as water-soluble molecular markers of sewage. Environ. Sci. Technol. 42 (17), 6347e6353. Ohte, N., Tayasu, I., Kohzu, A., Yoshimizu, C., Osaka, K., Makabe, A., Koba, K., Yoshida, N., Nagata, T., 2010. Spatial distribution of nitrate sources of rivers in the Lake Biwa watershed, Japan: controlling factors revealed by nitrogen and oxygen isotope values. Water Resour. Res. 46 (7), W07505. Oya, M., Kosaka, K., Asami, M., Kunikane, S., 2008. Formation of N-nitrosodimethylamine (NDMA) by ozonation of dyes and related compounds. Chemosphere 73 (11), 1724e1730. Pehlivanoglu-Mantas, E., Sedlak, D.L., 2006a. The fate of wastewater-derived NDMA precursors in the aquatic environment. Water Res. 40 (6), 1287e1293. Pehlivanoglu-Mantas, E., Sedlak, D.L., 2006b. Wastewater-derived dissolved organic nitrogen: analytical methods, characterization, and effects - a review. Environ. Sci. Technol. 36 (3), 261e285. Plumlee, M.H., Reinhard, M., 2007. Photochemical attenuation of N-nitrosodimethylamine (NDMA) and other nitrosamines in surface water. Environ. Sci. Technol. 41 (17), 6170e6176. Schreiber, I.M., Mitch, W.A., 2006. Occurrence and fate of nitrosamines and nitrosamine precursors in wastewaterimpacted surface waters using boron as a conservative tracer. Environ. Sci. Technol. 40 (10), 3203e3210. Sedlak, D.L., Deeb, R.A., Hawley, E.L., Mitch, W.A., Durbin, T.D., Mowbray, S., Carr, S., 2005. Sources and fate of nitrosodimethylamine and its precursors in municipal wastewater treatment plants. Water Environ. Res. 77 (1), 32e39. Singer, G.M., Taylor, H.W., Lijinsky, W., 1977. Liposolubility as an aspect of nitrosamine carcinogenicity: quantitative correlations and qualitative observations. Chem.Biol. Interact. 19 (2), 133e142. Toyoda, S., Iwai, H., Koba, K., Yoshida, N., 2009. Isotopomeric analysis of N2O dissolved in a river in the Tokyo metropolitan area, rapid Commun. Mass Spectrom. 23 (6), 809e821.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 6 9 e3 3 7 7
USEPA, 1987. N-Nitrosodimethylamine (CASRN 62-75-9). http:// www.epa.gov/ncea/iris/subst/0045.htm (accessed 18.07.10.). Valentine, R.L., Jafvert, C.T., 1988. General acid catalysis of monochloramine disproportionation. Environ. Sci. Technol. 22 (6), 691e696. Vikesland, P.J., Ozekin, K., Valentine, R.L., 1998. Effect of natural organic matter on monochloramine decomposition: pathway elucidation through the use of mass and redox balances. Environ. Sci. Technol. 32 (10), 1409e1416. WHO, 2002. N-Nitrosodimethylamine (Concise International Chemical Assessment Document, 38) Geneva: 52.
3377
WHO, 2008. N-Nitrosodimethylamine in Drinking-water Background Document for Development of WHO Guidelines for Drinking-water Quality. Zhou, W.J., Boyd, J.M., Qin, F., Hrudey, S.E., Li X-, F., 2009a. Formation of N-nitrosodiphenylamine and two new N-containing disinfection byproducts from chloramination of water containing diphenylamine. Environ. Sci. Technol. 43 (21), 8443e8448. Zhou, Q.L., McCraven, S., Garcia, J., Gasca, M., Johnson, T.A., Motzer, W.E., 2009b. Field evidence of biodegradation of Nnitrosodimethylamine (NDMA) in groundwater with incidental and active recycled water recharge. Water Res. 43 (3), 793e805.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 7 8 e3 3 8 8
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Evaluation of low-copy genetic targets for waterborne bacterial pathogen detection via qPCR Shawn T. Clark a, Kimberley A. Gilbride a,*, Mehrab Mehrvar b, Andrew E. Laursen a, Vadim Bostan a, Ronald Pushchak c, Lynda H. McCarthy a a
Department of Chemistry and Biology, Ryerson University, 350 Victoria Street, Toronto, Ontario M5B 2K3, Canada Department of Chemical Engineering, Ryerson University, 350 Victoria Street, Toronto, Ontario M5B 2K3, Canada c School of Occupational and Public Health, Ryerson University, 350 Victoria Street, Toronto, Ontario M5B 2K3, Canada b
article info
abstract
Article history:
Recent developments in water quality research have highlighted difficulties in accurately
Received 12 October 2010
predicting the incidence of pathogens within freshwater based on the viability, culturability
Received in revised form
and metabolic activity of indicator organisms. QPCR-driven assays are candidates to replace
8 March 2011
standard culture-based methods, however, protocols suitable for routine use have yet to be
Accepted 27 March 2011
sufficiently validated. The objective of this study was to evaluate five oligonucleotide primers
Available online 5 April 2011
sets (ETIR, SINV, exoT, VS1 and ipaH2) for their potential applicability in qPCR assays to detect contamination from five waterborne bacterial pathogens (Escherichia coli O157:H7, Salmonella
Keywords:
Typhimurium, Campylobacter jejuni, Pseudomonas aeruginosa, and Shigella flexneri). An
Pathogen detection
enrichment-free qPCR protocol was also tested using S. Typhimurium-seeded source water,
Real-time quantitative PCR
combining membrane filtration and mechanical, chemical and enzymatic lysis techniques to
E. coli O157:H7
recover the bacterial cells. All five primer sets were found to have high specificity and
Water quality monitoring
sensitivity for the tested organisms. Four of the primers were able to detect pathogen loads as low as 10 cells/mL while 200 cells/mL of C. jejuni were detectable in pure culture. Although sensitivity decreased in an artificially contaminated environmental matrix, it was still possible to detect as few as 10 S. Typhimurium cells without enrichment. The primers and protocols evaluated in this study have demonstrated potential for further validation for possible application alongside traditional indicator techniques. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
For many years, public health risks from recreational, receiving and drinking waters have been assessed by routine monitoring of indicator organisms such as thermotolerant Escherichia coli to predict fecal contamination. Recent literature suggests that indicator presence correlates poorly with the incidence of various bacterial, viral and eukaryotic pathogens. A universal indicator that is representative of the actual number and multitude of current pathogens does not
presently exist. Differences in environmental conditions, increases in pathogen diversity, variation in the spatial distribution of pathogenic cells within the water column, and the sometimes unpredictable persistence of common indicators from non-fecal sources are believed to contribute toward these discrepancies (Polo et al., 1998; Lemarchand and Lebaron, 2003; Harwood et al., 2005; Dorner et al., 2007; Ahmed et al., 2008, 2009; Wilkes et al., 2009). Such findings have reduced confidence in relying solely on indicator organisms to assess water quality.
* Corresponding author. Tel.: þ1 416 979 5000x6354; fax: þ1 416 979 5044. E-mail address:
[email protected] (K.A. Gilbride). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.03.050
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 7 8 e3 3 8 8
Furthermore, standard culture-based detection procedures for estimating indicator numbers from water samples can be prone to under- or over-estimating indicator densities. Specificity issues have been documented such as the growth of nonindicator bacteria, including several Pseudomonas spp., which can mimic the physical appearance of fecal coliforms, express the b-D-glucuronidase enzyme and therefore hydrolyze the 4-methylumbelliferyl-b-D-glucuronide (MUG) substrate used for coliform selection (Griffin et al., 2001; Rompre´ et al., 2002). Interference from non-coliform bacteria with high levels of b-D-glucuronidase activity such as Aeromonas spp. has also been reported when present in equivalent concentrations to the indicators (Tryland and Fiksdal, 1998). False negatives on MUG-based selective media can also be prevalent, since up to 50% of environmental E. coli O157:H7 isolates lack glucuronidase activity (Martins et al., 1993; Momba et al., 2006; Walters et al., 2007; Khan et al., 2007; Hora´kova´ et al., 2008). Finally, detection media restricts the growth of many fastidious bacteria in addition to indicator cells in viable but nonculturable (VBNC), highly stressed and weakened or injured states (Khan et al., 2007). This may severely underestimate true indicator concentrations and thus mask public health threats, since less than 10% of environmental isolates are fully culturable (Rompre´ et al., 2002). These techniques, however, are still applied on the basis of cost-effectiveness and ease-of-use. Current thought has shifted towards the development of molecular assays to supplement existing procedures. DNAbased techniques such as the quantitative real-time polymerase chain reaction (qPCR) among others offer many advantages over biochemical methods. QPCR is an ideal candidate for water quality-directed assays due to their sensitivity and rapid cycling times. Using PCR protocols, as few as 2 to 20 cells/mL have been detected when enumerating E. coli and/or E. coli O157:H7 (Heijnen and Medema, 2006; Khan et al., 2007; Ram et al., 2008), Salmonella Typhimurium (Thompson et al., 2006; Wolffs et al., 2006), Aeromonas hydrophila (Khan et al., 2009) or Campylobacter jejuni (Yang et al., 2003, 2004) in environmental samples. As of yet these PCRbased protocols have not been suited for standardized use because of: lengthy analysis times (specifically those involving 18 h enrichment and extended sample processing times prior to confirmation), poor primer specificity and detection limits above the proposed infectious dose thresholds. Importantly, there is no universal consensus on the specific primer sets required for the detection of common waterborne pathogens that would consistently amplify the desired targets with high reproducibility, sensitivity and specificity. Numerous primers have been designed to target many different gene classes from both pathogenic and nonpathogenic forms of E. coli as well as other waterborne pathogens. These include genus specific primers for segments of the 16S rRNA gene (Khan et al., 2009), the 16-23S internal transcribed spacer (ITS) region (Khan et al., 2007, 2009) and the uidA gene of E. coli encoding b-D-glucuronidase (Bej et al., 1991; Juck et al., 1996; Heijnen and Medema, 2006; Khan et al., 2007; Maheux et al., 2009). In most cases when the uidA gene has been used as a target, non-specificity and cross reactivity with members of the closely-related Shigella genus and other uidA carrying organisms (Hora´kova´ et al., 2008; Maheux et al., 2009) has interfered with immediate positive identifications. This
3379
suggests that routine PCR analysis of indicators may produce false positives since species specific determinants are rare. Genetic determinants encoding virulence-associated traits such as the shiga toxin (stx) and the intimin producing attaching and effacing (eae) genes of diarrheagenic E. coli O157:H7 and non-O157 strains (Heijnen and Medema, 2006; Bonetta et al., 2010), the invasion associated gene A (invA) of S. Typhimurium (Wolffs et al., 2006; Thompson et al., 2006; Ahmed et al., 2009) and the invasion plasmid antigen H (ipaH ) gene of Shigella flexneri (Theron et al., 2001; Fan et al., 2008) involved in the entry into intestinal epithelial cells have also been examined. These are preferential targets because they exist in single or low-copy number since they are generally exclusive to the pathogenic variants of a given species or genus (Hacker et al., 1997) and can help facilitate quantification by direct conversion between copy and cell numbers (Ahmed et al., 2009). It contrast, it has been reported that detecting these rare sequences within the natural environment is difficult due to sample complexity and low probability of capturing pathogenic cells in small volume grab-samples (Straub and Chandler, 2003). Enrichment methods have been applied to dilute samples to increase sensitivity by resuscitating injured and weakened cells (Ibekwe et al., 2002; Yang et al., 2003; Nam et al., 2005; Mull and Hill, 2009; Bonetta et al., 2010). In most cases, while detection limits are enhanced, the original numbers of target and non-target cells increase uncontrollably, adding a degree of difficulty in accurately estimating initial target concentrations (Wolffs et al., 2006). Increased levels of background DNA may also either negatively affect PCR sensitivity by concealing low concentrations of the desired target (Heijnen and Medema, 2006; Mull and Hill, 2009) or function as a carrier for target DNA by inducing co-precipitation during isolation (Wilson, 1997). Attempts to concentrate the total nucleic acids in large volume samples by centrifugation (Khan et al., 2007; Liu et al., 2008; Mull and Hill, 2009) and filtration (Bertrand and Roig, 2007; Hora´kova´ et al., 2008; Liu et al., 2008) have been employed. While centrifugation increases the probability of detecting rare targets; it does not differentiate between DNA from viable cells, dead cells or the extracellular DNA from dead cells. This can cause false positives in qPCR (D’Urso et al., 2009). As a result, extraction/purification schemes must be sufficiently optimized in order to allow even the most dilute DNA to be recovered with high purity. Limited literature is available on enrichment-free qPCR. Membrane filtration has been suggested to remove DNA from dead and severely damaged cells in complex samples (D’Urso et al., 2009) and has been tested in recent studies (Bertrand and Roig, 2007; and Liu et al., 2008) as enrichment replacements with satisfying results. The objective of this study was to evaluate five PCR primer sets having low-copy gene targets for their specificity and sensitivity in detecting the bacterial pathogens E. coli O157:H7, C. jejuni, S. Typhimurium, S. flexneri and Pseudomonas aeruginosa, by qPCR. The translocated intimin receptor (tir) gene involved in epithelial attachment and found in two copies chromosomally within the locus of enterocyte effacement (LEE) was selected for diarrheagenic E. coli O157:H7 detection (Trabulsi et al., 2002). The single-copy chromosomal genes invA of S. Typhimurium (Fey et al., 2004) and VS1 of C. jejuni which
3380
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 7 8 e3 3 8 8
includes regions specific to C. jejuni (Stonnet and Guesdon, 1993) were also chosen. The tri-copy effector protein exotoxin T (exoT ) which inhibits internalization of P. aeruginosa in eukaryotic cells was also selected (Feltman et al., 2001). Finally, the chromosomal and plasmid-borne ipaH gene of S. flexneri involved in the epithelial invasion mechanism and present in an estimated seven copies (Ashida et al., 2007) was also selected. An enrichment-free filtrationeextraction procedure was also tested in an attempt to reduce extracellular DNA prior to qPCR analysis. Detection limits were tested both in pure culture and in artificially contaminated source water using S. Typhimurium to predict sensitivity of the procedure.
2.
Materials and methods
2.1.
Reference strains and growth conditions
Strains of S. Typhimurium ATCC 14028 and C. jejuni NCTC 11168 were obtained from Dr. Dae-Young Lee (University of Guelph, Guelph, ON, Canada) and Dr. Eytan Wine (University of Alberta, Edmonton, AB, Canada) respectively. Enterohemorragic E. coli O157:H7 ATCC 700927, S. flexneri ATCC 12022 and P. aeruginosa ATCC 27853 (Cedarlane, Burlington, ON, Canada) were also used as reference strains in this study. Pathogens were selected on the basis of infectivity and potential role in waterborne disease. All bacterial strains with the exception of C. jejuni, were grown for 16e18 h (37 C, 30 C for P. aeruginosa) to a cell density of 1 109 cells/mL in 3 g/L Tryptic Soy Broth (TSB) (EMD Chemicals Inc., Mississauga, ON, Canada). The C. jejuni strain was grown on 5% anti-coagulated sheep’s blood agar under microaerophilic conditions using the BBL CampyPak Plus system (Beckton, Dickinson and Company, Franklin Lakes, NJ, USA) and incubated at 42 C for 5 days as per manufacturer’s recommendation.
2.2.
PCR primer design and selection
A total of five primer sets were evaluated on the basis of specificity and sensitivity (Table 1). The ipaH2 and exoT primers were custom designed to target the ipaH gene of S. flexneri (GenBank Accession No. M32063) and exoT gene of P. aeruginosa (Genbank Accession No. L46800) using the Primer3 (http://frodo. wi.mit.edu/primer3/) and LightCycler Probe Design2 softwares
(Roche Diagnostics, Laval, QC, Canada) respectively. The VS1 primers designed to target the VS1 gene from C. jejuni (Stonnet and Guesdon, 1993) were chosen since a detection limit of 1 cell/mL (Yang et al., 2003, 2004) was previously reported. The remaining primer sets were retrieved from a previous publication in our laboratory (Haffar and Gilbride, 2010) for the tir gene of several diarrheagenic pathotypes of E. coli O157:H7 and invA gene of S. Typhimurium (Table 1). Custom designs were chosen from the top five computer-generated primer sets based on user-defined selection criteria adapted from both Apte and Daniel (2003) and Dieffenbach et al. (1993). Sequence information was obtained from the GenBank database (http://www. ncbi.nlm.nih.gov/Genbank/index.html). Predictions of the formation of hairpins, self-dimers and hetero-dimers were also examined using the IDT SciTools OligoAnalyzer 3.1 program (http://www.idtdna.com/SCITOOLS/scitools.aspx) (Integrated DNA Technologies Inc., USA). Initial specificity was determined using the BLAST feature from the NCBI database (http://www. ncbi.nlm.nih.gov/BLAST/). Primers were synthesized at the DNA Synthesis Facility in the Center for Applied Genomics (Hospital for Sick Children, Toronto, ON, Canada).
2.3.
DNA extraction for specificity studies
Aliquots (1 mL) of the 1 109 cells/mL cultures were concentrated at 13 000 g (15 min) and pellets were re-suspended in 200 mL of 1 PBS solution. For C. jejuni, individual colonies were aseptically removed from the culture medium and resuspended in 200 mL of 1 PBS solution. Genomic DNA was extracted and purified from each strain using the High Pure PCR Template Purification Kit (Roche Diagnostics, Laval, QC, Canada) according to the manufacturer’s protocol. Purified DNA was stored at 20 C at a concentration of 20 ng/mL. DNA concentration and purity (A260/280) were determined spectrophotometrically with the Eppendorf BioPhotometer (Eppendorf AG, Hamburg, Germany) prior to use in PCR. Only DNA with an A260/280 ratio larger than 1.8 was used in PCR.
2.4. End-point PCR assays and specificity determinations PCR reactions were performed using the MyCycler Thermal cycler (BioRad Laboratories Inc., Mississauga, ON, Canada).
Table 1 e Characteristics of the oligonucleotide primers selected to target the specific waterborne bacterial pathogens chosen in this study. Pathogen
Primer
Target gene
E. coli O157:H7
ETIR
tir
S. flexneri
ipaH2
ipaH
S. Typhimurium
SINV
invA
P. aeruginosa
exoT
exoT
C. jejuni
VS15/6
VS1
Sequence (50 /30 )
Length (bp)
Annealing temperature ( C)
Amplicon size (bp)
Reference
F: GTCAGCTCATTAACTCTACGGG R: GCCTGTTAAGAGTATCGAGCG F: ATAATGATACCGGCGCTCTG R: CGGCTTCTGACCATAGCTTC F: TATGCCCGGTAAACAGATGAG R: GTATAAGTAGACAGAGCGGAGG F: GGTCTCTATACCAACGGCGA R: GAACAGGGTGGTTATCGTGC F: GAATGAAATTTTAGAATGGGG R: GATATGTATGATTTTATCCTGC
22 21 20 20 21 22 20 20 21 22
59
207
64
247
Haffar and Gilbride (2010) This study
59
252
65
285
56
358
Haffar and Gilbride (2010) This study Stonnet and Guesdon (1993)
3381
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 7 8 e3 3 8 8
Reactions (25 mL) contained 0.5 mL of 50 TITANIUM Taq, 2.5 mL of 10 TITANIUM Taq PCR buffer and 0.2 mM dNTP mixture from the TITANIUM Taq PCR kit (Clontech Laboratories Inc., Mountainview, CA, USA) in addition to 0.2 mM of the respective primer pairs, 2 mg/mL BSA (Roche Diagnostics, Laval, QC, Canada) and 2 mL of positive control template (20 ng/ mL). Volumes were adjusted to 25 mL with sterile MilliQ water, which was also used in place of DNA template for no template control (NTC) reactions. Cycling consisted of an initial denaturation at 95 C (2 min) followed by of 35 cycles of denaturation at 95 C (1 min), annealing between 49 and 65 C (30 s) (Table 1), extension at 72 C (1 min), followed by a second extension at 72 C (1 min). PCR products were confirmed by agarose gel electrophoresis (2%, 120 V for 35 min) using the iMupid 2-Plus Mini Agarose Gel Electrophoresis System (Helixx Technologies Inc., Scarborough, ON, Canada). Ethidium bromide stained gels (30 mg/mL) were visualized using the BioDoc-It Imaging System (UVP, Upland, CA, USA). Amplicon size was confirmed by comparison with the GeneRuler 100 bp DNA Ladder (Fermentas Life Sciences, Burlington, ON, Canada). Primer specificity was examined using the respective test strain as a positive control against template DNA from a variety of related enteric bacteria (Table 2). Reactions (25 mL) were prepared as above using 2 mL of the various DNA templates (20 ng/mL). Annealing temperatures for ETIR, SINV and VS1 were performed according to Stonnet and Guesdon (1993) and Haffar and Gilbride (2010) (Table 1). The presence of a single amplicon of the anticipated size generated with only the positive control template deemed primer sets suitable for qPCR.
2.5.
qPCR assays
Amplification reactions (20 mL) consisted of 4 mL of 5 Reaction Mix from the LightCycler FastStart DNA MasterPLUS SYBR Green I kit (Roche Diagnostics, Laval, QC, Canada), 0.2 mg/mL BSA, 0.6 mM MgCl2, 0.25 mM of each primer, 2 mL of template
(ranging from 20 ng to approximately 2 fg) and the remaining volume with sterile MilliQ water. The amplification protocols for each primer set are listed in Table 3. Melting curves were generated following amplification to confirm the identity of the desired end products. Cycling consisted of an initial heating at 95 C (0 s) followed by a 0.1 C/s increase from 65 to 95 C for 1 min. Standard curves of crossing point (CP) relative to the logarithm of defined bacterial cell numbers were generated through qPCR for each primer set and used to define the limit of quantification (LOQ) and limit of detection (LOD) for each assay. The LOQ refers to the lowest concentration of cells that was accurately quantified by the LightCycler instrument and within the linear dynamic range. The LOD was the lowest cell concentration which generated a positive product as confirmed via melting curve analysis but was unquantifiable. Curves were constructed in the LightCycler 2.0 system (Roche Diagnostics, Laval, QC, Canada) by amplifying genomic DNA from 10-fold serial cell dilutions of each reference strain (ranging from 1 108 cells/mL to 1 cell/mL) using the LightCycler Software 4.0 (v.4.0.0.23). Cell concentrations were determined by triplicate spread plating of the appropriate dilution onto 3 g/L Tryptic Soy Agar (TSA) (37 C, 16e18 h). Enumeration of C. jejuni was achieved by converting the DNA concentration in nanograms to cell equivalents, knowing a single cell contains 1.68 fg of DNA (Stonnet and Guesdon, 1993).
2.6.
Membrane filtration protocols for cell recovery
The 0.22 mm GSWP (Millipore Canada Ltd., Etobicoke, ON, Canada), 0.22 mm GVWP Durapore (Millipore Canada Ltd., Etobicoke, ON, Canada) and 0.45 mm GN-6 (Pall Corporation, USA) membrane filters were tested for their ability to isolate bacterial cells from water. Cell recovery from filters was examined using combinations of mechanical, enzymatic and chemical treatments modified from Ahmed et al. (2008, 2009) and Krsek and Wellington (1999) (Table 4). In each case S.
Table 2 e Primer specificity determination in end-point PCR using DNA from target and non-target bacterial strains. Organisms tested for specificity
Campylobacter jejuni Enterobacter aerogenes Enterococcus faecalis Escherichia coli Escherichia coli Escherichia coli Escherichia coli Escherichia coli O157:H7 Klebsiella pneumoniae Klebsiella pneumoniae Pseudomonas aeruginosa Salmonella enterica Salmonella Typhimurium Shigella flexneri Streptococcus faecalis
Primer seta,b
Type strain no.
NCTC 11168 ATCC 13048 ATCC 19433 ATCC 11229 ATCC 23723 ATCC 23725 ATCC 25922 ATCC 700927 ATCC 13882 ATCC 13887 ATCC 27853 ATCC 13314 ATCC 14028 ATCC 12022 ATCC 19432
ETIR
SINV
ipaH2
exoT
VS1
þ
þ
þ
þ
þ
a () No amplicon was generated following PCR using DNA from the specified organism. b (þ) Expected amplicon was generated following PCR with DNA from the specified organism.
3382
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 7 8 e3 3 8 8
Table 3 e QPCR detection limits for each test pathogen with their respective primer set under reference conditions. Primer set
qPCR protocol Denaturation
ETIR ipaH2 SINV exoT VS1 a b c d e f
95 C 95 C 95 C 95 C 95 C
for for for for for
15s 20s 20s 15s 0s
Annealing
58 C 64 C 60 C 60 C 56 C
for for for for for
10s 10s 20s 15s 10s
Elongation 72 C 72 C 72 C 72 C 72 C
for for for for for
10s 10s 20s 15s 20s
Correlation coefficient (R2)
Reaction efficiency
I.D. of target pathogena (# cells)
0.995 0.996 0.962 0.938 0.994
93.3% 92.8% 95.8% 105.3% 90.0%
10e100b 10e104c 104e107d 108e109e 500f
Melting peak ( C) 85.4 86.4 85.1 91.7 76.3
0.18 0.08 0.29 0.19 0.18
LOD (cells/mL) 10 10 10 10 200
I.D. or infectious dose represents the minimum number of cells required for infection in a suitable host. Liu et al., 2008. Theron et al., 2001. Maki and Hicks, 2002. Rusin et al., 1997. Yang et al., 2004.
Typhimurium ATCC 14028 was chosen to demonstrate the ability to detect a single-copy target from an artificially contaminated environmental matrix without enrichment. The invA gene is present in single copy on the S. Typhimurium chromosome (Fey et al., 2004). Samples (100 mL, n ¼ 3) of autoclaved (121 C for 20 min), low turbidity source water from DeCew Falls (Thorold, ON, Canada) were seeded with 1 107 S. Typhimurium ATCC 14028 cells, filtered and processed one of the following ways: Method 1: 0.45 mm filters were placed into 2 mL screwcap tubes containing a lysis buffer formulation [250 mM TriseHCl (pH 8.0), 20% SDS] and 0.2 g of 0.5 mm diameter Zirconia/Silica beads (BioSpec Products Inc., Bartlesville, OK, USA). Tubes
were bead beat for 10 min by vortexing and the supernatant was placed into a new 2 mL tube. Following the addition of a second aliquot of lysis buffer (300 mL) tubes were bead beat (10 min) and supernatants pooled. Method 2: The method described by Ahmed et al. (2008, 2009) was modified by the addition of bead solution from the UltraClean Soil DNA Isolation kit (MOBIO Laboratories Inc., Carlsbad, CA, USA) to 15 mL tubes of sterile STE buffer. Tubes were then bead beat for 10 min by vortexing. Method 3: The UltraClean Water DNA Isolation Kit (MOBIO Laboratories Inc., Carlsbad, CA, USA) was tested with the 0.22 mm GSWP filters as per manufacturer’s protocol. Method 4: The modifications to the filtration method from Ahmed et al. (2008, 2009) (Method 2) were used to test the
Table 4 e Percent recovery of S. Typhimurium DNA from artificially contaminated source water using various filtrationepurification methods. Method 1 2 3 4 5 6 7 8 9 10 Non-filtered
Components TriseHCl, 20% SDS (2), Zirconia/ Silica beads (0.5 mm diameter), bead beating STE buffer, coarse beads from UltraClean Soil DNA Isolation kit, bead beating UltraClean Water DNA Isolation kit (manufacturer’s protocol) STE buffer, coarse beads from UltraClean Soil DNA Isolation kit, bead beating TriseHCl, 20% SDS (2), coarse beads from UltraClean Soil DNA Isolation kit, bead beating TriseHCl, 20% SDS (2), Zirconia/Silica beads (0.5 mm diameter), bead beating TriseHCl, STE buffer, Zirconia/Silica beads (0.5 mm diameter) TriseHCl, proteinase K, lysozyme, Zirconia/Silica beads (0.5 mm diameter), bead beating TriseHCl, 20% SDS, proteinase K, lysozyme, Zirconia/Silica beads (0.5 mm diameter), bead beating TriseHCl, 20% SDS, proteinase K, lysozyme, Zirconia/Silica beads (0.5 mm diameter), bead beating Treated with HighPure PCR Template preparation kit (Manufacturer’s protocol)
Volume (mL)
% Recoverya
0.45
15
33.7 % 1.2%
0.45
15
34.2% 2.0%
þ
0.22 GSWP
15
10.4 % 1.8%
0.22 GSWP
2
11.5% 1.4%
0.22 GSWP
2
12.0% 3.0%
þþ
0.22 GSWP
2
21.0% 0.4%
þ
0.22 GSWP
2
32% 1.6%
þ
0.22 GSWP
2
40% 3.0%
0.22 Durapore
2
61.6% 1.5%
þþ
0.22 GSWP
2
73.3% 5.2%
þþ
None
2
80.4%b 6.5%
þþ
Filter size (mm)
Purityc
a A ratio of the amount of DNA (in ng) recovered post-filtration relative to the amount of DNA (in ng) isolated without filtration. b Calculated using a theoretical DNA yield of 50 ng, assuming each S . Typhimurium chromosome has a weight of 5 fg/cell (Malorny et al., 2003). c () DNA had low purity (A260/280 < 1.8) and was unamplified in end-point PCR, (þ) DNA had suitable purity (A260/280 ¼ 1.8) but PCR-amplifiable, (þþ) DNA had high purity (A260/280 >1.8) and was PCR-amplifiable.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 7 8 e3 3 8 8
0.22 mm GSWP membrane filters. Tubes were then bead beat for 10 min by vortexing. Method 5: The 0.22 mm GSWP filters were tested using the procedure for Method 1, however the Zirconia/Silica beads were replaced with the bead solution from the UltraClean Soil DNA Isolation kit (MOBIO Laboratories Inc., Carlsbad, CA, USA). Tubes were then bead beat for 10 min by vortexing. Method 6: The procedure for Method 1 was modified to use the 0.22 mm GSWP filters. Method 7: Method 1 was modified to replace the STE buffer solution with 20% SDS. Method 8: The procedure for Method 6 was modified with the addition of lysozyme (10 mg/mL) (37 C for 15 min) and proteinase K (70 C for 10 min) treatments as modified from Krsek and Wellington (1999). A 20% SDS solution then was added to the tubes following enzymatic treatment and subsequent bead beating for 10 min was then performed. Method 9: The 0.22 mm Durapore filters were placed in 2 mL screwcap tubes containing the novel lysis buffer formulation [250 mM TriseHCl, 20% SDS], Zirconia/Silica beads (BioSpec Products Inc., Bartlesville, OK, USA). The enzymatic treatment described in Method 8 was then applied and tubes were bead beat (10 min) by vortexing. Method 10: The procedure from Method 9 was tested with the 0.22 mm GSWP filters. Following treatment, all tubes were centrifuged (10 000 g, 10 min), pellets were re-suspended in 1 PBS solution, and DNA purified by the High Pure PCR Template Purification Kit (Roche Diagnostics, Laval, QC, Canada). Recoveries were assessed semi-quantitatively by comparing the ratio between the DNA content recovered from each filtration variant to that of a nonfiltered sample treated as per reference strains (Section 2.3).
2.8.
Source water processing and pathogen detection
To verify the ability to detect low-copy targets from environmental samples by the enrichment-free filtration method, S. Typhimurium ATCC 14028 was used to artificially contaminate source water from DeCew Falls (100 mL). Defined cell concentrations (1 109 to 10 cells/mL) were inoculated into 9 aliquots (100 mL each) of autoclaved source water in triplicate. Each sample was then filtered and treated as per extraction Method 10, which produced the highest cell recovery following earlier testing. QPCR reactions were prepared using the purified template from each cell dilution and tested with the SINV primers (Section 2.5). The optimal temperature profile was cycled 50 times (Table 3). The LOQ and LOD were defined by creating a second standard plot of CP relative to cell number. Amplicons were confirmed by melting curve analysis and compared to those generated in pure culture.
3.
Results
A total of five primer sets were evaluated. Three were selected from previous work, ETIR and SINV (Haffar and Gilbride, 2010) and VS1 (Stonnet and Guesdon, 1993) to target E. coli O157:H7, S. Typhimurium and C. jejuni respectively. Two additional primer sets were newly designed (exoT and ipaH2) to target P. aeruginosa and S. flexneri respectively. Target sequences were chosen because of their existence in single or low-copy numbers (Ashida et al., 2007; Ahmed et al., 2009; Feltman et al., 2001; Goosney et al., 2000; Stonnet and Guesdon, 1993), as well as being associated in some way with their virulence.
3.1. 2.7.
3383
PCR primer selection, optimization and specificity
Microscopy
The effect of filtration was examined qualitatively by assessing the ability to remove viable and/or dead cells from each of the 3 filter types tested in this study using viability staining and microscopy. Samples of 1 107 S. Typhimurium ATCC 14028 cells/mL were pre-stained (15 min at 25 C) with a 1:1 mixture of SYTO 9 and propidium iodide (0.3% DMSO) using the LIVE/DEAD BacLight Bacterial Viability kit for microscopy and quantitative assays (Molecular Probes Inc., Willow Creek, OR, USA) as per manufacturer’s protocol. Stained cells were then inoculated into duplicate 100 mL samples of distilled water and filtered through each of the 3 filter types. Method 10 was selected for evaluation as it was the most effective at removing the largest fraction of cells from the membrane filters. Adherence and removal of cells pre- and post-filtration was examined microscopically under epifluorescence with both GFP and dsRed filters using a Leica Model DM5000B upright microscope (Leica Microsystems, Canada) at 630 magnification. Prior to microscopy, filters were mounted onto glass slides with a drop of BacLight mounting oil (Molecular Probes Inc., Willow Creek, OR, USA) and coated with a drop of CITIFLUOR Anti-Fading Agent AF2 (Citifluor Ltd., London, UK). Controls consisted of 1 mL aliquots each of S. Typhimurium cells, distilled water and autoclaved environmental water to check for autofluorescence.
Results for the BLAST search at both the genus and species level revealed that the primer sets were highly specific for the desired organisms. In the case of the Shigella-specific ipaH2 primers, Shigella boydii was also matched to the target sequence. Since this strain is pathogenic and also possesses the ipaH gene such information did not eliminate the ipaH2 primer set from further analysis. An amplicon of the anticipated size (207e358 bp) was obtained with each primer set (Fig. 1) following protocol optimization in end-point PCR. This confirmed that only the product of interest was generated using optimal cycling conditions. In addition, strong amplification signals were produced in the qPCR assays when tested alongside their respective targets. Reaction artifacts (<100 bp) were also occasionally observed, however they were easily distinguished from the products of interest on melting curves and had no effect on amplification efficiencies. Primer sets showed 100% specificity for the intended gene targets when tested experimentally against non-target DNA in end-point PCR (Table 2). This was indicated by an amplicon generated only with the respective control strain (Table 2) and supports the earlier BLAST result. Importantly, no primer set exhibited cross-reactivity with DNA from any of the 5 pathogens selected in the study. Specificity had not been widely examined against an array of non-Campylobacter DNA in any of the three previous studies where VS1 primers have been applied (Stonnet and Guesdon, 1993; Yang et al., 2003; 2004).
3384
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 7 8 e3 3 8 8
Fig. 1 e Amplicons generated following end-point PCR amplification of each primer set with DNA from their respective positive control strain (2% agarose gel, 120 V). (1), (7) 100 bp DNA ladder (Fermentas), (2) C. jejuni NCTC 11168 (358 bp), (3) P. aeruginosa ATCC 27853 (285 bp), (4) S. Typhimurium ATCC 14028 (252 bp), (5) S. flexneri ATCC 12022 (247bp), (6) E. coli O157:H7 ATCC 700927.
3.2.
Sensitivity assays
Linear relationships between Cp and cell concentrations were observed in all cases when DNA from serial cell dilutions was tested in qPCR. Correlation coefficients (R2) ranged from 0.939 to 0.995 depending on the primer set (Table 3). Reaction efficiencies fell within the ideal range of 90e100% when determined by the LightCycler analysis program (V.4.0) (Roche Diagnostics, Laval, QC, Canada) (Table 3). The LOQ was found to be 1 102 cells/mL with four of the five primer sets (ETIR, SINV, ipaH2 and exoT). Although quantification below this threshold could not be accurately determined using only standard curves, detection (LOD) as low as 10 cells/mL was possible by observing a peak corresponding to a positive reaction in the melting curve analysis (Table 3). A similar trend was noted by Khan et al. (2007) where 1 cell/mL was detectable but unquantifiable within the defined linear range of the standard curve. With the VS1 primers, the LOD was found to be 200 cells/mL of C. jejuni NCTC 11168. The 1 cell/mL LOD obtained by Yang et al. (2003, 2004) was not reproducible by the present study. Data for all primer sets (Table 3) indicated that the LOD values in pure culture were at or below the minimum infectious doses for each tested pathogen (Table 3). In addition, the amplification protocols (Table 3) for any of the five primer sets were rapid, with a maximum length of 1 h required to amplify, quantify and confirm each target.
3.3.
Optimization of cell and DNA recovery
Ten methods were tested for whole cell and DNA recovery by enrichment-free concentration on membrane filters. The average percentage of the original 1 107 membrane-bound S. Typhimurium cells recovered with each of the 10 methods ranged from 10.4% 1.3%e73.3% 5.2% (Table 4). This indicates a relatively large variability in cell and subsequent DNA
recovery across the filterepurification variants, when examined semi-quantitatively. The majority of the methods were found to be inefficient at removing cells from membrane filters. The commercially available UltraClean Water DNA isolation kit produced the lowest recoverable yields (Table 4). For methods with low recoveries, it was assumed that cells and DNA were unsuccessfully removed from filters by the extraction formulations. Membrane-bound cells were recovered most efficiently by methods that used 0.22 mm filters treated with TriseHCl, 20% SDS, proteinase K, lysozyme and Zirconia/Silica beads (0.5 mm diameter). In particular, the highest recovery (73.3%) and purest template was achieved using Method 10 (Table 4). For these reasons, this method was selected to examine detection sensitivity. To determine recoveries, the amount of DNA obtained by each method was compared to a non-filtered S. Typhimurium sample where DNA was extracted by the purification scheme described in Section 2.3, without filtration. An estimated 80.4% of the theoretical DNA content from 1 107 S. Typhimurium cells was isolated by conventional purification. This value was thus used as a reference in place of the theoretical 50 ng yield for recovery comparisons, since it is known that 100% recovery is unachievable by any DNA purification scheme. A 7.1% difference in yield was observed between samples treated with Method 10 and those that were nonfiltered. This indicates that filtration reduced cell recovery, however, standard error may also account for some of this difference. Examining cells under epifluorescence following viability staining determined that 100% of the cells in the initial overnight culture were live prior to the source water seeding experiments. The successful entrapment and removal of cells from each filter type using Method 10 was confirmed by the presence and absence of live cells respectively. Examination of the cell suspension after removal from the filters revealed that approximately 95% of the original cells were recovered and retained viability prior to lytic treatment (Data not shown).
3.4. Detection sensitivity following filtrationepurification The standard curve generated post-filtration by amplifying 10-fold serially diluted S. Typhimurium cells with the SINV primer set produced a linear regression (R2 ¼ 0.995) having an LOQ of 1 103 cells/mL which appeared at a CP of 39. Melting curve analysis allowed us to detect the LOD of 10 cells for the single-copy invA gene, with a melting peak of approximately 85.05 C. This would indicate that our enrichment-free protocol for the environmental sample was as sensitive for detecting low numbers of a single-copy gene as pure culture conditions. However, relative to pure culture, the CP values were shifted by an average of 4 cycles per dilution, suggesting that some DNA was lost following the filtration procedure.
4.
Discussion
The enhanced sensitivity and rapid analysis times of qPCR are currently not achievable with many other existing molecular detection techniques such as microarrays and biosensors.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 7 8 e3 3 8 8
While it is not a recent discovery, there has been renewed interest to integrate these systems for standardized biomonitoring within the food industry, clinical diagnostic settings and water quality monitoring for the aforementioned reasons. Several analytical methods have been issued for standardized qPCR-based testing such as the Qualicon BAX System (DuPont Qualicon, Wilmington, DE, USA) to detect Listeria monocytogenes, E. coli O157:H7 and Salmonella spp. in foods, as well as specific Foodproof qPCR pathogen testing kits (EMD Chemicals Inc., Gibbstown, NJ, USA). It is only after scrutinization through intensive testing before accreditation by the European Commission, Health Canada and Association of Analytical Communities (AOAC) occurs (Hoorfar and Cook, 2003). Similar procedures have not been adequately developed for water-related analyses. Thus, the goal of present study was to evaluate several primer sets designed to target genes from common waterborne pathogens by examining their effectiveness and potential applicability in such assays. Primer suitability was evaluated on the basis of target specificity and detection limit. Each primer set was highly specific for the respective genetic target, as confirmed by sequence alignments and experimental determinations. No products were amplified in the presence of template DNA from non-target bacteria. As expected, cross-reactivity was not observed amongst any primer set and the five test strains, giving all primers a degree of robustness and validating the initial selection criteria. For our purposes, primer specificity was considered a critical parameter because only amplification from the particular gene target would be desired with a complex environmental water sample. Maheux et al. (2009) demonstrated the effect of cross-reactivity between many previously published E. coli-targeted primers and Shigella spp. It was suggested that detecting metabolic genes such as uidA does not give the specificity required for standard PCR assays since they may exaggerate cell numbers and complicate identification since they are not primarily associated with pathogens. Virulenceassociated targets were used in the present study to increase the likelihood of detecting pathogens. Many of the pathogenicity islands and virulence-encoding plasmids which harbor these genes are generally present only in pathogenic strains. In the case of the ETIR primers specifically, the tir gene is located chromosomally within several diarrheagenic pathotypes of E. coli O157:H7, some non-O157 serotypes (O26, O103, O111, and O145) and other E. coli pathotypes containing the LEE pathogenicity island (Kaper et al., 2004). Additional extraintestinal (ExPEC) pathotypes of E. coli and some diarrheagenic pathotypes lacking the LEE pathogenicity island which can persist in the environment would be undetected by the ETIR primers. However, EHEC and the enteropathogenic (EPEC) pathotypes are the main causative agents of mortality from recreational and drinking water outbreaks in North America (Kaper et al., 2004) and are those targeted by our primer selection. Importantly, these pathotypes are not detected by culture-based methods (Mull and Hill, 2009) but would be detectable by the present qPCR methods. The LOD values adopted in this evaluation were at or below the minimum infectious doses of each pathogen when tested with pure culture and artificially contaminated source water. The importance of these assays being protective to infectious doses is that it would aid in risk assessment. For example,
3385
should a positive reaction occur at a water treatment facility, operators would have adequate time to secure water resources and prevent potential health risks posed by any pathogen containing the tested genes. Presently, several days are required for positive identification of fecal coliforms, which could cause the use or release of contaminated water for human activities. The current LOD are in agreement with previous studies where SYBR Green has been used for qPCR, including those by Wolffs et al. (2006) and Nam et al. (2005) for the invA gene, where 1 102 to 1 108 S. Typhimurium cells/mL were detected, alongside assays by Heijnen and Medema (2006) for the stx genes of E. coli O157:H7 where as few as 10 cell equivalents were detected. In the present study, because of the high CP values at which single cell amplification was observed, fluorescence could not be accurately quantified by the LightCycler instrument below 1 102e2 102 cells/mL in pure culture. A similar trend was reported by Khan et al. (2007) where a 1 cell/mL dilution of an E. coli ATCC 35128 reference strain was unquantifiable at high cycle number. Ahmed et al. (2008) also noted the occurrence of non-reproducible results for single copies of the invA gene (S. Typhimurium) and mapA genes (C. jejuni), consistent with current observations. It is not necessary for our purpose to detect at the single cell level for two main reasons. Firstly, the likelihood that a single pathogenic cell would be viable and pose a direct risk to human health in this type of environment is extremely low. Secondly, as we have attempted to introduce a concentration based procedure, filtration of large volumes can amass cell numbers to within our defined detection ranges. As previously mentioned, testing the primer sets with artificially contaminated environmental samples was used as a secondary evaluation. Reducing false positives from complex samples was crucial for this study because PCR does not differentiate between DNA originating from live or dead cells or that which exists extracellularly after death. It is also known that some virulence genes can persist outside of a cell for up to 3 months in aquatic environments (DuPray et al., 1997). Early research by Phillips (1969) and Krasna (1970) has suggested that approximately 85% of free-floating native DNA can be removed by filtration. Membrane filtration was introduced prior to the DNA purification procedure which served two purposes; (i) to enhance the direct detection of pathogens in dilute field samples, and (ii) to limit the detection of extracellular DNA from dead cells. The 7.1% reduction in DNA yield observed between filtered and non-filtered samples is believed to be attributed in part to the fact that extracellular DNA may have passed through the filter, thus contributing to the reduced final yield and subsequent post-filtration shifts in CP. This assumption may be feasible since according to the microscopy results, 5% of the initial live bacterial cells lost viability and may have contributed to extracellular DNA loss. However, we cannot rule out that some cells may have been trapped within the filter matrix (Thomas, 1988) and were thus not available for DNA extraction. A more thorough examination of the effect of filtration on cell recoveries would be beneficial, since maximum recovery is desirable to eliminate false negatives. Treating the 0.22 mm GSWP mixed cellulose ester filters chemically, mechanically and enzymatically was efficient for isolating cells and their DNA from artificially contaminated source water. Removing cells from the filter prior to PCR was
3386
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 7 8 e3 3 8 8
determined to be an effective alternative to the direct filterPCR methods tested by Bej et al. (1991) and Oyofo and Rollins (1993). For this study, smaller pore size filters (0.22 mm relative to 0.45 mm) were preferred for the capture of cells exposed to harsh environmental conditions because of their reduced size (Young, 2006). The results are supported by Hora´kova´ et al. (2008) where PCR-amplifiable DNA was retrieved following concentration, bead beating and purification. Similar recoveries to those obtained in the present study with 0.22 mm filters have been reported by Bernhardt et al. (1991), where 57e90% of seeded cells in liquid samples were recoverable by filtration and centrifugation. The 0.22 mm Durapore filters were also tested, but gave lower yields relative to similar studies where the release of 100% of filtered cells has been documented (Wolffs et al., 2006). In the present study, to define the LOQ with environmental samples following cell recovery by the enrichment-free filtration method, qPCR was performed on artificially contaminated source water. It was determined that a 10-fold decrease in sensitivity was observed with the artificially contaminated samples relative to reference conditions. Similar studies have also documented a minimum10-fold decrease in sensitivity when their methods are tested with field samples (Ibekwe and Grieve, 2003; Khan et al., 2007; Liu et al., 2008; Ram et al., 2008). It is believed that this decrease in sensitivity is unavoidable since we are unable to predict and eliminate all possible sources of inhibition from the environment. A portion of this DNA may have passed through the filter in accordance with our microscopy results, thus contributing to the post-filtration shifts in CP and sensitivity. It is important to state that while sensitivity was mildly reduced with artificially contaminated samples, the chosen targets would ensure that specificity was not sacrificed. Since membrane filtration allows for the concentration of large volume samples, a sufficient cell density could be amassed to enable detection of low targets in environmental samples, although there is also a risk of co-concentrating inhibitors. Finally, the total cycling times for each primer set were under an hour in length, while the filtrationepurification scheme added only an extra hour to the analysis. Time-wise, this is an important advancement, since samples could be prepared and analyzed by the individual qPCR protocols in a single afternoon. Further work in this field is still needed before standardized real-time water quality monitoring practices can become a reality. The primers tested in this study have been deemed suitable for the proposed application, however, further scrutiny is required to ensure robustness is achieved on a larger scale before this method could be proposed to public health officials. While the existing culture-based identification schemes appear more simplistic, the end result obtained via molecular identification has a much greater meaning. Again, not only is it more rapid, but also more specific and produces a higher level of confidence with respect to quantitative outputs.
5.
Conclusions
The present study has developed and evaluated qPCR assays suitable for direct, rapid, culture-independent detection of pathogenic cells which can be completed within 12 h of initial sampling.
Low-copy targets offered increased specificity for pathogen detection and were shown to exhibit minimal crossreactivity between enteric strains. Sensitivities equivalent to minimum infectious doses were defined for test strains in both pure culture and artificially contaminated source water allowing for a good estimate of possible health risks from persistent pathogens. False positives from extracellular DNA can be eliminated by the application of membrane filtration during environmental sample processing. This study has shown that pre-enrichment can be avoided entirely by sample concentration.
Acknowledgments This research was supported by contributions from the National Science and Engineering Research Council of Canada (NSERC) through the NSERC Strategic Grant Program and Ryerson University. Thanks are extended to Dr. Gideon Wolfaardt (Ryerson University) for providing the qPCR instrument and Dr. Marie Killeen (Ryerson University) for use of the upright microscope. Thanks are also given to Dr. D-Y. Lee (University of Guelph), Dr. Eytan Wine (University of Alberta) and Liberty Victorio-Walz (Ryerson University) for providing the reference strains used in the optimization and specificity assays.
references
Ahmed, W., Huygens, F., Goonetilleke, A., Gardner, T., 2008. Realtime PCR detection of pathogenic microorganisms in roofharvested rainwater in southeast Queensland, Australia. Applied and Environmental Microbiology 74 (17), 5490e5496. Ahmed, W., Sawant, S., Huygens, F., Goonetilleke, A., Gardner, T., 2009. Prevalence and occurrence of zoonotic bacterial pathogens in surface waters determined by quantitative PCR. Water Research 43 (19), 4918e4928. Ashida, H., Toyotome, T., Nagai, T., Sasakawa, C., 2007. Shigella chromosomal ipaH proteins are secreted via the type III secretion system and act as effectors. Molecular Microbiology 63 (3), 680e693. Apte, A., Daniel, S., 2003. PCR Primer Design. In: Dieffenbach, C.W., Dveksler, G.S. (Eds.), PCR Primer: A Laboratory Manual. Cold Spring Harbour Laboratory Press, Cold Spring Harbour, New York, pp. 61e64. Bej, A.K., Mahbubani, M.H., Dicesare, J.L., Atlas, R.M., 1991. Polymerase chain reaction-gene probe detection of microorganisms using filter-concentrated samples. Applied and Environmental Microbiology 57 (12), 3529e3534. Bernhardt, M., Pennell, D.R., Almer, L.S., Schell, R.F., 1991. Detection of bacteria in blood by centrifugation and filtration. Journal of Clinical Microbiology 29, 422e425. Bertrand, R., Roig, B., 2007. Evaluation of enrichment-free PCR based detection on the rfbE gene of Escherichia coli O157: H7eApplication to municipal wastewater. Water Research 41 (6), 1280e1286. Bonetta, S., Borrelli, E., Bonetta, S., Conio, O., Palumbo, F., Carraro, E., 2010. Development of a PCR protocol for the detection of Escherichia coli O157:H7 and Salmonella spp. in surface water. Environmental Monitoring and Assessment (Aug. 17, 2010, Ahead of Print).
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 7 8 e3 3 8 8
Dieffenbach, C.W., Lowe, T.M.J., Dvelsker, G.S., 1993. General concepts for PCR primer design. Genome Research 3, S30eS37. D’Urso, O.F., Poltronieri, P., Marsigliante, S., Storelli, C., Herna´ndez, M., Rodrı´guez-La´zaro, D., 2009. A filtration-based real-time PCR method for the quantitative detection of viable Salmonella enterica and Listeria monocytogenes in food samples. Food Microbiology 26 (3), 311e316. Dorner, S.M., Anderson, W.B., Gaulin, T., Candon, H.L., Slawson, R.L., Payment, P., Huck, P.M., 2007. Pathogen and indicator variability in a heavily impacted watershed. Journal of Water and Health 5 (2), 241e257. DuPray, E., Caprais, M.P., Derrien, A., Fach, P., 1997. Salmonella DNA persistence in natural seawaters using PCR analysis. Journal of Applied Microbiology 82 (4), 507e510. Fan, H., Wu, Q., Kou, X., 2008. Co-detection of five species of waterborne bacteria by multiplex PCR. Life Science Journal 5, 47e54. Feltman, H., Schulert, G., Khan, S., Jain, M., Peterson, L., Hauser, A.R., 2001. Prevalence of type III secretion genes in clinical and environmental isolates of Pseudomonas aeruginosa. Microbiology 147 (10), 2659e2669. Goosney, D.L., DeVinney, R., Pfeutzner, R.A., Frey, E.A., Strynadka, N.C., Finlay, B.B., 2000. Enteropathogenic E. coli translocated intimin receptor, Tir, interacts directly with a-actinin. Current Biology 10, 735e738. Griffin, D.W., Lipp, E.K., McLaughlin, M.R., Rose, J.B., 2001. Marine recreation and public health microbiology: quest for the ideal indicator. BioScience 51, 817e825. Fey, A., Eichler, S., Flavier, S., Christen, R., Ho¨fle, M., Guzman, C., 2004. Establishment of a real-time PCR-based approach for accurate quantification of bacteria RNA targets in water using Salmonella as a model organism. Applied and Environmental Microbiology 70, 3618e3623. Hacker, J., Blum-Oehler, G., Mu¨hldorfer, I., Tscha¨pe, H., 1997. Pathogenicity islands of virulent bacteria: structure, function and impact on microbial evolution. Molecular Microbiology 23 (6), 1089e1097. Haffar, M., Gilbride, K.A., 2010. The utility and application of realtime PCR and FISH in the detection of single copy gene targets in Escherichia coli and Salmonella Typhimurium. Canadian Journal of Microbiology 56 (3), 254e262. Harwood, V.J., Levine, A.D., Scott, T.M., Chivukula, V., Lukasik, J., Farrah, S.R., Rose, J.B., 2005. Validity of the indicator organism paradigm for pathogen reduction in reclaimed water and public health protection. Applied and Environmental Microbiology 71 (6), 3163e3170. Heijnen, L., Medema, G., 2006. Quantitative detection of E. coli, E. coli O157 and other shiga toxin producing E. coli in water samples using a culture method combined with real-time PCR. Journal of Water and Health 4 (4), 487e498. Hoorfar, J., Cook, N., 2003. Critical aspects of standardization of PCR. In: Sachse, K., Frey, J. (Eds.), Methods in Molecular Biology (Vol. 216): PCR Detection of Microbial Pathogens. Humana Press, New Jersey, USA, pp. 51e64. Hora´kova´, K., Mlejnkova´, H., Mlejnek, P., 2008. Evaluation of methods for isolation of DNA for polymerase chain reaction (PCR)-based identification of pathogenic bacteria from pure cultures and water samples. Water Science and Technology 58 (5), 995e999. Ibekwe, A.M., Grieve, C.M., 2003. Detection and quantification of Escherichia coli O157:H7 in environmental samples by real-time PCR. Journal of Applied Microbiology 94, 421e431. Ibekwe, A.M., Watt, P.M., Grieve, C.M., Sharma, V.K., Lyons, S.R., 2002. Multiplex fluorogenic real-time detection and quantification of Escherichia coli O157:H7 in dairy wastewater wetlands. Applied and Environmental Microbiology 68 (10), 4853e4862. Juck, D., Ingram, J., Pre´vost, M., Coallier, J., Greer, C., 1996. Nested PCR protocol for the rapid detection of Escherichia coli in potable water. Canadian Journal of Microbiology 42 (8), 862e866.
3387
Kaper, J.B., Nataro, J.P., Mobley, H.L.T., 2004. Pathogenic Escherichia coli. Nature Reviews Microbiology 2, 123e140. Khan, I.U.H., Gannon, V., Kent, R., Koning, W., Lapen, D.R., Miller, J., Neumann, N., Phillips, R., Robertson, W., Topp, E., van Bochove, E., Edge, T.A., 2007. Development of a rapid quantitative PCR assay for direct detection and quantification of culturable and non-culturable Escherichia coli from agricultural watersheds. Journal of Microbiological Methods 69 (3), 480e488. Khan, I.U.H., Loughborough, A., Edge, T.A., 2009. DNA-based realtime detection and quantification of aeromonads from fresh water beaches on Lake Ontario. Journal of Water and Health 7 (2), 312e323. Krasna, A.I., 1970. Low-angle light-scattering studies on alkaliand heat-denatured DNA. Biopolymers 9 (9), 1029e1038. Krsek, M., Wellington, E.M.H., 1999. Comparison of different methods for the isolation and purification of total community DNA from soil. Journal of Microbiological Methods 39 (1), 1e16. Lemarchand, K., Lebaron, P., 2003. Occurrence of Salmonella spp. and Cryptosporidium spp. in a French coastal watershed: relationship with fecal indicators. FEMS Microbiology Letters 218 (1), 203e209. Liu, Y., Gilchrist, A., Zhang, J., Li, X.-F., 2008. Detection of viable but nonculturable Escherichia coli O157:H7 bacteria in drinking water and river water. Applied and Environmental Microbiology 74 (5), 1502e1507. Maheux, A.F., Picard, F.J., Boissinot, M., Bissonette, L., Paradis, S., Bergeron, M.G., 2009. Analytical comparison of nine PCR primer sets designed to detect the presence of Escherichia coli/ Shigella in water samples. Water Research 43 (12), 3019e3028. Maki, R.P., Hicks, R.E., 2002. Salmonella typhimurium survival and viability is unaltered by suspended particles in freshwater. Journal of Environmental Quality 31 (5), 1702e1709. Malorny, B., Hoorfar, J., Bunge, C., Helmuth, R., 2003. Multicenter validation of the analytical accuracy of Salmonella PCR: towards an international standard. Applied and Environmental Microbiology 69 (1), 290e296. Martins, M.T., Rivera, I.G., Clark, D.L., Stewart, M.H., Wolfe, R.L., Olson, B.H., 1993. Distribution of uidA gene sequences in Escherichia coli isolates in water sources and comparison with the expression of b-glucuronidase activity in 4-methylumbelliferyl-b-ᴅ-glucuronide media. Applied and Environmental Microbiology 59 (7), 2271e2276. Momba, M.N.B., Malakate, V.K., Theron, J., 2006. Abundance of pathogenic Escherichia coli, Salmonella typhimurium and Vibrio cholerae in Nkonkobe drinking water sources. Journal of Water and Health 4 (3), 289e296. Mull, B., Hill, V.R., 2009. Recovery and detection of Escherichia coli O157:H7 in surface water using ultrafiltration and real-time PCR. Applied and Environmental Microbiology 75 (11), 3593e3597. Nam, H.-M., Srinivasan, V., Gillespie, B.E., Murinda, S.E., Oliver, S. P., 2005. Application of SYBR Green real-time PCR assay for specific detection of Salmonella spp. in dairy farm environmental samples. International Journal of Food Microbiology 102 (2), 161e171. Oyofo, B.A., Rollins, D.M., 1993. Efficacy of filter types for detecting Campylobacter jejuni and Campylobacter coli in environmental water samples by polymerase chain reaction. Applied and Environmental Microbiology 59 (12), 4090e4095. Phillips, A.P., 1969. A study on factors influencing the retention of DNA by membrane filters. Biochimica et Biophysica ActaNucleic Acids and Protein Synthesis 195 (1), 186e196. Polo, F., Figueras, M.J., Inza, I., Sala, J., Fleisher, J.M., Guarro, J., 1998. Relationship between presence of Salmonella and indicators of fecal pollution in aquatic habitats. FEMS Microbiology Letters 160 (2), 253e256.
3388
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 7 8 e3 3 8 8
Ram, S., Vajpayee, P., Shanker, R., 2008. Rapid cultureindependent quantitative detection of enterotoxigenic Escherichia coli in surface waters by real-time PCR with molecular beacon. Environmental Science and Technology 42 (12), 4577e4582. Rompre´, A., Servais, P., Baudart, J., de-Roubin, M.-R., Laurent, P., 2002. Detection and enumeration of coliforms in drinking water: current methods and emerging approaches. Journal of Microbiological Methods 49 (1), 31e54. Rusin, P.A., Rose, J.B., Haas, C.N., Gerba, C.P., 1997. Risk assessment of opportunistic bacterial pathogens in drinking water. Reviews of Environmental Contamination and Toxicology 152, 52e83. Stonnet, V., Guesdon, J.-L., 1993. Campylobacter jejuni: specific oligonucleotides and DNA probes for use in polymerase chain reaction-based diagnosis. FEMS Immunology and Medical Microbiology 7 (4), 337e344. Straub, T.M., Chandler, D.P., 2003. Towards a unified system for detecting waterborne pathogens. Journal of Microbiological Methods 53 (2), 185e197. Theron, J., Morar, D., Du Preez, M., Bro¨zel, V.S., Venter, S.N., 2001. A sensitive seminested PCR method for the detection of Shigella in spiked environmental water samples. Water Research 35 (4), 869e874. Thomas, D.S., 1988. Electropositively charged filters for the recovery of yeasts and bacteria from beverages. Journal of Applied Bacteriology 65 (1), 35e41. Thompson, D.E., Rajal, V.B., De Batz, S., Wuertz, S., 2006. Detection of Salmonella spp. in water using magnetic capture hybridization combined with PCR or real-time PCR. Journal of Water and Health 4 (1), 67e75. Trabulsi, L.R., Keller, R., Tardelli Gomes, T.A., 2002. Typical and atypical enteropathogenic Escherichia coli. Emerging Infectious Diseases 8 (5), 508e513.
Tryland, I., Fiksdal, L., 1998. Enzyme characteristics of b-ᴅ-galactosidase- and b-ᴅ-glucuronidase- positive bacteria and their interference in rapid methods for detection of waterborne coliforms and Escherichia coli. Applied and Environmental Microbiology 64 (3), 1018e1023. Walters, S.P., Gannon, V.P.J., Field, K.G., 2007. Detection of Bacteroidales fecal indicators and the zoonotic pathogens E. coli O157:H7, Salmonella and Campylobacter in river water. Environmental Science and Technology 41 (6), 1856e1862. Wilkes, G., Edge, T., Gannon, V., Jokinen, C., Lyautey, E., Medeiros, D., Neumann, N., Ruecker, N., Topp, E., Lapen, D.R., 2009. Seasonal relationships among indicator bacteria, pathogenic bacteria, Cryptosporidium oocysts, Giardia cysts, and hydrogeological indices for surface waters within an agricultural landscape. Water Research 43 (8), 2209e2223. Wilson, I.G., 1997. Inhibition and facilitation of nucleic acid amplification. Applied and Environmental Microbiology 63 (10), 3741e3751. Wolffs, P.F.G., Glencross, K., Thibaudeau, R., Griffiths, M.W., 2006. Direct quantitation and detection of Salmonellae in biological samples without enrichment, using two-step filtration and real-time PCR. Applied and Environmental Microbiology 72 (6), 3896e3900. Yang, C., Jiang, Y., Huang, K., Zhu, C., Yin, Y., 2003. Application of real-time PCR for quantitative detection of Campylobacter jejuni in poultry, milk and environmental water. FEMS Immunology and Medical Microbiology 38 (3), 265e271. Yang, C., Jiang, Y., Huang, K., Zhu, C., Yin, Y., Gong, J.H., Yu, H., 2004. A real-time PCR assay for the detection and quantitation of Campylobacter jejuni using SYBR Green I and the LightCycler. Yale Journal of Biology and Medicine 77 (5e6), 125e132. Young, K.D., 2006. The selective value of bacterial shape. Microbiology and Molecular Biology Reviews 70 (3), 660e703.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 8 9 e3 3 9 7
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Sorption and biodegradation of sulfonamide antibiotics by activated sludge: Experimental assessment using batch data obtained under aerobic conditions Sheng-Fu Yang a, Cheng-Fang Lin a,*, Angela Yu-Chen Lin a, Pui-Kwan Andy Hong b a
Graduate Institute of Environmental Engineering, National Taiwan University, No. 1, Sec. 4, Roosevelt Road, Taipei 10617, Taiwan Department of Civil and Environmental Engineering, University of Utah, 122 South Central Campus Drive, 104 CME, Salt Lake City, UT 84112, USA b
article info
abstract
Article history:
This study investigated the adsorption, desorption, and biodegradation characteristics of
Received 15 November 2010
sulfonamide antibiotics in the presence of activated sludge with and without being sub-
Received in revised form
jected to NaN3 biocide. Batch experiments were conducted and the relative contributions of
28 March 2011
adsorption and biodegradation to the observed removal of sulfonamide antibiotics were
Accepted 28 March 2011
determined. Three sulfonamide antibiotics including sulfamethoxazole (SMX), sulfadime-
Available online 5 April 2011
thoxine (SDM), and sulfamonomethoxine (SMM), which had been detected in the influent and the activated sludge of wastewater treatment plants (WWTP) in Taiwan, were selected
Keywords:
for this study. Experimental results showed that the antibiotic compounds were removed
Sorption/desorption
via sorption and biodegradation by the activated sludge, though biodegradation was
Biodegradation
inhibited in the first 12 h possibly due to competitive inhibition of xenobiotic oxidation by
Sulfonamide antibiotics
readily biodegradable substances. The affinity of sulfonamides to sterilized sludge was in
Activated sludge
the order of SDM > SMM > SMX. The sulfonamides existed predominantly as anions at the study pH of 6.8, which resulted in a low level of adsorption to the activated sludge. The adsorption/desorption isotherms were of a linear form, as well described by the Freundlich isotherm with the n value approximating unity. The linear distribution coefficients (Kd) were determined from batch equilibrium experiments with values of 28.6 1.9, 55.7 2.2, and 110.0 4.6 mL/g for SMX, SMM, and SDM, respectively. SMX, SMM, and SDM desorb reversibly from the activated sludge leaving behind on the solids 0.9%, 1.6%, and 5.2% of the original sorption dose of 100 mg/L. The sorbed antibiotics can be introduced into the environment if no further treatments were employed to remove them from the biomass. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
Antibiotics have been recently investigated as a source of emerging environmental contaminants. Researchers are increasingly concerned with prevalent exposure of pathogenic microorganisms to antibiotics that may lead to development of
multi-resistant strains and antibiotics-resistant genes in the bacteria and to diminished effectiveness of conventional antibiotics (Boxall et al., 2003; Cabello, 2006; Hernando et al., 2006; Kummerer, 2004; Schwartz et al., 2003). As an important sector in prescription pharmaceuticals, antibiotic substances are employed mainly for protection from infection and diseases
* Corresponding author. Tel.: þ886 2 2362 7427; fax: þ886 2 2392 8830. E-mail address:
[email protected] (C.-F. Lin). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.03.052
3390
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 8 9 e3 3 9 7
and as growth promoters in veterinary clinics. Among the major classes of antibiotics, sulfonamide antibiotics were the first antimicrobial drugs utilized worldwide. Sulfonamides are amphoteric that they have both functional groups that can donate and accept a proton. Transfers of protons occur through protonation and deprotonation of the groups depending on the solution pH. In general, sulfonamides possess amine group (eNH3þ) of pKa1 ¼ 2.5 and sulfonamide group (eSO2NHe) of pKa2 ¼ 5e11. Hence, a sulfonamide compound can be cationic, neutral, or anionic depending on the solution pH relative to the pKa values of the compound (Thiele-Bruhn, 2003; Sarmah et al., 2006). Sulfonamide antibiotics excreted by human or animals were found to enter wastewater treatment plants (WWTPs) through the sewage system (Ingerslev and Halling-Sørensen, 2000; Rabolle and Spliid, 2000). Antibiotics were detected in the influents to wastewater treatment plants, and they were not completely removed by the activated sludge processes (Linberg et al., 2006; Xu et al., 2007; Lin et al., 2009; Kasprzyk-Hordern et al., 2009; Loganathan et al., 2009; Zorita et al., 2009; Gros et al., 2010; Ort et al., 2010; Plosz et al., 2010a, 2010b; Zuccato et al., 2010). These compounds were found at concentrations ranging from ng/L to a few mg/L. In WWTPs, the activated sludge may act as a reservoir that interacts with the compounds through sorption and biodegradation (Clara et al., 2004; Xia et al., 2005). Antibiotics including sulfonamides, fluoroquinolone, macrolides, and trimethoprim have been found in the activated sludge and the digested sludge in most WWTPs (Golet et al., 2002; Go¨bel et al., 2005a, 2005b; Kinney et al., 2006; Yang and Lin, 2009; Nieto et al., 2010). They have been detected at levels of micrograms per kilogram in sewage sludge. To date, little has been quantified for the interactions, biodegradation and sorption, of sulfonamide antibiotics with activated sludge. An aim of this work is to investigate the fate of sulfonamides as they are subjected to the activated sludge process. Experiments were designed and carried out to assess the relative contributions of adsorption, desorption, and biodegradation in the observed removal of sulfonamide antibiotics. Three sulfonamide antibiotics were selected for this investigation that had been previously found in both the domestic wastewater and in the activated sludge (Lin et al., 2009; Yang and Lin, 2009). In this research, sorption kinetics, sorption isotherms, and effects of chemical speciation on the fate of sulfonamide antibiotics are presented. In addition, measured solidewater distribution coefficients (Kd) are compared with values predicted based on octanolewater distribution coefficient (Kow) to provide insight on the interaction of antibiotics with the biomass.
amber bottles and stored in the dark at 20 C until use. Working solutions of 1 mg/L and 0.1 mg/L were prepared by dilution of stock solutions prior to each experimental run.
2.2.
The activated sludge sample was collected from an aerobic sequence batch reactor (SBR) of a wastewater treatment plant in the food manufacturing complex of Uni-President Enterprises Corporation in Taiwan. The wastewaters were generated from processes manufacturing instant noodles, tea beverages, and dairy products. Treatment processes employed in the plant included screen, equalization, dissolved air flotation, acidification, upward-flow anaerobic sludge bed process, aerobic process (via SBR), and final clarification. The flow rate, chemical oxygen demand (COD), pH, and suspended solid (SS) of the wastewater influent were 3500 m3/d, 3200 mg/L, 5e11, and 660 mg/L, respectively. The pH in the sampled SBR was 6.8, within the range of 6.6e7.1 typically observed for several other plants located at Northern, Central, and Southern Taiwan. After sampling, the activated sludge was cultivated under aerobic conditions in 20-L batch reactors using a synthetic wastewater feed (COD ¼ 300 mg/L) comprising: C12H22O11 (sucrose), 268 mg/L; (NH4)2SO4, 134 mg/L; MnSO4.H2O, 2.68 mg/L; MgSO4.7H2O, 21.4 mg/L; FeCl3.6H2O, 0.134 mg/L; CaCl2, 3.8 mg/L; KH2PO4, 141 mg/L; and K2HPO4, 287 mg/L (Yang et al., 2003). Cultivation was carried out at 25 0.5 C, pH of 6.8e7.0, and dissolved oxygen (DO) at 3 mg/L in the reactor. Aeration was by means of a digital mass flow meter (XFM; AALBORG, USA) and a porous diffuser with agitation at 90 rpm. Preexisting sulfonamides in the sludge sample were analyzed for and not found (i.e. <detection limit of 0.5 mg/kg). The activated sludge thus prepared was used in biodegradation, adsorption, and desorption experiments.
2.3.
Materials and methods
2.1.
Chemicals and reagents
Sulfamethoxazole (SMX), sulfadimethoxine (SDM), and sulfamonomethoxine (SMM) were purchased from SigmaeAldrich (St. Louis, MO, USA). HPLC-grade methanol, formic acid (FA), and sodium azide (NaN3) were from Merck (Darmstadt, Germany). Milli-Q water (18.2 MU) was produced from a Millipore purification system (Billerica, Calif., USA). Individual stock solutions of sulfonamide antibiotics were prepared by dissolving 1 mg of each compound in 1 mL of methanol in
Biodegradation experiments
In batch biodegradation experiments, a 1.5-L glass beaker containing 1 L of activated sludge suspension was spiked with 100 mg each of the sulfonamide standards. The operating conditions were 25 0.5 C, pH of 6.8, and with DO of 3 mg/L in the reactor. The aeration and mixing (JLT6; VELT, Italy) were the same as during cultivation of the activated sludge. The synthetic wastewater was introduced daily. Aqueous samples (500 mL) were taken at designated time intervals and analyzed by liquid chromatography tandem mass spectrometry (LC/MS/MS).
2.4.
2.
Activated sludge
Sorption experiments
Batch experiments were conducted to reveal the sorption of sulfonamide antibiotics with the activated sludge. Adsorption and desorption experiments with a sterilized sludge were conducted separately for SMX, SMM, and SDM. Activated sludge samples were diluted to 2.56 g MLSS/L with DI water and dispensed into 1-L screw top Erlenmeyer flasks. Sodium azide was added (1.0 g/L) to inhibit microbial activity of the sludge and minimize the loss of sulfonamide in the solution due to biodegradation. The mixture with the added biocide was agitated by a magnetic stirrer (SP135935, BarnsteadThermolyne, USA) at 90 rpm for sorption kinetics and
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 8 9 e3 3 9 7
isotherm measurements. All adsorption and desorption studies were conducted in triplicate at each concentration level at 25 C in a programmable temperature chamber (GTH225, Giant force, Taiwan). Sulfonamide solutions of varied concentration were prepared for experiments, e.g., 100 mg/L for adsorption experiments, and 10 mg/L, 50 mg/L, 100 mg/L, 200 mg/L, 300 mg/L, and 500 mg/L for isotherm determinations. The needed equilibration time for the sterilized activated sludge and sulfonamides was determined by observed kinetics in the adsorption experiment, in which aqueous samples (500 mL) were taken over 48 h of the experiment and analyzed by LC/MS/MS. The equilibration of 12 h was used in subsequent isotherm experiments. Two blank experiments, one with the reactor system containing sulfonamides without sludge and another containing sludge without sulfonamides, were carried out to ensure that no sorption of sulfonamides occurred on the flask surface and that no contamination of sulfonamides was introduced by the sampled sludge.
2.5.
analysis. The analytical methods adopted for chromatographic separation of analytes and mass spectrometric measurements of the aqueous samples were as previously reported (Lin and Tsai, 2009). An Agilent 1200 module (Agilent, Palo Alto, CA, USA) equipped with a 150 4.6 mm ZOBRAX Eclipse XDB-C18 column (5 mm, Agilent, Palo Alto, CA, USA) was employed to separate the analytes. A binary gradient with a flow rate of 1.0 mL/min was used. Mobile phase A contained 0.1% formic acid (v/v) in water. Mobile phase B contained 0.1% formic acid (v/v) in methanol. The sample injection volume was 25 mL and the autosampler was operated at room temperature. The mass spectrometric measurements were taken on a triple quadrupole mass spectrometer (Sciex API 4000; Applied Biosystems, Foster City, CA, USA) equipped with electrospray ionization (ESI). Detection was performed in positive mode. The curtain gas, nebulizer gas, turbo gas at 10, 50, 60 L/h, respectively, and ion spray voltage at 5.0 kV, heated capillary temperature at 400 C, and collisionally activated dissociation of 7 were used.
Desorption experiments 2.7.
After adsorption equilibrium, the sludge suspension was centrifuged at 3200 rpm for 15 min. The supernatant was removed and the residual sludge was re-suspended by addition of 1 L of DI water to initiate the desorption experiment. Test conditions of the desorption experiments were identical to those in the adsorption experiments.
2.6.
3391
Analytical methods
Aqueous samples from the sorption and biodegradation experiments were immediately filtered through a polyrinylidene difluoride (PVDF) syringe filter of 0.22 mm in pore size (Millipore, Billerica, Calif., USA) and stored at 20 C until
Sorption and desorption isotherm
Three isotherms including the Freundlich, linear, and Langmuir isotherms were applied to describe the sulfonamide adsorption/desorption equilibrium. During sorption experiments, sulfonamide concentrations in the aqueous phase were monitored and were used to determine their partition onto the sludge, q (mg of compound sorbed/g of sludge), according to: q¼
ðC0 Ce ÞV C0 Ce ¼ X XV
Where C0 is the initial sulfonamide concentration (mg/L), Ce the residual sulfonamide concentration (mg/L) in the aqueous
Fig. 1 e Removal of antibiotic compounds (SMX, (A); SMM, (B); SDM, (C)) from aqueous solution by activated sludge via sorption and biodegradation (circle, C) and by NaN3-treated activated sludge via sorption only (square, -), with low-lying curves (triangle, :) indicating the effect of biocide, i.e., the presence of biodegradation without the biocide. Conditions: initial sulfonamide concentration at 100 mg/L, sludge concentration at 2.56 g MLSS/L, 25 C, pH 6.8.
3392
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 8 9 e3 3 9 7
Table 1 e Molecular structure, CAS registry number, and physico-chemical properties of sulfonamide antibiotics. Name
Abbreviation
Sulfamethoxazole
Molecular structure
Log Kowa
pKaa
610 (37 C)
0.89
1.9/5.7
280.3
4030 (25 C)
0.70
2.0/6.0
310.3
343
1.63
2.1/6.3
CAS number
Formula
M.W. (g/mol)
SMX
723-46-6
C10H11N3O3S
253.3
Sulfamonomethoxine
SMM
1220-83-3
C11H12N4O3S
Sulfadimethoxine
SDM
122-11-2
C12H14N4O4S
Solubilitya (mg/L)
a Solubility, Log Kow, and pKa are from references: (Kan and Petz, 2000; Qiang and Adams, 2004; Sakurai and Ishimitsu, 1980; http://www.syrres. com/what-we-do/databaseforms.aspx?id¼386; http://logkow.cisti.nrc.ca/logkow/search.html).
phase at a specific moment, V the suspension volume (L), and X is the concentration (mg/L) of the sludge. The Freundlich model describes the relationship of sorption density of compounds on solid surface and the equilibrium concentration in liquid phase empirically (Aboul-Kassim and Simoneit, 2001): 1=n qeq ¼ Kf Ceq
Where qeq (mg/g) is the amount of compounds adsorbed onto the sludge at equilibrium, Kf (mg1nmLn/g) the Freundlich adsorption coefficient, Ceq (mg/L) the equilibrium concentration in the liquid phase, and 1/n the measure of nonlinearity. The Freundlich isotherm can be linearized in the logarithmic form: 1 log qeq ¼ log Kf þ log Ceq n The experimental data were fitted to the Freundlich isotherm and the coefficients (Kf and 1/n) were determined. A linear isotherm results when the constant 1/n of the Freundlich model approximates unity. The linear isotherm describes well the partition of a compound at low mass loading or when there is no specific bonding between the adsorbate and the adsorbent (Aboul-Kassim and Simoneit, 2001): qeq ¼ Kd Ceq Where Kd (mL/g) is the distribution coefficient, and qeq and Ceq are as previously defined. Thus, Kd also defines the ratio of concentration in the aqueous phase to that in the solid phase. The parameter KOM (mL/gVSS) describes the sorption potential of a compound toward organic matter (OM); it reflects the compound’s distribution toward the sludge solid phase and is related to Kd by: KOM ¼
MLSS Kd VSS
Where MLSS is the mixed liquor suspended solids concentration (mg/L), VSS the volatile suspended solids concentration (mg/L), and Kd the linear distribution coefficient (mL/g) as previously defined. The Langmuir model describes the monolayer sorption of a compound onto a surface with a finite number of identical sites without surface diffusion, as:
qeq ¼
Q b Ceq 1 þ bCeq
Where Q (mg/g) indicates the binding strength, b (L/mg) the maximum amount of the compound adsorbed per amount of sludge, and qeq and Ceq are as previously defined. The Langmuir model can be rewritten and plotted linearly as Ceq/qeq vs. Ceq:
Ceq Ceq 1 ¼ þ qeq Qb Q
2.8.
Speciation of sulfonamide in solution
The amphoteric sulfonamides with functional groups that readily undergo acidebase equilibrium processes (Sakurai and Ishimitsu, 1980):
H3NþC6H4SO2NHR 4 H2NC6H4SO2NHR 4 H2NC6H5SO2NR At pH 6.8 (>pK2) used in this study, anionic sulfonamides were the predominant form. Sulfonamide speciation is a function of solution pH relative to its pKa values. The extent of various amphoteric forms (neutral and charged species) in the solution can be determined by the following equations:
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 8 9 e3 3 9 7
Desorption efficiency (%)
100 80 60 40 Sulfamethoxazole Sulfamonomethoxine Sulfadimethoxin
20 0 0.0
0.5
1.0
1.5
2.0
Contact time (h)
Desorption efficiency (%)
100
80
60 Sulfamethoxazole Sulfamonomethoxine Sulfadimethoxin
40
20 0
4
8
12
16
20
24 48
Contact time (h) Fig. 2 e Desorption of test sulfonamides from NaN3-dosed activated sludge over time (sulfonamide initially at 100 mg/L, sludge at 2.56 g/L, 25 C, pH 6.8).
The fraction of neutralðfn Þ ¼
1 1 þ 10pHpKa; acid þ 10pKa; base pH
The fraction of anion ðf Þ ¼ fn 10pHpKa; acid The fraction of cation ðfþ Þ ¼ fn 10pKa; base pH
3.
Results and discussion
3.1.
Interaction of sulfonamides with activated sludge
Batch incubation experiments were conducted to determine the interaction of sulfonamide antibiotics with activated sludge at pH 6.8, 25 C, activated sludge concentration of 2.56 g MLSS/L, and initial sulfonamide concentration of 100 mg/L. The concentration changes of SMX, SMM, and SDM with time are presented in Fig. 1A, B, and C, respectively. The circles show changes upon contact with activated sludge without the biocide; the squares show changes upon contact with the sterilized sludge (with the biocide); whereas the triangles illustrate the net effect of biodegradation (i.e., square minus circle). For each sulfonamide, the concentration changes in the first 2 h of contact with the sludge (active and sterilized) are highlighted as magnified plots in Fig. 1, which reveal
3393
within this initial period a relatively rapid decrease to a stable concentration. For each sulfonamide, the rapid decreases (see enlarged plots of Fig. 1) in both active and sterilized sludges are very similar, which suggest the initial decreases are due to the adsorptive removal of sulfonamides from the aqueous phase by the solid phase; these decreases dampen to steady values over the next 10 h. In the ensuing 36 h (i.e. period 12e48 h), the concentrations of sulfonamides (see main plots of Fig. 1) in contact with the active sludge decrease gradually while the concentrations of sulfonamides in contact with the sterilized sludge remain unchanged. Thus, Fig. 1 appears to have shown an apparent point of transition at 12 h when the sulfonamide concentrations in the active sludge suspension begin to deviate from those of the sterilized suspension and decrease continually. These continual decreases of the antibiotics are attributed to biodegradation by the live activated sludge. Therefore, the removal of sulfonamides from aqueous solution was explained in two stages: the first stage between 0 and 12 h when adsorptive removal occurs predominantly reaching 6.5%, 11%, and 19% removal of SMX, SMM, and SDM, respectively, at the end of 12 h and the second stage between 12 and 48 h when biodegradative removal continues reaching 24.0%, 18.8%, and 29.5% removal of SMX, SMM, and SDM, respectively, at the end of 48 h. Thus, the sorption of sulfonamide antibiotics onto the activated sludge is a vital initial step for the removal of antibiotics. This step was followed by continual removal via biodegradation. The role of biodegradation became significant after 12 h, when the sulfonamides have fully established sorption equilibrium with the activated sludge. The lag phase before the onset of biodegradation of the antibiotics (2e10 h) could be due to readily biodegradable substrates in the reactor that could cause competitive inhibition on xenobiotic oxidation (Plosz et al., 2010b).
3.2. Interaction of sulfonamides with sterilized activated sludge 3.2.1.
Adsorption of sulfonamides
As shown in Fig. 1, the adsorptive removals of (SMX, 7.2%; SMM, 11%; and SDM, 19%) over the initial hours were modest compared to the removals of tetracycline (75e95%) (Kim et al., 2005) and other pharmaceutical compounds (Clara et al., 2004; Urase and Kikuta, 2005). Sulfonamides with low n-octanolewater distribution coefficients (log Kow) dissolve relatively well in water. According to their acid dissociation constants (all pK1 and pK2 < 6.8 as shown in Table 1), the predominant sulfonamide species would be the anionic form at the study pH of 6.8. Other studied compounds could exist in a neutral form at pH 6.8 that were more amenable to adsorptive removal (Sakurai and Ishimitsu, 1980; Schwarzenbach et al., 2003). Neutral sulfonamide species with log Kow values of 0.70e1.63 can more readily partition onto the activated sludge (Wang et al., 1993). Contrarily, organic compounds in their anionic forms adsorb less due to electrostatic repulsion by the negatively charged surface of the activated sludge (Mikkelsen and Keiding, 2002). The observed order of adsorptive removal, i.e., SDM (19%) > SMM (11%) > SMX (6.5%) consistently follows the order of their predominance in neutral form (i.e., the neutral fractions to total for SDM (24.0%, pK2 of 6.3) > SMM (13.7%, pK2 of
3394
8
6
6
4
4
2
2
50 100 200 300 Spike concentration (µg/L)
35 30
30
25
25
20
20
15
15
10
10
5
5
0
10
50 100 200 300 Spike concentration (µg/L)
12
8
8
4
4
500
0
10 35
16
12
500
c
b
0
0
10
16
50 100 200 300 Spike concentration (µg/L)
Proportion of desorbed compound (%)
8
0
Proportion of compound in sludge (%)
10
Proportion of compound in sludge (%)
a
Proportion of desorbed compound (%)
10
Proportion of desorbed compound (%)
Proportion of compound in sludge (%)
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 8 9 e3 3 9 7
500
(a) Sulfamethoxazole (b) Sulfamonomethoxine (c) Sulfadimethoxin After sorption After desorption Desorption proportion
0
Fig. 3 e Amounts (%) of spike sulfonamides that partitioned to NaN3-dosed activated sludge after adsorption (blank column) and after subsequent desorption (hatched column) equilibria, along reversibly desorbed amount shown (filled column). Conditions: sulfonamide initially at 10e500 mg/L, sludge at 2.56 g/L, 25 C, pH 6.8.
6.0) > SMX (9.1%, pK2 of 5.7)), which is explained by the increasing predominance of the anionic form and the increasing charge repulsion in the same order. It should be noted that a high pH employed above the pK2 of a diprotic acid (which the studied sulfonamides are) would result in increasing predominance of the anionic form. Thus, the adsorptive removal of sulfonamides is generally lower than other investigated pharmaceutical compounds (Clara et al., 2004; Kim et al., 2005; Urase and Kikuta, 2005).
have desorbed to the aqueous phase leaving behind 0.9%, 1.6%, and 5.2% (relative to the original dose of 100 mg/L), respectively, on the solid phase. Thus, the residual sulfonamides still in the sludge were 0.4 mg/g, 0.6 mg/g, and 2.0 mg/g for SMX, SMM, and SDM, respectively. These results suggest that, in the absence of biodegradation, the partition of sulfonamides to the activated sludge is reversible, as reported by other researchers for surfactant and azoprotein compounds (Guellil et al., 2001; Conrad et al., 2006).
3.2.2.
3.3.
Adsorption/desorption isotherms
3.3.1.
Adsorption/desorption equilibrium
Desorption of sulfonamides
Initial SMX, SMM, and SDM loadings on the sterilized activated sludge were 2.9 mg/g, 4.3 mg/g, and 7.8 mg/g, respectively. Desorption of the compounds over time is shown in Fig. 2. The aqueous concentrations of sulfonamides increased rapidly during the first 0.5 h; they approach steady values in the next 2 h and then held constant for the next 2 day. Fig. 3 presents the percentages of sulfonamide compounds partitioned onto the sterilized sludge after equilibrating (for 12 h) the sludge (2.56 g/L) with varying quantities of sulfonamides (10e500 mg/L) and again the percentages after re-equilibrating (24 h) the loaded sludge once more with deionized water (desorption). Taking the spike dose of 100 mg/L as an example, the percentages of SMX, SMM, and SDM that adsorbed onto the sludge at equilibrium were 7.4%, 11%, and 19%, respectively. After desorption equilibrium with deionized water, most of the SMX, SMM, and SDM were found to
To determine the time required to reach adsorption/desorption equilibrium for each sulfonamide, batch experiments were carried out to bring SMX, SMM, and SDM into contact with sterilized activated sludge. Fig. 1 shows the sorption of sulfonamides with sterilized activated sludge (square symbol) as a function of contact time. Sorption equilibrium was reached within 8 h, once reaching it the aqueous sulfonamide concentrations remained little changed for as long as the study period of 48 h. Thus, equilibration time of 12 h was adequate and used in this study for sorption of SMX, SMM, and SDM to the sludge. The study compounds showed similar desorption kinetics, achieving equilibrium in the first few hours (Fig. 2). To maintain consistency, desorption experiments were performed with a contact time of 24 h. Based on the observed time required for
3395
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 8 9 e3 3 9 7
a
Sulfamethoxazole
60
Sulfamonomethoxine Sulfadimethoxin
50 40 30 20 10 0 0.0
0.1
0.2
0.3
0.4
0.5
20
Sorbed concentration (µg/g VSS)
Sorbed concentration (µg/g VSS)
70
b
Sulfamethoxazole Sulfamonomethoxine Sulfadimethoxin
15
10
5
0 0.00
Solution concentration (µg/mL)
0.02
0.04
0.06
0.08
0.10
Solution concentration (µg/mL)
Fig. 4 e (a) Linear adsorption isotherm and (b) linear desorption isotherm of sulfonamides with NaN3-dosed activated sludge as adsorbent. Conditions: sulfonamide initially at 10e500 mg/L, sludge at 2.56 g/L, 25 C, pH 6.
sorption equilibrium (8e12 h) and a lag time of 12 h prior to biodegradation, the hydraulic retention time (4e6 h) provided by conventional activated sludge processes in typical WWTPs may not have allowed sufficient time for the antibiotics to be substantially biodegraded.
3.3.2.
Sorption isotherms
Freundlich and Langmuir adsorption isotherms were used to describe sorption equilibrium for the sulfonamides. While the Langmuir isotherm fitted poorly, the Freundlich isotherm described the data well with the n value close to 1, indicating a linear relationship between the sorption density (mg/g VSS) and the equilibrium concentration (mg/mL). The linear adsorption isotherms for SMX, SMM, and SDM with activated sludge are shown in Fig. 4(a). The fitted coefficients Kf and 1/n of the Freundlich model as well as the fitted distribution coefficients (Kd) of the linear model are presented in Table 2. Over the studied concentration range, the Freundlich isotherms for the free sulfonamides were linear with the 1/n value approximating unity. A linear isotherm is common where the adsorbate concentration is low relative to the adsorptive capacity of the solid, i.e., adsorption conditions far below saturation. However, extending the concentrations used in this study to higher values in order to observe adsorption saturation characterized by the Langmuir
isotherm is not warranted, because the concentrations would be considerably higher than those typically encountered in the natural or engineered aquatic systems. Linear distribution coefficients for the sulfonamide antibiotics were determined by applying the Freundlich isotherm with 1/n ¼ 1 to the experimental results, and the fitted Kd values (in lieu of Kf) for SMX, SMM, and SDM are presented in Table 2. The parameter Kd, indicating the affinity of sulfonamide for the activated sludge, was found to be 28.6 1.9, 55.7 2.2, and 110 4.6 mL/g for SMX, SMM, and SDM, respectively. In addition to Kd, the distribution coefficient normalized to the content of organic matter (KOM) is also presented in Table 2. Because the activated sludge comprised 75% of organic matter, the KOM values were expected to be greater than the Kd values. Ternes et al. (2004) determined that the Kd values of selected pharmaceuticals in their study to be between <1 and 500 mL/g. Field experiments were conducted by Go¨bel et al. (2005b) to investigate the sorption behavior of sulfonamides, such as sulfapyridine and SMX, in the presence of activated sludge. The Kd values were determined to be between 114 and 295 mL/g for sulfonamides, which were higher than the Kd values obtained in the present work. This might be due to other processes in the field study, such as biotic or abiotic degradation, resulting in reduced concentrations of sulfonamides in the field experiments (Sarmah et al., 2006).
Table 2 e Parameters of adsorption and desorption isotherms (mean ± SD) for sulfonamide antibiotics with NaN3-dosed sludge as adsorbent. Compounds Sorption Sulfamethoxazole Sulfamonomethoxine Sulfadimethoxine Desorption Sulfamethoxazole Sulfamonomethoxine Sulfadimethoxine
Freundlich parameters 1n
Kf (mg
n
mL /g)
35.2 1.2 79.2 1.1 133.6 1.1 Kf (mg1n mLn/g) 77.4 1.3 106.5 1.2 190.9 1.0
Linear parameters
1/n
R
Kd (mL/g)
Kom (mL/g VSS)
R2
0.93 0.09 1.04 0.01 0.94 0.06
0.98 0.98 0.99
28.6 1.9 55.7 2.2 110.0 4.6
38.1 2.6 74.2 2.9 146.6 6.2
0.99 0.99 0.99
1/n 1.05 0.06 1.01 0.03 0.94 0.01
2
R2 0.98 0.97 0.95
Kd (mL/g) 46.7 1.0 71.7 3.6 171.1 3.8
Kom (mL/g VSS) 62.3 1.4 95.6 4.7 228.1 5.1
R2 0.99 0.99 0.98
3396
3.3.3.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 8 9 e3 3 9 7
Desorption isotherms
Desorption isotherms were determined similarly according to the procedure used for adsorption isotherms. The desorption isotherms are presented in Fig. 4(b) with parameters summarized in Table 2. The desorption data agree well with the Freundlich isotherm with 1/n approximating unity, indicating linear isotherms. The average 1/n values for desorption are 1.05 0.06, 1.01 0.03, and 0.94 0.01 for SMX, SMM, and SDM, respectively. The Kd values of all sulfonamides in adsorption are lower than those in desorption, and are listed in Table 2.
3.4. Comparison of observed Kd with Kd, from Kow
predicted
derived
Xia et al. (2005) suggested that for pharmaceutical compounds the Kd values could be predicted based on their Kow values according to the equation: log Kd, predicted ¼ 0.58 log KOW þ 1.14, which was developed by Dobbs et al. (1989) for chlorinated organic compounds. The Kd, predicted values for SMX, SMM, and SDM were 45.3, 35.2, and 122, respectively, in comparison to experimentally determined Kd of 28.6, 55.7, and 110, respectively. Apparently, the Kd, predicted values and Kd values are reasonably close. That adsorption constants, Kd, can be predicted with KOW values is helpful for assessing the role of adsorption of sulfonamides in conventional biological wastewater treatment processes.
4.
Conclusions
This study has presented a quantitative assessment of the sorption of sulfonamide antibiotics to activated sludge. Adsorption of sulfonamide antibiotics to activated sludge occurs initially that accounts for the early removal of the antibiotic compounds from the water column. The compounds adsorb onto the activated sludge relatively quickly in the first 2 h. After the first 12 h that allows adsorption equilibrium as well as presumed acclimation of the microbes to the antibiotics, biodegradation of the antibiotic compounds occurs in the ensuing study period of 36 h (i.e., 12e48 h) resulting in at the end residual concentrations of SMX, SMM, and SDM at 76%, 81%, and 70%, respectively. The initial 12-h period prior to the onset of biodegradation could be due to readily biodegradable substrates in the reactor that caused competitive inhibition on xenobiotic oxidation. At the experimental pH 6.8, the sorption affinity of sulfonamide antibiotics to activated sludge follows the order of SDM > SMM > SMX, which is consistent with the order of abundance in neutral sulfonamide species that adsorb more readily to the sludge. The adsorption/desorption isotherms were well described by the Freundlich model in the linear regime. The Kd values determined from batch equilibrium experiments are in good agreement with model values predicted based on Kow of the sulfonamide compounds. The Kd values were 28.6 1.9, 55.7 2.2, and 110.0 4.6 mL/g for SMX, SMM, and SDM, respectively. Desorption results indicate that SMX, SMM, and SDM desorb extensively to the aqueous phase leaving behind 0.9%,
1.6%, and 5.2% (of the original dose of 100 mg/L), respectively, on the solid phase. The results affirm that sorption of sulfonamides to the activated sludge is reversible and rapid relative to their biodegradation. Indeed, the contact time required for the activated sludge to degrade sulfonamide antibiotics is longer than the hydraulic retention time of 4e6 h provided by conventional activated sludge processes in domestic wastewater treatment plants. The reversibility in the sorption of the studied sulfonamides implies that they can be released from the activated sludge upon release to the natural aquatic environment. Residual sulfonamides pose a potential risk for the environment if no suitable processes were to eliminate them from the sludge.
references
Aboul-Kassim, T.A.T., Simoneit, B.R.T., 2001. Pollutant-solid Phase Interactions: Mechanism, Chemistry and Modeling. Springer-Verlag, Berlin, Heidelberg, pp. 129e132. Boxall, A.B.A., Kolpin, D.W., Halling-Sorensen, B., Tolls, J., 2003. Peer reviewed: are veterinary medicines causing environmental risks? Environ. Sci. Technol. 37 (15), 286Ae294A. Cabello, F.C., 2006. Heavy use of prophylactic antibiotics in aquaculture: a growing problem for human and animal health and for the environment. Environ. Microbiol. 8 (7), 1137e1144. Clara, M., Strenn, B., Saracevic, E., Kreuzinger, N., 2004. Adsorption of bisphenol-A, 17b-estradiole and 17a-ethinylestradiole to sewage sludge. Chemosphere 56 (9), 843e851. Conrad, A., Cadoret, A., Corteel, P., Leroy, P., Block, J.-C., 2006. Adsorption/desorption of linear alkylbenzenesulfonate (LAS) and azoproteins by/from activated sludge flocs. Chemosphere 62 (1), 53e60. Dobbs, R.A., Wang, L., Govind, R., 1989. Sorption of toxic organic compounds on wastewater solids: correlation with fundamental properties. Environ. Sci. Technol. 23 (9), 1092e1097. Go¨bel, A., Thomsen, A., McArdell, C.S., Alder, A.C., Giger, W., Theiss, N., Lo¨ffler, D., Ternes, T.A., 2005a. Extraction and determination of sulfonamides, macrolides, and trimethoprim in sewage sludge. J. Chromatogr. A 1085 (2), 179e189. Go¨bel, A., Thomsen, A., McArdell, C.S., Joss, A., Giger, W., 2005b. Occurrence and sorption behavior of sulfonamides, macrolides, and trimethoprim in activated sludge treatment. Environ. Sci. Technol. 39 (11), 3981e3989. Golet, E.M., Strehler, A., Alder, A.C., Giger, W., 2002. Determination of fluoroquinolone antibacterial agents in sewage sludge and sludge-treated soil using accelerated solvent extraction followed by solid-phase extraction. Anal. Chem. 74 (21), 5455e5462. Gros, M., Petrovic, M., Ginebreda, A., Barcelo, D., 2010. Removal of pharmaceuticals during wastewater treatment and environmental risk assessment using hazard indexes. Environ. Int. 36 (1), 15e26. Guellil, A., Thomas, F., Block, F.-C., Bersillon, J.-L., Ginestet, P., 2001. Transfer of organic matter between wastewater and activated sludge flocs. Water Res. 35 (1), 143e150. Hernando, M.D., Mezcua, M., Fernandez-Alba, A.R., Barcel, D., 2006. Environmental risk assessment of pharmaceutical residues in wastewater effluents, surface waters and sediments. Talanta 69 (2), 334e342.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 8 9 e3 3 9 7
Ingerslev, F., Halling-Sørensen, B., 2000. Biodegradability properties of sulfonamides in activated sludge. Environ. Toxicol. Chem. 19 (10), 2467e2473. Kan, C.A., Petz, M., 2000. Residues of veterinary drugs in eggs and their distribution between yolk and white. J. Agric. Food Chem. 48 (12), 6397e6403. Kasprzyk-Hordern, B., Dinsdale, R.M., Guwy, A.J., 2009. The removal of pharmaceuticals, personal care products, endocrine disruptors and illicit drugs during wastewater treatment and its impact on the quality of receiving waters. Water Res. 43 (2), 363e380. Kim, S., Eichhorn, P., Jensen, J.N., Weber, A.S., Aga, D.S., 2005. Removal of antibiotics in wastewater: effect of hydraulic and solid retention times on the fate of tetracycline in the activated sludge process. Environ. Sci. Technol. 39 (15), 5816e5823. Kinney, C.A., Furlong, E.T., Zaugg, S.D., Burkhardt, M.R., Werner, S.L., Cahill, J.D., Jorgensen, G.R., 2006. Survey of organic wastewater contaminants in biosolids destined for land application. Environ. Sci. Technol. 40 (23), 7207e7215. Kummerer, K., 2004. Resistance in the environment. J. Antimicrob. Chemother. 54 (2), 311e320. Lin, A.Y.C., Tsai, Y.T., 2009. Occurrence of pharmaceuticals in Taiwan’s surface waters: impact of waste streams from hospitals and pharmaceutical production facilities. Sci. Total. Environ. 407 (12), 3793e3802. Lin, A.Y.C., Yu, T.H., Lateef, S.K., 2009. Removal of pharmaceuticals in secondary wastewater treatment processes in Taiwan. J. Hazard. Mater. 167 (1e3), 1163e1169. Linberg, R.H., Olofsson, U., Rendahl, P., Johansson, M.I., Tysklind, M., Andersson, B.A.V., 2006. Behavior of fluoroquinolones and trimethoprim during mechanical, chemical, and active sludge treatment of sewage water and digestion of sludge. Environ. Sci. Technol. 40 (3), 1042e1048. Loganathan, B., Philips, M., Mowery, H., Jones-Lepp, T.L., 2009. Contamination profiles and mass loadings of macrolide antibiotics and illicit drugs from a small urban wastewater treatment plant. Chemosphere 40 (1), 70e77. Mikkelsen, L.H., Keiding, K., 2002. Physico-chemical characteristics of full scale sewage sludges with implications to dewatering. Water Res. 36 (10), 2451e2462. Nieto, A., Borrull, F., Pocurull, E., Marce, R.M., 2010. Occurrence of pharmaceuticals and hormones in sewage sludge. Environ. Toxicol. Chem. 29 (7), 1484e1489. Ort, C., Lawrence, M.G., Rieckermann, J., Joss, A., 2010. 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. Plosz, B.G., Leknes, H., Liltved, H., Thomas, K.V., 2010a. Diurnal variations in the occurrence and the fate of hormones and antibiotics in activated sludge wastewater treatment in Oslo, Norway. Sci. Total. Environ. 408 (8), 1915e1924. Plosz, B.G., Leknes, H., Thomas, K.V., 2010b. Impacts of competitive inhibition, parent compound formation and
3397
partitioning behavior on the removal of antibiotics in municipal wastewater treatment. Environ. Sci. Technol. 44 (2), 734e742. Qiang, Z., Adams, C., 2004. Potentiometric determination of acid dissociation constants (pKa) for human and veterinary antibiotics. Water Res. 38 (12), 2874e2890. Rabolle, M., Spliid, N.H., 2000. Sorption and mobility of metronidazole, olaquindox, oxytetracycline and tylosin in soil. Chemosphere 40 (7), 715e722. Sakurai, H., Ishimitsu, T., 1980. Microionization constants of sulphonamides. Talanta 27 (3), 293e298. Sarmah, A.K., Meyer, M.T., Boxall, A.B.A., 2006. A global perspective on the use, sales exposure pathways, occurrence, fate and effects of veterinary antibiotics (VAs) in the environment. Chemosphere 65 (5), 725e759. Schwartz, T., Kohnen, W., Jansen, B., Obst, U., 2003. Detection of antibiotic-resistant bacteria and their resistance genes in wastewater, surface water, and drinking water biofilms. FEMS Microbiol. Ecol. 43 (3), 323e335. Schwarzenbach, R.P., Gschwend, P.M., Imboden, D.M., 2003. Environmental Organic Chemistry. John Wiley & Sons Publishers, Hoboken, NJ, pp. 254e256. Ternes, T.A., Herrmann, N., Bonerz, M., Knacker, T., Siegrist, H., Joss, A., 2004. A rapid method to measure the solid-water distribution coefficient (Kd) for pharmaceuticals and musk fragrances in sewage sludge. Water Res. 38 (19), 4075e4084. Thiele-Bruhn, S., 2003. Pharmaceutical antibiotic compounds in soils e a review. J. Plant Nutr. Soil Sci. 166 (2), 145e167. Urase, T., Kikuta, T., 2005. Sorption and biodegradation of 17b-estradiol by acclimated activated sludge under anaerobic conditions. Water Res. 39 (7), 1289e1300. Wang, L., Govind, R., Dobbs, R.A., 1993. Sorption of toxic organic compounds on wastewater solids: mechanism and modeling. Environ. Sci. Technol. 27 (1), 152e158. Xia, K., Bhandari, A., Das, K., Pillar, G., 2005. Occurrence and fate of pharmaceuticals and personal care products (PPCPs) in biosolids. J. Environ. Qual. 34 (1), 91e104. Xu, W., Zhang, G., Li, X., Zou, S., Li, P., Hu, Z., Li, J., 2007. Occurrence and elimination of antibiotics at four sewage treatment plants in the pearl river delta (PRD), South China. Water Res. 41 (19), 4526e4534. Yang, S.F., Lin, C.F., 2009. Distribution of antibiotics in municipal sewage sludge. In: The 3rd IWA-ASPIRE Conference, Oct. 18e22, Taipei, Taiwan. Yang, P.Y., Su, R., Kim, S.J., 2003. EMMC process for combined removal of organics, nitrogen and an odor producing substance. J. Environ. Manage. 69 (4), 381e389. Zorita, S., Martensson, L., Marthiasson, L., 2009. Occurrence and removal of pharmaceuticals in a municipal sewage treatment system in the south of Sweden. Sci. Total. Environ. 407 (8), 2760e2770. Zuccato, E., Castiglioni, S., Bagnati, R., Melis, M., Fanelli, R., 2010. Source, occurrence and fate of antibiotics in the Italian aquatic environment. J. Hazard. Mater. 179 (1e3), 1042e1048.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 9 8 e3 4 0 6
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Effects of selected pharmaceutically active compounds on treatment performance in sequencing batch reactors mimicking wastewater treatment plants operations Shuyi Wang, Claudia K. Gunsch* Department of Civil and Environmental Engineering, Duke University, Box 90287, Durham, NC 27708, USA
article info
abstract
Article history:
The impact of four pharmaceutically active compounds (PhACs) introduced both individ-
Received 4 January 2011
ually and in mixtures was ascertained on the performance of laboratory-scale wastewater
Received in revised form
treatment sequencing batch reactors (SBRs). When introduced individually at concentra-
22 March 2011
tions of 0.1, 1 and 10 mM, no significant differences were observed with respect to chemical
Accepted 30 March 2011
oxygen demand (COD) and ammonia removal. Microbial community analyses reveal that
Available online 6 April 2011
although similarity index values generally decreased over time with an increase in PhAC concentrations as compared to the controls, no major microbial community shifts were
Keywords:
observed for total bacteria and ammonia-oxidizing bacteria (AOB) communities. However,
Pharmaceutically active compounds
when some PhACs were introduced in mixtures, they were found to both inhibit nitrifi-
(PhACs)
cation and alter AOB community structure. Ammonia removal decreased by up to 45% in
Nitrification
the presence of 0.25 mM gemfibrozil and 0.75 mM naproxen. PhAC mixtures did not however
Wastewater treatment
affect COD removal performance suggesting that heterotrophic bacteria are more robust to
Sequencing batch reactors
PhACs than AOB. These results highlight that the joint action of PhACs in mixtures may
DGGE
have significantly different effects on nitrification than the individual PhACs. This phenomenon should be further investigated with a wider range of PhACs so that toxicity effects can more accurately be predicted. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
Every year, large quantities of human pharmaceutical products are consumed worldwide. Many drugs and their metabolites, referred to as pharmaceutically active compounds (PhACs), are not fully metabolized prior to household discharge resulting in their common occurrence in wastewater treatment plants (WWTPs) (Bound and Voulvoulis, 2005; Sedlak et al., 2000). PhACs have been detected in aquatic environments in the United States and even in drinking water samples (Kinney et al., 2006; Kolpin et al., 2002). The extent and magnitude of the risks posed by PhACs is not yet known
due to a lack of research data. However, there are many concerns that PhACs may threaten the physiological and reproductive processes of micro and macro aquatic organisms (Kinney et al., 2006). In most instances, WWTPs present the first treatment opportunity for removing PhACs and preventing significant environmental exposure. Because most municipal WWTPs rely heavily on the microbial component of the activated sludge process for nitrogen and organic removal, it is important to evaluate the impact of anthropogenic compounds in wastewater influent on WWTP treatment performance. The importance of microbial activity and community composition
* Corresponding author. Tel.: þ1 919 660 5208; fax: þ1 919 660 5219. E-mail address:
[email protected] (C.K. Gunsch). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.03.055
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 9 8 e3 4 0 6
has been reported in studies investigating the impact of mercury as well as other anthropogenic contaminants (e.g., nanomaterial and antibiotics) on bioreactor performance (Carucci et al., 2006; Nyberg et al., 2008; von Canstein et al., 2002). However, very few studies have focused on studying the ecological impacts of PhACs on WWTP microorganisms as well as to determine their effects on treatment performance. Thus, there is a need to determine if the presence of PhACs in wastewater has the potential of negatively impacting activated sludge microbial communities thereby potentially affecting treatment performance. One of the primary functions of WWTPs is the removal of nitrogen which is mainly present in the form of ammonia in wastewater influent. Ammonia-oxidizing bacteria (AOB) are a key group of bacteria which carry out the initial reaction in the transformation and removal of ammonia. AOB oxidize ammonia to nitrite in a two-step process: the first step is the oxidization of ammonia to hydroxylamine by the enzyme ammonia monooxygenase; the subsequent oxidation of hydroxylamine to nitrite is catalyzed by hydroxylamine oxidoreductase (Hooper et al., 1997). AOB are believed to be particularly susceptible to inhibition by certain chemical compounds at low concentrations and for this reason are ideal candidates to test the toxicity of chemicals (Hooper et al., 1997; Wood et al., 1981). In the present study, the impacts of four ubiquitous PhACs introduced either individually or in binary mixtures were investigated on the composition and function of activated sludge microbial communities. Functional impacts were measured in terms of reactor performance (i.e., total chemical oxygen demand (COD) and ammonia removal) while microbial community structure of both total bacteria and AOB were measured using denaturing gradient gel electrophoresis (DGGE). The selected PhACs consisted of ketoprofen, naproxen, carbamazepine, and gemfibrozil. These PhACs were selected because they have been detected in both aquatic environments and WWTPs (Sanderson et al., 2003; Terners, 1998). Furthermore, ketoprofen, naproxen and carbamazepine have previously been shown to inhibit microbial growth in batch reactors seeded with microorganisms originating from a wastewater treatment plant (Wang et al., 2008) as well as inhibit nitrification in the AOB Nitrosomonas europaea (Wang and Gunsch, 2011).
2.
Material and methods
2.1.
Bioreactor design and sampling
Three 2-L laboratory-scale sequencing batch reactors (SBR, Figure S1) with final liquid volumes of 1.5 L were constructed out of 2 L glass beakers (VWR, West Chester, PA). The reactors were operated with an 8 h cycle consisting of a 30 min feeding, 5 h aeration, 2 h settling and 30 min decanting. A 12 h hydraulic residence time was selected because it has been reported to be an efficient amount of time to remove pollutants in industrial wastewater (Franta and Wilderer, 1997; Ganesh et al., 2006). The sludge retention time (SRT) was maintained at 10 days to maximize the growth of slowgrowing nitrifiers (Mace and Mata-Alvarez, 2002).
3399
The bioreactor received synthetic wastewater (average COD 500 mg/L, ammonia 20 mg/L and pH 7) as influent. The exact composition is shown in Table S1. PhACs were added as stock solutions consisting of 50 mM PhAC dissolved in pure ethanol. Briefly, quantities of stock solution sufficient to yield 0.1, 1 or 10 mM final concentrations were added to empty Erlenmeyer flasks, the ethanol was then blown off using a gentle stream of nitrogen gas. Finally, synthetic wastewater was added and the medium was continuously stirred using magnetic stir bars to dissolve the PhAC. The recovery ratios for all PhACs were found to be over 90%. Although the treatment concentrations are significantly higher than those expected to be observed in a WWTP setting, they were selected in order to ascertain if these compounds would exert a toxic response on the mixed microbial community originating from a WWTP. Four groups of binary mixtures were introduced in Phase III: mixture group 1 (M1) e 0.5 mM ketoprofen and 0.5 mM naproxen; M2 e 0.25 mM ketoprofen and 0.75 mM naproxen; M3 e 0.5 mM gemfibrozil and 0.5 mM naproxen; and M4 e 0.25 mM gemfibrozil and 0.75 mM naproxen. Air, feed and effluent pumps (Cole Parmer, Masterflex L/S, Vernon Hills, IL) were automatically cycled using Intermatic TN311C timer-controllers (Grove, IL). Throughout the operation, mixed liquor suspended solids levels were maintained between 2000 and 2500 mg/L by wasting sludge to maintain the SRT. Dissolved oxygen was maintained between 5.8 and 6.5 mg/L by adjusting air flow rates and pH between 6.8 and 7.5 by adding 100 mM NaHCO3 as needed. All reactors were maintained at room temperature (18e22 C). During each experimental phase, three bioreactors were operated concurrently. At the beginning of operation, each bioreactor was inoculated with 500 mL activated sludge obtained from the South Durham WWTP (Chapel Hill, NC). This WWTP receives on a monthly average 6.29105 lbs BOD and 6.17104 lbs NH3. Removal efficiencies for both parameters average 99%. The plant is operated with an average sludge retention time of 20.75 days. A control reactor was operated without PhAC while the other two reactors received PhACs in their influent after the startup phase. The PhAC loading as well as the timeline for each reactor is shown in Table 1. All reactors were subjected to a startup phase lasting 30 days at the end of which steady state operation was verified prior to PhAC addition. Verification was performed by checking that microbial community structure as well as COD and ammonia removals were not statistically significantly different across reactors. All reactors were covered in aluminum foil to prevent photolysis. New bioreactors were prepared for each experimental phase by sterilizing and re-seeding with fresh activated sludge.
2.2.
Analytical methods
One mL samples were collected from each reactor every two days for PhAC concentration. Prior to analysis, biomass was removed by filtering each sample using a VWR 0.2 mm polypropylene filter (Westchester, PA). The recovery rates for ketoprofen, naproxen, carbamazepine and gemfibrozil were 81, 85, 99 and 95%, respectively. Ketoprofen, naproxen and carbamazepine concentrations were monitored by high pressure liquid chromatography (HPLC) as described in Wang
3400
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 9 8 e3 4 0 6
Table 1 e SBR experimental phases and treatment conditions. Phase I
II
III
Days of operation
SBR1
SBR2
SBR3 (Control)
0e30 31e60 61e90 0e30 31e60 61e90 90e120 0e30 31e60 61e90
Startup (No PhAC) 1 mM Ketoprofen 10 mM Ketoprofen Startup (No PhAC) 0.1 mM Carbamazepine 1 mM Carbamazepine 10 mM Carbamazepine Startup (No PhAC) 0.1 mM Ketoprofen 0.5 mM ketoprofen and 0.5 mM naproxen (M1) 0.5 mM Gemfibrozil and 0.5 mM naproxen (M3)
Startup (No PhAC) 1 mM Naproxen 10 mM Naproxen Startup (No PhAC) 0.1 mM Gemfibrozil 1 mM Gemfibrozil 10 mM Gemfibrozil Startup (No PhAC) 0.1 mM Naproxen 0.25 mM ketoprofen and 0.75 mM naproxen (M2)
Startup (No PhAC) No PhAC No PhAC Startup (No PhAC) No PhAC No PhAC No PhAC Startup (No PhAC) No PhAC No PhAC
0.25 mM Gemfibrozil and 0.75 mM naproxen (M4)
No PhAC
90e120
et al. (2008). Gemfibrozil was separated using a mobile phase consisting of methanol and water (80:20, v/v) at a flow rate of 1.1 mL/min with the UV detection set at 280 nm (Ulu, 2006). All measurements were performed in triplicates on three independent samples. The limit of quantification (LOQ) for ketoprofen, naproxen, carbamazepine and gemfibrozil was 0.5, 0.1 and 1 mM, respectively. To detect PhAC concentrations lower than the LOQ, solid phase extraction (SPE) was used to concentrate the PhACs according to a method developed by Pedrouzo et al. (2007). Reactor performance was monitored by measuring total COD, NH4eN, NO2eN and NO3eN following standard operating procedures using Hach reagents (Loveland, CO).
2.3.
Sorption batch tests
Sorption tests were performed by mixing activated sludge, PhACs and synthetic wastewater in 250 mL Erlenmeyer flasks. Briefly, 100 mL of activated sludge obtained from the North Durham WWTP (Durham, NC) were centrifuged at 5000 rpm for 5 min to obtain a bacterial pellet using an Eppendorf Centrifuge 5804 (Westbury, NY). The pellet was re-suspended in synthetic wastewater to a final volume of 100 mL. The final suspended solids concentration was approximately 2000 200 mg/L (Beltran et al., 2000). Inactivated sludge controls were prepared by adding sodium azide to a final concentration of 1% (v/v). An additional set of bottles with PhACs and sterile synthetic wastewater media but without any activated sludge was used as an abiotic control to determine if any hydrolysis or photolysis was occurring. All samples were spiked with PhACs stocks to a final concentration of 10 mM. Triplicates of each treatment condition were prepared. All experiments were performed at 20e22 C. All reactors were incubated in the dark on a VWR DS2-500E-1 orbital shaker table (Batavia, IL) at 150 rpm. One mL samples were collected for chemical analysis every 4 h over a 24-h period.
2.4.
Denaturing gradient gel electrophoresis (DGGE)
To determine the effect of PhAC on microbial community structure, two different bacterial subgroups were followed: total bacterial and AOB. In both cases, a nested polymerase chain reaction (PCR) approach was used and all analyses were
carried out in triplicate on three independent samples. A summary of the primers sequences and thermal cycling conditions used are shown in Table S2. For all PCR runs, a positive control (1 mL of purified DNA from strain N. europaea ATCC 19718) and a negative control (1 mL of DNase/RNase-free sterilized water) were used. All PCR amplifications were performed using a GeneAmp PCR System 9700 (Applied Biosystems, Foster City, CA). All reagents were obtained from the Eppendorf MasterTaq kit (Hamburg, Germany) and used following the manufacturer’s instructions. A D-Code Universal Mutation Detection System (Bio-Rad Laboratories, Hercules, CA) was used for DGGE analysis. A detailed protocol can be found in Wang et al. (2008). Pairwise comparisons of DGGE bands were performed using the Sorenson similarity index (SI) and calculated using Equation (1). SI ¼
2c aþb
(1)
where a and b are the number of distinct bands in two separate samples and c is the number of bands shared between those samples (Turpeinen et al., 2004). This index ranges from 0 (no common bands) to 1 (identical band patterns).
2.5.
Statistical analysis
Standard deviations were calculated and are shown in the figures. The Student’s t-test was used to assess statistical significance with a 95% confidence interval.
3.
Results
3.1.
PhAC removal
The amount of PhAC removed in the SBRs ranged from 7.6% to 91.3% (Table 2). In general, effluent PhAC concentration was
Table 2 e Steady state PhAC removal (%). Ketoprofen Naproxen Carbamazepine Gemfibrozil 0.1 mM 1 mM 10 mM
21.9 3.0 47.0 10.3 48.6 5.0
23.3 2.8 36.4 3.6 91.3 2.9
8.3 3.0 7.6 1.8 12.9 2.0
44.0 10.5 19.9 5.3 21.6 4.9
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 9 8 e3 4 0 6
found to gradually decrease during the first 10 days following a change in the influent concentration and then stabilize (data not shown). Naproxen and ketoprofen showed the highest degree of degradation with 91.3 2.9 and 48.6 5.0% removal at the highest influent concentration (10 mM) influent, respectively. Carbamazepine removal was much lower with a maximum of 12.9 2.0%. In the case of naproxen, ketoprofen and carbamazepine, removal increased with increasing influent concentrations. However, this was not the case for gemfibrozil where the highest removal (44.0 10.5%) was observed at the lowest influent concentration (0.1 mM). In order to determine how much of the removal could be attributed to sorption, short term batch experiments were performed with the same sludge used to seed the SBRs. As shown in Fig. 1, approximately 17.5 0.1% of ketoprofen was removed from the aquatic phase. In the case of naproxen, carbamazepine and gemfobrizil, approximately 58.7 4.0%, 9.0 3.7% and 12.4 6.3% was removed, respectively. Sorption data obtained in the batch reactors as well as the measured SBR PhAC aqueous concentrations were used to estimate PhAC sorption and biodegradation in the SBRs. Based on these results, a maximum biodegradation of 90.4% was found (10 mM naproxen). It should be noted however, that while these results provide some insights into the
3401
distribution, for more accurate estimates of PhAC biodegradation, sorption tests should be run directly on SBR sludge. It is likely that newly regrown sludge has different sorption characteristics as compared to aged sludge. Therefore, further sorption characterization may be needed to quantify PhAC distribution under those conditions, however, these experiments were not performed in the present study.
3.2.
Reactor performance
The presence of PhACs did not significantly affect COD removal whether they were introduced individually or in binary mixtures ( p > 0.05). In reactors receiving PhAC in their influent, the average COD removal ranged from 95 9% to 103 6% of that in the no PhAC controls (Table 3). Similarly, in the bioreactors exposed to the individual PhACs (0.1, 1 and 10 mM), ammonia removal was not significantly different between the no PhAC control and the reactors receiving PhACs in their influent ( p > 0.05, Table 3). Furthermore, when the first binary mixture M1 was introduced in the influent, no effect was observed. However, the ammonia removal was significantly lower ( p < 0.05) in the presence of PhAC mixtures M2, M3 and M4. Ammonia removal was 55 24% that of the control reactor in the presence of mixture M4 while the
Fig. 1 e Relative PhAC distribution during a) batch and b) SBR tests at a concentration of 10 mM. Error bars represent ± one standard deviation.
3402
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 9 8 e3 4 0 6
Table 3 e Comparisons of COD and NH4 removal in reactors receiving PhACs relative to the no PhAC control. Removals are shown as the ratio of COD (or NH4) removal in the presence of PhAC to the no PhAC control. Star (*) indicates significant difference relative to the no PhAC control with a 95% confidence level.
0.1 mM 1 mM 10 mM Mixtures
Group 1
COD
NH4
Group 2
COD
NH4
Ketoprofen (K) Naproxen(N) Ketoprofen Naproxen Ketoprofen Naproxen 0.5 mM Kþ 0.5 mM N (M1) 0.25 mM Kþ 0.75 mM N (M2)
1 0.02 1 0.03 0.98 0.04 0.96 0.06 0.96 0.06 0.98 0.06 1 0.02 1.02 0.02
0.98 0.14 0.83 0.20 0.95 0.12 0.95 0.12 0.99 0.08 1 0.07 0.98 0.14 0.60 0.09*
Carbamazepine Gemfibrozil (G) Carbamazepine Gemfibrozil Carbamazepine Gemfibrozil 0.5 mM Gþ 0.5 mM N (M3) 0.25 mM Gþ 0.75 mM N (M4)
0.98 0.06 0.95 0.09 1.03 0.03 1.02 0.04 1.03 0.06 1.02 0.07 1 0.00 1 0.01
0.93 0.15 1.1 0.14 1.05 0.16 1.09 0.15 1.05 0.23 1.09 0.26 0.74 0.15* 0.55 0.24*
relative ammonia removal was 60 9 and 74 15% that of the control reactor in the presence of mixtures M2 and M3, respectively.
3.3.
Total bacteria community structure
Several microbial community shifts for total bacteria (as measured by 16S rDNA) were observed when comparing the various treatments as well as the different time points during a given experimental phase (Figure S2). For instance, Band A intensified in the presence of 0.1 mM naproxen while Bands B and C faded away in the presence of 0.1 mM ketoprofen and naproxen, respectively. However, it is difficult to draw any conclusions from these data as there were no consistent trends, and similar shifts did not occur in the bioreactors treated with the higher concentrations (1 and 10 mM) of PhACs (data not shown) suggesting that those shift may not be associated with PhAC exposure. The microbial community structure analysis based on SI calculations show that the difference between the microbial communities exposed to PhACs generally decreased over time, however the relative changes were not found to be significant (data not shown). For example, the microbial fingerprints at Day 0 were essentially identical for 0.1 mM carbamazepine and gemfibrozil (SI ¼ 0.983). However, on Day 30, the SI value slightly decreased to 0.963 and 0.868 for these two conditions, respectively.
3.4.
is still mainly considered to be a largely qualitative tool (Muyzer et al., 1993). Thus, in order to further identify impacted AOB, our analysis focused on locating bands which either clearly appeared or disappeared or changed in intensity relative to the control treatment. Four bands were selected for further sequencing (Fig. 3). The bacteria were identified as b- Proteobacterium (Bands 1 and 4), Uncultured bacterium clone (Band 2) and Nitrosomonas sp. (Band 3).
4.
Discussion
4.1.
PhAC removal
PhAC removals reported herein are consistent with previously published studies on full scale WWTPs where removals on the order 50e82%, 23e78% and 16e69% have been reported for naproxen, ketoprofen and gemfibrozil, respectively (Heberer, 2002; Quintana and Reemtsma, 2004; Stumpf et al., 1999; Suarez et al., 2005; Terners, 1998). Very low removals of carbamazepine (<10%) have been previously reported which is also consistent with the results obtained in this study (Joss, 2005; Metcalfe et al., 2003; Radjenovic et al., 2007). The PhAC removals observed are likely a combination of biodegradation and sorption. Because all reactors were
AOB community structure
In general, the AOB community structure did not vary significantly between the PhAC treatments and the control within each experimental group when PhACs were introduced individually (Fig. 2). However, significant AOB community shifts were observed in the presence of M2, M3 and M4 as compared to the control reactor (Fig. 3a). The numbers of bands decreased throughout the experimental period suggesting a reduction in the overall number of AOB species. The microbial community structure analysis based on SI calculations reinforce data seen in the DGGE gels (Fig. 3b). In the presence of M1, the microbial fingerprints at Day 0 and 30 were essentially identical (SI ¼ 0.983 and 0.947, respectively). However, the SI values decreased significantly for M2, M3 and M4 where their values were 1.000, 0.923 and 0.677 on Day 0 and 0.286, 0.800 and 0.222 on Day 30, respectively. Even though DGGE has long been considered to be a powerful technique, it
Fig. 2 e DGGE results for SBR ammonia-oxidizing bacteria (AOB) communities exposed to 10 mM ketoprofen (K) and naproxen(N). Control reactors (C) are also shown. Microbial DGGE fingerprints are shown 0, 15 and 30 days after treatment change.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 9 8 e3 4 0 6
3403
Fig. 3 e a) DGGE results for SBR ammonia-oxidizing bacteria (AOB) communities exposed to PhAC mixtures (M1: 0.5 mM ketoprofen and 0.5 mM naproxen; M2: 0.25 mM ketoprofen and 0.75 mM naproxen; M3: 0.5 mM gemfibrozil and 0.5 mM naproxen; M4: 0.25 mM gemfibrozil and 0.75 mM naproxen). Control reactors (C) are also shown.b) Similarity index (SI) values for the AOB profile exposed to PhACs.
covered, photolytic degradation was likely not a major contributor to PhAC removal. In the short term batch experiments, sorption appeared to account for 8.1 1.8 to 17.9 3.4% of the total PhAC removal (Fig. 1). Because the sorption experiments were performed with the same sludge used to seed the SBRs, the sorption capacity is likely similar to that observed in the SBRs. However, because the microbial community evolved throughout the SBR operation, the biodegradation rates were likely different than those in the batch experiments and therefore no direct conclusions can be reached from the short term experiments about the relative contribution of biodegradation in the SBRs. Biodegradation has been previously reported for all four PhACs used herein (Daughton and Ternes, 1999; Kolpin et al., 2002). Carbamazepine has been suggested as one of the most recalcitrant PhAC with low removals even at very
high SRTs (Bernhard et al., 2006). It is important to note that the amount of ketoprofen, naproxen and carbamazepine removed increased with increasing concentration (Table 2). The increase in removal efficiency at higher concentration may be attributed to multiple factors. First, it is likely that microbial populations capable of PhAC degradation evolved. Second, it is possible that PhAC sorption is stronger at the higher concentration than at the lower concentration. Lin et al. (1994) reported this phenomenon with benzene and trichloroethylene. In general, this phenomenon is observed in media with high mineral content (Schwarzenbach and Westall, 1981) suggesting that biodegradation is likely the key contributor to increased removal in this study. While to date there are no published studies looking at the biodegradation pathways of the selected PhACs in pure cultures, others have reported on the biodegradation of steroidal
3404
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 9 8 e3 4 0 6
estrogens as well as antimicrobials (Halling-Sorensen, 2001; Roh et al., 2009).
4.2. Impact of PhACs on COD removal and total bacteria community structure The steady state COD removal for both the no PhAC control and PhAC reactors (Table 3) suggest that these PhACs do not significantly impact the performance of COD removal even at concentrations as high as 10 mM. Suarez et al. (2005) reported similar results in a nitrifying and denitrifying CSTR system exposed to a mixture of PhACs consisting of carbamazepine, diazepam, fluoxetine, citalopram, ibuprofen, naproxen and diclofenac at a combined concentration of 130 ppb (approximately 0.65 nM). Furthermore, even though a slight change in band pattern was observed (data not shown), this microbial community change did not result in any effects on treatment performance in terms of COD removal. These data suggest that there is functional redundancy in the activated sludge microbial community which can help overcome slight microbial shifts such as those observed herein even at PhACs concentrations as high as 10 mM. The relatively small change in SI values between treatments is likely due to the chaotic dynamics of the bacterial community. Chaotic dynamics are based on the assumption that competition between taxa is finely poised. It is generally recognized that competition for three or more growth-limiting resources may generate oscillations and chaotic fluctuations in species abundances (Huisman and Weissing, 1999). This theory has been applied in several studies to explain the difference in the bacterial community structure of replicate reactors (Curtis and Sloan, 2004; Kaewpipat and Grady, 2002; Saikaly et al., 2005). These dynamics in the bacterial community are especially likely in small-scale biological treatment plants (Curtis and Sloan, 2004) which are modeled by SBRs such as those used herein.
4.3. Impact of PhACs on ammonia removal and AOB community structure Ammonia removal was not significantly impacted at any PhAC concentration when PhACs were introduced individually, which may be due to the functional redundancy of AOB. It is possible that, similarly to what was observed with total bacteria, the selected PhACs may inhibit some AOB strains but not others which perform the same ecological function (i.e., ammonia oxidation). However, some PhACs were found to have synergistic effects on AOB. A decrease in nitrification and AOB diversity was observed for PhAC mixtures M2, M3 and M4 even though the same PhACs, when introduced individually, had no effect at higher concentrations (up to 10 mM). These data suggest that the combination of selected PhACs may have a more significant impact than the individual components themselves. This finding has important ramifications as PhACs are rarely (if ever) found individually in wastewater streams. In a previous study performed on N. europaea, Wang and Gunsch (2011) reported that ketoprofen, naproxen, carbamazepine and gemfibrozil inhibited nitrification by 25, 29, 22 and 26%, respectively, at a 10 mM loading. However, because no nitrification inhibition was observed in the present study with individual PhACs, this finding suggests that the AOB in the
SBRs may have been more resistant to PhACs at the concentrations of 1 and 10 mM than in the pure culture N. europaea. This result may be due to functional redundancy present in the SBR mixed microbial community. Nitrification inhibition in N. europaea was found to be correlated to membrane integrity and to be irreversible as it continued even after PhACs were removed (Wang and Gunsch, 2011). A possible explanation for the lack of measured nitrification inhibition in reactors receiving single PhACs is that AOB with membrane structures more resistant to PhACs than N. europaea may have been present in the SBRs. However, if the PhACs were impacting the activity of AOB with the less resistant membranes, microbial structure variations may still have been expected but were not observed. A possible reason for not observing such changes is that the primers utilized in the present study only target the b-subclass AOB (Nicolaisen and Ramsing, 2002). Thus, the data might not provide a complete picture of the actual AOB diversity found in the SBRs. Another possibility is that, because the inoculum for this study was obtained from an operating WWTP, the AOB likely had previously been exposed to PhACs, albeit at lower loadings. Thus, it is possible that the inoculum was already enriched with more resistant AOB strains. It should be noted that the present study was performed in SBRs and their microbial communities are likely different than those encountered in a traditional continuously operated WWTP. Still, because differences were observed in SBRs, these results suggest that differences would likely be observed in continuous reactors as well, although the microbial fingerprints will likely be quite different. While the synergistic inhibition mechanism at play for PhAC mixtures is still unknown, it appears to be specific to PhAC identity, relative concentration as well as the order in which they are introduced. For instance, even though ketoprofen and naproxen were introduced simultaneously both in mixtures M1 and M2, only M2 was found to inhibit nitrification. Similarly, both M3 and M4 were composed of gemfibrozil and naproxen however M4 had a much stronger effect both on AOB activity and diversity. Finally, it is possible that the introduction of mixture M4 into a stressed reactor (i.e., following the introduction of M2) contributed to the enhanced inhibition observed, however more studies are needed to determine if the order in which PhACs are introduced contributes to the degree of observed inhibition significantly. While the present study is very limited in scope since it only deals with four out of hundreds of PhACs and only a few concentration permutations, these results are significant as they suggest that joint action of PhACs may play an important role in determining nitrification inhibition potential. In a study of multicomponent mixtures of xenoestrogens and estrogenic agents. Silva et al. (2002) found that these compounds were able to act together to produce significant effects when combined at concentrations below their individual No Observable Effect Concentrations (NOECs) similarly to what was observed herein. Independent action and effect summation underestimated the responses observed experimentally. Silva et al. (2002) found that the toxicity equivalency factor approach predicted the mixture effects well. It was speculated that this may be the case because all mixture components acted through the same estrogen receptor activation pathway (Silva et al., 2002). It should be noted that the
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 9 8 e3 4 0 6
PhACs used herein are likely to affect different pathways since their characteristics are quite different. Therefore, more in depth studies should be performed to determine the exact mechanisms which are impacted. The findings of the present study could have important engineering significance because PhACs are present in mixtures in WWTPs. The results from this study suggest that the combined effect as opposed to the individual effect of each PhAC may affect treatment performance and, therefore, more data with complex influent mixtures are needed to follow up on the present study. It should be noted that wastewater treatment plant are operated daily with influents containing hundreds of different PhACs and that limited evidence has previously been shown on their effects on nitrification. Therefore, although our study shows that the correct combination of PhACs may negatively impact nitrification, caution should be exercised when extrapolating these data. Additional research should be performed to characterize PhAC mixture toxicity as well as their mode of action so that their synergistic or antagonistic effects can be more accurately predicted.
5.
Conclusions
The presence of PhACs did not significantly inhibit the performance of bioreactors or microbial community structure when PhACs were introduced individually. When introduced in mixtures, some PhAC combinations caused significant inhibition of ammonia removal as well as a decrease in AOB diversity suggesting that PhAC mixture effects may play an important role in an overall treatment’s nitrification potential.
Acknowledgment This research was funded by the Pratt School of Engineering at Duke University.
Appendix. Supplementary data Supplementary data associated with this article can be found in online version at doi:10.1016/j.watres.2011.03.055.
references
Beltran, F.J., Garcia-Araya, J.F., Alvarez, P.M., 2000. Sodium dodecylbenzenesulfonate removal from water and wastewater. 1. Kinetics of decomposition by ozonation. Industrial & Engineering Chemistry Research 39, 2214e2220. Bernhard, M., Muller, J., Knepper, T.R., 2006. Biodegradation of persistent polar pollutants in wastewater: comparison of an optimised lab-scale membrane bioreactor and activated sludge treatment. Water Research 40, 3419e3428. Bound, J.P., Voulvoulis, N., 2005. Household disposal of pharmaceuticals as a pathway for aquatic contamination in the United Kingdom. Environmental Health Perspectives 113, 1705e1711.
3405
Carucci, A., Cappai, G., Piredda, M., 2006. Biodegradability and toxicity of pharmaceuticals in biological wastewater treatment plants. Journal of Environmental Science and Health Part A: Environmental Science and Engineering 41, 1831e1842. Curtis, T.P., Sloan, W.T., 2004. Prokaryotic diversity and its limits: microbial community structure in nature and implications for microbial ecology. Current Opinion in Microbiology 7, 221e226. Daughton, C.G., Ternes, T.A., 1999. Pharmaceuticals and personal care products in the environment: agents of subtle change? Environmental Health Perspectives 107, 907e938. Franta, J.R., Wilderer, P.A., 1997. Biological treatment of papermill wastewater by sequencing batch reactor technology to reduce residual organics. Water Science and Technology 35, 129e136. Ganesh, R., Balaji, G., Ramanujam, R.A., 2006. Biodegradation of tannery wastewater using sequencing batch reactor e respirometric assessment. Bioresource Technology 97, 1815e1821. Halling-Sorensen, B., 2001. Inhibition of aerobic growth and nitrification of bacteria in sewage sludge by antibacterial agents. Archives of Environmental Contamination and Toxicology 40, 451e460. Heberer, T., 2002. Occurrence, fate, and removal of pharmaceutical residues in the aquatic environment: a review of recent research data. Toxicology Letters 131, 5e17. Hooper, A.B., Vannelli, T., Bergmann, D.J., Arciero, D.M., 1997. Enzymology of the oxidation of ammonia to nitrite by bacteria. Antonie Van Leeuwenhoek International Journal of General and Molecular Microbiology 71, 59e67. Huisman, J., Weissing, F.J., 1999. Biodiversity of plankton by species oscillations and chaos. Nature 402, 407e410. Joss, A., 2005. Removal of pharmaceuticals and fragrances in biological wastewater treatment. Water Research 39, 3139e3152. Kaewpipat, K., Grady, C.P.L., 2002. Microbial population dynamics in laboratory-scale activated sludge reactors. Water Science and Technology 46, 19e27. Kinney, C.A., Furlong, E.T., Werner, S.L., Cahill, J.D., 2006. Presence and distribution of wastewater-derived pharmaceuticals in soil irrigated with reclaimed water. Environmental Toxicology and Chemistry 25, 317e326. 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, T.F., Little, J.C., Nazaroff, W.W., 1994. Transport and Sorption of Volatile Organic-Compounds and Water-Vapor within Dry Soil Grains. Environmental Science and Technology 28, 322e330. Mace, S., Mata-Alvarez, J., 2002. Utilization of SBR technology for wastewater treatment: an overview. Industrial & Engineering Chemistry Research 41, 5539e5553. Metcalfe, C.D., Koenig, B.G., Bennie, D.T., Servos, M., Ternes, T.A., Hirsch, R., 2003. Occurrence of neutral and acidic drugs in the effluents of Canadian sewage treatment plants. Environmental Toxicology and Chemistry 22, 2872e2880. Muyzer, G., Dewaal, E.C., Uitterlinden, A.G., 1993. Profiling of complex microbial-populations by denaturing gradient gelelectrophoresis analysis of polymerase chain reactionamplified genes-coding for 16s ribosomal-RNA. Applied and Environmental Microbiology 59, 695e700. Nicolaisen, M.H., Ramsing, N.B., 2002. Denaturing gradient gel electrophoresis (DGGE) approaches to study the diversity of ammonia-oxidizing bacteria. Journal of Microbiological Methods 50, 189e203. Nyberg, L., Turco, R.F., Nies, L., 2008. Assessing the impact of nanomaterials on anaerobic microbial communities. Environmental Science & Technology 42, 1938e1943.
3406
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 3 9 8 e3 4 0 6
Pedrouzo, M., Reverte, S., Borrull, F., Pocurull, E., Marce, R.M., 2007. Pharmaceutical determination in surface and wastewaters using high-performance liquid chromatography-(electrospray)mass spectrometry. Journal of Separation Science 30, 297e303. Quintana, J.B., Reemtsma, T., 2004. Sensitive determination of acidic drugs and triclosan in surface and wastewater by ionpair reverse-phase liquid chromatography/tandem mass spectrometry. Rapid Communications in Mass Spectrometry 18, 765e774. Radjenovic, J., Petrovic, M., Barcelo, D., 2007. Analysis of pharmaceuticals in wastewater and removal using a membrane bioreactor. Analytical and Bioanalytical Chemistry 387, 1365e1377. Roh, H., Subramanya, N., Zhao, F.M., Yu, C.P., Sandt, J., Chu, K.H., 2009. Biodegradation potential of wastewater micropollutants by ammonia-oxidizing bacteria. Chemosphere 77, 1084e1089. Saikaly, P.E., Stroot, P.G., Oerther, D.B., 2005. Use of 16S rRNA gene terminal restriction fragment analysis to assess the impact of solids retention time on the bacterial diversity of activated sludge. Applied and Environmental Microbiology 71, 5814e5822. Sanderson, H., Johnson, D.J., Wilson, C.J., Brain, R.A., Solomon, K. R., 2003. Probabilistic hazard assessment of environmentally occurring pharmaceuticals toxicity to fish, daphnids and algae by ECOSAR screening. Toxicology Letters 144, 383e395. Schwarzenbach, R.P., Westall, J, 1981. Transport of nonpolar organic compounds from surface water to groundwater. Laboratory sorption studies. Environmental Science and Technology 15, 1360e1365. Sedlak, D.L., Gray, J.L., Pinkston, K.E., 2000. Understanding microcontaminants in recycled water. Environmental Science and Technology 34, 509Ae515A. Silva, E., Rajapakse, N., Kortenkamp, A., 2002. Something from "nothing"eeight weak estrogenic chemicals combined at
concentrations below NOECs produce significant mixture effects. Environmental Science and Technology 36, 1751e1756. Stumpf, M., Ternes, T.A., Wilken, R.D., Rodrigues, S.V., Baumann, W., 1999. Polar drug residues in sewage and natural waters in the state of Rio de Janeiro, Brazil. Science of the Total Environment 225, 135e141. Suarez, S., Ramill, M., Omil, F., Lema, J.M., 2005. Removal of pharmaceutically active compounds in nitrifying-denitrifying plants. Water Science and Technology 52, 9e14. Terners, T.A., 1998. Occurrence of drugs in German sewage treatment plants and rivers. Water Research 32, 3245e3260. Turpeinen, R., Kairesalo, T., Haggblom, M.M., 2004. Microbial community structure and activity in arsenic-, chromium- and copper-contaminated soils. Fems Microbiology Ecology 47, 39e50. Ulu, S.T., 2006. LC determination of gemfibrozil in tablets. Chromatographia 64, 447e451. von Canstein, H., Kelly, S., Li, Y., Wagner-Dobler, I., 2002. Species diversity improves the efficiency of mercury-reducing biofilms under changing environmental conditions. Applied and Environmental Microbiology 68, 2829e2837. Wang, S., Gunsch, C.K., 2011. Effects of Selected Pharmaceutically Active Compounds on the Ammonia Oxidizing Bacterium Nitrosomonas europaea. Chemosphere 82, 565e572. Wang, S., Holzem, R.M., Gunsch, C.K., 2008. Effects of pharmaceutically active compounds on a mixed microbial community originating from a municipal wastewater treatment plant. Environmental Science and Technology 42, 1091e1095. Wood, L.B., Hurley, B.J.E., Matthews, P.J., 1981. Some observations on the biochemistry and inhibition of nitrification. Water Research 15, 543e551.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 4 0 7 e3 4 1 6
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Degradation of antibiotics in water by non-thermal plasma treatment M. Magureanu a,*, D. Piroi a, N.B. Mandache a, V. David b, A. Medvedovici b, C. Bradu c, V.I. Parvulescu c,* a
National Institute for Lasers, Plasma and Radiation Physics, Department for Plasma Physics and Nuclear Fusion, Atomistilor Str. 409, Magurele-Bucharest 077125, Romania b University of Bucharest, Department of Analytical Chemistry, Bd. Regina Elisabeta 4-12, Bucharest 030016, Romania c University of Bucharest, Department of Organic Chemistry, Biochemistry and Catalysis, Bd. Regina Elisabeta 4-12, Bucharest 030016, Romania
article info
abstract
Article history:
The decomposition of three b-lactam antibiotics (amoxicillin, oxacillin and ampicillin) in
Received 20 December 2010
aqueous solution was investigated using a dielectric barrier discharge (DBD) in coaxial
Received in revised form
configuration. Solutions of concentration 100 mg/L were made to flow as a film over the
23 March 2011
surface of the inner electrode of the plasma reactor, so the discharge was generated at the
Accepted 30 March 2011
gaseliquid interface. The electrical discharge was operated in pulsed regime, at room
Available online 6 April 2011
temperature and atmospheric pressure, in oxygen. Amoxicillin was degraded after 10 min plasma treatment, while the other two antibiotics required about 30 min for decomposi-
Keywords:
tion. The evolution of the degradation process was continuously followed using liquid
Non-thermal plasma
chromatographyemass spectrometry (LCeMS), total organic carbon (TOC) and chemical
Dielectric barrier discharge
oxygen demand (COD) analyses. ª 2011 Elsevier Ltd. All rights reserved.
Antibiotics Water pollutants removal
1.
Introduction
Pharmaceutical compounds are an important class of water pollutants due to their large variety and high consumption over the last years, as well as due to their persistence in the environment. Their presence in water can have potential health effects on humans and may also affect aquatic organisms in an unpredictable way. The presence of low concentrations of various pharmaceutical substances in sewage treatment plant effluents and surface water has been detected in several studies, indicating their poor biodegradability in sewage treatment plants [Arslan-Alaton and Dogruel, 2004; Esplugas et al., 2007; Trovo et al., 2008, 2008]. Although the main source of pharmaceuticals release into the environment
is municipal wastewater, there are also other sources, such as inadequate treatment of manufacturing effluents or direct disposal of unused medicine [Magureanu et al., 2010; PerezEstrada et al., 2005]. A major concern about pharmaceuticals has been focused on antibiotics, which may promote resistance in natural bacterial populations [La¨ngin et al., 2009; Trovo et al., 2008]. Various advanced oxidation processes (AOPs) have been investigated with the aim of antibiotics removal from water, such as ozonation [Andreozzi et al., 2005; Arslan-Alaton and Dogruel, 2004; Benitez et al., 2009, 2009], Fenton and photoFenton processes [Arslan-Alaton and Dogruel, 2004; Ay and Kargi, 2010; Elmolla and Chaudhuri, 2009; Mavronikola et al., 2009; Rozas et al., 2010; Trovo et al., 2008], photochemical
* Corresponding authors. E-mail address:
[email protected] (M. Magureanu). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.03.057
3408
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 4 0 7 e3 4 1 6
Fig. 1 e Experimental set-up: plasma reactor and solution reservoir.
oxidation [Ay and Kargi, 2010; Benitez et al., 2009], photocatalysis [Elmolla and Chaudhuri, 2010a; Elmolla and Chaudhuri, 2010b; Klauson et al., 2010; Rizzo et al., 2009]. These processes are aimed at in-situ generation of strong oxygen-based oxidizers, especially hydroxyl radicals, which are among the strongest oxidizers and they react nonselectively with various types of pollutants [Esplugas et al., 2007; Trovo et al., 2008]. Non-thermal plasma generated in electrical discharges in liquid or at the gaseliquid interface leads also to the formation of oxidizing species: radicals (H$, O$, OH$) and molecules (H2O2, O3, etc) [Lukes et al., 2004; Lukes et al., 2005], which are effective for the removal of pollutants. The degradation of pharmaceutical compounds using non-thermal plasma has been recently studied by several authors [Gerrity et al., 2010; Krause et al., 2009; Magureanu et al., 2010]. Krause et al. (2009) used a corona discharge over water in order to remove several endocrine disrupting chemicals, carbamazepine, clofibric acid, and iopromide and obtained almost 100% conversion after 30 min plasma treatment. However, the power introduced in the plasma was very high (500 W), leading to low energy yields. Recently, Magureanu et al. (2010) reported the degradation of pentoxifylline in water by a pulsed dielectric barrier discharge and obtained over 90% removal after 60 min plasma treatment with a power of w1 W, corresponding to a decomposition yield of 16 g/kWh. Gerrity et al. (2010) evaluated a pilot-scale plasma reactor based on a pulsed corona discharge above water for the degradation of trace organic compounds such as pharmaceuticals and potential endocrine disrupting compounds (EDCs). They monitored the degradation of seven indicator compounds in tertiary-treated
Fig. 2 e Typical waveforms of the discharge voltage (upper graph) and discharge current (lower graph).
wastewater and spiked surface water and suggested that nonthermal plasma may be a viable alternative to more common AOPs due to its comparable energy requirements for contaminant degradation and its ability to operate without any additional feed chemicals. To the authors’ knowledge the removal of antibiotics from water by plasma treatment together with the study of the resulting degradation products has not been investigated up to now. In the present work the oxidation of three b-lactam antibiotics (amoxicillin, oxacillin and ampicillin) in water was done using a pulsed dielectric barrier discharge. These antibiotics belong to the penicillin class and are widely used in human and veterinary medicine to treat various bacterial infections. The degradation products resulting from the antibiotics decomposition were identified by LCeMS and their temporal evolution was followed.
2.
Experimental
2.1.
Plasma reactor and electrical circuit
The experimental set-up was described in detail in Magureanu et al. (2008); Magureanu et al. (2010). Briefly, the experiments were performed in a DBD reactor in coaxial configuration, with the solution flowing as a film on the surface of the inner electrode (Fig. 1). The discharge takes place at the interface between the gas (oxygen, flow rate 600 sccm) and the solution. Cooling of the reactor was not necessary since the average power in the experiments was
Scheme 1 e Structures of the investigated antibiotics.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 4 0 7 e3 4 1 6
low (w2 W) and gas circulation contributes eventually to cooling as well. The discharge was operated in pulsed mode. The high voltage pulses were generated by discharging a 2 nF capacitor by means of a rotating spark-gap switch, which controls the frequency of the voltage pulses. The discharge voltage was measured by means of a high voltage probe (Tektronix P6015, 1000, RP ¼ 100 MU) and the discharge current was determined from the voltage fall on a non-inductive shunt resistor (RS ¼ 3 U) connected in series with the outer electrode.
2.2.
Solution characteristics
Typical experiments used 200 mL solutions, containing 100 mg/L antibiotics (oxacillin, amoxicillin and ampicillin (Scheme 1)) in tap water. Separate experiments were done for each antibiotic. Tap water was used in order to simulate a real
3409
situation and was previously analyzed for identifying eventual unknown pollutants. The initial solution conductivity was 900 mS/cm, and the pH was 8. For comparison, in order to check the effect of inorganic carbon from tap water on the process, experiments using demineralized water have been carried out under the same compositional conditions.
2.3.
Analytical procedure
Analytical investigations were performed with an LCeMSeMS system (Agilent series 1200), using an atmospheric pressure electrospray ion source and quadrupole mass analyzer. MS conditions were as follows: positive ion monitoring was performed; N2 dry gas temperature - 350 C; flow N2 dry gas 13 L/min; N2 pressure for nebulization - 60 psi; capillary voltage - 4000 V; fragmentor set-up to 140 V; mass scan between 100 and 1000 a.m.u.
Fig. 3 e Chromatograms of the initial oxacillin solution and of the treated solution for various plasma treatment durations, up to 120 min.
3410
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 4 0 7 e3 4 1 6
Table 1 e Information on chemical species observed by LCeMS. #
Retention time (min)
1 2 3 4
4.73 5.02 5.16 6.17
MS fragments (m/z) 144; 114; 114; 114;
185; 203; 418; 123; 160; 376; 123; 160; 376; 144; 160; 243;
436 420 420 402; 424
Observations Identified as P3 Identified as P2 Identified as P1 Identified as Oxacillin (A)
Chromatographic separations were realized on a Zorbax SB C18 analytical column (5 cm length; 4.6 mm internal diameter; 1.8 mm particle diameter) at 35 C. The mobile phase consisted in two components: water (with 0.1% formic acid) and acetonitrile. The chromatographic elution was conducted in gradient mode with the following elution profile: starting from 5% acetonitrile and reaching to 100% in 9 min. Column equilibration was set-up to 1 min. The flow rate of mobile phase was 0.8 mL/min. Injection volume applied in all determinations was 1 mL. Freshly prepared stock solutions of each studied compound were used at a concentration level of 100 mg/mL. The degree of mineralization was followed by Total Organic Carbon (TOC) analysis. For each sample, the TOC was calculated by subtracting from the value of the Total Carbon (TC) the value of the Total Inorganic Carbon (TIC). TC and TIC measurements were performed using an analyzer based on carbon dioxide infrared absorption (HiPerTOC Thermo Electron). The TIC method involves the sample acidification (to convert HCO3 and CO32 to CO2) and the quantification of the released CO2. The TC was determined by UV-Persulfate oxidation method [Koprivnjak et al., 2006; Wallace et al., 2002].
Chemical oxygen demand (COD) analysis was achieved using a double beam spectrophotometer Unicam Helios a. The sample digestion was performed on the basis of the standard dichromate reflux method [Moore et al., 1949]. The concentration of ozone in the effluent gas was measured using an ozone detector (Anseros Ozomat MP).
3.
Results and discussion
3.1.
Discharge characteristics
Typical waveforms of the discharge voltage and discharge current are shown in Fig. 2. Voltage pulses with amplitudes of 17 kV and short rise time (w16 ns to 10e90%) were used. The current pulses had amplitudes of w110 A and duration of about 30 ns (FWHM). The pulse repetition rate was 50 Hz. The energy per pulse, calculated by integrating the product of discharge voltage and current over time, was 40 mJ and the average power dissipated in the discharge was 2 W.
3.2.
Antibiotics degradation
Although structurally related (Scheme 1), the three antibiotics exhibited quite different behavior during experiments, revealed by means of LCeMS technique. In spite of the complex structure of studied compounds, the chromatograms showed a few degradation products after interaction with plasma. Almost all identified products resulted from hydrolysis, decarboxylation or soft oxidation. However, some specific transformations of these compounds such as dimerization or
Fig. 4 e Degradation pathway of oxacillin under plasma treatment.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 4 0 7 e3 4 1 6
3411
proposed in order to overcome difficulties of similar MS spectra of the investigated compounds [Chong and Hu, 2008].
3.3.
Fig. 5 e Time-dependence of peak areas corresponding to oxacillin and to the main degradation compounds of oxacillin identified.
adduct formation, typically occurring in pharmaceutical formulations, reported for these compounds [Wu, 2000], have not been observed in our experiments. The identification of all species relied on MS spectra which were correlated with initial structure of investigated compounds [Medvedovici et al., 2010]. The procedure of identification b-lactam antibiotics using LC/ ESI/MS library combined with retention data has been already
Oxacillin degradation
Fig. 3 shows the chromatograms corresponding to the pure oxacillin initial solution and the oxacillin solutions treated in plasma for durations of 1, 2, 3, 4, 5, 10, 15, 20, 30, 60, 90, 120 min, respectively. The chromatograms recorded for the solutions exposed to plasma showed only four major chromatographic peaks, which corresponded to oxacillin and three degradation compounds (P1, P2 and P3). MS and retention data regarding to these peaks are summarized in Table 1. The compounds P1 and P2 correspond to a pair of diastereoisomers, which exhibit identical MS pattern, but which can be separated under RPeLC mechanism. This pair of diastereoisomers results almost instantly in aqueous medium, and does not represent a step of the degradative pathway. However, they become the initial substrates in degradation process under plasma conditions. The structures of the degradation products of oxacillin after plasma treatment, advanced according to the MS spectra acquired during the chromatographic run, are given in the scheme in Fig. 4. The quantitative profiles of oxacillin, of the diastereoisomers (P1, P2) and of the degradation product (P3), expressed as peak area measured after LCeMS determinations, are
Fig. 6 e Fragmentation route of oxacillin.
3412
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 4 0 7 e3 4 1 6
Fig. 7 e Variation of TC, TIC and TOC for oxacillin solution in tap water as a function of plasma treatment time.
depicted in Fig. 5. The degradation of P1, P2 and oxacillin occurs rapidly, with an almost complete disappearance after 30 min treatment. The energy efficiency for their decomposition was 27 g/kWh at a conversion of more than 90%. Among the reaction products, only the compound denoted by P3 showed a maximum value (resulted at 5 min exposure time). After this period the peak assigned to this component showed also a decrease with treatment time. Finally, after 120 min exposure, the LC analysis indicated no component in the analyzed waters. Correlating this with the mass scan interval used in this study (i.e. > 100 amu.), it results the remained fragments should correspond to very small entities. An example of the fragmentation route for oxacillin is given in Fig. 6; the MS spectral lines can be explained by the retro DielseAlder pathway. The fragmentation pathway provides additional arguments in this sense: the decomposition products should exhibit structures smaller than that corresponding to m/z of 114. Fig. 7 shows the evolution of TC, TIC and TOC as a function of time. After 120 min approximately 25% of the compound has been mineralized. By correlation with LCeMS analysis we can assume that the initial molecule suffered an advanced
fragmentation (entities with m/z smaller than 100) and about a quarter was removed as carbon dioxide. Interestingly, the decrease of TOC was accompanied by a decrease of TIC. Thus, about 23% of inorganic carbon was removed as carbon dioxide in the same interval of time. To check the possible interference from carbonate/dicarbonate species present in tap water, plasma assisted experiments in demineralized water have also been carried out. Data presented in Fig. 8 show no influence of these species, at least for this concentration. They also show no scavenger effect of these inorganic species on both ozone and radicals generated by ozone in water. However, they reveal another specific feature of the process. As expected, TIC increased in time as a result of the advanced oxidation of oxacillin but approximately 70% from it is kept in water. The evolution of COD shows a faster decrease with treatment time as compared to TOC. It corresponds to about 40% for oxacillin providing additional evidence of the fact that the persistent organic part exists indeed in an advanced oxidation state.
3.4.
Amoxicillin degradation
The degradation of Amoxicillin under plasma conditions followed a similar pathway as that observed for oxacillin and some of the products were identical to those recently reported from degradation under solar radiation [Klauson et al., 2010]. Table 2 summarizes the MS fragments identified by LCeMS for the four degradation products that have been detected. The identification of the structures corresponding to these fragments led to the degradation pathway depicted in Fig. 9. Fig. 10 shows the time evolution of the peak areas corresponding to amoxicillin and to the identified D1eD4 species. It is worth noticing that the degradation of the resulting diastereoisomers (D1, D2) occurred faster as compared with those resulted from oxacillin, and in consequence they are completely decomposed after 10 min plasma treatment. Except for D4, after 120 min exposure, all other species were completely removed and no other compounds with m/z higher than 100 has been detected. Like in the case of oxacillin, this accounts for an advanced degradation of this antibiotic. The energy efficiency for the degradation of amoxicillin was 105 g/kWh. Fig. 11 shows the evolution of TC, TIC, TOC and COD as a function of plasma treatment time. After 120 min the mineralization of amoxicillin in tap water was similar to that of oxacillin resulting in a proportion
Table 2 e Information on chemical species observed by LCeMS.
Fig. 8 e Variation of TC, TIC, TOC and COD for oxacillin solution in demineralized water as a function of plasma treatment time.
#
Retention time (min)
MS fragments (m/z)
Observations
1 2 3
1.53 1.81 2.27
160; 323; 340; 367; 384 160; 323; 340; 367; 384 114; 160; 208; 349; 366; 388
4 5
3.39 3.69
114; 123; 189; 227; 249 114; 123; 160; 249; 340; 362
Identified as D1 Identified as D2 Identified as Amoxicillin (A) Identified as D3 Identified as D4
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 4 0 7 e3 4 1 6
3413
Fig. 9 e Degradation pathway of amoxicillin under plasma treatment.
of 22.5%. The advanced fragmentation was also confirmed by COD values. They indicated a decrease of COD of 43% during the time of the experiment. Like in the case of oxacillin, the experiments carried out in demineralized water (data not shown) indicated that the presence of carbonate/dicarbonate species induces no scavenger effect on ozone and radicals generated by ozone and they accumulate in time. However, in this case the kinetics of the process is more evident from the time-dependence of TOC curve. After a quite stationary step in the first 30 min, the fragmentation goes faster. This correlates well with the LCeMS measurements.
3.5.
Ampicillin degradation
Like the other two investigated antibiotics, immediately after solubilization in water, the ampicillin molecule hydrolyzes by b-lactam ring opening into two diastereoisomers (R5, R6). The
Fig. 10 e Time-dependences of peak area corresponding to the main identified degradation compounds of amoxicillin.
degradation under plasma treatment led to several compounds, denoted by R1eR4 and R7, in accordance to their increasing retention time in chromatograms obtained from LCeMS investigations. Their structure is confirmed by MS spectra, whose spectral lines are mentioned in Table 3. However, three MS spectra could not be correlated to any possible structure derived from the initial molecule, i.e. ampicillin, and therefore, they were characterized only from quantitative point of view (peak area). These degradation compounds are R1, R3 and R4. The degradation pathway of ampicillin (limited to the identified compounds) under plasma treatment is shown in Fig. 12. Fig. 13 depicts the time evolution of ampicillin and of the LCeMS identified products (R1eR7). After around 30 min all these species suffered an advanced degradation and no other
Fig. 11 e Variation of TC, TIC, TOC and COD for amoxicillin solution in tap water as a function of plasma treatment time.
3414
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 4 0 7 e3 4 1 6
Table 3 e Information on chemical species observed by LCeMS. #
Retention time (min)
MS fragments (m/z)
Observations
1
2.30
114; 123; 324; 368
2 3
2.41 2.55
114; 123; 305; 322 114; 123; 324; 368
4
2.77
114; 123; 249; 349; 368
5 6 7 8
2.94 3.08 3.20 3.44
160; 160; 114; 106; 333;
Unidentified, denoted by R1 Identified as R2 Unidentified, denoted by R3 Unidentified, denoted by R4 Identified as R5 Identified as R6 Identified as R7 Identified as Ampicillin (A)
is very similar as it is demonstrated by the evolution of the CODeTOC ratios.
3.6.
324; 368 324; 368 249; 324 114; 160; 174; 192; 350
products with m/z higher than 100 were identified. The energy efficiency for ampicillin decomposition was 29 g/kWh. TC, TIC, TOC and COD measurements confirmed the above measurements (Fig. 14). The mineralization of ampicillin occurred in a 29% extent and the faster decrease of COD (54%) comparative with TOC (29%) indicated that among the investigated antibiotics this one suffers the most advanced degradation. The evolution of the ratio COD/TOC provides clear information of the process of fragmentation and, accordingly, it shows that it indeed occurs for ampicillin. In conclusion, comparing the effect of the exposure time on the degradation of the three investigated antibiotics, it appears that although they have a quite close molecular structure their stability is different and this results in different treatment times. However, the mechanism of the degradation
Ozone influence on the degradation of antibiotics
It was found that ozone is formed in significant amounts in hybrid gaseliquid electrical discharge reactors in oxygencontaining atmospheres [Lukes et al., 2004; Lukes et al., 2005; Magureanu et al., 2008, 2008]. This strong oxidizer can diffuse from the gas into the liquid and react with the pollutant molecules [Grabowski et al., 2007; Magureanu et al., 2008]. Therefore, ozone can play a very important role in the present case as well, for the oxidation of the investigated antibiotics. In order to investigate the influence of ozone generated in the plasma on the degradation of antibiotics the concentration of ozone in the effluent gas was measured in two situations: (a) in the flow containing pure water, and (b) in the flow containing the different antibiotics. The experiments were performed under the same discharge operating conditions. In case (a) the ozone concentration was constant in time and its value was 1.5 g/m3. The ozone concentration in the effluent gas was lower in case (b) as compared with case (a), since a part of the ozone generated in the plasma reacts with the antibiotics and with their degradation products. The difference between the two concentrations, showing the ozone consumed, is plotted in Fig. 15 as a function of the treatment time. They correlate relatively well with chromatographic data indicating the temporal evolution of the antibiotics and of their degradation products. In case of amoxicillin oxidation the consumption of ozone was high in the first 5 min of plasma treatment and started to decrease fast afterward, so that after 20 min the entire
Fig. 12 e Degradation pathway of ampicillin under plasma treatment.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 4 0 7 e3 4 1 6
Fig. 13 e Time-dependences of peak area corresponding to the main identified degradation compounds of ampicillin. amount of ozone generated in the plasma can be found in the effluent gas. As shown in Fig. 10, amoxicillin and the diastereoisomers resulting from its hydrolysis disappeared after 10 min treatment time and the only persistent product was D4, which apparently does not react with ozone. For the oxidation of oxacillin only w0.2 g/m3 of ozone was consumed in the first minutes and the decrease in consumed ozone was slower, lasting about 30 min. The decrease of the concentrations of oxacillin and its hydrolysis products was also slower than that of amoxicillin, after 30 min treatment over 90% conversion was obtained (Fig. 5). However, the decrease of the concentration of the degradation product P3 was much slower and did not seem to be due to ozone reactions. Ozone consumption in case of the ampicillin solution remained approximately constant, around 0.2 g/m3, for almost 30 min plasma treatment and then decreased slowly. Fig. 13 shows that ampicillin and its degradation products disappear after 30 min treatment. The consumption of ozone after that time might be due to the formation of small degradation products (not identified) which react also with ozone.
3415
Fig. 15 e Ozone consumed in oxidation reactions of antibiotics (initial concentration 100 mg/L) as a function of treatment time.
4.
Conclusions
Non-thermal plasma generated in a pulsed dielectric barrier discharge was investigated with the aim of the removal of three b-lactam antibiotics (amoxicillin, oxacillin and ampicillin) from water. The discharge was generated at the gaseliquid interface at room temperature and atmospheric pressure, in oxygen. The degradation products resulting from the antibiotics decomposition were identified by LCeMS and their temporal evolution was followed up to 120 min treatment time. It was found that immediately after solubilization in water the antibiotic molecules hydrolyze by b-lactam ring opening into two diastereoisomers. The formation of this pair of diastereoisomers does not represent a step of the degradative pathway. However, they become the initial substrates in degradation process under plasma conditions. Amoxicillin was degraded after 10 min plasma treatment, while oxacillin required about 30 min and ampicillin 20 min for decomposition. The concentrations of their degradation products decrease as well after plasma treatment and most of them are removed after longer treatment times. TC, TIC, TOC and COD measurements provided additional information about the mineralization and fragmentation of these molecules. They showed the advanced oxidation of these antibiotics (compounds with m/z smaller than 100) with releasing of 25e30% of the initial carbon content as carbon dioxide.
Acknowledgment The authors acknowledge financial support from UEFISCSU, project number ID-223.
Appendix. Supplementary material Fig. 14 e Variation of TC, TIC, TOC and COD for ampicillin solution in tap water as a function of plasma treatment time.
Supplementary data associated with this article can be found in the online version, at doi:10.1016/j.watres.2011.03.057.
3416
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 4 0 7 e3 4 1 6
references
Andreozzi, R., Canterino, M., Marotta, R., Paxeus, N., 2005. Antibiotic removal from wastewaters: the ozonation of amoxicillin. J. Hazard. Mater. 122, 243e250. Arslan-Alaton, I., Dogruel, S., 2004. Pre-treatment of penicillin formulation effluent by advanced oxidation processes. J. Hazard. Mater. B112, 105e113. Ay, F., Kargi, F., 2010. Advanced oxidation of amoxicillin by Fenton’s reagent treatment. J. Hazard. Mater. 179, 622e627. Benitez, F.J., Acero, J.L., Real, F.J., Roldan, G., 2009. Ozonation of pharmaceutical compounds: rate constants and elimination in various water matrices. Chemosphere 77, 53e59. Chong, X.M., Hu, C.Q., 2008. Compilation of an ESI-MS library of b-lactam antibiotics for rapid identification of drugs. Chromatographia 68, 759e766. Elmolla, E.S., Chaudhuri, M., 2009. Degradation of the antibiotics amoxicillin, ampicillin and cloxacillin in aqueous solution by the photo-Fenton process. J. Hazard. Mater. 172, 1476e1481. Elmolla, E.S., Chaudhuri, M., 2010a. Degradation of amoxicillin, ampicillin and cloxacillin antibiotics in aqueous solution by the UV/ZnO photocatalytic process. J. Hazard. Mater. 173, 445e449. Elmolla, E.S., Chaudhuri, M., 2010b. Photocatalytic degradation of amoxicillin, ampicillin and cloxacillin antibiotics in aqueous solution using UV/TiO2 and UV/H2O2/TiO2 photocatalysis. Desalination 252, 46e52. Esplugas, S., Bila, D.M., Krause, L.G.T., Dezotti, M., 2007. Ozonation and advanced oxidation technologies to remove endocrine disrupting chemicals (EDCs) and pharmaceuticals and personal care products (PPCPs) in water effluents. J. Hazard. Mater. 149, 631e642. Gerrity, D., Stanford, B.D., Trenholm, R.A., Snyder, S.A., 2010. An evaluation of a pilot-scale nonthermal plasma advanced oxidation process for trace organic compound degradation. Water Res. 44, 493e504. Grabowski, L.R., van Veldhuizen, E.M., Pemen, A.J.M., Rutgers, W.R., 2007. Breakdown of methylene blue and methyl orange by pulsed corona discharge. Plasma Sources Sci. Technol. 16, 226e232. Klauson, D., Babkina, J., Stepanova, K., Krichevskaya, M., Preis, S., 2010. Aqueous photocatalytic oxidation of amoxicillin. Catal. Today 151, 39e45. Koprivnjak, J.F., Perdue, E.M., Pfromm, P.H., 2006. Coupling reverse osmosis with electrodialysis to isolate natural organic matter from fresh waters. Water Res. 40, 3385e3392. Krause, H., Schweiger, B., Schuhmacher, J., Scholl, S., Steinfeld, U., 2009. Degradation of the endocrine disrupting chemicals (EDCs) carbamazepine, clofibric acid, and iopromide by corona discharge over water. Chemosphere 75, 163e168. La¨ngin, A., Alexy, R., Ko¨nig, A., Ku¨mmerer, K., 2009. Deactivation and transformation products in biodegradability testing of ß-lactams amoxicillin and piperacillin. Chemosphere 75, 347e354.
Lukes, P., Appleton, A.T., Locke, B.R., 2004. Hydrogen peroxide and ozone formation in hybrid gaseliquid electrical discharge reactors. IEEE Trans. Ind. Appl. 40, 60e67. Lukes, P., Clupek, M., Babicky, V., Janda, V., Sunka, P., 2005. Generation of ozone by pulsed corona discharge over water surface in hybrid gaseliquid electrical discharge reactor. J. Phys. D: Appl. Phys. 38, 409e416. Magureanu, M., Piroi, D., Mandache, N.B., Parvulescu, V., 2008. Decomposition of methylene blue in water using a dielectric barrier discharge: optimization of the operating parameters. J. Appl. Phys. 104 (art. no. 103306). Magureanu, M., Piroi, D., Mandache, N.B., David, V., Medvedovici, A., Parvulescu, V.I., 2010. Degradation of pharmaceutical compound pentoxifylline in water by nonthermal plasma treatment. Water Res. 44, 3445e3453. Mavronikola, C., Demetriou, M., Hapeshi, E., Partassides, D., Michael, C., Mantzavinos, D., Kassinos, D., 2009. Mineralisation of the antibiotic amoxicillin in pure and surface waters by artificial UVA-and sunlight-induced Fenton oxidation. J. Chem. Technol. Biotechnol. 84, 1211e1217. Medvedovici, A., Albu, F., David, V., 2010. Handling drawbacks of mass spectrometric detection coupled to liquid chromatography in bioanalysis. J. Liq. Chromatogr. Relat. Technol. 33, 1255e1286. Moore, W.A., Kroner, R.C., Ruchhoft, C.C., 1949. Dichromate reflux method for determination of oxygen consumed. Anal. Chem. 21, 953e957. Perez-Estrada, L.A., Maldonado, M.I., Gernjak, W., Aguera, A., Fernandez-Alba, A.R., Ballesteros, M.M., Malato, S., 2005. Decomposition of diclofenac by solar driven photocatalysis at pilot plant scale. Catal. Today 101, 219e226. Rizzo, L., Meric, S., Guida, M., Kassinos, D., Belgiorno, V., 2009. Heterogenous photocatalytic degradation kinetics and detoxification of an urban wastewater treatment plant effluent contaminated with pharmaceuticals. Water Res. 43, 4070e4078. Rozas, O., Contreras, D., Mondaca, M.A., Perez-Moya, M., Mansilla, H.D., 2010. Experimental design of Fenton and photo-Fenton reactions for the treatment of ampicillin solutions. J. Hazard. Mater. 177, 1025e1030. Trovo, A.G., Melo, S.A.S., Nogueira, R.F.P., 2008. Photodegradation of the pharmaceuticals amoxicillin, bezafibrate and paracetamol by the photo-Fenton process: application to sewage treatment plant effluent. J. Photochem. Photobiol. A: Chem. 198, 215e220. Wallace, B., Purcell, M., Furlong, J., 2002. Total organic carbon analysis as a precursor to disinfection byproducts in potable water: oxidation technique considerations. J. Environ. Monit. 4, 35e42. Wu, Y., 2000. The use of liquid chromatographyemass spectrometry for the identification of drug degradation products in pharmaceutical formulations. Biomed. Chromatogr. 14, 384e396.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 4 1 7 e3 4 2 6
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Sorption of emerging trace organic compounds onto wastewater sludge solids John Stevens-Garmon a, Jo¨rg E. Drewes a, Stuart J. Khan b, James A. McDonald b, Eric R.V. Dickenson a,* a
Advanced Water Technology Center (AQWATEC), Environmental Science and Engineering Division, Colorado School of Mines, Golden, CO 80401, USA b UNSW Water Research Centre, School of Civil and Environmental Engineering, The University of New South Wales, NSW 2052, Australia
article info
abstract
Article history:
This work examined the sorption potential to wastewater primary- and activated-sludge
Received 23 August 2010
solids for 34 emerging trace organic chemicals at environmentally relevant concentrations.
Received in revised form
These compounds represent a diverse range of physical and chemical properties, such as
14 March 2011
hydrophobicity and charge state, and a diverse range of classes, including steroidal hormones,
Accepted 30 March 2011
pharmaceutically-active compounds, personal care products, and household chemicals.
Available online 6 April 2011
Solid-water partitioning coefficients (Kd) were measured where 19 chemicals did not have previously reported values. Sludge solids were inactivated by a nonchemical lyophilization
Keywords:
and dry-heat technique, which provided similar sorption behavior for recalcitrant compounds
Wastewater treatment
as compared to fresh activated-sludge. Sorption behavior was similar between primary- and
Primary and activated sludge
activated-sludge solids from the same plant and between activated-sludge solids from two
Inactivation
nitrified processes from different wastewater treatment systems. Positively-charged phar-
Sorption
maceutically-active compounds, amitriptyline, clozapine, verapamil, risperidone, and
Steroidal hormones
hydroxyzine, had the highest sorption potential, log Kd ¼ 2.8e3.8 as compared to the neutral
Pharmaceutically-active
and negatively-charged chemicals. Sorption potentials correlated with a compound’s hydro-
compounds
phobicity, however the higher sorption potentials observed for positively-charged compounds
Personal care products
for a given log Dow indicate additional sorption mechanisms, such as electrostatic interactions,
Household chemicals
are important for these compounds. Previously published soil-based one-parameter models for predicting sorption from hydrophobicity (log Kow > 2) can be used to predict sorption for emerging nonionic compounds to wastewater sludge solids. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
A wide range of trace organic chemicals (TOrCs) regularly enters municipal wastewater treatment systems, including steroidal hormones, pharmaceutically-active compounds, personal care products, and household chemicals. Some of these are of concern due to their persistence and toxicological effects in the environment (Snyder et al., 2003; Stackelberg
et al., 2004; Carucci et al., 2006; Kim et al., 2007). Treatment plant operators, regulatory agencies, and the public concerned about discharges of such TOrCs are interested in the fate of these chemicals during wastewater treatment. TOrC removal mechanisms during activated-sludge treatment include biological transformation and degradation, volatilization, and sorption. The low Henry’s constant for most pharmaceutically-active compounds and compounds targeted
* Corresponding author. E-mail address:
[email protected] (E.R.V. Dickenson). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.03.056
3418
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 4 1 7 e3 4 2 6
in this study (H < 105, H is the dimensionless Henry gas water partitioning coefficient (Lwastewater/Lair)), indicates volatilization is negligible as a removal mechanism for these type of compounds (Schwarzenbach et al., 2003). Biotransformation is often the dominant removal process for some TOrCs, as described by Clara et al. (2005), Joss et al. (2006), Urase and Kikuta (2005), and Wick et al. (2009). Also, sorption onto solids can be a key process controlling the physical removal of many natural and xenobiotic organic chemicals in municipal wastewater treatment plants. Sorption mechanisms of organic compounds on wastewater solids consist of two processes: (1) adsorption of the organic compounds from the bulk liquid onto the surface of solids and (2) partitioning of the organic compounds between the aqueous phase and the organic matter within solids. The equilibrium partitioning of an organic chemical between the solid and aqueous phases is described by the solids-water distribution coefficient (Kd) (Schwarzenbach et al., 2003). Kd is the ratio of the equilibrium concentration of the chemical on the solids to the corresponding equilibrium aqueous concentration. Previous studies have reported Kd values for the sorption of several pharmaceutically-active compounds, fragrances and steroidal hormones to various primary, activated and digested sludges (Golet et al., 2001, 2003; Simonich et al., 2002; Ternes et al., 2004; Artola-Garicano et al. 2003; Clara et al., 2004; Go¨bel et al., 2005; Urase and Kikuta, 2005; Andersen et al., 2005; Maurer et al., 2007; Yi and Harper, 2007; Carballa et al., 2008; Wick et al., 2009; Radjenovic et al., 2009). However, some of these studies have used TOrC concentrations in the mg/L to mg/L range to measure Kd (Ternes et al., 2004; Clara et al., 2004; Urase and Kikuta, 2005; Yi and Harper, 2007; Wick et al., 2009), which are orders of magnitude greater than concentrations observed in raw municipal wastewaters for many emerging contaminants. Some studies have relied on single point calculations rather than sorption isotherms for calculating Kd values (Ternes et al., 2004; Golet et al., 2003; Go¨bel et al., 2005; Urase and Kikuta, 2005; Maurer et al., 2007; Wick et al., 2009; Radjenovic et al., 2009), which may not be appropriate at other TOrC concentrations. Only a limited set of emerging TOrCs that typically enter wastewater treatment plants have been extensively studied in regards to their sorption onto wastewater solids; the sorption behavior of positively-charged compounds is especially limited (Siegrist et al., 2003; Golet et al., 2003). Also, some of these studies relied on chemical inactivation techniques (Clara et al., 2004; Maurer et al., 2007; Yi and Harper, 2007; Wick et al., 2009) as opposed to nonchemical techniques (Ternes et al. 2004; Andersen et al., 2005), such as the lyophilization and dry-heating technique (Kerr et al. 2000). Chemical inactivation techniques, such as sodium azide and mercury salts, have the potential to affect the sorption behavior of compounds via the chemicals altering the aqueous matrix or the solids surface characteristics. Large and diverse datasets of experimentally determined partitioning coefficients for environmentally relevant conditions are valuable, both directly as inputs into mass balance models and for developing predictive models for the estimation of partitioning coefficients for compounds for which no experimental data are available. Sorption of TOrCs to sludge solids can potentially depend on the fraction of organic carbon present on the solids.
Sorption of nonionic solutes has been accurately expressed by the octanol-carbon distribution coefficient normalized to the organic-carbon content (Schwarzenbach et al., 2003) Kd ¼ foc Koc where Kd is in L/kgsolid, Koc is the organic-carbon distribution coefficient (L/kgoc), and foc is the fraction of organic carbon present on the solid (kgoc/kgsolid). Linear free energy relationships (LFERs) have been used for decades to estimate Koc (Gawlik et al., 1997; Nguyen et al., 2005), where the octanolwater partitioning coefficient, Kow, has been widely used to estimate Koc in one-parameter models (Gerstl, 1990; Sabljic et al., 1995; Huuskonen, 2003). These estimation techniques have the following form: log Koc ¼ alog Kow þ b where a and b are constants estimated from empirical data. These models were developed using sorption datasets for soil/ water systems and their applicability toward wastewater solids, where the fraction of organic carbon is much higher, requires verification. The purpose of the current study was twofold: 1) to experimentally measure the partitioning coefficients of 34 emerging ionic and nonionic TOrCs onto sludge solids using a nonchemical lyophilization and air-heat inactivation technique, and 2) evaluate predictive models for partitioning based on log Kow.
2.
Methods
2.1.
Sorption experiments
Sludge solids for sorption experiments came from three sources: primary clarifier sludge from Denver’s Wastewater Reclamation District facility (Denver Metro), mixed liquor from Denver Metro’s nitrified aeration basin, and mixed liquor from the nitrified aeration basin of the Colorado School of Mines’s (CSM’s) Mines Park laboratory-scale wastewater treatment system. Mixed liquor is defined as the mixture of raw or settled wastewater and activated-sludge in a bioreactor. Denver Metro’s treatment train consisted of primary clarification, a Modified Ludzack-Ettinger (MLE) process with nitrification, denitrification and centrate side-stream treatment, secondary clarification and chlorination. The plant treated 344 MLD (344,000 m3/d) of municipal wastewater from the City of Denver, Colorado, which had an influent biochemical oxygen demand (BOD) of 300 mg/L. The plant was operated at a food to microbial (F/M) ratio of 0.3 and sludge retention time (SRT) of 5 days. The mixed liquor from Denver Metro had the following characteristics: pH 7, chemical oxygen demand (COD) ¼ 11 mg/L, ammonia ¼ 0.3 mg-N/L, nitrate ¼ 0.3 mg-N/L, total suspended solids (TSS) ¼ 2600 mg/L and total organic carbon (TOC) ¼ 12 mg-C/L. Mines Park’s treatment train consisted of preliminary screening, secondary treatment with nitrification, and secondary clarification. The system treated 28 L/d (0.028 m3/d) from a student residential community at CSM, which had an influent COD of 900 mg/L. The plant was operated at a F/M ratio of 0.5, and SRT of 6 days. The mixed liquor from Mines Park had the following
3419
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 4 1 7 e3 4 2 6
characteristics: pH 7, COD ¼ 27 mg/L, ammonia ¼ 0.5 mg-N/L, nitrate ¼ 15 mg-N/L, TSS ¼ 1500 mg/L and TOC ¼ 8 mg-C/L. Preliminary experiments were performed to determine the most appropriate method to biologically inactivate sludge solids, and these experiments and their results are described in Appendix B of the supplementary material. The lyophilization and dry-heat inactivation technique (Blackburn, 1985; USEPA, 1998; Kerr et al., 2000; Andersen et al., 2005) was chosen and used for isotherm sorption experiments since it is a nonchemical technique and it sufficiently inactivated sludge for the compounds examined. Using respirometry and measuring enzymatic activity Kerr et al. (2000) demonstrated this inactivation procedure selectively inhibited microbial activity for a period of approximately 24 h, and the type and degree of enzymatic activity was shown to be dramatically reduced in the sludge. Also the lyophilization and dry-heat method did not significantly alter the aqueous matrix as compared to the chemical inactivation technique, using 0.5% sodium azide and 5 mM barium chloride and 5 mM of nickel chloride (i.e., increase in ionic strength, trace divalent ions added). Preliminary inactivation comparison tests (shown in Appendix B) revealed the inactivation by chemical biocides affected the sorption of positively-charged compounds. Also, Kerr et al. (2000) demonstrated this lyophilization-dry-heat inactivation procedure did not significantly alter activated-sludge solids. Video-enhanced light microscopy revealed the structural integrity of the bacterial cell walls was maintained and zeta potential measurements confirmed the negatively-charged surface was maintained on the solids. Field sludge samples were processed immediately after collection. Sludge samples were allowed to settle and subsequently decanted in order to concentrate the solids. The remaining solids mixture was centrifuged (15 min., g-force y 1050 g) in 250-mL polypropylene centrifuge bottles. The centrifuged supernatant was decanted and ultra pure water was added. The bottles were then shaken for 5 min and centrifuged again for 15 min. This process was repeated two more times for a total of three ultra pure water rinses. The supernatant was decanted and the remaining solids were placed into capped 50 mL clear glass jars. The solids were then frozen at 80 C for at least 24 h. The samples were then lyophilized in a shelf-freeze dryer overnight (covered loosely with aluminum foil) using a LABCONCO FreeZone benchtop freeze drying system. For inactivation, the solids were lightly ground and then placed in an oven at 103 C overnight prior to being stored at 4 C. The foc of all dried sludge solids (Table S3) was measured by a UIC CM5014 coulometric solid-phase TOC analyzer. Prior to being used in a sorption experiment, freezedried solids were reheated at 103 C overnight. Oven-dried solids were mixed with buffered synthetic wastewater (recipe listed in Appendix B of the supplementary material) using a shaker table followed by centrifugation and decantation. For primary-sludge solids a blender was required to initially homogenize the mixture. The rinsing procedure was repeated until the final aqueous TOC was <10 mg/L (Table 1). The inactivated-sludge solids were then used in sorption experiments immediately after washing. Initial aqueous samples were collected (concentration data shown in Table S2 in supplementary material) before spiking TOrC to determine the initial levels of the 34 TOrCs in the
Table 1 e TSS, UV, TOC, pH, and conductivity measurements for each reactor in the sorption experiments. TSS (mg/L)
UV254 (1/cm)
TOC (mg/L)
pH
Conductivity (mS/cm)
Initial
Initial
Initial
Initial, 2 h
Initial
Denver Metro AS 529 0.02 998 0.04 2970 0.06 4600 0.10 6770 0.08 8770 0.04
1.38 2.62 3.74 3.68 4.37 2.67
7.8, 7.7, 7.4, 7.1, 6.9, 7.2,
7.8 7.7 7.6 7.4 7.4 7.5
449 397 214 245 205 171
Denver Metro Primary 438 0.02 913 0.04 3250 0.06 6800 0.17 6420 0.12 11,200 0.18
8.65 1.86 4.13 6.50 6.37 9.18
7.6, 7.7, 7.3, 7.3, 7.0, 7.0,
7.8 7.7 7.5 7.6 7.5 7.5
390 366 286 271 195 163
Mines Park AS 533 1240 2420 4680 6350 8530
1.24 1.75 3.66 3.83 2.80 3.47
7.9, 7.9 7.7, 7.8 7.5, 7.6 7.5, 7.8 6.7, 7.2 7.0, 7.0
432 427 272 258 42 55
0.07 0.04 0.11 0.15 0.08 0.10
aqueous phase of the reconstituted freeze-dried sludge solution. Table 2 lists target TOrCs and Appendix A presents their structures and CAS numbers. Background concentrations in the buffered synthetic wastewater were below detection limits (data shown in Table S2 in supplementary material). Isotherm experiments were performed for freeze-dried TSS concentrations ranging between 500 and 10,000 mg/L (6 isotherm data points, shown in Table 1), and TOrCs were initially spiked at w4 nM (exceptions being diazepam, omeprazole, and phenylphenol were spiked at lower concentrations; Table S1). All reactors were capped and mixed on a shaker table. Experiments were performed at ambient temperature (w19 C) for 2 h. Based on preliminary kinetic tests (presented in Appendices B and C of the supplementary material), 2 h was determined to be sufficient for compounds to achieve partitioning equilibrium. The experimental procedure is presented in Appendix B and the following kinetic points were sampled: initial, 30 s, 1 h, 2 h, 4 h, and 24 h. Two hours was also used because concentrations of highly bioamenable compounds, such as caffeine, were found to be reduced after 4 h during kinetic tests, suggesting the partial reactivation of sludge after this time. The initial conductivity was measured and pH was determined at the beginning and end of the experiments (Table 1). Isotherm tests were performed in duplicate. A chemical abiotic control (with 0.5% sodium azide, 5 mM barium chloride, and 5 mM nickel chloride) was performed in parallel which confirmed the freeze-dried sludge was sufficiently inactivated (recovery data shown in Table S5 in supplementary material). Abiotic controls were performed in duplicate. In addition, a sorption control was performed in
Compound
Denver Metro AS Kd (L/kgSS) log Koc Inactivated
Denver Metro Primary Kd (L/kgSS)
Mines Park AS
log Koc
Inactivated
Kd (L/kgSS)
Literature data log Koc
Inactivated
Fresh
Inactivated
4555 (386) 1642 (180) 1501 (77) 861 (119) 819 (125)
4.01 3.56 3.52 3.28 3.26
5694 (684) 1730 (245) 1644 (348) 964 (99) 778 (154)
4.06 3.54 3.52 3.29 3.19
2686 (506) 1153 (160) 1232 (149) 669 (70) 808 (171)
2343 (292) 1324 (6) 995 537 920 (190)
3.78 3.41 3.44 3.17 3.25
Trimethoprim Atenolol Ethynylestradiol (17a) Estrone Estradiol (17b) Androsterone Testosterone Androstenedione Estriol Bisphenol A Phenylphenol Diazepam Carbamazepine
119 (49) <30 1550 (223) 645 (87) 771 (108) 579 (108) 157 (36) 156(6) 63 431 (35) 347 (64) 241 (59) 50 (1)
2.42 3.54 3.16 3.23 3.11 2.54 2.54 2.15 2.98 2.89 2.73 2.05
251 (99) 46 (6) 1017 (105) 636 (104) 560 (67) 534 (81) 178 (28) 174 (22) 58 (22) 314 (66) 652 (161) 291 (50) 65 (5)
2.70 1.96 3.31 3.10 3.05 3.03 2.55 2.54 2.06 2.80 3.12 2.76 2.11
193 (104) 35 1103 (76) 607 (48) 533 (34) 419 (80) 136 (17) 134 (13) 54 505 (83) 259 (69) 197 (31) 36 (2)
135(68) BD8 BD8 BD8 BD8 BD8 BD8 BD8 BD8 BD8 BD8 161 <135
2.63 1.89 3.39 3.13 3.07 2.97 2.48 2.47 2.08 3.05 2.76 2.64 1.90
DEET Omeprazole
42 107 (25)
1.97 2.38
100 (19) 130 (25)
2.30 2.41
<30 169 (40)
BD8 226
2.57
Atrazine TCEP Primidone Meprobamate Acetaminophen Caffeine Atorvastatin Dilantin Gemfibrozil Ibuprofen Diclofenac Naproxen Sulfamethoxazole Enalapril
60 (2) 65 (20) <30 <30 <30 <30 198 (69) 81 45 <30 <30 <30 <30 <30
2.12 2.16
122 (24) 162 (72) 45 (10) 42 (12) <30 <30 216 (82) 45 (22) 45 (9) <30 <30 <30 <30 <30
2.39 2.51 1.95 1.92
<30 <30 <30 <30 <30 <30 93 32 30 <30 <30 <30 <30 <30
<89 231 <100 <190 BD8 BD8 106 (36) <104 BD8 BD8 <196 BD8 <150 BD8
2.64 2.26 2.00
2.64 1.95 1.95
p: 4271 s: 2531, 2082 p: 951 s: 641 p: 2783 s: 3493, 438e5054, 5846 s: 170e2364, 4026 s: 15064, 4766
s: 217e2734 p: 443 s: 213, 535 p: 3141,<203 s: 1351, 1.23, 664, 175
s: 75 p: 6.71 s: 11601 2.32 1.85 1.82
p: 231 s: 19.31, 1004 p: 9.51,<203 s: 7.13, 804 p: 1941,4593 s: 1181, 163, 324 s: 244 p: 3.11 s: 771, 2562
Positive (9.76/99.8) Pos./Neut. (7.35/69.1) Positive (9.68/99.8) Positive (8.76/98.3) Pos./Neut. (2.09,7.82/ 19.65,67.1) Pos./Neut. (7.16/58.7) Positive (9.67/99.8) Neutral (10.33/99.9) Neutral (10.33/100) Neutral (10.33/100) Neutral (18.30/100) Neutral (19.38/100) Neutral Neutral (10.33/99.6) Neutral (9.78/99.8) Neutral (9.69/99.8) Neutral (2.92/0.0) Neutral (15.96/100) Neutral Neutral (9.68,4.77/ 99.2,0.59) Neutral (3.2/0.01) Neutral Neutral (11.50/100) Neutral (15.17/100) Neutral (9.46/99.7) Neutral (0.92/0) Negative (4.33/0.2) Neut./Neg. (6.46/22.3) Negative (4.42/0.26) Negative (4.88/0.71) Negative (4.00/0.1) Negative (4.19/0.15) Neut./Neg. (6.16/12.65) Negative (3.67,5.28/0,1.52)
BD e compound degraded, so Kd could not be determined. a e percentage distribution of the acid species. 1 Radjenovic et al., 2009; 2Go¨bel et al., 2005; 3Ternes et al., 2004; 4Urase and Kikuta, 2005; 5Wick et al., 2009; 6Andersen et al., 2005; 7ChemAxon, 2010; 8Dickenson et al., 2010.
log Dow at pH 77
2.12 2.89 2.42 0.86 2.53 0.92 2.14 3.81 4.31 3.74 3.77 3.37 3.93 2.67 4.04 3.32 3.08 2.77 2.50 2.43 2.20 2.11 1.12 0.93 0.91 0.55 2.77 2.13 1.85 1.71 1.37 0.25 0.14 1.1
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 4 1 7 e3 4 2 6
Amitriptyline Clozapine Verapamil Risperidone Hydroxyzine
Kd (L/kgSS)
Charge at pH 7 (pKa1, pKa2/a1,a2)7
3420
Table 2 e List of measured Kd and log Koc values for experimental TOrCs in each sludge, previously reported Kd values for primary (p) secondary (s) sludge solids, charge state of the dominant species at pH 7, and log Dow at pH 7. Measured Kd values in italics are based on single point calculations. Numbers in parenthesis for measured Kd values are 95% confidence intervals with the exception of those numbers in italics which are standard deviations.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 4 1 7 e3 4 2 6
triplicate, which consisted of fresh sludge that had not been inactivated (activated-sludge from Mines Park) in its preexisting mixed liquor matrix (TSS ¼ 2900 mg/L). In order to assess the accuracy of the initial dose concentration, four replicates of the spiking technique into synthetic wastewater only (dosing controls) were performed (data shown Table S1 in supplementary material). For acidic and basic compounds the following relationships, respectively, were used to derive log Dow values (Schwarzenbach et al., 2003): log Dow ¼ log Kow þ log
1 1 þ 10pHpKa
log Dow ¼ log Kow þ log
1 1 þ 10pKa pH
Adsorption isotherms were developed for the partitioning of each compound on sludge solids where the amount in the aqueous phase (Maq) was taken as the measured concentration at 2 h minus the initial aqueous concentration as measured from aqueous blanks. Assuming equilibrium partitioning of initially present TOrCs had occurred before TOrCs were dosed, Maq is the amount of dosed TOrC that remained in the aqueous phase. As utilized in other studies (Andersen et al., 2005), the amount of any given TOrC in the solid phase (Ms) was calculated from the mass balance equation: MD ¼ Maq þ Ms, where the amount of TOrC dosed (MD) was measured from the dosing controls. This assumes the TOrCs will only be present either in the aqueous or solid phases. Data was omitted where the fraction of compound in the aqueous phase ( fw) was greater than 80%. Regression analysis using the remaining data was used to estimate a linear isotherm, a Kd value, and associated 95% confidence interval for some compounds. For low sorbing compounds the data did not fit a linear isotherm and single point Kd values were determined.
2.2.
Analytical methods
TOrCs were measured by an isotope-dilution LC-MS/MS method using an Applied Biosystems API 4000 Q-Trap. This method was based on the methods developed by Vanderford et al. (2003) and Vanderford and Snyder (2006). Vanderford and Snyder (2006) observed that without employing isotope dilution, TOrC levels can be greatly under estimated due to matrix suppression of ion formation. Accordingly, isotope dilution was used to correct for matrix suppression, solid-phase extraction losses, and reconstitution and instrument variability. Details of the LC-MS/MS method and the monitored mass spectral transitions are provided in Appendix E in the supplementary material. New isotopic dilution methods were developed for the following compounds: omeprazole, clozapine, amitriptyline, verapamil, and hydroxyzine. External calibration results for phenylphenol were uncorrected as the transitions for its isotope surrogate failed QA/QC criteria, and results for TCEP were also uncorrected because an isotope standard was not available. 250 mL of sludge samples for TOrC analysis were initially centrifuged (15 min., g-force y 1050 g) with 250-mL polypropylene centrifuge bottles and then the supernatant was filtered with a 1.2 mm filter (Whatman GF/C glass fiber filter)
3421
prior to solid-phase extraction. Centrifuge and filtered controls were performed to assess losses during centrifugation followed by filtration. For this test target compounds were spiked between 369 and 5260 ng/L in buffered synthetic wastewater (pH 7). Samples were collected prior to centrifugation and after centrifugation/filtration. Results are shown in Table S4. For most compounds, recoveries were near 100% indicating very little loss of compounds during centrifugation and filtration. After centrifugation/filtration a solution containing 100 ng of each of the isotopically labeled standards was added to samples at pH 6e7 prior to solid-phase extraction with plastic Waters Oasis HLB cartridges (6 cc/500 mg, 60 mm; Part# 186000115). Cartridges were sequentially preconditioned using 5 mL of MTBE, 5 mL of methanol, and 5 mL of ultra pure water. Then cartridges were loaded with sample at approximately 5 mL/min. After loading, the cartridges were rinsed with 5 mL of ultra pure water. Cartridges were then dried under nitrogen until they were visibly dry. Dried cartridges were stored and cooled at 4 C in sealed plastic bags pending analysis.
3.
Results and discussion
3.1.
Sorption isotherm experiments
Isotherm plots for 2 compounds, estrone and verapamil, on activated-sludge solids from Denver Metro and CSM Mines Park and primary solids from Denver Metro are illustrated in Fig. 1. Isotherm plots for other compounds are presented in Appendix D in the supplementary material. Table 2 lists Kd values for each compound in each sludge, as well as the compound’s charge state and octanol-water partitioning coefficient at pH 7, and, if available, previously reported partitioning coefficients for primary and secondary sludges. Single point Kd values should be considered rough estimates, though they are low, indicating these compounds sorb relatively poorly. Compounds with Kd < 30 L/kg for inactivated sludge are compounds with low sorption potential, since fw > 0.8 for these compounds. Likewise, Kd values for atenolol, carbamazepine, diclofenac, gemfibrozil, ibuprofen, naproxen, primidone, and sulfamethoxazole have been previously reported to be low (<100 L/kg) (Ternes et al., 2004; Andersen et al., 2005; Urase and Kikuta, 2005; Maurer et al., 2007; Wick et al., 2009; Radjenovic et al., 2009). Nineteen compounds in Table 2 do not have previously reported partitioning sorption coefficients for secondary and primary-sludge solids. Interestingly, positively-charged aromatic compounds, such as amitriptyline, clozapine, verapamil, risperidone, and hydroxyzine, have the highest Kd values 669e5694 L/kgSS. This suggests that the positively-charged compounds sorb to sludge solid surfaces partly by an electrostatic interaction. The microorganisms have a negativelycharged surface, which acts as a cation exchanger, where a stronger association will occur between this surface and a positively-charged species, than with a neutral compound (Schwarzenbach et al., 2003). Note, Kerr et al. (2000) confirmed by zeta potential measurements that a negatively-charged surface was maintained on lyophilized and dry-heated solids. A previous study observed the electrostatic interaction for positively-charged compounds, i.e., fluoroquinolones, which
3422
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 4 1 7 e3 4 2 6
Fig. 1 e Sorption isotherms for the hormone estrone and the positively-charged compound verapamil in Denver Metro activated-sludge, Denver Metro primary sludge, and Mines Park activated-sludge. Six concentrations of suspended solids were used in duplicate batch experiments to fit linear isotherms for each compound in each sludge.
also have high log Kd values of w4 (Golet et al., 2003). However, positively-charged compounds, trimethoprim and atenolol, had a noticeably lower potential to sorb to sludge solids. Similar Kd magnitudes for trimethoprim were observed by others (Go¨bel et al., 2005; Radjenovic et al., 2009). One chemical characteristic difference between atenolol and the other positively-charged compounds (i.e., amitriptyline, clozapine, verapamil, risperidone, hydroxyzine) is that the latter compounds are distinctly more hydrophobic at pH 7 (log Dow 1e3) as compared to atenolol (log Dow 2.14). This suggests the hydrophobic sorption interactions are still important for positively-charged compounds. However, trimethoprim and risperidone have similar log Dow values at pH 7, 0.92 and 0.86, respectively, though risperidone has a much higher sorption potential. The reason for this is unknown. It is important to point out the pH varied from 6.7 to 7.9 between the different sludge concentrations (Table 1), which likely affected the results for the partly charged compounds near neutral pH, like clozapine (pKa 7.35), hydroxyzine (pKa 7.82), and trimethoprim (pKa 7.16). The neutral hormones, 17a-ethinylestradiol, 17b-estradiol, androsterone, and estrone have high Kd values ranging from 419 to 1550 L/kgSS, though these are lower, except for 17aethinylestradiol, than the highly-sorbing positively-charged compounds. For these compounds the high Kd values corresponded to high log Kow values (3.7e4.3). Interestingly, these log Kow values are 1 log unit higher than the log Dow values for the positively-charged compounds that had a higher sorbing potential. Assuming hydrophobic interaction is a sorption driver for steroidal hormones it is unknown why androstenedione with a log Kow of 3.93 and similar in structure, did not have a higher sorption potential, Kd is only 134e174 L/kgSS. A set of other neutral compounds, testosterone, bisphenol A, phenylphenol and diazepam generally has lower Kd values, 136e652 L/kgSS, than the before mentioned high-sorbing steroidal compounds. These lower Kd values corresponded with lower log Kow values of 3.1e4.0. Interestingly, omeprazole has comparable Kd values of 107e169 L/kgSS, eventhough it has
a lower log Kow value of 2.43. Though the neutral species for omeprazole dominates at pH 7 (99.2%) the protonated form is 0.59% present, which could be responsible for the increased sorption due to electrostatic interactions. Neutral compounds, estriol, carbamazepine, DEET, atrazine, TCEP, primidone, meprobamate, acetaminophen, and caffeine, have lower Kd values <30e162 L/kgSS, which corresponded with lower log Kow values of 0.5 to 2.8. In general, increasing log Kow is indicative of increasing sorption potential for neutral compounds. Negatively-charged compounds, atorvastatin, dilantin, and gemfibrozil, have Kd values of 93e216, 32e81, and 30e45 L/kgSS, respectively, which correlates in descending order with their respective log Dow values of 2.8, 2.1 and 1.9. The other negativelycharged compounds, ibuprofen, diclofenac, naproxen, sulfamethoxazole and enalapril, have Kd values <30 L/kgSS, where log Dow is <1.7 for these compounds. Similar to the neutral compounds, increasing log Dow is indicative of increasing sorption potential for negatively-charged compounds. Mines Park Kd values determined using the lyophilized and dry-heated inactivation technique were comparable with those generated with fresh activated-sludge for positively-charged compounds, amitriptyline, clozapine, verapamil, risperidone, hydroxyzine, and trimethoprim, neutral compounds, omeprazole and diazepam, and the negatively-charged compound, atorvastatin (Table 2). These compounds were observed to be recalcitrant in complementary biodegradability studies using mixed liquor sample from the same Mines Park aerobic/nitrified treatment process (Dickenson et al., 2010). Kerr et al. (2000) further confirmed good agreement between Kd values determined by the lyophilized and dry-heated inactivation technique with those derived from fresh activated-sludge for four surfactant, quaternary ammonium and chelator compounds. These results indicate the sorption behavior on lyophilized and dry-heated sludge solids is comparable to fresh activatedsludge solids. An abiotic control was performed, which consisted of the addition of chemical biocides, 0.5% sodium azide, 5 mM
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 4 1 7 e3 4 2 6
barium chloride, and 5 mM nickel chloride, to reconstituted lyophilized and dry-heated sludge solids. For most compounds and sludges the compound concentrations after 2 h in the sorption tests were comparable to the levels observed after 2 h in the abiotic controls (recoveries are reported in Table S5), indicating the compounds were not attenuated by biological mechanisms during sorption tests. The abiotic tests also allowed for assessment of the impact of the applied chemical biocide in the abiotic control on sorption behavior. For example, sorption potential for lyophilized and dry-heated sludge solids was significantly higher for highly-sorbing positively-charged compounds, amitriptyline, clozapine, hydroxyzine, risperidone and verapamil, (indicative of recoveries>170% in Table S5). It is believed the increased removal was not due to biological attenuation mechanisms, since these compounds were observed to be recalcitrant in complementary biodegradability studies. Furthermore, since the sorption behavior of these compounds on lyophilized and dry-heated sludge solids and fresh solids are comparable, this suggests these chemical biocides are leading to an underestimation of the sorption potential and should not be used for assessing sorption for highly-sorbing positively-charged compounds. The sorption behavior of estrone, 17b-estradiol and 17a-ethynylestradiol to activated-sludge has been previously studied (Ternes et al., 2004; Urase and Kikuta, 2005; Andersen et al., 2005; Yi and Harper, 2007). For these nonionic compounds it is assumed that sorption is governed by partitioning to the organic phase ( foc) in the activated-sludge, and therefore a comparison of log Koc values was assessed. Koc was calculated using reported Kd values and measured or assumed foc values. The foc of 34 and 27.7% was measured by Ternes et al. (2004) and Urase and Kikuta (2005), respectively, and 40% was assumed for the Urase and Kikuta (2005) and Yi and Harper (2007) studies. Log Koc values obtained for estrone, 2.8e3.2 (Urase and Kikuta, 2005; Andersen et al., 2005), 17b-estradiol, 3.1e3.2 (Urase and Kikuta, 2005; Andersen et al., 2005), and 17a-ethinylestradiol, 2.9e3.3 (Ternes et al., 2004; Urase and Kikuta, 2005; Andersen et al., 2005; Yi and Harper, 2007) are similar to those measured for Denver Metro and Mines Park activated-sludge solids, 3.1e3.5, as reported in Table 2.
3423
Kd values are comparable between the three sludges (Fig. 2). The high correlation of coefficients between activated- and primary-sludge solids (r ¼ 0.96, n ¼ 23) indicates that sorptive behavior is relatively similar during these two stages of wastewater treatment at Denver Metro. Kd values obtained from the two activated-sludge sources are also highly correlated (r ¼ 0.97, n ¼ 21), indicating that these results are replicable across two aerated and nitrified sludges, eventhough the sludge solids derive from two systems operated at extremely different scales, 344,000 m3/d versus 0.028 m3/d, and treating different types of wastewater, municipal versus one from a university student population. Interestingly, Carballa et al. (2008) also found that sorption potential for digested sludges are similar to primary and secondary sludges for pharmaceutically-active compounds, such as carbamazepine, ibuprofen naproxen, diclofenac, and sulfamethoxazole, and estrogens, such as estrone, 17b-estradiol and 17aethynylestradiol. Based on observed effects of charge and hydrophobicity on sorption behavior, compounds were subdivided into three classes of compounds: 1) nonionic, 2) negatively-charged, and 3) positively-charged compounds at pH 7. Fig. 3 plots measured log Koc in Denver Metro activated-sludge against calculated log Dow as estimated by Marvin Calculator Plugins version 5.1.5 (ChemAxon, 2010) for the three classes of compounds. A similar Figure (Appendix F of the supplementary material) combines all the measured log Koc values for the three sludge-solids studies. The Denver Metro activated-sludge experimental results showed some correlation between log Koc and log Dow for nonionic compounds (r ¼ 0.82, n ¼ 14). Considering nonionic and negatively-charged compounds together, the correlation is slightly higher (r ¼ 0.83, n ¼ 17). For the positively-charged compounds alone a correlation is also observed (r ¼ 0.61, n ¼ 6), suggesting sorption of positively-charged compounds is related to hydrophobic interactions, but since their log Koc values are consistently higher than nonionic or negativelycharged compounds with similar log Dow values, this also suggests other types of sorption mechanisms (e.g., electrostatic interactions) are involved and, in some cases, may be dominating.
Fig. 2 e Comparison of Kd values obtained from Denver Metro primary sludge (left) and Mines Park activated-sludge (right) to Kd values obtained from Denver Metro activated-sludge.
3424
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 4 1 7 e3 4 2 6
Fig. 3 e Comparison of measured log Koc to log Dow, estimated from Marvin (ChemAxon, 2010), for neutral, positively-charged, and negatively-charged compounds.
3.2.
Evaluation of log Kow based models
Previous studies have developed simple LFER models for predicting log Koc from log Kow using soil-based systems. Three such models are listed below. log Koc ¼ 0.52 log Kow þ 1.02 (Sabljic et al., 1995)
Fig. 4 e Log Koc from Denver Metro activated-sludge versus log Kow for neutral compounds at pH 7. Lines indicate predicted log Koc based on log Kow from three previously published models.
log Dow þ 0.15. They found for the neutral compounds (log Kow > 2.5) the model consistently under predicted the measured Koc, but the predictions were still within 1 log unit. In comparison, the best-fit linear model possible for our dataset is log Koc ¼ 0.602 log Kow þ 0.695, which has a RMSE of 0.285. Similar simplified relationships are lacking and needed for positivelycharged compounds, since these compounds with log Dow > 1 have a strong sorbing tendency.
(n ¼ 390, R2 ¼ 0.63)
log Koc ¼ 0.679 log Kow þ 0.663 (Gerstl, 1990)
2
(n ¼ 419, R ¼ 0.83) log Koc ¼ 0.6 log Kow þ 0.84 (Huuskonen, 2003) (n ¼ 403, R2 ¼ 0.79) These models were developed using large, diverse sets of organic compounds. Fig. 4 plots log Koc observed in Denver Metro activated-sludge against log Kow for nonionic compounds. The three models above are represented by lines that overlay the data. Sabljic et al. (1995) provides the best-fit model, having the lowest root mean square error (RMSE) of 0.296. The RMSE incorporates both bias as a measure of accuracy and error variance as a measure of precision, giving a measure of overall fit. Even though these models are based on soils systems, they can be applied to sludge solids and are able to predict the log Koc within 1 log unit for log Kow > 2. Carballa et al. (2008) found similar results, where they compared measured Koc values for sorption of nonionic compounds to digested sludge with estimated Koc using the following linear equation: log Koc ¼ 0.74
4.
Conclusions
Experimental sorption partitioning coefficients, Kd, were quantified for a diverse suite of TOrCs for three different wastewater sludge solids, where 19 of these chemicals did not have previously reported values. Five positively-charged compounds had the highest sorption potential. Sludge solids were inactivated by a nonchemical lyophilization and dry-heat technique, which provided similar sorption behavior as compared to fresh sludge. The sorption behavior was comparable between primary- and activatedsludge solids and activated-sludge solids from two nitrifying basins from different wastewater treatment systems. Simple LFERs based on log Kow can be used to predict the solids-water partitioning of emerging TOrCs in primary or activated-sludge. However simplified empirical models to predict the sorption behavior for positively-charged compounds are lacking. These are critical fate estimating techniques, since many TOrCs, such as pharmaceuticals, are charged at wastewater pH conditions. For the development of robust prediction techniques additional highquality sorption data for structurally diverse compounds, specifically ionic compounds, is needed.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 4 1 7 e3 4 2 6
Acknowledgments The authors thank Anna Hoessle, Nathan Rothe, and Dr. Dean Heil at the Colorado School of Mines for laboratory assistance and Tanya Bayha at Denver’s Wastewater Reclamation District for assisting in sample collection. The team also would like to thank the following people for their insight, constructive critique and feedback of the work herein: Drs. Christopher Higgins, Drew McAvoy, Arthur Meyers, Robert Arnold, and Robert Hannah. The authors are also grateful to the Water Environment Research Foundation (WERF-U2R07) for its financial, technical, and administrative assistance in funding and managing the project through which this information was derived. The comments and views detailed herein may not necessarily reflect the views of the Water Environment Research Foundation, its officers, directors, affiliates, or agents.
Appendix. Supplementary material Supplementary material related to this article can be found online at doi:10.1016/j.watres.2011.03.056.
references
Andersen, H.R., Hansen, M., Kjølholt, J., Stuer-Lauridsen, F., Ternes, T., Halling-Sørensen, B., 2005. Assessment of the importance of sorption for steroid estrogens removal during activated sludge treatment. Chemosphere 61, 139e146. Artola-Garicano, E., Borkent, I., Damen, K., Jager, T., Vaes, W.H.J., 2003. Sorption kinetics and microbial biodegradation activity of hydrophobic chemicals in sewage sludge: model and measurements based on free concentrations. Environ. Sci. Technol. 37 (1), 116e122. Blackburn, J.M., 1985. Organic Chemical Fate Prediction in Activated Sludge Treatment Processes. EPA Report No, EPA/600/2-85/102. US Government, Washington, DC. Carballa, M., Fink, G., Omil, F., Lema, J.M., Ternes, T., 2008. Determination of the solid-water distribution coefficient (Kd) for pharmaceuticals, estrogens and musk fragrances in digested sludge. Water Res. 42, 287e295. Carucci, A., Cappai, G., Piredda, M., 2006. Biodegradability and toxicity of pharmaceuticals in biological wastewater treatment plants. Environ. Sci. Health A Toxic Hazard Subst. Environ. Eng. 41, 1831e1842. ChemAxon, 2010. Calculator Plugins, Marvin 5.1.5. www. chemaxon.com. Clara, M., Strenn, B., Saracevic, E., Kreuzinger, N., 2004. Adsorption of bisphenol-A, 17b-estradiole and 17aethinylestradiole to sewage sludge. Chemosphere 56, 843e851. Clara, M., Strenn, B., Gans, O., Martinez, E., Kreuzinger, N., Kroiss, H., 2005. Removal of selected pharmaceuticals, fragrances and endocrine disrupting compounds in a membrane bioreactor and conventional wastewater treatment plants. Water Res. 39, 4797e4807. Dickenson, E., Drewes, J., Stevens-Garmon, J., Khan, S., McDonald, J., 2010. Evaluation of QSPR Techniques for Wastewater Treatment Processes. Water Environment Research Foundation, Alexandria, VA. Gawlik, B.M., Sotiriou, N., Feicht, E.A., SchulteHostede, S., Kettrup, A., 1997. Alternatives for the determination of the soil
3425
adsorption coefficient, Koc, of non-ionicorganic compounds e a review. Chemosphere 34, 2525e2551. Gerstl, Z., 1990. Estimation of organic chemical sorption by soils. J. Contam. Hydrol. 6, 357e375. Go¨bel, A., Thomsen, A., McArdell, C.S., Joss, A., Giger, W., 2005. Occurrence and sorption behavior of sulfonamides, macrolides, and trimethoprim in activated sludge treatment. Environ. Sci. Technol. 39, 3981e3989. Golet, E.M., Alder, A.C., Hartmann, A., Ternes, T.A., Giger, W., 2001. Trace determination of fluoroquinolone antibacterial agents in solid-phase extraction urban wastewater by and liquid chromatography with fluorescence detection. Anal. Chem. 73, 3632e3638. Golet, E.M., Xifra, I., Siegrist, H., Alder, A.C., Giger, W., 2003. Environmental exposure assessment of fluoroquinolone antibacterial agents from sewage to soil. Environ. Sci. Technol. 37 (15), 3243e3249. Huuskonen, J., 2003. Prediction of soil sorption coefficient of a diverse set of organic chemicals from molecular structure. J. Chem. Inf. Comput. Sci. 43, 1457e1462. Joss, A., Zabczynski, S., Go¨bel, A., Hoffmann, B., Lo¨ffler, D., McArdell, C.S., Ternes, T.A., Thomsen, A., Siegrist, H., 2006. Biological degradation of pharmaceuticals in municipal wastewater treatment: proposing a classification scheme. Water Res. 40, 1686e1696. Kerr, K.M., Larson, R.J., McAvoy, D.C., 2000. Evaluation of an inactivation procedure for determining the sorption of organic compounds to activated sludge. Ecotoxicol. Environ. Saf. 47, 314e322. Kim, S.D., Cho, J., Kim, I.S., Vanderford, B.J., Snyder, S.A., 2007. Occurrence and removal of pharmaceuticals and endocrine disruptors in South Korean surface, drinking, and waste waters. Water Res. 41 (5), 1013e1021. Maurer, M., Escher, B.I., Richle, P., Schaffner, C., Alder, A.C., 2007. Elimination of [beta]-blockers in sewage treatment plants. Water Res. 41, 1614e1622. Nguyen, T.H., Goss, K.-U., Ball, W.P., 2005. Polyparameter linear free energy relationships for estimating the equilibrium partition of organic compounds between water and the natural organic matter in soils and sediments. Environ. Sci. Technol. 39, 913e924. Radjenovic, J., Petrovic, M., Barcelo´, D., 2009. Fate and distribution of pharmaceuticals in wastewater and sewage sludge of the conventional activated sludge (CAS) and advanced membrane bioreactor (MBR) treatment. Water Res. 43, 831e841. Sabljic, A., Gu¨sten, H., Verhaar, H., Hermens, J., 1995. QSAR modelling of soil sorption. Improvements and systematics of log Koc vs. log Kow correlations. Chemosphere 31, 4489e4514. Schwarzenbach, R.P., Gschwend, P.M., Imboden, D.M., 2003. Environmental Organic Chemistry. Wiley-Interscience, Hoboken, NJ. Siegrist, H., Joss, A., Alder, A., McArdell-Bu¨rgisser, C., Go¨bel, A., Keller, E., Ternes, T.A., 2003. Micropollutantsdnew challenge in wastewater disposal? EAWAG News 57, 7e10. Simonich, S.L., Federle, T.W., Eckhoff, W.S., Rottiers, A., Webb, S., Sabaliunas, D., de Wolf, W., 2002. Removal of fragrance materials during U.S. and European wastewater treatment. Environ. Sci. Technol. 36 (13), 2839e2847. Snyder, S.A., Westerhoff, P., Yoon, Y., Sedlak, D.L., 2003. Pharmaceuticals, personal care products, and endocrine disruptors in water: implications for the water industry. Environ. Eng. Sci. 20, 449e469. Stackelberg, P.E., Furlong, E.T., Meyer, M.T., Zaugg, S.D., Henderson, A.K., Reissman, D.B., 2004. Persistence of pharmaceutical compounds and other organic wastewater contaminants in a conventional drinking-water-treatment plant. Sci. Total Environ. 329, 99e113.
3426
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 4 1 7 e3 4 2 6
Ternes, T.A., Herrmann, N., Bonerz, M., Knacker, T., Siegrist, H., Joss, A., 2004. A rapid method to measure the solid-water distribution coefficient (Kd) for pharmaceuticals and musk fragrances in sewage sludge. Water Res. 38, 4075e4084. Urase, T., Kikuta, T., 2005. Separate estimation of adsorption and degradation of pharmaceutical substances and estrogens in the activated sludge process. Water Res. 39, 1289e1300. USEPA, 1998. Activated sludge sorption isotherm. In: Fate, Transport and Transformation Test Guidelines. USEPA EPA 712-C-98-298. Vanderford, B.J., Pearson, R.A., Rexing, D.J., Snyder, S.A., 2003. Analysis of endocrine disruptors, pharmaceuticals, and
personal care products in water using liquid chromatography/ tandem mass spectrometry. Anal. Chem. 75, 6265e6274. Vanderford, B.J., Snyder, S.A., 2006. Analysis of pharmaceuticals in water by isotope dilution liquid chromatography/tandem mass spectrometry. Environ. Sci. Technol. 40, 7312e7320. Wick, A., Fink, G., Joss, A., Siegrist, H., Ternes, T.A., 2009. Fate of beta blockers and psycho-active drugs in conventional wastewater treatment. Water Res. 43, 1060e1074. Yi, T., Harper Jr., W.F., 2007. The effect of biomass characteristics on the partitioning and sorption hysteresis of 17aethinylestradiol. Water Res. 41, 1543e1553.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 4 2 7 e3 4 3 8
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Morphological characterisation of ATAD thermophilic sludge; sludge ecology and settleability Anna V. Piterina a,b,c,*, John Bartlett d, J. Tony Pembroke a,c a
Molecular Biochemistry Laboratory, Department of Chemical and Environmental Sciences, University of Limerick, Limerick, Ireland Centre for Applied Biomedical Engineering Research (CABER), Department of Mechanical, Aeronautical and Biomedical Engineering, University of Limerick, Limerick, Ireland c Materials and Surface Science Institute (MSSI), University of Limerick, Limerick, Ireland d Centre for Sustainability, Institute of Technology Sligo, Sligo, Ireland b
article info
abstract
Article history:
Autothermal thermophilic aerobic digestion (ATAD) is a biological wastewater treatment
Received 10 July 2010
process used for stabilisation of domestic, animal, food and pharmaceutical sludges, and
Received in revised form
wastewater. It produces a high-quality effluent due to thermophilic processing conditions,
30 March 2011
however the stabilised sludge has poor settling characteristics, a high water content, low
Accepted 30 March 2011
compaction capacity and is difficult to dewater by mechanical processes alone. These
Available online 5 April 2011
factors impact transport and disposal of processed ATAD sludge. We have carried out a detailed morphological characterisation of ATAD sludge at all stages of the ATAD process
Keywords:
in an attempt to determine key characteristics of the sludge that might be responsible for
ATAD sludge morphology
its poor dewatering and settleability. A number of microscopic techniques including
Microscopy
electron, optical, wide field and laser scanning confocal microscopy were applied to fresh,
Biosolids
fixed or embedded sludge taken at various stages during a full scale ATAD process treating
Multilabeling techniques
domestic sludge. The spatial distributions of structural sludge matrix components were
Laser scanning confocal microscopy
determined and suggested a highly dynamic sludge morphology during the overall process. Large amounts of fibres were observed in the feed sludge, whereas thermophilic sludge liquor with low settleability was shown to have a lower protein to polysaccharide ratio (1:0.9) compared to the easily settled fraction where ratio values were in the range of (1:1.14e1:1.7) with a prevalence of protein constituents. ATAD sludge was also shown to contain colloids, slime, cellulose micro-particles and multiple hydrophobic droplets in the bulk liquor, factors that may markedly impact on sludge dewaterability characteristics. Laser scanning confocal microscopy demonstrated a superior ability to identify composition and spatial localisation of structural constituents in such a dispersed, high water content sludge. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
Autothermal thermophilic aerobic digestion (ATAD) is a wastewater stabilisation process which utilises the heat
generated by aerobic growth and metabolic activity of microbial populations in insulated reactors to pasteurise treated sludge and produce a Class A Biosolids product suitable for land application with no site restriction (US EPA,
* Corresponding author. Molecular Biochemistry Laboratory, Department of Chemical and Environmental Sciences, University of Limerick, Limerick, Ireland. Tel.: þ353 85 724 68 00; fax: þ353 61 202 568. E-mail addresses:
[email protected] (A.V. Piterina),
[email protected] (J. Bartlett),
[email protected] (J.T. Pembroke). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.03.054
3428
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 4 2 7 e3 4 3 8
1994; Piterina et al., 2010b). Although the basic design principles for ATAD systems are well established (Kelly et al., 1993; LaPara and Alleman, 1999; Layden et al., 2007a; Layden et al., 2007b; Mohaibes and Heinonen-Tanski 2004), it is becoming apparent that a greater understanding of the physico-chemical and biological parameters of such a system is required to optimise performance and ensure its stability and versatility for different feed types and loading rates. The ATAD sludge produced following treatment demonstrates poor settling and low compaction capacity (Murthy et al., 2000; Agarwal et al., 2005; Piterina et al., 2006; Layden, 2007) and it has been observed that sludge dewatering properties change during the ATAD process with sludge showing most resistance to dewatering following thermophilic treatment (Zhou et al., 2002; Agarwal et al., 2005; Layden, 2007). ATAD sludges have proven difficult to dewater by mechanical processing or by filtration (Murthy et al., 2000; Agarwal et al., 2005). The difficulty arises from two key areas, firstly the rate at which water can be removed from digested ATAD sludge is extremely slow, and secondly the total proportion of water that can be removed is too low to be practically significant (Liao et al., 2000). Conditioning of ATAD sludge to thicken it, is critical to the success of ATAD digestion, however, this is costly and ultimately may contribute to its poor dewaterability which in turn has economic consequences for disposal and transport of a high water content sludge. Many process techniques have also been employed to improve the settling characteristics of ATAD sludge, included varying the retention time, increased aeration during digestion and addition of inorganic chemicals and polymers to modify the sludge (Murthy et al., 2000; Agarwal et al., 2005). Such modifications have largely proven unsuccessful and do not provide a generic solution for ATAD biosolids conditioning (Murthy et al., 2000; Agarwal et al., 2005). The nature of the poor dewaterability of ATAD sludge is however little understood and detailed analysis of the micro-composition and physical structure of ATAD sludge is lacking even though such characteristics influence particle aggregation mechanisms (Mikkelsen and Keiding, 2002; Morgan-Sagastume and Allen, 2005; Yu et al., 2008). Organic matter particles and microorganisms together can form structured colonies in the form of flakes, glued together; with extracellular polymeric substances (EPS) of varying composition giving a sludge matrix (Li and Ganczarczyk, 1990). The size and nature of these flocs are determined by operating conditions, the ATAD process, the type of feed, the applied load, the age of the sludge, the hydraulic retention time, and the temperature and pH during operation. These conditions in turn influence the selection and adaptation of microbial species, the pattern of biodegradation and synthesis during processing and consequently determine the quality of the sludge matrix (Herwijn, 1996), which in turn affects the ability of the sludge to settle. Sludge structure on both a macro- and micro-scale are interconnected and effect other key process parameters such as rheological characteristics, rate of heat transfer and penetration within the sludge which effect the capacity of the ATAD sludge to provide specific ecological niches for various microorganisms allowing their metabolism and proliferation as well as efficient pathogen reduction (Li and
Ganczarczyk, 1990; Grijspeerdt and Verstraete, 1997). Factors such as floc size and heterogeneity, type of bacterial community, the amount and composition of extracellular polymeric substances (EPS) (Dignac et al., 1998), the presence of slime (Yu et al., 2008), or colloidal matter (Leppard, 1993; Buffle and Leppard, 1995; Lee et al., 2003), surface properties of flocs (Liao et al., 2001), stability and transformation of sludge properties during processing are considered key issues in understanding the physico-chemical mechanisms influencing settling. To address such issues we have carried out a detailed microscopic analysis of ATAD sludge at all stages during processing to evaluate its utility in determining key characteristics of the sludge that may be responsible for poor settling and dewatering.
2.
Material and methods
2.1. Description of the wastewater treatment facility and sample collection process ATAD sludges were sourced from the ATAD treatment plant in Killarney, Co. Kerry, Ireland. Primary and secondary treated sludges were thickened to 4e6% TS and fed to two ATAD reactors 1A (range 35e49 C), and 2A (range 58e65 C) both of 110 m3 operated in series. Operational parameters, pH, volatile and total solids, total amount of the protein and polysaccharides of the Killarney ATAD have been detailed (Piterina et al., 2006, 2009; Layden, 2007). Samples were taken from the middle of the reactor via a deep-water sampling device as described (Piterina et al., 2009).
2.2.
Optical microscopy
Following sampling, sludge was placed into the cavity of a microscopic slide and covered with a cover slip to maintain the hydrated state, morphological characteristics were observed via an optical microscope and photomicrograph taken by a “Nikon” digital camera. Crystal violet staining was performed as described (Cappuccino and Sherman, 2004). 0.05% aqueous Toluidine blue O was used to differentially stain polyphenols (lignin and tannins stain green to bluegreen) and cellulose (stains pink) (O’Brian et al., 1964).
2.3.
Scanning electron microscopy (SEM analysis)
10 ml of sludge sample was immobilised onto a poly-L-lysinecoated glass slide, allowed to air dry and then fixed with 3% paraformaldehydeephosphate-buffered saline (PBS) (Kiernan, 2000) followed by sequential fixation with PBS e 96% ethanol (1:1 (vol/vol)) for enhanced stabilisation of structural components within the cell wall of gram-positive bacteria (Stahl et al., 1991). This was washed three times with phosphatebuffered saline (10 mM sodium phosphate buffer, 130 mM sodium chloride [pH 7.2]) and the fixed sludge dehydrated via immersion in increasing ethanol concentrations (20e96%) and air dried. Samples were sputter coated with gold (100 nm) and the surface morphology and structural characteristics observed via scanning electron microscope (Jeol, UK).
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 4 2 7 e3 4 3 8
2.4.
Epi-fluorescence microscopy
2.4.1.
Total cell count
Bacterial numbers in ATAD sludge samples were enumerated by staining with 1.5 mg ml1 40 ,60 -diamidino-2-phenylindole (DAPI) (Wei et al., 1995) and examined under UV light (excitation filter BP330-385, barrier filter BA420) with an Nikon epifluorescence microscope. 30 randomly chosen microscopic fields were counted corresponding to 700 to 1500 DAPI-stained cells. Duplicates were used to average cell numbers and express counts as cells per ml of the sludge. Images were processed with Image J software (Collins, 2007).
2.5.
Extraction of EPS
Non clarified thermophilic sludge obtained from reactor 2A after 23 h of processing at 59 C was physically separated as easily settling particulate matter and colloidal matter, without addition of chemical extractants, via centrifugation at 8000g for 20 min and EPS was extracted from particulate matter as previously described in detail by Liu and Fang (2002).
2.5.1.
Formaldehyde and NaOH extraction
50 ml of 36% formaldehyde was added to 25 ml sludge and agitated at 4 C for 1 h. 10 ml of 1 M NaOH was then added and the mixtures kept at 4 C for 3 h before centrifugation at 10,000g for 20 min at 4 C. The supernatant was removed and dialysed using a membrane of 14,000 MWCO against ultra pure water for 2 days at 4 C and stored at 20 C. The protein and carbohydrate content of extracted EPS samples were measured by the modified Lowry (Lowry et al., 1951) and anthrone method (Gaudy, 1962), with BSA and glucose as standards (FrØlund et al., 1995).
2.6.
1968). All staining procedures were performed in the dark and sequentially with triple washes with PBS between staining by each fluorochrome. Laser Scanning Confocal Microscopy (LSCM) (Schmid et al., 2003; Prasad et al., 2007) and image acquisition was performed via a Meta 710 based on Axio Observer 2.1 (Carl Zeiss, Germany) microscope equipped with a 405-UV, argonekrypton laser and spectral detection system containing 32 photomultiplying tubes for sensitive highly precise multichannel detection The observations were performed at different magnification with several types of microscope objectives (10 (air), 20 (air), 40 (oil) and 63 (oil)). The xey images were presented as extended-focus images, produced by taking the confocal images at different depths of the embedded ATAD sludge samples and projecting them into a single image. The extended-focus images and vertical cross sections through the sludge flocs were generated by using the ZEN 2008 software package (Carl Ziess, Germany). Images were further processed for display using Adobe Photoshop software.
3.
Results
3.1.
Initial microscopic observations
Extraction with EDTA
10 ml of 2% EDTA was added to 25 ml sludge and kept at 4 C for 3 h, then centrifuged at 10,000 rpm for 20 min at 4 C and the supernatant (final volume of 35 ml) kept at 20 C. The resulting supernatant was dialysed using a membrane of 14,000 MWCO against ultra pure water for 2 days at 4 C.
2.5.2.
3429
Spatial structure analysis and confocal microscopy
To maximize preservation of the ATAD sludge flocs, the particle embedding method proposed by Ganczarczyk et al. (1992) and modified by Droppo et al. (1996a,b) using agarose blocks was used, allowing good penetration of dyes and washing buffers. The agarose discs generated were placed on a cover slip within an open designed chamber maintaining hydration and preventing sample compression. To visualise polymers and compounds in the ATAD sludge several fluorochromes were applied. Fluorescein isothiocyanate (FITC) (2 mg/ml, PBS) was used to visualise amino groups and protein (Darken, 1961). Nile Red (dissolved in DMSO 50 ug ml1) was used to determine hydrophobic domains of lipids and proteins (Greenspan and Fowler, 1985; Greenspan et al., 1985; Haugland, 2002) Calcofluor White M2R (F-6258 Sigma), was used to visualise chitin, crystalline and amorphous constituents of cellulose and other b-1,4-linked carbohydrates (Hughes and McCully, 1975; Harrington and Raper,
Sludge obtained from a full scale thermophilic ATAD reactor treating domestic sludge was utilised to examine its associated morphological features. Fresh samples of ATAD sludge (in the hydrated state) were found to be composed of suspended matter and particulates, rich in organic matter. Long filamentous chains of microorganisms (Fig. 1, Panel B (a)) were observed within the suspended matter and the majority stained as rodshaped gram-positive bacteria, connected by layers of mucous (Fig. 1, Panel B (b)). These may play a key role in contributing to the viscous nature of the ATAD sludge. The abundance of microorganisms present may be explained by the high concentration of solubilised nutrients in the ATAD sludge and the amount of aeration used to stimulate microbial growth during the process. Within 10 min of samples cooling to room temperature, the morphology of the ATAD sludge altered significantly with chains of microorganisms fragmenting to smaller and shorter chains and becoming dispersed in the bulk water (Fig. 1, Panel B (c)). Our microscopic examination of ATAD sludge revealed an absence of well-defined flocs or granulated material and under low (100) magnification we observed a large amount of dispersed particulate material (micro-particles) (Fig. 1, Panel A (a, b, c)) reactive with celluloseespecific stains (Fig. 1, Panel A (d, e)), suggesting a cellulose-containing material. More detailed analysis of these micro-particles with high-resolution electron microscopy indicated that they appeared to be in close association with sludge microorganisms (Fig. 1, Panel C) and may provide carrier support for ATAD associated microorganisms.
3.2.
Biopolymer content of sludge fractions
Fractionation of the ATAD thermophilic sludge by centrifugation was used to examine the biopolymer content within the various sludge fractions. The liquid fraction of the ATAD sludge contained a large amount of suspended colloidal
3430
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 4 2 7 e3 4 3 8
Fig. 1 e Low, intermediate and high resolution images of the morphological and compositional features of ATAD thermophilic sludge obtained via optical, epifluorecence and scanning electron microscopy (more detailed description and sample preparation description in the text). Panel A: (a, b, c) Phase-contract optical microscopy of thermophilic sludge particulate matter after 23 h operation of reactor 2A. (d, e) e Sludge particulate matter was stained for Cellulose via a cellulose specific stain (cellulose stained from yellowish to orange-red, Lignin stained weak, and appear as dark-brownish to black). Panel B: (a, b, c) morphological appearance of thermophilic sludge liquor, (b) after staining with crystal violet (c) after 10 min cooling to 20 C. Panel C: SEM image of thermophilic sludge. The sludge sample taken from thermophilic reactor 2A was mounted onto poly-L-lysine-coated cover slips, fixed with 2.5% gluteraldehyde overnight and dehydrated in an ethanol series [25% (v/v), 5 min; 50%, 5 min; 75%, 5 min; 95%, 10 min] followed by critical-point drying, then sputter coated with gold. The cells were examined using a Jeol SEM.
particles with a protein to carbohydrate ratio ¼ 0.9. The EPS component of the particulate matter was extracted by two different methods (Liu & Fang, 2002) from the ATAD sludge, to optimise extractability of ATAD EPS components with different chemistries. This revealed that protein was a major EPS constituent associated with this easily settleable particulate ATAD fraction (Table 1). The efficiency of EPS
extraction using the two methods is illustrated for comparison in Table 1 and indicates that the methodology using EDTA extracted more protein, whereas both methods extracted an equal amount of carbohydrate. The ratio of protein to carbohydrate within the EPS extracts obtained with the EDTA and the NaOH methods were 1.76 and 1.14 respectively.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 4 2 7 e3 4 3 8
Table 1 e Major biopolymers associated with the colloidal fraction and EPS extracted from the particulate matter of the ATAD sludge. EPS was extracted by two different methods (EDTA and Alkali) following centrifugation (40003g) of particulate matter. Ratios of biopolymers in ATAD sludge sub-fraction are shown. Characterisation
Particulate matter Extraction Method
Protein (g/l) Carbohydrates (g/l) Protein/ Carbohydrate Ratio
EDTA
NaOH þ formaldehyde
0.312 0.02 0.173 0.04 1.76
0.182 0.02 0.182 0.05 1.14
Colloidal matter
0.796 0.07 0.883 0.04 0.9
To determine the location and spatial distribution of various types of biopolymers within the ATAD particles, optical sectioning and multiple labelling techniques (Schmid et al., 2003) followed by LSCM were performed. Stabilisation of the ATAD sludge samples was performed by embedding in agarose to maintain the sludge material in its original state, to minimise particle dehydration, and to prevent structural deformation during manipulation and staining. Such techniques have previously been successfully used (Droppo et al., 1996a,b) to examine delicate sludge material and the resulting embedded samples were shown to be highly porous facilitating transport, staining and washing of the particulate material while minimising structural damage (Droppo et al., 1996a, 1996b, 1997; Liss et al., 1996). We observed that upon stabilisation in agarose, the ATAD particles remained as discrete entities and the distance between the particles was maintained. Although some deformation and loss of colloidal material occurred our observations indicated that immobilisation in agarose resulted in good morphological integrity of the ATAD particles with minimal observable disturbance allowing direct, non-invasive, serial optical sectioning by LSCM of intact embedded ATAD sludge giving excellent resolution. Fig. 2 shows a typical specimen from ATAD Reactor 2A (operating range 46e65 C) stained and imaged by LSCM. The bulk water associated with the sample contained emulsified lipid droplets (Fig. 2, Panel B), which may serve as a source of microbial nutrients and most likely inhabited by aerobic organisms with esterase activity (Piterina et al., unpublished). Such thermostable esterases may be a target exo-enzyme for investigation of bulk sludge activity as they are known to play an important role in the biodegradation of many natural substances in wastewater including lipids, proteins and synthetic chemicals (Boczar et al., 2001). Indeed many esterases have high activity at alkaline pH, and possess good thermal stability (Bornscheuer, 2002), conditions that pertain in the ATAD thermophilic reactor (Piterina et al., 2009) where esterase enzymes may be responsible for removal and further solubilisation of lipids from ATAD particulate matter. Numerous bacteria were found associated with the particle surfaces or loosely attached. The interface between the particles and the bulk water also contained a high density of
3431
bacteria, while within the internal structure other bacterial cells could be observed loosely attached to the particles with limited surface contact or freely motile in the bulk water. Small water channels could be observed inside the particles which appeared as a ‘labyrinth’ with several openings. Staining revealed the presence of b-carbohydrates underneath a protein layer with a hydrophobic coat which could provide a surface to allow formation and anchoring of a variety of microbial aggregates.
3.3.
Changes in morphology during the process
As the ATAD treatment process progresses from sludge inlet through the mesophilic reactor, then through the thermophilic reactor and finally to storage the ATAD sludge matrix undergoes major morphological changes in appearance and behaviour. Observations of the microbial cell numbers present at different stages of the ATAD are summarised in Table 2, and indicate that the size and density of biomass undergo dramatic variation during the treatment. The increase in cell numbers observed in Reactor 2A following 20 h of continuous elevated temperature treatment (compared to the numbers determined after 4 h) is indicative of the successful adaptation of the bacterial population to the extreme process conditions of elevated temperature and consequent release and solubilisation of nutrients under elevated conditions. Interestingly bacterial cell numbers observed at the mesophilic stage (Reactor 1A) of ATAD treatment appear to be 3e4 fold larger compared to cell numbers inhabiting the thermophilic stage (Reactor 2A). This may be reflective of lower numbers being able to adapt to the elevated temperature conditions, the difficulties in adapting to the changed physico-chemical conditions such as altered process pH or oxygen solubility and changes in nutrient spectrum within the reactor at later stages in the ATAD process (Piterina et al., 2009). No granular or well-defined flocs similar to those observed in anaerobic sludge (Hulshoff Pol et al., 2004; Gao et al., 2010) were observed in the ATAD sludge at any of the ATAD processing stages. SEM analysis of the sludge revealed, that numerous organic matter particles which originated from the inlet feed sludge become difficult to distinguish and highly masked on the electron microscopy images due to sludge aggregation following conditioning polymer addition, used to condition and thicken the sludge prior to ATAD processing (Fig. 3, Panel B(a)). This however, undergoes extensive transformation at elevated temperature and becomes highly dispersed within the ATAD sludge during subsequent processing (Fig. 3, Panel B(b)) and is further degraded and modified following 9 days of storage in the holding tank post processing (Fig. 3, Panel B (c)). LSCM analysis revealed that dispersed particles are observed at inlet (Fig. 3, Panel C(a),) which contain a hydrophobic surface with a high protein content which then becomes partially degraded and more hydrophilic with a lower protein content as the process proceeds. As degradation proceeds the material becomes exposed to the bulk water and cellulose polymers and lipid droplets are apparent within the sludge liquors at and after the thermophilic treatment (Fig. 3, Panel C (b)).
3432
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 4 2 7 e3 4 3 8
Fig. 2 e Spatial composition and structural features of ATAD thermophilic sludge. The figure illustrates merged LSCM micrographs of 200 Z-stack images illustrating stained ATAD sludge obtained from thermophilic reactor 2A. (A) illustrates the emission spectras for DAPI, Nile Red and FITC which were utilised in the staining procedure. An excitation wavelength range is noted for each of the detection channels on the spectral detection system of the Ziess Meta 710 laser confocal microscope system for each fluorophore used. Collected image were merged into a single combined image. Laser lines applied for excitation of 488, 546 and 405 nm. DNA-DAPI (shown in blue); Hydrophobic moieties-Nile Red (shown in red), Protein-FITC (shown in green). (B) LSCM micrographs taken at different depths of the ATAD particles (0.8 mm and 16 mm) as Z-stack images showing the structural relationships within the stained ATAD sludge samples obtained from thermophilic reactor 2A. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
3.4.
Table 2 e Bacterial enumeration within the ATAD sludge matrix at different stages of ATAD treatment as detected by DAPI staining and epi-fluorescence microscopy. Sample origin Sludge Inlet Reactor 1A (t ¼ 0 h) Reactor 2A (t ¼ 4 h) Reactor 2A (t ¼ 24 h) Product (1 day storage) Product (9 day storage)
Total count of bacteria (per ml of sludge) (7.3 0.4) (9.2 0.1) (6.2 0.2) (5.2 0.12) (3.1 0.1) (6.2 0.3)
108 1013 1010 1012 104 106
Fibrous material degradation
A highly auto-fluoresencent, long fibre-like material was also observed in high quantity (in qualitative terms) in the inlet ATAD sludge and (Fig. 4, Panel A (a)) embedded in the sludge particles. During the ATAD auto-heating process this fibrous material showed little evidence of reduction (Fig. 3, Panel A (b)) and only showed a significant degree of fragmentation after 4 h processing in Reactor 2A (59 C) (Fig. 3, Panel A(c)). Following exposure to elevated temperature in the Reactor 2A (59 C) for up to 23 h this fibrous material underwent dramatic morphological change and appeared as floating or suspended fragments, with decreased fibre length and number (Fig. 3, Panel A (c)). As the ATAD temperature rose in Reactor 2A the fibre morphology changed, and one could observe increased width, the morphology became bulky, swollen and irregular
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 4 2 7 e3 4 3 8
with some degree of agglomeration or gluing occurring (Fig. 3, Panel A(c)), the fibres underwent shortening of length (3e10 mm) and there was observable erosion.
3.5.
Sludge appearance on storage
Following ATAD processing the resultant biosolids are cooled and stored without aeration at ambient temperature. This material was examined following 9 days of storage under such conditions. LSCM images of this material revealed a branched network composed of partially digested cellulose micro fibrils (Fig. 4). Microorganisms were not detected in the bulk water or micro-channels associated with these networks but were observed surrounded via a compact proteinacious material (Fig. 4, b and d). Hydrophobic domains were observed at the interface of the cellulose micro fibrils and at the adhesion points of these microcolonies (Fig. 4, c and d). These however, were not present on the exterior of sludge aggregates which were exposed to the bulk liquor and water channels. The interface of the ATAD sludge particulate matter at this storage stage contained hydrophilic moieties composed of protein and cellulose.
4.
Discussion
There is currently no data on the detailed physical structure and micro-composition of ATAD sludge during or following thermophilic aerobic digestion or indeed how such characteristics influence floc/particle formation or aggregation within ATAD sludges. To address this issue we have carried out a detailed morphological characterisation of ATAD sludge at all stages of the ATAD process on a full scale ATAD plant treating domestic sludge (Piterina et al., 2006, 2009). A number of microscopic techniques ranging from electron, optical, wide field and laser scanning confocal microscopy were applied to fresh, fixed and embedded ATAD sludge taken at various stages during a full scale ATAD process in an attempt to determine key characteristics of the sludge that might be responsible for its poor dewatering and settling. Immediate analysis following sampling by optical microscopy revealed multiple free-swimming microorganisms within the sludge liquor and a dynamic interaction between them and the dispersed organic matter particles present. However, a decrease in temperature of samples (below 35 C) introduced dramatic changes to the sludge morphology which have not been previously described for thermophilic sludge. One would expect that thermophilic microorganisms adapted to the thermophilic stage would undergo dramatic changes upon temperature reduction which would result in changes to the bulk water properties and the particulate morphology. This may be very important post treatment once the treated ATAD sludge finally cools and is stored. The application of electron microscopic techniques to ATAD sludges as demonstrated here is problematic and poses limitations, similar to those reported for examination of liquid or colloidal system (Leppard, 1992). Fixation and dehydration steps associated with the preparation of any biological sample for electron microscopy and the high vacuum environment
3433
within the SEM chamber do not allow maintenance of the original sludge structure which is primarily liquid based making data on morphology problematic. The use of more advanced imaging technique such as LSCM combined with multiple labelling protocols (Lopez et al., 2005) to distinguish between major chemical groups and microorganisms within the particulate structure also has limitations. However we have utilised the embedding method proposed by (Droppo et al., 1996a, 1996b, 1997; Liss et al., 1996) which was originally applied to investigate planktonic organisms and found this extremely useful in overcoming many of the difficulties associated with using advanced microscopic techniques. The technique has been successfully applied to examine floc structure (Liss et al., 2002) to characterise activated sludge floc structure (Schmid et al., 2003), to examine river aggregates (Luef et al., 2009) and has proven useful in case studies, where maintenance of hydration is an essential characteristic for preservation of sample morphology and to generate highquality 3D-images. Application of these methods has allowed us to observe ATAD sludge and generate microcompositional images reproducibly allowing the study of the dynamics of sludge morphology as a function of the stage of the ATAD process. Confocal microscopy offers several advantages over traditional wide-field fluorescence microscopy, including the ability to control depth of field, the elimination or reduction of background from the focal plane which can lead to image degradation, and the ability to collect serial optical sections from thick specimens. Using targeted differential staining, the presence of cellulose material with fibrous and amorphous characteristics was confirmed. Its presence may have its origins in vegetable matter or paper associated with domestic effluent and the abundance of cellulose in the ATAD tertiary treatment stage suggests its low biodegradability and limited removal by prior primary and secondary treatments. During the ATAD process we observed fibres and cellulose particles undergoing swelling and alteration in shape while becoming more dispersed within the ATAD sludge material. There may be several contributing factors associated with the ATAD process including the alkaline conditions resulting from elevated ammonia content which results in chemical alterations to cellulose and fibre which in turn may allow further microbial attack and degradation. In fact such chemical changes have previously been detected by 13C NMR analysis (Piterina et al., 2009). Microscopic observation demonstrated that cellulose material underwent partitioning and fragmentation during the thermophilic stage (Fig. 2(g)) suggesting an active cellulolytic activity at this stage with the fibres appearing to act as bacterial carriers. The close association between microorganisms and cellulose may not only acts as an immobilisation support but also contribute some thermal shielding allowing some less thermotolerent organisms to survive longer as the temperatures in the ATAD thermophilic niche increase. Indeed many cellulolytic organisms contain cell-associated cellulases and their association with such cellulose fibres may improve their thermostability (Pembroke, 1998). We noted a large swelling capacity of cellulose within the ATAD sludge via microscopic examination and this water holding effect may contribute significantly to the poor settling properties of the ATAD biosolids. Indeed much of the fibrous material remains undigested but swollen as it
Fig. 3 e Comparative microphotograph of the transformation and evolution of chemical and structural properties within the ATAD sludge matrix as a function of processing. Panel A. Epi-fluorescence images of ATAD matrix illustrating the pattern of fibre degradation: (a) thickened ATAD sludge (at inlet); (b) sludge from reactor 1A (t [ 16 h of the operation 56 C); and (c) sludge from reactor 2A (t [ 23 h of the operation at 58 C) (d) sludge from the holding tank after 9 days of anaerobic storage without mixing. Sludge samples were fixed and stained with DAPI. Images were captured under UV light at maximum extinction. Two types of auto-fluorescence signal can be observed, firstly bright blue fluorescent material, and secondly a red-orange material (only observed on the inverted image). Panel B. SEM microphotographs of ATAD sludge morphology during the ATAD process. ATAD sludge samples were taken from the inlet feed, the mesophilic reactor 1A, the thermophilic reactor 2A and from the final biosolids product and mounted on poly-L-lysine-coated cover slips prepared as described (Section 2.3). Panel C. LSCM micrographs of single optical sections rendered into 3D-Z-stack images showing the structural relationships within the stained ATAD sludge samples obtained from (a) inlet (b) thermophilic sludge reactor 2A and (c) treated biosolids. An excitation wavelength range is noted for each of detection channel on the spectral detection system of the Ziess Meta 710 laser confocal microscope system for each fluorochrome. Collected image were merged in a single combined image. Laser lines were applied for excitation at 488, 546 and 40 5 nm. DNA-DAPI (shown in blue), CelluloseeCalculofluor white (shown in purple-pink); Hydrophobic moieties-Nile Red (shown in green), Protein-FITC (shown in green). An excitation wavelength range is noted for each of detection channel on the spectral detection system of the Zeiss Meta 710 laser confocal microscope system for each fluorophore; The collected images were merged to a single combined image. Laser lines applied for excitation of 488, 546 and 405,412 nm. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 4 2 7 e3 4 3 8
3435
Fig. 4 e 3D-LSCM micrographs showing the structural relationships within the stained ATAD biosolids product post treatment and following storage. (a) DNA-DAPI (shown in blue) and CelluloseeCalculofluor white (shown in purple-pink); (b) Protein-FITC (shown in green) and DNA-DAPI (shown in blue); (c) Hydrophobic moieties-Nile Red (shown in red) and CelluloseeCalculofluor white (shown in purple-pink); (d) Hydrophobic moieties e Nile Red (shown in red) and CelluloseeCalculofluor white (shown in purple-pink), and Protein-FITC (shown in green) and DNA-DAPI (shown in blue). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
proceeds for storage and may play a major role in water retention post processing (Scallan and Tigerstrom, 1992; Wistara and Young, 1999; Young, 1994). Thus removal of the cellulose via more prolonged processing or by masking exposed groups chemically could play a key role in resolving some of the poor settleability of ATAD sludge. Some of the fibrous material may have its origins from man-made sources such as viscose, nylon or polyesters originating in the primary effluent which are non-degradable during the standard mesophilic biological treatment process as such fibres are somewhat resistant to microbial attack (Kawai, 1995). The strong intermolecular cohesive force caused by hydrogen bonds between molecular chains and the hydrophobic nature of these polymers, coupled to their recalcitrant chemical structure are key factors determining their resistance (Tokiwa et al., 1992). Our data suggests that secondary sludge of domestic origin may be a source of fibrous material that remains undegraded following standard secondary treatment processes. ATAD feed sludge contains a high amount of this fibrous material but in contrast with previous published data for treatment processes operated at mesophilic temperatures (Zubris and Richards, 2005), we observed dramatic abrasion and transformation of the fibre
length and thickness during the thermophilic stage of ATAD processing in Reactor 2A. Although complete degradation does not occur its abrasion may provide significant nutrients for the microbial population operating at these elevated temperatures in Reactor 2a. In fact it has been reported that certain thermophilic Bacillus spp. are capable of degrading aliphatic polyamide Nylon 66 at high temperature and can grow on minimal media supplemented with nylon fibres (Tomita et al., 2003). Indeed preliminary molecular and microbiological studies of ATAD microbial populations (Piterina et al., 2006; LaPara and Alleman, 1999) have indicated that biodegradative Bacillus spp. are the predominant microorganisms active at the aerobic thermophilic stage of the ATAD process (Piterina et al., 2010a,b). As this type of material is resistant to biodegradation by standard mesophilic wastewater treatment processes its presence in the final sludge can cause significant pollution and remains in the environment for a long time as micro-particles and fibres (Habib et al., 1998; Zubris and Richards, 2005; Barnes et al., 2009). The efficiency of removal of such fibre during wastewater treatment has been largely overlooked in the literature and this study points to the potential of ATAD to contribute significantly to modifying and utilising such synthetic fibrous materials not only of
3436
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 4 2 7 e3 4 3 8
domestic origin but potentially for the treatment of textile and plastic polyesters wastes. During this analysis we also evaluated and characterised the colloidal fraction of ATAD sludge and associated EPS material from particulate matter, which are believed to be key factors associated with sludge settleability (Higgins and Novak, 1997; Urbain et al., 1993; Nielsen et al., 1996). The bulk liquid associated with ATAD sludge contained large amounts of carbohydrate, which may be present in two distinctive forms, soluble and colloidal. Microscopy revealed the presence of multiple micro-particles within the liquor with free floating colloidal material. Such colloidal material may also impact on the low settleability of ATAD sludge. Further studies dedicated to size distribution of these particles, their organic content and how bio-separation processes might lead to their removal from the bulk sludge would be of interest. We utilised two methodologies to extract EPS (see Materials and Methods) and each method gave a different spectrum of components. Use of NaOH and formaldehyde yielded more carbohydrate, potentially suggesting that carbohydrate components may cause binding between particulate material and bacteria within the EPS matrix. Use of EDTA in the extraction, as a chelating agent, appeared to dissociate more protein from the sludge matrix, which suggests that divalent cations may play an important role in aggregation and compaction of the sludge matrix within the particulate fraction of ATAD sludge and is again suggestive that EPS may play a role in the poor settleability of ATAD sludge post treatment. Based on our microscopic study of ATAD sludge several ecophysiological niches were observed within the ATAD thermophilic sludge which may support growth and activity of various groups of microorganisms having specific requirements for oxygen and nutrient. Microaerophilic or even anaerobic environments with lower shear stress than in the bulk sludge liquor where extensive mechanical mixing and aeration occur may be found associated with particulate fractions. Such niches may contribute to the development of particulate environments with unique physico-chemical conditions which can support the growth and proliferation of microbial species with particular metabolic capabilities adapted to ATAD conditions. The niches we have observed include a bulk water niche with solute exposed to shear stress and multiple lipid droplet, a floc/particulate matter surface niche exposed to water channels and shear stress and characterised by high protein content and with internal structures provided by cellulose units which may not be exposed to high shear stress produced by the mixing and aeration within the ATAD reactor. LSCM analysis revealed multiple lipid droplets within the ATAD sludge liquor which may present an excellent nutrient source for free-swimming bacteria and indeed may result in diminished agglomeration and flocculation within the ATAD sludge matrix. Removal of these lipid particles could potentially give rise to improved flocculation Following ATAD treatment the exterior of particular matter becomes more hydrophilic, less compact and less dense and this could be a decisive parameter in determining the absence of structural compactness observed for thermophilic sludge (Liu et al., 2004). Sludge with a higher hydrophobicity on the other hand may favour more microbial aggregation and result in
a more compact structure and facilitate flocculation (Liao et al., 2001; Liu et al., 2003; Liao et al. 2002). Following prolonged storage of ATAD sludge the protein domains on the exterior re-appear possibly due to re-growth of mesophilic populations which are producing EPS components and this may again be important in effecting settleability. At this post treatment stage there is a decline in the number of hydrophobic/lipid droplets within the sludge liquor.
5.
Conclusion
ATAD sludge has a number of unique morphological features, and structural characteristics that may be of key importance related to the difficulties of dewatering the stabilised sludge product following processing. It contains a large and dynamic microbial population that aggregate and adhere to the sludge and potentially contribute to aggregation. Because of these dynamic changes, specific pre-microscopy techniques must be applied to maintain the integrity of sludge samples during examination. The high water content of the ATAD sludge samples (up to 98%) can be an obstacle to morphological analysis especially when samples need to be prepared via fixation and dehydration for SEM analysis. Laser Scanning Confocal Microscopy was shown to be the superior approach to study the composition and spatial localisation of structural constituents in such dispersed high water content ATAD sludge, especially when combined with embedding of the samples in agarose prior to analysis. Unique features of the thermophilic ATAD sludge observed here were a low protein/polysaccharide ratio EPS, the presence of numerous cellulose micro-particles and multiple hydrophobic droplets in the bulk liquor of ATAD biosolids. We propose that these play a major role in the poor dewaterability of ATAD sludge. Extracellular bio-colloids are also present in concentrations sufficient to form gel-like structures that significantly restrict water mobility within the sludge whereas hydrophobic droplets may cause repulsion of small floc particles preventing their agglomeration. Studies that focus on chemical and spectral characterisation of ATAD colloidal matter and lipids present in the bulk liquor will be necessary to provide more insight into the molecular mechanisms of poor dewatering of ATAD sludges. However, the current results are suggestive of key factors that may contribute to poor dewaterability and could allow the design of ATAD process modifications that would address, eliminate or reduce these factors to address dewaterability specifically without compromising the process or its environmental safety.
Acknowledgements The authors acknowledge the assistance of staff at the Killarney Wastewater Treatment Facilities, Killarney, Co. Kerry, Ireland during sampling at the ATAD plant. The authors acknowledge financial support via a collaborative research grant from the Higher Education Authority of Ireland, PRTLI 4 scheme and via the a Postdoctoral Fellowship (AP) from the
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 4 2 7 e3 4 3 8
Enterpise Partnership Scheme co-funded by the Irish Research Council for Science, Engineering and Technology (IRCSET) (www.ircset.ie) and HKPB Scientific Ltd (www.hkpb.ie).
references
Agarwal, S., Abu-Orf, M., Novak, J.T., 2005. Sequential polymer dosing for effective dewatering of ATAD sludges. Water Research 39 (7), 1301e1310. Barnes, D.K.A., Galgani, F., Thompson, R.C., Barlaz, M., 2009. Accumulation and fragmentation of plastic debris in global environments. Philosophical Transactions of the Royal Society B 27, 1985e1998. Boczar, B.A., Forney, L.J., Begley, W.M., Larson, R.J., Federle, T.W., 2001. Characterization and distribution of esterase activity in activated sludge. Water Research 35, 4208e4216. Bornscheuer, U.T., 2002. Microbial carboxyl esterases: classification, properties and application in biocatalysis. FEMS Microbiology Reviews 26, 73e81. Buffle, J., Leppard, G.G., 1995. Characterization of aquatic colloids and macromolecules. Key role of physical structures on analytical results. Environmental Science and Technology 29, 2176e2184. Cappuccino, J.G., Sherman, N., 2004. Microbiology, A Laboratory Manual. Pearson Education, Inc., New Delhi, USA. Collins, T.J., 2007. ImageJ for microscopy. BioTechniques 43, 25e33. Darken, M.J., 1961. Applications of fluorescent brighteners in biological techniques. Science 133, 1704e1705. Dignac, M.F., Urbain, V., Rybacki, D., Bruchet, A., Snidaro, D., Scirbe, P., 1998. Chemical description of extracellular polymers: implications on activated sludge floc structure. Water Science and Technology 38, 45. Droppo, I.G., Flannigan, D.T., Leppard, G.G., Jaskot, C., Liss, S.N., 1996a. Floc stabilization for multiple microscopic techniques. Applied Environmental Microbiology 62 (9), 3508e3515. Droppo, I.G., Flannigan, D.T., Leppard, G.G., Liss, S.N., 1996b. Microbial floc stabilization and preparation for structural analysis by correlative microscopy. Water Science and Technology 34 (5e6), 155e162. Droppo, I.G., Leppard, G.G., Flannigan, D.T., Liss, S.N., 1997. The freshwater floc: a functional relationship of water and organic and inorganic floc constituents affecting suspended sediment properties. Water Air and Soil Pollution 99, 43e53. FrØlund, B., Griebe, T., Nielsen, P.H., 1995. Enzymatic activity in the activated-sludge floc matrix. Applied Microbiology and Biotechnology 43, 755e761. Ganczarczyk, J.J., Zahid, W.M., Li, D.H., 1992. Physical stabilization and embedding of microbial aggregates for light microscopy studies. Water Research 26, 1695e1699. Gao, D., Liu, L., Liang, H., Wu, W.M., 2010. Aerobic granular sludge: characterization, mechanism of granulation and application to wastewater treatment. Critical Reviews Biotechnology, 1e16. Gaudy, A.F., 1962. Colorimetric determination of protein and carbohydrate. Industrial Water Wastes 7, 17. Greenspan, P., Fowler, S.D., 1985. Spectrofluorometric studies of the lipid probe, Nile Red. Journal of Lipid Research 26, 781e789. Greenspan, P., Mayer, E.P., Fowler, S.D., 1985. Nile red: a selective fluorescent stain for intracellular lipid droplets. Journal of Cellular Biology 100 (3), 965e973. Grijspeerdt, K., Verstraete, W., 1997. Image analysis to estimate the settleability and concentration of activated sludge. Water Research 31, 1126e1134. Habib, D., Locke, D.C., Cannone, L.J., 1998. Synthetic fibers as indicators of municipal sewage sludge, sludge products, and
3437
sewage treatment plant effluents. Water, Air, & Soil Pollution 103, 1e8. Harrington, B.J., Raper, K.B., 1968. Use of a fluorescent brightener to demonstrate cellulose in the cellular slime molds. Applied Environmental Microbiology 16 (1), 106e113. Haugland, R.P., 2002. Handbook of fluorescent probes and research products. In: Gregory, J., Spence, M.T.Z. (Eds.), Molecular Probes, (9 Ed), Vol. 1. Herwijn, A.J.M., 1996. Fundamental aspects of sludge characterization. PhD Thesis. Eindhoven University, The Netherlands. Higgins, M.J., Novak, J.T., 1997. Characterization of exocellular protein and its role in bioflocculation. Journal of Environmental Engineering 123, 479e485. Hughes, J., McCully, M.E., 1975. The use of an optical brightener in the study of plant structure. Stain Technology 50, 319e329. Hulshoff Pol, L.W., de Castro Lopes, S.I., Lettinga, G., Lens, P.N., 2004. Anaerobic sludge granulation. Water Research 38 (6), 1376e1389. Kawai, F., 1995. Breakdown of plastics and polymers by microorganisms. Advances in Biochemical Engineering Biotechnology 52, 151e194. Kelly, G., Melcer, H., Mavinic, D.S., 1993. Autothermal thermophilic aerobic digestion of municipal sludges: a oneyear, full-scale demonstration project. Water Environmental Research 65, 849e861. Kiernan, JA., 2000. Formaldehyde, formalin, paraformaldehyde and glutaraldehyde: what they are and what they do. Microscopy Today 1, 8e12. LaPara, T.M., Alleman, J.E., 1999. Thermophilic aerobic biological wastewater treatment. Water Research 33 (4), 895e908. Layden, N.M., 2007. An evaluation of autothermal thermophilic aerobic digestion (ATAD) of municipal sludge in Ireland. Journal of Environmental Engineering and Science. Layden, N.M., Kelly, H.G., Mavinic, D.S., Moles, R., Bartlett, J., 2007a. Autothermal thermophilic aerobic digestion (ATAD) part I: review of origins, design, and process operation. Journal of Environmental Engineering and Science 6, 665e678. Layden, N.M., Kelly, H.G., Mavinic, D.S., Moles, R., Bartlett, J., 2007b. Autothermal thermophilic aerobic digestion (ATAD) part II: review of research and full-scale operating experiences. Journal of Environmental Engineering and Science 6, 679e690. Lee, W., Kang, S., Shin, H., 2003. Sludge characteristic and their contribution to microfiltration in submerged membrane bioreactors. Journal of Membrane Science 216, 217e227. Leppard, G.G., 1992. Evaluation of electron microscope techniques for the description of aquatic colloids. In: Buffle, J., van Leeuwen, H.P. (Eds.), Environmental Particles, vol. 1, pp. 231e289. Lewis Publishers, Chelsea, UK. Leppard, G.G., 1993. Organic flocs in surface waters: their native state and aggregation behaviour in relation to contaminant dispersion. In: Rao, S.S. (Ed.), Particulate Matter and Aquatic Contaminants. Lewis Publishers, Boca Raton, Florida, USA, pp. 169e195. Li, D.H., Ganczarczyk, J.J., 1990. Structure of activated sludge flocs. Biotechnology and BioEngineering 35, 57e65. Liao, B.Q., Allen, D.G., Droppo, I.G., Leppard, G.G., Liss, S.N., 2000. Bound water content of activated sludge and its relationship to solids retention time, floc structure, and surface properties. Water Environmental Research 72 (6), 722e730. Liao, B.Q., Allen, D.G., Droppo, I.G., Leppard, G.G., Liss, S.N., 2001. Surface properties of sludge and their role in bioflocculation and settleability. Water Research 35, 339e350. Liao, B.Q., Allen, D.G., Droppo, I.G., Leppard, G.G., Liss, S.N., 2002. Interparticle interactions affecting the stability of sludge flocs. Journal of Colloid and Interface Science 249, 372e380. Liss, S.N., Droppo, I.G., Flannigan, D.T., Leppard, G.G., 1996. Flocs architecture in wastewater and natural riverine systems. Environmental Science and Technology 30 (2), 680e686.
3438
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 4 2 7 e3 4 3 8
Liss, S.N., Liao, B.Q., Droppo, I.G., Allen, D.G., Leppard, G., 2002. Effect of solid retention time of floc structure. Water Science and Technology 46 (1e2), 431e438. Liu, H., Fang, H.H.P., 2002. Extraction of extracellular polymeric substances (EPS) of sludges. Journal of Biotechnology 95 (3), 249e256. Liu, Y., Yang, S.F., Liu, Q.S., Tay, J.H., 2003. The role of cell hydrophobicity in the formation of aerobic granules. Current Microbiology 46, 270e274. Liu, Y., Yang, S.F., Qin, L., Tay, J.H., 2004. A thermodynamic interpretation of cell hydrophobicity in aerobic granulation. Applied Microbiology and Biotechnology 64 (3), 410e415. Lopez, C., Pons, M.N., Morgenroth, E., 2005. Evaluation of microscopic techniques (epifluorescence microscopy, CLSM, TPE-LSM) as a basis for the quantitative image analysis of activated sludge. Water Research 39, 456e468. Lowry, O.H., Rosebrough, N.J., Farr, A.L., Randall, R.J., 1951. Protein measurement with the folin phenol reagent. Journal of Biological Chemistry 193, 265e275. Luef, B., Neu, T.R., Zweimuller, I., 2009. Structure and Composition of aggregates in two large European rivers based on confocal laser scanning microscopy and image and statistical analyses. Applied Environmental Microbiology 75 (18), 5952e5962. Mikkelsen, L.H., Keiding, K., 2002. Physico-chemical characteristics of full scale sewage sludges with implication to dewatering. Water Research 36, 2451e2462. Mohaibes, M., Heinonen-Tanski, H., 2004. Aerobic thermophilic treatment of farm slurry and food wastes. Bioresource Technology 95, 245e254. Morgan-Sagastume, F., Allen, D.G., 2005. Physicochemical properties and stability of activated sludge flocs under temperature upshifts from 30 to 45 degrees C. Journal of Colloid and Interface Science 281 (1), 136e145. Murthy, S.N., Novak, J.T., Holbrook, R.D., 2000. Optimizing dewatering of biosolids from autothermal thermophilic aerobic digesters (ATAD) using inorganic conditioners. Water Environmental Research 72, 714e721. Nielsen, P.H., FrØlund, B., Keiding, K., 1996. Changes in the composition of extracellular polymeric substances in activated sludge during anaerobic storage. Applied Microbiology and Biotechnology 44, 823. O’Brian, T.P., Feder, N., McCully, M.E., 1964. Polychromatic staining of plant cell walls by toluidine blue O. Protoplasma 59, 368e373. Pembroke, J.T., 1998. Function and biotechnology of cellulose binding domains in bacterial and fungal cellulose and xylan hydrolysing enzymes. Recent Research Developments in Biotechnology and Bioengineering 1 (2), 177e190. Piterina, A.V., MacCausland, C., Bartlett, J., Pembroke, J.T., 2006. Microbial ecology of autothermal aerobic digestion (ATAD): diversity, dynamics and activity of bacterial communities involved in treatment of municipal wastewater. In: MendezVilas, A. (Ed.), Modern Multidisciplinary Applied Microbiology, Exploiting Microbes and Their Interactions. Wiley-VCH, Weinheim, Germany, pp. 526e535. Piterina, A.V., Barlett, J., Pembroke, J.T., 2009. 13C-NMR assessment of the pattern of organic matter transformation during domestic wastewater treatment by autothermal aerobic digestion (ATAD). International Journal of Environmental Research and Public Health 6, 2288e2306.
Piterina, A.V., Bartlett, J., Pembroke, J.T., 2010a. Evaluation of the removal of indicator bacteria from domestic sludge processed by autothermal thermophilic aerobic digestion (ATAD). International Journal of Environmental Research and Public Health 7, 3422e3441. Piterina, A.V., Bartlett, J., Pembroke, J.T., 2010b. Molecular analysis of bacterial community DNA in sludge undergoing autothermal thermophilic aerobic digestion (ATAD): pitfalls and improved methodology to enhance diversity recovery. Diversity 2, 505e526. Prasad, V., Semwogerere, D., Weeks, E.R., 2007. Confocal microscopy of colloids. Journal of Physics: Condensed Matter 19 (11), 113102. Scallan, A.M., Tigerstrom, A.C., 1992. Swelling and elasticity of the cell walls of pulp fibers. Journal of Pulp and Paper Science 18 (5), 188e193. Schmid, M., Thill, A., Purkhold, U., Walcher, M., Bottero, J.Y., Ginestet, P., Nielsen, P.H., Wuertz, S., Wagner, M., 2003. Characterization of activated sludge flocs by confocal laser scanning microscopy and image analysis. Water Research 37 (9), 2043e2052. Stahl, D.A., Devereux, R., Amann, R.I., Flesher, B., Lin, Stahl C., Amann, D.A., 1991. Development and application of nucleic acid probes in bacterial systematics. In: Stackenbrandt, E., Goodfellow, M. (Eds.), Sequencing and Hybridixation Techniques in Bacterial Systematics. John Wiley, Chichester, pp. 205e208. Tokiwa, Y., Iwamoto, A., Koyama, M., Kataoka, N., Nishida, H., 1992. Biological recycling of plastics containing ester bonds. Macromolecular Chemistry Macromolecules Symposium 57, 273e279. Tomita, K., Ikeda, N., Ueno, A., 2003. Isolation and characterization of a thermophilic bacterium, Geobacillus thermocatenulatus, degrading nylon 12 and nylon 66. Biotechnology Letters 25, 1743e1746. US EPA, 1994. A Plain English Guide to the EPA Part 503 Biosolids Rule. EPA/832/R-93/003. U.S. Environmental Protection Agency, Washington, D.C, USA. Urbain, V., Block, J.C., Manem, J., 1993. Bioflocculation in activated sludge: an analytic approach. Water Research 27, 829. Wei, Y., Dodds, W.K., Banks, M.K., Skalsky, J., Strauss, E.A., 1995. Optimal staining and sample storage time for direct microscopic enumeration of total and active bacteria in soil with two fluorescent dyes. Applied and Environmental Microbiology 61 (9), 3367e3372. Wistara, N., Young, R.A., 1999. Properties and treatments of pulps from recycled paper. Part I. Physical and chemical properties of pulps. Cellulose 6 (4), 291e324. Young, R.A., 1994. Comparison of the properties of chemical cellulose pulps. Cellulose 1 (2), 107e130. Yu, G.H., He, P.J., Shao, L.M., He, P.P., 2008. Stratification structure of sludge flocs with implications to dewaterability. Environmental Science and Technology 42 (21), 7944e7949. Zhou, J., Mavinic, D.S., Kelly, H.G., Ramey, W.D., 2002. Effects of temperatures and extracellular proteins on dewaterability of thermophilically digested biosolids. Journal of Environmental Engineering and Science 1 (6), 409e415. Zubris, K.A., Richards, B.K., 2005. Synthetic fibres as an indicator of land application of sludge. Environmental Pollution 138 (2), 201e211.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 4 3 9 e3 4 5 2
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Study on two operating conditions of a full-scale oxidation ditch for optimization of energy consumption and effluent quality by using CFD model Yin Yang a, Jiakuan Yang a,*, Jiaolan Zuo a, Ye Li b, Shu He b, Xiao Yang a, Kai Zhang a a b
School of Environmental Science and Engineering, Huazhong University of Science and Technology, Wuhan, Hubei 430074, PR China Universtar Science & Technology (Shenzhen) Co., Ltd., Shenzhen, Guangdong 518057, PR China
article info
abstract
Article history:
The operating condition of an oxidation ditch (OD) has significant impact on energy
Received 30 August 2010
consumption and effluent quality of wastewater treatment plants (WWTPs). An experi-
Received in revised form
mentally validated numerical tool, based on computational fluid dynamics (CFD) model,
31 March 2011
was proposed to optimize the operating condition by considering two important factors:
Accepted 1 April 2011
flow field and dissolved oxygen (DO) concentration profiles. The model is capable of pre-
Available online 13 April 2011
dicting flow pattern and oxygen mass transfer characteristics in ODs equipped with surface aerators and submerged impellers. Performance demonstration and comparison of
Keywords:
two operating conditions (existing and improved) were carried out in two full-scale
CFD
Carrousel ODs at the Ping Dingshan WWTP in Henan, China. A moving wall model and
Energy consumption
a fan model were designed to simulate surface aerators and submerged impellers,
Effluent quality
respectively. Oxygen mass transfer in the ditch was predicted by using a unit analysis
Operating condition
method. In aeration zones, the mass inlets representing the surface aerators were set as
Oxidation ditch
one source of DO. In the whole straight channel, the oxygen consumption was modeled by using modified BOD-DO model. The following results were obtained: (1) the CFD model characterized flow pattern and DO concentration profiles in the full-scale OD. The predicted flow field values were within 1.98 4.28% difference from the actual measured values while the predicted DO concentration values were within 4.71 4.15% of the measured ones, (2) a surface aerator should be relocated to around 15 m from the curve bend entrance to reduce energy loss caused by fierce collisions at the wall of the curve bend, and (3) DO concentration gradients in the OD under the improved operating condition were more favorable for occurrence of simultaneous nitrification and denitrification (SND). ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
Oxidation ditch (OD) has been widely used as a modified activated sludge biological treatment process due to its reliability, simplicity of operation and low sludge production (U.S.EPA, 1992; Grady et al., 1999). However, successful
operation of an OD system may still be challenging for many wastewater treatment plants (WWTPs) for economical and technical reasons. The new and stringent effluent discharge requirements for municipal WWTPs in China, imposed by Ministry of Environmental Protection (Ministry of Environmental Protection of China, 2002), are likely to incur
* Corresponding author. Tel.: þ86 27 87792207; fax: þ86 27 87792101. E-mail addresses:
[email protected],
[email protected] (J. Yang). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.04.007
3440
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 4 3 9 e3 4 5 2
Nomenclature BOD5 CFD COD ck Cm C13 C23 DO F G Jk k Lk NH3-N OD P P0 P1 P10 r
5 day biological oxygen demand, mg/l computational fluid dynamics chemical oxygen demand, mg/l local instantaneous scalar mass in phase k, kg/m3 empirical constant of the k-3 model empirical constant of the k-3 model empirical constant of the k-3 model dissolved oxygen, mg O2/l axial thrust of the submerged impeller production term of turbulent energy by the mean velocity gradients, kg/m s3 flux due to molecular diffusion, mol/s m2 average turbulent kinetic energy per unit mass, m2/s2 interfacial transfer of mass between two phases, kg/m3 s ammonia-nitrogen, mg N/l oxidation ditch pressure, Pascal pressure of the flow at the surface AA0 , CC0 pressure of the flow in front of the surface BB0 pressure of the flow behind the surface BB0 radius of circle, m
higher operation costs and hefty penalties for noncompliance. Therefore, upgrading of existing OD systems to comply with new effluent standards and reduce operation costs is warranted for many WWTPs in China. Since the process efficiency depends heavily on the flow field in an OD, a good understanding of hydrodynamics of the ditch is needed for a successful design. However, due to its complexity, hydraulics of an OD is often poorly understood. An assumption of an ideal flow pattern, ignoring the real hydrodynamic characteristics of the ditch, was often made in design (Stamou, 1994, 2008; Luo et al., 2005; Littleton et al., 2007a). With improvement of computational power and availability of computational fluid dynamics (CFD) codes, flow pattern in ODs have been studied by using 1D models or models with hydrodynamic effects in recent years (Stamou, 1994, 1997; Clercq et al., 1999; Lesage et al., 2003). Stamou (1993) applied a 2D standard k-3 turbulence model in an OD simulation and Luo et al. (2005) used a 3D k-3 turbulence model to simulate the flow field in an integrative OD aerated with one set of brush aerator. However, ODs of actual WWTPs often have more than six or seven sets of surface aerators combined with submerged impellers in operation simultaneously to achieve the required flow velocities and process efficiency. They present challenges to modeling and simulation runs. To alleviate the problem, a moving wall model was designed to simulate many sets of surface aerators. A moving zone was set up for one set of surface aerator which contains 45 groups of rotation discs along the rotating shaft. Each rotation disc served as moving wall, which was the main source of the drag force. The fluid in each moving zone got the velocity and momentum from the moving walls firstly, and then the velocity and momentum was passed to the rest fluid in the
R SND t T TN Ui vk VA VA þ U1 VA þ U2 W WWTP Xi z
radius of cylinder, m simultaneous nitrification and denitrification time, s temperature of the mixed liquor, C total nitrogen, mg N/l velocity component in i direction (i ¼ 1, 2 and 3), m/s local instantaneous phase velocity of phase k, m/s velocity of the flow at the surface AA0 velocity of the flow at the surface BB0 velocity of the flow at the surface CC0 width of cylinder, m wastewater treatment plant coordinate component in i direction (i ¼ 1, 2 and 3), m vertical distance from the water surface, m
Greek letters retention of phase k, dimensionless ak 3 dissipation rate of the turbulent kinetic energy, m2/s3 gt eddy viscosity, kg/m s h transmission efficiency of the submerged impeller r density of water, kg/m3 sk turbulent Schmidt number for k turbulent Schmidt number for 3 s3
oxidation ditch following the continuity and momentum equations (Yang et al., 2010). Most of the above-mentioned literatures were focused on calibrating and evaluating the CFD model in order to simulate and analyze flow field in OD systems. However it has rarely been reported before that a calibrated CFD model was applied to compare and analyze the flow fields under different operating conditions of an OD system. Therefore, based on our previous work (Yang et al., 2010), an experimentally validated CFD model was used in this study to compare and analyze the flow fields under two operating conditions (existing and improved) of an OD system. Furthermore, except for surface aerator, submerged impeller in OD is another major energy consumer. However, submerged impellers were rarely investigated in CFD simulation in previous literatures. Thus, submerged impellers will be modeled to optimize energy consumption in this study. On the other hand, aeration in a biological treatment process can account for up to 70% of total energy consumption of a WWTP. Optimizing aeration can reduce energy consumption and operation costs as well as guarantee a reliable and efficient treatment in such installations (Fayolle et al., 2007). Oxygen mass transfer models developed by Gillot and Heduit (2000) showed a relationship between gas and liquid velocities and the mass transfer parameters. An experimentally validated numerical tool, based on CFD, was used by Fayolle et al. (2007) to predict flow and oxygen mass transfer characteristics in aeration tanks equipped with fine bubble diffusers and axial slow speed mixers. Several control strategies were focused on control of fine bubble aeration systems, based on regulating their air supply, to enhance nitrogen removal (Fiter et al., 2003; Insel et al., 2005). However, these control strategies and aeration systems are not readily
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 4 3 9 e3 4 5 2
applicable to many existing WWTPs in China, especially the Carrousel ODs with many sets of surface aerators and submerged impellers operating simultaneously. Furthermore, many previous studies were focused on reduction of organic and nutrient concentrations by optimizing aeration rather than on reduction of energy consumption and the associated costs in WWTPs (Littleton et al., 2007b; Blackburne et al., 2008; Liu et al., 2010). In the present paper, both economical aspects and effluent discharge requirements were considered in optimization of dissolved oxygen (DO) concentration profiles by improving operating conditions of surface aerators and submerged impellers. The objective of this paper was to present an experimentally validated numerical tool that can be used to optimize energy consumption without compromising effluent quality by improving operating conditions of full-scale ODs. A CFD model capable of predicting flow pattern and oxygen mass transfer characteristics was developed for the Carrousel ODs of the Ping Dingshan WWTP in Ping Dingshan City in Henan Province of China. Two operating conditions, existing and improved, were compared and analyzed by considering two important factors: a) flow field in the ditch that has close relationship with energy consumption and process efficiency and b) DO concentration profiles that influence simultaneous nitrification and denitrification (SND) in the ditch. Full-scale demonstration of system performance under these two operating conditions was carried out and effects of these two operating conditions on effluent quality and energy consumption were compared. Based on the results of this research, several suggestions were proposed to improve system performance of ODs. The results of the full-scale demonstration showed that the improved operating condition could reduce energy consumption and satisfy effluent standards. In this paper, the fluctuations of the influent flow rate, concentration and composition were not considered, so a simplified steady-state model was investigated.
2.
Materials and methods
2.1.
Experimental set up
3441
There are two Carrousel ODs at the Ping Dingshan WWTP. Both of them have the same configurations and the wastewater flow rate to each ditch is 50,000 m3/d. Only one OD was chosen for this simulation study. The full-scale bioreactor is a four-channel circular ditch, with a total working volume of 26,000 m3. The diagram of the ditch is shown in Fig. 1. Each channel is 10 m wide, with a water depth of 4 m. The straight portion of each channel is 130 m in length, and the radius of the big semi-circle is 20.4 m and the radius of the small semicircle is 10.15 m. Horizontal flow velocities in the straight and the semicircle portions of the full-scale OD were measured by an intelligent current meter (type: XZ-3; manufactured by Nanjing Automation Institute of Water Resources and Hydrology, Ministry of Water Resources of China; measurement range: 0.010e10.000 m/s). On-line measurements of DO were done by DO meters (type: EndressþHauser; made in Germany; measurement range: 0.00e20.00 mg/l; temperature range: 10 to 60 C), and the DO levels were controlled by manual adjustment of the operating conditions of the aerators. Measurements of the velocity and DO concentration were taken at several sampling locations and different water depths: 0.5 m below the water surface (the Surface layer), 1 m below the water surface (the Top layer), 2 m below the water surface (the Middle layer), and 1 m from the bottom of the ditch (the Bottom layer). The locations of sampling points are shown in Fig. 2. There are thirteen sets of surface aerators and nine sets of submerged impellers in the bioreactor. Each set of aerators consists of 45 discs (1.4 m in diameter) and part of each disc (about 0.5 m) is submerged. Two operating conditions of the
Fig. 1 e Diagram of the oxidation ditch (unit: m).
3442
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 4 3 9 e3 4 5 2
Fig. 2 e Schematic diagram of sampling locations (unit: m).
OD were compared and analyzed: one was the existing operating condition with nine surface aerators and three submerged impellers operating simultaneously, the other was the improved operating condition with six surface aerators and six submerged impellers operating simultaneously, as shown in Fig. 3.
2.2.
Mathematical modeling
Flow field and oxygen mass transfer computations were determined with the 6.3 version of a commercial CFD code FLUENT (ANSYS, Inc., Pennsylvania, U.S.A.). The CFD model typically includes a description of flow geometry, a set of differential equations describing the physics and chemistry of the flow, boundary and initial conditions, and mesh points at which these equations are solved (Anderson, 1995; Warsi, 1998). The software is based on the fundamental equations of fluid dynamics, i.e. the continuity, momentum, and energy equations. These governing equations are programmed in FLUENT for conservation of mass, momentum and energy, using the NaviereStokes equations (Littleton et al., 2007a).
2.2.1.
Flow field modeling
The Reynolds-averaged, NaviereStokes equations that govern the 3D, steady-state, incompressible flow in the Carrousel OD as shown in Fig. 1, are illustrated as follows:Continuity equation: vUi ¼ 0: vXi
(1)
Momentum equation: v 1 vP v vUi vUj gt ; þ þ Ui Uj ¼ vXj vXi vXj r vXi vXj
(2)
where P is the pressure, Ui is the velocity component in i direction, Xi is the coordinate component in i direction, gt is the eddy viscosity, r is the density of water, and the subscripts i, j ¼ 1,2,3. The gt in the above equation is determined by the k-3 turbulence model (Rodi, 1980): gt ¼ Cm
k2 ; 3
(3)
where Cm is an empirical constant of the k-3 model, gt is related to the turbulent kinetic energy, k, and the dissipation rate, 3. Here, some simplifications were made. The fluctuations of the influent flow rate, concentration and composition were not considered, so a simplified steady-state model was investigated. In addition, the flow was considered as incompressible flow, thus the temporal and spatial variations of density were neglected. Values of k and 3 are calculated from the following semiempirical transport equations: k equation: Ui
vk v gt vk þ G 3: ¼ vXi vXi sk vXi
(4)
3 equation: Ui
v3 v gt v3 3 32 þ C13 G C23 ; ¼ k vXi vXi s3 vXi k
(5)
where C13 and C23 are empirical constants of the k-3 model, sk is turbulent Schmidt number for k, s3 is turbulent Schmidt number for 3, and G is the production term of turbulent energy by the mean velocity gradients: G ¼ gt
vUi vUj vUi þ : vXj vXi vXj
(6)
Since the flow pattern was considered as turbulence in this work, the kinetic coefficient k and energy constant 3 could be introduced using time-average NaviereStokes equations, known as Reynolds-averaged equations, to solve the continuity equation and the momentum equations (Littleton et al., 2007a). Default values of the k-3 turbulence model constants were used (Launder and Spalding, 1972).
2.2.2.
Oxygen mass transfer modeling
A unit analysis method was created to describe DO concentration profiles in a full-scale OD. Taking a straight channel of the ditch as an example (Fig. 4a), there are aerobic zones and anoxic zones in the channel where DO concentration profiles could exist. As illustrated in Fig. 4b, the channel was evenly divided into 13 sections in length and 4 sections in width and in depth, thus a total of 208 units were obtained. Each unit was
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 4 3 9 e3 4 5 2
3443
Fig. 3 e Schematic diagram of two operating conditions ((a) existing operating condition, (b) improved operating condition).
assumed as a completely mixed reaction unit, so DO concentration in the center of one unit could represent the concentration of that unit. Under the existing operating condition, two surface aerators in the channel were in operation simultaneously and under the improved operating condition only one surface aerator in the front of the channel was in operation. By using the unit analysis method, DO concentration profiles in the channel were predicted and analyzed. The mass transport equation is expressed as / vak ck / ! þ V $ðak ck ! v k Þ ¼ V $ ak J k þ c0k v0k þ Lk ; vt
(7)
where Lk represents the interfacial transfer of mass, Jk is the flux due to the molecular diffusion, ak represents retention of phase k, ck represents local instantaneous scalar mass in phase k, vk represents local instantaneous phase velocity of
phase k and c0k v0k represents the turbulent diffusion of the mass (Fayolle et al., 2007). Oxygen mass transfer terms were added to the CFD model to determine if the DO concentration profiles could occur in the channel. In this study, simulation of DO included two parts: the mass inlets from surface aerators and oxygen consumption of biochemical reaction. As suggested by Littleton et al. (2007a), the system geometry of all 208 units was defined as the source region (zones containing the aeration discs) and the tank region (the rest of tank with the exclusion of the aeration disc zones). In the source region containing the aeration discs, the mass inlets representing the surface aerators in operation were set as one source of DO. In the whole straight channel including both the source region and the tank region, the oxygen consumption was modeled by using modified BOD-DO model. BOD-DO model was widely used in the modeling of organic
3444
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 4 3 9 e3 4 5 2
Fig. 4 e Set up of unit analysis method ((a) diagram of a typical straight channel, (b) partition of a typical straight channel in (i) horizontal section; (ii) vertical section (unit: m)).
pollutant and DO in the riverway. The whole straight channel was 130 m long and 10 m wide, which was comparable with traditional riverway. Therefore, BOD-DO model could be used to simulate the oxygen consumption in the whole straight channel. Here, an assumption was made, i.e. the oxygen consumption was expressed as sink term of BOD biochemical reaction in BOD-DO model. The FLUENT software itself has Transport & Reaction model. In order to adapt BOD-DO model to the OD, the oxygen source, sink terms, and momentum in BOD-DO model were added to FLUENT by using the C Language subroutine, called User Defined Function (UDF). Empirical values were used as the initial coefficients of the model, and then the coefficients were adjusted and validated by field data. Furthermore, the gas-phase dispersion transported by the turbulent fluid motion was taken into account in FLUENT by activating the drift velocity model. Drift velocity plays a vital part in simulation of gas hold-up (Talvy et al., 2005).
2.2.3.
Computational domain
The computed subject was a Carrousel OD in operation with a working volume of 26,000 m3. Stamou (1993) found that the inlet and outlet had little effect on the flow field in an OD. Therefore, the effects of inlet and outlet were not considered in the computation. There are thirteen sets of disc aerators and nine sets of submerged impellers in the ditch. The aerators and impellers are the main sources that drive the flow in the ditch. In addition, they keep the solids suspended and provide aeration for the biological process. The computational domain was set up by using the pre-processor GAMBIT (Yang et al., 2010).
2.2.4.
Simulation of disc aerators
There are 45 groups of rotation discs along the rotating shaft for each of the 13 disc aerators. If the model of Luo et al. (2005) was used in this study, a huge number of grids would increase
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 4 3 9 e3 4 5 2
Fig. 5 e Schematic diagram of the fan model.
computational complexity for simulation. To alleviate the problem, a moving wall model was used (Yang et al., 2010). A complete moving zone was formed by the sidewalls, the outer wall, and the surface. Each rotation disc served as a moving wall, which was the main source of the drag force. The fluid in each moving zone got the velocity and momentum from the moving walls first, and then the velocity and momentum were passed to the rest of the fluid in the OD. In this case, the total number of grids significantly decreased and the calculation task became acceptable to a personal computer.
2.2.5.
Simulation of submerged impellers
There are nine sets of submerged impellers in the ditch, and the radius of each impeller is 1 m. Apart from producing axial flow velocity, the rotating impeller would also create eddy currents around itself. The eddy current would affect the distribution of flow field near the impeller. However, its effect was small when compared to the whole flow field in the ditch,
3445
so it was ignored in the simulation. The main function of the impeller is to pass energy to the fluid to obtain an axial flow velocity. A fan model was adopted for simulation of the submerged impellers, in which each submerged impeller was assumed as an infinitely thin circle and the pressure difference across the impeller was calculated by the area of the disc, density of the fluid and the average flow velocity in the ditch. As shown in Fig. 5, the submerged impeller was assumed as an infinitely thin circle BB0 . Suppose the velocity and pressure of the flow at AA0 in front of the submerged impeller were VA and P0, respectively. When the flow passed the surface BB0 , the velocity changed to VA þ U1. The pressure in front of the surface BB0 was P1 and the pressure behind BB0 was P10 . When the flow arrived at the surface CC0 behind the submerged impeller, the velocity and pressure were VA þ U2 and P0, respectively. According to Bernoulli equation, we can have: 1 1 P0 þ rVA2 ¼ P1 þ rðVA þ U1 Þ2 2 2
(8)
1 1 P0 þ rðVA þ U2 Þ2 ¼ P01 þ rðVA þ U1 Þ2 2 2
(9)
where r is the density of water. Suppose the axial thrust produced by the pressure difference of the submerged impeller was F, then we can have F ¼ hDPA0 ¼ hðP01 P1 ÞA0
(10)
where A0 is the area of the surface, h is the transmission efficiency of the submerged impeller which is 0.5 according to the data of the submerged impeller manufacturer.
Fig. 6 e The fan model for a submerged impeller.
3446
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 4 3 9 e3 4 5 2
Table 1 e The size information of grids for the studied ditch. Ditch Volume (m3) Number of Number of Number of cells faces nodes 1
26,000
1,657,499
3,460,355
344,687
A cylindrical moving zone with 0.5 m in width and 1.5 m in radius was created for each impeller, as shown in Fig. 6. The surface of the submerged impeller was defined as pressure inlet representing the axial pressure difference. The fluid in the moving zone obtained momentum and energy by axial thrust and the momentum and energy passed to the rest of the fluid in the ditch following the momentum and energy equations.
2.2.6.
Boundary conditions
The slip wall boundary condition was assigned at the water surface, and the rigid-lid assumption was adopted. The rotation disc in operation was set as the moving wall, and then the
rotational speed, rotation-axis origin and direction of the moving wall were set, respectively. The roughness constant and the roughness height of the moving wall were set as 1 and 0.02 m, respectively. The no-slip boundary conditions were assigned for all the other walls, including the bottom surface, the side and central walls of the ditch. The fan model was used for the submerged impeller.
2.2.7.
Mesh grid
For the studied ditch, 3D grid meshes were created by using the pre-processor GAMBIT. Owing to many boundaries and irregular shape of the studied ditch, unstructured meshes were used. The size information of the resulted grids is summarized in Table 1. The grids were refined near the moving zones of the rotation disc and impeller for more precise simulation. The grid size was chosen to be small enough to fulfill a grid-independent solution. The calculation time is approximately three days for a steady-state calculation (on an Intel Core computer, with a 2.93 GHz i3 bi-processor and 2 GB RAM).
Fig. 7 e Velocity profiles in the horizontal section of the ditch under two operating conditions ((a) existing operating condition, (b) improved operating condition).
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 4 3 9 e3 4 5 2
Fig. 8 e Flow velocity (m/s) versus distance from the internal wall (m) ((a) existing operating condition, (b) improved operating condition).
3447
3448
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 4 3 9 e3 4 5 2
3.
Results and discussion
3.1.
Flow field
3.1.1.
Flow velocity profiles
Fig. 7 shows the simulated velocity contours at a horizontal cross-section 1 m from the bottom of the ditch (the Bottom layer) under two operating conditions (existing and improved). The CFD model indicates that a heterogeneous flow pattern can develop within the ditch. As shown in Fig. 7, distributions of flow field under two operating conditions were different, especially in the curve bends of the ditch. Under the existing operating condition, the surface aerator at the entrance of the curve bend was in operation, and the flow velocity near the external wall was obviously higher than that near the internal wall. Under the improved operating condition, that surface aerator was not in operation, and instead, the submerged impeller in the curve bend was in operation to avoid sludge sedimentation. Consequently, the flow velocities were more homogeneous than those under the existing operating condition.
3.1.2.
Comparison of CFD results and measured data
The flow velocities were measured in Test Location 1 in the curve bend of the ditch (Fig. 1) using an intelligent current meter. Fig. 8 plots the flow velocities with distances from the internal wall under two operating conditions. Velocity measurements were made at Location 1 at four distances from
the water surface: 0.5 m (the Surface layer), 1 m (the Top layer), 2 m (the Middle layer) and 3 m (the Bottom layer). As shown in Fig. 8a, the flow velocities under the existing operating condition increased from 0.280 to 0.562 m/s (the Surface layer) and from 0.264 to 0.511 m/s (the Top layer) as the distance from the internal wall increased. For the Middle and Bottom layers, the measured flow velocities increased from 0.175 to 0.388 m/s and from 0.051 to 0.152 m/s, respectively. The predicted velocity values were within 4.10 5.19% difference from the measured velocity values. The finding of higher velocities near the external wall in the curve bend than those near the internal wall is consistent with the velocity profiles of the horizontal section of the ditch shown in Fig. 7a. This phenomenon was readily understandable because the disc aerator at the entrance of the curve bend was in operation. Due to inertia force and centrifugal force, the fluid was dragged towards the external wall and energy loss was caused by fierce collisions that occurred at the wall of the curve bend. As shown in Fig. 8b, the flow velocities under the improved operating condition increased from 0.330 to 0.361 m/s (the Surface layer) and from 0.314 to 0.339 m/s (the Top layer) as the distance from the internal wall increased. For the Middle and Bottom layers, the measured flow velocities increased from 0.262 to 0.284 m/s and from 0.187 to 0.216 m/s, respectively. The prediction was within 0.13 1.30% difference from the measured velocity values. The flow field in the ditch under the improved operating condition was more homogeneous than that under the
Fig. 9 e DO concentration profiles along the ditch (mg/l) ((a) existing operating condition, (b) improved operating condition (z represents the vertical distance from the water surface (m))).
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 4 3 9 e3 4 5 2
existing operating condition. Based on the simulation results and field measurements, a surface aerator should be relocated to about 15 m from the entrance of the curve bend. In this way, the faster flow at the top layer could be spread into the middle and the bottom layer, and then to the curve bend. This improved location of the surface aerator can reduce energy loss caused by fierce collisions at the wall of the curve bend. The integral of the predicted flow field values were within 1.98 4.28% difference from the actual measured values, indicating that the flow pattern in oxidation ditch with surface aerators and submerged impellers operating simultaneously could be captured by the CFD model.
3.2.
Oxygen mass transfer
3.2.1.
DO concentration profiles
3449
Fig. 9 illustrates the variations of DO concentrations along the ditch by using the unit analysis method. As shown in Fig. 9, higher DO concentrations were observed near the surface aerator than near the end of the channel. In this study, the DO profiles were more important than the specific DO values. The objective was to propose an operating condition that could lead to better DO concentration gradients for SND. The CFD results showed that DO concentration gradients were more apparent under the improved operating condition than those under the existing operating
Fig. 10 e DO concentration (mg/l) versus distance from the external wall (m) ((a) Test Location 2 (aerobic zone) under the existing operating condition, (b) Test Location 3 (anoxic zone) under the existing operating condition, (c) Test Location 2 (aerobic zone) under the improved operating condition and (d) Test Location 3 (anoxic zone) under the improved operating condition).
3450
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 4 3 9 e3 4 5 2
Table 2 e Characteristics of the influent of the full-scale oxidation ditch.
Table 3 e Comparison of the average effluent qualities under two operating conditions.
Parameters
Range
Average
Effluent quality (mg/l)
BOD5 (mg/l) COD (mg/l) NH3-N (mg/l) TN (mg/l) pH T ( C)
40.8e150.5 210.3e514.2 18.7e28.7 29.5e39.2 7.1e7.9 18.2e25.5
92.2 333.1 24.7 35.1 7.4 23.3
Existing operating condition Improved operating condition
condition. In addition, a DO concentration higher than about 0.5 mg/l in the anoxic zone is known to inhibit denitrification (Rittman and Langeland, 1985). As illustrated in Fig. 9a, the DO concentrations in the anoxic zone under the existing operating condition were greater than 1.0 mg/l and they could inhibit denitrification. Furthermore, the DO concentrations in the aerobic zone under the existing operating condition were much higher than 1.0e1.2 mg/l, which was a fitting DO range for nitrification in ODs (Liu et al., 2010a). It implied excessive aeration and waste of energy.
3.2.2.
Comparison of CFD results and measured data
DO concentrations were measured at two test locations (Test Location 2 in the aerobic zone; Test Location 3 in the anoxic zone) under the existing and improved operating conditions, respectively (Fig. 4a). As shown in Fig. 10, the differences of DO concentrations between Test Location 2 and Test Location 3 were obvious. It indicated that DO concentrations varied considerably along the ditch. Under the existing operating condition, the DO concentration varied from 2.75 to 5.32 mg/l in the aerobic zone and from 1.58 to 2.52 mg/l in the anoxic zone; while under the improved operating condition, the DO concentration varied from 1.75 to 3.38 mg/l in the aerobic zone and from 0.19 to 0.62 mg/l in the anoxic zone. DO concentration profiles were reflected with predicted values within 4.71 4.15% of the measured ones. The simulation error of DO concentration profiles was greater than that of flow field. This is mainly because of the complexity of biochemical reactions in oxidation ditches and the simplification of oxygen consumption and mass transfer modeling in the research. In general, the operating condition of surface aerators had a significant impact on the DO concentration profiles in the ditch. Under the existing operating condition, the DO concentrations were relatively high that caused waste of energy and the DO concentration gradients were not obvious;
BOD5
COD
NH3-N
TN
8.2 9.1
26.1 28.5
2.0 2.1
19.7 10.4
while under the improved operating condition, the DO concentrations showed apparent variations along the ditch and provided a more suitable environment for SND.
3.3. Demonstration of two operating conditions in a fullscale Carrousel OD There are two Carrousel ODs at the Ping Dingshan WWTP. Both of them have the same configurations and the wastewater flow rate to each ditch is 50,000 m3/d. Wastewater first flow into a surge tank before flowing into ODs, so the influent qualities in two ODs are essentially the same. A comparison on the performance of two operating conditions was carried out by running these two full-scale ODs concurrently for a month. The influent quality and effluent quality were monitored daily and energy consumptions for two operating conditions were calculated.
3.3.1. Comparison of effluent qualities under two operating conditions The characteristics of the influent are summarized in Table 2. Table 2 shows the main water quality indexes in OD systems at the Ping Dingshan WWTP. As illustrated in the table, the influent qualities fluctuated during the month and the actual operating condition adjusted accordingly. The average values of effluent quality under two operating conditions are summarized in Table 3. As shown in the table, BOD5, COD and NH3-N of two effluents were comparable and lower than 10, 50, 5 mg/l respectively, which were the first class A level of the new discharge standards of China (Ministry of Environmental Protection of China, 2002). However, total nitrogen (TN) of the effluent under the existing operating condition was much higher and exceeded the standard, 15 mg/l. This could be attributed to the relatively high DO concentrations in the anoxic zone that inhibit denitrification and the relatively uniform DO concentration gradients in the ditch that impair the possibility of occurrence of SND.
Table 4 e Comparison of energy consumptions under two operating conditions. Operating condition
Existing Improved
The number of surface aerators in operation
Total power input of surface aerators (kW)
The number of submerged impellers in operation
Total power input of submerged impellers (kW)
Total power input (kW)
9 6
333 222
3 6
12.57 25.14
345.57 247.14
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 4 3 9 e3 4 5 2
3.3.2. Comparison of energy consumptions under two operating conditions The power input of one surface aerator and one submerged impeller are 37 kW and 4.19 kW, respectively. Table 4 compares the energy consumptions under two operating conditions. The total energy consumption per hour under the improved operating condition is 98.43 kW h (28.5%) less than that under the existing operating condition. With an energy cost of 0.07 V/kW h (Vanrolleghem and Gillot, 2001), the annual energy saving would be about 60,000 V. Additional benefits could also be realized from less or no penalties from non-compliance.
4.
Conclusions
An experimentally validated numerical tool capable of predicting flow pattern and oxygen mass transfer characteristics was developed for OD systems. Two operating conditions, existing and improved, were compared and analyzed by considering two important factors: a) flow field in the ditch that has close relationship with energy consumption and process efficiency and b) DO concentration profiles that influence occurrence of SND in the ditch. The performance demonstration and comparison of the two operating conditions were carried out in two full-scale ODs for a month. The influent quality and effluent quality were monitored daily and energy consumptions were calculated. The following conclusions arise from this study: (1) The moving wall model and the fan model could successfully simulate surface aerators and submerged impellers of a full-scale OD, respectively. The total number of grids and computational complexity decreased significantly. The models characterized the flow pattern and oxygen mass transfer characteristics in the full-scale OD. (2) Based on the results of simulation and field measurements, a surface aerator should be relocated to about 15 m from the curve bend entrance. This location of the surface aerator can reduce energy loss caused by fierce collisions at the wall of the curve bend. (3) DO concentration profiles were established by incorporating the unit analysis method to the CFD model. Both the mass inlets from surface aerators and oxygen consumption of biochemical reaction were modeled. The results showed that DO concentration gradients in the OD under the improved operating condition were more favorable for occurrence of SND. (4) The results of the full-scale demonstration showed that the improved operating condition could reduce energy consumption and satisfy effluent standards. However, the influent flow rate, concentration and composition are likely to fluctuate day to day, so the validity of optimal operating condition chosen based on a steady-state model might be limited. Rather, an optimal operating condition should be derived from a dynamic model updated to accommodate actual process state and perturbations. The development of such a model is currently under investigation.
3451
Acknowledgements The authors are grateful to the persons who were involved in the field measurements. The authors thank for the financial supports from both Ping Dingshan WWTP and New Century Excellent Talents Project of Ministry of Education (NCET-09-0392).
references
Anderson Jr., J.D., 1995. Computational Fluid Dynamics. McGrawHill Inc, New York. Blackburne, R., Yuan, Z.Q., Keller, J., 2008. Demonstration of nitrogen removal via nitrite in a sequencing batch reactor treating domestic wastewater. Water Research 42 (8e9), 2166e2176. Clercq, B.D., Coen, F., Vanderhaegen, B., Vanrolleghem, P.A., 1999. Calibrating simple model for mixing and flow propagation in waste water treatment plants. Water Science and Technology 39 (4), 61e69. Fayolle, Y., Cockx, A., Gillot, S., Roustan, M., Heduit, A., 2007. Oxygen transfer prediction in aeration tanks using CFD. Chemical Engineering Science 62 (24), 7163e7171. Fiter, M., Colprim, J., Poch, M., Rodriguez-Roda, I., 2003. Enhancing biological nitrogen removal in a small wastewater treatment plant by regulating the air supply. Water Science and Technology 48 (11e12), 445e452. Gillot, S., Heduit, A., 2000. Effect of air flow rate on oxygen transfer in an oxidation ditch equipped with fine bubble diffusers and slow speed mixers. Water Research 34 (5), 1756e1762. Grady, C.P., Daigger, G.T., Lim, H.C., 1999. Biological Wastewater Treatment. Marcel Dekker Inc., New York. Insel, G., Artan, N., Orhon, D., 2005. Effect of aeration on nutrient removal performance of oxidation ditch systems. Environmental Engineering Science 22 (6), 802e815. Launder, B.E., Spalding, D.B., 1972. Lectures in Mathematical Models of Turbulence. Academic Press, England. Lesage, N., Sperandio, M., Lafforgue, C., Cockx, A., 2003. Calibration and application of a 1-D model for oxidation ditches. Chemical Engineering Research and Design 81 (9), 1259e1264. Littleton, H.X., Daigger, G.T., Strom, P.F., 2007a. Application of computational fluid dynamics to closed-loop bioreactors: I. Characterization and simulation of fluid-flow pattern and oxygen transfer. Water Environment Research 79 (6), 600e612. Littleton, H.X., Daigger, G.T., Strom, P.F., 2007b. Application of computational fluid dynamics to closed-loop bioreactors: II. Simulation of biological phosphorus removal using computational fluid dynamics. Water Environment Research 79 (6), 613e624. Liu, Y., Shi, H., Xia, L., Shi, H., Shen, T., Wang, Z., Wang, G., Wang, Y., 2010. Study of operational conditions of simultaneous nitrification and denitrification in a Carrousel oxidation ditch for domestic wastewater treatment. Bioresource Technology 101 (3), 901e906. Liu, Y., Shi, H., Shi, H., Wang, Z., 2010. Study on a discrete-time dynamic control model to enhance nitrogen removal with fluctuation of influent in oxidation ditches. Water Research 44 (18), 5150e5157. Luo, L., Li, W.M., Deng, Y.S., Wang, T., 2005. Numerical simulation of a combined oxidation ditch flow using 3D k-3 turbulence model. Journal of Environmental Sciences 17 (5), 808e812. Ministry of Environmental Protection of China, 2002. Discharge Standard of Pollutants for Municipal Wastewater Treatment
3452
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 4 3 9 e3 4 5 2
Plant, No: GB 18918e2002. Ministry of Environmental Protection of China, Beijing. Rittman, B.E., Langeland, W.E., 1985. Simultaneous denitrification with nitrification in single-channel oxidation ditches. Water Pollution Control Federation 57 (4), 300e308. Rodi, W., 1980. Prediction Method for Turbulent Flows. McGrawHill International Book Company, New York, USA. Stamou, A.I., 1993. Prediction of hydrodynamic characteristics of oxidation ditches using the k-3 turbulence model. In: Second Int. Symposium on Eng. Turbulence Modeling and Measurements, Florence, Italy, pp. 261e271. Stamou, A.I., 1994. Modeling oxidation ditches using the IAWPRC activated sludge model with hydrodynamic effects. Water Science and Technology 30 (2), 185e192. Stamou, A.I., 1997. Modeling of oxidation ditches using an open channel flow 1-D advection dispersion equation and ASM1 process description. Water Science and Technology 36 (5), 269e276.
Stamou, A.I., 2008. Improving the hydraulic efficiency of water process tanks using CFD models. Chemical Engineering and Processing: Process Intensification 47 (8), 1179e1189. Talvy, S., Cockx, A., Line, A., 2005. Global modelling of a gaseliquidesolid airlift reactor. Chemical Engineering Science 60 (22), 5991e6003. U.S.EPA, 1992. Evaluation of Oxidation Ditches for Nutrient Removal. Technical Report No: 832; R-92-003. U.S.EPA, Washington, DC. Vanrolleghem, P., Gillot, S., 2001. Robustness and economic measures as control benchmark performance criteria. Water Science and Technology 45 (4e5), 117e126. Warsi, Z.U.A., 1998. Fluid Dynamics: Theoretical and Computational Approaches, second ed. CRC Press, Boca Raton, Florida. Yang, Y., Wu, Y.Y., Yang, X., Zhang, K., Yang, J.K., 2010. Flow field prediction in full-scale Carrousel oxidation ditch by using computational fluid dynamics. Water Science and Technology 62 (2), 256e265.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 4 5 3 e3 4 6 0
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Modelling biocide leaching from facades Irene K. Wittmer a,b,*, Ruth Scheidegger a, Christian Stamm a, Willi Gujer a,c, Hans-Peter Bader a a
Eawag, U¨berlandstrasse 133, 8600 Du¨bendorf, Switzerland Institute of Biochemistry and Pollutant Dynamics, ETH Zu¨rich, Switzerland c Institute of Environmental Engineering, ETH Zu¨rich, Switzerland b
article info
abstract
Article history:
Biocides leach from facades during rain events and subsequently enter the aquatic envi-
Received 1 June 2010
ronment with storm water. Little is known about the losses of an entire settlement, since
Received in revised form
most studies referred to wash-off experiments conducted under laboratory conditions.
1 April 2011
Their results show a fast decrease of concentrations in the beginning, which subsequently
Accepted 1 April 2011
slows down. The aim of this study is to develop a simple model to understand the
Available online 8 April 2011
mechanisms leading to these losses as well as to simulate losses under various rainfall and application conditions.
Keywords:
We developed a four-box model based on the knowledge gained from fits of an expo-
Runoff
nential function to an existing experimental data set of a wash-off experiment. The model
Diuron
consists of two mobile stocks from which biocides are washed off during a rain event.
Storm water
These mobile stocks are supplied with biocides from storage stocks by diffusion-type
Material protection
processes. The model accurately reproduced the measured data of wash-off during
Surface waters
single cycles as well as peak wash-offs over all cycles. Our model results for diuron losses showed that a large proportion (w70%) of the applied biocides are still in the stocks even after a rain volume corresponding to several years (1100 mm y1, Swiss Plateau). Applications to realistic outdoor conditions showed that losses can not be neglected for urban environments and that knowledge about the amount of rainfall turned into runoff and the decay constants of the biocides in the facades are crucial. The model increased our understanding of the processes leading to the observed dynamic in laboratory experiments and was used to simulate losses for various rainfall and application conditions. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
Biocides are commonly found in surface waters of urban catchments throughout the year (Gerecke et al., 2002; Hollender et al., 2009; Wittmer et al., 2010). They are used to control unwanted organisms, for example in cosmetic products to prevent fungal growth or in building envelopes to control algae, fungi etc. (FriedliPartner, 2007; Paulus, 2005).
A recent study of biocide contributions to surface waters showed that for certain compounds, facades can be the major source (Wittmer et al., 2011). Some biocides (e.g., diuron), are of environmental concern even at low concentrations of a few hundred nanograms per litre (e.g., Che`vre et al., 2006). Such concentrations have been observed in several surface waters (e.g., Quednow and Pu¨ttmann, 2007; Wittmer et al., 2010) but little is understood about what influences their level. To
¨ berlandstrasse 133, 8600 Du¨bendorf, Switzerland. * Corresponding author. Eawag, U E-mail address:
[email protected] (I.K. Wittmer). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.04.003
3454
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 4 5 3 e3 4 6 0
develop mitigation strategies, it is important to acquire a better understanding of the losses from the sources as well as of the transport to surface waters. The leaching of biocides from facades was investigated by several studies. They carried out either controlled indoor rain experiments (Burkhardt et al., 2009a; Schoknecht et al., 2009) or submerged samples of facades in water for a certain period of time (Jungnickel et al., 2008; Schoknecht et al., 2009, 2003). All studies under controlled conditions found the same leaching dynamic for various compounds (e.g., carbendazim, terbutryn, diuron), namely a fast decrease of the concentrations at the beginning, which then slowed down. Many authors assumed that diffusion from the render or paint into the water phase is the main process controlling this decrease. One of the few outdoor studies was done by Burkhardt et al., (2007), who showed that biocide wash-off actually occurs from a real building and that the relevant compounds are found in the corresponding sewer system. To further understand the data, previous studies fitted a single or the sum of two exponential functions to the measured data (Jungnickel et al., 2008; Schoknecht et al., 2009). Their conclusion was that biocide losses from facades show an exponential decline as a function of time. However, an exponential fit does not allow us to extrapolate to different rainfall intensities and sequences. The aim of our study is to develop a simple biocide washoff model to simulate the losses from facades based on the applied biocide amounts and on rainfall. Our model is based on the knowledge gained from laboratory experiments by Burkhardt et al. (2009a). It gives insight into possible mechanisms leading to the observed biocide losses in order to simulate the dynamic of biocide loads from facades to surface waters. In a second step we applied the model to more realistic outdoor conditions in order to assess the importance of biocide leaching under realistic outdoor conditions.
2.
Methods
The method used here is the extended substance flow analysis introduced by Baccini and Bader (1996). The procedure is described in many publications (see for example Zeltner et al. (1999)) and consists of four steps: 1. system analysis, 2. model approach, 3. calibration and 4. simulation, including a sensitivity and uncertainty analysis.
2.1.
System analysis and data fit
The study carried out by Burkhardt et al. (2009a) is one of the most detailed indoor experiments ever carried out in which a sample facade was exposed to controlled rainfall. In the following we will focus on the experimental data published by Burkhardt et al. (2009a) for the development of the model and comparison of model results with experimental results. In the experiment (by Burkhardt et al. 2009a), leaching of biocides from render was studied. Two layers of render were applied to the sample facade (2 m 0.75 m). The lower layer contained no biocides and the top layer of render contained eight different ones (incl. diuron, terbutryn and carbendazim). The applied biocide amounts per kg of render were 0.5 g kg1
(1.5 g m2) for each biocide, which is in the range of common practice (0.1e2 g kg1, Burkhardt et al., 2009a). In the following, we focus on the biocide diuron as an example. For the wash-off experiment, facade samples were exposed to controlled rainfall. The rain application consisted of four times 20 rainfall cycles (Fig. 1). During one artificial rainfall cycle (60 min), 85 mm of water were evenly “rained” on the facade sample. Thus, all rain reached the facade (which is not the case for a realistic situation of a facade on a house). This resulted in a total of 6800 mm of rain, which is about six times the annual average rainfall of Zu¨rich (MeteoSchweiz). Thus, as Burkhardt et al. (2009) mentioned, an accelerated wash-off was generated. The leachate was sampled during the rainfall cycles. The time span between each cycle was 5 h. After a set of 20 cycles, there was a break of 48h. Detailed experimental conditions can be found in Burkhardt et al. (2009a). Burkhard et al. examined both the dynamic of the wash-off rate of several biocides during single rain events as well as the dynamic of the peak wash-off rates over all rainfall cycles. In the following we selected diuron as an example. However, other studied compounds showed a similar behaviour. For the studied single rain event numbers 1 and 61, they found that the dynamic of the diuron wash-off rate is characterised by a fast decrease at the beginning followed by a long slow-off (dots, Fig. 1a and b). Similar behaviour was observed for the peak wash-off rates of the single wash-off cycles as a function of time (dots, Fig. 1c). To understand this loss dynamic, we first fitted a single exponential function (first term on the right hand side of Eq. (1)) to the diuron wash-off rate of rain event numbers 1 and 61 as well as to the peak wash-off rates of all cycles (dashed lines, Fig. 1 aec). The single exponential function could only explain the fast decrease of the first three to 8 min, but not the long slow-off following this fast decrease during the remaining 50 min. Also for the dynamic of the peak wash-off rates of all cycles, only the fast decrease was explained, but not the slowdown. In a second step, we used a sum of two exponential functions (Eq. (1)) to describe the two types of dynamic of a fast decrease followed by a slow decrease and to fit to the data more accurately. Owash ðtÞ ¼ a,eb,t þ c,ed,t
(1)
where Owash ðtÞ is the wash-off rate of the biocide diuron, t is the time and aed are the parameters to be fitted. The sum of two exponential functions allowed perfect fits for single cycles (first and 61st cycles, Fig. 1a,b) and good fits for the peak wash-off rates over all cycles (Fig. 1c). However, the fitted parameters describing the exponential decays were different. This means that there is no single parameter set (aed) which explains the dynamic of a single wash-off cycle as well as that of the peaks of the wash-off rates. In conclusion, the wash-off rates of single events as well as the peak washoff rates show a two-fold exponential decay dynamic.
2.1.1.
Event dynamic
A single exponential decline can be modelled by a simple firstorder stock approach. A sum of two exponential functions requires a coupled two-stock approach. Each single event can
3455
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 4 5 3 e3 4 6 0
A
B D
C
Fig. 1 e Fit of sum of two exponential functions to the measured biocide (diuron) wash-off rates : A) wash-off rates of first cycle, and B) wash-off rates of 61st cycle C), peak wash-off rates of all cycles, D) values of the fitting parameter of the double exponential function. (Measured data was taken from Burkhardt et al. (2009a))
consequently be “modelled” by a coupled two-stock system, namely a mobile stock with a high depletion rate which is fed by diffusion from a storage stock with a lower depletion rate (diffusion rate). The mobile stock describes the mobilised biocides on the surface whereas the storage stock is the reservoir of the biocides in the facades.
2.1.2.
Peak dynamic
For such a two-box model, the peak wash-off rates of the single events decrease exponentially as a function of time, which differs from the observed fast decrease followed by a prolonged slow decrease. Such peak behaviour can be modelled by an overlap of two uncoupled two-box models with different time characteristics. This four-box model can be interpreted as follows: it represents the reservoirs, surface stocks, diffusion and wash-off of two different biocide “systems”, namely a system with “fast” diffusion and one with “slow” diffusion from the storage to the mobile phase, such as a biocide layer on top of render particles and the biocides in the particles. The corresponding system is shown in Fig. 2. In Fig. 2, O refers to the wash-off of biocides from the facade (g s1), A to the diffusion of biocides from the storage to the mobile stocks (g s1).
only differ by different storage and washout characteristics. All stock and flow variables are normalized to a unit area of facade. The unit of the stocks is therefore g m2 and that of the flows is g m2 s1. The biocides are incorporated in render or paints which are applied to the facades at the end of the construction phase and during possible renovation phases. This can be formulated as initial conditions: Mtot ðt ¼ 0Þ ¼ Papplied ðfastÞ
ðslowÞ
þ Mmobile ðt ¼ 0Þ
Model approach
(2)
where Mtot is the sum of all stocks and Papplied the total applied amount [g m2]. The separation into the four stocks ðfastÞ ðMstorage ; .; Mslow mobile Þ is defined by fractions of the total applied amount as follows:
O
(fast)
1 MMobile
A
(fast)
2 MStorage
Fas t diffus ion subsystem O
2.2.
ðfastÞ
ðslowÞ
¼ Mstorage ðt ¼ 0Þ þ Mstorage ðt ¼ 0Þ þ Mmobile ðt ¼ 0Þ
(slow)
3 MMobile
(slow)
A
4 MStorage
Slow diffus ion subsystem
The system described in Section 2.1 can be mathematically described as follows: the two subsystems of “fast diffusion” and “slow diffusion” are mathematically equivalent. They
Fig. 2 e Flows and stocks considered in the four-box model of biocide leaching from facades.
3456
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 4 5 3 e3 4 6 0
ðfastÞ
ðfastÞ
Mstorage ðt ¼ 0Þ ¼ Papplied ,kðfastÞ ,kstorage
(3)
ðfastÞ ðfastÞ Mmobile ðt ¼ 0Þ ¼ Papplied ,kðfastÞ , 1 kstorage
(4)
where kðfastÞ is the fraction which is initially part of the fast diffusion sub-system. The fraction kstorage of that amount goes to storage. The initial conditions of the stocks in the slow diffusion sub-system are defined accordingly. In case the biocides are applied several times over a certain time period, the initial conditions could be replaced by time-dependent input functions. The following functions are identical for the fast and slow diffusion subsystems; we therefore omitted the superscripts (fast) and (slow). The diffusion process from storage to mobile stock is described by a modified diffusion approach: Mstorage ðtÞ Mmobile ðtÞ AðtÞ ¼ Max 0; kdiffusion , dstorage dmobile
(5)
where kdiffusion is the substance exchange coefficient, Mstorage and Mmobile the biocide amount in the stocks, and dstorage and dmobile the thickness (Volume) of the two stocks. The flow A(t) is always positive from the storage to the mobile stock. Note that Mstorage ðtÞ=dstorage is the concentration of the biocide in the storage stock (g m3) since Mstorage represents the stock per unit area (g m2). A first-order kinetic process is assumed for the internal decay of stock as well as for the wash-off rates of the biocides. For example: stock decay Rstorage ðtÞ: Rstorage ðtÞ ¼ rstorage ,Mstorage ðtÞ
(6)
Wash-off rates: OðtÞ : OðtÞ ¼ kwash ,Prain ðtÞ,Mmobile ðtÞ
(7)
where kwash is the specific wash-off coefficient per mm of rain and Prain is the time-dependent rain intensity in mm s1. The model equations form a set of four (non-linear) ordinary differential equations for the four stocks, where the two equations for each sub-system are coupled:
ðfastÞ
diuron values. The rain Prain ðtÞ and the total application intensity Papplied were given by the lab experiments and are not calibrated. All the other parameter functions were manually calibrated as follows: 1) All parameters were assumed to be constant in function of time. 2) The simulated wash-off rates and peaks should fit the experimental data. 3) Biocide input to the facade was defined by initial conditions (Eq. (2)e(4)). The fraction of the amount applied to the “fast diffusion” sub-system was calibrated to be 25% of the total applied amount. Out of these, 18% goes into the mobile stock which is nearly washed off in the first cycle, and the rest goes to storage. In the “slow diffusion” sub-system, only the storage stock was initially filled. 4) The thickness of the entire render layer was estimated on the basis of the applied amount (kg m2) and the average density of the render (600 kg m3). The thickness of the ðfastÞ ðslowÞ single layers (dstorage ; .; dmobile ) was estimated on the basis of the parameters a and c of the double exponential fits (Eq. (1)). In a second step, the diffusion rates were adjusted. We assumed that the diffusion rates of both subsystems (fast and slow diffusion storage) are equal. 5) The decay in the facades was neglected for the simulation of the laboratory tests because they only lasted for less than three months. However, for the application of the model to realistic outdoor conditions various decay constants were considered (see Section 4 for details). 6) The calibration of the wash-off coefficients kwash was initially based on the parameter b of the exponential fits to the first and 61st rainfall cycles (parameter b of the first cycle and the 61st cycle respectively, Fig. 1D). For constant rain it can be shown that b ¼ kwash ,Prain ðtÞ. The finally calibrated values of kwash were a factor of 3 and 1.3 lower (for fast and slow diffusion storage respectively) then the ones estimated from b. The calibration procedure was an iterative process of the following steps a) simulating with a parameter set b)
ðslowÞ
dMmobile dMmobile ðfastÞ ðslowÞ ¼ AðfastÞ ðtÞ OðfastÞ ðtÞ Rmobile ðtÞ ¼ AðslowÞ ðtÞ OðslowÞ ðtÞ Rmobile ðtÞ dt dt and ðslowÞ ðfastÞ dMstorage dMstorage ðslowÞ ðfastÞ ¼ AðslowÞ ðtÞ Rstorage ðtÞ ¼ AðfastÞ ðtÞ Rstorage ðtÞ dt dt
The developed model includes all known relevant processes and can thus be used to simulate various compounds, application rates and rainfall conditions.
2.3.
Calibration
The above model contains a set of fifteen parameters1 and two experimental time series, which are available in Table 1. The model was calibrated based on the measured experimental 1 Here constant parameter functions are used. However, they could be replaced by time-dependent parameter functions considering e.g., the influence of temperature.
comparison of modelled with measured results c) adjust parameters in order to better fit the measured results. The “manual calibration procedure” for the parameter values turned out to be successful. Indeed, only small adjustments were necessary during the calibration. Manual calibration was chosen because it gives a good insight into the model, which is necessary to gain a system understanding and to assess the sensitivity. Furthermore, some parameters (diffusion rates and the thicknesses of the layers) of the model are not identifiable by the available data sets. If both are multiplied by the same factor, the model does not change.
3457
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 4 5 3 e3 4 6 0
Table 1 e Model parameters and calibrated parameter values with standard deviations (Stdv) and parameters for scenario calculations. The superscripts (fast) and (slow) indicate the subsystems “fast diffusion” and “slow diffusion”. Mean value
rel. Stdv a
Prain ðtÞ
85 mm h1 ¼ 0.024 mm s1
10%
Initial conditions Papplied
1.5 g m2
10%
Total amount of biocide applied to the facade
25%
30%
Fraction of total applied biocides ðPapplied Þ to the “fast diffusion” sub-system.
kstorage
82%
30%
Fraction of applied biocides in the “fast diffusion” sub-system to storage.
ðslowÞ kstorage
100%
30%
Fraction of applied biocides in the “slow diffusion” sub-system to storage.
30%
Substance exchange coefficient of diuron in the “fast diffusion” sub-system.
Parameter name
Description
Time series
k
ðfastÞ ðfastÞ
Rain intensity: four times 20 cycles of 1 h each (6800 mm in total, for details see Section 2.1)
Time independent ðfastÞ
kdiffusion
1.1 109 m s1
ðslowÞ kdiffusion
9
1.1 10
1
ms
30%
Substance exchange coefficient of diuron in the “slow diffusion” sub-system.
30%
Thickness of the storage in the “fast diffusion” sub-system.
ðfastÞ
2.5 105 m
dmobile
ðfastÞ
1 106 m
30%
Thickness of the mobile stock in the “fast diffusion” sub-system.
ðslowÞ dstorage
4.5 103 m
30%
Thickness of the storage in the “slow diffusion” sub-system.
ðslowÞ
1.5 106 m
30%
Thickness of the mobile stock in the “slow diffusion” sub-system.
1
0.11 mm
30%
Wash-off coefficient of the mobile stock in the “fast diffusion” sub-system.
kwash
0.29 mm1
30%
Wash-off coefficient of the mobile stock in the “slow diffusion” sub-system.
ðfastÞ ðslowÞ rstorage ; .; rmobile
no decay
dstorage
dmobile ðfastÞ kwash ðslowÞ
Parameters changed for scenarios ðfastÞ ðslowÞ 2.6 108 s1 rstorage ; rstorage ðfastÞ
ðslowÞ
Decay constant in the four different stocks. e
1.6 107 s1
e
7.9 109 s1
e
rmobile ; rmobile
1.6 108 s1
e
krunoff krunoff
1% 30%
e e
rmobile ; rmobile ðfastÞ
rstorage ; rslow storage ðfastÞ
ðslowÞ
Decay constant in storage stocks for fast decay (w92% in 3 years) Decay constant in mobile stocks for fast decay (w92% in 0.5 years) Decay constant in storage stocks for slow decay (w92% in 10 years) Decay constant in mobile stocks for slow decay (w92% in 5 years) Rainfall to runoff ratio Rainfall to runoff ration
a Relative standard deviation of the parameters.
3.
Results and discussion
Fig. 3 shows the measured and simulated diuron wash-off rates. The agreement between the measured and simulated values is good. This is also confirmed by two qualitative criteria namely, Nash-Sutcliffe criteria (0.89) (Nash and Sutcliffe, 1970) and the slope of the linear regression between measured and simulated values (0.97). Based on a lognormal distribution of the parameters, characterized by average and standard deviations given in Table 1, we conducted an uncertainty analysis using Monte Carlo procedure (500 samples). It was assumed that the different parameters were uncorrelated. All measured values were within the 5e95% uncertainty boundary and mostly within one standard deviation. A sensitivity analysis showed that, apart from rainfall and the total applied amount, the following processes most strongly influence the dynamic of single events: a) the washoff of biocides from the mobile stock (kwash ), and b) the diffusion from storage stock to mobile stock ðkdiffusion ; dstorage ; dmobile Þ.
The dynamic of peak wash-off rates during all cycles is strongly influenced by the split of biocides between the “fast diffusion” sub-system and the “slow diffusion” sub-system (kðfastÞ ). Additionally, the dynamic of the first twenty to forty cycles is dominated by the parameters of the “fast diffusion” sub-system. The remaining cycles are dominated by the parameters of the “slow diffusion” sub-system. According to the diffusion parameters, diffusion occurs fast enough to reach equilibrium between two rain cycles in less than 5 h. Furthermore, the diffusion process and the wash-off flow reach a steady state after a certain time of a rainfall cycle (Fig. 3a and b). This explains why biocide washout does not stop even during long rain events. After eighty cycles, the modelled losses due to wash-off were 30% of the applied amount. Both stocks (mobile and storage) of the “fast diffusion” sub-system were completely empty after this period of time. The stocks in the “slow diffusion” sub-system still contained 90% of their initial value (67.5% of the applied amount). Thus, the leaching out of the “slow diffusion” sub-system will continue for a longer time period with further rainfall.
3458
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 4 5 3 e3 4 6 0
A
B
B
Rainfall [mm]
Cumulative rainfall 8000 4000
2001 2002 2003 2004 Runoff to rainfall coefficient = 30%
2005
2006
2007
2005
2006
2007
2005
2006
2007
Cumulative wash-off [g m-2]
0.6
0.4
0.2
2001
C
2002
2003
2004
Runoff to rainfall coefficient = 1%
0.10
0.05
2001
2002
2003
2004
Accumlated wash-offs from: fast decay Total
slow decay
no decay
Fast diffusion fast decay
slow decay
no decay
Fig. 4 e Scenarios for cumulative wash-off under realistic outdoor conditions using measured rainfall time series of the years 2000e2007, graph A). Considered rainfall to runoff ratios were 30% in graph B) and 1% in graph C). Considered decay constants were: fast decay (90% within 0.5e3 years), slow decay (90% within 5e10 years), and no decay. Originally applied amount was 1.5 g mL2 diuron. Caution: axis in C) 4 times smaller than in B). The residual total stocks in percent of initially applied amount are in case of i) krunoff [ 30%: 60% for no decay, 13% for slow decay and 0 for fast decay and ii) krunoff [ 1%: 90% for no decay, 20% for slow decay, 0 for fast decay.
C
D
A
Cumulative wash-off [g m-2]
We also applied the model to simulate the losses of other compounds (e.g., Terbutryn). From the entire parameter set, we only adjusted the split of the initially applied amount to the “fast diffusion” sub-system and the “slow diffusion” subsystem. The results (not shown here) show good agreement with the experimental data, similar to those shown in Fig. 3. The NasheSutcliffe criteria for terbutryn was 0.84. This shows that the mechanisms represented by the model explain the dynamic of the biocide losses on facades. An interesting question is whether a simpler approach would be more appropriate. We initially started with a twobox model, followed by a three-box model, and ended with a four-box model approach. The two-box model approach (only one biocide stock subsystem) made it impossible to simulate the long-term dynamics of the losses. The three-box model consists of two storage stocks and one mobile stock. The two storage stocks have different diffusion characteristics to the mobile stock. This three-box model can simulate the experimental values if the wash-off coefficients do not differ. The available two data sets of single events did not allow us to determine whether the wash-off coefficients are identical or not. We therefore used the four-box model because it is a more general approach with two independent stocks. A chemical/physical interpretation of the fast/slow storages could be that a) the fast storages are biocides on the surface of the pores and the slow are the ones incorporated in the render or b) that not all biocides are incorporated in the render due to its solubility (500 mg/kgrender applied compared to 42 mg/lwater solubility (Paulus, 2005)). However, since neither the solubility of diuron in render nor the
E
distribution of biocides within the dried render is known these speculations would have to be justified by further experiments.
4. Model application to realistic outdoor conditions Fig. 3 e Measured (C) and modelled (-) diuron wash-off rates during: A) the first rain cycle, B) the 61st rain cycle, and C) peak wash-off rates during the entire experiment time (one tick represents one rainfall cycle), D) zoomed view of the days 14e20 and E) zoomed view of the days 21e27. (Measured data was taken from Burkhardt et al. (2009a))
We plan to use the model developed for laboratory data to calculate the biocide leaching from facades under realistic outdoor conditions. However to simulate realistic outdoor conditions two aspects have to be considered additionally: i) Under realistic conditions only a certain proportion of the
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 4 5 3 e3 4 6 0
rainfall reaches the facade and is turned into runoff and ii) decay has to be considered since the wash-off occurs over several months or years. The relationship of rainfall to facade runoff depends on many factors such as wind speed, rain intensity and location of the building (e.g., Blocken and Carmeliet, 2004; Blocken et al., 2009). Little is known about the average ratio of rainfall to runoff over an entire year. The few existing studies showed that the rainfall to runoff ratio can be in the range of <1% up to 50% (e.g., Burkhardt et al., 2009b; Steiner and Boller, 2004; Walser, 2007), being mostly between 0.2 and 3%. Therefore, we selected two scenarios for an average rainfall to runoff ratio of a) 30% and b) 1%. Our literature search yielded no results for decay constants of biocides in facades. However, it is known from field observation that the biocide protection lasts from about three to six years (M. Burkhardt pers. comm.). Thus, we selected three scenarios to account for various decay constants: 1) no decay, 2) 90% decayed after 0.5 years in the mobile stocks and three years in storage stocks and 3) 90% decayed after five years in mobile and ten years in the storage stock. We distinguished between decay constants for the mobile and the storage stock (Table 1), because we assumed the mobile stock to be more prone to decay (photo degradation, hydrolysis etc.). The developed model for laboratory conditions was used for the simulation of realistic outdoor conditions. Only Prain in Eq. (7) was replaced by Prunoff , which is the measured rainfall multiplied by the rainfall to runoff ratio (Prunoff ¼ Prain ,krunoff ). We calculated the diuron wash-off rate of an exemplary facade for measured rainfall series of the years 2000e2007 (Kloten, Switzerland, MeteoSchweiz, 2009). We calculated six scenarios including two rainfall to runoff ratios and the three decay constants (all parameters see Table 1). Besides rainfall and decay constants all parameters were kept as calibrated for diuron leaching under laboratory conditions. Fig. 4 shows the total cumulative wash-off and the washoff from the “fast diffusion” sub-system for the six scenarios. The results in Fig. 4 show that both, rainfall to runoff coefficient and the decay constant have a strong influence on the accumulated wash-off over seven years. The results further showed that in a first phase the “fast diffusion” sub-system (Boxes 1 and 2, Fig. 2) is dominant but in the second phase the “slow diffusion” sub-system (Boxes 3 and 4, Fig. 2) takes over. The duration of the first phase depends mainly on the rainfall to runoff ratio, in case of 30% it lasts for about 1.5 years and in case of 1% more than seven years. With a one box model only the beginning of the first phase could be simulated. It can be shown that the time until depletion of 92% of the stock in a one box model depends on runoff and decay rate as follows: kwash ,Srunoff ðDtÞ þ rmobile ,Dt ¼ 2:5 with Srunoff ðDtÞ being the total runoff over time. For such a one box model, the maximum duration until 92% are depleted would be 2.3 years (no decay, 1% wash-off, 1000 mm yr1 rain). This is in agreement with the discussion above, showing that with a one box model no long-term losses can be simulated. Depending on the decay constants in the various stocks longterm losses can be substantial.
3459
In summary, knowledge about decay constants and the rainfall to runoff ratio are crucial for the determination of the total amount washed out and the duration of the wash-off under realistic conditions.
5.
Conclusions
Biocide losses from facades typically show a fast initial decrease and long tailing during wash-off experiments (e.g., Burkhardt et al., 2009a). Although such a decline resembles an exponential decrease it deviates substantially from a single exponential function both within single wash-off cycles and over several cycles. By simulating the biocide losses with a four-box model instead of an exponential function, the experimental data was represented well. It further helped to improve the understanding of leaching processes of facades. The biocide wash-off dynamic could be described by four stocks representing the biocides in different layers. Two storage stocks feed by means of diffusion processes, one or two mobile stocks where biocides are washed off during rain events. These four stocks explain both the loss dynamic during a single rain event and the decrease of the peak of the single rain events on a long-term scale. The application of the model to experimental data for diuron revealed that roughly 2/3 of the applied amount is still in the facade after 6800 mm of artificial rain within 27 days. This amount of rainfall corresponds to rainfall of a period of about six years (w1100 mm yr1 in the Swiss Plateau). Considering that the factor for rainfall to runoff is in the range <1e50%, this would lead to an even longer leaching period. However, under realistic outdoor conditions also decay has to be considered. These two aspects were considered in simulations of various scenarios for realistic outdoor conditions and showed the importance of the rainfall to runoff ratio and decay constants. In order to better assess the losses of realistic outdoor conditions these aspects should be investigated in more detail. Overall these results confirm previous studies, both in laboratory and field that biocide losses from facades have to be considered in urban environments. To simulate long-term losses under realistic outdoor conditions a four-box model is needed. A one box model would only represent the very beginning of the washout. The model can be used to simulate various rainfall conditions and application amounts. It is not specific to one substance. However, for the prediction of different chemical compounds the model needs to be recalibrated. The model will be used to predict the losses of an entire settlement in a subsequent paper.
references
Baccini, P., Bader, H.-P., 1996. Regionaler Stoffhaushalt. Spektrum Akademischer Verlag GmbH, Washington D.C. Blocken, B., Carmeliet, J., 2004. A review of wind-driven rain research in building science. J. Wind Eng. Ind. Aerodynamics 92, 1079e1130. Blocken, B., Derome, D., Carmeliet, J., 2009. Overview of research on rainwater runoff from building facades. 12th Canadian
3460
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 4 5 3 e3 4 6 0
Conference on Building Science and Technology, Montreal, Quebec. Burkhardt, M., Kupper, T., Hean, S., Haag, R., Schmid, P., Kohler, M., Boller, M., 2007. Biocides used in building materials and their leaching behavior to sewer systems. Water Science and Technology 56 (12), 63e67. Burkhardt, M., Junghans, M., Zuleeg, S., Schoknecht, U., Lamani, X., Bester, K., Vonbank, R., Simmler, H., Boller, M., 2009a. Biozide in Geba¨udefassaden - o¨kotoxikologische Effekte, Auswaschung und Belastungsabscha¨tzung fu¨r Gewa¨sser. Umweltwiss Schadst Forsch 21, 36e47. Burkhardt, M., Zuleeg, S., Vonbank, R., Haag, R., Schmid, P., Hean, S., Bester, K., Boller, M., 2009b. Diffuse Belastung von Regenabwasser durch organische Problemstoffe und offene Fragen zum Regenwasser-Management. Hamburger Bericht zur Siedlungswasserwirtschaft 70, 37e42. Che`vre, N., Loepfe, C., Singer, H., Stamm, C., Fenner, K., Escher, B. I., 2006. Including mixtures in the determination of water quality criteria for herbicides in surface water. Environ. Sci. Technol. 40 (2), 426e435. FriedliPartner, 2007. Bafu, ERZ, Projekt Biomik -Biozide als Mikroverunreinigungen in Abwasser und Gewa¨sser. http://www. bafu.admin.ch/gewaesserschutz/03716/06387/ Accessed 15.12.08. Gerecke, A.C., Scha¨rer, M., Singer, H.P., Mu¨ller, S.R., Schwarzenbach, R.P., Sa¨gesser, M., Ochsenbein, U., Popow, G., 2002. Sources of pesticides in surface waters in Switzerland: pesticide load through wastewater treatment plants - current situation and reduction potential. Chemosphere 48, 307e315. Hollender, J., Zimmermann, S.G., Koepke, S., Krauss, M., McArdell, C. S., Ort, C., Singer, H., von Gunten, U., Siegrist, H., 2009. Elimination of organic micropollutants in a municipal wastewater treatment plant upgraded with a full-scale post-ozonation followed by sand filtration. Environ. Sci. Technol. 43 (20), 7862e7869. Jungnickel, C., Stock, F., Brandsch, T., Ranke, J., 2008. Risk assessment of biocides in roof paint. Environ. Sci. Pollut. Res. 15 (3), 258e265.
MeteoSchweiz, 2009. Klimadaten Schweiz. Nash, J.E., Sutcliffe, J.V., 1970. River flow forecasting through conceptual models part I e A discussion of principles. J. Hydrol 10 (3), 282e290. Paulus, W., 2005. Directory of Micorbiocides for the Protection of Materials, a Handbook. Springer, Dordrecht. Quednow, K., Pu¨ttmann, W., 2007. Monitoring terbutryn pollution in small rivers of Hesse. Germany. J. Envion. Monit. 9, 1337e1343. Schoknecht, U., Gruycheva, J., Mathies, H., Bergmann, H., Burkhardt, M., 2009. Leaching of biocides used in facade Coatings under laboratory test conditions. Environ. Sci. Technol. 43 (24), 9321e9328. Schoknecht, U., Wegner, R., Horn, W., Jann, O., 2003. Emission of biocides from Treated Materials. Environ. Sci. Pollut. Res. 10 (3), 154e161. Steiner, M., Boller, M., 2004. Kupferabtrag einer Kupferfassade und Wirksamkeit der Eisenhydroxid-Kalk-Adsorberschicht zur Abtrennung von Kupfer aus dem Fassadenwasser. Eawag. ¨ berWalser, A., Modellierung des Regenwasserabflusses einer U bauung ins Trennsystem und Analyse der potentiellen Gewa¨sserbelastung von Bioziden aus Geba¨udehu¨llen, 2007, Master thesis, Environmental Engineering, Zu¨rich, ETHZ. Wittmer, I., Singer, H., Scheidegger, R., Bader, H.P., Stamm, C., 2011. Urban biocide loss rates can exceed agricultural pesticide loss rates. Science of the total Environment 409, 920e932. Wittmer, I.K., Bader, H.P., Scheidegger, R., Singer, H., Lu¨ck, A., Hanke, I., Carlsson, C., Stamm, C., 2010. Significance of urban and agricultural land use for biocide and pesticide dynamics in surface waters. Water Res. 44, 2850e2862. Zeltner, C., Bader, H.P., Scheidegger, R., Baccini, P., 1999. Sustainable metal management exemplified by copper in the USA. Regional Environ. Change 1 (1), 31e46.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 5 1 2 e3 5 2 0
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Factors influencing 4-fluorobenzoate degradation in biofilm cultures of Pseudomonas knackmussii B13 Katarzyna Misiak a,c, Eoin Casey a,c,*, Cormac D. Murphy b,c a
School of Chemical and Bioprocess Engineering, University College Dublin, Ireland School of Biomolecular and Biomedical Science, University College Dublin, Ireland c Centre for Synthesis and Chemical Biology, University College Dublin, Ireland b
article info
abstract
Article history:
Membrane aerated biofilm reactors (MABRs) have potential in wastewater treatment as
Received 10 February 2011
they permit simultaneous COD minimisation, nitrification and denitrification. Here we
Received in revised form
report on the application of the MABR to the removal of fluorinated xenobiotics from
7 April 2011
wastewater, employing a Pseudomonas knackmussii monoculture to degrade the model
Accepted 13 April 2011
compound 4-fluorobenzoate. Growth of biofilm in the MABR using the fluorinated
Available online 21 April 2011
compound as the sole carbon source occurred in two distinct phases, with early rapid growth (up to 0.007 h1) followed by ten-fold slower growth after 200 h operation.
Keywords:
Furthermore, the specific 4-fluorobenzoate degradation rate decreased from 1.2 g g1 h1 to
Biofilm
0.2 g g1 h1, indicating a diminishing effectiveness of the biofilm as thickness increased. In
Fluorobenzoate
planktonic cultures stoichiometric conversion of substrate to the fluoride ion was
Fluoride
observed, however in the MABR, approximately 85% of the fluorine added was recovered as
Membrane
fluoride, suggesting accumulation of ‘fluorine’ in the biofilm might account for the
Biodegradation
decreasing efficiency. This was investigated by culturing the bacterium in a tubular biofilm reactor (TBR), revealing that there was significant fluoride accumulation within the biofilm (0.25 M), which might be responsible for inhibition of 4-fluorobenzoate degradation. This contention was supported by the observation of the inhibition of biofilm accumulation on glass cover slips in the presence of 40 mM fluoride. These experiments highlight the importance of fluoride ion accumulation on biofilm performance when applied to organofluorine remediation. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
The membrane-aerated biofilm reactor (MABR), in which oxygen is supplied to the biofilm solely from a gas permeable substratum, shows significant potential as a technology for high-rate biological wastewater treatment. In the MABR, the biofilm is naturally immobilized on an oxygen permeable membrane. Oxygen diffuses through the membrane into the
biofilm where oxidation of pollutants, supplied at the biofilmeliquid interface takes place. The oxygen supply rate can be controlled by the intra-membrane oxygen partial pressure and membrane surface area. Several investigators have reported performance advantages of MABRs for simultaneous COD oxidation, nitrification, and denitrification (Hibiya et al., 2003; Semmens et al., 2003; Timberlake et al., 1988; Yamagiwa et al., 1994), high oxygen utilisation efficiencies
* Corresponding author. UCD School of Chemical and Bioprocess Engineering, Engineering and Materials Science Centre, University College Dublin, Belfield, Dublin 4, Ireland. Tel.: þ353 1 7161877; fax: þ353 1 7161177. E-mail address:
[email protected] (E. Casey). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.04.020
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 5 1 2 e3 5 2 0
(Pankhania et al., 1994) and high specific organic reaction rates (Brindle et al., 1999; Ohandja and Stuckey, 2006). Much of the recent research has focused on nitrogen removal (Downing and Nerenberg, 2008; Lackner et al., 2008), however there is also interest in the use of the MABR for the aerobic treatment of xenobiotics. MABRs are viewed as particularly favourable in this context because bubbleless operation minimizes the airstripping of compounds with high Henry’s law constants such as xylene (Debus and Wanner, 1992) or acetonitrile (Li et al., 2008). The MABR is also of interest because the creation of, and ease of manipulation of, a defined oxic/anoxic micro-environment can be advantageous for the degradation of compounds with problematic intermediates such as perchloroethylene (Ohandja and Stuckey, 2006). In recent years the presence of fluorinated organic compounds in the environment has drastically increased as a result of the significant rise in the production of wide range of fluorinated pharmaceuticals and agrochemicals developed due to the unique and desirable properties of fluorine (Hansen et al., 2001; Muller et al., 2007). Environmentallyeunfriendly incineration is the mainstay for the management of most fluorinated waste (dos Santos et al., 2001), thus alternative methods are required to treat post-production water contaminated with fluorinated compounds, and biological processes might be the most economically and ecologically sound option. Biofilm reactors are ideally suited to the biological treatment of xenobiotics. It has been shown that biofilms have a higher resistance to toxic compounds (Morton et al., 1998) and this attribute could be highly beneficial when treating fluorinated aromatic compounds as transformation intermediates are often toxic. Strains have been isolated from industrial environments that have degradation capabilities towards fluorinated compounds (Carvalho et al., 2005). One such species, isolated from a sewage treatment plant and shown to have the ability to degrade fluorinated aromatic compounds, is Pseudomonas knackmussii, also known as Pseudomonas sp. B13. The strain has been also reported to form biofilm (Nielsen et al., 2000) and with its degradation properties (Schreiber et al., 1980) is potentially an ideal microorganism for the removal of fluorinated aromatic compounds in wastewater treatment. There are very few studies that have investigated the degradation of fluorinated compounds in biofilms, and none has investigated the degradation of fluorinated aromatic compounds in the MABR. Given the environmental significance of such compounds, we describe here investigations undertaken to evaluate the possible application of bacterial biofilms in the bioremediation of organofluorinecontaminated waste streams.
2.
Materials and methods
2.1.
Medium and culture conditions
P. knackmussii or Pseudomonas sp. B13 (DSM 6978) was obtained from the German Collection of Microorganisms and Cell Cultures (DSMZ, Germany). Brunners mineral medium (www. dsmz.de, medium 457) supplemented with the appropriate
3513
carbon source (10 mM) was the growth medium except for where stated otherwise.
2.2.
Planktonic studies
The planktonic growth rates were determined in 250 mL conical flasks with a working volume of 50 mL and incubated at 30 C with shaking at 150 rpm. The flasks were inoculated with 1 mL of a 24 h-old culture adjusted, only if needed, to an optical density of 0.8 at 660 nm with PBS (phosphate buffered saline). All planktonic trials were performed in triplicate. Biomass was established by dry cell weight and colony forming units (CFU) measurements.
2.3.
MABR
The MABR experiments were set-up as described previously by Heffernan et al. (2009b) with a modification of the sampling port (SI, Fig 4). The system operation was commenced by filling the reactor with 200 mL of medium and sterilisation of reactor by autoclaving at 121 C for 15 min in an Astel autoclave. Medium for the continuous phase of operation was sterilised separately by autoclaving at 121 C for 15 min in a 10 L carboy, which was subsequently connected aseptically to the reactor vessel. To prepare the inoculum; cells were first grown for 24 h in batch culture prior to reactor inoculation to obtain the exponential phase of growth; 10 mL of this culture was adjusted to a turbidity of approximately 0.8 at 660 nm in PBS and used to inoculate the reactor. Following inoculation the system was operated in batch mode for 48 h after which time the flow of medium was initiated at a flow rate of 40 mL h1 and maintained throughout the biofilm accumulation. Visible biofilm appeared within 48 h from commencing the feed. Experiments involving a change of the flow rate and pressure were performed after the biofilm had reached a steady state, which was established by thickness measurements. The reactor effluent was sampled with a 5 mL luer lock syringe attached to the circulation loop and analysed for pH, optical density (OD) and dissolved oxygen concentration (DO). The culture supernatant was also obtained by centrifuging the remaining liquid sample and stored at 20 C until analysis of the fluoride ion and other fluorometabolites were conducted.
2.4.
Tubular biofilm reactor (TBR)
The TBR experiments were set-up with some modifications of the system described previously by Heffernan et al. (2009a). Silicone tubing was sourced from VWR and Altec. The system was equipped only with one sample port located at the downstream section of the 90 cm reactor tubing and an additional 300 mL Duran bottle was introduced as an inoculation tank containing 200 mL of medium (SI, Fig 5), which was sterilised and then inoculated with 5 mL of 24 h-old cells (OD ¼ 0.8). Freshly inoculated medium was pumped through the system and after 48 h the feed was switched on and the first sample of culture supernatant was taken. TBR experiments with 10 mM of 4-fluorobenzoate as the carbon source were conducted for 216, 288, 360, 408, 432, and 504 h. Biofilm appeared within 96 h from the reactor inoculation. Samples were collected every 24 h during continuous culture using
3514
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 5 1 2 e3 5 2 0
a luer lock syringe connected to the sampling port. DO, OD and pH were measured and the remaining liquid sample was centrifuged and the supernatant stored frozen until analysis of the fluoride ion and other fluorometabolites were conducted. The initial investigation of free fluoride accumulation in TBR grown biofilm was performed in two experiments using medium containing 10 mM 4-fluorobenzoate as a carbon source. After 216 h operation of these two replicates was terminated and content of each reactor was harvested as three fractions: planktonic culture suspended in media (F1), sloughed off biofilm (F2) and mature biofilm (F3). F1 was obtained by draining the reactor tubing; F2 by washing cell content out with known volume of deionised water and F3 by pinching biofilm out and washing it out with known volume of deionised water. After each fraction was centrifuged, supernatants were examined for free fluoride ion concentration. Precipitated cells from each fraction were sonicated for 3 min (Ika Labortechnik U200S control) and centrifuged. Fluoride ion was measured in cell free extracts. Organic fluorometabolites were extracted with ethyl acetate from cell extracts and detected by 19F NMR.
2.5.
Biofilm thickness measurement
Biofilm thickness measurements in the MABR system were performed using the method described previously by Syron and Casey (2008b). The value obtained from the raw thickness was linearised (z) by using the equation z¼(ro þ d) ln((ro þ d)(ro)1), where d is the recorded thickness and ro is the radius of the silicone tube; the biofilm accumulation rate was then calculated from this value. Biofilm thickness in the TBR was established after each experiment was terminated and in a different manner than the online biofilm thickness measurement in the MABR system. Thickness analysis was performed using Able Image Analyser software (Mu Labs, Slovenia) on biofilm light microscopy images taken with an Olympus DP70 digital camera (4 magnification) by measuring the distance from the membrane to the biofilm liquid interface. The microscopy biofilm images were obtained after the reactor tubing (together with biofilm grown inside it) was cut and cryoembeded (embedded in freezing medium and frozen), the silicone tubing was removed and biofilm sample was sliced. The procedure was previously described by Heffernan et al. (2009a). From each section of the reactor six phase contrast images were analysed for thickness with 45 measurements taken from each image. These measurements were then averaged to give a final thickness.
2.6.
Biofilm cultivation in six well plates
Glass cover slips were treated with UV light, place in individual wells of 6-well plates and immersed in 8 mL sterile medium containing either 4-fluorobenzote (10 mM), benzoate (10 mM) or benzoate (10 mM) plus sodium fluoride (40 mM). Subsequently, the medium was inoculated with 1 mL of 24 hold culture (OD ¼ 0.8 at 660 nm) and incubated on a rocking platform at 30 C. Control wells contained medium that was not inoculated. At each sampling time (10, 24, 34, 48, 58 and
72 h) cover slips were removed and washed with phosphate buffered saline (PBS). The cover slips were stained with crystal violet and DAPI (40 ,6-diamidino-2-phenylindole) and microscopically examined using an Olympus BX51 epifluorescence microscope. The area of biofilm coverage of each slide was calculated by dividing the sum of all unit areas by the total area of a slide (141,050 mm2). The average biofilm coverage represents the average value obtained from 6 images, where each image represents separate slide taken from a well of a 6 well plate at each sampling time.
2.7.
Substrate and product analysis
Free fluoride concentration was measured by Fluoride/Fluoride Combination Electrode (Orion model 94-09) following the method described by Cooke (1972). The electrode was calibrated using NaF standards (1 mM, 10 mM and, if needed, 100 mM) in a mixture of H2SO4 (1 M) and KNO3/trisodium citrate buffer (0.5 M). Sample preparation involved mixing 1 mL of supernatant (or cell extract) with 1 mL of H2SO4 (1 M) solution and 8 mL of a KNO3/trisodium citrate buffer (0.5 M). 4-Fluorobenzoate and benzoate concentrations were determined by High Pressure Liquid Chromatography (HPLC) using a Varian ProStar system. Supernatant (10 mm) was eluted from a reverse phase column C18 (4.6 150 mm, 5 mm column Thermo Hypersil) using phosphoric acid (1 g L1) and acetonitrile (60:40) as the eluent. The 254 nm wavelength was monitored. The retention time under these conditions was 3.60 min for 4-fluorobenzoate and 3.29 min for benzoate. 4-fluorobenzoate was obtained from Fluorochem (Derbyshire, UK) other chemicals and various media components were obtained from a number of sources including BDH, Oxoid and SigmaeAldrich. Culture supernatants and extracts thereof were analysed by 19F nuclear magnetic resonance spectroscopy (19F NMR), using D2O as a solvent. Resonances detecting free fluoride, 4-fluorobenzoate and 4-fluorocyclohexadiene-cis,cis-1,2-diol1-carboxylate (4FDC) appear at d 120, 110 and 116 ppm, respectively (Boersma et al., 2004; SI, Fig 6). All 19F NMR analyses were performed using a Varian 400 MHz spectrometer.
3.
Results
3.1.
Development of biofilm in MABR
P. knackmussii was adapted for growth on 4-fluorobenzoate by continuous subculturing in medium containing 10 mM of the fluorinated substrate, eventually reaching a growth rate of 0.22 h1. This adapted strain was employed in all planktonic studies and in the establishment of the biofilm cultures. MABR experiments were conducted for 264 h (reactor I) and 600 h (reactors II and III), where 10 mM 4-fluorobenzoate was the sole source of carbon. Growth of the biofilm was assessed by measuring the thickness, and in the reactors II and III two distinct growth phases were observed: a period of relatively fast growth up to 200 h (0.004 and 0.007 h1, respectively) followed by a slower (factor 10) growth phase (Fig. 1A). Effluent from the reactors was collected every 24 h and the concentrations of fluoride ion and 4-fluorobenzoate were
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 5 1 2 e3 5 2 0
3515
0.2 g g1 h1 as the biofilm matured. Furthermore it was established during operation of reactor III, that there was no correlation between substrate degradation and oxygen supply; the highest specific utilisation rate was established at a pressure of 1 bar pure oxygen but only during the initial biofilm growth. The first increase of oxygen pressure from 0.4 to 0.7 bar resulted in increase of DO found in spent media from 3 to 6 mg/L. After the oxygen pressure was again increased the DO decreased back to the previous level of 2e3 mg/L and did not change overall during the rest of experiment even when the oxygen pressure was regularly elevated and declined. Therefore, some other factor(s), such as the accumulation of a toxic compound within the biofilm, was the cause of the poor substrate degradation rate (Adebusoye et al., 2008; Baggi and Zangrossi, 1999; Miguez et al., 1995). Approximately 85% of the fluorine initially added could be accounted for in terms of the fluoride ion and 4-fluorobenzoate; 19F NMR analysis of the effluent revealed only these metabolites (not shown), thus it is possible that a fluorinated intermediate, or fluoride ion, might accumulate in the biofilm. To investigate this, a TBR was employed because it facilitates recovery of biofilm for analysis of metabolites.
3.2. TBR
Fig. 1 e Linearised biofilm thickness measurements (Fig. A) reveal two distinct stages (A and B) of biofilm development, which biomass accumulation is correlated with declining specific utilisation rates (Fig. B) calculated as grams of utilised 4-fluorobenzoate by grams of dry cell weight per ), II ( ) and III ( ) hour in MABRs I ( operated with 10 mM 4-fluorobenzoate as a source of carbon. Substrate concentration measured in spent media supernatant is also shown (Fig. C).
measured, and used to calculate a specific utilisation rate (Fig. 1B). Although the biofilm thickness increased throughout the period of operation, the concentration of 4-fluorobenzoate (Fig. 1C) and fluoride ion in the effluent stabilised after approx. 100 h, thus the specific degradation rate decreased from 1.2 g g1 h1 in the early stage of biofilm growth (<200 h) to
Characteristics of 4-fluorobenzoate degradation in
P. knackmussii was grown in the TBR with 10 mM 4-fluorobenzoate as a source of carbon for experiments of 216e504 h duration and terminated when the system reached steady state based on unvarying and low OD values of the effluent. Initial attachment, microcolony formation, development of mature biofilm structure, sloughing events and biofilm recovery were observed to take place during each TBR experiment. Since each experiment was terminated at a different time it was possible to observe a relationship between cultivation time and biofilm thickness (Fig. 2). The utilisation of 4-fluorobenzoate was monitored by free fluoride and 4-fluorobenzoate measurements in the effluent, which together with biofilm dry cell weight measurements were used to calculate final specific utilisation rates. The specific utilisation decreased with biofilm increasing thickness, similar to what was observed in the MABR. A doubling of the thickness, from 25 to 50 mm, resulted in a 4-fold decrease in the degradation rate (Fig. 2). This difference is unlikely to be explained by substrate or oxygen limitation given the low thickness of the TBR grown biofilms. Furthermore the total fluorine (fluoride ion plus 4-fluorobenzoate) recovered in culture supernatants was approximately 85% of the starting substrate concentration. 19F NMR analyses of the effluent demonstrated that these comprised the major fluorine components, although a very small signal at d 116 ppm (4-fluorocyclohexadiene-cis,cis-1,2-diol-1carboxylate) was also observed (SI, Fig 6). Fluoride ion accumulation has been observed in oral biofilms (Engstrom et al., 2002; Watson et al., 2005) and it is a known enzyme inhibitor (Marquis et al., 2003; Nordstrom et al., 2009; Phan et al., 2002), thus the concentration of fluoride ion was determined in the biofilm. Three fractions of P. knackmussii culture obtained from the two TBR experiments operated for 216 h were analysed; the planktonic cells
3516
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 5 1 2 e3 5 2 0
Table 2 e Fluoride mass balance for each individual TBR experiment. TBR Amount of fluorine operation As substrate In spent In biofilm Recovered (h) (mmol) medium (F3) (mmol) (%) (mmol) 216 312 360 408 504
Fig. 2 e Data illustrating free fluoride accumulation increasing in time and its potential negative impact on utilisation abilities of P. knackmussii biofilm cultivated on 10 mM 4-fluorobenzoate in TBR systems operated for 216, 288, 360, 408, 432 and 504 h. Biofilm thickness ( )obtained from image software analysis and free fluoride concentration ( ) measured in biofilm cell free extract were established after each experiment was terminated. Final specific utilisation rates ( ) were calculated for each TBR experiment as a gram of substrate utilised by gram of ), dry cell weight per hour. Presented thickness ( ) and final specific fluoride ion concentration ( ) regression plots r2 [ 0.85, r2 [ 0.6 and utilisation ( 2 r [ 0.9, respectively.
collected along with spent medium (F1); the biofilm which detached after washing the TBR tubing with deionised water (F2); and the biofilm that was scraped off the TBR tubing (F3). In total, 0.095 mmol total fluoride ion was measured most of which was directly associated with the biofilm (Table 1). Fluoride ion was also measured in the biofilm extracts from TBR experiments that were conducted for 216e504 h. The amount of free fluoride as well as its total concentrations retained in the biofilm were found to increase with time, and consequently with thickness, from 0.14 to over 0.25 M (Fig. 2). Mass balance calculated for those TBR experiments indicates that accumulation of fluoride ion does not account for all of the fluorine in the biofilm (Table 2). At least some of the fluorine is also present as 4-fluorobenzoate (SI, Fig 7), and the
Table 1 e Free fluoride amount measured in Pseudomonas knackmussii culture retrieved from TBR system operated for 216 h. Data shown as average of two replicate experiments. Fraction Volume (mL)
Fluoride ion amount (mmol) Culture supernatant Cell extract
F1 F2 F3
6.164 0.019 0.175
0.004 0.007 0.003
0.038 0.018 0.025
Sum
6.359
0.014
0.081
13.5 27.9 35.1 42.3 56.2
11.4 23.1 27.9 34.5 46.4
0.025 0.020 0.040 0.036 0.054
85 83 80 82 83
remainder may be in the form of polymeric substances arising from auto-oxidation of fluorocatechols, which are known intermediates of fluorobenzoate degradation, and are not typically observable by 19F NMR.
3.3. The effect of fluoride ion on planktonic growth rates and biofilm formation The effects of fluoride have been studied in a variety of microorganisms, but nothing is known about the effects of the ion in P. knackmussii, thus experiments were undertaken by supplementing the medium with sodium fluoride in planktonic batch growth trials and biofilm cultures. To compare and eliminate possible growth limitations occurring as a result of fluorinated substrate degradation, benzoate was used as an alternative sole source of carbon in some experiments. Growth and specific utilisation rates were calculated in planktonic cultures incubated with 10, 20 and 40 mM NaF, by measuring dry cell weight and CFU (Table 3). In all planktonic batch experiments all applied fluorine, as 4-fluorobenzoate, was recovered in the culture supernatant as free fluoride ion at the end of logarithmic phase. Growth rates decreased as the fluoride ion increased, and the differences in growth rates were much more marked when benzoate was the carbon source. Similarly the specific utilisation rates of the substrate decreased as the concentration of fluoride increased. The effect of fluoride ion on biofilm development was investigated by culturing the bacterium in six-well plates
Table 3 e Specific growth rates based on CFU counts and specific utilisation rates calculated as gram of utilised substrate by gram of dry cell weight of biomass per hour obtained in growth trials, where planktonic culture of P. knackmussii was grown on 10 mM 4-fluorobenzoate and benzoate as sole sources of carbon with 0, 10, 20 and 40 mM supplemental sodium fluoride. All trials were performed in triplicates. Sodium 4-fluorobenzoate Benzoate fluoride mmax (h1) q (g g1 h1) mmax (h1) q (g g1 h1) (mM) 0 10 20 40
0.26 0.22 0.21 0.18
2.37 1.49 1.38 1.34
0.60 0.30 0.29 0.17
2.21 0.92 0.81 0.43
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 5 1 2 e3 5 2 0
3517
Fig. 3 e Average percentage coverage area data collected for 6-well plate experiments where media contained: 10 mM 4), 10 mM benzoate ( ) and 10 mM benzoate as a sole source of carbon plus 40 mM sodium fluorobenzoate ( ), complemented with OD ( ) recorded for sample supernatants and investigated substrate fluoride ( concentrations.
3518
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 5 1 2 e3 5 2 0
containing a glass coverslip, with medium containing benzoate and supplemented with 40 mM sodium fluoride. This additional fluoride concentration was established to have a clear negative impact on P. knackmussii biofilm culture. Crystal violet staining followed by microscopy and image software analysis was used to calculate the proportion of area covered by biofilm in all 6-well plate experiments (Fig. 3). The average percentage coverage area based on image analysis corresponded well to the observed visual result for each slide (SI Figs. 1e3) and complemented the OD recorded for planktonic cells; as biofilm grew on the glass cover slide the amount of the biomass in suspension decreased. It is clear that in the presence of 40 mM fluoride ion there is a dramatic reduction in coverage area.
experiments than in the starting medium, indicating that some ‘fluorine’ had accumulated in the biofilm. Thus a possible reason for declining degradation ability of biofilms in MABR was accumulation of free fluoride, which has been observed in dental plaques (Engstrom et al., 2002; Petersson et al., 2002; Twetman et al., 2003) but not previously reported in P. knackmussii biofilms. Since it is difficult to examine biofilm composition in the MABR system even after the experiment is terminated, biofilm grown in TBR was examined and high concentration of fluoride ion of up to 0.25 M was measured. Furthermore, accumulation of fluoride ion in the biofilm was determined to occur over time, and at such high concentrations might inhibit several key enzymes resulting in a diminished capability of the cells to degrade substrate.
4.
4.3. Fluoride ion impact on planktonic and biofilm culture
Discussion
Organofluorine compounds are used in an extensive range of applications, and consequently are widespread in the environment (Key et al., 1997). Microorganisms have long been known to degrade organofluorine compounds (Murphy, 2010) thus have considerable potential in bioremediation of polluted sites and waste streams. However, relatively few investigations have been conducted on the use of biofilms for organofluorine degradation (Emanuelsson et al., 2006; Osuna et al., 2008). The main objective of the present research was to assess the performance of continuously operated benchscale biofilm reactors for biodegradation of fluorinated compounds. P. knackmussii was used as a model organism since its utilisation abilities towards halogenated compounds is well studied (Dorn and Knackmuss, 1978; Schmidt et al., 1980; Schreiber et al., 1980). The MABR was chosen as the main system for investigation, since high loading rates can be applied and high oxygen transfer efficiencies are achievable through bubbleless aeration (Syron and Casey, 2008a).
4.1.
MABR performance
P. knackmussii planktonic and biofilm culture was shown to utilise 10 mM of 4-fluorobenzoate as a source of carbon. However, in the MABR the calculated specific 4-fluorobenzoate degradation rates unexpectedly declined with increasing biofilm thickness. A similar observation was made by Heffernan et al. (2009b) while investigating fluoroacetate degradation in Pseudomonas fluorescens in MABR. Mathematical modelling suggested that the rate of fluoroacetate degradation was affected by oxygen limitation and fluoride ion accumulation, but in the present study increasing the intramembrane oxygen pressure in the MABR did not affect the oxygen transfer rate. Furthermore, P. knackmussii biofilm cultivated in MABR on 4-fluorobenzoate did not achieve the thickness of P. fluorescens biofilm (1000 mm). Thus some other factor(s) causes the specific degradation rate to decline.
4.2. Free fluoride accumulation in biofilm continuous culture There was a lower concentration of total fluorine (fluoride ion plus 4-fluorobenzoate) in the spent medium from the MABR
Fluoride ion has been reported to impair transport mechanisms in bacteria (Marquis et al., 2003) and inhibit action of several enzymes (Belli et al., 1995; GuhaChowdhury et al., 1997; Phan et al., 2002; Todd and Hausinger, 2000). OchoaHerrera et al. (2009) investigated the effect of fluoride on microorganisms associated with wastewater systems, including denitrifying bacteria, aerobic heterotrophs and methanogens, and found that the groups of organisms investigated varied in their sensitivity to the ion. Thus, it was necessary to verify the influence of free fluoride concentrations on the planktonic and biofilm growth of P. knackmussii. Fluoride concentrations of 20 mM inhibited the growth of the strain in planktonic culture, and the formation of biofilms on cover slips was severely affected by the presence of 40 mM fluoride, strongly suggesting that the fluoride that accumulated in the biofilm as a result of 4-fluorobenzoate degradation in the MABR and TBR systems was responsible for the diminished degradation capacity, and highlights the problem of using such a system to remediate organofluorinecontaminated wastewater streams.
5.
Conclusions
The biodegradation of 4-fluorobenzoate was investigated in planktonic and biofilm cultures using P. knackmussii, a strain originally isolated from a wastewater treatment plant. It was established that the performance of the biofilm culture was comparable to planktonic culture. However, the specific utilisation of the substrate decreased with increasing biofilm thickness. The presence of fluoride ion, as the main product of 4-fluorobenzoate utilisation, was shown to be detrimental to planktonic growth and biofilm development. High concentrations (up to 0.25 M) of free fluoride were found to be retained within the biofilm and probably contributed to the decrease of the specific degradation rates. This work has broader implications for the use of biofilm-based wastewater treatment systems where organofluorine xenobiotics are to be degraded. Further investigation would be needed to establish the effect of fluoride accumulation on the degradation of organic matter in biofilm treatment processes for industrial wastewater.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 5 1 2 e3 5 2 0
Acknowledgement The authors acknowledge a CSCB studentship to KM.
Appendix. Supplementary material The supplementary data associated with this article can be found in the online version, at doi:10.1016/j.watres.2011.04. 020.
references
Adebusoye, S.A., Picardal, F.W., Ilori, M.O., Amund, O.O., 2008. Influence of chlorobenzoic acids on the growth and degradation potentials of PCB-degrading microorganisms. World Journal of Microbiology & Biotechnology 24 (7), 1203e1208. Baggi, G., Zangrossi, M., 1999. Degradation of chlorobenzoates in soil suspensions by indigenous populations and a specialized organism: interactions between growth and non-growth substrates. Fems Microbiology Ecology 29 (4), 311e318. Belli, W.A., Buckley, D.H., Marquis, R.E., 1995. Weak acid effects and fluoride inhibition of glycolysis by Streptococcus-mutans Gs-5. Canadian Journal of Microbiology 41 (9), 785e791. Boersma, F.G.H., McRoberts, W.C., Cobb, S.L., Murphy, C.D., 2004. A F-19 NMR study of fluorobenzoate biodegradation by Sphingomonas sp. HB-1. Fems Microbiology Letters 237 (2), 355e361. Brindle, K., Stephenson, T., Semmens, M.J., 1999. Pilot-plant treatment of a high-strength brewery wastewater using a membrane-aeration bioreactor. Water Environment Research 71 (6), 1197e1204. Carvalho, M.F., Jorge, R.F., Pacheco, C.C., De Marco, P., Castro, P.M.L., 2005. Isolation and properties of a pure bacterial strain capable of fluorobenzene degradation as sole carbon and energy source. Environmental Microbiology 7 (2), 294e298. Cooke, J.A., 1972. Fluoride compounds in plants: their occurrence, distribution and effects. Ph.D. thesis, University of Newcastleupon-Tyne, UK. Debus, O., Wanner, O., 1992. Degradation of xylene by a biofilm growing on a gas-permeable membrane. Water Science and Technology 26 (3e4), 607e616. Dorn, E., Knackmuss, H.J., 1978. Chemical-structure and biodegradability of halogenated aromatic-compounds e 2 catechol 1,2-dioxygenases from a 3-chlorobenzoate-grown Pseudomonad. Biochemical Journal 174 (1), 73e84. dos Santos, L.M.F., Spicq, A., New, A.P., Lo Biundo, G., Wolff, J.C., Edwards, A., 2001. Aerobic biotransformation of 4fluorocinnamic acid to 4-fluorobenzoic acid. Biodegradation 12 (1), 23e29. Downing, L.S., Nerenberg, R., 2008. Total nitrogen removal in a hybrid, membrane-aerated activated sludge process. Water Research 42 (12), 3697e3708. Emanuelsson, M.A.E., Henriques, I.S., Jorge, R.M.F., Castro, P.M.L., 2006. Biodegradation of 2-fluorobenzoate in upflow fixed bed bioreactors operated with different growth support materials. Journal of Chemical Technology and Biotechnology 81 (9), 1577e1585. Engstrom, K., Petersson, L.G., Twetman, S., 2002. Fluoride concentration in supragingival dental plaque after a single intake or habitual consumption of fluoridated milk. Acta Odontologica Scandinavica 60 (5), 311e314.
3519
GuhaChowdhury, N., Clark, A.G., Sissons, C.H., 1997. Inhibition of purified enolases from oral bacteria by fluoride. Oral Microbiology and Immunology 12 (2), 91e97. Hansen, K.J., Clemen, L.A., Ellefson, M.E., Johnson, H.O., 2001. Compound-specific, quantitative characterization of organic: fluorochemicals in biological matrices. Environmental Science & Technology 35 (4), 766e770. Heffernan, B., Murphy, C.D., Casey, E., 2009a. Comparison of planktonic and biofilm cultures of Pseudomonas fluorescens DSM 8341 cells grown on fluoroacetate. Applied and Environmental Microbiology 75 (9), 2899e2907. Heffernan, B., Murphy, C.D., Syron, E., Casey, E., 2009b. Treatment of fluoroacetate by a Pseudomonas fluorescens biofilm grown in membrane aerated biofilm reactor. Environmental Science & Technology 43 (17), 6776e6785. Hibiya, K., Terada, A., Tsuneda, S., Hirata, A., 2003. Simultaneous nitrification and denitrification by controlling vertical and horizontal microenvironment in a membrane-aerated biofilm reactor. Journal of Biotechnology 100 (1), 23e32. Key, B.D., Howell, R.D., Criddle, C.S., 1997. Fluorinated organics in the biosphere. Environmental Science & Technology 31 (9), 2445e2454. Lackner, S., Terada, A., Smets, B.F., 2008. Heterotrophic activity compromises autotrophic nitrogen removal in membraneaerated biofilms: results of a modeling study. Water Research 42 (4e5), 1102e1112. Li, T.G., Liu, J.X., Bai, R.B., Wong, F.S., 2008. Membrane-aerated biofilm reactor for the treatment of acetonitrile wastewater. Environmental Science & Technology 42 (6), 2099e2104. Marquis, R.E., Clock, S.A., Mota-Meira, M., 2003. Fluoride and organic weak acids as modulators of microbial physiology. Fems Microbiology Reviews 26 (5), 493e510. Miguez, C.B., Greer, C.W., Ingram, J.M., Macleod, R.A., 1995. Uptake of benzoic-acid and chlorosubstituted benzoic-acids by Alcaligenes Denitrificans Bri-3010 and Bri-6011. Applied and Environmental Microbiology 61 (12), 4152e4159. Morton, L.H.G., Greenway, D.L.A., Gaylarde, C.C., Surman, S.B., 1998. Consideration of some implications of the resistance of biofilms to biocides. International Biodeterioration & Biodegradation 41 (3e4), 247e259. Muller, K., Faeh, C., Diederich, F., 2007. Fluorine in pharmaceuticals: Looking beyond intuition. Science 317 (5846), 1881e1886. Murphy, C.D., 2010. Biodegradation and biotransformation of organofluorine compounds. Biotechnology Letters 32 (3), 351e359. Nielsen, A.T., Tolker-Nielsen, T., Barken, K.B., Molin, S., 2000. Role of commensal relationships on the spatial structure of a surface-attached microbial consortium. Environmental Microbiology 2 (1), 59e68. Nordstrom, A., Mystikos, C., Ramberg, P., Birkhed, D., 2009. Effect on de novo plaque formation of rinsing with toothpaste slurries and water solutions with a high fluoride concentration (5000 ppm). European Journal of Oral Sciences 117 (5), 563e567. Ochoa-Herrera, V., Banihani, Q., Leon, G., Khatri, C., Field, J.A., Sierra-Alvarez, R., 2009. Toxicity of fluoride to microorganisms in biological wastewater treatment systems. Water Research 43 (13), 3177e3186. Ohandja, D.G., Stuckey, D.C., 2006. Development of a membraneaerated biofilm reactor to completely mineralise perchloroethylene in wastewaters. Journal of Chemical Technology and Biotechnology 81 (11), 1736e1744. 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 (14), 3857e3869.
3520
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 5 1 2 e3 5 2 0
Pankhania, M., Stephenson, T., Semmens, M.J., 1994. Hollow-fiber bioreactor for waste-water treatment using bubbleless membrane aeration. Water Research 28 (10), 2233e2236. Petersson, L.G., Arvidsson, I., Lynch, E., Engstrom, K., Twetman, S. , 2002. Fluoride concentrations in saliva and dental plaque in young children after intake of fluoridated milk. Caries Research 36 (1), 40e43. Phan, T.N., Nguyen, P.T.M., Abranches, J., Marquis, R.E., 2002. Fluoride and organic weak acids as respiration inhibitors for oral streptococci in acidified environments. Oral Microbiology and Immunology 17 (2), 119e124. Schmidt, E., Remberg, G., Knackmuss, H.J., 1980. Chemicalstructure and biodegradability of halogenated aromaticcompounds e halogenated muconic acids as intermediates. Biochemical Journal 192 (1), 331e337. Schreiber, A., Hellwig, M., Dorn, E., Reineke, W., Knackmuss, H.J., 1980. Critical reactions in fluorobenzoic acid degradation by Pseudomonas sp. B13. Applied and Environmental Microbiology 39 (1), 58e67. Semmens, M.J., Dahm, K., Shanahan, J., Christianson, A., 2003. COD and nitrogen removal by biofilms growing on gas permeable membranes. Water Research 37 (18), 4343e4350. Syron, E., Casey, E., 2008a. Membrane-aerated biofilms for high rate biotreatment: performance appraisal, engineering
principles, scale-up, and development requirements. Environmental Science & Technology 42 (6), 1833e1844. Syron, E., Casey, E., 2008b. Model-based comparative performance analysis of membrane aerated biofilm reactor configurations. Biotechnology and Bioengineering 99 (6), 1361e1373. Timberlake, D.L., Strand, S.E., Williamson, K.J., 1988. Combined aerobic heterotrophic oxidation, nitrification and denitrification in a permeable-support biofilm. Water Research 22 (12), 1513e1517. Todd, M.J., Hausinger, R.P., 2000. Fluoride inhibition of Klebsiella aerogenes urease: Mechanistic implications of a pseudouncompetitive, slow-binding inhibitor. Biochemistry 39 (18), 5389e5396. Twetman, S., Engstrom, K., Petersson, L.G., 2003. Fluoride levels in plaque after a single or habitual intake of fluoridated milk. Journal of Dental Research 82 476e476. Watson, P.S., Pontefract, H.A., Devine, D.A., Shore, R.C., Nattress, B.R., Kirkham, J., Robinson, C., 2005. Penetration of fluoride into natural plaque biofilms. Journal of Dental Research 84 (5), 451e455. Yamagiwa, K., Ohkawa, A., Hirasa, O., 1994. Simultaneous organic-carbon removal and nitrification by biofilm formed on oxygen enrichment membrane. Journal of Chemical Engineering of Japan 27 (5), 638e643.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 5 2 1 e3 5 3 2
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Colloid transport in porous media: Impact of hyper-saline solutions Einat Magal a,b, Noam Weisbrod a,*, Yoseph Yechieli b, Sharon L. Walker c, Alexander Yakirevich a a
Department of Environmental Hydrology & Microbiology, Zuckerberg Institute for Water Research, Balustein Institutes for Desert Studies, Ben Gurion University of the Negev, Sede Boqer 84990, Israel b Geological Survey of Israel, Jerusalem 95501, Israel c Department of Chemical and Environmental Engineering, University of California, Riverside, Riverside, CA 92521, USA
article info
abstract
Article history:
The transport of colloids suspended in natural saline solutions with a wide range of ionic
Received 14 January 2011
strengths, up to that of Dead Sea brines (100.9 M) was explored. Migration of microspheres
Received in revised form
through saturated sand columns of different sizes was studied in laboratory experiments
12 April 2011
and simulated with mathematical models.
Accepted 14 April 2011 Available online 22 April 2011
Colloid transport was found to be related to the solution salinity as expected. The relative concentration of colloids at the columns outlet decreased (after 2e3 pore volumes) as the solution ionic strength increased until a critical value was reached (ionic
Keywords: Colloids
strength > 101.8 M) and then remained constant above this level of salinity. The colloids were found to be mobile even in the extremely saline brines of the Dead
Brine
Sea. At such high ionic strength no energetic barrier to colloid attachment was presumed
Breakthrough curve
to exist and colloid deposition was expected to be a favorable process. However, even at
Salinity
these salinity levels, colloid attachment was not complete and the transport of w30% of the
Microspheres
colloids through the 30-cm long columns was detected. To further explore the deposition of colloids on sand surfaces in Dead Sea brines, transport was studied using 7-cm long columns through which hundreds of pore volumes were introduced. The resulting breakthrough curves exhibited a bimodal shape whereby the relative concentration (C/C0) of colloids at the outlet rose to a value of 0.8, and it remained relatively constant (for the w18 pore volumes during which the colloid suspension was flushed through the column) and then the relative concentration increased to a value of one. The bimodal nature of the breakthrough suggests different rates of colloid attachment. Colloid transport processes were successfully modeled using the limited entrapment model, which assumes that the colloid attachment rate is dependent on the concentration of the attached colloids. Application of this model provided confirmation of the colloid aggregation and their accelerated attachment during transport through soil in high salinity solution. ª 2011 Elsevier Ltd. All rights reserved.
* Corresponding author. Tel.: þ972 8 6596903; fax: þ972 8 6596909. E-mail address:
[email protected] (N. Weisbrod). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.04.021
3522
1.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 5 2 1 e3 5 3 2
Introduction
Colloid transport in groundwater has been studied extensively in the past years and major findings have been summarized in several review papers (e.g., Bradford and Torkzaban, 2008.; Kretzschmar et al., 1999; McDowell-Boyer et al., 1986; Ryan and Elimelech, 1996; Sen and Khilar, 2006). The interaction energy between colloids and collectors is a key issue in retention processes of colloids in porous media. DLVO theory describes the interaction energy between surfaces as a sum of the electric double layer and London-van der Waals forces (Kretzschmar et al., 1999; Ryan and Elimelech, 1996). The ionic strength (IS) of the suspending solution has a major impact on the potential for interaction of colloids with surfaces. In response to an increase in the IS, the diffusive layer compresses enabling the attractive London-van der Waals forces to play a more dominate role, and in certain circumstances these forces may overcome electrostatic repulsion and result in an attractive energy interaction (McDowell-Boyer et al., 1986). For solutions composed of divalent and trivalent ions, the increase in colloid attachment can occur at substantial lower IS compared to mono-valent ions (Ryan and Elimelech, 1996). A DLVO interaction energy profile is constructed as the total interaction energy as a function of separation distance between a colloid and sediment grain surface or between colloids. A typical DLVO energy profile (low IS) of like-charged colloid and sediment grains (or colloids to colloids) is characterized by a deep attractive well (primary minimum) at a small separation distance, a sizable energy barrier, and a shallow attractive well (secondary minimum) at a larger distance (Shen et al., 2008). As the IS increases, there is a point at which the energy barrier to particleemedia interaction and attachment is eliminated based on the surfaces and solution chemistry (McDowell-Boyer et al., 1986). Deposition rates of colloids on sediment grains in
the presence of energy barriers (low IS) are slow or reactionlimited; whereas at high IS, as the repulsion energy barrier disappears every collision results in attachment and deposition kinetics is fast and transport-limited (Elimelech et al., 1995). The relative deposition (colloidecollector) and aggregation (colloidecolloid) rate constants follow very similar trends, featuring fast (favorable) and slow (unfavorable) regimes at high and low salt concentrations, respectively (Grolimund et al., 2001; Kretzschmar et al., 1997). Extensive research in recent years has focused on the improvement of the filtration theory by incorporating processes such as straining (entrapment of colloids within small pores and within pore-space constrictions associated with grainegrain contacts) (Bradford et al., 2003, 2006a,b, 2007; Shen et al., 2008; Torkzaban et al., 2008a), hydrodynamic effects (inhibition of particle deposition at high flow rates) (Ahfir et al., 2007; Bradford et al., 2006a). Research has also sought to account for the distribution in the interaction energies between particles (Tufenkji and Elimelech, 2005). Some work has also referred to the role of retained particles in the dynamics of colloid deposition in porous media (Ko and Elimelech, 2000; Kretzschmar et al., 1997; Kretzschmar and Sticher, 1997; Liu et al., 1995; Song and Elimelech, 1993). Particle transport is typically studied on the column scale using well-defined, simple model systems of uniform particles and sediment, low IS solutions, and a single type of electrolyte (Bradford et al., 2007; Compere et al., 2001; Grolimund et al., 2001; Kretzschmar et al., 1997; Nocito-Gobel and Tobiason, 1996). Particle release from packed columns has also been extensively studied in relatively high salinity environments (Blume et al., 2002, 2005; Khilar and Fogler, 1984; Khilar et al., 1983; Roy and Dzombak, 1996). Colloid detachment is a rather complicated process, but in general it is related to the decrease of solution salinity below a threshold value at which
Current Study Grolimund et al., 2001 (NaCl) Grolimund et al., 2001 (CaCl2) Tufenkji and Elimelech, 2005 Saiers and Ryan, 2006 Bradford et al., 2007 Kuhnen et al., 2000 Kretzschmar et al., 1997 (NaCl) Kretzschmar et al., 1997 (CaCl2) Johnson et al., 2007 Nocito-Gobel and Tobiason, 1996 Ko and Elimelech, 2000 Compere et al., 2001 0.0001
0.001
0.01
0.1
1
10
IS (molar) Fig. 1 e The ionic strengths of the solutions that were used in previous work on colloid transport. In these studies colloid transport was investigated in solutions of a single, mono-valence salt, unless specified otherwise. In this study (blue bar) natural saline solutions of DSW containing a variety of salts were used. Also high salinities on an order of magnitude more saline than that of all other studies were used. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
3523
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 5 2 1 e3 5 3 2
particles are released from the sediment (Blume et al., 2005; Khilar and Fogler, 1984; Roy and Dzombak, 1996; Fig. 1). Single-salt solutions typically used for exploring colloid deposition in porous media (Bradford et al., 2007; Compere et al., 2001; Grolimund et al., 2001; Johnson et al., 2007; Ko and Elimelech, 2000; Kretzschmar et al., 1997; Kuhnen et al., 2000; Nocito-Gobel and Tobiason, 1996; Saiers and Ryan, 2006; Tufenkji and Elimelech, 2005) do not adequately mimic the complex nature of groundwater containing a mixture of salts. Moreover, previous studies have focused mainly on low (up to w1 M) IS solutions. The current work studies the transport of colloid in the naturally extreme saline solution of the Dead Sea water (DSW, Table 1) with an IS of 8.5 M. Furthermore, studying colloid transport in a wide range of salinities and in a complex mixture of salts has been achieved by diluting the DSW. To the best of our knowledge the aforementioned conditions (wide range of salinities of natural and complex solutions) have not been examined before. The importance of colloid transport in DSW, beyond purely scientific interest, is its potential contribution to the creation of sinkholes around the Dead Sea coast. Numerous collapsed holes have developed along the Dead Sea shore causing a safety risk to people and infrastructure. It is well accepted that these sinkholes are generated as a result of subsurface dissolution of salt layers (Yechieli, 2006). This process is governed by the eastward and downward movement of the DSWefresh groundwater interface in response to the drop in the Dead Sea water level (Abelson et al., 2003; Yechieli, 2006). Consequently, the subsurface salt layers have been exposed to relatively fresh groundwater. In that dynamic environment, where the groundwater salinity changes over a timescale of months (Kiro et al., 2008), particle release and migration from the sediments was suggested as a supporting mechanism for the creation of the sinkholes (Arkin and Gilat, 2000). This is the motivation for this current work, which is aimed at exploring colloid transport in high salinity DSW using column experiments (on two scales). With experimental data and complementary mathematical modeling, the goal was to identify the potential for colloid transport in the complex solution of the DSW and, for the first time, compare between results from single-salt experiments to those of the natural system.
2. 2.1.
Materials and methods
than ocean water). DSW is typically Ca-chloridic brines, displaying the equivalent concentration relationship of Ca > (SO4 þ HCO3), Na/Cl < 0.86 and low SO4/Cl ratios relative to those of sea water due to precipitation of salts and several watererock interactions (Starinsky, 1974). In all the experiments, DSW taken from the En Gedi shore was used after filtration through a 0.45 mm filter. The DSW filtering procedure has been tested intensively in different experimental procedures including under the microscope and the filtered DSW did not contain any natural colloids. For certain experiments, noted below, DSW was diluted with deionized water (DI) water as follows: DSW/2 (i.e. 50% concentration of the DSW), DSW/5 (20% DSW), DSW/10 (10% DSW), DSW/100 (1% DSW), DSW/1000 (0.1% DSW), and DSW/5000 (0.02% DSW). The dilution process of the DSW increases the pH of the solution (Amit and Bentor, 1971) and the pH of DSW solutions that have been diluted by 100 (DSW/100) increases to 7.6 compared to pH w5.4 of undiluted DSW solution (Table 1). The increase of pH through the dilution of the DSW is an innate chemical property of saline solutions that carries more than 1.2e1.5% of total dissolved solids and is suspected to be connected to increased dissociation of bicarbonates salts (Amit and Bentor, 1971).
2.2.
Colloids
Fluorescent Carboxylate-Modified Latex (CML) microspheres of 1 mm diameter and excitation/emission wavelengths of 505/ 515 or 540/560 were used for visualization (FluoSpheres, Molecular probe, Eugene, Oregon). CML have often been used as model colloids due to their spherical and well-defined size and the ease in detection at low concentrations (Shani et al., 2008; Zvikelsky and Weisbrod, 2006). In these experiments, CML concentration of 3.64 107 colloids per ml (equivalent to 2 ppm) was used. The concentration was analyzed using fluorescence spectrophotometry (Cary Eclipse Fluorescence Spectrophotometer, Varian, Palo Alto, CA). Fluorescence intensity is slightly influenced by the solution IS, as was also demonstrated for fluorescent dye solutions (Magal et al., 2008); therefore, in each experiment, the concentration of colloids was determined utilizing linear calibration curves developed for the same ionic strength and composition as the test solution used (Magal, 2011).
2.2.1.
Solution chemistry
Dead Sea Water (DSW) is of marine-evaporitic origin with ionic strength of 100.9 M (ten times more concentrated
Conservative solute tracer
As a reference for the colloid transport, Sodium Naphthionate (Naph) was used as a conservative solute tracer. A previous study carried out on DSW solutions showed that Naph, with excitation/emission of 320/420, behaves as a conservative tracer even in highly saline brines (Magal et al., 2008).
Table 1 e Chemical analysis of undiluted Dead Sea water (DSW) including the concentration of key salts, total ionic strength, and total dissolved solids (TDS). Na
K
Ca
Mg
Sr
Cl
SO4
Br
Ionic strength (M)
Total dissolved solids (g/L)
8
70
8.5
324
mM/L 1416
189
873
3716
8
6065
3524
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 5 2 1 e3 5 3 2
Synchronous analysis proved a lack of spectrographic disturbances between the Naph and the CML used. The Naph was purchased as a dry powder (Fluka Chemika) and was dissolved in DI. In the experiment, solutions of different IS were prepared with the Naph tracer concentration of 150 ppb (6 1010 M). The solute tracer was analyzed by fluorescence spectrophotometry (Cary Eclipse Fluorescence Spectrophotometer, Varian, Palo Alto, CA). The tracer concentration was determined by transforming fluorescence results to concentration using calibration curves in solutions of similar compositions and salinities.
2.3.
Porous media
Natural sand originating from the Hatira Formation (Cretaceous) exposed in the Machtesh Ramon (Israel) was used as the granular porous media. The sand is commercially distributed by Machteshim LTD (Mizpe Ramon, Israel) after sieving to an average diameter of 0.8 mm and uniformity coefficient of 1.18 (d60/d10). It is composed of highly pure quartz sand (more than 99% SiO2), typically with <400e500 ppm of metal impurities (mostly iron, aluminum and titanium oxides). Sand pretreatment and purification procedures included washing with dilute, heated solutions of HCl, washing and rinsing with DI water, and drying in the oven at 50 C (Shani et al., 2008). Following the pretreatment, the quartz grains were packed in columns of two sizes, one a 7 cm-long polypropylene column with a 1.4 cm inner diameter, and the other a 30 cm-long Plexiglas column with a 5.5 cm inner diameter. The columns were dry-packed and the calculated porosity of the sand was of 37e38%.
2.4.
Colloid stability experiment
To assess the stability of the CML colloids used in the column experiment with respect to settling or floating, an experiment was conducted as a function of the density of solution and under hydrostatic conditions. Solutions of 2 ppm colloid concentration (3.64 107 colloids per ml) were prepared in DSW and DSW diluted to DSW/2, DSW/5, DSW/10, DSW/100 and DSW/1000 with DI. One L of the colloid suspension was placed in a beaker with a 10 cm depth of the water column. The beaker was strongly agitated to ensure a uniform colloid suspension prior to the initiation of the hydrostatic conditions. Next, samples of 2 mL were taken periodically (after 4, 14, 20, 30, 40, 55, 70, 95, 120, 145, 175, 205, 225 and 280 min) from the solution at three depths (<0.2 cm from the surface, 2.5 cm and 5 cm from the surface) in order to determine the extent of vertical transport of the CML colloids in the beaker.
2.5.
Colloid transport experiments
2.5.1.
30 cm columns
Solutions used in the long-column experiments were DSW and diluted DSW solutions (DSW/2, DSW/5, DSW/10, DSW/50, DSW/100, DSW/500, DSW/1000 and DSW/5000). At least two repetitions of the column experiments were made for each type of solution. The columns were set-up vertically, drypacked homogenously, and then first saturated from the bottom with CO2 for 20 min. The column inlet at the bottom
was connected to a three-point valve leading to two reservoirs, one containing the colloid suspension (colloids and conservative solute in DSW) and the other containing a colloid and tracer-free solution of DSW. The solution with colloids and tracer was continuously stirred in an opaque reservoir vessel, preventing light from entering the solution. Both solutions (with and without tracer/colloids) were pumped at a constant rate of 1 0.05 mL/min by a peristaltic pump (Gilson Minipuls 3, Gilson, Middleton, Wisconsin). Prior to the experiment, the column was flushed by the background solution (according to the salinity used in each experiment) with a solution volume identical to w3 pore volumes (PV). Next, the tracer solution was diverted into the column and was injected for 10e12 h (2.2e2.3 PVs) before the experiment was terminated. A fraction collector (Spectra/Chrom CF-1, Spectrum Laboratories, Houston, TX or Gilson FC 203B, Middleton, Wisconsin) was connected to the column outlet. Samples of 5 mL were collected from the column outflow at different frequencies over the course of the experiment every 15 min for the first 3.5 h, every 5 min for another 2.5 h, and then every 15 min until the end of the experiment. The 5 mL samples were collected in 10 mL opaque glass vials and were stored at 4 C until analysis, typically within 24 h from the end of the experiment. The flow rate in the column was measured periodically by weighing the effluent solution over a known time interval and accounting for solution density. The experiments were conducted at room temperature (25 2 C) and under dark conditions in order to prevent tracer photo-degradation.
2.5.2.
7 cm columns
Experiments were conducted for longer time periods (at least 100 h) in 7 cm columns. In these column experiments two solutions were used: DSW/5000 (IS ¼ 102.8 M) and slightly diluted DSW (90% DSW and 10% distilled water IS ¼ 100.9 M). A slight dilution of the DSW solution was necessary to prevent the precipitation of salts during the long experimental period. At least two repetitions were made for each solution chemistry condition tested. Apart from the columns’ size difference, the experimental procedure was conducted under the same conditions as those of the 30 cm column experiments described in Section 2.5.1. The procedure deviated only in the following ways: prior to the experiment, the column was flushed by the background solution for 7e8 PVs (versus the 3 PV of rinsing for the 30 cm-long column experiment) and the tracer solutions were applied for 100 h, (several thousand PVs) in contrast to w2 PV in the 30 cm-long column. Effluent collection was conducted as follows: once every 2 min during the first hour (2 mL aliquots), once every 5 min in the second hour (volume of 5 mL), and once every 9 min in the third hour (volume of 9 mL). Subsequently, the sampling frequency was decreased to once every 36 min for 8 h, followed by samples collected once every 100 min till the experiment terminated.
2.6.
Mathematical models
Two types of models were used in order to simulate the experimental results that are described below. The HYDRUS-1D model (Simunek et al., 2008) was used for simulating the results from the 30 cm-length column experiments. The experimental results on two scales (7 and 30 cm experiments)
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 5 2 1 e3 5 3 2
3525
were modeled using the more sophisticated model of limited entrapment (Pachepsky et al., 2006) assuming bimodal deposition coefficients of the colloids (as will be elaborated on below).
The mathematical model describes the bio-colloid transport according to Equation (1) with the mass balance equation for the bio-colloid attached to the solid phase given by:
2.6.1.
rb
HYDRUS-1D simulations
The HYDRUS-1D computer code (Simunek et al., 2008), that simulates water, colloid, and solute movement in onedimensional saturated porous media, was used for modeling the experimental results. HYDRUS-1D numerically solves the Fickian-based advectionedispersion equation with a nonlinear equilibrium and kinetic reactions. The code is coupled to a non-linear least-square optimization routine to find model parameters by fitting a simulated BTC to experimental BTC data. The transport of colloids through the sand columns was formulated as (Simunek et al., 2008): vC rb vS v2 C vC þ ¼ aL v 2 v vt ne vt vx vx
(1)
where C is the colloid concentration in the solution (Nc/L, Nc denotes number of colloids), S is the concentration of deposited colloids in the column (Nc/kg), t is the time (min), x is the distance (cm), aL is the dispersivity (cm), rb is the soil bulk density (kg/ dm3), ne is effective porosity, and v is the fluid velocity (cm/min). The mass balance for the colloids attached to the solid phase is given by (Simunek et al., 2008): rb
vS ¼ ne katt js C rb kdet S vt
Smax S Smax
(3)
where Smax is the maximum concentration of deposited colloids (on the solid phase, Nc/kg). It should be noted that the impact of Smax is mainly in advanced stages of the experiment after an injection of >2e3 PV of particle solution. The impact of Smax value is elaborated on below.
2.6.2.
(4)
where ka1 [min1] is the initial attachment rate that is applicable at early stages of transport when transport-related attachment has occurred, ka2 is the parameter responsible for the increase of the attachment rate with the increase in attached bacteria amount; a is a parameter which accounts for the acceleration of the attachment with the increase in the trapped amount of particles. The trapping rate grows linearly with the trapped amount when a ¼ 1. The trapping rate accelerates with the growth of S when a > 1. The trapping rate shows some retardation when a < 1 (Pachepsky et al., 2006). Note, that the original Pachepsky et al. (2006) model has been modified slightly by introducing a dimensionless colloid retention function js similar to that in Equation (2). Thus, if ka2 ¼ 0 the limited entrapment model is similar to that of the HYDRUS-1D. The least-square optimization was carried out to calculate the colloid transport parameters by fitting the model to the experimental results (inverse solution).
3.
Results and discussion
3.1.
Colloid stability
(2)
where katt is the colloid deposition coefficient (1/min), kdet is a first-order detachment rate coefficient (1/min) and js is a dimensionless function for deposited colloids. In order to simulate reduction in the attachment coefficient due to filling sorption sites on the sediment grains, js decreases with increasing adsorbed colloid mass as follows (Simunek et al., 2008): js ¼
vS ¼ ne ðka1 þ ka2 Sa Þjs C rb kdet S vt
Limited entrapment model simulations
The limited entrapment model (Pachepsky et al., 2006) was used for simulation of the experiments conducted in the 7 cm column. The model was developed for simulating transport of bio-colloid (bacteria) with bimodal breakthrough curves. The main assumption of the model (Pachepsky et al., 2006) is the existence of a limited attachment capacity of the sediments such that no bio-colloid attachment occurs after this capacity is reached. According to the model equations, the bio-colloid attachment rate is assumed to be a function of the number of previously attached bio-colloids. However, the attachment rate drops again as capacity for further attachment is exceeded. Consequently, the first maximum (or plateau) observed in the BTCs is consistent with the initial slow attachment rate (Pachepsky et al., 2006), while the second maximum (or plateau) is consistent with achieving attachment capacity.
In diluted DSW experiments (DSW/5000, IS ¼ 102.8 M), the concentration of colloids in the beaker was relatively uniform and stable over the 5 h duration of the experiment (with a coefficient of variance <3% for colloid concentration along the beaker). The visual appearance of colloids uniformity suspended in solution implies that colloids in the water column during the course of the experiment are quite stable and that gravitational settling is insignificant. The concentration of colloids was less uniform with time in the DSW experiment (IS ¼ 100.9 M). Specifically, an increase in colloid concentration (w20e30% higher value in comparison to the initial concentration) was observed at a depth of 0.2 cm after 3 h. A similar increase of colloid concentration was observed following 2 more hours, 2.5 cm below the water surface. It should be noted that at the same time, colloid concentration at the 5 cm depth remained quite constant across the entire 5 h experiment (with a coefficient of variance of 4.5% with time for colloid concentration with no distinctive trend). Upward transport of the colloids (density of 1.05 g/cm3) in the relatively dense solution of DSW (1.25 g/cm3) may occur; however, upward colloid movement could only be detected after relatively long time periods (3e5 h), suggesting any additional buoyancy of the colloids occurs at a relatively slow pace. This observation is supported by the low Stokes velocity (0.01 cm/h) for 1 mm colloids in DSW solution. To confirm whether the buoyancy of the colloids in the high IS solution contributed to transport in the column, BTCs of experiments conducted with both upward and downward vertical flows were compared. Quite similar BTCs were measured in these experiments (data not shown); hence, it was concluded that this mechanism is negligible in the dense DSW solution.
3526
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 5 2 1 e3 5 3 2
3.2.
Colloid transport experiments
3.2.1.
Transport in 30-cm columns
BTCs for colloids and the conservative tracer in solutions ranging across three orders of magnitude of IS are presented in Fig. 2. In the diluted solution, the colloids arrived at the outlet w0.2 PV (45 min) before the conservative tracer (IS ¼ 102.8 M or DSW/5000, Fig. 2a). The extent of early arrival of the colloids is reduced with the increase of IS, and in the concentrated DSW solution (IS ¼ 100.9 M, Fig. 2e) colloids and the conservative tracer arrived simultaneously. Early arrival of the colloids was detected in many studies (e.g., Bradford et al., 2003, 2006a; Kretzschmar et al., 1997; Mishurov et al., 2008; Nocito-Gobel and Tobiason, 1996; Shani et al., 2008) as was also the lack of correlation between preferential transport of colloids and the increase of IS (Harter et al., 2000; NocitoGobel and Tobiason, 1996). The transport velocity of colloids is known to be higher than that of the average water velocity in many cases. This occurs because not all pores are accessible for the colloids, which are excluded from the smaller pore throats (Harter et al., 2000). Subsequently, smaller effective pore volumes are available for colloid transport as compared to solution transport (Bradford et al., 2003, 2006a; Kretzschmar et al., 1997; Nocito-Gobel and Tobiason, 1996). Physical straining is unlikely considering the sand grain and colloid sizes (Bradford et al., 2006a). The early arrival of the colloids is more pronounced at low IS, due to the expansion of the electrical double layer at the solid surface (Harter et al., 2000). The shape of the BTCs in the concentrated DSW experiment is irregular, with the relative concentration (C/C0, C0 denotes concentration of colloids in the inlet solution) fluctuating between 0.18 and 0.38 (Fig. 2e). This seemingly
unstable phenomenon has not been observed in the diluted DSW experiments and is therefore considered to be related to the high salinity. The observed irregular shape of the BTC may be due to sporadic release of varying amounts of colloid aggregates in the column. The colloids in the concentrated DSW are likely to aggregate, since there is no energy barrier to colloid interaction at this high IS and every collision should result in attachment (Grolimund et al., 2001). Therefore, colloid aggregates may be presented in the injected suspension, as well as at the outlet due to creation of additional aggregates occurring in the column. An attempt was made to determine the colloid or aggregate size by a few techniques (dynamic light scattering and visualization under the microscope) with no conclusive results. Under the light microscope colloid aggregates were found in DSW solutions, while at the same time in the highly diluted solution of artificial rain water (Zvikelsky and Weisbrod, 2006) no aggregates has been inspected. This is attributed to the difficulties resulting from the complexity of dealing with suspensions in concentrated solutions such as DSW brines. Other routine methods, such as settling in a column or laser diffraction, could not be implemented for DSW solutions without major and inherent adjustments, which are subjects for future research. Comparing the shape of the colloid BTCs at the various IS conditions (Fig. 2) reveals that the breakthrough value (C/C0) after w1.5e2 PV reaches a maximum of 0.6 for IS < 101.8 M (DSW/500), while it is w0.3 for IS 101.1 M (DSW/100) (Fig. 2f). The corresponding colloid deposition rates were calculated based on filtration theory (e.g., Yao et al., 1971; Logan et al., 1995; Tufenkji and Elimelech, 2004) using the system physical parameters (such as collector diameter, column length,
Fig. 2 e BTCs of colloids and the conservative tracer in different solutions of DSW: (a) DSW/5000, IS [ 10L2.8 M, (b) DSW/500, IS [ 10L1.8 M, (c) DSW/100, IS [ 10L1.1 M, (d) DSW/10, IS [ 10L0.1 M, (e) DSW, IS [ 100.9 M, and (f) a compilation of all colloids BTCs.
3527
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 5 2 1 e3 5 3 2
porosity, fluid approach velocity). Here, the collector contact efficiency was assessed accounting also for the fluid and particles properties and the breakthrough curve plateau C/C0 (Tufenkji and Elimelech, 2004). Fig. 3 presents calculated deposition rate coefficients as a function of the solution IS up to IS ¼ 100.23 M (DSW/5). The deposition rate of colloids was w0.002 m1 for solutions up to IS of 101.8 M (DSW/500) and increased to w0.005 min1 for IS of 101.1 M (DSW/100) and higher (Fig. 3). Such an increase in colloid deposition rate with solution IS was expected according to the DLVO theory as the greater IS reduces the energy barrier for deposition of colloids (McDowell-Boyer et al., 1986). Previous experiments demonstrated the existence of an IS threshold value between the reaction-limited and transportlimited deposition regimes, the critical deposition concentration (CDC) (Grolimund et al., 2001; Kretzschmar et al., 1997). The CDC is in the range of 101 M for NaCl solutions and 102e103 M for CaCl2 solutions (Grolimund et al., 2001; Kretzschmar et al., 1997). Evaluation of the CDC in DSW is complicated since it is composed of substantial cation concentrations beyond simple Ca2þ and Naþ (Table 1). The CDC defined by the concentration of the monovalent salts in DSW is between 102.5 and 101.8 M (e.g., for solutions between DSW/500 and DSW/100) and is two orders of magnitude lower than the CDC of an artificial NaCl solution. Similarly, the CDC defined by the concentration of the DSW divalent ions is between 102.3 and 101.6 M and is on the same order of magnitude as the artificial CaCl2 solution. Grolimund et al. (2001) measured the CDC of colloids in mixed solutions of NaCl and CaCl2, at different ratios. The CDC of a mixed solution of Ca/Na with a molar ratio of 1.4 (similar to the ratio of mono to divalent cations in DSW) was w102.2 M, somewhat lower than the CDC of 101.9 M and 101.2 M in DSW/500 and DSW/100 suspensions respectively. The similarity between the CDC values of a simple mixture of salts and the complex composition of DSW indicates the role of cation composition on the attachment of colloids. This merits further study using natural (and complex) solutions
Fig. 3 e Average deposition rate of the colloids (dots) and standard deviations (bars) at different salinities as a function of the solution’s ionic strength in the 30 cm columns.
with varied cation compositions. Ocean water has a high salinity and a substantially different ionic composition with a di- to mono-valent cation ratio of 0.13 (compared to the ratio of 1.4 in DSW).
3.2.2.
HYDRUS-1D simulation results for the 30-cm columns
The above results from the 30 cm-long column experiments were inversely simulated successfully using the HYDRUS-1D model, as evidenced by the correlation coefficients (R2 of 0.69e1.00, Table 2). The colloid transport parameters derived from the simulation are presented in Table 2. The fitted values of the effective porosity was slightly lower than the measured porosity (0.36e0.37 and 0.37e0.39, respectively), possibly due to colloid size exclusion and the resulting phenomenon that colloids can only migrate through and access a fraction of the total voids. Colloid dispersivity was found to be on the same order of magnitude as determined for the experiments (Table 2). The colloid attachment coefficient (katt) at high IS was higher than at low IS, with the detachment coefficient (kdet) being at least an order of magnitude smaller than the attachment coefficient determined for the same solution conditions (Table 2). Colloid attachment is therefore considered an irreversible process, as was indicated in other studies (Compere et al., 2001; Johnson et al., 2007). HYDRUS-1D model simulations were highly insensitive in determining the maximum normalized concentration of deposited colloids, Smax ¼ Smax =N0c (N0c is the number of colloids in a unit volume of the influent colloid suspension
Table 2 e Colloid transport parameters calculated by fitting the HYDRUS-1D model for results obtained by the 30 cm column experiments. R2
kdet (min1)
Exp. Num.
ne
aL cm
Smax
katt (min1)
8.5
1 2 3a 4a
0.35 0.35 0.36 0.37
0.05 0.05 0.05 0.18
153 72 6.1 13.3
0.006 0.006 0.009 0.008
0 0 0 0
0.69 0.75 0.87 0.90
4.2
1 2
0.35 0.35
0.25 0.07
76 74
0.007 0.007
0 0
0.93 0.86
1.7
1 2
0.35 0.35
0.09 0.21
364 183
0.005 0.005
0 0
0.96 0.96
0.8
1 2
0.35 0.35
0.1 0.2
9 99
0.005 0.005
0 0
0.96 0.94
0.2
1 2
0.37 0.35
0.1 0.1
4 55
0.008 0.007
0.0003 0.0002
0.97 0.95
0.1
1 2
0.35 0.35
0.14 0.11
21 11
0.007 0.006
0.0002 0.0003
0.98 0.99
0.02
1 2
0.37 0.36
0.27 0.25
5 497
0.002 0.002
0 0
1.00 1.00
0.008
1 2
0.36 0.37
0.26 0.26
4 2
0.003 0.002
0 0.0007
0.99 0.99
0.002
1 2
0.36 0.36
0.26 0.27
1 30
0.002 0.002
0 0.0002
1.00 0.99
IS (M)
a Results of downward diversion of inlet solution experiment.
3528
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 5 2 1 e3 5 3 2
(Bradford et al., 2009)), as was demonstrated by two substantially different resulting fitted values for two repetitions of the same experiment (IS ¼ 0.02 M or DSW/500, Table 2). In order to understand the significance of Smax on colloid transport, the HYDRUS-1D model was used for forward simulations of the colloid BTCs. Simulations were conducted using the transport parameters obtained from fitting the DSW experimental results in the 30 cm-long column (aL ¼ 0.2 cm, katt¼0.008 min1, kdet ¼ 5 107 min1) and were repeated with several values of Smax (0.02, 0.2, 1 and 10). The simulations were also conducted for the 7 cm-long column, as described further below. The forward simulations (Fig. 4) indicated that at low Smax (0.02), the BTC shape is similar to that of the conservative tracer (Fig. 4); whereas, for higher Smax values, after the first increase to C/C0 of 0.6, the relative concentration gradually increases and approaches a value of one. The value of Smax is related to the time at which the C/C0 reaches a value of 1; for example, for greater Smax value a longer elution time is needed to achieve C/ C0 ¼ 1 (Fig. 4). The forward simulations demonstrated the Smax value influences the shape of BTC, especially at the advanced stages of the experiment after eluting colloid suspension of tens PV. The evaluation of Smax is highly important for understanding long term changes in colloid transport, especially for the evaluation of these processes in natural environments in time-scales of years rather than hours (or days) in the laboratory. Consequently, to accurately determine Smax the experiment should be significantly longer. It was decided, therefore, to repeat part of the experiments, using a small diameter column of 7 cm length, to reduce the elution time of one pore volume from hours (in the 30 cm-long column) to minutes (in the 7 cm-long column). This way experiments could be conducted over 100 s of PV over a reasonable period of time. These experiments are discussed below.
3.2.3.
Transport experiments in 7-cm columns
The short column (7 cm) experiments were conducted at two extreme salinity levels, low IS DSW (diluted by 5000, Fig. 5a) and a solution of highly concentrated DSW (Fig. 5b).
Fig. 4 e HYDRUS-1D forward simulations of colloid transport in DSW solution in a 7 cm-long column at different Smax values (lines for values of 0.02e10) and experimental results of colloid transport (circles).
Comparing longer duration experimental in the 7 cm columns (Fig. 5a,b) and the shorter duration experiments on the 30 cm columns (Fig. 5c,d) reveals some notable differences. First, the relative concentration of colloids eluted from the 7 cm column after injection of 2e3 pore volumes was greater than that from the 30 cm columns (C/C0 of 1 and 0.5 in the dilute experiment and 0.8 and 0.2 in the DSW experiment, for 7 and 30 cm columns, respectively). The value of C/C0 was confirmed to be directly related to the length of the column or to the pore volume of the column (Brown and Abramson, 2006). Another important trend was in the DSW experiments in the 7 cm column, after the first initial increase of the eluted colloids (C/C0) there was another increase approaching a value of one observed after w18 PVs. This two-step increase in C/C0 values could not be simulated by the HYDRUS-1D, which predicted for an identical experimental set-up a first initial breakthrough at C/C0 ¼ 0.6 followed by a more gradual increase of the C/C0 to the value of one (Fig. 4). All the attempts to model these results with the HYDRUS-1D model resulted in poor correlation coefficients. This bimodal behavior was observed in five repetitions of the DSW experiments in the 7 cm columns (not presented). Similar bimodal BTCs were reported previously for a few types of bio-colloids: Cryptosporidium (Harter et al., 2000), viruses (Jin et al., 1997) and bacteria (Pachepsky et al., 2006, Tufenkji et al., 2003). A bimodal shape of the colloid BTCs suggests a complex deposition process occurring in two stages. At the beginning of the experiment the fast increase of C/C0 up to 0.8 results from a low initial rate of colloidal deposition. As the colloid suspension continues to be injected (to approximately 18 PVs), the deposition rate increases further, leading to flattening of the BTC or even to a slight decrease of the relative concentration (C/C0). Subsequently, the deposition rate decreases and the eluted C/C0 increases again and approaches a value of one, possibly when the entrapment capacity has been exceeded. The exact point of transition between these stages is unknown. Jin et al. (1997) speculated that the second increase of virus concentration (followed the achievement of an initial steady-state value for w10 PV) reflects complete filling of the available retention space on the sediment surfaces. The available retention space for colloid deposition, however, is not the entire surface of the sediment grains. The surface area that is occupied by a monolayer of the colloids (deposited during the 18 PV of solution injected in to the column) is less than 0.1% of the total sediment grain surface area (103 m2 and 3.5 m2, cumulative surface of colloids and surface area of the sand, respectively). Jin et al. (1997) speculated that the active surfaces for the bio-colloid (virus) deposition are correlated with patches of positively charged sites on the edges of the sediment grain that occupy a small fraction of the sediment surface. Yet, in the high salinities of the DSW, these impurities would be insignificant due to the overall favorable conditions for colloid attachment. Also, Torkzaban et al. (2008b) demonstrated the existence of hydrodynamically disconnected regions in which the majority of colloid deposition takes place. A bimodal breakthrough curve may result, alternatively, from elution of colloid aggregates of two sizes that are retarded differently by the column due to size exclusion (Tufenkji et al., 2003).
3529
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 5 2 1 e3 5 3 2
Fig. 5 e Comparison between the results of colloid transport and model simulations of 7 cm-long and 30 cm-long columns: (a) diluted solution of DSW/5000, 7 cm column, and (b) DSW solution, 7 cm column; (c) diluted solution of DSW/1000; 30 cm column and (d) DSW solution, 30 cm column. The limited entrapment model simulation result is sketched as lines: continuous line for the conservative tracer and dashed line for the colloids.
3.2.4. model
a pronounced bimodality of the BTC is observed (Pachepsky et al., 2006). The shape of colloid BTCs observed in the experiment for the 7 cm column with DSW is most similar to those calculated when a ¼ 1. Therefore, we assumed the linear growth in colloid deposition rate with an increasing amount of attached colloids. The prescribed a value was also used to improve the uniqueness of the inverse solution. Thus, the limited entrapment model includes one additional parameter (ka2), which accounts for the enhanced deposition rate with the increase of the deposited particles concentration. Model parameters found by best fit are presented in Table 3. Large values of calculated dispersivity could be related to the mixing in a relatively large volume of solution sample (2 mL) compared to PV (w4.3 mL). The values of the detachment (kdet) and the initial attachment (ka1) rates in the experiments with DSW are an order of magnitude larger than those found for the longer column, while Smax is smaller for the 7 cm column
Modeling colloid BTCs using a limited entrapment
The conventional model (Equations (1) and (2)) did not provide fair agreement for the observed BTC’s of colloids in solution with high salinity after eluting more than 10 pore volumes. Therefore, the limited entrapment model was applied to perform inverse simulations of the BTCs obtained with 7 cmlong columns. A good fit was obtained between simulated and observed concentrations (Fig. 5), as evidenced by the high correlation coefficients (R2 of 0.87e0.95, Table 3). Simulations were carried out assuming the acceleration parameter, a, is equal to 1. This was based on the sensitivity analysis (Pachepsky et al., 2006) and the shape of the observed colloid BTCs. Following the results of model sensitivity, as the value of a decreases, less acceleration in colloid attachment occurs and an almost unimodal BTC is seen with a ¼ 0.5. For a close to 1, a first peak of concentration appears, and for a > 1
Table 3 e Colloid transport parameters calculated by fitting the limited entrapment model for results obtained by the 7 cmlong column experiments. Ionic strength (M)
8.5 0.002
Tracers
Colloids
aL (cm)
ne
R2
aL (cm)
ne
kdet (min1)
ka1 (min1)
ka2 (min1)
Smax
R2
2.1 1.4 1.1
0.39 0.37 0.37
0.99 0.99 0.97
2.1 1.4 1.1
0.33 0.35 0.32
0.0088 0.0015 0.0000
0.027 0.028 0.028
0.444 0.363 0
0.54 0.35 18.01
0.85 0.89 0.91
3530
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 5 2 1 e3 5 3 2
(Table 3). In the experiment with DSW/5000 it was determined that ka1 ¼ 0.028 min1, which is similar to the values found in the experiments with DSW. Such high values of this parameter are explained by the substantially greater flow velocity in the 7 cm-long column experiments (0.74 cm/min) in comparison to the velocity in the 30 cm-long column experiments (0.046 cm/min). The theoretical value of ka1 ¼ 0.0157 min1 (calculated using the filtration theory for DSW/5000) is almost half. Bradford et al. (2006a) noted that the predicted attachment rate may significantly underestimate deposition with increasing size of the sand or colloids, thus suggesting that the collector contact efficiency is not a complete descriptor of attachment and another mechanism may be involved in the colloid deposition. The values of ka2 are responsible for the increasing deposition rate with time and are presented in Table 3. They were estimated from the colloid breakthrough curves using a nonlinear optimization procedure. Constant or increased colloid deposition rates with time at high ionic strength are observed when attached colloids can act as additional collectors for the attachment leading to the formation of multilayer films on matrix surfaces (e.g., filter ripening) (Rajagopalan and Chu, 1982; Liu et al., 1995; Kuhnen et al., 2000). Kuhnen et al. (2000) developed a model that accounts for the additional deposition rate of colloid particles to previously deposited colloids attached to matrix surfaces. They demonstrated that at moderately high IS values (up to 101 M), the model accurately describes multilayer deposition assuming constant rate of particleeparticle deposition. Unlike the model of Kuhnen et al. (2000), Equation (4) accounts for increase in the overall deposition rate with time due to an increase in concentration of the deposited colloids and, as a result, an increase in the effective surface area available for colloid capture. Modeling results demonstrate that at very high IS values the ripening mechanism could be important.
4.
Conclusions
Colloid transport was found to decrease with greater solution IS, in natural solutions with salinities ranging widely from the hyper-saline brines of the Dead Sea to diluted solutions similar to fresh groundwater. Colloid transport on a column scale was found to be non-negligible even in the brines of DSW, and 30e90% of the colloids are eluted from the 30 and 7 cm-long columns, respectively. The current experimental results were found to be comparable to previously conducted, similar experiments using artificial solutions of a single salt (Grolimund et al., 2001; Kretzschmar et al., 1997). Like the single-salt solution studies, a threshold value of’ IS was found, above which the colloid deposition rate is constant (achieved at the critical deposition concentration, CDC). It was found that comparing the CDC of natural (and complex) solutions of the DSW to an artificial single-salt solution is possible only by accounting for the sum of the mono- or di-valent cations in solution. The values are generally in agreement, although the ratio between the mono- and di-valent ions in the solution should be taken into account due to its major impact on the CDC of the solution (Grolimund et al., 2001).
In the 7 cm-long column experiments that were long enough to enable achieving a steady-state (C/C0 ¼ 1), colloid transport in high salinities of DSW was found to be complex and composed of two stages e low initial attachment followed by an even lower stage of deposition rate. The bimodal shape of the colloid breakthrough curve was found to be quite similar to previously reported breakthrough curves of biocolloids in certain situations. The HYDRUS-1D model could only simulate the colloid transport experimental results before the achievement of C/C0 ¼ 1, but failed in simulating the bimodal BTC of the colloids in DSW. These experimental results were successfully simulated only by using the limited entrapment model (Pachepsky et al., 2006) developed for the transport of bacteria through a porous media. The current work shows that colloid deposition involves a two-step process in high salinities and that the exploration of colloid migration using a single-salt solution cannot truly represent natural conditions. Further work is needed in order to understand the role of the solution salt composition on the transport of colloids, especially the role of the balance between mono- and di-valent cations in the solution.
references
Abelson, M., Baer, G., Shtivelman, V., Wachs, D., Raz, E., Crouvi, O., Kurzon, I., Yechieli, Y., 2003. Collapse-sinkholes and radar interferometry reveal neotectonics concealed within the Dead Sea basin. Geophysical Research Letters 30 (10), 1545. doi:10. 1029/2003GL017103. Ahfir, N.-D., Wang, H., Benamar, A., Alem, A., Massei, N., Dupont, J.-P., 2007. Transport and deposition of suspended particles in saturated porous media: hydrodynamic effect. Hydrogeology Journal 15, 659e668. doi:10.1007/s10040-0060131-3. Amit, O., Bentor, Y.K., 1971. pH-dilution curves of saline waters. Chemical Geology 7, 307e313. Arkin, Y., Gilat, A., 2000. Dead Sea sinkholes e an ever-developing hazard. Environmental Geology 39, 711e722. doi:10.1007/ s002540050485. Blume, T., Weisbrod, N., Selker, J.S., 2002. Permeability changes in layered sediments: impact of particle release. Ground Water 40, 466e474. doi:10.1111/j.1745-6584.2002.tb02530.x. Blume, T., Weisbrod, N., Selker, J.S., 2005. On the critical salt concentrations for particle detachment in homogeneous sand and heterogeneous Hanford sediments. Geoderma 124, 121e132. doi:10.1016/, Geoderma.2004.04.007. Bradford, S.A., Simunek, J., Bettahar, M., Van Genuchten, M.T., Yates, S.R., 2003. Modeling colloid attachment, straining, and exclusion in saturated porous media. Environmental Science and Technology 37, 2242e2250. doi:10.1021/es025899u. Bradford, S.A., Simunek, J., Bettahar, M., Van Genuchten, M.T., Yates, S.R., 2006a. Significance of straining in colloid deposition: evidence and implications. Water Resources Research 42. doi:10.1029/2005WR004791, pp. W12S15. 1eW12S15.16. Bradford, S.A., Simunek, J., Walker, S.L., 2006b. Transport and straining of E. coli O157:H7 in saturated porous media. Water Resources Research 42, W12S12. doi:10.1029/2005 WR004805. Bradford, S.A., Torkzaban, S., 2008. Colloid transport and retention in unsaturated porous media: a review of interface-, collector-, and pore-scale processes and models. Vadose Zone Journal 7, 667e681. doi:10.2136/vzj2007.0092.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 5 2 1 e3 5 3 2
Bradford, S.A., Torkzaban, S., Walker, S.L., 2007. Coupling of physical and chemical mechanisms of colloid straining in saturated porous media. Water Research 41, 3012e3024. doi: 10.1016/j.watres.2007.03.030. Bradford, S.A., Torkzaban, S., Leij, F., Simunek, J., van Genuchten, M.T., 2009. Modeling the coupled effects of pore space geometry and velocity on colloid transport and retention. Water Resources Research 45, W02414. doi:10.1029/ 2008WR007096. Brown, D.G., Abramson, A., 2006. Collision efficiency distribution of a bacterial suspension flowing through porous media and implications for field-scale transport. Water Research 40, 1591e1598. doi:10.1016/j.watres.2006.02.016. Compere, F., Porel, G., Delay, F., 2001. Transport and retention of clay particles in saturated porous media. Influence of ionic strength and pore velocity. Journal of Contaminant Hydrology 49, 1e21. doi:10.1016/S0169-7722(00)00184-4. Elimelech, M., Gregory, J., Jia, X., Williams, R.A., 1995. Particle Deposition and Aggregation e Measurement, Modeling and Simulation. Elsevier, 441 p. Grolimund, D., Elimelech, M., Borkovec, M., 2001. Aggregation and deposition kinetics of mobile colloidal particles in natural porous media. Colloids and Surfaces A: Physicochemical and Engineering Aspects 191, 179e188. doi:10.1016/S0927-7757(01) 00773-7. Harter, T., Wagner, S., Atwill, E.R., 2000. Colloid transport and filtration of Cryptosporidium parvum in sandy soils and aquifer sediments. Environmental Science and Technology 34, 62e70. doi:10.1021/es990132w. Jin, Y., Yates, M.V., Thompson, S.S., Jury, W.A., 1997. Sorption of viruses during flow through saturated sand columns. Environmental Science and Technology 31, 548e555. doi:10. 1021/es9604323. Johnson, W.P., Li, X., Assemi, S., 2007. Deposition and reentrainment dynamics of microbes and non-biological colloids during non-perturbed transport in porous media in the presence of an energy barrier to deposition. Advances in Water Resources, Biological Processes in Porous Media: From the Pore Scale to the Field 30, 1432e1454. doi:10.1016/j. advwatres .2006.05.020. Khilar, K.C., Fogler, H.S., 1984. The existence of a critical salt concentration for particle release. Journal of Colloid and Interface Science 101, 214e224. doi:10.1016/0021-9797(84)90021-3. Khilar, K.C., Fogler, H.S., Ahluwalia, J.S., 1983. Sandstone water sensitivity: existence of a critical rate of salinity decrease for particle capture. Chemical Engineering Science 38, 789e800. Kiro, Y., Yechieli, Y., Lyakhovsky, V., Shalev, E., Starinsky, A., 2008. Time response of the water table and saltwater transition zone to a base level drop. Water Resources Research 44. doi:10.1029/2007WR006752. Ko, C.-H., Elimelech, M., 2000. The “Shadow Effect” in colloid transport and deposition dynamics in granular porous media. Measurements and Mechanisms: Environmental Science and Technology 34, 3681e3689. doi:10.1021/es0009323. Kretzschmar, R., Barmettler, K., Grolimund, D., Yan, Y.-D., Borkovec, M., Sticher, H., 1997. Experimental determination of colloid deposition rates and collision efficiencies in natural porous media. Water Resources Research 33, 1129e1137. Kretzschmar, R., Borkovec, M., Grolimund, D., Elimelech, M., 1999. Mobile subsurface colloids and their role in contaminant transport. Advances in Agronomy 66, 121e193. Kretzschmar, R., Sticher, H., 1997. Transport of humic-coated iron oxide colloids in a sandy soil: Influence of Ca2þ and trace metals. Environmental Science and Technology 31, 3497e3504. Kuhnen, F., Barmettler, K., Bhattacharjee, S., Elimelech, M., Kretzschmar, R., 2000. Transport of iron oxide colloids in packed quartz sand media. Monolayer and Multilayer
3531
Deposition: Journal of Colloid and Interface Science 231, 32e41. doi:10.1006/jcis.2000.7097. Liu, D., Johnson, P.R., Elimelech, M., 1995. Colloid deposition dynamics in flow through porous media: role of electrolyte concentration. Environmental Science and Technology 29, 2963e2973. doi:10.1021/es00012a012. Logan, B.E., Jewett, D.G., Arnold, R.G., Bouwer, E.J., O’Melia, C.R., 1995. Clarification of clean-bed filtration models. Journal of Environmental Engineering 21, 869e873. Magal, E., Weisbrod, N., Yakirevich, A., Yechieli, Y., 2008. The use of fluorescent dyes as tracers in highly saline groundwater. Journal of Hydrology 358, 124e133. doi:10.1016/j.jhydrol. 2008. 05.035. Magal, E., 2011, Flow and Transport Processes in an Extremely Saline Aquifers, Example from the Dead Sea Area, Ph.D. thesis. Ben-Gurion University of the Negev, Beer-Sheva, Israel. McDowell-Boyer, L.M., Hunt, J.R., Sitar, N., 1986. Particle transport through porous media. Water Resources Research 22, 1901e1921. Mishurov, M., Yakirevich, A., Weisbrod, N., 2008. Colloid transport in a heterogeneous partially saturated sand column. Environmental Science and Technology 42, 1066e1071. doi:10. 1021/es071780f. Nocito-Gobel, J., Tobiason, E., 1996. Effect of ionic strength on colloid deposition and release. Colloid and Surfaces 107, 223e231. doi:10.1016/0927-7757(95)03340-8. Pachepsky, Y.A., Devin, B.A., Polyanskaya, L.M., Shelton, D.R., Shein, E.V., Guber, A.K., 2006. Limited entrapment model to simulate the breakthrough of Arthrobacter and Aquaspirillum in soil columns. International Agrophysics 20, 207e218. Rajagopalan, R., Chu, R.Q., 1982. Dynamics of adsorption of colloidal particles in packed beds. Journal of Colloid and Interface Science 86 (2), 299e317. Roy, S.B., Dzombak, D.A., 1996. Colloid release and transport processes in natural and model porous media. Colloid and Surfaces 107, 245e262. doi:10.1016/0927-7757(95)03367-X. Ryan, J.N., Elimelech, M., 1996. Colloid mobilization and transport in groundwater. Colloids and Surfaces A: Physicochemical and Engineering Aspects 107, 1e56. doi:10.1016/0927-7757(95) 03384-X. Saiers, J.E., Ryan, J.N., 2006. Introduction to special section on colloid transport in subsurface environments. Water Resources Research 42, W12S01. doi:10.1029/2006WR005620. Sen, K.T., Khilar, K.C., 2006. Review on subsurface colloids and colloid-associated contaminant transport in saturated porous media. Advances in Colloid and Interface Science 119, 71e96. doi:10.1016/j.cis.2005.09.001. Shani, C., Weisbrod, N., Yakirevich, A., 2008. Colloid transport through saturated sand columns: influence of physical and chemical surface properties on deposition. Colloids and Surfaces A: Physicochemical and Engineering Aspects 316, 142e150. doi:10.1016/j.colsurfa. 2007. 08.047. Shen, C., Huang, Y., Li, B., Jin, Y., 2008. Effects of solution chemistry on straining of colloids in porous media under unfavorable conditions. Water Resources Research 44, W05419. doi:10.1029/2007WR006580. Simunek, J., Sejna, M., Saito, H., Sakai, M., Van Genuchten, M.T., 2008. The HYDRUS-1D Software Package for Simulating the One-dimensional Movement of Water, Heat, and Multiple Solutes in Variably-saturated Media. University of CaliforniaRiverside Research Reports, Riverside, California, p. 281. Song, L., Elimelech, M., 1993. Dynamics of colloid deposition in porous media: modeling the role of retained particles. Colloids and Surfaces A: Physicochemical and Engineering Aspects 73, 49e62. Starinsky, A. (1974) Relationship between Ca-chloride brines and sedimentary rocks in Israel, Ph.D. thesis. (in Hebrew, English summary), Hebrew University, Jerusalem.
3532
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 5 2 1 e3 5 3 2
Torkzaban, S., Bradford, S.A., van Genuchten, M.T., Walker, S.L., 2008a. Colloid transport in unsaturated porous media: the role of water content and ionic strength on particle straining. Journal of Contaminant Hydrology 96, 113e127. doi:10.1016/j. jconhyd.2007.10.006. Torkzaban, S., Tazehkand, S.S., Walker, S.L., Bradford, S.A., 2008b. Transport and fate of bacteria in porous media: coupled effects of chemical conditions and pore space geometry. Water Resources Research 44, W04403. doi:10.1029/ 2007WR006541. Tufenkji, N., Redman, J.A., Elimelech, M., 2003. Interpreting deposition patterns of microbial particles in laboratory-scale column experiments. Environmental Science and Technology 37 (3), 616e623. doi:10.1021/es025871i. Tufenkji, N., Elimelech, M., 2004. Correlation equation for predicting single-collector efficiency in physicochemical
filtration in saturated porous media. Environmental Science and Technology 38, 529e536. doi:10.1021/es034049r. Tufenkji, N., Elimelech, M., 2005. Breakdown of colloid filtration theory: role of the secondary energy minimum and surface charge heterogeneities. Langmuir 21, 841e852. doi:10.1021/ la048102g. Yao, K.M., Habibian, M.T., O’Melia, C.R., 1971. Water and waste water filtration concepts and applications. Environmental Science and Technology 5, 1105e1112. Yechieli, Y., 2006. The response of the groundwater system to changes in the Dead Sea level. In: Enzel, Y., Agnon, A., Stein, M. (Eds.), New Frontiers in Dead Sea Paleoenvironmental Research. Geological Society of America, Special Paper, pp. 113e126. Zvikelsky, O., Weisbrod, N., 2006. Impact of particle size on colloid transport in discrete fractures. Water Resources Research 42, W12S08. doi:10.1029/2006WR004873.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 5 3 3 e3 5 4 0
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Removal of residual dissolved methane gas in an upflow anaerobic sludge blanket reactor treating low-strength wastewater at low temperature with degassing membrane Wasala M.K.R.T.W. Bandara a, Hisashi Satoh a,*, Manabu Sasakawa b, Yoshihito Nakahara b, Masahiro Takahashi a, Satoshi Okabe a a b
Division of Environmental Engineering, Faculty of Engineering, Hokkaido University, North-13, West-8, Sapporo 060-8628, Japan Mitsubishi Rayon Co. Ltd., 1-2, Ushikawa-dori 4-chome, Toyohashi-shi, Aichi 440-8601, Japan
article info
abstract
Article history:
In this study, we investigated the efficiency of dissolved methane (D-CH4) collection by
Received 17 February 2011
degasification from the effluent of a bench-scale upflow anaerobic sludge blanket (UASB)
Received in revised form
reactor treating synthetic wastewater. A hollow-fiber degassing membrane module was
29 March 2011
used for degasification. This module was connected to the liquid outlet of the UASB reactor.
Accepted 15 April 2011
After chemical oxygen demand (COD) removal efficiency of the UASB reactor became
Available online 22 April 2011
stable, D-CH4 discharged from the UASB reactor was collected. Under 35 C and a hydraulic
Keywords:
63 mg COD L1 to 15 mg COD L1; this, in turn, resulted in an increase in total methane
Dissolved methane gas
(CH4) recovery efficiency from 89% to 97%. Furthermore, we investigated the effects of
Degassing membrane
temperature and HRT of the UASB reactor on degasification efficiency. Average D-CH4
Anaerobic wastewater treatment
concentration was as high as 104 mg COD L1 at 15 C because of the higher solubility of
Low temperature
CH4 gas in liquid; the average D-CH4 concentration was reduced to 14 mg COD L1 by
Low-strength wastewater
degasification. Accordingly, total CH4 recovery efficiency increased from 71% to 97% at
retention time (HRT) of 10 h, average D-CH4 concentration could be reduced from
15 C as a result of degasification. Moreover, degasification tended to cause an increase in particulate COD removal efficiency. The UASB reactor was operated at the same COD loading rate, but different wastewater feed rates and HRTs. Although average D-CH4 concentration in the UASB reactor was almost unchanged (ca. 70 mg COD L1) regardless of the HRT value, the CH4 discharge rate from the UASB reactor increased because of an increase in the wastewater feed rate. Because the D-CH4 concentration could be reduced down to 12 1 mg COD L1 by degasification at an HRT of 6.7 h, the CH4 recovery rate was 1.5 times higher under degasification than under normal operation. ª 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
Anaerobic wastewater treatment is a well-established and proven technology for the treatment of various categories of industrial wastewaters (Cakir and Stenstrom, 2005). This
technology has numerous advantages, such as low energy requirement and energy recovery as methane (CH4) gas, over aerobic wastewater treatment systems. Most anaerobic wastewater treatments have been conducted within mesophilic (30e40 C) or thermophilic (45e60 C) temperature
* Corresponding author. Tel./fax: þ81 11 706 6277. E-mail address:
[email protected] (H. Satoh). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.04.030
3534
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 5 3 3 e3 5 4 0
ranges (Kashyap et al., 2003; Dhaked et al., 2010; Fang et al., 2006; Lettinga et al., 2001). This is attributed to the fact that most of the biological reactions responsible for anaerobic biodegradation of organic matter proceeds much slower under psychrophilic (<20 C) conditions than under mesophilic conditions. However, domestic wastewater and a variety of industrial wastewaters, such as those from bottling, malting, and soft drinks manufacturing plants and breweries, are generally discharged at low ambient temperatures under temperate climatic conditions. Furthermore, domestic sewage wastewaters belong to the category of “lowstrength wastewaters” that have a chemical oxygen demand (COD) concentration of ca. 1.0 g L1 or lower. Therefore, a significant input of energy is required to heat the reactor to the treatment temperature (Lettinga et al., 2001; Angenent et al., 2001). If anaerobic wastewater treatment without heating the reactors can be applied to low-strength wastewater, the cost of anaerobic wastewater treatment can be reduced, thereby making this technology an attractive option for the treatment of a variety of wastewater categories. Improvements in reactor design and operational conditions have helped overcome many of the disadvantages of anaerobic wastewater treatment that limited its application to high-strength wastewater treatment at mesophilic or thermophilic temperatures and have allowed for successful operation of anaerobic wastewater treatment reactors at low temperatures (Dhaked et al., 2010; Lettinga et al., 1999, 2001; Trzcinski and Stuckey, 2010; Luostarinen and Rintala, 2005; Madden et al., 2010; Xing et al., 2009; McKeown et al., 2009). In addition, several studies have focused on anaerobic wastewater treatment of low-strength wastewaters at lower temperatures (Elmitwalli et al., 2003; Angenent et al., 2001; Gomec et al., 2008; Matsushige et al., 1990). A drop in temperature causes a change in the physical and chemical properties of wastewater, and this can significantly affect the reactor performance. For instance, dissolved methane gas (D-CH4) might play an important role in energy recovery efficiency of the reactor; however, to date, the role of D-CH4 has been overlooked. Discharge of residual D-CH4 in the reactor effluent represents a loss of energy that may be recovered; in addition, D-CH4 is a source of CH4, a greenhouse gas, that may be released into the environment (Hartley and Lant, 2006; Hatamoto et al., 2010; Matsuura et al., 2010). This knowledge is particularly important when low-strength wastewaters are anaerobically treated at low temperature, because the solubility of CH4 in the liquid phase increases with a decrease in temperature. Few studies have investigated the removal of D-CH4 in anaerobic wastewater treatment processes by biological oxidation, physical gasification based on gaseliquid equilibrium, and mixing with gas or a paddle (Hartley and Lant, 2006; Hatamoto et al., 2010; Matsuura et al., 2010; Pauss et al., 1990). However, the recovery efficiency of D-CH4 was low and/or the recovered CH4 gas composition was low in these processes. Hence, another technology is required for the removal of D-CH4 (Voolapalli and Stuckey, 1998). In this study, we employed a hollow-fiber membrane to recover residual D-CH4 in the effluent of an anaerobic wastewater treatment reactor by degasification. A bench-scale upflow anaerobic sludge blanket (UASB) reactor was operated, and the liquid outlet of the UASB reactor was connected
to another reactor for degasification. After achieving stable COD removal efficiency, the D-CH4 discharged from the UASB reactor was recovered by the degassing membrane (DM) reactor. In addition, we investigated the effects of temperature and hydraulic retention time (HRT) of the UASB reactor on the efficiency of degasification.
2.
Materials and methods
2.1.
Experimental setup and operational conditions
A bench-scale UASB reactor (height, 40 cm; diameter, 7 cm; working volume, 1.3 L) was operated in this study. The reactor was inoculated with 0.3 L of anaerobic granular sludge, which had total and volatile solids concentrations of 28 g L1 and 22 g L1, respectively, obtained from a full-scale UASB reactor treating wastewater from an isomerized sugar-processing plant. The UASB reactor was covered with a water jacket to maintain the reactor at a constant temperature (15, 25, or 35 C). The reactor was fed with synthetic wastewater containing powdered skim milk as carbon and energy sources, inorganic salts, and trace metals (Satoh et al., 2007). COD concentration was controlled by changing the concentration of powdered milk. HRT was adjusted by altering the wastewater feed rate. These operational conditions are summarized in Table 1. pH of the bulk liquid in the UASB reactor was maintained by adding NaOH at 7.6 after 51 days. After gas production reached a steady state, dissolved gas in the liquid was collected by the DM. A reactor for degasification (height, 30 cm; diameter, 7 cm; working volume, 1.1 L) was connected to the liquid outlet of the UASB reactor to collect the residual D-CH4 in the UASB effluent by degasification. A hollow-fiber membrane module (a multi-layered composite hollow-fiber membrane; MHF0504MBFT) provided by the Mitsubishi Rayon Co., Ltd. (Tokyo, Japan) was installed in the DM reactor. The DM reactor was completely filled with the wastewater treated in the UASB reactor. The liquid in the DM reactor was not mixed. The characteristics of the DM module are summarized in Table 2. The dissolved biogas diffuses into the lumen of the hollow fibers of the DM under vacuum generated using a vacuum pump (Model APN-110KV1; Iwaki Co., Ltd., Tokyo, Japan). Although gas molecules can pass through the non-porous layer of the membrane, liquids cannot. Thus, the DM effectively separates dissolved gas from the liquid. The DM reactor was operated at same temperatures as the UASB reactor. Transmembrane pressures were set at 50 kPa and 80 kPa (absolute pressure) by using a vacuum gauge, and HRTs of the DM reactor were altered in accordance with variations in the HRTs of the UASB reactor (Table 1). The operation of the DM reactor without degasification is referred to as normal operation.
2.2.
Analytical methods
The concentrations of total COD (T-COD) and dissolved fraction of COD (D-COD) after filtration with a 0.45-mm-pore-size membrane (Advantec Co., Ltd., Tokyo, Japan) were measured using a Hach method (Method 8000). Particulate COD (P-COD) concentration was calculated by subtracting the D-COD
3535
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 5 3 3 e3 5 4 0
Table 1 e Summary of operational conditions and reactor performances in the UASB and DM reactors. Phase 1.1 1.2 1.3 1.4 2.1 2.2 2.3 3.1 3.2 3.3 4 5.1 5.2 5.3 6.1 6.2 6.3 7
Influent T-COD concentration (mg/L) 1650 1580 1620 1640 1530 1410 1390 1390 1340 1400 1460 810 880 950 450 480 500 1510
132 95 48 21 13 0 18 37 30 55 4 48 5 46
T-COD loading rate (mg/L/h) 150 150 160 160 150 140 140 130 130 140 150 110 89 96 45 48 50 150
HRT of UASB (hour)
10 4 10
10.9 10.3 10.0 10.4 10.2 10.1 10.2 10.3 10.1 10.0 10.0 5.5 6.6 6.6 3.3 3.3 3.3 9.9
7 5 7 4
4 23 5 1 5 1
5
0.3 0.4 0.3 0.4 0.3 0.6 0.2
0.1
Phase
CH4 CH4 CH4 discharge rate CH4 discharge evolution collection from UASB (mgrate from DM rate rate COD/L/day) (mg-COD/L/day) (mg-COD/ (mg-COD/ L/day) L/day)
1.1 1.2 1.3 1.4 2.1 2.2 2.3 3.1 3.2 3.3 4 5.1 5.2 5.3 6.1 6.2 6.3 7
960 220 1000 94 1120 110 1010 230 880 91 930 69 580 93 530 25 700 73 720 1390 160 1220 100 1350 47 1150 190 830 63 770 33 960 280 810 540
180 85 210 26
300 84
370 100
480 130
370 12
140 14 150 7 150 4 150 3 190 16 210 27 190 3 240 19 240 7 250 20 160 12 240 14 250 19 240 14 480 35 500 41 510 19 150 125
110 12 49 10 35 11 110 4 160 13 32 11 160 12 220 24 33 4 230 6 150 7 230 11 45 4 210 12 460 31 370 25 410 17 140 10
concentration from the T-COD concentration. The concentrations of volatile fatty acids (VFAs; formate, acetate, propionate, lactate, i-butyrate, and n-butyrate) were determined by using a high-performance liquid chromatography system (LC10AD system; Shimadzu Co., Kyoto, Japan) equipped with a Shimadzu Shim-pack SCR-102H column (0.8 30 cm) after filtering it through a 0.2-mm-pore-size membrane. The oxidationereduction potential (ORP) and pH were directly determined by using an ORP and a pH electrode, respectively. Concentrations of CH4, CO2, N2, O2, and H2 in the headspace of the UASB reactor and inside the lumen of the hollow fibers of the DM were measured by using a gas chromatography system (GC-14B; Shimadzu Co., Kyoto, Japan) equipped with a thermal-conductivity detector and a Shincarbon-ST column (Shimadzu Co., Kyoto, Japan). The biogas volumes were measured at 25 C. The dissolved gas compositions were measured by using the headspace method (Bandara et al.,
0.1 1.0 0 0 0 0
HRT of DM (hour) 9.2 8.7 8.5 8.8 8.6 8.5 8.6 8.7 8.5 8.5 8.4 4.7 5.6 5.6 2.8 2.8 2.8 8.4
0.2 0.3 0.3 0.4 0.3 0.5 0.2 0 0.8 0 0 0 0 0
Temperature ( C)
Transmembrane pressure (kPa)
35 35 35 35 25 25 25 15 15 15 35 35 35 35 35 35 35 35
0 50 80 0 0 80 0 0 80 0 0 0 80 0 0 80 0 0
CH4 recovery Total CH4 production rate rate (mg(mg-COD/L/day) COD/L/day)
1090 230 1220 140 1370 130 1160 230 1080 85 1260 59 770 97 780 37 1100 150 610 490 1550 170 1460 110 1870 160 1390 190 1320 95 1500 16 1480 300 1220 600
960 220 1180 150 1340 130 1010 230 880 91 1230 54 580 93 530 25 1060 150 720 1390 160 1220 100 1830 160 1150 190 830 63 1140 31 960 280 1300 76
Total CH4 D-CH4 collection recovery efficiency efficiency () ()
68 7 77 7
85 4
86 2
82 1
26 7
89 3 96 1 97 1 90 2 85 2 97 1 78 1 71 2 97 1 76 90 1 84 1 98 0 84 2 64 1 76 2 69 5 90 1
submitted for publication). Liquid samples (50 mL) were obtained from the UASB and DM reactors and were injected with a syringe into a sealed vial (125 mL) that was prefilled with 100% argon gas. We then added 1 mL of 20 mM mercury (II) chloride to the vial to inhibit biological reactions. The vial was vigorously shaken for 5 min to allow for diffusion of the dissolved gas in the liquid sample into the headspace. The liquid sample was allowed to stand for 30 min at room temperature to equilibrate the gas and liquid phases. Then, the composition of the headspace gas was determined using gas chromatography. The standard liquid sample with known concentration of each standard gas was prepared as follows. A vial (125 mL) was filled with 50 mL of distilled water. Argon gas was blown into the distilled water in the vial for 5 min to remove air in the liquid and gas phases. The vial was then sealed with a butyl rubber septum. Next, we injected 1 mL of 100% of each gas, corresponding to 820 mmol at 25 C, to the
3536
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 5 3 3 e3 5 4 0
Table 2 e Characteristics of the degassing membrane module. Item
Characteristics
Non-porous layer for degassing Porous layer Outer diameter of a hollow fiber Inner diameter of a hollow fiber Thickness Length Number of hollow fiber Total membrane area Membrane volume
Polyethylene Polyurethane 280 mm 200 mm 40 mm 360 mm 5500 1.7 m2 0.4 L
vial. The vial was shaken and allowed to stand for 30 min, and then, the composition of the headspace gas was determined by gas chromatography. The concentrations of dissolved gases in the sample liquid were calculated on the basis of the ratio of the amount of gas in the sample to that in the standard sample. All measurements were conducted at a constant temperature of 25 C. On the basis of these results, D-CH4 collection efficiency was calculated as a ratio of D-CH4 concentration collected by the DM to D-CH4 discharged from the UASB reactor. The DCH4 concentration in the DM was calculated as the difference between the D-CH4 concentrations discharged from the UASB and DM reactors. Total CH4 recovery efficiency was calculated as a ratio of the CH4 recovery rate (mg COD L1 day1) to the total CH4 production rate (mg COD L1 day1). The CH4 recovery rate is defined as the sum of the CH4 evolution rate in the UASB headspace and CH4 collection rate in the DM. The total CH4 production rate is defined as the sum of the CH4 evolution rate, CH4 collection rate, and CH4 discharge rate from the DM reactor.
3.
Results and discussion
3.1.
Performance of the UASB reactor
The bench-scale UASB reactor was operated at different temperatures and HRTs for 170 days, and concentrations of influent T-COD and effluent T-COD and D-COD in the UASB reactor were measured (Fig. 1). The average T-COD concentration (standard deviation) of the influent was 1480 240 mg COD L1 from startup to day 128 (by phase 4), after which the influent T-COD concentrations decreased. Even at the beginning of reactor operation, D-COD removal efficiency was high (Fig. 1), probably because of the high amount of biomass, although the influent and effluent T-COD concentrations fluctuated. The average effluent D-COD concentration was 120 40 mg COD L1 by day 128, resulting in a D-COD removal efficiency as high as 92 2%. Total VFA concentration in the UASB reactor was less than 68 mg COD L1 for 170 days (data not shown). The dominant VFA was acetate, with a maximal concentration of 53 mg COD L1 and average concentration of 28 14 mg COD L1 for 170 days. The second dominant VFA was propionate (9 6 mg COD L1); the concentrations of other VFAs (lactate, formate, i-butyrate, and n-butyrate) were less than 1 mg COD L1.
Fig. 1 e Concentrations of influent total chemical oxygen demand (T-COD) and effluent T-COD and dissolved fraction of COD (D-COD) in the upflow anaerobic sludge blanket (UASB) reactor, and D-COD removal efficiency of the UASB reactor. The gray area represents a degasification period. Operational conditions in each phase are summarized in Table 1.
3.2.
Performance of the DM reactor
Since the reactor performance (e.g., the D-COD removal efficiency of the UASB reactor) became stable after adjusting the pH to 7.6 after day 51, degasification was conducted at a transmembrane pressure (absolute pressure) of 50 kPa from day 64 (phase 1.2) to recover residual dissolved CH4 (D-CH4) in the UASB effluent. The average D-CH4 concentration was 61 6 mg COD L1 during normal operation (phase 1.1); in contrast, it was 20 4 mg COD L1 in the effluent of the DM reactor during a degasification period (Phase 1.2) (Fig. 2). The difference between the D-CH4 concentrations in the effluents of the UASB and DM reactors indicated that D-CH4 was successfully collected by the DM during a degasification period. The transmembrane pressure was further reduced by 20 kPa (absolute pressure) at day 78 (phase 1.3) to reduce the residual D-CH4 in the UASB effluent. The D-CH4 concentration was 15 5 mg COD L1 in the DM reactor effluent in phase 1.3 (Fig. 2). On the basis of these results, D-CH4 collection efficiencies were calculated to be 68 7% and 77 7% in phases 1.2 and 1.3, respectively. When the DM reactor was applied to
Fig. 2 e Dissolved methane (D-CH4) concentrations in the upflow anaerobic sludge blanket (UASB) and degassing membrane (DM) reactors, and D-CH4 collection efficiency of the DM reactor. The gray area represents a degasification period. Operational conditions in each phase are summarized in Table 1.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 5 3 3 e3 5 4 0
pure water at 0.3 kPa, dissolved oxygen concentration was decreased to less than 0.1 mg L1 after 5 min, indicating that the theoretical dissolved gas collection efficiency of the DM was almost 100%. To check for reproducibility of the efficiency of degasification, vacuum was discontinued at day 85 (phase 1.4). The average D-CH4 concentration was 63 1 mg COD L1 in the DM reactor effluent in phase 1.4, and this concentration was comparable to that before degasification (phase 1.1) (Fig. 2). These results clearly indicated that D-CH4 was successfully collected by degasification with the DM. Discharge of residual D-CH4 in the effluent of wastewater treatment reactors contributes to an increase in atmospheric CH4, which is 21 times more potent than CO2 as a greenhouse gas (Hartley and Lant, 2006). In addition, the D-CH4 discharge represents a loss of energy that is generated in anaerobic wastewater treatment processes. Therefore, recovery of D-CH4 from the effluent of the anaerobic wastewater treatment reactors is a prerequisite for the discharge of the effluent into the environment. Gas compositions in the headspace of the UASB reactor (Fig. 3A) and inside the lumen of the hollow fibers of the DM (Fig. 3B) are shown in Fig. 3. Although gas compositions fluctuated until day 50, they became relatively stable thereafter because of pH adjustment to 7.6. Average CH4 compositions in the UASB headspace during the degasification period (phases 1.2 and 1.3) were 51 4% and 58 5%, respectively, and they were 49 12% and 56 3% during normal operation (phases 1.1 and 1.4, respectively) (Fig. 3A). Thus, degasification did not significantly affect the CH4 composition in the UASB
Fig. 3 e Gas compositions in the headspace of the upflow anaerobic sludge blanket (UASB) reactor (A) and inside the lumen of the hollow fibers of the degassing membrane (DM), and gas flux into the hollow fibers of the DM (B). The gray area represents a degasification period. Operational conditions in each phase are summarized in Table 1.
3537
headspace. Average compositions of CH4 and CO2 collected through the DM during the degasification period were 22 13% and 10 7%, respectively, in phase 1.2 and 20 2% and 27 6%, respectively, in phase 1.3 (Fig. 3B). O2 was detected in the biogas collected through the DM; this indicated the ingress of air into the dissolved gas collection line. The percentage of N2 plus O2 in the collected biogas varied from 39% to 87% in phase 1. The ratios of N2 to O2 were 3.9 or higher, and these ratios were greater than the ratio of N2 to O2 in air (79/21 or 3.76). This was because the dissolved biogas itself contained N2 but not O2. If air had been excluded from the collected biogas, the average CH4 composition in the collected gas would have been 63 15% in phase 1.2, which would have been comparable to that in the UASB headspace. Several approaches for removal and/or recovery of D-CH4 in wastewater treatment processes have been investigated. Hatamoto et al. (2010) used an encapsulated down-flow hanging sponge (DHS) reactor to biologically oxidize the D-CH4; although this technique enabled oxidization of 95% of the D-CH4, no D-CH4 was recovered as CH4 gas. In contrast, Matsuura et al. (2010) employed a two-stage DHS reactor for post-treatment of the effluent from a UASB reactor treating municipal sewage to recover and oxidize D-CH4. The firststage reactor could recover D-CH4 with 77% recovery efficiency. However, CH4 compositions in the recovered biogas were relatively low because of an inherent limitation of this methoddthe D-CH4 is released from a liquid in the DHS reactor by physical gasification based on the gaseliquid equilibrium. Another approach for removing D-CH4 involved mixing the liquid in an anaerobic wastewater treatment reactor with gas or a paddle. Hartley and Lant (2006) applied intermittent gas mixing by micro-aeration with biogascontaining air, and this resulted in the release of D-CH4 into the gas phase of the reactor. However, the CH4 composition in the recovered biogas was lower than that in the reactor headspace in the absence of aeration, because the gas used for aeration diluted the biogas. Recovered biogas with low CH4 concentration is unsuitable for subsequent energy generation processes. Pauss et al. (1990) mixed the liquid phase in reactors with a paddle or by recirculating the supernatant liquid to enhance the evolution of D-CH4 and gas bubbles attached to solid particles from the liquid phase to gas phase. However, in this method, the D-CH4 concentration would theoretically never be lower than the saturation concentration of D-CH4. Another possible process is to apply vacuum directly to the reactor. It does not seem to be realistic, because the reactor must be closed completely and a pressure tight case. In order to check the mass balance of CH4, the CH4 evolution rate in the UASB headspace, CH4 collection rate from the DM, and CH4 discharge rate from the DM reactor were calculated using the CH4 concentrations and flow rates of wastewater and biogas described above. Fig. 4 shows the rates of CH4 discharged from the UASB reactor and total CH4 recovery efficiency in addition to the 3 above-mentioned CH4 production rates and their averaged values were summarized in Table 1. The average CH4 evolution rates were 1000 94 mg COD L1 day1 and 1120 110 mg COD L1 day1 during the degasification period (phases 1.2 and 1.3, respectively). These rates were comparable to those during normal operation (960 220 mg COD L1 day1 in phase 1.1 and 1010 230 mg
3538
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 5 3 3 e3 5 4 0
Fig. 4 e Rates (mg COD LL1 dayL1) of CH4 evolution in the upflow anaerobic sludge blanket (UASB) headspace, CH4 collection from the degassing membrane (DM), and CH4 discharge from the UASB and DM reactors, as well as total CH4 recovery efficiency. The gray area represents a degasification period. Operational conditions in each phase are summarized in Table 1.
COD L1 day1 in phase 1.4). Thus, degasification did not significantly affect the CH4 evolution rate in the UASB reactor. In contrast, the average CH4 discharge rates were lower during the degasification period (49 10 mg COD L1 day1 in phase 1.2 and 35 11 mg COD L1 day1 in phase 1.3) than during normal operation (140 14 mg COD L1 day1 in phase 1.1 and 150 3 mg COD L1 day1 in phase 1.4) because of the collection of D-CH4 with the DM. The D-CH4 discharge rate accounted for 11 3% of the total CH4 production rate during the normal operation (Phase 1.1), whereas it was 2.6 0.8% during the degasification period (phase 1.3). The average CH4 collection rates were 180 85 mg COD L1 day1 in phase 1.2 and 210 26 mg COD L1 day1 in phase 1.3 (Fig. 4). On the basis of these results, the total CH4 production rates were calculated to be 1090 230 mg COD L1 day1 in phase 1.1 and 1160 230 mg COD L1 day1 in phase 1.4. In contrast, these rates were slightly higher (1220 140 mg COD L1 day1 in phase 1.2 and 1370 130 mg COD L1 day1 in phase 1.3) during the degasification period, probably because of CH4 production in the DM reactor. Accordingly, D-CH4 collection efficiencies were 68 7% and 77 7% in phases 1.2 and 1.3, respectively, and total CH4 recovery efficiencies increased up to 96 1% and 97 1% during the degasification period (phases 1.2 and 1.3, respectively), as compared to those during normal operation (89 3% in phase 1.1 and 90 2% in phase 1.4).
3.3.
Effect of temperature
The UASB and DM reactors were operated at low temperatures (25 C and 15 C in phases 2 and 3, respectively). The average DCH4 concentrations in the UASB reactor were 63 4 mg COD L1 at 35 C throughout phase 1, 82 7 mg COD L1 at 25 C throughout phase 2, and 104 5 mg COD L1 at 15 C throughout phase 3 (Fig. 2). The increase in D-CH4 concentrations at lower temperatures was attributed to an increase in the solubility of CH4 in the liquid phase with decreasing temperature. Corresponding with the increase in D-CH4 concentration, average DCH4 discharge rates increased from 150 12 mg COD L1 day1 at 35 C to 200 20 mg COD L1 day1 at 25 C and
250 13 mg COD L1 day1 at 15 C, thereby indicating that the loss of D-CH4 from the UASB reactor was more significant at lower temperatures. Degasification enabled successful collection of D-CH4 by the DM regardless of temperature. The D-CH4 concentrations decreased to 13 4 mg COD L1 at 25 C (phase 2.2) and 14 2 mg COD L1 at 15 C (phase 3.2) in the DM reactor. The ratio of D-CH4 concentration to CH4 composition in the UASB headspace increased with decreasing temperature. Therefore, the CH4 collection rates showed a relative increase from 210 26 mg COD L1 day1 at 35 C (phase 1.3) to 300 84 mg COD L1 day1 at 25 C (phase 2.2) and to 370 100 mg COD L1 day1 at 15 C (phase 3.2). Accordingly, total CH4 recovery efficiency increased from 71% in phase 3.1e97% at 15 C by degasification (phase 3.2), and D-CH4 collection efficiencies increased with a decrease in temperature from 77 7% at 35 C to 85 4% at 25 C and 86 2% at 15 C. CH4 evolution rates in the UASB headspace decreased with a decrease in temperature (Fig. 4); 1000 170 mg COD L1 day1 throughout phase 1, 860 140 mg COD L1 day1 throughout phase 2, and 650 100 mg COD L1 day1 throughout phase 3. D-COD removal efficiency was not considerably decreased at lower temperatures (94%, 89% and 91% in Phases 1.3, 2.2 and 3.2, respectively). This finding indicated that microbial CH4 production activity was maintained even at low temperatures probably because of the presence of sufficient amount of biomass and high bioavailability of influent organic materials. Hence, the decrease in the CH4 evolution rate might be due to an increase in the solubility of CH4 in the liquid phase (Lettinga et al., 2001). The CH4 recovery rates at 25 C (1230 54 mg COD L1 day1 in phase 2.2) and at 15 C (1060 150 mg COD L1 day1 in phase 3.2) were higher than the CH4 evolution rate at 35 C during normal operation (1000 170 mg COD L1 day1 throughout phase 1). Thus, it can be concluded that the degasification technology enables us to recover CH4 in a UASB process at low temperature without heating, and the amount of CH4 produced in the UASB process is comparable to that produced in a conventional process under mesophilic conditions. In general, many types of wastewaters are discharged at low ambient temperatures under temperate climatic conditions. To treat such wastewaters under mesophilic condition, their temperature must be increased to the optimal mesophilic range. This requires a significant amount of energy, and the high energy cost is a heavy economic burden on such wastewater treatment systems (Lettinga et al., 2001). A UASB process that does not require heating is economically attractive (Dhaked et al., 2010; Angenent et al., 2001). However, the degasification technology also consumed much energy in this study (42 J s1). The maximal CH4 collection rate of 660 mg COD L1 day1 (Fig. 4) was comparable only to 0.14 J s1, assuming that 1 L of CH4 has the energy of 35800 J. Future studies must aim to reduce the energy required for degasification so that the degasification technology may be applied in real wastewater treatment processes. Although the D-COD removal efficiency was slightly decreased at lower temperatures (phases 2 and 3), it was still higher than 87%, and the effluent D-COD concentration was less than 200 mg COD L1 (Fig. 1). In contrast, the P-COD concentration in the DM reactor, which was calculated as the
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 5 3 3 e3 5 4 0
difference between the effluent T-COD and D-COD concentrations, increased at lower temperatures, indicating a decrease in P-COD removal efficiency. At lower temperatures, particles might settle more slowly because of lower liquidesolid separation efficiency, which in turn may be attributed to higher liquid viscosity and/or attachment of gas bubbles onto the particles. Interestingly, the P-COD removal efficiency in the DM reactor was improved by degasification in phase 3. This finding may be explained by the fact that a decrease in the dissolved gas concentration in the liquid in the DM reactor caused by degasification, followed by redissolution of the gas bubbles attached to the particles, caused the particles to efficiently settle down. The degasification technology might improve the efficiency of particle separation in a clarifier for a UASB reactor. To verify the reproducibility of temperature effects, the temperature in the UASB reactor was set at 35 C again at day 123. The average D-COD removal efficiency (92 1%) and the D-CH4 concentration (66 5 mg COD L1) in phase 4 were comparable to those in phase 1 (94 2% and 63 4 mg COD L1, respectively). This finding indicated that changes in chemical parameters (e.g., the D-CH4 concentration and the CH4 evolution rate in the UASB reactor) in phases 2 and 3 were attributed to lower temperatures.
3.4.
Effect of HRT
The UASB reactor was again operated at 35 C but at lower COD concentrations (the concentrations throughout phase 5 and phase 6 were two-third and one-third, respectively, of that in phase 4) (Fig. 1 and Table 1). In order to keep the COD loading rate of the UASB reactor constant, HRTs in phase 5 (6.7 h) and phase 6 (3.3 h) were also reduced to two-third and one-third, respectively, of the HRT in phase 4 (10 h) by increasing the wastewater feed rate. The average D-CH4 concentration in the UASB reactor was ca. 70 mg COD L1 in phases 4 to 7 because of the identical solubility of CH4 at a constant temperature (35 C). The average D-CH4 concentration was ca. 90% of the theoretical saturation concentration of D-CH4 at 35 C (Hartley and Lant, 2006). However, an increase in the wastewater feed rate resulted in an increase in the CH4 discharge rate from the UASB reactor (Fig. 4). The D-CH4 discharge rates were 160 12 mg COD L1 day1 in phase 4, 240 15 mg COD L1 day1 throughout phase 5, and 500 35 mg COD L1 day1 throughout phase 6. Degasification could reduce the D-CH4 concentration to 12 1 mg COD L1 in the DM reactor at an HRT of 6.7 h (phase 5.2). The average CH4 collection rate was 480 130 mg COD L1 day1 in phase 5.2, and this was much higher than that in phase 1.3 (210 26 mg COD L1 day1) because of the higher D-CH4 concentration. The corresponding total CH4 recovery efficiency was 98% and the CH4 recovery rate was 1830 160 mg COD L1 day1, which was 1.5 times higher than that in normal operation at the same HRT in phase 5.1 (1220 100 mg COD L1 day1). These results indicate that the UASB process with degasification is a promising technology for effective CH4 recovery in low-strength wastewater treatment. Because of the economical benefits offered by the UASB process, its application to low-strength wastewater treatment has recently attracted much attention (Elmitwalli et al., 2003; Angenent et al., 2001; Gomec et al., 2008; Matsushige et al.,
3539
1990). At an HRT of 3.3 h, D-CH4 remained at a high concentration (51 3 mg COD L1), and hence, the total CH4 recovery efficiency was as low as 76 2% during the degasification period (phase 6.2). This was probably because of insufficient HRT to collect D-CH4; however, the average CH4 collection rate was still high (370 10 mg COD L1 day1). Future studies should aim to improve the gas flux rate of the DM, for example, by changing membrane materials (Liang et al., 2002). The average biogas collection rates calculated by excluding the amount of air from the collected biogas (i.e., biogas flux through the DM) were 150 14 mL m2 day1 in phase 1.3 and 150 24 mL m2 day1 in phase 6.2 (Fig. 3B), indicating that the decrease in biogas flux due to membrane fouling was negligible in this study.
4.
Conclusions
In this study, the bench-scale UASB reactor equipped with the DM reactor was operated at different temperatures and HRTs for 170 days. D-CH4 was successfully collected by degasification with the DM. Under lower temperatures or shorter HRTs, the D-CH4 concentrations increased; therefore, the D-CH4 collection efficiencies increased. Moreover, the P-COD concentration was decreased by degasification. These results indicated that degasification is a promising technology for improving CH4 recovery and P-COD removal efficiencies of the UASB process for treating low-strength wastewater at low temperature.
Acknowledgments This work was partially supported by the funding for Basic and Applied Researches on Construction Technologies from the Ministry of Land, Infrastructure, Transport, and Tourism of Japan.
references
Angenent, L.T., Banik, G.C., Sung, S.W., 2001. Anaerobic migrating blanket reactor treatment of low-strength wastewater at low temperatures. Water Environment Research 73 (5), 567e574. Bandara, W.M.K.R.T.W., Ikeda, M., Satoh, H., Sasakawa, M., Nakahara, Y., Takahashi, M., Okabe, S. Recovery of dissolved methane gas discharged from a UASB reactor with a degassing membrane. Water Science and Technology, Submitted for publication. Cakir, F.Y., Stenstrom, M.K., 2005. Greenhouse gas production: a comparison between aerobic and anaerobic wastewater treatment technology. Water Research 39 (17), 4197e4203. Dhaked, R.K., Singh, P., Singh, L., 2010. Biomethanation under psychrophilic conditions. Waste Management 30 (12), 2490e2496. Elmitwalli, T.A., van Lier, J., Zeeman, G., Lettinga, G., 2003. Treatment of domestic sewage at low temperature in a twoanaerobic step system followed by a trickling filter. Water Science and Technology 48 (11e12), 199e206. Fang, H.H.P., Liang, D.W., Zhang, T., Liu, Y., 2006. Anaerobic treatment of phenol in wastewater under thermophilic condition. Water Research 40 (3), 427e434.
3540
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 5 3 3 e3 5 4 0
Gomec, C.Y., Letsiou, I., Ozturk, I., Eroglu, Y., Wilderer, P.A., 2008. Identification of archaeal population in the granular sludge of an UASB reactor treating sewage at low temperatures. Journal of Environmental Science and Health, Part A 43 (13), 1504e1510. Hartley, K., Lant, P., 2006. Eliminating non-renewable CO2 emissions from sewage treatment: an anaerobic migrating bed reactor pilot plant study. Biotechnology and Bioengineering 95 (3), 384e398. Hatamoto, M., Yamamoto, H., Kindaichi, T., Ozaki, N., Ohashi, A., 2010. Biological oxidation of dissolved methane in effluents from anaerobic reactors using a down-flow hanging sponge reactor. Water Research 44 (5), 1409e1418. Kashyap, D.R., Dadhich, K.S., Sharma, S.K., 2003. Biomethanation under psychrophilic conditions: a review. Bioresource Technology 87 (2), 147e153. Lettinga, G., Rebac, S., Zeeman, G., 2001. Challenge of psychrophilic anaerobic wastewater treatment. Trends in Biotechnology 19 (9), 363e370. Lettinga, G., Rebac, S., Parshina, S., Nozhevnikova, A.N., van Lier, J.B., Stams, A.J., 1999. High-rate anaerobic treatment of wastewater at low temperatures. Applied and Environmental Microbiology 65 (4), 1696e1702. Liang, T.M., Cheng, S.S., Wu, K.L., 2002. Behavioral study on hydrogen fermentation reactor installed with silicone rubber membrane. International Journal of Hydrogen Energy 27 (11e12), 1157e1165. Luostarinen, S.A., Rintala, J.A., 2005. Anaerobic on-site treatment of black water and dairy parlour wastewater in UASB-septic tanks at low temperatures. Water Research 39 (2e3), 436e448. Madden, P., Chinalia, F.A., Enright, A.M., Collins, G., O’Flaherty, V., 2010. Perturbation-independent community development in low-temperature anaerobic biological wastewater treatment bioreactors. Biotechnology and Bioengineering 105 (1), 79e87. Matsushige, K., Inamori, Y., Mizuochi, M., Hosomi, M., Sudo, R., 1990. The effects of temperature on anaerobic filter treatment
for low-strength organic wastewater. Environmental Technology 11 (10), 899e910. Matsuura, N., Hatamoto, M., Sumino, H., Syutsubo, K., Yamaguchi, T., Ohashi, A., 2010. Closed DHS system to prevent dissolved methane emissions as greenhouse gas in anaerobic wastewater treatment by its recovery and biological oxidation. Water Science and Technology 61 (9), 2407e2415. McKeown, R.M., Scully, C., Mahony, T., Collins, G., O’Flaherty, V., 2009. Long-term (1243 days), low-temperature (4e15 C), anaerobic biotreatment of acidified wastewaters: bioprocess performance and physiological characteristics. Water Research 43 (6), 1611e1620. Pauss, A., Andre, G., Perrier, M., Guiot, S.R., 1990. Liquid-to-gas mass transfer in anaerobic processes: inevitable transfer limitations of methane and hydrogen in the biomethanation process. Applied and Environmental Microbiology 56 (6), 1636e1644. Satoh, H., Miura, Y., Tsushima, I., Okabe, S., 2007. Layered structure of bacterial and archaeal communities and their in situ activities in anaerobic granules. Applied and Environmental Microbiology 73 (22), 7300e7307. Trzcinski, A.P., Stuckey, D.C., 2010. Treatment of municipal solid waste leachate using a submerged anaerobic membrane bioreactor at mesophilic and psychrophilic temperatures: analysis of recalcitrants in the permeate using GC-MS. Water Research 44 (3), 671e680. Voolapalli, R.K., Stuckey, D.C., 1998. Stability enhancement of anaerobic digestion through membrane gas extraction under organic shock loads. Journal of Chemical Technology and Biotechnology 73 (2), 153e161. Xing, W., Zuo, J.E., Dai, N., Cheng, J., Li, J., 2009. Reactor performance and microbial community of an EGSB reactor operated at 20 and 15 C. Journal of Applied Microbiology 107 (3), 848e857.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 5 4 1 e3 5 4 9
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
An approach to improve the separation of solideliquid suspensions in inclined plate settlers: CFD simulation and experimental validation A.I. Salem*, G. Okoth, J. Tho¨ming University of Bremen, Centre for Environmental Research and Technology (UFT), Department of Chemical Engineering e Regeneration and Recycling, Leobener Strasse, D28359 Bremen, Germany
article info
abstract
Article history:
The most important requirements for achieving effective separation conditions in inclined
Received 11 November 2010
plate settler (IPS) are its hydraulic performance and the equal distribution of suspensions
Received in revised form
between settler channels, both of which depend on the inlet configuration. In this study,
7 April 2011
three different inlet structures were used to explore the effect of feeding a bench scale IPS
Accepted 13 April 2011
via a nozzle distributor on its hydraulic performance and separation efficiency. Experi-
Available online 21 April 2011
mental and Computational Fluid Dynamic (CFD) analyses were carried out to evaluate the hydraulic characteristics of the IPS. Comparing the experimental results with the predicted
Keywords:
results by CFD simulation implies that the CFD software can play a useful role in studying
Computational fluid dynamics (CFD)
the hydraulic performance of the IPS by employing residence time distribution (RTD)
Residence time distribution (RTD)
curves. The results also show that the use of a nozzle distributor can significantly enhance
Plug flow
the hydraulic performance of the IPS, which contributes to the improvement of its sepa-
Non-ideal flow
ration efficiency. ª 2011 Elsevier Ltd. All rights reserved.
Inclined plate settlers Sedimentation
1.
Introduction
A gravity settler is a common unit with wide applications in the treatment of water, domestic wastewater and industrial wastewater. Solideliquid separation efficiency of this type is directly related to the surface area available for settling, but under limited space conditions e as in industrial wastewater treatment e the use of an inclined plate settler (IPS) is preferred. It provides high spaceetime yield due to the short settling distance, and the available settling area is dependant on the total area of the plates projected on a horizontal surface. Separation efficiency of IPS is usually well below the theoretical performance due to many factors. Okoth et al. (2008) summarised these factors by the modelling of
suspensionesediment interaction phenomenologically. In their experimental study, they employed a nozzle distributor to improve the hydraulic performance of IPS which is one of the factors affecting the separation efficiency. However, they did not investigate in detail the effect of using this nozzle on both distribution of the total flow between the settler channels and flow patterns of the entire IPS system, both of which have a significant impact on the separation performance of the IPS. The equalized distribution of suspension within each settler is important to obtain an equal overflow velocity on every plate, which contributes to the improvement of the IPS efficiency, while the flow patterns are essential in determining the flow characteristics in order to achieve a reliable design (Lo´pez et al., 2008).
* Corresponding author. E-mail address:
[email protected] (A.I. Salem). 0043-1354/$ e see front matter ª 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2011.04.019
3542
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 5 4 1 e3 5 4 9
To explore operational shortcoming and characterize new separation equipment, it is important to understand the reactor hydrodynamics, which may be achieved by interpreting the reactor residence time distribution (RTD) (Behin and Aghajari, 2008). The residence time distribution of a reactor is one of the most informative characterizations of the flow pattern in a reactor (Gavrilescu and Tudose, 1999). Knowledge of the liquid RTD is important for a number of reasons, such as allowing for accurate modelling of the system and aiding in reactor design to achieve or preserve a desired flow pattern (Behin and Aghajari, 2008). The flow in tanks always deviates from the ideal plug flow or complete mixed flow and is usually described as a non-ideal flow pattern. Levenspiel (1999) described two methods for the characterization of a non-ideal flow pattern. One is the longitudinal dispersion (LD) model which represents a flow that deviates from plug flow, and the other is the tanks in series (TIS) model which describes the mixing within a single real reactor as a number of equally sized continuous stirredtank reactors in series (CSTRs). When the number of tanks in series is one, the model predicts the performance of an ideal CSTR. As the number of tanks in the TIS model increases, the flow within the reactor approaches that of a plug flow reactor (PFR) (Bircumshaw et al., 2006). Both the LD and TIS models are characterised by dispersion number and the number of tanks in series (NTIS) respectively. The TIS model was used in this study because it is mathematically much simpler than the LD model. Also, the NTIS does not depend on the definition of the inlet and exit boundary conditions. Computational fluid dynamics (CFD) has become a powerful tool in the reactor design process and provides useful and detailed information prevailing in the reactors, such as velocity field, concentration distribution, and phase hold-up distribution (Zhang et al., 2007). CFD is typically used in the simulation of RTD (Patwardhan, 2001; Choi et al., 2004; Moullec et al., 2008; Aubin et al., 2009). CFD codes used in software package such as Fluent, CFX and Cosmos offer different models for numerical solution of Navier Stokes partial differential equations. Generally, the model should be experimentally verified (Thy´n et al., 2002). Two turbulent models (kee model and keu model) were implemented in the present study to predict both velocity flow field and RTD curve. Thereafter, the predicted results were compared with the experimental results to determine which of the two models give the most realistic results. Furthermore, the influence of hydraulic performance on the separation efficiency of the bench scale IPS fed by a nozzle distributor is demonstrated in this study.
2.
Material and method
Because of the inlet structure plays a very important role in determining the flow characteristics in the downstream particle separation zone (He et al., 2008), three inlet configurations were used to investigate the impact of feeding the IPS via a nozzle distributor on its hydraulic behaviour and separation efficiency. Sketches of the employed inlet structures are shown in Fig. 1.
The IPS was made of plexiglas of 15 mm in thickness with internal dimensions of (100 mm 80 mm 480 mm) and was placed on a ramp with angle of inclination 45 in all tests. Three polyvinylchloride plates 300 mm long and 5 mm thick were used in the IPS. The three plates could be fixed at any distance from the nozzle apex, and the spacing between the plates was 15 mm.
2.1.
Experimental set-up and procedures
Two techniques were used to identify the hydraulic behaviour of the IPS. The first technique was measurement the velocity within every settler by using the colour velocity measurement (CVM) method (United States Department of the Interior Bureau of Reclamation, 1997). The second method was used to quantify the hydraulic behaviour of the IPS by using residence time distribution (RTD) experiment. Furthermore, to explore the impact of using distribution nozzle on the separation process, the removal of suspended solids (SS) efficiency for the IPS was determined. A small slug of concentration dye solution (potassium permanganate) was injected impulsively into the IPS inlet, and a high resolution digital camera was used to provide the data for the calculation of mean velocity within every settler by computing the required mean time for the dye to travel a known length. The RTD was measured by quickly injecting 3 mL of tracer (KCL, 3 g/l) into the IPS inlet and the tracer concentration at the outlet was measured with a conductivity probe every 5 s using a data acquisition system. The experimental procedures were repeated five times for each flow rate. The fitting of curve was then performed to minimize the deviation between the experimental data and the simulation data by using exponentially modified Gaussian peak function which has given us adjusted R2 values between 0.94 and 0.97. The separation efficiency was determined by specifying the concentration of SS in the samples which were collected from the inlet stream and outlet stream. The samples were filtered under pressure through a 0.45 mm pore size cellulose nitrate membrane using a compressed air filter model 16249 from Sartorius AG. This was followed by dry mass concentration analysis using a moisture analyser model MA45 also from Sartorius AG, Germany (Okoth et al., 2008). To carry out these tests, an experimental set-up was constructed. A process flow diagram of the experimental set-up is shown in Fig. 2. It consists of a 35 L storage tank with an aerator mounted at the bottom denoted as A. This tank was filled with tap water in both velocity measurement and RTD experiments, while it was filled with suspension e including crushed walnut shell particles e in the separation efficiency test. The feeding of fluid was achieved by using a centrifugal pump denoted as B. The flow rates were regulated by using variable direct-current voltage power supply in a range of 0e24 V. There is a flow meter between the pump and the inlet, denoted as C. Both the conductivity meter and data acquisition system e denoted as D and E e were used only in the RTD test. To determine the separation efficiency, three sets of samples were collected from both the inlet stream (Section 1) and the outlet stream (Section 2).
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 5 4 1 e3 5 4 9
3543
Fig. 1 e Sketches of the employed inlet structures. Where LS1 is fed by pipe, LS2 is fed by a nozzle distributor which is surrounded by the IPS wall, and LS3 is fed by a nozzle distributor which is surrounded by an additional cylindrical wall.
The crushed walnut shell particles had a density of 1.35 g/cm3 and their size distribution was analysed using the laser diffraction technique, with a Malvern (Mastersizer, 2000) analyser. Particle size distribution parameters like d(0.1), d(0.5), d(0.9) were analyzed and the corresponding values were 35, 115, and 235 mm respectively. The d(0.1), d(0.5) and d(0.9) values indicate that 10%, 50% and 90% of the particles have diameters which are smaller than or equal to the stated size. The separation experiments were carried out as follows: At the beginning of each experiment the suspension was mixed well in the feed tank for 5 min to achieve a uniform distribution. The mixture was then pumped to the IPS. Three samples were collected from the inlet stream after 5 min from the onset of pumping. On other hand, no fixed time was set for the
Fig. 2 e Process flow diagram of experimental set-up with aerator (A), pump (B), rotameter (C), conductivity probe (D), data acquisition system (E), camera (F), and test section for measuring dye velocity through settlers (L). (1) and (2) denote the sample collection points in the inlet and outlet streams, respectively, while (3) denotes the stream of withdrawal concentration sediment.
outlet stream sampling since this depended on the hydraulic residence time. At the end of every experimental run, the IPS was drained and the samples returned to the feed tank. The longest duration of each experiment was 15 min.
2.2.
CFD simulations
CFD analysis using CFX-10 from ANSYS was performed by employing the standard kee model and keu model, where k, e, and u denote turbulent kinetic energy, turbulent dissipation rate, and turbulent frequency, respectively. For wall-bounded flows, as in IPS, the standard kee model neglects the effects of viscosity in the near-wall region, and it is valid for turbulent core flow (Hrenya et al., 1995). A scalable wall function is adopted in CFX to improve the near wall treatment by limiting the value of the dimensionless distance from the wall (yþ) to be 11.06 (where 11.06 is the intersection between the logarithmic and linear near wall profile), which prevents the mesh points falling within the viscous sub-layer. Thus, all fine mesh inconsistencies are avoided (Grotjans and Menter, 1998). On the other hand, an automatic near wall treatment is used in the keu model by Wilcox (1988), which provides an analytical solution for u in both the logarithmic and the viscous regions. The idea behind the automatic wall treatment is that the model shifts gradually between a viscous sub-layer formulation and wall functions, based on the grid density near wall (Cheng and Tak, 2006). The two models were selected for this study for purpose of comparison, and to determine which model would be closed to experimental results. The governing equations of mass and momentum were determined using the Reynolds averaged NaviereStokes Equations. (1) and (2). vr þ V$ðrUÞ ¼ 0 vt
(1)
T v þB rU þ V$ðrU5UÞ ¼ Vp0 þ V$ meff VU þ meff VU vt
(2)
Where r is the liquid density, B is the sum of body forces, U is the mean velocity vector, p0 is the modified pressure and meff is the effective viscosity.
3544
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 5 4 1 e3 5 4 9
The calculations of p0 and meff are: 2 p0 ¼ p þ rk 3
(3)
meff ¼ m þ mt
(4)
Where mt is the turbulent viscosity, which is given by: mt ¼ Cm r
mt ¼ r
k2 e
(5)
k u
(6)
Equations (5) and (6) were employed for the kee and the keu models respectively. The values of k and e in the kee model are estimated from Equations (7) and (8) respectively. vðrkÞ m þ V$½rUk ¼ Pk re þ V$ eff Vk sk vt m vðreÞ e þ V$½rUe ¼ ðCe1 Pk Ce2 reÞ þ V$ eff Ve se vt k
(7)
vðruÞ u þ V$½rUu ¼ a Pk bru2 þ V$ vt k
m m þ t Vu su
(12)
The normalized concentration E(t) is used to compare RTD curves under different flow rate conditions, and it is defined by the following equation: EðtÞ ¼ ZN
(8)
CðtÞ
(13)
0
The mean residence time is given by: ZN tCðtÞd t 0
(9)
tm ¼ ZN
P ti Ci Dti yP Ci Dti
(14)
CðtÞd t 0
(10)
Where Pk is the production rate of turbulence The set of the kee model constants is Cm ¼ 0.09, Ce1 ¼ 0.1256, Ce2 ¼ 1.92, sk ¼ 0.9, se ¼ 1.3. While the values of the keu model constants are b0 ¼ 0.09, a ¼ 5/9, b ¼ 3/40, sk ¼ 2, su ¼ 2. The simulation was performed in two stages. The first stage e steady-state turbulent flow e was solved using both the kee model and keu model individually to determine the velocity profiles, kinetic energy, eddy energy, and eddy diffusivity. The second stage, the information obtained from the first stage was used to solve a non-reacting scalar transport equation, based on the assumption that the tracer is dispersed in the tank by convection and diffusion as follows: vF m F þ V$ðUFÞ ¼ V rDF þ t V$ þ SF Sct vt r
CðtÞ ¼ Foutlet ðtÞ
CðtÞd t
Where Pk is the shear production due to turbulence. The values of k and u in the keu model are estimated from Equations (9) and (10) respectively. vðrkÞ m þ V$½rUk ¼ Pk b0 rku þ V$ m þ t Vk sk3 vt
with a value from 55 to 97 g/s, whereas the boundary condition at the outlet was set as the average static pressure across the outlet area. The tetrahedral grid was used because it is the most common way of numerically solving problems in threedimensional domains of complex shapes (Mavriplis, 1997; Kruglyakova et al., 1998). Furthermore, the inflated mesh was used in the near-wall regions to capture the effects of the boundary layer. The mesh structures for LS1, LS2 and LS3 have 1.09 106, 1.2 106 and 1.07 106 elements respectively. By determining the tracer concentration F at the outlet from the transient results, the residence time distribution could be computed as follow:
The spread of the residence time curve (variance) is measured by: ZN t2 CðtÞd t s2 ¼ 0ZN
(15)
CðtÞd t 0
The normalized variance is calculated from the following equation: s2q ¼
s2 t2m
(16)
Finally the following equation estimates the NTIS: NTIS ¼
1 s2q
3.
Results and discussion
3.1.
Flow distribution study
(11)
Where F is the tracer concentration, F/r is the conserved quantity per unit mass, SF is a volumetric source term, DF is the kinematic diffusivity for the scalar and Sct is the turbulence Schmidt number, indicating the ratio between the rate of momentum transport and passive scalars (ANSYS Inc., 2005; Zhang et al., 2007). The used tracer was KCL which has molecular diffusivity of 1.703 109 m2/s (Griffiths et al., 1915). The boundary conditions for the system were as follows: IPS wall, nozzle distribution, and lamella plates were assumed as standard wall boundary conditions with no-slip flow. The boundary condition at the inlet was set as the mass flow rate
P ðti tm Þ2 Ci Dti P ðtm Þ2 y Ci Dti
(17)
Owing to the hypothesis made that the IPS efficiency depends on the quality of flow distribution within each settler, the velocity through every settler was determined. The flow velocity in the CFD was derived from the calculation of velocity at a specific section in the transverse direction, which was in the middle of the settler, while in the experiment it resulted from the required mean time for the dye to
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 5 4 1 e3 5 4 9
3545
travel a known length in the longitudinal direction. Thereafter, the standard deviation (SD) of a set of the values of flow velocities through every settler was utilized as a criterion for the flow distribution efficiency. As shown in Fig. 3aeC, the SD increases as the flow rate increases, indicating a decrease in the flow distribution efficiency. Furthermore, the distribution efficiency clearly depends on the type of inlet structure. It was optimal for LS3, where the average values of SD were between 0.09 and 0.22, while the inlet structure of LS1 showed the worst performance, with average SD values from 1.3 to 2.1, and highest dependency of SD on flow rate. Obviously, the flow distribution within the IPS can be significantly improved by using the nozzle distributor. Additionally, the simulation and experimental results clearly deviate significantly because two different methods were used in the determination of the flow velocity through the settlers, and the purpose here was only to perform a qualitative evaluation.
3.2.
Hydraulic behaviour study
3.2.1.
RTD experiments
Runs were carried out at four different flow rates, with values of 200,250, 300, and 350 l/h and RTD curves were plotted between E(t) versus (t). The influence of flow rate on the normalised concentration curves for different inlet structure is shown in Fig. 4. As expected, time to peak of the curve becomes shorter with increasing flow rate in the three cases (LS1, LS2 and LS3). Also
Fig. 4 e The normalised RTD curves as function in flow rate experimentally: (a) LS1, (b) LS2, and (c) LS3.
a strong tail of the RTD curve was observed in all cases. This leads to increase deviation from the ideal case (i.e. plug flow) which is not desirable for the performance of the separation process (Kuoppama¨ki, 1977; Wilkinson et al., 2000). Furthermore, the existence of a tail indicates presence of dead space,
Fig. 3 e The influence of both inlet configuration and flow rate on the efficiency of flow distribution between lamella plates: (a) experimental data, (b) kee model, and (c) keu model.
Fig. 5 e The impact of both inlet configuration and flow rate on the flow pattern of the IPS.
3546
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 5 4 1 e3 5 4 9
Fig. 6 e Comparison of normalised RTD curves between simulations and experimental results at flow rates: (a) Q [ 200 l/h; keu model. (b) Q [ 250 l/h; (c) Q [ 300 l/h; (d) Q [ 350 l/h; C experiment, d kee model, and
which gives an indication of stagnant pockets or recirculation regions. These regions reduce the effective volume and should be kept as small as possible. To quantify the hydraulic performance of the IPS, the NTIS is calculated from the RTD curve. Fig. 5 illustrates the impact
of both inlet configuration and flow rate on the NTIS. This figure reveals that the NTIS depends strongly on the inlet structure, additionally, as expected, the hydraulic behaviour of LS1 is significantly different from the plug flow due to nonuse of an effective inlet device to distribute the suspension. In
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 5 4 1 e3 5 4 9
contrast, the NTIS for LS2 and LS3 is about 7e10 which indicated tendencies to plug flow.
3.2.2. Comparison between numerical simulation and experimental results Fig. 6 illustrates a comparison between the simulated data and experimental measurements of the normalised RTD. As can be seen from the figure, the main characteristic for LS1 is an extreme initial peak with an average value of q w 0.50 which indicates a short-circuiting stream where q is dimensionless time [q ¼ ti/Tmean; Tmean ¼ theoretical mean retention time]. By contrast, the average values of initial peak for both LS2 and LS3 reached maximum concentration at q ¼ 0.78, and q ¼ 0.90 respectively. This means the hydraulic behaviour of LS3 approaches plug flow conditions (Tchobanoglous et al., 2003). Moreover, it can be observed from the figure that the simulated RTD curves of both the keu model and kee model are deviated slightly for LS1. On the other hand, for both LS2 and LS3, the predicated results by the kee model for both types differ considerably from the predicated results by the keu model. This means that a correct modelling of the nozzle distributor plays an important role in the final results. Also, it can be generally observed from the figures for both LS2 and LS3 that the predicted results obtained by the keu model better match the experimental observation when compared with the kee model. This is especially evident in the prediction of the time at which the peak concentration of the tracer is observed, the end of the tail and the normalized variance values. Whereas the discrepancy of the normalized variance values between the predicted results by keu model and experiments data was around 8% for LS2 and LS3, while the differences in the normalized variance values between the predicted results by the kee model and experiments data was within 20% for LS2 and 35% for LS3. This indicates that the keu model provides a good overall description of the IPS behaviour. As previously mentioned, having a correct model of the nozzle distributor has a large impact on the final results, and despite the use of inflated mesh to resolve the wall-layer accurately, there was little qualitative discrepancy between the predicted RTD curve by the keu model and the experimental results. This discrepancy could be because of existing eddies of a wide range of length scale in the boundary layer, which are totally modelled in the keu model. Further work needs to be undertaken to improve the prediction of the RTD curve by using other models, such as a large-eddy simulation model. This model has the capability of resolving the threedimensional time-dependent details of the large and medium (i.e., resolved) scales, whereas the effects of the small unresolved eddies are modelled with a sub-grid turbulence model (Fureby et al., 2004). It is important to study the effect of flow rate on RTD curves. It is noticeable that an increase in liquid flow rate leads to decreases in the average residence time as discussed in the RTD experiments section. Also, the normalized variance values of LS1 increase as flow rate increases for both experiment and simulation results. On the other hand, in both the LS2 and the LS3, flow rate did not change the normalized variance values significantly, despite approximately 2-fold increases in the flow rate. This indicates that a wide working range for both LS2 and LS3.
3547
Fig. 7 illustrates the hydraulic characteristics of the IPS as function in both the flow rate and inlet structure. This figure shows that the calculated NTIS from CFD simulations by using the keu model also match well with the experimental findings. Furthermore, it can be observed that the values of NTIS based on the normalized variances are equivalent to 2 to 3 NTIS for LS1, 7 to 9 NTIS for LS2 and 9 to 10 NTIS for LS3. These values again reveal that the LS3 provides significantly better hydraulic behaviour amongst the three.
3.3.
Separation efficiency study
The impact of flow rate and inlet configuration on the removal of suspended solids is illustrated in Fig. 8. The results reveal that the separation efficiency decreases as the flow rate increases for the three inlet structures. This could be expected based on the previous results, which indicated that the hydraulic behavior of IPS decreases with increases in the flow rate. It can also be observed that the separation efficiency for both LS1 and LS2 are closed to each other in the flow range of 200 l/h to 250 l/h which means that the separation efficiency at low flow rate does not depend significantly on the configuration of the inlet zone. Moreover it can be seen that the separation efficiency of LS1 drops rapidly after 250 l/h. On the contrary, the separation efficiency of both LS2 and LS3 decreases gradually after 250 l/h. In general, it can be said that the inlet configuration employed in LS3 improved both the distribution of flow within every settler and the hydraulic
Fig. 7 e Comparison of the predicted NTIS numerically with experimental data: (a) LS1, (b) LS2, and (c) LS3.
3548
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 5 4 1 e3 5 4 9
E(q) k NTIS p0 SF Sct tm Tmean ti U
Fig. 8 e Separation efficiency [h [ (SSin L SSout) 3 100/SSin] as a function of both flow rate and inlet configuration.
behaviour of the IPS, which led to an improvement in its separation efficiency.
4.
Conclusions
The impact of feeding the IPS via a nozzle on the quality of flow distribution between the IPS channels, flow patterns of the entire IPS system and separation efficiency could be clearly demonstrated via CFD simulation and experiments at different flow rates. The comparison between CFD simulation and experimental data showed better agreement for the keu model than the kee model. Furthermore, the results show that the inlet structure of LS3 is helpful in improving both the distribution of flow within every settler and hydraulic behaviour of the IPS, both of which could improve the separation efficiency of the IPS. Further work is needed to validate the use of this nozzle distributor in a larger size IPS instead of a small laboratory model.
Acknowledgements The authors are grateful to Michael Birkner for manufacturing the experimental set-up and assisting in some experiments. Also the authors would like to thank Lydia Achelis, for her support in the particles size analysis. Further, A. Salem gratefully acknowledges the Egyptian government (gs1) for providing the financial support for his scholarship through the Missions Department in the Ministry of Higher Education.
Nomenclature
B the sum of body forces, N Cm, Ce1, Ce2 constants in the kee model dimensionless kinematic diffusivity, m2/s DF E(t) residence time distribution, s1
dimensionless residence time distribution turbulent kinetic energy, m2/s2 number of tanks in series model modified pressure, Pa volumetric source term turbulence Schmidt number dimensionless mean residence time, s theoretical mean retention time, s time interval, s mean velocity vector, m/s
Greek symbols e turbulent dissipation rate, m2/s3 u turbulent frequency, s1 r Density, kg/m3 meff effective viscosity, Pa s turbulent viscosity, Pa s mt sk, se, b0 , a, b, su constants in the kee model and keu model dimensionless F tracer concentration, kg/m3 s2 spread of the residence time curve normalized variance of residence time distribution s2q q dimensionless time
references
ANSYS Inc., 2005. Germany GmbH. Aubin, J., Prat, L., Xuereb, C., Gourdon, C., 2009. Effect of microchannel aspect ratio on residence time distributions and the axial dispersion coefficient. Chemical Engineering and Processing: Process Intensification 48 (1), 554e559. Behin, J., Aghajari, M., 2008. Influence of water level on oilewater separation by residence time distribution curves investigations. Separation and Purification Technology 64 (1), 48e55. Bircumshaw, L., Changunda, K., Hansford, G., Rawatla, R., 2006. Development of a mathematical model for continuous tank bioleaching. The Journal of the South African Institute of Mining and Metallurgy 106, 277e282. Cheng, X., Tak, N., 2006. Investigation on turbulent heat transfer to leadebismuth eutectic flows in circular tubes for nuclear applications. Nuclear Engineering and Design 236 (4), 385e393. Choi, B.S., Wan, B., Philyaw, S., Dhanasekharan, K., Ring, T.A., 2004. Residence time distribution in a stirred tank: comparison of CFD predictions with experiment. Industrial and Engineering Chemistry Research 43 (20), 6548e6556. Fureby, C., Alin, N., Wilkstrom, N., Menon, S., Svanstedt, N., Persson, L., 2004. Large eddy simulation of high-Reynoldsnumber wall-bounded flows. The American Institute of Aeronautics and Astronautics Journal 42 (3), 457e468. Gavrilescu, M., Tudose, R.Z., 1999. Residence time distribution of the liquid phase in a concentric-tube airlift reactor. Chemical Engineering and Processing 38 (3), 225e238. Griffiths, A., Dickson, J.M., Griffiths, C.H., 1915. Determination of the coefficient of diffusion of potassium chloride by an analytical method. Proceedings of the Physical Society of London 28, 935e937. Grotjans, H., Menter, F.R., 1998. Wall functions for general application CFD codes. ECCOMAS 98. In: Proceedings of the Fourth European Computation Fluid Dynamics Conference, pp. 1112e1117. He, C., Wood, J., Marsalek, J., Rochfort, Q., 2008. Using CFD modeling to improve the inlet hydraulics and performance of a storm-water clarifier. Journal of Environmental Engineering 134 (9), 722e730.
w a t e r r e s e a r c h 4 5 ( 2 0 1 1 ) 3 5 4 1 e3 5 4 9
Hrenya, C.M., Bolio, E.J., Chakrabarti, D., Sinclair, J.L., 1995. Comparison of low Reynolds number kee turbulence models in predicting fully developed pipe flow. Chemical Engineering Science 50 (12), 1923e1941. Kruglyakova, D.V., Neledova, A.V., Tishkin, V.F., Filatov Yu, A., 1998. Unstructured adaptive grids for problems of mathematical physics (survey). Mathematical Modeling (in Russian) 10 (3), 93e116. Kuoppama¨ki, R., 1977. The applicability of tracer techniques for studies on sewage treatment process dynamics. The International Journal of Applied Radiation and Isotopes 28 (10e11), 833e837. Levenspiel, O., 1999. Chemical Reaction Engineering, third ed. John. Wiley & Sons Inc. Lo´pez, P.R., Lavı´n, A.G., Lo´pez, M.M.M., Heras, J.L.B., 2008. Flow models for rectangular sedimentation tanks. Chemical Engineering and Processing: Process Intensification 47 (9e10), 1705e1716. Mavriplis, D.J., 1997. Unstructured grid techniques. The Annual Review of Fluid Mechanics 29, 473e514. Moullec, Y.L., Potier, O., Gentric, C., Leclerc, J.P., 2008. Flow field and residence time distribution simulation of a cross-flow gaseliquid wastewater treatment reactor using CFD. Chemical Engineering Science 63 (9), 2436e2449. Okoth, G., Centikaya, S., Bru¨ggemann, J., Tho¨ming, J., 2008. On hydrodynamic optimisation of multi-channel counter-flow
3549
lamella settlers and separation efficiency of cohesive particles. Chemical Engineering and Processing: Process Intensification 47 (1), 90e100. Patwardhan, A.W., 2001. Prediction of residence time distribution of stirred reactors. Industrial and Engineering Chemistry Research 40 (24), 5686e5695. Tchobanoglous, G., Burton, F., Stensel, H.D., 2003. Wastewater Engineering Treatment and Reuse, fourth ed. Metcalf & Eddy, Inc. Thy´n, J., Novy´, M., Houdek, P., Portela, B., Zitny ´ , R, 2002. Verification of CFD model by stimulus response method. In: 15th International Congress of Chemical and Process Engineering CHISA. Praha. United States Department of the Interior Bureau of Reclamation, 1997. Water Measurement Manual. www.usbr.gov/pmts/ hydraulics_lab/pubs/wmm/. Wilcox, D.C., 1988. Multiscale model for turbulent flows. The American Institute of Aeronautics and Astronautics Journal 26 (11), 1211e1320. Wilkinson, D., Waldie, B., Nor, M.I.M., Lee, H.Y., 2000. Baffle plate configurations to enhance separation in horizontal primary separators. Chemical Engineering Journal 77 (3), 221e226. Zhang, L., Pan, Q., Rempel, G.L., 2007. Residence time distribution in a multistage agitated contactor with Newtonian fluids: CFD prediction and experimental validation. Industrial and Engineering Chemistry Research 46 (11), 3538e3546.
water research 45 (2011) 3550
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Erratum
Erratum to “Performance of CHROMagarä Staph aureus and CHROMagarä MRSA for detection of Staphylococcus aureus in seawater and beach sand e Comparison of culture, agglutination, and molecular analyses” [Water Research 43 (2009) 4802e4811] K.D. Goodwin a,*, M. Pobuda b a
National Oceanic and Atmospheric Administration (NOAA) Atlantic Oceanographic & Meteorological Laboratories (AOML), 4301 Rickenbacker Causeway, Miami, FL 33149, USA b Cooperative Institute of Marine and Atmospheric Studies, Rosenstiel School of Marine and Atmospheric Science, University of Miami, 4600 Rickenbacker Causeway, Miami, FL 33149, USA
The mecA F primer sequence given in Table 2 is the reverse complement of the actual primer sequence. The sequence should actually be as listed in Mason et al. (2001): TCCAGGAATGCAGAAAGACCAAAGC.
reference
Mason, W.J., Blevins, J.S., Beenken, K., Wibowo, N., Ojha, N., Smeltzer, M.S., 2001. Multiplex PCR protocol for the diagnosis of Staphylococcal infection. Journal of Clinical Microbiology 39, 3332e3338.
DOI of original article: 10.1016/j.watres.2009.06.025. * Corresponding author. Tel.: þ1 858 546 7142; fax: þ1 858 546 7003. E-mail address:
[email protected] (K.D. Goodwin). 0043-1354/$ e see front matter Published by Elsevier Ltd. doi:10.1016/j.watres.2011.04.022